New upstream version 6.9.6upstream/6.9.6

2 files changed, 117 insertions, 58 deletions

diff --git a/harnesses/base.c b/harnesses/base.c
index a88e6f2..1206217 100644
--- a/harnesses/base.c
+++ b/harnesses/base.c
@@ -10,16 +10,32 @@
 #include <sys/stat.h>
 #include <fcntl.h>
 #include <time.h>
-
 #include "oniguruma.h"
 
 #define PARSE_DEPTH_LIMIT           8
-#define RETRY_LIMIT              5000
 #define CALL_MAX_NEST_LEVEL         8
+#define SUBEXP_CALL_LIMIT         500
+#define BASE_RETRY_LIMIT        20000
+#define BASE_LENGTH              2048
+#define MATCH_STACK_LIMIT    10000000
+#define MAX_REM_SIZE          1048576
+#define MAX_SLOW_REM_SIZE        1024
+#define SLOW_RETRY_LIMIT         2000
+
 //#define EXEC_PRINT_INTERVAL    500000
 //#define DUMP_DATA_INTERVAL     100000
 //#define STAT_PATH              "fuzzer.stat_log"
 
+#define OPTIONS_AT_COMPILE   (ONIG_OPTION_IGNORECASE | ONIG_OPTION_EXTEND | ONIG_OPTION_MULTILINE | ONIG_OPTION_SINGLELINE | ONIG_OPTION_FIND_LONGEST | ONIG_OPTION_FIND_NOT_EMPTY | ONIG_OPTION_NEGATE_SINGLELINE | ONIG_OPTION_DONT_CAPTURE_GROUP | ONIG_OPTION_CAPTURE_GROUP | ONIG_OPTION_WORD_IS_ASCII | ONIG_OPTION_DIGIT_IS_ASCII | ONIG_OPTION_SPACE_IS_ASCII | ONIG_OPTION_POSIX_IS_ASCII | ONIG_OPTION_TEXT_SEGMENT_EXTENDED_GRAPHEME_CLUSTER | ONIG_OPTION_TEXT_SEGMENT_WORD )
+
+#define OPTIONS_AT_RUNTIME   (ONIG_OPTION_NOTBOL | ONIG_OPTION_NOTEOL | ONIG_OPTION_CHECK_VALIDITY_OF_STRING | ONIG_OPTION_NOT_BEGIN_STRING | ONIG_OPTION_NOT_END_STRING | ONIG_OPTION_NOT_BEGIN_POSITION)
+
+
+#define ADJUST_LEN(enc, len) do {\
+  int mlen = ONIGENC_MBC_MINLEN(enc);\
+  if (mlen != 1) { len -= len % mlen; }\
+} while (0)
+
 typedef unsigned char uint8_t;
 
 #ifdef DUMP_INPUT
@@ -103,14 +119,34 @@ output_current_time(FILE* fp)
 #endif
 
 static int
-search(regex_t* reg, unsigned char* str, unsigned char* end, int backward)
+search(regex_t* reg, unsigned char* str, unsigned char* end, OnigOptionType options, int backward, int sl)
 {
   int r;
   unsigned char *start, *range;
   OnigRegion *region;
+  unsigned int retry_limit;
+  size_t len;
 
   region = onig_region_new();
 
+  len = (size_t )(end - str);
+  if (len < BASE_LENGTH) {
+    if (sl >= 2)
+      retry_limit = (unsigned int )SLOW_RETRY_LIMIT;
+    else
+      retry_limit = (unsigned int )BASE_RETRY_LIMIT;
+  }
+  else
+    retry_limit = (unsigned int )(BASE_RETRY_LIMIT * BASE_LENGTH / len);
+
+#ifdef STANDALONE
+  fprintf(stdout, "retry limit: %u\n", retry_limit);
+#endif
+
+  onig_set_retry_limit_in_search(retry_limit);
+  onig_set_match_stack_limit_size(MATCH_STACK_LIMIT);
+  onig_set_subexp_call_limit_in_search(SUBEXP_CALL_LIMIT);
+
   if (backward != 0) {
     start = end;
     range = str;
@@ -120,7 +156,7 @@ search(regex_t* reg, unsigned char* str, unsigned char* end, int backward)
     range = end;
   }
 
