[ar7] add experimental support for 2.6.30
authorFlorian Fainelli <florian@openwrt.org>
Sun, 2 Aug 2009 13:01:18 +0000 (13:01 +0000)
committerFlorian Fainelli <florian@openwrt.org>
Sun, 2 Aug 2009 13:01:18 +0000 (13:01 +0000)
SVN-Revision: 17097

target/linux/ar7/files/arch/mips/include/asm/ar7 [new symlink]
target/linux/ar7/patches-2.6.30/100-board_support.patch [new file with mode: 0644]
target/linux/ar7/patches-2.6.30/110-flash.patch [new file with mode: 0644]
target/linux/ar7/patches-2.6.30/120-gpio_chrdev.patch [new file with mode: 0644]
target/linux/ar7/patches-2.6.30/130-vlynq.patch [new file with mode: 0644]
target/linux/ar7/patches-2.6.30/131-vlynq_fixes.patch [new file with mode: 0644]
target/linux/ar7/patches-2.6.30/150-cpmac_not_broken.patch [new file with mode: 0644]
target/linux/ar7/patches-2.6.30/500-serial_kludge.patch [new file with mode: 0644]
target/linux/ar7/patches-2.6.30/900-cpmac_multiqueue.patch [new file with mode: 0644]

diff --git a/target/linux/ar7/files/arch/mips/include/asm/ar7 b/target/linux/ar7/files/arch/mips/include/asm/ar7
new file mode 120000 (symlink)
index 0000000..4de8fbc
--- /dev/null
@@ -0,0 +1 @@
+../../../../include/asm-mips/ar7
\ No newline at end of file
diff --git a/target/linux/ar7/patches-2.6.30/100-board_support.patch b/target/linux/ar7/patches-2.6.30/100-board_support.patch
new file mode 100644 (file)
index 0000000..c3fad16
--- /dev/null
@@ -0,0 +1,86 @@
+--- a/arch/mips/Kconfig
++++ b/arch/mips/Kconfig
+@@ -19,6 +19,24 @@ choice
+       prompt "System type"
+       default SGI_IP22
++config AR7
++      bool "Texas Instruments AR7"
++      select BOOT_ELF32
++      select DMA_NONCOHERENT
++      select CEVT_R4K
++      select CSRC_R4K
++      select IRQ_CPU
++      select NO_EXCEPT_FILL
++      select SWAP_IO_SPACE
++      select SYS_HAS_CPU_MIPS32_R1
++      select SYS_HAS_EARLY_PRINTK
++      select SYS_SUPPORTS_32BIT_KERNEL
++      select SYS_SUPPORTS_KGDB
++      select SYS_SUPPORTS_LITTLE_ENDIAN
++      select SYS_SUPPORTS_BIG_ENDIAN
++      select GENERIC_GPIO
++      select GENERIC_HARDIRQS_NO__DO_IRQ
++
+ config MACH_ALCHEMY
+       bool "Alchemy processor based machines"
+--- a/arch/mips/kernel/traps.c
++++ b/arch/mips/kernel/traps.c
+@@ -1203,9 +1203,22 @@ void *set_except_vector(int n, void *add
+       exception_handlers[n] = handler;
+       if (n == 0 && cpu_has_divec) {
+-              *(u32 *)(ebase + 0x200) = 0x08000000 |
+-                                        (0x03ffffff & (handler >> 2));
+-              local_flush_icache_range(ebase + 0x200, ebase + 0x204);
++              if ((handler ^ (ebase + 4)) & 0xfc000000) {
++                      /* lui k0, 0x0000 */
++                      *(u32 *)(ebase + 0x200) = 0x3c1a0000 | (handler >> 16);
++                      /* ori k0, 0x0000 */
++                      *(u32 *)(ebase + 0x204) =
++                                      0x375a0000 | (handler & 0xffff);
++                      /* jr k0 */
++                      *(u32 *)(ebase + 0x208) = 0x03400008;
++                      /* nop */
++                      *(u32 *)(ebase + 0x20C) = 0x00000000;
++                      flush_icache_range(ebase + 0x200, ebase + 0x210);
++              } else {
++                      *(u32 *)(ebase + 0x200) =
++                              0x08000000 | (0x03ffffff & (handler >> 2));
++                      flush_icache_range(ebase + 0x200, ebase + 0x204);
++              }
+       }
+       return (void *)old_handler;
+ }
+--- a/arch/mips/Makefile
++++ b/arch/mips/Makefile
+@@ -167,6 +167,13 @@ libs-$(CONFIG_SIBYTE_CFE) += arch/mips/s
+ #
+ #
++# Texas Instruments AR7
++#
++core-$(CONFIG_AR7)            += arch/mips/ar7/
++cflags-$(CONFIG_AR7)          += -Iinclude/asm-mips/ar7
++load-$(CONFIG_AR7)            += 0xffffffff94100000
++
++#
+ # Acer PICA 61, Mips Magnum 4000 and Olivetti M700.
