target/linux/generic-2.6/patches-2.6.27/100-netfilter_layer7_2.17.patch

   1 diff --git a/include/linux/netfilter/xt_layer7.h b/include/linux/netfilter/xt_layer7.h
   2 new file mode 100644
   3 index 0000000..147cd64
   4 --- /dev/null
   5 +++ b/include/linux/netfilter/xt_layer7.h
   6 @@ -0,0 +1,13 @@
   7 +#ifndef _XT_LAYER7_H
   8 +#define _XT_LAYER7_H
   9 +
  10 +#define MAX_PATTERN_LEN 8192
  11 +#define MAX_PROTOCOL_LEN 256
  12 +
  13 +struct xt_layer7_info {
  14 +    char protocol[MAX_PROTOCOL_LEN];
  15 +    char pattern[MAX_PATTERN_LEN];
  16 +    u_int8_t invert;
  17 +};
  18 +
  19 +#endif /* _XT_LAYER7_H */
  20 diff --git a/include/net/netfilter/nf_conntrack.h b/include/net/netfilter/nf_conntrack.h
  21 index b76a868..2a0bb9b 100644
  22 --- a/include/net/netfilter/nf_conntrack.h
  23 +++ b/include/net/netfilter/nf_conntrack.h
  24 @@ -118,6 +118,22 @@ struct nf_conn
  25         u_int32_t secmark;
  26  #endif
  27
  28 +#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || \
  29 +    defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
  30 +       struct {
  31 +               /*
  32 +                * e.g. "http". NULL before decision. "unknown" after decision
  33 +                * if no match.
  34 +                */
  35 +               char *app_proto;
  36 +               /*
  37 +                * application layer data so far. NULL after match decision.
  38 +                */
  39 +               char *app_data;
  40 +               unsigned int app_data_len;
  41 +       } layer7;
  42 +#endif
  43 +
  44         /* Storage reserved for other modules: */
  45         union nf_conntrack_proto proto;
  46
  47 diff --git a/net/netfilter/Kconfig b/net/netfilter/Kconfig
  48 index 25dcef9..29fe15d 100644
  49 --- a/net/netfilter/Kconfig
  50 +++ b/net/netfilter/Kconfig
  51 @@ -795,6 +795,27 @@ config NETFILTER_XT_MATCH_STATE
  52
  53           To compile it as a module, choose M here.  If unsure, say N.
  54
  55 +config NETFILTER_XT_MATCH_LAYER7
  56 +       tristate '"layer7" match support'
  57 +       depends on NETFILTER_XTABLES
  58 +       depends on EXPERIMENTAL && (IP_NF_CONNTRACK || NF_CONNTRACK)
  59 +       depends on NF_CT_ACCT
  60 +       help
  61 +         Say Y if you want to be able to classify connections (and their
  62 +         packets) based on regular expression matching of their application
  63 +         layer data.   This is one way to classify applications such as
  64 +         peer-to-peer filesharing systems that do not always use the same
  65 +         port.
  66 +
  67 +         To compile it as a module, choose M here.  If unsure, say N.
  68 +
  69 +config NETFILTER_XT_MATCH_LAYER7_DEBUG
  70 +        bool 'Layer 7 debugging output'
  71 +        depends on NETFILTER_XT_MATCH_LAYER7
  72 +        help
  73 +          Say Y to get lots of debugging output.
  74 +
  75 +
  76  config NETFILTER_XT_MATCH_STATISTIC
  77         tristate '"statistic" match support'
  78         depends on NETFILTER_ADVANCED
  79 diff --git a/net/netfilter/Makefile b/net/netfilter/Makefile
  80 index da3d909..c15b75d 100644
  81 --- a/net/netfilter/Makefile
  82 +++ b/net/netfilter/Makefile
  83 @@ -84,6 +84,7 @@ obj-$(CONFIG_NETFILTER_XT_MATCH_RECENT) += xt_recent.o
  84  obj-$(CONFIG_NETFILTER_XT_MATCH_SCTP) += xt_sctp.o
  85  obj-$(CONFIG_NETFILTER_XT_MATCH_SOCKET) += xt_socket.o
  86  obj-$(CONFIG_NETFILTER_XT_MATCH_STATE) += xt_state.o
  87 +obj-$(CONFIG_NETFILTER_XT_MATCH_LAYER7) += xt_layer7.o
  88  obj-$(CONFIG_NETFILTER_XT_MATCH_STATISTIC) += xt_statistic.o
  89  obj-$(CONFIG_NETFILTER_XT_MATCH_STRING) += xt_string.o
  90  obj-$(CONFIG_NETFILTER_XT_MATCH_TCPMSS) += xt_tcpmss.o
  91 diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c
  92 index 622d7c6..32959cc 100644
  93 --- a/net/netfilter/nf_conntrack_core.c
  94 +++ b/net/netfilter/nf_conntrack_core.c
  95 @@ -201,6 +201,14 @@ destroy_conntrack(struct nf_conntrack *nfct)
  96          * too. */
  97         nf_ct_remove_expectations(ct);
  98
  99 +       #if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
 100 +       if(ct->layer7.app_proto)
 101 +               kfree(ct->layer7.app_proto);
 102 +       if(ct->layer7.app_data)
 103 +       kfree(ct->layer7.app_data);
 104 +       #endif
 105 +
 106 +
 107         /* We overload first tuple to link into unconfirmed list. */
 108         if (!nf_ct_is_confirmed(ct)) {
 109                 BUG_ON(hlist_unhashed(&ct->tuplehash[IP_CT_DIR_ORIGINAL].hnode));
 110 diff --git a/net/netfilter/nf_conntrack_standalone.c b/net/netfilter/nf_conntrack_standalone.c
 111 index 98106d4..774acad 100644
 112 --- a/net/netfilter/nf_conntrack_standalone.c
 113 +++ b/net/netfilter/nf_conntrack_standalone.c
 114 @@ -165,6 +165,12 @@ static int ct_seq_show(struct seq_file *s, void *v)
 115                 return -ENOSPC;
 116  #endif
 117
 118 +#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
 119 +       if(ct->layer7.app_proto &&
 120 +           seq_printf(s, "l7proto=%s ", ct->layer7.app_proto))
 121 +               return -ENOSPC;
 122 +#endif
 123 +
 124         if (seq_printf(s, "use=%u\n", atomic_read(&ct->ct_general.use)))
 125                 return -ENOSPC;
 126
 127 diff --git a/net/netfilter/regexp/regexp.c b/net/netfilter/regexp/regexp.c
 128 new file mode 100644
 129 index 0000000..9006988
 130 --- /dev/null
 131 +++ b/net/netfilter/regexp/regexp.c
 132 @@ -0,0 +1,1197 @@
 133 +/*
 134 + * regcomp and regexec -- regsub and regerror are elsewhere
 135 + * @(#)regexp.c        1.3 of 18 April 87
 136 + *
 137 + *     Copyright (c) 1986 by University of Toronto.
 138 + *     Written by Henry Spencer.  Not derived from licensed software.
 139 + *
 140 + *     Permission is granted to anyone to use this software for any
 141 + *     purpose on any computer system, and to redistribute it freely,
 142 + *     subject to the following restrictions:
 143 + *
 144 + *     1. The author is not responsible for the consequences of use of
 145 + *             this software, no matter how awful, even if they arise
 146 + *             from defects in it.
 147 + *
 148 + *     2. The origin of this software must not be misrepresented, either
 149 + *             by explicit claim or by omission.
 150 + *
 151 + *     3. Altered versions must be plainly marked as such, and must not
 152 + *             be misrepresented as being the original software.
 153 + *
 154 + * Beware that some of this code is subtly aware of the way operator
 155 + * precedence is structured in regular expressions.  Serious changes in
 156 + * regular-expression syntax might require a total rethink.
 157 + *
 158 + * This code was modified by Ethan Sommer to work within the kernel
 159 + * (it now uses kmalloc etc..)
 160 + *
 161 + * Modified slightly by Matthew Strait to use more modern C.
 162 + */
 163 +
 164 +#include "regexp.h"
 165 +#include "regmagic.h"
 166 +
 167 +/* added by ethan and matt.  Lets it work in both kernel and user space.
 168 +(So iptables can use it, for instance.)  Yea, it goes both ways... */
 169 +#if __KERNEL__
 170 +  #define malloc(foo) kmalloc(foo,GFP_ATOMIC)
 171 +#else
 172 +  #define printk(format,args...) printf(format,##args)
 173 +#endif
 174 +
 175 +void regerror(char * s)
 176 +{
 177 +        printk("<3>Regexp: %s\n", s);
 178 +        /* NOTREACHED */
 179 +}
 180 +
 181 +/*
 182 + * The "internal use only" fields in regexp.h are present to pass info from
 183 + * compile to execute that permits the execute phase to run lots faster on
 184 + * simple cases.  They are:
 185 + *
 186 + * regstart    char that must begin a match; '\0' if none obvious
 187 + * reganch     is the match anchored (at beginning-of-line only)?
 188 + * regmust     string (pointer into program) that match must include, or NULL
 189 + * regmlen     length of regmust string
 190 + *
 191 + * Regstart and reganch permit very fast decisions on suitable starting points
 192 + * for a match, cutting down the work a lot.  Regmust permits fast rejection
 193 + * of lines that cannot possibly match.  The regmust tests are costly enough
 194 + * that regcomp() supplies a regmust only if the r.e. contains something
 195 + * potentially expensive (at present, the only such thing detected is * or +
 196 + * at the start of the r.e., which can involve a lot of backup).  Regmlen is
 197 + * supplied because the test in regexec() needs it and regcomp() is computing
 198 + * it anyway.
 199 + */
 200 +
 201 +/*
 202 + * Structure for regexp "program".  This is essentially a linear encoding
 203 + * of a nondeterministic finite-state machine (aka syntax charts or
 204 + * "railroad normal form" in parsing technology).  Each node is an opcode
 205 + * plus a "next" pointer, possibly plus an operand.  "Next" pointers of
 206 + * all nodes except BRANCH implement concatenation; a "next" pointer with
 207 + * a BRANCH on both ends of it is connecting two alternatives.  (Here we
 208 + * have one of the subtle syntax dependencies:  an individual BRANCH (as
 209 + * opposed to a collection of them) is never concatenated with anything
 210 + * because of operator precedence.)  The operand of some types of node is
 211 + * a literal string; for others, it is a node leading into a sub-FSM.  In
 212 + * particular, the operand of a BRANCH node is the first node of the branch.
