fixed typo in generic squashfs split patch ... shame on you :)
[openwrt/svn-archive/archive.git] / target / linux / generic-2.6 / patches-2.6.23 / 130-netfilter_ipset.patch
1 diff -Nru linux-2.6.23/include/linux/netfilter_ipv4/ip_set.h linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set.h
2 --- linux-2.6.23/include/linux/netfilter_ipv4/ip_set.h 1970-01-01 01:00:00.000000000 +0100
3 +++ linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set.h 2007-10-12 11:52:37.000000000 +0200
4 @@ -0,0 +1,498 @@
5 +#ifndef _IP_SET_H
6 +#define _IP_SET_H
7 +
8 +/* Copyright (C) 2000-2002 Joakim Axelsson <gozem@linux.nu>
9 + * Patrick Schaaf <bof@bof.de>
10 + * Martin Josefsson <gandalf@wlug.westbo.se>
11 + * Copyright (C) 2003-2004 Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
12 + *
13 + * This program is free software; you can redistribute it and/or modify
14 + * it under the terms of the GNU General Public License version 2 as
15 + * published by the Free Software Foundation.
16 + */
17 +
18 +#if 0
19 +#define IP_SET_DEBUG
20 +#endif
21 +
22 +/*
23 + * A sockopt of such quality has hardly ever been seen before on the open
24 + * market! This little beauty, hardly ever used: above 64, so it's
25 + * traditionally used for firewalling, not touched (even once!) by the
26 + * 2.0, 2.2 and 2.4 kernels!
27 + *
28 + * Comes with its own certificate of authenticity, valid anywhere in the
29 + * Free world!
30 + *
31 + * Rusty, 19.4.2000
32 + */
33 +#define SO_IP_SET 83
34 +
35 +/*
36 + * Heavily modify by Joakim Axelsson 08.03.2002
37 + * - Made it more modulebased
38 + *
39 + * Additional heavy modifications by Jozsef Kadlecsik 22.02.2004
40 + * - bindings added
41 + * - in order to "deal with" backward compatibility, renamed to ipset
42 + */
43 +
44 +/*
45 + * Used so that the kernel module and ipset-binary can match their versions
46 + */
47 +#define IP_SET_PROTOCOL_VERSION 2
48 +
49 +#define IP_SET_MAXNAMELEN 32 /* set names and set typenames */
50 +
51 +/* Lets work with our own typedef for representing an IP address.
52 + * We hope to make the code more portable, possibly to IPv6...
53 + *
54 + * The representation works in HOST byte order, because most set types
55 + * will perform arithmetic operations and compare operations.
56 + *
57 + * For now the type is an uint32_t.
58 + *
59 + * Make sure to ONLY use the functions when translating and parsing
60 + * in order to keep the host byte order and make it more portable:
61 + * parse_ip()
62 + * parse_mask()
63 + * parse_ipandmask()
64 + * ip_tostring()
65 + * (Joakim: where are they???)
66 + */
67 +
68 +typedef uint32_t ip_set_ip_t;
69 +
70 +/* Sets are identified by an id in kernel space. Tweak with ip_set_id_t
71 + * and IP_SET_INVALID_ID if you want to increase the max number of sets.
72 + */
73 +typedef uint16_t ip_set_id_t;
74 +
75 +#define IP_SET_INVALID_ID 65535
76 +
77 +/* How deep we follow bindings */
78 +#define IP_SET_MAX_BINDINGS 6
79 +
80 +/*
81 + * Option flags for kernel operations (ipt_set_info)
82 + */
83 +#define IPSET_SRC 0x01 /* Source match/add */
84 +#define IPSET_DST 0x02 /* Destination match/add */
85 +#define IPSET_MATCH_INV 0x04 /* Inverse matching */
86 +
87 +/*
88 + * Set features
89 + */
90 +#define IPSET_TYPE_IP 0x01 /* IP address type of set */
91 +#define IPSET_TYPE_PORT 0x02 /* Port type of set */
92 +#define IPSET_DATA_SINGLE 0x04 /* Single data storage */
93 +#define IPSET_DATA_DOUBLE 0x08 /* Double data storage */
94 +
95 +/* Reserved keywords */
96 +#define IPSET_TOKEN_DEFAULT ":default:"
97 +#define IPSET_TOKEN_ALL ":all:"
98 +
99 +/* SO_IP_SET operation constants, and their request struct types.
100 + *
101 + * Operation ids:
102 + * 0-99: commands with version checking
103 + * 100-199: add/del/test/bind/unbind
104 + * 200-299: list, save, restore
105 + */
106 +
107 +/* Single shot operations:
108 + * version, create, destroy, flush, rename and swap
109 + *
110 + * Sets are identified by name.
111 + */
112 +
113 +#define IP_SET_REQ_STD \
114 + unsigned op; \
115 + unsigned version; \
116 + char name[IP_SET_MAXNAMELEN]
117 +
118 +#define IP_SET_OP_CREATE 0x00000001 /* Create a new (empty) set */
119 +struct ip_set_req_create {
120 + IP_SET_REQ_STD;
121 + char typename[IP_SET_MAXNAMELEN];
122 +};
123 +
124 +#define IP_SET_OP_DESTROY 0x00000002 /* Remove a (empty) set */
125 +struct ip_set_req_std {
126 + IP_SET_REQ_STD;
127 +};
128 +
129 +#define IP_SET_OP_FLUSH 0x00000003 /* Remove all IPs in a set */
130 +/* Uses ip_set_req_std */
131 +
132 +#define IP_SET_OP_RENAME 0x00000004 /* Rename a set */
133 +/* Uses ip_set_req_create */
134 +
135 +#define IP_SET_OP_SWAP 0x00000005 /* Swap two sets */
136 +/* Uses ip_set_req_create */
137 +
138 +union ip_set_name_index {
139 + char name[IP_SET_MAXNAMELEN];
140 + ip_set_id_t index;
141 +};
142 +
143 +#define IP_SET_OP_GET_BYNAME 0x00000006 /* Get set index by name */
144 +struct ip_set_req_get_set {
145 + unsigned op;
146 + unsigned version;
147 + union ip_set_name_index set;
148 +};
149 +
150 +#define IP_SET_OP_GET_BYINDEX 0x00000007 /* Get set name by index */
151 +/* Uses ip_set_req_get_set */
152 +
153 +#define IP_SET_OP_VERSION 0x00000100 /* Ask kernel version */
154 +struct ip_set_req_version {
155 + unsigned op;
156 + unsigned version;
157 +};
158 +
159 +/* Double shots operations:
160 + * add, del, test, bind and unbind.
161 + *
162 + * First we query the kernel to get the index and type of the target set,
163 + * then issue the command. Validity of IP is checked in kernel in order
164 + * to minimalize sockopt operations.
165 + */
166 +
167 +/* Get minimal set data for add/del/test/bind/unbind IP */
168 +#define IP_SET_OP_ADT_GET 0x00000010 /* Get set and type */
169 +struct ip_set_req_adt_get {
170 + unsigned op;
171 + unsigned version;
172 + union ip_set_name_index set;
173 + char typename[IP_SET_MAXNAMELEN];
174 +};
175 +
176 +#define IP_SET_REQ_BYINDEX \
177 + unsigned op; \
178 + ip_set_id_t index;
179 +
180 +struct ip_set_req_adt {
181 + IP_SET_REQ_BYINDEX;
182 +};
183 +
184 +#define IP_SET_OP_ADD_IP 0x00000101 /* Add an IP to a set */
185 +/* Uses ip_set_req_adt, with type specific addage */
186 +
187 +#define IP_SET_OP_DEL_IP 0x00000102 /* Remove an IP from a set */
188 +/* Uses ip_set_req_adt, with type specific addage */
189 +
190 +#define IP_SET_OP_TEST_IP 0x00000103 /* Test an IP in a set */
191 +/* Uses ip_set_req_adt, with type specific addage */
192 +
193 +#define IP_SET_OP_BIND_SET 0x00000104 /* Bind an IP to a set */
194 +/* Uses ip_set_req_bind, with type specific addage */
195 +struct ip_set_req_bind {
196 + IP_SET_REQ_BYINDEX;
197 + char binding[IP_SET_MAXNAMELEN];
198 +};
199 +
200 +#define IP_SET_OP_UNBIND_SET 0x00000105 /* Unbind an IP from a set */
201 +/* Uses ip_set_req_bind, with type speficic addage
202 + * index = 0 means unbinding for all sets */
203 +
204 +#define IP_SET_OP_TEST_BIND_SET 0x00000106 /* Test binding an IP to a set */
205 +/* Uses ip_set_req_bind, with type specific addage */
206 +
207 +/* Multiple shots operations: list, save, restore.
208 + *
209 + * - check kernel version and query the max number of sets
210 + * - get the basic information on all sets
211 + * and size required for the next step
212 + * - get actual set data: header, data, bindings
213 + */
214 +
215 +/* Get max_sets and the index of a queried set
216 + */
217 +#define IP_SET_OP_MAX_SETS 0x00000020
218 +struct ip_set_req_max_sets {
219 + unsigned op;
220 + unsigned version;
221 + ip_set_id_t max_sets; /* max_sets */
222 + ip_set_id_t sets; /* real number of sets */
223 + union ip_set_name_index set; /* index of set if name used */
224 +};
225 +
226 +/* Get the id and name of the sets plus size for next step */
227 +#define IP_SET_OP_LIST_SIZE 0x00000201
228 +#define IP_SET_OP_SAVE_SIZE 0x00000202
229 +struct ip_set_req_setnames {
230 + unsigned op;
231 + ip_set_id_t index; /* set to list/save */
232 + size_t size; /* size to get setdata/bindings */
233 + /* followed by sets number of struct ip_set_name_list */
234 +};
235 +
236 +struct ip_set_name_list {
237 + char name[IP_SET_MAXNAMELEN];
238 + char typename[IP_SET_MAXNAMELEN];
239 + ip_set_id_t index;
240 + ip_set_id_t id;
241 +};
242 +
243 +/* The actual list operation */
244 +#define IP_SET_OP_LIST 0x00000203
245 +struct ip_set_req_list {
246 + IP_SET_REQ_BYINDEX;
247 + /* sets number of struct ip_set_list in reply */
248 +};
249 +
250 +struct ip_set_list {
251 + ip_set_id_t index;
252 + ip_set_id_t binding;
253 + u_int32_t ref;
254 + size_t header_size; /* Set header data of header_size */
255 + size_t members_size; /* Set members data of members_size */
256 + size_t bindings_size; /* Set bindings data of bindings_size */
257 +};
258 +
259 +struct ip_set_hash_list {
260 + ip_set_ip_t ip;
261 + ip_set_id_t binding;
262 +};
263 +
264 +/* The save operation */
265 +#define IP_SET_OP_SAVE 0x00000204
266 +/* Uses ip_set_req_list, in the reply replaced by
267 + * sets number of struct ip_set_save plus a marker
268 + * ip_set_save followed by ip_set_hash_save structures.
269 + */
270 +struct ip_set_save {
271 + ip_set_id_t index;
272 + ip_set_id_t binding;
273 + size_t header_size; /* Set header data of header_size */
274 + size_t members_size; /* Set members data of members_size */
275 +};
276 +
277 +/* At restoring, ip == 0 means default binding for the given set: */
278 +struct ip_set_hash_save {
279 + ip_set_ip_t ip;
280 + ip_set_id_t id;
281 + ip_set_id_t binding;
282 +};
283 +
284 +/* The restore operation */
285 +#define IP_SET_OP_RESTORE 0x00000205
286 +/* Uses ip_set_req_setnames followed by ip_set_restore structures
287 + * plus a marker ip_set_restore, followed by ip_set_hash_save
288 + * structures.
289 + */
290 +struct ip_set_restore {
291 + char name[IP_SET_MAXNAMELEN];
292 + char typename[IP_SET_MAXNAMELEN];
293 + ip_set_id_t index;
294 + size_t header_size; /* Create data of header_size */
295 + size_t members_size; /* Set members data of members_size */
296 +};
297 +
298 +static inline int bitmap_bytes(ip_set_ip_t a, ip_set_ip_t b)
299 +{
300 + return 4 * ((((b - a + 8) / 8) + 3) / 4);
301 +}
302 +
303 +#ifdef __KERNEL__
304 +
305 +#define ip_set_printk(format, args...) \
306 + do { \
307 + printk("%s: %s: ", __FILE__, __FUNCTION__); \
308 + printk(format "\n" , ## args); \
309 + } while (0)
310 +
311 +#if defined(IP_SET_DEBUG)
312 +#define DP(format, args...) \
313 + do { \
314 + printk("%s: %s (DBG): ", __FILE__, __FUNCTION__);\
315 + printk(format "\n" , ## args); \
316 + } while (0)
317 +#define IP_SET_ASSERT(x) \
318 + do { \
319 + if (!(x)) \
320 + printk("IP_SET_ASSERT: %s:%i(%s)\n", \
321 + __FILE__, __LINE__, __FUNCTION__); \
322 + } while (0)
323 +#else
324 +#define DP(format, args...)
325 +#define IP_SET_ASSERT(x)
326 +#endif
327 +
328 +struct ip_set;
329 +
330 +/*
331 + * The ip_set_type definition - one per set type, e.g. "ipmap".
332 + *
333 + * Each individual set has a pointer, set->type, going to one
334 + * of these structures. Function pointers inside the structure implement
335 + * the real behaviour of the sets.
336 + *
337 + * If not mentioned differently, the implementation behind the function
338 + * pointers of a set_type, is expected to return 0 if ok, and a negative
339 + * errno (e.g. -EINVAL) on error.
340 + */
341 +struct ip_set_type {
342 + struct list_head list; /* next in list of set types */
343 +
344 + /* test for IP in set (kernel: iptables -m set src|dst)
345 + * return 0 if not in set, 1 if in set.
346 + */
347 + int (*testip_kernel) (struct ip_set *set,
348 + const struct sk_buff * skb,
349 + ip_set_ip_t *ip,
350 + const u_int32_t *flags,
351 + unsigned char index);
352 +
353 + /* test for IP in set (userspace: ipset -T set IP)
354 + * return 0 if not in set, 1 if in set.
355 + */
356 + int (*testip) (struct ip_set *set,
357 + const void *data, size_t size,
358 + ip_set_ip_t *ip);
359 +
360 + /*
361 + * Size of the data structure passed by when
362 + * adding/deletin/testing an entry.
363 + */
364 + size_t reqsize;
365 +
366 + /* Add IP into set (userspace: ipset -A set IP)
367 + * Return -EEXIST if the address is already in the set,
368 + * and -ERANGE if the address lies outside the set bounds.
369 + * If the address was not already in the set, 0 is returned.
370 + */
371 + int (*addip) (struct ip_set *set,
372 + const void *data, size_t size,
373 + ip_set_ip_t *ip);
374 +
375 + /* Add IP into set (kernel: iptables ... -j SET set src|dst)
376 + * Return -EEXIST if the address is already in the set,
377 + * and -ERANGE if the address lies outside the set bounds.
378 + * If the address was not already in the set, 0 is returned.
379 + */
380 + int (*addip_kernel) (struct ip_set *set,
381 + const struct sk_buff * skb,
382 + ip_set_ip_t *ip,
383 + const u_int32_t *flags,
384 + unsigned char index);
385 +
386 + /* remove IP from set (userspace: ipset -D set --entry x)
387 + * Return -EEXIST if the address is NOT in the set,
388 + * and -ERANGE if the address lies outside the set bounds.
389 + * If the address really was in the set, 0 is returned.
390 + */
391 + int (*delip) (struct ip_set *set,
392 + const void *data, size_t size,
393 + ip_set_ip_t *ip);
394 +
395 + /* remove IP from set (kernel: iptables ... -j SET --entry x)
396 + * Return -EEXIST if the address is NOT in the set,
397 + * and -ERANGE if the address lies outside the set bounds.
398 + * If the address really was in the set, 0 is returned.
399 + */
400 + int (*delip_kernel) (struct ip_set *set,
401 + const struct sk_buff * skb,
402 + ip_set_ip_t *ip,
403 + const u_int32_t *flags,
404 + unsigned char index);
405 +
406 + /* new set creation - allocated type specific items
407 + */
408 + int (*create) (struct ip_set *set,
409 + const void *data, size_t size);
410 +
411 + /* retry the operation after successfully tweaking the set
412 + */
413 + int (*retry) (struct ip_set *set);
414 +
415 + /* set destruction - free type specific items
416 + * There is no return value.
417 + * Can be called only when child sets are destroyed.
418 + */
419 + void (*destroy) (struct ip_set *set);
420 +
421 + /* set flushing - reset all bits in the set, or something similar.
422 + * There is no return value.
423 + */
424 + void (*flush) (struct ip_set *set);
425 +
426 + /* Listing: size needed for header
427 + */
428 + size_t header_size;
429 +
430 + /* Listing: Get the header
431 + *
432 + * Fill in the information in "data".