+ #
+ core-$(CONFIG_MACH_JAZZ)      += arch/mips/jazz/
+--- a/arch/mips/include/asm/page.h
++++ b/arch/mips/include/asm/page.h
+@@ -182,8 +182,10 @@ typedef struct { unsigned long pgprot; }
+ #define VM_DATA_DEFAULT_FLAGS (VM_READ | VM_WRITE | VM_EXEC | \
+                                VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC)
+-#define UNCAC_ADDR(addr)      ((addr) - PAGE_OFFSET + UNCAC_BASE)
+-#define CAC_ADDR(addr)                ((addr) - UNCAC_BASE + PAGE_OFFSET)
++#define UNCAC_ADDR(addr)      ((addr) - PAGE_OFFSET + UNCAC_BASE +    \
++                               PHYS_OFFSET)
++#define CAC_ADDR(addr)                ((addr) - UNCAC_BASE + PAGE_OFFSET -    \
++                               PHYS_OFFSET)
+ #include <asm-generic/memory_model.h>
+ #include <asm-generic/page.h>
diff --git a/target/linux/ar7/patches-2.6.30/110-flash.patch b/target/linux/ar7/patches-2.6.30/110-flash.patch
new file mode 100644 (file)
index 0000000..7311a67
--- /dev/null
@@ -0,0 +1,11 @@
+--- a/drivers/mtd/maps/physmap.c
++++ b/drivers/mtd/maps/physmap.c
+@@ -80,7 +80,7 @@ static const char *rom_probe_types[] = {
+                                       "map_rom",
+                                       NULL };
+ #ifdef CONFIG_MTD_PARTITIONS
+-static const char *part_probe_types[] = { "cmdlinepart", "RedBoot", NULL };
++static const char *part_probe_types[] = { "cmdlinepart", "RedBoot", "ar7part", NULL };
+ #endif
+ static int physmap_flash_probe(struct platform_device *dev)
diff --git a/target/linux/ar7/patches-2.6.30/120-gpio_chrdev.patch b/target/linux/ar7/patches-2.6.30/120-gpio_chrdev.patch
new file mode 100644 (file)
index 0000000..fa61e5c
--- /dev/null
@@ -0,0 +1,28 @@
+--- a/drivers/char/Kconfig
++++ b/drivers/char/Kconfig
+@@ -974,6 +974,15 @@ config MWAVE
+         To compile this driver as a module, choose M here: the
+         module will be called mwave.
++config AR7_GPIO
++      tristate "TI AR7 GPIO Support"
++      depends on AR7
++      help
++        Give userspace access to the GPIO pins on the Texas Instruments AR7 
++        processors.
++
++        If compiled as a module, it will be called ar7_gpio.
++
+ config SCx200_GPIO
+       tristate "NatSemi SCx200 GPIO Support"
+       depends on SCx200
+--- a/drivers/char/Makefile
++++ b/drivers/char/Makefile
+@@ -90,6 +90,7 @@ obj-$(CONFIG_HW_RANDOM)              += hw_random/
+ obj-$(CONFIG_PPDEV)           += ppdev.o
+ obj-$(CONFIG_NWBUTTON)                += nwbutton.o
+ obj-$(CONFIG_NWFLASH)         += nwflash.o
++obj-$(CONFIG_AR7_GPIO)                += ar7_gpio.o
+ obj-$(CONFIG_SCx200_GPIO)     += scx200_gpio.o
+ obj-$(CONFIG_PC8736x_GPIO)    += pc8736x_gpio.o
+ obj-$(CONFIG_NSC_GPIO)                += nsc_gpio.o
diff --git a/target/linux/ar7/patches-2.6.30/130-vlynq.patch b/target/linux/ar7/patches-2.6.30/130-vlynq.patch
new file mode 100644 (file)
index 0000000..9ab3638
--- /dev/null
@@ -0,0 +1,20 @@
+--- a/drivers/Kconfig
++++ b/drivers/Kconfig
+@@ -100,5 +100,7 @@ source "drivers/auxdisplay/Kconfig"
+ source "drivers/uio/Kconfig"
++source "drivers/vlynq/Kconfig"
++
+ source "drivers/xen/Kconfig"
+ endmenu
+--- a/drivers/Makefile
++++ b/drivers/Makefile
+@@ -96,6 +96,7 @@ obj-$(CONFIG_DCA)            += dca/
+ obj-$(CONFIG_HID)             += hid/
+ obj-$(CONFIG_PPC_PS3)         += ps3/
+ obj-$(CONFIG_OF)              += of/
++obj-$(CONFIG_VLYNQ)           += vlynq/
+ obj-$(CONFIG_SSB)             += ssb/
+ obj-$(CONFIG_VIRTIO)          += virtio/
+ obj-$(CONFIG_REGULATOR)               += regulator/
diff --git a/target/linux/ar7/patches-2.6.30/131-vlynq_fixes.patch b/target/linux/ar7/patches-2.6.30/131-vlynq_fixes.patch
new file mode 100644 (file)
index 0000000..f68ae4c
--- /dev/null
@@ -0,0 +1,548 @@
+--- a/drivers/vlynq/vlynq.c    2009-05-31 20:41:57.000000000 +0200
++++ b/drivers/vlynq/vlynq.c    2009-07-28 21:27:52.000000000 +0200
+@@ -14,6 +14,9 @@
+  * You should have received a copy of the GNU General Public License
+  * along with this program; if not, write to the Free Software
+  * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
++ *
++ * Parts of the VLYNQ specification can be found here:
++ * http://www.ti.com/litv/pdf/sprue36a
+  */
+ #include <linux/init.h>
+@@ -25,7 +28,6 @@
