target/linux/generic-2.6/patches-2.6.22/100-netfilter_layer7_2.8.patch

   1 diff -urN linux-2.6.21.1.old/include/linux/netfilter_ipv4/ipt_layer7.h linux-2.6.21.1.dev/include/linux/netfilter_ipv4/ipt_layer7.h
   2 --- linux-2.6.21.1.old/include/linux/netfilter_ipv4/ipt_layer7.h        1970-01-01 01:00:00.000000000 +0100
   3 +++ linux-2.6.21.1.dev/include/linux/netfilter_ipv4/ipt_layer7.h        2007-05-26 20:13:52.648130120 +0200
   4 @@ -0,0 +1,26 @@
   5 +/*
   6 +  By Matthew Strait <quadong@users.sf.net>, Dec 2003.
   7 +  http://l7-filter.sf.net
   8 +
   9 +  This program is free software; you can redistribute it and/or
  10 +  modify it under the terms of the GNU General Public License
  11 +  as published by the Free Software Foundation; either version
  12 +  2 of the License, or (at your option) any later version.
  13 +  http://www.gnu.org/licenses/gpl.txt
  14 +*/
  15 +
  16 +#ifndef _IPT_LAYER7_H
  17 +#define _IPT_LAYER7_H
  18 +
  19 +#define MAX_PATTERN_LEN 8192
  20 +#define MAX_PROTOCOL_LEN 256
  21 +
  22 +typedef char *(*proc_ipt_search) (char *, char, char *);
  23 +
  24 +struct ipt_layer7_info {
  25 +    char protocol[MAX_PROTOCOL_LEN];
  26 +    char invert:1;
  27 +    char pattern[MAX_PATTERN_LEN];
  28 +};
  29 +
  30 +#endif /* _IPT_LAYER7_H */
  31 diff -urN linux-2.6.21.1.old/net/netfilter/nf_conntrack_core.c linux-2.6.21.1.dev/net/netfilter/nf_conntrack_core.c
  32 --- linux-2.6.21.1.old/net/netfilter/nf_conntrack_core.c        2007-04-27 23:49:26.000000000 +0200
  33 +++ linux-2.6.21.1.dev/net/netfilter/nf_conntrack_core.c        2007-05-26 20:13:52.649129968 +0200
  34 @@ -330,6 +330,13 @@
  35          * too. */
  36         nf_ct_remove_expectations(ct);
  37
  38 +       #if defined(CONFIG_IP_NF_MATCH_LAYER7) || defined(CONFIG_IP_NF_MATCH_LAYER7_MODULE)
  39 +       if(ct->layer7.app_proto)
  40 +               kfree(ct->layer7.app_proto);
  41 +       if(ct->layer7.app_data)
  42 +               kfree(ct->layer7.app_data);
  43 +       #endif
  44 +
  45         /* We overload first tuple to link into unconfirmed list. */
  46         if (!nf_ct_is_confirmed(ct)) {
  47                 BUG_ON(list_empty(&ct->tuplehash[IP_CT_DIR_ORIGINAL].list));
  48 diff -urN linux-2.6.21.1.old/net/netfilter/nf_conntrack_standalone.c linux-2.6.21.1.dev/net/netfilter/nf_conntrack_standalone.c
  49 --- linux-2.6.21.1.old/net/netfilter/nf_conntrack_standalone.c  2007-04-27 23:49:26.000000000 +0200
  50 +++ linux-2.6.21.1.dev/net/netfilter/nf_conntrack_standalone.c  2007-05-26 20:13:52.649129968 +0200
  51 @@ -184,6 +184,12 @@
  52                 return -ENOSPC;
  53  #endif
  54
  55 +#if defined(CONFIG_IP_NF_MATCH_LAYER7) || defined(CONFIG_IP_NF_MATCH_LAYER7_MODULE)
  56 +       if(conntrack->layer7.app_proto)
  57 +               if (seq_printf(s, "l7proto=%s ",conntrack->layer7.app_proto))
  58 +                       return 1;
  59 +#endif
  60 +
  61         if (seq_printf(s, "use=%u\n", atomic_read(&conntrack->ct_general.use)))
  62                 return -ENOSPC;
  63
  64 diff -urN linux-2.6.21.1.old/net/ipv4/netfilter/ipt_layer7.c linux-2.6.21.1.dev/net/ipv4/netfilter/ipt_layer7.c
  65 --- linux-2.6.21.1.old/net/ipv4/netfilter/ipt_layer7.c  1970-01-01 01:00:00.000000000 +0100
  66 +++ linux-2.6.21.1.dev/net/ipv4/netfilter/ipt_layer7.c  2007-05-26 20:13:52.650129816 +0200
  67 @@ -0,0 +1,573 @@
  68 +/*
  69 +  Kernel module to match application layer (OSI layer 7) data in connections.
  70 +
  71 +  http://l7-filter.sf.net
  72 +
  73 +  By Matthew Strait and Ethan Sommer, 2003-2006.
  74 +
  75 +  This program is free software; you can redistribute it and/or
  76 +  modify it under the terms of the GNU General Public License
  77 +  as published by the Free Software Foundation; either version
  78 +  2 of the License, or (at your option) any later version.
  79 +  http://www.gnu.org/licenses/gpl.txt
  80 +
  81 +  Based on ipt_string.c (C) 2000 Emmanuel Roger <winfield@freegates.be>
  82 +  and cls_layer7.c (C) 2003 Matthew Strait, Ethan Sommer, Justin Levandoski
  83 +*/
  84 +
  85 +#include <linux/module.h>
  86 +#include <linux/skbuff.h>
  87 +#include <linux/netfilter_ipv4/ip_conntrack.h>
  88 +#include <linux/proc_fs.h>
  89 +#include <linux/ctype.h>
  90 +#include <net/ip.h>
  91 +#include <net/tcp.h>
  92 +#include <linux/spinlock.h>
  93 +
  94 +#include "regexp/regexp.c"
  95 +
  96 +#include <linux/netfilter_ipv4/ipt_layer7.h>
  97 +#include <linux/netfilter_ipv4/ip_tables.h>
  98 +
  99 +MODULE_AUTHOR("Matthew Strait <quadong@users.sf.net>, Ethan Sommer <sommere@users.sf.net>");
 100 +MODULE_LICENSE("GPL");
 101 +MODULE_DESCRIPTION("iptables application layer match module");
 102 +MODULE_VERSION("2.0");
 103 +
 104 +static int maxdatalen = 2048; // this is the default
 105 +module_param(maxdatalen, int, 0444);
 106 +MODULE_PARM_DESC(maxdatalen, "maximum bytes of data looked at by l7-filter");
 107 +
 108 +#ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
 109 +       #define DPRINTK(format,args...) printk(format,##args)
 110 +#else
 111 +       #define DPRINTK(format,args...)
 112 +#endif
 113 +
 114 +#define TOTAL_PACKETS master_conntrack->counters[IP_CT_DIR_ORIGINAL].packets + \
 115 +                     master_conntrack->counters[IP_CT_DIR_REPLY].packets
 116 +
 117 +/* Number of packets whose data we look at.
 118 +This can be modified through /proc/net/layer7_numpackets */
 119 +static int num_packets = 10;
 120 +
 121 +static struct pattern_cache {
 122 +       char * regex_string;
 123 +       regexp * pattern;
 124 +       struct pattern_cache * next;
 125 +} * first_pattern_cache = NULL;
 126 +
 127 +/* I'm new to locking.  Here are my assumptions:
 128 +
 129 +- No one will write to /proc/net/layer7_numpackets over and over very fast;
 130 +  if they did, nothing awful would happen.
 131 +
 132 +- This code will never be processing the same packet twice at the same time,
 133 +  because iptables rules are traversed in order.
 134 +
 135 +- It doesn't matter if two packets from different connections are in here at
 136 +  the same time, because they don't share any data.
 137 +
 138 +- It _does_ matter if two packets from the same connection (or one from a
 139 +  master and one from its child) are here at the same time.  In this case,
 140 +  we have to protect the conntracks and the list of compiled patterns.
 141 +*/
 142 +DEFINE_RWLOCK(ct_lock);
 143 +DEFINE_SPINLOCK(list_lock);
 144 +
 145 +#ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
 146 +/* Converts an unfriendly string into a friendly one by
 147 +replacing unprintables with periods and all whitespace with " ". */
 148 +static char * friendly_print(unsigned char * s)
 149 +{
 150 +       char * f = kmalloc(strlen(s) + 1, GFP_ATOMIC);
 151 +       int i;
 152 +
 153 +       if(!f) {
 154 +               if (net_ratelimit())
 155 +                       printk(KERN_ERR "layer7: out of memory in friendly_print, bailing.\n");
 156 +               return NULL;
 157 +       }
 158 +
 159 +       for(i = 0; i < strlen(s); i++){
 160 +               if(isprint(s[i]) && s[i] < 128) f[i] = s[i];
 161 +               else if(isspace(s[i]))          f[i] = ' ';
 162 +               else                            f[i] = '.';
 163 +       }
 164 +       f[i] = '\0';
 165 +       return f;
 166 +}
 167 +
 168 +static char dec2hex(int i)
 169 +{
 170 +       switch (i) {
 171 +               case 0 ... 9:
 172 +                       return (char)(i + '0');
 173 +                       break;
 174 +               case 10 ... 15:
 175 +                       return (char)(i - 10 + 'a');
 176 +                       break;
 177 +               default:
 178 +                       if (net_ratelimit())
 179 +                               printk("Problem in dec2hex\n");
 180 +                       return '\0';
 181 +       }
 182 +}
 183 +
 184 +static char * hex_print(unsigned char * s)
 185 +{
 186 +       char * g = kmalloc(strlen(s)*3 + 1, GFP_ATOMIC);
 187 +       int i;
 188 +
 189 +       if(!g) {
 190 +              if (net_ratelimit())
 191 +                       printk(KERN_ERR "layer7: out of memory in hex_print, bailing.\n");
 192 +              return NULL;
 193 +       }
 194 +
 195 +       for(i = 0; i < strlen(s); i++) {
 196 +               g[i*3    ] = dec2hex(s[i]/16);
 197 +               g[i*3 + 1] = dec2hex(s[i]%16);
 198 +               g[i*3 + 2] = ' ';
 199 +       }
 200 +       g[i*3] = '\0';
 201 +
 202 +       return g;
 203 +}
 204 +#endif // DEBUG
 205 +
 206 +/* Use instead of regcomp.  As we expect to be seeing the same regexps over and
 207 +over again, it make sense to cache the results. */
 208 +static regexp * compile_and_cache(char * regex_string, char * protocol)
 209 +{
 210 +       struct pattern_cache * node               = first_pattern_cache;
 211 +       struct pattern_cache * last_pattern_cache = first_pattern_cache;
 212 +       struct pattern_cache * tmp;
 213 +       unsigned int len;
 214 +
 215 +       while (node != NULL) {
 216 +               if (!strcmp(node->regex_string, regex_string))
 217 +               return node->pattern;
 218 +
 219 +               last_pattern_cache = node;/* points at the last non-NULL node */
 220 +               node = node->next;
 221 +       }
 222 +
 223 +       /* If we reach the end of the list, then we have not yet cached
 224 +          the pattern for this regex. Let's do that now.