433 + * This function is always run after list_header_size() under a
434 + * writelock on the set. Therefor is the length of "data" always
435 + * correct.
436 + */
437 + void (*list_header) (const struct ip_set *set,
438 + void *data);
439 +
440 + /* Listing: Get the size for the set members
441 + */
442 + int (*list_members_size) (const struct ip_set *set);
443 +
444 + /* Listing: Get the set members
445 + *
446 + * Fill in the information in "data".
447 + * This function is always run after list_member_size() under a
448 + * writelock on the set. Therefor is the length of "data" always
449 + * correct.
450 + */
451 + void (*list_members) (const struct ip_set *set,
452 + void *data);
453 +
454 + char typename[IP_SET_MAXNAMELEN];
455 + unsigned char features;
456 + int protocol_version;
457 +
458 + /* Set this to THIS_MODULE if you are a module, otherwise NULL */
459 + struct module *me;
460 +};
461 +
462 +extern int ip_set_register_set_type(struct ip_set_type *set_type);
463 +extern void ip_set_unregister_set_type(struct ip_set_type *set_type);
464 +
465 +/* A generic ipset */
466 +struct ip_set {
467 + char name[IP_SET_MAXNAMELEN]; /* the name of the set */
468 + rwlock_t lock; /* lock for concurrency control */
469 + ip_set_id_t id; /* set id for swapping */
470 + ip_set_id_t binding; /* default binding for the set */
471 + atomic_t ref; /* in kernel and in hash references */
472 + struct ip_set_type *type; /* the set types */
473 + void *data; /* pooltype specific data */
474 +};
475 +
476 +/* Structure to bind set elements to sets */
477 +struct ip_set_hash {
478 + struct list_head list; /* list of clashing entries in hash */
479 + ip_set_ip_t ip; /* ip from set */
480 + ip_set_id_t id; /* set id */
481 + ip_set_id_t binding; /* set we bind the element to */
482 +};
483 +
484 +/* register and unregister set references */
485 +extern ip_set_id_t ip_set_get_byname(const char name[IP_SET_MAXNAMELEN]);
486 +extern ip_set_id_t ip_set_get_byindex(ip_set_id_t id);
487 +extern void ip_set_put(ip_set_id_t id);
488 +
489 +/* API for iptables set match, and SET target */
490 +extern void ip_set_addip_kernel(ip_set_id_t id,
491 + const struct sk_buff *skb,
492 + const u_int32_t *flags);
493 +extern void ip_set_delip_kernel(ip_set_id_t id,
494 + const struct sk_buff *skb,
495 + const u_int32_t *flags);
496 +extern int ip_set_testip_kernel(ip_set_id_t id,
497 + const struct sk_buff *skb,
498 + const u_int32_t *flags);
499 +
500 +#endif /* __KERNEL__ */
501 +
502 +#endif /*_IP_SET_H*/
503 diff -Nru linux-2.6.23/include/linux/netfilter_ipv4/ip_set_iphash.h linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_iphash.h
504 --- linux-2.6.23/include/linux/netfilter_ipv4/ip_set_iphash.h 1970-01-01 01:00:00.000000000 +0100
505 +++ linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_iphash.h 2007-10-12 11:52:37.000000000 +0200
506 @@ -0,0 +1,30 @@
507 +#ifndef __IP_SET_IPHASH_H
508 +#define __IP_SET_IPHASH_H
509 +
510 +#include <linux/netfilter_ipv4/ip_set.h>
511 +
512 +#define SETTYPE_NAME "iphash"
513 +#define MAX_RANGE 0x0000FFFF
514 +
515 +struct ip_set_iphash {
516 + ip_set_ip_t *members; /* the iphash proper */
517 + uint32_t elements; /* number of elements */
518 + uint32_t hashsize; /* hash size */
519 + uint16_t probes; /* max number of probes */
520 + uint16_t resize; /* resize factor in percent */
521 + ip_set_ip_t netmask; /* netmask */
522 + void *initval[0]; /* initvals for jhash_1word */
523 +};
524 +
525 +struct ip_set_req_iphash_create {
526 + uint32_t hashsize;
527 + uint16_t probes;
528 + uint16_t resize;
529 + ip_set_ip_t netmask;
530 +};
531 +
532 +struct ip_set_req_iphash {
533 + ip_set_ip_t ip;
534 +};
535 +
536 +#endif /* __IP_SET_IPHASH_H */
537 diff -Nru linux-2.6.23/include/linux/netfilter_ipv4/ip_set_ipmap.h linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_ipmap.h
538 --- linux-2.6.23/include/linux/netfilter_ipv4/ip_set_ipmap.h 1970-01-01 01:00:00.000000000 +0100
539 +++ linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_ipmap.h 2007-10-12 11:52:37.000000000 +0200
540 @@ -0,0 +1,56 @@
541 +#ifndef __IP_SET_IPMAP_H
542 +#define __IP_SET_IPMAP_H
543 +
544 +#include <linux/netfilter_ipv4/ip_set.h>
545 +
546 +#define SETTYPE_NAME "ipmap"
547 +#define MAX_RANGE 0x0000FFFF
548 +
549 +struct ip_set_ipmap {
550 + void *members; /* the ipmap proper */
551 + ip_set_ip_t first_ip; /* host byte order, included in range */
552 + ip_set_ip_t last_ip; /* host byte order, included in range */
553 + ip_set_ip_t netmask; /* subnet netmask */
554 + ip_set_ip_t sizeid; /* size of set in IPs */
555 + ip_set_ip_t hosts; /* number of hosts in a subnet */
556 +};
557 +
558 +struct ip_set_req_ipmap_create {
559 + ip_set_ip_t from;
560 + ip_set_ip_t to;
561 + ip_set_ip_t netmask;
562 +};
563 +
564 +struct ip_set_req_ipmap {
565 + ip_set_ip_t ip;
566 +};
567 +
568 +unsigned int
569 +mask_to_bits(ip_set_ip_t mask)
570 +{
571 + unsigned int bits = 32;
572 + ip_set_ip_t maskaddr;
573 +
574 + if (mask == 0xFFFFFFFF)
575 + return bits;
576 +
577 + maskaddr = 0xFFFFFFFE;
578 + while (--bits >= 0 && maskaddr != mask)
579 + maskaddr <<= 1;
580 +
581 + return bits;
582 +}
583 +
584 +ip_set_ip_t
585 +range_to_mask(ip_set_ip_t from, ip_set_ip_t to, unsigned int *bits)
586 +{
587 + ip_set_ip_t mask = 0xFFFFFFFE;
588 +
589 + *bits = 32;
590 + while (--(*bits) >= 0 && mask && (to & mask) != from)
591 + mask <<= 1;
592 +
593 + return mask;
594 +}
595 +
596 +#endif /* __IP_SET_IPMAP_H */
597 diff -Nru linux-2.6.23/include/linux/netfilter_ipv4/ip_set_ipporthash.h linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_ipporthash.h
598 --- linux-2.6.23/include/linux/netfilter_ipv4/ip_set_ipporthash.h 1970-01-01 01:00:00.000000000 +0100
599 +++ linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_ipporthash.h 2007-10-12 11:52:37.000000000 +0200
600 @@ -0,0 +1,34 @@
601 +#ifndef __IP_SET_IPPORTHASH_H
602 +#define __IP_SET_IPPORTHASH_H
603 +
604 +#include <linux/netfilter_ipv4/ip_set.h>
605 +
606 +#define SETTYPE_NAME "ipporthash"
607 +#define MAX_RANGE 0x0000FFFF
608 +#define INVALID_PORT (MAX_RANGE + 1)
609 +
610 +struct ip_set_ipporthash {
611 + ip_set_ip_t *members; /* the ipporthash proper */
612 + uint32_t elements; /* number of elements */
613 + uint32_t hashsize; /* hash size */
614 + uint16_t probes; /* max number of probes */
615 + uint16_t resize; /* resize factor in percent */
616 + ip_set_ip_t first_ip; /* host byte order, included in range */
617 + ip_set_ip_t last_ip; /* host byte order, included in range */
618 + void *initval[0]; /* initvals for jhash_1word */
619 +};
620 +
621 +struct ip_set_req_ipporthash_create {
622 + uint32_t hashsize;
623 + uint16_t probes;
624 + uint16_t resize;
625 + ip_set_ip_t from;
626 + ip_set_ip_t to;
627 +};
628 +
629 +struct ip_set_req_ipporthash {
630 + ip_set_ip_t ip;
631 + ip_set_ip_t port;
632 +};
633 +
634 +#endif /* __IP_SET_IPPORTHASH_H */
635 diff -Nru linux-2.6.23/include/linux/netfilter_ipv4/ip_set_iptree.h linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_iptree.h
636 --- linux-2.6.23/include/linux/netfilter_ipv4/ip_set_iptree.h 1970-01-01 01:00:00.000000000 +0100
637 +++ linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_iptree.h 2007-10-12 11:52:37.000000000 +0200
638 @@ -0,0 +1,40 @@
639 +#ifndef __IP_SET_IPTREE_H
640 +#define __IP_SET_IPTREE_H
641 +
642 +#include <linux/netfilter_ipv4/ip_set.h>
643 +
644 +#define SETTYPE_NAME "iptree"
645 +#define MAX_RANGE 0x0000FFFF
646 +
647 +struct ip_set_iptreed {
648 + unsigned long expires[256]; /* x.x.x.ADDR */
649 +};
650 +
651 +struct ip_set_iptreec {
652 + struct ip_set_iptreed *tree[256]; /* x.x.ADDR.* */
653 +};
654 +
655 +struct ip_set_iptreeb {
656 + struct ip_set_iptreec *tree[256]; /* x.ADDR.*.* */
657 +};
658 +
659 +struct ip_set_iptree {
660 + unsigned int timeout;
661 + unsigned int gc_interval;
662 +#ifdef __KERNEL__
663 + uint32_t elements; /* number of elements */
664 + struct timer_list gc;
665 + struct ip_set_iptreeb *tree[256]; /* ADDR.*.*.* */
666 +#endif
667 +};
668 +
669 +struct ip_set_req_iptree_create {
670 + unsigned int timeout;
671 +};
672 +
673 +struct ip_set_req_iptree {
674 + ip_set_ip_t ip;
675 + unsigned int timeout;
676 +};
677 +
678 +#endif /* __IP_SET_IPTREE_H */
679 diff -Nru linux-2.6.23/include/linux/netfilter_ipv4/ip_set_iptreemap.h linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_iptreemap.h
680 --- linux-2.6.23/include/linux/netfilter_ipv4/ip_set_iptreemap.h 1970-01-01 01:00:00.000000000 +0100
681 +++ linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_iptreemap.h 2007-10-12 11:52:37.000000000 +0200
682 @@ -0,0 +1,40 @@
683 +#ifndef __IP_SET_IPTREEMAP_H
684 +#define __IP_SET_IPTREEMAP_H
685 +
686 +#include <linux/netfilter_ipv4/ip_set.h>
687 +
688 +#define SETTYPE_NAME "iptreemap"
689 +
690 +#ifdef __KERNEL__
691 +struct ip_set_iptreemap_d {
692 + unsigned char bitmap[32]; /* x.x.x.y */
693 +};
694 +
695 +struct ip_set_iptreemap_c {
696 + struct ip_set_iptreemap_d *tree[256]; /* x.x.y.x */
697 +};
698 +
699 +struct ip_set_iptreemap_b {
700 + struct ip_set_iptreemap_c *tree[256]; /* x.y.x.x */
701 + unsigned char dirty[32];
702 +};
703 +#endif
704 +
705 +struct ip_set_iptreemap {
706 + unsigned int gc_interval;
707 +#ifdef __KERNEL__
708 + struct timer_list gc;
709 + struct ip_set_iptreemap_b *tree[256]; /* y.x.x.x */
710 +#endif
711 +};
712 +
713 +struct ip_set_req_iptreemap_create {
714 + unsigned int gc_interval;
715 +};
716 +
717 +struct ip_set_req_iptreemap {
718 + ip_set_ip_t start;
719 + ip_set_ip_t end;
720 +};
721 +
722 +#endif /* __IP_SET_IPTREEMAP_H */
723 diff -Nru linux-2.6.23/include/linux/netfilter_ipv4/ip_set_jhash.h linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_jhash.h
724 --- linux-2.6.23/include/linux/netfilter_ipv4/ip_set_jhash.h 1970-01-01 01:00:00.000000000 +0100
725 +++ linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_jhash.h 2007-10-12 11:52:37.000000000 +0200
726 @@ -0,0 +1,148 @@
727 +#ifndef _LINUX_IPSET_JHASH_H
728 +#define _LINUX_IPSET_JHASH_H
729 +
730 +/* This is a copy of linux/jhash.h but the types u32/u8 are changed
731 + * to __u32/__u8 so that the header file can be included into
732 + * userspace code as well. Jozsef Kadlecsik (kadlec@blackhole.kfki.hu)
733 + */
734 +
735 +/* jhash.h: Jenkins hash support.
736 + *
737 + * Copyright (C) 1996 Bob Jenkins (bob_jenkins@burtleburtle.net)
738 + *
739 + * http://burtleburtle.net/bob/hash/
740 + *
741 + * These are the credits from Bob's sources:
742 + *
743 + * lookup2.c, by Bob Jenkins, December 1996, Public Domain.
744 + * hash(), hash2(), hash3, and mix() are externally useful functions.
745 + * Routines to test the hash are included if SELF_TEST is defined.
746 + * You can use this free for any purpose. It has no warranty.
747 + *
748 + * Copyright (C) 2003 David S. Miller (davem@redhat.com)
749 + *
750 + * I've modified Bob's hash to be useful in the Linux kernel, and
751 + * any bugs present are surely my fault. -DaveM
752 + */
753 +
754 +/* NOTE: Arguments are modified. */
755 +#define __jhash_mix(a, b, c) \
756 +{ \
757 + a -= b; a -= c; a ^= (c>>13); \
758 + b -= c; b -= a; b ^= (a<<8); \
759 + c -= a; c -= b; c ^= (b>>13); \
760 + a -= b; a -= c; a ^= (c>>12); \
761 + b -= c; b -= a; b ^= (a<<16); \
762 + c -= a; c -= b; c ^= (b>>5); \
763 + a -= b; a -= c; a ^= (c>>3); \
764 + b -= c; b -= a; b ^= (a<<10); \
765 + c -= a; c -= b; c ^= (b>>15); \
766 +}
767 +
768 +/* The golden ration: an arbitrary value */
769 +#define JHASH_GOLDEN_RATIO 0x9e3779b9
770 +
771 +/* The most generic version, hashes an arbitrary sequence
772 + * of bytes. No alignment or length assumptions are made about
773 + * the input key.
774 + */
775 +static inline __u32 jhash(void *key, __u32 length, __u32 initval)
776 +{
777 + __u32 a, b, c, len;
778 + __u8 *k = key;
779 +
780 + len = length;
781 + a = b = JHASH_GOLDEN_RATIO;
782 + c = initval;
783 +
784 + while (len >= 12) {
785 + a += (k[0] +((__u32)k[1]<<8) +((__u32)k[2]<<16) +((__u32)k[3]<<24));
786 + b += (k[4] +((__u32)k[5]<<8) +((__u32)k[6]<<16) +((__u32)k[7]<<24));
787 + c += (k[8] +((__u32)k[9]<<8) +((__u32)k[10]<<16)+((__u32)k[11]<<24));
788 +
789 + __jhash_mix(a,b,c);
790 +
791 + k += 12;
792 + len -= 12;
793 + }
794 +
795 + c += length;
796 + switch (len) {
797 + case 11: c += ((__u32)k[10]<<24);
798 + case 10: c += ((__u32)k[9]<<16);
799 + case 9 : c += ((__u32)k[8]<<8);
800 + case 8 : b += ((__u32)k[7]<<24);
801 + case 7 : b += ((__u32)k[6]<<16);
802 + case 6 : b += ((__u32)k[5]<<8);
803 + case 5 : b += k[4];
804 + case 4 : a += ((__u32)k[3]<<24);
805 + case 3 : a += ((__u32)k[2]<<16);
806 + case 2 : a += ((__u32)k[1]<<8);
807 + case 1 : a += k[0];
808 + };
809 +
810 + __jhash_mix(a,b,c);
811 +
812 + return c;
813 +}
814 +
815 +/* A special optimized version that handles 1 or more of __u32s.
816 + * The length parameter here is the number of __u32s in the key.
