kernel-layer7.patch

   1 diff -urNp -x '*.orig' linux-5.11/include/linux/netfilter/xt_layer7.h linux-5.11/include/linux/netfilter/xt_layer7.h
   2 --- linux-5.11/include/linux/netfilter/xt_layer7.h      1970-01-01 01:00:00.000000000 +0100
   3 +++ linux-5.11/include/linux/netfilter/xt_layer7.h      2021-02-24 13:21:48.338440845 +0100
   4 @@ -0,0 +1,13 @@
   5 +#ifndef _XT_LAYER7_H
   6 +#define _XT_LAYER7_H
   7 +
   8 +#define MAX_PATTERN_LEN 8192
   9 +#define MAX_PROTOCOL_LEN 256
  10 +
  11 +struct xt_layer7_info {
  12 +    char protocol[MAX_PROTOCOL_LEN];
  13 +    char pattern[MAX_PATTERN_LEN];
  14 +    u_int8_t invert;
  15 +};
  16 +
  17 +#endif /* _XT_LAYER7_H */
  18 diff -urNp -x '*.orig' linux-5.11/include/net/netfilter/nf_conntrack.h linux-5.11/include/net/netfilter/nf_conntrack.h
  19 --- linux-5.11/include/net/netfilter/nf_conntrack.h     2021-02-14 23:32:24.000000000 +0100
  20 +++ linux-5.11/include/net/netfilter/nf_conntrack.h     2021-02-24 13:21:48.338440845 +0100
  21 @@ -103,6 +103,22 @@ struct nf_conn {
  22         /* Extensions */
  23         struct nf_ct_ext *ext;
  24
  25 +#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || \
  26 +       defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
  27 +       struct {
  28 +               /*
  29 +                * e.g. "http". NULL before decision. "unknown" after decision
  30 +                * if no match.
  31 +                */
  32 +               char *app_proto;
  33 +               /*
  34 +                * application layer data so far. NULL after match decision.
  35 +                */
  36 +               char *app_data;
  37 +               unsigned int app_data_len;
  38 +       } layer7;
  39 +#endif
  40 +
  41         /* Storage reserved for other modules, must be the last member */
  42         union nf_conntrack_proto proto;
  43  };
  44 diff -urNp -x '*.orig' linux-5.11/net/netfilter/Kconfig linux-5.11/net/netfilter/Kconfig
  45 --- linux-5.11/net/netfilter/Kconfig    2021-02-14 23:32:24.000000000 +0100
  46 +++ linux-5.11/net/netfilter/Kconfig    2021-02-24 13:21:48.335107407 +0100
  47 @@ -1562,6 +1562,27 @@ config NETFILTER_XT_MATCH_STATE
  48
  49           To compile it as a module, choose M here.  If unsure, say N.
  50
  51 +config NETFILTER_XT_MATCH_LAYER7
  52 +       tristate '"layer7" match support'
  53 +       depends on NETFILTER_XTABLES
  54 +       depends on EXPERIMENTAL && (IP_NF_CONNTRACK || NF_CONNTRACK)
  55 +       help
  56 +         Say Y if you want to be able to classify connections (and their
  57 +         packets) based on regular expression matching of their application
  58 +         layer data.   This is one way to classify applications such as
  59 +         peer-to-peer filesharing systems that do not always use the same
  60 +         port.
  61 +
  62 +         To compile it as a module, choose M here.  If unsure, say N.
  63 +
  64 +
  65 +config NETFILTER_XT_MATCH_LAYER7_DEBUG
  66 +        bool 'Layer 7 debugging output'
  67 +        depends on NETFILTER_XT_MATCH_LAYER7
  68 +        help
  69 +          Say Y to get lots of debugging output.
  70 +
  71 +
  72  config NETFILTER_XT_MATCH_STATISTIC
  73         tristate '"statistic" match support'
  74         depends on NETFILTER_ADVANCED
  75 diff -urNp -x '*.orig' linux-5.11/net/netfilter/Makefile linux-5.11/net/netfilter/Makefile
  76 --- linux-5.11/net/netfilter/Makefile   2021-02-14 23:32:24.000000000 +0100
  77 +++ linux-5.11/net/netfilter/Makefile   2021-02-24 13:21:48.335107407 +0100
  78 @@ -204,6 +204,7 @@ obj-$(CONFIG_NETFILTER_XT_MATCH_RECENT)
  79  obj-$(CONFIG_NETFILTER_XT_MATCH_SCTP) += xt_sctp.o
  80  obj-$(CONFIG_NETFILTER_XT_MATCH_SOCKET) += xt_socket.o
  81  obj-$(CONFIG_NETFILTER_XT_MATCH_STATE) += xt_state.o
  82 +obj-$(CONFIG_NETFILTER_XT_MATCH_LAYER7) += xt_layer7.o
  83  obj-$(CONFIG_NETFILTER_XT_MATCH_STATISTIC) += xt_statistic.o
  84  obj-$(CONFIG_NETFILTER_XT_MATCH_STRING) += xt_string.o
  85  obj-$(CONFIG_NETFILTER_XT_MATCH_TCPMSS) += xt_tcpmss.o
  86 diff -urNp -x '*.orig' linux-5.11/net/netfilter/nf_conntrack_standalone.c linux-5.11/net/netfilter/nf_conntrack_standalone.c
  87 --- linux-5.11/net/netfilter/nf_conntrack_standalone.c  2021-02-14 23:32:24.000000000 +0100
  88 +++ linux-5.11/net/netfilter/nf_conntrack_standalone.c  2021-02-24 13:21:48.338440845 +0100
  89 @@ -366,6 +366,12 @@ static int ct_seq_show(struct seq_file *
  90         ct_show_zone(s, ct, NF_CT_DEFAULT_ZONE_DIR);
  91         ct_show_delta_time(s, ct);
  92
  93 +#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
  94 +       if(ct->layer7.app_proto &&
  95 +           seq_printf(s, "l7proto=%s ", ct->layer7.app_proto))
  96 +               return -ENOSPC;
  97 +#endif
  98 +
  99         seq_printf(s, "use=%u\n", refcount_read(&ct->ct_general.use));
 100
 101         if (seq_has_overflowed(s))
 102 diff -urNp -x '*.orig' linux-5.11/net/netfilter/regexp/regexp.c linux-5.11/net/netfilter/regexp/regexp.c
 103 --- linux-5.11/net/netfilter/regexp/regexp.c    1970-01-01 01:00:00.000000000 +0100
 104 +++ linux-5.11/net/netfilter/regexp/regexp.c    2021-02-24 13:21:48.335107407 +0100
 105 @@ -0,0 +1,1197 @@
 106 +/*
 107 + * regcomp and regexec -- regsub and regerror are elsewhere
 108 + * @(#)regexp.c        1.3 of 18 April 87
 109 + *
 110 + *     Copyright (c) 1986 by University of Toronto.
 111 + *     Written by Henry Spencer.  Not derived from licensed software.
 112 + *
 113 + *     Permission is granted to anyone to use this software for any
 114 + *     purpose on any computer system, and to redistribute it freely,
 115 + *     subject to the following restrictions:
 116 + *
 117 + *     1. The author is not responsible for the consequences of use of
 118 + *             this software, no matter how awful, even if they arise
 119 + *             from defects in it.
 120 + *
 121 + *     2. The origin of this software must not be misrepresented, either
 122 + *             by explicit claim or by omission.
 123 + *
 124 + *     3. Altered versions must be plainly marked as such, and must not
 125 + *             be misrepresented as being the original software.
 126 + *
 127 + * Beware that some of this code is subtly aware of the way operator
 128 + * precedence is structured in regular expressions.  Serious changes in
 129 + * regular-expression syntax might require a total rethink.
 130 + *
 131 + * This code was modified by Ethan Sommer to work within the kernel
 132 + * (it now uses kmalloc etc..)
 133 + *
 134 + * Modified slightly by Matthew Strait to use more modern C.
 135 + */
 136 +
 137 +#include "regexp.h"
 138 +#include "regmagic.h"
 139 +
 140 +/* added by ethan and matt.  Lets it work in both kernel and user space.
 141 +(So iptables can use it, for instance.)  Yea, it goes both ways... */
 142 +#if __KERNEL__
 143 +  #define malloc(foo) kmalloc(foo,GFP_ATOMIC)
 144 +#else
 145 +  #define printk(format,args...) printf(format,##args)
 146 +#endif
 147 +
 148 +void regerror(char * s)
 149 +{
 150 +        printk("<3>Regexp: %s\n", s);
 151 +        /* NOTREACHED */
 152 +}
 153 +
 154 +/*
 155 + * The "internal use only" fields in regexp.h are present to pass info from
 156 + * compile to execute that permits the execute phase to run lots faster on
 157 + * simple cases.  They are:
 158 + *
 159 + * regstart    char that must begin a match; '\0' if none obvious
 160 + * reganch     is the match anchored (at beginning-of-line only)?
 161 + * regmust     string (pointer into program) that match must include, or NULL
 162 + * regmlen     length of regmust string
 163 + *
 164 + * Regstart and reganch permit very fast decisions on suitable starting points
 165 + * for a match, cutting down the work a lot.  Regmust permits fast rejection
 166 + * of lines that cannot possibly match.  The regmust tests are costly enough
 167 + * that regcomp() supplies a regmust only if the r.e. contains something
 168 + * potentially expensive (at present, the only such thing detected is * or +
 169 + * at the start of the r.e., which can involve a lot of backup).  Regmlen is
 170 + * supplied because the test in regexec() needs it and regcomp() is computing
 171 + * it anyway.
 172 + */
 173 +
 174 +/*
 175 + * Structure for regexp "program".  This is essentially a linear encoding
 176 + * of a nondeterministic finite-state machine (aka syntax charts or
 177 + * "railroad normal form" in parsing technology).  Each node is an opcode
 178 + * plus a "next" pointer, possibly plus an operand.  "Next" pointers of
 179 + * all nodes except BRANCH implement concatenation; a "next" pointer with
 180 + * a BRANCH on both ends of it is connecting two alternatives.  (Here we
 181 + * have one of the subtle syntax dependencies:  an individual BRANCH (as
 182 + * opposed to a collection of them) is never concatenated with anything
 183 + * because of operator precedence.)  The operand of some types of node is
 184 + * a literal string; for others, it is a node leading into a sub-FSM.  In
 185 + * particular, the operand of a BRANCH node is the first node of the branch.
