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_core.c linux-5.11/net/netfilter/nf_conntrack_core.c
  87 --- linux-5.11/net/netfilter/nf_conntrack_core.c        2021-02-14 23:32:24.000000000 +0100
  88 +++ linux-5.11/net/netfilter/nf_conntrack_core.c        2021-02-24 13:21:48.335107407 +0100
  89 @@ -528,6 +528,14 @@ static void nf_ct_del_from_dying_or_unco
  90  {
  91         struct ct_pcpu *pcpu;
  92
  93 +       #if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
  94 +       if(ct->layer7.app_proto)
  95 +               kfree(ct->layer7.app_proto);
  96 +       if(ct->layer7.app_data)
  97 +       kfree(ct->layer7.app_data);
  98 +       #endif
  99 +
 100 +
 101         /* We overload first tuple to link into unconfirmed or dying list.*/
 102         pcpu = per_cpu_ptr(nf_ct_net(ct)->ct.pcpu_lists, ct->cpu);
 103
 104 diff -urNp -x '*.orig' linux-5.11/net/netfilter/nf_conntrack_standalone.c linux-5.11/net/netfilter/nf_conntrack_standalone.c
 105 --- linux-5.11/net/netfilter/nf_conntrack_standalone.c  2021-02-14 23:32:24.000000000 +0100
 106 +++ linux-5.11/net/netfilter/nf_conntrack_standalone.c  2021-02-24 13:21:48.338440845 +0100
 107 @@ -366,6 +366,12 @@ static int ct_seq_show(struct seq_file *
 108         ct_show_zone(s, ct, NF_CT_DEFAULT_ZONE_DIR);
 109         ct_show_delta_time(s, ct);
 110
 111 +#if defined(CONFIG_NETFILTER_XT_MATCH_LAYER7) || defined(CONFIG_NETFILTER_XT_MATCH_LAYER7_MODULE)
 112 +       if(ct->layer7.app_proto &&
 113 +           seq_printf(s, "l7proto=%s ", ct->layer7.app_proto))
 114 +               return -ENOSPC;
 115 +#endif
 116 +
 117         seq_printf(s, "use=%u\n", atomic_read(&ct->ct_general.use));
 118
 119         if (seq_has_overflowed(s))
 120 diff -urNp -x '*.orig' linux-5.11/net/netfilter/regexp/regexp.c linux-5.11/net/netfilter/regexp/regexp.c
 121 --- linux-5.11/net/netfilter/regexp/regexp.c    1970-01-01 01:00:00.000000000 +0100
 122 +++ linux-5.11/net/netfilter/regexp/regexp.c    2021-02-24 13:21:48.335107407 +0100
 123 @@ -0,0 +1,1197 @@
 124 +/*
 125 + * regcomp and regexec -- regsub and regerror are elsewhere
 126 + * @(#)regexp.c        1.3 of 18 April 87
 127 + *
 128 + *     Copyright (c) 1986 by University of Toronto.
 129 + *     Written by Henry Spencer.  Not derived from licensed software.
 130 + *
 131 + *     Permission is granted to anyone to use this software for any
 132 + *     purpose on any computer system, and to redistribute it freely,
 133 + *     subject to the following restrictions:
 134 + *
 135 + *     1. The author is not responsible for the consequences of use of
 136 + *             this software, no matter how awful, even if they arise
 137 + *             from defects in it.
 138 + *
 139 + *     2. The origin of this software must not be misrepresented, either
 140 + *             by explicit claim or by omission.
 141 + *
 142 + *     3. Altered versions must be plainly marked as such, and must not
 143 + *             be misrepresented as being the original software.
 144 + *
 145 + * Beware that some of this code is subtly aware of the way operator
 146 + * precedence is structured in regular expressions.  Serious changes in
 147 + * regular-expression syntax might require a total rethink.
 148 + *
 149 + * This code was modified by Ethan Sommer to work within the kernel
 150 + * (it now uses kmalloc etc..)
 151 + *
 152 + * Modified slightly by Matthew Strait to use more modern C.
 153 + */
 154 +
 155 +#include "regexp.h"
 156 +#include "regmagic.h"
 157 +
 158 +/* added by ethan and matt.  Lets it work in both kernel and user space.
 159 +(So iptables can use it, for instance.)  Yea, it goes both ways... */
 160 +#if __KERNEL__
 161 +  #define malloc(foo) kmalloc(foo,GFP_ATOMIC)
 162 +#else
 163 +  #define printk(format,args...) printf(format,##args)
 164 +#endif
 165 +
 166 +void regerror(char * s)
 167 +{
 168 +        printk("<3>Regexp: %s\n", s);
 169 +        /* NOTREACHED */
 170 +}
 171 +
 172 +/*
 173 + * The "internal use only" fields in regexp.h are present to pass info from
 174 + * compile to execute that permits the execute phase to run lots faster on
 175 + * simple cases.  They are:
 176 + *
 177 + * regstart    char that must begin a match; '\0' if none obvious
 178 + * reganch     is the match anchored (at beginning-of-line only)?
 179 + * regmust     string (pointer into program) that match must include, or NULL
 180 + * regmlen     length of regmust string
 181 + *
 182 + * Regstart and reganch permit very fast decisions on suitable starting points
 183 + * for a match, cutting down the work a lot.  Regmust permits fast rejection
 184 + * of lines that cannot possibly match.  The regmust tests are costly enough
 185 + * that regcomp() supplies a regmust only if the r.e. contains something
 186 + * potentially expensive (at present, the only such thing detected is * or +
 187 + * at the start of the r.e., which can involve a lot of backup).  Regmlen is
 188 + * supplied because the test in regexec() needs it and regcomp() is computing
 189 + * it anyway.
 190 + */
 191 +
 192 +/*
 193 + * Structure for regexp "program".  This is essentially a linear encoding
 194 + * of a nondeterministic finite-state machine (aka syntax charts or
 195 + * "railroad normal form" in parsing technology).  Each node is an opcode
 196 + * plus a "next" pointer, possibly plus an operand.  "Next" pointers of
 197 + * all nodes except BRANCH implement concatenation; a "next" pointer with
 198 + * a BRANCH on both ends of it is connecting two alternatives.  (Here we
 199 + * have one of the subtle syntax dependencies:  an individual BRANCH (as
 200 + * opposed to a collection of them) is never concatenated with anything
 201 + * because of operator precedence.)  The operand of some types of node is
 202 + * a literal string; for others, it is a node leading into a sub-FSM.  In
 203 + * particular, the operand of a BRANCH node is the first node of the branch.
 204 + * (NB this is *not* a tree structure:  the tail of the branch connects
 205 + * to the thing following the set of BRANCHes.)  The opcodes are:
 206 + */
 207 +
 208 +/* definition  number  opnd?   meaning */
 209 +#define        END     0       /* no   End of program. */
 210 +#define        BOL     1       /* no   Match "" at beginning of line. */
 211 +#define        EOL     2       /* no   Match "" at end of line. */
 212 +#define        ANY     3       /* no   Match any one character. */
 213 +#define        ANYOF   4       /* str  Match any character in this string. */
 214 +#define        ANYBUT  5       /* str  Match any character not in this string. */
 215 +#define        BRANCH  6       /* node Match this alternative, or the next... */
 216 +#define        BACK    7       /* no   Match "", "next" ptr points backward. */
 217 +#define        EXACTLY 8       /* str  Match this string. */
 218 +#define        NOTHING 9       /* no   Match empty string. */
 219 +#define        STAR    10      /* node Match this (simple) thing 0 or more times. */
 220 +#define        PLUS    11      /* node Match this (simple) thing 1 or more times. */
 221 +#define        OPEN    20      /* no   Mark this point in input as start of #n. */
 222 +                       /*      OPEN+1 is number 1, etc. */
 223 +#define        CLOSE   30      /* no   Analogous to OPEN. */
 224 +
 225 +/*
 226 + * Opcode notes:
 227 + *
 228 + * BRANCH      The set of branches constituting a single choice are hooked
 229 + *             together with their "next" pointers, since precedence prevents
 230 + *             anything being concatenated to any individual branch.  The
 231 + *             "next" pointer of the last BRANCH in a choice points to the
 232 + *             thing following the whole choice.  This is also where the
 233 + *             final "next" pointer of each individual branch points; each
 234 + *             branch starts with the operand node of a BRANCH node.
 235 + *
 236 + * BACK                Normal "next" pointers all implicitly point forward; BACK
 237 + *             exists to make loop structures possible.
 238 + *
 239 + * STAR,PLUS   '?', and complex '*' and '+', are implemented as circular
 240 + *             BRANCH structures using BACK.  Simple cases (one character
 241 + *             per match) are implemented with STAR and PLUS for speed
 242 + *             and to minimize recursive plunges.
 243 + *
 244 + * OPEN,CLOSE  ...are numbered at compile time.
 245 + */
 246 +
 247 +/*
 248 + * A node is one char of opcode followed by two chars of "next" pointer.
 249 + * "Next" pointers are stored as two 8-bit pieces, high order first.  The
 250 + * value is a positive offset from the opcode of the node containing it.
 251 + * An operand, if any, simply follows the node.  (Note that much of the
 252 + * code generation knows about this implicit relationship.)
 253 + *
 254 + * Using two bytes for the "next" pointer is vast overkill for most things,
 255 + * but allows patterns to get big without disasters.
 256 + */
 257 +#define        OP(p)   (*(p))
 258 +#define        NEXT(p) (((*((p)+1)&0377)<<8) + (*((p)+2)&0377))
 259 +#define        OPERAND(p)      ((p) + 3)
 260 +
 261 +/*
 262 + * See regmagic.h for one further detail of program structure.
 263 + */
 264 +
 265 +
 266 +/*
 267 + * Utility definitions.
 268 + */
 269 +#ifndef CHARBITS
 270 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
 271 +#else
 272 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
 273 +#endif
 274 +
 275 +#define        FAIL(m) { regerror(m); return(NULL); }
 276 +#define        ISMULT(c)       ((c) == '*' || (c) == '+' || (c) == '?')
 277 +#define        META    "^$.[()|?+*\\"
 278 +
 279 +/*
 280 + * Flags to be passed up and down.
 281 + */
 282 +#define        HASWIDTH        01      /* Known never to match null string. */
 283 +#define        SIMPLE          02      /* Simple enough to be STAR/PLUS operand. */
 284 +#define        SPSTART         04      /* Starts with * or +. */
 285 +#define        WORST           0       /* Worst case. */
 286 +
 287 +/*
 288 + * Global work variables for regcomp().
