target/linux/generic-2.6/patches-2.6.23/100-netfilter_layer7_2.21.patch

   1 --- /dev/null
   2 +++ b/include/linux/netfilter/xt_layer7.h
   3 @@ -0,0 +1,13 @@
   4 +#ifndef _XT_LAYER7_H
   5 +#define _XT_LAYER7_H
   6 +
   7 +#define MAX_PATTERN_LEN 8192
   8 +#define MAX_PROTOCOL_LEN 256
   9 +
  10 +struct xt_layer7_info {
  11 +    char protocol[MAX_PROTOCOL_LEN];
  12 +    char pattern[MAX_PATTERN_LEN];
  13 +    u_int8_t invert;
  14 +};
  15 +
  16 +#endif /* _XT_LAYER7_H */
  17 --- a/include/net/netfilter/nf_conntrack.h
  18 +++ b/include/net/netfilter/nf_conntrack.h
  19 @@ -127,6 +127,22 @@
  20         u_int32_t secmark;
  21  #endif
  22
  23 +#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || \
  24 +    defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
  25 +       struct {
  26 +               /*
  27 +                * e.g. "http". NULL before decision. "unknown" after decision
  28 +                * if no match.
  29 +                */
  30 +               char *app_proto;
  31 +               /*
  32 +                * application layer data so far. NULL after match decision.
  33 +                */
  34 +               char *app_data;
  35 +               unsigned int app_data_len;
  36 +       } layer7;
  37 +#endif
  38 +
  39         /* Storage reserved for other modules: */
  40         union nf_conntrack_proto proto;
  41
  42 --- a/net/netfilter/Kconfig
  43 +++ b/net/netfilter/Kconfig
  44 @@ -633,6 +633,27 @@
  45
  46           To compile it as a module, choose M here.  If unsure, say N.
  47
  48 +config NETFILTER_XT_MATCH_LAYER7
  49 +       tristate '"layer7" match support'
  50 +       depends on NETFILTER_XTABLES
  51 +       depends on EXPERIMENTAL && (IP_NF_CONNTRACK || NF_CONNTRACK)
  52 +       depends on NF_CT_ACCT
  53 +       help
  54 +         Say Y if you want to be able to classify connections (and their
  55 +         packets) based on regular expression matching of their application
  56 +         layer data.   This is one way to classify applications such as
  57 +         peer-to-peer filesharing systems that do not always use the same
  58 +         port.
  59 +
  60 +         To compile it as a module, choose M here.  If unsure, say N.
  61 +
  62 +config NETFILTER_XT_MATCH_LAYER7_DEBUG
  63 +        bool 'Layer 7 debugging output'
  64 +        depends on NETFILTER_XT_MATCH_LAYER7
  65 +        help
  66 +          Say Y to get lots of debugging output.
  67 +
  68 +
  69  config NETFILTER_XT_MATCH_STATISTIC
  70         tristate '"statistic" match support'
  71         depends on NETFILTER_XTABLES
  72 --- a/net/netfilter/Makefile
  73 +++ b/net/netfilter/Makefile
  74 @@ -71,6 +71,7 @@
  75  obj-$(CONFIG_NETFILTER_XT_MATCH_REALM) += xt_realm.o
  76  obj-$(CONFIG_NETFILTER_XT_MATCH_SCTP) += xt_sctp.o
  77  obj-$(CONFIG_NETFILTER_XT_MATCH_STATE) += xt_state.o
  78 +obj-$(CONFIG_NETFILTER_XT_MATCH_LAYER7) += xt_layer7.o
  79  obj-$(CONFIG_NETFILTER_XT_MATCH_STATISTIC) += xt_statistic.o
  80  obj-$(CONFIG_NETFILTER_XT_MATCH_STRING) += xt_string.o
  81  obj-$(CONFIG_NETFILTER_XT_MATCH_TCPMSS) += xt_tcpmss.o
  82 --- a/net/netfilter/nf_conntrack_core.c
  83 +++ b/net/netfilter/nf_conntrack_core.c
  84 @@ -207,6 +207,14 @@
  85          * too. */
  86         nf_ct_remove_expectations(ct);
  87
  88 +       #if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
  89 +       if(ct->layer7.app_proto)
  90 +               kfree(ct->layer7.app_proto);
  91 +       if(ct->layer7.app_data)
  92 +       kfree(ct->layer7.app_data);
  93 +       #endif
  94 +
  95 +
  96         /* We overload first tuple to link into unconfirmed list. */
  97         if (!nf_ct_is_confirmed(ct)) {
  98                 BUG_ON(hlist_unhashed(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnode));
  99 --- a/net/netfilter/nf_conntrack_standalone.c
 100 +++ b/net/netfilter/nf_conntrack_standalone.c
 101 @@ -179,6 +179,11 @@
 102                 return -ENOSPC;
 103  #endif
 104
 105 +#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
 106 +       if(conntrack->layer7.app_proto)
 107 +               if(seq_printf(s, "l7proto=%s ", conntrack->layer7.app_proto))
 108 +                       return -ENOSPC;
 109 +#endif
 110         if (seq_printf(s, "use=%u\n", atomic_read(&conntrack->ct_general.use)))
 111                 return -ENOSPC;
 112
 113 --- /dev/null
 114 +++ b/net/netfilter/regexp/regexp.c
 115 @@ -0,0 +1,1197 @@
 116 +/*
 117 + * regcomp and regexec -- regsub and regerror are elsewhere
 118 + * @(#)regexp.c        1.3 of 18 April 87
 119 + *
 120 + *     Copyright (c) 1986 by University of Toronto.
 121 + *     Written by Henry Spencer.  Not derived from licensed software.
 122 + *
 123 + *     Permission is granted to anyone to use this software for any
 124 + *     purpose on any computer system, and to redistribute it freely,
 125 + *     subject to the following restrictions:
 126 + *
 127 + *     1. The author is not responsible for the consequences of use of
 128 + *             this software, no matter how awful, even if they arise
 129 + *             from defects in it.
 130 + *
 131 + *     2. The origin of this software must not be misrepresented, either
 132 + *             by explicit claim or by omission.
 133 + *
 134 + *     3. Altered versions must be plainly marked as such, and must not
 135 + *             be misrepresented as being the original software.
 136 + *
 137 + * Beware that some of this code is subtly aware of the way operator
 138 + * precedence is structured in regular expressions.  Serious changes in
 139 + * regular-expression syntax might require a total rethink.
 140 + *
 141 + * This code was modified by Ethan Sommer to work within the kernel
 142 + * (it now uses kmalloc etc..)
 143 + *
 144 + * Modified slightly by Matthew Strait to use more modern C.
 145 + */
 146 +
 147 +#include "regexp.h"
 148 +#include "regmagic.h"
 149 +
 150 +/* added by ethan and matt.  Lets it work in both kernel and user space.
 151 +(So iptables can use it, for instance.)  Yea, it goes both ways... */
 152 +#if __KERNEL__
 153 +  #define malloc(foo) kmalloc(foo,GFP_ATOMIC)
 154 +#else
 155 +  #define printk(format,args...) printf(format,##args)
 156 +#endif
 157 +
 158 +void regerror(char * s)
 159 +{
 160 +        printk("<3>Regexp: %s\n", s);
 161 +        /* NOTREACHED */
 162 +}
 163 +
 164 +/*
 165 + * The "internal use only" fields in regexp.h are present to pass info from
 166 + * compile to execute that permits the execute phase to run lots faster on
 167 + * simple cases.  They are:
 168 + *
 169 + * regstart    char that must begin a match; '\0' if none obvious
 170 + * reganch     is the match anchored (at beginning-of-line only)?
 171 + * regmust     string (pointer into program) that match must include, or NULL
 172 + * regmlen     length of regmust string
 173 + *
 174 + * Regstart and reganch permit very fast decisions on suitable starting points
 175 + * for a match, cutting down the work a lot.  Regmust permits fast rejection
 176 + * of lines that cannot possibly match.  The regmust tests are costly enough
 177 + * that regcomp() supplies a regmust only if the r.e. contains something
 178 + * potentially expensive (at present, the only such thing detected is * or +
 179 + * at the start of the r.e., which can involve a lot of backup).  Regmlen is
 180 + * supplied because the test in regexec() needs it and regcomp() is computing
 181 + * it anyway.
 182 + */
 183 +
 184 +/*
 185 + * Structure for regexp "program".  This is essentially a linear encoding
 186 + * of a nondeterministic finite-state machine (aka syntax charts or
 187 + * "railroad normal form" in parsing technology).  Each node is an opcode
 188 + * plus a "next" pointer, possibly plus an operand.  "Next" pointers of
 189 + * all nodes except BRANCH implement concatenation; a "next" pointer with
 190 + * a BRANCH on both ends of it is connecting two alternatives.  (Here we
 191 + * have one of the subtle syntax dependencies:  an individual BRANCH (as
 192 + * opposed to a collection of them) is never concatenated with anything
 193 + * because of operator precedence.)  The operand of some types of node is
 194 + * a literal string; for others, it is a node leading into a sub-FSM.  In
 195 + * particular, the operand of a BRANCH node is the first node of the branch.
 196 + * (NB this is *not* a tree structure:  the tail of the branch connects
 197 + * to the thing following the set of BRANCHes.)  The opcodes are:
 198 + */
 199 +
 200 +/* definition  number  opnd?   meaning */
 201 +#define        END     0       /* no   End of program. */
 202 +#define        BOL     1       /* no   Match "" at beginning of line. */
 203 +#define        EOL     2       /* no   Match "" at end of line. */
 204 +#define        ANY     3       /* no   Match any one character. */
 205 +#define        ANYOF   4       /* str  Match any character in this string. */
 206 +#define        ANYBUT  5       /* str  Match any character not in this string. */
 207 +#define        BRANCH  6       /* node Match this alternative, or the next... */
 208 +#define        BACK    7       /* no   Match "", "next" ptr points backward. */
 209 +#define        EXACTLY 8       /* str  Match this string. */
 210 +#define        NOTHING 9       /* no   Match empty string. */
 211 +#define        STAR    10      /* node Match this (simple) thing 0 or more times. */
 212 +#define        PLUS    11      /* node Match this (simple) thing 1 or more times. */
 213 +#define        OPEN    20      /* no   Mark this point in input as start of #n. */
 214 +                       /*      OPEN+1 is number 1, etc. */
 215 +#define        CLOSE   30      /* no   Analogous to OPEN. */
 216 +
 217 +/*
 218 + * Opcode notes:
 219 + *
 220 + * BRANCH      The set of branches constituting a single choice are hooked
 221 + *             together with their "next" pointers, since precedence prevents
 222 + *             anything being concatenated to any individual branch.  The
 223 + *             "next" pointer of the last BRANCH in a choice points to the
 224 + *             thing following the whole choice.  This is also where the
 225 + *             final "next" pointer of each individual branch points; each
 226 + *             branch starts with the operand node of a BRANCH node.
