diff options
author | Harald Welte <laforge@netfilter.org> | 2006-01-12 16:30:04 -0500 |
---|---|---|
committer | David S. Miller <davem@sunset.davemloft.net> | 2006-01-12 17:06:43 -0500 |
commit | 2e4e6a17af35be359cc8f1c924f8f198fbd478cc (patch) | |
tree | cb4b5438dcf9ff9d57518a26124308bcbfffd214 /net/ipv4/netfilter/ip_tables.c | |
parent | 880b005f294454d989783d0984dc554dfe3c8214 (diff) |
[NETFILTER] x_tables: Abstraction layer for {ip,ip6,arp}_tables
This monster-patch tries to do the best job for unifying the data
structures and backend interfaces for the three evil clones ip_tables,
ip6_tables and arp_tables. In an ideal world we would never have
allowed this kind of copy+paste programming... but well, our world
isn't (yet?) ideal.
o introduce a new x_tables module
o {ip,arp,ip6}_tables depend on this x_tables module
o registration functions for tables, matches and targets are only
wrappers around x_tables provided functions
o all matches/targets that are used from ip_tables and ip6_tables
are now implemented as xt_FOOBAR.c files and provide module aliases
to ipt_FOOBAR and ip6t_FOOBAR
o header files for xt_matches are in include/linux/netfilter/,
include/linux/netfilter_{ipv4,ipv6} contains compatibility wrappers
around the xt_FOOBAR.h headers
Based on this patchset we're going to further unify the code,
gradually getting rid of all the layer 3 specific assumptions.
Signed-off-by: Harald Welte <laforge@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/ipv4/netfilter/ip_tables.c')
-rw-r--r-- | net/ipv4/netfilter/ip_tables.c | 842 |
1 files changed, 95 insertions, 747 deletions
diff --git a/net/ipv4/netfilter/ip_tables.c b/net/ipv4/netfilter/ip_tables.c index 877bc96d333..2371b2062c2 100644 --- a/net/ipv4/netfilter/ip_tables.c +++ b/net/ipv4/netfilter/ip_tables.c | |||
@@ -2,7 +2,7 @@ | |||
2 | * Packet matching code. | 2 | * Packet matching code. |
3 | * | 3 | * |
4 | * Copyright (C) 1999 Paul `Rusty' Russell & Michael J. Neuling | 4 | * Copyright (C) 1999 Paul `Rusty' Russell & Michael J. Neuling |
5 | * Copyright (C) 2000-2004 Netfilter Core Team <coreteam@netfilter.org> | 5 | * Copyright (C) 2000-2005 Netfilter Core Team <coreteam@netfilter.org> |
6 | * | 6 | * |
7 | * This program is free software; you can redistribute it and/or modify | 7 | * This program is free software; you can redistribute it and/or modify |
8 | * it under the terms of the GNU General Public License version 2 as | 8 | * it under the terms of the GNU General Public License version 2 as |
@@ -11,6 +11,8 @@ | |||
11 | * 19 Jan 2002 Harald Welte <laforge@gnumonks.org> | 11 | * 19 Jan 2002 Harald Welte <laforge@gnumonks.org> |
12 | * - increase module usage count as soon as we have rules inside | 12 | * - increase module usage count as soon as we have rules inside |
13 | * a table | 13 | * a table |
14 | * 08 Oct 2005 Harald Welte <lafore@netfilter.org> | ||
15 | * - Generalize into "x_tables" layer and "{ip,ip6,arp}_tables" | ||
14 | */ | 16 | */ |
15 | #include <linux/config.h> | 17 | #include <linux/config.h> |
16 | #include <linux/cache.h> | 18 | #include <linux/cache.h> |
@@ -20,8 +22,6 @@ | |||
20 | #include <linux/vmalloc.h> | 22 | #include <linux/vmalloc.h> |
21 | #include <linux/netdevice.h> | 23 | #include <linux/netdevice.h> |
22 | #include <linux/module.h> | 24 | #include <linux/module.h> |
23 | #include <linux/tcp.h> | ||
24 | #include <linux/udp.h> | ||
25 | #include <linux/icmp.h> | 25 | #include <linux/icmp.h> |
26 | #include <net/ip.h> | 26 | #include <net/ip.h> |
27 | #include <asm/uaccess.h> | 27 | #include <asm/uaccess.h> |
@@ -30,6 +30,7 @@ | |||
30 | #include <linux/err.h> | 30 | #include <linux/err.h> |
31 | #include <linux/cpumask.h> | 31 | #include <linux/cpumask.h> |
32 | 32 | ||
33 | #include <linux/netfilter/x_tables.h> | ||
33 | #include <linux/netfilter_ipv4/ip_tables.h> | 34 | #include <linux/netfilter_ipv4/ip_tables.h> |
34 | 35 | ||
35 | MODULE_LICENSE("GPL"); | 36 | MODULE_LICENSE("GPL"); |
@@ -62,14 +63,6 @@ do { \ | |||
62 | #else | 63 | #else |
63 | #define IP_NF_ASSERT(x) | 64 | #define IP_NF_ASSERT(x) |
64 | #endif | 65 | #endif |
65 | #define SMP_ALIGN(x) (((x) + SMP_CACHE_BYTES-1) & ~(SMP_CACHE_BYTES-1)) | ||
66 | |||
67 | static DECLARE_MUTEX(ipt_mutex); | ||
68 | |||
69 | /* Must have mutex */ | ||
70 | #define ASSERT_READ_LOCK(x) IP_NF_ASSERT(down_trylock(&ipt_mutex) != 0) | ||
71 | #define ASSERT_WRITE_LOCK(x) IP_NF_ASSERT(down_trylock(&ipt_mutex) != 0) | ||
72 | #include <linux/netfilter_ipv4/listhelp.h> | ||
73 | 66 | ||
74 | #if 0 | 67 | #if 0 |
75 | /* All the better to debug you with... */ | 68 | /* All the better to debug you with... */ |
@@ -86,36 +79,6 @@ static DECLARE_MUTEX(ipt_mutex); | |||
86 | 79 | ||
87 | Hence the start of any table is given by get_table() below. */ | 80 | Hence the start of any table is given by get_table() below. */ |
88 | 81 | ||
89 | /* The table itself */ | ||
90 | struct ipt_table_info | ||
91 | { | ||
92 | /* Size per table */ | ||
93 | unsigned int size; | ||
94 | /* Number of entries: FIXME. --RR */ | ||
95 | unsigned int number; | ||
96 | /* Initial number of entries. Needed for module usage count */ | ||
97 | unsigned int initial_entries; | ||
98 | |||
99 | /* Entry points and underflows */ | ||
100 | unsigned int hook_entry[NF_IP_NUMHOOKS]; | ||
101 | unsigned int underflow[NF_IP_NUMHOOKS]; | ||
102 | |||
103 | /* ipt_entry tables: one per CPU */ | ||
104 | void *entries[NR_CPUS]; | ||
105 | }; | ||
106 | |||
107 | static LIST_HEAD(ipt_target); | ||
108 | static LIST_HEAD(ipt_match); | ||
109 | static LIST_HEAD(ipt_tables); | ||
110 | #define SET_COUNTER(c,b,p) do { (c).bcnt = (b); (c).pcnt = (p); } while(0) | ||
111 | #define ADD_COUNTER(c,b,p) do { (c).bcnt += (b); (c).pcnt += (p); } while(0) | ||
112 | |||
113 | #if 0 | ||
114 | #define down(x) do { printk("DOWN:%u:" #x "\n", __LINE__); down(x); } while(0) | ||
115 | #define down_interruptible(x) ({ int __r; printk("DOWNi:%u:" #x "\n", __LINE__); __r = down_interruptible(x); if (__r != 0) printk("ABORT-DOWNi:%u\n", __LINE__); __r; }) | ||
116 | #define up(x) do { printk("UP:%u:" #x "\n", __LINE__); up(x); } while(0) | ||
117 | #endif | ||
118 | |||
119 | /* Returns whether matches rule or not. */ | 82 | /* Returns whether matches rule or not. */ |
120 | static inline int | 83 | static inline int |
121 | ip_packet_match(const struct iphdr *ip, | 84 | ip_packet_match(const struct iphdr *ip, |
@@ -234,7 +197,8 @@ int do_match(struct ipt_entry_match *m, | |||
234 | int *hotdrop) | 197 | int *hotdrop) |
235 | { | 198 | { |
236 | /* Stop iteration if it doesn't match */ | 199 | /* Stop iteration if it doesn't match */ |
237 | if (!m->u.kernel.match->match(skb, in, out, m->data, offset, hotdrop)) | 200 | if (!m->u.kernel.match->match(skb, in, out, m->data, offset, |
201 | skb->nh.iph->ihl*4, hotdrop)) | ||
238 | return 1; | 202 | return 1; |
239 | else | 203 | else |
240 | return 0; | 204 | return 0; |
@@ -265,6 +229,7 @@ ipt_do_table(struct sk_buff **pskb, | |||