 289 + */
 290 +struct match_globals {
 291 +char *reginput;                /* String-input pointer. */
 292 +char *regbol;          /* Beginning of input, for ^ check. */
 293 +char **regstartp;      /* Pointer to startp array. */
 294 +char **regendp;                /* Ditto for endp. */
 295 +char *regparse;                /* Input-scan pointer. */
 296 +int regnpar;           /* () count. */
 297 +char regdummy;
 298 +char *regcode;         /* Code-emit pointer; &regdummy = don't. */
 299 +long regsize;          /* Code size. */
 300 +};
 301 +
 302 +/*
 303 + * Forward declarations for regcomp()'s friends.
 304 + */
 305 +#ifndef STATIC
 306 +#define        STATIC  static
 307 +#endif
 308 +STATIC char *reg(struct match_globals *g, int paren,int *flagp);
 309 +STATIC char *regbranch(struct match_globals *g, int *flagp);
 310 +STATIC char *regpiece(struct match_globals *g, int *flagp);
 311 +STATIC char *regatom(struct match_globals *g, int *flagp);
 312 +STATIC char *regnode(struct match_globals *g, char op);
 313 +STATIC char *regnext(struct match_globals *g, char *p);
 314 +STATIC void regc(struct match_globals *g, char b);
 315 +STATIC void reginsert(struct match_globals *g, char op, char *opnd);
 316 +STATIC void regtail(struct match_globals *g, char *p, char *val);
 317 +STATIC void regoptail(struct match_globals *g, char *p, char *val);
 318 +
 319 +
 320 +__kernel_size_t my_strcspn(const char *s1,const char *s2)
 321 +{
 322 +        char *scan1;
 323 +        char *scan2;
 324 +        int count;
 325 +
 326 +        count = 0;
 327 +        for (scan1 = (char *)s1; *scan1 != '\0'; scan1++) {
 328 +                for (scan2 = (char *)s2; *scan2 != '\0';)       /* ++ moved down. */
 329 +                        if (*scan1 == *scan2++)
 330 +                                return(count);
 331 +                count++;
 332 +        }
 333 +        return(count);
 334 +}
 335 +
 336 +/*
 337 + - regcomp - compile a regular expression into internal code
 338 + *
 339 + * We can't allocate space until we know how big the compiled form will be,
 340 + * but we can't compile it (and thus know how big it is) until we've got a
 341 + * place to put the code.  So we cheat:  we compile it twice, once with code
 342 + * generation turned off and size counting turned on, and once "for real".
 343 + * This also means that we don't allocate space until we are sure that the
 344 + * thing really will compile successfully, and we never have to move the
 345 + * code and thus invalidate pointers into it.  (Note that it has to be in
 346 + * one piece because free() must be able to free it all.)
 347 + *
 348 + * Beware that the optimization-preparation code in here knows about some
 349 + * of the structure of the compiled regexp.
 350 + */
 351 +regexp *
 352 +regcomp(char *exp,int *patternsize)
 353 +{
 354 +       register regexp *r;
 355 +       register char *scan;
 356 +       register char *longest;
 357 +       register int len;
 358 +       int flags;
 359 +       struct match_globals g;
 360 +
 361 +       /* commented out by ethan
 362 +          extern char *malloc();
 363 +       */
 364 +
 365 +       if (exp == NULL)
 366 +               FAIL("NULL argument");
 367 +
 368 +       /* First pass: determine size, legality. */
 369 +       g.regparse = exp;
 370 +       g.regnpar = 1;
 371 +       g.regsize = 0L;
 372 +       g.regcode = &g.regdummy;
 373 +       regc(&g, MAGIC);
 374 +       if (reg(&g, 0, &flags) == NULL)
 375 +               return(NULL);
 376 +
 377 +       /* Small enough for pointer-storage convention? */
 378 +       if (g.regsize >= 32767L)                /* Probably could be 65535L. */
 379 +               FAIL("regexp too big");
 380 +
 381 +       /* Allocate space. */
 382 +       *patternsize=sizeof(regexp) + (unsigned)g.regsize;
 383 +       r = (regexp *)malloc(sizeof(regexp) + (unsigned)g.regsize);
 384 +       if (r == NULL)
 385 +               FAIL("out of space");
 386 +
 387 +       /* Second pass: emit code. */
 388 +       g.regparse = exp;
 389 +       g.regnpar = 1;
 390 +       g.regcode = r->program;
 391 +       regc(&g, MAGIC);
 392 +       if (reg(&g, 0, &flags) == NULL)
 393 +               return(NULL);
 394 +
 395 +       /* Dig out information for optimizations. */
 396 +       r->regstart = '\0';     /* Worst-case defaults. */
 397 +       r->reganch = 0;
 398 +       r->regmust = NULL;
 399 +       r->regmlen = 0;
 400 +       scan = r->program+1;                    /* First BRANCH. */
 401 +       if (OP(regnext(&g, scan)) == END) {             /* Only one top-level choice. */
 402 +               scan = OPERAND(scan);
 403 +
 404 +               /* Starting-point info. */
 405 +               if (OP(scan) == EXACTLY)
 406 +                       r->regstart = *OPERAND(scan);
 407 +               else if (OP(scan) == BOL)
 408 +                       r->reganch++;
 409 +
 410 +               /*
 411 +                * If there's something expensive in the r.e., find the
 412 +                * longest literal string that must appear and make it the
 413 +                * regmust.  Resolve ties in favor of later strings, since
 414 +                * the regstart check works with the beginning of the r.e.
 415 +                * and avoiding duplication strengthens checking.  Not a
 416 +                * strong reason, but sufficient in the absence of others.
 417 +                */
 418 +               if (flags&SPSTART) {
 419 +                       longest = NULL;
 420 +                       len = 0;
 421 +                       for (; scan != NULL; scan = regnext(&g, scan))
 422 +                               if (OP(scan) == EXACTLY && strlen(OPERAND(scan)) >= len) {
 423 +                                       longest = OPERAND(scan);
 424 +                                       len = strlen(OPERAND(scan));
 425 +                               }
 426 +                       r->regmust = longest;
 427 +                       r->regmlen = len;
 428 +               }
 429 +       }
 430 +
 431 +       return(r);
 432 +}
 433 +
 434 +/*
 435 + - reg - regular expression, i.e. main body or parenthesized thing
 436 + *
 437 + * Caller must absorb opening parenthesis.
 438 + *
 439 + * Combining parenthesis handling with the base level of regular expression
 440 + * is a trifle forced, but the need to tie the tails of the branches to what
 441 + * follows makes it hard to avoid.
 442 + */
 443 +static char *
 444 +reg(struct match_globals *g, int paren, int *flagp /* Parenthesized? */ )
 445 +{
 446 +       register char *ret;
 447 +       register char *br;
 448 +       register char *ender;
 449 +       register int parno = 0; /* 0 makes gcc happy */
 450 +       int flags;
 451 +
 452 +       *flagp = HASWIDTH;      /* Tentatively. */
 453 +
 454 +       /* Make an OPEN node, if parenthesized. */
 455 +       if (paren) {
 456 +               if (g->regnpar >= NSUBEXP)
 457 +                       FAIL("too many ()");
 458 +               parno = g->regnpar;
 459 +               g->regnpar++;
 460 +               ret = regnode(g, OPEN+parno);
 461 +       } else
 462 +               ret = NULL;
 463 +
 464 +       /* Pick up the branches, linking them together. */
 465 +       br = regbranch(g, &flags);
 466 +       if (br == NULL)
 467 +               return(NULL);
 468 +       if (ret != NULL)
 469 +               regtail(g, ret, br);    /* OPEN -> first. */
 470 +       else
 471 +               ret = br;
 472 +       if (!(flags&HASWIDTH))
 473 +               *flagp &= ~HASWIDTH;
 474 +       *flagp |= flags&SPSTART;
 475 +       while (*g->regparse == '|') {
 476 +               g->regparse++;
 477 +               br = regbranch(g, &flags);
 478 +               if (br == NULL)
 479 +                       return(NULL);
 480 +               regtail(g, ret, br);    /* BRANCH -> BRANCH. */
 481 +               if (!(flags&HASWIDTH))
 482 +                       *flagp &= ~HASWIDTH;
 483 +               *flagp |= flags&SPSTART;
 484 +       }
 485 +
 486 +       /* Make a closing node, and hook it on the end. */
 487 +       ender = regnode(g, (paren) ? CLOSE+parno : END);
 488 +       regtail(g, ret, ender);
 489 +
 490 +       /* Hook the tails of the branches to the closing node. */
 491 +       for (br = ret; br != NULL; br = regnext(g, br))
 492 +               regoptail(g, br, ender);
 493 +
 494 +       /* Check for proper termination. */
 495 +       if (paren && *g->regparse++ != ')') {
 496 +               FAIL("unmatched ()");
 497 +       } else if (!paren && *g->regparse != '\0') {
 498 +               if (*g->regparse == ')') {
 499 +                       FAIL("unmatched ()");
 500 +               } else
 501 +                       FAIL("junk on end");    /* "Can't happen". */
 502 +               /* NOTREACHED */
 503 +       }
 504 +
 505 +       return(ret);
 506 +}
 507 +
 508 +/*
 509 + - regbranch - one alternative of an | operator
 510 + *
 511 + * Implements the concatenation operator.
 512 + */
 513 +static char *
 514 +regbranch(struct match_globals *g, int *flagp)
 515 +{
 516 +       register char *ret;
 517 +       register char *chain;
 518 +       register char *latest;
 519 +       int flags;
 520 +
 521 +       *flagp = WORST;         /* Tentatively. */
 522 +
 523 +       ret = regnode(g, BRANCH);
 524 +       chain = NULL;
 525 +       while (*g->regparse != '\0' && *g->regparse != '|' && *g->regparse != ')') {
 526 +               latest = regpiece(g, &flags);
 527 +               if (latest == NULL)
 528 +                       return(NULL);
 529 +               *flagp |= flags&HASWIDTH;
 530 +               if (chain == NULL)      /* First piece. */
 531 +                       *flagp |= flags&SPSTART;
 532 +               else
 533 +                       regtail(g, chain, latest);
 534 +               chain = latest;
 535 +       }
 536 +       if (chain == NULL)      /* Loop ran zero times. */
 537 +               (void) regnode(g, NOTHING);
 538 +
 539 +       return(ret);
 540 +}
 541 +
 542 +/*
 543 + - regpiece - something followed by possible [*+?]
