From: Mirko Vogt Date: Sat, 22 Aug 2009 12:46:12 +0000 (+0000) Subject: add generic patchset levelled up to 2.6.31-rc6 X-Git-Url: http://git.openwrt.org/?p=openwrt%2Fstaging%2Fdedeckeh.git;a=commitdiff_plain;h=6e2a58ac3bdec3315fd9934cf46746143243594f add generic patchset levelled up to 2.6.31-rc6 SVN-Revision: 17351 --- diff --git a/target/linux/generic-2.6/patches-2.6.31/004-extra_optimization.patch b/target/linux/generic-2.6/patches-2.6.31/004-extra_optimization.patch new file mode 100644 index 0000000000..333acb612b --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/004-extra_optimization.patch @@ -0,0 +1,12 @@ +--- a/Makefile ++++ b/Makefile +@@ -559,6 +559,9 @@ endif + NOSTDINC_FLAGS += -nostdinc -isystem $(shell $(CC) -print-file-name=include) + CHECKFLAGS += $(NOSTDINC_FLAGS) + ++# improve gcc optimization ++CFLAGS += $(call cc-option,-funit-at-a-time,) ++ + # warn about C99 declaration after statement + KBUILD_CFLAGS += $(call cc-option,-Wdeclaration-after-statement,) + diff --git a/target/linux/generic-2.6/patches-2.6.31/006-gcc4_inline_fix.patch b/target/linux/generic-2.6/patches-2.6.31/006-gcc4_inline_fix.patch new file mode 100644 index 0000000000..642f11b21f --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/006-gcc4_inline_fix.patch @@ -0,0 +1,11 @@ +--- a/arch/mips/include/asm/system.h ++++ b/arch/mips/include/asm/system.h +@@ -187,7 +187,7 @@ extern __u64 __xchg_u64_unsupported_on_3 + if something tries to do an invalid xchg(). */ + extern void __xchg_called_with_bad_pointer(void); + +-static inline unsigned long __xchg(unsigned long x, volatile void * ptr, int size) ++static __always_inline unsigned long __xchg(unsigned long x, volatile void * ptr, int size) + { + switch (size) { + case 4: diff --git a/target/linux/generic-2.6/patches-2.6.31/007-samsung_flash.patch b/target/linux/generic-2.6/patches-2.6.31/007-samsung_flash.patch new file mode 100644 index 0000000000..5c835081b8 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/007-samsung_flash.patch @@ -0,0 +1,36 @@ +--- a/drivers/mtd/chips/cfi_cmdset_0002.c ++++ b/drivers/mtd/chips/cfi_cmdset_0002.c +@@ -51,6 +51,7 @@ + #define SST49LF040B 0x0050 + #define SST49LF008A 0x005a + #define AT49BV6416 0x00d6 ++#define MANUFACTURER_SAMSUNG 0x00ec + + static int cfi_amdstd_read (struct mtd_info *, loff_t, size_t, size_t *, u_char *); + static int cfi_amdstd_write_words(struct mtd_info *, loff_t, size_t, size_t *, const u_char *); +@@ -386,12 +387,19 @@ struct mtd_info *cfi_cmdset_0002(struct + + if (extp->MajorVersion != '1' || + (extp->MinorVersion < '0' || extp->MinorVersion > '4')) { +- printk(KERN_ERR " Unknown Amd/Fujitsu Extended Query " +- "version %c.%c.\n", extp->MajorVersion, +- extp->MinorVersion); +- kfree(extp); +- kfree(mtd); +- return NULL; ++ if (cfi->mfr == MANUFACTURER_SAMSUNG && ++ (extp->MajorVersion == '3' && extp->MinorVersion == '3')) { ++ printk(KERN_NOTICE " Newer Samsung flash detected, " ++ "should be compatibile with Amd/Fujitsu.\n"); ++ } ++ else { ++ printk(KERN_ERR " Unknown Amd/Fujitsu Extended Query " ++ "version %c.%c.\n", extp->MajorVersion, ++ extp->MinorVersion); ++ kfree(extp); ++ kfree(mtd); ++ return NULL; ++ } + } + + /* Install our own private info structure */ diff --git a/target/linux/generic-2.6/patches-2.6.31/011-mips_boot.patch b/target/linux/generic-2.6/patches-2.6.31/011-mips_boot.patch new file mode 100644 index 0000000000..c2a043acde --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/011-mips_boot.patch @@ -0,0 +1,11 @@ +--- a/arch/mips/kernel/head.S ++++ b/arch/mips/kernel/head.S +@@ -121,6 +121,8 @@ + #endif + .endm + ++ j kernel_entry ++ nop + #ifndef CONFIG_NO_EXCEPT_FILL + /* + * Reserved space for exception handlers. diff --git a/target/linux/generic-2.6/patches-2.6.31/020-mips_multi_machine_support.patch b/target/linux/generic-2.6/patches-2.6.31/020-mips_multi_machine_support.patch new file mode 100644 index 0000000000..f36ae38557 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/020-mips_multi_machine_support.patch @@ -0,0 +1,168 @@ +--- /dev/null ++++ b/include/asm-mips/mips_machine.h +@@ -0,0 +1,46 @@ ++/* ++ * Copyright (C) 2008-2009 Gabor Juhos ++ * ++ * This program is free software; you can redistribute it and/or modify it ++ * under the terms of the GNU General Public License version 2 as published ++ * by the Free Software Foundation. ++ * ++ */ ++ ++#ifndef __ASM_MIPS_MACHINE_H ++#define __ASM_MIPS_MACHINE_H ++ ++#include ++#include ++ ++struct mips_machine { ++ unsigned long mach_type; ++ void (*mach_setup)(void); ++ char *mach_name; ++ struct list_head list; ++}; ++ ++void mips_machine_register(struct mips_machine *) __init; ++void mips_machine_setup(unsigned long machtype) __init; ++ ++extern char *mips_machine_name; ++ ++#define MIPS_MACHINE(_type, _name, _setup) \ ++static char machine_name_##_type[] __initdata = _name; \ ++static struct mips_machine machine_##_type __initdata = \ ++{ \ ++ .mach_type = _type, \ ++ .mach_name = machine_name_##_type, \ ++ .mach_setup = _setup, \ ++}; \ ++ \ ++static int __init register_machine_##_type(void) \ ++{ \ ++ mips_machine_register(&machine_##_type); \ ++ return 0; \ ++} \ ++ \ ++pure_initcall(register_machine_##_type) ++ ++#endif /* __ASM_MIPS_MACHINE_H */ ++ +--- /dev/null ++++ b/arch/mips/kernel/mips_machine.c +@@ -0,0 +1,70 @@ ++/* ++ * Copyright (C) 2008-2009 Gabor Juhos ++ * ++ * This program is free software; you can redistribute it and/or modify it ++ * under the terms of the GNU General Public License version 2 as published ++ * by the Free Software Foundation. ++ * ++ */ ++#include ++ ++#include ++#include ++ ++static struct list_head mips_machines __initdata = ++ LIST_HEAD_INIT(mips_machines); ++ ++char *mips_machine_name = "Unknown"; ++ ++static struct mips_machine * __init mips_machine_find(unsigned long machtype) ++{ ++ struct list_head *this; ++ ++ list_for_each(this, &mips_machines) { ++ struct mips_machine *mach; ++ ++ mach = list_entry(this, struct mips_machine, list); ++ if (mach->mach_type == machtype) ++ return mach; ++ } ++ ++ return NULL; ++} ++ ++void __init mips_machine_register(struct mips_machine *mach) ++{ ++ list_add_tail(&mach->list, &mips_machines); ++} ++ ++void __init mips_machine_setup(unsigned long machtype) ++{ ++ struct mips_machine *mach; ++ ++ mach = mips_machine_find(machtype); ++ if (!mach) { ++ printk(KERN_ALERT "MIPS: no machine registered for " ++ "machtype %lu\n", machtype); ++ return; ++ } ++ ++ if (mach->mach_name) { ++ char *name; ++ unsigned int len; ++ ++ len = strlen(mach->mach_name); ++ name = kmalloc(len + 1, GFP_KERNEL); ++ if (name) { ++ strncpy(name, mach->mach_name,len); ++ name[len] = '\0'; ++ mips_machine_name = name; ++ } else { ++ printk(KERN_WARNING "MIPS: no memory for machine_name\n"); ++ } ++ } ++ ++ printk(KERN_INFO "MIPS: machine is %s\n", mips_machine_name); ++ ++ if (mach->mach_setup) ++ mach->mach_setup(); ++} ++ +--- a/arch/mips/kernel/Makefile ++++ b/arch/mips/kernel/Makefile +@@ -85,6 +85,7 @@ obj-$(CONFIG_GPIO_TXX9) += gpio_txx9.o + + obj-$(CONFIG_KEXEC) += machine_kexec.o relocate_kernel.o + obj-$(CONFIG_EARLY_PRINTK) += early_printk.o ++obj-$(CONFIG_MIPS_MACHINE) += mips_machine.o + + CFLAGS_cpu-bugs64.o = $(shell if $(CC) $(KBUILD_CFLAGS) -Wa,-mdaddi -c -o /dev/null -xc /dev/null >/dev/null 2>&1; then echo "-DHAVE_AS_SET_DADDI"; fi) + +--- a/arch/mips/Kconfig ++++ b/arch/mips/Kconfig +@@ -837,6 +837,9 @@ config MIPS_DISABLE_OBSOLETE_IDE + config SYNC_R4K + bool + ++config MIPS_MACHINE ++ def_bool n ++ + config NO_IOPORT + def_bool n + +--- a/arch/mips/kernel/proc.c ++++ b/arch/mips/kernel/proc.c +@@ -12,6 +12,7 @@ + #include + #include + #include ++#include + + unsigned int vced_count, vcei_count; + +@@ -31,8 +32,12 @@ static int show_cpuinfo(struct seq_file + /* + * For the first processor also print the system type + */ +- if (n == 0) ++ if (n == 0) { + seq_printf(m, "system type\t\t: %s\n", get_system_type()); ++#ifdef CONFIG_MIPS_MACHINE ++ seq_printf(m, "machine\t\t\t: %s\n", mips_machine_name); ++#endif ++ } + + seq_printf(m, "processor\t\t: %ld\n", n); + sprintf(fmt, "cpu model\t\t: %%s V%%d.%%d%s\n", diff --git a/target/linux/generic-2.6/patches-2.6.31/021-mips_simple_prom_emulator.patch b/target/linux/generic-2.6/patches-2.6.31/021-mips_simple_prom_emulator.patch new file mode 100644 index 0000000000..b53ca3f505 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/021-mips_simple_prom_emulator.patch @@ -0,0 +1,50 @@ +--- a/arch/mips/Kconfig ++++ b/arch/mips/Kconfig +@@ -839,6 +839,9 @@ config SYNC_R4K + + config MIPS_MACHINE + def_bool n ++ ++config PROM_EMU ++ def_bool n + + config NO_IOPORT + def_bool n +--- a/arch/mips/kernel/head.S ++++ b/arch/mips/kernel/head.S +@@ -143,6 +143,15 @@ FEXPORT(__kernel_entry) + j kernel_entry + #endif + ++#ifdef CONFIG_PROM_EMU ++EXPORT(prom_emu_argv) ++ .word 0 ++ .word prom_emu_cmdline ++ .ascii "CMDLINE:" ++EXPORT(prom_emu_cmdline) ++ .fill 0x400 ++#endif ++ + __REF + + NESTED(kernel_entry, 16, sp) # kernel entry point +@@ -183,6 +192,19 @@ NESTED(kernel_entry, 16, sp) # kernel + LONG_S zero, (t0) + bne t0, t1, 1b + ++#ifdef CONFIG_PROM_EMU ++ PTR_LA t0, prom_emu_cmdline ++ LONG_L t1, 0(t0) ++ beqz t1, 1f ++ ++ li a0, 2 ++ PTR_LA a1, prom_emu_argv ++ move a2, zero ++ move a3, zero ++ ++1: ++#endif /* CONFIG_PROM_EMU */ ++ + LONG_S a0, fw_arg0 # firmware arguments + LONG_S a1, fw_arg1 + LONG_S a2, fw_arg2 diff --git a/target/linux/generic-2.6/patches-2.6.31/025-mips_disable_fpu.patch b/target/linux/generic-2.6/patches-2.6.31/025-mips_disable_fpu.patch new file mode 100644 index 0000000000..98a321093d --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/025-mips_disable_fpu.patch @@ -0,0 +1,153 @@ +MIPS: allow disabling the kernel FPU emulator + +This patch allows turning off the in-kernel Algorithmics +FPU emulator support, which allows one to save a couple of +precious blocks on an embedded system. + +Signed-off-by: Florian Fainelli +-- +--- a/arch/mips/Kconfig ++++ b/arch/mips/Kconfig +@@ -825,6 +825,17 @@ config I8259 + config MIPS_BONITO64 + bool + ++config MIPS_FPU_EMU ++ bool ++ default n ++ help ++ This option allows building a kernel with or without the Algorithmics ++ FPU emulator enabled. Turning off this option results in a kernel which ++ does not catch floating operations exceptions. Make sure that your toolchain ++ is configured to enable software floating point emulation in that case. ++ ++ If unsure say Y here. ++ + config MIPS_MSC + bool + +--- a/arch/mips/math-emu/Makefile ++++ b/arch/mips/math-emu/Makefile +@@ -2,12 +2,14 @@ + # Makefile for the Linux/MIPS kernel FPU emulation. + # + +-obj-y := cp1emu.o ieee754m.o ieee754d.o ieee754dp.o ieee754sp.o ieee754.o \ ++obj-y := kernel_linkage.o dsemul.o cp1emu.o ++ ++obj-$(CONFIG_MIPS_FPU_EMU) += ieee754m.o ieee754d.o ieee754dp.o ieee754sp.o ieee754.o \ + ieee754xcpt.o dp_frexp.o dp_modf.o dp_div.o dp_mul.o dp_sub.o \ + dp_add.o dp_fsp.o dp_cmp.o dp_logb.o dp_scalb.o dp_simple.o \ + dp_tint.o dp_fint.o dp_tlong.o dp_flong.o sp_frexp.o sp_modf.o \ + sp_div.o sp_mul.o sp_sub.o sp_add.o sp_fdp.o sp_cmp.o sp_logb.o \ + sp_scalb.o sp_simple.o sp_tint.o sp_fint.o sp_tlong.o sp_flong.o \ +- dp_sqrt.o sp_sqrt.o kernel_linkage.o dsemul.o ++ dp_sqrt.o sp_sqrt.o + + EXTRA_CFLAGS += -Werror +--- a/arch/mips/math-emu/cp1emu.c ++++ b/arch/mips/math-emu/cp1emu.c +@@ -56,6 +56,12 @@ + #endif + #define __mips 4 + ++/* Further private data for which no space exists in mips_fpu_struct */ ++ ++struct mips_fpu_emulator_stats fpuemustats; ++ ++#ifdef CONFIG_MIPS_FPU_EMU ++ + /* Function which emulates a floating point instruction. */ + + static int fpu_emu(struct pt_regs *, struct mips_fpu_struct *, +@@ -66,10 +72,6 @@ static int fpux_emu(struct pt_regs *, + struct mips_fpu_struct *, mips_instruction); + #endif + +-/* Further private data for which no space exists in mips_fpu_struct */ +- +-struct mips_fpu_emulator_stats fpuemustats; +- + /* Control registers */ + + #define FPCREG_RID 0 /* $0 = revision id */ +@@ -1273,6 +1275,13 @@ int fpu_emulator_cop1Handler(struct pt_r + + return sig; + } ++#else ++int fpu_emulator_cop1Handler(struct pt_regs *xcp, struct mips_fpu_struct *ctx, ++ int has_fpu) ++{ ++ return 0; ++} ++#endif /* CONFIG_MIPS_FPU_EMU */ + + #ifdef CONFIG_DEBUG_FS + extern struct dentry *mips_debugfs_dir; +--- a/arch/mips/math-emu/dsemul.c ++++ b/arch/mips/math-emu/dsemul.c +@@ -109,6 +109,7 @@ int mips_dsemul(struct pt_regs *regs, mi + return SIGILL; /* force out of emulation loop */ + } + ++#ifdef CONFIG_MIPS_FPU_EMU + int do_dsemulret(struct pt_regs *xcp) + { + struct emuframe __user *fr; +@@ -165,3 +166,9 @@ int do_dsemulret(struct pt_regs *xcp) + + return 1; + } ++#else ++int do_dsemulret(struct pt_regs *xcp) ++{ ++ return 0; ++} ++#endif /* CONFIG_MIPS_FPU_EMU */ +--- a/arch/mips/math-emu/kernel_linkage.c ++++ b/arch/mips/math-emu/kernel_linkage.c +@@ -29,6 +29,7 @@ + + #define SIGNALLING_NAN 0x7ff800007ff80000LL + ++#ifdef CONFIG_MIPS_FPU_EMU + void fpu_emulator_init_fpu(void) + { + static int first = 1; +@@ -112,4 +113,34 @@ int fpu_emulator_restore_context32(struc + + return err; + } +-#endif ++#endif /* CONFIG_64BIT */ ++#else ++ ++void fpu_emulator_init_fpu(void) ++{ ++ return; ++} ++ ++int fpu_emulator_save_context(struct sigcontext __user *sc) ++{ ++ return 0; ++} ++ ++int fpu_emulator_restore_context(struct sigcontext __user *sc) ++{ ++ return 0; ++} ++ ++int fpu_emulator_save_context32(struct sigcontext32 __user *sc) ++{ ++ return 0; ++} ++ ++int fpu_emulator_restore_context32(struct sigcontext32 __user *sc) ++{ ++ return 0; ++} ++ ++#ifdef CONFIG_64BIT ++#endif /* CONFIG_64BIT */ ++#endif /* CONFIG_MIPS_FPU_EMU */ diff --git a/target/linux/generic-2.6/patches-2.6.31/027-mips_module_reloc.patch b/target/linux/generic-2.6/patches-2.6.31/027-mips_module_reloc.patch new file mode 100644 index 0000000000..f65a09c94e --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/027-mips_module_reloc.patch @@ -0,0 +1,331 @@ +--- a/arch/mips/Makefile ++++ b/arch/mips/Makefile +@@ -83,7 +83,7 @@ all-$(CONFIG_BOOT_ELF64) := $(vmlinux-64 + cflags-y += -G 0 -mno-abicalls -fno-pic -pipe + cflags-y += -msoft-float + LDFLAGS_vmlinux += -G 0 -static -n -nostdlib +-MODFLAGS += -mlong-calls ++MODFLAGS += -mno-long-calls + + cflags-y += -ffreestanding + +--- a/arch/mips/include/asm/module.h ++++ b/arch/mips/include/asm/module.h +@@ -9,6 +9,11 @@ struct mod_arch_specific { + struct list_head dbe_list; + const struct exception_table_entry *dbe_start; + const struct exception_table_entry *dbe_end; ++ ++ void *plt_tbl; ++ unsigned int core_plt_offset; ++ unsigned int core_plt_size; ++ unsigned int init_plt_offset; + }; + + typedef uint8_t Elf64_Byte; /* Type for a 8-bit quantity. */ +--- a/arch/mips/kernel/module.c ++++ b/arch/mips/kernel/module.c +@@ -43,6 +43,117 @@ static struct mips_hi16 *mips_hi16_list; + static LIST_HEAD(dbe_list); + static DEFINE_SPINLOCK(dbe_lock); + ++/* ++ * Get the potential max trampolines size required of the init and ++ * non-init sections. Only used if we cannot find enough contiguous ++ * physically mapped memory to put the module into. ++ */ ++static unsigned int ++get_plt_size(const Elf_Ehdr *hdr, const Elf_Shdr *sechdrs, ++ const char *secstrings, unsigned int symindex, bool is_init) ++{ ++ unsigned long ret = 0; ++ unsigned int i, j; ++ Elf_Sym *syms; ++ ++ /* Everything marked ALLOC (this includes the exported symbols) */ ++ for (i = 1; i < hdr->e_shnum; ++i) { ++ unsigned int info = sechdrs[i].sh_info; ++ ++ if (sechdrs[i].sh_type != SHT_REL ++ && sechdrs[i].sh_type != SHT_RELA) ++ continue; ++ ++ /* Not a valid relocation section? */ ++ if (info >= hdr->e_shnum) ++ continue; ++ ++ /* Don't bother with non-allocated sections */ ++ if (!(sechdrs[info].sh_flags & SHF_ALLOC)) ++ continue; ++ ++ /* If it's called *.init*, and we're not init, we're ++ not interested */ ++ if ((strstr(secstrings + sechdrs[i].sh_name, ".init") != 0) ++ != is_init) ++ continue; ++ ++ syms = (Elf_Sym *) sechdrs[symindex].sh_addr; ++ if (sechdrs[i].sh_type == SHT_REL) { ++ Elf_Mips_Rel *rel = (void *) sechdrs[i].sh_addr; ++ unsigned int size = sechdrs[i].sh_size / sizeof(*rel); ++ ++ for (j = 0; j < size; ++j) { ++ Elf_Sym *sym; ++ ++ if (ELF_MIPS_R_TYPE(rel[j]) != R_MIPS_26) ++ continue; ++ ++ sym = syms + ELF_MIPS_R_SYM(rel[j]); ++ if (!is_init && sym->st_shndx != SHN_UNDEF) ++ continue; ++ ++ ret += 4 * sizeof(int); ++ } ++ } else { ++ Elf_Mips_Rela *rela = (void *) sechdrs[i].sh_addr; ++ unsigned int size = sechdrs[i].sh_size / sizeof(*rela); ++ ++ for (j = 0; j < size; ++j) { ++ Elf_Sym *sym; ++ ++ if (ELF_MIPS_R_TYPE(rela[j]) != R_MIPS_26) ++ continue; ++ ++ sym = syms + ELF_MIPS_R_SYM(rela[j]); ++ if (!is_init && sym->st_shndx != SHN_UNDEF) ++ continue; ++ ++ ret += 4 * sizeof(int); ++ } ++ } ++ } ++ ++ return ret; ++} ++ ++#ifndef MODULE_START ++static void *alloc_phys(unsigned long size) ++{ ++ unsigned order; ++ struct page *page; ++ struct page *p; ++ ++ size = PAGE_ALIGN(size); ++ order = get_order(size); ++ ++ page = alloc_pages(GFP_KERNEL | __GFP_NORETRY | __GFP_NOWARN | ++ __GFP_THISNODE, order); ++ if (!page) ++ return NULL; ++ ++ split_page(page, order); ++ ++ for (p = page + (size >> PAGE_SHIFT); p < page + (1 << order); ++p) ++ __free_page(p); ++ ++ return page_address(page); ++} ++#endif ++ ++static void free_phys(void *ptr, unsigned long size) ++{ ++ struct page *page; ++ struct page *end; ++ ++ page = virt_to_page(ptr); ++ end = page + (PAGE_ALIGN(size) >> PAGE_SHIFT); ++ ++ for (; page < end; ++page) ++ __free_page(page); ++} ++ ++ + void *module_alloc(unsigned long size) + { + #ifdef MODULE_START +@@ -58,21 +169,68 @@ void *module_alloc(unsigned long size) + + return __vmalloc_area(area, GFP_KERNEL, PAGE_KERNEL); + #else ++ void *ptr; ++ + if (size == 0) + return NULL; +- return vmalloc(size); ++ ++ ptr = alloc_phys(size); ++ ++ /* If we failed to allocate physically contiguous memory, ++ * fall back to regular vmalloc. The module loader code will ++ * create jump tables to handle long jumps */ ++ if (!ptr) ++ return vmalloc(size); ++ ++ return ptr; ++#endif ++} ++ ++static inline bool is_phys_addr(void *ptr) ++{ ++#ifdef CONFIG_64BIT ++ return (KSEGX((unsigned long)ptr) == CKSEG0); ++#else ++ return (KSEGX(ptr) == KSEG0); + #endif + } + + /* Free memory returned from module_alloc */ + void module_free(struct module *mod, void *module_region) + { +- vfree(module_region); ++ if (is_phys_addr(module_region)) { ++ if (mod->module_init == module_region) ++ free_phys(module_region, mod->init_size); ++ else if (mod->module_core == module_region) ++ free_phys(module_region, mod->core_size); ++ else ++ BUG(); ++ } else { ++ vfree(module_region); ++ } + } + + int module_frob_arch_sections(Elf_Ehdr *hdr, Elf_Shdr *sechdrs, + char *secstrings, struct module *mod) + { ++ unsigned int symindex = 0; ++ unsigned int core_size, init_size; ++ int i; ++ ++ for (i = 1; i < hdr->e_shnum; i++) ++ if (sechdrs[i].sh_type == SHT_SYMTAB) ++ symindex = i; ++ ++ core_size = get_plt_size(hdr, sechdrs, secstrings, symindex, false); ++ init_size = get_plt_size(hdr, sechdrs, secstrings, symindex, true); ++ ++ mod->arch.core_plt_offset = 0; ++ mod->arch.core_plt_size = core_size; ++ mod->arch.init_plt_offset = core_size; ++ mod->arch.plt_tbl = kmalloc(core_size + init_size, GFP_KERNEL); ++ if (!mod->arch.plt_tbl) ++ return -ENOMEM; ++ + return 0; + } + +@@ -95,28 +253,40 @@ static int apply_r_mips_32_rela(struct m + return 0; + } + +-static int apply_r_mips_26_rel(struct module *me, u32 *location, Elf_Addr v) ++static Elf_Addr add_plt_entry_to(unsigned *plt_offset, ++ void *start, Elf_Addr v) + { +- if (v % 4) { +- pr_err("module %s: dangerous R_MIPS_26 REL relocation\n", +- me->name); +- return -ENOEXEC; +- } ++ unsigned *tramp = start + *plt_offset; ++ *plt_offset += 4 * sizeof(int); + +- if ((v & 0xf0000000) != (((unsigned long)location + 4) & 0xf0000000)) { +- printk(KERN_ERR +- "module %s: relocation overflow\n", +- me->name); +- return -ENOEXEC; +- } ++ /* adjust carry for addiu */ ++ if (v & 0x00008000) ++ v += 0x10000; ++ ++ tramp[0] = 0x3c190000 | (v >> 16); /* lui t9, hi16 */ ++ tramp[1] = 0x27390000 | (v & 0xffff); /* addiu t9, t9, lo16 */ ++ tramp[2] = 0x03200008; /* jr t9 */ ++ tramp[3] = 0x00000000; /* nop */ + +- *location = (*location & ~0x03ffffff) | +- ((*location + (v >> 2)) & 0x03ffffff); ++ return (Elf_Addr) tramp; ++} ++ ++static Elf_Addr add_plt_entry(struct module *me, void *location, Elf_Addr v) ++{ ++ if (location >= me->module_core && ++ location < me->module_core + me->core_size) ++ return add_plt_entry_to(&me->arch.core_plt_offset, ++ me->arch.plt_tbl, v); ++ ++ if (location >= me->module_init && ++ location < me->module_init + me->init_size) ++ return add_plt_entry_to(&me->arch.init_plt_offset, ++ me->arch.plt_tbl, v); + + return 0; + } + +-static int apply_r_mips_26_rela(struct module *me, u32 *location, Elf_Addr v) ++static int set_r_mips_26(struct module *me, u32 *location, u32 ofs, Elf_Addr v) + { + if (v % 4) { + pr_err("module %s: dangerous R_MIPS_26 RELArelocation\n", +@@ -125,17 +295,31 @@ static int apply_r_mips_26_rela(struct m + } + + if ((v & 0xf0000000) != (((unsigned long)location + 4) & 0xf0000000)) { +- printk(KERN_ERR ++ v = add_plt_entry(me, location, v + (ofs << 2)); ++ if (!v) { ++ printk(KERN_ERR + "module %s: relocation overflow\n", + me->name); +- return -ENOEXEC; ++ return -ENOEXEC; ++ } ++ ofs = 0; + } + +- *location = (*location & ~0x03ffffff) | ((v >> 2) & 0x03ffffff); ++ *location = (*location & ~0x03ffffff) | ((ofs + (v >> 2)) & 0x03ffffff); + + return 0; + } + ++static int apply_r_mips_26_rel(struct module *me, u32 *location, Elf_Addr v) ++{ ++ return set_r_mips_26(me, location, *location & 0x03ffffff, v); ++} ++ ++static int apply_r_mips_26_rela(struct module *me, u32 *location, Elf_Addr v) ++{ ++ return set_r_mips_26(me, location, 0, v); ++} ++ + static int apply_r_mips_hi16_rel(struct module *me, u32 *location, Elf_Addr v) + { + struct mips_hi16 *n; +@@ -400,11 +584,23 @@ int module_finalize(const Elf_Ehdr *hdr, + list_add(&me->arch.dbe_list, &dbe_list); + spin_unlock_irq(&dbe_lock); + } ++ ++ /* Get rid of the fixup trampoline if we're running the module ++ * from physically mapped address space */ ++ if (me->arch.core_plt_offset == 0 && ++ me->arch.init_plt_offset == me->arch.core_plt_size && ++ is_phys_addr(me->module_core)) { ++ kfree(me->arch.plt_tbl); ++ me->arch.plt_tbl = NULL; ++ } ++ + return 0; + } + + void module_arch_cleanup(struct module *mod) + { ++ if (mod->arch.plt_tbl) ++ kfree(mod->arch.plt_tbl); + spin_lock_irq(&dbe_lock); + list_del(&mod->arch.dbe_list); + spin_unlock_irq(&dbe_lock); diff --git a/target/linux/generic-2.6/patches-2.6.31/028-module_exports.patch b/target/linux/generic-2.6/patches-2.6.31/028-module_exports.patch new file mode 100644 index 0000000000..1af1a50367 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/028-module_exports.patch @@ -0,0 +1,212 @@ +--- a/include/asm-generic/vmlinux.lds.h ++++ b/include/asm-generic/vmlinux.lds.h +@@ -55,6 +55,27 @@ + #define LOAD_OFFSET 0 + #endif + ++#ifndef SYMTAB_KEEP_STR ++#define SYMTAB_KEEP_STR *(__ksymtab_strings.*) ++#define SYMTAB_DISCARD_STR ++#else ++#define SYMTAB_DISCARD_STR *(__ksymtab_strings.*) ++#endif ++ ++#ifndef SYMTAB_KEEP ++#define SYMTAB_KEEP *(__ksymtab.*) ++#define SYMTAB_DISCARD ++#else ++#define SYMTAB_DISCARD *(__ksymtab.*) ++#endif ++ ++#ifndef SYMTAB_KEEP_GPL ++#define SYMTAB_KEEP_GPL *(__ksymtab_gpl.*) ++#define SYMTAB_DISCARD_GPL ++#else ++#define SYMTAB_DISCARD_GPL *(__ksymtab_gpl.*) ++#endif ++ + #ifndef VMLINUX_SYMBOL + #define VMLINUX_SYMBOL(_sym_) _sym_ + #endif +@@ -256,35 +277,35 @@ + /* Kernel symbol table: Normal symbols */ \ + __ksymtab : AT(ADDR(__ksymtab) - LOAD_OFFSET) { \ + VMLINUX_SYMBOL(__start___ksymtab) = .; \ +- *(__ksymtab) \ ++ SYMTAB_KEEP \ + VMLINUX_SYMBOL(__stop___ksymtab) = .; \ + } \ + \ + /* Kernel symbol table: GPL-only symbols */ \ + __ksymtab_gpl : AT(ADDR(__ksymtab_gpl) - LOAD_OFFSET) { \ + VMLINUX_SYMBOL(__start___ksymtab_gpl) = .; \ +- *(__ksymtab_gpl) \ ++ SYMTAB_KEEP_GPL \ + VMLINUX_SYMBOL(__stop___ksymtab_gpl) = .; \ + } \ + \ + /* Kernel symbol table: Normal unused symbols */ \ + __ksymtab_unused : AT(ADDR(__ksymtab_unused) - LOAD_OFFSET) { \ + VMLINUX_SYMBOL(__start___ksymtab_unused) = .; \ +- *(__ksymtab_unused) \ ++ *(__ksymtab_unused.*) \ + VMLINUX_SYMBOL(__stop___ksymtab_unused) = .; \ + } \ + \ + /* Kernel symbol table: GPL-only unused symbols */ \ + __ksymtab_unused_gpl : AT(ADDR(__ksymtab_unused_gpl) - LOAD_OFFSET) { \ + VMLINUX_SYMBOL(__start___ksymtab_unused_gpl) = .; \ +- *(__ksymtab_unused_gpl) \ ++ *(__ksymtab_unused_gpl.*) \ + VMLINUX_SYMBOL(__stop___ksymtab_unused_gpl) = .; \ + } \ + \ + /* Kernel symbol table: GPL-future-only symbols */ \ + __ksymtab_gpl_future : AT(ADDR(__ksymtab_gpl_future) - LOAD_OFFSET) { \ + VMLINUX_SYMBOL(__start___ksymtab_gpl_future) = .; \ +- *(__ksymtab_gpl_future) \ ++ *(__ksymtab_gpl_future.*) \ + VMLINUX_SYMBOL(__stop___ksymtab_gpl_future) = .; \ + } \ + \ +@@ -325,7 +346,13 @@ + \ + /* Kernel symbol table: strings */ \ + __ksymtab_strings : AT(ADDR(__ksymtab_strings) - LOAD_OFFSET) { \ +- *(__ksymtab_strings) \ ++ SYMTAB_KEEP_STR \ ++ } \ ++ \ ++ /DISCARD/ : { \ ++ SYMTAB_DISCARD \ ++ SYMTAB_DISCARD_GPL \ ++ SYMTAB_DISCARD_STR \ + } \ + \ + /* __*init sections */ \ +--- a/include/linux/module.h ++++ b/include/linux/module.h +@@ -188,16 +188,24 @@ void *__symbol_get_gpl(const char *symbo + #define __CRC_SYMBOL(sym, sec) + #endif + ++#ifdef MODULE ++#define __EXPORT_SUFFIX(sym) ++#else ++#define __EXPORT_SUFFIX(sym) "." #sym ++#endif ++ + /* For every exported symbol, place a struct in the __ksymtab section */ + #define __EXPORT_SYMBOL(sym, sec) \ + extern typeof(sym) sym; \ + __CRC_SYMBOL(sym, sec) \ + static const char __kstrtab_##sym[] \ +- __attribute__((section("__ksymtab_strings"), aligned(1))) \ ++ __attribute__((section("__ksymtab_strings" \ ++ __EXPORT_SUFFIX(sym)), aligned(1))) \ + = MODULE_SYMBOL_PREFIX #sym; \ + static const struct kernel_symbol __ksymtab_##sym \ + __used \ +- __attribute__((section("__ksymtab" sec), unused)) \ ++ __attribute__((section("__ksymtab" sec \ ++ __EXPORT_SUFFIX(sym)), unused)) \ + = { (unsigned long)&sym, __kstrtab_##sym } + + #define EXPORT_SYMBOL(sym) \ +--- a/Makefile ++++ b/Makefile +@@ -986,7 +986,7 @@ prepare: prepare0 + # Leave this as default for preprocessing vmlinux.lds.S, which is now + # done in arch/$(ARCH)/kernel/Makefile + +-export CPPFLAGS_vmlinux.lds += -P -C -U$(ARCH) ++export CPPFLAGS_vmlinux.lds += -P -C -U$(ARCH) $(EXTRA_LDSFLAGS) + + # The asm symlink changes when $(ARCH) changes. + # Detect this and ask user to run make mrproper +--- a/arch/arm/kernel/vmlinux.lds.S ++++ b/arch/arm/kernel/vmlinux.lds.S +@@ -79,26 +79,6 @@ SECTIONS + #endif + } + +- /DISCARD/ : { /* Exit code and data */ +- EXIT_TEXT +- EXIT_DATA +- *(.exitcall.exit) +- *(.ARM.exidx.exit.text) +- *(.ARM.extab.exit.text) +-#ifndef CONFIG_HOTPLUG_CPU +- *(.ARM.exidx.cpuexit.text) +- *(.ARM.extab.cpuexit.text) +-#endif +-#ifndef CONFIG_HOTPLUG +- *(.ARM.exidx.devexit.text) +- *(.ARM.extab.devexit.text) +-#endif +-#ifndef CONFIG_MMU +- *(.fixup) +- *(__ex_table) +-#endif +- } +- + .text : { /* Real text segment */ + _text = .; /* Text and read-only data */ + __exception_text_start = .; +@@ -205,6 +185,28 @@ SECTIONS + __bss_stop = .; + _end = .; + } ++ ++ /DISCARD/ : { /* Exit code and data */ ++ EXIT_TEXT ++ EXIT_DATA ++ *(.discard) ++ *(.exitcall.exit) ++ *(.ARM.exidx.exit.text) ++ *(.ARM.extab.exit.text) ++#ifndef CONFIG_HOTPLUG_CPU ++ *(.ARM.exidx.cpuexit.text) ++ *(.ARM.extab.cpuexit.text) ++#endif ++#ifndef CONFIG_HOTPLUG ++ *(.ARM.exidx.devexit.text) ++ *(.ARM.extab.devexit.text) ++#endif ++#ifndef CONFIG_MMU ++ *(.fixup) ++ *(__ex_table) ++#endif ++ } ++ + /* Stabs debugging sections. */ + .stab 0 : { *(.stab) } + .stabstr 0 : { *(.stabstr) } +--- a/arch/powerpc/kernel/vmlinux.lds.S ++++ b/arch/powerpc/kernel/vmlinux.lds.S +@@ -37,12 +37,6 @@ jiffies = jiffies_64 + 4; + #endif + SECTIONS + { +- /* Sections to be discarded. */ +- /DISCARD/ : { +- *(.exitcall.exit) +- EXIT_DATA +- } +- + . = KERNELBASE; + + /* +@@ -295,6 +289,12 @@ SECTIONS + __bss_stop = .; + } + ++ /* Sections to be discarded. */ ++ /DISCARD/ : { ++ *(.exitcall.exit) ++ EXIT_DATA ++ } ++ + . = ALIGN(PAGE_SIZE); + _end = . ; + PROVIDE32 (end = .); diff --git a/target/linux/generic-2.6/patches-2.6.31/030-pci_disable_common_quirks.patch b/target/linux/generic-2.6/patches-2.6.31/030-pci_disable_common_quirks.patch new file mode 100644 index 0000000000..06153e15ee --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/030-pci_disable_common_quirks.patch @@ -0,0 +1,43 @@ +--- a/drivers/pci/Kconfig ++++ b/drivers/pci/Kconfig +@@ -51,6 +51,12 @@ config PCI_STUB + + When in doubt, say N. + ++config PCI_DISABLE_COMMON_QUIRKS ++ bool "PCI disable common quirks" ++ depends on PCI ++ help ++ If you don't know what to do here, say N. ++ + config HT_IRQ + bool "Interrupts on hypertransport devices" + default y +--- a/drivers/pci/quirks.c ++++ b/drivers/pci/quirks.c +@@ -98,6 +98,7 @@ static void __devinit quirk_resource_ali + } + DECLARE_PCI_FIXUP_HEADER(PCI_ANY_ID, PCI_ANY_ID, quirk_resource_alignment); + ++#ifndef CONFIG_PCI_DISABLE_COMMON_QUIRKS + /* The Mellanox Tavor device gives false positive parity errors + * Mark this device with a broken_parity_status, to allow + * PCI scanning code to "skip" this now blacklisted device. +@@ -1866,7 +1867,9 @@ static void __devinit fixup_rev1_53c810( + } + } + DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_NCR, PCI_DEVICE_ID_NCR_53C810, fixup_rev1_53c810); ++#endif /* !CONFIG_PCI_DISABLE_COMMON_QUIRKS */ + ++#ifndef CONFIG_PCI_DISABLE_COMMON_QUIRKS + /* Enable 1k I/O space granularity on the Intel P64H2 */ + static void __devinit quirk_p64h2_1k_io(struct pci_dev *dev) + { +@@ -2494,6 +2497,7 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_I + DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, 0x150a, quirk_i82576_sriov); + + #endif /* CONFIG_PCI_IOV */ ++#endif /* !CONFIG_PCI_DISABLE_COMMON_QUIRKS */ + + static void pci_do_fixups(struct pci_dev *dev, struct pci_fixup *f, + struct pci_fixup *end) diff --git a/target/linux/generic-2.6/patches-2.6.31/051-squashfs_pcomp.patch b/target/linux/generic-2.6/patches-2.6.31/051-squashfs_pcomp.patch new file mode 100644 index 0000000000..086adf651a --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/051-squashfs_pcomp.patch @@ -0,0 +1,234 @@ +--- a/fs/squashfs/Kconfig ++++ b/fs/squashfs/Kconfig +@@ -1,7 +1,8 @@ + config SQUASHFS + tristate "SquashFS 4.0 - Squashed file system support" + depends on BLOCK +- select ZLIB_INFLATE ++ select CRYPTO ++ select CRYPTO_ZLIB + help + Saying Y here includes support for SquashFS 4.0 (a Compressed + Read-Only File System). Squashfs is a highly compressed read-only +--- a/fs/squashfs/block.c ++++ b/fs/squashfs/block.c +@@ -32,7 +32,8 @@ + #include + #include + #include +-#include ++ ++#include + + #include "squashfs_fs.h" + #include "squashfs_fs_sb.h" +@@ -153,7 +154,8 @@ int squashfs_read_data(struct super_bloc + } + + if (compressed) { +- int zlib_err = 0, zlib_init = 0; ++ int res = 0, decomp_init = 0; ++ struct comp_request req; + + /* + * Uncompress block. +@@ -161,12 +163,13 @@ int squashfs_read_data(struct super_bloc + + mutex_lock(&msblk->read_data_mutex); + +- msblk->stream.avail_out = 0; +- msblk->stream.avail_in = 0; ++ req.avail_out = 0; ++ req.avail_in = 0; + + bytes = length; ++ length = 0; + do { +- if (msblk->stream.avail_in == 0 && k < b) { ++ if (req.avail_in == 0 && k < b) { + avail = min(bytes, msblk->devblksize - offset); + bytes -= avail; + wait_on_buffer(bh[k]); +@@ -179,45 +182,47 @@ int squashfs_read_data(struct super_bloc + continue; + } + +- msblk->stream.next_in = bh[k]->b_data + offset; +- msblk->stream.avail_in = avail; ++ req.next_in = bh[k]->b_data + offset; ++ req.avail_in = avail; + offset = 0; + } + +- if (msblk->stream.avail_out == 0 && page < pages) { +- msblk->stream.next_out = buffer[page++]; +- msblk->stream.avail_out = PAGE_CACHE_SIZE; ++ if (req.avail_out == 0 && page < pages) { ++ req.next_out = buffer[page++]; ++ req.avail_out = PAGE_CACHE_SIZE; + } + +- if (!zlib_init) { +- zlib_err = zlib_inflateInit(&msblk->stream); +- if (zlib_err != Z_OK) { +- ERROR("zlib_inflateInit returned" +- " unexpected result 0x%x," +- " srclength %d\n", zlib_err, +- srclength); ++ if (!decomp_init) { ++ res = crypto_decompress_init(msblk->tfm); ++ if (res) { ++ ERROR("crypto_decompress_init " ++ "returned %d, srclength %d\n", ++ res, srclength); + goto release_mutex; + } +- zlib_init = 1; ++ decomp_init = 1; + } + +- zlib_err = zlib_inflate(&msblk->stream, Z_SYNC_FLUSH); ++ res = crypto_decompress_update(msblk->tfm, &req); ++ if (res < 0) { ++ ERROR("crypto_decompress_update returned %d, " ++ "data probably corrupt\n", res); ++ goto release_mutex; ++ } ++ length += res; + +- if (msblk->stream.avail_in == 0 && k < b) ++ if (req.avail_in == 0 && k < b) + put_bh(bh[k++]); +- } while (zlib_err == Z_OK); ++ } while (bytes || res); + +- if (zlib_err != Z_STREAM_END) { +- ERROR("zlib_inflate error, data probably corrupt\n"); ++ res = crypto_decompress_final(msblk->tfm, &req); ++ if (res < 0) { ++ ERROR("crypto_decompress_final returned %d, data " ++ "probably corrupt\n", res); + goto release_mutex; + } ++ length += res; + +- zlib_err = zlib_inflateEnd(&msblk->stream); +- if (zlib_err != Z_OK) { +- ERROR("zlib_inflate error, data probably corrupt\n"); +- goto release_mutex; +- } +- length = msblk->stream.total_out; + mutex_unlock(&msblk->read_data_mutex); + } else { + /* +--- a/fs/squashfs/squashfs_fs_sb.h ++++ b/fs/squashfs/squashfs_fs_sb.h +@@ -64,7 +64,7 @@ struct squashfs_sb_info { + struct mutex read_data_mutex; + struct mutex meta_index_mutex; + struct meta_index *meta_index; +- z_stream stream; ++ struct crypto_pcomp *tfm; + __le64 *inode_lookup_table; + u64 inode_table; + u64 directory_table; +--- a/fs/squashfs/super.c ++++ b/fs/squashfs/super.c +@@ -38,11 +38,19 @@ + #include + #include + ++#include ++ ++#include ++ + #include "squashfs_fs.h" + #include "squashfs_fs_sb.h" + #include "squashfs_fs_i.h" + #include "squashfs.h" + ++ ++#define SQUASHFS_CRYPTO_ALG "zlib" ++ ++ + static struct file_system_type squashfs_fs_type; + static struct super_operations squashfs_super_ops; + +@@ -76,6 +84,16 @@ static int squashfs_fill_super(struct su + unsigned short flags; + unsigned int fragments; + u64 lookup_table_start; ++ struct { ++ struct nlattr nla; ++ int val; ++ } params = { ++ .nla = { ++ .nla_len = nla_attr_size(sizeof(int)), ++ .nla_type = ZLIB_DECOMP_WINDOWBITS, ++ }, ++ .val = DEF_WBITS, ++ }; + int err; + + TRACE("Entered squashfs_fill_superblock\n"); +@@ -87,16 +105,25 @@ static int squashfs_fill_super(struct su + } + msblk = sb->s_fs_info; + +- msblk->stream.workspace = kmalloc(zlib_inflate_workspacesize(), +- GFP_KERNEL); +- if (msblk->stream.workspace == NULL) { +- ERROR("Failed to allocate zlib workspace\n"); ++ msblk->tfm = crypto_alloc_pcomp(SQUASHFS_CRYPTO_ALG, 0, ++ CRYPTO_ALG_ASYNC); ++ if (IS_ERR(msblk->tfm)) { ++ ERROR("Failed to load %s crypto module\n", ++ SQUASHFS_CRYPTO_ALG); ++ err = PTR_ERR(msblk->tfm); ++ goto failed_pcomp; ++ } ++ ++ err = crypto_decompress_setup(msblk->tfm, ¶ms, sizeof(params)); ++ if (err) { ++ ERROR("Failed to set up decompression parameters\n"); + goto failure; + } + + sblk = kzalloc(sizeof(*sblk), GFP_KERNEL); + if (sblk == NULL) { + ERROR("Failed to allocate squashfs_super_block\n"); ++ err = -ENOMEM; + goto failure; + } + +@@ -295,17 +322,18 @@ failed_mount: + kfree(msblk->inode_lookup_table); + kfree(msblk->fragment_index); + kfree(msblk->id_table); +- kfree(msblk->stream.workspace); ++ crypto_free_pcomp(msblk->tfm); + kfree(sb->s_fs_info); + sb->s_fs_info = NULL; + kfree(sblk); + return err; + + failure: +- kfree(msblk->stream.workspace); ++ crypto_free_pcomp(msblk->tfm); ++failed_pcomp: + kfree(sb->s_fs_info); + sb->s_fs_info = NULL; +- return -ENOMEM; ++ return err; + } + + +@@ -349,7 +377,7 @@ static void squashfs_put_super(struct su + kfree(sbi->id_table); + kfree(sbi->fragment_index); + kfree(sbi->meta_index); +- kfree(sbi->stream.workspace); ++ crypto_free_pcomp(sbi->tfm); + kfree(sb->s_fs_info); + sb->s_fs_info = NULL; + } diff --git a/target/linux/generic-2.6/patches-2.6.31/052-pcomp_lzma_support.patch b/target/linux/generic-2.6/patches-2.6.31/052-pcomp_lzma_support.patch new file mode 100644 index 0000000000..0791555162 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/052-pcomp_lzma_support.patch @@ -0,0 +1,834 @@ +--- /dev/null ++++ b/crypto/unlzma.c +@@ -0,0 +1,723 @@ ++/* ++ * LZMA uncompresion module for pcomp ++ * Copyright (C) 2009 Felix Fietkau ++ * ++ * Based on: ++ * Initial Linux kernel adaptation ++ * Copyright (C) 2006 Alain < alain@knaff.lu > ++ * ++ * Based on small lzma deflate implementation/Small range coder ++ * implementation for lzma. ++ * Copyright (C) 2006 Aurelien Jacobs < aurel@gnuage.org > ++ * ++ * Based on LzmaDecode.c from the LZMA SDK 4.22 (http://www.7-zip.org/) ++ * Copyright (C) 1999-2005 Igor Pavlov ++ * ++ * This program is free software; you can redistribute it and/or modify it ++ * under the terms of the GNU General Public License version 2 as published ++ * by the Free Software Foundation. ++ * ++ * FIXME: the current implementation assumes that the caller will ++ * not free any output buffers until the whole decompression has been ++ * completed. This is necessary, because LZMA looks back at old output ++ * instead of doing a separate dictionary allocation, which saves RAM. ++ */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++#include "unlzma.h" ++ ++static int instance = 0; ++ ++struct unlzma_buffer { ++ struct unlzma_buffer *last; ++ int offset; ++ int size; ++ u8 *ptr; ++}; ++ ++struct unlzma_ctx { ++ struct task_struct *thread; ++ wait_queue_head_t next_req; ++ struct mutex mutex; ++ bool active; ++ bool cancel; ++ ++ const u8 *next_in; ++ int avail_in; ++ ++ u8 *next_out; ++ int avail_out; ++ ++ /* reader state */ ++ u32 code; ++ u32 range; ++ u32 bound; ++ ++ /* writer state */ ++ u8 previous_byte; ++ ssize_t pos; ++ struct unlzma_buffer *head; ++ int buf_full; ++ ++ /* cstate */ ++ int state; ++ u32 rep0, rep1, rep2, rep3; ++ ++ u32 dict_size; ++ ++ void *workspace; ++ int workspace_size; ++}; ++ ++static inline bool ++unlzma_should_stop(struct unlzma_ctx *ctx) ++{ ++ return unlikely(kthread_should_stop() || ctx->cancel); ++} ++ ++static void ++get_buffer(struct unlzma_ctx *ctx) ++{ ++ struct unlzma_buffer *bh; ++ ++ bh = kzalloc(sizeof(struct unlzma_buffer), GFP_KERNEL); ++ bh->ptr = ctx->next_out; ++ bh->offset = ctx->pos; ++ bh->last = ctx->head; ++ bh->size = ctx->avail_out; ++ ctx->head = bh; ++ ctx->buf_full = 0; ++} ++ ++static void ++unlzma_request_buffer(struct unlzma_ctx *ctx, int *avail) ++{ ++ do { ++ mutex_unlock(&ctx->mutex); ++ if (wait_event_interruptible(ctx->next_req, ++ unlzma_should_stop(ctx) || (*avail > 0))) ++ schedule(); ++ mutex_lock(&ctx->mutex); ++ } while (*avail <= 0 && !unlzma_should_stop(ctx)); ++ ++ if (!unlzma_should_stop(ctx) && ctx->buf_full) ++ get_buffer(ctx); ++} ++ ++static u8 ++rc_read(struct unlzma_ctx *ctx) ++{ ++ if (unlikely(ctx->avail_in <= 0)) ++ unlzma_request_buffer(ctx, &ctx->avail_in); ++ ++ if (unlzma_should_stop(ctx)) ++ return 0; ++ ++ ctx->avail_in--; ++ return *(ctx->next_in++); ++} ++ ++ ++static inline void ++rc_get_code(struct unlzma_ctx *ctx) ++{ ++ ctx->code = (ctx->code << 8) | rc_read(ctx); ++} ++ ++static void ++rc_normalize(struct unlzma_ctx *ctx) ++{ ++ if (ctx->range < (1 << RC_TOP_BITS)) { ++ ctx->range <<= 8; ++ rc_get_code(ctx); ++ } ++} ++ ++static int ++rc_is_bit_0(struct unlzma_ctx *ctx, u16 *p) ++{ ++ rc_normalize(ctx); ++ ctx->bound = *p * (ctx->range >> RC_MODEL_TOTAL_BITS); ++ return ctx->code < ctx->bound; ++} ++ ++static void ++rc_update_bit_0(struct unlzma_ctx *ctx, u16 *p) ++{ ++ ctx->range = ctx->bound; ++ *p += ((1 << RC_MODEL_TOTAL_BITS) - *p) >> RC_MOVE_BITS; ++} ++ ++static void ++rc_update_bit_1(struct unlzma_ctx *ctx, u16 *p) ++{ ++ ctx->range -= ctx->bound; ++ ctx->code -= ctx->bound; ++ *p -= *p >> RC_MOVE_BITS; ++} ++ ++static bool ++rc_get_bit(struct unlzma_ctx *ctx, u16 *p, int *symbol) ++{ ++ if (rc_is_bit_0(ctx, p)) { ++ rc_update_bit_0(ctx, p); ++ *symbol *= 2; ++ return 0; ++ } else { ++ rc_update_bit_1(ctx, p); ++ *symbol = *symbol * 2 + 1; ++ return 1; ++ } ++} ++ ++static int ++rc_direct_bit(struct unlzma_ctx *ctx) ++{ ++ rc_normalize(ctx); ++ ctx->range >>= 1; ++ if (ctx->code >= ctx->range) { ++ ctx->code -= ctx->range; ++ return 1; ++ } ++ return 0; ++} ++ ++static void ++rc_bit_tree_decode(struct unlzma_ctx *ctx, u16 *p, int num_levels, int *symbol) ++{ ++ int i = num_levels; ++ ++ *symbol = 1; ++ while (i--) ++ rc_get_bit(ctx, p + *symbol, symbol); ++ *symbol -= 1 << num_levels; ++} ++ ++static u8 ++peek_old_byte(struct unlzma_ctx *ctx, u32 offs) ++{ ++ struct unlzma_buffer *bh = ctx->head; ++ u32 pos; ++ ++ pos = ctx->pos - offs; ++ if (pos >= ctx->dict_size) { ++ pos = (~pos % ctx->dict_size); ++ } ++ ++ while (bh->offset > pos) { ++ bh = bh->last; ++ BUG_ON(!bh); ++ } ++ ++ pos -= bh->offset; ++ BUG_ON(pos >= bh->size); ++ ++ return bh->ptr[pos]; ++} ++ ++static void ++write_byte(struct unlzma_ctx *ctx, u8 byte) ++{ ++ if (unlikely(ctx->avail_out <= 0)) { ++ unlzma_request_buffer(ctx, &ctx->avail_out); ++ } ++ ++ if (!ctx->avail_out) ++ return; ++ ++ ctx->previous_byte = byte; ++ *(ctx->next_out++) = byte; ++ ctx->avail_out--; ++ if (ctx->avail_out == 0) ++ ctx->buf_full = 1; ++ ctx->pos++; ++} ++ ++ ++static inline void ++copy_byte(struct unlzma_ctx *ctx, u32 offs) ++{ ++ write_byte(ctx, peek_old_byte(ctx, offs)); ++} ++ ++static void ++copy_bytes(struct unlzma_ctx *ctx, u32 rep0, int len) ++{ ++ do { ++ copy_byte(ctx, rep0); ++ len--; ++ if (unlzma_should_stop(ctx)) ++ break; ++ } while (len != 0); ++} ++ ++static void ++process_bit0(struct unlzma_ctx *ctx, u16 *p, int pos_state, u16 *prob, ++ int lc, u32 literal_pos_mask) ++{ ++ int mi = 1; ++ rc_update_bit_0(ctx, prob); ++ prob = (p + LZMA_LITERAL + ++ (LZMA_LIT_SIZE ++ * (((ctx->pos & literal_pos_mask) << lc) ++ + (ctx->previous_byte >> (8 - lc)))) ++ ); ++ ++ if (ctx->state >= LZMA_NUM_LIT_STATES) { ++ int match_byte = peek_old_byte(ctx, ctx->rep0); ++ do { ++ u16 bit; ++ u16 *prob_lit; ++ ++ match_byte <<= 1; ++ bit = match_byte & 0x100; ++ prob_lit = prob + 0x100 + bit + mi; ++ if (rc_get_bit(ctx, prob_lit, &mi) != !!bit) ++ break; ++ } while (mi < 0x100); ++ } ++ while (mi < 0x100) { ++ u16 *prob_lit = prob + mi; ++ rc_get_bit(ctx, prob_lit, &mi); ++ } ++ write_byte(ctx, mi); ++ if (ctx->state < 4) ++ ctx->state = 0; ++ else if (ctx->state < 10) ++ ctx->state -= 3; ++ else ++ ctx->state -= 6; ++} ++ ++static void ++process_bit1(struct unlzma_ctx *ctx, u16 *p, int pos_state, u16 *prob) ++{ ++ int offset; ++ u16 *prob_len; ++ int num_bits; ++ int len; ++ ++ rc_update_bit_1(ctx, prob); ++ prob = p + LZMA_IS_REP + ctx->state; ++ if (rc_is_bit_0(ctx, prob)) { ++ rc_update_bit_0(ctx, prob); ++ ctx->rep3 = ctx->rep2; ++ ctx->rep2 = ctx->rep1; ++ ctx->rep1 = ctx->rep0; ++ ctx->state = ctx->state < LZMA_NUM_LIT_STATES ? 0 : 3; ++ prob = p + LZMA_LEN_CODER; ++ } else { ++ rc_update_bit_1(ctx, prob); ++ prob = p + LZMA_IS_REP_G0 + ctx->state; ++ if (rc_is_bit_0(ctx, prob)) { ++ rc_update_bit_0(ctx, prob); ++ prob = (p + LZMA_IS_REP_0_LONG ++ + (ctx->state << ++ LZMA_NUM_POS_BITS_MAX) + ++ pos_state); ++ if (rc_is_bit_0(ctx, prob)) { ++ rc_update_bit_0(ctx, prob); ++ ++ ctx->state = ctx->state < LZMA_NUM_LIT_STATES ? ++ 9 : 11; ++ copy_byte(ctx, ctx->rep0); ++ return; ++ } else { ++ rc_update_bit_1(ctx, prob); ++ } ++ } else { ++ u32 distance; ++ ++ rc_update_bit_1(ctx, prob); ++ prob = p + LZMA_IS_REP_G1 + ctx->state; ++ if (rc_is_bit_0(ctx, prob)) { ++ rc_update_bit_0(ctx, prob); ++ distance = ctx->rep1; ++ } else { ++ rc_update_bit_1(ctx, prob); ++ prob = p + LZMA_IS_REP_G2 + ctx->state; ++ if (rc_is_bit_0(ctx, prob)) { ++ rc_update_bit_0(ctx, prob); ++ distance = ctx->rep2; ++ } else { ++ rc_update_bit_1(ctx, prob); ++ distance = ctx->rep3; ++ ctx->rep3 = ctx->rep2; ++ } ++ ctx->rep2 = ctx->rep1; ++ } ++ ctx->rep1 = ctx->rep0; ++ ctx->rep0 = distance; ++ } ++ ctx->state = ctx->state < LZMA_NUM_LIT_STATES ? 8 : 11; ++ prob = p + LZMA_REP_LEN_CODER; ++ } ++ ++ prob_len = prob + LZMA_LEN_CHOICE; ++ if (rc_is_bit_0(ctx, prob_len)) { ++ rc_update_bit_0(ctx, prob_len); ++ prob_len = (prob + LZMA_LEN_LOW ++ + (pos_state << ++ LZMA_LEN_NUM_LOW_BITS)); ++ offset = 0; ++ num_bits = LZMA_LEN_NUM_LOW_BITS; ++ } else { ++ rc_update_bit_1(ctx, prob_len); ++ prob_len = prob + LZMA_LEN_CHOICE_2; ++ if (rc_is_bit_0(ctx, prob_len)) { ++ rc_update_bit_0(ctx, prob_len); ++ prob_len = (prob + LZMA_LEN_MID ++ + (pos_state << ++ LZMA_LEN_NUM_MID_BITS)); ++ offset = 1 << LZMA_LEN_NUM_LOW_BITS; ++ num_bits = LZMA_LEN_NUM_MID_BITS; ++ } else { ++ rc_update_bit_1(ctx, prob_len); ++ prob_len = prob + LZMA_LEN_HIGH; ++ offset = ((1 << LZMA_LEN_NUM_LOW_BITS) ++ + (1 << LZMA_LEN_NUM_MID_BITS)); ++ num_bits = LZMA_LEN_NUM_HIGH_BITS; ++ } ++ } ++ ++ rc_bit_tree_decode(ctx, prob_len, num_bits, &len); ++ len += offset; ++ ++ if (ctx->state < 4) { ++ int pos_slot; ++ ++ ctx->state += LZMA_NUM_LIT_STATES; ++ prob = ++ p + LZMA_POS_SLOT + ++ ((len < ++ LZMA_NUM_LEN_TO_POS_STATES ? len : ++ LZMA_NUM_LEN_TO_POS_STATES - 1) ++ << LZMA_NUM_POS_SLOT_BITS); ++ rc_bit_tree_decode(ctx, prob, ++ LZMA_NUM_POS_SLOT_BITS, ++ &pos_slot); ++ if (pos_slot >= LZMA_START_POS_MODEL_INDEX) { ++ int i, mi; ++ num_bits = (pos_slot >> 1) - 1; ++ ctx->rep0 = 2 | (pos_slot & 1); ++ if (pos_slot < LZMA_END_POS_MODEL_INDEX) { ++ ctx->rep0 <<= num_bits; ++ prob = p + LZMA_SPEC_POS + ++ ctx->rep0 - pos_slot - 1; ++ } else { ++ num_bits -= LZMA_NUM_ALIGN_BITS; ++ while (num_bits--) ++ ctx->rep0 = (ctx->rep0 << 1) | ++ rc_direct_bit(ctx); ++ prob = p + LZMA_ALIGN; ++ ctx->rep0 <<= LZMA_NUM_ALIGN_BITS; ++ num_bits = LZMA_NUM_ALIGN_BITS; ++ } ++ i = 1; ++ mi = 1; ++ while (num_bits--) { ++ if (rc_get_bit(ctx, prob + mi, &mi)) ++ ctx->rep0 |= i; ++ i <<= 1; ++ } ++ } else ++ ctx->rep0 = pos_slot; ++ if (++(ctx->rep0) == 0) ++ return; ++ } ++ ++ len += LZMA_MATCH_MIN_LEN; ++ ++ copy_bytes(ctx, ctx->rep0, len); ++} ++ ++ ++static int ++do_unlzma(struct unlzma_ctx *ctx) ++{ ++ u8 hdr_buf[sizeof(struct lzma_header)]; ++ struct lzma_header *header = (struct lzma_header *)hdr_buf; ++ u32 pos_state_mask; ++ u32 literal_pos_mask; ++ int lc, pb, lp; ++ int num_probs; ++ int i, mi; ++ u16 *p; ++ ++ for (i = 0; i < sizeof(struct lzma_header); i++) { ++ hdr_buf[i] = rc_read(ctx); ++ } ++ ++ ctx->pos = 0; ++ get_buffer(ctx); ++ ctx->active = true; ++ ctx->state = 0; ++ ctx->rep0 = ctx->rep1 = ctx->rep2 = ctx->rep3 = 1; ++ ++ ctx->previous_byte = 0; ++ ctx->code = 0; ++ ctx->range = 0xFFFFFFFF; ++ ++ ctx->dict_size = le32_to_cpu(header->dict_size); ++ ++ if (header->pos >= (9 * 5 * 5)) ++ return -1; ++ ++ mi = 0; ++ lc = header->pos; ++ while (lc >= 9) { ++ mi++; ++ lc -= 9; ++ } ++ pb = 0; ++ lp = mi; ++ while (lp >= 5) { ++ pb++; ++ lp -= 5; ++ } ++ pos_state_mask = (1 << pb) - 1; ++ literal_pos_mask = (1 << lp) - 1; ++ ++ if (ctx->dict_size == 0) ++ ctx->dict_size = 1; ++ ++ num_probs = LZMA_BASE_SIZE + (LZMA_LIT_SIZE << (lc + lp)); ++ if (ctx->workspace_size < num_probs * sizeof(*p)) { ++ if (ctx->workspace) ++ vfree(ctx->workspace); ++ ctx->workspace_size = num_probs * sizeof(*p); ++ ctx->workspace = vmalloc(ctx->workspace_size); ++ } ++ p = (u16 *) ctx->workspace; ++ if (!p) ++ return -1; ++ ++ num_probs = LZMA_LITERAL + (LZMA_LIT_SIZE << (lc + lp)); ++ for (i = 0; i < num_probs; i++) ++ p[i] = (1 << RC_MODEL_TOTAL_BITS) >> 1; ++ ++ for (i = 0; i < 5; i++) ++ rc_get_code(ctx); ++ ++ while (1) { ++ int pos_state = ctx->pos & pos_state_mask; ++ u16 *prob = p + LZMA_IS_MATCH + ++ (ctx->state << LZMA_NUM_POS_BITS_MAX) + pos_state; ++ if (rc_is_bit_0(ctx, prob)) ++ process_bit0(ctx, p, pos_state, prob, ++ lc, literal_pos_mask); ++ else { ++ process_bit1(ctx, p, pos_state, prob); ++ if (ctx->rep0 == 0) ++ break; ++ } ++ if (unlzma_should_stop(ctx)) ++ break; ++ } ++ if (likely(!unlzma_should_stop(ctx))) ++ rc_normalize(ctx); ++ ++ return ctx->pos; ++} ++ ++ ++static void ++unlzma_reset_buf(struct unlzma_ctx *ctx) ++{ ++ ctx->avail_in = 0; ++ ctx->next_in = NULL; ++ ctx->avail_out = 0; ++ ctx->next_out = NULL; ++} ++ ++static int ++unlzma_thread(void *data) ++{ ++ struct unlzma_ctx *ctx = data; ++ ++ mutex_lock(&ctx->mutex); ++ do { ++ if (do_unlzma(ctx) < 0) ++ ctx->pos = 0; ++ unlzma_reset_buf(ctx); ++ ctx->cancel = false; ++ ctx->active = false; ++ while (ctx->head) { ++ struct unlzma_buffer *bh = ctx->head; ++ ctx->head = bh->last; ++ kfree(bh); ++ } ++ } while (!kthread_should_stop()); ++ mutex_unlock(&ctx->mutex); ++ return 0; ++} ++ ++ ++static int ++unlzma_init(struct crypto_tfm *tfm) ++{ ++ return 0; ++} ++ ++static void ++unlzma_cancel(struct unlzma_ctx *ctx) ++{ ++ unlzma_reset_buf(ctx); ++ ++ if (!ctx->active) ++ return; ++ ++ ctx->cancel = true; ++ do { ++ mutex_unlock(&ctx->mutex); ++ wake_up(&ctx->next_req); ++ schedule(); ++ mutex_lock(&ctx->mutex); ++ } while (ctx->cancel); ++} ++ ++ ++static void ++unlzma_exit(struct crypto_tfm *tfm) ++{ ++ struct unlzma_ctx *ctx = crypto_tfm_ctx(tfm); ++ ++ if (ctx->thread) { ++ unlzma_cancel(ctx); ++ kthread_stop(ctx->thread); ++ ctx->thread = NULL; ++ } ++} ++ ++static int ++unlzma_decompress_setup(struct crypto_pcomp *tfm, void *p, unsigned int len) ++{ ++ struct unlzma_ctx *ctx = crypto_tfm_ctx(crypto_pcomp_tfm(tfm)); ++ int ret = 0; ++ ++ if (ctx->thread) ++ return 0; ++ ++ mutex_init(&ctx->mutex); ++ init_waitqueue_head(&ctx->next_req); ++ ctx->thread = kthread_run(unlzma_thread, ctx, "unlzma/%d", instance++); ++ if (IS_ERR(ctx->thread)) { ++ ret = PTR_ERR(ctx->thread); ++ ctx->thread = NULL; ++ } ++ ++ return ret; ++} ++ ++static int ++unlzma_decompress_init(struct crypto_pcomp *tfm) ++{ ++ struct unlzma_ctx *ctx = crypto_tfm_ctx(crypto_pcomp_tfm(tfm)); ++ ++ ctx->pos = 0; ++ return 0; ++} ++ ++static void ++unlzma_wait_complete(struct unlzma_ctx *ctx, bool finish) ++{ ++ do { ++ mutex_unlock(&ctx->mutex); ++ wake_up(&ctx->next_req); ++ schedule(); ++ mutex_lock(&ctx->mutex); ++ } while (ctx->active && (ctx->avail_in > 0) && (ctx->avail_out > 0)); ++} ++ ++static int ++unlzma_decompress_update(struct crypto_pcomp *tfm, struct comp_request *req) ++{ ++ struct unlzma_ctx *ctx = crypto_tfm_ctx(crypto_pcomp_tfm(tfm)); ++ size_t pos = 0; ++ ++ mutex_lock(&ctx->mutex); ++ if (!ctx->active && !req->avail_in) ++ goto out; ++ ++ pos = ctx->pos; ++ ctx->next_in = req->next_in; ++ ctx->avail_in = req->avail_in; ++ ctx->next_out = req->next_out; ++ ctx->avail_out = req->avail_out; ++ ++ unlzma_wait_complete(ctx, false); ++ ++ req->next_in = ctx->next_in; ++ req->avail_in = ctx->avail_in; ++ req->next_out = ctx->next_out; ++ req->avail_out = ctx->avail_out; ++ ctx->next_in = 0; ++ ctx->avail_in = 0; ++ pos = ctx->pos - pos; ++ ++out: ++ mutex_unlock(&ctx->mutex); ++ return pos; ++} ++ ++static int ++unlzma_decompress_final(struct crypto_pcomp *tfm, struct comp_request *req) ++{ ++ struct unlzma_ctx *ctx = crypto_tfm_ctx(crypto_pcomp_tfm(tfm)); ++ int ret = 0; ++ ++ /* cancel pending operation */ ++ mutex_lock(&ctx->mutex); ++ if (ctx->active) { ++ // ret = -EINVAL; ++ unlzma_cancel(ctx); ++ } ++ ctx->pos = 0; ++ mutex_unlock(&ctx->mutex); ++ return ret; ++} ++ ++ ++static struct pcomp_alg unlzma_alg = { ++ .decompress_setup = unlzma_decompress_setup, ++ .decompress_init = unlzma_decompress_init, ++ .decompress_update = unlzma_decompress_update, ++ .decompress_final = unlzma_decompress_final, ++ ++ .base = { ++ .cra_name = "lzma", ++ .cra_flags = CRYPTO_ALG_TYPE_PCOMPRESS, ++ .cra_ctxsize = sizeof(struct unlzma_ctx), ++ .cra_module = THIS_MODULE, ++ .cra_init = unlzma_init, ++ .cra_exit = unlzma_exit, ++ } ++}; ++ ++static int __init ++unlzma_mod_init(void) ++{ ++ return crypto_register_pcomp(&unlzma_alg); ++} ++ ++static void __exit ++unlzma_mod_exit(void) ++{ ++ crypto_unregister_pcomp(&unlzma_alg); ++} ++ ++module_init(unlzma_mod_init); ++module_exit(unlzma_mod_exit); ++ ++MODULE_LICENSE("GPL"); ++MODULE_DESCRIPTION("LZMA Decompression Algorithm"); ++MODULE_AUTHOR("Felix Fietkau "); +--- a/crypto/Kconfig ++++ b/crypto/Kconfig +@@ -768,6 +768,12 @@ config CRYPTO_ZLIB + help + This is the zlib algorithm. + ++config CRYPTO_UNLZMA ++ tristate "LZMA decompression" ++ select CRYPTO_PCOMP ++ help ++ This is the lzma decompression module. ++ + config CRYPTO_LZO + tristate "LZO compression algorithm" + select CRYPTO_ALGAPI +--- a/crypto/Makefile ++++ b/crypto/Makefile +@@ -75,6 +75,7 @@ obj-$(CONFIG_CRYPTO_SEED) += seed.o + obj-$(CONFIG_CRYPTO_SALSA20) += salsa20_generic.o + obj-$(CONFIG_CRYPTO_DEFLATE) += deflate.o + obj-$(CONFIG_CRYPTO_ZLIB) += zlib.o ++obj-$(CONFIG_CRYPTO_UNLZMA) += unlzma.o + obj-$(CONFIG_CRYPTO_MICHAEL_MIC) += michael_mic.o + obj-$(CONFIG_CRYPTO_CRC32C) += crc32c.o + obj-$(CONFIG_CRYPTO_AUTHENC) += authenc.o +--- /dev/null ++++ b/crypto/unlzma.h +@@ -0,0 +1,80 @@ ++/* LZMA uncompresion module for pcomp ++ * Copyright (C) 2009 Felix Fietkau ++ * ++ * Based on: ++ * Initial Linux kernel adaptation ++ * Copyright (C) 2006 Alain < alain@knaff.lu > ++ * ++ * Based on small lzma deflate implementation/Small range coder ++ * implementation for lzma. ++ * Copyright (C) 2006 Aurelien Jacobs < aurel@gnuage.org > ++ * ++ * Based on LzmaDecode.c from the LZMA SDK 4.22 (http://www.7-zip.org/) ++ * Copyright (C) 1999-2005 Igor Pavlov ++ * ++ * This program is free software; you can redistribute it and/or modify it ++ * under the terms of the GNU General Public License version 2 as published ++ * by the Free Software Foundation. ++ */ ++#ifndef __UNLZMA_H ++#define __UNLZMA_H ++ ++struct lzma_header { ++ __u8 pos; ++ __le32 dict_size; ++} __attribute__ ((packed)) ; ++ ++ ++#define RC_TOP_BITS 24 ++#define RC_MOVE_BITS 5 ++#define RC_MODEL_TOTAL_BITS 11 ++ ++#define LZMA_BASE_SIZE 1846 ++#define LZMA_LIT_SIZE 768 ++ ++#define LZMA_NUM_POS_BITS_MAX 4 ++ ++#define LZMA_LEN_NUM_LOW_BITS 3 ++#define LZMA_LEN_NUM_MID_BITS 3 ++#define LZMA_LEN_NUM_HIGH_BITS 8 ++ ++#define LZMA_LEN_CHOICE 0 ++#define LZMA_LEN_CHOICE_2 (LZMA_LEN_CHOICE + 1) ++#define LZMA_LEN_LOW (LZMA_LEN_CHOICE_2 + 1) ++#define LZMA_LEN_MID (LZMA_LEN_LOW \ ++ + (1 << (LZMA_NUM_POS_BITS_MAX + LZMA_LEN_NUM_LOW_BITS))) ++#define LZMA_LEN_HIGH (LZMA_LEN_MID \ ++ +(1 << (LZMA_NUM_POS_BITS_MAX + LZMA_LEN_NUM_MID_BITS))) ++#define LZMA_NUM_LEN_PROBS (LZMA_LEN_HIGH + (1 << LZMA_LEN_NUM_HIGH_BITS)) ++ ++#define LZMA_NUM_STATES 12 ++#define LZMA_NUM_LIT_STATES 7 ++ ++#define LZMA_START_POS_MODEL_INDEX 4 ++#define LZMA_END_POS_MODEL_INDEX 14 ++#define LZMA_NUM_FULL_DISTANCES (1 << (LZMA_END_POS_MODEL_INDEX >> 1)) ++ ++#define LZMA_NUM_POS_SLOT_BITS 6 ++#define LZMA_NUM_LEN_TO_POS_STATES 4 ++ ++#define LZMA_NUM_ALIGN_BITS 4 ++ ++#define LZMA_MATCH_MIN_LEN 2 ++ ++#define LZMA_IS_MATCH 0 ++#define LZMA_IS_REP (LZMA_IS_MATCH + (LZMA_NUM_STATES << LZMA_NUM_POS_BITS_MAX)) ++#define LZMA_IS_REP_G0 (LZMA_IS_REP + LZMA_NUM_STATES) ++#define LZMA_IS_REP_G1 (LZMA_IS_REP_G0 + LZMA_NUM_STATES) ++#define LZMA_IS_REP_G2 (LZMA_IS_REP_G1 + LZMA_NUM_STATES) ++#define LZMA_IS_REP_0_LONG (LZMA_IS_REP_G2 + LZMA_NUM_STATES) ++#define LZMA_POS_SLOT (LZMA_IS_REP_0_LONG \ ++ + (LZMA_NUM_STATES << LZMA_NUM_POS_BITS_MAX)) ++#define LZMA_SPEC_POS (LZMA_POS_SLOT \ ++ +(LZMA_NUM_LEN_TO_POS_STATES << LZMA_NUM_POS_SLOT_BITS)) ++#define LZMA_ALIGN (LZMA_SPEC_POS \ ++ + LZMA_NUM_FULL_DISTANCES - LZMA_END_POS_MODEL_INDEX) ++#define LZMA_LEN_CODER (LZMA_ALIGN + (1 << LZMA_NUM_ALIGN_BITS)) ++#define LZMA_REP_LEN_CODER (LZMA_LEN_CODER + LZMA_NUM_LEN_PROBS) ++#define LZMA_LITERAL (LZMA_REP_LEN_CODER + LZMA_NUM_LEN_PROBS) ++ ++#endif diff --git a/target/linux/generic-2.6/patches-2.6.31/053-squashfs_lzma.patch b/target/linux/generic-2.6/patches-2.6.31/053-squashfs_lzma.patch new file mode 100644 index 0000000000..c07239bb2d --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/053-squashfs_lzma.patch @@ -0,0 +1,214 @@ +--- a/fs/squashfs/Kconfig ++++ b/fs/squashfs/Kconfig +@@ -2,7 +2,6 @@ config SQUASHFS + tristate "SquashFS 4.0 - Squashed file system support" + depends on BLOCK + select CRYPTO +- select CRYPTO_ZLIB + help + Saying Y here includes support for SquashFS 4.0 (a Compressed + Read-Only File System). Squashfs is a highly compressed read-only +@@ -37,6 +36,26 @@ config SQUASHFS_EMBEDDED + + If unsure, say N. + ++config SQUASHFS_SUPPORT_ZLIB ++ bool ++ prompt "Support ZLIB compression" if SQUASHFS_SUPPORT_LZMA ++ depends on SQUASHFS ++ select CRYPTO_ZLIB ++ default y ++ help ++ ZLIB is the default compression used in squashfs. If you are ++ using LZMA compression instead, you can remove support for ZLIB ++ entirely. ++ ++config SQUASHFS_SUPPORT_LZMA ++ bool "Support LZMA compression" ++ depends on SQUASHFS ++ select CRYPTO_UNLZMA ++ help ++ By default SquashFS uses ZLIB compression, however (if your tools ++ support it, you can use LZMA instead, which saves space. ++ ++ + config SQUASHFS_FRAGMENT_CACHE_SIZE + int "Number of fragments cached" if SQUASHFS_EMBEDDED + depends on SQUASHFS +--- a/fs/squashfs/squashfs_fs.h ++++ b/fs/squashfs/squashfs_fs.h +@@ -212,6 +212,7 @@ struct meta_index { + * definitions for structures on disk + */ + #define ZLIB_COMPRESSION 1 ++#define LZMA_COMPRESSION 2 + + struct squashfs_super_block { + __le32 s_magic; +--- a/fs/squashfs/super.c ++++ b/fs/squashfs/super.c +@@ -48,13 +48,65 @@ + #include "squashfs.h" + + +-#define SQUASHFS_CRYPTO_ALG "zlib" ++static int squashfs_setup_zlib(struct squashfs_sb_info *msblk) ++{ ++ int err = -EOPNOTSUPP; ++ ++#ifdef CONFIG_SQUASHFS_SUPPORT_ZLIB ++ struct { ++ struct nlattr nla; ++ int val; ++ } params = { ++ .nla = { ++ .nla_len = nla_attr_size(sizeof(int)), ++ .nla_type = ZLIB_DECOMP_WINDOWBITS, ++ }, ++ .val = DEF_WBITS, ++ }; ++ ++ msblk->tfm = crypto_alloc_pcomp("zlib", 0, ++ CRYPTO_ALG_ASYNC); ++ if (IS_ERR(msblk->tfm)) { ++ ERROR("Failed to load zlib crypto module\n"); ++ return PTR_ERR(msblk->tfm); ++ } ++ ++ err = crypto_decompress_setup(msblk->tfm, ¶ms, sizeof(params)); ++ if (err) { ++ ERROR("Failed to set up decompression parameters\n"); ++ crypto_free_pcomp(msblk->tfm); ++ } ++#endif + ++ return err; ++} ++ ++static int squashfs_setup_lzma(struct squashfs_sb_info *msblk) ++{ ++ int err = -EOPNOTSUPP; ++ ++#ifdef CONFIG_SQUASHFS_SUPPORT_LZMA ++ msblk->tfm = crypto_alloc_pcomp("lzma", 0, ++ CRYPTO_ALG_ASYNC); ++ if (IS_ERR(msblk->tfm)) { ++ ERROR("Failed to load lzma crypto module\n"); ++ return PTR_ERR(msblk->tfm); ++ } ++ ++ err = crypto_decompress_setup(msblk->tfm, NULL, 0); ++ if (err) { ++ ERROR("Failed to set up decompression parameters\n"); ++ crypto_free_pcomp(msblk->tfm); ++ } ++#endif ++ ++ return err; ++} + + static struct file_system_type squashfs_fs_type; + static struct super_operations squashfs_super_ops; + +-static int supported_squashfs_filesystem(short major, short minor, short comp) ++static int supported_squashfs_filesystem(short major, short minor) + { + if (major < SQUASHFS_MAJOR) { + ERROR("Major/Minor mismatch, older Squashfs %d.%d " +@@ -67,9 +119,6 @@ static int supported_squashfs_filesystem + return -EINVAL; + } + +- if (comp != ZLIB_COMPRESSION) +- return -EINVAL; +- + return 0; + } + +@@ -84,16 +133,6 @@ static int squashfs_fill_super(struct su + unsigned short flags; + unsigned int fragments; + u64 lookup_table_start; +- struct { +- struct nlattr nla; +- int val; +- } params = { +- .nla = { +- .nla_len = nla_attr_size(sizeof(int)), +- .nla_type = ZLIB_DECOMP_WINDOWBITS, +- }, +- .val = DEF_WBITS, +- }; + int err; + + TRACE("Entered squashfs_fill_superblock\n"); +@@ -105,21 +144,6 @@ static int squashfs_fill_super(struct su + } + msblk = sb->s_fs_info; + +- msblk->tfm = crypto_alloc_pcomp(SQUASHFS_CRYPTO_ALG, 0, +- CRYPTO_ALG_ASYNC); +- if (IS_ERR(msblk->tfm)) { +- ERROR("Failed to load %s crypto module\n", +- SQUASHFS_CRYPTO_ALG); +- err = PTR_ERR(msblk->tfm); +- goto failed_pcomp; +- } +- +- err = crypto_decompress_setup(msblk->tfm, ¶ms, sizeof(params)); +- if (err) { +- ERROR("Failed to set up decompression parameters\n"); +- goto failure; +- } +- + sblk = kzalloc(sizeof(*sblk), GFP_KERNEL); + if (sblk == NULL) { + ERROR("Failed to allocate squashfs_super_block\n"); +@@ -159,8 +183,21 @@ static int squashfs_fill_super(struct su + + /* Check the MAJOR & MINOR versions and compression type */ + err = supported_squashfs_filesystem(le16_to_cpu(sblk->s_major), +- le16_to_cpu(sblk->s_minor), +- le16_to_cpu(sblk->compression)); ++ le16_to_cpu(sblk->s_minor)); ++ if (err < 0) ++ goto failed_mount; ++ ++ switch(le16_to_cpu(sblk->compression)) { ++ case ZLIB_COMPRESSION: ++ err = squashfs_setup_zlib(msblk); ++ break; ++ case LZMA_COMPRESSION: ++ err = squashfs_setup_lzma(msblk); ++ break; ++ default: ++ err = -EINVAL; ++ break; ++ } + if (err < 0) + goto failed_mount; + +@@ -316,21 +353,16 @@ allocate_root: + return 0; + + failed_mount: ++ if (msblk->tfm) ++ crypto_free_pcomp(msblk->tfm); + squashfs_cache_delete(msblk->block_cache); + squashfs_cache_delete(msblk->fragment_cache); + squashfs_cache_delete(msblk->read_page); + kfree(msblk->inode_lookup_table); + kfree(msblk->fragment_index); + kfree(msblk->id_table); +- crypto_free_pcomp(msblk->tfm); +- kfree(sb->s_fs_info); +- sb->s_fs_info = NULL; + kfree(sblk); +- return err; +- + failure: +- crypto_free_pcomp(msblk->tfm); +-failed_pcomp: + kfree(sb->s_fs_info); + sb->s_fs_info = NULL; + return err; diff --git a/target/linux/generic-2.6/patches-2.6.31/060-block2mtd_init.patch b/target/linux/generic-2.6/patches-2.6.31/060-block2mtd_init.patch new file mode 100644 index 0000000000..50df4aae0f --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/060-block2mtd_init.patch @@ -0,0 +1,110 @@ +--- a/drivers/mtd/devices/block2mtd.c ++++ b/drivers/mtd/devices/block2mtd.c +@@ -14,6 +14,7 @@ + #include + #include + #include ++#include + #include + #include + #include +@@ -232,10 +233,11 @@ static void block2mtd_free_device(struct + + + /* FIXME: ensure that mtd->size % erase_size == 0 */ +-static struct block2mtd_dev *add_device(char *devname, int erase_size) ++static struct block2mtd_dev *add_device(char *devname, int erase_size, const char *mtdname) + { + struct block_device *bdev; + struct block2mtd_dev *dev; ++ struct mtd_partition *part; + char *name; + + if (!devname) +@@ -273,17 +275,17 @@ static struct block2mtd_dev *add_device( + + mutex_init(&dev->write_mutex); + +- /* Setup the MTD structure */ +- /* make the name contain the block device in */ +- name = kmalloc(sizeof("block2mtd: ") + strlen(devname) + 1, +- GFP_KERNEL); ++ if (!mtdname) ++ mtdname = devname; ++ ++ name = kmalloc(strlen(mtdname) + 1, GFP_KERNEL); + if (!name) + goto devinit_err; + +- sprintf(name, "block2mtd: %s", devname); ++ strcpy(name, mtdname); + dev->mtd.name = name; + +- dev->mtd.size = dev->blkdev->bd_inode->i_size & PAGE_MASK; ++ dev->mtd.size = dev->blkdev->bd_inode->i_size & PAGE_MASK & ~(erase_size - 1); + dev->mtd.erasesize = erase_size; + dev->mtd.writesize = 1; + dev->mtd.type = MTD_RAM; +@@ -296,14 +298,17 @@ static struct block2mtd_dev *add_device( + dev->mtd.priv = dev; + dev->mtd.owner = THIS_MODULE; + +- if (add_mtd_device(&dev->mtd)) { ++ part = kzalloc(sizeof(struct mtd_partition), GFP_KERNEL); ++ part->name = dev->mtd.name; ++ part->offset = 0; ++ part->size = dev->mtd.size; ++ if (add_mtd_partitions(&dev->mtd, part, 1)) { + /* Device didnt get added, so free the entry */ + goto devinit_err; + } + list_add(&dev->list, &blkmtd_device_list); + INFO("mtd%d: [%s] erase_size = %dKiB [%d]", dev->mtd.index, +- dev->mtd.name + strlen("block2mtd: "), +- dev->mtd.erasesize >> 10, dev->mtd.erasesize); ++ mtdname, dev->mtd.erasesize >> 10, dev->mtd.erasesize); + return dev; + + devinit_err: +@@ -376,9 +381,9 @@ static char block2mtd_paramline[80 + 12] + + static int block2mtd_setup2(const char *val) + { +- char buf[80 + 12]; /* 80 for device, 12 for erase size */ ++ char buf[80 + 12 + 80]; /* 80 for device, 12 for erase size, 80 for name */ + char *str = buf; +- char *token[2]; ++ char *token[3]; + char *name; + size_t erase_size = PAGE_SIZE; + int i, ret; +@@ -389,7 +394,7 @@ static int block2mtd_setup2(const char * + strcpy(str, val); + kill_final_newline(str); + +- for (i = 0; i < 2; i++) ++ for (i = 0; i < 3; i++) + token[i] = strsep(&str, ","); + + if (str) +@@ -408,8 +413,10 @@ static int block2mtd_setup2(const char * + parse_err("illegal erase size"); + } + } ++ if (token[2] && (strlen(token[2]) + 1 > 80)) ++ parse_err("mtd device name too long"); + +- add_device(name, erase_size); ++ add_device(name, erase_size, token[2]); + + return 0; + } +@@ -443,7 +450,7 @@ static int block2mtd_setup(const char *v + + + module_param_call(block2mtd, block2mtd_setup, NULL, NULL, 0200); +-MODULE_PARM_DESC(block2mtd, "Device to use. \"block2mtd=[,]\""); ++MODULE_PARM_DESC(block2mtd, "Device to use. \"block2mtd=[,[,]]\""); + + static int __init block2mtd_init(void) + { diff --git a/target/linux/generic-2.6/patches-2.6.31/065-rootfs_split.patch b/target/linux/generic-2.6/patches-2.6.31/065-rootfs_split.patch new file mode 100644 index 0000000000..a642dee2ad --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/065-rootfs_split.patch @@ -0,0 +1,624 @@ +--- a/drivers/mtd/Kconfig ++++ b/drivers/mtd/Kconfig +@@ -53,6 +53,16 @@ config MTD_TESTS + should normally be compiled as kernel modules. The modules perform + various checks and verifications when loaded. + ++config MTD_ROOTFS_ROOT_DEV ++ bool "Automatically set 'rootfs' partition to be root filesystem" ++ depends on MTD_PARTITIONS ++ default y ++ ++config MTD_ROOTFS_SPLIT ++ bool "Automatically split 'rootfs' partition for squashfs" ++ depends on MTD_PARTITIONS ++ default y ++ + config MTD_REDBOOT_PARTS + tristate "RedBoot partition table parsing" + depends on MTD_PARTITIONS +--- a/drivers/mtd/mtdpart.c ++++ b/drivers/mtd/mtdpart.c +@@ -18,6 +18,8 @@ + #include + #include + #include ++#include ++#include + + /* Our partition linked list */ + static LIST_HEAD(mtd_partitions); +@@ -35,7 +37,7 @@ struct mtd_part { + * the pointer to that structure with this macro. + */ + #define PART(x) ((struct mtd_part *)(x)) +- ++#define IS_PART(mtd) (mtd->read == part_read) + + /* + * MTD methods which simply translate the effective address and pass through +@@ -502,6 +504,150 @@ out_register: + return slave; + } + ++#ifdef CONFIG_MTD_ROOTFS_SPLIT ++#define ROOTFS_SPLIT_NAME "rootfs_data" ++#define ROOTFS_REMOVED_NAME "" ++ ++struct squashfs_super_block { ++ __le32 s_magic; ++ __le32 pad0[9]; ++ __le64 bytes_used; ++}; ++ ++ ++static int split_squashfs(struct mtd_info *master, int offset, int *split_offset) ++{ ++ struct squashfs_super_block sb; ++ int len, ret; ++ ++ ret = master->read(master, offset, sizeof(sb), &len, (void *) &sb); ++ if (ret || (len != sizeof(sb))) { ++ printk(KERN_ALERT "split_squashfs: error occured while reading " ++ "from \"%s\"\n", master->name); ++ return -EINVAL; ++ } ++ ++ if (SQUASHFS_MAGIC != le32_to_cpu(sb.s_magic) ) { ++ printk(KERN_ALERT "split_squashfs: no squashfs found in \"%s\"\n", ++ master->name); ++ *split_offset = 0; ++ return 0; ++ } ++ ++ if (le64_to_cpu((sb.bytes_used)) <= 0) { ++ printk(KERN_ALERT "split_squashfs: squashfs is empty in \"%s\"\n", ++ master->name); ++ *split_offset = 0; ++ return 0; ++ } ++ ++ len = (u32) le64_to_cpu(sb.bytes_used); ++ len += (offset & 0x000fffff); ++ len += (master->erasesize - 1); ++ len &= ~(master->erasesize - 1); ++ len -= (offset & 0x000fffff); ++ *split_offset = offset + len; ++ ++ return 0; ++} ++ ++static int split_rootfs_data(struct mtd_info *master, struct mtd_info *rpart, const struct mtd_partition *part) ++{ ++ struct mtd_partition *dpart; ++ struct mtd_part *slave = NULL; ++ int split_offset = 0; ++ int ret; ++ ++ ret = split_squashfs(master, part->offset, &split_offset); ++ if (ret) ++ return ret; ++ ++ if (split_offset <= 0) ++ return 0; ++ ++ dpart = kmalloc(sizeof(*part)+sizeof(ROOTFS_SPLIT_NAME)+1, GFP_KERNEL); ++ if (dpart == NULL) { ++ printk(KERN_INFO "split_squashfs: no memory for partition \"%s\"\n", ++ ROOTFS_SPLIT_NAME); ++ return -ENOMEM; ++ } ++ ++ memcpy(dpart, part, sizeof(*part)); ++ dpart->name = (unsigned char *)&dpart[1]; ++ strcpy(dpart->name, ROOTFS_SPLIT_NAME); ++ ++ dpart->size -= split_offset - dpart->offset; ++ dpart->offset = split_offset; ++ ++ if (dpart == NULL) ++ return 1; ++ ++ printk(KERN_INFO "mtd: partition \"%s\" created automatically, ofs=%llX, len=%llX \n", ++ ROOTFS_SPLIT_NAME, dpart->offset, dpart->size); ++ ++ slave = add_one_partition(master, dpart, 0, split_offset); ++ if (!slave) { ++ kfree(dpart); ++ return -ENOMEM; ++ } ++ rpart->split = &slave->mtd; ++ ++ return 0; ++} ++ ++static int refresh_rootfs_split(struct mtd_info *mtd) ++{ ++ struct mtd_partition tpart; ++ struct mtd_part *part; ++ char *name; ++ //int index = 0; ++ int offset, size; ++ int ret; ++ ++ part = PART(mtd); ++ ++ /* check for the new squashfs offset first */ ++ ret = split_squashfs(part->master, part->offset, &offset); ++ if (ret) ++ return ret; ++ ++ if ((offset > 0) && !mtd->split) { ++ printk(KERN_INFO "%s: creating new split partition for \"%s\"\n", __func__, mtd->name); ++ /* if we don't have a rootfs split partition, create a new one */ ++ tpart.name = (char *) mtd->name; ++ tpart.size = mtd->size; ++ tpart.offset = part->offset; ++ ++ return split_rootfs_data(part->master, &part->mtd, &tpart); ++ } else if ((offset > 0) && mtd->split) { ++ /* update the offsets of the existing partition */ ++ size = mtd->size + part->offset - offset; ++ ++ part = PART(mtd->split); ++ part->offset = offset; ++ part->mtd.size = size; ++ printk(KERN_INFO "%s: %s partition \"" ROOTFS_SPLIT_NAME "\", offset: 0x%06x (0x%06x)\n", ++ __func__, (!strcmp(part->mtd.name, ROOTFS_SPLIT_NAME) ? "updating" : "creating"), ++ (u32) part->offset, (u32) part->mtd.size); ++ name = kmalloc(sizeof(ROOTFS_SPLIT_NAME) + 1, GFP_KERNEL); ++ strcpy(name, ROOTFS_SPLIT_NAME); ++ part->mtd.name = name; ++ } else if ((offset <= 0) && mtd->split) { ++ printk(KERN_INFO "%s: removing partition \"%s\"\n", __func__, mtd->split->name); ++ ++ /* mark existing partition as removed */ ++ part = PART(mtd->split); ++ name = kmalloc(sizeof(ROOTFS_SPLIT_NAME) + 1, GFP_KERNEL); ++ strcpy(name, ROOTFS_REMOVED_NAME); ++ part->mtd.name = name; ++ part->offset = 0; ++ part->mtd.size = 0; ++ } ++ ++ return 0; ++} ++#endif /* CONFIG_MTD_ROOTFS_SPLIT */ ++ + /* + * This function, given a master MTD object and a partition table, creates + * and registers slave MTD objects which are bound to the master according to +@@ -517,7 +663,7 @@ int add_mtd_partitions(struct mtd_info * + { + struct mtd_part *slave; + uint64_t cur_offset = 0; +- int i; ++ int i, ret; + + printk(KERN_NOTICE "Creating %d MTD partitions on \"%s\":\n", nbparts, master->name); + +@@ -525,6 +671,21 @@ int add_mtd_partitions(struct mtd_info * + slave = add_one_partition(master, parts + i, i, cur_offset); + if (!slave) + return -ENOMEM; ++ ++ if (!strcmp(parts[i].name, "rootfs")) { ++#ifdef CONFIG_MTD_ROOTFS_ROOT_DEV ++ if (ROOT_DEV == 0) { ++ printk(KERN_NOTICE "mtd: partition \"rootfs\" " ++ "set to be root filesystem\n"); ++ ROOT_DEV = MKDEV(MTD_BLOCK_MAJOR, i); ++ } ++#endif ++#ifdef CONFIG_MTD_ROOTFS_SPLIT ++ ret = split_rootfs_data(master, &slave->mtd, &parts[i]); ++ /* if (ret == 0) ++ j++; */ ++#endif ++ } + cur_offset = slave->offset + slave->mtd.size; + } + +@@ -532,6 +693,32 @@ int add_mtd_partitions(struct mtd_info * + } + EXPORT_SYMBOL(add_mtd_partitions); + ++int refresh_mtd_partitions(struct mtd_info *mtd) ++{ ++ int ret = 0; ++ ++ if (IS_PART(mtd)) { ++ struct mtd_part *part; ++ struct mtd_info *master; ++ ++ part = PART(mtd); ++ master = part->master; ++ if (master->refresh_device) ++ ret = master->refresh_device(master); ++ } ++ ++ if (!ret && mtd->refresh_device) ++ ret = mtd->refresh_device(mtd); ++ ++#ifdef CONFIG_MTD_ROOTFS_SPLIT ++ if (!ret && IS_PART(mtd) && !strcmp(mtd->name, "rootfs")) ++ refresh_rootfs_split(mtd); ++#endif ++ ++ return 0; ++} ++EXPORT_SYMBOL_GPL(refresh_mtd_partitions); ++ + static DEFINE_SPINLOCK(part_parser_lock); + static LIST_HEAD(part_parsers); + +--- a/drivers/mtd/devices/block2mtd.c ++++ b/drivers/mtd/devices/block2mtd.c +@@ -29,6 +29,8 @@ struct block2mtd_dev { + struct block_device *blkdev; + struct mtd_info mtd; + struct mutex write_mutex; ++ rwlock_t bdev_mutex; ++ char devname[0]; + }; + + +@@ -81,6 +83,12 @@ static int block2mtd_erase(struct mtd_in + size_t len = instr->len; + int err; + ++ read_lock(&dev->bdev_mutex); ++ if (!dev->blkdev) { ++ err = -EINVAL; ++ goto done; ++ } ++ + instr->state = MTD_ERASING; + mutex_lock(&dev->write_mutex); + err = _block2mtd_erase(dev, from, len); +@@ -93,6 +101,10 @@ static int block2mtd_erase(struct mtd_in + + instr->state = MTD_ERASE_DONE; + mtd_erase_callback(instr); ++ ++done: ++ read_unlock(&dev->bdev_mutex); ++ + return err; + } + +@@ -104,10 +116,14 @@ static int block2mtd_read(struct mtd_inf + struct page *page; + int index = from >> PAGE_SHIFT; + int offset = from & (PAGE_SIZE-1); +- int cpylen; ++ int cpylen, err = 0; ++ ++ read_lock(&dev->bdev_mutex); ++ if (!dev->blkdev || (from > mtd->size)) { ++ err = -EINVAL; ++ goto done; ++ } + +- if (from > mtd->size) +- return -EINVAL; + if (from + len > mtd->size) + len = mtd->size - from; + +@@ -122,10 +138,14 @@ static int block2mtd_read(struct mtd_inf + len = len - cpylen; + + page = page_read(dev->blkdev->bd_inode->i_mapping, index); +- if (!page) +- return -ENOMEM; +- if (IS_ERR(page)) +- return PTR_ERR(page); ++ if (!page) { ++ err = -ENOMEM; ++ goto done; ++ } ++ if (IS_ERR(page)) { ++ err = PTR_ERR(page); ++ goto done; ++ } + + memcpy(buf, page_address(page) + offset, cpylen); + page_cache_release(page); +@@ -136,7 +156,10 @@ static int block2mtd_read(struct mtd_inf + offset = 0; + index++; + } +- return 0; ++ ++done: ++ read_unlock(&dev->bdev_mutex); ++ return err; + } + + +@@ -188,12 +211,22 @@ static int block2mtd_write(struct mtd_in + size_t *retlen, const u_char *buf) + { + struct block2mtd_dev *dev = mtd->priv; +- int err; ++ int err = 0; ++ ++ read_lock(&dev->bdev_mutex); ++ if (!dev->blkdev) { ++ err = -EINVAL; ++ goto done; ++ } + + if (!len) +- return 0; +- if (to >= mtd->size) +- return -ENOSPC; ++ goto done; ++ ++ if (to >= mtd->size) { ++ err = -ENOSPC; ++ goto done; ++ } ++ + if (to + len > mtd->size) + len = mtd->size - to; + +@@ -202,6 +235,9 @@ static int block2mtd_write(struct mtd_in + mutex_unlock(&dev->write_mutex); + if (err > 0) + err = 0; ++ ++done: ++ read_unlock(&dev->bdev_mutex); + return err; + } + +@@ -210,52 +246,29 @@ static int block2mtd_write(struct mtd_in + static void block2mtd_sync(struct mtd_info *mtd) + { + struct block2mtd_dev *dev = mtd->priv; +- sync_blockdev(dev->blkdev); +- return; +-} +- +- +-static void block2mtd_free_device(struct block2mtd_dev *dev) +-{ +- if (!dev) +- return; +- +- kfree(dev->mtd.name); + +- if (dev->blkdev) { +- invalidate_mapping_pages(dev->blkdev->bd_inode->i_mapping, +- 0, -1); +- close_bdev_exclusive(dev->blkdev, FMODE_READ|FMODE_WRITE); +- } ++ read_lock(&dev->bdev_mutex); ++ if (dev->blkdev) ++ sync_blockdev(dev->blkdev); ++ read_unlock(&dev->bdev_mutex); + +- kfree(dev); ++ return; + } + + +-/* FIXME: ensure that mtd->size % erase_size == 0 */ +-static struct block2mtd_dev *add_device(char *devname, int erase_size, const char *mtdname) ++static int _open_bdev(struct block2mtd_dev *dev) + { + struct block_device *bdev; +- struct block2mtd_dev *dev; +- struct mtd_partition *part; +- char *name; +- +- if (!devname) +- return NULL; +- +- dev = kzalloc(sizeof(struct block2mtd_dev), GFP_KERNEL); +- if (!dev) +- return NULL; + + /* Get a handle on the device */ +- bdev = open_bdev_exclusive(devname, FMODE_READ|FMODE_WRITE, NULL); ++ bdev = open_bdev_exclusive(dev->devname, FMODE_READ|FMODE_WRITE, NULL); + #ifndef MODULE + if (IS_ERR(bdev)) { + + /* We might not have rootfs mounted at this point. Try + to resolve the device name by other means. */ + +- dev_t devt = name_to_dev_t(devname); ++ dev_t devt = name_to_dev_t(dev->devname); + if (devt) { + bdev = open_by_devnum(devt, FMODE_WRITE | FMODE_READ); + } +@@ -263,17 +276,97 @@ static struct block2mtd_dev *add_device( + #endif + + if (IS_ERR(bdev)) { +- ERROR("error: cannot open device %s", devname); +- goto devinit_err; ++ ERROR("error: cannot open device %s", dev->devname); ++ return 1; + } + dev->blkdev = bdev; + + if (MAJOR(bdev->bd_dev) == MTD_BLOCK_MAJOR) { + ERROR("attempting to use an MTD device as a block device"); +- goto devinit_err; ++ return 1; + } + ++ return 0; ++} ++ ++static void _close_bdev(struct block2mtd_dev *dev) ++{ ++ struct block_device *bdev; ++ ++ if (!dev->blkdev) ++ return; ++ ++ bdev = dev->blkdev; ++ invalidate_mapping_pages(dev->blkdev->bd_inode->i_mapping, 0, -1); ++ close_bdev_exclusive(dev->blkdev, FMODE_READ|FMODE_WRITE); ++ dev->blkdev = NULL; ++} ++ ++static void block2mtd_free_device(struct block2mtd_dev *dev) ++{ ++ if (!dev) ++ return; ++ ++ kfree(dev->mtd.name); ++ _close_bdev(dev); ++ kfree(dev); ++} ++ ++ ++static int block2mtd_refresh(struct mtd_info *mtd) ++{ ++ struct block2mtd_dev *dev = mtd->priv; ++ struct block_device *bdev; ++ dev_t devt; ++ int err = 0; ++ ++ /* no other mtd function can run at this point */ ++ write_lock(&dev->bdev_mutex); ++ ++ /* get the device number for the whole disk */ ++ devt = MKDEV(MAJOR(dev->blkdev->bd_dev), 0); ++ ++ /* close the old block device */ ++ _close_bdev(dev); ++ ++ /* open the whole disk, issue a partition rescan, then */ ++ bdev = open_by_devnum(devt, FMODE_WRITE | FMODE_READ); ++ if (!bdev || !bdev->bd_disk) ++ err = -EINVAL; ++ else { ++ err = rescan_partitions(bdev->bd_disk, bdev); ++ } ++ if (bdev) ++ close_bdev_exclusive(bdev, FMODE_READ|FMODE_WRITE); ++ ++ /* try to open the partition block device again */ ++ _open_bdev(dev); ++ write_unlock(&dev->bdev_mutex); ++ ++ return err; ++} ++ ++/* FIXME: ensure that mtd->size % erase_size == 0 */ ++static struct block2mtd_dev *add_device(char *devname, int erase_size, char *mtdname) ++{ ++ struct block2mtd_dev *dev; ++ struct mtd_partition *part; ++ char *name; ++ ++ if (!devname) ++ return NULL; ++ ++ dev = kzalloc(sizeof(struct block2mtd_dev) + strlen(devname) + 1, GFP_KERNEL); ++ if (!dev) ++ return NULL; ++ ++ strcpy(dev->devname, devname); ++ ++ if (_open_bdev(dev)) ++ goto devinit_err; ++ + mutex_init(&dev->write_mutex); ++ rwlock_init(&dev->bdev_mutex); + + if (!mtdname) + mtdname = devname; +@@ -297,6 +390,7 @@ static struct block2mtd_dev *add_device( + dev->mtd.read = block2mtd_read; + dev->mtd.priv = dev; + dev->mtd.owner = THIS_MODULE; ++ dev->mtd.refresh_device = block2mtd_refresh; + + part = kzalloc(sizeof(struct mtd_partition), GFP_KERNEL); + part->name = dev->mtd.name; +--- a/drivers/mtd/mtdchar.c ++++ b/drivers/mtd/mtdchar.c +@@ -18,6 +18,7 @@ + + #include + #include ++#include + + #include + +@@ -814,6 +815,13 @@ static int mtd_ioctl(struct inode *inode + file->f_pos = 0; + break; + } ++#ifdef CONFIG_MTD_PARTITIONS ++ case MTDREFRESH: ++ { ++ ret = refresh_mtd_partitions(mtd); ++ break; ++ } ++#endif + + default: + ret = -ENOTTY; +--- a/include/linux/mtd/mtd.h ++++ b/include/linux/mtd/mtd.h +@@ -101,6 +101,7 @@ struct mtd_oob_ops { + uint8_t *oobbuf; + }; + ++struct mtd_info; + struct mtd_info { + u_char type; + uint32_t flags; +@@ -241,6 +242,9 @@ struct mtd_info { + struct device dev; + int usecount; + ++ int (*refresh_device)(struct mtd_info *mtd); ++ struct mtd_info *split; ++ + /* If the driver is something smart, like UBI, it may need to maintain + * its own reference counting. The below functions are only for driver. + * The driver may register its callbacks. These callbacks are not +--- a/include/linux/mtd/partitions.h ++++ b/include/linux/mtd/partitions.h +@@ -34,12 +34,14 @@ + * erasesize aligned (e.g. use MTDPART_OFS_NEXTBLK). + */ + ++struct mtd_partition; + struct mtd_partition { + char *name; /* identifier string */ + uint64_t size; /* partition size */ + uint64_t offset; /* offset within the master MTD space */ + uint32_t mask_flags; /* master MTD flags to mask out for this partition */ + struct nand_ecclayout *ecclayout; /* out of band layout for this partition (NAND only)*/ ++ int (*refresh_partition)(struct mtd_info *); + }; + + #define MTDPART_OFS_NXTBLK (-2) +@@ -51,6 +53,7 @@ struct mtd_info; + + int add_mtd_partitions(struct mtd_info *, const struct mtd_partition *, int); + int del_mtd_partitions(struct mtd_info *); ++int refresh_mtd_partitions(struct mtd_info *); + + /* + * Functions dealing with the various ways of partitioning the space +--- a/include/mtd/mtd-abi.h ++++ b/include/mtd/mtd-abi.h +@@ -110,6 +110,7 @@ struct otp_info { + #define MEMERASE64 _IOW('M', 20, struct erase_info_user64) + #define MEMWRITEOOB64 _IOWR('M', 21, struct mtd_oob_buf64) + #define MEMREADOOB64 _IOWR('M', 22, struct mtd_oob_buf64) ++#define MTDREFRESH _IO('M', 23) + + /* + * Obsolete legacy interface. Keep it in order not to break userspace diff --git a/target/linux/generic-2.6/patches-2.6.31/070-redboot_space.patch b/target/linux/generic-2.6/patches-2.6.31/070-redboot_space.patch new file mode 100644 index 0000000000..6ace83c01a --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/070-redboot_space.patch @@ -0,0 +1,30 @@ +--- a/drivers/mtd/redboot.c ++++ b/drivers/mtd/redboot.c +@@ -249,14 +249,21 @@ static int parse_redboot_partitions(stru + #endif + names += strlen(names)+1; + +-#ifdef CONFIG_MTD_REDBOOT_PARTS_UNALLOCATED + if(fl->next && fl->img->flash_base + fl->img->size + master->erasesize <= fl->next->img->flash_base) { +- i++; +- parts[i].offset = parts[i-1].size + parts[i-1].offset; +- parts[i].size = fl->next->img->flash_base - parts[i].offset; +- parts[i].name = nullname; +- } ++ if (!strcmp(parts[i].name, "rootfs")) { ++ parts[i].size = fl->next->img->flash_base; ++ parts[i].size &= ~(master->erasesize - 1); ++ parts[i].size -= parts[i].offset; ++#ifdef CONFIG_MTD_REDBOOT_PARTS_UNALLOCATED ++ nrparts--; ++ } else { ++ i++; ++ parts[i].offset = parts[i-1].size + parts[i-1].offset; ++ parts[i].size = fl->next->img->flash_base - parts[i].offset; ++ parts[i].name = nullname; + #endif ++ } ++ } + tmp_fl = fl; + fl = fl->next; + kfree(tmp_fl); diff --git a/target/linux/generic-2.6/patches-2.6.31/071-redboot_boardconfig.patch b/target/linux/generic-2.6/patches-2.6.31/071-redboot_boardconfig.patch new file mode 100644 index 0000000000..f9bc4fe36b --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/071-redboot_boardconfig.patch @@ -0,0 +1,60 @@ +--- a/drivers/mtd/redboot.c ++++ b/drivers/mtd/redboot.c +@@ -11,6 +11,8 @@ + #include + #include + ++#define BOARD_CONFIG_PART "boardconfig" ++ + struct fis_image_desc { + unsigned char name[16]; // Null terminated name + uint32_t flash_base; // Address within FLASH of image +@@ -41,6 +43,7 @@ static int parse_redboot_partitions(stru + struct mtd_partition **pparts, + unsigned long fis_origin) + { ++ unsigned long max_offset = 0; + int nrparts = 0; + struct fis_image_desc *buf; + struct mtd_partition *parts; +@@ -209,14 +212,14 @@ static int parse_redboot_partitions(stru + } + } + #endif +- parts = kzalloc(sizeof(*parts)*nrparts + nulllen + namelen, GFP_KERNEL); ++ parts = kzalloc(sizeof(*parts) * (nrparts + 1) + nulllen + namelen + sizeof(BOARD_CONFIG_PART), GFP_KERNEL); + + if (!parts) { + ret = -ENOMEM; + goto out; + } + +- nullname = (char *)&parts[nrparts]; ++ nullname = (char *)&parts[nrparts + 1]; + #ifdef CONFIG_MTD_REDBOOT_PARTS_UNALLOCATED + if (nulllen > 0) { + strcpy(nullname, nullstring); +@@ -235,6 +238,8 @@ static int parse_redboot_partitions(stru + } + #endif + for ( ; iimg->size; + parts[i].offset = fl->img->flash_base; + parts[i].name = names; +@@ -268,6 +273,14 @@ static int parse_redboot_partitions(stru + fl = fl->next; + kfree(tmp_fl); + } ++ if(master->size - max_offset >= master->erasesize) ++ { ++ parts[nrparts].size = master->size - max_offset; ++ parts[nrparts].offset = max_offset; ++ parts[nrparts].name = names; ++ strcpy(names, BOARD_CONFIG_PART); ++ nrparts++; ++ } + ret = nrparts; + *pparts = parts; + out: diff --git a/target/linux/generic-2.6/patches-2.6.31/080-mtd_plat_nand_chip_fixup.patch b/target/linux/generic-2.6/patches-2.6.31/080-mtd_plat_nand_chip_fixup.patch new file mode 100644 index 0000000000..2566151ef9 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/080-mtd_plat_nand_chip_fixup.patch @@ -0,0 +1,32 @@ +--- a/include/linux/mtd/nand.h ++++ b/include/linux/mtd/nand.h +@@ -575,6 +575,7 @@ struct platform_nand_chip { + int chip_delay; + unsigned int options; + const char **part_probe_types; ++ int (*chip_fixup)(struct mtd_info *mtd); + void (*set_parts)(uint64_t size, + struct platform_nand_chip *chip); + void *priv; +--- a/drivers/mtd/nand/plat_nand.c ++++ b/drivers/mtd/nand/plat_nand.c +@@ -80,7 +80,18 @@ static int __devinit plat_nand_probe(str + } + + /* Scan to find existance of the device */ +- if (nand_scan(&data->mtd, 1)) { ++ if (nand_scan_ident(&data->mtd, 1)) { ++ res = -ENXIO; ++ goto out; ++ } ++ ++ if (pdata->chip.chip_fixup) { ++ res = pdata->chip.chip_fixup(&data->mtd); ++ if (res) ++ goto out; ++ } ++ ++ if (nand_scan_tail(&data->mtd)) { + res = -ENXIO; + goto out; + } diff --git a/target/linux/generic-2.6/patches-2.6.31/081-mtd_myloader_partition_parser.patch b/target/linux/generic-2.6/patches-2.6.31/081-mtd_myloader_partition_parser.patch new file mode 100644 index 0000000000..61b822ee88 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/081-mtd_myloader_partition_parser.patch @@ -0,0 +1,35 @@ +--- a/drivers/mtd/Kconfig ++++ b/drivers/mtd/Kconfig +@@ -181,6 +181,22 @@ config MTD_AR7_PARTS + ---help--- + TI AR7 partitioning support + ++config MTD_MYLOADER_PARTS ++ tristate "MyLoader partition parsing" ++ depends on MTD_PARTITIONS && (ADM5120 || ATHEROS_AR231X || ATHEROS_AR71XX) ++ ---help--- ++ MyLoader is a bootloader which allows the user to define partitions ++ in flash devices, by putting a table in the second erase block ++ on the device, similar to a partition table. This table gives the ++ offsets and lengths of the user defined partitions. ++ ++ If you need code which can detect and parse these tables, and ++ register MTD 'partitions' corresponding to each image detected, ++ enable this option. ++ ++ You will still need the parsing functions to be called by the driver ++ for your particular device. It won't happen automatically. ++ + comment "User Modules And Translation Layers" + + config MTD_CHAR +--- a/drivers/mtd/Makefile ++++ b/drivers/mtd/Makefile +@@ -13,6 +13,7 @@ obj-$(CONFIG_MTD_CMDLINE_PARTS) += cmdli + obj-$(CONFIG_MTD_AFS_PARTS) += afs.o + obj-$(CONFIG_MTD_AR7_PARTS) += ar7part.o + obj-$(CONFIG_MTD_OF_PARTS) += ofpart.o ++obj-$(CONFIG_MTD_MYLOADER_PARTS) += myloader.o + + # 'Users' - code which presents functionality to userspace. + obj-$(CONFIG_MTD_CHAR) += mtdchar.o diff --git a/target/linux/generic-2.6/patches-2.6.31/090-mtd_fix_nand_correct_data_return_code.patch b/target/linux/generic-2.6/patches-2.6.31/090-mtd_fix_nand_correct_data_return_code.patch new file mode 100644 index 0000000000..be2c7d7323 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/090-mtd_fix_nand_correct_data_return_code.patch @@ -0,0 +1,12 @@ +--- a/drivers/mtd/nand/nand_ecc.c ++++ b/drivers/mtd/nand/nand_ecc.c +@@ -492,8 +492,7 @@ int nand_correct_data(struct mtd_info *m + if ((bitsperbyte[b0] + bitsperbyte[b1] + bitsperbyte[b2]) == 1) + return 1; /* error in ecc data; no action needed */ + +- printk(KERN_ERR "uncorrectable error : "); +- return -1; ++ return -EBADMSG; + } + EXPORT_SYMBOL(nand_correct_data); + diff --git a/target/linux/generic-2.6/patches-2.6.31/100-netfilter_layer7_2.17.patch b/target/linux/generic-2.6/patches-2.6.31/100-netfilter_layer7_2.17.patch new file mode 100644 index 0000000000..e69de29bb2 diff --git a/target/linux/generic-2.6/patches-2.6.31/100-netfilter_layer7_2.21.patch b/target/linux/generic-2.6/patches-2.6.31/100-netfilter_layer7_2.21.patch new file mode 100644 index 0000000000..b846f78559 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/100-netfilter_layer7_2.21.patch @@ -0,0 +1,2132 @@ +--- /dev/null ++++ b/include/linux/netfilter/xt_layer7.h +@@ -0,0 +1,13 @@ ++#ifndef _XT_LAYER7_H ++#define _XT_LAYER7_H ++ ++#define MAX_PATTERN_LEN 8192 ++#define MAX_PROTOCOL_LEN 256 ++ ++struct xt_layer7_info { ++ char protocol[MAX_PROTOCOL_LEN]; ++ char pattern[MAX_PATTERN_LEN]; ++ u_int8_t invert; ++}; ++ ++#endif /* _XT_LAYER7_H */ +--- a/include/net/netfilter/nf_conntrack.h ++++ b/include/net/netfilter/nf_conntrack.h +@@ -116,6 +116,22 @@ struct nf_conn { + u_int32_t secmark; + #endif + ++#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || \ ++ defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE) ++ struct { ++ /* ++ * e.g. "http". NULL before decision. "unknown" after decision ++ * if no match. ++ */ ++ char *app_proto; ++ /* ++ * application layer data so far. NULL after match decision. ++ */ ++ char *app_data; ++ unsigned int app_data_len; ++ } layer7; ++#endif ++ + /* Storage reserved for other modules: */ + union nf_conntrack_proto proto; + +--- a/net/netfilter/Kconfig ++++ b/net/netfilter/Kconfig +@@ -858,6 +858,27 @@ config NETFILTER_XT_MATCH_STATE + + To compile it as a module, choose M here. If unsure, say N. + ++config NETFILTER_XT_MATCH_LAYER7 ++ tristate '"layer7" match support' ++ depends on NETFILTER_XTABLES ++ depends on EXPERIMENTAL && (IP_NF_CONNTRACK || NF_CONNTRACK) ++ depends on NF_CT_ACCT ++ help ++ Say Y if you want to be able to classify connections (and their ++ packets) based on regular expression matching of their application ++ layer data. This is one way to classify applications such as ++ peer-to-peer filesharing systems that do not always use the same ++ port. ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ ++config NETFILTER_XT_MATCH_LAYER7_DEBUG ++ bool 'Layer 7 debugging output' ++ depends on NETFILTER_XT_MATCH_LAYER7 ++ help ++ Say Y to get lots of debugging output. ++ ++ + config NETFILTER_XT_MATCH_STATISTIC + tristate '"statistic" match support' + depends on NETFILTER_ADVANCED +--- a/net/netfilter/Makefile ++++ b/net/netfilter/Makefile +@@ -89,6 +89,7 @@ obj-$(CONFIG_NETFILTER_XT_MATCH_RECENT) + obj-$(CONFIG_NETFILTER_XT_MATCH_SCTP) += xt_sctp.o + obj-$(CONFIG_NETFILTER_XT_MATCH_SOCKET) += xt_socket.o + obj-$(CONFIG_NETFILTER_XT_MATCH_STATE) += xt_state.o ++obj-$(CONFIG_NETFILTER_XT_MATCH_LAYER7) += xt_layer7.o + obj-$(CONFIG_NETFILTER_XT_MATCH_STATISTIC) += xt_statistic.o + obj-$(CONFIG_NETFILTER_XT_MATCH_STRING) += xt_string.o + obj-$(CONFIG_NETFILTER_XT_MATCH_TCPMSS) += xt_tcpmss.o +--- a/net/netfilter/nf_conntrack_core.c ++++ b/net/netfilter/nf_conntrack_core.c +@@ -200,6 +200,14 @@ destroy_conntrack(struct nf_conntrack *n + * too. */ + nf_ct_remove_expectations(ct); + ++ #if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE) ++ if(ct->layer7.app_proto) ++ kfree(ct->layer7.app_proto); ++ if(ct->layer7.app_data) ++ kfree(ct->layer7.app_data); ++ #endif ++ ++ + /* We overload first tuple to link into unconfirmed list. */ + if (!nf_ct_is_confirmed(ct)) { + BUG_ON(hlist_nulls_unhashed(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode)); +--- a/net/netfilter/nf_conntrack_standalone.c ++++ b/net/netfilter/nf_conntrack_standalone.c +@@ -171,6 +171,12 @@ static int ct_seq_show(struct seq_file * + goto release; + #endif + ++#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE) ++ if(ct->layer7.app_proto && ++ seq_printf(s, "l7proto=%s ", ct->layer7.app_proto)) ++ return -ENOSPC; ++#endif ++ + if (seq_printf(s, "use=%u\n", atomic_read(&ct->ct_general.use))) + goto release; + +--- /dev/null ++++ b/net/netfilter/regexp/regexp.c +@@ -0,0 +1,1197 @@ ++/* ++ * regcomp and regexec -- regsub and regerror are elsewhere ++ * @(#)regexp.c 1.3 of 18 April 87 ++ * ++ * Copyright (c) 1986 by University of Toronto. ++ * Written by Henry Spencer. Not derived from licensed software. ++ * ++ * Permission is granted to anyone to use this software for any ++ * purpose on any computer system, and to redistribute it freely, ++ * subject to the following restrictions: ++ * ++ * 1. The author is not responsible for the consequences of use of ++ * this software, no matter how awful, even if they arise ++ * from defects in it. ++ * ++ * 2. The origin of this software must not be misrepresented, either ++ * by explicit claim or by omission. ++ * ++ * 3. Altered versions must be plainly marked as such, and must not ++ * be misrepresented as being the original software. ++ * ++ * Beware that some of this code is subtly aware of the way operator ++ * precedence is structured in regular expressions. Serious changes in ++ * regular-expression syntax might require a total rethink. ++ * ++ * This code was modified by Ethan Sommer to work within the kernel ++ * (it now uses kmalloc etc..) ++ * ++ * Modified slightly by Matthew Strait to use more modern C. ++ */ ++ ++#include "regexp.h" ++#include "regmagic.h" ++ ++/* added by ethan and matt. Lets it work in both kernel and user space. ++(So iptables can use it, for instance.) Yea, it goes both ways... */ ++#if __KERNEL__ ++ #define malloc(foo) kmalloc(foo,GFP_ATOMIC) ++#else ++ #define printk(format,args...) printf(format,##args) ++#endif ++ ++void regerror(char * s) ++{ ++ printk("<3>Regexp: %s\n", s); ++ /* NOTREACHED */ ++} ++ ++/* ++ * The "internal use only" fields in regexp.h are present to pass info from ++ * compile to execute that permits the execute phase to run lots faster on ++ * simple cases. They are: ++ * ++ * regstart char that must begin a match; '\0' if none obvious ++ * reganch is the match anchored (at beginning-of-line only)? ++ * regmust string (pointer into program) that match must include, or NULL ++ * regmlen length of regmust string ++ * ++ * Regstart and reganch permit very fast decisions on suitable starting points ++ * for a match, cutting down the work a lot. Regmust permits fast rejection ++ * of lines that cannot possibly match. The regmust tests are costly enough ++ * that regcomp() supplies a regmust only if the r.e. contains something ++ * potentially expensive (at present, the only such thing detected is * or + ++ * at the start of the r.e., which can involve a lot of backup). Regmlen is ++ * supplied because the test in regexec() needs it and regcomp() is computing ++ * it anyway. ++ */ ++ ++/* ++ * Structure for regexp "program". This is essentially a linear encoding ++ * of a nondeterministic finite-state machine (aka syntax charts or ++ * "railroad normal form" in parsing technology). Each node is an opcode ++ * plus a "next" pointer, possibly plus an operand. "Next" pointers of ++ * all nodes except BRANCH implement concatenation; a "next" pointer with ++ * a BRANCH on both ends of it is connecting two alternatives. (Here we ++ * have one of the subtle syntax dependencies: an individual BRANCH (as ++ * opposed to a collection of them) is never concatenated with anything ++ * because of operator precedence.) The operand of some types of node is ++ * a literal string; for others, it is a node leading into a sub-FSM. In ++ * particular, the operand of a BRANCH node is the first node of the branch. ++ * (NB this is *not* a tree structure: the tail of the branch connects ++ * to the thing following the set of BRANCHes.) The opcodes are: ++ */ ++ ++/* definition number opnd? meaning */ ++#define END 0 /* no End of program. */ ++#define BOL 1 /* no Match "" at beginning of line. */ ++#define EOL 2 /* no Match "" at end of line. */ ++#define ANY 3 /* no Match any one character. */ ++#define ANYOF 4 /* str Match any character in this string. */ ++#define ANYBUT 5 /* str Match any character not in this string. */ ++#define BRANCH 6 /* node Match this alternative, or the next... */ ++#define BACK 7 /* no Match "", "next" ptr points backward. */ ++#define EXACTLY 8 /* str Match this string. */ ++#define NOTHING 9 /* no Match empty string. */ ++#define STAR 10 /* node Match this (simple) thing 0 or more times. */ ++#define PLUS 11 /* node Match this (simple) thing 1 or more times. */ ++#define OPEN 20 /* no Mark this point in input as start of #n. */ ++ /* OPEN+1 is number 1, etc. */ ++#define CLOSE 30 /* no Analogous to OPEN. */ ++ ++/* ++ * Opcode notes: ++ * ++ * BRANCH The set of branches constituting a single choice are hooked ++ * together with their "next" pointers, since precedence prevents ++ * anything being concatenated to any individual branch. The ++ * "next" pointer of the last BRANCH in a choice points to the ++ * thing following the whole choice. This is also where the ++ * final "next" pointer of each individual branch points; each ++ * branch starts with the operand node of a BRANCH node. ++ * ++ * BACK Normal "next" pointers all implicitly point forward; BACK ++ * exists to make loop structures possible. ++ * ++ * STAR,PLUS '?', and complex '*' and '+', are implemented as circular ++ * BRANCH structures using BACK. Simple cases (one character ++ * per match) are implemented with STAR and PLUS for speed ++ * and to minimize recursive plunges. ++ * ++ * OPEN,CLOSE ...are numbered at compile time. ++ */ ++ ++/* ++ * A node is one char of opcode followed by two chars of "next" pointer. ++ * "Next" pointers are stored as two 8-bit pieces, high order first. The ++ * value is a positive offset from the opcode of the node containing it. ++ * An operand, if any, simply follows the node. (Note that much of the ++ * code generation knows about this implicit relationship.) ++ * ++ * Using two bytes for the "next" pointer is vast overkill for most things, ++ * but allows patterns to get big without disasters. ++ */ ++#define OP(p) (*(p)) ++#define NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377)) ++#define OPERAND(p) ((p) + 3) ++ ++/* ++ * See regmagic.h for one further detail of program structure. ++ */ ++ ++ ++/* ++ * Utility definitions. ++ */ ++#ifndef CHARBITS ++#define UCHARAT(p) ((int)*(unsigned char *)(p)) ++#else ++#define UCHARAT(p) ((int)*(p)&CHARBITS) ++#endif ++ ++#define FAIL(m) { regerror(m); return(NULL); } ++#define ISMULT(c) ((c) == '*' || (c) == '+' || (c) == '?') ++#define META "^$.[()|?+*\\" ++ ++/* ++ * Flags to be passed up and down. ++ */ ++#define HASWIDTH 01 /* Known never to match null string. */ ++#define SIMPLE 02 /* Simple enough to be STAR/PLUS operand. */ ++#define SPSTART 04 /* Starts with * or +. */ ++#define WORST 0 /* Worst case. */ ++ ++/* ++ * Global work variables for regcomp(). ++ */ ++struct match_globals { ++char *reginput; /* String-input pointer. */ ++char *regbol; /* Beginning of input, for ^ check. */ ++char **regstartp; /* Pointer to startp array. */ ++char **regendp; /* Ditto for endp. */ ++char *regparse; /* Input-scan pointer. */ ++int regnpar; /* () count. */ ++char regdummy; ++char *regcode; /* Code-emit pointer; ®dummy = don't. */ ++long regsize; /* Code size. */ ++}; ++ ++/* ++ * Forward declarations for regcomp()'s friends. ++ */ ++#ifndef STATIC ++#define STATIC static ++#endif ++STATIC char *reg(struct match_globals *g, int paren,int *flagp); ++STATIC char *regbranch(struct match_globals *g, int *flagp); ++STATIC char *regpiece(struct match_globals *g, int *flagp); ++STATIC char *regatom(struct match_globals *g, int *flagp); ++STATIC char *regnode(struct match_globals *g, char op); ++STATIC char *regnext(struct match_globals *g, char *p); ++STATIC void regc(struct match_globals *g, char b); ++STATIC void reginsert(struct match_globals *g, char op, char *opnd); ++STATIC void regtail(struct match_globals *g, char *p, char *val); ++STATIC void regoptail(struct match_globals *g, char *p, char *val); ++ ++ ++__kernel_size_t my_strcspn(const char *s1,const char *s2) ++{ ++ char *scan1; ++ char *scan2; ++ int count; ++ ++ count = 0; ++ for (scan1 = (char *)s1; *scan1 != '\0'; scan1++) { ++ for (scan2 = (char *)s2; *scan2 != '\0';) /* ++ moved down. */ ++ if (*scan1 == *scan2++) ++ return(count); ++ count++; ++ } ++ return(count); ++} ++ ++/* ++ - regcomp - compile a regular expression into internal code ++ * ++ * We can't allocate space until we know how big the compiled form will be, ++ * but we can't compile it (and thus know how big it is) until we've got a ++ * place to put the code. So we cheat: we compile it twice, once with code ++ * generation turned off and size counting turned on, and once "for real". ++ * This also means that we don't allocate space until we are sure that the ++ * thing really will compile successfully, and we never have to move the ++ * code and thus invalidate pointers into it. (Note that it has to be in ++ * one piece because free() must be able to free it all.) ++ * ++ * Beware that the optimization-preparation code in here knows about some ++ * of the structure of the compiled regexp. ++ */ ++regexp * ++regcomp(char *exp,int *patternsize) ++{ ++ register regexp *r; ++ register char *scan; ++ register char *longest; ++ register int len; ++ int flags; ++ struct match_globals g; ++ ++ /* commented out by ethan ++ extern char *malloc(); ++ */ ++ ++ if (exp == NULL) ++ FAIL("NULL argument"); ++ ++ /* First pass: determine size, legality. */ ++ g.regparse = exp; ++ g.regnpar = 1; ++ g.regsize = 0L; ++ g.regcode = &g.regdummy; ++ regc(&g, MAGIC); ++ if (reg(&g, 0, &flags) == NULL) ++ return(NULL); ++ ++ /* Small enough for pointer-storage convention? */ ++ if (g.regsize >= 32767L) /* Probably could be 65535L. */ ++ FAIL("regexp too big"); ++ ++ /* Allocate space. */ ++ *patternsize=sizeof(regexp) + (unsigned)g.regsize; ++ r = (regexp *)malloc(sizeof(regexp) + (unsigned)g.regsize); ++ if (r == NULL) ++ FAIL("out of space"); ++ ++ /* Second pass: emit code. */ ++ g.regparse = exp; ++ g.regnpar = 1; ++ g.regcode = r->program; ++ regc(&g, MAGIC); ++ if (reg(&g, 0, &flags) == NULL) ++ return(NULL); ++ ++ /* Dig out information for optimizations. */ ++ r->regstart = '\0'; /* Worst-case defaults. */ ++ r->reganch = 0; ++ r->regmust = NULL; ++ r->regmlen = 0; ++ scan = r->program+1; /* First BRANCH. */ ++ if (OP(regnext(&g, scan)) == END) { /* Only one top-level choice. */ ++ scan = OPERAND(scan); ++ ++ /* Starting-point info. */ ++ if (OP(scan) == EXACTLY) ++ r->regstart = *OPERAND(scan); ++ else if (OP(scan) == BOL) ++ r->reganch++; ++ ++ /* ++ * If there's something expensive in the r.e., find the ++ * longest literal string that must appear and make it the ++ * regmust. Resolve ties in favor of later strings, since ++ * the regstart check works with the beginning of the r.e. ++ * and avoiding duplication strengthens checking. Not a ++ * strong reason, but sufficient in the absence of others. ++ */ ++ if (flags&SPSTART) { ++ longest = NULL; ++ len = 0; ++ for (; scan != NULL; scan = regnext(&g, scan)) ++ if (OP(scan) == EXACTLY && strlen(OPERAND(scan)) >= len) { ++ longest = OPERAND(scan); ++ len = strlen(OPERAND(scan)); ++ } ++ r->regmust = longest; ++ r->regmlen = len; ++ } ++ } ++ ++ return(r); ++} ++ ++/* ++ - reg - regular expression, i.e. main body or parenthesized thing ++ * ++ * Caller must absorb opening parenthesis. ++ * ++ * Combining parenthesis handling with the base level of regular expression ++ * is a trifle forced, but the need to tie the tails of the branches to what ++ * follows makes it hard to avoid. ++ */ ++static char * ++reg(struct match_globals *g, int paren, int *flagp /* Parenthesized? */ ) ++{ ++ register char *ret; ++ register char *br; ++ register char *ender; ++ register int parno = 0; /* 0 makes gcc happy */ ++ int flags; ++ ++ *flagp = HASWIDTH; /* Tentatively. */ ++ ++ /* Make an OPEN node, if parenthesized. */ ++ if (paren) { ++ if (g->regnpar >= NSUBEXP) ++ FAIL("too many ()"); ++ parno = g->regnpar; ++ g->regnpar++; ++ ret = regnode(g, OPEN+parno); ++ } else ++ ret = NULL; ++ ++ /* Pick up the branches, linking them together. */ ++ br = regbranch(g, &flags); ++ if (br == NULL) ++ return(NULL); ++ if (ret != NULL) ++ regtail(g, ret, br); /* OPEN -> first. */ ++ else ++ ret = br; ++ if (!(flags&HASWIDTH)) ++ *flagp &= ~HASWIDTH; ++ *flagp |= flags&SPSTART; ++ while (*g->regparse == '|') { ++ g->regparse++; ++ br = regbranch(g, &flags); ++ if (br == NULL) ++ return(NULL); ++ regtail(g, ret, br); /* BRANCH -> BRANCH. */ ++ if (!(flags&HASWIDTH)) ++ *flagp &= ~HASWIDTH; ++ *flagp |= flags&SPSTART; ++ } ++ ++ /* Make a closing node, and hook it on the end. */ ++ ender = regnode(g, (paren) ? CLOSE+parno : END); ++ regtail(g, ret, ender); ++ ++ /* Hook the tails of the branches to the closing node. */ ++ for (br = ret; br != NULL; br = regnext(g, br)) ++ regoptail(g, br, ender); ++ ++ /* Check for proper termination. */ ++ if (paren && *g->regparse++ != ')') { ++ FAIL("unmatched ()"); ++ } else if (!paren && *g->regparse != '\0') { ++ if (*g->regparse == ')') { ++ FAIL("unmatched ()"); ++ } else ++ FAIL("junk on end"); /* "Can't happen". */ ++ /* NOTREACHED */ ++ } ++ ++ return(ret); ++} ++ ++/* ++ - regbranch - one alternative of an | operator ++ * ++ * Implements the concatenation operator. ++ */ ++static char * ++regbranch(struct match_globals *g, int *flagp) ++{ ++ register char *ret; ++ register char *chain; ++ register char *latest; ++ int flags; ++ ++ *flagp = WORST; /* Tentatively. */ ++ ++ ret = regnode(g, BRANCH); ++ chain = NULL; ++ while (*g->regparse != '\0' && *g->regparse != '|' && *g->regparse != ')') { ++ latest = regpiece(g, &flags); ++ if (latest == NULL) ++ return(NULL); ++ *flagp |= flags&HASWIDTH; ++ if (chain == NULL) /* First piece. */ ++ *flagp |= flags&SPSTART; ++ else ++ regtail(g, chain, latest); ++ chain = latest; ++ } ++ if (chain == NULL) /* Loop ran zero times. */ ++ (void) regnode(g, NOTHING); ++ ++ return(ret); ++} ++ ++/* ++ - regpiece - something followed by possible [*+?] ++ * ++ * Note that the branching code sequences used for ? and the general cases ++ * of * and + are somewhat optimized: they use the same NOTHING node as ++ * both the endmarker for their branch list and the body of the last branch. ++ * It might seem that this node could be dispensed with entirely, but the ++ * endmarker role is not redundant. ++ */ ++static char * ++regpiece(struct match_globals *g, int *flagp) ++{ ++ register char *ret; ++ register char op; ++ register char *next; ++ int flags; ++ ++ ret = regatom(g, &flags); ++ if (ret == NULL) ++ return(NULL); ++ ++ op = *g->regparse; ++ if (!ISMULT(op)) { ++ *flagp = flags; ++ return(ret); ++ } ++ ++ if (!(flags&HASWIDTH) && op != '?') ++ FAIL("*+ operand could be empty"); ++ *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH); ++ ++ if (op == '*' && (flags&SIMPLE)) ++ reginsert(g, STAR, ret); ++ else if (op == '*') { ++ /* Emit x* as (x&|), where & means "self". */ ++ reginsert(g, BRANCH, ret); /* Either x */ ++ regoptail(g, ret, regnode(g, BACK)); /* and loop */ ++ regoptail(g, ret, ret); /* back */ ++ regtail(g, ret, regnode(g, BRANCH)); /* or */ ++ regtail(g, ret, regnode(g, NOTHING)); /* null. */ ++ } else if (op == '+' && (flags&SIMPLE)) ++ reginsert(g, PLUS, ret); ++ else if (op == '+') { ++ /* Emit x+ as x(&|), where & means "self". */ ++ next = regnode(g, BRANCH); /* Either */ ++ regtail(g, ret, next); ++ regtail(g, regnode(g, BACK), ret); /* loop back */ ++ regtail(g, next, regnode(g, BRANCH)); /* or */ ++ regtail(g, ret, regnode(g, NOTHING)); /* null. */ ++ } else if (op == '?') { ++ /* Emit x? as (x|) */ ++ reginsert(g, BRANCH, ret); /* Either x */ ++ regtail(g, ret, regnode(g, BRANCH)); /* or */ ++ next = regnode(g, NOTHING); /* null. */ ++ regtail(g, ret, next); ++ regoptail(g, ret, next); ++ } ++ g->regparse++; ++ if (ISMULT(*g->regparse)) ++ FAIL("nested *?+"); ++ ++ return(ret); ++} ++ ++/* ++ - regatom - the lowest level ++ * ++ * Optimization: gobbles an entire sequence of ordinary characters so that ++ * it can turn them into a single node, which is smaller to store and ++ * faster to run. Backslashed characters are exceptions, each becoming a ++ * separate node; the code is simpler that way and it's not worth fixing. ++ */ ++static char * ++regatom(struct match_globals *g, int *flagp) ++{ ++ register char *ret; ++ int flags; ++ ++ *flagp = WORST; /* Tentatively. */ ++ ++ switch (*g->regparse++) { ++ case '^': ++ ret = regnode(g, BOL); ++ break; ++ case '$': ++ ret = regnode(g, EOL); ++ break; ++ case '.': ++ ret = regnode(g, ANY); ++ *flagp |= HASWIDTH|SIMPLE; ++ break; ++ case '[': { ++ register int class; ++ register int classend; ++ ++ if (*g->regparse == '^') { /* Complement of range. */ ++ ret = regnode(g, ANYBUT); ++ g->regparse++; ++ } else ++ ret = regnode(g, ANYOF); ++ if (*g->regparse == ']' || *g->regparse == '-') ++ regc(g, *g->regparse++); ++ while (*g->regparse != '\0' && *g->regparse != ']') { ++ if (*g->regparse == '-') { ++ g->regparse++; ++ if (*g->regparse == ']' || *g->regparse == '\0') ++ regc(g, '-'); ++ else { ++ class = UCHARAT(g->regparse-2)+1; ++ classend = UCHARAT(g->regparse); ++ if (class > classend+1) ++ FAIL("invalid [] range"); ++ for (; class <= classend; class++) ++ regc(g, class); ++ g->regparse++; ++ } ++ } else ++ regc(g, *g->regparse++); ++ } ++ regc(g, '\0'); ++ if (*g->regparse != ']') ++ FAIL("unmatched []"); ++ g->regparse++; ++ *flagp |= HASWIDTH|SIMPLE; ++ } ++ break; ++ case '(': ++ ret = reg(g, 1, &flags); ++ if (ret == NULL) ++ return(NULL); ++ *flagp |= flags&(HASWIDTH|SPSTART); ++ break; ++ case '\0': ++ case '|': ++ case ')': ++ FAIL("internal urp"); /* Supposed to be caught earlier. */ ++ break; ++ case '?': ++ case '+': ++ case '*': ++ FAIL("?+* follows nothing"); ++ break; ++ case '\\': ++ if (*g->regparse == '\0') ++ FAIL("trailing \\"); ++ ret = regnode(g, EXACTLY); ++ regc(g, *g->regparse++); ++ regc(g, '\0'); ++ *flagp |= HASWIDTH|SIMPLE; ++ break; ++ default: { ++ register int len; ++ register char ender; ++ ++ g->regparse--; ++ len = my_strcspn((const char *)g->regparse, (const char *)META); ++ if (len <= 0) ++ FAIL("internal disaster"); ++ ender = *(g->regparse+len); ++ if (len > 1 && ISMULT(ender)) ++ len--; /* Back off clear of ?+* operand. */ ++ *flagp |= HASWIDTH; ++ if (len == 1) ++ *flagp |= SIMPLE; ++ ret = regnode(g, EXACTLY); ++ while (len > 0) { ++ regc(g, *g->regparse++); ++ len--; ++ } ++ regc(g, '\0'); ++ } ++ break; ++ } ++ ++ return(ret); ++} ++ ++/* ++ - regnode - emit a node ++ */ ++static char * /* Location. */ ++regnode(struct match_globals *g, char op) ++{ ++ register char *ret; ++ register char *ptr; ++ ++ ret = g->regcode; ++ if (ret == &g->regdummy) { ++ g->regsize += 3; ++ return(ret); ++ } ++ ++ ptr = ret; ++ *ptr++ = op; ++ *ptr++ = '\0'; /* Null "next" pointer. */ ++ *ptr++ = '\0'; ++ g->regcode = ptr; ++ ++ return(ret); ++} ++ ++/* ++ - regc - emit (if appropriate) a byte of code ++ */ ++static void ++regc(struct match_globals *g, char b) ++{ ++ if (g->regcode != &g->regdummy) ++ *g->regcode++ = b; ++ else ++ g->regsize++; ++} ++ ++/* ++ - reginsert - insert an operator in front of already-emitted operand ++ * ++ * Means relocating the operand. ++ */ ++static void ++reginsert(struct match_globals *g, char op, char* opnd) ++{ ++ register char *src; ++ register char *dst; ++ register char *place; ++ ++ if (g->regcode == &g->regdummy) { ++ g->regsize += 3; ++ return; ++ } ++ ++ src = g->regcode; ++ g->regcode += 3; ++ dst = g->regcode; ++ while (src > opnd) ++ *--dst = *--src; ++ ++ place = opnd; /* Op node, where operand used to be. */ ++ *place++ = op; ++ *place++ = '\0'; ++ *place++ = '\0'; ++} ++ ++/* ++ - regtail - set the next-pointer at the end of a node chain ++ */ ++static void ++regtail(struct match_globals *g, char *p, char *val) ++{ ++ register char *scan; ++ register char *temp; ++ register int offset; ++ ++ if (p == &g->regdummy) ++ return; ++ ++ /* Find last node. */ ++ scan = p; ++ for (;;) { ++ temp = regnext(g, scan); ++ if (temp == NULL) ++ break; ++ scan = temp; ++ } ++ ++ if (OP(scan) == BACK) ++ offset = scan - val; ++ else ++ offset = val - scan; ++ *(scan+1) = (offset>>8)&0377; ++ *(scan+2) = offset&0377; ++} ++ ++/* ++ - regoptail - regtail on operand of first argument; nop if operandless ++ */ ++static void ++regoptail(struct match_globals *g, char *p, char *val) ++{ ++ /* "Operandless" and "op != BRANCH" are synonymous in practice. */ ++ if (p == NULL || p == &g->regdummy || OP(p) != BRANCH) ++ return; ++ regtail(g, OPERAND(p), val); ++} ++ ++/* ++ * regexec and friends ++ */ ++ ++ ++/* ++ * Forwards. ++ */ ++STATIC int regtry(struct match_globals *g, regexp *prog, char *string); ++STATIC int regmatch(struct match_globals *g, char *prog); ++STATIC int regrepeat(struct match_globals *g, char *p); ++ ++#ifdef DEBUG ++int regnarrate = 0; ++void regdump(); ++STATIC char *regprop(char *op); ++#endif ++ ++/* ++ - regexec - match a regexp against a string ++ */ ++int ++regexec(regexp *prog, char *string) ++{ ++ register char *s; ++ struct match_globals g; ++ ++ /* Be paranoid... */ ++ if (prog == NULL || string == NULL) { ++ printk("<3>Regexp: NULL parameter\n"); ++ return(0); ++ } ++ ++ /* Check validity of program. */ ++ if (UCHARAT(prog->program) != MAGIC) { ++ printk("<3>Regexp: corrupted program\n"); ++ return(0); ++ } ++ ++ /* If there is a "must appear" string, look for it. */ ++ if (prog->regmust != NULL) { ++ s = string; ++ while ((s = strchr(s, prog->regmust[0])) != NULL) { ++ if (strncmp(s, prog->regmust, prog->regmlen) == 0) ++ break; /* Found it. */ ++ s++; ++ } ++ if (s == NULL) /* Not present. */ ++ return(0); ++ } ++ ++ /* Mark beginning of line for ^ . */ ++ g.regbol = string; ++ ++ /* Simplest case: anchored match need be tried only once. */ ++ if (prog->reganch) ++ return(regtry(&g, prog, string)); ++ ++ /* Messy cases: unanchored match. */ ++ s = string; ++ if (prog->regstart != '\0') ++ /* We know what char it must start with. */ ++ while ((s = strchr(s, prog->regstart)) != NULL) { ++ if (regtry(&g, prog, s)) ++ return(1); ++ s++; ++ } ++ else ++ /* We don't -- general case. */ ++ do { ++ if (regtry(&g, prog, s)) ++ return(1); ++ } while (*s++ != '\0'); ++ ++ /* Failure. */ ++ return(0); ++} ++ ++/* ++ - regtry - try match at specific point ++ */ ++static int /* 0 failure, 1 success */ ++regtry(struct match_globals *g, regexp *prog, char *string) ++{ ++ register int i; ++ register char **sp; ++ register char **ep; ++ ++ g->reginput = string; ++ g->regstartp = prog->startp; ++ g->regendp = prog->endp; ++ ++ sp = prog->startp; ++ ep = prog->endp; ++ for (i = NSUBEXP; i > 0; i--) { ++ *sp++ = NULL; ++ *ep++ = NULL; ++ } ++ if (regmatch(g, prog->program + 1)) { ++ prog->startp[0] = string; ++ prog->endp[0] = g->reginput; ++ return(1); ++ } else ++ return(0); ++} ++ ++/* ++ - regmatch - main matching routine ++ * ++ * Conceptually the strategy is simple: check to see whether the current ++ * node matches, call self recursively to see whether the rest matches, ++ * and then act accordingly. In practice we make some effort to avoid ++ * recursion, in particular by going through "ordinary" nodes (that don't ++ * need to know whether the rest of the match failed) by a loop instead of ++ * by recursion. ++ */ ++static int /* 0 failure, 1 success */ ++regmatch(struct match_globals *g, char *prog) ++{ ++ register char *scan = prog; /* Current node. */ ++ char *next; /* Next node. */ ++ ++#ifdef DEBUG ++ if (scan != NULL && regnarrate) ++ fprintf(stderr, "%s(\n", regprop(scan)); ++#endif ++ while (scan != NULL) { ++#ifdef DEBUG ++ if (regnarrate) ++ fprintf(stderr, "%s...\n", regprop(scan)); ++#endif ++ next = regnext(g, scan); ++ ++ switch (OP(scan)) { ++ case BOL: ++ if (g->reginput != g->regbol) ++ return(0); ++ break; ++ case EOL: ++ if (*g->reginput != '\0') ++ return(0); ++ break; ++ case ANY: ++ if (*g->reginput == '\0') ++ return(0); ++ g->reginput++; ++ break; ++ case EXACTLY: { ++ register int len; ++ register char *opnd; ++ ++ opnd = OPERAND(scan); ++ /* Inline the first character, for speed. */ ++ if (*opnd != *g->reginput) ++ return(0); ++ len = strlen(opnd); ++ if (len > 1 && strncmp(opnd, g->reginput, len) != 0) ++ return(0); ++ g->reginput += len; ++ } ++ break; ++ case ANYOF: ++ if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) == NULL) ++ return(0); ++ g->reginput++; ++ break; ++ case ANYBUT: ++ if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) != NULL) ++ return(0); ++ g->reginput++; ++ break; ++ case NOTHING: ++ case BACK: ++ break; ++ case OPEN+1: ++ case OPEN+2: ++ case OPEN+3: ++ case OPEN+4: ++ case OPEN+5: ++ case OPEN+6: ++ case OPEN+7: ++ case OPEN+8: ++ case OPEN+9: { ++ register int no; ++ register char *save; ++ ++ no = OP(scan) - OPEN; ++ save = g->reginput; ++ ++ if (regmatch(g, next)) { ++ /* ++ * Don't set startp if some later ++ * invocation of the same parentheses ++ * already has. ++ */ ++ if (g->regstartp[no] == NULL) ++ g->regstartp[no] = save; ++ return(1); ++ } else ++ return(0); ++ } ++ break; ++ case CLOSE+1: ++ case CLOSE+2: ++ case CLOSE+3: ++ case CLOSE+4: ++ case CLOSE+5: ++ case CLOSE+6: ++ case CLOSE+7: ++ case CLOSE+8: ++ case CLOSE+9: ++ { ++ register int no; ++ register char *save; ++ ++ no = OP(scan) - CLOSE; ++ save = g->reginput; ++ ++ if (regmatch(g, next)) { ++ /* ++ * Don't set endp if some later ++ * invocation of the same parentheses ++ * already has. ++ */ ++ if (g->regendp[no] == NULL) ++ g->regendp[no] = save; ++ return(1); ++ } else ++ return(0); ++ } ++ break; ++ case BRANCH: { ++ register char *save; ++ ++ if (OP(next) != BRANCH) /* No choice. */ ++ next = OPERAND(scan); /* Avoid recursion. */ ++ else { ++ do { ++ save = g->reginput; ++ if (regmatch(g, OPERAND(scan))) ++ return(1); ++ g->reginput = save; ++ scan = regnext(g, scan); ++ } while (scan != NULL && OP(scan) == BRANCH); ++ return(0); ++ /* NOTREACHED */ ++ } ++ } ++ break; ++ case STAR: ++ case PLUS: { ++ register char nextch; ++ register int no; ++ register char *save; ++ register int min; ++ ++ /* ++ * Lookahead to avoid useless match attempts ++ * when we know what character comes next. ++ */ ++ nextch = '\0'; ++ if (OP(next) == EXACTLY) ++ nextch = *OPERAND(next); ++ min = (OP(scan) == STAR) ? 0 : 1; ++ save = g->reginput; ++ no = regrepeat(g, OPERAND(scan)); ++ while (no >= min) { ++ /* If it could work, try it. */ ++ if (nextch == '\0' || *g->reginput == nextch) ++ if (regmatch(g, next)) ++ return(1); ++ /* Couldn't or didn't -- back up. */ ++ no--; ++ g->reginput = save + no; ++ } ++ return(0); ++ } ++ break; ++ case END: ++ return(1); /* Success! */ ++ break; ++ default: ++ printk("<3>Regexp: memory corruption\n"); ++ return(0); ++ break; ++ } ++ ++ scan = next; ++ } ++ ++ /* ++ * We get here only if there's trouble -- normally "case END" is ++ * the terminating point. ++ */ ++ printk("<3>Regexp: corrupted pointers\n"); ++ return(0); ++} ++ ++/* ++ - regrepeat - repeatedly match something simple, report how many ++ */ ++static int ++regrepeat(struct match_globals *g, char *p) ++{ ++ register int count = 0; ++ register char *scan; ++ register char *opnd; ++ ++ scan = g->reginput; ++ opnd = OPERAND(p); ++ switch (OP(p)) { ++ case ANY: ++ count = strlen(scan); ++ scan += count; ++ break; ++ case EXACTLY: ++ while (*opnd == *scan) { ++ count++; ++ scan++; ++ } ++ break; ++ case ANYOF: ++ while (*scan != '\0' && strchr(opnd, *scan) != NULL) { ++ count++; ++ scan++; ++ } ++ break; ++ case ANYBUT: ++ while (*scan != '\0' && strchr(opnd, *scan) == NULL) { ++ count++; ++ scan++; ++ } ++ break; ++ default: /* Oh dear. Called inappropriately. */ ++ printk("<3>Regexp: internal foulup\n"); ++ count = 0; /* Best compromise. */ ++ break; ++ } ++ g->reginput = scan; ++ ++ return(count); ++} ++ ++/* ++ - regnext - dig the "next" pointer out of a node ++ */ ++static char* ++regnext(struct match_globals *g, char *p) ++{ ++ register int offset; ++ ++ if (p == &g->regdummy) ++ return(NULL); ++ ++ offset = NEXT(p); ++ if (offset == 0) ++ return(NULL); ++ ++ if (OP(p) == BACK) ++ return(p-offset); ++ else ++ return(p+offset); ++} ++ ++#ifdef DEBUG ++ ++STATIC char *regprop(); ++ ++/* ++ - regdump - dump a regexp onto stdout in vaguely comprehensible form ++ */ ++void ++regdump(regexp *r) ++{ ++ register char *s; ++ register char op = EXACTLY; /* Arbitrary non-END op. */ ++ register char *next; ++ /* extern char *strchr(); */ ++ ++ ++ s = r->program + 1; ++ while (op != END) { /* While that wasn't END last time... */ ++ op = OP(s); ++ printf("%2d%s", s-r->program, regprop(s)); /* Where, what. */ ++ next = regnext(s); ++ if (next == NULL) /* Next ptr. */ ++ printf("(0)"); ++ else ++ printf("(%d)", (s-r->program)+(next-s)); ++ s += 3; ++ if (op == ANYOF || op == ANYBUT || op == EXACTLY) { ++ /* Literal string, where present. */ ++ while (*s != '\0') { ++ putchar(*s); ++ s++; ++ } ++ s++; ++ } ++ putchar('\n'); ++ } ++ ++ /* Header fields of interest. */ ++ if (r->regstart != '\0') ++ printf("start `%c' ", r->regstart); ++ if (r->reganch) ++ printf("anchored "); ++ if (r->regmust != NULL) ++ printf("must have \"%s\"", r->regmust); ++ printf("\n"); ++} ++ ++/* ++ - regprop - printable representation of opcode ++ */ ++static char * ++regprop(char *op) ++{ ++#define BUFLEN 50 ++ register char *p; ++ static char buf[BUFLEN]; ++ ++ strcpy(buf, ":"); ++ ++ switch (OP(op)) { ++ case BOL: ++ p = "BOL"; ++ break; ++ case EOL: ++ p = "EOL"; ++ break; ++ case ANY: ++ p = "ANY"; ++ break; ++ case ANYOF: ++ p = "ANYOF"; ++ break; ++ case ANYBUT: ++ p = "ANYBUT"; ++ break; ++ case BRANCH: ++ p = "BRANCH"; ++ break; ++ case EXACTLY: ++ p = "EXACTLY"; ++ break; ++ case NOTHING: ++ p = "NOTHING"; ++ break; ++ case BACK: ++ p = "BACK"; ++ break; ++ case END: ++ p = "END"; ++ break; ++ case OPEN+1: ++ case OPEN+2: ++ case OPEN+3: ++ case OPEN+4: ++ case OPEN+5: ++ case OPEN+6: ++ case OPEN+7: ++ case OPEN+8: ++ case OPEN+9: ++ snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "OPEN%d", OP(op)-OPEN); ++ p = NULL; ++ break; ++ case CLOSE+1: ++ case CLOSE+2: ++ case CLOSE+3: ++ case CLOSE+4: ++ case CLOSE+5: ++ case CLOSE+6: ++ case CLOSE+7: ++ case CLOSE+8: ++ case CLOSE+9: ++ snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "CLOSE%d", OP(op)-CLOSE); ++ p = NULL; ++ break; ++ case STAR: ++ p = "STAR"; ++ break; ++ case PLUS: ++ p = "PLUS"; ++ break; ++ default: ++ printk("<3>Regexp: corrupted opcode\n"); ++ break; ++ } ++ if (p != NULL) ++ strncat(buf, p, BUFLEN-strlen(buf)); ++ return(buf); ++} ++#endif ++ ++ +--- /dev/null ++++ b/net/netfilter/regexp/regexp.h +@@ -0,0 +1,41 @@ ++/* ++ * Definitions etc. for regexp(3) routines. ++ * ++ * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof], ++ * not the System V one. ++ */ ++ ++#ifndef REGEXP_H ++#define REGEXP_H ++ ++ ++/* ++http://www.opensource.apple.com/darwinsource/10.3/expect-1/expect/expect.h , ++which contains a version of this library, says: ++ ++ * ++ * NSUBEXP must be at least 10, and no greater than 117 or the parser ++ * will not work properly. ++ * ++ ++However, it looks rather like this library is limited to 10. If you think ++otherwise, let us know. ++*/ ++ ++#define NSUBEXP 10 ++typedef struct regexp { ++ char *startp[NSUBEXP]; ++ char *endp[NSUBEXP]; ++ char regstart; /* Internal use only. */ ++ char reganch; /* Internal use only. */ ++ char *regmust; /* Internal use only. */ ++ int regmlen; /* Internal use only. */ ++ char program[1]; /* Unwarranted chumminess with compiler. */ ++} regexp; ++ ++regexp * regcomp(char *exp, int *patternsize); ++int regexec(regexp *prog, char *string); ++void regsub(regexp *prog, char *source, char *dest); ++void regerror(char *s); ++ ++#endif +--- /dev/null ++++ b/net/netfilter/regexp/regmagic.h +@@ -0,0 +1,5 @@ ++/* ++ * The first byte of the regexp internal "program" is actually this magic ++ * number; the start node begins in the second byte. ++ */ ++#define MAGIC 0234 +--- /dev/null ++++ b/net/netfilter/regexp/regsub.c +@@ -0,0 +1,95 @@ ++/* ++ * regsub ++ * @(#)regsub.c 1.3 of 2 April 86 ++ * ++ * Copyright (c) 1986 by University of Toronto. ++ * Written by Henry Spencer. Not derived from licensed software. ++ * ++ * Permission is granted to anyone to use this software for any ++ * purpose on any computer system, and to redistribute it freely, ++ * subject to the following restrictions: ++ * ++ * 1. The author is not responsible for the consequences of use of ++ * this software, no matter how awful, even if they arise ++ * from defects in it. ++ * ++ * 2. The origin of this software must not be misrepresented, either ++ * by explicit claim or by omission. ++ * ++ * 3. Altered versions must be plainly marked as such, and must not ++ * be misrepresented as being the original software. ++ * ++ * ++ * This code was modified by Ethan Sommer to work within the kernel ++ * (it now uses kmalloc etc..) ++ * ++ */ ++#include "regexp.h" ++#include "regmagic.h" ++#include ++ ++ ++#ifndef CHARBITS ++#define UCHARAT(p) ((int)*(unsigned char *)(p)) ++#else ++#define UCHARAT(p) ((int)*(p)&CHARBITS) ++#endif ++ ++#if 0 ++//void regerror(char * s) ++//{ ++// printk("regexp(3): %s", s); ++// /* NOTREACHED */ ++//} ++#endif ++ ++/* ++ - regsub - perform substitutions after a regexp match ++ */ ++void ++regsub(regexp * prog, char * source, char * dest) ++{ ++ register char *src; ++ register char *dst; ++ register char c; ++ register int no; ++ register int len; ++ ++ /* Not necessary and gcc doesn't like it -MLS */ ++ /*extern char *strncpy();*/ ++ ++ if (prog == NULL || source == NULL || dest == NULL) { ++ regerror("NULL parm to regsub"); ++ return; ++ } ++ if (UCHARAT(prog->program) != MAGIC) { ++ regerror("damaged regexp fed to regsub"); ++ return; ++ } ++ ++ src = source; ++ dst = dest; ++ while ((c = *src++) != '\0') { ++ if (c == '&') ++ no = 0; ++ else if (c == '\\' && '0' <= *src && *src <= '9') ++ no = *src++ - '0'; ++ else ++ no = -1; ++ ++ if (no < 0) { /* Ordinary character. */ ++ if (c == '\\' && (*src == '\\' || *src == '&')) ++ c = *src++; ++ *dst++ = c; ++ } else if (prog->startp[no] != NULL && prog->endp[no] != NULL) { ++ len = prog->endp[no] - prog->startp[no]; ++ (void) strncpy(dst, prog->startp[no], len); ++ dst += len; ++ if (len != 0 && *(dst-1) == '\0') { /* strncpy hit NUL. */ ++ regerror("damaged match string"); ++ return; ++ } ++ } ++ } ++ *dst++ = '\0'; ++} +--- /dev/null ++++ b/net/netfilter/xt_layer7.c +@@ -0,0 +1,666 @@ ++/* ++ Kernel module to match application layer (OSI layer 7) data in connections. ++ ++ http://l7-filter.sf.net ++ ++ (C) 2003-2009 Matthew Strait and Ethan Sommer. ++ ++ This program is free software; you can redistribute it and/or ++ modify it under the terms of the GNU General Public License ++ as published by the Free Software Foundation; either version ++ 2 of the License, or (at your option) any later version. ++ http://www.gnu.org/licenses/gpl.txt ++ ++ Based on ipt_string.c (C) 2000 Emmanuel Roger , ++ xt_helper.c (C) 2002 Harald Welte and cls_layer7.c (C) 2003 Matthew Strait, ++ Ethan Sommer, Justin Levandoski. ++*/ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 27) ++#include ++#include ++#endif ++#include ++#include ++#include ++#include ++ ++#include "regexp/regexp.c" ++ ++MODULE_LICENSE("GPL"); ++MODULE_AUTHOR("Matthew Strait , Ethan Sommer "); ++MODULE_DESCRIPTION("iptables application layer match module"); ++MODULE_ALIAS("ipt_layer7"); ++MODULE_VERSION("2.21"); ++ ++static int maxdatalen = 2048; // this is the default ++module_param(maxdatalen, int, 0444); ++MODULE_PARM_DESC(maxdatalen, "maximum bytes of data looked at by l7-filter"); ++#ifdef CONFIG_NETFILTER_XT_MATCH_LAYER7_DEBUG ++ #define DPRINTK(format,args...) printk(format,##args) ++#else ++ #define DPRINTK(format,args...) ++#endif ++ ++/* Number of packets whose data we look at. ++This can be modified through /proc/net/layer7_numpackets */ ++static int num_packets = 10; ++ ++static struct pattern_cache { ++ char * regex_string; ++ regexp * pattern; ++ struct pattern_cache * next; ++} * first_pattern_cache = NULL; ++ ++DEFINE_SPINLOCK(l7_lock); ++ ++static int total_acct_packets(struct nf_conn *ct) ++{ ++#if LINUX_VERSION_CODE <= KERNEL_VERSION(2, 6, 26) ++ BUG_ON(ct == NULL); ++ return (ct->counters[IP_CT_DIR_ORIGINAL].packets + ct->counters[IP_CT_DIR_REPLY].packets); ++#else ++ struct nf_conn_counter *acct; ++ ++ BUG_ON(ct == NULL); ++ acct = nf_conn_acct_find(ct); ++ if (!acct) ++ return 0; ++ return (acct[IP_CT_DIR_ORIGINAL].packets + acct[IP_CT_DIR_REPLY].packets); ++#endif ++} ++ ++#ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG ++/* Converts an unfriendly string into a friendly one by ++replacing unprintables with periods and all whitespace with " ". */ ++static char * friendly_print(unsigned char * s) ++{ ++ char * f = kmalloc(strlen(s) + 1, GFP_ATOMIC); ++ int i; ++ ++ if(!f) { ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: out of memory in " ++ "friendly_print, bailing.\n"); ++ return NULL; ++ } ++ ++ for(i = 0; i < strlen(s); i++){ ++ if(isprint(s[i]) && s[i] < 128) f[i] = s[i]; ++ else if(isspace(s[i])) f[i] = ' '; ++ else f[i] = '.'; ++ } ++ f[i] = '\0'; ++ return f; ++} ++ ++static char dec2hex(int i) ++{ ++ switch (i) { ++ case 0 ... 9: ++ return (i + '0'); ++ break; ++ case 10 ... 15: ++ return (i - 10 + 'a'); ++ break; ++ default: ++ if (net_ratelimit()) ++ printk("layer7: Problem in dec2hex\n"); ++ return '\0'; ++ } ++} ++ ++static char * hex_print(unsigned char * s) ++{ ++ char * g = kmalloc(strlen(s)*3 + 1, GFP_ATOMIC); ++ int i; ++ ++ if(!g) { ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: out of memory in hex_print, " ++ "bailing.\n"); ++ return NULL; ++ } ++ ++ for(i = 0; i < strlen(s); i++) { ++ g[i*3 ] = dec2hex(s[i]/16); ++ g[i*3 + 1] = dec2hex(s[i]%16); ++ g[i*3 + 2] = ' '; ++ } ++ g[i*3] = '\0'; ++ ++ return g; ++} ++#endif // DEBUG ++ ++/* Use instead of regcomp. As we expect to be seeing the same regexps over and ++over again, it make sense to cache the results. */ ++static regexp * compile_and_cache(const char * regex_string, ++ const char * protocol) ++{ ++ struct pattern_cache * node = first_pattern_cache; ++ struct pattern_cache * last_pattern_cache = first_pattern_cache; ++ struct pattern_cache * tmp; ++ unsigned int len; ++ ++ while (node != NULL) { ++ if (!strcmp(node->regex_string, regex_string)) ++ return node->pattern; ++ ++ last_pattern_cache = node;/* points at the last non-NULL node */ ++ node = node->next; ++ } ++ ++ /* If we reach the end of the list, then we have not yet cached ++ the pattern for this regex. Let's do that now. ++ Be paranoid about running out of memory to avoid list corruption. */ ++ tmp = kmalloc(sizeof(struct pattern_cache), GFP_ATOMIC); ++ ++ if(!tmp) { ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: out of memory in " ++ "compile_and_cache, bailing.\n"); ++ return NULL; ++ } ++ ++ tmp->regex_string = kmalloc(strlen(regex_string) + 1, GFP_ATOMIC); ++ tmp->pattern = kmalloc(sizeof(struct regexp), GFP_ATOMIC); ++ tmp->next = NULL; ++ ++ if(!tmp->regex_string || !tmp->pattern) { ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: out of memory in " ++ "compile_and_cache, bailing.\n"); ++ kfree(tmp->regex_string); ++ kfree(tmp->pattern); ++ kfree(tmp); ++ return NULL; ++ } ++ ++ /* Ok. The new node is all ready now. */ ++ node = tmp; ++ ++ if(first_pattern_cache == NULL) /* list is empty */ ++ first_pattern_cache = node; /* make node the beginning */ ++ else ++ last_pattern_cache->next = node; /* attach node to the end */ ++ ++ /* copy the string and compile the regex */ ++ len = strlen(regex_string); ++ DPRINTK("About to compile this: \"%s\"\n", regex_string); ++ node->pattern = regcomp((char *)regex_string, &len); ++ if ( !node->pattern ) { ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: Error compiling regexp " ++ "\"%s\" (%s)\n", ++ regex_string, protocol); ++ /* pattern is now cached as NULL, so we won't try again. */ ++ } ++ ++ strcpy(node->regex_string, regex_string); ++ return node->pattern; ++} ++ ++static int can_handle(const struct sk_buff *skb) ++{ ++ if(!ip_hdr(skb)) /* not IP */ ++ return 0; ++ if(ip_hdr(skb)->protocol != IPPROTO_TCP && ++ ip_hdr(skb)->protocol != IPPROTO_UDP && ++ ip_hdr(skb)->protocol != IPPROTO_ICMP) ++ return 0; ++ return 1; ++} ++ ++/* Returns offset the into the skb->data that the application data starts */ ++static int app_data_offset(const struct sk_buff *skb) ++{ ++ /* In case we are ported somewhere (ebtables?) where ip_hdr(skb) ++ isn't set, this can be gotten from 4*(skb->data[0] & 0x0f) as well. */ ++ int ip_hl = 4*ip_hdr(skb)->ihl; ++ ++ if( ip_hdr(skb)->protocol == IPPROTO_TCP ) { ++ /* 12 == offset into TCP header for the header length field. ++ Can't get this with skb->h.th->doff because the tcphdr ++ struct doesn't get set when routing (this is confirmed to be ++ true in Netfilter as well as QoS.) */ ++ int tcp_hl = 4*(skb->data[ip_hl + 12] >> 4); ++ ++ return ip_hl + tcp_hl; ++ } else if( ip_hdr(skb)->protocol == IPPROTO_UDP ) { ++ return ip_hl + 8; /* UDP header is always 8 bytes */ ++ } else if( ip_hdr(skb)->protocol == IPPROTO_ICMP ) { ++ return ip_hl + 8; /* ICMP header is 8 bytes */ ++ } else { ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: tried to handle unknown " ++ "protocol!\n"); ++ return ip_hl + 8; /* something reasonable */ ++ } ++} ++ ++/* handles whether there's a match when we aren't appending data anymore */ ++static int match_no_append(struct nf_conn * conntrack, ++ struct nf_conn * master_conntrack, ++ enum ip_conntrack_info ctinfo, ++ enum ip_conntrack_info master_ctinfo, ++ const struct xt_layer7_info * info) ++{ ++ /* If we're in here, throw the app data away */ ++ if(master_conntrack->layer7.app_data != NULL) { ++ ++ #ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG ++ if(!master_conntrack->layer7.app_proto) { ++ char * f = ++ friendly_print(master_conntrack->layer7.app_data); ++ char * g = ++ hex_print(master_conntrack->layer7.app_data); ++ DPRINTK("\nl7-filter gave up after %d bytes " ++ "(%d packets):\n%s\n", ++ strlen(f), total_acct_packets(master_conntrack), f); ++ kfree(f); ++ DPRINTK("In hex: %s\n", g); ++ kfree(g); ++ } ++ #endif ++ ++ kfree(master_conntrack->layer7.app_data); ++ master_conntrack->layer7.app_data = NULL; /* don't free again */ ++ } ++ ++ if(master_conntrack->layer7.app_proto){ ++ /* Here child connections set their .app_proto (for /proc) */ ++ if(!conntrack->layer7.app_proto) { ++ conntrack->layer7.app_proto = ++ kmalloc(strlen(master_conntrack->layer7.app_proto)+1, ++ GFP_ATOMIC); ++ if(!conntrack->layer7.app_proto){ ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: out of memory " ++ "in match_no_append, " ++ "bailing.\n"); ++ return 1; ++ } ++ strcpy(conntrack->layer7.app_proto, ++ master_conntrack->layer7.app_proto); ++ } ++ ++ return (!strcmp(master_conntrack->layer7.app_proto, ++ info->protocol)); ++ } ++ else { ++ /* If not classified, set to "unknown" to distinguish from ++ connections that are still being tested. */ ++ master_conntrack->layer7.app_proto = ++ kmalloc(strlen("unknown")+1, GFP_ATOMIC); ++ if(!master_conntrack->layer7.app_proto){ ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: out of memory in " ++ "match_no_append, bailing.\n"); ++ return 1; ++ } ++ strcpy(master_conntrack->layer7.app_proto, "unknown"); ++ return 0; ++ } ++} ++ ++/* add the new app data to the conntrack. Return number of bytes added. */ ++static int add_data(struct nf_conn * master_conntrack, ++ char * app_data, int appdatalen) ++{ ++ int length = 0, i; ++ int oldlength = master_conntrack->layer7.app_data_len; ++ ++ /* This is a fix for a race condition by Deti Fliegl. However, I'm not ++ clear on whether the race condition exists or whether this really ++ fixes it. I might just be being dense... Anyway, if it's not really ++ a fix, all it does is waste a very small amount of time. */ ++ if(!master_conntrack->layer7.app_data) return 0; ++ ++ /* Strip nulls. Make everything lower case (our regex lib doesn't ++ do case insensitivity). Add it to the end of the current data. */ ++ for(i = 0; i < maxdatalen-oldlength-1 && ++ i < appdatalen; i++) { ++ if(app_data[i] != '\0') { ++ /* the kernel version of tolower mungs 'upper ascii' */ ++ master_conntrack->layer7.app_data[length+oldlength] = ++ isascii(app_data[i])? ++ tolower(app_data[i]) : app_data[i]; ++ length++; ++ } ++ } ++ ++ master_conntrack->layer7.app_data[length+oldlength] = '\0'; ++ master_conntrack->layer7.app_data_len = length + oldlength; ++ ++ return length; ++} ++ ++/* taken from drivers/video/modedb.c */ ++static int my_atoi(const char *s) ++{ ++ int val = 0; ++ ++ for (;; s++) { ++ switch (*s) { ++ case '0'...'9': ++ val = 10*val+(*s-'0'); ++ break; ++ default: ++ return val; ++ } ++ } ++} ++ ++/* write out num_packets to userland. */ ++static int layer7_read_proc(char* page, char ** start, off_t off, int count, ++ int* eof, void * data) ++{ ++ if(num_packets > 99 && net_ratelimit()) ++ printk(KERN_ERR "layer7: NOT REACHED. num_packets too big\n"); ++ ++ page[0] = num_packets/10 + '0'; ++ page[1] = num_packets%10 + '0'; ++ page[2] = '\n'; ++ page[3] = '\0'; ++ ++ *eof=1; ++ ++ return 3; ++} ++ ++/* Read in num_packets from userland */ ++static int layer7_write_proc(struct file* file, const char* buffer, ++ unsigned long count, void *data) ++{ ++ char * foo = kmalloc(count, GFP_ATOMIC); ++ ++ if(!foo){ ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: out of memory, bailing. " ++ "num_packets unchanged.\n"); ++ return count; ++ } ++ ++ if(copy_from_user(foo, buffer, count)) { ++ return -EFAULT; ++ } ++ ++ ++ num_packets = my_atoi(foo); ++ kfree (foo); ++ ++ /* This has an arbitrary limit to make the math easier. I'm lazy. ++ But anyway, 99 is a LOT! If you want more, you're doing it wrong! */ ++ if(num_packets > 99) { ++ printk(KERN_WARNING "layer7: num_packets can't be > 99.\n"); ++ num_packets = 99; ++ } else if(num_packets < 1) { ++ printk(KERN_WARNING "layer7: num_packets can't be < 1.\n"); ++ num_packets = 1; ++ } ++ ++ return count; ++} ++ ++static bool ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28) ++match(const struct sk_buff *skbin, const struct xt_match_param *par) ++#else ++match(const struct sk_buff *skbin, ++ const struct net_device *in, ++ const struct net_device *out, ++ const struct xt_match *match, ++ const void *matchinfo, ++ int offset, ++ unsigned int protoff, ++ bool *hotdrop) ++#endif ++{ ++ /* sidestep const without getting a compiler warning... */ ++ struct sk_buff * skb = (struct sk_buff *)skbin; ++ ++ const struct xt_layer7_info * info = ++ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28) ++ par->matchinfo; ++ #else ++ matchinfo; ++ #endif ++ ++ enum ip_conntrack_info master_ctinfo, ctinfo; ++ struct nf_conn *master_conntrack, *conntrack; ++ unsigned char * app_data; ++ unsigned int pattern_result, appdatalen; ++ regexp * comppattern; ++ ++ /* Be paranoid/incompetent - lock the entire match function. */ ++ spin_lock_bh(&l7_lock); ++ ++ if(!can_handle(skb)){ ++ DPRINTK("layer7: This is some protocol I can't handle.\n"); ++ spin_unlock_bh(&l7_lock); ++ return info->invert; ++ } ++ ++ /* Treat parent & all its children together as one connection, except ++ for the purpose of setting conntrack->layer7.app_proto in the actual ++ connection. This makes /proc/net/ip_conntrack more satisfying. */ ++ if(!(conntrack = nf_ct_get(skb, &ctinfo)) || ++ !(master_conntrack=nf_ct_get(skb,&master_ctinfo))){ ++ DPRINTK("layer7: couldn't get conntrack.\n"); ++ spin_unlock_bh(&l7_lock); ++ return info->invert; ++ } ++ ++ /* Try to get a master conntrack (and its master etc) for FTP, etc. */ ++ while (master_ct(master_conntrack) != NULL) ++ master_conntrack = master_ct(master_conntrack); ++ ++ /* if we've classified it or seen too many packets */ ++ if(total_acct_packets(master_conntrack) > num_packets || ++ master_conntrack->layer7.app_proto) { ++ ++ pattern_result = match_no_append(conntrack, master_conntrack, ++ ctinfo, master_ctinfo, info); ++ ++ /* skb->cb[0] == seen. Don't do things twice if there are ++ multiple l7 rules. I'm not sure that using cb for this purpose ++ is correct, even though it says "put your private variables ++ there". But it doesn't look like it is being used for anything ++ else in the skbs that make it here. */ ++ skb->cb[0] = 1; /* marking it seen here's probably irrelevant */ ++ ++ spin_unlock_bh(&l7_lock); ++ return (pattern_result ^ info->invert); ++ } ++ ++ if(skb_is_nonlinear(skb)){ ++ if(skb_linearize(skb) != 0){ ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: failed to linearize " ++ "packet, bailing.\n"); ++ spin_unlock_bh(&l7_lock); ++ return info->invert; ++ } ++ } ++ ++ /* now that the skb is linearized, it's safe to set these. */ ++ app_data = skb->data + app_data_offset(skb); ++ appdatalen = skb_tail_pointer(skb) - app_data; ++ ++ /* the return value gets checked later, when we're ready to use it */ ++ comppattern = compile_and_cache(info->pattern, info->protocol); ++ ++ /* On the first packet of a connection, allocate space for app data */ ++ if(total_acct_packets(master_conntrack) == 1 && !skb->cb[0] && ++ !master_conntrack->layer7.app_data){ ++ master_conntrack->layer7.app_data = ++ kmalloc(maxdatalen, GFP_ATOMIC); ++ if(!master_conntrack->layer7.app_data){ ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: out of memory in " ++ "match, bailing.\n"); ++ spin_unlock_bh(&l7_lock); ++ return info->invert; ++ } ++ ++ master_conntrack->layer7.app_data[0] = '\0'; ++ } ++ ++ /* Can be here, but unallocated, if numpackets is increased near ++ the beginning of a connection */ ++ if(master_conntrack->layer7.app_data == NULL){ ++ spin_unlock_bh(&l7_lock); ++ return info->invert; /* unmatched */ ++ } ++ ++ if(!skb->cb[0]){ ++ int newbytes; ++ newbytes = add_data(master_conntrack, app_data, appdatalen); ++ ++ if(newbytes == 0) { /* didn't add any data */ ++ skb->cb[0] = 1; ++ /* Didn't match before, not going to match now */ ++ spin_unlock_bh(&l7_lock); ++ return info->invert; ++ } ++ } ++ ++ /* If looking for "unknown", then never match. "Unknown" means that ++ we've given up; we're still trying with these packets. */ ++ if(!strcmp(info->protocol, "unknown")) { ++ pattern_result = 0; ++ /* If looking for "unset", then always match. "Unset" means that we ++ haven't yet classified the connection. */ ++ } else if(!strcmp(info->protocol, "unset")) { ++ pattern_result = 2; ++ DPRINTK("layer7: matched unset: not yet classified " ++ "(%d/%d packets)\n", ++ total_acct_packets(master_conntrack), num_packets); ++ /* If the regexp failed to compile, don't bother running it */ ++ } else if(comppattern && ++ regexec(comppattern, master_conntrack->layer7.app_data)){ ++ DPRINTK("layer7: matched %s\n", info->protocol); ++ pattern_result = 1; ++ } else pattern_result = 0; ++ ++ if(pattern_result == 1) { ++ master_conntrack->layer7.app_proto = ++ kmalloc(strlen(info->protocol)+1, GFP_ATOMIC); ++ if(!master_conntrack->layer7.app_proto){ ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: out of memory in " ++ "match, bailing.\n"); ++ spin_unlock_bh(&l7_lock); ++ return (pattern_result ^ info->invert); ++ } ++ strcpy(master_conntrack->layer7.app_proto, info->protocol); ++ } else if(pattern_result > 1) { /* cleanup from "unset" */ ++ pattern_result = 1; ++ } ++ ++ /* mark the packet seen */ ++ skb->cb[0] = 1; ++ ++ spin_unlock_bh(&l7_lock); ++ return (pattern_result ^ info->invert); ++} ++ ++// load nf_conntrack_ipv4 ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28) ++static bool check(const struct xt_mtchk_param *par) ++{ ++ if (nf_ct_l3proto_try_module_get(par->match->family) < 0) { ++ printk(KERN_WARNING "can't load conntrack support for " ++ "proto=%d\n", par->match->family); ++#else ++static bool check(const char *tablename, const void *inf, ++ const struct xt_match *match, void *matchinfo, ++ unsigned int hook_mask) ++{ ++ if (nf_ct_l3proto_try_module_get(match->family) < 0) { ++ printk(KERN_WARNING "can't load conntrack support for " ++ "proto=%d\n", match->family); ++#endif ++ return 0; ++ } ++ return 1; ++} ++ ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28) ++ static void destroy(const struct xt_mtdtor_param *par) ++ { ++ nf_ct_l3proto_module_put(par->match->family); ++ } ++#else ++ static void destroy(const struct xt_match *match, void *matchinfo) ++ { ++ nf_ct_l3proto_module_put(match->family); ++ } ++#endif ++ ++static struct xt_match xt_layer7_match[] __read_mostly = { ++{ ++ .name = "layer7", ++ .family = AF_INET, ++ .checkentry = check, ++ .match = match, ++ .destroy = destroy, ++ .matchsize = sizeof(struct xt_layer7_info), ++ .me = THIS_MODULE ++} ++}; ++ ++static void layer7_cleanup_proc(void) ++{ ++ remove_proc_entry("layer7_numpackets", init_net.proc_net); ++} ++ ++/* register the proc file */ ++static void layer7_init_proc(void) ++{ ++ struct proc_dir_entry* entry; ++ entry = create_proc_entry("layer7_numpackets", 0644, init_net.proc_net); ++ entry->read_proc = layer7_read_proc; ++ entry->write_proc = layer7_write_proc; ++} ++ ++static int __init xt_layer7_init(void) ++{ ++ need_conntrack(); ++ ++ layer7_init_proc(); ++ if(maxdatalen < 1) { ++ printk(KERN_WARNING "layer7: maxdatalen can't be < 1, " ++ "using 1\n"); ++ maxdatalen = 1; ++ } ++ /* This is not a hard limit. It's just here to prevent people from ++ bringing their slow machines to a grinding halt. */ ++ else if(maxdatalen > 65536) { ++ printk(KERN_WARNING "layer7: maxdatalen can't be > 65536, " ++ "using 65536\n"); ++ maxdatalen = 65536; ++ } ++ return xt_register_matches(xt_layer7_match, ++ ARRAY_SIZE(xt_layer7_match)); ++} ++ ++static void __exit xt_layer7_fini(void) ++{ ++ layer7_cleanup_proc(); ++ xt_unregister_matches(xt_layer7_match, ARRAY_SIZE(xt_layer7_match)); ++} ++ ++module_init(xt_layer7_init); ++module_exit(xt_layer7_fini); diff --git a/target/linux/generic-2.6/patches-2.6.31/101-netfilter_layer7_pktmatch.patch b/target/linux/generic-2.6/patches-2.6.31/101-netfilter_layer7_pktmatch.patch new file mode 100644 index 0000000000..f65e301fd1 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/101-netfilter_layer7_pktmatch.patch @@ -0,0 +1,108 @@ +--- a/include/linux/netfilter/xt_layer7.h ++++ b/include/linux/netfilter/xt_layer7.h +@@ -8,6 +8,7 @@ struct xt_layer7_info { + char protocol[MAX_PROTOCOL_LEN]; + char pattern[MAX_PATTERN_LEN]; + u_int8_t invert; ++ u_int8_t pkt; + }; + + #endif /* _XT_LAYER7_H */ +--- a/net/netfilter/xt_layer7.c ++++ b/net/netfilter/xt_layer7.c +@@ -314,33 +314,35 @@ static int match_no_append(struct nf_con + } + + /* add the new app data to the conntrack. Return number of bytes added. */ +-static int add_data(struct nf_conn * master_conntrack, +- char * app_data, int appdatalen) ++static int add_datastr(char *target, int offset, char *app_data, int len) + { + int length = 0, i; +- int oldlength = master_conntrack->layer7.app_data_len; +- +- /* This is a fix for a race condition by Deti Fliegl. However, I'm not +- clear on whether the race condition exists or whether this really +- fixes it. I might just be being dense... Anyway, if it's not really +- a fix, all it does is waste a very small amount of time. */ +- if(!master_conntrack->layer7.app_data) return 0; ++ if (!target) return 0; + + /* Strip nulls. Make everything lower case (our regex lib doesn't + do case insensitivity). Add it to the end of the current data. */ +- for(i = 0; i < maxdatalen-oldlength-1 && +- i < appdatalen; i++) { ++ for(i = 0; i < maxdatalen-offset-1 && i < len; i++) { + if(app_data[i] != '\0') { + /* the kernel version of tolower mungs 'upper ascii' */ +- master_conntrack->layer7.app_data[length+oldlength] = ++ target[length+offset] = + isascii(app_data[i])? + tolower(app_data[i]) : app_data[i]; + length++; + } + } ++ target[length+offset] = '\0'; ++ ++ return length; ++} ++ ++/* add the new app data to the conntrack. Return number of bytes added. */ ++static int add_data(struct nf_conn * master_conntrack, ++ char * app_data, int appdatalen) ++{ ++ int length; + +- master_conntrack->layer7.app_data[length+oldlength] = '\0'; +- master_conntrack->layer7.app_data_len = length + oldlength; ++ length = add_datastr(master_conntrack->layer7.app_data, master_conntrack->layer7.app_data_len, app_data, appdatalen); ++ master_conntrack->layer7.app_data_len += length; + + return length; + } +@@ -438,7 +440,7 @@ match(const struct sk_buff *skbin, + + enum ip_conntrack_info master_ctinfo, ctinfo; + struct nf_conn *master_conntrack, *conntrack; +- unsigned char * app_data; ++ unsigned char *app_data, *tmp_data; + unsigned int pattern_result, appdatalen; + regexp * comppattern; + +@@ -466,8 +468,8 @@ match(const struct sk_buff *skbin, + master_conntrack = master_ct(master_conntrack); + + /* if we've classified it or seen too many packets */ +- if(total_acct_packets(master_conntrack) > num_packets || +- master_conntrack->layer7.app_proto) { ++ if(!info->pkt && (total_acct_packets(master_conntrack) > num_packets || ++ master_conntrack->layer7.app_proto)) { + + pattern_result = match_no_append(conntrack, master_conntrack, + ctinfo, master_ctinfo, info); +@@ -500,6 +502,25 @@ match(const struct sk_buff *skbin, + /* the return value gets checked later, when we're ready to use it */ + comppattern = compile_and_cache(info->pattern, info->protocol); + ++ if (info->pkt) { ++ tmp_data = kmalloc(maxdatalen, GFP_ATOMIC); ++ if(!tmp_data){ ++ if (net_ratelimit()) ++ printk(KERN_ERR "layer7: out of memory in match, bailing.