265 | const char *indev, *outdev; | 229 | const char *indev, *outdev; |
266 | void *table_base; | 230 | void *table_base; |
267 | struct ipt_entry *e, *back; | 231 | struct ipt_entry *e, *back; |
232 | struct xt_table_info *private = table->private; | ||
268 | 233 | ||
269 | /* Initialization */ | 234 | /* Initialization */ |
270 | ip = (*pskb)->nh.iph; | 235 | ip = (*pskb)->nh.iph; |
@@ -281,24 +246,11 @@ ipt_do_table(struct sk_buff **pskb, | |||
281 | 246 | ||
282 | read_lock_bh(&table->lock); | 247 | read_lock_bh(&table->lock); |
283 | IP_NF_ASSERT(table->valid_hooks & (1 << hook)); | 248 | IP_NF_ASSERT(table->valid_hooks & (1 << hook)); |
284 | table_base = (void *)table->private->entries[smp_processor_id()]; | 249 | table_base = (void *)private->entries[smp_processor_id()]; |
285 | e = get_entry(table_base, table->private->hook_entry[hook]); | 250 | e = get_entry(table_base, private->hook_entry[hook]); |
286 | |||
287 | #ifdef CONFIG_NETFILTER_DEBUG | ||
288 | /* Check noone else using our table */ | ||
289 | if (((struct ipt_entry *)table_base)->comefrom != 0xdead57ac | ||
290 | && ((struct ipt_entry *)table_base)->comefrom != 0xeeeeeeec) { | ||
291 | printk("ASSERT: CPU #%u, %s comefrom(%p) = %X\n", | ||
292 | smp_processor_id(), | ||
293 | table->name, | ||
294 | &((struct ipt_entry *)table_base)->comefrom, | ||
295 | ((struct ipt_entry *)table_base)->comefrom); | ||
296 | } | ||
297 | ((struct ipt_entry *)table_base)->comefrom = 0x57acc001; | ||
298 | #endif | ||
299 | 251 | ||
300 | /* For return from builtin chain */ | 252 | /* For return from builtin chain */ |
301 | back = get_entry(table_base, table->private->underflow[hook]); | 253 | back = get_entry(table_base, private->underflow[hook]); |
302 | 254 | ||
303 | do { | 255 | do { |
304 | IP_NF_ASSERT(e); | 256 | IP_NF_ASSERT(e); |
@@ -384,9 +336,6 @@ ipt_do_table(struct sk_buff **pskb, | |||
384 | } | 336 | } |
385 | } while (!hotdrop); | 337 | } while (!hotdrop); |
386 | 338 | ||
387 | #ifdef CONFIG_NETFILTER_DEBUG | ||
388 | ((struct ipt_entry *)table_base)->comefrom = 0xdead57ac; | ||
389 | #endif | ||
390 | read_unlock_bh(&table->lock); | 339 | read_unlock_bh(&table->lock); |
391 | 340 | ||
392 | #ifdef DEBUG_ALLOW_ALL | 341 | #ifdef DEBUG_ALLOW_ALL |
@@ -398,145 +347,6 @@ ipt_do_table(struct sk_buff **pskb, | |||
398 | #endif | 347 | #endif |
399 | } | 348 | } |
400 | 349 | ||
401 | /* | ||
402 | * These are weird, but module loading must not be done with mutex | ||
403 | * held (since they will register), and we have to have a single | ||
404 | * function to use try_then_request_module(). | ||
405 | */ | ||
406 | |||
407 | /* Find table by name, grabs mutex & ref. Returns ERR_PTR() on error. */ | ||
408 | static inline struct ipt_table *find_table_lock(const char *name) | ||
409 | { | ||
410 | struct ipt_table *t; | ||
411 | |||
412 | if (down_interruptible(&ipt_mutex) != 0) | ||
413 | return ERR_PTR(-EINTR); | ||
414 | |||
415 | list_for_each_entry(t, &ipt_tables, list) | ||
416 | if (strcmp(t->name, name) == 0 && try_module_get(t->me)) | ||
417 | return t; | ||
418 | up(&ipt_mutex); | ||
419 | return NULL; | ||
420 | } | ||
421 | |||
422 | /* Find match, grabs ref. Returns ERR_PTR() on error. */ | ||
423 | static inline struct ipt_match *find_match(const char *name, u8 revision) | ||
424 | { | ||
425 | struct ipt_match *m; | ||
426 | int err = 0; | ||
427 | |||
428 | if (down_interruptible(&ipt_mutex) != 0) | ||
429 | return ERR_PTR(-EINTR); | ||
430 | |||
431 | list_for_each_entry(m, &ipt_match, list) { | ||
432 | if (strcmp(m->name, name) == 0) { | ||
433 | if (m->revision == revision) { | ||
434 | if (try_module_get(m->me)) { | ||
435 | up(&ipt_mutex); | ||
436 | return m; | ||
437 | } | ||
438 | } else | ||
439 | err = -EPROTOTYPE; /* Found something. */ | ||
440 | } | ||
441 | } | ||
442 | up(&ipt_mutex); | ||
443 | return ERR_PTR(err); | ||
444 | } | ||
445 | |||
446 | /* Find target, grabs ref. Returns ERR_PTR() on error. */ | ||
447 | static inline struct ipt_target *find_target(const char *name, u8 revision) | ||
448 | { | ||
449 | struct ipt_target *t; | ||
450 | int err = 0; | ||
451 | |||
452 | if (down_interruptible(&ipt_mutex) != 0) | ||
453 | return ERR_PTR(-EINTR); | ||
454 | |||
455 | list_for_each_entry(t, &ipt_target, list) { | ||
456 | if (strcmp(t->name, name) == 0) { | ||
457 | if (t->revision == revision) { | ||
458 | if (try_module_get(t->me)) { | ||
459 | up(&ipt_mutex); | ||
460 | return t; | ||
461 | } | ||
462 | } else | ||
463 | err = -EPROTOTYPE; /* Found something. */ | ||
464 | } | ||
465 | } | ||
466 | up(&ipt_mutex); | ||
467 | return ERR_PTR(err); | ||
468 | } | ||
469 | |||
470 | struct ipt_target *ipt_find_target(const char *name, u8 revision) | ||
471 | { | ||
472 | struct ipt_target *target; | ||
473 | |||
474 | target = try_then_request_module(find_target(name, revision), | ||
475 | "ipt_%s", name); | ||
476 | if (IS_ERR(target) || !target) | ||
477 | return NULL; | ||
478 | return target; | ||
479 | } | ||
480 | |||
481 | static int match_revfn(const char *name, u8 revision, int *bestp) | ||
482 | { | ||
483 | struct ipt_match *m; | ||
484 | int have_rev = 0; | ||
485 | |||
486 | list_for_each_entry(m, &ipt_match, list) { | ||
487 | if (strcmp(m->name, name) == 0) { | ||
488 | if (m->revision > *bestp) | ||
489 | *bestp = m->revision; | ||
490 | if (m->revision == revision) | ||
491 | have_rev = 1; | ||
492 | } | ||
493 | } | ||
494 | return have_rev; | ||
495 | } | ||
496 | |||
497 | static int target_revfn(const char *name, u8 revision, int *bestp) | ||
498 | { | ||
499 | struct ipt_target *t; | ||
500 | int have_rev = 0; | ||
501 | |||
502 | list_for_each_entry(t, &ipt_target, list) { | ||
503 | if (strcmp(t->name, name) == 0) { | ||
504 | if (t->revision > *bestp) | ||
505 | *bestp = t->revision; | ||
506 | if (t->revision == revision) | ||
507 | have_rev = 1; | ||
508 | } | ||
509 | } | ||
510 | return have_rev; | ||
511 | } | ||
512 | |||
513 | /* Returns true or false (if no such extension at all) */ | ||
514 | static inline int find_revision(const char *name, u8 revision, | ||
515 | int (*revfn)(const char *, u8, int *), | ||
516 | int *err) | ||
517 | { | ||
518 | int have_rev, best = -1; | ||
519 | |||
520 | if (down_interruptible(&ipt_mutex) != 0) { | ||
521 | *err = -EINTR; | ||
522 | return 1; | ||
523 | } | ||
524 | have_rev = revfn(name, revision, &best); | ||
525 | up(&ipt_mutex); | ||
526 | |||
527 | /* Nothing at all? Return 0 to try loading module. */ | ||
528 | if (best == -1) { | ||
529 | *err = -ENOENT; | ||
530 | return 0; | ||
531 | } | ||
532 | |||
533 | *err = best; | ||
534 | if (!have_rev) | ||
535 | *err = -EPROTONOSUPPORT; | ||
536 | return 1; | ||
537 | } | ||
538 | |||
539 | |||
540 | /* All zeroes == unconditional rule. */ | 350 | /* All zeroes == unconditional rule. */ |
541 | static inline int | 351 | static inline int |
542 | unconditional(const struct ipt_ip *ip) | 352 | unconditional(const struct ipt_ip *ip) |
