target/linux/generic-2.6/patches-2.6.21/100-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 invert:1;
  13 +    char pattern[MAX_PATTERN_LEN];
  14 +};
  15 +
  16 +#endif /* _XT_LAYER7_H */
  17 --- a/include/net/netfilter/nf_conntrack.h
  18 +++ b/include/net/netfilter/nf_conntrack.h
  19 @@ -128,6 +128,22 @@ 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/ipv4/netfilter/nf_conntrack_l3proto_ipv4_compat.c
  43 +++ b/net/ipv4/netfilter/nf_conntrack_l3proto_ipv4_compat.c
  44 @@ -163,6 +163,12 @@ static int ct_seq_show(struct seq_file *
  45                 return -ENOSPC;
  46  #endif
  47
  48 +#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
  49 +       if(ct->layer7.app_proto)
  50 +               if(seq_printf(s, "l7proto=%s ", ct->layer7.app_proto))
  51 +                       return -ENOSPC;
  52 +#endif
  53 +
  54         if (seq_printf(s, "use=%u\n", atomic_read(&ct->ct_general.use)))
  55                 return -ENOSPC;
  56
  57 --- a/net/netfilter/Kconfig
  58 +++ b/net/netfilter/Kconfig
  59 @@ -640,6 +640,27 @@ config NETFILTER_XT_MATCH_STATE
  60
  61           To compile it as a module, choose M here.  If unsure, say N.
  62
  63 +config NETFILTER_XT_MATCH_LAYER7
  64 +       tristate '"layer7" match support'
  65 +       depends on NETFILTER_XTABLES
  66 +       depends on IP_NF_CONNTRACK || NF_CONNTRACK
  67 +       depends on NF_CT_ACCT
  68 +       help
  69 +         Say Y if you want to be able to classify connections (and their
  70 +         packets) based on regular expression matching of their application
  71 +         layer data.   This is one way to classify applications such as
  72 +         peer-to-peer filesharing systems that do not always use the same
  73 +         port.
  74 +
  75 +         To compile it as a module, choose M here.  If unsure, say N.
  76 +
  77 +config NETFILTER_XT_MATCH_LAYER7_DEBUG
  78 +        bool 'layer7 debugging output'
  79 +        depends on NETFILTER_XT_MATCH_LAYER7
  80 +        help
  81 +          Say Y to get lots of debugging output.
  82 +
  83 +
  84  config NETFILTER_XT_MATCH_STATISTIC
  85         tristate '"statistic" match support'
  86         depends on NETFILTER_XTABLES
  87 --- a/net/netfilter/Makefile
  88 +++ b/net/netfilter/Makefile
  89 @@ -68,6 +68,7 @@ obj-$(CONFIG_NETFILTER_XT_MATCH_QUOTA) +
  90  obj-$(CONFIG_NETFILTER_XT_MATCH_REALM) += xt_realm.o
  91  obj-$(CONFIG_NETFILTER_XT_MATCH_SCTP) += xt_sctp.o
  92  obj-$(CONFIG_NETFILTER_XT_MATCH_STATE) += xt_state.o
  93 +obj-$(CONFIG_NETFILTER_XT_MATCH_LAYER7) += xt_layer7.o
  94  obj-$(CONFIG_NETFILTER_XT_MATCH_STATISTIC) += xt_statistic.o
  95  obj-$(CONFIG_NETFILTER_XT_MATCH_STRING) += xt_string.o
  96  obj-$(CONFIG_NETFILTER_XT_MATCH_TCPMSS) += xt_tcpmss.o
  97 --- a/net/netfilter/nf_conntrack_core.c
  98 +++ b/net/netfilter/nf_conntrack_core.c
  99 @@ -352,6 +352,14 @@ destroy_conntrack(struct nf_conntrack *n
 100          * too. */
 101         nf_ct_remove_expectations(ct);
 102
 103 +       #if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
 104 +       if(ct->layer7.app_proto)
 105 +               kfree(ct->layer7.app_proto);
 106 +       if(ct->layer7.app_data)
 107 +       kfree(ct->layer7.app_data);
 108 +       #endif
 109 +
 110 +
 111         /* We overload first tuple to link into unconfirmed list. */
 112         if (!nf_ct_is_confirmed(ct)) {
 113                 BUG_ON(list_empty(&ct->tuplehash[IP_CT_DIR_ORIGINAL].list));
 114 --- a/net/netfilter/nf_conntrack_standalone.c
 115 +++ b/net/netfilter/nf_conntrack_standalone.c
 116 @@ -195,6 +195,11 @@ static int ct_seq_show(struct seq_file *
 117                 return -ENOSPC;
 118  #endif
 119
 120 +#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
 121 +       if(conntrack->layer7.app_proto)
 122 +               if(seq_printf(s, "l7proto=%s ",conntrack->layer7.app_proto))
 123 +                       return -ENOSPC;
 124 +#endif
 125         if (seq_printf(s, "use=%u\n", atomic_read(&conntrack->ct_general.use)))
 126                 return -ENOSPC;
 127
 128 --- /dev/null
 129 +++ b/net/netfilter/regexp/regexp.c
 130 @@ -0,0 +1,1197 @@
 131 +/*
 132 + * regcomp and regexec -- regsub and regerror are elsewhere
 133 + * @(#)regexp.c        1.3 of 18 April 87
 134 + *
 135 + *     Copyright (c) 1986 by University of Toronto.
 136 + *     Written by Henry Spencer.  Not derived from licensed software.
 137 + *
 138 + *     Permission is granted to anyone to use this software for any
 139 + *     purpose on any computer system, and to redistribute it freely,
 140 + *     subject to the following restrictions:
 141 + *
 142 + *     1. The author is not responsible for the consequences of use of
 143 + *             this software, no matter how awful, even if they arise
 144 + *             from defects in it.
 145 + *
 146 + *     2. The origin of this software must not be misrepresented, either
 147 + *             by explicit claim or by omission.
 148 + *
 149 + *     3. Altered versions must be plainly marked as such, and must not
 150 + *             be misrepresented as being the original software.
 151 + *
 152 + * Beware that some of this code is subtly aware of the way operator
 153 + * precedence is structured in regular expressions.  Serious changes in
 154 + * regular-expression syntax might require a total rethink.
 155 + *
 156 + * This code was modified by Ethan Sommer to work within the kernel
 157 + * (it now uses kmalloc etc..)
 158 + *
 159 + * Modified slightly by Matthew Strait to use more modern C.
 160 + */
 161 +
 162 +#include "regexp.h"
 163 +#include "regmagic.h"
 164 +
 165 +/* added by ethan and matt.  Lets it work in both kernel and user space.
 166 +(So iptables can use it, for instance.)  Yea, it goes both ways... */
 167 +#if __KERNEL__
 168 +  #define malloc(foo) kmalloc(foo,GFP_ATOMIC)
 169 +#else
 170 +  #define printk(format,args...) printf(format,##args)
 171 +#endif
 172 +
 173 +void regerror(char * s)
 174 +{
 175 +        printk("<3>Regexp: %s\n", s);
 176 +        /* NOTREACHED */
 177 +}
 178 +
 179 +/*
 180 + * The "internal use only" fields in regexp.h are present to pass info from
 181 + * compile to execute that permits the execute phase to run lots faster on
 182 + * simple cases.  They are:
 183 + *
 184 + * regstart    char that must begin a match; '\0' if none obvious
 185 + * reganch     is the match anchored (at beginning-of-line only)?
 186 + * regmust     string (pointer into program) that match must include, or NULL
 187 + * regmlen     length of regmust string
 188 + *
 189 + * Regstart and reganch permit very fast decisions on suitable starting points
 190 + * for a match, cutting down the work a lot.  Regmust permits fast rejection
 191 + * of lines that cannot possibly match.  The regmust tests are costly enough
 192 + * that regcomp() supplies a regmust only if the r.e. contains something
 193 + * potentially expensive (at present, the only such thing detected is * or +
 194 + * at the start of the r.e., which can involve a lot of backup).  Regmlen is
 195 + * supplied because the test in regexec() needs it and regcomp() is computing
 196 + * it anyway.
 197 + */
 198 +
 199 +/*
 200 + * Structure for regexp "program".  This is essentially a linear encoding
 201 + * of a nondeterministic finite-state machine (aka syntax charts or
 202 + * "railroad normal form" in parsing technology).  Each node is an opcode
 203 + * plus a "next" pointer, possibly plus an operand.  "Next" pointers of
 204 + * all nodes except BRANCH implement concatenation; a "next" pointer with
 205 + * a BRANCH on both ends of it is connecting two alternatives.  (Here we
 206 + * have one of the subtle syntax dependencies:  an individual BRANCH (as
 207 + * opposed to a collection of them) is never concatenated with anything
 208 + * because of operator precedence.)  The operand of some types of node is
 209 + * a literal string; for others, it is a node leading into a sub-FSM.  In
 210 + * particular, the operand of a BRANCH node is the first node of the branch.
 211 + * (NB this is *not* a tree structure:  the tail of the branch connects
 212 + * to the thing following the set of BRANCHes.)  The opcodes are:
 213 + */
 214 +
 215 +/* definition  number  opnd?   meaning */
 216 +#define        END     0       /* no   End of program. */
 217 +#define        BOL     1       /* no   Match "" at beginning of line. */
 218 +#define        EOL     2       /* no   Match "" at end of line. */
 219 +#define        ANY     3       /* no   Match any one character. */
 220 +#define        ANYOF   4       /* str  Match any character in this string. */
 221 +#define        ANYBUT  5       /* str  Match any character not in this string. */
 222 +#define        BRANCH  6       /* node Match this alternative, or the next... */
 223 +#define        BACK    7       /* no   Match "", "next" ptr points backward. */
 224 +#define        EXACTLY 8       /* str  Match this string. */
 225 +#define        NOTHING 9       /* no   Match empty string. */
 226 +#define        STAR    10      /* node Match this (simple) thing 0 or more times. */
 227 +#define        PLUS    11      /* node Match this (simple) thing 1 or more times. */
 228 +#define        OPEN    20      /* no   Mark this point in input as start of #n. */
 229 +                       /*      OPEN+1 is number 1, etc. */
 230 +#define        CLOSE   30      /* no   Analogous to OPEN. */
 231 +
 232 +/*
 233 + * Opcode notes:
 234 + *
 235 + * BRANCH      The set of branches constituting a single choice are hooked
 236 + *             together with their "next" pointers, since precedence prevents
 237 + *             anything being concatenated to any individual branch.  The
 238 + *             "next" pointer of the last BRANCH in a choice points to the
 239 + *             thing following the whole choice.  This is also where the
 240 + *             final "next" pointer of each individual branch points; each
 241 + *             branch starts with the operand node of a BRANCH node.