\n"); ++ return info->invert; ++ } ++ ++ tmp_data[0] = '\0'; ++ add_datastr(tmp_data, 0, app_data, appdatalen); ++ pattern_result = ((comppattern && regexec(comppattern, tmp_data)) ? 1 : 0); ++ ++ kfree(tmp_data); ++ tmp_data = NULL; ++ spin_unlock_bh(&l7_lock); ++ ++ return (pattern_result ^ info->invert); ++ } ++ + /* On the first packet of a connection, allocate space for app data */ + if(total_acct_packets(master_conntrack) == 1 && !skb->cb[0] && + !master_conntrack->layer7.app_data){ diff --git a/target/linux/generic-2.6/patches-2.6.31/110-netfilter_match_speedup.patch b/target/linux/generic-2.6/patches-2.6.31/110-netfilter_match_speedup.patch new file mode 100644 index 0000000000..3dd114522a --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/110-netfilter_match_speedup.patch @@ -0,0 +1,121 @@ +--- a/include/linux/netfilter_ipv4/ip_tables.h ++++ b/include/linux/netfilter_ipv4/ip_tables.h +@@ -62,6 +62,7 @@ struct ipt_ip { + #define IPT_F_FRAG 0x01 /* Set if rule is a fragment rule */ + #define IPT_F_GOTO 0x02 /* Set if jump is a goto */ + #define IPT_F_MASK 0x03 /* All possible flag bits mask. */ ++#define IPT_F_NO_DEF_MATCH 0x80 /* Internal: no default match rules present */ + + /* Values for "inv" field in struct ipt_ip. */ + #define IPT_INV_VIA_IN 0x01 /* Invert the sense of IN IFACE. */ +--- a/net/ipv4/netfilter/ip_tables.c ++++ b/net/ipv4/netfilter/ip_tables.c +@@ -87,6 +87,9 @@ ip_packet_match(const struct iphdr *ip, + + #define FWINV(bool, invflg) ((bool) ^ !!(ipinfo->invflags & (invflg))) + ++ if (ipinfo->flags & IPT_F_NO_DEF_MATCH) ++ return true; ++ + if (FWINV((ip->saddr&ipinfo->smsk.s_addr) != ipinfo->src.s_addr, + IPT_INV_SRCIP) + || FWINV((ip->daddr&ipinfo->dmsk.s_addr) != ipinfo->dst.s_addr, +@@ -137,13 +140,35 @@ ip_packet_match(const struct iphdr *ip, + return false; + } + ++#undef FWINV + return true; + } + + static bool +-ip_checkentry(const struct ipt_ip *ip) ++ip_checkentry(struct ipt_ip *ip) + { +- if (ip->flags & ~IPT_F_MASK) { ++#define FWINV(bool, invflg) ((bool) || (ip->invflags & (invflg))) ++ ++ if (FWINV(ip->smsk.s_addr, IPT_INV_SRCIP) || ++ FWINV(ip->dmsk.s_addr, IPT_INV_DSTIP)) ++ goto has_match_rules; ++ ++ if (FWINV(!!((const unsigned long *)ip->iniface_mask)[0], ++ IPT_INV_VIA_IN) || ++ FWINV(!!((const unsigned long *)ip->outiface_mask)[0], ++ IPT_INV_VIA_OUT)) ++ goto has_match_rules; ++ ++ if (FWINV(ip->proto, IPT_INV_PROTO)) ++ goto has_match_rules; ++ ++ if (FWINV(ip->flags&IPT_F_FRAG, IPT_INV_FRAG)) ++ goto has_match_rules; ++ ++ ip->flags |= IPT_F_NO_DEF_MATCH; ++ ++has_match_rules: ++ if (ip->flags & ~(IPT_F_MASK|IPT_F_NO_DEF_MATCH)) { + duprintf("Unknown flag bits set: %08X\n", + ip->flags & ~IPT_F_MASK); + return false; +@@ -153,6 +178,8 @@ ip_checkentry(const struct ipt_ip *ip) + ip->invflags & ~IPT_INV_MASK); + return false; + } ++ ++#undef FWINV + return true; + } + +@@ -200,7 +227,6 @@ unconditional(const struct ipt_ip *ip) + return 0; + + return 1; +-#undef FWINV + } + + #if defined(CONFIG_NETFILTER_XT_TARGET_TRACE) || \ +@@ -326,8 +352,28 @@ ipt_do_table(struct sk_buff *skb, + struct xt_match_param mtpar; + struct xt_target_param tgpar; + +- /* Initialization */ + ip = ip_hdr(skb); ++ ++ IP_NF_ASSERT(table->valid_hooks & (1 << hook)); ++ xt_info_rdlock_bh(); ++ private = table->private; ++ table_base = private->entries[smp_processor_id()]; ++ e = get_entry(table_base, private->hook_entry[hook]); ++ ++ if (e->target_offset <= sizeof(struct ipt_entry) && ++ (e->ip.flags & IPT_F_NO_DEF_MATCH)) { ++ struct ipt_entry_target *t = ipt_get_target(e); ++ if (!t->u.kernel.target->target) { ++ int v = ((struct ipt_standard_target *)t)->verdict; ++ if ((v < 0) && (v != IPT_RETURN)) { ++ ADD_COUNTER(e->counters, ntohs(ip->tot_len), 1); ++ xt_info_rdunlock_bh(); ++ return (unsigned)(-v) - 1; ++ } ++ } ++ } ++ ++ /* Initialization */ + datalen = skb->len - ip->ihl * 4; + indev = in ? in->name : nulldevname; + outdev = out ? out->name : nulldevname; +@@ -345,13 +391,6 @@ ipt_do_table(struct sk_buff *skb, + mtpar.family = tgpar.family = NFPROTO_IPV4; + mtpar.hooknum = tgpar.hooknum = hook; + +- IP_NF_ASSERT(table->valid_hooks & (1 << hook)); +- xt_info_rdlock_bh(); +- private = table->private; +- table_base = private->entries[smp_processor_id()]; +- +- e = get_entry(table_base, private->hook_entry[hook]); +- + /* For return from builtin chain */ + back = get_entry(table_base, private->underflow[hook]); + diff --git a/target/linux/generic-2.6/patches-2.6.31/130-netfilter_ipset.patch b/target/linux/generic-2.6/patches-2.6.31/130-netfilter_ipset.patch new file mode 100644 index 0000000000..efe1041e6d --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/130-netfilter_ipset.patch @@ -0,0 +1,7688 @@ +--- a/include/linux/netfilter_ipv4/Kbuild ++++ b/include/linux/netfilter_ipv4/Kbuild +@@ -45,3 +45,14 @@ header-y += ipt_ttl.h + + unifdef-y += ip_queue.h + unifdef-y += ip_tables.h ++ ++unifdef-y += ip_set.h ++header-y += ip_set_iphash.h ++header-y += ip_set_ipmap.h ++header-y += ip_set_ipporthash.h ++unifdef-y += ip_set_iptree.h ++unifdef-y += ip_set_iptreemap.h ++header-y += ip_set_jhash.h ++header-y += ip_set_macipmap.h ++unifdef-y += ip_set_nethash.h ++header-y += ip_set_portmap.h +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ip_set.h +@@ -0,0 +1,498 @@ ++#ifndef _IP_SET_H ++#define _IP_SET_H ++ ++/* Copyright (C) 2000-2002 Joakim Axelsson ++ * Patrick Schaaf ++ * Martin Josefsson ++ * Copyright (C) 2003-2004 Jozsef Kadlecsik ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++#if 0 ++#define IP_SET_DEBUG ++#endif ++ ++/* ++ * A sockopt of such quality has hardly ever been seen before on the open ++ * market! This little beauty, hardly ever used: above 64, so it's ++ * traditionally used for firewalling, not touched (even once!) by the ++ * 2.0, 2.2 and 2.4 kernels! ++ * ++ * Comes with its own certificate of authenticity, valid anywhere in the ++ * Free world! ++ * ++ * Rusty, 19.4.2000 ++ */ ++#define SO_IP_SET 83 ++ ++/* ++ * Heavily modify by Joakim Axelsson 08.03.2002 ++ * - Made it more modulebased ++ * ++ * Additional heavy modifications by Jozsef Kadlecsik 22.02.2004 ++ * - bindings added ++ * - in order to "deal with" backward compatibility, renamed to ipset ++ */ ++ ++/* ++ * Used so that the kernel module and ipset-binary can match their versions ++ */ ++#define IP_SET_PROTOCOL_VERSION 2 ++ ++#define IP_SET_MAXNAMELEN 32 /* set names and set typenames */ ++ ++/* Lets work with our own typedef for representing an IP address. ++ * We hope to make the code more portable, possibly to IPv6... ++ * ++ * The representation works in HOST byte order, because most set types ++ * will perform arithmetic operations and compare operations. ++ * ++ * For now the type is an uint32_t. ++ * ++ * Make sure to ONLY use the functions when translating and parsing ++ * in order to keep the host byte order and make it more portable: ++ * parse_ip() ++ * parse_mask() ++ * parse_ipandmask() ++ * ip_tostring() ++ * (Joakim: where are they???) ++ */ ++ ++typedef uint32_t ip_set_ip_t; ++ ++/* Sets are identified by an id in kernel space. Tweak with ip_set_id_t ++ * and IP_SET_INVALID_ID if you want to increase the max number of sets. ++ */ ++typedef uint16_t ip_set_id_t; ++ ++#define IP_SET_INVALID_ID 65535 ++ ++/* How deep we follow bindings */ ++#define IP_SET_MAX_BINDINGS 6 ++ ++/* ++ * Option flags for kernel operations (ipt_set_info) ++ */ ++#define IPSET_SRC 0x01 /* Source match/add */ ++#define IPSET_DST 0x02 /* Destination match/add */ ++#define IPSET_MATCH_INV 0x04 /* Inverse matching */ ++ ++/* ++ * Set features ++ */ ++#define IPSET_TYPE_IP 0x01 /* IP address type of set */ ++#define IPSET_TYPE_PORT 0x02 /* Port type of set */ ++#define IPSET_DATA_SINGLE 0x04 /* Single data storage */ ++#define IPSET_DATA_DOUBLE 0x08 /* Double data storage */ ++ ++/* Reserved keywords */ ++#define IPSET_TOKEN_DEFAULT ":default:" ++#define IPSET_TOKEN_ALL ":all:" ++ ++/* SO_IP_SET operation constants, and their request struct types. ++ * ++ * Operation ids: ++ * 0-99: commands with version checking ++ * 100-199: add/del/test/bind/unbind ++ * 200-299: list, save, restore ++ */ ++ ++/* Single shot operations: ++ * version, create, destroy, flush, rename and swap ++ * ++ * Sets are identified by name. ++ */ ++ ++#define IP_SET_REQ_STD \ ++ unsigned op; \ ++ unsigned version; \ ++ char name[IP_SET_MAXNAMELEN] ++ ++#define IP_SET_OP_CREATE 0x00000001 /* Create a new (empty) set */ ++struct ip_set_req_create { ++ IP_SET_REQ_STD; ++ char typename[IP_SET_MAXNAMELEN]; ++}; ++ ++#define IP_SET_OP_DESTROY 0x00000002 /* Remove a (empty) set */ ++struct ip_set_req_std { ++ IP_SET_REQ_STD; ++}; ++ ++#define IP_SET_OP_FLUSH 0x00000003 /* Remove all IPs in a set */ ++/* Uses ip_set_req_std */ ++ ++#define IP_SET_OP_RENAME 0x00000004 /* Rename a set */ ++/* Uses ip_set_req_create */ ++ ++#define IP_SET_OP_SWAP 0x00000005 /* Swap two sets */ ++/* Uses ip_set_req_create */ ++ ++union ip_set_name_index { ++ char name[IP_SET_MAXNAMELEN]; ++ ip_set_id_t index; ++}; ++ ++#define IP_SET_OP_GET_BYNAME 0x00000006 /* Get set index by name */ ++struct ip_set_req_get_set { ++ unsigned op; ++ unsigned version; ++ union ip_set_name_index set; ++}; ++ ++#define IP_SET_OP_GET_BYINDEX 0x00000007 /* Get set name by index */ ++/* Uses ip_set_req_get_set */ ++ ++#define IP_SET_OP_VERSION 0x00000100 /* Ask kernel version */ ++struct ip_set_req_version { ++ unsigned op; ++ unsigned version; ++}; ++ ++/* Double shots operations: ++ * add, del, test, bind and unbind. ++ * ++ * First we query the kernel to get the index and type of the target set, ++ * then issue the command. Validity of IP is checked in kernel in order ++ * to minimalize sockopt operations. ++ */ ++ ++/* Get minimal set data for add/del/test/bind/unbind IP */ ++#define IP_SET_OP_ADT_GET 0x00000010 /* Get set and type */ ++struct ip_set_req_adt_get { ++ unsigned op; ++ unsigned version; ++ union ip_set_name_index set; ++ char typename[IP_SET_MAXNAMELEN]; ++}; ++ ++#define IP_SET_REQ_BYINDEX \ ++ unsigned op; \ ++ ip_set_id_t index; ++ ++struct ip_set_req_adt { ++ IP_SET_REQ_BYINDEX; ++}; ++ ++#define IP_SET_OP_ADD_IP 0x00000101 /* Add an IP to a set */ ++/* Uses ip_set_req_adt, with type specific addage */ ++ ++#define IP_SET_OP_DEL_IP 0x00000102 /* Remove an IP from a set */ ++/* Uses ip_set_req_adt, with type specific addage */ ++ ++#define IP_SET_OP_TEST_IP 0x00000103 /* Test an IP in a set */ ++/* Uses ip_set_req_adt, with type specific addage */ ++ ++#define IP_SET_OP_BIND_SET 0x00000104 /* Bind an IP to a set */ ++/* Uses ip_set_req_bind, with type specific addage */ ++struct ip_set_req_bind { ++ IP_SET_REQ_BYINDEX; ++ char binding[IP_SET_MAXNAMELEN]; ++}; ++ ++#define IP_SET_OP_UNBIND_SET 0x00000105 /* Unbind an IP from a set */ ++/* Uses ip_set_req_bind, with type speficic addage ++ * index = 0 means unbinding for all sets */ ++ ++#define IP_SET_OP_TEST_BIND_SET 0x00000106 /* Test binding an IP to a set */ ++/* Uses ip_set_req_bind, with type specific addage */ ++ ++/* Multiple shots operations: list, save, restore. ++ * ++ * - check kernel version and query the max number of sets ++ * - get the basic information on all sets ++ * and size required for the next step ++ * - get actual set data: header, data, bindings ++ */ ++ ++/* Get max_sets and the index of a queried set ++ */ ++#define IP_SET_OP_MAX_SETS 0x00000020 ++struct ip_set_req_max_sets { ++ unsigned op; ++ unsigned version; ++ ip_set_id_t max_sets; /* max_sets */ ++ ip_set_id_t sets; /* real number of sets */ ++ union ip_set_name_index set; /* index of set if name used */ ++}; ++ ++/* Get the id and name of the sets plus size for next step */ ++#define IP_SET_OP_LIST_SIZE 0x00000201 ++#define IP_SET_OP_SAVE_SIZE 0x00000202 ++struct ip_set_req_setnames { ++ unsigned op; ++ ip_set_id_t index; /* set to list/save */ ++ size_t size; /* size to get setdata/bindings */ ++ /* followed by sets number of struct ip_set_name_list */ ++}; ++ ++struct ip_set_name_list { ++ char name[IP_SET_MAXNAMELEN]; ++ char typename[IP_SET_MAXNAMELEN]; ++ ip_set_id_t index; ++ ip_set_id_t id; ++}; ++ ++/* The actual list operation */ ++#define IP_SET_OP_LIST 0x00000203 ++struct ip_set_req_list { ++ IP_SET_REQ_BYINDEX; ++ /* sets number of struct ip_set_list in reply */ ++}; ++ ++struct ip_set_list { ++ ip_set_id_t index; ++ ip_set_id_t binding; ++ u_int32_t ref; ++ size_t header_size; /* Set header data of header_size */ ++ size_t members_size; /* Set members data of members_size */ ++ size_t bindings_size; /* Set bindings data of bindings_size */ ++}; ++ ++struct ip_set_hash_list { ++ ip_set_ip_t ip; ++ ip_set_id_t binding; ++}; ++ ++/* The save operation */ ++#define IP_SET_OP_SAVE 0x00000204 ++/* Uses ip_set_req_list, in the reply replaced by ++ * sets number of struct ip_set_save plus a marker ++ * ip_set_save followed by ip_set_hash_save structures. ++ */ ++struct ip_set_save { ++ ip_set_id_t index; ++ ip_set_id_t binding; ++ size_t header_size; /* Set header data of header_size */ ++ size_t members_size; /* Set members data of members_size */ ++}; ++ ++/* At restoring, ip == 0 means default binding for the given set: */ ++struct ip_set_hash_save { ++ ip_set_ip_t ip; ++ ip_set_id_t id; ++ ip_set_id_t binding; ++}; ++ ++/* The restore operation */ ++#define IP_SET_OP_RESTORE 0x00000205 ++/* Uses ip_set_req_setnames followed by ip_set_restore structures ++ * plus a marker ip_set_restore, followed by ip_set_hash_save ++ * structures. ++ */ ++struct ip_set_restore { ++ char name[IP_SET_MAXNAMELEN]; ++ char typename[IP_SET_MAXNAMELEN]; ++ ip_set_id_t index; ++ size_t header_size; /* Create data of header_size */ ++ size_t members_size; /* Set members data of members_size */ ++}; ++ ++static inline int bitmap_bytes(ip_set_ip_t a, ip_set_ip_t b) ++{ ++ return 4 * ((((b - a + 8) / 8) + 3) / 4); ++} ++ ++#ifdef __KERNEL__ ++ ++#define ip_set_printk(format, args...) \ ++ do { \ ++ printk("%s: %s: ", __FILE__, __FUNCTION__); \ ++ printk(format "\n" , ## args); \ ++ } while (0) ++ ++#if defined(IP_SET_DEBUG) ++#define DP(format, args...) \ ++ do { \ ++ printk("%s: %s (DBG): ", __FILE__, __FUNCTION__);\ ++ printk(format "\n" , ## args); \ ++ } while (0) ++#define IP_SET_ASSERT(x) \ ++ do { \ ++ if (!(x)) \ ++ printk("IP_SET_ASSERT: %s:%i(%s)\n", \ ++ __FILE__, __LINE__, __FUNCTION__); \ ++ } while (0) ++#else ++#define DP(format, args...) ++#define IP_SET_ASSERT(x) ++#endif ++ ++struct ip_set; ++ ++/* ++ * The ip_set_type definition - one per set type, e.g. "ipmap". ++ * ++ * Each individual set has a pointer, set->type, going to one ++ * of these structures. Function pointers inside the structure implement ++ * the real behaviour of the sets. ++ * ++ * If not mentioned differently, the implementation behind the function ++ * pointers of a set_type, is expected to return 0 if ok, and a negative ++ * errno (e.g. -EINVAL) on error. ++ */ ++struct ip_set_type { ++ struct list_head list; /* next in list of set types */ ++ ++ /* test for IP in set (kernel: iptables -m set src|dst) ++ * return 0 if not in set, 1 if in set. ++ */ ++ int (*testip_kernel) (struct ip_set *set, ++ const struct sk_buff * skb, ++ ip_set_ip_t *ip, ++ const u_int32_t *flags, ++ unsigned char index); ++ ++ /* test for IP in set (userspace: ipset -T set IP) ++ * return 0 if not in set, 1 if in set. ++ */ ++ int (*testip) (struct ip_set *set, ++ const void *data, size_t size, ++ ip_set_ip_t *ip); ++ ++ /* ++ * Size of the data structure passed by when ++ * adding/deletin/testing an entry. ++ */ ++ size_t reqsize; ++ ++ /* Add IP into set (userspace: ipset -A set IP) ++ * Return -EEXIST if the address is already in the set, ++ * and -ERANGE if the address lies outside the set bounds. ++ * If the address was not already in the set, 0 is returned. ++ */ ++ int (*addip) (struct ip_set *set, ++ const void *data, size_t size, ++ ip_set_ip_t *ip); ++ ++ /* Add IP into set (kernel: iptables ... -j SET set src|dst) ++ * Return -EEXIST if the address is already in the set, ++ * and -ERANGE if the address lies outside the set bounds. ++ * If the address was not already in the set, 0 is returned. ++ */ ++ int (*addip_kernel) (struct ip_set *set, ++ const struct sk_buff * skb, ++ ip_set_ip_t *ip, ++ const u_int32_t *flags, ++ unsigned char index); ++ ++ /* remove IP from set (userspace: ipset -D set --entry x) ++ * Return -EEXIST if the address is NOT in the set, ++ * and -ERANGE if the address lies outside the set bounds. ++ * If the address really was in the set, 0 is returned. ++ */ ++ int (*delip) (struct ip_set *set, ++ const void *data, size_t size, ++ ip_set_ip_t *ip); ++ ++ /* remove IP from set (kernel: iptables ... -j SET --entry x) ++ * Return -EEXIST if the address is NOT in the set, ++ * and -ERANGE if the address lies outside the set bounds. ++ * If the address really was in the set, 0 is returned. ++ */ ++ int (*delip_kernel) (struct ip_set *set, ++ const struct sk_buff * skb, ++ ip_set_ip_t *ip, ++ const u_int32_t *flags, ++ unsigned char index); ++ ++ /* new set creation - allocated type specific items ++ */ ++ int (*create) (struct ip_set *set, ++ const void *data, size_t size); ++ ++ /* retry the operation after successfully tweaking the set ++ */ ++ int (*retry) (struct ip_set *set); ++ ++ /* set destruction - free type specific items ++ * There is no return value. ++ * Can be called only when child sets are destroyed. ++ */ ++ void (*destroy) (struct ip_set *set); ++ ++ /* set flushing - reset all bits in the set, or something similar. ++ * There is no return value. ++ */ ++ void (*flush) (struct ip_set *set); ++ ++ /* Listing: size needed for header ++ */ ++ size_t header_size; ++ ++ /* Listing: Get the header ++ * ++ * Fill in the information in "data". ++ * This function is always run after list_header_size() under a ++ * writelock on the set. Therefor is the length of "data" always ++ * correct. ++ */ ++ void (*list_header) (const struct ip_set *set, ++ void *data); ++ ++ /* Listing: Get the size for the set members ++ */ ++ int (*list_members_size) (const struct ip_set *set); ++ ++ /* Listing: Get the set members ++ * ++ * Fill in the information in "data". ++ * This function is always run after list_member_size() under a ++ * writelock on the set. Therefor is the length of "data" always ++ * correct. ++ */ ++ void (*list_members) (const struct ip_set *set, ++ void *data); ++ ++ char typename[IP_SET_MAXNAMELEN]; ++ unsigned char features; ++ int protocol_version; ++ ++ /* Set this to THIS_MODULE if you are a module, otherwise NULL */ ++ struct module *me; ++}; ++ ++extern int ip_set_register_set_type(struct ip_set_type *set_type); ++extern void ip_set_unregister_set_type(struct ip_set_type *set_type); ++ ++/* A generic ipset */ ++struct ip_set { ++ char name[IP_SET_MAXNAMELEN]; /* the name of the set */ ++ rwlock_t lock; /* lock for concurrency control */ ++ ip_set_id_t id; /* set id for swapping */ ++ ip_set_id_t binding; /* default binding for the set */ ++ atomic_t ref; /* in kernel and in hash references */ ++ struct ip_set_type *type; /* the set types */ ++ void *data; /* pooltype specific data */ ++}; ++ ++/* Structure to bind set elements to sets */ ++struct ip_set_hash { ++ struct list_head list; /* list of clashing entries in hash */ ++ ip_set_ip_t ip; /* ip from set */ ++ ip_set_id_t id; /* set id */ ++ ip_set_id_t binding; /* set we bind the element to */ ++}; ++ ++/* register and unregister set references */ ++extern ip_set_id_t ip_set_get_byname(const char name[IP_SET_MAXNAMELEN]); ++extern ip_set_id_t ip_set_get_byindex(ip_set_id_t id); ++extern void ip_set_put(ip_set_id_t id); ++ ++/* API for iptables set match, and SET target */ ++extern void ip_set_addip_kernel(ip_set_id_t id, ++ const struct sk_buff *skb, ++ const u_int32_t *flags); ++extern void ip_set_delip_kernel(ip_set_id_t id, ++ const struct sk_buff *skb, ++ const u_int32_t *flags); ++extern int ip_set_testip_kernel(ip_set_id_t id, ++ const struct sk_buff *skb, ++ const u_int32_t *flags); ++ ++#endif /* __KERNEL__ */ ++ ++#endif /*_IP_SET_H*/ +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ip_set_iphash.h +@@ -0,0 +1,30 @@ ++#ifndef __IP_SET_IPHASH_H ++#define __IP_SET_IPHASH_H ++ ++#include ++ ++#define SETTYPE_NAME "iphash" ++#define MAX_RANGE 0x0000FFFF ++ ++struct ip_set_iphash { ++ ip_set_ip_t *members; /* the iphash proper */ ++ uint32_t elements; /* number of elements */ ++ uint32_t hashsize; /* hash size */ ++ uint16_t probes; /* max number of probes */ ++ uint16_t resize; /* resize factor in percent */ ++ ip_set_ip_t netmask; /* netmask */ ++ void *initval[0]; /* initvals for jhash_1word */ ++}; ++ ++struct ip_set_req_iphash_create { ++ uint32_t hashsize; ++ uint16_t probes; ++ uint16_t resize; ++ ip_set_ip_t netmask; ++}; ++ ++struct ip_set_req_iphash { ++ ip_set_ip_t ip; ++}; ++ ++#endif /* __IP_SET_IPHASH_H */ +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ip_set_ipmap.h +@@ -0,0 +1,56 @@ ++#ifndef __IP_SET_IPMAP_H ++#define __IP_SET_IPMAP_H ++ ++#include ++ ++#define SETTYPE_NAME "ipmap" ++#define MAX_RANGE 0x0000FFFF ++ ++struct ip_set_ipmap { ++ void *members; /* the ipmap proper */ ++ ip_set_ip_t first_ip; /* host byte order, included in range */ ++ ip_set_ip_t last_ip; /* host byte order, included in range */ ++ ip_set_ip_t netmask; /* subnet netmask */ ++ ip_set_ip_t sizeid; /* size of set in IPs */ ++ ip_set_ip_t hosts; /* number of hosts in a subnet */ ++}; ++ ++struct ip_set_req_ipmap_create { ++ ip_set_ip_t from; ++ ip_set_ip_t to; ++ ip_set_ip_t netmask; ++}; ++ ++struct ip_set_req_ipmap { ++ ip_set_ip_t ip; ++}; ++ ++unsigned int ++mask_to_bits(ip_set_ip_t mask) ++{ ++ unsigned int bits = 32; ++ ip_set_ip_t maskaddr; ++ ++ if (mask == 0xFFFFFFFF) ++ return bits; ++ ++ maskaddr = 0xFFFFFFFE; ++ while (--bits >= 0 && maskaddr != mask) ++ maskaddr <<= 1; ++ ++ return bits; ++} ++ ++ip_set_ip_t ++range_to_mask(ip_set_ip_t from, ip_set_ip_t to, unsigned int *bits) ++{ ++ ip_set_ip_t mask = 0xFFFFFFFE; ++ ++ *bits = 32; ++ while (--(*bits) >= 0 && mask && (to & mask) != from) ++ mask <<= 1; ++ ++ return mask; ++} ++ ++#endif /* __IP_SET_IPMAP_H */ +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ip_set_ipporthash.h +@@ -0,0 +1,34 @@ ++#ifndef __IP_SET_IPPORTHASH_H ++#define __IP_SET_IPPORTHASH_H ++ ++#include ++ ++#define SETTYPE_NAME "ipporthash" ++#define MAX_RANGE 0x0000FFFF ++#define INVALID_PORT (MAX_RANGE + 1) ++ ++struct ip_set_ipporthash { ++ ip_set_ip_t *members; /* the ipporthash proper */ ++ uint32_t elements; /* number of elements */ ++ uint32_t hashsize; /* hash size */ ++ uint16_t probes; /* max number of probes */ ++ uint16_t resize; /* resize factor in percent */ ++ ip_set_ip_t first_ip; /* host byte order, included in range */ ++ ip_set_ip_t last_ip; /* host byte order, included in range */ ++ void *initval[0]; /* initvals for jhash_1word */ ++}; ++ ++struct ip_set_req_ipporthash_create { ++ uint32_t hashsize; ++ uint16_t probes; ++ uint16_t resize; ++ ip_set_ip_t from; ++ ip_set_ip_t to; ++}; ++ ++struct ip_set_req_ipporthash { ++ ip_set_ip_t ip; ++ ip_set_ip_t port; ++}; ++ ++#endif /* __IP_SET_IPPORTHASH_H */ +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ip_set_iptree.h +@@ -0,0 +1,40 @@ ++#ifndef __IP_SET_IPTREE_H ++#define __IP_SET_IPTREE_H ++ ++#include ++ ++#define SETTYPE_NAME "iptree" ++#define MAX_RANGE 0x0000FFFF ++ ++struct ip_set_iptreed { ++ unsigned long expires[256]; /* x.x.x.ADDR */ ++}; ++ ++struct ip_set_iptreec { ++ struct ip_set_iptreed *tree[256]; /* x.x.ADDR.* */ ++}; ++ ++struct ip_set_iptreeb { ++ struct ip_set_iptreec *tree[256]; /* x.ADDR.*.* */ ++}; ++ ++struct ip_set_iptree { ++ unsigned int timeout; ++ unsigned int gc_interval; ++#ifdef __KERNEL__ ++ uint32_t elements; /* number of elements */ ++ struct timer_list gc; ++ struct ip_set_iptreeb *tree[256]; /* ADDR.*.*.* */ ++#endif ++}; ++ ++struct ip_set_req_iptree_create { ++ unsigned int timeout; ++}; ++ ++struct ip_set_req_iptree { ++ ip_set_ip_t ip; ++ unsigned int timeout; ++}; ++ ++#endif /* __IP_SET_IPTREE_H */ +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ip_set_iptreemap.h +@@ -0,0 +1,40 @@ ++#ifndef __IP_SET_IPTREEMAP_H ++#define __IP_SET_IPTREEMAP_H ++ ++#include ++ ++#define SETTYPE_NAME "iptreemap" ++ ++#ifdef __KERNEL__ ++struct ip_set_iptreemap_d { ++ unsigned char bitmap[32]; /* x.x.x.y */ ++}; ++ ++struct ip_set_iptreemap_c { ++ struct ip_set_iptreemap_d *tree[256]; /* x.x.y.x */ ++}; ++ ++struct ip_set_iptreemap_b { ++ struct ip_set_iptreemap_c *tree[256]; /* x.y.x.x */ ++ unsigned char dirty[32]; ++}; ++#endif ++ ++struct ip_set_iptreemap { ++ unsigned int gc_interval; ++#ifdef __KERNEL__ ++ struct timer_list gc; ++ struct ip_set_iptreemap_b *tree[256]; /* y.x.x.x */ ++#endif ++}; ++ ++struct ip_set_req_iptreemap_create { ++ unsigned int gc_interval; ++}; ++ ++struct ip_set_req_iptreemap { ++ ip_set_ip_t start; ++ ip_set_ip_t end; ++}; ++ ++#endif /* __IP_SET_IPTREEMAP_H */ +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ip_set_jhash.h +@@ -0,0 +1,148 @@ ++#ifndef _LINUX_IPSET_JHASH_H ++#define _LINUX_IPSET_JHASH_H ++ ++/* This is a copy of linux/jhash.h but the types u32/u8 are changed ++ * to __u32/__u8 so that the header file can be included into ++ * userspace code as well. Jozsef Kadlecsik (kadlec@blackhole.kfki.hu) ++ */ ++ ++/* jhash.h: Jenkins hash support. ++ * ++ * Copyright (C) 1996 Bob Jenkins (bob_jenkins@burtleburtle.net) ++ * ++ * http://burtleburtle.net/bob/hash/ ++ * ++ * These are the credits from Bob's sources: ++ * ++ * lookup2.c, by Bob Jenkins, December 1996, Public Domain. ++ * hash(), hash2(), hash3, and mix() are externally useful functions. ++ * Routines to test the hash are included if SELF_TEST is defined. ++ * You can use this free for any purpose. It has no warranty. ++ * ++ * Copyright (C) 2003 David S. Miller (davem@redhat.com) ++ * ++ * I've modified Bob's hash to be useful in the Linux kernel, and ++ * any bugs present are surely my fault. -DaveM ++ */ ++ ++/* NOTE: Arguments are modified. */ ++#define __jhash_mix(a, b, c) \ ++{ \ ++ a -= b; a -= c; a ^= (c>>13); \ ++ b -= c; b -= a; b ^= (a<<8); \ ++ c -= a; c -= b; c ^= (b>>13); \ ++ a -= b; a -= c; a ^= (c>>12); \ ++ b -= c; b -= a; b ^= (a<<16); \ ++ c -= a; c -= b; c ^= (b>>5); \ ++ a -= b; a -= c; a ^= (c>>3); \ ++ b -= c; b -= a; b ^= (a<<10); \ ++ c -= a; c -= b; c ^= (b>>15); \ ++} ++ ++/* The golden ration: an arbitrary value */ ++#define JHASH_GOLDEN_RATIO 0x9e3779b9 ++ ++/* The most generic version, hashes an arbitrary sequence ++ * of bytes. No alignment or length assumptions are made about ++ * the input key. ++ */ ++static inline __u32 jhash(void *key, __u32 length, __u32 initval) ++{ ++ __u32 a, b, c, len; ++ __u8 *k = key; ++ ++ len = length; ++ a = b = JHASH_GOLDEN_RATIO; ++ c = initval; ++ ++ while (len >= 12) { ++ a += (k[0] +((__u32)k[1]<<8) +((__u32)k[2]<<16) +((__u32)k[3]<<24)); ++ b += (k[4] +((__u32)k[5]<<8) +((__u32)k[6]<<16) +((__u32)k[7]<<24)); ++ c += (k[8] +((__u32)k[9]<<8) +((__u32)k[10]<<16)+((__u32)k[11]<<24)); ++ ++ __jhash_mix(a,b,c); ++ ++ k += 12; ++ len -= 12; ++ } ++ ++ c += length; ++ switch (len) { ++ case 11: c += ((__u32)k[10]<<24); ++ case 10: c += ((__u32)k[9]<<16); ++ case 9 : c += ((__u32)k[8]<<8); ++ case 8 : b += ((__u32)k[7]<<24); ++ case 7 : b += ((__u32)k[6]<<16); ++ case 6 : b += ((__u32)k[5]<<8); ++ case 5 : b += k[4]; ++ case 4 : a += ((__u32)k[3]<<24); ++ case 3 : a += ((__u32)k[2]<<16); ++ case 2 : a += ((__u32)k[1]<<8); ++ case 1 : a += k[0]; ++ }; ++ ++ __jhash_mix(a,b,c); ++ ++ return c; ++} ++ ++/* A special optimized version that handles 1 or more of __u32s. ++ * The length parameter here is the number of __u32s in the key. ++ */ ++static inline __u32 jhash2(__u32 *k, __u32 length, __u32 initval) ++{ ++ __u32 a, b, c, len; ++ ++ a = b = JHASH_GOLDEN_RATIO; ++ c = initval; ++ len = length; ++ ++ while (len >= 3) { ++ a += k[0]; ++ b += k[1]; ++ c += k[2]; ++ __jhash_mix(a, b, c); ++ k += 3; len -= 3; ++ } ++ ++ c += length * 4; ++ ++ switch (len) { ++ case 2 : b += k[1]; ++ case 1 : a += k[0]; ++ }; ++ ++ __jhash_mix(a,b,c); ++ ++ return c; ++} ++ ++ ++/* A special ultra-optimized versions that knows they are hashing exactly ++ * 3, 2 or 1 word(s). ++ * ++ * NOTE: In partilar the "c += length; __jhash_mix(a,b,c);" normally ++ * done at the end is not done here. ++ */ ++static inline __u32 jhash_3words(__u32 a, __u32 b, __u32 c, __u32 initval) ++{ ++ a += JHASH_GOLDEN_RATIO; ++ b += JHASH_GOLDEN_RATIO; ++ c += initval; ++ ++ __jhash_mix(a, b, c); ++ ++ return c; ++} ++ ++static inline __u32 jhash_2words(__u32 a, __u32 b, __u32 initval) ++{ ++ return jhash_3words(a, b, 0, initval); ++} ++ ++static inline __u32 jhash_1word(__u32 a, __u32 initval) ++{ ++ return jhash_3words(a, 0, 0, initval); ++} ++ ++#endif /* _LINUX_IPSET_JHASH_H */ +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ip_set_macipmap.h +@@ -0,0 +1,38 @@ ++#ifndef __IP_SET_MACIPMAP_H ++#define __IP_SET_MACIPMAP_H ++ ++#include ++ ++#define SETTYPE_NAME "macipmap" ++#define MAX_RANGE 0x0000FFFF ++ ++/* general flags */ ++#define IPSET_MACIP_MATCHUNSET 1 ++ ++/* per ip flags */ ++#define IPSET_MACIP_ISSET 1 ++ ++struct ip_set_macipmap { ++ void *members; /* the macipmap proper */ ++ ip_set_ip_t first_ip; /* host byte order, included in range */ ++ ip_set_ip_t last_ip; /* host byte order, included in range */ ++ u_int32_t flags; ++}; ++ ++struct ip_set_req_macipmap_create { ++ ip_set_ip_t from; ++ ip_set_ip_t to; ++ u_int32_t flags; ++}; ++ ++struct ip_set_req_macipmap { ++ ip_set_ip_t ip; ++ unsigned char ethernet[ETH_ALEN]; ++}; ++ ++struct ip_set_macip { ++ unsigned short flags; ++ unsigned char ethernet[ETH_ALEN]; ++}; ++ ++#endif /* __IP_SET_MACIPMAP_H */ +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ip_set_malloc.h +@@ -0,0 +1,116 @@ ++#ifndef _IP_SET_MALLOC_H ++#define _IP_SET_MALLOC_H ++ ++#ifdef __KERNEL__ ++ ++/* Memory allocation and deallocation */ ++static size_t max_malloc_size = 0; ++ ++static inline void init_max_malloc_size(void) ++{ ++#define CACHE(x) max_malloc_size = x; ++#include ++#undef CACHE ++} ++ ++static inline void * ip_set_malloc(size_t bytes) ++{ ++ if (bytes > max_malloc_size) ++ return vmalloc(bytes); ++ else ++ return kmalloc(bytes, GFP_KERNEL); ++} ++ ++static inline void ip_set_free(void * data, size_t bytes) ++{ ++ if (bytes > max_malloc_size) ++ vfree(data); ++ else ++ kfree(data); ++} ++ ++struct harray { ++ size_t max_elements; ++ void *arrays[0]; ++}; ++ ++static inline void * ++harray_malloc(size_t hashsize, size_t typesize, int flags) ++{ ++ struct harray *harray; ++ size_t max_elements, size, i, j; ++ ++ if (!max_malloc_size) ++ init_max_malloc_size(); ++ ++ if (typesize > max_malloc_size) ++ return NULL; ++ ++ max_elements = max_malloc_size/typesize; ++ size = hashsize/max_elements; ++ if (hashsize % max_elements) ++ size++; ++ ++ /* Last pointer signals end of arrays */ ++ harray = kmalloc(sizeof(struct harray) + (size + 1) * sizeof(void *), ++ flags); ++ ++ if (!harray) ++ return NULL; ++ ++ for (i = 0; i < size - 1; i++) { ++ harray->arrays[i] = kmalloc(max_elements * typesize, flags); ++ if (!harray->arrays[i]) ++ goto undo; ++ memset(harray->arrays[i], 0, max_elements * typesize); ++ } ++ harray->arrays[i] = kmalloc((hashsize - i * max_elements) * typesize, ++ flags); ++ if (!harray->arrays[i]) ++ goto undo; ++ memset(harray->arrays[i], 0, (hashsize - i * max_elements) * typesize); ++ ++ harray->max_elements = max_elements; ++ harray->arrays[size] = NULL; ++ ++ return (void *)harray; ++ ++ undo: ++ for (j = 0; j < i; j++) { ++ kfree(harray->arrays[j]); ++ } ++ kfree(harray); ++ return NULL; ++} ++ ++static inline void harray_free(void *h) ++{ ++ struct harray *harray = (struct harray *) h; ++ size_t i; ++ ++ for (i = 0; harray->arrays[i] != NULL; i++) ++ kfree(harray->arrays[i]); ++ kfree(harray); ++} ++ ++static inline void harray_flush(void *h, size_t hashsize, size_t typesize) ++{ ++ struct harray *harray = (struct harray *) h; ++ size_t i; ++ ++ for (i = 0; harray->arrays[i+1] != NULL; i++) ++ memset(harray->arrays[i], 0, harray->max_elements * typesize); ++ memset(harray->arrays[i], 0, ++ (hashsize - i * harray->max_elements) * typesize); ++} ++ ++#define HARRAY_ELEM(h, type, which) \ ++({ \ ++ struct harray *__h = (struct harray *)(h); \ ++ ((type)((__h)->arrays[(which)/(__h)->max_elements]) \ ++ + (which)%(__h)->max_elements); \ ++}) ++ ++#endif /* __KERNEL__ */ ++ ++#endif /*_IP_SET_MALLOC_H*/ +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ip_set_nethash.h +@@ -0,0 +1,55 @@ ++#ifndef __IP_SET_NETHASH_H ++#define __IP_SET_NETHASH_H ++ ++#include ++ ++#define SETTYPE_NAME "nethash" ++#define MAX_RANGE 0x0000FFFF ++ ++struct ip_set_nethash { ++ ip_set_ip_t *members; /* the nethash proper */ ++ uint32_t elements; /* number of elements */ ++ uint32_t hashsize; /* hash size */ ++ uint16_t probes; /* max number of probes */ ++ uint16_t resize; /* resize factor in percent */ ++ unsigned char cidr[30]; /* CIDR sizes */ ++ void *initval[0]; /* initvals for jhash_1word */ ++}; ++ ++struct ip_set_req_nethash_create { ++ uint32_t hashsize; ++ uint16_t probes; ++ uint16_t resize; ++}; ++ ++struct ip_set_req_nethash { ++ ip_set_ip_t ip; ++ unsigned char cidr; ++}; ++ ++static unsigned char shifts[] = {255, 253, 249, 241, 225, 193, 129, 1}; ++ ++static inline ip_set_ip_t ++pack(ip_set_ip_t ip, unsigned char cidr) ++{ ++ ip_set_ip_t addr, *paddr = &addr; ++ unsigned char n, t, *a; ++ ++ addr = htonl(ip & (0xFFFFFFFF << (32 - (cidr)))); ++#ifdef __KERNEL__ ++ DP("ip:%u.%u.%u.%u/%u", NIPQUAD(addr), cidr); ++#endif ++ n = cidr / 8; ++ t = cidr % 8; ++ a = &((unsigned char *)paddr)[n]; ++ *a = *a /(1 << (8 - t)) + shifts[t]; ++#ifdef __KERNEL__ ++ DP("n: %u, t: %u, a: %u", n, t, *a); ++ DP("ip:%u.%u.%u.%u/%u, %u.%u.%u.%u", ++ HIPQUAD(ip), cidr, NIPQUAD(addr)); ++#endif ++ ++ return ntohl(addr); ++} ++ ++#endif /* __IP_SET_NETHASH_H */ +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ip_set_portmap.h +@@ -0,0 +1,25 @@ ++#ifndef __IP_SET_PORTMAP_H ++#define __IP_SET_PORTMAP_H ++ ++#include ++ ++#define SETTYPE_NAME "portmap" ++#define MAX_RANGE 0x0000FFFF ++#define INVALID_PORT (MAX_RANGE + 1) ++ ++struct ip_set_portmap { ++ void *members; /* the portmap proper */ ++ ip_set_ip_t first_port; /* host byte order, included in range */ ++ ip_set_ip_t last_port; /* host byte order, included in range */ ++}; ++ ++struct ip_set_req_portmap_create { ++ ip_set_ip_t from; ++ ip_set_ip_t to; ++}; ++ ++struct ip_set_req_portmap { ++ ip_set_ip_t port; ++}; ++ ++#endif /* __IP_SET_PORTMAP_H */ +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ipt_set.h +@@ -0,0 +1,21 @@ ++#ifndef _IPT_SET_H ++#define _IPT_SET_H ++ ++#include ++ ++struct ipt_set_info { ++ ip_set_id_t index; ++ u_int32_t flags[IP_SET_MAX_BINDINGS + 1]; ++}; ++ ++/* match info */ ++struct ipt_set_info_match { ++ struct ipt_set_info match_set; ++}; ++ ++struct ipt_set_info_target { ++ struct ipt_set_info add_set; ++ struct ipt_set_info del_set; ++}; ++ ++#endif /*_IPT_SET_H*/ +--- /dev/null ++++ b/net/ipv4/netfilter/ip_set.c +@@ -0,0 +1,2003 @@ ++/* Copyright (C) 2000-2002 Joakim Axelsson ++ * Patrick Schaaf ++ * Copyright (C) 2003-2004 Jozsef Kadlecsik ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++/* Kernel module for IP set management */ ++ ++#include ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) ++#include ++#endif ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#define ASSERT_READ_LOCK(x) ++#define ASSERT_WRITE_LOCK(x) ++#include ++ ++static struct list_head set_type_list; /* all registered sets */ ++static struct ip_set **ip_set_list; /* all individual sets */ ++static DEFINE_RWLOCK(ip_set_lock); /* protects the lists and the hash */ ++static DECLARE_MUTEX(ip_set_app_mutex); /* serializes user access */ ++static ip_set_id_t ip_set_max = CONFIG_IP_NF_SET_MAX; ++static ip_set_id_t ip_set_bindings_hash_size = CONFIG_IP_NF_SET_HASHSIZE; ++static struct list_head *ip_set_hash; /* hash of bindings */ ++static unsigned int ip_set_hash_random; /* random seed */ ++ ++/* ++ * Sets are identified either by the index in ip_set_list or by id. ++ * The id never changes and is used to find a key in the hash. ++ * The index may change by swapping and used at all other places ++ * (set/SET netfilter modules, binding value, etc.) ++ * ++ * Userspace requests are serialized by ip_set_mutex and sets can ++ * be deleted only from userspace. Therefore ip_set_list locking ++ * must obey the following rules: ++ * ++ * - kernel requests: read and write locking mandatory ++ * - user requests: read locking optional, write locking mandatory ++ */ ++ ++static inline void ++__ip_set_get(ip_set_id_t index) ++{ ++ atomic_inc(&ip_set_list[index]->ref); ++} ++ ++static inline void ++__ip_set_put(ip_set_id_t index) ++{ ++ atomic_dec(&ip_set_list[index]->ref); ++} ++ ++/* ++ * Binding routines ++ */ ++ ++static inline struct ip_set_hash * ++__ip_set_find(u_int32_t key, ip_set_id_t id, ip_set_ip_t ip) ++{ ++ struct ip_set_hash *set_hash; ++ ++ list_for_each_entry(set_hash, &ip_set_hash[key], list) ++ if (set_hash->id == id && set_hash->ip == ip) ++ return set_hash; ++ ++ return NULL; ++} ++ ++static ip_set_id_t ++ip_set_find_in_hash(ip_set_id_t id, ip_set_ip_t ip) ++{ ++ u_int32_t key = jhash_2words(id, ip, ip_set_hash_random) ++ % ip_set_bindings_hash_size; ++ struct ip_set_hash *set_hash; ++ ++ ASSERT_READ_LOCK(&ip_set_lock); ++ IP_SET_ASSERT(ip_set_list[id]); ++ DP("set: %s, ip: %u.%u.%u.%u", ip_set_list[id]->name, HIPQUAD(ip)); ++ ++ set_hash = __ip_set_find(key, id, ip); ++ ++ DP("set: %s, ip: %u.%u.%u.%u, binding: %s", ip_set_list[id]->name, ++ HIPQUAD(ip), ++ set_hash != NULL ? ip_set_list[set_hash->binding]->name : ""); ++ ++ return (set_hash != NULL ? set_hash->binding : IP_SET_INVALID_ID); ++} ++ ++static inline void ++__set_hash_del(struct ip_set_hash *set_hash) ++{ ++ ASSERT_WRITE_LOCK(&ip_set_lock); ++ IP_SET_ASSERT(ip_set_list[set_hash->binding]); ++ ++ __ip_set_put(set_hash->binding); ++ list_del(&set_hash->list); ++ kfree(set_hash); ++} ++ ++static int ++ip_set_hash_del(ip_set_id_t id, ip_set_ip_t ip) ++{ ++ u_int32_t key = jhash_2words(id, ip, ip_set_hash_random) ++ % ip_set_bindings_hash_size; ++ struct ip_set_hash *set_hash; ++ ++ IP_SET_ASSERT(ip_set_list[id]); ++ DP("set: %s, ip: %u.%u.%u.%u", ip_set_list[id]->name, HIPQUAD(ip)); ++ write_lock_bh(&ip_set_lock); ++ set_hash = __ip_set_find(key, id, ip); ++ DP("set: %s, ip: %u.%u.%u.%u, binding: %s", ip_set_list[id]->name, ++ HIPQUAD(ip), ++ set_hash != NULL ? ip_set_list[set_hash->binding]->name : ""); ++ ++ if (set_hash != NULL) ++ __set_hash_del(set_hash); ++ write_unlock_bh(&ip_set_lock); ++ return 0; ++} ++ ++static int ++ip_set_hash_add(ip_set_id_t id, ip_set_ip_t ip, ip_set_id_t binding) ++{ ++ u_int32_t key = jhash_2words(id, ip, ip_set_hash_random) ++ % ip_set_bindings_hash_size; ++ struct ip_set_hash *set_hash; ++ int ret = 0; ++ ++ IP_SET_ASSERT(ip_set_list[id]); ++ IP_SET_ASSERT(ip_set_list[binding]); ++ DP("set: %s, ip: %u.%u.%u.%u, binding: %s", ip_set_list[id]->name, ++ HIPQUAD(ip), ip_set_list[binding]->name); ++ write_lock_bh(&ip_set_lock); ++ set_hash = __ip_set_find(key, id, ip); ++ if (!set_hash) { ++ set_hash = kmalloc(sizeof(struct ip_set_hash), GFP_ATOMIC); ++ if (!set_hash) { ++ ret = -ENOMEM; ++ goto unlock; ++ } ++ INIT_LIST_HEAD(&set_hash->list); ++ set_hash->id = id; ++ set_hash->ip = ip; ++ list_add(&set_hash->list, &ip_set_hash[key]); ++ } else { ++ IP_SET_ASSERT(ip_set_list[set_hash->binding]); ++ DP("overwrite binding: %s", ++ ip_set_list[set_hash->binding]->name); ++ __ip_set_put(set_hash->binding); ++ } ++ set_hash->binding = binding; ++ __ip_set_get(set_hash->binding); ++ DP("stored: key %u, id %u (%s), ip %u.%u.%u.%u, binding %u (%s)", ++ key, id, ip_set_list[id]->name, ++ HIPQUAD(ip), binding, ip_set_list[binding]->name); ++ unlock: ++ write_unlock_bh(&ip_set_lock); ++ return ret; ++} ++ ++#define FOREACH_HASH_DO(fn, args...) \ ++({ \ ++ ip_set_id_t __key; \ ++ struct ip_set_hash *__set_hash; \ ++ \ ++ for (__key = 0; __key < ip_set_bindings_hash_size; __key++) { \ ++ list_for_each_entry(__set_hash, &ip_set_hash[__key], list) \ ++ fn(__set_hash , ## args); \ ++ } \ ++}) ++ ++#define FOREACH_HASH_RW_DO(fn, args...) \ ++({ \ ++ ip_set_id_t __key; \ ++ struct ip_set_hash *__set_hash, *__n; \ ++ \ ++ ASSERT_WRITE_LOCK(&ip_set_lock); \ ++ for (__key = 0; __key < ip_set_bindings_hash_size; __key++) { \ ++ list_for_each_entry_safe(__set_hash, __n, &ip_set_hash[__key], list)\ ++ fn(__set_hash , ## args); \ ++ } \ ++}) ++ ++/* Add, del and test set entries from kernel */ ++ ++#define follow_bindings(index, set, ip) \ ++((index = ip_set_find_in_hash((set)->id, ip)) != IP_SET_INVALID_ID \ ++ || (index = (set)->binding) != IP_SET_INVALID_ID) ++ ++int ++ip_set_testip_kernel(ip_set_id_t index, ++ const struct sk_buff *skb, ++ const u_int32_t *flags) ++{ ++ struct ip_set *set; ++ ip_set_ip_t ip; ++ int res; ++ unsigned char i = 0; ++ ++ IP_SET_ASSERT(flags[i]); ++ read_lock_bh(&ip_set_lock); ++ do { ++ set = ip_set_list[index]; ++ IP_SET_ASSERT(set); ++ DP("set %s, index %u", set->name, index); ++ read_lock_bh(&set->lock); ++ res = set->type->testip_kernel(set, skb, &ip, flags, i++); ++ read_unlock_bh(&set->lock); ++ i += !!(set->type->features & IPSET_DATA_DOUBLE); ++ } while (res > 0 ++ && flags[i] ++ && follow_bindings(index, set, ip)); ++ read_unlock_bh(&ip_set_lock); ++ ++ return res; ++} ++ ++void ++ip_set_addip_kernel(ip_set_id_t index, ++ const struct sk_buff *skb, ++ const u_int32_t *flags) ++{ ++ struct ip_set *set; ++ ip_set_ip_t ip; ++ int res; ++ unsigned char i = 0; ++ ++ IP_SET_ASSERT(flags[i]); ++ retry: ++ read_lock_bh(&ip_set_lock); ++ do { ++ set = ip_set_list[index]; ++ IP_SET_ASSERT(set); ++ DP("set %s, index %u", set->name, index); ++ write_lock_bh(&set->lock); ++ res = set->type->addip_kernel(set, skb, &ip, flags, i++); ++ write_unlock_bh(&set->lock); ++ i += !!(set->type->features & IPSET_DATA_DOUBLE); ++ } while ((res == 0 || res == -EEXIST) ++ && flags[i] ++ && follow_bindings(index, set, ip)); ++ read_unlock_bh(&ip_set_lock); ++ ++ if (res == -EAGAIN ++ && set->type->retry ++ && (res = set->type->retry(set)) == 0) ++ goto retry; ++} ++ ++void ++ip_set_delip_kernel(ip_set_id_t index, ++ const struct sk_buff *skb, ++ const u_int32_t *flags) ++{ ++ struct ip_set *set; ++ ip_set_ip_t ip; ++ int res; ++ unsigned char i = 0; ++ ++ IP_SET_ASSERT(flags[i]); ++ read_lock_bh(&ip_set_lock); ++ do { ++ set = ip_set_list[index]; ++ IP_SET_ASSERT(set); ++ DP("set %s, index %u", set->name, index); ++ write_lock_bh(&set->lock); ++ res = set->type->delip_kernel(set, skb, &ip, flags, i++); ++ write_unlock_bh(&set->lock); ++ i += !!(set->type->features & IPSET_DATA_DOUBLE); ++ } while ((res == 0 || res == -EEXIST) ++ && flags[i] ++ && follow_bindings(index, set, ip)); ++ read_unlock_bh(&ip_set_lock); ++} ++ ++/* Register and deregister settype */ ++ ++static inline struct ip_set_type * ++find_set_type(const char *name) ++{ ++ struct ip_set_type *set_type; ++ ++ list_for_each_entry(set_type, &set_type_list, list) ++ if (!strncmp(set_type->typename, name, IP_SET_MAXNAMELEN - 1)) ++ return set_type; ++ return NULL; ++} ++ ++int ++ip_set_register_set_type(struct ip_set_type *set_type) ++{ ++ int ret = 0; ++ ++ if (set_type->protocol_version != IP_SET_PROTOCOL_VERSION) { ++ ip_set_printk("'%s' uses wrong protocol version %u (want %u)", ++ set_type->typename, ++ set_type->protocol_version, ++ IP_SET_PROTOCOL_VERSION); ++ return -EINVAL; ++ } ++ ++ write_lock_bh(&ip_set_lock); ++ if (find_set_type(set_type->typename)) { ++ /* Duplicate! */ ++ ip_set_printk("'%s' already registered!", ++ set_type->typename); ++ ret = -EINVAL; ++ goto unlock; ++ } ++ if (!try_module_get(THIS_MODULE)) { ++ ret = -EFAULT; ++ goto unlock; ++ } ++ list_add(&set_type->list, &set_type_list); ++ DP("'%s' registered.", set_type->typename); ++ unlock: ++ write_unlock_bh(&ip_set_lock); ++ return ret; ++} ++ ++void ++ip_set_unregister_set_type(struct ip_set_type *set_type) ++{ ++ write_lock_bh(&ip_set_lock); ++ if (!find_set_type(set_type->typename)) { ++ ip_set_printk("'%s' not registered?", ++ set_type->typename); ++ goto unlock; ++ } ++ list_del(&set_type->list); ++ module_put(THIS_MODULE); ++ DP("'%s' unregistered.", set_type->typename); ++ unlock: ++ write_unlock_bh(&ip_set_lock); ++ ++} ++ ++/* ++ * Userspace routines ++ */ ++ ++/* ++ * Find set by name, reference it once. The reference makes sure the ++ * thing pointed to, does not go away under our feet. Drop the reference ++ * later, using ip_set_put(). ++ */ ++ip_set_id_t ++ip_set_get_byname(const char *name) ++{ ++ ip_set_id_t i, index = IP_SET_INVALID_ID; ++ ++ down(&ip_set_app_mutex); ++ for (i = 0; i < ip_set_max; i++) { ++ if (ip_set_list[i] != NULL ++ && strcmp(ip_set_list[i]->name, name) == 0) { ++ __ip_set_get(i); ++ index = i; ++ break; ++ } ++ } ++ up(&ip_set_app_mutex); ++ return index; ++} ++ ++/* ++ * Find set by index, reference it once. The reference makes sure the ++ * thing pointed to, does not go away under our feet. Drop the reference ++ * later, using ip_set_put(). ++ */ ++ip_set_id_t ++ip_set_get_byindex(ip_set_id_t index) ++{ ++ down(&ip_set_app_mutex); ++ ++ if (index >= ip_set_max) ++ return IP_SET_INVALID_ID; ++ ++ if (ip_set_list[index]) ++ __ip_set_get(index); ++ else ++ index = IP_SET_INVALID_ID; ++ ++ up(&ip_set_app_mutex); ++ return index; ++} ++ ++/* ++ * If the given set pointer points to a valid set, decrement ++ * reference count by 1. The caller shall not assume the index ++ * to be valid, after calling this function. ++ */ ++void ip_set_put(ip_set_id_t index) ++{ ++ down(&ip_set_app_mutex); ++ if (ip_set_list[index]) ++ __ip_set_put(index); ++ up(&ip_set_app_mutex); ++} ++ ++/* Find a set by name or index */ ++static ip_set_id_t ++ip_set_find_byname(const char *name) ++{ ++ ip_set_id_t i, index = IP_SET_INVALID_ID; ++ ++ for (i = 0; i < ip_set_max; i++) { ++ if (ip_set_list[i] != NULL ++ && strcmp(ip_set_list[i]->name, name) == 0) { ++ index = i; ++ break; ++ } ++ } ++ return index; ++} ++ ++static ip_set_id_t ++ip_set_find_byindex(ip_set_id_t index) ++{ ++ if (index >= ip_set_max || ip_set_list[index] == NULL) ++ index = IP_SET_INVALID_ID; ++ ++ return index; ++} ++ ++/* ++ * Add, del, test, bind and unbind ++ */ ++ ++static inline int ++__ip_set_testip(struct ip_set *set, ++ const void *data, ++ size_t size, ++ ip_set_ip_t *ip) ++{ ++ int res; ++ ++ read_lock_bh(&set->lock); ++ res = set->type->testip(set, data, size, ip); ++ read_unlock_bh(&set->lock); ++ ++ return res; ++} ++ ++static int ++__ip_set_addip(ip_set_id_t index, ++ const void *data, ++ size_t size) ++{ ++ struct ip_set *set = ip_set_list[index]; ++ ip_set_ip_t ip; ++ int res; ++ ++ IP_SET_ASSERT(set); ++ do { ++ write_lock_bh(&set->lock); ++ res = set->type->addip(set, data, size, &ip); ++ write_unlock_bh(&set->lock); ++ } while (res == -EAGAIN ++ && set->type->retry ++ && (res = set->type->retry(set)) == 0); ++ ++ return res; ++} ++ ++static int ++ip_set_addip(ip_set_id_t index, ++ const void *data, ++ size_t size) ++{ ++ ++ return __ip_set_addip(index, ++ data + sizeof(struct ip_set_req_adt), ++ size - sizeof(struct ip_set_req_adt)); ++} ++ ++static int ++ip_set_delip(ip_set_id_t index, ++ const void *data, ++ size_t size) ++{ ++ struct ip_set *set = ip_set_list[index]; ++ ip_set_ip_t ip; ++ int res; ++ ++ IP_SET_ASSERT(set); ++ write_lock_bh(&set->lock); ++ res = set->type->delip(set, ++ data + sizeof(struct ip_set_req_adt), ++ size - sizeof(struct ip_set_req_adt), ++ &ip); ++ write_unlock_bh(&set->lock); ++ ++ return res; ++} ++ ++static int ++ip_set_testip(ip_set_id_t index, ++ const void *data, ++ size_t size) ++{ ++ struct ip_set *set = ip_set_list[index]; ++ ip_set_ip_t ip; ++ int res; ++ ++ IP_SET_ASSERT(set); ++ res = __ip_set_testip(set, ++ data + sizeof(struct ip_set_req_adt), ++ size - sizeof(struct ip_set_req_adt), ++ &ip); ++ ++ return (res > 0 ? -EEXIST : res); ++} ++ ++static int ++ip_set_bindip(ip_set_id_t index, ++ const void *data, ++ size_t size) ++{ ++ struct ip_set *set = ip_set_list[index]; ++ struct ip_set_req_bind *req_bind; ++ ip_set_id_t binding; ++ ip_set_ip_t ip; ++ int res; ++ ++ IP_SET_ASSERT(set); ++ if (size < sizeof(struct ip_set_req_bind)) ++ return -EINVAL; ++ ++ req_bind = (struct ip_set_req_bind *) data; ++ req_bind->binding[IP_SET_MAXNAMELEN - 1] = '\0'; ++ ++ if (strcmp(req_bind->binding, IPSET_TOKEN_DEFAULT) == 0) { ++ /* Default binding of a set */ ++ char *binding_name; ++ ++ if (size != sizeof(struct ip_set_req_bind) + IP_SET_MAXNAMELEN) ++ return -EINVAL; ++ ++ binding_name = (char *)(data + sizeof(struct ip_set_req_bind)); ++ binding_name[IP_SET_MAXNAMELEN - 1] = '\0'; ++ ++ binding = ip_set_find_byname(binding_name); ++ if (binding == IP_SET_INVALID_ID) ++ return -ENOENT; ++ ++ write_lock_bh(&ip_set_lock); ++ /* Sets as binding values are referenced */ ++ if (set->binding != IP_SET_INVALID_ID) ++ __ip_set_put(set->binding); ++ set->binding = binding; ++ __ip_set_get(set->binding); ++ write_unlock_bh(&ip_set_lock); ++ ++ return 0; ++ } ++ binding = ip_set_find_byname(req_bind->binding); ++ if (binding == IP_SET_INVALID_ID) ++ return -ENOENT; ++ ++ res = __ip_set_testip(set, ++ data + sizeof(struct ip_set_req_bind), ++ size - sizeof(struct ip_set_req_bind), ++ &ip); ++ DP("set %s, ip: %u.%u.%u.%u, binding %s", ++ set->name, HIPQUAD(ip), ip_set_list[binding]->name); ++ ++ if (res >= 0) ++ res = ip_set_hash_add(set->id, ip, binding); ++ ++ return res; ++} ++ ++#define FOREACH_SET_DO(fn, args...) \ ++({ \ ++ ip_set_id_t __i; \ ++ struct ip_set *__set; \ ++ \ ++ for (__i = 0; __i < ip_set_max; __i++) { \ ++ __set = ip_set_list[__i]; \ ++ if (__set != NULL) \ ++ fn(__set , ##args); \ ++ } \ ++}) ++ ++static inline void ++__set_hash_del_byid(struct ip_set_hash *set_hash, ip_set_id_t id) ++{ ++ if (set_hash->id == id) ++ __set_hash_del(set_hash); ++} ++ ++static inline void ++__unbind_default(struct ip_set *set) ++{ ++ if (set->binding != IP_SET_INVALID_ID) { ++ /* Sets as binding values are referenced */ ++ __ip_set_put(set->binding); ++ set->binding = IP_SET_INVALID_ID; ++ } ++} ++ ++static int ++ip_set_unbindip(ip_set_id_t index, ++ const void *data, ++ size_t size) ++{ ++ struct ip_set *set; ++ struct ip_set_req_bind *req_bind; ++ ip_set_ip_t ip; ++ int res; ++ ++ DP(""); ++ if (size < sizeof(struct ip_set_req_bind)) ++ return -EINVAL; ++ ++ req_bind = (struct ip_set_req_bind *) data; ++ req_bind->binding[IP_SET_MAXNAMELEN - 1] = '\0'; ++ ++ DP("%u %s", index, req_bind->binding); ++ if (index == IP_SET_INVALID_ID) { ++ /* unbind :all: */ ++ if (strcmp(req_bind->binding, IPSET_TOKEN_DEFAULT) == 0) { ++ /* Default binding of sets */ ++ write_lock_bh(&ip_set_lock); ++ FOREACH_SET_DO(__unbind_default); ++ write_unlock_bh(&ip_set_lock); ++ return 0; ++ } else if (strcmp(req_bind->binding, IPSET_TOKEN_ALL) == 0) { ++ /* Flush all bindings of all sets*/ ++ write_lock_bh(&ip_set_lock); ++ FOREACH_HASH_RW_DO(__set_hash_del); ++ write_unlock_bh(&ip_set_lock); ++ return 0; ++ } ++ DP("unreachable reached!"); ++ return -EINVAL; ++ } ++ ++ set = ip_set_list[index]; ++ IP_SET_ASSERT(set); ++ if (strcmp(req_bind->binding, IPSET_TOKEN_DEFAULT) == 0) { ++ /* Default binding of set */ ++ ip_set_id_t binding = ip_set_find_byindex(set->binding); ++ ++ if (binding == IP_SET_INVALID_ID) ++ return -ENOENT; ++ ++ write_lock_bh(&ip_set_lock); ++ /* Sets in hash values are referenced */ ++ __ip_set_put(set->binding); ++ set->binding = IP_SET_INVALID_ID; ++ write_unlock_bh(&ip_set_lock); ++ ++ return 0; ++ } else if (strcmp(req_bind->binding, IPSET_TOKEN_ALL) == 0) { ++ /* Flush all bindings */ ++ ++ write_lock_bh(&ip_set_lock); ++ FOREACH_HASH_RW_DO(__set_hash_del_byid, set->id); ++ write_unlock_bh(&ip_set_lock); ++ return 0; ++ } ++ ++ res = __ip_set_testip(set, ++ data + sizeof(struct ip_set_req_bind), ++ size - sizeof(struct ip_set_req_bind), ++ &ip); ++ ++ DP("set %s, ip: %u.%u.%u.%u", set->name, HIPQUAD(ip)); ++ if (res >= 0) ++ res = ip_set_hash_del(set->id, ip); ++ ++ return res; ++} ++ ++static int ++ip_set_testbind(ip_set_id_t index, ++ const void *data, ++ size_t size) ++{ ++ struct ip_set *set = ip_set_list[index]; ++ struct ip_set_req_bind *req_bind; ++ ip_set_id_t binding; ++ ip_set_ip_t ip; ++ int res; ++ ++ IP_SET_ASSERT(set); ++ if (size < sizeof(struct ip_set_req_bind)) ++ return -EINVAL; ++ ++ req_bind = (struct ip_set_req_bind *) data; ++ req_bind->binding[IP_SET_MAXNAMELEN - 1] = '\0'; ++ ++ if (strcmp(req_bind->binding, IPSET_TOKEN_DEFAULT) == 0) { ++ /* Default binding of set */ ++ char *binding_name; ++ ++ if (size != sizeof(struct ip_set_req_bind) + IP_SET_MAXNAMELEN) ++ return -EINVAL; ++ ++ binding_name = (char *)(data + sizeof(struct ip_set_req_bind)); ++ binding_name[IP_SET_MAXNAMELEN - 1] = '\0'; ++ ++ binding = ip_set_find_byname(binding_name); ++ if (binding == IP_SET_INVALID_ID) ++ return -ENOENT; ++ ++ res = (set->binding == binding) ? -EEXIST : 0; ++ ++ return res; ++ } ++ binding = ip_set_find_byname(req_bind->binding); ++ if (binding == IP_SET_INVALID_ID) ++ return -ENOENT; ++ ++ ++ res = __ip_set_testip(set, ++ data + sizeof(struct ip_set_req_bind), ++ size - sizeof(struct ip_set_req_bind), ++ &ip); ++ DP("set %s, ip: %u.%u.%u.%u, binding %s", ++ set->name, HIPQUAD(ip), ip_set_list[binding]->name); ++ ++ if (res >= 0) ++ res = (ip_set_find_in_hash(set->id, ip) == binding) ++ ? -EEXIST : 0; ++ ++ return res; ++} ++ ++static struct ip_set_type * ++find_set_type_rlock(const char *typename) ++{ ++ struct ip_set_type *type; ++ ++ read_lock_bh(&ip_set_lock); ++ type = find_set_type(typename); ++ if (type == NULL) ++ read_unlock_bh(&ip_set_lock); ++ ++ return type; ++} ++ ++static int ++find_free_id(const char *name, ++ ip_set_id_t *index, ++ ip_set_id_t *id) ++{ ++ ip_set_id_t i; ++ ++ *id = IP_SET_INVALID_ID; ++ for (i = 0; i < ip_set_max; i++) { ++ if (ip_set_list[i] == NULL) { ++ if (*id == IP_SET_INVALID_ID) ++ *id = *index = i; ++ } else if (strcmp(name, ip_set_list[i]->name) == 0) ++ /* Name clash */ ++ return -EEXIST; ++ } ++ if (*id == IP_SET_INVALID_ID) ++ /* No free slot remained */ ++ return -ERANGE; ++ /* Check that index is usable as id (swapping) */ ++ check: ++ for (i = 0; i < ip_set_max; i++) { ++ if (ip_set_list[i] != NULL ++ && ip_set_list[i]->id == *id) { ++ *id = i; ++ goto check; ++ } ++ } ++ return 0; ++} ++ ++/* ++ * Create a set ++ */ ++static int ++ip_set_create(const char *name, ++ const char *typename, ++ ip_set_id_t restore, ++ const void *data, ++ size_t size) ++{ ++ struct ip_set *set; ++ ip_set_id_t index = 0, id; ++ int res = 0; ++ ++ DP("setname: %s, typename: %s, id: %u", name, typename, restore); ++ /* ++ * First, and without any locks, allocate and initialize ++ * a normal base set structure. ++ */ ++ set = kmalloc(sizeof(struct ip_set), GFP_KERNEL); ++ if (!set) ++ return -ENOMEM; ++ set->lock = RW_LOCK_UNLOCKED; ++ strncpy(set->name, name, IP_SET_MAXNAMELEN); ++ set->binding = IP_SET_INVALID_ID; ++ atomic_set(&set->ref, 0); ++ ++ /* ++ * Next, take the &ip_set_lock, check that we know the type, ++ * and take a reference on the type, to make sure it ++ * stays available while constructing our new set. ++ * ++ * After referencing the type, we drop the &ip_set_lock, ++ * and let the new set construction run without locks. ++ */ ++ set->type = find_set_type_rlock(typename); ++ if (set->type == NULL) { ++ /* Try loading the module */ ++ char modulename[IP_SET_MAXNAMELEN + strlen("ip_set_") + 1]; ++ strcpy(modulename, "ip_set_"); ++ strcat(modulename, typename); ++ DP("try to load %s", modulename); ++ request_module(modulename); ++ set->type = find_set_type_rlock(typename); ++ } ++ if (set->type == NULL) { ++ ip_set_printk("no set type '%s', set '%s' not created", ++ typename, name); ++ res = -ENOENT; ++ goto out; ++ } ++ if (!try_module_get(set->type->me)) { ++ read_unlock_bh(&ip_set_lock); ++ res = -EFAULT; ++ goto out; ++ } ++ read_unlock_bh(&ip_set_lock); ++ ++ /* ++ * Without holding any locks, create private part. ++ */ ++ res = set->type->create(set, data, size); ++ if (res != 0) ++ goto put_out; ++ ++ /* BTW, res==0 here. */ ++ ++ /* ++ * Here, we have a valid, constructed set. &ip_set_lock again, ++ * find free id/index and check that it is not already in ++ * ip_set_list. ++ */ ++ write_lock_bh(&ip_set_lock); ++ if ((res = find_free_id(set->name, &index, &id)) != 0) { ++ DP("no free id!"); ++ goto cleanup; ++ } ++ ++ /* Make sure restore gets the same index */ ++ if (restore != IP_SET_INVALID_ID && index != restore) { ++ DP("Can't restore, sets are screwed up"); ++ res = -ERANGE; ++ goto cleanup; ++ } ++ ++ /* ++ * Finally! Add our shiny new set to the list, and be done. ++ */ ++ DP("create: '%s' created with index %u, id %u!", set->name, index, id); ++ set->id = id; ++ ip_set_list[index] = set; ++ write_unlock_bh(&ip_set_lock); ++ return res; ++ ++ cleanup: ++ write_unlock_bh(&ip_set_lock); ++ set->type->destroy(set); ++ put_out: ++ module_put(set->type->me); ++ out: ++ kfree(set); ++ return res; ++} ++ ++/* ++ * Destroy a given existing set ++ */ ++static void ++ip_set_destroy_set(ip_set_id_t index) ++{ ++ struct ip_set *set = ip_set_list[index]; ++ ++ IP_SET_ASSERT(set); ++ DP("set: %s", set->name); ++ write_lock_bh(&ip_set_lock); ++ FOREACH_HASH_RW_DO(__set_hash_del_byid, set->id); ++ if (set->binding != IP_SET_INVALID_ID) ++ __ip_set_put(set->binding); ++ ip_set_list[index] = NULL; ++ write_unlock_bh(&ip_set_lock); ++ ++ /* Must call it without holding any lock */ ++ set->type->destroy(set); ++ module_put(set->type->me); ++ kfree(set); ++} ++ ++/* ++ * Destroy a set - or all sets ++ * Sets must not be referenced/used. ++ */ ++static int ++ip_set_destroy(ip_set_id_t index) ++{ ++ ip_set_id_t i; ++ ++ /* ref modification always protected by the mutex */ ++ if (index != IP_SET_INVALID_ID) { ++ if (atomic_read(&ip_set_list[index]->ref)) ++ return -EBUSY; ++ ip_set_destroy_set(index); ++ } else { ++ for (i = 0; i < ip_set_max; i++) { ++ if (ip_set_list[i] != NULL ++ && (atomic_read(&ip_set_list[i]->ref))) ++ return -EBUSY; ++ } ++ ++ for (i = 0; i < ip_set_max; i++) { ++ if (ip_set_list[i] != NULL) ++ ip_set_destroy_set(i); ++ } ++ } ++ return 0; ++} ++ ++static void ++ip_set_flush_set(struct ip_set *set) ++{ ++ DP("set: %s %u", set->name, set->id); ++ ++ write_lock_bh(&set->lock); ++ set->type->flush(set); ++ write_unlock_bh(&set->lock); ++} ++ ++/* ++ * Flush data in a set - or in all sets ++ */ ++static int ++ip_set_flush(ip_set_id_t index) ++{ ++ if (index != IP_SET_INVALID_ID) { ++ IP_SET_ASSERT(ip_set_list[index]); ++ ip_set_flush_set(ip_set_list[index]); ++ } else ++ FOREACH_SET_DO(ip_set_flush_set); ++ ++ return 0; ++} ++ ++/* Rename a set */ ++static int ++ip_set_rename(ip_set_id_t index, const char *name) ++{ ++ struct ip_set *set = ip_set_list[index]; ++ ip_set_id_t i; ++ int res = 0; ++ ++ DP("set: %s to %s", set->name, name); ++ write_lock_bh(&ip_set_lock); ++ for (i = 0; i < ip_set_max; i++) { ++ if (ip_set_list[i] != NULL ++ && strncmp(ip_set_list[i]->name, ++ name, ++ IP_SET_MAXNAMELEN - 1) == 0) { ++ res = -EEXIST; ++ goto unlock; ++ } ++ } ++ strncpy(set->name, name, IP_SET_MAXNAMELEN); ++ unlock: ++ write_unlock_bh(&ip_set_lock); ++ return res; ++} ++ ++/* ++ * Swap two sets so that name/index points to the other. ++ * References are also swapped. ++ */ ++static int ++ip_set_swap(ip_set_id_t from_index, ip_set_id_t to_index) ++{ ++ struct ip_set *from = ip_set_list[from_index]; ++ struct ip_set *to = ip_set_list[to_index]; ++ char from_name[IP_SET_MAXNAMELEN]; ++ u_int32_t from_ref; ++ ++ DP("set: %s to %s", from->name, to->name); ++ /* Features must not change. Artifical restriction. */ ++ if (from->type->features != to->type->features) ++ return -ENOEXEC; ++ ++ /* No magic here: ref munging protected by the mutex */ ++ write_lock_bh(&ip_set_lock); ++ strncpy(from_name, from->name, IP_SET_MAXNAMELEN); ++ from_ref = atomic_read(&from->ref); ++ ++ strncpy(from->name, to->name, IP_SET_MAXNAMELEN); ++ atomic_set(&from->ref, atomic_read(&to->ref)); ++ strncpy(to->name, from_name, IP_SET_MAXNAMELEN); ++ atomic_set(&to->ref, from_ref); ++ ++ ip_set_list[from_index] = to; ++ ip_set_list[to_index] = from; ++ ++ write_unlock_bh(&ip_set_lock); ++ return 0; ++} ++ ++/* ++ * List set data ++ */ ++ ++static inline void ++__set_hash_bindings_size_list(struct ip_set_hash *set_hash, ++ ip_set_id_t id, size_t *size) ++{ ++ if (set_hash->id == id) ++ *size += sizeof(struct ip_set_hash_list); ++} ++ ++static inline void ++__set_hash_bindings_size_save(struct ip_set_hash *set_hash, ++ ip_set_id_t id, size_t *size) ++{ ++ if (set_hash->id == id) ++ *size += sizeof(struct ip_set_hash_save); ++} ++ ++static inline void ++__set_hash_bindings(struct ip_set_hash *set_hash, ++ ip_set_id_t id, void *data, int *used) ++{ ++ if (set_hash->id == id) { ++ struct ip_set_hash_list *hash_list = ++ (struct ip_set_hash_list *)(data + *used); ++ ++ hash_list->ip = set_hash->ip; ++ hash_list->binding = set_hash->binding; ++ *used += sizeof(struct ip_set_hash_list); ++ } ++} ++ ++static int ip_set_list_set(ip_set_id_t index, ++ void *data, ++ int *used, ++ int len) ++{ ++ struct ip_set *set = ip_set_list[index]; ++ struct ip_set_list *set_list; ++ ++ /* Pointer to our header */ ++ set_list = (struct ip_set_list *) (data + *used); ++ ++ DP("set: %s, used: %d %p %p", set->name, *used, data, data + *used); ++ ++ /* Get and ensure header size */ ++ if (*used + sizeof(struct ip_set_list) > len) ++ goto not_enough_mem; ++ *used += sizeof(struct ip_set_list); ++ ++ read_lock_bh(&set->lock); ++ /* Get and ensure set specific header size */ ++ set_list->header_size = set->type->header_size; ++ if (*used + set_list->header_size > len) ++ goto unlock_set; ++ ++ /* Fill in the header */ ++ set_list->index = index; ++ set_list->binding = set->binding; ++ set_list->ref = atomic_read(&set->ref); ++ ++ /* Fill in set spefific header data */ ++ set->type->list_header(set, data + *used); ++ *used += set_list->header_size; ++ ++ /* Get and ensure set specific members size */ ++ set_list->members_size = set->type->list_members_size(set); ++ if (*used + set_list->members_size > len) ++ goto unlock_set; ++ ++ /* Fill in set spefific members data */ ++ set->type->list_members(set, data + *used); ++ *used += set_list->members_size; ++ read_unlock_bh(&set->lock); ++ ++ /* Bindings */ ++ ++ /* Get and ensure set specific bindings size */ ++ set_list->bindings_size = 0; ++ FOREACH_HASH_DO(__set_hash_bindings_size_list, ++ set->id, &set_list->bindings_size); ++ if (*used + set_list->bindings_size > len) ++ goto not_enough_mem; ++ ++ /* Fill in set spefific bindings data */ ++ FOREACH_HASH_DO(__set_hash_bindings, set->id, data, used); ++ ++ return 0; ++ ++ unlock_set: ++ read_unlock_bh(&set->lock); ++ not_enough_mem: ++ DP("not enough mem, try again"); ++ return -EAGAIN; ++} ++ ++/* ++ * Save sets ++ */ ++static int ip_set_save_set(ip_set_id_t index, ++ void *data, ++ int *used, ++ int len) ++{ ++ struct ip_set *set; ++ struct ip_set_save *set_save; ++ ++ /* Pointer to our header */ ++ set_save = (struct ip_set_save *) (data + *used); ++ ++ /* Get and ensure header size */ ++ if (*used + sizeof(struct ip_set_save) > len) ++ goto not_enough_mem; ++ *used += sizeof(struct ip_set_save); ++ ++ set = ip_set_list[index]; ++ DP("set: %s, used: %u(%u) %p %p", set->name, *used, len, ++ data, data + *used); ++ ++ read_lock_bh(&set->lock); ++ /* Get and ensure set specific header size */ ++ set_save->header_size = set->type->header_size; ++ if (*used + set_save->header_size > len) ++ goto unlock_set; ++ ++ /* Fill in the header */ ++ set_save->index = index; ++ set_save->binding = set->binding; ++ ++ /* Fill in set spefific header data */ ++ set->type->list_header(set, data + *used); ++ *used += set_save->header_size; ++ ++ DP("set header filled: %s, used: %u(%u) %p %p", set->name, *used, ++ set_save->header_size, data, data + *used); ++ /* Get and ensure set specific members size */ ++ set_save->members_size = set->type->list_members_size(set); ++ if (*used + set_save->members_size > len) ++ goto unlock_set; ++ ++ /* Fill in set spefific members data */ ++ set->type->list_members(set, data + *used); ++ *used += set_save->members_size; ++ read_unlock_bh(&set->lock); ++ DP("set members filled: %s, used: %u(%u) %p %p", set->name, *used, ++ set_save->members_size, data, data + *used); ++ return 0; ++ ++ unlock_set: ++ read_unlock_bh(&set->lock); ++ not_enough_mem: ++ DP("not enough mem, try again"); ++ return -EAGAIN; ++} ++ ++static inline void ++__set_hash_save_bindings(struct ip_set_hash *set_hash, ++ ip_set_id_t id, ++ void *data, ++ int *used, ++ int len, ++ int *res) ++{ ++ if (*res == 0 ++ && (id == IP_SET_INVALID_ID || set_hash->id == id)) { ++ struct ip_set_hash_save *hash_save = ++ (struct ip_set_hash_save *)(data + *used); ++ /* Ensure bindings size */ ++ if (*used + sizeof(struct ip_set_hash_save) > len) { ++ *res = -ENOMEM; ++ return; ++ } ++ hash_save->id = set_hash->id; ++ hash_save->ip = set_hash->ip; ++ hash_save->binding = set_hash->binding; ++ *used += sizeof(struct ip_set_hash_save); ++ } ++} ++ ++static int ip_set_save_bindings(ip_set_id_t index, ++ void *data, ++ int *used, ++ int len) ++{ ++ int res = 0; ++ struct ip_set_save *set_save; ++ ++ DP("used %u, len %u", *used, len); ++ /* Get and ensure header size */ ++ if (*used + sizeof(struct ip_set_save) > len) ++ return -ENOMEM; ++ ++ /* Marker */ ++ set_save = (struct ip_set_save *) (data + *used); ++ set_save->index = IP_SET_INVALID_ID; ++ set_save->header_size = 0; ++ set_save->members_size = 0; ++ *used += sizeof(struct ip_set_save); ++ ++ DP("marker added used %u, len %u", *used, len); ++ /* Fill in bindings data */ ++ if (index != IP_SET_INVALID_ID) ++ /* Sets are identified by id in hash */ ++ index = ip_set_list[index]->id; ++ FOREACH_HASH_DO(__set_hash_save_bindings, index, data, used, len, &res); ++ ++ return res; ++} ++ ++/* ++ * Restore sets ++ */ ++static int ip_set_restore(void *data, ++ int len) ++{ ++ int res = 0; ++ int line = 0, used = 0, members_size; ++ struct ip_set *set; ++ struct ip_set_hash_save *hash_save; ++ struct ip_set_restore *set_restore; ++ ip_set_id_t index; ++ ++ /* Loop to restore sets */ ++ while (1) { ++ line++; ++ ++ DP("%u %u %u", used, sizeof(struct ip_set_restore), len); ++ /* Get and ensure header size */ ++ if (used + sizeof(struct ip_set_restore) > len) ++ return line; ++ set_restore = (struct ip_set_restore *) (data + used); ++ used += sizeof(struct ip_set_restore); ++ ++ /* Ensure data size */ ++ if (used ++ + set_restore->header_size ++ + set_restore->members_size > len) ++ return line; ++ ++ /* Check marker */ ++ if (set_restore->index == IP_SET_INVALID_ID) { ++ line--; ++ goto bindings; ++ } ++ ++ /* Try to create the set */ ++ DP("restore %s %s", set_restore->name, set_restore->typename); ++ res = ip_set_create(set_restore->name, ++ set_restore->typename, ++ set_restore->index, ++ data + used, ++ set_restore->header_size); ++ ++ if (res != 0) ++ return line; ++ used += set_restore->header_size; ++ ++ index = ip_set_find_byindex(set_restore->index); ++ DP("index %u, restore_index %u", index, set_restore->index); ++ if (index != set_restore->index) ++ return line; ++ /* Try to restore members data */ ++ set = ip_set_list[index]; ++ members_size = 0; ++ DP("members_size %u reqsize %u", ++ set_restore->members_size, set->type->reqsize); ++ while (members_size + set->type->reqsize <= ++ set_restore->members_size) { ++ line++; ++ DP("members: %u, line %u", members_size, line); ++ res = __ip_set_addip(index, ++ data + used + members_size, ++ set->type->reqsize); ++ if (!(res == 0 || res == -EEXIST)) ++ return line; ++ members_size += set->type->reqsize; ++ } ++ ++ DP("members_size %u %u", ++ set_restore->members_size, members_size); ++ if (members_size != set_restore->members_size) ++ return line++; ++ used += set_restore->members_size; ++ } ++ ++ bindings: ++ /* Loop to restore bindings */ ++ while (used < len) { ++ line++; ++ ++ DP("restore binding, line %u", line); ++ /* Get and ensure size */ ++ if (used + sizeof(struct ip_set_hash_save) > len) ++ return line; ++ hash_save = (struct ip_set_hash_save *) (data + used); ++ used += sizeof(struct ip_set_hash_save); ++ ++ /* hash_save->id is used to store the index */ ++ index = ip_set_find_byindex(hash_save->id); ++ DP("restore binding index %u, id %u, %u -> %u", ++ index, hash_save->id, hash_save->ip, hash_save->binding); ++ if (index != hash_save->id) ++ return line; ++ if (ip_set_find_byindex(hash_save->binding) == IP_SET_INVALID_ID) { ++ DP("corrupt binding set index %u", hash_save->binding); ++ return line; ++ } ++ set = ip_set_list[hash_save->id]; ++ /* Null valued IP means default binding */ ++ if (hash_save->ip) ++ res = ip_set_hash_add(set->id, ++ hash_save->ip, ++ hash_save->binding); ++ else { ++ IP_SET_ASSERT(set->binding == IP_SET_INVALID_ID); ++ write_lock_bh(&ip_set_lock); ++ set->binding = hash_save->binding; ++ __ip_set_get(set->binding); ++ write_unlock_bh(&ip_set_lock); ++ DP("default binding: %u", set->binding); ++ } ++ if (res != 0) ++ return line; ++ } ++ if (used != len) ++ return line; ++ ++ return 0; ++} ++ ++static int ++ip_set_sockfn_set(struct sock *sk, int optval, void *user, unsigned int len) ++{ ++ void *data; ++ int res = 0; /* Assume OK */ ++ unsigned *op; ++ struct ip_set_req_adt *req_adt; ++ ip_set_id_t index = IP_SET_INVALID_ID; ++ int (*adtfn)(ip_set_id_t index, ++ const void *data, size_t size); ++ struct fn_table { ++ int (*fn)(ip_set_id_t index, ++ const void *data, size_t size); ++ } adtfn_table[] = ++ { { ip_set_addip }, { ip_set_delip }, { ip_set_testip}, ++ { ip_set_bindip}, { ip_set_unbindip }, { ip_set_testbind }, ++ }; ++ ++ DP("optval=%d, user=%p, len=%d", optval, user, len); ++ if (!capable(CAP_NET_ADMIN)) ++ return -EPERM; ++ if (optval != SO_IP_SET) ++ return -EBADF; ++ if (len <= sizeof(unsigned)) { ++ ip_set_printk("short userdata (want >%zu, got %u)", ++ sizeof(unsigned), len); ++ return -EINVAL; ++ } ++ data = vmalloc(len); ++ if (!data) { ++ DP("out of mem for %u bytes", len); ++ return -ENOMEM; ++ } ++ if (copy_from_user(data, user, len) != 0) { ++ res = -EFAULT; ++ goto done; ++ } ++ if (down_interruptible(&ip_set_app_mutex)) { ++ res = -EINTR; ++ goto done; ++ } ++ ++ op = (unsigned *)data; ++ DP("op=%x", *op); ++ ++ if (*op < IP_SET_OP_VERSION) { ++ /* Check the version at the beginning of operations */ ++ struct ip_set_req_version *req_version = ++ (struct ip_set_req_version *) data; ++ if (req_version->version != IP_SET_PROTOCOL_VERSION) { ++ res = -EPROTO; ++ goto done; ++ } ++ } ++ ++ switch (*op) { ++ case IP_SET_OP_CREATE:{ ++ struct ip_set_req_create *req_create ++ = (struct ip_set_req_create *) data; ++ ++ if (len < sizeof(struct ip_set_req_create)) { ++ ip_set_printk("short CREATE data (want >=%zu, got %u)", ++ sizeof(struct ip_set_req_create), len); ++ res = -EINVAL; ++ goto done; ++ } ++ req_create->name[IP_SET_MAXNAMELEN - 1] = '\0'; ++ req_create->typename[IP_SET_MAXNAMELEN - 1] = '\0'; ++ res = ip_set_create(req_create->name, ++ req_create->typename, ++ IP_SET_INVALID_ID, ++ data + sizeof(struct ip_set_req_create), ++ len - sizeof(struct ip_set_req_create)); ++ goto done; ++ } ++ case IP_SET_OP_DESTROY:{ ++ struct ip_set_req_std *req_destroy ++ = (struct ip_set_req_std *) data; ++ ++ if (len != sizeof(struct ip_set_req_std)) { ++ ip_set_printk("invalid DESTROY data (want %zu, got %u)", ++ sizeof(struct ip_set_req_std), len); ++ res = -EINVAL; ++ goto done; ++ } ++ if (strcmp(req_destroy->name, IPSET_TOKEN_ALL) == 0) { ++ /* Destroy all sets */ ++ index = IP_SET_INVALID_ID; ++ } else { ++ req_destroy->name[IP_SET_MAXNAMELEN - 1] = '\0'; ++ index = ip_set_find_byname(req_destroy->name); ++ ++ if (index == IP_SET_INVALID_ID) { ++ res = -ENOENT; ++ goto done; ++ } ++ } ++ ++ res = ip_set_destroy(index); ++ goto done; ++ } ++ case IP_SET_OP_FLUSH:{ ++ struct ip_set_req_std *req_flush = ++ (struct ip_set_req_std *) data; ++ ++ if (len != sizeof(struct ip_set_req_std)) { ++ ip_set_printk("invalid FLUSH data (want %zu, got %u)", ++ sizeof(struct ip_set_req_std), len); ++ res = -EINVAL; ++ goto done; ++ } ++ if (strcmp(req_flush->name, IPSET_TOKEN_ALL) == 0) { ++ /* Flush all sets */ ++ index = IP_SET_INVALID_ID; ++ } else { ++ req_flush->name[IP_SET_MAXNAMELEN - 1] = '\0'; ++ index = ip_set_find_byname(req_flush->name); ++ ++ if (index == IP_SET_INVALID_ID) { ++ res = -ENOENT; ++ goto done; ++ } ++ } ++ res = ip_set_flush(index); ++ goto done; ++ } ++ case IP_SET_OP_RENAME:{ ++ struct ip_set_req_create *req_rename ++ = (struct ip_set_req_create *) data; ++ ++ if (len != sizeof(struct ip_set_req_create)) { ++ ip_set_printk("invalid RENAME data (want %zu, got %u)", ++ sizeof(struct ip_set_req_create), len); ++ res = -EINVAL; ++ goto done; ++ } ++ ++ req_rename->name[IP_SET_MAXNAMELEN - 1] = '\0'; ++ req_rename->typename[IP_SET_MAXNAMELEN - 1] = '\0'; ++ ++ index = ip_set_find_byname(req_rename->name); ++ if (index == IP_SET_INVALID_ID) { ++ res = -ENOENT; ++ goto done; ++ } ++ res = ip_set_rename(index, req_rename->typename); ++ goto done; ++ } ++ case IP_SET_OP_SWAP:{ ++ struct ip_set_req_create *req_swap ++ = (struct ip_set_req_create *) data; ++ ip_set_id_t to_index; ++ ++ if (len != sizeof(struct ip_set_req_create)) { ++ ip_set_printk("invalid SWAP data (want %zu, got %u)", ++ sizeof(struct ip_set_req_create), len); ++ res = -EINVAL; ++ goto done; ++ } ++ ++ req_swap->name[IP_SET_MAXNAMELEN - 1] = '\0'; ++ req_swap->typename[IP_SET_MAXNAMELEN - 1] = '\0'; ++ ++ index = ip_set_find_byname(req_swap->name); ++ if (index == IP_SET_INVALID_ID) { ++ res = -ENOENT; ++ goto done; ++ } ++ to_index = ip_set_find_byname(req_swap->typename); ++ if (to_index == IP_SET_INVALID_ID) { ++ res = -ENOENT; ++ goto done; ++ } ++ res = ip_set_swap(index, to_index); ++ goto done; ++ } ++ default: ++ break; /* Set identified by id */ ++ } ++ ++ /* There we may have add/del/test/bind/unbind/test_bind operations */ ++ if (*op < IP_SET_OP_ADD_IP || *op > IP_SET_OP_TEST_BIND_SET) { ++ res = -EBADMSG; ++ goto done; ++ } ++ adtfn = adtfn_table[*op - IP_SET_OP_ADD_IP].fn; ++ ++ if (len < sizeof(struct ip_set_req_adt)) { ++ ip_set_printk("short data in adt request (want >=%zu, got %u)", ++ sizeof(struct ip_set_req_adt), len); ++ res = -EINVAL; ++ goto done; ++ } ++ req_adt = (struct ip_set_req_adt *) data; ++ ++ /* -U :all: :all:|:default: uses IP_SET_INVALID_ID */ ++ if (!(*op == IP_SET_OP_UNBIND_SET ++ && req_adt->index == IP_SET_INVALID_ID)) { ++ index = ip_set_find_byindex(req_adt->index); ++ if (index == IP_SET_INVALID_ID) { ++ res = -ENOENT; ++ goto done; ++ } ++ } ++ res = adtfn(index, data, len); ++ ++ done: ++ up(&ip_set_app_mutex); ++ vfree(data); ++ if (res > 0) ++ res = 0; ++ DP("final result %d", res); ++ return res; ++} ++ ++static int ++ip_set_sockfn_get(struct sock *sk, int optval, void *user, int *len) ++{ ++ int res = 0; ++ unsigned *op; ++ ip_set_id_t index = IP_SET_INVALID_ID; ++ void *data; ++ int copylen = *len; ++ ++ DP("optval=%d, user=%p, len=%d", optval, user, *len); ++ if (!capable(CAP_NET_ADMIN)) ++ return -EPERM; ++ if (optval != SO_IP_SET) ++ return -EBADF; ++ if (*len < sizeof(unsigned)) { ++ ip_set_printk("short userdata (want >=%zu, got %d)", ++ sizeof(unsigned), *len); ++ return -EINVAL; ++ } ++ data = vmalloc(*len); ++ if (!data) { ++ DP("out of mem for %d bytes", *len); ++ return -ENOMEM; ++ } ++ if (copy_from_user(data, user, *len) != 0) { ++ res = -EFAULT; ++ goto done; ++ } ++ if (down_interruptible(&ip_set_app_mutex)) { ++ res = -EINTR; ++ goto done; ++ } ++ ++ op = (unsigned *) data; ++ DP("op=%x", *op); ++ ++ if (*op < IP_SET_OP_VERSION) { ++ /* Check the version at the beginning of operations */ ++ struct ip_set_req_version *req_version = ++ (struct ip_set_req_version *) data; ++ if (req_version->version != IP_SET_PROTOCOL_VERSION) { ++ res = -EPROTO; ++ goto done; ++ } ++ } ++ ++ switch (*op) { ++ case IP_SET_OP_VERSION: { ++ struct ip_set_req_version *req_version = ++ (struct ip_set_req_version *) data; ++ ++ if (*len != sizeof(struct ip_set_req_version)) { ++ ip_set_printk("invalid VERSION (want %zu, got %d)", ++ sizeof(struct ip_set_req_version), ++ *len); ++ res = -EINVAL; ++ goto done; ++ } ++ ++ req_version->version = IP_SET_PROTOCOL_VERSION; ++ res = copy_to_user(user, req_version, ++ sizeof(struct ip_set_req_version)); ++ goto done; ++ } ++ case IP_SET_OP_GET_BYNAME: { ++ struct ip_set_req_get_set *req_get ++ = (struct ip_set_req_get_set *) data; ++ ++ if (*len != sizeof(struct ip_set_req_get_set)) { ++ ip_set_printk("invalid GET_BYNAME (want %zu, got %d)", ++ sizeof(struct ip_set_req_get_set), *len); ++ res = -EINVAL; ++ goto done; ++ } ++ req_get->set.name[IP_SET_MAXNAMELEN - 1] = '\0'; ++ index = ip_set_find_byname(req_get->set.name); ++ req_get->set.index = index; ++ goto copy; ++ } ++ case IP_SET_OP_GET_BYINDEX: { ++ struct ip_set_req_get_set *req_get ++ = (struct ip_set_req_get_set *) data; ++ ++ if (*len != sizeof(struct ip_set_req_get_set)) { ++ ip_set_printk("invalid GET_BYINDEX (want %zu, got %d)", ++ sizeof(struct ip_set_req_get_set), *len); ++ res = -EINVAL; ++ goto done; ++ } ++ req_get->set.name[IP_SET_MAXNAMELEN - 1] = '\0'; ++ index = ip_set_find_byindex(req_get->set.index); ++ strncpy(req_get->set.name, ++ index == IP_SET_INVALID_ID ? "" ++ : ip_set_list[index]->name, IP_SET_MAXNAMELEN); ++ goto copy; ++ } ++ case IP_SET_OP_ADT_GET: { ++ struct ip_set_req_adt_get *req_get ++ = (struct ip_set_req_adt_get *) data; ++ ++ if (*len != sizeof(struct ip_set_req_adt_get)) { ++ ip_set_printk("invalid ADT_GET (want %zu, got %d)", ++ sizeof(struct ip_set_req_adt_get), *len); ++ res = -EINVAL; ++ goto done; ++ } ++ req_get->set.name[IP_SET_MAXNAMELEN - 1] = '\0'; ++ index = ip_set_find_byname(req_get->set.name); ++ if (index != IP_SET_INVALID_ID) { ++ req_get->set.index = index; ++ strncpy(req_get->typename, ++ ip_set_list[index]->type->typename, ++ IP_SET_MAXNAMELEN - 1); ++ } else { ++ res = -ENOENT; ++ goto done; ++ } ++ goto copy; ++ } ++ case IP_SET_OP_MAX_SETS: { ++ struct ip_set_req_max_sets *req_max_sets ++ = (struct ip_set_req_max_sets *) data; ++ ip_set_id_t i; ++ ++ if (*len != sizeof(struct ip_set_req_max_sets)) { ++ ip_set_printk("invalid MAX_SETS (want %zu, got %d)", ++ sizeof(struct ip_set_req_max_sets), *len); ++ res = -EINVAL; ++ goto done; ++ } ++ ++ if (strcmp(req_max_sets->set.name, IPSET_TOKEN_ALL) == 0) { ++ req_max_sets->set.index = IP_SET_INVALID_ID; ++ } else { ++ req_max_sets->set.name[IP_SET_MAXNAMELEN - 1] = '\0'; ++ req_max_sets->set.index = ++ ip_set_find_byname(req_max_sets->set.name); ++ if (req_max_sets->set.index == IP_SET_INVALID_ID) { ++ res = -ENOENT; ++ goto done; ++ } ++ } ++ req_max_sets->max_sets = ip_set_max; ++ req_max_sets->sets = 0; ++ for (i = 0; i < ip_set_max; i++) { ++ if (ip_set_list[i] != NULL) ++ req_max_sets->sets++; ++ } ++ goto copy; ++ } ++ case IP_SET_OP_LIST_SIZE: ++ case IP_SET_OP_SAVE_SIZE: { ++ struct ip_set_req_setnames *req_setnames ++ = (struct ip_set_req_setnames *) data; ++ struct ip_set_name_list *name_list; ++ struct ip_set *set; ++ ip_set_id_t i; ++ int used; ++ ++ if (*len < sizeof(struct ip_set_req_setnames)) { ++ ip_set_printk("short LIST_SIZE (want >=%zu, got %d)", ++ sizeof(struct ip_set_req_setnames), *len); ++ res = -EINVAL; ++ goto done; ++ } ++ ++ req_setnames->size = 0; ++ used = sizeof(struct ip_set_req_setnames); ++ for (i = 0; i < ip_set_max; i++) { ++ if (ip_set_list[i] == NULL) ++ continue; ++ name_list = (struct ip_set_name_list *) ++ (data + used); ++ used += sizeof(struct ip_set_name_list); ++ if (used > copylen) { ++ res = -EAGAIN; ++ goto done; ++ } ++ set = ip_set_list[i]; ++ /* Fill in index, name, etc. */ ++ name_list->index = i; ++ name_list->id = set->id; ++ strncpy(name_list->name, ++ set->name, ++ IP_SET_MAXNAMELEN - 1); ++ strncpy(name_list->typename, ++ set->type->typename, ++ IP_SET_MAXNAMELEN - 1); ++ DP("filled %s of type %s, index %u\n", ++ name_list->name, name_list->typename, ++ name_list->index); ++ if (!(req_setnames->index == IP_SET_INVALID_ID ++ || req_setnames->index == i)) ++ continue; ++ /* Update size */ ++ switch (*op) { ++ case IP_SET_OP_LIST_SIZE: { ++ req_setnames->size += sizeof(struct ip_set_list) ++ + set->type->header_size ++ + set->type->list_members_size(set); ++ /* Sets are identified by id in the hash */ ++ FOREACH_HASH_DO(__set_hash_bindings_size_list, ++ set->id, &req_setnames->size); ++ break; ++ } ++ case IP_SET_OP_SAVE_SIZE: { ++ req_setnames->size += sizeof(struct ip_set_save) ++ + set->type->header_size ++ + set->type->list_members_size(set); ++ FOREACH_HASH_DO(__set_hash_bindings_size_save, ++ set->id, &req_setnames->size); ++ break; ++ } ++ default: ++ break; ++ } ++ } ++ if (copylen != used) { ++ res = -EAGAIN; ++ goto done; ++ } ++ goto copy; ++ } ++ case IP_SET_OP_LIST: { ++ struct ip_set_req_list *req_list ++ = (struct ip_set_req_list *) data; ++ ip_set_id_t i; ++ int used; ++ ++ if (*len < sizeof(struct ip_set_req_list)) { ++ ip_set_printk("short LIST (want >=%zu, got %d)", ++ sizeof(struct ip_set_req_list), *len); ++ res = -EINVAL; ++ goto done; ++ } ++ index = req_list->index; ++ if (index != IP_SET_INVALID_ID ++ && ip_set_find_byindex(index) != index) { ++ res = -ENOENT; ++ goto done; ++ } ++ used = 0; ++ if (index == IP_SET_INVALID_ID) { ++ /* List all sets */ ++ for (i = 0; i < ip_set_max && res == 0; i++) { ++ if (ip_set_list[i] != NULL) ++ res = ip_set_list_set(i, data, &used, *len); ++ } ++ } else { ++ /* List an individual set */ ++ res = ip_set_list_set(index, data, &used, *len); ++ } ++ if (res != 0) ++ goto done; ++ else if (copylen != used) { ++ res = -EAGAIN; ++ goto done; ++ } ++ goto copy; ++ } ++ case IP_SET_OP_SAVE: { ++ struct ip_set_req_list *req_save ++ = (struct ip_set_req_list *) data; ++ ip_set_id_t i; ++ int used; ++ ++ if (*len < sizeof(struct ip_set_req_list)) { ++ ip_set_printk("short SAVE (want >=%zu, got %d)", ++ sizeof(struct ip_set_req_list), *len); ++ res = -EINVAL; ++ goto done; ++ } ++ index = req_save->index; ++ if (index != IP_SET_INVALID_ID ++ && ip_set_find_byindex(index) != index) { ++ res = -ENOENT; ++ goto done; ++ } ++ used = 0; ++ if (index == IP_SET_INVALID_ID) { ++ /* Save all sets */ ++ for (i = 0; i < ip_set_max && res == 0; i++) { ++ if (ip_set_list[i] != NULL) ++ res = ip_set_save_set(i, data, &used, *len); ++ } ++ } else { ++ /* Save an individual set */ ++ res = ip_set_save_set(index, data, &used, *len); ++ } ++ if (res == 0) ++ res = ip_set_save_bindings(index, data, &used, *len); ++ ++ if (res != 0) ++ goto done; ++ else if (copylen != used) { ++ res = -EAGAIN; ++ goto done; ++ } ++ goto copy; ++ } ++ case IP_SET_OP_RESTORE: { ++ struct ip_set_req_setnames *req_restore ++ = (struct ip_set_req_setnames *) data; ++ int line; ++ ++ if (*len < sizeof(struct ip_set_req_setnames) ++ || *len != req_restore->size) { ++ ip_set_printk("invalid RESTORE (want =%zu, got %d)", ++ req_restore->size, *len); ++ res = -EINVAL; ++ goto done; ++ } ++ line = ip_set_restore(data + sizeof(struct ip_set_req_setnames), ++ req_restore->size - sizeof(struct ip_set_req_setnames)); ++ DP("ip_set_restore: %u", line); ++ if (line != 0) { ++ res = -EAGAIN; ++ req_restore->size = line; ++ copylen = sizeof(struct ip_set_req_setnames); ++ goto copy; ++ } ++ goto done; ++ } ++ default: ++ res = -EBADMSG; ++ goto done; ++ } /* end of switch(op) */ ++ ++ copy: ++ DP("set %s, copylen %u", index != IP_SET_INVALID_ID ++ && ip_set_list[index] ++ ? ip_set_list[index]->name ++ : ":all:", copylen); ++ res = copy_to_user(user, data, copylen); ++ ++ done: ++ up(&ip_set_app_mutex); ++ vfree(data); ++ if (res > 0) ++ res = 0; ++ DP("final result %d", res); ++ return res; ++} ++ ++static struct nf_sockopt_ops so_set = { ++ .pf = PF_INET, ++ .set_optmin = SO_IP_SET, ++ .set_optmax = SO_IP_SET + 1, ++ .set = &ip_set_sockfn_set, ++ .get_optmin = SO_IP_SET, ++ .get_optmax = SO_IP_SET + 1, ++ .get = &ip_set_sockfn_get, ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) ++ .owner = THIS_MODULE, ++#endif ++}; ++ ++static int max_sets, hash_size; ++module_param(max_sets, int, 0600); ++MODULE_PARM_DESC(max_sets, "maximal number of sets"); ++module_param(hash_size, int, 0600); ++MODULE_PARM_DESC(hash_size, "hash size for bindings"); ++MODULE_LICENSE("GPL"); ++MODULE_AUTHOR("Jozsef Kadlecsik "); ++MODULE_DESCRIPTION("module implementing core IP set support"); ++ ++static int __init ip_set_init(void) ++{ ++ int res; ++ ip_set_id_t i; ++ ++ get_random_bytes(&ip_set_hash_random, 4); ++ if (max_sets) ++ ip_set_max = max_sets; ++ ip_set_list = vmalloc(sizeof(struct ip_set *) * ip_set_max); ++ if (!ip_set_list) { ++ printk(KERN_ERR "Unable to create ip_set_list\n"); ++ return -ENOMEM; ++ } ++ memset(ip_set_list, 0, sizeof(struct ip_set *) * ip_set_max); ++ if (hash_size) ++ ip_set_bindings_hash_size = hash_size; ++ ip_set_hash = vmalloc(sizeof(struct list_head) * ip_set_bindings_hash_size); ++ if (!ip_set_hash) { ++ printk(KERN_ERR "Unable to create ip_set_hash\n"); ++ vfree(ip_set_list); ++ return -ENOMEM; ++ } ++ for (i = 0; i < ip_set_bindings_hash_size; i++) ++ INIT_LIST_HEAD(&ip_set_hash[i]); ++ ++ INIT_LIST_HEAD(&set_type_list); ++ ++ res = nf_register_sockopt(&so_set); ++ if (res != 0) { ++ ip_set_printk("SO_SET registry failed: %d", res); ++ vfree(ip_set_list); ++ vfree(ip_set_hash); ++ return res; ++ } ++ return 0; ++} ++ ++static void __exit ip_set_fini(void) ++{ ++ /* There can't be any existing set or binding */ ++ nf_unregister_sockopt(&so_set); ++ vfree(ip_set_list); ++ vfree(ip_set_hash); ++ DP("these are the famous last words"); ++} ++ ++EXPORT_SYMBOL(ip_set_register_set_type); ++EXPORT_SYMBOL(ip_set_unregister_set_type); ++ ++EXPORT_SYMBOL(ip_set_get_byname); ++EXPORT_SYMBOL(ip_set_get_byindex); ++EXPORT_SYMBOL(ip_set_put); ++ ++EXPORT_SYMBOL(ip_set_addip_kernel); ++EXPORT_SYMBOL(ip_set_delip_kernel); ++EXPORT_SYMBOL(ip_set_testip_kernel); ++ ++module_init(ip_set_init); ++module_exit(ip_set_fini); +--- /dev/null ++++ b/net/ipv4/netfilter/ip_set_iphash.c +@@ -0,0 +1,429 @@ ++/* Copyright (C) 2003-2004 Jozsef Kadlecsik ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++/* Kernel module implementing an ip hash set */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++ ++#include ++#include ++ ++static int limit = MAX_RANGE; ++ ++static inline __u32 ++jhash_ip(const struct ip_set_iphash *map, uint16_t i, ip_set_ip_t ip) ++{ ++ return jhash_1word(ip, *(((uint32_t *) map->initval) + i)); ++} ++ ++static inline __u32 ++hash_id(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_iphash *map = (struct ip_set_iphash *) set->data; ++ __u32 id; ++ u_int16_t i; ++ ip_set_ip_t *elem; ++ ++ *hash_ip = ip & map->netmask; ++ DP("set: %s, ip:%u.%u.%u.%u, %u.%u.%u.%u, %u.%u.%u.%u", ++ set->name, HIPQUAD(ip), HIPQUAD(*hash_ip), HIPQUAD(map->netmask)); ++ ++ for (i = 0; i < map->probes; i++) { ++ id = jhash_ip(map, i, *hash_ip) % map->hashsize; ++ DP("hash key: %u", id); ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); ++ if (*elem == *hash_ip) ++ return id; ++ /* No shortcut at testing - there can be deleted ++ * entries. */ ++ } ++ return UINT_MAX; ++} ++ ++static inline int ++__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ return (ip && hash_id(set, ip, hash_ip) != UINT_MAX); ++} ++ ++static int ++testip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_iphash *req = ++ (struct ip_set_req_iphash *) data; ++ ++ if (size != sizeof(struct ip_set_req_iphash)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_iphash), ++ size); ++ return -EINVAL; ++ } ++ return __testip(set, req->ip, hash_ip); ++} ++ ++static int ++testip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ return __testip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip); ++} ++ ++static inline int ++__addip(struct ip_set_iphash *map, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ __u32 probe; ++ u_int16_t i; ++ ip_set_ip_t *elem; ++ ++ if (!ip || map->elements >= limit) ++ return -ERANGE; ++ ++ *hash_ip = ip & map->netmask; ++ ++ for (i = 0; i < map->probes; i++) { ++ probe = jhash_ip(map, i, *hash_ip) % map->hashsize; ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, probe); ++ if (*elem == *hash_ip) ++ return -EEXIST; ++ if (!*elem) { ++ *elem = *hash_ip; ++ map->elements++; ++ return 0; ++ } ++ } ++ /* Trigger rehashing */ ++ return -EAGAIN; ++} ++ ++static int ++addip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_iphash *req = ++ (struct ip_set_req_iphash *) data; ++ ++ if (size != sizeof(struct ip_set_req_iphash)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_iphash), ++ size); ++ return -EINVAL; ++ } ++ return __addip((struct ip_set_iphash *) set->data, req->ip, hash_ip); ++} ++ ++static int ++addip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ return __addip((struct ip_set_iphash *) set->data, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip); ++} ++ ++static int retry(struct ip_set *set) ++{ ++ struct ip_set_iphash *map = (struct ip_set_iphash *) set->data; ++ ip_set_ip_t hash_ip, *elem; ++ void *members; ++ u_int32_t i, hashsize = map->hashsize; ++ int res; ++ struct ip_set_iphash *tmp; ++ ++ if (map->resize == 0) ++ return -ERANGE; ++ ++ again: ++ res = 0; ++ ++ /* Calculate new hash size */ ++ hashsize += (hashsize * map->resize)/100; ++ if (hashsize == map->hashsize) ++ hashsize++; ++ ++ ip_set_printk("rehashing of set %s triggered: " ++ "hashsize grows from %u to %u", ++ set->name, map->hashsize, hashsize); ++ ++ tmp = kmalloc(sizeof(struct ip_set_iphash) ++ + map->probes * sizeof(uint32_t), GFP_ATOMIC); ++ if (!tmp) { ++ DP("out of memory for %d bytes", ++ sizeof(struct ip_set_iphash) ++ + map->probes * sizeof(uint32_t)); ++ return -ENOMEM; ++ } ++ tmp->members = harray_malloc(hashsize, sizeof(ip_set_ip_t), GFP_ATOMIC); ++ if (!tmp->members) { ++ DP("out of memory for %d bytes", hashsize * sizeof(ip_set_ip_t)); ++ kfree(tmp); ++ return -ENOMEM; ++ } ++ tmp->hashsize = hashsize; ++ tmp->elements = 0; ++ tmp->probes = map->probes; ++ tmp->resize = map->resize; ++ tmp->netmask = map->netmask; ++ memcpy(tmp->initval, map->initval, map->probes * sizeof(uint32_t)); ++ ++ write_lock_bh(&set->lock); ++ map = (struct ip_set_iphash *) set->data; /* Play safe */ ++ for (i = 0; i < map->hashsize && res == 0; i++) { ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i); ++ if (*elem) ++ res = __addip(tmp, *elem, &hash_ip); ++ } ++ if (res) { ++ /* Failure, try again */ ++ write_unlock_bh(&set->lock); ++ harray_free(tmp->members); ++ kfree(tmp); ++ goto again; ++ } ++ ++ /* Success at resizing! */ ++ members = map->members; ++ ++ map->hashsize = tmp->hashsize; ++ map->members = tmp->members; ++ write_unlock_bh(&set->lock); ++ ++ harray_free(members); ++ kfree(tmp); ++ ++ return 0; ++} ++ ++static inline int ++__delip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_iphash *map = (struct ip_set_iphash *) set->data; ++ ip_set_ip_t id, *elem; ++ ++ if (!ip) ++ return -ERANGE; ++ ++ id = hash_id(set, ip, hash_ip); ++ if (id == UINT_MAX) ++ return -EEXIST; ++ ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); ++ *elem = 0; ++ map->elements--; ++ ++ return 0; ++} ++ ++static int ++delip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_iphash *req = ++ (struct ip_set_req_iphash *) data; ++ ++ if (size != sizeof(struct ip_set_req_iphash)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_iphash), ++ size); ++ return -EINVAL; ++ } ++ return __delip(set, req->ip, hash_ip); ++} ++ ++static int ++delip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ return __delip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip); ++} ++ ++static int create(struct ip_set *set, const void *data, size_t size) ++{ ++ struct ip_set_req_iphash_create *req = ++ (struct ip_set_req_iphash_create *) data; ++ struct ip_set_iphash *map; ++ uint16_t i; ++ ++ if (size != sizeof(struct ip_set_req_iphash_create)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_iphash_create), ++ size); ++ return -EINVAL; ++ } ++ ++ if (req->hashsize < 1) { ++ ip_set_printk("hashsize too small"); ++ return -ENOEXEC; ++ } ++ ++ if (req->probes < 1) { ++ ip_set_printk("probes too small"); ++ return -ENOEXEC; ++ } ++ ++ map = kmalloc(sizeof(struct ip_set_iphash) ++ + req->probes * sizeof(uint32_t), GFP_KERNEL); ++ if (!map) { ++ DP("out of memory for %d bytes", ++ sizeof(struct ip_set_iphash) ++ + req->probes * sizeof(uint32_t)); ++ return -ENOMEM; ++ } ++ for (i = 0; i < req->probes; i++) ++ get_random_bytes(((uint32_t *) map->initval)+i, 4); ++ map->elements = 0; ++ map->hashsize = req->hashsize; ++ map->probes = req->probes; ++ map->resize = req->resize; ++ map->netmask = req->netmask; ++ map->members = harray_malloc(map->hashsize, sizeof(ip_set_ip_t), GFP_KERNEL); ++ if (!map->members) { ++ DP("out of memory for %d bytes", map->hashsize * sizeof(ip_set_ip_t)); ++ kfree(map); ++ return -ENOMEM; ++ } ++ ++ set->data = map; ++ return 0; ++} ++ ++static void destroy(struct ip_set *set) ++{ ++ struct ip_set_iphash *map = (struct ip_set_iphash *) set->data; ++ ++ harray_free(map->members); ++ kfree(map); ++ ++ set->data = NULL; ++} ++ ++static void flush(struct ip_set *set) ++{ ++ struct ip_set_iphash *map = (struct ip_set_iphash *) set->data; ++ harray_flush(map->members, map->hashsize, sizeof(ip_set_ip_t)); ++ map->elements = 0; ++} ++ ++static void list_header(const struct ip_set *set, void *data) ++{ ++ struct ip_set_iphash *map = (struct ip_set_iphash *) set->data; ++ struct ip_set_req_iphash_create *header = ++ (struct ip_set_req_iphash_create *) data; ++ ++ header->hashsize = map->hashsize; ++ header->probes = map->probes; ++ header->resize = map->resize; ++ header->netmask = map->netmask; ++} ++ ++static int list_members_size(const struct ip_set *set) ++{ ++ struct ip_set_iphash *map = (struct ip_set_iphash *) set->data; ++ ++ return (map->hashsize * sizeof(ip_set_ip_t)); ++} ++ ++static void list_members(const struct ip_set *set, void *data) ++{ ++ struct ip_set_iphash *map = (struct ip_set_iphash *) set->data; ++ ip_set_ip_t i, *elem; ++ ++ for (i = 0; i < map->hashsize; i++) { ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i); ++ ((ip_set_ip_t *)data)[i] = *elem; ++ } ++} ++ ++static struct ip_set_type ip_set_iphash = { ++ .typename = SETTYPE_NAME, ++ .features = IPSET_TYPE_IP | IPSET_DATA_SINGLE, ++ .protocol_version = IP_SET_PROTOCOL_VERSION, ++ .create = &create, ++ .destroy = &destroy, ++ .flush = &flush, ++ .reqsize = sizeof(struct ip_set_req_iphash), ++ .addip = &addip, ++ .addip_kernel = &addip_kernel, ++ .retry = &retry, ++ .delip = &delip, ++ .delip_kernel = &delip_kernel, ++ .testip = &testip, ++ .testip_kernel = &testip_kernel, ++ .header_size = sizeof(struct ip_set_req_iphash_create), ++ .list_header = &list_header, ++ .list_members_size = &list_members_size, ++ .list_members = &list_members, ++ .me = THIS_MODULE, ++}; ++ ++MODULE_LICENSE("GPL"); ++MODULE_AUTHOR("Jozsef Kadlecsik "); ++MODULE_DESCRIPTION("iphash type of IP sets"); ++module_param(limit, int, 0600); ++MODULE_PARM_DESC(limit, "maximal number of elements stored in the sets"); ++ ++static int __init ip_set_iphash_init(void) ++{ ++ return ip_set_register_set_type(&ip_set_iphash); ++} ++ ++static void __exit ip_set_iphash_fini(void) ++{ ++ /* FIXME: possible race with ip_set_create() */ ++ ip_set_unregister_set_type(&ip_set_iphash); ++} ++ ++module_init(ip_set_iphash_init); ++module_exit(ip_set_iphash_fini); +--- /dev/null ++++ b/net/ipv4/netfilter/ip_set_ipmap.c +@@ -0,0 +1,336 @@ ++/* Copyright (C) 2000-2002 Joakim Axelsson ++ * Patrick Schaaf ++ * Copyright (C) 2003-2004 Jozsef Kadlecsik ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++/* Kernel module implementing an IP set type: the single bitmap type */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++ ++static inline ip_set_ip_t ++ip_to_id(const struct ip_set_ipmap *map, ip_set_ip_t ip) ++{ ++ return (ip - map->first_ip)/map->hosts; ++} ++ ++static inline int ++__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_ipmap *map = (struct ip_set_ipmap *) set->data; ++ ++ if (ip < map->first_ip || ip > map->last_ip) ++ return -ERANGE; ++ ++ *hash_ip = ip & map->netmask; ++ DP("set: %s, ip:%u.%u.%u.%u, %u.%u.%u.%u", ++ set->name, HIPQUAD(ip), HIPQUAD(*hash_ip)); ++ return !!test_bit(ip_to_id(map, *hash_ip), map->members); ++} ++ ++static int ++testip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_ipmap *req = ++ (struct ip_set_req_ipmap *) data; ++ ++ if (size != sizeof(struct ip_set_req_ipmap)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_ipmap), ++ size); ++ return -EINVAL; ++ } ++ return __testip(set, req->ip, hash_ip); ++} ++ ++static int ++testip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ int res = __testip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip); ++ return (res < 0 ? 0 : res); ++} ++ ++static inline int ++__addip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_ipmap *map = (struct ip_set_ipmap *) set->data; ++ ++ if (ip < map->first_ip || ip > map->last_ip) ++ return -ERANGE; ++ ++ *hash_ip = ip & map->netmask; ++ DP("%u.%u.%u.%u, %u.%u.%u.%u", HIPQUAD(ip), HIPQUAD(*hash_ip)); ++ if (test_and_set_bit(ip_to_id(map, *hash_ip), map->members)) ++ return -EEXIST; ++ ++ return 0; ++} ++ ++static int ++addip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_ipmap *req = ++ (struct ip_set_req_ipmap *) data; ++ ++ if (size != sizeof(struct ip_set_req_ipmap)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_ipmap), ++ size); ++ return -EINVAL; ++ } ++ DP("%u.%u.%u.%u", HIPQUAD(req->ip)); ++ return __addip(set, req->ip, hash_ip); ++} ++ ++static int ++addip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ return __addip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip); ++} ++ ++static inline int ++__delip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_ipmap *map = (struct ip_set_ipmap *) set->data; ++ ++ if (ip < map->first_ip || ip > map->last_ip) ++ return -ERANGE; ++ ++ *hash_ip = ip & map->netmask; ++ DP("%u.%u.%u.%u, %u.%u.%u.%u", HIPQUAD(ip), HIPQUAD(*hash_ip)); ++ if (!test_and_clear_bit(ip_to_id(map, *hash_ip), map->members)) ++ return -EEXIST; ++ ++ return 0; ++} ++ ++static int ++delip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_ipmap *req = ++ (struct ip_set_req_ipmap *) data; ++ ++ if (size != sizeof(struct ip_set_req_ipmap)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_ipmap), ++ size); ++ return -EINVAL; ++ } ++ return __delip(set, req->ip, hash_ip); ++} ++ ++static int ++delip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ return __delip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip); ++} ++ ++static int create(struct ip_set *set, const void *data, size_t size) ++{ ++ int newbytes; ++ struct ip_set_req_ipmap_create *req = ++ (struct ip_set_req_ipmap_create *) data; ++ struct ip_set_ipmap *map; ++ ++ if (size != sizeof(struct ip_set_req_ipmap_create)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_ipmap_create), ++ size); ++ return -EINVAL; ++ } ++ ++ DP("from %u.%u.%u.%u to %u.%u.%u.%u", ++ HIPQUAD(req->from), HIPQUAD(req->to)); ++ ++ if (req->from > req->to) { ++ DP("bad ip range"); ++ return -ENOEXEC; ++ } ++ ++ map = kmalloc(sizeof(struct ip_set_ipmap), GFP_KERNEL); ++ if (!map) { ++ DP("out of memory for %d bytes", ++ sizeof(struct ip_set_ipmap)); ++ return -ENOMEM; ++ } ++ map->first_ip = req->from; ++ map->last_ip = req->to; ++ map->netmask = req->netmask; ++ ++ if (req->netmask == 0xFFFFFFFF) { ++ map->hosts = 1; ++ map->sizeid = map->last_ip - map->first_ip + 1; ++ } else { ++ unsigned int mask_bits, netmask_bits; ++ ip_set_ip_t mask; ++ ++ map->first_ip &= map->netmask; /* Should we better bark? */ ++ ++ mask = range_to_mask(map->first_ip, map->last_ip, &mask_bits); ++ netmask_bits = mask_to_bits(map->netmask); ++ ++ if ((!mask && (map->first_ip || map->last_ip != 0xFFFFFFFF)) ++ || netmask_bits <= mask_bits) ++ return -ENOEXEC; ++ ++ DP("mask_bits %u, netmask_bits %u", ++ mask_bits, netmask_bits); ++ map->hosts = 2 << (32 - netmask_bits - 1); ++ map->sizeid = 2 << (netmask_bits - mask_bits - 1); ++ } ++ if (map->sizeid > MAX_RANGE + 1) { ++ ip_set_printk("range too big (max %d addresses)", ++ MAX_RANGE+1); ++ kfree(map); ++ return -ENOEXEC; ++ } ++ DP("hosts %u, sizeid %u", map->hosts, map->sizeid); ++ newbytes = bitmap_bytes(0, map->sizeid - 1); ++ map->members = kmalloc(newbytes, GFP_KERNEL); ++ if (!map->members) { ++ DP("out of memory for %d bytes", newbytes); ++ kfree(map); ++ return -ENOMEM; ++ } ++ memset(map->members, 0, newbytes); ++ ++ set->data = map; ++ return 0; ++} ++ ++static void destroy(struct ip_set *set) ++{ ++ struct ip_set_ipmap *map = (struct ip_set_ipmap *) set->data; ++ ++ kfree(map->members); ++ kfree(map); ++ ++ set->data = NULL; ++} ++ ++static void flush(struct ip_set *set) ++{ ++ struct ip_set_ipmap *map = (struct ip_set_ipmap *) set->data; ++ memset(map->members, 0, bitmap_bytes(0, map->sizeid - 1)); ++} ++ ++static void list_header(const struct ip_set *set, void *data) ++{ ++ struct ip_set_ipmap *map = (struct ip_set_ipmap *) set->data; ++ struct ip_set_req_ipmap_create *header = ++ (struct ip_set_req_ipmap_create *) data; ++ ++ header->from = map->first_ip; ++ header->to = map->last_ip; ++ header->netmask = map->netmask; ++} ++ ++static int list_members_size(const struct ip_set *set) ++{ ++ struct ip_set_ipmap *map = (struct ip_set_ipmap *) set->data; ++ ++ return bitmap_bytes(0, map->sizeid - 1); ++} ++ ++static void list_members(const struct ip_set *set, void *data) ++{ ++ struct ip_set_ipmap *map = (struct ip_set_ipmap *) set->data; ++ int bytes = bitmap_bytes(0, map->sizeid - 1); ++ ++ memcpy(data, map->members, bytes); ++} ++ ++static struct ip_set_type ip_set_ipmap = { ++ .typename = SETTYPE_NAME, ++ .features = IPSET_TYPE_IP | IPSET_DATA_SINGLE, ++ .protocol_version = IP_SET_PROTOCOL_VERSION, ++ .create = &create, ++ .destroy = &destroy, ++ .flush = &flush, ++ .reqsize = sizeof(struct ip_set_req_ipmap), ++ .addip = &addip, ++ .addip_kernel = &addip_kernel, ++ .delip = &delip, ++ .delip_kernel = &delip_kernel, ++ .testip = &testip, ++ .testip_kernel = &testip_kernel, ++ .header_size = sizeof(struct ip_set_req_ipmap_create), ++ .list_header = &list_header, ++ .list_members_size = &list_members_size, ++ .list_members = &list_members, ++ .me = THIS_MODULE, ++}; ++ ++MODULE_LICENSE("GPL"); ++MODULE_AUTHOR("Jozsef Kadlecsik "); ++MODULE_DESCRIPTION("ipmap type of IP sets"); ++ ++static int __init ip_set_ipmap_init(void) ++{ ++ return ip_set_register_set_type(&ip_set_ipmap); ++} ++ ++static void __exit ip_set_ipmap_fini(void) ++{ ++ /* FIXME: possible race with ip_set_create() */ ++ ip_set_unregister_set_type(&ip_set_ipmap); ++} ++ ++module_init(ip_set_ipmap_init); ++module_exit(ip_set_ipmap_fini); +--- /dev/null ++++ b/net/ipv4/netfilter/ip_set_ipporthash.c +@@ -0,0 +1,581 @@ ++/* Copyright (C) 2003-2004 Jozsef Kadlecsik ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++/* Kernel module implementing an ip+port hash set */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++ ++#include ++#include ++ ++static int limit = MAX_RANGE; ++ ++/* We must handle non-linear skbs */ ++static inline ip_set_ip_t ++get_port(const struct sk_buff *skb, u_int32_t flags) ++{ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ struct iphdr *iph = ip_hdr(skb); ++#else ++ struct iphdr *iph = skb->nh.iph; ++#endif ++ u_int16_t offset = ntohs(iph->frag_off) & IP_OFFSET; ++ ++ switch (iph->protocol) { ++ case IPPROTO_TCP: { ++ struct tcphdr tcph; ++ ++ /* See comments at tcp_match in ip_tables.c */ ++ if (offset) ++ return INVALID_PORT; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ if (skb_copy_bits(skb, ip_hdr(skb)->ihl*4, &tcph, sizeof(tcph)) < 0) ++#else ++ if (skb_copy_bits(skb, skb->nh.iph->ihl*4, &tcph, sizeof(tcph)) < 0) ++#endif ++ /* No choice either */ ++ return INVALID_PORT; ++ ++ return ntohs(flags & IPSET_SRC ? ++ tcph.source : tcph.dest); ++ } ++ case IPPROTO_UDP: { ++ struct udphdr udph; ++ ++ if (offset) ++ return INVALID_PORT; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ if (skb_copy_bits(skb, ip_hdr(skb)->ihl*4, &udph, sizeof(udph)) < 0) ++#else ++ if (skb_copy_bits(skb, skb->nh.iph->ihl*4, &udph, sizeof(udph)) < 0) ++#endif ++ /* No choice either */ ++ return INVALID_PORT; ++ ++ return ntohs(flags & IPSET_SRC ? ++ udph.source : udph.dest); ++ } ++ default: ++ return INVALID_PORT; ++ } ++} ++ ++static inline __u32 ++jhash_ip(const struct ip_set_ipporthash *map, uint16_t i, ip_set_ip_t ip) ++{ ++ return jhash_1word(ip, *(((uint32_t *) map->initval) + i)); ++} ++ ++#define HASH_IP(map, ip, port) (port + ((ip - ((map)->first_ip)) << 16)) ++ ++static inline __u32 ++hash_id(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t port, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_ipporthash *map = ++ (struct ip_set_ipporthash *) set->data; ++ __u32 id; ++ u_int16_t i; ++ ip_set_ip_t *elem; ++ ++ *hash_ip = HASH_IP(map, ip, port); ++ DP("set: %s, ipport:%u.%u.%u.%u:%u, %u.%u.%u.%u", ++ set->name, HIPQUAD(ip), port, HIPQUAD(*hash_ip)); ++ ++ for (i = 0; i < map->probes; i++) { ++ id = jhash_ip(map, i, *hash_ip) % map->hashsize; ++ DP("hash key: %u", id); ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); ++ if (*elem == *hash_ip) ++ return id; ++ /* No shortcut at testing - there can be deleted ++ * entries. */ ++ } ++ return UINT_MAX; ++} ++ ++static inline int ++__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t port, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_ipporthash *map = (struct ip_set_ipporthash *) set->data; ++ ++ if (ip < map->first_ip || ip > map->last_ip) ++ return -ERANGE; ++ ++ return (hash_id(set, ip, port, hash_ip) != UINT_MAX); ++} ++ ++static int ++testip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_ipporthash *req = ++ (struct ip_set_req_ipporthash *) data; ++ ++ if (size != sizeof(struct ip_set_req_ipporthash)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_ipporthash), ++ size); ++ return -EINVAL; ++ } ++ return __testip(set, req->ip, req->port, hash_ip); ++} ++ ++static int ++testip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ ip_set_ip_t port; ++ int res; ++ ++ if (flags[index+1] == 0) ++ return 0; ++ ++ port = get_port(skb, flags[index+1]); ++ ++ DP("flag: %s src: %u.%u.%u.%u dst: %u.%u.%u.%u", ++ flags[index] & IPSET_SRC ? "SRC" : "DST", ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ NIPQUAD(ip_hdr(skb)->saddr), ++ NIPQUAD(ip_hdr(skb)->daddr)); ++#else ++ NIPQUAD(skb->nh.iph->saddr), ++ NIPQUAD(skb->nh.iph->daddr)); ++#endif ++ DP("flag %s port %u", ++ flags[index+1] & IPSET_SRC ? "SRC" : "DST", ++ port); ++ if (port == INVALID_PORT) ++ return 0; ++ ++ res = __testip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ port, ++ hash_ip); ++ return (res < 0 ? 0 : res); ++ ++} ++ ++static inline int ++__add_haship(struct ip_set_ipporthash *map, ip_set_ip_t hash_ip) ++{ ++ __u32 probe; ++ u_int16_t i; ++ ip_set_ip_t *elem; ++ ++ for (i = 0; i < map->probes; i++) { ++ probe = jhash_ip(map, i, hash_ip) % map->hashsize; ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, probe); ++ if (*elem == hash_ip) ++ return -EEXIST; ++ if (!*elem) { ++ *elem = hash_ip; ++ map->elements++; ++ return 0; ++ } ++ } ++ /* Trigger rehashing */ ++ return -EAGAIN; ++} ++ ++static inline int ++__addip(struct ip_set_ipporthash *map, ip_set_ip_t ip, ip_set_ip_t port, ++ ip_set_ip_t *hash_ip) ++{ ++ if (map->elements > limit) ++ return -ERANGE; ++ if (ip < map->first_ip || ip > map->last_ip) ++ return -ERANGE; ++ ++ *hash_ip = HASH_IP(map, ip, port); ++ ++ return __add_haship(map, *hash_ip); ++} ++ ++static int ++addip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_ipporthash *req = ++ (struct ip_set_req_ipporthash *) data; ++ ++ if (size != sizeof(struct ip_set_req_ipporthash)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_ipporthash), ++ size); ++ return -EINVAL; ++ } ++ return __addip((struct ip_set_ipporthash *) set->data, ++ req->ip, req->port, hash_ip); ++} ++ ++static int ++addip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ ip_set_ip_t port; ++ ++ if (flags[index+1] == 0) ++ return -EINVAL; ++ ++ port = get_port(skb, flags[index+1]); ++ ++ DP("flag: %s src: %u.%u.%u.%u dst: %u.%u.%u.%u", ++ flags[index] & IPSET_SRC ? "SRC" : "DST", ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ NIPQUAD(ip_hdr(skb)->saddr), ++ NIPQUAD(ip_hdr(skb)->daddr)); ++#else ++ NIPQUAD(skb->nh.iph->saddr), ++ NIPQUAD(skb->nh.iph->daddr)); ++#endif ++ DP("flag %s port %u", ++ flags[index+1] & IPSET_SRC ? "SRC" : "DST", ++ port); ++ if (port == INVALID_PORT) ++ return -EINVAL; ++ ++ return __addip((struct ip_set_ipporthash *) set->data, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ port, ++ hash_ip); ++} ++ ++static int retry(struct ip_set *set) ++{ ++ struct ip_set_ipporthash *map = (struct ip_set_ipporthash *) set->data; ++ ip_set_ip_t *elem; ++ void *members; ++ u_int32_t i, hashsize = map->hashsize; ++ int res; ++ struct ip_set_ipporthash *tmp; ++ ++ if (map->resize == 0) ++ return -ERANGE; ++ ++ again: ++ res = 0; ++ ++ /* Calculate new hash size */ ++ hashsize += (hashsize * map->resize)/100; ++ if (hashsize == map->hashsize) ++ hashsize++; ++ ++ ip_set_printk("rehashing of set %s triggered: " ++ "hashsize grows from %u to %u", ++ set->name, map->hashsize, hashsize); ++ ++ tmp = kmalloc(sizeof(struct ip_set_ipporthash) ++ + map->probes * sizeof(uint32_t), GFP_ATOMIC); ++ if (!tmp) { ++ DP("out of memory for %d bytes", ++ sizeof(struct ip_set_ipporthash) ++ + map->probes * sizeof(uint32_t)); ++ return -ENOMEM; ++ } ++ tmp->members = harray_malloc(hashsize, sizeof(ip_set_ip_t), GFP_ATOMIC); ++ if (!tmp->members) { ++ DP("out of memory for %d bytes", hashsize * sizeof(ip_set_ip_t)); ++ kfree(tmp); ++ return -ENOMEM; ++ } ++ tmp->hashsize = hashsize; ++ tmp->elements = 0; ++ tmp->probes = map->probes; ++ tmp->resize = map->resize; ++ tmp->first_ip = map->first_ip; ++ tmp->last_ip = map->last_ip; ++ memcpy(tmp->initval, map->initval, map->probes * sizeof(uint32_t)); ++ ++ write_lock_bh(&set->lock); ++ map = (struct ip_set_ipporthash *) set->data; /* Play safe */ ++ for (i = 0; i < map->hashsize && res == 0; i++) { ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i); ++ if (*elem) ++ res = __add_haship(tmp, *elem); ++ } ++ if (res) { ++ /* Failure, try again */ ++ write_unlock_bh(&set->lock); ++ harray_free(tmp->members); ++ kfree(tmp); ++ goto again; ++ } ++ ++ /* Success at resizing! */ ++ members = map->members; ++ ++ map->hashsize = tmp->hashsize; ++ map->members = tmp->members; ++ write_unlock_bh(&set->lock); ++ ++ harray_free(members); ++ kfree(tmp); ++ ++ return 0; ++} ++ ++static inline int ++__delip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t port, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_ipporthash *map = (struct ip_set_ipporthash *) set->data; ++ ip_set_ip_t id; ++ ip_set_ip_t *elem; ++ ++ if (ip < map->first_ip || ip > map->last_ip) ++ return -ERANGE; ++ ++ id = hash_id(set, ip, port, hash_ip); ++ ++ if (id == UINT_MAX) ++ return -EEXIST; ++ ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); ++ *elem = 0; ++ map->elements--; ++ ++ return 0; ++} ++ ++static int ++delip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_ipporthash *req = ++ (struct ip_set_req_ipporthash *) data; ++ ++ if (size != sizeof(struct ip_set_req_ipporthash)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_ipporthash), ++ size); ++ return -EINVAL; ++ } ++ return __delip(set, req->ip, req->port, hash_ip); ++} ++ ++static int ++delip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ ip_set_ip_t port; ++ ++ if (flags[index+1] == 0) ++ return -EINVAL; ++ ++ port = get_port(skb, flags[index+1]); ++ ++ DP("flag: %s src: %u.%u.%u.%u dst: %u.%u.%u.%u", ++ flags[index] & IPSET_SRC ? "SRC" : "DST", ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ NIPQUAD(ip_hdr(skb)->saddr), ++ NIPQUAD(ip_hdr(skb)->daddr)); ++#else ++ NIPQUAD(skb->nh.iph->saddr), ++ NIPQUAD(skb->nh.iph->daddr)); ++#endif ++ DP("flag %s port %u", ++ flags[index+1] & IPSET_SRC ? "SRC" : "DST", ++ port); ++ if (port == INVALID_PORT) ++ return -EINVAL; ++ ++ return __delip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ port, ++ hash_ip); ++} ++ ++static int create(struct ip_set *set, const void *data, size_t size) ++{ ++ struct ip_set_req_ipporthash_create *req = ++ (struct ip_set_req_ipporthash_create *) data; ++ struct ip_set_ipporthash *map; ++ uint16_t i; ++ ++ if (size != sizeof(struct ip_set_req_ipporthash_create)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_ipporthash_create), ++ size); ++ return -EINVAL; ++ } ++ ++ if (req->hashsize < 1) { ++ ip_set_printk("hashsize too small"); ++ return -ENOEXEC; ++ } ++ ++ if (req->probes < 1) { ++ ip_set_printk("probes too small"); ++ return -ENOEXEC; ++ } ++ ++ map = kmalloc(sizeof(struct ip_set_ipporthash) ++ + req->probes * sizeof(uint32_t), GFP_KERNEL); ++ if (!map) { ++ DP("out of memory for %d bytes", ++ sizeof(struct ip_set_ipporthash) ++ + req->probes * sizeof(uint32_t)); ++ return -ENOMEM; ++ } ++ for (i = 0; i < req->probes; i++) ++ get_random_bytes(((uint32_t *) map->initval)+i, 4); ++ map->elements = 0; ++ map->hashsize = req->hashsize; ++ map->probes = req->probes; ++ map->resize = req->resize; ++ map->first_ip = req->from; ++ map->last_ip = req->to; ++ map->members = harray_malloc(map->hashsize, sizeof(ip_set_ip_t), GFP_KERNEL); ++ if (!map->members) { ++ DP("out of memory for %d bytes", map->hashsize * sizeof(ip_set_ip_t)); ++ kfree(map); ++ return -ENOMEM; ++ } ++ ++ set->data = map; ++ return 0; ++} ++ ++static void destroy(struct ip_set *set) ++{ ++ struct ip_set_ipporthash *map = (struct ip_set_ipporthash *) set->data; ++ ++ harray_free(map->members); ++ kfree(map); ++ ++ set->data = NULL; ++} ++ ++static void flush(struct ip_set *set) ++{ ++ struct ip_set_ipporthash *map = (struct ip_set_ipporthash *) set->data; ++ harray_flush(map->members, map->hashsize, sizeof(ip_set_ip_t)); ++ map->elements = 0; ++} ++ ++static void list_header(const struct ip_set *set, void *data) ++{ ++ struct ip_set_ipporthash *map = (struct ip_set_ipporthash *) set->data; ++ struct ip_set_req_ipporthash_create *header = ++ (struct ip_set_req_ipporthash_create *) data; ++ ++ header->hashsize = map->hashsize; ++ header->probes = map->probes; ++ header->resize = map->resize; ++ header->from = map->first_ip; ++ header->to = map->last_ip; ++} ++ ++static int list_members_size(const struct ip_set *set) ++{ ++ struct ip_set_ipporthash *map = (struct ip_set_ipporthash *) set->data; ++ ++ return (map->hashsize * sizeof(ip_set_ip_t)); ++} ++ ++static void list_members(const struct ip_set *set, void *data) ++{ ++ struct ip_set_ipporthash *map = (struct ip_set_ipporthash *) set->data; ++ ip_set_ip_t i, *elem; ++ ++ for (i = 0; i < map->hashsize; i++) { ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i); ++ ((ip_set_ip_t *)data)[i] = *elem; ++ } ++} ++ ++static struct ip_set_type ip_set_ipporthash = { ++ .typename = SETTYPE_NAME, ++ .features = IPSET_TYPE_IP | IPSET_TYPE_PORT | IPSET_DATA_DOUBLE, ++ .protocol_version = IP_SET_PROTOCOL_VERSION, ++ .create = &create, ++ .destroy = &destroy, ++ .flush = &flush, ++ .reqsize = sizeof(struct ip_set_req_ipporthash), ++ .addip = &addip, ++ .addip_kernel = &addip_kernel, ++ .retry = &retry, ++ .delip = &delip, ++ .delip_kernel = &delip_kernel, ++ .testip = &testip, ++ .testip_kernel = &testip_kernel, ++ .header_size = sizeof(struct ip_set_req_ipporthash_create), ++ .list_header = &list_header, ++ .list_members_size = &list_members_size, ++ .list_members = &list_members, ++ .me = THIS_MODULE, ++}; ++ ++MODULE_LICENSE("GPL"); ++MODULE_AUTHOR("Jozsef Kadlecsik "); ++MODULE_DESCRIPTION("ipporthash type of IP sets"); ++module_param(limit, int, 0600); ++MODULE_PARM_DESC(limit, "maximal number of elements stored in the sets"); ++ ++static int __init ip_set_ipporthash_init(void) ++{ ++ return ip_set_register_set_type(&ip_set_ipporthash); ++} ++ ++static void __exit ip_set_ipporthash_fini(void) ++{ ++ /* FIXME: possible race with ip_set_create() */ ++ ip_set_unregister_set_type(&ip_set_ipporthash); ++} ++ ++module_init(ip_set_ipporthash_init); ++module_exit(ip_set_ipporthash_fini); +--- /dev/null ++++ b/net/ipv4/netfilter/ip_set_iptree.c +@@ -0,0 +1,612 @@ ++/* Copyright (C) 2005 Jozsef Kadlecsik ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++/* Kernel module implementing an IP set type: the iptree type */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++/* Backward compatibility */ ++#ifndef __nocast ++#define __nocast ++#endif ++ ++#include ++ ++static int limit = MAX_RANGE; ++ ++/* Garbage collection interval in seconds: */ ++#define IPTREE_GC_TIME 5*60 ++/* Sleep so many milliseconds before trying again ++ * to delete the gc timer at destroying/flushing a set */ ++#define IPTREE_DESTROY_SLEEP 100 ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,21) ++static struct kmem_cache *branch_cachep; ++static struct kmem_cache *leaf_cachep; ++#else ++static kmem_cache_t *branch_cachep; ++static kmem_cache_t *leaf_cachep; ++#endif ++ ++#if defined(__LITTLE_ENDIAN) ++#define ABCD(a,b,c,d,addrp) do { \ ++ a = ((unsigned char *)addrp)[3]; \ ++ b = ((unsigned char *)addrp)[2]; \ ++ c = ((unsigned char *)addrp)[1]; \ ++ d = ((unsigned char *)addrp)[0]; \ ++} while (0) ++#elif defined(__BIG_ENDIAN) ++#define ABCD(a,b,c,d,addrp) do { \ ++ a = ((unsigned char *)addrp)[0]; \ ++ b = ((unsigned char *)addrp)[1]; \ ++ c = ((unsigned char *)addrp)[2]; \ ++ d = ((unsigned char *)addrp)[3]; \ ++} while (0) ++#else ++#error "Please fix asm/byteorder.h" ++#endif /* __LITTLE_ENDIAN */ ++ ++#define TESTIP_WALK(map, elem, branch) do { \ ++ if ((map)->tree[elem]) { \ ++ branch = (map)->tree[elem]; \ ++ } else \ ++ return 0; \ ++} while (0) ++ ++static inline int ++__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_iptree *map = (struct ip_set_iptree *) set->data; ++ struct ip_set_iptreeb *btree; ++ struct ip_set_iptreec *ctree; ++ struct ip_set_iptreed *dtree; ++ unsigned char a,b,c,d; ++ ++ if (!ip) ++ return -ERANGE; ++ ++ *hash_ip = ip; ++ ABCD(a, b, c, d, hash_ip); ++ DP("%u %u %u %u timeout %u", a, b, c, d, map->timeout); ++ TESTIP_WALK(map, a, btree); ++ TESTIP_WALK(btree, b, ctree); ++ TESTIP_WALK(ctree, c, dtree); ++ DP("%lu %lu", dtree->expires[d], jiffies); ++ return dtree->expires[d] ++ && (!map->timeout ++ || time_after(dtree->expires[d], jiffies)); ++} ++ ++static int ++testip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_iptree *req = ++ (struct ip_set_req_iptree *) data; ++ ++ if (size != sizeof(struct ip_set_req_iptree)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_iptree), ++ size); ++ return -EINVAL; ++ } ++ return __testip(set, req->ip, hash_ip); ++} ++ ++static int ++testip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ int res; ++ ++ DP("flag: %s src: %u.%u.%u.%u dst: %u.%u.%u.%u", ++ flags[index] & IPSET_SRC ? "SRC" : "DST", ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ NIPQUAD(ip_hdr(skb)->saddr), ++ NIPQUAD(ip_hdr(skb)->daddr)); ++#else ++ NIPQUAD(skb->nh.iph->saddr), ++ NIPQUAD(skb->nh.iph->daddr)); ++#endif ++ ++ res = __testip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip); ++ return (res < 0 ? 0 : res); ++} ++ ++#define ADDIP_WALK(map, elem, branch, type, cachep) do { \ ++ if ((map)->tree[elem]) { \ ++ DP("found %u", elem); \ ++ branch = (map)->tree[elem]; \ ++ } else { \ ++ branch = (type *) \ ++ kmem_cache_alloc(cachep, GFP_ATOMIC); \ ++ if (branch == NULL) \ ++ return -ENOMEM; \ ++ memset(branch, 0, sizeof(*branch)); \ ++ (map)->tree[elem] = branch; \ ++ DP("alloc %u", elem); \ ++ } \ ++} while (0) ++ ++static inline int ++__addip(struct ip_set *set, ip_set_ip_t ip, unsigned int timeout, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_iptree *map = (struct ip_set_iptree *) set->data; ++ struct ip_set_iptreeb *btree; ++ struct ip_set_iptreec *ctree; ++ struct ip_set_iptreed *dtree; ++ unsigned char a,b,c,d; ++ int ret = 0; ++ ++ if (!ip || map->elements >= limit) ++ /* We could call the garbage collector ++ * but it's probably overkill */ ++ return -ERANGE; ++ ++ *hash_ip = ip; ++ ABCD(a, b, c, d, hash_ip); ++ DP("%u %u %u %u timeout %u", a, b, c, d, timeout); ++ ADDIP_WALK(map, a, btree, struct ip_set_iptreeb, branch_cachep); ++ ADDIP_WALK(btree, b, ctree, struct ip_set_iptreec, branch_cachep); ++ ADDIP_WALK(ctree, c, dtree, struct ip_set_iptreed, leaf_cachep); ++ if (dtree->expires[d] ++ && (!map->timeout || time_after(dtree->expires[d], jiffies))) ++ ret = -EEXIST; ++ dtree->expires[d] = map->timeout ? (timeout * HZ + jiffies) : 1; ++ /* Lottery: I won! */ ++ if (dtree->expires[d] == 0) ++ dtree->expires[d] = 1; ++ DP("%u %lu", d, dtree->expires[d]); ++ if (ret == 0) ++ map->elements++; ++ return ret; ++} ++ ++static int ++addip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_iptree *map = (struct ip_set_iptree *) set->data; ++ struct ip_set_req_iptree *req = ++ (struct ip_set_req_iptree *) data; ++ ++ if (size != sizeof(struct ip_set_req_iptree)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_iptree), ++ size); ++ return -EINVAL; ++ } ++ DP("%u.%u.%u.%u %u", HIPQUAD(req->ip), req->timeout); ++ return __addip(set, req->ip, ++ req->timeout ? req->timeout : map->timeout, ++ hash_ip); ++} ++ ++static int ++addip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ struct ip_set_iptree *map = (struct ip_set_iptree *) set->data; ++ ++ return __addip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ map->timeout, ++ hash_ip); ++} ++ ++#define DELIP_WALK(map, elem, branch) do { \ ++ if ((map)->tree[elem]) { \ ++ branch = (map)->tree[elem]; \ ++ } else \ ++ return -EEXIST; \ ++} while (0) ++ ++static inline int ++__delip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_iptree *map = (struct ip_set_iptree *) set->data; ++ struct ip_set_iptreeb *btree; ++ struct ip_set_iptreec *ctree; ++ struct ip_set_iptreed *dtree; ++ unsigned char a,b,c,d; ++ ++ if (!ip) ++ return -ERANGE; ++ ++ *hash_ip = ip; ++ ABCD(a, b, c, d, hash_ip); ++ DELIP_WALK(map, a, btree); ++ DELIP_WALK(btree, b, ctree); ++ DELIP_WALK(ctree, c, dtree); ++ ++ if (dtree->expires[d]) { ++ dtree->expires[d] = 0; ++ map->elements--; ++ return 0; ++ } ++ return -EEXIST; ++} ++ ++static int ++delip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_iptree *req = ++ (struct ip_set_req_iptree *) data; ++ ++ if (size != sizeof(struct ip_set_req_iptree)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_iptree), ++ size); ++ return -EINVAL; ++ } ++ return __delip(set, req->ip, hash_ip); ++} ++ ++static int ++delip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ return __delip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip); ++} ++ ++#define LOOP_WALK_BEGIN(map, i, branch) \ ++ for (i = 0; i < 256; i++) { \ ++ if (!(map)->tree[i]) \ ++ continue; \ ++ branch = (map)->tree[i] ++ ++#define LOOP_WALK_END } ++ ++static void ip_tree_gc(unsigned long ul_set) ++{ ++ struct ip_set *set = (void *) ul_set; ++ struct ip_set_iptree *map = (struct ip_set_iptree *) set->data; ++ struct ip_set_iptreeb *btree; ++ struct ip_set_iptreec *ctree; ++ struct ip_set_iptreed *dtree; ++ unsigned int a,b,c,d; ++ unsigned char i,j,k; ++ ++ i = j = k = 0; ++ DP("gc: %s", set->name); ++ write_lock_bh(&set->lock); ++ LOOP_WALK_BEGIN(map, a, btree); ++ LOOP_WALK_BEGIN(btree, b, ctree); ++ LOOP_WALK_BEGIN(ctree, c, dtree); ++ for (d = 0; d < 256; d++) { ++ if (dtree->expires[d]) { ++ DP("gc: %u %u %u %u: expires %lu jiffies %lu", ++ a, b, c, d, ++ dtree->expires[d], jiffies); ++ if (map->timeout ++ && time_before(dtree->expires[d], jiffies)) { ++ dtree->expires[d] = 0; ++ map->elements--; ++ } else ++ k = 1; ++ } ++ } ++ if (k == 0) { ++ DP("gc: %s: leaf %u %u %u empty", ++ set->name, a, b, c); ++ kmem_cache_free(leaf_cachep, dtree); ++ ctree->tree[c] = NULL; ++ } else { ++ DP("gc: %s: leaf %u %u %u not empty", ++ set->name, a, b, c); ++ j = 1; ++ k = 0; ++ } ++ LOOP_WALK_END; ++ if (j == 0) { ++ DP("gc: %s: branch %u %u empty", ++ set->name, a, b); ++ kmem_cache_free(branch_cachep, ctree); ++ btree->tree[b] = NULL; ++ } else { ++ DP("gc: %s: branch %u %u not empty", ++ set->name, a, b); ++ i = 1; ++ j = k = 0; ++ } ++ LOOP_WALK_END; ++ if (i == 0) { ++ DP("gc: %s: branch %u empty", ++ set->name, a); ++ kmem_cache_free(branch_cachep, btree); ++ map->tree[a] = NULL; ++ } else { ++ DP("gc: %s: branch %u not empty", ++ set->name, a); ++ i = j = k = 0; ++ } ++ LOOP_WALK_END; ++ write_unlock_bh(&set->lock); ++ ++ map->gc.expires = jiffies + map->gc_interval * HZ; ++ add_timer(&map->gc); ++} ++ ++static inline void init_gc_timer(struct ip_set *set) ++{ ++ struct ip_set_iptree *map = (struct ip_set_iptree *) set->data; ++ ++ /* Even if there is no timeout for the entries, ++ * we still have to call gc because delete ++ * do not clean up empty branches */ ++ map->gc_interval = IPTREE_GC_TIME; ++ init_timer(&map->gc); ++ map->gc.data = (unsigned long) set; ++ map->gc.function = ip_tree_gc; ++ map->gc.expires = jiffies + map->gc_interval * HZ; ++ add_timer(&map->gc); ++} ++ ++static int create(struct ip_set *set, const void *data, size_t size) ++{ ++ struct ip_set_req_iptree_create *req = ++ (struct ip_set_req_iptree_create *) data; ++ struct ip_set_iptree *map; ++ ++ if (size != sizeof(struct ip_set_req_iptree_create)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_iptree_create), ++ size); ++ return -EINVAL; ++ } ++ ++ map = kmalloc(sizeof(struct ip_set_iptree), GFP_KERNEL); ++ if (!map) { ++ DP("out of memory for %d bytes", ++ sizeof(struct ip_set_iptree)); ++ return -ENOMEM; ++ } ++ memset(map, 0, sizeof(*map)); ++ map->timeout = req->timeout; ++ map->elements = 0; ++ set->data = map; ++ ++ init_gc_timer(set); ++ ++ return 0; ++} ++ ++static void __flush(struct ip_set_iptree *map) ++{ ++ struct ip_set_iptreeb *btree; ++ struct ip_set_iptreec *ctree; ++ struct ip_set_iptreed *dtree; ++ unsigned int a,b,c; ++ ++ LOOP_WALK_BEGIN(map, a, btree); ++ LOOP_WALK_BEGIN(btree, b, ctree); ++ LOOP_WALK_BEGIN(ctree, c, dtree); ++ kmem_cache_free(leaf_cachep, dtree); ++ LOOP_WALK_END; ++ kmem_cache_free(branch_cachep, ctree); ++ LOOP_WALK_END; ++ kmem_cache_free(branch_cachep, btree); ++ LOOP_WALK_END; ++ map->elements = 0; ++} ++ ++static void destroy(struct ip_set *set) ++{ ++ struct ip_set_iptree *map = (struct ip_set_iptree *) set->data; ++ ++ /* gc might be running */ ++ while (!del_timer(&map->gc)) ++ msleep(IPTREE_DESTROY_SLEEP); ++ __flush(map); ++ kfree(map); ++ set->data = NULL; ++} ++ ++static void flush(struct ip_set *set) ++{ ++ struct ip_set_iptree *map = (struct ip_set_iptree *) set->data; ++ unsigned int timeout = map->timeout; ++ ++ /* gc might be running */ ++ while (!del_timer(&map->gc)) ++ msleep(IPTREE_DESTROY_SLEEP); ++ __flush(map); ++ memset(map, 0, sizeof(*map)); ++ map->timeout = timeout; ++ ++ init_gc_timer(set); ++} ++ ++static void list_header(const struct ip_set *set, void *data) ++{ ++ struct ip_set_iptree *map = (struct ip_set_iptree *) set->data; ++ struct ip_set_req_iptree_create *header = ++ (struct ip_set_req_iptree_create *) data; ++ ++ header->timeout = map->timeout; ++} ++ ++static int list_members_size(const struct ip_set *set) ++{ ++ struct ip_set_iptree *map = (struct ip_set_iptree *) set->data; ++ struct ip_set_iptreeb *btree; ++ struct ip_set_iptreec *ctree; ++ struct ip_set_iptreed *dtree; ++ unsigned int a,b,c,d; ++ unsigned int count = 0; ++ ++ LOOP_WALK_BEGIN(map, a, btree); ++ LOOP_WALK_BEGIN(btree, b, ctree); ++ LOOP_WALK_BEGIN(ctree, c, dtree); ++ for (d = 0; d < 256; d++) { ++ if (dtree->expires[d] ++ && (!map->timeout || time_after(dtree->expires[d], jiffies))) ++ count++; ++ } ++ LOOP_WALK_END; ++ LOOP_WALK_END; ++ LOOP_WALK_END; ++ ++ DP("members %u", count); ++ return (count * sizeof(struct ip_set_req_iptree)); ++} ++ ++static void list_members(const struct ip_set *set, void *data) ++{ ++ struct ip_set_iptree *map = (struct ip_set_iptree *) set->data; ++ struct ip_set_iptreeb *btree; ++ struct ip_set_iptreec *ctree; ++ struct ip_set_iptreed *dtree; ++ unsigned int a,b,c,d; ++ size_t offset = 0; ++ struct ip_set_req_iptree *entry; ++ ++ LOOP_WALK_BEGIN(map, a, btree); ++ LOOP_WALK_BEGIN(btree, b, ctree); ++ LOOP_WALK_BEGIN(ctree, c, dtree); ++ for (d = 0; d < 256; d++) { ++ if (dtree->expires[d] ++ && (!map->timeout || time_after(dtree->expires[d], jiffies))) { ++ entry = (struct ip_set_req_iptree *)(data + offset); ++ entry->ip = ((a << 24) | (b << 16) | (c << 8) | d); ++ entry->timeout = !map->timeout ? 0 ++ : (dtree->expires[d] - jiffies)/HZ; ++ offset += sizeof(struct ip_set_req_iptree); ++ } ++ } ++ LOOP_WALK_END; ++ LOOP_WALK_END; ++ LOOP_WALK_END; ++} ++ ++static struct ip_set_type ip_set_iptree = { ++ .typename = SETTYPE_NAME, ++ .features = IPSET_TYPE_IP | IPSET_DATA_SINGLE, ++ .protocol_version = IP_SET_PROTOCOL_VERSION, ++ .create = &create, ++ .destroy = &destroy, ++ .flush = &flush, ++ .reqsize = sizeof(struct ip_set_req_iptree), ++ .addip = &addip, ++ .addip_kernel = &addip_kernel, ++ .delip = &delip, ++ .delip_kernel = &delip_kernel, ++ .testip = &testip, ++ .testip_kernel = &testip_kernel, ++ .header_size = sizeof(struct ip_set_req_iptree_create), ++ .list_header = &list_header, ++ .list_members_size = &list_members_size, ++ .list_members = &list_members, ++ .me = THIS_MODULE, ++}; ++ ++MODULE_LICENSE("GPL"); ++MODULE_AUTHOR("Jozsef Kadlecsik "); ++MODULE_DESCRIPTION("iptree type of IP sets"); ++module_param(limit, int, 0600); ++MODULE_PARM_DESC(limit, "maximal number of elements stored in the sets"); ++ ++static int __init ip_set_iptree_init(void) ++{ ++ int ret; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) ++ branch_cachep = kmem_cache_create("ip_set_iptreeb", ++ sizeof(struct ip_set_iptreeb), ++ 0, 0, NULL); ++#else ++ branch_cachep = kmem_cache_create("ip_set_iptreeb", ++ sizeof(struct ip_set_iptreeb), ++ 0, 0, NULL, NULL); ++#endif ++ if (!branch_cachep) { ++ printk(KERN_ERR "Unable to create ip_set_iptreeb slab cache\n"); ++ ret = -ENOMEM; ++ goto out; ++ } ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) ++ leaf_cachep = kmem_cache_create("ip_set_iptreed", ++ sizeof(struct ip_set_iptreed), ++ 0, 0, NULL); ++#else ++ leaf_cachep = kmem_cache_create("ip_set_iptreed", ++ sizeof(struct ip_set_iptreed), ++ 0, 0, NULL, NULL); ++#endif ++ if (!leaf_cachep) { ++ printk(KERN_ERR "Unable to create ip_set_iptreed slab cache\n"); ++ ret = -ENOMEM; ++ goto free_branch; ++ } ++ ret = ip_set_register_set_type(&ip_set_iptree); ++ if (ret == 0) ++ goto out; ++ ++ kmem_cache_destroy(leaf_cachep); ++ free_branch: ++ kmem_cache_destroy(branch_cachep); ++ out: ++ return ret; ++} ++ ++static void __exit ip_set_iptree_fini(void) ++{ ++ /* FIXME: possible race with ip_set_create() */ ++ ip_set_unregister_set_type(&ip_set_iptree); ++ kmem_cache_destroy(leaf_cachep); ++ kmem_cache_destroy(branch_cachep); ++} ++ ++module_init(ip_set_iptree_init); ++module_exit(ip_set_iptree_fini); +--- /dev/null ++++ b/net/ipv4/netfilter/ip_set_iptreemap.c +@@ -0,0 +1,829 @@ ++/* Copyright (C) 2007 Sven Wegener ++ * ++ * This program is free software; you can redistribute it and/or modify it ++ * under the terms of the GNU General Public License version 2 as published by ++ * the Free Software Foundation. ++ */ ++ ++/* This modules implements the iptreemap ipset type. It uses bitmaps to ++ * represent every single IPv4 address as a single bit. The bitmaps are managed ++ * in a tree structure, where the first three octets of an addresses are used ++ * as an index to find the bitmap and the last octet is used as the bit number. ++ */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++ ++#define IPTREEMAP_DEFAULT_GC_TIME (5 * 60) ++#define IPTREEMAP_DESTROY_SLEEP (100) ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,21) ++static struct kmem_cache *cachep_b; ++static struct kmem_cache *cachep_c; ++static struct kmem_cache *cachep_d; ++#else ++static kmem_cache_t *cachep_b; ++static kmem_cache_t *cachep_c; ++static kmem_cache_t *cachep_d; ++#endif ++ ++static struct ip_set_iptreemap_d *fullbitmap_d; ++static struct ip_set_iptreemap_c *fullbitmap_c; ++static struct ip_set_iptreemap_b *fullbitmap_b; ++ ++#if defined(__LITTLE_ENDIAN) ++#define ABCD(a, b, c, d, addr) \ ++ do { \ ++ a = ((unsigned char *)addr)[3]; \ ++ b = ((unsigned char *)addr)[2]; \ ++ c = ((unsigned char *)addr)[1]; \ ++ d = ((unsigned char *)addr)[0]; \ ++ } while (0) ++#elif defined(__BIG_ENDIAN) ++#define ABCD(a,b,c,d,addrp) do { \ ++ a = ((unsigned char *)addrp)[0]; \ ++ b = ((unsigned char *)addrp)[1]; \ ++ c = ((unsigned char *)addrp)[2]; \ ++ d = ((unsigned char *)addrp)[3]; \ ++} while (0) ++#else ++#error "Please fix asm/byteorder.h" ++#endif /* __LITTLE_ENDIAN */ ++ ++#define TESTIP_WALK(map, elem, branch, full) \ ++ do { \ ++ branch = (map)->tree[elem]; \ ++ if (!branch) \ ++ return 0; \ ++ else if (branch == full) \ ++ return 1; \ ++ } while (0) ++ ++#define ADDIP_WALK(map, elem, branch, type, cachep, full) \ ++ do { \ ++ branch = (map)->tree[elem]; \ ++ if (!branch) { \ ++ branch = (type *) kmem_cache_alloc(cachep, GFP_ATOMIC); \ ++ if (!branch) \ ++ return -ENOMEM; \ ++ memset(branch, 0, sizeof(*branch)); \ ++ (map)->tree[elem] = branch; \ ++ } else if (branch == full) { \ ++ return -EEXIST; \ ++ } \ ++ } while (0) ++ ++#define ADDIP_RANGE_LOOP(map, a, a1, a2, hint, branch, full, cachep, free) \ ++ for (a = a1; a <= a2; a++) { \ ++ branch = (map)->tree[a]; \ ++ if (branch != full) { \ ++ if ((a > a1 && a < a2) || (hint)) { \ ++ if (branch) \ ++ free(branch); \ ++ (map)->tree[a] = full; \ ++ continue; \ ++ } else if (!branch) { \ ++ branch = kmem_cache_alloc(cachep, GFP_ATOMIC); \ ++ if (!branch) \ ++ return -ENOMEM; \ ++ memset(branch, 0, sizeof(*branch)); \ ++ (map)->tree[a] = branch; \ ++ } ++ ++#define ADDIP_RANGE_LOOP_END() \ ++ } \ ++ } ++ ++#define DELIP_WALK(map, elem, branch, cachep, full, flags) \ ++ do { \ ++ branch = (map)->tree[elem]; \ ++ if (!branch) { \ ++ return -EEXIST; \ ++ } else if (branch == full) { \ ++ branch = kmem_cache_alloc(cachep, flags); \ ++ if (!branch) \ ++ return -ENOMEM; \ ++ memcpy(branch, full, sizeof(*full)); \ ++ (map)->tree[elem] = branch; \ ++ } \ ++ } while (0) ++ ++#define DELIP_RANGE_LOOP(map, a, a1, a2, hint, branch, full, cachep, free, flags) \ ++ for (a = a1; a <= a2; a++) { \ ++ branch = (map)->tree[a]; \ ++ if (branch) { \ ++ if ((a > a1 && a < a2) || (hint)) { \ ++ if (branch != full) \ ++ free(branch); \ ++ (map)->tree[a] = NULL; \ ++ continue; \ ++ } else if (branch == full) { \ ++ branch = kmem_cache_alloc(cachep, flags); \ ++ if (!branch) \ ++ return -ENOMEM; \ ++ memcpy(branch, full, sizeof(*branch)); \ ++ (map)->tree[a] = branch; \ ++ } ++ ++#define DELIP_RANGE_LOOP_END() \ ++ } \ ++ } ++ ++#define LOOP_WALK_BEGIN(map, i, branch) \ ++ for (i = 0; i < 256; i++) { \ ++ branch = (map)->tree[i]; \ ++ if (likely(!branch)) \ ++ continue; ++ ++#define LOOP_WALK_END() \ ++ } ++ ++#define LOOP_WALK_BEGIN_GC(map, i, branch, full, cachep, count) \ ++ count = -256; \ ++ for (i = 0; i < 256; i++) { \ ++ branch = (map)->tree[i]; \ ++ if (likely(!branch)) \ ++ continue; \ ++ count++; \ ++ if (branch == full) { \ ++ count++; \ ++ continue; \ ++ } ++ ++#define LOOP_WALK_END_GC(map, i, branch, full, cachep, count) \ ++ if (-256 == count) { \ ++ kmem_cache_free(cachep, branch); \ ++ (map)->tree[i] = NULL; \ ++ } else if (256 == count) { \ ++ kmem_cache_free(cachep, branch); \ ++ (map)->tree[i] = full; \ ++ } \ ++ } ++ ++#define LOOP_WALK_BEGIN_COUNT(map, i, branch, inrange, count) \ ++ for (i = 0; i < 256; i++) { \ ++ if (!(map)->tree[i]) { \ ++ if (inrange) { \ ++ count++; \ ++ inrange = 0; \ ++ } \ ++ continue; \ ++ } \ ++ branch = (map)->tree[i]; ++ ++#define LOOP_WALK_END_COUNT() \ ++ } ++ ++#define MIN(a, b) (a < b ? a : b) ++#define MAX(a, b) (a > b ? a : b) ++ ++#define GETVALUE1(a, a1, b1, r) \ ++ (a == a1 ? b1 : r) ++ ++#define GETVALUE2(a, b, a1, b1, c1, r) \ ++ (a == a1 && b == b1 ? c1 : r) ++ ++#define GETVALUE3(a, b, c, a1, b1, c1, d1, r) \ ++ (a == a1 && b == b1 && c == c1 ? d1 : r) ++ ++#define CHECK1(a, a1, a2, b1, b2, c1, c2, d1, d2) \ ++ ( \ ++ GETVALUE1(a, a1, b1, 0) == 0 \ ++ && GETVALUE1(a, a2, b2, 255) == 255 \ ++ && c1 == 0 \ ++ && c2 == 255 \ ++ && d1 == 0 \ ++ && d2 == 255 \ ++ ) ++ ++#define CHECK2(a, b, a1, a2, b1, b2, c1, c2, d1, d2) \ ++ ( \ ++ GETVALUE2(a, b, a1, b1, c1, 0) == 0 \ ++ && GETVALUE2(a, b, a2, b2, c2, 255) == 255 \ ++ && d1 == 0 \ ++ && d2 == 255 \ ++ ) ++ ++#define CHECK3(a, b, c, a1, a2, b1, b2, c1, c2, d1, d2) \ ++ ( \ ++ GETVALUE3(a, b, c, a1, b1, c1, d1, 0) == 0 \ ++ && GETVALUE3(a, b, c, a2, b2, c2, d2, 255) == 255 \ ++ ) ++ ++ ++static inline void ++free_d(struct ip_set_iptreemap_d *map) ++{ ++ kmem_cache_free(cachep_d, map); ++} ++ ++static inline void ++free_c(struct ip_set_iptreemap_c *map) ++{ ++ struct ip_set_iptreemap_d *dtree; ++ unsigned int i; ++ ++ LOOP_WALK_BEGIN(map, i, dtree) { ++ if (dtree != fullbitmap_d) ++ free_d(dtree); ++ } LOOP_WALK_END(); ++ ++ kmem_cache_free(cachep_c, map); ++} ++ ++static inline void ++free_b(struct ip_set_iptreemap_b *map) ++{ ++ struct ip_set_iptreemap_c *ctree; ++ unsigned int i; ++ ++ LOOP_WALK_BEGIN(map, i, ctree) { ++ if (ctree != fullbitmap_c) ++ free_c(ctree); ++ } LOOP_WALK_END(); ++ ++ kmem_cache_free(cachep_b, map); ++} ++ ++static inline int ++__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; ++ struct ip_set_iptreemap_b *btree; ++ struct ip_set_iptreemap_c *ctree; ++ struct ip_set_iptreemap_d *dtree; ++ unsigned char a, b, c, d; ++ ++ *hash_ip = ip; ++ ++ ABCD(a, b, c, d, hash_ip); ++ ++ TESTIP_WALK(map, a, btree, fullbitmap_b); ++ TESTIP_WALK(btree, b, ctree, fullbitmap_c); ++ TESTIP_WALK(ctree, c, dtree, fullbitmap_d); ++ ++ return !!test_bit(d, (void *) dtree->bitmap); ++} ++ ++static int ++testip(struct ip_set *set, const void *data, size_t size, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_iptreemap *req = (struct ip_set_req_iptreemap *) data; ++ ++ if (size != sizeof(struct ip_set_req_iptreemap)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", sizeof(struct ip_set_req_iptreemap), size); ++ return -EINVAL; ++ } ++ ++ return __testip(set, req->start, hash_ip); ++} ++ ++static int ++testip_kernel(struct ip_set *set, const struct sk_buff *skb, ip_set_ip_t *hash_ip, const u_int32_t *flags, unsigned char index) ++{ ++ int res; ++ ++ res = __testip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip); ++ ++ return (res < 0 ? 0 : res); ++} ++ ++static inline int ++__addip_single(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; ++ struct ip_set_iptreemap_b *btree; ++ struct ip_set_iptreemap_c *ctree; ++ struct ip_set_iptreemap_d *dtree; ++ unsigned char a, b, c, d; ++ ++ *hash_ip = ip; ++ ++ ABCD(a, b, c, d, hash_ip); ++ ++ ADDIP_WALK(map, a, btree, struct ip_set_iptreemap_b, cachep_b, fullbitmap_b); ++ ADDIP_WALK(btree, b, ctree, struct ip_set_iptreemap_c, cachep_c, fullbitmap_c); ++ ADDIP_WALK(ctree, c, dtree, struct ip_set_iptreemap_d, cachep_d, fullbitmap_d); ++ ++ if (test_and_set_bit(d, (void *) dtree->bitmap)) ++ return -EEXIST; ++ ++ set_bit(b, (void *) btree->dirty); ++ ++ return 0; ++} ++ ++static inline int ++__addip_range(struct ip_set *set, ip_set_ip_t start, ip_set_ip_t end, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; ++ struct ip_set_iptreemap_b *btree; ++ struct ip_set_iptreemap_c *ctree; ++ struct ip_set_iptreemap_d *dtree; ++ unsigned int a, b, c, d; ++ unsigned char a1, b1, c1, d1; ++ unsigned char a2, b2, c2, d2; ++ ++ if (start == end) ++ return __addip_single(set, start, hash_ip); ++ ++ *hash_ip = start; ++ ++ ABCD(a1, b1, c1, d1, &start); ++ ABCD(a2, b2, c2, d2, &end); ++ ++ /* This is sooo ugly... */ ++ ADDIP_RANGE_LOOP(map, a, a1, a2, CHECK1(a, a1, a2, b1, b2, c1, c2, d1, d2), btree, fullbitmap_b, cachep_b, free_b) { ++ ADDIP_RANGE_LOOP(btree, b, GETVALUE1(a, a1, b1, 0), GETVALUE1(a, a2, b2, 255), CHECK2(a, b, a1, a2, b1, b2, c1, c2, d1, d2), ctree, fullbitmap_c, cachep_c, free_c) { ++ ADDIP_RANGE_LOOP(ctree, c, GETVALUE2(a, b, a1, b1, c1, 0), GETVALUE2(a, b, a2, b2, c2, 255), CHECK3(a, b, c, a1, a2, b1, b2, c1, c2, d1, d2), dtree, fullbitmap_d, cachep_d, free_d) { ++ for (d = GETVALUE3(a, b, c, a1, b1, c1, d1, 0); d <= GETVALUE3(a, b, c, a2, b2, c2, d2, 255); d++) ++ set_bit(d, (void *) dtree->bitmap); ++ set_bit(b, (void *) btree->dirty); ++ } ADDIP_RANGE_LOOP_END(); ++ } ADDIP_RANGE_LOOP_END(); ++ } ADDIP_RANGE_LOOP_END(); ++ ++ return 0; ++} ++ ++static int ++addip(struct ip_set *set, const void *data, size_t size, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_iptreemap *req = (struct ip_set_req_iptreemap *) data; ++ ++ if (size != sizeof(struct ip_set_req_iptreemap)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", sizeof(struct ip_set_req_iptreemap), size); ++ return -EINVAL; ++ } ++ ++ return __addip_range(set, MIN(req->start, req->end), MAX(req->start, req->end), hash_ip); ++} ++ ++static int ++addip_kernel(struct ip_set *set, const struct sk_buff *skb, ip_set_ip_t *hash_ip, const u_int32_t *flags, unsigned char index) ++{ ++ ++ return __addip_single(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip); ++} ++ ++static inline int ++__delip_single(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip, unsigned int __nocast flags) ++{ ++ struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; ++ struct ip_set_iptreemap_b *btree; ++ struct ip_set_iptreemap_c *ctree; ++ struct ip_set_iptreemap_d *dtree; ++ unsigned char a,b,c,d; ++ ++ *hash_ip = ip; ++ ++ ABCD(a, b, c, d, hash_ip); ++ ++ DELIP_WALK(map, a, btree, cachep_b, fullbitmap_b, flags); ++ DELIP_WALK(btree, b, ctree, cachep_c, fullbitmap_c, flags); ++ DELIP_WALK(ctree, c, dtree, cachep_d, fullbitmap_d, flags); ++ ++ if (!test_and_clear_bit(d, (void *) dtree->bitmap)) ++ return -EEXIST; ++ ++ set_bit(b, (void *) btree->dirty); ++ ++ return 0; ++} ++ ++static inline int ++__delip_range(struct ip_set *set, ip_set_ip_t start, ip_set_ip_t end, ip_set_ip_t *hash_ip, unsigned int __nocast flags) ++{ ++ struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; ++ struct ip_set_iptreemap_b *btree; ++ struct ip_set_iptreemap_c *ctree; ++ struct ip_set_iptreemap_d *dtree; ++ unsigned int a, b, c, d; ++ unsigned char a1, b1, c1, d1; ++ unsigned char a2, b2, c2, d2; ++ ++ if (start == end) ++ return __delip_single(set, start, hash_ip, flags); ++ ++ *hash_ip = start; ++ ++ ABCD(a1, b1, c1, d1, &start); ++ ABCD(a2, b2, c2, d2, &end); ++ ++ /* This is sooo ugly... */ ++ DELIP_RANGE_LOOP(map, a, a1, a2, CHECK1(a, a1, a2, b1, b2, c1, c2, d1, d2), btree, fullbitmap_b, cachep_b, free_b, flags) { ++ DELIP_RANGE_LOOP(btree, b, GETVALUE1(a, a1, b1, 0), GETVALUE1(a, a2, b2, 255), CHECK2(a, b, a1, a2, b1, b2, c1, c2, d1, d2), ctree, fullbitmap_c, cachep_c, free_c, flags) { ++ DELIP_RANGE_LOOP(ctree, c, GETVALUE2(a, b, a1, b1, c1, 0), GETVALUE2(a, b, a2, b2, c2, 255), CHECK3(a, b, c, a1, a2, b1, b2, c1, c2, d1, d2), dtree, fullbitmap_d, cachep_d, free_d, flags) { ++ for (d = GETVALUE3(a, b, c, a1, b1, c1, d1, 0); d <= GETVALUE3(a, b, c, a2, b2, c2, d2, 255); d++) ++ clear_bit(d, (void *) dtree->bitmap); ++ set_bit(b, (void *) btree->dirty); ++ } DELIP_RANGE_LOOP_END(); ++ } DELIP_RANGE_LOOP_END(); ++ } DELIP_RANGE_LOOP_END(); ++ ++ return 0; ++} ++ ++static int ++delip(struct ip_set *set, const void *data, size_t size, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_iptreemap *req = (struct ip_set_req_iptreemap *) data; ++ ++ if (size != sizeof(struct ip_set_req_iptreemap)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", sizeof(struct ip_set_req_iptreemap), size); ++ return -EINVAL; ++ } ++ ++ return __delip_range(set, MIN(req->start, req->end), MAX(req->start, req->end), hash_ip, GFP_KERNEL); ++} ++ ++static int ++delip_kernel(struct ip_set *set, const struct sk_buff *skb, ip_set_ip_t *hash_ip, const u_int32_t *flags, unsigned char index) ++{ ++ return __delip_single(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip, ++ GFP_ATOMIC); ++} ++ ++/* Check the status of the bitmap ++ * -1 == all bits cleared ++ * 1 == all bits set ++ * 0 == anything else ++ */ ++static inline int ++bitmap_status(struct ip_set_iptreemap_d *dtree) ++{ ++ unsigned char first = dtree->bitmap[0]; ++ int a; ++ ++ for (a = 1; a < 32; a++) ++ if (dtree->bitmap[a] != first) ++ return 0; ++ ++ return (first == 0 ? -1 : (first == 255 ? 