 544 + *
 545 + * Note that the branching code sequences used for ? and the general cases
 546 + * of * and + are somewhat optimized:  they use the same NOTHING node as
 547 + * both the endmarker for their branch list and the body of the last branch.
 548 + * It might seem that this node could be dispensed with entirely, but the
 549 + * endmarker role is not redundant.
 550 + */
 551 +static char *
 552 +regpiece(struct match_globals *g, int *flagp)
 553 +{
 554 +       register char *ret;
 555 +       register char op;
 556 +       register char *next;
 557 +       int flags;
 558 +
 559 +       ret = regatom(g, &flags);
 560 +       if (ret == NULL)
 561 +               return(NULL);
 562 +
 563 +       op = *g->regparse;
 564 +       if (!ISMULT(op)) {
 565 +               *flagp = flags;
 566 +               return(ret);
 567 +       }
 568 +
 569 +       if (!(flags&HASWIDTH) && op != '?')
 570 +               FAIL("*+ operand could be empty");
 571 +       *flagp = (op != '+') ? (WORST|SPSTART) : (WORST|HASWIDTH);
 572 +
 573 +       if (op == '*' && (flags&SIMPLE))
 574 +               reginsert(g, STAR, ret);
 575 +       else if (op == '*') {
 576 +               /* Emit x* as (x&|), where & means "self". */
 577 +               reginsert(g, BRANCH, ret);                      /* Either x */
 578 +               regoptail(g, ret, regnode(g, BACK));            /* and loop */
 579 +               regoptail(g, ret, ret);                 /* back */
 580 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
 581 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
 582 +       } else if (op == '+' && (flags&SIMPLE))
 583 +               reginsert(g, PLUS, ret);
 584 +       else if (op == '+') {
 585 +               /* Emit x+ as x(&|), where & means "self". */
 586 +               next = regnode(g, BRANCH);                      /* Either */
 587 +               regtail(g, ret, next);
 588 +               regtail(g, regnode(g, BACK), ret);              /* loop back */
 589 +               regtail(g, next, regnode(g, BRANCH));           /* or */
 590 +               regtail(g, ret, regnode(g, NOTHING));           /* null. */
 591 +       } else if (op == '?') {
 592 +               /* Emit x? as (x|) */
 593 +               reginsert(g, BRANCH, ret);                      /* Either x */
 594 +               regtail(g, ret, regnode(g, BRANCH));            /* or */
 595 +               next = regnode(g, NOTHING);             /* null. */
 596 +               regtail(g, ret, next);
 597 +               regoptail(g, ret, next);
 598 +       }
 599 +       g->regparse++;
 600 +       if (ISMULT(*g->regparse))
 601 +               FAIL("nested *?+");
 602 +
 603 +       return(ret);
 604 +}
 605 +
 606 +/*
 607 + - regatom - the lowest level
 608 + *
 609 + * Optimization:  gobbles an entire sequence of ordinary characters so that
 610 + * it can turn them into a single node, which is smaller to store and
 611 + * faster to run.  Backslashed characters are exceptions, each becoming a
 612 + * separate node; the code is simpler that way and it's not worth fixing.
 613 + */
 614 +static char *
 615 +regatom(struct match_globals *g, int *flagp)
 616 +{
 617 +       register char *ret;
 618 +       int flags;
 619 +
 620 +       *flagp = WORST;         /* Tentatively. */
 621 +
 622 +       switch (*g->regparse++) {
 623 +       case '^':
 624 +               ret = regnode(g, BOL);
 625 +               break;
 626 +       case '$':
 627 +               ret = regnode(g, EOL);
 628 +               break;
 629 +       case '.':
 630 +               ret = regnode(g, ANY);
 631 +               *flagp |= HASWIDTH|SIMPLE;
 632 +               break;
 633 +       case '[': {
 634 +                       register int class;
 635 +                       register int classend;
 636 +
 637 +                       if (*g->regparse == '^') {      /* Complement of range. */
 638 +                               ret = regnode(g, ANYBUT);
 639 +                               g->regparse++;
 640 +                       } else
 641 +                               ret = regnode(g, ANYOF);
 642 +                       if (*g->regparse == ']' || *g->regparse == '-')
 643 +                               regc(g, *g->regparse++);
 644 +                       while (*g->regparse != '\0' && *g->regparse != ']') {
 645 +                               if (*g->regparse == '-') {
 646 +                                       g->regparse++;
 647 +                                       if (*g->regparse == ']' || *g->regparse == '\0')
 648 +                                               regc(g, '-');
 649 +                                       else {
 650 +                                               class = UCHARAT(g->regparse-2)+1;
 651 +                                               classend = UCHARAT(g->regparse);
 652 +                                               if (class > classend+1)
 653 +                                                       FAIL("invalid [] range");
 654 +                                               for (; class <= classend; class++)
 655 +                                                       regc(g, class);
 656 +                                               g->regparse++;
 657 +                                       }
 658 +                               } else
 659 +                                       regc(g, *g->regparse++);
 660 +                       }
 661 +                       regc(g, '\0');
 662 +                       if (*g->regparse != ']')
 663 +                               FAIL("unmatched []");
 664 +                       g->regparse++;
 665 +                       *flagp |= HASWIDTH|SIMPLE;
 666 +               }
 667 +               break;
 668 +       case '(':
 669 +               ret = reg(g, 1, &flags);
 670 +               if (ret == NULL)
 671 +                       return(NULL);
 672 +               *flagp |= flags&(HASWIDTH|SPSTART);
 673 +               break;
 674 +       case '\0':
 675 +       case '|':
 676 +       case ')':
 677 +               FAIL("internal urp");   /* Supposed to be caught earlier. */
 678 +               break;
 679 +       case '?':
 680 +       case '+':
 681 +       case '*':
 682 +               FAIL("?+* follows nothing");
 683 +               break;
 684 +       case '\\':
 685 +               if (*g->regparse == '\0')
 686 +                       FAIL("trailing \\");
 687 +               ret = regnode(g, EXACTLY);
 688 +               regc(g, *g->regparse++);
 689 +               regc(g, '\0');
 690 +               *flagp |= HASWIDTH|SIMPLE;
 691 +               break;
 692 +       default: {
 693 +                       register int len;
 694 +                       register char ender;
 695 +
 696 +                       g->regparse--;
 697 +                       len = my_strcspn((const char *)g->regparse, (const char *)META);
 698 +                       if (len <= 0)
 699 +                               FAIL("internal disaster");
 700 +                       ender = *(g->regparse+len);
 701 +                       if (len > 1 && ISMULT(ender))
 702 +                               len--;          /* Back off clear of ?+* operand. */
 703 +                       *flagp |= HASWIDTH;
 704 +                       if (len == 1)
 705 +                               *flagp |= SIMPLE;
 706 +                       ret = regnode(g, EXACTLY);
 707 +                       while (len > 0) {
 708 +                               regc(g, *g->regparse++);
 709 +                               len--;
 710 +                       }
 711 +                       regc(g, '\0');
 712 +               }
 713 +               break;
 714 +       }
 715 +
 716 +       return(ret);
 717 +}
 718 +
 719 +/*
 720 + - regnode - emit a node
 721 + */
 722 +static char *                  /* Location. */
 723 +regnode(struct match_globals *g, char op)
 724 +{
 725 +       register char *ret;
 726 +       register char *ptr;
 727 +
 728 +       ret = g->regcode;
 729 +       if (ret == &g->regdummy) {
 730 +               g->regsize += 3;
 731 +               return(ret);
 732 +       }
 733 +
 734 +       ptr = ret;
 735 +       *ptr++ = op;
 736 +       *ptr++ = '\0';          /* Null "next" pointer. */
 737 +       *ptr++ = '\0';
 738 +       g->regcode = ptr;
 739 +
 740 +       return(ret);
 741 +}
 742 +
 743 +/*
 744 + - regc - emit (if appropriate) a byte of code
 745 + */
 746 +static void
 747 +regc(struct match_globals *g, char b)
 748 +{
 749 +       if (g->regcode != &g->regdummy)
 750 +               *g->regcode++ = b;
 751 +       else
 752 +               g->regsize++;
 753 +}
 754 +
 755 +/*
 756 + - reginsert - insert an operator in front of already-emitted operand
 757 + *
 758 + * Means relocating the operand.
 759 + */
 760 +static void
 761 +reginsert(struct match_globals *g, char op, char* opnd)
 762 +{
 763 +       register char *src;
 764 +       register char *dst;
 765 +       register char *place;
 766 +
 767 +       if (g->regcode == &g->regdummy) {
 768 +               g->regsize += 3;
 769 +               return;
 770 +       }
 771 +
 772 +       src = g->regcode;
 773 +       g->regcode += 3;
 774 +       dst = g->regcode;
 775 +       while (src > opnd)
 776 +               *--dst = *--src;
 777 +
 778 +       place = opnd;           /* Op node, where operand used to be. */
 779 +       *place++ = op;
 780 +       *place++ = '\0';
 781 +       *place++ = '\0';
 782 +}
 783 +
 784 +/*
 785 + - regtail - set the next-pointer at the end of a node chain
 786 + */
 787 +static void
 788 +regtail(struct match_globals *g, char *p, char *val)
 789 +{
 790 +       register char *scan;
 791 +       register char *temp;
 792 +       register int offset;
 793 +
 794 +       if (p == &g->regdummy)
 795 +               return;
 796 +
 797 +       /* Find last node. */
 798 +       scan = p;
 799 +       for (;;) {
 800 +               temp = regnext(g, scan);
 801 +               if (temp == NULL)
 802 +                       break;
 803 +               scan = temp;
 804 +       }
 805 +
 806 +       if (OP(scan) == BACK)
 807 +               offset = scan - val;
 808 +       else
 809 +               offset = val - scan;
 810 +       *(scan+1) = (offset>>8)&0377;
 811 +       *(scan+2) = offset&0377;
 812 +}
 813 +
 814 +/*
 815 + - regoptail - regtail on operand of first argument; nop if operandless
 816 + */
 817 +static void
 818 +regoptail(struct match_globals *g, char *p, char *val)
 819 +{
 820 +       /* "Operandless" and "op != BRANCH" are synonymous in practice. */
 821 +       if (p == NULL || p == &g->regdummy || OP(p) != BRANCH)
 822 +               return;
 823 +       regtail(g, OPERAND(p), val);
 824 +}
 825 +
 826 +/*
 827 + * regexec and friends
 828 + */
 829 +
 830 +
 831 +/*
 832 + * Forwards.
