target/linux/etrax/patches/generic_2.6/100-netfilter_layer7_2.8.patch

   1 diff -urN linux.old/include/linux/netfilter_ipv4/ip_conntrack.h linux.dev/include/linux/netfilter_ipv4/ip_conntrack.h
   2 --- linux.old/include/linux/netfilter_ipv4/ip_conntrack.h       2007-01-01 05:17:07.000000000 +0100
   3 +++ linux.dev/include/linux/netfilter_ipv4/ip_conntrack.h       2007-01-01 05:18:48.000000000 +0100
   4 @@ -127,6 +127,15 @@
   5         /* Traversed often, so hopefully in different cacheline to top */
   6         /* These are my tuples; original and reply */
   7         struct ip_conntrack_tuple_hash tuplehash[IP_CT_DIR_MAX];
   8 +
   9 +#if defined(CONFIG_IP_NF_MATCH_LAYER7) || defined(CONFIG_IP_NF_MATCH_LAYER7_MODULE)
  10 +        struct {
  11 +                char * app_proto; /* e.g. "http". NULL before decision. "unknown" after decision if no match */
  12 +                char * app_data;  /* application layer data so far.  NULL after match decision */
  13 +                unsigned int app_data_len;
  14 +        } layer7;
  15 +#endif
  16 +
  17  };
  18
  19  struct ip_conntrack_expect
  20 diff -urN linux.old/include/linux/netfilter_ipv4/ipt_layer7.h linux.dev/include/linux/netfilter_ipv4/ipt_layer7.h
  21 --- linux.old/include/linux/netfilter_ipv4/ipt_layer7.h 1970-01-01 01:00:00.000000000 +0100
  22 +++ linux.dev/include/linux/netfilter_ipv4/ipt_layer7.h 2007-01-01 05:18:48.000000000 +0100
  23 @@ -0,0 +1,26 @@
  24 +/*
  25 +  By Matthew Strait <quadong@users.sf.net>, Dec 2003.
  26 +  http://l7-filter.sf.net
  27 +
  28 +  This program is free software; you can redistribute it and/or
  29 +  modify it under the terms of the GNU General Public License
  30 +  as published by the Free Software Foundation; either version
  31 +  2 of the License, or (at your option) any later version.
  32 +  http://www.gnu.org/licenses/gpl.txt
  33 +*/
  34 +
  35 +#ifndef _IPT_LAYER7_H
  36 +#define _IPT_LAYER7_H
  37 +
  38 +#define MAX_PATTERN_LEN 8192
  39 +#define MAX_PROTOCOL_LEN 256
  40 +
  41 +typedef char *(*proc_ipt_search) (char *, char, char *);
  42 +
  43 +struct ipt_layer7_info {
  44 +    char protocol[MAX_PROTOCOL_LEN];
  45 +    char invert:1;
  46 +    char pattern[MAX_PATTERN_LEN];
  47 +};
  48 +
  49 +#endif /* _IPT_LAYER7_H */
  50 diff -urN linux.old/net/ipv4/netfilter/ip_conntrack_core.c linux.dev/net/ipv4/netfilter/ip_conntrack_core.c
  51 --- linux.old/net/ipv4/netfilter/ip_conntrack_core.c    2007-01-01 05:17:07.000000000 +0100
  52 +++ linux.dev/net/ipv4/netfilter/ip_conntrack_core.c    2007-01-01 05:18:48.000000000 +0100
  53 @@ -337,6 +337,13 @@
  54          * too. */
  55         ip_ct_remove_expectations(ct);
  56
  57 +       #if defined(CONFIG_IP_NF_MATCH_LAYER7) || defined(CONFIG_IP_NF_MATCH_LAYER7_MODULE)
  58 +       if(ct->layer7.app_proto)
  59 +               kfree(ct->layer7.app_proto);
  60 +       if(ct->layer7.app_data)
  61 +               kfree(ct->layer7.app_data);
  62 +       #endif
  63 +
  64         /* We overload first tuple to link into unconfirmed list. */
  65         if (!is_confirmed(ct)) {
  66                 BUG_ON(list_empty(&ct->tuplehash[IP_CT_DIR_ORIGINAL].list));
  67 diff -urN linux.old/net/ipv4/netfilter/ip_conntrack_standalone.c linux.dev/net/ipv4/netfilter/ip_conntrack_standalone.c
  68 --- linux.old/net/ipv4/netfilter/ip_conntrack_standalone.c      2007-01-01 05:17:07.000000000 +0100
  69 +++ linux.dev/net/ipv4/netfilter/ip_conntrack_standalone.c      2007-01-01 05:18:48.000000000 +0100
  70 @@ -192,6 +192,12 @@
  71                 return -ENOSPC;
  72  #endif
  73
  74 +#if defined(CONFIG_IP_NF_MATCH_LAYER7) || defined(CONFIG_IP_NF_MATCH_LAYER7_MODULE)
  75 +       if(conntrack->layer7.app_proto)
  76 +               if (seq_printf(s, "l7proto=%s ",conntrack->layer7.app_proto))
  77 +                       return 1;
  78 +#endif
  79 +
  80         if (seq_printf(s, "use=%u\n", atomic_read(&conntrack->ct_general.use)))
  81                 return -ENOSPC;
  82
  83 diff -urN linux.old/net/ipv4/netfilter/ipt_layer7.c linux.dev/net/ipv4/netfilter/ipt_layer7.c
  84 --- linux.old/net/ipv4/netfilter/ipt_layer7.c   1970-01-01 01:00:00.000000000 +0100
  85 +++ linux.dev/net/ipv4/netfilter/ipt_layer7.c   2007-01-01 05:18:48.000000000 +0100
  86 @@ -0,0 +1,573 @@
  87 +/*
  88 +  Kernel module to match application layer (OSI layer 7) data in connections.
  89 +
  90 +  http://l7-filter.sf.net
  91 +
  92 +  By Matthew Strait and Ethan Sommer, 2003-2006.
  93 +
  94 +  This program is free software; you can redistribute it and/or
  95 +  modify it under the terms of the GNU General Public License
  96 +  as published by the Free Software Foundation; either version
  97 +  2 of the License, or (at your option) any later version.
  98 +  http://www.gnu.org/licenses/gpl.txt
  99 +
 100 +  Based on ipt_string.c (C) 2000 Emmanuel Roger <winfield@freegates.be>
 101 +  and cls_layer7.c (C) 2003 Matthew Strait, Ethan Sommer, Justin Levandoski
 102 +*/
 103 +
 104 +#include <linux/module.h>
 105 +#include <linux/skbuff.h>
 106 +#include <linux/netfilter_ipv4/ip_conntrack.h>
 107 +#include <linux/proc_fs.h>
 108 +#include <linux/ctype.h>
 109 +#include <net/ip.h>
 110 +#include <net/tcp.h>
 111 +#include <linux/spinlock.h>
 112 +
 113 +#include "regexp/regexp.c"
 114 +
 115 +#include <linux/netfilter_ipv4/ipt_layer7.h>
 116 +#include <linux/netfilter_ipv4/ip_tables.h>
 117 +
 118 +MODULE_AUTHOR("Matthew Strait <quadong@users.sf.net>, Ethan Sommer <sommere@users.sf.net>");
 119 +MODULE_LICENSE("GPL");
 120 +MODULE_DESCRIPTION("iptables application layer match module");
 121 +MODULE_VERSION("2.0");
 122 +
 123 +static int maxdatalen = 2048; // this is the default
 124 +module_param(maxdatalen, int, 0444);
 125 +MODULE_PARM_DESC(maxdatalen, "maximum bytes of data looked at by l7-filter");
 126 +
 127 +#ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
 128 +       #define DPRINTK(format,args...) printk(format,##args)
 129 +#else
 130 +       #define DPRINTK(format,args...)
 131 +#endif
 132 +
 133 +#define TOTAL_PACKETS master_conntrack->counters[IP_CT_DIR_ORIGINAL].packets + \
 134 +                     master_conntrack->counters[IP_CT_DIR_REPLY].packets
 135 +
 136 +/* Number of packets whose data we look at.
 137 +This can be modified through /proc/net/layer7_numpackets */
 138 +static int num_packets = 10;
 139 +
 140 +static struct pattern_cache {
 141 +       char * regex_string;
 142 +       regexp * pattern;
 143 +       struct pattern_cache * next;
 144 +} * first_pattern_cache = NULL;
 145 +
 146 +/* I'm new to locking.  Here are my assumptions:
 147 +
 148 +- No one will write to /proc/net/layer7_numpackets over and over very fast;
 149 +  if they did, nothing awful would happen.
 150 +
 151 +- This code will never be processing the same packet twice at the same time,
 152 +  because iptables rules are traversed in order.
 153 +
 154 +- It doesn't matter if two packets from different connections are in here at
 155 +  the same time, because they don't share any data.
 156 +
 157 +- It _does_ matter if two packets from the same connection (or one from a
 158 +  master and one from its child) are here at the same time.  In this case,
 159 +  we have to protect the conntracks and the list of compiled patterns.
 160 +*/
 161 +DEFINE_RWLOCK(ct_lock);
 162 +DEFINE_SPINLOCK(list_lock);
 163 +
 164 +#ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
 165 +/* Converts an unfriendly string into a friendly one by
 166 +replacing unprintables with periods and all whitespace with " ". */
 167 +static char * friendly_print(unsigned char * s)
 168 +{
 169 +       char * f = kmalloc(strlen(s) + 1, GFP_ATOMIC);
 170 +       int i;
 171 +
 172 +       if(!f) {
 173 +               if (net_ratelimit())
 174 +                       printk(KERN_ERR "layer7: out of memory in friendly_print, bailing.\n");
 175 +               return NULL;
 176 +       }
 177 +
 178 +       for(i = 0; i < strlen(s); i++){
 179 +               if(isprint(s[i]) && s[i] < 128) f[i] = s[i];
 180 +               else if(isspace(s[i]))          f[i] = ' ';
 181 +               else                            f[i] = '.';
 182 +       }
 183 +       f[i] = '\0';
 184 +       return f;
 185 +}
 186 +
 187 +static char dec2hex(int i)
 188 +{
 189 +       switch (i) {
 190 +               case 0 ... 9:
 191 +                       return (char)(i + '0');
 192 +                       break;
 193 +               case 10 ... 15:
 194 +                       return (char)(i - 10 + 'a');
 195 +                       break;
 196 +               default:
 197 +                       if (net_ratelimit())
 198 +                               printk("Problem in dec2hex\n");
 199 +                       return '\0';
 200 +       }
 201 +}
 202 +
 203 +static char * hex_print(unsigned char * s)
 204 +{
 205 +       char * g = kmalloc(strlen(s)*3 + 1, GFP_ATOMIC);
 206 +       int i;
 207 +
 208 +       if(!g) {
 209 +              if (net_ratelimit())
 210 +                       printk(KERN_ERR "layer7: out of memory in hex_print, bailing.\n");
 211 +              return NULL;
 212 +       }
 213 +
 214 +       for(i = 0; i < strlen(s); i++) {
 215 +               g[i*3    ] = dec2hex(s[i]/16);
 216 +               g[i*3 + 1] = dec2hex(s[i]%16);
 217 +               g[i*3 + 2] = ' ';
 218 +       }
 219 +       g[i*3] = '\0';
 220 +
 221 +       return g;
 222 +}
 223 +#endif // DEBUG
 224 +
 225 +/* Use instead of regcomp.  As we expect to be seeing the same regexps over and
 226 +over again, it make sense to cache the results. */
 227 +static regexp * compile_and_cache(char * regex_string, char * protocol)
 228 +{
 229 +       struct pattern_cache * node               = first_pattern_cache;
 230 +       struct pattern_cache * last_pattern_cache = first_pattern_cache;
 231 +       struct pattern_cache * tmp;
 232 +       unsigned int len;
 233 +
 234 +       while (node != NULL) {
 235 +               if (!strcmp(node->regex_string, regex_string))
 236 +               return node->pattern;
 237 +
 238 +               last_pattern_cache = node;/* points at the last non-NULL node */
 239 +               node = node->next;
 240 +       }
 241 +
 242 +       /* If we reach the end of the list, then we have not yet cached
 243 +          the pattern for this regex. Let's do that now.