 225 +          Be paranoid about running out of memory to avoid list corruption. */
 226 +       tmp = kmalloc(sizeof(struct pattern_cache), GFP_ATOMIC);
 227 +
 228 +       if(!tmp) {
 229 +               if (net_ratelimit())
 230 +                       printk(KERN_ERR "layer7: out of memory in compile_and_cache, bailing.\n");
 231 +               return NULL;
 232 +       }
 233 +
 234 +       tmp->regex_string  = kmalloc(strlen(regex_string) + 1, GFP_ATOMIC);
 235 +       tmp->pattern       = kmalloc(sizeof(struct regexp),    GFP_ATOMIC);
 236 +       tmp->next = NULL;
 237 +
 238 +       if(!tmp->regex_string || !tmp->pattern) {
 239 +               if (net_ratelimit())
 240 +                       printk(KERN_ERR "layer7: out of memory in compile_and_cache, bailing.\n");
 241 +               kfree(tmp->regex_string);
 242 +               kfree(tmp->pattern);
 243 +               kfree(tmp);
 244 +               return NULL;
 245 +       }
 246 +
 247 +       /* Ok.  The new node is all ready now. */
 248 +       node = tmp;
 249 +
 250 +       if(first_pattern_cache == NULL) /* list is empty */
 251 +               first_pattern_cache = node; /* make node the beginning */
 252 +       else
 253 +               last_pattern_cache->next = node; /* attach node to the end */
 254 +
 255 +       /* copy the string and compile the regex */
 256 +       len = strlen(regex_string);
 257 +       DPRINTK("About to compile this: \"%s\"\n", regex_string);
 258 +       node->pattern = regcomp(regex_string, &len);
 259 +       if ( !node->pattern ) {
 260 +               if (net_ratelimit())
 261 +                       printk(KERN_ERR "layer7: Error compiling regexp \"%s\" (%s)\n", regex_string, protocol);
 262 +               /* pattern is now cached as NULL, so we won't try again. */
 263 +       }
 264 +
 265 +       strcpy(node->regex_string, regex_string);
 266 +       return node->pattern;
 267 +}
 268 +
 269 +static int can_handle(const struct sk_buff *skb)
 270 +{
 271 +       if(!ip_hdr(skb)) /* not IP */
 272 +               return 0;
 273 +       if(ip_hdr(skb)->protocol != IPPROTO_TCP &&
 274 +          ip_hdr(skb)->protocol != IPPROTO_UDP &&
 275 +          ip_hdr(skb)->protocol != IPPROTO_ICMP)
 276 +               return 0;
 277 +       return 1;
 278 +}
 279 +
 280 +/* Returns offset the into the skb->data that the application data starts */
 281 +static int app_data_offset(const struct sk_buff *skb)
 282 +{
 283 +       /* In case we are ported somewhere (ebtables?) where ip_hdr(skb)
 284 +       isn't set, this can be gotten from 4*(skb->data[0] & 0x0f) as well. */
 285 +       int ip_hl = ip_hdrlen(skb);
 286 +
 287 +       if( ip_hdr(skb)->protocol == IPPROTO_TCP ) {
 288 +               /* 12 == offset into TCP header for the header length field.
 289 +               Can't get this with skb->h.th->doff because the tcphdr
 290 +               struct doesn't get set when routing (this is confirmed to be
 291 +               true in Netfilter as well as QoS.) */
 292 +               int tcp_hl = 4*(skb->data[ip_hl + 12] >> 4);
 293 +
 294 +               return ip_hl + tcp_hl;
 295 +       } else if( ip_hdr(skb)->protocol == IPPROTO_UDP  ) {
 296 +               return ip_hl + 8; /* UDP header is always 8 bytes */
 297 +       } else if( ip_hdr(skb)->protocol == IPPROTO_ICMP ) {
 298 +               return ip_hl + 8; /* ICMP header is 8 bytes */
 299 +       } else {
 300 +               if (net_ratelimit())
 301 +                       printk(KERN_ERR "layer7: tried to handle unknown protocol!\n");
 302 +               return ip_hl + 8; /* something reasonable */
 303 +       }
 304 +}
 305 +
 306 +/* handles whether there's a match when we aren't appending data anymore */
 307 +static int match_no_append(struct ip_conntrack * conntrack, struct ip_conntrack * master_conntrack,
 308 +                       enum ip_conntrack_info ctinfo, enum ip_conntrack_info master_ctinfo,
 309 +                       struct ipt_layer7_info * info)
 310 +{
 311 +       /* If we're in here, throw the app data away */
 312 +       write_lock(&ct_lock);
 313 +       if(master_conntrack->layer7.app_data != NULL) {
 314 +
 315 +       #ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
 316 +               if(!master_conntrack->layer7.app_proto) {
 317 +                       char * f = friendly_print(master_conntrack->layer7.app_data);
 318 +                       char * g = hex_print(master_conntrack->layer7.app_data);
 319 +                       DPRINTK("\nl7-filter gave up after %d bytes (%d packets):\n%s\n",
 320 +                               strlen(f), TOTAL_PACKETS, f);
 321 +                       kfree(f);
 322 +                       DPRINTK("In hex: %s\n", g);
 323 +                       kfree(g);
 324 +               }
 325 +       #endif
 326 +
 327 +               kfree(master_conntrack->layer7.app_data);
 328 +               master_conntrack->layer7.app_data = NULL; /* don't free again */
 329 +       }
 330 +       write_unlock(&ct_lock);
 331 +
 332 +       if(master_conntrack->layer7.app_proto){
 333 +               /* Here child connections set their .app_proto (for /proc/net/ip_conntrack) */
 334 +               write_lock(&ct_lock);
 335 +               if(!conntrack->layer7.app_proto) {
 336 +                       conntrack->layer7.app_proto = kmalloc(strlen(master_conntrack->layer7.app_proto)+1, GFP_ATOMIC);
 337 +                       if(!conntrack->layer7.app_proto){
 338 +                               if (net_ratelimit())
 339 +                                       printk(KERN_ERR "layer7: out of memory in match_no_append, bailing.\n");
 340 +                               write_unlock(&ct_lock);
 341 +                               return 1;
 342 +                       }
 343 +                       strcpy(conntrack->layer7.app_proto, master_conntrack->layer7.app_proto);
 344 +               }
 345 +               write_unlock(&ct_lock);
 346 +
 347 +               return (!strcmp(master_conntrack->layer7.app_proto, info->protocol));
 348 +       }
 349 +       else {
 350 +               /* If not classified, set to "unknown" to distinguish from
 351 +               connections that are still being tested. */
 352 +               write_lock(&ct_lock);
 353 +               master_conntrack->layer7.app_proto = kmalloc(strlen("unknown")+1, GFP_ATOMIC);
 354 +               if(!master_conntrack->layer7.app_proto){
 355 +                       if (net_ratelimit())
 356 +                               printk(KERN_ERR "layer7: out of memory in match_no_append, bailing.\n");
 357 +                       write_unlock(&ct_lock);
 358 +                       return 1;
 359 +               }
 360 +               strcpy(master_conntrack->layer7.app_proto, "unknown");
 361 +               write_unlock(&ct_lock);
 362 +               return 0;
 363 +       }
 364 +}
 365 +
 366 +/* add the new app data to the conntrack.  Return number of bytes added. */
 367 +static int add_data(struct ip_conntrack * master_conntrack,
 368 +                       char * app_data, int appdatalen)
 369 +{
 370 +       int length = 0, i;
 371 +       int oldlength = master_conntrack->layer7.app_data_len;
 372 +
 373 +       // This is a fix for a race condition by Deti Fliegl. However, I'm not
 374 +       // clear on whether the race condition exists or whether this really
 375 +       // fixes it.  I might just be being dense... Anyway, if it's not really
 376 +       // a fix, all it does is waste a very small amount of time.