 833 + */
 834 +STATIC int regtry(struct match_globals *g, regexp *prog, char *string);
 835 +STATIC int regmatch(struct match_globals *g, char *prog);
 836 +STATIC int regrepeat(struct match_globals *g, char *p);
 837 +
 838 +#ifdef DEBUG
 839 +int regnarrate = 0;
 840 +void regdump();
 841 +STATIC char *regprop(char *op);
 842 +#endif
 843 +
 844 +/*
 845 + - regexec - match a regexp against a string
 846 + */
 847 +int
 848 +regexec(regexp *prog, char *string)
 849 +{
 850 +       register char *s;
 851 +       struct match_globals g;
 852 +
 853 +       /* Be paranoid... */
 854 +       if (prog == NULL || string == NULL) {
 855 +               printk("<3>Regexp: NULL parameter\n");
 856 +               return(0);
 857 +       }
 858 +
 859 +       /* Check validity of program. */
 860 +       if (UCHARAT(prog->program) != MAGIC) {
 861 +               printk("<3>Regexp: corrupted program\n");
 862 +               return(0);
 863 +       }
 864 +
 865 +       /* If there is a "must appear" string, look for it. */
 866 +       if (prog->regmust != NULL) {
 867 +               s = string;
 868 +               while ((s = strchr(s, prog->regmust[0])) != NULL) {
 869 +                       if (strncmp(s, prog->regmust, prog->regmlen) == 0)
 870 +                               break;  /* Found it. */
 871 +                       s++;
 872 +               }
 873 +               if (s == NULL)  /* Not present. */
 874 +                       return(0);
 875 +       }
 876 +
 877 +       /* Mark beginning of line for ^ . */
 878 +       g.regbol = string;
 879 +
 880 +       /* Simplest case:  anchored match need be tried only once. */
 881 +       if (prog->reganch)
 882 +               return(regtry(&g, prog, string));
 883 +
 884 +       /* Messy cases:  unanchored match. */
 885 +       s = string;
 886 +       if (prog->regstart != '\0')
 887 +               /* We know what char it must start with. */
 888 +               while ((s = strchr(s, prog->regstart)) != NULL) {
 889 +                       if (regtry(&g, prog, s))
 890 +                               return(1);
 891 +                       s++;
 892 +               }
 893 +       else
 894 +               /* We don't -- general case. */
 895 +               do {
 896 +                       if (regtry(&g, prog, s))
 897 +                               return(1);
 898 +               } while (*s++ != '\0');
 899 +
 900 +       /* Failure. */
 901 +       return(0);
 902 +}
 903 +
 904 +/*
 905 + - regtry - try match at specific point
 906 + */
 907 +static int                     /* 0 failure, 1 success */
 908 +regtry(struct match_globals *g, regexp *prog, char *string)
 909 +{
 910 +       register int i;
 911 +       register char **sp;
 912 +       register char **ep;
 913 +
 914 +       g->reginput = string;
 915 +       g->regstartp = prog->startp;
 916 +       g->regendp = prog->endp;
 917 +
 918 +       sp = prog->startp;
 919 +       ep = prog->endp;
 920 +       for (i = NSUBEXP; i > 0; i--) {
 921 +               *sp++ = NULL;
 922 +               *ep++ = NULL;
 923 +       }
 924 +       if (regmatch(g, prog->program + 1)) {
 925 +               prog->startp[0] = string;
 926 +               prog->endp[0] = g->reginput;
 927 +               return(1);
 928 +       } else
 929 +               return(0);
 930 +}
 931 +
 932 +/*
 933 + - regmatch - main matching routine
 934 + *
 935 + * Conceptually the strategy is simple:  check to see whether the current
 936 + * node matches, call self recursively to see whether the rest matches,
 937 + * and then act accordingly.  In practice we make some effort to avoid
 938 + * recursion, in particular by going through "ordinary" nodes (that don't
 939 + * need to know whether the rest of the match failed) by a loop instead of
 940 + * by recursion.
 941 + */
 942 +static int                     /* 0 failure, 1 success */
 943 +regmatch(struct match_globals *g, char *prog)
 944 +{
 945 +       register char *scan = prog; /* Current node. */
 946 +       char *next;                 /* Next node. */
 947 +
 948 +#ifdef DEBUG
 949 +       if (scan != NULL && regnarrate)
 950 +               fprintf(stderr, "%s(\n", regprop(scan));
 951 +#endif
 952 +       while (scan != NULL) {
 953 +#ifdef DEBUG
 954 +               if (regnarrate)
 955 +                       fprintf(stderr, "%s...\n", regprop(scan));
 956 +#endif
 957 +               next = regnext(g, scan);
 958 +
 959 +               switch (OP(scan)) {
 960 +               case BOL:
 961 +                       if (g->reginput != g->regbol)
 962 +                               return(0);
 963 +                       break;
 964 +               case EOL:
 965 +                       if (*g->reginput != '\0')
 966 +                               return(0);
 967 +                       break;
 968 +               case ANY:
 969 +                       if (*g->reginput == '\0')
 970 +                               return(0);
 971 +                       g->reginput++;
 972 +                       break;
 973 +               case EXACTLY: {
 974 +                               register int len;
 975 +                               register char *opnd;
 976 +
 977 +                               opnd = OPERAND(scan);
 978 +                               /* Inline the first character, for speed. */
 979 +                               if (*opnd != *g->reginput)
 980 +                                       return(0);
 981 +                               len = strlen(opnd);
 982 +                               if (len > 1 && strncmp(opnd, g->reginput, len) != 0)
 983 +                                       return(0);
 984 +                               g->reginput += len;
 985 +                       }
 986 +                       break;
 987 +               case ANYOF:
 988 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) == NULL)
 989 +                               return(0);
 990 +                       g->reginput++;
 991 +                       break;
 992 +               case ANYBUT:
 993 +                       if (*g->reginput == '\0' || strchr(OPERAND(scan), *g->reginput) != NULL)
 994 +                               return(0);
 995 +                       g->reginput++;
 996 +                       break;
 997 +               case NOTHING:
 998 +               case BACK:
 999 +                       break;
1000 +               case OPEN+1:
1001 +               case OPEN+2:
1002 +               case OPEN+3:
1003 +               case OPEN+4:
1004 +               case OPEN+5:
1005 +               case OPEN+6:
1006 +               case OPEN+7:
1007 +               case OPEN+8:
1008 +               case OPEN+9: {
1009 +                               register int no;
1010 +                               register char *save;
1011 +
1012 +                               no = OP(scan) - OPEN;
1013 +                               save = g->reginput;
1014 +
1015 +                               if (regmatch(g, next)) {
1016 +                                       /*
1017 +                                        * Don't set startp if some later
1018 +                                        * invocation of the same parentheses
1019 +                                        * already has.
1020 +                                        */
1021 +                                       if (g->regstartp[no] == NULL)
1022 +                                               g->regstartp[no] = save;
1023 +                                       return(1);
1024 +                               } else
1025 +                                       return(0);
1026 +                       }
1027 +                       break;
1028 +               case CLOSE+1:
1029 +               case CLOSE+2:
1030 +               case CLOSE+3:
1031 +               case CLOSE+4:
1032 +               case CLOSE+5:
1033 +               case CLOSE+6:
1034 +               case CLOSE+7:
1035 +               case CLOSE+8:
1036 +               case CLOSE+9:
1037 +                       {
1038 +                               register int no;
1039 +                               register char *save;
1040 +
1041 +                               no = OP(scan) - CLOSE;
1042 +                               save = g->reginput;
1043 +
1044 +                               if (regmatch(g, next)) {
1045 +                                       /*
1046 +                                        * Don't set endp if some later
1047 +                                        * invocation of the same parentheses
1048 +                                        * already has.
1049 +                                        */
1050 +                                       if (g->regendp[no] == NULL)
1051 +                                               g->regendp[no] = save;
1052 +                                       return(1);
1053 +                               } else
1054 +                                       return(0);
1055 +                       }
1056 +                       break;
1057 +               case BRANCH: {
1058 +                               register char *save;
1059 +
1060 +                               if (OP(next) != BRANCH)         /* No choice. */
1061 +                                       next = OPERAND(scan);   /* Avoid recursion. */
1062 +                               else {
1063 +                                       do {
1064 +                                               save = g->reginput;
1065 +                                               if (regmatch(g, OPERAND(scan)))
1066 +                                                       return(1);
1067 +                                               g->reginput = save;
1068 +                                               scan = regnext(g, scan);
1069 +                                       } while (scan != NULL && OP(scan) == BRANCH);
1070 +                                       return(0);
1071 +                                       /* NOTREACHED */
1072 +                               }
1073 +                       }
1074 +                       break;
1075 +               case STAR:
1076 +               case PLUS: {
1077 +                               register char nextch;
1078 +                               register int no;
1079 +                               register char *save;
1080 +                               register int min;
1081 +
1082 +                               /*
1083 +                                * Lookahead to avoid useless match attempts
1084 +                                * when we know what character comes next.
1085 +                                */
1086 +                               nextch = '\0';
1087 +                               if (OP(next) == EXACTLY)
1088 +                                       nextch = *OPERAND(next);
1089 +                               min = (OP(scan) == STAR) ? 0 : 1;
1090 +                               save = g->reginput;
1091 +                               no = regrepeat(g, OPERAND(scan));
1092 +                               while (no >= min) {
1093 +                                       /* If it could work, try it. */
1094 +                                       if (nextch == '\0' || *g->reginput == nextch)
1095 +                                               if (regmatch(g, next))
1096 +                                                       return(1);
1097 +                                       /* Couldn't or didn't -- back up. */
1098 +                                       no--;
1099 +                                       g->reginput = save + no;
1100 +                               }
1101 +                               return(0);
1102 +                       }
1103 +                       break;
1104 +               case END:
1105 +                       return(1);      /* Success! */
1106 +                       break;
1107 +               default:
1108 +                       printk("<3>Regexp: memory corruption\n");
1109 +                       return(0);
1110 +                       break;
1111 +               }
1112 +
1113 +               scan = next;
1114 +       }
1115 +
1116 +       /*
1117 +        * We get here only if there's trouble -- normally "case END" is
1118 +        * the terminating point.