 242 + *
 243 + * BACK                Normal "next" pointers all implicitly point forward; BACK
 244 + *             exists to make loop structures possible.
 245 + *
 246 + * STAR,PLUS   '?', and complex '*' and '+', are implemented as circular
 247 + *             BRANCH structures using BACK.  Simple cases (one character
 248 + *             per match) are implemented with STAR and PLUS for speed
 249 + *             and to minimize recursive plunges.
 250 + *
 251 + * OPEN,CLOSE  ...are numbered at compile time.
 252 + */
 253 +
 254 +/*
 255 + * A node is one char of opcode followed by two chars of "next" pointer.
 256 + * "Next" pointers are stored as two 8-bit pieces, high order first.  The
 257 + * value is a positive offset from the opcode of the node containing it.
 258 + * An operand, if any, simply follows the node.  (Note that much of the
 259 + * code generation knows about this implicit relationship.)
 260 + *
 261 + * Using two bytes for the "next" pointer is vast overkill for most things,
 262 + * but allows patterns to get big without disasters.
 263 + */
 264 +#define        OP(p)   (*(p))
 265 +#define        NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377))
 266 +#define        OPERAND(p)      ((p) + 3)
 267 +
 268 +/*
 269 + * See regmagic.h for one further detail of program structure.
 270 + */
 271 +
 272 +
 273 +/*
 274 + * Utility definitions.
 275 + */
 276 +#ifndef CHARBITS
 277 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
 278 +#else
 279 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
 280 +#endif
 281 +
 282 +#define        FAIL(m) { regerror(m); return(NULL); }
 283 +#define        ISMULT(c)       ((c) == '*' || (c) == '+' || (c) == '?')
 284 +#define        META    "^$.[()|?+*\\"
 285 +
 286 +/*
 287 + * Flags to be passed up and down.
 288 + */
 289 +#define        HASWIDTH        01      /* Known never to match null string. */
 290 +#define        SIMPLE          02      /* Simple enough to be STAR/PLUS operand. */
 291 +#define        SPSTART         04      /* Starts with * or +. */
 292 +#define        WORST           0       /* Worst case. */
 293 +
 294 +/*
 295 + * Global work variables for regcomp().
 296 + */
 297 +struct match_globals {
 298 +char *reginput;                /* String-input pointer. */
 299 +char *regbol;          /* Beginning of input, for ^ check. */
 300 +char **regstartp;      /* Pointer to startp array. */
 301 +char **regendp;                /* Ditto for endp. */
 302 +char *regparse;                /* Input-scan pointer. */
 303 +int regnpar;           /* () count. */
 304 +char regdummy;
 305 +char *regcode;         /* Code-emit pointer; &regdummy = don't. */
 306 +long regsize;          /* Code size. */
 307 +};
 308 +
 309 +/*
 310 + * Forward declarations for regcomp()'s friends.
 311 + */
 312 +#ifndef STATIC
 313 +#define        STATIC  static
 314 +#endif
 315 +STATIC char *reg(struct match_globals *g, int paren,int *flagp);
 316 +STATIC char *regbranch(struct match_globals *g, int *flagp);
 317 +STATIC char *regpiece(struct match_globals *g, int *flagp);
 318 +STATIC char *regatom(struct match_globals *g, int *flagp);
 319 +STATIC char *regnode(struct match_globals *g, char op);
 320 +STATIC char *regnext(struct match_globals *g, char *p);
 321 +STATIC void regc(struct match_globals *g, char b);
 322 +STATIC void reginsert(struct match_globals *g, char op, char *opnd);
 323 +STATIC void regtail(struct match_globals *g, char *p, char *val);
 324 +STATIC void regoptail(struct match_globals *g, char *p, char *val);
 325 +
 326 +
 327 +__kernel_size_t my_strcspn(const char *s1,const char *s2)
 328 +{
 329 +        char *scan1;
 330 +        char *scan2;
 331 +        int count;
 332 +
 333 +        count = 0;
 334 +        for (scan1 = (char *)s1; *scan1 != '\0'; scan1++) {
 335 +                for (scan2 = (char *)s2; *scan2 != '\0';)       /* ++ moved down. */
 336 +                        if (*scan1 == *scan2++)
 337 +                                return(count);
 338 +                count++;
 339 +        }
 340 +        return(count);
 341 +}
 342 +
 343 +/*
 344 + - regcomp - compile a regular expression into internal code
 345 + *
 346 + * We can't allocate space until we know how big the compiled form will be,
 347 + * but we can't compile it (and thus know how big it is) until we've got a
 348 + * place to put the code.  So we cheat:  we compile it twice, once with code
 349 + * generation turned off and size counting turned on, and once "for real".
 350 + * This also means that we don't allocate space until we are sure that the
 351 + * thing really will compile successfully, and we never have to move the
 352 + * code and thus invalidate pointers into it.  (Note that it has to be in
 353 + * one piece because free() must be able to free it all.)
 354 + *
 355 + * Beware that the optimization-preparation code in here knows about some
 356 + * of the structure of the compiled regexp.
 357 + */
 358 +regexp *
 359 +regcomp(char *exp,int *patternsize)
 360 +{
 361 +       register regexp *r;
 362 +       register char *scan;
 363 +       register char *longest;
 364 +       register int len;
 365 +       int flags;
 366 +       struct match_globals g;
 367 +
 368 +       /* commented out by ethan
 369 +          extern char *malloc();
 370 +       */
 371 +
 372 +       if (exp == NULL)
 373 +               FAIL("NULL argument");
 374 +
 375 +       /* First pass: determine size, legality. */
 376 +       g.regparse = exp;
 377 +       g.regnpar = 1;
 378 +       g.regsize = 0L;
 379 +       g.regcode = &g.regdummy;
 380 +       regc(&g, MAGIC);
 381 +       if (reg(&g, 0, &flags) == NULL)
 382 +               return(NULL);
 383 +
 384 +       /* Small enough for pointer-storage convention? */
 385 +       if (g.regsize >= 32767L)                /* Probably could be 65535L. */
 386 +               FAIL("regexp too big");
 387 +
 388 +       /* Allocate space. */
 389 +       *patternsize=sizeof(regexp) + (unsigned)g.regsize;
 390 +       r = (regexp *)malloc(sizeof(regexp) + (unsigned)g.regsize);
 391 +       if (r == NULL)
 392 +               FAIL("out of space");
 393 +
 394 +       /* Second pass: emit code. */
 395 +       g.regparse = exp;
 396 +       g.regnpar = 1;
 397 +       g.regcode = r->program;
 398 +       regc(&g, MAGIC);
 399 +       if (reg(&g, 0, &flags) == NULL)
 400 +               return(NULL);
 401 +
 402 +       /* Dig out information for optimizations. */
 403 +       r->regstart = '\0';     /* Worst-case defaults. */
 404 +       r->reganch = 0;
 405 +       r->regmust = NULL;
 406 +       r->regmlen = 0;
 407 +       scan = r->program+1;                    /* First BRANCH. */
 408 +       if (OP(regnext(&g, scan)) == END) {             /* Only one top-level choice. */
 409 +               scan = OPERAND(scan);
 410 +
 411 +               /* Starting-point info. */
 412 +               if (OP(scan) == EXACTLY)
 413 +                       r->regstart = *OPERAND(scan);
 414 +               else if (OP(scan) == BOL)
 415 +                       r->reganch++;
 416 +
 417 +               /*
 418 +                * If there's something expensive in the r.e., find the
 419 +                * longest literal string that must appear and make it the
 420 +                * regmust.  Resolve ties in favor of later strings, since
 421 +                * the regstart check works with the beginning of the r.e.
 422 +                * and avoiding duplication strengthens checking.  Not a
 423 +                * strong reason, but sufficient in the absence of others.
 424 +                */
 425 +               if (flags&SPSTART) {
 426 +                       longest = NULL;
 427 +                       len = 0;
 428 +                       for (; scan != NULL; scan = regnext(&g, scan))
 429 +                               if (OP(scan) == EXACTLY && strlen(OPERAND(scan)) >= len) {
 430 +                                       longest = OPERAND(scan);
 431 +                                       len = strlen(OPERAND(scan));
 432 +                               }
 433 +                       r->regmust = longest;
 434 +                       r->regmlen = len;
 435 +               }
 436 +       }
 437 +
 438 +       return(r);
 439 +}
 440 +
 441 +/*
 442 + - reg - regular expression, i.e. main body or parenthesized thing
 443 + *
 444 + * Caller must absorb opening parenthesis.
 445 + *
 446 + * Combining parenthesis handling with the base level of regular expression
 447 + * is a trifle forced, but the need to tie the tails of the branches to what
 448 + * follows makes it hard to avoid.
 449 + */
 450 +static char *
 451 +reg(struct match_globals *g, int paren, int *flagp /* Parenthesized? */ )
 452 +{
 453 +       register char *ret;
 454 +       register char *br;
 455 +       register char *ender;
 456 +       register int parno = 0; /* 0 makes gcc happy */
 457 +       int flags;
 458 +
 459 +       *flagp = HASWIDTH;      /* Tentatively. */
 460 +
 461 +       /* Make an OPEN node, if parenthesized. */
 462 +       if (paren) {
 463 +               if (g->regnpar >= NSUBEXP)
 464 +                       FAIL("too many ()");
 465 +               parno = g->regnpar;
 466 +               g->regnpar++;
 467 +               ret = regnode(g, OPEN+parno);
 468 +       } else
 469 +               ret = NULL;
 470 +
 471 +       /* Pick up the branches, linking them together. */
 472 +       br = regbranch(g, &flags);
 473 +       if (br == NULL)
 474 +               return(NULL);
 475 +       if (ret != NULL)
 476 +               regtail(g, ret, br);    /* OPEN -> first. */
 477 +       else
 478 +               ret = br;
 479 +       if (!(flags&HASWIDTH))
 480 +               *flagp &= ~HASWIDTH;
 481 +       *flagp |= flags&SPSTART;
 482 +       while (*g->regparse == '|') {
 483 +               g->regparse++;
 484 +               br = regbranch(g, &flags);
 485 +               if (br == NULL)
 486 +                       return(NULL);
 487 +               regtail(g, ret, br);    /* BRANCH -> BRANCH. */
 488 +               if (!(flags&HASWIDTH))
 489 +                       *flagp &= ~HASWIDTH;
 490 +               *flagp |= flags&SPSTART;
 491 +       }
 492 +
 493 +       /* Make a closing node, and hook it on the end. */
 494 +       ender = regnode(g, (paren) ? CLOSE+parno : END);
 495 +       regtail(g, ret, ender);
 496 +
 497 +       /* Hook the tails of the branches to the closing node. */
 498 +       for (br = ret; br != NULL; br = regnext(g, br))
 499 +               regoptail(g, br, ender);
 500 +
 501 +       /* Check for proper termination. */
 502 +       if (paren && *g->regparse++ != ')') {
 503 +               FAIL("unmatched ()");
 504 +       } else if (!paren && *g->regparse != '\0') {
 505 +               if (*g->regparse == ')') {
 506 +                       FAIL("unmatched ()");
 507 +               } else
 508 +                       FAIL("junk on end");    /* "Can't happen". */
 509 +               /* NOTREACHED */
 510 +       }
 511 +
 512 +       return(ret);
 513 +}
 514 +
 515 +/*
 516 + - regbranch - one alternative of an | operator
 517 + *
 518 + * Implements the concatenation operator.