 377 +       if(!master_conntrack->layer7.app_data) return 0;
 378 +
 379 +       /* Strip nulls. Make everything lower case (our regex lib doesn't
 380 +       do case insensitivity).  Add it to the end of the current data. */
 381 +       for(i = 0; i < maxdatalen-oldlength-1 &&
 382 +                  i < appdatalen; i++) {
 383 +               if(app_data[i] != '\0') {
 384 +                       master_conntrack->layer7.app_data[length+oldlength] =
 385 +                               /* the kernel version of tolower mungs 'upper ascii' */
 386 +                               isascii(app_data[i])? tolower(app_data[i]) : app_data[i];
 387 +                       length++;
 388 +               }
 389 +       }
 390 +
 391 +       master_conntrack->layer7.app_data[length+oldlength] = '\0';
 392 +       master_conntrack->layer7.app_data_len = length + oldlength;
 393 +
 394 +       return length;
 395 +}
 396 +
 397 +/* Returns true on match and false otherwise.  */
 398 +static int match(const struct sk_buff *skbin,
 399 +       const struct net_device *in, const struct net_device *out,
 400 +       const struct xt_match *match, const void *matchinfo,
 401 +       int offset, unsigned int protoff, int *hotdrop)
 402 +{
 403 +       /* sidestep const without getting a compiler warning... */
 404 +       struct sk_buff * skb = (struct sk_buff *)skbin;
 405 +
 406 +       struct ipt_layer7_info * info = (struct ipt_layer7_info *)matchinfo;
 407 +       enum ip_conntrack_info master_ctinfo, ctinfo;
 408 +       struct ip_conntrack *master_conntrack, *conntrack;
 409 +       unsigned char * app_data;
 410 +       unsigned int pattern_result, appdatalen;
 411 +       regexp * comppattern;
 412 +
 413 +       if(!can_handle(skb)){
 414 +               DPRINTK("layer7: This is some protocol I can't handle.\n");
 415 +               return info->invert;
 416 +       }
 417 +
 418 +       /* Treat parent & all its children together as one connection, except
 419 +       for the purpose of setting conntrack->layer7.app_proto in the actual
 420 +       connection. This makes /proc/net/ip_conntrack more satisfying. */
 421 +       if(!(conntrack = ip_conntrack_get((struct sk_buff *)skb, &ctinfo)) ||
 422 +          !(master_conntrack = ip_conntrack_get((struct sk_buff *)skb, &master_ctinfo))) {
 423 +               //DPRINTK("layer7: packet is not from a known connection, giving up.\n");
 424 +               return info->invert;
 425 +       }
 426 +
 427 +       /* Try to get a master conntrack (and its master etc) for FTP, etc. */
 428 +       while (master_ct(master_conntrack) != NULL)
 429 +               master_conntrack = master_ct(master_conntrack);
 430 +
 431 +       /* if we've classified it or seen too many packets */
 432 +       if(TOTAL_PACKETS > num_packets ||
 433 +          master_conntrack->layer7.app_proto) {
 434 +
 435 +               pattern_result = match_no_append(conntrack, master_conntrack, ctinfo, master_ctinfo, info);
 436 +
 437 +               /* skb->cb[0] == seen. Avoid doing things twice if there are two l7
 438 +               rules. I'm not sure that using cb for this purpose is correct, although
 439 +               it says "put your private variables there". But it doesn't look like it
 440 +               is being used for anything else in the skbs that make it here. How can
 441 +               I write to cb without making the compiler angry? */
 442 +               skb->cb[0] = 1; /* marking it seen here is probably irrelevant, but consistant */
 443 +
 444 +               return (pattern_result ^ info->invert);
 445 +       }
 446 +
 447 +       if(skb_is_nonlinear(skb)){
 448 +               if(skb_linearize(skb) != 0){
 449 +                       if (net_ratelimit())
 450 +                               printk(KERN_ERR "layer7: failed to linearize packet, bailing.\n");
 451 +                       return info->invert;
 452 +               }
 453 +       }
 454 +
 455 +       /* now that the skb is linearized, it's safe to set these. */
 456 +       app_data = skb->data + app_data_offset(skb);
 457 +       appdatalen = skb->tail - app_data;
 458 +
 459 +       spin_lock_bh(&list_lock);
 460 +       /* the return value gets checked later, when we're ready to use it */
 461 +       comppattern = compile_and_cache(info->pattern, info->protocol);
 462 +       spin_unlock_bh(&list_lock);
 463 +
 464 +       /* On the first packet of a connection, allocate space for app data */
 465 +       write_lock(&ct_lock);
 466 +       if(TOTAL_PACKETS == 1 && !skb->cb[0] && !master_conntrack->layer7.app_data) {
 467 +               master_conntrack->layer7.app_data = kmalloc(maxdatalen, GFP_ATOMIC);
 468 +               if(!master_conntrack->layer7.app_data){
 469 +                       if (net_ratelimit())
 470 +                               printk(KERN_ERR "layer7: out of memory in match, bailing.\n");
 471 +                       write_unlock(&ct_lock);
 472 +                       return info->invert;
 473 +               }
 474 +
 475 +               master_conntrack->layer7.app_data[0] = '\0';
 476 +       }
 477 +       write_unlock(&ct_lock);
 478 +
 479 +       /* Can be here, but unallocated, if numpackets is increased near
 480 +       the beginning of a connection */
 481 +       if(master_conntrack->layer7.app_data == NULL)
 482 +               return (info->invert); /* unmatched */
 483 +
 484 +       if(!skb->cb[0]){
 485 +               int newbytes;
 486 +               write_lock(&ct_lock);
 487 +               newbytes = add_data(master_conntrack, app_data, appdatalen);
 488 +               write_unlock(&ct_lock);
 489 +
 490 +               if(newbytes == 0) { /* didn't add any data */
 491 +                       skb->cb[0] = 1;
 492 +                       /* Didn't match before, not going to match now */
 493 +                       return info->invert;
 494 +               }
 495 +       }
 496 +
 497 +       /* If looking for "unknown", then never match.  "Unknown" means that
 498 +       we've given up; we're still trying with these packets. */
 499 +       read_lock(&ct_lock);
 500 +       if(!strcmp(info->protocol, "unknown")) {
 501 +               pattern_result = 0;
 502 +       /* If the regexp failed to compile, don't bother running it */
 503 +       } else if(comppattern && regexec(comppattern, master_conntrack->layer7.app_data)) {
 504 +               DPRINTK("layer7: matched %s\n", info->protocol);
 505 +               pattern_result = 1;
 506 +       } else pattern_result = 0;
 507 +       read_unlock(&ct_lock);
 508 +
 509 +       if(pattern_result) {
 510 +               write_lock(&ct_lock);
 511 +               master_conntrack->layer7.app_proto = kmalloc(strlen(info->protocol)+1, GFP_ATOMIC);
 512 +               if(!master_conntrack->layer7.app_proto){
 513 +                       if (net_ratelimit())
 514 +                               printk(KERN_ERR "layer7: out of memory in match, bailing.\n");
 515 +                       write_unlock(&ct_lock);
 516 +                       return (pattern_result ^ info->invert);
 517 +               }
 518 +               strcpy(master_conntrack->layer7.app_proto, info->protocol);
 519 +               write_unlock(&ct_lock);
 520 +       }
 521 +
 522 +       /* mark the packet seen */
 523 +       skb->cb[0] = 1;
 524 +
 525 +       return (pattern_result ^ info->invert);
 526 +}
 527 +
 528 +static struct ipt_match layer7_match = {
 529 +       .name = "layer7",
 530 +       .match = &match,
 531 +       .matchsize  = sizeof(struct ipt_layer7_info),
 532 +       .me = THIS_MODULE
 533 +};
 534 +
 535 +/* taken from drivers/video/modedb.c */
 536 +static int my_atoi(const char *s)
 537 +{
 538 +       int val = 0;
 539 +
 540 +       for (;; s++) {
 541 +               switch (*s) {
 542 +                       case '0'...'9':
 543 +                       val = 10*val+(*s-'0');
 544 +                       break;
 545 +               default:
 546 +                       return val;
 547 +               }
 548 +       }
 549 +}
 550 +
 551 +/* write out num_packets to userland. */
 552 +static int layer7_read_proc(char* page, char ** start, off_t off, int count,
 553 +                    int* eof, void * data)
 554 +{
 555 +       if(num_packets > 99 && net_ratelimit())
 556 +               printk(KERN_ERR "layer7: NOT REACHED. num_packets too big\n");
 557 +
 558 +       page[0] = num_packets/10 + '0';
 559 +       page[1] = num_packets%10 + '0';
 560 +       page[2] = '\n';
 561 +       page[3] = '\0';
 562 +
 563 +       *eof=1;
 564 +
 565 +       return 3;
 566 +}
 567 +
 568 +/* Read in num_packets from userland */
 569 +static int layer7_write_proc(struct file* file, const char* buffer,
 570 +                     unsigned long count, void *data)
 571 +{
 572 +       char * foo = kmalloc(count, GFP_ATOMIC);
 573 +
 574 +       if(!foo){
 575 +               if (net_ratelimit())
 576 +                       printk(KERN_ERR "layer7: out of memory, bailing. num_packets unchanged.\n");
 577 +               return count;
 578 +       }
 579 +
 580 +       if(copy_from_user(foo, buffer, count)) {
 581 +               return -EFAULT;
 582 +       }
 583 +
 584 +
 585 +       num_packets = my_atoi(foo);
 586 +       kfree (foo);
 587 +
 588 +       /* This has an arbitrary limit to make the math easier. I'm lazy.
 589 +       But anyway, 99 is a LOT! If you want more, you're doing it wrong! */
 590 +       if(num_packets > 99) {
 591 +               printk(KERN_WARNING "layer7: num_packets can't be > 99.\n");
 592 +               num_packets = 99;
 593 +       } else if(num_packets < 1) {
 594 +               printk(KERN_WARNING "layer7: num_packets can't be < 1.\n");
 595 +               num_packets = 1;
 596 +       }
 597 +
 598 +       return count;
 599 +}
 600 +
 601 +/* register the proc file */
 602 +static void layer7_init_proc(void)
 603 +{
 604 +       struct proc_dir_entry* entry;
 605 +       entry = create_proc_entry("layer7_numpackets", 0644, proc_net);
 606 +       entry->read_proc = layer7_read_proc;
 607 +       entry->write_proc = layer7_write_proc;
 608 +}
 609 +
 610 +static void layer7_cleanup_proc(void)
 611 +{
 612 +       remove_proc_entry("layer7_numpackets", proc_net);
 613 +}
 614 +
 615 +static int __init ipt_layer7_init(void)
 616 +{
 617 +       need_conntrack();
 618 +
 619 +       layer7_init_proc();
 620 +       if(maxdatalen < 1) {
 621 +               printk(KERN_WARNING "layer7: maxdatalen can't be < 1, using 1\n");
 622 +               maxdatalen = 1;
 623 +       }
 624 +       /* This is not a hard limit.  It's just here to prevent people from
 625 +       bringing their slow machines to a grinding halt. */
 626 +       else if(maxdatalen > 65536) {
 627 +               printk(KERN_WARNING "layer7: maxdatalen can't be > 65536, using 65536\n");
 628 +               maxdatalen = 65536;
 629 +       }
 630 +       return ipt_register_match(&layer7_match);
 631 +}
 632 +
 633 +static void __exit ipt_layer7_fini(void)
 634 +{
 635 +       layer7_cleanup_proc();
 636 +       ipt_unregister_match(&layer7_match);
 637 +}
 638 +
 639 +module_init(ipt_layer7_init);
 640 +module_exit(ipt_layer7_fini);
 641 diff -urN linux-2.6.21.1.old/net/ipv4/netfilter/Kconfig linux-2.6.21.1.dev/net/ipv4/netfilter/Kconfig
 642 --- linux-2.6.21.1.old/net/ipv4/netfilter/Kconfig       2007-04-27 23:49:26.000000000 +0200
 643 +++ linux-2.6.21.1.dev/net/ipv4/netfilter/Kconfig       2007-05-26 20:13:52.650129816 +0200
 644 @@ -63,6 +63,24 @@
 645
 646           To compile it as a module, choose M here.  If unsure, say N.
 647
 648 +config IP_NF_MATCH_LAYER7
 649 +       tristate "Layer 7 match support (EXPERIMENTAL)"
 650 +       depends on IP_NF_IPTABLES && IP_NF_CT_ACCT && IP_NF_CONNTRACK && EXPERIMENTAL
 651 +       help
 652 +         Say Y if you want to be able to classify connections (and their
 653 +         packets) based on regular expression matching of their application
 654 +         layer data.   This is one way to classify applications such as
 655 +         peer-to-peer filesharing systems that do not always use the same
 656 +         port.
 657 +
 658 +         To compile it as a module, choose M here.  If unsure, say N.
 659 +
 660 +config IP_NF_MATCH_LAYER7_DEBUG
 661 +       bool "Layer 7 debugging output"
 662 +       depends on IP_NF_MATCH_LAYER7
 663 +       help
 664 +         Say Y to get lots of debugging output.