 213 + * (NB this is *not* a tree structure:  the tail of the branch connects
 214 + * to the thing following the set of BRANCHes.)  The opcodes are:
 215 + */
 216 +
 217 +/* definition  number  opnd?   meaning */
 218 +#define        END     0       /* no   End of program. */
 219 +#define        BOL     1       /* no   Match "" at beginning of line. */
 220 +#define        EOL     2       /* no   Match "" at end of line. */
 221 +#define        ANY     3       /* no   Match any one character. */
 222 +#define        ANYOF   4       /* str  Match any character in this string. */
 223 +#define        ANYBUT  5       /* str  Match any character not in this string. */
 224 +#define        BRANCH  6       /* node Match this alternative, or the next... */
 225 +#define        BACK    7       /* no   Match "", "next" ptr points backward. */
 226 +#define        EXACTLY 8       /* str  Match this string. */
 227 +#define        NOTHING 9       /* no   Match empty string. */
 228 +#define        STAR    10      /* node Match this (simple) thing 0 or more times. */
 229 +#define        PLUS    11      /* node Match this (simple) thing 1 or more times. */
 230 +#define        OPEN    20      /* no   Mark this point in input as start of #n. */
 231 +                       /*      OPEN+1 is number 1, etc. */
 232 +#define        CLOSE   30      /* no   Analogous to OPEN. */
 233 +
 234 +/*
 235 + * Opcode notes:
 236 + *
 237 + * BRANCH      The set of branches constituting a single choice are hooked
 238 + *             together with their "next" pointers, since precedence prevents
 239 + *             anything being concatenated to any individual branch.  The
 240 + *             "next" pointer of the last BRANCH in a choice points to the
 241 + *             thing following the whole choice.  This is also where the
 242 + *             final "next" pointer of each individual branch points; each
 243 + *             branch starts with the operand node of a BRANCH node.
 244 + *
 245 + * BACK                Normal "next" pointers all implicitly point forward; BACK
 246 + *             exists to make loop structures possible.
 247 + *
 248 + * STAR,PLUS   '?', and complex '*' and '+', are implemented as circular
 249 + *             BRANCH structures using BACK.  Simple cases (one character
 250 + *             per match) are implemented with STAR and PLUS for speed
 251 + *             and to minimize recursive plunges.
 252 + *
 253 + * OPEN,CLOSE  ...are numbered at compile time.
 254 + */
 255 +
 256 +/*
 257 + * A node is one char of opcode followed by two chars of "next" pointer.
 258 + * "Next" pointers are stored as two 8-bit pieces, high order first.  The
 259 + * value is a positive offset from the opcode of the node containing it.
 260 + * An operand, if any, simply follows the node.  (Note that much of the
 261 + * code generation knows about this implicit relationship.)
 262 + *
 263 + * Using two bytes for the "next" pointer is vast overkill for most things,
 264 + * but allows patterns to get big without disasters.
 265 + */
 266 +#define        OP(p)   (*(p))
 267 +#define        NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377))
 268 +#define        OPERAND(p)      ((p) + 3)
 269 +
 270 +/*
 271 + * See regmagic.h for one further detail of program structure.
 272 + */
 273 +
 274 +
 275 +/*
 276 + * Utility definitions.
 277 + */
 278 +#ifndef CHARBITS
 279 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
 280 +#else
 281 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
 282 +#endif
 283 +
 284 +#define        FAIL(m) { regerror(m); return(NULL); }
 285 +#define        ISMULT(c)       ((c) == '*' || (c) == '+' || (c) == '?')
 286 +#define        META    "^$.[()|?+*\\"
 287 +
 288 +/*
 289 + * Flags to be passed up and down.
 290 + */
 291 +#define        HASWIDTH        01      /* Known never to match null string. */
 292 +#define        SIMPLE          02      /* Simple enough to be STAR/PLUS operand. */
 293 +#define        SPSTART         04      /* Starts with * or +. */
 294 +#define        WORST           0       /* Worst case. */
 295 +
 296 +/*
 297 + * Global work variables for regcomp().
 298 + */
 299 +struct match_globals {
 300 +char *reginput;                /* String-input pointer. */
 301 +char *regbol;          /* Beginning of input, for ^ check. */
 302 +char **regstartp;      /* Pointer to startp array. */
 303 +char **regendp;                /* Ditto for endp. */
 304 +char *regparse;                /* Input-scan pointer. */
 305 +int regnpar;           /* () count. */
 306 +char regdummy;
 307 +char *regcode;         /* Code-emit pointer; &regdummy = don't. */
 308 +long regsize;          /* Code size. */
 309 +};
 310 +
 311 +/*
 312 + * Forward declarations for regcomp()'s friends.
 313 + */
 314 +#ifndef STATIC
 315 +#define        STATIC  static
 316 +#endif
 317 +STATIC char *reg(struct match_globals *g, int paren,int *flagp);
 318 +STATIC char *regbranch(struct match_globals *g, int *flagp);
 319 +STATIC char *regpiece(struct match_globals *g, int *flagp);
 320 +STATIC char *regatom(struct match_globals *g, int *flagp);
 321 +STATIC char *regnode(struct match_globals *g, char op);
 322 +STATIC char *regnext(struct match_globals *g, char *p);
 323 +STATIC void regc(struct match_globals *g, char b);
 324 +STATIC void reginsert(struct match_globals *g, char op, char *opnd);
 325 +STATIC void regtail(struct match_globals *g, char *p, char *val);
 326 +STATIC void regoptail(struct match_globals *g, char *p, char *val);
 327 +
 328 +
 329 +__kernel_size_t my_strcspn(const char *s1,const char *s2)
 330 +{
 331 +        char *scan1;
 332 +        char *scan2;
 333 +        int count;
 334 +
 335 +        count = 0;
 336 +        for (scan1 = (char *)s1; *scan1 != '\0'; scan1++) {
 337 +                for (scan2 = (char *)s2; *scan2 != '\0';)       /* ++ moved down. */
 338 +                        if (*scan1 == *scan2++)
 339 +                                return(count);
 340 +                count++;
 341 +        }
 342 +        return(count);
 343 +}
 344 +
 345 +/*
 346 + - regcomp - compile a regular expression into internal code
 347 + *
 348 + * We can't allocate space until we know how big the compiled form will be,
 349 + * but we can't compile it (and thus know how big it is) until we've got a
 350 + * place to put the code.  So we cheat:  we compile it twice, once with code
 351 + * generation turned off and size counting turned on, and once "for real".
 352 + * This also means that we don't allocate space until we are sure that the
 353 + * thing really will compile successfully, and we never have to move the
 354 + * code and thus invalidate pointers into it.  (Note that it has to be in
 355 + * one piece because free() must be able to free it all.)
 356 + *
 357 + * Beware that the optimization-preparation code in here knows about some
 358 + * of the structure of the compiled regexp.
 359 + */
 360 +regexp *
 361 +regcomp(char *exp,int *patternsize)
 362 +{
 363 +       register regexp *r;
 364 +       register char *scan;
 365 +       register char *longest;
 366 +       register int len;
 367 +       int flags;
 368 +       struct match_globals g;
 369 +
 370 +       /* commented out by ethan
 371 +          extern char *malloc();
 372 +       */
 373 +
 374 +       if (exp == NULL)
 375 +               FAIL("NULL argument");
 376 +
 377 +       /* First pass: determine size, legality. */
 378 +       g.regparse = exp;
 379 +       g.regnpar = 1;
 380 +       g.regsize = 0L;
 381 +       g.regcode = &g.regdummy;
 382 +       regc(&g, MAGIC);
 383 +       if (reg(&g, 0, &flags) == NULL)
 384 +               return(NULL);
 385 +
 386 +       /* Small enough for pointer-storage convention? */
 387 +       if (g.regsize >= 32767L)                /* Probably could be 65535L. */
 388 +               FAIL("regexp too big");
 389 +
 390 +       /* Allocate space. */
 391 +       *patternsize=sizeof(regexp) + (unsigned)g.regsize;
 392 +       r = (regexp *)malloc(sizeof(regexp) + (unsigned)g.regsize);
 393 +       if (r == NULL)
 394 +               FAIL("out of space");
 395 +
 396 +       /* Second pass: emit code. */
 397 +       g.regparse = exp;
 398 +       g.regnpar = 1;
 399 +       g.regcode = r->program;
 400 +       regc(&g, MAGIC);
 401 +       if (reg(&g, 0, &flags) == NULL)
 402 +               return(NULL);
 403 +
 404 +       /* Dig out information for optimizations. */
 405 +       r->regstart = '\0';     /* Worst-case defaults. */
 406 +       r->reganch = 0;
 407 +       r->regmust = NULL;
 408 +       r->regmlen = 0;
 409 +       scan = r->program+1;                    /* First BRANCH. */
 410 +       if (OP(regnext(&g, scan)) == END) {             /* Only one top-level choice. */
 411 +               scan = OPERAND(scan);
 412 +
 413 +               /* Starting-point info. */
 414 +               if (OP(scan) == EXACTLY)
 415 +                       r->regstart = *OPERAND(scan);
 416 +               else if (OP(scan) == BOL)
 417 +                       r->reganch++;
 418 +
 419 +               /*
 420 +                * If there's something expensive in the r.e., find the
 421 +                * longest literal string that must appear and make it the
 422 +                * regmust.  Resolve ties in favor of later strings, since
 423 +                * the regstart check works with the beginning of the r.e.
 424 +                * and avoiding duplication strengthens checking.  Not a
 425 +                * strong reason, but sufficient in the absence of others.
 426 +                */
 427 +               if (flags&SPSTART) {
 428 +                       longest = NULL;
 429 +                       len = 0;
 430 +                       for (; scan != NULL; scan = regnext(&g, scan))
 431 +                               if (OP(scan) == EXACTLY && strlen(OPERAND(scan)) >= len) {
 432 +                                       longest = OPERAND(scan);
 433 +                                       len = strlen(OPERAND(scan));
 434 +                               }
 435 +                       r->regmust = longest;
 436 +                       r->regmlen = len;
 437 +               }
 438 +       }
 439 +
 440 +       return(r);
 441 +}
 442 +
 443 +/*
 444 + - reg - regular expression, i.e. main body or parenthesized thing
 445 + *
 446 + * Caller must absorb opening parenthesis.
 447 + *
 448 + * Combining parenthesis handling with the base level of regular expression
 449 + * is a trifle forced, but the need to tie the tails of the branches to what
 450 + * follows makes it hard to avoid.
