kernel-layer7.patch

   1 --- linux-2.6.22-rc7/net/netfilter/Kconfig      2007-07-01 14:54:24.000000000 -0500
   2 +++ linux-2.6.22-rc7-layer7/net/netfilter/Kconfig       2007-07-03 03:21:32.000000000 -0500
   3 @@ -603,6 +603,26 @@ config NETFILTER_XT_MATCH_STATE
   4
   5           To compile it as a module, choose M here.  If unsure, say N.
   6
   7 +config NETFILTER_XT_MATCH_LAYER7
   8 +       tristate '"layer7" match support'
   9 +       depends on NETFILTER_XTABLES
  10 +       depends on EXPERIMENTAL && (IP_NF_CONNTRACK || NF_CONNTRACK)
  11 +       help
  12 +         Say Y if you want to be able to classify connections (and their
  13 +         packets) based on regular expression matching of their application
  14 +         layer data.   This is one way to classify applications such as
  15 +         peer-to-peer filesharing systems that do not always use the same
  16 +         port.
  17 +
  18 +         To compile it as a module, choose M here.  If unsure, say N.
  19 +
  20 +config NETFILTER_XT_MATCH_LAYER7_DEBUG
  21 +        bool 'Layer 7 debugging output'
  22 +        depends on NETFILTER_XT_MATCH_LAYER7
  23 +        help
  24 +          Say Y to get lots of debugging output.
  25 +
  26 +
  27  config NETFILTER_XT_MATCH_STATISTIC
  28         tristate '"statistic" match support'
  29         depends on NETFILTER_XTABLES
  30 --- linux-2.6.22-rc7/net/netfilter/Makefile     2007-07-01 14:54:24.000000000 -0500
  31 +++ linux-2.6.22-rc7-layer7/net/netfilter/Makefile      2007-07-03 01:32:35.000000000 -0500
  32 @@ -0,0 +0,1 @@
  33 +obj-$(CONFIG_NETFILTER_XT_MATCH_LAYER7) += xt_layer7.o
  34 --- linux-2.6.22-rc7/net/netfilter/xt_layer7.c  1969-12-31 18:00:00.000000000 -0600
  35 +++ linux-2.6.22-rc7-layer7/net/netfilter/xt_layer7.c   2007-07-03 07:57:38.000000000 -0500
  36 @@ -0,0 +1,626 @@
  37 +/*
  38 +  Kernel module to match application layer (OSI layer 7) data in connections.
  39 +
  40 +  http://l7-filter.sf.net
  41 +
  42 +  (C) 2003, 2004, 2005, 2006, 2007 Matthew Strait and Ethan Sommer.
  43 +
  44 +  This program is free software; you can redistribute it and/or
  45 +  modify it under the terms of the GNU General Public License
  46 +  as published by the Free Software Foundation; either version
  47 +  2 of the License, or (at your option) any later version.
  48 +  http://www.gnu.org/licenses/gpl.txt
  49 +
  50 +  Based on ipt_string.c (C) 2000 Emmanuel Roger <winfield@freegates.be>,
  51 +  xt_helper.c (C) 2002 Harald Welte and cls_layer7.c (C) 2003 Matthew Strait,
  52 +  Ethan Sommer, Justin Levandoski.
  53 +*/
  54 +
  55 +#include <linux/spinlock.h>
  56 +#include <net/ip.h>
  57 +#include <net/tcp.h>
  58 +#include <linux/module.h>
  59 +#include <linux/skbuff.h>
  60 +#include <linux/netfilter.h>
  61 +#include <net/netfilter/nf_conntrack.h>
  62 +#include <net/netfilter/nf_conntrack_core.h>
  63 +#include <linux/netfilter/x_tables.h>
  64 +#include <linux/netfilter/xt_layer7.h>
  65 +#include <linux/ctype.h>
  66 +#include <linux/proc_fs.h>
  67 +
  68 +#include "regexp/regexp.c"
  69 +
  70 +MODULE_LICENSE("GPL");
  71 +MODULE_AUTHOR("Matthew Strait <quadong@users.sf.net>, Ethan Sommer <sommere@users.sf.net>");
  72 +MODULE_DESCRIPTION("iptables application layer match module");
  73 +MODULE_ALIAS("ipt_layer7");
  74 +MODULE_VERSION("2.0");
  75 +
  76 +static int maxdatalen = 2048; // this is the default
  77 +module_param(maxdatalen, int, 0444);
  78 +MODULE_PARM_DESC(maxdatalen, "maximum bytes of data looked at by l7-filter");
  79 +#ifdef CONFIG_NETFILTER_XT_MATCH_LAYER7_DEBUG
  80 +       #define DPRINTK(format,args...) printk(format,##args)
  81 +#else
  82 +       #define DPRINTK(format,args...)
  83 +#endif
  84 +
  85 +#define TOTAL_PACKETS master_conntrack->counters[IP_CT_DIR_ORIGINAL].packets + \
  86 +                     master_conntrack->counters[IP_CT_DIR_REPLY].packets
  87 +
  88 +/* Number of packets whose data we look at.
  89 +This can be modified through /proc/net/layer7_numpackets */
  90 +static int num_packets = 10;
  91 +
  92 +static struct pattern_cache {
  93 +       char * regex_string;
  94 +       regexp * pattern;
  95 +       struct pattern_cache * next;
  96 +} * first_pattern_cache = NULL;
  97 +
  98 +DEFINE_SPINLOCK(l7_lock);
  99 +
 100 +#ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
 101 +/* Converts an unfriendly string into a friendly one by
 102 +replacing unprintables with periods and all whitespace with " ". */
 103 +static char * friendly_print(unsigned char * s)
 104 +{
 105 +       char * f = kmalloc(strlen(s) + 1, GFP_ATOMIC);
 106 +       int i;
 107 +
 108 +       if(!f) {
 109 +               if (net_ratelimit())
 110 +                       printk(KERN_ERR "layer7: out of memory in "
 111 +                                       "friendly_print, bailing.\n");
 112 +               return NULL;
 113 +       }
 114 +
 115 +       for(i = 0; i < strlen(s); i++){
 116 +               if(isprint(s[i]) && s[i] < 128) f[i] = s[i];
 117 +               else if(isspace(s[i]))          f[i] = ' ';
 118 +               else                            f[i] = '.';
 119 +       }
 120 +       f[i] = '\0';
 121 +       return f;
 122 +}
 123 +
 124 +static char dec2hex(int i)
 125 +{
 126 +       switch (i) {
 127 +               case 0 ... 9:
 128 +                       return (i + '0');
 129 +                       break;
 130 +               case 10 ... 15:
 131 +                       return (i - 10 + 'a');
 132 +                       break;
 133 +               default:
 134 +                       if (net_ratelimit())
 135 +                               printk("layer7: Problem in dec2hex\n");
 136 +                       return '\0';
 137 +       }
 138 +}
 139 +
 140 +static char * hex_print(unsigned char * s)
 141 +{
 142 +       char * g = kmalloc(strlen(s)*3 + 1, GFP_ATOMIC);
 143 +       int i;
 144 +
 145 +       if(!g) {
 146 +              if (net_ratelimit())
 147 +                       printk(KERN_ERR "layer7: out of memory in hex_print, "
 148 +                                       "bailing.\n");
 149 +              return NULL;
 150 +       }
 151 +
 152 +       for(i = 0; i < strlen(s); i++) {
 153 +               g[i*3    ] = dec2hex(s[i]/16);
 154 +               g[i*3 + 1] = dec2hex(s[i]%16);
 155 +               g[i*3 + 2] = ' ';
 156 +       }
 157 +       g[i*3] = '\0';
 158 +
 159 +       return g;
 160 +}
 161 +#endif // DEBUG
 162 +
 163 +/* Use instead of regcomp.  As we expect to be seeing the same regexps over and
 164 +over again, it make sense to cache the results. */
 165 +static regexp * compile_and_cache(const char * regex_string,
 166 +                                  const char * protocol)
 167 +{
 168 +       struct pattern_cache * node               = first_pattern_cache;
 169 +       struct pattern_cache * last_pattern_cache = first_pattern_cache;
 170 +       struct pattern_cache * tmp;
 171 +       unsigned int len;
 172 +
 173 +       while (node != NULL) {
 174 +               if (!strcmp(node->regex_string, regex_string))
 175 +               return node->pattern;
 176 +
 177 +               last_pattern_cache = node;/* points at the last non-NULL node */
 178 +               node = node->next;
 179 +       }
 180 +
 181 +       /* If we reach the end of the list, then we have not yet cached
 182 +          the pattern for this regex. Let's do that now.
 183 +          Be paranoid about running out of memory to avoid list corruption. */
 184 +       tmp = kmalloc(sizeof(struct pattern_cache), GFP_ATOMIC);
 185 +
 186 +       if(!tmp) {
 187 +               if (net_ratelimit())
 188 +                       printk(KERN_ERR "layer7: out of memory in "
 189 +                                       "compile_and_cache, bailing.\n");
 190 +               return NULL;
 191 +       }
 192 +
 193 +       tmp->regex_string  = kmalloc(strlen(regex_string) + 1, GFP_ATOMIC);
 194 +       tmp->pattern       = kmalloc(sizeof(struct regexp),    GFP_ATOMIC);
 195 +       tmp->next = NULL;
 196 +
 197 +       if(!tmp->regex_string || !tmp->pattern) {
 198 +               if (net_ratelimit())
 199 +                       printk(KERN_ERR "layer7: out of memory in "
 200 +                                       "compile_and_cache, bailing.\n");
 201 +               kfree(tmp->regex_string);
 202 +               kfree(tmp->pattern);
 203 +               kfree(tmp);
 204 +               return NULL;
 205 +       }
 206 +
 207 +       /* Ok.  The new node is all ready now. */
 208 +       node = tmp;
 209 +
 210 +       if(first_pattern_cache == NULL) /* list is empty */
 211 +               first_pattern_cache = node; /* make node the beginning */
 212 +       else
 213 +               last_pattern_cache->next = node; /* attach node to the end */
 214 +
 215 +       /* copy the string and compile the regex */
 216 +       len = strlen(regex_string);
 217 +       DPRINTK("About to compile this: \"%s\"\n", regex_string);
 218 +       node->pattern = regcomp((char *)regex_string, &len);
 219 +       if ( !node->pattern ) {
 220 +               if (net_ratelimit())
 221 +                       printk(KERN_ERR "layer7: Error compiling regexp "
 222 +                                       "\"%s\" (%s)\n",
 223 +                                       regex_string, protocol);
 224 +               /* pattern is now cached as NULL, so we won't try again. */
 225 +       }
 226 +
 227 +       strcpy(node->regex_string, regex_string);
 228 +       return node->pattern;
 229 +}
 230 +
 231 +static int can_handle(const struct sk_buff *skb)
 232 +{
 233 +       if(!ip_hdr(skb)) /* not IP */
 234 +               return 0;
 235 +       if(ip_hdr(skb)->protocol != IPPROTO_TCP &&
 236 +          ip_hdr(skb)->protocol != IPPROTO_UDP &&
 237 +          ip_hdr(skb)->protocol != IPPROTO_ICMP)
 238 +               return 0;
 239 +       return 1;
 240 +}
 241 +
 242 +/* Returns offset the into the skb->data that the application data starts */
 243 +static int app_data_offset(const struct sk_buff *skb)
 244 +{
 245 +       /* In case we are ported somewhere (ebtables?) where ip_hdr(skb)
 246 +       isn't set, this can be gotten from 4*(skb->data[0] & 0x0f) as well. */
 247 +       int ip_hl = 4*ip_hdr(skb)->ihl;
 248 +
 249 +       if( ip_hdr(skb)->protocol == IPPROTO_TCP ) {
 250 +               /* 12 == offset into TCP header for the header length field.