 186 + * (NB this is *not* a tree structure:  the tail of the branch connects
 187 + * to the thing following the set of BRANCHes.)  The opcodes are:
 188 + */
 189 +
 190 +/* definition  number  opnd?   meaning */
 191 +#define        END     0       /* no   End of program. */
 192 +#define        BOL     1       /* no   Match "" at beginning of line. */
 193 +#define        EOL     2       /* no   Match "" at end of line. */
 194 +#define        ANY     3       /* no   Match any one character. */
 195 +#define        ANYOF   4       /* str  Match any character in this string. */
 196 +#define        ANYBUT  5       /* str  Match any character not in this string. */
 197 +#define        BRANCH  6       /* node Match this alternative, or the next... */
 198 +#define        BACK    7       /* no   Match "", "next" ptr points backward. */
 199 +#define        EXACTLY 8       /* str  Match this string. */
 200 +#define        NOTHING 9       /* no   Match empty string. */
 201 +#define        STAR    10      /* node Match this (simple) thing 0 or more times. */
 202 +#define        PLUS    11      /* node Match this (simple) thing 1 or more times. */
 203 +#define        OPEN    20      /* no   Mark this point in input as start of #n. */
 204 +                       /*      OPEN+1 is number 1, etc. */
 205 +#define        CLOSE   30      /* no   Analogous to OPEN. */
 206 +
 207 +/*
 208 + * Opcode notes:
 209 + *
 210 + * BRANCH      The set of branches constituting a single choice are hooked
 211 + *             together with their "next" pointers, since precedence prevents
 212 + *             anything being concatenated to any individual branch.  The
 213 + *             "next" pointer of the last BRANCH in a choice points to the
 214 + *             thing following the whole choice.  This is also where the
 215 + *             final "next" pointer of each individual branch points; each
 216 + *             branch starts with the operand node of a BRANCH node.
 217 + *
 218 + * BACK                Normal "next" pointers all implicitly point forward; BACK
 219 + *             exists to make loop structures possible.
 220 + *
 221 + * STAR,PLUS   '?', and complex '*' and '+', are implemented as circular
 222 + *             BRANCH structures using BACK.  Simple cases (one character
 223 + *             per match) are implemented with STAR and PLUS for speed
 224 + *             and to minimize recursive plunges.
 225 + *
 226 + * OPEN,CLOSE  ...are numbered at compile time.
 227 + */
 228 +
 229 +/*
 230 + * A node is one char of opcode followed by two chars of "next" pointer.
 231 + * "Next" pointers are stored as two 8-bit pieces, high order first.  The
 232 + * value is a positive offset from the opcode of the node containing it.
 233 + * An operand, if any, simply follows the node.  (Note that much of the
 234 + * code generation knows about this implicit relationship.)
 235 + *
 236 + * Using two bytes for the "next" pointer is vast overkill for most things,
 237 + * but allows patterns to get big without disasters.
 238 + */
 239 +#define        OP(p)   (*(p))
 240 +#define        NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377))
 241 +#define        OPERAND(p)      ((p) + 3)
 242 +
 243 +/*
 244 + * See regmagic.h for one further detail of program structure.
 245 + */
 246 +
 247 +
 248 +/*
 249 + * Utility definitions.
 250 + */
 251 +#ifndef CHARBITS
 252 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
 253 +#else
 254 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
 255 +#endif
 256 +
 257 +#define        FAIL(m) { regerror(m); return(NULL); }
 258 +#define        ISMULT(c)       ((c) == '*' || (c) == '+' || (c) == '?')
 259 +#define        META    "^$.[()|?+*\\"
 260 +
 261 +/*
 262 + * Flags to be passed up and down.
 263 + */
 264 +#define        HASWIDTH        01      /* Known never to match null string. */
 265 +#define        SIMPLE          02      /* Simple enough to be STAR/PLUS operand. */
 266 +#define        SPSTART         04      /* Starts with * or +. */
 267 +#define        WORST           0       /* Worst case. */
 268 +
 269 +/*
 270 + * Global work variables for regcomp().
 271 + */
 272 +struct match_globals {
 273 +char *reginput;                /* String-input pointer. */
 274 +char *regbol;          /* Beginning of input, for ^ check. */
 275 +char **regstartp;      /* Pointer to startp array. */
 276 +char **regendp;                /* Ditto for endp. */
 277 +char *regparse;                /* Input-scan pointer. */
 278 +int regnpar;           /* () count. */
 279 +char regdummy;
 280 +char *regcode;         /* Code-emit pointer; &regdummy = don't. */
 281 +long regsize;          /* Code size. */
 282 +};
 283 +
 284 +/*
 285 + * Forward declarations for regcomp()'s friends.
 286 + */
 287 +#ifndef STATIC
 288 +#define        STATIC  static
 289 +#endif
 290 +STATIC char *reg(struct match_globals *g, int paren,int *flagp);
 291 +STATIC char *regbranch(struct match_globals *g, int *flagp);
 292 +STATIC char *regpiece(struct match_globals *g, int *flagp);
 293 +STATIC char *regatom(struct match_globals *g, int *flagp);
 294 +STATIC char *regnode(struct match_globals *g, char op);
 295 +STATIC char *regnext(struct match_globals *g, char *p);
 296 +STATIC void regc(struct match_globals *g, char b);
 297 +STATIC void reginsert(struct match_globals *g, char op, char *opnd);
 298 +STATIC void regtail(struct match_globals *g, char *p, char *val);
 299 +STATIC void regoptail(struct match_globals *g, char *p, char *val);
 300 +
 301 +
 302 +__kernel_size_t my_strcspn(const char *s1,const char *s2)
 303 +{
 304 +        char *scan1;
 305 +        char *scan2;
 306 +        int count;
 307 +
 308 +        count = 0;
 309 +        for (scan1 = (char *)s1; *scan1 != '\0'; scan1++) {
 310 +                for (scan2 = (char *)s2; *scan2 != '\0';)       /* ++ moved down. */
 311 +                        if (*scan1 == *scan2++)
 312 +                                return(count);
 313 +                count++;
 314 +        }
 315 +        return(count);
 316 +}
 317 +
 318 +/*
 319 + - regcomp - compile a regular expression into internal code
 320 + *
 321 + * We can't allocate space until we know how big the compiled form will be,
 322 + * but we can't compile it (and thus know how big it is) until we've got a
 323 + * place to put the code.  So we cheat:  we compile it twice, once with code
 324 + * generation turned off and size counting turned on, and once "for real".
 325 + * This also means that we don't allocate space until we are sure that the
 326 + * thing really will compile successfully, and we never have to move the
 327 + * code and thus invalidate pointers into it.  (Note that it has to be in
 328 + * one piece because free() must be able to free it all.)
 329 + *
 330 + * Beware that the optimization-preparation code in here knows about some
 331 + * of the structure of the compiled regexp.
 332 + */
 333 +regexp *
 334 +regcomp(char *exp,int *patternsize)
 335 +{
 336 +       register regexp *r;
 337 +       register char *scan;
 338 +       register char *longest;
 339 +       register int len;
 340 +       int flags;
 341 +       struct match_globals g;
 342 +
 343 +       /* commented out by ethan
 344 +          extern char *malloc();
 345 +       */
 346 +
 347 +       if (exp == NULL)
 348 +               FAIL("NULL argument");
 349 +
 350 +       /* First pass: determine size, legality. */
 351 +       g.regparse = exp;
 352 +       g.regnpar = 1;
 353 +       g.regsize = 0L;
 354 +       g.regcode = &g.regdummy;
 355 +       regc(&g, MAGIC);
 356 +       if (reg(&g, 0, &flags) == NULL)
 357 +               return(NULL);
 358 +
 359 +       /* Small enough for pointer-storage convention? */
 360 +       if (g.regsize >= 32767L)                /* Probably could be 65535L. */
 361 +               FAIL("regexp too big");
 362 +
 363 +       /* Allocate space. */
 364 +       *patternsize=sizeof(regexp) + (unsigned)g.regsize;
 365 +       r = (regexp *)malloc(sizeof(regexp) + (unsigned)g.regsize);
 366 +       if (r == NULL)
 367 +               FAIL("out of space");
 368 +
 369 +       /* Second pass: emit code. */
 370 +       g.regparse = exp;
 371 +       g.regnpar = 1;
 372 +       g.regcode = r->program;
 373 +       regc(&g, MAGIC);
 374 +       if (reg(&g, 0, &flags) == NULL)
 375 +               return(NULL);
 376 +
 377 +       /* Dig out information for optimizations. */
 378 +       r->regstart = '\0';     /* Worst-case defaults. */
 379 +       r->reganch = 0;
 380 +       r->regmust = NULL;
 381 +       r->regmlen = 0;
 382 +       scan = r->program+1;                    /* First BRANCH. */
 383 +       if (OP(regnext(&g, scan)) == END) {             /* Only one top-level choice. */
 384 +               scan = OPERAND(scan);
 385 +
 386 +               /* Starting-point info. */
 387 +               if (OP(scan) == EXACTLY)
 388 +                       r->regstart = *OPERAND(scan);
 389 +               else if (OP(scan) == BOL)
 390 +                       r->reganch++;
 391 +
 392 +               /*
 393 +                * If there's something expensive in the r.e., find the
 394 +                * longest literal string that must appear and make it the
 395 +                * regmust.  Resolve ties in favor of later strings, since
 396 +                * the regstart check works with the beginning of the r.e.
 397 +                * and avoiding duplication strengthens checking.  Not a
 398 +                * strong reason, but sufficient in the absence of others.
 399 +                */
 400 +               if (flags&SPSTART) {
 401 +                       longest = NULL;
 402 +                       len = 0;
 403 +                       for (; scan != NULL; scan = regnext(&g, scan))
 404 +                               if (OP(scan) == EXACTLY && strlen(OPERAND(scan)) >= len) {
 405 +                                       longest = OPERAND(scan);
 406 +                                       len = strlen(OPERAND(scan));
 407 +                               }
 408 +                       r->regmust = longest;
 409 +                       r->regmlen = len;
 410 +               }
 411 +       }
 412 +
 413 +       return(r);
 414 +}
 415 +
 416 +/*
 417 + - reg - regular expression, i.e. main body or parenthesized thing
 418 + *
 419 + * Caller must absorb opening parenthesis.
 420 + *
 421 + * Combining parenthesis handling with the base level of regular expression
 422 + * is a trifle forced, but the need to tie the tails of the branches to what
 423 + * follows makes it hard to avoid.