 519 + */
 520 +static char *
 521 +regbranch(struct match_globals *g, int *flagp)
 522 +{
 523 +       register char *ret;
 524 +       register char *chain;
 525 +       register char *latest;
 526 +       int flags;
 527 +
 528 +       *flagp = WORST;         /* Tentatively. */
 529 +
 530 +       ret = regnode(g, BRANCH);
 531 +       chain = NULL;
 532 +       while (*g->regparse != '\0' && *g->regparse != '|' && *g->regparse != ')') {
 533 +               latest = regpiece(g, &flags);
 534 +               if (latest == NULL)
 535 +                       return(NULL);
 536 +               *flagp |= flags&HASWIDTH;
 537 +               if (chain == NULL)      /* First piece. */
 538 +                       *flagp |= flags&SPSTART;
 539 +               else
 540 +                       regtail(g, chain, latest);
 541 +               chain = latest;
 542 +       }
 543 +       if (chain == NULL)      /* Loop ran zero times. */
 544 +               (void) regnode(g, NOTHING);
 545 +
 546 +       return(ret);
 547 +}
 548 +
 549 +/*
 550 + - regpiece - something followed by possible [*+?]
 551 + *
 552 + * Note that the branching code sequences used for ? and the general cases
 553 + * of * and + are somewhat optimized:  they use the same NOTHING node as
 554 + * both the endmarker for their branch list and the body of the last branch.
 555 + * It might seem that this node could be dispensed with entirely, but the
 556 + * endmarker role is not redundant.
 557 + */
 558 +static char *
 559 +regpiece(struct match_globals *g, int *flagp)
 560 +{
 561 +       register char *ret;
 562 +       register char op;
 563 +       register char *next;
 564 +       int flags;
 565 +
 566 +       ret = regatom(g, &flags);
 567 +       if (ret == NULL)
 568 +               return(NULL);
 569 +
 570 +       op = *g->regparse;
 571 +       if (!ISMULT(op)) {
 572 +               *flagp = flags;
 573 +               return(ret);
 574 +       }
 575 +
 576 +       if (!(flags&HASWIDTH) && op != '?')
 577 +               FAIL("*+ operand could be empty");
 578 +       *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
 579 +
 580 +       if (op == '*' && (flags&SIMPLE))
 581 +               reginsert(g, STAR, ret);
 582 +       else if (op == '*') {
 583 +               /* Emit x* as (x&|), where & means "self". */
 584 +               reginsert(g, BRANCH, ret);                      /* Either x */
 585 +               regoptail(g, ret, regnode(g, BACK));            /* and loop */
 586 +               regoptail(g, ret, ret);                 /* back */
 587 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
 588 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
 589 +       } else if (op == '+' && (flags&SIMPLE))
 590 +               reginsert(g, PLUS, ret);
 591 +       else if (op == '+') {
 592 +               /* Emit x+ as x(&|), where & means "self". */
 593 +               next = regnode(g, BRANCH);                      /* Either */
 594 +               regtail(g, ret, next);
 595 +               regtail(g, regnode(g, BACK), ret);              /* loop back */
 596 +               regtail(g, next, regnode(g, BRANCH));           /* or */
 597 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
 598 +       } else if (op == '?') {
 599 +               /* Emit x? as (x|) */
 600 +               reginsert(g, BRANCH, ret);                      /* Either x */
 601 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
 602 +               next = regnode(g, NOTHING);             /* null. */
 603 +               regtail(g, ret, next);
 604 +               regoptail(g, ret, next);
 605 +       }
 606 +       g->regparse++;
 607 +       if (ISMULT(*g->regparse))
 608 +               FAIL("nested *?+");
 609 +
 610 +       return(ret);
 611 +}
 612 +
 613 +/*
 614 + - regatom - the lowest level
 615 + *
 616 + * Optimization:  gobbles an entire sequence of ordinary characters so that
 617 + * it can turn them into a single node, which is smaller to store and
 618 + * faster to run.  Backslashed characters are exceptions, each becoming a
 619 + * separate node; the code is simpler that way and it's not worth fixing.
 620 + */
 621 +static char *
 622 +regatom(struct match_globals *g, int *flagp)
 623 +{
 624 +       register char *ret;
 625 +       int flags;
 626 +
 627 +       *flagp = WORST;         /* Tentatively. */
 628 +
 629 +       switch (*g->regparse++) {
 630 +       case '^':
 631 +               ret = regnode(g, BOL);
 632 +               break;
 633 +       case '$':
 634 +               ret = regnode(g, EOL);
 635 +               break;
 636 +       case '.':
 637 +               ret = regnode(g, ANY);
 638 +               *flagp |= HASWIDTH|SIMPLE;
 639 +               break;
 640 +       case '[': {
 641 +                       register int class;
 642 +                       register int classend;
 643 +
 644 +                       if (*g->regparse == '^') {      /* Complement of range. */
 645 +                               ret = regnode(g, ANYBUT);
 646 +                               g->regparse++;
 647 +                       } else
 648 +                               ret = regnode(g, ANYOF);
 649 +                       if (*g->regparse == ']' || *g->regparse == '-')
 650 +                               regc(g, *g->regparse++);
 651 +                       while (*g->regparse != '\0' && *g->regparse != ']') {
 652 +                               if (*g->regparse == '-') {
 653 +                                       g->regparse++;
 654 +                                       if (*g->regparse == ']' || *g->regparse == '\0')
 655 +                                               regc(g, '-');
 656 +                                       else {
 657 +                                               class = UCHARAT(g->regparse-2)+1;
 658 +                                               classend = UCHARAT(g->regparse);
 659 +                                               if (class > classend+1)
 660 +                                                       FAIL("invalid [] range");
 661 +                                               for (; class <= classend; class++)
 662 +                                                       regc(g, class);
 663 +                                               g->regparse++;
 664 +                                       }
 665 +                               } else
 666 +                                       regc(g, *g->regparse++);
 667 +                       }
 668 +                       regc(g, '\0');
 669 +                       if (*g->regparse != ']')
 670 +                               FAIL("unmatched []");
 671 +                       g->regparse++;
 672 +                       *flagp |= HASWIDTH|SIMPLE;
 673 +               }
 674 +               break;
 675 +       case '(':
 676 +               ret = reg(g, 1, &flags);
 677 +               if (ret == NULL)
 678 +                       return(NULL);
 679 +               *flagp |= flags&(HASWIDTH|SPSTART);
 680 +               break;
 681 +       case '\0':
 682 +       case '|':
 683 +       case ')':
 684 +               FAIL("internal urp");   /* Supposed to be caught earlier. */
 685 +               break;
 686 +       case '?':
 687 +       case '+':
 688 +       case '*':
 689 +               FAIL("?+* follows nothing");
 690 +               break;
 691 +       case '\\':
 692 +               if (*g->regparse == '\0')
 693 +                       FAIL("trailing \\");
 694 +               ret = regnode(g, EXACTLY);
 695 +               regc(g, *g->regparse++);
 696 +               regc(g, '\0');
 697 +               *flagp |= HASWIDTH|SIMPLE;
 698 +               break;
 699 +       default: {
 700 +                       register int len;
 701 +                       register char ender;
 702 +
 703 +                       g->regparse--;
 704 +                       len = my_strcspn((const char *)g->regparse, (const char *)META);
 705 +                       if (len <= 0)
 706 +                               FAIL("internal disaster");
 707 +                       ender = *(g->regparse+len);
 708 +                       if (len > 1 && ISMULT(ender))
 709 +                               len--;          /* Back off clear of ?+* operand. */
 710 +                       *flagp |= HASWIDTH;
 711 +                       if (len == 1)
 712 +                               *flagp |= SIMPLE;
 713 +                       ret = regnode(g, EXACTLY);
 714 +                       while (len > 0) {
 715 +                               regc(g, *g->regparse++);
 716 +                               len--;
 717 +                       }
 718 +                       regc(g, '\0');
 719 +               }
 720 +               break;
 721 +       }
 722 +
 723 +       return(ret);
 724 +}
 725 +
 726 +/*
 727 + - regnode - emit a node
 728 + */
 729 +static char *                  /* Location. */
 730 +regnode(struct match_globals *g, char op)
 731 +{
 732 +       register char *ret;
 733 +       register char *ptr;
 734 +
 735 +       ret = g->regcode;
 736 +       if (ret == &g->regdummy) {
 737 +               g->regsize += 3;
 738 +               return(ret);
 739 +       }
 740 +
 741 +       ptr = ret;
 742 +       *ptr++ = op;
 743 +       *ptr++ = '\0';          /* Null "next" pointer. */
 744 +       *ptr++ = '\0';
 745 +       g->regcode = ptr;
 746 +
 747 +       return(ret);
 748 +}
 749 +
 750 +/*
 751 + - regc - emit (if appropriate) a byte of code
 752 + */
 753 +static void
 754 +regc(struct match_globals *g, char b)
 755 +{
 756 +       if (g->regcode != &g->regdummy)
 757 +               *g->regcode++ = b;
 758 +       else
 759 +               g->regsize++;
 760 +}
 761 +
 762 +/*
 763 + - reginsert - insert an operator in front of already-emitted operand
 764 + *
 765 + * Means relocating the operand.