@@ -553,7 +363,7 @@ unconditional(const struct ipt_ip *ip) | |||
553 | /* Figures out from what hook each rule can be called: returns 0 if | 363 | /* Figures out from what hook each rule can be called: returns 0 if |
554 | there are loops. Puts hook bitmask in comefrom. */ | 364 | there are loops. Puts hook bitmask in comefrom. */ |
555 | static int | 365 | static int |
556 | mark_source_chains(struct ipt_table_info *newinfo, | 366 | mark_source_chains(struct xt_table_info *newinfo, |
557 | unsigned int valid_hooks, void *entry0) | 367 | unsigned int valid_hooks, void *entry0) |
558 | { | 368 | { |
559 | unsigned int hook; | 369 | unsigned int hook; |
@@ -699,7 +509,7 @@ check_match(struct ipt_entry_match *m, | |||
699 | { | 509 | { |
700 | struct ipt_match *match; | 510 | struct ipt_match *match; |
701 | 511 | ||
702 | match = try_then_request_module(find_match(m->u.user.name, | 512 | match = try_then_request_module(xt_find_match(AF_INET, m->u.user.name, |
703 | m->u.user.revision), | 513 | m->u.user.revision), |
704 | "ipt_%s", m->u.user.name); | 514 | "ipt_%s", m->u.user.name); |
705 | if (IS_ERR(match) || !match) { | 515 | if (IS_ERR(match) || !match) { |
@@ -744,7 +554,8 @@ check_entry(struct ipt_entry *e, const char *name, unsigned int size, | |||
744 | goto cleanup_matches; | 554 | goto cleanup_matches; |
745 | 555 | ||
746 | t = ipt_get_target(e); | 556 | t = ipt_get_target(e); |
747 | target = try_then_request_module(find_target(t->u.user.name, | 557 | target = try_then_request_module(xt_find_target(AF_INET, |
558 | t->u.user.name, | ||
748 | t->u.user.revision), | 559 | t->u.user.revision), |
749 | "ipt_%s", t->u.user.name); | 560 | "ipt_%s", t->u.user.name); |
750 | if (IS_ERR(target) || !target) { | 561 | if (IS_ERR(target) || !target) { |
@@ -781,7 +592,7 @@ check_entry(struct ipt_entry *e, const char *name, unsigned int size, | |||
781 | 592 | ||
782 | static inline int | 593 | static inline int |
783 | check_entry_size_and_hooks(struct ipt_entry *e, | 594 | check_entry_size_and_hooks(struct ipt_entry *e, |
784 | struct ipt_table_info *newinfo, | 595 | struct xt_table_info *newinfo, |
785 | unsigned char *base, | 596 | unsigned char *base, |
786 | unsigned char *limit, | 597 | unsigned char *limit, |
787 | const unsigned int *hook_entries, | 598 | const unsigned int *hook_entries, |
@@ -815,7 +626,7 @@ check_entry_size_and_hooks(struct ipt_entry *e, | |||
815 | < 0 (not IPT_RETURN). --RR */ | 626 | < 0 (not IPT_RETURN). --RR */ |
816 | 627 | ||
817 | /* Clear counters and comefrom */ | 628 | /* Clear counters and comefrom */ |
818 | e->counters = ((struct ipt_counters) { 0, 0 }); | 629 | e->counters = ((struct xt_counters) { 0, 0 }); |
819 | e->comefrom = 0; | 630 | e->comefrom = 0; |
820 | 631 | ||
821 | (*i)++; | 632 | (*i)++; |
@@ -845,7 +656,7 @@ cleanup_entry(struct ipt_entry *e, unsigned int *i) | |||
845 | static int | 656 | static int |
846 | translate_table(const char *name, | 657 | translate_table(const char *name, |
847 | unsigned int valid_hooks, | 658 | unsigned int valid_hooks, |
848 | struct ipt_table_info *newinfo, | 659 | struct xt_table_info *newinfo, |
849 | void *entry0, | 660 | void *entry0, |
850 | unsigned int size, | 661 | unsigned int size, |
851 | unsigned int number, | 662 | unsigned int number, |
@@ -922,48 +733,10 @@ translate_table(const char *name, | |||
922 | return ret; | 733 | return ret; |
923 | } | 734 | } |
924 | 735 | ||
925 | static struct ipt_table_info * | ||
926 | replace_table(struct ipt_table *table, | ||
927 | unsigned int num_counters, | ||
928 | struct ipt_table_info *newinfo, | ||
929 | int *error) | ||
930 | { | ||
931 | struct ipt_table_info *oldinfo; | ||
932 | |||
933 | #ifdef CONFIG_NETFILTER_DEBUG | ||
934 | { | ||
935 | int cpu; | ||
936 | |||
937 | for_each_cpu(cpu) { | ||
938 | struct ipt_entry *table_base = newinfo->entries[cpu]; | ||
939 | if (table_base) | ||
940 | table_base->comefrom = 0xdead57ac; | ||
941 | } | ||
942 | } | ||
943 | #endif | ||
944 | |||
945 | /* Do the substitution. */ | ||
946 | write_lock_bh(&table->lock); | ||
947 | /* Check inside lock: is the old number correct? */ | ||
948 | if (num_counters != table->private->number) { | ||
949 | duprintf("num_counters != table->private->number (%u/%u)\n", | ||
950 | num_counters, table->private->number); | ||
951 | write_unlock_bh(&table->lock); | ||
952 | *error = -EAGAIN; | ||
953 | return NULL; | ||
954 | } | ||
955 | oldinfo = table->private; | ||
956 | table->private = newinfo; | ||
957 | newinfo->initial_entries = oldinfo->initial_entries; | ||
958 | write_unlock_bh(&table->lock); | ||
959 | |||
960 | return oldinfo; | ||
961 | } | ||
962 | |||
963 | /* Gets counters. */ | 736 | /* Gets counters. */ |
964 | static inline int | 737 | static inline int |
965 | add_entry_to_counter(const struct ipt_entry *e, | 738 | add_entry_to_counter(const struct ipt_entry *e, |
966 | struct ipt_counters total[], | 739 | struct xt_counters total[], |
967 | unsigned int *i) | 740 | unsigned int *i) |
968 | { | 741 | { |
969 | ADD_COUNTER(total[*i], e->counters.bcnt, e->counters.pcnt); | 742 | ADD_COUNTER(total[*i], e->counters.bcnt, e->counters.pcnt); |
@@ -984,8 +757,8 @@ set_entry_to_counter(const struct ipt_entry *e, | |||
984 | } | 757 | } |
985 | 758 | ||
986 | static void | 759 | static void |
987 | get_counters(const struct ipt_table_info *t, | 760 | get_counters(const struct xt_table_info *t, |
988 | struct ipt_counters counters[]) | 761 | struct xt_counters counters[]) |
989 | { | 762 | { |
990 | unsigned int cpu; | 763 | unsigned int cpu; |
991 | unsigned int i; | 764 | unsigned int i; |
@@ -1024,14 +797,15 @@ copy_entries_to_user(unsigned int total_size, | |||
1024 | { | 797 | { |
1025 | unsigned int off, num, countersize; | 798 | unsigned int off, num, countersize; |
1026 | struct ipt_entry *e; | 799 | struct ipt_entry *e; |
1027 | struct ipt_counters *counters; | 800 | struct xt_counters *counters; |
801 | struct xt_table_info *private = table->private; | ||
1028 | int ret = 0; | 802 | int ret = 0; |
1029 | void *loc_cpu_entry; | 803 | void *loc_cpu_entry; |
1030 | 804 | ||
1031 | /* We need atomic snapshot of counters: rest doesn't change | 805 | /* We need atomic snapshot of counters: rest doesn't change |
1032 | (other than comefrom, which userspace doesn't care | 806 | (other than comefrom, which userspace doesn't care |
1033 | about). */ | 807 | about). */ |
1034 | countersize = sizeof(struct ipt_counters) * table->private->number; | 808 | countersize = sizeof(struct xt_counters) * private->number; |
1035 | counters = vmalloc_node(countersize, numa_node_id()); | 809 | counters = vmalloc_node(countersize, numa_node_id()); |
1036 | 810 | ||
1037 | if (counters == NULL) | 811 | if (counters == NULL) |
@@ -1039,14 +813,14 @@ copy_entries_to_user(unsigned int total_size, | |||
1039 | 813 | ||
1040 | /* First, sum counters... */ | 814 | /* First, sum counters... */ |