817 + */
818 +static inline __u32 jhash2(__u32 *k, __u32 length, __u32 initval)
819 +{
820 + __u32 a, b, c, len;
821 +
822 + a = b = JHASH_GOLDEN_RATIO;
823 + c = initval;
824 + len = length;
825 +
826 + while (len >= 3) {
827 + a += k[0];
828 + b += k[1];
829 + c += k[2];
830 + __jhash_mix(a, b, c);
831 + k += 3; len -= 3;
832 + }
833 +
834 + c += length * 4;
835 +
836 + switch (len) {
837 + case 2 : b += k[1];
838 + case 1 : a += k[0];
839 + };
840 +
841 + __jhash_mix(a,b,c);
842 +
843 + return c;
844 +}
845 +
846 +
847 +/* A special ultra-optimized versions that knows they are hashing exactly
848 + * 3, 2 or 1 word(s).
849 + *
850 + * NOTE: In partilar the "c += length; __jhash_mix(a,b,c);" normally
851 + * done at the end is not done here.
852 + */
853 +static inline __u32 jhash_3words(__u32 a, __u32 b, __u32 c, __u32 initval)
854 +{
855 + a += JHASH_GOLDEN_RATIO;
856 + b += JHASH_GOLDEN_RATIO;
857 + c += initval;
858 +
859 + __jhash_mix(a, b, c);
860 +
861 + return c;
862 +}
863 +
864 +static inline __u32 jhash_2words(__u32 a, __u32 b, __u32 initval)
865 +{
866 + return jhash_3words(a, b, 0, initval);
867 +}
868 +
869 +static inline __u32 jhash_1word(__u32 a, __u32 initval)
870 +{
871 + return jhash_3words(a, 0, 0, initval);
872 +}
873 +
874 +#endif /* _LINUX_IPSET_JHASH_H */
875 diff -Nru linux-2.6.23/include/linux/netfilter_ipv4/ip_set_macipmap.h linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_macipmap.h
876 --- linux-2.6.23/include/linux/netfilter_ipv4/ip_set_macipmap.h 1970-01-01 01:00:00.000000000 +0100
877 +++ linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_macipmap.h 2007-10-12 11:52:37.000000000 +0200
878 @@ -0,0 +1,38 @@
879 +#ifndef __IP_SET_MACIPMAP_H
880 +#define __IP_SET_MACIPMAP_H
881 +
882 +#include <linux/netfilter_ipv4/ip_set.h>
883 +
884 +#define SETTYPE_NAME "macipmap"
885 +#define MAX_RANGE 0x0000FFFF
886 +
887 +/* general flags */
888 +#define IPSET_MACIP_MATCHUNSET 1
889 +
890 +/* per ip flags */
891 +#define IPSET_MACIP_ISSET 1
892 +
893 +struct ip_set_macipmap {
894 + void *members; /* the macipmap proper */
895 + ip_set_ip_t first_ip; /* host byte order, included in range */
896 + ip_set_ip_t last_ip; /* host byte order, included in range */
897 + u_int32_t flags;
898 +};
899 +
900 +struct ip_set_req_macipmap_create {
901 + ip_set_ip_t from;
902 + ip_set_ip_t to;
903 + u_int32_t flags;
904 +};
905 +
906 +struct ip_set_req_macipmap {
907 + ip_set_ip_t ip;
908 + unsigned char ethernet[ETH_ALEN];
909 +};
910 +
911 +struct ip_set_macip {
912 + unsigned short flags;
913 + unsigned char ethernet[ETH_ALEN];
914 +};
915 +
916 +#endif /* __IP_SET_MACIPMAP_H */
917 diff -Nru linux-2.6.23/include/linux/netfilter_ipv4/ip_set_malloc.h linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_malloc.h
918 --- linux-2.6.23/include/linux/netfilter_ipv4/ip_set_malloc.h 1970-01-01 01:00:00.000000000 +0100
919 +++ linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_malloc.h 2007-10-12 11:52:37.000000000 +0200
920 @@ -0,0 +1,116 @@
921 +#ifndef _IP_SET_MALLOC_H
922 +#define _IP_SET_MALLOC_H
923 +
924 +#ifdef __KERNEL__
925 +
926 +/* Memory allocation and deallocation */
927 +static size_t max_malloc_size = 0;
928 +
929 +static inline void init_max_malloc_size(void)
930 +{
931 +#define CACHE(x) max_malloc_size = x;
932 +#include <linux/kmalloc_sizes.h>
933 +#undef CACHE
934 +}
935 +
936 +static inline void * ip_set_malloc(size_t bytes)
937 +{
938 + if (bytes > max_malloc_size)
939 + return vmalloc(bytes);
940 + else
941 + return kmalloc(bytes, GFP_KERNEL);
942 +}
943 +
944 +static inline void ip_set_free(void * data, size_t bytes)
945 +{
946 + if (bytes > max_malloc_size)
947 + vfree(data);
948 + else
949 + kfree(data);
950 +}
951 +
952 +struct harray {
953 + size_t max_elements;
954 + void *arrays[0];
955 +};
956 +
957 +static inline void *
958 +harray_malloc(size_t hashsize, size_t typesize, int flags)
959 +{
960 + struct harray *harray;
961 + size_t max_elements, size, i, j;
962 +
963 + if (!max_malloc_size)
964 + init_max_malloc_size();
965 +
966 + if (typesize > max_malloc_size)
967 + return NULL;
968 +
969 + max_elements = max_malloc_size/typesize;
970 + size = hashsize/max_elements;
971 + if (hashsize % max_elements)
972 + size++;
973 +
974 + /* Last pointer signals end of arrays */
975 + harray = kmalloc(sizeof(struct harray) + (size + 1) * sizeof(void *),
976 + flags);
977 +
978 + if (!harray)
979 + return NULL;
980 +
981 + for (i = 0; i < size - 1; i++) {
982 + harray->arrays[i] = kmalloc(max_elements * typesize, flags);
983 + if (!harray->arrays[i])
984 + goto undo;
985 + memset(harray->arrays[i], 0, max_elements * typesize);
986 + }
987 + harray->arrays[i] = kmalloc((hashsize - i * max_elements) * typesize,
988 + flags);
989 + if (!harray->arrays[i])
990 + goto undo;
991 + memset(harray->arrays[i], 0, (hashsize - i * max_elements) * typesize);
992 +
993 + harray->max_elements = max_elements;
994 + harray->arrays[size] = NULL;
995 +
996 + return (void *)harray;
997 +
998 + undo:
999 + for (j = 0; j < i; j++) {
1000 + kfree(harray->arrays[j]);
1001 + }
1002 + kfree(harray);
1003 + return NULL;
1004 +}
1005 +
1006 +static inline void harray_free(void *h)
1007 +{
1008 + struct harray *harray = (struct harray *) h;
1009 + size_t i;
1010 +
1011 + for (i = 0; harray->arrays[i] != NULL; i++)
1012 + kfree(harray->arrays[i]);
1013 + kfree(harray);
1014 +}
1015 +
1016 +static inline void harray_flush(void *h, size_t hashsize, size_t typesize)
1017 +{
1018 + struct harray *harray = (struct harray *) h;
1019 + size_t i;
1020 +
1021 + for (i = 0; harray->arrays[i+1] != NULL; i++)
1022 + memset(harray->arrays[i], 0, harray->max_elements * typesize);
1023 + memset(harray->arrays[i], 0,
1024 + (hashsize - i * harray->max_elements) * typesize);
1025 +}
1026 +
1027 +#define HARRAY_ELEM(h, type, which) \
1028 +({ \
1029 + struct harray *__h = (struct harray *)(h); \
1030 + ((type)((__h)->arrays[(which)/(__h)->max_elements]) \
1031 + + (which)%(__h)->max_elements); \
1032 +})
1033 +
1034 +#endif /* __KERNEL__ */
1035 +
1036 +#endif /*_IP_SET_MALLOC_H*/
1037 diff -Nru linux-2.6.23/include/linux/netfilter_ipv4/ip_set_nethash.h linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_nethash.h
1038 --- linux-2.6.23/include/linux/netfilter_ipv4/ip_set_nethash.h 1970-01-01 01:00:00.000000000 +0100
1039 +++ linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_nethash.h 2007-10-12 11:52:37.000000000 +0200
1040 @@ -0,0 +1,55 @@
1041 +#ifndef __IP_SET_NETHASH_H
1042 +#define __IP_SET_NETHASH_H
1043 +
1044 +#include <linux/netfilter_ipv4/ip_set.h>
1045 +
1046 +#define SETTYPE_NAME "nethash"
1047 +#define MAX_RANGE 0x0000FFFF
1048 +
1049 +struct ip_set_nethash {
1050 + ip_set_ip_t *members; /* the nethash proper */
1051 + uint32_t elements; /* number of elements */
1052 + uint32_t hashsize; /* hash size */
1053 + uint16_t probes; /* max number of probes */
1054 + uint16_t resize; /* resize factor in percent */
1055 + unsigned char cidr[30]; /* CIDR sizes */
1056 + void *initval[0]; /* initvals for jhash_1word */
1057 +};
1058 +
1059 +struct ip_set_req_nethash_create {
1060 + uint32_t hashsize;
1061 + uint16_t probes;
1062 + uint16_t resize;
1063 +};
1064 +
1065 +struct ip_set_req_nethash {
1066 + ip_set_ip_t ip;
1067 + unsigned char cidr;
1068 +};
1069 +
1070 +static unsigned char shifts[] = {255, 253, 249, 241, 225, 193, 129, 1};
1071 +
1072 +static inline ip_set_ip_t
1073 +pack(ip_set_ip_t ip, unsigned char cidr)
1074 +{
1075 + ip_set_ip_t addr, *paddr = &addr;
1076 + unsigned char n, t, *a;
1077 +
1078 + addr = htonl(ip & (0xFFFFFFFF << (32 - (cidr))));
1079 +#ifdef __KERNEL__
1080 + DP("ip:%u.%u.%u.%u/%u", NIPQUAD(addr), cidr);
1081 +#endif
1082 + n = cidr / 8;
1083 + t = cidr % 8;
1084 + a = &((unsigned char *)paddr)[n];
1085 + *a = *a /(1 << (8 - t)) + shifts[t];
1086 +#ifdef __KERNEL__
1087 + DP("n: %u, t: %u, a: %u", n, t, *a);
1088 + DP("ip:%u.%u.%u.%u/%u, %u.%u.%u.%u",
1089 + HIPQUAD(ip), cidr, NIPQUAD(addr));
1090 +#endif
1091 +
1092 + return ntohl(addr);
1093 +}
1094 +
1095 +#endif /* __IP_SET_NETHASH_H */
1096 diff -Nru linux-2.6.23/include/linux/netfilter_ipv4/ip_set_portmap.h linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_portmap.h
1097 --- linux-2.6.23/include/linux/netfilter_ipv4/ip_set_portmap.h 1970-01-01 01:00:00.000000000 +0100
1098 +++ linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ip_set_portmap.h 2007-10-12 11:52:37.000000000 +0200
1099 @@ -0,0 +1,25 @@
1100 +#ifndef __IP_SET_PORTMAP_H
1101 +#define __IP_SET_PORTMAP_H
1102 +
1103 +#include <linux/netfilter_ipv4/ip_set.h>
1104 +
1105 +#define SETTYPE_NAME "portmap"
1106 +#define MAX_RANGE 0x0000FFFF
1107 +#define INVALID_PORT (MAX_RANGE + 1)
1108 +
1109 +struct ip_set_portmap {
1110 + void *members; /* the portmap proper */
1111 + ip_set_ip_t first_port; /* host byte order, included in range */
1112 + ip_set_ip_t last_port; /* host byte order, included in range */
1113 +};
1114 +
1115 +struct ip_set_req_portmap_create {
1116 + ip_set_ip_t from;
1117 + ip_set_ip_t to;
1118 +};
1119 +
1120 +struct ip_set_req_portmap {
1121 + ip_set_ip_t port;
1122 +};
1123 +
1124 +#endif /* __IP_SET_PORTMAP_H */
1125 diff -Nru linux-2.6.23/include/linux/netfilter_ipv4/ipt_set.h linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ipt_set.h
1126 --- linux-2.6.23/include/linux/netfilter_ipv4/ipt_set.h 1970-01-01 01:00:00.000000000 +0100
1127 +++ linux-2.6.23.pom2patch.set/include/linux/netfilter_ipv4/ipt_set.h 2007-10-12 11:52:37.000000000 +0200
1128 @@ -0,0 +1,21 @@
1129 +#ifndef _IPT_SET_H
1130 +#define _IPT_SET_H
1131 +
1132 +#include <linux/netfilter_ipv4/ip_set.h>
1133 +
1134 +struct ipt_set_info {
1135 + ip_set_id_t index;
1136 + u_int32_t flags[IP_SET_MAX_BINDINGS + 1];
1137 +};
1138 +
1139 +/* match info */
1140 +struct ipt_set_info_match {
1141 + struct ipt_set_info match_set;
1142 +};
1143 +
1144 +struct ipt_set_info_target {
1145 + struct ipt_set_info add_set;
1146 + struct ipt_set_info del_set;
1147 +};
1148 +
1149 +#endif /*_IPT_SET_H*/
1150 diff -Nru linux-2.6.23/net/ipv4/netfilter/ip_set.c linux-2.6.23.pom2patch.set/net/ipv4/netfilter/ip_set.c
1151 --- linux-2.6.23/net/ipv4/netfilter/ip_set.c 1970-01-01 01:00:00.000000000 +0100
1152 +++ linux-2.6.23.pom2patch.set/net/ipv4/netfilter/ip_set.c 2007-10-12 11:52:37.000000000 +0200
1153 @@ -0,0 +1,2003 @@
1154 +/* Copyright (C) 2000-2002 Joakim Axelsson <gozem@linux.nu>
1155 + * Patrick Schaaf <bof@bof.de>
1156 + * Copyright (C) 2003-2004 Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
1157 + *
1158 + * This program is free software; you can redistribute it and/or modify
1159 + * it under the terms of the GNU General Public License version 2 as
1160 + * published by the Free Software Foundation.
1161 + */
1162 +
1163 +/* Kernel module for IP set management */
1164 +
1165 +#include <linux/version.h>
1166 +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19)
1167 +#include <linux/config.h>
1168 +#endif
1169 +#include <linux/module.h>
1170 +#include <linux/moduleparam.h>
1171 +#include <linux/kmod.h>
1172 +#include <linux/ip.h>
1173 +#include <linux/skbuff.h>
1174 +#include <linux/random.h>
1175 +#include <linux/jhash.h>
1176 +#include <linux/netfilter_ipv4/ip_tables.h>
1177 +#include <linux/errno.h>
1178 +#include <asm/uaccess.h>
1179 +#include <asm/bitops.h>
1180 +#include <asm/semaphore.h>
1181 +#include <linux/spinlock.h>
1182 +#include <linux/vmalloc.h>
1183 +
1184 +#define ASSERT_READ_LOCK(x)
1185 +#define ASSERT_WRITE_LOCK(x)
1186 +#include <linux/netfilter_ipv4/ip_set.h>
1187 +
1188 +static struct list_head set_type_list; /* all registered sets */
1189 +static struct ip_set **ip_set_list; /* all individual sets */
1190 +static DEFINE_RWLOCK(ip_set_lock); /* protects the lists and the hash */
1191 +static DECLARE_MUTEX(ip_set_app_mutex); /* serializes user access */
1192 +static ip_set_id_t ip_set_max = CONFIG_IP_NF_SET_MAX;
1193 +static ip_set_id_t ip_set_bindings_hash_size = CONFIG_IP_NF_SET_HASHSIZE;
1194 +static struct list_head *ip_set_hash; /* hash of bindings */
1195 +static unsigned int ip_set_hash_random; /* random seed */
1196 +
1197 +/*
1198 + * Sets are identified either by the index in ip_set_list or by id.
1199 + * The id never changes and is used to find a key in the hash.
1200 + * The index may change by swapping and used at all other places
1201 + * (set/SET netfilter modules, binding value, etc.)