1119 +        */
1120 +       printk("<3>Regexp: corrupted pointers\n");
1121 +       return(0);
1122 +}
1123 +
1124 +/*
1125 + - regrepeat - repeatedly match something simple, report how many
1126 + */
1127 +static int
1128 +regrepeat(struct match_globals *g, char *p)
1129 +{
1130 +       register int count = 0;
1131 +       register char *scan;
1132 +       register char *opnd;
1133 +
1134 +       scan = g->reginput;
1135 +       opnd = OPERAND(p);
1136 +       switch (OP(p)) {
1137 +       case ANY:
1138 +               count = strlen(scan);
1139 +               scan += count;
1140 +               break;
1141 +       case EXACTLY:
1142 +               while (*opnd == *scan) {
1143 +                       count++;
1144 +                       scan++;
1145 +               }
1146 +               break;
1147 +       case ANYOF:
1148 +               while (*scan != '\0' && strchr(opnd, *scan) != NULL) {
1149 +                       count++;
1150 +                       scan++;
1151 +               }
1152 +               break;
1153 +       case ANYBUT:
1154 +               while (*scan != '\0' && strchr(opnd, *scan) == NULL) {
1155 +                       count++;
1156 +                       scan++;
1157 +               }
1158 +               break;
1159 +       default:                /* Oh dear.  Called inappropriately. */
1160 +               printk("<3>Regexp: internal foulup\n");
1161 +               count = 0;      /* Best compromise. */
1162 +               break;
1163 +       }
1164 +       g->reginput = scan;
1165 +
1166 +       return(count);
1167 +}
1168 +
1169 +/*
1170 + - regnext - dig the "next" pointer out of a node
1171 + */
1172 +static char*
1173 +regnext(struct match_globals *g, char *p)
1174 +{
1175 +       register int offset;
1176 +
1177 +       if (p == &g->regdummy)
1178 +               return(NULL);
1179 +
1180 +       offset = NEXT(p);
1181 +       if (offset == 0)
1182 +               return(NULL);
1183 +
1184 +       if (OP(p) == BACK)
1185 +               return(p-offset);
1186 +       else
1187 +               return(p+offset);
1188 +}
1189 +
1190 +#ifdef DEBUG
1191 +
1192 +STATIC char *regprop();
1193 +
1194 +/*
1195 + - regdump - dump a regexp onto stdout in vaguely comprehensible form
1196 + */
1197 +void
1198 +regdump(regexp *r)
1199 +{
1200 +       register char *s;
1201 +       register char op = EXACTLY;     /* Arbitrary non-END op. */
1202 +       register char *next;
1203 +       /* extern char *strchr(); */
1204 +
1205 +
1206 +       s = r->program + 1;
1207 +       while (op != END) {     /* While that wasn't END last time... */
1208 +               op = OP(s);
1209 +               printf("%2d%s", s-r->program, regprop(s));      /* Where, what. */
1210 +               next = regnext(s);
1211 +               if (next == NULL)               /* Next ptr. */
1212 +                       printf("(0)");
1213 +               else
1214 +                       printf("(%d)", (s-r->program)+(next-s));
1215 +               s += 3;
1216 +               if (op == ANYOF || op == ANYBUT || op == EXACTLY) {
1217 +                       /* Literal string, where present. */
1218 +                       while (*s != '\0') {
1219 +                               putchar(*s);
1220 +                               s++;
1221 +                       }
1222 +                       s++;
1223 +               }
1224 +               putchar('\n');
1225 +       }
1226 +
1227 +       /* Header fields of interest. */
1228 +       if (r->regstart != '\0')
1229 +               printf("start `%c' ", r->regstart);
1230 +       if (r->reganch)
1231 +               printf("anchored ");
1232 +       if (r->regmust != NULL)
1233 +               printf("must have \"%s\"", r->regmust);
1234 +       printf("\n");
1235 +}
1236 +
1237 +/*
1238 + - regprop - printable representation of opcode
1239 + */
1240 +static char *
1241 +regprop(char *op)
1242 +{
1243 +#define BUFLEN 50
1244 +       register char *p;
1245 +       static char buf[BUFLEN];
1246 +
1247 +       strcpy(buf, ":");
1248 +
1249 +       switch (OP(op)) {
1250 +       case BOL:
1251 +               p = "BOL";
1252 +               break;
1253 +       case EOL:
1254 +               p = "EOL";
1255 +               break;
1256 +       case ANY:
1257 +               p = "ANY";
1258 +               break;
1259 +       case ANYOF:
1260 +               p = "ANYOF";
1261 +               break;
1262 +       case ANYBUT:
1263 +               p = "ANYBUT";
1264 +               break;
1265 +       case BRANCH:
1266 +               p = "BRANCH";
1267 +               break;
1268 +       case EXACTLY:
1269 +               p = "EXACTLY";
1270 +               break;
1271 +       case NOTHING:
1272 +               p = "NOTHING";
1273 +               break;
1274 +       case BACK:
1275 +               p = "BACK";
1276 +               break;
1277 +       case END:
1278 +               p = "END";
1279 +               break;
1280 +       case OPEN+1:
1281 +       case OPEN+2:
1282 +       case OPEN+3:
1283 +       case OPEN+4:
1284 +       case OPEN+5:
1285 +       case OPEN+6:
1286 +       case OPEN+7:
1287 +       case OPEN+8:
1288 +       case OPEN+9:
1289 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "OPEN%d", OP(op)-OPEN);
1290 +               p = NULL;
1291 +               break;
1292 +       case CLOSE+1:
1293 +       case CLOSE+2:
1294 +       case CLOSE+3:
1295 +       case CLOSE+4:
1296 +       case CLOSE+5:
1297 +       case CLOSE+6:
1298 +       case CLOSE+7:
1299 +       case CLOSE+8:
1300 +       case CLOSE+9:
1301 +               snprintf(buf+strlen(buf),BUFLEN-strlen(buf), "CLOSE%d", OP(op)-CLOSE);
1302 +               p = NULL;
1303 +               break;
1304 +       case STAR:
1305 +               p = "STAR";
1306 +               break;
1307 +       case PLUS:
1308 +               p = "PLUS";
1309 +               break;
1310 +       default:
1311 +               printk("<3>Regexp: corrupted opcode\n");
1312 +               break;
1313 +       }
1314 +       if (p != NULL)
1315 +               strncat(buf, p, BUFLEN-strlen(buf));
1316 +       return(buf);
1317 +}
1318 +#endif
1319 +
1320 +
1321 diff -urNp -x '*.orig' linux-5.11/net/netfilter/regexp/regexp.h linux-5.11/net/netfilter/regexp/regexp.h
1322 --- linux-5.11/net/netfilter/regexp/regexp.h    1970-01-01 01:00:00.000000000 +0100
1323 +++ linux-5.11/net/netfilter/regexp/regexp.h    2021-02-24 13:21:48.335107407 +0100
1324 @@ -0,0 +1,41 @@
1325 +/*
1326 + * Definitions etc. for regexp(3) routines.
1327 + *
1328 + * Caveat:  this is V8 regexp(3) [actually, a reimplementation thereof],
1329 + * not the System V one.
1330 + */
1331 +
1332 +#ifndef REGEXP_H
1333 +#define REGEXP_H
1334 +
1335 +
1336 +/*
1337 +http://www.opensource.apple.com/darwinsource/10.3/expect-1/expect/expect.h ,
1338 +which contains a version of this library, says:
1339 +
1340 + *
1341 + * NSUBEXP must be at least 10, and no greater than 117 or the parser
1342 + * will not work properly.
1343 + *
1344 +
1345 +However, it looks rather like this library is limited to 10.  If you think
1346 +otherwise, let us know.
1347 +*/
1348 +
1349 +#define NSUBEXP  10
1350 +typedef struct regexp {
1351 +       char *startp[NSUBEXP];
1352 +       char *endp[NSUBEXP];
1353 +       char regstart;          /* Internal use only. */
1354 +       char reganch;           /* Internal use only. */
1355 +       char *regmust;          /* Internal use only. */
1356 +       int regmlen;            /* Internal use only. */
1357 +       char program[1];        /* Unwarranted chumminess with compiler. */
1358 +} regexp;
1359 +
1360 +regexp * regcomp(char *exp, int *patternsize);
1361 +int regexec(regexp *prog, char *string);
1362 +void regsub(regexp *prog, char *source, char *dest);
1363 +void regerror(char *s);
1364 +
1365 +#endif
1366 diff -urNp -x '*.orig' linux-5.11/net/netfilter/regexp/regmagic.h linux-5.11/net/netfilter/regexp/regmagic.h
1367 --- linux-5.11/net/netfilter/regexp/regmagic.h  1970-01-01 01:00:00.000000000 +0100
1368 +++ linux-5.11/net/netfilter/regexp/regmagic.h  2021-02-24 13:21:48.335107407 +0100
1369 @@ -0,0 +1,5 @@
1370 +/*
1371 + * The first byte of the regexp internal "program" is actually this magic
1372 + * number; the start node begins in the second byte.
1373 + */
1374 +#define        MAGIC   0234
1375 diff -urNp -x '*.orig' linux-5.11/net/netfilter/regexp/regsub.c linux-5.11/net/netfilter/regexp/regsub.c
1376 --- linux-5.11/net/netfilter/regexp/regsub.c    1970-01-01 01:00:00.000000000 +0100
1377 +++ linux-5.11/net/netfilter/regexp/regsub.c    2021-02-24 13:21:48.335107407 +0100
1378 @@ -0,0 +1,95 @@
1379 +/*
1380 + * regsub
1381 + * @(#)regsub.c        1.3 of 2 April 86
1382 + *
1383 + *     Copyright (c) 1986 by University of Toronto.
1384 + *     Written by Henry Spencer.  Not derived from licensed software.
1385 + *
1386 + *     Permission is granted to anyone to use this software for any
1387 + *     purpose on any computer system, and to redistribute it freely,
1388 + *     subject to the following restrictions:
1389 + *
1390 + *     1. The author is not responsible for the consequences of use of
1391 + *             this software, no matter how awful, even if they arise
1392 + *             from defects in it.
1393 + *
1394 + *     2. The origin of this software must not be misrepresented, either
1395 + *             by explicit claim or by omission.
1396 + *
1397 + *     3. Altered versions must be plainly marked as such, and must not
1398 + *             be misrepresented as being the original software.
1399 + *
1400 + *
1401 + * This code was modified by Ethan Sommer to work within the kernel
1402 + * (it now uses kmalloc etc..)
