pcre: added disabling mechanism to metas and matcher

dconf: added field specific for pcre features
dconf: added overall support dconf for pcre

libclamav/dconf.c

@@ -136,6 +136,8 @@ static struct dconf_module modules[] = {
     { "STATS",      "DISABLED",     DCONF_STATS_DISABLED,   0 },
     { "STATS",      "PESECTION DISABLED", DCONF_STATS_PE_SECTION_DISABLED, 0 },
 
+    { "PCRE",       "SUPPORT",      PCRE_CONF_SUPPORT,   1 },
+
     { NULL,     NULL,       0,              0 }
 };
 
@@ -189,6 +191,9 @@ struct cli_dconf *cli_dconf_init(void)
         } else if (!strcmp(modules[i].mname, "STATS")) {
             if (modules[i].state)
                 dconf->stats |= modules[i].bflag;
+        } else if (!strcmp(modules[i].mname, "PCRE")) {
+            if (modules[i].state)
+                dconf->pcre |= modules[i].bflag;
         }
     }
 
@@ -198,7 +203,7 @@ struct cli_dconf *cli_dconf_init(void)
 void cli_dconf_print(struct cli_dconf *dconf)
 {
     unsigned int pe = 0, elf = 0, macho = 0, arch = 0, doc = 0, mail = 0;
-    unsigned int other = 0, phishing = 0, i, bytecode=0, stats=0;
+    unsigned int other = 0, phishing = 0, i, bytecode=0, stats=0, pcre=0;
 
 
     cli_dbgmsg("Dynamic engine configuration settings:\n");
@@ -292,7 +297,17 @@ void cli_dconf_print(struct cli_dconf *dconf)
             }
 
             if (dconf->stats)
-                cli_dbgmsg("    * Submodule %10s:\t%s\n", modules[i].sname, (dconf->stats & modules[i].bflag) ? "On" : "** Off **");
+                cli_dbgmsg("   * Submodule %10s:\t%s\n", modules[i].sname, (dconf->stats & modules[i].bflag) ? "On" : "** Off **");
+            else
+                continue;
+        } else if (!strcmp(modules[i].mname, "PCRE")) {
+            if (!pcre) {
+                cli_dbgmsg("Module PCRE %s\n", dconf->pcre ? "On" : "Off");
+                stats = 1;
+            }
+
+            if (dconf->pcre)
+                cli_dbgmsg("   * Submodule %10s:\t%s\n", modules[i].sname, (dconf->pcre & modules[i].bflag) ? "On" : "** Off **");
             else
                 continue;
         }
@@ -437,6 +452,15 @@ int cli_dconf_load(FILE *fs, struct cl_engine *engine, unsigned int options, str
                 break;
             }
         }
+
+        if(!strncmp(buffer, "PCRE:", 5) && chkflevel(buffer, 2)) {
+            if(sscanf(buffer + 5, "0x%x", &val) == 1) {
+                engine->dconf->pcre = val;
+            } else {
+                ret = CL_EMALFDB;
+                break;
+            }
+        }
     }
 
     if(ret) {

libclamav/dconf.h

@@ -42,6 +42,7 @@ struct cli_dconf {
     uint32_t phishing;
     uint32_t bytecode;
     uint32_t stats;
+    uint32_t pcre;
 };
 
 /* PE flags */
@@ -129,6 +130,9 @@ struct cli_dconf {
 #define DCONF_STATS_DISABLED            0x1
 #define DCONF_STATS_PE_SECTION_DISABLED 0x2
 
+/* PCRE flags */
+#define PCRE_CONF_SUPPORT 0x1
+
 #define BYTECODE_ENGINE_MASK (BYTECODE_INTERPRETER | BYTECODE_JIT_X86 | BYTECODE_JIT_PPC | BYTECODE_JIT_ARM)
 
 #ifdef USE_MPOOL

libclamav/matcher-pcre.c

@@ -29,6 +29,7 @@
 #if HAVE_PCRE
 #include "clamav.h"
 #include "cltypes.h"
+#include "dconf.h"
 #include "others.h"
 #include "matcher.h"
 #include "matcher-pcre.h"
@@ -142,7 +143,8 @@ int cli_pcre_addpatt(struct cli_matcher *root, const char *trigger, const char *
             }
             opt++;
         }
-
+        /* TODO - better debug? */
+        /*
         cli_dbgmsg("PCRE_CASELESS       %08x\n", PCRE_CASELESS);
         cli_dbgmsg("PCRE_DOTALL         %08x\n", PCRE_DOTALL);
         cli_dbgmsg("PCRE_MULTILINE      %08x\n", PCRE_MULTILINE);
@@ -153,6 +155,7 @@ int cli_pcre_addpatt(struct cli_matcher *root, const char *trigger, const char *
         cli_dbgmsg("PCRE_UNGREEDY       %08x\n", PCRE_UNGREEDY);
 
         cli_dbgmsg("PCRE_OPTIONS        %08x\n", pm->pdata.options);
+        */
     }
 
     /* add pcre data to root after reallocation */
@@ -181,15 +184,27 @@ int cli_pcre_addpatt(struct cli_matcher *root, const char *trigger, const char *
     return CL_SUCCESS;
 }
 