1202 + *
1203 + * Userspace requests are serialized by ip_set_mutex and sets can
1204 + * be deleted only from userspace. Therefore ip_set_list locking
1205 + * must obey the following rules:
1206 + *
1207 + * - kernel requests: read and write locking mandatory
1208 + * - user requests: read locking optional, write locking mandatory
1209 + */
1210 +
1211 +static inline void
1212 +__ip_set_get(ip_set_id_t index)
1213 +{
1214 + atomic_inc(&ip_set_list[index]->ref);
1215 +}
1216 +
1217 +static inline void
1218 +__ip_set_put(ip_set_id_t index)
1219 +{
1220 + atomic_dec(&ip_set_list[index]->ref);
1221 +}
1222 +
1223 +/*
1224 + * Binding routines
1225 + */
1226 +
1227 +static inline struct ip_set_hash *
1228 +__ip_set_find(u_int32_t key, ip_set_id_t id, ip_set_ip_t ip)
1229 +{
1230 + struct ip_set_hash *set_hash;
1231 +
1232 + list_for_each_entry(set_hash, &ip_set_hash[key], list)
1233 + if (set_hash->id == id && set_hash->ip == ip)
1234 + return set_hash;
1235 +
1236 + return NULL;
1237 +}
1238 +
1239 +static ip_set_id_t
1240 +ip_set_find_in_hash(ip_set_id_t id, ip_set_ip_t ip)
1241 +{
1242 + u_int32_t key = jhash_2words(id, ip, ip_set_hash_random)
1243 + % ip_set_bindings_hash_size;
1244 + struct ip_set_hash *set_hash;
1245 +
1246 + ASSERT_READ_LOCK(&ip_set_lock);
1247 + IP_SET_ASSERT(ip_set_list[id]);
1248 + DP("set: %s, ip: %u.%u.%u.%u", ip_set_list[id]->name, HIPQUAD(ip));
1249 +
1250 + set_hash = __ip_set_find(key, id, ip);
1251 +
1252 + DP("set: %s, ip: %u.%u.%u.%u, binding: %s", ip_set_list[id]->name,
1253 + HIPQUAD(ip),
1254 + set_hash != NULL ? ip_set_list[set_hash->binding]->name : "");
1255 +
1256 + return (set_hash != NULL ? set_hash->binding : IP_SET_INVALID_ID);
1257 +}
1258 +
1259 +static inline void
1260 +__set_hash_del(struct ip_set_hash *set_hash)
1261 +{
1262 + ASSERT_WRITE_LOCK(&ip_set_lock);
1263 + IP_SET_ASSERT(ip_set_list[set_hash->binding]);
1264 +
1265 + __ip_set_put(set_hash->binding);
1266 + list_del(&set_hash->list);
1267 + kfree(set_hash);
1268 +}
1269 +
1270 +static int
1271 +ip_set_hash_del(ip_set_id_t id, ip_set_ip_t ip)
1272 +{
1273 + u_int32_t key = jhash_2words(id, ip, ip_set_hash_random)
1274 + % ip_set_bindings_hash_size;
1275 + struct ip_set_hash *set_hash;
1276 +
1277 + IP_SET_ASSERT(ip_set_list[id]);
1278 + DP("set: %s, ip: %u.%u.%u.%u", ip_set_list[id]->name, HIPQUAD(ip));
1279 + write_lock_bh(&ip_set_lock);
1280 + set_hash = __ip_set_find(key, id, ip);
1281 + DP("set: %s, ip: %u.%u.%u.%u, binding: %s", ip_set_list[id]->name,
1282 + HIPQUAD(ip),
1283 + set_hash != NULL ? ip_set_list[set_hash->binding]->name : "");
1284 +
1285 + if (set_hash != NULL)
1286 + __set_hash_del(set_hash);
1287 + write_unlock_bh(&ip_set_lock);
1288 + return 0;
1289 +}
1290 +
1291 +static int
1292 +ip_set_hash_add(ip_set_id_t id, ip_set_ip_t ip, ip_set_id_t binding)
1293 +{
1294 + u_int32_t key = jhash_2words(id, ip, ip_set_hash_random)
1295 + % ip_set_bindings_hash_size;
1296 + struct ip_set_hash *set_hash;
1297 + int ret = 0;
1298 +
1299 + IP_SET_ASSERT(ip_set_list[id]);
1300 + IP_SET_ASSERT(ip_set_list[binding]);
1301 + DP("set: %s, ip: %u.%u.%u.%u, binding: %s", ip_set_list[id]->name,
1302 + HIPQUAD(ip), ip_set_list[binding]->name);
1303 + write_lock_bh(&ip_set_lock);
1304 + set_hash = __ip_set_find(key, id, ip);
1305 + if (!set_hash) {
1306 + set_hash = kmalloc(sizeof(struct ip_set_hash), GFP_ATOMIC);
1307 + if (!set_hash) {
1308 + ret = -ENOMEM;
1309 + goto unlock;
1310 + }
1311 + INIT_LIST_HEAD(&set_hash->list);
1312 + set_hash->id = id;
1313 + set_hash->ip = ip;
1314 + list_add(&set_hash->list, &ip_set_hash[key]);
1315 + } else {
1316 + IP_SET_ASSERT(ip_set_list[set_hash->binding]);
1317 + DP("overwrite binding: %s",
1318 + ip_set_list[set_hash->binding]->name);
1319 + __ip_set_put(set_hash->binding);
1320 + }
1321 + set_hash->binding = binding;
1322 + __ip_set_get(set_hash->binding);
1323 + DP("stored: key %u, id %u (%s), ip %u.%u.%u.%u, binding %u (%s)",
1324 + key, id, ip_set_list[id]->name,
1325 + HIPQUAD(ip), binding, ip_set_list[binding]->name);
1326 + unlock:
1327 + write_unlock_bh(&ip_set_lock);
1328 + return ret;
1329 +}
1330 +
1331 +#define FOREACH_HASH_DO(fn, args...) \
1332 +({ \
1333 + ip_set_id_t __key; \
1334 + struct ip_set_hash *__set_hash; \
1335 + \
1336 + for (__key = 0; __key < ip_set_bindings_hash_size; __key++) { \
1337 + list_for_each_entry(__set_hash, &ip_set_hash[__key], list) \
1338 + fn(__set_hash , ## args); \
1339 + } \
1340 +})
1341 +
1342 +#define FOREACH_HASH_RW_DO(fn, args...) \
1343 +({ \
1344 + ip_set_id_t __key; \
1345 + struct ip_set_hash *__set_hash, *__n; \
1346 + \
1347 + ASSERT_WRITE_LOCK(&ip_set_lock); \
1348 + for (__key = 0; __key < ip_set_bindings_hash_size; __key++) { \
1349 + list_for_each_entry_safe(__set_hash, __n, &ip_set_hash[__key], list)\
1350 + fn(__set_hash , ## args); \
1351 + } \
1352 +})
1353 +
1354 +/* Add, del and test set entries from kernel */
1355 +
1356 +#define follow_bindings(index, set, ip) \
1357 +((index = ip_set_find_in_hash((set)->id, ip)) != IP_SET_INVALID_ID \
1358 + || (index = (set)->binding) != IP_SET_INVALID_ID)
1359 +
1360 +int
1361 +ip_set_testip_kernel(ip_set_id_t index,
1362 + const struct sk_buff *skb,
1363 + const u_int32_t *flags)
1364 +{
1365 + struct ip_set *set;
1366 + ip_set_ip_t ip;
1367 + int res;
1368 + unsigned char i = 0;
1369 +
1370 + IP_SET_ASSERT(flags[i]);
1371 + read_lock_bh(&ip_set_lock);
1372 + do {
1373 + set = ip_set_list[index];
1374 + IP_SET_ASSERT(set);
1375 + DP("set %s, index %u", set->name, index);
1376 + read_lock_bh(&set->lock);
1377 + res = set->type->testip_kernel(set, skb, &ip, flags, i++);
1378 + read_unlock_bh(&set->lock);
1379 + i += !!(set->type->features & IPSET_DATA_DOUBLE);
1380 + } while (res > 0
1381 + && flags[i]
1382 + && follow_bindings(index, set, ip));
1383 + read_unlock_bh(&ip_set_lock);
1384 +
1385 + return res;
1386 +}
1387 +
1388 +void
1389 +ip_set_addip_kernel(ip_set_id_t index,
1390 + const struct sk_buff *skb,
1391 + const u_int32_t *flags)
1392 +{
1393 + struct ip_set *set;
1394 + ip_set_ip_t ip;
1395 + int res;
1396 + unsigned char i = 0;
1397 +
1398 + IP_SET_ASSERT(flags[i]);
1399 + retry:
1400 + read_lock_bh(&ip_set_lock);
1401 + do {
1402 + set = ip_set_list[index];
1403 + IP_SET_ASSERT(set);
1404 + DP("set %s, index %u", set->name, index);
1405 + write_lock_bh(&set->lock);
1406 + res = set->type->addip_kernel(set, skb, &ip, flags, i++);
1407 + write_unlock_bh(&set->lock);
1408 + i += !!(set->type->features & IPSET_DATA_DOUBLE);
1409 + } while ((res == 0 || res == -EEXIST)
1410 + && flags[i]
1411 + && follow_bindings(index, set, ip));
1412 + read_unlock_bh(&ip_set_lock);
1413 +
1414 + if (res == -EAGAIN
1415 + && set->type->retry
1416 + && (res = set->type->retry(set)) == 0)
1417 + goto retry;
1418 +}
1419 +
1420 +void
1421 +ip_set_delip_kernel(ip_set_id_t index,
1422 + const struct sk_buff *skb,
1423 + const u_int32_t *flags)
1424 +{
1425 + struct ip_set *set;
1426 + ip_set_ip_t ip;
1427 + int res;
1428 + unsigned char i = 0;
1429 +
1430 + IP_SET_ASSERT(flags[i]);
1431 + read_lock_bh(&ip_set_lock);
1432 + do {
1433 + set = ip_set_list[index];
1434 + IP_SET_ASSERT(set);
1435 + DP("set %s, index %u", set->name, index);
1436 + write_lock_bh(&set->lock);
1437 + res = set->type->delip_kernel(set, skb, &ip, flags, i++);
1438 + write_unlock_bh(&set->lock);
1439 + i += !!(set->type->features & IPSET_DATA_DOUBLE);
1440 + } while ((res == 0 || res == -EEXIST)
1441 + && flags[i]
1442 + && follow_bindings(index, set, ip));
1443 + read_unlock_bh(&ip_set_lock);
1444 +}
1445 +
1446 +/* Register and deregister settype */
1447 +
1448 +static inline struct ip_set_type *
1449 +find_set_type(const char *name)
1450 +{
1451 + struct ip_set_type *set_type;
1452 +
1453 + list_for_each_entry(set_type, &set_type_list, list)
1454 + if (!strncmp(set_type->typename, name, IP_SET_MAXNAMELEN - 1))
1455 + return set_type;
1456 + return NULL;
1457 +}
1458 +
1459 +int
1460 +ip_set_register_set_type(struct ip_set_type *set_type)
1461 +{
1462 + int ret = 0;
1463 +
1464 + if (set_type->protocol_version != IP_SET_PROTOCOL_VERSION) {
1465 + ip_set_printk("'%s' uses wrong protocol version %u (want %u)",
1466 + set_type->typename,
1467 + set_type->protocol_version,
1468 + IP_SET_PROTOCOL_VERSION);
1469 + return -EINVAL;
1470 + }
1471 +
1472 + write_lock_bh(&ip_set_lock);
1473 + if (find_set_type(set_type->typename)) {
1474 + /* Duplicate! */
1475 + ip_set_printk("'%s' already registered!",
1476 + set_type->typename);
1477 + ret = -EINVAL;
1478 + goto unlock;
1479 + }
1480 + if (!try_module_get(THIS_MODULE)) {
1481 + ret = -EFAULT;
1482 + goto unlock;
1483 + }
1484 + list_add(&set_type->list, &set_type_list);
1485 + DP("'%s' registered.", set_type->typename);
1486 + unlock:
1487 + write_unlock_bh(&ip_set_lock);
1488 + return ret;
1489 +}
1490 +
1491 +void
1492 +ip_set_unregister_set_type(struct ip_set_type *set_type)
1493 +{
1494 + write_lock_bh(&ip_set_lock);
1495 + if (!find_set_type(set_type->typename)) {
1496 + ip_set_printk("'%s' not registered?",
1497 + set_type->typename);
1498 + goto unlock;
1499 + }
1500 + list_del(&set_type->list);
1501 + module_put(THIS_MODULE);
1502 + DP("'%s' unregistered.", set_type->typename);
1503 + unlock:
1504 + write_unlock_bh(&ip_set_lock);
1505 +
1506 +}
1507 +
1508 +/*
1509 + * Userspace routines
1510 + */
1511 +
1512 +/*
1513 + * Find set by name, reference it once. The reference makes sure the
1514 + * thing pointed to, does not go away under our feet. Drop the reference
1515 + * later, using ip_set_put().
1516 + */
1517 +ip_set_id_t
1518 +ip_set_get_byname(const char *name)
1519 +{
1520 + ip_set_id_t i, index = IP_SET_INVALID_ID;
1521 +
1522 + down(&ip_set_app_mutex);
1523 + for (i = 0; i < ip_set_max; i++) {
1524 + if (ip_set_list[i] != NULL
1525 + && strcmp(ip_set_list[i]->name, name) == 0) {
1526 + __ip_set_get(i);
1527 + index = i;
1528 + break;
1529 + }
1530 + }
1531 + up(&ip_set_app_mutex);
1532 + return index;
1533 +}
1534 +
1535 +/*
1536 + * Find set by index, reference it once. The reference makes sure the
1537 + * thing pointed to, does not go away under our feet. Drop the reference
1538 + * later, using ip_set_put().
1539 + */
1540 +ip_set_id_t
1541 +ip_set_get_byindex(ip_set_id_t index)
1542 +{
1543 + down(&ip_set_app_mutex);
1544 +
1545 + if (index >= ip_set_max)
1546 + return IP_SET_INVALID_ID;
1547 +
1548 + if (ip_set_list[index])
1549 + __ip_set_get(index);
1550 + else
1551 + index = IP_SET_INVALID_ID;
1552 +
1553 + up(&ip_set_app_mutex);
1554 + return index;
1555 +}
1556 +
1557 +/*
1558 + * If the given set pointer points to a valid set, decrement
1559 + * reference count by 1. The caller shall not assume the index
1560 + * to be valid, after calling this function.