 665 +
 666  config IP_NF_MATCH_TOS
 667         tristate "TOS match support"
 668         depends on IP_NF_IPTABLES
 669 diff -urN linux-2.6.21.1.old/net/ipv4/netfilter/Makefile linux-2.6.21.1.dev/net/ipv4/netfilter/Makefile
 670 --- linux-2.6.21.1.old/net/ipv4/netfilter/Makefile      2007-04-27 23:49:26.000000000 +0200
 671 +++ linux-2.6.21.1.dev/net/ipv4/netfilter/Makefile      2007-05-26 20:13:52.651129664 +0200
 672 @@ -50,6 +50,8 @@
 673  obj-$(CONFIG_IP_NF_MATCH_TTL) += ipt_ttl.o
 674  obj-$(CONFIG_IP_NF_MATCH_ADDRTYPE) += ipt_addrtype.o
 675
 676 +obj-$(CONFIG_IP_NF_MATCH_LAYER7) += ipt_layer7.o
 677 +
 678  # targets
 679  obj-$(CONFIG_IP_NF_TARGET_REJECT) += ipt_REJECT.o
 680  obj-$(CONFIG_IP_NF_TARGET_TOS) += ipt_TOS.o
 681 diff -urN linux-2.6.21.1.old/net/ipv4/netfilter/regexp/regexp.c linux-2.6.21.1.dev/net/ipv4/netfilter/regexp/regexp.c
 682 --- linux-2.6.21.1.old/net/ipv4/netfilter/regexp/regexp.c       1970-01-01 01:00:00.000000000 +0100
 683 +++ linux-2.6.21.1.dev/net/ipv4/netfilter/regexp/regexp.c       2007-05-26 20:13:52.652129512 +0200
 684 @@ -0,0 +1,1197 @@
 685 +/*
 686 + * regcomp and regexec -- regsub and regerror are elsewhere
 687 + * @(#)regexp.c        1.3 of 18 April 87
 688 + *
 689 + *     Copyright (c) 1986 by University of Toronto.
 690 + *     Written by Henry Spencer.  Not derived from licensed software.
 691 + *
 692 + *     Permission is granted to anyone to use this software for any
 693 + *     purpose on any computer system, and to redistribute it freely,
 694 + *     subject to the following restrictions:
 695 + *
 696 + *     1. The author is not responsible for the consequences of use of
 697 + *             this software, no matter how awful, even if they arise
 698 + *             from defects in it.
 699 + *
 700 + *     2. The origin of this software must not be misrepresented, either
 701 + *             by explicit claim or by omission.
 702 + *
 703 + *     3. Altered versions must be plainly marked as such, and must not
 704 + *             be misrepresented as being the original software.
 705 + *
 706 + * Beware that some of this code is subtly aware of the way operator
 707 + * precedence is structured in regular expressions.  Serious changes in
 708 + * regular-expression syntax might require a total rethink.
 709 + *
 710 + * This code was modified by Ethan Sommer to work within the kernel
 711 + * (it now uses kmalloc etc..)
 712 + *
 713 + * Modified slightly by Matthew Strait to use more modern C.
 714 + */
 715 +
 716 +#include "regexp.h"
 717 +#include "regmagic.h"
 718 +
 719 +/* added by ethan and matt.  Lets it work in both kernel and user space.
 720 +(So iptables can use it, for instance.)  Yea, it goes both ways... */
 721 +#if __KERNEL__
 722 +  #define malloc(foo) kmalloc(foo,GFP_ATOMIC)
 723 +#else
 724 +  #define printk(format,args...) printf(format,##args)
 725 +#endif
 726 +
 727 +void regerror(char * s)
 728 +{
 729 +        printk("<3>Regexp: %s\n", s);
 730 +        /* NOTREACHED */
 731 +}
 732 +
 733 +/*
 734 + * The "internal use only" fields in regexp.h are present to pass info from
 735 + * compile to execute that permits the execute phase to run lots faster on
 736 + * simple cases.  They are:
 737 + *
 738 + * regstart    char that must begin a match; '\0' if none obvious
 739 + * reganch     is the match anchored (at beginning-of-line only)?
 740 + * regmust     string (pointer into program) that match must include, or NULL
 741 + * regmlen     length of regmust string
 742 + *
 743 + * Regstart and reganch permit very fast decisions on suitable starting points
 744 + * for a match, cutting down the work a lot.  Regmust permits fast rejection
 745 + * of lines that cannot possibly match.  The regmust tests are costly enough
 746 + * that regcomp() supplies a regmust only if the r.e. contains something
 747 + * potentially expensive (at present, the only such thing detected is * or +
 748 + * at the start of the r.e., which can involve a lot of backup).  Regmlen is
 749 + * supplied because the test in regexec() needs it and regcomp() is computing
 750 + * it anyway.
 751 + */
 752 +
 753 +/*
 754 + * Structure for regexp "program".  This is essentially a linear encoding
 755 + * of a nondeterministic finite-state machine (aka syntax charts or
 756 + * "railroad normal form" in parsing technology).  Each node is an opcode
 757 + * plus a "next" pointer, possibly plus an operand.  "Next" pointers of
 758 + * all nodes except BRANCH implement concatenation; a "next" pointer with
 759 + * a BRANCH on both ends of it is connecting two alternatives.  (Here we
 760 + * have one of the subtle syntax dependencies:  an individual BRANCH (as
 761 + * opposed to a collection of them) is never concatenated with anything
 762 + * because of operator precedence.)  The operand of some types of node is
 763 + * a literal string; for others, it is a node leading into a sub-FSM.  In
 764 + * particular, the operand of a BRANCH node is the first node of the branch.
 765 + * (NB this is *not* a tree structure:  the tail of the branch connects
 766 + * to the thing following the set of BRANCHes.)  The opcodes are:
 767 + */
 768 +
 769 +/* definition  number  opnd?   meaning */
 770 +#define        END     0       /* no   End of program. */
 771 +#define        BOL     1       /* no   Match "" at beginning of line. */
 772 +#define        EOL     2       /* no   Match "" at end of line. */
 773 +#define        ANY     3       /* no   Match any one character. */
 774 +#define        ANYOF   4       /* str  Match any character in this string. */
 775 +#define        ANYBUT  5       /* str  Match any character not in this string. */
 776 +#define        BRANCH  6       /* node Match this alternative, or the next... */
 777 +#define        BACK    7       /* no   Match "", "next" ptr points backward. */
 778 +#define        EXACTLY 8       /* str  Match this string. */
 779 +#define        NOTHING 9       /* no   Match empty string. */
 780 +#define        STAR    10      /* node Match this (simple) thing 0 or more times. */
 781 +#define        PLUS    11      /* node Match this (simple) thing 1 or more times. */
 782 +#define        OPEN    20      /* no   Mark this point in input as start of #n. */
 783 +                       /*      OPEN+1 is number 1, etc. */
 784 +#define        CLOSE   30      /* no   Analogous to OPEN. */
 785 +
 786 +/*
 787 + * Opcode notes:
 788 + *
 789 + * BRANCH      The set of branches constituting a single choice are hooked
 790 + *             together with their "next" pointers, since precedence prevents
 791 + *             anything being concatenated to any individual branch.  The
 792 + *             "next" pointer of the last BRANCH in a choice points to the
 793 + *             thing following the whole choice.  This is also where the
 794 + *             final "next" pointer of each individual branch points; each
 795 + *             branch starts with the operand node of a BRANCH node.
 796 + *
 797 + * BACK                Normal "next" pointers all implicitly point forward; BACK
 798 + *             exists to make loop structures possible.
 799 + *
 800 + * STAR,PLUS   '?', and complex '*' and '+', are implemented as circular
 801 + *             BRANCH structures using BACK.  Simple cases (one character
 802 + *             per match) are implemented with STAR and PLUS for speed
 803 + *             and to minimize recursive plunges.
 804 + *
 805 + * OPEN,CLOSE  ...are numbered at compile time.
 806 + */
 807 +
 808 +/*
 809 + * A node is one char of opcode followed by two chars of "next" pointer.
 810 + * "Next" pointers are stored as two 8-bit pieces, high order first.  The
 811 + * value is a positive offset from the opcode of the node containing it.
 812 + * An operand, if any, simply follows the node.  (Note that much of the
 813 + * code generation knows about this implicit relationship.)
 814 + *
 815 + * Using two bytes for the "next" pointer is vast overkill for most things,
 816 + * but allows patterns to get big without disasters.
 817 + */
 818 +#define        OP(p)   (*(p))
 819 +#define        NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377))
 820 +#define        OPERAND(p)      ((p) + 3)
 821 +
 822 +/*
 823 + * See regmagic.h for one further detail of program structure.
 824 + */
 825 +
 826 +
 827 +/*
 828 + * Utility definitions.
 829 + */
 830 +#ifndef CHARBITS
 831 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
 832 +#else
 833 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
 834 +#endif
 835 +
 836 +#define        FAIL(m) { regerror(m); return(NULL); }
 837 +#define        ISMULT(c)       ((c) == '*' || (c) == '+' || (c) == '?')
 838 +#define        META    "^$.[()|?+*\\"
 839 +
 840 +/*
 841 + * Flags to be passed up and down.
 842 + */
 843 +#define        HASWIDTH        01      /* Known never to match null string. */
 844 +#define        SIMPLE          02      /* Simple enough to be STAR/PLUS operand. */
 845 +#define        SPSTART         04      /* Starts with * or +. */
 846 +#define        WORST           0       /* Worst case. */
 847 +
 848 +/*
 849 + * Global work variables for regcomp().
 850 + */
 851 +struct match_globals {
 852 +char *reginput;                /* String-input pointer. */
 853 +char *regbol;          /* Beginning of input, for ^ check. */
 854 +char **regstartp;      /* Pointer to startp array. */
 855 +char **regendp;                /* Ditto for endp. */
 856 +char *regparse;                /* Input-scan pointer. */
 857 +int regnpar;           /* () count. */
 858 +char regdummy;
 859 +char *regcode;         /* Code-emit pointer; &regdummy = don't. */
 860 +long regsize;          /* Code size. */
 861 +};
 862 +
 863 +/*
 864 + * Forward declarations for regcomp()'s friends.