 251 +               Can't get this with skb->h.th->doff because the tcphdr
 252 +               struct doesn't get set when routing (this is confirmed to be
 253 +               true in Netfilter as well as QoS.) */
 254 +               int tcp_hl = 4*(skb->data[ip_hl + 12] >> 4);
 255 +
 256 +               return ip_hl + tcp_hl;
 257 +       } else if( ip_hdr(skb)->protocol == IPPROTO_UDP  ) {
 258 +               return ip_hl + 8; /* UDP header is always 8 bytes */
 259 +       } else if( ip_hdr(skb)->protocol == IPPROTO_ICMP ) {
 260 +               return ip_hl + 8; /* ICMP header is 8 bytes */
 261 +       } else {
 262 +               if (net_ratelimit())
 263 +                       printk(KERN_ERR "layer7: tried to handle unknown "
 264 +                                       "protocol!\n");
 265 +               return ip_hl + 8; /* something reasonable */
 266 +       }
 267 +}
 268 +
 269 +/* handles whether there's a match when we aren't appending data anymore */
 270 +static int match_no_append(struct nf_conn * conntrack,
 271 +                           struct nf_conn * master_conntrack,
 272 +                           enum ip_conntrack_info ctinfo,
 273 +                           enum ip_conntrack_info master_ctinfo,
 274 +                           const struct xt_layer7_info * info)
 275 +{
 276 +       /* If we're in here, throw the app data away */
 277 +       if(master_conntrack->layer7.app_data != NULL) {
 278 +
 279 +       #ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
 280 +               if(!master_conntrack->layer7.app_proto) {
 281 +                       char * f =
 282 +                         friendly_print(master_conntrack->layer7.app_data);
 283 +                       char * g =
 284 +                         hex_print(master_conntrack->layer7.app_data);
 285 +                       DPRINTK("\nl7-filter gave up after %d bytes "
 286 +                               "(%d packets):\n%s\n",
 287 +                               strlen(f), TOTAL_PACKETS, f);
 288 +                       kfree(f);
 289 +                       DPRINTK("In hex: %s\n", g);
 290 +                       kfree(g);
 291 +               }
 292 +       #endif
 293 +
 294 +               kfree(master_conntrack->layer7.app_data);
 295 +               master_conntrack->layer7.app_data = NULL; /* don't free again */
 296 +       }
 297 +
 298 +       if(master_conntrack->layer7.app_proto){
 299 +               /* Here child connections set their .app_proto (for /proc) */
 300 +               if(!conntrack->layer7.app_proto) {
 301 +                       conntrack->layer7.app_proto =
 302 +                         kmalloc(strlen(master_conntrack->layer7.app_proto)+1,
 303 +                           GFP_ATOMIC);
 304 +                       if(!conntrack->layer7.app_proto){
 305 +                               if (net_ratelimit())
 306 +                                       printk(KERN_ERR "layer7: out of memory "
 307 +                                                       "in match_no_append, "
 308 +                                                       "bailing.\n");
 309 +                               return 1;
 310 +                       }
 311 +                       strcpy(conntrack->layer7.app_proto,
 312 +                               master_conntrack->layer7.app_proto);
 313 +               }
 314 +
 315 +               return (!strcmp(master_conntrack->layer7.app_proto,
 316 +                               info->protocol));
 317 +       }
 318 +       else {
 319 +               /* If not classified, set to "unknown" to distinguish from
 320 +               connections that are still being tested. */
 321 +               master_conntrack->layer7.app_proto =
 322 +                       kmalloc(strlen("unknown")+1, GFP_ATOMIC);
 323 +               if(!master_conntrack->layer7.app_proto){
 324 +                       if (net_ratelimit())
 325 +                               printk(KERN_ERR "layer7: out of memory in "
 326 +                                               "match_no_append, bailing.\n");
 327 +                       return 1;
 328 +               }
 329 +               strcpy(master_conntrack->layer7.app_proto, "unknown");
 330 +               return 0;
 331 +       }
 332 +}
 333 +
 334 +/* add the new app data to the conntrack.  Return number of bytes added. */
 335 +static int add_data(struct nf_conn * master_conntrack,
 336 +                    char * app_data, int appdatalen)
 337 +{
 338 +       int length = 0, i;
 339 +       int oldlength = master_conntrack->layer7.app_data_len;
 340 +
 341 +       /* This is a fix for a race condition by Deti Fliegl. However, I'm not
 342 +          clear on whether the race condition exists or whether this really
 343 +          fixes it.  I might just be being dense... Anyway, if it's not really
 344 +          a fix, all it does is waste a very small amount of time. */
 345 +       if(!master_conntrack->layer7.app_data) return 0;
 346 +
 347 +       /* Strip nulls. Make everything lower case (our regex lib doesn't
 348 +       do case insensitivity).  Add it to the end of the current data. */
 349 +       for(i = 0; i < maxdatalen-oldlength-1 &&
 350 +                  i < appdatalen; i++) {
 351 +               if(app_data[i] != '\0') {
 352 +                       /* the kernel version of tolower mungs 'upper ascii' */
 353 +                       master_conntrack->layer7.app_data[length+oldlength] =
 354 +                               isascii(app_data[i])?
 355 +                                       tolower(app_data[i]) : app_data[i];
 356 +                       length++;
 357 +               }
 358 +       }
 359 +
 360 +       master_conntrack->layer7.app_data[length+oldlength] = '\0';
 361 +       master_conntrack->layer7.app_data_len = length + oldlength;
 362 +
 363 +       return length;
 364 +}
 365 +
 366 +/* taken from drivers/video/modedb.c */
 367 +static int my_atoi(const char *s)
 368 +{
 369 +       int val = 0;
 370 +
 371 +       for (;; s++) {
 372 +               switch (*s) {
 373 +                       case '0'...'9':
 374 +                       val = 10*val+(*s-'0');
 375 +                       break;
 376 +               default:
 377 +                       return val;
 378 +               }
 379 +       }
 380 +}
 381 +
 382 +/* write out num_packets to userland. */
 383 +static int layer7_read_proc(char* page, char ** start, off_t off, int count,
 384 +                            int* eof, void * data)
 385 +{
 386 +       if(num_packets > 99 && net_ratelimit())
 387 +               printk(KERN_ERR "layer7: NOT REACHED. num_packets too big\n");
 388 +
 389 +       page[0] = num_packets/10 + '0';
 390 +       page[1] = num_packets%10 + '0';
 391 +       page[2] = '\n';
 392 +       page[3] = '\0';
 393 +
 394 +       *eof=1;
 395 +
 396 +       return 3;
 397 +}
 398 +
 399 +/* Read in num_packets from userland */
 400 +static int layer7_write_proc(struct file* file, const char* buffer,
 401 +                             unsigned long count, void *data)
 402 +{
 403 +       char * foo = kmalloc(count, GFP_ATOMIC);
 404 +
 405 +       if(!foo){
 406 +               if (net_ratelimit())
 407 +                       printk(KERN_ERR "layer7: out of memory, bailing. "
 408 +                                       "num_packets unchanged.\n");
 409 +               return count;
 410 +       }
 411 +
 412 +       if(copy_from_user(foo, buffer, count)) {
 413 +               return -EFAULT;
 414 +       }
 415 +
 416 +
 417 +       num_packets = my_atoi(foo);
 418 +       kfree (foo);
 419 +
 420 +       /* This has an arbitrary limit to make the math easier. I'm lazy.