1561 + */
1562 +void ip_set_put(ip_set_id_t index)
1563 +{
1564 + down(&ip_set_app_mutex);
1565 + if (ip_set_list[index])
1566 + __ip_set_put(index);
1567 + up(&ip_set_app_mutex);
1568 +}
1569 +
1570 +/* Find a set by name or index */
1571 +static ip_set_id_t
1572 +ip_set_find_byname(const char *name)
1573 +{
1574 + ip_set_id_t i, index = IP_SET_INVALID_ID;
1575 +
1576 + for (i = 0; i < ip_set_max; i++) {
1577 + if (ip_set_list[i] != NULL
1578 + && strcmp(ip_set_list[i]->name, name) == 0) {
1579 + index = i;
1580 + break;
1581 + }
1582 + }
1583 + return index;
1584 +}
1585 +
1586 +static ip_set_id_t
1587 +ip_set_find_byindex(ip_set_id_t index)
1588 +{
1589 + if (index >= ip_set_max || ip_set_list[index] == NULL)
1590 + index = IP_SET_INVALID_ID;
1591 +
1592 + return index;
1593 +}
1594 +
1595 +/*
1596 + * Add, del, test, bind and unbind
1597 + */
1598 +
1599 +static inline int
1600 +__ip_set_testip(struct ip_set *set,
1601 + const void *data,
1602 + size_t size,
1603 + ip_set_ip_t *ip)
1604 +{
1605 + int res;
1606 +
1607 + read_lock_bh(&set->lock);
1608 + res = set->type->testip(set, data, size, ip);
1609 + read_unlock_bh(&set->lock);
1610 +
1611 + return res;
1612 +}
1613 +
1614 +static int
1615 +__ip_set_addip(ip_set_id_t index,
1616 + const void *data,
1617 + size_t size)
1618 +{
1619 + struct ip_set *set = ip_set_list[index];
1620 + ip_set_ip_t ip;
1621 + int res;
1622 +
1623 + IP_SET_ASSERT(set);
1624 + do {
1625 + write_lock_bh(&set->lock);
1626 + res = set->type->addip(set, data, size, &ip);
1627 + write_unlock_bh(&set->lock);
1628 + } while (res == -EAGAIN
1629 + && set->type->retry
1630 + && (res = set->type->retry(set)) == 0);
1631 +
1632 + return res;
1633 +}
1634 +
1635 +static int
1636 +ip_set_addip(ip_set_id_t index,
1637 + const void *data,
1638 + size_t size)
1639 +{
1640 +
1641 + return __ip_set_addip(index,
1642 + data + sizeof(struct ip_set_req_adt),
1643 + size - sizeof(struct ip_set_req_adt));
1644 +}
1645 +
1646 +static int
1647 +ip_set_delip(ip_set_id_t index,
1648 + const void *data,
1649 + size_t size)
1650 +{
1651 + struct ip_set *set = ip_set_list[index];
1652 + ip_set_ip_t ip;
1653 + int res;
1654 +
1655 + IP_SET_ASSERT(set);
1656 + write_lock_bh(&set->lock);
1657 + res = set->type->delip(set,
1658 + data + sizeof(struct ip_set_req_adt),
1659 + size - sizeof(struct ip_set_req_adt),
1660 + &ip);
1661 + write_unlock_bh(&set->lock);
1662 +
1663 + return res;
1664 +}
1665 +
1666 +static int
1667 +ip_set_testip(ip_set_id_t index,
1668 + const void *data,
1669 + size_t size)
1670 +{
1671 + struct ip_set *set = ip_set_list[index];
1672 + ip_set_ip_t ip;
1673 + int res;
1674 +
1675 + IP_SET_ASSERT(set);
1676 + res = __ip_set_testip(set,
1677 + data + sizeof(struct ip_set_req_adt),
1678 + size - sizeof(struct ip_set_req_adt),
1679 + &ip);
1680 +
1681 + return (res > 0 ? -EEXIST : res);
1682 +}
1683 +
1684 +static int
1685 +ip_set_bindip(ip_set_id_t index,
1686 + const void *data,
1687 + size_t size)
1688 +{
1689 + struct ip_set *set = ip_set_list[index];
1690 + struct ip_set_req_bind *req_bind;
1691 + ip_set_id_t binding;
1692 + ip_set_ip_t ip;
1693 + int res;
1694 +
1695 + IP_SET_ASSERT(set);
1696 + if (size < sizeof(struct ip_set_req_bind))
1697 + return -EINVAL;
1698 +
1699 + req_bind = (struct ip_set_req_bind *) data;
1700 + req_bind->binding[IP_SET_MAXNAMELEN - 1] = '\0';
1701 +
1702 + if (strcmp(req_bind->binding, IPSET_TOKEN_DEFAULT) == 0) {
1703 + /* Default binding of a set */
1704 + char *binding_name;
1705 +
1706 + if (size != sizeof(struct ip_set_req_bind) + IP_SET_MAXNAMELEN)
1707 + return -EINVAL;
1708 +
1709 + binding_name = (char *)(data + sizeof(struct ip_set_req_bind));
1710 + binding_name[IP_SET_MAXNAMELEN - 1] = '\0';
1711 +
1712 + binding = ip_set_find_byname(binding_name);
1713 + if (binding == IP_SET_INVALID_ID)
1714 + return -ENOENT;
1715 +
1716 + write_lock_bh(&ip_set_lock);
1717 + /* Sets as binding values are referenced */
1718 + if (set->binding != IP_SET_INVALID_ID)
1719 + __ip_set_put(set->binding);
1720 + set->binding = binding;
1721 + __ip_set_get(set->binding);
1722 + write_unlock_bh(&ip_set_lock);
1723 +
1724 + return 0;
1725 + }
1726 + binding = ip_set_find_byname(req_bind->binding);
1727 + if (binding == IP_SET_INVALID_ID)
1728 + return -ENOENT;
1729 +
1730 + res = __ip_set_testip(set,
1731 + data + sizeof(struct ip_set_req_bind),
1732 + size - sizeof(struct ip_set_req_bind),
1733 + &ip);
1734 + DP("set %s, ip: %u.%u.%u.%u, binding %s",
1735 + set->name, HIPQUAD(ip), ip_set_list[binding]->name);
1736 +
1737 + if (res >= 0)
1738 + res = ip_set_hash_add(set->id, ip, binding);
1739 +
1740 + return res;
1741 +}
1742 +
1743 +#define FOREACH_SET_DO(fn, args...) \
1744 +({ \
1745 + ip_set_id_t __i; \
1746 + struct ip_set *__set; \
1747 + \
1748 + for (__i = 0; __i < ip_set_max; __i++) { \
1749 + __set = ip_set_list[__i]; \
1750 + if (__set != NULL) \
1751 + fn(__set , ##args); \
1752 + } \
1753 +})
1754 +
1755 +static inline void
1756 +__set_hash_del_byid(struct ip_set_hash *set_hash, ip_set_id_t id)
1757 +{
1758 + if (set_hash->id == id)
1759 + __set_hash_del(set_hash);
1760 +}
1761 +
1762 +static inline void
1763 +__unbind_default(struct ip_set *set)
1764 +{
1765 + if (set->binding != IP_SET_INVALID_ID) {
1766 + /* Sets as binding values are referenced */
1767 + __ip_set_put(set->binding);
1768 + set->binding = IP_SET_INVALID_ID;
1769 + }
1770 +}
1771 +
1772 +static int
1773 +ip_set_unbindip(ip_set_id_t index,
1774 + const void *data,
1775 + size_t size)
1776 +{
1777 + struct ip_set *set;
1778 + struct ip_set_req_bind *req_bind;
1779 + ip_set_ip_t ip;
1780 + int res;
1781 +
1782 + DP("");
1783 + if (size < sizeof(struct ip_set_req_bind))
1784 + return -EINVAL;
1785 +
1786 + req_bind = (struct ip_set_req_bind *) data;
1787 + req_bind->binding[IP_SET_MAXNAMELEN - 1] = '\0';
1788 +
1789 + DP("%u %s", index, req_bind->binding);
1790 + if (index == IP_SET_INVALID_ID) {
1791 + /* unbind :all: */
1792 + if (strcmp(req_bind->binding, IPSET_TOKEN_DEFAULT) == 0) {
1793 + /* Default binding of sets */
1794 + write_lock_bh(&ip_set_lock);
1795 + FOREACH_SET_DO(__unbind_default);
1796 + write_unlock_bh(&ip_set_lock);
1797 + return 0;
1798 + } else if (strcmp(req_bind->binding, IPSET_TOKEN_ALL) == 0) {
1799 + /* Flush all bindings of all sets*/
1800 + write_lock_bh(&ip_set_lock);
1801 + FOREACH_HASH_RW_DO(__set_hash_del);
1802 + write_unlock_bh(&ip_set_lock);
1803 + return 0;
1804 + }
1805 + DP("unreachable reached!");
1806 + return -EINVAL;
1807 + }
1808 +
1809 + set = ip_set_list[index];
1810 + IP_SET_ASSERT(set);
1811 + if (strcmp(req_bind->binding, IPSET_TOKEN_DEFAULT) == 0) {
1812 + /* Default binding of set */
1813 + ip_set_id_t binding = ip_set_find_byindex(set->binding);
1814 +
1815 + if (binding == IP_SET_INVALID_ID)
1816 + return -ENOENT;
1817 +
1818 + write_lock_bh(&ip_set_lock);
1819 + /* Sets in hash values are referenced */
1820 + __ip_set_put(set->binding);
1821 + set->binding = IP_SET_INVALID_ID;
1822 + write_unlock_bh(&ip_set_lock);
1823 +
1824 + return 0;
1825 + } else if (strcmp(req_bind->binding, IPSET_TOKEN_ALL) == 0) {
1826 + /* Flush all bindings */
1827 +
1828 + write_lock_bh(&ip_set_lock);
1829 + FOREACH_HASH_RW_DO(__set_hash_del_byid, set->id);
1830 + write_unlock_bh(&ip_set_lock);
1831 + return 0;
1832 + }
1833 +
1834 + res = __ip_set_testip(set,
1835 + data + sizeof(struct ip_set_req_bind),
1836 + size - sizeof(struct ip_set_req_bind),
1837 + &ip);
1838 +
1839 + DP("set %s, ip: %u.%u.%u.%u", set->name, HIPQUAD(ip));
1840 + if (res >= 0)
1841 + res = ip_set_hash_del(set->id, ip);
1842 +
1843 + return res;
1844 +}
1845 +
1846 +static int
1847 +ip_set_testbind(ip_set_id_t index,
1848 + const void *data,
1849 + size_t size)
1850 +{
1851 + struct ip_set *set = ip_set_list[index];
1852 + struct ip_set_req_bind *req_bind;
1853 + ip_set_id_t binding;
1854 + ip_set_ip_t ip;
1855 + int res;
1856 +
1857 + IP_SET_ASSERT(set);
1858 + if (size < sizeof(struct ip_set_req_bind))
1859 + return -EINVAL;
1860 +
1861 + req_bind = (struct ip_set_req_bind *) data;
1862 + req_bind->binding[IP_SET_MAXNAMELEN - 1] = '\0';
1863 +
1864 + if (strcmp(req_bind->binding, IPSET_TOKEN_DEFAULT) == 0) {
1865 + /* Default binding of set */
1866 + char *binding_name;
1867 +
1868 + if (size != sizeof(struct ip_set_req_bind) + IP_SET_MAXNAMELEN)
1869 + return -EINVAL;
1870 +
1871 + binding_name = (char *)(data + sizeof(struct ip_set_req_bind));
1872 + binding_name[IP_SET_MAXNAMELEN - 1] = '\0';
1873 +
1874 + binding = ip_set_find_byname(binding_name);
1875 + if (binding == IP_SET_INVALID_ID)
1876 + return -ENOENT;
1877 +
1878 + res = (set->binding == binding) ? -EEXIST : 0;
1879 +
1880 + return res;
1881 + }
1882 + binding = ip_set_find_byname(req_bind->binding);
1883 + if (binding == IP_SET_INVALID_ID)
1884 + return -ENOENT;
1885 +
1886 +
1887 + res = __ip_set_testip(set,
1888 + data + sizeof(struct ip_set_req_bind),
1889 + size - sizeof(struct ip_set_req_bind),
1890 + &ip);
1891 + DP("set %s, ip: %u.%u.%u.%u, binding %s",
1892 + set->name, HIPQUAD(ip), ip_set_list[binding]->name);
1893 +
1894 + if (res >= 0)
1895 + res = (ip_set_find_in_hash(set->id, ip) == binding)
1896 + ? -EEXIST : 0;
1897 +
1898 + return res;
1899 +}
1900 +
1901 +static struct ip_set_type *
1902 +find_set_type_rlock(const char *typename)
1903 +{
1904 + struct ip_set_type *type;
1905 +
1906 + read_lock_bh(&ip_set_lock);
1907 + type = find_set_type(typename);
1908 + if (type == NULL)
1909 + read_unlock_bh(&ip_set_lock);
1910 +
1911 + return type;
1912 +}
1913 +
1914 +static int
1915 +find_free_id(const char *name,
1916 + ip_set_id_t *index,
1917 + ip_set_id_t *id)
1918 +{
1919 + ip_set_id_t i;
1920 +
1921 + *id = IP_SET_INVALID_ID;
1922 + for (i = 0; i < ip_set_max; i++) {
1923 + if (ip_set_list[i] == NULL) {
1924 + if (*id == IP_SET_INVALID_ID)
1925 + *id = *index = i;
1926 + } else if (strcmp(name, ip_set_list[i]->name) == 0)
1927 + /* Name clash */
1928 + return -EEXIST;
1929 + }
1930 + if (*id == IP_SET_INVALID_ID)
1931 + /* No free slot remained */
1932 + return -ERANGE;
1933 + /* Check that index is usable as id (swapping) */
1934 + check:
1935 + for (i = 0; i < ip_set_max; i++) {
1936 + if (ip_set_list[i] != NULL
1937 + && ip_set_list[i]->id == *id) {
1938 + *id = i;
1939 + goto check;
1940 + }
1941 + }
1942 + return 0;
1943 +}
1944 +
1945 +/*
1946 + * Create a set
1947 + */
1948 +static int
1949 +ip_set_create(const char *name,
1950 + const char *typename,
1951 + ip_set_id_t restore,
1952 + const void *data,
1953 + size_t size)
1954 +{
1955 + struct ip_set *set;
1956 + ip_set_id_t index = 0, id;
1957 + int res = 0;
1958 +
1959 + DP("setname: %s, typename: %s, id: %u", name, typename, restore);
1960 + /*
1961 + * First, and without any locks, allocate and initialize
1962 + * a normal base set structure.
1963 + */
1964 + set = kmalloc(sizeof(struct ip_set), GFP_KERNEL);
1965 + if (!set)
1966 + return -ENOMEM;
1967 + set->lock = RW_LOCK_UNLOCKED;
1968 + strncpy(set->name, name, IP_SET_MAXNAMELEN);
1969 + set->binding = IP_SET_INVALID_ID;
1970 + atomic_set(&set->ref, 0);
1971 +
1972 + /*
1973 + * Next, take the &ip_set_lock, check that we know the type,
1974 + * and take a reference on the type, to make sure it
1975 + * stays available while constructing our new set.
1976 + *
1977 + * After referencing the type, we drop the &ip_set_lock,
1978 + * and let the new set construction run without locks.
1979 + */
1980 + set->type = find_set_type_rlock(typename);
1981 + if (set->type == NULL) {
1982 + /* Try loading the module */
1983 + char modulename[IP_SET_MAXNAMELEN + strlen("ip_set_") + 1];
1984 + strcpy(modulename, "ip_set_");
1985 + strcat(modulename, typename);
1986 + DP("try to load %s", modulename);
1987 + request_module(modulename);
1988 + set->type = find_set_type_rlock(typename);
1989 + }
1990 + if (set->type == NULL) {
1991 + ip_set_printk("no set type '%s', set '%s' not created",
1992 + typename, name);
1993 + res = -ENOENT;
1994 + goto out;
1995 + }
1996 + if (!try_module_get(set->type->me)) {
1997 + read_unlock_bh(&ip_set_lock);
1998 + res = -EFAULT;
1999 + goto out;
2000 + }
2001 + read_unlock_bh(&ip_set_lock);
2002 +
2003 + /*
2004 + * Without holding any locks, create private part.
2005 + */
2006 + res = set->type->create(set, data, size);
2007 + if (res != 0)
2008 + goto put_out;
2009 +
2010 + /* BTW, res==0 here. */
2011 +
2012 + /*
2013 + * Here, we have a valid, constructed set. &ip_set_lock again,
2014 + * find free id/index and check that it is not already in
2015 + * ip_set_list.
2016 + */
2017 + write_lock_bh(&ip_set_lock);
2018 + if ((res = find_free_id(set->name, &index, &id)) != 0) {
2019 + DP("no free id!");
2020 + goto cleanup;
2021 + }
2022 +
2023 + /* Make sure restore gets the same index */
2024 + if (restore != IP_SET_INVALID_ID && index != restore) {
2025 + DP("Can't restore, sets are screwed up");
2026 + res = -ERANGE;
2027 + goto cleanup;
2028 + }
2029 +
2030 + /*
2031 + * Finally! Add our shiny new set to the list, and be done.
2032 + */
2033 + DP("create: '%s' created with index %u, id %u!", set->name, index, id);
2034 + set->id = id;
2035 + ip_set_list[index] = set;
2036 + write_unlock_bh(&ip_set_lock);
2037 + return res;
2038 +
2039 + cleanup:
2040 + write_unlock_bh(&ip_set_lock);
2041 + set->type->destroy(set);
2042 + put_out:
2043 + module_put(set->type->me);
2044 + out:
2045 + kfree(set);
2046 + return res;
2047 +}
2048 +
2049 +/*
2050 + * Destroy a given existing set
2051 + */
2052 +static void
2053 +ip_set_destroy_set(ip_set_id_t index)
2054 +{
2055 + struct ip_set *set = ip_set_list[index];
2056 +
2057 + IP_SET_ASSERT(set);
2058 + DP("set: %s", set->name);
2059 + write_lock_bh(&ip_set_lock);
2060 + FOREACH_HASH_RW_DO(__set_hash_del_byid, set->id);
2061 + if (set->binding != IP_SET_INVALID_ID)
2062 + __ip_set_put(set->binding);
2063 + ip_set_list[index] = NULL;
2064 + write_unlock_bh(&ip_set_lock);
2065 +
2066 + /* Must call it without holding any lock */
2067 + set->type->destroy(set);
2068 + module_put(set->type->me);
2069 + kfree(set);
2070 +}
2071 +
2072 +/*
2073 + * Destroy a set - or all sets
2074 + * Sets must not be referenced/used.
2075 + */
2076 +static int
2077 +ip_set_destroy(ip_set_id_t index)
2078 +{
2079 + ip_set_id_t i;
2080 +
2081 + /* ref modification always protected by the mutex */
2082 + if (index != IP_SET_INVALID_ID) {
2083 + if (atomic_read(&ip_set_list[index]->ref))
2084 + return -EBUSY;
2085 + ip_set_destroy_set(index);
2086 + } else {
2087 + for (i = 0; i < ip_set_max; i++) {
2088 + if (ip_set_list[i] != NULL
2089 + && (atomic_read(&ip_set_list[i]->ref)))
2090 + return -EBUSY;
2091 + }
2092 +
2093 + for (i = 0; i < ip_set_max; i++) {
2094 + if (ip_set_list[i] != NULL)
2095 + ip_set_destroy_set(i);
2096 + }
2097 + }
2098 + return 0;
2099 +}
2100 +
2101 +static void
2102 +ip_set_flush_set(struct ip_set *set)
2103 +{
2104 + DP("set: %s %u", set->name, set->id);
2105 +
2106 + write_lock_bh(&set->lock);
2107 + set->type->flush(set);
2108 + write_unlock_bh(&set->lock);
2109 +}
2110 +
2111 +/*
2112 + * Flush data in a set - or in all sets
2113 + */
2114 +static int
2115 +ip_set_flush(ip_set_id_t index)
2116 +{
2117 + if (index != IP_SET_INVALID_ID) {
2118 + IP_SET_ASSERT(ip_set_list[index]);
2119 + ip_set_flush_set(ip_set_list[index]);
2120 + } else
2121 + FOREACH_SET_DO(ip_set_flush_set);
2122 +
2123 + return 0;
2124 +}
2125 +
2126 +/* Rename a set */
2127 +static int
2128 +ip_set_rename(ip_set_id_t index, const char *name)
2129 +{
2130 + struct ip_set *set = ip_set_list[index];
2131 + ip_set_id_t i;
2132 + int res = 0;
2133 +
2134 + DP("set: %s to %s", set->name, name);
2135 + write_lock_bh(&ip_set_lock);
2136 + for (i = 0; i < ip_set_max; i++) {
2137 + if (ip_set_list[i] != NULL
2138 + && strncmp(ip_set_list[i]->name,
2139 + name,
2140 + IP_SET_MAXNAMELEN - 1) == 0) {
2141 + res = -EEXIST;
2142 + goto unlock;
2143 + }
2144 + }
2145 + strncpy(set->name, name, IP_SET_MAXNAMELEN);
2146 + unlock:
2147 + write_unlock_bh(&ip_set_lock);
2148 + return res;
2149 +}
2150 +
2151 +/*
2152 + * Swap two sets so that name/index points to the other.