 865 + */
 866 +#ifndef STATIC
 867 +#define        STATIC  static
 868 +#endif
 869 +STATIC char *reg(struct match_globals *g, int paren,int *flagp);
 870 +STATIC char *regbranch(struct match_globals *g, int *flagp);
 871 +STATIC char *regpiece(struct match_globals *g, int *flagp);
 872 +STATIC char *regatom(struct match_globals *g, int *flagp);
 873 +STATIC char *regnode(struct match_globals *g, char op);
 874 +STATIC char *regnext(struct match_globals *g, char *p);
 875 +STATIC void regc(struct match_globals *g, char b);
 876 +STATIC void reginsert(struct match_globals *g, char op, char *opnd);
 877 +STATIC void regtail(struct match_globals *g, char *p, char *val);
 878 +STATIC void regoptail(struct match_globals *g, char *p, char *val);
 879 +
 880 +
 881 +__kernel_size_t my_strcspn(const char *s1,const char *s2)
 882 +{
 883 +        char *scan1;
 884 +        char *scan2;
 885 +        int count;
 886 +
 887 +        count = 0;
 888 +        for (scan1 = (char *)s1; *scan1 != '\0'; scan1++) {
 889 +                for (scan2 = (char *)s2; *scan2 != '\0';)       /* ++ moved down. */
 890 +                        if (*scan1 == *scan2++)
 891 +                                return(count);
 892 +                count++;
 893 +        }
 894 +        return(count);
 895 +}
 896 +
 897 +/*
 898 + - regcomp - compile a regular expression into internal code
 899 + *
 900 + * We can't allocate space until we know how big the compiled form will be,
 901 + * but we can't compile it (and thus know how big it is) until we've got a
 902 + * place to put the code.  So we cheat:  we compile it twice, once with code
 903 + * generation turned off and size counting turned on, and once "for real".
 904 + * This also means that we don't allocate space until we are sure that the
 905 + * thing really will compile successfully, and we never have to move the
 906 + * code and thus invalidate pointers into it.  (Note that it has to be in
 907 + * one piece because free() must be able to free it all.)
 908 + *
 909 + * Beware that the optimization-preparation code in here knows about some
 910 + * of the structure of the compiled regexp.
 911 + */
 912 +regexp *
 913 +regcomp(char *exp,int *patternsize)
 914 +{
 915 +       register regexp *r;
 916 +       register char *scan;
 917 +       register char *longest;
 918 +       register int len;
 919 +       int flags;
 920 +       struct match_globals g;
 921 +
 922 +       /* commented out by ethan
 923 +          extern char *malloc();
 924 +       */
 925 +
 926 +       if (exp == NULL)
 927 +               FAIL("NULL argument");
 928 +
 929 +       /* First pass: determine size, legality. */
 930 +       g.regparse = exp;
 931 +       g.regnpar = 1;
 932 +       g.regsize = 0L;
 933 +       g.regcode = &g.regdummy;
 934 +       regc(&g, MAGIC);
 935 +       if (reg(&g, 0, &flags) == NULL)
 936 +               return(NULL);
 937 +
 938 +       /* Small enough for pointer-storage convention? */
 939 +       if (g.regsize >= 32767L)                /* Probably could be 65535L. */
 940 +               FAIL("regexp too big");
 941 +
 942 +       /* Allocate space. */
 943 +       *patternsize=sizeof(regexp) + (unsigned)g.regsize;
 944 +       r = (regexp *)malloc(sizeof(regexp) + (unsigned)g.regsize);
 945 +       if (r == NULL)
 946 +               FAIL("out of space");
 947 +
 948 +       /* Second pass: emit code. */
 949 +       g.regparse = exp;
 950 +       g.regnpar = 1;
 951 +       g.regcode = r->program;
 952 +       regc(&g, MAGIC);
 953 +       if (reg(&g, 0, &flags) == NULL)
 954 +               return(NULL);
 955 +
 956 +       /* Dig out information for optimizations. */
 957 +       r->regstart = '\0';     /* Worst-case defaults. */
 958 +       r->reganch = 0;
 959 +       r->regmust = NULL;
 960 +       r->regmlen = 0;
 961 +       scan = r->program+1;                    /* First BRANCH. */
 962 +       if (OP(regnext(&g, scan)) == END) {             /* Only one top-level choice. */
 963 +               scan = OPERAND(scan);
 964 +
 965 +               /* Starting-point info. */
 966 +               if (OP(scan) == EXACTLY)
 967 +                       r->regstart = *OPERAND(scan);
 968 +               else if (OP(scan) == BOL)
 969 +                       r->reganch++;
 970 +
 971 +               /*
 972 +                * If there's something expensive in the r.e., find the
 973 +                * longest literal string that must appear and make it the
 974 +                * regmust.  Resolve ties in favor of later strings, since
 975 +                * the regstart check works with the beginning of the r.e.
 976 +                * and avoiding duplication strengthens checking.  Not a
 977 +                * strong reason, but sufficient in the absence of others.
 978 +                */
 979 +               if (flags&SPSTART) {
 980 +                       longest = NULL;
 981 +                       len = 0;
 982 +                       for (; scan != NULL; scan = regnext(&g, scan))
 983 +                               if (OP(scan) == EXACTLY && strlen(OPERAND(scan)) >= len) {
 984 +                                       longest = OPERAND(scan);
 985 +                                       len = strlen(OPERAND(scan));
 986 +                               }
 987 +                       r->regmust = longest;
 988 +                       r->regmlen = len;
 989 +               }
 990 +       }
 991 +
 992 +       return(r);
 993 +}
 994 +
 995 +/*
 996 + - reg - regular expression, i.e. main body or parenthesized thing
 997 + *
 998 + * Caller must absorb opening parenthesis.
 999 + *
1000 + * Combining parenthesis handling with the base level of regular expression
1001 + * is a trifle forced, but the need to tie the tails of the branches to what
1002 + * follows makes it hard to avoid.
1003 + */
1004 +static char *
1005 +reg(struct match_globals *g, int paren, int *flagp /* Parenthesized? */ )
1006 +{
1007 +       register char *ret;
1008 +       register char *br;
1009 +       register char *ender;
1010 +       register int parno = 0; /* 0 makes gcc happy */
1011 +       int flags;
1012 +
1013 +       *flagp = HASWIDTH;      /* Tentatively. */
1014 +
1015 +       /* Make an OPEN node, if parenthesized. */
1016 +       if (paren) {
1017 +               if (g->regnpar >= NSUBEXP)
1018 +                       FAIL("too many ()");
1019 +               parno = g->regnpar;
1020 +               g->regnpar++;
1021 +               ret = regnode(g, OPEN+parno);
1022 +       } else
1023 +               ret = NULL;
1024 +
1025 +       /* Pick up the branches, linking them together. */
1026 +       br = regbranch(g, &flags);
1027 +       if (br == NULL)
1028 +               return(NULL);
1029 +       if (ret != NULL)
1030 +               regtail(g, ret, br);    /* OPEN -> first. */
1031 +       else
1032 +               ret = br;
1033 +       if (!(flags&HASWIDTH))
1034 +               *flagp &= ~HASWIDTH;
1035 +       *flagp |= flags&SPSTART;
1036 +       while (*g->regparse == '|') {
1037 +               g->regparse++;
1038 +               br = regbranch(g, &flags);
1039 +               if (br == NULL)
1040 +                       return(NULL);
1041 +               regtail(g, ret, br);    /* BRANCH -> BRANCH. */
1042 +               if (!(flags&HASWIDTH))
1043 +                       *flagp &= ~HASWIDTH;
1044 +               *flagp |= flags&SPSTART;
1045 +       }
1046 +
1047 +       /* Make a closing node, and hook it on the end. */
1048 +       ender = regnode(g, (paren) ? CLOSE+parno : END);
1049 +       regtail(g, ret, ender);
1050 +
1051 +       /* Hook the tails of the branches to the closing node. */
1052 +       for (br = ret; br != NULL; br = regnext(g, br))
1053 +               regoptail(g, br, ender);
1054 +
1055 +       /* Check for proper termination. */
1056 +       if (paren && *g->regparse++ != ')') {
1057 +               FAIL("unmatched ()");
1058 +       } else if (!paren && *g->regparse != '\0') {
1059 +               if (*g->regparse == ')') {
1060 +                       FAIL("unmatched ()");
1061 +               } else
1062 +                       FAIL("junk on end");    /* "Can't happen". */
1063 +               /* NOTREACHED */
1064 +       }
1065 +
1066 +       return(ret);
1067 +}
1068 +
1069 +/*
1070 + - regbranch - one alternative of an | operator
1071 + *
1072 + * Implements the concatenation operator.
1073 + */
1074 +static char *
1075 +regbranch(struct match_globals *g, int *flagp)
1076 +{
1077 +       register char *ret;
1078 +       register char *chain;
1079 +       register char *latest;
1080 +       int flags;
1081 +
1082 +       *flagp = WORST;         /* Tentatively. */
1083 +
1084 +       ret = regnode(g, BRANCH);
1085 +       chain = NULL;
1086 +       while (*g->regparse != '\0' && *g->regparse != '|' && *g->regparse != ')') {
1087 +               latest = regpiece(g, &flags);
1088 +               if (latest == NULL)
1089 +                       return(NULL);
1090 +               *flagp |= flags&HASWIDTH;
1091 +               if (chain == NULL)      /* First piece. */
1092 +                       *flagp |= flags&SPSTART;
1093 +               else
1094 +                       regtail(g, chain, latest);
1095 +               chain = latest;
1096 +       }
1097 +       if (chain == NULL)      /* Loop ran zero times. */
1098 +               (void) regnode(g, NOTHING);
1099 +
1100 +       return(ret);
1101 +}
1102 +
1103 +/*
1104 + - regpiece - something followed by possible [*+?]
1105 + *
1106 + * Note that the branching code sequences used for ? and the general cases
1107 + * of * and + are somewhat optimized:  they use the same NOTHING node as
1108 + * both the endmarker for their branch list and the body of the last branch.
1109 + * It might seem that this node could be dispensed with entirely, but the
1110 + * endmarker role is not redundant.
1111 + */
1112 +static char *
1113 +regpiece(struct match_globals *g, int *flagp)
1114 +{
1115 +       register char *ret;
1116 +       register char op;
1117 +       register char *next;
1118 +       int flags;
1119 +
1120 +       ret = regatom(g, &flags);
1121 +       if (ret == NULL)
1122 +               return(NULL);
1123 +
1124 +       op = *g->regparse;
1125 +       if (!ISMULT(op)) {
1126 +               *flagp = flags;
1127 +               return(ret);
1128 +       }
1129 +
1130 +       if (!(flags&HASWIDTH) && op != '?')