 766 + */
 767 +static void
 768 +reginsert(struct match_globals *g, char op, char* opnd)
 769 +{
 770 +       register char *src;
 771 +       register char *dst;
 772 +       register char *place;
 773 +
 774 +       if (g->regcode == &g->regdummy) {
 775 +               g->regsize += 3;
 776 +               return;
 777 +       }
 778 +
 779 +       src = g->regcode;
 780 +       g->regcode += 3;
 781 +       dst = g->regcode;
 782 +       while (src > opnd)
 783 +               *--dst = *--src;
 784 +
 785 +       place = opnd;           /* Op node, where operand used to be. */
 786 +       *place++ = op;
 787 +       *place++ = '\0';
 788 +       *place++ = '\0';
 789 +}
 790 +
 791 +/*
 792 + - regtail - set the next-pointer at the end of a node chain
 793 + */
 794 +static void
 795 +regtail(struct match_globals *g, char *p, char *val)
 796 +{
 797 +       register char *scan;
 798 +       register char *temp;
 799 +       register int offset;
 800 +
 801 +       if (p == &g->regdummy)
 802 +               return;
 803 +
 804 +       /* Find last node. */
 805 +       scan = p;
 806 +       for (;;) {
 807 +               temp = regnext(g, scan);
 808 +               if (temp == NULL)
 809 +                       break;
 810 +               scan = temp;
 811 +       }
 812 +
 813 +       if (OP(scan) == BACK)
 814 +               offset = scan - val;
 815 +       else
 816 +               offset = val - scan;
 817 +       *(scan+1) = (offset>>8)&0377;
 818 +       *(scan+2) = offset&0377;
 819 +}
 820 +
 821 +/*
 822 + - regoptail - regtail on operand of first argument; nop if operandless
 823 + */
 824 +static void
 825 +regoptail(struct match_globals *g, char *p, char *val)
 826 +{
 827 +       /* "Operandless" and "op != BRANCH" are synonymous in practice. */
 828 +       if (p == NULL || p == &g->regdummy || OP(p) != BRANCH)
 829 +               return;
 830 +       regtail(g, OPERAND(p), val);
 831 +}
 832 +
 833 +/*
 834 + * regexec and friends
 835 + */
 836 +
 837 +
 838 +/*
 839 + * Forwards.
 840 + */
 841 +STATIC int regtry(struct match_globals *g, regexp *prog, char *string);
 842 +STATIC int regmatch(struct match_globals *g, char *prog);
 843 +STATIC int regrepeat(struct match_globals *g, char *p);
 844 +
 845 +#ifdef DEBUG
 846 +int regnarrate = 0;
 847 +void regdump();
 848 +STATIC char *regprop(char *op);
 849 +#endif
 850 +
 851 +/*
 852 + - regexec - match a regexp against a string
 853 + */
 854 +int
 855 +regexec(regexp *prog, char *string)
 856 +{
 857 +       register char *s;
 858 +       struct match_globals g;
 859 +
 860 +       /* Be paranoid... */
 861 +       if (prog == NULL || string == NULL) {
 862 +               printk("<3>Regexp: NULL parameter\n");
 863 +               return(0);
 864 +       }
 865 +
 866 +       /* Check validity of program. */
 867 +       if (UCHARAT(prog->program) != MAGIC) {
 868 +               printk("<3>Regexp: corrupted program\n");
 869 +               return(0);
 870 +       }
 871 +
 872 +       /* If there is a "must appear" string, look for it. */
 873 +       if (prog->regmust != NULL) {
 874 +               s = string;
 875 +               while ((s = strchr(s, prog->regmust[0])) != NULL) {
 876 +                       if (strncmp(s, prog->regmust, prog->regmlen) == 0)
 877 +                               break;  /* Found it. */
 878 +                       s++;
 879 +               }
 880 +               if (s == NULL)  /* Not present. */
 881 +                       return(0);
 882 +       }
 883 +
 884 +       /* Mark beginning of line for ^ . */
 885 +       g.regbol = string;
 886 +
 887 +       /* Simplest case:  anchored match need be tried only once. */
 888 +       if (prog->reganch)
 889 +               return(regtry(&g, prog, string));
 890 +
 891 +       /* Messy cases:  unanchored match. */
 892 +       s = string;
 893 +       if (prog->regstart != '\0')
 894 +               /* We know what char it must start with. */
 895 +               while ((s = strchr(s, prog->regstart)) != NULL) {
 896 +                       if (regtry(&g, prog, s))
 897 +                               return(1);
 898 +                       s++;
 899 +               }
 900 +       else
 901 +               /* We don't -- general case. */
 902 +               do {
 903 +                       if (regtry(&g, prog, s))
 904 +                               return(1);
 905 +               } while (*s++ != '\0');
 906 +
 907 +       /* Failure. */
 908 +       return(0);
 909 +}
 910 +
 911 +/*
 912 + - regtry - try match at specific point
 913 + */
 914 +static int                     /* 0 failure, 1 success */
 915 +regtry(struct match_globals *g, regexp *prog, char *string)
 916 +{
 917 +       register int i;
 918 +       register char **sp;
 919 +       register char **ep;
 920 +
 921 +       g->reginput = string;
 922 +       g->regstartp = prog->startp;
 923 +       g->regendp = prog->endp;
 924 +
 925 +       sp = prog->startp;
 926 +       ep = prog->endp;
 927 +       for (i = NSUBEXP; i > 0; i--) {
 928 +               *sp++ = NULL;
 929 +               *ep++ = NULL;
 930 +       }
 931 +       if (regmatch(g, prog->program + 1)) {
 932 +               prog->startp[0] = string;
 933 +               prog->endp[0] = g->reginput;
 934 +               return(1);
 935 +       } else
 936 +               return(0);
 937 +}
 938 +
 939 +/*
 940 + - regmatch - main matching routine
 941 + *
 942 + * Conceptually the strategy is simple:  check to see whether the current
 943 + * node matches, call self recursively to see whether the rest matches,
 944 + * and then act accordingly.  In practice we make some effort to avoid
 945 + * recursion, in particular by going through "ordinary" nodes (that don't
 946 + * need to know whether the rest of the match failed) by a loop instead of
 947 + * by recursion.
 948 + */
 949 +static int                     /* 0 failure, 1 success */
 950 +regmatch(struct match_globals *g, char *prog)
 951 +{
 952 +       register char *scan = prog; /* Current node. */
 953 +       char *next;                 /* Next node. */
 954 +
 955 +#ifdef DEBUG
 956 +       if (scan != NULL && regnarrate)
 957 +               fprintf(stderr, "%s(\n", regprop(scan));
 958 +#endif
 959 +       while (scan != NULL) {
 960 +#ifdef DEBUG
 961 +               if (regnarrate)
 962 +                       fprintf(stderr, "%s...\n", regprop(scan));
 963 +#endif
 964 +               next = regnext(g, scan);
 965 +
 966 +               switch (OP(scan)) {
 967 +               case BOL:
 968 +                       if (g->reginput != g->regbol)
 969 +                               return(0);
 970 +                       break;
 971 +               case EOL:
 972 +                       if (*g->reginput != '\0')
 973 +                               return(0);
 974 +                       break;
 975 +               case ANY:
 976 +                       if (*g->reginput == '\0')
 977 +                               return(0);
 978 +                       g->reginput++;
 979 +                       break;
 980 +               case EXACTLY: {
 981 +                               register int len;
 982 +                               register char *opnd;
 983 +
 984 +                               opnd = OPERAND(scan);
 985 +                               /* Inline the first character, for speed. */
 986 +                               if (*opnd != *g->reginput)
 987 +                                       return(0);
 988 +                               len = strlen(opnd);
 989 +                               if (len > 1 && strncmp(opnd, g->reginput, len) != 0)
 990 +                                       return(0);
 991 +                               g->reginput += len;
 992 +                       }
 993 +                       break;
 994 +               case ANYOF:
 995 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) == NULL)
 996 +                               return(0);
 997 +                       g->reginput++;
 998 +                       break;
 999 +               case ANYBUT:
1000 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) != NULL)
1001 +                               return(0);
1002 +                       g->reginput++;
1003 +                       break;
1004 +               case NOTHING:
1005 +               case BACK:
1006 +                       break;
1007 +               case OPEN+1:
1008 +               case OPEN+2:
1009 +               case OPEN+3:
1010 +               case OPEN+4:
1011 +               case OPEN+5:
1012 +               case OPEN+6:
1013 +               case OPEN+7:
1014 +               case OPEN+8:
1015 +               case OPEN+9: {
1016 +                               register int no;
1017 +                               register char *save;
1018 +
1019 +                               no = OP(scan) - OPEN;
1020 +                               save = g->reginput;
1021 +
1022 +                               if (regmatch(g, next)) {
1023 +                                       /*
1024 +                                        * Don't set startp if some later
1025 +                                        * invocation of the same parentheses
1026 +                                        * already has.
1027 +                                        */
1028 +                                       if (g->regstartp[no] == NULL)
1029 +                                               g->regstartp[no] = save;
1030 +                                       return(1);
1031 +                               } else
1032 +                                       return(0);
1033 +                       }
1034 +                       break;
1035 +               case CLOSE+1:
1036 +               case CLOSE+2:
1037 +               case CLOSE+3:
1038 +               case CLOSE+4:
1039 +               case CLOSE+5:
1040 +               case CLOSE+6:
1041 +               case CLOSE+7:
1042 +               case CLOSE+8:
1043 +               case CLOSE+9:
1044 +                       {
1045 +                               register int no;
1046 +                               register char *save;
1047 +
1048 +                               no = OP(scan) - CLOSE;
1049 +                               save = g->reginput;
1050 +
1051 +                               if (regmatch(g, next)) {
1052 +                                       /*
1053 +                                        * Don't set endp if some later
1054 +                                        * invocation of the same parentheses
1055 +                                        * already has.