 421 +       But anyway, 99 is a LOT! If you want more, you're doing it wrong! */
 422 +       if(num_packets > 99) {
 423 +               printk(KERN_WARNING "layer7: num_packets can't be > 99.\n");
 424 +               num_packets = 99;
 425 +       } else if(num_packets < 1) {
 426 +               printk(KERN_WARNING "layer7: num_packets can't be < 1.\n");
 427 +               num_packets = 1;
 428 +       }
 429 +
 430 +       return count;
 431 +}
 432 +
 433 +static int
 434 +match(const struct sk_buff *skbin,
 435 +      const struct net_device *in,
 436 +      const struct net_device *out,
 437 +      const struct xt_match *match,
 438 +      const void *matchinfo,
 439 +      int offset,
 440 +      unsigned int protoff,
 441 +      int *hotdrop)
 442 +{
 443 +       /* sidestep const without getting a compiler warning... */
 444 +       struct sk_buff * skb = (struct sk_buff *)skbin;
 445 +
 446 +       const struct xt_layer7_info * info = matchinfo;
 447 +       enum ip_conntrack_info master_ctinfo, ctinfo;
 448 +       struct nf_conn *master_conntrack, *conntrack;
 449 +       unsigned char * app_data;
 450 +       unsigned int pattern_result, appdatalen;
 451 +       regexp * comppattern;
 452 +
 453 +       /* Be paranoid/incompetent - lock the entire match function. */
 454 +       spin_lock_bh(&l7_lock);
 455 +
 456 +       if(!can_handle(skb)){
 457 +               DPRINTK("layer7: This is some protocol I can't handle.\n");
 458 +               spin_unlock_bh(&l7_lock);
 459 +               return info->invert;
 460 +       }
 461 +
 462 +       /* Treat parent & all its children together as one connection, except
 463 +       for the purpose of setting conntrack->layer7.app_proto in the actual
 464 +       connection. This makes /proc/net/ip_conntrack more satisfying. */
 465 +       if(!(conntrack = nf_ct_get(skb, &ctinfo)) ||
 466 +          !(master_conntrack=nf_ct_get(skb,&master_ctinfo))){
 467 +               DPRINTK("layer7: couldn't get conntrack.\n");
 468 +               spin_unlock_bh(&l7_lock);
 469 +               return info->invert;
 470 +       }
 471 +
 472 +       /* Try to get a master conntrack (and its master etc) for FTP, etc. */
 473 +       while (master_ct(master_conntrack) != NULL)
 474 +               master_conntrack = master_ct(master_conntrack);
 475 +
 476 +       /* if we've classified it or seen too many packets */
 477 +       if(TOTAL_PACKETS > num_packets ||
 478 +          master_conntrack->layer7.app_proto) {
 479 +
 480 +               pattern_result = match_no_append(conntrack, master_conntrack,
 481 +                                                ctinfo, master_ctinfo, info);
 482 +
 483 +               /* skb->cb[0] == seen. Don't do things twice if there are
 484 +               multiple l7 rules. I'm not sure that using cb for this purpose
 485 +               is correct, even though it says "put your private variables
 486 +               there". But it doesn't look like it is being used for anything
 487 +               else in the skbs that make it here. */
 488 +               skb->cb[0] = 1; /* marking it seen here's probably irrelevant */
 489 +
 490 +               spin_unlock_bh(&l7_lock);
 491 +               return (pattern_result ^ info->invert);
 492 +       }
 493 +
 494 +       if(skb_is_nonlinear(skb)){
 495 +               if(skb_linearize(skb) != 0){
 496 +                       if (net_ratelimit())
 497 +                               printk(KERN_ERR "layer7: failed to linearize "
 498 +                                               "packet, bailing.\n");
 499 +                       spin_unlock_bh(&l7_lock);
 500 +                       return info->invert;
 501 +               }
 502 +       }
 503 +
 504 +       /* now that the skb is linearized, it's safe to set these. */
 505 +       app_data = skb->data + app_data_offset(skb);
 506 +       appdatalen = skb_tail_pointer(skb) - app_data;
 507 +
 508 +       /* the return value gets checked later, when we're ready to use it */
 509 +       comppattern = compile_and_cache(info->pattern, info->protocol);
 510 +
 511 +       /* On the first packet of a connection, allocate space for app data */
 512 +       if(TOTAL_PACKETS == 1 && !skb->cb[0] &&
 513 +          !master_conntrack->layer7.app_data){
 514 +               master_conntrack->layer7.app_data =
 515 +                       kmalloc(maxdatalen, GFP_ATOMIC);
 516 +               if(!master_conntrack->layer7.app_data){
 517 +                       if (net_ratelimit())
 518 +                               printk(KERN_ERR "layer7: out of memory in "
 519 +                                               "match, bailing.\n");
 520 +                       spin_unlock_bh(&l7_lock);
 521 +                       return info->invert;
 522 +               }
 523 +
 524 +               master_conntrack->layer7.app_data[0] = '\0';
 525 +       }
 526 +
 527 +       /* Can be here, but unallocated, if numpackets is increased near
 528 +       the beginning of a connection */
 529 +       if(master_conntrack->layer7.app_data == NULL){
 530 +               spin_unlock_bh(&l7_lock);
 531 +               return (info->invert); /* unmatched */
 532 +       }
 533 +
 534 +       if(!skb->cb[0]){
 535 +               int newbytes;
 536 +               newbytes = add_data(master_conntrack, app_data, appdatalen);
 537 +
 538 +               if(newbytes == 0) { /* didn't add any data */
 539 +                       skb->cb[0] = 1;
 540 +                       /* Didn't match before, not going to match now */
 541 +                       spin_unlock_bh(&l7_lock);
 542 +                       return info->invert;
 543 +               }
 544 +       }
 545 +
 546 +       /* If looking for "unknown", then never match.  "Unknown" means that
 547 +       we've given up; we're still trying with these packets. */
 548 +       if(!strcmp(info->protocol, "unknown")) {
 549 +               pattern_result = 0;
 550 +       /* If looking for "unset", then always match. "Unset" means that we
 551 +       haven't yet classified the connection. */
 552 +       } else if(!strcmp(info->protocol, "unset")) {
 553 +               pattern_result = 2;
 554 +               DPRINTK("layer7: matched unset: not yet classified "
 555 +                       "(%d/%d packets)\n", TOTAL_PACKETS, num_packets);
 556 +       /* If the regexp failed to compile, don't bother running it */
 557 +       } else if(comppattern &&
 558 +                 regexec(comppattern, master_conntrack->layer7.app_data)){
 559 +               DPRINTK("layer7: matched %s\n", info->protocol);
 560 +               pattern_result = 1;
 561 +       } else pattern_result = 0;
 562 +
 563 +       if(pattern_result == 1) {
 564 +               master_conntrack->layer7.app_proto =
 565 +                       kmalloc(strlen(info->protocol)+1, GFP_ATOMIC);
 566 +               if(!master_conntrack->layer7.app_proto){
 567 +                       if (net_ratelimit())
 568 +                               printk(KERN_ERR "layer7: out of memory in "
 569 +                                               "match, bailing.\n");
 570 +                       spin_unlock_bh(&l7_lock);
 571 +                       return (pattern_result ^ info->invert);
 572 +               }
 573 +               strcpy(master_conntrack->layer7.app_proto, info->protocol);
 574 +       } else if(pattern_result > 1) { /* cleanup from "unset" */
 575 +               pattern_result = 1;
 576 +       }
 577 +
 578 +       /* mark the packet seen */
 579 +       skb->cb[0] = 1;
 580 +
 581 +       spin_unlock_bh(&l7_lock);
 582 +       return (pattern_result ^ info->invert);
 583 +}
 584 +
 585 +static int check(const char *tablename,
 586 +                const void *inf,
 587 +                const struct xt_match *match,
 588 +                void *matchinfo,
 589 +                unsigned int hook_mask)
 590 +
 591 +{
 592 +       // load nf_conntrack_ipv4
 593 +        if (nf_ct_l3proto_try_module_get(match->family) < 0) {
 594 +                printk(KERN_WARNING "can't load conntrack support for "
 595 +                                    "proto=%d\n", match->family);
 596 +                return 0;
 597 +        }
 598 +       return 1;
 599 +}
 600 +
 601 +static void
 602 +destroy(const struct xt_match *match, void *matchinfo)
 603 +{
 604 +       nf_ct_l3proto_module_put(match->family);
 605 +}
 606 +
 607 +static struct xt_match xt_layer7_match[] = {
 608 +{
 609 +       .name           = "layer7",
 610 +       .family         = AF_INET,
 611 +       .checkentry     = check,
 612 +       .match          = match,
 613 +       .destroy        = destroy,
 614 +       .matchsize      = sizeof(struct xt_layer7_info),
 615 +       .me             = THIS_MODULE
 616 +}
 617 +};
 618 +
 619 +static void layer7_cleanup_proc(void)
 620 +{
 621 +       remove_proc_entry("layer7_numpackets", proc_net);
 622 +}
 623 +
 624 +/* register the proc file */
 625 +static void layer7_init_proc(void)
 626 +{
 627 +       struct proc_dir_entry* entry;
 628 +       entry = create_proc_entry("layer7_numpackets", 0644, proc_net);
 629 +       entry->read_proc = layer7_read_proc;
 630 +       entry->write_proc = layer7_write_proc;
 631 +}
 632 +
 633 +static int __init xt_layer7_init(void)
 634 +{
 635 +       need_conntrack();
 636 +
 637 +       layer7_init_proc();
 638 +       if(maxdatalen < 1) {
 639 +               printk(KERN_WARNING "layer7: maxdatalen can't be < 1, "
 640 +                       "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, "
 647 +                       "using 65536\n");
 648 +               maxdatalen = 65536;
 649 +       }
 650 +       return xt_register_matches(xt_layer7_match,
 651 +                                  ARRAY_SIZE(xt_layer7_match));
 652 +}
 653 +
 654 +static void __exit xt_layer7_fini(void)
 655 +{
 656 +       layer7_cleanup_proc();
 657 +       xt_unregister_matches(xt_layer7_match, ARRAY_SIZE(xt_layer7_match));
 658 +}
 659 +
 660 +module_init(xt_layer7_init);
 661 +module_exit(xt_layer7_fini);
 662 +
 663 --- linux-2.6.22-rc7/net/netfilter/regexp/regexp.c      1969-12-31 18:00:00.000000000 -0600
 664 +++ linux-2.6.22-rc7-layer7/net/netfilter/regexp/regexp.c       2007-07-03 01:32:35.000000000 -0500
 665 @@ -0,0 +1,1197 @@
 666 +/*
 667 + * regcomp and regexec -- regsub and regerror are elsewhere
 668 + * @(#)regexp.c        1.3 of 18 April 87
 669 + *
 670 + *     Copyright (c) 1986 by University of Toronto.
 671 + *     Written by Henry Spencer.  Not derived from licensed software.
 672 + *
 673 + *     Permission is granted to anyone to use this software for any
 674 + *     purpose on any computer system, and to redistribute it freely,
 675 + *     subject to the following restrictions:
 676 + *
 677 + *     1. The author is not responsible for the consequences of use of
 678 + *             this software, no matter how awful, even if they arise
 679 + *             from defects in it.
 680 + *
 681 + *     2. The origin of this software must not be misrepresented, either
 682 + *             by explicit claim or by omission.
 683 + *
 684 + *     3. Altered versions must be plainly marked as such, and must not
 685 + *             be misrepresented as being the original software.
 686 + *
 687 + * Beware that some of this code is subtly aware of the way operator
 688 + * precedence is structured in regular expressions.  Serious changes in
 689 + * regular-expression syntax might require a total rethink.
 690 + *
 691 + * This code was modified by Ethan Sommer to work within the kernel
 692 + * (it now uses kmalloc etc..)
 693 + *
 694 + * Modified slightly by Matthew Strait to use more modern C.
 695 + */
 696 +
 697 +#include "regexp.h"
 698 +#include "regmagic.h"
 699 +
 700 +/* added by ethan and matt.  Lets it work in both kernel and user space.