2153 + * References are also swapped.
2154 + */
2155 +static int
2156 +ip_set_swap(ip_set_id_t from_index, ip_set_id_t to_index)
2157 +{
2158 + struct ip_set *from = ip_set_list[from_index];
2159 + struct ip_set *to = ip_set_list[to_index];
2160 + char from_name[IP_SET_MAXNAMELEN];
2161 + u_int32_t from_ref;
2162 +
2163 + DP("set: %s to %s", from->name, to->name);
2164 + /* Features must not change. Artifical restriction. */
2165 + if (from->type->features != to->type->features)
2166 + return -ENOEXEC;
2167 +
2168 + /* No magic here: ref munging protected by the mutex */
2169 + write_lock_bh(&ip_set_lock);
2170 + strncpy(from_name, from->name, IP_SET_MAXNAMELEN);
2171 + from_ref = atomic_read(&from->ref);
2172 +
2173 + strncpy(from->name, to->name, IP_SET_MAXNAMELEN);
2174 + atomic_set(&from->ref, atomic_read(&to->ref));
2175 + strncpy(to->name, from_name, IP_SET_MAXNAMELEN);
2176 + atomic_set(&to->ref, from_ref);
2177 +
2178 + ip_set_list[from_index] = to;
2179 + ip_set_list[to_index] = from;
2180 +
2181 + write_unlock_bh(&ip_set_lock);
2182 + return 0;
2183 +}
2184 +
2185 +/*
2186 + * List set data
2187 + */
2188 +
2189 +static inline void
2190 +__set_hash_bindings_size_list(struct ip_set_hash *set_hash,
2191 + ip_set_id_t id, size_t *size)
2192 +{
2193 + if (set_hash->id == id)
2194 + *size += sizeof(struct ip_set_hash_list);
2195 +}
2196 +
2197 +static inline void
2198 +__set_hash_bindings_size_save(struct ip_set_hash *set_hash,
2199 + ip_set_id_t id, size_t *size)
2200 +{
2201 + if (set_hash->id == id)
2202 + *size += sizeof(struct ip_set_hash_save);
2203 +}
2204 +
2205 +static inline void
2206 +__set_hash_bindings(struct ip_set_hash *set_hash,
2207 + ip_set_id_t id, void *data, int *used)
2208 +{
2209 + if (set_hash->id == id) {
2210 + struct ip_set_hash_list *hash_list =
2211 + (struct ip_set_hash_list *)(data + *used);
2212 +
2213 + hash_list->ip = set_hash->ip;
2214 + hash_list->binding = set_hash->binding;
2215 + *used += sizeof(struct ip_set_hash_list);
2216 + }
2217 +}
2218 +
2219 +static int ip_set_list_set(ip_set_id_t index,
2220 + void *data,
2221 + int *used,
2222 + int len)
2223 +{
2224 + struct ip_set *set = ip_set_list[index];
2225 + struct ip_set_list *set_list;
2226 +
2227 + /* Pointer to our header */
2228 + set_list = (struct ip_set_list *) (data + *used);
2229 +
2230 + DP("set: %s, used: %d %p %p", set->name, *used, data, data + *used);
2231 +
2232 + /* Get and ensure header size */
2233 + if (*used + sizeof(struct ip_set_list) > len)
2234 + goto not_enough_mem;
2235 + *used += sizeof(struct ip_set_list);
2236 +
2237 + read_lock_bh(&set->lock);
2238 + /* Get and ensure set specific header size */
2239 + set_list->header_size = set->type->header_size;
2240 + if (*used + set_list->header_size > len)
2241 + goto unlock_set;
2242 +
2243 + /* Fill in the header */
2244 + set_list->index = index;
2245 + set_list->binding = set->binding;
2246 + set_list->ref = atomic_read(&set->ref);
2247 +
2248 + /* Fill in set spefific header data */
2249 + set->type->list_header(set, data + *used);
2250 + *used += set_list->header_size;
2251 +
2252 + /* Get and ensure set specific members size */
2253 + set_list->members_size = set->type->list_members_size(set);
2254 + if (*used + set_list->members_size > len)
2255 + goto unlock_set;
2256 +
2257 + /* Fill in set spefific members data */
2258 + set->type->list_members(set, data + *used);
2259 + *used += set_list->members_size;
2260 + read_unlock_bh(&set->lock);
2261 +
2262 + /* Bindings */
2263 +
2264 + /* Get and ensure set specific bindings size */
2265 + set_list->bindings_size = 0;
2266 + FOREACH_HASH_DO(__set_hash_bindings_size_list,
2267 + set->id, &set_list->bindings_size);
2268 + if (*used + set_list->bindings_size > len)
2269 + goto not_enough_mem;
2270 +
2271 + /* Fill in set spefific bindings data */
2272 + FOREACH_HASH_DO(__set_hash_bindings, set->id, data, used);
2273 +
2274 + return 0;
2275 +
2276 + unlock_set:
2277 + read_unlock_bh(&set->lock);
2278 + not_enough_mem:
2279 + DP("not enough mem, try again");
2280 + return -EAGAIN;
2281 +}
2282 +
2283 +/*
2284 + * Save sets
2285 + */
2286 +static int ip_set_save_set(ip_set_id_t index,
2287 + void *data,
2288 + int *used,
2289 + int len)
2290 +{
2291 + struct ip_set *set;
2292 + struct ip_set_save *set_save;
2293 +
2294 + /* Pointer to our header */
2295 + set_save = (struct ip_set_save *) (data + *used);
2296 +
2297 + /* Get and ensure header size */
2298 + if (*used + sizeof(struct ip_set_save) > len)
2299 + goto not_enough_mem;
2300 + *used += sizeof(struct ip_set_save);
2301 +
2302 + set = ip_set_list[index];
2303 + DP("set: %s, used: %u(%u) %p %p", set->name, *used, len,
2304 + data, data + *used);
2305 +
2306 + read_lock_bh(&set->lock);
2307 + /* Get and ensure set specific header size */
2308 + set_save->header_size = set->type->header_size;
2309 + if (*used + set_save->header_size > len)
2310 + goto unlock_set;
2311 +
2312 + /* Fill in the header */
2313 + set_save->index = index;
2314 + set_save->binding = set->binding;
2315 +
2316 + /* Fill in set spefific header data */
2317 + set->type->list_header(set, data + *used);
2318 + *used += set_save->header_size;
2319 +
2320 + DP("set header filled: %s, used: %u(%u) %p %p", set->name, *used,
2321 + set_save->header_size, data, data + *used);
2322 + /* Get and ensure set specific members size */
2323 + set_save->members_size = set->type->list_members_size(set);
2324 + if (*used + set_save->members_size > len)
2325 + goto unlock_set;
2326 +
2327 + /* Fill in set spefific members data */
2328 + set->type->list_members(set, data + *used);
2329 + *used += set_save->members_size;
2330 + read_unlock_bh(&set->lock);
2331 + DP("set members filled: %s, used: %u(%u) %p %p", set->name, *used,
2332 + set_save->members_size, data, data + *used);
2333 + return 0;
2334 +
2335 + unlock_set:
2336 + read_unlock_bh(&set->lock);
2337 + not_enough_mem:
2338 + DP("not enough mem, try again");
2339 + return -EAGAIN;
2340 +}
2341 +
2342 +static inline void
2343 +__set_hash_save_bindings(struct ip_set_hash *set_hash,
2344 + ip_set_id_t id,
2345 + void *data,
2346 + int *used,
2347 + int len,
2348 + int *res)
2349 +{
2350 + if (*res == 0
2351 + && (id == IP_SET_INVALID_ID || set_hash->id == id)) {
2352 + struct ip_set_hash_save *hash_save =
2353 + (struct ip_set_hash_save *)(data + *used);
2354 + /* Ensure bindings size */
2355 + if (*used + sizeof(struct ip_set_hash_save) > len) {
2356 + *res = -ENOMEM;
2357 + return;
2358 + }
2359 + hash_save->id = set_hash->id;
2360 + hash_save->ip = set_hash->ip;
2361 + hash_save->binding = set_hash->binding;
2362 + *used += sizeof(struct ip_set_hash_save);
2363 + }
2364 +}
2365 +
2366 +static int ip_set_save_bindings(ip_set_id_t index,
2367 + void *data,
2368 + int *used,
2369 + int len)
2370 +{
2371 + int res = 0;
2372 + struct ip_set_save *set_save;
2373 +
2374 + DP("used %u, len %u", *used, len);
2375 + /* Get and ensure header size */
2376 + if (*used + sizeof(struct ip_set_save) > len)
2377 + return -ENOMEM;
2378 +
2379 + /* Marker */
2380 + set_save = (struct ip_set_save *) (data + *used);
2381 + set_save->index = IP_SET_INVALID_ID;
2382 + set_save->header_size = 0;
2383 + set_save->members_size = 0;
2384 + *used += sizeof(struct ip_set_save);
2385 +
2386 + DP("marker added used %u, len %u", *used, len);
2387 + /* Fill in bindings data */
2388 + if (index != IP_SET_INVALID_ID)
2389 + /* Sets are identified by id in hash */
2390 + index = ip_set_list[index]->id;
2391 + FOREACH_HASH_DO(__set_hash_save_bindings, index, data, used, len, &res);
2392 +
2393 + return res;
2394 +}
2395 +
2396 +/*
2397 + * Restore sets
2398 + */
2399 +static int ip_set_restore(void *data,
2400 + int len)
2401 +{
2402 + int res = 0;
2403 + int line = 0, used = 0, members_size;
2404 + struct ip_set *set;
2405 + struct ip_set_hash_save *hash_save;
2406 + struct ip_set_restore *set_restore;
2407 + ip_set_id_t index;
2408 +
2409 + /* Loop to restore sets */
2410 + while (1) {
2411 + line++;
2412 +
2413 + DP("%u %u %u", used, sizeof(struct ip_set_restore), len);
2414 + /* Get and ensure header size */
2415 + if (used + sizeof(struct ip_set_restore) > len)
2416 + return line;
2417 + set_restore = (struct ip_set_restore *) (data + used);
2418 + used += sizeof(struct ip_set_restore);
2419 +
2420 + /* Ensure data size */
2421 + if (used
2422 + + set_restore->header_size
2423 + + set_restore->members_size > len)
2424 + return line;
2425 +
2426 + /* Check marker */
2427 + if (set_restore->index == IP_SET_INVALID_ID) {
2428 + line--;
2429 + goto bindings;
2430 + }
2431 +
2432 + /* Try to create the set */
2433 + DP("restore %s %s", set_restore->name, set_restore->typename);
2434 + res = ip_set_create(set_restore->name,
2435 + set_restore->typename,
2436 + set_restore->index,
2437 + data + used,
2438 + set_restore->header_size);
2439 +
2440 + if (res != 0)
2441 + return line;
2442 + used += set_restore->header_size;
2443 +
2444 + index = ip_set_find_byindex(set_restore->index);
2445 + DP("index %u, restore_index %u", index, set_restore->index);
2446 + if (index != set_restore->index)
2447 + return line;
2448 + /* Try to restore members data */
2449 + set = ip_set_list[index];
2450 + members_size = 0;
2451 + DP("members_size %u reqsize %u",
2452 + set_restore->members_size, set->type->reqsize);
2453 + while (members_size + set->type->reqsize <=
2454 + set_restore->members_size) {
2455 + line++;
2456 + DP("members: %u, line %u", members_size, line);
2457 + res = __ip_set_addip(index,
2458 + data + used + members_size,
2459 + set->type->reqsize);
2460 + if (!(res == 0 || res == -EEXIST))
2461 + return line;
2462 + members_size += set->type->reqsize;
2463 + }
2464 +
2465 + DP("members_size %u %u",
2466 + set_restore->members_size, members_size);
2467 + if (members_size != set_restore->members_size)
2468 + return line++;
2469 + used += set_restore->members_size;
2470 + }
2471 +
2472 + bindings:
2473 + /* Loop to restore bindings */
2474 + while (used < len) {
2475 + line++;
2476 +
2477 + DP("restore binding, line %u", line);
2478 + /* Get and ensure size */
2479 + if (used + sizeof(struct ip_set_hash_save) > len)
2480 + return line;
2481 + hash_save = (struct ip_set_hash_save *) (data + used);
2482 + used += sizeof(struct ip_set_hash_save);
2483 +
2484 + /* hash_save->id is used to store the index */
2485 + index = ip_set_find_byindex(hash_save->id);
2486 + DP("restore binding index %u, id %u, %u -> %u",
2487 + index, hash_save->id, hash_save->ip, hash_save->binding);
2488 + if (index != hash_save->id)
2489 + return line;
2490 + if (ip_set_find_byindex(hash_save->binding) == IP_SET_INVALID_ID) {
2491 + DP("corrupt binding set index %u", hash_save->binding);
2492 + return line;
2493 + }
2494 + set = ip_set_list[hash_save->id];
2495 + /* Null valued IP means default binding */
2496 + if (hash_save->ip)
2497 + res = ip_set_hash_add(set->id,
2498 + hash_save->ip,
2499 + hash_save->binding);
2500 + else {
2501 + IP_SET_ASSERT(set->binding == IP_SET_INVALID_ID);
2502 + write_lock_bh(&ip_set_lock);
2503 + set->binding = hash_save->binding;
2504 + __ip_set_get(set->binding);
2505 + write_unlock_bh(&ip_set_lock);
2506 + DP("default binding: %u", set->binding);
2507 + }
2508 + if (res != 0)
2509 + return line;
2510 + }
2511 + if (used != len)
2512 + return line;
2513 +
2514 + return 0;
2515 +}
2516 +
2517 +static int
2518 +ip_set_sockfn_set(struct sock *sk, int optval, void *user, unsigned int len)
2519 +{
2520 + void *data;
2521 + int res = 0; /* Assume OK */
2522 + unsigned *op;
2523 + struct ip_set_req_adt *req_adt;
2524 + ip_set_id_t index = IP_SET_INVALID_ID;
2525 + int (*adtfn)(ip_set_id_t index,
2526 + const void *data, size_t size);
2527 + struct fn_table {
2528 + int (*fn)(ip_set_id_t index,
2529 + const void *data, size_t size);
2530 + } adtfn_table[] =
2531 + { { ip_set_addip }, { ip_set_delip }, { ip_set_testip},
2532 + { ip_set_bindip}, { ip_set_unbindip }, { ip_set_testbind },
2533 + };
2534 +
2535 + DP("optval=%d, user=%p, len=%d", optval, user, len);
2536 + if (!capable(CAP_NET_ADMIN))
2537 + return -EPERM;
2538 + if (optval != SO_IP_SET)
2539 + return -EBADF;
2540 + if (len <= sizeof(unsigned)) {
2541 + ip_set_printk("short userdata (want >%zu, got %u)",
2542 + sizeof(unsigned), len);
2543 + return -EINVAL;
2544 + }
2545 + data = vmalloc(len);
2546 + if (!data) {
2547 + DP("out of mem for %u bytes", len);
2548 + return -ENOMEM;
2549 + }
2550 + if (copy_from_user(data, user, len) != 0) {
2551 + res = -EFAULT;
2552 + goto done;
2553 + }
2554 + if (down_interruptible(&ip_set_app_mutex)) {
2555 + res = -EINTR;
2556 + goto done;
2557 + }
2558 +
2559 + op = (unsigned *)data;
2560 + DP("op=%x", *op);
2561 +
2562 + if (*op < IP_SET_OP_VERSION) {
2563 + /* Check the version at the beginning of operations */