 424 + */
 425 +static char *
 426 +reg(struct match_globals *g, int paren, int *flagp /* Parenthesized? */ )
 427 +{
 428 +       register char *ret;
 429 +       register char *br;
 430 +       register char *ender;
 431 +       register int parno = 0; /* 0 makes gcc happy */
 432 +       int flags;
 433 +
 434 +       *flagp = HASWIDTH;      /* Tentatively. */
 435 +
 436 +       /* Make an OPEN node, if parenthesized. */
 437 +       if (paren) {
 438 +               if (g->regnpar >= NSUBEXP)
 439 +                       FAIL("too many ()");
 440 +               parno = g->regnpar;
 441 +               g->regnpar++;
 442 +               ret = regnode(g, OPEN+parno);
 443 +       } else
 444 +               ret = NULL;
 445 +
 446 +       /* Pick up the branches, linking them together. */
 447 +       br = regbranch(g, &flags);
 448 +       if (br == NULL)
 449 +               return(NULL);
 450 +       if (ret != NULL)
 451 +               regtail(g, ret, br);    /* OPEN -> first. */
 452 +       else
 453 +               ret = br;
 454 +       if (!(flags&HASWIDTH))
 455 +               *flagp &= ~HASWIDTH;
 456 +       *flagp |= flags&SPSTART;
 457 +       while (*g->regparse == '|') {
 458 +               g->regparse++;
 459 +               br = regbranch(g, &flags);
 460 +               if (br == NULL)
 461 +                       return(NULL);
 462 +               regtail(g, ret, br);    /* BRANCH -> BRANCH. */
 463 +               if (!(flags&HASWIDTH))
 464 +                       *flagp &= ~HASWIDTH;
 465 +               *flagp |= flags&SPSTART;
 466 +       }
 467 +
 468 +       /* Make a closing node, and hook it on the end. */
 469 +       ender = regnode(g, (paren) ? CLOSE+parno : END);
 470 +       regtail(g, ret, ender);
 471 +
 472 +       /* Hook the tails of the branches to the closing node. */
 473 +       for (br = ret; br != NULL; br = regnext(g, br))
 474 +               regoptail(g, br, ender);
 475 +
 476 +       /* Check for proper termination. */
 477 +       if (paren && *g->regparse++ != ')') {
 478 +               FAIL("unmatched ()");
 479 +       } else if (!paren && *g->regparse != '\0') {
 480 +               if (*g->regparse == ')') {
 481 +                       FAIL("unmatched ()");
 482 +               } else
 483 +                       FAIL("junk on end");    /* "Can't happen". */
 484 +               /* NOTREACHED */
 485 +       }
 486 +
 487 +       return(ret);
 488 +}
 489 +
 490 +/*
 491 + - regbranch - one alternative of an | operator
 492 + *
 493 + * Implements the concatenation operator.
 494 + */
 495 +static char *
 496 +regbranch(struct match_globals *g, int *flagp)
 497 +{
 498 +       register char *ret;
 499 +       register char *chain;
 500 +       register char *latest;
 501 +       int flags;
 502 +
 503 +       *flagp = WORST;         /* Tentatively. */
 504 +
 505 +       ret = regnode(g, BRANCH);
 506 +       chain = NULL;
 507 +       while (*g->regparse != '\0' && *g->regparse != '|' && *g->regparse != ')') {
 508 +               latest = regpiece(g, &flags);
 509 +               if (latest == NULL)
 510 +                       return(NULL);
 511 +               *flagp |= flags&HASWIDTH;
 512 +               if (chain == NULL)      /* First piece. */
 513 +                       *flagp |= flags&SPSTART;
 514 +               else
 515 +                       regtail(g, chain, latest);
 516 +               chain = latest;
 517 +       }
 518 +       if (chain == NULL)      /* Loop ran zero times. */
 519 +               (void) regnode(g, NOTHING);
 520 +
 521 +       return(ret);
 522 +}
 523 +
 524 +/*
 525 + - regpiece - something followed by possible [*+?]
 526 + *
 527 + * Note that the branching code sequences used for ? and the general cases
 528 + * of * and + are somewhat optimized:  they use the same NOTHING node as
 529 + * both the endmarker for their branch list and the body of the last branch.
 530 + * It might seem that this node could be dispensed with entirely, but the
 531 + * endmarker role is not redundant.
 532 + */
 533 +static char *
 534 +regpiece(struct match_globals *g, int *flagp)
 535 +{
 536 +       register char *ret;
 537 +       register char op;
 538 +       register char *next;
 539 +       int flags;
 540 +
 541 +       ret = regatom(g, &flags);
 542 +       if (ret == NULL)
 543 +               return(NULL);
 544 +
 545 +       op = *g->regparse;
 546 +       if (!ISMULT(op)) {
 547 +               *flagp = flags;
 548 +               return(ret);
 549 +       }
 550 +
 551 +       if (!(flags&HASWIDTH) && op != '?')
 552 +               FAIL("*+ operand could be empty");
 553 +       *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
 554 +
 555 +       if (op == '*' && (flags&SIMPLE))
 556 +               reginsert(g, STAR, ret);
 557 +       else if (op == '*') {
 558 +               /* Emit x* as (x&|), where & means "self". */
 559 +               reginsert(g, BRANCH, ret);                      /* Either x */
 560 +               regoptail(g, ret, regnode(g, BACK));            /* and loop */
 561 +               regoptail(g, ret, ret);                 /* back */
 562 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
 563 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
 564 +       } else if (op == '+' && (flags&SIMPLE))
 565 +               reginsert(g, PLUS, ret);
 566 +       else if (op == '+') {
 567 +               /* Emit x+ as x(&|), where & means "self". */
 568 +               next = regnode(g, BRANCH);                      /* Either */
 569 +               regtail(g, ret, next);
 570 +               regtail(g, regnode(g, BACK), ret);              /* loop back */
 571 +               regtail(g, next, regnode(g, BRANCH));           /* or */
 572 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
 573 +       } else if (op == '?') {
 574 +               /* Emit x? as (x|) */
 575 +               reginsert(g, BRANCH, ret);                      /* Either x */
 576 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
 577 +               next = regnode(g, NOTHING);             /* null. */
 578 +               regtail(g, ret, next);
 579 +               regoptail(g, ret, next);
 580 +       }
 581 +       g->regparse++;
 582 +       if (ISMULT(*g->regparse))
 583 +               FAIL("nested *?+");
 584 +
 585 +       return(ret);
 586 +}
 587 +
 588 +/*
 589 + - regatom - the lowest level
 590 + *
 591 + * Optimization:  gobbles an entire sequence of ordinary characters so that
 592 + * it can turn them into a single node, which is smaller to store and
 593 + * faster to run.  Backslashed characters are exceptions, each becoming a
 594 + * separate node; the code is simpler that way and it's not worth fixing.
 595 + */
 596 +static char *
 597 +regatom(struct match_globals *g, int *flagp)
 598 +{
 599 +       register char *ret;
 600 +       int flags;
 601 +
 602 +       *flagp = WORST;         /* Tentatively. */
 603 +
 604 +       switch (*g->regparse++) {
 605 +       case '^':
 606 +               ret = regnode(g, BOL);
 607 +               break;
 608 +       case '$':
 609 +               ret = regnode(g, EOL);
 610 +               break;
 611 +       case '.':
 612 +               ret = regnode(g, ANY);
 613 +               *flagp |= HASWIDTH|SIMPLE;
 614 +               break;
 615 +       case '[': {
 616 +                       register int class;
 617 +                       register int classend;
 618 +
 619 +                       if (*g->regparse == '^') {      /* Complement of range. */
 620 +                               ret = regnode(g, ANYBUT);
 621 +                               g->regparse++;
 622 +                       } else
 623 +                               ret = regnode(g, ANYOF);
 624 +                       if (*g->regparse == ']' || *g->regparse == '-')
 625 +                               regc(g, *g->regparse++);
 626 +                       while (*g->regparse != '\0' && *g->regparse != ']') {
 627 +                               if (*g->regparse == '-') {
 628 +                                       g->regparse++;
 629 +                                       if (*g->regparse == ']' || *g->regparse == '\0')
 630 +                                               regc(g, '-');
 631 +                                       else {
 632 +                                               class = UCHARAT(g->regparse-2)+1;
 633 +                                               classend = UCHARAT(g->regparse);
 634 +                                               if (class > classend+1)
 635 +                                                       FAIL("invalid [] range");
 636 +                                               for (; class <= classend; class++)
 637 +                                                       regc(g, class);
 638 +                                               g->regparse++;
 639 +                                       }
 640 +                               } else
 641 +                                       regc(g, *g->regparse++);
 642 +                       }
 643 +                       regc(g, '\0');
 644 +                       if (*g->regparse != ']')
 645 +                               FAIL("unmatched []");
 646 +                       g->regparse++;
 647 +                       *flagp |= HASWIDTH|SIMPLE;
 648 +               }
 649 +               break;
 650 +       case '(':
 651 +               ret = reg(g, 1, &flags);
 652 +               if (ret == NULL)
 653 +                       return(NULL);
 654 +               *flagp |= flags&(HASWIDTH|SPSTART);
 655 +               break;
 656 +       case '\0':
 657 +       case '|':
 658 +       case ')':
 659 +               FAIL("internal urp");   /* Supposed to be caught earlier. */
 660 +               break;
 661 +       case '?':
 662 +       case '+':
 663 +       case '*':
 664 +               FAIL("?+* follows nothing");
 665 +               break;
 666 +       case '\\':
 667 +               if (*g->regparse == '\0')
 668 +                       FAIL("trailing \\");
 669 +               ret = regnode(g, EXACTLY);
 670 +               regc(g, *g->regparse++);
 671 +               regc(g, '\0');
 672 +               *flagp |= HASWIDTH|SIMPLE;
 673 +               break;
 674 +       default: {
 675 +                       register int len;
 676 +                       register char ender;
 677 +
 678 +                       g->regparse--;
 679 +                       len = my_strcspn((const char *)g->regparse, (const char *)META);
 680 +                       if (len <= 0)
 681 +                               FAIL("internal disaster");
 682 +                       ender = *(g->regparse+len);
 683 +                       if (len > 1 && ISMULT(ender))
 684 +                               len--;          /* Back off clear of ?+* operand. */
 685 +                       *flagp |= HASWIDTH;
 686 +                       if (len == 1)
 687 +                               *flagp |= SIMPLE;
 688 +                       ret = regnode(g, EXACTLY);
 689 +                       while (len > 0) {
 690 +                               regc(g, *g->regparse++);
 691 +                               len--;
 692 +                       }
 693 +                       regc(g, '\0');
 694 +               }
 695 +               break;
 696 +       }
 697 +
 698 +       return(ret);
 699 +}
 700 +
 701 +/*
 702 + - regnode - emit a node
 703 + */
 704 +static char *                  /* Location. */
 705 +regnode(struct match_globals *g, char op)
 706 +{
 707 +       register char *ret;
 708 +       register char *ptr;
 709 +
 710 +       ret = g->regcode;
 711 +       if (ret == &g->regdummy) {
 712 +               g->regsize += 3;
 713 +               return(ret);
 714 +       }
 715 +
 716 +       ptr = ret;
 717 +       *ptr++ = op;
 718 +       *ptr++ = '\0';          /* Null "next" pointer. */
 719 +       *ptr++ = '\0';
 720 +       g->regcode = ptr;
 721 +
 722 +       return(ret);
 723 +}
 724 +
 725 +/*
 726 + - regc - emit (if appropriate) a byte of code
 727 + */
 728 +static void
 729 +regc(struct match_globals *g, char b)
 730 +{
 731 +       if (g->regcode != &g->regdummy)
 732 +               *g->regcode++ = b;
 733 +       else
 734 +               g->regsize++;
 735 +}
 736 +
 737 +/*
 738 + - reginsert - insert an operator in front of already-emitted operand
 739 + *
 740 + * Means relocating the operand.