1041 | write_lock_bh(&table->lock); | 815 | write_lock_bh(&table->lock); |
1042 | get_counters(table->private, counters); | 816 | get_counters(private, counters); |
1043 | write_unlock_bh(&table->lock); | 817 | write_unlock_bh(&table->lock); |
1044 | 818 | ||
1045 | /* choose the copy that is on our node/cpu, ... | 819 | /* choose the copy that is on our node/cpu, ... |
1046 | * This choice is lazy (because current thread is | 820 | * This choice is lazy (because current thread is |
1047 | * allowed to migrate to another cpu) | 821 | * allowed to migrate to another cpu) |
1048 | */ | 822 | */ |
1049 | loc_cpu_entry = table->private->entries[raw_smp_processor_id()]; | 823 | loc_cpu_entry = private->entries[raw_smp_processor_id()]; |
1050 | /* ... then copy entire thing ... */ | 824 | /* ... then copy entire thing ... */ |
1051 | if (copy_to_user(userptr, loc_cpu_entry, total_size) != 0) { | 825 | if (copy_to_user(userptr, loc_cpu_entry, total_size) != 0) { |
1052 | ret = -EFAULT; | 826 | ret = -EFAULT; |
@@ -1108,74 +882,36 @@ get_entries(const struct ipt_get_entries *entries, | |||
1108 | int ret; | 882 | int ret; |
1109 | struct ipt_table *t; | 883 | struct ipt_table *t; |
1110 | 884 | ||
1111 | t = find_table_lock(entries->name); | 885 | t = xt_find_table_lock(AF_INET, entries->name); |
1112 | if (t && !IS_ERR(t)) { | 886 | if (t && !IS_ERR(t)) { |
887 | struct xt_table_info *private = t->private; | ||
1113 | duprintf("t->private->number = %u\n", | 888 | duprintf("t->private->number = %u\n", |
1114 | t->private->number); | 889 | private->number); |
1115 | if (entries->size == t->private->size) | 890 | if (entries->size == private->size) |
1116 | ret = copy_entries_to_user(t->private->size, | 891 | ret = copy_entries_to_user(private->size, |
1117 | t, uptr->entrytable); | 892 | t, uptr->entrytable); |
1118 | else { | 893 | else { |
1119 | duprintf("get_entries: I've got %u not %u!\n", | 894 | duprintf("get_entries: I've got %u not %u!\n", |
1120 | t->private->size, | 895 | private->size, |
1121 | entries->size); | 896 | entries->size); |
1122 | ret = -EINVAL; | 897 | ret = -EINVAL; |
1123 | } | 898 | } |
1124 | module_put(t->me); | 899 | module_put(t->me); |
1125 | up(&ipt_mutex); | 900 | xt_table_unlock(t); |
1126 | } else | 901 | } else |
1127 | ret = t ? PTR_ERR(t) : -ENOENT; | 902 | ret = t ? PTR_ERR(t) : -ENOENT; |
1128 | 903 | ||
1129 | return ret; | 904 | return ret; |
1130 | } | 905 | } |
1131 | 906 | ||
1132 | static void free_table_info(struct ipt_table_info *info) | ||
1133 | { | ||
1134 | int cpu; | ||
1135 | for_each_cpu(cpu) { | ||
1136 | if (info->size <= PAGE_SIZE) | ||
1137 | kfree(info->entries[cpu]); | ||
1138 | else | ||
1139 | vfree(info->entries[cpu]); | ||
1140 | } | ||
1141 | kfree(info); | ||
1142 | } | ||
1143 | |||
1144 | static struct ipt_table_info *alloc_table_info(unsigned int size) | ||
1145 | { | ||
1146 | struct ipt_table_info *newinfo; | ||
1147 | int cpu; | ||
1148 | |||
1149 | newinfo = kzalloc(sizeof(struct ipt_table_info), GFP_KERNEL); | ||
1150 | if (!newinfo) | ||
1151 | return NULL; | ||
1152 | |||
1153 | newinfo->size = size; | ||
1154 | |||
1155 | for_each_cpu(cpu) { | ||
1156 | if (size <= PAGE_SIZE) | ||
1157 | newinfo->entries[cpu] = kmalloc_node(size, | ||
1158 | GFP_KERNEL, | ||
1159 | cpu_to_node(cpu)); | ||
1160 | else | ||
1161 | newinfo->entries[cpu] = vmalloc_node(size, cpu_to_node(cpu)); | ||
1162 | if (newinfo->entries[cpu] == 0) { | ||
1163 | free_table_info(newinfo); | ||
1164 | return NULL; | ||
1165 | } | ||
1166 | } | ||
1167 | |||
1168 | return newinfo; | ||
1169 | } | ||
1170 | |||
1171 | static int | 907 | static int |
1172 | do_replace(void __user *user, unsigned int len) | 908 | do_replace(void __user *user, unsigned int len) |
1173 | { | 909 | { |
1174 | int ret; | 910 | int ret; |
1175 | struct ipt_replace tmp; | 911 | struct ipt_replace tmp; |
1176 | struct ipt_table *t; | 912 | struct ipt_table *t; |
1177 | struct ipt_table_info *newinfo, *oldinfo; | 913 | struct xt_table_info *newinfo, *oldinfo; |
1178 | struct ipt_counters *counters; | 914 | struct xt_counters *counters; |
1179 | void *loc_cpu_entry, *loc_cpu_old_entry; | 915 | void *loc_cpu_entry, *loc_cpu_old_entry; |
1180 | 916 | ||
1181 | if (copy_from_user(&tmp, user, sizeof(tmp)) != 0) | 917 | if (copy_from_user(&tmp, user, sizeof(tmp)) != 0) |
@@ -1185,11 +921,7 @@ do_replace(void __user *user, unsigned int len) | |||
1185 | if (len != sizeof(tmp) + tmp.size) | 921 | if (len != sizeof(tmp) + tmp.size) |
1186 | return -ENOPROTOOPT; | 922 | return -ENOPROTOOPT; |
1187 | 923 | ||
1188 | /* Pedantry: prevent them from hitting BUG() in vmalloc.c --RR */ | 924 | newinfo = xt_alloc_table_info(tmp.size); |
1189 | if ((SMP_ALIGN(tmp.size) >> PAGE_SHIFT) + 2 > num_physpages) | ||
1190 | return -ENOMEM; | ||
1191 | |||
1192 | newinfo = alloc_table_info(tmp.size); | ||
1193 | if (!newinfo) | 925 | if (!newinfo) |
1194 | return -ENOMEM; | 926 | return -ENOMEM; |
1195 | 927 | ||
@@ -1201,7 +933,7 @@ do_replace(void __user *user, unsigned int len) | |||
1201 | goto free_newinfo; | 933 | goto free_newinfo; |
1202 | } | 934 | } |
1203 | 935 | ||
1204 | counters = vmalloc(tmp.num_counters * sizeof(struct ipt_counters)); | 936 | counters = vmalloc(tmp.num_counters * sizeof(struct xt_counters)); |
1205 | if (!counters) { | 937 | if (!counters) { |
1206 | ret = -ENOMEM; | 938 | ret = -ENOMEM; |
1207 | goto free_newinfo; | 939 | goto free_newinfo; |
@@ -1215,7 +947,7 @@ do_replace(void __user *user, unsigned int len) | |||
1215 | 947 | ||
1216 | duprintf("ip_tables: Translated table\n"); | 948 | duprintf("ip_tables: Translated table\n"); |
1217 | 949 | ||
1218 | t = try_then_request_module(find_table_lock(tmp.name), | 950 | t = try_then_request_module(xt_find_table_lock(AF_INET, tmp.name), |
1219 | "iptable_%s", tmp.name); | 951 | "iptable_%s", tmp.name); |
1220 | if (!t || IS_ERR(t)) { | 952 | if (!t || IS_ERR(t)) { |
1221 | ret = t ? PTR_ERR(t) : -ENOENT; | 953 | ret = t ? PTR_ERR(t) : -ENOENT; |
@@ -1230,7 +962,7 @@ do_replace(void __user *user, unsigned int len) | |||
1230 | goto put_module; | 962 | goto put_module; |
1231 | } | 963 | } |
1232 | 964 | ||
1233 | oldinfo = replace_table(t, tmp.num_counters, newinfo, &ret); | 965 | oldinfo = xt_replace_table(t, tmp.num_counters, newinfo, &ret); |
1234 | if (!oldinfo) | 966 | if (!oldinfo) |
1235 | goto put_module; | 967 | goto put_module; |
1236 | 968 | ||
@@ -1249,23 +981,23 @@ do_replace(void __user *user, unsigned int len) | |||
1249 | /* Decrease module usage counts and free resource */ | 981 | /* Decrease module usage counts and free resource */ |
1250 | loc_cpu_old_entry = oldinfo->entries[raw_smp_processor_id()]; | 982 | loc_cpu_old_entry = oldinfo->entries[raw_smp_processor_id()]; |
1251 | IPT_ENTRY_ITERATE(loc_cpu_old_entry, oldinfo->size, cleanup_entry,NULL); | 983 | IPT_ENTRY_ITERATE(loc_cpu_old_entry, oldinfo->size, cleanup_entry,NULL); |
1252 | free_table_info(oldinfo); | 984 | xt_free_table_info(oldinfo); |
