target/linux/generic-2.6/patches-2.6.29/100-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 @@ -118,6 +118,22 @@ struct nf_conn
  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 @@ -794,6 +794,27 @@ config NETFILTER_XT_MATCH_STATE
  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_ADVANCED
  72 --- a/net/netfilter/Makefile
  73 +++ b/net/netfilter/Makefile
  74 @@ -84,6 +84,7 @@ obj-$(CONFIG_NETFILTER_XT_MATCH_RECENT)
  75  obj-$(CONFIG_NETFILTER_XT_MATCH_SCTP) += xt_sctp.o
  76  obj-$(CONFIG_NETFILTER_XT_MATCH_SOCKET) += xt_socket.o
  77  obj-$(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 @@ -202,6 +202,14 @@ destroy_conntrack(struct nf_conntrack *n
  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 @@ -165,6 +165,12 @@ static int ct_seq_show(struct seq_file *
 102                 return -ENOSPC;
 103  #endif
 104
 105 +#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
 106 +       if(ct->layer7.app_proto &&
 107 +           seq_printf(s, "l7proto=%s ", ct->layer7.app_proto))
 108 +               return -ENOSPC;
 109 +#endif
 110 +
 111         if (seq_printf(s, "use=%u\n", atomic_read(&ct->ct_general.use)))
 112                 return -ENOSPC;
 113
 114 --- /dev/null
 115 +++ b/net/netfilter/regexp/regexp.c
 116 @@ -0,0 +1,1197 @@
 117 +/*
 118 + * regcomp and regexec -- regsub and regerror are elsewhere
 119 + * @(#)regexp.c        1.3 of 18 April 87
 120 + *
 121 + *     Copyright (c) 1986 by University of Toronto.
 122 + *     Written by Henry Spencer.  Not derived from licensed software.
 123 + *
 124 + *     Permission is granted to anyone to use this software for any
 125 + *     purpose on any computer system, and to redistribute it freely,
 126 + *     subject to the following restrictions:
 127 + *
 128 + *     1. The author is not responsible for the consequences of use of
 129 + *             this software, no matter how awful, even if they arise
 130 + *             from defects in it.
 131 + *
 132 + *     2. The origin of this software must not be misrepresented, either
 133 + *             by explicit claim or by omission.
 134 + *
 135 + *     3. Altered versions must be plainly marked as such, and must not
 136 + *             be misrepresented as being the original software.
 137 + *
 138 + * Beware that some of this code is subtly aware of the way operator
 139 + * precedence is structured in regular expressions.  Serious changes in
 140 + * regular-expression syntax might require a total rethink.
 141 + *
 142 + * This code was modified by Ethan Sommer to work within the kernel
 143 + * (it now uses kmalloc etc..)
 144 + *
 145 + * Modified slightly by Matthew Strait to use more modern C.
 146 + */
 147 +
 148 +#include "regexp.h"
 149 +#include "regmagic.h"
 150 +
 151 +/* added by ethan and matt.  Lets it work in both kernel and user space.
 152 +(So iptables can use it, for instance.)  Yea, it goes both ways... */
 153 +#if __KERNEL__
 154 +  #define malloc(foo) kmalloc(foo,GFP_ATOMIC)
 155 +#else
 156 +  #define printk(format,args...) printf(format,##args)
 157 +#endif
 158 +
 159 +void regerror(char * s)
 160 +{
 161 +        printk("<3>Regexp: %s\n", s);
 162 +        /* NOTREACHED */
 163 +}
 164 +
 165 +/*
 166 + * The "internal use only" fields in regexp.h are present to pass info from
 167 + * compile to execute that permits the execute phase to run lots faster on
 168 + * simple cases.  They are:
 169 + *
 170 + * regstart    char that must begin a match; '\0' if none obvious
 171 + * reganch     is the match anchored (at beginning-of-line only)?
 172 + * regmust     string (pointer into program) that match must include, or NULL
 173 + * regmlen     length of regmust string
 174 + *
 175 + * Regstart and reganch permit very fast decisions on suitable starting points
 176 + * for a match, cutting down the work a lot.  Regmust permits fast rejection
 177 + * of lines that cannot possibly match.  The regmust tests are costly enough
 178 + * that regcomp() supplies a regmust only if the r.e. contains something
 179 + * potentially expensive (at present, the only such thing detected is * or +
 180 + * at the start of the r.e., which can involve a lot of backup).  Regmlen is
 181 + * supplied because the test in regexec() needs it and regcomp() is computing
 182 + * it anyway.
 183 + */
 184 +
 185 +/*
 186 + * Structure for regexp "program".  This is essentially a linear encoding
 187 + * of a nondeterministic finite-state machine (aka syntax charts or
 188 + * "railroad normal form" in parsing technology).  Each node is an opcode
 189 + * plus a "next" pointer, possibly plus an operand.  "Next" pointers of
 190 + * all nodes except BRANCH implement concatenation; a "next" pointer with
 191 + * a BRANCH on both ends of it is connecting two alternatives.  (Here we
 192 + * have one of the subtle syntax dependencies:  an individual BRANCH (as
 193 + * opposed to a collection of them) is never concatenated with anything
 194 + * because of operator precedence.)  The operand of some types of node is
 195 + * a literal string; for others, it is a node leading into a sub-FSM.  In
 196 + * particular, the operand of a BRANCH node is the first node of the branch.
 197 + * (NB this is *not* a tree structure:  the tail of the branch connects
 198 + * to the thing following the set of BRANCHes.)  The opcodes are:
 199 + */
 200 +
 201 +/* definition  number  opnd?   meaning */
 202 +#define        END     0       /* no   End of program. */
 203 +#define        BOL     1       /* no   Match "" at beginning of line. */
 204 +#define        EOL     2       /* no   Match "" at end of line. */
 205 +#define        ANY     3       /* no   Match any one character. */
 206 +#define        ANYOF   4       /* str  Match any character in this string. */
 207 +#define        ANYBUT  5       /* str  Match any character not in this string. */
 208 +#define        BRANCH  6       /* node Match this alternative, or the next... */
 209 +#define        BACK    7       /* no   Match "", "next" ptr points backward. */
 210 +#define        EXACTLY 8       /* str  Match this string. */
 211 +#define        NOTHING 9       /* no   Match empty string. */
 212 +#define        STAR    10      /* node Match this (simple) thing 0 or more times. */
 213 +#define        PLUS    11      /* node Match this (simple) thing 1 or more times. */
 214 +#define        OPEN    20      /* no   Mark this point in input as start of #n. */
 215 +                       /*      OPEN+1 is number 1, etc. */
 216 +#define        CLOSE   30      /* no   Analogous to OPEN. */
 217 +
 218 +/*
 219 + * Opcode notes:
 220 + *
 221 + * BRANCH      The set of branches constituting a single choice are hooked
 222 + *             together with their "next" pointers, since precedence prevents
 223 + *             anything being concatenated to any individual branch.  The
 224 + *             "next" pointer of the last BRANCH in a choice points to the
 225 + *             thing following the whole choice.  This is also where the
 226 + *             final "next" pointer of each individual branch points; each
 227 + *             branch starts with the operand node of a BRANCH node.
 228 + *
 229 + * BACK                Normal "next" pointers all implicitly point forward; BACK
 230 + *             exists to make loop structures possible.
 231 + *
 232 + * STAR,PLUS   '?', and complex '*' and '+', are implemented as circular
 233 + *             BRANCH structures using BACK.  Simple cases (one character
 234 + *             per match) are implemented with STAR and PLUS for speed
 235 + *             and to minimize recursive plunges.
 236 + *
 237 + * OPEN,CLOSE  ...are numbered at compile time.
 238 + */
 239 +
 240 +/*
 241 + * A node is one char of opcode followed by two chars of "next" pointer.
 242 + * "Next" pointers are stored as two 8-bit pieces, high order first.  The
 243 + * value is a positive offset from the opcode of the node containing it.
 244 + * An operand, if any, simply follows the node.  (Note that much of the
 245 + * code generation knows about this implicit relationship.)
 246 + *
 247 + * Using two bytes for the "next" pointer is vast overkill for most things,
 248 + * but allows patterns to get big without disasters.
 249 + */
 250 +#define        OP(p)   (*(p))
 251 +#define        NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377))
 252 +#define        OPERAND(p)      ((p) + 3)
 253 +
 254 +/*
 255 + * See regmagic.h for one further detail of program structure.
 256 + */
 257 +
 258 +
 259 +/*
 260 + * Utility definitions.
 261 + */
 262 +#ifndef CHARBITS
 263 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
 264 +#else
 265 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
 266 +#endif
 267 +
 268 +#define        FAIL(m) { regerror(m); return(NULL); }
 269 +#define        ISMULT(c)       ((c) == '*' || (c) == '+' || (c) == '?')
 270 +#define        META    "^$.[()|?+*\\"
 271 +
 272 +/*
 273 + * Flags to be passed up and down.
 274 + */
 275 +#define        HASWIDTH        01      /* Known never to match null string. */
 276 +#define        SIMPLE          02      /* Simple enough to be STAR/PLUS operand. */
 277 +#define        SPSTART         04      /* Starts with * or +. */
 278 +#define        WORST           0       /* Worst case. */
 279 +
 280 +/*
 281 + * Global work variables for regcomp().
 282 + */
 283 +struct match_globals {
 284 +char *reginput;                /* String-input pointer. */
 285 +char *regbol;          /* Beginning of input, for ^ check. */
 286 +char **regstartp;      /* Pointer to startp array. */
 287 +char **regendp;                /* Ditto for endp. */
 288 +char *regparse;                /* Input-scan pointer. */
 289 +int regnpar;           /* () count. */
 290 +char regdummy;
 291 +char *regcode;         /* Code-emit pointer; &regdummy = don't. */
 292 +long regsize;          /* Code size. */
 293 +};
 294 +
 295 +/*
 296 + * Forward declarations for regcomp()'s friends.