 701 +(So iptables can use it, for instance.)  Yea, it goes both ways... */
 702 +#if __KERNEL__
 703 +  #define malloc(foo) kmalloc(foo,GFP_ATOMIC)
 704 +#else
 705 +  #define printk(format,args...) printf(format,##args)
 706 +#endif
 707 +
 708 +void regerror(char * s)
 709 +{
 710 +        printk("<3>Regexp: %s\n", s);
 711 +        /* NOTREACHED */
 712 +}
 713 +
 714 +/*
 715 + * The "internal use only" fields in regexp.h are present to pass info from
 716 + * compile to execute that permits the execute phase to run lots faster on
 717 + * simple cases.  They are:
 718 + *
 719 + * regstart    char that must begin a match; '\0' if none obvious
 720 + * reganch     is the match anchored (at beginning-of-line only)?
 721 + * regmust     string (pointer into program) that match must include, or NULL
 722 + * regmlen     length of regmust string
 723 + *
 724 + * Regstart and reganch permit very fast decisions on suitable starting points
 725 + * for a match, cutting down the work a lot.  Regmust permits fast rejection
 726 + * of lines that cannot possibly match.  The regmust tests are costly enough
 727 + * that regcomp() supplies a regmust only if the r.e. contains something
 728 + * potentially expensive (at present, the only such thing detected is * or +
 729 + * at the start of the r.e., which can involve a lot of backup).  Regmlen is
 730 + * supplied because the test in regexec() needs it and regcomp() is computing
 731 + * it anyway.
 732 + */
 733 +
 734 +/*
 735 + * Structure for regexp "program".  This is essentially a linear encoding
 736 + * of a nondeterministic finite-state machine (aka syntax charts or
 737 + * "railroad normal form" in parsing technology).  Each node is an opcode
 738 + * plus a "next" pointer, possibly plus an operand.  "Next" pointers of
 739 + * all nodes except BRANCH implement concatenation; a "next" pointer with
 740 + * a BRANCH on both ends of it is connecting two alternatives.  (Here we
 741 + * have one of the subtle syntax dependencies:  an individual BRANCH (as
 742 + * opposed to a collection of them) is never concatenated with anything
 743 + * because of operator precedence.)  The operand of some types of node is
 744 + * a literal string; for others, it is a node leading into a sub-FSM.  In
 745 + * particular, the operand of a BRANCH node is the first node of the branch.
 746 + * (NB this is *not* a tree structure:  the tail of the branch connects
 747 + * to the thing following the set of BRANCHes.)  The opcodes are:
 748 + */
 749 +
 750 +/* definition  number  opnd?   meaning */
 751 +#define        END     0       /* no   End of program. */
 752 +#define        BOL     1       /* no   Match "" at beginning of line. */
 753 +#define        EOL     2       /* no   Match "" at end of line. */
 754 +#define        ANY     3       /* no   Match any one character. */
 755 +#define        ANYOF   4       /* str  Match any character in this string. */
 756 +#define        ANYBUT  5       /* str  Match any character not in this string. */
 757 +#define        BRANCH  6       /* node Match this alternative, or the next... */
 758 +#define        BACK    7       /* no   Match "", "next" ptr points backward. */
 759 +#define        EXACTLY 8       /* str  Match this string. */
 760 +#define        NOTHING 9       /* no   Match empty string. */
 761 +#define        STAR    10      /* node Match this (simple) thing 0 or more times. */
 762 +#define        PLUS    11      /* node Match this (simple) thing 1 or more times. */
 763 +#define        OPEN    20      /* no   Mark this point in input as start of #n. */
 764 +                       /*      OPEN+1 is number 1, etc. */
 765 +#define        CLOSE   30      /* no   Analogous to OPEN. */
 766 +
 767 +/*
 768 + * Opcode notes:
 769 + *
 770 + * BRANCH      The set of branches constituting a single choice are hooked
 771 + *             together with their "next" pointers, since precedence prevents
 772 + *             anything being concatenated to any individual branch.  The
 773 + *             "next" pointer of the last BRANCH in a choice points to the
 774 + *             thing following the whole choice.  This is also where the
 775 + *             final "next" pointer of each individual branch points; each
 776 + *             branch starts with the operand node of a BRANCH node.
 777 + *
 778 + * BACK                Normal "next" pointers all implicitly point forward; BACK
 779 + *             exists to make loop structures possible.
 780 + *
 781 + * STAR,PLUS   '?', and complex '*' and '+', are implemented as circular
 782 + *             BRANCH structures using BACK.  Simple cases (one character
 783 + *             per match) are implemented with STAR and PLUS for speed
 784 + *             and to minimize recursive plunges.
 785 + *
 786 + * OPEN,CLOSE  ...are numbered at compile time.
 787 + */
 788 +
 789 +/*
 790 + * A node is one char of opcode followed by two chars of "next" pointer.
 791 + * "Next" pointers are stored as two 8-bit pieces, high order first.  The
 792 + * value is a positive offset from the opcode of the node containing it.
 793 + * An operand, if any, simply follows the node.  (Note that much of the
 794 + * code generation knows about this implicit relationship.)
 795 + *
 796 + * Using two bytes for the "next" pointer is vast overkill for most things,
 797 + * but allows patterns to get big without disasters.
 798 + */
 799 +#define        OP(p)   (*(p))
 800 +#define        NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377))
 801 +#define        OPERAND(p)      ((p) + 3)
 802 +
 803 +/*
 804 + * See regmagic.h for one further detail of program structure.
 805 + */
 806 +
 807 +
 808 +/*
 809 + * Utility definitions.
 810 + */
 811 +#ifndef CHARBITS
 812 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
 813 +#else
 814 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
 815 +#endif
 816 +
 817 +#define        FAIL(m) { regerror(m); return(NULL); }
 818 +#define        ISMULT(c)       ((c) == '*' || (c) == '+' || (c) == '?')
 819 +#define        META    "^$.[()|?+*\\"
 820 +
 821 +/*
 822 + * Flags to be passed up and down.
 823 + */
 824 +#define        HASWIDTH        01      /* Known never to match null string. */
 825 +#define        SIMPLE          02      /* Simple enough to be STAR/PLUS operand. */
 826 +#define        SPSTART         04      /* Starts with * or +. */
 827 +#define        WORST           0       /* Worst case. */
 828 +
 829 +/*
 830 + * Global work variables for regcomp().
 831 + */
 832 +struct match_globals {
 833 +char *reginput;                /* String-input pointer. */
 834 +char *regbol;          /* Beginning of input, for ^ check. */
 835 +char **regstartp;      /* Pointer to startp array. */
 836 +char **regendp;                /* Ditto for endp. */
 837 +char *regparse;                /* Input-scan pointer. */
 838 +int regnpar;           /* () count. */
 839 +char regdummy;
 840 +char *regcode;         /* Code-emit pointer; &regdummy = don't. */
 841 +long regsize;          /* Code size. */
 842 +};
 843 +
 844 +/*
 845 + * Forward declarations for regcomp()'s friends.
 846 + */
 847 +#ifndef STATIC
 848 +#define        STATIC  static
 849 +#endif
 850 +STATIC char *reg(struct match_globals *g, int paren,int *flagp);
 851 +STATIC char *regbranch(struct match_globals *g, int *flagp);
 852 +STATIC char *regpiece(struct match_globals *g, int *flagp);
 853 +STATIC char *regatom(struct match_globals *g, int *flagp);
 854 +STATIC char *regnode(struct match_globals *g, char op);
 855 +STATIC char *regnext(struct match_globals *g, char *p);
 856 +STATIC void regc(struct match_globals *g, char b);
 857 +STATIC void reginsert(struct match_globals *g, char op, char *opnd);
 858 +STATIC void regtail(struct match_globals *g, char *p, char *val);
 859 +STATIC void regoptail(struct match_globals *g, char *p, char *val);
 860 +
 861 +
 862 +__kernel_size_t my_strcspn(const char *s1,const char *s2)
 863 +{
 864 +        char *scan1;
 865 +        char *scan2;
 866 +        int count;
 867 +
 868 +        count = 0;
 869 +        for (scan1 = (char *)s1; *scan1 != '\0'; scan1++) {
 870 +                for (scan2 = (char *)s2; *scan2 != '\0';)       /* ++ moved down. */
 871 +                        if (*scan1 == *scan2++)
 872 +                                return(count);
 873 +                count++;
 874 +        }
 875 +        return(count);
 876 +}
 877 +
 878 +/*
 879 + - regcomp - compile a regular expression into internal code
 880 + *
 881 + * We can't allocate space until we know how big the compiled form will be,
 882 + * but we can't compile it (and thus know how big it is) until we've got a
 883 + * place to put the code.  So we cheat:  we compile it twice, once with code
 884 + * generation turned off and size counting turned on, and once "for real".
 885 + * This also means that we don't allocate space until we are sure that the
 886 + * thing really will compile successfully, and we never have to move the
 887 + * code and thus invalidate pointers into it.  (Note that it has to be in
 888 + * one piece because free() must be able to free it all.)
 889 + *
 890 + * Beware that the optimization-preparation code in here knows about some
 891 + * of the structure of the compiled regexp.
 892 + */
 893 +regexp *
 894 +regcomp(char *exp,int *patternsize)
 895 +{
 896 +       register regexp *r;
 897 +       register char *scan;
 898 +       register char *longest;
 899 +       register int len;
 900 +       int flags;
 901 +       struct match_globals g;
 902 +
 903 +       /* commented out by ethan
 904 +          extern char *malloc();
 905 +       */
 906 +
 907 +       if (exp == NULL)
 908 +               FAIL("NULL argument");
 909 +
 910 +       /* First pass: determine size, legality. */
 911 +       g.regparse = exp;
 912 +       g.regnpar = 1;
 913 +       g.regsize = 0L;
 914 +       g.regcode = &g.regdummy;
 915 +       regc(&g, MAGIC);
 916 +       if (reg(&g, 0, &flags) == NULL)
 917 +               return(NULL);
 918 +
 919 +       /* Small enough for pointer-storage convention? */
 920 +       if (g.regsize >= 32767L)                /* Probably could be 65535L. */
 921 +               FAIL("regexp too big");
 922 +
 923 +       /* Allocate space. */
 924 +       *patternsize=sizeof(regexp) + (unsigned)g.regsize;
 925 +       r = (regexp *)malloc(sizeof(regexp) + (unsigned)g.regsize);
 926 +       if (r == NULL)
 927 +               FAIL("out of space");
 928 +
 929 +       /* Second pass: emit code. */
 930 +       g.regparse = exp;
 931 +       g.regnpar = 1;
 932 +       g.regcode = r->program;
 933 +       regc(&g, MAGIC);
 934 +       if (reg(&g, 0, &flags) == NULL)
 935 +               return(NULL);
 936 +
 937 +       /* Dig out information for optimizations. */
 938 +       r->regstart = '\0';     /* Worst-case defaults. */
 939 +       r->reganch = 0;
 940 +       r->regmust = NULL;
 941 +       r->regmlen = 0;
 942 +       scan = r->program+1;                    /* First BRANCH. */
 943 +       if (OP(regnext(&g, scan)) == END) {             /* Only one top-level choice. */
 944 +               scan = OPERAND(scan);
 945 +
 946 +               /* Starting-point info. */
 947 +               if (OP(scan) == EXACTLY)
 948 +                       r->regstart = *OPERAND(scan);
 949 +               else if (OP(scan) == BOL)
 950 +                       r->reganch++;
 951 +
 952 +               /*
 953 +                * If there's something expensive in the r.e., find the
 954 +                * longest literal string that must appear and make it the
 955 +                * regmust.  Resolve ties in favor of later strings, since
 956 +                * the regstart check works with the beginning of the r.e.
