package/linux/kernel-patches/306-netfilter-layer7-0.9.1

   1 diff -Nurp linux-2.4.26-stock/Documentation/Configure.help linux-2.4.26-layer7-clean/Documentation/Configure.help
   2 --- linux-2.4.26-stock/Documentation/Configure.help     2004-04-14 08:05:24.000000000 -0500
   3 +++ linux-2.4.26-layer7-clean/Documentation/Configure.help      2004-06-21 00:18:14.000000000 -0500
   4 @@ -28819,6 +28819,23 @@ CONFIG_SOUND_WM97XX
   5
   6    If unsure, say N.
   7
   8 +CONFIG_IP_NF_MATCH_LAYER7
   9 +   Say Y if you want to be able to classify connections (and their
  10 +   packets) based on regular expression matching of their application
  11 +   layer data.   This is one way to classify applications such as
  12 +   peer-to-peer filesharing systems that do not always use the same
  13 +   port.
  14 +
  15 +   To compile it as a module, choose M here.  If unsure, say N.
  16 +
  17 +CONFIG_IP_NF_MATCH_LAYER7_DEBUG
  18 +   Say Y to get lots of debugging output.
  19 +
  20 +CONFIG_IP_NF_MATCH_LAYER7_MAXDATALEN
  21 +   Size of the buffer that the application layer data is stored in.
  22 +   Unless you know what you're doing, leave it at the default of 2048
  23 +   Bytes.
  24 +
  25  #
  26  # A couple of things I keep forgetting:
  27  #   capitalize: AppleTalk, Ethernet, DOS, DMA, FAT, FTP, Internet,
  28 diff -Nurp linux-2.4.26-stock/include/linux/netfilter_ipv4/ip_conntrack.h linux-2.4.26-layer7-clean/include/linux/netfilter_ipv4/ip_conntrack.h
  29 --- linux-2.4.26-stock/include/linux/netfilter_ipv4/ip_conntrack.h      2004-04-14 08:05:40.000000000 -0500
  30 +++ linux-2.4.26-layer7-clean/include/linux/netfilter_ipv4/ip_conntrack.h       2004-06-21 00:18:28.000000000 -0500
  31 @@ -207,6 +207,17 @@ struct ip_conntrack
  32         } nat;
  33  #endif /* CONFIG_IP_NF_NAT_NEEDED */
  34
  35 +#if defined(CONFIG_IP_NF_MATCH_LAYER7) || defined(CONFIG_IP_NF_MATCH_LAYER7_MODULE)
  36 +       struct {
  37 +               unsigned int numpackets; /* surely this is kept track of somewhere else, right? I can't find it... */
  38 +               char * app_proto; /* "http", "ftp", etc.  NULL if unclassifed */
  39 +
  40 +               /* the application layer data so far.  NULL if ->numpackets > numpackets */
  41 +               char * app_data;
  42 +
  43 +               unsigned int app_data_len;
  44 +       } layer7;
  45 +#endif
  46  };
  47
  48  /* get master conntrack via master expectation */
  49 diff -Nurp linux-2.4.26-stock/include/linux/netfilter_ipv4/ipt_layer7.h linux-2.4.26-layer7-clean/include/linux/netfilter_ipv4/ipt_layer7.h
  50 --- linux-2.4.26-stock/include/linux/netfilter_ipv4/ipt_layer7.h        1969-12-31 18:00:00.000000000 -0600
  51 +++ linux-2.4.26-layer7-clean/include/linux/netfilter_ipv4/ipt_layer7.h 2004-06-21 00:18:28.000000000 -0500
  52 @@ -0,0 +1,26 @@
  53 +/*
  54 +  By Matthew Strait <quadong@users.sf.net>, Dec 2003.
  55 +  http://l7-filter.sf.net
  56 +
  57 +  This program is free software; you can redistribute it and/or
  58 +  modify it under the terms of the GNU General Public License
  59 +  as published by the Free Software Foundation; either version
  60 +  2 of the License, or (at your option) any later version.
  61 +  http://www.gnu.org/licenses/gpl.txt
  62 +*/
  63 +
  64 +#ifndef _IPT_LAYER7_H
  65 +#define _IPT_LAYER7_H
  66 +
  67 +#define MAX_PATTERN_LEN 8192
  68 +#define MAX_PROTOCOL_LEN 256
  69 +
  70 +typedef char *(*proc_ipt_search) (char *, char, char *);
  71 +
  72 +struct ipt_layer7_info {
  73 +    char protocol[MAX_PROTOCOL_LEN];
  74 +    char invert:1;
  75 +    char pattern[MAX_PATTERN_LEN];
  76 +};
  77 +
  78 +#endif /* _IPT_LAYER7_H */
  79 diff -Nurp linux-2.4.26-stock/net/ipv4/netfilter/Config.in linux-2.4.26-layer7-clean/net/ipv4/netfilter/Config.in
  80 --- linux-2.4.26-stock/net/ipv4/netfilter/Config.in     2003-08-25 06:44:44.000000000 -0500
  81 +++ linux-2.4.26-layer7-clean/net/ipv4/netfilter/Config.in      2004-06-21 00:15:23.000000000 -0500
  82 @@ -43,6 +43,10 @@ if [ "$CONFIG_IP_NF_IPTABLES" != "n" ];
  83    if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
  84      dep_tristate '  Unclean match support (EXPERIMENTAL)' CONFIG_IP_NF_MATCH_UNCLEAN $CONFIG_IP_NF_IPTABLES
  85      dep_tristate '  Owner match support (EXPERIMENTAL)' CONFIG_IP_NF_MATCH_OWNER $CONFIG_IP_NF_IPTABLES
  86 +    dep_tristate '  Layer 7 match support (EXPERIMENTAL)' CONFIG_IP_NF_MATCH_LAYER7 $CONFIG_IP_NF_CONNTRACK
  87 +    dep_mbool '  Layer 7 debugging output (EXPERIMENTAL)' CONFIG_IP_NF_MATCH_LAYER7_DEBUG $CONFIG_IP_NF_MATCH_LAYER7
  88 +    int  '  Buffer size for application layer data (256-65536)' CONFIG_IP_NF_MATCH_LAYER7_MAXDATALEN 2048
  89 +
  90    fi
  91  # The targets
  92    dep_tristate '  Packet filtering' CONFIG_IP_NF_FILTER $CONFIG_IP_NF_IPTABLES
  93 diff -Nurp linux-2.4.26-stock/net/ipv4/netfilter/Makefile linux-2.4.26-layer7-clean/net/ipv4/netfilter/Makefile
  94 --- linux-2.4.26-stock/net/ipv4/netfilter/Makefile      2003-08-25 06:44:44.000000000 -0500
  95 +++ linux-2.4.26-layer7-clean/net/ipv4/netfilter/Makefile       2004-06-21 00:15:22.000000000 -0500
  96 @@ -87,6 +87,8 @@ obj-$(CONFIG_IP_NF_MATCH_CONNTRACK) += i
  97  obj-$(CONFIG_IP_NF_MATCH_UNCLEAN) += ipt_unclean.o
  98  obj-$(CONFIG_IP_NF_MATCH_TCPMSS) += ipt_tcpmss.o
  99
 100 +obj-$(CONFIG_IP_NF_MATCH_LAYER7) += ipt_layer7.o
 101 +
 102  # targets
 103  obj-$(CONFIG_IP_NF_TARGET_REJECT) += ipt_REJECT.o
 104  obj-$(CONFIG_IP_NF_TARGET_MIRROR) += ipt_MIRROR.o
 105 diff -Nurp linux-2.4.26-stock/net/ipv4/netfilter/ip_conntrack_core.c linux-2.4.26-layer7-clean/net/ipv4/netfilter/ip_conntrack_core.c
 106 --- linux-2.4.26-stock/net/ipv4/netfilter/ip_conntrack_core.c   2004-02-18 07:36:32.000000000 -0600
 107 +++ linux-2.4.26-layer7-clean/net/ipv4/netfilter/ip_conntrack_core.c    2004-06-21 00:15:22.000000000 -0500
 108 @@ -339,6 +339,15 @@ destroy_conntrack(struct nf_conntrack *n
 109                 }
 110                 kfree(ct->master);
 111         }
 112 +
 113 +       #if defined(CONFIG_IP_NF_MATCH_LAYER7) || defined(CONFIG_IP_NF_MATCH_LAYER7_MODULE)
 114 +       /* This ought to get free'd somewhere.  How about here? */
 115 +       if(ct->layer7.app_proto) /* this is sufficient, right? */
 116 +               kfree(ct->layer7.app_proto);
 117 +       if(ct->layer7.app_data)
 118 +               kfree(ct->layer7.app_data);
 119 +       #endif
 120 +
 121         WRITE_UNLOCK(&ip_conntrack_lock);
 122
 123         if (master)
 124 diff -Nurp linux-2.4.26-stock/net/ipv4/netfilter/ip_conntrack_standalone.c linux-2.4.26-layer7-clean/net/ipv4/netfilter/ip_conntrack_standalone.c
 125 --- linux-2.4.26-stock/net/ipv4/netfilter/ip_conntrack_standalone.c     2004-02-18 07:36:32.000000000 -0600
 126 +++ linux-2.4.26-layer7-clean/net/ipv4/netfilter/ip_conntrack_standalone.c      2004-06-21 00:15:22.000000000 -0500
 127 @@ -107,6 +107,13 @@ print_conntrack(char *buffer, struct ip_
 128                 len += sprintf(buffer + len, "[ASSURED] ");
 129         len += sprintf(buffer + len, "use=%u ",
 130                        atomic_read(&conntrack->ct_general.use));
 131 +
 132 +       #if defined(CONFIG_IP_NF_MATCH_LAYER7) || defined(CONFIG_IP_NF_MATCH_LAYER7_MODULE)
 133 +       if(conntrack->layer7.app_proto)
 134 +               len += sprintf(buffer + len, "l7proto=%s ",
 135 +                               conntrack->layer7.app_proto);
 136 +       #endif
 137 +
 138         len += sprintf(buffer + len, "\n");
 139
 140         return len;
 141 diff -Nurp linux-2.4.26-stock/net/ipv4/netfilter/ipt_layer7.c linux-2.4.26-layer7-clean/net/ipv4/netfilter/ipt_layer7.c
 142 --- linux-2.4.26-stock/net/ipv4/netfilter/ipt_layer7.c  1969-12-31 18:00:00.000000000 -0600
 143 +++ linux-2.4.26-layer7-clean/net/ipv4/netfilter/ipt_layer7.c   2004-06-27 19:06:51.000000000 -0500
 144 @@ -0,0 +1,540 @@
 145 +/*
 146 +  Kernel module to match application layer (OSI layer 7)
 147 +  data in connections.