 297 + */
 298 +#ifndef STATIC
 299 +#define        STATIC  static
 300 +#endif
 301 +STATIC char *reg(struct match_globals *g, int paren,int *flagp);
 302 +STATIC char *regbranch(struct match_globals *g, int *flagp);
 303 +STATIC char *regpiece(struct match_globals *g, int *flagp);
 304 +STATIC char *regatom(struct match_globals *g, int *flagp);
 305 +STATIC char *regnode(struct match_globals *g, char op);
 306 +STATIC char *regnext(struct match_globals *g, char *p);
 307 +STATIC void regc(struct match_globals *g, char b);
 308 +STATIC void reginsert(struct match_globals *g, char op, char *opnd);
 309 +STATIC void regtail(struct match_globals *g, char *p, char *val);
 310 +STATIC void regoptail(struct match_globals *g, char *p, char *val);
 311 +
 312 +
 313 +__kernel_size_t my_strcspn(const char *s1,const char *s2)
 314 +{
 315 +        char *scan1;
 316 +        char *scan2;
 317 +        int count;
 318 +
 319 +        count = 0;
 320 +        for (scan1 = (char *)s1; *scan1 != '\0'; scan1++) {
 321 +                for (scan2 = (char *)s2; *scan2 != '\0';)       /* ++ moved down. */
 322 +                        if (*scan1 == *scan2++)
 323 +                                return(count);
 324 +                count++;
 325 +        }
 326 +        return(count);
 327 +}
 328 +
 329 +/*
 330 + - regcomp - compile a regular expression into internal code
 331 + *
 332 + * We can't allocate space until we know how big the compiled form will be,
 333 + * but we can't compile it (and thus know how big it is) until we've got a
 334 + * place to put the code.  So we cheat:  we compile it twice, once with code
 335 + * generation turned off and size counting turned on, and once "for real".
 336 + * This also means that we don't allocate space until we are sure that the
 337 + * thing really will compile successfully, and we never have to move the
 338 + * code and thus invalidate pointers into it.  (Note that it has to be in
 339 + * one piece because free() must be able to free it all.)
 340 + *
 341 + * Beware that the optimization-preparation code in here knows about some
 342 + * of the structure of the compiled regexp.
 343 + */
 344 +regexp *
 345 +regcomp(char *exp,int *patternsize)
 346 +{
 347 +       register regexp *r;
 348 +       register char *scan;
 349 +       register char *longest;
 350 +       register int len;
 351 +       int flags;
 352 +       struct match_globals g;
 353 +
 354 +       /* commented out by ethan
 355 +          extern char *malloc();
 356 +       */
 357 +
 358 +       if (exp == NULL)
 359 +               FAIL("NULL argument");
 360 +
 361 +       /* First pass: determine size, legality. */
 362 +       g.regparse = exp;
 363 +       g.regnpar = 1;
 364 +       g.regsize = 0L;
 365 +       g.regcode = &g.regdummy;
 366 +       regc(&g, MAGIC);
 367 +       if (reg(&g, 0, &flags) == NULL)
 368 +               return(NULL);
 369 +
 370 +       /* Small enough for pointer-storage convention? */
 371 +       if (g.regsize >= 32767L)                /* Probably could be 65535L. */
 372 +               FAIL("regexp too big");
 373 +
 374 +       /* Allocate space. */
 375 +       *patternsize=sizeof(regexp) + (unsigned)g.regsize;
 376 +       r = (regexp *)malloc(sizeof(regexp) + (unsigned)g.regsize);
 377 +       if (r == NULL)
 378 +               FAIL("out of space");
 379 +
 380 +       /* Second pass: emit code. */
 381 +       g.regparse = exp;
 382 +       g.regnpar = 1;
 383 +       g.regcode = r->program;
 384 +       regc(&g, MAGIC);
 385 +       if (reg(&g, 0, &flags) == NULL)
 386 +               return(NULL);
 387 +
 388 +       /* Dig out information for optimizations. */
 389 +       r->regstart = '\0';     /* Worst-case defaults. */
 390 +       r->reganch = 0;
 391 +       r->regmust = NULL;
 392 +       r->regmlen = 0;
 393 +       scan = r->program+1;                    /* First BRANCH. */
 394 +       if (OP(regnext(&g, scan)) == END) {             /* Only one top-level choice. */
 395 +               scan = OPERAND(scan);
 396 +
 397 +               /* Starting-point info. */
 398 +               if (OP(scan) == EXACTLY)
 399 +                       r->regstart = *OPERAND(scan);
 400 +               else if (OP(scan) == BOL)
 401 +                       r->reganch++;
 402 +
 403 +               /*
 404 +                * If there's something expensive in the r.e., find the
 405 +                * longest literal string that must appear and make it the
 406 +                * regmust.  Resolve ties in favor of later strings, since
 407 +                * the regstart check works with the beginning of the r.e.
 408 +                * and avoiding duplication strengthens checking.  Not a
 409 +                * strong reason, but sufficient in the absence of others.
 410 +                */
 411 +               if (flags&SPSTART) {
 412 +                       longest = NULL;
 413 +                       len = 0;
 414 +                       for (; scan != NULL; scan = regnext(&g, scan))
 415 +                               if (OP(scan) == EXACTLY && strlen(OPERAND(scan)) >= len) {
 416 +                                       longest = OPERAND(scan);
 417 +                                       len = strlen(OPERAND(scan));
 418 +                               }
 419 +                       r->regmust = longest;
 420 +                       r->regmlen = len;
 421 +               }
 422 +       }
 423 +
 424 +       return(r);
 425 +}
 426 +
 427 +/*
 428 + - reg - regular expression, i.e. main body or parenthesized thing
 429 + *
 430 + * Caller must absorb opening parenthesis.
 431 + *
 432 + * Combining parenthesis handling with the base level of regular expression
 433 + * is a trifle forced, but the need to tie the tails of the branches to what
 434 + * follows makes it hard to avoid.
 435 + */
 436 +static char *
 437 +reg(struct match_globals *g, int paren, int *flagp /* Parenthesized? */ )
 438 +{
 439 +       register char *ret;
 440 +       register char *br;
 441 +       register char *ender;
 442 +       register int parno = 0; /* 0 makes gcc happy */
 443 +       int flags;
 444 +
 445 +       *flagp = HASWIDTH;      /* Tentatively. */
 446 +
 447 +       /* Make an OPEN node, if parenthesized. */
 448 +       if (paren) {
 449 +               if (g->regnpar >= NSUBEXP)
 450 +                       FAIL("too many ()");
 451 +               parno = g->regnpar;
 452 +               g->regnpar++;
 453 +               ret = regnode(g, OPEN+parno);
 454 +       } else
 455 +               ret = NULL;
 456 +
 457 +       /* Pick up the branches, linking them together. */
 458 +       br = regbranch(g, &flags);
 459 +       if (br == NULL)
 460 +               return(NULL);
 461 +       if (ret != NULL)
 462 +               regtail(g, ret, br);    /* OPEN -> first. */
 463 +       else
 464 +               ret = br;
 465 +       if (!(flags&HASWIDTH))
 466 +               *flagp &= ~HASWIDTH;
 467 +       *flagp |= flags&SPSTART;
 468 +       while (*g->regparse == '|') {
 469 +               g->regparse++;
 470 +               br = regbranch(g, &flags);
 471 +               if (br == NULL)
 472 +                       return(NULL);
 473 +               regtail(g, ret, br);    /* BRANCH -> BRANCH. */
 474 +               if (!(flags&HASWIDTH))
 475 +                       *flagp &= ~HASWIDTH;
 476 +               *flagp |= flags&SPSTART;
 477 +       }
 478 +
 479 +       /* Make a closing node, and hook it on the end. */
 480 +       ender = regnode(g, (paren) ? CLOSE+parno : END);
 481 +       regtail(g, ret, ender);
 482 +
 483 +       /* Hook the tails of the branches to the closing node. */
 484 +       for (br = ret; br != NULL; br = regnext(g, br))
 485 +               regoptail(g, br, ender);
 486 +
 487 +       /* Check for proper termination. */
 488 +       if (paren && *g->regparse++ != ')') {
 489 +               FAIL("unmatched ()");
 490 +       } else if (!paren && *g->regparse != '\0') {
 491 +               if (*g->regparse == ')') {
 492 +                       FAIL("unmatched ()");
 493 +               } else
 494 +                       FAIL("junk on end");    /* "Can't happen". */
 495 +               /* NOTREACHED */
 496 +       }
 497 +
 498 +       return(ret);
 499 +}
 500 +
 501 +/*
 502 + - regbranch - one alternative of an | operator
 503 + *
 504 + * Implements the concatenation operator.
 505 + */
 506 +static char *
 507 +regbranch(struct match_globals *g, int *flagp)
 508 +{
 509 +       register char *ret;
 510 +       register char *chain;
 511 +       register char *latest;
 512 +       int flags;
 513 +
 514 +       *flagp = WORST;         /* Tentatively. */
 515 +
 516 +       ret = regnode(g, BRANCH);
 517 +       chain = NULL;
 518 +       while (*g->regparse != '\0' && *g->regparse != '|' && *g->regparse != ')') {
 519 +               latest = regpiece(g, &flags);
 520 +               if (latest == NULL)
 521 +                       return(NULL);
 522 +               *flagp |= flags&HASWIDTH;
 523 +               if (chain == NULL)      /* First piece. */
 524 +                       *flagp |= flags&SPSTART;
 525 +               else
 526 +                       regtail(g, chain, latest);
 527 +               chain = latest;
 528 +       }
 529 +       if (chain == NULL)      /* Loop ran zero times. */
 530 +               (void) regnode(g, NOTHING);
 531 +
 532 +       return(ret);
 533 +}
 534 +
 535 +/*
 536 + - regpiece - something followed by possible [*+?]
 537 + *
 538 + * Note that the branching code sequences used for ? and the general cases
 539 + * of * and + are somewhat optimized:  they use the same NOTHING node as
 540 + * both the endmarker for their branch list and the body of the last branch.
 541 + * It might seem that this node could be dispensed with entirely, but the
 542 + * endmarker role is not redundant.