1056 +                                        */
1057 +                                       if (g->regendp[no] == NULL)
1058 +                                               g->regendp[no] = save;
1059 +                                       return(1);
1060 +                               } else
1061 +                                       return(0);
1062 +                       }
1063 +                       break;
1064 +               case BRANCH: {
1065 +                               register char *save;
1066 +
1067 +                               if (OP(next) != BRANCH)         /* No choice. */
1068 +                                       next = OPERAND(scan);   /* Avoid recursion. */
1069 +                               else {
1070 +                                       do {
1071 +                                               save = g->reginput;
1072 +                                               if (regmatch(g, OPERAND(scan)))
1073 +                                                       return(1);
1074 +                                               g->reginput = save;
1075 +                                               scan = regnext(g, scan);
1076 +                                       } while (scan != NULL && OP(scan) == BRANCH);
1077 +                                       return(0);
1078 +                                       /* NOTREACHED */
1079 +                               }
1080 +                       }
1081 +                       break;
1082 +               case STAR:
1083 +               case PLUS: {
1084 +                               register char nextch;
1085 +                               register int no;
1086 +                               register char *save;
1087 +                               register int min;
1088 +
1089 +                               /*
1090 +                                * Lookahead to avoid useless match attempts
1091 +                                * when we know what character comes next.
1092 +                                */
1093 +                               nextch = '\0';
1094 +                               if (OP(next) == EXACTLY)
1095 +                                       nextch = *OPERAND(next);
1096 +                               min = (OP(scan) == STAR) ? 0 : 1;
1097 +                               save = g->reginput;
1098 +                               no = regrepeat(g, OPERAND(scan));
1099 +                               while (no >= min) {
1100 +                                       /* If it could work, try it. */
1101 +                                       if (nextch == '\0' || *g->reginput == nextch)
1102 +                                               if (regmatch(g, next))
1103 +                                                       return(1);
1104 +                                       /* Couldn't or didn't -- back up. */
1105 +                                       no--;
1106 +                                       g->reginput = save + no;
1107 +                               }
1108 +                               return(0);
1109 +                       }
1110 +                       break;
1111 +               case END:
1112 +                       return(1);      /* Success! */
1113 +                       break;
1114 +               default:
1115 +                       printk("<3>Regexp: memory corruption\n");
1116 +                       return(0);
1117 +                       break;
1118 +               }
1119 +
1120 +               scan = next;
1121 +       }
1122 +
1123 +       /*
1124 +        * We get here only if there's trouble -- normally "case END" is
1125 +        * the terminating point.
1126 +        */
1127 +       printk("<3>Regexp: corrupted pointers\n");
1128 +       return(0);
1129 +}
1130 +
1131 +/*
1132 + - regrepeat - repeatedly match something simple, report how many
1133 + */
1134 +static int
1135 +regrepeat(struct match_globals *g, char *p)
1136 +{
1137 +       register int count = 0;
1138 +       register char *scan;
1139 +       register char *opnd;
1140 +
1141 +       scan = g->reginput;
1142 +       opnd = OPERAND(p);
1143 +       switch (OP(p)) {
1144 +       case ANY:
1145 +               count = strlen(scan);
1146 +               scan += count;
1147 +               break;
1148 +       case EXACTLY:
1149 +               while (*opnd == *scan) {
1150 +                       count++;
1151 +                       scan++;
1152 +               }
1153 +               break;
1154 +       case ANYOF:
1155 +               while (*scan != '\0' && strchr(opnd, *scan) != NULL) {
1156 +                       count++;
1157 +                       scan++;
1158 +               }
1159 +               break;
1160 +       case ANYBUT:
1161 +               while (*scan != '\0' && strchr(opnd, *scan) == NULL) {
1162 +                       count++;
1163 +                       scan++;
1164 +               }
1165 +               break;
1166 +       default:                /* Oh dear.  Called inappropriately. */
1167 +               printk("<3>Regexp: internal foulup\n");
1168 +               count = 0;      /* Best compromise. */
1169 +               break;
1170 +       }
1171 +       g->reginput = scan;
1172 +
1173 +       return(count);
1174 +}
1175 +
1176 +/*
1177 + - regnext - dig the "next" pointer out of a node
1178 + */
1179 +static char*
1180 +regnext(struct match_globals *g, char *p)
1181 +{
1182 +       register int offset;
1183 +
1184 +       if (p == &g->regdummy)
1185 +               return(NULL);
1186 +
1187 +       offset = NEXT(p);
1188 +       if (offset == 0)
1189 +               return(NULL);
1190 +
1191 +       if (OP(p) == BACK)
1192 +               return(p-offset);
1193 +       else
1194 +               return(p+offset);
1195 +}
1196 +
1197 +#ifdef DEBUG
1198 +
1199 +STATIC char *regprop();
1200 +
1201 +/*
1202 + - regdump - dump a regexp onto stdout in vaguely comprehensible form
1203 + */
1204 +void
1205 +regdump(regexp *r)
1206 +{
1207 +       register char *s;
1208 +       register char op = EXACTLY;     /* Arbitrary non-END op. */
1209 +       register char *next;
1210 +       /* extern char *strchr(); */
1211 +
1212 +
1213 +       s = r->program + 1;
1214 +       while (op != END) {     /* While that wasn't END last time... */
1215 +               op = OP(s);
1216 +               printf("%2d%s", s-r->program, regprop(s));      /* Where, what. */
1217 +               next = regnext(s);
1218 +               if (next == NULL)               /* Next ptr. */
1219 +                       printf("(0)");
1220 +               else
1221 +                       printf("(%d)", (s-r->program)+(next-s));
1222 +               s += 3;
1223 +               if (op == ANYOF || op == ANYBUT || op == EXACTLY) {
1224 +                       /* Literal string, where present. */
1225 +                       while (*s != '\0') {
1226 +                               putchar(*s);
1227 +                               s++;
1228 +                       }
1229 +                       s++;
1230 +               }
1231 +               putchar('\n');
1232 +       }
1233 +
1234 +       /* Header fields of interest. */
1235 +       if (r->regstart != '\0')
1236 +               printf("start `%c' ", r->regstart);
1237 +       if (r->reganch)
1238 +               printf("anchored ");
1239 +       if (r->regmust != NULL)
1240 +               printf("must have \"%s\"", r->regmust);
1241 +       printf("\n");
1242 +}
1243 +
1244 +/*
1245 + - regprop - printable representation of opcode
1246 + */
1247 +static char *
1248 +regprop(char *op)
1249 +{
1250 +#define BUFLEN 50
1251 +       register char *p;
1252 +       static char buf[BUFLEN];
1253 +
1254 +       strcpy(buf, ":");
1255 +
1256 +       switch (OP(op)) {
1257 +       case BOL:
1258 +               p = "BOL";
1259 +               break;
1260 +       case EOL:
1261 +               p = "EOL";
1262 +               break;
1263 +       case ANY:
1264 +               p = "ANY";
1265 +               break;
1266 +       case ANYOF:
1267 +               p = "ANYOF";
1268 +               break;
1269 +       case ANYBUT:
1270 +               p = "ANYBUT";
1271 +               break;
1272 +       case BRANCH:
1273 +               p = "BRANCH";
1274 +               break;
1275 +       case EXACTLY:
1276 +               p = "EXACTLY";
1277 +               break;
1278 +       case NOTHING:
1279 +               p = "NOTHING";
1280 +               break;
1281 +       case BACK:
1282 +               p = "BACK";
1283 +               break;
1284 +       case END:
1285 +               p = "END";
1286 +               break;
1287 +       case OPEN+1:
1288 +       case OPEN+2:
1289 +       case OPEN+3:
1290 +       case OPEN+4:
1291 +       case OPEN+5:
1292 +       case OPEN+6:
1293 +       case OPEN+7:
1294 +       case OPEN+8:
1295 +       case OPEN+9:
1296 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "OPEN%d", OP(op)-OPEN);
1297 +               p = NULL;
1298 +               break;
1299 +       case CLOSE+1:
1300 +       case CLOSE+2:
1301 +       case CLOSE+3:
1302 +       case CLOSE+4:
1303 +       case CLOSE+5:
1304 +       case CLOSE+6:
1305 +       case CLOSE+7:
1306 +       case CLOSE+8:
1307 +       case CLOSE+9:
1308 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "CLOSE%d", OP(op)-CLOSE);
1309 +               p = NULL;
1310 +               break;
1311 +       case STAR:
1312 +               p = "STAR";
1313 +               break;
1314 +       case PLUS:
1315 +               p = "PLUS";
1316 +               break;
1317 +       default:
1318 +               printk("<3>Regexp: corrupted opcode\n");
1319 +               break;
1320 +       }
1321 +       if (p != NULL)
1322 +               strncat(buf, p, BUFLEN-strlen(buf));
1323 +       return(buf);
1324 +}
1325 +#endif
1326 +
1327 +
1328 --- /dev/null
1329 +++ b/net/netfilter/regexp/regexp.h
1330 @@ -0,0 +1,41 @@
1331 +/*
1332 + * Definitions etc. for regexp(3) routines.
1333 + *
1334 + * Caveat:  this is V8 regexp(3) [actually, a reimplementation thereof],
1335 + * not the System V one.
1336 + */
1337 +
1338 +#ifndef REGEXP_H
1339 +#define REGEXP_H
1340 +
1341 +
1342 +/*
1343 +http://www.opensource.apple.com/darwinsource/10.3/expect-1/expect/expect.h ,
1344 +which contains a version of this library, says:
1345 +
1346 + *
1347 + * NSUBEXP must be at least 10, and no greater than 117 or the parser
1348 + * will not work properly.
1349 + *
1350 +
1351 +However, it looks rather like this library is limited to 10.  If you think
1352 +otherwise, let us know.
1353 +*/
1354 +
1355 +#define NSUBEXP  10
1356 +typedef struct regexp {
1357 +       char *startp[NSUBEXP];
1358 +       char *endp[NSUBEXP];
1359 +       char regstart;          /* Internal use only. */
1360 +       char reganch;           /* Internal use only. */
1361 +       char *regmust;          /* Internal use only. */
1362 +       int regmlen;            /* Internal use only. */
1363 +       char program[1];        /* Unwarranted chumminess with compiler. */
1364 +} regexp;
1365 +
1366 +regexp * regcomp(char *exp, int *patternsize);
1367 +int regexec(regexp *prog, char *string);
1368 +void regsub(regexp *prog, char *source, char *dest);
1369 +void regerror(char *s);
1370 +
1371 +#endif
1372 --- /dev/null
1373 +++ b/net/netfilter/regexp/regmagic.h
1374 @@ -0,0 +1,5 @@
1375 +/*
1376 + * The first byte of the regexp internal "program" is actually this magic
1377 + * number; the start node begins in the second byte.