 227 + *
 228 + * BACK                Normal "next" pointers all implicitly point forward; BACK
 229 + *             exists to make loop structures possible.
 230 + *
 231 + * STAR,PLUS   '?', and complex '*' and '+', are implemented as circular
 232 + *             BRANCH structures using BACK.  Simple cases (one character
 233 + *             per match) are implemented with STAR and PLUS for speed
 234 + *             and to minimize recursive plunges.
 235 + *
 236 + * OPEN,CLOSE  ...are numbered at compile time.
 237 + */
 238 +
 239 +/*
 240 + * A node is one char of opcode followed by two chars of "next" pointer.
 241 + * "Next" pointers are stored as two 8-bit pieces, high order first.  The
 242 + * value is a positive offset from the opcode of the node containing it.
 243 + * An operand, if any, simply follows the node.  (Note that much of the
 244 + * code generation knows about this implicit relationship.)
 245 + *
 246 + * Using two bytes for the "next" pointer is vast overkill for most things,
 247 + * but allows patterns to get big without disasters.
 248 + */
 249 +#define        OP(p)   (*(p))
 250 +#define        NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377))
 251 +#define        OPERAND(p)      ((p) + 3)
 252 +
 253 +/*
 254 + * See regmagic.h for one further detail of program structure.
 255 + */
 256 +
 257 +
 258 +/*
 259 + * Utility definitions.
 260 + */
 261 +#ifndef CHARBITS
 262 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
 263 +#else
 264 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
 265 +#endif
 266 +
 267 +#define        FAIL(m) { regerror(m); return(NULL); }
 268 +#define        ISMULT(c)       ((c) == '*' || (c) == '+' || (c) == '?')
 269 +#define        META    "^$.[()|?+*\\"
 270 +
 271 +/*
 272 + * Flags to be passed up and down.
 273 + */
 274 +#define        HASWIDTH        01      /* Known never to match null string. */
 275 +#define        SIMPLE          02      /* Simple enough to be STAR/PLUS operand. */
 276 +#define        SPSTART         04      /* Starts with * or +. */
 277 +#define        WORST           0       /* Worst case. */
 278 +
 279 +/*
 280 + * Global work variables for regcomp().
 281 + */
 282 +struct match_globals {
 283 +char *reginput;                /* String-input pointer. */
 284 +char *regbol;          /* Beginning of input, for ^ check. */
 285 +char **regstartp;      /* Pointer to startp array. */
 286 +char **regendp;                /* Ditto for endp. */
 287 +char *regparse;                /* Input-scan pointer. */
 288 +int regnpar;           /* () count. */
 289 +char regdummy;
 290 +char *regcode;         /* Code-emit pointer; &regdummy = don't. */
 291 +long regsize;          /* Code size. */
 292 +};
 293 +
 294 +/*
 295 + * Forward declarations for regcomp()'s friends.
 296 + */
 297 +#ifndef STATIC
 298 +#define        STATIC  static
 299 +#endif
 300 +STATIC char *reg(struct match_globals *g, int paren,int *flagp);
 301 +STATIC char *regbranch(struct match_globals *g, int *flagp);
 302 +STATIC char *regpiece(struct match_globals *g, int *flagp);
 303 +STATIC char *regatom(struct match_globals *g, int *flagp);
 304 +STATIC char *regnode(struct match_globals *g, char op);
 305 +STATIC char *regnext(struct match_globals *g, char *p);
 306 +STATIC void regc(struct match_globals *g, char b);
 307 +STATIC void reginsert(struct match_globals *g, char op, char *opnd);
 308 +STATIC void regtail(struct match_globals *g, char *p, char *val);
 309 +STATIC void regoptail(struct match_globals *g, char *p, char *val);
 310 +
 311 +
 312 +__kernel_size_t my_strcspn(const char *s1,const char *s2)
 313 +{
 314 +        char *scan1;
 315 +        char *scan2;
 316 +        int count;
 317 +
 318 +        count = 0;
 319 +        for (scan1 = (char *)s1; *scan1 != '\0'; scan1++) {
 320 +                for (scan2 = (char *)s2; *scan2 != '\0';)       /* ++ moved down. */
 321 +                        if (*scan1 == *scan2++)
 322 +                                return(count);
 323 +                count++;
 324 +        }
 325 +        return(count);
 326 +}
 327 +
 328 +/*
 329 + - regcomp - compile a regular expression into internal code
 330 + *
 331 + * We can't allocate space until we know how big the compiled form will be,
 332 + * but we can't compile it (and thus know how big it is) until we've got a
 333 + * place to put the code.  So we cheat:  we compile it twice, once with code
 334 + * generation turned off and size counting turned on, and once "for real".
 335 + * This also means that we don't allocate space until we are sure that the
 336 + * thing really will compile successfully, and we never have to move the
 337 + * code and thus invalidate pointers into it.  (Note that it has to be in
 338 + * one piece because free() must be able to free it all.)
 339 + *
 340 + * Beware that the optimization-preparation code in here knows about some
 341 + * of the structure of the compiled regexp.
 342 + */
 343 +regexp *
 344 +regcomp(char *exp,int *patternsize)
 345 +{
 346 +       register regexp *r;
 347 +       register char *scan;
 348 +       register char *longest;
 349 +       register int len;
 350 +       int flags;
 351 +       struct match_globals g;
 352 +
 353 +       /* commented out by ethan
 354 +          extern char *malloc();
 355 +       */
 356 +
 357 +       if (exp == NULL)
 358 +               FAIL("NULL argument");
 359 +
 360 +       /* First pass: determine size, legality. */
 361 +       g.regparse = exp;
 362 +       g.regnpar = 1;
 363 +       g.regsize = 0L;
 364 +       g.regcode = &g.regdummy;
 365 +       regc(&g, MAGIC);
 366 +       if (reg(&g, 0, &flags) == NULL)
 367 +               return(NULL);
 368 +
 369 +       /* Small enough for pointer-storage convention? */
 370 +       if (g.regsize >= 32767L)                /* Probably could be 65535L. */
 371 +               FAIL("regexp too big");
 372 +
 373 +       /* Allocate space. */
 374 +       *patternsize=sizeof(regexp) + (unsigned)g.regsize;
 375 +       r = (regexp *)malloc(sizeof(regexp) + (unsigned)g.regsize);
 376 +       if (r == NULL)
 377 +               FAIL("out of space");
 378 +
 379 +       /* Second pass: emit code. */
 380 +       g.regparse = exp;
 381 +       g.regnpar = 1;
 382 +       g.regcode = r->program;
 383 +       regc(&g, MAGIC);
 384 +       if (reg(&g, 0, &flags) == NULL)
 385 +               return(NULL);
 386 +
 387 +       /* Dig out information for optimizations. */
 388 +       r->regstart = '\0';     /* Worst-case defaults. */
 389 +       r->reganch = 0;
 390 +       r->regmust = NULL;
 391 +       r->regmlen = 0;
 392 +       scan = r->program+1;                    /* First BRANCH. */
 393 +       if (OP(regnext(&g, scan)) == END) {             /* Only one top-level choice. */
 394 +               scan = OPERAND(scan);
 395 +
 396 +               /* Starting-point info. */
 397 +               if (OP(scan) == EXACTLY)
 398 +                       r->regstart = *OPERAND(scan);
 399 +               else if (OP(scan) == BOL)
 400 +                       r->reganch++;
 401 +
 402 +               /*
 403 +                * If there's something expensive in the r.e., find the
 404 +                * longest literal string that must appear and make it the
 405 +                * regmust.  Resolve ties in favor of later strings, since
 406 +                * the regstart check works with the beginning of the r.e.
 407 +                * and avoiding duplication strengthens checking.  Not a
 408 +                * strong reason, but sufficient in the absence of others.
 409 +                */
 410 +               if (flags&SPSTART) {
 411 +                       longest = NULL;
 412 +                       len = 0;
 413 +                       for (; scan != NULL; scan = regnext(&g, scan))
 414 +                               if (OP(scan) == EXACTLY && strlen(OPERAND(scan)) >= len) {
 415 +                                       longest = OPERAND(scan);
 416 +                                       len = strlen(OPERAND(scan));
 417 +                               }
 418 +                       r->regmust = longest;
 419 +                       r->regmlen = len;
 420 +               }
 421 +       }
 422 +
 423 +       return(r);
 424 +}
 425 +
 426 +/*
 427 + - reg - regular expression, i.e. main body or parenthesized thing
 428 + *
 429 + * Caller must absorb opening parenthesis.
 430 + *
 431 + * Combining parenthesis handling with the base level of regular expression
 432 + * is a trifle forced, but the need to tie the tails of the branches to what
 433 + * follows makes it hard to avoid.