 148 +
 149 +  http://l7-filter.sf.net
 150 +
 151 +  By Matthew Strait and Ethan Sommer, 2003.
 152 +
 153 +  This program is free software; you can redistribute it and/or
 154 +  modify it under the terms of the GNU General Public License
 155 +  as published by the Free Software Foundation; either version
 156 +  2 of the License, or (at your option) any later version.
 157 +  http://www.gnu.org/licenses/gpl.txt
 158 +
 159 +  Based on ipt_string.c (C) 2000 Emmanuel Roger <winfield@freegates.be>
 160 +  and cls_layer7.c (C) 2003 Matthew Strait, Ethan Sommer, Justin Levandoski
 161 +*/
 162 +
 163 +#include <linux/module.h>
 164 +#include <linux/skbuff.h>
 165 +#include <linux/netfilter_ipv4/ip_conntrack.h>
 166 +#include <linux/proc_fs.h>
 167 +#include <linux/ctype.h>
 168 +#include <net/ip.h>
 169 +#include <net/tcp.h>
 170 +
 171 +#include "regexp/regexp.c"
 172 +
 173 +#include <linux/netfilter_ipv4/ipt_layer7.h>
 174 +#include <linux/netfilter_ipv4/ip_tables.h>
 175 +
 176 +MODULE_AUTHOR("Matthew Strait <quadong@users.sf.net>, Ethan Sommer <sommere@users.sf.net>");
 177 +MODULE_LICENSE("GPL");
 178 +MODULE_DESCRIPTION("iptables application layer match module");
 179 +
 180 +#if defined(CONFIG_IP_NF_MATCH_LAYER7_DEBUG)
 181 +  #define DPRINTK(format,args...) printk(format,##args)
 182 +#else
 183 +  #define DPRINTK(format,args...)
 184 +#endif
 185 +
 186 +/* Number of packets whose data we look at.
 187 +This can be modified through /proc/net/layer7_numpackets */
 188 +static int num_packets = 8;
 189 +
 190 +static struct pattern_cache {
 191 +  char * regex_string;
 192 +  regexp * pattern;
 193 +  struct pattern_cache * next;
 194 +} * first_pattern_cache = NULL;
 195 +
 196 +/* I'm new to locking.  Here are my assumptions:
 197 +
 198 +- No one is going to write to /proc/net/layer7_numpackets over and over
 199 +  within a short period of time, and if they did, nothing awful would happen.
 200 +
 201 +- This code will never be processing the same packet twice at the same time,
 202 +  because iptables rules need to be traversed in order.
 203 +
 204 +- It doesn't matter if two packets from different connections are in here at
 205 +  the same time, because they don't share any data.
 206 +
 207 +- It _does_ matter if two packets from the same connection are here at the same
 208 +  time.  In this case, the things we have to protect are the conntracks and
 209 +  the list of compiled patterns.
 210 +*/
 211 +DECLARE_RWLOCK(ct_lock);
 212 +DECLARE_LOCK(list_lock);
 213 +
 214 +/* Use instead of regcomp.  As we expect to be seeing the same regexps over and
 215 +over again, it make sense to cache the results. */
 216 +static regexp * compile_and_cache(char * regex_string, char * protocol)
 217 +{
 218 +        struct pattern_cache * node               = first_pattern_cache;
 219 +        struct pattern_cache * last_pattern_cache = first_pattern_cache;
 220 +       struct pattern_cache * tmp;
 221 +        unsigned int len;
 222 +
 223 +        while (node != NULL) {
 224 +                if (!strcmp(node->regex_string, regex_string))
 225 +                        return node->pattern;
 226 +
 227 +                last_pattern_cache = node;/* points at the last non-NULL node */
 228 +                node = node->next;
 229 +        }
 230 +
 231 +        /* If we reach the end of the list, then we have not yet cached
 232 +           the pattern for this regex. Let's do that now.
 233 +          Be paranoid about running out of memory to avoid list corruption. */
 234 +        tmp = kmalloc(sizeof(struct pattern_cache), GFP_ATOMIC);
 235 +
 236 +       if(!tmp) {
 237 +               printk(KERN_ERR "layer7: out of memory in compile_and_cache, bailing.\n");
 238 +               return NULL;
 239 +       }
 240 +
 241 +        tmp->regex_string  = kmalloc(strlen(regex_string) + 1, GFP_ATOMIC);
 242 +        tmp->pattern       = kmalloc(sizeof(struct regexp),    GFP_ATOMIC);
 243 +
 244 +       if(!tmp->regex_string || !tmp->pattern) {
 245 +               printk(KERN_ERR "layer7: out of memory in compile_and_cache, bailing.\n");
 246 +               kfree(tmp->regex_string);
 247 +               kfree(tmp->pattern);
 248 +               kfree(tmp);
 249 +               return NULL;
 250 +       }
 251 +
 252 +        tmp->next = NULL;
 253 +       /* Ok.  The new node is all ready now. */
 254 +       node = tmp;
 255 +
 256 +        if(first_pattern_cache == NULL) /* list is empty */
 257 +                first_pattern_cache = node; /* make node the beginning */
 258 +        else
 259 +                last_pattern_cache->next = node; /* attach node to the end */
 260 +
 261 +        /* copy the string and compile the regex */
 262 +        len = strlen(regex_string);
 263 +       node->pattern = regcomp(regex_string, &len);
 264 +        if ( !node->pattern ) {
 265 +                printk(KERN_ERR "layer7: Error compiling regexp \"%s\" (%s)\n",
 266 +                       regex_string, protocol);
 267 +                /* pattern is now cached as NULL, so we won't try again. */
 268 +        }
 269 +
 270 +        strcpy(node->regex_string, regex_string);
 271 +        return node->pattern;
 272 +}
 273 +
 274 +#if CONFIG_IP_NF_MATCH_LAYER7_DEBUG
 275 +/* Converts an unfriendly string into a friendly one by
 276 +replacing unprintables with periods and all whitespace with " ". */
 277 +static char * friendly_print(unsigned char * s)
 278 +{
 279 +       char * f = kmalloc(strlen(s) + 1, GFP_ATOMIC);
 280 +       int i;
 281 +
 282 +       if(!f) {
 283 +               printk(KERN_ERR "layer7: out of memory in friendly_print, bailing.\n");
 284 +               return NULL;
 285 +        }
 286 +
 287 +       for(i = 0; i < strlen(s); i++){
 288 +               if(isprint(s[i]) && s[i] < 128) f[i] = s[i];
 289 +               else if(isspace(s[i]))          f[i] = ' ';
 290 +               else                            f[i] = '.';
 291 +       }
 292 +       f[i] = '\0';
 293 +       return f;
 294 +}
 295 +
 296 +static char dec2hex(int i)
 297 +{
 298 +       switch (i) {
 299 +               case 0 ... 9:
 300 +                       return (char)(i + '0');
 301 +                       break;
 302 +               case 10 ... 15:
 303 +                       return (char)(i - 10 + 'a');
 304 +                       break;
 305 +               default:
 306 +                       printk("Problem in dec2hex\n");
 307 +                       return '\0';
 308 +       }
 309 +}
 310 +
 311 +static char * hex_print(unsigned char * s)
 312 +{
 313 +       char * g = kmalloc(strlen(s)*3 + 1, GFP_ATOMIC);
 314 +       int i;
 315 +
 316 +       if(!g) {
 317 +               printk(KERN_ERR "layer7: out of memory in hex_print, bailing.\n");
 318 +               return NULL;
 319 +        }
 320 +
 321 +       for(i = 0; i < strlen(s); i++) {
 322 +               g[i*3    ] = dec2hex(s[i]/16);
 323 +               g[i*3 + 1] = dec2hex(s[i]%16);
 324 +               g[i*3 + 2] = ' ';
 325 +       }
 326 +       g[i*3] = '\0';
 327 +
 328 +       return g;
 329 +}
 330 +#endif // DEBUG
 331 +
 332 +
 333 +/* The following functions are here in case we get ported into an environment
 334 +(ebtables?) where skb->nh.iph->protocol isn't set. They assume that skb->data
 335 +points at the beginning of the IP datagram, which is true for iptables (but in
 336 +QoS it points to the beginning of the Ethernet frame). */
 337 +#if 0
 338 +#define IP_PROTO_OFFSET 9
 339 +static int is_tcp_over_ipv4 (const struct sk_buff *skb){return(skb->data[IP_PROTO_OFFSET]==IPPROTO_TCP );}
 340 +static int is_udp_over_ipv4 (const struct sk_buff *skb){return(skb->data[IP_PROTO_OFFSET]==IPPROTO_UDP );}
 341 +static int is_icmp_over_ipv4(const struct sk_buff *skb){return(skb->data[IP_PROTO_OFFSET]==IPPROTO_ICMP);}
 342 +#endif
 343 +
 344 +static int can_handle(const struct sk_buff *skb)
 345 +{
 346 +       if(!skb->nh.iph) /* not IP */
 347 +               return 0;
 348 +       if(skb->nh.iph->protocol != IPPROTO_TCP &&
 349 +          skb->nh.iph->protocol != IPPROTO_UDP &&
 350 +          skb->nh.iph->protocol != IPPROTO_ICMP)
 351 +               return 0;
 352 +       return 1;
 353 +}
 354 +
 355 +/* Returns offset the into the skb->data that the application data starts */
 356 +static int app_data_offset(const struct sk_buff *skb)
 357 +{
 358 +       /* In case we are ported somewhere (ebtables?) where skb->nh.iph
 359 +       isn't set, this can be gotten from 4*(skb->data[0] & 0x0f) as well. */
 360 +        int ip_hl = 4*skb->nh.iph->ihl;
 361 +
 362 +        if( skb->nh.iph->protocol == IPPROTO_TCP ) {
 363 +                /* 12 == offset into TCP header for the header length field.