 543 + */
 544 +static char *
 545 +regpiece(struct match_globals *g, int *flagp)
 546 +{
 547 +       register char *ret;
 548 +       register char op;
 549 +       register char *next;
 550 +       int flags;
 551 +
 552 +       ret = regatom(g, &flags);
 553 +       if (ret == NULL)
 554 +               return(NULL);
 555 +
 556 +       op = *g->regparse;
 557 +       if (!ISMULT(op)) {
 558 +               *flagp = flags;
 559 +               return(ret);
 560 +       }
 561 +
 562 +       if (!(flags&HASWIDTH) && op != '?')
 563 +               FAIL("*+ operand could be empty");
 564 +       *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
 565 +
 566 +       if (op == '*' && (flags&SIMPLE))
 567 +               reginsert(g, STAR, ret);
 568 +       else if (op == '*') {
 569 +               /* Emit x* as (x&|), where & means "self". */
 570 +               reginsert(g, BRANCH, ret);                      /* Either x */
 571 +               regoptail(g, ret, regnode(g, BACK));            /* and loop */
 572 +               regoptail(g, ret, ret);                 /* back */
 573 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
 574 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
 575 +       } else if (op == '+' && (flags&SIMPLE))
 576 +               reginsert(g, PLUS, ret);
 577 +       else if (op == '+') {
 578 +               /* Emit x+ as x(&|), where & means "self". */
 579 +               next = regnode(g, BRANCH);                      /* Either */
 580 +               regtail(g, ret, next);
 581 +               regtail(g, regnode(g, BACK), ret);              /* loop back */
 582 +               regtail(g, next, regnode(g, BRANCH));           /* or */
 583 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
 584 +       } else if (op == '?') {
 585 +               /* Emit x? as (x|) */
 586 +               reginsert(g, BRANCH, ret);                      /* Either x */
 587 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
 588 +               next = regnode(g, NOTHING);             /* null. */
 589 +               regtail(g, ret, next);
 590 +               regoptail(g, ret, next);
 591 +       }
 592 +       g->regparse++;
 593 +       if (ISMULT(*g->regparse))
 594 +               FAIL("nested *?+");
 595 +
 596 +       return(ret);
 597 +}
 598 +
 599 +/*
 600 + - regatom - the lowest level
 601 + *
 602 + * Optimization:  gobbles an entire sequence of ordinary characters so that
 603 + * it can turn them into a single node, which is smaller to store and
 604 + * faster to run.  Backslashed characters are exceptions, each becoming a
 605 + * separate node; the code is simpler that way and it's not worth fixing.
 606 + */
 607 +static char *
 608 +regatom(struct match_globals *g, int *flagp)
 609 +{
 610 +       register char *ret;
 611 +       int flags;
 612 +
 613 +       *flagp = WORST;         /* Tentatively. */
 614 +
 615 +       switch (*g->regparse++) {
 616 +       case '^':
 617 +               ret = regnode(g, BOL);
 618 +               break;
 619 +       case '$':
 620 +               ret = regnode(g, EOL);
 621 +               break;
 622 +       case '.':
 623 +               ret = regnode(g, ANY);
 624 +               *flagp |= HASWIDTH|SIMPLE;
 625 +               break;
 626 +       case '[': {
 627 +                       register int class;
 628 +                       register int classend;
 629 +
 630 +                       if (*g->regparse == '^') {      /* Complement of range. */
 631 +                               ret = regnode(g, ANYBUT);
 632 +                               g->regparse++;
 633 +                       } else
 634 +                               ret = regnode(g, ANYOF);
 635 +                       if (*g->regparse == ']' || *g->regparse == '-')
 636 +                               regc(g, *g->regparse++);
 637 +                       while (*g->regparse != '\0' && *g->regparse != ']') {
 638 +                               if (*g->regparse == '-') {
 639 +                                       g->regparse++;
 640 +                                       if (*g->regparse == ']' || *g->regparse == '\0')
 641 +                                               regc(g, '-');
 642 +                                       else {
 643 +                                               class = UCHARAT(g->regparse-2)+1;
 644 +                                               classend = UCHARAT(g->regparse);
 645 +                                               if (class > classend+1)
 646 +                                                       FAIL("invalid [] range");
 647 +                                               for (; class <= classend; class++)
 648 +                                                       regc(g, class);
 649 +                                               g->regparse++;
 650 +                                       }
 651 +                               } else
 652 +                                       regc(g, *g->regparse++);
 653 +                       }
 654 +                       regc(g, '\0');
 655 +                       if (*g->regparse != ']')
 656 +                               FAIL("unmatched []");
 657 +                       g->regparse++;
 658 +                       *flagp |= HASWIDTH|SIMPLE;
 659 +               }
 660 +               break;
 661 +       case '(':
 662 +               ret = reg(g, 1, &flags);
 663 +               if (ret == NULL)
 664 +                       return(NULL);
 665 +               *flagp |= flags&(HASWIDTH|SPSTART);
 666 +               break;
 667 +       case '\0':
 668 +       case '|':
 669 +       case ')':
 670 +               FAIL("internal urp");   /* Supposed to be caught earlier. */
 671 +               break;
 672 +       case '?':
 673 +       case '+':
 674 +       case '*':
 675 +               FAIL("?+* follows nothing");
 676 +               break;
 677 +       case '\\':
 678 +               if (*g->regparse == '\0')
 679 +                       FAIL("trailing \\");
 680 +               ret = regnode(g, EXACTLY);
 681 +               regc(g, *g->regparse++);
 682 +               regc(g, '\0');
 683 +               *flagp |= HASWIDTH|SIMPLE;
 684 +               break;
 685 +       default: {
 686 +                       register int len;
 687 +                       register char ender;
 688 +
 689 +                       g->regparse--;
 690 +                       len = my_strcspn((const char *)g->regparse, (const char *)META);
 691 +                       if (len <= 0)
 692 +                               FAIL("internal disaster");
 693 +                       ender = *(g->regparse+len);
 694 +                       if (len > 1 && ISMULT(ender))
 695 +                               len--;          /* Back off clear of ?+* operand. */
 696 +                       *flagp |= HASWIDTH;
 697 +                       if (len == 1)
 698 +                               *flagp |= SIMPLE;
 699 +                       ret = regnode(g, EXACTLY);
 700 +                       while (len > 0) {
 701 +                               regc(g, *g->regparse++);
 702 +                               len--;
 703 +                       }
 704 +                       regc(g, '\0');
 705 +               }
 706 +               break;
 707 +       }
 708 +
 709 +       return(ret);
 710 +}
 711 +
 712 +/*
 713 + - regnode - emit a node
 714 + */
 715 +static char *                  /* Location. */
 716 +regnode(struct match_globals *g, char op)
 717 +{
 718 +       register char *ret;
 719 +       register char *ptr;
 720 +
 721 +       ret = g->regcode;
 722 +       if (ret == &g->regdummy) {
 723 +               g->regsize += 3;
 724 +               return(ret);
 725 +       }
 726 +
 727 +       ptr = ret;
 728 +       *ptr++ = op;
 729 +       *ptr++ = '\0';          /* Null "next" pointer. */
 730 +       *ptr++ = '\0';
 731 +       g->regcode = ptr;
 732 +
 733 +       return(ret);
 734 +}
 735 +
 736 +/*
 737 + - regc - emit (if appropriate) a byte of code
 738 + */
 739 +static void
 740 +regc(struct match_globals *g, char b)
 741 +{
 742 +       if (g->regcode != &g->regdummy)
 743 +               *g->regcode++ = b;
 744 +       else
 745 +               g->regsize++;
 746 +}
 747 +
 748 +/*
 749 + - reginsert - insert an operator in front of already-emitted operand
 750 + *
 751 + * Means relocating the operand.
 752 + */
 753 +static void
 754 +reginsert(struct match_globals *g, char op, char* opnd)
 755 +{
 756 +       register char *src;
 757 +       register char *dst;
 758 +       register char *place;
 759 +
 760 +       if (g->regcode == &g->regdummy) {
 761 +               g->regsize += 3;
 762 +               return;
 763 +       }
 764 +
 765 +       src = g->regcode;
 766 +       g->regcode += 3;
 767 +       dst = g->regcode;
 768 +       while (src > opnd)
 769 +               *--dst = *--src;
 770 +
 771 +       place = opnd;           /* Op node, where operand used to be. */
 772 +       *place++ = op;
 773 +       *place++ = '\0';
 774 +       *place++ = '\0';
 775 +}
 776 +
 777 +/*
 778 + - regtail - set the next-pointer at the end of a node chain
 779 + */
 780 +static void
 781 +regtail(struct match_globals *g, char *p, char *val)
 782 +{
 783 +       register char *scan;
 784 +       register char *temp;
 785 +       register int offset;
 786 +
 787 +       if (p == &g->regdummy)
 788 +               return;
 789 +
 790 +       /* Find last node. */
 791 +       scan = p;
 792 +       for (;;) {
 793 +               temp = regnext(g, scan);
 794 +               if (temp == NULL)
 795 +                       break;
 796 +               scan = temp;
 797 +       }
 798 +
 799 +       if (OP(scan) == BACK)
 800 +               offset = scan - val;
 801 +       else
 802 +               offset = val - scan;
 803 +       *(scan+1) = (offset>>8)&0377;
 804 +       *(scan+2) = offset&0377;
 805 +}
 806 +
 807 +/*
 808 + - regoptail - regtail on operand of first argument; nop if operandless
 809 + */
 810 +static void
 811 +regoptail(struct match_globals *g, char *p, char *val)
 812 +{
 813 +       /* "Operandless" and "op != BRANCH" are synonymous in practice. */
 814 +       if (p == NULL || p == &g->regdummy || OP(p) != BRANCH)
 815 +               return;
 816 +       regtail(g, OPERAND(p), val);
 817 +}
 818 +
 819 +/*
 820 + * regexec and friends
 821 + */
 822 +
 823 +
 824 +/*
 825 + * Forwards.