 434 + */
 435 +static char *
 436 +reg(struct match_globals *g, int paren, int *flagp /* Parenthesized? */ )
 437 +{
 438 +       register char *ret;
 439 +       register char *br;
 440 +       register char *ender;
 441 +       register int parno = 0; /* 0 makes gcc happy */
 442 +       int flags;
 443 +
 444 +       *flagp = HASWIDTH;      /* Tentatively. */
 445 +
 446 +       /* Make an OPEN node, if parenthesized. */
 447 +       if (paren) {
 448 +               if (g->regnpar >= NSUBEXP)
 449 +                       FAIL("too many ()");
 450 +               parno = g->regnpar;
 451 +               g->regnpar++;
 452 +               ret = regnode(g, OPEN+parno);
 453 +       } else
 454 +               ret = NULL;
 455 +
 456 +       /* Pick up the branches, linking them together. */
 457 +       br = regbranch(g, &flags);
 458 +       if (br == NULL)
 459 +               return(NULL);
 460 +       if (ret != NULL)
 461 +               regtail(g, ret, br);    /* OPEN -> first. */
 462 +       else
 463 +               ret = br;
 464 +       if (!(flags&HASWIDTH))
 465 +               *flagp &= ~HASWIDTH;
 466 +       *flagp |= flags&SPSTART;
 467 +       while (*g->regparse == '|') {
 468 +               g->regparse++;
 469 +               br = regbranch(g, &flags);
 470 +               if (br == NULL)
 471 +                       return(NULL);
 472 +               regtail(g, ret, br);    /* BRANCH -> BRANCH. */
 473 +               if (!(flags&HASWIDTH))
 474 +                       *flagp &= ~HASWIDTH;
 475 +               *flagp |= flags&SPSTART;
 476 +       }
 477 +
 478 +       /* Make a closing node, and hook it on the end. */
 479 +       ender = regnode(g, (paren) ? CLOSE+parno : END);
 480 +       regtail(g, ret, ender);
 481 +
 482 +       /* Hook the tails of the branches to the closing node. */
 483 +       for (br = ret; br != NULL; br = regnext(g, br))
 484 +               regoptail(g, br, ender);
 485 +
 486 +       /* Check for proper termination. */
 487 +       if (paren && *g->regparse++ != ')') {
 488 +               FAIL("unmatched ()");
 489 +       } else if (!paren && *g->regparse != '\0') {
 490 +               if (*g->regparse == ')') {
 491 +                       FAIL("unmatched ()");
 492 +               } else
 493 +                       FAIL("junk on end");    /* "Can't happen". */
 494 +               /* NOTREACHED */
 495 +       }
 496 +
 497 +       return(ret);
 498 +}
 499 +
 500 +/*
 501 + - regbranch - one alternative of an | operator
 502 + *
 503 + * Implements the concatenation operator.
 504 + */
 505 +static char *
 506 +regbranch(struct match_globals *g, int *flagp)
 507 +{
 508 +       register char *ret;
 509 +       register char *chain;
 510 +       register char *latest;
 511 +       int flags;
 512 +
 513 +       *flagp = WORST;         /* Tentatively. */
 514 +
 515 +       ret = regnode(g, BRANCH);
 516 +       chain = NULL;
 517 +       while (*g->regparse != '\0' && *g->regparse != '|' && *g->regparse != ')') {
 518 +               latest = regpiece(g, &flags);
 519 +               if (latest == NULL)
 520 +                       return(NULL);
 521 +               *flagp |= flags&HASWIDTH;
 522 +               if (chain == NULL)      /* First piece. */
 523 +                       *flagp |= flags&SPSTART;
 524 +               else
 525 +                       regtail(g, chain, latest);
 526 +               chain = latest;
 527 +       }
 528 +       if (chain == NULL)      /* Loop ran zero times. */
 529 +               (void) regnode(g, NOTHING);
 530 +
 531 +       return(ret);
 532 +}
 533 +
 534 +/*
 535 + - regpiece - something followed by possible [*+?]
 536 + *
 537 + * Note that the branching code sequences used for ? and the general cases
 538 + * of * and + are somewhat optimized:  they use the same NOTHING node as
 539 + * both the endmarker for their branch list and the body of the last branch.
 540 + * It might seem that this node could be dispensed with entirely, but the
 541 + * endmarker role is not redundant.
 542 + */
 543 +static char *
 544 +regpiece(struct match_globals *g, int *flagp)
 545 +{
 546 +       register char *ret;
 547 +       register char op;
 548 +       register char *next;
 549 +       int flags;
 550 +
 551 +       ret = regatom(g, &flags);
 552 +       if (ret == NULL)
 553 +               return(NULL);
 554 +
 555 +       op = *g->regparse;
 556 +       if (!ISMULT(op)) {
 557 +               *flagp = flags;
 558 +               return(ret);
 559 +       }
 560 +
 561 +       if (!(flags&HASWIDTH) && op != '?')
 562 +               FAIL("*+ operand could be empty");
 563 +       *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
 564 +
 565 +       if (op == '*' && (flags&SIMPLE))
 566 +               reginsert(g, STAR, ret);
 567 +       else if (op == '*') {
 568 +               /* Emit x* as (x&|), where & means "self". */
 569 +               reginsert(g, BRANCH, ret);                      /* Either x */
 570 +               regoptail(g, ret, regnode(g, BACK));            /* and loop */
 571 +               regoptail(g, ret, ret);                 /* back */
 572 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
 573 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
 574 +       } else if (op == '+' && (flags&SIMPLE))
 575 +               reginsert(g, PLUS, ret);
 576 +       else if (op == '+') {
 577 +               /* Emit x+ as x(&|), where & means "self". */
 578 +               next = regnode(g, BRANCH);                      /* Either */
 579 +               regtail(g, ret, next);
 580 +               regtail(g, regnode(g, BACK), ret);              /* loop back */
 581 +               regtail(g, next, regnode(g, BRANCH));           /* or */
 582 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
 583 +       } else if (op == '?') {
 584 +               /* Emit x? as (x|) */
 585 +               reginsert(g, BRANCH, ret);                      /* Either x */
 586 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
 587 +               next = regnode(g, NOTHING);             /* null. */
 588 +               regtail(g, ret, next);
 589 +               regoptail(g, ret, next);
 590 +       }
 591 +       g->regparse++;
 592 +       if (ISMULT(*g->regparse))
 593 +               FAIL("nested *?+");
 594 +
 595 +       return(ret);
 596 +}
 597 +
 598 +/*
 599 + - regatom - the lowest level
 600 + *
 601 + * Optimization:  gobbles an entire sequence of ordinary characters so that
 602 + * it can turn them into a single node, which is smaller to store and
 603 + * faster to run.  Backslashed characters are exceptions, each becoming a
 604 + * separate node; the code is simpler that way and it's not worth fixing.
 605 + */
 606 +static char *
 607 +regatom(struct match_globals *g, int *flagp)
 608 +{
 609 +       register char *ret;
 610 +       int flags;
 611 +
 612 +       *flagp = WORST;         /* Tentatively. */
 613 +
 614 +       switch (*g->regparse++) {
 615 +       case '^':
 616 +               ret = regnode(g, BOL);
 617 +               break;
 618 +       case '$':
 619 +               ret = regnode(g, EOL);
 620 +               break;
 621 +       case '.':
 622 +               ret = regnode(g, ANY);
 623 +               *flagp |= HASWIDTH|SIMPLE;
 624 +               break;
 625 +       case '[': {
 626 +                       register int class;
 627 +                       register int classend;
 628 +
 629 +                       if (*g->regparse == '^') {      /* Complement of range. */
 630 +                               ret = regnode(g, ANYBUT);
 631 +                               g->regparse++;
 632 +                       } else
 633 +                               ret = regnode(g, ANYOF);
 634 +                       if (*g->regparse == ']' || *g->regparse == '-')
 635 +                               regc(g, *g->regparse++);
 636 +                       while (*g->regparse != '\0' && *g->regparse != ']') {
 637 +                               if (*g->regparse == '-') {
 638 +                                       g->regparse++;
 639 +                                       if (*g->regparse == ']' || *g->regparse == '\0')
 640 +                                               regc(g, '-');
 641 +                                       else {
 642 +                                               class = UCHARAT(g->regparse-2)+1;
 643 +                                               classend = UCHARAT(g->regparse);
 644 +                                               if (class > classend+1)
 645 +                                                       FAIL("invalid [] range");
 646 +                                               for (; class <= classend; class++)
 647 +                                                       regc(g, class);
 648 +                                               g->regparse++;
 649 +                                       }
 650 +                               } else
 651 +                                       regc(g, *g->regparse++);
 652 +                       }
 653 +                       regc(g, '\0');
 654 +                       if (*g->regparse != ']')
 655 +                               FAIL("unmatched []");
 656 +                       g->regparse++;
 657 +                       *flagp |= HASWIDTH|SIMPLE;
 658 +               }
 659 +               break;
 660 +       case '(':
 661 +               ret = reg(g, 1, &flags);
 662 +               if (ret == NULL)
 663 +                       return(NULL);
 664 +               *flagp |= flags&(HASWIDTH|SPSTART);
 665 +               break;
 666 +       case '\0':
 667 +       case '|':
 668 +       case ')':
 669 +               FAIL("internal urp");   /* Supposed to be caught earlier. */
 670 +               break;
 671 +       case '?':
 672 +       case '+':
 673 +       case '*':
 674 +               FAIL("?+* follows nothing");
 675 +               break;
 676 +       case '\\':
 677 +               if (*g->regparse == '\0')
 678 +                       FAIL("trailing \\");
 679 +               ret = regnode(g, EXACTLY);
 680 +               regc(g, *g->regparse++);
 681 +               regc(g, '\0');
 682 +               *flagp |= HASWIDTH|SIMPLE;
 683 +               break;
 684 +       default: {
 685 +                       register int len;
 686 +                       register char ender;
 687 +
 688 +                       g->regparse--;
 689 +                       len = my_strcspn((const char *)g->regparse, (const char *)META);
 690 +                       if (len <= 0)
 691 +                               FAIL("internal disaster");
 692 +                       ender = *(g->regparse+len);
 693 +                       if (len > 1 && ISMULT(ender))
 694 +                               len--;          /* Back off clear of ?+* operand. */
 695 +                       *flagp |= HASWIDTH;
 696 +                       if (len == 1)
 697 +                               *flagp |= SIMPLE;
 698 +                       ret = regnode(g, EXACTLY);
 699 +                       while (len > 0) {
 700 +                               regc(g, *g->regparse++);
 701 +                               len--;
 702 +                       }
 703 +                       regc(g, '\0');
 704 +               }
 705 +               break;
 706 +       }
 707 +
 708 +       return(ret);
 709 +}
 710 +
 711 +/*
 712 + - regnode - emit a node
 713 + */
 714 +static char *                  /* Location. */
 715 +regnode(struct match_globals *g, char op)
 716 +{
 717 +       register char *ret;
 718 +       register char *ptr;
 719 +
 720 +       ret = g->regcode;
 721 +       if (ret == &g->regdummy) {
 722 +               g->regsize += 3;
 723 +               return(ret);
 724 +       }
 725 +
 726 +       ptr = ret;
 727 +       *ptr++ = op;
 728 +       *ptr++ = '\0';          /* Null "next" pointer. */
 729 +       *ptr++ = '\0';
 730 +       g->regcode = ptr;
 731 +
 732 +       return(ret);
 733 +}
 734 +
 735 +/*
 736 + - regc - emit (if appropriate) a byte of code
 737 + */
 738 +static void
 739 +regc(struct match_globals *g, char b)
 740 +{
 741 +       if (g->regcode != &g->regdummy)
 742 +               *g->regcode++ = b;
 743 +       else
 744 +               g->regsize++;
 745 +}
 746 +
 747 +/*
 748 + - reginsert - insert an operator in front of already-emitted operand
 749 + *
 750 + * Means relocating the operand.