-int cli_pcre_build(struct cli_matcher *root, long long unsigned match_limit, long long unsigned recmatch_limit)
+ int cli_pcre_build(struct cli_matcher *root, long long unsigned match_limit, long long unsigned recmatch_limit, const struct cli_dconf *dconf)
 {
     unsigned int i;
     int ret;
     struct cli_pcre_meta *pm = NULL;
+    int disable_all = !(dconf->pcre & PCRE_CONF_SUPPORT);
 
     for (i = 0; i < root->pcre_metas; ++i) {
         pm = root->pcre_metatable[i];
 
+        /* for safety, disable all pcre */
+        if (disable_all) {
+            pm->flags |= CLI_PCRE_DISABLED;
+            continue;
+        }
+
+        if (pm->flags & CLI_PCRE_DISABLED) {
+            cli_dbgmsg("cli_pcre_build: Skip compiling regex: %s (disabled)\n", pm->pdata.expression);
+            continue;
+        }
+
         if (!pm) {
             cli_errmsg("cli_pcre_build: metadata for pcre %d is missing\n", i);
             return CL_ENULLARG;
@@ -204,7 +219,8 @@ int cli_pcre_build(struct cli_matcher *root, long long unsigned match_limit, lon
 
         /* parse the regex, no options override *wink* */
         if ((ret = cli_pcre_compile(&(pm->pdata), match_limit, recmatch_limit, 0, 0)) != CL_SUCCESS) {
-            cli_errmsg("cli_pcre_build: failed to parse pcre regex\n");
+            cli_errmsg("cli_pcre_build: failed to build pcre regex\n");
+            pm->flags |= CLI_PCRE_DISABLED; /* disable the pcre */
             return ret;
         }
     }
@@ -212,7 +228,7 @@ int cli_pcre_build(struct cli_matcher *root, long long unsigned match_limit, lon
     return CL_SUCCESS;
 }
 
-int cli_pcre_recaloff(struct cli_matcher *root, struct cli_pcre_off *data, struct cli_target_info *info)
+int cli_pcre_recaloff(struct cli_matcher *root, struct cli_pcre_off *data, struct cli_target_info *info, cli_ctx *ctx)
 {
     /* TANGENT: maintain relative offset data in cli_ac_data? */
     int ret;
@@ -223,7 +239,8 @@ int cli_pcre_recaloff(struct cli_matcher *root, struct cli_pcre_off *data, struc
     if (!data) {
         return CL_ENULLARG;
     }
-    if (!root || !root->pcre_metatable || !info) {
+
+    if (!(ctx->dconf->pcre & PCRE_CONF_SUPPORT) || !root || !root->pcre_metatable || !info) {
         data->shift = NULL;
         data->offset = NULL;
         return CL_SUCCESS;
@@ -246,6 +263,13 @@ int cli_pcre_recaloff(struct cli_matcher *root, struct cli_pcre_off *data, struc
     for (i = 0; i < root->pcre_metas; ++i) {
         pm = root->pcre_metatable[i];
 
+        /* skip broken pcres, not getting executed always */
+        if (pm->flags & CLI_PCRE_DISABLED) {
+            data->offset[i] = CLI_OFF_NONE;
+            data->shift[i] = CLI_OFF_NONE;
+            continue;
+        }
+
         if (pm->offdata[0] == CLI_OFF_ANY) {
             data->offset[i] = 0;
             data->shift[i] = 0;
@@ -269,6 +293,8 @@ int cli_pcre_recaloff(struct cli_matcher *root, struct cli_pcre_off *data, struc
             data->shift[i] = endoff-(data->offset[i]);
         }
 
+        /* TODO: better debug? */
+        /*
         cli_dbgmsg("info->fsize: %lu\n", (long unsigned)info->fsize);
         if (pm->offdata[0]>9)
             cli_dbgmsg("offdata[0] type:     %x\n", pm->offdata[0]);
@@ -279,7 +305,7 @@ int cli_pcre_recaloff(struct cli_matcher *root, struct cli_pcre_off *data, struc
         cli_dbgmsg("offdata[3] section:  %u\n", pm->offdata[3]);
         cli_dbgmsg("offset_min: %u\n", pm->offset_min);
         cli_dbgmsg("offset_max: %u\n", pm->offset_max);
-
+        */
     }
 
     for (i = 0; i < root->pcre_metas; ++i) {
@@ -360,7 +386,7 @@ int cli_pcre_scanbuf(const unsigned char *buffer, uint32_t length, const struct
     uint32_t global, encompass;
     int rc, offset, ovector[OVECCOUNT];
 