 364 +               Can't get this with skb->h.th->doff because the tcphdr
 365 +               struct doesn't get set when routing (this is confirmed to be
 366 +               true in Netfilter as well as QoS.) */
 367 +                int tcp_hl = 4*(skb->data[ip_hl + 12] >> 4);
 368 +
 369 +                return ip_hl + tcp_hl;
 370 +        } else if( skb->nh.iph->protocol == IPPROTO_UDP  ) {
 371 +                return ip_hl + 8; /* UDP header is always 8 bytes */
 372 +        } else if( skb->nh.iph->protocol == IPPROTO_ICMP ) {
 373 +                return ip_hl + 8; /* ICMP header is 8 bytes */
 374 +        } else {
 375 +                printk(KERN_ERR "layer7: tried to handle unknown protocol!\n");
 376 +                return ip_hl + 8; /* something reasonable */
 377 +        }
 378 +}
 379 +
 380 +/* handles whether there's a match when we aren't appending data anymore */
 381 +static int match_no_append(struct ip_conntrack * conntrack,
 382 +                       struct ip_conntrack * master_conntrack,
 383 +                       struct ipt_layer7_info * info)
 384 +{
 385 +       /* If we're in here, we don't care about the app data anymore */
 386 +       WRITE_LOCK(&ct_lock);
 387 +       if(master_conntrack->layer7.app_data != NULL) {
 388 +
 389 +                       #ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
 390 +                       if(!master_conntrack->layer7.app_proto) {
 391 +                               char * f = friendly_print(master_conntrack->layer7.app_data);
 392 +                       char * g = hex_print(master_conntrack->layer7.app_data);
 393 +                               DPRINTK("\nGave up on the %d length stream: \n%s\n",
 394 +                              master_conntrack->layer7.app_data_len, f);
 395 +                       DPRINTK("\nIn hex: %s\n", g);
 396 +                               kfree(f);
 397 +                       kfree(g);
 398 +                       }
 399 +                       #endif
 400 +
 401 +               kfree(master_conntrack->layer7.app_data);
 402 +               master_conntrack->layer7.app_data = NULL; /* don't free again */
 403 +       }
 404 +       WRITE_UNLOCK(&ct_lock);
 405 +
 406 +       /* Is top-level master (possibly self) classified? */
 407 +       if(master_conntrack->layer7.app_proto) {
 408 +               if(!strcmp(master_conntrack->layer7.app_proto, info->protocol))
 409 +               {
 410 +                       /* set own .protocol (for /proc/net/ip_conntrack) */
 411 +                       WRITE_LOCK(&ct_lock);
 412 +                       if(!conntrack->layer7.app_proto) {
 413 +                               conntrack->layer7.app_proto = kmalloc(strlen(info->protocol), GFP_ATOMIC);
 414 +                                if(!conntrack->layer7.app_proto){
 415 +                                       printk(KERN_ERR "layer7: out of memory in match_no_append, bailing.\n");
 416 +                                       WRITE_UNLOCK(&ct_lock);
 417 +                                       return 1;
 418 +                                }
 419 +
 420 +                               strcpy(conntrack->layer7.app_proto, info->protocol);
 421 +                       }
 422 +                       WRITE_UNLOCK(&ct_lock);
 423 +
 424 +                       return 1;
 425 +               } else return 0;
 426 +       } else return 0; /* no clasification */
 427 +}
 428 +
 429 +/* add the new app data to the conntrack.  Return number of bytes added. */
 430 +static int add_data(struct ip_conntrack * master_conntrack,
 431 +                       char * app_data, int appdatalen)
 432 +{
 433 +       int length = 0, i;
 434 +       int oldlength = master_conntrack->layer7.app_data_len;
 435 +
 436 +        /* Strip nulls. Make everything lower case (our regex lib doesn't
 437 +        do case insensitivity).  Add it to the end of the current data. */
 438 +       for(i = 0; i < CONFIG_IP_NF_MATCH_LAYER7_MAXDATALEN-oldlength-1 &&
 439 +                  i < appdatalen; i++) {
 440 +               if(app_data[i] != '\0') {
 441 +                       master_conntrack->layer7.app_data[length+oldlength] =
 442 +                               /* the kernel version of tolower mungs 'upper ascii' */
 443 +                               isascii(app_data[i])? tolower(app_data[i]) : app_data[i];
 444 +                       length++;
 445 +               }
 446 +       }
 447 +
 448 +       master_conntrack->layer7.app_data[length+oldlength] = '\0';
 449 +       master_conntrack->layer7.app_data_len = length + oldlength;
 450 +
 451 +       return length;
 452 +}
 453 +
 454 +/* Returns true on match and false otherwise.  */
 455 +static int match(/* const */struct sk_buff *skb, const struct net_device *in,
 456 +                 const struct net_device *out, const void *matchinfo,
 457 +                 int offset,                   int *hotdrop)
 458 +{
 459 +       struct ipt_layer7_info * info = (struct ipt_layer7_info *)matchinfo;
 460 +        enum ip_conntrack_info master_ctinfo, ctinfo;
 461 +        struct ip_conntrack *master_conntrack, *conntrack;
 462 +       unsigned char * app_data;
 463 +       unsigned int pattern_result, appdatalen;
 464 +       regexp * comppattern;
 465 +
 466 +       if(!can_handle(skb)){
 467 +               DPRINTK("layer7: This is some protocol I can't handle\n");
 468 +               return info->invert;
 469 +       }
 470 +
 471 +       LOCK_BH(&list_lock);
 472 +       comppattern = compile_and_cache(info->pattern, info->protocol);
 473 +       UNLOCK_BH(&list_lock);
 474 +       /* the return value gets checked later, when we're ready to use it */
 475 +
 476 +       app_data = skb->data + app_data_offset(skb);
 477 +       appdatalen = skb->tail - app_data;
 478 +
 479 +       /* Treat the parent and all its children together as one connection,
 480 +       except for the purpose of setting conntrack->layer7.pattern in the
 481 +       actual connection. This makes /proc/net/ip_conntrack somewhat more
 482 +       satisfying. */
 483 +        if(!(conntrack        = ip_conntrack_get((struct sk_buff *)skb, &ctinfo)) ||
 484 +           !(master_conntrack = ip_conntrack_get((struct sk_buff *)skb, &master_ctinfo))) {
 485 +                DPRINTK("layer7: packet is not from a known connection, giving up.\n");
 486 +               return info->invert;
 487 +        }
 488 +
 489 +        /* Try to get a master conntrack (and its master etc) for FTP, etc. */
 490 +        while (master_ct(master_conntrack) != NULL)
 491 +                master_conntrack = master_ct(master_conntrack);
 492 +
 493 +       /* skb->cb[0] == seen. Avoid doing things twice if there are two layer7
 494 +       rules. I'm not sure that using cb for this purpose is correct, although
 495 +       it says "put your private variables there" and this seems to qualify.
 496 +       But it doesn't look like it's being used for anything else in the
 497 +       sk_buffs that make it here. I'm open to suggestions for how to be able
 498 +       to write to cb without making the compiler angry.  That I can't figure
 499 +       this out is an argument against this being correct. */
 500 +       if(!skb->cb[0]){
 501 +               WRITE_LOCK(&ct_lock);
 502 +               master_conntrack->layer7.numpackets++;/*starts at 0 via memset*/
 503 +               WRITE_UNLOCK(&ct_lock);
 504 +       }
 505 +
 506 +       /* On the first packet of a connection, allocate space for app data */
 507 +       WRITE_LOCK(&ct_lock);
 508 +       if(master_conntrack->layer7.numpackets == 1 && !skb->cb[0]) {
 509 +               master_conntrack->layer7.app_data = kmalloc(CONFIG_IP_NF_MATCH_LAYER7_MAXDATALEN, GFP_ATOMIC);
 510 +                if(!master_conntrack->layer7.app_data){
 511 +                        printk(KERN_ERR "layer7: out of memory in match, bailing.\n");
 512 +                        WRITE_UNLOCK(&ct_lock);
 513 +                        return info->invert;
 514 +                }
 515 +
 516 +               master_conntrack->layer7.app_data[0] = '\0';
 517 +       }
 518 +       WRITE_UNLOCK(&ct_lock);
 519 +
 520 +       /* if we've classified it or seen too many packets */
 521 +       if(master_conntrack->layer7.numpackets > num_packets ||
 522 +          master_conntrack->layer7.app_proto) {
 523 +
 524 +               pattern_result = match_no_append(conntrack, master_conntrack, info);
 525 +
 526 +               /* mark the packet seen (probably irrelevant, but consistant) */
 527 +               skb->cb[0] = 1;
 528 +
 529 +               return (pattern_result ^ info->invert);
 530 +       }
 531 +
 532 +       /* Can end up here, but unallocated, if numpackets is increased during
 533 +       the beginning of a connection */
 534 +       if(master_conntrack->layer7.app_data == NULL)
 535 +               return (info->invert); /* unmatched */
 536 +
 537 +       if(!skb->cb[0]) {
 538 +               int newbytes;
 539 +               WRITE_LOCK(&ct_lock);
 540 +               newbytes = add_data(master_conntrack, app_data, appdatalen);
 541 +               WRITE_UNLOCK(&ct_lock);
 542 +
 543 +               if(newbytes == 0) { /* didn't add any data */
 544 +                       skb->cb[0] = 1;
 545 +                       /* Didn't match before, not going to match now */
 546 +                       return info->invert;
 547 +               }
 548 +       }
 549 +
 550 +       /* If the regexp failed to compile, don't bother running it */
 551 +       if(comppattern && regexec(comppattern, master_conntrack->layer7.app_data)) {
 552 +               DPRINTK("layer7: regexec positive: %s!\n", info->protocol);
 553 +               pattern_result = 1;
 554 +       } else pattern_result = 0;
 555 +
 556 +       if(pattern_result) {
 557 +               WRITE_LOCK(&ct_lock);
 558 +               conntrack->layer7.app_proto = kmalloc(strlen(info->protocol), GFP_ATOMIC);
 559 +                if(!conntrack->layer7.app_proto){
 560 +                        printk(KERN_ERR "layer7: out of memory in match, bailing.\n");
 561 +                        WRITE_UNLOCK(&ct_lock);
 562 +                        return (pattern_result ^ info->invert);
 563 +                        }
 564 +
 565 +               strcpy(conntrack->layer7.app_proto, info->protocol);
 566 +               WRITE_UNLOCK(&ct_lock);
 567 +       }
 568 +
 569 +       /* mark the packet seen */
 570 +       skb->cb[0] = 1;
 571 +
 572 +       return (pattern_result ^ info->invert);
 573 +}
 574 +
 575 +static int checkentry(const char *tablename, const struct ipt_ip *ip,
 576 +           void *matchinfo, unsigned int matchsize, unsigned int hook_mask)
 577 +{
 578 +       if (matchsize != IPT_ALIGN(sizeof(struct ipt_layer7_info)))
 579 +               return 0;
 580 +       return 1;
 581 +}
 582 +
 583 +static struct ipt_match layer7_match = {
 584 +       .name = "layer7",
 585 +       .match = &match,
 586 +       .checkentry = &checkentry,
 587 +       .me = THIS_MODULE
 588 +};
 589 +
 590 +/* taken from drivers/video/modedb.c */
 591 +static int my_atoi(const char *s)
 592 +{
 593 +    int val = 0;
 594 +
 595 +    for (;; s++) {
 596 +        switch (*s) {
 597 +            case '0'...'9':
 598 +                val = 10*val+(*s-'0');
 599 +                break;
 600 +            default:
 601 +                return val;
 602 +        }
 603 +    }
 604 +}
 605 +
 606 +/* write out num_packets to userland. */
 607 +static int layer7_read_proc(char* page, char ** start, off_t off, int count,
 608 +                     int* eof, void * data)
 609 +{
 610 +        if(num_packets > 99)
 611 +                printk(KERN_ERR "layer7: NOT REACHED. num_packets too big\n");
 612 +
 613 +        page[0] = num_packets/10 + '0';
 614 +        page[1] = num_packets%10 + '0';
 615 +        page[2] = '\n';
 616 +        page[3] = '\0';
 617 +
 618 +        *eof=1;
 619 +
 620 +        return 3;
 621 +}
 622 +
 623 +/* Read in num_packets from userland */
 624 +static int layer7_write_proc(struct file* file, const char* buffer,
 625 +                      unsigned long count, void *data)
 626 +{
 627 +        char * foo = kmalloc(count, GFP_ATOMIC);
 628 +
 629 +        if(!foo){
 630 +                printk(KERN_ERR "layer7: out of memory, bailing.  num_packets unchanged.\n");
 631 +                return count;
 632 +        }
 633 +
 634 +        /* copy in the data from userland */
 635 +        copy_from_user(foo, buffer, count);
 636 +
 637 +        num_packets = my_atoi(foo);
 638 +       kfree (foo);
 639 +
 640 +        /* This has an arbitrary limit to make the math easier. I'm lazy.