1253 | if (copy_to_user(tmp.counters, counters, | 985 | if (copy_to_user(tmp.counters, counters, |
1254 | sizeof(struct ipt_counters) * tmp.num_counters) != 0) | 986 | sizeof(struct xt_counters) * tmp.num_counters) != 0) |
1255 | ret = -EFAULT; | 987 | ret = -EFAULT; |
1256 | vfree(counters); | 988 | vfree(counters); |
1257 | up(&ipt_mutex); | 989 | xt_table_unlock(t); |
1258 | return ret; | 990 | return ret; |
1259 | 991 | ||
1260 | put_module: | 992 | put_module: |
1261 | module_put(t->me); | 993 | module_put(t->me); |
1262 | up(&ipt_mutex); | 994 | xt_table_unlock(t); |
1263 | free_newinfo_counters_untrans: | 995 | free_newinfo_counters_untrans: |
1264 | IPT_ENTRY_ITERATE(loc_cpu_entry, newinfo->size, cleanup_entry,NULL); | 996 | IPT_ENTRY_ITERATE(loc_cpu_entry, newinfo->size, cleanup_entry,NULL); |
1265 | free_newinfo_counters: | 997 | free_newinfo_counters: |
1266 | vfree(counters); | 998 | vfree(counters); |
1267 | free_newinfo: | 999 | free_newinfo: |
1268 | free_table_info(newinfo); | 1000 | xt_free_table_info(newinfo); |
1269 | return ret; | 1001 | return ret; |
1270 | } | 1002 | } |
1271 | 1003 | ||
@@ -1273,7 +1005,7 @@ do_replace(void __user *user, unsigned int len) | |||
1273 | * and everything is OK. */ | 1005 | * and everything is OK. */ |
1274 | static inline int | 1006 | static inline int |
1275 | add_counter_to_entry(struct ipt_entry *e, | 1007 | add_counter_to_entry(struct ipt_entry *e, |
1276 | const struct ipt_counters addme[], | 1008 | const struct xt_counters addme[], |
1277 | unsigned int *i) | 1009 | unsigned int *i) |
1278 | { | 1010 | { |
1279 | #if 0 | 1011 | #if 0 |
@@ -1295,15 +1027,16 @@ static int | |||
1295 | do_add_counters(void __user *user, unsigned int len) | 1027 | do_add_counters(void __user *user, unsigned int len) |
1296 | { | 1028 | { |
1297 | unsigned int i; | 1029 | unsigned int i; |
1298 | struct ipt_counters_info tmp, *paddc; | 1030 | struct xt_counters_info tmp, *paddc; |
1299 | struct ipt_table *t; | 1031 | struct ipt_table *t; |
1032 | struct xt_table_info *private; | ||
1300 | int ret = 0; | 1033 | int ret = 0; |
1301 | void *loc_cpu_entry; | 1034 | void *loc_cpu_entry; |
1302 | 1035 | ||
1303 | if (copy_from_user(&tmp, user, sizeof(tmp)) != 0) | 1036 | if (copy_from_user(&tmp, user, sizeof(tmp)) != 0) |
1304 | return -EFAULT; | 1037 | return -EFAULT; |
1305 | 1038 | ||
1306 | if (len != sizeof(tmp) + tmp.num_counters*sizeof(struct ipt_counters)) | 1039 | if (len != sizeof(tmp) + tmp.num_counters*sizeof(struct xt_counters)) |
1307 | return -EINVAL; | 1040 | return -EINVAL; |
1308 | 1041 | ||
1309 | paddc = vmalloc_node(len, numa_node_id()); | 1042 | paddc = vmalloc_node(len, numa_node_id()); |
@@ -1315,29 +1048,30 @@ do_add_counters(void __user *user, unsigned int len) | |||
1315 | goto free; | 1048 | goto free; |
1316 | } | 1049 | } |
1317 | 1050 | ||
1318 | t = find_table_lock(tmp.name); | 1051 | t = xt_find_table_lock(AF_INET, tmp.name); |
1319 | if (!t || IS_ERR(t)) { | 1052 | if (!t || IS_ERR(t)) { |
1320 | ret = t ? PTR_ERR(t) : -ENOENT; | 1053 | ret = t ? PTR_ERR(t) : -ENOENT; |
1321 | goto free; | 1054 | goto free; |
1322 | } | 1055 | } |
1323 | 1056 | ||
1324 | write_lock_bh(&t->lock); | 1057 | write_lock_bh(&t->lock); |
1325 | if (t->private->number != paddc->num_counters) { | 1058 | private = t->private; |
1059 | if (private->number != paddc->num_counters) { | ||
1326 | ret = -EINVAL; | 1060 | ret = -EINVAL; |
1327 | goto unlock_up_free; | 1061 | goto unlock_up_free; |
1328 | } | 1062 | } |
1329 | 1063 | ||
1330 | i = 0; | 1064 | i = 0; |
1331 | /* Choose the copy that is on our node */ | 1065 | /* Choose the copy that is on our node */ |
1332 | loc_cpu_entry = t->private->entries[raw_smp_processor_id()]; | 1066 | loc_cpu_entry = private->entries[raw_smp_processor_id()]; |
1333 | IPT_ENTRY_ITERATE(loc_cpu_entry, | 1067 | IPT_ENTRY_ITERATE(loc_cpu_entry, |
1334 | t->private->size, | 1068 | private->size, |
1335 | add_counter_to_entry, | 1069 | add_counter_to_entry, |
1336 | paddc->counters, | 1070 | paddc->counters, |
1337 | &i); | 1071 | &i); |
1338 | unlock_up_free: | 1072 | unlock_up_free: |
1339 | write_unlock_bh(&t->lock); | 1073 | write_unlock_bh(&t->lock); |
1340 | up(&ipt_mutex); | 1074 | xt_table_unlock(t); |
1341 | module_put(t->me); | 1075 | module_put(t->me); |
1342 | free: | 1076 | free: |
1343 | vfree(paddc); | 1077 | vfree(paddc); |
@@ -1396,25 +1130,26 @@ do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len) | |||
1396 | } | 1130 | } |
1397 | name[IPT_TABLE_MAXNAMELEN-1] = '\0'; | 1131 | name[IPT_TABLE_MAXNAMELEN-1] = '\0'; |
1398 | 1132 | ||
1399 | t = try_then_request_module(find_table_lock(name), | 1133 | t = try_then_request_module(xt_find_table_lock(AF_INET, name), |
1400 | "iptable_%s", name); | 1134 | "iptable_%s", name); |
1401 | if (t && !IS_ERR(t)) { | 1135 | if (t && !IS_ERR(t)) { |
1402 | struct ipt_getinfo info; | 1136 | struct ipt_getinfo info; |
1137 | struct xt_table_info *private = t->private; | ||
1403 | 1138 | ||
1404 | info.valid_hooks = t->valid_hooks; | 1139 | info.valid_hooks = t->valid_hooks; |
1405 | memcpy(info.hook_entry, t->private->hook_entry, | 1140 | memcpy(info.hook_entry, private->hook_entry, |
1406 | sizeof(info.hook_entry)); | 1141 | sizeof(info.hook_entry)); |
1407 | memcpy(info.underflow, t->private->underflow, | 1142 | memcpy(info.underflow, private->underflow, |
1408 | sizeof(info.underflow)); | 1143 | sizeof(info.underflow)); |
1409 | info.num_entries = t->private->number; | 1144 | info.num_entries = private->number; |
1410 | info.size = t->private->size; | 1145 | info.size = private->size; |
1411 | memcpy(info.name, name, sizeof(info.name)); | 1146 | memcpy(info.name, name, sizeof(info.name)); |
1412 | 1147 | ||
1413 | if (copy_to_user(user, &info, *len) != 0) | 1148 | if (copy_to_user(user, &info, *len) != 0) |
1414 | ret = -EFAULT; | 1149 | ret = -EFAULT; |
1415 | else | 1150 | else |
1416 | ret = 0; | 1151 | ret = 0; |
1417 | up(&ipt_mutex); | 1152 | xt_table_unlock(t); |
1418 | module_put(t->me); | 1153 | module_put(t->me); |
1419 | } else | 1154 | } else |
1420 | ret = t ? PTR_ERR(t) : -ENOENT; | 1155 | ret = t ? PTR_ERR(t) : -ENOENT; |
@@ -1441,7 +1176,7 @@ do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len) | |||
1441 | case IPT_SO_GET_REVISION_MATCH: | 1176 | case IPT_SO_GET_REVISION_MATCH: |
1442 | case IPT_SO_GET_REVISION_TARGET: { | 1177 | case IPT_SO_GET_REVISION_TARGET: { |
1443 | struct ipt_get_revision rev; | 1178 | struct ipt_get_revision rev; |
1444 | int (*revfn)(const char *, u8, int *); | 1179 | int target; |
1445 | 1180 | ||
1446 | if (*len != sizeof(rev)) { | 1181 | if (*len != sizeof(rev)) { |
1447 | ret = -EINVAL; | 1182 | ret = -EINVAL; |
@@ -1453,12 +1188,13 @@ do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len) | |||
1453 | } | 1188 | } |
1454 | 1189 | ||
1455 | if (cmd == IPT_SO_GET_REVISION_TARGET) | 1190 | if (cmd == IPT_SO_GET_REVISION_TARGET) |
1456 | revfn = target_revfn; | 1191 | target = 1; |