1378 + */
1379 +#define        MAGIC   0234
1380 --- /dev/null
1381 +++ b/net/netfilter/regexp/regsub.c
1382 @@ -0,0 +1,95 @@
1383 +/*
1384 + * regsub
1385 + * @(#)regsub.c        1.3 of 2 April 86
1386 + *
1387 + *     Copyright (c) 1986 by University of Toronto.
1388 + *     Written by Henry Spencer.  Not derived from licensed software.
1389 + *
1390 + *     Permission is granted to anyone to use this software for any
1391 + *     purpose on any computer system, and to redistribute it freely,
1392 + *     subject to the following restrictions:
1393 + *
1394 + *     1. The author is not responsible for the consequences of use of
1395 + *             this software, no matter how awful, even if they arise
1396 + *             from defects in it.
1397 + *
1398 + *     2. The origin of this software must not be misrepresented, either
1399 + *             by explicit claim or by omission.
1400 + *
1401 + *     3. Altered versions must be plainly marked as such, and must not
1402 + *             be misrepresented as being the original software.
1403 + *
1404 + *
1405 + * This code was modified by Ethan Sommer to work within the kernel
1406 + * (it now uses kmalloc etc..)
1407 + *
1408 + */
1409 +#include "regexp.h"
1410 +#include "regmagic.h"
1411 +#include <linux/string.h>
1412 +
1413 +
1414 +#ifndef CHARBITS
1415 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
1416 +#else
1417 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
1418 +#endif
1419 +
1420 +#if 0
1421 +//void regerror(char * s)
1422 +//{
1423 +//        printk("regexp(3): %s", s);
1424 +//        /* NOTREACHED */
1425 +//}
1426 +#endif
1427 +
1428 +/*
1429 + - regsub - perform substitutions after a regexp match
1430 + */
1431 +void
1432 +regsub(regexp * prog, char * source, char * dest)
1433 +{
1434 +       register char *src;
1435 +       register char *dst;
1436 +       register char c;
1437 +       register int no;
1438 +       register int len;
1439 +
1440 +       /* Not necessary and gcc doesn't like it -MLS */
1441 +       /*extern char *strncpy();*/
1442 +
1443 +       if (prog == NULL || source == NULL || dest == NULL) {
1444 +               regerror("NULL parm to regsub");
1445 +               return;
1446 +       }
1447 +       if (UCHARAT(prog->program) != MAGIC) {
1448 +               regerror("damaged regexp fed to regsub");
1449 +               return;
1450 +       }
1451 +
1452 +       src = source;
1453 +       dst = dest;
1454 +       while ((c = *src++) != '\0') {
1455 +               if (c == '&')
1456 +                       no = 0;
1457 +               else if (c == '\\' && '0' <= *src && *src <= '9')
1458 +                       no = *src++ - '0';
1459 +               else
1460 +                       no = -1;
1461 +
1462 +               if (no < 0) {   /* Ordinary character. */
1463 +                       if (c == '\\' && (*src == '\\' || *src == '&'))
1464 +                               c = *src++;
1465 +                       *dst++ = c;
1466 +               } else if (prog->startp[no] != NULL && prog->endp[no] != NULL) {
1467 +                       len = prog->endp[no] - prog->startp[no];
1468 +                       (void) strncpy(dst, prog->startp[no], len);
1469 +                       dst += len;
1470 +                       if (len != 0 && *(dst-1) == '\0') {     /* strncpy hit NUL. */
1471 +                               regerror("damaged match string");
1472 +                               return;
1473 +                       }
1474 +               }
1475 +       }
1476 +       *dst++ = '\0';
1477 +}
1478 --- /dev/null
1479 +++ b/net/netfilter/xt_layer7.c
1480 @@ -0,0 +1,626 @@
1481 +/*
1482 +  Kernel module to match application layer (OSI layer 7) data in connections.
1483 +
1484 +  http://l7-filter.sf.net
1485 +
1486 +  (C) 2003, 2004, 2005, 2006, 2007 Matthew Strait and Ethan Sommer.
1487 +
1488 +  This program is free software; you can redistribute it and/or
1489 +  modify it under the terms of the GNU General Public License
1490 +  as published by the Free Software Foundation; either version
1491 +  2 of the License, or (at your option) any later version.
1492 +  http://www.gnu.org/licenses/gpl.txt
1493 +
1494 +  Based on ipt_string.c (C) 2000 Emmanuel Roger <winfield@freegates.be>,
1495 +  xt_helper.c (C) 2002 Harald Welte and cls_layer7.c (C) 2003 Matthew Strait,
1496 +  Ethan Sommer, Justin Levandoski.
1497 +*/
1498 +
1499 +#include <linux/spinlock.h>
1500 +#include <net/ip.h>
1501 +#include <net/tcp.h>
1502 +#include <linux/module.h>
1503 +#include <linux/skbuff.h>
1504 +#include <linux/netfilter.h>
1505 +#include <net/netfilter/nf_conntrack.h>
1506 +#include <net/netfilter/nf_conntrack_core.h>
1507 +#include <linux/netfilter/x_tables.h>
1508 +#include <linux/netfilter/xt_layer7.h>
1509 +#include <net/netfilter/nf_conntrack_compat.h>
1510 +#include <linux/ctype.h>
1511 +#include <linux/proc_fs.h>
1512 +
1513 +#include "regexp/regexp.c"
1514 +
1515 +MODULE_LICENSE("GPL");
1516 +MODULE_AUTHOR("Matthew Strait <quadong@users.sf.net>, Ethan Sommer <sommere@users.sf.net>");
1517 +MODULE_DESCRIPTION("iptables application layer match module");
1518 +MODULE_ALIAS("ipt_layer7");
1519 +MODULE_VERSION("2.16.1");
1520 +
1521 +static int maxdatalen = 2048; // this is the default
1522 +module_param(maxdatalen, int, 0444);
1523 +MODULE_PARM_DESC(maxdatalen, "maximum bytes of data looked at by l7-filter");
1524 +#ifdef CONFIG_NETFILTER_XT_MATCH_LAYER7_DEBUG
1525 +       #define DPRINTK(format,args...) printk(format,##args)
1526 +#else
1527 +       #define DPRINTK(format,args...)
1528 +#endif
1529 +
1530 +#define TOTAL_PACKETS master_conntrack->counters[IP_CT_DIR_ORIGINAL].packets + \
1531 +                     master_conntrack->counters[IP_CT_DIR_REPLY].packets
1532 +
1533 +/* Number of packets whose data we look at.
1534 +This can be modified through /proc/net/layer7_numpackets */
1535 +static int num_packets = 10;
1536 +
1537 +static struct pattern_cache {
1538 +       char * regex_string;
1539 +       regexp * pattern;
1540 +       struct pattern_cache * next;
1541 +} * first_pattern_cache = NULL;
1542 +
1543 +DEFINE_SPINLOCK(l7_lock);
1544 +
1545 +#ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
1546 +/* Converts an unfriendly string into a friendly one by
1547 +replacing unprintables with periods and all whitespace with " ". */
1548 +static char * friendly_print(unsigned char * s)
1549 +{
1550 +       char * f = kmalloc(strlen(s) + 1, GFP_ATOMIC);
1551 +       int i;
1552 +
1553 +       if(!f) {
1554 +               if (net_ratelimit())
1555 +                       printk(KERN_ERR "layer7: out of memory in "
1556 +                                       "friendly_print, bailing.\n");
1557 +               return NULL;
1558 +       }
1559 +
1560 +       for(i = 0; i < strlen(s); i++){
1561 +               if(isprint(s[i]) && s[i] < 128) f[i] = s[i];
1562 +               else if(isspace(s[i]))          f[i] = ' ';
1563 +               else                            f[i] = '.';
1564 +       }
1565 +       f[i] = '\0';
1566 +       return f;
1567 +}
1568 +
1569 +static char dec2hex(int i)
1570 +{
1571 +       switch (i) {
1572 +               case 0 ... 9:
1573 +                       return (char)(i + '0');
1574 +                       break;
1575 +               case 10 ... 15:
1576 +                       return (char)(i - 10 + 'a');
1577 +                       break;
1578 +               default:
1579 +                       if (net_ratelimit())
1580 +                               printk("Problem in dec2hex\n");
1581 +                       return '\0';
1582 +       }
1583 +}
1584 +
1585 +static char * hex_print(unsigned char * s)
1586 +{
1587 +       char * g = kmalloc(strlen(s)*3 + 1, GFP_ATOMIC);
1588 +       int i;
1589 +
1590 +       if(!g) {
1591 +              if (net_ratelimit())
1592 +                       printk(KERN_ERR "layer7: out of memory in hex_print, "
1593 +                                       "bailing.\n");
1594 +              return NULL;
1595 +       }
1596 +
1597 +       for(i = 0; i < strlen(s); i++) {
1598 +               g[i*3    ] = dec2hex(s[i]/16);
1599 +               g[i*3 + 1] = dec2hex(s[i]%16);
1600 +               g[i*3 + 2] = ' ';
1601 +       }
1602 +       g[i*3] = '\0';
1603 +
1604 +       return g;
1605 +}
1606 +#endif // DEBUG
1607 +
1608 +/* Use instead of regcomp.  As we expect to be seeing the same regexps over and
1609 +over again, it make sense to cache the results. */
1610 +static regexp * compile_and_cache(char * regex_string, char * protocol)
1611 +{
1612 +       struct pattern_cache * node               = first_pattern_cache;
1613 +       struct pattern_cache * last_pattern_cache = first_pattern_cache;
1614 +       struct pattern_cache * tmp;
1615 +       unsigned int len;
1616 +
1617 +       while (node != NULL) {
1618 +               if (!strcmp(node->regex_string, regex_string))
1619 +               return node->pattern;
1620 +
1621 +               last_pattern_cache = node;/* points at the last non-NULL node */
1622 +               node = node->next;
1623 +       }
1624 +
1625 +       /* If we reach the end of the list, then we have not yet cached
1626 +          the pattern for this regex. Let's do that now.