 741 + */
 742 +static void
 743 +reginsert(struct match_globals *g, char op, char* opnd)
 744 +{
 745 +       register char *src;
 746 +       register char *dst;
 747 +       register char *place;
 748 +
 749 +       if (g->regcode == &g->regdummy) {
 750 +               g->regsize += 3;
 751 +               return;
 752 +       }
 753 +
 754 +       src = g->regcode;
 755 +       g->regcode += 3;
 756 +       dst = g->regcode;
 757 +       while (src > opnd)
 758 +               *--dst = *--src;
 759 +
 760 +       place = opnd;           /* Op node, where operand used to be. */
 761 +       *place++ = op;
 762 +       *place++ = '\0';
 763 +       *place++ = '\0';
 764 +}
 765 +
 766 +/*
 767 + - regtail - set the next-pointer at the end of a node chain
 768 + */
 769 +static void
 770 +regtail(struct match_globals *g, char *p, char *val)
 771 +{
 772 +       register char *scan;
 773 +       register char *temp;
 774 +       register int offset;
 775 +
 776 +       if (p == &g->regdummy)
 777 +               return;
 778 +
 779 +       /* Find last node. */
 780 +       scan = p;
 781 +       for (;;) {
 782 +               temp = regnext(g, scan);
 783 +               if (temp == NULL)
 784 +                       break;
 785 +               scan = temp;
 786 +       }
 787 +
 788 +       if (OP(scan) == BACK)
 789 +               offset = scan - val;
 790 +       else
 791 +               offset = val - scan;
 792 +       *(scan+1) = (offset>>8)&0377;
 793 +       *(scan+2) = offset&0377;
 794 +}
 795 +
 796 +/*
 797 + - regoptail - regtail on operand of first argument; nop if operandless
 798 + */
 799 +static void
 800 +regoptail(struct match_globals *g, char *p, char *val)
 801 +{
 802 +       /* "Operandless" and "op != BRANCH" are synonymous in practice. */
 803 +       if (p == NULL || p == &g->regdummy || OP(p) != BRANCH)
 804 +               return;
 805 +       regtail(g, OPERAND(p), val);
 806 +}
 807 +
 808 +/*
 809 + * regexec and friends
 810 + */
 811 +
 812 +
 813 +/*
 814 + * Forwards.
 815 + */
 816 +STATIC int regtry(struct match_globals *g, regexp *prog, char *string);
 817 +STATIC int regmatch(struct match_globals *g, char *prog);
 818 +STATIC int regrepeat(struct match_globals *g, char *p);
 819 +
 820 +#ifdef DEBUG
 821 +int regnarrate = 0;
 822 +void regdump();
 823 +STATIC char *regprop(char *op);
 824 +#endif
 825 +
 826 +/*
 827 + - regexec - match a regexp against a string
 828 + */
 829 +int
 830 +regexec(regexp *prog, char *string)
 831 +{
 832 +       register char *s;
 833 +       struct match_globals g;
 834 +
 835 +       /* Be paranoid... */
 836 +       if (prog == NULL || string == NULL) {
 837 +               printk("<3>Regexp: NULL parameter\n");
 838 +               return(0);
 839 +       }
 840 +
 841 +       /* Check validity of program. */
 842 +       if (UCHARAT(prog->program) != MAGIC) {
 843 +               printk("<3>Regexp: corrupted program\n");
 844 +               return(0);
 845 +       }
 846 +
 847 +       /* If there is a "must appear" string, look for it. */
 848 +       if (prog->regmust != NULL) {
 849 +               s = string;
 850 +               while ((s = strchr(s, prog->regmust[0])) != NULL) {
 851 +                       if (strncmp(s, prog->regmust, prog->regmlen) == 0)
 852 +                               break;  /* Found it. */
 853 +                       s++;
 854 +               }
 855 +               if (s == NULL)  /* Not present. */
 856 +                       return(0);
 857 +       }
 858 +
 859 +       /* Mark beginning of line for ^ . */
 860 +       g.regbol = string;
 861 +
 862 +       /* Simplest case:  anchored match need be tried only once. */
 863 +       if (prog->reganch)
 864 +               return(regtry(&g, prog, string));
 865 +
 866 +       /* Messy cases:  unanchored match. */
 867 +       s = string;
 868 +       if (prog->regstart != '\0')
 869 +               /* We know what char it must start with. */
 870 +               while ((s = strchr(s, prog->regstart)) != NULL) {
 871 +                       if (regtry(&g, prog, s))
 872 +                               return(1);
 873 +                       s++;
 874 +               }
 875 +       else
 876 +               /* We don't -- general case. */
 877 +               do {
 878 +                       if (regtry(&g, prog, s))
 879 +                               return(1);
 880 +               } while (*s++ != '\0');
 881 +
 882 +       /* Failure. */
 883 +       return(0);
 884 +}
 885 +
 886 +/*
 887 + - regtry - try match at specific point
 888 + */
 889 +static int                     /* 0 failure, 1 success */
 890 +regtry(struct match_globals *g, regexp *prog, char *string)
 891 +{
 892 +       register int i;
 893 +       register char **sp;
 894 +       register char **ep;
 895 +
 896 +       g->reginput = string;
 897 +       g->regstartp = prog->startp;
 898 +       g->regendp = prog->endp;
 899 +
 900 +       sp = prog->startp;
 901 +       ep = prog->endp;
 902 +       for (i = NSUBEXP; i > 0; i--) {
 903 +               *sp++ = NULL;
 904 +               *ep++ = NULL;
 905 +       }
 906 +       if (regmatch(g, prog->program + 1)) {
 907 +               prog->startp[0] = string;
 908 +               prog->endp[0] = g->reginput;
 909 +               return(1);
 910 +       } else
 911 +               return(0);
 912 +}
 913 +
 914 +/*
 915 + - regmatch - main matching routine
 916 + *
 917 + * Conceptually the strategy is simple:  check to see whether the current
 918 + * node matches, call self recursively to see whether the rest matches,
 919 + * and then act accordingly.  In practice we make some effort to avoid
 920 + * recursion, in particular by going through "ordinary" nodes (that don't
 921 + * need to know whether the rest of the match failed) by a loop instead of
 922 + * by recursion.
 923 + */
 924 +static int                     /* 0 failure, 1 success */
 925 +regmatch(struct match_globals *g, char *prog)
 926 +{
 927 +       register char *scan = prog; /* Current node. */
 928 +       char *next;                 /* Next node. */
 929 +
 930 +#ifdef DEBUG
 931 +       if (scan != NULL && regnarrate)
 932 +               fprintf(stderr, "%s(\n", regprop(scan));
 933 +#endif
 934 +       while (scan != NULL) {
 935 +#ifdef DEBUG
 936 +               if (regnarrate)
 937 +                       fprintf(stderr, "%s...\n", regprop(scan));
 938 +#endif
 939 +               next = regnext(g, scan);
 940 +
 941 +               switch (OP(scan)) {
 942 +               case BOL:
 943 +                       if (g->reginput != g->regbol)
 944 +                               return(0);
 945 +                       break;
 946 +               case EOL:
 947 +                       if (*g->reginput != '\0')
 948 +                               return(0);
 949 +                       break;
 950 +               case ANY:
 951 +                       if (*g->reginput == '\0')
 952 +                               return(0);
 953 +                       g->reginput++;
 954 +                       break;
 955 +               case EXACTLY: {
 956 +                               register int len;
 957 +                               register char *opnd;
 958 +
 959 +                               opnd = OPERAND(scan);
 960 +                               /* Inline the first character, for speed. */
 961 +                               if (*opnd != *g->reginput)
 962 +                                       return(0);
 963 +                               len = strlen(opnd);
 964 +                               if (len > 1 && strncmp(opnd, g->reginput, len) != 0)
 965 +                                       return(0);
 966 +                               g->reginput += len;
 967 +                       }
 968 +                       break;
 969 +               case ANYOF:
 970 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) == NULL)
 971 +                               return(0);
 972 +                       g->reginput++;
 973 +                       break;
 974 +               case ANYBUT:
 975 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) != NULL)
 976 +                               return(0);
 977 +                       g->reginput++;
 978 +                       break;
 979 +               case NOTHING:
 980 +               case BACK:
 981 +                       break;
 982 +               case OPEN+1:
 983 +               case OPEN+2:
 984 +               case OPEN+3:
 985 +               case OPEN+4:
 986 +               case OPEN+5:
 987 +               case OPEN+6:
 988 +               case OPEN+7:
 989 +               case OPEN+8:
 990 +               case OPEN+9: {
 991 +                               register int no;
 992 +                               register char *save;
 993 +
 994 +                               no = OP(scan) - OPEN;
 995 +                               save = g->reginput;
 996 +
 997 +                               if (regmatch(g, next)) {
 998 +                                       /*
 999 +                                        * Don't set startp if some later
1000 +                                        * invocation of the same parentheses
1001 +                                        * already has.
1002 +                                        */
1003 +                                       if (g->regstartp[no] == NULL)
1004 +                                               g->regstartp[no] = save;
1005 +                                       return(1);
1006 +                               } else
1007 +                                       return(0);
1008 +                       }
1009 +                       break;
1010 +               case CLOSE+1:
1011 +               case CLOSE+2:
1012 +               case CLOSE+3:
1013 +               case CLOSE+4:
1014 +               case CLOSE+5:
1015 +               case CLOSE+6:
1016 +               case CLOSE+7:
1017 +               case CLOSE+8:
1018 +               case CLOSE+9:
1019 +                       {
1020 +                               register int no;
1021 +                               register char *save;
1022 +
1023 +                               no = OP(scan) - CLOSE;
1024 +                               save = g->reginput;
1025 +
1026 +                               if (regmatch(g, next)) {
1027 +                                       /*
1028 +                                        * Don't set endp if some later
1029 +                                        * invocation of the same parentheses
1030 +                                        * already has.
1031 +                                        */
1032 +                                       if (g->regendp[no] == NULL)
1033 +                                               g->regendp[no] = save;
1034 +                                       return(1);
1035 +                               } else
1036 +                                       return(0);
1037 +                       }
1038 +                       break;
1039 +               case BRANCH: {
1040 +                               register char *save;
1041 +
1042 +                               if (OP(next) != BRANCH)         /* No choice. */
1043 +                                       next = OPERAND(scan);   /* Avoid recursion. */
1044 +                               else {
1045 +                                       do {
1046 +                                               save = g->reginput;
1047 +                                               if (regmatch(g, OPERAND(scan)))
1048 +                                                       return(1);
1049 +                                               g->reginput = save;
1050 +                                               scan = regnext(g, scan);
1051 +                                       } while (scan != NULL && OP(scan) == BRANCH);
1052 +                                       return(0);
1053 +                                       /* NOTREACHED */
1054 +                               }
1055 +                       }
1056 +                       break;
1057 +               case STAR:
1058 +               case PLUS: {
1059 +                               register char nextch;
1060 +                               register int no;
1061 +                               register char *save;
1062 +                               register int min;
1063 +
1064 +                               /*
1065 +                                * Lookahead to avoid useless match attempts
1066 +                                * when we know what character comes next.