1457 | else | 1192 | else |
1458 | revfn = match_revfn; | 1193 | target = 0; |
1459 | 1194 | ||
1460 | try_then_request_module(find_revision(rev.name, rev.revision, | 1195 | try_then_request_module(xt_find_revision(AF_INET, rev.name, |
1461 | revfn, &ret), | 1196 | rev.revision, |
1197 | target, &ret), | ||
1462 | "ipt_%s", rev.name); | 1198 | "ipt_%s", rev.name); |
1463 | break; | 1199 | break; |
1464 | } | 1200 | } |
@@ -1471,60 +1207,15 @@ do_ipt_get_ctl(struct sock *sk, int cmd, void __user *user, int *len) | |||
1471 | return ret; | 1207 | return ret; |
1472 | } | 1208 | } |
1473 | 1209 | ||
1474 | /* Registration hooks for targets. */ | 1210 | int ipt_register_table(struct xt_table *table, const struct ipt_replace *repl) |
1475 | int | ||
1476 | ipt_register_target(struct ipt_target *target) | ||
1477 | { | 1211 | { |
1478 | int ret; | 1212 | int ret; |
1479 | 1213 | struct xt_table_info *newinfo; | |
1480 | ret = down_interruptible(&ipt_mutex); | 1214 | static struct xt_table_info bootstrap |
1481 | if (ret != 0) | ||
1482 | return ret; | ||
1483 | list_add(&target->list, &ipt_target); | ||
1484 | up(&ipt_mutex); | ||
1485 | return ret; | ||
1486 | } | ||
1487 | |||
1488 | void | ||
1489 | ipt_unregister_target(struct ipt_target *target) | ||
1490 | { | ||
1491 | down(&ipt_mutex); | ||
1492 | LIST_DELETE(&ipt_target, target); | ||
1493 | up(&ipt_mutex); | ||
1494 | } | ||
1495 | |||
1496 | int | ||
1497 | ipt_register_match(struct ipt_match *match) | ||
1498 | { | ||
1499 | int ret; | ||
1500 | |||
1501 | ret = down_interruptible(&ipt_mutex); | ||
1502 | if (ret != 0) | ||
1503 | return ret; | ||
1504 | |||
1505 | list_add(&match->list, &ipt_match); | ||
1506 | up(&ipt_mutex); | ||
1507 | |||
1508 | return ret; | ||
1509 | } | ||
1510 | |||
1511 | void | ||
1512 | ipt_unregister_match(struct ipt_match *match) | ||
1513 | { | ||
1514 | down(&ipt_mutex); | ||
1515 | LIST_DELETE(&ipt_match, match); | ||
1516 | up(&ipt_mutex); | ||
1517 | } | ||
1518 | |||
1519 | int ipt_register_table(struct ipt_table *table, const struct ipt_replace *repl) | ||
1520 | { | ||
1521 | int ret; | ||
1522 | struct ipt_table_info *newinfo; | ||
1523 | static struct ipt_table_info bootstrap | ||
1524 | = { 0, 0, 0, { 0 }, { 0 }, { } }; | 1215 | = { 0, 0, 0, { 0 }, { 0 }, { } }; |
1525 | void *loc_cpu_entry; | 1216 | void *loc_cpu_entry; |
1526 | 1217 | ||
1527 | newinfo = alloc_table_info(repl->size); | 1218 | newinfo = xt_alloc_table_info(repl->size); |
1528 | if (!newinfo) | 1219 | if (!newinfo) |
1529 | return -ENOMEM; | 1220 | return -ENOMEM; |
1530 | 1221 | ||
@@ -1540,246 +1231,29 @@ int ipt_register_table(struct ipt_table *table, const struct ipt_replace *repl) | |||
1540 | repl->hook_entry, | 1231 | repl->hook_entry, |
1541 | repl->underflow); | 1232 | repl->underflow); |
1542 | if (ret != 0) { | 1233 | if (ret != 0) { |
1543 | free_table_info(newinfo); | 1234 | xt_free_table_info(newinfo); |
1544 | return ret; | 1235 | return ret; |
1545 | } | 1236 | } |
1546 | 1237 | ||
1547 | ret = down_interruptible(&ipt_mutex); | 1238 | if (xt_register_table(table, &bootstrap, newinfo) != 0) { |
1548 | if (ret != 0) { | 1239 | xt_free_table_info(newinfo); |
1549 | free_table_info(newinfo); | ||
1550 | return ret; | 1240 | return ret; |
1551 | } | 1241 | } |
1552 | 1242 | ||
1553 | /* Don't autoload: we'd eat our tail... */ | 1243 | return 0; |
1554 | if (list_named_find(&ipt_tables, table->name)) { | ||
1555 | ret = -EEXIST; | ||
1556 | goto free_unlock; | ||
1557 | } | ||
1558 | |||
1559 | /* Simplifies replace_table code. */ | ||
1560 | table->private = &bootstrap; | ||
1561 | if (!replace_table(table, 0, newinfo, &ret)) | ||
1562 | goto free_unlock; | ||
1563 | |||
1564 | duprintf("table->private->number = %u\n", | ||
1565 | table->private->number); | ||
1566 | |||
1567 | /* save number of initial entries */ | ||
1568 | table->private->initial_entries = table->private->number; | ||
1569 | |||
1570 | rwlock_init(&table->lock); | ||
1571 | list_prepend(&ipt_tables, table); | ||
1572 | |||
1573 | unlock: | ||
1574 | up(&ipt_mutex); | ||
1575 | return ret; | ||
1576 | |||
1577 | free_unlock: | ||
1578 | free_table_info(newinfo); | ||
1579 | goto unlock; | ||
1580 | } | 1244 | } |
1581 | 1245 | ||
1582 | void ipt_unregister_table(struct ipt_table *table) | 1246 | void ipt_unregister_table(struct ipt_table *table) |
1583 | { | 1247 | { |
1248 | struct xt_table_info *private; | ||
1584 | void *loc_cpu_entry; | 1249 | void *loc_cpu_entry; |
1585 | 1250 | ||
1586 | down(&ipt_mutex); | 1251 | private = xt_unregister_table(table); |
1587 | LIST_DELETE(&ipt_tables, table); | ||
1588 | up(&ipt_mutex); | ||
1589 | 1252 | ||
1590 | /* Decrease module usage counts and free resources */ | 1253 | /* Decrease module usage counts and free resources */ |
1591 | loc_cpu_entry = table->private->entries[raw_smp_processor_id()]; | 1254 | loc_cpu_entry = private->entries[raw_smp_processor_id()]; |
1592 | IPT_ENTRY_ITERATE(loc_cpu_entry, table->private->size, | 1255 | IPT_ENTRY_ITERATE(loc_cpu_entry, private->size, cleanup_entry, NULL); |
1593 | cleanup_entry, NULL); | 1256 | xt_free_table_info(private); |
1594 | free_table_info(table->private); | ||
1595 | } | ||
1596 | |||
1597 | /* Returns 1 if the port is matched by the range, 0 otherwise */ | ||
1598 | static inline int | ||
1599 | port_match(u_int16_t min, u_int16_t max, u_int16_t port, int invert) | ||
1600 | { | ||
1601 | int ret; | ||
1602 | |||
1603 | ret = (port >= min && port <= max) ^ invert; | ||
1604 | return ret; | ||
1605 | } | ||
1606 | |||
1607 | static int | ||
1608 | tcp_find_option(u_int8_t option, | ||
1609 | const struct sk_buff *skb, | ||
1610 | unsigned int optlen, | ||
1611 | int invert, | ||
1612 | int *hotdrop) | ||
1613 | { | ||
1614 | /* tcp.doff is only 4 bits, ie. max 15 * 4 bytes */ | ||
1615 | u_int8_t _opt[60 - sizeof(struct tcphdr)], *op; | ||
1616 | unsigned int i; | ||
1617 | |||
1618 | duprintf("tcp_match: finding option\n"); | ||
1619 | |||
1620 | if (!optlen) | ||
1621 | return invert; | ||
1622 | |||
1623 | /* If we don't have the whole header, drop packet. */ | ||
1624 | op = skb_header_pointer(skb, | ||
1625 | skb->nh.iph->ihl*4 + sizeof(struct tcphdr), | ||
1626 | optlen, _opt); | ||
1627 | if (op == NULL) { | ||
1628 | *hotdrop = 1; | ||
1629 | return 0; | ||
1630 | } | ||
1631 | |||
1632 | for (i = 0; i < optlen; ) { | ||
1633 | if (op[i] == option) return !invert; | ||
1634 | if (op[i] < 2) i++; | ||
1635 | else i += op[i+1]?:1; | ||
1636 | } | ||
1637 | |||
1638 | return invert; | ||
1639 | } | ||
1640 | |||
1641 | static int | ||
1642 | tcp_match(const struct sk_buff *skb, | ||
1643 | const struct net_device *in, | ||
1644 | const struct net_device *out, | ||
1645 | const void *matchinfo, | ||
1646 | int offset, | ||
1647 | int *hotdrop) | ||
1648 | { | ||
1649 | struct tcphdr _tcph, *th; | ||
1650 | const struct ipt_tcp *tcpinfo = matchinfo; | ||
1651 | |||
1652 | if (offset) { | ||
1653 | /* To quote Alan: | ||