+ #include <linux/errno.h>
+ #include <linux/platform_device.h>
+ #include <linux/interrupt.h>
+-#include <linux/device.h>
+ #include <linux/delay.h>
+ #include <linux/io.h>
+@@ -73,15 +75,11 @@
+       u32 int_device[8];
+ };
+-#define vlynq_reg_read(reg) readl(&(reg))
+-#define vlynq_reg_write(reg, val)  writel(val, &(reg))
+-
+-static int __vlynq_enable_device(struct vlynq_device *dev);
+-
+-#ifdef VLYNQ_DEBUG
++#ifdef CONFIG_VLYNQ_DEBUG
+ static void vlynq_dump_regs(struct vlynq_device *dev)
+ {
+       int i;
++
+       printk(KERN_DEBUG "VLYNQ local=%p remote=%p\n",
+                       dev->local, dev->remote);
+       for (i = 0; i < 32; i++) {
+@@ -95,20 +93,23 @@
+ static void vlynq_dump_mem(u32 *base, int count)
+ {
+       int i;
++
+       for (i = 0; i < (count + 3) / 4; i++) {
+-              if (i % 4 == 0) printk(KERN_DEBUG "\nMEM[0x%04x]:", i * 4);
++              if (i % 4 == 0)
++                      printk(KERN_DEBUG "\nMEM[0x%04x]:", i * 4);
+               printk(KERN_DEBUG " 0x%08x", *(base + i));
+       }
+       printk(KERN_DEBUG "\n");
+ }
+ #endif
+-int vlynq_linked(struct vlynq_device *dev)
++/* Check the VLYNQ link status with a given device */
++static int vlynq_linked(struct vlynq_device *dev)
+ {
+       int i;
+       for (i = 0; i < 100; i++)
+-              if (vlynq_reg_read(dev->local->status) & VLYNQ_STATUS_LINK)
++              if (readl(&dev->local->status) & VLYNQ_STATUS_LINK)
+                       return 1;
+               else
+                       cpu_relax();
+@@ -118,17 +119,15 @@
+ static void vlynq_reset(struct vlynq_device *dev)
+ {
+-      vlynq_reg_write(dev->local->control,
+-                      vlynq_reg_read(dev->local->control) |
+-                      VLYNQ_CTRL_RESET);
++      writel(readl(&dev->local->control) | VLYNQ_CTRL_RESET,
++                      &dev->local->control);
+       /* Wait for the devices to finish resetting */
+       msleep(5);
+       /* Remove reset bit */
+-      vlynq_reg_write(dev->local->control,
+-                      vlynq_reg_read(dev->local->control) &
+-                      ~VLYNQ_CTRL_RESET);
++      writel(readl(&dev->local->control) & ~VLYNQ_CTRL_RESET,
++                      &dev->local->control);
+       /* Give some time for the devices to settle */
+       msleep(5);
+@@ -142,9 +141,9 @@
+       BUG_ON(!dev);
+       virq = irq - dev->irq_start;
+-      val = vlynq_reg_read(dev->remote->int_device[virq >> 2]);
++      val = readl(&dev->remote->int_device[virq >> 2]);
+       val |= (VINT_ENABLE | virq) << VINT_OFFSET(virq);
+-      vlynq_reg_write(dev->remote->int_device[virq >> 2], val);
++      writel(val, &dev->remote->int_device[virq >> 2]);
+ }
+ static void vlynq_irq_mask(unsigned int irq)
+@@ -155,9 +154,9 @@
+       BUG_ON(!dev);
+       virq = irq - dev->irq_start;
+-      val = vlynq_reg_read(dev->remote->int_device[virq >> 2]);
++      val = readl(&dev->remote->int_device[virq >> 2]);
+       val &= ~(VINT_ENABLE << VINT_OFFSET(virq));
+-      vlynq_reg_write(dev->remote->int_device[virq >> 2], val);
++      writel(val, &dev->remote->int_device[virq >> 2]);
+ }
+ static int vlynq_irq_type(unsigned int irq, unsigned int flow_type)
+@@ -168,7 +167,7 @@
+       BUG_ON(!dev);
+       virq = irq - dev->irq_start;
+-      val = vlynq_reg_read(dev->remote->int_device[virq >> 2]);
++      val = readl(&dev->remote->int_device[virq >> 2]);
+       switch (flow_type & IRQ_TYPE_SENSE_MASK) {
+       case IRQ_TYPE_EDGE_RISING:
+       case IRQ_TYPE_EDGE_FALLING:
+@@ -187,28 +186,30 @@
+       default:
+               return -EINVAL;
+       }
+-      vlynq_reg_write(dev->remote->int_device[virq >> 2], val);
++      writel(val, &dev->remote->int_device[virq >> 2]);
+       return 0;
+ }
+ static void vlynq_local_ack(unsigned int irq)
+ {
+       struct vlynq_device *dev = get_irq_chip_data(irq);
+-      u32 status = vlynq_reg_read(dev->local->status);
+-      if (printk_ratelimit())
+-              printk(KERN_DEBUG "%s: local status: 0x%08x\n",
+-                     dev->dev.bus_id, status);
+-      vlynq_reg_write(dev->local->status, status);
++
++      u32 status = readl(&dev->local->status);
++
++      pr_debug("%s: local status: 0x%08x\n",
++                     dev_name(&dev->dev), status);