 751 + */
 752 +static void
 753 +reginsert(struct match_globals *g, char op, char* opnd)
 754 +{
 755 +       register char *src;
 756 +       register char *dst;
 757 +       register char *place;
 758 +
 759 +       if (g->regcode == &g->regdummy) {
 760 +               g->regsize += 3;
 761 +               return;
 762 +       }
 763 +
 764 +       src = g->regcode;
 765 +       g->regcode += 3;
 766 +       dst = g->regcode;
 767 +       while (src > opnd)
 768 +               *--dst = *--src;
 769 +
 770 +       place = opnd;           /* Op node, where operand used to be. */
 771 +       *place++ = op;
 772 +       *place++ = '\0';
 773 +       *place++ = '\0';
 774 +}
 775 +
 776 +/*
 777 + - regtail - set the next-pointer at the end of a node chain
 778 + */
 779 +static void
 780 +regtail(struct match_globals *g, char *p, char *val)
 781 +{
 782 +       register char *scan;
 783 +       register char *temp;
 784 +       register int offset;
 785 +
 786 +       if (p == &g->regdummy)
 787 +               return;
 788 +
 789 +       /* Find last node. */
 790 +       scan = p;
 791 +       for (;;) {
 792 +               temp = regnext(g, scan);
 793 +               if (temp == NULL)
 794 +                       break;
 795 +               scan = temp;
 796 +       }
 797 +
 798 +       if (OP(scan) == BACK)
 799 +               offset = scan - val;
 800 +       else
 801 +               offset = val - scan;
 802 +       *(scan+1) = (offset>>8)&0377;
 803 +       *(scan+2) = offset&0377;
 804 +}
 805 +
 806 +/*
 807 + - regoptail - regtail on operand of first argument; nop if operandless
 808 + */
 809 +static void
 810 +regoptail(struct match_globals *g, char *p, char *val)
 811 +{
 812 +       /* "Operandless" and "op != BRANCH" are synonymous in practice. */
 813 +       if (p == NULL || p == &g->regdummy || OP(p) != BRANCH)
 814 +               return;
 815 +       regtail(g, OPERAND(p), val);
 816 +}
 817 +
 818 +/*
 819 + * regexec and friends
 820 + */
 821 +
 822 +
 823 +/*
 824 + * Forwards.
 825 + */
 826 +STATIC int regtry(struct match_globals *g, regexp *prog, char *string);
 827 +STATIC int regmatch(struct match_globals *g, char *prog);
 828 +STATIC int regrepeat(struct match_globals *g, char *p);
 829 +
 830 +#ifdef DEBUG
 831 +int regnarrate = 0;
 832 +void regdump();
 833 +STATIC char *regprop(char *op);
 834 +#endif
 835 +
 836 +/*
 837 + - regexec - match a regexp against a string
 838 + */
 839 +int
 840 +regexec(regexp *prog, char *string)
 841 +{
 842 +       register char *s;
 843 +       struct match_globals g;
 844 +
 845 +       /* Be paranoid... */
 846 +       if (prog == NULL || string == NULL) {
 847 +               printk("<3>Regexp: NULL parameter\n");
 848 +               return(0);
 849 +       }
 850 +
 851 +       /* Check validity of program. */
 852 +       if (UCHARAT(prog->program) != MAGIC) {
 853 +               printk("<3>Regexp: corrupted program\n");
 854 +               return(0);
 855 +       }
 856 +
 857 +       /* If there is a "must appear" string, look for it. */
 858 +       if (prog->regmust != NULL) {
 859 +               s = string;
 860 +               while ((s = strchr(s, prog->regmust[0])) != NULL) {
 861 +                       if (strncmp(s, prog->regmust, prog->regmlen) == 0)
 862 +                               break;  /* Found it. */
 863 +                       s++;
 864 +               }
 865 +               if (s == NULL)  /* Not present. */
 866 +                       return(0);
 867 +       }
 868 +
 869 +       /* Mark beginning of line for ^ . */
 870 +       g.regbol = string;
 871 +
 872 +       /* Simplest case:  anchored match need be tried only once. */
 873 +       if (prog->reganch)
 874 +               return(regtry(&g, prog, string));
 875 +
 876 +       /* Messy cases:  unanchored match. */
 877 +       s = string;
 878 +       if (prog->regstart != '\0')
 879 +               /* We know what char it must start with. */
 880 +               while ((s = strchr(s, prog->regstart)) != NULL) {
 881 +                       if (regtry(&g, prog, s))
 882 +                               return(1);
 883 +                       s++;
 884 +               }
 885 +       else
 886 +               /* We don't -- general case. */
 887 +               do {
 888 +                       if (regtry(&g, prog, s))
 889 +                               return(1);
 890 +               } while (*s++ != '\0');
 891 +
 892 +       /* Failure. */
 893 +       return(0);
 894 +}
 895 +
 896 +/*
 897 + - regtry - try match at specific point
 898 + */
 899 +static int                     /* 0 failure, 1 success */
 900 +regtry(struct match_globals *g, regexp *prog, char *string)
 901 +{
 902 +       register int i;
 903 +       register char **sp;
 904 +       register char **ep;
 905 +
 906 +       g->reginput = string;
 907 +       g->regstartp = prog->startp;
 908 +       g->regendp = prog->endp;
 909 +
 910 +       sp = prog->startp;
 911 +       ep = prog->endp;
 912 +       for (i = NSUBEXP; i > 0; i--) {
 913 +               *sp++ = NULL;
 914 +               *ep++ = NULL;
 915 +       }
 916 +       if (regmatch(g, prog->program + 1)) {
 917 +               prog->startp[0] = string;
 918 +               prog->endp[0] = g->reginput;
 919 +               return(1);
 920 +       } else
 921 +               return(0);
 922 +}
 923 +
 924 +/*
 925 + - regmatch - main matching routine
 926 + *
 927 + * Conceptually the strategy is simple:  check to see whether the current
 928 + * node matches, call self recursively to see whether the rest matches,
 929 + * and then act accordingly.  In practice we make some effort to avoid
 930 + * recursion, in particular by going through "ordinary" nodes (that don't
 931 + * need to know whether the rest of the match failed) by a loop instead of
 932 + * by recursion.
 933 + */
 934 +static int                     /* 0 failure, 1 success */
 935 +regmatch(struct match_globals *g, char *prog)
 936 +{
 937 +       register char *scan = prog; /* Current node. */
 938 +       char *next;                 /* Next node. */
 939 +
 940 +#ifdef DEBUG
 941 +       if (scan != NULL && regnarrate)
 942 +               fprintf(stderr, "%s(\n", regprop(scan));
 943 +#endif
 944 +       while (scan != NULL) {
 945 +#ifdef DEBUG
 946 +               if (regnarrate)
 947 +                       fprintf(stderr, "%s...\n", regprop(scan));
 948 +#endif
 949 +               next = regnext(g, scan);
 950 +
 951 +               switch (OP(scan)) {
 952 +               case BOL:
 953 +                       if (g->reginput != g->regbol)
 954 +                               return(0);
 955 +                       break;
 956 +               case EOL:
 957 +                       if (*g->reginput != '\0')
 958 +                               return(0);
 959 +                       break;
 960 +               case ANY:
 961 +                       if (*g->reginput == '\0')
 962 +                               return(0);
 963 +                       g->reginput++;
 964 +                       break;
 965 +               case EXACTLY: {
 966 +                               register int len;
 967 +                               register char *opnd;
 968 +
 969 +                               opnd = OPERAND(scan);
 970 +                               /* Inline the first character, for speed. */
 971 +                               if (*opnd != *g->reginput)
 972 +                                       return(0);
 973 +                               len = strlen(opnd);
 974 +                               if (len > 1 && strncmp(opnd, g->reginput, len) != 0)
 975 +                                       return(0);
 976 +                               g->reginput += len;
 977 +                       }
 978 +                       break;
 979 +               case ANYOF:
 980 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) == NULL)
 981 +                               return(0);
 982 +                       g->reginput++;
 983 +                       break;
 984 +               case ANYBUT:
 985 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) != NULL)
 986 +                               return(0);
 987 +                       g->reginput++;
 988 +                       break;
 989 +               case NOTHING:
 990 +               case BACK:
 991 +                       break;
 992 +               case OPEN+1:
 993 +               case OPEN+2:
 994 +               case OPEN+3:
 995 +               case OPEN+4:
 996 +               case OPEN+5:
 997 +               case OPEN+6:
 998 +               case OPEN+7:
 999 +               case OPEN+8:
1000 +               case OPEN+9: {
1001 +                               register int no;
1002 +                               register char *save;
1003 +
1004 +                               no = OP(scan) - OPEN;
1005 +                               save = g->reginput;
1006 +
1007 +                               if (regmatch(g, next)) {
1008 +                                       /*
1009 +                                        * Don't set startp if some later
1010 +                                        * invocation of the same parentheses
1011 +                                        * already has.
1012 +                                        */
1013 +                                       if (g->regstartp[no] == NULL)
1014 +                                               g->regstartp[no] = save;
1015 +                                       return(1);
1016 +                               } else
1017 +                                       return(0);
1018 +                       }
1019 +                       break;
1020 +               case CLOSE+1:
1021 +               case CLOSE+2:
1022 +               case CLOSE+3:
1023 +               case CLOSE+4:
1024 +               case CLOSE+5:
1025 +               case CLOSE+6:
1026 +               case CLOSE+7:
1027 +               case CLOSE+8:
1028 +               case CLOSE+9:
1029 +                       {
1030 +                               register int no;
1031 +                               register char *save;
1032 +
1033 +                               no = OP(scan) - CLOSE;
1034 +                               save = g->reginput;
1035 +
1036 +                               if (regmatch(g, next)) {
1037 +                                       /*
1038 +                                        * Don't set endp if some later
1039 +                                        * invocation of the same parentheses
1040 +                                        * already has.