1 : 0)); ++} ++ ++static void ++gc(unsigned long addr) ++{ ++ struct ip_set *set = (struct ip_set *) addr; ++ struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; ++ struct ip_set_iptreemap_b *btree; ++ struct ip_set_iptreemap_c *ctree; ++ struct ip_set_iptreemap_d *dtree; ++ unsigned int a, b, c; ++ int i, j, k; ++ ++ write_lock_bh(&set->lock); ++ ++ LOOP_WALK_BEGIN_GC(map, a, btree, fullbitmap_b, cachep_b, i) { ++ LOOP_WALK_BEGIN_GC(btree, b, ctree, fullbitmap_c, cachep_c, j) { ++ if (!test_and_clear_bit(b, (void *) btree->dirty)) ++ continue; ++ LOOP_WALK_BEGIN_GC(ctree, c, dtree, fullbitmap_d, cachep_d, k) { ++ switch (bitmap_status(dtree)) { ++ case -1: ++ kmem_cache_free(cachep_d, dtree); ++ ctree->tree[c] = NULL; ++ k--; ++ break; ++ case 1: ++ kmem_cache_free(cachep_d, dtree); ++ ctree->tree[c] = fullbitmap_d; ++ k++; ++ break; ++ } ++ } LOOP_WALK_END(); ++ } LOOP_WALK_END_GC(btree, b, ctree, fullbitmap_c, cachep_c, k); ++ } LOOP_WALK_END_GC(map, a, btree, fullbitmap_b, cachep_b, j); ++ ++ write_unlock_bh(&set->lock); ++ ++ map->gc.expires = jiffies + map->gc_interval * HZ; ++ add_timer(&map->gc); ++} ++ ++static inline void ++init_gc_timer(struct ip_set *set) ++{ ++ struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; ++ ++ init_timer(&map->gc); ++ map->gc.data = (unsigned long) set; ++ map->gc.function = gc; ++ map->gc.expires = jiffies + map->gc_interval * HZ; ++ add_timer(&map->gc); ++} ++ ++static int create(struct ip_set *set, const void *data, size_t size) ++{ ++ struct ip_set_req_iptreemap_create *req = (struct ip_set_req_iptreemap_create *) data; ++ struct ip_set_iptreemap *map; ++ ++ if (size != sizeof(struct ip_set_req_iptreemap_create)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", sizeof(struct ip_set_req_iptreemap_create), size); ++ return -EINVAL; ++ } ++ ++ map = kzalloc(sizeof(*map), GFP_KERNEL); ++ if (!map) ++ return -ENOMEM; ++ ++ map->gc_interval = req->gc_interval ? req->gc_interval : IPTREEMAP_DEFAULT_GC_TIME; ++ set->data = map; ++ ++ init_gc_timer(set); ++ ++ return 0; ++} ++ ++static inline void __flush(struct ip_set_iptreemap *map) ++{ ++ struct ip_set_iptreemap_b *btree; ++ unsigned int a; ++ ++ LOOP_WALK_BEGIN(map, a, btree); ++ if (btree != fullbitmap_b) ++ free_b(btree); ++ LOOP_WALK_END(); ++} ++ ++static void destroy(struct ip_set *set) ++{ ++ struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; ++ ++ while (!del_timer(&map->gc)) ++ msleep(IPTREEMAP_DESTROY_SLEEP); ++ ++ __flush(map); ++ kfree(map); ++ ++ set->data = NULL; ++} ++ ++static void flush(struct ip_set *set) ++{ ++ struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; ++ ++ while (!del_timer(&map->gc)) ++ msleep(IPTREEMAP_DESTROY_SLEEP); ++ ++ __flush(map); ++ ++ memset(map, 0, sizeof(*map)); ++ ++ init_gc_timer(set); ++} ++ ++static void list_header(const struct ip_set *set, void *data) ++{ ++ struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; ++ struct ip_set_req_iptreemap_create *header = (struct ip_set_req_iptreemap_create *) data; ++ ++ header->gc_interval = map->gc_interval; ++} ++ ++static int list_members_size(const struct ip_set *set) ++{ ++ struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; ++ struct ip_set_iptreemap_b *btree; ++ struct ip_set_iptreemap_c *ctree; ++ struct ip_set_iptreemap_d *dtree; ++ unsigned int a, b, c, d, inrange = 0, count = 0; ++ ++ LOOP_WALK_BEGIN_COUNT(map, a, btree, inrange, count) { ++ LOOP_WALK_BEGIN_COUNT(btree, b, ctree, inrange, count) { ++ LOOP_WALK_BEGIN_COUNT(ctree, c, dtree, inrange, count) { ++ for (d = 0; d < 256; d++) { ++ if (test_bit(d, (void *) dtree->bitmap)) { ++ inrange = 1; ++ } else if (inrange) { ++ count++; ++ inrange = 0; ++ } ++ } ++ } LOOP_WALK_END_COUNT(); ++ } LOOP_WALK_END_COUNT(); ++ } LOOP_WALK_END_COUNT(); ++ ++ if (inrange) ++ count++; ++ ++ return (count * sizeof(struct ip_set_req_iptreemap)); ++} ++ ++static inline size_t add_member(void *data, size_t offset, ip_set_ip_t start, ip_set_ip_t end) ++{ ++ struct ip_set_req_iptreemap *entry = (struct ip_set_req_iptreemap *) (data + offset); ++ ++ entry->start = start; ++ entry->end = end; ++ ++ return sizeof(*entry); ++} ++ ++static void list_members(const struct ip_set *set, void *data) ++{ ++ struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; ++ struct ip_set_iptreemap_b *btree; ++ struct ip_set_iptreemap_c *ctree; ++ struct ip_set_iptreemap_d *dtree; ++ unsigned int a, b, c, d, inrange = 0; ++ size_t offset = 0; ++ ip_set_ip_t start = 0, end = 0, ip; ++ ++ LOOP_WALK_BEGIN(map, a, btree) { ++ LOOP_WALK_BEGIN(btree, b, ctree) { ++ LOOP_WALK_BEGIN(ctree, c, dtree) { ++ for (d = 0; d < 256; d++) { ++ if (test_bit(d, (void *) dtree->bitmap)) { ++ ip = ((a << 24) | (b << 16) | (c << 8) | d); ++ if (!inrange) { ++ inrange = 1; ++ start = ip; ++ } else if (end < ip - 1) { ++ offset += add_member(data, offset, start, end); ++ start = ip; ++ } ++ end = ip; ++ } else if (inrange) { ++ offset += add_member(data, offset, start, end); ++ inrange = 0; ++ } ++ } ++ } LOOP_WALK_END(); ++ } LOOP_WALK_END(); ++ } LOOP_WALK_END(); ++ ++ if (inrange) ++ add_member(data, offset, start, end); ++} ++ ++static struct ip_set_type ip_set_iptreemap = { ++ .typename = SETTYPE_NAME, ++ .features = IPSET_TYPE_IP | IPSET_DATA_SINGLE, ++ .protocol_version = IP_SET_PROTOCOL_VERSION, ++ .create = create, ++ .destroy = destroy, ++ .flush = flush, ++ .reqsize = sizeof(struct ip_set_req_iptreemap), ++ .addip = addip, ++ .addip_kernel = addip_kernel, ++ .delip = delip, ++ .delip_kernel = delip_kernel, ++ .testip = testip, ++ .testip_kernel = testip_kernel, ++ .header_size = sizeof(struct ip_set_req_iptreemap_create), ++ .list_header = list_header, ++ .list_members_size = list_members_size, ++ .list_members = list_members, ++ .me = THIS_MODULE, ++}; ++ ++MODULE_LICENSE("GPL"); ++MODULE_AUTHOR("Sven Wegener "); ++MODULE_DESCRIPTION("iptreemap type of IP sets"); ++ ++static int __init ip_set_iptreemap_init(void) ++{ ++ int ret = -ENOMEM; ++ int a; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) ++ cachep_b = kmem_cache_create("ip_set_iptreemap_b", ++ sizeof(struct ip_set_iptreemap_b), ++ 0, 0, NULL); ++#else ++ cachep_b = kmem_cache_create("ip_set_iptreemap_b", ++ sizeof(struct ip_set_iptreemap_b), ++ 0, 0, NULL, NULL); ++#endif ++ if (!cachep_b) { ++ ip_set_printk("Unable to create ip_set_iptreemap_b slab cache"); ++ goto out; ++ } ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) ++ cachep_c = kmem_cache_create("ip_set_iptreemap_c", ++ sizeof(struct ip_set_iptreemap_c), ++ 0, 0, NULL); ++#else ++ cachep_c = kmem_cache_create("ip_set_iptreemap_c", ++ sizeof(struct ip_set_iptreemap_c), ++ 0, 0, NULL, NULL); ++#endif ++ if (!cachep_c) { ++ ip_set_printk("Unable to create ip_set_iptreemap_c slab cache"); ++ goto outb; ++ } ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) ++ cachep_d = kmem_cache_create("ip_set_iptreemap_d", ++ sizeof(struct ip_set_iptreemap_d), ++ 0, 0, NULL); ++#else ++ cachep_d = kmem_cache_create("ip_set_iptreemap_d", ++ sizeof(struct ip_set_iptreemap_d), ++ 0, 0, NULL, NULL); ++#endif ++ if (!cachep_d) { ++ ip_set_printk("Unable to create ip_set_iptreemap_d slab cache"); ++ goto outc; ++ } ++ ++ fullbitmap_d = kmem_cache_alloc(cachep_d, GFP_KERNEL); ++ if (!fullbitmap_d) ++ goto outd; ++ ++ fullbitmap_c = kmem_cache_alloc(cachep_c, GFP_KERNEL); ++ if (!fullbitmap_c) ++ goto outbitmapd; ++ ++ fullbitmap_b = kmem_cache_alloc(cachep_b, GFP_KERNEL); ++ if (!fullbitmap_b) ++ goto outbitmapc; ++ ++ ret = ip_set_register_set_type(&ip_set_iptreemap); ++ if (0 > ret) ++ goto outbitmapb; ++ ++ /* Now init our global bitmaps */ ++ memset(fullbitmap_d->bitmap, 0xff, sizeof(fullbitmap_d->bitmap)); ++ ++ for (a = 0; a < 256; a++) ++ fullbitmap_c->tree[a] = fullbitmap_d; ++ ++ for (a = 0; a < 256; a++) ++ fullbitmap_b->tree[a] = fullbitmap_c; ++ memset(fullbitmap_b->dirty, 0, sizeof(fullbitmap_b->dirty)); ++ ++ return 0; ++ ++outbitmapb: ++ kmem_cache_free(cachep_b, fullbitmap_b); ++outbitmapc: ++ kmem_cache_free(cachep_c, fullbitmap_c); ++outbitmapd: ++ kmem_cache_free(cachep_d, fullbitmap_d); ++outd: ++ kmem_cache_destroy(cachep_d); ++outc: ++ kmem_cache_destroy(cachep_c); ++outb: ++ kmem_cache_destroy(cachep_b); ++out: ++ ++ return ret; ++} ++ ++static void __exit ip_set_iptreemap_fini(void) ++{ ++ ip_set_unregister_set_type(&ip_set_iptreemap); ++ kmem_cache_free(cachep_d, fullbitmap_d); ++ kmem_cache_free(cachep_c, fullbitmap_c); ++ kmem_cache_free(cachep_b, fullbitmap_b); ++ kmem_cache_destroy(cachep_d); ++ kmem_cache_destroy(cachep_c); ++ kmem_cache_destroy(cachep_b); ++} ++ ++module_init(ip_set_iptreemap_init); ++module_exit(ip_set_iptreemap_fini); +--- /dev/null ++++ b/net/ipv4/netfilter/ip_set_macipmap.c +@@ -0,0 +1,375 @@ ++/* Copyright (C) 2000-2002 Joakim Axelsson ++ * Patrick Schaaf ++ * Martin Josefsson ++ * Copyright (C) 2003-2004 Jozsef Kadlecsik ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++/* Kernel module implementing an IP set type: the macipmap type */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++#include ++ ++static int ++testip(struct ip_set *set, const void *data, size_t size, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_macipmap *map = (struct ip_set_macipmap *) set->data; ++ struct ip_set_macip *table = (struct ip_set_macip *) map->members; ++ struct ip_set_req_macipmap *req = (struct ip_set_req_macipmap *) data; ++ ++ if (size != sizeof(struct ip_set_req_macipmap)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_macipmap), ++ size); ++ return -EINVAL; ++ } ++ ++ if (req->ip < map->first_ip || req->ip > map->last_ip) ++ return -ERANGE; ++ ++ *hash_ip = req->ip; ++ DP("set: %s, ip:%u.%u.%u.%u, %u.%u.%u.%u", ++ set->name, HIPQUAD(req->ip), HIPQUAD(*hash_ip)); ++ if (test_bit(IPSET_MACIP_ISSET, ++ (void *) &table[req->ip - map->first_ip].flags)) { ++ return (memcmp(req->ethernet, ++ &table[req->ip - map->first_ip].ethernet, ++ ETH_ALEN) == 0); ++ } else { ++ return (map->flags & IPSET_MACIP_MATCHUNSET ? 1 : 0); ++ } ++} ++ ++static int ++testip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ struct ip_set_macipmap *map = ++ (struct ip_set_macipmap *) set->data; ++ struct ip_set_macip *table = ++ (struct ip_set_macip *) map->members; ++ ip_set_ip_t ip; ++ ++ ip = ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr); ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr); ++#endif ++ ++ if (ip < map->first_ip || ip > map->last_ip) ++ return 0; ++ ++ *hash_ip = ip; ++ DP("set: %s, ip:%u.%u.%u.%u, %u.%u.%u.%u", ++ set->name, HIPQUAD(ip), HIPQUAD(*hash_ip)); ++ if (test_bit(IPSET_MACIP_ISSET, ++ (void *) &table[ip - map->first_ip].flags)) { ++ /* Is mac pointer valid? ++ * If so, compare... */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ return (skb_mac_header(skb) >= skb->head ++ && (skb_mac_header(skb) + ETH_HLEN) <= skb->data ++#else ++ return (skb->mac.raw >= skb->head ++ && (skb->mac.raw + ETH_HLEN) <= skb->data ++#endif ++ && (memcmp(eth_hdr(skb)->h_source, ++ &table[ip - map->first_ip].ethernet, ++ ETH_ALEN) == 0)); ++ } else { ++ return (map->flags & IPSET_MACIP_MATCHUNSET ? 1 : 0); ++ } ++} ++ ++/* returns 0 on success */ ++static inline int ++__addip(struct ip_set *set, ++ ip_set_ip_t ip, unsigned char *ethernet, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_macipmap *map = ++ (struct ip_set_macipmap *) set->data; ++ struct ip_set_macip *table = ++ (struct ip_set_macip *) map->members; ++ ++ if (ip < map->first_ip || ip > map->last_ip) ++ return -ERANGE; ++ if (test_and_set_bit(IPSET_MACIP_ISSET, ++ (void *) &table[ip - map->first_ip].flags)) ++ return -EEXIST; ++ ++ *hash_ip = ip; ++ DP("%u.%u.%u.%u, %u.%u.%u.%u", HIPQUAD(ip), HIPQUAD(*hash_ip)); ++ memcpy(&table[ip - map->first_ip].ethernet, ethernet, ETH_ALEN); ++ return 0; ++} ++ ++static int ++addip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_macipmap *req = ++ (struct ip_set_req_macipmap *) data; ++ ++ if (size != sizeof(struct ip_set_req_macipmap)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_macipmap), ++ size); ++ return -EINVAL; ++ } ++ return __addip(set, req->ip, req->ethernet, hash_ip); ++} ++ ++static int ++addip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ ip_set_ip_t ip; ++ ++ ip = ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr); ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr); ++#endif ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ if (!(skb_mac_header(skb) >= skb->head ++ && (skb_mac_header(skb) + ETH_HLEN) <= skb->data)) ++#else ++ if (!(skb->mac.raw >= skb->head ++ && (skb->mac.raw + ETH_HLEN) <= skb->data)) ++#endif ++ return -EINVAL; ++ ++ return __addip(set, ip, eth_hdr(skb)->h_source, hash_ip); ++} ++ ++static inline int ++__delip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_macipmap *map = ++ (struct ip_set_macipmap *) set->data; ++ struct ip_set_macip *table = ++ (struct ip_set_macip *) map->members; ++ ++ if (ip < map->first_ip || ip > map->last_ip) ++ return -ERANGE; ++ if (!test_and_clear_bit(IPSET_MACIP_ISSET, ++ (void *)&table[ip - map->first_ip].flags)) ++ return -EEXIST; ++ ++ *hash_ip = ip; ++ DP("%u.%u.%u.%u, %u.%u.%u.%u", HIPQUAD(ip), HIPQUAD(*hash_ip)); ++ return 0; ++} ++ ++static int ++delip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_macipmap *req = ++ (struct ip_set_req_macipmap *) data; ++ ++ if (size != sizeof(struct ip_set_req_macipmap)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_macipmap), ++ size); ++ return -EINVAL; ++ } ++ return __delip(set, req->ip, hash_ip); ++} ++ ++static int ++delip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ return __delip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip); ++} ++ ++static inline size_t members_size(ip_set_id_t from, ip_set_id_t to) ++{ ++ return (size_t)((to - from + 1) * sizeof(struct ip_set_macip)); ++} ++ ++static int create(struct ip_set *set, const void *data, size_t size) ++{ ++ int newbytes; ++ struct ip_set_req_macipmap_create *req = ++ (struct ip_set_req_macipmap_create *) data; ++ struct ip_set_macipmap *map; ++ ++ if (size != sizeof(struct ip_set_req_macipmap_create)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_macipmap_create), ++ size); ++ return -EINVAL; ++ } ++ ++ DP("from %u.%u.%u.%u to %u.%u.%u.%u", ++ HIPQUAD(req->from), HIPQUAD(req->to)); ++ ++ if (req->from > req->to) { ++ DP("bad ip range"); ++ return -ENOEXEC; ++ } ++ ++ if (req->to - req->from > MAX_RANGE) { ++ ip_set_printk("range too big (max %d addresses)", ++ MAX_RANGE+1); ++ return -ENOEXEC; ++ } ++ ++ map = kmalloc(sizeof(struct ip_set_macipmap), GFP_KERNEL); ++ if (!map) { ++ DP("out of memory for %d bytes", ++ sizeof(struct ip_set_macipmap)); ++ return -ENOMEM; ++ } ++ map->flags = req->flags; ++ map->first_ip = req->from; ++ map->last_ip = req->to; ++ newbytes = members_size(map->first_ip, map->last_ip); ++ map->members = ip_set_malloc(newbytes); ++ DP("members: %u %p", newbytes, map->members); ++ if (!map->members) { ++ DP("out of memory for %d bytes", newbytes); ++ kfree(map); ++ return -ENOMEM; ++ } ++ memset(map->members, 0, newbytes); ++ ++ set->data = map; ++ return 0; ++} ++ ++static void destroy(struct ip_set *set) ++{ ++ struct ip_set_macipmap *map = ++ (struct ip_set_macipmap *) set->data; ++ ++ ip_set_free(map->members, members_size(map->first_ip, map->last_ip)); ++ kfree(map); ++ ++ set->data = NULL; ++} ++ ++static void flush(struct ip_set *set) ++{ ++ struct ip_set_macipmap *map = ++ (struct ip_set_macipmap *) set->data; ++ memset(map->members, 0, members_size(map->first_ip, map->last_ip)); ++} ++ ++static void list_header(const struct ip_set *set, void *data) ++{ ++ struct ip_set_macipmap *map = ++ (struct ip_set_macipmap *) set->data; ++ struct ip_set_req_macipmap_create *header = ++ (struct ip_set_req_macipmap_create *) data; ++ ++ DP("list_header %x %x %u", map->first_ip, map->last_ip, ++ map->flags); ++ ++ header->from = map->first_ip; ++ header->to = map->last_ip; ++ header->flags = map->flags; ++} ++ ++static int list_members_size(const struct ip_set *set) ++{ ++ struct ip_set_macipmap *map = ++ (struct ip_set_macipmap *) set->data; ++ ++ DP("%u", members_size(map->first_ip, map->last_ip)); ++ return members_size(map->first_ip, map->last_ip); ++} ++ ++static void list_members(const struct ip_set *set, void *data) ++{ ++ struct ip_set_macipmap *map = ++ (struct ip_set_macipmap *) set->data; ++ ++ int bytes = members_size(map->first_ip, map->last_ip); ++ ++ DP("members: %u %p", bytes, map->members); ++ memcpy(data, map->members, bytes); ++} ++ ++static struct ip_set_type ip_set_macipmap = { ++ .typename = SETTYPE_NAME, ++ .features = IPSET_TYPE_IP | IPSET_DATA_SINGLE, ++ .protocol_version = IP_SET_PROTOCOL_VERSION, ++ .create = &create, ++ .destroy = &destroy, ++ .flush = &flush, ++ .reqsize = sizeof(struct ip_set_req_macipmap), ++ .addip = &addip, ++ .addip_kernel = &addip_kernel, ++ .delip = &delip, ++ .delip_kernel = &delip_kernel, ++ .testip = &testip, ++ .testip_kernel = &testip_kernel, ++ .header_size = sizeof(struct ip_set_req_macipmap_create), ++ .list_header = &list_header, ++ .list_members_size = &list_members_size, ++ .list_members = &list_members, ++ .me = THIS_MODULE, ++}; ++ ++MODULE_LICENSE("GPL"); ++MODULE_AUTHOR("Jozsef Kadlecsik "); ++MODULE_DESCRIPTION("macipmap type of IP sets"); ++ ++static int __init ip_set_macipmap_init(void) ++{ ++ init_max_malloc_size(); ++ return ip_set_register_set_type(&ip_set_macipmap); ++} ++ ++static void __exit ip_set_macipmap_fini(void) ++{ ++ /* FIXME: possible race with ip_set_create() */ ++ ip_set_unregister_set_type(&ip_set_macipmap); ++} ++ ++module_init(ip_set_macipmap_init); ++module_exit(ip_set_macipmap_fini); +--- /dev/null ++++ b/net/ipv4/netfilter/ip_set_nethash.c +@@ -0,0 +1,497 @@ ++/* Copyright (C) 2003-2004 Jozsef Kadlecsik ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++/* Kernel module implementing a cidr nethash set */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++ ++#include ++#include ++ ++static int limit = MAX_RANGE; ++ ++static inline __u32 ++jhash_ip(const struct ip_set_nethash *map, uint16_t i, ip_set_ip_t ip) ++{ ++ return jhash_1word(ip, *(((uint32_t *) map->initval) + i)); ++} ++ ++static inline __u32 ++hash_id_cidr(struct ip_set_nethash *map, ++ ip_set_ip_t ip, ++ unsigned char cidr, ++ ip_set_ip_t *hash_ip) ++{ ++ __u32 id; ++ u_int16_t i; ++ ip_set_ip_t *elem; ++ ++ *hash_ip = pack(ip, cidr); ++ ++ for (i = 0; i < map->probes; i++) { ++ id = jhash_ip(map, i, *hash_ip) % map->hashsize; ++ DP("hash key: %u", id); ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); ++ if (*elem == *hash_ip) ++ return id; ++ } ++ return UINT_MAX; ++} ++ ++static inline __u32 ++hash_id(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_nethash *map = (struct ip_set_nethash *) set->data; ++ __u32 id = UINT_MAX; ++ int i; ++ ++ for (i = 0; i < 30 && map->cidr[i]; i++) { ++ id = hash_id_cidr(map, ip, map->cidr[i], hash_ip); ++ if (id != UINT_MAX) ++ break; ++ } ++ return id; ++} ++ ++static inline int ++__testip_cidr(struct ip_set *set, ip_set_ip_t ip, unsigned char cidr, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_nethash *map = (struct ip_set_nethash *) set->data; ++ ++ return (ip && hash_id_cidr(map, ip, cidr, hash_ip) != UINT_MAX); ++} ++ ++static inline int ++__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) ++{ ++ return (ip && hash_id(set, ip, hash_ip) != UINT_MAX); ++} ++ ++static int ++testip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_nethash *req = ++ (struct ip_set_req_nethash *) data; ++ ++ if (size != sizeof(struct ip_set_req_nethash)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_nethash), ++ size); ++ return -EINVAL; ++ } ++ return (req->cidr == 32 ? __testip(set, req->ip, hash_ip) ++ : __testip_cidr(set, req->ip, req->cidr, hash_ip)); ++} ++ ++static int ++testip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ return __testip(set, ++ ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr), ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr), ++#endif ++ hash_ip); ++} ++ ++static inline int ++__addip_base(struct ip_set_nethash *map, ip_set_ip_t ip) ++{ ++ __u32 probe; ++ u_int16_t i; ++ ip_set_ip_t *elem; ++ ++ for (i = 0; i < map->probes; i++) { ++ probe = jhash_ip(map, i, ip) % map->hashsize; ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, probe); ++ if (*elem == ip) ++ return -EEXIST; ++ if (!*elem) { ++ *elem = ip; ++ map->elements++; ++ return 0; ++ } ++ } ++ /* Trigger rehashing */ ++ return -EAGAIN; ++} ++ ++static inline int ++__addip(struct ip_set_nethash *map, ip_set_ip_t ip, unsigned char cidr, ++ ip_set_ip_t *hash_ip) ++{ ++ if (!ip || map->elements >= limit) ++ return -ERANGE; ++ ++ *hash_ip = pack(ip, cidr); ++ DP("%u.%u.%u.%u/%u, %u.%u.%u.%u", HIPQUAD(ip), cidr, HIPQUAD(*hash_ip)); ++ ++ return __addip_base(map, *hash_ip); ++} ++ ++static void ++update_cidr_sizes(struct ip_set_nethash *map, unsigned char cidr) ++{ ++ unsigned char next; ++ int i; ++ ++ for (i = 0; i < 30 && map->cidr[i]; i++) { ++ if (map->cidr[i] == cidr) { ++ return; ++ } else if (map->cidr[i] < cidr) { ++ next = map->cidr[i]; ++ map->cidr[i] = cidr; ++ cidr = next; ++ } ++ } ++ if (i < 30) ++ map->cidr[i] = cidr; ++} ++ ++static int ++addip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_nethash *req = ++ (struct ip_set_req_nethash *) data; ++ int ret; ++ ++ if (size != sizeof(struct ip_set_req_nethash)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_nethash), ++ size); ++ return -EINVAL; ++ } ++ ret = __addip((struct ip_set_nethash *) set->data, ++ req->ip, req->cidr, hash_ip); ++ ++ if (ret == 0) ++ update_cidr_sizes((struct ip_set_nethash *) set->data, ++ req->cidr); ++ ++ return ret; ++} ++ ++static int ++addip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ struct ip_set_nethash *map = (struct ip_set_nethash *) set->data; ++ int ret = -ERANGE; ++ ip_set_ip_t ip = ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr); ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr); ++#endif ++ ++ if (map->cidr[0]) ++ ret = __addip(map, ip, map->cidr[0], hash_ip); ++ ++ return ret; ++} ++ ++static int retry(struct ip_set *set) ++{ ++ struct ip_set_nethash *map = (struct ip_set_nethash *) set->data; ++ ip_set_ip_t *elem; ++ void *members; ++ u_int32_t i, hashsize = map->hashsize; ++ int res; ++ struct ip_set_nethash *tmp; ++ ++ if (map->resize == 0) ++ return -ERANGE; ++ ++ again: ++ res = 0; ++ ++ /* Calculate new parameters */ ++ hashsize += (hashsize * map->resize)/100; ++ if (hashsize == map->hashsize) ++ hashsize++; ++ ++ ip_set_printk("rehashing of set %s triggered: " ++ "hashsize grows from %u to %u", ++ set->name, map->hashsize, hashsize); ++ ++ tmp = kmalloc(sizeof(struct ip_set_nethash) ++ + map->probes * sizeof(uint32_t), GFP_ATOMIC); ++ if (!tmp) { ++ DP("out of memory for %d bytes", ++ sizeof(struct ip_set_nethash) ++ + map->probes * sizeof(uint32_t)); ++ return -ENOMEM; ++ } ++ tmp->members = harray_malloc(hashsize, sizeof(ip_set_ip_t), GFP_ATOMIC); ++ if (!tmp->members) { ++ DP("out of memory for %d bytes", hashsize * sizeof(ip_set_ip_t)); ++ kfree(tmp); ++ return -ENOMEM; ++ } ++ tmp->hashsize = hashsize; ++ tmp->elements = 0; ++ tmp->probes = map->probes; ++ tmp->resize = map->resize; ++ memcpy(tmp->initval, map->initval, map->probes * sizeof(uint32_t)); ++ memcpy(tmp->cidr, map->cidr, 30 * sizeof(unsigned char)); ++ ++ write_lock_bh(&set->lock); ++ map = (struct ip_set_nethash *) set->data; /* Play safe */ ++ for (i = 0; i < map->hashsize && res == 0; i++) { ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i); ++ if (*elem) ++ res = __addip_base(tmp, *elem); ++ } ++ if (res) { ++ /* Failure, try again */ ++ write_unlock_bh(&set->lock); ++ harray_free(tmp->members); ++ kfree(tmp); ++ goto again; ++ } ++ ++ /* Success at resizing! */ ++ members = map->members; ++ ++ map->hashsize = tmp->hashsize; ++ map->members = tmp->members; ++ write_unlock_bh(&set->lock); ++ ++ harray_free(members); ++ kfree(tmp); ++ ++ return 0; ++} ++ ++static inline int ++__delip(struct ip_set_nethash *map, ip_set_ip_t ip, unsigned char cidr, ++ ip_set_ip_t *hash_ip) ++{ ++ ip_set_ip_t id, *elem; ++ ++ if (!ip) ++ return -ERANGE; ++ ++ id = hash_id_cidr(map, ip, cidr, hash_ip); ++ if (id == UINT_MAX) ++ return -EEXIST; ++ ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); ++ *elem = 0; ++ map->elements--; ++ return 0; ++} ++ ++static int ++delip(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_ip) ++{ ++ struct ip_set_req_nethash *req = ++ (struct ip_set_req_nethash *) data; ++ ++ if (size != sizeof(struct ip_set_req_nethash)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_nethash), ++ size); ++ return -EINVAL; ++ } ++ /* TODO: no garbage collection in map->cidr */ ++ return __delip((struct ip_set_nethash *) set->data, ++ req->ip, req->cidr, hash_ip); ++} ++ ++static int ++delip_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_ip, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ struct ip_set_nethash *map = (struct ip_set_nethash *) set->data; ++ int ret = -ERANGE; ++ ip_set_ip_t ip = ntohl(flags[index] & IPSET_SRC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ ? ip_hdr(skb)->saddr ++ : ip_hdr(skb)->daddr); ++#else ++ ? skb->nh.iph->saddr ++ : skb->nh.iph->daddr); ++#endif ++ ++ if (map->cidr[0]) ++ ret = __delip(map, ip, map->cidr[0], hash_ip); ++ ++ return ret; ++} ++ ++static int create(struct ip_set *set, const void *data, size_t size) ++{ ++ struct ip_set_req_nethash_create *req = ++ (struct ip_set_req_nethash_create *) data; ++ struct ip_set_nethash *map; ++ uint16_t i; ++ ++ if (size != sizeof(struct ip_set_req_nethash_create)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_nethash_create), ++ size); ++ return -EINVAL; ++ } ++ ++ if (req->hashsize < 1) { ++ ip_set_printk("hashsize too small"); ++ return -ENOEXEC; ++ } ++ if (req->probes < 1) { ++ ip_set_printk("probes too small"); ++ return -ENOEXEC; ++ } ++ ++ map = kmalloc(sizeof(struct ip_set_nethash) ++ + req->probes * sizeof(uint32_t), GFP_KERNEL); ++ if (!map) { ++ DP("out of memory for %d bytes", ++ sizeof(struct ip_set_nethash) ++ + req->probes * sizeof(uint32_t)); ++ return -ENOMEM; ++ } ++ for (i = 0; i < req->probes; i++) ++ get_random_bytes(((uint32_t *) map->initval)+i, 4); ++ map->elements = 0; ++ map->hashsize = req->hashsize; ++ map->probes = req->probes; ++ map->resize = req->resize; ++ memset(map->cidr, 0, 30 * sizeof(unsigned char)); ++ map->members = harray_malloc(map->hashsize, sizeof(ip_set_ip_t), GFP_KERNEL); ++ if (!map->members) { ++ DP("out of memory for %d bytes", map->hashsize * sizeof(ip_set_ip_t)); ++ kfree(map); ++ return -ENOMEM; ++ } ++ ++ set->data = map; ++ return 0; ++} ++ ++static void destroy(struct ip_set *set) ++{ ++ struct ip_set_nethash *map = (struct ip_set_nethash *) set->data; ++ ++ harray_free(map->members); ++ kfree(map); ++ ++ set->data = NULL; ++} ++ ++static void flush(struct ip_set *set) ++{ ++ struct ip_set_nethash *map = (struct ip_set_nethash *) set->data; ++ harray_flush(map->members, map->hashsize, sizeof(ip_set_ip_t)); ++ memset(map->cidr, 0, 30 * sizeof(unsigned char)); ++ map->elements = 0; ++} ++ ++static void list_header(const struct ip_set *set, void *data) ++{ ++ struct ip_set_nethash *map = (struct ip_set_nethash *) set->data; ++ struct ip_set_req_nethash_create *header = ++ (struct ip_set_req_nethash_create *) data; ++ ++ header->hashsize = map->hashsize; ++ header->probes = map->probes; ++ header->resize = map->resize; ++} ++ ++static int list_members_size(const struct ip_set *set) ++{ ++ struct ip_set_nethash *map = (struct ip_set_nethash *) set->data; ++ ++ return (map->hashsize * sizeof(ip_set_ip_t)); ++} ++ ++static void list_members(const struct ip_set *set, void *data) ++{ ++ struct ip_set_nethash *map = (struct ip_set_nethash *) set->data; ++ ip_set_ip_t i, *elem; ++ ++ for (i = 0; i < map->hashsize; i++) { ++ elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i); ++ ((ip_set_ip_t *)data)[i] = *elem; ++ } ++} ++ ++static struct ip_set_type ip_set_nethash = { ++ .typename = SETTYPE_NAME, ++ .features = IPSET_TYPE_IP | IPSET_DATA_SINGLE, ++ .protocol_version = IP_SET_PROTOCOL_VERSION, ++ .create = &create, ++ .destroy = &destroy, ++ .flush = &flush, ++ .reqsize = sizeof(struct ip_set_req_nethash), ++ .addip = &addip, ++ .addip_kernel = &addip_kernel, ++ .retry = &retry, ++ .delip = &delip, ++ .delip_kernel = &delip_kernel, ++ .testip = &testip, ++ .testip_kernel = &testip_kernel, ++ .header_size = sizeof(struct ip_set_req_nethash_create), ++ .list_header = &list_header, ++ .list_members_size = &list_members_size, ++ .list_members = &list_members, ++ .me = THIS_MODULE, ++}; ++ ++MODULE_LICENSE("GPL"); ++MODULE_AUTHOR("Jozsef Kadlecsik "); ++MODULE_DESCRIPTION("nethash type of IP sets"); ++module_param(limit, int, 0600); ++MODULE_PARM_DESC(limit, "maximal number of elements stored in the sets"); ++ ++static int __init ip_set_nethash_init(void) ++{ ++ return ip_set_register_set_type(&ip_set_nethash); ++} ++ ++static void __exit ip_set_nethash_fini(void) ++{ ++ /* FIXME: possible race with ip_set_create() */ ++ ip_set_unregister_set_type(&ip_set_nethash); ++} ++ ++module_init(ip_set_nethash_init); ++module_exit(ip_set_nethash_fini); +--- /dev/null ++++ b/net/ipv4/netfilter/ip_set_portmap.c +@@ -0,0 +1,346 @@ ++/* Copyright (C) 2003-2004 Jozsef Kadlecsik ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++/* Kernel module implementing a port set type as a bitmap */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++ ++#include ++ ++/* We must handle non-linear skbs */ ++static inline ip_set_ip_t ++get_port(const struct sk_buff *skb, u_int32_t flags) ++{ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ struct iphdr *iph = ip_hdr(skb); ++#else ++ struct iphdr *iph = skb->nh.iph; ++#endif ++ u_int16_t offset = ntohs(iph->frag_off) & IP_OFFSET; ++ switch (iph->protocol) { ++ case IPPROTO_TCP: { ++ struct tcphdr tcph; ++ ++ /* See comments at tcp_match in ip_tables.c */ ++ if (offset) ++ return INVALID_PORT; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ if (skb_copy_bits(skb, ip_hdr(skb)->ihl*4, &tcph, sizeof(tcph)) < 0) ++#else ++ if (skb_copy_bits(skb, skb->nh.iph->ihl*4, &tcph, sizeof(tcph)) < 0) ++#endif ++ /* No choice either */ ++ return INVALID_PORT; ++ ++ return ntohs(flags & IPSET_SRC ? ++ tcph.source : tcph.dest); ++ } ++ case IPPROTO_UDP: { ++ struct udphdr udph; ++ ++ if (offset) ++ return INVALID_PORT; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) ++ if (skb_copy_bits(skb, ip_hdr(skb)->ihl*4, &udph, sizeof(udph)) < 0) ++#else ++ if (skb_copy_bits(skb, skb->nh.iph->ihl*4, &udph, sizeof(udph)) < 0) ++#endif ++ /* No choice either */ ++ return INVALID_PORT; ++ ++ return ntohs(flags & IPSET_SRC ? ++ udph.source : udph.dest); ++ } ++ default: ++ return INVALID_PORT; ++ } ++} ++ ++static inline int ++__testport(struct ip_set *set, ip_set_ip_t port, ip_set_ip_t *hash_port) ++{ ++ struct ip_set_portmap *map = (struct ip_set_portmap *) set->data; ++ ++ if (port < map->first_port || port > map->last_port) ++ return -ERANGE; ++ ++ *hash_port = port; ++ DP("set: %s, port:%u, %u", set->name, port, *hash_port); ++ return !!test_bit(port - map->first_port, map->members); ++} ++ ++static int ++testport(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_port) ++{ ++ struct ip_set_req_portmap *req = ++ (struct ip_set_req_portmap *) data; ++ ++ if (size != sizeof(struct ip_set_req_portmap)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_portmap), ++ size); ++ return -EINVAL; ++ } ++ return __testport(set, req->port, hash_port); ++} ++ ++static int ++testport_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_port, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ int res; ++ ip_set_ip_t port = get_port(skb, flags[index]); ++ ++ DP("flag %s port %u", flags[index] & IPSET_SRC ? "SRC" : "DST", port); ++ if (port == INVALID_PORT) ++ return 0; ++ ++ res = __testport(set, port, hash_port); ++ ++ return (res < 0 ? 0 : res); ++} ++ ++static inline int ++__addport(struct ip_set *set, ip_set_ip_t port, ip_set_ip_t *hash_port) ++{ ++ struct ip_set_portmap *map = (struct ip_set_portmap *) set->data; ++ ++ if (port < map->first_port || port > map->last_port) ++ return -ERANGE; ++ if (test_and_set_bit(port - map->first_port, map->members)) ++ return -EEXIST; ++ ++ *hash_port = port; ++ DP("port %u", port); ++ return 0; ++} ++ ++static int ++addport(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_port) ++{ ++ struct ip_set_req_portmap *req = ++ (struct ip_set_req_portmap *) data; ++ ++ if (size != sizeof(struct ip_set_req_portmap)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_portmap), ++ size); ++ return -EINVAL; ++ } ++ return __addport(set, req->port, hash_port); ++} ++ ++static int ++addport_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_port, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ ip_set_ip_t port = get_port(skb, flags[index]); ++ ++ if (port == INVALID_PORT) ++ return -EINVAL; ++ ++ return __addport(set, port, hash_port); ++} ++ ++static inline int ++__delport(struct ip_set *set, ip_set_ip_t port, ip_set_ip_t *hash_port) ++{ ++ struct ip_set_portmap *map = (struct ip_set_portmap *) set->data; ++ ++ if (port < map->first_port || port > map->last_port) ++ return -ERANGE; ++ if (!test_and_clear_bit(port - map->first_port, map->members)) ++ return -EEXIST; ++ ++ *hash_port = port; ++ DP("port %u", port); ++ return 0; ++} ++ ++static int ++delport(struct ip_set *set, const void *data, size_t size, ++ ip_set_ip_t *hash_port) ++{ ++ struct ip_set_req_portmap *req = ++ (struct ip_set_req_portmap *) data; ++ ++ if (size != sizeof(struct ip_set_req_portmap)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_portmap), ++ size); ++ return -EINVAL; ++ } ++ return __delport(set, req->port, hash_port); ++} ++ ++static int ++delport_kernel(struct ip_set *set, ++ const struct sk_buff *skb, ++ ip_set_ip_t *hash_port, ++ const u_int32_t *flags, ++ unsigned char index) ++{ ++ ip_set_ip_t port = get_port(skb, flags[index]); ++ ++ if (port == INVALID_PORT) ++ return -EINVAL; ++ ++ return __delport(set, port, hash_port); ++} ++ ++static int create(struct ip_set *set, const void *data, size_t size) ++{ ++ int newbytes; ++ struct ip_set_req_portmap_create *req = ++ (struct ip_set_req_portmap_create *) data; ++ struct ip_set_portmap *map; ++ ++ if (size != sizeof(struct ip_set_req_portmap_create)) { ++ ip_set_printk("data length wrong (want %zu, have %zu)", ++ sizeof(struct ip_set_req_portmap_create), ++ size); ++ return -EINVAL; ++ } ++ ++ DP("from %u to %u", req->from, req->to); ++ ++ if (req->from > req->to) { ++ DP("bad port range"); ++ return -ENOEXEC; ++ } ++ ++ if (req->to - req->from > MAX_RANGE) { ++ ip_set_printk("range too big (max %d ports)", ++ MAX_RANGE+1); ++ return -ENOEXEC; ++ } ++ ++ map = kmalloc(sizeof(struct ip_set_portmap), GFP_KERNEL); ++ if (!map) { ++ DP("out of memory for %d bytes", ++ sizeof(struct ip_set_portmap)); ++ return -ENOMEM; ++ } ++ map->first_port = req->from; ++ map->last_port = req->to; ++ newbytes = bitmap_bytes(req->from, req->to); ++ map->members = kmalloc(newbytes, GFP_KERNEL); ++ if (!map->members) { ++ DP("out of memory for %d bytes", newbytes); ++ kfree(map); ++ return -ENOMEM; ++ } ++ memset(map->members, 0, newbytes); ++ ++ set->data = map; ++ return 0; ++} ++ ++static void destroy(struct ip_set *set) ++{ ++ struct ip_set_portmap *map = (struct ip_set_portmap *) set->data; ++ ++ kfree(map->members); ++ kfree(map); ++ ++ set->data = NULL; ++} ++ ++static void flush(struct ip_set *set) ++{ ++ struct ip_set_portmap *map = (struct ip_set_portmap *) set->data; ++ memset(map->members, 0, bitmap_bytes(map->first_port, map->last_port)); ++} ++ ++static void list_header(const struct ip_set *set, void *data) ++{ ++ struct ip_set_portmap *map = (struct ip_set_portmap *) set->data; ++ struct ip_set_req_portmap_create *header = ++ (struct ip_set_req_portmap_create *) data; ++ ++ DP("list_header %u %u", map->first_port, map->last_port); ++ ++ header->from = map->first_port; ++ header->to = map->last_port; ++} ++ ++static int list_members_size(const struct ip_set *set) ++{ ++ struct ip_set_portmap *map = (struct ip_set_portmap *) set->data; ++ ++ return bitmap_bytes(map->first_port, map->last_port); ++} ++ ++static void list_members(const struct ip_set *set, void *data) ++{ ++ struct ip_set_portmap *map = (struct ip_set_portmap *) set->data; ++ int bytes = bitmap_bytes(map->first_port, map->last_port); ++ ++ memcpy(data, map->members, bytes); ++} ++ ++static struct ip_set_type ip_set_portmap = { ++ .typename = SETTYPE_NAME, ++ .features = IPSET_TYPE_PORT | IPSET_DATA_SINGLE, ++ .protocol_version = IP_SET_PROTOCOL_VERSION, ++ .create = &create, ++ .destroy = &destroy, ++ .flush = &flush, ++ .reqsize = sizeof(struct ip_set_req_portmap), ++ .addip = &addport, ++ .addip_kernel = &addport_kernel, ++ .delip = &delport, ++ .delip_kernel = &delport_kernel, ++ .testip = &testport, ++ .testip_kernel = &testport_kernel, ++ .header_size = sizeof(struct ip_set_req_portmap_create), ++ .list_header = &list_header, ++ .list_members_size = &list_members_size, ++ .list_members = &list_members, ++ .me = THIS_MODULE, ++}; ++ ++MODULE_LICENSE("GPL"); ++MODULE_AUTHOR("Jozsef Kadlecsik "); ++MODULE_DESCRIPTION("portmap type of IP sets"); ++ ++static int __init ip_set_portmap_init(void) ++{ ++ return ip_set_register_set_type(&ip_set_portmap); ++} ++ ++static void __exit ip_set_portmap_fini(void) ++{ ++ /* FIXME: possible race with ip_set_create() */ ++ ip_set_unregister_set_type(&ip_set_portmap); ++} ++ ++module_init(ip_set_portmap_init); ++module_exit(ip_set_portmap_fini); +--- /dev/null ++++ b/net/ipv4/netfilter/ipt_set.c +@@ -0,0 +1,160 @@ ++/* Copyright (C) 2000-2002 Joakim Axelsson ++ * Patrick Schaaf ++ * Martin Josefsson ++ * Copyright (C) 2003-2004 Jozsef Kadlecsik ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++/* Kernel module to match an IP set. */ ++ ++#include ++#include ++#include ++#include ++ ++#include ++#include ++#include ++ ++static inline int ++match_set(const struct ipt_set_info *info, ++ const struct sk_buff *skb, ++ int inv) ++{ ++ if (ip_set_testip_kernel(info->index, skb, info->flags)) ++ inv = !inv; ++ return inv; ++} ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) ++static bool ++#else ++static int ++#endif ++match(const struct sk_buff *skb, ++ const struct net_device *in, ++ const struct net_device *out, ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) ++ const struct xt_match *match, ++#endif ++ const void *matchinfo, ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) ++ int offset, unsigned int protoff, bool *hotdrop) ++#elif LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ int offset, unsigned int protoff, int *hotdrop) ++#else ++ int offset, int *hotdrop) ++#endif ++{ ++ const struct ipt_set_info_match *info = matchinfo; ++ ++ return match_set(&info->match_set, ++ skb, ++ info->match_set.flags[0] & IPSET_MATCH_INV); ++} ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) ++bool ++#else ++static int ++#endif ++checkentry(const char *tablename, ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ const void *inf, ++#else ++ const struct ipt_ip *ip, ++#endif ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) ++ const struct xt_match *match, ++#endif ++ void *matchinfo, ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) ++ unsigned int matchsize, ++#endif ++ unsigned int hook_mask) ++{ ++ struct ipt_set_info_match *info = ++ (struct ipt_set_info_match *) matchinfo; ++ ip_set_id_t index; ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) ++ if (matchsize != IPT_ALIGN(sizeof(struct ipt_set_info_match))) { ++ ip_set_printk("invalid matchsize %d", matchsize); ++ return 0; ++ } ++#endif ++ ++ index = ip_set_get_byindex(info->match_set.index); ++ ++ if (index == IP_SET_INVALID_ID) { ++ ip_set_printk("Cannot find set indentified by id %u to match", ++ info->match_set.index); ++ return 0; /* error */ ++ } ++ if (info->match_set.flags[IP_SET_MAX_BINDINGS] != 0) { ++ ip_set_printk("That's nasty!"); ++ return 0; /* error */ ++ } ++ ++ return 1; ++} ++ ++static void destroy( ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) ++ const struct xt_match *match, ++#endif ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) ++ void *matchinfo, unsigned int matchsize) ++#else ++ void *matchinfo) ++#endif ++{ ++ struct ipt_set_info_match *info = matchinfo; ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) ++ if (matchsize != IPT_ALIGN(sizeof(struct ipt_set_info_match))) { ++ ip_set_printk("invalid matchsize %d", matchsize); ++ return; ++ } ++#endif ++ ip_set_put(info->match_set.index); ++} ++ ++static struct ipt_match set_match = { ++ .name = "set", ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,21) ++ .family = AF_INET, ++#endif ++ .match = &match, ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) ++ .matchsize = sizeof(struct ipt_set_info_match), ++#endif ++ .checkentry = &checkentry, ++ .destroy = &destroy, ++ .me = THIS_MODULE ++}; ++ ++MODULE_LICENSE("GPL"); ++MODULE_AUTHOR("Jozsef Kadlecsik "); ++MODULE_DESCRIPTION("iptables IP set match module"); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,21) ++#define ipt_register_match xt_register_match ++#define ipt_unregister_match xt_unregister_match ++#endif ++ ++static int __init ipt_ipset_init(void) ++{ ++ return ipt_register_match(&set_match); ++} ++ ++static void __exit ipt_ipset_fini(void) ++{ ++ ipt_unregister_match(&set_match); ++} ++ ++module_init(ipt_ipset_init); ++module_exit(ipt_ipset_fini); +--- /dev/null ++++ b/net/ipv4/netfilter/ipt_SET.c +@@ -0,0 +1,179 @@ ++/* Copyright (C) 2000-2002 Joakim Axelsson ++ * Patrick Schaaf ++ * Martin Josefsson ++ * Copyright (C) 2003-2004 Jozsef Kadlecsik ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++/* ipt_SET.c - netfilter target to manipulate IP sets */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++static unsigned int ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,24) ++target(struct sk_buff *skb, ++#else ++target(struct sk_buff **pskb, ++#endif ++ const struct net_device *in, ++ const struct net_device *out, ++ unsigned int hooknum, ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) ++ const struct xt_target *target, ++#endif ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) ++ const void *targinfo, ++ void *userinfo) ++#else ++ const void *targinfo) ++#endif ++{ ++ const struct ipt_set_info_target *info = targinfo; ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24) ++ struct sk_buff *skb = *pskb; ++#endif ++ ++ if (info->add_set.index != IP_SET_INVALID_ID) ++ ip_set_addip_kernel(info->add_set.index, ++ skb, ++ info->add_set.flags); ++ if (info->del_set.index != IP_SET_INVALID_ID) ++ ip_set_delip_kernel(info->del_set.index, ++ skb, ++ info->del_set.flags); ++ ++ return IPT_CONTINUE; ++} ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) ++static bool ++#else ++static int ++#endif ++checkentry(const char *tablename, ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ const void *e, ++#else ++ const struct ipt_entry *e, ++#endif ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) ++ const struct xt_target *target, ++#endif ++ void *targinfo, ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) ++ unsigned int targinfosize, ++#endif ++ unsigned int hook_mask) ++{ ++ struct ipt_set_info_target *info = ++ (struct ipt_set_info_target *) targinfo; ++ ip_set_id_t index; ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) ++ if (targinfosize != IPT_ALIGN(sizeof(*info))) { ++ DP("bad target info size %u", targinfosize); ++ return 0; ++ } ++#endif ++ ++ if (info->add_set.index != IP_SET_INVALID_ID) { ++ index = ip_set_get_byindex(info->add_set.index); ++ if (index == IP_SET_INVALID_ID) { ++ ip_set_printk("cannot find add_set index %u as target", ++ info->add_set.index); ++ return 0; /* error */ ++ } ++ } ++ ++ if (info->del_set.index != IP_SET_INVALID_ID) { ++ index = ip_set_get_byindex(info->del_set.index); ++ if (index == IP_SET_INVALID_ID) { ++ ip_set_printk("cannot find del_set index %u as target", ++ info->del_set.index); ++ return 0; /* error */ ++ } ++ } ++ if (info->add_set.flags[IP_SET_MAX_BINDINGS] != 0 ++ || info->del_set.flags[IP_SET_MAX_BINDINGS] != 0) { ++ ip_set_printk("That's nasty!"); ++ return 0; /* error */ ++ } ++ ++ return 1; ++} ++ ++static void destroy( ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) ++ const struct xt_target *target, ++#endif ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) ++ void *targetinfo, unsigned int targetsize) ++#else ++ void *targetinfo) ++#endif ++{ ++ struct ipt_set_info_target *info = targetinfo; ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) ++ if (targetsize != IPT_ALIGN(sizeof(struct ipt_set_info_target))) { ++ ip_set_printk("invalid targetsize %d", targetsize); ++ return; ++ } ++#endif ++ if (info->add_set.index != IP_SET_INVALID_ID) ++ ip_set_put(info->add_set.index); ++ if (info->del_set.index != IP_SET_INVALID_ID) ++ ip_set_put(info->del_set.index); ++} ++ ++static struct ipt_target SET_target = { ++ .name = "SET", ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,21) ++ .family = AF_INET, ++#endif ++ .target = target, ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) ++ .targetsize = sizeof(struct ipt_set_info_target), ++#endif ++ .checkentry = checkentry, ++ .destroy = destroy, ++ .me = THIS_MODULE ++}; ++ ++MODULE_LICENSE("GPL"); ++MODULE_AUTHOR("Jozsef Kadlecsik "); ++MODULE_DESCRIPTION("iptables IP set target module"); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,21) ++#define ipt_register_target xt_register_target ++#define ipt_unregister_target xt_unregister_target ++#endif ++ ++static int __init ipt_SET_init(void) ++{ ++ return ipt_register_target(&SET_target); ++} ++ ++static void __exit ipt_SET_fini(void) ++{ ++ ipt_unregister_target(&SET_target); ++} ++ ++module_init(ipt_SET_init); ++module_exit(ipt_SET_fini); +--- a/net/ipv4/netfilter/Kconfig ++++ b/net/ipv4/netfilter/Kconfig +@@ -388,5 +388,122 @@ config IP_NF_ARP_MANGLE + + endif # IP_NF_ARPTABLES + ++config IP_NF_SET ++ tristate "IP set support" ++ depends on INET && NETFILTER ++ help ++ This option adds IP set support to the kernel. ++ In order to define and use sets, you need the userspace utility ++ ipset(8). ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ ++config IP_NF_SET_MAX ++ int "Maximum number of IP sets" ++ default 256 ++ range 2 65534 ++ depends on IP_NF_SET ++ help ++ You can define here default value of the maximum number ++ of IP sets for the kernel. ++ ++ The value can be overriden by the 'max_sets' module ++ parameter of the 'ip_set' module. ++ ++config IP_NF_SET_HASHSIZE ++ int "Hash size for bindings of IP sets" ++ default 1024 ++ depends on IP_NF_SET ++ help ++ You can define here default value of the hash size for ++ bindings of IP sets. ++ ++ The value can be overriden by the 'hash_size' module ++ parameter of the 'ip_set' module. ++ ++config IP_NF_SET_IPMAP ++ tristate "ipmap set support" ++ depends on IP_NF_SET ++ help ++ This option adds the ipmap set type support. ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ ++config IP_NF_SET_MACIPMAP ++ tristate "macipmap set support" ++ depends on IP_NF_SET ++ help ++ This option adds the macipmap set type support. ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ ++config IP_NF_SET_PORTMAP ++ tristate "portmap set support" ++ depends on IP_NF_SET ++ help ++ This option adds the portmap set type support. ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ ++config IP_NF_SET_IPHASH ++ tristate "iphash set support" ++ depends on IP_NF_SET ++ help ++ This option adds the iphash set type support. ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ ++config IP_NF_SET_NETHASH ++ tristate "nethash set support" ++ depends on IP_NF_SET ++ help ++ This option adds the nethash set type support. ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ ++config IP_NF_SET_IPPORTHASH ++ tristate "ipporthash set support" ++ depends on IP_NF_SET ++ help ++ This option adds the ipporthash set type support. ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ ++config IP_NF_SET_IPTREE ++ tristate "iptree set support" ++ depends on IP_NF_SET ++ help ++ This option adds the iptree set type support. ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ ++config IP_NF_SET_IPTREEMAP ++ tristate "iptreemap set support" ++ depends on IP_NF_SET ++ help ++ This option adds the iptreemap set type support. ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ ++config IP_NF_MATCH_SET ++ tristate "set match support" ++ depends on IP_NF_SET ++ help ++ Set matching matches against given IP sets. ++ You need the ipset utility to create and set up the sets. ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ ++config IP_NF_TARGET_SET ++ tristate "SET target support" ++ depends on IP_NF_SET ++ help ++ The SET target makes possible to add/delete entries ++ in IP sets. ++ You need the ipset utility to create and set up the sets. ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ ++ + endmenu + +--- a/net/ipv4/netfilter/Makefile ++++ b/net/ipv4/netfilter/Makefile +@@ -51,6 +51,7 @@ obj-$(CONFIG_IP_NF_SECURITY) += iptable_ + obj-$(CONFIG_IP_NF_MATCH_ADDRTYPE) += ipt_addrtype.o + obj-$(CONFIG_IP_NF_MATCH_AH) += ipt_ah.o + obj-$(CONFIG_IP_NF_MATCH_ECN) += ipt_ecn.o ++obj-$(CONFIG_IP_NF_MATCH_SET) += ipt_set.o + + # targets + obj-$(CONFIG_IP_NF_TARGET_CLUSTERIP) += ipt_CLUSTERIP.o +@@ -61,6 +62,18 @@ obj-$(CONFIG_IP_NF_TARGET_NETMAP) += ipt + obj-$(CONFIG_IP_NF_TARGET_REDIRECT) += ipt_REDIRECT.o + obj-$(CONFIG_IP_NF_TARGET_REJECT) += ipt_REJECT.o + obj-$(CONFIG_IP_NF_TARGET_ULOG) += ipt_ULOG.o ++obj-$(CONFIG_IP_NF_TARGET_SET) += ipt_SET.o ++ ++# sets ++obj-$(CONFIG_IP_NF_SET) += ip_set.o ++obj-$(CONFIG_IP_NF_SET_IPMAP) += ip_set_ipmap.o ++obj-$(CONFIG_IP_NF_SET_PORTMAP) += ip_set_portmap.o ++obj-$(CONFIG_IP_NF_SET_MACIPMAP) += ip_set_macipmap.o ++obj-$(CONFIG_IP_NF_SET_IPHASH) += ip_set_iphash.o ++obj-$(CONFIG_IP_NF_SET_NETHASH) += ip_set_nethash.o ++obj-$(CONFIG_IP_NF_SET_IPPORTHASH) += ip_set_ipporthash.o ++obj-$(CONFIG_IP_NF_SET_IPTREE) += ip_set_iptree.o ++obj-$(CONFIG_IP_NF_SET_IPTREEMAP) += ip_set_iptreemap.o + + # generic ARP tables + obj-$(CONFIG_IP_NF_ARPTABLES) += arp_tables.o diff --git a/target/linux/generic-2.6/patches-2.6.31/150-netfilter_imq.patch b/target/linux/generic-2.6/patches-2.6.31/150-netfilter_imq.patch new file mode 100644 index 0000000000..70c80c81db --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/150-netfilter_imq.patch @@ -0,0 +1,1260 @@ +--- /dev/null ++++ b/drivers/net/imq.c +@@ -0,0 +1,571 @@ ++/* ++ * Pseudo-driver for the intermediate queue device. ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ * ++ * Authors: Patrick McHardy, ++ * ++ * The first version was written by Martin Devera, ++ * ++ * Credits: Jan Rafaj ++ * - Update patch to 2.4.21 ++ * Sebastian Strollo ++ * - Fix "Dead-loop on netdevice imq"-issue ++ * Marcel Sebek ++ * - Update to 2.6.2-rc1 ++ * ++ * After some time of inactivity there is a group taking care ++ * of IMQ again: http://www.linuximq.net ++ * ++ * ++ * 2004/06/30 - New version of IMQ patch to kernels <=2.6.7 ++ * including the following changes: ++ * ++ * - Correction of ipv6 support "+"s issue (Hasso Tepper) ++ * - Correction of imq_init_devs() issue that resulted in ++ * kernel OOPS unloading IMQ as module (Norbert Buchmuller) ++ * - Addition of functionality to choose number of IMQ devices ++ * during kernel config (Andre Correa) ++ * - Addition of functionality to choose how IMQ hooks on ++ * PRE and POSTROUTING (after or before NAT) (Andre Correa) ++ * - Cosmetic corrections (Norbert Buchmuller) (Andre Correa) ++ * ++ * ++ * 2005/12/16 - IMQ versions between 2.6.7 and 2.6.13 were ++ * released with almost no problems. 2.6.14-x was released ++ * with some important changes: nfcache was removed; After ++ * some weeks of trouble we figured out that some IMQ fields ++ * in skb were missing in skbuff.c - skb_clone and copy_skb_header. ++ * These functions are correctly patched by this new patch version. ++ * ++ * Thanks for all who helped to figure out all the problems with ++ * 2.6.14.x: Patrick McHardy, Rune Kock, VeNoMouS, Max CtRiX, ++ * Kevin Shanahan, Richard Lucassen, Valery Dachev (hopefully ++ * I didn't forget anybody). I apologize again for my lack of time. ++ * ++ * ++ * 2008/06/17 - 2.6.25 - Changed imq.c to use qdisc_run() instead ++ * of qdisc_restart() and moved qdisc_run() to tasklet to avoid ++ * recursive locking. New initialization routines to fix 'rmmod' not ++ * working anymore. Used code from ifb.c. (Jussi Kivilinna) ++ * ++ * 2008/08/06 - 2.6.26 - (JK) ++ * - Replaced tasklet with 'netif_schedule()'. ++ * - Cleaned up and added comments for imq_nf_queue(). ++ * ++ * 2009/04/12 ++ * - Add skb_save_cb/skb_restore_cb helper functions for backuping ++ * control buffer. This is needed because qdisc-layer on kernels ++ * 2.6.27 and newer overwrite control buffer. (Jussi Kivilinna) ++ * - Add better locking for IMQ device. Hopefully this will solve ++ * SMP issues. (Jussi Kivilinna) ++ * - Port to 2.6.27 ++ * - Port to 2.6.28 ++ * - Port to 2.6.29 + fix rmmod not working ++ * ++ * 2009/04/20 - (Jussi Kivilinna) ++ * - Use netdevice feature flags to avoid extra packet handling ++ * by core networking layer and possibly increase performance. ++ * ++ * Also, many thanks to pablo Sebastian Greco for making the initial ++ * patch and to those who helped the testing. ++ * ++ * More info at: http://www.linuximq.net/ (Andre Correa) ++ */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) ++ #include ++#endif ++#include ++#include ++#include ++ ++static nf_hookfn imq_nf_hook; ++ ++static struct nf_hook_ops imq_ingress_ipv4 = { ++ .hook = imq_nf_hook, ++ .owner = THIS_MODULE, ++ .pf = PF_INET, ++ .hooknum = NF_INET_PRE_ROUTING, ++#if defined(CONFIG_IMQ_BEHAVIOR_BA) || defined(CONFIG_IMQ_BEHAVIOR_BB) ++ .priority = NF_IP_PRI_MANGLE + 1 ++#else ++ .priority = NF_IP_PRI_NAT_DST + 1 ++#endif ++}; ++ ++static struct nf_hook_ops imq_egress_ipv4 = { ++ .hook = imq_nf_hook, ++ .owner = THIS_MODULE, ++ .pf = PF_INET, ++ .hooknum = NF_INET_POST_ROUTING, ++#if defined(CONFIG_IMQ_BEHAVIOR_AA) || defined(CONFIG_IMQ_BEHAVIOR_BA) ++ .priority = NF_IP_PRI_LAST ++#else ++ .priority = NF_IP_PRI_NAT_SRC - 1 ++#endif ++}; ++ ++#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) ++static struct nf_hook_ops imq_ingress_ipv6 = { ++ .hook = imq_nf_hook, ++ .owner = THIS_MODULE, ++ .pf = PF_INET6, ++ .hooknum = NF_INET_PRE_ROUTING, ++#if defined(CONFIG_IMQ_BEHAVIOR_BA) || defined(CONFIG_IMQ_BEHAVIOR_BB) ++ .priority = NF_IP6_PRI_MANGLE + 1 ++#else ++ .priority = NF_IP6_PRI_NAT_DST + 1 ++#endif ++}; ++ ++static struct nf_hook_ops imq_egress_ipv6 = { ++ .hook = imq_nf_hook, ++ .owner = THIS_MODULE, ++ .pf = PF_INET6, ++ .hooknum = NF_INET_POST_ROUTING, ++#if defined(CONFIG_IMQ_BEHAVIOR_AA) || defined(CONFIG_IMQ_BEHAVIOR_BA) ++ .priority = NF_IP6_PRI_LAST ++#else ++ .priority = NF_IP6_PRI_NAT_SRC - 1 ++#endif ++}; ++#endif ++ ++#if defined(CONFIG_IMQ_NUM_DEVS) ++static unsigned int numdevs = CONFIG_IMQ_NUM_DEVS; ++#else ++static unsigned int numdevs = IMQ_MAX_DEVS; ++#endif ++ ++static DEFINE_SPINLOCK(imq_nf_queue_lock); ++ ++static struct net_device *imq_devs_cache[IMQ_MAX_DEVS]; ++ ++ ++static struct net_device_stats *imq_get_stats(struct net_device *dev) ++{ ++ return &dev->stats; ++} ++ ++/* called for packets kfree'd in qdiscs at places other than enqueue */ ++static void imq_skb_destructor(struct sk_buff *skb) ++{ ++ struct nf_queue_entry *entry = skb->nf_queue_entry; ++ ++ if (entry) { ++ nf_queue_entry_release_refs(entry); ++ kfree(entry); ++ } ++ ++ skb_restore_cb(skb); /* kfree backup */ ++} ++ ++static void imq_nf_reinject(struct nf_queue_entry *entry, unsigned int verdict) ++{ ++ int status; ++ ++ if (!entry->next_outfn) { ++ spin_lock_bh(&imq_nf_queue_lock); ++ nf_reinject(entry, verdict); ++ spin_unlock_bh(&imq_nf_queue_lock); ++ return; ++ } ++ ++ rcu_read_lock(); ++ local_bh_disable(); ++ status = entry->next_outfn(entry, entry->next_queuenum); ++ local_bh_enable(); ++ if (status < 0) { ++ nf_queue_entry_release_refs(entry); ++ kfree_skb(entry->skb); ++ kfree(entry); ++ } ++ ++ rcu_read_unlock(); ++} ++ ++static int imq_dev_xmit(struct sk_buff *skb, struct net_device *dev) ++{ ++ dev->stats.tx_bytes += skb->len; ++ dev->stats.tx_packets++; ++ ++ skb->imq_flags = 0; ++ skb->destructor = NULL; ++ ++ skb_restore_cb(skb); /* restore skb->cb */ ++ ++ dev->trans_start = jiffies; ++ imq_nf_reinject(skb->nf_queue_entry, NF_ACCEPT); ++ return 0; ++} ++ ++static int imq_nf_queue(struct nf_queue_entry *entry, unsigned queue_num) ++{ ++ struct net_device *dev; ++ struct sk_buff *skb_orig, *skb, *skb_shared; ++ struct Qdisc *q; ++ struct netdev_queue *txq; ++ int users, index; ++ int retval = -EINVAL; ++ ++ index = entry->skb->imq_flags & IMQ_F_IFMASK; ++ if (unlikely(index > numdevs - 1)) { ++ if (net_ratelimit()) ++ printk(KERN_WARNING ++ "IMQ: invalid device specified, highest is %u\n", ++ numdevs - 1); ++ retval = -EINVAL; ++ goto out; ++ } ++ ++ /* check for imq device by index from cache */ ++ dev = imq_devs_cache[index]; ++ if (unlikely(!dev)) { ++ char buf[8]; ++ ++ /* get device by name and cache result */ ++ snprintf(buf, sizeof(buf), "imq%d", index); ++ dev = dev_get_by_name(&init_net, buf); ++ if (!dev) { ++ /* not found ?!*/ ++ BUG(); ++ retval = -ENODEV; ++ goto out; ++ } ++ ++ imq_devs_cache[index] = dev; ++ dev_put(dev); ++ } ++ ++ if (unlikely(!(dev->flags & IFF_UP))) { ++ entry->skb->imq_flags = 0; ++ imq_nf_reinject(entry, NF_ACCEPT); ++ retval = 0; ++ goto out; ++ } ++ dev->last_rx = jiffies; ++ ++ skb = entry->skb; ++ skb_orig = NULL; ++ ++ /* skb has owner? => make clone */ ++ if (unlikely(skb->destructor)) { ++ skb_orig = skb; ++ skb = skb_clone(skb, GFP_ATOMIC); ++ if (!skb) { ++ retval = -ENOMEM; ++ goto out; ++ } ++ entry->skb = skb; ++ } ++ ++ skb->nf_queue_entry = entry; ++ ++ dev->stats.rx_bytes += skb->len; ++ dev->stats.rx_packets++; ++ ++ txq = dev_pick_tx(dev, skb); ++ ++ q = rcu_dereference(txq->qdisc); ++ if (unlikely(!q->enqueue)) ++ goto packet_not_eaten_by_imq_dev; ++ ++ spin_lock_bh(qdisc_lock(q)); ++ ++ users = atomic_read(&skb->users); ++ ++ skb_shared = skb_get(skb); /* increase reference count by one */ ++ skb_save_cb(skb_shared); /* backup skb->cb, as qdisc layer will ++ overwrite it */ ++ qdisc_enqueue_root(skb_shared, q); /* might kfree_skb */ ++ ++ if (likely(atomic_read(&skb_shared->users) == users + 1)) { ++ kfree_skb(skb_shared); /* decrease reference count by one */ ++ ++ skb->destructor = &imq_skb_destructor; ++ ++ /* cloned? */ ++ if (skb_orig) ++ kfree_skb(skb_orig); /* free original */ ++ ++ spin_unlock_bh(qdisc_lock(q)); ++ ++ /* schedule qdisc dequeue */ ++ __netif_schedule(q); ++ ++ retval = 0; ++ goto out; ++ } else { ++ skb_restore_cb(skb_shared); /* restore skb->cb */ ++ /* qdisc dropped packet and decreased skb reference count of ++ * skb, so we don't really want to and try refree as that would ++ * actually destroy the skb. */ ++ spin_unlock_bh(qdisc_lock(q)); ++ goto packet_not_eaten_by_imq_dev; ++ } ++ ++packet_not_eaten_by_imq_dev: ++ /* cloned? restore original */ ++ if (skb_orig) { ++ kfree_skb(skb); ++ entry->skb = skb_orig; ++ } ++ retval = -1; ++out: ++ return retval; ++} ++ ++static struct nf_queue_handler nfqh = { ++ .name = "imq", ++ .outfn = imq_nf_queue, ++}; ++ ++static unsigned int imq_nf_hook(unsigned int hook, struct sk_buff *pskb, ++ const struct net_device *indev, ++ const struct net_device *outdev, ++ int (*okfn)(struct sk_buff *)) ++{ ++ if (pskb->imq_flags & IMQ_F_ENQUEUE) ++ return NF_QUEUE; ++ ++ return NF_ACCEPT; ++} ++ ++static int imq_close(struct net_device *dev) ++{ ++ netif_stop_queue(dev); ++ return 0; ++} ++ ++static int imq_open(struct net_device *dev) ++{ ++ netif_start_queue(dev); ++ return 0; ++} ++ ++static const struct net_device_ops imq_netdev_ops = { ++ .ndo_open = imq_open, ++ .ndo_stop = imq_close, ++ .ndo_start_xmit = imq_dev_xmit, ++ .ndo_get_stats = imq_get_stats, ++}; ++ ++static void imq_setup(struct net_device *dev) ++{ ++ dev->netdev_ops = &imq_netdev_ops; ++ dev->type = ARPHRD_VOID; ++ dev->mtu = 16000; ++ dev->tx_queue_len = 11000; ++ dev->flags = IFF_NOARP; ++ dev->features = NETIF_F_SG | NETIF_F_FRAGLIST | ++ NETIF_F_GSO | NETIF_F_HW_CSUM | ++ NETIF_F_HIGHDMA; ++} ++ ++static int imq_validate(struct nlattr *tb[], struct nlattr *data[]) ++{ ++ int ret = 0; ++ ++ if (tb[IFLA_ADDRESS]) { ++ if (nla_len(tb[IFLA_ADDRESS]) != ETH_ALEN) { ++ ret = -EINVAL; ++ goto end; ++ } ++ if (!is_valid_ether_addr(nla_data(tb[IFLA_ADDRESS]))) { ++ ret = -EADDRNOTAVAIL; ++ goto end; ++ } ++ } ++ return 0; ++end: ++ printk(KERN_WARNING "IMQ: imq_validate failed (%d)\n", ret); ++ return ret; ++} ++ ++static struct rtnl_link_ops imq_link_ops __read_mostly = { ++ .kind = "imq", ++ .priv_size = 0, ++ .setup = imq_setup, ++ .validate = imq_validate, ++}; ++ ++static int __init imq_init_hooks(void) ++{ ++ int err; ++ ++ nf_register_queue_imq_handler(&nfqh); ++ ++ err = nf_register_hook(&imq_ingress_ipv4); ++ if (err) ++ goto err1; ++ ++ err = nf_register_hook(&imq_egress_ipv4); ++ if (err) ++ goto err2; ++ ++#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) ++ err = nf_register_hook(&imq_ingress_ipv6); ++ if (err) ++ goto err3; ++ ++ err = nf_register_hook(&imq_egress_ipv6); ++ if (err) ++ goto err4; ++#endif ++ ++ return 0; ++ ++#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) ++err4: ++ nf_unregister_hook(&imq_ingress_ipv6); ++err3: ++ nf_unregister_hook(&imq_egress_ipv4); ++#endif ++err2: ++ nf_unregister_hook(&imq_ingress_ipv4); ++err1: ++ nf_unregister_queue_imq_handler(); ++ return err; ++} ++ ++static int __init imq_init_one(int index) ++{ ++ struct net_device *dev; ++ int ret; ++ ++ dev = alloc_netdev(0, "imq%d", imq_setup); ++ if (!dev) ++ return -ENOMEM; ++ ++ ret = dev_alloc_name(dev, dev->name); ++ if (ret < 0) ++ goto fail; ++ ++ dev->rtnl_link_ops = &imq_link_ops; ++ ret = register_netdevice(dev); ++ if (ret < 0) ++ goto fail; ++ ++ return 0; ++fail: ++ free_netdev(dev); ++ return ret; ++} ++ ++static int __init imq_init_devs(void) ++{ ++ int err, i; ++ ++ if (numdevs < 1 || numdevs > IMQ_MAX_DEVS) { ++ printk(KERN_ERR "IMQ: numdevs has to be betweed 1 and %u\n", ++ IMQ_MAX_DEVS); ++ return -EINVAL; ++ } ++ ++ rtnl_lock(); ++ err = __rtnl_link_register(&imq_link_ops); ++ ++ for (i = 0; i < numdevs && !err; i++) ++ err = imq_init_one(i); ++ ++ if (err) { ++ __rtnl_link_unregister(&imq_link_ops); ++ memset(imq_devs_cache, 0, sizeof(imq_devs_cache)); ++ } ++ rtnl_unlock(); ++ ++ return err; ++} ++ ++static int __init imq_init_module(void) ++{ ++ int err; ++ ++#if defined(CONFIG_IMQ_NUM_DEVS) ++ BUILD_BUG_ON(CONFIG_IMQ_NUM_DEVS > 16); ++ BUILD_BUG_ON(CONFIG_IMQ_NUM_DEVS < 2); ++ BUILD_BUG_ON(CONFIG_IMQ_NUM_DEVS - 1 > IMQ_F_IFMASK); ++#endif ++ ++ err = imq_init_devs(); ++ if (err) { ++ printk(KERN_ERR "IMQ: Error trying imq_init_devs(net)\n"); ++ return err; ++ } ++ ++ err = imq_init_hooks(); ++ if (err) { ++ printk(KERN_ERR "IMQ: Error trying imq_init_hooks()\n"); ++ rtnl_link_unregister(&imq_link_ops); ++ memset(imq_devs_cache, 0, sizeof(imq_devs_cache)); ++ return err; ++ } ++ ++ printk(KERN_INFO "IMQ driver loaded successfully.\n"); ++ ++#if defined(CONFIG_IMQ_BEHAVIOR_BA) || defined(CONFIG_IMQ_BEHAVIOR_BB) ++ printk(KERN_INFO "\tHooking IMQ before NAT on PREROUTING.\n"); ++#else ++ printk(KERN_INFO "\tHooking IMQ after NAT on PREROUTING.\n"); ++#endif ++#if defined(CONFIG_IMQ_BEHAVIOR_AB) || defined(CONFIG_IMQ_BEHAVIOR_BB) ++ printk(KERN_INFO "\tHooking IMQ before NAT on POSTROUTING.\n"); ++#else ++ printk(KERN_INFO "\tHooking IMQ after NAT on POSTROUTING.\n"); ++#endif ++ ++ return 0; ++} ++ ++static void __exit imq_unhook(void) ++{ ++#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) ++ nf_unregister_hook(&imq_ingress_ipv6); ++ nf_unregister_hook(&imq_egress_ipv6); ++#endif ++ nf_unregister_hook(&imq_ingress_ipv4); ++ nf_unregister_hook(&imq_egress_ipv4); ++ ++ nf_unregister_queue_imq_handler(); ++} ++ ++static void __exit imq_cleanup_devs(void) ++{ ++ rtnl_link_unregister(&imq_link_ops); ++ memset(imq_devs_cache, 0, sizeof(imq_devs_cache)); ++} ++ ++static void __exit imq_exit_module(void) ++{ ++ imq_unhook(); ++ imq_cleanup_devs(); ++ printk(KERN_INFO "IMQ driver unloaded successfully.\n"); ++} ++ ++module_init(imq_init_module); ++module_exit(imq_exit_module); ++ ++module_param(numdevs, int, 0); ++MODULE_PARM_DESC(numdevs, "number of IMQ devices (how many imq* devices will " ++ "be created)"); ++MODULE_AUTHOR("http://www.linuximq.net"); ++MODULE_DESCRIPTION("Pseudo-driver for the intermediate queue device. See " ++ "http://www.linuximq.net/ for more information."); ++MODULE_LICENSE("GPL"); ++MODULE_ALIAS_RTNL_LINK("imq"); ++ +--- a/drivers/net/Kconfig ++++ b/drivers/net/Kconfig +@@ -109,6 +109,129 @@ config EQUALIZER + To compile this driver as a module, choose M here: the module + will be called eql. If unsure, say N. + ++config IMQ ++ tristate "IMQ (intermediate queueing device) support" ++ depends on NETDEVICES && NETFILTER ++ ---help--- ++ The IMQ device(s) is used as placeholder for QoS queueing ++ disciplines. Every packet entering/leaving the IP stack can be ++ directed through the IMQ device where it's enqueued/dequeued to the ++ attached qdisc. This allows you to treat network devices as classes ++ and distribute bandwidth among them. Iptables is used to specify ++ through which IMQ device, if any, packets travel. ++ ++ More information at: http://www.linuximq.net/ ++ ++ To compile this driver as a module, choose M here: the module ++ will be called imq. If unsure, say N. ++ ++choice ++ prompt "IMQ behavior (PRE/POSTROUTING)" ++ depends on IMQ ++ default IMQ_BEHAVIOR_AB ++ help ++ ++ This settings defines how IMQ behaves in respect to its ++ hooking in PREROUTING and POSTROUTING. ++ ++ IMQ can work in any of the following ways: ++ ++ PREROUTING | POSTROUTING ++ -----------------|------------------- ++ #1 After NAT | After NAT ++ #2 After NAT | Before NAT ++ #3 Before NAT | After NAT ++ #4 Before NAT | Before NAT ++ ++ The default behavior is to hook before NAT on PREROUTING ++ and after NAT on POSTROUTING (#3). ++ ++ This settings are specially usefull when trying to use IMQ ++ to shape NATed clients. ++ ++ More information can be found at: www.linuximq.net ++ ++ If not sure leave the default settings alone. ++ ++config IMQ_BEHAVIOR_AA ++ bool "IMQ AA" ++ help ++ This settings defines how IMQ behaves in respect to its ++ hooking in PREROUTING and POSTROUTING. ++ ++ Choosing this option will make IMQ hook like this: ++ ++ PREROUTING: After NAT ++ POSTROUTING: After NAT ++ ++ More information can be found at: www.linuximq.net ++ ++ If not sure leave the default settings alone. ++ ++config IMQ_BEHAVIOR_AB ++ bool "IMQ AB" ++ help ++ This settings defines how IMQ behaves in respect to its ++ hooking in PREROUTING and POSTROUTING. ++ ++ Choosing this option will make IMQ hook like this: ++ ++ PREROUTING: After NAT ++ POSTROUTING: Before NAT ++ ++ More information can be found at: www.linuximq.net ++ ++ If not sure leave the default settings alone. ++ ++config IMQ_BEHAVIOR_BA ++ bool "IMQ BA" ++ help ++ This settings defines how IMQ behaves in respect to its ++ hooking in PREROUTING and POSTROUTING. ++ ++ Choosing this option will make IMQ hook like this: ++ ++ PREROUTING: Before NAT ++ POSTROUTING: After NAT ++ ++ More information can be found at: www.linuximq.net ++ ++ If not sure leave the default settings alone. ++ ++config IMQ_BEHAVIOR_BB ++ bool "IMQ BB" ++ help ++ This settings defines how IMQ behaves in respect to its ++ hooking in PREROUTING and POSTROUTING. ++ ++ Choosing this option will make IMQ hook like this: ++ ++ PREROUTING: Before NAT ++ POSTROUTING: Before NAT ++ ++ More information can be found at: www.linuximq.net ++ ++ If not sure leave the default settings alone. ++ ++endchoice ++ ++config IMQ_NUM_DEVS ++ ++ int "Number of IMQ devices" ++ range 2 16 ++ depends on IMQ ++ default "16" ++ help ++ ++ This settings defines how many IMQ devices will be ++ created. ++ ++ The default value is 16. ++ ++ More information can be found at: www.linuximq.net ++ ++ If not sure leave the default settings alone. ++ + config TUN + tristate "Universal TUN/TAP device driver support" + select CRC32 +--- a/drivers/net/Makefile ++++ b/drivers/net/Makefile +@@ -160,6 +160,7 @@ obj-$(CONFIG_SLHC) += slhc.o + obj-$(CONFIG_XEN_NETDEV_FRONTEND) += xen-netfront.o + + obj-$(CONFIG_DUMMY) += dummy.o ++obj-$(CONFIG_IMQ) += imq.o + obj-$(CONFIG_IFB) += ifb.o + obj-$(CONFIG_MACVLAN) += macvlan.o + obj-$(CONFIG_DE600) += de600.o +--- /dev/null ++++ b/include/linux/imq.h +@@ -0,0 +1,13 @@ ++#ifndef _IMQ_H ++#define _IMQ_H ++ ++/* IFMASK (16 device indexes, 0 to 15) and flag(s) fit in 5 bits */ ++#define IMQ_F_BITS 5 ++ ++#define IMQ_F_IFMASK 0x0f ++#define IMQ_F_ENQUEUE 0x10 ++ ++#define IMQ_MAX_DEVS (IMQ_F_IFMASK + 1) ++ ++#endif /* _IMQ_H */ ++ +--- /dev/null ++++ b/include/linux/netfilter_ipv4/ipt_IMQ.h +@@ -0,0 +1,10 @@ ++#ifndef _IPT_IMQ_H ++#define _IPT_IMQ_H ++ ++/* Backwards compatibility for old userspace */ ++#include ++ ++#define ipt_imq_info xt_imq_info ++ ++#endif /* _IPT_IMQ_H */ ++ +--- /dev/null ++++ b/include/linux/netfilter_ipv6/ip6t_IMQ.h +@@ -0,0 +1,10 @@ ++#ifndef _IP6T_IMQ_H ++#define _IP6T_IMQ_H ++ ++/* Backwards compatibility for old userspace */ ++#include ++ ++#define ip6t_imq_info xt_imq_info ++ ++#endif /* _IP6T_IMQ_H */ ++ +--- a/include/linux/skbuff.h ++++ b/include/linux/skbuff.h +@@ -29,6 +29,9 @@ + #include + #include + #include ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++#include ++#endif + + /* Don't change this without changing skb_csum_unnecessary! */ + #define CHECKSUM_NONE 0 +@@ -331,6 +334,9 @@ struct sk_buff { + * first. This is owned by whoever has the skb queued ATM. + */ + char cb[48]; ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ void *cb_next; ++#endif + + unsigned int len, + data_len; +@@ -363,6 +369,9 @@ struct sk_buff { + struct nf_conntrack *nfct; + struct sk_buff *nfct_reasm; + #endif ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ struct nf_queue_entry *nf_queue_entry; ++#endif + #ifdef CONFIG_BRIDGE_NETFILTER + struct nf_bridge_info *nf_bridge; + #endif +@@ -386,6 +395,9 @@ struct sk_buff { + kmemcheck_bitfield_end(flags2); + + /* 0/13/14 bit hole */ ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ __u8 imq_flags:IMQ_F_BITS; ++#endif + + #ifdef CONFIG_NET_DMA + dma_cookie_t dma_cookie; +@@ -441,6 +453,12 @@ static inline struct rtable *skb_rtable( + return (struct rtable *)skb_dst(skb); + } + ++ ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++extern int skb_save_cb(struct sk_buff *skb); ++extern int skb_restore_cb(struct sk_buff *skb); ++#endif ++ + extern void kfree_skb(struct sk_buff *skb); + extern void consume_skb(struct sk_buff *skb); + extern void __kfree_skb(struct sk_buff *skb); +@@ -1974,6 +1992,10 @@ static inline void __nf_copy(struct sk_b + dst->nfct_reasm = src->nfct_reasm; + nf_conntrack_get_reasm(src->nfct_reasm); + #endif ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ dst->imq_flags = src->imq_flags; ++ dst->nf_queue_entry = src->nf_queue_entry; ++#endif + #ifdef CONFIG_BRIDGE_NETFILTER + dst->nf_bridge = src->nf_bridge; + nf_bridge_get(src->nf_bridge); +--- a/net/core/dev.c ++++ b/net/core/dev.c +@@ -96,6 +96,9 @@ + #include + #include + #include ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++#include ++#endif + #include + #include + #include +@@ -1686,7 +1689,11 @@ int dev_hard_start_xmit(struct sk_buff * + int rc; + + if (likely(!skb->next)) { +- if (!list_empty(&ptype_all)) ++ if (!list_empty(&ptype_all) ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ && !(skb->imq_flags & IMQ_F_ENQUEUE) ++#endif ++ ) + dev_queue_xmit_nit(skb, dev); + + if (netif_needs_gso(dev, skb)) { +@@ -1771,8 +1778,7 @@ u16 skb_tx_hash(const struct net_device + } + EXPORT_SYMBOL(skb_tx_hash); + +-static struct netdev_queue *dev_pick_tx(struct net_device *dev, +- struct sk_buff *skb) ++struct netdev_queue *dev_pick_tx(struct net_device *dev, struct sk_buff *skb) + { + const struct net_device_ops *ops = dev->netdev_ops; + u16 queue_index = 0; +@@ -1785,6 +1791,7 @@ static struct netdev_queue *dev_pick_tx( + skb_set_queue_mapping(skb, queue_index); + return netdev_get_tx_queue(dev, queue_index); + } ++EXPORT_SYMBOL(dev_pick_tx); + + /** + * dev_queue_xmit - transmit a buffer +--- a/include/linux/netdevice.h ++++ b/include/linux/netdevice.h +@@ -1102,6 +1102,7 @@ extern int dev_alloc_name(struct net_de + extern int dev_open(struct net_device *dev); + extern int dev_close(struct net_device *dev); + extern void dev_disable_lro(struct net_device *dev); ++extern struct netdev_queue *dev_pick_tx(struct net_device *dev, struct sk_buff *skb); + extern int dev_queue_xmit(struct sk_buff *skb); + extern int register_netdevice(struct net_device *dev); + extern void unregister_netdevice(struct net_device *dev); +--- /dev/null ++++ b/include/linux/netfilter/xt_IMQ.h +@@ -0,0 +1,9 @@ ++#ifndef _XT_IMQ_H ++#define _XT_IMQ_H ++ ++struct xt_imq_info { ++ unsigned int todev; /* target imq device */ ++}; ++ ++#endif /* _XT_IMQ_H */ ++ +--- a/include/net/netfilter/nf_queue.h ++++ b/include/net/netfilter/nf_queue.h +@@ -13,6 +13,12 @@ struct nf_queue_entry { + struct net_device *indev; + struct net_device *outdev; + int (*okfn)(struct sk_buff *); ++ ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ int (*next_outfn)(struct nf_queue_entry *entry, ++ unsigned int queuenum); ++ unsigned int next_queuenum; ++#endif + }; + + #define nf_queue_entry_reroute(x) ((void *)x + sizeof(struct nf_queue_entry)) +@@ -30,5 +36,11 @@ extern int nf_unregister_queue_handler(u + const struct nf_queue_handler *qh); + extern void nf_unregister_queue_handlers(const struct nf_queue_handler *qh); + extern void nf_reinject(struct nf_queue_entry *entry, unsigned int verdict); ++extern void nf_queue_entry_release_refs(struct nf_queue_entry *entry); ++ ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++extern void nf_register_queue_imq_handler(const struct nf_queue_handler *qh); ++extern void nf_unregister_queue_imq_handler(void); ++#endif + + #endif /* _NF_QUEUE_H */ +--- a/net/core/skbuff.c ++++ b/net/core/skbuff.c +@@ -72,6 +72,9 @@ + + static struct kmem_cache *skbuff_head_cache __read_mostly; + static struct kmem_cache *skbuff_fclone_cache __read_mostly; ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++static struct kmem_cache *skbuff_cb_store_cache __read_mostly; ++#endif + + static void sock_pipe_buf_release(struct pipe_inode_info *pipe, + struct pipe_buffer *buf) +@@ -91,6 +94,80 @@ static int sock_pipe_buf_steal(struct pi + return 1; + } + ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++/* Control buffer save/restore for IMQ devices */ ++struct skb_cb_table { ++ void *cb_next; ++ atomic_t refcnt; ++ char cb[48]; ++}; ++ ++static DEFINE_SPINLOCK(skb_cb_store_lock); ++ ++int skb_save_cb(struct sk_buff *skb) ++{ ++ struct skb_cb_table *next; ++ ++ next = kmem_cache_alloc(skbuff_cb_store_cache, GFP_ATOMIC); ++ if (!next) ++ return -ENOMEM; ++ ++ BUILD_BUG_ON(sizeof(skb->cb) != sizeof(next->cb)); ++ ++ memcpy(next->cb, skb->cb, sizeof(skb->cb)); ++ next->cb_next = skb->cb_next; ++ ++ atomic_set(&next->refcnt, 1); ++ ++ skb->cb_next = next; ++ return 0; ++} ++EXPORT_SYMBOL(skb_save_cb); ++ ++int skb_restore_cb(struct sk_buff *skb) ++{ ++ struct skb_cb_table *next; ++ ++ if (!skb->cb_next) ++ return 0; ++ ++ next = skb->cb_next; ++ ++ BUILD_BUG_ON(sizeof(skb->cb) != sizeof(next->cb)); ++ ++ memcpy(skb->cb, next->cb, sizeof(skb->cb)); ++ skb->cb_next = next->cb_next; ++ ++ spin_lock(&skb_cb_store_lock); ++ ++ if (atomic_dec_and_test(&next->refcnt)) { ++ kmem_cache_free(skbuff_cb_store_cache, next); ++ } ++ ++ spin_unlock(&skb_cb_store_lock); ++ ++ return 0; ++} ++EXPORT_SYMBOL(skb_restore_cb); ++ ++static void skb_copy_stored_cb(struct sk_buff *new, struct sk_buff *old) ++{ ++ struct skb_cb_table *next; ++ ++ if (!old->cb_next) { ++ new->cb_next = 0; ++ return; ++ } ++ ++ spin_lock(&skb_cb_store_lock); ++ ++ next = old->cb_next; ++ atomic_inc(&next->refcnt); ++ new->cb_next = next; ++ ++ spin_unlock(&skb_cb_store_lock); ++} ++#endif + + /* Pipe buffer operations for a socket. */ + static struct pipe_buf_operations sock_pipe_buf_ops = { +@@ -398,6 +475,15 @@ static void skb_release_head_state(struc + WARN_ON(in_irq()); + skb->destructor(skb); + } ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ /* This should not happen. When it does, avoid memleak by restoring ++ the chain of cb-backups. */ ++ while(skb->cb_next != NULL) { ++ printk(KERN_WARNING "kfree_skb: skb->cb_next: %08x\n", ++ skb->cb_next); ++ skb_restore_cb(skb); ++ } ++#endif + #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE) + nf_conntrack_put(skb->nfct); + nf_conntrack_put_reasm(skb->nfct_reasm); +@@ -535,6 +621,9 @@ static void __copy_skb_header(struct sk_ + new->sp = secpath_get(old->sp); + #endif + memcpy(new->cb, old->cb, sizeof(old->cb)); ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ skb_copy_stored_cb(new, old); ++#endif + new->csum = old->csum; + new->local_df = old->local_df; + new->pkt_type = old->pkt_type; +@@ -2778,6 +2867,13 @@ void __init skb_init(void) + 0, + SLAB_HWCACHE_ALIGN|SLAB_PANIC, + NULL); ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ skbuff_cb_store_cache = kmem_cache_create("skbuff_cb_store_cache", ++ sizeof(struct skb_cb_table), ++ 0, ++ SLAB_HWCACHE_ALIGN|SLAB_PANIC, ++ NULL); ++#endif + } + + /** +--- a/net/netfilter/Kconfig ++++ b/net/netfilter/Kconfig +@@ -396,6 +396,18 @@ config NETFILTER_XT_TARGET_LED + For more information on the LEDs available on your system, see + Documentation/leds-class.txt + ++config NETFILTER_XT_TARGET_IMQ ++ tristate '"IMQ" target support' ++ depends on NETFILTER_XTABLES ++ depends on IP_NF_MANGLE || IP6_NF_MANGLE ++ select IMQ ++ default m if NETFILTER_ADVANCED=n ++ help ++ This option adds a `IMQ' target which is used to specify if and ++ to which imq device packets should get enqueued/dequeued. ++ ++ To compile it as a module, choose M here. If unsure, say N. ++ + config NETFILTER_XT_TARGET_MARK + tristate '"MARK" target support' + default m if NETFILTER_ADVANCED=n +--- a/net/netfilter/Makefile ++++ b/net/netfilter/Makefile +@@ -46,6 +46,7 @@ obj-$(CONFIG_NETFILTER_XT_TARGET_CONNMAR + obj-$(CONFIG_NETFILTER_XT_TARGET_CONNSECMARK) += xt_CONNSECMARK.o + obj-$(CONFIG_NETFILTER_XT_TARGET_DSCP) += xt_DSCP.o + obj-$(CONFIG_NETFILTER_XT_TARGET_HL) += xt_HL.o ++obj-$(CONFIG_NETFILTER_XT_TARGET_IMQ) += xt_IMQ.o + obj-$(CONFIG_NETFILTER_XT_TARGET_LED) += xt_LED.o + obj-$(CONFIG_NETFILTER_XT_TARGET_MARK) += xt_MARK.o + obj-$(CONFIG_NETFILTER_XT_TARGET_NFLOG) += xt_NFLOG.o +--- a/net/netfilter/nf_queue.c ++++ b/net/netfilter/nf_queue.c +@@ -20,6 +20,26 @@ static const struct nf_queue_handler *qu + + static DEFINE_MUTEX(queue_handler_mutex); + ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++static const struct nf_queue_handler *queue_imq_handler; ++ ++void nf_register_queue_imq_handler(const struct nf_queue_handler *qh) ++{ ++ mutex_lock(&queue_handler_mutex); ++ rcu_assign_pointer(queue_imq_handler, qh); ++ mutex_unlock(&queue_handler_mutex); ++} ++EXPORT_SYMBOL(nf_register_queue_imq_handler); ++ ++void nf_unregister_queue_imq_handler(void) ++{ ++ mutex_lock(&queue_handler_mutex); ++ rcu_assign_pointer(queue_imq_handler, NULL); ++ mutex_unlock(&queue_handler_mutex); ++} ++EXPORT_SYMBOL(nf_unregister_queue_imq_handler); ++#endif ++ + /* return EBUSY when somebody else is registered, return EEXIST if the + * same handler is registered, return 0 in case of success. */ + int nf_register_queue_handler(u_int8_t pf, const struct nf_queue_handler *qh) +@@ -80,7 +100,7 @@ void nf_unregister_queue_handlers(const + } + EXPORT_SYMBOL_GPL(nf_unregister_queue_handlers); + +-static void nf_queue_entry_release_refs(struct nf_queue_entry *entry) ++void nf_queue_entry_release_refs(struct nf_queue_entry *entry) + { + /* Release those devices we held, or Alexey will kill me. */ + if (entry->indev) +@@ -100,6 +120,7 @@ static void nf_queue_entry_release_refs( + /* Drop reference to owner of hook which queued us. */ + module_put(entry->elem->owner); + } ++EXPORT_SYMBOL_GPL(nf_queue_entry_release_refs); + + /* + * Any packet that leaves via this function must come back +@@ -121,12 +142,26 @@ static int __nf_queue(struct sk_buff *sk + #endif + const struct nf_afinfo *afinfo; + const struct nf_queue_handler *qh; ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ const struct nf_queue_handler *qih = NULL; ++#endif + + /* QUEUE == DROP if noone is waiting, to be safe. */ + rcu_read_lock(); + + qh = rcu_dereference(queue_handler[pf]); ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) ++ if (pf == PF_INET || pf == PF_INET6) ++#else ++ if (pf == PF_INET) ++#endif ++ qih = rcu_dereference(queue_imq_handler); ++ ++ if (!qh && !qih) ++#else /* !IMQ */ + if (!qh) ++#endif + goto err_unlock; + + afinfo = nf_get_afinfo(pf); +@@ -145,6 +180,10 @@ static int __nf_queue(struct sk_buff *sk + .indev = indev, + .outdev = outdev, + .okfn = okfn, ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ .next_outfn = qh ? qh->outfn : NULL, ++ .next_queuenum = queuenum, ++#endif + }; + + /* If it's going away, ignore hook. */ +@@ -170,8 +209,19 @@ static int __nf_queue(struct sk_buff *sk + } + #endif + afinfo->saveroute(skb, entry); ++ ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++ if (qih) { ++ status = qih->outfn(entry, queuenum); ++ goto imq_skip_queue; ++ } ++#endif ++ + status = qh->outfn(entry, queuenum); + ++#if defined(CONFIG_IMQ) || defined(CONFIG_IMQ_MODULE) ++imq_skip_queue: ++#endif + rcu_read_unlock(); + + if (status < 0) { +--- /dev/null ++++ b/net/netfilter/xt_IMQ.c +@@ -0,0 +1,73 @@ ++/* ++ * This target marks packets to be enqueued to an imq device ++ */ ++#include ++#include ++#include ++#include ++#include ++ ++static unsigned int imq_target(struct sk_buff *pskb, ++ const struct xt_target_param *par) ++{ ++ const struct xt_imq_info *mr = par->targinfo; ++ ++ pskb->imq_flags = (mr->todev & IMQ_F_IFMASK) | IMQ_F_ENQUEUE; ++ ++ return XT_CONTINUE; ++} ++ ++static bool imq_checkentry(const struct xt_tgchk_param *par) ++{ ++ struct xt_imq_info *mr = par->targinfo; ++ ++ if (mr->todev > IMQ_MAX_DEVS - 1) { ++ printk(KERN_WARNING ++ "IMQ: invalid device specified, highest is %u\n", ++ IMQ_MAX_DEVS - 1); ++ return 0; ++ } ++ ++ return 1; ++} ++ ++static struct xt_target xt_imq_reg[] __read_mostly = { ++ { ++ .name = "IMQ", ++ .family = AF_INET, ++ .checkentry = imq_checkentry, ++ .target = imq_target, ++ .targetsize = sizeof(struct xt_imq_info), ++ .table = "mangle", ++ .me = THIS_MODULE ++ }, ++ { ++ .name = "IMQ", ++ .family = AF_INET6, ++ .checkentry = imq_checkentry, ++ .target = imq_target, ++ .targetsize = sizeof(struct xt_imq_info), ++ .table = "mangle", ++ .me = THIS_MODULE ++ }, ++}; ++ ++static int __init imq_init(void) ++{ ++ return xt_register_targets(xt_imq_reg, ARRAY_SIZE(xt_imq_reg)); ++} ++ ++static void __exit imq_fini(void) ++{ ++ xt_unregister_targets(xt_imq_reg, ARRAY_SIZE(xt_imq_reg)); ++} ++ ++module_init(imq_init); ++module_exit(imq_fini); ++ ++MODULE_AUTHOR("http://www.linuximq.net"); ++MODULE_DESCRIPTION("Pseudo-driver for the intermediate queue device. See http://www.linuximq.net/ for more information."); ++MODULE_LICENSE("GPL"); ++MODULE_ALIAS("ipt_IMQ"); ++MODULE_ALIAS("ip6t_IMQ"); ++ diff --git a/target/linux/generic-2.6/patches-2.6.31/180-netfilter_depends.patch b/target/linux/generic-2.6/patches-2.6.31/180-netfilter_depends.patch new file mode 100644 index 0000000000..fc00d159c3 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/180-netfilter_depends.patch @@ -0,0 +1,18 @@ +--- a/net/netfilter/Kconfig ++++ b/net/netfilter/Kconfig +@@ -160,7 +160,6 @@ config NF_CONNTRACK_FTP + + config NF_CONNTRACK_H323 + tristate "H.323 protocol support" +- depends on (IPV6 || IPV6=n) + depends on NETFILTER_ADVANCED + help + H.323 is a VoIP signalling protocol from ITU-T. As one of the most +@@ -505,7 +504,6 @@ config NETFILTER_XT_TARGET_SECMARK + + config NETFILTER_XT_TARGET_TCPMSS + tristate '"TCPMSS" target support' +- depends on (IPV6 || IPV6=n) + default m if NETFILTER_ADVANCED=n + ---help--- + This option adds a `TCPMSS' target, which allows you to alter the diff --git a/target/linux/generic-2.6/patches-2.6.31/190-netfilter_rtsp.patch b/target/linux/generic-2.6/patches-2.6.31/190-netfilter_rtsp.patch new file mode 100644 index 0000000000..29e82084ca --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/190-netfilter_rtsp.patch @@ -0,0 +1,1366 @@ +--- /dev/null ++++ b/include/linux/netfilter/nf_conntrack_rtsp.h +@@ -0,0 +1,63 @@ ++/* ++ * RTSP extension for IP connection tracking. ++ * (C) 2003 by Tom Marshall ++ * based on ip_conntrack_irc.h ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ */ ++#ifndef _IP_CONNTRACK_RTSP_H ++#define _IP_CONNTRACK_RTSP_H ++ ++//#define IP_NF_RTSP_DEBUG 1 ++#define IP_NF_RTSP_VERSION "0.6.21" ++ ++#ifdef __KERNEL__ ++/* port block types */ ++typedef enum { ++ pb_single, /* client_port=x */ ++ pb_range, /* client_port=x-y */ ++ pb_discon /* client_port=x/y (rtspbis) */ ++} portblock_t; ++ ++/* We record seq number and length of rtsp headers here, all in host order. */ ++ ++/* ++ * This structure is per expected connection. It is a member of struct ++ * ip_conntrack_expect. The TCP SEQ for the conntrack expect is stored ++ * there and we are expected to only store the length of the data which ++ * needs replaced. If a packet contains multiple RTSP messages, we create ++ * one expected connection per message. ++ * ++ * We use these variables to mark the entire header block. This may seem ++ * like overkill, but the nature of RTSP requires it. A header may appear ++ * multiple times in a message. We must treat two Transport headers the ++ * same as one Transport header with two entries. ++ */ ++struct ip_ct_rtsp_expect ++{ ++ u_int32_t len; /* length of header block */ ++ portblock_t pbtype; /* Type of port block that was requested */ ++ u_int16_t loport; /* Port that was requested, low or first */ ++ u_int16_t hiport; /* Port that was requested, high or second */ ++#if 0 ++ uint method; /* RTSP method */ ++ uint cseq; /* CSeq from request */ ++#endif ++}; ++ ++extern unsigned int (*nf_nat_rtsp_hook)(struct sk_buff *skb, ++ enum ip_conntrack_info ctinfo, ++ unsigned int matchoff, unsigned int matchlen, ++ struct ip_ct_rtsp_expect *prtspexp, ++ struct nf_conntrack_expect *exp); ++ ++extern void (*nf_nat_rtsp_hook_expectfn)(struct nf_conn *ct, struct nf_conntrack_expect *exp); ++ ++#define RTSP_PORT 554 ++ ++#endif /* __KERNEL__ */ ++ ++#endif /* _IP_CONNTRACK_RTSP_H */ +--- /dev/null ++++ b/include/linux/netfilter_helpers.h +@@ -0,0 +1,133 @@ ++/* ++ * Helpers for netfiler modules. This file provides implementations for basic ++ * functions such as strncasecmp(), etc. ++ * ++ * gcc will warn for defined but unused functions, so we only include the ++ * functions requested. The following macros are used: ++ * NF_NEED_STRNCASECMP nf_strncasecmp() ++ * NF_NEED_STRTOU16 nf_strtou16() ++ * NF_NEED_STRTOU32 nf_strtou32() ++ */ ++#ifndef _NETFILTER_HELPERS_H ++#define _NETFILTER_HELPERS_H ++ ++/* Only include these functions for kernel code. */ ++#ifdef __KERNEL__ ++ ++#include ++#define iseol(c) ( (c) == '\r' || (c) == '\n' ) ++ ++/* ++ * The standard strncasecmp() ++ */ ++#ifdef NF_NEED_STRNCASECMP ++static int ++nf_strncasecmp(const char* s1, const char* s2, u_int32_t len) ++{ ++ if (s1 == NULL || s2 == NULL) ++ { ++ if (s1 == NULL && s2 == NULL) ++ { ++ return 0; ++ } ++ return (s1 == NULL) ? -1 : 1; ++ } ++ while (len > 0 && tolower(*s1) == tolower(*s2)) ++ { ++ len--; ++ s1++; ++ s2++; ++ } ++ return ( (len == 0) ? 0 : (tolower(*s1) - tolower(*s2)) ); ++} ++#endif /* NF_NEED_STRNCASECMP */ ++ ++/* ++ * Parse a string containing a 16-bit unsigned integer. ++ * Returns the number of chars used, or zero if no number is found. ++ */ ++#ifdef NF_NEED_STRTOU16 ++static int ++nf_strtou16(const char* pbuf, u_int16_t* pval) ++{ ++ int n = 0; ++ ++ *pval = 0; ++ while (isdigit(pbuf[n])) ++ { ++ *pval = (*pval * 10) + (pbuf[n] - '0'); ++ n++; ++ } ++ ++ return n; ++} ++#endif /* NF_NEED_STRTOU16 */ ++ ++/* ++ * Parse a string containing a 32-bit unsigned integer. ++ * Returns the number of chars used, or zero if no number is found. ++ */ ++#ifdef NF_NEED_STRTOU32 ++static int ++nf_strtou32(const char* pbuf, u_int32_t* pval) ++{ ++ int n = 0; ++ ++ *pval = 0; ++ while (pbuf[n] >= '0' && pbuf[n] <= '9') ++ { ++ *pval = (*pval * 10) + (pbuf[n] - '0'); ++ n++; ++ } ++ ++ return n; ++} ++#endif /* NF_NEED_STRTOU32 */ ++ ++/* ++ * Given a buffer and length, advance to the next line and mark the current ++ * line. ++ */ ++#ifdef NF_NEED_NEXTLINE ++static int ++nf_nextline(char* p, uint len, uint* poff, uint* plineoff, uint* plinelen) ++{ ++ uint off = *poff; ++ uint physlen = 0; ++ ++ if (off >= len) ++ { ++ return 0; ++ } ++ ++ while (p[off] != '\n') ++ { ++ if (len-off <= 1) ++ { ++ return 0; ++ } ++ ++ physlen++; ++ off++; ++ } ++ ++ /* if we saw a crlf, physlen needs adjusted */ ++ if (physlen > 0 && p[off] == '\n' && p[off-1] == '\r') ++ { ++ physlen--; ++ } ++ ++ /* advance past the newline */ ++ off++; ++ ++ *plineoff = *poff; ++ *plinelen = physlen; ++ *poff = off; ++ ++ return 1; ++} ++#endif /* NF_NEED_NEXTLINE */ ++ ++#endif /* __KERNEL__ */ ++ ++#endif /* _NETFILTER_HELPERS_H */ +--- /dev/null ++++ b/include/linux/netfilter_mime.h +@@ -0,0 +1,89 @@ ++/* ++ * MIME functions for netfilter modules. This file provides implementations ++ * for basic MIME parsing. MIME headers are used in many protocols, such as ++ * HTTP, RTSP, SIP, etc. ++ * ++ * gcc will warn for defined but unused functions, so we only include the ++ * functions requested. The following macros are used: ++ * NF_NEED_MIME_NEXTLINE nf_mime_nextline() ++ */ ++#ifndef _NETFILTER_MIME_H ++#define _NETFILTER_MIME_H ++ ++/* Only include these functions for kernel code. */ ++#ifdef __KERNEL__ ++ ++#include ++ ++/* ++ * Given a buffer and length, advance to the next line and mark the current ++ * line. If the current line is empty, *plinelen will be set to zero. If ++ * not, it will be set to the actual line length (including CRLF). ++ * ++ * 'line' in this context means logical line (includes LWS continuations). ++ * Returns 1 on success, 0 on failure. ++ */ ++#ifdef NF_NEED_MIME_NEXTLINE ++static int ++nf_mime_nextline(char* p, uint len, uint* poff, uint* plineoff, uint* plinelen) ++{ ++ uint off = *poff; ++ uint physlen = 0; ++ int is_first_line = 1; ++ ++ if (off >= len) ++ { ++ return 0; ++ } ++ ++ do ++ { ++ while (p[off] != '\n') ++ { ++ if (len-off <= 1) ++ { ++ return 0; ++ } ++ ++ physlen++; ++ off++; ++ } ++ ++ /* if we saw a crlf, physlen needs adjusted */ ++ if (physlen > 0 && p[off] == '\n' && p[off-1] == '\r') ++ { ++ physlen--; ++ } ++ ++ /* advance past the newline */ ++ off++; ++ ++ /* check for an empty line */ ++ if (physlen == 0) ++ { ++ break; ++ } ++ ++ /* check for colon on the first physical line */ ++ if (is_first_line) ++ { ++ is_first_line = 0; ++ if (memchr(p+(*poff), ':', physlen) == NULL) ++ { ++ return 0; ++ } ++ } ++ } ++ while (p[off] == ' ' || p[off] == '\t'); ++ ++ *plineoff = *poff; ++ *plinelen = (physlen == 0) ? 0 : (off - *poff); ++ *poff = off; ++ ++ return 1; ++} ++#endif /* NF_NEED_MIME_NEXTLINE */ ++ ++#endif /* __KERNEL__ */ ++ ++#endif /* _NETFILTER_MIME_H */ +--- a/net/ipv4/netfilter/Makefile ++++ b/net/ipv4/netfilter/Makefile +@@ -26,6 +26,7 @@ obj-$(CONFIG_NF_NAT_AMANDA) += nf_nat_am + obj-$(CONFIG_NF_NAT_FTP) += nf_nat_ftp.o + obj-$(CONFIG_NF_NAT_H323) += nf_nat_h323.o + obj-$(CONFIG_NF_NAT_IRC) += nf_nat_irc.o ++obj-$(CONFIG_NF_NAT_RTSP) += nf_nat_rtsp.o + obj-$(CONFIG_NF_NAT_PPTP) += nf_nat_pptp.o + obj-$(CONFIG_NF_NAT_SIP) += nf_nat_sip.o + obj-$(CONFIG_NF_NAT_SNMP_BASIC) += nf_nat_snmp_basic.o +--- a/net/netfilter/Kconfig ++++ b/net/netfilter/Kconfig +@@ -267,6 +267,16 @@ config NF_CONNTRACK_TFTP + + To compile it as a module, choose M here. If unsure, say N. + ++config NF_CONNTRACK_RTSP ++ tristate "RTSP protocol support" ++ depends on NF_CONNTRACK ++ help ++ Support the RTSP protocol. This allows UDP transports to be setup ++ properly, including RTP and RDT. ++ ++ If you want to compile it as a module, say 'M' here and read ++ Documentation/modules.txt. If unsure, say 'Y'. ++ + config NF_CT_NETLINK + tristate 'Connection tracking netlink interface' + select NETFILTER_NETLINK +--- a/net/netfilter/Makefile ++++ b/net/netfilter/Makefile +@@ -33,6 +33,7 @@ obj-$(CONFIG_NF_CONNTRACK_PPTP) += nf_co + obj-$(CONFIG_NF_CONNTRACK_SANE) += nf_conntrack_sane.o + obj-$(CONFIG_NF_CONNTRACK_SIP) += nf_conntrack_sip.o + obj-$(CONFIG_NF_CONNTRACK_TFTP) += nf_conntrack_tftp.o ++obj-$(CONFIG_NF_CONNTRACK_RTSP) += nf_conntrack_rtsp.o + + # transparent proxy support + obj-$(CONFIG_NETFILTER_TPROXY) += nf_tproxy_core.o +--- a/net/ipv4/netfilter/Kconfig ++++ b/net/ipv4/netfilter/Kconfig +@@ -257,6 +257,11 @@ config NF_NAT_IRC + depends on NF_CONNTRACK && NF_NAT + default NF_NAT && NF_CONNTRACK_IRC + ++config NF_NAT_RTSP ++ tristate ++ depends on IP_NF_IPTABLES && NF_CONNTRACK && NF_NAT ++ default NF_NAT && NF_CONNTRACK_RTSP ++ + config NF_NAT_TFTP + tristate + depends on NF_CONNTRACK && NF_NAT +--- /dev/null ++++ b/net/netfilter/nf_conntrack_rtsp.c +@@ -0,0 +1,517 @@ ++/* ++ * RTSP extension for IP connection tracking ++ * (C) 2003 by Tom Marshall ++ * based on ip_conntrack_irc.c ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ * ++ * Module load syntax: ++ * insmod nf_conntrack_rtsp.o ports=port1,port2,...port ++ * max_outstanding=n setup_timeout=secs ++ * ++ * If no ports are specified, the default will be port 554. ++ * ++ * With max_outstanding you can define the maximum number of not yet ++ * answered SETUP requests per RTSP session (default 8). ++ * With setup_timeout you can specify how long the system waits for ++ * an expected data channel (default 300 seconds). ++ * ++ * 2005-02-13: Harald Welte ++ * - port to 2.6 ++ * - update to recent post-2.6.11 api changes ++ * 2006-09-14: Steven Van Acker ++ * - removed calls to NAT code from conntrack helper: NAT no longer needed to use rtsp-conntrack ++ * 2007-04-18: Michael Guntsche ++ * - Port to new NF API ++ */ ++ ++#include ++#include ++#include ++#include ++#include ++ ++#include ++#include ++#include ++#include ++ ++#define NF_NEED_STRNCASECMP ++#define NF_NEED_STRTOU16 ++#define NF_NEED_STRTOU32 ++#define NF_NEED_NEXTLINE ++#include ++#define NF_NEED_MIME_NEXTLINE ++#include ++ ++#include ++#define MAX_SIMUL_SETUP 8 /* XXX: use max_outstanding */ ++#define INFOP(fmt, args...) printk(KERN_INFO "%s: %s: " fmt, __FILE__, __FUNCTION__ , ## args) ++#if 0 ++#define DEBUGP(fmt, args...) printk(KERN_DEBUG "%s: %s: " fmt, __FILE__, __FUNCTION__ , ## args) ++#else ++#define DEBUGP(fmt, args...) ++#endif ++ ++#define MAX_PORTS 8 ++static int ports[MAX_PORTS]; ++static int num_ports = 0; ++static int max_outstanding = 8; ++static unsigned int setup_timeout = 300; ++ ++MODULE_AUTHOR("Tom Marshall "); ++MODULE_DESCRIPTION("RTSP connection tracking module"); ++MODULE_LICENSE("GPL"); ++module_param_array(ports, int, &num_ports, 0400); ++MODULE_PARM_DESC(ports, "port numbers of RTSP servers"); ++module_param(max_outstanding, int, 0400); ++MODULE_PARM_DESC(max_outstanding, "max number of outstanding SETUP requests per RTSP session"); ++module_param(setup_timeout, int, 0400); ++MODULE_PARM_DESC(setup_timeout, "timeout on for unestablished data channels"); ++ ++static char *rtsp_buffer; ++static DEFINE_SPINLOCK(rtsp_buffer_lock); ++ ++unsigned int (*nf_nat_rtsp_hook)(struct sk_buff *skb, ++ enum ip_conntrack_info ctinfo, ++ unsigned int matchoff, unsigned int matchlen,struct ip_ct_rtsp_expect* prtspexp, ++ struct nf_conntrack_expect *exp); ++void (*nf_nat_rtsp_hook_expectfn)(struct nf_conn *ct, struct nf_conntrack_expect *exp); ++ ++EXPORT_SYMBOL_GPL(nf_nat_rtsp_hook); ++ ++/* ++ * Max mappings we will allow for one RTSP connection (for RTP, the number ++ * of allocated ports is twice this value). Note that SMIL burns a lot of ++ * ports so keep this reasonably high. If this is too low, you will see a ++ * lot of "no free client map entries" messages. ++ */ ++#define MAX_PORT_MAPS 16 ++ ++/*** default port list was here in the masq code: 554, 3030, 4040 ***/ ++ ++#define SKIP_WSPACE(ptr,len,off) while(off < len && isspace(*(ptr+off))) { off++; } ++ ++/* ++ * Parse an RTSP packet. ++ * ++ * Returns zero if parsing failed. ++ * ++ * Parameters: ++ * IN ptcp tcp data pointer ++ * IN tcplen tcp data len ++ * IN/OUT ptcpoff points to current tcp offset ++ * OUT phdrsoff set to offset of rtsp headers ++ * OUT phdrslen set to length of rtsp headers ++ * OUT pcseqoff set to offset of CSeq header ++ * OUT pcseqlen set to length of CSeq header ++ */ ++static int ++rtsp_parse_message(char* ptcp, uint tcplen, uint* ptcpoff, ++ uint* phdrsoff, uint* phdrslen, ++ uint* pcseqoff, uint* pcseqlen, ++ uint* transoff, uint* translen) ++{ ++ uint entitylen = 0; ++ uint lineoff; ++ uint linelen; ++ ++ if (!nf_nextline(ptcp, tcplen, ptcpoff, &lineoff, &linelen)) ++ return 0; ++ ++ *phdrsoff = *ptcpoff; ++ while (nf_mime_nextline(ptcp, tcplen, ptcpoff, &lineoff, &linelen)) { ++ if (linelen == 0) { ++ if (entitylen > 0) ++ *ptcpoff += min(entitylen, tcplen - *ptcpoff); ++ break; ++ } ++ if (lineoff+linelen > tcplen) { ++ INFOP("!! overrun !!\n"); ++ break; ++ } ++ ++ if (nf_strncasecmp(ptcp+lineoff, "CSeq:", 5) == 0) { ++ *pcseqoff = lineoff; ++ *pcseqlen = linelen; ++ } ++ ++ if (nf_strncasecmp(ptcp+lineoff, "Transport:", 10) == 0) { ++ *transoff = lineoff; ++ *translen = linelen; ++ } ++ ++ if (nf_strncasecmp(ptcp+lineoff, "Content-Length:", 15) == 0) { ++ uint off = lineoff+15; ++ SKIP_WSPACE(ptcp+lineoff, linelen, off); ++ nf_strtou32(ptcp+off, &entitylen); ++ } ++ } ++ *phdrslen = (*ptcpoff) - (*phdrsoff); ++ ++ return 1; ++} ++ ++/* ++ * Find lo/hi client ports (if any) in transport header ++ * In: ++ * ptcp, tcplen = packet ++ * tranoff, tranlen = buffer to search ++ * ++ * Out: ++ * pport_lo, pport_hi = lo/hi ports (host endian) ++ * ++ * Returns nonzero if any client ports found ++ * ++ * Note: it is valid (and expected) for the client to request multiple ++ * transports, so we need to parse the entire line. ++ */ ++static int ++rtsp_parse_transport(char* ptran, uint tranlen, ++ struct ip_ct_rtsp_expect* prtspexp) ++{ ++ int rc = 0; ++ uint off = 0; ++ ++ if (tranlen < 10 || !iseol(ptran[tranlen-1]) || ++ nf_strncasecmp(ptran, "Transport:", 10) != 0) { ++ INFOP("sanity check failed\n"); ++ return 0; ++ } ++ ++ DEBUGP("tran='%.*s'\n", (int)tranlen, ptran); ++ off += 10; ++ SKIP_WSPACE(ptran, tranlen, off); ++ ++ /* Transport: tran;field;field=val,tran;field;field=val,... */ ++ while (off < tranlen) { ++ const char* pparamend; ++ uint nextparamoff; ++ ++ pparamend = memchr(ptran+off, ',', tranlen-off); ++ pparamend = (pparamend == NULL) ? ptran+tranlen : pparamend+1; ++ nextparamoff = pparamend-ptran; ++ ++ while (off < nextparamoff) { ++ const char* pfieldend; ++ uint nextfieldoff; ++ ++ pfieldend = memchr(ptran+off, ';', nextparamoff-off); ++ nextfieldoff = (pfieldend == NULL) ? nextparamoff : pfieldend-ptran+1; ++ ++ if (strncmp(ptran+off, "client_port=", 12) == 0) { ++ u_int16_t port; ++ uint numlen; ++ ++ off += 12; ++ numlen = nf_strtou16(ptran+off, &port); ++ off += numlen; ++ if (prtspexp->loport != 0 && prtspexp->loport != port) ++ DEBUGP("multiple ports found, port %hu ignored\n", port); ++ else { ++ DEBUGP("lo port found : %hu\n", port); ++ prtspexp->loport = prtspexp->hiport = port; ++ if (ptran[off] == '-') { ++ off++; ++ numlen = nf_strtou16(ptran+off, &port); ++ off += numlen; ++ prtspexp->pbtype = pb_range; ++ prtspexp->hiport = port; ++ ++ // If we have a range, assume rtp: ++ // loport must be even, hiport must be loport+1 ++ if ((prtspexp->loport & 0x0001) != 0 || ++ prtspexp->hiport != prtspexp->loport+1) { ++ DEBUGP("incorrect range: %hu-%hu, correcting\n", ++ prtspexp->loport, prtspexp->hiport); ++ prtspexp->loport &= 0xfffe; ++ prtspexp->hiport = prtspexp->loport+1; ++ } ++ } else if (ptran[off] == '/') { ++ off++; ++ numlen = nf_strtou16(ptran+off, &port); ++ off += numlen; ++ prtspexp->pbtype = pb_discon; ++ prtspexp->hiport = port; ++ } ++ rc = 1; ++ } ++ } ++ ++ /* ++ * Note we don't look for the destination parameter here. ++ * If we are using NAT, the NAT module will handle it. If not, ++ * and the client is sending packets elsewhere, the expectation ++ * will quietly time out. ++ */ ++ ++ off = nextfieldoff; ++ } ++ ++ off = nextparamoff; ++ } ++ ++ return rc; ++} ++ ++void expected(struct nf_conn *ct, struct nf_conntrack_expect *exp) ++{ ++ if(nf_nat_rtsp_hook_expectfn) { ++ nf_nat_rtsp_hook_expectfn(ct,exp); ++ } ++} ++ ++/*** conntrack functions ***/ ++ ++/* outbound packet: client->server */ ++ ++static inline int ++help_out(struct sk_buff *skb, unsigned char *rb_ptr, unsigned int datalen, ++ struct nf_conn *ct, enum ip_conntrack_info ctinfo) ++{ ++ struct ip_ct_rtsp_expect expinfo; ++ ++ int dir = CTINFO2DIR(ctinfo); /* = IP_CT_DIR_ORIGINAL */ ++ //struct tcphdr* tcph = (void*)iph + iph->ihl * 4; ++ //uint tcplen = pktlen - iph->ihl * 4; ++ char* pdata = rb_ptr; ++ //uint datalen = tcplen - tcph->doff * 4; ++ uint dataoff = 0; ++ int ret = NF_ACCEPT; ++ ++ struct nf_conntrack_expect *exp; ++ ++ __be16 be_loport; ++ ++ memset(&expinfo, 0, sizeof(expinfo)); ++ ++ while (dataoff < datalen) { ++ uint cmdoff = dataoff; ++ uint hdrsoff = 0; ++ uint hdrslen = 0; ++ uint cseqoff = 0; ++ uint cseqlen = 0; ++ uint transoff = 0; ++ uint translen = 0; ++ uint off; ++ ++ if (!rtsp_parse_message(pdata, datalen, &dataoff, ++ &hdrsoff, &hdrslen, ++ &cseqoff, &cseqlen, ++ &transoff, &translen)) ++ break; /* not a valid message */ ++ ++ if (strncmp(pdata+cmdoff, "SETUP ", 6) != 0) ++ continue; /* not a SETUP message */ ++ DEBUGP("found a setup message\n"); ++ ++ off = 0; ++ if(translen) { ++ rtsp_parse_transport(pdata+transoff, translen, &expinfo); ++ } ++ ++ if (expinfo.loport == 0) { ++ DEBUGP("no udp transports found\n"); ++ continue; /* no udp transports found */ ++ } ++ ++ DEBUGP("udp transport found, ports=(%d,%hu,%hu)\n", ++ (int)expinfo.pbtype, expinfo.loport, expinfo.hiport); ++ ++ exp = nf_ct_expect_alloc(ct); ++ if (!exp) { ++ ret = NF_DROP; ++ goto out; ++ } ++ ++ be_loport = htons(expinfo.loport); ++ ++ nf_ct_expect_init(exp, NF_CT_EXPECT_CLASS_DEFAULT, ++ ct->tuplehash[!dir].tuple.src.l3num, ++ &ct->tuplehash[!dir].tuple.src.u3, &ct->tuplehash[!dir].tuple.dst.u3, ++ IPPROTO_UDP, NULL, &be_loport); ++ ++ exp->master = ct; ++ ++ exp->expectfn = expected; ++ exp->flags = 0; ++ ++ if (expinfo.pbtype == pb_range) { ++ DEBUGP("Changing expectation mask to handle multiple ports\n"); ++ exp->mask.src.u.udp.port = 0xfffe; ++ } ++ ++ DEBUGP("expect_related %u.%u.%u.%u:%u-%u.%u.%u.%u:%u\n", ++ NIPQUAD(exp->tuple.src.u3.ip), ++ ntohs(exp->tuple.src.u.udp.port), ++ NIPQUAD(exp->tuple.dst.u3.ip), ++ ntohs(exp->tuple.dst.u.udp.port)); ++ ++ if (nf_nat_rtsp_hook) ++ /* pass the request off to the nat helper */ ++ ret = nf_nat_rtsp_hook(skb, ctinfo, hdrsoff, hdrslen, &expinfo, exp); ++ else if (nf_ct_expect_related(exp) != 0) { ++ INFOP("nf_ct_expect_related failed\n"); ++ ret = NF_DROP; ++ } ++ nf_ct_expect_put(exp); ++ goto out; ++ } ++out: ++ ++ return ret; ++} ++ ++ ++static inline int ++help_in(struct sk_buff *skb, size_t pktlen, ++ struct nf_conn* ct, enum ip_conntrack_info ctinfo) ++{ ++ return NF_ACCEPT; ++} ++ ++static int help(struct sk_buff *skb, unsigned int protoff, ++ struct nf_conn *ct, enum ip_conntrack_info ctinfo) ++{ ++ struct tcphdr _tcph, *th; ++ unsigned int dataoff, datalen; ++ char *rb_ptr; ++ int ret = NF_DROP; ++ ++ /* Until there's been traffic both ways, don't look in packets. */ ++ if (ctinfo != IP_CT_ESTABLISHED && ++ ctinfo != IP_CT_ESTABLISHED + IP_CT_IS_REPLY) { ++ DEBUGP("conntrackinfo = %u\n", ctinfo); ++ return NF_ACCEPT; ++ } ++ ++ /* Not whole TCP header? */ ++ th = skb_header_pointer(skb, protoff, sizeof(_tcph), &_tcph); ++ ++ if (!th) ++ return NF_ACCEPT; ++ ++ /* No data ? */ ++ dataoff = protoff + th->doff*4; ++ datalen = skb->len - dataoff; ++ if (dataoff >= skb->len) ++ return NF_ACCEPT; ++ ++ spin_lock_bh(&rtsp_buffer_lock); ++ rb_ptr = skb_header_pointer(skb, dataoff, ++ skb->len - dataoff, rtsp_buffer); ++ BUG_ON(rb_ptr == NULL); ++ ++#if 0 ++ /* Checksum invalid? Ignore. */ ++ /* FIXME: Source route IP option packets --RR */ ++ if (tcp_v4_check(tcph, tcplen, iph->saddr, iph->daddr, ++ csum_partial((char*)tcph, tcplen, 0))) ++ { ++ DEBUGP("bad csum: %p %u %u.%u.%u.%u %u.%u.%u.%u\n", ++ tcph, tcplen, NIPQUAD(iph->saddr), NIPQUAD(iph->daddr)); ++ return NF_ACCEPT; ++ } ++#endif ++ ++ switch (CTINFO2DIR(ctinfo)) { ++ case IP_CT_DIR_ORIGINAL: ++ ret = help_out(skb, rb_ptr, datalen, ct, ctinfo); ++ break; ++ case IP_CT_DIR_REPLY: ++ DEBUGP("IP_CT_DIR_REPLY\n"); ++ /* inbound packet: server->client */ ++ ret = NF_ACCEPT; ++ break; ++ } ++ ++ spin_unlock_bh(&rtsp_buffer_lock); ++ ++ return ret; ++} ++ ++static struct nf_conntrack_helper rtsp_helpers[MAX_PORTS]; ++static char rtsp_names[MAX_PORTS][10]; ++static struct nf_conntrack_expect_policy rtsp_expect_policy; ++ ++/* This function is intentionally _NOT_ defined as __exit */ ++static void ++fini(void) ++{ ++ int i; ++ for (i = 0; i < num_ports; i++) { ++ DEBUGP("unregistering port %d\n", ports[i]); ++ nf_conntrack_helper_unregister(&rtsp_helpers[i]); ++ } ++ kfree(rtsp_buffer); ++} ++ ++static int __init ++init(void) ++{ ++ int i, ret; ++ struct nf_conntrack_helper *hlpr; ++ char *tmpname; ++ ++ printk("nf_conntrack_rtsp v" IP_NF_RTSP_VERSION " loading\n"); ++ ++ if (max_outstanding < 1) { ++ printk("nf_conntrack_rtsp: max_outstanding must be a positive integer\n"); ++ return -EBUSY; ++ } ++ if (setup_timeout < 0) { ++ printk("nf_conntrack_rtsp: setup_timeout must be a positive integer\n"); ++ return -EBUSY; ++ } ++ ++ rtsp_expect_policy.max_expected = max_outstanding; ++ rtsp_expect_policy.timeout = setup_timeout; ++ ++ rtsp_buffer = kmalloc(65536, GFP_KERNEL); ++ if (!rtsp_buffer) ++ return -ENOMEM; ++ ++ /* If no port given, default to standard rtsp port */ ++ if (ports[0] == 0) { ++ ports[0] = RTSP_PORT; ++ } ++ ++ for (i = 0; (i < MAX_PORTS) && ports[i]; i++) { ++ hlpr = &rtsp_helpers[i]; ++ memset(hlpr, 0, sizeof(struct nf_conntrack_helper)); ++ hlpr->tuple.src.u.tcp.port = htons(ports[i]); ++ hlpr->tuple.dst.protonum = IPPROTO_TCP; ++ hlpr->expect_policy = &rtsp_expect_policy; ++ hlpr->me = THIS_MODULE; ++ hlpr->help = help; ++ ++ tmpname = &rtsp_names[i][0]; ++ if (ports[i] == RTSP_PORT) { ++ sprintf(tmpname, "rtsp"); ++ } else { ++ sprintf(tmpname, "rtsp-%d", i); ++ } ++ hlpr->name = tmpname; ++ ++ DEBUGP("port #%d: %d\n", i, ports[i]); ++ ++ ret = nf_conntrack_helper_register(hlpr); ++ ++ if (ret) { ++ printk("nf_conntrack_rtsp: ERROR registering port %d\n", ports[i]); ++ fini(); ++ return -EBUSY; ++ } ++ num_ports++; ++ } ++ return 0; ++} ++ ++module_init(init); ++module_exit(fini); ++ ++EXPORT_SYMBOL(nf_nat_rtsp_hook_expectfn); ++ +--- /dev/null ++++ b/net/ipv4/netfilter/nf_nat_rtsp.c +@@ -0,0 +1,496 @@ ++/* ++ * RTSP extension for TCP NAT alteration ++ * (C) 2003 by Tom Marshall ++ * based on ip_nat_irc.c ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ * ++ * Module load syntax: ++ * insmod nf_nat_rtsp.o ports=port1,port2,...port ++ * stunaddr=
++ * destaction=[auto|strip|none] ++ * ++ * If no ports are specified, the default will be port 554 only. ++ * ++ * stunaddr specifies the address used to detect that a client is using STUN. ++ * If this address is seen in the destination parameter, it is assumed that ++ * the client has already punched a UDP hole in the firewall, so we don't ++ * mangle the client_port. If none is specified, it is autodetected. It ++ * only needs to be set if you have multiple levels of NAT. It should be ++ * set to the external address that the STUN clients detect. Note that in ++ * this case, it will not be possible for clients to use UDP with servers ++ * between the NATs. ++ * ++ * If no destaction is specified, auto is used. ++ * destaction=auto: strip destination parameter if it is not stunaddr. ++ * destaction=strip: always strip destination parameter (not recommended). ++ * destaction=none: do not touch destination parameter (not recommended). ++ */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#include ++#include ++#define NF_NEED_STRNCASECMP ++#define NF_NEED_STRTOU16 ++#include ++#define NF_NEED_MIME_NEXTLINE ++#include ++ ++#define INFOP(fmt, args...) printk(KERN_INFO "%s: %s: " fmt, __FILE__, __FUNCTION__ , ## args) ++#if 0 ++#define DEBUGP(fmt, args...) printk(KERN_DEBUG "%s: %s: " fmt, __FILE__, __FUNCTION__ , ## args) ++#else ++#define DEBUGP(fmt, args...) ++#endif ++ ++#define MAX_PORTS 8 ++#define DSTACT_AUTO 0 ++#define DSTACT_STRIP 1 ++#define DSTACT_NONE 2 ++ ++static char* stunaddr = NULL; ++static char* destaction = NULL; ++ ++static u_int32_t extip = 0; ++static int dstact = 0; ++ ++MODULE_AUTHOR("Tom Marshall "); ++MODULE_DESCRIPTION("RTSP network address translation module"); ++MODULE_LICENSE("GPL"); ++module_param(stunaddr, charp, 0644); ++MODULE_PARM_DESC(stunaddr, "Address for detecting STUN"); ++module_param(destaction, charp, 0644); ++MODULE_PARM_DESC(destaction, "Action for destination parameter (auto/strip/none)"); ++ ++#define SKIP_WSPACE(ptr,len,off) while(off < len && isspace(*(ptr+off))) { off++; } ++ ++/*** helper functions ***/ ++ ++static void ++get_skb_tcpdata(struct sk_buff* skb, char** pptcpdata, uint* ptcpdatalen) ++{ ++ struct iphdr* iph = ip_hdr(skb); ++ struct tcphdr* tcph = (void *)iph + ip_hdrlen(skb); ++ ++ *pptcpdata = (char*)tcph + tcph->doff*4; ++ *ptcpdatalen = ((char*)skb_transport_header(skb) + skb->len) - *pptcpdata; ++} ++ ++/*** nat functions ***/ ++ ++/* ++ * Mangle the "Transport:" header: ++ * - Replace all occurences of "client_port=" ++ * - Handle destination parameter ++ * ++ * In: ++ * ct, ctinfo = conntrack context ++ * skb = packet ++ * tranoff = Transport header offset from TCP data ++ * tranlen = Transport header length (incl. CRLF) ++ * rport_lo = replacement low port (host endian) ++ * rport_hi = replacement high port (host endian) ++ * ++ * Returns packet size difference. ++ * ++ * Assumes that a complete transport header is present, ending with CR or LF ++ */ ++static int ++rtsp_mangle_tran(enum ip_conntrack_info ctinfo, ++ struct nf_conntrack_expect* exp, ++ struct ip_ct_rtsp_expect* prtspexp, ++ struct sk_buff* skb, uint tranoff, uint tranlen) ++{ ++ char* ptcp; ++ uint tcplen; ++ char* ptran; ++ char rbuf1[16]; /* Replacement buffer (one port) */ ++ uint rbuf1len; /* Replacement len (one port) */ ++ char rbufa[16]; /* Replacement buffer (all ports) */ ++ uint rbufalen; /* Replacement len (all ports) */ ++ u_int32_t newip; ++ u_int16_t loport, hiport; ++ uint off = 0; ++ uint diff; /* Number of bytes we removed */ ++ ++ struct nf_conn *ct = exp->master; ++ struct nf_conntrack_tuple *t; ++ ++ char szextaddr[15+1]; ++ uint extaddrlen; ++ int is_stun; ++ ++ get_skb_tcpdata(skb, &ptcp, &tcplen); ++ ptran = ptcp+tranoff; ++ ++ if (tranoff+tranlen > tcplen || tcplen-tranoff < tranlen || ++ tranlen < 10 || !iseol(ptran[tranlen-1]) || ++ nf_strncasecmp(ptran, "Transport:", 10) != 0) ++ { ++ INFOP("sanity check failed\n"); ++ return 0; ++ } ++ off += 10; ++ SKIP_WSPACE(ptcp+tranoff, tranlen, off); ++ ++ newip = ct->tuplehash[IP_CT_DIR_REPLY].tuple.dst.u3.ip; ++ t = &exp->tuple; ++ t->dst.u3.ip = newip; ++ ++ extaddrlen = extip ? sprintf(szextaddr, "%u.%u.%u.%u", NIPQUAD(extip)) ++ : sprintf(szextaddr, "%u.%u.%u.%u", NIPQUAD(newip)); ++ DEBUGP("stunaddr=%s (%s)\n", szextaddr, (extip?"forced":"auto")); ++ ++ rbuf1len = rbufalen = 0; ++ switch (prtspexp->pbtype) ++ { ++ case pb_single: ++ for (loport = prtspexp->loport; loport != 0; loport++) /* XXX: improper wrap? */ ++ { ++ t->dst.u.udp.port = htons(loport); ++ if (nf_ct_expect_related(exp) == 0) ++ { ++ DEBUGP("using port %hu\n", loport); ++ break; ++ } ++ } ++ if (loport != 0) ++ { ++ rbuf1len = sprintf(rbuf1, "%hu", loport); ++ rbufalen = sprintf(rbufa, "%hu", loport); ++ } ++ break; ++ case pb_range: ++ for (loport = prtspexp->loport; loport != 0; loport += 2) /* XXX: improper wrap? */ ++ { ++ t->dst.u.udp.port = htons(loport); ++ if (nf_ct_expect_related(exp) == 0) ++ { ++ hiport = loport + ~exp->mask.src.u.udp.port; ++ DEBUGP("using ports %hu-%hu\n", loport, hiport); ++ break; ++ } ++ } ++ if (loport != 0) ++ { ++ rbuf1len = sprintf(rbuf1, "%hu", loport); ++ rbufalen = sprintf(rbufa, "%hu-%hu", loport, loport+1); ++ } ++ break; ++ case pb_discon: ++ for (loport = prtspexp->loport; loport != 0; loport++) /* XXX: improper wrap? */ ++ { ++ t->dst.u.udp.port = htons(loport); ++ if (nf_ct_expect_related(exp) == 0) ++ { ++ DEBUGP("using port %hu (1 of 2)\n", loport); ++ break; ++ } ++ } ++ for (hiport = prtspexp->hiport; hiport != 0; hiport++) /* XXX: improper wrap? */ ++ { ++ t->dst.u.udp.port = htons(hiport); ++ if (nf_ct_expect_related(exp) == 0) ++ { ++ DEBUGP("using port %hu (2 of 2)\n", hiport); ++ break; ++ } ++ } ++ if (loport != 0 && hiport != 0) ++ { ++ rbuf1len = sprintf(rbuf1, "%hu", loport); ++ if (hiport == loport+1) ++ { ++ rbufalen = sprintf(rbufa, "%hu-%hu", loport, hiport); ++ } ++ else ++ { ++ rbufalen = sprintf(rbufa, "%hu/%hu", loport, hiport); ++ } ++ } ++ break; ++ } ++ ++ if (rbuf1len == 0) ++ { ++ return 0; /* cannot get replacement port(s) */ ++ } ++ ++ /* Transport: tran;field;field=val,tran;field;field=val,... */ ++ while (off < tranlen) ++ { ++ uint saveoff; ++ const char* pparamend; ++ uint nextparamoff; ++ ++ pparamend = memchr(ptran+off, ',', tranlen-off); ++ pparamend = (pparamend == NULL) ? ptran+tranlen : pparamend+1; ++ nextparamoff = pparamend-ptcp; ++ ++ /* ++ * We pass over each param twice. On the first pass, we look for a ++ * destination= field. It is handled by the security policy. If it ++ * is present, allowed, and equal to our external address, we assume ++ * that STUN is being used and we leave the client_port= field alone. ++ */ ++ is_stun = 0; ++ saveoff = off; ++ while (off < nextparamoff) ++ { ++ const char* pfieldend; ++ uint nextfieldoff; ++ ++ pfieldend = memchr(ptran+off, ';', nextparamoff-off); ++ nextfieldoff = (pfieldend == NULL) ? nextparamoff : pfieldend-ptran+1; ++ ++ if (dstact != DSTACT_NONE && strncmp(ptran+off, "destination=", 12) == 0) ++ { ++ if (strncmp(ptran+off+12, szextaddr, extaddrlen) == 0) ++ { ++ is_stun = 1; ++ } ++ if (dstact == DSTACT_STRIP || (dstact == DSTACT_AUTO && !is_stun)) ++ { ++ diff = nextfieldoff-off; ++ if (!nf_nat_mangle_tcp_packet(skb, ct, ctinfo, ++ off, diff, NULL, 0)) ++ { ++ /* mangle failed, all we can do is bail */ ++ nf_ct_unexpect_related(exp); ++ return 0; ++ } ++ get_skb_tcpdata(skb, &ptcp, &tcplen); ++ ptran = ptcp+tranoff; ++ tranlen -= diff; ++ nextparamoff -= diff; ++ nextfieldoff -= diff; ++ } ++ } ++ ++ off = nextfieldoff; ++ } ++ if (is_stun) ++ { ++ continue; ++ } ++ off = saveoff; ++ while (off < nextparamoff) ++ { ++ const char* pfieldend; ++ uint nextfieldoff; ++ ++ pfieldend = memchr(ptran+off, ';', nextparamoff-off); ++ nextfieldoff = (pfieldend == NULL) ? nextparamoff : pfieldend-ptran+1; ++ ++ if (strncmp(ptran+off, "client_port=", 12) == 0) ++ { ++ u_int16_t port; ++ uint numlen; ++ uint origoff; ++ uint origlen; ++ char* rbuf = rbuf1; ++ uint rbuflen = rbuf1len; ++ ++ off += 12; ++ origoff = (ptran-ptcp)+off; ++ origlen = 0; ++ numlen = nf_strtou16(ptran+off, &port); ++ off += numlen; ++ origlen += numlen; ++ if (port != prtspexp->loport) ++ { ++ DEBUGP("multiple ports found, port %hu ignored\n", port); ++ } ++ else ++ { ++ if (ptran[off] == '-' || ptran[off] == '/') ++ { ++ off++; ++ origlen++; ++ numlen = nf_strtou16(ptran+off, &port); ++ off += numlen; ++ origlen += numlen; ++ rbuf = rbufa; ++ rbuflen = rbufalen; ++ } ++ ++ /* ++ * note we cannot just memcpy() if the sizes are the same. ++ * the mangle function does skb resizing, checks for a ++ * cloned skb, and updates the checksums. ++ * ++ * parameter 4 below is offset from start of tcp data. ++ */ ++ diff = origlen-rbuflen; ++ if (!nf_nat_mangle_tcp_packet(skb, ct, ctinfo, ++ origoff, origlen, rbuf, rbuflen)) ++ { ++ /* mangle failed, all we can do is bail */ ++ nf_ct_unexpect_related(exp); ++ return 0; ++ } ++ get_skb_tcpdata(skb, &ptcp, &tcplen); ++ ptran = ptcp+tranoff; ++ tranlen -= diff; ++ nextparamoff -= diff; ++ nextfieldoff -= diff; ++ } ++ } ++ ++ off = nextfieldoff; ++ } ++ ++ off = nextparamoff; ++ } ++ ++ return 1; ++} ++ ++static uint ++help_out(struct sk_buff *skb, enum ip_conntrack_info ctinfo, ++ unsigned int matchoff, unsigned int matchlen, struct ip_ct_rtsp_expect* prtspexp, ++ struct nf_conntrack_expect* exp) ++{ ++ char* ptcp; ++ uint tcplen; ++ uint hdrsoff; ++ uint hdrslen; ++ uint lineoff; ++ uint linelen; ++ uint off; ++ ++ //struct iphdr* iph = (struct iphdr*)skb->nh.iph; ++ //struct tcphdr* tcph = (struct tcphdr*)((void*)iph + iph->ihl*4); ++ ++ get_skb_tcpdata(skb, &ptcp, &tcplen); ++ hdrsoff = matchoff;//exp->seq - ntohl(tcph->seq); ++ hdrslen = matchlen; ++ off = hdrsoff; ++ DEBUGP("NAT rtsp help_out\n"); ++ ++ while (nf_mime_nextline(ptcp, hdrsoff+hdrslen, &off, &lineoff, &linelen)) ++ { ++ if (linelen == 0) ++ { ++ break; ++ } ++ if (off > hdrsoff+hdrslen) ++ { ++ INFOP("!! overrun !!"); ++ break; ++ } ++ DEBUGP("hdr: len=%u, %.*s", linelen, (int)linelen, ptcp+lineoff); ++ ++ if (nf_strncasecmp(ptcp+lineoff, "Transport:", 10) == 0) ++ { ++ uint oldtcplen = tcplen; ++ DEBUGP("hdr: Transport\n"); ++ if (!rtsp_mangle_tran(ctinfo, exp, prtspexp, skb, lineoff, linelen)) ++ { ++ DEBUGP("hdr: Transport mangle failed"); ++ break; ++ } ++ get_skb_tcpdata(skb, &ptcp, &tcplen); ++ hdrslen -= (oldtcplen-tcplen); ++ off -= (oldtcplen-tcplen); ++ lineoff -= (oldtcplen-tcplen); ++ linelen -= (oldtcplen-tcplen); ++ DEBUGP("rep: len=%u, %.*s", linelen, (int)linelen, ptcp+lineoff); ++ } ++ } ++ ++ return NF_ACCEPT; ++} ++ ++static unsigned int ++help(struct sk_buff *skb, enum ip_conntrack_info ctinfo, ++ unsigned int matchoff, unsigned int matchlen, struct ip_ct_rtsp_expect* prtspexp, ++ struct nf_conntrack_expect* exp) ++{ ++ int dir = CTINFO2DIR(ctinfo); ++ int rc = NF_ACCEPT; ++ ++ switch (dir) ++ { ++ case IP_CT_DIR_ORIGINAL: ++ rc = help_out(skb, ctinfo, matchoff, matchlen, prtspexp, exp); ++ break; ++ case IP_CT_DIR_REPLY: ++ DEBUGP("unmangle ! %u\n", ctinfo); ++ /* XXX: unmangle */ ++ rc = NF_ACCEPT; ++ break; ++ } ++ //UNLOCK_BH(&ip_rtsp_lock); ++ ++ return rc; ++} ++ ++static void expected(struct nf_conn* ct, struct nf_conntrack_expect *exp) ++{ ++ struct nf_nat_multi_range_compat mr; ++ u_int32_t newdstip, newsrcip, newip; ++ ++ struct nf_conn *master = ct->master; ++ ++ newdstip = master->tuplehash[IP_CT_DIR_ORIGINAL].tuple.src.u3.ip; ++ newsrcip = ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple.src.u3.ip; ++ //FIXME (how to port that ?) ++ //code from 2.4 : newip = (HOOK2MANIP(hooknum) == IP_NAT_MANIP_SRC) ? newsrcip : newdstip; ++ newip = newdstip; ++ ++ DEBUGP("newsrcip=%u.%u.%u.%u, newdstip=%u.%u.%u.%u, newip=%u.%u.%u.%u\n", ++ NIPQUAD(newsrcip), NIPQUAD(newdstip), NIPQUAD(newip)); ++ ++ mr.rangesize = 1; ++ // We don't want to manip the per-protocol, just the IPs. ++ mr.range[0].flags = IP_NAT_RANGE_MAP_IPS; ++ mr.range[0].min_ip = mr.range[0].max_ip = newip; ++ ++ nf_nat_setup_info(ct, &mr.range[0], IP_NAT_MANIP_DST); ++} ++ ++ ++static void __exit fini(void) ++{ ++ nf_nat_rtsp_hook = NULL; ++ nf_nat_rtsp_hook_expectfn = NULL; ++ synchronize_net(); ++} ++ ++static int __init init(void) ++{ ++ printk("nf_nat_rtsp v" IP_NF_RTSP_VERSION " loading\n"); ++ ++ BUG_ON(nf_nat_rtsp_hook); ++ nf_nat_rtsp_hook = help; ++ nf_nat_rtsp_hook_expectfn = &expected; ++ ++ if (stunaddr != NULL) ++ extip = in_aton(stunaddr); ++ ++ if (destaction != NULL) { ++ if (strcmp(destaction, "auto") == 0) ++ dstact = DSTACT_AUTO; ++ ++ if (strcmp(destaction, "strip") == 0) ++ dstact = DSTACT_STRIP; ++ ++ if (strcmp(destaction, "none") == 0) ++ dstact = DSTACT_NONE; ++ } ++ ++ return 0; ++} ++ ++module_init(init); ++module_exit(fini); diff --git a/target/linux/generic-2.6/patches-2.6.31/200-sched_esfq.patch b/target/linux/generic-2.6/patches-2.6.31/200-sched_esfq.patch new file mode 100644 index 0000000000..f72c0e3705 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/200-sched_esfq.patch @@ -0,0 +1,793 @@ +--- a/include/linux/pkt_sched.h ++++ b/include/linux/pkt_sched.h +@@ -182,8 +182,37 @@ struct tc_sfq_xstats + * + * The only reason for this is efficiency, it is possible + * to change these parameters in compile time. ++ * ++ * If you need to play with these values, use esfq instead. + */ + ++/* ESFQ section */ ++ ++enum ++{ ++ /* traditional */ ++ TCA_SFQ_HASH_CLASSIC, ++ TCA_SFQ_HASH_DST, ++ TCA_SFQ_HASH_SRC, ++ TCA_SFQ_HASH_FWMARK, ++ /* conntrack */ ++ TCA_SFQ_HASH_CTORIGDST, ++ TCA_SFQ_HASH_CTORIGSRC, ++ TCA_SFQ_HASH_CTREPLDST, ++ TCA_SFQ_HASH_CTREPLSRC, ++ TCA_SFQ_HASH_CTNATCHG, ++}; ++ ++struct tc_esfq_qopt ++{ ++ unsigned quantum; /* Bytes per round allocated to flow */ ++ int perturb_period; /* Period of hash perturbation */ ++ __u32 limit; /* Maximal packets in queue */ ++ unsigned divisor; /* Hash divisor */ ++ unsigned flows; /* Maximal number of flows */ ++ unsigned hash_kind; /* Hash function to use for flow identification */ ++}; ++ + /* RED section */ + + enum +--- a/net/sched/Kconfig ++++ b/net/sched/Kconfig +@@ -137,6 +137,37 @@ config NET_SCH_SFQ + To compile this code as a module, choose M here: the + module will be called sch_sfq. + ++config NET_SCH_ESFQ ++ tristate "Enhanced Stochastic Fairness Queueing (ESFQ)" ++ ---help--- ++ Say Y here if you want to use the Enhanced Stochastic Fairness ++ Queueing (ESFQ) packet scheduling algorithm for some of your network ++ devices or as a leaf discipline for a classful qdisc such as HTB or ++ CBQ (see the top of for details and ++ references to the SFQ algorithm). ++ ++ This is an enchanced SFQ version which allows you to control some ++ hardcoded values in the SFQ scheduler. ++ ++ ESFQ also adds control of the hash function used to identify packet ++ flows. The original SFQ discipline hashes by connection; ESFQ add ++ several other hashing methods, such as by src IP or by dst IP, which ++ can be more fair to users in some networking situations. ++ ++ To compile this code as a module, choose M here: the ++ module will be called sch_esfq. ++ ++config NET_SCH_ESFQ_NFCT ++ bool "Connection Tracking Hash Types" ++ depends on NET_SCH_ESFQ && NF_CONNTRACK ++ ---help--- ++ Say Y here to enable support for hashing based on netfilter connection ++ tracking information. This is useful for a router that is also using ++ NAT to connect privately-addressed hosts to the Internet. If you want ++ to provide fair distribution of upstream bandwidth, ESFQ must use ++ connection tracking information, since all outgoing packets will share ++ the same source address. ++ + config NET_SCH_TEQL + tristate "True Link Equalizer (TEQL)" + ---help--- +--- a/net/sched/Makefile ++++ b/net/sched/Makefile +@@ -24,6 +24,7 @@ obj-$(CONFIG_NET_SCH_GRED) += sch_gred.o + obj-$(CONFIG_NET_SCH_INGRESS) += sch_ingress.o + obj-$(CONFIG_NET_SCH_DSMARK) += sch_dsmark.o + obj-$(CONFIG_NET_SCH_SFQ) += sch_sfq.o ++obj-$(CONFIG_NET_SCH_ESFQ) += sch_esfq.o + obj-$(CONFIG_NET_SCH_TBF) += sch_tbf.o + obj-$(CONFIG_NET_SCH_TEQL) += sch_teql.o + obj-$(CONFIG_NET_SCH_PRIO) += sch_prio.o +--- /dev/null ++++ b/net/sched/sch_esfq.c +@@ -0,0 +1,700 @@ ++/* ++ * net/sched/sch_esfq.c Extended Stochastic Fairness Queueing discipline. ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ * ++ * Authors: Alexey Kuznetsov, ++ * ++ * Changes: Alexander Atanasov, ++ * Added dynamic depth,limit,divisor,hash_kind options. ++ * Added dst and src hashes. ++ * ++ * Alexander Clouter, ++ * Ported ESFQ to Linux 2.6. ++ * ++ * Corey Hickey, ++ * Maintenance of the Linux 2.6 port. ++ * Added fwmark hash (thanks to Robert Kurjata). ++ * Added usage of jhash. ++ * Added conntrack support. ++ * Added ctnatchg hash (thanks to Ben Pfountz). ++ */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++/* Stochastic Fairness Queuing algorithm. ++ For more comments look at sch_sfq.c. ++ The difference is that you can change limit, depth, ++ hash table size and choose alternate hash types. ++ ++ classic: same as in sch_sfq.c ++ dst: destination IP address ++ src: source IP address ++ fwmark: netfilter mark value ++ ctorigdst: original destination IP address ++ ctorigsrc: original source IP address ++ ctrepldst: reply destination IP address ++ ctreplsrc: reply source IP ++ ++*/ ++ ++#define ESFQ_HEAD 0 ++#define ESFQ_TAIL 1 ++ ++/* This type should contain at least SFQ_DEPTH*2 values */ ++typedef unsigned int esfq_index; ++ ++struct esfq_head ++{ ++ esfq_index next; ++ esfq_index prev; ++}; ++ ++struct esfq_sched_data ++{ ++/* Parameters */ ++ int perturb_period; ++ unsigned quantum; /* Allotment per round: MUST BE >= MTU */ ++ int limit; ++ unsigned depth; ++ unsigned hash_divisor; ++ unsigned hash_kind; ++/* Variables */ ++ struct timer_list perturb_timer; ++ int perturbation; ++ esfq_index tail; /* Index of current slot in round */ ++ esfq_index max_depth; /* Maximal depth */ ++ ++ esfq_index *ht; /* Hash table */ ++ esfq_index *next; /* Active slots link */ ++ short *allot; /* Current allotment per slot */ ++ unsigned short *hash; /* Hash value indexed by slots */ ++ struct sk_buff_head *qs; /* Slot queue */ ++ struct esfq_head *dep; /* Linked list of slots, indexed by depth */ ++}; ++ ++/* This contains the info we will hash. */ ++struct esfq_packet_info ++{ ++ u32 proto; /* protocol or port */ ++ u32 src; /* source from packet header */ ++ u32 dst; /* destination from packet header */ ++ u32 ctorigsrc; /* original source from conntrack */ ++ u32 ctorigdst; /* original destination from conntrack */ ++ u32 ctreplsrc; /* reply source from conntrack */ ++ u32 ctrepldst; /* reply destination from conntrack */ ++ u32 mark; /* netfilter mark (fwmark) */ ++}; ++ ++static __inline__ unsigned esfq_jhash_1word(struct esfq_sched_data *q,u32 a) ++{ ++ return jhash_1word(a, q->perturbation) & (q->hash_divisor-1); ++} ++ ++static __inline__ unsigned esfq_jhash_2words(struct esfq_sched_data *q, u32 a, u32 b) ++{ ++ return jhash_2words(a, b, q->perturbation) & (q->hash_divisor-1); ++} ++ ++static __inline__ unsigned esfq_jhash_3words(struct esfq_sched_data *q, u32 a, u32 b, u32 c) ++{ ++ return jhash_3words(a, b, c, q->perturbation) & (q->hash_divisor-1); ++} ++ ++static unsigned esfq_hash(struct esfq_sched_data *q, struct sk_buff *skb) ++{ ++ struct esfq_packet_info info; ++#ifdef CONFIG_NET_SCH_ESFQ_NFCT ++ enum ip_conntrack_info ctinfo; ++ struct nf_conn *ct = nf_ct_get(skb, &ctinfo); ++#endif ++ ++ switch (skb->protocol) { ++ case __constant_htons(ETH_P_IP): ++ { ++ struct iphdr *iph = ip_hdr(skb); ++ info.dst = iph->daddr; ++ info.src = iph->saddr; ++ if (!(iph->frag_off&htons(IP_MF|IP_OFFSET)) && ++ (iph->protocol == IPPROTO_TCP || ++ iph->protocol == IPPROTO_UDP || ++ iph->protocol == IPPROTO_SCTP || ++ iph->protocol == IPPROTO_DCCP || ++ iph->protocol == IPPROTO_ESP)) ++ info.proto = *(((u32*)iph) + iph->ihl); ++ else ++ info.proto = iph->protocol; ++ break; ++ } ++ case __constant_htons(ETH_P_IPV6): ++ { ++ struct ipv6hdr *iph = ipv6_hdr(skb); ++ /* Hash ipv6 addresses into a u32. This isn't ideal, ++ * but the code is simple. */ ++ info.dst = jhash2(iph->daddr.s6_addr32, 4, q->perturbation); ++ info.src = jhash2(iph->saddr.s6_addr32, 4, q->perturbation); ++ if (iph->nexthdr == IPPROTO_TCP || ++ iph->nexthdr == IPPROTO_UDP || ++ iph->nexthdr == IPPROTO_SCTP || ++ iph->nexthdr == IPPROTO_DCCP || ++ iph->nexthdr == IPPROTO_ESP) ++ info.proto = *(u32*)&iph[1]; ++ else ++ info.proto = iph->nexthdr; ++ break; ++ } ++ default: ++ info.dst = (u32)(unsigned long)skb->dst; ++ info.src = (u32)(unsigned long)skb->sk; ++ info.proto = skb->protocol; ++ } ++ ++ info.mark = skb->mark; ++ ++#ifdef CONFIG_NET_SCH_ESFQ_NFCT ++ /* defaults if there is no conntrack info */ ++ info.ctorigsrc = info.src; ++ info.ctorigdst = info.dst; ++ info.ctreplsrc = info.dst; ++ info.ctrepldst = info.src; ++ /* collect conntrack info */ ++ if (ct && ct != &nf_conntrack_untracked) { ++ if (skb->protocol == __constant_htons(ETH_P_IP)) { ++ info.ctorigsrc = ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple.src.u3.ip; ++ info.ctorigdst = ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple.dst.u3.ip; ++ info.ctreplsrc = ct->tuplehash[IP_CT_DIR_REPLY].tuple.src.u3.ip; ++ info.ctrepldst = ct->tuplehash[IP_CT_DIR_REPLY].tuple.dst.u3.ip; ++ } ++ else if (skb->protocol == __constant_htons(ETH_P_IPV6)) { ++ /* Again, hash ipv6 addresses into a single u32. */ ++ info.ctorigsrc = jhash2(ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple.src.u3.ip6, 4, q->perturbation); ++ info.ctorigdst = jhash2(ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple.dst.u3.ip6, 4, q->perturbation); ++ info.ctreplsrc = jhash2(ct->tuplehash[IP_CT_DIR_REPLY].tuple.src.u3.ip6, 4, q->perturbation); ++ info.ctrepldst = jhash2(ct->tuplehash[IP_CT_DIR_REPLY].tuple.dst.u3.ip6, 4, q->perturbation); ++ } ++ ++ } ++#endif ++ ++ switch(q->hash_kind) { ++ case TCA_SFQ_HASH_CLASSIC: ++ return esfq_jhash_3words(q, info.dst, info.src, info.proto); ++ case TCA_SFQ_HASH_DST: ++ return esfq_jhash_1word(q, info.dst); ++ case TCA_SFQ_HASH_SRC: ++ return esfq_jhash_1word(q, info.src); ++ case TCA_SFQ_HASH_FWMARK: ++ return esfq_jhash_1word(q, info.mark); ++#ifdef CONFIG_NET_SCH_ESFQ_NFCT ++ case TCA_SFQ_HASH_CTORIGDST: ++ return esfq_jhash_1word(q, info.ctorigdst); ++ case TCA_SFQ_HASH_CTORIGSRC: ++ return esfq_jhash_1word(q, info.ctorigsrc); ++ case TCA_SFQ_HASH_CTREPLDST: ++ return esfq_jhash_1word(q, info.ctrepldst); ++ case TCA_SFQ_HASH_CTREPLSRC: ++ return esfq_jhash_1word(q, info.ctreplsrc); ++ case TCA_SFQ_HASH_CTNATCHG: ++ { ++ if (info.ctorigdst == info.ctreplsrc) ++ return esfq_jhash_1word(q, info.ctorigsrc); ++ return esfq_jhash_1word(q, info.ctreplsrc); ++ } ++#endif ++ default: ++ if (net_ratelimit()) ++ printk(KERN_WARNING "ESFQ: Unknown hash method. Falling back to classic.\n"); ++ } ++ return esfq_jhash_3words(q, info.dst, info.src, info.proto); ++} ++ ++static inline void esfq_link(struct esfq_sched_data *q, esfq_index x) ++{ ++ esfq_index p, n; ++ int d = q->qs[x].qlen + q->depth; ++ ++ p = d; ++ n = q->dep[d].next; ++ q->dep[x].next = n; ++ q->dep[x].prev = p; ++ q->dep[p].next = q->dep[n].prev = x; ++} ++ ++static inline void esfq_dec(struct esfq_sched_data *q, esfq_index x) ++{ ++ esfq_index p, n; ++ ++ n = q->dep[x].next; ++ p = q->dep[x].prev; ++ q->dep[p].next = n; ++ q->dep[n].prev = p; ++ ++ if (n == p && q->max_depth == q->qs[x].qlen + 1) ++ q->max_depth--; ++ ++ esfq_link(q, x); ++} ++ ++static inline void esfq_inc(struct esfq_sched_data *q, esfq_index x) ++{ ++ esfq_index p, n; ++ int d; ++ ++ n = q->dep[x].next; ++ p = q->dep[x].prev; ++ q->dep[p].next = n; ++ q->dep[n].prev = p; ++ d = q->qs[x].qlen; ++ if (q->max_depth < d) ++ q->max_depth = d; ++ ++ esfq_link(q, x); ++} ++ ++static unsigned int esfq_drop(struct Qdisc *sch) ++{ ++ struct esfq_sched_data *q = qdisc_priv(sch); ++ esfq_index d = q->max_depth; ++ struct sk_buff *skb; ++ unsigned int len; ++ ++ /* Queue is full! Find the longest slot and ++ drop a packet from it */ ++ ++ if (d > 1) { ++ esfq_index x = q->dep[d+q->depth].next; ++ skb = q->qs[x].prev; ++ len = skb->len; ++ __skb_unlink(skb, &q->qs[x]); ++ kfree_skb(skb); ++ esfq_dec(q, x); ++ sch->q.qlen--; ++ sch->qstats.drops++; ++ sch->qstats.backlog -= len; ++ return len; ++ } ++ ++ if (d == 1) { ++ /* It is difficult to believe, but ALL THE SLOTS HAVE LENGTH 1. */ ++ d = q->next[q->tail]; ++ q->next[q->tail] = q->next[d]; ++ q->allot[q->next[d]] += q->quantum; ++ skb = q->qs[d].prev; ++ len = skb->len; ++ __skb_unlink(skb, &q->qs[d]); ++ kfree_skb(skb); ++ esfq_dec(q, d); ++ sch->q.qlen--; ++ q->ht[q->hash[d]] = q->depth; ++ sch->qstats.drops++; ++ sch->qstats.backlog -= len; ++ return len; ++ } ++ ++ return 0; ++} ++ ++static void esfq_q_enqueue(struct sk_buff *skb, struct esfq_sched_data *q, unsigned int end) ++{ ++ unsigned hash = esfq_hash(q, skb); ++ unsigned depth = q->depth; ++ esfq_index x; ++ ++ x = q->ht[hash]; ++ if (x == depth) { ++ q->ht[hash] = x = q->dep[depth].next; ++ q->hash[x] = hash; ++ } ++ ++ if (end == ESFQ_TAIL) ++ __skb_queue_tail(&q->qs[x], skb); ++ else ++ __skb_queue_head(&q->qs[x], skb); ++ ++ esfq_inc(q, x); ++ if (q->qs[x].qlen == 1) { /* The flow is new */ ++ if (q->tail == depth) { /* It is the first flow */ ++ q->tail = x; ++ q->next[x] = x; ++ q->allot[x] = q->quantum; ++ } else { ++ q->next[x] = q->next[q->tail]; ++ q->next[q->tail] = x; ++ q->tail = x; ++ } ++ } ++} ++ ++static int esfq_enqueue(struct sk_buff *skb, struct Qdisc* sch) ++{ ++ struct esfq_sched_data *q = qdisc_priv(sch); ++ esfq_q_enqueue(skb, q, ESFQ_TAIL); ++ sch->qstats.backlog += skb->len; ++ if (++sch->q.qlen < q->limit-1) { ++ sch->bstats.bytes += skb->len; ++ sch->bstats.packets++; ++ return 0; ++ } ++ ++ sch->qstats.drops++; ++ esfq_drop(sch); ++ return NET_XMIT_CN; ++} ++ ++static struct sk_buff *esfq_peek(struct Qdisc* sch) ++{ ++ struct esfq_sched_data *q = qdisc_priv(sch); ++ esfq_index a; ++ ++ /* No active slots */ ++ if (q->tail == q->depth) ++ return NULL; ++ ++ a = q->next[q->tail]; ++ return skb_peek(&q->qs[a]); ++} ++ ++static struct sk_buff *esfq_q_dequeue(struct esfq_sched_data *q) ++{ ++ struct sk_buff *skb; ++ unsigned depth = q->depth; ++ esfq_index a, old_a; ++ ++ /* No active slots */ ++ if (q->tail == depth) ++ return NULL; ++ ++ a = old_a = q->next[q->tail]; ++ ++ /* Grab packet */ ++ skb = __skb_dequeue(&q->qs[a]); ++ esfq_dec(q, a); ++ ++ /* Is the slot empty? */ ++ if (q->qs[a].qlen == 0) { ++ q->ht[q->hash[a]] = depth; ++ a = q->next[a]; ++ if (a == old_a) { ++ q->tail = depth; ++ return skb; ++ } ++ q->next[q->tail] = a; ++ q->allot[a] += q->quantum; ++ } else if ((q->allot[a] -= skb->len) <= 0) { ++ q->tail = a; ++ a = q->next[a]; ++ q->allot[a] += q->quantum; ++ } ++ ++ return skb; ++} ++ ++static struct sk_buff *esfq_dequeue(struct Qdisc* sch) ++{ ++ struct esfq_sched_data *q = qdisc_priv(sch); ++ struct sk_buff *skb; ++ ++ skb = esfq_q_dequeue(q); ++ if (skb == NULL) ++ return NULL; ++ sch->q.qlen--; ++ sch->qstats.backlog -= skb->len; ++ return skb; ++} ++ ++static void esfq_q_destroy(struct esfq_sched_data *q) ++{ ++ del_timer(&q->perturb_timer); ++ if(q->ht) ++ kfree(q->ht); ++ if(q->dep) ++ kfree(q->dep); ++ if(q->next) ++ kfree(q->next); ++ if(q->allot) ++ kfree(q->allot); ++ if(q->hash) ++ kfree(q->hash); ++ if(q->qs) ++ kfree(q->qs); ++} ++ ++static void esfq_destroy(struct Qdisc *sch) ++{ ++ struct esfq_sched_data *q = qdisc_priv(sch); ++ esfq_q_destroy(q); ++} ++ ++ ++static void esfq_reset(struct Qdisc* sch) ++{ ++ struct sk_buff *skb; ++ ++ while ((skb = esfq_dequeue(sch)) != NULL) ++ kfree_skb(skb); ++} ++ ++static void esfq_perturbation(unsigned long arg) ++{ ++ struct Qdisc *sch = (struct Qdisc*)arg; ++ struct esfq_sched_data *q = qdisc_priv(sch); ++ ++ q->perturbation = net_random()&0x1F; ++ ++ if (q->perturb_period) { ++ q->perturb_timer.expires = jiffies + q->perturb_period; ++ add_timer(&q->perturb_timer); ++ } ++} ++ ++static unsigned int esfq_check_hash(unsigned int kind) ++{ ++ switch (kind) { ++ case TCA_SFQ_HASH_CTORIGDST: ++ case TCA_SFQ_HASH_CTORIGSRC: ++ case TCA_SFQ_HASH_CTREPLDST: ++ case TCA_SFQ_HASH_CTREPLSRC: ++ case TCA_SFQ_HASH_CTNATCHG: ++#ifndef CONFIG_NET_SCH_ESFQ_NFCT ++ { ++ if (net_ratelimit()) ++ printk(KERN_WARNING "ESFQ: Conntrack hash types disabled in kernel config. Falling back to classic.\n"); ++ return TCA_SFQ_HASH_CLASSIC; ++ } ++#endif ++ case TCA_SFQ_HASH_CLASSIC: ++ case TCA_SFQ_HASH_DST: ++ case TCA_SFQ_HASH_SRC: ++ case TCA_SFQ_HASH_FWMARK: ++ return kind; ++ default: ++ { ++ if (net_ratelimit()) ++ printk(KERN_WARNING "ESFQ: Unknown hash type. Falling back to classic.\n"); ++ return TCA_SFQ_HASH_CLASSIC; ++ } ++ } ++} ++ ++static int esfq_q_init(struct esfq_sched_data *q, struct nlattr *opt) ++{ ++ struct tc_esfq_qopt *ctl = nla_data(opt); ++ esfq_index p = ~0U/2; ++ int i; ++ ++ if (opt && opt->nla_len < nla_attr_size(sizeof(*ctl))) ++ return -EINVAL; ++ ++ q->perturbation = 0; ++ q->hash_kind = TCA_SFQ_HASH_CLASSIC; ++ q->max_depth = 0; ++ if (opt == NULL) { ++ q->perturb_period = 0; ++ q->hash_divisor = 1024; ++ q->tail = q->limit = q->depth = 128; ++ ++ } else { ++ struct tc_esfq_qopt *ctl = nla_data(opt); ++ if (ctl->quantum) ++ q->quantum = ctl->quantum; ++ q->perturb_period = ctl->perturb_period*HZ; ++ q->hash_divisor = ctl->divisor ? : 1024; ++ q->tail = q->limit = q->depth = ctl->flows ? : 128; ++ ++ if ( q->depth > p - 1 ) ++ return -EINVAL; ++ ++ if (ctl->limit) ++ q->limit = min_t(u32, ctl->limit, q->depth); ++ ++ if (ctl->hash_kind) { ++ q->hash_kind = esfq_check_hash(ctl->hash_kind); ++ } ++ } ++ ++ q->ht = kmalloc(q->hash_divisor*sizeof(esfq_index), GFP_KERNEL); ++ if (!q->ht) ++ goto err_case; ++ q->dep = kmalloc((1+q->depth*2)*sizeof(struct esfq_head), GFP_KERNEL); ++ if (!q->dep) ++ goto err_case; ++ q->next = kmalloc(q->depth*sizeof(esfq_index), GFP_KERNEL); ++ if (!q->next) ++ goto err_case; ++ q->allot = kmalloc(q->depth*sizeof(short), GFP_KERNEL); ++ if (!q->allot) ++ goto err_case; ++ q->hash = kmalloc(q->depth*sizeof(unsigned short), GFP_KERNEL); ++ if (!q->hash) ++ goto err_case; ++ q->qs = kmalloc(q->depth*sizeof(struct sk_buff_head), GFP_KERNEL); ++ if (!q->qs) ++ goto err_case; ++ ++ for (i=0; i< q->hash_divisor; i++) ++ q->ht[i] = q->depth; ++ for (i=0; idepth; i++) { ++ skb_queue_head_init(&q->qs[i]); ++ q->dep[i+q->depth].next = i+q->depth; ++ q->dep[i+q->depth].prev = i+q->depth; ++ } ++ ++ for (i=0; idepth; i++) ++ esfq_link(q, i); ++ return 0; ++err_case: ++ esfq_q_destroy(q); ++ return -ENOBUFS; ++} ++ ++static int esfq_init(struct Qdisc *sch, struct nlattr *opt) ++{ ++ struct esfq_sched_data *q = qdisc_priv(sch); ++ int err; ++ ++ q->quantum = psched_mtu(qdisc_dev(sch)); /* default */ ++ if ((err = esfq_q_init(q, opt))) ++ return err; ++ ++ init_timer(&q->perturb_timer); ++ q->perturb_timer.data = (unsigned long)sch; ++ q->perturb_timer.function = esfq_perturbation; ++ if (q->perturb_period) { ++ q->perturb_timer.expires = jiffies + q->perturb_period; ++ add_timer(&q->perturb_timer); ++ } ++ ++ return 0; ++} ++ ++static int esfq_change(struct Qdisc *sch, struct nlattr *opt) ++{ ++ struct esfq_sched_data *q = qdisc_priv(sch); ++ struct esfq_sched_data new; ++ struct sk_buff *skb; ++ int err; ++ ++ /* set up new queue */ ++ memset(&new, 0, sizeof(struct esfq_sched_data)); ++ new.quantum = psched_mtu(qdisc_dev(sch)); /* default */ ++ if ((err = esfq_q_init(&new, opt))) ++ return err; ++ ++ /* copy all packets from the old queue to the new queue */ ++ sch_tree_lock(sch); ++ while ((skb = esfq_q_dequeue(q)) != NULL) ++ esfq_q_enqueue(skb, &new, ESFQ_TAIL); ++ ++ /* clean up the old queue */ ++ esfq_q_destroy(q); ++ ++ /* copy elements of the new queue into the old queue */ ++ q->perturb_period = new.perturb_period; ++ q->quantum = new.quantum; ++ q->limit = new.limit; ++ q->depth = new.depth; ++ q->hash_divisor = new.hash_divisor; ++ q->hash_kind = new.hash_kind; ++ q->tail = new.tail; ++ q->max_depth = new.max_depth; ++ q->ht = new.ht; ++ q->dep = new.dep; ++ q->next = new.next; ++ q->allot = new.allot; ++ q->hash = new.hash; ++ q->qs = new.qs; ++ ++ /* finish up */ ++ if (q->perturb_period) { ++ q->perturb_timer.expires = jiffies + q->perturb_period; ++ add_timer(&q->perturb_timer); ++ } else { ++ q->perturbation = 0; ++ } ++ sch_tree_unlock(sch); ++ return 0; ++} ++ ++static int esfq_dump(struct Qdisc *sch, struct sk_buff *skb) ++{ ++ struct esfq_sched_data *q = qdisc_priv(sch); ++ unsigned char *b = skb_tail_pointer(skb); ++ struct tc_esfq_qopt opt; ++ ++ opt.quantum = q->quantum; ++ opt.perturb_period = q->perturb_period/HZ; ++ ++ opt.limit = q->limit; ++ opt.divisor = q->hash_divisor; ++ opt.flows = q->depth; ++ opt.hash_kind = q->hash_kind; ++ ++ NLA_PUT(skb, TCA_OPTIONS, sizeof(opt), &opt); ++ ++ return skb->len; ++ ++nla_put_failure: ++ nlmsg_trim(skb, b); ++ return -1; ++} ++ ++static struct Qdisc_ops esfq_qdisc_ops = ++{ ++ .next = NULL, ++ .cl_ops = NULL, ++ .id = "esfq", ++ .priv_size = sizeof(struct esfq_sched_data), ++ .enqueue = esfq_enqueue, ++ .dequeue = esfq_dequeue, ++ .peek = esfq_peek, ++ .drop = esfq_drop, ++ .init = esfq_init, ++ .reset = esfq_reset, ++ .destroy = esfq_destroy, ++ .change = esfq_change, ++ .dump = esfq_dump, ++ .owner = THIS_MODULE, ++}; ++ ++static int __init esfq_module_init(void) ++{ ++ return register_qdisc(&esfq_qdisc_ops); ++} ++static void __exit esfq_module_exit(void) ++{ ++ unregister_qdisc(&esfq_qdisc_ops); ++} ++module_init(esfq_module_init) ++module_exit(esfq_module_exit) ++MODULE_LICENSE("GPL"); diff --git a/target/linux/generic-2.6/patches-2.6.31/201-jhash3.patch b/target/linux/generic-2.6/patches-2.6.31/201-jhash3.patch new file mode 100644 index 0000000000..0218fa19c5 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/201-jhash3.patch @@ -0,0 +1,227 @@ +--- a/include/linux/jhash.h ++++ b/include/linux/jhash.h +@@ -3,80 +3,95 @@ + + /* jhash.h: Jenkins hash support. + * +- * Copyright (C) 1996 Bob Jenkins (bob_jenkins@burtleburtle.net) ++ * Copyright (C) 2006. Bob Jenkins (bob_jenkins@burtleburtle.net) + * + * http://burtleburtle.net/bob/hash/ + * + * These are the credits from Bob's sources: + * +- * lookup2.c, by Bob Jenkins, December 1996, Public Domain. +- * hash(), hash2(), hash3, and mix() are externally useful functions. +- * Routines to test the hash are included if SELF_TEST is defined. +- * You can use this free for any purpose. It has no warranty. ++ * lookup3.c, by Bob Jenkins, May 2006, Public Domain. + * +- * Copyright (C) 2003 David S. Miller (davem@redhat.com) ++ * These are functions for producing 32-bit hashes for hash table lookup. ++ * hashword(), hashlittle(), hashlittle2(), hashbig(), mix(), and final() ++ * are externally useful functions. Routines to test the hash are included ++ * if SELF_TEST is defined. You can use this free for any purpose. It's in ++ * the public domain. It has no warranty. ++ * ++ * Copyright (C) 2009 Jozsef Kadlecsik (kadlec@blackhole.kfki.hu) + * + * I've modified Bob's hash to be useful in the Linux kernel, and +- * any bugs present are surely my fault. -DaveM ++ * any bugs present are my fault. Jozsef + */ + +-/* NOTE: Arguments are modified. */ +-#define __jhash_mix(a, b, c) \ ++#define __rot(x,k) (((x)<<(k)) | ((x)>>(32-(k)))) ++ ++/* __jhash_mix - mix 3 32-bit values reversibly. */ ++#define __jhash_mix(a,b,c) \ ++{ \ ++ a -= c; a ^= __rot(c, 4); c += b; \ ++ b -= a; b ^= __rot(a, 6); a += c; \ ++ c -= b; c ^= __rot(b, 8); b += a; \ ++ a -= c; a ^= __rot(c,16); c += b; \ ++ b -= a; b ^= __rot(a,19); a += c; \ ++ c -= b; c ^= __rot(b, 4); b += a; \ ++} ++ ++/* __jhash_final - final mixing of 3 32-bit values (a,b,c) into c */ ++#define __jhash_final(a,b,c) \ + { \ +- a -= b; a -= c; a ^= (c>>13); \ +- b -= c; b -= a; b ^= (a<<8); \ +- c -= a; c -= b; c ^= (b>>13); \ +- a -= b; a -= c; a ^= (c>>12); \ +- b -= c; b -= a; b ^= (a<<16); \ +- c -= a; c -= b; c ^= (b>>5); \ +- a -= b; a -= c; a ^= (c>>3); \ +- b -= c; b -= a; b ^= (a<<10); \ +- c -= a; c -= b; c ^= (b>>15); \ ++ c ^= b; c -= __rot(b,14); \ ++ a ^= c; a -= __rot(c,11); \ ++ b ^= a; b -= __rot(a,25); \ ++ c ^= b; c -= __rot(b,16); \ ++ a ^= c; a -= __rot(c,4); \ ++ b ^= a; b -= __rot(a,14); \ ++ c ^= b; c -= __rot(b,24); \ + } + +-/* The golden ration: an arbitrary value */ +-#define JHASH_GOLDEN_RATIO 0x9e3779b9 ++/* An arbitrary initial parameter */ ++#define JHASH_GOLDEN_RATIO 0xdeadbeef + + /* The most generic version, hashes an arbitrary sequence + * of bytes. No alignment or length assumptions are made about +- * the input key. ++ * the input key. The result depends on endianness. + */ + static inline u32 jhash(const void *key, u32 length, u32 initval) + { +- u32 a, b, c, len; ++ u32 a,b,c; + const u8 *k = key; + +- len = length; +- a = b = JHASH_GOLDEN_RATIO; +- c = initval; +- +- while (len >= 12) { +- a += (k[0] +((u32)k[1]<<8) +((u32)k[2]<<16) +((u32)k[3]<<24)); +- b += (k[4] +((u32)k[5]<<8) +((u32)k[6]<<16) +((u32)k[7]<<24)); +- c += (k[8] +((u32)k[9]<<8) +((u32)k[10]<<16)+((u32)k[11]<<24)); +- +- __jhash_mix(a,b,c); ++ /* Set up the internal state */ ++ a = b = c = JHASH_GOLDEN_RATIO + length + initval; + ++ /* all but the last block: affect some 32 bits of (a,b,c) */ ++ while (length > 12) { ++ a += (k[0] + ((u32)k[1]<<8) + ((u32)k[2]<<16) + ((u32)k[3]<<24)); ++ b += (k[4] + ((u32)k[5]<<8) + ((u32)k[6]<<16) + ((u32)k[7]<<24)); ++ c += (k[8] + ((u32)k[9]<<8) + ((u32)k[10]<<16) + ((u32)k[11]<<24)); ++ __jhash_mix(a, b, c); ++ length -= 12; + k += 12; +- len -= 12; + } + +- c += length; +- switch (len) { +- case 11: c += ((u32)k[10]<<24); +- case 10: c += ((u32)k[9]<<16); +- case 9 : c += ((u32)k[8]<<8); +- case 8 : b += ((u32)k[7]<<24); +- case 7 : b += ((u32)k[6]<<16); +- case 6 : b += ((u32)k[5]<<8); ++ /* last block: affect all 32 bits of (c) */ ++ /* all the case statements fall through */ ++ switch (length) { ++ case 12: c += (u32)k[11]<<24; ++ case 11: c += (u32)k[10]<<16; ++ case 10: c += (u32)k[9]<<8; ++ case 9 : c += k[8]; ++ case 8 : b += (u32)k[7]<<24; ++ case 7 : b += (u32)k[6]<<16; ++ case 6 : b += (u32)k[5]<<8; + case 5 : b += k[4]; +- case 4 : a += ((u32)k[3]<<24); +- case 3 : a += ((u32)k[2]<<16); +- case 2 : a += ((u32)k[1]<<8); ++ case 4 : a += (u32)k[3]<<24; ++ case 3 : a += (u32)k[2]<<16; ++ case 2 : a += (u32)k[1]<<8; + case 1 : a += k[0]; +- }; +- +- __jhash_mix(a,b,c); ++ __jhash_final(a, b, c); ++ case 0 : ++ break; ++ } + + return c; + } +@@ -86,58 +101,57 @@ static inline u32 jhash(const void *key, + */ + static inline u32 jhash2(const u32 *k, u32 length, u32 initval) + { +- u32 a, b, c, len; ++ u32 a, b, c; + +- a = b = JHASH_GOLDEN_RATIO; +- c = initval; +- len = length; ++ /* Set up the internal state */ ++ a = b = c = JHASH_GOLDEN_RATIO + (length<<2) + initval; + +- while (len >= 3) { ++ /* handle most of the key */ ++ while (length > 3) { + a += k[0]; + b += k[1]; + c += k[2]; + __jhash_mix(a, b, c); +- k += 3; len -= 3; ++ length -= 3; ++ k += 3; + } + +- c += length * 4; +- +- switch (len) { +- case 2 : b += k[1]; +- case 1 : a += k[0]; +- }; +- +- __jhash_mix(a,b,c); ++ /* handle the last 3 u32's */ ++ /* all the case statements fall through */ ++ switch (length) { ++ case 3: c += k[2]; ++ case 2: b += k[1]; ++ case 1: a += k[0]; ++ __jhash_final(a, b, c); ++ case 0: /* case 0: nothing left to add */ ++ break; ++ } + + return c; + } + +- + /* A special ultra-optimized versions that knows they are hashing exactly + * 3, 2 or 1 word(s). +- * +- * NOTE: In partilar the "c += length; __jhash_mix(a,b,c);" normally +- * done at the end is not done here. + */ + static inline u32 jhash_3words(u32 a, u32 b, u32 c, u32 initval) + { +- a += JHASH_GOLDEN_RATIO; +- b += JHASH_GOLDEN_RATIO; +- c += initval; ++ a += JHASH_GOLDEN_RATIO + initval; ++ b += JHASH_GOLDEN_RATIO + initval; ++ c += JHASH_GOLDEN_RATIO + initval; + +- __jhash_mix(a, b, c); ++ __jhash_final(a, b, c); + + return c; + } + + static inline u32 jhash_2words(u32 a, u32 b, u32 initval) + { +- return jhash_3words(a, b, 0, initval); ++ return jhash_3words(0, a, b, initval); + } + + static inline u32 jhash_1word(u32 a, u32 initval) + { +- return jhash_3words(a, 0, 0, initval); ++ return jhash_3words(0, 0, a, initval); + } + + #endif /* _LINUX_JHASH_H */ diff --git a/target/linux/generic-2.6/patches-2.6.31/202-mips-freestanding.patch b/target/linux/generic-2.6/patches-2.6.31/202-mips-freestanding.patch new file mode 100644 index 0000000000..021d072823 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/202-mips-freestanding.patch @@ -0,0 +1,12 @@ +--- a/arch/mips/Makefile ++++ b/arch/mips/Makefile +@@ -612,6 +612,9 @@ else + load-$(CONFIG_CPU_CAVIUM_OCTEON) += 0xffffffff81100000 + endif + ++# temporary until string.h is fixed ++cflags-y += -ffreestanding ++ + cflags-y += -I$(srctree)/arch/mips/include/asm/mach-generic + drivers-$(CONFIG_PCI) += arch/mips/pci/ + diff --git a/target/linux/generic-2.6/patches-2.6.31/203-slab_maxsize.patch b/target/linux/generic-2.6/patches-2.6.31/203-slab_maxsize.patch new file mode 100644 index 0000000000..0c1ae72a4e --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/203-slab_maxsize.patch @@ -0,0 +1,13 @@ +--- a/include/linux/slab.h ++++ b/include/linux/slab.h +@@ -124,8 +124,8 @@ int kmem_ptr_validate(struct kmem_cache + * to do various tricks to work around compiler limitations in order to + * ensure proper constant folding. + */ +-#define KMALLOC_SHIFT_HIGH ((MAX_ORDER + PAGE_SHIFT - 1) <= 25 ? \ +- (MAX_ORDER + PAGE_SHIFT - 1) : 25) ++#define KMALLOC_SHIFT_HIGH ((MAX_ORDER + PAGE_SHIFT - 1) <= 17 ? \ ++ (MAX_ORDER + PAGE_SHIFT - 1) : 17) + + #define KMALLOC_MAX_SIZE (1UL << KMALLOC_SHIFT_HIGH) + #define KMALLOC_MAX_ORDER (KMALLOC_SHIFT_HIGH - PAGE_SHIFT) diff --git a/target/linux/generic-2.6/patches-2.6.31/204-jffs2_eofdetect.patch b/target/linux/generic-2.6/patches-2.6.31/204-jffs2_eofdetect.patch new file mode 100644 index 0000000000..16d3e1f8e8 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/204-jffs2_eofdetect.patch @@ -0,0 +1,132 @@ +--- a/fs/jffs2/build.c ++++ b/fs/jffs2/build.c +@@ -111,6 +111,17 @@ static int jffs2_build_filesystem(struct + dbg_fsbuild("scanned flash completely\n"); + jffs2_dbg_dump_block_lists_nolock(c); + ++ if (c->flags & (1 << 7)) { ++ printk("%s(): unlocking the mtd device... ", __func__); ++ if (c->mtd->unlock) ++ c->mtd->unlock(c->mtd, 0, c->mtd->size); ++ printk("done.