-  r = onig_search(reg, str, end, start, range, region, ONIG_OPTION_NONE);
+  r = onig_search(reg, str, end, start, range, region, (options & OPTIONS_AT_RUNTIME));
   if (r >= 0) {
 #ifdef STANDALONE
     int i;
@@ -168,7 +204,8 @@ static long VALID_STRING_COUNT;
 
 static int
 exec(OnigEncoding enc, OnigOptionType options, OnigSyntaxType* syntax,
-     char* apattern, char* apattern_end, char* astr, UChar* end, int backward)
+     char* apattern, char* apattern_end, char* astr, UChar* end, int backward,
+     int sl)
 {
   int r;
   regex_t* reg;
@@ -181,14 +218,13 @@ exec(OnigEncoding enc, OnigOptionType options, OnigSyntaxType* syntax,
   EXEC_COUNT_INTERVAL++;
 
   onig_initialize(&enc, 1);
-  onig_set_retry_limit_in_search(RETRY_LIMIT);
 #ifdef PARSE_DEPTH_LIMIT
   onig_set_parse_depth_limit(PARSE_DEPTH_LIMIT);
 #endif
   onig_set_subexp_call_max_nest_level(CALL_MAX_NEST_LEVEL);
 
   r = onig_new(&reg, pattern, pattern_end,
-               options, enc, syntax, &einfo);
+               (options & OPTIONS_AT_COMPILE), enc, syntax, &einfo);
   if (r != ONIG_NORMAL) {
     char s[ONIG_MAX_ERROR_MESSAGE_LEN];
     onig_error_code_to_str((UChar* )s, r, &einfo);
@@ -208,12 +244,12 @@ exec(OnigEncoding enc, OnigOptionType options, OnigSyntaxType* syntax,
   }
   REGEX_SUCCESS_COUNT++;
 
-  r = search(reg, pattern, pattern_end, backward);
+  r = search(reg, pattern, pattern_end, options, backward, sl);
   if (r == -2) return -2;
 
   if (onigenc_is_valid_mbc_string(enc, str, end) != 0) {
     VALID_STRING_COUNT++;
-    r = search(reg, str, end, backward);
+    r = search(reg, str, end, options, backward, sl);
     if (r == -2) return -2;
   }
 
@@ -224,43 +260,52 @@ exec(OnigEncoding enc, OnigOptionType options, OnigSyntaxType* syntax,
 
 static int
 alloc_exec(OnigEncoding enc, OnigOptionType options, OnigSyntaxType* syntax,
-           int backward, int pattern_size, size_t remaining_size, unsigned char *data)
+           int backward, int pattern_size, size_t rem_size, unsigned char *data)
 {
+  extern int onig_detect_can_be_slow_pattern(const UChar* pattern, const UChar* pattern_end, OnigOptionType option, OnigEncoding enc, OnigSyntaxType* syntax);
+
   int r;
+  int sl;
+  unsigned char *pattern;
   unsigned char *pattern_end;
   unsigned char *str_null_end;
 
-  // copy first PATTERN_SIZE bytes off to be the pattern
-  unsigned char *pattern = (unsigned char *)malloc(pattern_size != 0 ? pattern_size : 1);
+  pattern = (unsigned char *)malloc(pattern_size != 0 ? pattern_size : 1);
   memcpy(pattern, data, pattern_size);
   pattern_end = pattern + pattern_size;
   data += pattern_size;
-  remaining_size -= pattern_size;
+  rem_size -= pattern_size;
+
+  if (rem_size > MAX_REM_SIZE) rem_size = MAX_REM_SIZE;
+
+  sl = onig_detect_can_be_slow_pattern(pattern, pattern_end, options, enc, syntax);
+  if (sl > 0) {
+    if (rem_size > MAX_SLOW_REM_SIZE)
+      rem_size = MAX_SLOW_REM_SIZE;
+  }
 
-#if defined(UTF16_BE) || defined(UTF16_LE)
-  if (remaining_size % 2 == 1) remaining_size--;
+  ADJUST_LEN(enc, rem_size);
+#ifdef STANDALONE
+  fprintf(stdout, "rem_size: %ld\n", rem_size);
 #endif
 