1654 | |||
1655 | Don't allow a fragment of TCP 8 bytes in. Nobody normal | ||
1656 | causes this. Its a cracker trying to break in by doing a | ||
1657 | flag overwrite to pass the direction checks. | ||
1658 | */ | ||
1659 | if (offset == 1) { | ||
1660 | duprintf("Dropping evil TCP offset=1 frag.\n"); | ||
1661 | *hotdrop = 1; | ||
1662 | } | ||
1663 | /* Must not be a fragment. */ | ||
1664 | return 0; | ||
1665 | } | ||
1666 | |||
1667 | #define FWINVTCP(bool,invflg) ((bool) ^ !!(tcpinfo->invflags & invflg)) | ||
1668 | |||
1669 | th = skb_header_pointer(skb, skb->nh.iph->ihl*4, | ||
1670 | sizeof(_tcph), &_tcph); | ||
1671 | if (th == NULL) { | ||
1672 | /* We've been asked to examine this packet, and we | ||
1673 | can't. Hence, no choice but to drop. */ | ||
1674 | duprintf("Dropping evil TCP offset=0 tinygram.\n"); | ||
1675 | *hotdrop = 1; | ||
1676 | return 0; | ||
1677 | } | ||
1678 | |||
1679 | if (!port_match(tcpinfo->spts[0], tcpinfo->spts[1], | ||
1680 | ntohs(th->source), | ||
1681 | !!(tcpinfo->invflags & IPT_TCP_INV_SRCPT))) | ||
1682 | return 0; | ||
1683 | if (!port_match(tcpinfo->dpts[0], tcpinfo->dpts[1], | ||
1684 | ntohs(th->dest), | ||
1685 | !!(tcpinfo->invflags & IPT_TCP_INV_DSTPT))) | ||
1686 | return 0; | ||
1687 | if (!FWINVTCP((((unsigned char *)th)[13] & tcpinfo->flg_mask) | ||
1688 | == tcpinfo->flg_cmp, | ||
1689 | IPT_TCP_INV_FLAGS)) | ||
1690 | return 0; | ||
1691 | if (tcpinfo->option) { | ||
1692 | if (th->doff * 4 < sizeof(_tcph)) { | ||
1693 | *hotdrop = 1; | ||
1694 | return 0; | ||
1695 | } | ||
1696 | if (!tcp_find_option(tcpinfo->option, skb, | ||
1697 | th->doff*4 - sizeof(_tcph), | ||
1698 | tcpinfo->invflags & IPT_TCP_INV_OPTION, | ||
1699 | hotdrop)) | ||
1700 | return 0; | ||
1701 | } | ||
1702 | return 1; | ||
1703 | } | ||
1704 | |||
1705 | /* Called when user tries to insert an entry of this type. */ | ||
1706 | static int | ||
1707 | tcp_checkentry(const char *tablename, | ||
1708 | const struct ipt_ip *ip, | ||
1709 | void *matchinfo, | ||
1710 | unsigned int matchsize, | ||
1711 | unsigned int hook_mask) | ||
1712 | { | ||
1713 | const struct ipt_tcp *tcpinfo = matchinfo; | ||
1714 | |||
1715 | /* Must specify proto == TCP, and no unknown invflags */ | ||
1716 | return ip->proto == IPPROTO_TCP | ||
1717 | && !(ip->invflags & IPT_INV_PROTO) | ||
1718 | && matchsize == IPT_ALIGN(sizeof(struct ipt_tcp)) | ||
1719 | && !(tcpinfo->invflags & ~IPT_TCP_INV_MASK); | ||
1720 | } | ||
1721 | |||
1722 | static int | ||
1723 | udp_match(const struct sk_buff *skb, | ||
1724 | const struct net_device *in, | ||
1725 | const struct net_device *out, | ||
1726 | const void *matchinfo, | ||
1727 | int offset, | ||
1728 | int *hotdrop) | ||
1729 | { | ||
1730 | struct udphdr _udph, *uh; | ||
1731 | const struct ipt_udp *udpinfo = matchinfo; | ||
1732 | |||
1733 | /* Must not be a fragment. */ | ||
1734 | if (offset) | ||
1735 | return 0; | ||
1736 | |||
1737 | uh = skb_header_pointer(skb, skb->nh.iph->ihl*4, | ||
1738 | sizeof(_udph), &_udph); | ||
1739 | if (uh == NULL) { | ||
1740 | /* We've been asked to examine this packet, and we | ||
1741 | can't. Hence, no choice but to drop. */ | ||
1742 | duprintf("Dropping evil UDP tinygram.\n"); | ||
1743 | *hotdrop = 1; | ||
1744 | return 0; | ||
1745 | } | ||
1746 | |||
1747 | return port_match(udpinfo->spts[0], udpinfo->spts[1], | ||
1748 | ntohs(uh->source), | ||
1749 | !!(udpinfo->invflags & IPT_UDP_INV_SRCPT)) | ||
1750 | && port_match(udpinfo->dpts[0], udpinfo->dpts[1], | ||
1751 | ntohs(uh->dest), | ||
1752 | !!(udpinfo->invflags & IPT_UDP_INV_DSTPT)); | ||
1753 | } | ||
1754 | |||
1755 | /* Called when user tries to insert an entry of this type. */ | ||
1756 | static int | ||
1757 | udp_checkentry(const char *tablename, | ||
1758 | const struct ipt_ip *ip, | ||
1759 | void *matchinfo, | ||
1760 | unsigned int matchinfosize, | ||
1761 | unsigned int hook_mask) | ||
1762 | { | ||
1763 | const struct ipt_udp *udpinfo = matchinfo; | ||
1764 | |||
1765 | /* Must specify proto == UDP, and no unknown invflags */ | ||
1766 | if (ip->proto != IPPROTO_UDP || (ip->invflags & IPT_INV_PROTO)) { | ||
1767 | duprintf("ipt_udp: Protocol %u != %u\n", ip->proto, | ||
1768 | IPPROTO_UDP); | ||
1769 | return 0; | ||
1770 | } | ||
1771 | if (matchinfosize != IPT_ALIGN(sizeof(struct ipt_udp))) { | ||
1772 | duprintf("ipt_udp: matchsize %u != %u\n", | ||
1773 | matchinfosize, IPT_ALIGN(sizeof(struct ipt_udp))); | ||
1774 | return 0; | ||
1775 | } | ||
1776 | if (udpinfo->invflags & ~IPT_UDP_INV_MASK) { | ||
1777 | duprintf("ipt_udp: unknown flags %X\n", | ||
1778 | udpinfo->invflags); | ||
1779 | return 0; | ||
1780 | } | ||
1781 | |||
1782 | return 1; | ||
1783 | } | 1257 | } |
1784 | 1258 | ||
1785 | /* Returns 1 if the type and code is matched by the range, 0 otherwise */ | 1259 | /* Returns 1 if the type and code is matched by the range, 0 otherwise */ |
@@ -1798,6 +1272,7 @@ icmp_match(const struct sk_buff *skb, | |||
1798 | const struct net_device *out, | 1272 | const struct net_device *out, |
1799 | const void *matchinfo, | 1273 | const void *matchinfo, |
1800 | int offset, | 1274 | int offset, |
1275 | unsigned int protoff, | ||
1801 | int *hotdrop) | 1276 | int *hotdrop) |
1802 | { | 1277 | { |
1803 | struct icmphdr _icmph, *ic; | 1278 | struct icmphdr _icmph, *ic; |
@@ -1807,8 +1282,7 @@ icmp_match(const struct sk_buff *skb, | |||
1807 | if (offset) | 1282 | if (offset) |
1808 | return 0; | 1283 | return 0; |
1809 | 1284 | ||
1810 | ic = skb_header_pointer(skb, skb->nh.iph->ihl*4, | 1285 | ic = skb_header_pointer(skb, protoff, sizeof(_icmph), &_icmph); |
1811 | sizeof(_icmph), &_icmph); | ||
1812 | if (ic == NULL) { | 1286 | if (ic == NULL) { |
1813 | /* We've been asked to examine this packet, and we | 1287 | /* We've been asked to examine this packet, and we |
1814 | * can't. Hence, no choice but to drop. | 1288 | * can't. Hence, no choice but to drop. |
@@ -1828,11 +1302,12 @@ icmp_match(const struct sk_buff *skb, | |||
1828 | /* Called when user tries to insert an entry of this type. */ | 1302 | /* Called when user tries to insert an entry of this type. */ |
1829 | static int | 1303 | static int |
1830 | icmp_checkentry(const char *tablename, | 1304 | icmp_checkentry(const char *tablename, |
1831 | const struct ipt_ip *ip, | 1305 | const void *info, |
1832 | void *matchinfo, | 1306 | void *matchinfo, |
1833 | unsigned int matchsize, | 1307 | unsigned int matchsize, |
1834 | unsigned int hook_mask) | 1308 | unsigned int hook_mask) |
1835 | { | 1309 | { |
1310 | const struct ipt_ip *ip = info; | ||
1836 | const struct ipt_icmp *icmpinfo = matchinfo; | 1311 | const struct ipt_icmp *icmpinfo = matchinfo; |
1837 | 1312 | ||
1838 | /* Must specify proto == ICMP, and no unknown invflags */ | 1313 | /* Must specify proto == ICMP, and no unknown invflags */ |
@@ -1862,123 +1337,22 @@ static struct nf_sockopt_ops ipt_sockopts = { | |||
1862 | .get = do_ipt_get_ctl, | 1337 | .get = do_ipt_get_ctl, |