 641 +       But anyway, 99 is a LOT! If you want more, you're doing it wrong! */
 642 +        if(num_packets > 99) {
 643 +                printk(KERN_WARNING "layer7: num_packets can't be > 99.\n");
 644 +                num_packets = 99;
 645 +        } else if(num_packets < 1) {
 646 +                printk(KERN_WARNING "layer7: num_packets can't be < 1.\n");
 647 +                num_packets = 1;
 648 +        }
 649 +
 650 +        return count;
 651 +}
 652 +
 653 +/* register the proc file */
 654 +static void layer7_init_proc(void)
 655 +{
 656 +        struct proc_dir_entry* entry;
 657 +
 658 +        /* create the file */
 659 +        entry = create_proc_entry("layer7_numpackets", 0644, proc_net);
 660 +
 661 +        /* set the callback functions */
 662 +       entry->read_proc = layer7_read_proc;
 663 +       entry->write_proc = layer7_write_proc;
 664 +}
 665 +
 666 +static void layer7_cleanup_proc(void)
 667 +{
 668 +        remove_proc_entry("layer7_numpackets", proc_net);
 669 +}
 670 +
 671 +static int __init init(void)
 672 +{
 673 +        layer7_init_proc();
 674 +       return ipt_register_match(&layer7_match);
 675 +}
 676 +
 677 +static void __exit fini(void)
 678 +{
 679 +       layer7_cleanup_proc();
 680 +       ipt_unregister_match(&layer7_match);
 681 +}
 682 +
 683 +module_init(init);
 684 +module_exit(fini);
 685 diff -Nurp linux-2.4.26-stock/net/ipv4/netfilter/regexp/regexp.c linux-2.4.26-layer7-clean/net/ipv4/netfilter/regexp/regexp.c
 686 --- linux-2.4.26-stock/net/ipv4/netfilter/regexp/regexp.c       1969-12-31 18:00:00.000000000 -0600
 687 +++ linux-2.4.26-layer7-clean/net/ipv4/netfilter/regexp/regexp.c        2004-06-27 19:07:00.000000000 -0500
 688 @@ -0,0 +1,1195 @@
 689 +/*
 690 + * regcomp and regexec -- regsub and regerror are elsewhere
 691 + * @(#)regexp.c        1.3 of 18 April 87
 692 + *
 693 + *     Copyright (c) 1986 by University of Toronto.
 694 + *     Written by Henry Spencer.  Not derived from licensed software.
 695 + *
 696 + *     Permission is granted to anyone to use this software for any
 697 + *     purpose on any computer system, and to redistribute it freely,
 698 + *     subject to the following restrictions:
 699 + *
 700 + *     1. The author is not responsible for the consequences of use of
 701 + *             this software, no matter how awful, even if they arise
 702 + *             from defects in it.
 703 + *
 704 + *     2. The origin of this software must not be misrepresented, either
 705 + *             by explicit claim or by omission.
 706 + *
 707 + *     3. Altered versions must be plainly marked as such, and must not
 708 + *             be misrepresented as being the original software.
 709 + *
 710 + * Beware that some of this code is subtly aware of the way operator
 711 + * precedence is structured in regular expressions.  Serious changes in
 712 + * regular-expression syntax might require a total rethink.
 713 + *
 714 + * This code was modified by Ethan Sommer to work within the kernel
 715 + * (it now uses kmalloc etc..)
 716 + *
 717 + * Modified slightly by Matthew Strait to use more modern C.
 718 + */
 719 +
 720 +#include "regexp.h"
 721 +#include "regmagic.h"
 722 +
 723 +/* added by ethan and matt.  Lets it work in both kernel and user space.
 724 +(So iptables can use it, for instance.)  Yea, it goes both ways... */
 725 +#if __KERNEL__
 726 +  #define malloc(foo) kmalloc(foo,GFP_ATOMIC)
 727 +#else
 728 +  #define printk(format,args...) printf(format,##args)
 729 +#endif
 730 +
 731 +void regerror(char * s)
 732 +{
 733 +        printk("<3>Regexp: %s\n", s);
 734 +        /* NOTREACHED */
 735 +}
 736 +
 737 +/*
 738 + * The "internal use only" fields in regexp.h are present to pass info from
 739 + * compile to execute that permits the execute phase to run lots faster on
 740 + * simple cases.  They are:
 741 + *
 742 + * regstart    char that must begin a match; '\0' if none obvious
 743 + * reganch     is the match anchored (at beginning-of-line only)?
 744 + * regmust     string (pointer into program) that match must include, or NULL
 745 + * regmlen     length of regmust string
 746 + *
 747 + * Regstart and reganch permit very fast decisions on suitable starting points
 748 + * for a match, cutting down the work a lot.  Regmust permits fast rejection
 749 + * of lines that cannot possibly match.  The regmust tests are costly enough
 750 + * that regcomp() supplies a regmust only if the r.e. contains something
 751 + * potentially expensive (at present, the only such thing detected is * or +
 752 + * at the start of the r.e., which can involve a lot of backup).  Regmlen is
 753 + * supplied because the test in regexec() needs it and regcomp() is computing
 754 + * it anyway.
 755 + */
 756 +
 757 +/*
 758 + * Structure for regexp "program".  This is essentially a linear encoding
 759 + * of a nondeterministic finite-state machine (aka syntax charts or
 760 + * "railroad normal form" in parsing technology).  Each node is an opcode
 761 + * plus a "next" pointer, possibly plus an operand.  "Next" pointers of
 762 + * all nodes except BRANCH implement concatenation; a "next" pointer with
 763 + * a BRANCH on both ends of it is connecting two alternatives.  (Here we
 764 + * have one of the subtle syntax dependencies:  an individual BRANCH (as
 765 + * opposed to a collection of them) is never concatenated with anything
 766 + * because of operator precedence.)  The operand of some types of node is
 767 + * a literal string; for others, it is a node leading into a sub-FSM.  In
 768 + * particular, the operand of a BRANCH node is the first node of the branch.
 769 + * (NB this is *not* a tree structure:  the tail of the branch connects
 770 + * to the thing following the set of BRANCHes.)  The opcodes are:
 771 + */
 772 +
 773 +/* definition  number  opnd?   meaning */
 774 +#define        END     0       /* no   End of program. */
 775 +#define        BOL     1       /* no   Match "" at beginning of line. */
 776 +#define        EOL     2       /* no   Match "" at end of line. */
 777 +#define        ANY     3       /* no   Match any one character. */
 778 +#define        ANYOF   4       /* str  Match any character in this string. */
 779 +#define        ANYBUT  5       /* str  Match any character not in this string. */
 780 +#define        BRANCH  6       /* node Match this alternative, or the next... */
 781 +#define        BACK    7       /* no   Match "", "next" ptr points backward. */
 782 +#define        EXACTLY 8       /* str  Match this string. */
 783 +#define        NOTHING 9       /* no   Match empty string. */
 784 +#define        STAR    10      /* node Match this (simple) thing 0 or more times. */
 785 +#define        PLUS    11      /* node Match this (simple) thing 1 or more times. */
 786 +#define        OPEN    20      /* no   Mark this point in input as start of #n. */
 787 +                       /*      OPEN+1 is number 1, etc. */
 788 +#define        CLOSE   30      /* no   Analogous to OPEN. */
 789 +
 790 +/*
 791 + * Opcode notes:
 792 + *
 793 + * BRANCH      The set of branches constituting a single choice are hooked
 794 + *             together with their "next" pointers, since precedence prevents
 795 + *             anything being concatenated to any individual branch.  The
 796 + *             "next" pointer of the last BRANCH in a choice points to the
 797 + *             thing following the whole choice.  This is also where the
 798 + *             final "next" pointer of each individual branch points; each
 799 + *             branch starts with the operand node of a BRANCH node.
 800 + *
 801 + * BACK                Normal "next" pointers all implicitly point forward; BACK
 802 + *             exists to make loop structures possible.
 803 + *
 804 + * STAR,PLUS   '?', and complex '*' and '+', are implemented as circular
 805 + *             BRANCH structures using BACK.  Simple cases (one character
 806 + *             per match) are implemented with STAR and PLUS for speed
 807 + *             and to minimize recursive plunges.
 808 + *
 809 + * OPEN,CLOSE  ...are numbered at compile time.
 810 + */
 811 +
 812 +/*
 813 + * A node is one char of opcode followed by two chars of "next" pointer.
 814 + * "Next" pointers are stored as two 8-bit pieces, high order first.  The
 815 + * value is a positive offset from the opcode of the node containing it.
 816 + * An operand, if any, simply follows the node.  (Note that much of the
 817 + * code generation knows about this implicit relationship.)
 818 + *
 819 + * Using two bytes for the "next" pointer is vast overkill for most things,
 820 + * but allows patterns to get big without disasters.
 821 + */
 822 +#define        OP(p)   (*(p))
 823 +#define        NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377))
 824 +#define        OPERAND(p)      ((p) + 3)
 825 +
 826 +/*
 827 + * See regmagic.h for one further detail of program structure.
 828 + */
 829 +
 830 +
 831 +/*
 832 + * Utility definitions.
 833 + */
 834 +#ifndef CHARBITS
 835 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
 836 +#else
 837 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
 838 +#endif
 839 +
 840 +#define        FAIL(m) { regerror(m); return(NULL); }
 841 +#define        ISMULT(c)       ((c) == '*' || (c) == '+' || (c) == '?')