++      writel(status, &dev->local->status);
+ }
+ static void vlynq_remote_ack(unsigned int irq)
+ {
+       struct vlynq_device *dev = get_irq_chip_data(irq);
+-      u32 status = vlynq_reg_read(dev->remote->status);
+-      if (printk_ratelimit())
+-              printk(KERN_DEBUG "%s: remote status: 0x%08x\n",
+-                     dev->dev.bus_id, status);
+-      vlynq_reg_write(dev->remote->status, status);
++
++      u32 status = readl(&dev->remote->status);
++
++      pr_debug("%s: remote status: 0x%08x\n",
++                     dev_name(&dev->dev), status);
++      writel(status, &dev->remote->status);
+ }
+ static irqreturn_t vlynq_irq(int irq, void *dev_id)
+@@ -217,8 +218,8 @@
+       u32 status;
+       int virq = 0;
+-      status = vlynq_reg_read(dev->local->int_status);
+-      vlynq_reg_write(dev->local->int_status, status);
++      status = readl(&dev->local->int_status);
++      writel(status, &dev->local->int_status);
+       if (unlikely(!status))
+               spurious_interrupt();
+@@ -262,28 +263,28 @@
+       if (dev->local_irq == dev->remote_irq) {
+               printk(KERN_ERR
+                      "%s: local vlynq irq should be different from remote\n",
+-                     dev->dev.bus_id);
++                     dev_name(&dev->dev));
+               return -EINVAL;
+       }
+       /* Clear local and remote error bits */
+-      vlynq_reg_write(dev->local->status, vlynq_reg_read(dev->local->status));
+-      vlynq_reg_write(dev->remote->status,
+-                      vlynq_reg_read(dev->remote->status));
++      writel(readl(&dev->local->status), &dev->local->status);
++      writel(readl(&dev->remote->status), &dev->remote->status);
+       /* Now setup interrupts */
+       val = VLYNQ_CTRL_INT_VECTOR(dev->local_irq);
+       val |= VLYNQ_CTRL_INT_ENABLE | VLYNQ_CTRL_INT_LOCAL |
+               VLYNQ_CTRL_INT2CFG;
+-      val |= vlynq_reg_read(dev->local->control);
+-      vlynq_reg_write(dev->local->int_ptr, VLYNQ_INT_OFFSET);
+-      vlynq_reg_write(dev->local->control, val);
++      val |= readl(&dev->local->control);
++      writel(VLYNQ_INT_OFFSET, &dev->local->int_ptr);
++      writel(val, &dev->local->control);
+       val = VLYNQ_CTRL_INT_VECTOR(dev->remote_irq);
+       val |= VLYNQ_CTRL_INT_ENABLE;
+-      val |= vlynq_reg_read(dev->remote->control);
+-      vlynq_reg_write(dev->remote->int_ptr, VLYNQ_INT_OFFSET);
+-      vlynq_reg_write(dev->remote->control, val);
++      val |= readl(&dev->remote->control);
++      writel(VLYNQ_INT_OFFSET, &dev->remote->int_ptr);
++      writel(val, &dev->remote->int_ptr);
++      writel(val, &dev->remote->control);
+       for (i = dev->irq_start; i <= dev->irq_end; i++) {
+               virq = i - dev->irq_start;
+@@ -299,12 +300,13 @@
+                       set_irq_chip_and_handler(i, &vlynq_irq_chip,
+                                                handle_simple_irq);
+                       set_irq_chip_data(i, dev);
+-                      vlynq_reg_write(dev->remote->int_device[virq >> 2], 0);
++                      writel(0, &dev->remote->int_device[virq >> 2]);
+               }
+       }
+       if (request_irq(dev->irq, vlynq_irq, IRQF_SHARED, "vlynq", dev)) {
+-              printk(KERN_ERR "%s: request_irq failed\n", dev->dev.bus_id);
++              printk(KERN_ERR "%s: request_irq failed\n",
++                                      dev_name(&dev->dev));
+               return -EAGAIN;
+       }
+@@ -328,11 +330,11 @@
+               if (ids->id == vdev->dev_id) {
+                       vdev->divisor = ids->divisor;
+                       vlynq_set_drvdata(vdev, ids);
+-                      printk(KERN_INFO "Driver found for VLYNQ " \
++                      printk(KERN_INFO "Driver found for VLYNQ "
+                               "device: %08x\n", vdev->dev_id);
+                       return 1;
+               }
+-              printk(KERN_DEBUG "Not using the %08x VLYNQ device's driver" \
++              printk(KERN_DEBUG "Not using the %08x VLYNQ device's driver"
+                       " for VLYNQ device: %08x\n", ids->id, vdev->dev_id);
+               ids++;
+       }
+@@ -346,8 +348,7 @@
+       struct vlynq_device_id *id = vlynq_get_drvdata(vdev);