 244 +          Be paranoid about running out of memory to avoid list corruption. */
 245 +       tmp = kmalloc(sizeof(struct pattern_cache), GFP_ATOMIC);
 246 +
 247 +       if(!tmp) {
 248 +               if (net_ratelimit())
 249 +                       printk(KERN_ERR "layer7: out of memory in compile_and_cache, bailing.\n");
 250 +               return NULL;
 251 +       }
 252 +
 253 +       tmp->regex_string  = kmalloc(strlen(regex_string) + 1, GFP_ATOMIC);
 254 +       tmp->pattern       = kmalloc(sizeof(struct regexp),    GFP_ATOMIC);
 255 +       tmp->next = NULL;
 256 +
 257 +       if(!tmp->regex_string || !tmp->pattern) {
 258 +               if (net_ratelimit())
 259 +                       printk(KERN_ERR "layer7: out of memory in compile_and_cache, bailing.\n");
 260 +               kfree(tmp->regex_string);
 261 +               kfree(tmp->pattern);
 262 +               kfree(tmp);
 263 +               return NULL;
 264 +       }
 265 +
 266 +       /* Ok.  The new node is all ready now. */
 267 +       node = tmp;
 268 +
 269 +       if(first_pattern_cache == NULL) /* list is empty */
 270 +               first_pattern_cache = node; /* make node the beginning */
 271 +       else
 272 +               last_pattern_cache->next = node; /* attach node to the end */
 273 +
 274 +       /* copy the string and compile the regex */
 275 +       len = strlen(regex_string);
 276 +       DPRINTK("About to compile this: \"%s\"\n", regex_string);
 277 +       node->pattern = regcomp(regex_string, &len);
 278 +       if ( !node->pattern ) {
 279 +               if (net_ratelimit())
 280 +                       printk(KERN_ERR "layer7: Error compiling regexp \"%s\" (%s)\n", regex_string, protocol);
 281 +               /* pattern is now cached as NULL, so we won't try again. */
 282 +       }
 283 +
 284 +       strcpy(node->regex_string, regex_string);
 285 +       return node->pattern;
 286 +}
 287 +
 288 +static int can_handle(const struct sk_buff *skb)
 289 +{
 290 +       if(!skb->nh.iph) /* not IP */
 291 +               return 0;
 292 +       if(skb->nh.iph->protocol != IPPROTO_TCP &&
 293 +          skb->nh.iph->protocol != IPPROTO_UDP &&
 294 +          skb->nh.iph->protocol != IPPROTO_ICMP)
 295 +               return 0;
 296 +       return 1;
 297 +}
 298 +
 299 +/* Returns offset the into the skb->data that the application data starts */
 300 +static int app_data_offset(const struct sk_buff *skb)
 301 +{
 302 +       /* In case we are ported somewhere (ebtables?) where skb->nh.iph
 303 +       isn't set, this can be gotten from 4*(skb->data[0] & 0x0f) as well. */
 304 +       int ip_hl = 4*skb->nh.iph->ihl;
 305 +
 306 +       if( skb->nh.iph->protocol == IPPROTO_TCP ) {
 307 +               /* 12 == offset into TCP header for the header length field.
 308 +               Can't get this with skb->h.th->doff because the tcphdr
 309 +               struct doesn't get set when routing (this is confirmed to be
 310 +               true in Netfilter as well as QoS.) */
 311 +               int tcp_hl = 4*(skb->data[ip_hl + 12] >> 4);
 312 +
 313 +               return ip_hl + tcp_hl;
 314 +       } else if( skb->nh.iph->protocol == IPPROTO_UDP  ) {
 315 +               return ip_hl + 8; /* UDP header is always 8 bytes */
 316 +       } else if( skb->nh.iph->protocol == IPPROTO_ICMP ) {
 317 +               return ip_hl + 8; /* ICMP header is 8 bytes */
 318 +       } else {
 319 +               if (net_ratelimit())
 320 +                       printk(KERN_ERR "layer7: tried to handle unknown protocol!\n");
 321 +               return ip_hl + 8; /* something reasonable */
 322 +       }
 323 +}
 324 +
 325 +/* handles whether there's a match when we aren't appending data anymore */
 326 +static int match_no_append(struct ip_conntrack * conntrack, struct ip_conntrack * master_conntrack,
 327 +                       enum ip_conntrack_info ctinfo, enum ip_conntrack_info master_ctinfo,
 328 +                       struct ipt_layer7_info * info)
 329 +{
 330 +       /* If we're in here, throw the app data away */
 331 +       write_lock(&ct_lock);
 332 +       if(master_conntrack->layer7.app_data != NULL) {
 333 +
 334 +       #ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
 335 +               if(!master_conntrack->layer7.app_proto) {
 336 +                       char * f = friendly_print(master_conntrack->layer7.app_data);
 337 +                       char * g = hex_print(master_conntrack->layer7.app_data);
 338 +                       DPRINTK("\nl7-filter gave up after %d bytes (%d packets):\n%s\n",
 339 +                               strlen(f), TOTAL_PACKETS, f);
 340 +                       kfree(f);
 341 +                       DPRINTK("In hex: %s\n", g);
 342 +                       kfree(g);
 343 +               }
 344 +       #endif
 345 +
 346 +               kfree(master_conntrack->layer7.app_data);
 347 +               master_conntrack->layer7.app_data = NULL; /* don't free again */
 348 +       }
 349 +       write_unlock(&ct_lock);
 350 +
 351 +       if(master_conntrack->layer7.app_proto){
 352 +               /* Here child connections set their .app_proto (for /proc/net/ip_conntrack) */
 353 +               write_lock(&ct_lock);
 354 +               if(!conntrack->layer7.app_proto) {
 355 +                       conntrack->layer7.app_proto = kmalloc(strlen(master_conntrack->layer7.app_proto)+1, GFP_ATOMIC);
 356 +                       if(!conntrack->layer7.app_proto){
 357 +                               if (net_ratelimit())
 358 +                                       printk(KERN_ERR "layer7: out of memory in match_no_append, bailing.\n");
 359 +                               write_unlock(&ct_lock);
 360 +                               return 1;
 361 +                       }
 362 +                       strcpy(conntrack->layer7.app_proto, master_conntrack->layer7.app_proto);
 363 +               }
 364 +               write_unlock(&ct_lock);
 365 +
 366 +               return (!strcmp(master_conntrack->layer7.app_proto, info->protocol));
 367 +       }
 368 +       else {
 369 +               /* If not classified, set to "unknown" to distinguish from
 370 +               connections that are still being tested. */
 371 +               write_lock(&ct_lock);
 372 +               master_conntrack->layer7.app_proto = kmalloc(strlen("unknown")+1, GFP_ATOMIC);
 373 +               if(!master_conntrack->layer7.app_proto){
 374 +                       if (net_ratelimit())
 375 +                               printk(KERN_ERR "layer7: out of memory in match_no_append, bailing.\n");
 376 +                       write_unlock(&ct_lock);
 377 +                       return 1;
 378 +               }
 379 +               strcpy(master_conntrack->layer7.app_proto, "unknown");
 380 +               write_unlock(&ct_lock);
 381 +               return 0;
 382 +       }
 383 +}
 384 +
 385 +/* add the new app data to the conntrack.  Return number of bytes added. */
 386 +static int add_data(struct ip_conntrack * master_conntrack,
 387 +                       char * app_data, int appdatalen)
 388 +{
 389 +       int length = 0, i;
 390 +       int oldlength = master_conntrack->layer7.app_data_len;
 391 +
 392 +       // This is a fix for a race condition by Deti Fliegl. However, I'm not
 393 +       // clear on whether the race condition exists or whether this really
 394 +       // fixes it.  I might just be being dense... Anyway, if it's not really
 395 +       // a fix, all it does is waste a very small amount of time.