\n"); ++ ++ printk("%s(): erasing all blocks after the end marker... ", __func__); ++ jffs2_erase_pending_blocks(c, -1); ++ printk("done.\n"); ++ } ++ + dbg_fsbuild("pass 1 starting\n"); + c->flags |= JFFS2_SB_FLAG_BUILDING; + /* Now scan the directory tree, increasing nlink according to every dirent found. */ +--- a/fs/jffs2/scan.c ++++ b/fs/jffs2/scan.c +@@ -72,7 +72,7 @@ static int file_dirty(struct jffs2_sb_in + return ret; + if ((ret = jffs2_scan_dirty_space(c, jeb, jeb->free_size))) + return ret; +- /* Turned wasted size into dirty, since we apparently ++ /* Turned wasted size into dirty, since we apparently + think it's recoverable now. */ + jeb->dirty_size += jeb->wasted_size; + c->dirty_size += jeb->wasted_size; +@@ -144,8 +144,11 @@ int jffs2_scan_medium(struct jffs2_sb_in + /* reset summary info for next eraseblock scan */ + jffs2_sum_reset_collected(s); + +- ret = jffs2_scan_eraseblock(c, jeb, buf_size?flashbuf:(flashbuf+jeb->offset), +- buf_size, s); ++ if (c->flags & (1 << 7)) ++ ret = BLK_STATE_ALLFF; ++ else ++ ret = jffs2_scan_eraseblock(c, jeb, buf_size?flashbuf:(flashbuf+jeb->offset), ++ buf_size, s); + + if (ret < 0) + goto out; +@@ -400,7 +403,7 @@ static int jffs2_scan_xref_node(struct j + if (!ref) + return -ENOMEM; + +- /* BEFORE jffs2_build_xattr_subsystem() called, ++ /* BEFORE jffs2_build_xattr_subsystem() called, + * and AFTER xattr_ref is marked as a dead xref, + * ref->xid is used to store 32bit xid, xd is not used + * ref->ino is used to store 32bit inode-number, ic is not used +@@ -473,7 +476,7 @@ static int jffs2_scan_eraseblock (struct + struct jffs2_sum_marker *sm; + void *sumptr = NULL; + uint32_t sumlen; +- ++ + if (!buf_size) { + /* XIP case. Just look, point at the summary if it's there */ + sm = (void *)buf + c->sector_size - sizeof(*sm); +@@ -489,9 +492,9 @@ static int jffs2_scan_eraseblock (struct + buf_len = sizeof(*sm); + + /* Read as much as we want into the _end_ of the preallocated buffer */ +- err = jffs2_fill_scan_buf(c, buf + buf_size - buf_len, ++ err = jffs2_fill_scan_buf(c, buf + buf_size - buf_len, + jeb->offset + c->sector_size - buf_len, +- buf_len); ++ buf_len); + if (err) + return err; + +@@ -510,9 +513,9 @@ static int jffs2_scan_eraseblock (struct + } + if (buf_len < sumlen) { + /* Need to read more so that the entire summary node is present */ +- err = jffs2_fill_scan_buf(c, sumptr, ++ err = jffs2_fill_scan_buf(c, sumptr, + jeb->offset + c->sector_size - sumlen, +- sumlen - buf_len); ++ sumlen - buf_len); + if (err) + return err; + } +@@ -525,7 +528,7 @@ static int jffs2_scan_eraseblock (struct + + if (buf_size && sumlen > buf_size) + kfree(sumptr); +- /* If it returns with a real error, bail. ++ /* If it returns with a real error, bail. + If it returns positive, that's a block classification + (i.e. BLK_STATE_xxx) so return that too. + If it returns zero, fall through to full scan. */ +@@ -546,6 +549,17 @@ static int jffs2_scan_eraseblock (struct + return err; + } + ++ if ((buf[0] == 0xde) && ++ (buf[1] == 0xad) && ++ (buf[2] == 0xc0) && ++ (buf[3] == 0xde)) { ++ /* end of filesystem. erase everything after this point */ ++ printk("%s(): End of filesystem marker found at 0x%x\n", __func__, jeb->offset); ++ c->flags |= (1 << 7); ++ ++ return BLK_STATE_ALLFF; ++ } ++ + /* We temporarily use 'ofs' as a pointer into the buffer/jeb */ + ofs = 0; + +@@ -671,7 +685,7 @@ scan_more: + scan_end = buf_len; + goto more_empty; + } +- ++ + /* See how much more there is to read in this eraseblock... */ + buf_len = min_t(uint32_t, buf_size, jeb->offset + c->sector_size - ofs); + if (!buf_len) { +@@ -907,7 +921,7 @@ scan_more: + + D1(printk(KERN_DEBUG "Block at 0x%08x: free 0x%08x, dirty 0x%08x, unchecked 0x%08x, used 0x%08x, wasted 0x%08x\n", + jeb->offset,jeb->free_size, jeb->dirty_size, jeb->unchecked_size, jeb->used_size, jeb->wasted_size)); +- ++ + /* mark_node_obsolete can add to wasted !! */ + if (jeb->wasted_size) { + jeb->dirty_size += jeb->wasted_size; diff --git a/target/linux/generic-2.6/patches-2.6.31/205-skb_padding.patch b/target/linux/generic-2.6/patches-2.6.31/205-skb_padding.patch new file mode 100644 index 0000000000..80c7e704e8 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/205-skb_padding.patch @@ -0,0 +1,56 @@ +--- a/include/linux/skbuff.h ++++ b/include/linux/skbuff.h +@@ -1387,11 +1387,18 @@ static inline int skb_network_offset(con + * + * Various parts of the networking layer expect at least 32 bytes of + * headroom, you should not reduce this. ++ * ++ * This has been changed to 64 to acommodate for routing between ethernet ++ * and wireless, but only for new allocations + */ + #ifndef NET_SKB_PAD + #define NET_SKB_PAD 32 + #endif + ++#ifndef NET_SKB_PAD_ALLOC ++#define NET_SKB_PAD_ALLOC 64 ++#endif ++ + extern int ___pskb_trim(struct sk_buff *skb, unsigned int len); + + static inline void __skb_trim(struct sk_buff *skb, unsigned int len) +@@ -1481,9 +1488,9 @@ static inline void __skb_queue_purge(str + static inline struct sk_buff *__dev_alloc_skb(unsigned int length, + gfp_t gfp_mask) + { +- struct sk_buff *skb = alloc_skb(length + NET_SKB_PAD, gfp_mask); ++ struct sk_buff *skb = alloc_skb(length + NET_SKB_PAD_ALLOC, gfp_mask); + if (likely(skb)) +- skb_reserve(skb, NET_SKB_PAD); ++ skb_reserve(skb, NET_SKB_PAD_ALLOC); + return skb; + } + +@@ -1556,7 +1563,7 @@ static inline int __skb_cow(struct sk_bu + delta = headroom - skb_headroom(skb); + + if (delta || cloned) +- return pskb_expand_head(skb, ALIGN(delta, NET_SKB_PAD), 0, ++ return pskb_expand_head(skb, ALIGN(delta, NET_SKB_PAD_ALLOC), 0, + GFP_ATOMIC); + return 0; + } +--- a/net/core/skbuff.c ++++ b/net/core/skbuff.c +@@ -336,9 +336,9 @@ struct sk_buff *__netdev_alloc_skb(struc + int node = dev->dev.parent ? dev_to_node(dev->dev.parent) : -1; + struct sk_buff *skb; + +- skb = __alloc_skb(length + NET_SKB_PAD, gfp_mask, 0, node); ++ skb = __alloc_skb(length + NET_SKB_PAD_ALLOC, gfp_mask, 0, node); + if (likely(skb)) { +- skb_reserve(skb, NET_SKB_PAD); ++ skb_reserve(skb, NET_SKB_PAD_ALLOC); + skb->dev = dev; + } + return skb; diff --git a/target/linux/generic-2.6/patches-2.6.31/207-powerpc_asm_segment_h.patch b/target/linux/generic-2.6/patches-2.6.31/207-powerpc_asm_segment_h.patch new file mode 100644 index 0000000000..1272e82c75 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/207-powerpc_asm_segment_h.patch @@ -0,0 +1,9 @@ +--- /dev/null ++++ b/include/asm-powerpc/segment.h +@@ -0,0 +1,6 @@ ++#ifndef _ASM_SEGMENT_H ++#define _ASM_SEGMENT_H ++ ++/* Only here because we have some old header files that expect it.. */ ++ ++#endif /* _ASM_SEGMENT_H */ diff --git a/target/linux/generic-2.6/patches-2.6.31/209-mini_fo.patch b/target/linux/generic-2.6/patches-2.6.31/209-mini_fo.patch new file mode 100644 index 0000000000..e3ce162ad7 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/209-mini_fo.patch @@ -0,0 +1,7780 @@ +--- a/fs/Kconfig ++++ b/fs/Kconfig +@@ -178,6 +178,7 @@ source "fs/ubifs/Kconfig" + source "fs/cramfs/Kconfig" + source "fs/squashfs/Kconfig" + source "fs/freevxfs/Kconfig" ++source "fs/mini_fo/Kconfig" + source "fs/minix/Kconfig" + source "fs/omfs/Kconfig" + source "fs/hpfs/Kconfig" +--- a/fs/Makefile ++++ b/fs/Makefile +@@ -77,6 +77,7 @@ obj-$(CONFIG_SQUASHFS) += squashfs/ + obj-y += ramfs/ + obj-$(CONFIG_HUGETLBFS) += hugetlbfs/ + obj-$(CONFIG_CODA_FS) += coda/ ++obj-$(CONFIG_MINI_FO) += mini_fo/ + obj-$(CONFIG_MINIX_FS) += minix/ + obj-$(CONFIG_FAT_FS) += fat/ + obj-$(CONFIG_BFS_FS) += bfs/ +--- /dev/null ++++ b/fs/mini_fo/aux.c +@@ -0,0 +1,577 @@ ++/* ++ * Copyright (c) 1997-2003 Erez Zadok ++ * Copyright (c) 2001-2003 Stony Brook University ++ * ++ * For specific licensing information, see the COPYING file distributed with ++ * this package, or get one from ftp://ftp.filesystems.org/pub/fist/COPYING. ++ * ++ * This Copyright notice must be kept intact and distributed with all ++ * fistgen sources INCLUDING sources generated by fistgen. ++ */ ++/* ++ * Copyright (C) 2004, 2005 Markus Klotzbuecher ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ */ ++/* ++ * $Id$ ++ */ ++ ++#ifdef HAVE_CONFIG_H ++# include ++#endif ++ ++#include "fist.h" ++#include "mini_fo.h" ++ ++/* check if file exists in storage */ ++int exists_in_storage(dentry_t *dentry) ++{ ++ check_mini_fo_dentry(dentry); ++ if(dtost(dentry) == MODIFIED || dtost(dentry) == CREATED || dtost(dentry) == DEL_REWRITTEN) ++ return 1; ++ return 0; ++} ++ ++/* check if dentry is in an existing state */ ++int is_mini_fo_existant(dentry_t *dentry) ++{ ++ check_mini_fo_dentry(dentry); ++ ++ if(dtost(dentry) == DELETED || dtost(dentry) == NON_EXISTANT) ++ return 0; ++ else ++ return 1; ++} ++ ++/* ++ * This function will create a negative storage dentry for ++ * dentry, what is required for many create like options. ++ * It will create the storage structure if necessary. ++ */ ++int get_neg_sto_dentry(dentry_t *dentry) ++{ ++ int err = 0; ++ unsigned int len; ++ const unsigned char *name; ++ ++ if(!dentry || ++ !dtopd(dentry) || ++ !(dtost(dentry) == UNMODIFIED || ++ dtost(dentry) == NON_EXISTANT || ++ dtost(dentry) == DELETED)) { ++ printk(KERN_CRIT "mini_fo: get_neg_sto_dentry: invalid dentry passed.\n"); ++ err = -1; ++ goto out; ++ } ++ /* Have we got a neg. dentry already? */ ++ if(dtohd2(dentry)) { ++ err = 0; ++ goto out; ++ } ++ if(dtost(dentry->d_parent) == UNMODIFIED) { ++ /* build sto struct */ ++ err = build_sto_structure(dentry->d_parent->d_parent, dentry->d_parent); ++ if(err || ++ dtost(dentry->d_parent) != MODIFIED) { ++ printk(KERN_CRIT "mini_fo: get_neg_sto_dentry: ERROR building sto structure.\n"); ++ err = -1; ++ goto out; ++ } ++ } ++ ++ len = dentry->d_name.len; ++ name = dentry->d_name.name; ++ ++ dtohd2(dentry) = ++ lookup_one_len(name, dtohd2(dentry->d_parent), len); ++ ++ out: ++ return err; ++} ++ ++int check_mini_fo_dentry(dentry_t *dentry) ++{ ++ ASSERT(dentry != NULL); ++ ASSERT(dtopd(dentry) != NULL); ++ ASSERT((dtohd(dentry) != NULL) || (dtohd2(dentry) != NULL)); ++ ++/* if(dtost(dentry) == MODIFIED) { */ ++/* ASSERT(dentry->d_inode != NULL); */ ++/* ASSERT(dtohd(dentry) != NULL); */ ++/* ASSERT(dtohd(dentry)->d_inode != NULL); */ ++/* ASSERT(dtohd2(dentry) != NULL); */ ++/* ASSERT(dtohd2(dentry)->d_inode != NULL); */ ++/* } */ ++/* else if(dtost(dentry) == UNMODIFIED) { */ ++/* ASSERT(dentry->d_inode != NULL); */ ++/* ASSERT( */ ++/* } */ ++ return 0; ++} ++ ++int check_mini_fo_file(file_t *file) ++{ ++ ASSERT(file != NULL); ++ ASSERT(ftopd(file) != NULL); ++ ASSERT(file->f_dentry != NULL); ++ ++ /* violent checking, check depending of state and type ++ * if(S_ISDIR(file->f_dentry->d_inode->i_mode)) {} ++ */ ++ ASSERT((ftohf(file) != NULL) || (ftohf2(file) != NULL)); ++ return 0; ++} ++ ++int check_mini_fo_inode(inode_t *inode) ++{ ++ ASSERT(inode != NULL); ++ ASSERT(itopd(inode) != NULL); ++ ASSERT((itohi(inode) != NULL) || (itohi2(inode) != NULL)); ++ return 0; ++} ++ ++/* ++ * will walk a base path as provided by get_mini_fo_bpath and return ++ * the (hopefully ;-) ) positive dentry of the renamed base dir. ++ * ++ * This does some work of path_init. ++ */ ++dentry_t *bpath_walk(super_block_t *sb, char *bpath) ++{ ++ int err; ++ struct vfsmount *mnt; ++ struct nameidata nd; ++ ++ /* be paranoid */ ++ if(!bpath || bpath[0] != '/') { ++ printk(KERN_CRIT "mini_fo: bpath_walk: Invalid string.\n"); ++ return NULL; ++ } ++ if(!sb || !stopd(sb)) { ++ printk(KERN_CRIT "mini_fo: bpath_walk: Invalid sb.\n"); ++ return NULL; ++ } ++ ++ /* fix this: how do I reach this lock? ++ * read_lock(¤t->fs->lock); */ ++ mnt = mntget(stopd(sb)->hidden_mnt); ++ /* read_unlock(¤t->fs->lock); */ ++ ++ err = vfs_path_lookup(mnt->mnt_root, mnt, bpath+1, 0, &nd); ++ ++ /* validate */ ++ if (err || !nd.dentry || !nd.dentry->d_inode) { ++ printk(KERN_CRIT "mini_fo: bpath_walk: path_walk failed.\n"); ++ return NULL; ++ } ++ return nd.dentry; ++} ++ ++ ++/* returns the full path of the basefile incl. its name */ ++int get_mini_fo_bpath(dentry_t *dentry, char **bpath, int *bpath_len) ++{ ++ char *buf_walker; ++ int len = 0; ++ dentry_t *sky_walker; ++ ++ if(!dentry || !dtohd(dentry)) { ++ printk(KERN_CRIT "mini_fo: get_mini_fo_bpath: invalid dentry passed.\n"); ++ return -1; ++ } ++ sky_walker = dtohd(dentry); ++ ++ do { ++ len += sky_walker->d_name.len + 1 ; /* 1 for '/' */ ++ sky_walker = sky_walker->d_parent; ++ } while(sky_walker != stopd(dentry->d_inode->i_sb)->base_dir_dentry); ++ ++ /* 1 to oil the loop */ ++ *bpath = (char*) kmalloc(len + 1, GFP_KERNEL); ++ if(!*bpath) { ++ printk(KERN_CRIT "mini_fo: get_mini_fo_bpath: out of mem.\n"); ++ return -1; ++ } ++ buf_walker = *bpath+len; /* put it on last char */ ++ *buf_walker = '\n'; ++ sky_walker = dtohd(dentry); ++ ++ do { ++ buf_walker -= sky_walker->d_name.len; ++ strncpy(buf_walker, ++ sky_walker->d_name.name, ++ sky_walker->d_name.len); ++ *(--buf_walker) = '/'; ++ sky_walker = sky_walker->d_parent; ++ } while(sky_walker != stopd(dentry->d_inode->i_sb)->base_dir_dentry); ++ ++ /* bpath_len doesn't count newline! */ ++ *bpath_len = len; ++ return 0; ++} ++ ++int mini_fo_cp_cont(dentry_t *tgt_dentry, struct vfsmount *tgt_mnt, ++ dentry_t *src_dentry, struct vfsmount *src_mnt) ++{ ++ void *buf; ++ mm_segment_t old_fs; ++ file_t *tgt_file; ++ file_t *src_file; ++ int bytes, len, tmp, err; ++ err = 0; ++ ++ if(!(tgt_dentry->d_inode && src_dentry->d_inode)) { ++ printk(KERN_CRIT "mini_fo_cp_cont: ERROR, neg. dentry passed.\n"); ++ err = -EINVAL; ++ goto out; ++ } ++ ++ dget(tgt_dentry); ++ dget(src_dentry); ++ mntget(tgt_mnt); ++ mntget(src_mnt); ++ ++ /* open file write only */ ++ tgt_file = dentry_open(tgt_dentry, tgt_mnt, 0x1); ++ if(!tgt_file || IS_ERR(tgt_file)) { ++ printk(KERN_CRIT "mini_fo_cp_cont: ERROR opening target file.\n"); ++ err = PTR_ERR(tgt_file); ++ goto out_err; ++ } ++ ++ /* open file read only */ ++ src_file = dentry_open(src_dentry, src_mnt, 0x0); ++ if(!src_file || IS_ERR(src_file)) { ++ printk(KERN_CRIT "mini_fo_cp_cont: ERROR opening source file.\n"); ++ err = PTR_ERR(src_file); ++ ++ /* close target file */ ++ fput(tgt_file); ++ goto out_err; ++ } ++ ++ /* check if the filesystem(s) support read respective write */ ++ if(!src_file->f_op->read || !tgt_file->f_op->write) { ++ printk(KERN_CRIT "mini_fo_cp_cont: ERROR, no fs read or write support.\n"); ++ err = -EPERM; ++ goto out_close; ++ } ++ ++ /* allocate a page for transfering the data */ ++ buf = (void *) __get_free_page(GFP_KERNEL); ++ if(!buf) { ++ printk(KERN_CRIT "mini_fo_cp_cont: ERROR, out of kernel mem.\n"); ++ goto out_err; ++ } ++ ++ tgt_file->f_pos = 0; ++ src_file->f_pos = 0; ++ ++ old_fs = get_fs(); ++ set_fs(KERNEL_DS); ++ ++ /* Doing this I assume that a read operation will return a full ++ * buffer while there is still data to read, and a less than ++ * full buffer when all data has been read. ++ */ ++ bytes = len = PAGE_SIZE; ++ while(bytes == len) { ++ bytes = src_file->f_op->read(src_file, buf, len, ++ &src_file->f_pos); ++ tmp = tgt_file->f_op->write(tgt_file, buf, bytes, ++ &tgt_file->f_pos); ++ if(tmp != bytes) { ++ printk(KERN_CRIT "mini_fo_cp_cont: ERROR writing.\n"); ++ goto out_close_unset; ++ } ++ } ++ ++ free_page((unsigned long) buf); ++ set_fs(old_fs); ++ fput(tgt_file); ++ fput(src_file); ++ goto out; ++ ++ out_close_unset: ++ free_page((unsigned long) buf); ++ set_fs(old_fs); ++ ++ out_close: ++ fput(tgt_file); ++ fput(src_file); ++ ++ out_err: ++ dput(tgt_dentry); ++ dput(src_dentry); ++ ++ /* mk: not sure if this need to be done */ ++ mntput(tgt_mnt); ++ mntput(src_mnt); ++ ++ out: ++ return err; ++} ++ ++/* mk: ++ * ndl (no-duplicate list) stuff ++ * This is used in mini_fo_readdir, to save the storage directory contents ++ * and later when reading base, match them against the list in order ++ * to avoid duplicates. ++ */ ++ ++/* add a file specified by name and len to the ndl ++ * Return values: 0 on success, <0 on failure. ++ */ ++int ndl_add_entry(struct readdir_data *rd, const char *name, int len) ++{ ++ struct ndl_entry *tmp_entry; ++ ++ tmp_entry = (struct ndl_entry *) ++ kmalloc(sizeof(struct ndl_entry), GFP_KERNEL); ++ if(!tmp_entry) { ++ printk(KERN_CRIT "mini_fo: ndl_add_entry: out of mem.\n"); ++ return -ENOMEM; ++ } ++ tmp_entry->name = (char*) kmalloc(len, GFP_KERNEL); ++ if(!tmp_entry->name) { ++ printk(KERN_CRIT "mini_fo: ndl_add_entry: out of mem.\n"); ++ return -ENOMEM; ++ } ++ strncpy(tmp_entry->name, name, len); ++ tmp_entry->len = len; ++ ++ list_add(&tmp_entry->list, &rd->ndl_list); ++ rd->ndl_size++; ++ return 0; ++} ++ ++/* delete all list entries and free memory */ ++void ndl_put_list(struct readdir_data *rd) ++{ ++ struct list_head *tmp; ++ struct ndl_entry *tmp_entry; ++ ++ if(rd->ndl_size <= 0) ++ return; ++ while(!list_empty(&rd->ndl_list)) { ++ tmp = rd->ndl_list.next; ++ list_del(tmp); ++ tmp_entry = list_entry(tmp, struct ndl_entry, list); ++ kfree(tmp_entry->name); ++ kfree(tmp_entry); ++ } ++ rd->ndl_size = 0; ++} ++ ++/* Check if a file specified by name and len is in the ndl ++ * Return value: 0 if not in list, 1 if file is found in ndl. ++ */ ++int ndl_check_entry(struct readdir_data *rd, const char *name, int len) ++{ ++ struct list_head *tmp; ++ struct ndl_entry *tmp_entry; ++ ++ if(rd->ndl_size <= 0) ++ return 0; ++ ++ list_for_each(tmp, &rd->ndl_list) { ++ tmp_entry = list_entry(tmp, struct ndl_entry, list); ++ if(tmp_entry->len != len) ++ continue; ++ if(!strncmp(tmp_entry->name, name, len)) ++ return 1; ++ } ++ return 0; ++} ++ ++/* mk: ++ * Recursive function to create corresponding directorys in the storage fs. ++ * The function will build the storage directorys up to dentry. ++ */ ++int build_sto_structure(dentry_t *dir, dentry_t *dentry) ++{ ++ int err; ++ dentry_t *hidden_sto_dentry; ++ dentry_t *hidden_sto_dir_dentry; ++ ++ if(dentry->d_parent != dir) { ++ printk(KERN_CRIT "mini_fo: build_sto_structure: invalid parameter or meta data corruption [1].\n"); ++ return 1; ++ } ++ ++ if(dtost(dir) != MODIFIED) { ++ err = build_sto_structure(dir->d_parent, dentry->d_parent); ++ if(err) ++ return err; ++ } ++ ++ /* ok, coming back again. */ ++ check_mini_fo_dentry(dentry); ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ if(!hidden_sto_dentry) { ++ /* ++ * This is the case after creating the first ++ * hidden_sto_dentry. ++ * After one negative storage_dentry, all pointers to ++ * hidden_storage dentries are set to NULL. We need to ++ * create the negative dentry before we create the storage ++ * file. ++ */ ++ unsigned int len; ++ const unsigned char *name; ++ len = dtohd(dentry)->d_name.len; ++ name = dtohd(dentry)->d_name.name; ++ hidden_sto_dentry = lookup_one_len(name, dtohd2(dir), len); ++ dtohd2(dentry) = hidden_sto_dentry; ++ } ++ ++ /* was: hidden_sto_dir_dentry = lock_parent(hidden_sto_dentry); */ ++ hidden_sto_dir_dentry = dget(hidden_sto_dentry->d_parent); ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ /* lets be safe */ ++ if(dtohd2(dir) != hidden_sto_dir_dentry) { ++ printk(KERN_CRIT "mini_fo: build_sto_structure: invalid parameter or meta data corruption [2].\n"); ++ return 1; ++ } ++ ++ /* check for errors in lock_parent */ ++ err = PTR_ERR(hidden_sto_dir_dentry); ++ if(IS_ERR(hidden_sto_dir_dentry)) { ++ printk(KERN_CRIT "mini_fo: build_sto_structure: lock_parent failed.\n"); ++ return err; ++ } ++ ++ err = vfs_mkdir(hidden_sto_dir_dentry->d_inode, ++ hidden_sto_dentry, ++ dir->d_inode->i_mode); ++ ++ if(err) { ++ printk(KERN_CRIT "mini_fo: build_sto_structure: failed to create storage dir [1].\n"); ++ /* was: unlock_dir(dir); */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&dir->d_inode->i_mutex); ++#else ++ up(&dir->d_inode->i_sem); ++#endif ++ dput(dir); ++ return err; ++ } ++ ++ /* everything ok! */ ++ if(!dtohd2(dentry)->d_inode) { ++ printk(KERN_CRIT "mini_fo: build_sto_structure: failed to create storage dir [2].\n"); ++ /* was: unlock_dir(dir); */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&dir->d_inode->i_mutex); ++#else ++ up(&dir->d_inode->i_sem); ++#endif ++ dput(dir); ++ return 1; ++ } ++ ++ /* interpose the new inode and set new state */ ++ itohi2(dentry->d_inode) = igrab(dtohd2(dentry)->d_inode); ++ dtopd(dentry)->state = MODIFIED; ++ ++ /* initalize the wol list */ ++ itopd(dentry->d_inode)->deleted_list_size = -1; ++ itopd(dentry->d_inode)->renamed_list_size = -1; ++ meta_build_lists(dentry); ++ ++ fist_copy_attr_all(dentry->d_inode, itohi2(dentry->d_inode)); ++ fist_copy_attr_timesizes(dir->d_inode, ++ hidden_sto_dir_dentry->d_inode); ++ dir->d_inode->i_nlink++; ++ /* was: unlock_dir(hidden_sto_dir_dentry); */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ dput(hidden_sto_dir_dentry); ++ return 0; ++} ++ ++ ++#if 0 /* unused */ ++ ++/* ++ * Read "len" bytes from "filename" into "buf". ++ * "buf" is in kernel space. ++ */ ++int ++mini_fo_read_file(const char *filename, void *buf, int len) ++{ ++ file_t *filp; ++ mm_segment_t oldfs; ++ int bytes; ++ /* Chroot? Maybe NULL isn't right here */ ++ filp = filp_open(filename, O_RDONLY, 0); ++ if (!filp || IS_ERR(filp)) { ++ printk("mini_fo_read_file err %d\n", (int) PTR_ERR(filp)); ++ return -1; /* or do something else */ ++ } ++ ++ if (!filp->f_op->read) ++ return -2; /* file(system) doesn't allow reads */ ++ ++ /* now read len bytes from offset 0 */ ++ filp->f_pos = 0; /* start offset */ ++ oldfs = get_fs(); ++ set_fs(KERNEL_DS); ++ bytes = filp->f_op->read(filp, buf, len, &filp->f_pos); ++ set_fs(oldfs); ++ ++ /* close the file */ ++ fput(filp); ++ ++ return bytes; ++} ++ ++ ++ ++/* ++ * Write "len" bytes from "buf" to "filename" ++ * "buf" is in kernel space. ++ */ ++int ++mini_fo_write_file(const char *filename, void *buf, int len) ++{ ++ file_t *filp; ++ mm_segment_t oldfs; ++ int bytes; ++ /* Chroot? Maybe NULL isn't right here */ ++ filp = filp_open(filename, O_RDWR|O_CREAT, 0640); ++ if (!filp || IS_ERR(filp)) { ++ printk("mini_fo_write_file err %d\n", (int) PTR_ERR(filp)); ++ return -1; /* or do something else */ ++ } ++ ++ if (!filp->f_op->write) ++ return -2; /* file(system) doesn't allow writes */ ++ ++ /* now write len bytes from offset 0 */ ++ filp->f_pos = 0; /* start offset */ ++ oldfs = get_fs(); ++ set_fs(KERNEL_DS); ++ bytes = filp->f_op->write(filp, buf, len, &filp->f_pos); ++ set_fs(oldfs); ++ ++ /* close the file */ ++ fput(filp); ++ ++ return bytes; ++} ++ ++#endif /* unused */ ++ +--- /dev/null ++++ b/fs/mini_fo/ChangeLog +@@ -0,0 +1,281 @@ ++2006-01-24 Markus Klotzbuecher ++ ++ * Add tons of ugly ifdefs to Ed L. Cashin's mutex patch to ++ retain backwards compatibility. ++ ++2006-01-24 Ed L. Cashin ++ ++ * Support for the new mutex infrastructure ++ (7892f2f48d165a34b0b8130c8a195dfd807b8cb6) ++ ++2005-10-15 Markus Klotzbuecher ++ ++ * Bugfix for a serious memory leak in mini_fo_follow_link. ++ ++2005-09-21 Markus Klotzbuecher ++ ++ * new release 0.6.1 ++ ++ * fix of a compiler warning due to changes in 2.6.13 ++ ++2005-09-21 Klaus Wenninger ++ ++ * file.c: readdir: fix for a bug that caused directory entries ++ to show up twice when using storage filesystems such as ++ minixfs or pramfs. ++ ++2005-06-30 Eric Lammerts ++ ++ * fix for an oops when overwriting a binary thats beeing ++ executed. ++ ++2005-06-09 ++ ++ * Renamed overlay to mini_fo-overlay. ++ ++ * Added mini_fo-merge script to allow merging of storage and base ++ after making modifications. ++ ++2005-05-22 root ++ ++ * Added overlay script that allows to easily mount mini_fo ontop ++ of a given base directory ++ ++2005-05-10 ++ ++ * inode.c: xattr functions return -EOPNOSUPP instead of ++ -ENOSUPP, what confuses "ls -l" ++ ++ * Changed license from LGPL to GPL. ++ ++2005-05-08 root ++ ++ * Makefile: clean it up and added make install and make ++ uninstall. ++ ++2005-05-06 ++ ++ * merged devel branch back to main. [v0-6-0-pre3] ++ ++ * removed unused files print.c and fist_ioctl. [devel-0-0-18] ++ ++ * ioctl: removed fist_ioctl stuff, that is not needed for ++ now. ++ ++2005-05-03 ++ ++ * file.c: simplified mini_fo_open and mini_fo_setattr using ++ new state changing functions. [devel-0-0-17] ++ ++ * inode.c: Fixed getattr state bug (see below) in 2.4 function ++ mini_fo_inode revalidate. ++ ++ * inode.c: found an other bug in mini_fo_getattr. States are not ++ reliable in this function, as a file can be opened, unlinked and ++ the getattr function called. This results in a deleted dentry ++ with an inode. Fix is to ignore states and simply use the inode ++ available. ++ ++2005-04-29 ++ ++ * file.c: Bugfix and cleanup in fasync and fsync. [devel-0-0-16] ++ ++ * file.c: do not use mini_fo_lock so the generic version is ++ used (I guess). ++ ++ * inode.c: getattr, never call getattr on lower files, as this ++ will cause the inum to change. ++ ++ * inode.c: rename_reg_file renamed to rename_nondir, as it ++ doesn't matter as long it't not a dir. Removed all ++ rename_dev_file etc. ++ ++ * tagged as devel-0-0-15 ++ ++ * inode.c: added support for chosing support for extended ++ attrs at compile time by XATTR define in mini_fo.h . ++ ++ * inode.c: fixed mini_fo_getattr to use mini_fo inode and not ++ lower again, what avoids inode number changes that confused ++ rm again. This is the proper solution. ++ ++2005-04-24 ++ ++ * all files: updated Copyright notive to 2005. [devel-0-0-14] ++ ++ * inode.c: fixed mini_fo_getattr to not change the inode ++ number, even if lower files change. ++ ++ * super.c: fixed a bug that caused deleted base file to show ++ up suddenly after some time, or after creating a special ++ file. The problem was that after some time or after special ++ file creating sync_sb_inodes is called by the vfs, that ++ called our mini_fo_put_inode. There was (wrongly) called ++ __meta_put_lists, that nuked the lists, although the inode ++ was going to continue its life. Moving __meta_put_lists to ++ mini_fo_clear_inode, where an inode is really destroyed, ++ solved the problem. ++ ++ ++2005-04-23 ++ ++ * state.c, aux.c: more cleaning up and ++ simplifications. [devel-0-0-13] ++ ++ * inode.c: implemented mini_fo_getattr, that was required for ++ 2.6 because inode_revalidate has been remove there, and the ++ old "du" bug returned. ++ ++ ++2005-04-20 ++ ++ * aux.c: get_neg_sto_dentry(): allow to be called for dentries ++ in state UNMODIFIED, NON_EXISTANT _and_ DELETED. ++ ++2005-04-19 ++ ++ * Fixed a bug under 2.6 that caused files deleted via mini_fo ++ not to be deleted properly and therefore the fs filled up ++ untill no memory was left. [devel-0-0-12] ++ ++ * Added basic hard link support. This means that creating ++ hardlinks will work, but existing ones will be treated as ++ individual files. [devel-0-0-11] ++ ++2005-04-17 ++ ++ * Bugfixes ++ ++2005-04-13 root ++ ++ * Added file state.c for the state transition ++ functions. Doesn't work very well yet, though... ++ ++2005-04-12 ++ ++ * Porting to 2.6 started, which is easier than expected, also ++ due to Olivier previous work. ++ ++2005-04-08 ++ ++ * Fixed the bug that caused du to return invalid sizes of ++ directory trees. The problem was that ++ mini_fo_inode_revalidate didn't always copy the attributes ++ from the base inode properly. ++ ++2005-04-01 Markus Klotzbuecher ++ ++ * Merged devel branch back to main trunk and updated the ++ RELEASE notes. This will be 0-6-0-pre1. ++ ++2005-03-31 Markus Klotzbuecher ++ ++ * Fixed some bugs in rename_reg_file, that only showed up in ++ the kernel compile test. Kernel compiles cleanly ontop of ++ mini_fo, now also make mrproper etc. work. Seems pretty stable. ++ ++2005-03-28 Markus Klotzbuecher ++ ++ * Many, many directory renaming bugfixes and a lot of other ++ cleanup. Dir renaming seems to work relatively stable. ++ ++2005-03-22 Markus Klotzbuecher ++ ++ * Finished implementing lightweight directory renaming. Some ++ basic testing indicates it works fine. ++ Next is to implement testcases for the testsuite and confirm ++ everything is really working ok. ++ ++2005-03-18 Markus Klotzbuecher ++ ++ * Finished implementing meta.c stuff required for directory ++ renaming. ++ ++2005-03-17 Markus Klotzbuecher ++ ++ * Fixed all compile warnings + an extremly old bug that ++ somehow crept in while reworking the wol stuff to the META ++ system. Turning on -Werror again... :-) ++ ++ * Fixed some bugs in the new rename_reg_file function. ++ ++ * Rewrote mini_fo rename and split it into several ++ subfunctions, that handle the different types ++ seperately. Rewrote the regular file function aswell, as it ++ was implemented somewhat inefficient. ++ ++2005-03-16 Markus Klotzbuecher ++ ++ * Implemented new META subsystem, removed old WOL stuff in favor ++ if it. ++ ++ * After some basic testing everything seems ok... ++ ++2005-03-11 Markus Klotzbuecher ++ ++ * Renaming a non regular file caused trouble because I always ++ tried to copy the contents. Now I only do this for regular ++ files. mini_fo_rename still isn't implemented properly, renaming ++ of device files, symlinks etc. results in a empty regular file ++ instead of the proper type. ++ ++ * Directory renaming suddenly works! What a surprise! I guess ++ this is because renaming is implemented as making a copy and ++ removing the original. Still this might not work ++ everywhere... ++ ++2005-03-09 Markus Klotzbuecher ++ ++ * Bugfix, when a mini_fo directory that exists in storage ++ (state: MODIFIED, CREATED and DEL_REWRITTEN) is deleted, a ++ possibly existing WOL file contained in it needs to be ++ deleted too. ++ ++ * Starting cleanup: defined state names in order to get rid of ++ the state numbers. ++ ++2005-03-08 Markus Klotzbuecher ++ ++ * Makefile fix, fist_ioctl was built against wrong sources if ARCH=um ++ ++ * Fixed a bug in dentry.c, mini_fo_d_hash. In state 4 = ++ DEL_REWRITTEN the hash was calculated from the base dentry, ++ which was wrong and and caused assertions in ++ __mini_fo_hidden_dentry to fail. ++ ++2005-02-21 ++ ++ * Implemented directory deleting (inode.c) ++ ++ * main.c: made mini_fo_parse_options a little more robust. ++ ++2004-12-22 ++ ++ * Makefile cleanup and uml stuff, removed unneccessary files ++ ++ * Created a new and hopefully more informative README ++ ++ * CHANGELOG: created a new CHANGELOG and added old entries reversely ++ ++ ++2004-10-24 Gleb Natapov ++ ++ * Fix: owner and group where not correctly copied from base to ++ storage. ++ ++ ++2004-10-05 Gleb Natapov ++ ++ * Implementation of fsync, fasync and lock mini_fo functions. ++ ++ ++2004-09-29 Bob Lee ++ ++ * Fix of a serious pointer bug ++ ++ ++2004-09-28 Gleb Natapov ++ ++ * Implementation of mini_fo_mknod and mini_fo_rename, support ++ for device files. ++ +--- /dev/null ++++ b/fs/mini_fo/dentry.c +@@ -0,0 +1,244 @@ ++/* ++ * Copyright (c) 1997-2003 Erez Zadok ++ * Copyright (c) 2001-2003 Stony Brook University ++ * ++ * For specific licensing information, see the COPYING file distributed with ++ * this package, or get one from ftp://ftp.filesystems.org/pub/fist/COPYING. ++ * ++ * This Copyright notice must be kept intact and distributed with all ++ * fistgen sources INCLUDING sources generated by fistgen. ++ */ ++/* ++ * Copyright (C) 2004, 2005 Markus Klotzbuecher ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ */ ++ ++/* ++ * $Id$ ++ */ ++ ++#ifdef HAVE_CONFIG_H ++# include ++#endif ++ ++#include "fist.h" ++#include "mini_fo.h" ++ ++/* ++ * THIS IS A BOOLEAN FUNCTION: returns 1 if valid, 0 otherwise. ++ */ ++STATIC int ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++mini_fo_d_revalidate(dentry_t *dentry, struct nameidata *nd) ++#else ++mini_fo_d_revalidate(dentry_t *dentry, int flags) ++#endif ++{ ++ int err1 = 1; /* valid = 1, invalid = 0 */ ++ int err2 = 1; ++ dentry_t *hidden_dentry; ++ dentry_t *hidden_sto_dentry; ++ ++ ++ check_mini_fo_dentry(dentry); ++ ++ hidden_dentry = dtohd(dentry); ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ if(hidden_dentry && ++ hidden_dentry->d_op && ++ hidden_dentry->d_op->d_revalidate) { ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ err1 = hidden_dentry->d_op->d_revalidate(hidden_dentry, nd); ++#else ++ err1 = hidden_dentry->d_op->d_revalidate(hidden_dentry, flags); ++#endif ++ } ++ if(hidden_sto_dentry && ++ hidden_sto_dentry->d_op && ++ hidden_sto_dentry->d_op->d_revalidate) { ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ err2 = hidden_sto_dentry->d_op->d_revalidate(hidden_sto_dentry, ++ nd); ++#else ++ err2 = hidden_sto_dentry->d_op->d_revalidate(hidden_sto_dentry, ++ flags); ++#endif ++ } ++ ++ /* mk: if one of the lower level dentries are valid, ++ * the mini_fo dentry is too. ++ */ ++ return (err1 || err2); ++} ++ ++ ++STATIC int ++mini_fo_d_hash(dentry_t *dentry, qstr_t *name) ++{ ++ int err = 0; ++ dentry_t *hidden_dentry; ++ dentry_t *hidden_sto_dentry; ++ ++ /* hidden_dentry = mini_fo_hidden_dentry(dentry); ++ * hidden_sto_dentry = mini_fo_hidden_sto_dentry(dentry); */ ++ ++ /* state 1, 3, 4, 5: build the hash for the storage dentry */ ++ if((dtopd(dentry)->state == MODIFIED) || ++ (dtopd(dentry)->state == CREATED) || ++ (dtopd(dentry)->state == DEL_REWRITTEN) || ++ (dtopd(dentry)->state == DELETED)) { ++ hidden_sto_dentry = dtohd2(dentry); ++ if(hidden_sto_dentry && ++ hidden_sto_dentry->d_op && ++ hidden_sto_dentry->d_op->d_hash) { ++ err = hidden_sto_dentry->d_op->d_hash(hidden_sto_dentry, name); ++ } ++ goto out; ++ } ++ /* state 2: build the hash for the base dentry */ ++ if(dtopd(dentry)->state == UNMODIFIED) { ++ hidden_dentry = dtohd(dentry); ++ if(hidden_dentry && ++ hidden_dentry->d_op && ++ hidden_dentry->d_op->d_hash) { ++ err = hidden_dentry->d_op->d_hash(hidden_dentry, name); ++ } ++ goto out; ++ } ++ /* state 6: build hash for the dentry that exists */ ++ if(dtopd(dentry)->state == NON_EXISTANT) { ++ hidden_sto_dentry = dtohd2(dentry); ++ if(hidden_sto_dentry && ++ hidden_sto_dentry->d_op && ++ hidden_sto_dentry->d_op->d_hash) { ++ err = hidden_sto_dentry->d_op->d_hash(hidden_sto_dentry, name); ++ goto out; ++ } ++ hidden_dentry = dtohd(dentry); ++ if(hidden_dentry && ++ hidden_dentry->d_op && ++ hidden_dentry->d_op->d_hash) { ++ err = hidden_dentry->d_op->d_hash(hidden_dentry, name); ++ goto out; ++ } ++ } ++ ++ printk(KERN_CRIT "mini_fo: d_hash: invalid state detected.\n"); ++ ++ out: ++ return err; ++} ++ ++ ++STATIC int ++mini_fo_d_compare(dentry_t *dentry, qstr_t *a, qstr_t *b) ++{ ++ int err; ++ dentry_t *hidden_dentry=NULL; ++ ++ /* hidden_dentry = mini_fo_hidden_dentry(dentry); */ ++ if(dtohd2(dentry)) ++ hidden_dentry = dtohd2(dentry); ++ else if(dtohd(dentry)) ++ hidden_dentry = dtohd(dentry); ++ ++ if (hidden_dentry && hidden_dentry->d_op && hidden_dentry->d_op->d_compare) { ++ err = hidden_dentry->d_op->d_compare(hidden_dentry, a, b); ++ } else { ++ err = ((a->len != b->len) || memcmp(a->name, b->name, b->len)); ++ } ++ ++ return err; ++} ++ ++ ++int ++mini_fo_d_delete(dentry_t *dentry) ++{ ++ dentry_t *hidden_dentry; ++ dentry_t *hidden_sto_dentry; ++ int err = 0; ++ ++ /* this could be a negative dentry, so check first */ ++ if (!dtopd(dentry)) { ++ printk(KERN_CRIT "mini_fo_d_delete: negative dentry passed.\n"); ++ goto out; ++ } ++ hidden_dentry = dtohd(dentry); ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ if(hidden_dentry) { ++ if(hidden_dentry->d_op && ++ hidden_dentry->d_op->d_delete) { ++ err = hidden_dentry->d_op->d_delete(hidden_dentry); ++ } ++ } ++ if(hidden_sto_dentry) { ++ if(hidden_sto_dentry->d_op && ++ hidden_sto_dentry->d_op->d_delete) { ++ err = hidden_sto_dentry->d_op->d_delete(hidden_sto_dentry); ++ } ++ } ++ ++ out: ++ return err; ++} ++ ++ ++void ++mini_fo_d_release(dentry_t *dentry) ++{ ++ dentry_t *hidden_dentry; ++ dentry_t *hidden_sto_dentry; ++ ++ ++ /* this could be a negative dentry, so check first */ ++ if (!dtopd(dentry)) { ++ printk(KERN_CRIT "mini_fo_d_release: no private data.\n"); ++ goto out; ++ } ++ hidden_dentry = dtohd(dentry); ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ if(hidden_dentry) { ++ /* decrement hidden dentry's counter and free its inode */ ++ dput(hidden_dentry); ++ } ++ if(hidden_sto_dentry) { ++ /* decrement hidden dentry's counter and free its inode */ ++ dput(hidden_sto_dentry); ++ } ++ ++ /* free private data (mini_fo_dentry_info) here */ ++ kfree(dtopd(dentry)); ++ __dtopd(dentry) = NULL; /* just to be safe */ ++ out: ++ return; ++} ++ ++ ++/* ++ * we don't really need mini_fo_d_iput, because dentry_iput will call iput() if ++ * mini_fo_d_iput is not defined. We left this implemented for ease of ++ * tracing/debugging. ++ */ ++void ++mini_fo_d_iput(dentry_t *dentry, inode_t *inode) ++{ ++ iput(inode); ++} ++ ++ ++struct dentry_operations mini_fo_dops = { ++ d_revalidate: mini_fo_d_revalidate, ++ d_hash: mini_fo_d_hash, ++ d_compare: mini_fo_d_compare, ++ d_release: mini_fo_d_release, ++ d_delete: mini_fo_d_delete, ++ d_iput: mini_fo_d_iput, ++}; +--- /dev/null ++++ b/fs/mini_fo/file.c +@@ -0,0 +1,713 @@ ++/* ++ * Copyright (c) 1997-2003 Erez Zadok ++ * Copyright (c) 2001-2003 Stony Brook University ++ * ++ * For specific licensing information, see the COPYING file distributed with ++ * this package, or get one from ftp://ftp.filesystems.org/pub/fist/COPYING. ++ * ++ * This Copyright notice must be kept intact and distributed with all ++ * fistgen sources INCLUDING sources generated by fistgen. ++ */ ++/* ++ * Copyright (C) 2004, 2005 Markus Klotzbuecher ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ */ ++ ++/* ++ * $Id$ ++ */ ++ ++#ifdef HAVE_CONFIG_H ++# include ++#endif ++ ++#include "fist.h" ++#include "mini_fo.h" ++#define ROUND_UP(x) (((x)+sizeof(long)-1) & ~(sizeof(long)-1)) ++ ++/******************* ++ * File Operations * ++ *******************/ ++ ++STATIC loff_t ++mini_fo_llseek(file_t *file, loff_t offset, int origin) ++{ ++ loff_t err; ++ file_t *hidden_file = NULL; ++ ++ if(S_ISDIR(file->f_dentry->d_inode->i_mode)) { ++ /* Check if trying to llseek from a directory */ ++ err = -EISDIR; ++ goto out; ++ } ++ if (ftopd(file) != NULL) { ++ if(ftohf2(file)) { ++ hidden_file = ftohf2(file); ++ } else { ++ hidden_file = ftohf(file); ++ } ++ } ++ ++ /* always set hidden position to this one */ ++ hidden_file->f_pos = file->f_pos; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ memcpy(&(hidden_file->f_ra), ++ &(file->f_ra), ++ sizeof(struct file_ra_state)); ++#else ++ if (file->f_reada) { /* update readahead information if needed */ ++ hidden_file->f_reada = file->f_reada; ++ hidden_file->f_ramax = file->f_ramax; ++ hidden_file->f_raend = file->f_raend; ++ hidden_file->f_ralen = file->f_ralen; ++ hidden_file->f_rawin = file->f_rawin; ++ } ++#endif ++ if (hidden_file->f_op && hidden_file->f_op->llseek) ++ err = hidden_file->f_op->llseek(hidden_file, offset, origin); ++ else ++ err = generic_file_llseek(hidden_file, offset, origin); ++ ++ if (err < 0) ++ goto out; ++ ++ if (err != file->f_pos) { ++ file->f_pos = err; ++ // ION maybe this? ++ // file->f_pos = hidden_file->f_pos; ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++ file->f_reada = 0; ++#endif ++ file->f_version++; ++ } ++ ++ out: ++ return err; ++} ++ ++ ++/* mk: fanout capable */ ++STATIC ssize_t ++mini_fo_read(file_t *file, char *buf, size_t count, loff_t *ppos) ++{ ++ int err = -EINVAL; ++ file_t *hidden_file = NULL; ++ loff_t pos = *ppos; ++ ++ if(S_ISDIR(file->f_dentry->d_inode->i_mode)) { ++ /* Check if trying to read from a directory */ ++ /* printk(KERN_CRIT "mini_fo_read: ERROR: trying to read data from a directory.\n"); */ ++ err = -EISDIR; ++ goto out; ++ } ++ ++ if (ftopd(file) != NULL) { ++ if(ftohf2(file)) { ++ hidden_file = ftohf2(file); ++ } else { ++ hidden_file = ftohf(file); ++ } ++ } ++ ++ if (!hidden_file->f_op || !hidden_file->f_op->read) ++ goto out; ++ ++ err = hidden_file->f_op->read(hidden_file, buf, count, &pos); ++ *ppos = pos; ++ ++ if (err >= 0) { ++ /* atime should also be updated for reads of size zero or more */ ++ fist_copy_attr_atime(file->f_dentry->d_inode, ++ hidden_file->f_dentry->d_inode); ++ } ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++ /* ++ * MAJOR HACK ++ * because pread() does not have any way to tell us that it is ++ * our caller, then we don't know for sure if we have to update ++ * the file positions. This hack relies on read() having passed us ++ * the "real" pointer of its struct file's f_pos field. ++ */ ++ if (ppos == &file->f_pos) ++ hidden_file->f_pos = *ppos = pos; ++ if (hidden_file->f_reada) { /* update readahead information if needed */ ++ file->f_reada = hidden_file->f_reada; ++ file->f_ramax = hidden_file->f_ramax; ++ file->f_raend = hidden_file->f_raend; ++ file->f_ralen = hidden_file->f_ralen; ++ file->f_rawin = hidden_file->f_rawin; ++ } ++#else ++ memcpy(&(file->f_ra),&(hidden_file->f_ra),sizeof(struct file_ra_state)); ++#endif ++ ++ out: ++ return err; ++} ++ ++ ++/* this mini_fo_write() does not modify data pages! */ ++STATIC ssize_t ++mini_fo_write(file_t *file, const char *buf, size_t count, loff_t *ppos) ++{ ++ int err = -EINVAL; ++ file_t *hidden_file = NULL; ++ inode_t *inode; ++ inode_t *hidden_inode; ++ loff_t pos = *ppos; ++ ++ /* mk: fan out: */ ++ if (ftopd(file) != NULL) { ++ if(ftohf2(file)) { ++ hidden_file = ftohf2(file); ++ } else { ++ /* This is bad! We have no storage file to write to. This ++ * should never happen because if a file is opened for ++ * writing, a copy should have been made earlier. ++ */ ++ printk(KERN_CRIT "mini_fo: write : ERROR, no storage file to write.\n"); ++ err = -EINVAL; ++ goto out; ++ } ++ } ++ ++ inode = file->f_dentry->d_inode; ++ hidden_inode = itohi2(inode); ++ if(!hidden_inode) { ++ printk(KERN_CRIT "mini_fo: write: no sto inode found, not good.\n"); ++ goto out; ++ } ++ ++ if (!hidden_file->f_op || !hidden_file->f_op->write) ++ goto out; ++ ++ /* adjust for append -- seek to the end of the file */ ++ if (file->f_flags & O_APPEND) ++ pos = inode->i_size; ++ ++ err = hidden_file->f_op->write(hidden_file, buf, count, &pos); ++ ++ /* ++ * copy ctime and mtime from lower layer attributes ++ * atime is unchanged for both layers ++ */ ++ if (err >= 0) ++ fist_copy_attr_times(inode, hidden_inode); ++ ++ *ppos = pos; ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++ /* ++ * XXX: MAJOR HACK ++ * ++ * because pwrite() does not have any way to tell us that it is ++ * our caller, then we don't know for sure if we have to update ++ * the file positions. This hack relies on write() having passed us ++ * the "real" pointer of its struct file's f_pos field. ++ */ ++ if (ppos == &file->f_pos) ++ hidden_file->f_pos = *ppos = pos; ++#endif ++ /* update this inode's size */ ++ if (pos > inode->i_size) ++ inode->i_size = pos; ++ ++ out: ++ return err; ++} ++ ++/* Global variable to hold a file_t pointer. ++ * This serves to allow mini_fo_filldir function to know which file is ++ * beeing read, which is required for two reasons: ++ * ++ * - be able to call wol functions in order to avoid listing deleted ++ * base files. ++ * - if we're reading a directory which is in state 1, we need to ++ * maintain a list (in mini_fo_filldir) of which files allready ++ * have been copied to userspace,to detect files existing in base ++ * and storage and not list them twice. ++ */ ++filldir_t mini_fo_filldir_orig; ++file_t *mini_fo_filldir_file; ++ ++/* mainly copied from fs/readdir.c */ ++STATIC int ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18) ++mini_fo_filldir(void * __buf, const char * name, int namlen, loff_t offset, ++ u64 ino, unsigned int d_type) ++#else ++mini_fo_filldir(void * __buf, const char * name, int namlen, loff_t offset, ++ ino_t ino, unsigned int d_type) ++#endif ++{ ++ struct getdents_callback * buf = (struct getdents_callback *) __buf; ++ file_t* file = mini_fo_filldir_file; ++ ++ /* In theses states we filter meta files in storage (WOL) */ ++ if(file && (dtopd(file->f_dentry)->state == MODIFIED || ++ dtopd(file->f_dentry)->state == CREATED || ++ dtopd(file->f_dentry)->state == DEL_REWRITTEN)) { ++ ++ int tmp = strlen(META_FILENAME); ++ if(tmp == namlen) { ++ if(!strncmp(name, META_FILENAME, namlen)) ++ return 0; ++ } ++ } ++ ++ /* check if we are merging the contents of storage and base */ ++ if(file && dtopd(file->f_dentry)->state == MODIFIED) { ++ /* check if we are still reading storage contents, if ++ * yes, we just save the name of the file for duplicate ++ * checking later. */ ++ ++ if(!ftopd(file)->rd.sto_done) { ++ /* put file into ndl list */ ++ if(ndl_add_entry(&ftopd(file)->rd, name, namlen)) ++ printk(KERN_CRIT "mini_fo_filldir: Error adding to ndl.\n"); ++ } else { ++ /* check if file has been deleted */ ++ if(meta_check_d_entry(file->f_dentry, name, namlen)) ++ return 0; ++ ++ /* do duplicate checking */ ++ if(ndl_check_entry(&ftopd(file)->rd, name, namlen)) ++ return 0; ++ } ++ } ++ ++ return mini_fo_filldir_orig(buf, name, namlen, offset, ino, d_type); ++} ++ ++ ++STATIC int ++mini_fo_readdir(file_t *file, void *dirent, filldir_t filldir) ++{ ++ int err = 0;/* mk: ??? -ENOTDIR; */ ++ file_t *hidden_file = NULL; ++ file_t *hidden_sto_file = NULL; ++ inode_t *inode; ++ struct getdents_callback *buf; ++ int oldcount; ++ ++#if defined(FIST_FILTER_NAME) || defined(FIST_FILTER_SCA) ++ struct mini_fo_getdents_callback buf; ++#endif /* FIST_FILTER_NAME || FIST_FILTER_SCA */ ++ ++ buf = (struct getdents_callback *) dirent; ++ oldcount = buf->count; ++ inode = file->f_dentry->d_inode; ++ mini_fo_filldir_file = file; ++ mini_fo_filldir_orig = filldir; ++ ++ ftopd(file)->rd.sto_done = 0; ++ do { ++ if (ftopd(file) != NULL) { ++ if(ftohf2(file)) { ++ hidden_sto_file = ftohf2(file); ++ err = vfs_readdir(hidden_sto_file, mini_fo_filldir, dirent); ++ file->f_pos = hidden_sto_file->f_pos; ++ if (err > 0) ++ fist_copy_attr_atime(inode, hidden_sto_file->f_dentry->d_inode); ++ /* not finshed yet, we'll be called again */ ++ if (buf->count != oldcount) ++ break; ++ } ++ ++ ftopd(file)->rd.sto_done = 1; ++ ++ if(ftohf(file)) { ++ hidden_file = ftohf(file); ++ err = vfs_readdir(hidden_file, mini_fo_filldir, dirent); ++ file->f_pos = hidden_file->f_pos; ++ if (err > 0) ++ fist_copy_attr_atime(inode, hidden_file->f_dentry->d_inode); ++ } ++ ++ } ++ } while (0); ++ ++ /* mk: ++ * we need to check if all the directory data has been copied to userspace, ++ * or if we will be called again by userspace to complete the operation. ++ */ ++ if(buf->count == oldcount) { ++ ndl_put_list(&ftopd(file)->rd); ++ } ++ ++ /* unset this, safe */ ++ mini_fo_filldir_file = NULL; ++ return err; ++} ++ ++ ++STATIC unsigned int ++mini_fo_poll(file_t *file, poll_table *wait) ++{ ++ unsigned int mask = DEFAULT_POLLMASK; ++ file_t *hidden_file = NULL; ++ ++ if (ftopd(file) != NULL) { ++ if(ftohf2(file)) { ++ hidden_file = ftohf2(file); ++ } else { ++ hidden_file = ftohf(file); ++ } ++ } ++ ++ if (!hidden_file->f_op || !hidden_file->f_op->poll) ++ goto out; ++ ++ mask = hidden_file->f_op->poll(hidden_file, wait); ++ ++ out: ++ return mask; ++} ++ ++/* FIST-LITE special version of mmap */ ++STATIC int ++mini_fo_mmap(file_t *file, vm_area_t *vma) ++{ ++ int err = 0; ++ file_t *hidden_file = NULL; ++ ++ /* fanout capability */ ++ if (ftopd(file) != NULL) { ++ if(ftohf2(file)) { ++ hidden_file = ftohf2(file); ++ } else { ++ hidden_file = ftohf(file); ++ } ++ } ++ ++ ASSERT(hidden_file != NULL); ++ ASSERT(hidden_file->f_op != NULL); ++ ASSERT(hidden_file->f_op->mmap != NULL); ++ ++ vma->vm_file = hidden_file; ++ err = hidden_file->f_op->mmap(hidden_file, vma); ++ get_file(hidden_file); /* make sure it doesn't get freed on us */ ++ fput(file); /* no need to keep extra ref on ours */ ++ ++ return err; ++} ++ ++ ++ ++STATIC int ++mini_fo_open(inode_t *inode, file_t *file) ++{ ++ int err = 0; ++ int hidden_flags; ++ file_t *hidden_file = NULL; ++ dentry_t *hidden_dentry = NULL; ++ ++ /* fanout stuff */ ++ file_t *hidden_sto_file = NULL; ++ dentry_t *hidden_sto_dentry = NULL; ++ ++ __ftopd(file) = ++ kmalloc(sizeof(struct mini_fo_file_info), GFP_KERNEL); ++ if (!ftopd(file)) { ++ err = -ENOMEM; ++ goto out; ++ } ++ ++ /* init the readdir_helper structure */ ++ INIT_LIST_HEAD(&ftopd(file)->rd.ndl_list); ++ ftopd(file)->rd.ndl_size = 0; ++ ++ /* In certain paths this could stay uninitalized and cause trouble */ ++ ftohf(file) = NULL; ++ ftohf2(file) = NULL; ++ hidden_flags = file->f_flags; ++ ++ /* create storage files? */ ++ if(dtost(file->f_dentry) == UNMODIFIED) { ++ if(!IS_WRITE_FLAG(file->f_flags)) { ++ hidden_dentry = dtohd(file->f_dentry); ++ dget(hidden_dentry); ++ /* dentry_open will decrement mnt refcnt if err. ++ * otherwise fput() will do an mntput() for us upon file close. */ ++ mntget(stopd(inode->i_sb)->hidden_mnt); ++ hidden_file = dentry_open(hidden_dentry, ++ stopd(inode->i_sb)->hidden_mnt, ++ hidden_flags); ++ if (IS_ERR(hidden_file)) { ++ err = PTR_ERR(hidden_file); ++ dput(hidden_dentry); ++ goto out; ++ } ++ ftohf(file) = hidden_file; /* link two files */ ++ goto out; ++ } ++ else { ++ if(S_ISDIR(file->f_dentry->d_inode->i_mode)) { ++ err = dir_unmod_to_mod(file->f_dentry); ++ } else ++ err = nondir_unmod_to_mod(file->f_dentry, 1); ++ ++ if (err) { ++ printk("mini_fo_open: ERROR creating storage file.\n"); ++ goto out; ++ } ++ } ++ } ++ hidden_sto_dentry = dtohd2(file->f_dentry); ++ dget(hidden_sto_dentry); ++ ++ if(dtopd(file->f_dentry)->state == MODIFIED) { ++ /* Directorys are special, interpose on both lower level files */ ++ if(S_ISDIR(itohi(inode)->i_mode)) { ++ /* check for invalid file types of lower level files */ ++ if(!(S_ISDIR(itohi(inode)->i_mode) && S_ISDIR(itohi2(inode)->i_mode))) { ++ printk(KERN_CRIT "mini_fo_open: meta data corruption detected.\n"); ++ dput(hidden_sto_dentry); ++ err = -EINVAL; ++ goto out; ++ } ++ ++ /* lower level directorys are ok, open the base file */ ++ hidden_dentry = dtohd(file->f_dentry); ++ dget(hidden_dentry); ++ ++ mntget(stopd(inode->i_sb)->hidden_mnt); ++ hidden_file = dentry_open(hidden_dentry, ++ stopd(inode->i_sb)->hidden_mnt, ++ hidden_flags); ++ if (IS_ERR(hidden_file)) { ++ err = PTR_ERR(hidden_file); ++ dput(hidden_dentry); ++ dput(hidden_sto_dentry); ++ goto out; ++ } ++ ftohf(file) = hidden_file; /* link the two files */ ++ } ++ } ++ ++ if(!exists_in_storage(file->f_dentry)) { ++ printk(KERN_CRIT "mini_fo_open: invalid file state detected.\n"); ++ err = -EINVAL; ++ dput(hidden_sto_dentry); ++ ++ /* If the base file has been opened, we need to close it here */ ++ if(ftohf(file)) { ++ if (hidden_file->f_op && hidden_file->f_op->flush) ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18) ++ hidden_file->f_op->flush(hidden_file, NULL); ++#else ++ hidden_file->f_op->flush(hidden_file); ++#endif ++ dput(hidden_dentry); ++ } ++ goto out; ++ } ++ ++ /* ok, now we can safely open the storage file */ ++ mntget(stopd(inode->i_sb)->hidden_mnt2); ++ hidden_sto_file = dentry_open(hidden_sto_dentry, ++ stopd(inode->i_sb)->hidden_mnt2, ++ hidden_flags); ++ ++ /* dentry_open dputs the dentry if it fails */ ++ if (IS_ERR(hidden_sto_file)) { ++ err = PTR_ERR(hidden_sto_file); ++ /* close base file if open */ ++ if(ftohf(file)) { ++ if (hidden_file->f_op && hidden_file->f_op->flush) ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18) ++ hidden_file->f_op->flush(hidden_file, NULL); ++#else ++ hidden_file->f_op->flush(hidden_file); ++#endif ++ dput(hidden_dentry); ++ } ++ goto out; ++ } ++ ftohf2(file) = hidden_sto_file; /* link storage file */ ++ ++ out: ++ if (err < 0 && ftopd(file)) { ++ kfree(ftopd(file)); ++ } ++ return err; ++} ++ ++STATIC int ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18) ++mini_fo_flush(file_t *file, fl_owner_t id) ++#else ++mini_fo_flush(file_t *file) ++#endif ++{ ++ int err1 = 0; /* assume ok (see open.c:close_fp) */ ++ int err2 = 0; ++ file_t *hidden_file = NULL; ++ ++ check_mini_fo_file(file); ++ ++ /* mk: we don't do any state checking here, as its not worth the time. ++ * Just flush the lower level files if they exist. ++ */ ++ if(ftopd(file) != NULL) { ++ if(ftohf(file) != NULL) { ++ hidden_file = ftohf(file); ++ if (hidden_file->f_op && hidden_file->f_op->flush) ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18) ++ err1 = hidden_file->f_op->flush(hidden_file, id); ++#else ++ err1 = hidden_file->f_op->flush(hidden_file); ++#endif ++ } ++ if(ftohf2(file) != NULL) { ++ hidden_file = ftohf2(file); ++ if (hidden_file->f_op && hidden_file->f_op->flush) ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18) ++ err2 = hidden_file->f_op->flush(hidden_file, id); ++#else ++ err2 = hidden_file->f_op->flush(hidden_file); ++#endif ++ } ++ } ++ return (err1 | err2); ++} ++ ++ ++STATIC int ++mini_fo_release(inode_t *inode, file_t *file) ++{ ++ int err = 0; ++ file_t *hidden_file = NULL; ++ ++ if (ftopd(file) != NULL) { ++ if(ftohf(file)) { ++ hidden_file = ftohf(file); ++ fput(hidden_file); ++ } ++ if(ftohf2(file)) { ++ hidden_file = ftohf2(file); ++ fput(hidden_file); ++ } ++ kfree(ftopd(file)); ++ } ++ return err; ++} ++ ++STATIC int ++mini_fo_fsync(file_t *file, dentry_t *dentry, int datasync) ++{ ++ int err1 = 0; ++ int err2 = 0; ++ file_t *hidden_file = NULL; ++ dentry_t *hidden_dentry; ++ ++ check_mini_fo_file(file); ++ ++ if ((hidden_file = ftohf(file)) != NULL) { ++ hidden_dentry = dtohd(dentry); ++ if (hidden_file->f_op && hidden_file->f_op->fsync) { ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_dentry->d_inode->i_sem); ++#endif ++ err1 = hidden_file->f_op->fsync(hidden_file, hidden_dentry, datasync); ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_dentry->d_inode->i_sem); ++#endif ++ } ++ } ++ ++ if ((hidden_file = ftohf2(file)) != NULL) { ++ hidden_dentry = dtohd2(dentry); ++ if (hidden_file->f_op && hidden_file->f_op->fsync) { ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_dentry->d_inode->i_sem); ++#endif ++ err2 = hidden_file->f_op->fsync(hidden_file, hidden_dentry, datasync); ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_dentry->d_inode->i_sem); ++#endif ++ } ++ } ++ else ++ goto err; ++ ++err: ++ return (err1 || err2); ++} ++ ++ ++STATIC int ++mini_fo_fasync(int fd, file_t *file, int flag) ++{ ++ int err1 = 0; ++ int err2 = 0; ++ ++ file_t *hidden_file = NULL; ++ ++ check_mini_fo_file(file); ++ ++ if((hidden_file = ftohf(file)) != NULL) { ++ err1 = hidden_file->f_op->fasync(fd, hidden_file, flag); ++ } ++ if((hidden_file = ftohf2(file)) != NULL) { ++ err2 = hidden_file->f_op->fasync(fd, hidden_file, flag); ++ } ++ ++ return (err1 || err2); ++} ++ ++ ++ ++struct file_operations mini_fo_dir_fops = ++ { ++ read: generic_read_dir, ++ write: mini_fo_write, ++ readdir: mini_fo_readdir, ++ poll: mini_fo_poll, ++ /* ioctl: mini_fo_ioctl, */ ++ mmap: mini_fo_mmap, ++ open: mini_fo_open, ++ flush: mini_fo_flush, ++ release: mini_fo_release, ++ fsync: mini_fo_fsync, ++ fasync: mini_fo_fasync, ++ /* not needed lock: mini_fo_lock, */ ++ /* not needed: readv */ ++ /* not needed: writev */ ++ /* not implemented: sendpage */ ++ /* not implemented: get_unmapped_area */ ++ }; ++ ++struct file_operations mini_fo_main_fops = ++ { ++ llseek: mini_fo_llseek, ++ read: mini_fo_read, ++ write: mini_fo_write, ++ readdir: mini_fo_readdir, ++ poll: mini_fo_poll, ++ /* ioctl: mini_fo_ioctl, */ ++ mmap: mini_fo_mmap, ++ open: mini_fo_open, ++ flush: mini_fo_flush, ++ release: mini_fo_release, ++ fsync: mini_fo_fsync, ++ fasync: mini_fo_fasync, ++ /* not needed: lock: mini_fo_lock, */ ++ /* not needed: readv */ ++ /* not needed: writev */ ++ /* not implemented: sendpage */ ++ /* not implemented: get_unmapped_area */ ++ }; +--- /dev/null ++++ b/fs/mini_fo/fist.h +@@ -0,0 +1,252 @@ ++/* ++ * Copyright (c) 1997-2003 Erez Zadok ++ * Copyright (c) 2001-2003 Stony Brook University ++ * ++ * For specific licensing information, see the COPYING file distributed with ++ * this package, or get one from ftp://ftp.filesystems.org/pub/fist/COPYING. ++ * ++ * This Copyright notice must be kept intact and distributed with all ++ * fistgen sources INCLUDING sources generated by fistgen. ++ */ ++/* ++ * Copyright (C) 2004, 2005 Markus Klotzbuecher ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ */ ++ ++ ++/* ++ * $Id$ ++ */ ++ ++#ifndef __FIST_H_ ++#define __FIST_H_ ++ ++/* ++ * KERNEL ONLY CODE: ++ */ ++#ifdef __KERNEL__ ++#include ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,19) ++#include ++#else ++#include ++#endif ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++#ifdef CONFIG_MODVERSIONS ++# define MODVERSIONS ++# include ++#endif /* CONFIG_MODVERSIONS */ ++#endif /* KERNEL_VERSION < 2.6.0 */ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++#include ++#else ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#endif ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,20) ++#include ++#endif ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++#include ++#endif ++ ++#include ++ ++#include ++/* #include */ ++#include ++#include ++ ++/* ++ * MACROS: ++ */ ++ ++/* those mapped to ATTR_* were copied from linux/fs.h */ ++#define FA_MODE ATTR_MODE ++#define FA_UID ATTR_UID ++#define FA_GID ATTR_GID ++#define FA_SIZE ATTR_SIZE ++#define FA_ATIME ATTR_ATIME ++#define FA_MTIME ATTR_MTIME ++#define FA_CTIME ATTR_CTIME ++#define FA_ATIME_SET ATTR_ATIME_SET ++#define FA_MTIME_SET ATTR_MTIME_SET ++#define FA_FORCE ATTR_FORCE ++#define FA_ATTR_FLAGS ATTR_ATTR_FLAG ++ ++/* must be greater than all other ATTR_* flags! */ ++#define FA_NLINK 2048 ++#define FA_BLKSIZE 4096 ++#define FA_BLOCKS 8192 ++#define FA_TIMES (FA_ATIME|FA_MTIME|FA_CTIME) ++#define FA_ALL 0 ++ ++/* macros to manage changes between kernels */ ++#define INODE_DATA(i) (&(i)->i_data) ++ ++#define MIN(x,y) ((x < y) ? (x) : (y)) ++#define MAX(x,y) ((x > y) ? (x) : (y)) ++#define MAXPATHLEN PATH_MAX ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,4,5) ++# define lookup_one_len(a,b,c) lookup_one(a,b) ++#endif /* LINUX_VERSION_CODE < KERNEL_VERSION(2,4,5) */ ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,4,8) ++# define generic_file_llseek default_llseek ++#endif /* LINUX_VERSION_CODE < KERNEL_VERSION(2,4,8) */ ++ ++#ifndef SEEK_SET ++# define SEEK_SET 0 ++#endif /* not SEEK_SET */ ++ ++#ifndef SEEK_CUR ++# define SEEK_CUR 1 ++#endif /* not SEEK_CUR */ ++ ++#ifndef SEEK_END ++# define SEEK_END 2 ++#endif /* not SEEK_END */ ++ ++#ifndef DEFAULT_POLLMASK ++# define DEFAULT_POLLMASK (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM) ++#endif /* not DEFAULT_POLLMASK */ ++ ++/* XXX: fix this so fistgen generates kfree() code directly */ ++#define kfree_s(a,b) kfree(a) ++ ++/* ++ * TYPEDEFS: ++ */ ++typedef struct dentry dentry_t; ++typedef struct file file_t; ++typedef struct inode inode_t; ++typedef inode_t vnode_t; ++typedef struct page page_t; ++typedef struct qstr qstr_t; ++typedef struct super_block super_block_t; ++typedef super_block_t vfs_t; ++typedef struct vm_area_struct vm_area_t; ++ ++ ++/* ++ * EXTERNALS: ++ */ ++ ++#define FPPF(str,page) printk("PPF %s 0x%x/%d: Lck:%d Err:%d Ref:%d Upd:%d Other::%d:%d:%d:%d:\n", \ ++ str, \ ++ (int) page, \ ++ (int) page->index, \ ++ (PageLocked(page) ? 1 : 0), \ ++ (PageError(page) ? 1 : 0), \ ++ (PageReferenced(page) ? 1 : 0), \ ++ (Page_Uptodate(page) ? 1 : 0), \ ++ (PageDecrAfter(page) ? 1 : 0), \ ++ (PageSlab(page) ? 1 : 0), \ ++ (PageSwapCache(page) ? 1 : 0), \ ++ (PageReserved(page) ? 1 : 0) \ ++ ) ++#define EZKDBG printk("EZK %s:%d:%s\n",__FILE__,__LINE__,__FUNCTION__) ++#if 0 ++# define EZKDBG1 printk("EZK %s:%d\n",__FILE__,__LINE__) ++#else ++# define EZKDBG1 ++#endif ++ ++extern int fist_get_debug_value(void); ++extern int fist_set_debug_value(int val); ++#if 0 /* mini_fo doesn't need these */ ++extern void fist_dprint_internal(int level, char *str,...); ++extern void fist_print_dentry(char *str, const dentry_t *dentry); ++extern void fist_print_inode(char *str, const inode_t *inode); ++extern void fist_print_file(char *str, const file_t *file); ++extern void fist_print_buffer_flags(char *str, struct buffer_head *buffer); ++extern void fist_print_page_flags(char *str, page_t *page); ++extern void fist_print_page_bytes(char *str, page_t *page); ++extern void fist_print_pte_flags(char *str, const page_t *page); ++extern void fist_checkinode(inode_t *inode, char *msg); ++extern void fist_print_sb(char *str, const super_block_t *sb); ++ ++/* §$% by mk: special debug functions */ ++extern void fist_mk_print_dentry(char *str, const dentry_t *dentry); ++extern void fist_mk_print_inode(char *str, const inode_t *inode); ++ ++extern char *add_indent(void); ++extern char *del_indent(void); ++#endif/* mini_fo doesn't need these */ ++ ++ ++#define STATIC ++#define ASSERT(EX) \ ++do { \ ++ if (!(EX)) { \ ++ printk(KERN_CRIT "ASSERTION FAILED: %s at %s:%d (%s)\n", #EX, \ ++ __FILE__, __LINE__, __FUNCTION__); \ ++ (*((char *)0))=0; \ ++ } \ ++} while (0) ++/* same ASSERT, but tell me who was the caller of the function */ ++#define ASSERT2(EX) \ ++do { \ ++ if (!(EX)) { \ ++ printk(KERN_CRIT "ASSERTION FAILED (caller): %s at %s:%d (%s)\n", #EX, \ ++ file, line, func); \ ++ (*((char *)0))=0; \ ++ } \ ++} while (0) ++ ++#if 0 /* mini_fo doesn't need these */ ++#define dprintk(format, args...) printk(KERN_DEBUG format, ##args) ++#define fist_dprint(level, str, args...) fist_dprint_internal(level, KERN_DEBUG str, ## args) ++#define print_entry_location() fist_dprint(4, "%sIN: %s %s:%d\n", add_indent(), __FUNCTION__, __FILE__, __LINE__) ++#define print_exit_location() fist_dprint(4, "%s OUT: %s %s:%d\n", del_indent(), __FUNCTION__, __FILE__, __LINE__) ++#define print_exit_status(status) fist_dprint(4, "%s OUT: %s %s:%d, STATUS: %d\n", del_indent(), __FUNCTION__, __FILE__, __LINE__, status) ++#define print_exit_pointer(status) \ ++do { \ ++ if (IS_ERR(status)) \ ++ fist_dprint(4, "%s OUT: %s %s:%d, RESULT: %ld\n", del_indent(), __FUNCTION__, __FILE__, __LINE__, PTR_ERR(status)); \ ++ else \ ++ fist_dprint(4, "%s OUT: %s %s:%d, RESULT: 0x%x\n", del_indent(), __FUNCTION__, __FILE__, __LINE__, PTR_ERR(status)); \ ++} while (0) ++#endif/* mini_fo doesn't need these */ ++ ++#endif /* __KERNEL__ */ ++ ++ ++/* ++ * DEFINITIONS FOR USER AND KERNEL CODE: ++ * (Note: ioctl numbers 1--9 are reserved for fistgen, the rest ++ * are auto-generated automatically based on the user's .fist file.) ++ */ ++# define FIST_IOCTL_GET_DEBUG_VALUE _IOR(0x15, 1, int) ++# define FIST_IOCTL_SET_DEBUG_VALUE _IOW(0x15, 2, int) ++ ++#endif /* not __FIST_H_ */ +--- /dev/null ++++ b/fs/mini_fo/inode.c +@@ -0,0 +1,1564 @@ ++/* ++ * Copyright (c) 1997-2003 Erez Zadok ++ * Copyright (c) 2001-2003 Stony Brook University ++ * ++ * For specific licensing information, see the COPYING file distributed with ++ * this package, or get one from ftp://ftp.filesystems.org/pub/fist/COPYING. ++ * ++ * This Copyright notice must be kept intact and distributed with all ++ * fistgen sources INCLUDING sources generated by fistgen. ++ */ ++/* ++ * Copyright (C) 2004, 2005 Markus Klotzbuecher ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ */ ++ ++/* ++ * $Id$ ++ */ ++ ++#ifdef HAVE_CONFIG_H ++# include ++#endif ++ ++#include "fist.h" ++#include "mini_fo.h" ++ ++STATIC int ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++mini_fo_create(inode_t *dir, dentry_t *dentry, int mode, struct nameidata *nd) ++#else ++mini_fo_create(inode_t *dir, dentry_t *dentry, int mode) ++#endif ++{ ++ int err = 0; ++ ++ check_mini_fo_dentry(dentry); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ err = create_sto_reg_file(dentry, mode, nd); ++#else ++ err = create_sto_reg_file(dentry, mode); ++#endif ++ check_mini_fo_dentry(dentry); ++ return err; ++} ++ ++ ++STATIC dentry_t * ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++mini_fo_lookup(inode_t *dir, dentry_t *dentry, struct nameidata* nd) ++#else ++mini_fo_lookup(inode_t *dir, dentry_t *dentry) ++#endif ++{ ++ int err = 0; ++ dentry_t *hidden_dir_dentry; ++ dentry_t *hidden_dentry = NULL; ++ ++ dentry_t *hidden_sto_dir_dentry; ++ dentry_t *hidden_sto_dentry = NULL; ++ ++ /* whiteout flag */ ++ int del_flag = 0; ++ char *bpath = NULL; ++ ++ const char *name; ++ unsigned int namelen; ++ ++ /* Don't allow lookups of META-files */ ++ namelen = strlen(META_FILENAME); ++ if(namelen == dentry->d_name.len) { ++ if(!strncmp(dentry->d_name.name, META_FILENAME, namelen)) { ++ err = -ENOENT; ++ goto out; ++ } ++ } ++ ++ hidden_dir_dentry = dtohd(dentry->d_parent); ++ hidden_sto_dir_dentry = dtohd2(dentry->d_parent); ++ ++ name = dentry->d_name.name; ++ namelen = dentry->d_name.len; ++ ++ /* must initialize dentry operations */ ++ dentry->d_op = &mini_fo_dops; ++ ++ /* setup the del_flag */ ++ del_flag = __meta_check_d_entry(dir, name, namelen); ++ bpath = __meta_check_r_entry(dir, name, namelen); ++ ++ /* perform the lookups of base and storage files: ++ * ++ * This caused some serious trouble, as a lookup_one_len passing ++ * a negative dentry oopses. Solution is to only do the lookup ++ * if the dentry is positive, else we set it to NULL ++ * More trouble, who said a *_dir_dentry can't be NULL? ++ */ ++ if(bpath) { ++ /* Cross-Interposing (C), yeah! */ ++ hidden_dentry = bpath_walk(dir->i_sb, bpath); ++ if(!hidden_dentry || !hidden_dentry->d_inode) { ++ printk(KERN_CRIT "mini_fo_lookup: bpath_walk failed.\n"); ++ err= -EINVAL; ++ goto out; ++ } ++ ++ /* this can be set up safely without fear of spaghetti ++ * interposing as it is only used for copying times */ ++ hidden_dir_dentry = hidden_dentry->d_parent; ++ kfree(bpath); ++ } ++ else if(hidden_dir_dentry && hidden_dir_dentry->d_inode) ++ hidden_dentry = ++ lookup_one_len(name, hidden_dir_dentry, namelen); ++ else ++ hidden_dentry = NULL; ++ ++ if(hidden_sto_dir_dentry && hidden_sto_dir_dentry->d_inode) ++ hidden_sto_dentry = ++ lookup_one_len(name, hidden_sto_dir_dentry, namelen); ++ else ++ hidden_sto_dentry = NULL; ++ ++ /* catch error in lookup */ ++ if (IS_ERR(hidden_dentry) || IS_ERR(hidden_sto_dentry)) { ++ /* mk: we need to call dput on the dentry, whose ++ * lookup_one_len operation failed, in order to avoid ++ * unmount trouble. ++ */ ++ if(IS_ERR(hidden_dentry)) { ++ printk(KERN_CRIT "mini_fo_lookup: ERR from base dentry, lookup failed.\n"); ++ err = PTR_ERR(hidden_dentry); ++ } else { ++ dput(hidden_dentry); ++ } ++ if(IS_ERR(hidden_sto_dentry)) { ++ printk(KERN_CRIT "mini_fo_lookup: ERR from storage dentry, lookup failed.\n"); ++ err = PTR_ERR(hidden_sto_dentry); ++ } else { ++ dput(hidden_sto_dentry); ++ } ++ goto out; ++ } ++ ++ /* allocate dentry private data */ ++ __dtopd(dentry) = (struct mini_fo_dentry_info *) ++ kmalloc(sizeof(struct mini_fo_dentry_info), GFP_KERNEL); ++ ++ if (!dtopd(dentry)) { ++ err = -ENOMEM; ++ goto out_dput; ++ } ++ ++ /* check for different states of the mini_fo file to be looked up. */ ++ ++ /* state 1, file has been modified */ ++ if(hidden_dentry && hidden_sto_dentry && ++ hidden_dentry->d_inode && hidden_sto_dentry->d_inode && !del_flag) { ++ ++ /* update parent directory's atime */ ++ fist_copy_attr_atime(dir, hidden_sto_dir_dentry->d_inode); ++ ++ dtopd(dentry)->state = MODIFIED; ++ dtohd(dentry) = hidden_dentry; ++ dtohd2(dentry) = hidden_sto_dentry; ++ ++ err = mini_fo_tri_interpose(hidden_dentry, ++ hidden_sto_dentry, ++ dentry, dir->i_sb, 1); ++ if (err) { ++ printk(KERN_CRIT "mini_fo_lookup: error interposing (state1).\n"); ++ goto out_free; ++ } ++ goto out; ++ } ++ /* state 2, file is unmodified */ ++ if(hidden_dentry && hidden_dentry->d_inode && !del_flag) { ++ ++ fist_copy_attr_atime(dir, hidden_dir_dentry->d_inode); ++ ++ dtopd(dentry)->state = UNMODIFIED; ++ dtohd(dentry) = hidden_dentry; ++ dtohd2(dentry) = hidden_sto_dentry; /* could be negative */ ++ ++ err = mini_fo_tri_interpose(hidden_dentry, ++ hidden_sto_dentry, ++ dentry, dir->i_sb, 1); ++ if (err) { ++ printk(KERN_CRIT "mini_fo_lookup: error interposing (state2).\n"); ++ goto out_free; ++ } ++ goto out; ++ } ++ /* state 3, file has been newly created */ ++ if(hidden_sto_dentry && hidden_sto_dentry->d_inode && !del_flag) { ++ ++ fist_copy_attr_atime(dir, hidden_sto_dir_dentry->d_inode); ++ dtopd(dentry)->state = CREATED; ++ dtohd(dentry) = hidden_dentry; /* could be negative */ ++ dtohd2(dentry) = hidden_sto_dentry; ++ ++ err = mini_fo_tri_interpose(hidden_dentry, ++ hidden_sto_dentry, ++ dentry, dir->i_sb, 1); ++ if (err) { ++ printk(KERN_CRIT "mini_fo_lookup: error interposing (state3).\n"); ++ goto out_free; ++ } ++ goto out; ++ } ++ ++ /* state 4, file has deleted and created again. */ ++ if(hidden_dentry && hidden_sto_dentry && ++ hidden_dentry->d_inode && ++ hidden_sto_dentry->d_inode && del_flag) { ++ ++ fist_copy_attr_atime(dir, hidden_sto_dir_dentry->d_inode); ++ dtopd(dentry)->state = DEL_REWRITTEN; ++ dtohd(dentry) = NULL; ++ dtohd2(dentry) = hidden_sto_dentry; ++ ++ err = mini_fo_tri_interpose(NULL, ++ hidden_sto_dentry, ++ dentry, dir->i_sb, 1); ++ if (err) { ++ printk(KERN_CRIT "mini_fo_lookup: error interposing (state4).\n"); ++ goto out_free; ++ } ++ /* We will never need this dentry again, as the file has been ++ * deleted from base */ ++ dput(hidden_dentry); ++ goto out; ++ } ++ /* state 5, file has been deleted in base */ ++ if(hidden_dentry && hidden_sto_dentry && ++ hidden_dentry->d_inode && ++ !hidden_sto_dentry->d_inode && del_flag) { ++ ++ /* check which parents atime we need for updating */ ++ if(hidden_sto_dir_dentry->d_inode) ++ fist_copy_attr_atime(dir, ++ hidden_sto_dir_dentry->d_inode); ++ else ++ fist_copy_attr_atime(dir, ++ hidden_dir_dentry->d_inode); ++ ++ dtopd(dentry)->state = DELETED; ++ dtohd(dentry) = NULL; ++ dtohd2(dentry) = hidden_sto_dentry; ++ ++ /* add negative dentry to dcache to speed up lookups */ ++ d_add(dentry, NULL); ++ dput(hidden_dentry); ++ goto out; ++ } ++ /* state 6, file does not exist */ ++ if(((hidden_dentry && !hidden_dentry->d_inode) || ++ (hidden_sto_dentry && !hidden_sto_dentry->d_inode)) && !del_flag) ++ { ++ /* check which parents atime we need for updating */ ++ if(hidden_sto_dir_dentry && hidden_sto_dir_dentry->d_inode) ++ fist_copy_attr_atime(dir, hidden_sto_dir_dentry->d_inode); ++ else ++ fist_copy_attr_atime(dir, hidden_dir_dentry->d_inode); ++ ++ dtopd(dentry)->state = NON_EXISTANT; ++ dtohd(dentry) = hidden_dentry; ++ dtohd2(dentry) = hidden_sto_dentry; ++ d_add(dentry, NULL); ++ goto out; ++ } ++ ++ /* if we get to here, were in an invalid state. bad. */ ++ printk(KERN_CRIT "mini_fo_lookup: ERROR, meta data corruption detected.\n"); ++ ++ /* end state checking */ ++ out_free: ++ d_drop(dentry); /* so that our bad dentry will get destroyed */ ++ kfree(dtopd(dentry)); ++ __dtopd(dentry) = NULL; /* be safe */ ++ ++ out_dput: ++ if(hidden_dentry) ++ dput(hidden_dentry); ++ if(hidden_sto_dentry) ++ dput(hidden_sto_dentry); /* drops usage count and marks for release */ ++ ++ out: ++ /* initalize wol if file exists and is directory */ ++ if(dentry->d_inode) { ++ if(S_ISDIR(dentry->d_inode->i_mode)) { ++ itopd(dentry->d_inode)->deleted_list_size = -1; ++ itopd(dentry->d_inode)->renamed_list_size = -1; ++ meta_build_lists(dentry); ++ } ++ } ++ return ERR_PTR(err); ++} ++ ++ ++STATIC int ++mini_fo_link(dentry_t *old_dentry, inode_t *dir, dentry_t *new_dentry) ++{ ++ int err; ++ dentry_t *hidden_old_dentry; ++ dentry_t *hidden_new_dentry; ++ dentry_t *hidden_dir_dentry; ++ ++ ++ check_mini_fo_dentry(old_dentry); ++ check_mini_fo_dentry(new_dentry); ++ check_mini_fo_inode(dir); ++ ++ /* no links to directorys and existing targets target allowed */ ++ if(S_ISDIR(old_dentry->d_inode->i_mode) || ++ is_mini_fo_existant(new_dentry)) { ++ err = -EPERM; ++ goto out; ++ } ++ ++ /* bring it directly from unmod to del_rew */ ++ if(dtost(old_dentry) == UNMODIFIED) { ++ err = nondir_unmod_to_mod(old_dentry, 1); ++ if(err) { ++ err = -EINVAL; ++ goto out; ++ } ++ err = meta_add_d_entry(old_dentry->d_parent, ++ old_dentry->d_name.name, ++ old_dentry->d_name.len); ++ if(err) { ++ err = -EINVAL; ++ goto out; ++ } ++ dput(dtohd(old_dentry)); ++ dtohd(old_dentry) = NULL; ++ dtost(old_dentry) = DEL_REWRITTEN; ++ } ++ ++ err = get_neg_sto_dentry(new_dentry); ++ if(err) { ++ err = -EINVAL; ++ goto out; ++ } ++ ++ hidden_old_dentry = dtohd2(old_dentry); ++ hidden_new_dentry = dtohd2(new_dentry); ++ ++ dget(hidden_old_dentry); ++ dget(hidden_new_dentry); ++ ++ /* was: hidden_dir_dentry = lock_parent(hidden_new_dentry); */ ++ hidden_dir_dentry = dget(hidden_new_dentry->d_parent); ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_dir_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_dir_dentry->d_inode->i_sem); ++#endif ++ ++ err = vfs_link(hidden_old_dentry, ++ hidden_dir_dentry->d_inode, ++ hidden_new_dentry); ++ if (err || !hidden_new_dentry->d_inode) ++ goto out_lock; ++ ++ dtost(new_dentry) = CREATED; ++ err = mini_fo_tri_interpose(NULL, hidden_new_dentry, new_dentry, dir->i_sb, 0); ++ if (err) ++ goto out_lock; ++ ++ fist_copy_attr_timesizes(dir, hidden_new_dentry->d_inode); ++ /* propagate number of hard-links */ ++ old_dentry->d_inode->i_nlink = itohi2(old_dentry->d_inode)->i_nlink; ++ ++ out_lock: ++ /* was: unlock_dir(hidden_dir_dentry); */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_dir_dentry->d_inode->i_sem); ++#endif ++ dput(hidden_dir_dentry); ++ ++ dput(hidden_new_dentry); ++ dput(hidden_old_dentry); ++ if (!new_dentry->d_inode) ++ d_drop(new_dentry); ++ ++ out: ++ return err; ++} ++ ++ ++STATIC int ++mini_fo_unlink(inode_t *dir, dentry_t *dentry) ++{ ++ int err = 0; ++ ++ dget(dentry); ++ if(dtopd(dentry)->state == MODIFIED) { ++ err = nondir_mod_to_del(dentry); ++ goto out; ++ } ++ else if(dtopd(dentry)->state == UNMODIFIED) { ++ err = nondir_unmod_to_del(dentry); ++ goto out; ++ } ++ else if(dtopd(dentry)->state == CREATED) { ++ err = nondir_creat_to_del(dentry); ++ goto out; ++ } ++ else if(dtopd(dentry)->state == DEL_REWRITTEN) { ++ err = nondir_del_rew_to_del(dentry); ++ goto out; ++ } ++ ++ printk(KERN_CRIT "mini_fo_unlink: ERROR, invalid state detected.\n"); ++ ++ out: ++ fist_copy_attr_times(dir, itohi2(dentry->d_parent->d_inode)); ++ ++ if(!err) { ++ /* is this causing my pain? d_delete(dentry); */ ++ d_drop(dentry); ++ } ++ ++ dput(dentry); ++ return err; ++} ++ ++ ++STATIC int ++mini_fo_symlink(inode_t *dir, dentry_t *dentry, const char *symname) ++{ ++ int err=0; ++ dentry_t *hidden_sto_dentry; ++ dentry_t *hidden_sto_dir_dentry; ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ umode_t mode; ++#endif ++ ++ /* Fail if the symlink file exists */ ++ if(!(dtost(dentry) == DELETED || ++ dtost(dentry) == NON_EXISTANT)) { ++ err = -EEXIST; ++ goto out; ++ } ++ ++ err = get_neg_sto_dentry(dentry); ++ if(err) { ++ err = -EINVAL; ++ goto out; ++ } ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ dget(hidden_sto_dentry); ++ /* was: hidden_sto_dir_dentry = lock_parent(hidden_sto_dentry); */ ++ hidden_sto_dir_dentry = dget(hidden_sto_dentry->d_parent); ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ mode = S_IALLUGO; ++ err = vfs_symlink(hidden_sto_dir_dentry->d_inode, ++ hidden_sto_dentry, symname, mode); ++#else ++ err = vfs_symlink(hidden_sto_dir_dentry->d_inode, ++ hidden_sto_dentry, ++ symname); ++#endif ++ if (err || !hidden_sto_dentry->d_inode) ++ goto out_lock; ++ ++ if(dtost(dentry) == DELETED) { ++ dtost(dentry) = DEL_REWRITTEN; ++ err = mini_fo_tri_interpose(NULL, hidden_sto_dentry, dentry, dir->i_sb, 0); ++ if(err) ++ goto out_lock; ++ } else if(dtost(dentry) == NON_EXISTANT) { ++ dtost(dentry) = CREATED; ++ err = mini_fo_tri_interpose(dtohd(dentry), hidden_sto_dentry, dentry, dir->i_sb, 0); ++ if(err) ++ goto out_lock; ++ } ++ fist_copy_attr_timesizes(dir, hidden_sto_dir_dentry->d_inode); ++ ++ out_lock: ++ /* was: unlock_dir(hidden_sto_dir_dentry); */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ dput(hidden_sto_dir_dentry); ++ ++ dput(hidden_sto_dentry); ++ if (!dentry->d_inode) ++ d_drop(dentry); ++ out: ++ return err; ++} ++ ++STATIC int ++mini_fo_mkdir(inode_t *dir, dentry_t *dentry, int mode) ++{ ++ int err; ++ ++ err = create_sto_dir(dentry, mode); ++ ++ check_mini_fo_dentry(dentry); ++ ++ return err; ++} ++ ++ ++STATIC int ++mini_fo_rmdir(inode_t *dir, dentry_t *dentry) ++{ ++ int err = 0; ++ ++ dentry_t *hidden_sto_dentry; ++ dentry_t *hidden_sto_dir_dentry; ++ dentry_t *meta_dentry; ++ inode_t *hidden_sto_dir = NULL; ++ ++ check_mini_fo_dentry(dentry); ++ check_mini_fo_inode(dir); ++ ++ dget(dentry); ++ if(dtopd(dentry)->state == MODIFIED) { ++ /* XXX: disabled, because it does not bother to check files on ++ * the original filesystem - just a hack, but better than simply ++ * removing it without testing */ ++ err = -EINVAL; ++ goto out; ++ ++ hidden_sto_dir = itohi2(dir); ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ /* was:hidden_sto_dir_dentry = lock_parent(hidden_sto_dentry); */ ++ hidden_sto_dir_dentry = dget(hidden_sto_dentry->d_parent); ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ ++ /* Delete an old WOL file contained in the storage dir */ ++ meta_dentry = lookup_one_len(META_FILENAME, ++ hidden_sto_dentry, ++ strlen(META_FILENAME)); ++ if(meta_dentry->d_inode) { ++ err = vfs_unlink(hidden_sto_dentry->d_inode, meta_dentry); ++ dput(meta_dentry); ++ if(!err) ++ d_delete(meta_dentry); ++ } ++ ++ err = vfs_rmdir(hidden_sto_dir, hidden_sto_dentry); ++ dput(hidden_sto_dentry); ++ if(!err) ++ d_delete(hidden_sto_dentry); ++ ++ /* propagate number of hard-links */ ++ dentry->d_inode->i_nlink = itohi2(dentry->d_inode)->i_nlink; ++ ++ dput(dtohd(dentry)); ++ ++ dtohd(dentry) = NULL; ++ dtopd(dentry)->state = DELETED; ++ ++ /* carefull with R files */ ++ if( __meta_is_r_entry(dir, ++ dentry->d_name.name, ++ dentry->d_name.len) == 1) { ++ err = meta_remove_r_entry(dentry->d_parent, ++ dentry->d_name.name, ++ dentry->d_name.len); ++ if(err) { ++ printk(KERN_CRIT "mini_fo: rmdir: meta_remove_r_entry failed.\n"); ++ goto out; ++ } ++ } ++ else { ++ /* ok, add deleted file to META */ ++ meta_add_d_entry(dentry->d_parent, ++ dentry->d_name.name, ++ dentry->d_name.len); ++ } ++ /* was: unlock_dir(hidden_sto_dir_dentry); */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ dput(hidden_sto_dir_dentry); ++ goto out; ++ } ++ else if(dtopd(dentry)->state == UNMODIFIED) { ++ /* XXX: simply adding it to the delete list here is fscking dangerous! ++ * as a temporary hack, i will disable rmdir on unmodified directories ++ * for now. ++ */ ++ err = -EINVAL; ++ goto out; ++ ++ err = get_neg_sto_dentry(dentry); ++ if(err) { ++ err = -EINVAL; ++ goto out; ++ } ++ ++ /* dput base dentry, this will relase the inode and free the ++ * dentry, as we will never need it again. */ ++ dput(dtohd(dentry)); ++ dtohd(dentry) = NULL; ++ dtopd(dentry)->state = DELETED; ++ ++ /* add deleted file to META-file */ ++ meta_add_d_entry(dentry->d_parent, ++ dentry->d_name.name, ++ dentry->d_name.len); ++ goto out; ++ } ++ else if(dtopd(dentry)->state == CREATED) { ++ hidden_sto_dir = itohi2(dir); ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ /* was: hidden_sto_dir_dentry = lock_parent(hidden_sto_dentry);*/ ++ hidden_sto_dir_dentry = dget(hidden_sto_dentry->d_parent); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ ++ /* Delete an old WOL file contained in the storage dir */ ++ meta_dentry = lookup_one_len(META_FILENAME, ++ hidden_sto_dentry, ++ strlen(META_FILENAME)); ++ if(meta_dentry->d_inode) { ++ /* is this necessary? dget(meta_dentry); */ ++ err = vfs_unlink(hidden_sto_dentry->d_inode, ++ meta_dentry); ++ dput(meta_dentry); ++ if(!err) ++ d_delete(meta_dentry); ++ } ++ ++ err = vfs_rmdir(hidden_sto_dir, hidden_sto_dentry); ++ dput(hidden_sto_dentry); ++ if(!err) ++ d_delete(hidden_sto_dentry); ++ ++ /* propagate number of hard-links */ ++ dentry->d_inode->i_nlink = itohi2(dentry->d_inode)->i_nlink; ++ dtopd(dentry)->state = NON_EXISTANT; ++ ++ /* was: unlock_dir(hidden_sto_dir_dentry); */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ dput(hidden_sto_dir_dentry); ++ ++ goto out; ++ } ++ else if(dtopd(dentry)->state == DEL_REWRITTEN) { ++ hidden_sto_dir = itohi2(dir); ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ /* was: hidden_sto_dir_dentry = lock_parent(hidden_sto_dentry);*/ ++ hidden_sto_dir_dentry = dget(hidden_sto_dentry->d_parent); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ ++ /* Delete an old WOL file contained in the storage dir */ ++ meta_dentry = lookup_one_len(META_FILENAME, ++ hidden_sto_dentry, ++ strlen(META_FILENAME)); ++ if(meta_dentry->d_inode) { ++ /* is this necessary? dget(meta_dentry); */ ++ err = vfs_unlink(hidden_sto_dentry->d_inode, ++ meta_dentry); ++ dput(meta_dentry); ++ if(!err) ++ d_delete(meta_dentry); ++ } ++ ++ err = vfs_rmdir(hidden_sto_dir, hidden_sto_dentry); ++ dput(hidden_sto_dentry); ++ if(!err) ++ d_delete(hidden_sto_dentry); ++ ++ /* propagate number of hard-links */ ++ dentry->d_inode->i_nlink = itohi2(dentry->d_inode)->i_nlink; ++ dtopd(dentry)->state = DELETED; ++ /* was: unlock_dir(hidden_sto_dir_dentry); */ ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ dput(hidden_sto_dir_dentry); ++ goto out; ++ } ++ ++ printk(KERN_CRIT "mini_fo_rmdir: ERROR, invalid state detected.\n"); ++ ++ out: ++ if(!err) { ++ d_drop(dentry); ++ } ++ ++ fist_copy_attr_times(dir, itohi2(dentry->d_parent->d_inode)); ++ dput(dentry); ++ ++ return err; ++} ++ ++ ++STATIC int ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++mini_fo_mknod(inode_t *dir, dentry_t *dentry, int mode, dev_t dev) ++#else ++mini_fo_mknod(inode_t *dir, dentry_t *dentry, int mode, int dev) ++#endif ++{ ++ int err = 0; ++ ++ check_mini_fo_dentry(dentry); ++ ++ err = create_sto_nod(dentry, mode, dev); ++ if(err) { ++ printk(KERN_CRIT "mini_fo_mknod: creating sto nod failed.\n"); ++ err = -EINVAL; ++ } ++ ++ check_mini_fo_dentry(dentry); ++ return err; ++} ++ ++ ++STATIC int ++mini_fo_rename(inode_t *old_dir, dentry_t *old_dentry, ++ inode_t *new_dir, dentry_t *new_dentry) ++{ ++ /* dispatch */ ++ if(S_ISDIR(old_dentry->d_inode->i_mode)) ++ return rename_directory(old_dir, old_dentry, new_dir, new_dentry); ++ return rename_nondir(old_dir, old_dentry, new_dir, new_dentry); ++ ++} ++ ++int rename_directory(inode_t *old_dir, dentry_t *old_dentry, ++ inode_t *new_dir, dentry_t *new_dentry) ++{ ++ int err, bpath_len; ++ char *bpath; ++ ++ dentry_t *hidden_old_dentry; ++ dentry_t *hidden_new_dentry; ++ dentry_t *hidden_old_dir_dentry; ++ dentry_t *hidden_new_dir_dentry; ++ ++ err = 0; ++ bpath = NULL; ++ bpath_len = 0; ++ ++ /* this is a test, chuck out if it works */ ++ if(!(dtopd(new_dentry)->state == DELETED || ++ dtopd(new_dentry)->state == NON_EXISTANT)) { ++ printk(KERN_CRIT "mini_fo: rename_directory: \ ++ uh, ah, new_dentry not negative.\n"); ++ /* return -1; */ ++ } ++ ++ /* state = UNMODIFIED */ ++ if(dtopd(old_dentry)->state == UNMODIFIED) { ++ err = dir_unmod_to_mod(old_dentry); ++ if (err) ++ goto out; ++ } ++ ++ /* state = MODIFIED */ ++ if(dtopd(old_dentry)->state == MODIFIED) { ++ bpath = meta_check_r_entry(old_dentry->d_parent, ++ old_dentry->d_name.name, ++ old_dentry->d_name.len); ++ if(bpath) { ++ err = meta_remove_r_entry(old_dentry->d_parent, ++ old_dentry->d_name.name, ++ old_dentry->d_name.len); ++ if(err) { ++ printk(KERN_CRIT "mini_fo: rename_directory:\ ++ meta_remove_r_entry \ ++ failed.\n"); ++ goto out; ++ } ++ err = meta_add_r_entry(new_dentry->d_parent, ++ bpath, ++ strlen(bpath), ++ new_dentry->d_name.name, ++ new_dentry->d_name.len); ++ kfree(bpath); ++ } ++ else {/* wol it */ ++ err = meta_add_d_entry(old_dentry->d_parent, ++ old_dentry->d_name.name, ++ old_dentry->d_name.len); ++ if (err) ++ goto out; ++ /* put it on rename list */ ++ err = get_mini_fo_bpath(old_dentry, ++ &bpath, ++ &bpath_len); ++ if (err) ++ goto out; ++ err = meta_add_r_entry(new_dentry->d_parent, ++ bpath, bpath_len, ++ new_dentry->d_name.name, ++ new_dentry->d_name.len); ++ if (err) ++ goto out; ++ } ++ /* no state change, MODIFIED stays MODIFIED */ ++ } ++ /* state = CREATED */ ++ if(dtopd(old_dentry)->state == CREATED || ++ dtopd(old_dentry)->state == DEL_REWRITTEN) { ++ if(dtohd(old_dentry)) ++ dput(dtohd(old_dentry)); ++ ++ if(dtopd(new_dentry)->state == DELETED) { ++ dtopd(old_dentry)->state = DEL_REWRITTEN; ++ dtohd(old_dentry) = NULL; ++ } ++ else if(dtopd(new_dentry)->state == NON_EXISTANT) { ++ dtopd(old_dentry)->state = CREATED; ++ /* steal new dentry's neg. base dentry */ ++ dtohd(old_dentry) = dtohd(new_dentry); ++ dtohd(new_dentry) = NULL; ++ } ++ } ++ if(dtopd(new_dentry)->state == UNMODIFIED || ++ dtopd(new_dentry)->state == NON_EXISTANT) { ++ err = get_neg_sto_dentry(new_dentry); ++ if(err) ++ goto out; ++ } ++ ++ /* now move sto file */ ++ hidden_old_dentry = dtohd2(old_dentry); ++ hidden_new_dentry = dtohd2(new_dentry); ++ ++ dget(hidden_old_dentry); ++ dget(hidden_new_dentry); ++ ++ hidden_old_dir_dentry = dget(hidden_old_dentry->d_parent); ++ hidden_new_dir_dentry = dget(hidden_new_dentry->d_parent); ++ double_lock(hidden_old_dir_dentry, hidden_new_dir_dentry); ++ ++ err = vfs_rename(hidden_old_dir_dentry->d_inode, hidden_old_dentry, ++ hidden_new_dir_dentry->d_inode, hidden_new_dentry); ++ if(err) ++ goto out_lock; ++ ++ fist_copy_attr_all(new_dir, hidden_new_dir_dentry->d_inode); ++ if (new_dir != old_dir) ++ fist_copy_attr_all(old_dir, ++ hidden_old_dir_dentry->d_inode); ++ ++ out_lock: ++ /* double_unlock will dput the new/old parent dentries ++ * whose refcnts were incremented via get_parent above. */ ++ double_unlock(hidden_old_dir_dentry, hidden_new_dir_dentry); ++ dput(hidden_new_dentry); ++ dput(hidden_old_dentry); ++ ++ out: ++ return err; ++} ++ ++int rename_nondir(inode_t *old_dir, dentry_t *old_dentry, ++ inode_t *new_dir, dentry_t *new_dentry) ++{ ++ int err=0; ++ ++ check_mini_fo_dentry(old_dentry); ++ check_mini_fo_dentry(new_dentry); ++ check_mini_fo_inode(old_dir); ++ check_mini_fo_inode(new_dir); ++ ++ /* state: UNMODIFIED */ ++ if(dtost(old_dentry) == UNMODIFIED) { ++ err = nondir_unmod_to_mod(old_dentry, 1); ++ if(err) { ++ err = -EINVAL; ++ goto out; ++ } ++ } ++ ++ /* the easy states */ ++ if(exists_in_storage(old_dentry)) { ++ ++ dentry_t *hidden_old_dentry; ++ dentry_t *hidden_new_dentry; ++ dentry_t *hidden_old_dir_dentry; ++ dentry_t *hidden_new_dir_dentry; ++ ++ /* if old file is MODIFIED, add it to the deleted_list */ ++ if(dtopd(old_dentry)->state == MODIFIED) { ++ meta_add_d_entry(old_dentry->d_parent, ++ old_dentry->d_name.name, ++ old_dentry->d_name.len); ++ ++ dput(dtohd(old_dentry)); ++ } ++ /* if old file is CREATED, we only release the base dentry */ ++ if(dtopd(old_dentry)->state == CREATED) { ++ if(dtohd(old_dentry)) ++ dput(dtohd(old_dentry)); ++ } ++ ++ /* now setup the new states (depends on new_dentry state) */ ++ /* new dentry state = MODIFIED */ ++ if(dtopd(new_dentry)->state == MODIFIED) { ++ meta_add_d_entry(new_dentry->d_parent, ++ new_dentry->d_name.name, ++ new_dentry->d_name.len); ++ ++ /* new dentry will be d_put'ed later by the vfs ++ * so don't do it here ++ * dput(dtohd(new_dentry)); ++ */ ++ dtohd(old_dentry) = NULL; ++ dtopd(old_dentry)->state = DEL_REWRITTEN; ++ } ++ /* new dentry state = UNMODIFIED */ ++ else if(dtopd(new_dentry)->state == UNMODIFIED) { ++ if(get_neg_sto_dentry(new_dentry)) ++ return -EINVAL; ++ ++ meta_add_d_entry(new_dentry->d_parent, ++ new_dentry->d_name.name, ++ new_dentry->d_name.len); ++ ++ /* is this right??? */ ++ /*dput(dtohd(new_dentry));*/ ++ dtohd(old_dentry) = NULL; ++ dtopd(old_dentry)->state = DEL_REWRITTEN; ++ } ++ /* new dentry state = CREATED */ ++ else if(dtopd(new_dentry)->state == CREATED) { ++ /* we keep the neg. base dentry (if exists) */ ++ dtohd(old_dentry) = dtohd(new_dentry); ++ /* ...and set it to Null, or we'll get ++ * dcache.c:345 if it gets dput twice... */ ++ dtohd(new_dentry) = NULL; ++ dtopd(old_dentry)->state = CREATED; ++ } ++ /* new dentry state = NON_EXISTANT */ ++ else if(dtopd(new_dentry)->state == NON_EXISTANT) { ++ if(get_neg_sto_dentry(new_dentry)) ++ return -EINVAL; ++ ++ /* we keep the neg. base dentry (if exists) */ ++ dtohd(old_dentry) = dtohd(new_dentry); ++ /* ...and set it to Null, or we'll get ++ * Dr. dcache.c:345 if it gets dput twice... */ ++ dtohd(new_dentry) = NULL; ++ dtopd(old_dentry)->state = CREATED; ++ } ++ /* new dentry state = DEL_REWRITTEN or DELETED */ ++ else if(dtopd(new_dentry)->state == DEL_REWRITTEN || ++ dtopd(new_dentry)->state == DELETED) { ++ dtohd(old_dentry) = NULL; ++ dtopd(old_dentry)->state = DEL_REWRITTEN; ++ } ++ else { /* not possible, uhh, ahh */ ++ printk(KERN_CRIT ++ "mini_fo: rename_reg_file: invalid state detected [1].\n"); ++ return -1; ++ } ++ ++ /* now we definitely have a sto file */ ++ hidden_old_dentry = dtohd2(old_dentry); ++ hidden_new_dentry = dtohd2(new_dentry); ++ ++ dget(hidden_old_dentry); ++ dget(hidden_new_dentry); ++ ++ hidden_old_dir_dentry = dget(hidden_old_dentry->d_parent); ++ hidden_new_dir_dentry = dget(hidden_new_dentry->d_parent); ++ double_lock(hidden_old_dir_dentry, hidden_new_dir_dentry); ++ ++ err = vfs_rename(hidden_old_dir_dentry->d_inode, ++ hidden_old_dentry, ++ hidden_new_dir_dentry->d_inode, ++ hidden_new_dentry); ++ if(err) ++ goto out_lock; ++ ++ fist_copy_attr_all(new_dir, hidden_new_dir_dentry->d_inode); ++ if (new_dir != old_dir) ++ fist_copy_attr_all(old_dir, hidden_old_dir_dentry->d_inode); ++ ++ out_lock: ++ /* double_unlock will dput the new/old parent dentries ++ * whose refcnts were incremented via get_parent above. ++ */ ++ double_unlock(hidden_old_dir_dentry, hidden_new_dir_dentry); ++ dput(hidden_new_dentry); ++ dput(hidden_old_dentry); ++ out: ++ return err; ++ } ++ else { /* invalid state */ ++ printk(KERN_CRIT "mini_fo: rename_reg_file: ERROR: invalid state detected [2].\n"); ++ return -1; ++ } ++} ++ ++ ++STATIC int ++mini_fo_readlink(dentry_t *dentry, char *buf, int bufsiz) ++{ ++ int err=0; ++ dentry_t *hidden_dentry = NULL; ++ ++ if(dtohd2(dentry) && dtohd2(dentry)->d_inode) { ++ hidden_dentry = dtohd2(dentry); ++ } else if(dtohd(dentry) && dtohd(dentry)->d_inode) { ++ hidden_dentry = dtohd(dentry); ++ } else { ++ goto out; ++ } ++ ++ if (!hidden_dentry->d_inode->i_op || ++ !hidden_dentry->d_inode->i_op->readlink) { ++ err = -EINVAL; goto out; ++ } ++ ++ err = hidden_dentry->d_inode->i_op->readlink(hidden_dentry, ++ buf, ++ bufsiz); ++ if (err > 0) ++ fist_copy_attr_atime(dentry->d_inode, hidden_dentry->d_inode); ++ ++ out: ++ return err; ++} ++ ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,13) ++static int mini_fo_follow_link(dentry_t *dentry, struct nameidata *nd) ++#else ++static void* mini_fo_follow_link(dentry_t *dentry, struct nameidata *nd) ++#endif ++{ ++ char *buf; ++ int len = PAGE_SIZE, err; ++ mm_segment_t old_fs; ++ ++ /* in 2.6 this is freed by mini_fo_put_link called by __do_follow_link */ ++ buf = kmalloc(len, GFP_KERNEL); ++ if (!buf) { ++ err = -ENOMEM; ++ goto out; ++ } ++ ++ /* read the symlink, and then we will follow it */ ++ old_fs = get_fs(); ++ set_fs(KERNEL_DS); ++ err = dentry->d_inode->i_op->readlink(dentry, buf, len); ++ set_fs(old_fs); ++ if (err < 0) { ++ kfree(buf); ++ buf = NULL; ++ goto out; ++ } ++ buf[err] = 0; ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ nd_set_link(nd, buf); ++ err = 0; ++#else ++ err = vfs_follow_link(nd, buf); ++#endif ++ ++ out: ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++ kfree(buf); ++#endif ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,13) ++ return err; ++#else ++ return ERR_PTR(err); ++#endif ++} ++ ++STATIC ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,13) ++void mini_fo_put_link(struct dentry *dentry, struct nameidata *nd) ++#else ++void mini_fo_put_link(struct dentry *dentry, struct nameidata *nd, void *cookie) ++#endif ++{ ++ char *link; ++ link = nd_get_link(nd); ++ kfree(link); ++} ++#endif ++ ++STATIC int ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++mini_fo_permission(inode_t *inode, int mask, struct nameidata *nd) ++#else ++mini_fo_permission(inode_t *inode, int mask) ++#endif ++{ ++ inode_t *hidden_inode; ++ int mode; ++ int err; ++ ++ if(itohi2(inode)) { ++ hidden_inode = itohi2(inode); ++ } else { ++ hidden_inode = itohi(inode); ++ } ++ mode = inode->i_mode; ++ ++ /* not really needed, as permission handles everything: ++ * err = vfs_permission(inode, mask); ++ * if (err) ++ * goto out; ++ */ ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ err = permission(hidden_inode, mask, nd); ++#else ++ err = permission(hidden_inode, mask); ++#endif ++ ++ /* out: */ ++ return err; ++} ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++STATIC int ++mini_fo_inode_revalidate(dentry_t *dentry) ++{ ++ int err = 0; ++ dentry_t *hidden_dentry; ++ inode_t *hidden_inode; ++ ++ ASSERT(dentry->d_inode); ++ ASSERT(itopd(dentry->d_inode)); ++ ++ if(itohi2(dentry->d_inode)) { ++ hidden_dentry = dtohd2(dentry); ++ hidden_inode = hidden_dentry->d_inode; ++ } else if(itohi(dentry->d_inode)) { ++ hidden_dentry = dtohd(dentry); ++ hidden_inode = hidden_dentry->d_inode; ++ } else { ++ printk(KERN_CRIT "mini_fo_inode_revalidate: ERROR, invalid state detected.\n"); ++ err = -ENOENT; ++ goto out; ++ } ++ if (hidden_inode && hidden_inode->i_op && hidden_inode->i_op->revalidate){ ++ err = hidden_inode->i_op->revalidate(hidden_dentry); ++ if (err) ++ goto out; ++ } ++ fist_copy_attr_all(dentry->d_inode, hidden_inode); ++ out: ++ return err; ++} ++#endif ++ ++STATIC int ++mini_fo_setattr(dentry_t *dentry, struct iattr *ia) ++{ ++ int err = 0; ++ ++ check_mini_fo_dentry(dentry); ++ ++ if(!is_mini_fo_existant(dentry)) { ++ printk(KERN_CRIT "mini_fo_setattr: ERROR, invalid state detected [1].\n"); ++ goto out; ++ } ++ ++ if(dtost(dentry) == UNMODIFIED) { ++ if(!IS_COPY_FLAG(ia->ia_valid)) ++ goto out; /* we ignore these changes to base */ ++ ++ if(S_ISDIR(dentry->d_inode->i_mode)) { ++ err = dir_unmod_to_mod(dentry); ++ } else { ++ /* we copy contents if file is not beeing truncated */ ++ if(S_ISREG(dentry->d_inode->i_mode) && ++ !(ia->ia_size == 0 && (ia->ia_valid & ATTR_SIZE))) { ++ err = nondir_unmod_to_mod(dentry, 1); ++ } else ++ err = nondir_unmod_to_mod(dentry, 0); ++ } ++ if(err) { ++ err = -EINVAL; ++ printk(KERN_CRIT "mini_fo_setattr: ERROR changing states.\n"); ++ goto out; ++ } ++ } ++ if(!exists_in_storage(dentry)) { ++ printk(KERN_CRIT "mini_fo_setattr: ERROR, invalid state detected [2].\n"); ++ err = -EINVAL; ++ goto out; ++ } ++ ASSERT(dentry->d_inode); ++ ASSERT(dtohd2(dentry)); ++ ASSERT(itopd(dentry->d_inode)); ++ ASSERT(itohi2(dentry->d_inode)); ++ ++ err = notify_change(dtohd2(dentry), ia); ++ fist_copy_attr_all(dentry->d_inode, itohi2(dentry->d_inode)); ++ out: ++ return err; ++} ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++STATIC int ++mini_fo_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat) ++{ ++ int err = 0; ++ dentry_t *hidden_dentry; ++ ++ ASSERT(dentry->d_inode); ++ ASSERT(itopd(dentry->d_inode)); ++ ++ if(itohi2(dentry->d_inode)) { ++ hidden_dentry = dtohd2(dentry); ++ } else if(itohi(dentry->d_inode)) { ++ hidden_dentry = dtohd(dentry); ++ } else { ++ printk(KERN_CRIT "mini_fo_getattr: ERROR, invalid state detected.\n"); ++ err = -ENOENT; ++ goto out; ++ } ++ fist_copy_attr_all(dentry->d_inode, hidden_dentry->d_inode); ++ ++ ASSERT(hidden_dentry); ++ ASSERT(hidden_dentry->d_inode); ++ ASSERT(hidden_dentry->d_inode->i_op); ++ ++ generic_fillattr(dentry->d_inode, stat); ++ if (!stat->blksize) { ++ struct super_block *s = hidden_dentry->d_inode->i_sb; ++ unsigned blocks; ++ blocks = (stat->size+s->s_blocksize-1) >> s->s_blocksize_bits; ++ stat->blocks = (s->s_blocksize / 512) * blocks; ++ stat->blksize = s->s_blocksize; ++ } ++ out: ++ return err; ++} ++#endif ++ ++#if defined(XATTR) && (LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,20)) ++#if 0 /* no xattr_alloc() and xattr_free() */ ++/* This is lifted from fs/xattr.c */ ++static void * ++xattr_alloc(size_t size, size_t limit) ++{ ++ void *ptr; ++ ++ if (size > limit) ++ return ERR_PTR(-E2BIG); ++ ++ if (!size) /* size request, no buffer is needed */ ++ return NULL; ++ else if (size <= PAGE_SIZE) ++ ptr = kmalloc((unsigned long) size, GFP_KERNEL); ++ else ++ ptr = vmalloc((unsigned long) size); ++ if (!ptr) ++ return ERR_PTR(-ENOMEM); ++ return ptr; ++} ++ ++static void ++xattr_free(void *ptr, size_t size) ++{ ++ if (!size) /* size request, no buffer was needed */ ++ return; ++ else if (size <= PAGE_SIZE) ++ kfree(ptr); ++ else ++ vfree(ptr); ++} ++#endif /* no xattr_alloc() and xattr_free() */ ++ ++/* BKL held by caller. ++ * dentry->d_inode->i_sem down ++ */ ++STATIC int ++mini_fo_getxattr(struct dentry *dentry, const char *name, void *value, size_t size) { ++ struct dentry *hidden_dentry = NULL; ++ int err = -EOPNOTSUPP; ++ /* Define these anyway so we don't need as much ifdef'ed code. */ ++ char *encoded_name = NULL; ++ char *encoded_value = NULL; ++ ++ check_mini_fo_dentry(dentry); ++ ++ if(exists_in_storage(dentry)) ++ hidden_dentry = dtohd2(dentry); ++ else ++ hidden_dentry = dtohd(dentry); ++ ++ ASSERT(hidden_dentry); ++ ASSERT(hidden_dentry->d_inode); ++ ASSERT(hidden_dentry->d_inode->i_op); ++ ++ if (hidden_dentry->d_inode->i_op->getxattr) { ++ encoded_name = (char *)name; ++ encoded_value = (char *)value; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_dentry->d_inode->i_sem); ++#endif ++ /* lock_kernel() already done by caller. */ ++ err = hidden_dentry->d_inode->i_op->getxattr(hidden_dentry, encoded_name, encoded_value, size); ++ /* unlock_kernel() will be done by caller. */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_dentry->d_inode->i_sem); ++#endif ++ } ++ return err; ++} ++ ++/* BKL held by caller. ++ * dentry->d_inode->i_sem down ++ */ ++STATIC int ++#if ((LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,21) \ ++ && LINUX_VERSION_CODE <= KERNEL_VERSION(2,4,23)) \ ++ || LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0)) ++mini_fo_setxattr(struct dentry *dentry, const char *name, ++ const void *value, size_t size, int flags) ++#else ++mini_fo_setxattr(struct dentry *dentry, const char *name, ++ void *value, size_t size, int flags) ++#endif ++ ++{ ++ struct dentry *hidden_dentry = NULL; ++ int err = -EOPNOTSUPP; ++ ++ /* Define these anyway, so we don't have as much ifdef'ed code. */ ++ char *encoded_value = NULL; ++ char *encoded_name = NULL; ++ ++ check_mini_fo_dentry(dentry); ++ ++ if(exists_in_storage(dentry)) ++ hidden_dentry = dtohd2(dentry); ++ else ++ hidden_dentry = dtohd(dentry); ++ ++ ASSERT(hidden_dentry); ++ ASSERT(hidden_dentry->d_inode); ++ ASSERT(hidden_dentry->d_inode->i_op); ++ ++ if (hidden_dentry->d_inode->i_op->setxattr) { ++ encoded_name = (char *)name; ++ encoded_value = (char *)value; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_dentry->d_inode->i_sem); ++#endif ++ /* lock_kernel() already done by caller. */ ++ err = hidden_dentry->d_inode->i_op->setxattr(hidden_dentry, encoded_name, encoded_value, size, flags); ++ /* unlock_kernel() will be done by caller. */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_dentry->d_inode->i_sem); ++#endif ++ } ++ return err; ++} ++ ++/* BKL held by caller. ++ * dentry->d_inode->i_sem down ++ */ ++STATIC int ++mini_fo_removexattr(struct dentry *dentry, const char *name) { ++ struct dentry *hidden_dentry = NULL; ++ int err = -EOPNOTSUPP; ++ char *encoded_name; ++ ++ check_mini_fo_dentry(dentry); ++ ++ if(exists_in_storage(dentry)) ++ hidden_dentry = dtohd2(dentry); ++ else ++ hidden_dentry = dtohd(dentry); ++ ++ ASSERT(hidden_dentry); ++ ASSERT(hidden_dentry->d_inode); ++ ASSERT(hidden_dentry->d_inode->i_op); ++ ++ if (hidden_dentry->d_inode->i_op->removexattr) { ++ encoded_name = (char *)name; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_dentry->d_inode->i_sem); ++#endif ++ /* lock_kernel() already done by caller. */ ++ err = hidden_dentry->d_inode->i_op->removexattr(hidden_dentry, encoded_name); ++ /* unlock_kernel() will be done by caller. */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_dentry->d_inode->i_sem); ++#endif ++ } ++ return err; ++} ++ ++/* BKL held by caller. ++ * dentry->d_inode->i_sem down ++ */ ++STATIC int ++mini_fo_listxattr(struct dentry *dentry, char *list, size_t size) { ++ struct dentry *hidden_dentry = NULL; ++ int err = -EOPNOTSUPP; ++ char *encoded_list = NULL; ++ ++ check_mini_fo_dentry(dentry); ++ ++ if(exists_in_storage(dentry)) ++ hidden_dentry = dtohd2(dentry); ++ else ++ hidden_dentry = dtohd(dentry); ++ ++ ASSERT(hidden_dentry); ++ ASSERT(hidden_dentry->d_inode); ++ ASSERT(hidden_dentry->d_inode->i_op); ++ ++ if (hidden_dentry->d_inode->i_op->listxattr) { ++ encoded_list = list; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_dentry->d_inode->i_sem); ++#endif ++ /* lock_kernel() already done by caller. */ ++ err = hidden_dentry->d_inode->i_op->listxattr(hidden_dentry, encoded_list, size); ++ /* unlock_kernel() will be done by caller. */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_dentry->d_inode->i_sem); ++#endif ++ } ++ return err; ++} ++# endif /* defined(XATTR) && (LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,20)) */ ++ ++struct inode_operations mini_fo_symlink_iops = ++ { ++ readlink: mini_fo_readlink, ++ follow_link: mini_fo_follow_link, ++ /* mk: permission: mini_fo_permission, */ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++ revalidate: mini_fo_inode_revalidate, ++#endif ++ setattr: mini_fo_setattr, ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ getattr: mini_fo_getattr, ++ put_link: mini_fo_put_link, ++#endif ++ ++#if defined(XATTR) && (LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,20)) ++ setxattr: mini_fo_setxattr, ++ getxattr: mini_fo_getxattr, ++ listxattr: mini_fo_listxattr, ++ removexattr: mini_fo_removexattr ++# endif /* defined(XATTR) && (LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,20)) */ ++ }; ++ ++struct inode_operations mini_fo_dir_iops = ++ { ++ create: mini_fo_create, ++ lookup: mini_fo_lookup, ++ link: mini_fo_link, ++ unlink: mini_fo_unlink, ++ symlink: mini_fo_symlink, ++ mkdir: mini_fo_mkdir, ++ rmdir: mini_fo_rmdir, ++ mknod: mini_fo_mknod, ++ rename: mini_fo_rename, ++ /* no readlink/follow_link for non-symlinks */ ++ // off because we have setattr ++ // truncate: mini_fo_truncate, ++ /* mk:permission: mini_fo_permission, */ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++ revalidate: mini_fo_inode_revalidate, ++#endif ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ getattr: mini_fo_getattr, ++#endif ++ setattr: mini_fo_setattr, ++#if defined(XATTR) && (LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,20)) ++ setxattr: mini_fo_setxattr, ++ getxattr: mini_fo_getxattr, ++ listxattr: mini_fo_listxattr, ++ removexattr: mini_fo_removexattr ++# endif /* XATTR && LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,20) */ ++ }; ++ ++struct inode_operations mini_fo_main_iops = ++ { ++ /* permission: mini_fo_permission, */ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++ revalidate: mini_fo_inode_revalidate, ++#endif ++ setattr: mini_fo_setattr, ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ getattr: mini_fo_getattr, ++#endif ++#if defined(XATTR) && (LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,20)) ++ setxattr: mini_fo_setxattr, ++ getxattr: mini_fo_getxattr, ++ listxattr: mini_fo_listxattr, ++ removexattr: mini_fo_removexattr ++# endif /* XATTR && LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,20) */ ++ }; +--- /dev/null ++++ b/fs/mini_fo/main.c +@@ -0,0 +1,423 @@ ++/* ++ * Copyright (c) 1997-2003 Erez Zadok ++ * Copyright (c) 2001-2003 Stony Brook University ++ * ++ * For specific licensing information, see the COPYING file distributed with ++ * this package, or get one from ftp://ftp.filesystems.org/pub/fist/COPYING. ++ * ++ * This Copyright notice must be kept intact and distributed with all ++ * fistgen sources INCLUDING sources generated by fistgen. ++ */ ++/* ++ * Copyright (C) 2004, 2005 Markus Klotzbuecher ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ */ ++ ++/* ++ * $Id$ ++ */ ++ ++#ifdef HAVE_CONFIG_H ++# include ++#endif ++ ++#include "fist.h" ++#include "mini_fo.h" ++#include ++ ++/* This definition must only appear after we include */ ++#ifndef MODULE_LICENSE ++# define MODULE_LICENSE(bison) ++#endif /* not MODULE_LICENSE */ ++ ++/* ++ * This is the mini_fo tri interpose function, which extends the ++ * functionality of the regular interpose by interposing a higher ++ * level inode on top of two lower level ones: the base filesystem ++ * inode and the storage filesystem inode. ++ * ++ * sb we pass is mini_fo's super_block ++ */ ++int ++mini_fo_tri_interpose(dentry_t *hidden_dentry, ++ dentry_t *hidden_sto_dentry, ++ dentry_t *dentry, super_block_t *sb, int flag) ++{ ++ inode_t *hidden_inode = NULL; ++ inode_t *hidden_sto_inode = NULL; /* store corresponding storage inode */ ++ int err = 0; ++ inode_t *inode; ++ ++ /* Pointer to hidden_sto_inode if exists, else to hidden_inode. ++ * This is used to copy the attributes of the correct inode. */ ++ inode_t *master_inode; ++ ++ if(hidden_dentry) ++ hidden_inode = hidden_dentry->d_inode; ++ if(hidden_sto_dentry) ++ hidden_sto_inode = hidden_sto_dentry->d_inode; ++ ++ ASSERT(dentry->d_inode == NULL); ++ ++ /* mk: One of the inodes associated with the dentrys is likely to ++ * be NULL, so carefull: ++ */ ++ ASSERT((hidden_inode != NULL) || (hidden_sto_inode != NULL)); ++ ++ if(hidden_sto_inode) ++ master_inode = hidden_sto_inode; ++ else ++ master_inode = hidden_inode; ++ ++ /* ++ * We allocate our new inode below, by calling iget. ++ * iget will call our read_inode which will initialize some ++ * of the new inode's fields ++ */ ++ ++ /* ++ * original: inode = iget(sb, hidden_inode->i_ino); ++ */ ++ inode = iget(sb, iunique(sb, 25)); ++ if (!inode) { ++ err = -EACCES; /* should be impossible??? */ ++ goto out; ++ } ++ ++ /* ++ * interpose the inode if not already interposed ++ * this is possible if the inode is being reused ++ * XXX: what happens if we get_empty_inode() but there's another already? ++ * for now, ASSERT() that this can't happen; fix later. ++ */ ++ if (itohi(inode) != NULL) { ++ printk(KERN_CRIT "mini_fo_tri_interpose: itohi(inode) != NULL.\n"); ++ } ++ if (itohi2(inode) != NULL) { ++ printk(KERN_CRIT "mini_fo_tri_interpose: itohi2(inode) != NULL.\n"); ++ } ++ ++ /* mk: Carefull, igrab can't handle NULL inodes (ok, why should it?), so ++ * we need to check here: ++ */ ++ if(hidden_inode) ++ itohi(inode) = igrab(hidden_inode); ++ else ++ itohi(inode) = NULL; ++ ++ if(hidden_sto_inode) ++ itohi2(inode) = igrab(hidden_sto_inode); ++ else ++ itohi2(inode) = NULL; ++ ++ ++ /* Use different set of inode ops for symlinks & directories*/ ++ if (S_ISLNK(master_inode->i_mode)) ++ inode->i_op = &mini_fo_symlink_iops; ++ else if (S_ISDIR(master_inode->i_mode)) ++ inode->i_op = &mini_fo_dir_iops; ++ ++ /* Use different set of file ops for directories */ ++ if (S_ISDIR(master_inode->i_mode)) ++ inode->i_fop = &mini_fo_dir_fops; ++ ++ /* properly initialize special inodes */ ++ if (S_ISBLK(master_inode->i_mode) || S_ISCHR(master_inode->i_mode) || ++ S_ISFIFO(master_inode->i_mode) || S_ISSOCK(master_inode->i_mode)) { ++ init_special_inode(inode, master_inode->i_mode, master_inode->i_rdev); ++ } ++ ++ /* Fix our inode's address operations to that of the lower inode */ ++ if (inode->i_mapping->a_ops != master_inode->i_mapping->a_ops) { ++ inode->i_mapping->a_ops = master_inode->i_mapping->a_ops; ++ } ++ ++ /* only (our) lookup wants to do a d_add */ ++ if (flag) ++ d_add(dentry, inode); ++ else ++ d_instantiate(dentry, inode); ++ ++ ASSERT(dtopd(dentry) != NULL); ++ ++ /* all well, copy inode attributes */ ++ fist_copy_attr_all(inode, master_inode); ++ ++ out: ++ return err; ++} ++ ++/* parse mount options "base=" and "sto=" */ ++dentry_t * ++mini_fo_parse_options(super_block_t *sb, char *options) ++{ ++ dentry_t *hidden_root = ERR_PTR(-EINVAL); ++ dentry_t *hidden_root2 = ERR_PTR(-EINVAL); ++ struct nameidata nd, nd2; ++ char *name, *tmp, *end; ++ int err = 0; ++ ++ /* We don't want to go off the end of our arguments later on. */ ++ for (end = options; *end; end++); ++ ++ while (options < end) { ++ tmp = options; ++ while (*tmp && *tmp != ',') ++ tmp++; ++ *tmp = '\0'; ++ if (!strncmp("base=", options, 5)) { ++ name = options + 5; ++ printk(KERN_INFO "mini_fo: using base directory: %s\n", name); ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++ if (path_init(name, LOOKUP_FOLLOW, &nd)) ++ err = path_walk(name, &nd); ++#else ++ err = path_lookup(name, LOOKUP_FOLLOW, &nd); ++#endif ++ if (err) { ++ printk(KERN_CRIT "mini_fo: error accessing hidden directory '%s'\n", name); ++ hidden_root = ERR_PTR(err); ++ goto out; ++ } ++ hidden_root = nd.dentry; ++ stopd(sb)->base_dir_dentry = nd.dentry; ++ stopd(sb)->hidden_mnt = nd.mnt; ++ ++ } else if(!strncmp("sto=", options, 4)) { ++ /* parse the storage dir */ ++ name = options + 4; ++ printk(KERN_INFO "mini_fo: using storage directory: %s\n", name); ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++ if(path_init(name, LOOKUP_FOLLOW, &nd2)) ++ err = path_walk(name, &nd2); ++#else ++ err = path_lookup(name, LOOKUP_FOLLOW, &nd2); ++#endif ++ if(err) { ++ printk(KERN_CRIT "mini_fo: error accessing hidden storage directory '%s'\n", name); ++ ++ hidden_root2 = ERR_PTR(err); ++ goto out; ++ } ++ hidden_root2 = nd2.dentry; ++ stopd(sb)->storage_dir_dentry = nd2.dentry; ++ stopd(sb)->hidden_mnt2 = nd2.mnt; ++ stohs2(sb) = hidden_root2->d_sb; ++ ++ /* validate storage dir, this is done in ++ * mini_fo_read_super for the base directory. ++ */ ++ if (IS_ERR(hidden_root2)) { ++ printk(KERN_WARNING "mini_fo_parse_options: storage dentry lookup failed (err = %ld)\n", PTR_ERR(hidden_root2)); ++ goto out; ++ } ++ if (!hidden_root2->d_inode) { ++ printk(KERN_WARNING "mini_fo_parse_options: no storage dir to interpose on.\n"); ++ goto out; ++ } ++ stohs2(sb) = hidden_root2->d_sb; ++ } else { ++ printk(KERN_WARNING "mini_fo: unrecognized option '%s'\n", options); ++ hidden_root = ERR_PTR(-EINVAL); ++ goto out; ++ } ++ options = tmp + 1; ++ } ++ ++ out: ++ if(IS_ERR(hidden_root2)) ++ return hidden_root2; ++ return hidden_root; ++} ++ ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++static int ++#else ++super_block_t * ++#endif ++mini_fo_read_super(super_block_t *sb, void *raw_data, int silent) ++{ ++ dentry_t *hidden_root; ++ int err = 0; ++ ++ if (!raw_data) { ++ printk(KERN_WARNING "mini_fo_read_super: missing argument\n"); ++ err = -EINVAL; ++ goto out; ++ } ++ /* ++ * Allocate superblock private data ++ */ ++ __stopd(sb) = kmalloc(sizeof(struct mini_fo_sb_info), GFP_KERNEL); ++ if (!stopd(sb)) { ++ printk(KERN_WARNING "%s: out of memory\n", __FUNCTION__); ++ err = -ENOMEM; ++ goto out; ++ } ++ stohs(sb) = NULL; ++ ++ hidden_root = mini_fo_parse_options(sb, raw_data); ++ if (IS_ERR(hidden_root)) { ++ printk(KERN_WARNING "mini_fo_read_super: lookup_dentry failed (err = %ld)\n", PTR_ERR(hidden_root)); ++ err = PTR_ERR(hidden_root); ++ goto out_free; ++ } ++ if (!hidden_root->d_inode) { ++ printk(KERN_WARNING "mini_fo_read_super: no directory to interpose on\n"); ++ goto out_free; ++ } ++ stohs(sb) = hidden_root->d_sb; ++ ++ /* ++ * Linux 2.4.2-ac3 and beyond has code in ++ * mm/filemap.c:generic_file_write() that requires sb->s_maxbytes ++ * to be populated. If not set, all write()s under that sb will ++ * return 0. ++ * ++ * Linux 2.4.4+ automatically sets s_maxbytes to MAX_NON_LFS; ++ * the filesystem should override it only if it supports LFS. ++ */ ++ /* non-SCA code is good to go with LFS */ ++ sb->s_maxbytes = hidden_root->d_sb->s_maxbytes; ++ ++ sb->s_op = &mini_fo_sops; ++ /* ++ * we can't use d_alloc_root if we want to use ++ * our own interpose function unchanged, ++ * so we simply replicate *most* of the code in d_alloc_root here ++ */ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++ sb->s_root = d_alloc(NULL, &(const struct qstr) { "/", 1, 0 }); ++#else ++ sb->s_root = d_alloc(NULL, &(const struct qstr){hash: 0, name: "/", len : 1}); ++#endif ++ if (IS_ERR(sb->s_root)) { ++ printk(KERN_WARNING "mini_fo_read_super: d_alloc failed\n"); ++ err = -ENOMEM; ++ goto out_dput; ++ } ++ ++ sb->s_root->d_op = &mini_fo_dops; ++ sb->s_root->d_sb = sb; ++ sb->s_root->d_parent = sb->s_root; ++ ++ /* link the upper and lower dentries */ ++ __dtopd(sb->s_root) = (struct mini_fo_dentry_info *) ++ kmalloc(sizeof(struct mini_fo_dentry_info), GFP_KERNEL); ++ if (!dtopd(sb->s_root)) { ++ err = -ENOMEM; ++ goto out_dput2; ++ } ++ dtopd(sb->s_root)->state = MODIFIED; ++ dtohd(sb->s_root) = hidden_root; ++ ++ /* fanout relevant, interpose on storage root dentry too */ ++ dtohd2(sb->s_root) = stopd(sb)->storage_dir_dentry; ++ ++ /* ...and call tri-interpose to interpose root dir inodes ++ * if (mini_fo_interpose(hidden_root, sb->s_root, sb, 0)) ++ */ ++ if(mini_fo_tri_interpose(hidden_root, dtohd2(sb->s_root), sb->s_root, sb, 0)) ++ goto out_dput2; ++ ++ /* initalize the wol list */ ++ itopd(sb->s_root->d_inode)->deleted_list_size = -1; ++ itopd(sb->s_root->d_inode)->renamed_list_size = -1; ++ meta_build_lists(sb->s_root); ++ ++ goto out; ++ ++ out_dput2: ++ dput(sb->s_root); ++ out_dput: ++ dput(hidden_root); ++ dput(dtohd2(sb->s_root)); /* release the hidden_sto_dentry too */ ++ out_free: ++ kfree(stopd(sb)); ++ __stopd(sb) = NULL; ++ out: ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ return err; ++#else ++ if (err) { ++ return ERR_PTR(err); ++ } else { ++ return sb; ++ } ++#endif ++} ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18) ++static int mini_fo_get_sb(struct file_system_type *fs_type, ++ int flags, const char *dev_name, ++ void *raw_data, struct vfsmount *mnt) ++{ ++ return get_sb_nodev(fs_type, flags, raw_data, mini_fo_read_super, mnt); ++} ++#else ++static struct super_block *mini_fo_get_sb(struct file_system_type *fs_type, ++ int flags, const char *dev_name, ++ void *raw_data) ++{ ++ return get_sb_nodev(fs_type, flags, raw_data, mini_fo_read_super); ++} ++#endif ++ ++void mini_fo_kill_block_super(struct super_block *sb) ++{ ++ generic_shutdown_super(sb); ++ /* ++ * XXX: BUG: Halcrow: Things get unstable sometime after this point: ++ * lib/rwsem-spinlock.c:127: spin_is_locked on uninitialized ++ * fs/fs-writeback.c:402: spin_lock(fs/super.c:a0381828) already ++ * locked by fs/fs-writeback.c/402 ++ * ++ * Apparently, someone's not releasing a lock on sb_lock... ++ */ ++} ++ ++static struct file_system_type mini_fo_fs_type = { ++ .owner = THIS_MODULE, ++ .name = "mini_fo", ++ .get_sb = mini_fo_get_sb, ++ .kill_sb = mini_fo_kill_block_super, ++ .fs_flags = 0, ++}; ++ ++ ++#else ++static DECLARE_FSTYPE(mini_fo_fs_type, "mini_fo", mini_fo_read_super, 0); ++#endif ++ ++static int __init init_mini_fo_fs(void) ++{ ++ printk("Registering mini_fo version $Id$\n"); ++ return register_filesystem(&mini_fo_fs_type); ++} ++static void __exit exit_mini_fo_fs(void) ++{ ++ printk("Unregistering mini_fo version $Id$\n"); ++ unregister_filesystem(&mini_fo_fs_type); ++} ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++EXPORT_NO_SYMBOLS; ++#endif ++ ++MODULE_AUTHOR("Erez Zadok "); ++MODULE_DESCRIPTION("FiST-generated mini_fo filesystem"); ++MODULE_LICENSE("GPL"); ++ ++/* MODULE_PARM(fist_debug_var, "i"); */ ++/* MODULE_PARM_DESC(fist_debug_var, "Debug level"); */ ++ ++module_init(init_mini_fo_fs) ++module_exit(exit_mini_fo_fs) +--- /dev/null ++++ b/fs/mini_fo/Makefile +@@ -0,0 +1,17 @@ ++# ++# Makefile for mini_fo 2.4 and 2.6 Linux kernels ++# ++# Copyright (C) 2004, 2005 Markus Klotzbuecher ++# ++# This program is free software; you can redistribute it and/or ++# modify it under the terms of the GNU General Public License ++# as published by the Free Software Foundation; either version ++# 2 of the License, or (at your option) any later version. ++# ++ ++obj-$(CONFIG_MINI_FO) := mini_fo.o ++mini_fo-objs := meta.o dentry.o file.o inode.o main.o super.o state.o aux.o ++ ++# dependencies ++${mini_fo-objs}: mini_fo.h fist.h ++ +--- /dev/null ++++ b/fs/mini_fo/meta.c +@@ -0,0 +1,1000 @@ ++/* ++ * Copyright (C) 2004, 2005 Markus Klotzbuecher ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ */ ++ ++#ifdef HAVE_CONFIG_H ++# include ++#endif /* HAVE_CONFIG_H */ ++#include "fist.h" ++#include "mini_fo.h" ++ ++int meta_build_lists(dentry_t *dentry) ++{ ++ struct mini_fo_inode_info *inode_info; ++ ++ dentry_t *meta_dentry = 0; ++ file_t *meta_file = 0; ++ mm_segment_t old_fs; ++ void *buf; ++ ++ int bytes, len; ++ struct vfsmount *meta_mnt; ++ char *entry; ++ ++ inode_info = itopd(dentry->d_inode); ++ if(!(inode_info->deleted_list_size == -1 && ++ inode_info->renamed_list_size == -1)) { ++ printk(KERN_CRIT "mini_fo: meta_build_lists: \ ++ Error, list(s) not virgin.\n"); ++ return -1; ++ } ++ ++ /* init our meta lists */ ++ INIT_LIST_HEAD(&inode_info->deleted_list); ++ inode_info->deleted_list_size = 0; ++ ++ INIT_LIST_HEAD(&inode_info->renamed_list); ++ inode_info->renamed_list_size = 0; ++ ++ /* might there be a META-file? */ ++ if(dtohd2(dentry) && dtohd2(dentry)->d_inode) { ++ meta_dentry = lookup_one_len(META_FILENAME, ++ dtohd2(dentry), ++ strlen(META_FILENAME)); ++ if(!meta_dentry->d_inode) { ++ dput(meta_dentry); ++ goto out_ok; ++ } ++ /* $%& err, is this correct? */ ++ meta_mnt = stopd(dentry->d_inode->i_sb)->hidden_mnt2; ++ mntget(meta_mnt); ++ ++ ++ /* open META-file for reading */ ++ meta_file = dentry_open(meta_dentry, meta_mnt, 0x0); ++ if(!meta_file || IS_ERR(meta_file)) { ++ printk(KERN_CRIT "mini_fo: meta_build_lists: \ ++ ERROR opening META file.\n"); ++ goto out_err; ++ } ++ ++ /* check if fs supports reading */ ++ if(!meta_file->f_op->read) { ++ printk(KERN_CRIT "mini_fo: meta_build_lists: \ ++ ERROR, fs does not support reading.\n"); ++ goto out_err_close; ++ } ++ ++ /* allocate a page for transfering the data */ ++ buf = (void *) __get_free_page(GFP_KERNEL); ++ if(!buf) { ++ printk(KERN_CRIT "mini_fo: meta_build_lists: \ ++ ERROR, out of mem.\n"); ++ goto out_err_close; ++ } ++ meta_file->f_pos = 0; ++ old_fs = get_fs(); ++ set_fs(KERNEL_DS); ++ do { ++ char *c; ++ bytes = meta_file->f_op->read(meta_file, buf, PAGE_SIZE, &meta_file->f_pos); ++ if(bytes == PAGE_SIZE) { ++ /* trim a cut off filename and adjust f_pos to get it next time */ ++ for(c = (char*) buf+PAGE_SIZE; ++ *c != '\n'; ++ c--, bytes--, meta_file->f_pos--); ++ } ++ entry = (char *) buf; ++ while(entry < (char *) buf+bytes) { ++ ++ char *old_path; ++ char *dir_name; ++ int old_len, new_len; ++ ++ /* len without '\n'*/ ++ len = (int) (strchr(entry, '\n') - entry); ++ switch (*entry) { ++ case 'D': ++ /* format: "D filename" */ ++ meta_list_add_d_entry(dentry, ++ entry+2, ++ len-2); ++ break; ++ case 'R': ++ /* format: "R path/xy/dir newDir" */ ++ old_path = entry+2; ++ dir_name = strchr(old_path, ' ') + 1; ++ old_len = dir_name - old_path - 1; ++ new_len = ((int) entry) + len - ((int ) dir_name); ++ meta_list_add_r_entry(dentry, ++ old_path, ++ old_len, ++ dir_name, ++ new_len); ++ break; ++ default: ++ /* unknown entry type detected */ ++ break; ++ } ++ entry += len+1; ++ } ++ ++ } while(meta_file->f_pos < meta_dentry->d_inode->i_size); ++ ++ free_page((unsigned long) buf); ++ set_fs(old_fs); ++ fput(meta_file); ++ } ++ goto out_ok; ++ ++ out_err_close: ++ fput(meta_file); ++ out_err: ++ mntput(meta_mnt); ++ dput(meta_dentry); ++ return -1; ++ out_ok: ++ return 1; /* check this!!! inode_info->wol_size; */ ++} ++ ++/* cleanups up all lists and free's the mem by dentry */ ++int meta_put_lists(dentry_t *dentry) ++{ ++ if(!dentry || !dentry->d_inode) { ++ printk("mini_fo: meta_put_lists: invalid dentry passed.\n"); ++ return -1; ++ } ++ return __meta_put_lists(dentry->d_inode); ++} ++ ++/* cleanups up all lists and free's the mem by inode */ ++int __meta_put_lists(inode_t *inode) ++{ ++ int err = 0; ++ if(!inode || !itopd(inode)) { ++ printk("mini_fo: __meta_put_lists: invalid inode passed.\n"); ++ return -1; ++ } ++ err = __meta_put_d_list(inode); ++ err |= __meta_put_r_list(inode); ++ return err; ++} ++ ++int meta_sync_lists(dentry_t *dentry) ++{ ++ int err = 0; ++ if(!dentry || !dentry->d_inode) { ++ printk("mini_fo: meta_sync_lists: \ ++ invalid dentry passed.\n"); ++ return -1; ++ } ++ err = meta_sync_d_list(dentry, 0); ++ err |= meta_sync_r_list(dentry, 1); ++ return err; ++} ++ ++ ++/* remove all D entries from the renamed list and free the mem */ ++int __meta_put_d_list(inode_t *inode) ++{ ++ struct list_head *tmp; ++ struct deleted_entry *del_entry; ++ struct mini_fo_inode_info *inode_info; ++ ++ if(!inode || !itopd(inode)) { ++ printk(KERN_CRIT "mini_fo: __meta_put_d_list: \ ++ invalid inode passed.\n"); ++ return -1; ++ } ++ inode_info = itopd(inode); ++ ++ /* nuke the DELETED-list */ ++ if(inode_info->deleted_list_size <= 0) ++ return 0; ++ ++ while(!list_empty(&inode_info->deleted_list)) { ++ tmp = inode_info->deleted_list.next; ++ list_del(tmp); ++ del_entry = list_entry(tmp, struct deleted_entry, list); ++ kfree(del_entry->name); ++ kfree(del_entry); ++ } ++ inode_info->deleted_list_size = 0; ++ ++ return 0; ++} ++ ++/* remove all R entries from the renamed list and free the mem */ ++int __meta_put_r_list(inode_t *inode) ++{ ++ struct list_head *tmp; ++ struct renamed_entry *ren_entry; ++ struct mini_fo_inode_info *inode_info; ++ ++ if(!inode || !itopd(inode)) { ++ printk(KERN_CRIT "mini_fo: meta_put_r_list: invalid inode.\n"); ++ return -1; ++ } ++ inode_info = itopd(inode); ++ ++ /* nuke the RENAMED-list */ ++ if(inode_info->renamed_list_size <= 0) ++ return 0; ++ ++ while(!list_empty(&inode_info->renamed_list)) { ++ tmp = inode_info->renamed_list.next; ++ list_del(tmp); ++ ren_entry = list_entry(tmp, struct renamed_entry, list); ++ kfree(ren_entry->new_name); ++ kfree(ren_entry->old_name); ++ kfree(ren_entry); ++ } ++ inode_info->renamed_list_size = 0; ++ ++ return 0; ++} ++ ++int meta_add_d_entry(dentry_t *dentry, const char *name, int len) ++{ ++ int err = 0; ++ err = meta_list_add_d_entry(dentry, name, len); ++ err |= meta_write_d_entry(dentry,name,len); ++ return err; ++} ++ ++/* add a D entry to the deleted list */ ++int meta_list_add_d_entry(dentry_t *dentry, const char *name, int len) ++{ ++ struct deleted_entry *del_entry; ++ struct mini_fo_inode_info *inode_info; ++ ++ if(!dentry || !dentry->d_inode) { ++ printk(KERN_CRIT "mini_fo: meta_list_add_d_entry: \ ++ invalid dentry passed.\n"); ++ return -1; ++ } ++ inode_info = itopd(dentry->d_inode); ++ ++ if(inode_info->deleted_list_size < 0) ++ return -1; ++ ++ del_entry = (struct deleted_entry *) ++ kmalloc(sizeof(struct deleted_entry), GFP_KERNEL); ++ del_entry->name = (char*) kmalloc(len, GFP_KERNEL); ++ if(!del_entry || !del_entry->name) { ++ printk(KERN_CRIT "mini_fo: meta_list_add_d_entry: \ ++ out of mem.\n"); ++ kfree(del_entry->name); ++ kfree(del_entry); ++ return -ENOMEM; ++ } ++ ++ strncpy(del_entry->name, name, len); ++ del_entry->len = len; ++ ++ list_add(&del_entry->list, &inode_info->deleted_list); ++ inode_info->deleted_list_size++; ++ return 0; ++} ++ ++int meta_add_r_entry(dentry_t *dentry, ++ const char *old_name, int old_len, ++ const char *new_name, int new_len) ++{ ++ int err = 0; ++ err = meta_list_add_r_entry(dentry, ++ old_name, old_len, ++ new_name, new_len); ++ err |= meta_write_r_entry(dentry, ++ old_name, old_len, ++ new_name, new_len); ++ return err; ++} ++ ++/* add a R entry to the renamed list */ ++int meta_list_add_r_entry(dentry_t *dentry, ++ const char *old_name, int old_len, ++ const char *new_name, int new_len) ++{ ++ struct renamed_entry *ren_entry; ++ struct mini_fo_inode_info *inode_info; ++ ++ if(!dentry || !dentry->d_inode) { ++ printk(KERN_CRIT "mini_fo: meta_list_add_r_entry: \ ++ invalid dentry passed.\n"); ++ return -1; ++ } ++ inode_info = itopd(dentry->d_inode); ++ ++ if(inode_info->renamed_list_size < 0) ++ return -1; ++ ++ ren_entry = (struct renamed_entry *) ++ kmalloc(sizeof(struct renamed_entry), GFP_KERNEL); ++ ren_entry->old_name = (char*) kmalloc(old_len, GFP_KERNEL); ++ ren_entry->new_name = (char*) kmalloc(new_len, GFP_KERNEL); ++ ++ if(!ren_entry || !ren_entry->old_name || !ren_entry->new_name) { ++ printk(KERN_CRIT "mini_fo: meta_list_add_r_entry: \ ++ out of mem.\n"); ++ kfree(ren_entry->new_name); ++ kfree(ren_entry->old_name); ++ kfree(ren_entry); ++ return -ENOMEM; ++ } ++ ++ strncpy(ren_entry->old_name, old_name, old_len); ++ ren_entry->old_len = old_len; ++ strncpy(ren_entry->new_name, new_name, new_len); ++ ren_entry->new_len = new_len; ++ ++ list_add(&ren_entry->list, &inode_info->renamed_list); ++ inode_info->renamed_list_size++; ++ return 0; ++} ++ ++ ++int meta_remove_r_entry(dentry_t *dentry, const char *name, int len) ++{ ++ int err = 0; ++ if(!dentry || !dentry->d_inode) { ++ printk(KERN_CRIT ++ "mini_fo: meta_remove_r_entry: \ ++ invalid dentry passed.\n"); ++ return -1; ++ } ++ ++ err = meta_list_remove_r_entry(dentry, name, len); ++ err |= meta_sync_lists(dentry); ++ return err; ++} ++ ++int meta_list_remove_r_entry(dentry_t *dentry, const char *name, int len) ++{ ++ if(!dentry || !dentry->d_inode) { ++ printk(KERN_CRIT ++ "mini_fo: meta_list_remove_r_entry: \ ++ invalid dentry passed.\n"); ++ return -1; ++ } ++ return __meta_list_remove_r_entry(dentry->d_inode, name, len); ++} ++ ++int __meta_list_remove_r_entry(inode_t *inode, const char *name, int len) ++{ ++ struct list_head *tmp; ++ struct renamed_entry *ren_entry; ++ struct mini_fo_inode_info *inode_info; ++ ++ if(!inode || !itopd(inode)) ++ printk(KERN_CRIT ++ "mini_fo: __meta_list_remove_r_entry: \ ++ invalid inode passed.\n"); ++ inode_info = itopd(inode); ++ ++ if(inode_info->renamed_list_size < 0) ++ return -1; ++ if(inode_info->renamed_list_size == 0) ++ return 1; ++ ++ list_for_each(tmp, &inode_info->renamed_list) { ++ ren_entry = list_entry(tmp, struct renamed_entry, list); ++ if(ren_entry->new_len != len) ++ continue; ++ ++ if(!strncmp(ren_entry->new_name, name, len)) { ++ list_del(tmp); ++ kfree(ren_entry->new_name); ++ kfree(ren_entry->old_name); ++ kfree(ren_entry); ++ inode_info->renamed_list_size--; ++ return 0; ++ } ++ } ++ return 1; ++} ++ ++ ++/* append a single D entry to the meta file */ ++int meta_write_d_entry(dentry_t *dentry, const char *name, int len) ++{ ++ dentry_t *meta_dentry = 0; ++ file_t *meta_file = 0; ++ mm_segment_t old_fs; ++ ++ int bytes, err; ++ struct vfsmount *meta_mnt = 0; ++ char *buf; ++ ++ err = 0; ++ ++ if(itopd(dentry->d_inode)->deleted_list_size < 0) { ++ err = -1; ++ goto out; ++ } ++ ++ if(dtopd(dentry)->state == UNMODIFIED) { ++ err = build_sto_structure(dentry->d_parent, dentry); ++ if(err) { ++ printk(KERN_CRIT "mini_fo: meta_write_d_entry: \ ++ build_sto_structure failed.\n"); ++ goto out; ++ } ++ } ++ meta_dentry = lookup_one_len(META_FILENAME, ++ dtohd2(dentry), strlen (META_FILENAME)); ++ ++ /* We need to create a META-file */ ++ if(!meta_dentry->d_inode) { ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ vfs_create(dtohd2(dentry)->d_inode, ++ meta_dentry, ++ S_IRUSR | S_IWUSR, ++ NULL); ++#else ++ vfs_create(dtohd2(dentry)->d_inode, ++ meta_dentry, ++ S_IRUSR | S_IWUSR); ++#endif ++ } ++ /* open META-file for writing */ ++ meta_file = dentry_open(meta_dentry, meta_mnt, 0x1); ++ if(!meta_file || IS_ERR(meta_file)) { ++ printk(KERN_CRIT "mini_fo: meta_write_d_entry: \ ++ ERROR opening meta file.\n"); ++ mntput(meta_mnt); /* $%& is this necessary? */ ++ dput(meta_dentry); ++ err = -1; ++ goto out; ++ } ++ ++ /* check if fs supports writing */ ++ if(!meta_file->f_op->write) { ++ printk(KERN_CRIT "mini_fo: meta_write_d_entry: \ ++ ERROR, fs does not support writing.\n"); ++ goto out_err_close; ++ } ++ ++ meta_file->f_pos = meta_dentry->d_inode->i_size; /* append */ ++ old_fs = get_fs(); ++ set_fs(KERNEL_DS); ++ ++ /* size: len for name, 1 for \n and 2 for "D " */ ++ buf = (char *) kmalloc(len+3, GFP_KERNEL); ++ if (!buf) { ++ printk(KERN_CRIT "mini_fo: meta_write_d_entry: \ ++ out of mem.\n"); ++ return -ENOMEM; ++ } ++ ++ buf[0] = 'D'; ++ buf[1] = ' '; ++ strncpy(buf+2, name, len); ++ buf[len+2] = '\n'; ++ bytes = meta_file->f_op->write(meta_file, buf, len+3, ++ &meta_file->f_pos); ++ if(bytes != len+3) { ++ printk(KERN_CRIT "mini_fo: meta_write_d_entry: \ ++ ERROR writing.\n"); ++ err = -1; ++ } ++ kfree(buf); ++ set_fs(old_fs); ++ ++ out_err_close: ++ fput(meta_file); ++ out: ++ return err; ++} ++ ++/* append a single R entry to the meta file */ ++int meta_write_r_entry(dentry_t *dentry, ++ const char *old_name, int old_len, ++ const char *new_name, int new_len) ++{ ++ dentry_t *meta_dentry = 0; ++ file_t *meta_file = 0; ++ mm_segment_t old_fs; ++ ++ int bytes, err, buf_len; ++ struct vfsmount *meta_mnt = 0; ++ char *buf; ++ ++ ++ err = 0; ++ ++ if(itopd(dentry->d_inode)->renamed_list_size < 0) { ++ err = -1; ++ goto out; ++ } ++ ++ /* build the storage structure? */ ++ if(dtopd(dentry)->state == UNMODIFIED) { ++ err = build_sto_structure(dentry->d_parent, dentry); ++ if(err) { ++ printk(KERN_CRIT "mini_fo: meta_write_r_entry: \ ++ build_sto_structure failed.\n"); ++ goto out; ++ } ++ } ++ meta_dentry = lookup_one_len(META_FILENAME, ++ dtohd2(dentry), ++ strlen (META_FILENAME)); ++ if(!meta_dentry->d_inode) { ++ /* We need to create a META-file */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ vfs_create(dtohd2(dentry)->d_inode, ++ meta_dentry, S_IRUSR | S_IWUSR, NULL); ++#else ++ vfs_create(dtohd2(dentry)->d_inode, ++ meta_dentry, S_IRUSR | S_IWUSR); ++#endif ++ } ++ /* open META-file for writing */ ++ meta_file = dentry_open(meta_dentry, meta_mnt, 0x1); ++ if(!meta_file || IS_ERR(meta_file)) { ++ printk(KERN_CRIT "mini_fo: meta_write_r_entry: \ ++ ERROR opening meta file.\n"); ++ mntput(meta_mnt); ++ dput(meta_dentry); ++ err = -1; ++ goto out; ++ } ++ ++ /* check if fs supports writing */ ++ if(!meta_file->f_op->write) { ++ printk(KERN_CRIT "mini_fo: meta_write_r_entry: \ ++ ERROR, fs does not support writing.\n"); ++ goto out_err_close; ++ } ++ ++ meta_file->f_pos = meta_dentry->d_inode->i_size; /* append */ ++ old_fs = get_fs(); ++ set_fs(KERNEL_DS); ++ ++ /* size: 2 for "R ", old_len+new_len for names, 1 blank+1 \n */ ++ buf_len = old_len + new_len + 4; ++ buf = (char *) kmalloc(buf_len, GFP_KERNEL); ++ if (!buf) { ++ printk(KERN_CRIT "mini_fo: meta_write_r_entry: out of mem.\n"); ++ return -ENOMEM; ++ } ++ ++ buf[0] = 'R'; ++ buf[1] = ' '; ++ strncpy(buf + 2, old_name, old_len); ++ buf[old_len + 2] = ' '; ++ strncpy(buf + old_len + 3, new_name, new_len); ++ buf[buf_len -1] = '\n'; ++ bytes = meta_file->f_op->write(meta_file, buf, buf_len, &meta_file->f_pos); ++ if(bytes != buf_len) { ++ printk(KERN_CRIT "mini_fo: meta_write_r_entry: ERROR writing.\n"); ++ err = -1; ++ } ++ ++ kfree(buf); ++ set_fs(old_fs); ++ ++ out_err_close: ++ fput(meta_file); ++ out: ++ return err; ++} ++ ++/* sync D list to disk, append data if app_flag is 1 */ ++/* check the meta_mnt, which seems not to be used (properly) */ ++ ++int meta_sync_d_list(dentry_t *dentry, int app_flag) ++{ ++ dentry_t *meta_dentry; ++ file_t *meta_file; ++ mm_segment_t old_fs; ++ ++ int bytes, err; ++ struct vfsmount *meta_mnt; ++ char *buf; ++ ++ struct list_head *tmp; ++ struct deleted_entry *del_entry; ++ struct mini_fo_inode_info *inode_info; ++ ++ err = 0; ++ meta_file=0; ++ meta_mnt=0; ++ ++ if(!dentry || !dentry->d_inode) { ++ printk(KERN_CRIT "mini_fo: meta_sync_d_list: \ ++ invalid inode passed.\n"); ++ err = -1; ++ goto out; ++ } ++ inode_info = itopd(dentry->d_inode); ++ ++ if(inode_info->deleted_list_size < 0) { ++ err = -1; ++ goto out; ++ } ++ ++ /* ok, there is something to sync */ ++ ++ /* build the storage structure? */ ++ if(!dtohd2(dentry) && !itohi2(dentry->d_inode)) { ++ err = build_sto_structure(dentry->d_parent, dentry); ++ if(err) { ++ printk(KERN_CRIT "mini_fo: meta_sync_d_list: \ ++ build_sto_structure failed.\n"); ++ goto out; ++ } ++ } ++ meta_dentry = lookup_one_len(META_FILENAME, ++ dtohd2(dentry), ++ strlen(META_FILENAME)); ++ if(!meta_dentry->d_inode) { ++ /* We need to create a META-file */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ vfs_create(dtohd2(dentry)->d_inode, ++ meta_dentry, S_IRUSR | S_IWUSR, NULL); ++#else ++ vfs_create(dtohd2(dentry)->d_inode, ++ meta_dentry, S_IRUSR | S_IWUSR); ++#endif ++ app_flag = 0; ++ } ++ /* need we truncate the meta file? */ ++ if(!app_flag) { ++ struct iattr newattrs; ++ newattrs.ia_size = 0; ++ newattrs.ia_valid = ATTR_SIZE | ATTR_CTIME; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&meta_dentry->d_inode->i_mutex); ++#else ++ down(&meta_dentry->d_inode->i_sem); ++#endif ++ err = notify_change(meta_dentry, &newattrs); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&meta_dentry->d_inode->i_mutex); ++#else ++ up(&meta_dentry->d_inode->i_sem); ++#endif ++ ++ if(err || meta_dentry->d_inode->i_size != 0) { ++ printk(KERN_CRIT "mini_fo: meta_sync_d_list: \ ++ ERROR truncating meta file.\n"); ++ goto out_err_close; ++ } ++ } ++ ++ /* open META-file for writing */ ++ meta_file = dentry_open(meta_dentry, meta_mnt, 0x1); ++ if(!meta_file || IS_ERR(meta_file)) { ++ printk(KERN_CRIT "mini_fo: meta_sync_d_list: \ ++ ERROR opening meta file.\n"); ++ /* we don't mntget so we dont't mntput (for now) ++ * mntput(meta_mnt); ++ */ ++ dput(meta_dentry); ++ err = -1; ++ goto out; ++ } ++ ++ /* check if fs supports writing */ ++ if(!meta_file->f_op->write) { ++ printk(KERN_CRIT "mini_fo: meta_sync_d_list: \ ++ ERROR, fs does not support writing.\n"); ++ goto out_err_close; ++ } ++ ++ meta_file->f_pos = meta_dentry->d_inode->i_size; /* append */ ++ old_fs = get_fs(); ++ set_fs(KERNEL_DS); ++ ++ /* here we go... */ ++ list_for_each(tmp, &inode_info->deleted_list) { ++ del_entry = list_entry(tmp, struct deleted_entry, list); ++ ++ /* size: len for name, 1 for \n and 2 for "D " */ ++ buf = (char *) kmalloc(del_entry->len+3, GFP_KERNEL); ++ if (!buf) { ++ printk(KERN_CRIT "mini_fo: meta_sync_d_list: \ ++ out of mem.\n"); ++ return -ENOMEM; ++ } ++ ++ buf[0] = 'D'; ++ buf[1] = ' '; ++ strncpy(buf+2, del_entry->name, del_entry->len); ++ buf[del_entry->len+2] = '\n'; ++ bytes = meta_file->f_op->write(meta_file, buf, ++ del_entry->len+3, ++ &meta_file->f_pos); ++ if(bytes != del_entry->len+3) { ++ printk(KERN_CRIT "mini_fo: meta_sync_d_list: \ ++ ERROR writing.\n"); ++ err |= -1; ++ } ++ kfree(buf); ++ } ++ set_fs(old_fs); ++ ++ out_err_close: ++ fput(meta_file); ++ out: ++ return err; ++ ++} ++ ++int meta_sync_r_list(dentry_t *dentry, int app_flag) ++{ ++ dentry_t *meta_dentry; ++ file_t *meta_file; ++ mm_segment_t old_fs; ++ ++ int bytes, err, buf_len; ++ struct vfsmount *meta_mnt; ++ char *buf; ++ ++ struct list_head *tmp; ++ struct renamed_entry *ren_entry; ++ struct mini_fo_inode_info *inode_info; ++ ++ err = 0; ++ meta_file=0; ++ meta_mnt=0; ++ ++ if(!dentry || !dentry->d_inode) { ++ printk(KERN_CRIT "mini_fo: meta_sync_r_list: \ ++ invalid dentry passed.\n"); ++ err = -1; ++ goto out; ++ } ++ inode_info = itopd(dentry->d_inode); ++ ++ if(inode_info->deleted_list_size < 0) { ++ err = -1; ++ goto out; ++ } ++ ++ /* ok, there is something to sync */ ++ ++ /* build the storage structure? */ ++ if(!dtohd2(dentry) && !itohi2(dentry->d_inode)) { ++ err = build_sto_structure(dentry->d_parent, dentry); ++ if(err) { ++ printk(KERN_CRIT "mini_fo: meta_sync_r_list: \ ++ build_sto_structure failed.\n"); ++ goto out; ++ } ++ } ++ meta_dentry = lookup_one_len(META_FILENAME, ++ dtohd2(dentry), ++ strlen(META_FILENAME)); ++ if(!meta_dentry->d_inode) { ++ /* We need to create a META-file */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ vfs_create(dtohd2(dentry)->d_inode, ++ meta_dentry, S_IRUSR | S_IWUSR, NULL); ++#else ++ vfs_create(dtohd2(dentry)->d_inode, ++ meta_dentry, S_IRUSR | S_IWUSR); ++#endif ++ app_flag = 0; ++ } ++ /* need we truncate the meta file? */ ++ if(!app_flag) { ++ struct iattr newattrs; ++ newattrs.ia_size = 0; ++ newattrs.ia_valid = ATTR_SIZE | ATTR_CTIME; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&meta_dentry->d_inode->i_mutex); ++#else ++ down(&meta_dentry->d_inode->i_sem); ++#endif ++ err = notify_change(meta_dentry, &newattrs); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&meta_dentry->d_inode->i_mutex); ++#else ++ up(&meta_dentry->d_inode->i_sem); ++#endif ++ if(err || meta_dentry->d_inode->i_size != 0) { ++ printk(KERN_CRIT "mini_fo: meta_sync_r_list: \ ++ ERROR truncating meta file.\n"); ++ goto out_err_close; ++ } ++ } ++ ++ /* open META-file for writing */ ++ meta_file = dentry_open(meta_dentry, meta_mnt, 0x1); ++ if(!meta_file || IS_ERR(meta_file)) { ++ printk(KERN_CRIT "mini_fo: meta_sync_r_list: \ ++ ERROR opening meta file.\n"); ++ /* we don't mntget so we dont't mntput (for now) ++ * mntput(meta_mnt); ++ */ ++ dput(meta_dentry); ++ err = -1; ++ goto out; ++ } ++ ++ /* check if fs supports writing */ ++ if(!meta_file->f_op->write) { ++ printk(KERN_CRIT "mini_fo: meta_sync_r_list: \ ++ ERROR, fs does not support writing.\n"); ++ goto out_err_close; ++ } ++ ++ meta_file->f_pos = meta_dentry->d_inode->i_size; /* append */ ++ old_fs = get_fs(); ++ set_fs(KERNEL_DS); ++ ++ /* here we go... */ ++ list_for_each(tmp, &inode_info->renamed_list) { ++ ren_entry = list_entry(tmp, struct renamed_entry, list); ++ /* size: ++ * 2 for "R ", old_len+new_len for names, 1 blank+1 \n */ ++ buf_len = ren_entry->old_len + ren_entry->new_len + 4; ++ buf = (char *) kmalloc(buf_len, GFP_KERNEL); ++ if (!buf) { ++ printk(KERN_CRIT "mini_fo: meta_sync_r_list: \ ++ out of mem.\n"); ++ return -ENOMEM; ++ } ++ buf[0] = 'R'; ++ buf[1] = ' '; ++ strncpy(buf + 2, ren_entry->old_name, ren_entry->old_len); ++ buf[ren_entry->old_len + 2] = ' '; ++ strncpy(buf + ren_entry->old_len + 3, ++ ren_entry->new_name, ren_entry->new_len); ++ buf[buf_len - 1] = '\n'; ++ bytes = meta_file->f_op->write(meta_file, buf, ++ buf_len, &meta_file->f_pos); ++ if(bytes != buf_len) { ++ printk(KERN_CRIT "mini_fo: meta_sync_r_list: \ ++ ERROR writing.\n"); ++ err |= -1; ++ } ++ kfree(buf); ++ } ++ set_fs(old_fs); ++ ++ out_err_close: ++ fput(meta_file); ++ out: ++ return err; ++} ++ ++int meta_check_d_entry(dentry_t *dentry, const char *name, int len) ++{ ++ if(!dentry || !dentry->d_inode) ++ printk(KERN_CRIT "mini_fo: meta_check_d_dentry: \ ++ invalid dentry passed.\n"); ++ return __meta_check_d_entry(dentry->d_inode, name, len); ++} ++ ++int __meta_check_d_entry(inode_t *inode, const char *name, int len) ++{ ++ struct list_head *tmp; ++ struct deleted_entry *del_entry; ++ struct mini_fo_inode_info *inode_info; ++ ++ if(!inode || !itopd(inode)) ++ printk(KERN_CRIT "mini_fo: __meta_check_d_dentry: \ ++ invalid inode passed.\n"); ++ ++ inode_info = itopd(inode); ++ ++ if(inode_info->deleted_list_size <= 0) ++ return 0; ++ ++ list_for_each(tmp, &inode_info->deleted_list) { ++ del_entry = list_entry(tmp, struct deleted_entry, list); ++ if(del_entry->len != len) ++ continue; ++ ++ if(!strncmp(del_entry->name, name, len)) ++ return 1; ++ } ++ return 0; ++} ++ ++/* ++ * check if file has been renamed and return path to orig. base dir. ++ * Implements no error return values so far, what of course sucks. ++ * String is null terminated.' ++ */ ++char* meta_check_r_entry(dentry_t *dentry, const char *name, int len) ++{ ++ if(!dentry || !dentry->d_inode) { ++ printk(KERN_CRIT "mini_fo: meta_check_r_dentry: \ ++ invalid dentry passed.\n"); ++ return NULL; ++ } ++ return __meta_check_r_entry(dentry->d_inode, name, len); ++} ++ ++char* __meta_check_r_entry(inode_t *inode, const char *name, int len) ++{ ++ struct list_head *tmp; ++ struct renamed_entry *ren_entry; ++ struct mini_fo_inode_info *inode_info; ++ char *old_path; ++ ++ if(!inode || !itopd(inode)) { ++ printk(KERN_CRIT "mini_fo: meta_check_r_dentry: \ ++ invalid inode passed.\n"); ++ return NULL; ++ } ++ inode_info = itopd(inode); ++ ++ if(inode_info->renamed_list_size <= 0) ++ return NULL; ++ ++ list_for_each(tmp, &inode_info->renamed_list) { ++ ren_entry = list_entry(tmp, struct renamed_entry, list); ++ if(ren_entry->new_len != len) ++ continue; ++ ++ if(!strncmp(ren_entry->new_name, name, len)) { ++ old_path = (char *) ++ kmalloc(ren_entry->old_len+1, GFP_KERNEL); ++ strncpy(old_path, ++ ren_entry->old_name, ++ ren_entry->old_len); ++ old_path[ren_entry->old_len]='\0'; ++ return old_path; ++ } ++ } ++ return NULL; ++} ++ ++/* ++ * This version only checks if entry exists and return: ++ * 1 if exists, ++ * 0 if not, ++ * -1 if error. ++ */ ++int meta_is_r_entry(dentry_t *dentry, const char *name, int len) ++{ ++ if(!dentry || !dentry->d_inode) { ++ printk(KERN_CRIT "mini_fo: meta_check_r_dentry [2]: \ ++ invalid dentry passed.\n"); ++ return -1; ++ } ++ return __meta_is_r_entry(dentry->d_inode, name, len); ++} ++ ++int __meta_is_r_entry(inode_t *inode, const char *name, int len) ++{ ++ struct list_head *tmp; ++ struct renamed_entry *ren_entry; ++ struct mini_fo_inode_info *inode_info; ++ ++ if(!inode || !itopd(inode)) { ++ printk(KERN_CRIT "mini_fo: meta_check_r_dentry [2]: \ ++ invalid inode passed.