 957 +                * and avoiding duplication strengthens checking.  Not a
 958 +                * strong reason, but sufficient in the absence of others.
 959 +                */
 960 +               if (flags&SPSTART) {
 961 +                       longest = NULL;
 962 +                       len = 0;
 963 +                       for (; scan != NULL; scan = regnext(&g, scan))
 964 +                               if (OP(scan) == EXACTLY && strlen(OPERAND(scan)) >= len) {
 965 +                                       longest = OPERAND(scan);
 966 +                                       len = strlen(OPERAND(scan));
 967 +                               }
 968 +                       r->regmust = longest;
 969 +                       r->regmlen = len;
 970 +               }
 971 +       }
 972 +
 973 +       return(r);
 974 +}
 975 +
 976 +/*
 977 + - reg - regular expression, i.e. main body or parenthesized thing
 978 + *
 979 + * Caller must absorb opening parenthesis.
 980 + *
 981 + * Combining parenthesis handling with the base level of regular expression
 982 + * is a trifle forced, but the need to tie the tails of the branches to what
 983 + * follows makes it hard to avoid.
 984 + */
 985 +static char *
 986 +reg(struct match_globals *g, int paren, int *flagp /* Parenthesized? */ )
 987 +{
 988 +       register char *ret;
 989 +       register char *br;
 990 +       register char *ender;
 991 +       register int parno = 0; /* 0 makes gcc happy */
 992 +       int flags;
 993 +
 994 +       *flagp = HASWIDTH;      /* Tentatively. */
 995 +
 996 +       /* Make an OPEN node, if parenthesized. */
 997 +       if (paren) {
 998 +               if (g->regnpar >= NSUBEXP)
 999 +                       FAIL("too many ()");
1000 +               parno = g->regnpar;
1001 +               g->regnpar++;
1002 +               ret = regnode(g, OPEN+parno);
1003 +       } else
1004 +               ret = NULL;
1005 +
1006 +       /* Pick up the branches, linking them together. */
1007 +       br = regbranch(g, &flags);
1008 +       if (br == NULL)
1009 +               return(NULL);
1010 +       if (ret != NULL)
1011 +               regtail(g, ret, br);    /* OPEN -> first. */
1012 +       else
1013 +               ret = br;
1014 +       if (!(flags&HASWIDTH))
1015 +               *flagp &= ~HASWIDTH;
1016 +       *flagp |= flags&SPSTART;
1017 +       while (*g->regparse == '|') {
1018 +               g->regparse++;
1019 +               br = regbranch(g, &flags);
1020 +               if (br == NULL)
1021 +                       return(NULL);
1022 +               regtail(g, ret, br);    /* BRANCH -> BRANCH. */
1023 +               if (!(flags&HASWIDTH))
1024 +                       *flagp &= ~HASWIDTH;
1025 +               *flagp |= flags&SPSTART;
1026 +       }
1027 +
1028 +       /* Make a closing node, and hook it on the end. */
1029 +       ender = regnode(g, (paren) ? CLOSE+parno : END);
1030 +       regtail(g, ret, ender);
1031 +
1032 +       /* Hook the tails of the branches to the closing node. */
1033 +       for (br = ret; br != NULL; br = regnext(g, br))
1034 +               regoptail(g, br, ender);
1035 +
1036 +       /* Check for proper termination. */
1037 +       if (paren && *g->regparse++ != ')') {
1038 +               FAIL("unmatched ()");
1039 +       } else if (!paren && *g->regparse != '\0') {
1040 +               if (*g->regparse == ')') {
1041 +                       FAIL("unmatched ()");
1042 +               } else
1043 +                       FAIL("junk on end");    /* "Can't happen". */
1044 +               /* NOTREACHED */
1045 +       }
1046 +
1047 +       return(ret);
1048 +}
1049 +
1050 +/*
1051 + - regbranch - one alternative of an | operator
1052 + *
1053 + * Implements the concatenation operator.
1054 + */
1055 +static char *
1056 +regbranch(struct match_globals *g, int *flagp)
1057 +{
1058 +       register char *ret;
1059 +       register char *chain;
1060 +       register char *latest;
1061 +       int flags;
1062 +
1063 +       *flagp = WORST;         /* Tentatively. */
1064 +
1065 +       ret = regnode(g, BRANCH);
1066 +       chain = NULL;
1067 +       while (*g->regparse != '\0' && *g->regparse != '|' && *g->regparse != ')') {
1068 +               latest = regpiece(g, &flags);
1069 +               if (latest == NULL)
1070 +                       return(NULL);
1071 +               *flagp |= flags&HASWIDTH;
1072 +               if (chain == NULL)      /* First piece. */
1073 +                       *flagp |= flags&SPSTART;
1074 +               else
1075 +                       regtail(g, chain, latest);
1076 +               chain = latest;
1077 +       }
1078 +       if (chain == NULL)      /* Loop ran zero times. */
1079 +               (void) regnode(g, NOTHING);
1080 +
1081 +       return(ret);
1082 +}
1083 +
1084 +/*
1085 + - regpiece - something followed by possible [*+?]
1086 + *
1087 + * Note that the branching code sequences used for ? and the general cases
1088 + * of * and + are somewhat optimized:  they use the same NOTHING node as
1089 + * both the endmarker for their branch list and the body of the last branch.
1090 + * It might seem that this node could be dispensed with entirely, but the
1091 + * endmarker role is not redundant.
1092 + */
1093 +static char *
1094 +regpiece(struct match_globals *g, int *flagp)
1095 +{
1096 +       register char *ret;
1097 +       register char op;
1098 +       register char *next;
1099 +       int flags;
1100 +
1101 +       ret = regatom(g, &flags);
1102 +       if (ret == NULL)
1103 +               return(NULL);
1104 +
1105 +       op = *g->regparse;
1106 +       if (!ISMULT(op)) {
1107 +               *flagp = flags;
1108 +               return(ret);
1109 +       }
1110 +
1111 +       if (!(flags&HASWIDTH) && op != '?')
1112 +               FAIL("*+ operand could be empty");
1113 +       *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
1114 +
1115 +       if (op == '*' && (flags&SIMPLE))
1116 +               reginsert(g, STAR, ret);
1117 +       else if (op == '*') {
1118 +               /* Emit x* as (x&|), where & means "self". */
1119 +               reginsert(g, BRANCH, ret);                      /* Either x */
1120 +               regoptail(g, ret, regnode(g, BACK));            /* and loop */
1121 +               regoptail(g, ret, ret);                 /* back */
1122 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
1123 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
1124 +       } else if (op == '+' && (flags&SIMPLE))
1125 +               reginsert(g, PLUS, ret);
1126 +       else if (op == '+') {
1127 +               /* Emit x+ as x(&|), where & means "self". */
1128 +               next = regnode(g, BRANCH);                      /* Either */
1129 +               regtail(g, ret, next);
1130 +               regtail(g, regnode(g, BACK), ret);              /* loop back */
1131 +               regtail(g, next, regnode(g, BRANCH));           /* or */
1132 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
1133 +       } else if (op == '?') {
1134 +               /* Emit x? as (x|) */
1135 +               reginsert(g, BRANCH, ret);                      /* Either x */
1136 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
1137 +               next = regnode(g, NOTHING);             /* null. */
1138 +               regtail(g, ret, next);
1139 +               regoptail(g, ret, next);
1140 +       }
1141 +       g->regparse++;
1142 +       if (ISMULT(*g->regparse))
1143 +               FAIL("nested *?+");
1144 +
1145 +       return(ret);
1146 +}
1147 +
1148 +/*
1149 + - regatom - the lowest level
1150 + *
1151 + * Optimization:  gobbles an entire sequence of ordinary characters so that
1152 + * it can turn them into a single node, which is smaller to store and
1153 + * faster to run.  Backslashed characters are exceptions, each becoming a
1154 + * separate node; the code is simpler that way and it's not worth fixing.