1067 +                                */
1068 +                               nextch = '\0';
1069 +                               if (OP(next) == EXACTLY)
1070 +                                       nextch = *OPERAND(next);
1071 +                               min = (OP(scan) == STAR) ? 0 : 1;
1072 +                               save = g->reginput;
1073 +                               no = regrepeat(g, OPERAND(scan));
1074 +                               while (no >= min) {
1075 +                                       /* If it could work, try it. */
1076 +                                       if (nextch == '\0' || *g->reginput == nextch)
1077 +                                               if (regmatch(g, next))
1078 +                                                       return(1);
1079 +                                       /* Couldn't or didn't -- back up. */
1080 +                                       no--;
1081 +                                       g->reginput = save + no;
1082 +                               }
1083 +                               return(0);
1084 +                       }
1085 +                       break;
1086 +               case END:
1087 +                       return(1);      /* Success! */
1088 +                       break;
1089 +               default:
1090 +                       printk("<3>Regexp: memory corruption\n");
1091 +                       return(0);
1092 +                       break;
1093 +               }
1094 +
1095 +               scan = next;
1096 +       }
1097 +
1098 +       /*
1099 +        * We get here only if there's trouble -- normally "case END" is
1100 +        * the terminating point.
1101 +        */
1102 +       printk("<3>Regexp: corrupted pointers\n");
1103 +       return(0);
1104 +}
1105 +
1106 +/*
1107 + - regrepeat - repeatedly match something simple, report how many
1108 + */
1109 +static int
1110 +regrepeat(struct match_globals *g, char *p)
1111 +{
1112 +       register int count = 0;
1113 +       register char *scan;
1114 +       register char *opnd;
1115 +
1116 +       scan = g->reginput;
1117 +       opnd = OPERAND(p);
1118 +       switch (OP(p)) {
1119 +       case ANY:
1120 +               count = strlen(scan);
1121 +               scan += count;
1122 +               break;
1123 +       case EXACTLY:
1124 +               while (*opnd == *scan) {
1125 +                       count++;
1126 +                       scan++;
1127 +               }
1128 +               break;
1129 +       case ANYOF:
1130 +               while (*scan != '\0' && strchr(opnd, *scan) != NULL) {
1131 +                       count++;
1132 +                       scan++;
1133 +               }
1134 +               break;
1135 +       case ANYBUT:
1136 +               while (*scan != '\0' && strchr(opnd, *scan) == NULL) {
1137 +                       count++;
1138 +                       scan++;
1139 +               }
1140 +               break;
1141 +       default:                /* Oh dear.  Called inappropriately. */
1142 +               printk("<3>Regexp: internal foulup\n");
1143 +               count = 0;      /* Best compromise. */
1144 +               break;
1145 +       }
1146 +       g->reginput = scan;
1147 +
1148 +       return(count);
1149 +}
1150 +
1151 +/*
1152 + - regnext - dig the "next" pointer out of a node
1153 + */
1154 +static char*
1155 +regnext(struct match_globals *g, char *p)
1156 +{
1157 +       register int offset;
1158 +
1159 +       if (p == &g->regdummy)
1160 +               return(NULL);
1161 +
1162 +       offset = NEXT(p);
1163 +       if (offset == 0)
1164 +               return(NULL);
1165 +
1166 +       if (OP(p) == BACK)
1167 +               return(p-offset);
1168 +       else
1169 +               return(p+offset);
1170 +}
1171 +
1172 +#ifdef DEBUG
1173 +
1174 +STATIC char *regprop();
1175 +
1176 +/*
1177 + - regdump - dump a regexp onto stdout in vaguely comprehensible form
1178 + */
1179 +void
1180 +regdump(regexp *r)
1181 +{
1182 +       register char *s;
1183 +       register char op = EXACTLY;     /* Arbitrary non-END op. */
1184 +       register char *next;
1185 +       /* extern char *strchr(); */
1186 +
1187 +
1188 +       s = r->program + 1;
1189 +       while (op != END) {     /* While that wasn't END last time... */
1190 +               op = OP(s);
1191 +               printf("%2d%s", s-r->program, regprop(s));      /* Where, what. */
1192 +               next = regnext(s);
1193 +               if (next == NULL)               /* Next ptr. */
1194 +                       printf("(0)");
1195 +               else
1196 +                       printf("(%d)", (s-r->program)+(next-s));
1197 +               s += 3;
1198 +               if (op == ANYOF || op == ANYBUT || op == EXACTLY) {
1199 +                       /* Literal string, where present. */
1200 +                       while (*s != '\0') {
1201 +                               putchar(*s);
1202 +                               s++;
1203 +                       }
1204 +                       s++;
1205 +               }
1206 +               putchar('\n');
1207 +       }
1208 +
1209 +       /* Header fields of interest. */
1210 +       if (r->regstart != '\0')
1211 +               printf("start `%c' ", r->regstart);
1212 +       if (r->reganch)
1213 +               printf("anchored ");
1214 +       if (r->regmust != NULL)
1215 +               printf("must have \"%s\"", r->regmust);
1216 +       printf("\n");
1217 +}
1218 +
1219 +/*
1220 + - regprop - printable representation of opcode
1221 + */
1222 +static char *
1223 +regprop(char *op)
1224 +{
1225 +#define BUFLEN 50
1226 +       register char *p;
1227 +       static char buf[BUFLEN];
1228 +
1229 +       strcpy(buf, ":");
1230 +
1231 +       switch (OP(op)) {
1232 +       case BOL:
1233 +               p = "BOL";
1234 +               break;
1235 +       case EOL:
1236 +               p = "EOL";
1237 +               break;
1238 +       case ANY:
1239 +               p = "ANY";
1240 +               break;
1241 +       case ANYOF:
1242 +               p = "ANYOF";
1243 +               break;
1244 +       case ANYBUT:
1245 +               p = "ANYBUT";
1246 +               break;
1247 +       case BRANCH:
1248 +               p = "BRANCH";
1249 +               break;
1250 +       case EXACTLY:
1251 +               p = "EXACTLY";
1252 +               break;
1253 +       case NOTHING:
1254 +               p = "NOTHING";
1255 +               break;
1256 +       case BACK:
1257 +               p = "BACK";
1258 +               break;
1259 +       case END:
1260 +               p = "END";
1261 +               break;
1262 +       case OPEN+1:
1263 +       case OPEN+2:
1264 +       case OPEN+3:
1265 +       case OPEN+4:
1266 +       case OPEN+5:
1267 +       case OPEN+6:
1268 +       case OPEN+7:
1269 +       case OPEN+8:
1270 +       case OPEN+9:
1271 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "OPEN%d", OP(op)-OPEN);
1272 +               p = NULL;
1273 +               break;
1274 +       case CLOSE+1:
1275 +       case CLOSE+2:
1276 +       case CLOSE+3:
1277 +       case CLOSE+4:
1278 +       case CLOSE+5:
1279 +       case CLOSE+6:
1280 +       case CLOSE+7:
1281 +       case CLOSE+8:
1282 +       case CLOSE+9:
1283 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "CLOSE%d", OP(op)-CLOSE);
1284 +               p = NULL;
1285 +               break;
1286 +       case STAR:
1287 +               p = "STAR";
1288 +               break;
1289 +       case PLUS:
1290 +               p = "PLUS";
1291 +               break;
1292 +       default:
1293 +               printk("<3>Regexp: corrupted opcode\n");
1294 +               break;
1295 +       }
1296 +       if (p != NULL)
1297 +               strncat(buf, p, BUFLEN-strlen(buf));
1298 +       return(buf);
1299 +}
1300 +#endif
1301 +
1302 +
1303 diff -urNp -x '*.orig' linux-5.11/net/netfilter/regexp/regexp.h linux-5.11/net/netfilter/regexp/regexp.h
1304 --- linux-5.11/net/netfilter/regexp/regexp.h    1970-01-01 01:00:00.000000000 +0100
1305 +++ linux-5.11/net/netfilter/regexp/regexp.h    2021-02-24 13:21:48.335107407 +0100
1306 @@ -0,0 +1,41 @@
1307 +/*
1308 + * Definitions etc. for regexp(3) routines.
1309 + *
1310 + * Caveat:  this is V8 regexp(3) [actually, a reimplementation thereof],
1311 + * not the System V one.
1312 + */
1313 +
1314 +#ifndef REGEXP_H
1315 +#define REGEXP_H
1316 +
1317 +
1318 +/*
1319 +http://www.opensource.apple.com/darwinsource/10.3/expect-1/expect/expect.h ,
1320 +which contains a version of this library, says:
1321 +
1322 + *
1323 + * NSUBEXP must be at least 10, and no greater than 117 or the parser
1324 + * will not work properly.
1325 + *
1326 +
1327 +However, it looks rather like this library is limited to 10.  If you think
1328 +otherwise, let us know.
1329 +*/
1330 +
1331 +#define NSUBEXP  10
1332 +typedef struct regexp {
1333 +       char *startp[NSUBEXP];
1334 +       char *endp[NSUBEXP];
1335 +       char regstart;          /* Internal use only. */
1336 +       char reganch;           /* Internal use only. */
1337 +       char *regmust;          /* Internal use only. */
1338 +       int regmlen;            /* Internal use only. */
1339 +       char program[1];        /* Unwarranted chumminess with compiler. */
1340 +} regexp;
1341 +
1342 +regexp * regcomp(char *exp, int *patternsize);
1343 +int regexec(regexp *prog, char *string);
1344 +void regsub(regexp *prog, char *source, char *dest);
1345 +void regerror(char *s);
1346 +
1347 +#endif
1348 diff -urNp -x '*.orig' linux-5.11/net/netfilter/regexp/regmagic.h linux-5.11/net/netfilter/regexp/regmagic.h
1349 --- linux-5.11/net/netfilter/regexp/regmagic.h  1970-01-01 01:00:00.000000000 +0100
1350 +++ linux-5.11/net/netfilter/regexp/regmagic.h  2021-02-24 13:21:48.335107407 +0100
1351 @@ -0,0 +1,5 @@
1352 +/*
1353 + * The first byte of the regexp internal "program" is actually this magic
1354 + * number; the start node begins in the second byte.