 842 +#define        META    "^$.[()|?+*\\"
 843 +
 844 +/*
 845 + * Flags to be passed up and down.
 846 + */
 847 +#define        HASWIDTH        01      /* Known never to match null string. */
 848 +#define        SIMPLE          02      /* Simple enough to be STAR/PLUS operand. */
 849 +#define        SPSTART         04      /* Starts with * or +. */
 850 +#define        WORST           0       /* Worst case. */
 851 +
 852 +/*
 853 + * Global work variables for regcomp().
 854 + */
 855 +static char *regparse;         /* Input-scan pointer. */
 856 +static int regnpar;            /* () count. */
 857 +static char regdummy;
 858 +static char *regcode;          /* Code-emit pointer; &regdummy = don't. */
 859 +static long regsize;           /* Code size. */
 860 +
 861 +/*
 862 + * Forward declarations for regcomp()'s friends.
 863 + */
 864 +#ifndef STATIC
 865 +#define        STATIC  static
 866 +#endif
 867 +STATIC char *reg(int paren,int *flagp);
 868 +STATIC char *regbranch(int *flagp);
 869 +STATIC char *regpiece(int *flagp);
 870 +STATIC char *regatom(int *flagp);
 871 +STATIC char *regnode(char op);
 872 +STATIC char *regnext(char *p);
 873 +STATIC void regc(char b);
 874 +STATIC void reginsert(char op, char *opnd);
 875 +STATIC void regtail(char *p, char *val);
 876 +STATIC void regoptail(char *p, char *val);
 877 +
 878 +
 879 +__kernel_size_t my_strcspn(const char *s1,const char *s2)
 880 +{
 881 +        char *scan1;
 882 +        char *scan2;
 883 +        int count;
 884 +
 885 +        count = 0;
 886 +        for (scan1 = (char *)s1; *scan1 != '\0'; scan1++) {
 887 +                for (scan2 = (char *)s2; *scan2 != '\0';)       /* ++ moved down. */
 888 +                        if (*scan1 == *scan2++)
 889 +                                return(count);
 890 +                count++;
 891 +        }
 892 +        return(count);
 893 +}
 894 +
 895 +/*
 896 + - regcomp - compile a regular expression into internal code
 897 + *
 898 + * We can't allocate space until we know how big the compiled form will be,
 899 + * but we can't compile it (and thus know how big it is) until we've got a
 900 + * place to put the code.  So we cheat:  we compile it twice, once with code
 901 + * generation turned off and size counting turned on, and once "for real".
 902 + * This also means that we don't allocate space until we are sure that the
 903 + * thing really will compile successfully, and we never have to move the
 904 + * code and thus invalidate pointers into it.  (Note that it has to be in
 905 + * one piece because free() must be able to free it all.)
 906 + *
 907 + * Beware that the optimization-preparation code in here knows about some
 908 + * of the structure of the compiled regexp.
 909 + */
 910 +regexp *
 911 +regcomp(char *exp,int *patternsize)
 912 +{
 913 +       register regexp *r;
 914 +       register char *scan;
 915 +       register char *longest;
 916 +       register int len;
 917 +       int flags;
 918 +       /* commented out by ethan
 919 +          extern char *malloc();
 920 +       */
 921 +
 922 +       if (exp == NULL)
 923 +               FAIL("NULL argument");
 924 +
 925 +       /* First pass: determine size, legality. */
 926 +       regparse = exp;
 927 +       regnpar = 1;
 928 +       regsize = 0L;
 929 +       regcode = &regdummy;
 930 +       regc(MAGIC);
 931 +       if (reg(0, &flags) == NULL)
 932 +               return(NULL);
 933 +
 934 +       /* Small enough for pointer-storage convention? */
 935 +       if (regsize >= 32767L)          /* Probably could be 65535L. */
 936 +               FAIL("regexp too big");
 937 +
 938 +       /* Allocate space. */
 939 +       *patternsize=sizeof(regexp) + (unsigned)regsize;
 940 +       r = (regexp *)malloc(sizeof(regexp) + (unsigned)regsize);
 941 +       if (r == NULL)
 942 +               FAIL("out of space");
 943 +
 944 +       /* Second pass: emit code. */
 945 +       regparse = exp;
 946 +       regnpar = 1;
 947 +       regcode = r->program;
 948 +       regc(MAGIC);
 949 +       if (reg(0, &flags) == NULL)
 950 +               return(NULL);
 951 +
 952 +       /* Dig out information for optimizations. */
 953 +       r->regstart = '\0';     /* Worst-case defaults. */
 954 +       r->reganch = 0;
 955 +       r->regmust = NULL;
 956 +       r->regmlen = 0;
 957 +       scan = r->program+1;                    /* First BRANCH. */
 958 +       if (OP(regnext(scan)) == END) {         /* Only one top-level choice. */
 959 +               scan = OPERAND(scan);
 960 +
 961 +               /* Starting-point info. */
 962 +               if (OP(scan) == EXACTLY)
 963 +                       r->regstart = *OPERAND(scan);
 964 +               else if (OP(scan) == BOL)
 965 +                       r->reganch++;
 966 +
 967 +               /*
 968 +                * If there's something expensive in the r.e., find the
 969 +                * longest literal string that must appear and make it the
 970 +                * regmust.  Resolve ties in favor of later strings, since
 971 +                * the regstart check works with the beginning of the r.e.
 972 +                * and avoiding duplication strengthens checking.  Not a
 973 +                * strong reason, but sufficient in the absence of others.
 974 +                */
 975 +               if (flags&SPSTART) {
 976 +                       longest = NULL;
 977 +                       len = 0;
 978 +                       for (; scan != NULL; scan = regnext(scan))
 979 +                               if (OP(scan) == EXACTLY && strlen(OPERAND(scan)) >= len) {
 980 +                                       longest = OPERAND(scan);
 981 +                                       len = strlen(OPERAND(scan));
 982 +                               }
 983 +                       r->regmust = longest;
 984 +                       r->regmlen = len;
 985 +               }
 986 +       }
 987 +
 988 +       return(r);
 989 +}
 990 +
 991 +/*
 992 + - reg - regular expression, i.e. main body or parenthesized thing
 993 + *
 994 + * Caller must absorb opening parenthesis.
 995 + *
 996 + * Combining parenthesis handling with the base level of regular expression
 997 + * is a trifle forced, but the need to tie the tails of the branches to what
 998 + * follows makes it hard to avoid.
 999 + */
1000 +static char *
1001 +reg(int paren, int *flagp /* Parenthesized? */ )
1002 +{
1003 +       register char *ret;
1004 +       register char *br;
1005 +       register char *ender;
1006 +       register int parno = 0; /* 0 makes gcc happy */
1007 +       int flags;
1008 +
1009 +       *flagp = HASWIDTH;      /* Tentatively. */
1010 +
1011 +       /* Make an OPEN node, if parenthesized. */
1012 +       if (paren) {
1013 +               if (regnpar >= NSUBEXP)
1014 +                       FAIL("too many ()");
1015 +               parno = regnpar;
1016 +               regnpar++;
1017 +               ret = regnode(OPEN+parno);
1018 +       } else
1019 +               ret = NULL;
1020 +
1021 +       /* Pick up the branches, linking them together. */
1022 +       br = regbranch(&flags);
1023 +       if (br == NULL)
1024 +               return(NULL);
1025 +       if (ret != NULL)
1026 +               regtail(ret, br);       /* OPEN -> first. */
1027 +       else
1028 +               ret = br;
1029 +       if (!(flags&HASWIDTH))
1030 +               *flagp &= ~HASWIDTH;
1031 +       *flagp |= flags&SPSTART;
1032 +       while (*regparse == '|') {
1033 +               regparse++;
1034 +               br = regbranch(&flags);
1035 +               if (br == NULL)
1036 +                       return(NULL);
1037 +               regtail(ret, br);       /* BRANCH -> BRANCH. */
1038 +               if (!(flags&HASWIDTH))
1039 +                       *flagp &= ~HASWIDTH;
1040 +               *flagp |= flags&SPSTART;
1041 +       }
1042 +
1043 +       /* Make a closing node, and hook it on the end. */
1044 +       ender = regnode((paren) ? CLOSE+parno : END);
1045 +       regtail(ret, ender);
1046 +
1047 +       /* Hook the tails of the branches to the closing node. */
1048 +       for (br = ret; br != NULL; br = regnext(br))
1049 +               regoptail(br, ender);
1050 +
1051 +       /* Check for proper termination. */
1052 +       if (paren && *regparse++ != ')') {
1053 +               FAIL("unmatched ()");
1054 +       } else if (!paren && *regparse != '\0') {
1055 +               if (*regparse == ')') {
1056 +                       FAIL("unmatched ()");
1057 +               } else
1058 +                       FAIL("junk on end");    /* "Can't happen". */
1059 +               /* NOTREACHED */
1060 +       }
1061 +
1062 +       return(ret);
1063 +}
1064 +
1065 +/*
1066 + - regbranch - one alternative of an | operator
1067 + *
1068 + * Implements the concatenation operator.
1069 + */
1070 +static char *
1071 +regbranch(int *flagp)
1072 +{
1073 +       register char *ret;
1074 +       register char *chain;
1075 +       register char *latest;
1076 +       int flags;
1077 +
1078 +       *flagp = WORST;         /* Tentatively. */
1079 +
1080 +       ret = regnode(BRANCH);
1081 +       chain = NULL;
1082 +       while (*regparse != '\0' && *regparse != '|' && *regparse != ')') {
1083 +               latest = regpiece(&flags);
1084 +               if (latest == NULL)
1085 +                       return(NULL);
1086 +               *flagp |= flags&HASWIDTH;
1087 +               if (chain == NULL)      /* First piece. */
1088 +                       *flagp |= flags&SPSTART;
1089 +               else
1090 +                       regtail(chain, latest);
1091 +               chain = latest;
1092 +       }
1093 +       if (chain == NULL)      /* Loop ran zero times. */
1094 +               (void) regnode(NOTHING);
1095 +
1096 +       return(ret);
1097 +}
1098 +
1099 +/*
1100 + - regpiece - something followed by possible [*+?]
1101 + *
1102 + * Note that the branching code sequences used for ? and the general cases
1103 + * of * and + are somewhat optimized:  they use the same NOTHING node as
1104 + * both the endmarker for their branch list and the body of the last branch.
1105 + * It might seem that this node could be dispensed with entirely, but the
1106 + * endmarker role is not redundant.