1627 +          Be paranoid about running out of memory to avoid list corruption. */
1628 +       tmp = kmalloc(sizeof(struct pattern_cache), GFP_ATOMIC);
1629 +
1630 +       if(!tmp) {
1631 +               if (net_ratelimit())
1632 +                       printk(KERN_ERR "layer7: out of memory in "
1633 +                                       "compile_and_cache, bailing.\n");
1634 +               return NULL;
1635 +       }
1636 +
1637 +       tmp->regex_string  = kmalloc(strlen(regex_string) + 1, GFP_ATOMIC);
1638 +       tmp->pattern       = kmalloc(sizeof(struct regexp),    GFP_ATOMIC);
1639 +       tmp->next = NULL;
1640 +
1641 +       if(!tmp->regex_string || !tmp->pattern) {
1642 +               if (net_ratelimit())
1643 +                       printk(KERN_ERR "layer7: out of memory in "
1644 +                                       "compile_and_cache, bailing.\n");
1645 +               kfree(tmp->regex_string);
1646 +               kfree(tmp->pattern);
1647 +               kfree(tmp);
1648 +               return NULL;
1649 +       }
1650 +
1651 +       /* Ok.  The new node is all ready now. */
1652 +       node = tmp;
1653 +
1654 +       if(first_pattern_cache == NULL) /* list is empty */
1655 +               first_pattern_cache = node; /* make node the beginning */
1656 +       else
1657 +               last_pattern_cache->next = node; /* attach node to the end */
1658 +
1659 +       /* copy the string and compile the regex */
1660 +       len = strlen(regex_string);
1661 +       DPRINTK("About to compile this: \"%s\"\n", regex_string);
1662 +       node->pattern = regcomp(regex_string, &len);
1663 +       if ( !node->pattern ) {
1664 +               if (net_ratelimit())
1665 +                       printk(KERN_ERR "layer7: Error compiling regexp "
1666 +                                       "\"%s\" (%s)\n",
1667 +                                       regex_string, protocol);
1668 +               /* pattern is now cached as NULL, so we won't try again. */
1669 +       }
1670 +
1671 +       strcpy(node->regex_string, regex_string);
1672 +       return node->pattern;
1673 +}
1674 +
1675 +static int can_handle(const struct sk_buff *skb)
1676 +{
1677 +       if(!skb->nh.iph) /* not IP */
1678 +               return 0;
1679 +       if(skb->nh.iph->protocol != IPPROTO_TCP &&
1680 +          skb->nh.iph->protocol != IPPROTO_UDP &&
1681 +          skb->nh.iph->protocol != IPPROTO_ICMP)
1682 +               return 0;
1683 +       return 1;
1684 +}
1685 +
1686 +/* Returns offset the into the skb->data that the application data starts */
1687 +static int app_data_offset(const struct sk_buff *skb)
1688 +{
1689 +       /* In case we are ported somewhere (ebtables?) where skb->nh.iph
1690 +       isn't set, this can be gotten from 4*(skb->data[0] & 0x0f) as well. */
1691 +       int ip_hl = 4*skb->nh.iph->ihl;
1692 +
1693 +       if( skb->nh.iph->protocol == IPPROTO_TCP ) {
1694 +               /* 12 == offset into TCP header for the header length field.
1695 +               Can't get this with skb->h.th->doff because the tcphdr
1696 +               struct doesn't get set when routing (this is confirmed to be
1697 +               true in Netfilter as well as QoS.) */
1698 +               int tcp_hl = 4*(skb->data[ip_hl + 12] >> 4);
1699 +
1700 +               return ip_hl + tcp_hl;
1701 +       } else if( skb->nh.iph->protocol == IPPROTO_UDP  ) {
1702 +               return ip_hl + 8; /* UDP header is always 8 bytes */
1703 +       } else if( skb->nh.iph->protocol == IPPROTO_ICMP ) {
1704 +               return ip_hl + 8; /* ICMP header is 8 bytes */
1705 +       } else {
1706 +               if (net_ratelimit())
1707 +                       printk(KERN_ERR "layer7: tried to handle unknown "
1708 +                                       "protocol!\n");
1709 +               return ip_hl + 8; /* something reasonable */
1710 +       }
1711 +}
1712 +
1713 +/* handles whether there's a match when we aren't appending data anymore */
1714 +static int match_no_append(struct nf_conn * conntrack,
1715 +                           struct nf_conn * master_conntrack,
1716 +                           enum ip_conntrack_info ctinfo,
1717 +                           enum ip_conntrack_info master_ctinfo,
1718 +                           struct xt_layer7_info * info)
1719 +{
1720 +       /* If we're in here, throw the app data away */
1721 +       if(master_conntrack->layer7.app_data != NULL) {
1722 +
1723 +       #ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
1724 +               if(!master_conntrack->layer7.app_proto) {
1725 +                       char * f =
1726 +                         friendly_print(master_conntrack->layer7.app_data);
1727 +                       char * g =
1728 +                         hex_print(master_conntrack->layer7.app_data);
1729 +                       DPRINTK("\nl7-filter gave up after %d bytes "
1730 +                               "(%d packets):\n%s\n",
1731 +                               strlen(f), TOTAL_PACKETS, f);
1732 +                       kfree(f);
1733 +                       DPRINTK("In hex: %s\n", g);
1734 +                       kfree(g);
1735 +               }
1736 +       #endif
1737 +
1738 +               kfree(master_conntrack->layer7.app_data);
1739 +               master_conntrack->layer7.app_data = NULL; /* don't free again */
1740 +       }
1741 +
1742 +       if(master_conntrack->layer7.app_proto){
1743 +               /* Here child connections set their .app_proto (for /proc) */
1744 +               if(!conntrack->layer7.app_proto) {
1745 +                       conntrack->layer7.app_proto =
1746 +                         kmalloc(strlen(master_conntrack->layer7.app_proto)+1,
1747 +                           GFP_ATOMIC);
1748 +                       if(!conntrack->layer7.app_proto){
1749 +                               if (net_ratelimit())
1750 +                                       printk(KERN_ERR "layer7: out of memory "
1751 +                                                       "in match_no_append, "
1752 +                                                       "bailing.\n");
1753 +                               return 1;
1754 +                       }
1755 +                       strcpy(conntrack->layer7.app_proto,
1756 +                               master_conntrack->layer7.app_proto);
1757 +               }
1758 +
1759 +               return (!strcmp(master_conntrack->layer7.app_proto,
1760 +                               info->protocol));
1761 +       }
1762 +       else {
1763 +               /* If not classified, set to "unknown" to distinguish from
1764 +               connections that are still being tested. */
1765 +               master_conntrack->layer7.app_proto =
1766 +                       kmalloc(strlen("unknown")+1, GFP_ATOMIC);
1767 +               if(!master_conntrack->layer7.app_proto){
1768 +                       if (net_ratelimit())
1769 +                               printk(KERN_ERR "layer7: out of memory in "
1770 +                                               "match_no_append, bailing.\n");
1771 +                       return 1;
1772 +               }
1773 +               strcpy(master_conntrack->layer7.app_proto, "unknown");
1774 +               return 0;
1775 +       }
1776 +}
1777 +
1778 +/* add the new app data to the conntrack.  Return number of bytes added. */
1779 +static int add_data(struct nf_conn * master_conntrack,
1780 +                    char * app_data, int appdatalen)
1781 +{
1782 +       int length = 0, i;
1783 +       int oldlength = master_conntrack->layer7.app_data_len;
1784 +
1785 +       /* This is a fix for a race condition by Deti Fliegl. However, I'm not
1786 +          clear on whether the race condition exists or whether this really
1787 +          fixes it.  I might just be being dense... Anyway, if it's not really
1788 +          a fix, all it does is waste a very small amount of time. */
1789 +       if(!master_conntrack->layer7.app_data) return 0;
1790 +
1791 +       /* Strip nulls. Make everything lower case (our regex lib doesn't
1792 +       do case insensitivity).  Add it to the end of the current data. */
1793 +       for(i = 0; i < maxdatalen-oldlength-1 &&
1794 +                  i < appdatalen; i++) {
1795 +               if(app_data[i] != '\0') {
1796 +                       /* the kernel version of tolower mungs 'upper ascii' */
1797 +                       master_conntrack->layer7.app_data[length+oldlength] =
1798 +                               isascii(app_data[i])?
1799 +                                       tolower(app_data[i]) : app_data[i];
1800 +                       length++;
1801 +               }
1802 +       }
1803 +
1804 +       master_conntrack->layer7.app_data[length+oldlength] = '\0';
1805 +       master_conntrack->layer7.app_data_len = length + oldlength;
1806 +
1807 +       return length;
1808 +}
1809 +
1810 +/* taken from drivers/video/modedb.c */
1811 +static int my_atoi(const char *s)
1812 +{
1813 +       int val = 0;
1814 +
1815 +       for (;; s++) {
1816 +               switch (*s) {
1817 +                       case '0'...'9':
1818 +                       val = 10*val+(*s-'0');
1819 +                       break;
1820 +               default:
1821 +                       return val;
1822 +               }
1823 +       }
1824 +}
1825 +
1826 +/* write out num_packets to userland. */
1827 +static int layer7_read_proc(char* page, char ** start, off_t off, int count,
1828 +                            int* eof, void * data)
1829 +{
1830 +       if(num_packets > 99 && net_ratelimit())
1831 +               printk(KERN_ERR "layer7: NOT REACHED. num_packets too big\n");
1832 +
1833 +       page[0] = num_packets/10 + '0';
1834 +       page[1] = num_packets%10 + '0';
1835 +       page[2] = '\n';
1836 +       page[3] = '\0';
1837 +
1838 +       *eof=1;
1839 +
1840 +       return 3;
1841 +}
1842 +
1843 +/* Read in num_packets from userland */
1844 +static int layer7_write_proc(struct file* file, const char* buffer,
1845 +                             unsigned long count, void *data)
1846 +{
1847 +       char * foo = kmalloc(count, GFP_ATOMIC);
1848 +
1849 +       if(!foo){
1850 +               if (net_ratelimit())
1851 +                       printk(KERN_ERR "layer7: out of memory, bailing. "
1852 +                                       "num_packets unchanged.\n");
1853 +               return count;
1854 +       }
1855 +
1856 +       if(copy_from_user(foo, buffer, count)) {
1857 +               return -EFAULT;
1858 +       }
1859 +
1860 +
1861 +       num_packets = my_atoi(foo);
1862 +       kfree (foo);
1863 +
1864 +       /* This has an arbitrary limit to make the math easier. I'm lazy.