+       int result = -ENODEV;
+-      get_device(dev);
+-      if (drv && drv->probe)
++      if (drv->probe)
+               result = drv->probe(vdev, id);
+       if (result)
+               put_device(dev);
+@@ -357,9 +358,10 @@
+ static int vlynq_device_remove(struct device *dev)
+ {
+       struct vlynq_driver *drv = to_vlynq_driver(dev->driver);
+-      if (drv && drv->remove)
++
++      if (drv->remove)
+               drv->remove(to_vlynq_device(dev));
+-      put_device(dev);
++
+       return 0;
+ }
+@@ -377,6 +379,14 @@
+ }
+ EXPORT_SYMBOL(vlynq_unregister_driver);
++/*
++ * A VLYNQ remote device can clock the VLYNQ bus master
++ * using a dedicated clock line. In that case, both the
++ * remove device and the bus master should have the same
++ * serial clock dividers configured. Iterate through the
++ * 8 possible dividers until we actually link with the
++ * device.
++ */
+ static int __vlynq_try_remote(struct vlynq_device *dev)
+ {
+       int i;
+@@ -389,21 +399,21 @@
+               if (!vlynq_linked(dev))
+                       break;
+-              vlynq_reg_write(dev->remote->control,
+-                              (vlynq_reg_read(dev->remote->control) &
++              writel((readl(&dev->remote->control) &
+                               ~VLYNQ_CTRL_CLOCK_MASK) |
+                               VLYNQ_CTRL_CLOCK_INT |
+-                              VLYNQ_CTRL_CLOCK_DIV(i - vlynq_rdiv1));
+-              vlynq_reg_write(dev->local->control,
+-                              ((vlynq_reg_read(dev->local->control)
++                              VLYNQ_CTRL_CLOCK_DIV(i - vlynq_rdiv1),
++                              &dev->remote->control);
++              writel((readl(&dev->local->control)
+                               & ~(VLYNQ_CTRL_CLOCK_INT |
+                               VLYNQ_CTRL_CLOCK_MASK)) |
+-                              VLYNQ_CTRL_CLOCK_DIV(i - vlynq_rdiv1)));
++                              VLYNQ_CTRL_CLOCK_DIV(i - vlynq_rdiv1),
++                              &dev->local->control);
+               if (vlynq_linked(dev)) {
+                       printk(KERN_DEBUG
+                               "%s: using remote clock divisor %d\n",
+-                              dev->dev.bus_id, i - vlynq_rdiv1 + 1);
++                              dev_name(&dev->dev), i - vlynq_rdiv1 + 1);
+                       dev->divisor = i;
+                       return 0;
+               } else {
+@@ -414,26 +424,33 @@
+       return -ENODEV;
+ }
++/*
++ * A VLYNQ remote device can be clocked by the VLYNQ bus
++ * master using a dedicated clock line. In that case, only
++ * the bus master configures the serial clock divider.
++ * Iterate through the 8 possible dividers until we
++ * actually get a link with the device.
++ */
+ static int __vlynq_try_local(struct vlynq_device *dev)
+ {
+       int i;
+-      
++
+       vlynq_reset(dev);
+       for (i = dev->dev_id ? vlynq_ldiv2 : vlynq_ldiv8; dev->dev_id ?
+                       i <= vlynq_ldiv8 : i >= vlynq_ldiv2;
+               dev->dev_id ? i++ : i--) {
+-              vlynq_reg_write(dev->local->control,
+-                              (vlynq_reg_read(dev->local->control) &
++              writel((readl(&dev->local->control) &
+                               ~VLYNQ_CTRL_CLOCK_MASK) |
+                               VLYNQ_CTRL_CLOCK_INT |
+-                              VLYNQ_CTRL_CLOCK_DIV(i - vlynq_ldiv1));
++                              VLYNQ_CTRL_CLOCK_DIV(i - vlynq_ldiv1),
++                              &dev->local->control);
+               if (vlynq_linked(dev)) {
+                       printk(KERN_DEBUG
+                               "%s: using local clock divisor %d\n",
+-                              dev->dev.bus_id, i - vlynq_ldiv1 + 1);
++                              dev_name(&dev->dev), i - vlynq_ldiv1 + 1);
+                       dev->divisor = i;
+                       return 0;
+               } else {
+@@ -444,27 +461,33 @@
+       return -ENODEV;
+ }
++/*
++ * When using external clocking method, serial clock
++ * is supplied by an external oscillator, therefore we
++ * should mask the local clock bit in the clock control
++ * register for both the bus master and the remote device.