1107 + */
1108 +static char *
1109 +regpiece(int *flagp)
1110 +{
1111 +       register char *ret;
1112 +       register char op;
1113 +       register char *next;
1114 +       int flags;
1115 +
1116 +       ret = regatom(&flags);
1117 +       if (ret == NULL)
1118 +               return(NULL);
1119 +
1120 +       op = *regparse;
1121 +       if (!ISMULT(op)) {
1122 +               *flagp = flags;
1123 +               return(ret);
1124 +       }
1125 +
1126 +       if (!(flags&HASWIDTH) && op != '?')
1127 +               FAIL("*+ operand could be empty");
1128 +       *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
1129 +
1130 +       if (op == '*' && (flags&SIMPLE))
1131 +               reginsert(STAR, ret);
1132 +       else if (op == '*') {
1133 +               /* Emit x* as (x&|), where & means "self". */
1134 +               reginsert(BRANCH, ret);                 /* Either x */
1135 +               regoptail(ret, regnode(BACK));          /* and loop */
1136 +               regoptail(ret, ret);                    /* back */
1137 +               regtail(ret, regnode(BRANCH));          /* or */
1138 +               regtail(ret, regnode(NOTHING));         /* null. */
1139 +       } else if (op == '+' && (flags&SIMPLE))
1140 +               reginsert(PLUS, ret);
1141 +       else if (op == '+') {
1142 +               /* Emit x+ as x(&|), where & means "self". */
1143 +               next = regnode(BRANCH);                 /* Either */
1144 +               regtail(ret, next);
1145 +               regtail(regnode(BACK), ret);            /* loop back */
1146 +               regtail(next, regnode(BRANCH));         /* or */
1147 +               regtail(ret, regnode(NOTHING));         /* null. */
1148 +       } else if (op == '?') {
1149 +               /* Emit x? as (x|) */
1150 +               reginsert(BRANCH, ret);                 /* Either x */
1151 +               regtail(ret, regnode(BRANCH));          /* or */
1152 +               next = regnode(NOTHING);                /* null. */
1153 +               regtail(ret, next);
1154 +               regoptail(ret, next);
1155 +       }
1156 +       regparse++;
1157 +       if (ISMULT(*regparse))
1158 +               FAIL("nested *?+");
1159 +
1160 +       return(ret);
1161 +}
1162 +
1163 +/*
1164 + - regatom - the lowest level
1165 + *
1166 + * Optimization:  gobbles an entire sequence of ordinary characters so that
1167 + * it can turn them into a single node, which is smaller to store and
1168 + * faster to run.  Backslashed characters are exceptions, each becoming a
1169 + * separate node; the code is simpler that way and it's not worth fixing.
1170 + */
1171 +static char *
1172 +regatom(int *flagp)
1173 +{
1174 +       register char *ret;
1175 +       int flags;
1176 +
1177 +       *flagp = WORST;         /* Tentatively. */
1178 +
1179 +       switch (*regparse++) {
1180 +       case '^':
1181 +               ret = regnode(BOL);
1182 +               break;
1183 +       case '$':
1184 +               ret = regnode(EOL);
1185 +               break;
1186 +       case '.':
1187 +               ret = regnode(ANY);
1188 +               *flagp |= HASWIDTH|SIMPLE;
1189 +               break;
1190 +       case '[': {
1191 +                       register int class;
1192 +                       register int classend;
1193 +
1194 +                       if (*regparse == '^') { /* Complement of range. */
1195 +                               ret = regnode(ANYBUT);
1196 +                               regparse++;
1197 +                       } else
1198 +                               ret = regnode(ANYOF);
1199 +                       if (*regparse == ']' || *regparse == '-')
1200 +                               regc(*regparse++);
1201 +                       while (*regparse != '\0' && *regparse != ']') {
1202 +                               if (*regparse == '-') {
1203 +                                       regparse++;
1204 +                                       if (*regparse == ']' || *regparse == '\0')
1205 +                                               regc('-');
1206 +                                       else {
1207 +                                               class = UCHARAT(regparse-2)+1;
1208 +                                               classend = UCHARAT(regparse);
1209 +                                               if (class > classend+1)
1210 +                                                       FAIL("invalid [] range");
1211 +                                               for (; class <= classend; class++)
1212 +                                                       regc(class);
1213 +                                               regparse++;
1214 +                                       }
1215 +                               } else
1216 +                                       regc(*regparse++);
1217 +                       }
1218 +                       regc('\0');
1219 +                       if (*regparse != ']')
1220 +                               FAIL("unmatched []");
1221 +                       regparse++;
1222 +                       *flagp |= HASWIDTH|SIMPLE;
1223 +               }
1224 +               break;
1225 +       case '(':
1226 +               ret = reg(1, &flags);
1227 +               if (ret == NULL)
1228 +                       return(NULL);
1229 +               *flagp |= flags&(HASWIDTH|SPSTART);
1230 +               break;
1231 +       case '\0':
1232 +       case '|':
1233 +       case ')':
1234 +               FAIL("internal urp");   /* Supposed to be caught earlier. */
1235 +               break;
1236 +       case '?':
1237 +       case '+':
1238 +       case '*':
1239 +               FAIL("?+* follows nothing");
1240 +               break;
1241 +       case '\\':
1242 +               if (*regparse == '\0')
1243 +                       FAIL("trailing \\");
1244 +               ret = regnode(EXACTLY);
1245 +               regc(*regparse++);
1246 +               regc('\0');
1247 +               *flagp |= HASWIDTH|SIMPLE;
1248 +               break;
1249 +       default: {
1250 +                       register int len;
1251 +                       register char ender;
1252 +
1253 +                       regparse--;
1254 +                       len = my_strcspn((const char *)regparse, (const char *)META);
1255 +                       if (len <= 0)
1256 +                               FAIL("internal disaster");
1257 +                       ender = *(regparse+len);
1258 +                       if (len > 1 && ISMULT(ender))
1259 +                               len--;          /* Back off clear of ?+* operand. */
1260 +                       *flagp |= HASWIDTH;
1261 +                       if (len == 1)
1262 +                               *flagp |= SIMPLE;
1263 +                       ret = regnode(EXACTLY);
1264 +                       while (len > 0) {
1265 +                               regc(*regparse++);
1266 +                               len--;
1267 +                       }
1268 +                       regc('\0');
1269 +               }
1270 +               break;
1271 +       }
1272 +
1273 +       return(ret);
1274 +}
1275 +
1276 +/*
1277 + - regnode - emit a node
1278 + */
1279 +static char *                  /* Location. */
1280 +regnode(char op)
1281 +{
1282 +       register char *ret;
1283 +       register char *ptr;
1284 +
1285 +       ret = regcode;
1286 +       if (ret == &regdummy) {
1287 +               regsize += 3;
1288 +               return(ret);
1289 +       }
1290 +
1291 +       ptr = ret;
1292 +       *ptr++ = op;
1293 +       *ptr++ = '\0';          /* Null "next" pointer. */
1294 +       *ptr++ = '\0';
1295 +       regcode = ptr;
1296 +
1297 +       return(ret);
1298 +}
1299 +
1300 +/*
1301 + - regc - emit (if appropriate) a byte of code
1302 + */
1303 +static void
1304 +regc(char b)
1305 +{
1306 +       if (regcode != &regdummy)
1307 +               *regcode++ = b;
1308 +       else
1309 +               regsize++;
1310 +}
1311 +
1312 +/*
1313 + - reginsert - insert an operator in front of already-emitted operand
1314 + *
1315 + * Means relocating the operand.
1316 + */
1317 +static void
1318 +reginsert(char op, char* opnd)
1319 +{
1320 +       register char *src;
1321 +       register char *dst;
1322 +       register char *place;
1323 +
1324 +       if (regcode == &regdummy) {
1325 +               regsize += 3;
1326 +               return;
1327 +       }
1328 +
1329 +       src = regcode;
1330 +       regcode += 3;
1331 +       dst = regcode;
1332 +       while (src > opnd)
1333 +               *--dst = *--src;
1334 +
1335 +       place = opnd;           /* Op node, where operand used to be. */
1336 +       *place++ = op;
1337 +       *place++ = '\0';
1338 +       *place++ = '\0';
1339 +}
1340 +
1341 +/*
1342 + - regtail - set the next-pointer at the end of a node chain
1343 + */
1344 +static void
1345 +regtail(char *p, char *val)
1346 +{
1347 +       register char *scan;
1348 +       register char *temp;
1349 +       register int offset;
1350 +
1351 +       if (p == &regdummy)
1352 +               return;
1353 +
1354 +       /* Find last node. */
1355 +       scan = p;
1356 +       for (;;) {
1357 +               temp = regnext(scan);
1358 +               if (temp == NULL)
1359 +                       break;
1360 +               scan = temp;
1361 +       }
1362 +
1363 +       if (OP(scan) == BACK)
1364 +               offset = scan - val;
1365 +       else
1366 +               offset = val - scan;
1367 +       *(scan+1) = (offset>>8)&0377;
1368 +       *(scan+2) = offset&0377;
1369 +}
1370 +
1371 +/*
1372 + - regoptail - regtail on operand of first argument; nop if operandless
1373 + */
1374 +static void
1375 +regoptail(char *p, char *val)
1376 +{
1377 +       /* "Operandless" and "op != BRANCH" are synonymous in practice. */
1378 +       if (p == NULL || p == &regdummy || OP(p) != BRANCH)
1379 +               return;
1380 +       regtail(OPERAND(p), val);
1381 +}
1382 +
1383 +/*
1384 + * regexec and friends
1385 + */
1386 +
1387 +/*
1388 + * Global work variables for regexec().
1389 + */
1390 +static char *reginput;         /* String-input pointer. */
1391 +static char *regbol;           /* Beginning of input, for ^ check. */
1392 +static char **regstartp;       /* Pointer to startp array. */
1393 +static char **regendp;         /* Ditto for endp. */
1394 +
1395 +/*
1396 + * Forwards.