1041 +                                        */
1042 +                                       if (g->regendp[no] == NULL)
1043 +                                               g->regendp[no] = save;
1044 +                                       return(1);
1045 +                               } else
1046 +                                       return(0);
1047 +                       }
1048 +                       break;
1049 +               case BRANCH: {
1050 +                               register char *save;
1051 +
1052 +                               if (OP(next) != BRANCH)         /* No choice. */
1053 +                                       next = OPERAND(scan);   /* Avoid recursion. */
1054 +                               else {
1055 +                                       do {
1056 +                                               save = g->reginput;
1057 +                                               if (regmatch(g, OPERAND(scan)))
1058 +                                                       return(1);
1059 +                                               g->reginput = save;
1060 +                                               scan = regnext(g, scan);
1061 +                                       } while (scan != NULL && OP(scan) == BRANCH);
1062 +                                       return(0);
1063 +                                       /* NOTREACHED */
1064 +                               }
1065 +                       }
1066 +                       break;
1067 +               case STAR:
1068 +               case PLUS: {
1069 +                               register char nextch;
1070 +                               register int no;
1071 +                               register char *save;
1072 +                               register int min;
1073 +
1074 +                               /*
1075 +                                * Lookahead to avoid useless match attempts
1076 +                                * when we know what character comes next.
1077 +                                */
1078 +                               nextch = '\0';
1079 +                               if (OP(next) == EXACTLY)
1080 +                                       nextch = *OPERAND(next);
1081 +                               min = (OP(scan) == STAR) ? 0 : 1;
1082 +                               save = g->reginput;
1083 +                               no = regrepeat(g, OPERAND(scan));
1084 +                               while (no >= min) {
1085 +                                       /* If it could work, try it. */
1086 +                                       if (nextch == '\0' || *g->reginput == nextch)
1087 +                                               if (regmatch(g, next))
1088 +                                                       return(1);
1089 +                                       /* Couldn't or didn't -- back up. */
1090 +                                       no--;
1091 +                                       g->reginput = save + no;
1092 +                               }
1093 +                               return(0);
1094 +                       }
1095 +                       break;
1096 +               case END:
1097 +                       return(1);      /* Success! */
1098 +                       break;
1099 +               default:
1100 +                       printk("<3>Regexp: memory corruption\n");
1101 +                       return(0);
1102 +                       break;
1103 +               }
1104 +
1105 +               scan = next;
1106 +       }
1107 +
1108 +       /*
1109 +        * We get here only if there's trouble -- normally "case END" is
1110 +        * the terminating point.
1111 +        */
1112 +       printk("<3>Regexp: corrupted pointers\n");
1113 +       return(0);
1114 +}
1115 +
1116 +/*
1117 + - regrepeat - repeatedly match something simple, report how many
1118 + */
1119 +static int
1120 +regrepeat(struct match_globals *g, char *p)
1121 +{
1122 +       register int count = 0;
1123 +       register char *scan;
1124 +       register char *opnd;
1125 +
1126 +       scan = g->reginput;
1127 +       opnd = OPERAND(p);
1128 +       switch (OP(p)) {
1129 +       case ANY:
1130 +               count = strlen(scan);
1131 +               scan += count;
1132 +               break;
1133 +       case EXACTLY:
1134 +               while (*opnd == *scan) {
1135 +                       count++;
1136 +                       scan++;
1137 +               }
1138 +               break;
1139 +       case ANYOF:
1140 +               while (*scan != '\0' && strchr(opnd, *scan) != NULL) {
1141 +                       count++;
1142 +                       scan++;
1143 +               }
1144 +               break;
1145 +       case ANYBUT:
1146 +               while (*scan != '\0' && strchr(opnd, *scan) == NULL) {
1147 +                       count++;
1148 +                       scan++;
1149 +               }
1150 +               break;
1151 +       default:                /* Oh dear.  Called inappropriately. */
1152 +               printk("<3>Regexp: internal foulup\n");
1153 +               count = 0;      /* Best compromise. */
1154 +               break;
1155 +       }
1156 +       g->reginput = scan;
1157 +
1158 +       return(count);
1159 +}
1160 +
1161 +/*
1162 + - regnext - dig the "next" pointer out of a node
1163 + */
1164 +static char*
1165 +regnext(struct match_globals *g, char *p)
1166 +{
1167 +       register int offset;
1168 +
1169 +       if (p == &g->regdummy)
1170 +               return(NULL);
1171 +
1172 +       offset = NEXT(p);
1173 +       if (offset == 0)
1174 +               return(NULL);
1175 +
1176 +       if (OP(p) == BACK)
1177 +               return(p-offset);
1178 +       else
1179 +               return(p+offset);
1180 +}
1181 +
1182 +#ifdef DEBUG
1183 +
1184 +STATIC char *regprop();
1185 +
1186 +/*
1187 + - regdump - dump a regexp onto stdout in vaguely comprehensible form
1188 + */
1189 +void
1190 +regdump(regexp *r)
1191 +{
1192 +       register char *s;
1193 +       register char op = EXACTLY;     /* Arbitrary non-END op. */
1194 +       register char *next;
1195 +       /* extern char *strchr(); */
1196 +
1197 +
1198 +       s = r->program + 1;
1199 +       while (op != END) {     /* While that wasn't END last time... */
1200 +               op = OP(s);
1201 +               printf("%2d%s", s-r->program, regprop(s));      /* Where, what. */
1202 +               next = regnext(s);
1203 +               if (next == NULL)               /* Next ptr. */
1204 +                       printf("(0)");
1205 +               else
1206 +                       printf("(%d)", (s-r->program)+(next-s));
1207 +               s += 3;
1208 +               if (op == ANYOF || op == ANYBUT || op == EXACTLY) {
1209 +                       /* Literal string, where present. */
1210 +                       while (*s != '\0') {
1211 +                               putchar(*s);
1212 +                               s++;
1213 +                       }
1214 +                       s++;
1215 +               }
1216 +               putchar('\n');
1217 +       }
1218 +
1219 +       /* Header fields of interest. */
1220 +       if (r->regstart != '\0')
1221 +               printf("start `%c' ", r->regstart);
1222 +       if (r->reganch)
1223 +               printf("anchored ");
1224 +       if (r->regmust != NULL)
1225 +               printf("must have \"%s\"", r->regmust);
1226 +       printf("\n");
1227 +}
1228 +
1229 +/*
1230 + - regprop - printable representation of opcode
1231 + */
1232 +static char *
1233 +regprop(char *op)
1234 +{
1235 +#define BUFLEN 50
1236 +       register char *p;
1237 +       static char buf[BUFLEN];
1238 +
1239 +       strcpy(buf, ":");
1240 +
1241 +       switch (OP(op)) {
1242 +       case BOL:
1243 +               p = "BOL";
1244 +               break;
1245 +       case EOL:
1246 +               p = "EOL";
1247 +               break;
1248 +       case ANY:
1249 +               p = "ANY";
1250 +               break;
1251 +       case ANYOF:
1252 +               p = "ANYOF";
1253 +               break;
1254 +       case ANYBUT:
1255 +               p = "ANYBUT";
1256 +               break;
1257 +       case BRANCH:
1258 +               p = "BRANCH";
1259 +               break;
1260 +       case EXACTLY:
1261 +               p = "EXACTLY";
1262 +               break;
1263 +       case NOTHING:
1264 +               p = "NOTHING";
1265 +               break;
1266 +       case BACK:
1267 +               p = "BACK";
1268 +               break;
1269 +       case END:
1270 +               p = "END";
1271 +               break;
1272 +       case OPEN+1:
1273 +       case OPEN+2:
1274 +       case OPEN+3:
1275 +       case OPEN+4:
1276 +       case OPEN+5:
1277 +       case OPEN+6:
1278 +       case OPEN+7:
1279 +       case OPEN+8:
1280 +       case OPEN+9:
1281 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "OPEN%d", OP(op)-OPEN);
1282 +               p = NULL;
1283 +               break;
1284 +       case CLOSE+1:
1285 +       case CLOSE+2:
1286 +       case CLOSE+3:
1287 +       case CLOSE+4:
1288 +       case CLOSE+5:
1289 +       case CLOSE+6:
1290 +       case CLOSE+7:
1291 +       case CLOSE+8:
1292 +       case CLOSE+9:
1293 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "CLOSE%d", OP(op)-CLOSE);
1294 +               p = NULL;
1295 +               break;
1296 +       case STAR:
1297 +               p = "STAR";
1298 +               break;
1299 +       case PLUS:
1300 +               p = "PLUS";
1301 +               break;
1302 +       default:
1303 +               printk("<3>Regexp: corrupted opcode\n");
1304 +               break;
1305 +       }
1306 +       if (p != NULL)
1307 +               strncat(buf, p, BUFLEN-strlen(buf));
1308 +       return(buf);
1309 +}
1310 +#endif
1311 +
1312 +
1313 --- /dev/null
1314 +++ b/net/netfilter/regexp/regexp.h
1315 @@ -0,0 +1,41 @@
1316 +/*
1317 + * Definitions etc. for regexp(3) routines.
1318 + *
1319 + * Caveat:  this is V8 regexp(3) [actually, a reimplementation thereof],
1320 + * not the System V one.
1321 + */
1322 +
1323 +#ifndef REGEXP_H
1324 +#define REGEXP_H
1325 +
1326 +
1327 +/*
1328 +http://www.opensource.apple.com/darwinsource/10.3/expect-1/expect/expect.h ,
1329 +which contains a version of this library, says:
1330 +
1331 + *
1332 + * NSUBEXP must be at least 10, and no greater than 117 or the parser
1333 + * will not work properly.
1334 + *
1335 +
1336 +However, it looks rather like this library is limited to 10.  If you think
1337 +otherwise, let us know.
1338 +*/
1339 +
1340 +#define NSUBEXP  10
1341 +typedef struct regexp {
1342 +       char *startp[NSUBEXP];
1343 +       char *endp[NSUBEXP];
1344 +       char regstart;          /* Internal use only. */
1345 +       char reganch;           /* Internal use only. */
1346 +       char *regmust;          /* Internal use only. */
1347 +       int regmlen;            /* Internal use only. */
1348 +       char program[1];        /* Unwarranted chumminess with compiler. */
1349 +} regexp;
1350 +
1351 +regexp * regcomp(char *exp, int *patternsize);
1352 +int regexec(regexp *prog, char *string);
1353 +void regsub(regexp *prog, char *source, char *dest);
1354 +void regerror(char *s);
1355 +
1356 +#endif
1357 --- /dev/null
1358 +++ b/net/netfilter/regexp/regmagic.h
1359 @@ -0,0 +1,5 @@
1360 +/*
1361 + * The first byte of the regexp internal "program" is actually this magic
1362 + * number; the start node begins in the second byte.