 396 +       if(!master_conntrack->layer7.app_data) return 0;
 397 +
 398 +       /* Strip nulls. Make everything lower case (our regex lib doesn't
 399 +       do case insensitivity).  Add it to the end of the current data. */
 400 +       for(i = 0; i < maxdatalen-oldlength-1 &&
 401 +                  i < appdatalen; i++) {
 402 +               if(app_data[i] != '\0') {
 403 +                       master_conntrack->layer7.app_data[length+oldlength] =
 404 +                               /* the kernel version of tolower mungs 'upper ascii' */
 405 +                               isascii(app_data[i])? tolower(app_data[i]) : app_data[i];
 406 +                       length++;
 407 +               }
 408 +       }
 409 +
 410 +       master_conntrack->layer7.app_data[length+oldlength] = '\0';
 411 +       master_conntrack->layer7.app_data_len = length + oldlength;
 412 +
 413 +       return length;
 414 +}
 415 +
 416 +/* Returns true on match and false otherwise.  */
 417 +static int match(const struct sk_buff *skbin,
 418 +       const struct net_device *in, const struct net_device *out,
 419 +       const struct xt_match *match, const void *matchinfo,
 420 +       int offset, unsigned int protoff, int *hotdrop)
 421 +{
 422 +       /* sidestep const without getting a compiler warning... */
 423 +       struct sk_buff * skb = (struct sk_buff *)skbin;
 424 +
 425 +       struct ipt_layer7_info * info = (struct ipt_layer7_info *)matchinfo;
 426 +       enum ip_conntrack_info master_ctinfo, ctinfo;
 427 +       struct ip_conntrack *master_conntrack, *conntrack;
 428 +       unsigned char * app_data;
 429 +       unsigned int pattern_result, appdatalen;
 430 +       regexp * comppattern;
 431 +
 432 +       if(!can_handle(skb)){
 433 +               DPRINTK("layer7: This is some protocol I can't handle.\n");
 434 +               return info->invert;
 435 +       }
 436 +
 437 +       /* Treat parent & all its children together as one connection, except
 438 +       for the purpose of setting conntrack->layer7.app_proto in the actual
 439 +       connection. This makes /proc/net/ip_conntrack more satisfying. */
 440 +       if(!(conntrack = ip_conntrack_get((struct sk_buff *)skb, &ctinfo)) ||
 441 +          !(master_conntrack = ip_conntrack_get((struct sk_buff *)skb, &master_ctinfo))) {
 442 +               //DPRINTK("layer7: packet is not from a known connection, giving up.\n");
 443 +               return info->invert;
 444 +       }
 445 +
 446 +       /* Try to get a master conntrack (and its master etc) for FTP, etc. */
 447 +       while (master_ct(master_conntrack) != NULL)
 448 +               master_conntrack = master_ct(master_conntrack);
 449 +
 450 +       /* if we've classified it or seen too many packets */
 451 +       if(TOTAL_PACKETS > num_packets ||
 452 +          master_conntrack->layer7.app_proto) {
 453 +
 454 +               pattern_result = match_no_append(conntrack, master_conntrack, ctinfo, master_ctinfo, info);
 455 +
 456 +               /* skb->cb[0] == seen. Avoid doing things twice if there are two l7
 457 +               rules. I'm not sure that using cb for this purpose is correct, although
 458 +               it says "put your private variables there". But it doesn't look like it
 459 +               is being used for anything else in the skbs that make it here. How can
 460 +               I write to cb without making the compiler angry? */
 461 +               skb->cb[0] = 1; /* marking it seen here is probably irrelevant, but consistant */
 462 +
 463 +               return (pattern_result ^ info->invert);
 464 +       }
 465 +
 466 +       if(skb_is_nonlinear(skb)){
 467 +               if(skb_linearize(skb) != 0){
 468 +                       if (net_ratelimit())
 469 +                               printk(KERN_ERR "layer7: failed to linearize packet, bailing.\n");
 470 +                       return info->invert;
 471 +               }
 472 +       }
 473 +
 474 +       /* now that the skb is linearized, it's safe to set these. */
 475 +       app_data = skb->data + app_data_offset(skb);
 476 +       appdatalen = skb->tail - app_data;
 477 +
 478 +       spin_lock_bh(&list_lock);
 479 +       /* the return value gets checked later, when we're ready to use it */
 480 +       comppattern = compile_and_cache(info->pattern, info->protocol);
 481 +       spin_unlock_bh(&list_lock);
 482 +
 483 +       /* On the first packet of a connection, allocate space for app data */
 484 +       write_lock(&ct_lock);
 485 +       if(TOTAL_PACKETS == 1 && !skb->cb[0] && !master_conntrack->layer7.app_data) {
 486 +               master_conntrack->layer7.app_data = kmalloc(maxdatalen, GFP_ATOMIC);
 487 +               if(!master_conntrack->layer7.app_data){
 488 +                       if (net_ratelimit())
 489 +                               printk(KERN_ERR "layer7: out of memory in match, bailing.\n");
 490 +                       write_unlock(&ct_lock);
 491 +                       return info->invert;
 492 +               }
 493 +
 494 +               master_conntrack->layer7.app_data[0] = '\0';
 495 +       }
 496 +       write_unlock(&ct_lock);
 497 +
 498 +       /* Can be here, but unallocated, if numpackets is increased near
 499 +       the beginning of a connection */
 500 +       if(master_conntrack->layer7.app_data == NULL)
 501 +               return (info->invert); /* unmatched */
 502 +
 503 +       if(!skb->cb[0]){
 504 +               int newbytes;
 505 +               write_lock(&ct_lock);
 506 +               newbytes = add_data(master_conntrack, app_data, appdatalen);
 507 +               write_unlock(&ct_lock);
 508 +
 509 +               if(newbytes == 0) { /* didn't add any data */
 510 +                       skb->cb[0] = 1;
 511 +                       /* Didn't match before, not going to match now */
 512 +                       return info->invert;
 513 +               }
 514 +       }
 515 +
 516 +       /* If looking for "unknown", then never match.  "Unknown" means that
 517 +       we've given up; we're still trying with these packets. */
 518 +       read_lock(&ct_lock);
 519 +       if(!strcmp(info->protocol, "unknown")) {
 520 +               pattern_result = 0;
 521 +       /* If the regexp failed to compile, don't bother running it */
 522 +       } else if(comppattern && regexec(comppattern, master_conntrack->layer7.app_data)) {
 523 +               DPRINTK("layer7: matched %s\n", info->protocol);
 524 +               pattern_result = 1;
 525 +       } else pattern_result = 0;
 526 +       read_unlock(&ct_lock);
 527 +
 528 +       if(pattern_result) {
 529 +               write_lock(&ct_lock);
 530 +               master_conntrack->layer7.app_proto = kmalloc(strlen(info->protocol)+1, GFP_ATOMIC);
 531 +               if(!master_conntrack->layer7.app_proto){
 532 +                       if (net_ratelimit())
 533 +                               printk(KERN_ERR "layer7: out of memory in match, bailing.\n");
 534 +                       write_unlock(&ct_lock);
 535 +                       return (pattern_result ^ info->invert);
 536 +               }
 537 +               strcpy(master_conntrack->layer7.app_proto, info->protocol);
 538 +               write_unlock(&ct_lock);
 539 +       }
 540 +
 541 +       /* mark the packet seen */
 542 +       skb->cb[0] = 1;
 543 +
 544 +       return (pattern_result ^ info->invert);
 545 +}
 546 +
 547 +static struct ipt_match layer7_match = {
 548 +       .name = "layer7",
 549 +       .match = &match,
 550 +       .matchsize  = sizeof(struct ipt_layer7_info),
 551 +       .me = THIS_MODULE
 552 +};
 553 +
 554 +/* taken from drivers/video/modedb.c */
 555 +static int my_atoi(const char *s)
 556 +{
 557 +       int val = 0;
 558 +
 559 +       for (;; s++) {
 560 +               switch (*s) {
 561 +                       case '0'...'9':
 562 +                       val = 10*val+(*s-'0');
 563 +                       break;
 564 +               default:
 565 +                       return val;
 566 +               }
 567 +       }
 568 +}
 569 +
 570 +/* write out num_packets to userland. */
 571 +static int layer7_read_proc(char* page, char ** start, off_t off, int count,
 572 +                    int* eof, void * data)
 573 +{
 574 +       if(num_packets > 99 && net_ratelimit())
 575 +               printk(KERN_ERR "layer7: NOT REACHED. num_packets too big\n");
 576 +
 577 +       page[0] = num_packets/10 + '0';
 578 +       page[1] = num_packets%10 + '0';
 579 +       page[2] = '\n';
 580 +       page[3] = '\0';
 581 +
 582 +       *eof=1;
 583 +
 584 +       return 3;
 585 +}
 586 +
 587 +/* Read in num_packets from userland */
 588 +static int layer7_write_proc(struct file* file, const char* buffer,
 589 +                     unsigned long count, void *data)
 590 +{
 591 +       char * foo = kmalloc(count, GFP_ATOMIC);
 592 +
 593 +       if(!foo){
 594 +               if (net_ratelimit())
 595 +                       printk(KERN_ERR "layer7: out of memory, bailing. num_packets unchanged.\n");
 596 +               return count;
 597 +       }
 598 +
 599 +       if(copy_from_user(foo, buffer, count)) {
 600 +               return -EFAULT;
 601 +       }
 602 +
 603 +
 604 +       num_packets = my_atoi(foo);
 605 +       kfree (foo);
 606 +
 607 +       /* This has an arbitrary limit to make the math easier. I'm lazy.
 608 +       But anyway, 99 is a LOT! If you want more, you're doing it wrong! */
 609 +       if(num_packets > 99) {
 610 +               printk(KERN_WARNING "layer7: num_packets can't be > 99.\n");
 611 +               num_packets = 99;
 612 +       } else if(num_packets < 1) {
 613 +               printk(KERN_WARNING "layer7: num_packets can't be < 1.\n");
 614 +               num_packets = 1;
 615 +       }
 616 +
 617 +       return count;
 618 +}
 619 +
 620 +/* register the proc file */
 621 +static void layer7_init_proc(void)
 622 +{
 623 +       struct proc_dir_entry* entry;
 624 +       entry = create_proc_entry("layer7_numpackets", 0644, proc_net);
 625 +       entry->read_proc = layer7_read_proc;
 626 +       entry->write_proc = layer7_write_proc;
 627 +}
 628 +
 629 +static void layer7_cleanup_proc(void)
 630 +{
 631 +       remove_proc_entry("layer7_numpackets", proc_net);
 632 +}
 633 +
 634 +static int __init ipt_layer7_init(void)
 635 +{
 636 +       need_conntrack();
 637 +
 638 +       layer7_init_proc();
 639 +       if(maxdatalen < 1) {
 640 +               printk(KERN_WARNING "layer7: maxdatalen can't be < 1, using 1\n");
 641 +               maxdatalen = 1;
 642 +       }
 643 +       /* This is not a hard limit.  It's just here to prevent people from
 644 +       bringing their slow machines to a grinding halt. */
 645 +       else if(maxdatalen > 65536) {
 646 +               printk(KERN_WARNING "layer7: maxdatalen can't be > 65536, using 65536\n");
 647 +               maxdatalen = 65536;
 648 +       }
 649 +       return ipt_register_match(&layer7_match);
 650 +}
 651 +
 652 +static void __exit ipt_layer7_fini(void)
 653 +{
 654 +       layer7_cleanup_proc();
 655 +       ipt_unregister_match(&layer7_match);
 656 +}
 657 +
 658 +module_init(ipt_layer7_init);
 659 +module_exit(ipt_layer7_fini);
 660 diff -urN linux.old/net/ipv4/netfilter/Kconfig linux.dev/net/ipv4/netfilter/Kconfig
 661 --- linux.old/net/ipv4/netfilter/Kconfig        2007-01-01 05:17:07.000000000 +0100
 662 +++ linux.dev/net/ipv4/netfilter/Kconfig        2007-01-01 05:18:48.000000000 +0100
 663 @@ -248,6 +248,24 @@
 664
 665           To compile it as a module, choose M here.  If unsure, say N.
 666
 667 +config IP_NF_MATCH_LAYER7
 668 +       tristate "Layer 7 match support (EXPERIMENTAL)"
 669 +       depends on IP_NF_IPTABLES && IP_NF_CT_ACCT && IP_NF_CONNTRACK && EXPERIMENTAL
 670 +       help
 671 +         Say Y if you want to be able to classify connections (and their
 672 +         packets) based on regular expression matching of their application
 673 +         layer data.   This is one way to classify applications such as
 674 +         peer-to-peer filesharing systems that do not always use the same
 675 +         port.