++ */
+ static int __vlynq_try_external(struct vlynq_device *dev)
+ {
+       vlynq_reset(dev);
+       if (!vlynq_linked(dev))
+               return -ENODEV;
+-      vlynq_reg_write(dev->remote->control,
+-                      (vlynq_reg_read(dev->remote->control) &
+-                      ~VLYNQ_CTRL_CLOCK_INT));
+-
+-      vlynq_reg_write(dev->local->control,
+-                      (vlynq_reg_read(dev->local->control) &
+-                      ~VLYNQ_CTRL_CLOCK_INT));
++      writel((readl(&dev->remote->control) &
++                      ~VLYNQ_CTRL_CLOCK_INT),
++                      &dev->remote->control);
++
++      writel((readl(&dev->local->control) &
++                      ~VLYNQ_CTRL_CLOCK_INT),
++                      &dev->local->control);
+       if (vlynq_linked(dev)) {
+               printk(KERN_DEBUG "%s: using external clock\n",
+-                      dev->dev.bus_id);
++                      dev_name(&dev->dev));
+                       dev->divisor = vlynq_div_external;
+               return 0;
+       }
+-      
++
+       return -ENODEV;
+ }
+@@ -481,10 +504,10 @@
+       case vlynq_div_external:
+       case vlynq_div_auto:
+               /* When the device is brought from reset it should have clock
+-              generation negotiated by hardware.
+-              Check which device is generating clocks and perform setup
+-              accordingly */
+-              if (vlynq_linked(dev) && vlynq_reg_read(dev->remote->control) &
++               * generation negotiated by hardware.
++               * Check which device is generating clocks and perform setup
++               * accordingly */
++              if (vlynq_linked(dev) && readl(&dev->remote->control) &
+                  VLYNQ_CTRL_CLOCK_INT) {
+                       if (!__vlynq_try_remote(dev) ||
+                               !__vlynq_try_local(dev)  ||
+@@ -497,31 +520,43 @@
+                               return 0;
+               }
+               break;
+-      case vlynq_ldiv1: case vlynq_ldiv2: case vlynq_ldiv3: case vlynq_ldiv4:
+-      case vlynq_ldiv5: case vlynq_ldiv6: case vlynq_ldiv7: case vlynq_ldiv8:
+-              vlynq_reg_write(dev->local->control,
+-                              VLYNQ_CTRL_CLOCK_INT |
+-                              VLYNQ_CTRL_CLOCK_DIV(dev->divisor -
+-                                                   vlynq_ldiv1));
+-              vlynq_reg_write(dev->remote->control, 0);
++      case vlynq_ldiv1:
++      case vlynq_ldiv2:
++      case vlynq_ldiv3:
++      case vlynq_ldiv4:
++      case vlynq_ldiv5:
++      case vlynq_ldiv6:
++      case vlynq_ldiv7:
++      case vlynq_ldiv8:
++              writel(VLYNQ_CTRL_CLOCK_INT |
++                      VLYNQ_CTRL_CLOCK_DIV(dev->divisor -
++                      vlynq_ldiv1), &dev->local->control);
++              writel(0, &dev->remote->control);
+               if (vlynq_linked(dev)) {
+                       printk(KERN_DEBUG
+-                             "%s: using local clock divisor %d\n",
+-                             dev->dev.bus_id, dev->divisor - vlynq_ldiv1 + 1);
++                              "%s: using local clock divisor %d\n",
++                              dev_name(&dev->dev),
++                              dev->divisor - vlynq_ldiv1 + 1);
+                       return 0;
+               }
+               break;
+-      case vlynq_rdiv1: case vlynq_rdiv2: case vlynq_rdiv3: case vlynq_rdiv4:
+-      case vlynq_rdiv5: case vlynq_rdiv6: case vlynq_rdiv7: case vlynq_rdiv8:
+-              vlynq_reg_write(dev->local->control, 0);
+-              vlynq_reg_write(dev->remote->control,
+-                              VLYNQ_CTRL_CLOCK_INT |
+-                              VLYNQ_CTRL_CLOCK_DIV(dev->divisor -
+-                                                   vlynq_rdiv1));
++      case vlynq_rdiv1:
++      case vlynq_rdiv2:
++      case vlynq_rdiv3:
++      case vlynq_rdiv4:
++      case vlynq_rdiv5:
++      case vlynq_rdiv6:
++      case vlynq_rdiv7:
++      case vlynq_rdiv8:
++              writel(0, &dev->local->control);
++              writel(VLYNQ_CTRL_CLOCK_INT |
++                      VLYNQ_CTRL_CLOCK_DIV(dev->divisor -
++                      vlynq_rdiv1), &dev->remote->control);
+               if (vlynq_linked(dev)) {
+                       printk(KERN_DEBUG