1155 + */
1156 +static char *
1157 +regatom(struct match_globals *g, int *flagp)
1158 +{
1159 +       register char *ret;
1160 +       int flags;
1161 +
1162 +       *flagp = WORST;         /* Tentatively. */
1163 +
1164 +       switch (*g->regparse++) {
1165 +       case '^':
1166 +               ret = regnode(g, BOL);
1167 +               break;
1168 +       case '$':
1169 +               ret = regnode(g, EOL);
1170 +               break;
1171 +       case '.':
1172 +               ret = regnode(g, ANY);
1173 +               *flagp |= HASWIDTH|SIMPLE;
1174 +               break;
1175 +       case '[': {
1176 +                       register int class;
1177 +                       register int classend;
1178 +
1179 +                       if (*g->regparse == '^') {      /* Complement of range. */
1180 +                               ret = regnode(g, ANYBUT);
1181 +                               g->regparse++;
1182 +                       } else
1183 +                               ret = regnode(g, ANYOF);
1184 +                       if (*g->regparse == ']' || *g->regparse == '-')
1185 +                               regc(g, *g->regparse++);
1186 +                       while (*g->regparse != '\0' && *g->regparse != ']') {
1187 +                               if (*g->regparse == '-') {
1188 +                                       g->regparse++;
1189 +                                       if (*g->regparse == ']' || *g->regparse == '\0')
1190 +                                               regc(g, '-');
1191 +                                       else {
1192 +                                               class = UCHARAT(g->regparse-2)+1;
1193 +                                               classend = UCHARAT(g->regparse);
1194 +                                               if (class > classend+1)
1195 +                                                       FAIL("invalid [] range");
1196 +                                               for (; class <= classend; class++)
1197 +                                                       regc(g, class);
1198 +                                               g->regparse++;
1199 +                                       }
1200 +                               } else
1201 +                                       regc(g, *g->regparse++);
1202 +                       }
1203 +                       regc(g, '\0');
1204 +                       if (*g->regparse != ']')
1205 +                               FAIL("unmatched []");
1206 +                       g->regparse++;
1207 +                       *flagp |= HASWIDTH|SIMPLE;
1208 +               }
1209 +               break;
1210 +       case '(':
1211 +               ret = reg(g, 1, &flags);
1212 +               if (ret == NULL)
1213 +                       return(NULL);
1214 +               *flagp |= flags&(HASWIDTH|SPSTART);
1215 +               break;
1216 +       case '\0':
1217 +       case '|':
1218 +       case ')':
1219 +               FAIL("internal urp");   /* Supposed to be caught earlier. */
1220 +               break;
1221 +       case '?':
1222 +       case '+':
1223 +       case '*':
1224 +               FAIL("?+* follows nothing");
1225 +               break;
1226 +       case '\\':
1227 +               if (*g->regparse == '\0')
1228 +                       FAIL("trailing \\");
1229 +               ret = regnode(g, EXACTLY);
1230 +               regc(g, *g->regparse++);
1231 +               regc(g, '\0');
1232 +               *flagp |= HASWIDTH|SIMPLE;
1233 +               break;
1234 +       default: {
1235 +                       register int len;
1236 +                       register char ender;
1237 +
1238 +                       g->regparse--;
1239 +                       len = my_strcspn((const char *)g->regparse, (const char *)META);
1240 +                       if (len <= 0)
1241 +                               FAIL("internal disaster");
1242 +                       ender = *(g->regparse+len);
1243 +                       if (len > 1 && ISMULT(ender))
1244 +                               len--;          /* Back off clear of ?+* operand. */
1245 +                       *flagp |= HASWIDTH;
1246 +                       if (len == 1)
1247 +                               *flagp |= SIMPLE;
1248 +                       ret = regnode(g, EXACTLY);
1249 +                       while (len > 0) {
1250 +                               regc(g, *g->regparse++);
1251 +                               len--;
1252 +                       }
1253 +                       regc(g, '\0');
1254 +               }
1255 +               break;
1256 +       }
1257 +
1258 +       return(ret);
1259 +}
1260 +
1261 +/*
1262 + - regnode - emit a node
1263 + */
1264 +static char *                  /* Location. */
1265 +regnode(struct match_globals *g, char op)
1266 +{
1267 +       register char *ret;
1268 +       register char *ptr;
1269 +
1270 +       ret = g->regcode;
1271 +       if (ret == &g->regdummy) {
1272 +               g->regsize += 3;
1273 +               return(ret);
1274 +       }
1275 +
1276 +       ptr = ret;
1277 +       *ptr++ = op;
1278 +       *ptr++ = '\0';          /* Null "next" pointer. */
1279 +       *ptr++ = '\0';
1280 +       g->regcode = ptr;
1281 +
1282 +       return(ret);
1283 +}
1284 +
1285 +/*
1286 + - regc - emit (if appropriate) a byte of code
1287 + */
1288 +static void
1289 +regc(struct match_globals *g, char b)
1290 +{
1291 +       if (g->regcode != &g->regdummy)
1292 +               *g->regcode++ = b;
1293 +       else
1294 +               g->regsize++;
1295 +}
1296 +
1297 +/*
1298 + - reginsert - insert an operator in front of already-emitted operand
1299 + *
1300 + * Means relocating the operand.
1301 + */
1302 +static void
1303 +reginsert(struct match_globals *g, char op, char* opnd)
1304 +{
1305 +       register char *src;
1306 +       register char *dst;
1307 +       register char *place;
1308 +
1309 +       if (g->regcode == &g->regdummy) {
1310 +               g->regsize += 3;
1311 +               return;
1312 +       }
1313 +
1314 +       src = g->regcode;
1315 +       g->regcode += 3;
1316 +       dst = g->regcode;
1317 +       while (src > opnd)
1318 +               *--dst = *--src;
1319 +
1320 +       place = opnd;           /* Op node, where operand used to be. */
1321 +       *place++ = op;
1322 +       *place++ = '\0';
1323 +       *place++ = '\0';
1324 +}
1325 +
1326 +/*
1327 + - regtail - set the next-pointer at the end of a node chain
1328 + */
1329 +static void
1330 +regtail(struct match_globals *g, char *p, char *val)
1331 +{
1332 +       register char *scan;
1333 +       register char *temp;
1334 +       register int offset;
1335 +
1336 +       if (p == &g->regdummy)
1337 +               return;
1338 +
1339 +       /* Find last node. */
1340 +       scan = p;
1341 +       for (;;) {
1342 +               temp = regnext(g, scan);
1343 +               if (temp == NULL)
1344 +                       break;
1345 +               scan = temp;
1346 +       }
1347 +
1348 +       if (OP(scan) == BACK)
1349 +               offset = scan - val;
1350 +       else
1351 +               offset = val - scan;
1352 +       *(scan+1) = (offset>>8)&0377;
1353 +       *(scan+2) = offset&0377;
1354 +}
1355 +
1356 +/*
1357 + - regoptail - regtail on operand of first argument; nop if operandless
1358 + */
1359 +static void
1360 +regoptail(struct match_globals *g, char *p, char *val)
1361 +{
1362 +       /* "Operandless" and "op != BRANCH" are synonymous in practice. */
1363 +       if (p == NULL || p == &g->regdummy || OP(p) != BRANCH)
1364 +               return;
1365 +       regtail(g, OPERAND(p), val);
1366 +}
1367 +
1368 +/*
1369 + * regexec and friends
1370 + */
1371 +
1372 +
1373 +/*
1374 + * Forwards.
1375 + */
1376 +STATIC int regtry(struct match_globals *g, regexp *prog, char *string);
1377 +STATIC int regmatch(struct match_globals *g, char *prog);
1378 +STATIC int regrepeat(struct match_globals *g, char *p);
1379 +
1380 +#ifdef DEBUG
1381 +int regnarrate = 0;
1382 +void regdump();
1383 +STATIC char *regprop(char *op);
1384 +#endif
1385 +
1386 +/*
1387 + - regexec - match a regexp against a string
1388 + */
1389 +int
1390 +regexec(regexp *prog, char *string)
1391 +{
1392 +       register char *s;
1393 +       struct match_globals g;
1394 +
1395 +       /* Be paranoid... */
1396 +       if (prog == NULL || string == NULL) {
1397 +               printk("<3>Regexp: NULL parameter\n");
1398 +               return(0);
1399 +       }
1400 +
1401 +       /* Check validity of program. */
1402 +       if (UCHARAT(prog->program) != MAGIC) {
1403 +               printk("<3>Regexp: corrupted program\n");
1404 +               return(0);
1405 +       }
1406 +
1407 +       /* If there is a "must appear" string, look for it. */
1408 +       if (prog->regmust != NULL) {
1409 +               s = string;
1410 +               while ((s = strchr(s, prog->regmust[0])) != NULL) {
1411 +                       if (strncmp(s, prog->regmust, prog->regmlen) == 0)
1412 +                               break;  /* Found it. */
1413 +                       s++;
1414 +               }
1415 +               if (s == NULL)  /* Not present. */
1416 +                       return(0);
1417 +       }
1418 +
1419 +       /* Mark beginning of line for ^ . */
1420 +       g.regbol = string;
1421 +
1422 +       /* Simplest case:  anchored match need be tried only once. */
1423 +       if (prog->reganch)
1424 +               return(regtry(&g, prog, string));
1425 +
1426 +       /* Messy cases:  unanchored match. */
1427 +       s = string;
1428 +       if (prog->regstart != '\0')
1429 +               /* We know what char it must start with. */
1430 +               while ((s = strchr(s, prog->regstart)) != NULL) {
1431 +                       if (regtry(&g, prog, s))
1432 +                               return(1);
1433 +                       s++;
1434 +               }
1435 +       else
1436 +               /* We don't -- general case. */
1437 +               do {
1438 +                       if (regtry(&g, prog, s))
1439 +                               return(1);
1440 +               } while (*s++ != '\0');
1441 +
1442 +       /* Failure. */
1443 +       return(0);
1444 +}
1445 +
1446 +/*
1447 + - regtry - try match at specific point
1448 + */
1449 +static int                     /* 0 failure, 1 success */
1450 +regtry(struct match_globals *g, regexp *prog, char *string)
1451 +{
1452 +       register int i;
1453 +       register char **sp;
1454 +       register char **ep;
1455 +
1456 +       g->reginput = string;
1457 +       g->regstartp = prog->startp;
1458 +       g->regendp = prog->endp;
1459 +
1460 +       sp = prog->startp;
1461 +       ep = prog->endp;
1462 +       for (i = NSUBEXP; i > 0; i--) {
1463 +               *sp++ = NULL;
1464 +               *ep++ = NULL;
1465 +       }
1466 +       if (regmatch(g, prog->program + 1)) {
1467 +               prog->startp[0] = string;
1468 +               prog->endp[0] = g->reginput;
1469 +               return(1);
1470 +       } else
1471 +               return(0);
1472 +}
1473 +
1474 +/*
1475 + - regmatch - main matching routine
1476 + *
1477 + * Conceptually the strategy is simple:  check to see whether the current
1478 + * node matches, call self recursively to see whether the rest matches,
1479 + * and then act accordingly.  In practice we make some effort to avoid
1480 + * recursion, in particular by going through "ordinary" nodes (that don't
1481 + * need to know whether the rest of the match failed) by a loop instead of
1482 + * by recursion.