 676 +
 677 +         To compile it as a module, choose M here.  If unsure, say N.
 678 +
 679 +config IP_NF_MATCH_LAYER7_DEBUG
 680 +       bool "Layer 7 debugging output"
 681 +       depends on IP_NF_MATCH_LAYER7
 682 +       help
 683 +         Say Y to get lots of debugging output.
 684 +
 685  config IP_NF_MATCH_TOS
 686         tristate "TOS match support"
 687         depends on IP_NF_IPTABLES
 688 diff -urN linux.old/net/ipv4/netfilter/Makefile linux.dev/net/ipv4/netfilter/Makefile
 689 --- linux.old/net/ipv4/netfilter/Makefile       2007-01-01 05:17:07.000000000 +0100
 690 +++ linux.dev/net/ipv4/netfilter/Makefile       2007-01-01 05:18:48.000000000 +0100
 691 @@ -63,6 +63,8 @@
 692  obj-$(CONFIG_IP_NF_MATCH_TTL) += ipt_ttl.o
 693  obj-$(CONFIG_IP_NF_MATCH_ADDRTYPE) += ipt_addrtype.o
 694
 695 +obj-$(CONFIG_IP_NF_MATCH_LAYER7) += ipt_layer7.o
 696 +
 697  # targets
 698  obj-$(CONFIG_IP_NF_TARGET_REJECT) += ipt_REJECT.o
 699  obj-$(CONFIG_IP_NF_TARGET_TOS) += ipt_TOS.o
 700 diff -urN linux.old/net/ipv4/netfilter/regexp/regexp.c linux.dev/net/ipv4/netfilter/regexp/regexp.c
 701 --- linux.old/net/ipv4/netfilter/regexp/regexp.c        1970-01-01 01:00:00.000000000 +0100
 702 +++ linux.dev/net/ipv4/netfilter/regexp/regexp.c        2007-01-01 05:18:48.000000000 +0100
 703 @@ -0,0 +1,1197 @@
 704 +/*
 705 + * regcomp and regexec -- regsub and regerror are elsewhere
 706 + * @(#)regexp.c        1.3 of 18 April 87
 707 + *
 708 + *     Copyright (c) 1986 by University of Toronto.
 709 + *     Written by Henry Spencer.  Not derived from licensed software.
 710 + *
 711 + *     Permission is granted to anyone to use this software for any
 712 + *     purpose on any computer system, and to redistribute it freely,
 713 + *     subject to the following restrictions:
 714 + *
 715 + *     1. The author is not responsible for the consequences of use of
 716 + *             this software, no matter how awful, even if they arise
 717 + *             from defects in it.
 718 + *
 719 + *     2. The origin of this software must not be misrepresented, either
 720 + *             by explicit claim or by omission.
 721 + *
 722 + *     3. Altered versions must be plainly marked as such, and must not
 723 + *             be misrepresented as being the original software.
 724 + *
 725 + * Beware that some of this code is subtly aware of the way operator
 726 + * precedence is structured in regular expressions.  Serious changes in
 727 + * regular-expression syntax might require a total rethink.
 728 + *
 729 + * This code was modified by Ethan Sommer to work within the kernel
 730 + * (it now uses kmalloc etc..)
 731 + *
 732 + * Modified slightly by Matthew Strait to use more modern C.
 733 + */
 734 +
 735 +#include "regexp.h"
 736 +#include "regmagic.h"
 737 +
 738 +/* added by ethan and matt.  Lets it work in both kernel and user space.
 739 +(So iptables can use it, for instance.)  Yea, it goes both ways... */
 740 +#if __KERNEL__
 741 +  #define malloc(foo) kmalloc(foo,GFP_ATOMIC)
 742 +#else
 743 +  #define printk(format,args...) printf(format,##args)
 744 +#endif
 745 +
 746 +void regerror(char * s)
 747 +{
 748 +        printk("<3>Regexp: %s\n", s);
 749 +        /* NOTREACHED */
 750 +}
 751 +
 752 +/*
 753 + * The "internal use only" fields in regexp.h are present to pass info from
 754 + * compile to execute that permits the execute phase to run lots faster on
 755 + * simple cases.  They are:
 756 + *
 757 + * regstart    char that must begin a match; '\0' if none obvious
 758 + * reganch     is the match anchored (at beginning-of-line only)?
 759 + * regmust     string (pointer into program) that match must include, or NULL
 760 + * regmlen     length of regmust string
 761 + *
 762 + * Regstart and reganch permit very fast decisions on suitable starting points
 763 + * for a match, cutting down the work a lot.  Regmust permits fast rejection
 764 + * of lines that cannot possibly match.  The regmust tests are costly enough
 765 + * that regcomp() supplies a regmust only if the r.e. contains something
 766 + * potentially expensive (at present, the only such thing detected is * or +
 767 + * at the start of the r.e., which can involve a lot of backup).  Regmlen is
 768 + * supplied because the test in regexec() needs it and regcomp() is computing
 769 + * it anyway.
 770 + */
 771 +
 772 +/*
 773 + * Structure for regexp "program".  This is essentially a linear encoding
 774 + * of a nondeterministic finite-state machine (aka syntax charts or
 775 + * "railroad normal form" in parsing technology).  Each node is an opcode
 776 + * plus a "next" pointer, possibly plus an operand.  "Next" pointers of
 777 + * all nodes except BRANCH implement concatenation; a "next" pointer with
 778 + * a BRANCH on both ends of it is connecting two alternatives.  (Here we
 779 + * have one of the subtle syntax dependencies:  an individual BRANCH (as
 780 + * opposed to a collection of them) is never concatenated with anything
 781 + * because of operator precedence.)  The operand of some types of node is
 782 + * a literal string; for others, it is a node leading into a sub-FSM.  In
 783 + * particular, the operand of a BRANCH node is the first node of the branch.
 784 + * (NB this is *not* a tree structure:  the tail of the branch connects
 785 + * to the thing following the set of BRANCHes.)  The opcodes are:
 786 + */
 787 +
 788 +/* definition  number  opnd?   meaning */
 789 +#define        END     0       /* no   End of program. */
 790 +#define        BOL     1       /* no   Match "" at beginning of line. */
 791 +#define        EOL     2       /* no   Match "" at end of line. */
 792 +#define        ANY     3       /* no   Match any one character. */
 793 +#define        ANYOF   4       /* str  Match any character in this string. */
 794 +#define        ANYBUT  5       /* str  Match any character not in this string. */
 795 +#define        BRANCH  6       /* node Match this alternative, or the next... */
 796 +#define        BACK    7       /* no   Match "", "next" ptr points backward. */
 797 +#define        EXACTLY 8       /* str  Match this string. */
 798 +#define        NOTHING 9       /* no   Match empty string. */
 799 +#define        STAR    10      /* node Match this (simple) thing 0 or more times. */
 800 +#define        PLUS    11      /* node Match this (simple) thing 1 or more times. */
 801 +#define        OPEN    20      /* no   Mark this point in input as start of #n. */
 802 +                       /*      OPEN+1 is number 1, etc. */
 803 +#define        CLOSE   30      /* no   Analogous to OPEN. */
 804 +
 805 +/*
 806 + * Opcode notes:
 807 + *
 808 + * BRANCH      The set of branches constituting a single choice are hooked
 809 + *             together with their "next" pointers, since precedence prevents
 810 + *             anything being concatenated to any individual branch.  The
 811 + *             "next" pointer of the last BRANCH in a choice points to the
 812 + *             thing following the whole choice.  This is also where the
 813 + *             final "next" pointer of each individual branch points; each
 814 + *             branch starts with the operand node of a BRANCH node.
 815 + *
 816 + * BACK                Normal "next" pointers all implicitly point forward; BACK
 817 + *             exists to make loop structures possible.
 818 + *
 819 + * STAR,PLUS   '?', and complex '*' and '+', are implemented as circular
 820 + *             BRANCH structures using BACK.  Simple cases (one character
 821 + *             per match) are implemented with STAR and PLUS for speed
 822 + *             and to minimize recursive plunges.
 823 + *
 824 + * OPEN,CLOSE  ...are numbered at compile time.
 825 + */
 826 +
 827 +/*
 828 + * A node is one char of opcode followed by two chars of "next" pointer.
 829 + * "Next" pointers are stored as two 8-bit pieces, high order first.  The
 830 + * value is a positive offset from the opcode of the node containing it.
 831 + * An operand, if any, simply follows the node.  (Note that much of the
 832 + * code generation knows about this implicit relationship.)
 833 + *
 834 + * Using two bytes for the "next" pointer is vast overkill for most things,
 835 + * but allows patterns to get big without disasters.
 836 + */
 837 +#define        OP(p)   (*(p))
 838 +#define        NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377))
 839 +#define        OPERAND(p)      ((p) + 3)
 840 +
 841 +/*
 842 + * See regmagic.h for one further detail of program structure.
 843 + */
 844 +
 845 +
 846 +/*
 847 + * Utility definitions.
 848 + */
 849 +#ifndef CHARBITS
 850 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
 851 +#else
 852 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
 853 +#endif
 854 +
 855 +#define        FAIL(m) { regerror(m); return(NULL); }
 856 +#define        ISMULT(c)       ((c) == '*' || (c) == '+' || (c) == '?')
 857 +#define        META    "^$.[()|?+*\\"
 858 +
 859 +/*
 860 + * Flags to be passed up and down.
 861 + */
 862 +#define        HASWIDTH        01      /* Known never to match null string. */
 863 +#define        SIMPLE          02      /* Simple enough to be STAR/PLUS operand. */
 864 +#define        SPSTART         04      /* Starts with * or +. */
 865 +#define        WORST           0       /* Worst case. */
 866 +
 867 +/*
 868 + * Global work variables for regcomp().
 869 + */
 870 +struct match_globals {
 871 +char *reginput;                /* String-input pointer. */
 872 +char *regbol;          /* Beginning of input, for ^ check. */
 873 +char **regstartp;      /* Pointer to startp array. */
 874 +char **regendp;                /* Ditto for endp. */
 875 +char *regparse;                /* Input-scan pointer. */
 876 +int regnpar;           /* () count. */
 877 +char regdummy;
 878 +char *regcode;         /* Code-emit pointer; &regdummy = don't. */
 879 +long regsize;          /* Code size. */
 880 +};
 881 +
 882 +/*
 883 + * Forward declarations for regcomp()'s friends.