1355 + */
1356 +#define        MAGIC   0234
1357 diff -urNp -x '*.orig' linux-5.11/net/netfilter/regexp/regsub.c linux-5.11/net/netfilter/regexp/regsub.c
1358 --- linux-5.11/net/netfilter/regexp/regsub.c    1970-01-01 01:00:00.000000000 +0100
1359 +++ linux-5.11/net/netfilter/regexp/regsub.c    2021-02-24 13:21:48.335107407 +0100
1360 @@ -0,0 +1,95 @@
1361 +/*
1362 + * regsub
1363 + * @(#)regsub.c        1.3 of 2 April 86
1364 + *
1365 + *     Copyright (c) 1986 by University of Toronto.
1366 + *     Written by Henry Spencer.  Not derived from licensed software.
1367 + *
1368 + *     Permission is granted to anyone to use this software for any
1369 + *     purpose on any computer system, and to redistribute it freely,
1370 + *     subject to the following restrictions:
1371 + *
1372 + *     1. The author is not responsible for the consequences of use of
1373 + *             this software, no matter how awful, even if they arise
1374 + *             from defects in it.
1375 + *
1376 + *     2. The origin of this software must not be misrepresented, either
1377 + *             by explicit claim or by omission.
1378 + *
1379 + *     3. Altered versions must be plainly marked as such, and must not
1380 + *             be misrepresented as being the original software.
1381 + *
1382 + *
1383 + * This code was modified by Ethan Sommer to work within the kernel
1384 + * (it now uses kmalloc etc..)
1385 + *
1386 + */
1387 +#include "regexp.h"
1388 +#include "regmagic.h"
1389 +#include <linux/string.h>
1390 +
1391 +
1392 +#ifndef CHARBITS
1393 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
1394 +#else
1395 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
1396 +#endif
1397 +
1398 +#if 0
1399 +//void regerror(char * s)
1400 +//{
1401 +//        printk("regexp(3): %s", s);
1402 +//        /* NOTREACHED */
1403 +//}
1404 +#endif
1405 +
1406 +/*
1407 + - regsub - perform substitutions after a regexp match
1408 + */
1409 +void
1410 +regsub(regexp * prog, char * source, char * dest)
1411 +{
1412 +       register char *src;
1413 +       register char *dst;
1414 +       register char c;
1415 +       register int no;
1416 +       register int len;
1417 +
1418 +       /* Not necessary and gcc doesn't like it -MLS */
1419 +       /*extern char *strncpy();*/
1420 +
1421 +       if (prog == NULL || source == NULL || dest == NULL) {
1422 +               regerror("NULL parm to regsub");
1423 +               return;
1424 +       }
1425 +       if (UCHARAT(prog->program) != MAGIC) {
1426 +               regerror("damaged regexp fed to regsub");
1427 +               return;
1428 +       }
1429 +
1430 +       src = source;
1431 +       dst = dest;
1432 +       while ((c = *src++) != '\0') {
1433 +               if (c == '&')
1434 +                       no = 0;
1435 +               else if (c == '\\' && '0' <= *src && *src <= '9')
1436 +                       no = *src++ - '0';
1437 +               else
1438 +                       no = -1;
1439 +
1440 +               if (no < 0) {   /* Ordinary character. */
1441 +                       if (c == '\\' && (*src == '\\' || *src == '&'))
1442 +                               c = *src++;
1443 +                       *dst++ = c;
1444 +               } else if (prog->startp[no] != NULL && prog->endp[no] != NULL) {
1445 +                       len = prog->endp[no] - prog->startp[no];
1446 +                       (void) strncpy(dst, prog->startp[no], len);
1447 +                       dst += len;
1448 +                       if (len != 0 && *(dst-1) == '\0') {     /* strncpy hit NUL. */
1449 +                               regerror("damaged match string");
1450 +                               return;
1451 +                       }
1452 +               }
1453 +       }
1454 +       *dst++ = '\0';
1455 +}
1456 diff -urNp -x '*.orig' linux-5.11/net/netfilter/xt_layer7.c linux-5.11/net/netfilter/xt_layer7.c
1457 --- linux-5.11/net/netfilter/xt_layer7.c        1970-01-01 01:00:00.000000000 +0100
1458 +++ linux-5.11/net/netfilter/xt_layer7.c        2021-02-24 13:21:48.335107407 +0100
1459 @@ -0,0 +1,656 @@
1460 +/*
1461 +  Kernel module to match application layer (OSI layer 7) data in connections.
1462 +
1463 +  http://l7-filter.sf.net
1464 +
1465 +  (C) 2003-2009 Matthew Strait and Ethan Sommer.
1466 +
1467 +  This program is free software; you can redistribute it and/or
1468 +  modify it under the terms of the GNU General Public License
1469 +  as published by the Free Software Foundation; either version
1470 +  2 of the License, or (at your option) any later version.
1471 +  http://www.gnu.org/licenses/gpl.txt
1472 +
1473 +  Based on ipt_string.c (C) 2000 Emmanuel Roger <winfield@freegates.be>,
1474 +  xt_helper.c (C) 2002 Harald Welte and cls_layer7.c (C) 2003 Matthew Strait,
1475 +  Ethan Sommer, Justin Levandoski.
1476 +*/
1477 +
1478 +#include <linux/spinlock.h>
1479 +#include <linux/version.h>
1480 +#include <net/ip.h>
1481 +#include <net/tcp.h>
1482 +#include <linux/module.h>
1483 +#include <linux/skbuff.h>
1484 +#include <linux/netfilter.h>
1485 +#include <net/netfilter/nf_conntrack.h>
1486 +#include <net/netfilter/nf_conntrack_core.h>
1487 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 27)
1488 +#include <net/netfilter/nf_conntrack_extend.h>
1489 +#include <net/netfilter/nf_conntrack_acct.h>
1490 +#endif
1491 +#include <linux/netfilter/x_tables.h>
1492 +#include <linux/netfilter/xt_layer7.h>
1493 +#include <linux/ctype.h>
1494 +#include <linux/proc_fs.h>
1495 +
1496 +#include "regexp/regexp.c"
1497 +
1498 +MODULE_LICENSE("GPL");
1499 +MODULE_AUTHOR("Matthew Strait <quadong@users.sf.net>, Ethan Sommer <sommere@users.sf.net>");
1500 +MODULE_DESCRIPTION("iptables application layer match module");
1501 +MODULE_ALIAS("ipt_layer7");
1502 +MODULE_VERSION("2.21");
1503 +
1504 +static int maxdatalen = 2048; // this is the default
1505 +module_param(maxdatalen, int, 0444);
1506 +MODULE_PARM_DESC(maxdatalen, "maximum bytes of data looked at by l7-filter");
1507 +#ifdef CONFIG_NETFILTER_XT_MATCH_LAYER7_DEBUG
1508 +       #define DPRINTK(format,args...) printk(format,##args)
1509 +#else
1510 +       #define DPRINTK(format,args...)
1511 +#endif
1512 +
1513 +/* Number of packets whose data we look at.
1514 +This can be modified through /proc/net/layer7_numpackets */
1515 +static int num_packets = 10;
1516 +
1517 +static struct pattern_cache {
1518 +       char * regex_string;
1519 +       regexp * pattern;
1520 +       struct pattern_cache * next;
1521 +} * first_pattern_cache = NULL;
1522 +
1523 +DEFINE_SPINLOCK(l7_lock);
1524 +
1525 +static int total_acct_packets(struct nf_conn *ct)
1526 +{
1527 +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2, 6, 26)
1528 +       BUG_ON(ct == NULL);
1529 +       return (ct->counters[IP_CT_DIR_ORIGINAL].packets + ct->counters[IP_CT_DIR_REPLY].packets);
1530 +#else
1531 +       struct nf_conn_counter *acct;
1532 +
1533 +       BUG_ON(ct == NULL);
1534 +       acct = nf_conn_acct_find(ct);
1535 +       if (!acct)
1536 +               return 0;
1537 +       return (atomic64_read(&acct[IP_CT_DIR_ORIGINAL].packets) + atomic64_read(&acct[IP_CT_DIR_REPLY].packets));
1538 +#endif
1539 +}
1540 +
1541 +#ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
1542 +/* Converts an unfriendly string into a friendly one by
1543 +replacing unprintables with periods and all whitespace with " ". */
1544 +static char * friendly_print(unsigned char * s)
1545 +{
1546 +       char * f = kmalloc(strlen(s) + 1, GFP_ATOMIC);
1547 +       int i;
1548 +
1549 +       if(!f) {
1550 +               if (net_ratelimit())
1551 +                       printk(KERN_ERR "layer7: out of memory in "
1552 +                                       "friendly_print, bailing.\n");
1553 +               return NULL;
1554 +       }
1555 +
1556 +       for(i = 0; i < strlen(s); i++){
1557 +               if(isprint(s[i]) && s[i] < 128) f[i] = s[i];
1558 +               else if(isspace(s[i]))          f[i] = ' ';
1559 +               else                            f[i] = '.';
1560 +       }
1561 +       f[i] = '\0';
1562 +       return f;
1563 +}
1564 +
1565 +static char dec2hex(int i)
1566 +{
1567 +       switch (i) {
1568 +               case 0 ... 9:
1569 +                       return (i + '0');
1570 +                       break;
1571 +               case 10 ... 15:
1572 +                       return (i - 10 + 'a');
1573 +                       break;
1574 +               default:
1575 +                       if (net_ratelimit())
1576 +                               printk("layer7: Problem in dec2hex\n");
1577 +                       return '\0';
1578 +       }
1579 +}
1580 +
1581 +static char * hex_print(unsigned char * s)
1582 +{
1583 +       char * g = kmalloc(strlen(s)*3 + 1, GFP_ATOMIC);
1584 +       int i;
1585 +
1586 +       if(!g) {
1587 +              if (net_ratelimit())
1588 +                       printk(KERN_ERR "layer7: out of memory in hex_print, "
1589 +                                       "bailing.\n");
1590 +              return NULL;
1591 +       }
1592 +
1593 +       for(i = 0; i < strlen(s); i++) {
1594 +               g[i*3    ] = dec2hex(s[i]/16);
1595 +               g[i*3 + 1] = dec2hex(s[i]%16);
1596 +               g[i*3 + 2] = ' ';
1597 +       }
1598 +       g[i*3] = '\0';
1599 +
1600 +       return g;
1601 +}
1602 +#endif // DEBUG
1603 +
1604 +/* Use instead of regcomp.  As we expect to be seeing the same regexps over and
1605 +over again, it make sense to cache the results. */
1606 +static regexp * compile_and_cache(const char * regex_string,
1607 +                                  const char * protocol)
1608 +{
1609 +       struct pattern_cache * node               = first_pattern_cache;
1610 +       struct pattern_cache * last_pattern_cache = first_pattern_cache;
1611 +       struct pattern_cache * tmp;
1612 +       unsigned int len;
1613 +
1614 +       while (node != NULL) {
1615 +               if (!strcmp(node->regex_string, regex_string))
1616 +               return node->pattern;
1617 +
1618 +               last_pattern_cache = node;/* points at the last non-NULL node */
1619 +               node = node->next;
1620 +       }
1621 +
1622 +       /* If we reach the end of the list, then we have not yet cached
1623 +          the pattern for this regex. Let's do that now.