 826 + */
 827 +STATIC int regtry(struct match_globals *g, regexp *prog, char *string);
 828 +STATIC int regmatch(struct match_globals *g, char *prog);
 829 +STATIC int regrepeat(struct match_globals *g, char *p);
 830 +
 831 +#ifdef DEBUG
 832 +int regnarrate = 0;
 833 +void regdump();
 834 +STATIC char *regprop(char *op);
 835 +#endif
 836 +
 837 +/*
 838 + - regexec - match a regexp against a string
 839 + */
 840 +int
 841 +regexec(regexp *prog, char *string)
 842 +{
 843 +       register char *s;
 844 +       struct match_globals g;
 845 +
 846 +       /* Be paranoid... */
 847 +       if (prog == NULL || string == NULL) {
 848 +               printk("<3>Regexp: NULL parameter\n");
 849 +               return(0);
 850 +       }
 851 +
 852 +       /* Check validity of program. */
 853 +       if (UCHARAT(prog->program) != MAGIC) {
 854 +               printk("<3>Regexp: corrupted program\n");
 855 +               return(0);
 856 +       }
 857 +
 858 +       /* If there is a "must appear" string, look for it. */
 859 +       if (prog->regmust != NULL) {
 860 +               s = string;
 861 +               while ((s = strchr(s, prog->regmust[0])) != NULL) {
 862 +                       if (strncmp(s, prog->regmust, prog->regmlen) == 0)
 863 +                               break;  /* Found it. */
 864 +                       s++;
 865 +               }
 866 +               if (s == NULL)  /* Not present. */
 867 +                       return(0);
 868 +       }
 869 +
 870 +       /* Mark beginning of line for ^ . */
 871 +       g.regbol = string;
 872 +
 873 +       /* Simplest case:  anchored match need be tried only once. */
 874 +       if (prog->reganch)
 875 +               return(regtry(&g, prog, string));
 876 +
 877 +       /* Messy cases:  unanchored match. */
 878 +       s = string;
 879 +       if (prog->regstart != '\0')
 880 +               /* We know what char it must start with. */
 881 +               while ((s = strchr(s, prog->regstart)) != NULL) {
 882 +                       if (regtry(&g, prog, s))
 883 +                               return(1);
 884 +                       s++;
 885 +               }
 886 +       else
 887 +               /* We don't -- general case. */
 888 +               do {
 889 +                       if (regtry(&g, prog, s))
 890 +                               return(1);
 891 +               } while (*s++ != '\0');
 892 +
 893 +       /* Failure. */
 894 +       return(0);
 895 +}
 896 +
 897 +/*
 898 + - regtry - try match at specific point
 899 + */
 900 +static int                     /* 0 failure, 1 success */
 901 +regtry(struct match_globals *g, regexp *prog, char *string)
 902 +{
 903 +       register int i;
 904 +       register char **sp;
 905 +       register char **ep;
 906 +
 907 +       g->reginput = string;
 908 +       g->regstartp = prog->startp;
 909 +       g->regendp = prog->endp;
 910 +
 911 +       sp = prog->startp;
 912 +       ep = prog->endp;
 913 +       for (i = NSUBEXP; i > 0; i--) {
 914 +               *sp++ = NULL;
 915 +               *ep++ = NULL;
 916 +       }
 917 +       if (regmatch(g, prog->program + 1)) {
 918 +               prog->startp[0] = string;
 919 +               prog->endp[0] = g->reginput;
 920 +               return(1);
 921 +       } else
 922 +               return(0);
 923 +}
 924 +
 925 +/*
 926 + - regmatch - main matching routine
 927 + *
 928 + * Conceptually the strategy is simple:  check to see whether the current
 929 + * node matches, call self recursively to see whether the rest matches,
 930 + * and then act accordingly.  In practice we make some effort to avoid
 931 + * recursion, in particular by going through "ordinary" nodes (that don't
 932 + * need to know whether the rest of the match failed) by a loop instead of
 933 + * by recursion.
 934 + */
 935 +static int                     /* 0 failure, 1 success */
 936 +regmatch(struct match_globals *g, char *prog)
 937 +{
 938 +       register char *scan = prog; /* Current node. */
 939 +       char *next;                 /* Next node. */
 940 +
 941 +#ifdef DEBUG
 942 +       if (scan != NULL && regnarrate)
 943 +               fprintf(stderr, "%s(\n", regprop(scan));
 944 +#endif
 945 +       while (scan != NULL) {
 946 +#ifdef DEBUG
 947 +               if (regnarrate)
 948 +                       fprintf(stderr, "%s...\n", regprop(scan));
 949 +#endif
 950 +               next = regnext(g, scan);
 951 +
 952 +               switch (OP(scan)) {
 953 +               case BOL:
 954 +                       if (g->reginput != g->regbol)
 955 +                               return(0);
 956 +                       break;
 957 +               case EOL:
 958 +                       if (*g->reginput != '\0')
 959 +                               return(0);
 960 +                       break;
 961 +               case ANY:
 962 +                       if (*g->reginput == '\0')
 963 +                               return(0);
 964 +                       g->reginput++;
 965 +                       break;
 966 +               case EXACTLY: {
 967 +                               register int len;
 968 +                               register char *opnd;
 969 +
 970 +                               opnd = OPERAND(scan);
 971 +                               /* Inline the first character, for speed. */
 972 +                               if (*opnd != *g->reginput)
 973 +                                       return(0);
 974 +                               len = strlen(opnd);
 975 +                               if (len > 1 && strncmp(opnd, g->reginput, len) != 0)
 976 +                                       return(0);
 977 +                               g->reginput += len;
 978 +                       }
 979 +                       break;
 980 +               case ANYOF:
 981 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) == NULL)
 982 +                               return(0);
 983 +                       g->reginput++;
 984 +                       break;
 985 +               case ANYBUT:
 986 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) != NULL)
 987 +                               return(0);
 988 +                       g->reginput++;
 989 +                       break;
 990 +               case NOTHING:
 991 +               case BACK:
 992 +                       break;
 993 +               case OPEN+1:
 994 +               case OPEN+2:
 995 +               case OPEN+3:
 996 +               case OPEN+4:
 997 +               case OPEN+5:
 998 +               case OPEN+6:
 999 +               case OPEN+7:
1000 +               case OPEN+8:
1001 +               case OPEN+9: {
1002 +                               register int no;
1003 +                               register char *save;
1004 +
1005 +                               no = OP(scan) - OPEN;
1006 +                               save = g->reginput;
1007 +
1008 +                               if (regmatch(g, next)) {
1009 +                                       /*
1010 +                                        * Don't set startp if some later
1011 +                                        * invocation of the same parentheses
1012 +                                        * already has.
1013 +                                        */
1014 +                                       if (g->regstartp[no] == NULL)
1015 +                                               g->regstartp[no] = save;
1016 +                                       return(1);
1017 +                               } else
1018 +                                       return(0);
1019 +                       }
1020 +                       break;
1021 +               case CLOSE+1:
1022 +               case CLOSE+2:
1023 +               case CLOSE+3:
1024 +               case CLOSE+4:
1025 +               case CLOSE+5:
1026 +               case CLOSE+6:
1027 +               case CLOSE+7:
1028 +               case CLOSE+8:
1029 +               case CLOSE+9:
1030 +                       {
1031 +                               register int no;
1032 +                               register char *save;
1033 +
1034 +                               no = OP(scan) - CLOSE;
1035 +                               save = g->reginput;
1036 +
1037 +                               if (regmatch(g, next)) {
1038 +                                       /*
1039 +                                        * Don't set endp if some later
1040 +                                        * invocation of the same parentheses
1041 +                                        * already has.
1042 +                                        */
1043 +                                       if (g->regendp[no] == NULL)
1044 +                                               g->regendp[no] = save;
1045 +                                       return(1);
1046 +                               } else
1047 +                                       return(0);
1048 +                       }
1049 +                       break;
1050 +               case BRANCH: {
1051 +                               register char *save;
1052 +
1053 +                               if (OP(next) != BRANCH)         /* No choice. */
1054 +                                       next = OPERAND(scan);   /* Avoid recursion. */
1055 +                               else {
1056 +                                       do {
1057 +                                               save = g->reginput;
1058 +                                               if (regmatch(g, OPERAND(scan)))
1059 +                                                       return(1);
1060 +                                               g->reginput = save;
1061 +                                               scan = regnext(g, scan);
1062 +                                       } while (scan != NULL && OP(scan) == BRANCH);
1063 +                                       return(0);
1064 +                                       /* NOTREACHED */
1065 +                               }
1066 +                       }
1067 +                       break;
1068 +               case STAR:
1069 +               case PLUS: {
1070 +                               register char nextch;
1071 +                               register int no;
1072 +                               register char *save;
1073 +                               register int min;
1074 +
1075 +                               /*
1076 +                                * Lookahead to avoid useless match attempts
1077 +                                * when we know what character comes next.
1078 +                                */
1079 +                               nextch = '\0';
1080 +                               if (OP(next) == EXACTLY)
1081 +                                       nextch = *OPERAND(next);
1082 +                               min = (OP(scan) == STAR) ? 0 : 1;
1083 +                               save = g->reginput;
1084 +                               no = regrepeat(g, OPERAND(scan));
1085 +                               while (no >= min) {
1086 +                                       /* If it could work, try it. */
1087 +                                       if (nextch == '\0' || *g->reginput == nextch)
1088 +                                               if (regmatch(g, next))
1089 +                                                       return(1);
1090 +                                       /* Couldn't or didn't -- back up. */
1091 +                                       no--;
1092 +                                       g->reginput = save + no;
1093 +                               }
1094 +                               return(0);
1095 +                       }
1096 +                       break;
1097 +               case END:
1098 +                       return(1);      /* Success! */
1099 +                       break;
1100 +               default:
1101 +                       printk("<3>Regexp: memory corruption\n");
1102 +                       return(0);
1103 +                       break;
1104 +               }
1105 +
1106 +               scan = next;
1107 +       }
1108 +
1109 +       /*
1110 +        * We get here only if there's trouble -- normally "case END" is
1111 +        * the terminating point.