-  unsigned char *str = (unsigned char*)malloc(remaining_size != 0 ? remaining_size : 1);
-  memcpy(str, data, remaining_size);
-  str_null_end = str + remaining_size;
+  unsigned char *str = (unsigned char*)malloc(rem_size != 0 ? rem_size : 1);
+  memcpy(str, data, rem_size);
+  str_null_end = str + rem_size;
 
   r = exec(enc, options, syntax,
            (char *)pattern, (char *)pattern_end,
-           (char *)str, str_null_end, backward);
+           (char *)str, str_null_end, backward, sl);
 
   free(pattern);
   free(str);
   return r;
 }
 
-#define OPTIONS_MASK  (ONIG_OPTION_IGNORECASE | ONIG_OPTION_EXTEND | ONIG_OPTION_MULTILINE | ONIG_OPTION_SINGLELINE | ONIG_OPTION_FIND_LONGEST | ONIG_OPTION_FIND_NOT_EMPTY | ONIG_OPTION_NEGATE_SINGLELINE | ONIG_OPTION_DONT_CAPTURE_GROUP | ONIG_OPTION_CAPTURE_GROUP)
-
-
 #ifdef SYNTAX_TEST
-#define NUM_CONTROL_BYTES      6
+#define NUM_CONTROL_BYTES      7
 #else
-#define NUM_CONTROL_BYTES      5
+#define NUM_CONTROL_BYTES      6
 #endif
 
 int LLVMFuzzerTestOneInput(const uint8_t * Data, size_t Size)
@@ -285,14 +330,14 @@ int LLVMFuzzerTestOneInput(const uint8_t * Data, size_t Size)
     ONIG_ENCODING_CP1251,
     ONIG_ENCODING_BIG5,
     ONIG_ENCODING_GB18030,
-    ONIG_ENCODING_UTF8,
-    ONIG_ENCODING_UTF8,
-    ONIG_ENCODING_UTF8,
-    ONIG_ENCODING_UTF8,
-    ONIG_ENCODING_UTF8,
-    ONIG_ENCODING_UTF8,
-    ONIG_ENCODING_UTF8,
-    ONIG_ENCODING_UTF8,
+    ONIG_ENCODING_UTF16_BE,
+    ONIG_ENCODING_UTF16_LE,
+    ONIG_ENCODING_UTF16_BE,
+    ONIG_ENCODING_UTF16_LE,
+    ONIG_ENCODING_UTF32_BE,
+    ONIG_ENCODING_UTF32_LE,
+    ONIG_ENCODING_UTF32_BE,
+    ONIG_ENCODING_UTF32_LE,
     ONIG_ENCODING_ISO_8859_1,
     ONIG_ENCODING_ISO_8859_2,
     ONIG_ENCODING_ISO_8859_3,
@@ -341,7 +386,7 @@ int LLVMFuzzerTestOneInput(const uint8_t * Data, size_t Size)
   int r;
   int backward;
   int pattern_size;
-  size_t remaining_size;
+  size_t rem_size;
   unsigned char *data;
   unsigned char pattern_size_choice;
   OnigOptionType  options;
@@ -364,7 +409,7 @@ int LLVMFuzzerTestOneInput(const uint8_t * Data, size_t Size)
 
   if (Size < NUM_CONTROL_BYTES) return 0;
 
-  remaining_size = Size;
+  rem_size = Size;
   data = (unsigned char* )(Data);
 
 #ifdef UTF16_BE
@@ -375,7 +420,7 @@ int LLVMFuzzerTestOneInput(const uint8_t * Data, size_t Size)
 #else
   encoding_choice = data[0];
   data++;
-  remaining_size--;
+  rem_size--;
 
   int num_encodings = sizeof(encodings)/sizeof(encodings[0]);
   enc = encodings[encoding_choice % num_encodings];
@@ -385,7 +430,7 @@ int LLVMFuzzerTestOneInput(const uint8_t * Data, size_t Size)
 #ifdef SYNTAX_TEST
   syntax_choice = data[0];
   data++;
-  remaining_size--;
+  rem_size--;
 
   int num_syntaxes = sizeof(syntaxes)/sizeof(syntaxes[0]);
   syntax = syntaxes[syntax_choice % num_syntaxes];
@@ -393,31 +438,30 @@ int LLVMFuzzerTestOneInput(const uint8_t * Data, size_t Size)
   syntax = ONIG_SYNTAX_DEFAULT;
 #endif
 
-  if ((data[1] & 0xc0) == 0)
-    options = (data[0] | (data[1] << 8)) & OPTIONS_MASK;
+  if ((data[2] & 0xc0) == 0)
+    options = data[0] | (data[1] << 8) | (data[2] << 16);
   else
     options = data[0] & ONIG_OPTION_IGNORECASE;
 