1403 + *
1404 + */
1405 +#include "regexp.h"
1406 +#include "regmagic.h"
1407 +#include <linux/string.h>
1408 +
1409 +
1410 +#ifndef CHARBITS
1411 +#define        UCHARAT(p)      ((int)*(unsigned char *)(p))
1412 +#else
1413 +#define        UCHARAT(p)      ((int)*(p)&CHARBITS)
1414 +#endif
1415 +
1416 +#if 0
1417 +//void regerror(char * s)
1418 +//{
1419 +//        printk("regexp(3): %s", s);
1420 +//        /* NOTREACHED */
1421 +//}
1422 +#endif
1423 +
1424 +/*
1425 + - regsub - perform substitutions after a regexp match
1426 + */
1427 +void
1428 +regsub(regexp * prog, char * source, char * dest)
1429 +{
1430 +       register char *src;
1431 +       register char *dst;
1432 +       register char c;
1433 +       register int no;
1434 +       register int len;
1435 +
1436 +       /* Not necessary and gcc doesn't like it -MLS */
1437 +       /*extern char *strncpy();*/
1438 +
1439 +       if (prog == NULL || source == NULL || dest == NULL) {
1440 +               regerror("NULL parm to regsub");
1441 +               return;
1442 +       }
1443 +       if (UCHARAT(prog->program) != MAGIC) {
1444 +               regerror("damaged regexp fed to regsub");
1445 +               return;
1446 +       }
1447 +
1448 +       src = source;
1449 +       dst = dest;
1450 +       while ((c = *src++) != '\0') {
1451 +               if (c == '&')
1452 +                       no = 0;
1453 +               else if (c == '\\' && '0' <= *src && *src <= '9')
1454 +                       no = *src++ - '0';
1455 +               else
1456 +                       no = -1;
1457 +
1458 +               if (no < 0) {   /* Ordinary character. */
1459 +                       if (c == '\\' && (*src == '\\' || *src == '&'))
1460 +                               c = *src++;
1461 +                       *dst++ = c;
1462 +               } else if (prog->startp[no] != NULL && prog->endp[no] != NULL) {
1463 +                       len = prog->endp[no] - prog->startp[no];
1464 +                       (void) strncpy(dst, prog->startp[no], len);
1465 +                       dst += len;
1466 +                       if (len != 0 && *(dst-1) == '\0') {     /* strncpy hit NUL. */
1467 +                               regerror("damaged match string");
1468 +                               return;
1469 +                       }
1470 +               }
1471 +       }
1472 +       *dst++ = '\0';
1473 +}
1474 diff -urNp -x '*.orig' linux-5.11/net/netfilter/xt_layer7.c linux-5.11/net/netfilter/xt_layer7.c
1475 --- linux-5.11/net/netfilter/xt_layer7.c        1970-01-01 01:00:00.000000000 +0100
1476 +++ linux-5.11/net/netfilter/xt_layer7.c        2021-02-24 13:21:48.335107407 +0100
1477 @@ -0,0 +1,656 @@
1478 +/*
1479 +  Kernel module to match application layer (OSI layer 7) data in connections.
1480 +
1481 +  http://l7-filter.sf.net
1482 +
1483 +  (C) 2003-2009 Matthew Strait and Ethan Sommer.
1484 +
1485 +  This program is free software; you can redistribute it and/or
1486 +  modify it under the terms of the GNU General Public License
1487 +  as published by the Free Software Foundation; either version
1488 +  2 of the License, or (at your option) any later version.
1489 +  http://www.gnu.org/licenses/gpl.txt
1490 +
1491 +  Based on ipt_string.c (C) 2000 Emmanuel Roger <winfield@freegates.be>,
1492 +  xt_helper.c (C) 2002 Harald Welte and cls_layer7.c (C) 2003 Matthew Strait,
1493 +  Ethan Sommer, Justin Levandoski.
1494 +*/
1495 +
1496 +#include <linux/spinlock.h>
1497 +#include <linux/version.h>
1498 +#include <net/ip.h>
1499 +#include <net/tcp.h>
1500 +#include <linux/module.h>
1501 +#include <linux/skbuff.h>
1502 +#include <linux/netfilter.h>
1503 +#include <net/netfilter/nf_conntrack.h>
1504 +#include <net/netfilter/nf_conntrack_core.h>
1505 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 27)
1506 +#include <net/netfilter/nf_conntrack_extend.h>
1507 +#include <net/netfilter/nf_conntrack_acct.h>
1508 +#endif
1509 +#include <linux/netfilter/x_tables.h>
1510 +#include <linux/netfilter/xt_layer7.h>
1511 +#include <linux/ctype.h>
1512 +#include <linux/proc_fs.h>
1513 +
1514 +#include "regexp/regexp.c"
1515 +
1516 +MODULE_LICENSE("GPL");
1517 +MODULE_AUTHOR("Matthew Strait <quadong@users.sf.net>, Ethan Sommer <sommere@users.sf.net>");
1518 +MODULE_DESCRIPTION("iptables application layer match module");
1519 +MODULE_ALIAS("ipt_layer7");
1520 +MODULE_VERSION("2.21");
1521 +
1522 +static int maxdatalen = 2048; // this is the default
1523 +module_param(maxdatalen, int, 0444);
1524 +MODULE_PARM_DESC(maxdatalen, "maximum bytes of data looked at by l7-filter");
1525 +#ifdef CONFIG_NETFILTER_XT_MATCH_LAYER7_DEBUG
1526 +       #define DPRINTK(format,args...) printk(format,##args)
1527 +#else
1528 +       #define DPRINTK(format,args...)
1529 +#endif
1530 +
1531 +/* Number of packets whose data we look at.
1532 +This can be modified through /proc/net/layer7_numpackets */
1533 +static int num_packets = 10;
1534 +
1535 +static struct pattern_cache {
1536 +       char * regex_string;
1537 +       regexp * pattern;
1538 +       struct pattern_cache * next;
1539 +} * first_pattern_cache = NULL;
1540 +
1541 +DEFINE_SPINLOCK(l7_lock);
1542 +
1543 +static int total_acct_packets(struct nf_conn *ct)
1544 +{
1545 +#if LINUX_VERSION_CODE <= KERNEL_VERSION(2, 6, 26)
1546 +       BUG_ON(ct == NULL);
1547 +       return (ct->counters[IP_CT_DIR_ORIGINAL].packets + ct->counters[IP_CT_DIR_REPLY].packets);
1548 +#else
1549 +       struct nf_conn_counter *acct;
1550 +
1551 +       BUG_ON(ct == NULL);
1552 +       acct = nf_conn_acct_find(ct);
1553 +       if (!acct)
1554 +               return 0;
1555 +       return (atomic64_read(&acct[IP_CT_DIR_ORIGINAL].packets) + atomic64_read(&acct[IP_CT_DIR_REPLY].packets));
1556 +#endif
1557 +}
1558 +
1559 +#ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
1560 +/* Converts an unfriendly string into a friendly one by
1561 +replacing unprintables with periods and all whitespace with " ". */
1562 +static char * friendly_print(unsigned char * s)
1563 +{
1564 +       char * f = kmalloc(strlen(s) + 1, GFP_ATOMIC);
1565 +       int i;
1566 +
1567 +       if(!f) {
1568 +               if (net_ratelimit())
1569 +                       printk(KERN_ERR "layer7: out of memory in "
1570 +                                       "friendly_print, bailing.\n");
1571 +               return NULL;
1572 +       }
1573 +
1574 +       for(i = 0; i < strlen(s); i++){
1575 +               if(isprint(s[i]) && s[i] < 128) f[i] = s[i];
1576 +               else if(isspace(s[i]))          f[i] = ' ';
1577 +               else                            f[i] = '.';
1578 +       }
1579 +       f[i] = '\0';
1580 +       return f;
1581 +}
1582 +
1583 +static char dec2hex(int i)
1584 +{
1585 +       switch (i) {
1586 +               case 0 ... 9:
1587 +                       return (i + '0');
1588 +                       break;
1589 +               case 10 ... 15:
1590 +                       return (i - 10 + 'a');
1591 +                       break;
1592 +               default:
1593 +                       if (net_ratelimit())
1594 +                               printk("layer7: Problem in dec2hex\n");
1595 +                       return '\0';
1596 +       }
1597 +}
1598 +
1599 +static char * hex_print(unsigned char * s)
1600 +{
1601 +       char * g = kmalloc(strlen(s)*3 + 1, GFP_ATOMIC);
1602 +       int i;
1603 +
1604 +       if(!g) {
1605 +              if (net_ratelimit())
1606 +                       printk(KERN_ERR "layer7: out of memory in hex_print, "
1607 +                                       "bailing.\n");
1608 +              return NULL;
1609 +       }
1610 +
1611 +       for(i = 0; i < strlen(s); i++) {
1612 +               g[i*3    ] = dec2hex(s[i]/16);
1613 +               g[i*3 + 1] = dec2hex(s[i]%16);
1614 +               g[i*3 + 2] = ' ';
1615 +       }
1616 +       g[i*3] = '\0';
1617 +
1618 +       return g;
1619 +}
1620 +#endif // DEBUG
1621 +
1622 +/* Use instead of regcomp.  As we expect to be seeing the same regexps over and
1623 +over again, it make sense to cache the results. */
1624 +static regexp * compile_and_cache(const char * regex_string,
1625 +                                  const char * protocol)
1626 +{
1627 +       struct pattern_cache * node               = first_pattern_cache;
1628 +       struct pattern_cache * last_pattern_cache = first_pattern_cache;
1629 +       struct pattern_cache * tmp;
1630 +       unsigned int len;
1631 +
1632 +       while (node != NULL) {
1633 +               if (!strcmp(node->regex_string, regex_string))
1634 +               return node->pattern;
1635 +
1636 +               last_pattern_cache = node;/* points at the last non-NULL node */
1637 +               node = node->next;
1638 +       }
1639 +
1640 +       /* If we reach the end of the list, then we have not yet cached
1641 +          the pattern for this regex. Let's do that now.