 451 + */
 452 +static char *
 453 +reg(struct match_globals *g, int paren, int *flagp /* Parenthesized? */ )
 454 +{
 455 +       register char *ret;
 456 +       register char *br;
 457 +       register char *ender;
 458 +       register int parno = 0; /* 0 makes gcc happy */
 459 +       int flags;
 460 +
 461 +       *flagp = HASWIDTH;      /* Tentatively. */
 462 +
 463 +       /* Make an OPEN node, if parenthesized. */
 464 +       if (paren) {
 465 +               if (g->regnpar >= NSUBEXP)
 466 +                       FAIL("too many ()");
 467 +               parno = g->regnpar;
 468 +               g->regnpar++;
 469 +               ret = regnode(g, OPEN+parno);
 470 +       } else
 471 +               ret = NULL;
 472 +
 473 +       /* Pick up the branches, linking them together. */
 474 +       br = regbranch(g, &flags);
 475 +       if (br == NULL)
 476 +               return(NULL);
 477 +       if (ret != NULL)
 478 +               regtail(g, ret, br);    /* OPEN -> first. */
 479 +       else
 480 +               ret = br;
 481 +       if (!(flags&HASWIDTH))
 482 +               *flagp &= ~HASWIDTH;
 483 +       *flagp |= flags&SPSTART;
 484 +       while (*g->regparse == '|') {
 485 +               g->regparse++;
 486 +               br = regbranch(g, &flags);
 487 +               if (br == NULL)
 488 +                       return(NULL);
 489 +               regtail(g, ret, br);    /* BRANCH -> BRANCH. */
 490 +               if (!(flags&HASWIDTH))
 491 +                       *flagp &= ~HASWIDTH;
 492 +               *flagp |= flags&SPSTART;
 493 +       }
 494 +
 495 +       /* Make a closing node, and hook it on the end. */
 496 +       ender = regnode(g, (paren) ? CLOSE+parno : END);
 497 +       regtail(g, ret, ender);
 498 +
 499 +       /* Hook the tails of the branches to the closing node. */
 500 +       for (br = ret; br != NULL; br = regnext(g, br))
 501 +               regoptail(g, br, ender);
 502 +
 503 +       /* Check for proper termination. */
 504 +       if (paren && *g->regparse++ != ')') {
 505 +               FAIL("unmatched ()");
 506 +       } else if (!paren && *g->regparse != '\0') {
 507 +               if (*g->regparse == ')') {
 508 +                       FAIL("unmatched ()");
 509 +               } else
 510 +                       FAIL("junk on end");    /* "Can't happen". */
 511 +               /* NOTREACHED */
 512 +       }
 513 +
 514 +       return(ret);
 515 +}
 516 +
 517 +/*
 518 + - regbranch - one alternative of an | operator
 519 + *
 520 + * Implements the concatenation operator.
 521 + */
 522 +static char *
 523 +regbranch(struct match_globals *g, int *flagp)
 524 +{
 525 +       register char *ret;
 526 +       register char *chain;
 527 +       register char *latest;
 528 +       int flags;
 529 +
 530 +       *flagp = WORST;         /* Tentatively. */
 531 +
 532 +       ret = regnode(g, BRANCH);
 533 +       chain = NULL;
 534 +       while (*g->regparse != '\0' && *g->regparse != '|' && *g->regparse != ')') {
 535 +               latest = regpiece(g, &flags);
 536 +               if (latest == NULL)
 537 +                       return(NULL);
 538 +               *flagp |= flags&HASWIDTH;
 539 +               if (chain == NULL)      /* First piece. */
 540 +                       *flagp |= flags&SPSTART;
 541 +               else
 542 +                       regtail(g, chain, latest);
 543 +               chain = latest;
 544 +       }
 545 +       if (chain == NULL)      /* Loop ran zero times. */
 546 +               (void) regnode(g, NOTHING);
 547 +
 548 +       return(ret);
 549 +}
 550 +
 551 +/*
 552 + - regpiece - something followed by possible [*+?]
 553 + *
 554 + * Note that the branching code sequences used for ? and the general cases
 555 + * of * and + are somewhat optimized:  they use the same NOTHING node as
 556 + * both the endmarker for their branch list and the body of the last branch.
 557 + * It might seem that this node could be dispensed with entirely, but the
 558 + * endmarker role is not redundant.
 559 + */
 560 +static char *
 561 +regpiece(struct match_globals *g, int *flagp)
 562 +{
 563 +       register char *ret;
 564 +       register char op;
 565 +       register char *next;
 566 +       int flags;
 567 +
 568 +       ret = regatom(g, &flags);
 569 +       if (ret == NULL)
 570 +               return(NULL);
 571 +
 572 +       op = *g->regparse;
 573 +       if (!ISMULT(op)) {
 574 +               *flagp = flags;
 575 +               return(ret);
 576 +       }
 577 +
 578 +       if (!(flags&HASWIDTH) && op != '?')
 579 +               FAIL("*+ operand could be empty");
 580 +       *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
 581 +
 582 +       if (op == '*' && (flags&SIMPLE))
 583 +               reginsert(g, STAR, ret);
 584 +       else if (op == '*') {
 585 +               /* Emit x* as (x&|), where & means "self". */
 586 +               reginsert(g, BRANCH, ret);                      /* Either x */
 587 +               regoptail(g, ret, regnode(g, BACK));            /* and loop */
 588 +               regoptail(g, ret, ret);                 /* back */
 589 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
 590 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
 591 +       } else if (op == '+' && (flags&SIMPLE))
 592 +               reginsert(g, PLUS, ret);
 593 +       else if (op == '+') {
 594 +               /* Emit x+ as x(&|), where & means "self". */
 595 +               next = regnode(g, BRANCH);                      /* Either */
 596 +               regtail(g, ret, next);
 597 +               regtail(g, regnode(g, BACK), ret);              /* loop back */
 598 +               regtail(g, next, regnode(g, BRANCH));           /* or */
 599 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
 600 +       } else if (op == '?') {
 601 +               /* Emit x? as (x|) */
 602 +               reginsert(g, BRANCH, ret);                      /* Either x */
 603 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
 604 +               next = regnode(g, NOTHING);             /* null. */
 605 +               regtail(g, ret, next);
 606 +               regoptail(g, ret, next);
 607 +       }
 608 +       g->regparse++;
 609 +       if (ISMULT(*g->regparse))
 610 +               FAIL("nested *?+");
 611 +
 612 +       return(ret);
 613 +}
 614 +
 615 +/*
 616 + - regatom - the lowest level
 617 + *
 618 + * Optimization:  gobbles an entire sequence of ordinary characters so that
 619 + * it can turn them into a single node, which is smaller to store and
 620 + * faster to run.  Backslashed characters are exceptions, each becoming a
 621 + * separate node; the code is simpler that way and it's not worth fixing.
 622 + */
 623 +static char *
 624 +regatom(struct match_globals *g, int *flagp)
 625 +{
 626 +       register char *ret;
 627 +       int flags;
 628 +
 629 +       *flagp = WORST;         /* Tentatively. */
 630 +
 631 +       switch (*g->regparse++) {
 632 +       case '^':
 633 +               ret = regnode(g, BOL);
 634 +               break;
 635 +       case '$':
 636 +               ret = regnode(g, EOL);
 637 +               break;
 638 +       case '.':
 639 +               ret = regnode(g, ANY);
 640 +               *flagp |= HASWIDTH|SIMPLE;
 641 +               break;
 642 +       case '[': {
 643 +                       register int class;
 644 +                       register int classend;
 645 +
 646 +                       if (*g->regparse == '^') {      /* Complement of range. */
 647 +                               ret = regnode(g, ANYBUT);
 648 +                               g->regparse++;
 649 +                       } else
 650 +                               ret = regnode(g, ANYOF);
 651 +                       if (*g->regparse == ']' || *g->regparse == '-')
 652 +                               regc(g, *g->regparse++);
 653 +                       while (*g->regparse != '\0' && *g->regparse != ']') {
 654 +                               if (*g->regparse == '-') {
 655 +                                       g->regparse++;
 656 +                                       if (*g->regparse == ']' || *g->regparse == '\0')
 657 +                                               regc(g, '-');
 658 +                                       else {
 659 +                                               class = UCHARAT(g->regparse-2)+1;
 660 +                                               classend = UCHARAT(g->regparse);
 661 +                                               if (class > classend+1)
 662 +                                                       FAIL("invalid [] range");
 663 +                                               for (; class <= classend; class++)
 664 +                                                       regc(g, class);
 665 +                                               g->regparse++;
 666 +                                       }
 667 +                               } else
 668 +                                       regc(g, *g->regparse++);
 669 +                       }
 670 +                       regc(g, '\0');
 671 +                       if (*g->regparse != ']')
 672 +                               FAIL("unmatched []");
 673 +                       g->regparse++;
 674 +                       *flagp |= HASWIDTH|SIMPLE;
 675 +               }
 676 +               break;
 677 +       case '(':
 678 +               ret = reg(g, 1, &flags);
 679 +               if (ret == NULL)
 680 +                       return(NULL);
 681 +               *flagp |= flags&(HASWIDTH|SPSTART);
 682 +               break;
 683 +       case '\0':
 684 +       case '|':
 685 +       case ')':
 686 +               FAIL("internal urp");   /* Supposed to be caught earlier. */
 687 +               break;
 688 +       case '?':
 689 +       case '+':
 690 +       case '*':
 691 +               FAIL("?+* follows nothing");
 692 +               break;
 693 +       case '\\':
 694 +               if (*g->regparse == '\0')
 695 +                       FAIL("trailing \\");
 696 +               ret = regnode(g, EXACTLY);
 697 +               regc(g, *g->regparse++);
 698 +               regc(g, '\0');
 699 +               *flagp |= HASWIDTH|SIMPLE;
 700 +               break;
 701 +       default: {
 702 +                       register int len;
 703 +                       register char ender;
 704 +
 705 +                       g->regparse--;
 706 +                       len = my_strcspn((const char *)g->regparse, (const char *)META);
 707 +                       if (len <= 0)
 708 +                               FAIL("internal disaster");
 709 +                       ender = *(g->regparse+len);
 710 +                       if (len > 1 && ISMULT(ender))
 711 +                               len--;          /* Back off clear of ?+* operand. */
 712 +                       *flagp |= HASWIDTH;
 713 +                       if (len == 1)
 714 +                               *flagp |= SIMPLE;
 715 +                       ret = regnode(g, EXACTLY);
 716 +                       while (len > 0) {
 717 +                               regc(g, *g->regparse++);
 718 +                               len--;
 719 +                       }
 720 +                       regc(g, '\0');
 721 +               }
 722 +               break;
 723 +       }
 724 +
 725 +       return(ret);
 726 +}
 727 +
 728 +/*
 729 + - regnode - emit a node
 730 + */
 731 +static char *                  /* Location. */
 732 +regnode(struct match_globals *g, char op)
 733 +{
 734 +       register char *ret;
 735 +       register char *ptr;
 736 +
 737 +       ret = g->regcode;
 738 +       if (ret == &g->regdummy) {
 739 +               g->regsize += 3;
 740 +               return(ret);
 741 +       }
 742 +
 743 +       ptr = ret;
 744 +       *ptr++ = op;
 745 +       *ptr++ = '\0';          /* Null "next" pointer. */
 746 +       *ptr++ = '\0';
 747 +       g->regcode = ptr;
 748 +
 749 +       return(ret);
 750 +}
 751 +
 752 +/*
 753 + - regc - emit (if appropriate) a byte of code
 754 + */
 755 +static void
 756 +regc(struct match_globals *g, char b)
 757 +{
 758 +       if (g->regcode != &g->regdummy)
 759 +               *g->regcode++ = b;
 760 +       else
 761 +               g->regsize++;
 762 +}
 763 +
 764 +/*
 765 + - reginsert - insert an operator in front of already-emitted operand
 766 + *
 767 + * Means relocating the operand.