1363 + */
1364 +#define        MAGIC   0234
1365 --- /dev/null
1366 +++ b/net/netfilter/regexp/regsub.c
1367 @@ -0,0 +1,95 @@
1368 +/*
1369 + * regsub
1370 + * @(#)regsub.c        1.3 of 2 April 86
1371 + *
1372 + *     Copyright (c) 1986 by University of Toronto.
1373 + *     Written by Henry Spencer.  Not derived from licensed software.
1374 + *
1375 + *     Permission is granted to anyone to use this software for any
1376 + *     purpose on any computer system, and to redistribute it freely,
1377 + *     subject to the following restrictions:
1378 + *
1379 + *     1. The author is not responsible for the consequences of use of
1380 + *             this software, no matter how awful, even if they arise
1381 + *             from defects in it.
1382 + *
1383 + *     2. The origin of this software must not be misrepresented, either
1384 + *             by explicit claim or by omission.
1385 + *
1386 + *     3. Altered versions must be plainly marked as such, and must not
1387 + *             be misrepresented as being the original software.
1388 + *
1389 + *
1390 + * This code was modified by Ethan Sommer to work within the kernel
1391 + * (it now uses kmalloc etc..)
1392 + *
1393 + */
1394 +#include "regexp.h"
1395 +#include "regmagic.h"
1396 +#include <linux/string.h>
1397 +
1398 +
1399 +#ifndef CHARBITS
1400 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
1401 +#else
1402 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
1403 +#endif
1404 +
1405 +#if 0
1406 +//void regerror(char * s)
1407 +//{
1408 +//        printk("regexp(3): %s", s);
1409 +//        /* NOTREACHED */
1410 +//}
1411 +#endif
1412 +
1413 +/*
1414 + - regsub - perform substitutions after a regexp match
1415 + */
1416 +void
1417 +regsub(regexp * prog, char * source, char * dest)
1418 +{
1419 +       register char *src;
1420 +       register char *dst;
1421 +       register char c;
1422 +       register int no;
1423 +       register int len;
1424 +
1425 +       /* Not necessary and gcc doesn't like it -MLS */
1426 +       /*extern char *strncpy();*/
1427 +
1428 +       if (prog == NULL || source == NULL || dest == NULL) {
1429 +               regerror("NULL parm to regsub");
1430 +               return;
1431 +       }
1432 +       if (UCHARAT(prog->program) != MAGIC) {
1433 +               regerror("damaged regexp fed to regsub");
1434 +               return;
1435 +       }
1436 +
1437 +       src = source;
1438 +       dst = dest;
1439 +       while ((c = *src++) != '\0') {
1440 +               if (c == '&')
1441 +                       no = 0;
1442 +               else if (c == '\\' && '0' <= *src && *src <= '9')
1443 +                       no = *src++ - '0';
1444 +               else
1445 +                       no = -1;
1446 +
1447 +               if (no < 0) {   /* Ordinary character. */
1448 +                       if (c == '\\' && (*src == '\\' || *src == '&'))
1449 +                               c = *src++;
1450 +                       *dst++ = c;
1451 +               } else if (prog->startp[no] != NULL && prog->endp[no] != NULL) {
1452 +                       len = prog->endp[no] - prog->startp[no];
1453 +                       (void) strncpy(dst, prog->startp[no], len);
1454 +                       dst += len;
1455 +                       if (len != 0 && *(dst-1) == '\0') {     /* strncpy hit NUL. */
1456 +                               regerror("damaged match string");
1457 +                               return;
1458 +                       }
1459 +               }
1460 +       }
1461 +       *dst++ = '\0';
1462 +}
1463 --- /dev/null
1464 +++ b/net/netfilter/xt_layer7.c
1465 @@ -0,0 +1,634 @@
1466 +/*
1467 +  Kernel module to match application layer (OSI layer 7) data in connections.
1468 +
1469 +  http://l7-filter.sf.net
1470 +
1471 +  (C) 2003, 2004, 2005, 2006, 2007 Matthew Strait and Ethan Sommer.
1472 +
1473 +  This program is free software; you can redistribute it and/or
1474 +  modify it under the terms of the GNU General Public License
1475 +  as published by the Free Software Foundation; either version
1476 +  2 of the License, or (at your option) any later version.
1477 +  http://www.gnu.org/licenses/gpl.txt
1478 +
1479 +  Based on ipt_string.c (C) 2000 Emmanuel Roger <winfield@freegates.be>,
1480 +  xt_helper.c (C) 2002 Harald Welte and cls_layer7.c (C) 2003 Matthew Strait,
1481 +  Ethan Sommer, Justin Levandoski.
1482 +*/
1483 +
1484 +#include <linux/spinlock.h>
1485 +#include <linux/version.h>
1486 +#include <net/ip.h>
1487 +#include <net/tcp.h>
1488 +#include <linux/module.h>
1489 +#include <linux/skbuff.h>
1490 +#include <linux/netfilter.h>
1491 +#include <net/netfilter/nf_conntrack.h>
1492 +#include <net/netfilter/nf_conntrack_core.h>
1493 +#include <linux/netfilter/x_tables.h>
1494 +#include <linux/netfilter/xt_layer7.h>
1495 +#include <linux/ctype.h>
1496 +#include <linux/proc_fs.h>
1497 +
1498 +#include "regexp/regexp.c"
1499 +
1500 +MODULE_LICENSE("GPL");
1501 +MODULE_AUTHOR("Matthew Strait <quadong@users.sf.net>, Ethan Sommer <sommere@users.sf.net>");
1502 +MODULE_DESCRIPTION("iptables application layer match module");
1503 +MODULE_ALIAS("ipt_layer7");
1504 +MODULE_VERSION("2.18");
1505 +
1506 +static int maxdatalen = 2048; // this is the default
1507 +module_param(maxdatalen, int, 0444);
1508 +MODULE_PARM_DESC(maxdatalen, "maximum bytes of data looked at by l7-filter");
1509 +#ifdef CONFIG_NETFILTER_XT_MATCH_LAYER7_DEBUG
1510 +       #define DPRINTK(format,args...) printk(format,##args)
1511 +#else
1512 +       #define DPRINTK(format,args...)
1513 +#endif
1514 +
1515 +#define TOTAL_PACKETS master_conntrack->counters[IP_CT_DIR_ORIGINAL].packets + \
1516 +                     master_conntrack->counters[IP_CT_DIR_REPLY].packets
1517 +
1518 +/* Number of packets whose data we look at.
1519 +This can be modified through /proc/net/layer7_numpackets */
1520 +static int num_packets = 10;
1521 +
1522 +static struct pattern_cache {
1523 +       char * regex_string;
1524 +       regexp * pattern;
1525 +       struct pattern_cache * next;
1526 +} * first_pattern_cache = NULL;
1527 +
1528 +DEFINE_SPINLOCK(l7_lock);
1529 +
1530 +#ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
1531 +/* Converts an unfriendly string into a friendly one by
1532 +replacing unprintables with periods and all whitespace with " ". */
1533 +static char * friendly_print(unsigned char * s)
1534 +{
1535 +       char * f = kmalloc(strlen(s) + 1, GFP_ATOMIC);
1536 +       int i;
1537 +
1538 +       if(!f) {
1539 +               if (net_ratelimit())
1540 +                       printk(KERN_ERR "layer7: out of memory in "
1541 +                                       "friendly_print, bailing.\n");
1542 +               return NULL;
1543 +       }
1544 +
1545 +       for(i = 0; i < strlen(s); i++){
1546 +               if(isprint(s[i]) && s[i] < 128) f[i] = s[i];
1547 +               else if(isspace(s[i]))          f[i] = ' ';
1548 +               else                            f[i] = '.';
1549 +       }
1550 +       f[i] = '\0';
1551 +       return f;
1552 +}
1553 +
1554 +static char dec2hex(int i)
1555 +{
1556 +       switch (i) {
1557 +               case 0 ... 9:
1558 +                       return (i + '0');
1559 +                       break;
1560 +               case 10 ... 15:
1561 +                       return (i - 10 + 'a');
1562 +                       break;
1563 +               default:
1564 +                       if (net_ratelimit())
1565 +                               printk("layer7: Problem in dec2hex\n");
1566 +                       return '\0';
1567 +       }
1568 +}
1569 +
1570 +static char * hex_print(unsigned char * s)
1571 +{
1572 +       char * g = kmalloc(strlen(s)*3 + 1, GFP_ATOMIC);
1573 +       int i;
1574 +
1575 +       if(!g) {
1576 +              if (net_ratelimit())
1577 +                       printk(KERN_ERR "layer7: out of memory in hex_print, "
1578 +                                       "bailing.\n");
1579 +              return NULL;
1580 +       }
1581 +
1582 +       for(i = 0; i < strlen(s); i++) {
1583 +               g[i*3    ] = dec2hex(s[i]/16);
1584 +               g[i*3 + 1] = dec2hex(s[i]%16);
1585 +               g[i*3 + 2] = ' ';
1586 +       }
1587 +       g[i*3] = '\0';
1588 +
1589 +       return g;
1590 +}
1591 +#endif // DEBUG
1592 +
1593 +/* Use instead of regcomp.  As we expect to be seeing the same regexps over and
1594 +over again, it make sense to cache the results. */
1595 +static regexp * compile_and_cache(const char * regex_string,
1596 +                                  const char * protocol)
1597 +{
1598 +       struct pattern_cache * node               = first_pattern_cache;
1599 +       struct pattern_cache * last_pattern_cache = first_pattern_cache;
1600 +       struct pattern_cache * tmp;
1601 +       unsigned int len;
1602 +
1603 +       while (node != NULL) {
1604 +               if (!strcmp(node->regex_string, regex_string))
1605 +               return node->pattern;
1606 +
1607 +               last_pattern_cache = node;/* points at the last non-NULL node */
1608 +               node = node->next;
1609 +       }
1610 +
1611 +       /* If we reach the end of the list, then we have not yet cached
1612 +          the pattern for this regex. Let's do that now.