-    if (!root->pcre_metatable) {
+    if (!(ctx->dconf->pcre & PCRE_CONF_SUPPORT) || (!root->pcre_metatable)) {
         return CL_SUCCESS;
     }
 
@@ -368,6 +394,12 @@ int cli_pcre_scanbuf(const unsigned char *buffer, uint32_t length, const struct
         pm = root->pcre_metatable[i];
         pd = &(pm->pdata);
 
+        /* skip checking and running disabled pcres */
+        if (pm->flags & CLI_PCRE_DISABLED) {
+            cli_dbgmsg("cli_pcre_scanbuf: skipping disabled regex /%s/\n", pd->expression);
+            continue;
+        }
+
         /* check the evaluation of the trigger */
         if (pm->lsigid[0]) {
             cli_dbgmsg("cli_pcre_scanbuf: checking %s; running regex /%s/\n", pm->trigger, pd->expression);

libclamav/matcher-pcre.h

@@ -33,6 +33,7 @@
 #include <sys/types.h>
 
 #include "cltypes.h"
+#include "dconf.h"
 #include "mpool.h"
 #include "regex_pcre.h"
 
@@ -40,6 +41,8 @@
 #define CLI_PCRE_GLOBAL    0x00000001 /* g */
 #define CLI_PCRE_ENCOMPASS 0x00000002 /* e */
 
+#define CLI_PCRE_DISABLED  0x80000000 /* used for dconf or fail to build */
+
 struct cli_pcre_meta {
     char *trigger;
     uint32_t lsigid[3]; /* 0=valid, 1=lsigid, 2=subsigid */
@@ -57,8 +60,8 @@ struct cli_pcre_off {
 };
 
 int cli_pcre_addpatt(struct cli_matcher *root, const char *trigger,  const char *pattern, const char *cflags, const char *offset, const uint32_t *lsigid);
-int cli_pcre_build(struct cli_matcher *root, long long unsigned match_limit, long long unsigned recmatch_limit);
-int cli_pcre_recaloff(struct cli_matcher *root, struct cli_pcre_off *data, struct cli_target_info *info);
+int cli_pcre_build(struct cli_matcher *root, long long unsigned match_limit, long long unsigned recmatch_limit, const struct cli_dconf *dconf);
+int cli_pcre_recaloff(struct cli_matcher *root, struct cli_pcre_off *data, struct cli_target_info *info, cli_ctx *ctx);
 void cli_pcre_freeoff(struct cli_pcre_off *data);
 int cli_pcre_scanbuf(const unsigned char *buffer, uint32_t length, const struct cli_matcher *root, struct cli_ac_data *mdata,  struct cli_ac_result **res, const struct cli_pcre_off *data, cli_ctx *ctx);
 void cli_pcre_freemeta(struct cli_pcre_meta *pm);

libclamav/matcher.c

@@ -967,7 +967,7 @@ int cli_fmap_scandesc(cli_ctx *ctx, cli_file_t ftype, uint8_t ftonly, struct cli
             struct cli_pcre_off poff;
 
             /* calculate the relative offsets */
-            ret = cli_pcre_recaloff(groot, &poff, &info);
+            ret = cli_pcre_recaloff(groot, &poff, &info, ctx);
             if (ret != CL_SUCCESS) {
                 cli_ac_freedata(&gdata);
                 cli_ac_freedata(&tdata);
@@ -1009,7 +1009,7 @@ int cli_fmap_scandesc(cli_ctx *ctx, cli_file_t ftype, uint8_t ftonly, struct cli
             struct cli_pcre_off poff;
 
             /* calculate the relative offsets */
-            ret = cli_pcre_recaloff(troot, &poff, &info);
+            ret = cli_pcre_recaloff(troot, &poff, &info, ctx);
             if (ret != CL_SUCCESS) {
                 cli_ac_freedata(&gdata);
                 cli_ac_freedata(&tdata);

libclamav/readdb.c

@@ -3505,7 +3505,7 @@ int cl_engine_compile(struct cl_engine *engine)
 	    if((ret = cli_ac_buildtrie(root)))
 		return ret;
 #if HAVE_PCRE
-            if((ret = cli_pcre_build(root, engine->pcre_match_limit, engine->pcre_recmatch_limit)))
+            if((ret = cli_pcre_build(root, engine->pcre_match_limit, engine->pcre_recmatch_limit, engine->dconf)))
                 return ret;
 
 	    cli_dbgmsg("Matcher[%u]: %s: AC sigs: %u (reloff: %u, absoff: %u) BM sigs: %u (reloff: %u, absoff: %u) PCREs: %u (reloff: %u, absoff: %u) maxpatlen %u %s\n", i, cli_mtargets[i].name, root->ac_patterns, root->ac_reloff_num, root->ac_absoff_num, root->bm_patterns, root->bm_reloff_num, root->bm_absoff_num, root->pcre_metas, root->pcre_reloff_num, root->pcre_absoff_num, root->maxpatlen, root->ac_only ? "(ac_only mode)" : "");