1624 +          Be paranoid about running out of memory to avoid list corruption. */
1625 +       tmp = kmalloc(sizeof(struct pattern_cache), GFP_ATOMIC);
1626 +
1627 +       if(!tmp) {
1628 +               if (net_ratelimit())
1629 +                       printk(KERN_ERR "layer7: out of memory in "
1630 +                                       "compile_and_cache, bailing.\n");
1631 +               return NULL;
1632 +       }
1633 +
1634 +       tmp->regex_string  = kmalloc(strlen(regex_string) + 1, GFP_ATOMIC);
1635 +       tmp->pattern       = kmalloc(sizeof(struct regexp),    GFP_ATOMIC);
1636 +       tmp->next = NULL;
1637 +
1638 +       if(!tmp->regex_string || !tmp->pattern) {
1639 +               if (net_ratelimit())
1640 +                       printk(KERN_ERR "layer7: out of memory in "
1641 +                                       "compile_and_cache, bailing.\n");
1642 +               kfree(tmp->regex_string);
1643 +               kfree(tmp->pattern);
1644 +               kfree(tmp);
1645 +               return NULL;
1646 +       }
1647 +
1648 +       /* Ok.  The new node is all ready now. */
1649 +       node = tmp;
1650 +
1651 +       if(first_pattern_cache == NULL) /* list is empty */
1652 +               first_pattern_cache = node; /* make node the beginning */
1653 +       else
1654 +               last_pattern_cache->next = node; /* attach node to the end */
1655 +
1656 +       /* copy the string and compile the regex */
1657 +       len = strlen(regex_string);
1658 +       DPRINTK("About to compile this: \"%s\"\n", regex_string);
1659 +       node->pattern = regcomp((char *)regex_string, &len);
1660 +       if ( !node->pattern ) {
1661 +               if (net_ratelimit())
1662 +                       printk(KERN_ERR "layer7: Error compiling regexp "
1663 +                                       "\"%s\" (%s)\n",
1664 +                                       regex_string, protocol);
1665 +               /* pattern is now cached as NULL, so we won't try again. */
1666 +       }
1667 +
1668 +       strcpy(node->regex_string, regex_string);
1669 +       return node->pattern;
1670 +}
1671 +
1672 +static int can_handle(const struct sk_buff *skb)
1673 +{
1674 +       if(!ip_hdr(skb)) /* not IP */
1675 +               return 0;
1676 +       if(ip_hdr(skb)->protocol != IPPROTO_TCP &&
1677 +          ip_hdr(skb)->protocol != IPPROTO_UDP &&
1678 +          ip_hdr(skb)->protocol != IPPROTO_ICMP)
1679 +               return 0;
1680 +       return 1;
1681 +}
1682 +
1683 +/* Returns offset the into the skb->data that the application data starts */
1684 +static int app_data_offset(const struct sk_buff *skb)
1685 +{
1686 +       /* In case we are ported somewhere (ebtables?) where ip_hdr(skb)
1687 +       isn't set, this can be gotten from 4*(skb->data[0] & 0x0f) as well. */
1688 +       int ip_hl = 4*ip_hdr(skb)->ihl;
1689 +
1690 +       if( ip_hdr(skb)->protocol == IPPROTO_TCP ) {
1691 +               /* 12 == offset into TCP header for the header length field.
1692 +               Can't get this with skb->h.th->doff because the tcphdr
1693 +               struct doesn't get set when routing (this is confirmed to be
1694 +               true in Netfilter as well as QoS.) */
1695 +               int tcp_hl = 4*(skb->data[ip_hl + 12] >> 4);
1696 +
1697 +               return ip_hl + tcp_hl;
1698 +       } else if( ip_hdr(skb)->protocol == IPPROTO_UDP  ) {
1699 +               return ip_hl + 8; /* UDP header is always 8 bytes */
1700 +       } else if( ip_hdr(skb)->protocol == IPPROTO_ICMP ) {
1701 +               return ip_hl + 8; /* ICMP header is 8 bytes */
1702 +       } else {
1703 +               if (net_ratelimit())
1704 +                       printk(KERN_ERR "layer7: tried to handle unknown "
1705 +                                       "protocol!\n");
1706 +               return ip_hl + 8; /* something reasonable */
1707 +       }
1708 +}
1709 +
1710 +/* handles whether there's a match when we aren't appending data anymore */
1711 +static int match_no_append(struct nf_conn * conntrack,
1712 +                           struct nf_conn * master_conntrack,
1713 +                           enum ip_conntrack_info ctinfo,
1714 +                           enum ip_conntrack_info master_ctinfo,
1715 +                           const struct xt_layer7_info * info)
1716 +{
1717 +       /* If we're in here, throw the app data away */
1718 +       if(master_conntrack->layer7.app_data != NULL) {
1719 +
1720 +       #ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
1721 +               if(!master_conntrack->layer7.app_proto) {
1722 +                       char * f =
1723 +                         friendly_print(master_conntrack->layer7.app_data);
1724 +                       char * g =
1725 +                         hex_print(master_conntrack->layer7.app_data);
1726 +                       DPRINTK("\nl7-filter gave up after %d bytes "
1727 +                               "(%d packets):\n%s\n",
1728 +                               strlen(f), total_acct_packets(master_conntrack), f);
1729 +                       kfree(f);
1730 +                       DPRINTK("In hex: %s\n", g);
1731 +                       kfree(g);
1732 +               }
1733 +       #endif
1734 +
1735 +               kfree(master_conntrack->layer7.app_data);
1736 +               master_conntrack->layer7.app_data = NULL; /* don't free again */
1737 +       }
1738 +
1739 +       if(master_conntrack->layer7.app_proto){
1740 +               /* Here child connections set their .app_proto (for /proc) */
1741 +               if(!conntrack->layer7.app_proto) {
1742 +                       conntrack->layer7.app_proto =
1743 +                         kmalloc(strlen(master_conntrack->layer7.app_proto)+1,
1744 +                           GFP_ATOMIC);
1745 +                       if(!conntrack->layer7.app_proto){
1746 +                               if (net_ratelimit())
1747 +                                       printk(KERN_ERR "layer7: out of memory "
1748 +                                                       "in match_no_append, "
1749 +                                                       "bailing.\n");
1750 +                               return 1;
1751 +                       }
1752 +                       strcpy(conntrack->layer7.app_proto,
1753 +                               master_conntrack->layer7.app_proto);
1754 +               }
1755 +
1756 +               return (!strcmp(master_conntrack->layer7.app_proto,
1757 +                               info->protocol));
1758 +       }
1759 +       else {
1760 +               /* If not classified, set to "unknown" to distinguish from
1761 +               connections that are still being tested. */
1762 +               master_conntrack->layer7.app_proto =
1763 +                       kmalloc(strlen("unknown")+1, GFP_ATOMIC);
1764 +               if(!master_conntrack->layer7.app_proto){
1765 +                       if (net_ratelimit())
1766 +                               printk(KERN_ERR "layer7: out of memory in "
1767 +                                               "match_no_append, bailing.\n");
1768 +                       return 1;
1769 +               }
1770 +               strcpy(master_conntrack->layer7.app_proto, "unknown");
1771 +               return 0;
1772 +       }
1773 +}
1774 +
1775 +/* add the new app data to the conntrack.  Return number of bytes added. */
1776 +static int add_data(struct nf_conn * master_conntrack,
1777 +                    char * app_data, int appdatalen)
1778 +{
1779 +       int length = 0, i;
1780 +       int oldlength = master_conntrack->layer7.app_data_len;
1781 +
1782 +       /* This is a fix for a race condition by Deti Fliegl. However, I'm not
1783 +          clear on whether the race condition exists or whether this really
1784 +          fixes it.  I might just be being dense... Anyway, if it's not really
1785 +          a fix, all it does is waste a very small amount of time. */
1786 +       if(!master_conntrack->layer7.app_data) return 0;
1787 +
1788 +       /* Strip nulls. Make everything lower case (our regex lib doesn't
1789 +       do case insensitivity).  Add it to the end of the current data. */
1790 +       for(i = 0; i < maxdatalen-oldlength-1 &&
1791 +                  i < appdatalen; i++) {
1792 +               if(app_data[i] != '\0') {
1793 +                       /* the kernel version of tolower mungs 'upper ascii' */
1794 +                       master_conntrack->layer7.app_data[length+oldlength] =
1795 +                               isascii(app_data[i])?
1796 +                                       tolower(app_data[i]) : app_data[i];
1797 +                       length++;
1798 +               }
1799 +       }
1800 +
1801 +       master_conntrack->layer7.app_data[length+oldlength] = '\0';
1802 +       master_conntrack->layer7.app_data_len = length + oldlength;
1803 +
1804 +       return length;
1805 +}
1806 +
1807 +/* taken from drivers/video/modedb.c */
1808 +static int my_atoi(const char *s)
1809 +{
1810 +       int val = 0;
1811 +
1812 +       for (;; s++) {
1813 +               switch (*s) {
1814 +                       case '0'...'9':
1815 +                       val = 10*val+(*s-'0');
1816 +                       break;
1817 +               default:
1818 +                       return val;
1819 +               }
1820 +       }
1821 +}
1822 +
1823 +/* write out num_packets to userland. */
1824 +static int layer7_read_proc(char* page, char ** start, off_t off, int count,
1825 +                            int* eof, void * data)
1826 +{
1827 +       if(num_packets > 99 && net_ratelimit())
1828 +               printk(KERN_ERR "layer7: NOT REACHED. num_packets too big\n");
1829 +
1830 +       page[0] = num_packets/10 + '0';
1831 +       page[1] = num_packets%10 + '0';
1832 +       page[2] = '\n';
1833 +       page[3] = '\0';
1834 +
1835 +       *eof=1;
1836 +
1837 +       return 3;
1838 +}
1839 +
1840 +/* Read in num_packets from userland */
1841 +static int layer7_write_proc(struct file* file, const char* buffer,
1842 +                             unsigned long count, void *data)
1843 +{
1844 +       char * foo = kmalloc(count, GFP_ATOMIC);
1845 +
1846 +       if(!foo){
1847 +               if (net_ratelimit())
1848 +                       printk(KERN_ERR "layer7: out of memory, bailing. "
1849 +                                       "num_packets unchanged.\n");
1850 +               return count;
1851 +       }
1852 +
1853 +       if(copy_from_user(foo, buffer, count)) {
1854 +               return -EFAULT;
1855 +       }
1856 +
1857 +
1858 +       num_packets = my_atoi(foo);
1859 +       kfree (foo);
1860 +
1861 +       /* This has an arbitrary limit to make the math easier. I'm lazy.