+-                             "%s: using remote clock divisor %d\n",
+-                             dev->dev.bus_id, dev->divisor - vlynq_rdiv1 + 1);
++                              "%s: using remote clock divisor %d\n",
++                              dev_name(&dev->dev),
++                              dev->divisor - vlynq_rdiv1 + 1);
+                       return 0;
+               }
+               break;
+@@ -568,12 +603,10 @@
+       if (!dev->enabled)
+               return -ENXIO;
+-      vlynq_reg_write(dev->local->tx_offset, tx_offset);
++      writel(tx_offset, &dev->local->tx_offset);
+       for (i = 0; i < 4; i++) {
+-              vlynq_reg_write(dev->local->rx_mapping[i].offset,
+-                                                      mapping[i].offset);
+-              vlynq_reg_write(dev->local->rx_mapping[i].size,
+-                                                      mapping[i].size);
++              writel(mapping[i].offset, &dev->local->rx_mapping[i].offset);
++              writel(mapping[i].size, &dev->local->rx_mapping[i].size);
+       }
+       return 0;
+ }
+@@ -587,12 +620,10 @@
+       if (!dev->enabled)
+               return -ENXIO;
+-      vlynq_reg_write(dev->remote->tx_offset, tx_offset);
++      writel(tx_offset, &dev->remote->tx_offset);
+       for (i = 0; i < 4; i++) {
+-              vlynq_reg_write(dev->remote->rx_mapping[i].offset,
+-                                                      mapping[i].offset);
+-              vlynq_reg_write(dev->remote->rx_mapping[i].size,
+-                                                      mapping[i].size);
++              writel(mapping[i].offset, &dev->remote->rx_mapping[i].offset);
++              writel(mapping[i].size, &dev->remote->rx_mapping[i].size);
+       }
+       return 0;
+ }
+@@ -662,8 +693,7 @@
+       dev->id = pdev->id;
+       dev->dev.bus = &vlynq_bus_type;
+       dev->dev.parent = &pdev->dev;
+-      snprintf(dev->dev.bus_id, BUS_ID_SIZE, "vlynq%d", dev->id);
+-      dev->dev.bus_id[BUS_ID_SIZE - 1] = 0;
++      dev_set_name(&dev->dev, "vlynq%d", dev->id);
+       dev->dev.platform_data = pdev->dev.platform_data;
+       dev->dev.release = vlynq_device_release;
+@@ -673,9 +703,9 @@
+       dev->mem_end = mem_res->end;
+       len = regs_res->end - regs_res->start;
+-      if (!request_mem_region(regs_res->start, len, dev->dev.bus_id)) {
++      if (!request_mem_region(regs_res->start, len, dev_name(&dev->dev))) {
+               printk(KERN_ERR "%s: Can't request vlynq registers\n",
+-                     dev->dev.bus_id);
++                     dev_name(&dev->dev));
+               result = -ENXIO;
+               goto fail_request;
+       }
+@@ -683,7 +713,7 @@
+       dev->local = ioremap(regs_res->start, len);
+       if (!dev->local) {
+               printk(KERN_ERR "%s: Can't remap vlynq registers\n",
+-                     dev->dev.bus_id);
++                     dev_name(&dev->dev));
+               result = -ENXIO;
+               goto fail_remap;
+       }
+@@ -702,14 +732,14 @@
+       platform_set_drvdata(pdev, dev);
+       printk(KERN_INFO "%s: regs 0x%p, irq %d, mem 0x%p\n",
+-             dev->dev.bus_id, (void *)dev->regs_start, dev->irq,
++             dev_name(&dev->dev), (void *)dev->regs_start, dev->irq,
+              (void *)dev->mem_start);
+       dev->dev_id = 0;
+       dev->divisor = vlynq_div_auto;
+       result = __vlynq_enable_device(dev);
+       if (result == 0) {
+-              dev->dev_id = vlynq_reg_read(dev->remote->chip);
++              dev->dev_id = readl(&dev->remote->chip);
+               ((struct plat_vlynq_ops *)(dev->dev.platform_data))->off(dev);
+       }
+       if (dev->dev_id)
diff --git a/target/linux/ar7/patches-2.6.30/150-cpmac_not_broken.patch b/target/linux/ar7/patches-2.6.30/150-cpmac_not_broken.patch
new file mode 100644 (file)
index 0000000..0c8a20d
--- /dev/null
@@ -0,0 +1,11 @@
+--- a/drivers/net/Kconfig
++++ b/drivers/net/Kconfig
+@@ -1883,7 +1883,7 @@ config SC92031
+ config CPMAC
+       tristate "TI AR7 CPMAC Ethernet support (EXPERIMENTAL)"
+-      depends on NET_ETHERNET && EXPERIMENTAL && AR7 && BROKEN
++      depends on NET_ETHERNET && EXPERIMENTAL && AR7
+       select PHYLIB
+       help
+         TI AR7 CPMAC Ethernet support