2564 + struct ip_set_req_version *req_version =
2565 + (struct ip_set_req_version *) data;
2566 + if (req_version->version != IP_SET_PROTOCOL_VERSION) {
2567 + res = -EPROTO;
2568 + goto done;
2569 + }
2570 + }
2571 +
2572 + switch (*op) {
2573 + case IP_SET_OP_CREATE:{
2574 + struct ip_set_req_create *req_create
2575 + = (struct ip_set_req_create *) data;
2576 +
2577 + if (len < sizeof(struct ip_set_req_create)) {
2578 + ip_set_printk("short CREATE data (want >=%zu, got %u)",
2579 + sizeof(struct ip_set_req_create), len);
2580 + res = -EINVAL;
2581 + goto done;
2582 + }
2583 + req_create->name[IP_SET_MAXNAMELEN - 1] = '\0';
2584 + req_create->typename[IP_SET_MAXNAMELEN - 1] = '\0';
2585 + res = ip_set_create(req_create->name,
2586 + req_create->typename,
2587 + IP_SET_INVALID_ID,
2588 + data + sizeof(struct ip_set_req_create),
2589 + len - sizeof(struct ip_set_req_create));
2590 + goto done;
2591 + }
2592 + case IP_SET_OP_DESTROY:{
2593 + struct ip_set_req_std *req_destroy
2594 + = (struct ip_set_req_std *) data;
2595 +
2596 + if (len != sizeof(struct ip_set_req_std)) {
2597 + ip_set_printk("invalid DESTROY data (want %zu, got %u)",
2598 + sizeof(struct ip_set_req_std), len);
2599 + res = -EINVAL;
2600 + goto done;
2601 + }
2602 + if (strcmp(req_destroy->name, IPSET_TOKEN_ALL) == 0) {
2603 + /* Destroy all sets */
2604 + index = IP_SET_INVALID_ID;
2605 + } else {
2606 + req_destroy->name[IP_SET_MAXNAMELEN - 1] = '\0';
2607 + index = ip_set_find_byname(req_destroy->name);
2608 +
2609 + if (index == IP_SET_INVALID_ID) {
2610 + res = -ENOENT;
2611 + goto done;
2612 + }
2613 + }
2614 +
2615 + res = ip_set_destroy(index);
2616 + goto done;
2617 + }
2618 + case IP_SET_OP_FLUSH:{
2619 + struct ip_set_req_std *req_flush =
2620 + (struct ip_set_req_std *) data;
2621 +
2622 + if (len != sizeof(struct ip_set_req_std)) {
2623 + ip_set_printk("invalid FLUSH data (want %zu, got %u)",
2624 + sizeof(struct ip_set_req_std), len);
2625 + res = -EINVAL;
2626 + goto done;
2627 + }
2628 + if (strcmp(req_flush->name, IPSET_TOKEN_ALL) == 0) {
2629 + /* Flush all sets */
2630 + index = IP_SET_INVALID_ID;
2631 + } else {
2632 + req_flush->name[IP_SET_MAXNAMELEN - 1] = '\0';
2633 + index = ip_set_find_byname(req_flush->name);
2634 +
2635 + if (index == IP_SET_INVALID_ID) {
2636 + res = -ENOENT;
2637 + goto done;
2638 + }
2639 + }
2640 + res = ip_set_flush(index);
2641 + goto done;
2642 + }
2643 + case IP_SET_OP_RENAME:{
2644 + struct ip_set_req_create *req_rename
2645 + = (struct ip_set_req_create *) data;
2646 +
2647 + if (len != sizeof(struct ip_set_req_create)) {
2648 + ip_set_printk("invalid RENAME data (want %zu, got %u)",
2649 + sizeof(struct ip_set_req_create), len);
2650 + res = -EINVAL;
2651 + goto done;
2652 + }
2653 +
2654 + req_rename->name[IP_SET_MAXNAMELEN - 1] = '\0';
2655 + req_rename->typename[IP_SET_MAXNAMELEN - 1] = '\0';
2656 +
2657 + index = ip_set_find_byname(req_rename->name);
2658 + if (index == IP_SET_INVALID_ID) {
2659 + res = -ENOENT;
2660 + goto done;
2661 + }
2662 + res = ip_set_rename(index, req_rename->typename);
2663 + goto done;
2664 + }
2665 + case IP_SET_OP_SWAP:{
2666 + struct ip_set_req_create *req_swap
2667 + = (struct ip_set_req_create *) data;
2668 + ip_set_id_t to_index;
2669 +
2670 + if (len != sizeof(struct ip_set_req_create)) {
2671 + ip_set_printk("invalid SWAP data (want %zu, got %u)",
2672 + sizeof(struct ip_set_req_create), len);
2673 + res = -EINVAL;
2674 + goto done;
2675 + }
2676 +
2677 + req_swap->name[IP_SET_MAXNAMELEN - 1] = '\0';
2678 + req_swap->typename[IP_SET_MAXNAMELEN - 1] = '\0';
2679 +
2680 + index = ip_set_find_byname(req_swap->name);
2681 + if (index == IP_SET_INVALID_ID) {
2682 + res = -ENOENT;
2683 + goto done;
2684 + }
2685 + to_index = ip_set_find_byname(req_swap->typename);
2686 + if (to_index == IP_SET_INVALID_ID) {
2687 + res = -ENOENT;
2688 + goto done;
2689 + }
2690 + res = ip_set_swap(index, to_index);
2691 + goto done;
2692 + }
2693 + default:
2694 + break; /* Set identified by id */
2695 + }
2696 +
2697 + /* There we may have add/del/test/bind/unbind/test_bind operations */
2698 + if (*op < IP_SET_OP_ADD_IP || *op > IP_SET_OP_TEST_BIND_SET) {
2699 + res = -EBADMSG;
2700 + goto done;
2701 + }
2702 + adtfn = adtfn_table[*op - IP_SET_OP_ADD_IP].fn;
2703 +
2704 + if (len < sizeof(struct ip_set_req_adt)) {
2705 + ip_set_printk("short data in adt request (want >=%zu, got %u)",
2706 + sizeof(struct ip_set_req_adt), len);
2707 + res = -EINVAL;
2708 + goto done;
2709 + }
2710 + req_adt = (struct ip_set_req_adt *) data;
2711 +
2712 + /* -U :all: :all:|:default: uses IP_SET_INVALID_ID */
2713 + if (!(*op == IP_SET_OP_UNBIND_SET
2714 + && req_adt->index == IP_SET_INVALID_ID)) {
2715 + index = ip_set_find_byindex(req_adt->index);
2716 + if (index == IP_SET_INVALID_ID) {
2717 + res = -ENOENT;
2718 + goto done;
2719 + }
2720 + }
2721 + res = adtfn(index, data, len);
2722 +
2723 + done:
2724 + up(&ip_set_app_mutex);
2725 + vfree(data);
2726 + if (res > 0)
2727 + res = 0;
2728 + DP("final result %d", res);
2729 + return res;
2730 +}
2731 +
2732 +static int
2733 +ip_set_sockfn_get(struct sock *sk, int optval, void *user, int *len)
2734 +{
2735 + int res = 0;
2736 + unsigned *op;
2737 + ip_set_id_t index = IP_SET_INVALID_ID;
2738 + void *data;
2739 + int copylen = *len;
2740 +
2741 + DP("optval=%d, user=%p, len=%d", optval, user, *len);
2742 + if (!capable(CAP_NET_ADMIN))
2743 + return -EPERM;
2744 + if (optval != SO_IP_SET)
2745 + return -EBADF;
2746 + if (*len < sizeof(unsigned)) {
2747 + ip_set_printk("short userdata (want >=%zu, got %d)",
2748 + sizeof(unsigned), *len);
2749 + return -EINVAL;
2750 + }
2751 + data = vmalloc(*len);
2752 + if (!data) {
2753 + DP("out of mem for %d bytes", *len);
2754 + return -ENOMEM;
2755 + }
2756 + if (copy_from_user(data, user, *len) != 0) {
2757 + res = -EFAULT;
2758 + goto done;
2759 + }
2760 + if (down_interruptible(&ip_set_app_mutex)) {
2761 + res = -EINTR;
2762 + goto done;
2763 + }
2764 +
2765 + op = (unsigned *) data;
2766 + DP("op=%x", *op);
2767 +
2768 + if (*op < IP_SET_OP_VERSION) {
2769 + /* Check the version at the beginning of operations */
2770 + struct ip_set_req_version *req_version =
2771 + (struct ip_set_req_version *) data;
2772 + if (req_version->version != IP_SET_PROTOCOL_VERSION) {
2773 + res = -EPROTO;
2774 + goto done;
2775 + }
2776 + }
2777 +
2778 + switch (*op) {
2779 + case IP_SET_OP_VERSION: {
2780 + struct ip_set_req_version *req_version =
2781 + (struct ip_set_req_version *) data;
2782 +
2783 + if (*len != sizeof(struct ip_set_req_version)) {
2784 + ip_set_printk("invalid VERSION (want %zu, got %d)",
2785 + sizeof(struct ip_set_req_version),
2786 + *len);
2787 + res = -EINVAL;
2788 + goto done;
2789 + }
2790 +
2791 + req_version->version = IP_SET_PROTOCOL_VERSION;
2792 + res = copy_to_user(user, req_version,
2793 + sizeof(struct ip_set_req_version));
2794 + goto done;
2795 + }
2796 + case IP_SET_OP_GET_BYNAME: {
2797 + struct ip_set_req_get_set *req_get
2798 + = (struct ip_set_req_get_set *) data;
2799 +
2800 + if (*len != sizeof(struct ip_set_req_get_set)) {
2801 + ip_set_printk("invalid GET_BYNAME (want %zu, got %d)",
2802 + sizeof(struct ip_set_req_get_set), *len);
2803 + res = -EINVAL;
2804 + goto done;
2805 + }
2806 + req_get->set.name[IP_SET_MAXNAMELEN - 1] = '\0';
2807 + index = ip_set_find_byname(req_get->set.name);
2808 + req_get->set.index = index;
2809 + goto copy;
2810 + }
2811 + case IP_SET_OP_GET_BYINDEX: {
2812 + struct ip_set_req_get_set *req_get
2813 + = (struct ip_set_req_get_set *) data;
2814 +
2815 + if (*len != sizeof(struct ip_set_req_get_set)) {
2816 + ip_set_printk("invalid GET_BYINDEX (want %zu, got %d)",
2817 + sizeof(struct ip_set_req_get_set), *len);
2818 + res = -EINVAL;
2819 + goto done;
2820 + }
2821 + req_get->set.name[IP_SET_MAXNAMELEN - 1] = '\0';
2822 + index = ip_set_find_byindex(req_get->set.index);
2823 + strncpy(req_get->set.name,
2824 + index == IP_SET_INVALID_ID ? ""
2825 + : ip_set_list[index]->name, IP_SET_MAXNAMELEN);
2826 + goto copy;
2827 + }
2828 + case IP_SET_OP_ADT_GET: {
2829 + struct ip_set_req_adt_get *req_get
2830 + = (struct ip_set_req_adt_get *) data;
2831 +
2832 + if (*len != sizeof(struct ip_set_req_adt_get)) {
2833 + ip_set_printk("invalid ADT_GET (want %zu, got %d)",
2834 + sizeof(struct ip_set_req_adt_get), *len);
2835 + res = -EINVAL;
2836 + goto done;
2837 + }
2838 + req_get->set.name[IP_SET_MAXNAMELEN - 1] = '\0';
2839 + index = ip_set_find_byname(req_get->set.name);
2840 + if (index != IP_SET_INVALID_ID) {
2841 + req_get->set.index = index;
2842 + strncpy(req_get->typename,
2843 + ip_set_list[index]->type->typename,
2844 + IP_SET_MAXNAMELEN - 1);
2845 + } else {
2846 + res = -ENOENT;
2847 + goto done;
2848 + }
2849 + goto copy;
2850 + }
2851 + case IP_SET_OP_MAX_SETS: {
2852 + struct ip_set_req_max_sets *req_max_sets
2853 + = (struct ip_set_req_max_sets *) data;
2854 + ip_set_id_t i;
2855 +
2856 + if (*len != sizeof(struct ip_set_req_max_sets)) {
2857 + ip_set_printk("invalid MAX_SETS (want %zu, got %d)",
2858 + sizeof(struct ip_set_req_max_sets), *len);
2859 + res = -EINVAL;
2860 + goto done;
2861 + }
2862 +
2863 + if (strcmp(req_max_sets->set.name, IPSET_TOKEN_ALL) == 0) {
2864 + req_max_sets->set.index = IP_SET_INVALID_ID;
2865 + } else {
2866 + req_max_sets->set.name[IP_SET_MAXNAMELEN - 1] = '\0';
2867 + req_max_sets->set.index =
2868 + ip_set_find_byname(req_max_sets->set.name);
2869 + if (req_max_sets->set.index == IP_SET_INVALID_ID) {
2870 + res = -ENOENT;
2871 + goto done;
2872 + }
2873 + }
2874 + req_max_sets->max_sets = ip_set_max;
2875 + req_max_sets->sets = 0;
2876 + for (i = 0; i < ip_set_max; i++) {
2877 + if (ip_set_list[i] != NULL)
2878 + req_max_sets->sets++;
2879 + }
2880 + goto copy;
2881 + }
2882 + case IP_SET_OP_LIST_SIZE:
2883 + case IP_SET_OP_SAVE_SIZE: {
2884 + struct ip_set_req_setnames *req_setnames
2885 + = (struct ip_set_req_setnames *) data;
2886 + struct ip_set_name_list *name_list;
2887 + struct ip_set *set;
2888 + ip_set_id_t i;
2889 + int used;
2890 +
2891 + if (*len < sizeof(struct ip_set_req_setnames)) {
2892 + ip_set_printk("short LIST_SIZE (want >=%zu, got %d)",
2893 + sizeof(struct ip_set_req_setnames), *len);
2894 + res = -EINVAL;
2895 + goto done;
2896 + }
2897 +
2898 + req_setnames->size = 0;
2899 + used = sizeof(struct ip_set_req_setnames);
2900 + for (i = 0; i < ip_set_max; i++) {
2901 + if (ip_set_list[i] == NULL)
2902 + continue;
2903 + name_list = (struct ip_set_name_list *)
2904 + (data + used);
2905 + used += sizeof(struct ip_set_name_list);
2906 + if (used > copylen) {
2907 + res = -EAGAIN;
2908 + goto done;
2909 + }
2910 + set = ip_set_list[i];
2911 + /* Fill in index, name, etc. */
2912 + name_list->index = i;
2913 + name_list->id = set->id;
2914 + strncpy(name_list->name,
2915 + set->name,
2916 + IP_SET_MAXNAMELEN - 1);
2917 + strncpy(name_list->typename,
2918 + set->type->typename,
2919 + IP_SET_MAXNAMELEN - 1);
2920 + DP("filled %s of type %s, index %u\n",
2921 + name_list->name, name_list->typename,
2922 + name_list->index);
2923 + if (!(req_setnames->index == IP_SET_INVALID_ID
2924 + || req_setnames->index == i))
2925 + continue;
2926 + /* Update size */
2927 + switch (*op) {
2928 + case IP_SET_OP_LIST_SIZE: {
2929 + req_setnames->size += sizeof(struct ip_set_list)
2930 + + set->type->header_size
2931 + + set->type->list_members_size(set);
2932 + /* Sets are identified by id in the hash */
2933 + FOREACH_HASH_DO(__set_hash_bindings_size_list,
2934 + set->id, &req_setnames->size);
2935 + break;
2936 + }
2937 + case IP_SET_OP_SAVE_SIZE: {
2938 + req_setnames->size += sizeof(struct ip_set_save)
2939 + + set->type->header_size
2940 + + set->type->list_members_size(set);
2941 + FOREACH_HASH_DO(__set_hash_bindings_size_save,
2942 + set->id, &req_setnames->size);
2943 + break;
2944 + }
2945 + default:
2946 + break;
2947 + }
2948 + }
2949 + if (copylen != used) {
2950 + res = -EAGAIN;
2951 + goto done;
2952 + }
2953 + goto copy;
2954 + }
2955 + case IP_SET_OP_LIST: {
2956 + struct ip_set_req_list *req_list
2957 + = (struct ip_set_req_list *) data;
2958 + ip_set_id_t i;
2959 + int used;
2960 +
2961 + if (*len < sizeof(struct ip_set_req_list)) {
2962 + ip_set_printk("short LIST (want >=%zu, got %d)",
2963 + sizeof(struct ip_set_req_list), *len);
2964 + res = -EINVAL;
2965 + goto done;
2966 + }
2967 + index = req_list->index;
2968 + if (index != IP_SET_INVALID_ID
2969 + && ip_set_find_byindex(index) != index) {
2970 + res = -ENOENT;
2971 + goto done;
2972 + }
2973 + used = 0;
2974 + if (index == IP_SET_INVALID_ID) {
2975 + /* List all sets */
2976 + for (i = 0; i < ip_set_max && res == 0; i++) {
2977 + if (ip_set_list[i] != NULL)
2978 + res = ip_set_list_set(i, data, &used, *len);
2979 + }
2980 + } else {
2981 + /* List an individual set */
2982 + res = ip_set_list_set(index, data, &used, *len);