1862 +       But anyway, 99 is a LOT! If you want more, you're doing it wrong! */
1863 +       if(num_packets > 99) {
1864 +               printk(KERN_WARNING "layer7: num_packets can't be > 99.\n");
1865 +               num_packets = 99;
1866 +       } else if(num_packets < 1) {
1867 +               printk(KERN_WARNING "layer7: num_packets can't be < 1.\n");
1868 +               num_packets = 1;
1869 +       }
1870 +
1871 +       return count;
1872 +}
1873 +
1874 +static bool
1875 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28)
1876 +match(const struct sk_buff *skbin, struct xt_action_param *par)
1877 +#else
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 +      bool *hotdrop)
1886 +#endif
1887 +{
1888 +       /* sidestep const without getting a compiler warning... */
1889 +       struct sk_buff * skb = (struct sk_buff *)skbin;
1890 +
1891 +       const struct xt_layer7_info * info =
1892 +       #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28)
1893 +               par->matchinfo;
1894 +       #else
1895 +               matchinfo;
1896 +       #endif
1897 +
1898 +       enum ip_conntrack_info master_ctinfo, ctinfo;
1899 +       struct nf_conn *master_conntrack, *conntrack;
1900 +       unsigned char * app_data;
1901 +       unsigned int pattern_result, appdatalen;
1902 +       regexp * comppattern;
1903 +
1904 +       /* Be paranoid/incompetent - lock the entire match function. */
1905 +       spin_lock_bh(&l7_lock);
1906 +
1907 +       if(!can_handle(skb)){
1908 +               DPRINTK("layer7: This is some protocol I can't handle.\n");
1909 +               spin_unlock_bh(&l7_lock);
1910 +               return info->invert;
1911 +       }
1912 +
1913 +       /* Treat parent & all its children together as one connection, except
1914 +       for the purpose of setting conntrack->layer7.app_proto in the actual
1915 +       connection. This makes /proc/net/ip_conntrack more satisfying. */
1916 +       if(!(conntrack = nf_ct_get(skb, &ctinfo)) ||
1917 +          !(master_conntrack=nf_ct_get(skb,&master_ctinfo))){
1918 +               DPRINTK("layer7: couldn't get conntrack.\n");
1919 +               spin_unlock_bh(&l7_lock);
1920 +               return info->invert;
1921 +       }
1922 +
1923 +       /* Try to get a master conntrack (and its master etc) for FTP, etc. */
1924 +       while (master_ct(master_conntrack) != NULL)
1925 +               master_conntrack = master_ct(master_conntrack);
1926 +
1927 +       /* if we've classified it or seen too many packets */
1928 +       if(total_acct_packets(master_conntrack) > num_packets ||
1929 +          master_conntrack->layer7.app_proto) {
1930 +
1931 +               pattern_result = match_no_append(conntrack, master_conntrack,
1932 +                                                ctinfo, master_ctinfo, info);
1933 +
1934 +               /* skb->cb[0] == seen. Don't do things twice if there are
1935 +               multiple l7 rules. I'm not sure that using cb for this purpose
1936 +               is correct, even though it says "put your private variables
1937 +               there". But it doesn't look like it is being used for anything
1938 +               else in the skbs that make it here. */
1939 +               skb->cb[0] = 1; /* marking it seen here's probably irrelevant */
1940 +
1941 +               spin_unlock_bh(&l7_lock);
1942 +               return (pattern_result ^ info->invert);
1943 +       }
1944 +
1945 +       if(skb_is_nonlinear(skb)){
1946 +               if(skb_linearize(skb) != 0){
1947 +                       if (net_ratelimit())
1948 +                               printk(KERN_ERR "layer7: failed to linearize "
1949 +                                               "packet, bailing.\n");
1950 +                       spin_unlock_bh(&l7_lock);
1951 +                       return info->invert;
1952 +               }
1953 +       }
1954 +
1955 +       /* now that the skb is linearized, it's safe to set these. */
1956 +       app_data = skb->data + app_data_offset(skb);
1957 +       appdatalen = skb_tail_pointer(skb) - app_data;
1958 +
1959 +       /* the return value gets checked later, when we're ready to use it */
1960 +       comppattern = compile_and_cache(info->pattern, info->protocol);
1961 +
1962 +       /* On the first packet of a connection, allocate space for app data */
1963 +       if(total_acct_packets(master_conntrack) == 1 && !skb->cb[0] &&
1964 +          !master_conntrack->layer7.app_data){
1965 +               master_conntrack->layer7.app_data =
1966 +                       kmalloc(maxdatalen, GFP_ATOMIC);
1967 +               if(!master_conntrack->layer7.app_data){
1968 +                       if (net_ratelimit())
1969 +                               printk(KERN_ERR "layer7: out of memory in "
1970 +                                               "match, bailing.\n");
1971 +                       spin_unlock_bh(&l7_lock);
1972 +                       return info->invert;
1973 +               }
1974 +
1975 +               master_conntrack->layer7.app_data[0] = '\0';
1976 +       }
1977 +
1978 +       /* Can be here, but unallocated, if numpackets is increased near
1979 +       the beginning of a connection */
1980 +       if(master_conntrack->layer7.app_data == NULL){
1981 +               spin_unlock_bh(&l7_lock);
1982 +               return info->invert; /* unmatched */
1983 +       }
1984 +
1985 +       if(!skb->cb[0]){
1986 +               int newbytes;
1987 +               newbytes = add_data(master_conntrack, app_data, appdatalen);
1988 +
1989 +               if(newbytes == 0) { /* didn't add any data */
1990 +                       skb->cb[0] = 1;
1991 +                       /* Didn't match before, not going to match now */
1992 +                       spin_unlock_bh(&l7_lock);
1993 +                       return info->invert;
1994 +               }
1995 +       }
1996 +
1997 +       /* If looking for "unknown", then never match.  "Unknown" means that
1998 +       we've given up; we're still trying with these packets. */
1999 +       if(!strcmp(info->protocol, "unknown")) {
2000 +               pattern_result = 0;
2001 +       /* If looking for "unset", then always match. "Unset" means that we
2002 +       haven't yet classified the connection. */
2003 +       } else if(!strcmp(info->protocol, "unset")) {
2004 +               pattern_result = 2;
2005 +               DPRINTK("layer7: matched unset: not yet classified "
2006 +                       "(%d/%d packets)\n",
2007 +                        total_acct_packets(master_conntrack), num_packets);
2008 +       /* If the regexp failed to compile, don't bother running it */
2009 +       } else if(comppattern &&
2010 +                 regexec(comppattern, master_conntrack->layer7.app_data)){
2011 +               DPRINTK("layer7: matched %s\n", info->protocol);
2012 +               pattern_result = 1;
2013 +       } else pattern_result = 0;
2014 +
2015 +       if(pattern_result == 1) {
2016 +               master_conntrack->layer7.app_proto =
2017 +                       kmalloc(strlen(info->protocol)+1, GFP_ATOMIC);
2018 +               if(!master_conntrack->layer7.app_proto){
2019 +                       if (net_ratelimit())
2020 +                               printk(KERN_ERR "layer7: out of memory in "
2021 +                                               "match, bailing.\n");
2022 +                       spin_unlock_bh(&l7_lock);
2023 +                       return (pattern_result ^ info->invert);
2024 +               }
2025 +               strcpy(master_conntrack->layer7.app_proto, info->protocol);
2026 +       } else if(pattern_result > 1) { /* cleanup from "unset" */
2027 +               pattern_result = 1;
2028 +       }
2029 +
2030 +       /* mark the packet seen */
2031 +       skb->cb[0] = 1;
2032 +
2033 +       spin_unlock_bh(&l7_lock);
2034 +       return (pattern_result ^ info->invert);
2035 +}
2036 +
2037 +// load nf_conntrack_ipv4
2038 +static int check(const struct xt_mtchk_param *par)
2039 +{
2040 +        if (nf_ct_l3proto_try_module_get(par->family) < 0) {
2041 +                pr_info("can't load conntrack support for "
2042 +                                    "proto=%d\n", par->family);
2043 +                return -EINVAL;
2044 +        }
2045 +       return 0;
2046 +}
2047 +
2048 +
2049 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28)
2050 +       static void destroy(const struct xt_mtdtor_param *par)
2051 +       {
2052 +               nf_ct_l3proto_module_put(par->family);
2053 +       }
2054 +#else
2055 +       static void destroy(const struct xt_match *match, void *matchinfo)
2056 +       {
2057 +               nf_ct_l3proto_module_put(match->family);
2058 +       }
2059 +#endif
2060 +
2061 +static struct xt_match xt_layer7_match[] __read_mostly = {
2062 +{
2063 +       .name           = "layer7",
2064 +       .family         = AF_INET,
2065 +       .checkentry     = check,
2066 +       .match          = match,
2067 +       .destroy        = destroy,
2068 +       .matchsize      = sizeof(struct xt_layer7_info),
2069 +       .me             = THIS_MODULE
2070 +}
2071 +};
2072 +
2073 +static void layer7_cleanup_proc(void)
2074 +{
2075 +       remove_proc_entry("layer7_numpackets", init_net.proc_net);
2076 +}
2077 +
2078 +/* register the proc file */
2079 +static void layer7_init_proc(void)
2080 +{
2081 +       struct proc_dir_entry* entry;
2082 +       entry = create_proc_entry("layer7_numpackets", 0644, init_net.proc_net);
2083 +       entry->read_proc = layer7_read_proc;
2084 +       entry->write_proc = layer7_write_proc;
2085 +}
2086 +
2087 +static int __init xt_layer7_init(void)
2088 +{
2089 +       need_conntrack();
2090 +
2091 +       layer7_init_proc();
2092 +       if(maxdatalen < 1) {
2093 +               printk(KERN_WARNING "layer7: maxdatalen can't be < 1, "
2094 +                       "using 1\n");
2095 +               maxdatalen = 1;
2096 +       }
2097 +       /* This is not a hard limit.  It's just here to prevent people from
2098 +       bringing their slow machines to a grinding halt. */
2099 +       else if(maxdatalen > 65536) {
2100 +               printk(KERN_WARNING "layer7: maxdatalen can't be > 65536, "
2101 +                       "using 65536\n");
2102 +               maxdatalen = 65536;
2103 +       }
2104 +       return xt_register_matches(xt_layer7_match,
2105 +                                  ARRAY_SIZE(xt_layer7_match));
2106 +}
2107 +
2108 +static void __exit xt_layer7_fini(void)
2109 +{
2110 +       layer7_cleanup_proc();
2111 +       xt_unregister_matches(xt_layer7_match, ARRAY_SIZE(xt_layer7_match));
2112 +}
2113 +
2114 +module_init(xt_layer7_init);
2115 +module_exit(xt_layer7_fini);