diff --git a/target/linux/ar7/patches-2.6.30/500-serial_kludge.patch b/target/linux/ar7/patches-2.6.30/500-serial_kludge.patch
new file mode 100644 (file)
index 0000000..d4b02bc
--- /dev/null
@@ -0,0 +1,38 @@
+--- a/drivers/serial/8250.c
++++ b/drivers/serial/8250.c
+@@ -287,6 +287,13 @@ static const struct serial8250_config ua
+               .fcr            = UART_FCR_ENABLE_FIFO | UART_FCR_R_TRIG_10,
+               .flags          = UART_CAP_FIFO,
+       },
++      [PORT_AR7] = {
++              .name           = "TI-AR7",
++              .fifo_size      = 16,
++              .tx_loadsz      = 16,
++              .fcr            = UART_FCR_ENABLE_FIFO | UART_FCR_R_TRIG_00,
++              .flags          = UART_CAP_FIFO | UART_CAP_AFE,
++      },
+ };
+ #if defined (CONFIG_SERIAL_8250_AU1X00)
+@@ -2702,7 +2709,11 @@ static void serial8250_console_putchar(s
+ {
+       struct uart_8250_port *up = (struct uart_8250_port *)port;
++#ifdef CONFIG_AR7
++      wait_for_xmitr(up, BOTH_EMPTY);
++#else
+       wait_for_xmitr(up, UART_LSR_THRE);
++#endif
+       serial_out(up, UART_TX, ch);
+ }
+--- a/include/linux/serial_core.h
++++ b/include/linux/serial_core.h
+@@ -41,6 +41,7 @@
+ #define PORT_XSCALE   15
+ #define PORT_RM9000   16      /* PMC-Sierra RM9xxx internal UART */
+ #define PORT_OCTEON   17      /* Cavium OCTEON internal UART */
++#define PORT_AR7      18      /* TI AR7 internal UART */
+ #define PORT_MAX_8250 17      /* max port ID */
+ /*
diff --git a/target/linux/ar7/patches-2.6.30/900-cpmac_multiqueue.patch b/target/linux/ar7/patches-2.6.30/900-cpmac_multiqueue.patch
new file mode 100644 (file)
index 0000000..3df3d68
--- /dev/null
@@ -0,0 +1,70 @@
+This patch fixes the network driver cpmac.c for compilation with
+configuration option CONFIG_NETDEVICES_MULTIQUEUE.
+
+These compiler warnings are fixed by the patch:
+drivers/net/cpmac.c: In function 'cpmac_end_xmit':
+drivers/net/cpmac.c:630: warning: passing argument 2 of 'netif_subqueue_stopped' makes pointer from integer without a cast
+drivers/net/cpmac.c:641: warning: passing argument 2 of 'netif_subqueue_stopped' makes pointer from integer without a cast
+drivers/net/cpmac.c: In function 'cpmac_probe':
+drivers/net/cpmac.c:1128: warning: unused variable 'i'
+
+During runtime, the unpatched driver raises a fatal runtime exception.
+This is fixed by calling __netif_subqueue_stopped instead
+of netif_subqueue_stopped, too.
+
+Two additional code parts were modified for CONFIG_NETDEVICES_MULTIQUEUE
+because other drivers do it in the same way.
+
+        Signed-off-by: Stefan Weil <weil@mail.berlios.de>
+
+--- a/drivers/net/cpmac.c
++++ b/drivers/net/cpmac.c
+@@ -615,13 +615,13 @@ static void cpmac_end_xmit(struct net_de
+               dev_kfree_skb_irq(desc->skb);
+               desc->skb = NULL;
+-              if (netif_subqueue_stopped(dev, queue))
++              if (__netif_subqueue_stopped(dev, queue))
+                       netif_wake_subqueue(dev, queue);
+       } else {
+               if (netif_msg_tx_err(priv) && net_ratelimit())
+                       printk(KERN_WARNING
+                              "%s: end_xmit: spurious interrupt\n", dev->name);
+-              if (netif_subqueue_stopped(dev, queue))
++              if (__netif_subqueue_stopped(dev, queue))
+                       netif_wake_subqueue(dev, queue);
+       }
+ }
+@@ -731,7 +731,6 @@ static void cpmac_clear_tx(struct net_de
+ static void cpmac_hw_error(struct work_struct *work)
+ {
+-      int i;
+       struct cpmac_priv *priv =
+               container_of(work, struct cpmac_priv, reset_work);
+@@ -818,7 +817,6 @@ static irqreturn_t cpmac_irq(int irq, vo
+ static void cpmac_tx_timeout(struct net_device *dev)
+ {
+-      int i;
+       struct cpmac_priv *priv = netdev_priv(dev);
+       spin_lock(&priv->lock);
+@@ -1097,7 +1095,7 @@ static int external_switch;
+ static int __devinit cpmac_probe(struct platform_device *pdev)
+ {
+-      int rc, phy_id, i;
++      int rc, phy_id;
+       char *mdio_bus_id = "0";
+       struct resource *mem;
+       struct cpmac_priv *priv;
+@@ -1125,6 +1123,7 @@ static int __devinit cpmac_probe(struct 
+       }
+       dev = alloc_etherdev_mq(sizeof(*priv), CPMAC_QUEUES);
++      //~ dev = alloc_etherdev(sizeof(*priv));
+       if (!dev) {
+               printk(KERN_ERR "cpmac: Unable to allocate net_device\n");