1397 + */
1398 +STATIC int regtry(regexp *prog, char *string);
1399 +STATIC int regmatch(char *prog);
1400 +STATIC int regrepeat(char *p);
1401 +
1402 +#ifdef DEBUG
1403 +int regnarrate = 0;
1404 +void regdump();
1405 +STATIC char *regprop(char *op);
1406 +#endif
1407 +
1408 +/*
1409 + - regexec - match a regexp against a string
1410 + */
1411 +int
1412 +regexec(regexp *prog, char *string)
1413 +{
1414 +       register char *s;
1415 +
1416 +       /* Be paranoid... */
1417 +       if (prog == NULL || string == NULL) {
1418 +               printk("<3>Regexp: NULL parameter\n");
1419 +               return(0);
1420 +       }
1421 +
1422 +       /* Check validity of program. */
1423 +       if (UCHARAT(prog->program) != MAGIC) {
1424 +               printk("<3>Regexp: corrupted program\n");
1425 +               return(0);
1426 +       }
1427 +
1428 +       /* If there is a "must appear" string, look for it. */
1429 +       if (prog->regmust != NULL) {
1430 +               s = string;
1431 +               while ((s = strchr(s, prog->regmust[0])) != NULL) {
1432 +                       if (strncmp(s, prog->regmust, prog->regmlen) == 0)
1433 +                               break;  /* Found it. */
1434 +                       s++;
1435 +               }
1436 +               if (s == NULL)  /* Not present. */
1437 +                       return(0);
1438 +       }
1439 +
1440 +       /* Mark beginning of line for ^ . */
1441 +       regbol = string;
1442 +
1443 +       /* Simplest case:  anchored match need be tried only once. */
1444 +       if (prog->reganch)
1445 +               return(regtry(prog, string));
1446 +
1447 +       /* Messy cases:  unanchored match. */
1448 +       s = string;
1449 +       if (prog->regstart != '\0')
1450 +               /* We know what char it must start with. */
1451 +               while ((s = strchr(s, prog->regstart)) != NULL) {
1452 +                       if (regtry(prog, s))
1453 +                               return(1);
1454 +                       s++;
1455 +               }
1456 +       else
1457 +               /* We don't -- general case. */
1458 +               do {
1459 +                       if (regtry(prog, s))
1460 +                               return(1);
1461 +               } while (*s++ != '\0');
1462 +
1463 +       /* Failure. */
1464 +       return(0);
1465 +}
1466 +
1467 +/*
1468 + - regtry - try match at specific point
1469 + */
1470 +static int                     /* 0 failure, 1 success */
1471 +regtry(regexp *prog, char *string)
1472 +{
1473 +       register int i;
1474 +       register char **sp;
1475 +       register char **ep;
1476 +
1477 +       reginput = string;
1478 +       regstartp = prog->startp;
1479 +       regendp = prog->endp;
1480 +
1481 +       sp = prog->startp;
1482 +       ep = prog->endp;
1483 +       for (i = NSUBEXP; i > 0; i--) {
1484 +               *sp++ = NULL;
1485 +               *ep++ = NULL;
1486 +       }
1487 +       if (regmatch(prog->program + 1)) {
1488 +               prog->startp[0] = string;
1489 +               prog->endp[0] = reginput;
1490 +               return(1);
1491 +       } else
1492 +               return(0);
1493 +}
1494 +
1495 +/*
1496 + - regmatch - main matching routine
1497 + *
1498 + * Conceptually the strategy is simple:  check to see whether the current
1499 + * node matches, call self recursively to see whether the rest matches,
1500 + * and then act accordingly.  In practice we make some effort to avoid
1501 + * recursion, in particular by going through "ordinary" nodes (that don't
1502 + * need to know whether the rest of the match failed) by a loop instead of
1503 + * by recursion.
1504 + */
1505 +static int                     /* 0 failure, 1 success */
1506 +regmatch(char *prog)
1507 +{
1508 +       register char *scan = prog; /* Current node. */
1509 +       char *next;                 /* Next node. */
1510 +
1511 +#ifdef DEBUG
1512 +       if (scan != NULL && regnarrate)
1513 +               fprintf(stderr, "%s(\n", regprop(scan));
1514 +#endif
1515 +       while (scan != NULL) {
1516 +#ifdef DEBUG
1517 +               if (regnarrate)
1518 +                       fprintf(stderr, "%s...\n", regprop(scan));
1519 +#endif
1520 +               next = regnext(scan);
1521 +
1522 +               switch (OP(scan)) {
1523 +               case BOL:
1524 +                       if (reginput != regbol)
1525 +                               return(0);
1526 +                       break;
1527 +               case EOL:
1528 +                       if (*reginput != '\0')
1529 +                               return(0);
1530 +                       break;
1531 +               case ANY:
1532 +                       if (*reginput == '\0')
1533 +                               return(0);
1534 +                       reginput++;
1535 +                       break;
1536 +               case EXACTLY: {
1537 +                               register int len;
1538 +                               register char *opnd;
1539 +
1540 +                               opnd = OPERAND(scan);
1541 +                               /* Inline the first character, for speed. */
1542 +                               if (*opnd != *reginput)
1543 +                                       return(0);
1544 +                               len = strlen(opnd);
1545 +                               if (len > 1 && strncmp(opnd, reginput, len) != 0)
1546 +                                       return(0);
1547 +                               reginput += len;
1548 +                       }
1549 +                       break;
1550 +               case ANYOF:
1551 +                       if (*reginput == '\0' || strchr(OPERAND(scan), *reginput) == NULL)
1552 +                               return(0);
1553 +                       reginput++;
1554 +                       break;
1555 +               case ANYBUT:
1556 +                       if (*reginput == '\0' || strchr(OPERAND(scan), *reginput) != NULL)
1557 +                               return(0);
1558 +                       reginput++;
1559 +                       break;
1560 +               case NOTHING:
1561 +               case BACK:
1562 +                       break;
1563 +               case OPEN+1:
1564 +               case OPEN+2:
1565 +               case OPEN+3:
1566 +               case OPEN+4:
1567 +               case OPEN+5:
1568 +               case OPEN+6:
1569 +               case OPEN+7:
1570 +               case OPEN+8:
1571 +               case OPEN+9: {
1572 +                               register int no;
1573 +                               register char *save;
1574 +
1575 +                               no = OP(scan) - OPEN;
1576 +                               save = reginput;
1577 +
1578 +                               if (regmatch(next)) {
1579 +                                       /*
1580 +                                        * Don't set startp if some later
1581 +                                        * invocation of the same parentheses
1582 +                                        * already has.
1583 +                                        */
1584 +                                       if (regstartp[no] == NULL)
1585 +                                               regstartp[no] = save;
1586 +                                       return(1);
1587 +                               } else
1588 +                                       return(0);
1589 +                       }
1590 +                       break;
1591 +               case CLOSE+1:
1592 +               case CLOSE+2:
1593 +               case CLOSE+3:
1594 +               case CLOSE+4:
1595 +               case CLOSE+5:
1596 +               case CLOSE+6:
1597 +               case CLOSE+7:
1598 +               case CLOSE+8:
1599 +               case CLOSE+9:
1600 +                       {
1601 +                               register int no;
1602 +                               register char *save;
1603 +
1604 +                               no = OP(scan) - CLOSE;
1605 +                               save = reginput;
1606 +
1607 +                               if (regmatch(next)) {
1608 +                                       /*
1609 +                                        * Don't set endp if some later
1610 +                                        * invocation of the same parentheses
1611 +                                        * already has.
1612 +                                        */
1613 +                                       if (regendp[no] == NULL)
1614 +                                               regendp[no] = save;
1615 +                                       return(1);
1616 +                               } else
1617 +                                       return(0);
1618 +                       }
1619 +                       break;
1620 +               case BRANCH: {
1621 +                               register char *save;
1622 +
1623 +                               if (OP(next) != BRANCH)         /* No choice. */
1624 +                                       next = OPERAND(scan);   /* Avoid recursion. */
1625 +                               else {
1626 +                                       do {
1627 +                                               save = reginput;
1628 +                                               if (regmatch(OPERAND(scan)))
1629 +                                                       return(1);
1630 +                                               reginput = save;
1631 +                                               scan = regnext(scan);
1632 +                                       } while (scan != NULL && OP(scan) == BRANCH);
1633 +                                       return(0);
1634 +                                       /* NOTREACHED */
1635 +                               }
1636 +                       }
1637 +                       break;
1638 +               case STAR:
1639 +               case PLUS: {
1640 +                               register char nextch;
1641 +                               register int no;
1642 +                               register char *save;
1643 +                               register int min;
1644 +
1645 +                               /*
1646 +                                * Lookahead to avoid useless match attempts
1647 +                                * when we know what character comes next.
1648 +                                */
1649 +                               nextch = '\0';
1650 +                               if (OP(next) == EXACTLY)
1651 +                                       nextch = *OPERAND(next);
1652 +                               min = (OP(scan) == STAR) ? 0 : 1;
1653 +                               save = reginput;
1654 +                               no = regrepeat(OPERAND(scan));
1655 +                               while (no >= min) {
1656 +                                       /* If it could work, try it. */
1657 +                                       if (nextch == '\0' || *reginput == nextch)
1658 +                                               if (regmatch(next))
1659 +                                                       return(1);
1660 +                                       /* Couldn't or didn't -- back up. */
1661 +                                       no--;
1662 +                                       reginput = save + no;
1663 +                               }
1664 +                               return(0);
1665 +                       }
1666 +                       break;
1667 +               case END:
1668 +                       return(1);      /* Success! */
1669 +                       break;
1670 +               default:
1671 +                       printk("<3>Regexp: memory corruption\n");
1672 +                       return(0);
1673 +                       break;
1674 +               }
1675 +
1676 +               scan = next;
1677 +       }
1678 +
1679 +       /*
1680 +        * We get here only if there's trouble -- normally "case END" is
1681 +        * the terminating point.