 884 + */
 885 +#ifndef STATIC
 886 +#define        STATIC  static
 887 +#endif
 888 +STATIC char *reg(struct match_globals *g, int paren,int *flagp);
 889 +STATIC char *regbranch(struct match_globals *g, int *flagp);
 890 +STATIC char *regpiece(struct match_globals *g, int *flagp);
 891 +STATIC char *regatom(struct match_globals *g, int *flagp);
 892 +STATIC char *regnode(struct match_globals *g, char op);
 893 +STATIC char *regnext(struct match_globals *g, char *p);
 894 +STATIC void regc(struct match_globals *g, char b);
 895 +STATIC void reginsert(struct match_globals *g, char op, char *opnd);
 896 +STATIC void regtail(struct match_globals *g, char *p, char *val);
 897 +STATIC void regoptail(struct match_globals *g, char *p, char *val);
 898 +
 899 +
 900 +__kernel_size_t my_strcspn(const char *s1,const char *s2)
 901 +{
 902 +        char *scan1;
 903 +        char *scan2;
 904 +        int count;
 905 +
 906 +        count = 0;
 907 +        for (scan1 = (char *)s1; *scan1 != '\0'; scan1++) {
 908 +                for (scan2 = (char *)s2; *scan2 != '\0';)       /* ++ moved down. */
 909 +                        if (*scan1 == *scan2++)
 910 +                                return(count);
 911 +                count++;
 912 +        }
 913 +        return(count);
 914 +}
 915 +
 916 +/*
 917 + - regcomp - compile a regular expression into internal code
 918 + *
 919 + * We can't allocate space until we know how big the compiled form will be,
 920 + * but we can't compile it (and thus know how big it is) until we've got a
 921 + * place to put the code.  So we cheat:  we compile it twice, once with code
 922 + * generation turned off and size counting turned on, and once "for real".
 923 + * This also means that we don't allocate space until we are sure that the
 924 + * thing really will compile successfully, and we never have to move the
 925 + * code and thus invalidate pointers into it.  (Note that it has to be in
 926 + * one piece because free() must be able to free it all.)
 927 + *
 928 + * Beware that the optimization-preparation code in here knows about some
 929 + * of the structure of the compiled regexp.
 930 + */
 931 +regexp *
 932 +regcomp(char *exp,int *patternsize)
 933 +{
 934 +       register regexp *r;
 935 +       register char *scan;
 936 +       register char *longest;
 937 +       register int len;
 938 +       int flags;
 939 +       struct match_globals g;
 940 +
 941 +       /* commented out by ethan
 942 +          extern char *malloc();
 943 +       */
 944 +
 945 +       if (exp == NULL)
 946 +               FAIL("NULL argument");
 947 +
 948 +       /* First pass: determine size, legality. */
 949 +       g.regparse = exp;
 950 +       g.regnpar = 1;
 951 +       g.regsize = 0L;
 952 +       g.regcode = &g.regdummy;
 953 +       regc(&g, MAGIC);
 954 +       if (reg(&g, 0, &flags) == NULL)
 955 +               return(NULL);
 956 +
 957 +       /* Small enough for pointer-storage convention? */
 958 +       if (g.regsize >= 32767L)                /* Probably could be 65535L. */
 959 +               FAIL("regexp too big");
 960 +
 961 +       /* Allocate space. */
 962 +       *patternsize=sizeof(regexp) + (unsigned)g.regsize;
 963 +       r = (regexp *)malloc(sizeof(regexp) + (unsigned)g.regsize);
 964 +       if (r == NULL)
 965 +               FAIL("out of space");
 966 +
 967 +       /* Second pass: emit code. */
 968 +       g.regparse = exp;
 969 +       g.regnpar = 1;
 970 +       g.regcode = r->program;
 971 +       regc(&g, MAGIC);
 972 +       if (reg(&g, 0, &flags) == NULL)
 973 +               return(NULL);
 974 +
 975 +       /* Dig out information for optimizations. */
 976 +       r->regstart = '\0';     /* Worst-case defaults. */
 977 +       r->reganch = 0;
 978 +       r->regmust = NULL;
 979 +       r->regmlen = 0;
 980 +       scan = r->program+1;                    /* First BRANCH. */
 981 +       if (OP(regnext(&g, scan)) == END) {             /* Only one top-level choice. */
 982 +               scan = OPERAND(scan);
 983 +
 984 +               /* Starting-point info. */
 985 +               if (OP(scan) == EXACTLY)
 986 +                       r->regstart = *OPERAND(scan);
 987 +               else if (OP(scan) == BOL)
 988 +                       r->reganch++;
 989 +
 990 +               /*
 991 +                * If there's something expensive in the r.e., find the
 992 +                * longest literal string that must appear and make it the
 993 +                * regmust.  Resolve ties in favor of later strings, since
 994 +                * the regstart check works with the beginning of the r.e.
 995 +                * and avoiding duplication strengthens checking.  Not a
 996 +                * strong reason, but sufficient in the absence of others.
 997 +                */
 998 +               if (flags&SPSTART) {
 999 +                       longest = NULL;
1000 +                       len = 0;
1001 +                       for (; scan != NULL; scan = regnext(&g, scan))
1002 +                               if (OP(scan) == EXACTLY && strlen(OPERAND(scan)) >= len) {
1003 +                                       longest = OPERAND(scan);
1004 +                                       len = strlen(OPERAND(scan));
1005 +                               }
1006 +                       r->regmust = longest;
1007 +                       r->regmlen = len;
1008 +               }
1009 +       }
1010 +
1011 +       return(r);
1012 +}
1013 +
1014 +/*
1015 + - reg - regular expression, i.e. main body or parenthesized thing
1016 + *
1017 + * Caller must absorb opening parenthesis.
1018 + *
1019 + * Combining parenthesis handling with the base level of regular expression
1020 + * is a trifle forced, but the need to tie the tails of the branches to what
1021 + * follows makes it hard to avoid.
1022 + */
1023 +static char *
1024 +reg(struct match_globals *g, int paren, int *flagp /* Parenthesized? */ )
1025 +{
1026 +       register char *ret;
1027 +       register char *br;
1028 +       register char *ender;
1029 +       register int parno = 0; /* 0 makes gcc happy */
1030 +       int flags;
1031 +
1032 +       *flagp = HASWIDTH;      /* Tentatively. */
1033 +
1034 +       /* Make an OPEN node, if parenthesized. */
1035 +       if (paren) {
1036 +               if (g->regnpar >= NSUBEXP)
1037 +                       FAIL("too many ()");
1038 +               parno = g->regnpar;
1039 +               g->regnpar++;
1040 +               ret = regnode(g, OPEN+parno);
1041 +       } else
1042 +               ret = NULL;
1043 +
1044 +       /* Pick up the branches, linking them together. */
1045 +       br = regbranch(g, &flags);
1046 +       if (br == NULL)
1047 +               return(NULL);
1048 +       if (ret != NULL)
1049 +               regtail(g, ret, br);    /* OPEN -> first. */
1050 +       else
1051 +               ret = br;
1052 +       if (!(flags&HASWIDTH))
1053 +               *flagp &= ~HASWIDTH;
1054 +       *flagp |= flags&SPSTART;
1055 +       while (*g->regparse == '|') {
1056 +               g->regparse++;
1057 +               br = regbranch(g, &flags);
1058 +               if (br == NULL)
1059 +                       return(NULL);
1060 +               regtail(g, ret, br);    /* BRANCH -> BRANCH. */
1061 +               if (!(flags&HASWIDTH))
1062 +                       *flagp &= ~HASWIDTH;
1063 +               *flagp |= flags&SPSTART;
1064 +       }
1065 +
1066 +       /* Make a closing node, and hook it on the end. */
1067 +       ender = regnode(g, (paren) ? CLOSE+parno : END);
1068 +       regtail(g, ret, ender);
1069 +
1070 +       /* Hook the tails of the branches to the closing node. */
1071 +       for (br = ret; br != NULL; br = regnext(g, br))
1072 +               regoptail(g, br, ender);
1073 +
1074 +       /* Check for proper termination. */
1075 +       if (paren && *g->regparse++ != ')') {
1076 +               FAIL("unmatched ()");
1077 +       } else if (!paren && *g->regparse != '\0') {
1078 +               if (*g->regparse == ')') {
1079 +                       FAIL("unmatched ()");
1080 +               } else
1081 +                       FAIL("junk on end");    /* "Can't happen". */
1082 +               /* NOTREACHED */
1083 +       }
1084 +
1085 +       return(ret);
1086 +}
1087 +
1088 +/*
1089 + - regbranch - one alternative of an | operator
1090 + *
1091 + * Implements the concatenation operator.
1092 + */
1093 +static char *
1094 +regbranch(struct match_globals *g, int *flagp)
1095 +{
1096 +       register char *ret;
1097 +       register char *chain;
1098 +       register char *latest;
1099 +       int flags;
1100 +
1101 +       *flagp = WORST;         /* Tentatively. */
1102 +
1103 +       ret = regnode(g, BRANCH);
1104 +       chain = NULL;
1105 +       while (*g->regparse != '\0' && *g->regparse != '|' && *g->regparse != ')') {
1106 +               latest = regpiece(g, &flags);
1107 +               if (latest == NULL)
1108 +                       return(NULL);
1109 +               *flagp |= flags&HASWIDTH;
1110 +               if (chain == NULL)      /* First piece. */
1111 +                       *flagp |= flags&SPSTART;
1112 +               else
1113 +                       regtail(g, chain, latest);
1114 +               chain = latest;
1115 +       }
1116 +       if (chain == NULL)      /* Loop ran zero times. */
1117 +               (void) regnode(g, NOTHING);
1118 +
1119 +       return(ret);
1120 +}
1121 +
1122 +/*
1123 + - regpiece - something followed by possible [*+?]
1124 + *
1125 + * Note that the branching code sequences used for ? and the general cases
1126 + * of * and + are somewhat optimized:  they use the same NOTHING node as
1127 + * both the endmarker for their branch list and the body of the last branch.
1128 + * It might seem that this node could be dispensed with entirely, but the
1129 + * endmarker role is not redundant.
1130 + */
1131 +static char *
1132 +regpiece(struct match_globals *g, int *flagp)
1133 +{
1134 +       register char *ret;
1135 +       register char op;
1136 +       register char *next;
1137 +       int flags;
1138 +
1139 +       ret = regatom(g, &flags);
1140 +       if (ret == NULL)
1141 +               return(NULL);
1142 +
1143 +       op = *g->regparse;
1144 +       if (!ISMULT(op)) {
1145 +               *flagp = flags;
1146 +               return(ret);
1147 +       }
1148 +
1149 +       if (!(flags&HASWIDTH) && op != '?')