1131 +               FAIL("*+ operand could be empty");
1132 +       *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
1133 +
1134 +       if (op == '*' && (flags&SIMPLE))
1135 +               reginsert(g, STAR, ret);
1136 +       else if (op == '*') {
1137 +               /* Emit x* as (x&|), where & means "self". */
1138 +               reginsert(g, BRANCH, ret);                      /* Either x */
1139 +               regoptail(g, ret, regnode(g, BACK));            /* and loop */
1140 +               regoptail(g, ret, ret);                 /* back */
1141 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
1142 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
1143 +       } else if (op == '+' && (flags&SIMPLE))
1144 +               reginsert(g, PLUS, ret);
1145 +       else if (op == '+') {
1146 +               /* Emit x+ as x(&|), where & means "self". */
1147 +               next = regnode(g, BRANCH);                      /* Either */
1148 +               regtail(g, ret, next);
1149 +               regtail(g, regnode(g, BACK), ret);              /* loop back */
1150 +               regtail(g, next, regnode(g, BRANCH));           /* or */
1151 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
1152 +       } else if (op == '?') {
1153 +               /* Emit x? as (x|) */
1154 +               reginsert(g, BRANCH, ret);                      /* Either x */
1155 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
1156 +               next = regnode(g, NOTHING);             /* null. */
1157 +               regtail(g, ret, next);
1158 +               regoptail(g, ret, next);
1159 +       }
1160 +       g->regparse++;
1161 +       if (ISMULT(*g->regparse))
1162 +               FAIL("nested *?+");
1163 +
1164 +       return(ret);
1165 +}
1166 +
1167 +/*
1168 + - regatom - the lowest level
1169 + *
1170 + * Optimization:  gobbles an entire sequence of ordinary characters so that
1171 + * it can turn them into a single node, which is smaller to store and
1172 + * faster to run.  Backslashed characters are exceptions, each becoming a
1173 + * separate node; the code is simpler that way and it's not worth fixing.
1174 + */
1175 +static char *
1176 +regatom(struct match_globals *g, int *flagp)
1177 +{
1178 +       register char *ret;
1179 +       int flags;
1180 +
1181 +       *flagp = WORST;         /* Tentatively. */
1182 +
1183 +       switch (*g->regparse++) {
1184 +       case '^':
1185 +               ret = regnode(g, BOL);
1186 +               break;
1187 +       case '$':
1188 +               ret = regnode(g, EOL);
1189 +               break;
1190 +       case '.':
1191 +               ret = regnode(g, ANY);
1192 +               *flagp |= HASWIDTH|SIMPLE;
1193 +               break;
1194 +       case '[': {
1195 +                       register int class;
1196 +                       register int classend;
1197 +
1198 +                       if (*g->regparse == '^') {      /* Complement of range. */
1199 +                               ret = regnode(g, ANYBUT);
1200 +                               g->regparse++;
1201 +                       } else
1202 +                               ret = regnode(g, ANYOF);
1203 +                       if (*g->regparse == ']' || *g->regparse == '-')
1204 +                               regc(g, *g->regparse++);
1205 +                       while (*g->regparse != '\0' && *g->regparse != ']') {
1206 +                               if (*g->regparse == '-') {
1207 +                                       g->regparse++;
1208 +                                       if (*g->regparse == ']' || *g->regparse == '\0')
1209 +                                               regc(g, '-');
1210 +                                       else {
1211 +                                               class = UCHARAT(g->regparse-2)+1;
1212 +                                               classend = UCHARAT(g->regparse);
1213 +                                               if (class > classend+1)
1214 +                                                       FAIL("invalid [] range");
1215 +                                               for (; class <= classend; class++)
1216 +                                                       regc(g, class);
1217 +                                               g->regparse++;
1218 +                                       }
1219 +                               } else
1220 +                                       regc(g, *g->regparse++);
1221 +                       }
1222 +                       regc(g, '\0');
1223 +                       if (*g->regparse != ']')
1224 +                               FAIL("unmatched []");
1225 +                       g->regparse++;
1226 +                       *flagp |= HASWIDTH|SIMPLE;
1227 +               }
1228 +               break;
1229 +       case '(':
1230 +               ret = reg(g, 1, &flags);
1231 +               if (ret == NULL)
1232 +                       return(NULL);
1233 +               *flagp |= flags&(HASWIDTH|SPSTART);
1234 +               break;
1235 +       case '\0':
1236 +       case '|':
1237 +       case ')':
1238 +               FAIL("internal urp");   /* Supposed to be caught earlier. */
1239 +               break;
1240 +       case '?':
1241 +       case '+':
1242 +       case '*':
1243 +               FAIL("?+* follows nothing");
1244 +               break;
1245 +       case '\\':
1246 +               if (*g->regparse == '\0')
1247 +                       FAIL("trailing \\");
1248 +               ret = regnode(g, EXACTLY);
1249 +               regc(g, *g->regparse++);
1250 +               regc(g, '\0');
1251 +               *flagp |= HASWIDTH|SIMPLE;
1252 +               break;
1253 +       default: {
1254 +                       register int len;
1255 +                       register char ender;
1256 +
1257 +                       g->regparse--;
1258 +                       len = my_strcspn((const char *)g->regparse, (const char *)META);
1259 +                       if (len <= 0)
1260 +                               FAIL("internal disaster");
1261 +                       ender = *(g->regparse+len);
1262 +                       if (len > 1 && ISMULT(ender))
1263 +                               len--;          /* Back off clear of ?+* operand. */
1264 +                       *flagp |= HASWIDTH;
1265 +                       if (len == 1)
1266 +                               *flagp |= SIMPLE;
1267 +                       ret = regnode(g, EXACTLY);
1268 +                       while (len > 0) {
1269 +                               regc(g, *g->regparse++);
1270 +                               len--;
1271 +                       }
1272 +                       regc(g, '\0');
1273 +               }
1274 +               break;
1275 +       }
1276 +
1277 +       return(ret);
1278 +}
1279 +
1280 +/*
1281 + - regnode - emit a node
1282 + */
1283 +static char *                  /* Location. */
1284 +regnode(struct match_globals *g, char op)
1285 +{
1286 +       register char *ret;
1287 +       register char *ptr;
1288 +
1289 +       ret = g->regcode;
1290 +       if (ret == &g->regdummy) {
1291 +               g->regsize += 3;
1292 +               return(ret);
1293 +       }
1294 +
1295 +       ptr = ret;
1296 +       *ptr++ = op;
1297 +       *ptr++ = '\0';          /* Null "next" pointer. */
1298 +       *ptr++ = '\0';
1299 +       g->regcode = ptr;
1300 +
1301 +       return(ret);
1302 +}
1303 +
1304 +/*
1305 + - regc - emit (if appropriate) a byte of code
1306 + */
1307 +static void
1308 +regc(struct match_globals *g, char b)
1309 +{
1310 +       if (g->regcode != &g->regdummy)
1311 +               *g->regcode++ = b;
1312 +       else
1313 +               g->regsize++;
1314 +}
1315 +
1316 +/*
1317 + - reginsert - insert an operator in front of already-emitted operand
1318 + *
1319 + * Means relocating the operand.
1320 + */
1321 +static void
1322 +reginsert(struct match_globals *g, char op, char* opnd)
1323 +{
1324 +       register char *src;
1325 +       register char *dst;
1326 +       register char *place;
1327 +
1328 +       if (g->regcode == &g->regdummy) {
1329 +               g->regsize += 3;
1330 +               return;
1331 +       }
1332 +
1333 +       src = g->regcode;
1334 +       g->regcode += 3;
1335 +       dst = g->regcode;
1336 +       while (src > opnd)
1337 +               *--dst = *--src;
1338 +
1339 +       place = opnd;           /* Op node, where operand used to be. */
1340 +       *place++ = op;
1341 +       *place++ = '\0';
1342 +       *place++ = '\0';
1343 +}
1344 +
1345 +/*
1346 + - regtail - set the next-pointer at the end of a node chain
1347 + */
1348 +static void
1349 +regtail(struct match_globals *g, char *p, char *val)
1350 +{
1351 +       register char *scan;
1352 +       register char *temp;
1353 +       register int offset;
1354 +
1355 +       if (p == &g->regdummy)
1356 +               return;
1357 +
1358 +       /* Find last node. */
1359 +       scan = p;
1360 +       for (;;) {
1361 +               temp = regnext(g, scan);
1362 +               if (temp == NULL)
1363 +                       break;
1364 +               scan = temp;
1365 +       }
1366 +
1367 +       if (OP(scan) == BACK)
1368 +               offset = scan - val;
1369 +       else
1370 +               offset = val - scan;
1371 +       *(scan+1) = (offset>>8)&0377;
1372 +       *(scan+2) = offset&0377;
1373 +}
1374 +
1375 +/*
1376 + - regoptail - regtail on operand of first argument; nop if operandless
1377 + */
1378 +static void
1379 +regoptail(struct match_globals *g, char *p, char *val)
1380 +{
1381 +       /* "Operandless" and "op != BRANCH" are synonymous in practice. */
1382 +       if (p == NULL || p == &g->regdummy || OP(p) != BRANCH)
1383 +               return;
1384 +       regtail(g, OPERAND(p), val);
1385 +}
1386 +
1387 +/*
1388 + * regexec and friends
1389 + */
1390 +
1391 +
1392 +/*
1393 + * Forwards.