1112 +        */
1113 +       printk("<3>Regexp: corrupted pointers\n");
1114 +       return(0);
1115 +}
1116 +
1117 +/*
1118 + - regrepeat - repeatedly match something simple, report how many
1119 + */
1120 +static int
1121 +regrepeat(struct match_globals *g, char *p)
1122 +{
1123 +       register int count = 0;
1124 +       register char *scan;
1125 +       register char *opnd;
1126 +
1127 +       scan = g->reginput;
1128 +       opnd = OPERAND(p);
1129 +       switch (OP(p)) {
1130 +       case ANY:
1131 +               count = strlen(scan);
1132 +               scan += count;
1133 +               break;
1134 +       case EXACTLY:
1135 +               while (*opnd == *scan) {
1136 +                       count++;
1137 +                       scan++;
1138 +               }
1139 +               break;
1140 +       case ANYOF:
1141 +               while (*scan != '\0' && strchr(opnd, *scan) != NULL) {
1142 +                       count++;
1143 +                       scan++;
1144 +               }
1145 +               break;
1146 +       case ANYBUT:
1147 +               while (*scan != '\0' && strchr(opnd, *scan) == NULL) {
1148 +                       count++;
1149 +                       scan++;
1150 +               }
1151 +               break;
1152 +       default:                /* Oh dear.  Called inappropriately. */
1153 +               printk("<3>Regexp: internal foulup\n");
1154 +               count = 0;      /* Best compromise. */
1155 +               break;
1156 +       }
1157 +       g->reginput = scan;
1158 +
1159 +       return(count);
1160 +}
1161 +
1162 +/*
1163 + - regnext - dig the "next" pointer out of a node
1164 + */
1165 +static char*
1166 +regnext(struct match_globals *g, char *p)
1167 +{
1168 +       register int offset;
1169 +
1170 +       if (p == &g->regdummy)
1171 +               return(NULL);
1172 +
1173 +       offset = NEXT(p);
1174 +       if (offset == 0)
1175 +               return(NULL);
1176 +
1177 +       if (OP(p) == BACK)
1178 +               return(p-offset);
1179 +       else
1180 +               return(p+offset);
1181 +}
1182 +
1183 +#ifdef DEBUG
1184 +
1185 +STATIC char *regprop();
1186 +
1187 +/*
1188 + - regdump - dump a regexp onto stdout in vaguely comprehensible form
1189 + */
1190 +void
1191 +regdump(regexp *r)
1192 +{
1193 +       register char *s;
1194 +       register char op = EXACTLY;     /* Arbitrary non-END op. */
1195 +       register char *next;
1196 +       /* extern char *strchr(); */
1197 +
1198 +
1199 +       s = r->program + 1;
1200 +       while (op != END) {     /* While that wasn't END last time... */
1201 +               op = OP(s);
1202 +               printf("%2d%s", s-r->program, regprop(s));      /* Where, what. */
1203 +               next = regnext(s);
1204 +               if (next == NULL)               /* Next ptr. */
1205 +                       printf("(0)");
1206 +               else
1207 +                       printf("(%d)", (s-r->program)+(next-s));
1208 +               s += 3;
1209 +               if (op == ANYOF || op == ANYBUT || op == EXACTLY) {
1210 +                       /* Literal string, where present. */
1211 +                       while (*s != '\0') {
1212 +                               putchar(*s);
1213 +                               s++;
1214 +                       }
1215 +                       s++;
1216 +               }
1217 +               putchar('\n');
1218 +       }
1219 +
1220 +       /* Header fields of interest. */
1221 +       if (r->regstart != '\0')
1222 +               printf("start `%c' ", r->regstart);
1223 +       if (r->reganch)
1224 +               printf("anchored ");
1225 +       if (r->regmust != NULL)
1226 +               printf("must have \"%s\"", r->regmust);
1227 +       printf("\n");
1228 +}
1229 +
1230 +/*
1231 + - regprop - printable representation of opcode
1232 + */
1233 +static char *
1234 +regprop(char *op)
1235 +{
1236 +#define BUFLEN 50
1237 +       register char *p;
1238 +       static char buf[BUFLEN];
1239 +
1240 +       strcpy(buf, ":");
1241 +
1242 +       switch (OP(op)) {
1243 +       case BOL:
1244 +               p = "BOL";
1245 +               break;
1246 +       case EOL:
1247 +               p = "EOL";
1248 +               break;
1249 +       case ANY:
1250 +               p = "ANY";
1251 +               break;
1252 +       case ANYOF:
1253 +               p = "ANYOF";
1254 +               break;
1255 +       case ANYBUT:
1256 +               p = "ANYBUT";
1257 +               break;
1258 +       case BRANCH:
1259 +               p = "BRANCH";
1260 +               break;
1261 +       case EXACTLY:
1262 +               p = "EXACTLY";
1263 +               break;
1264 +       case NOTHING:
1265 +               p = "NOTHING";
1266 +               break;
1267 +       case BACK:
1268 +               p = "BACK";
1269 +               break;
1270 +       case END:
1271 +               p = "END";
1272 +               break;
1273 +       case OPEN+1:
1274 +       case OPEN+2:
1275 +       case OPEN+3:
1276 +       case OPEN+4:
1277 +       case OPEN+5:
1278 +       case OPEN+6:
1279 +       case OPEN+7:
1280 +       case OPEN+8:
1281 +       case OPEN+9:
1282 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "OPEN%d", OP(op)-OPEN);
1283 +               p = NULL;
1284 +               break;
1285 +       case CLOSE+1:
1286 +       case CLOSE+2:
1287 +       case CLOSE+3:
1288 +       case CLOSE+4:
1289 +       case CLOSE+5:
1290 +       case CLOSE+6:
1291 +       case CLOSE+7:
1292 +       case CLOSE+8:
1293 +       case CLOSE+9:
1294 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "CLOSE%d", OP(op)-CLOSE);
1295 +               p = NULL;
1296 +               break;
1297 +       case STAR:
1298 +               p = "STAR";
1299 +               break;
1300 +       case PLUS:
1301 +               p = "PLUS";
1302 +               break;
1303 +       default:
1304 +               printk("<3>Regexp: corrupted opcode\n");
1305 +               break;
1306 +       }
1307 +       if (p != NULL)
1308 +               strncat(buf, p, BUFLEN-strlen(buf));
1309 +       return(buf);
1310 +}
1311 +#endif
1312 +
1313 +
1314 --- /dev/null
1315 +++ b/net/netfilter/regexp/regexp.h
1316 @@ -0,0 +1,41 @@
1317 +/*
1318 + * Definitions etc. for regexp(3) routines.
1319 + *
1320 + * Caveat:  this is V8 regexp(3) [actually, a reimplementation thereof],
1321 + * not the System V one.
1322 + */
1323 +
1324 +#ifndef REGEXP_H
1325 +#define REGEXP_H
1326 +
1327 +
1328 +/*
1329 +http://www.opensource.apple.com/darwinsource/10.3/expect-1/expect/expect.h ,
1330 +which contains a version of this library, says:
1331 +
1332 + *
1333 + * NSUBEXP must be at least 10, and no greater than 117 or the parser
1334 + * will not work properly.
1335 + *
1336 +
1337 +However, it looks rather like this library is limited to 10.  If you think
1338 +otherwise, let us know.
1339 +*/
1340 +
1341 +#define NSUBEXP  10
1342 +typedef struct regexp {
1343 +       char *startp[NSUBEXP];
1344 +       char *endp[NSUBEXP];
1345 +       char regstart;          /* Internal use only. */
1346 +       char reganch;           /* Internal use only. */
1347 +       char *regmust;          /* Internal use only. */
1348 +       int regmlen;            /* Internal use only. */
1349 +       char program[1];        /* Unwarranted chumminess with compiler. */
1350 +} regexp;
1351 +
1352 +regexp * regcomp(char *exp, int *patternsize);
1353 +int regexec(regexp *prog, char *string);
1354 +void regsub(regexp *prog, char *source, char *dest);
1355 +void regerror(char *s);
1356 +
1357 +#endif
1358 --- /dev/null
1359 +++ b/net/netfilter/regexp/regmagic.h
1360 @@ -0,0 +1,5 @@
1361 +/*
1362 + * The first byte of the regexp internal "program" is actually this magic
1363 + * number; the start node begins in the second byte.
1364 + */
1365 +#define        MAGIC   0234
1366 --- /dev/null
1367 +++ b/net/netfilter/regexp/regsub.c
1368 @@ -0,0 +1,95 @@
1369 +/*
1370 + * regsub
1371 + * @(#)regsub.c        1.3 of 2 April 86
1372 + *
1373 + *     Copyright (c) 1986 by University of Toronto.
1374 + *     Written by Henry Spencer.  Not derived from licensed software.
1375 + *
1376 + *     Permission is granted to anyone to use this software for any
1377 + *     purpose on any computer system, and to redistribute it freely,
1378 + *     subject to the following restrictions:
1379 + *
1380 + *     1. The author is not responsible for the consequences of use of
1381 + *             this software, no matter how awful, even if they arise
1382 + *             from defects in it.
1383 + *
1384 + *     2. The origin of this software must not be misrepresented, either
1385 + *             by explicit claim or by omission.
1386 + *
1387 + *     3. Altered versions must be plainly marked as such, and must not
1388 + *             be misrepresented as being the original software.
1389 + *
1390 + *
1391 + * This code was modified by Ethan Sommer to work within the kernel
1392 + * (it now uses kmalloc etc..)