1150 +               FAIL("*+ operand could be empty");
1151 +       *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
1152 +
1153 +       if (op == '*' && (flags&SIMPLE))
1154 +               reginsert(g, STAR, ret);
1155 +       else if (op == '*') {
1156 +               /* Emit x* as (x&|), where & means "self". */
1157 +               reginsert(g, BRANCH, ret);                      /* Either x */
1158 +               regoptail(g, ret, regnode(g, BACK));            /* and loop */
1159 +               regoptail(g, ret, ret);                 /* back */
1160 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
1161 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
1162 +       } else if (op == '+' && (flags&SIMPLE))
1163 +               reginsert(g, PLUS, ret);
1164 +       else if (op == '+') {
1165 +               /* Emit x+ as x(&|), where & means "self". */
1166 +               next = regnode(g, BRANCH);                      /* Either */
1167 +               regtail(g, ret, next);
1168 +               regtail(g, regnode(g, BACK), ret);              /* loop back */
1169 +               regtail(g, next, regnode(g, BRANCH));           /* or */
1170 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
1171 +       } else if (op == '?') {
1172 +               /* Emit x? as (x|) */
1173 +               reginsert(g, BRANCH, ret);                      /* Either x */
1174 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
1175 +               next = regnode(g, NOTHING);             /* null. */
1176 +               regtail(g, ret, next);
1177 +               regoptail(g, ret, next);
1178 +       }
1179 +       g->regparse++;
1180 +       if (ISMULT(*g->regparse))
1181 +               FAIL("nested *?+");
1182 +
1183 +       return(ret);
1184 +}
1185 +
1186 +/*
1187 + - regatom - the lowest level
1188 + *
1189 + * Optimization:  gobbles an entire sequence of ordinary characters so that
1190 + * it can turn them into a single node, which is smaller to store and
1191 + * faster to run.  Backslashed characters are exceptions, each becoming a
1192 + * separate node; the code is simpler that way and it's not worth fixing.
1193 + */
1194 +static char *
1195 +regatom(struct match_globals *g, int *flagp)
1196 +{
1197 +       register char *ret;
1198 +       int flags;
1199 +
1200 +       *flagp = WORST;         /* Tentatively. */
1201 +
1202 +       switch (*g->regparse++) {
1203 +       case '^':
1204 +               ret = regnode(g, BOL);
1205 +               break;
1206 +       case '$':
1207 +               ret = regnode(g, EOL);
1208 +               break;
1209 +       case '.':
1210 +               ret = regnode(g, ANY);
1211 +               *flagp |= HASWIDTH|SIMPLE;
1212 +               break;
1213 +       case '[': {
1214 +                       register int class;
1215 +                       register int classend;
1216 +
1217 +                       if (*g->regparse == '^') {      /* Complement of range. */
1218 +                               ret = regnode(g, ANYBUT);
1219 +                               g->regparse++;
1220 +                       } else
1221 +                               ret = regnode(g, ANYOF);
1222 +                       if (*g->regparse == ']' || *g->regparse == '-')
1223 +                               regc(g, *g->regparse++);
1224 +                       while (*g->regparse != '\0' && *g->regparse != ']') {
1225 +                               if (*g->regparse == '-') {
1226 +                                       g->regparse++;
1227 +                                       if (*g->regparse == ']' || *g->regparse == '\0')
1228 +                                               regc(g, '-');
1229 +                                       else {
1230 +                                               class = UCHARAT(g->regparse-2)+1;
1231 +                                               classend = UCHARAT(g->regparse);
1232 +                                               if (class > classend+1)
1233 +                                                       FAIL("invalid [] range");
1234 +                                               for (; class <= classend; class++)
1235 +                                                       regc(g, class);
1236 +                                               g->regparse++;
1237 +                                       }
1238 +                               } else
1239 +                                       regc(g, *g->regparse++);
1240 +                       }
1241 +                       regc(g, '\0');
1242 +                       if (*g->regparse != ']')
1243 +                               FAIL("unmatched []");
1244 +                       g->regparse++;
1245 +                       *flagp |= HASWIDTH|SIMPLE;
1246 +               }
1247 +               break;
1248 +       case '(':
1249 +               ret = reg(g, 1, &flags);
1250 +               if (ret == NULL)
1251 +                       return(NULL);
1252 +               *flagp |= flags&(HASWIDTH|SPSTART);
1253 +               break;
1254 +       case '\0':
1255 +       case '|':
1256 +       case ')':
1257 +               FAIL("internal urp");   /* Supposed to be caught earlier. */
1258 +               break;
1259 +       case '?':
1260 +       case '+':
1261 +       case '*':
1262 +               FAIL("?+* follows nothing");
1263 +               break;
1264 +       case '\\':
1265 +               if (*g->regparse == '\0')
1266 +                       FAIL("trailing \\");
1267 +               ret = regnode(g, EXACTLY);
1268 +               regc(g, *g->regparse++);
1269 +               regc(g, '\0');
1270 +               *flagp |= HASWIDTH|SIMPLE;
1271 +               break;
1272 +       default: {
1273 +                       register int len;
1274 +                       register char ender;
1275 +
1276 +                       g->regparse--;
1277 +                       len = my_strcspn((const char *)g->regparse, (const char *)META);
1278 +                       if (len <= 0)
1279 +                               FAIL("internal disaster");
1280 +                       ender = *(g->regparse+len);
1281 +                       if (len > 1 && ISMULT(ender))
1282 +                               len--;          /* Back off clear of ?+* operand. */
1283 +                       *flagp |= HASWIDTH;
1284 +                       if (len == 1)
1285 +                               *flagp |= SIMPLE;
1286 +                       ret = regnode(g, EXACTLY);
1287 +                       while (len > 0) {
1288 +                               regc(g, *g->regparse++);
1289 +                               len--;
1290 +                       }
1291 +                       regc(g, '\0');
1292 +               }
1293 +               break;
1294 +       }
1295 +
1296 +       return(ret);
1297 +}
1298 +
1299 +/*
1300 + - regnode - emit a node
1301 + */
1302 +static char *                  /* Location. */
1303 +regnode(struct match_globals *g, char op)
1304 +{
1305 +       register char *ret;
1306 +       register char *ptr;
1307 +
1308 +       ret = g->regcode;
1309 +       if (ret == &g->regdummy) {
1310 +               g->regsize += 3;
1311 +               return(ret);
1312 +       }
1313 +
1314 +       ptr = ret;
1315 +       *ptr++ = op;
1316 +       *ptr++ = '\0';          /* Null "next" pointer. */
1317 +       *ptr++ = '\0';
1318 +       g->regcode = ptr;
1319 +
1320 +       return(ret);
1321 +}
1322 +
1323 +/*
1324 + - regc - emit (if appropriate) a byte of code
1325 + */
1326 +static void
1327 +regc(struct match_globals *g, char b)
1328 +{
1329 +       if (g->regcode != &g->regdummy)
1330 +               *g->regcode++ = b;
1331 +       else
1332 +               g->regsize++;
1333 +}
1334 +
1335 +/*
1336 + - reginsert - insert an operator in front of already-emitted operand
1337 + *
1338 + * Means relocating the operand.
1339 + */
1340 +static void
1341 +reginsert(struct match_globals *g, char op, char* opnd)
1342 +{
1343 +       register char *src;
1344 +       register char *dst;
1345 +       register char *place;
1346 +
1347 +       if (g->regcode == &g->regdummy) {
1348 +               g->regsize += 3;
1349 +               return;
1350 +       }
1351 +
1352 +       src = g->regcode;
1353 +       g->regcode += 3;
1354 +       dst = g->regcode;
1355 +       while (src > opnd)
1356 +               *--dst = *--src;
1357 +
1358 +       place = opnd;           /* Op node, where operand used to be. */
1359 +       *place++ = op;
1360 +       *place++ = '\0';
1361 +       *place++ = '\0';
1362 +}
1363 +
1364 +/*
1365 + - regtail - set the next-pointer at the end of a node chain
1366 + */
1367 +static void
1368 +regtail(struct match_globals *g, char *p, char *val)
1369 +{
1370 +       register char *scan;
1371 +       register char *temp;
1372 +       register int offset;
1373 +
1374 +       if (p == &g->regdummy)
1375 +               return;
1376 +
1377 +       /* Find last node. */
1378 +       scan = p;
1379 +       for (;;) {
1380 +               temp = regnext(g, scan);
1381 +               if (temp == NULL)
1382 +                       break;
1383 +               scan = temp;
1384 +       }
1385 +
1386 +       if (OP(scan) == BACK)
1387 +               offset = scan - val;
1388 +       else
1389 +               offset = val - scan;
1390 +       *(scan+1) = (offset>>8)&0377;
1391 +       *(scan+2) = offset&0377;
1392 +}
1393 +
1394 +/*
1395 + - regoptail - regtail on operand of first argument; nop if operandless
1396 + */
1397 +static void
1398 +regoptail(struct match_globals *g, char *p, char *val)
1399 +{
1400 +       /* "Operandless" and "op != BRANCH" are synonymous in practice. */
1401 +       if (p == NULL || p == &g->regdummy || OP(p) != BRANCH)
1402 +               return;
1403 +       regtail(g, OPERAND(p), val);
1404 +}
1405 +
1406 +/*
1407 + * regexec and friends
1408 + */
1409 +
1410 +
1411 +/*
1412 + * Forwards.