1394 + */
1395 +STATIC int regtry(struct match_globals *g, regexp *prog, char *string);
1396 +STATIC int regmatch(struct match_globals *g, char *prog);
1397 +STATIC int regrepeat(struct match_globals *g, char *p);
1398 +
1399 +#ifdef DEBUG
1400 +int regnarrate = 0;
1401 +void regdump();
1402 +STATIC char *regprop(char *op);
1403 +#endif
1404 +
1405 +/*
1406 + - regexec - match a regexp against a string
1407 + */
1408 +int
1409 +regexec(regexp *prog, char *string)
1410 +{
1411 +       register char *s;
1412 +       struct match_globals g;
1413 +
1414 +       /* Be paranoid... */
1415 +       if (prog == NULL || string == NULL) {
1416 +               printk("<3>Regexp: NULL parameter\n");
1417 +               return(0);
1418 +       }
1419 +
1420 +       /* Check validity of program. */
1421 +       if (UCHARAT(prog->program) != MAGIC) {
1422 +               printk("<3>Regexp: corrupted program\n");
1423 +               return(0);
1424 +       }
1425 +
1426 +       /* If there is a "must appear" string, look for it. */
1427 +       if (prog->regmust != NULL) {
1428 +               s = string;
1429 +               while ((s = strchr(s, prog->regmust[0])) != NULL) {
1430 +                       if (strncmp(s, prog->regmust, prog->regmlen) == 0)
1431 +                               break;  /* Found it. */
1432 +                       s++;
1433 +               }
1434 +               if (s == NULL)  /* Not present. */
1435 +                       return(0);
1436 +       }
1437 +
1438 +       /* Mark beginning of line for ^ . */
1439 +       g.regbol = string;
1440 +
1441 +       /* Simplest case:  anchored match need be tried only once. */
1442 +       if (prog->reganch)
1443 +               return(regtry(&g, prog, string));
1444 +
1445 +       /* Messy cases:  unanchored match. */
1446 +       s = string;
1447 +       if (prog->regstart != '\0')
1448 +               /* We know what char it must start with. */
1449 +               while ((s = strchr(s, prog->regstart)) != NULL) {
1450 +                       if (regtry(&g, prog, s))
1451 +                               return(1);
1452 +                       s++;
1453 +               }
1454 +       else
1455 +               /* We don't -- general case. */
1456 +               do {
1457 +                       if (regtry(&g, prog, s))
1458 +                               return(1);
1459 +               } while (*s++ != '\0');
1460 +
1461 +       /* Failure. */
1462 +       return(0);
1463 +}
1464 +
1465 +/*
1466 + - regtry - try match at specific point
1467 + */
1468 +static int                     /* 0 failure, 1 success */
1469 +regtry(struct match_globals *g, regexp *prog, char *string)
1470 +{
1471 +       register int i;
1472 +       register char **sp;
1473 +       register char **ep;
1474 +
1475 +       g->reginput = string;
1476 +       g->regstartp = prog->startp;
1477 +       g->regendp = prog->endp;
1478 +
1479 +       sp = prog->startp;
1480 +       ep = prog->endp;
1481 +       for (i = NSUBEXP; i > 0; i--) {
1482 +               *sp++ = NULL;
1483 +               *ep++ = NULL;
1484 +       }
1485 +       if (regmatch(g, prog->program + 1)) {
1486 +               prog->startp[0] = string;
1487 +               prog->endp[0] = g->reginput;
1488 +               return(1);
1489 +       } else
1490 +               return(0);
1491 +}
1492 +
1493 +/*
1494 + - regmatch - main matching routine
1495 + *
1496 + * Conceptually the strategy is simple:  check to see whether the current
1497 + * node matches, call self recursively to see whether the rest matches,
1498 + * and then act accordingly.  In practice we make some effort to avoid
1499 + * recursion, in particular by going through "ordinary" nodes (that don't
1500 + * need to know whether the rest of the match failed) by a loop instead of
1501 + * by recursion.
1502 + */
1503 +static int                     /* 0 failure, 1 success */
1504 +regmatch(struct match_globals *g, char *prog)
1505 +{
1506 +       register char *scan = prog; /* Current node. */
1507 +       char *next;                 /* Next node. */
1508 +
1509 +#ifdef DEBUG
1510 +       if (scan != NULL && regnarrate)
1511 +               fprintf(stderr, "%s(\n", regprop(scan));
1512 +#endif
1513 +       while (scan != NULL) {
1514 +#ifdef DEBUG
1515 +               if (regnarrate)
1516 +                       fprintf(stderr, "%s...\n", regprop(scan));
1517 +#endif
1518 +               next = regnext(g, scan);
1519 +
1520 +               switch (OP(scan)) {
1521 +               case BOL:
1522 +                       if (g->reginput != g->regbol)
1523 +                               return(0);
1524 +                       break;
1525 +               case EOL:
1526 +                       if (*g->reginput != '\0')
1527 +                               return(0);
1528 +                       break;
1529 +               case ANY:
1530 +                       if (*g->reginput == '\0')
1531 +                               return(0);
1532 +                       g->reginput++;
1533 +                       break;
1534 +               case EXACTLY: {
1535 +                               register int len;
1536 +                               register char *opnd;
1537 +
1538 +                               opnd = OPERAND(scan);
1539 +                               /* Inline the first character, for speed. */
1540 +                               if (*opnd != *g->reginput)
1541 +                                       return(0);
1542 +                               len = strlen(opnd);
1543 +                               if (len > 1 && strncmp(opnd, g->reginput, len) != 0)
1544 +                                       return(0);
1545 +                               g->reginput += len;
1546 +                       }
1547 +                       break;
1548 +               case ANYOF:
1549 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) == NULL)
1550 +                               return(0);
1551 +                       g->reginput++;
1552 +                       break;
1553 +               case ANYBUT:
1554 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) != NULL)
1555 +                               return(0);
1556 +                       g->reginput++;
1557 +                       break;
1558 +               case NOTHING:
1559 +               case BACK:
1560 +                       break;
1561 +               case OPEN+1:
1562 +               case OPEN+2:
1563 +               case OPEN+3:
1564 +               case OPEN+4:
1565 +               case OPEN+5:
1566 +               case OPEN+6:
1567 +               case OPEN+7:
1568 +               case OPEN+8:
1569 +               case OPEN+9: {
1570 +                               register int no;
1571 +                               register char *save;
1572 +
1573 +                               no = OP(scan) - OPEN;
1574 +                               save = g->reginput;
1575 +
1576 +                               if (regmatch(g, next)) {
1577 +                                       /*
1578 +                                        * Don't set startp if some later
1579 +                                        * invocation of the same parentheses
1580 +                                        * already has.
1581 +                                        */
1582 +                                       if (g->regstartp[no] == NULL)
1583 +                                               g->regstartp[no] = save;
1584 +                                       return(1);
1585 +                               } else
1586 +                                       return(0);
1587 +                       }
1588 +                       break;
1589 +               case CLOSE+1:
1590 +               case CLOSE+2:
1591 +               case CLOSE+3:
1592 +               case CLOSE+4:
1593 +               case CLOSE+5:
1594 +               case CLOSE+6:
1595 +               case CLOSE+7:
1596 +               case CLOSE+8:
1597 +               case CLOSE+9:
1598 +                       {
1599 +                               register int no;
1600 +                               register char *save;
1601 +
1602 +                               no = OP(scan) - CLOSE;
1603 +                               save = g->reginput;
1604 +
1605 +                               if (regmatch(g, next)) {
1606 +                                       /*
1607 +                                        * Don't set endp if some later
1608 +                                        * invocation of the same parentheses
1609 +                                        * already has.
1610 +                                        */
1611 +                                       if (g->regendp[no] == NULL)
1612 +                                               g->regendp[no] = save;
1613 +                                       return(1);
1614 +                               } else
1615 +                                       return(0);
1616 +                       }
1617 +                       break;
1618 +               case BRANCH: {
1619 +                               register char *save;
1620 +
1621 +                               if (OP(next) != BRANCH)         /* No choice. */
1622 +                                       next = OPERAND(scan);   /* Avoid recursion. */
1623 +                               else {
1624 +                                       do {
1625 +                                               save = g->reginput;
1626 +                                               if (regmatch(g, OPERAND(scan)))
1627 +                                                       return(1);
1628 +                                               g->reginput = save;
1629 +                                               scan = regnext(g, scan);
1630 +                                       } while (scan != NULL && OP(scan) == BRANCH);
1631 +                                       return(0);
1632 +                                       /* NOTREACHED */
1633 +                               }
1634 +                       }
1635 +                       break;
1636 +               case STAR:
1637 +               case PLUS: {
1638 +                               register char nextch;
1639 +                               register int no;
1640 +                               register char *save;
1641 +                               register int min;
1642 +
1643 +                               /*
1644 +                                * Lookahead to avoid useless match attempts
1645 +                                * when we know what character comes next.
1646 +                                */
1647 +                               nextch = '\0';
1648 +                               if (OP(next) == EXACTLY)
1649 +                                       nextch = *OPERAND(next);
1650 +                               min = (OP(scan) == STAR) ? 0 : 1;
1651 +                               save = g->reginput;
1652 +                               no = regrepeat(g, OPERAND(scan));
1653 +                               while (no >= min) {
1654 +                                       /* If it could work, try it. */
1655 +                                       if (nextch == '\0' || *g->reginput == nextch)
1656 +                                               if (regmatch(g, next))
1657 +                                                       return(1);
1658 +                                       /* Couldn't or didn't -- back up. */
1659 +                                       no--;
1660 +                                       g->reginput = save + no;
1661 +                               }
1662 +                               return(0);
1663 +                       }
1664 +                       break;
1665 +               case END:
1666 +                       return(1);      /* Success! */
1667 +                       break;
1668 +               default:
1669 +                       printk("<3>Regexp: memory corruption\n");
1670 +                       return(0);
1671 +                       break;
1672 +               }
1673 +
1674 +               scan = next;
1675 +       }
1676 +
1677 +       /*
1678 +        * We get here only if there's trouble -- normally "case END" is
1679 +        * the terminating point.