\n"); ++ return -1; ++ } ++ inode_info = itopd(inode); ++ ++ if(inode_info->renamed_list_size <= 0) ++ return -1; ++ ++ list_for_each(tmp, &inode_info->renamed_list) { ++ ren_entry = list_entry(tmp, struct renamed_entry, list); ++ if(ren_entry->new_len != len) ++ continue; ++ ++ if(!strncmp(ren_entry->new_name, name, len)) ++ return 1; ++ } ++ return 0; ++} ++ +--- /dev/null ++++ b/fs/mini_fo/mini_fo.h +@@ -0,0 +1,510 @@ ++/* ++ * Copyright (c) 1997-2003 Erez Zadok ++ * Copyright (c) 2001-2003 Stony Brook University ++ * ++ * For specific licensing information, see the COPYING file distributed with ++ * this package, or get one from ftp://ftp.filesystems.org/pub/fist/COPYING. ++ * ++ * This Copyright notice must be kept intact and distributed with all ++ * fistgen sources INCLUDING sources generated by fistgen. ++ */ ++/* ++ * Copyright (C) 2004, 2005 Markus Klotzbuecher ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ */ ++ ++/* ++ * $Id$ ++ */ ++ ++#ifndef __MINI_FO_H_ ++#define __MINI_FO_H_ ++ ++#ifdef __KERNEL__ ++ ++/* META stuff */ ++#define META_FILENAME "META_dAfFgHE39ktF3HD2sr" ++ ++/* use xattrs? */ ++#define XATTR ++ ++/* File attributes that when changed, result in a file beeing copied to storage */ ++#define COPY_FLAGS ATTR_MODE | ATTR_UID | ATTR_GID | ATTR_SIZE ++ ++/* ++ * mini_fo filestates ++ */ ++#define MODIFIED 1 ++#define UNMODIFIED 2 ++#define CREATED 3 ++#define DEL_REWRITTEN 4 ++#define DELETED 5 ++#define NON_EXISTANT 6 ++ ++/* fist file systems superblock magic */ ++# define MINI_FO_SUPER_MAGIC 0xf15f ++ ++/* ++ * STRUCTURES: ++ */ ++ ++/* mini_fo inode data in memory */ ++struct mini_fo_inode_info { ++ inode_t *wii_inode; ++ inode_t *wii_inode2; /* pointer to storage inode */ ++ ++ /* META-data lists */ ++ /* deleted list, ex wol */ ++ struct list_head deleted_list; ++ int deleted_list_size; ++ ++ /* renamed list */ ++ struct list_head renamed_list; ++ int renamed_list_size; ++ ++ /* add other lists here ... */ ++}; ++ ++/* mini_fo dentry data in memory */ ++struct mini_fo_dentry_info { ++ dentry_t *wdi_dentry; ++ dentry_t *wdi_dentry2; /* pointer to storage dentry */ ++ unsigned int state; /* state of the mini_fo dentry */ ++}; ++ ++ ++/* mini_fo super-block data in memory */ ++struct mini_fo_sb_info { ++ super_block_t *wsi_sb, *wsi_sb2; /* mk: might point to the same sb */ ++ struct vfsmount *hidden_mnt, *hidden_mnt2; ++ dentry_t *base_dir_dentry; ++ dentry_t *storage_dir_dentry; ++ ; ++}; ++ ++/* readdir_data, readdir helper struct */ ++struct readdir_data { ++ struct list_head ndl_list; /* linked list head ptr */ ++ int ndl_size; /* list size */ ++ int sto_done; /* flag to show that the storage dir entries have ++ * all been read an now follow base entries */ ++}; ++ ++/* file private data. */ ++struct mini_fo_file_info { ++ struct file *wfi_file; ++ struct file *wfi_file2; /* pointer to storage file */ ++ struct readdir_data rd; ++}; ++ ++/* struct ndl_entry */ ++struct ndl_entry { ++ struct list_head list; ++ char *name; ++ int len; ++}; ++ ++/******************************** ++ * META-data structures ++ ********************************/ ++ ++/* deleted entry */ ++struct deleted_entry { ++ struct list_head list; ++ char *name; ++ int len; ++}; ++ ++/* renamed entry */ ++struct renamed_entry { ++ struct list_head list; ++ char *old_name; /* old directory with full path */ ++ int old_len; /* length of above string */ ++ char *new_name; /* new directory name */ ++ int new_len; /* length of above string */ ++}; ++ ++/* attr_change entry */ ++struct attr_change_entry { ++ struct list_head list; ++ char *name; ++ int len; ++}; ++ ++/* link entry */ ++struct link_entry { ++ struct list_head list; ++ int links_moved; ++ int inum_base; ++ int inum_sto; ++ char *weird_name; ++ int weird_name_len; ++}; ++ ++ ++/* Some other stuff required for mini_fo_filldir64, copied from ++ * fs/readdir.c ++ */ ++ ++#define ROUND_UP64(x) (((x)+sizeof(u64)-1) & ~(sizeof(u64)-1)) ++#define NAME_OFFSET(de) ((int) ((de)->d_name - (char *) (de))) ++ ++ ++struct linux_dirent64 { ++ u64 d_ino; ++ s64 d_off; ++ unsigned short d_reclen; ++ unsigned char d_type; ++ char d_name[0]; ++}; ++ ++ ++struct getdents_callback64 { ++ struct linux_dirent64 * current_dir; ++ struct linux_dirent64 * previous; ++ int count; ++ int error; ++}; ++ ++struct linux_dirent { ++ unsigned long d_ino; ++ unsigned long d_off; ++ unsigned short d_reclen; ++ char d_name[1]; ++}; ++ ++struct getdents_callback { ++ struct linux_dirent * current_dir; ++ struct linux_dirent * previous; ++ int count; ++ int error; ++}; ++ ++ ++/* ++ * MACROS: ++ */ ++ ++/* file TO private_data */ ++# define ftopd(file) ((struct mini_fo_file_info *)((file)->private_data)) ++# define __ftopd(file) ((file)->private_data) ++/* file TO hidden_file */ ++# define ftohf(file) ((ftopd(file))->wfi_file) ++# define ftohf2(file) ((ftopd(file))->wfi_file2) ++ ++/* inode TO private_data */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18) ++# define itopd(ino) ((struct mini_fo_inode_info *)(ino)->i_private) ++# define __itopd(ino) ((ino)->i_private) ++#else ++# define itopd(ino) ((struct mini_fo_inode_info *)(ino)->u.generic_ip) ++# define __itopd(ino) ((ino)->u.generic_ip) ++#endif ++/* inode TO hidden_inode */ ++# define itohi(ino) (itopd(ino)->wii_inode) ++# define itohi2(ino) (itopd(ino)->wii_inode2) ++ ++/* superblock TO private_data */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++# define stopd(super) ((struct mini_fo_sb_info *)(super)->s_fs_info) ++# define __stopd(super) ((super)->s_fs_info) ++#else ++# define stopd(super) ((struct mini_fo_sb_info *)(super)->u.generic_sbp) ++# define __stopd(super) ((super)->u.generic_sbp) ++#endif ++ ++/* unused? # define vfs2priv stopd */ ++/* superblock TO hidden_superblock */ ++ ++# define stohs(super) (stopd(super)->wsi_sb) ++# define stohs2(super) (stopd(super)->wsi_sb2) ++ ++/* dentry TO private_data */ ++# define dtopd(dentry) ((struct mini_fo_dentry_info *)(dentry)->d_fsdata) ++# define __dtopd(dentry) ((dentry)->d_fsdata) ++/* dentry TO hidden_dentry */ ++# define dtohd(dent) (dtopd(dent)->wdi_dentry) ++# define dtohd2(dent) (dtopd(dent)->wdi_dentry2) ++ ++/* dentry to state */ ++# define dtost(dent) (dtopd(dent)->state) ++# define sbt(sb) ((sb)->s_type->name) ++ ++#define IS_WRITE_FLAG(flag) (flag & (O_RDWR | O_WRONLY | O_APPEND)) ++#define IS_COPY_FLAG(flag) (flag & (COPY_FLAGS)) ++ ++/* macros to simplify non-SCA code */ ++# define MALLOC_PAGE_POINTERS(hidden_pages, num_hidden_pages) ++# define MALLOC_PAGEDATA_POINTERS(hidden_pages_data, num_hidden_pages) ++# define FREE_PAGE_POINTERS(hidden_pages, num) ++# define FREE_PAGEDATA_POINTERS(hidden_pages_data, num) ++# define FOR_EACH_PAGE ++# define CURRENT_HIDDEN_PAGE hidden_page ++# define CURRENT_HIDDEN_PAGEDATA hidden_page_data ++# define CURRENT_HIDDEN_PAGEINDEX page->index ++ ++/* ++ * EXTERNALS: ++ */ ++extern struct file_operations mini_fo_main_fops; ++extern struct file_operations mini_fo_dir_fops; ++extern struct inode_operations mini_fo_main_iops; ++extern struct inode_operations mini_fo_dir_iops; ++extern struct inode_operations mini_fo_symlink_iops; ++extern struct super_operations mini_fo_sops; ++extern struct dentry_operations mini_fo_dops; ++extern struct vm_operations_struct mini_fo_shared_vmops; ++extern struct vm_operations_struct mini_fo_private_vmops; ++extern struct address_space_operations mini_fo_aops; ++ ++#if 0 /* unused by mini_fo */ ++extern int mini_fo_interpose(dentry_t *hidden_dentry, dentry_t *this_dentry, super_block_t *sb, int flag); ++#if defined(FIST_FILTER_DATA) || defined(FIST_FILTER_SCA) ++extern page_t *mini_fo_get1page(file_t *file, int index); ++extern int mini_fo_fill_zeros(file_t *file, page_t *page, unsigned from); ++# endif /* FIST_FILTER_DATA || FIST_FILTER_SCA */ ++ ++ ++# define mini_fo_hidden_dentry(d) __mini_fo_hidden_dentry(__FILE__,__FUNCTION__,__LINE__,(d)) ++# define mini_fo_hidden_sto_dentry(d) __mini_fo_hidden_sto_dentry(__FILE__,__FUNCTION__,__LINE__,(d)) ++ ++extern dentry_t *__mini_fo_hidden_dentry(char *file, char *func, int line, dentry_t *this_dentry); ++extern dentry_t *__mini_fo_hidden_sto_dentry(char *file, char *func, int line, dentry_t *this_dentry); ++ ++extern int mini_fo_read_file(const char *filename, void *buf, int len); ++extern int mini_fo_write_file(const char *filename, void *buf, int len); ++extern dentry_t *fist_lookup(dentry_t *dir, const char *name, vnode_t **out, uid_t uid, gid_t gid); ++#endif /* unused by mini_fo */ ++ ++/* state transition functions */ ++extern int nondir_unmod_to_mod(dentry_t *dentry, int cp_flag); ++extern int nondir_del_rew_to_del(dentry_t *dentry); ++extern int nondir_creat_to_del(dentry_t *dentry); ++extern int nondir_mod_to_del(dentry_t *dentry); ++extern int nondir_unmod_to_del(dentry_t *dentry); ++ ++extern int dir_unmod_to_mod(dentry_t *dentry); ++ ++/* rename specials */ ++extern int rename_directory(inode_t *old_dir, dentry_t *old_dentry, inode_t *new_dir, dentry_t *new_dentry); ++extern int rename_nondir(inode_t *old_dir, dentry_t *old_dentry, inode_t *new_dir, dentry_t *new_dentry); ++ ++/* misc stuff */ ++extern int mini_fo_tri_interpose(dentry_t *hidden_dentry, ++ dentry_t *hidden_sto_dentry, ++ dentry_t *dentry, ++ super_block_t *sb, int flag); ++ ++extern int mini_fo_cp_cont(dentry_t *tgt_dentry, struct vfsmount *tgt_mnt, ++ dentry_t *src_dentry, struct vfsmount *src_mnt); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++extern int mini_fo_create(inode_t *dir, dentry_t *dentry, int mode, struct nameidata *nd); ++ ++extern int create_sto_nod(dentry_t *dentry, int mode, dev_t dev); ++extern int create_sto_reg_file(dentry_t *dentry, int mode, struct nameidata *nd); ++#else ++extern int mini_fo_create(inode_t *dir, dentry_t *dentry, int mode); ++ ++extern int create_sto_nod(dentry_t *dentry, int mode, int dev); ++extern int create_sto_reg_file(dentry_t *dentry, int mode); ++#endif ++ ++extern int create_sto_dir(dentry_t *dentry, int mode); ++ ++extern int exists_in_storage(dentry_t *dentry); ++extern int is_mini_fo_existant(dentry_t *dentry); ++extern int get_neg_sto_dentry(dentry_t *dentry); ++extern int build_sto_structure(dentry_t *dir, dentry_t *dentry); ++extern int get_mini_fo_bpath(dentry_t *dentry, char **bpath, int *bpath_len); ++extern dentry_t *bpath_walk(super_block_t *sb, char *bpath); ++extern int bpath_put(dentry_t *dentry); ++ ++/* check_mini_fo types functions */ ++extern int check_mini_fo_dentry(dentry_t *dentry); ++extern int check_mini_fo_file(file_t *file); ++extern int check_mini_fo_inode(inode_t *inode); ++ ++/* General meta functions, can be called from outside of meta.c */ ++extern int meta_build_lists(dentry_t *dentry); ++extern int meta_put_lists(dentry_t *dentry); ++extern int __meta_put_lists(inode_t *inode); ++ ++extern int meta_add_d_entry(dentry_t *dentry, const char *name, int len); ++extern int meta_add_r_entry(dentry_t *dentry, ++ const char *old_name, int old_len, ++ const char *new_name, int new_len); ++ ++extern int meta_remove_r_entry(dentry_t *dentry, const char *name, int len); ++ ++extern int meta_check_d_entry(dentry_t *dentry, const char *name, int len); ++extern int __meta_check_d_entry(inode_t *inode, const char *name, int len); ++ ++extern char* meta_check_r_entry(dentry_t *dentry, const char *name, int len); ++extern char* __meta_check_r_entry(inode_t *inode, const char *name, int len); ++extern int meta_is_r_entry(dentry_t *dentry, const char *name, int len); ++extern int __meta_is_r_entry(inode_t *inode, const char *name, int len); ++ ++/* Specific meta functions, should be called only inside meta.c */ ++extern int __meta_put_d_list(inode_t *inode); ++extern int __meta_put_r_list(inode_t *inode); ++ ++extern int meta_list_add_d_entry(dentry_t *dentry, ++ const char *name, int len); ++extern int meta_list_add_r_entry(dentry_t *dentry, ++ const char *old_name, int old_len, ++ const char *new_name, int new_len); ++ ++extern int meta_list_remove_r_entry(dentry_t *dentry, ++ const char *name, int len); ++ ++extern int __meta_list_remove_r_entry(inode_t *inode, ++ const char *name, int len); ++ ++extern int meta_write_d_entry(dentry_t *dentry, const char *name, int len); ++extern int meta_write_r_entry(dentry_t *dentry, ++ const char *old_name, int old_len, ++ const char *new_name, int new_len); ++ ++extern int meta_sync_lists(dentry_t *dentry); ++extern int meta_sync_d_list(dentry_t *dentry, int app_flag); ++extern int meta_sync_r_list(dentry_t *dentry, int app_flag); ++ ++/* ndl stuff */ ++extern int ndl_add_entry(struct readdir_data *rd, const char *name, int len); ++extern void ndl_put_list(struct readdir_data *rd); ++extern int ndl_check_entry(struct readdir_data *rd, ++ const char *name, int len); ++ ++ ++# define copy_inode_size(dst, src) \ ++ dst->i_size = src->i_size; \ ++ dst->i_blocks = src->i_blocks; ++ ++static inline void ++fist_copy_attr_atime(inode_t *dest, const inode_t *src) ++{ ++ ASSERT(dest != NULL); ++ ASSERT(src != NULL); ++ dest->i_atime = src->i_atime; ++} ++static inline void ++fist_copy_attr_times(inode_t *dest, const inode_t *src) ++{ ++ ASSERT(dest != NULL); ++ ASSERT(src != NULL); ++ dest->i_atime = src->i_atime; ++ dest->i_mtime = src->i_mtime; ++ dest->i_ctime = src->i_ctime; ++} ++static inline void ++fist_copy_attr_timesizes(inode_t *dest, const inode_t *src) ++{ ++ ASSERT(dest != NULL); ++ ASSERT(src != NULL); ++ dest->i_atime = src->i_atime; ++ dest->i_mtime = src->i_mtime; ++ dest->i_ctime = src->i_ctime; ++ copy_inode_size(dest, src); ++} ++static inline void ++fist_copy_attr_all(inode_t *dest, const inode_t *src) ++{ ++ ASSERT(dest != NULL); ++ ASSERT(src != NULL); ++ dest->i_mode = src->i_mode; ++ dest->i_nlink = src->i_nlink; ++ dest->i_uid = src->i_uid; ++ dest->i_gid = src->i_gid; ++ dest->i_rdev = src->i_rdev; ++ dest->i_atime = src->i_atime; ++ dest->i_mtime = src->i_mtime; ++ dest->i_ctime = src->i_ctime; ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) ++ dest->i_blksize = src->i_blksize; ++#endif ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,12) ++ dest->i_blkbits = src->i_blkbits; ++# endif /* linux 2.4.12 and newer */ ++ copy_inode_size(dest, src); ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) ++ dest->i_attr_flags = src->i_attr_flags; ++#else ++ dest->i_flags = src->i_flags; ++#endif ++} ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++/* copied from linux/fs.h */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++static inline void double_lock(struct dentry *d1, struct dentry *d2) ++{ ++ struct mutex *m1 = &d1->d_inode->i_mutex; ++ struct mutex *m2 = &d2->d_inode->i_mutex; ++ if (m1 != m2) { ++ if ((unsigned long) m1 < (unsigned long) m2) { ++ struct mutex *tmp = m2; ++ m2 = m1; m1 = tmp; ++ } ++ mutex_lock(m1); ++ } ++ mutex_lock(m2); ++} ++ ++static inline void double_unlock(struct dentry *d1, struct dentry *d2) ++{ ++ struct mutex *m1 = &d1->d_inode->i_mutex; ++ struct mutex *m2 = &d2->d_inode->i_mutex; ++ mutex_unlock(m1); ++ if (m1 != m2) ++ mutex_unlock(m2); ++ dput(d1); ++ dput(d2); ++} ++ ++#else ++static inline void double_down(struct semaphore *s1, struct semaphore *s2) ++{ ++ if (s1 != s2) { ++ if ((unsigned long) s1 < (unsigned long) s2) { ++ struct semaphore *tmp = s2; ++ s2 = s1; s1 = tmp; ++ } ++ down(s1); ++ } ++ down(s2); ++} ++ ++static inline void double_up(struct semaphore *s1, struct semaphore *s2) ++{ ++ up(s1); ++ if (s1 != s2) ++ up(s2); ++} ++ ++static inline void double_lock(struct dentry *d1, struct dentry *d2) ++{ ++ double_down(&d1->d_inode->i_sem, &d2->d_inode->i_sem); ++} ++ ++static inline void double_unlock(struct dentry *d1, struct dentry *d2) ++{ ++ double_up(&d1->d_inode->i_sem,&d2->d_inode->i_sem); ++ dput(d1); ++ dput(d2); ++} ++#endif /* if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) */ ++#endif /* if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) */ ++#endif /* __KERNEL__ */ ++ ++/* ++ * Definitions for user and kernel code ++ */ ++ ++/* ioctls */ ++ ++#endif /* not __MINI_FO_H_ */ +--- /dev/null ++++ b/fs/mini_fo/mini_fo-merge +@@ -0,0 +1,180 @@ ++#!/bin/bash ++# ++# Copyright (C) 2005 Markus Klotzbuecher ++# This program is free software; you can redistribute it and/or ++# modify it under the terms of the GNU General Public License ++# as published by the Free Software Foundation; either version ++# 2 of the License, or (at your option) any later version. ++# ++ ++BASE= ++STO= ++HELP= ++DRYRUN= ++VERBOSE= ++TMP="/tmp/" ++META_NAME="META_dAfFgHE39ktF3HD2sr" ++SKIP_DEL_LIST="skip-delete-list.mini_fo-merge" ++ ++COMMAND= ++exec_command() ++{ ++ if [ x$DRYRUN == "xset" ]; then ++ echo " would run: $COMMAND" ++ elif ! [ x$DRYRUN == "xset" ]; then ++ if [ x$VERBOSE == "xset" ]; then ++ echo " running: $COMMAND" ++ fi ++ eval $COMMAND ++ fi ++} ++ ++usage() ++{ ++cat < -s ++Version 0.1 ++ ++This script merges the contents of a mini_fo storage file system back ++to the base file system. ++ ++!!! Warning: This will modify the base filesystem and can destroy data ++ if used wrongly. ++ ++Options: ++ -b ++ the directory of the base file system. ++ ++ -s ++ the directory of the storage file system. ++ ++ -d dry run, will not change anything and print the commands that ++ would be executed. ++ ++ -t tmp dir for storing temporary file. default: $TMP ++ ++ -v show what operations are performed. ++ ++ -h displays this message. ++ ++EOF ++} ++ ++# parse parameters ++while getopts hdvt:b:s: OPTS ++ do ++ case $OPTS in ++ h) HELP="set";; ++ d) DRYRUN="set";; ++ v) VERBOSE="set";; ++ b) BASE="$OPTARG";; ++ s) STO="$OPTARG";; ++ t) TMP="$OPTARG";; ++ ?) usage ++ exit 1;; ++ esac ++done ++ ++if [ "x$HELP" == "xset" ]; then ++ usage ++ exit -1 ++fi ++ ++if ! [ -d "$BASE" ] || ! [ -d "$STO" ]; then ++ echo -e "$0:\n Error, -s and/or -b argument missing. type $0 -h for help." ++ exit -1; ++fi ++ ++# get full paths ++pushd $STO; STO=`pwd`; popd ++pushd $BASE; BASE=`pwd`; popd ++TMP=${TMP%/} ++ ++ ++cat< /dev/null ++find . -name $META_NAME -type f -print0 | xargs -0 -e grep -e '^R ' | tr -s ':R' ' ' | while read ENTRY; do ++ echo "entry: $ENTRY" ++ META_FILE=`echo $ENTRY | cut -d ' ' -f 1` ++ OLD_B_DIR=`echo $ENTRY | cut -d ' ' -f 2 | sed -e 's/\///'` ++ NEW_NAME=`echo $ENTRY | cut -d ' ' -f 3` ++ NEW_B_DIR=`echo $META_FILE | sed -e "s/$META_NAME/$NEW_NAME/" | sed -e 's/^\.\///'` ++ echo "META_FILE: $META_FILE" ++ echo "OLD_B_DIR: $OLD_B_DIR" ++ echo "NEW_NAME: $NEW_NAME" ++ echo "NEW_B_DIR: $NEW_B_DIR" ++ ++ pushd $BASE &> /dev/null ++ # remove an existing dir in storage ++ COMMAND="rm -rf $NEW_B_DIR"; exec_command ++ COMMAND="cp -R $OLD_B_DIR $NEW_B_DIR"; exec_command ++ echo "" ++ popd &> /dev/null ++ ++ # remember this dir to exclude it from deleting later ++ echo $NEW_B_DIR >> $TMP/$SKIP_DEL_LIST ++done ++ ++# delete all whiteouted files from base ++echo -e "\nDeleting whiteout'ed files from base file system..." ++find . -name $META_NAME -type f -print0 | xargs -0 -e grep -e '^D ' | sed -e 's/:D//' | while read ENTRY; do ++ META_FILE=`echo $ENTRY | cut -d ' ' -f 1` ++ DEL_NAME=`echo $ENTRY | cut -d ' ' -f 2` ++ DEL_FILE=`echo $META_FILE | sed -e "s/$META_NAME/$DEL_NAME/" | sed -e 's/^\.\///'` ++ grep -x $DEL_FILE $TMP/$SKIP_DEL_LIST &> /dev/null ++ if [ $? -ne 0 ]; then ++ pushd $BASE &> /dev/null ++ COMMAND="rm -rf $DEL_FILE"; exec_command ++ popd &> /dev/null ++ else ++ echo " excluding: $DEL_FILE as in skip-del-list." ++ fi ++done ++ ++# create all dirs and update permissions ++echo -e "\nSetting up directory structures in base file system..." ++find . -type d | sed -e 's/^\.\///' | while read DIR; do ++ PERMS=`stat -c %a $DIR` ++ DIR_UID=`stat -c %u $DIR` ++ DIR_GID=`stat -c %g $DIR` ++ pushd $BASE &> /dev/null ++ if ! [ -d $DIR ]; then ++ COMMAND="mkdir -p $DIR"; exec_command ++ fi ++ COMMAND="chmod $PERMS $DIR"; exec_command ++ COMMAND="chown $DIR_UID:$DIR_GID $DIR"; exec_command ++ popd &> /dev/null ++done ++ ++# merge all non-directory files ++echo -e "\nMerging all non-directory files...." ++for i in b c p f l s; do ++ find . -type $i | sed -e 's/^\.\///' | grep -v "$META_NAME" | while read FILE; do ++ pushd $BASE #&> /dev/null ++ COMMAND="cp -df $STO/$FILE $BASE/$FILE"; exec_command ++ popd &> /dev/null ++ done ++done ++popd &> /dev/null ++ ++#rm $TMP/$SKIP_DEL_LIST ++ ++echo "Done!" +--- /dev/null ++++ b/fs/mini_fo/mini_fo-overlay +@@ -0,0 +1,130 @@ ++#!/bin/bash ++# ++# Copyright (C) 2005 Markus Klotzbuecher ++# This program is free software; you can redistribute it and/or ++# modify it under the terms of the GNU General Public License ++# as published by the Free Software Foundation; either version ++# 2 of the License, or (at your option) any later version. ++# ++ ++HELP= ++SUFF= ++MNTP= ++MNT_DIR="/mnt" ++STO= ++STO_DIR="/tmp" ++BASE= ++ ++usage() ++{ ++cat <" in $STO_DIR, ++and mount point "mini_fo-" in $MNT_DIR. ++ ++Options: ++ -s ++ add given suffix to storage directory and the mount ++ point. This is usefull for overlaying one base directory ++ several times and avoiding conflicts with storage directory ++ names and mount points. ++ ++ -d ++ change the directory in which the storage directory will be ++ created (default is currently "$STO_DIR". ++ ++ -m ++ use an alternative directory to create the mini_fo ++ mountpoint (default is currently "$MNT_DIR". ++ ++ -h displays this message. ++ ++EOF ++exit 1; ++} ++ ++while getopts hm:s:d: OPTS ++ do ++ case $OPTS in ++ s) SUFF="$OPTARG";; ++ d) STO_DIR="$OPTARG";; ++ m) MNT_DIR="$OPTARG";; ++ h) HELP="set";; ++ ?) usage ++ exit 1;; ++ esac ++done ++shift $(($OPTIND - 1)) ++ ++BASE="$1" ++ ++if [ "x$HELP" == "xset" ]; then ++ usage ++ exit -1 ++fi ++ ++# fix suffix ++if [ "x$SUFF" != "x" ]; then ++ SUFF="-$SUFF" ++fi ++ ++# kill trailing slashes ++MNT_DIR=${MNT_DIR%/} ++STO_DIR=${STO_DIR%/} ++BASE=${BASE%/} ++ ++ ++if ! [ -d "$BASE" ]; then ++ echo "invalid base dir $BASE, run $0 -h for help." ++ exit -1 ++fi ++ ++# check opts ++if ! [ -d "$MNT_DIR" ]; then ++ echo "invalid mount dir $MNT_DIR, run $0 -h for help." ++ exit -1 ++fi ++ ++if ! [ -d "$STO_DIR" ]; then ++ echo "invalid sto_dir_dir $STO_DIR, run $0 -h for help." ++ exit -1 ++fi ++ ++MNTP="$MNT_DIR/mini_fo-`basename $BASE`$SUFF" ++STO="$STO_DIR/sto-`basename $BASE`$SUFF" ++ ++# create the mount point if it doesn't exist ++mkdir -p $MNTP ++if [ $? -ne 0 ]; then ++ echo "Error, failed to create mount point $MNTP" ++fi ++ ++mkdir -p $STO ++if [ $? -ne 0 ]; then ++ echo "Error, failed to create storage dir $STO" ++fi ++ ++# check if fs is already mounted ++mount | grep mini_fo | grep $MNTP &> /dev/null ++if [ $? -eq 0 ]; then ++ echo "Error, existing mini_fo mount at $MNTP." ++ exit -1 ++fi ++ ++mount | grep mini_fo | grep $STO &> /dev/null ++if [ $? -eq 0 ]; then ++ echo "Error, $STO seems to be used already." ++ exit -1 ++fi ++ ++# mount ++mount -t mini_fo -o base=$BASE,sto=$STO $BASE $MNTP ++ ++if [ $? -ne 0 ]; then ++ echo "Error, mounting failed, maybe no permisson to mount?" ++fi +--- /dev/null ++++ b/fs/mini_fo/mmap.c +@@ -0,0 +1,637 @@ ++/* ++ * Copyright (c) 1997-2003 Erez Zadok ++ * Copyright (c) 2001-2003 Stony Brook University ++ * ++ * For specific licensing information, see the COPYING file distributed with ++ * this package, or get one from ftp://ftp.filesystems.org/pub/fist/COPYING. ++ * ++ * This Copyright notice must be kept intact and distributed with all ++ * fistgen sources INCLUDING sources generated by fistgen. ++ */ ++/* ++ * Copyright (C) 2004, 2005 Markus Klotzbuecher ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ */ ++ ++/* ++ * $Id$ ++ */ ++ ++#ifdef HAVE_CONFIG_H ++# include ++#endif /* HAVE_CONFIG_H */ ++ ++#include "fist.h" ++#include "mini_fo.h" ++ ++ ++#ifdef FIST_COUNT_WRITES ++/* for counting writes in the middle vs. regular writes */ ++unsigned long count_writes = 0, count_writes_middle = 0; ++#endif /* FIST_COUNT_WRITES */ ++ ++/* forward declaration of commit write and prepare write */ ++STATIC int mini_fo_commit_write(file_t *file, page_t *page, unsigned from, unsigned to); ++STATIC int mini_fo_prepare_write(file_t *file, page_t *page, unsigned from, unsigned to); ++ ++ ++/* ++ * Function for handling creation of holes when lseek-ing past the ++ * end of the file and then writing some data. ++ */ ++int ++mini_fo_fill_zeros(file_t* file, page_t *page, unsigned from) ++{ ++ int err = 0; ++ dentry_t *dentry = file->f_dentry; ++ inode_t *inode = dentry->d_inode; ++ page_t *tmp_page; ++ int index; ++ ++ print_entry_location(); ++ ++ for (index = inode->i_size >> PAGE_CACHE_SHIFT; index < page->index; index++) { ++ tmp_page = mini_fo_get1page(file, index); ++ if (IS_ERR(tmp_page)) { ++ err = PTR_ERR(tmp_page); ++ goto out; ++ } ++ ++ /* ++ * zero out rest of the contents of the page between the appropriate ++ * offsets. ++ */ ++ memset((char*)page_address(tmp_page) + (inode->i_size & ~PAGE_CACHE_MASK), 0, PAGE_CACHE_SIZE - (inode->i_size & ~PAGE_CACHE_MASK)); ++ ++ if (! (err = mini_fo_prepare_write(file, tmp_page, 0, PAGE_CACHE_SIZE))) ++ err = mini_fo_commit_write(file, tmp_page, 0, PAGE_CACHE_SIZE); ++ ++ page_cache_release(tmp_page); ++ if (err < 0) ++ goto out; ++ if (current->need_resched) ++ schedule(); ++ } ++ ++ /* zero out appropriate parts of last page */ ++ ++ /* ++ * if the encoding type is block, then adjust the 'from' (where the ++ * zeroing will start) offset appropriately ++ */ ++ from = from & (~(FIST_ENCODING_BLOCKSIZE - 1)); ++ ++ if ((from - (inode->i_size & ~PAGE_CACHE_MASK)) > 0) { ++ ++ memset((char*)page_address(page) + (inode->i_size & ~PAGE_CACHE_MASK), 0, from - (inode->i_size & ~PAGE_CACHE_MASK)); ++ if (! (err = mini_fo_prepare_write(file, page, 0, PAGE_CACHE_SIZE))) ++ err = mini_fo_commit_write(file, page, 0, PAGE_CACHE_SIZE); ++ ++ if (err < 0) ++ goto out; ++ if (current->need_resched) ++ schedule(); ++ } ++ ++ out: ++ print_exit_status(err); ++ return err; ++} ++ ++ ++ ++STATIC int ++mini_fo_writepage(page_t *page) ++{ ++ int err = -EIO; ++ inode_t *inode; ++ inode_t *hidden_inode; ++ page_t *hidden_page; ++ char *kaddr, *hidden_kaddr; ++ ++ print_entry_location(); ++ ++ inode = page->mapping->host; ++ hidden_inode = itohi(inode); ++ ++ /* ++ * writepage is called when shared mmap'ed files need to write ++ * their pages, while prepare/commit_write are called from the ++ * non-paged write() interface. (However, in 2.3 the two interfaces ++ * share the same cache, while in 2.2 they didn't.) ++ * ++ * So we pretty much have to duplicate much of what commit_write does. ++ */ ++ ++ /* find lower page (returns a locked page) */ ++ hidden_page = grab_cache_page(hidden_inode->i_mapping, page->index); ++ if (!hidden_page) ++ goto out; ++ ++ /* get page address, and encode it */ ++ kaddr = (char *) kmap(page); ++ hidden_kaddr = (char*) kmap(hidden_page); ++ mini_fo_encode_block(kaddr, hidden_kaddr, PAGE_CACHE_SIZE, inode, inode->i_sb, page->index); ++ /* if encode_block could fail, then return error */ ++ kunmap(page); ++ kunmap(hidden_page); ++ ++ /* call lower writepage (expects locked page) */ ++ err = hidden_inode->i_mapping->a_ops->writepage(hidden_page); ++ ++ /* ++ * update mtime and ctime of lower level file system ++ * mini_fo' mtime and ctime are updated by generic_file_write ++ */ ++ hidden_inode->i_mtime = hidden_inode->i_ctime = CURRENT_TIME; ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,4,1) ++ UnlockPage(hidden_page); /* b/c grab_cache_page locked it */ ++# endif /* kernel older than 2.4.1 */ ++ page_cache_release(hidden_page); /* b/c grab_cache_page increased refcnt */ ++ ++ if (err) ++ ClearPageUptodate(page); ++ else ++ SetPageUptodate(page); ++ out: ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,1) ++ UnlockPage(page); ++# endif /* kernel 2.4.1 and newer */ ++ print_exit_status(err); ++ return err; ++} ++ ++ ++/* ++ * get one page from cache or lower f/s, return error otherwise. ++ * returns unlocked, up-to-date page (if ok), with increased refcnt. ++ */ ++page_t * ++mini_fo_get1page(file_t *file, int index) ++{ ++ page_t *page; ++ dentry_t *dentry; ++ inode_t *inode; ++ struct address_space *mapping; ++ int err; ++ ++ print_entry_location(); ++ ++ dentry = file->f_dentry; /* CPW: Moved below print_entry_location */ ++ inode = dentry->d_inode; ++ mapping = inode->i_mapping; ++ ++ fist_dprint(8, "%s: read page index %d pid %d\n", __FUNCTION__, index, current->pid); ++ if (index < 0) { ++ printk("%s BUG: index=%d\n", __FUNCTION__, index); ++ page = ERR_PTR(-EIO); ++ goto out; ++ } ++ page = read_cache_page(mapping, ++ index, ++ (filler_t *) mapping->a_ops->readpage, ++ (void *) file); ++ if (IS_ERR(page)) ++ goto out; ++ wait_on_page(page); ++ if (!Page_Uptodate(page)) { ++ lock_page(page); ++ err = mapping->a_ops->readpage(file, page); ++ if (err) { ++ page = ERR_PTR(err); ++ goto out; ++ } ++ wait_on_page(page); ++ if (!Page_Uptodate(page)) { ++ page = ERR_PTR(-EIO); ++ goto out; ++ } ++ } ++ ++ out: ++ print_exit_pointer(page); ++ return page; ++} ++ ++ ++/* ++ * get one page from cache or lower f/s, return error otherwise. ++ * similar to get1page, but doesn't guarantee that it will return ++ * an unlocked page. ++ */ ++page_t * ++mini_fo_get1page_cached(file_t *file, int index) ++{ ++ page_t *page; ++ dentry_t *dentry; ++ inode_t *inode; ++ struct address_space *mapping; ++ int err; ++ ++ print_entry_location(); ++ ++ dentry = file->f_dentry; /* CPW: Moved below print_entry_location */ ++ inode = dentry->d_inode; ++ mapping = inode->i_mapping; ++ ++ fist_dprint(8, "%s: read page index %d pid %d\n", __FUNCTION__, index, current->pid); ++ if (index < 0) { ++ printk("%s BUG: index=%d\n", __FUNCTION__, index); ++ page = ERR_PTR(-EIO); ++ goto out; ++ } ++ page = read_cache_page(mapping, ++ index, ++ (filler_t *) mapping->a_ops->readpage, ++ (void *) file); ++ if (IS_ERR(page)) ++ goto out; ++ ++ out: ++ print_exit_pointer(page); ++ return page; ++} ++ ++ ++/* ++ * readpage is called from generic_page_read and the fault handler. ++ * If your file system uses generic_page_read for the read op, it ++ * must implement readpage. ++ * ++ * Readpage expects a locked page, and must unlock it. ++ */ ++STATIC int ++mini_fo_do_readpage(file_t *file, page_t *page) ++{ ++ int err = -EIO; ++ dentry_t *dentry; ++ file_t *hidden_file = NULL; ++ dentry_t *hidden_dentry; ++ inode_t *inode; ++ inode_t *hidden_inode; ++ char *page_data; ++ page_t *hidden_page; ++ char *hidden_page_data; ++ int real_size; ++ ++ print_entry_location(); ++ ++ dentry = file->f_dentry; /* CPW: Moved below print_entry_location */ ++ if (ftopd(file) != NULL) ++ hidden_file = ftohf(file); ++ hidden_dentry = dtohd(dentry); ++ inode = dentry->d_inode; ++ hidden_inode = itohi(inode); ++ ++ fist_dprint(7, "%s: requesting page %d from file %s\n", __FUNCTION__, page->index, dentry->d_name.name); ++ ++ MALLOC_PAGE_POINTERS(hidden_pages, num_hidden_pages); ++ MALLOC_PAGEDATA_POINTERS(hidden_pages_data, num_hidden_pages); ++ FOR_EACH_PAGE ++ CURRENT_HIDDEN_PAGE = NULL; ++ ++ /* find lower page (returns a locked page) */ ++ FOR_EACH_PAGE { ++ fist_dprint(8, "%s: Current page index = %d\n", __FUNCTION__, CURRENT_HIDDEN_PAGEINDEX); ++ CURRENT_HIDDEN_PAGE = read_cache_page(hidden_inode->i_mapping, ++ CURRENT_HIDDEN_PAGEINDEX, ++ (filler_t *) hidden_inode->i_mapping->a_ops->readpage, ++ (void *) hidden_file); ++ if (IS_ERR(CURRENT_HIDDEN_PAGE)) { ++ err = PTR_ERR(CURRENT_HIDDEN_PAGE); ++ CURRENT_HIDDEN_PAGE = NULL; ++ goto out_release; ++ } ++ } ++ ++ /* ++ * wait for the page data to show up ++ * (signaled by readpage as unlocking the page) ++ */ ++ FOR_EACH_PAGE { ++ wait_on_page(CURRENT_HIDDEN_PAGE); ++ if (!Page_Uptodate(CURRENT_HIDDEN_PAGE)) { ++ /* ++ * call readpage() again if we returned from wait_on_page with a ++ * page that's not up-to-date; that can happen when a partial ++ * page has a few buffers which are ok, but not the whole ++ * page. ++ */ ++ lock_page(CURRENT_HIDDEN_PAGE); ++ err = hidden_inode->i_mapping->a_ops->readpage(hidden_file, ++ CURRENT_HIDDEN_PAGE); ++ if (err) { ++ CURRENT_HIDDEN_PAGE = NULL; ++ goto out_release; ++ } ++ wait_on_page(CURRENT_HIDDEN_PAGE); ++ if (!Page_Uptodate(CURRENT_HIDDEN_PAGE)) { ++ err = -EIO; ++ goto out_release; ++ } ++ } ++ } ++ ++ /* map pages, get their addresses */ ++ page_data = (char *) kmap(page); ++ FOR_EACH_PAGE ++ CURRENT_HIDDEN_PAGEDATA = (char *) kmap(CURRENT_HIDDEN_PAGE); ++ ++ /* if decode_block could fail, then return error */ ++ err = 0; ++ real_size = hidden_inode->i_size - (page->index << PAGE_CACHE_SHIFT); ++ if (real_size <= 0) ++ memset(page_data, 0, PAGE_CACHE_SIZE); ++ else if (real_size < PAGE_CACHE_SIZE) { ++ mini_fo_decode_block(hidden_page_data, page_data, real_size, inode, inode->i_sb, page->index); ++ memset(page_data + real_size, 0, PAGE_CACHE_SIZE - real_size); ++ } else ++ mini_fo_decode_block(hidden_page_data, page_data, PAGE_CACHE_SIZE, inode, inode->i_sb, page->index); ++ ++ FOR_EACH_PAGE ++ kunmap(CURRENT_HIDDEN_PAGE); ++ kunmap(page); ++ ++ out_release: ++ FOR_EACH_PAGE ++ if (CURRENT_HIDDEN_PAGE) ++ page_cache_release(CURRENT_HIDDEN_PAGE); /* undo read_cache_page */ ++ ++ FREE_PAGE_POINTERS(hidden_pages, num_hidden_pages); ++ FREE_PAGEDATA_POINTERS(hidden_pages_data, num_hidden_pages); ++ ++ out: ++ if (err == 0) ++ SetPageUptodate(page); ++ else ++ ClearPageUptodate(page); ++ ++ print_exit_status(err); ++ return err; ++} ++ ++ ++STATIC int ++mini_fo_readpage(file_t *file, page_t *page) ++{ ++ int err; ++ print_entry_location(); ++ ++ err = mini_fo_do_readpage(file, page); ++ ++ /* ++ * we have to unlock our page, b/c we _might_ have gotten a locked page. ++ * but we no longer have to wakeup on our page here, b/c UnlockPage does ++ * it ++ */ ++ UnlockPage(page); ++ ++ print_exit_status(err); ++ return err; ++} ++ ++ ++STATIC int ++mini_fo_prepare_write(file_t *file, page_t *page, unsigned from, unsigned to) ++{ ++ int err = 0; ++ ++ print_entry_location(); ++ ++ /* ++ * we call kmap(page) only here, and do the kunmap ++ * and the actual downcalls, including unlockpage and uncache ++ * in commit_write. ++ */ ++ kmap(page); ++ ++ /* fast path for whole page writes */ ++ if (from == 0 && to == PAGE_CACHE_SIZE) ++ goto out; ++ /* read the page to "revalidate" our data */ ++ /* call the helper function which doesn't unlock the page */ ++ if (!Page_Uptodate(page)) ++ err = mini_fo_do_readpage(file, page); ++ ++ out: ++ print_exit_status(err); ++ return err; ++} ++ ++ ++ ++STATIC int ++mini_fo_commit_write(file_t *file, page_t *page, unsigned from, unsigned to) ++{ ++ int err = -ENOMEM; ++ inode_t *inode; ++ inode_t *hidden_inode; ++ page_t *hidden_page; ++ file_t *hidden_file = NULL; ++ loff_t pos; ++ unsigned bytes = to - from; ++ unsigned hidden_from, hidden_to, hidden_bytes; ++ ++ print_entry_location(); ++ ++ inode = page->mapping->host; /* CPW: Moved below print_entry_location */ ++ hidden_inode = itohi(inode); ++ ++ ASSERT(file != NULL); ++ /* ++ * here we have a kmapped page, with data from the user copied ++ * into it. we need to encode_block it, and then call the lower ++ * commit_write. We also need to simulate same behavior of ++ * generic_file_write, and call prepare_write on the lower f/s first. ++ */ ++#ifdef FIST_COUNT_WRITES ++ count_writes++; ++# endif /* FIST_COUNT_WRITES */ ++ ++ /* this is append and/or extend -- we can't have holes so fill them in */ ++ if (page->index > (hidden_inode->i_size >> PAGE_CACHE_SHIFT)) { ++ page_t *tmp_page; ++ int index; ++ for (index = hidden_inode->i_size >> PAGE_CACHE_SHIFT; index < page->index; index++) { ++ tmp_page = mini_fo_get1page(file, index); ++ if (IS_ERR(tmp_page)) { ++ err = PTR_ERR(tmp_page); ++ goto out; ++ } ++ /* zero out the contents of the page at the appropriate offsets */ ++ memset((char*)page_address(tmp_page) + (inode->i_size & ~PAGE_CACHE_MASK), 0, PAGE_CACHE_SIZE - (inode->i_size & ~PAGE_CACHE_MASK)); ++ if (!(err = mini_fo_prepare_write(file, tmp_page, 0, PAGE_CACHE_SIZE))) ++ err = mini_fo_commit_write(file, tmp_page, 0, PAGE_CACHE_SIZE); ++ page_cache_release(tmp_page); ++ if (err < 0) ++ goto out; ++ if (current->need_resched) ++ schedule(); ++ } ++ } ++ ++ if (ftopd(file) != NULL) ++ hidden_file = ftohf(file); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_inode->i_mutex); ++#else ++ down(&hidden_inode->i_sem); ++#endif ++ /* find lower page (returns a locked page) */ ++ hidden_page = grab_cache_page(hidden_inode->i_mapping, page->index); ++ if (!hidden_page) ++ goto out; ++ ++#if FIST_ENCODING_BLOCKSIZE > 1 ++# error encoding_blocksize greater than 1 is not yet supported ++# endif /* FIST_ENCODING_BLOCKSIZE > 1 */ ++ ++ hidden_from = from & (~(FIST_ENCODING_BLOCKSIZE - 1)); ++ hidden_to = ((to + FIST_ENCODING_BLOCKSIZE - 1) & (~(FIST_ENCODING_BLOCKSIZE - 1))); ++ if ((page->index << PAGE_CACHE_SHIFT) + to > hidden_inode->i_size) { ++ ++ /* ++ * if this call to commit_write had introduced holes and the code ++ * for handling holes was invoked, then the beginning of this page ++ * must be zeroed out ++ * zero out bytes from 'size_of_file%pagesize' to 'from'. ++ */ ++ if ((hidden_from - (inode->i_size & ~PAGE_CACHE_MASK)) > 0) ++ memset((char*)page_address(page) + (inode->i_size & ~PAGE_CACHE_MASK), 0, hidden_from - (inode->i_size & ~PAGE_CACHE_MASK)); ++ ++ } ++ hidden_bytes = hidden_to - hidden_from; ++ ++ /* call lower prepare_write */ ++ err = -EINVAL; ++ if (hidden_inode->i_mapping && ++ hidden_inode->i_mapping->a_ops && ++ hidden_inode->i_mapping->a_ops->prepare_write) ++ err = hidden_inode->i_mapping->a_ops->prepare_write(hidden_file, ++ hidden_page, ++ hidden_from, ++ hidden_to); ++ if (err) ++ /* don't leave locked pages behind, esp. on an ENOSPC */ ++ goto out_unlock; ++ ++ fist_dprint(8, "%s: encoding %d bytes\n", __FUNCTION__, hidden_bytes); ++ mini_fo_encode_block((char *) page_address(page) + hidden_from, (char*) page_address(hidden_page) + hidden_from, hidden_bytes, inode, inode->i_sb, page->index); ++ /* if encode_block could fail, then goto unlock and return error */ ++ ++ /* call lower commit_write */ ++ err = hidden_inode->i_mapping->a_ops->commit_write(hidden_file, ++ hidden_page, ++ hidden_from, ++ hidden_to); ++ ++ if (err < 0) ++ goto out_unlock; ++ ++ err = bytes; /* convert error to no. of bytes */ ++ ++ inode->i_blocks = hidden_inode->i_blocks; ++ /* we may have to update i_size */ ++ pos = (page->index << PAGE_CACHE_SHIFT) + to; ++ if (pos > inode->i_size) ++ inode->i_size = pos; ++ ++ /* ++ * update mtime and ctime of lower level file system ++ * mini_fo' mtime and ctime are updated by generic_file_write ++ */ ++ hidden_inode->i_mtime = hidden_inode->i_ctime = CURRENT_TIME; ++ ++ mark_inode_dirty_sync(inode); ++ ++ out_unlock: ++ UnlockPage(hidden_page); ++ page_cache_release(hidden_page); ++ kunmap(page); /* kmap was done in prepare_write */ ++ out: ++ /* we must set our page as up-to-date */ ++ if (err < 0) ++ ClearPageUptodate(page); ++ else ++ SetPageUptodate(page); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_inode->i_mutex); ++#else ++ up(&hidden_inode->i_sem); ++#endif ++ print_exit_status(err); ++ return err; /* assume all is ok */ ++} ++ ++ ++STATIC int ++mini_fo_bmap(struct address_space *mapping, long block) ++{ ++ int err = 0; ++ inode_t *inode; ++ inode_t *hidden_inode; ++ ++ print_entry_location(); ++ ++ inode = (inode_t *) mapping->host; ++ hidden_inode = itohi(inode); ++ ++ if (hidden_inode->i_mapping->a_ops->bmap) ++ err = hidden_inode->i_mapping->a_ops->bmap(hidden_inode->i_mapping, block); ++ print_exit_location(); ++ return err; ++} ++ ++ ++/* ++ * This function is copied verbatim from mm/filemap.c. ++ * XXX: It should be simply moved to some header file instead -- bug Al about it! ++ */ ++static inline int sync_page(struct page *page) ++{ ++ struct address_space *mapping = page->mapping; ++ ++ if (mapping && mapping->a_ops && mapping->a_ops->sync_page) ++ return mapping->a_ops->sync_page(page); ++ return 0; ++} ++ ++ ++/* ++ * XXX: we may not need this function if not FIST_FILTER_DATA. ++ * FIXME: for FIST_FILTER_SCA, get all lower pages and sync them each. ++ */ ++STATIC int ++mini_fo_sync_page(page_t *page) ++{ ++ int err = 0; ++ inode_t *inode; ++ inode_t *hidden_inode; ++ page_t *hidden_page; ++ ++ print_entry_location(); ++ ++ inode = page->mapping->host; /* CPW: Moved below print_entry_location */ ++ hidden_inode = itohi(inode); ++ ++ /* find lower page (returns a locked page) */ ++ hidden_page = grab_cache_page(hidden_inode->i_mapping, page->index); ++ if (!hidden_page) ++ goto out; ++ ++ err = sync_page(hidden_page); ++ ++ UnlockPage(hidden_page); /* b/c grab_cache_page locked it */ ++ page_cache_release(hidden_page); /* b/c grab_cache_page increased refcnt */ ++ ++ out: ++ print_exit_status(err); ++ return err; ++} +--- /dev/null ++++ b/fs/mini_fo/README +@@ -0,0 +1,163 @@ ++README for the mini_fo overlay file system ++========================================= ++ ++ ++WHAT IS MINI_FO? ++---------------- ++ ++mini_fo is a virtual kernel file system that can make read-only ++file systems writable. This is done by redirecting modifying operations ++to a writeable location called "storage directory", and leaving the ++original data in the "base directory" untouched. When reading, the ++file system merges the modifed and original data so that only the ++newest versions will appear. This occurs transparently to the user, ++who can access the data like on any other read-write file system. ++ ++Base and storage directories may be located on the same or on ++different partitions and may be of different file system types. While ++the storage directory obviously needs to be writable, the base may or ++may not be writable, what doesn't matter as it will no be modified ++anyway. ++ ++ ++WHAT IS GOOD FOR? ++----------------- ++ ++The primary purpose of the mini_fo file system is to allow easy ++software updates to embedded systems, that often store their root ++file system in a read-only flash file system, but there are many ++more as for example sandboxing, or for allowing live-cds to ++permanently store information. ++ ++ ++BUILDING ++-------- ++This should be simple. Adjust the Makefile to point to the correct ++kernel headers you want to build the module for. Then: ++ ++ # make ++ ++should build "mini_fo.o" for a 2.4 kernel or "mini_fo.ko" for a 2.6 ++kernel. ++ ++If you are building the module for you current kernel, you can install ++the module (as root): ++ ++ # make install ++ ++or uninstall with ++ ++ # make uninstall ++ ++ ++USING THE FILE SYSTEM ++-------------------- ++ ++the general mount syntax is: ++ ++ mount -t mini_fo -o base=,sto=\ ++ ++ ++Example: ++ ++You have mounted a cdrom to /mnt/cdrom and want to modifiy some files ++on it: ++ ++load the module (as root) ++ ++ # insmod mini_fo.o for a 2.4 kernel or ++ ++ # insmod mini_fo.ko for a 2.6 kernel ++ ++ ++create a storage dir in tmp and a mountpoint for mini_fo: ++ ++ # mkdir /tmp/sto ++ # mkdir /mnt/mini_fo ++ ++and mount the mini_fo file system: ++ ++ # mount -t mini_fo -o base=/mnt/cdrom,sto=/tmp/sto /mnt/cdrom /mnt/mini_fo ++ ++ ++Now the data stored on the cd can be accessed via the mini_fo ++mountpoint just like any read-write file system, files can be modified ++and deleted, new ones can be created and so on. When done unmount the ++file system: ++ ++ # unmount /mnt/mini_fo ++ ++Note that if the file system is mounted again using the same storage ++file system, of course it will appear in the modified state again. If ++you remount it using an new empty storage directory, it will be ++unmodified. Therefore by executing: ++ ++ # cd /tmp/sto ++ # rm -rf * ++ ++you can nuke all the changes you made to the original file system. But ++ remember NEVER do this while the mini_fo file system is mounted! ++ ++ ++Alternatively you can use the mini_fo-overlay bash script, that ++simplifies managing mini_fo mounts. See TOOLS Section. ++ ++ ++TOOLS ++----- ++ ++mini_fo-merge (experimental): ++ ++This is a bash script that will merge changes contained in the storage ++directory back to the base directory. This allows mini_fo to function ++as a cache file system by overlaying a slow (network, ...) file system ++and using a fast (ramdisk, ...) as storage. When done, changes can be ++merged back to the (slow) base with mini_fo-merge. See "mini_fo-merge ++-h" for details. ++ ++It can be usefull for merging changes back after a successfull test ++(patches, software updates...) ++ ++ ++mini_fo-overlay: ++ ++This bash script simplifies managing one or more mini_fo mounts. For ++overlaying a directory called "basedir1", you can just call: ++ ++ # mini_fo-overlay basedir1 ++ ++This will mount mini_fo with "basedir1" as base, "/tmp/sto-basedir1/" ++as storage to "/mnt/mini_fo-basedir1/". It has more options though, ++type "mini_fo-overlay -h" for details. ++ ++ ++DOCUMENTATION, REPORTING BUGS, GETTING HELP ++------------------------------------------- ++ ++Please visit the mini_fo project page at: ++ ++http://www.denx.de/twiki/bin/view/Know/MiniFOHome ++ ++ ++WARNINGS ++-------- ++ ++Never modify the base or the storage directorys while the mini_fo ++file system is mounted, or you might crash you system. Simply accessing ++and reading should not cause any trouble. ++ ++Exporting a mini_fo mount point via NFS has not been tested, and may ++or may not work. ++ ++Check the RELEASE_NOTES for details on bugs and features. ++ ++ ++ ++Copyright (C) 2004, 2005 Markus Klotzbuecher ++ ++This program is free software; you can redistribute it and/or ++modify it under the terms of the GNU General Public License ++as published by the Free Software Foundation; either version ++2 of the License, or (at your option) any later version. ++ ++ +--- /dev/null ++++ b/fs/mini_fo/RELEASE_NOTES +@@ -0,0 +1,111 @@ ++Release: mini_fo-0.6.1 (v0-6-1) ++Date: 21.09.2005 ++ ++ ++Changes: ++-------- ++v0-6-1: ++ ++- bugfixes (see ChangeLog) ++ ++- two helper scripts "mini_fo_merge" and "mini_fo_overlay" (see ++ README for details). ++ ++v0-6-0: ++ ++- Support for 2.4 and 2.6 (see Makefile) ++ ++- Partial hard link support (creating works as expected, but already ++ existing links in the base file system will be treated as if they ++ were individual files). ++ ++- Various bugfixes and cleanups. ++ ++ ++v0-6-0-pre1: ++ ++- This is mini_fo-0-6-0-pre1! This release is a complete rewrite of ++ many vital mini_fo parts such as the old whiteout list code which ++ has been replaced by the new META subsystem. ++ ++- Light weight directory renaming implemented. This means if a ++ directory is renamed via the mini_fo filesystem this will no longer ++ result in a complete copy in storage, instead only one empty ++ directory will be created. All base filed contained in the original ++ directory stay there until modified. ++ ++- Special files (creating, renaming, deleting etc.) now working. ++ ++- Many bugfixes and cleanup, mini_fo is now a lot more stable. ++ ++ ++v0-5-10: ++ ++- Final release of the 0-5-* versions. Next will be a complete rewrite ++ of many features. This release contains several bugfixes related to ++ directory renaming. ++ ++ ++v0-5-10-pre6: ++ ++- Lots of cleanup and several bugfixes related to directory deleting ++ ++- Directory renaming suddenly works, what is most likely due to the ++ fact tha that "mv" is smart: if the classic rename doesn't work it ++ will assume that source and target file are on different fs and will ++ copy the directory and try to remove the source directory. Until ++ directory removing wasn't implemented, it would fail to do this and ++ rollback. ++ So, directory renaming works for now, but it doesn't yet do what you ++ would expect from a overlay fs, so use with care. ++ ++ ++v0-5-10-pre5: ++ ++- implemented directory deleting ++- made parsing of mount options more stable ++- New format of mount options! (See README) ++- I can't reproduce the unknown panic with 2.4.25 anymore, so I'll ++ happily assume it never existed! ++ ++ ++Implemented features: ++--------------------- ++ ++- creating hard links (see BUGS on already existing hard links) ++- lightweight directory renaming ++- renaming device files, pipes, sockets, etc. ++- creating, renaming, deleting of special files ++- deleting directorys ++- general directory reading (simple "ls" ) ++- creating files in existing directorys ++- creating directorys ++- renaming files. ++- reading and writing files (involves opening) ++- appending to files (creates copy in storage) ++- deleting files ++- llseek works too, what allows editors to work ++- persistency (a deleted file stay deleted over remounts) ++- use of symbolic links ++- creating of device files ++ ++ ++Not (yet) implemented features: ++------------------------------- ++ ++- full hard link support. ++ ++ ++ ++BUGS: ++----- ++ ++Hard links in the base file system will be treated as individual ++files, not as links to one inode. ++ ++The main problem with hard links isn't allowing to create them, but ++their pure existence. If you modify a base hard link, the changes made ++will only show up on this link, the other link will remain in the ++original state. I hope to fix this someday. Please note that this does ++not effect the special hard links '.' and '..', that are handled ++seperately by the lower fs. +--- /dev/null ++++ b/fs/mini_fo/state.c +@@ -0,0 +1,620 @@ ++/* ++ * Copyright (C) 2005 Markus Klotzbuecher ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ */ ++ ++#ifdef HAVE_CONFIG_H ++# include ++#endif /* HAVE_CONFIG_H */ ++ ++#include "fist.h" ++#include "mini_fo.h" ++ ++ ++/* create the storage file, setup new states */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++int create_sto_reg_file(dentry_t *dentry, int mode, struct nameidata *nd) ++#else ++int create_sto_reg_file(dentry_t *dentry, int mode) ++#endif ++{ ++ int err = 0; ++ inode_t *dir; ++ dentry_t *hidden_sto_dentry; ++ dentry_t *hidden_sto_dir_dentry; ++ ++ if(exists_in_storage(dentry)) { ++ printk(KERN_CRIT "mini_fo: create_sto_file: wrong type or state.\n"); ++ err = -EINVAL; ++ goto out; ++ } ++ err = get_neg_sto_dentry(dentry); ++ ++ if (err) { ++ printk(KERN_CRIT "mini_fo: create_sto_file: ERROR getting neg. sto dentry.\n"); ++ goto out; ++ } ++ ++ dir = dentry->d_parent->d_inode; ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ /* lock parent */ ++ hidden_sto_dir_dentry = dget(hidden_sto_dentry->d_parent); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ ++ err = PTR_ERR(hidden_sto_dir_dentry); ++ if (IS_ERR(hidden_sto_dir_dentry)) ++ goto out; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ err = vfs_create(hidden_sto_dir_dentry->d_inode, ++ hidden_sto_dentry, ++ mode, nd); ++#else ++ err = vfs_create(hidden_sto_dir_dentry->d_inode, ++ hidden_sto_dentry, ++ mode); ++#endif ++ if(err) { ++ printk(KERN_CRIT "mini_fo: create_sto_file: ERROR creating sto file.\n"); ++ goto out_lock; ++ } ++ ++ if(!dtohd2(dentry)->d_inode) { ++ printk(KERN_CRIT "mini_fo: create_sto_file: ERROR creating sto file [2].\n"); ++ err = -EINVAL; ++ goto out_lock; ++ } ++ ++ /* interpose the new inode */ ++ if(dtost(dentry) == DELETED) { ++ dtost(dentry) = DEL_REWRITTEN; ++ err = mini_fo_tri_interpose(NULL, hidden_sto_dentry, dentry, dir->i_sb, 0); ++ if(err) ++ goto out_lock; ++ } ++ else if(dtost(dentry) == NON_EXISTANT) { ++ dtost(dentry) = CREATED; ++ err = mini_fo_tri_interpose(dtohd(dentry), hidden_sto_dentry, dentry, dir->i_sb, 0); ++ if(err) ++ goto out_lock; ++ } ++ else if(dtost(dentry) == UNMODIFIED) { ++ dtost(dentry) = MODIFIED; ++ /* interpose on new inode */ ++ if(itohi2(dentry->d_inode) != NULL) { ++ printk(KERN_CRIT "mini_fo: create_sto_file: invalid inode detected.\n"); ++ err = -EINVAL; ++ goto out_lock; ++ } ++ itohi2(dentry->d_inode) = igrab(dtohd2(dentry)->d_inode); ++ } ++ fist_copy_attr_timesizes(dentry->d_parent->d_inode, ++ hidden_sto_dir_dentry->d_inode); ++ ++ out_lock: ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ dput(hidden_sto_dir_dentry); ++ out: ++ return err; ++} ++ ++/* create the sto dir, setup states */ ++int create_sto_dir(dentry_t *dentry, int mode) ++{ ++ int err = 0; ++ inode_t *dir; ++ dentry_t *hidden_sto_dentry; ++ dentry_t *hidden_sto_dir_dentry; ++ ++ /* had to take the "!S_ISDIR(mode))" check out, because it failed */ ++ if(exists_in_storage(dentry)) { ++ printk(KERN_CRIT "mini_fo: create_sto_dir: wrong type or state.\\ ++n"); ++ err = -EINVAL; ++ goto out; ++ } ++ ++ err = get_neg_sto_dentry(dentry); ++ if(err) { ++ err = -EINVAL; ++ goto out; ++ } ++ ++ dir = dentry->d_parent->d_inode; ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ /* was: hidden_sto_dir_dentry = lock_parent(hidden_sto_dentry); */ ++ hidden_sto_dir_dentry = dget(hidden_sto_dentry->d_parent); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ ++ err = PTR_ERR(hidden_sto_dir_dentry); ++ if (IS_ERR(hidden_sto_dir_dentry)) ++ goto out; ++ ++ err = vfs_mkdir(hidden_sto_dir_dentry->d_inode, ++ hidden_sto_dentry, ++ mode); ++ if(err) { ++ printk(KERN_CRIT "mini_fo: create_sto_dir: ERROR creating sto dir.\n"); ++ goto out_lock; ++ } ++ ++ if(!dtohd2(dentry)->d_inode) { ++ printk(KERN_CRIT "mini_fo: create_sto_dir: ERROR creating sto dir [2].\n"); ++ err = -EINVAL; ++ goto out_lock; ++ } ++ ++ /* interpose the new inode */ ++ if(dtost(dentry) == DELETED) { ++ dtost(dentry) = DEL_REWRITTEN; ++ err = mini_fo_tri_interpose(NULL, hidden_sto_dentry, dentry, dir->i_sb, 0); ++ if(err) ++ goto out_lock; ++ } ++ else if(dtopd(dentry)->state == NON_EXISTANT) { ++ dtopd(dentry)->state = CREATED; ++ err = mini_fo_tri_interpose(dtohd(dentry), hidden_sto_dentry, dentry, dir->i_sb, 0); ++ if(err) ++ goto out_lock; ++ } ++ else if(dtopd(dentry)->state == UNMODIFIED) { ++ dtopd(dentry)->state = MODIFIED; ++ /* interpose on new inode */ ++ if(itohi2(dentry->d_inode) != NULL) { ++ printk(KERN_CRIT "mini_fo: create_sto_dir: ERROR, invalid inode detected.\n"); ++ err = -EINVAL; ++ goto out_lock; ++ } ++ itohi2(dentry->d_inode) = igrab(dtohd2(dentry)->d_inode); ++ } ++ ++ fist_copy_attr_timesizes(dir, hidden_sto_dir_dentry->d_inode); ++ ++ /* initalize the wol list */ ++ itopd(dentry->d_inode)->deleted_list_size = -1; ++ itopd(dentry->d_inode)->renamed_list_size = -1; ++ meta_build_lists(dentry); ++ ++ ++ out_lock: ++ /* was: unlock_dir(hidden_sto_dir_dentry); */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ dput(hidden_sto_dir_dentry); ++ out: ++ return err; ++} ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++int create_sto_nod(dentry_t *dentry, int mode, dev_t dev) ++#else ++int create_sto_nod(dentry_t *dentry, int mode, int dev) ++#endif ++{ ++ int err = 0; ++ inode_t *dir; ++ dentry_t *hidden_sto_dentry; ++ dentry_t *hidden_sto_dir_dentry; ++ ++ if(exists_in_storage(dentry)) { ++ err = -EEXIST; ++ goto out; ++ } ++ err = get_neg_sto_dentry(dentry); ++ ++ if (err) { ++ printk(KERN_CRIT "mini_fo: create_sto_nod: ERROR getting neg. sto dentry.\n"); ++ goto out; ++ } ++ ++ dir = dentry->d_parent->d_inode; ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ /* lock parent */ ++ hidden_sto_dir_dentry = dget(hidden_sto_dentry->d_parent); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ ++ err = PTR_ERR(hidden_sto_dir_dentry); ++ if (IS_ERR(hidden_sto_dir_dentry)) ++ goto out; ++ ++ err = vfs_mknod(hidden_sto_dir_dentry->d_inode, hidden_sto_dentry, mode, dev); ++ if(err) ++ goto out_lock; ++ ++ if(!dtohd2(dentry)->d_inode) { ++ printk(KERN_CRIT "mini_fo: create_sto_nod: creating storage inode failed [1].\n"); ++ err = -EINVAL; /* return something indicating failure */ ++ goto out_lock; ++ } ++ ++ /* interpose the new inode */ ++ if(dtost(dentry) == DELETED) { ++ dtost(dentry) = DEL_REWRITTEN; ++ err = mini_fo_tri_interpose(NULL, hidden_sto_dentry, dentry, dir->i_sb, 0); ++ if(err) ++ goto out_lock; ++ } ++ else if(dtost(dentry) == NON_EXISTANT) { ++ dtost(dentry) = CREATED; ++ err = mini_fo_tri_interpose(dtohd(dentry), hidden_sto_dentry, dentry, dir->i_sb, 0); ++ if(err) ++ goto out_lock; ++ } ++ else if(dtost(dentry) == UNMODIFIED) { ++ dtost(dentry) = MODIFIED; ++ /* interpose on new inode */ ++ if(itohi2(dentry->d_inode) != NULL) { ++ printk(KERN_CRIT "mini_fo: create_sto_nod: error, invalid inode detected.\n"); ++ err = -EINVAL; ++ goto out_lock; ++ } ++ itohi2(dentry->d_inode) = igrab(dtohd2(dentry)->d_inode); ++ } ++ ++ fist_copy_attr_timesizes(dir, hidden_sto_dir_dentry->d_inode); ++ ++ out_lock: ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ dput(hidden_sto_dir_dentry); ++ out: ++ return err; ++} ++ ++ ++/* unimplemented (and possibly not usefull): ++ ++ nondir-del_to_del_rew ++ nondir-non_exist_to_creat ++ ++ dir-unmod_to_del ++ dir-mod_to_del ++ dir-creat_to_del ++ dir-del_rew_to_del ++ dir-del_to_del_rew ++ dir-non_exist_to_creat ++*/ ++ ++ ++/* bring a file of any type from state UNMODIFIED to MODIFIED */ ++int nondir_unmod_to_mod(dentry_t *dentry, int cp_flag) ++{ ++ int err = 0; ++ struct vfsmount *tgt_mnt; ++ struct vfsmount *src_mnt; ++ dentry_t *tgt_dentry; ++ dentry_t *src_dentry; ++ dentry_t *hidden_sto_dentry; ++ dentry_t *hidden_sto_dir_dentry; ++ ++ check_mini_fo_dentry(dentry); ++ ++ if((dtost(dentry) != UNMODIFIED) || ++ S_ISDIR(dentry->d_inode->i_mode)) { ++ printk(KERN_CRIT "mini_fo: nondir_unmod_to_mod: \ ++ wrong type or state.\n"); ++ err = -1; ++ goto out; ++ } ++ err = get_neg_sto_dentry(dentry); ++ ++ if (err) { ++ printk(KERN_CRIT "mini_fo: nondir_unmod_to_mod: \ ++ ERROR getting neg. sto dentry.\n"); ++ goto out; ++ } ++ ++ /* create sto file */ ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ /* lock parent */ ++ hidden_sto_dir_dentry = dget(hidden_sto_dentry->d_parent); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ ++ err = PTR_ERR(hidden_sto_dir_dentry); ++ if (IS_ERR(hidden_sto_dir_dentry)) ++ goto out; ++ ++ /* handle different types of nondirs */ ++ if(S_ISCHR(dentry->d_inode->i_mode) || ++ S_ISBLK(dentry->d_inode->i_mode)) { ++ err = vfs_mknod(hidden_sto_dir_dentry->d_inode, ++ hidden_sto_dentry, ++ dtohd(dentry)->d_inode->i_mode, ++ dtohd(dentry)->d_inode->i_rdev); ++ } ++ ++ else if(S_ISREG(dentry->d_inode->i_mode)) { ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ err = vfs_create(hidden_sto_dir_dentry->d_inode, ++ hidden_sto_dentry, ++ dtohd(dentry)->d_inode->i_mode, NULL); ++#else ++ err = vfs_create(hidden_sto_dir_dentry->d_inode, ++ hidden_sto_dentry, ++ dtohd(dentry)->d_inode->i_mode); ++#endif ++ } ++ if(err) { ++ printk(KERN_CRIT "mini_fo: nondir_unmod_to_mod: \ ++ ERROR creating sto file.\n"); ++ goto out_lock; ++ } ++ ++ /* interpose on new inode */ ++ if(itohi2(dentry->d_inode) != NULL) { ++ printk(KERN_CRIT "mini_fo: nondir_unmod_to_mod: \ ++ ERROR, invalid inode detected.\n"); ++ err = -EINVAL; ++ goto out_lock; ++ } ++ ++ itohi2(dentry->d_inode) = igrab(dtohd2(dentry)->d_inode); ++ ++ fist_copy_attr_timesizes(dentry->d_parent->d_inode, ++ hidden_sto_dir_dentry->d_inode); ++ dtost(dentry) = MODIFIED; ++ ++ /* copy contents if regular file and cp_flag = 1 */ ++ if((cp_flag == 1) && S_ISREG(dentry->d_inode->i_mode)) { ++ ++ /* unlock first */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ ++ dput(hidden_sto_dir_dentry); ++ ++ tgt_dentry = dtohd2(dentry); ++ tgt_mnt = stopd(dentry->d_inode->i_sb)->hidden_mnt2; ++ src_dentry = dtohd(dentry); ++ src_mnt = stopd(dentry->d_inode->i_sb)->hidden_mnt; ++ ++ err = mini_fo_cp_cont(tgt_dentry, tgt_mnt, ++ src_dentry, src_mnt); ++ if(err) { ++ printk(KERN_CRIT "mini_fo: nondir_unmod_to_mod: \ ++ ERROR copying contents.\n"); ++ } ++ goto out; ++ } ++ ++ out_lock: ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ dput(hidden_sto_dir_dentry); ++ out: ++ return err; ++} ++ ++/* this function is currently identical to nondir_creat_to_del */ ++int nondir_del_rew_to_del(dentry_t *dentry) ++{ ++ return nondir_creat_to_del(dentry); ++} ++ ++int nondir_creat_to_del(dentry_t *dentry) ++{ ++ int err = 0; ++ ++ inode_t *hidden_sto_dir_inode; ++ dentry_t *hidden_sto_dir_dentry; ++ dentry_t *hidden_sto_dentry; ++ ++ check_mini_fo_dentry(dentry); ++ ++ /* for now this function serves for both state DEL_REWRITTEN and ++ * CREATED */ ++ if(!(dtost(dentry) == CREATED || (dtost(dentry) == DEL_REWRITTEN)) || ++ S_ISDIR(dentry->d_inode->i_mode)) { ++ printk(KERN_CRIT "mini_fo: nondir_mod_to_del/del_rew_to_del: \ ++ wrong type or state.\n"); ++ err = -1; ++ goto out; ++ } ++ ++ hidden_sto_dir_inode = itohi2(dentry->d_parent->d_inode); ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ /* was: hidden_sto_dir_dentry = lock_parent(hidden_sto_dentry);*/ ++ hidden_sto_dir_dentry = dget(hidden_sto_dentry->d_parent); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ ++ /* avoid destroying the hidden inode if the file is in use */ ++ dget(hidden_sto_dentry); ++ err = vfs_unlink(hidden_sto_dir_inode, hidden_sto_dentry); ++ dput(hidden_sto_dentry); ++ if(!err) ++ d_delete(hidden_sto_dentry); ++ ++ /* propagate number of hard-links */ ++ dentry->d_inode->i_nlink = itohi2(dentry->d_inode)->i_nlink; ++ ++ dtost(dentry) = NON_EXISTANT; ++ ++ /* was: unlock_dir(hidden_sto_dir_dentry); */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ dput(hidden_sto_dir_dentry); ++ ++ out: ++ return err; ++} ++ ++int nondir_mod_to_del(dentry_t *dentry) ++{ ++ int err; ++ dentry_t *hidden_sto_dentry; ++ inode_t *hidden_sto_dir_inode; ++ dentry_t *hidden_sto_dir_dentry; ++ ++ check_mini_fo_dentry(dentry); ++ ++ if(dtost(dentry) != MODIFIED || ++ S_ISDIR(dentry->d_inode->i_mode)) { ++ printk(KERN_CRIT "mini_fo: nondir_mod_to_del: \ ++ wrong type or state.\n"); ++ err = -1; ++ goto out; ++ } ++ ++ hidden_sto_dir_inode = itohi2(dentry->d_parent->d_inode); ++ hidden_sto_dentry = dtohd2(dentry); ++ ++ /* was hidden_sto_dir_dentry = lock_parent(hidden_sto_dentry); */ ++ hidden_sto_dir_dentry = dget(hidden_sto_dentry->d_parent); ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_lock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ down(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ ++ /* avoid destroying the hidden inode if the file is in use */ ++ dget(hidden_sto_dentry); ++ err = vfs_unlink(hidden_sto_dir_inode, hidden_sto_dentry); ++ dput(hidden_sto_dentry); ++ if(!err) ++ d_delete(hidden_sto_dentry); ++ ++ /* propagate number of hard-links */ ++ dentry->d_inode->i_nlink = itohi2(dentry->d_inode)->i_nlink; ++ ++ /* dput base dentry, this will relase the inode and free the ++ * dentry, as we will never need it again. */ ++ dput(dtohd(dentry)); ++ dtohd(dentry) = NULL; ++ dtost(dentry) = DELETED; ++ ++ /* add deleted file to META-file */ ++ meta_add_d_entry(dentry->d_parent, ++ dentry->d_name.name, ++ dentry->d_name.len); ++ ++ /* was: unlock_dir(hidden_sto_dir_dentry); */ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++#else ++ up(&hidden_sto_dir_dentry->d_inode->i_sem); ++#endif ++ dput(hidden_sto_dir_dentry); ++ ++ out: ++ return err; ++} ++ ++int nondir_unmod_to_del(dentry_t *dentry) ++{ ++ int err = 0; ++ ++ check_mini_fo_dentry(dentry); ++ ++ if(dtost(dentry) != UNMODIFIED || ++ S_ISDIR(dentry->d_inode->i_mode)) { ++ printk(KERN_CRIT "mini_fo: nondir_unmod_to_del: \ ++ wrong type or state.\n"); ++ err = -1; ++ goto out; ++ } ++ ++ /* next we have to get a negative dentry for the storage file */ ++ err = get_neg_sto_dentry(dentry); ++ ++ if(err) ++ goto out; ++ ++ /* add deleted file to META lists */ ++ err = meta_add_d_entry(dentry->d_parent, ++ dentry->d_name.name, ++ dentry->d_name.len); ++ ++ if(err) ++ goto out; ++ ++ /* dput base dentry, this will relase the inode and free the ++ * dentry, as we will never need it again. */ ++ dput(dtohd(dentry)); ++ dtohd(dentry) = NULL; ++ dtost(dentry) = DELETED; ++ ++ out: ++ return err; ++} ++ ++/* bring a dir from state UNMODIFIED to MODIFIED */ ++int dir_unmod_to_mod(dentry_t *dentry) ++{ ++ int err; ++ ++ check_mini_fo_dentry(dentry); ++ ++ if(dtost(dentry) != UNMODIFIED || ++ !S_ISDIR(dentry->d_inode->i_mode)) { ++ printk(KERN_CRIT "mini_fo: dir_unmod_to_mod: \ ++ wrong type or state.\n"); ++ err = -1; ++ goto out; ++ } ++ ++ /* this creates our dir incl. sto. structure */ ++ err = build_sto_structure(dentry->d_parent, dentry); ++ if(err) { ++ printk(KERN_CRIT "mini_fo: dir_unmod_to_mod: \ ++ build_sto_structure failed.\n"); ++ goto out; ++ } ++ out: ++ return err; ++} ++ +--- /dev/null ++++ b/fs/mini_fo/super.c +@@ -0,0 +1,281 @@ ++/* ++ * Copyright (c) 1997-2003 Erez Zadok ++ * Copyright (c) 2001-2003 Stony Brook University ++ * ++ * For specific licensing information, see the COPYING file distributed with ++ * this package, or get one from ftp://ftp.filesystems.org/pub/fist/COPYING. ++ * ++ * This Copyright notice must be kept intact and distributed with all ++ * fistgen sources INCLUDING sources generated by fistgen. ++ */ ++/* ++ * Copyright (C) 2004, 2005 Markus Klotzbuecher ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version ++ * 2 of the License, or (at your option) any later version. ++ */ ++ ++/* ++ * $Id$ ++ */ ++ ++#ifdef HAVE_CONFIG_H ++# include ++#endif ++ ++#include "fist.h" ++#include "mini_fo.h" ++ ++ ++STATIC void ++mini_fo_read_inode(inode_t *inode) ++{ ++ static struct address_space_operations mini_fo_empty_aops; ++ ++ __itopd(inode) = kmalloc(sizeof(struct mini_fo_inode_info), GFP_KERNEL); ++ if (!itopd(inode)) { ++ printk("<0>%s:%s:%d: No kernel memory!\n", __FILE__, __FUNCTION__, __LINE__); ++ ASSERT(NULL); ++ } ++ itohi(inode) = NULL; ++ itohi2(inode) = NULL; ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++ inode->i_version++; ++#else ++ inode->i_version = ++event; /* increment inode version */ ++#endif ++ inode->i_op = &mini_fo_main_iops; ++ inode->i_fop = &mini_fo_main_fops; ++#if 0 ++ /* ++ * XXX: To export a file system via NFS, it has to have the ++ * FS_REQUIRES_DEV flag, so turn it on. But should we inherit it from ++ * the lower file system, or can we allow our file system to be exported ++ * even if the lower one cannot be natively exported. ++ */ ++ inode->i_sb->s_type->fs_flags |= FS_REQUIRES_DEV; ++ /* ++ * OK, the above was a hack, which is now turned off because it may ++ * cause a panic/oops on some systems. The correct way to export a ++ * "nodev" filesystem is via using nfs-utils > 1.0 and the "fsid=" export ++ * parameter, which requires 2.4.20 or later. ++ */ ++#endif ++ /* I don't think ->a_ops is ever allowed to be NULL */ ++ inode->i_mapping->a_ops = &mini_fo_empty_aops; ++} ++ ++ ++#if defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) ++/* ++ * No need to call write_inode() on the lower inode, as it ++ * will have been marked 'dirty' anyway. But we might need ++ * to write some of our own stuff to disk. ++ */ ++STATIC void ++mini_fo_write_inode(inode_t *inode, int sync) ++{ ++ print_entry_location(); ++ print_exit_location(); ++} ++#endif /* defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) */ ++ ++ ++STATIC void ++mini_fo_put_inode(inode_t *inode) ++{ ++ /* ++ * This is really funky stuff: ++ * Basically, if i_count == 1, iput will then decrement it and this inode will be destroyed. ++ * It is currently holding a reference to the hidden inode. ++ * Therefore, it needs to release that reference by calling iput on the hidden inode. ++ * iput() _will_ do it for us (by calling our clear_inode), but _only_ if i_nlink == 0. ++ * The problem is, NFS keeps i_nlink == 1 for silly_rename'd files. ++ * So we must for our i_nlink to 0 here to trick iput() into calling our clear_inode. ++ */ ++ if (atomic_read(&inode->i_count) == 1) ++ inode->i_nlink = 0; ++} ++ ++ ++#if defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) ++/* ++ * we now define delete_inode, because there are two VFS paths that may ++ * destroy an inode: one of them calls clear inode before doing everything ++ * else that's needed, and the other is fine. This way we truncate the inode ++ * size (and its pages) and then clear our own inode, which will do an iput ++ * on our and the lower inode. ++ */ ++STATIC void ++mini_fo_delete_inode(inode_t *inode) ++{ ++ print_entry_location(); ++ ++ fist_checkinode(inode, "mini_fo_delete_inode IN"); ++ inode->i_size = 0; /* every f/s seems to do that */ ++ clear_inode(inode); ++ ++ print_exit_location(); ++} ++#endif /* defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) */ ++ ++ ++/* final actions when unmounting a file system */ ++STATIC void ++mini_fo_put_super(super_block_t *sb) ++{ ++ if (stopd(sb)) { ++ mntput(stopd(sb)->hidden_mnt); ++ mntput(stopd(sb)->hidden_mnt2); ++ ++ /* mk: no! dput(stopd(sb)->base_dir_dentry); ++ dput(stopd(sb)->storage_dir_dentry); */ ++ ++ kfree(stopd(sb)); ++ __stopd(sb) = NULL; ++ } ++} ++ ++ ++#ifdef NOT_NEEDED ++/* ++ * This is called in do_umount before put_super. ++ * The superblock lock is not held yet. ++ * We probably do not need to define this or call write_super ++ * on the hidden_sb, because sync_supers() will get to hidden_sb ++ * sooner or later. But it is also called from file_fsync()... ++ */ ++STATIC void ++mini_fo_write_super(super_block_t *sb) ++{ ++ return; ++} ++#endif /* NOT_NEEDED */ ++ ++ ++STATIC int ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18) ++mini_fo_statfs(struct dentry *d, struct kstatfs *buf) ++#elif LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++mini_fo_statfs(super_block_t *sb, struct kstatfs *buf) ++#else ++mini_fo_statfs(super_block_t *sb, struct statfs *buf) ++#endif ++{ ++ int err = 0; ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18) ++ struct dentry *hidden_d; ++ ++ hidden_d = dtohd(d); ++ err = vfs_statfs(hidden_d, buf); ++#else ++ super_block_t *hidden_sb; ++ ++ hidden_sb = stohs(sb); ++ err = vfs_statfs(hidden_sb, buf); ++#endif ++ ++ return err; ++} ++ ++ ++/* ++ * XXX: not implemented. This is not allowed yet. ++ * Should we call this on the hidden_sb? Probably not. ++ */ ++STATIC int ++mini_fo_remount_fs(super_block_t *sb, int *flags, char *data) ++{ ++ //printk(KERN_CRIT "mini_fo_remount_fs: WARNING, this function is umimplemented.\n"); ++ return -ENOSYS; ++} ++ ++ ++/* ++ * Called by iput() when the inode reference count reached zero ++ * and the inode is not hashed anywhere. Used to clear anything ++ * that needs to be, before the inode is completely destroyed and put ++ * on the inode free list. ++ */ ++STATIC void ++mini_fo_clear_inode(inode_t *inode) ++{ ++ /* ++ * Decrement a reference to a hidden_inode, which was incremented ++ * by our read_inode when it was created initially. ++ */ ++ ++ /* release the wol_list */ ++ if(S_ISDIR(inode->i_mode)) { ++ __meta_put_lists(inode); ++ } ++ ++ /* mk: fan out fun */ ++ if(itohi(inode)) ++ iput(itohi(inode)); ++ if(itohi2(inode)) ++ iput(itohi2(inode)); ++ ++ // XXX: why this assertion fails? ++ // because it doesn't like us ++ // ASSERT((inode->i_state & I_DIRTY) == 0); ++ kfree(itopd(inode)); ++ __itopd(inode) = NULL; ++} ++ ++ ++/* ++ * Called in do_umount() if the MNT_FORCE flag was used and this ++ * function is defined. See comment in linux/fs/super.c:do_umount(). ++ * Used only in nfs, to kill any pending RPC tasks, so that subsequent ++ * code can actually succeed and won't leave tasks that need handling. ++ * ++ * PS. I wonder if this is somehow useful to undo damage that was ++ * left in the kernel after a user level file server (such as amd) ++ * dies. ++ */ ++STATIC void ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18) ++mini_fo_umount_begin(struct vfsmount *mnt, int flags) ++{ ++ struct vfsmount *hidden_mnt; ++ ++ hidden_mnt = stopd(mnt->mnt_sb)->hidden_mnt; ++ ++ if (hidden_mnt->mnt_sb->s_op->umount_begin) ++ hidden_mnt->mnt_sb->s_op->umount_begin(hidden_mnt, flags); ++ ++} ++#else ++mini_fo_umount_begin(super_block_t *sb) ++{ ++ super_block_t *hidden_sb; ++ ++ hidden_sb = stohs(sb); ++ ++ if (hidden_sb->s_op->umount_begin) ++ hidden_sb->s_op->umount_begin(hidden_sb); ++ ++} ++#endif ++ ++ ++struct super_operations mini_fo_sops = ++{ ++ read_inode: mini_fo_read_inode, ++#if defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) ++ write_inode: mini_fo_write_inode, ++#endif /* defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) */ ++ put_inode: mini_fo_put_inode, ++#if defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) ++ delete_inode: mini_fo_delete_inode, ++#endif /* defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) */ ++ put_super: mini_fo_put_super, ++ statfs: mini_fo_statfs, ++ remount_fs: mini_fo_remount_fs, ++ clear_inode: mini_fo_clear_inode, ++ umount_begin: mini_fo_umount_begin, ++}; +--- /dev/null ++++ b/fs/mini_fo/Kconfig +@@ -0,0 +1,3 @@ ++config MINI_FO ++ tristate "Mini fanout overlay filesystem" ++ diff --git a/target/linux/generic-2.6/patches-2.6.31/210-mini_fo_2.6.25_fixes.patch b/target/linux/generic-2.6/patches-2.6.31/210-mini_fo_2.6.25_fixes.patch new file mode 100644 index 0000000000..d71e3b6faa --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/210-mini_fo_2.6.25_fixes.patch @@ -0,0 +1,143 @@ +--- a/fs/mini_fo/main.c ++++ b/fs/mini_fo/main.c +@@ -79,6 +79,7 @@ mini_fo_tri_interpose(dentry_t *hidden_d + * of the new inode's fields + */ + ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,25) + /* + * original: inode = iget(sb, hidden_inode->i_ino); + */ +@@ -87,6 +88,13 @@ mini_fo_tri_interpose(dentry_t *hidden_d + err = -EACCES; /* should be impossible??? */ + goto out; + } ++#else ++ inode = mini_fo_iget(sb, iunique(sb, 25)); ++ if (IS_ERR(inode)) { ++ err = PTR_ERR(inode); ++ goto out; ++ } ++#endif + + /* + * interpose the inode if not already interposed +@@ -184,9 +192,9 @@ mini_fo_parse_options(super_block_t *sb, + hidden_root = ERR_PTR(err); + goto out; + } +- hidden_root = nd.dentry; +- stopd(sb)->base_dir_dentry = nd.dentry; +- stopd(sb)->hidden_mnt = nd.mnt; ++ hidden_root = nd_get_dentry(&nd); ++ stopd(sb)->base_dir_dentry = nd_get_dentry(&nd); ++ stopd(sb)->hidden_mnt = nd_get_mnt(&nd); + + } else if(!strncmp("sto=", options, 4)) { + /* parse the storage dir */ +@@ -204,9 +212,9 @@ mini_fo_parse_options(super_block_t *sb, + hidden_root2 = ERR_PTR(err); + goto out; + } +- hidden_root2 = nd2.dentry; +- stopd(sb)->storage_dir_dentry = nd2.dentry; +- stopd(sb)->hidden_mnt2 = nd2.mnt; ++ hidden_root2 = nd_get_dentry(&nd2); ++ stopd(sb)->storage_dir_dentry = nd_get_dentry(&nd2); ++ stopd(sb)->hidden_mnt2 = nd_get_mnt(&nd2); + stohs2(sb) = hidden_root2->d_sb; + + /* validate storage dir, this is done in +--- a/fs/mini_fo/mini_fo.h ++++ b/fs/mini_fo/mini_fo.h +@@ -302,6 +302,10 @@ extern int mini_fo_tri_interpose(dentry_ + extern int mini_fo_cp_cont(dentry_t *tgt_dentry, struct vfsmount *tgt_mnt, + dentry_t *src_dentry, struct vfsmount *src_mnt); + ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,25) ++extern struct inode *mini_fo_iget(struct super_block *sb, unsigned long ino); ++#endif ++ + #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) + extern int mini_fo_create(inode_t *dir, dentry_t *dentry, int mode, struct nameidata *nd); + +@@ -501,6 +505,29 @@ static inline void double_unlock(struct + #endif /* if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) */ + #endif /* __KERNEL__ */ + ++ ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,25) ++static inline dentry_t *nd_get_dentry(struct nameidata *nd) ++{ ++ return (nd->path.dentry); ++} ++ ++static inline struct vfsmount *nd_get_mnt(struct nameidata *nd) ++{ ++ return (nd->path.mnt); ++} ++#else ++static inline dentry_t *nd_get_dentry(struct nameidata *nd) ++{ ++ return (nd->dentry); ++} ++ ++static inline struct vfsmount *nd_get_mnt(struct nameidata *nd) ++{ ++ return (nd->mnt); ++} ++#endif ++ + /* + * Definitions for user and kernel code + */ +--- a/fs/mini_fo/super.c ++++ b/fs/mini_fo/super.c +@@ -262,10 +262,31 @@ mini_fo_umount_begin(super_block_t *sb) + } + #endif + ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,25) ++struct inode * ++mini_fo_iget(struct super_block *sb, unsigned long ino) ++{ ++ struct inode *inode; ++ ++ inode = iget_locked(sb, ino); ++ if (!inode) ++ return ERR_PTR(-ENOMEM); ++ ++ if (!(inode->i_state & I_NEW)) ++ return inode; ++ ++ mini_fo_read_inode(inode); ++ ++ unlock_new_inode(inode); ++ return inode; ++} ++#endif /* if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,25) */ + + struct super_operations mini_fo_sops = + { ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,25) + read_inode: mini_fo_read_inode, ++#endif + #if defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) + write_inode: mini_fo_write_inode, + #endif /* defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) */ +--- a/fs/mini_fo/aux.c ++++ b/fs/mini_fo/aux.c +@@ -164,11 +164,11 @@ dentry_t *bpath_walk(super_block_t *sb, + err = vfs_path_lookup(mnt->mnt_root, mnt, bpath+1, 0, &nd); + + /* validate */ +- if (err || !nd.dentry || !nd.dentry->d_inode) { ++ if (err || !nd_get_dentry(&nd) || !nd_get_dentry(&nd)->d_inode) { + printk(KERN_CRIT "mini_fo: bpath_walk: path_walk failed.\n"); + return NULL; + } +- return nd.dentry; ++ return nd_get_dentry(&nd); + } + + diff --git a/target/linux/generic-2.6/patches-2.6.31/211-mini_fo_2.6.25_dentry_open_war.patch b/target/linux/generic-2.6/patches-2.6.31/211-mini_fo_2.6.25_dentry_open_war.patch new file mode 100644 index 0000000000..48a19429b9 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/211-mini_fo_2.6.25_dentry_open_war.patch @@ -0,0 +1,66 @@ +--- a/fs/mini_fo/meta.c ++++ b/fs/mini_fo/meta.c +@@ -442,6 +442,11 @@ int meta_write_d_entry(dentry_t *dentry, + S_IRUSR | S_IWUSR); + #endif + } ++ ++ /* $%& err, is this correct? */ ++ meta_mnt = stopd(dentry->d_inode->i_sb)->hidden_mnt2; ++ mntget(meta_mnt); ++ + /* open META-file for writing */ + meta_file = dentry_open(meta_dentry, meta_mnt, 0x1); + if(!meta_file || IS_ERR(meta_file)) { +@@ -535,6 +540,11 @@ int meta_write_r_entry(dentry_t *dentry, + meta_dentry, S_IRUSR | S_IWUSR); + #endif + } ++ ++ /* $%& err, is this correct? */ ++ meta_mnt = stopd(dentry->d_inode->i_sb)->hidden_mnt2; ++ mntget(meta_mnt); ++ + /* open META-file for writing */ + meta_file = dentry_open(meta_dentry, meta_mnt, 0x1); + if(!meta_file || IS_ERR(meta_file)) { +@@ -671,14 +681,16 @@ int meta_sync_d_list(dentry_t *dentry, i + } + } + ++ /* $%& err, is this correct? */ ++ meta_mnt = stopd(dentry->d_inode->i_sb)->hidden_mnt2; ++ mntget(meta_mnt); ++ + /* open META-file for writing */ + meta_file = dentry_open(meta_dentry, meta_mnt, 0x1); + if(!meta_file || IS_ERR(meta_file)) { + printk(KERN_CRIT "mini_fo: meta_sync_d_list: \ + ERROR opening meta file.\n"); +- /* we don't mntget so we dont't mntput (for now) +- * mntput(meta_mnt); +- */ ++ mntput(meta_mnt); + dput(meta_dentry); + err = -1; + goto out; +@@ -811,14 +823,16 @@ int meta_sync_r_list(dentry_t *dentry, i + } + } + ++ /* $%& err, is this correct? */ ++ meta_mnt = stopd(dentry->d_inode->i_sb)->hidden_mnt2; ++ mntget(meta_mnt); ++ + /* open META-file for writing */ + meta_file = dentry_open(meta_dentry, meta_mnt, 0x1); + if(!meta_file || IS_ERR(meta_file)) { + printk(KERN_CRIT "mini_fo: meta_sync_r_list: \ + ERROR opening meta file.\n"); +- /* we don't mntget so we dont't mntput (for now) +- * mntput(meta_mnt); +- */ ++ mntput(meta_mnt); + dput(meta_dentry); + err = -1; + goto out; diff --git a/target/linux/generic-2.6/patches-2.6.31/212-mini_fo_2.6.26_fixes.patch b/target/linux/generic-2.6/patches-2.6.31/212-mini_fo_2.6.26_fixes.patch new file mode 100644 index 0000000000..8bd9ba3244 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/212-mini_fo_2.6.26_fixes.patch @@ -0,0 +1,37 @@ +--- a/fs/mini_fo/super.c ++++ b/fs/mini_fo/super.c +@@ -84,6 +84,7 @@ mini_fo_write_inode(inode_t *inode, int + #endif /* defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) */ + + ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,26) + STATIC void + mini_fo_put_inode(inode_t *inode) + { +@@ -99,6 +100,7 @@ mini_fo_put_inode(inode_t *inode) + if (atomic_read(&inode->i_count) == 1) + inode->i_nlink = 0; + } ++#endif /* LINUX_VERSION_CODE < KERNEL_VERSION(2,6,26) */ + + + #if defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) +@@ -238,7 +240,7 @@ mini_fo_clear_inode(inode_t *inode) + * dies. + */ + STATIC void +-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18) ++#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,18)) && (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,26)) + mini_fo_umount_begin(struct vfsmount *mnt, int flags) + { + struct vfsmount *hidden_mnt; +@@ -290,7 +292,9 @@ struct super_operations mini_fo_sops = + #if defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) + write_inode: mini_fo_write_inode, + #endif /* defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) */ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,26) + put_inode: mini_fo_put_inode, ++#endif /* LINUX_VERSION_CODE < KERNEL_VERSION(2,6,26) */ + #if defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) + delete_inode: mini_fo_delete_inode, + #endif /* defined(FIST_DEBUG) || defined(FIST_FILTER_SCA) */ diff --git a/target/linux/generic-2.6/patches-2.6.31/213-mini_fo_2.6.27_fixes.patch b/target/linux/generic-2.6/patches-2.6.31/213-mini_fo_2.6.27_fixes.patch new file mode 100644 index 0000000000..f92c18b254 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/213-mini_fo_2.6.27_fixes.patch @@ -0,0 +1,41 @@ +--- a/fs/mini_fo/inode.c ++++ b/fs/mini_fo/inode.c +@@ -439,7 +439,7 @@ mini_fo_symlink(inode_t *dir, dentry_t * + int err=0; + dentry_t *hidden_sto_dentry; + dentry_t *hidden_sto_dir_dentry; +-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0)) && (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,27)) + umode_t mode; + #endif + +@@ -466,7 +466,7 @@ mini_fo_symlink(inode_t *dir, dentry_t * + down(&hidden_sto_dir_dentry->d_inode->i_sem); + #endif + +-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0)) && (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,27)) + mode = S_IALLUGO; + err = vfs_symlink(hidden_sto_dir_dentry->d_inode, + hidden_sto_dentry, symname, mode); +@@ -1128,7 +1128,7 @@ void mini_fo_put_link(struct dentry *den + #endif + + STATIC int +-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0)) && (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,27)) + mini_fo_permission(inode_t *inode, int mask, struct nameidata *nd) + #else + mini_fo_permission(inode_t *inode, int mask) +@@ -1150,8 +1150,9 @@ mini_fo_permission(inode_t *inode, int m + * if (err) + * goto out; + */ +- +-#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) ++#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,27) ++ err = inode_permission(hidden_inode, mask); ++#elif LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) + err = permission(hidden_inode, mask, nd); + #else + err = permission(hidden_inode, mask); diff --git a/target/linux/generic-2.6/patches-2.6.31/214-mini_fo_2.6.29.patch b/target/linux/generic-2.6/patches-2.6.31/214-mini_fo_2.6.29.patch new file mode 100644 index 0000000000..63d704bf51 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/214-mini_fo_2.6.29.patch @@ -0,0 +1,96 @@ +--- a/fs/mini_fo/aux.c ++++ b/fs/mini_fo/aux.c +@@ -236,7 +236,7 @@ int mini_fo_cp_cont(dentry_t *tgt_dentry + mntget(src_mnt); + + /* open file write only */ +- tgt_file = dentry_open(tgt_dentry, tgt_mnt, 0x1); ++ tgt_file = dentry_open(tgt_dentry, tgt_mnt, 0x1, current_cred()); + if(!tgt_file || IS_ERR(tgt_file)) { + printk(KERN_CRIT "mini_fo_cp_cont: ERROR opening target file.\n"); + err = PTR_ERR(tgt_file); +@@ -244,7 +244,7 @@ int mini_fo_cp_cont(dentry_t *tgt_dentry + } + + /* open file read only */ +- src_file = dentry_open(src_dentry, src_mnt, 0x0); ++ src_file = dentry_open(src_dentry, src_mnt, 0x0, current_cred()); + if(!src_file || IS_ERR(src_file)) { + printk(KERN_CRIT "mini_fo_cp_cont: ERROR opening source file.\n"); + err = PTR_ERR(src_file); +--- a/fs/mini_fo/file.c ++++ b/fs/mini_fo/file.c +@@ -437,7 +437,7 @@ mini_fo_open(inode_t *inode, file_t *fil + mntget(stopd(inode->i_sb)->hidden_mnt); + hidden_file = dentry_open(hidden_dentry, + stopd(inode->i_sb)->hidden_mnt, +- hidden_flags); ++ hidden_flags, file->f_cred); + if (IS_ERR(hidden_file)) { + err = PTR_ERR(hidden_file); + dput(hidden_dentry); +@@ -479,7 +479,7 @@ mini_fo_open(inode_t *inode, file_t *fil + mntget(stopd(inode->i_sb)->hidden_mnt); + hidden_file = dentry_open(hidden_dentry, + stopd(inode->i_sb)->hidden_mnt, +- hidden_flags); ++ hidden_flags, file->f_cred); + if (IS_ERR(hidden_file)) { + err = PTR_ERR(hidden_file); + dput(hidden_dentry); +@@ -512,7 +512,7 @@ mini_fo_open(inode_t *inode, file_t *fil + mntget(stopd(inode->i_sb)->hidden_mnt2); + hidden_sto_file = dentry_open(hidden_sto_dentry, + stopd(inode->i_sb)->hidden_mnt2, +- hidden_flags); ++ hidden_flags, file->f_cred); + + /* dentry_open dputs the dentry if it fails */ + if (IS_ERR(hidden_sto_file)) { +--- a/fs/mini_fo/meta.c ++++ b/fs/mini_fo/meta.c +@@ -56,7 +56,7 @@ int meta_build_lists(dentry_t *dentry) + + + /* open META-file for reading */ +- meta_file = dentry_open(meta_dentry, meta_mnt, 0x0); ++ meta_file = dentry_open(meta_dentry, meta_mnt, 0x0, current_cred()); + if(!meta_file || IS_ERR(meta_file)) { + printk(KERN_CRIT "mini_fo: meta_build_lists: \ + ERROR opening META file.\n"); +@@ -448,7 +448,7 @@ int meta_write_d_entry(dentry_t *dentry, + mntget(meta_mnt); + + /* open META-file for writing */ +- meta_file = dentry_open(meta_dentry, meta_mnt, 0x1); ++ meta_file = dentry_open(meta_dentry, meta_mnt, 0x1, current_cred()); + if(!meta_file || IS_ERR(meta_file)) { + printk(KERN_CRIT "mini_fo: meta_write_d_entry: \ + ERROR opening meta file.\n"); +@@ -546,7 +546,7 @@ int meta_write_r_entry(dentry_t *dentry, + mntget(meta_mnt); + + /* open META-file for writing */ +- meta_file = dentry_open(meta_dentry, meta_mnt, 0x1); ++ meta_file = dentry_open(meta_dentry, meta_mnt, 0x1, current_cred()); + if(!meta_file || IS_ERR(meta_file)) { + printk(KERN_CRIT "mini_fo: meta_write_r_entry: \ + ERROR opening meta file.\n"); +@@ -686,7 +686,7 @@ int meta_sync_d_list(dentry_t *dentry, i + mntget(meta_mnt); + + /* open META-file for writing */ +- meta_file = dentry_open(meta_dentry, meta_mnt, 0x1); ++ meta_file = dentry_open(meta_dentry, meta_mnt, 0x1, current_cred()); + if(!meta_file || IS_ERR(meta_file)) { + printk(KERN_CRIT "mini_fo: meta_sync_d_list: \ + ERROR opening meta file.\n"); +@@ -828,7 +828,7 @@ int meta_sync_r_list(dentry_t *dentry, i + mntget(meta_mnt); + + /* open META-file for writing */ +- meta_file = dentry_open(meta_dentry, meta_mnt, 0x1); ++ meta_file = dentry_open(meta_dentry, meta_mnt, 0x1, current_cred()); + if(!meta_file || IS_ERR(meta_file)) { + printk(KERN_CRIT "mini_fo: meta_sync_r_list: \ + ERROR opening meta file.\n"); diff --git a/target/linux/generic-2.6/patches-2.6.31/215-mini_fo_2.6.30.patch b/target/linux/generic-2.6/patches-2.6.31/215-mini_fo_2.6.30.patch new file mode 100644 index 0000000000..2d20920614 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/215-mini_fo_2.6.30.patch @@ -0,0 +1,157 @@ +--- a/fs/mini_fo/aux.c ++++ b/fs/mini_fo/aux.c +@@ -86,8 +86,10 @@ int get_neg_sto_dentry(dentry_t *dentry) + len = dentry->d_name.len; + name = dentry->d_name.name; + ++ mutex_lock(&dtohd2(dentry->d_parent)->d_inode->i_mutex); + dtohd2(dentry) = + lookup_one_len(name, dtohd2(dentry->d_parent), len); ++ mutex_unlock(&dtohd2(dentry->d_parent)->d_inode->i_mutex); + + out: + return err; +@@ -426,7 +428,9 @@ int build_sto_structure(dentry_t *dir, d + const unsigned char *name; + len = dtohd(dentry)->d_name.len; + name = dtohd(dentry)->d_name.name; ++ mutex_lock(&dtohd2(dir)->d_inode->i_mutex); + hidden_sto_dentry = lookup_one_len(name, dtohd2(dir), len); ++ mutex_unlock(&dtohd2(dir)->d_inode->i_mutex); + dtohd2(dentry) = hidden_sto_dentry; + } + +--- a/fs/mini_fo/inode.c ++++ b/fs/mini_fo/inode.c +@@ -113,17 +113,23 @@ mini_fo_lookup(inode_t *dir, dentry_t *d + hidden_dir_dentry = hidden_dentry->d_parent; + kfree(bpath); + } +- else if(hidden_dir_dentry && hidden_dir_dentry->d_inode) ++ else if(hidden_dir_dentry && hidden_dir_dentry->d_inode) { ++ mutex_lock(&hidden_dir_dentry->d_inode->i_mutex); + hidden_dentry = + lookup_one_len(name, hidden_dir_dentry, namelen); +- else ++ mutex_unlock(&hidden_dir_dentry->d_inode->i_mutex); ++ } else { + hidden_dentry = NULL; ++ } + +- if(hidden_sto_dir_dentry && hidden_sto_dir_dentry->d_inode) ++ if(hidden_sto_dir_dentry && hidden_sto_dir_dentry->d_inode) { ++ mutex_lock(&hidden_sto_dir_dentry->d_inode->i_mutex); + hidden_sto_dentry = + lookup_one_len(name, hidden_sto_dir_dentry, namelen); +- else ++ mutex_unlock(&hidden_sto_dir_dentry->d_inode->i_mutex); ++ } else { + hidden_sto_dentry = NULL; ++ } + + /* catch error in lookup */ + if (IS_ERR(hidden_dentry) || IS_ERR(hidden_sto_dentry)) { +@@ -553,9 +559,11 @@ mini_fo_rmdir(inode_t *dir, dentry_t *de + #endif + + /* Delete an old WOL file contained in the storage dir */ ++ mutex_lock(&hidden_sto_dentry->d_inode->i_mutex); + meta_dentry = lookup_one_len(META_FILENAME, + hidden_sto_dentry, + strlen(META_FILENAME)); ++ mutex_unlock(&hidden_sto_dentry->d_inode->i_mutex); + if(meta_dentry->d_inode) { + err = vfs_unlink(hidden_sto_dentry->d_inode, meta_dentry); + dput(meta_dentry); +@@ -643,9 +651,11 @@ mini_fo_rmdir(inode_t *dir, dentry_t *de + #endif + + /* Delete an old WOL file contained in the storage dir */ ++ mutex_lock(&hidden_sto_dentry->d_inode->i_mutex); + meta_dentry = lookup_one_len(META_FILENAME, + hidden_sto_dentry, + strlen(META_FILENAME)); ++ mutex_unlock(&hidden_sto_dentry->d_inode->i_mutex); + if(meta_dentry->d_inode) { + /* is this necessary? dget(meta_dentry); */ + err = vfs_unlink(hidden_sto_dentry->d_inode, +@@ -688,9 +698,11 @@ mini_fo_rmdir(inode_t *dir, dentry_t *de + #endif + + /* Delete an old WOL file contained in the storage dir */ ++ mutex_lock(&hidden_sto_dentry->d_inode->i_mutex); + meta_dentry = lookup_one_len(META_FILENAME, + hidden_sto_dentry, + strlen(META_FILENAME)); ++ mutex_unlock(&hidden_sto_dentry->d_inode->i_mutex); + if(meta_dentry->d_inode) { + /* is this necessary? dget(meta_dentry); */ + err = vfs_unlink(hidden_sto_dentry->d_inode, +--- a/fs/mini_fo/meta.c ++++ b/fs/mini_fo/meta.c +@@ -43,9 +43,11 @@ int meta_build_lists(dentry_t *dentry) + + /* might there be a META-file? */ + if(dtohd2(dentry) && dtohd2(dentry)->d_inode) { ++ mutex_lock(&dtohd2(dentry)->d_inode->i_mutex); + meta_dentry = lookup_one_len(META_FILENAME, + dtohd2(dentry), + strlen(META_FILENAME)); ++ mutex_unlock(&dtohd2(dentry)->d_inode->i_mutex); + if(!meta_dentry->d_inode) { + dput(meta_dentry); + goto out_ok; +@@ -426,8 +428,11 @@ int meta_write_d_entry(dentry_t *dentry, + goto out; + } + } ++ ++ mutex_lock(&dtohd2(dentry)->d_inode->i_mutex); + meta_dentry = lookup_one_len(META_FILENAME, + dtohd2(dentry), strlen (META_FILENAME)); ++ mutex_unlock(&dtohd2(dentry)->d_inode->i_mutex); + + /* We need to create a META-file */ + if(!meta_dentry->d_inode) { +@@ -527,9 +532,13 @@ int meta_write_r_entry(dentry_t *dentry, + goto out; + } + } ++ ++ mutex_lock(&dtohd2(dentry)->d_inode->i_mutex); + meta_dentry = lookup_one_len(META_FILENAME, + dtohd2(dentry), + strlen (META_FILENAME)); ++ mutex_unlock(&dtohd2(dentry)->d_inode->i_mutex); ++ + if(!meta_dentry->d_inode) { + /* We need to create a META-file */ + #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) +@@ -641,9 +650,13 @@ int meta_sync_d_list(dentry_t *dentry, i + goto out; + } + } ++ ++ mutex_lock(&dtohd2(dentry)->d_inode->i_mutex); + meta_dentry = lookup_one_len(META_FILENAME, + dtohd2(dentry), + strlen(META_FILENAME)); ++ mutex_unlock(&dtohd2(dentry)->d_inode->i_mutex); ++ + if(!meta_dentry->d_inode) { + /* We need to create a META-file */ + #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) +@@ -784,9 +797,13 @@ int meta_sync_r_list(dentry_t *dentry, i + goto out; + } + } ++ ++ mutex_lock(&dtohd2(dentry)->d_inode->i_mutex); + meta_dentry = lookup_one_len(META_FILENAME, + dtohd2(dentry), + strlen(META_FILENAME)); ++ mutex_unlock(&dtohd2(dentry)->d_inode->i_mutex); ++ + if(!meta_dentry->d_inode) { + /* We need to create a META-file */ + #if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,0) diff --git a/target/linux/generic-2.6/patches-2.6.31/219-kobject_uevent.patch b/target/linux/generic-2.6/patches-2.6.31/219-kobject_uevent.patch new file mode 100644 index 0000000000..7e00b224cc --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/219-kobject_uevent.patch @@ -0,0 +1,42 @@ +--- a/lib/kobject_uevent.c ++++ b/lib/kobject_uevent.c +@@ -29,7 +29,8 @@ u64 uevent_seqnum; + char uevent_helper[UEVENT_HELPER_PATH_LEN] = CONFIG_UEVENT_HELPER_PATH; + static DEFINE_SPINLOCK(sequence_lock); + #if defined(CONFIG_NET) +-static struct sock *uevent_sock; ++struct sock *uevent_sock = NULL; ++EXPORT_SYMBOL_GPL(uevent_sock); + #endif + + /* the strings here must match the enum in include/linux/kobject.h */ +@@ -42,6 +43,18 @@ static const char *kobject_actions[] = { + [KOBJ_OFFLINE] = "offline", + }; + ++u64 uevent_next_seqnum(void) ++{ ++ u64 seq; ++ ++ spin_lock(&sequence_lock); ++ seq = ++uevent_seqnum; ++ spin_unlock(&sequence_lock); ++ ++ return seq; ++} ++EXPORT_SYMBOL_GPL(uevent_next_seqnum); ++ + /** + * kobject_action_type - translate action string to numeric type + * +@@ -201,9 +214,7 @@ int kobject_uevent_env(struct kobject *k + kobj->state_remove_uevent_sent = 1; + + /* we will send an event, so request a new sequence number */ +- spin_lock(&sequence_lock); +- seq = ++uevent_seqnum; +- spin_unlock(&sequence_lock); ++ seq = uevent_next_seqnum(); + retval = add_uevent_var(env, "SEQNUM=%llu", (unsigned long long)seq); + if (retval) + goto exit; diff --git a/target/linux/generic-2.6/patches-2.6.31/220-sound_kconfig.patch b/target/linux/generic-2.6/patches-2.6.31/220-sound_kconfig.patch new file mode 100644 index 0000000000..f374009a65 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/220-sound_kconfig.patch @@ -0,0 +1,11 @@ +--- a/sound/core/Kconfig ++++ b/sound/core/Kconfig +@@ -7,7 +7,7 @@ config SND_PCM + select SND_TIMER + + config SND_HWDEP +- tristate ++ tristate "Sound hardware support" + + config SND_RAWMIDI + tristate diff --git a/target/linux/generic-2.6/patches-2.6.31/221-binfmt_elf_gcc4.1.patch b/target/linux/generic-2.6/patches-2.6.31/221-binfmt_elf_gcc4.1.patch new file mode 100644 index 0000000000..fa891e0f3d --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/221-binfmt_elf_gcc4.1.patch @@ -0,0 +1,11 @@ +--- a/fs/binfmt_elf.c ++++ b/fs/binfmt_elf.c +@@ -1193,7 +1193,7 @@ static unsigned long vma_dump_size(struc + if (FILTER(ELF_HEADERS) && + vma->vm_pgoff == 0 && (vma->vm_flags & VM_READ)) { + u32 __user *header = (u32 __user *) vma->vm_start; +- u32 word; ++ u32 word = 0; + mm_segment_t fs = get_fs(); + /* + * Doing it this way gets the constant folded by GCC. diff --git a/target/linux/generic-2.6/patches-2.6.31/240-packet_socket_type.patch b/target/linux/generic-2.6/patches-2.6.31/240-packet_socket_type.patch new file mode 100644 index 0000000000..92553de0f4 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/240-packet_socket_type.patch @@ -0,0 +1,132 @@ +This patch allows the user to specify desired packet types (outgoing, +broadcast, unicast, etc.) on packet sockets via setsockopt. +This can reduce the load in situations where only a limited number +of packet types are necessary + +Signed-off-by: Felix Fietkau + +--- a/include/linux/if_packet.h ++++ b/include/linux/if_packet.h +@@ -31,6 +31,8 @@ struct sockaddr_ll + /* These ones are invisible by user level */ + #define PACKET_LOOPBACK 5 /* MC/BRD frame looped back */ + #define PACKET_FASTROUTE 6 /* Fastrouted frame */ ++#define PACKET_MASK_ANY 0xffffffff /* mask for packet type bits */ ++ + + /* Packet socket options */ + +@@ -48,6 +50,7 @@ struct sockaddr_ll + #define PACKET_RESERVE 12 + #define PACKET_TX_RING 13 + #define PACKET_LOSS 14 ++#define PACKET_RECV_TYPE 15 + + struct tpacket_stats + { +--- a/net/packet/af_packet.c ++++ b/net/packet/af_packet.c +@@ -206,6 +206,7 @@ struct packet_sock { + unsigned int tp_reserve; + unsigned int tp_loss:1; + #endif ++ unsigned int pkt_type; + }; + + struct packet_skb_cb { +@@ -343,6 +344,7 @@ static int packet_rcv_spkt(struct sk_buf + { + struct sock *sk; + struct sockaddr_pkt *spkt; ++ struct packet_sock *po; + + /* + * When we registered the protocol we saved the socket in the data +@@ -350,6 +352,7 @@ static int packet_rcv_spkt(struct sk_buf + */ + + sk = pt->af_packet_priv; ++ po = pkt_sk(sk); + + /* + * Yank back the headers [hope the device set this +@@ -362,7 +365,7 @@ static int packet_rcv_spkt(struct sk_buf + * so that this procedure is noop. + */ + +- if (skb->pkt_type == PACKET_LOOPBACK) ++ if (!(po->pkt_type & (1 << skb->pkt_type))) + goto out; + + if (dev_net(dev) != sock_net(sk)) +@@ -546,12 +549,12 @@ static int packet_rcv(struct sk_buff *sk + int skb_len = skb->len; + unsigned int snaplen, res; + +- if (skb->pkt_type == PACKET_LOOPBACK) +- goto drop; +- + sk = pt->af_packet_priv; + po = pkt_sk(sk); + ++ if (!(po->pkt_type & (1 << skb->pkt_type))) ++ goto drop; ++ + if (dev_net(dev) != sock_net(sk)) + goto drop; + +@@ -667,12 +670,12 @@ static int tpacket_rcv(struct sk_buff *s + struct timeval tv; + struct timespec ts; + +- if (skb->pkt_type == PACKET_LOOPBACK) +- goto drop; +- + sk = pt->af_packet_priv; + po = pkt_sk(sk); + ++ if (!(po->pkt_type & (1 << skb->pkt_type))) ++ goto drop; ++ + if (dev_net(dev) != sock_net(sk)) + goto drop; + +@@ -1387,6 +1390,7 @@ static int packet_create(struct net *net + spin_lock_init(&po->bind_lock); + mutex_init(&po->pg_vec_lock); + po->prot_hook.func = packet_rcv; ++ po->pkt_type = PACKET_MASK_ANY & ~(1 << PACKET_LOOPBACK); + + if (sock->type == SOCK_PACKET) + po->prot_hook.func = packet_rcv_spkt; +@@ -1733,6 +1737,16 @@ packet_setsockopt(struct socket *sock, i + ret = packet_mc_drop(sk, &mreq); + return ret; + } ++ case PACKET_RECV_TYPE: ++ { ++ unsigned int val; ++ if (optlen != sizeof(val)) ++ return -EINVAL; ++ if (copy_from_user(&val, optval, sizeof(val))) ++ return -EFAULT; ++ po->pkt_type = val & ~PACKET_LOOPBACK; ++ return 0; ++ } + + #ifdef CONFIG_PACKET_MMAP + case PACKET_RX_RING: +@@ -1878,6 +1892,13 @@ static int packet_getsockopt(struct sock + + data = &val; + break; ++ case PACKET_RECV_TYPE: ++ if (len > sizeof(unsigned int)) ++ len = sizeof(unsigned int); ++ val = po->pkt_type; ++ ++ data = &val; ++ break; + #ifdef CONFIG_PACKET_MMAP + case PACKET_VERSION: + if (len > sizeof(int)) diff --git a/target/linux/generic-2.6/patches-2.6.31/250-pppoe_header_pad.patch b/target/linux/generic-2.6/patches-2.6.31/250-pppoe_header_pad.patch new file mode 100644 index 0000000000..042546f317 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/250-pppoe_header_pad.patch @@ -0,0 +1,20 @@ +--- a/drivers/net/pppoe.c ++++ b/drivers/net/pppoe.c +@@ -863,7 +863,7 @@ static int pppoe_sendmsg(struct kiocb *i + goto end; + + +- skb = sock_wmalloc(sk, total_len + dev->hard_header_len + 32, ++ skb = sock_wmalloc(sk, total_len + dev->hard_header_len + 32 + NET_SKB_PAD, + 0, GFP_KERNEL); + if (!skb) { + error = -ENOMEM; +@@ -871,7 +871,7 @@ static int pppoe_sendmsg(struct kiocb *i + } + + /* Reserve space for headers. */ +- skb_reserve(skb, dev->hard_header_len); ++ skb_reserve(skb, dev->hard_header_len + NET_SKB_PAD); + skb_reset_network_header(skb); + + skb->dev = dev; diff --git a/target/linux/generic-2.6/patches-2.6.31/260-extend_pfifo_fast.patch b/target/linux/generic-2.6/patches-2.6.31/260-extend_pfifo_fast.patch new file mode 100644 index 0000000000..f306e861ff --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/260-extend_pfifo_fast.patch @@ -0,0 +1,144 @@ +--- a/net/sched/sch_generic.c ++++ b/net/sched/sch_generic.c +@@ -391,16 +391,50 @@ static const u8 prio2band[TC_PRIO_MAX+1] + + #define PFIFO_FAST_BANDS 3 + ++struct pfifo_fast_sched_data { ++ struct tcf_proto *filter_list; ++ struct sk_buff_head list[PFIFO_FAST_BANDS]; ++}; ++ + static inline struct sk_buff_head *prio2list(struct sk_buff *skb, + struct Qdisc *qdisc) + { +- struct sk_buff_head *list = qdisc_priv(qdisc); ++ struct pfifo_fast_sched_data *q = qdisc_priv(qdisc); ++ struct sk_buff_head *list = q->list; + return list + prio2band[skb->priority & TC_PRIO_MAX]; + } + ++static int pfifo_fast_filter(struct sk_buff *skb, struct Qdisc* qdisc) ++{ ++#ifdef CONFIG_NET_CLS_ACT ++ struct pfifo_fast_sched_data *q = qdisc_priv(qdisc); ++ int result = 0, ret = NET_XMIT_SUCCESS | __NET_XMIT_STOLEN; ++ struct tcf_result res; ++ ++ if (q->filter_list != NULL) ++ result = tc_classify(skb, q->filter_list, &res); ++ if (result >= 0) { ++ switch (result) { ++ case TC_ACT_STOLEN: ++ case TC_ACT_QUEUED: ++ ret = NET_XMIT_SUCCESS | __NET_XMIT_STOLEN; ++ case TC_ACT_SHOT: ++ kfree_skb(skb); ++ return ret; ++ } ++ } ++#endif ++ return 0; ++} ++ + static int pfifo_fast_enqueue(struct sk_buff *skb, struct Qdisc* qdisc) + { + struct sk_buff_head *list = prio2list(skb, qdisc); ++ int ret; ++ ++ ret = pfifo_fast_filter(skb, qdisc); ++ if (ret) ++ return ret; + + if (skb_queue_len(list) < qdisc_dev(qdisc)->tx_queue_len) { + qdisc->q.qlen++; +@@ -412,8 +446,9 @@ static int pfifo_fast_enqueue(struct sk_ + + static struct sk_buff *pfifo_fast_dequeue(struct Qdisc* qdisc) + { ++ struct pfifo_fast_sched_data *q = qdisc_priv(qdisc); ++ struct sk_buff_head *list = q->list; + int prio; +- struct sk_buff_head *list = qdisc_priv(qdisc); + + for (prio = 0; prio < PFIFO_FAST_BANDS; prio++) { + if (!skb_queue_empty(list + prio)) { +@@ -440,8 +475,9 @@ static struct sk_buff *pfifo_fast_peek(s + + static void pfifo_fast_reset(struct Qdisc* qdisc) + { ++ struct pfifo_fast_sched_data *q = qdisc_priv(qdisc); ++ struct sk_buff_head *list = q->list; + int prio; +- struct sk_buff_head *list = qdisc_priv(qdisc); + + for (prio = 0; prio < PFIFO_FAST_BANDS; prio++) + __qdisc_reset_queue(qdisc, list + prio); +@@ -464,8 +500,9 @@ nla_put_failure: + + static int pfifo_fast_init(struct Qdisc *qdisc, struct nlattr *opt) + { ++ struct pfifo_fast_sched_data *q = qdisc_priv(qdisc); ++ struct sk_buff_head *list = q->list; + int prio; +- struct sk_buff_head *list = qdisc_priv(qdisc); + + for (prio = 0; prio < PFIFO_FAST_BANDS; prio++) + skb_queue_head_init(list + prio); +@@ -473,9 +510,36 @@ static int pfifo_fast_init(struct Qdisc + return 0; + } + ++static int pfifo_fast_change_class(struct Qdisc *qdisc, u32 classid, u32 parentid, ++ struct nlattr **tca, unsigned long *arg) ++{ ++ return -EOPNOTSUPP; ++} ++ ++static unsigned long pfifo_fast_get(struct Qdisc *qdisc, u32 classid) ++{ ++ return 0; ++} ++ ++static struct tcf_proto **pfifo_fast_find_tcf(struct Qdisc *qdisc, unsigned long cl) ++{ ++ struct pfifo_fast_sched_data *q = qdisc_priv(qdisc); ++ ++ if (cl) ++ return NULL; ++ return &q->filter_list; ++} ++ ++static const struct Qdisc_class_ops pfifo_fast_class_ops = { ++ .get = pfifo_fast_get, ++ .change = pfifo_fast_change_class, ++ .tcf_chain = pfifo_fast_find_tcf, ++}; ++ + static struct Qdisc_ops pfifo_fast_ops __read_mostly = { + .id = "pfifo_fast", +- .priv_size = PFIFO_FAST_BANDS * sizeof(struct sk_buff_head), ++ .cl_ops = &pfifo_fast_class_ops, ++ .priv_size = sizeof(struct pfifo_fast_sched_data), + .enqueue = pfifo_fast_enqueue, + .dequeue = pfifo_fast_dequeue, + .peek = pfifo_fast_peek, +@@ -757,3 +821,18 @@ void dev_shutdown(struct net_device *dev + shutdown_scheduler_queue(dev, &dev->rx_queue, &noop_qdisc); + WARN_ON(timer_pending(&dev->watchdog_timer)); + } ++ ++#ifdef CONFIG_NET_SCHED ++static int __init sch_generic_init(void) ++{ ++ return register_qdisc(&pfifo_fast_ops); ++} ++ ++static void __exit sch_generic_exit(void) ++{ ++ unregister_qdisc(&pfifo_fast_ops); ++} ++ ++module_init(sch_generic_init) ++module_exit(sch_generic_exit) ++#endif diff --git a/target/linux/generic-2.6/patches-2.6.31/400-ledtrig_morse.patch b/target/linux/generic-2.6/patches-2.6.31/400-ledtrig_morse.patch new file mode 100644 index 0000000000..18c504f3d1 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/400-ledtrig_morse.patch @@ -0,0 +1,18 @@ +--- a/drivers/leds/Kconfig ++++ b/drivers/leds/Kconfig +@@ -297,4 +297,8 @@ config LEDS_TRIGGER_DEFAULT_ON + comment "iptables trigger is under Netfilter config (LED target)" + depends on LEDS_TRIGGERS + ++config LEDS_TRIGGER_MORSE ++ tristate "LED Morse Trigger" ++ depends on LEDS_TRIGGERS ++ + endif # NEW_LEDS +--- a/drivers/leds/Makefile ++++ b/drivers/leds/Makefile +@@ -39,3 +39,4 @@ obj-$(CONFIG_LEDS_TRIGGER_HEARTBEAT) += + obj-$(CONFIG_LEDS_TRIGGER_BACKLIGHT) += ledtrig-backlight.o + obj-$(CONFIG_LEDS_TRIGGER_GPIO) += ledtrig-gpio.o + obj-$(CONFIG_LEDS_TRIGGER_DEFAULT_ON) += ledtrig-default-on.o ++obj-$(CONFIG_LEDS_TRIGGER_MORSE) += ledtrig-morse.o diff --git a/target/linux/generic-2.6/patches-2.6.31/402-ledtrig_netdev.patch b/target/linux/generic-2.6/patches-2.6.31/402-ledtrig_netdev.patch new file mode 100644 index 0000000000..16d80030f8 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/402-ledtrig_netdev.patch @@ -0,0 +1,21 @@ +--- a/drivers/leds/Kconfig ++++ b/drivers/leds/Kconfig +@@ -301,4 +301,11 @@ config LEDS_TRIGGER_MORSE + tristate "LED Morse Trigger" + depends on LEDS_TRIGGERS + ++config LEDS_TRIGGER_NETDEV ++ tristate "LED Netdev Trigger" ++ depends on LEDS_TRIGGERS ++ help ++ This allows LEDs to be controlled by network device activity. ++ If unsure, say Y. ++ + endif # NEW_LEDS +--- a/drivers/leds/Makefile ++++ b/drivers/leds/Makefile +@@ -40,3 +40,4 @@ obj-$(CONFIG_LEDS_TRIGGER_BACKLIGHT) += + obj-$(CONFIG_LEDS_TRIGGER_GPIO) += ledtrig-gpio.o + obj-$(CONFIG_LEDS_TRIGGER_DEFAULT_ON) += ledtrig-default-on.o + obj-$(CONFIG_LEDS_TRIGGER_MORSE) += ledtrig-morse.o ++obj-$(CONFIG_LEDS_TRIGGER_NETDEV) += ledtrig-netdev.o diff --git a/target/linux/generic-2.6/patches-2.6.31/410-gpio_buttons.patch b/target/linux/generic-2.6/patches-2.6.31/410-gpio_buttons.patch new file mode 100644 index 0000000000..688f6fe10f --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/410-gpio_buttons.patch @@ -0,0 +1,31 @@ +--- a/drivers/input/misc/Kconfig ++++ b/drivers/input/misc/Kconfig +@@ -269,4 +269,21 @@ config INPUT_DM355EVM + + To compile this driver as a module, choose M here: the + module will be called dm355evm_keys. ++ ++config INPUT_GPIO_BUTTONS ++ tristate "Polled GPIO buttons interface" ++ depends on GENERIC_GPIO ++ select INPUT_POLLDEV ++ help ++ This driver implements support for buttons connected ++ to GPIO pins of various CPUs (and some other chips). ++ ++ Say Y here if your device has buttons connected ++ directly to such GPIO pins. Your board-specific ++ setup logic must also provide a platform device, ++ with configuration data saying which GPIOs are used. ++ ++ To compile this driver as a module, choose M here: the ++ module will be called gpio-buttons. ++ + endif +--- a/drivers/input/misc/Makefile ++++ b/drivers/input/misc/Makefile +@@ -26,3 +26,4 @@ obj-$(CONFIG_INPUT_TWL4030_PWRBUTTON) += + obj-$(CONFIG_INPUT_UINPUT) += uinput.o + obj-$(CONFIG_INPUT_WISTRON_BTNS) += wistron_btns.o + obj-$(CONFIG_INPUT_YEALINK) += yealink.o ++obj-$(CONFIG_INPUT_GPIO_BUTTONS) += gpio_buttons.o diff --git a/target/linux/generic-2.6/patches-2.6.31/420-gpiodev.patch b/target/linux/generic-2.6/patches-2.6.31/420-gpiodev.patch new file mode 100644 index 0000000000..5a364d69ad --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/420-gpiodev.patch @@ -0,0 +1,27 @@ +--- a/drivers/char/Kconfig ++++ b/drivers/char/Kconfig +@@ -1029,6 +1029,14 @@ config CS5535_GPIO + + If compiled as a module, it will be called cs5535_gpio. + ++config GPIO_DEVICE ++ tristate "GPIO device support" ++ depends on GENERIC_GPIO ++ help ++ Say Y to enable Linux GPIO device support. This allows control of ++ GPIO pins using a character device ++ ++ + config RAW_DRIVER + tristate "RAW driver (/dev/raw/rawN)" + depends on BLOCK +--- a/drivers/char/Makefile ++++ b/drivers/char/Makefile +@@ -95,6 +95,7 @@ obj-$(CONFIG_SCx200_GPIO) += scx200_gpio + obj-$(CONFIG_PC8736x_GPIO) += pc8736x_gpio.o + obj-$(CONFIG_NSC_GPIO) += nsc_gpio.o + obj-$(CONFIG_CS5535_GPIO) += cs5535_gpio.o ++obj-$(CONFIG_GPIO_DEVICE) += gpio_dev.o + obj-$(CONFIG_GPIO_TB0219) += tb0219.o + obj-$(CONFIG_TELCLOCK) += tlclk.o + diff --git a/target/linux/generic-2.6/patches-2.6.31/510-yaffs_support.patch b/target/linux/generic-2.6/patches-2.6.31/510-yaffs_support.patch new file mode 100644 index 0000000000..08a1efc29c --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/510-yaffs_support.patch @@ -0,0 +1,17 @@ +--- a/fs/Kconfig ++++ b/fs/Kconfig +@@ -188,6 +188,7 @@ source "fs/sysv/Kconfig" + source "fs/ufs/Kconfig" + source "fs/exofs/Kconfig" + source "fs/nilfs2/Kconfig" ++source "fs/yaffs2/Kconfig" + + endif # MISC_FILESYSTEMS + +--- a/fs/Makefile ++++ b/fs/Makefile +@@ -125,3 +125,4 @@ obj-$(CONFIG_OCFS2_FS) += ocfs2/ + obj-$(CONFIG_BTRFS_FS) += btrfs/ + obj-$(CONFIG_GFS2_FS) += gfs2/ + obj-$(CONFIG_EXOFS_FS) += exofs/ ++obj-$(CONFIG_YAFFS_FS) += yaffs2/ diff --git a/target/linux/generic-2.6/patches-2.6.31/511-yaffs-cvs-2009-04-24.patch b/target/linux/generic-2.6/patches-2.6.31/511-yaffs-cvs-2009-04-24.patch new file mode 100644 index 0000000000..5c70e79a05 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/511-yaffs-cvs-2009-04-24.patch @@ -0,0 +1,12344 @@ +--- a/fs/yaffs2/devextras.h ++++ b/fs/yaffs2/devextras.h +@@ -14,194 +14,135 @@ + */ + + /* +- * This file is just holds extra declarations used during development. +- * Most of these are from kernel includes placed here so we can use them in +- * applications. ++ * This file is just holds extra declarations of macros that would normally ++ * be providesd in the Linux kernel. These macros have been written from ++ * scratch but are functionally equivalent to the Linux ones. + * + */ + + #ifndef __EXTRAS_H__ + #define __EXTRAS_H__ + +-#if defined WIN32 +-#define __inline__ __inline +-#define new newHack +-#endif +- +-#if !(defined __KERNEL__) || (defined WIN32) + +-/* User space defines */ ++#if !(defined __KERNEL__) + ++/* Definition of types */ + typedef unsigned char __u8; + typedef unsigned short __u16; + typedef unsigned __u32; + ++#endif ++ + /* +- * Simple doubly linked list implementation. +- * +- * Some of the internal functions ("__xxx") are useful when +- * manipulating whole lists rather than single entries, as +- * sometimes we already know the next/prev entries and we can +- * generate better code by using them directly rather than +- * using the generic single-entry routines. ++ * This is a simple doubly linked list implementation that matches the ++ * way the Linux kernel doubly linked list implementation works. + */ + +-#define prefetch(x) 1 +- +-struct list_head { +- struct list_head *next, *prev; ++struct ylist_head { ++ struct ylist_head *next; /* next in chain */ ++ struct ylist_head *prev; /* previous in chain */ + }; + +-#define LIST_HEAD_INIT(name) { &(name), &(name) } + +-#define LIST_HEAD(name) \ +- struct list_head name = LIST_HEAD_INIT(name) ++/* Initialise a static list */ ++#define YLIST_HEAD(name) \ ++struct ylist_head name = { &(name), &(name)} ++ + +-#define INIT_LIST_HEAD(ptr) do { \ +- (ptr)->next = (ptr); (ptr)->prev = (ptr); \ ++ ++/* Initialise a list head to an empty list */ ++#define YINIT_LIST_HEAD(p) \ ++do { \ ++ (p)->next = (p);\ ++ (p)->prev = (p); \ + } while (0) + +-/* +- * Insert a new entry between two known consecutive entries. +- * +- * This is only for internal list manipulation where we know +- * the prev/next entries already! +- */ +-static __inline__ void __list_add(struct list_head *new, +- struct list_head *prev, +- struct list_head *next) +-{ +- next->prev = new; +- new->next = next; +- new->prev = prev; +- prev->next = new; +-} + +-/** +- * list_add - add a new entry +- * @new: new entry to be added +- * @head: list head to add it after +- * +- * Insert a new entry after the specified head. +- * This is good for implementing stacks. +- */ +-static __inline__ void list_add(struct list_head *new, struct list_head *head) ++/* Add an element to a list */ ++static __inline__ void ylist_add(struct ylist_head *newEntry, ++ struct ylist_head *list) + { +- __list_add(new, head, head->next); +-} ++ struct ylist_head *listNext = list->next; ++ ++ list->next = newEntry; ++ newEntry->prev = list; ++ newEntry->next = listNext; ++ listNext->prev = newEntry; + +-/** +- * list_add_tail - add a new entry +- * @new: new entry to be added +- * @head: list head to add it before +- * +- * Insert a new entry before the specified head. +- * This is useful for implementing queues. +- */ +-static __inline__ void list_add_tail(struct list_head *new, +- struct list_head *head) +-{ +- __list_add(new, head->prev, head); + } + +-/* +- * Delete a list entry by making the prev/next entries +- * point to each other. +- * +- * This is only for internal list manipulation where we know +- * the prev/next entries already! +- */ +-static __inline__ void __list_del(struct list_head *prev, +- struct list_head *next) ++static __inline__ void ylist_add_tail(struct ylist_head *newEntry, ++ struct ylist_head *list) + { +- next->prev = prev; +- prev->next = next; ++ struct ylist_head *listPrev = list->prev; ++ ++ list->prev = newEntry; ++ newEntry->next = list; ++ newEntry->prev = listPrev; ++ listPrev->next = newEntry; ++ + } + +-/** +- * list_del - deletes entry from list. +- * @entry: the element to delete from the list. +- * Note: list_empty on entry does not return true after this, the entry is +- * in an undefined state. +- */ +-static __inline__ void list_del(struct list_head *entry) ++ ++/* Take an element out of its current list, with or without ++ * reinitialising the links.of the entry*/ ++static __inline__ void ylist_del(struct ylist_head *entry) + { +- __list_del(entry->prev, entry->next); ++ struct ylist_head *listNext = entry->next; ++ struct ylist_head *listPrev = entry->prev; ++ ++ listNext->prev = listPrev; ++ listPrev->next = listNext; ++ + } + +-/** +- * list_del_init - deletes entry from list and reinitialize it. +- * @entry: the element to delete from the list. +- */ +-static __inline__ void list_del_init(struct list_head *entry) ++static __inline__ void ylist_del_init(struct ylist_head *entry) + { +- __list_del(entry->prev, entry->next); +- INIT_LIST_HEAD(entry); ++ ylist_del(entry); ++ entry->next = entry->prev = entry; + } + +-/** +- * list_empty - tests whether a list is empty +- * @head: the list to test. +- */ +-static __inline__ int list_empty(struct list_head *head) ++ ++/* Test if the list is empty */ ++static __inline__ int ylist_empty(struct ylist_head *entry) + { +- return head->next == head; ++ return (entry->next == entry); + } + +-/** +- * list_splice - join two lists +- * @list: the new list to add. +- * @head: the place to add it in the first list. ++ ++/* ylist_entry takes a pointer to a list entry and offsets it to that ++ * we can find a pointer to the object it is embedded in. + */ +-static __inline__ void list_splice(struct list_head *list, +- struct list_head *head) +-{ +- struct list_head *first = list->next; + +- if (first != list) { +- struct list_head *last = list->prev; +- struct list_head *at = head->next; +- +- first->prev = head; +- head->next = first; +- +- last->next = at; +- at->prev = last; +- } +-} + +-/** +- * list_entry - get the struct for this entry +- * @ptr: the &struct list_head pointer. +- * @type: the type of the struct this is embedded in. +- * @member: the name of the list_struct within the struct. +- */ +-#define list_entry(ptr, type, member) \ +- ((type *)((char *)(ptr)-(unsigned long)(&((type *)0)->member))) +- +-/** +- * list_for_each - iterate over a list +- * @pos: the &struct list_head to use as a loop counter. +- * @head: the head for your list. +- */ +-#define list_for_each(pos, head) \ +- for (pos = (head)->next, prefetch(pos->next); pos != (head); \ +- pos = pos->next, prefetch(pos->next)) +- +-/** +- * list_for_each_safe - iterate over a list safe against removal +- * of list entry +- * @pos: the &struct list_head to use as a loop counter. +- * @n: another &struct list_head to use as temporary storage +- * @head: the head for your list. +- */ +-#define list_for_each_safe(pos, n, head) \ +- for (pos = (head)->next, n = pos->next; pos != (head); \ +- pos = n, n = pos->next) ++#define ylist_entry(entry, type, member) \ ++ ((type *)((char *)(entry)-(unsigned long)(&((type *)NULL)->member))) + +-/* +- * File types ++ ++/* ylist_for_each and list_for_each_safe iterate over lists. ++ * ylist_for_each_safe uses temporary storage to make the list delete safe + */ ++ ++#define ylist_for_each(itervar, list) \ ++ for (itervar = (list)->next; itervar != (list); itervar = itervar->next) ++ ++#define ylist_for_each_safe(itervar, saveVar, list) \ ++ for (itervar = (list)->next, saveVar = (list)->next->next; \ ++ itervar != (list); itervar = saveVar, saveVar = saveVar->next) ++ ++ ++#if !(defined __KERNEL__) ++ ++ ++#ifndef WIN32 ++#include ++#endif ++ ++ ++#ifdef CONFIG_YAFFS_PROVIDE_DEFS ++/* File types */ ++ ++ + #define DT_UNKNOWN 0 + #define DT_FIFO 1 + #define DT_CHR 2 +@@ -212,6 +153,7 @@ static __inline__ void list_splice(struc + #define DT_SOCK 12 + #define DT_WHT 14 + ++ + #ifndef WIN32 + #include + #endif +@@ -227,10 +169,6 @@ static __inline__ void list_splice(struc + #define ATTR_ATIME 16 + #define ATTR_MTIME 32 + #define ATTR_CTIME 64 +-#define ATTR_ATIME_SET 128 +-#define ATTR_MTIME_SET 256 +-#define ATTR_FORCE 512 /* Not a change, but a change it */ +-#define ATTR_ATTR_FLAG 1024 + + struct iattr { + unsigned int ia_valid; +@@ -244,21 +182,15 @@ struct iattr { + unsigned int ia_attr_flags; + }; + +-#define KERN_DEBUG ++#endif + + #else + +-#ifndef WIN32 + #include +-#include + #include + #include +-#endif + + #endif + +-#if defined WIN32 +-#undef new +-#endif + + #endif +--- a/fs/yaffs2/Kconfig ++++ b/fs/yaffs2/Kconfig +@@ -5,7 +5,7 @@ + config YAFFS_FS + tristate "YAFFS2 file system support" + default n +- depends on MTD ++ depends on MTD_BLOCK + select YAFFS_YAFFS1 + select YAFFS_YAFFS2 + help +@@ -43,7 +43,8 @@ config YAFFS_9BYTE_TAGS + format that you need to continue to support. New data written + also uses the older-style format. Note: Use of this option + generally requires that MTD's oob layout be adjusted to use the +- older-style format. See notes on tags formats and MTD versions. ++ older-style format. See notes on tags formats and MTD versions ++ in yaffs_mtdif1.c. + + If unsure, say N. + +@@ -109,26 +110,6 @@ config YAFFS_DISABLE_LAZY_LOAD + + If unsure, say N. + +-config YAFFS_CHECKPOINT_RESERVED_BLOCKS +- int "Reserved blocks for checkpointing" +- depends on YAFFS_YAFFS2 +- default 10 +- help +- Give the number of Blocks to reserve for checkpointing. +- Checkpointing saves the state at unmount so that mounting is +- much faster as a scan of all the flash to regenerate this state +- is not needed. These Blocks are reserved per partition, so if +- you have very small partitions the default (10) may be a mess +- for you. You can set this value to 0, but that does not mean +- checkpointing is disabled at all. There only won't be any +- specially reserved blocks for checkpointing, so if there is +- enough free space on the filesystem, it will be used for +- checkpointing. +- +- If unsure, leave at default (10), but don't wonder if there are +- always 2MB used on your large page device partition (10 x 2k +- pagesize). When using small partitions or when being very small +- on space, you probably want to set this to zero. + + config YAFFS_DISABLE_WIDE_TNODES + bool "Turn off wide tnodes" +--- a/fs/yaffs2/Makefile ++++ b/fs/yaffs2/Makefile +@@ -5,7 +5,6 @@ + obj-$(CONFIG_YAFFS_FS) += yaffs.o + + yaffs-y := yaffs_ecc.o yaffs_fs.o yaffs_guts.o yaffs_checkptrw.o +-yaffs-y += yaffs_packedtags2.o yaffs_nand.o yaffs_qsort.o ++yaffs-y += yaffs_packedtags1.o yaffs_packedtags2.o yaffs_nand.o yaffs_qsort.o + yaffs-y += yaffs_tagscompat.o yaffs_tagsvalidity.o +-yaffs-y += yaffs_mtdif1.o yaffs_packedtags1.o +-yaffs-y += yaffs_mtdif.o yaffs_mtdif2.o ++yaffs-y += yaffs_mtdif.o yaffs_mtdif1.o yaffs_mtdif2.o +--- a/fs/yaffs2/moduleconfig.h ++++ b/fs/yaffs2/moduleconfig.h +@@ -27,12 +27,12 @@ + + /* Default: Not selected */ + /* Meaning: Yaffs does its own ECC, rather than using MTD ECC */ +-//#define CONFIG_YAFFS_DOES_ECC ++/* #define CONFIG_YAFFS_DOES_ECC */ + + /* Default: Not selected */ + /* Meaning: ECC byte order is 'wrong'. Only meaningful if */ + /* CONFIG_YAFFS_DOES_ECC is set */ +-//#define CONFIG_YAFFS_ECC_WRONG_ORDER ++/* #define CONFIG_YAFFS_ECC_WRONG_ORDER */ + + /* Default: Selected */ + /* Meaning: Disables testing whether chunks are erased before writing to them*/ +@@ -54,11 +54,11 @@ that you need to continue to support. N + older-style format. + Note: Use of this option generally requires that MTD's oob layout be + adjusted to use the older-style format. See notes on tags formats and +-MTD versions. ++MTD versions in yaffs_mtdif1.c. + */ + /* Default: Not selected */ + /* Meaning: Use older-style on-NAND data format with pageStatus byte */ +-#define CONFIG_YAFFS_9BYTE_TAGS ++/* #define CONFIG_YAFFS_9BYTE_TAGS */ + + #endif /* YAFFS_OUT_OF_TREE */ + +--- a/fs/yaffs2/yaffs_checkptrw.c ++++ b/fs/yaffs2/yaffs_checkptrw.c +@@ -12,48 +12,43 @@ + */ + + const char *yaffs_checkptrw_c_version = +- "$Id: yaffs_checkptrw.c,v 1.14 2007-05-15 20:07:40 charles Exp $"; ++ "$Id: yaffs_checkptrw.c,v 1.18 2009-03-06 17:20:49 wookey Exp $"; + + + #include "yaffs_checkptrw.h" +- ++#include "yaffs_getblockinfo.h" + + static int yaffs_CheckpointSpaceOk(yaffs_Device *dev) + { +- + int blocksAvailable = dev->nErasedBlocks - dev->nReservedBlocks; + + T(YAFFS_TRACE_CHECKPOINT, + (TSTR("checkpt blocks available = %d" TENDSTR), + blocksAvailable)); + +- + return (blocksAvailable <= 0) ? 0 : 1; + } + + + static int yaffs_CheckpointErase(yaffs_Device *dev) + { +- + int i; + +- +- if(!dev->eraseBlockInNAND) ++ if (!dev->eraseBlockInNAND) + return 0; +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("checking blocks %d to %d"TENDSTR), +- dev->internalStartBlock,dev->internalEndBlock)); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("checking blocks %d to %d"TENDSTR), ++ dev->internalStartBlock, dev->internalEndBlock)); + +- for(i = dev->internalStartBlock; i <= dev->internalEndBlock; i++) { +- yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev,i); +- if(bi->blockState == YAFFS_BLOCK_STATE_CHECKPOINT){ +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("erasing checkpt block %d"TENDSTR),i)); +- if(dev->eraseBlockInNAND(dev,i- dev->blockOffset /* realign */)){ ++ for (i = dev->internalStartBlock; i <= dev->internalEndBlock; i++) { ++ yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev, i); ++ if (bi->blockState == YAFFS_BLOCK_STATE_CHECKPOINT) { ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("erasing checkpt block %d"TENDSTR), i)); ++ if (dev->eraseBlockInNAND(dev, i - dev->blockOffset /* realign */)) { + bi->blockState = YAFFS_BLOCK_STATE_EMPTY; + dev->nErasedBlocks++; + dev->nFreeChunks += dev->nChunksPerBlock; +- } +- else { +- dev->markNANDBlockBad(dev,i); ++ } else { ++ dev->markNANDBlockBad(dev, i); + bi->blockState = YAFFS_BLOCK_STATE_DEAD; + } + } +@@ -71,23 +66,23 @@ static void yaffs_CheckpointFindNextEras + int blocksAvailable = dev->nErasedBlocks - dev->nReservedBlocks; + T(YAFFS_TRACE_CHECKPOINT, + (TSTR("allocating checkpt block: erased %d reserved %d avail %d next %d "TENDSTR), +- dev->nErasedBlocks,dev->nReservedBlocks,blocksAvailable,dev->checkpointNextBlock)); ++ dev->nErasedBlocks, dev->nReservedBlocks, blocksAvailable, dev->checkpointNextBlock)); + +- if(dev->checkpointNextBlock >= 0 && +- dev->checkpointNextBlock <= dev->internalEndBlock && +- blocksAvailable > 0){ +- +- for(i = dev->checkpointNextBlock; i <= dev->internalEndBlock; i++){ +- yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev,i); +- if(bi->blockState == YAFFS_BLOCK_STATE_EMPTY){ ++ if (dev->checkpointNextBlock >= 0 && ++ dev->checkpointNextBlock <= dev->internalEndBlock && ++ blocksAvailable > 0) { ++ ++ for (i = dev->checkpointNextBlock; i <= dev->internalEndBlock; i++) { ++ yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev, i); ++ if (bi->blockState == YAFFS_BLOCK_STATE_EMPTY) { + dev->checkpointNextBlock = i + 1; + dev->checkpointCurrentBlock = i; +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("allocating checkpt block %d"TENDSTR),i)); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("allocating checkpt block %d"TENDSTR), i)); + return; + } + } + } +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("out of checkpt blocks"TENDSTR))); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("out of checkpt blocks"TENDSTR))); + + dev->checkpointNextBlock = -1; + dev->checkpointCurrentBlock = -1; +@@ -98,30 +93,31 @@ static void yaffs_CheckpointFindNextChec + int i; + yaffs_ExtendedTags tags; + +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("find next checkpt block: start: blocks %d next %d" TENDSTR), ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("find next checkpt block: start: blocks %d next %d" TENDSTR), + dev->blocksInCheckpoint, dev->checkpointNextBlock)); + +- if(dev->blocksInCheckpoint < dev->checkpointMaxBlocks) +- for(i = dev->checkpointNextBlock; i <= dev->internalEndBlock; i++){ ++ if (dev->blocksInCheckpoint < dev->checkpointMaxBlocks) ++ for (i = dev->checkpointNextBlock; i <= dev->internalEndBlock; i++) { + int chunk = i * dev->nChunksPerBlock; + int realignedChunk = chunk - dev->chunkOffset; + +- dev->readChunkWithTagsFromNAND(dev,realignedChunk,NULL,&tags); +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("find next checkpt block: search: block %d oid %d seq %d eccr %d" TENDSTR), +- i, tags.objectId,tags.sequenceNumber,tags.eccResult)); ++ dev->readChunkWithTagsFromNAND(dev, realignedChunk, ++ NULL, &tags); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("find next checkpt block: search: block %d oid %d seq %d eccr %d" TENDSTR), ++ i, tags.objectId, tags.sequenceNumber, tags.eccResult)); + +- if(tags.sequenceNumber == YAFFS_SEQUENCE_CHECKPOINT_DATA){ ++ if (tags.sequenceNumber == YAFFS_SEQUENCE_CHECKPOINT_DATA) { + /* Right kind of block */ + dev->checkpointNextBlock = tags.objectId; + dev->checkpointCurrentBlock = i; + dev->checkpointBlockList[dev->blocksInCheckpoint] = i; + dev->blocksInCheckpoint++; +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("found checkpt block %d"TENDSTR),i)); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("found checkpt block %d"TENDSTR), i)); + return; + } + } + +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("found no more checkpt blocks"TENDSTR))); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("found no more checkpt blocks"TENDSTR))); + + dev->checkpointNextBlock = -1; + dev->checkpointCurrentBlock = -1; +@@ -133,17 +129,17 @@ int yaffs_CheckpointOpen(yaffs_Device *d + + /* Got the functions we need? */ + if (!dev->writeChunkWithTagsToNAND || +- !dev->readChunkWithTagsFromNAND || +- !dev->eraseBlockInNAND || +- !dev->markNANDBlockBad) ++ !dev->readChunkWithTagsFromNAND || ++ !dev->eraseBlockInNAND || ++ !dev->markNANDBlockBad) + return 0; + +- if(forWriting && !yaffs_CheckpointSpaceOk(dev)) ++ if (forWriting && !yaffs_CheckpointSpaceOk(dev)) + return 0; + +- if(!dev->checkpointBuffer) +- dev->checkpointBuffer = YMALLOC_DMA(dev->nDataBytesPerChunk); +- if(!dev->checkpointBuffer) ++ if (!dev->checkpointBuffer) ++ dev->checkpointBuffer = YMALLOC_DMA(dev->totalBytesPerChunk); ++ if (!dev->checkpointBuffer) + return 0; + + +@@ -159,12 +155,10 @@ int yaffs_CheckpointOpen(yaffs_Device *d + dev->checkpointNextBlock = dev->internalStartBlock; + + /* Erase all the blocks in the checkpoint area */ +- if(forWriting){ +- memset(dev->checkpointBuffer,0,dev->nDataBytesPerChunk); ++ if (forWriting) { ++ memset(dev->checkpointBuffer, 0, dev->nDataBytesPerChunk); + dev->checkpointByteOffset = 0; + return yaffs_CheckpointErase(dev); +- +- + } else { + int i; + /* Set to a value that will kick off a read */ +@@ -174,7 +168,7 @@ int yaffs_CheckpointOpen(yaffs_Device *d + dev->blocksInCheckpoint = 0; + dev->checkpointMaxBlocks = (dev->internalEndBlock - dev->internalStartBlock)/16 + 2; + dev->checkpointBlockList = YMALLOC(sizeof(int) * dev->checkpointMaxBlocks); +- for(i = 0; i < dev->checkpointMaxBlocks; i++) ++ for (i = 0; i < dev->checkpointMaxBlocks; i++) + dev->checkpointBlockList[i] = -1; + } + +@@ -191,18 +185,17 @@ int yaffs_GetCheckpointSum(yaffs_Device + + static int yaffs_CheckpointFlushBuffer(yaffs_Device *dev) + { +- + int chunk; + int realignedChunk; + + yaffs_ExtendedTags tags; + +- if(dev->checkpointCurrentBlock < 0){ ++ if (dev->checkpointCurrentBlock < 0) { + yaffs_CheckpointFindNextErasedBlock(dev); + dev->checkpointCurrentChunk = 0; + } + +- if(dev->checkpointCurrentBlock < 0) ++ if (dev->checkpointCurrentBlock < 0) + return 0; + + tags.chunkDeleted = 0; +@@ -210,10 +203,10 @@ static int yaffs_CheckpointFlushBuffer(y + tags.chunkId = dev->checkpointPageSequence + 1; + tags.sequenceNumber = YAFFS_SEQUENCE_CHECKPOINT_DATA; + tags.byteCount = dev->nDataBytesPerChunk; +- if(dev->checkpointCurrentChunk == 0){ ++ if (dev->checkpointCurrentChunk == 0) { + /* First chunk we write for the block? Set block state to + checkpoint */ +- yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev,dev->checkpointCurrentBlock); ++ yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev, dev->checkpointCurrentBlock); + bi->blockState = YAFFS_BLOCK_STATE_CHECKPOINT; + dev->blocksInCheckpoint++; + } +@@ -221,28 +214,29 @@ static int yaffs_CheckpointFlushBuffer(y + chunk = dev->checkpointCurrentBlock * dev->nChunksPerBlock + dev->checkpointCurrentChunk; + + +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("checkpoint wite buffer nand %d(%d:%d) objid %d chId %d" TENDSTR), +- chunk, dev->checkpointCurrentBlock, dev->checkpointCurrentChunk,tags.objectId,tags.chunkId)); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("checkpoint wite buffer nand %d(%d:%d) objid %d chId %d" TENDSTR), ++ chunk, dev->checkpointCurrentBlock, dev->checkpointCurrentChunk, tags.objectId, tags.chunkId)); + + realignedChunk = chunk - dev->chunkOffset; + +- dev->writeChunkWithTagsToNAND(dev,realignedChunk,dev->checkpointBuffer,&tags); ++ dev->writeChunkWithTagsToNAND(dev, realignedChunk, ++ dev->checkpointBuffer, &tags); + dev->checkpointByteOffset = 0; + dev->checkpointPageSequence++; + dev->checkpointCurrentChunk++; +- if(dev->checkpointCurrentChunk >= dev->nChunksPerBlock){ ++ if (dev->checkpointCurrentChunk >= dev->nChunksPerBlock) { + dev->checkpointCurrentChunk = 0; + dev->checkpointCurrentBlock = -1; + } +- memset(dev->checkpointBuffer,0,dev->nDataBytesPerChunk); ++ memset(dev->checkpointBuffer, 0, dev->nDataBytesPerChunk); + + return 1; + } + + +-int yaffs_CheckpointWrite(yaffs_Device *dev,const void *data, int nBytes) ++int yaffs_CheckpointWrite(yaffs_Device *dev, const void *data, int nBytes) + { +- int i=0; ++ int i = 0; + int ok = 1; + + +@@ -250,17 +244,14 @@ int yaffs_CheckpointWrite(yaffs_Device * + + + +- if(!dev->checkpointBuffer) ++ if (!dev->checkpointBuffer) + return 0; + +- if(!dev->checkpointOpenForWrite) ++ if (!dev->checkpointOpenForWrite) + return -1; + +- while(i < nBytes && ok) { +- +- +- +- dev->checkpointBuffer[dev->checkpointByteOffset] = *dataBytes ; ++ while (i < nBytes && ok) { ++ dev->checkpointBuffer[dev->checkpointByteOffset] = *dataBytes; + dev->checkpointSum += *dataBytes; + dev->checkpointXor ^= *dataBytes; + +@@ -270,18 +261,17 @@ int yaffs_CheckpointWrite(yaffs_Device * + dev->checkpointByteCount++; + + +- if(dev->checkpointByteOffset < 0 || ++ if (dev->checkpointByteOffset < 0 || + dev->checkpointByteOffset >= dev->nDataBytesPerChunk) + ok = yaffs_CheckpointFlushBuffer(dev); +- + } + +- return i; ++ return i; + } + + int yaffs_CheckpointRead(yaffs_Device *dev, void *data, int nBytes) + { +- int i=0; ++ int i = 0; + int ok = 1; + yaffs_ExtendedTags tags; + +@@ -291,52 +281,54 @@ int yaffs_CheckpointRead(yaffs_Device *d + + __u8 *dataBytes = (__u8 *)data; + +- if(!dev->checkpointBuffer) ++ if (!dev->checkpointBuffer) + return 0; + +- if(dev->checkpointOpenForWrite) ++ if (dev->checkpointOpenForWrite) + return -1; + +- while(i < nBytes && ok) { ++ while (i < nBytes && ok) { + + +- if(dev->checkpointByteOffset < 0 || +- dev->checkpointByteOffset >= dev->nDataBytesPerChunk) { ++ if (dev->checkpointByteOffset < 0 || ++ dev->checkpointByteOffset >= dev->nDataBytesPerChunk) { + +- if(dev->checkpointCurrentBlock < 0){ ++ if (dev->checkpointCurrentBlock < 0) { + yaffs_CheckpointFindNextCheckpointBlock(dev); + dev->checkpointCurrentChunk = 0; + } + +- if(dev->checkpointCurrentBlock < 0) ++ if (dev->checkpointCurrentBlock < 0) + ok = 0; + else { +- +- chunk = dev->checkpointCurrentBlock * dev->nChunksPerBlock + +- dev->checkpointCurrentChunk; ++ chunk = dev->checkpointCurrentBlock * ++ dev->nChunksPerBlock + ++ dev->checkpointCurrentChunk; + + realignedChunk = chunk - dev->chunkOffset; + +- /* read in the next chunk */ +- /* printf("read checkpoint page %d\n",dev->checkpointPage); */ +- dev->readChunkWithTagsFromNAND(dev, realignedChunk, +- dev->checkpointBuffer, +- &tags); +- +- if(tags.chunkId != (dev->checkpointPageSequence + 1) || +- tags.sequenceNumber != YAFFS_SEQUENCE_CHECKPOINT_DATA) +- ok = 0; ++ /* read in the next chunk */ ++ /* printf("read checkpoint page %d\n",dev->checkpointPage); */ ++ dev->readChunkWithTagsFromNAND(dev, ++ realignedChunk, ++ dev->checkpointBuffer, ++ &tags); ++ ++ if (tags.chunkId != (dev->checkpointPageSequence + 1) || ++ tags.eccResult > YAFFS_ECC_RESULT_FIXED || ++ tags.sequenceNumber != YAFFS_SEQUENCE_CHECKPOINT_DATA) ++ ok = 0; + + dev->checkpointByteOffset = 0; + dev->checkpointPageSequence++; + dev->checkpointCurrentChunk++; + +- if(dev->checkpointCurrentChunk >= dev->nChunksPerBlock) ++ if (dev->checkpointCurrentChunk >= dev->nChunksPerBlock) + dev->checkpointCurrentBlock = -1; + } + } + +- if(ok){ ++ if (ok) { + *dataBytes = dev->checkpointBuffer[dev->checkpointByteOffset]; + dev->checkpointSum += *dataBytes; + dev->checkpointXor ^= *dataBytes; +@@ -353,17 +345,17 @@ int yaffs_CheckpointRead(yaffs_Device *d + int yaffs_CheckpointClose(yaffs_Device *dev) + { + +- if(dev->checkpointOpenForWrite){ +- if(dev->checkpointByteOffset != 0) ++ if (dev->checkpointOpenForWrite) { ++ if (dev->checkpointByteOffset != 0) + yaffs_CheckpointFlushBuffer(dev); + } else { + int i; +- for(i = 0; i < dev->blocksInCheckpoint && dev->checkpointBlockList[i] >= 0; i++){ +- yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev,dev->checkpointBlockList[i]); +- if(bi->blockState == YAFFS_BLOCK_STATE_EMPTY) ++ for (i = 0; i < dev->blocksInCheckpoint && dev->checkpointBlockList[i] >= 0; i++) { ++ yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev, dev->checkpointBlockList[i]); ++ if (bi->blockState == YAFFS_BLOCK_STATE_EMPTY) + bi->blockState = YAFFS_BLOCK_STATE_CHECKPOINT; + else { +- // Todo this looks odd... ++ /* Todo this looks odd... */ + } + } + YFREE(dev->checkpointBlockList); +@@ -374,27 +366,25 @@ int yaffs_CheckpointClose(yaffs_Device * + dev->nErasedBlocks -= dev->blocksInCheckpoint; + + +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("checkpoint byte count %d" TENDSTR), ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("checkpoint byte count %d" TENDSTR), + dev->checkpointByteCount)); + +- if(dev->checkpointBuffer){ ++ if (dev->checkpointBuffer) { + /* free the buffer */ + YFREE(dev->checkpointBuffer); + dev->checkpointBuffer = NULL; + return 1; +- } +- else ++ } else + return 0; +- + } + + int yaffs_CheckpointInvalidateStream(yaffs_Device *dev) + { + /* Erase the first checksum block */ + +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("checkpoint invalidate"TENDSTR))); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("checkpoint invalidate"TENDSTR))); + +- if(!yaffs_CheckpointSpaceOk(dev)) ++ if (!yaffs_CheckpointSpaceOk(dev)) + return 0; + + return yaffs_CheckpointErase(dev); +--- a/fs/yaffs2/yaffs_checkptrw.h ++++ b/fs/yaffs2/yaffs_checkptrw.h +@@ -20,9 +20,9 @@ + + int yaffs_CheckpointOpen(yaffs_Device *dev, int forWriting); + +-int yaffs_CheckpointWrite(yaffs_Device *dev,const void *data, int nBytes); ++int yaffs_CheckpointWrite(yaffs_Device *dev, const void *data, int nBytes); + +-int yaffs_CheckpointRead(yaffs_Device *dev,void *data, int nBytes); ++int yaffs_CheckpointRead(yaffs_Device *dev, void *data, int nBytes); + + int yaffs_GetCheckpointSum(yaffs_Device *dev, __u32 *sum); + +--- a/fs/yaffs2/yaffs_ecc.c ++++ b/fs/yaffs2/yaffs_ecc.c +@@ -29,7 +29,7 @@ + */ + + const char *yaffs_ecc_c_version = +- "$Id: yaffs_ecc.c,v 1.9 2007-02-14 01:09:06 wookey Exp $"; ++ "$Id: yaffs_ecc.c,v 1.11 2009-03-06 17:20:50 wookey Exp $"; + + #include "yportenv.h" + +@@ -109,12 +109,10 @@ void yaffs_ECCCalculate(const unsigned c + b = column_parity_table[*data++]; + col_parity ^= b; + +- if (b & 0x01) // odd number of bits in the byte +- { ++ if (b & 0x01) { /* odd number of bits in the byte */ + line_parity ^= i; + line_parity_prime ^= ~i; + } +- + } + + ecc[2] = (~col_parity) | 0x03; +@@ -158,7 +156,7 @@ void yaffs_ECCCalculate(const unsigned c + ecc[0] = ~t; + + #ifdef CONFIG_YAFFS_ECC_WRONG_ORDER +- // Swap the bytes into the wrong order ++ /* Swap the bytes into the wrong order */ + t = ecc[0]; + ecc[0] = ecc[1]; + ecc[1] = t; +@@ -189,7 +187,7 @@ int yaffs_ECCCorrect(unsigned char *data + unsigned bit; + + #ifdef CONFIG_YAFFS_ECC_WRONG_ORDER +- // swap the bytes to correct for the wrong order ++ /* swap the bytes to correct for the wrong order */ + unsigned char t; + + t = d0; +@@ -251,7 +249,7 @@ int yaffs_ECCCorrect(unsigned char *data + * ECCxxxOther does ECC calcs on arbitrary n bytes of data + */ + void yaffs_ECCCalculateOther(const unsigned char *data, unsigned nBytes, +- yaffs_ECCOther * eccOther) ++ yaffs_ECCOther *eccOther) + { + unsigned int i; + +@@ -278,8 +276,8 @@ void yaffs_ECCCalculateOther(const unsig + } + + int yaffs_ECCCorrectOther(unsigned char *data, unsigned nBytes, +- yaffs_ECCOther * read_ecc, +- const yaffs_ECCOther * test_ecc) ++ yaffs_ECCOther *read_ecc, ++ const yaffs_ECCOther *test_ecc) + { + unsigned char cDelta; /* column parity delta */ + unsigned lDelta; /* line parity delta */ +@@ -294,8 +292,7 @@ int yaffs_ECCCorrectOther(unsigned char + return 0; /* no error */ + + if (lDelta == ~lDeltaPrime && +- (((cDelta ^ (cDelta >> 1)) & 0x15) == 0x15)) +- { ++ (((cDelta ^ (cDelta >> 1)) & 0x15) == 0x15)) { + /* Single bit (recoverable) error in data */ + + bit = 0; +@@ -307,7 +304,7 @@ int yaffs_ECCCorrectOther(unsigned char + if (cDelta & 0x02) + bit |= 0x01; + +- if(lDelta >= nBytes) ++ if (lDelta >= nBytes) + return -1; + + data[lDelta] ^= (1 << bit); +@@ -316,7 +313,7 @@ int yaffs_ECCCorrectOther(unsigned char + } + + if ((yaffs_CountBits32(lDelta) + yaffs_CountBits32(lDeltaPrime) + +- yaffs_CountBits(cDelta)) == 1) { ++ yaffs_CountBits(cDelta)) == 1) { + /* Reccoverable error in ecc */ + + *read_ecc = *test_ecc; +@@ -326,6 +323,4 @@ int yaffs_ECCCorrectOther(unsigned char + /* Unrecoverable error */ + + return -1; +- + } +- +--- a/fs/yaffs2/yaffs_ecc.h ++++ b/fs/yaffs2/yaffs_ecc.h +@@ -13,15 +13,15 @@ + * Note: Only YAFFS headers are LGPL, YAFFS C code is covered by GPL. + */ + +- /* +- * This code implements the ECC algorithm used in SmartMedia. +- * +- * The ECC comprises 22 bits of parity information and is stuffed into 3 bytes. +- * The two unused bit are set to 1. +- * The ECC can correct single bit errors in a 256-byte page of data. Thus, two such ECC +- * blocks are used on a 512-byte NAND page. +- * +- */ ++/* ++ * This code implements the ECC algorithm used in SmartMedia. ++ * ++ * The ECC comprises 22 bits of parity information and is stuffed into 3 bytes. ++ * The two unused bit are set to 1. ++ * The ECC can correct single bit errors in a 256-byte page of data. Thus, two such ECC ++ * blocks are used on a 512-byte NAND page. ++ * ++ */ + + #ifndef __YAFFS_ECC_H__ + #define __YAFFS_ECC_H__ +@@ -34,11 +34,11 @@ typedef struct { + + void yaffs_ECCCalculate(const unsigned char *data, unsigned char *ecc); + int yaffs_ECCCorrect(unsigned char *data, unsigned char *read_ecc, +- const unsigned char *test_ecc); ++ const unsigned char *test_ecc); + + void yaffs_ECCCalculateOther(const unsigned char *data, unsigned nBytes, +- yaffs_ECCOther * ecc); ++ yaffs_ECCOther *ecc); + int yaffs_ECCCorrectOther(unsigned char *data, unsigned nBytes, +- yaffs_ECCOther * read_ecc, +- const yaffs_ECCOther * test_ecc); ++ yaffs_ECCOther *read_ecc, ++ const yaffs_ECCOther *test_ecc); + #endif +--- a/fs/yaffs2/yaffs_fs.c ++++ b/fs/yaffs2/yaffs_fs.c +@@ -1,7 +1,7 @@ + /* + * YAFFS: Yet Another Flash File System. A NAND-flash specific file system. + * +- * Copyright (C) 2002-2007 Aleph One Ltd. ++ * Copyright (C) 2002-2009 Aleph One Ltd. + * for Toby Churchill Ltd and Brightstar Engineering + * + * Created by Charles Manning +@@ -32,18 +32,17 @@ + */ + + const char *yaffs_fs_c_version = +- "$Id: yaffs_fs.c,v 1.63 2007-09-19 20:35:40 imcd Exp $"; ++ "$Id: yaffs_fs.c,v 1.79 2009-03-17 01:12:00 wookey Exp $"; + extern const char *yaffs_guts_c_version; + + #include +-#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19)) ++#if (LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 19)) + #include + #endif + #include + #include + #include + #include +-#include + #include + #include + #include +@@ -53,10 +52,12 @@ extern const char *yaffs_guts_c_version; + #include + #include + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#include "asm/div64.h" ++ ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + + #include /* Added NCB 15-8-2003 */ +-#include ++#include + #define UnlockPage(p) unlock_page(p) + #define Page_Uptodate(page) test_bit(PG_uptodate, &(page)->flags) + +@@ -69,22 +70,45 @@ extern const char *yaffs_guts_c_version; + #define BDEVNAME_SIZE 0 + #define yaffs_devname(sb, buf) kdevname(sb->s_dev) + +-#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 0)) + /* added NCB 26/5/2006 for 2.4.25-vrs2-tcl1 kernel */ + #define __user + #endif + + #endif + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 26)) ++#define YPROC_ROOT (&proc_root) ++#else ++#define YPROC_ROOT NULL ++#endif ++ ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + #define WRITE_SIZE_STR "writesize" +-#define WRITE_SIZE(mtd) (mtd)->writesize ++#define WRITE_SIZE(mtd) ((mtd)->writesize) + #else + #define WRITE_SIZE_STR "oobblock" +-#define WRITE_SIZE(mtd) (mtd)->oobblock ++#define WRITE_SIZE(mtd) ((mtd)->oobblock) + #endif + +-#include ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 27)) ++#define YAFFS_USE_WRITE_BEGIN_END 1 ++#else ++#define YAFFS_USE_WRITE_BEGIN_END 0 ++#endif ++ ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 28)) ++static uint32_t YCALCBLOCKS(uint64_t partition_size, uint32_t block_size) ++{ ++ uint64_t result = partition_size; ++ do_div(result, block_size); ++ return (uint32_t)result; ++} ++#else ++#define YCALCBLOCKS(s, b) ((s)/(b)) ++#endif ++ ++#include + + #include "yportenv.h" + #include "yaffs_guts.h" +@@ -96,28 +120,44 @@ extern const char *yaffs_guts_c_version; + + unsigned int yaffs_traceMask = YAFFS_TRACE_BAD_BLOCKS; + unsigned int yaffs_wr_attempts = YAFFS_WR_ATTEMPTS; ++unsigned int yaffs_auto_checkpoint = 1; + + /* Module Parameters */ +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) +-module_param(yaffs_traceMask,uint,0644); +-module_param(yaffs_wr_attempts,uint,0644); ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) ++module_param(yaffs_traceMask, uint, 0644); ++module_param(yaffs_wr_attempts, uint, 0644); ++module_param(yaffs_auto_checkpoint, uint, 0644); ++#else ++MODULE_PARM(yaffs_traceMask, "i"); ++MODULE_PARM(yaffs_wr_attempts, "i"); ++MODULE_PARM(yaffs_auto_checkpoint, "i"); ++#endif ++ ++#if (LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 25)) ++/* use iget and read_inode */ ++#define Y_IGET(sb, inum) iget((sb), (inum)) ++static void yaffs_read_inode(struct inode *inode); ++ + #else +-MODULE_PARM(yaffs_traceMask,"i"); +-MODULE_PARM(yaffs_wr_attempts,"i"); ++/* Call local equivalent */ ++#define YAFFS_USE_OWN_IGET ++#define Y_IGET(sb, inum) yaffs_iget((sb), (inum)) ++ ++static struct inode *yaffs_iget(struct super_block *sb, unsigned long ino); + #endif + + /*#define T(x) printk x */ + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,18)) +-#define yaffs_InodeToObjectLV(iptr) (iptr)->i_private ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 18)) ++#define yaffs_InodeToObjectLV(iptr) ((iptr)->i_private) + #else +-#define yaffs_InodeToObjectLV(iptr) (iptr)->u.generic_ip ++#define yaffs_InodeToObjectLV(iptr) ((iptr)->u.generic_ip) + #endif + + #define yaffs_InodeToObject(iptr) ((yaffs_Object *)(yaffs_InodeToObjectLV(iptr))) + #define yaffs_DentryToObject(dptr) yaffs_InodeToObject((dptr)->d_inode) + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + #define yaffs_SuperToDevice(sb) ((yaffs_Device *)sb->s_fs_info) + #else + #define yaffs_SuperToDevice(sb) ((yaffs_Device *)sb->u.generic_sbp) +@@ -126,47 +166,49 @@ MODULE_PARM(yaffs_wr_attempts,"i"); + static void yaffs_put_super(struct super_block *sb); + + static ssize_t yaffs_file_write(struct file *f, const char *buf, size_t n, +- loff_t * pos); ++ loff_t *pos); ++static ssize_t yaffs_hold_space(struct file *f); ++static void yaffs_release_space(struct file *f); + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + static int yaffs_file_flush(struct file *file, fl_owner_t id); + #else + static int yaffs_file_flush(struct file *file); + #endif + + static int yaffs_sync_object(struct file *file, struct dentry *dentry, +- int datasync); ++ int datasync); + + static int yaffs_readdir(struct file *f, void *dirent, filldir_t filldir); + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + static int yaffs_create(struct inode *dir, struct dentry *dentry, int mode, + struct nameidata *n); + static struct dentry *yaffs_lookup(struct inode *dir, struct dentry *dentry, +- struct nameidata *n); ++ struct nameidata *n); + #else + static int yaffs_create(struct inode *dir, struct dentry *dentry, int mode); + static struct dentry *yaffs_lookup(struct inode *dir, struct dentry *dentry); + #endif + static int yaffs_link(struct dentry *old_dentry, struct inode *dir, +- struct dentry *dentry); ++ struct dentry *dentry); + static int yaffs_unlink(struct inode *dir, struct dentry *dentry); + static int yaffs_symlink(struct inode *dir, struct dentry *dentry, +- const char *symname); ++ const char *symname); + static int yaffs_mkdir(struct inode *dir, struct dentry *dentry, int mode); + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + static int yaffs_mknod(struct inode *dir, struct dentry *dentry, int mode, +- dev_t dev); ++ dev_t dev); + #else + static int yaffs_mknod(struct inode *dir, struct dentry *dentry, int mode, +- int dev); ++ int dev); + #endif + static int yaffs_rename(struct inode *old_dir, struct dentry *old_dentry, + struct inode *new_dir, struct dentry *new_dentry); + static int yaffs_setattr(struct dentry *dentry, struct iattr *attr); + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + static int yaffs_sync_fs(struct super_block *sb, int wait); + static void yaffs_write_super(struct super_block *sb); + #else +@@ -174,33 +216,47 @@ static int yaffs_sync_fs(struct super_bl + static int yaffs_write_super(struct super_block *sb); + #endif + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + static int yaffs_statfs(struct dentry *dentry, struct kstatfs *buf); +-#elif (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#elif (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + static int yaffs_statfs(struct super_block *sb, struct kstatfs *buf); + #else + static int yaffs_statfs(struct super_block *sb, struct statfs *buf); + #endif +-static void yaffs_read_inode(struct inode *inode); + ++#ifdef YAFFS_HAS_PUT_INODE + static void yaffs_put_inode(struct inode *inode); ++#endif ++ + static void yaffs_delete_inode(struct inode *); + static void yaffs_clear_inode(struct inode *); + + static int yaffs_readpage(struct file *file, struct page *page); +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + static int yaffs_writepage(struct page *page, struct writeback_control *wbc); + #else + static int yaffs_writepage(struct page *page); + #endif ++ ++ ++#if (YAFFS_USE_WRITE_BEGIN_END != 0) ++static int yaffs_write_begin(struct file *filp, struct address_space *mapping, ++ loff_t pos, unsigned len, unsigned flags, ++ struct page **pagep, void **fsdata); ++static int yaffs_write_end(struct file *filp, struct address_space *mapping, ++ loff_t pos, unsigned len, unsigned copied, ++ struct page *pg, void *fsdadata); ++#else + static int yaffs_prepare_write(struct file *f, struct page *pg, +- unsigned offset, unsigned to); ++ unsigned offset, unsigned to); + static int yaffs_commit_write(struct file *f, struct page *pg, unsigned offset, +- unsigned to); ++ unsigned to); + +-static int yaffs_readlink(struct dentry *dentry, char __user * buffer, +- int buflen); +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,13)) ++#endif ++ ++static int yaffs_readlink(struct dentry *dentry, char __user *buffer, ++ int buflen); ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 13)) + static void *yaffs_follow_link(struct dentry *dentry, struct nameidata *nd); + #else + static int yaffs_follow_link(struct dentry *dentry, struct nameidata *nd); +@@ -209,12 +265,17 @@ static int yaffs_follow_link(struct dent + static struct address_space_operations yaffs_file_address_operations = { + .readpage = yaffs_readpage, + .writepage = yaffs_writepage, ++#if (YAFFS_USE_WRITE_BEGIN_END > 0) ++ .write_begin = yaffs_write_begin, ++ .write_end = yaffs_write_end, ++#else + .prepare_write = yaffs_prepare_write, + .commit_write = yaffs_commit_write, ++#endif + }; + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,22)) +-static struct file_operations yaffs_file_operations = { ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 22)) ++static const struct file_operations yaffs_file_operations = { + .read = do_sync_read, + .write = do_sync_write, + .aio_read = generic_file_aio_read, +@@ -224,11 +285,12 @@ static struct file_operations yaffs_file + .fsync = yaffs_sync_object, + .splice_read = generic_file_splice_read, + .splice_write = generic_file_splice_write, ++ .llseek = generic_file_llseek, + }; + +-#elif (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,18)) ++#elif (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 18)) + +-static struct file_operations yaffs_file_operations = { ++static const struct file_operations yaffs_file_operations = { + .read = do_sync_read, + .write = do_sync_write, + .aio_read = generic_file_aio_read, +@@ -241,29 +303,29 @@ static struct file_operations yaffs_file + + #else + +-static struct file_operations yaffs_file_operations = { ++static const struct file_operations yaffs_file_operations = { + .read = generic_file_read, + .write = generic_file_write, + .mmap = generic_file_mmap, + .flush = yaffs_file_flush, + .fsync = yaffs_sync_object, +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + .sendfile = generic_file_sendfile, + #endif + }; + #endif + +-static struct inode_operations yaffs_file_inode_operations = { ++static const struct inode_operations yaffs_file_inode_operations = { + .setattr = yaffs_setattr, + }; + +-static struct inode_operations yaffs_symlink_inode_operations = { ++static const struct inode_operations yaffs_symlink_inode_operations = { + .readlink = yaffs_readlink, + .follow_link = yaffs_follow_link, + .setattr = yaffs_setattr, + }; + +-static struct inode_operations yaffs_dir_inode_operations = { ++static const struct inode_operations yaffs_dir_inode_operations = { + .create = yaffs_create, + .lookup = yaffs_lookup, + .link = yaffs_link, +@@ -276,16 +338,21 @@ static struct inode_operations yaffs_dir + .setattr = yaffs_setattr, + }; + +-static struct file_operations yaffs_dir_operations = { ++static const struct file_operations yaffs_dir_operations = { + .read = generic_read_dir, + .readdir = yaffs_readdir, + .fsync = yaffs_sync_object, + }; + +-static struct super_operations yaffs_super_ops = { ++static const struct super_operations yaffs_super_ops = { + .statfs = yaffs_statfs, ++ ++#ifndef YAFFS_USE_OWN_IGET + .read_inode = yaffs_read_inode, ++#endif ++#ifdef YAFFS_HAS_PUT_INODE + .put_inode = yaffs_put_inode, ++#endif + .put_super = yaffs_put_super, + .delete_inode = yaffs_delete_inode, + .clear_inode = yaffs_clear_inode, +@@ -293,22 +360,21 @@ static struct super_operations yaffs_sup + .write_super = yaffs_write_super, + }; + +-static void yaffs_GrossLock(yaffs_Device * dev) ++static void yaffs_GrossLock(yaffs_Device *dev) + { +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs locking\n")); +- ++ T(YAFFS_TRACE_OS, ("yaffs locking %p\n", current)); + down(&dev->grossLock); ++ T(YAFFS_TRACE_OS, ("yaffs locked %p\n", current)); + } + +-static void yaffs_GrossUnlock(yaffs_Device * dev) ++static void yaffs_GrossUnlock(yaffs_Device *dev) + { +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs unlocking\n")); ++ T(YAFFS_TRACE_OS, ("yaffs unlocking %p\n", current)); + up(&dev->grossLock); +- + } + +-static int yaffs_readlink(struct dentry *dentry, char __user * buffer, +- int buflen) ++static int yaffs_readlink(struct dentry *dentry, char __user *buffer, ++ int buflen) + { + unsigned char *alias; + int ret; +@@ -329,7 +395,7 @@ static int yaffs_readlink(struct dentry + return ret; + } + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,13)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 13)) + static void *yaffs_follow_link(struct dentry *dentry, struct nameidata *nd) + #else + static int yaffs_follow_link(struct dentry *dentry, struct nameidata *nd) +@@ -345,32 +411,31 @@ static int yaffs_follow_link(struct dent + + yaffs_GrossUnlock(dev); + +- if (!alias) +- { ++ if (!alias) { + ret = -ENOMEM; + goto out; +- } ++ } + + ret = vfs_follow_link(nd, alias); + kfree(alias); + out: +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,13)) +- return ERR_PTR (ret); ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 13)) ++ return ERR_PTR(ret); + #else + return ret; + #endif + } + + struct inode *yaffs_get_inode(struct super_block *sb, int mode, int dev, +- yaffs_Object * obj); ++ yaffs_Object *obj); + + /* + * Lookup is used to find objects in the fs + */ +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + + static struct dentry *yaffs_lookup(struct inode *dir, struct dentry *dentry, +- struct nameidata *n) ++ struct nameidata *n) + #else + static struct dentry *yaffs_lookup(struct inode *dir, struct dentry *dentry) + #endif +@@ -383,12 +448,11 @@ static struct dentry *yaffs_lookup(struc + yaffs_GrossLock(dev); + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_lookup for %d:%s\n", +- yaffs_InodeToObject(dir)->objectId, dentry->d_name.name)); ++ ("yaffs_lookup for %d:%s\n", ++ yaffs_InodeToObject(dir)->objectId, dentry->d_name.name)); + +- obj = +- yaffs_FindObjectByName(yaffs_InodeToObject(dir), +- dentry->d_name.name); ++ obj = yaffs_FindObjectByName(yaffs_InodeToObject(dir), ++ dentry->d_name.name); + + obj = yaffs_GetEquivalentObject(obj); /* in case it was a hardlink */ + +@@ -397,13 +461,13 @@ static struct dentry *yaffs_lookup(struc + + if (obj) { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_lookup found %d\n", obj->objectId)); ++ ("yaffs_lookup found %d\n", obj->objectId)); + + inode = yaffs_get_inode(dir->i_sb, obj->yst_mode, 0, obj); + + if (inode) { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_loookup dentry \n")); ++ ("yaffs_loookup dentry \n")); + /* #if 0 asserted by NCB for 2.5/6 compatability - falls through to + * d_add even if NULL inode */ + #if 0 +@@ -416,7 +480,7 @@ static struct dentry *yaffs_lookup(struc + } + + } else { +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_lookup not found\n")); ++ T(YAFFS_TRACE_OS, ("yaffs_lookup not found\n")); + + } + +@@ -425,20 +489,22 @@ static struct dentry *yaffs_lookup(struc + d_add(dentry, inode); + + return NULL; +- /* return (ERR_PTR(-EIO)); */ +- + } + ++ ++#ifdef YAFFS_HAS_PUT_INODE ++ + /* For now put inode is just for debugging + * Put inode is called when the inode **structure** is put. + */ + static void yaffs_put_inode(struct inode *inode) + { + T(YAFFS_TRACE_OS, +- ("yaffs_put_inode: ino %d, count %d\n", (int)inode->i_ino, +- atomic_read(&inode->i_count))); ++ ("yaffs_put_inode: ino %d, count %d\n", (int)inode->i_ino, ++ atomic_read(&inode->i_count))); + + } ++#endif + + /* clear is called to tell the fs to release any per-inode data it holds */ + static void yaffs_clear_inode(struct inode *inode) +@@ -449,9 +515,9 @@ static void yaffs_clear_inode(struct ino + obj = yaffs_InodeToObject(inode); + + T(YAFFS_TRACE_OS, +- ("yaffs_clear_inode: ino %d, count %d %s\n", (int)inode->i_ino, +- atomic_read(&inode->i_count), +- obj ? "object exists" : "null object")); ++ ("yaffs_clear_inode: ino %d, count %d %s\n", (int)inode->i_ino, ++ atomic_read(&inode->i_count), ++ obj ? "object exists" : "null object")); + + if (obj) { + dev = obj->myDev; +@@ -486,23 +552,23 @@ static void yaffs_delete_inode(struct in + yaffs_Device *dev; + + T(YAFFS_TRACE_OS, +- ("yaffs_delete_inode: ino %d, count %d %s\n", (int)inode->i_ino, +- atomic_read(&inode->i_count), +- obj ? "object exists" : "null object")); ++ ("yaffs_delete_inode: ino %d, count %d %s\n", (int)inode->i_ino, ++ atomic_read(&inode->i_count), ++ obj ? "object exists" : "null object")); + + if (obj) { + dev = obj->myDev; + yaffs_GrossLock(dev); +- yaffs_DeleteFile(obj); ++ yaffs_DeleteObject(obj); + yaffs_GrossUnlock(dev); + } +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,13)) +- truncate_inode_pages (&inode->i_data, 0); ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 13)) ++ truncate_inode_pages(&inode->i_data, 0); + #endif + clear_inode(inode); + } + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + static int yaffs_file_flush(struct file *file, fl_owner_t id) + #else + static int yaffs_file_flush(struct file *file) +@@ -513,8 +579,8 @@ static int yaffs_file_flush(struct file + yaffs_Device *dev = obj->myDev; + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_file_flush object %d (%s)\n", obj->objectId, +- obj->dirty ? "dirty" : "clean")); ++ ("yaffs_file_flush object %d (%s)\n", obj->objectId, ++ obj->dirty ? "dirty" : "clean")); + + yaffs_GrossLock(dev); + +@@ -535,15 +601,15 @@ static int yaffs_readpage_nolock(struct + + yaffs_Device *dev; + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_readpage at %08x, size %08x\n", +- (unsigned)(pg->index << PAGE_CACHE_SHIFT), +- (unsigned)PAGE_CACHE_SIZE)); ++ T(YAFFS_TRACE_OS, ("yaffs_readpage at %08x, size %08x\n", ++ (unsigned)(pg->index << PAGE_CACHE_SHIFT), ++ (unsigned)PAGE_CACHE_SIZE)); + + obj = yaffs_DentryToObject(f->f_dentry); + + dev = obj->myDev; + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + BUG_ON(!PageLocked(pg)); + #else + if (!PageLocked(pg)) +@@ -555,9 +621,9 @@ static int yaffs_readpage_nolock(struct + + yaffs_GrossLock(dev); + +- ret = +- yaffs_ReadDataFromFile(obj, pg_buf, pg->index << PAGE_CACHE_SHIFT, +- PAGE_CACHE_SIZE); ++ ret = yaffs_ReadDataFromFile(obj, pg_buf, ++ pg->index << PAGE_CACHE_SHIFT, ++ PAGE_CACHE_SIZE); + + yaffs_GrossUnlock(dev); + +@@ -575,7 +641,7 @@ static int yaffs_readpage_nolock(struct + flush_dcache_page(pg); + kunmap(pg); + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_readpage done\n")); ++ T(YAFFS_TRACE_OS, ("yaffs_readpage done\n")); + return ret; + } + +@@ -593,7 +659,7 @@ static int yaffs_readpage(struct file *f + + /* writepage inspired by/stolen from smbfs */ + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + static int yaffs_writepage(struct page *page, struct writeback_control *wbc) + #else + static int yaffs_writepage(struct page *page) +@@ -616,12 +682,11 @@ static int yaffs_writepage(struct page * + + if (offset > inode->i_size) { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG +- "yaffs_writepage at %08x, inode size = %08x!!!