 768 + */
 769 +static void
 770 +reginsert(struct match_globals *g, char op, char* opnd)
 771 +{
 772 +       register char *src;
 773 +       register char *dst;
 774 +       register char *place;
 775 +
 776 +       if (g->regcode == &g->regdummy) {
 777 +               g->regsize += 3;
 778 +               return;
 779 +       }
 780 +
 781 +       src = g->regcode;
 782 +       g->regcode += 3;
 783 +       dst = g->regcode;
 784 +       while (src > opnd)
 785 +               *--dst = *--src;
 786 +
 787 +       place = opnd;           /* Op node, where operand used to be. */
 788 +       *place++ = op;
 789 +       *place++ = '\0';
 790 +       *place++ = '\0';
 791 +}
 792 +
 793 +/*
 794 + - regtail - set the next-pointer at the end of a node chain
 795 + */
 796 +static void
 797 +regtail(struct match_globals *g, char *p, char *val)
 798 +{
 799 +       register char *scan;
 800 +       register char *temp;
 801 +       register int offset;
 802 +
 803 +       if (p == &g->regdummy)
 804 +               return;
 805 +
 806 +       /* Find last node. */
 807 +       scan = p;
 808 +       for (;;) {
 809 +               temp = regnext(g, scan);
 810 +               if (temp == NULL)
 811 +                       break;
 812 +               scan = temp;
 813 +       }
 814 +
 815 +       if (OP(scan) == BACK)
 816 +               offset = scan - val;
 817 +       else
 818 +               offset = val - scan;
 819 +       *(scan+1) = (offset>>8)&0377;
 820 +       *(scan+2) = offset&0377;
 821 +}
 822 +
 823 +/*
 824 + - regoptail - regtail on operand of first argument; nop if operandless
 825 + */
 826 +static void
 827 +regoptail(struct match_globals *g, char *p, char *val)
 828 +{
 829 +       /* "Operandless" and "op != BRANCH" are synonymous in practice. */
 830 +       if (p == NULL || p == &g->regdummy || OP(p) != BRANCH)
 831 +               return;
 832 +       regtail(g, OPERAND(p), val);
 833 +}
 834 +
 835 +/*
 836 + * regexec and friends
 837 + */
 838 +
 839 +
 840 +/*
 841 + * Forwards.
 842 + */
 843 +STATIC int regtry(struct match_globals *g, regexp *prog, char *string);
 844 +STATIC int regmatch(struct match_globals *g, char *prog);
 845 +STATIC int regrepeat(struct match_globals *g, char *p);
 846 +
 847 +#ifdef DEBUG
 848 +int regnarrate = 0;
 849 +void regdump();
 850 +STATIC char *regprop(char *op);
 851 +#endif
 852 +
 853 +/*
 854 + - regexec - match a regexp against a string
 855 + */
 856 +int
 857 +regexec(regexp *prog, char *string)
 858 +{
 859 +       register char *s;
 860 +       struct match_globals g;
 861 +
 862 +       /* Be paranoid... */
 863 +       if (prog == NULL || string == NULL) {
 864 +               printk("<3>Regexp: NULL parameter\n");
 865 +               return(0);
 866 +       }
 867 +
 868 +       /* Check validity of program. */
 869 +       if (UCHARAT(prog->program) != MAGIC) {
 870 +               printk("<3>Regexp: corrupted program\n");
 871 +               return(0);
 872 +       }
 873 +
 874 +       /* If there is a "must appear" string, look for it. */
 875 +       if (prog->regmust != NULL) {
 876 +               s = string;
 877 +               while ((s = strchr(s, prog->regmust[0])) != NULL) {
 878 +                       if (strncmp(s, prog->regmust, prog->regmlen) == 0)
 879 +                               break;  /* Found it. */
 880 +                       s++;
 881 +               }
 882 +               if (s == NULL)  /* Not present. */
 883 +                       return(0);
 884 +       }
 885 +
 886 +       /* Mark beginning of line for ^ . */
 887 +       g.regbol = string;
 888 +
 889 +       /* Simplest case:  anchored match need be tried only once. */
 890 +       if (prog->reganch)
 891 +               return(regtry(&g, prog, string));
 892 +
 893 +       /* Messy cases:  unanchored match. */
 894 +       s = string;
 895 +       if (prog->regstart != '\0')
 896 +               /* We know what char it must start with. */
 897 +               while ((s = strchr(s, prog->regstart)) != NULL) {
 898 +                       if (regtry(&g, prog, s))
 899 +                               return(1);
 900 +                       s++;
 901 +               }
 902 +       else
 903 +               /* We don't -- general case. */
 904 +               do {
 905 +                       if (regtry(&g, prog, s))
 906 +                               return(1);
 907 +               } while (*s++ != '\0');
 908 +
 909 +       /* Failure. */
 910 +       return(0);
 911 +}
 912 +
 913 +/*
 914 + - regtry - try match at specific point
 915 + */
 916 +static int                     /* 0 failure, 1 success */
 917 +regtry(struct match_globals *g, regexp *prog, char *string)
 918 +{
 919 +       register int i;
 920 +       register char **sp;
 921 +       register char **ep;
 922 +
 923 +       g->reginput = string;
 924 +       g->regstartp = prog->startp;
 925 +       g->regendp = prog->endp;
 926 +
 927 +       sp = prog->startp;
 928 +       ep = prog->endp;
 929 +       for (i = NSUBEXP; i > 0; i--) {
 930 +               *sp++ = NULL;
 931 +               *ep++ = NULL;
 932 +       }
 933 +       if (regmatch(g, prog->program + 1)) {
 934 +               prog->startp[0] = string;
 935 +               prog->endp[0] = g->reginput;
 936 +               return(1);
 937 +       } else
 938 +               return(0);
 939 +}
 940 +
 941 +/*
 942 + - regmatch - main matching routine
 943 + *
 944 + * Conceptually the strategy is simple:  check to see whether the current
 945 + * node matches, call self recursively to see whether the rest matches,
 946 + * and then act accordingly.  In practice we make some effort to avoid
 947 + * recursion, in particular by going through "ordinary" nodes (that don't
 948 + * need to know whether the rest of the match failed) by a loop instead of
 949 + * by recursion.
 950 + */
 951 +static int                     /* 0 failure, 1 success */
 952 +regmatch(struct match_globals *g, char *prog)
 953 +{
 954 +       register char *scan = prog; /* Current node. */
 955 +       char *next;                 /* Next node. */
 956 +
 957 +#ifdef DEBUG
 958 +       if (scan != NULL && regnarrate)
 959 +               fprintf(stderr, "%s(\n", regprop(scan));
 960 +#endif
 961 +       while (scan != NULL) {
 962 +#ifdef DEBUG
 963 +               if (regnarrate)
 964 +                       fprintf(stderr, "%s...\n", regprop(scan));
 965 +#endif
 966 +               next = regnext(g, scan);
 967 +
 968 +               switch (OP(scan)) {
 969 +               case BOL:
 970 +                       if (g->reginput != g->regbol)
 971 +                               return(0);
 972 +                       break;
 973 +               case EOL:
 974 +                       if (*g->reginput != '\0')
 975 +                               return(0);
 976 +                       break;
 977 +               case ANY:
 978 +                       if (*g->reginput == '\0')
 979 +                               return(0);
 980 +                       g->reginput++;
 981 +                       break;
 982 +               case EXACTLY: {
 983 +                               register int len;
 984 +                               register char *opnd;
 985 +
 986 +                               opnd = OPERAND(scan);
 987 +                               /* Inline the first character, for speed. */
 988 +                               if (*opnd != *g->reginput)
 989 +                                       return(0);
 990 +                               len = strlen(opnd);
 991 +                               if (len > 1 && strncmp(opnd, g->reginput, len) != 0)
 992 +                                       return(0);
 993 +                               g->reginput += len;
 994 +                       }
 995 +                       break;
 996 +               case ANYOF:
 997 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) == NULL)
 998 +                               return(0);
 999 +                       g->reginput++;
1000 +                       break;
1001 +               case ANYBUT:
1002 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) != NULL)
1003 +                               return(0);
1004 +                       g->reginput++;
1005 +                       break;
1006 +               case NOTHING:
1007 +               case BACK:
1008 +                       break;
1009 +               case OPEN+1:
1010 +               case OPEN+2:
1011 +               case OPEN+3:
1012 +               case OPEN+4:
1013 +               case OPEN+5:
1014 +               case OPEN+6:
1015 +               case OPEN+7:
1016 +               case OPEN+8:
1017 +               case OPEN+9: {
1018 +                               register int no;
1019 +                               register char *save;
1020 +
1021 +                               no = OP(scan) - OPEN;
1022 +                               save = g->reginput;
1023 +
1024 +                               if (regmatch(g, next)) {
1025 +                                       /*
1026 +                                        * Don't set startp if some later
1027 +                                        * invocation of the same parentheses
1028 +                                        * already has.
1029 +                                        */
1030 +                                       if (g->regstartp[no] == NULL)
1031 +                                               g->regstartp[no] = save;
1032 +                                       return(1);
1033 +                               } else
1034 +                                       return(0);
1035 +                       }
1036 +                       break;
1037 +               case CLOSE+1:
1038 +               case CLOSE+2:
1039 +               case CLOSE+3:
1040 +               case CLOSE+4:
1041 +               case CLOSE+5:
1042 +               case CLOSE+6:
1043 +               case CLOSE+7:
1044 +               case CLOSE+8:
1045 +               case CLOSE+9:
1046 +                       {
1047 +                               register int no;
1048 +                               register char *save;
1049 +
1050 +                               no = OP(scan) - CLOSE;
1051 +                               save = g->reginput;
1052 +
1053 +                               if (regmatch(g, next)) {
1054 +                                       /*
1055 +                                        * Don't set endp if some later
1056 +                                        * invocation of the same parentheses
1057 +                                        * already has.