1483 + */
1484 +static int                     /* 0 failure, 1 success */
1485 +regmatch(struct match_globals *g, char *prog)
1486 +{
1487 +       register char *scan = prog; /* Current node. */
1488 +       char *next;                 /* Next node. */
1489 +
1490 +#ifdef DEBUG
1491 +       if (scan != NULL && regnarrate)
1492 +               fprintf(stderr, "%s(\n", regprop(scan));
1493 +#endif
1494 +       while (scan != NULL) {
1495 +#ifdef DEBUG
1496 +               if (regnarrate)
1497 +                       fprintf(stderr, "%s...\n", regprop(scan));
1498 +#endif
1499 +               next = regnext(g, scan);
1500 +
1501 +               switch (OP(scan)) {
1502 +               case BOL:
1503 +                       if (g->reginput != g->regbol)
1504 +                               return(0);
1505 +                       break;
1506 +               case EOL:
1507 +                       if (*g->reginput != '\0')
1508 +                               return(0);
1509 +                       break;
1510 +               case ANY:
1511 +                       if (*g->reginput == '\0')
1512 +                               return(0);
1513 +                       g->reginput++;
1514 +                       break;
1515 +               case EXACTLY: {
1516 +                               register int len;
1517 +                               register char *opnd;
1518 +
1519 +                               opnd = OPERAND(scan);
1520 +                               /* Inline the first character, for speed. */
1521 +                               if (*opnd != *g->reginput)
1522 +                                       return(0);
1523 +                               len = strlen(opnd);
1524 +                               if (len > 1 && strncmp(opnd, g->reginput, len) != 0)
1525 +                                       return(0);
1526 +                               g->reginput += len;
1527 +                       }
1528 +                       break;
1529 +               case ANYOF:
1530 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) == NULL)
1531 +                               return(0);
1532 +                       g->reginput++;
1533 +                       break;
1534 +               case ANYBUT:
1535 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) != NULL)
1536 +                               return(0);
1537 +                       g->reginput++;
1538 +                       break;
1539 +               case NOTHING:
1540 +               case BACK:
1541 +                       break;
1542 +               case OPEN+1:
1543 +               case OPEN+2:
1544 +               case OPEN+3:
1545 +               case OPEN+4:
1546 +               case OPEN+5:
1547 +               case OPEN+6:
1548 +               case OPEN+7:
1549 +               case OPEN+8:
1550 +               case OPEN+9: {
1551 +                               register int no;
1552 +                               register char *save;
1553 +
1554 +                               no = OP(scan) - OPEN;
1555 +                               save = g->reginput;
1556 +
1557 +                               if (regmatch(g, next)) {
1558 +                                       /*
1559 +                                        * Don't set startp if some later
1560 +                                        * invocation of the same parentheses
1561 +                                        * already has.
1562 +                                        */
1563 +                                       if (g->regstartp[no] == NULL)
1564 +                                               g->regstartp[no] = save;
1565 +                                       return(1);
1566 +                               } else
1567 +                                       return(0);
1568 +                       }
1569 +                       break;
1570 +               case CLOSE+1:
1571 +               case CLOSE+2:
1572 +               case CLOSE+3:
1573 +               case CLOSE+4:
1574 +               case CLOSE+5:
1575 +               case CLOSE+6:
1576 +               case CLOSE+7:
1577 +               case CLOSE+8:
1578 +               case CLOSE+9:
1579 +                       {
1580 +                               register int no;
1581 +                               register char *save;
1582 +
1583 +                               no = OP(scan) - CLOSE;
1584 +                               save = g->reginput;
1585 +
1586 +                               if (regmatch(g, next)) {
1587 +                                       /*
1588 +                                        * Don't set endp if some later
1589 +                                        * invocation of the same parentheses
1590 +                                        * already has.
1591 +                                        */
1592 +                                       if (g->regendp[no] == NULL)
1593 +                                               g->regendp[no] = save;
1594 +                                       return(1);
1595 +                               } else
1596 +                                       return(0);
1597 +                       }
1598 +                       break;
1599 +               case BRANCH: {
1600 +                               register char *save;
1601 +
1602 +                               if (OP(next) != BRANCH)         /* No choice. */
1603 +                                       next = OPERAND(scan);   /* Avoid recursion. */
1604 +                               else {
1605 +                                       do {
1606 +                                               save = g->reginput;
1607 +                                               if (regmatch(g, OPERAND(scan)))
1608 +                                                       return(1);
1609 +                                               g->reginput = save;
1610 +                                               scan = regnext(g, scan);
1611 +                                       } while (scan != NULL && OP(scan) == BRANCH);
1612 +                                       return(0);
1613 +                                       /* NOTREACHED */
1614 +                               }
1615 +                       }
1616 +                       break;
1617 +               case STAR:
1618 +               case PLUS: {
1619 +                               register char nextch;
1620 +                               register int no;
1621 +                               register char *save;
1622 +                               register int min;
1623 +
1624 +                               /*
1625 +                                * Lookahead to avoid useless match attempts
1626 +                                * when we know what character comes next.
1627 +                                */
1628 +                               nextch = '\0';
1629 +                               if (OP(next) == EXACTLY)
1630 +                                       nextch = *OPERAND(next);
1631 +                               min = (OP(scan) == STAR) ? 0 : 1;
1632 +                               save = g->reginput;
1633 +                               no = regrepeat(g, OPERAND(scan));
1634 +                               while (no >= min) {
1635 +                                       /* If it could work, try it. */
1636 +                                       if (nextch == '\0' || *g->reginput == nextch)
1637 +                                               if (regmatch(g, next))
1638 +                                                       return(1);
1639 +                                       /* Couldn't or didn't -- back up. */
1640 +                                       no--;
1641 +                                       g->reginput = save + no;
1642 +                               }
1643 +                               return(0);
1644 +                       }
1645 +                       break;
1646 +               case END:
1647 +                       return(1);      /* Success! */
1648 +                       break;
1649 +               default:
1650 +                       printk("<3>Regexp: memory corruption\n");
1651 +                       return(0);
1652 +                       break;
1653 +               }
1654 +
1655 +               scan = next;
1656 +       }
1657 +
1658 +       /*
1659 +        * We get here only if there's trouble -- normally "case END" is
1660 +        * the terminating point.
1661 +        */
1662 +       printk("<3>Regexp: corrupted pointers\n");
1663 +       return(0);
1664 +}
1665 +
1666 +/*
1667 + - regrepeat - repeatedly match something simple, report how many
1668 + */
1669 +static int
1670 +regrepeat(struct match_globals *g, char *p)
1671 +{
1672 +       register int count = 0;
1673 +       register char *scan;
1674 +       register char *opnd;
1675 +
1676 +       scan = g->reginput;
1677 +       opnd = OPERAND(p);
1678 +       switch (OP(p)) {
1679 +       case ANY:
1680 +               count = strlen(scan);
1681 +               scan += count;
1682 +               break;
1683 +       case EXACTLY:
1684 +               while (*opnd == *scan) {
1685 +                       count++;
1686 +                       scan++;
1687 +               }
1688 +               break;
1689 +       case ANYOF:
1690 +               while (*scan != '\0' && strchr(opnd, *scan) != NULL) {
1691 +                       count++;
1692 +                       scan++;
1693 +               }
1694 +               break;
1695 +       case ANYBUT:
1696 +               while (*scan != '\0' && strchr(opnd, *scan) == NULL) {
1697 +                       count++;
1698 +                       scan++;
1699 +               }
1700 +               break;
1701 +       default:                /* Oh dear.  Called inappropriately. */
1702 +               printk("<3>Regexp: internal foulup\n");
1703 +               count = 0;      /* Best compromise. */
1704 +               break;
1705 +       }
1706 +       g->reginput = scan;
1707 +
1708 +       return(count);
1709 +}
1710 +
1711 +/*
1712 + - regnext - dig the "next" pointer out of a node
1713 + */
1714 +static char*
1715 +regnext(struct match_globals *g, char *p)
1716 +{
1717 +       register int offset;
1718 +
1719 +       if (p == &g->regdummy)
1720 +               return(NULL);
1721 +
1722 +       offset = NEXT(p);
1723 +       if (offset == 0)
1724 +               return(NULL);
1725 +
1726 +       if (OP(p) == BACK)
1727 +               return(p-offset);
1728 +       else
1729 +               return(p+offset);
1730 +}
1731 +
1732 +#ifdef DEBUG
1733 +
1734 +STATIC char *regprop();
1735 +
1736 +/*
1737 + - regdump - dump a regexp onto stdout in vaguely comprehensible form
1738 + */
1739 +void
1740 +regdump(regexp *r)
1741 +{
1742 +       register char *s;
1743 +       register char op = EXACTLY;     /* Arbitrary non-END op. */
1744 +       register char *next;
1745 +       /* extern char *strchr(); */
1746 +
1747 +
1748 +       s = r->program + 1;
1749 +       while (op != END) {     /* While that wasn't END last time... */
1750 +               op = OP(s);
1751 +               printf("%2d%s", s-r->program, regprop(s));      /* Where, what. */
1752 +               next = regnext(s);
1753 +               if (next == NULL)               /* Next ptr. */
1754 +                       printf("(0)");
1755 +               else
1756 +                       printf("(%d)", (s-r->program)+(next-s));
1757 +               s += 3;
1758 +               if (op == ANYOF || op == ANYBUT || op == EXACTLY) {
1759 +                       /* Literal string, where present. */
1760 +                       while (*s != '\0') {
1761 +                               putchar(*s);
1762 +                               s++;
1763 +                       }
1764 +                       s++;
1765 +               }
1766 +               putchar('\n');
1767 +       }
1768 +
1769 +       /* Header fields of interest. */
1770 +       if (r->regstart != '\0')
1771 +               printf("start `%c' ", r->regstart);
1772 +       if (r->reganch)
1773 +               printf("anchored ");
1774 +       if (r->regmust != NULL)
1775 +               printf("must have \"%s\"", r->regmust);
1776 +       printf("\n");
1777 +}
1778 +
1779 +/*
1780 + - regprop - printable representation of opcode
1781 + */
1782 +static char *
1783 +regprop(char *op)
1784 +{
1785 +#define BUFLEN 50
1786 +       register char *p;
1787 +       static char buf[BUFLEN];
1788 +
1789 +       strcpy(buf, ":");
1790 +
1791 +       switch (OP(op)) {
1792 +       case BOL:
1793 +               p = "BOL";
1794 +               break;
1795 +       case EOL:
1796 +               p = "EOL";
1797 +               break;
1798 +       case ANY:
1799 +               p = "ANY";
1800 +               break;
1801 +       case ANYOF:
1802 +               p = "ANYOF";
1803 +               break;
1804 +       case ANYBUT:
1805 +               p = "ANYBUT";
1806 +               break;
1807 +       case BRANCH:
1808 +               p = "BRANCH";
1809 +               break;
1810 +       case EXACTLY:
1811 +               p = "EXACTLY";
1812 +               break;
1813 +       case NOTHING:
1814 +               p = "NOTHING";
1815 +               break;
1816 +       case BACK:
1817 +               p = "BACK";
1818 +               break;
1819 +       case END:
1820 +               p = "END";
1821 +               break;
1822 +       case OPEN+1:
1823 +       case OPEN+2:
1824 +       case OPEN+3:
1825 +       case OPEN+4:
1826 +       case OPEN+5:
1827 +       case OPEN+6:
1828 +       case OPEN+7:
1829 +       case OPEN+8:
1830 +       case OPEN+9:
1831 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "OPEN%d", OP(op)-OPEN);
1832 +               p = NULL;
1833 +               break;
1834 +       case CLOSE+1:
1835 +       case CLOSE+2:
1836 +       case CLOSE+3:
1837 +       case CLOSE+4:
1838 +       case CLOSE+5:
1839 +       case CLOSE+6:
1840 +       case CLOSE+7:
1841 +       case CLOSE+8:
1842 +       case CLOSE+9:
1843 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "CLOSE%d", OP(op)-CLOSE);
1844 +               p = NULL;
1845 +               break;
1846 +       case STAR:
1847 +               p = "STAR";
1848 +               break;
1849 +       case PLUS:
1850 +               p = "PLUS";
1851 +               break;
1852 +       default:
1853 +               printk("<3>Regexp: corrupted opcode\n");
1854 +               break;
1855 +       }
1856 +       if (p != NULL)
1857 +               strncat(buf, p, BUFLEN-strlen(buf));
1858 +       return(buf);
1859 +}
1860 +#endif
1861 +
1862 +
1863 --- linux-2.6.22-rc7/net/netfilter/regexp/regexp.h      1969-12-31 18:00:00.000000000 -0600
1864 +++ linux-2.6.22-rc7-layer7/net/netfilter/regexp/regexp.h       2007-07-03 01:32:35.000000000 -0500
1865 @@ -0,0 +1,41 @@
1866 +/*
1867 + * Definitions etc. for regexp(3) routines.