\n", +- (unsigned)(page->index << PAGE_CACHE_SHIFT), +- (unsigned)inode->i_size)); ++ ("yaffs_writepage at %08x, inode size = %08x!!!\n", ++ (unsigned)(page->index << PAGE_CACHE_SHIFT), ++ (unsigned)inode->i_size)); + T(YAFFS_TRACE_OS, +- (KERN_DEBUG " -> don't care!!\n")); ++ (" -> don't care!!\n")); + unlock_page(page); + return 0; + } +@@ -629,11 +694,10 @@ static int yaffs_writepage(struct page * + end_index = inode->i_size >> PAGE_CACHE_SHIFT; + + /* easy case */ +- if (page->index < end_index) { ++ if (page->index < end_index) + nBytes = PAGE_CACHE_SIZE; +- } else { ++ else + nBytes = inode->i_size & (PAGE_CACHE_SIZE - 1); +- } + + get_page(page); + +@@ -643,19 +707,18 @@ static int yaffs_writepage(struct page * + yaffs_GrossLock(obj->myDev); + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_writepage at %08x, size %08x\n", +- (unsigned)(page->index << PAGE_CACHE_SHIFT), nBytes)); ++ ("yaffs_writepage at %08x, size %08x\n", ++ (unsigned)(page->index << PAGE_CACHE_SHIFT), nBytes)); + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "writepag0: obj = %05x, ino = %05x\n", +- (int)obj->variant.fileVariant.fileSize, (int)inode->i_size)); ++ ("writepag0: obj = %05x, ino = %05x\n", ++ (int)obj->variant.fileVariant.fileSize, (int)inode->i_size)); + +- nWritten = +- yaffs_WriteDataToFile(obj, buffer, page->index << PAGE_CACHE_SHIFT, +- nBytes, 0); ++ nWritten = yaffs_WriteDataToFile(obj, buffer, ++ page->index << PAGE_CACHE_SHIFT, nBytes, 0); + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "writepag1: obj = %05x, ino = %05x\n", +- (int)obj->variant.fileVariant.fileSize, (int)inode->i_size)); ++ ("writepag1: obj = %05x, ino = %05x\n", ++ (int)obj->variant.fileVariant.fileSize, (int)inode->i_size)); + + yaffs_GrossUnlock(obj->myDev); + +@@ -667,100 +730,207 @@ static int yaffs_writepage(struct page * + return (nWritten == nBytes) ? 0 : -ENOSPC; + } + ++ ++#if (YAFFS_USE_WRITE_BEGIN_END > 0) ++static int yaffs_write_begin(struct file *filp, struct address_space *mapping, ++ loff_t pos, unsigned len, unsigned flags, ++ struct page **pagep, void **fsdata) ++{ ++ struct page *pg = NULL; ++ pgoff_t index = pos >> PAGE_CACHE_SHIFT; ++ uint32_t offset = pos & (PAGE_CACHE_SIZE - 1); ++ uint32_t to = offset + len; ++ ++ int ret = 0; ++ int space_held = 0; ++ ++ T(YAFFS_TRACE_OS, ("start yaffs_write_begin\n")); ++ /* Get a page */ ++#if LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 28) ++ pg = grab_cache_page_write_begin(mapping, index, flags); ++#else ++ pg = __grab_cache_page(mapping, index); ++#endif ++ ++ *pagep = pg; ++ if (!pg) { ++ ret = -ENOMEM; ++ goto out; ++ } ++ /* Get fs space */ ++ space_held = yaffs_hold_space(filp); ++ ++ if (!space_held) { ++ ret = -ENOSPC; ++ goto out; ++ } ++ ++ /* Update page if required */ ++ ++ if (!Page_Uptodate(pg) && (offset || to < PAGE_CACHE_SIZE)) ++ ret = yaffs_readpage_nolock(filp, pg); ++ ++ if (ret) ++ goto out; ++ ++ /* Happy path return */ ++ T(YAFFS_TRACE_OS, ("end yaffs_write_begin - ok\n")); ++ ++ return 0; ++ ++out: ++ T(YAFFS_TRACE_OS, ("end yaffs_write_begin fail returning %d\n", ret)); ++ if (space_held) ++ yaffs_release_space(filp); ++ if (pg) { ++ unlock_page(pg); ++ page_cache_release(pg); ++ } ++ return ret; ++} ++ ++#else ++ + static int yaffs_prepare_write(struct file *f, struct page *pg, +- unsigned offset, unsigned to) ++ unsigned offset, unsigned to) + { ++ T(YAFFS_TRACE_OS, ("yaffs_prepair_write\n")); + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_prepair_write\n")); + if (!Page_Uptodate(pg) && (offset || to < PAGE_CACHE_SIZE)) + return yaffs_readpage_nolock(f, pg); +- + return 0; ++} ++#endif ++ ++#if (YAFFS_USE_WRITE_BEGIN_END > 0) ++static int yaffs_write_end(struct file *filp, struct address_space *mapping, ++ loff_t pos, unsigned len, unsigned copied, ++ struct page *pg, void *fsdadata) ++{ ++ int ret = 0; ++ void *addr, *kva; ++ uint32_t offset_into_page = pos & (PAGE_CACHE_SIZE - 1); ++ ++ kva = kmap(pg); ++ addr = kva + offset_into_page; ++ ++ T(YAFFS_TRACE_OS, ++ ("yaffs_write_end addr %x pos %x nBytes %d\n", ++ (unsigned) addr, ++ (int)pos, copied)); ++ ++ ret = yaffs_file_write(filp, addr, copied, &pos); ++ ++ if (ret != copied) { ++ T(YAFFS_TRACE_OS, ++ ("yaffs_write_end not same size ret %d copied %d\n", ++ ret, copied)); ++ SetPageError(pg); ++ ClearPageUptodate(pg); ++ } else { ++ SetPageUptodate(pg); ++ } ++ ++ kunmap(pg); + ++ yaffs_release_space(filp); ++ unlock_page(pg); ++ page_cache_release(pg); ++ return ret; + } ++#else + + static int yaffs_commit_write(struct file *f, struct page *pg, unsigned offset, +- unsigned to) ++ unsigned to) + { ++ void *addr, *kva; + +- void *addr = page_address(pg) + offset; + loff_t pos = (((loff_t) pg->index) << PAGE_CACHE_SHIFT) + offset; + int nBytes = to - offset; + int nWritten; + + unsigned spos = pos; +- unsigned saddr = (unsigned)addr; ++ unsigned saddr; ++ ++ kva = kmap(pg); ++ addr = kva + offset; ++ ++ saddr = (unsigned) addr; + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_commit_write addr %x pos %x nBytes %d\n", saddr, +- spos, nBytes)); ++ ("yaffs_commit_write addr %x pos %x nBytes %d\n", ++ saddr, spos, nBytes)); + + nWritten = yaffs_file_write(f, addr, nBytes, &pos); + + if (nWritten != nBytes) { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG +- "yaffs_commit_write not same size nWritten %d nBytes %d\n", +- nWritten, nBytes)); ++ ("yaffs_commit_write not same size nWritten %d nBytes %d\n", ++ nWritten, nBytes)); + SetPageError(pg); + ClearPageUptodate(pg); + } else { + SetPageUptodate(pg); + } + ++ kunmap(pg); ++ + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_commit_write returning %d\n", +- nWritten == nBytes ? 0 : nWritten)); ++ ("yaffs_commit_write returning %d\n", ++ nWritten == nBytes ? 0 : nWritten)); + + return nWritten == nBytes ? 0 : nWritten; +- + } ++#endif ++ + +-static void yaffs_FillInodeFromObject(struct inode *inode, yaffs_Object * obj) ++static void yaffs_FillInodeFromObject(struct inode *inode, yaffs_Object *obj) + { + if (inode && obj) { + + + /* Check mode against the variant type and attempt to repair if broken. */ +- __u32 mode = obj->yst_mode; +- switch( obj->variantType ){ +- case YAFFS_OBJECT_TYPE_FILE : +- if( ! S_ISREG(mode) ){ +- obj->yst_mode &= ~S_IFMT; +- obj->yst_mode |= S_IFREG; +- } +- +- break; +- case YAFFS_OBJECT_TYPE_SYMLINK : +- if( ! S_ISLNK(mode) ){ +- obj->yst_mode &= ~S_IFMT; +- obj->yst_mode |= S_IFLNK; +- } +- +- break; +- case YAFFS_OBJECT_TYPE_DIRECTORY : +- if( ! S_ISDIR(mode) ){ +- obj->yst_mode &= ~S_IFMT; +- obj->yst_mode |= S_IFDIR; +- } +- +- break; +- case YAFFS_OBJECT_TYPE_UNKNOWN : +- case YAFFS_OBJECT_TYPE_HARDLINK : +- case YAFFS_OBJECT_TYPE_SPECIAL : +- default: +- /* TODO? */ +- break; +- } ++ __u32 mode = obj->yst_mode; ++ switch (obj->variantType) { ++ case YAFFS_OBJECT_TYPE_FILE: ++ if (!S_ISREG(mode)) { ++ obj->yst_mode &= ~S_IFMT; ++ obj->yst_mode |= S_IFREG; ++ } ++ ++ break; ++ case YAFFS_OBJECT_TYPE_SYMLINK: ++ if (!S_ISLNK(mode)) { ++ obj->yst_mode &= ~S_IFMT; ++ obj->yst_mode |= S_IFLNK; ++ } ++ ++ break; ++ case YAFFS_OBJECT_TYPE_DIRECTORY: ++ if (!S_ISDIR(mode)) { ++ obj->yst_mode &= ~S_IFMT; ++ obj->yst_mode |= S_IFDIR; ++ } ++ ++ break; ++ case YAFFS_OBJECT_TYPE_UNKNOWN: ++ case YAFFS_OBJECT_TYPE_HARDLINK: ++ case YAFFS_OBJECT_TYPE_SPECIAL: ++ default: ++ /* TODO? */ ++ break; ++ } ++ ++ inode->i_flags |= S_NOATIME; + + inode->i_ino = obj->objectId; + inode->i_mode = obj->yst_mode; + inode->i_uid = obj->yst_uid; + inode->i_gid = obj->yst_gid; +-#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19)) ++#if (LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 19)) + inode->i_blksize = inode->i_sb->s_blocksize; + #endif +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + + inode->i_rdev = old_decode_dev(obj->yst_rdev); + inode->i_atime.tv_sec = (time_t) (obj->yst_atime); +@@ -781,26 +951,25 @@ static void yaffs_FillInodeFromObject(st + inode->i_nlink = yaffs_GetObjectLinkCount(obj); + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG +- "yaffs_FillInode mode %x uid %d gid %d size %d count %d\n", +- inode->i_mode, inode->i_uid, inode->i_gid, +- (int)inode->i_size, atomic_read(&inode->i_count))); ++ ("yaffs_FillInode mode %x uid %d gid %d size %d count %d\n", ++ inode->i_mode, inode->i_uid, inode->i_gid, ++ (int)inode->i_size, atomic_read(&inode->i_count))); + + switch (obj->yst_mode & S_IFMT) { + default: /* fifo, device or socket */ +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + init_special_inode(inode, obj->yst_mode, +- old_decode_dev(obj->yst_rdev)); ++ old_decode_dev(obj->yst_rdev)); + #else + init_special_inode(inode, obj->yst_mode, +- (dev_t) (obj->yst_rdev)); ++ (dev_t) (obj->yst_rdev)); + #endif + break; + case S_IFREG: /* file */ + inode->i_op = &yaffs_file_inode_operations; + inode->i_fop = &yaffs_file_operations; + inode->i_mapping->a_ops = +- &yaffs_file_address_operations; ++ &yaffs_file_address_operations; + break; + case S_IFDIR: /* directory */ + inode->i_op = &yaffs_dir_inode_operations; +@@ -817,34 +986,36 @@ static void yaffs_FillInodeFromObject(st + + } else { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_FileInode invalid parameters\n")); ++ ("yaffs_FileInode invalid parameters\n")); + } + + } + + struct inode *yaffs_get_inode(struct super_block *sb, int mode, int dev, +- yaffs_Object * obj) ++ yaffs_Object *obj) + { + struct inode *inode; + + if (!sb) { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_get_inode for NULL super_block!!\n")); ++ ("yaffs_get_inode for NULL super_block!!\n")); + return NULL; + + } + + if (!obj) { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_get_inode for NULL object!!\n")); ++ ("yaffs_get_inode for NULL object!!\n")); + return NULL; + + } + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_get_inode for object %d\n", obj->objectId)); ++ ("yaffs_get_inode for object %d\n", obj->objectId)); + +- inode = iget(sb, obj->objectId); ++ inode = Y_IGET(sb, obj->objectId); ++ if (IS_ERR(inode)) ++ return NULL; + + /* NB Side effect: iget calls back to yaffs_read_inode(). */ + /* iget also increments the inode's i_count */ +@@ -854,7 +1025,7 @@ struct inode *yaffs_get_inode(struct sup + } + + static ssize_t yaffs_file_write(struct file *f, const char *buf, size_t n, +- loff_t * pos) ++ loff_t *pos) + { + yaffs_Object *obj; + int nWritten, ipos; +@@ -869,28 +1040,26 @@ static ssize_t yaffs_file_write(struct f + + inode = f->f_dentry->d_inode; + +- if (!S_ISBLK(inode->i_mode) && f->f_flags & O_APPEND) { ++ if (!S_ISBLK(inode->i_mode) && f->f_flags & O_APPEND) + ipos = inode->i_size; +- } else { ++ else + ipos = *pos; +- } + +- if (!obj) { ++ if (!obj) + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_file_write: hey obj is null!\n")); +- } else { ++ ("yaffs_file_write: hey obj is null!\n")); ++ else + T(YAFFS_TRACE_OS, +- (KERN_DEBUG +- "yaffs_file_write about to write writing %d bytes" +- "to object %d at %d\n", +- n, obj->objectId, ipos)); +- } ++ ("yaffs_file_write about to write writing %zu bytes" ++ "to object %d at %d\n", ++ n, obj->objectId, ipos)); + + nWritten = yaffs_WriteDataToFile(obj, buf, ipos, n, 0); + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_file_write writing %d bytes, %d written at %d\n", +- n, nWritten, ipos)); ++ ("yaffs_file_write writing %zu bytes, %d written at %d\n", ++ n, nWritten, ipos)); ++ + if (nWritten > 0) { + ipos += nWritten; + *pos = ipos; +@@ -899,10 +1068,9 @@ static ssize_t yaffs_file_write(struct f + inode->i_blocks = (ipos + 511) >> 9; + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG +- "yaffs_file_write size updated to %d bytes, " +- "%d blocks\n", +- ipos, (int)(inode->i_blocks))); ++ ("yaffs_file_write size updated to %d bytes, " ++ "%d blocks\n", ++ ipos, (int)(inode->i_blocks))); + } + + } +@@ -910,13 +1078,54 @@ static ssize_t yaffs_file_write(struct f + return nWritten == 0 ? -ENOSPC : nWritten; + } + ++/* Space holding and freeing is done to ensure we have space available for write_begin/end */ ++/* For now we just assume few parallel writes and check against a small number. */ ++/* Todo: need to do this with a counter to handle parallel reads better */ ++ ++static ssize_t yaffs_hold_space(struct file *f) ++{ ++ yaffs_Object *obj; ++ yaffs_Device *dev; ++ ++ int nFreeChunks; ++ ++ ++ obj = yaffs_DentryToObject(f->f_dentry); ++ ++ dev = obj->myDev; ++ ++ yaffs_GrossLock(dev); ++ ++ nFreeChunks = yaffs_GetNumberOfFreeChunks(dev); ++ ++ yaffs_GrossUnlock(dev); ++ ++ return (nFreeChunks > 20) ? 1 : 0; ++} ++ ++static void yaffs_release_space(struct file *f) ++{ ++ yaffs_Object *obj; ++ yaffs_Device *dev; ++ ++ ++ obj = yaffs_DentryToObject(f->f_dentry); ++ ++ dev = obj->myDev; ++ ++ yaffs_GrossLock(dev); ++ ++ ++ yaffs_GrossUnlock(dev); ++} ++ + static int yaffs_readdir(struct file *f, void *dirent, filldir_t filldir) + { + yaffs_Object *obj; + yaffs_Device *dev; + struct inode *inode = f->f_dentry->d_inode; + unsigned long offset, curoffs; +- struct list_head *i; ++ struct ylist_head *i; + yaffs_Object *l; + + char name[YAFFS_MAX_NAME_LENGTH + 1]; +@@ -932,24 +1141,20 @@ static int yaffs_readdir(struct file *f, + + if (offset == 0) { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_readdir: entry . ino %d \n", +- (int)inode->i_ino)); +- if (filldir(dirent, ".", 1, offset, inode->i_ino, DT_DIR) +- < 0) { ++ ("yaffs_readdir: entry . ino %d \n", ++ (int)inode->i_ino)); ++ if (filldir(dirent, ".", 1, offset, inode->i_ino, DT_DIR) < 0) + goto out; +- } + offset++; + f->f_pos++; + } + if (offset == 1) { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_readdir: entry .. ino %d \n", +- (int)f->f_dentry->d_parent->d_inode->i_ino)); +- if (filldir +- (dirent, "..", 2, offset, +- f->f_dentry->d_parent->d_inode->i_ino, DT_DIR) < 0) { ++ ("yaffs_readdir: entry .. ino %d \n", ++ (int)f->f_dentry->d_parent->d_inode->i_ino)); ++ if (filldir(dirent, "..", 2, offset, ++ f->f_dentry->d_parent->d_inode->i_ino, DT_DIR) < 0) + goto out; +- } + offset++; + f->f_pos++; + } +@@ -965,35 +1170,32 @@ static int yaffs_readdir(struct file *f, + f->f_version = inode->i_version; + } + +- list_for_each(i, &obj->variant.directoryVariant.children) { ++ ylist_for_each(i, &obj->variant.directoryVariant.children) { + curoffs++; + if (curoffs >= offset) { +- l = list_entry(i, yaffs_Object, siblings); ++ l = ylist_entry(i, yaffs_Object, siblings); + + yaffs_GetObjectName(l, name, + YAFFS_MAX_NAME_LENGTH + 1); + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_readdir: %s inode %d\n", name, ++ ("yaffs_readdir: %s inode %d\n", name, + yaffs_GetObjectInode(l))); + + if (filldir(dirent, +- name, +- strlen(name), +- offset, +- yaffs_GetObjectInode(l), +- yaffs_GetObjectType(l)) +- < 0) { ++ name, ++ strlen(name), ++ offset, ++ yaffs_GetObjectInode(l), ++ yaffs_GetObjectType(l)) < 0) + goto up_and_out; +- } + + offset++; + f->f_pos++; + } + } + +- up_and_out: +- out: +- ++up_and_out: ++out: + yaffs_GrossUnlock(dev); + + return 0; +@@ -1002,12 +1204,19 @@ static int yaffs_readdir(struct file *f, + /* + * File creation. Allocate an inode, and we're done.. + */ +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++ ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 29) ++#define YCRED(x) x ++#else ++#define YCRED(x) (x->cred) ++#endif ++ ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + static int yaffs_mknod(struct inode *dir, struct dentry *dentry, int mode, +- dev_t rdev) ++ dev_t rdev) + #else + static int yaffs_mknod(struct inode *dir, struct dentry *dentry, int mode, +- int rdev) ++ int rdev) + #endif + { + struct inode *inode; +@@ -1018,25 +1227,25 @@ static int yaffs_mknod(struct inode *dir + yaffs_Object *parent = yaffs_InodeToObject(dir); + + int error = -ENOSPC; +- uid_t uid = current->fsuid; +- gid_t gid = (dir->i_mode & S_ISGID) ? dir->i_gid : current->fsgid; ++ uid_t uid = YCRED(current)->fsuid; ++ gid_t gid = (dir->i_mode & S_ISGID) ? dir->i_gid : YCRED(current)->fsgid; + +- if((dir->i_mode & S_ISGID) && S_ISDIR(mode)) ++ if ((dir->i_mode & S_ISGID) && S_ISDIR(mode)) + mode |= S_ISGID; + + if (parent) { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_mknod: parent object %d type %d\n", +- parent->objectId, parent->variantType)); ++ ("yaffs_mknod: parent object %d type %d\n", ++ parent->objectId, parent->variantType)); + } else { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_mknod: could not get parent object\n")); ++ ("yaffs_mknod: could not get parent object\n")); + return -EPERM; + } + + T(YAFFS_TRACE_OS, ("yaffs_mknod: making oject for %s, " +- "mode %x dev %x\n", +- dentry->d_name.name, mode, rdev)); ++ "mode %x dev %x\n", ++ dentry->d_name.name, mode, rdev)); + + dev = parent->myDev; + +@@ -1045,33 +1254,28 @@ static int yaffs_mknod(struct inode *dir + switch (mode & S_IFMT) { + default: + /* Special (socket, fifo, device...) */ +- T(YAFFS_TRACE_OS, (KERN_DEBUG +- "yaffs_mknod: making special\n")); +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) +- obj = +- yaffs_MknodSpecial(parent, dentry->d_name.name, mode, uid, +- gid, old_encode_dev(rdev)); +-#else +- obj = +- yaffs_MknodSpecial(parent, dentry->d_name.name, mode, uid, +- gid, rdev); ++ T(YAFFS_TRACE_OS, ("yaffs_mknod: making special\n")); ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) ++ obj = yaffs_MknodSpecial(parent, dentry->d_name.name, mode, uid, ++ gid, old_encode_dev(rdev)); ++#else ++ obj = yaffs_MknodSpecial(parent, dentry->d_name.name, mode, uid, ++ gid, rdev); + #endif + break; + case S_IFREG: /* file */ +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_mknod: making file\n")); +- obj = +- yaffs_MknodFile(parent, dentry->d_name.name, mode, uid, +- gid); ++ T(YAFFS_TRACE_OS, ("yaffs_mknod: making file\n")); ++ obj = yaffs_MknodFile(parent, dentry->d_name.name, mode, uid, ++ gid); + break; + case S_IFDIR: /* directory */ + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_mknod: making directory\n")); +- obj = +- yaffs_MknodDirectory(parent, dentry->d_name.name, mode, +- uid, gid); ++ ("yaffs_mknod: making directory\n")); ++ obj = yaffs_MknodDirectory(parent, dentry->d_name.name, mode, ++ uid, gid); + break; + case S_IFLNK: /* symlink */ +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_mknod: making file\n")); ++ T(YAFFS_TRACE_OS, ("yaffs_mknod: making symlink\n")); + obj = NULL; /* Do we ever get here? */ + break; + } +@@ -1083,12 +1287,12 @@ static int yaffs_mknod(struct inode *dir + inode = yaffs_get_inode(dir->i_sb, mode, rdev, obj); + d_instantiate(dentry, inode); + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_mknod created object %d count = %d\n", +- obj->objectId, atomic_read(&inode->i_count))); ++ ("yaffs_mknod created object %d count = %d\n", ++ obj->objectId, atomic_read(&inode->i_count))); + error = 0; + } else { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_mknod failed making object\n")); ++ ("yaffs_mknod failed making object\n")); + error = -ENOMEM; + } + +@@ -1098,25 +1302,19 @@ static int yaffs_mknod(struct inode *dir + static int yaffs_mkdir(struct inode *dir, struct dentry *dentry, int mode) + { + int retVal; +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_mkdir\n")); ++ T(YAFFS_TRACE_OS, ("yaffs_mkdir\n")); + retVal = yaffs_mknod(dir, dentry, mode | S_IFDIR, 0); +-#if 0 +- /* attempt to fix dir bug - didn't work */ +- if (!retVal) { +- dget(dentry); +- } +-#endif + return retVal; + } + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + static int yaffs_create(struct inode *dir, struct dentry *dentry, int mode, + struct nameidata *n) + #else + static int yaffs_create(struct inode *dir, struct dentry *dentry, int mode) + #endif + { +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_create\n")); ++ T(YAFFS_TRACE_OS, ("yaffs_create\n")); + return yaffs_mknod(dir, dentry, mode | S_IFREG, 0); + } + +@@ -1127,8 +1325,8 @@ static int yaffs_unlink(struct inode *di + yaffs_Device *dev; + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_unlink %d:%s\n", (int)(dir->i_ino), +- dentry->d_name.name)); ++ ("yaffs_unlink %d:%s\n", (int)(dir->i_ino), ++ dentry->d_name.name)); + + dev = yaffs_InodeToObject(dir)->myDev; + +@@ -1151,82 +1349,74 @@ static int yaffs_unlink(struct inode *di + * Create a link... + */ + static int yaffs_link(struct dentry *old_dentry, struct inode *dir, +- struct dentry *dentry) ++ struct dentry *dentry) + { + struct inode *inode = old_dentry->d_inode; + yaffs_Object *obj = NULL; + yaffs_Object *link = NULL; + yaffs_Device *dev; + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_link\n")); ++ T(YAFFS_TRACE_OS, ("yaffs_link\n")); + + obj = yaffs_InodeToObject(inode); + dev = obj->myDev; + + yaffs_GrossLock(dev); + +- if (!S_ISDIR(inode->i_mode)) /* Don't link directories */ +- { +- link = +- yaffs_Link(yaffs_InodeToObject(dir), dentry->d_name.name, +- obj); +- } ++ if (!S_ISDIR(inode->i_mode)) /* Don't link directories */ ++ link = yaffs_Link(yaffs_InodeToObject(dir), dentry->d_name.name, ++ obj); + + if (link) { + old_dentry->d_inode->i_nlink = yaffs_GetObjectLinkCount(obj); + d_instantiate(dentry, old_dentry->d_inode); + atomic_inc(&old_dentry->d_inode->i_count); + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_link link count %d i_count %d\n", +- old_dentry->d_inode->i_nlink, +- atomic_read(&old_dentry->d_inode->i_count))); +- ++ ("yaffs_link link count %d i_count %d\n", ++ old_dentry->d_inode->i_nlink, ++ atomic_read(&old_dentry->d_inode->i_count))); + } + + yaffs_GrossUnlock(dev); + +- if (link) { +- ++ if (link) + return 0; +- } + + return -EPERM; + } + + static int yaffs_symlink(struct inode *dir, struct dentry *dentry, +- const char *symname) ++ const char *symname) + { + yaffs_Object *obj; + yaffs_Device *dev; +- uid_t uid = current->fsuid; +- gid_t gid = (dir->i_mode & S_ISGID) ? dir->i_gid : current->fsgid; ++ uid_t uid = YCRED(current)->fsuid; ++ gid_t gid = (dir->i_mode & S_ISGID) ? dir->i_gid : YCRED(current)->fsgid; + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_symlink\n")); ++ T(YAFFS_TRACE_OS, ("yaffs_symlink\n")); + + dev = yaffs_InodeToObject(dir)->myDev; + yaffs_GrossLock(dev); + obj = yaffs_MknodSymLink(yaffs_InodeToObject(dir), dentry->d_name.name, +- S_IFLNK | S_IRWXUGO, uid, gid, symname); ++ S_IFLNK | S_IRWXUGO, uid, gid, symname); + yaffs_GrossUnlock(dev); + + if (obj) { +- + struct inode *inode; + + inode = yaffs_get_inode(dir->i_sb, obj->yst_mode, 0, obj); + d_instantiate(dentry, inode); +- T(YAFFS_TRACE_OS, (KERN_DEBUG "symlink created OK\n")); ++ T(YAFFS_TRACE_OS, ("symlink created OK\n")); + return 0; + } else { +- T(YAFFS_TRACE_OS, (KERN_DEBUG "symlink not created\n")); +- ++ T(YAFFS_TRACE_OS, ("symlink not created\n")); + } + + return -ENOMEM; + } + + static int yaffs_sync_object(struct file *file, struct dentry *dentry, +- int datasync) ++ int datasync) + { + + yaffs_Object *obj; +@@ -1236,7 +1426,7 @@ static int yaffs_sync_object(struct file + + dev = obj->myDev; + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_sync_object\n")); ++ T(YAFFS_TRACE_OS, ("yaffs_sync_object\n")); + yaffs_GrossLock(dev); + yaffs_FlushFile(obj, 1); + yaffs_GrossUnlock(dev); +@@ -1255,41 +1445,36 @@ static int yaffs_rename(struct inode *ol + int retVal = YAFFS_FAIL; + yaffs_Object *target; + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_rename\n")); ++ T(YAFFS_TRACE_OS, ("yaffs_rename\n")); + dev = yaffs_InodeToObject(old_dir)->myDev; + + yaffs_GrossLock(dev); + + /* Check if the target is an existing directory that is not empty. */ +- target = +- yaffs_FindObjectByName(yaffs_InodeToObject(new_dir), +- new_dentry->d_name.name); ++ target = yaffs_FindObjectByName(yaffs_InodeToObject(new_dir), ++ new_dentry->d_name.name); + + + +- if (target && +- target->variantType == YAFFS_OBJECT_TYPE_DIRECTORY && +- !list_empty(&target->variant.directoryVariant.children)) { ++ if (target && target->variantType == YAFFS_OBJECT_TYPE_DIRECTORY && ++ !ylist_empty(&target->variant.directoryVariant.children)) { + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "target is non-empty dir\n")); ++ T(YAFFS_TRACE_OS, ("target is non-empty dir\n")); + + retVal = YAFFS_FAIL; + } else { +- + /* Now does unlinking internally using shadowing mechanism */ +- T(YAFFS_TRACE_OS, (KERN_DEBUG "calling yaffs_RenameObject\n")); +- +- retVal = +- yaffs_RenameObject(yaffs_InodeToObject(old_dir), +- old_dentry->d_name.name, +- yaffs_InodeToObject(new_dir), +- new_dentry->d_name.name); ++ T(YAFFS_TRACE_OS, ("calling yaffs_RenameObject\n")); + ++ retVal = yaffs_RenameObject(yaffs_InodeToObject(old_dir), ++ old_dentry->d_name.name, ++ yaffs_InodeToObject(new_dir), ++ new_dentry->d_name.name); + } + yaffs_GrossUnlock(dev); + + if (retVal == YAFFS_OK) { +- if(target) { ++ if (target) { + new_dentry->d_inode->i_nlink--; + mark_inode_dirty(new_dentry->d_inode); + } +@@ -1298,7 +1483,6 @@ static int yaffs_rename(struct inode *ol + } else { + return -ENOTEMPTY; + } +- + } + + static int yaffs_setattr(struct dentry *dentry, struct iattr *attr) +@@ -1308,15 +1492,15 @@ static int yaffs_setattr(struct dentry * + yaffs_Device *dev; + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_setattr of object %d\n", +- yaffs_InodeToObject(inode)->objectId)); +- +- if ((error = inode_change_ok(inode, attr)) == 0) { ++ ("yaffs_setattr of object %d\n", ++ yaffs_InodeToObject(inode)->objectId)); + ++ error = inode_change_ok(inode, attr); ++ if (error == 0) { + dev = yaffs_InodeToObject(inode)->myDev; + yaffs_GrossLock(dev); + if (yaffs_SetAttributes(yaffs_InodeToObject(inode), attr) == +- YAFFS_OK) { ++ YAFFS_OK) { + error = 0; + } else { + error = -EPERM; +@@ -1328,12 +1512,12 @@ static int yaffs_setattr(struct dentry * + return error; + } + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + static int yaffs_statfs(struct dentry *dentry, struct kstatfs *buf) + { + yaffs_Device *dev = yaffs_DentryToObject(dentry)->myDev; + struct super_block *sb = dentry->d_sb; +-#elif (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#elif (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + static int yaffs_statfs(struct super_block *sb, struct kstatfs *buf) + { + yaffs_Device *dev = yaffs_SuperToDevice(sb); +@@ -1343,32 +1527,53 @@ static int yaffs_statfs(struct super_blo + yaffs_Device *dev = yaffs_SuperToDevice(sb); + #endif + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_statfs\n")); ++ T(YAFFS_TRACE_OS, ("yaffs_statfs\n")); + + yaffs_GrossLock(dev); + + buf->f_type = YAFFS_MAGIC; + buf->f_bsize = sb->s_blocksize; + buf->f_namelen = 255; +- if (sb->s_blocksize > dev->nDataBytesPerChunk) { ++ ++ if (dev->nDataBytesPerChunk & (dev->nDataBytesPerChunk - 1)) { ++ /* Do this if chunk size is not a power of 2 */ ++ ++ uint64_t bytesInDev; ++ uint64_t bytesFree; ++ ++ bytesInDev = ((uint64_t)((dev->endBlock - dev->startBlock + 1))) * ++ ((uint64_t)(dev->nChunksPerBlock * dev->nDataBytesPerChunk)); ++ ++ do_div(bytesInDev, sb->s_blocksize); /* bytesInDev becomes the number of blocks */ ++ buf->f_blocks = bytesInDev; ++ ++ bytesFree = ((uint64_t)(yaffs_GetNumberOfFreeChunks(dev))) * ++ ((uint64_t)(dev->nDataBytesPerChunk)); ++ ++ do_div(bytesFree, sb->s_blocksize); ++ ++ buf->f_bfree = bytesFree; ++ ++ } else if (sb->s_blocksize > dev->nDataBytesPerChunk) { + + buf->f_blocks = +- (dev->endBlock - dev->startBlock + +- 1) * dev->nChunksPerBlock / (sb->s_blocksize / +- dev->nDataBytesPerChunk); ++ (dev->endBlock - dev->startBlock + 1) * ++ dev->nChunksPerBlock / ++ (sb->s_blocksize / dev->nDataBytesPerChunk); + buf->f_bfree = +- yaffs_GetNumberOfFreeChunks(dev) / (sb->s_blocksize / +- dev->nDataBytesPerChunk); ++ yaffs_GetNumberOfFreeChunks(dev) / ++ (sb->s_blocksize / dev->nDataBytesPerChunk); + } else { +- + buf->f_blocks = +- (dev->endBlock - dev->startBlock + +- 1) * dev->nChunksPerBlock * (dev->nDataBytesPerChunk / +- sb->s_blocksize); ++ (dev->endBlock - dev->startBlock + 1) * ++ dev->nChunksPerBlock * ++ (dev->nDataBytesPerChunk / sb->s_blocksize); ++ + buf->f_bfree = +- yaffs_GetNumberOfFreeChunks(dev) * (dev->nDataBytesPerChunk / +- sb->s_blocksize); ++ yaffs_GetNumberOfFreeChunks(dev) * ++ (dev->nDataBytesPerChunk / sb->s_blocksize); + } ++ + buf->f_files = 0; + buf->f_ffree = 0; + buf->f_bavail = buf->f_bfree; +@@ -1378,18 +1583,19 @@ static int yaffs_statfs(struct super_blo + } + + +-/** + static int yaffs_do_sync_fs(struct super_block *sb) + { + + yaffs_Device *dev = yaffs_SuperToDevice(sb); +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_do_sync_fs\n")); ++ T(YAFFS_TRACE_OS, ("yaffs_do_sync_fs\n")); + +- if(sb->s_dirt) { ++ if (sb->s_dirt) { + yaffs_GrossLock(dev); + +- if(dev) ++ if (dev) { ++ yaffs_FlushEntireDeviceCache(dev); + yaffs_CheckpointSave(dev); ++ } + + yaffs_GrossUnlock(dev); + +@@ -1397,35 +1603,73 @@ static int yaffs_do_sync_fs(struct super + } + return 0; + } +-**/ + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++ ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + static void yaffs_write_super(struct super_block *sb) + #else + static int yaffs_write_super(struct super_block *sb) + #endif + { + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_write_super\n")); +-#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,18)) +- return 0; /* yaffs_do_sync_fs(sb);*/ ++ T(YAFFS_TRACE_OS, ("yaffs_write_super\n")); ++ if (yaffs_auto_checkpoint >= 2) ++ yaffs_do_sync_fs(sb); ++#if (LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 18)) ++ return 0; + #endif + } + + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + static int yaffs_sync_fs(struct super_block *sb, int wait) + #else + static int yaffs_sync_fs(struct super_block *sb) + #endif + { ++ T(YAFFS_TRACE_OS, ("yaffs_sync_fs\n")); ++ ++ if (yaffs_auto_checkpoint >= 1) ++ yaffs_do_sync_fs(sb); ++ ++ return 0; ++} ++ ++#ifdef YAFFS_USE_OWN_IGET ++ ++static struct inode *yaffs_iget(struct super_block *sb, unsigned long ino) ++{ ++ struct inode *inode; ++ yaffs_Object *obj; ++ yaffs_Device *dev = yaffs_SuperToDevice(sb); ++ ++ T(YAFFS_TRACE_OS, ++ ("yaffs_iget for %lu\n", ino)); + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_sync_fs\n")); ++ inode = iget_locked(sb, ino); ++ if (!inode) ++ return ERR_PTR(-ENOMEM); ++ if (!(inode->i_state & I_NEW)) ++ return inode; ++ ++ /* NB This is called as a side effect of other functions, but ++ * we had to release the lock to prevent deadlocks, so ++ * need to lock again. ++ */ + +- return 0; /* yaffs_do_sync_fs(sb);*/ ++ yaffs_GrossLock(dev); + ++ obj = yaffs_FindObjectByNumber(dev, inode->i_ino); ++ ++ yaffs_FillInodeFromObject(inode, obj); ++ ++ yaffs_GrossUnlock(dev); ++ ++ unlock_new_inode(inode); ++ return inode; + } + ++#else + + static void yaffs_read_inode(struct inode *inode) + { +@@ -1438,7 +1682,7 @@ static void yaffs_read_inode(struct inod + yaffs_Device *dev = yaffs_SuperToDevice(inode->i_sb); + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_read_inode for %d\n", (int)inode->i_ino)); ++ ("yaffs_read_inode for %d\n", (int)inode->i_ino)); + + yaffs_GrossLock(dev); + +@@ -1449,18 +1693,20 @@ static void yaffs_read_inode(struct inod + yaffs_GrossUnlock(dev); + } + +-static LIST_HEAD(yaffs_dev_list); ++#endif ++ ++static YLIST_HEAD(yaffs_dev_list); + +-#if 0 // not used ++#if 0 /* not used */ + static int yaffs_remount_fs(struct super_block *sb, int *flags, char *data) + { + yaffs_Device *dev = yaffs_SuperToDevice(sb); + +- if( *flags & MS_RDONLY ) { ++ if (*flags & MS_RDONLY) { + struct mtd_info *mtd = yaffs_SuperToDevice(sb)->genericDevice; + + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_remount_fs: %s: RO\n", dev->name )); ++ ("yaffs_remount_fs: %s: RO\n", dev->name)); + + yaffs_GrossLock(dev); + +@@ -1472,10 +1718,9 @@ static int yaffs_remount_fs(struct super + mtd->sync(mtd); + + yaffs_GrossUnlock(dev); +- } +- else { ++ } else { + T(YAFFS_TRACE_OS, +- (KERN_DEBUG "yaffs_remount_fs: %s: RW\n", dev->name )); ++ ("yaffs_remount_fs: %s: RW\n", dev->name)); + } + + return 0; +@@ -1486,7 +1731,7 @@ static void yaffs_put_super(struct super + { + yaffs_Device *dev = yaffs_SuperToDevice(sb); + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_put_super\n")); ++ T(YAFFS_TRACE_OS, ("yaffs_put_super\n")); + + yaffs_GrossLock(dev); + +@@ -1494,18 +1739,17 @@ static void yaffs_put_super(struct super + + yaffs_CheckpointSave(dev); + +- if (dev->putSuperFunc) { ++ if (dev->putSuperFunc) + dev->putSuperFunc(sb); +- } + + yaffs_Deinitialise(dev); + + yaffs_GrossUnlock(dev); + + /* we assume this is protected by lock_kernel() in mount/umount */ +- list_del(&dev->devList); ++ ylist_del(&dev->devList); + +- if(dev->spareBuffer){ ++ if (dev->spareBuffer) { + YFREE(dev->spareBuffer); + dev->spareBuffer = NULL; + } +@@ -1516,12 +1760,10 @@ static void yaffs_put_super(struct super + + static void yaffs_MTDPutSuper(struct super_block *sb) + { +- + struct mtd_info *mtd = yaffs_SuperToDevice(sb)->genericDevice; + +- if (mtd->sync) { ++ if (mtd->sync) + mtd->sync(mtd); +- } + + put_mtd_device(mtd); + } +@@ -1531,9 +1773,9 @@ static void yaffs_MarkSuperBlockDirty(vo + { + struct super_block *sb = (struct super_block *)vsb; + +- T(YAFFS_TRACE_OS, (KERN_DEBUG "yaffs_MarkSuperBlockDirty() sb = %p\n",sb)); +-// if(sb) +-// sb->s_dirt = 1; ++ T(YAFFS_TRACE_OS, ("yaffs_MarkSuperBlockDirty() sb = %p\n", sb)); ++ if (sb) ++ sb->s_dirt = 1; + } + + typedef struct { +@@ -1546,48 +1788,48 @@ typedef struct { + #define MAX_OPT_LEN 20 + static int yaffs_parse_options(yaffs_options *options, const char *options_str) + { +- char cur_opt[MAX_OPT_LEN+1]; ++ char cur_opt[MAX_OPT_LEN + 1]; + int p; + int error = 0; + + /* Parse through the options which is a comma seperated list */ + +- while(options_str && *options_str && !error){ +- memset(cur_opt,0,MAX_OPT_LEN+1); ++ while (options_str && *options_str && !error) { ++ memset(cur_opt, 0, MAX_OPT_LEN + 1); + p = 0; + +- while(*options_str && *options_str != ','){ +- if(p < MAX_OPT_LEN){ ++ while (*options_str && *options_str != ',') { ++ if (p < MAX_OPT_LEN) { + cur_opt[p] = *options_str; + p++; + } + options_str++; + } + +- if(!strcmp(cur_opt,"inband-tags")) ++ if (!strcmp(cur_opt, "inband-tags")) + options->inband_tags = 1; +- else if(!strcmp(cur_opt,"no-cache")) ++ else if (!strcmp(cur_opt, "no-cache")) + options->no_cache = 1; +- else if(!strcmp(cur_opt,"no-checkpoint-read")) ++ else if (!strcmp(cur_opt, "no-checkpoint-read")) + options->skip_checkpoint_read = 1; +- else if(!strcmp(cur_opt,"no-checkpoint-write")) ++ else if (!strcmp(cur_opt, "no-checkpoint-write")) + options->skip_checkpoint_write = 1; +- else if(!strcmp(cur_opt,"no-checkpoint")){ ++ else if (!strcmp(cur_opt, "no-checkpoint")) { + options->skip_checkpoint_read = 1; + options->skip_checkpoint_write = 1; + } else { +- printk(KERN_INFO "yaffs: Bad mount option \"%s\"\n",cur_opt); ++ printk(KERN_INFO "yaffs: Bad mount option \"%s\"\n", ++ cur_opt); + error = 1; + } +- + } + + return error; + } + + static struct super_block *yaffs_internal_read_super(int yaffsVersion, +- struct super_block *sb, +- void *data, int silent) ++ struct super_block *sb, ++ void *data, int silent) + { + int nBlocks; + struct inode *inode = NULL; +@@ -1602,6 +1844,7 @@ static struct super_block *yaffs_interna + + sb->s_magic = YAFFS_MAGIC; + sb->s_op = &yaffs_super_ops; ++ sb->s_flags |= MS_NOATIME; + + if (!sb) + printk(KERN_INFO "yaffs: sb is NULL\n"); +@@ -1614,14 +1857,14 @@ static struct super_block *yaffs_interna + sb->s_dev, + yaffs_devname(sb, devname_buf)); + +- if(!data_str) ++ if (!data_str) + data_str = ""; + +- printk(KERN_INFO "yaffs: passed flags \"%s\"\n",data_str); ++ printk(KERN_INFO "yaffs: passed flags \"%s\"\n", data_str); + +- memset(&options,0,sizeof(options)); ++ memset(&options, 0, sizeof(options)); + +- if(yaffs_parse_options(&options,data_str)){ ++ if (yaffs_parse_options(&options, data_str)) { + /* Option parsing failed */ + return NULL; + } +@@ -1645,9 +1888,9 @@ static struct super_block *yaffs_interna + yaffs_devname(sb, devname_buf))); + + /* Check it's an mtd device..... */ +- if (MAJOR(sb->s_dev) != MTD_BLOCK_MAJOR) { ++ if (MAJOR(sb->s_dev) != MTD_BLOCK_MAJOR) + return NULL; /* This isn't an mtd device */ +- } ++ + /* Get the device */ + mtd = get_mtd_device(NULL, MINOR(sb->s_dev)); + if (!mtd) { +@@ -1673,29 +1916,23 @@ static struct super_block *yaffs_interna + T(YAFFS_TRACE_OS, (" %s %d\n", WRITE_SIZE_STR, WRITE_SIZE(mtd))); + T(YAFFS_TRACE_OS, (" oobsize %d\n", mtd->oobsize)); + T(YAFFS_TRACE_OS, (" erasesize %d\n", mtd->erasesize)); +- T(YAFFS_TRACE_OS, (" size %d\n", mtd->size)); ++#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 29) ++ T(YAFFS_TRACE_OS, (" size %u\n", mtd->size)); ++#else ++ T(YAFFS_TRACE_OS, (" size %lld\n", mtd->size)); ++#endif + + #ifdef CONFIG_YAFFS_AUTO_YAFFS2 + +- if (yaffsVersion == 1 && +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) +- mtd->writesize >= 2048) { +-#else +- mtd->oobblock >= 2048) { +-#endif +- T(YAFFS_TRACE_ALWAYS,("yaffs: auto selecting yaffs2\n")); +- yaffsVersion = 2; ++ if (yaffsVersion == 1 && WRITE_SIZE(mtd) >= 2048) { ++ T(YAFFS_TRACE_ALWAYS, ("yaffs: auto selecting yaffs2\n")); ++ yaffsVersion = 2; + } + + /* Added NCB 26/5/2006 for completeness */ +- if (yaffsVersion == 2 && +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) +- mtd->writesize == 512) { +-#else +- mtd->oobblock == 512) { +-#endif +- T(YAFFS_TRACE_ALWAYS,("yaffs: auto selecting yaffs1\n")); +- yaffsVersion = 1; ++ if (yaffsVersion == 2 && !options.inband_tags && WRITE_SIZE(mtd) == 512) { ++ T(YAFFS_TRACE_ALWAYS, ("yaffs: auto selecting yaffs1\n")); ++ yaffsVersion = 1; + } + + #endif +@@ -1707,7 +1944,7 @@ static struct super_block *yaffs_interna + !mtd->block_markbad || + !mtd->read || + !mtd->write || +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + !mtd->read_oob || !mtd->write_oob) { + #else + !mtd->write_ecc || +@@ -1719,12 +1956,9 @@ static struct super_block *yaffs_interna + return NULL; + } + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) +- if (mtd->writesize < YAFFS_MIN_YAFFS2_CHUNK_SIZE || +-#else +- if (mtd->oobblock < YAFFS_MIN_YAFFS2_CHUNK_SIZE || +-#endif +- mtd->oobsize < YAFFS_MIN_YAFFS2_SPARE_SIZE) { ++ if ((WRITE_SIZE(mtd) < YAFFS_MIN_YAFFS2_CHUNK_SIZE || ++ mtd->oobsize < YAFFS_MIN_YAFFS2_SPARE_SIZE) && ++ !options.inband_tags) { + T(YAFFS_TRACE_ALWAYS, + ("yaffs: MTD device does not have the " + "right page sizes\n")); +@@ -1735,7 +1969,7 @@ static struct super_block *yaffs_interna + if (!mtd->erase || + !mtd->read || + !mtd->write || +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + !mtd->read_oob || !mtd->write_oob) { + #else + !mtd->write_ecc || +@@ -1761,7 +1995,7 @@ static struct super_block *yaffs_interna + * Set the yaffs_Device up for mtd + */ + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + sb->s_fs_info = dev = kmalloc(sizeof(yaffs_Device), GFP_KERNEL); + #else + sb->u.generic_sbp = dev = kmalloc(sizeof(yaffs_Device), GFP_KERNEL); +@@ -1780,13 +2014,15 @@ static struct super_block *yaffs_interna + + /* Set up the memory size parameters.... */ + +- nBlocks = mtd->size / (YAFFS_CHUNKS_PER_BLOCK * YAFFS_BYTES_PER_CHUNK); ++ nBlocks = YCALCBLOCKS(mtd->size, (YAFFS_CHUNKS_PER_BLOCK * YAFFS_BYTES_PER_CHUNK)); ++ + dev->startBlock = 0; + dev->endBlock = nBlocks - 1; + dev->nChunksPerBlock = YAFFS_CHUNKS_PER_BLOCK; +- dev->nDataBytesPerChunk = YAFFS_BYTES_PER_CHUNK; ++ dev->totalBytesPerChunk = YAFFS_BYTES_PER_CHUNK; + dev->nReservedBlocks = 5; + dev->nShortOpCaches = (options.no_cache) ? 0 : 10; ++ dev->inbandTags = options.inband_tags; + + /* ... and the functions. */ + if (yaffsVersion == 2) { +@@ -1798,20 +2034,19 @@ static struct super_block *yaffs_interna + dev->queryNANDBlock = nandmtd2_QueryNANDBlock; + dev->spareBuffer = YMALLOC(mtd->oobsize); + dev->isYaffs2 = 1; +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) +- dev->nDataBytesPerChunk = mtd->writesize; ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) ++ dev->totalBytesPerChunk = mtd->writesize; + dev->nChunksPerBlock = mtd->erasesize / mtd->writesize; + #else +- dev->nDataBytesPerChunk = mtd->oobblock; ++ dev->totalBytesPerChunk = mtd->oobblock; + dev->nChunksPerBlock = mtd->erasesize / mtd->oobblock; + #endif +- nBlocks = mtd->size / mtd->erasesize; ++ nBlocks = YCALCBLOCKS(mtd->size, mtd->erasesize); + +- dev->nCheckpointReservedBlocks = CONFIG_YAFFS_CHECKPOINT_RESERVED_BLOCKS; + dev->startBlock = 0; + dev->endBlock = nBlocks - 1; + } else { +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + /* use the MTD interface in yaffs_mtdif1.c */ + dev->writeChunkWithTagsToNAND = + nandmtd1_WriteChunkWithTagsToNAND; +@@ -1847,7 +2082,7 @@ static struct super_block *yaffs_interna + dev->skipCheckpointWrite = options.skip_checkpoint_write; + + /* we assume this is protected by lock_kernel() in mount/umount */ +- list_add_tail(&dev->devList, &yaffs_dev_list); ++ ylist_add_tail(&dev->devList, &yaffs_dev_list); + + init_MUTEX(&dev->grossLock); + +@@ -1884,20 +2119,23 @@ static struct super_block *yaffs_interna + return NULL; + } + sb->s_root = root; ++ sb->s_dirt = !dev->isCheckpointed; ++ T(YAFFS_TRACE_ALWAYS, ++ ("yaffs_read_super: isCheckpointed %d\n", dev->isCheckpointed)); + + T(YAFFS_TRACE_OS, ("yaffs_read_super: done\n")); + return sb; + } + + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + static int yaffs_internal_read_super_mtd(struct super_block *sb, void *data, + int silent) + { + return yaffs_internal_read_super(1, sb, data, silent) ? 0 : -EINVAL; + } + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + static int yaffs_read_super(struct file_system_type *fs, + int flags, const char *dev_name, + void *data, struct vfsmount *mnt) +@@ -1938,14 +2176,14 @@ static DECLARE_FSTYPE(yaffs_fs_type, "ya + + #ifdef CONFIG_YAFFS_YAFFS2 + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + static int yaffs2_internal_read_super_mtd(struct super_block *sb, void *data, + int silent) + { + return yaffs_internal_read_super(2, sb, data, silent) ? 0 : -EINVAL; + } + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) + static int yaffs2_read_super(struct file_system_type *fs, + int flags, const char *dev_name, void *data, + struct vfsmount *mnt) +@@ -1990,12 +2228,12 @@ static char *yaffs_dump_dev(char *buf, y + { + buf += sprintf(buf, "startBlock......... %d\n", dev->startBlock); + buf += sprintf(buf, "endBlock........... %d\n", dev->endBlock); ++ buf += sprintf(buf, "totalBytesPerChunk. %d\n", dev->totalBytesPerChunk); + buf += sprintf(buf, "nDataBytesPerChunk. %d\n", dev->nDataBytesPerChunk); + buf += sprintf(buf, "chunkGroupBits..... %d\n", dev->chunkGroupBits); + buf += sprintf(buf, "chunkGroupSize..... %d\n", dev->chunkGroupSize); + buf += sprintf(buf, "nErasedBlocks...... %d\n", dev->nErasedBlocks); + buf += sprintf(buf, "nReservedBlocks.... %d\n", dev->nReservedBlocks); +- buf += sprintf(buf, "nCheckptResBlocks.. %d\n", dev->nCheckpointReservedBlocks); + buf += sprintf(buf, "blocksInCheckpoint. %d\n", dev->blocksInCheckpoint); + buf += sprintf(buf, "nTnodesCreated..... %d\n", dev->nTnodesCreated); + buf += sprintf(buf, "nFreeTnodes........ %d\n", dev->nFreeTnodes); +@@ -2006,10 +2244,8 @@ static char *yaffs_dump_dev(char *buf, y + buf += sprintf(buf, "nPageReads......... %d\n", dev->nPageReads); + buf += sprintf(buf, "nBlockErasures..... %d\n", dev->nBlockErasures); + buf += sprintf(buf, "nGCCopies.......... %d\n", dev->nGCCopies); +- buf += +- sprintf(buf, "garbageCollections. %d\n", dev->garbageCollections); +- buf += +- sprintf(buf, "passiveGCs......... %d\n", ++ buf += sprintf(buf, "garbageCollections. %d\n", dev->garbageCollections); ++ buf += sprintf(buf, "passiveGCs......... %d\n", + dev->passiveGarbageCollections); + buf += sprintf(buf, "nRetriedWrites..... %d\n", dev->nRetriedWrites); + buf += sprintf(buf, "nShortOpCaches..... %d\n", dev->nShortOpCaches); +@@ -2025,6 +2261,7 @@ static char *yaffs_dump_dev(char *buf, y + sprintf(buf, "nBackgroudDeletions %d\n", dev->nBackgroundDeletions); + buf += sprintf(buf, "useNANDECC......... %d\n", dev->useNANDECC); + buf += sprintf(buf, "isYaffs2........... %d\n", dev->isYaffs2); ++ buf += sprintf(buf, "inbandTags......... %d\n", dev->inbandTags); + + return buf; + } +@@ -2033,7 +2270,7 @@ static int yaffs_proc_read(char *page, + char **start, + off_t offset, int count, int *eof, void *data) + { +- struct list_head *item; ++ struct ylist_head *item; + char *buf = page; + int step = offset; + int n = 0; +@@ -2057,8 +2294,8 @@ static int yaffs_proc_read(char *page, + lock_kernel(); + + /* Locate and print the Nth entry. Order N-squared but N is small. */ +- list_for_each(item, &yaffs_dev_list) { +- yaffs_Device *dev = list_entry(item, yaffs_Device, devList); ++ ylist_for_each(item, &yaffs_dev_list) { ++ yaffs_Device *dev = ylist_entry(item, yaffs_Device, devList); + if (n < step) { + n++; + continue; +@@ -2119,7 +2356,7 @@ static int yaffs_proc_write(struct file + char *end; + char *mask_name; + const char *x; +- char substring[MAX_MASK_NAME_LENGTH+1]; ++ char substring[MAX_MASK_NAME_LENGTH + 1]; + int i; + int done = 0; + int add, len = 0; +@@ -2129,9 +2366,8 @@ static int yaffs_proc_write(struct file + + while (!done && (pos < count)) { + done = 1; +- while ((pos < count) && isspace(buf[pos])) { ++ while ((pos < count) && isspace(buf[pos])) + pos++; +- } + + switch (buf[pos]) { + case '+': +@@ -2148,20 +2384,21 @@ static int yaffs_proc_write(struct file + mask_name = NULL; + + mask_bitfield = simple_strtoul(buf + pos, &end, 0); ++ + if (end > buf + pos) { + mask_name = "numeral"; + len = end - (buf + pos); + pos += len; + done = 0; + } else { +- for(x = buf + pos, i = 0; +- (*x == '_' || (*x >='a' && *x <= 'z')) && +- i = 'a' && *x <= 'z')) && ++ i < MAX_MASK_NAME_LENGTH; x++, i++, pos++) ++ substring[i] = *x; + substring[i] = '\0'; + + for (i = 0; mask_flags[i].mask_name != NULL; i++) { +- if(strcmp(substring,mask_flags[i].mask_name) == 0){ ++ if (strcmp(substring, mask_flags[i].mask_name) == 0) { + mask_name = mask_flags[i].mask_name; + mask_bitfield = mask_flags[i].mask_bitfield; + done = 0; +@@ -2172,7 +2409,7 @@ static int yaffs_proc_write(struct file + + if (mask_name != NULL) { + done = 0; +- switch(add) { ++ switch (add) { + case '-': + rg &= ~mask_bitfield; + break; +@@ -2191,13 +2428,13 @@ static int yaffs_proc_write(struct file + + yaffs_traceMask = rg | YAFFS_TRACE_ALWAYS; + +- printk("new trace = 0x%08X\n",yaffs_traceMask); ++ printk(KERN_DEBUG "new trace = 0x%08X\n", yaffs_traceMask); + + if (rg & YAFFS_TRACE_ALWAYS) { + for (i = 0; mask_flags[i].mask_name != NULL; i++) { + char flag; + flag = ((rg & mask_flags[i].mask_bitfield) == mask_flags[i].mask_bitfield) ? '+' : '-'; +- printk("%c%s\n", flag, mask_flags[i].mask_name); ++ printk(KERN_DEBUG "%c%s\n", flag, mask_flags[i].mask_name); + } + } + +@@ -2211,12 +2448,8 @@ struct file_system_to_install { + }; + + static struct file_system_to_install fs_to_install[] = { +-//#ifdef CONFIG_YAFFS_YAFFS1 + {&yaffs_fs_type, 0}, +-//#endif +-//#ifdef CONFIG_YAFFS_YAFFS2 + {&yaffs2_fs_type, 0}, +-//#endif + {NULL, 0} + }; + +@@ -2231,15 +2464,14 @@ static int __init init_yaffs_fs(void) + /* Install the proc_fs entry */ + my_proc_entry = create_proc_entry("yaffs", + S_IRUGO | S_IFREG, +- &proc_root); ++ YPROC_ROOT); + + if (my_proc_entry) { + my_proc_entry->write_proc = yaffs_proc_write; + my_proc_entry->read_proc = yaffs_proc_read; + my_proc_entry->data = NULL; +- } else { ++ } else + return -ENOMEM; +- } + + /* Now add the file system entries */ + +@@ -2247,9 +2479,8 @@ static int __init init_yaffs_fs(void) + + while (fsinst->fst && !error) { + error = register_filesystem(fsinst->fst); +- if (!error) { ++ if (!error) + fsinst->installed = 1; +- } + fsinst++; + } + +@@ -2277,7 +2508,7 @@ static void __exit exit_yaffs_fs(void) + T(YAFFS_TRACE_ALWAYS, ("yaffs " __DATE__ " " __TIME__ + " removing. \n")); + +- remove_proc_entry("yaffs", &proc_root); ++ remove_proc_entry("yaffs", YPROC_ROOT); + + fsinst = fs_to_install; + +@@ -2288,7 +2519,6 @@ static void __exit exit_yaffs_fs(void) + } + fsinst++; + } +- + } + + module_init(init_yaffs_fs) +--- /dev/null ++++ b/fs/yaffs2/yaffs_getblockinfo.h +@@ -0,0 +1,34 @@ ++/* ++ * YAFFS: Yet another Flash File System . A NAND-flash specific file system. ++ * ++ * Copyright (C) 2002-2007 Aleph One Ltd. ++ * for Toby Churchill Ltd and Brightstar Engineering ++ * ++ * Created by Charles Manning ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU Lesser General Public License version 2.1 as ++ * published by the Free Software Foundation. ++ * ++ * Note: Only YAFFS headers are LGPL, YAFFS C code is covered by GPL. ++ */ ++ ++#ifndef __YAFFS_GETBLOCKINFO_H__ ++#define __YAFFS_GETBLOCKINFO_H__ ++ ++#include "yaffs_guts.h" ++ ++/* Function to manipulate block info */ ++static Y_INLINE yaffs_BlockInfo *yaffs_GetBlockInfo(yaffs_Device * dev, int blk) ++{ ++ if (blk < dev->internalStartBlock || blk > dev->internalEndBlock) { ++ T(YAFFS_TRACE_ERROR, ++ (TSTR ++ ("**>> yaffs: getBlockInfo block %d is not valid" TENDSTR), ++ blk)); ++ YBUG(); ++ } ++ return &dev->blockInfo[blk - dev->internalStartBlock]; ++} ++ ++#endif +--- a/fs/yaffs2/yaffs_guts.c ++++ b/fs/yaffs2/yaffs_guts.c +@@ -12,16 +12,17 @@ + */ + + const char *yaffs_guts_c_version = +- "$Id: yaffs_guts.c,v 1.49 2007-05-15 20:07:40 charles Exp $"; ++ "$Id: yaffs_guts.c,v 1.82 2009-03-09 04:24:17 charles Exp $"; + + #include "yportenv.h" + + #include "yaffsinterface.h" + #include "yaffs_guts.h" + #include "yaffs_tagsvalidity.h" ++#include "yaffs_getblockinfo.h" + + #include "yaffs_tagscompat.h" +-#ifndef CONFIG_YAFFS_USE_OWN_SORT ++#ifndef CONFIG_YAFFS_USE_OWN_SORT + #include "yaffs_qsort.h" + #endif + #include "yaffs_nand.h" +@@ -32,116 +33,116 @@ const char *yaffs_guts_c_version = + #include "yaffs_packedtags2.h" + + +-#ifdef CONFIG_YAFFS_WINCE +-void yfsd_LockYAFFS(BOOL fsLockOnly); +-void yfsd_UnlockYAFFS(BOOL fsLockOnly); +-#endif +- + #define YAFFS_PASSIVE_GC_CHUNKS 2 + + #include "yaffs_ecc.h" + + + /* Robustification (if it ever comes about...) */ +-static void yaffs_RetireBlock(yaffs_Device * dev, int blockInNAND); +-static void yaffs_HandleWriteChunkError(yaffs_Device * dev, int chunkInNAND, int erasedOk); +-static void yaffs_HandleWriteChunkOk(yaffs_Device * dev, int chunkInNAND, +- const __u8 * data, +- const yaffs_ExtendedTags * tags); +-static void yaffs_HandleUpdateChunk(yaffs_Device * dev, int chunkInNAND, +- const yaffs_ExtendedTags * tags); ++static void yaffs_RetireBlock(yaffs_Device *dev, int blockInNAND); ++static void yaffs_HandleWriteChunkError(yaffs_Device *dev, int chunkInNAND, ++ int erasedOk); ++static void yaffs_HandleWriteChunkOk(yaffs_Device *dev, int chunkInNAND, ++ const __u8 *data, ++ const yaffs_ExtendedTags *tags); ++static void yaffs_HandleUpdateChunk(yaffs_Device *dev, int chunkInNAND, ++ const yaffs_ExtendedTags *tags); + + /* Other local prototypes */ +-static int yaffs_UnlinkObject( yaffs_Object *obj); ++static int yaffs_UnlinkObject(yaffs_Object *obj); + static int yaffs_ObjectHasCachedWriteData(yaffs_Object *obj); + + static void yaffs_HardlinkFixup(yaffs_Device *dev, yaffs_Object *hardList); + +-static int yaffs_WriteNewChunkWithTagsToNAND(yaffs_Device * dev, +- const __u8 * buffer, +- yaffs_ExtendedTags * tags, +- int useReserve); +-static int yaffs_PutChunkIntoFile(yaffs_Object * in, int chunkInInode, +- int chunkInNAND, int inScan); +- +-static yaffs_Object *yaffs_CreateNewObject(yaffs_Device * dev, int number, +- yaffs_ObjectType type); +-static void yaffs_AddObjectToDirectory(yaffs_Object * directory, +- yaffs_Object * obj); +-static int yaffs_UpdateObjectHeader(yaffs_Object * in, const YCHAR * name, +- int force, int isShrink, int shadows); +-static void yaffs_RemoveObjectFromDirectory(yaffs_Object * obj); ++static int yaffs_WriteNewChunkWithTagsToNAND(yaffs_Device *dev, ++ const __u8 *buffer, ++ yaffs_ExtendedTags *tags, ++ int useReserve); ++static int yaffs_PutChunkIntoFile(yaffs_Object *in, int chunkInInode, ++ int chunkInNAND, int inScan); ++ ++static yaffs_Object *yaffs_CreateNewObject(yaffs_Device *dev, int number, ++ yaffs_ObjectType type); ++static void yaffs_AddObjectToDirectory(yaffs_Object *directory, ++ yaffs_Object *obj); ++static int yaffs_UpdateObjectHeader(yaffs_Object *in, const YCHAR *name, ++ int force, int isShrink, int shadows); ++static void yaffs_RemoveObjectFromDirectory(yaffs_Object *obj); + static int yaffs_CheckStructures(void); +-static int yaffs_DeleteWorker(yaffs_Object * in, yaffs_Tnode * tn, __u32 level, +- int chunkOffset, int *limit); +-static int yaffs_DoGenericObjectDeletion(yaffs_Object * in); +- +-static yaffs_BlockInfo *yaffs_GetBlockInfo(yaffs_Device * dev, int blockNo); +- +-static __u8 *yaffs_GetTempBuffer(yaffs_Device * dev, int lineNo); +-static void yaffs_ReleaseTempBuffer(yaffs_Device * dev, __u8 * buffer, +- int lineNo); ++static int yaffs_DeleteWorker(yaffs_Object *in, yaffs_Tnode *tn, __u32 level, ++ int chunkOffset, int *limit); ++static int yaffs_DoGenericObjectDeletion(yaffs_Object *in); ++ ++static yaffs_BlockInfo *yaffs_GetBlockInfo(yaffs_Device *dev, int blockNo); + +-static int yaffs_CheckChunkErased(struct yaffs_DeviceStruct *dev, +- int chunkInNAND); + +-static int yaffs_UnlinkWorker(yaffs_Object * obj); +-static void yaffs_DestroyObject(yaffs_Object * obj); ++static int yaffs_CheckChunkErased(struct yaffs_DeviceStruct *dev, ++ int chunkInNAND); + +-static int yaffs_TagsMatch(const yaffs_ExtendedTags * tags, int objectId, +- int chunkInObject); ++static int yaffs_UnlinkWorker(yaffs_Object *obj); + +-loff_t yaffs_GetFileSize(yaffs_Object * obj); ++static int yaffs_TagsMatch(const yaffs_ExtendedTags *tags, int objectId, ++ int chunkInObject); + +-static int yaffs_AllocateChunk(yaffs_Device * dev, int useReserve, yaffs_BlockInfo **blockUsedPtr); ++static int yaffs_AllocateChunk(yaffs_Device *dev, int useReserve, ++ yaffs_BlockInfo **blockUsedPtr); + +-static void yaffs_VerifyFreeChunks(yaffs_Device * dev); ++static void yaffs_VerifyFreeChunks(yaffs_Device *dev); + + static void yaffs_CheckObjectDetailsLoaded(yaffs_Object *in); + ++static void yaffs_VerifyDirectory(yaffs_Object *directory); + #ifdef YAFFS_PARANOID +-static int yaffs_CheckFileSanity(yaffs_Object * in); ++static int yaffs_CheckFileSanity(yaffs_Object *in); + #else + #define yaffs_CheckFileSanity(in) + #endif + +-static void yaffs_InvalidateWholeChunkCache(yaffs_Object * in); +-static void yaffs_InvalidateChunkCache(yaffs_Object * object, int chunkId); ++static void yaffs_InvalidateWholeChunkCache(yaffs_Object *in); ++static void yaffs_InvalidateChunkCache(yaffs_Object *object, int chunkId); + + static void yaffs_InvalidateCheckpoint(yaffs_Device *dev); + +-static int yaffs_FindChunkInFile(yaffs_Object * in, int chunkInInode, +- yaffs_ExtendedTags * tags); ++static int yaffs_FindChunkInFile(yaffs_Object *in, int chunkInInode, ++ yaffs_ExtendedTags *tags); + +-static __u32 yaffs_GetChunkGroupBase(yaffs_Device *dev, yaffs_Tnode *tn, unsigned pos); +-static yaffs_Tnode *yaffs_FindLevel0Tnode(yaffs_Device * dev, +- yaffs_FileStructure * fStruct, +- __u32 chunkId); ++static __u32 yaffs_GetChunkGroupBase(yaffs_Device *dev, yaffs_Tnode *tn, ++ unsigned pos); ++static yaffs_Tnode *yaffs_FindLevel0Tnode(yaffs_Device *dev, ++ yaffs_FileStructure *fStruct, ++ __u32 chunkId); + + + /* Function to calculate chunk and offset */ + +-static void yaffs_AddrToChunk(yaffs_Device *dev, loff_t addr, __u32 *chunk, __u32 *offset) ++static void yaffs_AddrToChunk(yaffs_Device *dev, loff_t addr, int *chunkOut, ++ __u32 *offsetOut) + { +- if(dev->chunkShift){ +- /* Easy-peasy power of 2 case */ +- *chunk = (__u32)(addr >> dev->chunkShift); +- *offset = (__u32)(addr & dev->chunkMask); +- } +- else if(dev->crumbsPerChunk) +- { +- /* Case where we're using "crumbs" */ +- *offset = (__u32)(addr & dev->crumbMask); +- addr >>= dev->crumbShift; +- *chunk = ((__u32)addr)/dev->crumbsPerChunk; +- *offset += ((addr - (*chunk * dev->crumbsPerChunk)) << dev->crumbShift); ++ int chunk; ++ __u32 offset; ++ ++ chunk = (__u32)(addr >> dev->chunkShift); ++ ++ if (dev->chunkDiv == 1) { ++ /* easy power of 2 case */ ++ offset = (__u32)(addr & dev->chunkMask); ++ } else { ++ /* Non power-of-2 case */ ++ ++ loff_t chunkBase; ++ ++ chunk /= dev->chunkDiv; ++ ++ chunkBase = ((loff_t)chunk) * dev->nDataBytesPerChunk; ++ offset = (__u32)(addr - chunkBase); + } +- else +- YBUG(); ++ ++ *chunkOut = chunk; ++ *offsetOut = offset; + } + +-/* Function to return the number of shifts for a power of 2 greater than or equal +- * to the given number ++/* Function to return the number of shifts for a power of 2 greater than or ++ * equal to the given number + * Note we don't try to cater for all possible numbers and this does not have to + * be hellishly efficient. + */ +@@ -153,13 +154,14 @@ static __u32 ShiftsGE(__u32 x) + + nShifts = extraBits = 0; + +- while(x>1){ +- if(x & 1) extraBits++; +- x>>=1; ++ while (x > 1) { ++ if (x & 1) ++ extraBits++; ++ x >>= 1; + nShifts++; + } + +- if(extraBits) ++ if (extraBits) + nShifts++; + + return nShifts; +@@ -168,16 +170,17 @@ static __u32 ShiftsGE(__u32 x) + /* Function to return the number of shifts to get a 1 in bit 0 + */ + +-static __u32 ShiftDiv(__u32 x) ++static __u32 Shifts(__u32 x) + { + int nShifts; + + nShifts = 0; + +- if(!x) return 0; ++ if (!x) ++ return 0; + +- while( !(x&1)){ +- x>>=1; ++ while (!(x&1)) { ++ x >>= 1; + nShifts++; + } + +@@ -195,21 +198,25 @@ static int yaffs_InitialiseTempBuffers(y + int i; + __u8 *buf = (__u8 *)1; + +- memset(dev->tempBuffer,0,sizeof(dev->tempBuffer)); ++ memset(dev->tempBuffer, 0, sizeof(dev->tempBuffer)); + + for (i = 0; buf && i < YAFFS_N_TEMP_BUFFERS; i++) { + dev->tempBuffer[i].line = 0; /* not in use */ + dev->tempBuffer[i].buffer = buf = +- YMALLOC_DMA(dev->nDataBytesPerChunk); ++ YMALLOC_DMA(dev->totalBytesPerChunk); + } + + return buf ? YAFFS_OK : YAFFS_FAIL; +- + } + +-static __u8 *yaffs_GetTempBuffer(yaffs_Device * dev, int lineNo) ++__u8 *yaffs_GetTempBuffer(yaffs_Device *dev, int lineNo) + { + int i, j; ++ ++ dev->tempInUse++; ++ if (dev->tempInUse > dev->maxTemp) ++ dev->maxTemp = dev->tempInUse; ++ + for (i = 0; i < YAFFS_N_TEMP_BUFFERS; i++) { + if (dev->tempBuffer[i].line == 0) { + dev->tempBuffer[i].line = lineNo; +@@ -227,9 +234,9 @@ static __u8 *yaffs_GetTempBuffer(yaffs_D + T(YAFFS_TRACE_BUFFERS, + (TSTR("Out of temp buffers at line %d, other held by lines:"), + lineNo)); +- for (i = 0; i < YAFFS_N_TEMP_BUFFERS; i++) { ++ for (i = 0; i < YAFFS_N_TEMP_BUFFERS; i++) + T(YAFFS_TRACE_BUFFERS, (TSTR(" %d "), dev->tempBuffer[i].line)); +- } ++ + T(YAFFS_TRACE_BUFFERS, (TSTR(" " TENDSTR))); + + /* +@@ -242,10 +249,13 @@ static __u8 *yaffs_GetTempBuffer(yaffs_D + + } + +-static void yaffs_ReleaseTempBuffer(yaffs_Device * dev, __u8 * buffer, ++void yaffs_ReleaseTempBuffer(yaffs_Device *dev, __u8 *buffer, + int lineNo) + { + int i; ++ ++ dev->tempInUse--; ++ + for (i = 0; i < YAFFS_N_TEMP_BUFFERS; i++) { + if (dev->tempBuffer[i].buffer == buffer) { + dev->tempBuffer[i].line = 0; +@@ -267,27 +277,26 @@ static void yaffs_ReleaseTempBuffer(yaff + /* + * Determine if we have a managed buffer. + */ +-int yaffs_IsManagedTempBuffer(yaffs_Device * dev, const __u8 * buffer) ++int yaffs_IsManagedTempBuffer(yaffs_Device *dev, const __u8 *buffer) + { + int i; ++ + for (i = 0; i < YAFFS_N_TEMP_BUFFERS; i++) { + if (dev->tempBuffer[i].buffer == buffer) + return 1; ++ } + ++ for (i = 0; i < dev->nShortOpCaches; i++) { ++ if (dev->srCache[i].data == buffer) ++ return 1; + } + +- for (i = 0; i < dev->nShortOpCaches; i++) { +- if( dev->srCache[i].data == buffer ) +- return 1; +- +- } +- +- if (buffer == dev->checkpointBuffer) +- return 1; +- +- T(YAFFS_TRACE_ALWAYS, +- (TSTR("yaffs: unmaged buffer detected.\n" TENDSTR))); +- return 0; ++ if (buffer == dev->checkpointBuffer) ++ return 1; ++ ++ T(YAFFS_TRACE_ALWAYS, ++ (TSTR("yaffs: unmaged buffer detected.\n" TENDSTR))); ++ return 0; + } + + +@@ -296,62 +305,63 @@ int yaffs_IsManagedTempBuffer(yaffs_Devi + * Chunk bitmap manipulations + */ + +-static Y_INLINE __u8 *yaffs_BlockBits(yaffs_Device * dev, int blk) ++static Y_INLINE __u8 *yaffs_BlockBits(yaffs_Device *dev, int blk) + { + if (blk < dev->internalStartBlock || blk > dev->internalEndBlock) { + T(YAFFS_TRACE_ERROR, +- (TSTR("**>> yaffs: BlockBits block %d is not valid" TENDSTR), +- blk)); ++ (TSTR("**>> yaffs: BlockBits block %d is not valid" TENDSTR), ++ blk)); + YBUG(); + } + return dev->chunkBits + +- (dev->chunkBitmapStride * (blk - dev->internalStartBlock)); ++ (dev->chunkBitmapStride * (blk - dev->internalStartBlock)); + } + + static Y_INLINE void yaffs_VerifyChunkBitId(yaffs_Device *dev, int blk, int chunk) + { +- if(blk < dev->internalStartBlock || blk > dev->internalEndBlock || +- chunk < 0 || chunk >= dev->nChunksPerBlock) { +- T(YAFFS_TRACE_ERROR, +- (TSTR("**>> yaffs: Chunk Id (%d:%d) invalid"TENDSTR),blk,chunk)); +- YBUG(); ++ if (blk < dev->internalStartBlock || blk > dev->internalEndBlock || ++ chunk < 0 || chunk >= dev->nChunksPerBlock) { ++ T(YAFFS_TRACE_ERROR, ++ (TSTR("**>> yaffs: Chunk Id (%d:%d) invalid"TENDSTR), ++ blk, chunk)); ++ YBUG(); + } + } + +-static Y_INLINE void yaffs_ClearChunkBits(yaffs_Device * dev, int blk) ++static Y_INLINE void yaffs_ClearChunkBits(yaffs_Device *dev, int blk) + { + __u8 *blkBits = yaffs_BlockBits(dev, blk); + + memset(blkBits, 0, dev->chunkBitmapStride); + } + +-static Y_INLINE void yaffs_ClearChunkBit(yaffs_Device * dev, int blk, int chunk) ++static Y_INLINE void yaffs_ClearChunkBit(yaffs_Device *dev, int blk, int chunk) + { + __u8 *blkBits = yaffs_BlockBits(dev, blk); + +- yaffs_VerifyChunkBitId(dev,blk,chunk); ++ yaffs_VerifyChunkBitId(dev, blk, chunk); + + blkBits[chunk / 8] &= ~(1 << (chunk & 7)); + } + +-static Y_INLINE void yaffs_SetChunkBit(yaffs_Device * dev, int blk, int chunk) ++static Y_INLINE void yaffs_SetChunkBit(yaffs_Device *dev, int blk, int chunk) + { + __u8 *blkBits = yaffs_BlockBits(dev, blk); + +- yaffs_VerifyChunkBitId(dev,blk,chunk); ++ yaffs_VerifyChunkBitId(dev, blk, chunk); + + blkBits[chunk / 8] |= (1 << (chunk & 7)); + } + +-static Y_INLINE int yaffs_CheckChunkBit(yaffs_Device * dev, int blk, int chunk) ++static Y_INLINE int yaffs_CheckChunkBit(yaffs_Device *dev, int blk, int chunk) + { + __u8 *blkBits = yaffs_BlockBits(dev, blk); +- yaffs_VerifyChunkBitId(dev,blk,chunk); ++ yaffs_VerifyChunkBitId(dev, blk, chunk); + + return (blkBits[chunk / 8] & (1 << (chunk & 7))) ? 1 : 0; + } + +-static Y_INLINE int yaffs_StillSomeChunkBits(yaffs_Device * dev, int blk) ++static Y_INLINE int yaffs_StillSomeChunkBits(yaffs_Device *dev, int blk) + { + __u8 *blkBits = yaffs_BlockBits(dev, blk); + int i; +@@ -363,17 +373,17 @@ static Y_INLINE int yaffs_StillSomeChunk + return 0; + } + +-static int yaffs_CountChunkBits(yaffs_Device * dev, int blk) ++static int yaffs_CountChunkBits(yaffs_Device *dev, int blk) + { + __u8 *blkBits = yaffs_BlockBits(dev, blk); + int i; + int n = 0; + for (i = 0; i < dev->chunkBitmapStride; i++) { + __u8 x = *blkBits; +- while(x){ +- if(x & 1) ++ while (x) { ++ if (x & 1) + n++; +- x >>=1; ++ x >>= 1; + } + + blkBits++; +@@ -400,7 +410,7 @@ static int yaffs_SkipNANDVerification(ya + return !(yaffs_traceMask & (YAFFS_TRACE_VERIFY_NAND)); + } + +-static const char * blockStateName[] = { ++static const char *blockStateName[] = { + "Unknown", + "Needs scanning", + "Scanning", +@@ -413,64 +423,65 @@ static const char * blockStateName[] = { + "Dead" + }; + +-static void yaffs_VerifyBlock(yaffs_Device *dev,yaffs_BlockInfo *bi,int n) ++static void yaffs_VerifyBlock(yaffs_Device *dev, yaffs_BlockInfo *bi, int n) + { + int actuallyUsed; + int inUse; + +- if(yaffs_SkipVerification(dev)) ++ if (yaffs_SkipVerification(dev)) + return; + + /* Report illegal runtime states */ +- if(bi->blockState <0 || bi->blockState >= YAFFS_NUMBER_OF_BLOCK_STATES) +- T(YAFFS_TRACE_VERIFY,(TSTR("Block %d has undefined state %d"TENDSTR),n,bi->blockState)); ++ if (bi->blockState >= YAFFS_NUMBER_OF_BLOCK_STATES) ++ T(YAFFS_TRACE_VERIFY, (TSTR("Block %d has undefined state %d"TENDSTR), n, bi->blockState)); + +- switch(bi->blockState){ +- case YAFFS_BLOCK_STATE_UNKNOWN: +- case YAFFS_BLOCK_STATE_SCANNING: +- case YAFFS_BLOCK_STATE_NEEDS_SCANNING: +- T(YAFFS_TRACE_VERIFY,(TSTR("Block %d has bad run-state %s"TENDSTR), +- n,blockStateName[bi->blockState])); ++ switch (bi->blockState) { ++ case YAFFS_BLOCK_STATE_UNKNOWN: ++ case YAFFS_BLOCK_STATE_SCANNING: ++ case YAFFS_BLOCK_STATE_NEEDS_SCANNING: ++ T(YAFFS_TRACE_VERIFY, (TSTR("Block %d has bad run-state %s"TENDSTR), ++ n, blockStateName[bi->blockState])); + } + + /* Check pages in use and soft deletions are legal */ + + actuallyUsed = bi->pagesInUse - bi->softDeletions; + +- if(bi->pagesInUse < 0 || bi->pagesInUse > dev->nChunksPerBlock || ++ if (bi->pagesInUse < 0 || bi->pagesInUse > dev->nChunksPerBlock || + bi->softDeletions < 0 || bi->softDeletions > dev->nChunksPerBlock || + actuallyUsed < 0 || actuallyUsed > dev->nChunksPerBlock) +- T(YAFFS_TRACE_VERIFY,(TSTR("Block %d has illegal values pagesInUsed %d softDeletions %d"TENDSTR), +- n,bi->pagesInUse,bi->softDeletions)); ++ T(YAFFS_TRACE_VERIFY, (TSTR("Block %d has illegal values pagesInUsed %d softDeletions %d"TENDSTR), ++ n, bi->pagesInUse, bi->softDeletions)); + + + /* Check chunk bitmap legal */ +- inUse = yaffs_CountChunkBits(dev,n); +- if(inUse != bi->pagesInUse) +- T(YAFFS_TRACE_VERIFY,(TSTR("Block %d has inconsistent values pagesInUse %d counted chunk bits %d"TENDSTR), +- n,bi->pagesInUse,inUse)); ++ inUse = yaffs_CountChunkBits(dev, n); ++ if (inUse != bi->pagesInUse) ++ T(YAFFS_TRACE_VERIFY, (TSTR("Block %d has inconsistent values pagesInUse %d counted chunk bits %d"TENDSTR), ++ n, bi->pagesInUse, inUse)); + + /* Check that the sequence number is valid. + * Ten million is legal, but is very unlikely + */ +- if(dev->isYaffs2 && ++ if (dev->isYaffs2 && + (bi->blockState == YAFFS_BLOCK_STATE_ALLOCATING || bi->blockState == YAFFS_BLOCK_STATE_FULL) && +- (bi->sequenceNumber < YAFFS_LOWEST_SEQUENCE_NUMBER || bi->sequenceNumber > 10000000 )) +- T(YAFFS_TRACE_VERIFY,(TSTR("Block %d has suspect sequence number of %d"TENDSTR), +- n,bi->sequenceNumber)); +- ++ (bi->sequenceNumber < YAFFS_LOWEST_SEQUENCE_NUMBER || bi->sequenceNumber > 10000000)) ++ T(YAFFS_TRACE_VERIFY, (TSTR("Block %d has suspect sequence number of %d"TENDSTR), ++ n, bi->sequenceNumber)); + } + +-static void yaffs_VerifyCollectedBlock(yaffs_Device *dev,yaffs_BlockInfo *bi,int n) ++static void yaffs_VerifyCollectedBlock(yaffs_Device *dev, yaffs_BlockInfo *bi, ++ int n) + { +- yaffs_VerifyBlock(dev,bi,n); ++ yaffs_VerifyBlock(dev, bi, n); + + /* After collection the block should be in the erased state */ +- /* TODO: This will need to change if we do partial gc */ ++ /* This will need to change if we do partial gc */ + +- if(bi->blockState != YAFFS_BLOCK_STATE_EMPTY){ +- T(YAFFS_TRACE_ERROR,(TSTR("Block %d is in state %d after gc, should be erased"TENDSTR), +- n,bi->blockState)); ++ if (bi->blockState != YAFFS_BLOCK_STATE_COLLECTING && ++ bi->blockState != YAFFS_BLOCK_STATE_EMPTY) { ++ T(YAFFS_TRACE_ERROR, (TSTR("Block %d is in state %d after gc, should be erased"TENDSTR), ++ n, bi->blockState)); + } + } + +@@ -480,52 +491,49 @@ static void yaffs_VerifyBlocks(yaffs_Dev + int nBlocksPerState[YAFFS_NUMBER_OF_BLOCK_STATES]; + int nIllegalBlockStates = 0; + +- +- if(yaffs_SkipVerification(dev)) ++ if (yaffs_SkipVerification(dev)) + return; + +- memset(nBlocksPerState,0,sizeof(nBlocksPerState)); +- ++ memset(nBlocksPerState, 0, sizeof(nBlocksPerState)); + +- for(i = dev->internalStartBlock; i <= dev->internalEndBlock; i++){ +- yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev,i); +- yaffs_VerifyBlock(dev,bi,i); ++ for (i = dev->internalStartBlock; i <= dev->internalEndBlock; i++) { ++ yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev, i); ++ yaffs_VerifyBlock(dev, bi, i); + +- if(bi->blockState >=0 && bi->blockState < YAFFS_NUMBER_OF_BLOCK_STATES) ++ if (bi->blockState < YAFFS_NUMBER_OF_BLOCK_STATES) + nBlocksPerState[bi->blockState]++; + else + nIllegalBlockStates++; +- + } + +- T(YAFFS_TRACE_VERIFY,(TSTR(""TENDSTR))); +- T(YAFFS_TRACE_VERIFY,(TSTR("Block summary"TENDSTR))); ++ T(YAFFS_TRACE_VERIFY, (TSTR(""TENDSTR))); ++ T(YAFFS_TRACE_VERIFY, (TSTR("Block summary"TENDSTR))); + +- T(YAFFS_TRACE_VERIFY,(TSTR("%d blocks have illegal states"TENDSTR),nIllegalBlockStates)); +- if(nBlocksPerState[YAFFS_BLOCK_STATE_ALLOCATING] > 1) +- T(YAFFS_TRACE_VERIFY,(TSTR("Too many allocating blocks"TENDSTR))); ++ T(YAFFS_TRACE_VERIFY, (TSTR("%d blocks have illegal states"TENDSTR), nIllegalBlockStates)); ++ if (nBlocksPerState[YAFFS_BLOCK_STATE_ALLOCATING] > 1) ++ T(YAFFS_TRACE_VERIFY, (TSTR("Too many allocating blocks"TENDSTR))); + +- for(i = 0; i < YAFFS_NUMBER_OF_BLOCK_STATES; i++) ++ for (i = 0; i < YAFFS_NUMBER_OF_BLOCK_STATES; i++) + T(YAFFS_TRACE_VERIFY, + (TSTR("%s %d blocks"TENDSTR), +- blockStateName[i],nBlocksPerState[i])); ++ blockStateName[i], nBlocksPerState[i])); + +- if(dev->blocksInCheckpoint != nBlocksPerState[YAFFS_BLOCK_STATE_CHECKPOINT]) ++ if (dev->blocksInCheckpoint != nBlocksPerState[YAFFS_BLOCK_STATE_CHECKPOINT]) + T(YAFFS_TRACE_VERIFY, + (TSTR("Checkpoint block count wrong dev %d count %d"TENDSTR), + dev->blocksInCheckpoint, nBlocksPerState[YAFFS_BLOCK_STATE_CHECKPOINT])); + +- if(dev->nErasedBlocks != nBlocksPerState[YAFFS_BLOCK_STATE_EMPTY]) ++ if (dev->nErasedBlocks != nBlocksPerState[YAFFS_BLOCK_STATE_EMPTY]) + T(YAFFS_TRACE_VERIFY, + (TSTR("Erased block count wrong dev %d count %d"TENDSTR), + dev->nErasedBlocks, nBlocksPerState[YAFFS_BLOCK_STATE_EMPTY])); + +- if(nBlocksPerState[YAFFS_BLOCK_STATE_COLLECTING] > 1) ++ if (nBlocksPerState[YAFFS_BLOCK_STATE_COLLECTING] > 1) + T(YAFFS_TRACE_VERIFY, + (TSTR("Too many collecting blocks %d (max is 1)"TENDSTR), + nBlocksPerState[YAFFS_BLOCK_STATE_COLLECTING])); + +- T(YAFFS_TRACE_VERIFY,(TSTR(""TENDSTR))); ++ T(YAFFS_TRACE_VERIFY, (TSTR(""TENDSTR))); + + } + +@@ -535,26 +543,26 @@ static void yaffs_VerifyBlocks(yaffs_Dev + */ + static void yaffs_VerifyObjectHeader(yaffs_Object *obj, yaffs_ObjectHeader *oh, yaffs_ExtendedTags *tags, int parentCheck) + { +- if(yaffs_SkipVerification(obj->myDev)) ++ if (obj && yaffs_SkipVerification(obj->myDev)) + return; + +- if(!(tags && obj && oh)){ +- T(YAFFS_TRACE_VERIFY, +- (TSTR("Verifying object header tags %x obj %x oh %x"TENDSTR), +- (__u32)tags,(__u32)obj,(__u32)oh)); ++ if (!(tags && obj && oh)) { ++ T(YAFFS_TRACE_VERIFY, ++ (TSTR("Verifying object header tags %x obj %x oh %x"TENDSTR), ++ (__u32)tags, (__u32)obj, (__u32)oh)); + return; + } + +- if(oh->type <= YAFFS_OBJECT_TYPE_UNKNOWN || +- oh->type > YAFFS_OBJECT_TYPE_MAX) +- T(YAFFS_TRACE_VERIFY, +- (TSTR("Obj %d header type is illegal value 0x%x"TENDSTR), +- tags->objectId, oh->type)); +- +- if(tags->objectId != obj->objectId) +- T(YAFFS_TRACE_VERIFY, +- (TSTR("Obj %d header mismatch objectId %d"TENDSTR), +- tags->objectId, obj->objectId)); ++ if (oh->type <= YAFFS_OBJECT_TYPE_UNKNOWN || ++ oh->type > YAFFS_OBJECT_TYPE_MAX) ++ T(YAFFS_TRACE_VERIFY, ++ (TSTR("Obj %d header type is illegal value 0x%x"TENDSTR), ++ tags->objectId, oh->type)); ++ ++ if (tags->objectId != obj->objectId) ++ T(YAFFS_TRACE_VERIFY, ++ (TSTR("Obj %d header mismatch objectId %d"TENDSTR), ++ tags->objectId, obj->objectId)); + + + /* +@@ -563,46 +571,43 @@ static void yaffs_VerifyObjectHeader(yaf + * Tests do not apply to the root object. + */ + +- if(parentCheck && tags->objectId > 1 && !obj->parent) +- T(YAFFS_TRACE_VERIFY, +- (TSTR("Obj %d header mismatch parentId %d obj->parent is NULL"TENDSTR), +- tags->objectId, oh->parentObjectId)); +- +- +- if(parentCheck && obj->parent && +- oh->parentObjectId != obj->parent->objectId && +- (oh->parentObjectId != YAFFS_OBJECTID_UNLINKED || +- obj->parent->objectId != YAFFS_OBJECTID_DELETED)) +- T(YAFFS_TRACE_VERIFY, +- (TSTR("Obj %d header mismatch parentId %d parentObjectId %d"TENDSTR), +- tags->objectId, oh->parentObjectId, obj->parent->objectId)); ++ if (parentCheck && tags->objectId > 1 && !obj->parent) ++ T(YAFFS_TRACE_VERIFY, ++ (TSTR("Obj %d header mismatch parentId %d obj->parent is NULL"TENDSTR), ++ tags->objectId, oh->parentObjectId)); + ++ if (parentCheck && obj->parent && ++ oh->parentObjectId != obj->parent->objectId && ++ (oh->parentObjectId != YAFFS_OBJECTID_UNLINKED || ++ obj->parent->objectId != YAFFS_OBJECTID_DELETED)) ++ T(YAFFS_TRACE_VERIFY, ++ (TSTR("Obj %d header mismatch parentId %d parentObjectId %d"TENDSTR), ++ tags->objectId, oh->parentObjectId, obj->parent->objectId)); + +- if(tags->objectId > 1 && oh->name[0] == 0) /* Null name */ ++ if (tags->objectId > 1 && oh->name[0] == 0) /* Null name */ + T(YAFFS_TRACE_VERIFY, +- (TSTR("Obj %d header name is NULL"TENDSTR), +- obj->objectId)); ++ (TSTR("Obj %d header name is NULL"TENDSTR), ++ obj->objectId)); + +- if(tags->objectId > 1 && ((__u8)(oh->name[0])) == 0xff) /* Trashed name */ ++ if (tags->objectId > 1 && ((__u8)(oh->name[0])) == 0xff) /* Trashed name */ + T(YAFFS_TRACE_VERIFY, +- (TSTR("Obj %d header name is 0xFF"TENDSTR), +- obj->objectId)); ++ (TSTR("Obj %d header name is 0xFF"TENDSTR), ++ obj->objectId)); + } + + + +-static int yaffs_VerifyTnodeWorker(yaffs_Object * obj, yaffs_Tnode * tn, +- __u32 level, int chunkOffset) ++static int yaffs_VerifyTnodeWorker(yaffs_Object *obj, yaffs_Tnode *tn, ++ __u32 level, int chunkOffset) + { + int i; + yaffs_Device *dev = obj->myDev; + int ok = 1; +- int nTnodeBytes = (dev->tnodeWidth * YAFFS_NTNODES_LEVEL0)/8; + + if (tn) { + if (level > 0) { + +- for (i = 0; i < YAFFS_NTNODES_INTERNAL && ok; i++){ ++ for (i = 0; i < YAFFS_NTNODES_INTERNAL && ok; i++) { + if (tn->internal[i]) { + ok = yaffs_VerifyTnodeWorker(obj, + tn->internal[i], +@@ -611,20 +616,19 @@ static int yaffs_VerifyTnodeWorker(yaffs + } + } + } else if (level == 0) { +- int i; + yaffs_ExtendedTags tags; + __u32 objectId = obj->objectId; + + chunkOffset <<= YAFFS_TNODES_LEVEL0_BITS; + +- for(i = 0; i < YAFFS_NTNODES_LEVEL0; i++){ +- __u32 theChunk = yaffs_GetChunkGroupBase(dev,tn,i); ++ for (i = 0; i < YAFFS_NTNODES_LEVEL0; i++) { ++ __u32 theChunk = yaffs_GetChunkGroupBase(dev, tn, i); + +- if(theChunk > 0){ ++ if (theChunk > 0) { + /* T(~0,(TSTR("verifying (%d:%d) %d"TENDSTR),tags.objectId,tags.chunkId,theChunk)); */ +- yaffs_ReadChunkWithTagsFromNAND(dev,theChunk,NULL, &tags); +- if(tags.objectId != objectId || tags.chunkId != chunkOffset){ +- T(~0,(TSTR("Object %d chunkId %d NAND mismatch chunk %d tags (%d:%d)"TENDSTR), ++ yaffs_ReadChunkWithTagsFromNAND(dev, theChunk, NULL, &tags); ++ if (tags.objectId != objectId || tags.chunkId != chunkOffset) { ++ T(~0, (TSTR("Object %d chunkId %d NAND mismatch chunk %d tags (%d:%d)"TENDSTR), + objectId, chunkOffset, theChunk, + tags.objectId, tags.chunkId)); + } +@@ -646,13 +650,15 @@ static void yaffs_VerifyFile(yaffs_Objec + __u32 lastChunk; + __u32 x; + __u32 i; +- int ok; + yaffs_Device *dev; + yaffs_ExtendedTags tags; + yaffs_Tnode *tn; + __u32 objectId; + +- if(obj && yaffs_SkipVerification(obj->myDev)) ++ if (!obj) ++ return; ++ ++ if (yaffs_SkipVerification(obj->myDev)) + return; + + dev = obj->myDev; +@@ -662,17 +668,17 @@ static void yaffs_VerifyFile(yaffs_Objec + lastChunk = obj->variant.fileVariant.fileSize / dev->nDataBytesPerChunk + 1; + x = lastChunk >> YAFFS_TNODES_LEVEL0_BITS; + requiredTallness = 0; +- while (x> 0) { ++ while (x > 0) { + x >>= YAFFS_TNODES_INTERNAL_BITS; + requiredTallness++; + } + + actualTallness = obj->variant.fileVariant.topLevel; + +- if(requiredTallness > actualTallness ) ++ if (requiredTallness > actualTallness) + T(YAFFS_TRACE_VERIFY, + (TSTR("Obj %d had tnode tallness %d, needs to be %d"TENDSTR), +- obj->objectId,actualTallness, requiredTallness)); ++ obj->objectId, actualTallness, requiredTallness)); + + + /* Check that the chunks in the tnode tree are all correct. +@@ -680,39 +686,31 @@ static void yaffs_VerifyFile(yaffs_Objec + * checking the tags for every chunk match. + */ + +- if(yaffs_SkipNANDVerification(dev)) ++ if (yaffs_SkipNANDVerification(dev)) + return; + +- for(i = 1; i <= lastChunk; i++){ +- tn = yaffs_FindLevel0Tnode(dev, &obj->variant.fileVariant,i); ++ for (i = 1; i <= lastChunk; i++) { ++ tn = yaffs_FindLevel0Tnode(dev, &obj->variant.fileVariant, i); + + if (tn) { +- __u32 theChunk = yaffs_GetChunkGroupBase(dev,tn,i); +- if(theChunk > 0){ ++ __u32 theChunk = yaffs_GetChunkGroupBase(dev, tn, i); ++ if (theChunk > 0) { + /* T(~0,(TSTR("verifying (%d:%d) %d"TENDSTR),objectId,i,theChunk)); */ +- yaffs_ReadChunkWithTagsFromNAND(dev,theChunk,NULL, &tags); +- if(tags.objectId != objectId || tags.chunkId != i){ +- T(~0,(TSTR("Object %d chunkId %d NAND mismatch chunk %d tags (%d:%d)"TENDSTR), ++ yaffs_ReadChunkWithTagsFromNAND(dev, theChunk, NULL, &tags); ++ if (tags.objectId != objectId || tags.chunkId != i) { ++ T(~0, (TSTR("Object %d chunkId %d NAND mismatch chunk %d tags (%d:%d)"TENDSTR), + objectId, i, theChunk, + tags.objectId, tags.chunkId)); + } + } + } +- + } +- + } + +-static void yaffs_VerifyDirectory(yaffs_Object *obj) +-{ +- if(obj && yaffs_SkipVerification(obj->myDev)) +- return; +- +-} + + static void yaffs_VerifyHardLink(yaffs_Object *obj) + { +- if(obj && yaffs_SkipVerification(obj->myDev)) ++ if (obj && yaffs_SkipVerification(obj->myDev)) + return; + + /* Verify sane equivalent object */ +@@ -720,7 +718,7 @@ static void yaffs_VerifyHardLink(yaffs_O + + static void yaffs_VerifySymlink(yaffs_Object *obj) + { +- if(obj && yaffs_SkipVerification(obj->myDev)) ++ if (obj && yaffs_SkipVerification(obj->myDev)) + return; + + /* Verify symlink string */ +@@ -728,7 +726,7 @@ static void yaffs_VerifySymlink(yaffs_Ob + + static void yaffs_VerifySpecial(yaffs_Object *obj) + { +- if(obj && yaffs_SkipVerification(obj->myDev)) ++ if (obj && yaffs_SkipVerification(obj->myDev)) + return; + } + +@@ -740,14 +738,19 @@ static void yaffs_VerifyObject(yaffs_Obj + __u32 chunkMax; + + __u32 chunkIdOk; +- __u32 chunkIsLive; ++ __u32 chunkInRange; ++ __u32 chunkShouldNotBeDeleted; ++ __u32 chunkValid; ++ ++ if (!obj) ++ return; + +- if(!obj) ++ if (obj->beingCreated) + return; + + dev = obj->myDev; + +- if(yaffs_SkipVerification(dev)) ++ if (yaffs_SkipVerification(dev)) + return; + + /* Check sane object header chunk */ +@@ -755,50 +758,54 @@ static void yaffs_VerifyObject(yaffs_Obj + chunkMin = dev->internalStartBlock * dev->nChunksPerBlock; + chunkMax = (dev->internalEndBlock+1) * dev->nChunksPerBlock - 1; + +- chunkIdOk = (obj->chunkId >= chunkMin && obj->chunkId <= chunkMax); +- chunkIsLive = chunkIdOk && ++ chunkInRange = (((unsigned)(obj->hdrChunk)) >= chunkMin && ((unsigned)(obj->hdrChunk)) <= chunkMax); ++ chunkIdOk = chunkInRange || obj->hdrChunk == 0; ++ chunkValid = chunkInRange && + yaffs_CheckChunkBit(dev, +- obj->chunkId / dev->nChunksPerBlock, +- obj->chunkId % dev->nChunksPerBlock); +- if(!obj->fake && +- (!chunkIdOk || !chunkIsLive)) { +- T(YAFFS_TRACE_VERIFY, +- (TSTR("Obj %d has chunkId %d %s %s"TENDSTR), +- obj->objectId,obj->chunkId, +- chunkIdOk ? "" : ",out of range", +- chunkIsLive || !chunkIdOk ? "" : ",marked as deleted")); ++ obj->hdrChunk / dev->nChunksPerBlock, ++ obj->hdrChunk % dev->nChunksPerBlock); ++ chunkShouldNotBeDeleted = chunkInRange && !chunkValid; ++ ++ if (!obj->fake && ++ (!chunkIdOk || chunkShouldNotBeDeleted)) { ++ T(YAFFS_TRACE_VERIFY, ++ (TSTR("Obj %d has chunkId %d %s %s"TENDSTR), ++ obj->objectId, obj->hdrChunk, ++ chunkIdOk ? "" : ",out of range", ++ chunkShouldNotBeDeleted ? ",marked as deleted" : "")); + } + +- if(chunkIdOk && chunkIsLive &&!yaffs_SkipNANDVerification(dev)) { ++ if (chunkValid && !yaffs_SkipNANDVerification(dev)) { + yaffs_ExtendedTags tags; + yaffs_ObjectHeader *oh; +- __u8 *buffer = yaffs_GetTempBuffer(dev,__LINE__); ++ __u8 *buffer = yaffs_GetTempBuffer(dev, __LINE__); + + oh = (yaffs_ObjectHeader *)buffer; + +- yaffs_ReadChunkWithTagsFromNAND(dev, obj->chunkId,buffer, &tags); ++ yaffs_ReadChunkWithTagsFromNAND(dev, obj->hdrChunk, buffer, ++ &tags); + +- yaffs_VerifyObjectHeader(obj,oh,&tags,1); ++ yaffs_VerifyObjectHeader(obj, oh, &tags, 1); + +- yaffs_ReleaseTempBuffer(dev,buffer,__LINE__); ++ yaffs_ReleaseTempBuffer(dev, buffer, __LINE__); + } + + /* Verify it has a parent */ +- if(obj && !obj->fake && +- (!obj->parent || obj->parent->myDev != dev)){ +- T(YAFFS_TRACE_VERIFY, +- (TSTR("Obj %d has parent pointer %p which does not look like an object"TENDSTR), +- obj->objectId,obj->parent)); ++ if (obj && !obj->fake && ++ (!obj->parent || obj->parent->myDev != dev)) { ++ T(YAFFS_TRACE_VERIFY, ++ (TSTR("Obj %d has parent pointer %p which does not look like an object"TENDSTR), ++ obj->objectId, obj->parent)); + } + + /* Verify parent is a directory */ +- if(obj->parent && obj->parent->variantType != YAFFS_OBJECT_TYPE_DIRECTORY){ +- T(YAFFS_TRACE_VERIFY, +- (TSTR("Obj %d's parent is not a directory (type %d)"TENDSTR), +- obj->objectId,obj->parent->variantType)); ++ if (obj->parent && obj->parent->variantType != YAFFS_OBJECT_TYPE_DIRECTORY) { ++ T(YAFFS_TRACE_VERIFY, ++ (TSTR("Obj %d's parent is not a directory (type %d)"TENDSTR), ++ obj->objectId, obj->parent->variantType)); + } + +- switch(obj->variantType){ ++ switch (obj->variantType) { + case YAFFS_OBJECT_TYPE_FILE: + yaffs_VerifyFile(obj); + break; +@@ -818,33 +825,30 @@ static void yaffs_VerifyObject(yaffs_Obj + default: + T(YAFFS_TRACE_VERIFY, + (TSTR("Obj %d has illegaltype %d"TENDSTR), +- obj->objectId,obj->variantType)); ++ obj->objectId, obj->variantType)); + break; + } +- +- + } + + static void yaffs_VerifyObjects(yaffs_Device *dev) + { + yaffs_Object *obj; + int i; +- struct list_head *lh; ++ struct ylist_head *lh; + +- if(yaffs_SkipVerification(dev)) ++ if (yaffs_SkipVerification(dev)) + return; + + /* Iterate through the objects in each hash entry */ + +- for(i = 0; i < YAFFS_NOBJECT_BUCKETS; i++){ +- list_for_each(lh, &dev->objectBucket[i].list) { ++ for (i = 0; i < YAFFS_NOBJECT_BUCKETS; i++) { ++ ylist_for_each(lh, &dev->objectBucket[i].list) { + if (lh) { +- obj = list_entry(lh, yaffs_Object, hashLink); ++ obj = ylist_entry(lh, yaffs_Object, hashLink); + yaffs_VerifyObject(obj); + } + } +- } +- ++ } + } + + +@@ -855,19 +859,20 @@ static void yaffs_VerifyObjects(yaffs_De + static Y_INLINE int yaffs_HashFunction(int n) + { + n = abs(n); +- return (n % YAFFS_NOBJECT_BUCKETS); ++ return n % YAFFS_NOBJECT_BUCKETS; + } + + /* +- * Access functions to useful fake objects ++ * Access functions to useful fake objects. ++ * Note that root might have a presence in NAND if permissions are set. + */ + +-yaffs_Object *yaffs_Root(yaffs_Device * dev) ++yaffs_Object *yaffs_Root(yaffs_Device *dev) + { + return dev->rootDir; + } + +-yaffs_Object *yaffs_LostNFound(yaffs_Device * dev) ++yaffs_Object *yaffs_LostNFound(yaffs_Device *dev) + { + return dev->lostNFoundDir; + } +@@ -877,7 +882,7 @@ yaffs_Object *yaffs_LostNFound(yaffs_Dev + * Erased NAND checking functions + */ + +-int yaffs_CheckFF(__u8 * buffer, int nBytes) ++int yaffs_CheckFF(__u8 *buffer, int nBytes) + { + /* Horrible, slow implementation */ + while (nBytes--) { +@@ -889,9 +894,8 @@ int yaffs_CheckFF(__u8 * buffer, int nBy + } + + static int yaffs_CheckChunkErased(struct yaffs_DeviceStruct *dev, +- int chunkInNAND) ++ int chunkInNAND) + { +- + int retval = YAFFS_OK; + __u8 *data = yaffs_GetTempBuffer(dev, __LINE__); + yaffs_ExtendedTags tags; +@@ -899,10 +903,9 @@ static int yaffs_CheckChunkErased(struct + + result = yaffs_ReadChunkWithTagsFromNAND(dev, chunkInNAND, data, &tags); + +- if(tags.eccResult > YAFFS_ECC_RESULT_NO_ERROR) ++ if (tags.eccResult > YAFFS_ECC_RESULT_NO_ERROR) + retval = YAFFS_FAIL; + +- + if (!yaffs_CheckFF(data, dev->nDataBytesPerChunk) || tags.chunkUsed) { + T(YAFFS_TRACE_NANDACCESS, + (TSTR("Chunk %d not erased" TENDSTR), chunkInNAND)); +@@ -915,11 +918,10 @@ static int yaffs_CheckChunkErased(struct + + } + +- + static int yaffs_WriteNewChunkWithTagsToNAND(struct yaffs_DeviceStruct *dev, +- const __u8 * data, +- yaffs_ExtendedTags * tags, +- int useReserve) ++ const __u8 *data, ++ yaffs_ExtendedTags *tags, ++ int useReserve) + { + int attempts = 0; + int writeOk = 0; +@@ -972,7 +974,7 @@ static int yaffs_WriteNewChunkWithTagsTo + erasedOk = yaffs_CheckChunkErased(dev, chunk); + if (erasedOk != YAFFS_OK) { + T(YAFFS_TRACE_ERROR, +- (TSTR ("**>> yaffs chunk %d was not erased" ++ (TSTR("**>> yaffs chunk %d was not erased" + TENDSTR), chunk)); + + /* try another chunk */ +@@ -992,7 +994,11 @@ static int yaffs_WriteNewChunkWithTagsTo + /* Copy the data into the robustification buffer */ + yaffs_HandleWriteChunkOk(dev, chunk, data, tags); + +- } while (writeOk != YAFFS_OK && attempts < yaffs_wr_attempts); ++ } while (writeOk != YAFFS_OK && ++ (yaffs_wr_attempts <= 0 || attempts <= yaffs_wr_attempts)); ++ ++ if (!writeOk) ++ chunk = -1; + + if (attempts > 1) { + T(YAFFS_TRACE_ERROR, +@@ -1009,13 +1015,35 @@ static int yaffs_WriteNewChunkWithTagsTo + * Block retiring for handling a broken block. + */ + +-static void yaffs_RetireBlock(yaffs_Device * dev, int blockInNAND) ++static void yaffs_RetireBlock(yaffs_Device *dev, int blockInNAND) + { + yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev, blockInNAND); + + yaffs_InvalidateCheckpoint(dev); + +- yaffs_MarkBlockBad(dev, blockInNAND); ++ if (yaffs_MarkBlockBad(dev, blockInNAND) != YAFFS_OK) { ++ if (yaffs_EraseBlockInNAND(dev, blockInNAND) != YAFFS_OK) { ++ T(YAFFS_TRACE_ALWAYS, (TSTR( ++ "yaffs: Failed to mark bad and erase block %d" ++ TENDSTR), blockInNAND)); ++ } else { ++ yaffs_ExtendedTags tags; ++ int chunkId = blockInNAND * dev->nChunksPerBlock; ++ ++ __u8 *buffer = yaffs_GetTempBuffer(dev, __LINE__); ++ ++ memset(buffer, 0xff, dev->nDataBytesPerChunk); ++ yaffs_InitialiseTags(&tags); ++ tags.sequenceNumber = YAFFS_SEQUENCE_BAD_BLOCK; ++ if (dev->writeChunkWithTagsToNAND(dev, chunkId - ++ dev->chunkOffset, buffer, &tags) != YAFFS_OK) ++ T(YAFFS_TRACE_ALWAYS, (TSTR("yaffs: Failed to " ++ TCONT("write bad block marker to block %d") ++ TENDSTR), blockInNAND)); ++ ++ yaffs_ReleaseTempBuffer(dev, buffer, __LINE__); ++ } ++ } + + bi->blockState = YAFFS_BLOCK_STATE_DEAD; + bi->gcPrioritise = 0; +@@ -1029,49 +1057,45 @@ static void yaffs_RetireBlock(yaffs_Devi + * + */ + +-static void yaffs_HandleWriteChunkOk(yaffs_Device * dev, int chunkInNAND, +- const __u8 * data, +- const yaffs_ExtendedTags * tags) ++static void yaffs_HandleWriteChunkOk(yaffs_Device *dev, int chunkInNAND, ++ const __u8 *data, ++ const yaffs_ExtendedTags *tags) + { + } + +-static void yaffs_HandleUpdateChunk(yaffs_Device * dev, int chunkInNAND, +- const yaffs_ExtendedTags * tags) ++static void yaffs_HandleUpdateChunk(yaffs_Device *dev, int chunkInNAND, ++ const yaffs_ExtendedTags *tags) + { + } + + void yaffs_HandleChunkError(yaffs_Device *dev, yaffs_BlockInfo *bi) + { +- if(!bi->gcPrioritise){ ++ if (!bi->gcPrioritise) { + bi->gcPrioritise = 1; + dev->hasPendingPrioritisedGCs = 1; +- bi->chunkErrorStrikes ++; ++ bi->chunkErrorStrikes++; + +- if(bi->chunkErrorStrikes > 3){ ++ if (bi->chunkErrorStrikes > 3) { + bi->needsRetiring = 1; /* Too many stikes, so retire this */ + T(YAFFS_TRACE_ALWAYS, (TSTR("yaffs: Block struck out" TENDSTR))); + + } +- + } + } + +-static void yaffs_HandleWriteChunkError(yaffs_Device * dev, int chunkInNAND, int erasedOk) ++static void yaffs_HandleWriteChunkError(yaffs_Device *dev, int chunkInNAND, ++ int erasedOk) + { +- + int blockInNAND = chunkInNAND / dev->nChunksPerBlock; + yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev, blockInNAND); + +- yaffs_HandleChunkError(dev,bi); ++ yaffs_HandleChunkError(dev, bi); + +- +- if(erasedOk ) { ++ if (erasedOk) { + /* Was an actual write failure, so mark the block for retirement */ + bi->needsRetiring = 1; + T(YAFFS_TRACE_ERROR | YAFFS_TRACE_BAD_BLOCKS, + (TSTR("**>> Block %d needs retiring" TENDSTR), blockInNAND)); +- +- + } + + /* Delete the chunk */ +@@ -1081,12 +1105,12 @@ static void yaffs_HandleWriteChunkError( + + /*---------------- Name handling functions ------------*/ + +-static __u16 yaffs_CalcNameSum(const YCHAR * name) ++static __u16 yaffs_CalcNameSum(const YCHAR *name) + { + __u16 sum = 0; + __u16 i = 1; + +- YUCHAR *bname = (YUCHAR *) name; ++ const YUCHAR *bname = (const YUCHAR *) name; + if (bname) { + while ((*bname) && (i < (YAFFS_MAX_NAME_LENGTH/2))) { + +@@ -1102,14 +1126,14 @@ static __u16 yaffs_CalcNameSum(const YCH + return sum; + } + +-static void yaffs_SetObjectName(yaffs_Object * obj, const YCHAR * name) ++static void yaffs_SetObjectName(yaffs_Object *obj, const YCHAR *name) + { + #ifdef CONFIG_YAFFS_SHORT_NAMES_IN_RAM +- if (name && yaffs_strlen(name) <= YAFFS_SHORT_NAME_LENGTH) { ++ memset(obj->shortName, 0, sizeof(YCHAR) * (YAFFS_SHORT_NAME_LENGTH+1)); ++ if (name && yaffs_strlen(name) <= YAFFS_SHORT_NAME_LENGTH) + yaffs_strcpy(obj->shortName, name); +- } else { ++ else + obj->shortName[0] = _Y('\0'); +- } + #endif + obj->sum = yaffs_CalcNameSum(name); + } +@@ -1126,7 +1150,7 @@ static void yaffs_SetObjectName(yaffs_Ob + * Don't use this function directly + */ + +-static int yaffs_CreateTnodes(yaffs_Device * dev, int nTnodes) ++static int yaffs_CreateTnodes(yaffs_Device *dev, int nTnodes) + { + int i; + int tnodeSize; +@@ -1143,6 +1167,9 @@ static int yaffs_CreateTnodes(yaffs_Devi + * Must be a multiple of 32-bits */ + tnodeSize = (dev->tnodeWidth * YAFFS_NTNODES_LEVEL0)/8; + ++ if (tnodeSize < sizeof(yaffs_Tnode)) ++ tnodeSize = sizeof(yaffs_Tnode); ++ + /* make these things */ + + newTnodes = YMALLOC(nTnodes * tnodeSize); +@@ -1150,7 +1177,7 @@ static int yaffs_CreateTnodes(yaffs_Devi + + if (!newTnodes) { + T(YAFFS_TRACE_ERROR, +- (TSTR("yaffs: Could not allocate Tnodes" TENDSTR))); ++ (TSTR("yaffs: Could not allocate Tnodes" TENDSTR))); + return YAFFS_FAIL; + } + +@@ -1170,7 +1197,7 @@ static int yaffs_CreateTnodes(yaffs_Devi + dev->freeTnodes = newTnodes; + #else + /* New hookup for wide tnodes */ +- for(i = 0; i < nTnodes -1; i++) { ++ for (i = 0; i < nTnodes - 1; i++) { + curr = (yaffs_Tnode *) &mem[i * tnodeSize]; + next = (yaffs_Tnode *) &mem[(i+1) * tnodeSize]; + curr->internal[0] = next; +@@ -1197,7 +1224,6 @@ static int yaffs_CreateTnodes(yaffs_Devi + (TSTR + ("yaffs: Could not add tnodes to management list" TENDSTR))); + return YAFFS_FAIL; +- + } else { + tnl->tnodes = newTnodes; + tnl->next = dev->allocatedTnodeList; +@@ -1211,14 +1237,13 @@ static int yaffs_CreateTnodes(yaffs_Devi + + /* GetTnode gets us a clean tnode. Tries to make allocate more if we run out */ + +-static yaffs_Tnode *yaffs_GetTnodeRaw(yaffs_Device * dev) ++static yaffs_Tnode *yaffs_GetTnodeRaw(yaffs_Device *dev) + { + yaffs_Tnode *tn = NULL; + + /* If there are none left make more */ +- if (!dev->freeTnodes) { ++ if (!dev->freeTnodes) + yaffs_CreateTnodes(dev, YAFFS_ALLOCATION_NTNODES); +- } + + if (dev->freeTnodes) { + tn = dev->freeTnodes; +@@ -1233,21 +1258,27 @@ static yaffs_Tnode *yaffs_GetTnodeRaw(ya + dev->nFreeTnodes--; + } + ++ dev->nCheckpointBlocksRequired = 0; /* force recalculation*/ ++ + return tn; + } + +-static yaffs_Tnode *yaffs_GetTnode(yaffs_Device * dev) ++static yaffs_Tnode *yaffs_GetTnode(yaffs_Device *dev) + { + yaffs_Tnode *tn = yaffs_GetTnodeRaw(dev); ++ int tnodeSize = (dev->tnodeWidth * YAFFS_NTNODES_LEVEL0)/8; + +- if(tn) +- memset(tn, 0, (dev->tnodeWidth * YAFFS_NTNODES_LEVEL0)/8); ++ if (tnodeSize < sizeof(yaffs_Tnode)) ++ tnodeSize = sizeof(yaffs_Tnode); ++ ++ if (tn) ++ memset(tn, 0, tnodeSize); + + return tn; + } + + /* FreeTnode frees up a tnode and puts it back on the free list */ +-static void yaffs_FreeTnode(yaffs_Device * dev, yaffs_Tnode * tn) ++static void yaffs_FreeTnode(yaffs_Device *dev, yaffs_Tnode *tn) + { + if (tn) { + #ifdef CONFIG_YAFFS_TNODE_LIST_DEBUG +@@ -1262,9 +1293,10 @@ static void yaffs_FreeTnode(yaffs_Device + dev->freeTnodes = tn; + dev->nFreeTnodes++; + } ++ dev->nCheckpointBlocksRequired = 0; /* force recalculation*/ + } + +-static void yaffs_DeinitialiseTnodes(yaffs_Device * dev) ++static void yaffs_DeinitialiseTnodes(yaffs_Device *dev) + { + /* Free the list of allocated tnodes */ + yaffs_TnodeList *tmp; +@@ -1282,71 +1314,72 @@ static void yaffs_DeinitialiseTnodes(yaf + dev->nFreeTnodes = 0; + } + +-static void yaffs_InitialiseTnodes(yaffs_Device * dev) ++static void yaffs_InitialiseTnodes(yaffs_Device *dev) + { + dev->allocatedTnodeList = NULL; + dev->freeTnodes = NULL; + dev->nFreeTnodes = 0; + dev->nTnodesCreated = 0; +- + } + + +-void yaffs_PutLevel0Tnode(yaffs_Device *dev, yaffs_Tnode *tn, unsigned pos, unsigned val) ++void yaffs_PutLevel0Tnode(yaffs_Device *dev, yaffs_Tnode *tn, unsigned pos, ++ unsigned val) + { +- __u32 *map = (__u32 *)tn; +- __u32 bitInMap; +- __u32 bitInWord; +- __u32 wordInMap; +- __u32 mask; ++ __u32 *map = (__u32 *)tn; ++ __u32 bitInMap; ++ __u32 bitInWord; ++ __u32 wordInMap; ++ __u32 mask; + +- pos &= YAFFS_TNODES_LEVEL0_MASK; +- val >>= dev->chunkGroupBits; ++ pos &= YAFFS_TNODES_LEVEL0_MASK; ++ val >>= dev->chunkGroupBits; + +- bitInMap = pos * dev->tnodeWidth; +- wordInMap = bitInMap /32; +- bitInWord = bitInMap & (32 -1); ++ bitInMap = pos * dev->tnodeWidth; ++ wordInMap = bitInMap / 32; ++ bitInWord = bitInMap & (32 - 1); + +- mask = dev->tnodeMask << bitInWord; ++ mask = dev->tnodeMask << bitInWord; + +- map[wordInMap] &= ~mask; +- map[wordInMap] |= (mask & (val << bitInWord)); ++ map[wordInMap] &= ~mask; ++ map[wordInMap] |= (mask & (val << bitInWord)); + +- if(dev->tnodeWidth > (32-bitInWord)) { +- bitInWord = (32 - bitInWord); +- wordInMap++;; +- mask = dev->tnodeMask >> (/*dev->tnodeWidth -*/ bitInWord); +- map[wordInMap] &= ~mask; +- map[wordInMap] |= (mask & (val >> bitInWord)); +- } ++ if (dev->tnodeWidth > (32 - bitInWord)) { ++ bitInWord = (32 - bitInWord); ++ wordInMap++;; ++ mask = dev->tnodeMask >> (/*dev->tnodeWidth -*/ bitInWord); ++ map[wordInMap] &= ~mask; ++ map[wordInMap] |= (mask & (val >> bitInWord)); ++ } + } + +-static __u32 yaffs_GetChunkGroupBase(yaffs_Device *dev, yaffs_Tnode *tn, unsigned pos) ++static __u32 yaffs_GetChunkGroupBase(yaffs_Device *dev, yaffs_Tnode *tn, ++ unsigned pos) + { +- __u32 *map = (__u32 *)tn; +- __u32 bitInMap; +- __u32 bitInWord; +- __u32 wordInMap; +- __u32 val; ++ __u32 *map = (__u32 *)tn; ++ __u32 bitInMap; ++ __u32 bitInWord; ++ __u32 wordInMap; ++ __u32 val; + +- pos &= YAFFS_TNODES_LEVEL0_MASK; ++ pos &= YAFFS_TNODES_LEVEL0_MASK; + +- bitInMap = pos * dev->tnodeWidth; +- wordInMap = bitInMap /32; +- bitInWord = bitInMap & (32 -1); ++ bitInMap = pos * dev->tnodeWidth; ++ wordInMap = bitInMap / 32; ++ bitInWord = bitInMap & (32 - 1); + +- val = map[wordInMap] >> bitInWord; ++ val = map[wordInMap] >> bitInWord; + +- if(dev->tnodeWidth > (32-bitInWord)) { +- bitInWord = (32 - bitInWord); +- wordInMap++;; +- val |= (map[wordInMap] << bitInWord); +- } ++ if (dev->tnodeWidth > (32 - bitInWord)) { ++ bitInWord = (32 - bitInWord); ++ wordInMap++;; ++ val |= (map[wordInMap] << bitInWord); ++ } + +- val &= dev->tnodeMask; +- val <<= dev->chunkGroupBits; ++ val &= dev->tnodeMask; ++ val <<= dev->chunkGroupBits; + +- return val; ++ return val; + } + + /* ------------------- End of individual tnode manipulation -----------------*/ +@@ -1357,24 +1390,21 @@ static __u32 yaffs_GetChunkGroupBase(yaf + */ + + /* FindLevel0Tnode finds the level 0 tnode, if one exists. */ +-static yaffs_Tnode *yaffs_FindLevel0Tnode(yaffs_Device * dev, +- yaffs_FileStructure * fStruct, +- __u32 chunkId) ++static yaffs_Tnode *yaffs_FindLevel0Tnode(yaffs_Device *dev, ++ yaffs_FileStructure *fStruct, ++ __u32 chunkId) + { +- + yaffs_Tnode *tn = fStruct->top; + __u32 i; + int requiredTallness; + int level = fStruct->topLevel; + + /* Check sane level and chunk Id */ +- if (level < 0 || level > YAFFS_TNODES_MAX_LEVEL) { ++ if (level < 0 || level > YAFFS_TNODES_MAX_LEVEL) + return NULL; +- } + +- if (chunkId > YAFFS_MAX_CHUNK_ID) { ++ if (chunkId > YAFFS_MAX_CHUNK_ID) + return NULL; +- } + + /* First check we're tall enough (ie enough topLevel) */ + +@@ -1385,22 +1415,17 @@ static yaffs_Tnode *yaffs_FindLevel0Tnod + requiredTallness++; + } + +- if (requiredTallness > fStruct->topLevel) { +- /* Not tall enough, so we can't find it, return NULL. */ +- return NULL; +- } ++ if (requiredTallness > fStruct->topLevel) ++ return NULL; /* Not tall enough, so we can't find it */ + + /* Traverse down to level 0 */ + while (level > 0 && tn) { +- tn = tn-> +- internal[(chunkId >> +- ( YAFFS_TNODES_LEVEL0_BITS + +- (level - 1) * +- YAFFS_TNODES_INTERNAL_BITS) +- ) & +- YAFFS_TNODES_INTERNAL_MASK]; ++ tn = tn->internal[(chunkId >> ++ (YAFFS_TNODES_LEVEL0_BITS + ++ (level - 1) * ++ YAFFS_TNODES_INTERNAL_BITS)) & ++ YAFFS_TNODES_INTERNAL_MASK]; + level--; +- + } + + return tn; +@@ -1417,12 +1442,11 @@ static yaffs_Tnode *yaffs_FindLevel0Tnod + * be plugged into the ttree. + */ + +-static yaffs_Tnode *yaffs_AddOrFindLevel0Tnode(yaffs_Device * dev, +- yaffs_FileStructure * fStruct, +- __u32 chunkId, +- yaffs_Tnode *passedTn) ++static yaffs_Tnode *yaffs_AddOrFindLevel0Tnode(yaffs_Device *dev, ++ yaffs_FileStructure *fStruct, ++ __u32 chunkId, ++ yaffs_Tnode *passedTn) + { +- + int requiredTallness; + int i; + int l; +@@ -1432,13 +1456,11 @@ static yaffs_Tnode *yaffs_AddOrFindLevel + + + /* Check sane level and page Id */ +- if (fStruct->topLevel < 0 || fStruct->topLevel > YAFFS_TNODES_MAX_LEVEL) { ++ if (fStruct->topLevel < 0 || fStruct->topLevel > YAFFS_TNODES_MAX_LEVEL) + return NULL; +- } + +- if (chunkId > YAFFS_MAX_CHUNK_ID) { ++ if (chunkId > YAFFS_MAX_CHUNK_ID) + return NULL; +- } + + /* First check we're tall enough (ie enough topLevel) */ + +@@ -1451,7 +1473,7 @@ static yaffs_Tnode *yaffs_AddOrFindLevel + + + if (requiredTallness > fStruct->topLevel) { +- /* Not tall enough,gotta make the tree taller */ ++ /* Not tall enough, gotta make the tree taller */ + for (i = fStruct->topLevel; i < requiredTallness; i++) { + + tn = yaffs_GetTnode(dev); +@@ -1473,27 +1495,27 @@ static yaffs_Tnode *yaffs_AddOrFindLevel + l = fStruct->topLevel; + tn = fStruct->top; + +- if(l > 0) { ++ if (l > 0) { + while (l > 0 && tn) { + x = (chunkId >> +- ( YAFFS_TNODES_LEVEL0_BITS + ++ (YAFFS_TNODES_LEVEL0_BITS + + (l - 1) * YAFFS_TNODES_INTERNAL_BITS)) & + YAFFS_TNODES_INTERNAL_MASK; + + +- if((l>1) && !tn->internal[x]){ ++ if ((l > 1) && !tn->internal[x]) { + /* Add missing non-level-zero tnode */ + tn->internal[x] = yaffs_GetTnode(dev); + +- } else if(l == 1) { ++ } else if (l == 1) { + /* Looking from level 1 at level 0 */ +- if (passedTn) { ++ if (passedTn) { + /* If we already have one, then release it.*/ +- if(tn->internal[x]) +- yaffs_FreeTnode(dev,tn->internal[x]); ++ if (tn->internal[x]) ++ yaffs_FreeTnode(dev, tn->internal[x]); + tn->internal[x] = passedTn; + +- } else if(!tn->internal[x]) { ++ } else if (!tn->internal[x]) { + /* Don't have one, none passed in */ + tn->internal[x] = yaffs_GetTnode(dev); + } +@@ -1504,31 +1526,29 @@ static yaffs_Tnode *yaffs_AddOrFindLevel + } + } else { + /* top is level 0 */ +- if(passedTn) { +- memcpy(tn,passedTn,(dev->tnodeWidth * YAFFS_NTNODES_LEVEL0)/8); +- yaffs_FreeTnode(dev,passedTn); ++ if (passedTn) { ++ memcpy(tn, passedTn, (dev->tnodeWidth * YAFFS_NTNODES_LEVEL0)/8); ++ yaffs_FreeTnode(dev, passedTn); + } + } + + return tn; + } + +-static int yaffs_FindChunkInGroup(yaffs_Device * dev, int theChunk, +- yaffs_ExtendedTags * tags, int objectId, +- int chunkInInode) ++static int yaffs_FindChunkInGroup(yaffs_Device *dev, int theChunk, ++ yaffs_ExtendedTags *tags, int objectId, ++ int chunkInInode) + { + int j; + + for (j = 0; theChunk && j < dev->chunkGroupSize; j++) { +- if (yaffs_CheckChunkBit +- (dev, theChunk / dev->nChunksPerBlock, +- theChunk % dev->nChunksPerBlock)) { ++ if (yaffs_CheckChunkBit(dev, theChunk / dev->nChunksPerBlock, ++ theChunk % dev->nChunksPerBlock)) { + yaffs_ReadChunkWithTagsFromNAND(dev, theChunk, NULL, + tags); + if (yaffs_TagsMatch(tags, objectId, chunkInInode)) { + /* found it; */ + return theChunk; +- + } + } + theChunk++; +@@ -1543,7 +1563,7 @@ static int yaffs_FindChunkInGroup(yaffs_ + * Returns 0 if it stopped early due to hitting the limit and the delete is incomplete. + */ + +-static int yaffs_DeleteWorker(yaffs_Object * in, yaffs_Tnode * tn, __u32 level, ++static int yaffs_DeleteWorker(yaffs_Object *in, yaffs_Tnode *tn, __u32 level, + int chunkOffset, int *limit) + { + int i; +@@ -1557,7 +1577,6 @@ static int yaffs_DeleteWorker(yaffs_Obje + + if (tn) { + if (level > 0) { +- + for (i = YAFFS_NTNODES_INTERNAL - 1; allDone && i >= 0; + i--) { + if (tn->internal[i]) { +@@ -1565,17 +1584,17 @@ static int yaffs_DeleteWorker(yaffs_Obje + allDone = 0; + } else { + allDone = +- yaffs_DeleteWorker(in, +- tn-> +- internal +- [i], +- level - +- 1, +- (chunkOffset ++ yaffs_DeleteWorker(in, ++ tn-> ++ internal ++ [i], ++ level - ++ 1, ++ (chunkOffset + << + YAFFS_TNODES_INTERNAL_BITS) +- + i, +- limit); ++ + i, ++ limit); + } + if (allDone) { + yaffs_FreeTnode(dev, +@@ -1584,27 +1603,25 @@ static int yaffs_DeleteWorker(yaffs_Obje + tn->internal[i] = NULL; + } + } +- + } + return (allDone) ? 1 : 0; + } else if (level == 0) { + int hitLimit = 0; + + for (i = YAFFS_NTNODES_LEVEL0 - 1; i >= 0 && !hitLimit; +- i--) { +- theChunk = yaffs_GetChunkGroupBase(dev,tn,i); ++ i--) { ++ theChunk = yaffs_GetChunkGroupBase(dev, tn, i); + if (theChunk) { + +- chunkInInode = +- (chunkOffset << +- YAFFS_TNODES_LEVEL0_BITS) + i; ++ chunkInInode = (chunkOffset << ++ YAFFS_TNODES_LEVEL0_BITS) + i; + + foundChunk = +- yaffs_FindChunkInGroup(dev, +- theChunk, +- &tags, +- in->objectId, +- chunkInInode); ++ yaffs_FindChunkInGroup(dev, ++ theChunk, ++ &tags, ++ in->objectId, ++ chunkInInode); + + if (foundChunk > 0) { + yaffs_DeleteChunk(dev, +@@ -1613,14 +1630,13 @@ static int yaffs_DeleteWorker(yaffs_Obje + in->nDataChunks--; + if (limit) { + *limit = *limit - 1; +- if (*limit <= 0) { ++ if (*limit <= 0) + hitLimit = 1; +- } + } + + } + +- yaffs_PutLevel0Tnode(dev,tn,i,0); ++ yaffs_PutLevel0Tnode(dev, tn, i, 0); + } + + } +@@ -1634,9 +1650,8 @@ static int yaffs_DeleteWorker(yaffs_Obje + + } + +-static void yaffs_SoftDeleteChunk(yaffs_Device * dev, int chunk) ++static void yaffs_SoftDeleteChunk(yaffs_Device *dev, int chunk) + { +- + yaffs_BlockInfo *theBlock; + + T(YAFFS_TRACE_DELETION, (TSTR("soft delete chunk %d" TENDSTR), chunk)); +@@ -1654,7 +1669,7 @@ static void yaffs_SoftDeleteChunk(yaffs_ + * Thus, essentially this is the same as DeleteWorker except that the chunks are soft deleted. + */ + +-static int yaffs_SoftDeleteWorker(yaffs_Object * in, yaffs_Tnode * tn, ++static int yaffs_SoftDeleteWorker(yaffs_Object *in, yaffs_Tnode *tn, + __u32 level, int chunkOffset) + { + int i; +@@ -1691,14 +1706,14 @@ static int yaffs_SoftDeleteWorker(yaffs_ + } else if (level == 0) { + + for (i = YAFFS_NTNODES_LEVEL0 - 1; i >= 0; i--) { +- theChunk = yaffs_GetChunkGroupBase(dev,tn,i); ++ theChunk = yaffs_GetChunkGroupBase(dev, tn, i); + if (theChunk) { + /* Note this does not find the real chunk, only the chunk group. + * We make an assumption that a chunk group is not larger than + * a block. + */ + yaffs_SoftDeleteChunk(dev, theChunk); +- yaffs_PutLevel0Tnode(dev,tn,i,0); ++ yaffs_PutLevel0Tnode(dev, tn, i, 0); + } + + } +@@ -1712,7 +1727,7 @@ static int yaffs_SoftDeleteWorker(yaffs_ + + } + +-static void yaffs_SoftDeleteFile(yaffs_Object * obj) ++static void yaffs_SoftDeleteFile(yaffs_Object *obj) + { + if (obj->deleted && + obj->variantType == YAFFS_OBJECT_TYPE_FILE && !obj->softDeleted) { +@@ -1746,8 +1761,8 @@ static void yaffs_SoftDeleteFile(yaffs_O + * by a special case. + */ + +-static yaffs_Tnode *yaffs_PruneWorker(yaffs_Device * dev, yaffs_Tnode * tn, +- __u32 level, int del0) ++static yaffs_Tnode *yaffs_PruneWorker(yaffs_Device *dev, yaffs_Tnode *tn, ++ __u32 level, int del0) + { + int i; + int hasData; +@@ -1763,9 +1778,8 @@ static yaffs_Tnode *yaffs_PruneWorker(ya + (i == 0) ? del0 : 1); + } + +- if (tn->internal[i]) { ++ if (tn->internal[i]) + hasData++; +- } + } + + if (hasData == 0 && del0) { +@@ -1781,8 +1795,8 @@ static yaffs_Tnode *yaffs_PruneWorker(ya + + } + +-static int yaffs_PruneFileStructure(yaffs_Device * dev, +- yaffs_FileStructure * fStruct) ++static int yaffs_PruneFileStructure(yaffs_Device *dev, ++ yaffs_FileStructure *fStruct) + { + int i; + int hasData; +@@ -1805,9 +1819,8 @@ static int yaffs_PruneFileStructure(yaff + + hasData = 0; + for (i = 1; i < YAFFS_NTNODES_INTERNAL; i++) { +- if (tn->internal[i]) { ++ if (tn->internal[i]) + hasData++; +- } + } + + if (!hasData) { +@@ -1828,7 +1841,7 @@ static int yaffs_PruneFileStructure(yaff + /* yaffs_CreateFreeObjects creates a bunch more objects and + * adds them to the object free list. + */ +-static int yaffs_CreateFreeObjects(yaffs_Device * dev, int nObjects) ++static int yaffs_CreateFreeObjects(yaffs_Device *dev, int nObjects) + { + int i; + yaffs_Object *newObjects; +@@ -1842,9 +1855,9 @@ static int yaffs_CreateFreeObjects(yaffs + list = YMALLOC(sizeof(yaffs_ObjectList)); + + if (!newObjects || !list) { +- if(newObjects) ++ if (newObjects) + YFREE(newObjects); +- if(list) ++ if (list) + YFREE(list); + T(YAFFS_TRACE_ALLOCATE, + (TSTR("yaffs: Could not allocate more objects" TENDSTR))); +@@ -1854,7 +1867,7 @@ static int yaffs_CreateFreeObjects(yaffs + /* Hook them into the free list */ + for (i = 0; i < nObjects - 1; i++) { + newObjects[i].siblings.next = +- (struct list_head *)(&newObjects[i + 1]); ++ (struct ylist_head *)(&newObjects[i + 1]); + } + + newObjects[nObjects - 1].siblings.next = (void *)dev->freeObjects; +@@ -1873,85 +1886,109 @@ static int yaffs_CreateFreeObjects(yaffs + + + /* AllocateEmptyObject gets us a clean Object. Tries to make allocate more if we run out */ +-static yaffs_Object *yaffs_AllocateEmptyObject(yaffs_Device * dev) ++static yaffs_Object *yaffs_AllocateEmptyObject(yaffs_Device *dev) + { + yaffs_Object *tn = NULL; + ++#ifdef VALGRIND_TEST ++ tn = YMALLOC(sizeof(yaffs_Object)); ++#else + /* If there are none left make more */ +- if (!dev->freeObjects) { ++ if (!dev->freeObjects) + yaffs_CreateFreeObjects(dev, YAFFS_ALLOCATION_NOBJECTS); +- } + + if (dev->freeObjects) { + tn = dev->freeObjects; + dev->freeObjects = +- (yaffs_Object *) (dev->freeObjects->siblings.next); ++ (yaffs_Object *) (dev->freeObjects->siblings.next); + dev->nFreeObjects--; +- ++ } ++#endif ++ if (tn) { + /* Now sweeten it up... */ + + memset(tn, 0, sizeof(yaffs_Object)); ++ tn->beingCreated = 1; ++ + tn->myDev = dev; +- tn->chunkId = -1; ++ tn->hdrChunk = 0; + tn->variantType = YAFFS_OBJECT_TYPE_UNKNOWN; +- INIT_LIST_HEAD(&(tn->hardLinks)); +- INIT_LIST_HEAD(&(tn->hashLink)); +- INIT_LIST_HEAD(&tn->siblings); ++ YINIT_LIST_HEAD(&(tn->hardLinks)); ++ YINIT_LIST_HEAD(&(tn->hashLink)); ++ YINIT_LIST_HEAD(&tn->siblings); ++ ++ ++ /* Now make the directory sane */ ++ if (dev->rootDir) { ++ tn->parent = dev->rootDir; ++ ylist_add(&(tn->siblings), &dev->rootDir->variant.directoryVariant.children); ++ } + + /* Add it to the lost and found directory. + * NB Can't put root or lostNFound in lostNFound so + * check if lostNFound exists first + */ +- if (dev->lostNFoundDir) { ++ if (dev->lostNFoundDir) + yaffs_AddObjectToDirectory(dev->lostNFoundDir, tn); +- } ++ ++ tn->beingCreated = 0; + } + ++ dev->nCheckpointBlocksRequired = 0; /* force recalculation*/ ++ + return tn; + } + +-static yaffs_Object *yaffs_CreateFakeDirectory(yaffs_Device * dev, int number, ++static yaffs_Object *yaffs_CreateFakeDirectory(yaffs_Device *dev, int number, + __u32 mode) + { + + yaffs_Object *obj = + yaffs_CreateNewObject(dev, number, YAFFS_OBJECT_TYPE_DIRECTORY); + if (obj) { +- obj->fake = 1; /* it is fake so it has no NAND presence... */ ++ obj->fake = 1; /* it is fake so it might have no NAND presence... */ + obj->renameAllowed = 0; /* ... and we're not allowed to rename it... */ + obj->unlinkAllowed = 0; /* ... or unlink it */ + obj->deleted = 0; + obj->unlinked = 0; + obj->yst_mode = mode; + obj->myDev = dev; +- obj->chunkId = 0; /* Not a valid chunk. */ ++ obj->hdrChunk = 0; /* Not a valid chunk. */ + } + + return obj; + + } + +-static void yaffs_UnhashObject(yaffs_Object * tn) ++static void yaffs_UnhashObject(yaffs_Object *tn) + { + int bucket; + yaffs_Device *dev = tn->myDev; + + /* If it is still linked into the bucket list, free from the list */ +- if (!list_empty(&tn->hashLink)) { +- list_del_init(&tn->hashLink); ++ if (!ylist_empty(&tn->hashLink)) { ++ ylist_del_init(&tn->hashLink); + bucket = yaffs_HashFunction(tn->objectId); + dev->objectBucket[bucket].count--; + } +- + } + + /* FreeObject frees up a Object and puts it back on the free list */ +-static void yaffs_FreeObject(yaffs_Object * tn) ++static void yaffs_FreeObject(yaffs_Object *tn) + { +- + yaffs_Device *dev = tn->myDev; + +-#ifdef __KERNEL__ ++#ifdef __KERNEL__ ++ T(YAFFS_TRACE_OS, (TSTR("FreeObject %p inode %p"TENDSTR), tn, tn->myInode)); ++#endif ++ ++ if (tn->parent) ++ YBUG(); ++ if (!ylist_empty(&tn->siblings)) ++ YBUG(); ++ ++ ++#ifdef __KERNEL__ + if (tn->myInode) { + /* We're still hooked up to a cached inode. + * Don't delete now, but mark for later deletion +@@ -1963,24 +2000,28 @@ static void yaffs_FreeObject(yaffs_Objec + + yaffs_UnhashObject(tn); + ++#ifdef VALGRIND_TEST ++ YFREE(tn); ++#else + /* Link into the free list. */ +- tn->siblings.next = (struct list_head *)(dev->freeObjects); ++ tn->siblings.next = (struct ylist_head *)(dev->freeObjects); + dev->freeObjects = tn; + dev->nFreeObjects++; ++#endif ++ dev->nCheckpointBlocksRequired = 0; /* force recalculation*/ + } + + #ifdef __KERNEL__ + +-void yaffs_HandleDeferedFree(yaffs_Object * obj) ++void yaffs_HandleDeferedFree(yaffs_Object *obj) + { +- if (obj->deferedFree) { ++ if (obj->deferedFree) + yaffs_FreeObject(obj); +- } + } + + #endif + +-static void yaffs_DeinitialiseObjects(yaffs_Device * dev) ++static void yaffs_DeinitialiseObjects(yaffs_Device *dev) + { + /* Free the list of allocated Objects */ + +@@ -1998,7 +2039,7 @@ static void yaffs_DeinitialiseObjects(ya + dev->nFreeObjects = 0; + } + +-static void yaffs_InitialiseObjects(yaffs_Device * dev) ++static void yaffs_InitialiseObjects(yaffs_Device *dev) + { + int i; + +@@ -2007,15 +2048,14 @@ static void yaffs_InitialiseObjects(yaff + dev->nFreeObjects = 0; + + for (i = 0; i < YAFFS_NOBJECT_BUCKETS; i++) { +- INIT_LIST_HEAD(&dev->objectBucket[i].list); ++ YINIT_LIST_HEAD(&dev->objectBucket[i].list); + dev->objectBucket[i].count = 0; + } +- + } + +-static int yaffs_FindNiceObjectBucket(yaffs_Device * dev) ++static int yaffs_FindNiceObjectBucket(yaffs_Device *dev) + { +- static int x = 0; ++ static int x; + int i; + int l = 999; + int lowest = 999999; +@@ -2049,7 +2089,7 @@ static int yaffs_FindNiceObjectBucket(ya + return l; + } + +-static int yaffs_CreateNewObjectNumber(yaffs_Device * dev) ++static int yaffs_CreateNewObjectNumber(yaffs_Device *dev) + { + int bucket = yaffs_FindNiceObjectBucket(dev); + +@@ -2058,7 +2098,7 @@ static int yaffs_CreateNewObjectNumber(y + */ + + int found = 0; +- struct list_head *i; ++ struct ylist_head *i; + + __u32 n = (__u32) bucket; + +@@ -2068,41 +2108,38 @@ static int yaffs_CreateNewObjectNumber(y + found = 1; + n += YAFFS_NOBJECT_BUCKETS; + if (1 || dev->objectBucket[bucket].count > 0) { +- list_for_each(i, &dev->objectBucket[bucket].list) { ++ ylist_for_each(i, &dev->objectBucket[bucket].list) { + /* If there is already one in the list */ +- if (i +- && list_entry(i, yaffs_Object, +- hashLink)->objectId == n) { ++ if (i && ylist_entry(i, yaffs_Object, ++ hashLink)->objectId == n) { + found = 0; + } + } + } + } + +- + return n; + } + +-static void yaffs_HashObject(yaffs_Object * in) ++static void yaffs_HashObject(yaffs_Object *in) + { + int bucket = yaffs_HashFunction(in->objectId); + yaffs_Device *dev = in->myDev; + +- list_add(&in->hashLink, &dev->objectBucket[bucket].list); ++ ylist_add(&in->hashLink, &dev->objectBucket[bucket].list); + dev->objectBucket[bucket].count++; +- + } + +-yaffs_Object *yaffs_FindObjectByNumber(yaffs_Device * dev, __u32 number) ++yaffs_Object *yaffs_FindObjectByNumber(yaffs_Device *dev, __u32 number) + { + int bucket = yaffs_HashFunction(number); +- struct list_head *i; ++ struct ylist_head *i; + yaffs_Object *in; + +- list_for_each(i, &dev->objectBucket[bucket].list) { ++ ylist_for_each(i, &dev->objectBucket[bucket].list) { + /* Look if it is in the list */ + if (i) { +- in = list_entry(i, yaffs_Object, hashLink); ++ in = ylist_entry(i, yaffs_Object, hashLink); + if (in->objectId == number) { + #ifdef __KERNEL__ + /* Don't tell the VFS about this one if it is defered free */ +@@ -2118,31 +2155,27 @@ yaffs_Object *yaffs_FindObjectByNumber(y + return NULL; + } + +-yaffs_Object *yaffs_CreateNewObject(yaffs_Device * dev, int number, ++yaffs_Object *yaffs_CreateNewObject(yaffs_Device *dev, int number, + yaffs_ObjectType type) + { +- + yaffs_Object *theObject; +- yaffs_Tnode *tn; ++ yaffs_Tnode *tn = NULL; + +- if (number < 0) { ++ if (number < 0) + number = yaffs_CreateNewObjectNumber(dev); +- } + + theObject = yaffs_AllocateEmptyObject(dev); +- if(!theObject) ++ if (!theObject) + return NULL; + +- if(type == YAFFS_OBJECT_TYPE_FILE){ ++ if (type == YAFFS_OBJECT_TYPE_FILE) { + tn = yaffs_GetTnode(dev); +- if(!tn){ ++ if (!tn) { + yaffs_FreeObject(theObject); + return NULL; + } + } + +- +- + if (theObject) { + theObject->fake = 0; + theObject->renameAllowed = 1; +@@ -2171,8 +2204,8 @@ yaffs_Object *yaffs_CreateNewObject(yaff + theObject->variant.fileVariant.top = tn; + break; + case YAFFS_OBJECT_TYPE_DIRECTORY: +- INIT_LIST_HEAD(&theObject->variant.directoryVariant. +- children); ++ YINIT_LIST_HEAD(&theObject->variant.directoryVariant. ++ children); + break; + case YAFFS_OBJECT_TYPE_SYMLINK: + case YAFFS_OBJECT_TYPE_HARDLINK: +@@ -2188,32 +2221,30 @@ yaffs_Object *yaffs_CreateNewObject(yaff + return theObject; + } + +-static yaffs_Object *yaffs_FindOrCreateObjectByNumber(yaffs_Device * dev, ++static yaffs_Object *yaffs_FindOrCreateObjectByNumber(yaffs_Device *dev, + int number, + yaffs_ObjectType type) + { + yaffs_Object *theObject = NULL; + +- if (number > 0) { ++ if (number > 0) + theObject = yaffs_FindObjectByNumber(dev, number); +- } + +- if (!theObject) { ++ if (!theObject) + theObject = yaffs_CreateNewObject(dev, number, type); +- } + + return theObject; + + } + + +-static YCHAR *yaffs_CloneString(const YCHAR * str) ++static YCHAR *yaffs_CloneString(const YCHAR *str) + { + YCHAR *newStr = NULL; + + if (str && *str) { + newStr = YMALLOC((yaffs_strlen(str) + 1) * sizeof(YCHAR)); +- if(newStr) ++ if (newStr) + yaffs_strcpy(newStr, str); + } + +@@ -2229,29 +2260,31 @@ static YCHAR *yaffs_CloneString(const YC + */ + + static yaffs_Object *yaffs_MknodObject(yaffs_ObjectType type, +- yaffs_Object * parent, +- const YCHAR * name, ++ yaffs_Object *parent, ++ const YCHAR *name, + __u32 mode, + __u32 uid, + __u32 gid, +- yaffs_Object * equivalentObject, +- const YCHAR * aliasString, __u32 rdev) ++ yaffs_Object *equivalentObject, ++ const YCHAR *aliasString, __u32 rdev) + { + yaffs_Object *in; +- YCHAR *str; ++ YCHAR *str = NULL; + + yaffs_Device *dev = parent->myDev; + + /* Check if the entry exists. If it does then fail the call since we don't want a dup.*/ +- if (yaffs_FindObjectByName(parent, name)) { ++ if (yaffs_FindObjectByName(parent, name)) + return NULL; +- } + + in = yaffs_CreateNewObject(dev, -1, type); + +- if(type == YAFFS_OBJECT_TYPE_SYMLINK){ ++ if (!in) ++ return YAFFS_FAIL; ++ ++ if (type == YAFFS_OBJECT_TYPE_SYMLINK) { + str = yaffs_CloneString(aliasString); +- if(!str){ ++ if (!str) { + yaffs_FreeObject(in); + return NULL; + } +@@ -2260,7 +2293,7 @@ static yaffs_Object *yaffs_MknodObject(y + + + if (in) { +- in->chunkId = -1; ++ in->hdrChunk = 0; + in->valid = 1; + in->variantType = type; + +@@ -2293,10 +2326,10 @@ static yaffs_Object *yaffs_MknodObject(y + break; + case YAFFS_OBJECT_TYPE_HARDLINK: + in->variant.hardLinkVariant.equivalentObject = +- equivalentObject; ++ equivalentObject; + in->variant.hardLinkVariant.equivalentObjectId = +- equivalentObject->objectId; +- list_add(&in->hardLinks, &equivalentObject->hardLinks); ++ equivalentObject->objectId; ++ ylist_add(&in->hardLinks, &equivalentObject->hardLinks); + break; + case YAFFS_OBJECT_TYPE_FILE: + case YAFFS_OBJECT_TYPE_DIRECTORY: +@@ -2308,7 +2341,7 @@ static yaffs_Object *yaffs_MknodObject(y + + if (yaffs_UpdateObjectHeader(in, name, 0, 0, 0) < 0) { + /* Could not create the object header, fail the creation */ +- yaffs_DestroyObject(in); ++ yaffs_DeleteObject(in); + in = NULL; + } + +@@ -2317,38 +2350,38 @@ static yaffs_Object *yaffs_MknodObject(y + return in; + } + +-yaffs_Object *yaffs_MknodFile(yaffs_Object * parent, const YCHAR * name, +- __u32 mode, __u32 uid, __u32 gid) ++yaffs_Object *yaffs_MknodFile(yaffs_Object *parent, const YCHAR *name, ++ __u32 mode, __u32 uid, __u32 gid) + { + return yaffs_MknodObject(YAFFS_OBJECT_TYPE_FILE, parent, name, mode, +- uid, gid, NULL, NULL, 0); ++ uid, gid, NULL, NULL, 0); + } + +-yaffs_Object *yaffs_MknodDirectory(yaffs_Object * parent, const YCHAR * name, +- __u32 mode, __u32 uid, __u32 gid) ++yaffs_Object *yaffs_MknodDirectory(yaffs_Object *parent, const YCHAR *name, ++ __u32 mode, __u32 uid, __u32 gid) + { + return yaffs_MknodObject(YAFFS_OBJECT_TYPE_DIRECTORY, parent, name, + mode, uid, gid, NULL, NULL, 0); + } + +-yaffs_Object *yaffs_MknodSpecial(yaffs_Object * parent, const YCHAR * name, +- __u32 mode, __u32 uid, __u32 gid, __u32 rdev) ++yaffs_Object *yaffs_MknodSpecial(yaffs_Object *parent, const YCHAR *name, ++ __u32 mode, __u32 uid, __u32 gid, __u32 rdev) + { + return yaffs_MknodObject(YAFFS_OBJECT_TYPE_SPECIAL, parent, name, mode, + uid, gid, NULL, NULL, rdev); + } + +-yaffs_Object *yaffs_MknodSymLink(yaffs_Object * parent, const YCHAR * name, +- __u32 mode, __u32 uid, __u32 gid, +- const YCHAR * alias) ++yaffs_Object *yaffs_MknodSymLink(yaffs_Object *parent, const YCHAR *name, ++ __u32 mode, __u32 uid, __u32 gid, ++ const YCHAR *alias) + { + return yaffs_MknodObject(YAFFS_OBJECT_TYPE_SYMLINK, parent, name, mode, +- uid, gid, NULL, alias, 0); ++ uid, gid, NULL, alias, 0); + } + + /* yaffs_Link returns the object id of the equivalent object.*/ +-yaffs_Object *yaffs_Link(yaffs_Object * parent, const YCHAR * name, +- yaffs_Object * equivalentObject) ++yaffs_Object *yaffs_Link(yaffs_Object *parent, const YCHAR *name, ++ yaffs_Object *equivalentObject) + { + /* Get the real object in case we were fed a hard link as an equivalent object */ + equivalentObject = yaffs_GetEquivalentObject(equivalentObject); +@@ -2363,33 +2396,31 @@ yaffs_Object *yaffs_Link(yaffs_Object * + + } + +-static int yaffs_ChangeObjectName(yaffs_Object * obj, yaffs_Object * newDir, +- const YCHAR * newName, int force, int shadows) ++static int yaffs_ChangeObjectName(yaffs_Object *obj, yaffs_Object *newDir, ++ const YCHAR *newName, int force, int shadows) + { + int unlinkOp; + int deleteOp; + + yaffs_Object *existingTarget; + +- if (newDir == NULL) { ++ if (newDir == NULL) + newDir = obj->parent; /* use the old directory */ +- } + + if (newDir->variantType != YAFFS_OBJECT_TYPE_DIRECTORY) { + T(YAFFS_TRACE_ALWAYS, + (TSTR +- ("tragendy: yaffs_ChangeObjectName: newDir is not a directory" ++ ("tragedy: yaffs_ChangeObjectName: newDir is not a directory" + TENDSTR))); + YBUG(); + } + + /* TODO: Do we need this different handling for YAFFS2 and YAFFS1?? */ +- if (obj->myDev->isYaffs2) { ++ if (obj->myDev->isYaffs2) + unlinkOp = (newDir == obj->myDev->unlinkedDir); +- } else { ++ else + unlinkOp = (newDir == obj->myDev->unlinkedDir + && obj->variantType == YAFFS_OBJECT_TYPE_FILE); +- } + + deleteOp = (newDir == obj->myDev->deletedDir); + +@@ -2415,40 +2446,40 @@ static int yaffs_ChangeObjectName(yaffs_ + obj->unlinked = 1; + + /* If it is a deletion then we mark it as a shrink for gc purposes. */ +- if (yaffs_UpdateObjectHeader(obj, newName, 0, deleteOp, shadows)>= 0) ++ if (yaffs_UpdateObjectHeader(obj, newName, 0, deleteOp, shadows) >= 0) + return YAFFS_OK; + } + + return YAFFS_FAIL; + } + +-int yaffs_RenameObject(yaffs_Object * oldDir, const YCHAR * oldName, +- yaffs_Object * newDir, const YCHAR * newName) ++int yaffs_RenameObject(yaffs_Object *oldDir, const YCHAR *oldName, ++ yaffs_Object *newDir, const YCHAR *newName) + { +- yaffs_Object *obj; +- yaffs_Object *existingTarget; ++ yaffs_Object *obj = NULL; ++ yaffs_Object *existingTarget = NULL; + int force = 0; + ++ ++ if (!oldDir || oldDir->variantType != YAFFS_OBJECT_TYPE_DIRECTORY) ++ YBUG(); ++ if (!newDir || newDir->variantType != YAFFS_OBJECT_TYPE_DIRECTORY) ++ YBUG(); ++ + #ifdef CONFIG_YAFFS_CASE_INSENSITIVE + /* Special case for case insemsitive systems (eg. WinCE). + * While look-up is case insensitive, the name isn't. + * Therefore we might want to change x.txt to X.txt + */ +- if (oldDir == newDir && yaffs_strcmp(oldName, newName) == 0) { ++ if (oldDir == newDir && yaffs_strcmp(oldName, newName) == 0) + force = 1; +- } + #endif + ++ else if (yaffs_strlen(newName) > YAFFS_MAX_NAME_LENGTH) ++ /* ENAMETOOLONG */ ++ return YAFFS_FAIL; ++ + obj = yaffs_FindObjectByName(oldDir, oldName); +- /* Check new name to long. */ +- if (obj->variantType == YAFFS_OBJECT_TYPE_SYMLINK && +- yaffs_strlen(newName) > YAFFS_MAX_ALIAS_LENGTH) +- /* ENAMETOOLONG */ +- return YAFFS_FAIL; +- else if (obj->variantType != YAFFS_OBJECT_TYPE_SYMLINK && +- yaffs_strlen(newName) > YAFFS_MAX_NAME_LENGTH) +- /* ENAMETOOLONG */ +- return YAFFS_FAIL; + + if (obj && obj->renameAllowed) { + +@@ -2456,8 +2487,8 @@ int yaffs_RenameObject(yaffs_Object * ol + + existingTarget = yaffs_FindObjectByName(newDir, newName); + if (existingTarget && +- existingTarget->variantType == YAFFS_OBJECT_TYPE_DIRECTORY && +- !list_empty(&existingTarget->variant.directoryVariant.children)) { ++ existingTarget->variantType == YAFFS_OBJECT_TYPE_DIRECTORY && ++ !ylist_empty(&existingTarget->variant.directoryVariant.children)) { + /* There is a target that is a non-empty directory, so we fail */ + return YAFFS_FAIL; /* EEXIST or ENOTEMPTY */ + } else if (existingTarget && existingTarget != obj) { +@@ -2465,7 +2496,7 @@ int yaffs_RenameObject(yaffs_Object * ol + * but only if it isn't the same object + */ + yaffs_ChangeObjectName(obj, newDir, newName, force, +- existingTarget->objectId); ++ existingTarget->objectId); + yaffs_UnlinkObject(existingTarget); + } + +@@ -2476,7 +2507,7 @@ int yaffs_RenameObject(yaffs_Object * ol + + /*------------------------- Block Management and Page Allocation ----------------*/ + +-static int yaffs_InitialiseBlocks(yaffs_Device * dev) ++static int yaffs_InitialiseBlocks(yaffs_Device *dev) + { + int nBlocks = dev->internalEndBlock - dev->internalStartBlock + 1; + +@@ -2487,23 +2518,20 @@ static int yaffs_InitialiseBlocks(yaffs_ + + /* If the first allocation strategy fails, thry the alternate one */ + dev->blockInfo = YMALLOC(nBlocks * sizeof(yaffs_BlockInfo)); +- if(!dev->blockInfo){ ++ if (!dev->blockInfo) { + dev->blockInfo = YMALLOC_ALT(nBlocks * sizeof(yaffs_BlockInfo)); + dev->blockInfoAlt = 1; +- } +- else ++ } else + dev->blockInfoAlt = 0; + +- if(dev->blockInfo){ +- ++ if (dev->blockInfo) { + /* Set up dynamic blockinfo stuff. */ + dev->chunkBitmapStride = (dev->nChunksPerBlock + 7) / 8; /* round up bytes */ + dev->chunkBits = YMALLOC(dev->chunkBitmapStride * nBlocks); +- if(!dev->chunkBits){ ++ if (!dev->chunkBits) { + dev->chunkBits = YMALLOC_ALT(dev->chunkBitmapStride * nBlocks); + dev->chunkBitsAlt = 1; +- } +- else ++ } else + dev->chunkBitsAlt = 0; + } + +@@ -2514,30 +2542,29 @@ static int yaffs_InitialiseBlocks(yaffs_ + } + + return YAFFS_FAIL; +- + } + +-static void yaffs_DeinitialiseBlocks(yaffs_Device * dev) ++static void yaffs_DeinitialiseBlocks(yaffs_Device *dev) + { +- if(dev->blockInfoAlt && dev->blockInfo) ++ if (dev->blockInfoAlt && dev->blockInfo) + YFREE_ALT(dev->blockInfo); +- else if(dev->blockInfo) ++ else if (dev->blockInfo) + YFREE(dev->blockInfo); + + dev->blockInfoAlt = 0; + + dev->blockInfo = NULL; + +- if(dev->chunkBitsAlt && dev->chunkBits) ++ if (dev->chunkBitsAlt && dev->chunkBits) + YFREE_ALT(dev->chunkBits); +- else if(dev->chunkBits) ++ else if (dev->chunkBits) + YFREE(dev->chunkBits); + dev->chunkBitsAlt = 0; + dev->chunkBits = NULL; + } + +-static int yaffs_BlockNotDisqualifiedFromGC(yaffs_Device * dev, +- yaffs_BlockInfo * bi) ++static int yaffs_BlockNotDisqualifiedFromGC(yaffs_Device *dev, ++ yaffs_BlockInfo *bi) + { + int i; + __u32 seq; +@@ -2556,7 +2583,7 @@ static int yaffs_BlockNotDisqualifiedFro + seq = dev->sequenceNumber; + + for (i = dev->internalStartBlock; i <= dev->internalEndBlock; +- i++) { ++ i++) { + b = yaffs_GetBlockInfo(dev, i); + if (b->blockState == YAFFS_BLOCK_STATE_FULL && + (b->pagesInUse - b->softDeletions) < +@@ -2571,38 +2598,36 @@ static int yaffs_BlockNotDisqualifiedFro + * discarded pages. + */ + return (bi->sequenceNumber <= dev->oldestDirtySequence); +- + } + + /* FindDiretiestBlock is used to select the dirtiest block (or close enough) + * for garbage collection. + */ + +-static int yaffs_FindBlockForGarbageCollection(yaffs_Device * dev, +- int aggressive) ++static int yaffs_FindBlockForGarbageCollection(yaffs_Device *dev, ++ int aggressive) + { +- + int b = dev->currentDirtyChecker; + + int i; + int iterations; + int dirtiest = -1; + int pagesInUse = 0; +- int prioritised=0; ++ int prioritised = 0; + yaffs_BlockInfo *bi; + int pendingPrioritisedExist = 0; + + /* First let's see if we need to grab a prioritised block */ +- if(dev->hasPendingPrioritisedGCs){ +- for(i = dev->internalStartBlock; i < dev->internalEndBlock && !prioritised; i++){ ++ if (dev->hasPendingPrioritisedGCs) { ++ for (i = dev->internalStartBlock; i < dev->internalEndBlock && !prioritised; i++) { + + bi = yaffs_GetBlockInfo(dev, i); +- //yaffs_VerifyBlock(dev,bi,i); ++ /* yaffs_VerifyBlock(dev,bi,i); */ + +- if(bi->gcPrioritise) { ++ if (bi->gcPrioritise) { + pendingPrioritisedExist = 1; +- if(bi->blockState == YAFFS_BLOCK_STATE_FULL && +- yaffs_BlockNotDisqualifiedFromGC(dev, bi)){ ++ if (bi->blockState == YAFFS_BLOCK_STATE_FULL && ++ yaffs_BlockNotDisqualifiedFromGC(dev, bi)) { + pagesInUse = (bi->pagesInUse - bi->softDeletions); + dirtiest = i; + prioritised = 1; +@@ -2611,7 +2636,7 @@ static int yaffs_FindBlockForGarbageColl + } + } + +- if(!pendingPrioritisedExist) /* None found, so we can clear this */ ++ if (!pendingPrioritisedExist) /* None found, so we can clear this */ + dev->hasPendingPrioritisedGCs = 0; + } + +@@ -2623,31 +2648,28 @@ static int yaffs_FindBlockForGarbageColl + + dev->nonAggressiveSkip--; + +- if (!aggressive && (dev->nonAggressiveSkip > 0)) { ++ if (!aggressive && (dev->nonAggressiveSkip > 0)) + return -1; +- } + +- if(!prioritised) ++ if (!prioritised) + pagesInUse = +- (aggressive) ? dev->nChunksPerBlock : YAFFS_PASSIVE_GC_CHUNKS + 1; ++ (aggressive) ? dev->nChunksPerBlock : YAFFS_PASSIVE_GC_CHUNKS + 1; + +- if (aggressive) { ++ if (aggressive) + iterations = + dev->internalEndBlock - dev->internalStartBlock + 1; +- } else { ++ else { + iterations = + dev->internalEndBlock - dev->internalStartBlock + 1; + iterations = iterations / 16; +- if (iterations > 200) { ++ if (iterations > 200) + iterations = 200; +- } + } + + for (i = 0; i <= iterations && pagesInUse > 0 && !prioritised; i++) { + b++; +- if (b < dev->internalStartBlock || b > dev->internalEndBlock) { ++ if (b < dev->internalStartBlock || b > dev->internalEndBlock) + b = dev->internalStartBlock; +- } + + if (b < dev->internalStartBlock || b > dev->internalEndBlock) { + T(YAFFS_TRACE_ERROR, +@@ -2657,17 +2679,9 @@ static int yaffs_FindBlockForGarbageColl + + bi = yaffs_GetBlockInfo(dev, b); + +-#if 0 +- if (bi->blockState == YAFFS_BLOCK_STATE_CHECKPOINT) { +- dirtiest = b; +- pagesInUse = 0; +- } +- else +-#endif +- + if (bi->blockState == YAFFS_BLOCK_STATE_FULL && +- (bi->pagesInUse - bi->softDeletions) < pagesInUse && +- yaffs_BlockNotDisqualifiedFromGC(dev, bi)) { ++ (bi->pagesInUse - bi->softDeletions) < pagesInUse && ++ yaffs_BlockNotDisqualifiedFromGC(dev, bi)) { + dirtiest = b; + pagesInUse = (bi->pagesInUse - bi->softDeletions); + } +@@ -2678,19 +2692,18 @@ static int yaffs_FindBlockForGarbageColl + if (dirtiest > 0) { + T(YAFFS_TRACE_GC, + (TSTR("GC Selected block %d with %d free, prioritised:%d" TENDSTR), dirtiest, +- dev->nChunksPerBlock - pagesInUse,prioritised)); ++ dev->nChunksPerBlock - pagesInUse, prioritised)); + } + + dev->oldestDirtySequence = 0; + +- if (dirtiest > 0) { ++ if (dirtiest > 0) + dev->nonAggressiveSkip = 4; +- } + + return dirtiest; + } + +-static void yaffs_BlockBecameDirty(yaffs_Device * dev, int blockNo) ++static void yaffs_BlockBecameDirty(yaffs_Device *dev, int blockNo) + { + yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev, blockNo); + +@@ -2752,7 +2765,7 @@ static void yaffs_BlockBecameDirty(yaffs + } + } + +-static int yaffs_FindBlockForAllocation(yaffs_Device * dev) ++static int yaffs_FindBlockForAllocation(yaffs_Device *dev) + { + int i; + +@@ -2763,7 +2776,7 @@ static int yaffs_FindBlockForAllocation( + * Can't get space to gc + */ + T(YAFFS_TRACE_ERROR, +- (TSTR("yaffs tragedy: no more eraased blocks" TENDSTR))); ++ (TSTR("yaffs tragedy: no more erased blocks" TENDSTR))); + + return -1; + } +@@ -2794,31 +2807,74 @@ static int yaffs_FindBlockForAllocation( + + T(YAFFS_TRACE_ALWAYS, + (TSTR +- ("yaffs tragedy: no more eraased blocks, but there should have been %d" ++ ("yaffs tragedy: no more erased blocks, but there should have been %d" + TENDSTR), dev->nErasedBlocks)); + + return -1; + } + + +-// Check if there's space to allocate... +-// Thinks.... do we need top make this ths same as yaffs_GetFreeChunks()? +-static int yaffs_CheckSpaceForAllocation(yaffs_Device * dev) ++ ++static int yaffs_CalcCheckpointBlocksRequired(yaffs_Device *dev) ++{ ++ if (!dev->nCheckpointBlocksRequired && ++ dev->isYaffs2) { ++ /* Not a valid value so recalculate */ ++ int nBytes = 0; ++ int nBlocks; ++ int devBlocks = (dev->endBlock - dev->startBlock + 1); ++ int tnodeSize; ++ ++ tnodeSize = (dev->tnodeWidth * YAFFS_NTNODES_LEVEL0)/8; ++ ++ if (tnodeSize < sizeof(yaffs_Tnode)) ++ tnodeSize = sizeof(yaffs_Tnode); ++ ++ nBytes += sizeof(yaffs_CheckpointValidity); ++ nBytes += sizeof(yaffs_CheckpointDevice); ++ nBytes += devBlocks * sizeof(yaffs_BlockInfo); ++ nBytes += devBlocks * dev->chunkBitmapStride; ++ nBytes += (sizeof(yaffs_CheckpointObject) + sizeof(__u32)) * (dev->nObjectsCreated - dev->nFreeObjects); ++ nBytes += (tnodeSize + sizeof(__u32)) * (dev->nTnodesCreated - dev->nFreeTnodes); ++ nBytes += sizeof(yaffs_CheckpointValidity); ++ nBytes += sizeof(__u32); /* checksum*/ ++ ++ /* Round up and add 2 blocks to allow for some bad blocks, so add 3 */ ++ ++ nBlocks = (nBytes/(dev->nDataBytesPerChunk * dev->nChunksPerBlock)) + 3; ++ ++ dev->nCheckpointBlocksRequired = nBlocks; ++ } ++ ++ return dev->nCheckpointBlocksRequired; ++} ++ ++/* ++ * Check if there's space to allocate... ++ * Thinks.... do we need top make this ths same as yaffs_GetFreeChunks()? ++ */ ++static int yaffs_CheckSpaceForAllocation(yaffs_Device *dev) + { + int reservedChunks; + int reservedBlocks = dev->nReservedBlocks; + int checkpointBlocks; + +- checkpointBlocks = dev->nCheckpointReservedBlocks - dev->blocksInCheckpoint; +- if(checkpointBlocks < 0) ++ if (dev->isYaffs2) { ++ checkpointBlocks = yaffs_CalcCheckpointBlocksRequired(dev) - ++ dev->blocksInCheckpoint; ++ if (checkpointBlocks < 0) ++ checkpointBlocks = 0; ++ } else { + checkpointBlocks = 0; ++ } + + reservedChunks = ((reservedBlocks + checkpointBlocks) * dev->nChunksPerBlock); + + return (dev->nFreeChunks > reservedChunks); + } + +-static int yaffs_AllocateChunk(yaffs_Device * dev, int useReserve, yaffs_BlockInfo **blockUsedPtr) ++static int yaffs_AllocateChunk(yaffs_Device *dev, int useReserve, ++ yaffs_BlockInfo **blockUsedPtr) + { + int retVal; + yaffs_BlockInfo *bi; +@@ -2835,7 +2891,7 @@ static int yaffs_AllocateChunk(yaffs_Dev + } + + if (dev->nErasedBlocks < dev->nReservedBlocks +- && dev->allocationPage == 0) { ++ && dev->allocationPage == 0) { + T(YAFFS_TRACE_ALLOCATE, (TSTR("Allocating reserve" TENDSTR))); + } + +@@ -2844,10 +2900,10 @@ static int yaffs_AllocateChunk(yaffs_Dev + bi = yaffs_GetBlockInfo(dev, dev->allocationBlock); + + retVal = (dev->allocationBlock * dev->nChunksPerBlock) + +- dev->allocationPage; ++ dev->allocationPage; + bi->pagesInUse++; + yaffs_SetChunkBit(dev, dev->allocationBlock, +- dev->allocationPage); ++ dev->allocationPage); + + dev->allocationPage++; + +@@ -2859,43 +2915,43 @@ static int yaffs_AllocateChunk(yaffs_Dev + dev->allocationBlock = -1; + } + +- if(blockUsedPtr) ++ if (blockUsedPtr) + *blockUsedPtr = bi; + + return retVal; + } + + T(YAFFS_TRACE_ERROR, +- (TSTR("!!!!!!!!! Allocator out !!!!!!!!!!!!!!!!!" TENDSTR))); ++ (TSTR("!!!!!!!!! Allocator out !!!!!!!!!!!!!!!!!" TENDSTR))); + + return -1; + } + +-static int yaffs_GetErasedChunks(yaffs_Device * dev) ++static int yaffs_GetErasedChunks(yaffs_Device *dev) + { + int n; + + n = dev->nErasedBlocks * dev->nChunksPerBlock; + +- if (dev->allocationBlock > 0) { ++ if (dev->allocationBlock > 0) + n += (dev->nChunksPerBlock - dev->allocationPage); +- } + + return n; + + } + +-static int yaffs_GarbageCollectBlock(yaffs_Device * dev, int block) ++static int yaffs_GarbageCollectBlock(yaffs_Device *dev, int block, ++ int wholeBlock) + { + int oldChunk; + int newChunk; +- int chunkInBlock; + int markNAND; + int retVal = YAFFS_OK; + int cleanups = 0; + int i; + int isCheckpointBlock; + int matchingChunk; ++ int maxCopies; + + int chunksBefore = yaffs_GetErasedChunks(dev); + int chunksAfter; +@@ -2911,8 +2967,11 @@ static int yaffs_GarbageCollectBlock(yaf + bi->blockState = YAFFS_BLOCK_STATE_COLLECTING; + + T(YAFFS_TRACE_TRACING, +- (TSTR("Collecting block %d, in use %d, shrink %d, " TENDSTR), block, +- bi->pagesInUse, bi->hasShrinkHeader)); ++ (TSTR("Collecting block %d, in use %d, shrink %d, wholeBlock %d" TENDSTR), ++ block, ++ bi->pagesInUse, ++ bi->hasShrinkHeader, ++ wholeBlock)); + + /*yaffs_VerifyFreeChunks(dev); */ + +@@ -2926,26 +2985,33 @@ static int yaffs_GarbageCollectBlock(yaf + dev->isDoingGC = 1; + + if (isCheckpointBlock || +- !yaffs_StillSomeChunkBits(dev, block)) { ++ !yaffs_StillSomeChunkBits(dev, block)) { + T(YAFFS_TRACE_TRACING, +- (TSTR +- ("Collecting block %d that has no chunks in use" TENDSTR), +- block)); ++ (TSTR ++ ("Collecting block %d that has no chunks in use" TENDSTR), ++ block)); + yaffs_BlockBecameDirty(dev, block); + } else { + + __u8 *buffer = yaffs_GetTempBuffer(dev, __LINE__); + +- yaffs_VerifyBlock(dev,bi,block); ++ yaffs_VerifyBlock(dev, bi, block); + +- for (chunkInBlock = 0, oldChunk = block * dev->nChunksPerBlock; +- chunkInBlock < dev->nChunksPerBlock +- && yaffs_StillSomeChunkBits(dev, block); +- chunkInBlock++, oldChunk++) { +- if (yaffs_CheckChunkBit(dev, block, chunkInBlock)) { ++ maxCopies = (wholeBlock) ? dev->nChunksPerBlock : 10; ++ oldChunk = block * dev->nChunksPerBlock + dev->gcChunk; ++ ++ for (/* init already done */; ++ retVal == YAFFS_OK && ++ dev->gcChunk < dev->nChunksPerBlock && ++ (bi->blockState == YAFFS_BLOCK_STATE_COLLECTING) && ++ maxCopies > 0; ++ dev->gcChunk++, oldChunk++) { ++ if (yaffs_CheckChunkBit(dev, block, dev->gcChunk)) { + + /* This page is in use and might need to be copied off */ + ++ maxCopies--; ++ + markNAND = 1; + + yaffs_InitialiseTags(&tags); +@@ -2959,22 +3025,22 @@ static int yaffs_GarbageCollectBlock(yaf + + T(YAFFS_TRACE_GC_DETAIL, + (TSTR +- ("Collecting page %d, %d %d %d " TENDSTR), +- chunkInBlock, tags.objectId, tags.chunkId, ++ ("Collecting chunk in block %d, %d %d %d " TENDSTR), ++ dev->gcChunk, tags.objectId, tags.chunkId, + tags.byteCount)); + +- if(object && !yaffs_SkipVerification(dev)){ +- if(tags.chunkId == 0) +- matchingChunk = object->chunkId; +- else if(object->softDeleted) ++ if (object && !yaffs_SkipVerification(dev)) { ++ if (tags.chunkId == 0) ++ matchingChunk = object->hdrChunk; ++ else if (object->softDeleted) + matchingChunk = oldChunk; /* Defeat the test */ + else +- matchingChunk = yaffs_FindChunkInFile(object,tags.chunkId,NULL); ++ matchingChunk = yaffs_FindChunkInFile(object, tags.chunkId, NULL); + +- if(oldChunk != matchingChunk) ++ if (oldChunk != matchingChunk) + T(YAFFS_TRACE_ERROR, + (TSTR("gc: page in gc mismatch: %d %d %d %d"TENDSTR), +- oldChunk,matchingChunk,tags.objectId, tags.chunkId)); ++ oldChunk, matchingChunk, tags.objectId, tags.chunkId)); + + } + +@@ -2986,9 +3052,11 @@ static int yaffs_GarbageCollectBlock(yaf + tags.objectId, tags.chunkId, tags.byteCount)); + } + +- if (object && object->deleted +- && tags.chunkId != 0) { +- /* Data chunk in a deleted file, throw it away ++ if (object && ++ object->deleted && ++ object->softDeleted && ++ tags.chunkId != 0) { ++ /* Data chunk in a soft deleted file, throw it away + * It's a soft deleted data chunk, + * No need to copy this, just forget about it and + * fix up the object. +@@ -3003,13 +3071,12 @@ static int yaffs_GarbageCollectBlock(yaf + cleanups++; + } + markNAND = 0; +- } else if (0 +- /* Todo object && object->deleted && object->nDataChunks == 0 */ +- ) { ++ } else if (0) { ++ /* Todo object && object->deleted && object->nDataChunks == 0 */ + /* Deleted object header with no data chunks. + * Can be discarded and the file deleted. + */ +- object->chunkId = 0; ++ object->hdrChunk = 0; + yaffs_FreeTnode(object->myDev, + object->variant. + fileVariant.top); +@@ -3031,17 +3098,14 @@ static int yaffs_GarbageCollectBlock(yaf + * We need to nuke the shrinkheader flags first + * We no longer want the shrinkHeader flag since its work is done + * and if it is left in place it will mess up scanning. +- * Also, clear out any shadowing stuff + */ + + yaffs_ObjectHeader *oh; + oh = (yaffs_ObjectHeader *)buffer; + oh->isShrink = 0; +- oh->shadowsObject = -1; +- tags.extraShadows = 0; + tags.extraIsShrinkHeader = 0; + +- yaffs_VerifyObjectHeader(object,oh,&tags,1); ++ yaffs_VerifyObjectHeader(object, oh, &tags, 1); + } + + newChunk = +@@ -3055,7 +3119,7 @@ static int yaffs_GarbageCollectBlock(yaf + + if (tags.chunkId == 0) { + /* It's a header */ +- object->chunkId = newChunk; ++ object->hdrChunk = newChunk; + object->serial = tags.serialNumber; + } else { + /* It's a data chunk */ +@@ -3067,7 +3131,8 @@ static int yaffs_GarbageCollectBlock(yaf + } + } + +- yaffs_DeleteChunk(dev, oldChunk, markNAND, __LINE__); ++ if (retVal == YAFFS_OK) ++ yaffs_DeleteChunk(dev, oldChunk, markNAND, __LINE__); + + } + } +@@ -3098,18 +3163,25 @@ static int yaffs_GarbageCollectBlock(yaf + + } + +- yaffs_VerifyCollectedBlock(dev,bi,block); ++ yaffs_VerifyCollectedBlock(dev, bi, block); + +- if (chunksBefore >= (chunksAfter = yaffs_GetErasedChunks(dev))) { ++ chunksAfter = yaffs_GetErasedChunks(dev); ++ if (chunksBefore >= chunksAfter) { + T(YAFFS_TRACE_GC, + (TSTR + ("gc did not increase free chunks before %d after %d" + TENDSTR), chunksBefore, chunksAfter)); + } + ++ /* If the gc completed then clear the current gcBlock so that we find another. */ ++ if (bi->blockState != YAFFS_BLOCK_STATE_COLLECTING) { ++ dev->gcBlock = -1; ++ dev->gcChunk = 0; ++ } ++ + dev->isDoingGC = 0; + +- return YAFFS_OK; ++ return retVal; + } + + /* New garbage collector +@@ -3121,7 +3193,7 @@ static int yaffs_GarbageCollectBlock(yaf + * The idea is to help clear out space in a more spread-out manner. + * Dunno if it really does anything useful. + */ +-static int yaffs_CheckGarbageCollection(yaffs_Device * dev) ++static int yaffs_CheckGarbageCollection(yaffs_Device *dev) + { + int block; + int aggressive; +@@ -3142,8 +3214,8 @@ static int yaffs_CheckGarbageCollection( + do { + maxTries++; + +- checkpointBlockAdjust = (dev->nCheckpointReservedBlocks - dev->blocksInCheckpoint); +- if(checkpointBlockAdjust < 0) ++ checkpointBlockAdjust = yaffs_CalcCheckpointBlocksRequired(dev) - dev->blocksInCheckpoint; ++ if (checkpointBlockAdjust < 0) + checkpointBlockAdjust = 0; + + if (dev->nErasedBlocks < (dev->nReservedBlocks + checkpointBlockAdjust + 2)) { +@@ -3154,20 +3226,24 @@ static int yaffs_CheckGarbageCollection( + aggressive = 0; + } + +- block = yaffs_FindBlockForGarbageCollection(dev, aggressive); ++ if (dev->gcBlock <= 0) { ++ dev->gcBlock = yaffs_FindBlockForGarbageCollection(dev, aggressive); ++ dev->gcChunk = 0; ++ } ++ ++ block = dev->gcBlock; + + if (block > 0) { + dev->garbageCollections++; +- if (!aggressive) { ++ if (!aggressive) + dev->passiveGarbageCollections++; +- } + + T(YAFFS_TRACE_GC, + (TSTR + ("yaffs: GC erasedBlocks %d aggressive %d" TENDSTR), + dev->nErasedBlocks, aggressive)); + +- gcOk = yaffs_GarbageCollectBlock(dev, block); ++ gcOk = yaffs_GarbageCollectBlock(dev, block, aggressive); + } + + if (dev->nErasedBlocks < (dev->nReservedBlocks) && block > 0) { +@@ -3176,15 +3252,16 @@ static int yaffs_CheckGarbageCollection( + ("yaffs: GC !!!no reclaim!!! erasedBlocks %d after try %d block %d" + TENDSTR), dev->nErasedBlocks, maxTries, block)); + } +- } while ((dev->nErasedBlocks < dev->nReservedBlocks) && (block > 0) +- && (maxTries < 2)); ++ } while ((dev->nErasedBlocks < dev->nReservedBlocks) && ++ (block > 0) && ++ (maxTries < 2)); + + return aggressive ? gcOk : YAFFS_OK; + } + + /*------------------------- TAGS --------------------------------*/ + +-static int yaffs_TagsMatch(const yaffs_ExtendedTags * tags, int objectId, ++static int yaffs_TagsMatch(const yaffs_ExtendedTags *tags, int objectId, + int chunkInObject) + { + return (tags->chunkId == chunkInObject && +@@ -3195,8 +3272,8 @@ static int yaffs_TagsMatch(const yaffs_E + + /*-------------------- Data file manipulation -----------------*/ + +-static int yaffs_FindChunkInFile(yaffs_Object * in, int chunkInInode, +- yaffs_ExtendedTags * tags) ++static int yaffs_FindChunkInFile(yaffs_Object *in, int chunkInInode, ++ yaffs_ExtendedTags *tags) + { + /*Get the Tnode, then get the level 0 offset chunk offset */ + yaffs_Tnode *tn; +@@ -3214,7 +3291,7 @@ static int yaffs_FindChunkInFile(yaffs_O + tn = yaffs_FindLevel0Tnode(dev, &in->variant.fileVariant, chunkInInode); + + if (tn) { +- theChunk = yaffs_GetChunkGroupBase(dev,tn,chunkInInode); ++ theChunk = yaffs_GetChunkGroupBase(dev, tn, chunkInInode); + + retVal = + yaffs_FindChunkInGroup(dev, theChunk, tags, in->objectId, +@@ -3223,8 +3300,8 @@ static int yaffs_FindChunkInFile(yaffs_O + return retVal; + } + +-static int yaffs_FindAndDeleteChunkInFile(yaffs_Object * in, int chunkInInode, +- yaffs_ExtendedTags * tags) ++static int yaffs_FindAndDeleteChunkInFile(yaffs_Object *in, int chunkInInode, ++ yaffs_ExtendedTags *tags) + { + /* Get the Tnode, then get the level 0 offset chunk offset */ + yaffs_Tnode *tn; +@@ -3243,29 +3320,23 @@ static int yaffs_FindAndDeleteChunkInFil + + if (tn) { + +- theChunk = yaffs_GetChunkGroupBase(dev,tn,chunkInInode); ++ theChunk = yaffs_GetChunkGroupBase(dev, tn, chunkInInode); + + retVal = + yaffs_FindChunkInGroup(dev, theChunk, tags, in->objectId, + chunkInInode); + + /* Delete the entry in the filestructure (if found) */ +- if (retVal != -1) { +- yaffs_PutLevel0Tnode(dev,tn,chunkInInode,0); +- } +- } else { +- /*T(("No level 0 found for %d\n", chunkInInode)); */ ++ if (retVal != -1) ++ yaffs_PutLevel0Tnode(dev, tn, chunkInInode, 0); + } + +- if (retVal == -1) { +- /* T(("Could not find %d to delete\n",chunkInInode)); */ +- } + return retVal; + } + + #ifdef YAFFS_PARANOID + +-static int yaffs_CheckFileSanity(yaffs_Object * in) ++static int yaffs_CheckFileSanity(yaffs_Object *in) + { + int chunk; + int nChunks; +@@ -3278,10 +3349,8 @@ static int yaffs_CheckFileSanity(yaffs_O + int theChunk; + int chunkDeleted; + +- if (in->variantType != YAFFS_OBJECT_TYPE_FILE) { +- /* T(("Object not a file\n")); */ ++ if (in->variantType != YAFFS_OBJECT_TYPE_FILE) + return YAFFS_FAIL; +- } + + objId = in->objectId; + fSize = in->variant.fileVariant.fileSize; +@@ -3294,7 +3363,7 @@ static int yaffs_CheckFileSanity(yaffs_O + + if (tn) { + +- theChunk = yaffs_GetChunkGroupBase(dev,tn,chunk); ++ theChunk = yaffs_GetChunkGroupBase(dev, tn, chunk); + + if (yaffs_CheckChunkBits + (dev, theChunk / dev->nChunksPerBlock, +@@ -3323,7 +3392,7 @@ static int yaffs_CheckFileSanity(yaffs_O + + #endif + +-static int yaffs_PutChunkIntoFile(yaffs_Object * in, int chunkInInode, ++static int yaffs_PutChunkIntoFile(yaffs_Object *in, int chunkInInode, + int chunkInNAND, int inScan) + { + /* NB inScan is zero unless scanning. +@@ -3358,11 +3427,10 @@ static int yaffs_PutChunkIntoFile(yaffs_ + &in->variant.fileVariant, + chunkInInode, + NULL); +- if (!tn) { ++ if (!tn) + return YAFFS_FAIL; +- } + +- existingChunk = yaffs_GetChunkGroupBase(dev,tn,chunkInInode); ++ existingChunk = yaffs_GetChunkGroupBase(dev, tn, chunkInInode); + + if (inScan != 0) { + /* If we're scanning then we need to test for duplicates +@@ -3374,7 +3442,7 @@ static int yaffs_PutChunkIntoFile(yaffs_ + * Update: For backward scanning we don't need to re-read tags so this is quite cheap. + */ + +- if (existingChunk != 0) { ++ if (existingChunk > 0) { + /* NB Right now existing chunk will not be real chunkId if the device >= 32MB + * thus we have to do a FindChunkInFile to get the real chunk id. + * +@@ -3411,8 +3479,10 @@ static int yaffs_PutChunkIntoFile(yaffs_ + * not be loaded during a scan + */ + +- newSerial = newTags.serialNumber; +- existingSerial = existingTags.serialNumber; ++ if (inScan > 0) { ++ newSerial = newTags.serialNumber; ++ existingSerial = existingTags.serialNumber; ++ } + + if ((inScan > 0) && + (in->myDev->isYaffs2 || +@@ -3437,24 +3507,23 @@ static int yaffs_PutChunkIntoFile(yaffs_ + + } + +- if (existingChunk == 0) { ++ if (existingChunk == 0) + in->nDataChunks++; +- } + +- yaffs_PutLevel0Tnode(dev,tn,chunkInInode,chunkInNAND); ++ yaffs_PutLevel0Tnode(dev, tn, chunkInInode, chunkInNAND); + + return YAFFS_OK; + } + +-static int yaffs_ReadChunkDataFromObject(yaffs_Object * in, int chunkInInode, +- __u8 * buffer) ++static int yaffs_ReadChunkDataFromObject(yaffs_Object *in, int chunkInInode, ++ __u8 *buffer) + { + int chunkInNAND = yaffs_FindChunkInFile(in, chunkInInode, NULL); + +- if (chunkInNAND >= 0) { ++ if (chunkInNAND >= 0) + return yaffs_ReadChunkWithTagsFromNAND(in->myDev, chunkInNAND, +- buffer,NULL); +- } else { ++ buffer, NULL); ++ else { + T(YAFFS_TRACE_NANDACCESS, + (TSTR("Chunk %d not found zero instead" TENDSTR), + chunkInNAND)); +@@ -3465,7 +3534,7 @@ static int yaffs_ReadChunkDataFromObject + + } + +-void yaffs_DeleteChunk(yaffs_Device * dev, int chunkId, int markNAND, int lyn) ++void yaffs_DeleteChunk(yaffs_Device *dev, int chunkId, int markNAND, int lyn) + { + int block; + int page; +@@ -3475,16 +3544,15 @@ void yaffs_DeleteChunk(yaffs_Device * de + if (chunkId <= 0) + return; + +- + dev->nDeletions++; + block = chunkId / dev->nChunksPerBlock; + page = chunkId % dev->nChunksPerBlock; + + +- if(!yaffs_CheckChunkBit(dev,block,page)) ++ if (!yaffs_CheckChunkBit(dev, block, page)) + T(YAFFS_TRACE_VERIFY, +- (TSTR("Deleting invalid chunk %d"TENDSTR), +- chunkId)); ++ (TSTR("Deleting invalid chunk %d"TENDSTR), ++ chunkId)); + + bi = yaffs_GetBlockInfo(dev, block); + +@@ -3524,14 +3592,12 @@ void yaffs_DeleteChunk(yaffs_Device * de + yaffs_BlockBecameDirty(dev, block); + } + +- } else { +- /* T(("Bad news deleting chunk %d\n",chunkId)); */ + } + + } + +-static int yaffs_WriteChunkDataToObject(yaffs_Object * in, int chunkInInode, +- const __u8 * buffer, int nBytes, ++static int yaffs_WriteChunkDataToObject(yaffs_Object *in, int chunkInInode, ++ const __u8 *buffer, int nBytes, + int useReserve) + { + /* Find old chunk Need to do this to get serial number +@@ -3561,6 +3627,12 @@ static int yaffs_WriteChunkDataToObject( + (prevChunkId >= 0) ? prevTags.serialNumber + 1 : 1; + newTags.byteCount = nBytes; + ++ if (nBytes < 1 || nBytes > dev->totalBytesPerChunk) { ++ T(YAFFS_TRACE_ERROR, ++ (TSTR("Writing %d bytes to chunk!!!!!!!!!" TENDSTR), nBytes)); ++ YBUG(); ++ } ++ + newChunkId = + yaffs_WriteNewChunkWithTagsToNAND(dev, buffer, &newTags, + useReserve); +@@ -3568,11 +3640,9 @@ static int yaffs_WriteChunkDataToObject( + if (newChunkId >= 0) { + yaffs_PutChunkIntoFile(in, chunkInInode, newChunkId, 0); + +- if (prevChunkId >= 0) { ++ if (prevChunkId >= 0) + yaffs_DeleteChunk(dev, prevChunkId, 1, __LINE__); + +- } +- + yaffs_CheckFileSanity(in); + } + return newChunkId; +@@ -3582,7 +3652,7 @@ static int yaffs_WriteChunkDataToObject( + /* UpdateObjectHeader updates the header on NAND for an object. + * If name is not NULL, then that new name is used. + */ +-int yaffs_UpdateObjectHeader(yaffs_Object * in, const YCHAR * name, int force, ++int yaffs_UpdateObjectHeader(yaffs_Object *in, const YCHAR *name, int force, + int isShrink, int shadows) + { + +@@ -3603,9 +3673,12 @@ int yaffs_UpdateObjectHeader(yaffs_Objec + + yaffs_ObjectHeader *oh = NULL; + +- yaffs_strcpy(oldName,"silly old name"); ++ yaffs_strcpy(oldName, _Y("silly old name")); + +- if (!in->fake || force) { ++ ++ if (!in->fake || ++ in == dev->rootDir || /* The rootDir should also be saved */ ++ force) { + + yaffs_CheckGarbageCollection(dev); + yaffs_CheckObjectDetailsLoaded(in); +@@ -3613,13 +3686,13 @@ int yaffs_UpdateObjectHeader(yaffs_Objec + buffer = yaffs_GetTempBuffer(in->myDev, __LINE__); + oh = (yaffs_ObjectHeader *) buffer; + +- prevChunkId = in->chunkId; ++ prevChunkId = in->hdrChunk; + +- if (prevChunkId >= 0) { ++ if (prevChunkId > 0) { + result = yaffs_ReadChunkWithTagsFromNAND(dev, prevChunkId, + buffer, &oldTags); + +- yaffs_VerifyObjectHeader(in,oh,&oldTags,0); ++ yaffs_VerifyObjectHeader(in, oh, &oldTags, 0); + + memcpy(oldName, oh->name, sizeof(oh->name)); + } +@@ -3628,7 +3701,7 @@ int yaffs_UpdateObjectHeader(yaffs_Objec + + oh->type = in->variantType; + oh->yst_mode = in->yst_mode; +- oh->shadowsObject = shadows; ++ oh->shadowsObject = oh->inbandShadowsObject = shadows; + + #ifdef CONFIG_YAFFS_WINCE + oh->win_atime[0] = in->win_atime[0]; +@@ -3645,20 +3718,18 @@ int yaffs_UpdateObjectHeader(yaffs_Objec + oh->yst_ctime = in->yst_ctime; + oh->yst_rdev = in->yst_rdev; + #endif +- if (in->parent) { ++ if (in->parent) + oh->parentObjectId = in->parent->objectId; +- } else { ++ else + oh->parentObjectId = 0; +- } + + if (name && *name) { + memset(oh->name, 0, sizeof(oh->name)); + yaffs_strncpy(oh->name, name, YAFFS_MAX_NAME_LENGTH); +- } else if (prevChunkId>=0) { ++ } else if (prevChunkId >= 0) + memcpy(oh->name, oldName, sizeof(oh->name)); +- } else { ++ else + memset(oh->name, 0, sizeof(oh->name)); +- } + + oh->isShrink = isShrink; + +@@ -3708,7 +3779,7 @@ int yaffs_UpdateObjectHeader(yaffs_Objec + newTags.extraShadows = (oh->shadowsObject > 0) ? 