1393 + *
1394 + */
1395 +#include "regexp.h"
1396 +#include "regmagic.h"
1397 +#include <linux/string.h>
1398 +
1399 +
1400 +#ifndef CHARBITS
1401 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
1402 +#else
1403 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
1404 +#endif
1405 +
1406 +#if 0
1407 +//void regerror(char * s)
1408 +//{
1409 +//        printk("regexp(3): %s", s);
1410 +//        /* NOTREACHED */
1411 +//}
1412 +#endif
1413 +
1414 +/*
1415 + - regsub - perform substitutions after a regexp match
1416 + */
1417 +void
1418 +regsub(regexp * prog, char * source, char * dest)
1419 +{
1420 +       register char *src;
1421 +       register char *dst;
1422 +       register char c;
1423 +       register int no;
1424 +       register int len;
1425 +
1426 +       /* Not necessary and gcc doesn't like it -MLS */
1427 +       /*extern char *strncpy();*/
1428 +
1429 +       if (prog == NULL || source == NULL || dest == NULL) {
1430 +               regerror("NULL parm to regsub");
1431 +               return;
1432 +       }
1433 +       if (UCHARAT(prog->program) != MAGIC) {
1434 +               regerror("damaged regexp fed to regsub");
1435 +               return;
1436 +       }
1437 +
1438 +       src = source;
1439 +       dst = dest;
1440 +       while ((c = *src++) != '\0') {
1441 +               if (c == '&')
1442 +                       no = 0;
1443 +               else if (c == '\\' && '0' <= *src && *src <= '9')
1444 +                       no = *src++ - '0';
1445 +               else
1446 +                       no = -1;
1447 +
1448 +               if (no < 0) {   /* Ordinary character. */
1449 +                       if (c == '\\' && (*src == '\\' || *src == '&'))
1450 +                               c = *src++;
1451 +                       *dst++ = c;
1452 +               } else if (prog->startp[no] != NULL && prog->endp[no] != NULL) {
1453 +                       len = prog->endp[no] - prog->startp[no];
1454 +                       (void) strncpy(dst, prog->startp[no], len);
1455 +                       dst += len;
1456 +                       if (len != 0 && *(dst-1) == '\0') {     /* strncpy hit NUL. */
1457 +                               regerror("damaged match string");
1458 +                               return;
1459 +                       }
1460 +               }
1461 +       }
1462 +       *dst++ = '\0';
1463 +}
1464 --- /dev/null
1465 +++ b/net/netfilter/xt_layer7.c
1466 @@ -0,0 +1,666 @@
1467 +/*
1468 +  Kernel module to match application layer (OSI layer 7) data in connections.
1469 +
1470 +  http://l7-filter.sf.net
1471 +
1472 +  (C) 2003-2009 Matthew Strait and Ethan Sommer.
1473 +
1474 +  This program is free software; you can redistribute it and/or
1475 +  modify it under the terms of the GNU General Public License
1476 +  as published by the Free Software Foundation; either version
1477 +  2 of the License, or (at your option) any later version.
1478 +  http://www.gnu.org/licenses/gpl.txt
1479 +
1480 +  Based on ipt_string.c (C) 2000 Emmanuel Roger <winfield@freegates.be>,
1481 +  xt_helper.c (C) 2002 Harald Welte and cls_layer7.c (C) 2003 Matthew Strait,
1482 +  Ethan Sommer, Justin Levandoski.
1483 +*/
1484 +
1485 +#include <linux/spinlock.h>
1486 +#include <linux/version.h>
1487 +#include <net/ip.h>
1488 +#include <net/tcp.h>
1489 +#include <linux/module.h>
1490 +#include <linux/skbuff.h>
1491 +#include <linux/netfilter.h>
1492 +#include <net/netfilter/nf_conntrack.h>
1493 +#include <net/netfilter/nf_conntrack_core.h>
1494 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 27)
1495 +#include <net/netfilter/nf_conntrack_extend.h>
1496 +#include <net/netfilter/nf_conntrack_acct.h>
1497 +#endif
1498 +#include <linux/netfilter/x_tables.h>
1499 +#include <linux/netfilter/xt_layer7.h>
1500 +#include <linux/ctype.h>
1501 +#include <linux/proc_fs.h>
1502 +
1503 +#include "regexp/regexp.c"
1504 +
1505 +MODULE_LICENSE("GPL");
1506 +MODULE_AUTHOR("Matthew Strait <quadong@users.sf.net>, Ethan Sommer <sommere@users.sf.net>");
1507 +MODULE_DESCRIPTION("iptables application layer match module");
1508 +MODULE_ALIAS("ipt_layer7");
1509 +MODULE_VERSION("2.21");
1510 +
1511 +static int maxdatalen = 2048; // this is the default
1512 +module_param(maxdatalen, int, 0444);
1513 +MODULE_PARM_DESC(maxdatalen, "maximum bytes of data looked at by l7-filter");
1514 +#ifdef CONFIG_NETFILTER_XT_MATCH_LAYER7_DEBUG
1515 +       #define DPRINTK(format,args...) printk(format,##args)
1516 +#else
1517 +       #define DPRINTK(format,args...)
1518 +#endif
1519 +
1520 +/* Number of packets whose data we look at.
1521 +This can be modified through /proc/net/layer7_numpackets */
1522 +static int num_packets = 10;
1523 +
1524 +static struct pattern_cache {
1525 +       char * regex_string;
1526 +       regexp * pattern;
1527 +       struct pattern_cache * next;
1528 +} * first_pattern_cache = NULL;
1529 +
1530 +DEFINE_SPINLOCK(l7_lock);
1531 +
1532 +static int total_acct_packets(struct nf_conn *ct)
1533 +{
1534 +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2, 6, 26)
1535 +       BUG_ON(ct == NULL);
1536 +       return (ct->counters[IP_CT_DIR_ORIGINAL].packets + ct->counters[IP_CT_DIR_REPLY].packets);
1537 +#else
1538 +       struct nf_conn_counter *acct;
1539 +
1540 +       BUG_ON(ct == NULL);
1541 +       acct = nf_conn_acct_find(ct);
1542 +       if (!acct)
1543 +               return 0;
1544 +       return (acct[IP_CT_DIR_ORIGINAL].packets + acct[IP_CT_DIR_REPLY].packets);
1545 +#endif
1546 +}
1547 +
1548 +#ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
1549 +/* Converts an unfriendly string into a friendly one by
1550 +replacing unprintables with periods and all whitespace with " ". */
1551 +static char * friendly_print(unsigned char * s)
1552 +{
1553 +       char * f = kmalloc(strlen(s) + 1, GFP_ATOMIC);
1554 +       int i;
1555 +
1556 +       if(!f) {
1557 +               if (net_ratelimit())
1558 +                       printk(KERN_ERR "layer7: out of memory in "
1559 +                                       "friendly_print, bailing.\n");
1560 +               return NULL;
1561 +       }
1562 +
1563 +       for(i = 0; i < strlen(s); i++){
1564 +               if(isprint(s[i]) && s[i] < 128) f[i] = s[i];
1565 +               else if(isspace(s[i]))          f[i] = ' ';
1566 +               else                            f[i] = '.';
1567 +       }
1568 +       f[i] = '\0';
1569 +       return f;
1570 +}
1571 +
1572 +static char dec2hex(int i)
1573 +{
1574 +       switch (i) {
1575 +               case 0 ... 9:
1576 +                       return (i + '0');
1577 +                       break;
1578 +               case 10 ... 15:
1579 +                       return (i - 10 + 'a');
1580 +                       break;
1581 +               default:
1582 +                       if (net_ratelimit())
1583 +                               printk("layer7: Problem in dec2hex\n");
1584 +                       return '\0';
1585 +       }
1586 +}
1587 +
1588 +static char * hex_print(unsigned char * s)
1589 +{
1590 +       char * g = kmalloc(strlen(s)*3 + 1, GFP_ATOMIC);
1591 +       int i;
1592 +
1593 +       if(!g) {
1594 +              if (net_ratelimit())
1595 +                       printk(KERN_ERR "layer7: out of memory in hex_print, "
1596 +                                       "bailing.\n");
1597 +              return NULL;
1598 +       }
1599 +
1600 +       for(i = 0; i < strlen(s); i++) {
1601 +               g[i*3    ] = dec2hex(s[i]/16);
1602 +               g[i*3 + 1] = dec2hex(s[i]%16);
1603 +               g[i*3 + 2] = ' ';
1604 +       }
1605 +       g[i*3] = '\0';
1606 +
1607 +       return g;
1608 +}
1609 +#endif // DEBUG
1610 +
1611 +/* Use instead of regcomp.  As we expect to be seeing the same regexps over and
1612 +over again, it make sense to cache the results. */
1613 +static regexp * compile_and_cache(const char * regex_string,
1614 +                                  const char * protocol)
1615 +{
1616 +       struct pattern_cache * node               = first_pattern_cache;
1617 +       struct pattern_cache * last_pattern_cache = first_pattern_cache;
1618 +       struct pattern_cache * tmp;
1619 +       unsigned int len;
1620 +
1621 +       while (node != NULL) {
1622 +               if (!strcmp(node->regex_string, regex_string))
1623 +               return node->pattern;
1624 +
1625 +               last_pattern_cache = node;/* points at the last non-NULL node */
1626 +               node = node->next;
1627 +       }
1628 +
1629 +       /* If we reach the end of the list, then we have not yet cached
1630 +          the pattern for this regex. Let's do that now.
1631 +          Be paranoid about running out of memory to avoid list corruption. */
1632 +       tmp = kmalloc(sizeof(struct pattern_cache), GFP_ATOMIC);
1633 +
1634 +       if(!tmp) {
1635 +               if (net_ratelimit())
1636 +                       printk(KERN_ERR "layer7: out of memory in "
1637 +                                       "compile_and_cache, bailing.\n");
1638 +               return NULL;
1639 +       }
1640 +
1641 +       tmp->regex_string  = kmalloc(strlen(regex_string) + 1, GFP_ATOMIC);
1642 +       tmp->pattern       = kmalloc(sizeof(struct regexp),    GFP_ATOMIC);
1643 +       tmp->next = NULL;
1644 +
1645 +       if(!tmp->regex_string || !tmp->pattern) {
1646 +               if (net_ratelimit())
1647 +                       printk(KERN_ERR "layer7: out of memory in "
1648 +                                       "compile_and_cache, bailing.\n");
1649 +               kfree(tmp->regex_string);
1650 +               kfree(tmp->pattern);
1651 +               kfree(tmp);
1652 +               return NULL;
1653 +       }
1654 +
1655 +       /* Ok.  The new node is all ready now. */
1656 +       node = tmp;
1657 +
1658 +       if(first_pattern_cache == NULL) /* list is empty */
1659 +               first_pattern_cache = node; /* make node the beginning */
1660 +       else
1661 +               last_pattern_cache->next = node; /* attach node to the end */
1662 +
1663 +       /* copy the string and compile the regex */
1664 +       len = strlen(regex_string);
1665 +       DPRINTK("About to compile this: \"%s\"\n", regex_string);
1666 +       node->pattern = regcomp((char *)regex_string, &len);
1667 +       if ( !node->pattern ) {
1668 +               if (net_ratelimit())
1669 +                       printk(KERN_ERR "layer7: Error compiling regexp "
1670 +                                       "\"%s\" (%s)\n",
1671 +                                       regex_string, protocol);
1672 +               /* pattern is now cached as NULL, so we won't try again. */
1673 +       }
1674 +
1675 +       strcpy(node->regex_string, regex_string);
1676 +       return node->pattern;
1677 +}
1678 +
1679 +static int can_handle(const struct sk_buff *skb)
1680 +{
1681 +       if(!ip_hdr(skb)) /* not IP */
1682 +               return 0;
1683 +       if(ip_hdr(skb)->protocol != IPPROTO_TCP &&
1684 +          ip_hdr(skb)->protocol != IPPROTO_UDP &&
1685 +          ip_hdr(skb)->protocol != IPPROTO_ICMP)
1686 +               return 0;
1687 +       return 1;
1688 +}
1689 +
1690 +/* Returns offset the into the skb->data that the application data starts */
1691 +static int app_data_offset(const struct sk_buff *skb)
1692 +{
1693 +       /* In case we are ported somewhere (ebtables?) where ip_hdr(skb)
1694 +       isn't set, this can be gotten from 4*(skb->data[0] & 0x0f) as well. */
1695 +       int ip_hl = 4*ip_hdr(skb)->ihl;
1696 +
1697 +       if( ip_hdr(skb)->protocol == IPPROTO_TCP ) {
1698 +               /* 12 == offset into TCP header for the header length field.