1413 + */
1414 +STATIC int regtry(struct match_globals *g, regexp *prog, char *string);
1415 +STATIC int regmatch(struct match_globals *g, char *prog);
1416 +STATIC int regrepeat(struct match_globals *g, char *p);
1417 +
1418 +#ifdef DEBUG
1419 +int regnarrate = 0;
1420 +void regdump();
1421 +STATIC char *regprop(char *op);
1422 +#endif
1423 +
1424 +/*
1425 + - regexec - match a regexp against a string
1426 + */
1427 +int
1428 +regexec(regexp *prog, char *string)
1429 +{
1430 +       register char *s;
1431 +       struct match_globals g;
1432 +
1433 +       /* Be paranoid... */
1434 +       if (prog == NULL || string == NULL) {
1435 +               printk("<3>Regexp: NULL parameter\n");
1436 +               return(0);
1437 +       }
1438 +
1439 +       /* Check validity of program. */
1440 +       if (UCHARAT(prog->program) != MAGIC) {
1441 +               printk("<3>Regexp: corrupted program\n");
1442 +               return(0);
1443 +       }
1444 +
1445 +       /* If there is a "must appear" string, look for it. */
1446 +       if (prog->regmust != NULL) {
1447 +               s = string;
1448 +               while ((s = strchr(s, prog->regmust[0])) != NULL) {
1449 +                       if (strncmp(s, prog->regmust, prog->regmlen) == 0)
1450 +                               break;  /* Found it. */
1451 +                       s++;
1452 +               }
1453 +               if (s == NULL)  /* Not present. */
1454 +                       return(0);
1455 +       }
1456 +
1457 +       /* Mark beginning of line for ^ . */
1458 +       g.regbol = string;
1459 +
1460 +       /* Simplest case:  anchored match need be tried only once. */
1461 +       if (prog->reganch)
1462 +               return(regtry(&g, prog, string));
1463 +
1464 +       /* Messy cases:  unanchored match. */
1465 +       s = string;
1466 +       if (prog->regstart != '\0')
1467 +               /* We know what char it must start with. */
1468 +               while ((s = strchr(s, prog->regstart)) != NULL) {
1469 +                       if (regtry(&g, prog, s))
1470 +                               return(1);
1471 +                       s++;
1472 +               }
1473 +       else
1474 +               /* We don't -- general case. */
1475 +               do {
1476 +                       if (regtry(&g, prog, s))
1477 +                               return(1);
1478 +               } while (*s++ != '\0');
1479 +
1480 +       /* Failure. */
1481 +       return(0);
1482 +}
1483 +
1484 +/*
1485 + - regtry - try match at specific point
1486 + */
1487 +static int                     /* 0 failure, 1 success */
1488 +regtry(struct match_globals *g, regexp *prog, char *string)
1489 +{
1490 +       register int i;
1491 +       register char **sp;
1492 +       register char **ep;
1493 +
1494 +       g->reginput = string;
1495 +       g->regstartp = prog->startp;
1496 +       g->regendp = prog->endp;
1497 +
1498 +       sp = prog->startp;
1499 +       ep = prog->endp;
1500 +       for (i = NSUBEXP; i > 0; i--) {
1501 +               *sp++ = NULL;
1502 +               *ep++ = NULL;
1503 +       }
1504 +       if (regmatch(g, prog->program + 1)) {
1505 +               prog->startp[0] = string;
1506 +               prog->endp[0] = g->reginput;
1507 +               return(1);
1508 +       } else
1509 +               return(0);
1510 +}
1511 +
1512 +/*
1513 + - regmatch - main matching routine
1514 + *
1515 + * Conceptually the strategy is simple:  check to see whether the current
1516 + * node matches, call self recursively to see whether the rest matches,
1517 + * and then act accordingly.  In practice we make some effort to avoid
1518 + * recursion, in particular by going through "ordinary" nodes (that don't
1519 + * need to know whether the rest of the match failed) by a loop instead of
1520 + * by recursion.
1521 + */
1522 +static int                     /* 0 failure, 1 success */
1523 +regmatch(struct match_globals *g, char *prog)
1524 +{
1525 +       register char *scan = prog; /* Current node. */
1526 +       char *next;                 /* Next node. */
1527 +
1528 +#ifdef DEBUG
1529 +       if (scan != NULL && regnarrate)
1530 +               fprintf(stderr, "%s(\n", regprop(scan));
1531 +#endif
1532 +       while (scan != NULL) {
1533 +#ifdef DEBUG
1534 +               if (regnarrate)
1535 +                       fprintf(stderr, "%s...\n", regprop(scan));
1536 +#endif
1537 +               next = regnext(g, scan);
1538 +
1539 +               switch (OP(scan)) {
1540 +               case BOL:
1541 +                       if (g->reginput != g->regbol)
1542 +                               return(0);
1543 +                       break;
1544 +               case EOL:
1545 +                       if (*g->reginput != '\0')
1546 +                               return(0);
1547 +                       break;
1548 +               case ANY:
1549 +                       if (*g->reginput == '\0')
1550 +                               return(0);
1551 +                       g->reginput++;
1552 +                       break;
1553 +               case EXACTLY: {
1554 +                               register int len;
1555 +                               register char *opnd;
1556 +
1557 +                               opnd = OPERAND(scan);
1558 +                               /* Inline the first character, for speed. */
1559 +                               if (*opnd != *g->reginput)
1560 +                                       return(0);
1561 +                               len = strlen(opnd);
1562 +                               if (len > 1 && strncmp(opnd, g->reginput, len) != 0)
1563 +                                       return(0);
1564 +                               g->reginput += len;
1565 +                       }
1566 +                       break;
1567 +               case ANYOF:
1568 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) == NULL)
1569 +                               return(0);
1570 +                       g->reginput++;
1571 +                       break;
1572 +               case ANYBUT:
1573 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) != NULL)
1574 +                               return(0);
1575 +                       g->reginput++;
1576 +                       break;
1577 +               case NOTHING:
1578 +               case BACK:
1579 +                       break;
1580 +               case OPEN+1:
1581 +               case OPEN+2:
1582 +               case OPEN+3:
1583 +               case OPEN+4:
1584 +               case OPEN+5:
1585 +               case OPEN+6:
1586 +               case OPEN+7:
1587 +               case OPEN+8:
1588 +               case OPEN+9: {
1589 +                               register int no;
1590 +                               register char *save;
1591 +
1592 +                               no = OP(scan) - OPEN;
1593 +                               save = g->reginput;
1594 +
1595 +                               if (regmatch(g, next)) {
1596 +                                       /*
1597 +                                        * Don't set startp if some later
1598 +                                        * invocation of the same parentheses
1599 +                                        * already has.
1600 +                                        */
1601 +                                       if (g->regstartp[no] == NULL)
1602 +                                               g->regstartp[no] = save;
1603 +                                       return(1);
1604 +                               } else
1605 +                                       return(0);
1606 +                       }
1607 +                       break;
1608 +               case CLOSE+1:
1609 +               case CLOSE+2:
1610 +               case CLOSE+3:
1611 +               case CLOSE+4:
1612 +               case CLOSE+5:
1613 +               case CLOSE+6:
1614 +               case CLOSE+7:
1615 +               case CLOSE+8:
1616 +               case CLOSE+9:
1617 +                       {
1618 +                               register int no;
1619 +                               register char *save;
1620 +
1621 +                               no = OP(scan) - CLOSE;
1622 +                               save = g->reginput;
1623 +
1624 +                               if (regmatch(g, next)) {
1625 +                                       /*
1626 +                                        * Don't set endp if some later
1627 +                                        * invocation of the same parentheses
1628 +                                        * already has.
1629 +                                        */
1630 +                                       if (g->regendp[no] == NULL)
1631 +                                               g->regendp[no] = save;
1632 +                                       return(1);
1633 +                               } else
1634 +                                       return(0);
1635 +                       }
1636 +                       break;
1637 +               case BRANCH: {
1638 +                               register char *save;
1639 +
1640 +                               if (OP(next) != BRANCH)         /* No choice. */
1641 +                                       next = OPERAND(scan);   /* Avoid recursion. */
1642 +                               else {
1643 +                                       do {
1644 +                                               save = g->reginput;
1645 +                                               if (regmatch(g, OPERAND(scan)))
1646 +                                                       return(1);
1647 +                                               g->reginput = save;
1648 +                                               scan = regnext(g, scan);
1649 +                                       } while (scan != NULL && OP(scan) == BRANCH);
1650 +                                       return(0);
1651 +                                       /* NOTREACHED */
1652 +                               }
1653 +                       }
1654 +                       break;
1655 +               case STAR:
1656 +               case PLUS: {
1657 +                               register char nextch;
1658 +                               register int no;
1659 +                               register char *save;
1660 +                               register int min;
1661 +
1662 +                               /*
1663 +                                * Lookahead to avoid useless match attempts
1664 +                                * when we know what character comes next.
1665 +                                */
1666 +                               nextch = '\0';
1667 +                               if (OP(next) == EXACTLY)
1668 +                                       nextch = *OPERAND(next);
1669 +                               min = (OP(scan) == STAR) ? 0 : 1;
1670 +                               save = g->reginput;
1671 +                               no = regrepeat(g, OPERAND(scan));
1672 +                               while (no >= min) {
1673 +                                       /* If it could work, try it. */
1674 +                                       if (nextch == '\0' || *g->reginput == nextch)
1675 +                                               if (regmatch(g, next))
1676 +                                                       return(1);
1677 +                                       /* Couldn't or didn't -- back up. */
1678 +                                       no--;
1679 +                                       g->reginput = save + no;
1680 +                               }
1681 +                               return(0);
1682 +                       }
1683 +                       break;
1684 +               case END:
1685 +                       return(1);      /* Success! */
1686 +                       break;
1687 +               default:
1688 +                       printk("<3>Regexp: memory corruption\n");
1689 +                       return(0);
1690 +                       break;
1691 +               }
1692 +
1693 +               scan = next;
1694 +       }
1695 +
1696 +       /*
1697 +        * We get here only if there's trouble -- normally "case END" is
1698 +        * the terminating point.