1682 +        */
1683 +       printk("<3>Regexp: corrupted pointers\n");
1684 +       return(0);
1685 +}
1686 +
1687 +/*
1688 + - regrepeat - repeatedly match something simple, report how many
1689 + */
1690 +static int
1691 +regrepeat(char *p)
1692 +{
1693 +       register int count = 0;
1694 +       register char *scan;
1695 +       register char *opnd;
1696 +
1697 +       scan = reginput;
1698 +       opnd = OPERAND(p);
1699 +       switch (OP(p)) {
1700 +       case ANY:
1701 +               count = strlen(scan);
1702 +               scan += count;
1703 +               break;
1704 +       case EXACTLY:
1705 +               while (*opnd == *scan) {
1706 +                       count++;
1707 +                       scan++;
1708 +               }
1709 +               break;
1710 +       case ANYOF:
1711 +               while (*scan != '\0' && strchr(opnd, *scan) != NULL) {
1712 +                       count++;
1713 +                       scan++;
1714 +               }
1715 +               break;
1716 +       case ANYBUT:
1717 +               while (*scan != '\0' && strchr(opnd, *scan) == NULL) {
1718 +                       count++;
1719 +                       scan++;
1720 +               }
1721 +               break;
1722 +       default:                /* Oh dear.  Called inappropriately. */
1723 +               printk("<3>Regexp: internal foulup\n");
1724 +               count = 0;      /* Best compromise. */
1725 +               break;
1726 +       }
1727 +       reginput = scan;
1728 +
1729 +       return(count);
1730 +}
1731 +
1732 +/*
1733 + - regnext - dig the "next" pointer out of a node
1734 + */
1735 +static char*
1736 +regnext(char *p)
1737 +{
1738 +       register int offset;
1739 +
1740 +       if (p == &regdummy)
1741 +               return(NULL);
1742 +
1743 +       offset = NEXT(p);
1744 +       if (offset == 0)
1745 +               return(NULL);
1746 +
1747 +       if (OP(p) == BACK)
1748 +               return(p-offset);
1749 +       else
1750 +               return(p+offset);
1751 +}
1752 +
1753 +#ifdef DEBUG
1754 +
1755 +STATIC char *regprop();
1756 +
1757 +/*
1758 + - regdump - dump a regexp onto stdout in vaguely comprehensible form
1759 + */
1760 +void
1761 +regdump(regexp *r)
1762 +{
1763 +       register char *s;
1764 +       register char op = EXACTLY;     /* Arbitrary non-END op. */
1765 +       register char *next;
1766 +       /* extern char *strchr(); */
1767 +
1768 +
1769 +       s = r->program + 1;
1770 +       while (op != END) {     /* While that wasn't END last time... */
1771 +               op = OP(s);
1772 +               printf("%2d%s", s-r->program, regprop(s));      /* Where, what. */
1773 +               next = regnext(s);
1774 +               if (next == NULL)               /* Next ptr. */
1775 +                       printf("(0)");
1776 +               else
1777 +                       printf("(%d)", (s-r->program)+(next-s));
1778 +               s += 3;
1779 +               if (op == ANYOF || op == ANYBUT || op == EXACTLY) {
1780 +                       /* Literal string, where present. */
1781 +                       while (*s != '\0') {
1782 +                               putchar(*s);
1783 +                               s++;
1784 +                       }
1785 +                       s++;
1786 +               }
1787 +               putchar('\n');
1788 +       }
1789 +
1790 +       /* Header fields of interest. */
1791 +       if (r->regstart != '\0')
1792 +               printf("start `%c' ", r->regstart);
1793 +       if (r->reganch)
1794 +               printf("anchored ");
1795 +       if (r->regmust != NULL)
1796 +               printf("must have \"%s\"", r->regmust);
1797 +       printf("\n");
1798 +}
1799 +
1800 +/*
1801 + - regprop - printable representation of opcode
1802 + */
1803 +static char *
1804 +regprop(char *op)
1805 +{
1806 +#define BUFLEN 50
1807 +       register char *p;
1808 +       static char buf[BUFLEN];
1809 +
1810 +       strcpy(buf, ":");
1811 +
1812 +       switch (OP(op)) {
1813 +       case BOL:
1814 +               p = "BOL";
1815 +               break;
1816 +       case EOL:
1817 +               p = "EOL";
1818 +               break;
1819 +       case ANY:
1820 +               p = "ANY";
1821 +               break;
1822 +       case ANYOF:
1823 +               p = "ANYOF";
1824 +               break;
1825 +       case ANYBUT:
1826 +               p = "ANYBUT";
1827 +               break;
1828 +       case BRANCH:
1829 +               p = "BRANCH";
1830 +               break;
1831 +       case EXACTLY:
1832 +               p = "EXACTLY";
1833 +               break;
1834 +       case NOTHING:
1835 +               p = "NOTHING";
1836 +               break;
1837 +       case BACK:
1838 +               p = "BACK";
1839 +               break;
1840 +       case END:
1841 +               p = "END";
1842 +               break;
1843 +       case OPEN+1:
1844 +       case OPEN+2:
1845 +       case OPEN+3:
1846 +       case OPEN+4:
1847 +       case OPEN+5:
1848 +       case OPEN+6:
1849 +       case OPEN+7:
1850 +       case OPEN+8:
1851 +       case OPEN+9:
1852 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "OPEN%d", OP(op)-OPEN);
1853 +               p = NULL;
1854 +               break;
1855 +       case CLOSE+1:
1856 +       case CLOSE+2:
1857 +       case CLOSE+3:
1858 +       case CLOSE+4:
1859 +       case CLOSE+5:
1860 +       case CLOSE+6:
1861 +       case CLOSE+7:
1862 +       case CLOSE+8:
1863 +       case CLOSE+9:
1864 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "CLOSE%d", OP(op)-CLOSE);
1865 +               p = NULL;
1866 +               break;
1867 +       case STAR:
1868 +               p = "STAR";
1869 +               break;
1870 +       case PLUS:
1871 +               p = "PLUS";
1872 +               break;
1873 +       default:
1874 +               printk("<3>Regexp: corrupted opcode\n");
1875 +               break;
1876 +       }
1877 +       if (p != NULL)
1878 +               strncat(buf, p, BUFLEN-strlen(buf));
1879 +       return(buf);
1880 +}
1881 +#endif
1882 +
1883 +
1884 diff -Nurp linux-2.4.26-stock/net/ipv4/netfilter/regexp/regexp.h linux-2.4.26-layer7-clean/net/ipv4/netfilter/regexp/regexp.h
1885 --- linux-2.4.26-stock/net/ipv4/netfilter/regexp/regexp.h       1969-12-31 18:00:00.000000000 -0600
1886 +++ linux-2.4.26-layer7-clean/net/ipv4/netfilter/regexp/regexp.h        2004-06-27 19:07:00.000000000 -0500
1887 @@ -0,0 +1,27 @@
1888 +/*
1889 + * Definitions etc. for regexp(3) routines.
1890 + *
1891 + * Caveat:  this is V8 regexp(3) [actually, a reimplementation thereof],
1892 + * not the System V one.
1893 + */
1894 +
1895 +#ifndef REGEXP_H
1896 +#define REGEXP_H
1897 +
1898 +#define NSUBEXP  10
1899 +typedef struct regexp {
1900 +       char *startp[NSUBEXP];
1901 +       char *endp[NSUBEXP];
1902 +       char regstart;          /* Internal use only. */
1903 +       char reganch;           /* Internal use only. */
1904 +       char *regmust;          /* Internal use only. */
1905 +       int regmlen;            /* Internal use only. */
1906 +       char program[1];        /* Unwarranted chumminess with compiler. */
1907 +} regexp;
1908 +
1909 +regexp * regcomp(char *exp, int *patternsize);
1910 +int regexec(regexp *prog, char *string);
1911 +void regsub(regexp *prog, char *source, char *dest);
1912 +void regerror(char *s);
1913 +
1914 +#endif
1915 diff -Nurp linux-2.4.26-stock/net/ipv4/netfilter/regexp/regmagic.h linux-2.4.26-layer7-clean/net/ipv4/netfilter/regexp/regmagic.h
1916 --- linux-2.4.26-stock/net/ipv4/netfilter/regexp/regmagic.h     1969-12-31 18:00:00.000000000 -0600
1917 +++ linux-2.4.26-layer7-clean/net/ipv4/netfilter/regexp/regmagic.h      2004-06-27 19:07:00.000000000 -0500
1918 @@ -0,0 +1,5 @@
1919 +/*
1920 + * The first byte of the regexp internal "program" is actually this magic
1921 + * number; the start node begins in the second byte.
1922 + */
1923 +#define        MAGIC   0234
1924 diff -Nurp linux-2.4.26-stock/net/ipv4/netfilter/regexp/regsub.c linux-2.4.26-layer7-clean/net/ipv4/netfilter/regexp/regsub.c
1925 --- linux-2.4.26-stock/net/ipv4/netfilter/regexp/regsub.c       1969-12-31 18:00:00.000000000 -0600
1926 +++ linux-2.4.26-layer7-clean/net/ipv4/netfilter/regexp/regsub.c        2004-06-27 19:07:00.000000000 -0500
1927 @@ -0,0 +1,95 @@
1928 +/*
1929 + * regsub
1930 + * @(#)regsub.c        1.3 of 2 April 86
1931 + *
1932 + *     Copyright (c) 1986 by University of Toronto.
1933 + *     Written by Henry Spencer.  Not derived from licensed software.
1934 + *
1935 + *     Permission is granted to anyone to use this software for any
1936 + *     purpose on any computer system, and to redistribute it freely,
1937 + *     subject to the following restrictions:
1938 + *
1939 + *     1. The author is not responsible for the consequences of use of
1940 + *             this software, no matter how awful, even if they arise
1941 + *             from defects in it.
1942 + *
1943 + *     2. The origin of this software must not be misrepresented, either
1944 + *             by explicit claim or by omission.
1945 + *
1946 + *     3. Altered versions must be plainly marked as such, and must not
1947 + *             be misrepresented as being the original software.
1948 + *
1949 + *
1950 + * This code was modified by Ethan Sommer to work within the kernel
1951 + * (it now uses kmalloc etc..)
1952 + *
1953 + */
1954 +#include "regexp.h"
1955 +#include "regmagic.h"
1956 +#include <linux/string.h>
1957 +
1958 +
1959 +#ifndef CHARBITS
1960 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
1961 +#else
1962 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
1963 +#endif
1964 +
1965 +#if 0
1966 +//void regerror(char * s)
1967 +//{
1968 +//        printk("regexp(3): %s", s);
1969 +//        /* NOTREACHED */
1970 +//}
1971 +#endif
1972 +
1973 +/*
1974 + - regsub - perform substitutions after a regexp match
1975 + */
1976 +void
1977 +regsub(regexp * prog, char * source, char * dest)
1978 +{
1979 +       register char *src;
1980 +       register char *dst;
1981 +       register char c;
1982 +       register int no;
1983 +       register int len;
1984 +
1985 +       /* Not necessary and gcc doesn't like it -MLS */
1986 +       /*extern char *strncpy();*/
1987 +
1988 +       if (prog == NULL || source == NULL || dest == NULL) {
1989 +               regerror("NULL parm to regsub");
1990 +               return;
1991 +       }
1992 +       if (UCHARAT(prog->program) != MAGIC) {
1993 +               regerror("damaged regexp fed to regsub");
1994 +               return;
1995 +       }
1996 +
1997 +       src = source;
1998 +       dst = dest;
1999 +       while ((c = *src++) != '\0') {
2000 +               if (c == '&')
2001 +                       no = 0;
2002 +               else if (c == '\\' && '0' <= *src && *src <= '9')
2003 +                       no = *src++ - '0';
2004 +               else
2005 +                       no = -1;
2006 +
2007 +               if (no < 0) {   /* Ordinary character. */
2008 +                       if (c == '\\' && (*src == '\\' || *src == '&'))
2009 +                               c = *src++;
2010 +                       *dst++ = c;
2011 +               } else if (prog->startp[no] != NULL && prog->endp[no] != NULL) {
2012 +                       len = prog->endp[no] - prog->startp[no];
2013 +                       (void) strncpy(dst, prog->startp[no], len);
2014 +                       dst += len;
2015 +                       if (len != 0 && *(dst-1) == '\0') {     /* strncpy hit NUL. */
2016 +                               regerror("damaged match string");
2017 +                               return;
2018 +                       }
2019 +               }
2020 +       }
2021 +       *dst++ = '\0';
2022 +}