1699 +               Can't get this with skb->h.th->doff because the tcphdr
1700 +               struct doesn't get set when routing (this is confirmed to be
1701 +               true in Netfilter as well as QoS.) */
1702 +               int tcp_hl = 4*(skb->data[ip_hl + 12] >> 4);
1703 +
1704 +               return ip_hl + tcp_hl;
1705 +       } else if( ip_hdr(skb)->protocol == IPPROTO_UDP  ) {
1706 +               return ip_hl + 8; /* UDP header is always 8 bytes */
1707 +       } else if( ip_hdr(skb)->protocol == IPPROTO_ICMP ) {
1708 +               return ip_hl + 8; /* ICMP header is 8 bytes */
1709 +       } else {
1710 +               if (net_ratelimit())
1711 +                       printk(KERN_ERR "layer7: tried to handle unknown "
1712 +                                       "protocol!\n");
1713 +               return ip_hl + 8; /* something reasonable */
1714 +       }
1715 +}
1716 +
1717 +/* handles whether there's a match when we aren't appending data anymore */
1718 +static int match_no_append(struct nf_conn * conntrack,
1719 +                           struct nf_conn * master_conntrack,
1720 +                           enum ip_conntrack_info ctinfo,
1721 +                           enum ip_conntrack_info master_ctinfo,
1722 +                           const struct xt_layer7_info * info)
1723 +{
1724 +       /* If we're in here, throw the app data away */
1725 +       if(master_conntrack->layer7.app_data != NULL) {
1726 +
1727 +       #ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
1728 +               if(!master_conntrack->layer7.app_proto) {
1729 +                       char * f =
1730 +                         friendly_print(master_conntrack->layer7.app_data);
1731 +                       char * g =
1732 +                         hex_print(master_conntrack->layer7.app_data);
1733 +                       DPRINTK("\nl7-filter gave up after %d bytes "
1734 +                               "(%d packets):\n%s\n",
1735 +                               strlen(f), total_acct_packets(master_conntrack), f);
1736 +                       kfree(f);
1737 +                       DPRINTK("In hex: %s\n", g);
1738 +                       kfree(g);
1739 +               }
1740 +       #endif
1741 +
1742 +               kfree(master_conntrack->layer7.app_data);
1743 +               master_conntrack->layer7.app_data = NULL; /* don't free again */
1744 +       }
1745 +
1746 +       if(master_conntrack->layer7.app_proto){
1747 +               /* Here child connections set their .app_proto (for /proc) */
1748 +               if(!conntrack->layer7.app_proto) {
1749 +                       conntrack->layer7.app_proto =
1750 +                         kmalloc(strlen(master_conntrack->layer7.app_proto)+1,
1751 +                           GFP_ATOMIC);
1752 +                       if(!conntrack->layer7.app_proto){
1753 +                               if (net_ratelimit())
1754 +                                       printk(KERN_ERR "layer7: out of memory "
1755 +                                                       "in match_no_append, "
1756 +                                                       "bailing.\n");
1757 +                               return 1;
1758 +                       }
1759 +                       strcpy(conntrack->layer7.app_proto,
1760 +                               master_conntrack->layer7.app_proto);
1761 +               }
1762 +
1763 +               return (!strcmp(master_conntrack->layer7.app_proto,
1764 +                               info->protocol));
1765 +       }
1766 +       else {
1767 +               /* If not classified, set to "unknown" to distinguish from
1768 +               connections that are still being tested. */
1769 +               master_conntrack->layer7.app_proto =
1770 +                       kmalloc(strlen("unknown")+1, GFP_ATOMIC);
1771 +               if(!master_conntrack->layer7.app_proto){
1772 +                       if (net_ratelimit())
1773 +                               printk(KERN_ERR "layer7: out of memory in "
1774 +                                               "match_no_append, bailing.\n");
1775 +                       return 1;
1776 +               }
1777 +               strcpy(master_conntrack->layer7.app_proto, "unknown");
1778 +               return 0;
1779 +       }
1780 +}
1781 +
1782 +/* add the new app data to the conntrack.  Return number of bytes added. */
1783 +static int add_data(struct nf_conn * master_conntrack,
1784 +                    char * app_data, int appdatalen)
1785 +{
1786 +       int length = 0, i;
1787 +       int oldlength = master_conntrack->layer7.app_data_len;
1788 +
1789 +       /* This is a fix for a race condition by Deti Fliegl. However, I'm not
1790 +          clear on whether the race condition exists or whether this really
1791 +          fixes it.  I might just be being dense... Anyway, if it's not really
1792 +          a fix, all it does is waste a very small amount of time. */
1793 +       if(!master_conntrack->layer7.app_data) return 0;
1794 +
1795 +       /* Strip nulls. Make everything lower case (our regex lib doesn't
1796 +       do case insensitivity).  Add it to the end of the current data. */
1797 +       for(i = 0; i < maxdatalen-oldlength-1 &&
1798 +                  i < appdatalen; i++) {
1799 +               if(app_data[i] != '\0') {
1800 +                       /* the kernel version of tolower mungs 'upper ascii' */
1801 +                       master_conntrack->layer7.app_data[length+oldlength] =
1802 +                               isascii(app_data[i])?
1803 +                                       tolower(app_data[i]) : app_data[i];
1804 +                       length++;
1805 +               }
1806 +       }
1807 +
1808 +       master_conntrack->layer7.app_data[length+oldlength] = '\0';
1809 +       master_conntrack->layer7.app_data_len = length + oldlength;
1810 +
1811 +       return length;
1812 +}
1813 +
1814 +/* taken from drivers/video/modedb.c */
1815 +static int my_atoi(const char *s)
1816 +{
1817 +       int val = 0;
1818 +
1819 +       for (;; s++) {
1820 +               switch (*s) {
1821 +                       case '0'...'9':
1822 +                       val = 10*val+(*s-'0');
1823 +                       break;
1824 +               default:
1825 +                       return val;
1826 +               }
1827 +       }
1828 +}
1829 +
1830 +/* write out num_packets to userland. */
1831 +static int layer7_read_proc(char* page, char ** start, off_t off, int count,
1832 +                            int* eof, void * data)
1833 +{
1834 +       if(num_packets > 99 && net_ratelimit())
1835 +               printk(KERN_ERR "layer7: NOT REACHED. num_packets too big\n");
1836 +
1837 +       page[0] = num_packets/10 + '0';
1838 +       page[1] = num_packets%10 + '0';
1839 +       page[2] = '\n';
1840 +       page[3] = '\0';
1841 +
1842 +       *eof=1;
1843 +
1844 +       return 3;
1845 +}
1846 +
1847 +/* Read in num_packets from userland */
1848 +static int layer7_write_proc(struct file* file, const char* buffer,
1849 +                             unsigned long count, void *data)
1850 +{
1851 +       char * foo = kmalloc(count, GFP_ATOMIC);
1852 +
1853 +       if(!foo){
1854 +               if (net_ratelimit())
1855 +                       printk(KERN_ERR "layer7: out of memory, bailing. "
1856 +                                       "num_packets unchanged.\n");
1857 +               return count;
1858 +       }
1859 +
1860 +       if(copy_from_user(foo, buffer, count)) {
1861 +               return -EFAULT;
1862 +       }
1863 +
1864 +
1865 +       num_packets = my_atoi(foo);
1866 +       kfree (foo);
1867 +
1868 +       /* This has an arbitrary limit to make the math easier. I'm lazy.