1642 +          Be paranoid about running out of memory to avoid list corruption. */
1643 +       tmp = kmalloc(sizeof(struct pattern_cache), GFP_ATOMIC);
1644 +
1645 +       if(!tmp) {
1646 +               if (net_ratelimit())
1647 +                       printk(KERN_ERR "layer7: out of memory in "
1648 +                                       "compile_and_cache, bailing.\n");
1649 +               return NULL;
1650 +       }
1651 +
1652 +       tmp->regex_string  = kmalloc(strlen(regex_string) + 1, GFP_ATOMIC);
1653 +       tmp->pattern       = kmalloc(sizeof(struct regexp),    GFP_ATOMIC);
1654 +       tmp->next = NULL;
1655 +
1656 +       if(!tmp->regex_string || !tmp->pattern) {
1657 +               if (net_ratelimit())
1658 +                       printk(KERN_ERR "layer7: out of memory in "
1659 +                                       "compile_and_cache, bailing.\n");
1660 +               kfree(tmp->regex_string);
1661 +               kfree(tmp->pattern);
1662 +               kfree(tmp);
1663 +               return NULL;
1664 +       }
1665 +
1666 +       /* Ok.  The new node is all ready now. */
1667 +       node = tmp;
1668 +
1669 +       if(first_pattern_cache == NULL) /* list is empty */
1670 +               first_pattern_cache = node; /* make node the beginning */
1671 +       else
1672 +               last_pattern_cache->next = node; /* attach node to the end */
1673 +
1674 +       /* copy the string and compile the regex */
1675 +       len = strlen(regex_string);
1676 +       DPRINTK("About to compile this: \"%s\"\n", regex_string);
1677 +       node->pattern = regcomp((char *)regex_string, &len);
1678 +       if ( !node->pattern ) {
1679 +               if (net_ratelimit())
1680 +                       printk(KERN_ERR "layer7: Error compiling regexp "
1681 +                                       "\"%s\" (%s)\n",
1682 +                                       regex_string, protocol);
1683 +               /* pattern is now cached as NULL, so we won't try again. */
1684 +       }
1685 +
1686 +       strcpy(node->regex_string, regex_string);
1687 +       return node->pattern;
1688 +}
1689 +
1690 +static int can_handle(const struct sk_buff *skb)
1691 +{
1692 +       if(!ip_hdr(skb)) /* not IP */
1693 +               return 0;
1694 +       if(ip_hdr(skb)->protocol != IPPROTO_TCP &&
1695 +          ip_hdr(skb)->protocol != IPPROTO_UDP &&
1696 +          ip_hdr(skb)->protocol != IPPROTO_ICMP)
1697 +               return 0;
1698 +       return 1;
1699 +}
1700 +
1701 +/* Returns offset the into the skb->data that the application data starts */
1702 +static int app_data_offset(const struct sk_buff *skb)
1703 +{
1704 +       /* In case we are ported somewhere (ebtables?) where ip_hdr(skb)
1705 +       isn't set, this can be gotten from 4*(skb->data[0] & 0x0f) as well. */
1706 +       int ip_hl = 4*ip_hdr(skb)->ihl;
1707 +
1708 +       if( ip_hdr(skb)->protocol == IPPROTO_TCP ) {
1709 +               /* 12 == offset into TCP header for the header length field.
1710 +               Can't get this with skb->h.th->doff because the tcphdr
1711 +               struct doesn't get set when routing (this is confirmed to be
1712 +               true in Netfilter as well as QoS.) */
1713 +               int tcp_hl = 4*(skb->data[ip_hl + 12] >> 4);
1714 +
1715 +               return ip_hl + tcp_hl;
1716 +       } else if( ip_hdr(skb)->protocol == IPPROTO_UDP  ) {
1717 +               return ip_hl + 8; /* UDP header is always 8 bytes */
1718 +       } else if( ip_hdr(skb)->protocol == IPPROTO_ICMP ) {
1719 +               return ip_hl + 8; /* ICMP header is 8 bytes */
1720 +       } else {
1721 +               if (net_ratelimit())
1722 +                       printk(KERN_ERR "layer7: tried to handle unknown "
1723 +                                       "protocol!\n");
1724 +               return ip_hl + 8; /* something reasonable */
1725 +       }
1726 +}
1727 +
1728 +/* handles whether there's a match when we aren't appending data anymore */
1729 +static int match_no_append(struct nf_conn * conntrack,
1730 +                           struct nf_conn * master_conntrack,
1731 +                           enum ip_conntrack_info ctinfo,
1732 +                           enum ip_conntrack_info master_ctinfo,
1733 +                           const struct xt_layer7_info * info)
1734 +{
1735 +       /* If we're in here, throw the app data away */
1736 +       if(master_conntrack->layer7.app_data != NULL) {
1737 +
1738 +       #ifdef CONFIG_IP_NF_MATCH_LAYER7_DEBUG
1739 +               if(!master_conntrack->layer7.app_proto) {
1740 +                       char * f =
1741 +                         friendly_print(master_conntrack->layer7.app_data);
1742 +                       char * g =
1743 +                         hex_print(master_conntrack->layer7.app_data);
1744 +                       DPRINTK("\nl7-filter gave up after %d bytes "
1745 +                               "(%d packets):\n%s\n",
1746 +                               strlen(f), total_acct_packets(master_conntrack), f);
1747 +                       kfree(f);
1748 +                       DPRINTK("In hex: %s\n", g);
1749 +                       kfree(g);
1750 +               }
1751 +       #endif
1752 +
1753 +               kfree(master_conntrack->layer7.app_data);
1754 +               master_conntrack->layer7.app_data = NULL; /* don't free again */
1755 +       }
1756 +
1757 +       if(master_conntrack->layer7.app_proto){
1758 +               /* Here child connections set their .app_proto (for /proc) */
1759 +               if(!conntrack->layer7.app_proto) {
1760 +                       conntrack->layer7.app_proto =
1761 +                         kmalloc(strlen(master_conntrack->layer7.app_proto)+1,
1762 +                           GFP_ATOMIC);
1763 +                       if(!conntrack->layer7.app_proto){
1764 +                               if (net_ratelimit())
1765 +                                       printk(KERN_ERR "layer7: out of memory "
1766 +                                                       "in match_no_append, "
1767 +                                                       "bailing.\n");
1768 +                               return 1;
1769 +                       }
1770 +                       strcpy(conntrack->layer7.app_proto,
1771 +                               master_conntrack->layer7.app_proto);
1772 +               }
1773 +
1774 +               return (!strcmp(master_conntrack->layer7.app_proto,
1775 +                               info->protocol));
1776 +       }
1777 +       else {
1778 +               /* If not classified, set to "unknown" to distinguish from
1779 +               connections that are still being tested. */
1780 +               master_conntrack->layer7.app_proto =
1781 +                       kmalloc(strlen("unknown")+1, GFP_ATOMIC);
1782 +               if(!master_conntrack->layer7.app_proto){
1783 +                       if (net_ratelimit())
1784 +                               printk(KERN_ERR "layer7: out of memory in "
1785 +                                               "match_no_append, bailing.\n");
1786 +                       return 1;
1787 +               }
1788 +               strcpy(master_conntrack->layer7.app_proto, "unknown");
1789 +               return 0;
1790 +       }
1791 +}
1792 +
1793 +/* add the new app data to the conntrack.  Return number of bytes added. */
1794 +static int add_data(struct nf_conn * master_conntrack,
1795 +                    char * app_data, int appdatalen)
1796 +{
1797 +       int length = 0, i;
1798 +       int oldlength = master_conntrack->layer7.app_data_len;
1799 +
1800 +       /* This is a fix for a race condition by Deti Fliegl. However, I'm not
1801 +          clear on whether the race condition exists or whether this really
1802 +          fixes it.  I might just be being dense... Anyway, if it's not really
1803 +          a fix, all it does is waste a very small amount of time. */
1804 +       if(!master_conntrack->layer7.app_data) return 0;
1805 +
1806 +       /* Strip nulls. Make everything lower case (our regex lib doesn't
1807 +       do case insensitivity).  Add it to the end of the current data. */
1808 +       for(i = 0; i < maxdatalen-oldlength-1 &&
1809 +                  i < appdatalen; i++) {
1810 +               if(app_data[i] != '\0') {
1811 +                       /* the kernel version of tolower mungs 'upper ascii' */
1812 +                       master_conntrack->layer7.app_data[length+oldlength] =
1813 +                               isascii(app_data[i])?
1814 +                                       tolower(app_data[i]) : app_data[i];
1815 +                       length++;
1816 +               }
1817 +       }
1818 +
1819 +       master_conntrack->layer7.app_data[length+oldlength] = '\0';
1820 +       master_conntrack->layer7.app_data_len = length + oldlength;
1821 +
1822 +       return length;
1823 +}
1824 +
1825 +/* taken from drivers/video/modedb.c */
1826 +static int my_atoi(const char *s)
1827 +{
1828 +       int val = 0;
1829 +
1830 +       for (;; s++) {
1831 +               switch (*s) {
1832 +                       case '0'...'9':
1833 +                       val = 10*val+(*s-'0');
1834 +                       break;
1835 +               default:
1836 +                       return val;
1837 +               }
1838 +       }
1839 +}
1840 +
1841 +/* write out num_packets to userland. */
1842 +static int layer7_read_proc(char* page, char ** start, off_t off, int count,
1843 +                            int* eof, void * data)
1844 +{
1845 +       if(num_packets > 99 && net_ratelimit())
1846 +               printk(KERN_ERR "layer7: NOT REACHED. num_packets too big\n");
1847 +
1848 +       page[0] = num_packets/10 + '0';
1849 +       page[1] = num_packets%10 + '0';
1850 +       page[2] = '\n';
1851 +       page[3] = '\0';
1852 +
1853 +       *eof=1;
1854 +
1855 +       return 3;
1856 +}
1857 +
1858 +/* Read in num_packets from userland */
1859 +static int layer7_write_proc(struct file* file, const char* buffer,
1860 +                             unsigned long count, void *data)
1861 +{
1862 +       char * foo = kmalloc(count, GFP_ATOMIC);
1863 +
1864 +       if(!foo){
1865 +               if (net_ratelimit())
1866 +                       printk(KERN_ERR "layer7: out of memory, bailing. "
1867 +                                       "num_packets unchanged.\n");
1868 +               return count;
1869 +       }
1870 +
1871 +       if(copy_from_user(foo, buffer, count)) {
1872 +               return -EFAULT;
1873 +       }
1874 +
1875 +
1876 +       num_packets = my_atoi(foo);
1877 +       kfree (foo);
1878 +
1879 +       /* This has an arbitrary limit to make the math easier. I'm lazy.