1868 + *
1869 + * Caveat:  this is V8 regexp(3) [actually, a reimplementation thereof],
1870 + * not the System V one.
1871 + */
1872 +
1873 +#ifndef REGEXP_H
1874 +#define REGEXP_H
1875 +
1876 +
1877 +/*
1878 +http://www.opensource.apple.com/darwinsource/10.3/expect-1/expect/expect.h ,
1879 +which contains a version of this library, says:
1880 +
1881 + *
1882 + * NSUBEXP must be at least 10, and no greater than 117 or the parser
1883 + * will not work properly.
1884 + *
1885 +
1886 +However, it looks rather like this library is limited to 10.  If you think
1887 +otherwise, let us know.
1888 +*/
1889 +
1890 +#define NSUBEXP  10
1891 +typedef struct regexp {
1892 +       char *startp[NSUBEXP];
1893 +       char *endp[NSUBEXP];
1894 +       char regstart;          /* Internal use only. */
1895 +       char reganch;           /* Internal use only. */
1896 +       char *regmust;          /* Internal use only. */
1897 +       int regmlen;            /* Internal use only. */
1898 +       char program[1];        /* Unwarranted chumminess with compiler. */
1899 +} regexp;
1900 +
1901 +regexp * regcomp(char *exp, int *patternsize);
1902 +int regexec(regexp *prog, char *string);
1903 +void regsub(regexp *prog, char *source, char *dest);
1904 +void regerror(char *s);
1905 +
1906 +#endif
1907 --- linux-2.6.22-rc7/net/netfilter/regexp/regmagic.h    1969-12-31 18:00:00.000000000 -0600
1908 +++ linux-2.6.22-rc7-layer7/net/netfilter/regexp/regmagic.h     2007-07-03 01:32:35.000000000 -0500
1909 @@ -0,0 +1,5 @@
1910 +/*
1911 + * The first byte of the regexp internal "program" is actually this magic
1912 + * number; the start node begins in the second byte.
1913 + */
1914 +#define        MAGIC   0234
1915 --- linux-2.6.22-rc7/net/netfilter/regexp/regsub.c      1969-12-31 18:00:00.000000000 -0600
1916 +++ linux-2.6.22-rc7-layer7/net/netfilter/regexp/regsub.c       2007-07-03 01:32:35.000000000 -0500
1917 @@ -0,0 +1,95 @@
1918 +/*
1919 + * regsub
1920 + * @(#)regsub.c        1.3 of 2 April 86
1921 + *
1922 + *     Copyright (c) 1986 by University of Toronto.
1923 + *     Written by Henry Spencer.  Not derived from licensed software.
1924 + *
1925 + *     Permission is granted to anyone to use this software for any
1926 + *     purpose on any computer system, and to redistribute it freely,
1927 + *     subject to the following restrictions:
1928 + *
1929 + *     1. The author is not responsible for the consequences of use of
1930 + *             this software, no matter how awful, even if they arise
1931 + *             from defects in it.
1932 + *
1933 + *     2. The origin of this software must not be misrepresented, either
1934 + *             by explicit claim or by omission.
1935 + *
1936 + *     3. Altered versions must be plainly marked as such, and must not
1937 + *             be misrepresented as being the original software.
1938 + *
1939 + *
1940 + * This code was modified by Ethan Sommer to work within the kernel
1941 + * (it now uses kmalloc etc..)
1942 + *
1943 + */
1944 +#include "regexp.h"
1945 +#include "regmagic.h"
1946 +#include <linux/string.h>
1947 +
1948 +
1949 +#ifndef CHARBITS
1950 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
1951 +#else
1952 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
1953 +#endif
1954 +
1955 +#if 0
1956 +//void regerror(char * s)
1957 +//{
1958 +//        printk("regexp(3): %s", s);
1959 +//        /* NOTREACHED */
1960 +//}
1961 +#endif
1962 +
1963 +/*
1964 + - regsub - perform substitutions after a regexp match
1965 + */
1966 +void
1967 +regsub(regexp * prog, char * source, char * dest)
1968 +{
1969 +       register char *src;
1970 +       register char *dst;
1971 +       register char c;
1972 +       register int no;
1973 +       register int len;
1974 +
1975 +       /* Not necessary and gcc doesn't like it -MLS */
1976 +       /*extern char *strncpy();*/
1977 +
1978 +       if (prog == NULL || source == NULL || dest == NULL) {
1979 +               regerror("NULL parm to regsub");
1980 +               return;
1981 +       }
1982 +       if (UCHARAT(prog->program) != MAGIC) {
1983 +               regerror("damaged regexp fed to regsub");
1984 +               return;
1985 +       }
1986 +
1987 +       src = source;
1988 +       dst = dest;
1989 +       while ((c = *src++) != '\0') {
1990 +               if (c == '&')
1991 +                       no = 0;
1992 +               else if (c == '\\' && '0' <= *src && *src <= '9')
1993 +                       no = *src++ - '0';
1994 +               else
1995 +                       no = -1;
1996 +
1997 +               if (no < 0) {   /* Ordinary character. */
1998 +                       if (c == '\\' && (*src == '\\' || *src == '&'))
1999 +                               c = *src++;
2000 +                       *dst++ = c;
2001 +               } else if (prog->startp[no] != NULL && prog->endp[no] != NULL) {
2002 +                       len = prog->endp[no] - prog->startp[no];
2003 +                       (void) strncpy(dst, prog->startp[no], len);
2004 +                       dst += len;
2005 +                       if (len != 0 && *(dst-1) == '\0') {     /* strncpy hit NUL. */
2006 +                               regerror("damaged match string");
2007 +                               return;
2008 +                       }
2009 +               }
2010 +       }
2011 +       *dst++ = '\0';
2012 +}
2013 --- linux-2.6.22-rc7/net/netfilter/nf_conntrack_core.c  2007-07-01 14:54:24.000000000 -0500
2014 +++ linux-2.6.22-rc7-layer7/net/netfilter/nf_conntrack_core.c   2007-07-03 01:32:35.000000000 -0500
2015 @@ -330,6 +330,14 @@ destroy_conntrack(struct nf_conntrack *n
2016          * too. */
2017         nf_ct_remove_expectations(ct);
2018
2019 +       #if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
2020 +       if(ct->layer7.app_proto)
2021 +               kfree(ct->layer7.app_proto);
2022 +       if(ct->layer7.app_data)
2023 +       kfree(ct->layer7.app_data);
2024 +       #endif
2025 +
2026 +
2027         /* We overload first tuple to link into unconfirmed list. */
2028         if (!nf_ct_is_confirmed(ct)) {
2029                 BUG_ON(list_empty(&ct->tuplehash[IP_CT_DIR_ORIGINAL].list));
2030 --- linux-2.6.22-rc7/net/netfilter/nf_conntrack_standalone.c    2007-07-01 14:54:24.000000000 -0500
2031 +++ linux-2.6.22-rc7-layer7/net/netfilter/nf_conntrack_standalone.c     2007-07-03 01:32:35.000000000 -0500
2032 @@ -184,6 +184,11 @@ static int ct_seq_show(struct seq_file *
2033                 return -ENOSPC;
2034  #endif
2035
2036 +#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
2037 +       if(conntrack->layer7.app_proto)
2038 +               if(seq_printf(s, "l7proto=%s ",conntrack->layer7.app_proto))
2039 +                       return -ENOSPC;
2040 +#endif
2041         if (seq_printf(s, "use=%u\n", atomic_read(&conntrack->ct_general.use)))
2042                 return -ENOSPC;
2043
2044 --- linux-2.6.22-rc7/include/net/netfilter/nf_conntrack.h       2007-07-01 14:54:24.000000000 -0500
2045 +++ linux-2.6.22-rc7-layer7/include/net/netfilter/nf_conntrack.h        2007-07-03 01:32:35.000000000 -0500
2046 @@ -128,6 +128,22 @@ struct nf_conn
2047         u_int32_t secmark;
2048  #endif
2049
2050 +#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || \
2051 +    defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
2052 +       struct {
2053 +               /*
2054 +                * e.g. "http". NULL before decision. "unknown" after decision
2055 +                * if no match.
2056 +                */
2057 +               char *app_proto;
2058 +               /*
2059 +                * application layer data so far. NULL after match decision.
2060 +                */
2061 +               char *app_data;
2062 +               unsigned int app_data_len;
2063 +       } layer7;
2064 +#endif
2065 +
2066         /* Storage reserved for other modules: */
2067         union nf_conntrack_proto proto;
2068
2069 --- linux-2.6.22-rc7/include/linux/netfilter/xt_layer7.h        1969-12-31 18:00:00.000000000 -0600
2070 +++ linux-2.6.22-rc7-layer7/include/linux/netfilter/xt_layer7.h 2007-07-03 07:48:19.000000000 -0500
2071 @@ -0,0 +1,13 @@
2072 +#ifndef _XT_LAYER7_H
2073 +#define _XT_LAYER7_H
2074 +
2075 +#define MAX_PATTERN_LEN 8192
2076 +#define MAX_PROTOCOL_LEN 256
2077 +
2078 +struct xt_layer7_info {
2079 +    char protocol[MAX_PROTOCOL_LEN];
2080 +    char pattern[MAX_PATTERN_LEN];
2081 +    u_int8_t invert;
2082 +};
2083 +
2084 +#endif /* _XT_LAYER7_H */