1058 +                                        */
1059 +                                       if (g->regendp[no] == NULL)
1060 +                                               g->regendp[no] = save;
1061 +                                       return(1);
1062 +                               } else
1063 +                                       return(0);
1064 +                       }
1065 +                       break;
1066 +               case BRANCH: {
1067 +                               register char *save;
1068 +
1069 +                               if (OP(next) != BRANCH)         /* No choice. */
1070 +                                       next = OPERAND(scan);   /* Avoid recursion. */
1071 +                               else {
1072 +                                       do {
1073 +                                               save = g->reginput;
1074 +                                               if (regmatch(g, OPERAND(scan)))
1075 +                                                       return(1);
1076 +                                               g->reginput = save;
1077 +                                               scan = regnext(g, scan);
1078 +                                       } while (scan != NULL && OP(scan) == BRANCH);
1079 +                                       return(0);
1080 +                                       /* NOTREACHED */
1081 +                               }
1082 +                       }
1083 +                       break;
1084 +               case STAR:
1085 +               case PLUS: {
1086 +                               register char nextch;
1087 +                               register int no;
1088 +                               register char *save;
1089 +                               register int min;
1090 +
1091 +                               /*
1092 +                                * Lookahead to avoid useless match attempts
1093 +                                * when we know what character comes next.
1094 +                                */
1095 +                               nextch = '\0';
1096 +                               if (OP(next) == EXACTLY)
1097 +                                       nextch = *OPERAND(next);
1098 +                               min = (OP(scan) == STAR) ? 0 : 1;
1099 +                               save = g->reginput;
1100 +                               no = regrepeat(g, OPERAND(scan));
1101 +                               while (no >= min) {
1102 +                                       /* If it could work, try it. */
1103 +                                       if (nextch == '\0' || *g->reginput == nextch)
1104 +                                               if (regmatch(g, next))
1105 +                                                       return(1);
1106 +                                       /* Couldn't or didn't -- back up. */
1107 +                                       no--;
1108 +                                       g->reginput = save + no;
1109 +                               }
1110 +                               return(0);
1111 +                       }
1112 +                       break;
1113 +               case END:
1114 +                       return(1);      /* Success! */
1115 +                       break;
1116 +               default:
1117 +                       printk("<3>Regexp: memory corruption\n");
1118 +                       return(0);
1119 +                       break;
1120 +               }
1121 +
1122 +               scan = next;
1123 +       }
1124 +
1125 +       /*
1126 +        * We get here only if there's trouble -- normally "case END" is
1127 +        * the terminating point.
1128 +        */
1129 +       printk("<3>Regexp: corrupted pointers\n");
1130 +       return(0);
1131 +}
1132 +
1133 +/*
1134 + - regrepeat - repeatedly match something simple, report how many
1135 + */
1136 +static int
1137 +regrepeat(struct match_globals *g, char *p)
1138 +{
1139 +       register int count = 0;
1140 +       register char *scan;
1141 +       register char *opnd;
1142 +
1143 +       scan = g->reginput;
1144 +       opnd = OPERAND(p);
1145 +       switch (OP(p)) {
1146 +       case ANY:
1147 +               count = strlen(scan);
1148 +               scan += count;
1149 +               break;
1150 +       case EXACTLY:
1151 +               while (*opnd == *scan) {
1152 +                       count++;
1153 +                       scan++;
1154 +               }
1155 +               break;
1156 +       case ANYOF:
1157 +               while (*scan != '\0' && strchr(opnd, *scan) != NULL) {
1158 +                       count++;
1159 +                       scan++;
1160 +               }
1161 +               break;
1162 +       case ANYBUT:
1163 +               while (*scan != '\0' && strchr(opnd, *scan) == NULL) {
1164 +                       count++;
1165 +                       scan++;
1166 +               }
1167 +               break;
1168 +       default:                /* Oh dear.  Called inappropriately. */
1169 +               printk("<3>Regexp: internal foulup\n");
1170 +               count = 0;      /* Best compromise. */
1171 +               break;
1172 +       }
1173 +       g->reginput = scan;
1174 +
1175 +       return(count);
1176 +}
1177 +
1178 +/*
1179 + - regnext - dig the "next" pointer out of a node
1180 + */
1181 +static char*
1182 +regnext(struct match_globals *g, char *p)
1183 +{
1184 +       register int offset;
1185 +
1186 +       if (p == &g->regdummy)
1187 +               return(NULL);
1188 +
1189 +       offset = NEXT(p);
1190 +       if (offset == 0)
1191 +               return(NULL);
1192 +
1193 +       if (OP(p) == BACK)
1194 +               return(p-offset);
1195 +       else
1196 +               return(p+offset);
1197 +}
1198 +
1199 +#ifdef DEBUG
1200 +
1201 +STATIC char *regprop();
1202 +
1203 +/*
1204 + - regdump - dump a regexp onto stdout in vaguely comprehensible form
1205 + */
1206 +void
1207 +regdump(regexp *r)
1208 +{
1209 +       register char *s;
1210 +       register char op = EXACTLY;     /* Arbitrary non-END op. */
1211 +       register char *next;
1212 +       /* extern char *strchr(); */
1213 +
1214 +
1215 +       s = r->program + 1;
1216 +       while (op != END) {     /* While that wasn't END last time... */
1217 +               op = OP(s);
1218 +               printf("%2d%s", s-r->program, regprop(s));      /* Where, what. */
1219 +               next = regnext(s);
1220 +               if (next == NULL)               /* Next ptr. */
1221 +                       printf("(0)");
1222 +               else
1223 +                       printf("(%d)", (s-r->program)+(next-s));
1224 +               s += 3;
1225 +               if (op == ANYOF || op == ANYBUT || op == EXACTLY) {
1226 +                       /* Literal string, where present. */
1227 +                       while (*s != '\0') {
1228 +                               putchar(*s);
1229 +                               s++;
1230 +                       }
1231 +                       s++;
1232 +               }
1233 +               putchar('\n');
1234 +       }
1235 +
1236 +       /* Header fields of interest. */
1237 +       if (r->regstart != '\0')
1238 +               printf("start `%c' ", r->regstart);
1239 +       if (r->reganch)
1240 +               printf("anchored ");
1241 +       if (r->regmust != NULL)
1242 +               printf("must have \"%s\"", r->regmust);
1243 +       printf("\n");
1244 +}
1245 +
1246 +/*
1247 + - regprop - printable representation of opcode
1248 + */
1249 +static char *
1250 +regprop(char *op)
1251 +{
1252 +#define BUFLEN 50
1253 +       register char *p;
1254 +       static char buf[BUFLEN];
1255 +
1256 +       strcpy(buf, ":");
1257 +
1258 +       switch (OP(op)) {
1259 +       case BOL:
1260 +               p = "BOL";
1261 +               break;
1262 +       case EOL:
1263 +               p = "EOL";
1264 +               break;
1265 +       case ANY:
1266 +               p = "ANY";
1267 +               break;
1268 +       case ANYOF:
1269 +               p = "ANYOF";
1270 +               break;
1271 +       case ANYBUT:
1272 +               p = "ANYBUT";
1273 +               break;
1274 +       case BRANCH:
1275 +               p = "BRANCH";
1276 +               break;
1277 +       case EXACTLY:
1278 +               p = "EXACTLY";
1279 +               break;
1280 +       case NOTHING:
1281 +               p = "NOTHING";
1282 +               break;
1283 +       case BACK:
1284 +               p = "BACK";
1285 +               break;
1286 +       case END:
1287 +               p = "END";
1288 +               break;
1289 +       case OPEN+1:
1290 +       case OPEN+2:
1291 +       case OPEN+3:
1292 +       case OPEN+4:
1293 +       case OPEN+5:
1294 +       case OPEN+6:
1295 +       case OPEN+7:
1296 +       case OPEN+8:
1297 +       case OPEN+9:
1298 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "OPEN%d", OP(op)-OPEN);
1299 +               p = NULL;
1300 +               break;
1301 +       case CLOSE+1:
1302 +       case CLOSE+2:
1303 +       case CLOSE+3:
1304 +       case CLOSE+4:
1305 +       case CLOSE+5:
1306 +       case CLOSE+6:
1307 +       case CLOSE+7:
1308 +       case CLOSE+8:
1309 +       case CLOSE+9:
1310 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "CLOSE%d", OP(op)-CLOSE);
1311 +               p = NULL;
1312 +               break;
1313 +       case STAR:
1314 +               p = "STAR";
1315 +               break;
1316 +       case PLUS:
1317 +               p = "PLUS";
1318 +               break;
1319 +       default:
1320 +               printk("<3>Regexp: corrupted opcode\n");
1321 +               break;
1322 +       }
1323 +       if (p != NULL)
1324 +               strncat(buf, p, BUFLEN-strlen(buf));
1325 +       return(buf);
1326 +}
1327 +#endif
1328 +
1329 +
1330 diff --git a/net/netfilter/regexp/regexp.h b/net/netfilter/regexp/regexp.h
1331 new file mode 100644
1332 index 0000000..a72eba7
1333 --- /dev/null
1334 +++ b/net/netfilter/regexp/regexp.h
1335 @@ -0,0 +1,41 @@
1336 +/*
1337 + * Definitions etc. for regexp(3) routines.
1338 + *
1339 + * Caveat:  this is V8 regexp(3) [actually, a reimplementation thereof],
1340 + * not the System V one.
1341 + */
1342 +
1343 +#ifndef REGEXP_H
1344 +#define REGEXP_H
1345 +
1346 +
1347 +/*
1348 +http://www.opensource.apple.com/darwinsource/10.3/expect-1/expect/expect.h ,
1349 +which contains a version of this library, says:
1350 +
1351 + *
1352 + * NSUBEXP must be at least 10, and no greater than 117 or the parser
1353 + * will not work properly.
1354 + *
1355 +
1356 +However, it looks rather like this library is limited to 10.  If you think
1357 +otherwise, let us know.
1358 +*/
1359 +
1360 +#define NSUBEXP  10
1361 +typedef struct regexp {
1362 +       char *startp[NSUBEXP];
1363 +       char *endp[NSUBEXP];
1364 +       char regstart;          /* Internal use only. */
1365 +       char reganch;           /* Internal use only. */
1366 +       char *regmust;          /* Internal use only. */
1367 +       int regmlen;            /* Internal use only. */
1368 +       char program[1];        /* Unwarranted chumminess with compiler. */
1369 +} regexp;
1370 +
1371 +regexp * regcomp(char *exp, int *patternsize);
1372 +int regexec(regexp *prog, char *string);
1373 +void regsub(regexp *prog, char *source, char *dest);
1374 +void regerror(char *s);
1375 +
1376 +#endif
1377 diff --git a/net/netfilter/regexp/regmagic.h b/net/netfilter/regexp/regmagic.h
1378 new file mode 100644
1379 index 0000000..5acf447
1380 --- /dev/null
1381 +++ b/net/netfilter/regexp/regmagic.h
1382 @@ -0,0 +1,5 @@
1383 +/*
1384 + * The first byte of the regexp internal "program" is actually this magic
1385 + * number; the start node begins in the second byte.