1880 +       But anyway, 99 is a LOT! If you want more, you're doing it wrong! */
1881 +       if(num_packets > 99) {
1882 +               printk(KERN_WARNING "layer7: num_packets can't be > 99.\n");
1883 +               num_packets = 99;
1884 +       } else if(num_packets < 1) {
1885 +               printk(KERN_WARNING "layer7: num_packets can't be < 1.\n");
1886 +               num_packets = 1;
1887 +       }
1888 +
1889 +       return count;
1890 +}
1891 +
1892 +static bool
1893 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28)
1894 +match(const struct sk_buff *skbin, struct xt_action_param *par)
1895 +#else
1896 +match(const struct sk_buff *skbin,
1897 +      const struct net_device *in,
1898 +      const struct net_device *out,
1899 +      const struct xt_match *match,
1900 +      const void *matchinfo,
1901 +      int offset,
1902 +      unsigned int protoff,
1903 +      bool *hotdrop)
1904 +#endif
1905 +{
1906 +       /* sidestep const without getting a compiler warning... */
1907 +       struct sk_buff * skb = (struct sk_buff *)skbin;
1908 +
1909 +       const struct xt_layer7_info * info =
1910 +       #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28)
1911 +               par->matchinfo;
1912 +       #else
1913 +               matchinfo;
1914 +       #endif
1915 +
1916 +       enum ip_conntrack_info master_ctinfo, ctinfo;
1917 +       struct nf_conn *master_conntrack, *conntrack;
1918 +       unsigned char * app_data;
1919 +       unsigned int pattern_result, appdatalen;
1920 +       regexp * comppattern;
1921 +
1922 +       /* Be paranoid/incompetent - lock the entire match function. */
1923 +       spin_lock_bh(&l7_lock);
1924 +
1925 +       if(!can_handle(skb)){
1926 +               DPRINTK("layer7: This is some protocol I can't handle.\n");
1927 +               spin_unlock_bh(&l7_lock);
1928 +               return info->invert;
1929 +       }
1930 +
1931 +       /* Treat parent & all its children together as one connection, except
1932 +       for the purpose of setting conntrack->layer7.app_proto in the actual
1933 +       connection. This makes /proc/net/ip_conntrack more satisfying. */
1934 +       if(!(conntrack = nf_ct_get(skb, &ctinfo)) ||
1935 +          !(master_conntrack=nf_ct_get(skb,&master_ctinfo))){
1936 +               DPRINTK("layer7: couldn't get conntrack.\n");
1937 +               spin_unlock_bh(&l7_lock);
1938 +               return info->invert;
1939 +       }
1940 +
1941 +       /* Try to get a master conntrack (and its master etc) for FTP, etc. */
1942 +       while (master_ct(master_conntrack) != NULL)
1943 +               master_conntrack = master_ct(master_conntrack);
1944 +
1945 +       /* if we've classified it or seen too many packets */
1946 +       if(total_acct_packets(master_conntrack) > num_packets ||
1947 +          master_conntrack->layer7.app_proto) {
1948 +
1949 +               pattern_result = match_no_append(conntrack, master_conntrack,
1950 +                                                ctinfo, master_ctinfo, info);
1951 +
1952 +               /* skb->cb[0] == seen. Don't do things twice if there are
1953 +               multiple l7 rules. I'm not sure that using cb for this purpose
1954 +               is correct, even though it says "put your private variables
1955 +               there". But it doesn't look like it is being used for anything
1956 +               else in the skbs that make it here. */
1957 +               skb->cb[0] = 1; /* marking it seen here's probably irrelevant */
1958 +
1959 +               spin_unlock_bh(&l7_lock);
1960 +               return (pattern_result ^ info->invert);
1961 +       }
1962 +
1963 +       if(skb_is_nonlinear(skb)){
1964 +               if(skb_linearize(skb) != 0){
1965 +                       if (net_ratelimit())
1966 +                               printk(KERN_ERR "layer7: failed to linearize "
1967 +                                               "packet, bailing.\n");
1968 +                       spin_unlock_bh(&l7_lock);
1969 +                       return info->invert;
1970 +               }
1971 +       }
1972 +
1973 +       /* now that the skb is linearized, it's safe to set these. */
1974 +       app_data = skb->data + app_data_offset(skb);
1975 +       appdatalen = skb_tail_pointer(skb) - app_data;
1976 +
1977 +       /* the return value gets checked later, when we're ready to use it */
1978 +       comppattern = compile_and_cache(info->pattern, info->protocol);
1979 +
1980 +       /* On the first packet of a connection, allocate space for app data */
1981 +       if(total_acct_packets(master_conntrack) == 1 && !skb->cb[0] &&
1982 +          !master_conntrack->layer7.app_data){
1983 +               master_conntrack->layer7.app_data =
1984 +                       kmalloc(maxdatalen, GFP_ATOMIC);
1985 +               if(!master_conntrack->layer7.app_data){
1986 +                       if (net_ratelimit())
1987 +                               printk(KERN_ERR "layer7: out of memory in "
1988 +                                               "match, bailing.\n");
1989 +                       spin_unlock_bh(&l7_lock);
1990 +                       return info->invert;
1991 +               }
1992 +
1993 +               master_conntrack->layer7.app_data[0] = '\0';
1994 +       }
1995 +
1996 +       /* Can be here, but unallocated, if numpackets is increased near
1997 +       the beginning of a connection */
1998 +       if(master_conntrack->layer7.app_data == NULL){
1999 +               spin_unlock_bh(&l7_lock);
2000 +               return info->invert; /* unmatched */
2001 +       }
2002 +
2003 +       if(!skb->cb[0]){
2004 +               int newbytes;
2005 +               newbytes = add_data(master_conntrack, app_data, appdatalen);
2006 +
2007 +               if(newbytes == 0) { /* didn't add any data */
2008 +                       skb->cb[0] = 1;
2009 +                       /* Didn't match before, not going to match now */
2010 +                       spin_unlock_bh(&l7_lock);
2011 +                       return info->invert;
2012 +               }
2013 +       }
2014 +
2015 +       /* If looking for "unknown", then never match.  "Unknown" means that
2016 +       we've given up; we're still trying with these packets. */
2017 +       if(!strcmp(info->protocol, "unknown")) {
2018 +               pattern_result = 0;
2019 +       /* If looking for "unset", then always match. "Unset" means that we
2020 +       haven't yet classified the connection. */
2021 +       } else if(!strcmp(info->protocol, "unset")) {
2022 +               pattern_result = 2;
2023 +               DPRINTK("layer7: matched unset: not yet classified "
2024 +                       "(%d/%d packets)\n",
2025 +                        total_acct_packets(master_conntrack), num_packets);
2026 +       /* If the regexp failed to compile, don't bother running it */
2027 +       } else if(comppattern &&
2028 +                 regexec(comppattern, master_conntrack->layer7.app_data)){
2029 +               DPRINTK("layer7: matched %s\n", info->protocol);
2030 +               pattern_result = 1;
2031 +       } else pattern_result = 0;
2032 +
2033 +       if(pattern_result == 1) {
2034 +               master_conntrack->layer7.app_proto =
2035 +                       kmalloc(strlen(info->protocol)+1, GFP_ATOMIC);
2036 +               if(!master_conntrack->layer7.app_proto){
2037 +                       if (net_ratelimit())
2038 +                               printk(KERN_ERR "layer7: out of memory in "
2039 +                                               "match, bailing.\n");
2040 +                       spin_unlock_bh(&l7_lock);
2041 +                       return (pattern_result ^ info->invert);
2042 +               }
2043 +               strcpy(master_conntrack->layer7.app_proto, info->protocol);
2044 +       } else if(pattern_result > 1) { /* cleanup from "unset" */
2045 +               pattern_result = 1;
2046 +       }
2047 +
2048 +       /* mark the packet seen */
2049 +       skb->cb[0] = 1;
2050 +
2051 +       spin_unlock_bh(&l7_lock);
2052 +       return (pattern_result ^ info->invert);
2053 +}
2054 +
2055 +// load nf_conntrack_ipv4
2056 +static int check(const struct xt_mtchk_param *par)
2057 +{
2058 +        if (nf_ct_l3proto_try_module_get(par->family) < 0) {
2059 +                pr_info("can't load conntrack support for "
2060 +                                    "proto=%d\n", par->family);
2061 +                return -EINVAL;
2062 +        }
2063 +       return 0;
2064 +}
2065 +
2066 +
2067 +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28)
2068 +       static void destroy(const struct xt_mtdtor_param *par)
2069 +       {
2070 +               nf_ct_l3proto_module_put(par->family);
2071 +       }
2072 +#else
2073 +       static void destroy(const struct xt_match *match, void *matchinfo)
2074 +       {
2075 +               nf_ct_l3proto_module_put(match->family);
2076 +       }
2077 +#endif
2078 +
2079 +static struct xt_match xt_layer7_match[] __read_mostly = {
2080 +{
2081 +       .name           = "layer7",
2082 +       .family         = AF_INET,
2083 +       .checkentry     = check,
2084 +       .match          = match,
2085 +       .destroy        = destroy,
2086 +       .matchsize      = sizeof(struct xt_layer7_info),
2087 +       .me             = THIS_MODULE
2088 +}
2089 +};
2090 +
2091 +static void layer7_cleanup_proc(void)
2092 +{
2093 +       remove_proc_entry("layer7_numpackets", init_net.proc_net);
2094 +}
2095 +
2096 +/* register the proc file */
2097 +static void layer7_init_proc(void)
2098 +{
2099 +       struct proc_dir_entry* entry;
2100 +       entry = create_proc_entry("layer7_numpackets", 0644, init_net.proc_net);
2101 +       entry->read_proc = layer7_read_proc;
2102 +       entry->write_proc = layer7_write_proc;
2103 +}
2104 +
2105 +static int __init xt_layer7_init(void)
2106 +{
2107 +       need_conntrack();
2108 +
2109 +       layer7_init_proc();
2110 +       if(maxdatalen < 1) {
2111 +               printk(KERN_WARNING "layer7: maxdatalen can't be < 1, "
2112 +                       "using 1\n");
2113 +               maxdatalen = 1;
2114 +       }
2115 +       /* This is not a hard limit.  It's just here to prevent people from
2116 +       bringing their slow machines to a grinding halt. */
2117 +       else if(maxdatalen > 65536) {
2118 +               printk(KERN_WARNING "layer7: maxdatalen can't be > 65536, "
2119 +                       "using 65536\n");
2120 +               maxdatalen = 65536;
2121 +       }
2122 +       return xt_register_matches(xt_layer7_match,
2123 +                                  ARRAY_SIZE(xt_layer7_match));
2124 +}
2125 +
2126 +static void __exit xt_layer7_fini(void)
2127 +{
2128 +       layer7_cleanup_proc();
2129 +       xt_unregister_matches(xt_layer7_match, ARRAY_SIZE(xt_layer7_match));
2130 +}
2131 +
2132 +module_init(xt_layer7_init);
2133 +module_exit(xt_layer7_fini);