1613 +          Be paranoid about running out of memory to avoid list corruption. */
1614 +       tmp = kmalloc(sizeof(struct pattern_cache), GFP_ATOMIC);
1615 +
1616 +       if(!tmp) {
1617 +               if (net_ratelimit())
1618 +                       printk(KERN_ERR "layer7: out of memory in "
1619 +                                       "compile_and_cache, bailing.\n");
1620 +               return NULL;
1621 +       }
1622 +
1623 +       tmp->regex_string  = kmalloc(strlen(regex_string) + 1, GFP_ATOMIC);
1624 +       tmp->pattern       = kmalloc(sizeof(struct regexp),    GFP_ATOMIC);
1625 +       tmp->next = NULL;
1626 +
1627 +       if(!tmp->regex_string || !tmp->pattern) {
1628 +               if (net_ratelimit())
1629 +                       printk(KERN_ERR "layer7: out of memory in "
1630 +                                       "compile_and_cache, bailing.\n");
1631 +               kfree(tmp->regex_string);
1632 +               kfree(tmp->pattern);
1633 +               kfree(tmp);
1634 +               return NULL;
1635 +       }
1636 +
1637 +       /* Ok.  The new node is all ready now. */
1638 +       node = tmp;
1639 +
1640 +       if(first_pattern_cache == NULL) /* list is empty */
1641 +               first_pattern_cache = node; /* make node the beginning */
1642 +       else
1643 +               last_pattern_cache->next = node; /* attach node to the end */
1644 +
1645 +       /* copy the string and compile the regex */
1646 +       len = strlen(regex_string);
1647 +       DPRINTK("About to compile this: \"%s\"\n", regex_string);
1648 +       node->pattern = regcomp((char *)regex_string, &len);
1649 +       if ( !node->pattern ) {
1650 +               if (net_ratelimit())
1651 +                       printk(KERN_ERR "layer7: Error compiling regexp "
1652 +                                       "\"%s\" (%s)\n",
1653 +                                       regex_string, protocol);
1654 +               /* pattern is now cached as NULL, so we won't try again. */
1655 +       }
1656 +
1657 +       strcpy(node->regex_string, regex_string);
1658 +       return node->pattern;
1659 +}
1660 +
1661 +static int can_handle(const struct sk_buff *skb)
1662 +{
1663 +       if(!ip_hdr(skb)) /* not IP */
1664 +               return 0;
1665 +       if(ip_hdr(skb)->protocol != IPPROTO_TCP &&
1666 +          ip_hdr(skb)->protocol != IPPROTO_UDP &&
1667 +          ip_hdr(skb)->protocol != IPPROTO_ICMP)
1668 +               return 0;
1669 +       return 1;
1670 +}
1671 +
1672 +/* Returns offset the into the skb->data that the application data starts */
1673 +static int app_data_offset(const struct sk_buff *skb)
1674 +{
1675 +       /* In case we are ported somewhere (ebtables?) where ip_hdr(skb)
1676 +       isn't set, this can be gotten from 4*(skb->data[0] & 0x0f) as well. */
1677 +       int ip_hl = 4*ip_hdr(skb)->ihl;
1678 +
1679 +       if( ip_hdr(skb)->protocol == IPPROTO_TCP ) {
1680 +               /* 12 == offset into TCP header for the header length field.
1681 +               Can't get this with skb->h.th->doff because the tcphdr
1682 +               struct doesn't get set when routing (this is confirmed to be
1683 +               true in Netfilter as well as QoS.) */
1684 +               int tcp_hl = 4*(skb->data[ip_hl + 12] >> 4);
1685 +
1686 +               return ip_hl + tcp_hl;
1687 +       } else if( ip_hdr(skb)->protocol == IPPROTO_UDP  ) {
1688 +               return ip_hl + 8; /* UDP header is always 8 bytes */
1689 +       } else if( ip_hdr(skb)->protocol == IPPROTO_ICMP ) {
1690 +               return ip_hl + 8; /* ICMP header is 8 bytes */
1691 +       } else {
1692 +               if (net_ratelimit())
1693 +                       printk(KERN_ERR "layer7: tried to handle unknown "
1694 +                                       "protocol!\n");
1695 +               return ip_hl + 8; /* something reasonable */
1696 +       }
1697 +}
1698 +
1699 +/* handles whether there's a match when we aren't appending data anymore */
1700 +static int match_no_append(struct nf_conn * conntrack,
1701 +                           struct nf_conn * master_conntrack,
1702 +                           enum ip_conntrack_info ctinfo,
1703 +                           enum ip_conntrack_info master_ctinfo,
1704 +                           const struct xt_layer7_info * info)
1705 +{
1706 +       /* If we're in here, throw the app data away */
1707 +       if(master_conntrack->layer7.app_data != NULL) {
1708 +
1709 +       #ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
1710 +               if(!master_conntrack->layer7.app_proto) {
1711 +                       char * f =
1712 +                         friendly_print(master_conntrack->layer7.app_data);
1713 +                       char * g =
1714 +                         hex_print(master_conntrack->layer7.app_data);
1715 +                       DPRINTK("\nl7-filter gave up after %d bytes "
1716 +                               "(%d packets):\n%s\n",
1717 +                               strlen(f), TOTAL_PACKETS, f);
1718 +                       kfree(f);
1719 +                       DPRINTK("In hex: %s\n", g);
1720 +                       kfree(g);
1721 +               }
1722 +       #endif
1723 +
1724 +               kfree(master_conntrack->layer7.app_data);
1725 +               master_conntrack->layer7.app_data = NULL; /* don't free again */
1726 +       }
1727 +
1728 +       if(master_conntrack->layer7.app_proto){
1729 +               /* Here child connections set their .app_proto (for /proc) */
1730 +               if(!conntrack->layer7.app_proto) {
1731 +                       conntrack->layer7.app_proto =
1732 +                         kmalloc(strlen(master_conntrack->layer7.app_proto)+1,
1733 +                           GFP_ATOMIC);
1734 +                       if(!conntrack->layer7.app_proto){
1735 +                               if (net_ratelimit())
1736 +                                       printk(KERN_ERR "layer7: out of memory "
1737 +                                                       "in match_no_append, "
1738 +                                                       "bailing.\n");
1739 +                               return 1;
1740 +                       }
1741 +                       strcpy(conntrack->layer7.app_proto,
1742 +                               master_conntrack->layer7.app_proto);
1743 +               }
1744 +
1745 +               return (!strcmp(master_conntrack->layer7.app_proto,
1746 +                               info->protocol));
1747 +       }
1748 +       else {
1749 +               /* If not classified, set to "unknown" to distinguish from
1750 +               connections that are still being tested. */
1751 +               master_conntrack->layer7.app_proto =
1752 +                       kmalloc(strlen("unknown")+1, GFP_ATOMIC);
1753 +               if(!master_conntrack->layer7.app_proto){
1754 +                       if (net_ratelimit())
1755 +                               printk(KERN_ERR "layer7: out of memory in "
1756 +                                               "match_no_append, bailing.\n");
1757 +                       return 1;
1758 +               }
1759 +               strcpy(master_conntrack->layer7.app_proto, "unknown");
1760 +               return 0;
1761 +       }
1762 +}
1763 +
1764 +/* add the new app data to the conntrack.  Return number of bytes added. */
1765 +static int add_data(struct nf_conn * master_conntrack,
1766 +                    char * app_data, int appdatalen)
1767 +{
1768 +       int length = 0, i;
1769 +       int oldlength = master_conntrack->layer7.app_data_len;
1770 +
1771 +       /* This is a fix for a race condition by Deti Fliegl. However, I'm not
1772 +          clear on whether the race condition exists or whether this really
1773 +          fixes it.  I might just be being dense... Anyway, if it's not really
1774 +          a fix, all it does is waste a very small amount of time. */
1775 +       if(!master_conntrack->layer7.app_data) return 0;
1776 +
1777 +       /* Strip nulls. Make everything lower case (our regex lib doesn't
1778 +       do case insensitivity).  Add it to the end of the current data. */
1779 +       for(i = 0; i < maxdatalen-oldlength-1 &&
1780 +                  i < appdatalen; i++) {
1781 +               if(app_data[i] != '\0') {
1782 +                       /* the kernel version of tolower mungs 'upper ascii' */
1783 +                       master_conntrack->layer7.app_data[length+oldlength] =
1784 +                               isascii(app_data[i])?
1785 +                                       tolower(app_data[i]) : app_data[i];
1786 +                       length++;
1787 +               }
1788 +       }
1789 +
1790 +       master_conntrack->layer7.app_data[length+oldlength] = '\0';
1791 +       master_conntrack->layer7.app_data_len = length + oldlength;
1792 +
1793 +       return length;
1794 +}
1795 +
1796 +/* taken from drivers/video/modedb.c */
1797 +static int my_atoi(const char *s)
1798 +{
1799 +       int val = 0;
1800 +
1801 +       for (;; s++) {
1802 +               switch (*s) {
1803 +                       case '0'...'9':
1804 +                       val = 10*val+(*s-'0');
1805 +                       break;
1806 +               default:
1807 +                       return val;
1808 +               }
1809 +       }
1810 +}
1811 +
1812 +/* write out num_packets to userland. */
1813 +static int layer7_read_proc(char* page, char ** start, off_t off, int count,
1814 +                            int* eof, void * data)
1815 +{
1816 +       if(num_packets > 99 && net_ratelimit())
1817 +               printk(KERN_ERR "layer7: NOT REACHED. num_packets too big\n");
1818 +
1819 +       page[0] = num_packets/10 + '0';
1820 +       page[1] = num_packets%10 + '0';
1821 +       page[2] = '\n';
1822 +       page[3] = '\0';
1823 +
1824 +       *eof=1;
1825 +
1826 +       return 3;
1827 +}
1828 +
1829 +/* Read in num_packets from userland */
1830 +static int layer7_write_proc(struct file* file, const char* buffer,
1831 +                             unsigned long count, void *data)
1832 +{
1833 +       char * foo = kmalloc(count, GFP_ATOMIC);
1834 +
1835 +       if(!foo){
1836 +               if (net_ratelimit())
1837 +                       printk(KERN_ERR "layer7: out of memory, bailing. "
1838 +                                       "num_packets unchanged.\n");
1839 +               return count;
1840 +       }
1841 +
1842 +       if(copy_from_user(foo, buffer, count)) {
1843 +               return -EFAULT;
1844 +       }
1845 +
1846 +
1847 +       num_packets = my_atoi(foo);
1848 +       kfree (foo);
1849 +
1850 +       /* This has an arbitrary limit to make the math easier. I'm lazy.