2983 + }
2984 + if (res != 0)
2985 + goto done;
2986 + else if (copylen != used) {
2987 + res = -EAGAIN;
2988 + goto done;
2989 + }
2990 + goto copy;
2991 + }
2992 + case IP_SET_OP_SAVE: {
2993 + struct ip_set_req_list *req_save
2994 + = (struct ip_set_req_list *) data;
2995 + ip_set_id_t i;
2996 + int used;
2997 +
2998 + if (*len < sizeof(struct ip_set_req_list)) {
2999 + ip_set_printk("short SAVE (want >=%zu, got %d)",
3000 + sizeof(struct ip_set_req_list), *len);
3001 + res = -EINVAL;
3002 + goto done;
3003 + }
3004 + index = req_save->index;
3005 + if (index != IP_SET_INVALID_ID
3006 + && ip_set_find_byindex(index) != index) {
3007 + res = -ENOENT;
3008 + goto done;
3009 + }
3010 + used = 0;
3011 + if (index == IP_SET_INVALID_ID) {
3012 + /* Save all sets */
3013 + for (i = 0; i < ip_set_max && res == 0; i++) {
3014 + if (ip_set_list[i] != NULL)
3015 + res = ip_set_save_set(i, data, &used, *len);
3016 + }
3017 + } else {
3018 + /* Save an individual set */
3019 + res = ip_set_save_set(index, data, &used, *len);
3020 + }
3021 + if (res == 0)
3022 + res = ip_set_save_bindings(index, data, &used, *len);
3023 +
3024 + if (res != 0)
3025 + goto done;
3026 + else if (copylen != used) {
3027 + res = -EAGAIN;
3028 + goto done;
3029 + }
3030 + goto copy;
3031 + }
3032 + case IP_SET_OP_RESTORE: {
3033 + struct ip_set_req_setnames *req_restore
3034 + = (struct ip_set_req_setnames *) data;
3035 + int line;
3036 +
3037 + if (*len < sizeof(struct ip_set_req_setnames)
3038 + || *len != req_restore->size) {
3039 + ip_set_printk("invalid RESTORE (want =%zu, got %d)",
3040 + req_restore->size, *len);
3041 + res = -EINVAL;
3042 + goto done;
3043 + }
3044 + line = ip_set_restore(data + sizeof(struct ip_set_req_setnames),
3045 + req_restore->size - sizeof(struct ip_set_req_setnames));
3046 + DP("ip_set_restore: %u", line);
3047 + if (line != 0) {
3048 + res = -EAGAIN;
3049 + req_restore->size = line;
3050 + copylen = sizeof(struct ip_set_req_setnames);
3051 + goto copy;
3052 + }
3053 + goto done;
3054 + }
3055 + default:
3056 + res = -EBADMSG;
3057 + goto done;
3058 + } /* end of switch(op) */
3059 +
3060 + copy:
3061 + DP("set %s, copylen %u", index != IP_SET_INVALID_ID
3062 + && ip_set_list[index]
3063 + ? ip_set_list[index]->name
3064 + : ":all:", copylen);
3065 + res = copy_to_user(user, data, copylen);
3066 +
3067 + done:
3068 + up(&ip_set_app_mutex);
3069 + vfree(data);
3070 + if (res > 0)
3071 + res = 0;
3072 + DP("final result %d", res);
3073 + return res;
3074 +}
3075 +
3076 +static struct nf_sockopt_ops so_set = {
3077 + .pf = PF_INET,
3078 + .set_optmin = SO_IP_SET,
3079 + .set_optmax = SO_IP_SET + 1,
3080 + .set = &ip_set_sockfn_set,
3081 + .get_optmin = SO_IP_SET,
3082 + .get_optmax = SO_IP_SET + 1,
3083 + .get = &ip_set_sockfn_get,
3084 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23)
3085 + .owner = THIS_MODULE,
3086 +#endif
3087 +};
3088 +
3089 +static int max_sets, hash_size;
3090 +module_param(max_sets, int, 0600);
3091 +MODULE_PARM_DESC(max_sets, "maximal number of sets");
3092 +module_param(hash_size, int, 0600);
3093 +MODULE_PARM_DESC(hash_size, "hash size for bindings");
3094 +MODULE_LICENSE("GPL");
3095 +MODULE_AUTHOR("Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>");
3096 +MODULE_DESCRIPTION("module implementing core IP set support");
3097 +
3098 +static int __init ip_set_init(void)
3099 +{
3100 + int res;
3101 + ip_set_id_t i;
3102 +
3103 + get_random_bytes(&ip_set_hash_random, 4);
3104 + if (max_sets)
3105 + ip_set_max = max_sets;
3106 + ip_set_list = vmalloc(sizeof(struct ip_set *) * ip_set_max);
3107 + if (!ip_set_list) {
3108 + printk(KERN_ERR "Unable to create ip_set_list\n");
3109 + return -ENOMEM;
3110 + }
3111 + memset(ip_set_list, 0, sizeof(struct ip_set *) * ip_set_max);
3112 + if (hash_size)
3113 + ip_set_bindings_hash_size = hash_size;
3114 + ip_set_hash = vmalloc(sizeof(struct list_head) * ip_set_bindings_hash_size);
3115 + if (!ip_set_hash) {
3116 + printk(KERN_ERR "Unable to create ip_set_hash\n");
3117 + vfree(ip_set_list);
3118 + return -ENOMEM;
3119 + }
3120 + for (i = 0; i < ip_set_bindings_hash_size; i++)
3121 + INIT_LIST_HEAD(&ip_set_hash[i]);
3122 +
3123 + INIT_LIST_HEAD(&set_type_list);
3124 +
3125 + res = nf_register_sockopt(&so_set);
3126 + if (res != 0) {
3127 + ip_set_printk("SO_SET registry failed: %d", res);
3128 + vfree(ip_set_list);
3129 + vfree(ip_set_hash);
3130 + return res;
3131 + }
3132 + return 0;
3133 +}
3134 +
3135 +static void __exit ip_set_fini(void)
3136 +{
3137 + /* There can't be any existing set or binding */
3138 + nf_unregister_sockopt(&so_set);
3139 + vfree(ip_set_list);
3140 + vfree(ip_set_hash);
3141 + DP("these are the famous last words");
3142 +}
3143 +
3144 +EXPORT_SYMBOL(ip_set_register_set_type);
3145 +EXPORT_SYMBOL(ip_set_unregister_set_type);
3146 +
3147 +EXPORT_SYMBOL(ip_set_get_byname);
3148 +EXPORT_SYMBOL(ip_set_get_byindex);
3149 +EXPORT_SYMBOL(ip_set_put);
3150 +
3151 +EXPORT_SYMBOL(ip_set_addip_kernel);
3152 +EXPORT_SYMBOL(ip_set_delip_kernel);
3153 +EXPORT_SYMBOL(ip_set_testip_kernel);
3154 +
3155 +module_init(ip_set_init);
3156 +module_exit(ip_set_fini);
3157 diff -Nru linux-2.6.23/net/ipv4/netfilter/ip_set_iphash.c linux-2.6.23.pom2patch.set/net/ipv4/netfilter/ip_set_iphash.c
3158 --- linux-2.6.23/net/ipv4/netfilter/ip_set_iphash.c 1970-01-01 01:00:00.000000000 +0100
3159 +++ linux-2.6.23.pom2patch.set/net/ipv4/netfilter/ip_set_iphash.c 2007-10-12 11:52:37.000000000 +0200
3160 @@ -0,0 +1,429 @@
3161 +/* Copyright (C) 2003-2004 Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
3162 + *
3163 + * This program is free software; you can redistribute it and/or modify
3164 + * it under the terms of the GNU General Public License version 2 as
3165 + * published by the Free Software Foundation.
3166 + */
3167 +
3168 +/* Kernel module implementing an ip hash set */
3169 +
3170 +#include <linux/module.h>
3171 +#include <linux/ip.h>
3172 +#include <linux/skbuff.h>
3173 +#include <linux/version.h>
3174 +#include <linux/jhash.h>
3175 +#include <linux/netfilter_ipv4/ip_tables.h>
3176 +#include <linux/netfilter_ipv4/ip_set.h>
3177 +#include <linux/errno.h>
3178 +#include <asm/uaccess.h>
3179 +#include <asm/bitops.h>
3180 +#include <linux/spinlock.h>
3181 +#include <linux/vmalloc.h>
3182 +#include <linux/random.h>
3183 +
3184 +#include <net/ip.h>
3185 +
3186 +#include <linux/netfilter_ipv4/ip_set_malloc.h>
3187 +#include <linux/netfilter_ipv4/ip_set_iphash.h>
3188 +
3189 +static int limit = MAX_RANGE;
3190 +
3191 +static inline __u32
3192 +jhash_ip(const struct ip_set_iphash *map, uint16_t i, ip_set_ip_t ip)
3193 +{
3194 + return jhash_1word(ip, *(((uint32_t *) map->initval) + i));
3195 +}
3196 +
3197 +static inline __u32
3198 +hash_id(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip)
3199 +{
3200 + struct ip_set_iphash *map = (struct ip_set_iphash *) set->data;
3201 + __u32 id;
3202 + u_int16_t i;
3203 + ip_set_ip_t *elem;
3204 +
3205 + *hash_ip = ip & map->netmask;
3206 + DP("set: %s, ip:%u.%u.%u.%u, %u.%u.%u.%u, %u.%u.%u.%u",
3207 + set->name, HIPQUAD(ip), HIPQUAD(*hash_ip), HIPQUAD(map->netmask));
3208 +
3209 + for (i = 0; i < map->probes; i++) {
3210 + id = jhash_ip(map, i, *hash_ip) % map->hashsize;
3211 + DP("hash key: %u", id);
3212 + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id);
3213 + if (*elem == *hash_ip)
3214 + return id;
3215 + /* No shortcut at testing - there can be deleted
3216 + * entries. */
3217 + }
3218 + return UINT_MAX;
3219 +}
3220 +
3221 +static inline int
3222 +__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip)
3223 +{
3224 + return (ip && hash_id(set, ip, hash_ip) != UINT_MAX);
3225 +}
3226 +
3227 +static int
3228 +testip(struct ip_set *set, const void *data, size_t size,
3229 + ip_set_ip_t *hash_ip)
3230 +{
3231 + struct ip_set_req_iphash *req =
3232 + (struct ip_set_req_iphash *) data;
3233 +
3234 + if (size != sizeof(struct ip_set_req_iphash)) {
3235 + ip_set_printk("data length wrong (want %zu, have %zu)",
3236 + sizeof(struct ip_set_req_iphash),
3237 + size);
3238 + return -EINVAL;
3239 + }
3240 + return __testip(set, req->ip, hash_ip);
3241 +}
3242 +
3243 +static int
3244 +testip_kernel(struct ip_set *set,
3245 + const struct sk_buff *skb,
3246 + ip_set_ip_t *hash_ip,
3247 + const u_int32_t *flags,
3248 + unsigned char index)
3249 +{
3250 + return __testip(set,
3251 + ntohl(flags[index] & IPSET_SRC
3252 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22)
3253 + ? ip_hdr(skb)->saddr
3254 + : ip_hdr(skb)->daddr),
3255 +#else
3256 + ? skb->nh.iph->saddr
3257 + : skb->nh.iph->daddr),
3258 +#endif
3259 + hash_ip);
3260 +}
3261 +
3262 +static inline int
3263 +__addip(struct ip_set_iphash *map, ip_set_ip_t ip, ip_set_ip_t *hash_ip)
3264 +{
3265 + __u32 probe;
3266 + u_int16_t i;
3267 + ip_set_ip_t *elem;
3268 +
3269 + if (!ip || map->elements >= limit)
3270 + return -ERANGE;
3271 +
3272 + *hash_ip = ip & map->netmask;
3273 +
3274 + for (i = 0; i < map->probes; i++) {
3275 + probe = jhash_ip(map, i, *hash_ip) % map->hashsize;
3276 + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, probe);
3277 + if (*elem == *hash_ip)
3278 + return -EEXIST;
3279 + if (!*elem) {
3280 + *elem = *hash_ip;
3281 + map->elements++;
3282 + return 0;
3283 + }
3284 + }
3285 + /* Trigger rehashing */
3286 + return -EAGAIN;
3287 +}
3288 +
3289 +static int
3290 +addip(struct ip_set *set, const void *data, size_t size,
3291 + ip_set_ip_t *hash_ip)
3292 +{
3293 + struct ip_set_req_iphash *req =
3294 + (struct ip_set_req_iphash *) data;
3295 +
3296 + if (size != sizeof(struct ip_set_req_iphash)) {
3297 + ip_set_printk("data length wrong (want %zu, have %zu)",
3298 + sizeof(struct ip_set_req_iphash),
3299 + size);
3300 + return -EINVAL;
3301 + }
3302 + return __addip((struct ip_set_iphash *) set->data, req->ip, hash_ip);
3303 +}
3304 +
3305 +static int
3306 +addip_kernel(struct ip_set *set,
3307 + const struct sk_buff *skb,
3308 + ip_set_ip_t *hash_ip,
3309 + const u_int32_t *flags,
3310 + unsigned char index)
3311 +{
3312 + return __addip((struct ip_set_iphash *) set->data,
3313 + ntohl(flags[index] & IPSET_SRC
3314 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22)
3315 + ? ip_hdr(skb)->saddr
3316 + : ip_hdr(skb)->daddr),
3317 +#else
3318 + ? skb->nh.iph->saddr
3319 + : skb->nh.iph->daddr),
3320 +#endif
3321 + hash_ip);
3322 +}
3323 +
3324 +static int retry(struct ip_set *set)
3325 +{
3326 + struct ip_set_iphash *map = (struct ip_set_iphash *) set->data;
3327 + ip_set_ip_t hash_ip, *elem;
3328 + void *members;
3329 + u_int32_t i, hashsize = map->hashsize;
3330 + int res;
3331 + struct ip_set_iphash *tmp;
3332 +
3333 + if (map->resize == 0)
3334 + return -ERANGE;
3335 +
3336 + again:
3337 + res = 0;
3338 +
3339 + /* Calculate new hash size */
3340 + hashsize += (hashsize * map->resize)/100;
3341 + if (hashsize == map->hashsize)
3342 + hashsize++;
3343 +
3344 + ip_set_printk("rehashing of set %s triggered: "
3345 + "hashsize grows from %u to %u",
3346 + set->name, map->hashsize, hashsize);
3347 +
3348 + tmp = kmalloc(sizeof(struct ip_set_iphash)
3349 + + map->probes * sizeof(uint32_t), GFP_ATOMIC);
3350 + if (!tmp) {
3351 + DP("out of memory for %d bytes",
3352 + sizeof(struct ip_set_iphash)
3353 + + map->probes * sizeof(uint32_t));
3354 + return -ENOMEM;
3355 + }
3356 + tmp->members = harray_malloc(hashsize, sizeof(ip_set_ip_t), GFP_ATOMIC);
3357 + if (!tmp->members) {
3358 + DP("out of memory for %d bytes", hashsize * sizeof(ip_set_ip_t));
3359 + kfree(tmp);
3360 + return -ENOMEM;
3361 + }
3362 + tmp->hashsize = hashsize;
3363 + tmp->elements = 0;
3364 + tmp->probes = map->probes;
3365 + tmp->resize = map->resize;
3366 + tmp->netmask = map->netmask;
3367 + memcpy(tmp->initval, map->initval, map->probes * sizeof(uint32_t));
3368 +
3369 + write_lock_bh(&set->lock);
3370 + map = (struct ip_set_iphash *) set->data; /* Play safe */
3371 + for (i = 0; i < map->hashsize && res == 0; i++) {
3372 + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i);
3373 + if (*elem)
3374 + res = __addip(tmp, *elem, &hash_ip);
3375 + }
3376 + if (res) {
3377 + /* Failure, try again */
3378 + write_unlock_bh(&set->lock);
3379 + harray_free(tmp->members);
3380 + kfree(tmp);
3381 + goto again;
3382 + }
3383 +
3384 + /* Success at resizing! */
3385 + members = map->members;
3386 +
3387 + map->hashsize = tmp->hashsize;
3388 + map->members = tmp->members;
3389 + write_unlock_bh(&set->lock);
3390 +
3391 + harray_free(members);
3392 + kfree(tmp);
3393 +
3394 + return 0;
3395 +}
3396 +
3397 +static inline int
3398 +__delip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip)
3399 +{
3400 + struct ip_set_iphash *map = (struct ip_set_iphash *) set->data;
3401 + ip_set_ip_t id, *elem;
3402 +
3403 + if (!ip)
3404 + return -ERANGE;
3405 +
3406 + id = hash_id(set, ip, hash_ip);