1680 +        */
1681 +       printk("<3>Regexp: corrupted pointers\n");
1682 +       return(0);
1683 +}
1684 +
1685 +/*
1686 + - regrepeat - repeatedly match something simple, report how many
1687 + */
1688 +static int
1689 +regrepeat(struct match_globals *g, char *p)
1690 +{
1691 +       register int count = 0;
1692 +       register char *scan;
1693 +       register char *opnd;
1694 +
1695 +       scan = g->reginput;
1696 +       opnd = OPERAND(p);
1697 +       switch (OP(p)) {
1698 +       case ANY:
1699 +               count = strlen(scan);
1700 +               scan += count;
1701 +               break;
1702 +       case EXACTLY:
1703 +               while (*opnd == *scan) {
1704 +                       count++;
1705 +                       scan++;
1706 +               }
1707 +               break;
1708 +       case ANYOF:
1709 +               while (*scan != '\0' && strchr(opnd, *scan) != NULL) {
1710 +                       count++;
1711 +                       scan++;
1712 +               }
1713 +               break;
1714 +       case ANYBUT:
1715 +               while (*scan != '\0' && strchr(opnd, *scan) == NULL) {
1716 +                       count++;
1717 +                       scan++;
1718 +               }
1719 +               break;
1720 +       default:                /* Oh dear.  Called inappropriately. */
1721 +               printk("<3>Regexp: internal foulup\n");
1722 +               count = 0;      /* Best compromise. */
1723 +               break;
1724 +       }
1725 +       g->reginput = scan;
1726 +
1727 +       return(count);
1728 +}
1729 +
1730 +/*
1731 + - regnext - dig the "next" pointer out of a node
1732 + */
1733 +static char*
1734 +regnext(struct match_globals *g, char *p)
1735 +{
1736 +       register int offset;
1737 +
1738 +       if (p == &g->regdummy)
1739 +               return(NULL);
1740 +
1741 +       offset = NEXT(p);
1742 +       if (offset == 0)
1743 +               return(NULL);
1744 +
1745 +       if (OP(p) == BACK)
1746 +               return(p-offset);
1747 +       else
1748 +               return(p+offset);
1749 +}
1750 +
1751 +#ifdef DEBUG
1752 +
1753 +STATIC char *regprop();
1754 +
1755 +/*
1756 + - regdump - dump a regexp onto stdout in vaguely comprehensible form
1757 + */
1758 +void
1759 +regdump(regexp *r)
1760 +{
1761 +       register char *s;
1762 +       register char op = EXACTLY;     /* Arbitrary non-END op. */
1763 +       register char *next;
1764 +       /* extern char *strchr(); */
1765 +
1766 +
1767 +       s = r->program + 1;
1768 +       while (op != END) {     /* While that wasn't END last time... */
1769 +               op = OP(s);
1770 +               printf("%2d%s", s-r->program, regprop(s));      /* Where, what. */
1771 +               next = regnext(s);
1772 +               if (next == NULL)               /* Next ptr. */
1773 +                       printf("(0)");
1774 +               else
1775 +                       printf("(%d)", (s-r->program)+(next-s));
1776 +               s += 3;
1777 +               if (op == ANYOF || op == ANYBUT || op == EXACTLY) {
1778 +                       /* Literal string, where present. */
1779 +                       while (*s != '\0') {
1780 +                               putchar(*s);
1781 +                               s++;
1782 +                       }
1783 +                       s++;
1784 +               }
1785 +               putchar('\n');
1786 +       }
1787 +
1788 +       /* Header fields of interest. */
1789 +       if (r->regstart != '\0')
1790 +               printf("start `%c' ", r->regstart);
1791 +       if (r->reganch)
1792 +               printf("anchored ");
1793 +       if (r->regmust != NULL)
1794 +               printf("must have \"%s\"", r->regmust);
1795 +       printf("\n");
1796 +}
1797 +
1798 +/*
1799 + - regprop - printable representation of opcode
1800 + */
1801 +static char *
1802 +regprop(char *op)
1803 +{
1804 +#define BUFLEN 50
1805 +       register char *p;
1806 +       static char buf[BUFLEN];
1807 +
1808 +       strcpy(buf, ":");
1809 +
1810 +       switch (OP(op)) {
1811 +       case BOL:
1812 +               p = "BOL";
1813 +               break;
1814 +       case EOL:
1815 +               p = "EOL";
1816 +               break;
1817 +       case ANY:
1818 +               p = "ANY";
1819 +               break;
1820 +       case ANYOF:
1821 +               p = "ANYOF";
1822 +               break;
1823 +       case ANYBUT:
1824 +               p = "ANYBUT";
1825 +               break;
1826 +       case BRANCH:
1827 +               p = "BRANCH";
1828 +               break;
1829 +       case EXACTLY:
1830 +               p = "EXACTLY";
1831 +               break;
1832 +       case NOTHING:
1833 +               p = "NOTHING";
1834 +               break;
1835 +       case BACK:
1836 +               p = "BACK";
1837 +               break;
1838 +       case END:
1839 +               p = "END";
1840 +               break;
1841 +       case OPEN+1:
1842 +       case OPEN+2:
1843 +       case OPEN+3:
1844 +       case OPEN+4:
1845 +       case OPEN+5:
1846 +       case OPEN+6:
1847 +       case OPEN+7:
1848 +       case OPEN+8:
1849 +       case OPEN+9:
1850 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "OPEN%d", OP(op)-OPEN);
1851 +               p = NULL;
1852 +               break;
1853 +       case CLOSE+1:
1854 +       case CLOSE+2:
1855 +       case CLOSE+3:
1856 +       case CLOSE+4:
1857 +       case CLOSE+5:
1858 +       case CLOSE+6:
1859 +       case CLOSE+7:
1860 +       case CLOSE+8:
1861 +       case CLOSE+9:
1862 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "CLOSE%d", OP(op)-CLOSE);
1863 +               p = NULL;
1864 +               break;
1865 +       case STAR:
1866 +               p = "STAR";
1867 +               break;
1868 +       case PLUS:
1869 +               p = "PLUS";
1870 +               break;
1871 +       default:
1872 +               printk("<3>Regexp: corrupted opcode\n");
1873 +               break;
1874 +       }
1875 +       if (p != NULL)
1876 +               strncat(buf, p, BUFLEN-strlen(buf));
1877 +       return(buf);
1878 +}
1879 +#endif
1880 +
1881 +
1882 diff -urN linux-2.6.21.1.old/net/ipv4/netfilter/regexp/regexp.h linux-2.6.21.1.dev/net/ipv4/netfilter/regexp/regexp.h
1883 --- linux-2.6.21.1.old/net/ipv4/netfilter/regexp/regexp.h       1970-01-01 01:00:00.000000000 +0100
1884 +++ linux-2.6.21.1.dev/net/ipv4/netfilter/regexp/regexp.h       2007-05-26 20:13:52.652129512 +0200
1885 @@ -0,0 +1,41 @@
1886 +/*
1887 + * Definitions etc. for regexp(3) routines.
1888 + *
1889 + * Caveat:  this is V8 regexp(3) [actually, a reimplementation thereof],
1890 + * not the System V one.
1891 + */
1892 +
1893 +#ifndef REGEXP_H
1894 +#define REGEXP_H
1895 +
1896 +
1897 +/*
1898 +http://www.opensource.apple.com/darwinsource/10.3/expect-1/expect/expect.h ,
1899 +which contains a version of this library, says:
1900 +
1901 + *
1902 + * NSUBEXP must be at least 10, and no greater than 117 or the parser
1903 + * will not work properly.
1904 + *
1905 +
1906 +However, it looks rather like this library is limited to 10.  If you think
1907 +otherwise, let us know.
1908 +*/
1909 +
1910 +#define NSUBEXP  10
1911 +typedef struct regexp {
1912 +       char *startp[NSUBEXP];
1913 +       char *endp[NSUBEXP];
1914 +       char regstart;          /* Internal use only. */
1915 +       char reganch;           /* Internal use only. */
1916 +       char *regmust;          /* Internal use only. */
1917 +       int regmlen;            /* Internal use only. */
1918 +       char program[1];        /* Unwarranted chumminess with compiler. */
1919 +} regexp;
1920 +
1921 +regexp * regcomp(char *exp, int *patternsize);
1922 +int regexec(regexp *prog, char *string);
1923 +void regsub(regexp *prog, char *source, char *dest);
1924 +void regerror(char *s);
1925 +
1926 +#endif
1927 diff -urN linux-2.6.21.1.old/net/ipv4/netfilter/regexp/regmagic.h linux-2.6.21.1.dev/net/ipv4/netfilter/regexp/regmagic.h
1928 --- linux-2.6.21.1.old/net/ipv4/netfilter/regexp/regmagic.h     1970-01-01 01:00:00.000000000 +0100
1929 +++ linux-2.6.21.1.dev/net/ipv4/netfilter/regexp/regmagic.h     2007-05-26 20:13:52.652129512 +0200
1930 @@ -0,0 +1,5 @@
1931 +/*
1932 + * The first byte of the regexp internal "program" is actually this magic
1933 + * number; the start node begins in the second byte.
1934 + */
1935 +#define        MAGIC   0234
1936 diff -urN linux-2.6.21.1.old/net/ipv4/netfilter/regexp/regsub.c linux-2.6.21.1.dev/net/ipv4/netfilter/regexp/regsub.c
1937 --- linux-2.6.21.1.old/net/ipv4/netfilter/regexp/regsub.c       1970-01-01 01:00:00.000000000 +0100
1938 +++ linux-2.6.21.1.dev/net/ipv4/netfilter/regexp/regsub.c       2007-05-26 20:13:52.653129360 +0200
1939 @@ -0,0 +1,95 @@
1940 +/*
1941 + * regsub
1942 + * @(#)regsub.c        1.3 of 2 April 86
1943 + *
1944 + *     Copyright (c) 1986 by University of Toronto.
1945 + *     Written by Henry Spencer.  Not derived from licensed software.
1946 + *
1947 + *     Permission is granted to anyone to use this software for any
1948 + *     purpose on any computer system, and to redistribute it freely,
1949 + *     subject to the following restrictions:
1950 + *
1951 + *     1. The author is not responsible for the consequences of use of
1952 + *             this software, no matter how awful, even if they arise
1953 + *             from defects in it.
1954 + *
1955 + *     2. The origin of this software must not be misrepresented, either
1956 + *             by explicit claim or by omission.
1957 + *
1958 + *     3. Altered versions must be plainly marked as such, and must not
1959 + *             be misrepresented as being the original software.
1960 + *
1961 + *
1962 + * This code was modified by Ethan Sommer to work within the kernel
1963 + * (it now uses kmalloc etc..)
1964 + *
1965 + */
1966 +#include "regexp.h"
1967 +#include "regmagic.h"
1968 +#include <linux/string.h>
1969 +
1970 +
1971 +#ifndef CHARBITS
1972 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
1973 +#else
1974 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
1975 +#endif
1976 +
1977 +#if 0
1978 +//void regerror(char * s)
1979 +//{
1980 +//        printk("regexp(3): %s", s);
1981 +//        /* NOTREACHED */
1982 +//}
1983 +#endif
1984 +
1985 +/*
1986 + - regsub - perform substitutions after a regexp match
1987 + */
1988 +void
1989 +regsub(regexp * prog, char * source, char * dest)
1990 +{
1991 +       register char *src;
1992 +       register char *dst;
1993 +       register char c;
1994 +       register int no;
1995 +       register int len;
1996 +
1997 +       /* Not necessary and gcc doesn't like it -MLS */
1998 +       /*extern char *strncpy();*/
1999 +
2000 +       if (prog == NULL || source == NULL || dest == NULL) {
2001 +               regerror("NULL parm to regsub");
2002 +               return;
2003 +       }
2004 +       if (UCHARAT(prog->program) != MAGIC) {
2005 +               regerror("damaged regexp fed to regsub");
2006 +               return;
2007 +       }
2008 +
2009 +       src = source;
2010 +       dst = dest;
2011 +       while ((c = *src++) != '\0') {
2012 +               if (c == '&')
2013 +                       no = 0;
2014 +               else if (c == '\\' && '0' <= *src && *src <= '9')
2015 +                       no = *src++ - '0';
2016 +               else
2017 +                       no = -1;
2018 +
2019 +               if (no < 0) {   /* Ordinary character. */
2020 +                       if (c == '\\' && (*src == '\\' || *src == '&'))
2021 +                               c = *src++;
2022 +                       *dst++ = c;
2023 +               } else if (prog->startp[no] != NULL && prog->endp[no] != NULL) {
2024 +                       len = prog->endp[no] - prog->startp[no];
2025 +                       (void) strncpy(dst, prog->startp[no], len);
2026 +                       dst += len;
2027 +                       if (len != 0 && *(dst-1) == '\0') {     /* strncpy hit NUL. */
2028 +                               regerror("damaged match string");
2029 +                               return;
2030 +                       }
2031 +               }
2032 +       }
2033 +       *dst++ = '\0';
2034 +}