-  data++;
-  remaining_size--;
-  data++;
-  remaining_size--;
+  data++; rem_size--;
+  data++; rem_size--;
+  data++; rem_size--;
 
   pattern_size_choice = data[0];
-  data++;
-  remaining_size--;
+  data++; rem_size--;
 
   backward = (data[0] == 0xbb);
-  data++;
-  remaining_size--;
+  data++; rem_size--;
 
-  if (remaining_size == 0)
+  if (backward != 0) {
+    options = options & ~ONIG_OPTION_FIND_LONGEST;
+  }
+
+  if (rem_size == 0)
     pattern_size = 0;
   else {
-    pattern_size = (int )pattern_size_choice % remaining_size;
-#if defined(UTF16_BE) || defined(UTF16_LE)
-    if (pattern_size % 2 == 1) pattern_size--;
-#endif
+    pattern_size = (int )pattern_size_choice % rem_size;
+    ADJUST_LEN(enc, pattern_size);
   }
 
 #ifdef STANDALONE
@@ -440,7 +484,7 @@ int LLVMFuzzerTestOneInput(const uint8_t * Data, size_t Size)
 #endif
 
   r = alloc_exec(enc, options, syntax, backward, pattern_size,
-                 remaining_size, data);
+                 rem_size, data);
   if (r == -2) exit(-2);
 
 #ifndef STANDALONE
@@ -485,15 +529,25 @@ int LLVMFuzzerTestOneInput(const uint8_t * Data, size_t Size)
 
 #ifdef STANDALONE
 
+#define MAX_INPUT_DATA_SIZE  4194304
+
 extern int main(int argc, char* argv[])
 {
+  size_t max_size;
   size_t n;
-  uint8_t Data[10000];
+  uint8_t Data[MAX_INPUT_DATA_SIZE];
 
-  n = read(0, Data, sizeof(Data));
-  fprintf(stdout, "n: %ld\n", n);
-  LLVMFuzzerTestOneInput(Data, n);
+  if (argc > 1) {
+    max_size = (size_t )atoi(argv[1]);
+  }
+  else {
+    max_size = sizeof(Data);
+  }
 
+  n = read(0, Data, max_size);
+  fprintf(stdout, "read size: %ld, max_size: %ld\n", n, max_size);
+
+  LLVMFuzzerTestOneInput(Data, n);
   return 0;
 }
 #endif /* STANDALONE */
diff --git a/harnesses/makefile b/harnesses/makefile
index b324295..d4fcfb6 100644
--- a/harnesses/makefile
+++ b/harnesses/makefile
@@ -1,4 +1,8 @@
 # makefile for harness
+DEBUG_OUT =
+#DEBUG_OUT = -DONIG_DEBUG_PARSE -DONIG_DEBUG_COMPILE
+#DEBUG_OUT = -DONIG_DEBUG_PARSE -DONIG_DEBUG_COMPILE -DONIG_DEBUG_MATCH_COUNTER
+
 SRC = ../src
 CFLAGS   = -I$(SRC) -Wall -g -fsanitize=fuzzer,address -fno-omit-frame-pointer
 CFLAGS_M = -I$(SRC) -Wall -g -fsanitize=fuzzer-no-link,address -fno-omit-frame-pointer -DSTANDALONE
@@ -12,7 +16,8 @@ TARGETS = fuzzer-encode fuzzer-syntax fuzzer-utf16-be fuzzer-utf16-le \
 OTHER_TARGETS = libfuzzer-onig libfuzzer-onig-full fuzzer-deluxe read-deluxe
 
 
-default: $(TARGETS)
+#default: $(TARGETS)
+default: read-syntax
 
 fuzzer-encode: base.c $(ONIG_LIB)
 	clang $(CFLAGS) $< $(LIBS) -o $@
@@ -60,7 +65,7 @@ libfuzzer-onig-full: libfuzzer-onig.cpp $(ONIG_LIB)
 $(ONIG_LIB):
 	cd ..; make clean
 	#cd ..; autoreconf -vfi
-	cd ..; ./configure CC=clang LD=clang CFLAGS="-g -fsanitize=address -fno-omit-frame-pointer" LDFLAGS="-g -fsanitize=address -fno-omit-frame-pointer"
+	cd ..; ./configure CC=clang LD=clang CFLAGS="-g -fsanitize=address -fno-omit-frame-pointer $(DEBUG_OUT)" LDFLAGS="-g -fsanitize=address -fno-omit-frame-pointer"
 	cd ..; make -j4