1869 +       But anyway, 99 is a LOT! If you want more, you're doing it wrong! */
1870 +       if(num_packets > 99) {
1871 +               printk(KERN_WARNING "layer7: num_packets can't be > 99.\n");
1872 +               num_packets = 99;
1873 +       } else if(num_packets < 1) {
1874 +               printk(KERN_WARNING "layer7: num_packets can't be < 1.\n");
1875 +               num_packets = 1;
1876 +       }
1877 +
1878 +       return count;
1879 +}
1880 +
1881 +static bool
1882 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28)
1883 +match(const struct sk_buff *skbin, const struct xt_match_param *par)
1884 +#else
1885 +match(const struct sk_buff *skbin,
1886 +      const struct net_device *in,
1887 +      const struct net_device *out,
1888 +      const struct xt_match *match,
1889 +      const void *matchinfo,
1890 +      int offset,
1891 +      unsigned int protoff,
1892 +      bool *hotdrop)
1893 +#endif
1894 +{
1895 +       /* sidestep const without getting a compiler warning... */
1896 +       struct sk_buff * skb = (struct sk_buff *)skbin;
1897 +
1898 +       const struct xt_layer7_info * info =
1899 +       #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28)
1900 +               par->matchinfo;
1901 +       #else
1902 +               matchinfo;
1903 +       #endif
1904 +
1905 +       enum ip_conntrack_info master_ctinfo, ctinfo;
1906 +       struct nf_conn *master_conntrack, *conntrack;
1907 +       unsigned char * app_data;
1908 +       unsigned int pattern_result, appdatalen;
1909 +       regexp * comppattern;
1910 +
1911 +       /* Be paranoid/incompetent - lock the entire match function. */
1912 +       spin_lock_bh(&l7_lock);
1913 +
1914 +       if(!can_handle(skb)){
1915 +               DPRINTK("layer7: This is some protocol I can't handle.\n");
1916 +               spin_unlock_bh(&l7_lock);
1917 +               return info->invert;
1918 +       }
1919 +
1920 +       /* Treat parent & all its children together as one connection, except
1921 +       for the purpose of setting conntrack->layer7.app_proto in the actual
1922 +       connection. This makes /proc/net/ip_conntrack more satisfying. */
1923 +       if(!(conntrack = nf_ct_get(skb, &ctinfo)) ||
1924 +          !(master_conntrack=nf_ct_get(skb,&master_ctinfo))){
1925 +               DPRINTK("layer7: couldn't get conntrack.\n");
1926 +               spin_unlock_bh(&l7_lock);
1927 +               return info->invert;
1928 +       }
1929 +
1930 +       /* Try to get a master conntrack (and its master etc) for FTP, etc. */
1931 +       while (master_ct(master_conntrack) != NULL)
1932 +               master_conntrack = master_ct(master_conntrack);
1933 +
1934 +       /* if we've classified it or seen too many packets */
1935 +       if(total_acct_packets(master_conntrack) > num_packets ||
1936 +          master_conntrack->layer7.app_proto) {
1937 +
1938 +               pattern_result = match_no_append(conntrack, master_conntrack,
1939 +                                                ctinfo, master_ctinfo, info);
1940 +
1941 +               /* skb->cb[0] == seen. Don't do things twice if there are
1942 +               multiple l7 rules. I'm not sure that using cb for this purpose
1943 +               is correct, even though it says "put your private variables
1944 +               there". But it doesn't look like it is being used for anything
1945 +               else in the skbs that make it here. */
1946 +               skb->cb[0] = 1; /* marking it seen here's probably irrelevant */
1947 +
1948 +               spin_unlock_bh(&l7_lock);
1949 +               return (pattern_result ^ info->invert);
1950 +       }
1951 +
1952 +       if(skb_is_nonlinear(skb)){
1953 +               if(skb_linearize(skb) != 0){
1954 +                       if (net_ratelimit())
1955 +                               printk(KERN_ERR "layer7: failed to linearize "
1956 +                                               "packet, bailing.\n");
1957 +                       spin_unlock_bh(&l7_lock);
1958 +                       return info->invert;
1959 +               }
1960 +       }
1961 +
1962 +       /* now that the skb is linearized, it's safe to set these. */
1963 +       app_data = skb->data + app_data_offset(skb);
1964 +       appdatalen = skb_tail_pointer(skb) - app_data;
1965 +
1966 +       /* the return value gets checked later, when we're ready to use it */
1967 +       comppattern = compile_and_cache(info->pattern, info->protocol);
1968 +
1969 +       /* On the first packet of a connection, allocate space for app data */
1970 +       if(total_acct_packets(master_conntrack) == 1 && !skb->cb[0] &&
1971 +          !master_conntrack->layer7.app_data){
1972 +               master_conntrack->layer7.app_data =
1973 +                       kmalloc(maxdatalen, GFP_ATOMIC);
1974 +               if(!master_conntrack->layer7.app_data){
1975 +                       if (net_ratelimit())
1976 +                               printk(KERN_ERR "layer7: out of memory in "
1977 +                                               "match, bailing.\n");
1978 +                       spin_unlock_bh(&l7_lock);
1979 +                       return info->invert;
1980 +               }
1981 +
1982 +               master_conntrack->layer7.app_data[0] = '\0';
1983 +       }
1984 +
1985 +       /* Can be here, but unallocated, if numpackets is increased near
1986 +       the beginning of a connection */
1987 +       if(master_conntrack->layer7.app_data == NULL){
1988 +               spin_unlock_bh(&l7_lock);
1989 +               return info->invert; /* unmatched */
1990 +       }
1991 +
1992 +       if(!skb->cb[0]){
1993 +               int newbytes;
1994 +               newbytes = add_data(master_conntrack, app_data, appdatalen);
1995 +
1996 +               if(newbytes == 0) { /* didn't add any data */
1997 +                       skb->cb[0] = 1;
1998 +                       /* Didn't match before, not going to match now */
1999 +                       spin_unlock_bh(&l7_lock);
2000 +                       return info->invert;
2001 +               }
2002 +       }
2003 +
2004 +       /* If looking for "unknown", then never match.  "Unknown" means that
2005 +       we've given up; we're still trying with these packets. */
2006 +       if(!strcmp(info->protocol, "unknown")) {
2007 +               pattern_result = 0;
2008 +       /* If looking for "unset", then always match. "Unset" means that we
2009 +       haven't yet classified the connection. */
2010 +       } else if(!strcmp(info->protocol, "unset")) {
2011 +               pattern_result = 2;
2012 +               DPRINTK("layer7: matched unset: not yet classified "
2013 +                       "(%d/%d packets)\n",
2014 +                        total_acct_packets(master_conntrack), num_packets);
2015 +       /* If the regexp failed to compile, don't bother running it */
2016 +       } else if(comppattern &&
2017 +                 regexec(comppattern, master_conntrack->layer7.app_data)){
2018 +               DPRINTK("layer7: matched %s\n", info->protocol);
2019 +               pattern_result = 1;
2020 +       } else pattern_result = 0;
2021 +
2022 +       if(pattern_result == 1) {
2023 +               master_conntrack->layer7.app_proto =
2024 +                       kmalloc(strlen(info->protocol)+1, GFP_ATOMIC);
2025 +               if(!master_conntrack->layer7.app_proto){
2026 +                       if (net_ratelimit())
2027 +                               printk(KERN_ERR "layer7: out of memory in "
2028 +                                               "match, bailing.\n");
2029 +                       spin_unlock_bh(&l7_lock);
2030 +                       return (pattern_result ^ info->invert);
2031 +               }
2032 +               strcpy(master_conntrack->layer7.app_proto, info->protocol);
2033 +       } else if(pattern_result > 1) { /* cleanup from "unset" */
2034 +               pattern_result = 1;
2035 +       }
2036 +
2037 +       /* mark the packet seen */
2038 +       skb->cb[0] = 1;
2039 +
2040 +       spin_unlock_bh(&l7_lock);
2041 +       return (pattern_result ^ info->invert);
2042 +}
2043 +
2044 +// load nf_conntrack_ipv4
2045 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28)
2046 +static bool check(const struct xt_mtchk_param *par)
2047 +{
2048 +        if (nf_ct_l3proto_try_module_get(par->match->family) < 0) {
2049 +                printk(KERN_WARNING "can't load conntrack support for "
2050 +                                    "proto=%d\n", par->match->family);
2051 +#else
2052 +static bool check(const char *tablename, const void *inf,
2053 +                const struct xt_match *match, void *matchinfo,
2054 +                unsigned int hook_mask)
2055 +{
2056 +        if (nf_ct_l3proto_try_module_get(match->family) < 0) {
2057 +                printk(KERN_WARNING "can't load conntrack support for "
2058 +                                    "proto=%d\n", match->family);
2059 +#endif
2060 +                return 0;
2061 +        }
2062 +       return 1;
2063 +}
2064 +
2065 +
2066 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28)
2067 +       static void destroy(const struct xt_mtdtor_param *par)
2068 +       {
2069 +               nf_ct_l3proto_module_put(par->match->family);
2070 +       }
2071 +#else
2072 +       static void destroy(const struct xt_match *match, void *matchinfo)
2073 +       {
2074 +               nf_ct_l3proto_module_put(match->family);
2075 +       }
2076 +#endif
2077 +
2078 +static struct xt_match xt_layer7_match[] __read_mostly = {
2079 +{
2080 +       .name           = "layer7",
2081 +       .family         = AF_INET,
2082 +       .checkentry     = check,
2083 +       .match          = match,
2084 +       .destroy        = destroy,
2085 +       .matchsize      = sizeof(struct xt_layer7_info),
2086 +       .me             = THIS_MODULE
2087 +}
2088 +};
2089 +
2090 +static void layer7_cleanup_proc(void)
2091 +{
2092 +       remove_proc_entry("layer7_numpackets", init_net.proc_net);
2093 +}
2094 +
2095 +/* register the proc file */
2096 +static void layer7_init_proc(void)
2097 +{
2098 +       struct proc_dir_entry* entry;
2099 +       entry = create_proc_entry("layer7_numpackets", 0644, init_net.proc_net);
2100 +       entry->read_proc = layer7_read_proc;
2101 +       entry->write_proc = layer7_write_proc;
2102 +}
2103 +
2104 +static int __init xt_layer7_init(void)
2105 +{
2106 +       need_conntrack();
2107 +
2108 +       layer7_init_proc();
2109 +       if(maxdatalen < 1) {
2110 +               printk(KERN_WARNING "layer7: maxdatalen can't be < 1, "
2111 +                       "using 1\n");
2112 +               maxdatalen = 1;
2113 +       }
2114 +       /* This is not a hard limit.  It's just here to prevent people from
2115 +       bringing their slow machines to a grinding halt. */
2116 +       else if(maxdatalen > 65536) {
2117 +               printk(KERN_WARNING "layer7: maxdatalen can't be > 65536, "
2118 +                       "using 65536\n");
2119 +               maxdatalen = 65536;
2120 +       }
2121 +       return xt_register_matches(xt_layer7_match,
2122 +                                  ARRAY_SIZE(xt_layer7_match));
2123 +}
2124 +
2125 +static void __exit xt_layer7_fini(void)
2126 +{
2127 +       layer7_cleanup_proc();
2128 +       xt_unregister_matches(xt_layer7_match, ARRAY_SIZE(xt_layer7_match));
2129 +}
2130 +
2131 +module_init(xt_layer7_init);
2132 +module_exit(xt_layer7_fini);