1851 +       But anyway, 99 is a LOT! If you want more, you're doing it wrong! */
1852 +       if(num_packets > 99) {
1853 +               printk(KERN_WARNING "layer7: num_packets can't be > 99.\n");
1854 +               num_packets = 99;
1855 +       } else if(num_packets < 1) {
1856 +               printk(KERN_WARNING "layer7: num_packets can't be < 1.\n");
1857 +               num_packets = 1;
1858 +       }
1859 +
1860 +       return count;
1861 +}
1862 +
1863 +static int
1864 +match(const struct sk_buff *skbin,
1865 +      const struct net_device *in,
1866 +      const struct net_device *out,
1867 +      const struct xt_match *match,
1868 +      const void *matchinfo,
1869 +      int offset,
1870 +      unsigned int protoff,
1871 +      int *hotdrop)
1872 +{
1873 +       /* sidestep const without getting a compiler warning... */
1874 +       struct sk_buff * skb = (struct sk_buff *)skbin;
1875 +
1876 +       const struct xt_layer7_info * info = matchinfo;
1877 +       enum ip_conntrack_info master_ctinfo, ctinfo;
1878 +       struct nf_conn *master_conntrack, *conntrack;
1879 +       unsigned char * app_data;
1880 +       unsigned int pattern_result, appdatalen;
1881 +       regexp * comppattern;
1882 +
1883 +       /* Be paranoid/incompetent - lock the entire match function. */
1884 +       spin_lock_bh(&l7_lock);
1885 +
1886 +       if(!can_handle(skb)){
1887 +               DPRINTK("layer7: This is some protocol I can't handle.\n");
1888 +               spin_unlock_bh(&l7_lock);
1889 +               return info->invert;
1890 +       }
1891 +
1892 +       /* Treat parent & all its children together as one connection, except
1893 +       for the purpose of setting conntrack->layer7.app_proto in the actual
1894 +       connection. This makes /proc/net/ip_conntrack more satisfying. */
1895 +       if(!(conntrack = nf_ct_get(skb, &ctinfo)) ||
1896 +          !(master_conntrack=nf_ct_get(skb,&master_ctinfo))){
1897 +               DPRINTK("layer7: couldn't get conntrack.\n");
1898 +               spin_unlock_bh(&l7_lock);
1899 +               return info->invert;
1900 +       }
1901 +
1902 +       /* Try to get a master conntrack (and its master etc) for FTP, etc. */
1903 +       while (master_ct(master_conntrack) != NULL)
1904 +               master_conntrack = master_ct(master_conntrack);
1905 +
1906 +       /* if we've classified it or seen too many packets */
1907 +       if(TOTAL_PACKETS > num_packets ||
1908 +          master_conntrack->layer7.app_proto) {
1909 +
1910 +               pattern_result = match_no_append(conntrack, master_conntrack,
1911 +                                                ctinfo, master_ctinfo, info);
1912 +
1913 +               /* skb->cb[0] == seen. Don't do things twice if there are
1914 +               multiple l7 rules. I'm not sure that using cb for this purpose
1915 +               is correct, even though it says "put your private variables
1916 +               there". But it doesn't look like it is being used for anything
1917 +               else in the skbs that make it here. */
1918 +               skb->cb[0] = 1; /* marking it seen here's probably irrelevant */
1919 +
1920 +               spin_unlock_bh(&l7_lock);
1921 +               return (pattern_result ^ info->invert);
1922 +       }
1923 +
1924 +       if(skb_is_nonlinear(skb)){
1925 +               if(skb_linearize(skb) != 0){
1926 +                       if (net_ratelimit())
1927 +                               printk(KERN_ERR "layer7: failed to linearize "
1928 +                                               "packet, bailing.\n");
1929 +                       spin_unlock_bh(&l7_lock);
1930 +                       return info->invert;
1931 +               }
1932 +       }
1933 +
1934 +       /* now that the skb is linearized, it's safe to set these. */
1935 +       app_data = skb->data + app_data_offset(skb);
1936 +       appdatalen = skb_tail_pointer(skb) - app_data;
1937 +
1938 +       /* the return value gets checked later, when we're ready to use it */
1939 +       comppattern = compile_and_cache(info->pattern, info->protocol);
1940 +
1941 +       /* On the first packet of a connection, allocate space for app data */
1942 +       if(TOTAL_PACKETS == 1 && !skb->cb[0] &&
1943 +          !master_conntrack->layer7.app_data){
1944 +               master_conntrack->layer7.app_data =
1945 +                       kmalloc(maxdatalen, GFP_ATOMIC);
1946 +               if(!master_conntrack->layer7.app_data){
1947 +                       if (net_ratelimit())
1948 +                               printk(KERN_ERR "layer7: out of memory in "
1949 +                                               "match, bailing.\n");
1950 +                       spin_unlock_bh(&l7_lock);
1951 +                       return info->invert;
1952 +               }
1953 +
1954 +               master_conntrack->layer7.app_data[0] = '\0';
1955 +       }
1956 +
1957 +       /* Can be here, but unallocated, if numpackets is increased near
1958 +       the beginning of a connection */
1959 +       if(master_conntrack->layer7.app_data == NULL){
1960 +               spin_unlock_bh(&l7_lock);
1961 +               return (info->invert); /* unmatched */
1962 +       }
1963 +
1964 +       if(!skb->cb[0]){
1965 +               int newbytes;
1966 +               newbytes = add_data(master_conntrack, app_data, appdatalen);
1967 +
1968 +               if(newbytes == 0) { /* didn't add any data */
1969 +                       skb->cb[0] = 1;
1970 +                       /* Didn't match before, not going to match now */
1971 +                       spin_unlock_bh(&l7_lock);
1972 +                       return info->invert;
1973 +               }
1974 +       }
1975 +
1976 +       /* If looking for "unknown", then never match.  "Unknown" means that
1977 +       we've given up; we're still trying with these packets. */
1978 +       if(!strcmp(info->protocol, "unknown")) {
1979 +               pattern_result = 0;
1980 +       /* If looking for "unset", then always match. "Unset" means that we
1981 +       haven't yet classified the connection. */
1982 +       } else if(!strcmp(info->protocol, "unset")) {
1983 +               pattern_result = 2;
1984 +               DPRINTK("layer7: matched unset: not yet classified "
1985 +                       "(%d/%d packets)\n", TOTAL_PACKETS, num_packets);
1986 +       /* If the regexp failed to compile, don't bother running it */
1987 +       } else if(comppattern &&
1988 +                 regexec(comppattern, master_conntrack->layer7.app_data)){
1989 +               DPRINTK("layer7: matched %s\n", info->protocol);
1990 +               pattern_result = 1;
1991 +       } else pattern_result = 0;
1992 +
1993 +       if(pattern_result == 1) {
1994 +               master_conntrack->layer7.app_proto =
1995 +                       kmalloc(strlen(info->protocol)+1, GFP_ATOMIC);
1996 +               if(!master_conntrack->layer7.app_proto){
1997 +                       if (net_ratelimit())
1998 +                               printk(KERN_ERR "layer7: out of memory in "
1999 +                                               "match, bailing.\n");
2000 +                       spin_unlock_bh(&l7_lock);
2001 +                       return (pattern_result ^ info->invert);
2002 +               }
2003 +               strcpy(master_conntrack->layer7.app_proto, info->protocol);
2004 +       } else if(pattern_result > 1) { /* cleanup from "unset" */
2005 +               pattern_result = 1;
2006 +       }
2007 +
2008 +       /* mark the packet seen */
2009 +       skb->cb[0] = 1;
2010 +
2011 +       spin_unlock_bh(&l7_lock);
2012 +       return (pattern_result ^ info->invert);
2013 +}
2014 +
2015 +static int check(const char *tablename,
2016 +                const void *inf,
2017 +                const struct xt_match *match,
2018 +                void *matchinfo,
2019 +                unsigned int hook_mask)
2020 +
2021 +{
2022 +       // load nf_conntrack_ipv4
2023 +        if (nf_ct_l3proto_try_module_get(match->family) < 0) {
2024 +                printk(KERN_WARNING "can't load conntrack support for "
2025 +                                    "proto=%d\n", match->family);
2026 +                return 0;
2027 +        }
2028 +       return 1;
2029 +}
2030 +
2031 +static void
2032 +destroy(const struct xt_match *match, void *matchinfo)
2033 +{
2034 +       nf_ct_l3proto_module_put(match->family);
2035 +}
2036 +
2037 +static struct xt_match xt_layer7_match[] = {
2038 +{
2039 +       .name           = "layer7",
2040 +       .family         = AF_INET,
2041 +       .checkentry     = check,
2042 +       .match          = match,
2043 +       .destroy        = destroy,
2044 +       .matchsize      = sizeof(struct xt_layer7_info),
2045 +       .me             = THIS_MODULE
2046 +}
2047 +};
2048 +
2049 +static void layer7_cleanup_proc(void)
2050 +{
2051 +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23)
2052 +       remove_proc_entry("layer7_numpackets", proc_net);
2053 +#else
2054 +       remove_proc_entry("layer7_numpackets", init_net.proc_net);
2055 +#endif
2056 +}
2057 +
2058 +/* register the proc file */
2059 +static void layer7_init_proc(void)
2060 +{
2061 +       struct proc_dir_entry* entry;
2062 +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23)
2063 +       entry = create_proc_entry("layer7_numpackets", 0644, proc_net);
2064 +#else
2065 +       entry = create_proc_entry("layer7_numpackets", 0644, init_net.proc_net);
2066 +#endif
2067 +       entry->read_proc = layer7_read_proc;
2068 +       entry->write_proc = layer7_write_proc;
2069 +}
2070 +
2071 +static int __init xt_layer7_init(void)
2072 +{
2073 +       need_conntrack();
2074 +
2075 +       layer7_init_proc();
2076 +       if(maxdatalen < 1) {
2077 +               printk(KERN_WARNING "layer7: maxdatalen can't be < 1, "
2078 +                       "using 1\n");
2079 +               maxdatalen = 1;
2080 +       }
2081 +       /* This is not a hard limit.  It's just here to prevent people from
2082 +       bringing their slow machines to a grinding halt. */
2083 +       else if(maxdatalen > 65536) {
2084 +               printk(KERN_WARNING "layer7: maxdatalen can't be > 65536, "
2085 +                       "using 65536\n");
2086 +               maxdatalen = 65536;
2087 +       }
2088 +       return xt_register_matches(xt_layer7_match,
2089 +                                  ARRAY_SIZE(xt_layer7_match));
2090 +}
2091 +
2092 +static void __exit xt_layer7_fini(void)
2093 +{
2094 +       layer7_cleanup_proc();
2095 +       xt_unregister_matches(xt_layer7_match, ARRAY_SIZE(xt_layer7_match));
2096 +}
2097 +
2098 +module_init(xt_layer7_init);
2099 +module_exit(xt_layer7_fini);