1863 | }; | 1338 | }; |
1864 | 1339 | ||
1865 | static struct ipt_match tcp_matchstruct = { | ||
1866 | .name = "tcp", | ||
1867 | .match = &tcp_match, | ||
1868 | .checkentry = &tcp_checkentry, | ||
1869 | }; | ||
1870 | |||
1871 | static struct ipt_match udp_matchstruct = { | ||
1872 | .name = "udp", | ||
1873 | .match = &udp_match, | ||
1874 | .checkentry = &udp_checkentry, | ||
1875 | }; | ||
1876 | |||
1877 | static struct ipt_match icmp_matchstruct = { | 1340 | static struct ipt_match icmp_matchstruct = { |
1878 | .name = "icmp", | 1341 | .name = "icmp", |
1879 | .match = &icmp_match, | 1342 | .match = &icmp_match, |
1880 | .checkentry = &icmp_checkentry, | 1343 | .checkentry = &icmp_checkentry, |
1881 | }; | 1344 | }; |
1882 | 1345 | ||
1883 | #ifdef CONFIG_PROC_FS | ||
1884 | static inline int print_name(const char *i, | ||
1885 | off_t start_offset, char *buffer, int length, | ||
1886 | off_t *pos, unsigned int *count) | ||
1887 | { | ||
1888 | if ((*count)++ >= start_offset) { | ||
1889 | unsigned int namelen; | ||
1890 | |||
1891 | namelen = sprintf(buffer + *pos, "%s\n", | ||
1892 | i + sizeof(struct list_head)); | ||
1893 | if (*pos + namelen > length) { | ||
1894 | /* Stop iterating */ | ||
1895 | return 1; | ||
1896 | } | ||
1897 | *pos += namelen; | ||
1898 | } | ||
1899 | return 0; | ||
1900 | } | ||
1901 | |||
1902 | static inline int print_target(const struct ipt_target *t, | ||
1903 | off_t start_offset, char *buffer, int length, | ||
1904 | off_t *pos, unsigned int *count) | ||
1905 | { | ||
1906 | if (t == &ipt_standard_target || t == &ipt_error_target) | ||
1907 | return 0; | ||
1908 | return print_name((char *)t, start_offset, buffer, length, pos, count); | ||
1909 | } | ||
1910 | |||
1911 | static int ipt_get_tables(char *buffer, char **start, off_t offset, int length) | ||
1912 | { | ||
1913 | off_t pos = 0; | ||
1914 | unsigned int count = 0; | ||
1915 | |||
1916 | if (down_interruptible(&ipt_mutex) != 0) | ||
1917 | return 0; | ||
1918 | |||
1919 | LIST_FIND(&ipt_tables, print_name, void *, | ||
1920 | offset, buffer, length, &pos, &count); | ||
1921 | |||
1922 | up(&ipt_mutex); | ||
1923 | |||
1924 | /* `start' hack - see fs/proc/generic.c line ~105 */ | ||
1925 | *start=(char *)((unsigned long)count-offset); | ||
1926 | return pos; | ||
1927 | } | ||
1928 | |||
1929 | static int ipt_get_targets(char *buffer, char **start, off_t offset, int length) | ||
1930 | { | ||
1931 | off_t pos = 0; | ||
1932 | unsigned int count = 0; | ||
1933 | |||
1934 | if (down_interruptible(&ipt_mutex) != 0) | ||
1935 | return 0; | ||
1936 | |||
1937 | LIST_FIND(&ipt_target, print_target, struct ipt_target *, | ||
1938 | offset, buffer, length, &pos, &count); | ||
1939 | |||
1940 | up(&ipt_mutex); | ||
1941 | |||
1942 | *start = (char *)((unsigned long)count - offset); | ||
1943 | return pos; | ||
1944 | } | ||
1945 | |||
1946 | static int ipt_get_matches(char *buffer, char **start, off_t offset, int length) | ||
1947 | { | ||
1948 | off_t pos = 0; | ||
1949 | unsigned int count = 0; | ||
1950 | |||
1951 | if (down_interruptible(&ipt_mutex) != 0) | ||
1952 | return 0; | ||
1953 | |||
1954 | LIST_FIND(&ipt_match, print_name, void *, | ||
1955 | offset, buffer, length, &pos, &count); | ||
1956 | |||
1957 | up(&ipt_mutex); | ||
1958 | |||
1959 | *start = (char *)((unsigned long)count - offset); | ||
1960 | return pos; | ||
1961 | } | ||
1962 | |||
1963 | static const struct { char *name; get_info_t *get_info; } ipt_proc_entry[] = | ||
1964 | { { "ip_tables_names", ipt_get_tables }, | ||
1965 | { "ip_tables_targets", ipt_get_targets }, | ||
1966 | { "ip_tables_matches", ipt_get_matches }, | ||
1967 | { NULL, NULL} }; | ||
1968 | #endif /*CONFIG_PROC_FS*/ | ||
1969 | |||
1970 | static int __init init(void) | 1346 | static int __init init(void) |
1971 | { | 1347 | { |
1972 | int ret; | 1348 | int ret; |
1973 | 1349 | ||
1350 | xt_proto_init(AF_INET); | ||
1351 | |||
1974 | /* Noone else will be downing sem now, so we won't sleep */ | 1352 | /* Noone else will be downing sem now, so we won't sleep */ |
1975 | down(&ipt_mutex); | 1353 | xt_register_target(AF_INET, &ipt_standard_target); |
1976 | list_append(&ipt_target, &ipt_standard_target); | 1354 | xt_register_target(AF_INET, &ipt_error_target); |
1977 | list_append(&ipt_target, &ipt_error_target); | 1355 | xt_register_match(AF_INET, &icmp_matchstruct); |
1978 | list_append(&ipt_match, &tcp_matchstruct); | ||
1979 | list_append(&ipt_match, &udp_matchstruct); | ||
1980 | list_append(&ipt_match, &icmp_matchstruct); | ||
1981 | up(&ipt_mutex); | ||
1982 | 1356 | ||
1983 | /* Register setsockopt */ | 1357 | /* Register setsockopt */ |
1984 | ret = nf_register_sockopt(&ipt_sockopts); | 1358 | ret = nf_register_sockopt(&ipt_sockopts); |
@@ -1987,49 +1361,23 @@ static int __init init(void) | |||
1987 | return ret; | 1361 | return ret; |
1988 | } | 1362 | } |
1989 | 1363 | ||
1990 | #ifdef CONFIG_PROC_FS | 1364 | printk("ip_tables: (C) 2000-2006 Netfilter Core Team\n"); |
1991 | { | ||
1992 | struct proc_dir_entry *proc; | ||
1993 | int i; | ||
1994 | |||
1995 | for (i = 0; ipt_proc_entry[i].name; i++) { | ||
1996 | proc = proc_net_create(ipt_proc_entry[i].name, 0, | ||
1997 | ipt_proc_entry[i].get_info); | ||
1998 | if (!proc) { | ||
1999 | while (--i >= 0) | ||
2000 | proc_net_remove(ipt_proc_entry[i].name); | ||
2001 | nf_unregister_sockopt(&ipt_sockopts); | ||
2002 | return -ENOMEM; | ||
2003 | } | ||
2004 | proc->owner = THIS_MODULE; | ||
2005 | } | ||
2006 | } | ||
2007 | #endif | ||
2008 | |||
2009 | printk("ip_tables: (C) 2000-2002 Netfilter core team\n"); | ||
2010 | return 0; | 1365 | return 0; |
2011 | } | 1366 | } |
2012 | 1367 | ||
2013 | static void __exit fini(void) | 1368 | static void __exit fini(void) |
2014 | { | 1369 | { |
2015 | nf_unregister_sockopt(&ipt_sockopts); | 1370 | nf_unregister_sockopt(&ipt_sockopts); |
2016 | #ifdef CONFIG_PROC_FS | 1371 | |
2017 | { | 1372 | xt_unregister_match(AF_INET, &icmp_matchstruct); |
2018 | int i; | 1373 | xt_unregister_target(AF_INET, &ipt_error_target); |
2019 | for (i = 0; ipt_proc_entry[i].name; i++) | 1374 | xt_unregister_target(AF_INET, &ipt_standard_target); |
2020 | proc_net_remove(ipt_proc_entry[i].name); | 1375 | |
2021 | } | 1376 | xt_proto_fini(AF_INET); |
2022 | #endif | ||
2023 | } | 1377 | } |
2024 | 1378 | ||
2025 | EXPORT_SYMBOL(ipt_register_table); | 1379 | EXPORT_SYMBOL(ipt_register_table); |
2026 | EXPORT_SYMBOL(ipt_unregister_table); | 1380 | EXPORT_SYMBOL(ipt_unregister_table); |
2027 | EXPORT_SYMBOL(ipt_register_match); | ||
2028 | EXPORT_SYMBOL(ipt_unregister_match); | ||
2029 | EXPORT_SYMBOL(ipt_do_table); | 1381 | EXPORT_SYMBOL(ipt_do_table); |
2030 | EXPORT_SYMBOL(ipt_register_target); | ||
2031 | EXPORT_SYMBOL(ipt_unregister_target); | ||
2032 | EXPORT_SYMBOL(ipt_find_target); | ||
2033 | |||
2034 | module_init(init); | 1382 | module_init(init); |
2035 | module_exit(fini); | 1383 | module_exit(fini); |