1386 + */
1387 +#define        MAGIC   0234
1388 diff --git a/net/netfilter/regexp/regsub.c b/net/netfilter/regexp/regsub.c
1389 new file mode 100644
1390 index 0000000..339631f
1391 --- /dev/null
1392 +++ b/net/netfilter/regexp/regsub.c
1393 @@ -0,0 +1,95 @@
1394 +/*
1395 + * regsub
1396 + * @(#)regsub.c        1.3 of 2 April 86
1397 + *
1398 + *     Copyright (c) 1986 by University of Toronto.
1399 + *     Written by Henry Spencer.  Not derived from licensed software.
1400 + *
1401 + *     Permission is granted to anyone to use this software for any
1402 + *     purpose on any computer system, and to redistribute it freely,
1403 + *     subject to the following restrictions:
1404 + *
1405 + *     1. The author is not responsible for the consequences of use of
1406 + *             this software, no matter how awful, even if they arise
1407 + *             from defects in it.
1408 + *
1409 + *     2. The origin of this software must not be misrepresented, either
1410 + *             by explicit claim or by omission.
1411 + *
1412 + *     3. Altered versions must be plainly marked as such, and must not
1413 + *             be misrepresented as being the original software.
1414 + *
1415 + *
1416 + * This code was modified by Ethan Sommer to work within the kernel
1417 + * (it now uses kmalloc etc..)
1418 + *
1419 + */
1420 +#include "regexp.h"
1421 +#include "regmagic.h"
1422 +#include <linux/string.h>
1423 +
1424 +
1425 +#ifndef CHARBITS
1426 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
1427 +#else
1428 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
1429 +#endif
1430 +
1431 +#if 0
1432 +//void regerror(char * s)
1433 +//{
1434 +//        printk("regexp(3): %s", s);
1435 +//        /* NOTREACHED */
1436 +//}
1437 +#endif
1438 +
1439 +/*
1440 + - regsub - perform substitutions after a regexp match
1441 + */
1442 +void
1443 +regsub(regexp * prog, char * source, char * dest)
1444 +{
1445 +       register char *src;
1446 +       register char *dst;
1447 +       register char c;
1448 +       register int no;
1449 +       register int len;
1450 +
1451 +       /* Not necessary and gcc doesn't like it -MLS */
1452 +       /*extern char *strncpy();*/
1453 +
1454 +       if (prog == NULL || source == NULL || dest == NULL) {
1455 +               regerror("NULL parm to regsub");
1456 +               return;
1457 +       }
1458 +       if (UCHARAT(prog->program) != MAGIC) {
1459 +               regerror("damaged regexp fed to regsub");
1460 +               return;
1461 +       }
1462 +
1463 +       src = source;
1464 +       dst = dest;
1465 +       while ((c = *src++) != '\0') {
1466 +               if (c == '&')
1467 +                       no = 0;
1468 +               else if (c == '\\' && '0' <= *src && *src <= '9')
1469 +                       no = *src++ - '0';
1470 +               else
1471 +                       no = -1;
1472 +
1473 +               if (no < 0) {   /* Ordinary character. */
1474 +                       if (c == '\\' && (*src == '\\' || *src == '&'))
1475 +                               c = *src++;
1476 +                       *dst++ = c;
1477 +               } else if (prog->startp[no] != NULL && prog->endp[no] != NULL) {
1478 +                       len = prog->endp[no] - prog->startp[no];
1479 +                       (void) strncpy(dst, prog->startp[no], len);
1480 +                       dst += len;
1481 +                       if (len != 0 && *(dst-1) == '\0') {     /* strncpy hit NUL. */
1482 +                               regerror("damaged match string");
1483 +                               return;
1484 +                       }
1485 +               }
1486 +       }
1487 +       *dst++ = '\0';
1488 +}
1489 diff --git a/net/netfilter/xt_layer7.c b/net/netfilter/xt_layer7.c
1490 new file mode 100644
1491 index 0000000..a9b88d3
1492 --- /dev/null
1493 +++ b/net/netfilter/xt_layer7.c
1494 @@ -0,0 +1,651 @@
1495 +/*
1496 +  Kernel module to match application layer (OSI layer 7) data in connections.
1497 +
1498 +  http://l7-filter.sf.net
1499 +
1500 +  (C) 2003, 2004, 2005, 2006, 2007 Matthew Strait and Ethan Sommer.
1501 +
1502 +  This program is free software; you can redistribute it and/or
1503 +  modify it under the terms of the GNU General Public License
1504 +  as published by the Free Software Foundation; either version
1505 +  2 of the License, or (at your option) any later version.
1506 +  http://www.gnu.org/licenses/gpl.txt
1507 +
1508 +  Based on ipt_string.c (C) 2000 Emmanuel Roger <winfield@freegates.be>,
1509 +  xt_helper.c (C) 2002 Harald Welte and cls_layer7.c (C) 2003 Matthew Strait,
1510 +  Ethan Sommer, Justin Levandoski.
1511 +*/
1512 +
1513 +#include <linux/spinlock.h>
1514 +#include <linux/version.h>
1515 +#include <net/ip.h>
1516 +#include <net/tcp.h>
1517 +#include <linux/module.h>
1518 +#include <linux/skbuff.h>
1519 +#include <linux/netfilter.h>
1520 +#include <net/netfilter/nf_conntrack.h>
1521 +#include <net/netfilter/nf_conntrack_core.h>
1522 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 27)
1523 +#include <net/netfilter/nf_conntrack_extend.h>
1524 +#include <net/netfilter/nf_conntrack_acct.h>
1525 +#endif
1526 +#include <linux/netfilter/x_tables.h>
1527 +#include <linux/netfilter/xt_layer7.h>
1528 +#include <linux/ctype.h>
1529 +#include <linux/proc_fs.h>
1530 +
1531 +#include "regexp/regexp.c"
1532 +
1533 +MODULE_LICENSE("GPL");
1534 +MODULE_AUTHOR("Matthew Strait <quadong@users.sf.net>, Ethan Sommer <sommere@users.sf.net>");
1535 +MODULE_DESCRIPTION("iptables application layer match module");
1536 +MODULE_ALIAS("ipt_layer7");
1537 +MODULE_VERSION("2.19");
1538 +
1539 +static int maxdatalen = 2048; // this is the default
1540 +module_param(maxdatalen, int, 0444);
1541 +MODULE_PARM_DESC(maxdatalen, "maximum bytes of data looked at by l7-filter");
1542 +#ifdef CONFIG_NETFILTER_XT_MATCH_LAYER7_DEBUG
1543 +       #define DPRINTK(format,args...) printk(format,##args)
1544 +#else
1545 +       #define DPRINTK(format,args...)
1546 +#endif
1547 +
1548 +/* Number of packets whose data we look at.
1549 +This can be modified through /proc/net/layer7_numpackets */
1550 +static int num_packets = 10;
1551 +
1552 +static struct pattern_cache {
1553 +       char * regex_string;
1554 +       regexp * pattern;
1555 +       struct pattern_cache * next;
1556 +} * first_pattern_cache = NULL;
1557 +
1558 +DEFINE_SPINLOCK(l7_lock);
1559 +
1560 +static int total_acct_packets(struct nf_conn *ct)
1561 +{
1562 +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2, 6, 26)
1563 +       BUG_ON(ct == NULL);
1564 +       return (ct->counters[IP_CT_DIR_ORIGINAL].packets + ct->counters[IP_CT_DIR_REPLY].packets);
1565 +#else
1566 +       struct nf_conn_counter *acct;
1567 +
1568 +       BUG_ON(ct == NULL);
1569 +       acct = nf_conn_acct_find(ct);
1570 +       if (!acct)
1571 +               return 0;
1572 +       return (acct[IP_CT_DIR_ORIGINAL].packets + acct[IP_CT_DIR_REPLY].packets);
1573 +#endif
1574 +}
1575 +
1576 +#ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
1577 +/* Converts an unfriendly string into a friendly one by
1578 +replacing unprintables with periods and all whitespace with " ". */
1579 +static char * friendly_print(unsigned char * s)
1580 +{
1581 +       char * f = kmalloc(strlen(s) + 1, GFP_ATOMIC);
1582 +       int i;
1583 +
1584 +       if(!f) {
1585 +               if (net_ratelimit())
1586 +                       printk(KERN_ERR "layer7: out of memory in "
1587 +                                       "friendly_print, bailing.\n");
1588 +               return NULL;
1589 +       }
1590 +
1591 +       for(i = 0; i < strlen(s); i++){
1592 +               if(isprint(s[i]) && s[i] < 128) f[i] = s[i];
1593 +               else if(isspace(s[i]))          f[i] = ' ';
1594 +               else                            f[i] = '.';
1595 +       }
1596 +       f[i] = '\0';
1597 +       return f;
1598 +}
1599 +
1600 +static char dec2hex(int i)
1601 +{
1602 +       switch (i) {
1603 +               case 0 ... 9:
1604 +                       return (i + '0');
1605 +                       break;
1606 +               case 10 ... 15:
1607 +                       return (i - 10 + 'a');
1608 +                       break;
1609 +               default:
1610 +                       if (net_ratelimit())
1611 +                               printk("layer7: Problem in dec2hex\n");
1612 +                       return '\0';
1613 +       }
1614 +}
1615 +
1616 +static char * hex_print(unsigned char * s)
1617 +{
1618 +       char * g = kmalloc(strlen(s)*3 + 1, GFP_ATOMIC);
1619 +       int i;
1620 +
1621 +       if(!g) {
1622 +              if (net_ratelimit())
1623 +                       printk(KERN_ERR "layer7: out of memory in hex_print, "
1624 +                                       "bailing.\n");
1625 +              return NULL;
1626 +       }
1627 +
1628 +       for(i = 0; i < strlen(s); i++) {
1629 +               g[i*3    ] = dec2hex(s[i]/16);
1630 +               g[i*3 + 1] = dec2hex(s[i]%16);
1631 +               g[i*3 + 2] = ' ';
1632 +       }
1633 +       g[i*3] = '\0';
1634 +
1635 +       return g;
1636 +}
1637 +#endif // DEBUG
1638 +
1639 +/* Use instead of regcomp.  As we expect to be seeing the same regexps over and
1640 +over again, it make sense to cache the results. */
1641 +static regexp * compile_and_cache(const char * regex_string,
1642 +                                  const char * protocol)
1643 +{
1644 +       struct pattern_cache * node               = first_pattern_cache;
1645 +       struct pattern_cache * last_pattern_cache = first_pattern_cache;
1646 +       struct pattern_cache * tmp;
1647 +       unsigned int len;
1648 +
1649 +       while (node != NULL) {
1650 +               if (!strcmp(node->regex_string, regex_string))
1651 +               return node->pattern;
1652 +
1653 +               last_pattern_cache = node;/* points at the last non-NULL node */
1654 +               node = node->next;
1655 +       }
1656 +
1657 +       /* If we reach the end of the list, then we have not yet cached
1658 +          the pattern for this regex. Let's do that now.