1699 +        */
1700 +       printk("<3>Regexp: corrupted pointers\n");
1701 +       return(0);
1702 +}
1703 +
1704 +/*
1705 + - regrepeat - repeatedly match something simple, report how many
1706 + */
1707 +static int
1708 +regrepeat(struct match_globals *g, char *p)
1709 +{
1710 +       register int count = 0;
1711 +       register char *scan;
1712 +       register char *opnd;
1713 +
1714 +       scan = g->reginput;
1715 +       opnd = OPERAND(p);
1716 +       switch (OP(p)) {
1717 +       case ANY:
1718 +               count = strlen(scan);
1719 +               scan += count;
1720 +               break;
1721 +       case EXACTLY:
1722 +               while (*opnd == *scan) {
1723 +                       count++;
1724 +                       scan++;
1725 +               }
1726 +               break;
1727 +       case ANYOF:
1728 +               while (*scan != '\0' && strchr(opnd, *scan) != NULL) {
1729 +                       count++;
1730 +                       scan++;
1731 +               }
1732 +               break;
1733 +       case ANYBUT:
1734 +               while (*scan != '\0' && strchr(opnd, *scan) == NULL) {
1735 +                       count++;
1736 +                       scan++;
1737 +               }
1738 +               break;
1739 +       default:                /* Oh dear.  Called inappropriately. */
1740 +               printk("<3>Regexp: internal foulup\n");
1741 +               count = 0;      /* Best compromise. */
1742 +               break;
1743 +       }
1744 +       g->reginput = scan;
1745 +
1746 +       return(count);
1747 +}
1748 +
1749 +/*
1750 + - regnext - dig the "next" pointer out of a node
1751 + */
1752 +static char*
1753 +regnext(struct match_globals *g, char *p)
1754 +{
1755 +       register int offset;
1756 +
1757 +       if (p == &g->regdummy)
1758 +               return(NULL);
1759 +
1760 +       offset = NEXT(p);
1761 +       if (offset == 0)
1762 +               return(NULL);
1763 +
1764 +       if (OP(p) == BACK)
1765 +               return(p-offset);
1766 +       else
1767 +               return(p+offset);
1768 +}
1769 +
1770 +#ifdef DEBUG
1771 +
1772 +STATIC char *regprop();
1773 +
1774 +/*
1775 + - regdump - dump a regexp onto stdout in vaguely comprehensible form
1776 + */
1777 +void
1778 +regdump(regexp *r)
1779 +{
1780 +       register char *s;
1781 +       register char op = EXACTLY;     /* Arbitrary non-END op. */
1782 +       register char *next;
1783 +       /* extern char *strchr(); */
1784 +
1785 +
1786 +       s = r->program + 1;
1787 +       while (op != END) {     /* While that wasn't END last time... */
1788 +               op = OP(s);
1789 +               printf("%2d%s", s-r->program, regprop(s));      /* Where, what. */
1790 +               next = regnext(s);
1791 +               if (next == NULL)               /* Next ptr. */
1792 +                       printf("(0)");
1793 +               else
1794 +                       printf("(%d)", (s-r->program)+(next-s));
1795 +               s += 3;
1796 +               if (op == ANYOF || op == ANYBUT || op == EXACTLY) {
1797 +                       /* Literal string, where present. */
1798 +                       while (*s != '\0') {
1799 +                               putchar(*s);
1800 +                               s++;
1801 +                       }
1802 +                       s++;
1803 +               }
1804 +               putchar('\n');
1805 +       }
1806 +
1807 +       /* Header fields of interest. */
1808 +       if (r->regstart != '\0')
1809 +               printf("start `%c' ", r->regstart);
1810 +       if (r->reganch)
1811 +               printf("anchored ");
1812 +       if (r->regmust != NULL)
1813 +               printf("must have \"%s\"", r->regmust);
1814 +       printf("\n");
1815 +}
1816 +
1817 +/*
1818 + - regprop - printable representation of opcode
1819 + */
1820 +static char *
1821 +regprop(char *op)
1822 +{
1823 +#define BUFLEN 50
1824 +       register char *p;
1825 +       static char buf[BUFLEN];
1826 +
1827 +       strcpy(buf, ":");
1828 +
1829 +       switch (OP(op)) {
1830 +       case BOL:
1831 +               p = "BOL";
1832 +               break;
1833 +       case EOL:
1834 +               p = "EOL";
1835 +               break;
1836 +       case ANY:
1837 +               p = "ANY";
1838 +               break;
1839 +       case ANYOF:
1840 +               p = "ANYOF";
1841 +               break;
1842 +       case ANYBUT:
1843 +               p = "ANYBUT";
1844 +               break;
1845 +       case BRANCH:
1846 +               p = "BRANCH";
1847 +               break;
1848 +       case EXACTLY:
1849 +               p = "EXACTLY";
1850 +               break;
1851 +       case NOTHING:
1852 +               p = "NOTHING";
1853 +               break;
1854 +       case BACK:
1855 +               p = "BACK";
1856 +               break;
1857 +       case END:
1858 +               p = "END";
1859 +               break;
1860 +       case OPEN+1:
1861 +       case OPEN+2:
1862 +       case OPEN+3:
1863 +       case OPEN+4:
1864 +       case OPEN+5:
1865 +       case OPEN+6:
1866 +       case OPEN+7:
1867 +       case OPEN+8:
1868 +       case OPEN+9:
1869 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "OPEN%d", OP(op)-OPEN);
1870 +               p = NULL;
1871 +               break;
1872 +       case CLOSE+1:
1873 +       case CLOSE+2:
1874 +       case CLOSE+3:
1875 +       case CLOSE+4:
1876 +       case CLOSE+5:
1877 +       case CLOSE+6:
1878 +       case CLOSE+7:
1879 +       case CLOSE+8:
1880 +       case CLOSE+9:
1881 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "CLOSE%d", OP(op)-CLOSE);
1882 +               p = NULL;
1883 +               break;
1884 +       case STAR:
1885 +               p = "STAR";
1886 +               break;
1887 +       case PLUS:
1888 +               p = "PLUS";
1889 +               break;
1890 +       default:
1891 +               printk("<3>Regexp: corrupted opcode\n");
1892 +               break;
1893 +       }
1894 +       if (p != NULL)
1895 +               strncat(buf, p, BUFLEN-strlen(buf));
1896 +       return(buf);
1897 +}
1898 +#endif
1899 +
1900 +
1901 diff -urN linux.old/net/ipv4/netfilter/regexp/regexp.h linux.dev/net/ipv4/netfilter/regexp/regexp.h
1902 --- linux.old/net/ipv4/netfilter/regexp/regexp.h        1970-01-01 01:00:00.000000000 +0100
1903 +++ linux.dev/net/ipv4/netfilter/regexp/regexp.h        2007-01-01 05:18:48.000000000 +0100
1904 @@ -0,0 +1,41 @@
1905 +/*
1906 + * Definitions etc. for regexp(3) routines.
1907 + *
1908 + * Caveat:  this is V8 regexp(3) [actually, a reimplementation thereof],
1909 + * not the System V one.
1910 + */
1911 +
1912 +#ifndef REGEXP_H
1913 +#define REGEXP_H
1914 +
1915 +
1916 +/*
1917 +http://www.opensource.apple.com/darwinsource/10.3/expect-1/expect/expect.h ,
1918 +which contains a version of this library, says:
1919 +
1920 + *
1921 + * NSUBEXP must be at least 10, and no greater than 117 or the parser
1922 + * will not work properly.
1923 + *
1924 +
1925 +However, it looks rather like this library is limited to 10.  If you think
1926 +otherwise, let us know.
1927 +*/
1928 +
1929 +#define NSUBEXP  10
1930 +typedef struct regexp {
1931 +       char *startp[NSUBEXP];
1932 +       char *endp[NSUBEXP];
1933 +       char regstart;          /* Internal use only. */
1934 +       char reganch;           /* Internal use only. */
1935 +       char *regmust;          /* Internal use only. */
1936 +       int regmlen;            /* Internal use only. */
1937 +       char program[1];        /* Unwarranted chumminess with compiler. */
1938 +} regexp;
1939 +
1940 +regexp * regcomp(char *exp, int *patternsize);
1941 +int regexec(regexp *prog, char *string);
1942 +void regsub(regexp *prog, char *source, char *dest);
1943 +void regerror(char *s);
1944 +
1945 +#endif
1946 diff -urN linux.old/net/ipv4/netfilter/regexp/regmagic.h linux.dev/net/ipv4/netfilter/regexp/regmagic.h
1947 --- linux.old/net/ipv4/netfilter/regexp/regmagic.h      1970-01-01 01:00:00.000000000 +0100
1948 +++ linux.dev/net/ipv4/netfilter/regexp/regmagic.h      2007-01-01 05:18:48.000000000 +0100
1949 @@ -0,0 +1,5 @@
1950 +/*
1951 + * The first byte of the regexp internal "program" is actually this magic
1952 + * number; the start node begins in the second byte.
1953 + */
1954 +#define        MAGIC   0234
1955 diff -urN linux.old/net/ipv4/netfilter/regexp/regsub.c linux.dev/net/ipv4/netfilter/regexp/regsub.c
1956 --- linux.old/net/ipv4/netfilter/regexp/regsub.c        1970-01-01 01:00:00.000000000 +0100
1957 +++ linux.dev/net/ipv4/netfilter/regexp/regsub.c        2007-01-01 05:18:48.000000000 +0100
1958 @@ -0,0 +1,95 @@
1959 +/*
1960 + * regsub
1961 + * @(#)regsub.c        1.3 of 2 April 86
1962 + *
1963 + *     Copyright (c) 1986 by University of Toronto.
1964 + *     Written by Henry Spencer.  Not derived from licensed software.
1965 + *
1966 + *     Permission is granted to anyone to use this software for any
1967 + *     purpose on any computer system, and to redistribute it freely,
1968 + *     subject to the following restrictions:
1969 + *
1970 + *     1. The author is not responsible for the consequences of use of
1971 + *             this software, no matter how awful, even if they arise
1972 + *             from defects in it.
1973 + *
1974 + *     2. The origin of this software must not be misrepresented, either
1975 + *             by explicit claim or by omission.
1976 + *
1977 + *     3. Altered versions must be plainly marked as such, and must not
1978 + *             be misrepresented as being the original software.
1979 + *
1980 + *
1981 + * This code was modified by Ethan Sommer to work within the kernel
1982 + * (it now uses kmalloc etc..)
1983 + *
1984 + */
1985 +#include "regexp.h"
1986 +#include "regmagic.h"
1987 +#include <linux/string.h>
1988 +
1989 +
1990 +#ifndef CHARBITS
1991 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
1992 +#else
1993 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
1994 +#endif
1995 +
1996 +#if 0
1997 +//void regerror(char * s)
1998 +//{
1999 +//        printk("regexp(3): %s", s);
2000 +//        /* NOTREACHED */
2001 +//}
2002 +#endif
2003 +
2004 +/*
2005 + - regsub - perform substitutions after a regexp match
2006 + */
2007 +void
2008 +regsub(regexp * prog, char * source, char * dest)
2009 +{
2010 +       register char *src;
2011 +       register char *dst;
2012 +       register char c;
2013 +       register int no;
2014 +       register int len;
2015 +
2016 +       /* Not necessary and gcc doesn't like it -MLS */
2017 +       /*extern char *strncpy();*/
2018 +
2019 +       if (prog == NULL || source == NULL || dest == NULL) {
2020 +               regerror("NULL parm to regsub");
2021 +               return;
2022 +       }
2023 +       if (UCHARAT(prog->program) != MAGIC) {
2024 +               regerror("damaged regexp fed to regsub");
2025 +               return;
2026 +       }
2027 +
2028 +       src = source;
2029 +       dst = dest;
2030 +       while ((c = *src++) != '\0') {
2031 +               if (c == '&')
2032 +                       no = 0;
2033 +               else if (c == '\\' && '0' <= *src && *src <= '9')
2034 +                       no = *src++ - '0';
2035 +               else
2036 +                       no = -1;
2037 +
2038 +               if (no < 0) {   /* Ordinary character. */
2039 +                       if (c == '\\' && (*src == '\\' || *src == '&'))
2040 +                               c = *src++;
2041 +                       *dst++ = c;
2042 +               } else if (prog->startp[no] != NULL && prog->endp[no] != NULL) {
2043 +                       len = prog->endp[no] - prog->startp[no];
2044 +                       (void) strncpy(dst, prog->startp[no], len);
2045 +                       dst += len;
2046 +                       if (len != 0 && *(dst-1) == '\0') {     /* strncpy hit NUL. */
2047 +                               regerror("damaged match string");
2048 +                               return;
2049 +                       }
2050 +               }
2051 +       }
2052 +       *dst++ = '\0';
2053 +}