1865 +       But anyway, 99 is a LOT! If you want more, you're doing it wrong! */
1866 +       if(num_packets > 99) {
1867 +               printk(KERN_WARNING "layer7: num_packets can't be > 99.\n");
1868 +               num_packets = 99;
1869 +       } else if(num_packets < 1) {
1870 +               printk(KERN_WARNING "layer7: num_packets can't be < 1.\n");
1871 +               num_packets = 1;
1872 +       }
1873 +
1874 +       return count;
1875 +}
1876 +
1877 +static int
1878 +match(const struct sk_buff *skbin,
1879 +      const struct net_device *in,
1880 +      const struct net_device *out,
1881 +      const struct xt_match *match,
1882 +      const void *matchinfo,
1883 +      int offset,
1884 +      unsigned int protoff,
1885 +      int *hotdrop)
1886 +{
1887 +       /* sidestep const without getting a compiler warning... */
1888 +       struct sk_buff * skb = (struct sk_buff *)skbin;
1889 +
1890 +       struct xt_layer7_info * info = (struct xt_layer7_info *)matchinfo;
1891 +       enum ip_conntrack_info master_ctinfo, ctinfo;
1892 +       struct nf_conn *master_conntrack, *conntrack;
1893 +       unsigned char * app_data;
1894 +       unsigned int pattern_result, appdatalen;
1895 +       regexp * comppattern;
1896 +
1897 +       /* Be paranoid/incompetent - lock the entire match function. */
1898 +       spin_lock_bh(&l7_lock);
1899 +
1900 +       if(!can_handle(skb)){
1901 +               DPRINTK("layer7: This is some protocol I can't handle.\n");
1902 +               spin_unlock_bh(&l7_lock);
1903 +               return info->invert;
1904 +       }
1905 +
1906 +       /* Treat parent & all its children together as one connection, except
1907 +       for the purpose of setting conntrack->layer7.app_proto in the actual
1908 +       connection. This makes /proc/net/ip_conntrack more satisfying. */
1909 +       if(!(conntrack = nf_ct_get((struct sk_buff *)skb, &ctinfo)) ||
1910 +          !(master_conntrack=nf_ct_get((struct sk_buff *)skb,&master_ctinfo))){
1911 +               DPRINTK("layer7: couldn't get conntrack.\n");
1912 +               spin_unlock_bh(&l7_lock);
1913 +               return info->invert;
1914 +       }
1915 +
1916 +       /* Try to get a master conntrack (and its master etc) for FTP, etc. */
1917 +       while (master_ct(master_conntrack) != NULL)
1918 +               master_conntrack = master_ct(master_conntrack);
1919 +
1920 +       /* if we've classified it or seen too many packets */
1921 +       if(TOTAL_PACKETS > num_packets ||
1922 +          master_conntrack->layer7.app_proto) {
1923 +
1924 +               pattern_result = match_no_append(conntrack, master_conntrack,
1925 +                                                ctinfo, master_ctinfo, info);
1926 +
1927 +               /* skb->cb[0] == seen. Don't do things twice if there are
1928 +               multiple l7 rules. I'm not sure that using cb for this purpose
1929 +               is correct, even though it says "put your private variables
1930 +               there". But it doesn't look like it is being used for anything
1931 +               else in the skbs that make it here. */
1932 +               skb->cb[0] = 1; /* marking it seen here's probably irrelevant */
1933 +
1934 +               spin_unlock_bh(&l7_lock);
1935 +               return (pattern_result ^ info->invert);
1936 +       }
1937 +
1938 +       if(skb_is_nonlinear(skb)){
1939 +               if(skb_linearize(skb) != 0){
1940 +                       if (net_ratelimit())
1941 +                               printk(KERN_ERR "layer7: failed to linearize "
1942 +                                               "packet, bailing.\n");
1943 +                       spin_unlock_bh(&l7_lock);
1944 +                       return info->invert;
1945 +               }
1946 +       }
1947 +
1948 +       /* now that the skb is linearized, it's safe to set these. */
1949 +       app_data = skb->data + app_data_offset(skb);
1950 +       appdatalen = skb->tail - app_data;
1951 +
1952 +       /* the return value gets checked later, when we're ready to use it */
1953 +       comppattern = compile_and_cache(info->pattern, info->protocol);
1954 +
1955 +       /* On the first packet of a connection, allocate space for app data */
1956 +       if(TOTAL_PACKETS == 1 && !skb->cb[0] &&
1957 +          !master_conntrack->layer7.app_data){
1958 +               master_conntrack->layer7.app_data =
1959 +                       kmalloc(maxdatalen, GFP_ATOMIC);
1960 +               if(!master_conntrack->layer7.app_data){
1961 +                       if (net_ratelimit())
1962 +                               printk(KERN_ERR "layer7: out of memory in "
1963 +                                               "match, bailing.\n");
1964 +                       spin_unlock_bh(&l7_lock);
1965 +                       return info->invert;
1966 +               }
1967 +
1968 +               master_conntrack->layer7.app_data[0] = '\0';
1969 +       }
1970 +
1971 +       /* Can be here, but unallocated, if numpackets is increased near
1972 +       the beginning of a connection */
1973 +       if(master_conntrack->layer7.app_data == NULL){
1974 +               spin_unlock_bh(&l7_lock);
1975 +               return (info->invert); /* unmatched */
1976 +       }
1977 +
1978 +       if(!skb->cb[0]){
1979 +               int newbytes;
1980 +               newbytes = add_data(master_conntrack, app_data, appdatalen);
1981 +
1982 +               if(newbytes == 0) { /* didn't add any data */
1983 +                       skb->cb[0] = 1;
1984 +                       /* Didn't match before, not going to match now */
1985 +                       spin_unlock_bh(&l7_lock);
1986 +                       return info->invert;
1987 +               }
1988 +       }
1989 +
1990 +       /* If looking for "unknown", then never match.  "Unknown" means that
1991 +       we've given up; we're still trying with these packets. */
1992 +       if(!strcmp(info->protocol, "unknown")) {
1993 +               pattern_result = 0;
1994 +       /* If looking for "unset", then always match. "Unset" means that we
1995 +       haven't yet classified the connection. */
1996 +       } else if(!strcmp(info->protocol, "unset")) {
1997 +               pattern_result = 2;
1998 +               DPRINTK("layer7: matched unset: not yet classified "
1999 +                       "(%d/%d packets)\n", TOTAL_PACKETS, num_packets);
2000 +       /* If the regexp failed to compile, don't bother running it */
2001 +       } else if(comppattern &&
2002 +                 regexec(comppattern, master_conntrack->layer7.app_data)){
2003 +               DPRINTK("layer7: matched %s\n", info->protocol);
2004 +               pattern_result = 1;
2005 +       } else pattern_result = 0;
2006 +
2007 +       if(pattern_result == 1) {
2008 +               master_conntrack->layer7.app_proto =
2009 +                       kmalloc(strlen(info->protocol)+1, GFP_ATOMIC);
2010 +               if(!master_conntrack->layer7.app_proto){
2011 +                       if (net_ratelimit())
2012 +                               printk(KERN_ERR "layer7: out of memory in "
2013 +                                               "match, bailing.\n");
2014 +                       spin_unlock_bh(&l7_lock);
2015 +                       return (pattern_result ^ info->invert);
2016 +               }
2017 +               strcpy(master_conntrack->layer7.app_proto, info->protocol);
2018 +       } else if(pattern_result > 1) { /* cleanup from "unset" */
2019 +               pattern_result = 1;
2020 +       }
2021 +
2022 +       /* mark the packet seen */
2023 +       skb->cb[0] = 1;
2024 +
2025 +       spin_unlock_bh(&l7_lock);
2026 +       return (pattern_result ^ info->invert);
2027 +}
2028 +
2029 +static int check(const char *tablename,
2030 +                const void *inf,
2031 +                const struct xt_match *match,
2032 +                void *matchinfo,
2033 +                unsigned int hook_mask)
2034 +
2035 +{
2036 +       // load nf_conntrack_ipv4
2037 +        if (nf_ct_l3proto_try_module_get(match->family) < 0) {
2038 +                printk(KERN_WARNING "can't load conntrack support for "
2039 +                                    "proto=%d\n", match->family);
2040 +                return 0;
2041 +        }
2042 +       return 1;
2043 +}
2044 +
2045 +static void
2046 +destroy(const struct xt_match *match, void *matchinfo)
2047 +{
2048 +       nf_ct_l3proto_module_put(match->family);
2049 +}
2050 +
2051 +static struct xt_match xt_layer7_match[] = {
2052 +{
2053 +       .name           = "layer7",
2054 +       .family         = AF_INET,
2055 +       .checkentry     = check,
2056 +       .match          = match,
2057 +       .destroy        = destroy,
2058 +       .matchsize      = sizeof(struct xt_layer7_info),
2059 +       .me             = THIS_MODULE
2060 +}
2061 +};
2062 +
2063 +static void layer7_cleanup_proc(void)
2064 +{
2065 +       remove_proc_entry("layer7_numpackets", proc_net);
2066 +}
2067 +
2068 +/* register the proc file */
2069 +static void layer7_init_proc(void)
2070 +{
2071 +       struct proc_dir_entry* entry;
2072 +       entry = create_proc_entry("layer7_numpackets", 0644, proc_net);
2073 +       entry->read_proc = layer7_read_proc;
2074 +       entry->write_proc = layer7_write_proc;
2075 +}
2076 +
2077 +static int __init xt_layer7_init(void)
2078 +{
2079 +       need_conntrack();
2080 +
2081 +       layer7_init_proc();
2082 +       if(maxdatalen < 1) {
2083 +               printk(KERN_WARNING "layer7: maxdatalen can't be < 1, "
2084 +                       "using 1\n");
2085 +               maxdatalen = 1;
2086 +       }
2087 +       /* This is not a hard limit.  It's just here to prevent people from
2088 +       bringing their slow machines to a grinding halt. */
2089 +       else if(maxdatalen > 65536) {
2090 +               printk(KERN_WARNING "layer7: maxdatalen can't be > 65536, "
2091 +                       "using 65536\n");
2092 +               maxdatalen = 65536;
2093 +       }
2094 +       return xt_register_matches(xt_layer7_match,
2095 +                                  ARRAY_SIZE(xt_layer7_match));
2096 +}
2097 +
2098 +static void __exit xt_layer7_fini(void)
2099 +{
2100 +       layer7_cleanup_proc();
2101 +       xt_unregister_matches(xt_layer7_match, ARRAY_SIZE(xt_layer7_match));
2102 +}
2103 +
2104 +module_init(xt_layer7_init);
2105 +module_exit(xt_layer7_fini);
2106 +