1659 +          Be paranoid about running out of memory to avoid list corruption. */
1660 +       tmp = kmalloc(sizeof(struct pattern_cache), GFP_ATOMIC);
1661 +
1662 +       if(!tmp) {
1663 +               if (net_ratelimit())
1664 +                       printk(KERN_ERR "layer7: out of memory in "
1665 +                                       "compile_and_cache, bailing.\n");
1666 +               return NULL;
1667 +       }
1668 +
1669 +       tmp->regex_string  = kmalloc(strlen(regex_string) + 1, GFP_ATOMIC);
1670 +       tmp->pattern       = kmalloc(sizeof(struct regexp),    GFP_ATOMIC);
1671 +       tmp->next = NULL;
1672 +
1673 +       if(!tmp->regex_string || !tmp->pattern) {
1674 +               if (net_ratelimit())
1675 +                       printk(KERN_ERR "layer7: out of memory in "
1676 +                                       "compile_and_cache, bailing.\n");
1677 +               kfree(tmp->regex_string);
1678 +               kfree(tmp->pattern);
1679 +               kfree(tmp);
1680 +               return NULL;
1681 +       }
1682 +
1683 +       /* Ok.  The new node is all ready now. */
1684 +       node = tmp;
1685 +
1686 +       if(first_pattern_cache == NULL) /* list is empty */
1687 +               first_pattern_cache = node; /* make node the beginning */
1688 +       else
1689 +               last_pattern_cache->next = node; /* attach node to the end */
1690 +
1691 +       /* copy the string and compile the regex */
1692 +       len = strlen(regex_string);
1693 +       DPRINTK("About to compile this: \"%s\"\n", regex_string);
1694 +       node->pattern = regcomp((char *)regex_string, &len);
1695 +       if ( !node->pattern ) {
1696 +               if (net_ratelimit())
1697 +                       printk(KERN_ERR "layer7: Error compiling regexp "
1698 +                                       "\"%s\" (%s)\n",
1699 +                                       regex_string, protocol);
1700 +               /* pattern is now cached as NULL, so we won't try again. */
1701 +       }
1702 +
1703 +       strcpy(node->regex_string, regex_string);
1704 +       return node->pattern;
1705 +}
1706 +
1707 +static int can_handle(const struct sk_buff *skb)
1708 +{
1709 +       if(!ip_hdr(skb)) /* not IP */
1710 +               return 0;
1711 +       if(ip_hdr(skb)->protocol != IPPROTO_TCP &&
1712 +          ip_hdr(skb)->protocol != IPPROTO_UDP &&
1713 +          ip_hdr(skb)->protocol != IPPROTO_ICMP)
1714 +               return 0;
1715 +       return 1;
1716 +}
1717 +
1718 +/* Returns offset the into the skb->data that the application data starts */
1719 +static int app_data_offset(const struct sk_buff *skb)
1720 +{
1721 +       /* In case we are ported somewhere (ebtables?) where ip_hdr(skb)
1722 +       isn't set, this can be gotten from 4*(skb->data[0] & 0x0f) as well. */
1723 +       int ip_hl = 4*ip_hdr(skb)->ihl;
1724 +
1725 +       if( ip_hdr(skb)->protocol == IPPROTO_TCP ) {
1726 +               /* 12 == offset into TCP header for the header length field.
1727 +               Can't get this with skb->h.th->doff because the tcphdr
1728 +               struct doesn't get set when routing (this is confirmed to be
1729 +               true in Netfilter as well as QoS.) */
1730 +               int tcp_hl = 4*(skb->data[ip_hl + 12] >> 4);
1731 +
1732 +               return ip_hl + tcp_hl;
1733 +       } else if( ip_hdr(skb)->protocol == IPPROTO_UDP  ) {
1734 +               return ip_hl + 8; /* UDP header is always 8 bytes */
1735 +       } else if( ip_hdr(skb)->protocol == IPPROTO_ICMP ) {
1736 +               return ip_hl + 8; /* ICMP header is 8 bytes */
1737 +       } else {
1738 +               if (net_ratelimit())
1739 +                       printk(KERN_ERR "layer7: tried to handle unknown "
1740 +                                       "protocol!\n");
1741 +               return ip_hl + 8; /* something reasonable */
1742 +       }
1743 +}
1744 +
1745 +/* handles whether there's a match when we aren't appending data anymore */
1746 +static int match_no_append(struct nf_conn * conntrack,
1747 +                           struct nf_conn * master_conntrack,
1748 +                           enum ip_conntrack_info ctinfo,
1749 +                           enum ip_conntrack_info master_ctinfo,
1750 +                           const struct xt_layer7_info * info)
1751 +{
1752 +       /* If we're in here, throw the app data away */
1753 +       if(master_conntrack->layer7.app_data != NULL) {
1754 +
1755 +       #ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
1756 +               if(!master_conntrack->layer7.app_proto) {
1757 +                       char * f =
1758 +                         friendly_print(master_conntrack->layer7.app_data);
1759 +                       char * g =
1760 +                         hex_print(master_conntrack->layer7.app_data);
1761 +                       DPRINTK("\nl7-filter gave up after %d bytes "
1762 +                               "(%d packets):\n%s\n",
1763 +                               strlen(f), total_acct_packets(master_conntrack), f);
1764 +                       kfree(f);
1765 +                       DPRINTK("In hex: %s\n", g);
1766 +                       kfree(g);
1767 +               }
1768 +       #endif
1769 +
1770 +               kfree(master_conntrack->layer7.app_data);
1771 +               master_conntrack->layer7.app_data = NULL; /* don't free again */
1772 +       }
1773 +
1774 +       if(master_conntrack->layer7.app_proto){
1775 +               /* Here child connections set their .app_proto (for /proc) */
1776 +               if(!conntrack->layer7.app_proto) {
1777 +                       conntrack->layer7.app_proto =
1778 +                         kmalloc(strlen(master_conntrack->layer7.app_proto)+1,
1779 +                           GFP_ATOMIC);
1780 +                       if(!conntrack->layer7.app_proto){
1781 +                               if (net_ratelimit())
1782 +                                       printk(KERN_ERR "layer7: out of memory "
1783 +                                                       "in match_no_append, "
1784 +                                                       "bailing.\n");
1785 +                               return 1;
1786 +                       }
1787 +                       strcpy(conntrack->layer7.app_proto,
1788 +                               master_conntrack->layer7.app_proto);
1789 +               }
1790 +
1791 +               return (!strcmp(master_conntrack->layer7.app_proto,
1792 +                               info->protocol));
1793 +       }
1794 +       else {
1795 +               /* If not classified, set to "unknown" to distinguish from
1796 +               connections that are still being tested. */
1797 +               master_conntrack->layer7.app_proto =
1798 +                       kmalloc(strlen("unknown")+1, GFP_ATOMIC);
1799 +               if(!master_conntrack->layer7.app_proto){
1800 +                       if (net_ratelimit())
1801 +                               printk(KERN_ERR "layer7: out of memory in "
1802 +                                               "match_no_append, bailing.\n");
1803 +                       return 1;
1804 +               }
1805 +               strcpy(master_conntrack->layer7.app_proto, "unknown");
1806 +               return 0;
1807 +       }
1808 +}
1809 +
1810 +/* add the new app data to the conntrack.  Return number of bytes added. */
1811 +static int add_data(struct nf_conn * master_conntrack,
1812 +                    char * app_data, int appdatalen)
1813 +{
1814 +       int length = 0, i;
1815 +       int oldlength = master_conntrack->layer7.app_data_len;
1816 +
1817 +       /* This is a fix for a race condition by Deti Fliegl. However, I'm not
1818 +          clear on whether the race condition exists or whether this really
1819 +          fixes it.  I might just be being dense... Anyway, if it's not really
1820 +          a fix, all it does is waste a very small amount of time. */
1821 +       if(!master_conntrack->layer7.app_data) return 0;
1822 +
1823 +       /* Strip nulls. Make everything lower case (our regex lib doesn't
1824 +       do case insensitivity).  Add it to the end of the current data. */
1825 +       for(i = 0; i < maxdatalen-oldlength-1 &&
1826 +                  i < appdatalen; i++) {
1827 +               if(app_data[i] != '\0') {
1828 +                       /* the kernel version of tolower mungs 'upper ascii' */
1829 +                       master_conntrack->layer7.app_data[length+oldlength] =
1830 +                               isascii(app_data[i])?
1831 +                                       tolower(app_data[i]) : app_data[i];
1832 +                       length++;
1833 +               }
1834 +       }
1835 +
1836 +       master_conntrack->layer7.app_data[length+oldlength] = '\0';
1837 +       master_conntrack->layer7.app_data_len = length + oldlength;
1838 +
1839 +       return length;
1840 +}
1841 +
1842 +/* taken from drivers/video/modedb.c */
1843 +static int my_atoi(const char *s)
1844 +{
1845 +       int val = 0;
1846 +
1847 +       for (;; s++) {
1848 +               switch (*s) {
1849 +                       case '0'...'9':
1850 +                       val = 10*val+(*s-'0');
1851 +                       break;
1852 +               default:
1853 +                       return val;
1854 +               }
1855 +       }
1856 +}
1857 +
1858 +/* write out num_packets to userland. */
1859 +static int layer7_read_proc(char* page, char ** start, off_t off, int count,
1860 +                            int* eof, void * data)
1861 +{
1862 +       if(num_packets > 99 && net_ratelimit())
1863 +               printk(KERN_ERR "layer7: NOT REACHED. num_packets too big\n");
1864 +
1865 +       page[0] = num_packets/10 + '0';
1866 +       page[1] = num_packets%10 + '0';
1867 +       page[2] = '\n';
1868 +       page[3] = '\0';
1869 +
1870 +       *eof=1;
1871 +
1872 +       return 3;
1873 +}
1874 +
1875 +/* Read in num_packets from userland */
1876 +static int layer7_write_proc(struct file* file, const char* buffer,
1877 +                             unsigned long count, void *data)
1878 +{
1879 +       char * foo = kmalloc(count, GFP_ATOMIC);
1880 +
1881 +       if(!foo){
1882 +               if (net_ratelimit())
1883 +                       printk(KERN_ERR "layer7: out of memory, bailing. "
1884 +                                       "num_packets unchanged.\n");
1885 +               return count;
1886 +       }
1887 +
1888 +       if(copy_from_user(foo, buffer, count)) {
1889 +               return -EFAULT;
1890 +       }
1891 +
1892 +
1893 +       num_packets = my_atoi(foo);
1894 +       kfree (foo);
1895 +
1896 +       /* This has an arbitrary limit to make the math easier. I'm lazy.