1 : 0; + newTags.extraObjectType = in->variantType; + +- yaffs_VerifyObjectHeader(in,oh,&newTags,1); ++ yaffs_VerifyObjectHeader(in, oh, &newTags, 1); + + /* Create new chunk in NAND */ + newChunkId = +@@ -3717,20 +3788,20 @@ int yaffs_UpdateObjectHeader(yaffs_Objec + + if (newChunkId >= 0) { + +- in->chunkId = newChunkId; ++ in->hdrChunk = newChunkId; + + if (prevChunkId >= 0) { + yaffs_DeleteChunk(dev, prevChunkId, 1, + __LINE__); + } + +- if(!yaffs_ObjectHasCachedWriteData(in)) ++ if (!yaffs_ObjectHasCachedWriteData(in)) + in->dirty = 0; + + /* If this was a shrink, then mark the block that the chunk lives on */ + if (isShrink) { + bi = yaffs_GetBlockInfo(in->myDev, +- newChunkId /in->myDev-> nChunksPerBlock); ++ newChunkId / in->myDev->nChunksPerBlock); + bi->hasShrinkHeader = 1; + } + +@@ -3766,7 +3837,7 @@ static int yaffs_ObjectHasCachedWriteDat + yaffs_ChunkCache *cache; + int nCaches = obj->myDev->nShortOpCaches; + +- for(i = 0; i < nCaches; i++){ ++ for (i = 0; i < nCaches; i++) { + cache = &dev->srCache[i]; + if (cache->object == obj && + cache->dirty) +@@ -3777,7 +3848,7 @@ static int yaffs_ObjectHasCachedWriteDat + } + + +-static void yaffs_FlushFilesChunkCache(yaffs_Object * obj) ++static void yaffs_FlushFilesChunkCache(yaffs_Object *obj) + { + yaffs_Device *dev = obj->myDev; + int lowest = -99; /* Stop compiler whining. */ +@@ -3844,16 +3915,16 @@ void yaffs_FlushEntireDeviceCache(yaffs_ + */ + do { + obj = NULL; +- for( i = 0; i < nCaches && !obj; i++) { ++ for (i = 0; i < nCaches && !obj; i++) { + if (dev->srCache[i].object && + dev->srCache[i].dirty) + obj = dev->srCache[i].object; + + } +- if(obj) ++ if (obj) + yaffs_FlushFilesChunkCache(obj); + +- } while(obj); ++ } while (obj); + + } + +@@ -3863,41 +3934,21 @@ void yaffs_FlushEntireDeviceCache(yaffs_ + * Then look for the least recently used non-dirty one. + * Then look for the least recently used dirty one...., flush and look again. + */ +-static yaffs_ChunkCache *yaffs_GrabChunkCacheWorker(yaffs_Device * dev) ++static yaffs_ChunkCache *yaffs_GrabChunkCacheWorker(yaffs_Device *dev) + { + int i; +- int usage; +- int theOne; + + if (dev->nShortOpCaches > 0) { + for (i = 0; i < dev->nShortOpCaches; i++) { + if (!dev->srCache[i].object) + return &dev->srCache[i]; + } ++ } + +- return NULL; ++ return NULL; ++} + +- theOne = -1; +- usage = 0; /* just to stop the compiler grizzling */ +- +- for (i = 0; i < dev->nShortOpCaches; i++) { +- if (!dev->srCache[i].dirty && +- ((dev->srCache[i].lastUse < usage && theOne >= 0) || +- theOne < 0)) { +- usage = dev->srCache[i].lastUse; +- theOne = i; +- } +- } +- +- +- return theOne >= 0 ? &dev->srCache[theOne] : NULL; +- } else { +- return NULL; +- } +- +-} +- +-static yaffs_ChunkCache *yaffs_GrabChunkCache(yaffs_Device * dev) ++static yaffs_ChunkCache *yaffs_GrabChunkCache(yaffs_Device *dev) + { + yaffs_ChunkCache *cache; + yaffs_Object *theObj; +@@ -3927,8 +3978,7 @@ static yaffs_ChunkCache *yaffs_GrabChunk + for (i = 0; i < dev->nShortOpCaches; i++) { + if (dev->srCache[i].object && + !dev->srCache[i].locked && +- (dev->srCache[i].lastUse < usage || !cache)) +- { ++ (dev->srCache[i].lastUse < usage || !cache)) { + usage = dev->srCache[i].lastUse; + theObj = dev->srCache[i].object; + cache = &dev->srCache[i]; +@@ -3950,7 +4000,7 @@ static yaffs_ChunkCache *yaffs_GrabChunk + } + + /* Find a cached chunk */ +-static yaffs_ChunkCache *yaffs_FindChunkCache(const yaffs_Object * obj, ++static yaffs_ChunkCache *yaffs_FindChunkCache(const yaffs_Object *obj, + int chunkId) + { + yaffs_Device *dev = obj->myDev; +@@ -3969,7 +4019,7 @@ static yaffs_ChunkCache *yaffs_FindChunk + } + + /* Mark the chunk for the least recently used algorithym */ +-static void yaffs_UseChunkCache(yaffs_Device * dev, yaffs_ChunkCache * cache, ++static void yaffs_UseChunkCache(yaffs_Device *dev, yaffs_ChunkCache *cache, + int isAWrite) + { + +@@ -3977,9 +4027,9 @@ static void yaffs_UseChunkCache(yaffs_De + if (dev->srLastUse < 0 || dev->srLastUse > 100000000) { + /* Reset the cache usages */ + int i; +- for (i = 1; i < dev->nShortOpCaches; i++) { ++ for (i = 1; i < dev->nShortOpCaches; i++) + dev->srCache[i].lastUse = 0; +- } ++ + dev->srLastUse = 0; + } + +@@ -3987,9 +4037,8 @@ static void yaffs_UseChunkCache(yaffs_De + + cache->lastUse = dev->srLastUse; + +- if (isAWrite) { ++ if (isAWrite) + cache->dirty = 1; +- } + } + } + +@@ -3997,21 +4046,20 @@ static void yaffs_UseChunkCache(yaffs_De + * Do this when a whole page gets written, + * ie the short cache for this page is no longer valid. + */ +-static void yaffs_InvalidateChunkCache(yaffs_Object * object, int chunkId) ++static void yaffs_InvalidateChunkCache(yaffs_Object *object, int chunkId) + { + if (object->myDev->nShortOpCaches > 0) { + yaffs_ChunkCache *cache = yaffs_FindChunkCache(object, chunkId); + +- if (cache) { ++ if (cache) + cache->object = NULL; +- } + } + } + + /* Invalidate all the cache pages associated with this object + * Do this whenever ther file is deleted or resized. + */ +-static void yaffs_InvalidateWholeChunkCache(yaffs_Object * in) ++static void yaffs_InvalidateWholeChunkCache(yaffs_Object *in) + { + int i; + yaffs_Device *dev = in->myDev; +@@ -4019,9 +4067,8 @@ static void yaffs_InvalidateWholeChunkCa + if (dev->nShortOpCaches > 0) { + /* Invalidate it. */ + for (i = 0; i < dev->nShortOpCaches; i++) { +- if (dev->srCache[i].object == in) { ++ if (dev->srCache[i].object == in) + dev->srCache[i].object = NULL; +- } + } + } + } +@@ -4029,18 +4076,18 @@ static void yaffs_InvalidateWholeChunkCa + /*--------------------- Checkpointing --------------------*/ + + +-static int yaffs_WriteCheckpointValidityMarker(yaffs_Device *dev,int head) ++static int yaffs_WriteCheckpointValidityMarker(yaffs_Device *dev, int head) + { + yaffs_CheckpointValidity cp; + +- memset(&cp,0,sizeof(cp)); ++ memset(&cp, 0, sizeof(cp)); + + cp.structType = sizeof(cp); + cp.magic = YAFFS_MAGIC; + cp.version = YAFFS_CHECKPOINT_VERSION; + cp.head = (head) ? 1 : 0; + +- return (yaffs_CheckpointWrite(dev,&cp,sizeof(cp)) == sizeof(cp))? ++ return (yaffs_CheckpointWrite(dev, &cp, sizeof(cp)) == sizeof(cp)) ? + 1 : 0; + } + +@@ -4049,9 +4096,9 @@ static int yaffs_ReadCheckpointValidityM + yaffs_CheckpointValidity cp; + int ok; + +- ok = (yaffs_CheckpointRead(dev,&cp,sizeof(cp)) == sizeof(cp)); ++ ok = (yaffs_CheckpointRead(dev, &cp, sizeof(cp)) == sizeof(cp)); + +- if(ok) ++ if (ok) + ok = (cp.structType == sizeof(cp)) && + (cp.magic == YAFFS_MAGIC) && + (cp.version == YAFFS_CHECKPOINT_VERSION) && +@@ -4100,21 +4147,21 @@ static int yaffs_WriteCheckpointDevice(y + int ok; + + /* Write device runtime values*/ +- yaffs_DeviceToCheckpointDevice(&cp,dev); ++ yaffs_DeviceToCheckpointDevice(&cp, dev); + cp.structType = sizeof(cp); + +- ok = (yaffs_CheckpointWrite(dev,&cp,sizeof(cp)) == sizeof(cp)); ++ ok = (yaffs_CheckpointWrite(dev, &cp, sizeof(cp)) == sizeof(cp)); + + /* Write block info */ +- if(ok) { ++ if (ok) { + nBytes = nBlocks * sizeof(yaffs_BlockInfo); +- ok = (yaffs_CheckpointWrite(dev,dev->blockInfo,nBytes) == nBytes); ++ ok = (yaffs_CheckpointWrite(dev, dev->blockInfo, nBytes) == nBytes); + } + + /* Write chunk bits */ +- if(ok) { ++ if (ok) { + nBytes = nBlocks * dev->chunkBitmapStride; +- ok = (yaffs_CheckpointWrite(dev,dev->chunkBits,nBytes) == nBytes); ++ ok = (yaffs_CheckpointWrite(dev, dev->chunkBits, nBytes) == nBytes); + } + return ok ? 1 : 0; + +@@ -4128,25 +4175,25 @@ static int yaffs_ReadCheckpointDevice(ya + + int ok; + +- ok = (yaffs_CheckpointRead(dev,&cp,sizeof(cp)) == sizeof(cp)); +- if(!ok) ++ ok = (yaffs_CheckpointRead(dev, &cp, sizeof(cp)) == sizeof(cp)); ++ if (!ok) + return 0; + +- if(cp.structType != sizeof(cp)) ++ if (cp.structType != sizeof(cp)) + return 0; + + +- yaffs_CheckpointDeviceToDevice(dev,&cp); ++ yaffs_CheckpointDeviceToDevice(dev, &cp); + + nBytes = nBlocks * sizeof(yaffs_BlockInfo); + +- ok = (yaffs_CheckpointRead(dev,dev->blockInfo,nBytes) == nBytes); ++ ok = (yaffs_CheckpointRead(dev, dev->blockInfo, nBytes) == nBytes); + +- if(!ok) ++ if (!ok) + return 0; + nBytes = nBlocks * dev->chunkBitmapStride; + +- ok = (yaffs_CheckpointRead(dev,dev->chunkBits,nBytes) == nBytes); ++ ok = (yaffs_CheckpointRead(dev, dev->chunkBits, nBytes) == nBytes); + + return ok ? 1 : 0; + } +@@ -4157,7 +4204,7 @@ static void yaffs_ObjectToCheckpointObje + + cp->objectId = obj->objectId; + cp->parentId = (obj->parent) ? obj->parent->objectId : 0; +- cp->chunkId = obj->chunkId; ++ cp->hdrChunk = obj->hdrChunk; + cp->variantType = obj->variantType; + cp->deleted = obj->deleted; + cp->softDeleted = obj->softDeleted; +@@ -4168,20 +4215,28 @@ static void yaffs_ObjectToCheckpointObje + cp->serial = obj->serial; + cp->nDataChunks = obj->nDataChunks; + +- if(obj->variantType == YAFFS_OBJECT_TYPE_FILE) ++ if (obj->variantType == YAFFS_OBJECT_TYPE_FILE) + cp->fileSizeOrEquivalentObjectId = obj->variant.fileVariant.fileSize; +- else if(obj->variantType == YAFFS_OBJECT_TYPE_HARDLINK) ++ else if (obj->variantType == YAFFS_OBJECT_TYPE_HARDLINK) + cp->fileSizeOrEquivalentObjectId = obj->variant.hardLinkVariant.equivalentObjectId; + } + +-static void yaffs_CheckpointObjectToObject( yaffs_Object *obj,yaffs_CheckpointObject *cp) ++static int yaffs_CheckpointObjectToObject(yaffs_Object *obj, yaffs_CheckpointObject *cp) + { + + yaffs_Object *parent; + ++ if (obj->variantType != cp->variantType) { ++ T(YAFFS_TRACE_ERROR, (TSTR("Checkpoint read object %d type %d " ++ TCONT("chunk %d does not match existing object type %d") ++ TENDSTR), cp->objectId, cp->variantType, cp->hdrChunk, ++ obj->variantType)); ++ return 0; ++ } ++ + obj->objectId = cp->objectId; + +- if(cp->parentId) ++ if (cp->parentId) + parent = yaffs_FindOrCreateObjectByNumber( + obj->myDev, + cp->parentId, +@@ -4189,10 +4244,19 @@ static void yaffs_CheckpointObjectToObje + else + parent = NULL; + +- if(parent) ++ if (parent) { ++ if (parent->variantType != YAFFS_OBJECT_TYPE_DIRECTORY) { ++ T(YAFFS_TRACE_ALWAYS, (TSTR("Checkpoint read object %d parent %d type %d" ++ TCONT(" chunk %d Parent type, %d, not directory") ++ TENDSTR), ++ cp->objectId, cp->parentId, cp->variantType, ++ cp->hdrChunk, parent->variantType)); ++ return 0; ++ } + yaffs_AddObjectToDirectory(parent, obj); ++ } + +- obj->chunkId = cp->chunkId; ++ obj->hdrChunk = cp->hdrChunk; + obj->variantType = cp->variantType; + obj->deleted = cp->deleted; + obj->softDeleted = cp->softDeleted; +@@ -4203,29 +4267,34 @@ static void yaffs_CheckpointObjectToObje + obj->serial = cp->serial; + obj->nDataChunks = cp->nDataChunks; + +- if(obj->variantType == YAFFS_OBJECT_TYPE_FILE) ++ if (obj->variantType == YAFFS_OBJECT_TYPE_FILE) + obj->variant.fileVariant.fileSize = cp->fileSizeOrEquivalentObjectId; +- else if(obj->variantType == YAFFS_OBJECT_TYPE_HARDLINK) ++ else if (obj->variantType == YAFFS_OBJECT_TYPE_HARDLINK) + obj->variant.hardLinkVariant.equivalentObjectId = cp->fileSizeOrEquivalentObjectId; + +- if(obj->objectId >= YAFFS_NOBJECT_BUCKETS) ++ if (obj->hdrChunk > 0) + obj->lazyLoaded = 1; ++ return 1; + } + + + +-static int yaffs_CheckpointTnodeWorker(yaffs_Object * in, yaffs_Tnode * tn, +- __u32 level, int chunkOffset) ++static int yaffs_CheckpointTnodeWorker(yaffs_Object *in, yaffs_Tnode *tn, ++ __u32 level, int chunkOffset) + { + int i; + yaffs_Device *dev = in->myDev; + int ok = 1; +- int nTnodeBytes = (dev->tnodeWidth * YAFFS_NTNODES_LEVEL0)/8; ++ int tnodeSize = (dev->tnodeWidth * YAFFS_NTNODES_LEVEL0)/8; ++ ++ if (tnodeSize < sizeof(yaffs_Tnode)) ++ tnodeSize = sizeof(yaffs_Tnode); ++ + + if (tn) { + if (level > 0) { + +- for (i = 0; i < YAFFS_NTNODES_INTERNAL && ok; i++){ ++ for (i = 0; i < YAFFS_NTNODES_INTERNAL && ok; i++) { + if (tn->internal[i]) { + ok = yaffs_CheckpointTnodeWorker(in, + tn->internal[i], +@@ -4235,10 +4304,9 @@ static int yaffs_CheckpointTnodeWorker(y + } + } else if (level == 0) { + __u32 baseOffset = chunkOffset << YAFFS_TNODES_LEVEL0_BITS; +- /* printf("write tnode at %d\n",baseOffset); */ +- ok = (yaffs_CheckpointWrite(dev,&baseOffset,sizeof(baseOffset)) == sizeof(baseOffset)); +- if(ok) +- ok = (yaffs_CheckpointWrite(dev,tn,nTnodeBytes) == nTnodeBytes); ++ ok = (yaffs_CheckpointWrite(dev, &baseOffset, sizeof(baseOffset)) == sizeof(baseOffset)); ++ if (ok) ++ ok = (yaffs_CheckpointWrite(dev, tn, tnodeSize) == tnodeSize); + } + } + +@@ -4251,13 +4319,13 @@ static int yaffs_WriteCheckpointTnodes(y + __u32 endMarker = ~0; + int ok = 1; + +- if(obj->variantType == YAFFS_OBJECT_TYPE_FILE){ ++ if (obj->variantType == YAFFS_OBJECT_TYPE_FILE) { + ok = yaffs_CheckpointTnodeWorker(obj, + obj->variant.fileVariant.top, + obj->variant.fileVariant.topLevel, + 0); +- if(ok) +- ok = (yaffs_CheckpointWrite(obj->myDev,&endMarker,sizeof(endMarker)) == ++ if (ok) ++ ok = (yaffs_CheckpointWrite(obj->myDev, &endMarker, sizeof(endMarker)) == + sizeof(endMarker)); + } + +@@ -4272,38 +4340,38 @@ static int yaffs_ReadCheckpointTnodes(ya + yaffs_FileStructure *fileStructPtr = &obj->variant.fileVariant; + yaffs_Tnode *tn; + int nread = 0; ++ int tnodeSize = (dev->tnodeWidth * YAFFS_NTNODES_LEVEL0)/8; + +- ok = (yaffs_CheckpointRead(dev,&baseChunk,sizeof(baseChunk)) == sizeof(baseChunk)); ++ if (tnodeSize < sizeof(yaffs_Tnode)) ++ tnodeSize = sizeof(yaffs_Tnode); + +- while(ok && (~baseChunk)){ ++ ok = (yaffs_CheckpointRead(dev, &baseChunk, sizeof(baseChunk)) == sizeof(baseChunk)); ++ ++ while (ok && (~baseChunk)) { + nread++; + /* Read level 0 tnode */ + + +- /* printf("read tnode at %d\n",baseChunk); */ + tn = yaffs_GetTnodeRaw(dev); +- if(tn) +- ok = (yaffs_CheckpointRead(dev,tn,(dev->tnodeWidth * YAFFS_NTNODES_LEVEL0)/8) == +- (dev->tnodeWidth * YAFFS_NTNODES_LEVEL0)/8); ++ if (tn) ++ ok = (yaffs_CheckpointRead(dev, tn, tnodeSize) == tnodeSize); + else + ok = 0; + +- if(tn && ok){ ++ if (tn && ok) + ok = yaffs_AddOrFindLevel0Tnode(dev, +- fileStructPtr, +- baseChunk, +- tn) ? 1 : 0; ++ fileStructPtr, ++ baseChunk, ++ tn) ? 1 : 0; + +- } +- +- if(ok) +- ok = (yaffs_CheckpointRead(dev,&baseChunk,sizeof(baseChunk)) == sizeof(baseChunk)); ++ if (ok) ++ ok = (yaffs_CheckpointRead(dev, &baseChunk, sizeof(baseChunk)) == sizeof(baseChunk)); + + } + +- T(YAFFS_TRACE_CHECKPOINT,( ++ T(YAFFS_TRACE_CHECKPOINT, ( + TSTR("Checkpoint read tnodes %d records, last %d. ok %d" TENDSTR), +- nread,baseChunk,ok)); ++ nread, baseChunk, ok)); + + return ok ? 1 : 0; + } +@@ -4315,41 +4383,40 @@ static int yaffs_WriteCheckpointObjects( + yaffs_CheckpointObject cp; + int i; + int ok = 1; +- struct list_head *lh; ++ struct ylist_head *lh; + + + /* Iterate through the objects in each hash entry, + * dumping them to the checkpointing stream. + */ + +- for(i = 0; ok && i < YAFFS_NOBJECT_BUCKETS; i++){ +- list_for_each(lh, &dev->objectBucket[i].list) { ++ for (i = 0; ok && i < YAFFS_NOBJECT_BUCKETS; i++) { ++ ylist_for_each(lh, &dev->objectBucket[i].list) { + if (lh) { +- obj = list_entry(lh, yaffs_Object, hashLink); ++ obj = ylist_entry(lh, yaffs_Object, hashLink); + if (!obj->deferedFree) { +- yaffs_ObjectToCheckpointObject(&cp,obj); ++ yaffs_ObjectToCheckpointObject(&cp, obj); + cp.structType = sizeof(cp); + +- T(YAFFS_TRACE_CHECKPOINT,( ++ T(YAFFS_TRACE_CHECKPOINT, ( + TSTR("Checkpoint write object %d parent %d type %d chunk %d obj addr %x" TENDSTR), +- cp.objectId,cp.parentId,cp.variantType,cp.chunkId,(unsigned) obj)); ++ cp.objectId, cp.parentId, cp.variantType, cp.hdrChunk, (unsigned) obj)); + +- ok = (yaffs_CheckpointWrite(dev,&cp,sizeof(cp)) == sizeof(cp)); ++ ok = (yaffs_CheckpointWrite(dev, &cp, sizeof(cp)) == sizeof(cp)); + +- if(ok && obj->variantType == YAFFS_OBJECT_TYPE_FILE){ ++ if (ok && obj->variantType == YAFFS_OBJECT_TYPE_FILE) + ok = yaffs_WriteCheckpointTnodes(obj); +- } + } + } + } +- } ++ } + +- /* Dump end of list */ +- memset(&cp,0xFF,sizeof(yaffs_CheckpointObject)); ++ /* Dump end of list */ ++ memset(&cp, 0xFF, sizeof(yaffs_CheckpointObject)); + cp.structType = sizeof(cp); + +- if(ok) +- ok = (yaffs_CheckpointWrite(dev,&cp,sizeof(cp)) == sizeof(cp)); ++ if (ok) ++ ok = (yaffs_CheckpointWrite(dev, &cp, sizeof(cp)) == sizeof(cp)); + + return ok ? 1 : 0; + } +@@ -4362,38 +4429,39 @@ static int yaffs_ReadCheckpointObjects(y + int done = 0; + yaffs_Object *hardList = NULL; + +- while(ok && !done) { +- ok = (yaffs_CheckpointRead(dev,&cp,sizeof(cp)) == sizeof(cp)); +- if(cp.structType != sizeof(cp)) { +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("struct size %d instead of %d ok %d"TENDSTR), +- cp.structType,sizeof(cp),ok)); ++ while (ok && !done) { ++ ok = (yaffs_CheckpointRead(dev, &cp, sizeof(cp)) == sizeof(cp)); ++ if (cp.structType != sizeof(cp)) { ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("struct size %d instead of %d ok %d"TENDSTR), ++ cp.structType, sizeof(cp), ok)); + ok = 0; + } + +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("Checkpoint read object %d parent %d type %d chunk %d " TENDSTR), +- cp.objectId,cp.parentId,cp.variantType,cp.chunkId)); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("Checkpoint read object %d parent %d type %d chunk %d " TENDSTR), ++ cp.objectId, cp.parentId, cp.variantType, cp.hdrChunk)); + +- if(ok && cp.objectId == ~0) ++ if (ok && cp.objectId == ~0) + done = 1; +- else if(ok){ +- obj = yaffs_FindOrCreateObjectByNumber(dev,cp.objectId, cp.variantType); +- if(obj) { +- yaffs_CheckpointObjectToObject(obj,&cp); +- if(obj->variantType == YAFFS_OBJECT_TYPE_FILE) { ++ else if (ok) { ++ obj = yaffs_FindOrCreateObjectByNumber(dev, cp.objectId, cp.variantType); ++ if (obj) { ++ ok = yaffs_CheckpointObjectToObject(obj, &cp); ++ if (!ok) ++ break; ++ if (obj->variantType == YAFFS_OBJECT_TYPE_FILE) { + ok = yaffs_ReadCheckpointTnodes(obj); +- } else if(obj->variantType == YAFFS_OBJECT_TYPE_HARDLINK) { ++ } else if (obj->variantType == YAFFS_OBJECT_TYPE_HARDLINK) { + obj->hardLinks.next = +- (struct list_head *) +- hardList; ++ (struct ylist_head *) hardList; + hardList = obj; + } +- +- } ++ } else ++ ok = 0; + } + } + +- if(ok) +- yaffs_HardlinkFixup(dev,hardList); ++ if (ok) ++ yaffs_HardlinkFixup(dev, hardList); + + return ok ? 1 : 0; + } +@@ -4403,11 +4471,11 @@ static int yaffs_WriteCheckpointSum(yaff + __u32 checkpointSum; + int ok; + +- yaffs_GetCheckpointSum(dev,&checkpointSum); ++ yaffs_GetCheckpointSum(dev, &checkpointSum); + +- ok = (yaffs_CheckpointWrite(dev,&checkpointSum,sizeof(checkpointSum)) == sizeof(checkpointSum)); ++ ok = (yaffs_CheckpointWrite(dev, &checkpointSum, sizeof(checkpointSum)) == sizeof(checkpointSum)); + +- if(!ok) ++ if (!ok) + return 0; + + return 1; +@@ -4419,14 +4487,14 @@ static int yaffs_ReadCheckpointSum(yaffs + __u32 checkpointSum1; + int ok; + +- yaffs_GetCheckpointSum(dev,&checkpointSum0); ++ yaffs_GetCheckpointSum(dev, &checkpointSum0); + +- ok = (yaffs_CheckpointRead(dev,&checkpointSum1,sizeof(checkpointSum1)) == sizeof(checkpointSum1)); ++ ok = (yaffs_CheckpointRead(dev, &checkpointSum1, sizeof(checkpointSum1)) == sizeof(checkpointSum1)); + +- if(!ok) ++ if (!ok) + return 0; + +- if(checkpointSum0 != checkpointSum1) ++ if (checkpointSum0 != checkpointSum1) + return 0; + + return 1; +@@ -4435,46 +4503,43 @@ static int yaffs_ReadCheckpointSum(yaffs + + static int yaffs_WriteCheckpointData(yaffs_Device *dev) + { +- + int ok = 1; + +- if(dev->skipCheckpointWrite || !dev->isYaffs2){ +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("skipping checkpoint write" TENDSTR))); ++ if (dev->skipCheckpointWrite || !dev->isYaffs2) { ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("skipping checkpoint write" TENDSTR))); + ok = 0; + } + +- if(ok) +- ok = yaffs_CheckpointOpen(dev,1); ++ if (ok) ++ ok = yaffs_CheckpointOpen(dev, 1); + +- if(ok){ +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("write checkpoint validity" TENDSTR))); +- ok = yaffs_WriteCheckpointValidityMarker(dev,1); ++ if (ok) { ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("write checkpoint validity" TENDSTR))); ++ ok = yaffs_WriteCheckpointValidityMarker(dev, 1); + } +- if(ok){ +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("write checkpoint device" TENDSTR))); ++ if (ok) { ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("write checkpoint device" TENDSTR))); + ok = yaffs_WriteCheckpointDevice(dev); + } +- if(ok){ +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("write checkpoint objects" TENDSTR))); ++ if (ok) { ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("write checkpoint objects" TENDSTR))); + ok = yaffs_WriteCheckpointObjects(dev); + } +- if(ok){ +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("write checkpoint validity" TENDSTR))); +- ok = yaffs_WriteCheckpointValidityMarker(dev,0); ++ if (ok) { ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("write checkpoint validity" TENDSTR))); ++ ok = yaffs_WriteCheckpointValidityMarker(dev, 0); + } + +- if(ok){ ++ if (ok) + ok = yaffs_WriteCheckpointSum(dev); +- } +- + +- if(!yaffs_CheckpointClose(dev)) +- ok = 0; ++ if (!yaffs_CheckpointClose(dev)) ++ ok = 0; + +- if(ok) +- dev->isCheckpointed = 1; +- else +- dev->isCheckpointed = 0; ++ if (ok) ++ dev->isCheckpointed = 1; ++ else ++ dev->isCheckpointed = 0; + + return dev->isCheckpointed; + } +@@ -4483,43 +4548,43 @@ static int yaffs_ReadCheckpointData(yaff + { + int ok = 1; + +- if(dev->skipCheckpointRead || !dev->isYaffs2){ +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("skipping checkpoint read" TENDSTR))); ++ if (dev->skipCheckpointRead || !dev->isYaffs2) { ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("skipping checkpoint read" TENDSTR))); + ok = 0; + } + +- if(ok) +- ok = yaffs_CheckpointOpen(dev,0); /* open for read */ ++ if (ok) ++ ok = yaffs_CheckpointOpen(dev, 0); /* open for read */ + +- if(ok){ +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("read checkpoint validity" TENDSTR))); +- ok = yaffs_ReadCheckpointValidityMarker(dev,1); ++ if (ok) { ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("read checkpoint validity" TENDSTR))); ++ ok = yaffs_ReadCheckpointValidityMarker(dev, 1); + } +- if(ok){ +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("read checkpoint device" TENDSTR))); ++ if (ok) { ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("read checkpoint device" TENDSTR))); + ok = yaffs_ReadCheckpointDevice(dev); + } +- if(ok){ +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("read checkpoint objects" TENDSTR))); ++ if (ok) { ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("read checkpoint objects" TENDSTR))); + ok = yaffs_ReadCheckpointObjects(dev); + } +- if(ok){ +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("read checkpoint validity" TENDSTR))); +- ok = yaffs_ReadCheckpointValidityMarker(dev,0); ++ if (ok) { ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("read checkpoint validity" TENDSTR))); ++ ok = yaffs_ReadCheckpointValidityMarker(dev, 0); + } + +- if(ok){ ++ if (ok) { + ok = yaffs_ReadCheckpointSum(dev); +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("read checkpoint checksum %d" TENDSTR),ok)); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("read checkpoint checksum %d" TENDSTR), ok)); + } + +- if(!yaffs_CheckpointClose(dev)) ++ if (!yaffs_CheckpointClose(dev)) + ok = 0; + +- if(ok) +- dev->isCheckpointed = 1; +- else +- dev->isCheckpointed = 0; ++ if (ok) ++ dev->isCheckpointed = 1; ++ else ++ dev->isCheckpointed = 0; + + return ok ? 1 : 0; + +@@ -4527,11 +4592,11 @@ static int yaffs_ReadCheckpointData(yaff + + static void yaffs_InvalidateCheckpoint(yaffs_Device *dev) + { +- if(dev->isCheckpointed || +- dev->blocksInCheckpoint > 0){ ++ if (dev->isCheckpointed || ++ dev->blocksInCheckpoint > 0) { + dev->isCheckpointed = 0; + yaffs_CheckpointInvalidateStream(dev); +- if(dev->superBlock && dev->markSuperBlockDirty) ++ if (dev->superBlock && dev->markSuperBlockDirty) + dev->markSuperBlockDirty(dev->superBlock); + } + } +@@ -4540,18 +4605,18 @@ static void yaffs_InvalidateCheckpoint(y + int yaffs_CheckpointSave(yaffs_Device *dev) + { + +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("save entry: isCheckpointed %d"TENDSTR),dev->isCheckpointed)); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("save entry: isCheckpointed %d"TENDSTR), dev->isCheckpointed)); + + yaffs_VerifyObjects(dev); + yaffs_VerifyBlocks(dev); + yaffs_VerifyFreeChunks(dev); + +- if(!dev->isCheckpointed) { ++ if (!dev->isCheckpointed) { + yaffs_InvalidateCheckpoint(dev); + yaffs_WriteCheckpointData(dev); + } + +- T(YAFFS_TRACE_ALWAYS,(TSTR("save exit: isCheckpointed %d"TENDSTR),dev->isCheckpointed)); ++ T(YAFFS_TRACE_ALWAYS, (TSTR("save exit: isCheckpointed %d"TENDSTR), dev->isCheckpointed)); + + return dev->isCheckpointed; + } +@@ -4559,17 +4624,17 @@ int yaffs_CheckpointSave(yaffs_Device *d + int yaffs_CheckpointRestore(yaffs_Device *dev) + { + int retval; +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("restore entry: isCheckpointed %d"TENDSTR),dev->isCheckpointed)); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("restore entry: isCheckpointed %d"TENDSTR), dev->isCheckpointed)); + + retval = yaffs_ReadCheckpointData(dev); + +- if(dev->isCheckpointed){ ++ if (dev->isCheckpointed) { + yaffs_VerifyObjects(dev); + yaffs_VerifyBlocks(dev); + yaffs_VerifyFreeChunks(dev); + } + +- T(YAFFS_TRACE_CHECKPOINT,(TSTR("restore exit: isCheckpointed %d"TENDSTR),dev->isCheckpointed)); ++ T(YAFFS_TRACE_CHECKPOINT, (TSTR("restore exit: isCheckpointed %d"TENDSTR), dev->isCheckpointed)); + + return retval; + } +@@ -4584,12 +4649,12 @@ int yaffs_CheckpointRestore(yaffs_Device + * Curve-balls: the first chunk might also be the last chunk. + */ + +-int yaffs_ReadDataFromFile(yaffs_Object * in, __u8 * buffer, loff_t offset, +- int nBytes) ++int yaffs_ReadDataFromFile(yaffs_Object *in, __u8 *buffer, loff_t offset, ++ int nBytes) + { + + int chunk; +- int start; ++ __u32 start; + int nToCopy; + int n = nBytes; + int nDone = 0; +@@ -4600,27 +4665,26 @@ int yaffs_ReadDataFromFile(yaffs_Object + dev = in->myDev; + + while (n > 0) { +- //chunk = offset / dev->nDataBytesPerChunk + 1; +- //start = offset % dev->nDataBytesPerChunk; +- yaffs_AddrToChunk(dev,offset,&chunk,&start); ++ /* chunk = offset / dev->nDataBytesPerChunk + 1; */ ++ /* start = offset % dev->nDataBytesPerChunk; */ ++ yaffs_AddrToChunk(dev, offset, &chunk, &start); + chunk++; + + /* OK now check for the curveball where the start and end are in + * the same chunk. + */ +- if ((start + n) < dev->nDataBytesPerChunk) { ++ if ((start + n) < dev->nDataBytesPerChunk) + nToCopy = n; +- } else { ++ else + nToCopy = dev->nDataBytesPerChunk - start; +- } + + cache = yaffs_FindChunkCache(in, chunk); + + /* If the chunk is already in the cache or it is less than a whole chunk +- * then use the cache (if there is caching) ++ * or we're using inband tags then use the cache (if there is caching) + * else bypass the cache. + */ +- if (cache || nToCopy != dev->nDataBytesPerChunk) { ++ if (cache || nToCopy != dev->nDataBytesPerChunk || dev->inbandTags) { + if (dev->nShortOpCaches > 0) { + + /* If we can't find the data in the cache, then load it up. */ +@@ -4641,14 +4705,9 @@ int yaffs_ReadDataFromFile(yaffs_Object + + cache->locked = 1; + +-#ifdef CONFIG_YAFFS_WINCE +- yfsd_UnlockYAFFS(TRUE); +-#endif ++ + memcpy(buffer, &cache->data[start], nToCopy); + +-#ifdef CONFIG_YAFFS_WINCE +- yfsd_LockYAFFS(TRUE); +-#endif + cache->locked = 0; + } else { + /* Read into the local buffer then copy..*/ +@@ -4657,41 +4716,19 @@ int yaffs_ReadDataFromFile(yaffs_Object + yaffs_GetTempBuffer(dev, __LINE__); + yaffs_ReadChunkDataFromObject(in, chunk, + localBuffer); +-#ifdef CONFIG_YAFFS_WINCE +- yfsd_UnlockYAFFS(TRUE); +-#endif ++ + memcpy(buffer, &localBuffer[start], nToCopy); + +-#ifdef CONFIG_YAFFS_WINCE +- yfsd_LockYAFFS(TRUE); +-#endif ++ + yaffs_ReleaseTempBuffer(dev, localBuffer, + __LINE__); + } + + } else { +-#ifdef CONFIG_YAFFS_WINCE +- __u8 *localBuffer = yaffs_GetTempBuffer(dev, __LINE__); +- +- /* Under WinCE can't do direct transfer. Need to use a local buffer. +- * This is because we otherwise screw up WinCE's memory mapper +- */ +- yaffs_ReadChunkDataFromObject(in, chunk, localBuffer); +- +-#ifdef CONFIG_YAFFS_WINCE +- yfsd_UnlockYAFFS(TRUE); +-#endif +- memcpy(buffer, localBuffer, dev->nDataBytesPerChunk); + +-#ifdef CONFIG_YAFFS_WINCE +- yfsd_LockYAFFS(TRUE); +- yaffs_ReleaseTempBuffer(dev, localBuffer, __LINE__); +-#endif +- +-#else + /* A full chunk. Read directly into the supplied buffer. */ + yaffs_ReadChunkDataFromObject(in, chunk, buffer); +-#endif ++ + } + + n -= nToCopy; +@@ -4704,28 +4741,37 @@ int yaffs_ReadDataFromFile(yaffs_Object + return nDone; + } + +-int yaffs_WriteDataToFile(yaffs_Object * in, const __u8 * buffer, loff_t offset, +- int nBytes, int writeThrough) ++int yaffs_WriteDataToFile(yaffs_Object *in, const __u8 *buffer, loff_t offset, ++ int nBytes, int writeThrough) + { + + int chunk; +- int start; ++ __u32 start; + int nToCopy; + int n = nBytes; + int nDone = 0; + int nToWriteBack; + int startOfWrite = offset; + int chunkWritten = 0; +- int nBytesRead; ++ __u32 nBytesRead; ++ __u32 chunkStart; + + yaffs_Device *dev; + + dev = in->myDev; + + while (n > 0 && chunkWritten >= 0) { +- //chunk = offset / dev->nDataBytesPerChunk + 1; +- //start = offset % dev->nDataBytesPerChunk; +- yaffs_AddrToChunk(dev,offset,&chunk,&start); ++ /* chunk = offset / dev->nDataBytesPerChunk + 1; */ ++ /* start = offset % dev->nDataBytesPerChunk; */ ++ yaffs_AddrToChunk(dev, offset, &chunk, &start); ++ ++ if (chunk * dev->nDataBytesPerChunk + start != offset || ++ start >= dev->nDataBytesPerChunk) { ++ T(YAFFS_TRACE_ERROR, ( ++ TSTR("AddrToChunk of offset %d gives chunk %d start %d" ++ TENDSTR), ++ (int)offset, chunk, start)); ++ } + chunk++; + + /* OK now check for the curveball where the start and end are in +@@ -4740,25 +4786,32 @@ int yaffs_WriteDataToFile(yaffs_Object * + * we need to write back as much as was there before. + */ + +- nBytesRead = +- in->variant.fileVariant.fileSize - +- ((chunk - 1) * dev->nDataBytesPerChunk); ++ chunkStart = ((chunk - 1) * dev->nDataBytesPerChunk); ++ ++ if (chunkStart > in->variant.fileVariant.fileSize) ++ nBytesRead = 0; /* Past end of file */ ++ else ++ nBytesRead = in->variant.fileVariant.fileSize - chunkStart; + +- if (nBytesRead > dev->nDataBytesPerChunk) { ++ if (nBytesRead > dev->nDataBytesPerChunk) + nBytesRead = dev->nDataBytesPerChunk; +- } + + nToWriteBack = + (nBytesRead > + (start + n)) ? nBytesRead : (start + n); + ++ if (nToWriteBack < 0 || nToWriteBack > dev->nDataBytesPerChunk) ++ YBUG(); ++ + } else { + nToCopy = dev->nDataBytesPerChunk - start; + nToWriteBack = dev->nDataBytesPerChunk; + } + +- if (nToCopy != dev->nDataBytesPerChunk) { +- /* An incomplete start or end chunk (or maybe both start and end chunk) */ ++ if (nToCopy != dev->nDataBytesPerChunk || dev->inbandTags) { ++ /* An incomplete start or end chunk (or maybe both start and end chunk), ++ * or we're using inband tags, so we want to use the cache buffers. ++ */ + if (dev->nShortOpCaches > 0) { + yaffs_ChunkCache *cache; + /* If we can't find the data in the cache, then load the cache */ +@@ -4775,10 +4828,9 @@ int yaffs_WriteDataToFile(yaffs_Object * + yaffs_ReadChunkDataFromObject(in, chunk, + cache-> + data); +- } +- else if(cache && +- !cache->dirty && +- !yaffs_CheckSpaceForAllocation(in->myDev)){ ++ } else if (cache && ++ !cache->dirty && ++ !yaffs_CheckSpaceForAllocation(in->myDev)) { + /* Drop the cache if it was a read cache item and + * no space check has been made for it. + */ +@@ -4788,16 +4840,12 @@ int yaffs_WriteDataToFile(yaffs_Object * + if (cache) { + yaffs_UseChunkCache(dev, cache, 1); + cache->locked = 1; +-#ifdef CONFIG_YAFFS_WINCE +- yfsd_UnlockYAFFS(TRUE); +-#endif ++ + + memcpy(&cache->data[start], buffer, + nToCopy); + +-#ifdef CONFIG_YAFFS_WINCE +- yfsd_LockYAFFS(TRUE); +-#endif ++ + cache->locked = 0; + cache->nBytes = nToWriteBack; + +@@ -4825,15 +4873,10 @@ int yaffs_WriteDataToFile(yaffs_Object * + yaffs_ReadChunkDataFromObject(in, chunk, + localBuffer); + +-#ifdef CONFIG_YAFFS_WINCE +- yfsd_UnlockYAFFS(TRUE); +-#endif ++ + + memcpy(&localBuffer[start], buffer, nToCopy); + +-#ifdef CONFIG_YAFFS_WINCE +- yfsd_LockYAFFS(TRUE); +-#endif + chunkWritten = + yaffs_WriteChunkDataToObject(in, chunk, + localBuffer, +@@ -4846,31 +4889,15 @@ int yaffs_WriteDataToFile(yaffs_Object * + } + + } else { +- +-#ifdef CONFIG_YAFFS_WINCE +- /* Under WinCE can't do direct transfer. Need to use a local buffer. +- * This is because we otherwise screw up WinCE's memory mapper +- */ +- __u8 *localBuffer = yaffs_GetTempBuffer(dev, __LINE__); +-#ifdef CONFIG_YAFFS_WINCE +- yfsd_UnlockYAFFS(TRUE); +-#endif +- memcpy(localBuffer, buffer, dev->nDataBytesPerChunk); +-#ifdef CONFIG_YAFFS_WINCE +- yfsd_LockYAFFS(TRUE); +-#endif +- chunkWritten = +- yaffs_WriteChunkDataToObject(in, chunk, localBuffer, +- dev->nDataBytesPerChunk, +- 0); +- yaffs_ReleaseTempBuffer(dev, localBuffer, __LINE__); +-#else + /* A full chunk. Write directly from the supplied buffer. */ ++ ++ ++ + chunkWritten = + yaffs_WriteChunkDataToObject(in, chunk, buffer, + dev->nDataBytesPerChunk, + 0); +-#endif ++ + /* Since we've overwritten the cached data, we better invalidate it. */ + yaffs_InvalidateChunkCache(in, chunk); + } +@@ -4886,9 +4913,8 @@ int yaffs_WriteDataToFile(yaffs_Object * + + /* Update file object */ + +- if ((startOfWrite + nDone) > in->variant.fileVariant.fileSize) { ++ if ((startOfWrite + nDone) > in->variant.fileVariant.fileSize) + in->variant.fileVariant.fileSize = (startOfWrite + nDone); +- } + + in->dirty = 1; + +@@ -4898,7 +4924,7 @@ int yaffs_WriteDataToFile(yaffs_Object * + + /* ---------------------- File resizing stuff ------------------ */ + +-static void yaffs_PruneResizedChunks(yaffs_Object * in, int newSize) ++static void yaffs_PruneResizedChunks(yaffs_Object *in, int newSize) + { + + yaffs_Device *dev = in->myDev; +@@ -4939,11 +4965,11 @@ static void yaffs_PruneResizedChunks(yaf + + } + +-int yaffs_ResizeFile(yaffs_Object * in, loff_t newSize) ++int yaffs_ResizeFile(yaffs_Object *in, loff_t newSize) + { + + int oldFileSize = in->variant.fileVariant.fileSize; +- int newSizeOfPartialChunk; ++ __u32 newSizeOfPartialChunk; + int newFullChunks; + + yaffs_Device *dev = in->myDev; +@@ -4955,13 +4981,11 @@ int yaffs_ResizeFile(yaffs_Object * in, + + yaffs_CheckGarbageCollection(dev); + +- if (in->variantType != YAFFS_OBJECT_TYPE_FILE) { +- return yaffs_GetFileSize(in); +- } ++ if (in->variantType != YAFFS_OBJECT_TYPE_FILE) ++ return YAFFS_FAIL; + +- if (newSize == oldFileSize) { +- return oldFileSize; +- } ++ if (newSize == oldFileSize) ++ return YAFFS_OK; + + if (newSize < oldFileSize) { + +@@ -4994,21 +5018,20 @@ int yaffs_ResizeFile(yaffs_Object * in, + } + + +- + /* Write a new object header. + * show we've shrunk the file, if need be + * Do this only if the file is not in the deleted directories. + */ +- if (in->parent->objectId != YAFFS_OBJECTID_UNLINKED && +- in->parent->objectId != YAFFS_OBJECTID_DELETED) { ++ if (in->parent && ++ in->parent->objectId != YAFFS_OBJECTID_UNLINKED && ++ in->parent->objectId != YAFFS_OBJECTID_DELETED) + yaffs_UpdateObjectHeader(in, NULL, 0, + (newSize < oldFileSize) ? 1 : 0, 0); +- } + +- return newSize; ++ return YAFFS_OK; + } + +-loff_t yaffs_GetFileSize(yaffs_Object * obj) ++loff_t yaffs_GetFileSize(yaffs_Object *obj) + { + obj = yaffs_GetEquivalentObject(obj); + +@@ -5024,7 +5047,7 @@ loff_t yaffs_GetFileSize(yaffs_Object * + + + +-int yaffs_FlushFile(yaffs_Object * in, int updateTime) ++int yaffs_FlushFile(yaffs_Object *in, int updateTime) + { + int retVal; + if (in->dirty) { +@@ -5039,9 +5062,8 @@ int yaffs_FlushFile(yaffs_Object * in, i + #endif + } + +- retVal = +- (yaffs_UpdateObjectHeader(in, NULL, 0, 0, 0) >= +- 0) ? YAFFS_OK : YAFFS_FAIL; ++ retVal = (yaffs_UpdateObjectHeader(in, NULL, 0, 0, 0) >= ++ 0) ? YAFFS_OK : YAFFS_FAIL; + } else { + retVal = YAFFS_OK; + } +@@ -5050,7 +5072,7 @@ int yaffs_FlushFile(yaffs_Object * in, i + + } + +-static int yaffs_DoGenericObjectDeletion(yaffs_Object * in) ++static int yaffs_DoGenericObjectDeletion(yaffs_Object *in) + { + + /* First off, invalidate the file's data in the cache, without flushing. */ +@@ -5058,13 +5080,13 @@ static int yaffs_DoGenericObjectDeletion + + if (in->myDev->isYaffs2 && (in->parent != in->myDev->deletedDir)) { + /* Move to the unlinked directory so we have a record that it was deleted. */ +- yaffs_ChangeObjectName(in, in->myDev->deletedDir,"deleted", 0, 0); ++ yaffs_ChangeObjectName(in, in->myDev->deletedDir, _Y("deleted"), 0, 0); + + } + + yaffs_RemoveObjectFromDirectory(in); +- yaffs_DeleteChunk(in->myDev, in->chunkId, 1, __LINE__); +- in->chunkId = -1; ++ yaffs_DeleteChunk(in->myDev, in->hdrChunk, 1, __LINE__); ++ in->hdrChunk = 0; + + yaffs_FreeObject(in); + return YAFFS_OK; +@@ -5075,62 +5097,63 @@ static int yaffs_DoGenericObjectDeletion + * and the inode associated with the file. + * It does not delete the links associated with the file. + */ +-static int yaffs_UnlinkFile(yaffs_Object * in) ++static int yaffs_UnlinkFileIfNeeded(yaffs_Object *in) + { + + int retVal; + int immediateDeletion = 0; + +- if (1) { + #ifdef __KERNEL__ +- if (!in->myInode) { +- immediateDeletion = 1; +- +- } ++ if (!in->myInode) ++ immediateDeletion = 1; + #else +- if (in->inUse <= 0) { +- immediateDeletion = 1; +- +- } ++ if (in->inUse <= 0) ++ immediateDeletion = 1; + #endif +- if (immediateDeletion) { +- retVal = +- yaffs_ChangeObjectName(in, in->myDev->deletedDir, +- "deleted", 0, 0); +- T(YAFFS_TRACE_TRACING, +- (TSTR("yaffs: immediate deletion of file %d" TENDSTR), +- in->objectId)); +- in->deleted = 1; +- in->myDev->nDeletedFiles++; +- if (0 && in->myDev->isYaffs2) { +- yaffs_ResizeFile(in, 0); +- } +- yaffs_SoftDeleteFile(in); +- } else { +- retVal = +- yaffs_ChangeObjectName(in, in->myDev->unlinkedDir, +- "unlinked", 0, 0); +- } + ++ if (immediateDeletion) { ++ retVal = ++ yaffs_ChangeObjectName(in, in->myDev->deletedDir, ++ _Y("deleted"), 0, 0); ++ T(YAFFS_TRACE_TRACING, ++ (TSTR("yaffs: immediate deletion of file %d" TENDSTR), ++ in->objectId)); ++ in->deleted = 1; ++ in->myDev->nDeletedFiles++; ++ if (1 || in->myDev->isYaffs2) ++ yaffs_ResizeFile(in, 0); ++ yaffs_SoftDeleteFile(in); ++ } else { ++ retVal = ++ yaffs_ChangeObjectName(in, in->myDev->unlinkedDir, ++ _Y("unlinked"), 0, 0); + } ++ ++ + return retVal; + } + +-int yaffs_DeleteFile(yaffs_Object * in) ++int yaffs_DeleteFile(yaffs_Object *in) + { + int retVal = YAFFS_OK; ++ int deleted = in->deleted; ++ ++ yaffs_ResizeFile(in, 0); + + if (in->nDataChunks > 0) { +- /* Use soft deletion if there is data in the file */ +- if (!in->unlinked) { +- retVal = yaffs_UnlinkFile(in); +- } ++ /* Use soft deletion if there is data in the file. ++ * That won't be the case if it has been resized to zero. ++ */ ++ if (!in->unlinked) ++ retVal = yaffs_UnlinkFileIfNeeded(in); ++ + if (retVal == YAFFS_OK && in->unlinked && !in->deleted) { + in->deleted = 1; ++ deleted = 1; + in->myDev->nDeletedFiles++; + yaffs_SoftDeleteFile(in); + } +- return in->deleted ? YAFFS_OK : YAFFS_FAIL; ++ return deleted ? YAFFS_OK : YAFFS_FAIL; + } else { + /* The file has no data chunks so we toss it immediately */ + yaffs_FreeTnode(in->myDev, in->variant.fileVariant.top); +@@ -5141,62 +5164,75 @@ int yaffs_DeleteFile(yaffs_Object * in) + } + } + +-static int yaffs_DeleteDirectory(yaffs_Object * in) ++static int yaffs_DeleteDirectory(yaffs_Object *in) + { + /* First check that the directory is empty. */ +- if (list_empty(&in->variant.directoryVariant.children)) { ++ if (ylist_empty(&in->variant.directoryVariant.children)) + return yaffs_DoGenericObjectDeletion(in); +- } + + return YAFFS_FAIL; + + } + +-static int yaffs_DeleteSymLink(yaffs_Object * in) ++static int yaffs_DeleteSymLink(yaffs_Object *in) + { + YFREE(in->variant.symLinkVariant.alias); + + return yaffs_DoGenericObjectDeletion(in); + } + +-static int yaffs_DeleteHardLink(yaffs_Object * in) ++static int yaffs_DeleteHardLink(yaffs_Object *in) + { + /* remove this hardlink from the list assocaited with the equivalent + * object + */ +- list_del(&in->hardLinks); ++ ylist_del_init(&in->hardLinks); + return yaffs_DoGenericObjectDeletion(in); + } + +-static void yaffs_DestroyObject(yaffs_Object * obj) ++int yaffs_DeleteObject(yaffs_Object *obj) + { ++int retVal = -1; + switch (obj->variantType) { + case YAFFS_OBJECT_TYPE_FILE: +- yaffs_DeleteFile(obj); ++ retVal = yaffs_DeleteFile(obj); + break; + case YAFFS_OBJECT_TYPE_DIRECTORY: +- yaffs_DeleteDirectory(obj); ++ return yaffs_DeleteDirectory(obj); + break; + case YAFFS_OBJECT_TYPE_SYMLINK: +- yaffs_DeleteSymLink(obj); ++ retVal = yaffs_DeleteSymLink(obj); + break; + case YAFFS_OBJECT_TYPE_HARDLINK: +- yaffs_DeleteHardLink(obj); ++ retVal = yaffs_DeleteHardLink(obj); + break; + case YAFFS_OBJECT_TYPE_SPECIAL: +- yaffs_DoGenericObjectDeletion(obj); ++ retVal = yaffs_DoGenericObjectDeletion(obj); + break; + case YAFFS_OBJECT_TYPE_UNKNOWN: ++ retVal = 0; + break; /* should not happen. */ + } ++ ++ return retVal; + } + +-static int yaffs_UnlinkWorker(yaffs_Object * obj) ++static int yaffs_UnlinkWorker(yaffs_Object *obj) + { + ++ int immediateDeletion = 0; ++ ++#ifdef __KERNEL__ ++ if (!obj->myInode) ++ immediateDeletion = 1; ++#else ++ if (obj->inUse <= 0) ++ immediateDeletion = 1; ++#endif ++ + if (obj->variantType == YAFFS_OBJECT_TYPE_HARDLINK) { + return yaffs_DeleteHardLink(obj); +- } else if (!list_empty(&obj->hardLinks)) { ++ } else if (!ylist_empty(&obj->hardLinks)) { + /* Curve ball: We're unlinking an object that has a hardlink. + * + * This problem arises because we are not strictly following +@@ -5215,24 +5251,24 @@ static int yaffs_UnlinkWorker(yaffs_Obje + int retVal; + YCHAR name[YAFFS_MAX_NAME_LENGTH + 1]; + +- hl = list_entry(obj->hardLinks.next, yaffs_Object, hardLinks); ++ hl = ylist_entry(obj->hardLinks.next, yaffs_Object, hardLinks); + +- list_del_init(&hl->hardLinks); +- list_del_init(&hl->siblings); ++ ylist_del_init(&hl->hardLinks); ++ ylist_del_init(&hl->siblings); + + yaffs_GetObjectName(hl, name, YAFFS_MAX_NAME_LENGTH + 1); + + retVal = yaffs_ChangeObjectName(obj, hl->parent, name, 0, 0); + +- if (retVal == YAFFS_OK) { ++ if (retVal == YAFFS_OK) + retVal = yaffs_DoGenericObjectDeletion(hl); +- } ++ + return retVal; + +- } else { ++ } else if (immediateDeletion) { + switch (obj->variantType) { + case YAFFS_OBJECT_TYPE_FILE: +- return yaffs_UnlinkFile(obj); ++ return yaffs_DeleteFile(obj); + break; + case YAFFS_OBJECT_TYPE_DIRECTORY: + return yaffs_DeleteDirectory(obj); +@@ -5248,21 +5284,22 @@ static int yaffs_UnlinkWorker(yaffs_Obje + default: + return YAFFS_FAIL; + } +- } ++ } else ++ return yaffs_ChangeObjectName(obj, obj->myDev->unlinkedDir, ++ _Y("unlinked"), 0, 0); + } + + +-static int yaffs_UnlinkObject( yaffs_Object *obj) ++static int yaffs_UnlinkObject(yaffs_Object *obj) + { + +- if (obj && obj->unlinkAllowed) { ++ if (obj && obj->unlinkAllowed) + return yaffs_UnlinkWorker(obj); +- } + + return YAFFS_FAIL; + + } +-int yaffs_Unlink(yaffs_Object * dir, const YCHAR * name) ++int yaffs_Unlink(yaffs_Object *dir, const YCHAR *name) + { + yaffs_Object *obj; + +@@ -5272,8 +5309,8 @@ int yaffs_Unlink(yaffs_Object * dir, con + + /*----------------------- Initialisation Scanning ---------------------- */ + +-static void yaffs_HandleShadowedObject(yaffs_Device * dev, int objId, +- int backwardScanning) ++static void yaffs_HandleShadowedObject(yaffs_Device *dev, int objId, ++ int backwardScanning) + { + yaffs_Object *obj; + +@@ -5286,9 +5323,8 @@ static void yaffs_HandleShadowedObject(y + /* Handle YAFFS2 case (backward scanning) + * If the shadowed object exists then ignore. + */ +- if (yaffs_FindObjectByNumber(dev, objId)) { ++ if (yaffs_FindObjectByNumber(dev, objId)) + return; +- } + } + + /* Let's create it (if it does not exist) assuming it is a file so that it can do shrinking etc. +@@ -5297,6 +5333,8 @@ static void yaffs_HandleShadowedObject(y + obj = + yaffs_FindOrCreateObjectByNumber(dev, objId, + YAFFS_OBJECT_TYPE_FILE); ++ if (!obj) ++ return; + yaffs_AddObjectToDirectory(dev->unlinkedDir, obj); + obj->variant.fileVariant.shrinkSize = 0; + obj->valid = 1; /* So that we don't read any other info for this file */ +@@ -5325,44 +5363,77 @@ static void yaffs_HardlinkFixup(yaffs_De + if (in) { + /* Add the hardlink pointers */ + hl->variant.hardLinkVariant.equivalentObject = in; +- list_add(&hl->hardLinks, &in->hardLinks); ++ ylist_add(&hl->hardLinks, &in->hardLinks); + } else { + /* Todo Need to report/handle this better. + * Got a problem... hardlink to a non-existant object + */ + hl->variant.hardLinkVariant.equivalentObject = NULL; +- INIT_LIST_HEAD(&hl->hardLinks); ++ YINIT_LIST_HEAD(&hl->hardLinks); + + } +- + } ++} ++ ++ + ++ ++ ++static int ybicmp(const void *a, const void *b) ++{ ++ register int aseq = ((yaffs_BlockIndex *)a)->seq; ++ register int bseq = ((yaffs_BlockIndex *)b)->seq; ++ register int ablock = ((yaffs_BlockIndex *)a)->block; ++ register int bblock = ((yaffs_BlockIndex *)b)->block; ++ if (aseq == bseq) ++ return ablock - bblock; ++ else ++ return aseq - bseq; + } + + ++struct yaffs_ShadowFixerStruct { ++ int objectId; ++ int shadowedId; ++ struct yaffs_ShadowFixerStruct *next; ++}; ++ + ++static void yaffs_StripDeletedObjects(yaffs_Device *dev) ++{ ++ /* ++ * Sort out state of unlinked and deleted objects after scanning. ++ */ ++ struct ylist_head *i; ++ struct ylist_head *n; ++ yaffs_Object *l; + ++ /* Soft delete all the unlinked files */ ++ ylist_for_each_safe(i, n, ++ &dev->unlinkedDir->variant.directoryVariant.children) { ++ if (i) { ++ l = ylist_entry(i, yaffs_Object, siblings); ++ yaffs_DeleteObject(l); ++ } ++ } + +-static int ybicmp(const void *a, const void *b){ +- register int aseq = ((yaffs_BlockIndex *)a)->seq; +- register int bseq = ((yaffs_BlockIndex *)b)->seq; +- register int ablock = ((yaffs_BlockIndex *)a)->block; +- register int bblock = ((yaffs_BlockIndex *)b)->block; +- if( aseq == bseq ) +- return ablock - bblock; +- else +- return aseq - bseq; ++ ylist_for_each_safe(i, n, ++ &dev->deletedDir->variant.directoryVariant.children) { ++ if (i) { ++ l = ylist_entry(i, yaffs_Object, siblings); ++ yaffs_DeleteObject(l); ++ } ++ } + + } + +-static int yaffs_Scan(yaffs_Device * dev) ++static int yaffs_Scan(yaffs_Device *dev) + { + yaffs_ExtendedTags tags; + int blk; + int blockIterator; + int startIterator; + int endIterator; +- int nBlocksToScan = 0; + int result; + + int chunk; +@@ -5371,26 +5442,19 @@ static int yaffs_Scan(yaffs_Device * dev + yaffs_BlockState state; + yaffs_Object *hardList = NULL; + yaffs_BlockInfo *bi; +- int sequenceNumber; ++ __u32 sequenceNumber; + yaffs_ObjectHeader *oh; + yaffs_Object *in; + yaffs_Object *parent; +- int nBlocks = dev->internalEndBlock - dev->internalStartBlock + 1; + + int alloc_failed = 0; + ++ struct yaffs_ShadowFixerStruct *shadowFixerList = NULL; ++ + + __u8 *chunkData; + +- yaffs_BlockIndex *blockIndex = NULL; + +- if (dev->isYaffs2) { +- T(YAFFS_TRACE_SCAN, +- (TSTR("yaffs_Scan is not for YAFFS2!" TENDSTR))); +- return YAFFS_FAIL; +- } +- +- //TODO Throw all the yaffs2 stuuf out of yaffs_Scan since it is only for yaffs1 format. + + T(YAFFS_TRACE_SCAN, + (TSTR("yaffs_Scan starts intstartblk %d intendblk %d..." TENDSTR), +@@ -5400,12 +5464,6 @@ static int yaffs_Scan(yaffs_Device * dev + + dev->sequenceNumber = YAFFS_LOWEST_SEQUENCE_NUMBER; + +- if (dev->isYaffs2) { +- blockIndex = YMALLOC(nBlocks * sizeof(yaffs_BlockIndex)); +- if(!blockIndex) +- return YAFFS_FAIL; +- } +- + /* Scan all the blocks to determine their state */ + for (blk = dev->internalStartBlock; blk <= dev->internalEndBlock; blk++) { + bi = yaffs_GetBlockInfo(dev, blk); +@@ -5418,6 +5476,9 @@ static int yaffs_Scan(yaffs_Device * dev + bi->blockState = state; + bi->sequenceNumber = sequenceNumber; + ++ if (bi->sequenceNumber == YAFFS_SEQUENCE_BAD_BLOCK) ++ bi->blockState = state = YAFFS_BLOCK_STATE_DEAD; ++ + T(YAFFS_TRACE_SCAN_DEBUG, + (TSTR("Block scanning block %d state %d seq %d" TENDSTR), blk, + state, sequenceNumber)); +@@ -5430,70 +5491,21 @@ static int yaffs_Scan(yaffs_Device * dev + (TSTR("Block empty " TENDSTR))); + dev->nErasedBlocks++; + dev->nFreeChunks += dev->nChunksPerBlock; +- } else if (state == YAFFS_BLOCK_STATE_NEEDS_SCANNING) { +- +- /* Determine the highest sequence number */ +- if (dev->isYaffs2 && +- sequenceNumber >= YAFFS_LOWEST_SEQUENCE_NUMBER && +- sequenceNumber < YAFFS_HIGHEST_SEQUENCE_NUMBER) { +- +- blockIndex[nBlocksToScan].seq = sequenceNumber; +- blockIndex[nBlocksToScan].block = blk; +- +- nBlocksToScan++; +- +- if (sequenceNumber >= dev->sequenceNumber) { +- dev->sequenceNumber = sequenceNumber; +- } +- } else if (dev->isYaffs2) { +- /* TODO: Nasty sequence number! */ +- T(YAFFS_TRACE_SCAN, +- (TSTR +- ("Block scanning block %d has bad sequence number %d" +- TENDSTR), blk, sequenceNumber)); +- +- } + } + } + +- /* Sort the blocks +- * Dungy old bubble sort for now... +- */ +- if (dev->isYaffs2) { +- yaffs_BlockIndex temp; +- int i; +- int j; +- +- for (i = 0; i < nBlocksToScan; i++) +- for (j = i + 1; j < nBlocksToScan; j++) +- if (blockIndex[i].seq > blockIndex[j].seq) { +- temp = blockIndex[j]; +- blockIndex[j] = blockIndex[i]; +- blockIndex[i] = temp; +- } +- } +- +- /* Now scan the blocks looking at the data. */ +- if (dev->isYaffs2) { +- startIterator = 0; +- endIterator = nBlocksToScan - 1; +- T(YAFFS_TRACE_SCAN_DEBUG, +- (TSTR("%d blocks to be scanned" TENDSTR), nBlocksToScan)); +- } else { +- startIterator = dev->internalStartBlock; +- endIterator = dev->internalEndBlock; +- } ++ startIterator = dev->internalStartBlock; ++ endIterator = dev->internalEndBlock; + + /* For each block.... */ + for (blockIterator = startIterator; !alloc_failed && blockIterator <= endIterator; + blockIterator++) { + +- if (dev->isYaffs2) { +- /* get the block to scan in the correct order */ +- blk = blockIndex[blockIterator].block; +- } else { +- blk = blockIterator; +- } ++ YYIELD(); ++ ++ YYIELD(); ++ ++ blk = blockIterator; + + bi = yaffs_GetBlockInfo(dev, blk); + state = bi->blockState; +@@ -5511,7 +5523,7 @@ static int yaffs_Scan(yaffs_Device * dev + + /* Let's have a good look at this chunk... */ + +- if (!dev->isYaffs2 && tags.chunkDeleted) { ++ if (tags.eccResult == YAFFS_ECC_RESULT_UNFIXED || tags.chunkDeleted) { + /* YAFFS1 only... + * A deleted chunk + */ +@@ -5540,18 +5552,6 @@ static int yaffs_Scan(yaffs_Device * dev + dev->allocationBlockFinder = blk; + /* Set it to here to encourage the allocator to go forth from here. */ + +- /* Yaffs2 sanity check: +- * This should be the one with the highest sequence number +- */ +- if (dev->isYaffs2 +- && (dev->sequenceNumber != +- bi->sequenceNumber)) { +- T(YAFFS_TRACE_ALWAYS, +- (TSTR +- ("yaffs: Allocation block %d was not highest sequence id:" +- " block seq = %d, dev seq = %d" +- TENDSTR), blk,bi->sequenceNumber,dev->sequenceNumber)); +- } + } + + dev->nFreeChunks += (dev->nChunksPerBlock - c); +@@ -5570,11 +5570,11 @@ static int yaffs_Scan(yaffs_Device * dev + * the same chunkId). + */ + +- if(!in) ++ if (!in) + alloc_failed = 1; + +- if(in){ +- if(!yaffs_PutChunkIntoFile(in, tags.chunkId, chunk,1)) ++ if (in) { ++ if (!yaffs_PutChunkIntoFile(in, tags.chunkId, chunk, 1)) + alloc_failed = 1; + } + +@@ -5617,7 +5617,7 @@ static int yaffs_Scan(yaffs_Device * dev + * deleted, and worse still it has changed type. Delete the old object. + */ + +- yaffs_DestroyObject(in); ++ yaffs_DeleteObject(in); + + in = 0; + } +@@ -5627,14 +5627,20 @@ static int yaffs_Scan(yaffs_Device * dev + objectId, + oh->type); + +- if(!in) ++ if (!in) + alloc_failed = 1; + + if (in && oh->shadowsObject > 0) { +- yaffs_HandleShadowedObject(dev, +- oh-> +- shadowsObject, +- 0); ++ ++ struct yaffs_ShadowFixerStruct *fixer; ++ fixer = YMALLOC(sizeof(struct yaffs_ShadowFixerStruct)); ++ if (fixer) { ++ fixer->next = shadowFixerList; ++ shadowFixerList = fixer; ++ fixer->objectId = tags.objectId; ++ fixer->shadowedId = oh->shadowsObject; ++ } ++ + } + + if (in && in->valid) { +@@ -5643,12 +5649,10 @@ static int yaffs_Scan(yaffs_Device * dev + unsigned existingSerial = in->serial; + unsigned newSerial = tags.serialNumber; + +- if (dev->isYaffs2 || +- ((existingSerial + 1) & 3) == +- newSerial) { ++ if (((existingSerial + 1) & 3) == newSerial) { + /* Use new one - destroy the exisiting one */ + yaffs_DeleteChunk(dev, +- in->chunkId, ++ in->hdrChunk, + 1, __LINE__); + in->valid = 0; + } else { +@@ -5681,7 +5685,8 @@ static int yaffs_Scan(yaffs_Device * dev + in->yst_ctime = oh->yst_ctime; + in->yst_rdev = oh->yst_rdev; + #endif +- in->chunkId = chunk; ++ in->hdrChunk = chunk; ++ in->serial = tags.serialNumber; + + } else if (in && !in->valid) { + /* we need to load this info */ +@@ -5705,7 +5710,8 @@ static int yaffs_Scan(yaffs_Device * dev + in->yst_ctime = oh->yst_ctime; + in->yst_rdev = oh->yst_rdev; + #endif +- in->chunkId = chunk; ++ in->hdrChunk = chunk; ++ in->serial = tags.serialNumber; + + yaffs_SetObjectName(in, oh->name); + in->dirty = 0; +@@ -5718,25 +5724,25 @@ static int yaffs_Scan(yaffs_Device * dev + yaffs_FindOrCreateObjectByNumber + (dev, oh->parentObjectId, + YAFFS_OBJECT_TYPE_DIRECTORY); +- if (parent->variantType == ++ if (!parent) ++ alloc_failed = 1; ++ if (parent && parent->variantType == + YAFFS_OBJECT_TYPE_UNKNOWN) { + /* Set up as a directory */ + parent->variantType = +- YAFFS_OBJECT_TYPE_DIRECTORY; +- INIT_LIST_HEAD(&parent->variant. +- directoryVariant. +- children); +- } else if (parent->variantType != +- YAFFS_OBJECT_TYPE_DIRECTORY) +- { ++ YAFFS_OBJECT_TYPE_DIRECTORY; ++ YINIT_LIST_HEAD(&parent->variant. ++ directoryVariant. ++ children); ++ } else if (!parent || parent->variantType != ++ YAFFS_OBJECT_TYPE_DIRECTORY) { + /* Hoosterman, another problem.... + * We're trying to use a non-directory as a directory + */ + + T(YAFFS_TRACE_ERROR, + (TSTR +- ("yaffs tragedy: attempting to use non-directory as" +- " a directory in scan. Put in lost+found." ++ ("yaffs tragedy: attempting to use non-directory as a directory in scan. Put in lost+found." + TENDSTR))); + parent = dev->lostNFoundDir; + } +@@ -5760,15 +5766,6 @@ static int yaffs_Scan(yaffs_Device * dev + /* Todo got a problem */ + break; + case YAFFS_OBJECT_TYPE_FILE: +- if (dev->isYaffs2 +- && oh->isShrink) { +- /* Prune back the shrunken chunks */ +- yaffs_PruneResizedChunks +- (in, oh->fileSize); +- /* Mark the block as having a shrinkHeader */ +- bi->hasShrinkHeader = 1; +- } +- + if (dev->useHeaderFileSize) + + in->variant.fileVariant. +@@ -5778,11 +5775,11 @@ static int yaffs_Scan(yaffs_Device * dev + break; + case YAFFS_OBJECT_TYPE_HARDLINK: + in->variant.hardLinkVariant. +- equivalentObjectId = +- oh->equivalentObjectId; ++ equivalentObjectId = ++ oh->equivalentObjectId; + in->hardLinks.next = +- (struct list_head *) +- hardList; ++ (struct ylist_head *) ++ hardList; + hardList = in; + break; + case YAFFS_OBJECT_TYPE_DIRECTORY: +@@ -5794,15 +5791,17 @@ static int yaffs_Scan(yaffs_Device * dev + case YAFFS_OBJECT_TYPE_SYMLINK: + in->variant.symLinkVariant.alias = + yaffs_CloneString(oh->alias); +- if(!in->variant.symLinkVariant.alias) ++ if (!in->variant.symLinkVariant.alias) + alloc_failed = 1; + break; + } + ++/* + if (parent == dev->deletedDir) { + yaffs_DestroyObject(in); + bi->hasShrinkHeader = 1; + } ++*/ + } + } + } +@@ -5823,10 +5822,6 @@ static int yaffs_Scan(yaffs_Device * dev + + } + +- if (blockIndex) { +- YFREE(blockIndex); +- } +- + + /* Ok, we've done all the scanning. + * Fix up the hard link chains. +@@ -5834,32 +5829,36 @@ static int yaffs_Scan(yaffs_Device * dev + * hardlinks. + */ + +- yaffs_HardlinkFixup(dev,hardList); ++ yaffs_HardlinkFixup(dev, hardList); + +- /* Handle the unlinked files. Since they were left in an unlinked state we should +- * just delete them. +- */ ++ /* Fix up any shadowed objects */ + { +- struct list_head *i; +- struct list_head *n; ++ struct yaffs_ShadowFixerStruct *fixer; ++ yaffs_Object *obj; + +- yaffs_Object *l; +- /* Soft delete all the unlinked files */ +- list_for_each_safe(i, n, +- &dev->unlinkedDir->variant.directoryVariant. +- children) { +- if (i) { +- l = list_entry(i, yaffs_Object, siblings); +- yaffs_DestroyObject(l); +- } ++ while (shadowFixerList) { ++ fixer = shadowFixerList; ++ shadowFixerList = fixer->next; ++ /* Complete the rename transaction by deleting the shadowed object ++ * then setting the object header to unshadowed. ++ */ ++ obj = yaffs_FindObjectByNumber(dev, fixer->shadowedId); ++ if (obj) ++ yaffs_DeleteObject(obj); ++ ++ obj = yaffs_FindObjectByNumber(dev, fixer->objectId); ++ ++ if (obj) ++ yaffs_UpdateObjectHeader(obj, NULL, 1, 0, 0); ++ ++ YFREE(fixer); + } + } + + yaffs_ReleaseTempBuffer(dev, chunkData, __LINE__); + +- if(alloc_failed){ ++ if (alloc_failed) + return YAFFS_FAIL; +- } + + T(YAFFS_TRACE_SCAN, (TSTR("yaffs_Scan ends" TENDSTR))); + +@@ -5871,25 +5870,27 @@ static void yaffs_CheckObjectDetailsLoad + { + __u8 *chunkData; + yaffs_ObjectHeader *oh; +- yaffs_Device *dev = in->myDev; ++ yaffs_Device *dev; + yaffs_ExtendedTags tags; + int result; + int alloc_failed = 0; + +- if(!in) ++ if (!in) + return; + ++ dev = in->myDev; ++ + #if 0 +- T(YAFFS_TRACE_SCAN,(TSTR("details for object %d %s loaded" TENDSTR), ++ T(YAFFS_TRACE_SCAN, (TSTR("details for object %d %s loaded" TENDSTR), + in->objectId, + in->lazyLoaded ? "not yet" : "already")); + #endif + +- if(in->lazyLoaded){ ++ if (in->lazyLoaded && in->hdrChunk > 0) { + in->lazyLoaded = 0; + chunkData = yaffs_GetTempBuffer(dev, __LINE__); + +- result = yaffs_ReadChunkWithTagsFromNAND(dev,in->chunkId,chunkData,&tags); ++ result = yaffs_ReadChunkWithTagsFromNAND(dev, in->hdrChunk, chunkData, &tags); + oh = (yaffs_ObjectHeader *) chunkData; + + in->yst_mode = oh->yst_mode; +@@ -5911,18 +5912,18 @@ static void yaffs_CheckObjectDetailsLoad + #endif + yaffs_SetObjectName(in, oh->name); + +- if(in->variantType == YAFFS_OBJECT_TYPE_SYMLINK){ +- in->variant.symLinkVariant.alias = ++ if (in->variantType == YAFFS_OBJECT_TYPE_SYMLINK) { ++ in->variant.symLinkVariant.alias = + yaffs_CloneString(oh->alias); +- if(!in->variant.symLinkVariant.alias) ++ if (!in->variant.symLinkVariant.alias) + alloc_failed = 1; /* Not returned to caller */ + } + +- yaffs_ReleaseTempBuffer(dev,chunkData, __LINE__); ++ yaffs_ReleaseTempBuffer(dev, chunkData, __LINE__); + } + } + +-static int yaffs_ScanBackwards(yaffs_Device * dev) ++static int yaffs_ScanBackwards(yaffs_Device *dev) + { + yaffs_ExtendedTags tags; + int blk; +@@ -5938,7 +5939,7 @@ static int yaffs_ScanBackwards(yaffs_Dev + yaffs_BlockState state; + yaffs_Object *hardList = NULL; + yaffs_BlockInfo *bi; +- int sequenceNumber; ++ __u32 sequenceNumber; + yaffs_ObjectHeader *oh; + yaffs_Object *in; + yaffs_Object *parent; +@@ -5972,12 +5973,12 @@ static int yaffs_ScanBackwards(yaffs_Dev + + blockIndex = YMALLOC(nBlocks * sizeof(yaffs_BlockIndex)); + +- if(!blockIndex) { ++ if (!blockIndex) { + blockIndex = YMALLOC_ALT(nBlocks * sizeof(yaffs_BlockIndex)); + altBlockIndex = 1; + } + +- if(!blockIndex) { ++ if (!blockIndex) { + T(YAFFS_TRACE_SCAN, + (TSTR("yaffs_Scan() could not allocate block index!" TENDSTR))); + return YAFFS_FAIL; +@@ -5999,15 +6000,17 @@ static int yaffs_ScanBackwards(yaffs_Dev + bi->blockState = state; + bi->sequenceNumber = sequenceNumber; + +- if(bi->sequenceNumber == YAFFS_SEQUENCE_CHECKPOINT_DATA) ++ if (bi->sequenceNumber == YAFFS_SEQUENCE_CHECKPOINT_DATA) + bi->blockState = state = YAFFS_BLOCK_STATE_CHECKPOINT; ++ if (bi->sequenceNumber == YAFFS_SEQUENCE_BAD_BLOCK) ++ bi->blockState = state = YAFFS_BLOCK_STATE_DEAD; + + T(YAFFS_TRACE_SCAN_DEBUG, + (TSTR("Block scanning block %d state %d seq %d" TENDSTR), blk, + state, sequenceNumber)); + + +- if(state == YAFFS_BLOCK_STATE_CHECKPOINT){ ++ if (state == YAFFS_BLOCK_STATE_CHECKPOINT) { + dev->blocksInCheckpoint++; + + } else if (state == YAFFS_BLOCK_STATE_DEAD) { +@@ -6021,8 +6024,7 @@ static int yaffs_ScanBackwards(yaffs_Dev + } else if (state == YAFFS_BLOCK_STATE_NEEDS_SCANNING) { + + /* Determine the highest sequence number */ +- if (dev->isYaffs2 && +- sequenceNumber >= YAFFS_LOWEST_SEQUENCE_NUMBER && ++ if (sequenceNumber >= YAFFS_LOWEST_SEQUENCE_NUMBER && + sequenceNumber < YAFFS_HIGHEST_SEQUENCE_NUMBER) { + + blockIndex[nBlocksToScan].seq = sequenceNumber; +@@ -6030,10 +6032,9 @@ static int yaffs_ScanBackwards(yaffs_Dev + + nBlocksToScan++; + +- if (sequenceNumber >= dev->sequenceNumber) { ++ if (sequenceNumber >= dev->sequenceNumber) + dev->sequenceNumber = sequenceNumber; +- } +- } else if (dev->isYaffs2) { ++ } else { + /* TODO: Nasty sequence number! */ + T(YAFFS_TRACE_SCAN, + (TSTR +@@ -6053,11 +6054,13 @@ static int yaffs_ScanBackwards(yaffs_Dev + + /* Sort the blocks */ + #ifndef CONFIG_YAFFS_USE_OWN_SORT +- yaffs_qsort(blockIndex, nBlocksToScan, +- sizeof(yaffs_BlockIndex), ybicmp); ++ { ++ /* Use qsort now. */ ++ yaffs_qsort(blockIndex, nBlocksToScan, sizeof(yaffs_BlockIndex), ybicmp); ++ } + #else + { +- /* Dungy old bubble sort... */ ++ /* Dungy old bubble sort... */ + + yaffs_BlockIndex temp; + int i; +@@ -6075,7 +6078,7 @@ static int yaffs_ScanBackwards(yaffs_Dev + + YYIELD(); + +- T(YAFFS_TRACE_SCAN, (TSTR("...done" TENDSTR))); ++ T(YAFFS_TRACE_SCAN, (TSTR("...done" TENDSTR))); + + /* Now scan the blocks looking at the data. */ + startIterator = 0; +@@ -6085,10 +6088,10 @@ static int yaffs_ScanBackwards(yaffs_Dev + + /* For each block.... backwards */ + for (blockIterator = endIterator; !alloc_failed && blockIterator >= startIterator; +- blockIterator--) { +- /* Cooperative multitasking! This loop can run for so ++ blockIterator--) { ++ /* Cooperative multitasking! This loop can run for so + long that watchdog timers expire. */ +- YYIELD(); ++ YYIELD(); + + /* get the block to scan in the correct order */ + blk = blockIndex[blockIterator].block; +@@ -6127,10 +6130,8 @@ static int yaffs_ScanBackwards(yaffs_Dev + * this is the one being allocated from + */ + +- if(foundChunksInBlock) +- { ++ if (foundChunksInBlock) { + /* This is a chunk that was skipped due to failing the erased check */ +- + } else if (c == 0) { + /* We're looking at the first chunk in the block so the block is unused */ + state = YAFFS_BLOCK_STATE_EMPTY; +@@ -6138,7 +6139,7 @@ static int yaffs_ScanBackwards(yaffs_Dev + } else { + if (state == YAFFS_BLOCK_STATE_NEEDS_SCANNING || + state == YAFFS_BLOCK_STATE_ALLOCATING) { +- if(dev->sequenceNumber == bi->sequenceNumber) { ++ if (dev->sequenceNumber == bi->sequenceNumber) { + /* this is the block being allocated from */ + + T(YAFFS_TRACE_SCAN, +@@ -6150,27 +6151,31 @@ static int yaffs_ScanBackwards(yaffs_Dev + dev->allocationBlock = blk; + dev->allocationPage = c; + dev->allocationBlockFinder = blk; +- } +- else { ++ } else { + /* This is a partially written block that is not + * the current allocation block. This block must have + * had a write failure, so set up for retirement. + */ + +- bi->needsRetiring = 1; ++ /* bi->needsRetiring = 1; ??? TODO */ + bi->gcPrioritise = 1; + + T(YAFFS_TRACE_ALWAYS, +- (TSTR("Partially written block %d being set for retirement" TENDSTR), ++ (TSTR("Partially written block %d detected" TENDSTR), + blk)); + } +- + } +- + } + + dev->nFreeChunks++; + ++ } else if (tags.eccResult == YAFFS_ECC_RESULT_UNFIXED) { ++ T(YAFFS_TRACE_SCAN, ++ (TSTR(" Unfixed ECC in chunk(%d:%d), chunk ignored"TENDSTR), ++ blk, c)); ++ ++ dev->nFreeChunks++; ++ + } else if (tags.chunkId > 0) { + /* chunkId > 0 so it is a data chunk... */ + unsigned int endpos; +@@ -6187,7 +6192,7 @@ static int yaffs_ScanBackwards(yaffs_Dev + tags. + objectId, + YAFFS_OBJECT_TYPE_FILE); +- if(!in){ ++ if (!in) { + /* Out of memory */ + alloc_failed = 1; + } +@@ -6197,8 +6202,8 @@ static int yaffs_ScanBackwards(yaffs_Dev + && chunkBase < + in->variant.fileVariant.shrinkSize) { + /* This has not been invalidated by a resize */ +- if(!yaffs_PutChunkIntoFile(in, tags.chunkId, +- chunk, -1)){ ++ if (!yaffs_PutChunkIntoFile(in, tags.chunkId, ++ chunk, -1)) { + alloc_failed = 1; + } + +@@ -6221,7 +6226,7 @@ static int yaffs_ScanBackwards(yaffs_Dev + scannedFileSize; + } + +- } else if(in) { ++ } else if (in) { + /* This chunk has been invalidated by a resize, so delete */ + yaffs_DeleteChunk(dev, chunk, 1, __LINE__); + +@@ -6242,6 +6247,8 @@ static int yaffs_ScanBackwards(yaffs_Dev + in = yaffs_FindOrCreateObjectByNumber + (dev, tags.objectId, + tags.extraObjectType); ++ if (!in) ++ alloc_failed = 1; + } + + if (!in || +@@ -6251,8 +6258,7 @@ static int yaffs_ScanBackwards(yaffs_Dev + tags.extraShadows || + (!in->valid && + (tags.objectId == YAFFS_OBJECTID_ROOT || +- tags.objectId == YAFFS_OBJECTID_LOSTNFOUND)) +- ) { ++ tags.objectId == YAFFS_OBJECTID_LOSTNFOUND))) { + + /* If we don't have valid info then we need to read the chunk + * TODO In future we can probably defer reading the chunk and +@@ -6266,8 +6272,17 @@ static int yaffs_ScanBackwards(yaffs_Dev + + oh = (yaffs_ObjectHeader *) chunkData; + +- if (!in) ++ if (dev->inbandTags) { ++ /* Fix up the header if they got corrupted by inband tags */ ++ oh->shadowsObject = oh->inbandShadowsObject; ++ oh->isShrink = oh->inbandIsShrink; ++ } ++ ++ if (!in) { + in = yaffs_FindOrCreateObjectByNumber(dev, tags.objectId, oh->type); ++ if (!in) ++ alloc_failed = 1; ++ } + + } + +@@ -6275,10 +6290,9 @@ static int yaffs_ScanBackwards(yaffs_Dev + /* TODO Hoosterman we have a problem! */ + T(YAFFS_TRACE_ERROR, + (TSTR +- ("yaffs tragedy: Could not make object for object %d " +- "at chunk %d during scan" ++ ("yaffs tragedy: Could not make object for object %d at chunk %d during scan" + TENDSTR), tags.objectId, chunk)); +- ++ continue; + } + + if (in->valid) { +@@ -6289,10 +6303,9 @@ static int yaffs_ScanBackwards(yaffs_Dev + + if ((in->variantType == YAFFS_OBJECT_TYPE_FILE) && + ((oh && +- oh-> type == YAFFS_OBJECT_TYPE_FILE)|| ++ oh->type == YAFFS_OBJECT_TYPE_FILE) || + (tags.extraHeaderInfoAvailable && +- tags.extraObjectType == YAFFS_OBJECT_TYPE_FILE)) +- ) { ++ tags.extraObjectType == YAFFS_OBJECT_TYPE_FILE))) { + __u32 thisSize = + (oh) ? oh->fileSize : tags. + extraFileLength; +@@ -6300,7 +6313,9 @@ static int yaffs_ScanBackwards(yaffs_Dev + (oh) ? oh-> + parentObjectId : tags. + extraParentObjectId; +- unsigned isShrink = ++ ++ ++ isShrink = + (oh) ? oh->isShrink : tags. + extraIsShrinkHeader; + +@@ -6323,9 +6338,8 @@ static int yaffs_ScanBackwards(yaffs_Dev + thisSize; + } + +- if (isShrink) { ++ if (isShrink) + bi->hasShrinkHeader = 1; +- } + + } + /* Use existing - destroy this one. */ +@@ -6333,6 +6347,17 @@ static int yaffs_ScanBackwards(yaffs_Dev + + } + ++ if (!in->valid && in->variantType != ++ (oh ? oh->type : tags.extraObjectType)) ++ T(YAFFS_TRACE_ERROR, ( ++ TSTR("yaffs tragedy: Bad object type, " ++ TCONT("%d != %d, for object %d at chunk ") ++ TCONT("%d during scan") ++ TENDSTR), oh ? ++ oh->type : tags.extraObjectType, ++ in->variantType, tags.objectId, ++ chunk)); ++ + if (!in->valid && + (tags.objectId == YAFFS_OBJECTID_ROOT || + tags.objectId == +@@ -6340,7 +6365,7 @@ static int yaffs_ScanBackwards(yaffs_Dev + /* We only load some info, don't fiddle with directory structure */ + in->valid = 1; + +- if(oh) { ++ if (oh) { + in->variantType = oh->type; + + in->yst_mode = oh->yst_mode; +@@ -6365,15 +6390,15 @@ static int yaffs_ScanBackwards(yaffs_Dev + in->lazyLoaded = 1; + } + +- in->chunkId = chunk; ++ in->hdrChunk = chunk; + + } else if (!in->valid) { + /* we need to load this info */ + + in->valid = 1; +- in->chunkId = chunk; ++ in->hdrChunk = chunk; + +- if(oh) { ++ if (oh) { + in->variantType = oh->type; + + in->yst_mode = oh->yst_mode; +@@ -6403,20 +6428,19 @@ static int yaffs_ScanBackwards(yaffs_Dev + yaffs_SetObjectName(in, oh->name); + parent = + yaffs_FindOrCreateObjectByNumber +- (dev, oh->parentObjectId, +- YAFFS_OBJECT_TYPE_DIRECTORY); ++ (dev, oh->parentObjectId, ++ YAFFS_OBJECT_TYPE_DIRECTORY); + + fileSize = oh->fileSize; +- isShrink = oh->isShrink; ++ isShrink = oh->isShrink; + equivalentObjectId = oh->equivalentObjectId; + +- } +- else { ++ } else { + in->variantType = tags.extraObjectType; + parent = + yaffs_FindOrCreateObjectByNumber +- (dev, tags.extraParentObjectId, +- YAFFS_OBJECT_TYPE_DIRECTORY); ++ (dev, tags.extraParentObjectId, ++ YAFFS_OBJECT_TYPE_DIRECTORY); + fileSize = tags.extraFileLength; + isShrink = tags.extraIsShrinkHeader; + equivalentObjectId = tags.extraEquivalentObjectId; +@@ -6425,29 +6449,30 @@ static int yaffs_ScanBackwards(yaffs_Dev + } + in->dirty = 0; + ++ if (!parent) ++ alloc_failed = 1; ++ + /* directory stuff... + * hook up to parent + */ + +- if (parent->variantType == ++ if (parent && parent->variantType == + YAFFS_OBJECT_TYPE_UNKNOWN) { + /* Set up as a directory */ + parent->variantType = +- YAFFS_OBJECT_TYPE_DIRECTORY; +- INIT_LIST_HEAD(&parent->variant. +- directoryVariant. +- children); +- } else if (parent->variantType != +- YAFFS_OBJECT_TYPE_DIRECTORY) +- { ++ YAFFS_OBJECT_TYPE_DIRECTORY; ++ YINIT_LIST_HEAD(&parent->variant. ++ directoryVariant. ++ children); ++ } else if (!parent || parent->variantType != ++ YAFFS_OBJECT_TYPE_DIRECTORY) { + /* Hoosterman, another problem.... + * We're trying to use a non-directory as a directory + */ + + T(YAFFS_TRACE_ERROR, + (TSTR +- ("yaffs tragedy: attempting to use non-directory as" +- " a directory in scan. Put in lost+found." ++ ("yaffs tragedy: attempting to use non-directory as a directory in scan. Put in lost+found." + TENDSTR))); + parent = dev->lostNFoundDir; + } +@@ -6494,12 +6519,12 @@ static int yaffs_ScanBackwards(yaffs_Dev + + break; + case YAFFS_OBJECT_TYPE_HARDLINK: +- if(!itsUnlinked) { +- in->variant.hardLinkVariant.equivalentObjectId = +- equivalentObjectId; +- in->hardLinks.next = +- (struct list_head *) hardList; +- hardList = in; ++ if (!itsUnlinked) { ++ in->variant.hardLinkVariant.equivalentObjectId = ++ equivalentObjectId; ++ in->hardLinks.next = ++ (struct ylist_head *) hardList; ++ hardList = in; + } + break; + case YAFFS_OBJECT_TYPE_DIRECTORY: +@@ -6509,12 +6534,11 @@ static int yaffs_ScanBackwards(yaffs_Dev + /* Do nothing */ + break; + case YAFFS_OBJECT_TYPE_SYMLINK: +- if(oh){ +- in->variant.symLinkVariant.alias = +- yaffs_CloneString(oh-> +- alias); +- if(!in->variant.symLinkVariant.alias) +- alloc_failed = 1; ++ if (oh) { ++ in->variant.symLinkVariant.alias = ++ yaffs_CloneString(oh->alias); ++ if (!in->variant.symLinkVariant.alias) ++ alloc_failed = 1; + } + break; + } +@@ -6551,75 +6575,129 @@ static int yaffs_ScanBackwards(yaffs_Dev + * We should now have scanned all the objects, now it's time to add these + * hardlinks. + */ +- yaffs_HardlinkFixup(dev,hardList); ++ yaffs_HardlinkFixup(dev, hardList); + + +- /* +- * Sort out state of unlinked and deleted objects. +- */ +- { +- struct list_head *i; +- struct list_head *n; ++ yaffs_ReleaseTempBuffer(dev, chunkData, __LINE__); + +- yaffs_Object *l; ++ if (alloc_failed) ++ return YAFFS_FAIL; + +- /* Soft delete all the unlinked files */ +- list_for_each_safe(i, n, +- &dev->unlinkedDir->variant.directoryVariant. +- children) { +- if (i) { +- l = list_entry(i, yaffs_Object, siblings); +- yaffs_DestroyObject(l); +- } +- } ++ T(YAFFS_TRACE_SCAN, (TSTR("yaffs_ScanBackwards ends" TENDSTR))); + +- /* Soft delete all the deletedDir files */ +- list_for_each_safe(i, n, +- &dev->deletedDir->variant.directoryVariant. +- children) { +- if (i) { +- l = list_entry(i, yaffs_Object, siblings); +- yaffs_DestroyObject(l); ++ return YAFFS_OK; ++} + +- } ++/*------------------------------ Directory Functions ----------------------------- */ ++ ++static void yaffs_VerifyObjectInDirectory(yaffs_Object *obj) ++{ ++ struct ylist_head *lh; ++ yaffs_Object *listObj; ++ ++ int count = 0; ++ ++ if (!obj) { ++ T(YAFFS_TRACE_ALWAYS, (TSTR("No object to verify" TENDSTR))); ++ YBUG(); ++ return; ++ } ++ ++ if (yaffs_SkipVerification(obj->myDev)) ++ return; ++ ++ if (!obj->parent) { ++ T(YAFFS_TRACE_ALWAYS, (TSTR("Object does not have parent" TENDSTR))); ++ YBUG(); ++ return; ++ } ++ ++ if (obj->parent->variantType != YAFFS_OBJECT_TYPE_DIRECTORY) { ++ T(YAFFS_TRACE_ALWAYS, (TSTR("Parent is not directory" TENDSTR))); ++ YBUG(); ++ } ++ ++ /* Iterate through the objects in each hash entry */ ++ ++ ylist_for_each(lh, &obj->parent->variant.directoryVariant.children) { ++ if (lh) { ++ listObj = ylist_entry(lh, yaffs_Object, siblings); ++ yaffs_VerifyObject(listObj); ++ if (obj == listObj) ++ count++; + } ++ } ++ ++ if (count != 1) { ++ T(YAFFS_TRACE_ALWAYS, (TSTR("Object in directory %d times" TENDSTR), count)); ++ YBUG(); + } ++} + +- yaffs_ReleaseTempBuffer(dev, chunkData, __LINE__); ++static void yaffs_VerifyDirectory(yaffs_Object *directory) ++{ ++ struct ylist_head *lh; ++ yaffs_Object *listObj; + +- if(alloc_failed){ +- return YAFFS_FAIL; ++ if (!directory) { ++ YBUG(); ++ return; + } + +- T(YAFFS_TRACE_SCAN, (TSTR("yaffs_ScanBackwards ends" TENDSTR))); ++ if (yaffs_SkipFullVerification(directory->myDev)) ++ return; + +- return YAFFS_OK; ++ if (directory->variantType != YAFFS_OBJECT_TYPE_DIRECTORY) { ++ T(YAFFS_TRACE_ALWAYS, (TSTR("Directory has wrong type: %d" TENDSTR), directory->variantType)); ++ YBUG(); ++ } ++ ++ /* Iterate through the objects in each hash entry */ ++ ++ ylist_for_each(lh, &directory->variant.directoryVariant.children) { ++ if (lh) { ++ listObj = ylist_entry(lh, yaffs_Object, siblings); ++ if (listObj->parent != directory) { ++ T(YAFFS_TRACE_ALWAYS, (TSTR("Object in directory list has wrong parent %p" TENDSTR), listObj->parent)); ++ YBUG(); ++ } ++ yaffs_VerifyObjectInDirectory(listObj); ++ } ++ } + } + +-/*------------------------------ Directory Functions ----------------------------- */ + +-static void yaffs_RemoveObjectFromDirectory(yaffs_Object * obj) ++static void yaffs_RemoveObjectFromDirectory(yaffs_Object *obj) + { + yaffs_Device *dev = obj->myDev; ++ yaffs_Object *parent; ++ ++ yaffs_VerifyObjectInDirectory(obj); ++ parent = obj->parent; ++ ++ yaffs_VerifyDirectory(parent); + +- if(dev && dev->removeObjectCallback) ++ if (dev && dev->removeObjectCallback) + dev->removeObjectCallback(obj); + +- list_del_init(&obj->siblings); ++ ++ ylist_del_init(&obj->siblings); + obj->parent = NULL; ++ ++ yaffs_VerifyDirectory(parent); + } + + +-static void yaffs_AddObjectToDirectory(yaffs_Object * directory, +- yaffs_Object * obj) ++static void yaffs_AddObjectToDirectory(yaffs_Object *directory, ++ yaffs_Object *obj) + { +- + if (!directory) { + T(YAFFS_TRACE_ALWAYS, + (TSTR + ("tragedy: Trying to add an object to a null pointer directory" + TENDSTR))); + YBUG(); ++ return; + } + if (directory->variantType != YAFFS_OBJECT_TYPE_DIRECTORY) { + T(YAFFS_TRACE_ALWAYS, +@@ -6631,37 +6709,42 @@ static void yaffs_AddObjectToDirectory(y + + if (obj->siblings.prev == NULL) { + /* Not initialised */ +- INIT_LIST_HEAD(&obj->siblings); +- +- } else if (!list_empty(&obj->siblings)) { +- /* If it is holed up somewhere else, un hook it */ +- yaffs_RemoveObjectFromDirectory(obj); ++ YBUG(); + } ++ ++ ++ yaffs_VerifyDirectory(directory); ++ ++ yaffs_RemoveObjectFromDirectory(obj); ++ ++ + /* Now add it */ +- list_add(&obj->siblings, &directory->variant.directoryVariant.children); ++ ylist_add(&obj->siblings, &directory->variant.directoryVariant.children); + obj->parent = directory; + + if (directory == obj->myDev->unlinkedDir +- || directory == obj->myDev->deletedDir) { ++ || directory == obj->myDev->deletedDir) { + obj->unlinked = 1; + obj->myDev->nUnlinkedFiles++; + obj->renameAllowed = 0; + } ++ ++ yaffs_VerifyDirectory(directory); ++ yaffs_VerifyObjectInDirectory(obj); + } + +-yaffs_Object *yaffs_FindObjectByName(yaffs_Object * directory, +- const YCHAR * name) ++yaffs_Object *yaffs_FindObjectByName(yaffs_Object *directory, ++ const YCHAR *name) + { + int sum; + +- struct list_head *i; ++ struct ylist_head *i; + YCHAR buffer[YAFFS_MAX_NAME_LENGTH + 1]; + + yaffs_Object *l; + +- if (!name) { ++ if (!name) + return NULL; +- } + + if (!directory) { + T(YAFFS_TRACE_ALWAYS, +@@ -6669,6 +6752,7 @@ yaffs_Object *yaffs_FindObjectByName(yaf + ("tragedy: yaffs_FindObjectByName: null pointer directory" + TENDSTR))); + YBUG(); ++ return NULL; + } + if (directory->variantType != YAFFS_OBJECT_TYPE_DIRECTORY) { + T(YAFFS_TRACE_ALWAYS, +@@ -6679,28 +6763,27 @@ yaffs_Object *yaffs_FindObjectByName(yaf + + sum = yaffs_CalcNameSum(name); + +- list_for_each(i, &directory->variant.directoryVariant.children) { ++ ylist_for_each(i, &directory->variant.directoryVariant.children) { + if (i) { +- l = list_entry(i, yaffs_Object, siblings); ++ l = ylist_entry(i, yaffs_Object, siblings); ++ ++ if (l->parent != directory) ++ YBUG(); + + yaffs_CheckObjectDetailsLoaded(l); + + /* Special case for lost-n-found */ + if (l->objectId == YAFFS_OBJECTID_LOSTNFOUND) { +- if (yaffs_strcmp(name, YAFFS_LOSTNFOUND_NAME) == 0) { ++ if (yaffs_strcmp(name, YAFFS_LOSTNFOUND_NAME) == 0) + return l; +- } +- } else if (yaffs_SumCompare(l->sum, sum) || l->chunkId <= 0) +- { +- /* LostnFound cunk called Objxxx ++ } else if (yaffs_SumCompare(l->sum, sum) || l->hdrChunk <= 0) { ++ /* LostnFound chunk called Objxxx + * Do a real check + */ + yaffs_GetObjectName(l, buffer, + YAFFS_MAX_NAME_LENGTH); +- if (yaffs_strncmp(name, buffer,YAFFS_MAX_NAME_LENGTH) == 0) { ++ if (yaffs_strncmp(name, buffer, YAFFS_MAX_NAME_LENGTH) == 0) + return l; +- } +- + } + } + } +@@ -6710,10 +6793,10 @@ yaffs_Object *yaffs_FindObjectByName(yaf + + + #if 0 +-int yaffs_ApplyToDirectoryChildren(yaffs_Object * theDir, +- int (*fn) (yaffs_Object *)) ++int yaffs_ApplyToDirectoryChildren(yaffs_Object *theDir, ++ int (*fn) (yaffs_Object *)) + { +- struct list_head *i; ++ struct ylist_head *i; + yaffs_Object *l; + + if (!theDir) { +@@ -6722,20 +6805,21 @@ int yaffs_ApplyToDirectoryChildren(yaffs + ("tragedy: yaffs_FindObjectByName: null pointer directory" + TENDSTR))); + YBUG(); ++ return YAFFS_FAIL; + } + if (theDir->variantType != YAFFS_OBJECT_TYPE_DIRECTORY) { + T(YAFFS_TRACE_ALWAYS, + (TSTR + ("tragedy: yaffs_FindObjectByName: non-directory" TENDSTR))); + YBUG(); ++ return YAFFS_FAIL; + } + +- list_for_each(i, &theDir->variant.directoryVariant.children) { ++ ylist_for_each(i, &theDir->variant.directoryVariant.children) { + if (i) { +- l = list_entry(i, yaffs_Object, siblings); +- if (l && !fn(l)) { ++ l = ylist_entry(i, yaffs_Object, siblings); ++ if (l && !fn(l)) + return YAFFS_FAIL; +- } + } + } + +@@ -6748,7 +6832,7 @@ int yaffs_ApplyToDirectoryChildren(yaffs + * actual object. + */ + +-yaffs_Object *yaffs_GetEquivalentObject(yaffs_Object * obj) ++yaffs_Object *yaffs_GetEquivalentObject(yaffs_Object *obj) + { + if (obj && obj->variantType == YAFFS_OBJECT_TYPE_HARDLINK) { + /* We want the object id of the equivalent object, not this one */ +@@ -6756,10 +6840,9 @@ yaffs_Object *yaffs_GetEquivalentObject( + yaffs_CheckObjectDetailsLoaded(obj); + } + return obj; +- + } + +-int yaffs_GetObjectName(yaffs_Object * obj, YCHAR * name, int buffSize) ++int yaffs_GetObjectName(yaffs_Object *obj, YCHAR *name, int buffSize) + { + memset(name, 0, buffSize * sizeof(YCHAR)); + +@@ -6767,18 +6850,26 @@ int yaffs_GetObjectName(yaffs_Object * o + + if (obj->objectId == YAFFS_OBJECTID_LOSTNFOUND) { + yaffs_strncpy(name, YAFFS_LOSTNFOUND_NAME, buffSize - 1); +- } else if (obj->chunkId <= 0) { ++ } else if (obj->hdrChunk <= 0) { + YCHAR locName[20]; ++ YCHAR numString[20]; ++ YCHAR *x = &numString[19]; ++ unsigned v = obj->objectId; ++ numString[19] = 0; ++ while (v > 0) { ++ x--; ++ *x = '0' + (v % 10); ++ v /= 10; ++ } + /* make up a name */ +- yaffs_sprintf(locName, _Y("%s%d"), YAFFS_LOSTNFOUND_PREFIX, +- obj->objectId); ++ yaffs_strcpy(locName, YAFFS_LOSTNFOUND_PREFIX); ++ yaffs_strcat(locName, x); + yaffs_strncpy(name, locName, buffSize - 1); + + } + #ifdef CONFIG_YAFFS_SHORT_NAMES_IN_RAM +- else if (obj->shortName[0]) { ++ else if (obj->shortName[0]) + yaffs_strcpy(name, obj->shortName); +- } + #endif + else { + int result; +@@ -6788,9 +6879,9 @@ int yaffs_GetObjectName(yaffs_Object * o + + memset(buffer, 0, obj->myDev->nDataBytesPerChunk); + +- if (obj->chunkId >= 0) { ++ if (obj->hdrChunk > 0) { + result = yaffs_ReadChunkWithTagsFromNAND(obj->myDev, +- obj->chunkId, buffer, ++ obj->hdrChunk, buffer, + NULL); + } + yaffs_strncpy(name, oh->name, buffSize - 1); +@@ -6801,46 +6892,43 @@ int yaffs_GetObjectName(yaffs_Object * o + return yaffs_strlen(name); + } + +-int yaffs_GetObjectFileLength(yaffs_Object * obj) ++int yaffs_GetObjectFileLength(yaffs_Object *obj) + { +- + /* Dereference any hard linking */ + obj = yaffs_GetEquivalentObject(obj); + +- if (obj->variantType == YAFFS_OBJECT_TYPE_FILE) { ++ if (obj->variantType == YAFFS_OBJECT_TYPE_FILE) + return obj->variant.fileVariant.fileSize; +- } +- if (obj->variantType == YAFFS_OBJECT_TYPE_SYMLINK) { ++ if (obj->variantType == YAFFS_OBJECT_TYPE_SYMLINK) + return yaffs_strlen(obj->variant.symLinkVariant.alias); +- } else { ++ else { + /* Only a directory should drop through to here */ + return obj->myDev->nDataBytesPerChunk; + } + } + +-int yaffs_GetObjectLinkCount(yaffs_Object * obj) ++int yaffs_GetObjectLinkCount(yaffs_Object *obj) + { + int count = 0; +- struct list_head *i; ++ struct ylist_head *i; + +- if (!obj->unlinked) { +- count++; /* the object itself */ +- } +- list_for_each(i, &obj->hardLinks) { +- count++; /* add the hard links; */ +- } +- return count; ++ if (!obj->unlinked) ++ count++; /* the object itself */ ++ ++ ylist_for_each(i, &obj->hardLinks) ++ count++; /* add the hard links; */ + ++ return count; + } + +-int yaffs_GetObjectInode(yaffs_Object * obj) ++int yaffs_GetObjectInode(yaffs_Object *obj) + { + obj = yaffs_GetEquivalentObject(obj); + + return obj->objectId; + } + +-unsigned yaffs_GetObjectType(yaffs_Object * obj) ++unsigned yaffs_GetObjectType(yaffs_Object *obj) + { + obj = yaffs_GetEquivalentObject(obj); + +@@ -6872,19 +6960,18 @@ unsigned yaffs_GetObjectType(yaffs_Objec + } + } + +-YCHAR *yaffs_GetSymlinkAlias(yaffs_Object * obj) ++YCHAR *yaffs_GetSymlinkAlias(yaffs_Object *obj) + { + obj = yaffs_GetEquivalentObject(obj); +- if (obj->variantType == YAFFS_OBJECT_TYPE_SYMLINK) { ++ if (obj->variantType == YAFFS_OBJECT_TYPE_SYMLINK) + return yaffs_CloneString(obj->variant.symLinkVariant.alias); +- } else { ++ else + return yaffs_CloneString(_Y("")); +- } + } + + #ifndef CONFIG_YAFFS_WINCE + +-int yaffs_SetAttributes(yaffs_Object * obj, struct iattr *attr) ++int yaffs_SetAttributes(yaffs_Object *obj, struct iattr *attr) + { + unsigned int valid = attr->ia_valid; + +@@ -6910,7 +6997,7 @@ int yaffs_SetAttributes(yaffs_Object * o + return YAFFS_OK; + + } +-int yaffs_GetAttributes(yaffs_Object * obj, struct iattr *attr) ++int yaffs_GetAttributes(yaffs_Object *obj, struct iattr *attr) + { + unsigned int valid = 0; + +@@ -6934,13 +7021,12 @@ int yaffs_GetAttributes(yaffs_Object * o + attr->ia_valid = valid; + + return YAFFS_OK; +- + } + + #endif + + #if 0 +-int yaffs_DumpObject(yaffs_Object * obj) ++int yaffs_DumpObject(yaffs_Object *obj) + { + YCHAR name[257]; + +@@ -6951,7 +7037,7 @@ int yaffs_DumpObject(yaffs_Object * obj) + ("Object %d, inode %d \"%s\"\n dirty %d valid %d serial %d sum %d" + " chunk %d type %d size %d\n" + TENDSTR), obj->objectId, yaffs_GetObjectInode(obj), name, +- obj->dirty, obj->valid, obj->serial, obj->sum, obj->chunkId, ++ obj->dirty, obj->valid, obj->serial, obj->sum, obj->hdrChunk, + yaffs_GetObjectType(obj), yaffs_GetObjectFileLength(obj))); + + return YAFFS_OK; +@@ -6960,7 +7046,7 @@ int yaffs_DumpObject(yaffs_Object * obj) + + /*---------------------------- Initialisation code -------------------------------------- */ + +-static int yaffs_CheckDevFunctions(const yaffs_Device * dev) ++static int yaffs_CheckDevFunctions(const yaffs_Device *dev) + { + + /* Common functions, gotta have */ +@@ -7011,7 +7097,7 @@ static int yaffs_CreateInitialDirectorie + yaffs_CreateFakeDirectory(dev, YAFFS_OBJECTID_LOSTNFOUND, + YAFFS_LOSTNFOUND_MODE | S_IFDIR); + +- if(dev->lostNFoundDir && dev->rootDir && dev->unlinkedDir && dev->deletedDir){ ++ if (dev->lostNFoundDir && dev->rootDir && dev->unlinkedDir && dev->deletedDir) { + yaffs_AddObjectToDirectory(dev->rootDir, dev->lostNFoundDir); + return YAFFS_OK; + } +@@ -7019,7 +7105,7 @@ static int yaffs_CreateInitialDirectorie + return YAFFS_FAIL; + } + +-int yaffs_GutsInitialise(yaffs_Device * dev) ++int yaffs_GutsInitialise(yaffs_Device *dev) + { + int init_failed = 0; + unsigned x; +@@ -7040,6 +7126,8 @@ int yaffs_GutsInitialise(yaffs_Device * + dev->chunkOffset = 0; + dev->nFreeChunks = 0; + ++ dev->gcBlock = -1; ++ + if (dev->startBlock == 0) { + dev->internalStartBlock = dev->startBlock + 1; + dev->internalEndBlock = dev->endBlock + 1; +@@ -7049,18 +7137,18 @@ int yaffs_GutsInitialise(yaffs_Device * + + /* Check geometry parameters. */ + +- if ((dev->isYaffs2 && dev->nDataBytesPerChunk < 1024) || +- (!dev->isYaffs2 && dev->nDataBytesPerChunk != 512) || ++ if ((!dev->inbandTags && dev->isYaffs2 && dev->totalBytesPerChunk < 1024) || ++ (!dev->isYaffs2 && dev->totalBytesPerChunk < 512) || ++ (dev->inbandTags && !dev->isYaffs2) || + dev->nChunksPerBlock < 2 || + dev->nReservedBlocks < 2 || + dev->internalStartBlock <= 0 || + dev->internalEndBlock <= 0 || +- dev->internalEndBlock <= (dev->internalStartBlock + dev->nReservedBlocks + 2) // otherwise it is too small +- ) { ++ dev->internalEndBlock <= (dev->internalStartBlock + dev->nReservedBlocks + 2)) { /* otherwise it is too small */ + T(YAFFS_TRACE_ALWAYS, + (TSTR +- ("yaffs: NAND geometry problems: chunk size %d, type is yaffs%s " +- TENDSTR), dev->nDataBytesPerChunk, dev->isYaffs2 ? "2" : "")); ++ ("yaffs: NAND geometry problems: chunk size %d, type is yaffs%s, inbandTags %d " ++ TENDSTR), dev->totalBytesPerChunk, dev->isYaffs2 ? "2" : "", dev->inbandTags)); + return YAFFS_FAIL; + } + +@@ -7070,6 +7158,12 @@ int yaffs_GutsInitialise(yaffs_Device * + return YAFFS_FAIL; + } + ++ /* Sort out space for inband tags, if required */ ++ if (dev->inbandTags) ++ dev->nDataBytesPerChunk = dev->totalBytesPerChunk - sizeof(yaffs_PackedTags2TagsPart); ++ else ++ dev->nDataBytesPerChunk = dev->totalBytesPerChunk; ++ + /* Got the right mix of functions? */ + if (!yaffs_CheckDevFunctions(dev)) { + /* Function missing */ +@@ -7097,31 +7191,18 @@ int yaffs_GutsInitialise(yaffs_Device * + + dev->isMounted = 1; + +- +- + /* OK now calculate a few things for the device */ + + /* + * Calculate all the chunk size manipulation numbers: + */ +- /* Start off assuming it is a power of 2 */ +- dev->chunkShift = ShiftDiv(dev->nDataBytesPerChunk); +- dev->chunkMask = (1<chunkShift) - 1; +- +- if(dev->nDataBytesPerChunk == (dev->chunkMask + 1)){ +- /* Yes it is a power of 2, disable crumbs */ +- dev->crumbMask = 0; +- dev->crumbShift = 0; +- dev->crumbsPerChunk = 0; +- } else { +- /* Not a power of 2, use crumbs instead */ +- dev->crumbShift = ShiftDiv(sizeof(yaffs_PackedTags2TagsPart)); +- dev->crumbMask = (1<crumbShift)-1; +- dev->crumbsPerChunk = dev->nDataBytesPerChunk/(1 << dev->crumbShift); +- dev->chunkShift = 0; +- dev->chunkMask = 0; +- } +- ++ x = dev->nDataBytesPerChunk; ++ /* We always use dev->chunkShift and dev->chunkDiv */ ++ dev->chunkShift = Shifts(x); ++ x >>= dev->chunkShift; ++ dev->chunkDiv = x; ++ /* We only use chunk mask if chunkDiv is 1 */ ++ dev->chunkMask = (1<chunkShift) - 1; + + /* + * Calculate chunkGroupBits. +@@ -7133,16 +7214,15 @@ int yaffs_GutsInitialise(yaffs_Device * + bits = ShiftsGE(x); + + /* Set up tnode width if wide tnodes are enabled. */ +- if(!dev->wideTnodesDisabled){ ++ if (!dev->wideTnodesDisabled) { + /* bits must be even so that we end up with 32-bit words */ +- if(bits & 1) ++ if (bits & 1) + bits++; +- if(bits < 16) ++ if (bits < 16) + dev->tnodeWidth = 16; + else + dev->tnodeWidth = bits; +- } +- else ++ } else + dev->tnodeWidth = 16; + + dev->tnodeMask = (1<tnodeWidth)-1; +@@ -7193,7 +7273,7 @@ int yaffs_GutsInitialise(yaffs_Device * + dev->hasPendingPrioritisedGCs = 1; /* Assume the worst for now, will get fixed on first GC */ + + /* Initialise temporary buffers and caches. */ +- if(!yaffs_InitialiseTempBuffers(dev)) ++ if (!yaffs_InitialiseTempBuffers(dev)) + init_failed = 1; + + dev->srCache = NULL; +@@ -7203,25 +7283,26 @@ int yaffs_GutsInitialise(yaffs_Device * + if (!init_failed && + dev->nShortOpCaches > 0) { + int i; +- __u8 *buf; ++ void *buf; + int srCacheBytes = dev->nShortOpCaches * sizeof(yaffs_ChunkCache); + +- if (dev->nShortOpCaches > YAFFS_MAX_SHORT_OP_CACHES) { ++ if (dev->nShortOpCaches > YAFFS_MAX_SHORT_OP_CACHES) + dev->nShortOpCaches = YAFFS_MAX_SHORT_OP_CACHES; +- } + +- buf = dev->srCache = YMALLOC(srCacheBytes); ++ dev->srCache = YMALLOC(srCacheBytes); + +- if(dev->srCache) +- memset(dev->srCache,0,srCacheBytes); ++ buf = (__u8 *) dev->srCache; ++ ++ if (dev->srCache) ++ memset(dev->srCache, 0, srCacheBytes); + + for (i = 0; i < dev->nShortOpCaches && buf; i++) { + dev->srCache[i].object = NULL; + dev->srCache[i].lastUse = 0; + dev->srCache[i].dirty = 0; +- dev->srCache[i].data = buf = YMALLOC_DMA(dev->nDataBytesPerChunk); ++ dev->srCache[i].data = buf = YMALLOC_DMA(dev->totalBytesPerChunk); + } +- if(!buf) ++ if (!buf) + init_failed = 1; + + dev->srLastUse = 0; +@@ -7229,29 +7310,30 @@ int yaffs_GutsInitialise(yaffs_Device * + + dev->cacheHits = 0; + +- if(!init_failed){ ++ if (!init_failed) { + dev->gcCleanupList = YMALLOC(dev->nChunksPerBlock * sizeof(__u32)); +- if(!dev->gcCleanupList) ++ if (!dev->gcCleanupList) + init_failed = 1; + } + +- if (dev->isYaffs2) { ++ if (dev->isYaffs2) + dev->useHeaderFileSize = 1; +- } +- if(!init_failed && !yaffs_InitialiseBlocks(dev)) ++ ++ if (!init_failed && !yaffs_InitialiseBlocks(dev)) + init_failed = 1; + + yaffs_InitialiseTnodes(dev); + yaffs_InitialiseObjects(dev); + +- if(!init_failed && !yaffs_CreateInitialDirectories(dev)) ++ if (!init_failed && !yaffs_CreateInitialDirectories(dev)) + init_failed = 1; + + +- if(!init_failed){ ++ if (!init_failed) { + /* Now scan the flash. */ + if (dev->isYaffs2) { +- if(yaffs_CheckpointRestore(dev)) { ++ if (yaffs_CheckpointRestore(dev)) { ++ yaffs_CheckObjectDetailsLoaded(dev->rootDir); + T(YAFFS_TRACE_ALWAYS, + (TSTR("yaffs: restored from checkpoint" TENDSTR))); + } else { +@@ -7273,24 +7355,25 @@ int yaffs_GutsInitialise(yaffs_Device * + dev->nBackgroundDeletions = 0; + dev->oldestDirtySequence = 0; + +- if(!init_failed && !yaffs_InitialiseBlocks(dev)) ++ if (!init_failed && !yaffs_InitialiseBlocks(dev)) + init_failed = 1; + + yaffs_InitialiseTnodes(dev); + yaffs_InitialiseObjects(dev); + +- if(!init_failed && !yaffs_CreateInitialDirectories(dev)) ++ if (!init_failed && !yaffs_CreateInitialDirectories(dev)) + init_failed = 1; + +- if(!init_failed && !yaffs_ScanBackwards(dev)) ++ if (!init_failed && !yaffs_ScanBackwards(dev)) + init_failed = 1; + } +- }else +- if(!yaffs_Scan(dev)) ++ } else if (!yaffs_Scan(dev)) + init_failed = 1; ++ ++ yaffs_StripDeletedObjects(dev); + } + +- if(init_failed){ ++ if (init_failed) { + /* Clean up the mess */ + T(YAFFS_TRACE_TRACING, + (TSTR("yaffs: yaffs_GutsInitialise() aborted.\n" TENDSTR))); +@@ -7318,7 +7401,7 @@ int yaffs_GutsInitialise(yaffs_Device * + + } + +-void yaffs_Deinitialise(yaffs_Device * dev) ++void yaffs_Deinitialise(yaffs_Device *dev) + { + if (dev->isMounted) { + int i; +@@ -7330,7 +7413,7 @@ void yaffs_Deinitialise(yaffs_Device * d + dev->srCache) { + + for (i = 0; i < dev->nShortOpCaches; i++) { +- if(dev->srCache[i].data) ++ if (dev->srCache[i].data) + YFREE(dev->srCache[i].data); + dev->srCache[i].data = NULL; + } +@@ -7341,16 +7424,17 @@ void yaffs_Deinitialise(yaffs_Device * d + + YFREE(dev->gcCleanupList); + +- for (i = 0; i < YAFFS_N_TEMP_BUFFERS; i++) { ++ for (i = 0; i < YAFFS_N_TEMP_BUFFERS; i++) + YFREE(dev->tempBuffer[i].buffer); +- } + + dev->isMounted = 0; +- } + ++ if (dev->deinitialiseNAND) ++ dev->deinitialiseNAND(dev); ++ } + } + +-static int yaffs_CountFreeChunks(yaffs_Device * dev) ++static int yaffs_CountFreeChunks(yaffs_Device *dev) + { + int nFree; + int b; +@@ -7358,7 +7442,7 @@ static int yaffs_CountFreeChunks(yaffs_D + yaffs_BlockInfo *blk; + + for (nFree = 0, b = dev->internalStartBlock; b <= dev->internalEndBlock; +- b++) { ++ b++) { + blk = yaffs_GetBlockInfo(dev, b); + + switch (blk->blockState) { +@@ -7373,19 +7457,19 @@ static int yaffs_CountFreeChunks(yaffs_D + default: + break; + } +- + } + + return nFree; + } + +-int yaffs_GetNumberOfFreeChunks(yaffs_Device * dev) ++int yaffs_GetNumberOfFreeChunks(yaffs_Device *dev) + { + /* This is what we report to the outside world */ + + int nFree; + int nDirtyCacheChunks; + int blocksForCheckpoint; ++ int i; + + #if 1 + nFree = dev->nFreeChunks; +@@ -7397,12 +7481,9 @@ int yaffs_GetNumberOfFreeChunks(yaffs_De + + /* Now count the number of dirty chunks in the cache and subtract those */ + +- { +- int i; +- for (nDirtyCacheChunks = 0, i = 0; i < dev->nShortOpCaches; i++) { +- if (dev->srCache[i].dirty) +- nDirtyCacheChunks++; +- } ++ for (nDirtyCacheChunks = 0, i = 0; i < dev->nShortOpCaches; i++) { ++ if (dev->srCache[i].dirty) ++ nDirtyCacheChunks++; + } + + nFree -= nDirtyCacheChunks; +@@ -7410,8 +7491,8 @@ int yaffs_GetNumberOfFreeChunks(yaffs_De + nFree -= ((dev->nReservedBlocks + 1) * dev->nChunksPerBlock); + + /* Now we figure out how much to reserve for the checkpoint and report that... */ +- blocksForCheckpoint = dev->nCheckpointReservedBlocks - dev->blocksInCheckpoint; +- if(blocksForCheckpoint < 0) ++ blocksForCheckpoint = yaffs_CalcCheckpointBlocksRequired(dev) - dev->blocksInCheckpoint; ++ if (blocksForCheckpoint < 0) + blocksForCheckpoint = 0; + + nFree -= (blocksForCheckpoint * dev->nChunksPerBlock); +@@ -7425,12 +7506,12 @@ int yaffs_GetNumberOfFreeChunks(yaffs_De + + static int yaffs_freeVerificationFailures; + +-static void yaffs_VerifyFreeChunks(yaffs_Device * dev) ++static void yaffs_VerifyFreeChunks(yaffs_Device *dev) + { + int counted; + int difference; + +- if(yaffs_SkipVerification(dev)) ++ if (yaffs_SkipVerification(dev)) + return; + + counted = yaffs_CountFreeChunks(dev); +@@ -7447,23 +7528,25 @@ static void yaffs_VerifyFreeChunks(yaffs + + /*---------------------------------------- YAFFS test code ----------------------*/ + +-#define yaffs_CheckStruct(structure,syze, name) \ +- if(sizeof(structure) != syze) \ +- { \ +- T(YAFFS_TRACE_ALWAYS,(TSTR("%s should be %d but is %d\n" TENDSTR),\ +- name,syze,sizeof(structure))); \ +- return YAFFS_FAIL; \ +- } ++#define yaffs_CheckStruct(structure, syze, name) \ ++ do { \ ++ if (sizeof(structure) != syze) { \ ++ T(YAFFS_TRACE_ALWAYS, (TSTR("%s should be %d but is %d\n" TENDSTR),\ ++ name, syze, sizeof(structure))); \ ++ return YAFFS_FAIL; \ ++ } \ ++ } while (0) + + static int yaffs_CheckStructures(void) + { +-/* yaffs_CheckStruct(yaffs_Tags,8,"yaffs_Tags") */ +-/* yaffs_CheckStruct(yaffs_TagsUnion,8,"yaffs_TagsUnion") */ +-/* yaffs_CheckStruct(yaffs_Spare,16,"yaffs_Spare") */ ++/* yaffs_CheckStruct(yaffs_Tags,8,"yaffs_Tags"); */ ++/* yaffs_CheckStruct(yaffs_TagsUnion,8,"yaffs_TagsUnion"); */ ++/* yaffs_CheckStruct(yaffs_Spare,16,"yaffs_Spare"); */ + #ifndef CONFIG_YAFFS_TNODE_LIST_DEBUG +- yaffs_CheckStruct(yaffs_Tnode, 2 * YAFFS_NTNODES_LEVEL0, "yaffs_Tnode") ++ yaffs_CheckStruct(yaffs_Tnode, 2 * YAFFS_NTNODES_LEVEL0, "yaffs_Tnode"); + #endif +- yaffs_CheckStruct(yaffs_ObjectHeader, 512, "yaffs_ObjectHeader") +- +- return YAFFS_OK; ++#ifndef CONFIG_YAFFS_WINCE ++ yaffs_CheckStruct(yaffs_ObjectHeader, 512, "yaffs_ObjectHeader"); ++#endif ++ return YAFFS_OK; + } +--- a/fs/yaffs2/yaffs_guts.h ++++ b/fs/yaffs2/yaffs_guts.h +@@ -90,7 +90,7 @@ + + #define YAFFS_MAX_SHORT_OP_CACHES 20 + +-#define YAFFS_N_TEMP_BUFFERS 4 ++#define YAFFS_N_TEMP_BUFFERS 6 + + /* We limit the number attempts at sucessfully saving a chunk of data. + * Small-page devices have 32 pages per block; large-page devices have 64. +@@ -108,6 +108,9 @@ + #define YAFFS_LOWEST_SEQUENCE_NUMBER 0x00001000 + #define YAFFS_HIGHEST_SEQUENCE_NUMBER 0xEFFFFF00 + ++/* Special sequence number for bad block that failed to be marked bad */ ++#define YAFFS_SEQUENCE_BAD_BLOCK 0xFFFF0000 ++ + /* ChunkCache is used for short read/write operations.*/ + typedef struct { + struct yaffs_ObjectStruct *object; +@@ -134,11 +137,10 @@ typedef struct { + typedef struct { + unsigned chunkId:20; + unsigned serialNumber:2; +- unsigned byteCount:10; ++ unsigned byteCountLSB:10; + unsigned objectId:18; + unsigned ecc:12; +- unsigned unusedStuff:2; +- ++ unsigned byteCountMSB:2; + } yaffs_Tags; + + typedef union { +@@ -277,13 +279,13 @@ typedef struct { + + int softDeletions:10; /* number of soft deleted pages */ + int pagesInUse:10; /* number of pages in use */ +- yaffs_BlockState blockState:4; /* One of the above block states */ ++ unsigned blockState:4; /* One of the above block states. NB use unsigned because enum is sometimes an int */ + __u32 needsRetiring:1; /* Data has failed on this block, need to get valid data off */ +- /* and retire the block. */ +- __u32 skipErasedCheck: 1; /* If this is set we can skip the erased check on this block */ +- __u32 gcPrioritise: 1; /* An ECC check or blank check has failed on this block. ++ /* and retire the block. */ ++ __u32 skipErasedCheck:1; /* If this is set we can skip the erased check on this block */ ++ __u32 gcPrioritise:1; /* An ECC check or blank check has failed on this block. + It should be prioritised for GC */ +- __u32 chunkErrorStrikes:3; /* How many times we've had ecc etc failures on this block and tried to reuse it */ ++ __u32 chunkErrorStrikes:3; /* How many times we've had ecc etc failures on this block and tried to reuse it */ + + #ifdef CONFIG_YAFFS_YAFFS2 + __u32 hasShrinkHeader:1; /* This block has at least one shrink object header */ +@@ -300,11 +302,11 @@ typedef struct { + + /* Apply to everything */ + int parentObjectId; +- __u16 sum__NoLongerUsed; /* checksum of name. No longer used */ ++ __u16 sum__NoLongerUsed; /* checksum of name. No longer used */ + YCHAR name[YAFFS_MAX_NAME_LENGTH + 1]; + +- /* Thes following apply to directories, files, symlinks - not hard links */ +- __u32 yst_mode; /* protection */ ++ /* The following apply to directories, files, symlinks - not hard links */ ++ __u32 yst_mode; /* protection */ + + #ifdef CONFIG_YAFFS_WINCE + __u32 notForWinCE[5]; +@@ -331,11 +333,14 @@ typedef struct { + __u32 win_ctime[2]; + __u32 win_atime[2]; + __u32 win_mtime[2]; +- __u32 roomToGrow[4]; + #else +- __u32 roomToGrow[10]; ++ __u32 roomToGrow[6]; ++ + #endif ++ __u32 inbandShadowsObject; ++ __u32 inbandIsShrink; + ++ __u32 reservedSpace[2]; + int shadowsObject; /* This object header shadows the specified object if > 0 */ + + /* isShrink applies to object headers written when we shrink the file (ie resize) */ +@@ -381,7 +386,7 @@ typedef struct { + } yaffs_FileStructure; + + typedef struct { +- struct list_head children; /* list of child links */ ++ struct ylist_head children; /* list of child links */ + } yaffs_DirectoryStructure; + + typedef struct { +@@ -418,23 +423,24 @@ struct yaffs_ObjectStruct { + * still in the inode cache. Free of object is defered. + * until the inode is released. + */ ++ __u8 beingCreated:1; /* This object is still being created so skip some checks. */ + + __u8 serial; /* serial number of chunk in NAND. Cached here */ + __u16 sum; /* sum of the name to speed searching */ + +- struct yaffs_DeviceStruct *myDev; /* The device I'm on */ ++ struct yaffs_DeviceStruct *myDev; /* The device I'm on */ + +- struct list_head hashLink; /* list of objects in this hash bucket */ ++ struct ylist_head hashLink; /* list of objects in this hash bucket */ + +- struct list_head hardLinks; /* all the equivalent hard linked objects */ ++ struct ylist_head hardLinks; /* all the equivalent hard linked objects */ + + /* directory structure stuff */ + /* also used for linking up the free list */ + struct yaffs_ObjectStruct *parent; +- struct list_head siblings; ++ struct ylist_head siblings; + + /* Where's my object header in NAND? */ +- int chunkId; ++ int hdrChunk; + + int nDataChunks; /* Number of data chunks attached to the file. */ + +@@ -485,7 +491,7 @@ struct yaffs_ObjectList_struct { + typedef struct yaffs_ObjectList_struct yaffs_ObjectList; + + typedef struct { +- struct list_head list; ++ struct ylist_head list; + int count; + } yaffs_ObjectBucket; + +@@ -495,11 +501,10 @@ typedef struct { + */ + + typedef struct { +- int structType; ++ int structType; + __u32 objectId; + __u32 parentId; +- int chunkId; +- ++ int hdrChunk; + yaffs_ObjectType variantType:3; + __u8 deleted:1; + __u8 softDeleted:1; +@@ -511,8 +516,7 @@ typedef struct { + + int nDataChunks; + __u32 fileSizeOrEquivalentObjectId; +- +-}yaffs_CheckpointObject; ++} yaffs_CheckpointObject; + + /*--------------------- Temporary buffers ---------------- + * +@@ -528,13 +532,13 @@ typedef struct { + /*----------------- Device ---------------------------------*/ + + struct yaffs_DeviceStruct { +- struct list_head devList; ++ struct ylist_head devList; + const char *name; + + /* Entry parameters set up way early. Yaffs sets up the rest.*/ + int nDataBytesPerChunk; /* Should be a power of 2 >= 512 */ + int nChunksPerBlock; /* does not need to be a power of 2 */ +- int nBytesPerSpare; /* spare area size */ ++ int spareBytesPerChunk; /* spare area size */ + int startBlock; /* Start block we're allowed to use */ + int endBlock; /* End block we're allowed to use */ + int nReservedBlocks; /* We want this tuneable so that we can reduce */ +@@ -544,9 +548,7 @@ struct yaffs_DeviceStruct { + /* Stuff used by the shared space checkpointing mechanism */ + /* If this value is zero, then this mechanism is disabled */ + +- int nCheckpointReservedBlocks; /* Blocks to reserve for checkpoint data */ +- +- ++/* int nCheckpointReservedBlocks; */ /* Blocks to reserve for checkpoint data */ + + + int nShortOpCaches; /* If <= 0, then short op caching is disabled, else +@@ -560,30 +562,31 @@ struct yaffs_DeviceStruct { + void *genericDevice; /* Pointer to device context + * On an mtd this holds the mtd pointer. + */ +- void *superBlock; ++ void *superBlock; + + /* NAND access functions (Must be set before calling YAFFS)*/ + +- int (*writeChunkToNAND) (struct yaffs_DeviceStruct * dev, +- int chunkInNAND, const __u8 * data, +- const yaffs_Spare * spare); +- int (*readChunkFromNAND) (struct yaffs_DeviceStruct * dev, +- int chunkInNAND, __u8 * data, +- yaffs_Spare * spare); +- int (*eraseBlockInNAND) (struct yaffs_DeviceStruct * dev, +- int blockInNAND); +- int (*initialiseNAND) (struct yaffs_DeviceStruct * dev); ++ int (*writeChunkToNAND) (struct yaffs_DeviceStruct *dev, ++ int chunkInNAND, const __u8 *data, ++ const yaffs_Spare *spare); ++ int (*readChunkFromNAND) (struct yaffs_DeviceStruct *dev, ++ int chunkInNAND, __u8 *data, ++ yaffs_Spare *spare); ++ int (*eraseBlockInNAND) (struct yaffs_DeviceStruct *dev, ++ int blockInNAND); ++ int (*initialiseNAND) (struct yaffs_DeviceStruct *dev); ++ int (*deinitialiseNAND) (struct yaffs_DeviceStruct *dev); + + #ifdef CONFIG_YAFFS_YAFFS2 +- int (*writeChunkWithTagsToNAND) (struct yaffs_DeviceStruct * dev, +- int chunkInNAND, const __u8 * data, +- const yaffs_ExtendedTags * tags); +- int (*readChunkWithTagsFromNAND) (struct yaffs_DeviceStruct * dev, +- int chunkInNAND, __u8 * data, +- yaffs_ExtendedTags * tags); +- int (*markNANDBlockBad) (struct yaffs_DeviceStruct * dev, int blockNo); +- int (*queryNANDBlock) (struct yaffs_DeviceStruct * dev, int blockNo, +- yaffs_BlockState * state, int *sequenceNumber); ++ int (*writeChunkWithTagsToNAND) (struct yaffs_DeviceStruct *dev, ++ int chunkInNAND, const __u8 *data, ++ const yaffs_ExtendedTags *tags); ++ int (*readChunkWithTagsFromNAND) (struct yaffs_DeviceStruct *dev, ++ int chunkInNAND, __u8 *data, ++ yaffs_ExtendedTags *tags); ++ int (*markNANDBlockBad) (struct yaffs_DeviceStruct *dev, int blockNo); ++ int (*queryNANDBlock) (struct yaffs_DeviceStruct *dev, int blockNo, ++ yaffs_BlockState *state, __u32 *sequenceNumber); + #endif + + int isYaffs2; +@@ -595,10 +598,12 @@ struct yaffs_DeviceStruct { + void (*removeObjectCallback)(struct yaffs_ObjectStruct *obj); + + /* Callback to mark the superblock dirsty */ +- void (*markSuperBlockDirty)(void * superblock); ++ void (*markSuperBlockDirty)(void *superblock); + + int wideTnodesDisabled; /* Set to disable wide tnodes */ + ++ YCHAR *pathDividers; /* String of legal path dividers */ ++ + + /* End of stuff that must be set before initialisation. */ + +@@ -615,16 +620,14 @@ struct yaffs_DeviceStruct { + __u32 tnodeWidth; + __u32 tnodeMask; + +- /* Stuff to support various file offses to chunk/offset translations */ +- /* "Crumbs" for nDataBytesPerChunk not being a power of 2 */ +- __u32 crumbMask; +- __u32 crumbShift; +- __u32 crumbsPerChunk; +- +- /* Straight shifting for nDataBytesPerChunk being a power of 2 */ +- __u32 chunkShift; +- __u32 chunkMask; +- ++ /* Stuff for figuring out file offset to chunk conversions */ ++ __u32 chunkShift; /* Shift value */ ++ __u32 chunkDiv; /* Divisor after shifting: 1 for power-of-2 sizes */ ++ __u32 chunkMask; /* Mask to use for power-of-2 case */ ++ ++ /* Stuff to handle inband tags */ ++ int inbandTags; ++ __u32 totalBytesPerChunk; + + #ifdef __KERNEL__ + +@@ -633,7 +636,7 @@ struct yaffs_DeviceStruct { + __u8 *spareBuffer; /* For mtdif2 use. Don't know the size of the buffer + * at compile time so we have to allocate it. + */ +- void (*putSuperFunc) (struct super_block * sb); ++ void (*putSuperFunc) (struct super_block *sb); + #endif + + int isMounted; +@@ -663,6 +666,8 @@ struct yaffs_DeviceStruct { + __u32 checkpointSum; + __u32 checkpointXor; + ++ int nCheckpointBlocksRequired; /* Number of blocks needed to store current checkpoint set */ ++ + /* Block Info */ + yaffs_BlockInfo *blockInfo; + __u8 *chunkBits; /* bitmap of chunks in use */ +@@ -684,11 +689,15 @@ struct yaffs_DeviceStruct { + yaffs_TnodeList *allocatedTnodeList; + + int isDoingGC; ++ int gcBlock; ++ int gcChunk; + + int nObjectsCreated; + yaffs_Object *freeObjects; + int nFreeObjects; + ++ int nHardLinks; ++ + yaffs_ObjectList *allocatedObjectList; + + yaffs_ObjectBucket objectBucket[YAFFS_NOBJECT_BUCKETS]; +@@ -745,8 +754,10 @@ struct yaffs_DeviceStruct { + int nBackgroundDeletions; /* Count of background deletions. */ + + ++ /* Temporary buffer management */ + yaffs_TempBuffer tempBuffer[YAFFS_N_TEMP_BUFFERS]; + int maxTemp; ++ int tempInUse; + int unmanagedTempAllocations; + int unmanagedTempDeallocations; + +@@ -758,9 +769,9 @@ struct yaffs_DeviceStruct { + + typedef struct yaffs_DeviceStruct yaffs_Device; + +-/* The static layout of bllock usage etc is stored in the super block header */ ++/* The static layout of block usage etc is stored in the super block header */ + typedef struct { +- int StructType; ++ int StructType; + int version; + int checkpointStartBlock; + int checkpointEndBlock; +@@ -773,7 +784,7 @@ typedef struct { + * must be preserved over unmount/mount cycles. + */ + typedef struct { +- int structType; ++ int structType; + int nErasedBlocks; + int allocationBlock; /* Current block being allocated off */ + __u32 allocationPage; +@@ -791,57 +802,45 @@ typedef struct { + + + typedef struct { +- int structType; +- __u32 magic; +- __u32 version; +- __u32 head; ++ int structType; ++ __u32 magic; ++ __u32 version; ++ __u32 head; + } yaffs_CheckpointValidity; + +-/* Function to manipulate block info */ +-static Y_INLINE yaffs_BlockInfo *yaffs_GetBlockInfo(yaffs_Device * dev, int blk) +-{ +- if (blk < dev->internalStartBlock || blk > dev->internalEndBlock) { +- T(YAFFS_TRACE_ERROR, +- (TSTR +- ("**>> yaffs: getBlockInfo block %d is not valid" TENDSTR), +- blk)); +- YBUG(); +- } +- return &dev->blockInfo[blk - dev->internalStartBlock]; +-} + + /*----------------------- YAFFS Functions -----------------------*/ + +-int yaffs_GutsInitialise(yaffs_Device * dev); +-void yaffs_Deinitialise(yaffs_Device * dev); ++int yaffs_GutsInitialise(yaffs_Device *dev); ++void yaffs_Deinitialise(yaffs_Device *dev); + +-int yaffs_GetNumberOfFreeChunks(yaffs_Device * dev); ++int yaffs_GetNumberOfFreeChunks(yaffs_Device *dev); + +-int yaffs_RenameObject(yaffs_Object * oldDir, const YCHAR * oldName, +- yaffs_Object * newDir, const YCHAR * newName); ++int yaffs_RenameObject(yaffs_Object *oldDir, const YCHAR *oldName, ++ yaffs_Object *newDir, const YCHAR *newName); + +-int yaffs_Unlink(yaffs_Object * dir, const YCHAR * name); +-int yaffs_DeleteFile(yaffs_Object * obj); ++int yaffs_Unlink(yaffs_Object *dir, const YCHAR *name); ++int yaffs_DeleteObject(yaffs_Object *obj); + +-int yaffs_GetObjectName(yaffs_Object * obj, YCHAR * name, int buffSize); +-int yaffs_GetObjectFileLength(yaffs_Object * obj); +-int yaffs_GetObjectInode(yaffs_Object * obj); +-unsigned yaffs_GetObjectType(yaffs_Object * obj); +-int yaffs_GetObjectLinkCount(yaffs_Object * obj); ++int yaffs_GetObjectName(yaffs_Object *obj, YCHAR *name, int buffSize); ++int yaffs_GetObjectFileLength(yaffs_Object *obj); ++int yaffs_GetObjectInode(yaffs_Object *obj); ++unsigned yaffs_GetObjectType(yaffs_Object *obj); ++int yaffs_GetObjectLinkCount(yaffs_Object *obj); + +-int yaffs_SetAttributes(yaffs_Object * obj, struct iattr *attr); +-int yaffs_GetAttributes(yaffs_Object * obj, struct iattr *attr); ++int yaffs_SetAttributes(yaffs_Object *obj, struct iattr *attr); ++int yaffs_GetAttributes(yaffs_Object *obj, struct iattr *attr); + + /* File operations */ +-int yaffs_ReadDataFromFile(yaffs_Object * obj, __u8 * buffer, loff_t offset, +- int nBytes); +-int yaffs_WriteDataToFile(yaffs_Object * obj, const __u8 * buffer, loff_t offset, +- int nBytes, int writeThrough); +-int yaffs_ResizeFile(yaffs_Object * obj, loff_t newSize); +- +-yaffs_Object *yaffs_MknodFile(yaffs_Object * parent, const YCHAR * name, +- __u32 mode, __u32 uid, __u32 gid); +-int yaffs_FlushFile(yaffs_Object * obj, int updateTime); ++int yaffs_ReadDataFromFile(yaffs_Object *obj, __u8 *buffer, loff_t offset, ++ int nBytes); ++int yaffs_WriteDataToFile(yaffs_Object *obj, const __u8 *buffer, loff_t offset, ++ int nBytes, int writeThrough); ++int yaffs_ResizeFile(yaffs_Object *obj, loff_t newSize); ++ ++yaffs_Object *yaffs_MknodFile(yaffs_Object *parent, const YCHAR *name, ++ __u32 mode, __u32 uid, __u32 gid); ++int yaffs_FlushFile(yaffs_Object *obj, int updateTime); + + /* Flushing and checkpointing */ + void yaffs_FlushEntireDeviceCache(yaffs_Device *dev); +@@ -850,33 +849,33 @@ int yaffs_CheckpointSave(yaffs_Device *d + int yaffs_CheckpointRestore(yaffs_Device *dev); + + /* Directory operations */ +-yaffs_Object *yaffs_MknodDirectory(yaffs_Object * parent, const YCHAR * name, +- __u32 mode, __u32 uid, __u32 gid); +-yaffs_Object *yaffs_FindObjectByName(yaffs_Object * theDir, const YCHAR * name); +-int yaffs_ApplyToDirectoryChildren(yaffs_Object * theDir, ++yaffs_Object *yaffs_MknodDirectory(yaffs_Object *parent, const YCHAR *name, ++ __u32 mode, __u32 uid, __u32 gid); ++yaffs_Object *yaffs_FindObjectByName(yaffs_Object *theDir, const YCHAR *name); ++int yaffs_ApplyToDirectoryChildren(yaffs_Object *theDir, + int (*fn) (yaffs_Object *)); + +-yaffs_Object *yaffs_FindObjectByNumber(yaffs_Device * dev, __u32 number); ++yaffs_Object *yaffs_FindObjectByNumber(yaffs_Device *dev, __u32 number); + + /* Link operations */ +-yaffs_Object *yaffs_Link(yaffs_Object * parent, const YCHAR * name, +- yaffs_Object * equivalentObject); ++yaffs_Object *yaffs_Link(yaffs_Object *parent, const YCHAR *name, ++ yaffs_Object *equivalentObject); + +-yaffs_Object *yaffs_GetEquivalentObject(yaffs_Object * obj); ++yaffs_Object *yaffs_GetEquivalentObject(yaffs_Object *obj); + + /* Symlink operations */ +-yaffs_Object *yaffs_MknodSymLink(yaffs_Object * parent, const YCHAR * name, ++yaffs_Object *yaffs_MknodSymLink(yaffs_Object *parent, const YCHAR *name, + __u32 mode, __u32 uid, __u32 gid, +- const YCHAR * alias); +-YCHAR *yaffs_GetSymlinkAlias(yaffs_Object * obj); ++ const YCHAR *alias); ++YCHAR *yaffs_GetSymlinkAlias(yaffs_Object *obj); + + /* Special inodes (fifos, sockets and devices) */ +-yaffs_Object *yaffs_MknodSpecial(yaffs_Object * parent, const YCHAR * name, ++yaffs_Object *yaffs_MknodSpecial(yaffs_Object *parent, const YCHAR *name, + __u32 mode, __u32 uid, __u32 gid, __u32 rdev); + + /* Special directories */ +-yaffs_Object *yaffs_Root(yaffs_Device * dev); +-yaffs_Object *yaffs_LostNFound(yaffs_Device * dev); ++yaffs_Object *yaffs_Root(yaffs_Device *dev); ++yaffs_Object *yaffs_LostNFound(yaffs_Device *dev); + + #ifdef CONFIG_YAFFS_WINCE + /* CONFIG_YAFFS_WINCE special stuff */ +@@ -885,18 +884,21 @@ void yfsd_WinFileTimeNow(__u32 target[2] + + #ifdef __KERNEL__ + +-void yaffs_HandleDeferedFree(yaffs_Object * obj); ++void yaffs_HandleDeferedFree(yaffs_Object *obj); + #endif + + /* Debug dump */ +-int yaffs_DumpObject(yaffs_Object * obj); ++int yaffs_DumpObject(yaffs_Object *obj); + +-void yaffs_GutsTest(yaffs_Device * dev); ++void yaffs_GutsTest(yaffs_Device *dev); + + /* A few useful functions */ +-void yaffs_InitialiseTags(yaffs_ExtendedTags * tags); +-void yaffs_DeleteChunk(yaffs_Device * dev, int chunkId, int markNAND, int lyn); +-int yaffs_CheckFF(__u8 * buffer, int nBytes); ++void yaffs_InitialiseTags(yaffs_ExtendedTags *tags); ++void yaffs_DeleteChunk(yaffs_Device *dev, int chunkId, int markNAND, int lyn); ++int yaffs_CheckFF(__u8 *buffer, int nBytes); + void yaffs_HandleChunkError(yaffs_Device *dev, yaffs_BlockInfo *bi); + ++__u8 *yaffs_GetTempBuffer(yaffs_Device *dev, int lineNo); ++void yaffs_ReleaseTempBuffer(yaffs_Device *dev, __u8 *buffer, int lineNo); ++ + #endif +--- a/fs/yaffs2/yaffs_mtdif1.c ++++ b/fs/yaffs2/yaffs_mtdif1.c +@@ -26,7 +26,7 @@ + #include "yportenv.h" + #include "yaffs_guts.h" + #include "yaffs_packedtags1.h" +-#include "yaffs_tagscompat.h" // for yaffs_CalcTagsECC ++#include "yaffs_tagscompat.h" /* for yaffs_CalcTagsECC */ + + #include "linux/kernel.h" + #include "linux/version.h" +@@ -34,9 +34,9 @@ + #include "linux/mtd/mtd.h" + + /* Don't compile this module if we don't have MTD's mtd_oob_ops interface */ +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (MTD_VERSION_CODE > MTD_VERSION(2, 6, 17)) + +-const char *yaffs_mtdif1_c_version = "$Id: yaffs_mtdif1.c,v 1.3 2007/05/15 20:16:11 ian Exp $"; ++const char *yaffs_mtdif1_c_version = "$Id: yaffs_mtdif1.c,v 1.10 2009-03-09 07:41:10 charles Exp $"; + + #ifndef CONFIG_YAFFS_9BYTE_TAGS + # define YTAG1_SIZE 8 +@@ -89,9 +89,9 @@ static struct nand_ecclayout nand_oob_16 + * Returns YAFFS_OK or YAFFS_FAIL. + */ + int nandmtd1_WriteChunkWithTagsToNAND(yaffs_Device *dev, +- int chunkInNAND, const __u8 * data, const yaffs_ExtendedTags * etags) ++ int chunkInNAND, const __u8 *data, const yaffs_ExtendedTags *etags) + { +- struct mtd_info * mtd = dev->genericDevice; ++ struct mtd_info *mtd = dev->genericDevice; + int chunkBytes = dev->nDataBytesPerChunk; + loff_t addr = ((loff_t)chunkInNAND) * chunkBytes; + struct mtd_oob_ops ops; +@@ -146,7 +146,7 @@ int nandmtd1_WriteChunkWithTagsToNAND(ya + + /* Return with empty ExtendedTags but add eccResult. + */ +-static int rettags(yaffs_ExtendedTags * etags, int eccResult, int retval) ++static int rettags(yaffs_ExtendedTags *etags, int eccResult, int retval) + { + if (etags) { + memset(etags, 0, sizeof(*etags)); +@@ -169,9 +169,9 @@ static int rettags(yaffs_ExtendedTags * + * Returns YAFFS_OK or YAFFS_FAIL. + */ + int nandmtd1_ReadChunkWithTagsFromNAND(yaffs_Device *dev, +- int chunkInNAND, __u8 * data, yaffs_ExtendedTags * etags) ++ int chunkInNAND, __u8 *data, yaffs_ExtendedTags *etags) + { +- struct mtd_info * mtd = dev->genericDevice; ++ struct mtd_info *mtd = dev->genericDevice; + int chunkBytes = dev->nDataBytesPerChunk; + loff_t addr = ((loff_t)chunkInNAND) * chunkBytes; + int eccres = YAFFS_ECC_RESULT_NO_ERROR; +@@ -189,7 +189,7 @@ int nandmtd1_ReadChunkWithTagsFromNAND(y + ops.datbuf = data; + ops.oobbuf = (__u8 *)&pt1; + +-#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,20)) ++#if (MTD_VERSION_CODE < MTD_VERSION(2, 6, 20)) + /* In MTD 2.6.18 to 2.6.19 nand_base.c:nand_do_read_oob() has a bug; + * help it out with ops.len = ops.ooblen when ops.datbuf == NULL. + */ +@@ -284,11 +284,11 @@ int nandmtd1_ReadChunkWithTagsFromNAND(y + */ + int nandmtd1_MarkNANDBlockBad(struct yaffs_DeviceStruct *dev, int blockNo) + { +- struct mtd_info * mtd = dev->genericDevice; ++ struct mtd_info *mtd = dev->genericDevice; + int blocksize = dev->nChunksPerBlock * dev->nDataBytesPerChunk; + int retval; + +- yaffs_trace(YAFFS_TRACE_BAD_BLOCKS, "marking block %d bad", blockNo); ++ yaffs_trace(YAFFS_TRACE_BAD_BLOCKS, "marking block %d bad\n", blockNo); + + retval = mtd->block_markbad(mtd, (loff_t)blocksize * blockNo); + return (retval) ? YAFFS_FAIL : YAFFS_OK; +@@ -298,7 +298,7 @@ int nandmtd1_MarkNANDBlockBad(struct yaf + * + * Returns YAFFS_OK or YAFFS_FAIL. + */ +-static int nandmtd1_TestPrerequists(struct mtd_info * mtd) ++static int nandmtd1_TestPrerequists(struct mtd_info *mtd) + { + /* 2.6.18 has mtd->ecclayout->oobavail */ + /* 2.6.21 has mtd->ecclayout->oobavail and mtd->oobavail */ +@@ -323,10 +323,11 @@ static int nandmtd1_TestPrerequists(stru + * Always returns YAFFS_OK. + */ + int nandmtd1_QueryNANDBlock(struct yaffs_DeviceStruct *dev, int blockNo, +- yaffs_BlockState * pState, int *pSequenceNumber) ++ yaffs_BlockState *pState, __u32 *pSequenceNumber) + { +- struct mtd_info * mtd = dev->genericDevice; ++ struct mtd_info *mtd = dev->genericDevice; + int chunkNo = blockNo * dev->nChunksPerBlock; ++ loff_t addr = (loff_t)chunkNo * dev->nDataBytesPerChunk; + yaffs_ExtendedTags etags; + int state = YAFFS_BLOCK_STATE_DEAD; + int seqnum = 0; +@@ -335,21 +336,22 @@ int nandmtd1_QueryNANDBlock(struct yaffs + /* We don't yet have a good place to test for MTD config prerequists. + * Do it here as we are called during the initial scan. + */ +- if (nandmtd1_TestPrerequists(mtd) != YAFFS_OK) { ++ if (nandmtd1_TestPrerequists(mtd) != YAFFS_OK) + return YAFFS_FAIL; +- } + + retval = nandmtd1_ReadChunkWithTagsFromNAND(dev, chunkNo, NULL, &etags); ++ etags.blockBad = (mtd->block_isbad)(mtd, addr); + if (etags.blockBad) { + yaffs_trace(YAFFS_TRACE_BAD_BLOCKS, +- "block %d is marked bad", blockNo); ++ "block %d is marked bad\n", blockNo); + state = YAFFS_BLOCK_STATE_DEAD; +- } +- else if (etags.chunkUsed) { ++ } else if (etags.eccResult != YAFFS_ECC_RESULT_NO_ERROR) { ++ /* bad tags, need to look more closely */ ++ state = YAFFS_BLOCK_STATE_NEEDS_SCANNING; ++ } else if (etags.chunkUsed) { + state = YAFFS_BLOCK_STATE_NEEDS_SCANNING; + seqnum = etags.sequenceNumber; +- } +- else { ++ } else { + state = YAFFS_BLOCK_STATE_EMPTY; + } + +@@ -360,4 +362,4 @@ int nandmtd1_QueryNANDBlock(struct yaffs + return YAFFS_OK; + } + +-#endif /*KERNEL_VERSION*/ ++#endif /*MTD_VERSION*/ +--- a/fs/yaffs2/yaffs_mtdif1.h ++++ b/fs/yaffs2/yaffs_mtdif1.h +@@ -14,15 +14,15 @@ + #ifndef __YAFFS_MTDIF1_H__ + #define __YAFFS_MTDIF1_H__ + +-int nandmtd1_WriteChunkWithTagsToNAND(yaffs_Device * dev, int chunkInNAND, +- const __u8 * data, const yaffs_ExtendedTags * tags); ++int nandmtd1_WriteChunkWithTagsToNAND(yaffs_Device *dev, int chunkInNAND, ++ const __u8 *data, const yaffs_ExtendedTags *tags); + +-int nandmtd1_ReadChunkWithTagsFromNAND(yaffs_Device * dev, int chunkInNAND, +- __u8 * data, yaffs_ExtendedTags * tags); ++int nandmtd1_ReadChunkWithTagsFromNAND(yaffs_Device *dev, int chunkInNAND, ++ __u8 *data, yaffs_ExtendedTags *tags); + + int nandmtd1_MarkNANDBlockBad(struct yaffs_DeviceStruct *dev, int blockNo); + + int nandmtd1_QueryNANDBlock(struct yaffs_DeviceStruct *dev, int blockNo, +- yaffs_BlockState * state, int *sequenceNumber); ++ yaffs_BlockState *state, __u32 *sequenceNumber); + + #endif +--- a/fs/yaffs2/yaffs_mtdif2.c ++++ b/fs/yaffs2/yaffs_mtdif2.c +@@ -14,7 +14,7 @@ + /* mtd interface for YAFFS2 */ + + const char *yaffs_mtdif2_c_version = +- "$Id: yaffs_mtdif2.c,v 1.17 2007-02-14 01:09:06 wookey Exp $"; ++ "$Id: yaffs_mtdif2.c,v 1.23 2009-03-06 17:20:53 wookey Exp $"; + + #include "yportenv.h" + +@@ -27,19 +27,23 @@ const char *yaffs_mtdif2_c_version = + + #include "yaffs_packedtags2.h" + +-int nandmtd2_WriteChunkWithTagsToNAND(yaffs_Device * dev, int chunkInNAND, +- const __u8 * data, +- const yaffs_ExtendedTags * tags) ++/* NB For use with inband tags.... ++ * We assume that the data buffer is of size totalBytersPerChunk so that we can also ++ * use it to load the tags. ++ */ ++int nandmtd2_WriteChunkWithTagsToNAND(yaffs_Device *dev, int chunkInNAND, ++ const __u8 *data, ++ const yaffs_ExtendedTags *tags) + { + struct mtd_info *mtd = (struct mtd_info *)(dev->genericDevice); +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (MTD_VERSION_CODE > MTD_VERSION(2, 6, 17)) + struct mtd_oob_ops ops; + #else + size_t dummy; + #endif + int retval = 0; + +- loff_t addr = ((loff_t) chunkInNAND) * dev->nDataBytesPerChunk; ++ loff_t addr; + + yaffs_PackedTags2 pt; + +@@ -48,46 +52,40 @@ int nandmtd2_WriteChunkWithTagsToNAND(ya + ("nandmtd2_WriteChunkWithTagsToNAND chunk %d data %p tags %p" + TENDSTR), chunkInNAND, data, tags)); + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) +- if (tags) +- yaffs_PackTags2(&pt, tags); +- else +- BUG(); /* both tags and data should always be present */ + +- if (data) { +- ops.mode = MTD_OOB_AUTO; +- ops.ooblen = sizeof(pt); +- ops.len = dev->nDataBytesPerChunk; +- ops.ooboffs = 0; +- ops.datbuf = (__u8 *)data; +- ops.oobbuf = (void *)&pt; +- retval = mtd->write_oob(mtd, addr, &ops); ++ addr = ((loff_t) chunkInNAND) * dev->totalBytesPerChunk; ++ ++ /* For yaffs2 writing there must be both data and tags. ++ * If we're using inband tags, then the tags are stuffed into ++ * the end of the data buffer. ++ */ ++ if (!data || !tags) ++ BUG(); ++ else if (dev->inbandTags) { ++ yaffs_PackedTags2TagsPart *pt2tp; ++ pt2tp = (yaffs_PackedTags2TagsPart *)(data + dev->nDataBytesPerChunk); ++ yaffs_PackTags2TagsPart(pt2tp, tags); + } else +- BUG(); /* both tags and data should always be present */ +-#else +- if (tags) { + yaffs_PackTags2(&pt, tags); +- } + +- if (data && tags) { +- if (dev->useNANDECC) +- retval = +- mtd->write_ecc(mtd, addr, dev->nDataBytesPerChunk, +- &dummy, data, (__u8 *) & pt, NULL); +- else +- retval = +- mtd->write_ecc(mtd, addr, dev->nDataBytesPerChunk, +- &dummy, data, (__u8 *) & pt, NULL); +- } else { +- if (data) +- retval = +- mtd->write(mtd, addr, dev->nDataBytesPerChunk, &dummy, +- data); +- if (tags) +- retval = +- mtd->write_oob(mtd, addr, mtd->oobsize, &dummy, +- (__u8 *) & pt); ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) ++ ops.mode = MTD_OOB_AUTO; ++ ops.ooblen = (dev->inbandTags) ? 0 : sizeof(pt); ++ ops.len = dev->totalBytesPerChunk; ++ ops.ooboffs = 0; ++ ops.datbuf = (__u8 *)data; ++ ops.oobbuf = (dev->inbandTags) ? NULL : (void *)&pt; ++ retval = mtd->write_oob(mtd, addr, &ops); + ++#else ++ if (!dev->inbandTags) { ++ retval = ++ mtd->write_ecc(mtd, addr, dev->nDataBytesPerChunk, ++ &dummy, data, (__u8 *) &pt, NULL); ++ } else { ++ retval = ++ mtd->write(mtd, addr, dev->totalBytesPerChunk, &dummy, ++ data); + } + #endif + +@@ -97,17 +95,18 @@ int nandmtd2_WriteChunkWithTagsToNAND(ya + return YAFFS_FAIL; + } + +-int nandmtd2_ReadChunkWithTagsFromNAND(yaffs_Device * dev, int chunkInNAND, +- __u8 * data, yaffs_ExtendedTags * tags) ++int nandmtd2_ReadChunkWithTagsFromNAND(yaffs_Device *dev, int chunkInNAND, ++ __u8 *data, yaffs_ExtendedTags *tags) + { + struct mtd_info *mtd = (struct mtd_info *)(dev->genericDevice); +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (MTD_VERSION_CODE > MTD_VERSION(2, 6, 17)) + struct mtd_oob_ops ops; + #endif + size_t dummy; + int retval = 0; ++ int localData = 0; + +- loff_t addr = ((loff_t) chunkInNAND) * dev->nDataBytesPerChunk; ++ loff_t addr = ((loff_t) chunkInNAND) * dev->totalBytesPerChunk; + + yaffs_PackedTags2 pt; + +@@ -116,9 +115,20 @@ int nandmtd2_ReadChunkWithTagsFromNAND(y + ("nandmtd2_ReadChunkWithTagsFromNAND chunk %d data %p tags %p" + TENDSTR), chunkInNAND, data, tags)); + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) +- if (data && !tags) +- retval = mtd->read(mtd, addr, dev->nDataBytesPerChunk, ++ if (dev->inbandTags) { ++ ++ if (!data) { ++ localData = 1; ++ data = yaffs_GetTempBuffer(dev, __LINE__); ++ } ++ ++ ++ } ++ ++ ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 17)) ++ if (dev->inbandTags || (data && !tags)) ++ retval = mtd->read(mtd, addr, dev->totalBytesPerChunk, + &dummy, data); + else if (tags) { + ops.mode = MTD_OOB_AUTO; +@@ -130,38 +140,42 @@ int nandmtd2_ReadChunkWithTagsFromNAND(y + retval = mtd->read_oob(mtd, addr, &ops); + } + #else +- if (data && tags) { +- if (dev->useNANDECC) { +- retval = +- mtd->read_ecc(mtd, addr, dev->nDataBytesPerChunk, +- &dummy, data, dev->spareBuffer, +- NULL); +- } else { +- retval = +- mtd->read_ecc(mtd, addr, dev->nDataBytesPerChunk, ++ if (!dev->inbandTags && data && tags) { ++ ++ retval = mtd->read_ecc(mtd, addr, dev->nDataBytesPerChunk, + &dummy, data, dev->spareBuffer, + NULL); +- } + } else { + if (data) + retval = + mtd->read(mtd, addr, dev->nDataBytesPerChunk, &dummy, + data); +- if (tags) ++ if (!dev->inbandTags && tags) + retval = + mtd->read_oob(mtd, addr, mtd->oobsize, &dummy, + dev->spareBuffer); + } + #endif + +- memcpy(&pt, dev->spareBuffer, sizeof(pt)); + +- if (tags) +- yaffs_UnpackTags2(tags, &pt); ++ if (dev->inbandTags) { ++ if (tags) { ++ yaffs_PackedTags2TagsPart *pt2tp; ++ pt2tp = (yaffs_PackedTags2TagsPart *)&data[dev->nDataBytesPerChunk]; ++ yaffs_UnpackTags2TagsPart(tags, pt2tp); ++ } ++ } else { ++ if (tags) { ++ memcpy(&pt, dev->spareBuffer, sizeof(pt)); ++ yaffs_UnpackTags2(tags, &pt); ++ } ++ } + +- if(tags && retval == -EBADMSG && tags->eccResult == YAFFS_ECC_RESULT_NO_ERROR) +- tags->eccResult = YAFFS_ECC_RESULT_UNFIXED; ++ if (localData) ++ yaffs_ReleaseTempBuffer(dev, data, __LINE__); + ++ if (tags && retval == -EBADMSG && tags->eccResult == YAFFS_ECC_RESULT_NO_ERROR) ++ tags->eccResult = YAFFS_ECC_RESULT_UNFIXED; + if (retval == 0) + return YAFFS_OK; + else +@@ -178,7 +192,7 @@ int nandmtd2_MarkNANDBlockBad(struct yaf + retval = + mtd->block_markbad(mtd, + blockNo * dev->nChunksPerBlock * +- dev->nDataBytesPerChunk); ++ dev->totalBytesPerChunk); + + if (retval == 0) + return YAFFS_OK; +@@ -188,7 +202,7 @@ int nandmtd2_MarkNANDBlockBad(struct yaf + } + + int nandmtd2_QueryNANDBlock(struct yaffs_DeviceStruct *dev, int blockNo, +- yaffs_BlockState * state, int *sequenceNumber) ++ yaffs_BlockState *state, __u32 *sequenceNumber) + { + struct mtd_info *mtd = (struct mtd_info *)(dev->genericDevice); + int retval; +@@ -198,7 +212,7 @@ int nandmtd2_QueryNANDBlock(struct yaffs + retval = + mtd->block_isbad(mtd, + blockNo * dev->nChunksPerBlock * +- dev->nDataBytesPerChunk); ++ dev->totalBytesPerChunk); + + if (retval) { + T(YAFFS_TRACE_MTD, (TSTR("block is bad" TENDSTR))); +--- a/fs/yaffs2/yaffs_mtdif2.h ++++ b/fs/yaffs2/yaffs_mtdif2.h +@@ -17,13 +17,13 @@ + #define __YAFFS_MTDIF2_H__ + + #include "yaffs_guts.h" +-int nandmtd2_WriteChunkWithTagsToNAND(yaffs_Device * dev, int chunkInNAND, +- const __u8 * data, +- const yaffs_ExtendedTags * tags); +-int nandmtd2_ReadChunkWithTagsFromNAND(yaffs_Device * dev, int chunkInNAND, +- __u8 * data, yaffs_ExtendedTags * tags); ++int nandmtd2_WriteChunkWithTagsToNAND(yaffs_Device *dev, int chunkInNAND, ++ const __u8 *data, ++ const yaffs_ExtendedTags *tags); ++int nandmtd2_ReadChunkWithTagsFromNAND(yaffs_Device *dev, int chunkInNAND, ++ __u8 *data, yaffs_ExtendedTags *tags); + int nandmtd2_MarkNANDBlockBad(struct yaffs_DeviceStruct *dev, int blockNo); + int nandmtd2_QueryNANDBlock(struct yaffs_DeviceStruct *dev, int blockNo, +- yaffs_BlockState * state, int *sequenceNumber); ++ yaffs_BlockState *state, __u32 *sequenceNumber); + + #endif +--- a/fs/yaffs2/yaffs_mtdif.c ++++ b/fs/yaffs2/yaffs_mtdif.c +@@ -12,7 +12,7 @@ + */ + + const char *yaffs_mtdif_c_version = +- "$Id: yaffs_mtdif.c,v 1.19 2007-02-14 01:09:06 wookey Exp $"; ++ "$Id: yaffs_mtdif.c,v 1.22 2009-03-06 17:20:51 wookey Exp $"; + + #include "yportenv.h" + +@@ -24,7 +24,7 @@ const char *yaffs_mtdif_c_version = + #include "linux/time.h" + #include "linux/mtd/nand.h" + +-#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,18)) ++#if (MTD_VERSION_CODE < MTD_VERSION(2, 6, 18)) + static struct nand_oobinfo yaffs_oobinfo = { + .useecc = 1, + .eccbytes = 6, +@@ -36,7 +36,7 @@ static struct nand_oobinfo yaffs_noeccin + }; + #endif + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (MTD_VERSION_CODE > MTD_VERSION(2, 6, 17)) + static inline void translate_spare2oob(const yaffs_Spare *spare, __u8 *oob) + { + oob[0] = spare->tagByte0; +@@ -45,8 +45,8 @@ static inline void translate_spare2oob(c + oob[3] = spare->tagByte3; + oob[4] = spare->tagByte4; + oob[5] = spare->tagByte5 & 0x3f; +- oob[5] |= spare->blockStatus == 'Y' ? 0: 0x80; +- oob[5] |= spare->pageStatus == 0 ? 0: 0x40; ++ oob[5] |= spare->blockStatus == 'Y' ? 0 : 0x80; ++ oob[5] |= spare->pageStatus == 0 ? 0 : 0x40; + oob[6] = spare->tagByte6; + oob[7] = spare->tagByte7; + } +@@ -71,18 +71,18 @@ static inline void translate_oob2spare(y + } + #endif + +-int nandmtd_WriteChunkToNAND(yaffs_Device * dev, int chunkInNAND, +- const __u8 * data, const yaffs_Spare * spare) ++int nandmtd_WriteChunkToNAND(yaffs_Device *dev, int chunkInNAND, ++ const __u8 *data, const yaffs_Spare *spare) + { + struct mtd_info *mtd = (struct mtd_info *)(dev->genericDevice); +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (MTD_VERSION_CODE > MTD_VERSION(2, 6, 17)) + struct mtd_oob_ops ops; + #endif + size_t dummy; + int retval = 0; + + loff_t addr = ((loff_t) chunkInNAND) * dev->nDataBytesPerChunk; +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (MTD_VERSION_CODE > MTD_VERSION(2, 6, 17)) + __u8 spareAsBytes[8]; /* OOB */ + + if (data && !spare) +@@ -135,18 +135,18 @@ int nandmtd_WriteChunkToNAND(yaffs_Devic + return YAFFS_FAIL; + } + +-int nandmtd_ReadChunkFromNAND(yaffs_Device * dev, int chunkInNAND, __u8 * data, +- yaffs_Spare * spare) ++int nandmtd_ReadChunkFromNAND(yaffs_Device *dev, int chunkInNAND, __u8 *data, ++ yaffs_Spare *spare) + { + struct mtd_info *mtd = (struct mtd_info *)(dev->genericDevice); +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (MTD_VERSION_CODE > MTD_VERSION(2, 6, 17)) + struct mtd_oob_ops ops; + #endif + size_t dummy; + int retval = 0; + + loff_t addr = ((loff_t) chunkInNAND) * dev->nDataBytesPerChunk; +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,6,17)) ++#if (MTD_VERSION_CODE > MTD_VERSION(2, 6, 17)) + __u8 spareAsBytes[8]; /* OOB */ + + if (data && !spare) +@@ -205,7 +205,7 @@ int nandmtd_ReadChunkFromNAND(yaffs_Devi + return YAFFS_FAIL; + } + +-int nandmtd_EraseBlockInNAND(yaffs_Device * dev, int blockNumber) ++int nandmtd_EraseBlockInNAND(yaffs_Device *dev, int blockNumber) + { + struct mtd_info *mtd = (struct mtd_info *)(dev->genericDevice); + __u32 addr = +@@ -234,7 +234,7 @@ int nandmtd_EraseBlockInNAND(yaffs_Devic + return YAFFS_FAIL; + } + +-int nandmtd_InitialiseNAND(yaffs_Device * dev) ++int nandmtd_InitialiseNAND(yaffs_Device *dev) + { + return YAFFS_OK; + } +--- a/fs/yaffs2/yaffs_mtdif.h ++++ b/fs/yaffs2/yaffs_mtdif.h +@@ -18,10 +18,15 @@ + + #include "yaffs_guts.h" + +-int nandmtd_WriteChunkToNAND(yaffs_Device * dev, int chunkInNAND, +- const __u8 * data, const yaffs_Spare * spare); +-int nandmtd_ReadChunkFromNAND(yaffs_Device * dev, int chunkInNAND, __u8 * data, +- yaffs_Spare * spare); +-int nandmtd_EraseBlockInNAND(yaffs_Device * dev, int blockNumber); +-int nandmtd_InitialiseNAND(yaffs_Device * dev); ++#if (MTD_VERSION_CODE < MTD_VERSION(2, 6, 18)) ++extern struct nand_oobinfo yaffs_oobinfo; ++extern struct nand_oobinfo yaffs_noeccinfo; ++#endif ++ ++int nandmtd_WriteChunkToNAND(yaffs_Device *dev, int chunkInNAND, ++ const __u8 *data, const yaffs_Spare *spare); ++int nandmtd_ReadChunkFromNAND(yaffs_Device *dev, int chunkInNAND, __u8 *data, ++ yaffs_Spare *spare); ++int nandmtd_EraseBlockInNAND(yaffs_Device *dev, int blockNumber); ++int nandmtd_InitialiseNAND(yaffs_Device *dev); + #endif +--- a/fs/yaffs2/yaffs_nand.c ++++ b/fs/yaffs2/yaffs_nand.c +@@ -12,16 +12,17 @@ + */ + + const char *yaffs_nand_c_version = +- "$Id: yaffs_nand.c,v 1.7 2007-02-14 01:09:06 wookey Exp $"; ++ "$Id: yaffs_nand.c,v 1.10 2009-03-06 17:20:54 wookey Exp $"; + + #include "yaffs_nand.h" + #include "yaffs_tagscompat.h" + #include "yaffs_tagsvalidity.h" + ++#include "yaffs_getblockinfo.h" + +-int yaffs_ReadChunkWithTagsFromNAND(yaffs_Device * dev, int chunkInNAND, +- __u8 * buffer, +- yaffs_ExtendedTags * tags) ++int yaffs_ReadChunkWithTagsFromNAND(yaffs_Device *dev, int chunkInNAND, ++ __u8 *buffer, ++ yaffs_ExtendedTags *tags) + { + int result; + yaffs_ExtendedTags localTags; +@@ -29,7 +30,7 @@ int yaffs_ReadChunkWithTagsFromNAND(yaff + int realignedChunkInNAND = chunkInNAND - dev->chunkOffset; + + /* If there are no tags provided, use local tags to get prioritised gc working */ +- if(!tags) ++ if (!tags) + tags = &localTags; + + if (dev->readChunkWithTagsFromNAND) +@@ -40,20 +41,20 @@ int yaffs_ReadChunkWithTagsFromNAND(yaff + realignedChunkInNAND, + buffer, + tags); +- if(tags && +- tags->eccResult > YAFFS_ECC_RESULT_NO_ERROR){ ++ if (tags && ++ tags->eccResult > YAFFS_ECC_RESULT_NO_ERROR) { + + yaffs_BlockInfo *bi = yaffs_GetBlockInfo(dev, chunkInNAND/dev->nChunksPerBlock); +- yaffs_HandleChunkError(dev,bi); ++ yaffs_HandleChunkError(dev, bi); + } + + return result; + } + +-int yaffs_WriteChunkWithTagsToNAND(yaffs_Device * dev, ++int yaffs_WriteChunkWithTagsToNAND(yaffs_Device *dev, + int chunkInNAND, +- const __u8 * buffer, +- yaffs_ExtendedTags * tags) ++ const __u8 *buffer, ++ yaffs_ExtendedTags *tags) + { + chunkInNAND -= dev->chunkOffset; + +@@ -84,7 +85,7 @@ int yaffs_WriteChunkWithTagsToNAND(yaffs + tags); + } + +-int yaffs_MarkBlockBad(yaffs_Device * dev, int blockNo) ++int yaffs_MarkBlockBad(yaffs_Device *dev, int blockNo) + { + blockNo -= dev->blockOffset; + +@@ -95,10 +96,10 @@ int yaffs_MarkBlockBad(yaffs_Device * de + return yaffs_TagsCompatabilityMarkNANDBlockBad(dev, blockNo); + } + +-int yaffs_QueryInitialBlockState(yaffs_Device * dev, ++int yaffs_QueryInitialBlockState(yaffs_Device *dev, + int blockNo, +- yaffs_BlockState * state, +- unsigned *sequenceNumber) ++ yaffs_BlockState *state, ++ __u32 *sequenceNumber) + { + blockNo -= dev->blockOffset; + +--- a/fs/yaffs2/yaffs_nandemul2k.h ++++ b/fs/yaffs2/yaffs_nandemul2k.h +@@ -21,14 +21,14 @@ + #include "yaffs_guts.h" + + int nandemul2k_WriteChunkWithTagsToNAND(struct yaffs_DeviceStruct *dev, +- int chunkInNAND, const __u8 * data, +- yaffs_ExtendedTags * tags); ++ int chunkInNAND, const __u8 *data, ++ const yaffs_ExtendedTags *tags); + int nandemul2k_ReadChunkWithTagsFromNAND(struct yaffs_DeviceStruct *dev, +- int chunkInNAND, __u8 * data, +- yaffs_ExtendedTags * tags); ++ int chunkInNAND, __u8 *data, ++ yaffs_ExtendedTags *tags); + int nandemul2k_MarkNANDBlockBad(struct yaffs_DeviceStruct *dev, int blockNo); + int nandemul2k_QueryNANDBlock(struct yaffs_DeviceStruct *dev, int blockNo, +- yaffs_BlockState * state, int *sequenceNumber); ++ yaffs_BlockState *state, __u32 *sequenceNumber); + int nandemul2k_EraseBlockInNAND(struct yaffs_DeviceStruct *dev, + int blockInNAND); + int nandemul2k_InitialiseNAND(struct yaffs_DeviceStruct *dev); +--- a/fs/yaffs2/yaffs_nand.h ++++ b/fs/yaffs2/yaffs_nand.h +@@ -19,21 +19,21 @@ + + + +-int yaffs_ReadChunkWithTagsFromNAND(yaffs_Device * dev, int chunkInNAND, +- __u8 * buffer, +- yaffs_ExtendedTags * tags); +- +-int yaffs_WriteChunkWithTagsToNAND(yaffs_Device * dev, +- int chunkInNAND, +- const __u8 * buffer, +- yaffs_ExtendedTags * tags); +- +-int yaffs_MarkBlockBad(yaffs_Device * dev, int blockNo); +- +-int yaffs_QueryInitialBlockState(yaffs_Device * dev, +- int blockNo, +- yaffs_BlockState * state, +- unsigned *sequenceNumber); ++int yaffs_ReadChunkWithTagsFromNAND(yaffs_Device *dev, int chunkInNAND, ++ __u8 *buffer, ++ yaffs_ExtendedTags *tags); ++ ++int yaffs_WriteChunkWithTagsToNAND(yaffs_Device *dev, ++ int chunkInNAND, ++ const __u8 *buffer, ++ yaffs_ExtendedTags *tags); ++ ++int yaffs_MarkBlockBad(yaffs_Device *dev, int blockNo); ++ ++int yaffs_QueryInitialBlockState(yaffs_Device *dev, ++ int blockNo, ++ yaffs_BlockState *state, ++ unsigned *sequenceNumber); + + int yaffs_EraseBlockInNAND(struct yaffs_DeviceStruct *dev, + int blockInNAND); +--- a/fs/yaffs2/yaffs_packedtags1.c ++++ b/fs/yaffs2/yaffs_packedtags1.c +@@ -14,7 +14,7 @@ + #include "yaffs_packedtags1.h" + #include "yportenv.h" + +-void yaffs_PackTags1(yaffs_PackedTags1 * pt, const yaffs_ExtendedTags * t) ++void yaffs_PackTags1(yaffs_PackedTags1 *pt, const yaffs_ExtendedTags *t) + { + pt->chunkId = t->chunkId; + pt->serialNumber = t->serialNumber; +@@ -27,7 +27,7 @@ void yaffs_PackTags1(yaffs_PackedTags1 * + + } + +-void yaffs_UnpackTags1(yaffs_ExtendedTags * t, const yaffs_PackedTags1 * pt) ++void yaffs_UnpackTags1(yaffs_ExtendedTags *t, const yaffs_PackedTags1 *pt) + { + static const __u8 allFF[] = + { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, +@@ -35,9 +35,8 @@ void yaffs_UnpackTags1(yaffs_ExtendedTag + + if (memcmp(allFF, pt, sizeof(yaffs_PackedTags1))) { + t->blockBad = 0; +- if (pt->shouldBeFF != 0xFFFFFFFF) { ++ if (pt->shouldBeFF != 0xFFFFFFFF) + t->blockBad = 1; +- } + t->chunkUsed = 1; + t->objectId = pt->objectId; + t->chunkId = pt->chunkId; +@@ -47,6 +46,5 @@ void yaffs_UnpackTags1(yaffs_ExtendedTag + t->serialNumber = pt->serialNumber; + } else { + memset(t, 0, sizeof(yaffs_ExtendedTags)); +- + } + } +--- a/fs/yaffs2/yaffs_packedtags1.h ++++ b/fs/yaffs2/yaffs_packedtags1.h +@@ -32,6 +32,6 @@ typedef struct { + + } yaffs_PackedTags1; + +-void yaffs_PackTags1(yaffs_PackedTags1 * pt, const yaffs_ExtendedTags * t); +-void yaffs_UnpackTags1(yaffs_ExtendedTags * t, const yaffs_PackedTags1 * pt); ++void yaffs_PackTags1(yaffs_PackedTags1 *pt, const yaffs_ExtendedTags *t); ++void yaffs_UnpackTags1(yaffs_ExtendedTags *t, const yaffs_PackedTags1 *pt); + #endif +--- a/fs/yaffs2/yaffs_packedtags2.c ++++ b/fs/yaffs2/yaffs_packedtags2.c +@@ -37,60 +37,68 @@ + #define EXTRA_OBJECT_TYPE_SHIFT (28) + #define EXTRA_OBJECT_TYPE_MASK ((0x0F) << EXTRA_OBJECT_TYPE_SHIFT) + +-static void yaffs_DumpPackedTags2(const yaffs_PackedTags2 * pt) ++ ++static void yaffs_DumpPackedTags2TagsPart(const yaffs_PackedTags2TagsPart *ptt) + { + T(YAFFS_TRACE_MTD, + (TSTR("packed tags obj %d chunk %d byte %d seq %d" TENDSTR), +- pt->t.objectId, pt->t.chunkId, pt->t.byteCount, +- pt->t.sequenceNumber)); ++ ptt->objectId, ptt->chunkId, ptt->byteCount, ++ ptt->sequenceNumber)); ++} ++static void yaffs_DumpPackedTags2(const yaffs_PackedTags2 *pt) ++{ ++ yaffs_DumpPackedTags2TagsPart(&pt->t); + } + +-static void yaffs_DumpTags2(const yaffs_ExtendedTags * t) ++static void yaffs_DumpTags2(const yaffs_ExtendedTags *t) + { + T(YAFFS_TRACE_MTD, + (TSTR +- ("ext.tags eccres %d blkbad %d chused %d obj %d chunk%d byte " +- "%d del %d ser %d seq %d" ++ ("ext.tags eccres %d blkbad %d chused %d obj %d chunk%d byte %d del %d ser %d seq %d" + TENDSTR), t->eccResult, t->blockBad, t->chunkUsed, t->objectId, + t->chunkId, t->byteCount, t->chunkDeleted, t->serialNumber, + t->sequenceNumber)); + + } + +-void yaffs_PackTags2(yaffs_PackedTags2 * pt, const yaffs_ExtendedTags * t) ++void yaffs_PackTags2TagsPart(yaffs_PackedTags2TagsPart *ptt, ++ const yaffs_ExtendedTags *t) + { +- pt->t.chunkId = t->chunkId; +- pt->t.sequenceNumber = t->sequenceNumber; +- pt->t.byteCount = t->byteCount; +- pt->t.objectId = t->objectId; ++ ptt->chunkId = t->chunkId; ++ ptt->sequenceNumber = t->sequenceNumber; ++ ptt->byteCount = t->byteCount; ++ ptt->objectId = t->objectId; + + if (t->chunkId == 0 && t->extraHeaderInfoAvailable) { + /* Store the extra header info instead */ + /* We save the parent object in the chunkId */ +- pt->t.chunkId = EXTRA_HEADER_INFO_FLAG ++ ptt->chunkId = EXTRA_HEADER_INFO_FLAG + | t->extraParentObjectId; +- if (t->extraIsShrinkHeader) { +- pt->t.chunkId |= EXTRA_SHRINK_FLAG; +- } +- if (t->extraShadows) { +- pt->t.chunkId |= EXTRA_SHADOWS_FLAG; +- } ++ if (t->extraIsShrinkHeader) ++ ptt->chunkId |= EXTRA_SHRINK_FLAG; ++ if (t->extraShadows) ++ ptt->chunkId |= EXTRA_SHADOWS_FLAG; + +- pt->t.objectId &= ~EXTRA_OBJECT_TYPE_MASK; +- pt->t.objectId |= ++ ptt->objectId &= ~EXTRA_OBJECT_TYPE_MASK; ++ ptt->objectId |= + (t->extraObjectType << EXTRA_OBJECT_TYPE_SHIFT); + +- if (t->extraObjectType == YAFFS_OBJECT_TYPE_HARDLINK) { +- pt->t.byteCount = t->extraEquivalentObjectId; +- } else if (t->extraObjectType == YAFFS_OBJECT_TYPE_FILE) { +- pt->t.byteCount = t->extraFileLength; +- } else { +- pt->t.byteCount = 0; +- } ++ if (t->extraObjectType == YAFFS_OBJECT_TYPE_HARDLINK) ++ ptt->byteCount = t->extraEquivalentObjectId; ++ else if (t->extraObjectType == YAFFS_OBJECT_TYPE_FILE) ++ ptt->byteCount = t->extraFileLength; ++ else ++ ptt->byteCount = 0; + } + +- yaffs_DumpPackedTags2(pt); ++ yaffs_DumpPackedTags2TagsPart(ptt); + yaffs_DumpTags2(t); ++} ++ ++ ++void yaffs_PackTags2(yaffs_PackedTags2 *pt, const yaffs_ExtendedTags *t) ++{ ++ yaffs_PackTags2TagsPart(&pt->t, t); + + #ifndef YAFFS_IGNORE_TAGS_ECC + { +@@ -101,82 +109,98 @@ void yaffs_PackTags2(yaffs_PackedTags2 * + #endif + } + +-void yaffs_UnpackTags2(yaffs_ExtendedTags * t, yaffs_PackedTags2 * pt) ++ ++void yaffs_UnpackTags2TagsPart(yaffs_ExtendedTags *t, ++ yaffs_PackedTags2TagsPart *ptt) + { + + memset(t, 0, sizeof(yaffs_ExtendedTags)); + + yaffs_InitialiseTags(t); + +- if (pt->t.sequenceNumber != 0xFFFFFFFF) { +- /* Page is in use */ +-#ifdef YAFFS_IGNORE_TAGS_ECC +- { +- t->eccResult = YAFFS_ECC_RESULT_NO_ERROR; +- } +-#else +- { +- yaffs_ECCOther ecc; +- int result; +- yaffs_ECCCalculateOther((unsigned char *)&pt->t, +- sizeof +- (yaffs_PackedTags2TagsPart), +- &ecc); +- result = +- yaffs_ECCCorrectOther((unsigned char *)&pt->t, +- sizeof +- (yaffs_PackedTags2TagsPart), +- &pt->ecc, &ecc); +- switch(result){ +- case 0: +- t->eccResult = YAFFS_ECC_RESULT_NO_ERROR; +- break; +- case 1: +- t->eccResult = YAFFS_ECC_RESULT_FIXED; +- break; +- case -1: +- t->eccResult = YAFFS_ECC_RESULT_UNFIXED; +- break; +- default: +- t->eccResult = YAFFS_ECC_RESULT_UNKNOWN; +- } +- } +-#endif ++ if (ptt->sequenceNumber != 0xFFFFFFFF) { + t->blockBad = 0; + t->chunkUsed = 1; +- t->objectId = pt->t.objectId; +- t->chunkId = pt->t.chunkId; +- t->byteCount = pt->t.byteCount; ++ t->objectId = ptt->objectId; ++ t->chunkId = ptt->chunkId; ++ t->byteCount = ptt->byteCount; + t->chunkDeleted = 0; + t->serialNumber = 0; +- t->sequenceNumber = pt->t.sequenceNumber; ++ t->sequenceNumber = ptt->sequenceNumber; + + /* Do extra header info stuff */ + +- if (pt->t.chunkId & EXTRA_HEADER_INFO_FLAG) { ++ if (ptt->chunkId & EXTRA_HEADER_INFO_FLAG) { + t->chunkId = 0; + t->byteCount = 0; + + t->extraHeaderInfoAvailable = 1; + t->extraParentObjectId = +- pt->t.chunkId & (~(ALL_EXTRA_FLAGS)); ++ ptt->chunkId & (~(ALL_EXTRA_FLAGS)); + t->extraIsShrinkHeader = +- (pt->t.chunkId & EXTRA_SHRINK_FLAG) ? 