1897 +       But anyway, 99 is a LOT! If you want more, you're doing it wrong! */
1898 +       if(num_packets > 99) {
1899 +               printk(KERN_WARNING "layer7: num_packets can't be > 99.\n");
1900 +               num_packets = 99;
1901 +       } else if(num_packets < 1) {
1902 +               printk(KERN_WARNING "layer7: num_packets can't be < 1.\n");
1903 +               num_packets = 1;
1904 +       }
1905 +
1906 +       return count;
1907 +}
1908 +
1909 +static bool
1910 +match(const struct sk_buff *skbin,
1911 +      const struct net_device *in,
1912 +      const struct net_device *out,
1913 +      const struct xt_match *match,
1914 +      const void *matchinfo,
1915 +      int offset,
1916 +      unsigned int protoff,
1917 +      bool *hotdrop)
1918 +{
1919 +       /* sidestep const without getting a compiler warning... */
1920 +       struct sk_buff * skb = (struct sk_buff *)skbin;
1921 +
1922 +       const struct xt_layer7_info * info = matchinfo;
1923 +       enum ip_conntrack_info master_ctinfo, ctinfo;
1924 +       struct nf_conn *master_conntrack, *conntrack;
1925 +       unsigned char * app_data;
1926 +       unsigned int pattern_result, appdatalen;
1927 +       regexp * comppattern;
1928 +
1929 +       /* Be paranoid/incompetent - lock the entire match function. */
1930 +       spin_lock_bh(&l7_lock);
1931 +
1932 +       if(!can_handle(skb)){
1933 +               DPRINTK("layer7: This is some protocol I can't handle.\n");
1934 +               spin_unlock_bh(&l7_lock);
1935 +               return info->invert;
1936 +       }
1937 +
1938 +       /* Treat parent & all its children together as one connection, except
1939 +       for the purpose of setting conntrack->layer7.app_proto in the actual
1940 +       connection. This makes /proc/net/ip_conntrack more satisfying. */
1941 +       if(!(conntrack = nf_ct_get(skb, &ctinfo)) ||
1942 +          !(master_conntrack=nf_ct_get(skb,&master_ctinfo))){
1943 +               DPRINTK("layer7: couldn't get conntrack.\n");
1944 +               spin_unlock_bh(&l7_lock);
1945 +               return info->invert;
1946 +       }
1947 +
1948 +       /* Try to get a master conntrack (and its master etc) for FTP, etc. */
1949 +       while (master_ct(master_conntrack) != NULL)
1950 +               master_conntrack = master_ct(master_conntrack);
1951 +
1952 +       /* if we've classified it or seen too many packets */
1953 +       if(total_acct_packets(master_conntrack) > num_packets ||
1954 +          master_conntrack->layer7.app_proto) {
1955 +
1956 +               pattern_result = match_no_append(conntrack, master_conntrack,
1957 +                                                ctinfo, master_ctinfo, info);
1958 +
1959 +               /* skb->cb[0] == seen. Don't do things twice if there are
1960 +               multiple l7 rules. I'm not sure that using cb for this purpose
1961 +               is correct, even though it says "put your private variables
1962 +               there". But it doesn't look like it is being used for anything
1963 +               else in the skbs that make it here. */
1964 +               skb->cb[0] = 1; /* marking it seen here's probably irrelevant */
1965 +
1966 +               spin_unlock_bh(&l7_lock);
1967 +               return (pattern_result ^ info->invert);
1968 +       }
1969 +
1970 +       if(skb_is_nonlinear(skb)){
1971 +               if(skb_linearize(skb) != 0){
1972 +                       if (net_ratelimit())
1973 +                               printk(KERN_ERR "layer7: failed to linearize "
1974 +                                               "packet, bailing.\n");
1975 +                       spin_unlock_bh(&l7_lock);
1976 +                       return info->invert;
1977 +               }
1978 +       }
1979 +
1980 +       /* now that the skb is linearized, it's safe to set these. */
1981 +       app_data = skb->data + app_data_offset(skb);
1982 +       appdatalen = skb_tail_pointer(skb) - app_data;
1983 +
1984 +       /* the return value gets checked later, when we're ready to use it */
1985 +       comppattern = compile_and_cache(info->pattern, info->protocol);
1986 +
1987 +       /* On the first packet of a connection, allocate space for app data */
1988 +       if(total_acct_packets(master_conntrack) == 1 && !skb->cb[0] &&
1989 +          !master_conntrack->layer7.app_data){
1990 +               master_conntrack->layer7.app_data =
1991 +                       kmalloc(maxdatalen, GFP_ATOMIC);
1992 +               if(!master_conntrack->layer7.app_data){
1993 +                       if (net_ratelimit())
1994 +                               printk(KERN_ERR "layer7: out of memory in "
1995 +                                               "match, bailing.\n");
1996 +                       spin_unlock_bh(&l7_lock);
1997 +                       return info->invert;
1998 +               }
1999 +
2000 +               master_conntrack->layer7.app_data[0] = '\0';
2001 +       }
2002 +
2003 +       /* Can be here, but unallocated, if numpackets is increased near
2004 +       the beginning of a connection */
2005 +       if(master_conntrack->layer7.app_data == NULL){
2006 +               spin_unlock_bh(&l7_lock);
2007 +               return (info->invert); /* unmatched */
2008 +       }
2009 +
2010 +       if(!skb->cb[0]){
2011 +               int newbytes;
2012 +               newbytes = add_data(master_conntrack, app_data, appdatalen);
2013 +
2014 +               if(newbytes == 0) { /* didn't add any data */
2015 +                       skb->cb[0] = 1;
2016 +                       /* Didn't match before, not going to match now */
2017 +                       spin_unlock_bh(&l7_lock);
2018 +                       return info->invert;
2019 +               }
2020 +       }
2021 +
2022 +       /* If looking for "unknown", then never match.  "Unknown" means that
2023 +       we've given up; we're still trying with these packets. */
2024 +       if(!strcmp(info->protocol, "unknown")) {
2025 +               pattern_result = 0;
2026 +       /* If looking for "unset", then always match. "Unset" means that we
2027 +       haven't yet classified the connection. */
2028 +       } else if(!strcmp(info->protocol, "unset")) {
2029 +               pattern_result = 2;
2030 +               DPRINTK("layer7: matched unset: not yet classified "
2031 +                       "(%d/%d packets)\n", total_acct_packets(master_conntrack), num_packets);
2032 +       /* If the regexp failed to compile, don't bother running it */
2033 +       } else if(comppattern &&
2034 +                 regexec(comppattern, master_conntrack->layer7.app_data)){
2035 +               DPRINTK("layer7: matched %s\n", info->protocol);
2036 +               pattern_result = 1;
2037 +       } else pattern_result = 0;
2038 +
2039 +       if(pattern_result == 1) {
2040 +               master_conntrack->layer7.app_proto =
2041 +                       kmalloc(strlen(info->protocol)+1, GFP_ATOMIC);
2042 +               if(!master_conntrack->layer7.app_proto){
2043 +                       if (net_ratelimit())
2044 +                               printk(KERN_ERR "layer7: out of memory in "
2045 +                                               "match, bailing.\n");
2046 +                       spin_unlock_bh(&l7_lock);
2047 +                       return (pattern_result ^ info->invert);
2048 +               }
2049 +               strcpy(master_conntrack->layer7.app_proto, info->protocol);
2050 +       } else if(pattern_result > 1) { /* cleanup from "unset" */
2051 +               pattern_result = 1;
2052 +       }
2053 +
2054 +       /* mark the packet seen */
2055 +       skb->cb[0] = 1;
2056 +
2057 +       spin_unlock_bh(&l7_lock);
2058 +       return (pattern_result ^ info->invert);
2059 +}
2060 +
2061 +static bool check(const char *tablename,
2062 +                const void *inf,
2063 +                const struct xt_match *match,
2064 +                void *matchinfo,
2065 +                unsigned int hook_mask)
2066 +
2067 +{
2068 +       // load nf_conntrack_ipv4
2069 +        if (nf_ct_l3proto_try_module_get(match->family) < 0) {
2070 +                printk(KERN_WARNING "can't load conntrack support for "
2071 +                                    "proto=%d\n", match->family);
2072 +                return 0;
2073 +        }
2074 +       return 1;
2075 +}
2076 +
2077 +static void
2078 +destroy(const struct xt_match *match, void *matchinfo)
2079 +{
2080 +       nf_ct_l3proto_module_put(match->family);
2081 +}
2082 +
2083 +static struct xt_match xt_layer7_match[] __read_mostly = {
2084 +{
2085 +       .name           = "layer7",
2086 +       .family         = AF_INET,
2087 +       .checkentry     = check,
2088 +       .match          = match,
2089 +       .destroy        = destroy,
2090 +       .matchsize      = sizeof(struct xt_layer7_info),
2091 +       .me             = THIS_MODULE
2092 +}
2093 +};
2094 +
2095 +static void layer7_cleanup_proc(void)
2096 +{
2097 +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23)
2098 +       remove_proc_entry("layer7_numpackets", proc_net);
2099 +#else
2100 +       remove_proc_entry("layer7_numpackets", init_net.proc_net);
2101 +#endif
2102 +}
2103 +
2104 +/* register the proc file */
2105 +static void layer7_init_proc(void)
2106 +{
2107 +       struct proc_dir_entry* entry;
2108 +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,23)
2109 +       entry = create_proc_entry("layer7_numpackets", 0644, proc_net);
2110 +#else
2111 +       entry = create_proc_entry("layer7_numpackets", 0644, init_net.proc_net);
2112 +#endif
2113 +       entry->read_proc = layer7_read_proc;
2114 +       entry->write_proc = layer7_write_proc;
2115 +}
2116 +
2117 +static int __init xt_layer7_init(void)
2118 +{
2119 +       need_conntrack();
2120 +
2121 +       layer7_init_proc();
2122 +       if(maxdatalen < 1) {
2123 +               printk(KERN_WARNING "layer7: maxdatalen can't be < 1, "
2124 +                       "using 1\n");
2125 +               maxdatalen = 1;
2126 +       }
2127 +       /* This is not a hard limit.  It's just here to prevent people from
2128 +       bringing their slow machines to a grinding halt. */
2129 +       else if(maxdatalen > 65536) {
2130 +               printk(KERN_WARNING "layer7: maxdatalen can't be > 65536, "
2131 +                       "using 65536\n");
2132 +               maxdatalen = 65536;
2133 +       }
2134 +       return xt_register_matches(xt_layer7_match,
2135 +                                  ARRAY_SIZE(xt_layer7_match));
2136 +}
2137 +
2138 +static void __exit xt_layer7_fini(void)
2139 +{
2140 +       layer7_cleanup_proc();
2141 +       xt_unregister_matches(xt_layer7_match, ARRAY_SIZE(xt_layer7_match));
2142 +}
2143 +
2144 +module_init(xt_layer7_init);
2145 +module_exit(xt_layer7_fini);