1 : 0; ++ (ptt->chunkId & EXTRA_SHRINK_FLAG) ? 1 : 0; + t->extraShadows = +- (pt->t.chunkId & EXTRA_SHADOWS_FLAG) ? 1 : 0; ++ (ptt->chunkId & EXTRA_SHADOWS_FLAG) ? 1 : 0; + t->extraObjectType = +- pt->t.objectId >> EXTRA_OBJECT_TYPE_SHIFT; ++ ptt->objectId >> EXTRA_OBJECT_TYPE_SHIFT; + t->objectId &= ~EXTRA_OBJECT_TYPE_MASK; + +- if (t->extraObjectType == YAFFS_OBJECT_TYPE_HARDLINK) { +- t->extraEquivalentObjectId = pt->t.byteCount; +- } else { +- t->extraFileLength = pt->t.byteCount; ++ if (t->extraObjectType == YAFFS_OBJECT_TYPE_HARDLINK) ++ t->extraEquivalentObjectId = ptt->byteCount; ++ else ++ t->extraFileLength = ptt->byteCount; ++ } ++ } ++ ++ yaffs_DumpPackedTags2TagsPart(ptt); ++ yaffs_DumpTags2(t); ++ ++} ++ ++ ++void yaffs_UnpackTags2(yaffs_ExtendedTags *t, yaffs_PackedTags2 *pt) ++{ ++ ++ yaffs_ECCResult eccResult = YAFFS_ECC_RESULT_NO_ERROR; ++ ++ if (pt->t.sequenceNumber != 0xFFFFFFFF) { ++ /* Page is in use */ ++#ifndef YAFFS_IGNORE_TAGS_ECC ++ { ++ yaffs_ECCOther ecc; ++ int result; ++ yaffs_ECCCalculateOther((unsigned char *)&pt->t, ++ sizeof ++ (yaffs_PackedTags2TagsPart), ++ &ecc); ++ result = ++ yaffs_ECCCorrectOther((unsigned char *)&pt->t, ++ sizeof ++ (yaffs_PackedTags2TagsPart), ++ &pt->ecc, &ecc); ++ switch (result) { ++ case 0: ++ eccResult = YAFFS_ECC_RESULT_NO_ERROR; ++ break; ++ case 1: ++ eccResult = YAFFS_ECC_RESULT_FIXED; ++ break; ++ case -1: ++ eccResult = YAFFS_ECC_RESULT_UNFIXED; ++ break; ++ default: ++ eccResult = YAFFS_ECC_RESULT_UNKNOWN; + } + } ++#endif + } + ++ yaffs_UnpackTags2TagsPart(t, &pt->t); ++ ++ t->eccResult = eccResult; ++ + yaffs_DumpPackedTags2(pt); + yaffs_DumpTags2(t); + + } ++ +--- a/fs/yaffs2/yaffs_packedtags2.h ++++ b/fs/yaffs2/yaffs_packedtags2.h +@@ -33,6 +33,11 @@ typedef struct { + yaffs_ECCOther ecc; + } yaffs_PackedTags2; + +-void yaffs_PackTags2(yaffs_PackedTags2 * pt, const yaffs_ExtendedTags * t); +-void yaffs_UnpackTags2(yaffs_ExtendedTags * t, yaffs_PackedTags2 * pt); ++/* Full packed tags with ECC, used for oob tags */ ++void yaffs_PackTags2(yaffs_PackedTags2 *pt, const yaffs_ExtendedTags *t); ++void yaffs_UnpackTags2(yaffs_ExtendedTags *t, yaffs_PackedTags2 *pt); ++ ++/* Only the tags part (no ECC for use with inband tags */ ++void yaffs_PackTags2TagsPart(yaffs_PackedTags2TagsPart *pt, const yaffs_ExtendedTags *t); ++void yaffs_UnpackTags2TagsPart(yaffs_ExtendedTags *t, yaffs_PackedTags2TagsPart *pt); + #endif +--- a/fs/yaffs2/yaffs_qsort.c ++++ b/fs/yaffs2/yaffs_qsort.c +@@ -28,12 +28,12 @@ + */ + + #include "yportenv.h" +-//#include ++/* #include */ + + /* + * Qsort routine from Bentley & McIlroy's "Engineering a Sort Function". + */ +-#define swapcode(TYPE, parmi, parmj, n) { \ ++#define swapcode(TYPE, parmi, parmj, n) do { \ + long i = (n) / sizeof (TYPE); \ + register TYPE *pi = (TYPE *) (parmi); \ + register TYPE *pj = (TYPE *) (parmj); \ +@@ -41,28 +41,29 @@ + register TYPE t = *pi; \ + *pi++ = *pj; \ + *pj++ = t; \ +- } while (--i > 0); \ +-} ++ } while (--i > 0); \ ++} while (0) + + #define SWAPINIT(a, es) swaptype = ((char *)a - (char *)0) % sizeof(long) || \ +- es % sizeof(long) ? 2 : es == sizeof(long)? 0 : 1; ++ es % sizeof(long) ? 2 : es == sizeof(long) ? 0 : 1; + + static __inline void + swapfunc(char *a, char *b, int n, int swaptype) + { + if (swaptype <= 1) +- swapcode(long, a, b, n) ++ swapcode(long, a, b, n); + else +- swapcode(char, a, b, n) ++ swapcode(char, a, b, n); + } + +-#define swap(a, b) \ ++#define yswap(a, b) do { \ + if (swaptype == 0) { \ + long t = *(long *)(a); \ + *(long *)(a) = *(long *)(b); \ + *(long *)(b) = t; \ + } else \ +- swapfunc(a, b, es, swaptype) ++ swapfunc(a, b, es, swaptype); \ ++} while (0) + + #define vecswap(a, b, n) if ((n) > 0) swapfunc(a, b, n, swaptype) + +@@ -70,12 +71,12 @@ static __inline char * + med3(char *a, char *b, char *c, int (*cmp)(const void *, const void *)) + { + return cmp(a, b) < 0 ? +- (cmp(b, c) < 0 ? b : (cmp(a, c) < 0 ? c : a )) +- :(cmp(b, c) > 0 ? b : (cmp(a, c) < 0 ? a : c )); ++ (cmp(b, c) < 0 ? b : (cmp(a, c) < 0 ? c : a)) ++ : (cmp(b, c) > 0 ? b : (cmp(a, c) < 0 ? a : c)); + } + + #ifndef min +-#define min(a,b) (((a) < (b)) ? (a) : (b)) ++#define min(a, b) (((a) < (b)) ? (a) : (b)) + #endif + + void +@@ -92,7 +93,7 @@ loop: SWAPINIT(a, es); + for (pm = (char *)a + es; pm < (char *) a + n * es; pm += es) + for (pl = pm; pl > (char *) a && cmp(pl - es, pl) > 0; + pl -= es) +- swap(pl, pl - es); ++ yswap(pl, pl - es); + return; + } + pm = (char *)a + (n / 2) * es; +@@ -107,7 +108,7 @@ loop: SWAPINIT(a, es); + } + pm = med3(pl, pm, pn, cmp); + } +- swap(a, pm); ++ yswap(a, pm); + pa = pb = (char *)a + es; + + pc = pd = (char *)a + (n - 1) * es; +@@ -115,7 +116,7 @@ loop: SWAPINIT(a, es); + while (pb <= pc && (r = cmp(pb, a)) <= 0) { + if (r == 0) { + swap_cnt = 1; +- swap(pa, pb); ++ yswap(pa, pb); + pa += es; + } + pb += es; +@@ -123,14 +124,14 @@ loop: SWAPINIT(a, es); + while (pb <= pc && (r = cmp(pc, a)) >= 0) { + if (r == 0) { + swap_cnt = 1; +- swap(pc, pd); ++ yswap(pc, pd); + pd -= es; + } + pc -= es; + } + if (pb > pc) + break; +- swap(pb, pc); ++ yswap(pb, pc); + swap_cnt = 1; + pb += es; + pc -= es; +@@ -139,7 +140,7 @@ loop: SWAPINIT(a, es); + for (pm = (char *) a + es; pm < (char *) a + n * es; pm += es) + for (pl = pm; pl > (char *) a && cmp(pl - es, pl) > 0; + pl -= es) +- swap(pl, pl - es); ++ yswap(pl, pl - es); + return; + } + +@@ -148,9 +149,11 @@ loop: SWAPINIT(a, es); + vecswap(a, pb - r, r); + r = min((long)(pd - pc), (long)(pn - pd - es)); + vecswap(pb, pn - r, r); +- if ((r = pb - pa) > es) ++ r = pb - pa; ++ if (r > es) + yaffs_qsort(a, r / es, es, cmp); +- if ((r = pd - pc) > es) { ++ r = pd - pc; ++ if (r > es) { + /* Iterate rather than recurse to save stack space */ + a = pn - r; + n = r / es; +--- a/fs/yaffs2/yaffs_qsort.h ++++ b/fs/yaffs2/yaffs_qsort.h +@@ -17,7 +17,7 @@ + #ifndef __YAFFS_QSORT_H__ + #define __YAFFS_QSORT_H__ + +-extern void yaffs_qsort (void *const base, size_t total_elems, size_t size, +- int (*cmp)(const void *, const void *)); ++extern void yaffs_qsort(void *const base, size_t total_elems, size_t size, ++ int (*cmp)(const void *, const void *)); + + #endif +--- a/fs/yaffs2/yaffs_tagscompat.c ++++ b/fs/yaffs2/yaffs_tagscompat.c +@@ -14,16 +14,17 @@ + #include "yaffs_guts.h" + #include "yaffs_tagscompat.h" + #include "yaffs_ecc.h" ++#include "yaffs_getblockinfo.h" + +-static void yaffs_HandleReadDataError(yaffs_Device * dev, int chunkInNAND); ++static void yaffs_HandleReadDataError(yaffs_Device *dev, int chunkInNAND); + #ifdef NOTYET +-static void yaffs_CheckWrittenBlock(yaffs_Device * dev, int chunkInNAND); +-static void yaffs_HandleWriteChunkOk(yaffs_Device * dev, int chunkInNAND, +- const __u8 * data, +- const yaffs_Spare * spare); +-static void yaffs_HandleUpdateChunk(yaffs_Device * dev, int chunkInNAND, +- const yaffs_Spare * spare); +-static void yaffs_HandleWriteChunkError(yaffs_Device * dev, int chunkInNAND); ++static void yaffs_CheckWrittenBlock(yaffs_Device *dev, int chunkInNAND); ++static void yaffs_HandleWriteChunkOk(yaffs_Device *dev, int chunkInNAND, ++ const __u8 *data, ++ const yaffs_Spare *spare); ++static void yaffs_HandleUpdateChunk(yaffs_Device *dev, int chunkInNAND, ++ const yaffs_Spare *spare); ++static void yaffs_HandleWriteChunkError(yaffs_Device *dev, int chunkInNAND); + #endif + + static const char yaffs_countBitsTable[256] = { +@@ -54,13 +55,13 @@ int yaffs_CountBits(__u8 x) + + /********** Tags ECC calculations *********/ + +-void yaffs_CalcECC(const __u8 * data, yaffs_Spare * spare) ++void yaffs_CalcECC(const __u8 *data, yaffs_Spare *spare) + { + yaffs_ECCCalculate(data, spare->ecc1); + yaffs_ECCCalculate(&data[256], spare->ecc2); + } + +-void yaffs_CalcTagsECC(yaffs_Tags * tags) ++void yaffs_CalcTagsECC(yaffs_Tags *tags) + { + /* Calculate an ecc */ + +@@ -74,9 +75,8 @@ void yaffs_CalcTagsECC(yaffs_Tags * tags + for (i = 0; i < 8; i++) { + for (j = 1; j & 0xff; j <<= 1) { + bit++; +- if (b[i] & j) { ++ if (b[i] & j) + ecc ^= bit; +- } + } + } + +@@ -84,7 +84,7 @@ void yaffs_CalcTagsECC(yaffs_Tags * tags + + } + +-int yaffs_CheckECCOnTags(yaffs_Tags * tags) ++int yaffs_CheckECCOnTags(yaffs_Tags *tags) + { + unsigned ecc = tags->ecc; + +@@ -115,8 +115,8 @@ int yaffs_CheckECCOnTags(yaffs_Tags * ta + + /********** Tags **********/ + +-static void yaffs_LoadTagsIntoSpare(yaffs_Spare * sparePtr, +- yaffs_Tags * tagsPtr) ++static void yaffs_LoadTagsIntoSpare(yaffs_Spare *sparePtr, ++ yaffs_Tags *tagsPtr) + { + yaffs_TagsUnion *tu = (yaffs_TagsUnion *) tagsPtr; + +@@ -132,8 +132,8 @@ static void yaffs_LoadTagsIntoSpare(yaff + sparePtr->tagByte7 = tu->asBytes[7]; + } + +-static void yaffs_GetTagsFromSpare(yaffs_Device * dev, yaffs_Spare * sparePtr, +- yaffs_Tags * tagsPtr) ++static void yaffs_GetTagsFromSpare(yaffs_Device *dev, yaffs_Spare *sparePtr, ++ yaffs_Tags *tagsPtr) + { + yaffs_TagsUnion *tu = (yaffs_TagsUnion *) tagsPtr; + int result; +@@ -148,21 +148,20 @@ static void yaffs_GetTagsFromSpare(yaffs + tu->asBytes[7] = sparePtr->tagByte7; + + result = yaffs_CheckECCOnTags(tagsPtr); +- if (result > 0) { ++ if (result > 0) + dev->tagsEccFixed++; +- } else if (result < 0) { ++ else if (result < 0) + dev->tagsEccUnfixed++; +- } + } + +-static void yaffs_SpareInitialise(yaffs_Spare * spare) ++static void yaffs_SpareInitialise(yaffs_Spare *spare) + { + memset(spare, 0xFF, sizeof(yaffs_Spare)); + } + + static int yaffs_WriteChunkToNAND(struct yaffs_DeviceStruct *dev, +- int chunkInNAND, const __u8 * data, +- yaffs_Spare * spare) ++ int chunkInNAND, const __u8 *data, ++ yaffs_Spare *spare) + { + if (chunkInNAND < dev->startBlock * dev->nChunksPerBlock) { + T(YAFFS_TRACE_ERROR, +@@ -177,9 +176,9 @@ static int yaffs_WriteChunkToNAND(struct + + static int yaffs_ReadChunkFromNAND(struct yaffs_DeviceStruct *dev, + int chunkInNAND, +- __u8 * data, +- yaffs_Spare * spare, +- yaffs_ECCResult * eccResult, ++ __u8 *data, ++ yaffs_Spare *spare, ++ yaffs_ECCResult *eccResult, + int doErrorCorrection) + { + int retVal; +@@ -252,9 +251,11 @@ static int yaffs_ReadChunkFromNAND(struc + /* Must allocate enough memory for spare+2*sizeof(int) */ + /* for ecc results from device. */ + struct yaffs_NANDSpare nspare; +- retVal = +- dev->readChunkFromNAND(dev, chunkInNAND, data, +- (yaffs_Spare *) & nspare); ++ ++ memset(&nspare, 0, sizeof(nspare)); ++ ++ retVal = dev->readChunkFromNAND(dev, chunkInNAND, data, ++ (yaffs_Spare *) &nspare); + memcpy(spare, &nspare, sizeof(yaffs_Spare)); + if (data && doErrorCorrection) { + if (nspare.eccres1 > 0) { +@@ -302,8 +303,7 @@ static int yaffs_ReadChunkFromNAND(struc + static int yaffs_CheckChunkErased(struct yaffs_DeviceStruct *dev, + int chunkInNAND) + { +- +- static int init = 0; ++ static int init; + static __u8 cmpbuf[YAFFS_BYTES_PER_CHUNK]; + static __u8 data[YAFFS_BYTES_PER_CHUNK]; + /* Might as well always allocate the larger size for */ +@@ -331,12 +331,12 @@ static int yaffs_CheckChunkErased(struct + * Functions for robustisizing + */ + +-static void yaffs_HandleReadDataError(yaffs_Device * dev, int chunkInNAND) ++static void yaffs_HandleReadDataError(yaffs_Device *dev, int chunkInNAND) + { + int blockInNAND = chunkInNAND / dev->nChunksPerBlock; + + /* Mark the block for retirement */ +- yaffs_GetBlockInfo(dev, blockInNAND)->needsRetiring = 1; ++ yaffs_GetBlockInfo(dev, blockInNAND + dev->blockOffset)->needsRetiring = 1; + T(YAFFS_TRACE_ERROR | YAFFS_TRACE_BAD_BLOCKS, + (TSTR("**>>Block %d marked for retirement" TENDSTR), blockInNAND)); + +@@ -348,22 +348,22 @@ static void yaffs_HandleReadDataError(ya + } + + #ifdef NOTYET +-static void yaffs_CheckWrittenBlock(yaffs_Device * dev, int chunkInNAND) ++static void yaffs_CheckWrittenBlock(yaffs_Device *dev, int chunkInNAND) + { + } + +-static void yaffs_HandleWriteChunkOk(yaffs_Device * dev, int chunkInNAND, +- const __u8 * data, +- const yaffs_Spare * spare) ++static void yaffs_HandleWriteChunkOk(yaffs_Device *dev, int chunkInNAND, ++ const __u8 *data, ++ const yaffs_Spare *spare) + { + } + +-static void yaffs_HandleUpdateChunk(yaffs_Device * dev, int chunkInNAND, +- const yaffs_Spare * spare) ++static void yaffs_HandleUpdateChunk(yaffs_Device *dev, int chunkInNAND, ++ const yaffs_Spare *spare) + { + } + +-static void yaffs_HandleWriteChunkError(yaffs_Device * dev, int chunkInNAND) ++static void yaffs_HandleWriteChunkError(yaffs_Device *dev, int chunkInNAND) + { + int blockInNAND = chunkInNAND / dev->nChunksPerBlock; + +@@ -373,8 +373,8 @@ static void yaffs_HandleWriteChunkError( + yaffs_DeleteChunk(dev, chunkInNAND, 1, __LINE__); + } + +-static int yaffs_VerifyCompare(const __u8 * d0, const __u8 * d1, +- const yaffs_Spare * s0, const yaffs_Spare * s1) ++static int yaffs_VerifyCompare(const __u8 *d0, const __u8 *d1, ++ const yaffs_Spare *s0, const yaffs_Spare *s1) + { + + if (memcmp(d0, d1, YAFFS_BYTES_PER_CHUNK) != 0 || +@@ -398,28 +398,35 @@ static int yaffs_VerifyCompare(const __u + } + #endif /* NOTYET */ + +-int yaffs_TagsCompatabilityWriteChunkWithTagsToNAND(yaffs_Device * dev, +- int chunkInNAND, +- const __u8 * data, +- const yaffs_ExtendedTags * +- eTags) ++int yaffs_TagsCompatabilityWriteChunkWithTagsToNAND(yaffs_Device *dev, ++ int chunkInNAND, ++ const __u8 *data, ++ const yaffs_ExtendedTags *eTags) + { + yaffs_Spare spare; + yaffs_Tags tags; + + yaffs_SpareInitialise(&spare); + +- if (eTags->chunkDeleted) { ++ if (eTags->chunkDeleted) + spare.pageStatus = 0; +- } else { ++ else { + tags.objectId = eTags->objectId; + tags.chunkId = eTags->chunkId; +- tags.byteCount = eTags->byteCount; ++ ++ tags.byteCountLSB = eTags->byteCount & 0x3ff; ++ ++ if (dev->nDataBytesPerChunk >= 1024) ++ tags.byteCountMSB = (eTags->byteCount >> 10) & 3; ++ else ++ tags.byteCountMSB = 3; ++ ++ + tags.serialNumber = eTags->serialNumber; + +- if (!dev->useNANDECC && data) { ++ if (!dev->useNANDECC && data) + yaffs_CalcECC(data, &spare); +- } ++ + yaffs_LoadTagsIntoSpare(&spare, &tags); + + } +@@ -427,15 +434,15 @@ int yaffs_TagsCompatabilityWriteChunkWit + return yaffs_WriteChunkToNAND(dev, chunkInNAND, data, &spare); + } + +-int yaffs_TagsCompatabilityReadChunkWithTagsFromNAND(yaffs_Device * dev, ++int yaffs_TagsCompatabilityReadChunkWithTagsFromNAND(yaffs_Device *dev, + int chunkInNAND, +- __u8 * data, +- yaffs_ExtendedTags * eTags) ++ __u8 *data, ++ yaffs_ExtendedTags *eTags) + { + + yaffs_Spare spare; + yaffs_Tags tags; +- yaffs_ECCResult eccResult; ++ yaffs_ECCResult eccResult = YAFFS_ECC_RESULT_UNKNOWN; + + static yaffs_Spare spareFF; + static int init; +@@ -466,7 +473,11 @@ int yaffs_TagsCompatabilityReadChunkWith + + eTags->objectId = tags.objectId; + eTags->chunkId = tags.chunkId; +- eTags->byteCount = tags.byteCount; ++ eTags->byteCount = tags.byteCountLSB; ++ ++ if (dev->nDataBytesPerChunk >= 1024) ++ eTags->byteCount |= (((unsigned) tags.byteCountMSB) << 10); ++ + eTags->serialNumber = tags.serialNumber; + } + } +@@ -497,9 +508,9 @@ int yaffs_TagsCompatabilityMarkNANDBlock + } + + int yaffs_TagsCompatabilityQueryNANDBlock(struct yaffs_DeviceStruct *dev, +- int blockNo, yaffs_BlockState * +- state, +- int *sequenceNumber) ++ int blockNo, ++ yaffs_BlockState *state, ++ __u32 *sequenceNumber) + { + + yaffs_Spare spare0, spare1; +--- a/fs/yaffs2/yaffs_tagscompat.h ++++ b/fs/yaffs2/yaffs_tagscompat.h +@@ -17,24 +17,23 @@ + #define __YAFFS_TAGSCOMPAT_H__ + + #include "yaffs_guts.h" +-int yaffs_TagsCompatabilityWriteChunkWithTagsToNAND(yaffs_Device * dev, +- int chunkInNAND, +- const __u8 * data, +- const yaffs_ExtendedTags * +- tags); +-int yaffs_TagsCompatabilityReadChunkWithTagsFromNAND(yaffs_Device * dev, +- int chunkInNAND, +- __u8 * data, +- yaffs_ExtendedTags * +- tags); ++int yaffs_TagsCompatabilityWriteChunkWithTagsToNAND(yaffs_Device *dev, ++ int chunkInNAND, ++ const __u8 *data, ++ const yaffs_ExtendedTags *tags); ++int yaffs_TagsCompatabilityReadChunkWithTagsFromNAND(yaffs_Device *dev, ++ int chunkInNAND, ++ __u8 *data, ++ yaffs_ExtendedTags *tags); + int yaffs_TagsCompatabilityMarkNANDBlockBad(struct yaffs_DeviceStruct *dev, + int blockNo); + int yaffs_TagsCompatabilityQueryNANDBlock(struct yaffs_DeviceStruct *dev, +- int blockNo, yaffs_BlockState * +- state, int *sequenceNumber); ++ int blockNo, ++ yaffs_BlockState *state, ++ __u32 *sequenceNumber); + +-void yaffs_CalcTagsECC(yaffs_Tags * tags); +-int yaffs_CheckECCOnTags(yaffs_Tags * tags); ++void yaffs_CalcTagsECC(yaffs_Tags *tags); ++int yaffs_CheckECCOnTags(yaffs_Tags *tags); + int yaffs_CountBits(__u8 byte); + + #endif +--- a/fs/yaffs2/yaffs_tagsvalidity.c ++++ b/fs/yaffs2/yaffs_tagsvalidity.c +@@ -13,14 +13,14 @@ + + #include "yaffs_tagsvalidity.h" + +-void yaffs_InitialiseTags(yaffs_ExtendedTags * tags) ++void yaffs_InitialiseTags(yaffs_ExtendedTags *tags) + { + memset(tags, 0, sizeof(yaffs_ExtendedTags)); + tags->validMarker0 = 0xAAAAAAAA; + tags->validMarker1 = 0x55555555; + } + +-int yaffs_ValidateTags(yaffs_ExtendedTags * tags) ++int yaffs_ValidateTags(yaffs_ExtendedTags *tags) + { + return (tags->validMarker0 == 0xAAAAAAAA && + tags->validMarker1 == 0x55555555); +--- a/fs/yaffs2/yaffs_tagsvalidity.h ++++ b/fs/yaffs2/yaffs_tagsvalidity.h +@@ -19,6 +19,6 @@ + + #include "yaffs_guts.h" + +-void yaffs_InitialiseTags(yaffs_ExtendedTags * tags); +-int yaffs_ValidateTags(yaffs_ExtendedTags * tags); ++void yaffs_InitialiseTags(yaffs_ExtendedTags *tags); ++int yaffs_ValidateTags(yaffs_ExtendedTags *tags); + #endif +--- a/fs/yaffs2/yportenv.h ++++ b/fs/yaffs2/yportenv.h +@@ -17,17 +17,28 @@ + #ifndef __YPORTENV_H__ + #define __YPORTENV_H__ + ++/* ++ * Define the MTD version in terms of Linux Kernel versions ++ * This allows yaffs to be used independantly of the kernel ++ * as well as with it. ++ */ ++ ++#define MTD_VERSION(a, b, c) (((a) << 16) + ((b) << 8) + (c)) ++ + #if defined CONFIG_YAFFS_WINCE + + #include "ywinceenv.h" + +-#elif defined __KERNEL__ ++#elif defined __KERNEL__ + + #include "moduleconfig.h" + + /* Linux kernel */ ++ + #include +-#if (LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19)) ++#define MTD_VERSION_CODE LINUX_VERSION_CODE ++ ++#if (LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 19)) + #include + #endif + #include +@@ -40,12 +51,13 @@ + #define YCHAR char + #define YUCHAR unsigned char + #define _Y(x) x +-#define yaffs_strcpy(a,b) strcpy(a,b) +-#define yaffs_strncpy(a,b,c) strncpy(a,b,c) +-#define yaffs_strncmp(a,b,c) strncmp(a,b,c) +-#define yaffs_strlen(s) strlen(s) +-#define yaffs_sprintf sprintf +-#define yaffs_toupper(a) toupper(a) ++#define yaffs_strcat(a, b) strcat(a, b) ++#define yaffs_strcpy(a, b) strcpy(a, b) ++#define yaffs_strncpy(a, b, c) strncpy(a, b, c) ++#define yaffs_strncmp(a, b, c) strncmp(a, b, c) ++#define yaffs_strlen(s) strlen(s) ++#define yaffs_sprintf sprintf ++#define yaffs_toupper(a) toupper(a) + + #define Y_INLINE inline + +@@ -53,19 +65,19 @@ + #define YAFFS_LOSTNFOUND_PREFIX "obj" + + /* #define YPRINTF(x) printk x */ +-#define YMALLOC(x) kmalloc(x,GFP_KERNEL) ++#define YMALLOC(x) kmalloc(x, GFP_NOFS) + #define YFREE(x) kfree(x) + #define YMALLOC_ALT(x) vmalloc(x) + #define YFREE_ALT(x) vfree(x) + #define YMALLOC_DMA(x) YMALLOC(x) + +-// KR - added for use in scan so processes aren't blocked indefinitely. ++/* KR - added for use in scan so processes aren't blocked indefinitely. */ + #define YYIELD() schedule() + + #define YAFFS_ROOT_MODE 0666 + #define YAFFS_LOSTNFOUND_MODE 0666 + +-#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0)) ++#if (LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 0)) + #define Y_CURRENT_TIME CURRENT_TIME.tv_sec + #define Y_TIME_CONVERT(x) (x).tv_sec + #else +@@ -73,11 +85,12 @@ + #define Y_TIME_CONVERT(x) (x) + #endif + +-#define yaffs_SumCompare(x,y) ((x) == (y)) +-#define yaffs_strcmp(a,b) strcmp(a,b) ++#define yaffs_SumCompare(x, y) ((x) == (y)) ++#define yaffs_strcmp(a, b) strcmp(a, b) + + #define TENDSTR "\n" + #define TSTR(x) KERN_WARNING x ++#define TCONT(x) x + #define TOUT(p) printk p + + #define yaffs_trace(mask, fmt, args...) \ +@@ -90,6 +103,8 @@ + + #elif defined CONFIG_YAFFS_DIRECT + ++#define MTD_VERSION_CODE MTD_VERSION(2, 6, 22) ++ + /* Direct interface */ + #include "ydirectenv.h" + +@@ -111,11 +126,12 @@ + #define YCHAR char + #define YUCHAR unsigned char + #define _Y(x) x +-#define yaffs_strcpy(a,b) strcpy(a,b) +-#define yaffs_strncpy(a,b,c) strncpy(a,b,c) +-#define yaffs_strlen(s) strlen(s) +-#define yaffs_sprintf sprintf +-#define yaffs_toupper(a) toupper(a) ++#define yaffs_strcat(a, b) strcat(a, b) ++#define yaffs_strcpy(a, b) strcpy(a, b) ++#define yaffs_strncpy(a, b, c) strncpy(a, b, c) ++#define yaffs_strlen(s) strlen(s) ++#define yaffs_sprintf sprintf ++#define yaffs_toupper(a) toupper(a) + + #define Y_INLINE inline + +@@ -133,8 +149,8 @@ + #define YAFFS_ROOT_MODE 0666 + #define YAFFS_LOSTNFOUND_MODE 0666 + +-#define yaffs_SumCompare(x,y) ((x) == (y)) +-#define yaffs_strcmp(a,b) strcmp(a,b) ++#define yaffs_SumCompare(x, y) ((x) == (y)) ++#define yaffs_strcmp(a, b) strcmp(a, b) + + #else + /* Should have specified a configuration type */ +@@ -178,10 +194,10 @@ extern unsigned int yaffs_wr_attempts; + #define YAFFS_TRACE_ALWAYS 0xF0000000 + + +-#define T(mask,p) do{ if((mask) & (yaffs_traceMask | YAFFS_TRACE_ALWAYS)) TOUT(p);} while(0) ++#define T(mask, p) do { if ((mask) & (yaffs_traceMask | YAFFS_TRACE_ALWAYS)) TOUT(p); } while (0) + +-#ifndef CONFIG_YAFFS_WINCE +-#define YBUG() T(YAFFS_TRACE_BUG,(TSTR("==>> yaffs bug: " __FILE__ " %d" TENDSTR),__LINE__)) ++#ifndef YBUG ++#define YBUG() do {T(YAFFS_TRACE_BUG, (TSTR("==>> yaffs bug: " __FILE__ " %d" TENDSTR), __LINE__)); } while (0) + #endif + + #endif diff --git a/target/linux/generic-2.6/patches-2.6.31/600-phy_extension.patch b/target/linux/generic-2.6/patches-2.6.31/600-phy_extension.patch new file mode 100644 index 0000000000..473812895e --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/600-phy_extension.patch @@ -0,0 +1,83 @@ +--- a/drivers/net/phy/phy.c ++++ b/drivers/net/phy/phy.c +@@ -299,6 +299,50 @@ int phy_ethtool_gset(struct phy_device * + } + EXPORT_SYMBOL(phy_ethtool_gset); + ++int phy_ethtool_ioctl(struct phy_device *phydev, void *useraddr) ++{ ++ u32 cmd; ++ int tmp; ++ struct ethtool_cmd ecmd = { ETHTOOL_GSET }; ++ struct ethtool_value edata = { ETHTOOL_GLINK }; ++ ++ if (get_user(cmd, (u32 *) useraddr)) ++ return -EFAULT; ++ ++ switch (cmd) { ++ case ETHTOOL_GSET: ++ phy_ethtool_gset(phydev, &ecmd); ++ if (copy_to_user(useraddr, &ecmd, sizeof(ecmd))) ++ return -EFAULT; ++ return 0; ++ ++ case ETHTOOL_SSET: ++ if (copy_from_user(&ecmd, useraddr, sizeof(ecmd))) ++ return -EFAULT; ++ return phy_ethtool_sset(phydev, &ecmd); ++ ++ case ETHTOOL_NWAY_RST: ++ /* if autoneg is off, it's an error */ ++ tmp = phy_read(phydev, MII_BMCR); ++ if (tmp & BMCR_ANENABLE) { ++ tmp |= (BMCR_ANRESTART); ++ phy_write(phydev, MII_BMCR, tmp); ++ return 0; ++ } ++ return -EINVAL; ++ ++ case ETHTOOL_GLINK: ++ edata.data = (phy_read(phydev, ++ MII_BMSR) & BMSR_LSTATUS) ? 1 : 0; ++ if (copy_to_user(useraddr, &edata, sizeof(edata))) ++ return -EFAULT; ++ return 0; ++ } ++ ++ return -EOPNOTSUPP; ++} ++EXPORT_SYMBOL(phy_ethtool_ioctl); ++ + /** + * phy_mii_ioctl - generic PHY MII ioctl interface + * @phydev: the phy_device struct +@@ -355,8 +399,8 @@ int phy_mii_ioctl(struct phy_device *phy + } + + phy_write(phydev, mii_data->reg_num, val); +- +- if (mii_data->reg_num == MII_BMCR ++ ++ if (mii_data->reg_num == MII_BMCR + && val & BMCR_RESET + && phydev->drv->config_init) { + phy_scan_fixups(phydev); +@@ -471,7 +515,7 @@ static void phy_force_reduction(struct p + int idx; + + idx = phy_find_setting(phydev->speed, phydev->duplex); +- ++ + idx++; + + idx = phy_find_valid(idx, phydev->supported); +--- a/include/linux/phy.h ++++ b/include/linux/phy.h +@@ -489,6 +489,7 @@ void phy_start_machine(struct phy_device + void phy_stop_machine(struct phy_device *phydev); + int phy_ethtool_sset(struct phy_device *phydev, struct ethtool_cmd *cmd); + int phy_ethtool_gset(struct phy_device *phydev, struct ethtool_cmd *cmd); ++int phy_ethtool_ioctl(struct phy_device *phydev, void *useraddr); + int phy_mii_ioctl(struct phy_device *phydev, + struct mii_ioctl_data *mii_data, int cmd); + int phy_start_interrupts(struct phy_device *phydev); diff --git a/target/linux/generic-2.6/patches-2.6.31/620-phy_adm6996.patch b/target/linux/generic-2.6/patches-2.6.31/620-phy_adm6996.patch new file mode 100644 index 0000000000..8f61d95c2d --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/620-phy_adm6996.patch @@ -0,0 +1,24 @@ +--- a/drivers/net/phy/Kconfig ++++ b/drivers/net/phy/Kconfig +@@ -82,6 +82,11 @@ config LSI_ET1011C_PHY + ---help--- + Supports the LSI ET1011C PHY. + ++config ADM6996_PHY ++ tristate "Driver for ADM6996 switches" ++ ---help--- ++ Currently supports the ADM6996F switch ++ + config FIXED_PHY + bool "Driver for MDIO Bus/PHY emulation with fixed speed/link PHYs" + depends on PHYLIB=y +--- a/drivers/net/phy/Makefile ++++ b/drivers/net/phy/Makefile +@@ -12,6 +12,7 @@ obj-$(CONFIG_SMSC_PHY) += smsc.o + obj-$(CONFIG_VITESSE_PHY) += vitesse.o + obj-$(CONFIG_BROADCOM_PHY) += broadcom.o + obj-$(CONFIG_ICPLUS_PHY) += icplus.o ++obj-$(CONFIG_ADM6996_PHY) += adm6996.o + obj-$(CONFIG_REALTEK_PHY) += realtek.o + obj-$(CONFIG_LSI_ET1011C_PHY) += et1011c.o + obj-$(CONFIG_FIXED_PHY) += fixed.o diff --git a/target/linux/generic-2.6/patches-2.6.31/630-phy_packets.patch b/target/linux/generic-2.6/patches-2.6.31/630-phy_packets.patch new file mode 100644 index 0000000000..82f67bcee5 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/630-phy_packets.patch @@ -0,0 +1,63 @@ +--- a/drivers/net/phy/phy_device.c ++++ b/drivers/net/phy/phy_device.c +@@ -146,6 +146,18 @@ int phy_scan_fixups(struct phy_device *p + } + EXPORT_SYMBOL(phy_scan_fixups); + ++static int generic_receive_skb(struct sk_buff *skb) ++{ ++ skb->protocol = eth_type_trans(skb, skb->dev); ++ return netif_receive_skb(skb); ++} ++ ++static int generic_rx(struct sk_buff *skb) ++{ ++ skb->protocol = eth_type_trans(skb, skb->dev); ++ return netif_rx(skb); ++} ++ + struct phy_device* phy_device_create(struct mii_bus *bus, int addr, int phy_id) + { + struct phy_device *dev; +@@ -175,6 +187,8 @@ struct phy_device* phy_device_create(str + dev_set_name(&dev->dev, PHY_ID_FMT, bus->id, addr); + + dev->state = PHY_DOWN; ++ dev->netif_receive_skb = &generic_receive_skb; ++ dev->netif_rx = &generic_rx; + + mutex_init(&dev->lock); + +--- a/include/linux/phy.h ++++ b/include/linux/phy.h +@@ -325,6 +325,20 @@ struct phy_device { + void (*adjust_link)(struct net_device *dev); + + void (*adjust_state)(struct net_device *dev); ++ ++ /* ++ * By default these point to the original functions ++ * with the same name. adding them to the phy_device ++ * allows the phy driver to override them for packet ++ * mangling if the ethernet driver supports it ++ * This is required to support some really horrible ++ * switches such as the Marvell 88E6060 ++ */ ++ int (*netif_receive_skb)(struct sk_buff *skb); ++ int (*netif_rx)(struct sk_buff *skb); ++ ++ /* alignment offset for packets */ ++ int pkt_align; + }; + #define to_phy_device(d) container_of(d, struct phy_device, dev) + +--- a/include/linux/netdevice.h ++++ b/include/linux/netdevice.h +@@ -804,6 +804,7 @@ struct net_device + void *ax25_ptr; /* AX.25 specific data */ + struct wireless_dev *ieee80211_ptr; /* IEEE 802.11 specific data, + assign before registering */ ++ void *phy_ptr; /* PHY device specific data */ + + /* + * Cache line mostly used on receive path (including eth_type_trans()) diff --git a/target/linux/generic-2.6/patches-2.6.31/650-swconfig.patch b/target/linux/generic-2.6/patches-2.6.31/650-swconfig.patch new file mode 100644 index 0000000000..6825037ce9 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/650-swconfig.patch @@ -0,0 +1,25 @@ +--- a/drivers/net/phy/Kconfig ++++ b/drivers/net/phy/Kconfig +@@ -13,6 +13,12 @@ menuconfig PHYLIB + + if PHYLIB + ++config SWCONFIG ++ tristate "Switch configuration API" ++ ---help--- ++ Switch configuration API using netlink. This allows ++ you to configure the VLAN features of certain switches. ++ + comment "MII PHY device drivers" + + config MARVELL_PHY +--- a/drivers/net/phy/Makefile ++++ b/drivers/net/phy/Makefile +@@ -3,6 +3,7 @@ + libphy-objs := phy.o phy_device.o mdio_bus.o + + obj-$(CONFIG_PHYLIB) += libphy.o ++obj-$(CONFIG_SWCONFIG) += swconfig.o + obj-$(CONFIG_MARVELL_PHY) += marvell.o + obj-$(CONFIG_DAVICOM_PHY) += davicom.o + obj-$(CONFIG_CICADA_PHY) += cicada.o diff --git a/target/linux/generic-2.6/patches-2.6.31/660-phy_mvswitch.patch b/target/linux/generic-2.6/patches-2.6.31/660-phy_mvswitch.patch new file mode 100644 index 0000000000..731f30301f --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/660-phy_mvswitch.patch @@ -0,0 +1,22 @@ +--- a/drivers/net/phy/Kconfig ++++ b/drivers/net/phy/Kconfig +@@ -93,6 +93,9 @@ config ADM6996_PHY + ---help--- + Currently supports the ADM6996F switch + ++config MVSWITCH_PHY ++ tristate "Driver for Marvell 88E6060 switches" ++ + config FIXED_PHY + bool "Driver for MDIO Bus/PHY emulation with fixed speed/link PHYs" + depends on PHYLIB=y +--- a/drivers/net/phy/Makefile ++++ b/drivers/net/phy/Makefile +@@ -14,6 +14,7 @@ obj-$(CONFIG_VITESSE_PHY) += vitesse.o + obj-$(CONFIG_BROADCOM_PHY) += broadcom.o + obj-$(CONFIG_ICPLUS_PHY) += icplus.o + obj-$(CONFIG_ADM6996_PHY) += adm6996.o ++obj-$(CONFIG_MVSWITCH_PHY) += mvswitch.o + obj-$(CONFIG_REALTEK_PHY) += realtek.o + obj-$(CONFIG_LSI_ET1011C_PHY) += et1011c.o + obj-$(CONFIG_FIXED_PHY) += fixed.o diff --git a/target/linux/generic-2.6/patches-2.6.31/670-phy_ip175c.patch b/target/linux/generic-2.6/patches-2.6.31/670-phy_ip175c.patch new file mode 100644 index 0000000000..461df50465 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/670-phy_ip175c.patch @@ -0,0 +1,23 @@ +--- a/drivers/net/phy/Kconfig ++++ b/drivers/net/phy/Kconfig +@@ -96,6 +96,10 @@ config ADM6996_PHY + config MVSWITCH_PHY + tristate "Driver for Marvell 88E6060 switches" + ++config IP175C_PHY ++ tristate "Driver for IC+ IP175C/IP178C switches" ++ select SWCONFIG ++ + config FIXED_PHY + bool "Driver for MDIO Bus/PHY emulation with fixed speed/link PHYs" + depends on PHYLIB=y +--- a/drivers/net/phy/Makefile ++++ b/drivers/net/phy/Makefile +@@ -15,6 +15,7 @@ obj-$(CONFIG_BROADCOM_PHY) += broadcom.o + obj-$(CONFIG_ICPLUS_PHY) += icplus.o + obj-$(CONFIG_ADM6996_PHY) += adm6996.o + obj-$(CONFIG_MVSWITCH_PHY) += mvswitch.o ++obj-$(CONFIG_IP175C_PHY) += ip175c.o + obj-$(CONFIG_REALTEK_PHY) += realtek.o + obj-$(CONFIG_LSI_ET1011C_PHY) += et1011c.o + obj-$(CONFIG_FIXED_PHY) += fixed.o diff --git a/target/linux/generic-2.6/patches-2.6.31/680-phy_ar8216.patch b/target/linux/generic-2.6/patches-2.6.31/680-phy_ar8216.patch new file mode 100644 index 0000000000..d9d3ce7721 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/680-phy_ar8216.patch @@ -0,0 +1,23 @@ +--- a/drivers/net/phy/Kconfig ++++ b/drivers/net/phy/Kconfig +@@ -100,6 +100,10 @@ config IP175C_PHY + tristate "Driver for IC+ IP175C/IP178C switches" + select SWCONFIG + ++config AR8216_PHY ++ tristate "Driver for Atheros AR8216 switches" ++ select SWCONFIG ++ + config FIXED_PHY + bool "Driver for MDIO Bus/PHY emulation with fixed speed/link PHYs" + depends on PHYLIB=y +--- a/drivers/net/phy/Makefile ++++ b/drivers/net/phy/Makefile +@@ -17,6 +17,7 @@ obj-$(CONFIG_ADM6996_PHY) += adm6996.o + obj-$(CONFIG_MVSWITCH_PHY) += mvswitch.o + obj-$(CONFIG_IP175C_PHY) += ip175c.o + obj-$(CONFIG_REALTEK_PHY) += realtek.o ++obj-$(CONFIG_AR8216_PHY) += ar8216.o + obj-$(CONFIG_LSI_ET1011C_PHY) += et1011c.o + obj-$(CONFIG_FIXED_PHY) += fixed.o + obj-$(CONFIG_MDIO_BITBANG) += mdio-bitbang.o diff --git a/target/linux/generic-2.6/patches-2.6.31/690-phy_rtl8306.patch b/target/linux/generic-2.6/patches-2.6.31/690-phy_rtl8306.patch new file mode 100644 index 0000000000..bfebbcff51 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/690-phy_rtl8306.patch @@ -0,0 +1,23 @@ +--- a/drivers/net/phy/Kconfig ++++ b/drivers/net/phy/Kconfig +@@ -104,6 +104,10 @@ config AR8216_PHY + tristate "Driver for Atheros AR8216 switches" + select SWCONFIG + ++config RTL8306_PHY ++ tristate "Driver for Realtek RTL8306S switches" ++ select SWCONFIG ++ + config FIXED_PHY + bool "Driver for MDIO Bus/PHY emulation with fixed speed/link PHYs" + depends on PHYLIB=y +--- a/drivers/net/phy/Makefile ++++ b/drivers/net/phy/Makefile +@@ -18,6 +18,7 @@ obj-$(CONFIG_MVSWITCH_PHY) += mvswitch.o + obj-$(CONFIG_IP175C_PHY) += ip175c.o + obj-$(CONFIG_REALTEK_PHY) += realtek.o + obj-$(CONFIG_AR8216_PHY) += ar8216.o ++obj-$(CONFIG_RTL8306_PHY) += rtl8306.o + obj-$(CONFIG_LSI_ET1011C_PHY) += et1011c.o + obj-$(CONFIG_FIXED_PHY) += fixed.o + obj-$(CONFIG_MDIO_BITBANG) += mdio-bitbang.o diff --git a/target/linux/generic-2.6/patches-2.6.31/700-rtc7301.patch b/target/linux/generic-2.6/patches-2.6.31/700-rtc7301.patch new file mode 100644 index 0000000000..6e66654400 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/700-rtc7301.patch @@ -0,0 +1,250 @@ +--- a/drivers/rtc/Kconfig ++++ b/drivers/rtc/Kconfig +@@ -535,6 +535,15 @@ config RTC_DRV_PCF50633 + If you say yes here you get support for the RTC subsystem of the + NXP PCF50633 used in embedded systems. + ++config RTC_DRV_RTC7301 ++ tristate "Epson RTC-7301 SF/DG" ++ help ++ If you say Y here you will get support for the ++ Epson RTC-7301 SF/DG RTC chips. ++ ++ This driver can also be built as a module. If so, the module ++ will be called rtc-7301. ++ + comment "on-CPU RTC drivers" + + config RTC_DRV_OMAP +--- a/drivers/rtc/Makefile ++++ b/drivers/rtc/Makefile +@@ -62,6 +62,7 @@ obj-$(CONFIG_RTC_DRV_R9701) += rtc-r9701 + obj-$(CONFIG_RTC_DRV_RS5C313) += rtc-rs5c313.o + obj-$(CONFIG_RTC_DRV_RS5C348) += rtc-rs5c348.o + obj-$(CONFIG_RTC_DRV_RS5C372) += rtc-rs5c372.o ++obj-$(CONFIG_RTC_DRV_RTC7301) += rtc-rtc7301.o + obj-$(CONFIG_RTC_DRV_RX8025) += rtc-rx8025.o + obj-$(CONFIG_RTC_DRV_RX8581) += rtc-rx8581.o + obj-$(CONFIG_RTC_DRV_S35390A) += rtc-s35390a.o +--- /dev/null ++++ b/drivers/rtc/rtc-rtc7301.c +@@ -0,0 +1,219 @@ ++/* ++ * Driver for Epson RTC-7301SF/DG ++ * ++ * Copyright (C) 2009 Jose Vasconcellos ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++ ++#define RTC_NAME "rtc7301" ++#define RTC_VERSION "0.1" ++ ++/* Epson RTC-7301 register addresses */ ++#define RTC7301_SEC 0x00 ++#define RTC7301_SEC10 0x01 ++#define RTC7301_MIN 0x02 ++#define RTC7301_MIN10 0x03 ++#define RTC7301_HOUR 0x04 ++#define RTC7301_HOUR10 0x05 ++#define RTC7301_WEEKDAY 0x06 ++#define RTC7301_DAY 0x07 ++#define RTC7301_DAY10 0x08 ++#define RTC7301_MON 0x09 ++#define RTC7301_MON10 0x0A ++#define RTC7301_YEAR 0x0B ++#define RTC7301_YEAR10 0x0C ++#define RTC7301_YEAR100 0x0D ++#define RTC7301_YEAR1000 0x0E ++#define RTC7301_CTRLREG 0x0F ++ ++static uint8_t __iomem *rtc7301_base; ++ ++#define read_reg(offset) (readb(rtc7301_base + offset) & 0xf) ++#define write_reg(offset, data) writeb(data, rtc7301_base + (offset)) ++ ++#define rtc7301_isbusy() (read_reg(RTC7301_CTRLREG) & 1) ++ ++static void rtc7301_init_settings(void) ++{ ++ int i; ++ ++ write_reg(RTC7301_CTRLREG, 2); ++ write_reg(RTC7301_YEAR1000, 2); ++ udelay(122); ++ ++ /* bank 1 */ ++ write_reg(RTC7301_CTRLREG, 6); ++ for (i=0; i<15; i++) ++ write_reg(i, 0); ++ ++ /* bank 2 */ ++ write_reg(RTC7301_CTRLREG, 14); ++ for (i=0; i<15; i++) ++ write_reg(i, 0); ++ write_reg(RTC7301_CTRLREG, 0); ++} ++ ++static int rtc7301_get_datetime(struct device *dev, struct rtc_time *dt) ++{ ++ int cnt; ++ uint8_t buf[16]; ++ ++ cnt = 0; ++ while (rtc7301_isbusy()) { ++ udelay(244); ++ if (cnt++ > 100) { ++ dev_err(dev, "%s: timeout error %x\n", __func__, rtc7301_base[RTC7301_CTRLREG]); ++ return -EIO; ++ } ++ } ++ ++ for (cnt=0; cnt<16; cnt++) ++ buf[cnt] = read_reg(cnt); ++ ++ if (buf[RTC7301_SEC10] & 8) { ++ dev_err(dev, "%s: RTC not set\n", __func__); ++ return -EINVAL; ++ } ++ ++ memset(dt, 0, sizeof(*dt)); ++ ++ dt->tm_sec = buf[RTC7301_SEC] + buf[RTC7301_SEC10]*10; ++ dt->tm_min = buf[RTC7301_MIN] + buf[RTC7301_MIN10]*10; ++ dt->tm_hour = buf[RTC7301_HOUR] + buf[RTC7301_HOUR10]*10; ++ ++ dt->tm_mday = buf[RTC7301_DAY] + buf[RTC7301_DAY10]*10; ++ dt->tm_mon = buf[RTC7301_MON] + buf[RTC7301_MON10]*10 - 1; ++ dt->tm_year = buf[RTC7301_YEAR] + buf[RTC7301_YEAR10]*10 + ++ buf[RTC7301_YEAR100]*100 + ++ ((buf[RTC7301_YEAR1000] & 3)*1000) - 1900; ++ ++ /* the rtc device may contain illegal values on power up ++ * according to the data sheet. make sure they are valid. ++ */ ++ ++ return rtc_valid_tm(dt); ++} ++ ++static int rtc7301_set_datetime(struct device *dev, struct rtc_time *dt) ++{ ++ int data; ++ ++ data = dt->tm_year + 1900; ++ if (data >= 2100 || data < 1900) ++ return -EINVAL; ++ ++ write_reg(RTC7301_CTRLREG, 2); ++ udelay(122); ++ ++ data = bin2bcd(dt->tm_sec); ++ write_reg(RTC7301_SEC, data); ++ write_reg(RTC7301_SEC10, (data >> 4)); ++ ++ data = bin2bcd(dt->tm_min); ++ write_reg(RTC7301_MIN, data ); ++ write_reg(RTC7301_MIN10, (data >> 4)); ++ ++ data = bin2bcd(dt->tm_hour); ++ write_reg(RTC7301_HOUR, data); ++ write_reg(RTC7301_HOUR10, (data >> 4)); ++ ++ data = bin2bcd(dt->tm_mday); ++ write_reg(RTC7301_DAY, data); ++ write_reg(RTC7301_DAY10, (data>> 4)); ++ ++ data = bin2bcd(dt->tm_mon + 1); ++ write_reg(RTC7301_MON, data); ++ write_reg(RTC7301_MON10, (data >> 4)); ++ ++ data = bin2bcd(dt->tm_year % 100); ++ write_reg(RTC7301_YEAR, data); ++ write_reg(RTC7301_YEAR10, (data >> 4)); ++ data = bin2bcd((1900 + dt->tm_year) / 100); ++ write_reg(RTC7301_YEAR100, data); ++ ++ data = bin2bcd(dt->tm_wday); ++ write_reg(RTC7301_WEEKDAY, data); ++ ++ write_reg(RTC7301_CTRLREG, 0); ++ ++ return 0; ++} ++ ++static const struct rtc_class_ops rtc7301_rtc_ops = { ++ .read_time = rtc7301_get_datetime, ++ .set_time = rtc7301_set_datetime, ++}; ++ ++static int __devinit rtc7301_probe(struct platform_device *pdev) ++{ ++ struct rtc_device *rtc; ++ struct resource *res; ++ ++ res = platform_get_resource(pdev, IORESOURCE_MEM, 0); ++ if (!res) ++ return -ENOENT; ++ ++ rtc7301_base = ioremap_nocache(res->start, 0x1000 /*res->end - res->start + 1*/); ++ if (!rtc7301_base) ++ return -EINVAL; ++ ++ rtc = rtc_device_register(RTC_NAME, &pdev->dev, ++ &rtc7301_rtc_ops, THIS_MODULE); ++ if (IS_ERR(rtc)) { ++ iounmap(rtc7301_base); ++ return PTR_ERR(rtc); ++ } ++ ++ platform_set_drvdata(pdev, rtc); ++ ++ rtc7301_init_settings(); ++ return 0; ++} ++ ++static int __devexit rtc7301_remove(struct platform_device *pdev) ++{ ++ struct rtc_device *rtc = platform_get_drvdata(pdev); ++ ++ if (rtc) ++ rtc_device_unregister(rtc); ++ if (rtc7301_base) ++ iounmap(rtc7301_base); ++ return 0; ++} ++ ++static struct platform_driver rtc7301_driver = { ++ .driver = { ++ .name = RTC_NAME, ++ .owner = THIS_MODULE, ++ }, ++ .probe = rtc7301_probe, ++ .remove = __devexit_p(rtc7301_remove), ++}; ++ ++static __init int rtc7301_init(void) ++{ ++ return platform_driver_register(&rtc7301_driver); ++} ++module_init(rtc7301_init); ++ ++static __exit void rtc7301_exit(void) ++{ ++ platform_driver_unregister(&rtc7301_driver); ++} ++module_exit(rtc7301_exit); ++ ++MODULE_DESCRIPTION("Epson 7301 RTC driver"); ++MODULE_AUTHOR("Jose Vasconcellos "); ++MODULE_LICENSE("GPL"); ++MODULE_ALIAS("platform:" RTC_NAME); ++MODULE_VERSION(RTC_VERSION); diff --git a/target/linux/generic-2.6/patches-2.6.31/750-glamo-headers.patch b/target/linux/generic-2.6/patches-2.6.31/750-glamo-headers.patch new file mode 100644 index 0000000000..419d98a424 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/750-glamo-headers.patch @@ -0,0 +1,21 @@ +--- a/include/linux/fb.h ++++ b/include/linux/fb.h +@@ -124,6 +124,7 @@ struct dentry; + #define FB_ACCEL_TRIDENT_BLADE3D 52 /* Trident Blade3D */ + #define FB_ACCEL_TRIDENT_BLADEXP 53 /* Trident BladeXP */ + #define FB_ACCEL_CIRRUS_ALPINE 53 /* Cirrus Logic 543x/544x/5480 */ ++#define FB_ACCEL_GLAMO 50 /* SMedia Glamo */ + #define FB_ACCEL_NEOMAGIC_NM2070 90 /* NeoMagic NM2070 */ + #define FB_ACCEL_NEOMAGIC_NM2090 91 /* NeoMagic NM2090 */ + #define FB_ACCEL_NEOMAGIC_NM2093 92 /* NeoMagic NM2093 */ +--- a/include/linux/Kbuild ++++ b/include/linux/Kbuild +@@ -76,6 +76,8 @@ header-y += genetlink.h + header-y += gen_stats.h + header-y += gfs2_ondisk.h + header-y += gigaset_dev.h ++header-y += glamofb.h ++header-y += glamo-engine.h + header-y += hysdn_if.h + header-y += i2o-dev.h + header-y += i8k.h diff --git a/target/linux/generic-2.6/patches-2.6.31/780-fix-race-in-snd_soc_jack_add_gpios.patch b/target/linux/generic-2.6/patches-2.6.31/780-fix-race-in-snd_soc_jack_add_gpios.patch new file mode 100644 index 0000000000..b7214b9d6d --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/780-fix-race-in-snd_soc_jack_add_gpios.patch @@ -0,0 +1,33 @@ +commit 5ced436d549d911ce610ea47d85f71fae5bbfce4 +Author: Lars-Peter Clausen +Date: Fri Jul 31 18:26:48 2009 +0200 + + ASoC: jack: Fix race in snd_soc_jack_add_gpios + + The irq can fire as soon as it has been requested, thus all fields accessed + from within the irq handler must be initialized prior to requesting the irq. + + Signed-off-by: Lars-Peter Clausen + +--- a/sound/soc/soc-jack.c ++++ b/sound/soc/soc-jack.c +@@ -220,6 +220,9 @@ int snd_soc_jack_add_gpios(struct snd_so + if (ret) + goto err; + ++ INIT_WORK(&gpios[i].work, gpio_work); ++ gpios[i].jack = jack; ++ + ret = request_irq(gpio_to_irq(gpios[i].gpio), + gpio_handler, + IRQF_TRIGGER_RISING | IRQF_TRIGGER_FALLING, +@@ -227,9 +230,6 @@ int snd_soc_jack_add_gpios(struct snd_so + &gpios[i]); + if (ret) + goto err; +- +- INIT_WORK(&gpios[i].work, gpio_work); +- gpios[i].jack = jack; + } + + return 0; diff --git a/target/linux/generic-2.6/patches-2.6.31/801-usb_serial_endpoint_size.patch b/target/linux/generic-2.6/patches-2.6.31/801-usb_serial_endpoint_size.patch new file mode 100644 index 0000000000..6f02625381 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/801-usb_serial_endpoint_size.patch @@ -0,0 +1,25 @@ +--- a/drivers/usb/serial/usb-serial.c ++++ b/drivers/usb/serial/usb-serial.c +@@ -62,6 +62,7 @@ static struct usb_driver usb_serial_driv + drivers depend on it. + */ + ++static ushort maxSize = 0; + static int debug; + /* initially all NULL */ + static struct usb_serial *serial_table[SERIAL_TTY_MINORS]; +@@ -946,7 +947,7 @@ int usb_serial_probe(struct usb_interfac + dev_err(&interface->dev, "No free urbs available\n"); + goto probe_error; + } +- buffer_size = le16_to_cpu(endpoint->wMaxPacketSize); ++ buffer_size = (endpoint->wMaxPacketSize > maxSize) ? endpoint->wMaxPacketSize : maxSize; + port->bulk_in_size = buffer_size; + port->bulk_in_endpointAddress = endpoint->bEndpointAddress; + port->bulk_in_buffer = kmalloc(buffer_size, GFP_KERNEL); +@@ -1416,3 +1417,5 @@ MODULE_LICENSE("GPL"); + + module_param(debug, bool, S_IRUGO | S_IWUSR); + MODULE_PARM_DESC(debug, "Debug enabled or not"); ++module_param(maxSize, ushort,0); ++MODULE_PARM_DESC(maxSize,"User specified USB endpoint size"); diff --git a/target/linux/generic-2.6/patches-2.6.31/840-unable_to_open_console.patch b/target/linux/generic-2.6/patches-2.6.31/840-unable_to_open_console.patch new file mode 100644 index 0000000000..5c381f7133 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/840-unable_to_open_console.patch @@ -0,0 +1,11 @@ +--- a/init/main.c ++++ b/init/main.c +@@ -843,7 +843,7 @@ static noinline int init_post(void) + numa_default_policy(); + + if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) +- printk(KERN_WARNING "Warning: unable to open an initial console.\n"); ++ printk(KERN_WARNING "Please be patient, while OpenWrt loads ...\n"); + + (void) sys_dup(0); + (void) sys_dup(0); diff --git a/target/linux/generic-2.6/patches-2.6.31/902-darwin_scripts_include.patch b/target/linux/generic-2.6/patches-2.6.31/902-darwin_scripts_include.patch new file mode 100644 index 0000000000..e98a10d106 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/902-darwin_scripts_include.patch @@ -0,0 +1,102 @@ +--- a/scripts/genksyms/parse.c_shipped ++++ b/scripts/genksyms/parse.c_shipped +@@ -160,7 +160,9 @@ + + + #include ++#ifndef __APPLE__ + #include ++#endif + #include "genksyms.h" + + static int is_typedef; +--- a/scripts/genksyms/parse.y ++++ b/scripts/genksyms/parse.y +@@ -24,7 +24,9 @@ + %{ + + #include ++#ifndef __APPLE__ + #include ++#endif + #include "genksyms.h" + + static int is_typedef; +--- a/scripts/kallsyms.c ++++ b/scripts/kallsyms.c +@@ -22,6 +22,35 @@ + #include + #include + #include ++#ifdef __APPLE__ ++/* Darwin has no memmem implementation, this one is ripped of the uClibc-0.9.28 source */ ++void *memmem (const void *haystack, size_t haystack_len, ++ const void *needle, size_t needle_len) ++{ ++ const char *begin; ++ const char *const last_possible ++ = (const char *) haystack + haystack_len - needle_len; ++ ++ if (needle_len == 0) ++ /* The first occurrence of the empty string is deemed to occur at ++ the beginning of the string. */ ++ return (void *) haystack; ++ ++ /* Sanity check, otherwise the loop might search through the whole ++ memory. */ ++ if (__builtin_expect (haystack_len < needle_len, 0)) ++ return NULL; ++ ++ for (begin = (const char *) haystack; begin <= last_possible; ++begin) ++ if (begin[0] == ((const char *) needle)[0] && ++ !memcmp ((const void *) &begin[1], ++ (const void *) ((const char *) needle + 1), ++ needle_len - 1)) ++ return (void *) begin; ++ ++ return NULL; ++} ++#endif + + #ifndef ARRAY_SIZE + #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof(arr[0])) +--- a/scripts/kconfig/Makefile ++++ b/scripts/kconfig/Makefile +@@ -97,6 +97,9 @@ check-lxdialog := $(srctree)/$(src)/lxd + # we really need to do so. (Do not call gcc as part of make mrproper) + HOST_EXTRACFLAGS = $(shell $(CONFIG_SHELL) $(check-lxdialog) -ccflags) + HOST_LOADLIBES = $(shell $(CONFIG_SHELL) $(check-lxdialog) -ldflags $(HOSTCC)) ++ifeq ($(shell uname -s),Darwin) ++HOST_LOADLIBES += -lncurses ++endif + + HOST_EXTRACFLAGS += -DLOCALE + +--- a/scripts/mod/mk_elfconfig.c ++++ b/scripts/mod/mk_elfconfig.c +@@ -1,7 +1,11 @@ + #include + #include + #include ++#ifndef __APPLE__ + #include ++#else ++#include "../../../../../tools/sstrip/include/elf.h" ++#endif + + int + main(int argc, char **argv) +--- a/scripts/mod/modpost.h ++++ b/scripts/mod/modpost.h +@@ -7,7 +7,11 @@ + #include + #include + #include ++#ifndef __APPLE__ + #include ++#else ++#include "../../../../../tools/sstrip/include/elf.h" ++#endif + + #include "elfconfig.h" + diff --git a/target/linux/generic-2.6/patches-2.6.31/903-hostap_txpower.patch b/target/linux/generic-2.6/patches-2.6.31/903-hostap_txpower.patch new file mode 100644 index 0000000000..197d0a1e32 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/903-hostap_txpower.patch @@ -0,0 +1,154 @@ +--- a/drivers/net/wireless/hostap/hostap_ap.c ++++ b/drivers/net/wireless/hostap/hostap_ap.c +@@ -2335,13 +2335,13 @@ int prism2_ap_get_sta_qual(local_info_t + addr[count].sa_family = ARPHRD_ETHER; + memcpy(addr[count].sa_data, sta->addr, ETH_ALEN); + if (sta->last_rx_silence == 0) +- qual[count].qual = sta->last_rx_signal < 27 ? +- 0 : (sta->last_rx_signal - 27) * 92 / 127; ++ qual[count].qual = (sta->last_rx_signal - 156) == 0 ? ++ 0 : (sta->last_rx_signal - 156) * 92 / 64; + else +- qual[count].qual = sta->last_rx_signal - +- sta->last_rx_silence - 35; +- qual[count].level = HFA384X_LEVEL_TO_dBm(sta->last_rx_signal); +- qual[count].noise = HFA384X_LEVEL_TO_dBm(sta->last_rx_silence); ++ qual[count].qual = (sta->last_rx_signal - ++ sta->last_rx_silence) * 92 / 64; ++ qual[count].level = sta->last_rx_signal; ++ qual[count].noise = sta->last_rx_silence; + qual[count].updated = sta->last_rx_updated; + + sta->last_rx_updated = IW_QUAL_DBM; +@@ -2407,13 +2407,13 @@ int prism2_ap_translate_scan(struct net_ + memset(&iwe, 0, sizeof(iwe)); + iwe.cmd = IWEVQUAL; + if (sta->last_rx_silence == 0) +- iwe.u.qual.qual = sta->last_rx_signal < 27 ? +- 0 : (sta->last_rx_signal - 27) * 92 / 127; ++ iwe.u.qual.qual = (sta->last_rx_signal -156) == 0 ? ++ 0 : (sta->last_rx_signal - 156) * 92 / 64; + else +- iwe.u.qual.qual = sta->last_rx_signal - +- sta->last_rx_silence - 35; +- iwe.u.qual.level = HFA384X_LEVEL_TO_dBm(sta->last_rx_signal); +- iwe.u.qual.noise = HFA384X_LEVEL_TO_dBm(sta->last_rx_silence); ++ iwe.u.qual.qual = (sta->last_rx_signal - ++ sta->last_rx_silence) * 92 / 64; ++ iwe.u.qual.level = sta->last_rx_signal; ++ iwe.u.qual.noise = sta->last_rx_silence; + iwe.u.qual.updated = sta->last_rx_updated; + iwe.len = IW_EV_QUAL_LEN; + current_ev = iwe_stream_add_event(info, current_ev, end_buf, +--- a/drivers/net/wireless/hostap/hostap_config.h ++++ b/drivers/net/wireless/hostap/hostap_config.h +@@ -45,4 +45,9 @@ + */ + /* #define PRISM2_NO_STATION_MODES */ + ++/* Enable TX power Setting functions ++ * (min att = -128 , max att = 127) ++ */ ++#define RAW_TXPOWER_SETTING ++ + #endif /* HOSTAP_CONFIG_H */ +--- a/drivers/net/wireless/hostap/hostap.h ++++ b/drivers/net/wireless/hostap/hostap.h +@@ -90,6 +90,7 @@ extern const struct iw_handler_def hosta + extern const struct ethtool_ops prism2_ethtool_ops; + + int hostap_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd); ++int hostap_restore_power(struct net_device *dev); + + + #endif /* HOSTAP_H */ +--- a/drivers/net/wireless/hostap/hostap_hw.c ++++ b/drivers/net/wireless/hostap/hostap_hw.c +@@ -932,6 +932,7 @@ static int hfa384x_set_rid(struct net_de + prism2_hw_reset(dev); + } + ++ hostap_restore_power(dev); + return res; + } + +--- a/drivers/net/wireless/hostap/hostap_info.c ++++ b/drivers/net/wireless/hostap/hostap_info.c +@@ -431,6 +431,11 @@ static void handle_info_queue_linkstatus + } + + /* Get BSSID if we have a valid AP address */ ++ ++ if ( val == HFA384X_LINKSTATUS_CONNECTED || ++ val == HFA384X_LINKSTATUS_DISCONNECTED ) ++ hostap_restore_power(local->dev); ++ + if (connected) { + netif_carrier_on(local->dev); + netif_carrier_on(local->ddev); +--- a/drivers/net/wireless/hostap/hostap_ioctl.c ++++ b/drivers/net/wireless/hostap/hostap_ioctl.c +@@ -1475,23 +1475,20 @@ static int prism2_txpower_hfa386x_to_dBm + val = 255; + + tmp = val; +- tmp >>= 2; + +- return -12 - tmp; ++ return tmp; + } + + static u16 prism2_txpower_dBm_to_hfa386x(int val) + { + signed char tmp; + +- if (val > 20) +- return 128; +- else if (val < -43) ++ if (val > 127) + return 127; ++ else if (val < -128) ++ return 128; + + tmp = val; +- tmp = -12 - tmp; +- tmp <<= 2; + + return (unsigned char) tmp; + } +@@ -4055,3 +4052,35 @@ int hostap_ioctl(struct net_device *dev, + + return ret; + } ++ ++/* BUG FIX: Restore power setting value when lost due to F/W bug */ ++ ++int hostap_restore_power(struct net_device *dev) ++{ ++ struct hostap_interface *iface = netdev_priv(dev); ++ local_info_t *local = iface->local; ++ ++ u16 val; ++ int ret = 0; ++ ++ if (local->txpower_type == PRISM2_TXPOWER_OFF) { ++ val = 0xff; /* use all standby and sleep modes */ ++ ret = local->func->cmd(dev, HFA384X_CMDCODE_WRITEMIF, ++ HFA386X_CR_A_D_TEST_MODES2, ++ &val, NULL); ++ } ++ ++#ifdef RAW_TXPOWER_SETTING ++ if (local->txpower_type == PRISM2_TXPOWER_FIXED) { ++ val = HFA384X_TEST_CFG_BIT_ALC; ++ local->func->cmd(dev, HFA384X_CMDCODE_TEST | ++ (HFA384X_TEST_CFG_BITS << 8), 0, &val, NULL); ++ val = prism2_txpower_dBm_to_hfa386x(local->txpower); ++ ret = (local->func->cmd(dev, HFA384X_CMDCODE_WRITEMIF, ++ HFA386X_CR_MANUAL_TX_POWER, &val, NULL)); ++ } ++#endif /* RAW_TXPOWER_SETTING */ ++ return (ret ? -EOPNOTSUPP : 0); ++} ++ ++EXPORT_SYMBOL(hostap_restore_power); diff --git a/target/linux/generic-2.6/patches-2.6.31/903-stddef_include.patch b/target/linux/generic-2.6/patches-2.6.31/903-stddef_include.patch new file mode 100644 index 0000000000..7fe248d8d3 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/903-stddef_include.patch @@ -0,0 +1,17 @@ +--- a/include/linux/stddef.h ++++ b/include/linux/stddef.h +@@ -16,6 +16,7 @@ enum { + false = 0, + true = 1 + }; ++#endif /* __KERNEL__ */ + + #undef offsetof + #ifdef __compiler_offsetof +@@ -23,6 +24,5 @@ enum { + #else + #define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER) + #endif +-#endif /* __KERNEL__ */ + + #endif diff --git a/target/linux/generic-2.6/patches-2.6.31/905-i386_build.patch b/target/linux/generic-2.6/patches-2.6.31/905-i386_build.patch new file mode 100644 index 0000000000..c701fdaa95 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/905-i386_build.patch @@ -0,0 +1,10 @@ +--- a/arch/x86/boot/tools/build.c ++++ b/arch/x86/boot/tools/build.c +@@ -29,7 +29,6 @@ + #include + #include + #include +-#include + #include + #include + #include diff --git a/target/linux/generic-2.6/patches-2.6.31/920-01-hotpluggable-spi-gpio.patch b/target/linux/generic-2.6/patches-2.6.31/920-01-hotpluggable-spi-gpio.patch new file mode 100644 index 0000000000..b10bc9a65e --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/920-01-hotpluggable-spi-gpio.patch @@ -0,0 +1,60 @@ +Fix spi-gpio for hotplug. + +--mb + + + +--- a/drivers/spi/spi_gpio.c ++++ b/drivers/spi/spi_gpio.c +@@ -218,7 +218,7 @@ static void spi_gpio_cleanup(struct spi_ + spi_bitbang_cleanup(spi); + } + +-static int __init spi_gpio_alloc(unsigned pin, const char *label, bool is_in) ++static int __devinit spi_gpio_alloc(unsigned pin, const char *label, bool is_in) + { + int value; + +@@ -232,7 +232,7 @@ static int __init spi_gpio_alloc(unsigne + return value; + } + +-static int __init ++static int __devinit + spi_gpio_request(struct spi_gpio_platform_data *pdata, const char *label) + { + int value; +@@ -261,7 +261,7 @@ done: + return value; + } + +-static int __init spi_gpio_probe(struct platform_device *pdev) ++static int __devinit spi_gpio_probe(struct platform_device *pdev) + { + int status; + struct spi_master *master; +@@ -317,7 +317,7 @@ gpio_free: + return status; + } + +-static int __exit spi_gpio_remove(struct platform_device *pdev) ++static int __devexit spi_gpio_remove(struct platform_device *pdev) + { + struct spi_gpio *spi_gpio; + struct spi_gpio_platform_data *pdata; +@@ -344,12 +344,13 @@ MODULE_ALIAS("platform:" DRIVER_NAME); + static struct platform_driver spi_gpio_driver = { + .driver.name = DRIVER_NAME, + .driver.owner = THIS_MODULE, +- .remove = __exit_p(spi_gpio_remove), ++ .probe = spi_gpio_probe, ++ .remove = __devexit_p(spi_gpio_remove), + }; + + static int __init spi_gpio_init(void) + { +- return platform_driver_probe(&spi_gpio_driver, spi_gpio_probe); ++ return platform_driver_register(&spi_gpio_driver); + } + module_init(spi_gpio_init); + diff --git a/target/linux/generic-2.6/patches-2.6.31/920-04-spi-gpio-implement-spi-delay.patch b/target/linux/generic-2.6/patches-2.6.31/920-04-spi-gpio-implement-spi-delay.patch new file mode 100644 index 0000000000..19032881b8 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/920-04-spi-gpio-implement-spi-delay.patch @@ -0,0 +1,58 @@ +Implement the SPI-GPIO delay function for busses that need speed limitation. + +--mb + + + +--- a/drivers/spi/spi_gpio.c ++++ b/drivers/spi/spi_gpio.c +@@ -21,6 +21,7 @@ + #include + #include + #include ++#include + + #include + #include +@@ -69,6 +70,7 @@ struct spi_gpio { + * #define SPI_MOSI_GPIO 120 + * #define SPI_SCK_GPIO 121 + * #define SPI_N_CHIPSEL 4 ++ * #undef NEED_SPIDELAY + * #include "spi_gpio.c" + */ + +@@ -76,6 +78,7 @@ struct spi_gpio { + #define DRIVER_NAME "spi_gpio" + + #define GENERIC_BITBANG /* vs tight inlines */ ++#define NEED_SPIDELAY 1 + + /* all functions referencing these symbols must define pdata */ + #define SPI_MISO_GPIO ((pdata)->miso) +@@ -120,12 +123,20 @@ static inline int getmiso(const struct s + #undef pdata + + /* +- * NOTE: this clocks "as fast as we can". It "should" be a function of the +- * requested device clock. Software overhead means we usually have trouble +- * reaching even one Mbit/sec (except when we can inline bitops), so for now +- * we'll just assume we never need additional per-bit slowdowns. ++ * NOTE: to clock "as fast as we can", set spi_device.max_speed_hz ++ * and spi_transfer.speed_hz to 0. ++ * Otherwise this is a function of the requested device clock. ++ * Software overhead means we usually have trouble ++ * reaching even one Mbit/sec (except when we can inline bitops). So on small ++ * embedded devices with fast SPI slaves you usually don't need a delay. + */ +-#define spidelay(nsecs) do {} while (0) ++static inline void spidelay(unsigned nsecs) ++{ ++#ifdef NEED_SPIDELAY ++ if (unlikely(nsecs)) ++ ndelay(nsecs); ++#endif /* NEED_SPIDELAY */ ++} + + #define EXPAND_BITBANG_TXRX + #include diff --git a/target/linux/generic-2.6/patches-2.6.31/921-gpio_spi_driver.patch b/target/linux/generic-2.6/patches-2.6.31/921-gpio_spi_driver.patch new file mode 100644 index 0000000000..90e40e7054 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/921-gpio_spi_driver.patch @@ -0,0 +1,366 @@ +THIS CODE IS DEPRECATED. + +Please use the new mainline SPI-GPIO driver, as of 2.6.29. + +--mb + + + +--- /dev/null ++++ b/include/linux/spi/spi_gpio_old.h +@@ -0,0 +1,73 @@ ++/* ++ * spi_gpio interface to platform code ++ * ++ * Copyright (c) 2008 Piotr Skamruk ++ * Copyright (c) 2008 Michael Buesch ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++#ifndef _LINUX_SPI_SPI_GPIO ++#define _LINUX_SPI_SPI_GPIO ++ ++#include ++#include ++ ++ ++/** ++ * struct spi_gpio_platform_data - Data definitions for a SPI-GPIO device. ++ * ++ * This structure holds information about a GPIO-based SPI device. ++ * ++ * @pin_clk: The GPIO pin number of the CLOCK pin. ++ * ++ * @pin_miso: The GPIO pin number of the MISO pin. ++ * ++ * @pin_mosi: The GPIO pin number of the MOSI pin. ++ * ++ * @pin_cs: The GPIO pin number of the CHIPSELECT pin. ++ * ++ * @cs_activelow: If true, the chip is selected when the CS line is low. ++ * ++ * @no_spi_delay: If true, no delay is done in the lowlevel bitbanging. ++ * Note that doing no delay is not standards compliant, ++ * but it might be needed to speed up transfers on some ++ * slow embedded machines. ++ * ++ * @boardinfo_setup: This callback is called after the ++ * SPI master device was registered, but before the ++ * device is registered. ++ * @boardinfo_setup_data: Data argument passed to boardinfo_setup(). ++ */ ++struct spi_gpio_platform_data { ++ unsigned int pin_clk; ++ unsigned int pin_miso; ++ unsigned int pin_mosi; ++ unsigned int pin_cs; ++ bool cs_activelow; ++ bool no_spi_delay; ++ int (*boardinfo_setup)(struct spi_board_info *bi, ++ struct spi_master *master, ++ void *data); ++ void *boardinfo_setup_data; ++}; ++ ++/** ++ * SPI_GPIO_PLATDEV_NAME - The platform device name string. ++ * ++ * The name string that has to be used for platform_device_alloc ++ * when allocating a spi-gpio device. ++ */ ++#define SPI_GPIO_PLATDEV_NAME "spi-gpio" ++ ++/** ++ * spi_gpio_next_id - Get another platform device ID number. ++ * ++ * This returns the next platform device ID number that has to be used ++ * for platform_device_alloc. The ID is opaque and should not be used for ++ * anything else. ++ */ ++int spi_gpio_next_id(void); ++ ++#endif /* _LINUX_SPI_SPI_GPIO */ +--- /dev/null ++++ b/drivers/spi/spi_gpio_old.c +@@ -0,0 +1,251 @@ ++/* ++ * Bitbanging SPI bus driver using GPIO API ++ * ++ * Copyright (c) 2008 Piotr Skamruk ++ * Copyright (c) 2008 Michael Buesch ++ * ++ * based on spi_s3c2410_gpio.c ++ * Copyright (c) 2006 Ben Dooks ++ * Copyright (c) 2006 Simtec Electronics ++ * and on i2c-gpio.c ++ * Copyright (C) 2007 Atmel Corporation ++ * ++ * This program is free software; you can redistribute it and/or modify ++ * it under the terms of the GNU General Public License version 2 as ++ * published by the Free Software Foundation. ++ */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++ ++struct spi_gpio { ++ struct spi_bitbang bitbang; ++ struct spi_gpio_platform_data *info; ++ struct platform_device *pdev; ++ struct spi_board_info bi; ++}; ++ ++ ++static inline struct spi_gpio *spidev_to_sg(struct spi_device *dev) ++{ ++ return dev->controller_data; ++} ++ ++static inline void setsck(struct spi_device *dev, int val) ++{ ++ struct spi_gpio *sp = spidev_to_sg(dev); ++ gpio_set_value(sp->info->pin_clk, val ? 1 : 0); ++} ++ ++static inline void setmosi(struct spi_device *dev, int val) ++{ ++ struct spi_gpio *sp = spidev_to_sg(dev); ++ gpio_set_value(sp->info->pin_mosi, val ? 1 : 0); ++} ++ ++static inline u32 getmiso(struct spi_device *dev) ++{ ++ struct spi_gpio *sp = spidev_to_sg(dev); ++ return gpio_get_value(sp->info->pin_miso) ? 1 : 0; ++} ++ ++static inline void do_spidelay(struct spi_device *dev, unsigned nsecs) ++{ ++ struct spi_gpio *sp = spidev_to_sg(dev); ++ ++ if (!sp->info->no_spi_delay) ++ ndelay(nsecs); ++} ++ ++#define spidelay(nsecs) do { \ ++ /* Steal the spi_device pointer from our caller. \ ++ * The bitbang-API should probably get fixed here... */ \ ++ do_spidelay(spi, nsecs); \ ++ } while (0) ++ ++#define EXPAND_BITBANG_TXRX ++#include ++ ++static u32 spi_gpio_txrx_mode0(struct spi_device *spi, ++ unsigned nsecs, u32 word, u8 bits) ++{ ++ return bitbang_txrx_be_cpha0(spi, nsecs, 0, word, bits); ++} ++ ++static u32 spi_gpio_txrx_mode1(struct spi_device *spi, ++ unsigned nsecs, u32 word, u8 bits) ++{ ++ return bitbang_txrx_be_cpha1(spi, nsecs, 0, word, bits); ++} ++ ++static u32 spi_gpio_txrx_mode2(struct spi_device *spi, ++ unsigned nsecs, u32 word, u8 bits) ++{ ++ return bitbang_txrx_be_cpha0(spi, nsecs, 1, word, bits); ++} ++ ++static u32 spi_gpio_txrx_mode3(struct spi_device *spi, ++ unsigned nsecs, u32 word, u8 bits) ++{ ++ return bitbang_txrx_be_cpha1(spi, nsecs, 1, word, bits); ++} ++ ++static void spi_gpio_chipselect(struct spi_device *dev, int on) ++{ ++ struct spi_gpio *sp = spidev_to_sg(dev); ++ ++ if (sp->info->cs_activelow) ++ on = !on; ++ gpio_set_value(sp->info->pin_cs, on ? 1 : 0); ++} ++ ++static int spi_gpio_probe(struct platform_device *pdev) ++{ ++ struct spi_master *master; ++ struct spi_gpio_platform_data *pdata; ++ struct spi_gpio *sp; ++ struct spi_device *spidev; ++ int err; ++ ++ pdata = pdev->dev.platform_data; ++ if (!pdata) ++ return -ENXIO; ++ ++ err = -ENOMEM; ++ master = spi_alloc_master(&pdev->dev, sizeof(struct spi_gpio)); ++ if (!master) ++ goto err_alloc_master; ++ ++ sp = spi_master_get_devdata(master); ++ platform_set_drvdata(pdev, sp); ++ sp->info = pdata; ++ ++ err = gpio_request(pdata->pin_clk, "spi_clock"); ++ if (err) ++ goto err_request_clk; ++ err = gpio_request(pdata->pin_mosi, "spi_mosi"); ++ if (err) ++ goto err_request_mosi; ++ err = gpio_request(pdata->pin_miso, "spi_miso"); ++ if (err) ++ goto err_request_miso; ++ err = gpio_request(pdata->pin_cs, "spi_cs"); ++ if (err) ++ goto err_request_cs; ++ ++ sp->bitbang.master = spi_master_get(master); ++ sp->bitbang.master->bus_num = -1; ++ sp->bitbang.master->num_chipselect = 1; ++ sp->bitbang.chipselect = spi_gpio_chipselect; ++ sp->bitbang.txrx_word[SPI_MODE_0] = spi_gpio_txrx_mode0; ++ sp->bitbang.txrx_word[SPI_MODE_1] = spi_gpio_txrx_mode1; ++ sp->bitbang.txrx_word[SPI_MODE_2] = spi_gpio_txrx_mode2; ++ sp->bitbang.txrx_word[SPI_MODE_3] = spi_gpio_txrx_mode3; ++ ++ gpio_direction_output(pdata->pin_clk, 0); ++ gpio_direction_output(pdata->pin_mosi, 0); ++ gpio_direction_output(pdata->pin_cs, ++ pdata->cs_activelow ? 1 : 0); ++ gpio_direction_input(pdata->pin_miso); ++ ++ err = spi_bitbang_start(&sp->bitbang); ++ if (err) ++ goto err_no_bitbang; ++ err = pdata->boardinfo_setup(&sp->bi, master, ++ pdata->boardinfo_setup_data); ++ if (err) ++ goto err_bi_setup; ++ sp->bi.controller_data = sp; ++ spidev = spi_new_device(master, &sp->bi); ++ if (!spidev) ++ goto err_new_dev; ++ ++ return 0; ++ ++err_new_dev: ++err_bi_setup: ++ spi_bitbang_stop(&sp->bitbang); ++err_no_bitbang: ++ spi_master_put(sp->bitbang.master); ++ gpio_free(pdata->pin_cs); ++err_request_cs: ++ gpio_free(pdata->pin_miso); ++err_request_miso: ++ gpio_free(pdata->pin_mosi); ++err_request_mosi: ++ gpio_free(pdata->pin_clk); ++err_request_clk: ++ kfree(master); ++ ++err_alloc_master: ++ return err; ++} ++ ++static int __devexit spi_gpio_remove(struct platform_device *pdev) ++{ ++ struct spi_gpio *sp; ++ struct spi_gpio_platform_data *pdata; ++ ++ pdata = pdev->dev.platform_data; ++ sp = platform_get_drvdata(pdev); ++ ++ gpio_free(pdata->pin_clk); ++ gpio_free(pdata->pin_mosi); ++ gpio_free(pdata->pin_miso); ++ gpio_free(pdata->pin_cs); ++ spi_bitbang_stop(&sp->bitbang); ++ spi_master_put(sp->bitbang.master); ++ ++ return 0; ++} ++ ++static struct platform_driver spi_gpio_driver = { ++ .driver = { ++ .name = SPI_GPIO_PLATDEV_NAME, ++ .owner = THIS_MODULE, ++ }, ++ .probe = spi_gpio_probe, ++ .remove = __devexit_p(spi_gpio_remove), ++}; ++ ++int spi_gpio_next_id(void) ++{ ++ static atomic_t counter = ATOMIC_INIT(-1); ++ ++ return atomic_inc_return(&counter); ++} ++EXPORT_SYMBOL(spi_gpio_next_id); ++ ++static int __init spi_gpio_init(void) ++{ ++ int err; ++ ++ err = platform_driver_register(&spi_gpio_driver); ++ if (err) ++ printk(KERN_ERR "spi-gpio: register failed: %d\n", err); ++ ++ return err; ++} ++module_init(spi_gpio_init); ++ ++static void __exit spi_gpio_exit(void) ++{ ++ platform_driver_unregister(&spi_gpio_driver); ++} ++module_exit(spi_gpio_exit); ++ ++MODULE_AUTHOR("Piot Skamruk "); ++MODULE_AUTHOR("Michael Buesch"); ++MODULE_DESCRIPTION("Platform independent GPIO bitbanging SPI driver"); ++MODULE_LICENSE("GPL v2"); +--- a/drivers/spi/Kconfig ++++ b/drivers/spi/Kconfig +@@ -116,6 +116,15 @@ config SPI_GPIO + GPIO operations, you should be able to leverage that for better + speed with a custom version of this driver; see the source code. + ++config SPI_GPIO_OLD ++ tristate "Old GPIO API based bitbanging SPI controller (DEPRECATED)" ++ depends on SPI_MASTER && GENERIC_GPIO ++ select SPI_BITBANG ++ help ++ This code is deprecated. Please use the new mainline SPI-GPIO driver. ++ ++ If unsure, say N. ++ + config SPI_IMX + tristate "Freescale iMX SPI controller" + depends on ARCH_MX1 && EXPERIMENTAL +--- a/drivers/spi/Makefile ++++ b/drivers/spi/Makefile +@@ -17,6 +17,7 @@ obj-$(CONFIG_SPI_BITBANG) += spi_bitban + obj-$(CONFIG_SPI_AU1550) += au1550_spi.o + obj-$(CONFIG_SPI_BUTTERFLY) += spi_butterfly.o + obj-$(CONFIG_SPI_GPIO) += spi_gpio.o ++obj-$(CONFIG_SPI_GPIO_OLD) += spi_gpio_old.o + obj-$(CONFIG_SPI_IMX) += spi_imx.o + obj-$(CONFIG_SPI_LM70_LLP) += spi_lm70llp.o + obj-$(CONFIG_SPI_PXA2XX) += pxa2xx_spi.o diff --git a/target/linux/generic-2.6/patches-2.6.31/922-gpiommc.patch b/target/linux/generic-2.6/patches-2.6.31/922-gpiommc.patch new file mode 100644 index 0000000000..55c97233b4 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/922-gpiommc.patch @@ -0,0 +1,843 @@ +--- /dev/null ++++ b/drivers/mmc/host/gpiommc.c +@@ -0,0 +1,608 @@ ++/* ++ * Driver an MMC/SD card on a bitbanging GPIO SPI bus. ++ * This module hooks up the mmc_spi and spi_gpio modules and also ++ * provides a configfs interface. ++ * ++ * Copyright 2008 Michael Buesch ++ * ++ * Licensed under the GNU/GPL. See COPYING for details. ++ */ ++ ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++#include ++ ++ ++#define PFX "gpio-mmc: " ++ ++ ++struct gpiommc_device { ++ struct platform_device *pdev; ++ struct platform_device *spi_pdev; ++ struct spi_board_info boardinfo; ++}; ++ ++ ++MODULE_DESCRIPTION("GPIO based MMC driver"); ++MODULE_AUTHOR("Michael Buesch"); ++MODULE_LICENSE("GPL"); ++ ++ ++static int gpiommc_boardinfo_setup(struct spi_board_info *bi, ++ struct spi_master *master, ++ void *data) ++{ ++ struct gpiommc_device *d = data; ++ struct gpiommc_platform_data *pdata = d->pdev->dev.platform_data; ++ ++ /* Bind the SPI master to the MMC-SPI host driver. */ ++ strlcpy(bi->modalias, "mmc_spi", sizeof(bi->modalias)); ++ ++ bi->max_speed_hz = pdata->max_bus_speed; ++ bi->bus_num = master->bus_num; ++ bi->mode = pdata->mode; ++ ++ return 0; ++} ++ ++static int gpiommc_probe(struct platform_device *pdev) ++{ ++ struct gpiommc_platform_data *mmc_pdata = pdev->dev.platform_data; ++ struct spi_gpio_platform_data spi_pdata; ++ struct gpiommc_device *d; ++ int err; ++ ++ err = -ENXIO; ++ if (!mmc_pdata) ++ goto error; ++ ++#ifdef CONFIG_MMC_SPI_MODULE ++ err = request_module("mmc_spi"); ++ if (err) { ++ printk(KERN_WARNING PFX ++ "Failed to request mmc_spi module.\n"); ++ } ++#endif /* CONFIG_MMC_SPI_MODULE */ ++ ++ /* Allocate the GPIO-MMC device */ ++ err = -ENOMEM; ++ d = kzalloc(sizeof(*d), GFP_KERNEL); ++ if (!d) ++ goto error; ++ d->pdev = pdev; ++ ++ /* Create the SPI-GPIO device */ ++ d->spi_pdev = platform_device_alloc(SPI_GPIO_PLATDEV_NAME, ++ spi_gpio_next_id()); ++ if (!d->spi_pdev) ++ goto err_free_d; ++ ++ memset(&spi_pdata, 0, sizeof(spi_pdata)); ++ spi_pdata.pin_clk = mmc_pdata->pins.gpio_clk; ++ spi_pdata.pin_miso = mmc_pdata->pins.gpio_do; ++ spi_pdata.pin_mosi = mmc_pdata->pins.gpio_di; ++ spi_pdata.pin_cs = mmc_pdata->pins.gpio_cs; ++ spi_pdata.cs_activelow = mmc_pdata->pins.cs_activelow; ++ spi_pdata.no_spi_delay = mmc_pdata->no_spi_delay; ++ spi_pdata.boardinfo_setup = gpiommc_boardinfo_setup; ++ spi_pdata.boardinfo_setup_data = d; ++ ++ err = platform_device_add_data(d->spi_pdev, &spi_pdata, ++ sizeof(spi_pdata)); ++ if (err) ++ goto err_free_pdev; ++ err = platform_device_add(d->spi_pdev); ++ if (err) ++ goto err_free_pdata; ++ platform_set_drvdata(pdev, d); ++ ++ printk(KERN_INFO PFX "MMC-Card \"%s\" " ++ "attached to GPIO pins di=%u, do=%u, clk=%u, cs=%u\n", ++ mmc_pdata->name, mmc_pdata->pins.gpio_di, ++ mmc_pdata->pins.gpio_do, ++ mmc_pdata->pins.gpio_clk, ++ mmc_pdata->pins.gpio_cs); ++ ++ return 0; ++ ++err_free_pdata: ++ kfree(d->spi_pdev->dev.platform_data); ++ d->spi_pdev->dev.platform_data = NULL; ++err_free_pdev: ++ platform_device_put(d->spi_pdev); ++err_free_d: ++ kfree(d); ++error: ++ return err; ++} ++ ++static int gpiommc_remove(struct platform_device *pdev) ++{ ++ struct gpiommc_device *d = platform_get_drvdata(pdev); ++ struct gpiommc_platform_data *pdata = d->pdev->dev.platform_data; ++ ++ platform_device_unregister(d->spi_pdev); ++ printk(KERN_INFO PFX "GPIO based MMC-Card \"%s\" removed\n", ++ pdata->name); ++ platform_device_put(d->spi_pdev); ++ ++ return 0; ++} ++ ++#ifdef CONFIG_GPIOMMC_CONFIGFS ++ ++/* A device that was created through configfs */ ++struct gpiommc_configfs_device { ++ struct config_item item; ++ /* The platform device, after registration. */ ++ struct platform_device *pdev; ++ /* The configuration */ ++ struct gpiommc_platform_data pdata; ++}; ++ ++#define GPIO_INVALID -1 ++ ++static inline bool gpiommc_is_registered(struct gpiommc_configfs_device *dev) ++{ ++ return (dev->pdev != NULL); ++} ++ ++static inline struct gpiommc_configfs_device *ci_to_gpiommc(struct config_item *item) ++{ ++ return item ? container_of(item, struct gpiommc_configfs_device, item) : NULL; ++} ++ ++static struct configfs_attribute gpiommc_attr_DI = { ++ .ca_owner = THIS_MODULE, ++ .ca_name = "gpio_data_in", ++ .ca_mode = S_IRUGO | S_IWUSR, ++}; ++ ++static struct configfs_attribute gpiommc_attr_DO = { ++ .ca_owner = THIS_MODULE, ++ .ca_name = "gpio_data_out", ++ .ca_mode = S_IRUGO | S_IWUSR, ++}; ++ ++static struct configfs_attribute gpiommc_attr_CLK = { ++ .ca_owner = THIS_MODULE, ++ .ca_name = "gpio_clock", ++ .ca_mode = S_IRUGO | S_IWUSR, ++}; ++ ++static struct configfs_attribute gpiommc_attr_CS = { ++ .ca_owner = THIS_MODULE, ++ .ca_name = "gpio_chipselect", ++ .ca_mode = S_IRUGO | S_IWUSR, ++}; ++ ++static struct configfs_attribute gpiommc_attr_CS_activelow = { ++ .ca_owner = THIS_MODULE, ++ .ca_name = "gpio_chipselect_activelow", ++ .ca_mode = S_IRUGO | S_IWUSR, ++}; ++ ++static struct configfs_attribute gpiommc_attr_spimode = { ++ .ca_owner = THIS_MODULE, ++ .ca_name = "spi_mode", ++ .ca_mode = S_IRUGO | S_IWUSR, ++}; ++ ++static struct configfs_attribute gpiommc_attr_spidelay = { ++ .ca_owner = THIS_MODULE, ++ .ca_name = "spi_delay", ++ .ca_mode = S_IRUGO | S_IWUSR, ++}; ++ ++static struct configfs_attribute gpiommc_attr_max_bus_speed = { ++ .ca_owner = THIS_MODULE, ++ .ca_name = "max_bus_speed", ++ .ca_mode = S_IRUGO | S_IWUSR, ++}; ++ ++static struct configfs_attribute gpiommc_attr_register = { ++ .ca_owner = THIS_MODULE, ++ .ca_name = "register", ++ .ca_mode = S_IRUGO | S_IWUSR, ++}; ++ ++static struct configfs_attribute *gpiommc_config_attrs[] = { ++ &gpiommc_attr_DI, ++ &gpiommc_attr_DO, ++ &gpiommc_attr_CLK, ++ &gpiommc_attr_CS, ++ &gpiommc_attr_CS_activelow, ++ &gpiommc_attr_spimode, ++ &gpiommc_attr_spidelay, ++ &gpiommc_attr_max_bus_speed, ++ &gpiommc_attr_register, ++ NULL, ++}; ++ ++static ssize_t gpiommc_config_attr_show(struct config_item *item, ++ struct configfs_attribute *attr, ++ char *page) ++{ ++ struct gpiommc_configfs_device *dev = ci_to_gpiommc(item); ++ ssize_t count = 0; ++ unsigned int gpio; ++ int err = 0; ++ ++ if (attr == &gpiommc_attr_DI) { ++ gpio = dev->pdata.pins.gpio_di; ++ if (gpio == GPIO_INVALID) ++ count = snprintf(page, PAGE_SIZE, "not configured\n"); ++ else ++ count = snprintf(page, PAGE_SIZE, "%u\n", gpio); ++ goto out; ++ } ++ if (attr == &gpiommc_attr_DO) { ++ gpio = dev->pdata.pins.gpio_do; ++ if (gpio == GPIO_INVALID) ++ count = snprintf(page, PAGE_SIZE, "not configured\n"); ++ else ++ count = snprintf(page, PAGE_SIZE, "%u\n", gpio); ++ goto out; ++ } ++ if (attr == &gpiommc_attr_CLK) { ++ gpio = dev->pdata.pins.gpio_clk; ++ if (gpio == GPIO_INVALID) ++ count = snprintf(page, PAGE_SIZE, "not configured\n"); ++ else ++ count = snprintf(page, PAGE_SIZE, "%u\n", gpio); ++ goto out; ++ } ++ if (attr == &gpiommc_attr_CS) { ++ gpio = dev->pdata.pins.gpio_cs; ++ if (gpio == GPIO_INVALID) ++ count = snprintf(page, PAGE_SIZE, "not configured\n"); ++ else ++ count = snprintf(page, PAGE_SIZE, "%u\n", gpio); ++ goto out; ++ } ++ if (attr == &gpiommc_attr_CS_activelow) { ++ count = snprintf(page, PAGE_SIZE, "%u\n", ++ dev->pdata.pins.cs_activelow); ++ goto out; ++ } ++ if (attr == &gpiommc_attr_spimode) { ++ count = snprintf(page, PAGE_SIZE, "%u\n", ++ dev->pdata.mode); ++ goto out; ++ } ++ if (attr == &gpiommc_attr_spidelay) { ++ count = snprintf(page, PAGE_SIZE, "%u\n", ++ !dev->pdata.no_spi_delay); ++ goto out; ++ } ++ if (attr == &gpiommc_attr_max_bus_speed) { ++ count = snprintf(page, PAGE_SIZE, "%u\n", ++ dev->pdata.max_bus_speed); ++ goto out; ++ } ++ if (attr == &gpiommc_attr_register) { ++ count = snprintf(page, PAGE_SIZE, "%u\n", ++ gpiommc_is_registered(dev)); ++ goto out; ++ } ++ WARN_ON(1); ++ err = -ENOSYS; ++out: ++ return err ? err : count; ++} ++ ++static int gpiommc_do_register(struct gpiommc_configfs_device *dev, ++ const char *name) ++{ ++ int err; ++ ++ if (gpiommc_is_registered(dev)) ++ return 0; ++ ++ if (!gpio_is_valid(dev->pdata.pins.gpio_di) || ++ !gpio_is_valid(dev->pdata.pins.gpio_do) || ++ !gpio_is_valid(dev->pdata.pins.gpio_clk) || ++ !gpio_is_valid(dev->pdata.pins.gpio_cs)) { ++ printk(KERN_ERR PFX ++ "configfs: Invalid GPIO pin number(s)\n"); ++ return -EINVAL; ++ } ++ ++ strlcpy(dev->pdata.name, name, ++ sizeof(dev->pdata.name)); ++ ++ dev->pdev = platform_device_alloc(GPIOMMC_PLATDEV_NAME, ++ gpiommc_next_id()); ++ if (!dev->pdev) ++ return -ENOMEM; ++ err = platform_device_add_data(dev->pdev, &dev->pdata, ++ sizeof(dev->pdata)); ++ if (err) { ++ platform_device_put(dev->pdev); ++ return err; ++ } ++ err = platform_device_add(dev->pdev); ++ if (err) { ++ platform_device_put(dev->pdev); ++ return err; ++ } ++ ++ return 0; ++} ++ ++static void gpiommc_do_unregister(struct gpiommc_configfs_device *dev) ++{ ++ if (!gpiommc_is_registered(dev)) ++ return; ++ ++ platform_device_unregister(dev->pdev); ++ dev->pdev = NULL; ++} ++ ++static ssize_t gpiommc_config_attr_store(struct config_item *item, ++ struct configfs_attribute *attr, ++ const char *page, size_t count) ++{ ++ struct gpiommc_configfs_device *dev = ci_to_gpiommc(item); ++ int err = -EINVAL; ++ unsigned long data; ++ ++ if (attr == &gpiommc_attr_register) { ++ err = strict_strtoul(page, 10, &data); ++ if (err) ++ goto out; ++ err = -EINVAL; ++ if (data == 1) ++ err = gpiommc_do_register(dev, item->ci_name); ++ if (data == 0) { ++ gpiommc_do_unregister(dev); ++ err = 0; ++ } ++ goto out; ++ } ++ ++ if (gpiommc_is_registered(dev)) { ++ /* The rest of the config parameters can only be set ++ * as long as the device is not registered, yet. */ ++ err = -EBUSY; ++ goto out; ++ } ++ ++ if (attr == &gpiommc_attr_DI) { ++ err = strict_strtoul(page, 10, &data); ++ if (err) ++ goto out; ++ err = -EINVAL; ++ if (!gpio_is_valid(data)) ++ goto out; ++ dev->pdata.pins.gpio_di = data; ++ err = 0; ++ goto out; ++ } ++ if (attr == &gpiommc_attr_DO) { ++ err = strict_strtoul(page, 10, &data); ++ if (err) ++ goto out; ++ err = -EINVAL; ++ if (!gpio_is_valid(data)) ++ goto out; ++ dev->pdata.pins.gpio_do = data; ++ err = 0; ++ goto out; ++ } ++ if (attr == &gpiommc_attr_CLK) { ++ err = strict_strtoul(page, 10, &data); ++ if (err) ++ goto out; ++ err = -EINVAL; ++ if (!gpio_is_valid(data)) ++ goto out; ++ dev->pdata.pins.gpio_clk = data; ++ err = 0; ++ goto out; ++ } ++ if (attr == &gpiommc_attr_CS) { ++ err = strict_strtoul(page, 10, &data); ++ if (err) ++ goto out; ++ err = -EINVAL; ++ if (!gpio_is_valid(data)) ++ goto out; ++ dev->pdata.pins.gpio_cs = data; ++ err = 0; ++ goto out; ++ } ++ if (attr == &gpiommc_attr_CS_activelow) { ++ err = strict_strtoul(page, 10, &data); ++ if (err) ++ goto out; ++ err = -EINVAL; ++ if (data != 0 && data != 1) ++ goto out; ++ dev->pdata.pins.cs_activelow = data; ++ err = 0; ++ goto out; ++ } ++ if (attr == &gpiommc_attr_spimode) { ++ err = strict_strtoul(page, 10, &data); ++ if (err) ++ goto out; ++ err = -EINVAL; ++ switch (data) { ++ case 0: ++ dev->pdata.mode = SPI_MODE_0; ++ break; ++ case 1: ++ dev->pdata.mode = SPI_MODE_1; ++ break; ++ case 2: ++ dev->pdata.mode = SPI_MODE_2; ++ break; ++ case 3: ++ dev->pdata.mode = SPI_MODE_3; ++ break; ++ default: ++ goto out; ++ } ++ err = 0; ++ goto out; ++ } ++ if (attr == &gpiommc_attr_spidelay) { ++ err = strict_strtoul(page, 10, &data); ++ if (err) ++ goto out; ++ err = -EINVAL; ++ if (data != 0 && data != 1) ++ goto out; ++ dev->pdata.no_spi_delay = !data; ++ err = 0; ++ goto out; ++ } ++ if (attr == &gpiommc_attr_max_bus_speed) { ++ err = strict_strtoul(page, 10, &data); ++ if (err) ++ goto out; ++ err = -EINVAL; ++ if (data > UINT_MAX) ++ goto out; ++ dev->pdata.max_bus_speed = data; ++ err = 0; ++ goto out; ++ } ++ WARN_ON(1); ++ err = -ENOSYS; ++out: ++ return err ? err : count; ++} ++ ++static void gpiommc_config_item_release(struct config_item *item) ++{ ++ struct gpiommc_configfs_device *dev = ci_to_gpiommc(item); ++ ++ kfree(dev); ++} ++ ++static struct configfs_item_operations gpiommc_config_item_ops = { ++ .release = gpiommc_config_item_release, ++ .show_attribute = gpiommc_config_attr_show, ++ .store_attribute = gpiommc_config_attr_store, ++}; ++ ++static struct config_item_type gpiommc_dev_ci_type = { ++ .ct_item_ops = &gpiommc_config_item_ops, ++ .ct_attrs = gpiommc_config_attrs, ++ .ct_owner = THIS_MODULE, ++}; ++ ++static struct config_item *gpiommc_make_item(struct config_group *group, ++ const char *name) ++{ ++ struct gpiommc_configfs_device *dev; ++ ++ if (strlen(name) > GPIOMMC_MAX_NAMELEN) { ++ printk(KERN_ERR PFX "configfs: device name too long\n"); ++ return NULL; ++ } ++ ++ dev = kzalloc(sizeof(*dev), GFP_KERNEL); ++ if (!dev) ++ return NULL; ++ ++ config_item_init_type_name(&dev->item, name, ++ &gpiommc_dev_ci_type); ++ ++ /* Assign default configuration */ ++ dev->pdata.pins.gpio_di = GPIO_INVALID; ++ dev->pdata.pins.gpio_do = GPIO_INVALID; ++ dev->pdata.pins.gpio_clk = GPIO_INVALID; ++ dev->pdata.pins.gpio_cs = GPIO_INVALID; ++ dev->pdata.pins.cs_activelow = 1; ++ dev->pdata.mode = SPI_MODE_0; ++ dev->pdata.no_spi_delay = 0; ++ dev->pdata.max_bus_speed = 5000000; /* 5 MHz */ ++ ++ return &(dev->item); ++} ++ ++static void gpiommc_drop_item(struct config_group *group, ++ struct config_item *item) ++{ ++ struct gpiommc_configfs_device *dev = ci_to_gpiommc(item); ++ ++ gpiommc_do_unregister(dev); ++ kfree(dev); ++} ++ ++static struct configfs_group_operations gpiommc_ct_group_ops = { ++ .make_item = gpiommc_make_item, ++ .drop_item = gpiommc_drop_item, ++}; ++ ++static struct config_item_type gpiommc_ci_type = { ++ .ct_group_ops = &gpiommc_ct_group_ops, ++ .ct_owner = THIS_MODULE, ++}; ++ ++static struct configfs_subsystem gpiommc_subsys = { ++ .su_group = { ++ .cg_item = { ++ .ci_namebuf = GPIOMMC_PLATDEV_NAME, ++ .ci_type = &gpiommc_ci_type, ++ }, ++ }, ++ .su_mutex = __MUTEX_INITIALIZER(gpiommc_subsys.su_mutex), ++}; ++ ++#endif /* CONFIG_GPIOMMC_CONFIGFS */ ++ ++static struct platform_driver gpiommc_plat_driver = { ++ .probe = gpiommc_probe, ++ .remove = gpiommc_remove, ++ .driver = { ++ .name = GPIOMMC_PLATDEV_NAME, ++ .owner = THIS_MODULE, ++ }, ++}; ++ ++int gpiommc_next_id(void) ++{ ++ static atomic_t counter = ATOMIC_INIT(-1); ++ ++ return atomic_inc_return(&counter); ++} ++EXPORT_SYMBOL(gpiommc_next_id); ++ ++static int __init gpiommc_modinit(void) ++{ ++ int err; ++ ++ err = platform_driver_register(&gpiommc_plat_driver); ++ if (err) ++ return err; ++ ++#ifdef CONFIG_GPIOMMC_CONFIGFS ++ config_group_init(&gpiommc_subsys.su_group); ++ err = configfs_register_subsystem(&gpiommc_subsys); ++ if (err) { ++ platform_driver_unregister(&gpiommc_plat_driver); ++ return err; ++ } ++#endif /* CONFIG_GPIOMMC_CONFIGFS */ ++ ++ return 0; ++} ++module_init(gpiommc_modinit); ++ ++static void __exit gpiommc_modexit(void) ++{ ++#ifdef CONFIG_GPIOMMC_CONFIGFS ++ configfs_unregister_subsystem(&gpiommc_subsys); ++#endif ++ platform_driver_unregister(&gpiommc_plat_driver); ++} ++module_exit(gpiommc_modexit); +--- a/drivers/mmc/host/Kconfig ++++ b/drivers/mmc/host/Kconfig +@@ -278,6 +278,31 @@ config MMC_TMIO + This provides support for the SD/MMC cell found in TC6393XB, + T7L66XB and also HTC ASIC3 + ++config GPIOMMC ++ tristate "MMC/SD over GPIO-based SPI" ++ depends on MMC && MMC_SPI && SPI_GPIO ++ help ++ This driver hooks up the mmc_spi and spi_gpio modules so that ++ MMC/SD cards can be used on a GPIO based bus by bitbanging ++ the SPI protocol in software. ++ ++ This driver provides a configfs interface to dynamically create ++ and destroy GPIO-based MMC/SD card devices. It also provides ++ a platform device interface API. ++ See Documentation/gpiommc.txt for details. ++ ++ The module will be called gpiommc. ++ ++ If unsure, say N. ++ ++config GPIOMMC_CONFIGFS ++ bool ++ depends on GPIOMMC && CONFIGFS_FS ++ default y ++ help ++ This option automatically enables configfs support for gpiommc ++ if configfs is available. ++ + config MMC_CB710 + tristate "ENE CB710 MMC/SD Interface support" + depends on PCI +--- a/drivers/mmc/host/Makefile ++++ b/drivers/mmc/host/Makefile +@@ -33,6 +33,7 @@ obj-$(CONFIG_MMC_SDRICOH_CS) += sdricoh_ + obj-$(CONFIG_MMC_TMIO) += tmio_mmc.o + obj-$(CONFIG_MMC_CB710) += cb710-mmc.o + obj-$(CONFIG_MMC_VIA_SDMMC) += via-sdmmc.o ++obj-$(CONFIG_GPIOMMC) += gpiommc.o + + ifeq ($(CONFIG_CB710_DEBUG),y) + CFLAGS-cb710-mmc += -DDEBUG +--- /dev/null ++++ b/include/linux/mmc/gpiommc.h +@@ -0,0 +1,71 @@ ++/* ++ * Device driver for MMC/SD cards driven over a GPIO bus. ++ * ++ * Copyright (c) 2008 Michael Buesch ++ * ++ * Licensed under the GNU/GPL version 2. ++ */ ++#ifndef LINUX_GPIOMMC_H_ ++#define LINUX_GPIOMMC_H_ ++ ++#include ++ ++ ++#define GPIOMMC_MAX_NAMELEN 15 ++#define GPIOMMC_MAX_NAMELEN_STR __stringify(GPIOMMC_MAX_NAMELEN) ++ ++/** ++ * struct gpiommc_pins - Hardware pin assignments ++ * ++ * @gpio_di: The GPIO number of the DATA IN pin ++ * @gpio_do: The GPIO number of the DATA OUT pin ++ * @gpio_clk: The GPIO number of the CLOCK pin ++ * @gpio_cs: The GPIO number of the CHIPSELECT pin ++ * @cs_activelow: If true, the chip is considered selected if @gpio_cs is low. ++ */ ++struct gpiommc_pins { ++ unsigned int gpio_di; ++ unsigned int gpio_do; ++ unsigned int gpio_clk; ++ unsigned int gpio_cs; ++ bool cs_activelow; ++}; ++ ++/** ++ * struct gpiommc_platform_data - Platform data for a MMC-over-SPI-GPIO device. ++ * ++ * @name: The unique name string of the device. ++ * @pins: The hardware pin assignments. ++ * @mode: The hardware mode. This is either SPI_MODE_0, ++ * SPI_MODE_1, SPI_MODE_2 or SPI_MODE_3. See the SPI documentation. ++ * @no_spi_delay: Do not use delays in the lowlevel SPI bitbanging code. ++ * This is not standards compliant, but may be required for some ++ * embedded machines to gain reasonable speed. ++ * @max_bus_speed: The maximum speed of the SPI bus, in Hertz. ++ */ ++struct gpiommc_platform_data { ++ char name[GPIOMMC_MAX_NAMELEN + 1]; ++ struct gpiommc_pins pins; ++ u8 mode; ++ bool no_spi_delay; ++ unsigned int max_bus_speed; ++}; ++ ++/** ++ * GPIOMMC_PLATDEV_NAME - The platform device name string. ++ * ++ * The name string that has to be used for platform_device_alloc ++ * when allocating a gpiommc device. ++ */ ++#define GPIOMMC_PLATDEV_NAME "gpiommc" ++ ++/** ++ * gpiommc_next_id - Get another platform device ID number. ++ * ++ * This returns the next platform device ID number that has to be used ++ * for platform_device_alloc. The ID is opaque and should not be used for ++ * anything else. ++ */ ++int gpiommc_next_id(void); ++ ++#endif /* LINUX_GPIOMMC_H_ */ +--- /dev/null ++++ b/Documentation/gpiommc.txt +@@ -0,0 +1,97 @@ ++GPIOMMC - Driver for an MMC/SD card on a bitbanging GPIO SPI bus ++================================================================ ++ ++The gpiommc module hooks up the mmc_spi and spi_gpio modules for running an ++MMC or SD card on GPIO pins. ++ ++Two interfaces for registering a new MMC/SD card device are provided: ++A static platform-device based mechanism and a dynamic configfs based interface. ++ ++ ++Registering devices via platform-device ++======================================= ++ ++The platform-device interface is used for registering MMC/SD devices that are ++part of the hardware platform. This is most useful only for embedded machines ++with MMC/SD devices statically connected to the platform GPIO bus. ++ ++The data structures are declared in . ++ ++To register a new device, define an instance of struct gpiommc_platform_data. ++This structure holds any information about how the device is hooked up to the ++GPIO pins and what hardware modes the device supports. See the docbook-style ++documentation in the header file for more information on the struct fields. ++ ++Then allocate a new instance of a platform device by doing: ++ ++ pdev = platform_device_alloc(GPIOMMC_PLATDEV_NAME, gpiommc_next_id()); ++ ++This will allocate the platform device data structures and hook it up to the ++gpiommc driver. ++Then add the gpiommc_platform_data to the platform device. ++ ++ err = platform_device_add_data(pdev, pdata, sizeof(struct gpiommc_platform_data)); ++ ++You may free the local instance of struct gpiommc_platform_data now. (So the ++struct may be allocated on the stack, too). ++Now simply register the platform device. ++ ++ err = platform_device_add(pdev); ++ ++Done. The gpiommc probe routine will be invoked now and you should see a kernel ++log message for the added device. ++ ++ ++Registering devices via configfs ++================================ ++ ++MMC/SD cards connected via GPIO often are a pretty dynamic thing, as for example ++selfmade hacks for soldering an MMC/SD card to standard GPIO pins on embedded ++hardware are a common situation. ++So we provide a dynamic interface to conveniently handle adding and removing ++devices from userspace, without the need to recompile the kernel. ++ ++The "gpiommc" subdirectory at the configfs mountpoint is used for handling ++the dynamic configuration. ++ ++To create a new device, it must first be allocated with mkdir. ++The following command will allocate a device named "my_mmc": ++ mkdir /config/gpiommc/my_mmc ++ ++There are several configuration files available in the new ++/config/gpiommc/my_mmc/ directory: ++ ++gpio_data_in = The SPI data-IN GPIO pin number. ++gpio_data_out = The SPI data-OUT GPIO pin number. ++gpio_clock = The SPI Clock GPIO pin number. ++gpio_chipselect = The SPI Chipselect GPIO pin number. ++gpio_chipselect_activelow = Boolean. If 0, Chipselect is active-HIGH. ++ If 1, Chipselect is active-LOW. ++spi_mode = The SPI data mode. Can be 0-3. ++spi_delay = Enable all delays in the lowlevel bitbanging. ++max_bus_speed = The maximum SPI bus speed. In Hertz. ++ ++register = Not a configuration parameter. ++ Used to register the configured card ++ with the kernel. ++ ++The device must first get configured and then registered by writing "1" to ++the "register" file. ++The configuration parameters "gpio_data_in", "gpio_data_out", "gpio_clock" ++and "gpio_chipselect" are essential and _must_ be configured before writing ++"1" to the "register" file. The registration will fail, otherwise. ++ ++The default values for the other parameters are: ++gpio_chipselect_activelow = 1 (CS active-LOW) ++spi_mode = 0 (SPI_MODE_0) ++spi_delay = 1 (enabled) ++max_bus_speed = 5000000 (5 Mhz) ++ ++Configuration values can not be changed after registration. To unregister ++the device, write a "0" to the "register" file. The configuration can be ++changed again after unregistering. ++ ++To completely remove the device, simply rmdir the directory ++(/config/gpiommc/my_mmc in this example). ++There's no need to first unregister the device before removing it. That will ++be done automatically. +--- a/MAINTAINERS ++++ b/MAINTAINERS +@@ -2253,6 +2253,11 @@ T: git git://git.kernel.org/pub/scm/linu + S: Maintained + F: drivers/media/video/gspca/ + ++GPIOMMC DRIVER ++P: Michael Buesch ++M: mb@bu3sch.de ++S: Maintained ++ + HARDWARE MONITORING + L: lm-sensors@lm-sensors.org + W: http://www.lm-sensors.org/ diff --git a/target/linux/generic-2.6/patches-2.6.31/923-gpiommc-configfs-locking.patch b/target/linux/generic-2.6/patches-2.6.31/923-gpiommc-configfs-locking.patch new file mode 100644 index 0000000000..2e4e820b2b --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/923-gpiommc-configfs-locking.patch @@ -0,0 +1,58 @@ +The gpiommc configfs context structure needs locking, as configfs +does not lock access between files. + +--- a/drivers/mmc/host/gpiommc.c ++++ b/drivers/mmc/host/gpiommc.c +@@ -143,6 +143,8 @@ struct gpiommc_configfs_device { + struct platform_device *pdev; + /* The configuration */ + struct gpiommc_platform_data pdata; ++ /* Mutex to protect this structure */ ++ struct mutex mutex; + }; + + #define GPIO_INVALID -1 +@@ -233,6 +235,8 @@ static ssize_t gpiommc_config_attr_show( + unsigned int gpio; + int err = 0; + ++ mutex_lock(&dev->mutex); ++ + if (attr == &gpiommc_attr_DI) { + gpio = dev->pdata.pins.gpio_di; + if (gpio == GPIO_INVALID) +@@ -293,6 +297,8 @@ static ssize_t gpiommc_config_attr_show( + WARN_ON(1); + err = -ENOSYS; + out: ++ mutex_unlock(&dev->mutex); ++ + return err ? err : count; + } + +@@ -352,6 +358,8 @@ static ssize_t gpiommc_config_attr_store + int err = -EINVAL; + unsigned long data; + ++ mutex_lock(&dev->mutex); ++ + if (attr == &gpiommc_attr_register) { + err = strict_strtoul(page, 10, &data); + if (err) +@@ -477,6 +485,8 @@ static ssize_t gpiommc_config_attr_store + WARN_ON(1); + err = -ENOSYS; + out: ++ mutex_unlock(&dev->mutex); ++ + return err ? err : count; + } + +@@ -513,6 +523,7 @@ static struct config_item *gpiommc_make_ + if (!dev) + return NULL; + ++ mutex_init(&dev->mutex); + config_item_init_type_name(&dev->item, name, + &gpiommc_dev_ci_type); + diff --git a/target/linux/generic-2.6/patches-2.6.31/924-cs5535_gpio.patch b/target/linux/generic-2.6/patches-2.6.31/924-cs5535_gpio.patch new file mode 100644 index 0000000000..453affe583 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/924-cs5535_gpio.patch @@ -0,0 +1,102 @@ +--- a/drivers/char/cs5535_gpio.c ++++ b/drivers/char/cs5535_gpio.c +@@ -15,6 +15,7 @@ + #include + #include + #include ++#include + #include + #include + #include +@@ -48,6 +49,7 @@ static struct pci_device_id divil_pci[] + MODULE_DEVICE_TABLE(pci, divil_pci); + + static struct cdev cs5535_gpio_cdev; ++static struct class *cs5535_gpio_class; + + /* reserve 32 entries even though some aren't usable */ + #define CS5535_GPIO_COUNT 32 +@@ -66,9 +68,14 @@ static struct gpio_regmap rm[] = + { 0x30, 0x00, '1', '0' }, /* GPIOx_READ_BACK / GPIOx_OUT_VAL */ + { 0x20, 0x20, 'I', 'i' }, /* GPIOx_IN_EN */ + { 0x04, 0x04, 'O', 'o' }, /* GPIOx_OUT_EN */ ++ { 0x10, 0x10, 'A', 'a' }, /* GPIOx_OUT_AUX1_SEL */ ++ { 0x14, 0x14, 'B', 'b' }, /* GPIOx_OUT_AUX2_SEL */ + { 0x08, 0x08, 't', 'T' }, /* GPIOx_OUT_OD_EN */ + { 0x18, 0x18, 'P', 'p' }, /* GPIOx_OUT_PU_EN */ + { 0x1c, 0x1c, 'D', 'd' }, /* GPIOx_OUT_PD_EN */ ++ { 0x24, 0x24, 'N', 'n' }, /* GPIOx_IN_INV_EN */ ++ { 0x0c, 0x0c, 'X', 'x' }, /* GPIOx_OUT_INV_EN */ ++ { 0x00, 0x00, 'H', 'L' }, /* GPIOx_OUT_VAL */ + }; + + +@@ -177,7 +184,7 @@ static int __init cs5535_gpio_init(void) + { + dev_t dev_id; + u32 low, hi; +- int retval; ++ int retval, i; + + if (pci_dev_present(divil_pci) == 0) { + printk(KERN_WARNING NAME ": DIVIL not found\n"); +@@ -232,23 +239,54 @@ static int __init cs5535_gpio_init(void) + major = MAJOR(dev_id); + } + +- if (retval) { +- release_region(gpio_base, CS5535_GPIO_SIZE); +- return -1; +- } ++ if (retval) ++ goto error; + + printk(KERN_DEBUG NAME ": base=%#x mask=%#lx major=%d\n", + gpio_base, mask, major); + + cdev_init(&cs5535_gpio_cdev, &cs5535_gpio_fops); +- cdev_add(&cs5535_gpio_cdev, dev_id, CS5535_GPIO_COUNT); ++ retval = cdev_add(&cs5535_gpio_cdev, dev_id, CS5535_GPIO_COUNT); ++ if (retval) { ++ kobject_put(&cs5535_gpio_cdev.kobj); ++ goto error_region; ++ } ++ ++ cs5535_gpio_class = class_create(THIS_MODULE, "cs5535_gpio"); ++ if (IS_ERR(cs5535_gpio_class)) { ++ printk(KERN_ERR "Error creating cs5535_gpio class\n"); ++ cdev_del(&cs5535_gpio_cdev); ++ retval = PTR_ERR(cs5535_gpio_class); ++ goto error_region; ++ } ++ ++ for (i = 0; i < CS5535_GPIO_COUNT; i++) { ++ if (mask & (1< +- +- .file "crtsavres.S" +- .section ".text" +- +-#ifdef CONFIG_CC_OPTIMIZE_FOR_SIZE +- +-/* Routines for saving integer registers, called by the compiler. */ +-/* Called with r11 pointing to the stack header word of the caller of the */ +-/* function, just beyond the end of the integer save area. */ +- +-_GLOBAL(_savegpr_14) +-_GLOBAL(_save32gpr_14) +- stw 14,-72(11) /* save gp registers */ +-_GLOBAL(_savegpr_15) +-_GLOBAL(_save32gpr_15) +- stw 15,-68(11) +-_GLOBAL(_savegpr_16) +-_GLOBAL(_save32gpr_16) +- stw 16,-64(11) +-_GLOBAL(_savegpr_17) +-_GLOBAL(_save32gpr_17) +- stw 17,-60(11) +-_GLOBAL(_savegpr_18) +-_GLOBAL(_save32gpr_18) +- stw 18,-56(11) +-_GLOBAL(_savegpr_19) +-_GLOBAL(_save32gpr_19) +- stw 19,-52(11) +-_GLOBAL(_savegpr_20) +-_GLOBAL(_save32gpr_20) +- stw 20,-48(11) +-_GLOBAL(_savegpr_21) +-_GLOBAL(_save32gpr_21) +- stw 21,-44(11) +-_GLOBAL(_savegpr_22) +-_GLOBAL(_save32gpr_22) +- stw 22,-40(11) +-_GLOBAL(_savegpr_23) +-_GLOBAL(_save32gpr_23) +- stw 23,-36(11) +-_GLOBAL(_savegpr_24) +-_GLOBAL(_save32gpr_24) +- stw 24,-32(11) +-_GLOBAL(_savegpr_25) +-_GLOBAL(_save32gpr_25) +- stw 25,-28(11) +-_GLOBAL(_savegpr_26) +-_GLOBAL(_save32gpr_26) +- stw 26,-24(11) +-_GLOBAL(_savegpr_27) +-_GLOBAL(_save32gpr_27) +- stw 27,-20(11) +-_GLOBAL(_savegpr_28) +-_GLOBAL(_save32gpr_28) +- stw 28,-16(11) +-_GLOBAL(_savegpr_29) +-_GLOBAL(_save32gpr_29) +- stw 29,-12(11) +-_GLOBAL(_savegpr_30) +-_GLOBAL(_save32gpr_30) +- stw 30,-8(11) +-_GLOBAL(_savegpr_31) +-_GLOBAL(_save32gpr_31) +- stw 31,-4(11) +- blr +- +-/* Routines for restoring integer registers, called by the compiler. */ +-/* Called with r11 pointing to the stack header word of the caller of the */ +-/* function, just beyond the end of the integer restore area. */ +- +-_GLOBAL(_restgpr_14) +-_GLOBAL(_rest32gpr_14) +- lwz 14,-72(11) /* restore gp registers */ +-_GLOBAL(_restgpr_15) +-_GLOBAL(_rest32gpr_15) +- lwz 15,-68(11) +-_GLOBAL(_restgpr_16) +-_GLOBAL(_rest32gpr_16) +- lwz 16,-64(11) +-_GLOBAL(_restgpr_17) +-_GLOBAL(_rest32gpr_17) +- lwz 17,-60(11) +-_GLOBAL(_restgpr_18) +-_GLOBAL(_rest32gpr_18) +- lwz 18,-56(11) +-_GLOBAL(_restgpr_19) +-_GLOBAL(_rest32gpr_19) +- lwz 19,-52(11) +-_GLOBAL(_restgpr_20) +-_GLOBAL(_rest32gpr_20) +- lwz 20,-48(11) +-_GLOBAL(_restgpr_21) +-_GLOBAL(_rest32gpr_21) +- lwz 21,-44(11) +-_GLOBAL(_restgpr_22) +-_GLOBAL(_rest32gpr_22) +- lwz 22,-40(11) +-_GLOBAL(_restgpr_23) +-_GLOBAL(_rest32gpr_23) +- lwz 23,-36(11) +-_GLOBAL(_restgpr_24) +-_GLOBAL(_rest32gpr_24) +- lwz 24,-32(11) +-_GLOBAL(_restgpr_25) +-_GLOBAL(_rest32gpr_25) +- lwz 25,-28(11) +-_GLOBAL(_restgpr_26) +-_GLOBAL(_rest32gpr_26) +- lwz 26,-24(11) +-_GLOBAL(_restgpr_27) +-_GLOBAL(_rest32gpr_27) +- lwz 27,-20(11) +-_GLOBAL(_restgpr_28) +-_GLOBAL(_rest32gpr_28) +- lwz 28,-16(11) +-_GLOBAL(_restgpr_29) +-_GLOBAL(_rest32gpr_29) +- lwz 29,-12(11) +-_GLOBAL(_restgpr_30) +-_GLOBAL(_rest32gpr_30) +- lwz 30,-8(11) +-_GLOBAL(_restgpr_31) +-_GLOBAL(_rest32gpr_31) +- lwz 31,-4(11) +- blr +- +-/* Routines for restoring integer registers, called by the compiler. */ +-/* Called with r11 pointing to the stack header word of the caller of the */ +-/* function, just beyond the end of the integer restore area. */ +- +-_GLOBAL(_restgpr_14_x) +-_GLOBAL(_rest32gpr_14_x) +- lwz 14,-72(11) /* restore gp registers */ +-_GLOBAL(_restgpr_15_x) +-_GLOBAL(_rest32gpr_15_x) +- lwz 15,-68(11) +-_GLOBAL(_restgpr_16_x) +-_GLOBAL(_rest32gpr_16_x) +- lwz 16,-64(11) +-_GLOBAL(_restgpr_17_x) +-_GLOBAL(_rest32gpr_17_x) +- lwz 17,-60(11) +-_GLOBAL(_restgpr_18_x) +-_GLOBAL(_rest32gpr_18_x) +- lwz 18,-56(11) +-_GLOBAL(_restgpr_19_x) +-_GLOBAL(_rest32gpr_19_x) +- lwz 19,-52(11) +-_GLOBAL(_restgpr_20_x) +-_GLOBAL(_rest32gpr_20_x) +- lwz 20,-48(11) +-_GLOBAL(_restgpr_21_x) +-_GLOBAL(_rest32gpr_21_x) +- lwz 21,-44(11) +-_GLOBAL(_restgpr_22_x) +-_GLOBAL(_rest32gpr_22_x) +- lwz 22,-40(11) +-_GLOBAL(_restgpr_23_x) +-_GLOBAL(_rest32gpr_23_x) +- lwz 23,-36(11) +-_GLOBAL(_restgpr_24_x) +-_GLOBAL(_rest32gpr_24_x) +- lwz 24,-32(11) +-_GLOBAL(_restgpr_25_x) +-_GLOBAL(_rest32gpr_25_x) +- lwz 25,-28(11) +-_GLOBAL(_restgpr_26_x) +-_GLOBAL(_rest32gpr_26_x) +- lwz 26,-24(11) +-_GLOBAL(_restgpr_27_x) +-_GLOBAL(_rest32gpr_27_x) +- lwz 27,-20(11) +-_GLOBAL(_restgpr_28_x) +-_GLOBAL(_rest32gpr_28_x) +- lwz 28,-16(11) +-_GLOBAL(_restgpr_29_x) +-_GLOBAL(_rest32gpr_29_x) +- lwz 29,-12(11) +-_GLOBAL(_restgpr_30_x) +-_GLOBAL(_rest32gpr_30_x) +- lwz 30,-8(11) +-_GLOBAL(_restgpr_31_x) +-_GLOBAL(_rest32gpr_31_x) +- lwz 0,4(11) +- lwz 31,-4(11) +- mtlr 0 +- mr 1,11 +- blr +-#endif +--- a/arch/powerpc/lib/Makefile ++++ b/arch/powerpc/lib/Makefile +@@ -13,7 +13,7 @@ CFLAGS_REMOVE_feature-fixups.o = -pg + + obj-y := string.o alloc.o \ + checksum_$(CONFIG_WORD_SIZE).o +-obj-$(CONFIG_PPC32) += div64.o copy_32.o crtsavres.o ++obj-$(CONFIG_PPC32) += div64.o copy_32.o + obj-$(CONFIG_HAS_IOMEM) += devres.o + + obj-$(CONFIG_PPC64) += copypage_64.o copyuser_64.o \ +--- a/arch/powerpc/Makefile ++++ b/arch/powerpc/Makefile +@@ -92,8 +92,6 @@ endif + else + KBUILD_CFLAGS += $(call cc-option,-mtune=power4) + endif +-else +-LDFLAGS_MODULE += arch/powerpc/lib/crtsavres.o + endif + + ifeq ($(CONFIG_TUNE_CELL),y) diff --git a/target/linux/generic-2.6/patches-2.6.31/960-arm_lzma_loader.patch b/target/linux/generic-2.6/patches-2.6.31/960-arm_lzma_loader.patch new file mode 100644 index 0000000000..5f4f28757b --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/960-arm_lzma_loader.patch @@ -0,0 +1,710 @@ +--- a/arch/arm/boot/compressed/Makefile ++++ b/arch/arm/boot/compressed/Makefile +@@ -63,7 +63,7 @@ endif + + SEDFLAGS = s/TEXT_START/$(ZTEXTADDR)/;s/BSS_START/$(ZBSSADDR)/ + +-targets := vmlinux vmlinux.lds piggy.gz piggy.o font.o font.c \ ++targets := vmlinux vmlinux.lds piggy.lzma piggy.o font.o font.c \ + head.o misc.o $(OBJS) + + ifeq ($(CONFIG_FUNCTION_TRACER),y) +@@ -99,10 +99,10 @@ $(obj)/vmlinux: $(obj)/vmlinux.lds $(obj + $(call if_changed,ld) + @: + +-$(obj)/piggy.gz: $(obj)/../Image FORCE +- $(call if_changed,gzip) ++$(obj)/piggy.lzma: $(obj)/../Image FORCE ++ $(call if_changed,lzma) + +-$(obj)/piggy.o: $(obj)/piggy.gz FORCE ++$(obj)/piggy.o: $(obj)/piggy.lzma FORCE + + CFLAGS_font.o := -Dstatic= + +--- a/arch/arm/boot/compressed/misc.c ++++ b/arch/arm/boot/compressed/misc.c +@@ -185,36 +185,10 @@ static inline __ptr_t memcpy(__ptr_t __d + return __dest; + } + +-/* +- * gzip delarations +- */ +-#define OF(args) args +-#define STATIC static +- +-typedef unsigned char uch; +-typedef unsigned short ush; +-typedef unsigned long ulg; +- +-#define WSIZE 0x8000 /* Window size must be at least 32k, */ ++#define WSIZE 0x20000 /* Window size must be at least 128k, */ + /* and a power of two */ + +-static uch *inbuf; /* input buffer */ +-static uch window[WSIZE]; /* Sliding window buffer */ +- +-static unsigned insize; /* valid bytes in inbuf */ +-static unsigned inptr; /* index of next byte to be processed in inbuf */ +-static unsigned outcnt; /* bytes in output buffer */ +- +-/* gzip flag byte */ +-#define ASCII_FLAG 0x01 /* bit 0 set: file probably ascii text */ +-#define CONTINUATION 0x02 /* bit 1 set: continuation of multi-part gzip file */ +-#define EXTRA_FIELD 0x04 /* bit 2 set: extra field present */ +-#define ORIG_NAME 0x08 /* bit 3 set: original file name present */ +-#define COMMENT 0x10 /* bit 4 set: file comment present */ +-#define ENCRYPTED 0x20 /* bit 5 set: file is encrypted */ +-#define RESERVED 0xC0 /* bit 6,7: reserved */ +- +-#define get_byte() (inptr < insize ? inbuf[inptr++] : fill_inbuf()) ++static u8 window[WSIZE]; /* Sliding window buffer */ + + /* Diagnostic functions */ + #ifdef DEBUG +@@ -233,24 +207,21 @@ static unsigned outcnt; /* bytes in out + # define Tracecv(c,x) + #endif + +-static int fill_inbuf(void); +-static void flush_window(void); + static void error(char *m); + + extern char input_data[]; + extern char input_data_end[]; + +-static uch *output_data; +-static ulg output_ptr; +-static ulg bytes_out; ++static unsigned long output_ptr; ++static unsigned long bytes_out; + + static void error(char *m); + + static void putstr(const char *); + + extern int end; +-static ulg free_mem_ptr; +-static ulg free_mem_end_ptr; ++static unsigned long free_mem_ptr; ++static unsigned long free_mem_end_ptr; + + #ifdef STANDALONE_DEBUG + #define NO_INFLATE_MALLOC +@@ -258,50 +229,10 @@ static ulg free_mem_end_ptr; + + #define ARCH_HAS_DECOMP_WDOG + +-#include "../../../../lib/inflate.c" +- +-/* =========================================================================== +- * Fill the input buffer. This is called only when the buffer is empty +- * and at least one byte is really needed. +- */ +-int fill_inbuf(void) +-{ +- if (insize != 0) +- error("ran out of input data"); +- +- inbuf = input_data; +- insize = &input_data_end[0] - &input_data[0]; +- +- inptr = 1; +- return inbuf[0]; +-} +- +-/* =========================================================================== +- * Write the output window window[0..outcnt-1] and update crc and bytes_out. +- * (Used for the decompressed data only.) +- */ +-void flush_window(void) +-{ +- ulg c = crc; +- unsigned n; +- uch *in, *out, ch; +- +- in = window; +- out = &output_data[output_ptr]; +- for (n = 0; n < outcnt; n++) { +- ch = *out++ = *in++; +- c = crc_32_tab[((int)c ^ ch) & 0xff] ^ (c >> 8); +- } +- crc = c; +- bytes_out += (ulg)outcnt; +- output_ptr += (ulg)outcnt; +- outcnt = 0; +- putstr("."); +-} +- + #ifndef arch_error + #define arch_error(x) + #endif ++#include "unlzma.c" + + static void error(char *x) + { +@@ -316,20 +247,16 @@ static void error(char *x) + + #ifndef STANDALONE_DEBUG + +-ulg +-decompress_kernel(ulg output_start, ulg free_mem_ptr_p, ulg free_mem_ptr_end_p, ++unsigned long ++decompress_kernel(unsigned long output_start, unsigned long free_mem_ptr_p, unsigned long free_mem_ptr_end_p, + int arch_id) + { +- output_data = (uch *)output_start; /* Points to kernel start */ +- free_mem_ptr = free_mem_ptr_p; +- free_mem_end_ptr = free_mem_ptr_end_p; + __machine_arch_type = arch_id; + + arch_decomp_setup(); + +- makecrc(); + putstr("Uncompressing Linux..."); +- gunzip(); ++ output_ptr += unlzma((u8 *) output_start, input_data, window); + putstr(" done, booting the kernel.\n"); + return output_ptr; + } +@@ -339,11 +266,8 @@ char output_buffer[1500*1024]; + + int main() + { +- output_data = output_buffer; +- +- makecrc(); + putstr("Uncompressing Linux..."); +- gunzip(); ++ unlzma((u8 *) output_buffer, input_data, window); + putstr("done.\n"); + return 0; + } +--- a/arch/arm/boot/compressed/piggy.S ++++ b/arch/arm/boot/compressed/piggy.S +@@ -1,6 +1,6 @@ + .section .piggydata,#alloc + .globl input_data + input_data: +- .incbin "arch/arm/boot/compressed/piggy.gz" ++ .incbin "arch/arm/boot/compressed/piggy.lzma" + .globl input_data_end + input_data_end: +--- /dev/null ++++ b/arch/arm/boot/compressed/unlzma.c +@@ -0,0 +1,429 @@ ++/* ++ * Copyright (c) 2009 Felix Fietkau ++ * ++ * This program is free software; you can redistribute it and/or ++ * modify it under the terms of the GNU General Public License ++ * as published by the Free Software Foundation; either version 2, ++ * or (at your option) any later version. ++ * ++ * This program is distributed in the hope that it will be useful, ++ * but WITHOUT ANY WARRANTY; without even the implied warranty of ++ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the ++ * GNU General Public License for more details. ++ * ++ * You should have received a copy of the GNU General Public License ++ * along with this program; if not, write to the Free Software ++ * Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. ++ * ++ * uncompress.c ++ */ ++ ++#include ++#include ++#include "unlzma.h" ++ ++struct unlzma_ctx { ++ const u8 *next_in; ++ u8 *next_out; ++ u8 *outbuf; ++ ++ /* reader state */ ++ u32 code; ++ u32 range; ++ u32 bound; ++ ++ /* writer state */ ++ u8 previous_byte; ++ ssize_t pos; ++ ++ /* cstate */ ++ int state; ++ u32 rep0, rep1, rep2, rep3; ++ ++ void *workspace; ++} ctx; ++ ++static int inbs = 0; ++static inline u8 ++rc_read(void) ++{ ++#if 0 ++ if (unlikely(++inbs > 16 * 1024)) { ++ putstr("."); ++ inbs = 0; ++ } ++#endif ++ return *(ctx.next_in++); ++} ++ ++ ++static inline void ++rc_get_code(void) ++{ ++ ctx.code = (ctx.code << 8) | rc_read(); ++} ++ ++static inline void ++rc_normalize(void) ++{ ++ if (ctx.range < (1 << RC_TOP_BITS)) { ++ ctx.range <<= 8; ++ rc_get_code(); ++ } ++} ++ ++static inline int ++rc_is_bit_0(u16 *p) ++{ ++ rc_normalize(); ++ ctx.bound = *p * (ctx.range >> RC_MODEL_TOTAL_BITS); ++ return ctx.code < ctx.bound; ++} ++ ++static inline void ++rc_update_bit_0(u16 *p) ++{ ++ ctx.range = ctx.bound; ++ *p += ((1 << RC_MODEL_TOTAL_BITS) - *p) >> RC_MOVE_BITS; ++} ++ ++static inline void ++rc_update_bit_1(u16 *p) ++{ ++ ctx.range -= ctx.bound; ++ ctx.code -= ctx.bound; ++ *p -= *p >> RC_MOVE_BITS; ++} ++ ++static inline bool ++rc_get_bit(u16 *p, int *symbol) ++{ ++ if (rc_is_bit_0(p)) { ++ rc_update_bit_0(p); ++ *symbol *= 2; ++ return 0; ++ } else { ++ rc_update_bit_1(p); ++ *symbol = *symbol * 2 + 1; ++ return 1; ++ } ++} ++ ++static inline int ++rc_direct_bit(void) ++{ ++ rc_normalize(); ++ ctx.range >>= 1; ++ if (ctx.code >= ctx.range) { ++ ctx.code -= ctx.range; ++ return 1; ++ } ++ return 0; ++} ++ ++static inline void ++rc_bit_tree_decode(u16 *p, int num_levels, int *symbol) ++{ ++ int i = num_levels; ++ ++ *symbol = 1; ++ while (i--) ++ rc_get_bit(p + *symbol, symbol); ++ *symbol -= 1 << num_levels; ++} ++ ++static inline u8 ++peek_old_byte(u32 offs) ++{ ++ u32 pos = ctx.pos - offs; ++ return ctx.outbuf[pos]; ++} ++ ++static inline void ++write_byte(u8 byte) ++{ ++ ctx.previous_byte = byte; ++ *(ctx.next_out++) = byte; ++ ctx.pos++; ++} ++ ++ ++static inline void ++copy_byte(u32 offs) ++{ ++ write_byte(peek_old_byte(offs)); ++} ++ ++static inline void ++copy_bytes(u32 rep0, int len) ++{ ++ do { ++ copy_byte(rep0); ++ len--; ++ } while (len != 0); ++} ++ ++static inline void ++process_bit0(u16 *p, int pos_state, u16 *prob, ++ int lc, u32 literal_pos_mask) ++{ ++ int mi = 1; ++ rc_update_bit_0(prob); ++ prob = (p + LZMA_LITERAL + ++ (LZMA_LIT_SIZE ++ * (((ctx.pos & literal_pos_mask) << lc) ++ + (ctx.previous_byte >> (8 - lc)))) ++ ); ++ ++ if (ctx.state >= LZMA_NUM_LIT_STATES) { ++ int match_byte = peek_old_byte(ctx.rep0); ++ do { ++ u16 bit; ++ u16 *prob_lit; ++ ++ match_byte <<= 1; ++ bit = match_byte & 0x100; ++ prob_lit = prob + 0x100 + bit + mi; ++ if (rc_get_bit(prob_lit, &mi) != !!bit) ++ break; ++ } while (mi < 0x100); ++ } ++ while (mi < 0x100) { ++ u16 *prob_lit = prob + mi; ++ rc_get_bit(prob_lit, &mi); ++ } ++ write_byte(mi); ++ if (ctx.state < 4) ++ ctx.state = 0; ++ else if (ctx.state < 10) ++ ctx.state -= 3; ++ else ++ ctx.state -= 6; ++} ++ ++static inline void ++process_bit1(u16 *p, int pos_state, u16 *prob) ++{ ++ int offset; ++ u16 *prob_len; ++ int num_bits; ++ int len; ++ ++ rc_update_bit_1(prob); ++ prob = p + LZMA_IS_REP + ctx.state; ++ if (rc_is_bit_0(prob)) { ++ rc_update_bit_0(prob); ++ ctx.rep3 = ctx.rep2; ++ ctx.rep2 = ctx.rep1; ++ ctx.rep1 = ctx.rep0; ++ ctx.state = ctx.state < LZMA_NUM_LIT_STATES ? 0 : 3; ++ prob = p + LZMA_LEN_CODER; ++ } else { ++ rc_update_bit_1(prob); ++ prob = p + LZMA_IS_REP_G0 + ctx.state; ++ if (rc_is_bit_0(prob)) { ++ rc_update_bit_0(prob); ++ prob = (p + LZMA_IS_REP_0_LONG ++ + (ctx.state << ++ LZMA_NUM_POS_BITS_MAX) + ++ pos_state); ++ if (rc_is_bit_0(prob)) { ++ rc_update_bit_0(prob); ++ ++ ctx.state = ctx.state < LZMA_NUM_LIT_STATES ? ++ 9 : 11; ++ copy_byte(ctx.rep0); ++ return; ++ } else { ++ rc_update_bit_1(prob); ++ } ++ } else { ++ u32 distance; ++ ++ rc_update_bit_1(prob); ++ prob = p + LZMA_IS_REP_G1 + ctx.state; ++ if (rc_is_bit_0(prob)) { ++ rc_update_bit_0(prob); ++ distance = ctx.rep1; ++ } else { ++ rc_update_bit_1(prob); ++ prob = p + LZMA_IS_REP_G2 + ctx.state; ++ if (rc_is_bit_0(prob)) { ++ rc_update_bit_0(prob); ++ distance = ctx.rep2; ++ } else { ++ rc_update_bit_1(prob); ++ distance = ctx.rep3; ++ ctx.rep3 = ctx.rep2; ++ } ++ ctx.rep2 = ctx.rep1; ++ } ++ ctx.rep1 = ctx.rep0; ++ ctx.rep0 = distance; ++ } ++ ctx.state = ctx.state < LZMA_NUM_LIT_STATES ? 8 : 11; ++ prob = p + LZMA_REP_LEN_CODER; ++ } ++ ++ prob_len = prob + LZMA_LEN_CHOICE; ++ if (rc_is_bit_0(prob_len)) { ++ rc_update_bit_0(prob_len); ++ prob_len = (prob + LZMA_LEN_LOW ++ + (pos_state << ++ LZMA_LEN_NUM_LOW_BITS)); ++ offset = 0; ++ num_bits = LZMA_LEN_NUM_LOW_BITS; ++ } else { ++ rc_update_bit_1(prob_len); ++ prob_len = prob + LZMA_LEN_CHOICE_2; ++ if (rc_is_bit_0(prob_len)) { ++ rc_update_bit_0(prob_len); ++ prob_len = (prob + LZMA_LEN_MID ++ + (pos_state << ++ LZMA_LEN_NUM_MID_BITS)); ++ offset = 1 << LZMA_LEN_NUM_LOW_BITS; ++ num_bits = LZMA_LEN_NUM_MID_BITS; ++ } else { ++ rc_update_bit_1(prob_len); ++ prob_len = prob + LZMA_LEN_HIGH; ++ offset = ((1 << LZMA_LEN_NUM_LOW_BITS) ++ + (1 << LZMA_LEN_NUM_MID_BITS)); ++ num_bits = LZMA_LEN_NUM_HIGH_BITS; ++ } ++ } ++ ++ rc_bit_tree_decode(prob_len, num_bits, &len); ++ len += offset; ++ ++ if (ctx.state < 4) { ++ int pos_slot; ++ ++ ctx.state += LZMA_NUM_LIT_STATES; ++ prob = ++ p + LZMA_POS_SLOT + ++ ((len < ++ LZMA_NUM_LEN_TO_POS_STATES ? len : ++ LZMA_NUM_LEN_TO_POS_STATES - 1) ++ << LZMA_NUM_POS_SLOT_BITS); ++ rc_bit_tree_decode(prob, ++ LZMA_NUM_POS_SLOT_BITS, ++ &pos_slot); ++ if (pos_slot >= LZMA_START_POS_MODEL_INDEX) { ++ int i, mi; ++ num_bits = (pos_slot >> 1) - 1; ++ ctx.rep0 = 2 | (pos_slot & 1); ++ if (pos_slot < LZMA_END_POS_MODEL_INDEX) { ++ ctx.rep0 <<= num_bits; ++ prob = p + LZMA_SPEC_POS + ++ ctx.rep0 - pos_slot - 1; ++ } else { ++ num_bits -= LZMA_NUM_ALIGN_BITS; ++ while (num_bits--) ++ ctx.rep0 = (ctx.rep0 << 1) | ++ rc_direct_bit(); ++ prob = p + LZMA_ALIGN; ++ ctx.rep0 <<= LZMA_NUM_ALIGN_BITS; ++ num_bits = LZMA_NUM_ALIGN_BITS; ++ } ++ i = 1; ++ mi = 1; ++ while (num_bits--) { ++ if (rc_get_bit(prob + mi, &mi)) ++ ctx.rep0 |= i; ++ i <<= 1; ++ } ++ } else ++ ctx.rep0 = pos_slot; ++ if (++(ctx.rep0) == 0) ++ return; ++ } ++ ++ len += LZMA_MATCH_MIN_LEN; ++ ++ copy_bytes(ctx.rep0, len); ++} ++ ++ ++static int ++do_unlzma(void) ++{ ++ u8 hdr_buf[sizeof(struct lzma_header)]; ++ struct lzma_header *header = (struct lzma_header *)hdr_buf; ++ u32 pos_state_mask; ++ u32 literal_pos_mask; ++ int lc, pb, lp; ++ int num_probs; ++ int i, mi; ++ u16 *p; ++ ++ for (i = 0; i < sizeof(struct lzma_header); i++) { ++ hdr_buf[i] = rc_read(); ++ } ++ ++ ctx.pos = 0; ++ ctx.state = 0; ++ ctx.rep0 = ctx.rep1 = ctx.rep2 = ctx.rep3 = 1; ++ ++ ctx.previous_byte = 0; ++ ctx.code = 0; ++ ctx.range = 0xFFFFFFFF; ++ ++ if (header->pos >= (9 * 5 * 5)) ++ return -1; ++ ++ mi = 0; ++ lc = header->pos; ++ while (lc >= 9) { ++ mi++; ++ lc -= 9; ++ } ++ pb = 0; ++ lp = mi; ++ while (lp >= 5) { ++ pb++; ++ lp -= 5; ++ } ++ pos_state_mask = (1 << pb) - 1; ++ literal_pos_mask = (1 << lp) - 1; ++ ++ p = (u16 *) ctx.workspace; ++ if (!p) ++ return -1; ++ ++ num_probs = LZMA_LITERAL + (LZMA_LIT_SIZE << (lc + lp)); ++ for (i = 0; i < num_probs; i++) ++ p[i] = (1 << RC_MODEL_TOTAL_BITS) >> 1; ++ ++ for (i = 0; i < 5; i++) ++ rc_get_code(); ++ ++ while (1) { ++ int pos_state = ctx.pos & pos_state_mask; ++ u16 *prob = p + LZMA_IS_MATCH + ++ (ctx.state << LZMA_NUM_POS_BITS_MAX) + pos_state; ++ if (rc_is_bit_0(prob)) ++ process_bit0(p, pos_state, prob, ++ lc, literal_pos_mask); ++ else { ++ process_bit1(p, pos_state, prob); ++ if (ctx.rep0 == 0) ++ break; ++ } ++ } ++ ++ return ctx.pos; ++} ++ ++ ++static int unlzma(unsigned char *dest, const unsigned char *src, unsigned char *workspace) ++{ ++ memset(&ctx, 0, sizeof(ctx)); ++ ctx.outbuf = dest; ++ ctx.next_in = src; ++ ctx.next_out = dest; ++ ctx.workspace = workspace; ++ ++ return do_unlzma(); ++} ++ ++ +--- /dev/null ++++ b/arch/arm/boot/compressed/unlzma.h +@@ -0,0 +1,81 @@ ++/* LZMA uncompresion module for pcomp ++ * Copyright (C) 2009 Felix Fietkau ++ * ++ * Based on: ++ * Initial Linux kernel adaptation ++ * Copyright (C) 2006 Alain < alain@knaff.lu > ++ * ++ * Based on small lzma deflate implementation/Small range coder ++ * implementation for lzma. ++ * Copyright (C) 2006 Aurelien Jacobs < aurel@gnuage.org > ++ * ++ * Based on LzmaDecode.c from the LZMA SDK 4.22 (http://www.7-zip.org/) ++ * Copyright (C) 1999-2005 Igor Pavlov ++ * ++ * This program is free software; you can redistribute it and/or modify it ++ * under the terms of the GNU General Public License version 2 as published ++ * by the Free Software Foundation. ++ */ ++#ifndef __UNLZMA_H ++#define __UNLZMA_H ++ ++struct lzma_header { ++ __u8 pos; ++ __le32 dict_size; ++ __le64 uncompr_size; ++} __attribute__ ((packed)); ++ ++ ++#define RC_TOP_BITS 24 ++#define RC_MOVE_BITS 5 ++#define RC_MODEL_TOTAL_BITS 11 ++ ++#define LZMA_BASE_SIZE 1846 ++#define LZMA_LIT_SIZE 768 ++ ++#define LZMA_NUM_POS_BITS_MAX 4 ++ ++#define LZMA_LEN_NUM_LOW_BITS 3 ++#define LZMA_LEN_NUM_MID_BITS 3 ++#define LZMA_LEN_NUM_HIGH_BITS 8 ++ ++#define LZMA_LEN_CHOICE 0 ++#define LZMA_LEN_CHOICE_2 (LZMA_LEN_CHOICE + 1) ++#define LZMA_LEN_LOW (LZMA_LEN_CHOICE_2 + 1) ++#define LZMA_LEN_MID (LZMA_LEN_LOW \ ++ + (1 << (LZMA_NUM_POS_BITS_MAX + LZMA_LEN_NUM_LOW_BITS))) ++#define LZMA_LEN_HIGH (LZMA_LEN_MID \ ++ +(1 << (LZMA_NUM_POS_BITS_MAX + LZMA_LEN_NUM_MID_BITS))) ++#define LZMA_NUM_LEN_PROBS (LZMA_LEN_HIGH + (1 << LZMA_LEN_NUM_HIGH_BITS)) ++ ++#define LZMA_NUM_STATES 12 ++#define LZMA_NUM_LIT_STATES 7 ++ ++#define LZMA_START_POS_MODEL_INDEX 4 ++#define LZMA_END_POS_MODEL_INDEX 14 ++#define LZMA_NUM_FULL_DISTANCES (1 << (LZMA_END_POS_MODEL_INDEX >> 1)) ++ ++#define LZMA_NUM_POS_SLOT_BITS 6 ++#define LZMA_NUM_LEN_TO_POS_STATES 4 ++ ++#define LZMA_NUM_ALIGN_BITS 4 ++ ++#define LZMA_MATCH_MIN_LEN 2 ++ ++#define LZMA_IS_MATCH 0 ++#define LZMA_IS_REP (LZMA_IS_MATCH + (LZMA_NUM_STATES << LZMA_NUM_POS_BITS_MAX)) ++#define LZMA_IS_REP_G0 (LZMA_IS_REP + LZMA_NUM_STATES) ++#define LZMA_IS_REP_G1 (LZMA_IS_REP_G0 + LZMA_NUM_STATES) ++#define LZMA_IS_REP_G2 (LZMA_IS_REP_G1 + LZMA_NUM_STATES) ++#define LZMA_IS_REP_0_LONG (LZMA_IS_REP_G2 + LZMA_NUM_STATES) ++#define LZMA_POS_SLOT (LZMA_IS_REP_0_LONG \ ++ + (LZMA_NUM_STATES << LZMA_NUM_POS_BITS_MAX)) ++#define LZMA_SPEC_POS (LZMA_POS_SLOT \ ++ +(LZMA_NUM_LEN_TO_POS_STATES << LZMA_NUM_POS_SLOT_BITS)) ++#define LZMA_ALIGN (LZMA_SPEC_POS \ ++ + LZMA_NUM_FULL_DISTANCES - LZMA_END_POS_MODEL_INDEX) ++#define LZMA_LEN_CODER (LZMA_ALIGN + (1 << LZMA_NUM_ALIGN_BITS)) ++#define LZMA_REP_LEN_CODER (LZMA_LEN_CODER + LZMA_NUM_LEN_PROBS) ++#define LZMA_LITERAL (LZMA_REP_LEN_CODER + LZMA_NUM_LEN_PROBS) ++ ++#endif diff --git a/target/linux/generic-2.6/patches-2.6.31/970-ocf_kbuild_integration.patch b/target/linux/generic-2.6/patches-2.6.31/970-ocf_kbuild_integration.patch new file mode 100644 index 0000000000..54740df083 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/970-ocf_kbuild_integration.patch @@ -0,0 +1,25 @@ +--- a/crypto/Kconfig ++++ b/crypto/Kconfig +@@ -794,6 +794,8 @@ config CRYPTO_ANSI_CPRNG + for cryptographic modules. Uses the Algorithm specified in + ANSI X9.31 A.2.4 + ++source "crypto/ocf/Kconfig" ++ + source "drivers/crypto/Kconfig" + + endif # if CRYPTO +--- a/crypto/Makefile ++++ b/crypto/Makefile +@@ -86,6 +86,11 @@ obj-$(CONFIG_CRYPTO_ANSI_CPRNG) += ansi_ + obj-$(CONFIG_CRYPTO_TEST) += tcrypt.o + + # ++# OCF ++# ++obj-$(CONFIG_OCF_OCF) += ocf/ ++ ++# + # generic algorithms and the async_tx api + # + obj-$(CONFIG_XOR_BLOCKS) += xor.o diff --git a/target/linux/generic-2.6/patches-2.6.31/971-ocf_20080917.patch b/target/linux/generic-2.6/patches-2.6.31/971-ocf_20080917.patch new file mode 100644 index 0000000000..29b15d5f80 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/971-ocf_20080917.patch @@ -0,0 +1,152 @@ +--- a/drivers/char/random.c ++++ b/drivers/char/random.c +@@ -129,6 +129,9 @@ + * unsigned int value); + * void add_interrupt_randomness(int irq); + * ++ * void random_input_words(__u32 *buf, size_t wordcount, int ent_count) ++ * int random_input_wait(void); ++ * + * add_input_randomness() uses the input layer interrupt timing, as well as + * the event type information from the hardware. + * +@@ -140,6 +143,13 @@ + * a better measure, since the timing of the disk interrupts are more + * unpredictable. + * ++ * random_input_words() just provides a raw block of entropy to the input ++ * pool, such as from a hardware entropy generator. ++ * ++ * random_input_wait() suspends the caller until such time as the ++ * entropy pool falls below the write threshold, and returns a count of how ++ * much entropy (in bits) is needed to sustain the pool. ++ * + * All of these routines try to estimate how many bits of randomness a + * particular randomness source. They do this by keeping track of the + * first and second order deltas of the event timings. +@@ -712,6 +722,61 @@ void add_disk_randomness(struct gendisk + } + #endif + ++/* ++ * random_input_words - add bulk entropy to pool ++ * ++ * @buf: buffer to add ++ * @wordcount: number of __u32 words to add ++ * @ent_count: total amount of entropy (in bits) to credit ++ * ++ * this provides bulk input of entropy to the input pool ++ * ++ */ ++void random_input_words(__u32 *buf, size_t wordcount, int ent_count) ++{ ++ mix_pool_bytes(&input_pool, buf, wordcount*4); ++ ++ credit_entropy_bits(&input_pool, ent_count); ++ ++ DEBUG_ENT("crediting %d bits => %d\n", ++ ent_count, input_pool.entropy_count); ++ /* ++ * Wake up waiting processes if we have enough ++ * entropy. ++ */ ++ if (input_pool.entropy_count >= random_read_wakeup_thresh) ++ wake_up_interruptible(&random_read_wait); ++} ++EXPORT_SYMBOL(random_input_words); ++ ++/* ++ * random_input_wait - wait until random needs entropy ++ * ++ * this function sleeps until the /dev/random subsystem actually ++ * needs more entropy, and then return the amount of entropy ++ * that it would be nice to have added to the system. ++ */ ++int random_input_wait(void) ++{ ++ int count; ++ ++ wait_event_interruptible(random_write_wait, ++ input_pool.entropy_count < random_write_wakeup_thresh); ++ ++ count = random_write_wakeup_thresh - input_pool.entropy_count; ++ ++ /* likely we got woken up due to a signal */ ++ if (count <= 0) count = random_read_wakeup_thresh; ++ ++ DEBUG_ENT("requesting %d bits from input_wait()er %d<%d\n", ++ count, ++ input_pool.entropy_count, random_write_wakeup_thresh); ++ ++ return count; ++} ++EXPORT_SYMBOL(random_input_wait); ++ ++ + #define EXTRACT_SIZE 10 + + /********************************************************************* +--- a/fs/fcntl.c ++++ b/fs/fcntl.c +@@ -141,6 +141,7 @@ SYSCALL_DEFINE1(dup, unsigned int, filde + } + return ret; + } ++EXPORT_SYMBOL(sys_dup); + + #define SETFL_MASK (O_APPEND | O_NONBLOCK | O_NDELAY | O_DIRECT | O_NOATIME) + +--- a/include/linux/miscdevice.h ++++ b/include/linux/miscdevice.h +@@ -12,6 +12,7 @@ + #define APOLLO_MOUSE_MINOR 7 + #define PC110PAD_MINOR 9 + /*#define ADB_MOUSE_MINOR 10 FIXME OBSOLETE */ ++#define CRYPTODEV_MINOR 70 /* /dev/crypto */ + #define WATCHDOG_MINOR 130 /* Watchdog timer */ + #define TEMP_MINOR 131 /* Temperature Sensor */ + #define RTC_MINOR 135 +--- a/include/linux/random.h ++++ b/include/linux/random.h +@@ -34,6 +34,30 @@ + /* Clear the entropy pool and associated counters. (Superuser only.) */ + #define RNDCLEARPOOL _IO( 'R', 0x06 ) + ++#ifdef CONFIG_FIPS_RNG ++ ++/* Size of seed value - equal to AES blocksize */ ++#define AES_BLOCK_SIZE_BYTES 16 ++#define SEED_SIZE_BYTES AES_BLOCK_SIZE_BYTES ++/* Size of AES key */ ++#define KEY_SIZE_BYTES 16 ++ ++/* ioctl() structure used by FIPS 140-2 Tests */ ++struct rand_fips_test { ++ unsigned char key[KEY_SIZE_BYTES]; /* Input */ ++ unsigned char datetime[SEED_SIZE_BYTES]; /* Input */ ++ unsigned char seed[SEED_SIZE_BYTES]; /* Input */ ++ unsigned char result[SEED_SIZE_BYTES]; /* Output */ ++}; ++ ++/* FIPS 140-2 RNG Variable Seed Test. (Superuser only.) */ ++#define RNDFIPSVST _IOWR('R', 0x10, struct rand_fips_test) ++ ++/* FIPS 140-2 RNG Monte Carlo Test. (Superuser only.) */ ++#define RNDFIPSMCT _IOWR('R', 0x11, struct rand_fips_test) ++ ++#endif /* #ifdef CONFIG_FIPS_RNG */ ++ + struct rand_pool_info { + int entropy_count; + int buf_size; +@@ -50,6 +74,10 @@ extern void add_input_randomness(unsigne + unsigned int value); + extern void add_interrupt_randomness(int irq); + ++extern void random_input_words(__u32 *buf, size_t wordcount, int ent_count); ++extern int random_input_wait(void); ++#define HAS_RANDOM_INPUT_WAIT 1 ++ + extern void get_random_bytes(void *buf, int nbytes); + void generate_random_uuid(unsigned char uuid_out[16]); + diff --git a/target/linux/generic-2.6/patches-2.6.31/972-ocf_compile_fix.patch b/target/linux/generic-2.6/patches-2.6.31/972-ocf_compile_fix.patch new file mode 100644 index 0000000000..a3fa226814 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/972-ocf_compile_fix.patch @@ -0,0 +1,11 @@ +--- a/crypto/ocf/cryptosoft.c ++++ b/crypto/ocf/cryptosoft.c +@@ -47,7 +47,7 @@ + #include + #include + #include +-#include ++#include + + #include + #include diff --git a/target/linux/generic-2.6/patches-2.6.31/973-ocf_2.6.27_fix.patch b/target/linux/generic-2.6/patches-2.6.31/973-ocf_2.6.27_fix.patch new file mode 100644 index 0000000000..ecb9bef513 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/973-ocf_2.6.27_fix.patch @@ -0,0 +1,197 @@ +--- a/crypto/ocf/random.c ++++ b/crypto/ocf/random.c +@@ -49,6 +49,7 @@ + #include + #include + #include ++#include + #include + + #ifdef CONFIG_OCF_FIPS +@@ -81,7 +82,7 @@ struct random_op { + + static int random_proc(void *arg); + +-static pid_t randomproc = (pid_t) -1; ++static struct task_struct *random_task; + static spinlock_t random_lock; + + /* +@@ -141,13 +142,18 @@ crypto_rregister( + spin_lock_irqsave(&random_lock, flags); + list_add_tail(&rops->random_list, &random_ops); + if (!started) { +- randomproc = kernel_thread(random_proc, NULL, CLONE_FS|CLONE_FILES); +- if (randomproc < 0) { +- ret = randomproc; ++ struct task_struct *t; ++ ++ t = kthread_create(random_proc, NULL, "ocf-random"); ++ if (IS_ERR(t)) { ++ ret = PTR_ERR(t); + printk("crypto: crypto_rregister cannot start random thread; " + "error %d", ret); +- } else ++ } else { ++ random_task = t; ++ wake_up_process(t); + started = 1; ++ } + } + spin_unlock_irqrestore(&random_lock, flags); + +@@ -172,7 +178,7 @@ crypto_runregister_all(u_int32_t driveri + + spin_lock_irqsave(&random_lock, flags); + if (list_empty(&random_ops) && started) +- kill_proc(randomproc, SIGKILL, 1); ++ send_sig(SIGKILL, random_task, 1); + spin_unlock_irqrestore(&random_lock, flags); + return(0); + } +@@ -308,7 +314,7 @@ random_proc(void *arg) + + bad_alloc: + spin_lock_irq(&random_lock); +- randomproc = (pid_t) -1; ++ random_task = NULL; + started = 0; + spin_unlock_irq(&random_lock); + +--- a/crypto/ocf/crypto.c ++++ b/crypto/ocf/crypto.c +@@ -74,6 +74,7 @@ __FBSDID("$FreeBSD: src/sys/opencrypto/c + #include + #include + #include ++#include + #include + + /* +@@ -255,10 +256,10 @@ module_param(crypto_devallowsoft, int, 0 + MODULE_PARM_DESC(crypto_devallowsoft, + "Enable/disable use of software crypto support"); + +-static pid_t cryptoproc = (pid_t) -1; ++static struct task_struct *crypto_task; + static struct completion cryptoproc_exited; + static DECLARE_WAIT_QUEUE_HEAD(cryptoproc_wait); +-static pid_t cryptoretproc = (pid_t) -1; ++static struct task_struct *cryptoret_task; + static struct completion cryptoretproc_exited; + static DECLARE_WAIT_QUEUE_HEAD(cryptoretproc_wait); + +@@ -1401,7 +1402,7 @@ crypto_proc(void *arg) + wait_event_interruptible(cryptoproc_wait, + !(list_empty(&crp_q) || crypto_all_qblocked) || + !(list_empty(&crp_kq) || crypto_all_kqblocked) || +- cryptoproc == (pid_t) -1); ++ crypto_task == NULL); + crp_sleep = 0; + if (signal_pending (current)) { + #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) +@@ -1414,7 +1415,7 @@ crypto_proc(void *arg) + } + CRYPTO_Q_LOCK(); + dprintk("%s - awake\n", __FUNCTION__); +- if (cryptoproc == (pid_t) -1) ++ if (crypto_task == NULL) + break; + cryptostats.cs_intrs++; + } +@@ -1470,7 +1471,7 @@ crypto_ret_proc(void *arg) + dprintk("%s - sleeping\n", __FUNCTION__); + CRYPTO_RETQ_UNLOCK(); + wait_event_interruptible(cryptoretproc_wait, +- cryptoretproc == (pid_t) -1 || ++ cryptoret_task == NULL || + !list_empty(&crp_ret_q) || + !list_empty(&crp_ret_kq)); + if (signal_pending (current)) { +@@ -1484,7 +1485,7 @@ crypto_ret_proc(void *arg) + } + CRYPTO_RETQ_LOCK(); + dprintk("%s - awake\n", __FUNCTION__); +- if (cryptoretproc == (pid_t) -1) { ++ if (cryptoret_task == NULL) { + dprintk("%s - EXITING!\n", __FUNCTION__); + break; + } +@@ -1597,6 +1598,7 @@ DB_SHOW_COMMAND(kcrypto, db_show_kcrypto + static int + crypto_init(void) + { ++ struct task_struct *t; + int error; + + dprintk("%s(0x%x)\n", __FUNCTION__, (int) crypto_init); +@@ -1643,23 +1645,27 @@ crypto_init(void) + init_completion(&cryptoproc_exited); + init_completion(&cryptoretproc_exited); + +- cryptoproc = 0; /* to avoid race condition where proc runs first */ +- cryptoproc = kernel_thread(crypto_proc, NULL, CLONE_FS|CLONE_FILES); +- if (cryptoproc < 0) { +- error = cryptoproc; ++ crypto_task = NULL; /* to avoid race condition where proc runs first */ ++ t = kthread_create(crypto_proc, NULL, "ocf-crypto"); ++ if (IS_ERR(t)) { ++ error = PTR_ERR(t); + printk("crypto: crypto_init cannot start crypto thread; error %d", + error); + goto bad; + } ++ wake_up_process(t); ++ crypto_task = t; + +- cryptoretproc = 0; /* to avoid race condition where proc runs first */ +- cryptoretproc = kernel_thread(crypto_ret_proc, NULL, CLONE_FS|CLONE_FILES); +- if (cryptoretproc < 0) { +- error = cryptoretproc; ++ cryptoret_task = NULL; /* to avoid race condition where proc runs first */ ++ t = kthread_create(crypto_ret_proc, NULL, "ocf-cryptoret"); ++ if (IS_ERR(t)) { ++ error = PTR_ERR(t); + printk("crypto: crypto_init cannot start cryptoret thread; error %d", + error); + goto bad; + } ++ wake_up_process(t); ++ cryptoret_task = t; + + return 0; + bad: +@@ -1671,7 +1677,7 @@ bad: + static void + crypto_exit(void) + { +- pid_t p; ++ struct task_struct *t; + unsigned long d_flags; + + dprintk("%s()\n", __FUNCTION__); +@@ -1681,18 +1687,18 @@ crypto_exit(void) + */ + + CRYPTO_DRIVER_LOCK(); +- p = cryptoproc; +- cryptoproc = (pid_t) -1; +- kill_proc(p, SIGTERM, 1); ++ t = crypto_task; ++ crypto_task = NULL; ++ send_sig(SIGTERM, t, 1); + wake_up_interruptible(&cryptoproc_wait); + CRYPTO_DRIVER_UNLOCK(); + + wait_for_completion(&cryptoproc_exited); + + CRYPTO_DRIVER_LOCK(); +- p = cryptoretproc; +- cryptoretproc = (pid_t) -1; +- kill_proc(p, SIGTERM, 1); ++ t = cryptoret_task; ++ cryptoret_task = NULL; ++ send_sig(SIGTERM, t, 1); + wake_up_interruptible(&cryptoretproc_wait); + CRYPTO_DRIVER_UNLOCK(); + diff --git a/target/linux/generic-2.6/patches-2.6.31/974-ssb_b43_default_on.patch b/target/linux/generic-2.6/patches-2.6.31/974-ssb_b43_default_on.patch new file mode 100644 index 0000000000..98dde2a3c8 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/974-ssb_b43_default_on.patch @@ -0,0 +1,11 @@ +--- a/drivers/ssb/Kconfig ++++ b/drivers/ssb/Kconfig +@@ -49,7 +49,7 @@ config SSB_PCIHOST + config SSB_B43_PCI_BRIDGE + bool + depends on SSB_PCIHOST +- default n ++ default y + + config SSB_PCMCIAHOST_POSSIBLE + bool diff --git a/target/linux/generic-2.6/patches-2.6.31/977-textsearch_kconfig_hacks.patch b/target/linux/generic-2.6/patches-2.6.31/977-textsearch_kconfig_hacks.patch new file mode 100644 index 0000000000..655ebae645 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/977-textsearch_kconfig_hacks.patch @@ -0,0 +1,23 @@ +--- a/lib/Kconfig ++++ b/lib/Kconfig +@@ -145,16 +145,16 @@ config REED_SOLOMON_DEC16 + # Textsearch support is select'ed if needed + # + config TEXTSEARCH +- boolean ++ boolean "Textsearch support" + + config TEXTSEARCH_KMP +- tristate ++ tristate "Textsearch KMP" + + config TEXTSEARCH_BM +- tristate ++ tristate "Textsearch BM" + + config TEXTSEARCH_FSM +- tristate ++ tristate "Textsearch FSM" + + config HAS_IOMEM + boolean diff --git a/target/linux/generic-2.6/patches-2.6.31/978-lib80211_kconfig_hacks.patch b/target/linux/generic-2.6/patches-2.6.31/978-lib80211_kconfig_hacks.patch new file mode 100644 index 0000000000..696ad55602 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/978-lib80211_kconfig_hacks.patch @@ -0,0 +1,19 @@ +--- a/net/wireless/Kconfig ++++ b/net/wireless/Kconfig +@@ -72,13 +72,13 @@ config LIB80211 + you want this built into your kernel. + + config LIB80211_CRYPT_WEP +- tristate ++ tristate "LIB80211_CRYPT_WEP" + + config LIB80211_CRYPT_CCMP +- tristate ++ tristate "LIB80211_CRYPT_CCMP" + + config LIB80211_CRYPT_TKIP +- tristate ++ tristate "LIB80211_CRYPT_TKIP" + + config LIB80211_DEBUG + bool "lib80211 debugging messages" diff --git a/target/linux/generic-2.6/patches-2.6.31/979-crypto_add_kconfig_prompts.patch b/target/linux/generic-2.6/patches-2.6.31/979-crypto_add_kconfig_prompts.patch new file mode 100644 index 0000000000..03ad2a8590 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/979-crypto_add_kconfig_prompts.patch @@ -0,0 +1,47 @@ +--- a/crypto/Kconfig ++++ b/crypto/Kconfig +@@ -30,7 +30,7 @@ config CRYPTO_FIPS + this is. + + config CRYPTO_ALGAPI +- tristate ++ tristate "ALGAPI" + select CRYPTO_ALGAPI2 + help + This option provides the API for cryptographic algorithms. +@@ -39,7 +39,7 @@ config CRYPTO_ALGAPI2 + tristate + + config CRYPTO_AEAD +- tristate ++ tristate "AEAD" + select CRYPTO_AEAD2 + select CRYPTO_ALGAPI + +@@ -48,7 +48,7 @@ config CRYPTO_AEAD2 + select CRYPTO_ALGAPI2 + + config CRYPTO_BLKCIPHER +- tristate ++ tristate "BLKCIPHER" + select CRYPTO_BLKCIPHER2 + select CRYPTO_ALGAPI + +@@ -59,7 +59,7 @@ config CRYPTO_BLKCIPHER2 + select CRYPTO_WORKQUEUE + + config CRYPTO_HASH +- tristate ++ tristate "HASH" + select CRYPTO_HASH2 + select CRYPTO_ALGAPI + +@@ -68,7 +68,7 @@ config CRYPTO_HASH2 + select CRYPTO_ALGAPI2 + + config CRYPTO_RNG +- tristate ++ tristate "RNG" + select CRYPTO_RNG2 + select CRYPTO_ALGAPI + diff --git a/target/linux/generic-2.6/patches-2.6.31/980-vm_exports.patch b/target/linux/generic-2.6/patches-2.6.31/980-vm_exports.patch new file mode 100644 index 0000000000..0513bf3690 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/980-vm_exports.patch @@ -0,0 +1,135 @@ +--- a/mm/shmem.c ++++ b/mm/shmem.c +@@ -2602,6 +2602,16 @@ int shmem_unuse(swp_entry_t entry, struc + + /* common code */ + ++void shmem_set_file(struct vm_area_struct *vma, struct file *file) ++{ ++ ima_counts_get(file); ++ if (vma->vm_file) ++ fput(vma->vm_file); ++ vma->vm_file = file; ++ vma->vm_ops = &shmem_vm_ops; ++} ++EXPORT_SYMBOL_GPL(shmem_set_file); ++ + /** + * shmem_file_setup - get an unlinked file living in tmpfs + * @name: name for dentry (to be seen in /proc//maps +@@ -2681,10 +2691,7 @@ int shmem_zero_setup(struct vm_area_stru + if (IS_ERR(file)) + return PTR_ERR(file); + +- if (vma->vm_file) +- fput(vma->vm_file); +- vma->vm_file = file; +- vma->vm_ops = &shmem_vm_ops; ++ shmem_set_file(vma, file); + return 0; + } + +--- a/fs/file.c ++++ b/fs/file.c +@@ -270,6 +270,7 @@ int expand_files(struct files_struct *fi + /* All good, so we try */ + return expand_fdtable(files, nr); + } ++EXPORT_SYMBOL_GPL(expand_files); + + static int count_open_files(struct fdtable *fdt) + { +--- a/kernel/exit.c ++++ b/kernel/exit.c +@@ -507,6 +507,7 @@ struct files_struct *get_files_struct(st + + return files; + } ++EXPORT_SYMBOL_GPL(get_files_struct); + + void put_files_struct(struct files_struct *files) + { +@@ -526,6 +527,7 @@ void put_files_struct(struct files_struc + free_fdtable(fdt); + } + } ++EXPORT_SYMBOL_GPL(put_files_struct); + + void reset_files_struct(struct files_struct *files) + { +--- a/kernel/fork.c ++++ b/kernel/fork.c +@@ -159,6 +159,7 @@ void __put_task_struct(struct task_struc + if (!profile_handoff_task(tsk)) + free_task(tsk); + } ++EXPORT_SYMBOL_GPL(__put_task_struct); + + /* + * macro override instead of weak attribute alias, to workaround +--- a/kernel/sched.c ++++ b/kernel/sched.c +@@ -5981,6 +5981,7 @@ int can_nice(const struct task_struct *p + return (nice_rlim <= p->signal->rlim[RLIMIT_NICE].rlim_cur || + capable(CAP_SYS_NICE)); + } ++EXPORT_SYMBOL_GPL(can_nice); + + #ifdef __ARCH_WANT_SYS_NICE + +--- a/mm/memory.c ++++ b/mm/memory.c +@@ -1064,6 +1064,7 @@ unsigned long zap_page_range(struct vm_a + tlb_finish_mmu(tlb, address, end); + return end; + } ++EXPORT_SYMBOL_GPL(zap_page_range); + + /** + * zap_vma_ptes - remove ptes mapping the vma +@@ -2490,6 +2491,7 @@ int vmtruncate_range(struct inode *inode + + return 0; + } ++EXPORT_SYMBOL_GPL(vmtruncate_range); + + /* + * We enter with non-exclusive mmap_sem (to exclude vma changes, +--- a/mm/vmalloc.c ++++ b/mm/vmalloc.c +@@ -1101,6 +1101,7 @@ void unmap_kernel_range(unsigned long ad + vunmap_page_range(addr, end); + flush_tlb_kernel_range(addr, end); + } ++EXPORT_SYMBOL_GPL(unmap_kernel_range); + + int map_vm_area(struct vm_struct *area, pgprot_t prot, struct page ***pages) + { +@@ -1214,6 +1215,7 @@ struct vm_struct *get_vm_area(unsigned l + return __get_vm_area_node(size, flags, VMALLOC_START, VMALLOC_END, + -1, GFP_KERNEL, __builtin_return_address(0)); + } ++EXPORT_SYMBOL_GPL(get_vm_area); + + struct vm_struct *get_vm_area_caller(unsigned long size, unsigned long flags, + void *caller) +--- a/include/linux/mm.h ++++ b/include/linux/mm.h +@@ -726,6 +726,7 @@ static inline int shmem_lock(struct file + #endif + struct file *shmem_file_setup(const char *name, loff_t size, unsigned long flags); + ++void shmem_set_file(struct vm_area_struct *vma, struct file *file); + int shmem_zero_setup(struct vm_area_struct *); + + #ifndef CONFIG_MMU +--- a/kernel/signal.c ++++ b/kernel/signal.c +@@ -1061,6 +1061,7 @@ struct sighand_struct *lock_task_sighand + + return sighand; + } ++EXPORT_SYMBOL(lock_task_sighand); + + /* + * send signal info to all the members of a group diff --git a/target/linux/generic-2.6/patches-2.6.31/985-cris-headers.patch b/target/linux/generic-2.6/patches-2.6.31/985-cris-headers.patch new file mode 100644 index 0000000000..73ede933ba --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/985-cris-headers.patch @@ -0,0 +1,27 @@ +--- a/arch/cris/include/arch-v10/arch/Kbuild ++++ b/arch/cris/include/arch-v10/arch/Kbuild +@@ -1,3 +1,5 @@ ++header-y += elf.h ++header-y += ptrace.h + header-y += user.h + header-y += svinto.h + header-y += sv_addr_ag.h +--- a/arch/cris/include/asm/Kbuild ++++ b/arch/cris/include/asm/Kbuild +@@ -1,11 +1,14 @@ + include include/asm-generic/Kbuild.asm + +-header-y += arch-v10/ +-header-y += arch-v32/ ++header-y += ../arch-v10/arch/ ++header-y += ../arch-v32/arch/ + ++header-y += elf.h + header-y += ethernet.h ++header-y += page.h + header-y += rtc.h + header-y += sync_serial.h ++header-y += user.h + + unifdef-y += etraxgpio.h + unifdef-y += rs485.h diff --git a/target/linux/generic-2.6/patches-2.6.31/991-ppc4xx_optimization.patch b/target/linux/generic-2.6/patches-2.6.31/991-ppc4xx_optimization.patch new file mode 100644 index 0000000000..39d5fb0d00 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/991-ppc4xx_optimization.patch @@ -0,0 +1,30 @@ +Upstream doesn't optimize the kernel and bootwrappers for ppc44x because +they still want to support gcc 3.3 -- well, we don't. + +--- a/arch/powerpc/Makefile ++++ b/arch/powerpc/Makefile +@@ -123,7 +123,8 @@ ifeq ($(CONFIG_FUNCTION_TRACER),y) + KBUILD_CFLAGS += -mno-sched-epilog + endif + +-cpu-as-$(CONFIG_4xx) += -Wa,-m405 ++cpu-as-$(CONFIG_40x) += -Wa,-m405 ++cpu-as-$(CONFIG_44x) += -Wa,-m440 + cpu-as-$(CONFIG_6xx) += -Wa,-maltivec + cpu-as-$(CONFIG_POWER4) += -Wa,-maltivec + cpu-as-$(CONFIG_E500) += -Wa,-me500 +--- a/arch/powerpc/boot/Makefile ++++ b/arch/powerpc/boot/Makefile +@@ -38,9 +38,9 @@ BOOTCFLAGS += -I$(obj) -I$(srctree)/$(ob + DTS_FLAGS ?= -p 1024 + + $(obj)/4xx.o: BOOTCFLAGS += -mcpu=405 +-$(obj)/ebony.o: BOOTCFLAGS += -mcpu=405 +-$(obj)/cuboot-taishan.o: BOOTCFLAGS += -mcpu=405 +-$(obj)/cuboot-katmai.o: BOOTCFLAGS += -mcpu=405 ++$(obj)/ebony.o: BOOTCFLAGS += -mcpu=440 ++$(obj)/cuboot-taishan.o: BOOTCFLAGS += -mcpu=440 ++$(obj)/cuboot-katmai.o: BOOTCFLAGS += -mcpu=440 + $(obj)/cuboot-acadia.o: BOOTCFLAGS += -mcpu=405 + $(obj)/treeboot-walnut.o: BOOTCFLAGS += -mcpu=405 + $(obj)/virtex405-head.o: BOOTAFLAGS += -mcpu=405 diff --git a/target/linux/generic-2.6/patches-2.6.31/998-openwrt_lzma_options.patch b/target/linux/generic-2.6/patches-2.6.31/998-openwrt_lzma_options.patch new file mode 100644 index 0000000000..e6eb30a706 --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/998-openwrt_lzma_options.patch @@ -0,0 +1,42 @@ +--- a/scripts/Makefile.lib ++++ b/scripts/Makefile.lib +@@ -228,5 +228,5 @@ cmd_bzip2 = (cat $(filter-out FORCE,$^) + + quiet_cmd_lzma = LZMA $@ + cmd_lzma = (cat $(filter-out FORCE,$^) | \ +- lzma -9 && $(call size_append, $(filter-out FORCE,$^))) > $@ || \ ++ lzma -lc1 -lp2 -pb2 -eos -si -so && $(call size_append, $(filter-out FORCE,$^))) > $@ || \ + (rm -f $@ ; false) +--- a/scripts/gen_initramfs_list.sh ++++ b/scripts/gen_initramfs_list.sh +@@ -225,7 +225,7 @@ cpio_list= + output="/dev/stdout" + output_file="" + is_cpio_compressed= +-compr="gzip -9 -f" ++compr="gzip -9 -f -" + + arg="$1" + case "$arg" in +@@ -239,9 +239,9 @@ case "$arg" in + output_file="$1" + cpio_list="$(mktemp ${TMPDIR:-/tmp}/cpiolist.XXXXXX)" + output=${cpio_list} +- echo "$output_file" | grep -q "\.gz$" && compr="gzip -9 -f" +- echo "$output_file" | grep -q "\.bz2$" && compr="bzip2 -9 -f" +- echo "$output_file" | grep -q "\.lzma$" && compr="lzma -9 -f" ++ echo "$output_file" | grep -q "\.gz$" && compr="gzip -9 -f -" ++ echo "$output_file" | grep -q "\.bz2$" && compr="bzip2 -9 -f -" ++ echo "$output_file" | grep -q "\.lzma$" && compr="lzma e -lc1 -lp2 -pb2 -eos -si -so" + echo "$output_file" | grep -q "\.cpio$" && compr="cat" + shift + ;; +@@ -292,7 +292,7 @@ if [ ! -z ${output_file} ]; then + if [ "${is_cpio_compressed}" = "compressed" ]; then + cat ${cpio_tfile} > ${output_file} + else +- (cat ${cpio_tfile} | ${compr} - > ${output_file}) \ ++ (cat ${cpio_tfile} | ${compr} > ${output_file}) \ + || (rm -f ${output_file} ; false) + fi + [ -z ${cpio_file} ] && rm ${cpio_tfile} diff --git a/target/linux/generic-2.6/patches-2.6.31/999-use_preinit_as_init.patch b/target/linux/generic-2.6/patches-2.6.31/999-use_preinit_as_init.patch new file mode 100644 index 0000000000..25b1ab1f6f --- /dev/null +++ b/target/linux/generic-2.6/patches-2.6.31/999-use_preinit_as_init.patch @@ -0,0 +1,14 @@ +--- a/init/main.c ++++ b/init/main.c +@@ -867,10 +867,7 @@ static noinline int init_post(void) + printk(KERN_WARNING "Failed to execute %s. Attempting " + "defaults...\n", execute_command); + } +- run_init_process("/sbin/init"); +- run_init_process("/etc/init"); +- run_init_process("/bin/init"); +- run_init_process("/bin/sh"); ++ run_init_process("/etc/preinit"); + + panic("No init found. Try passing init= option to kernel."); + }