Add a bit more sanity checking to pcre2_serialize_decode(), and document.

ChangeLog

@@ -113,6 +113,9 @@ compiler warning.
 
 27. Minor code refactor to avoid "left shift of negative number" warning. 
 
+28. Add a bit more sanity checking to pcre2_serialize_decode() and document 
+that it expects trusted data.
+
 
 Version 10.21 12-January-2016
 -----------------------------

doc/pcre2serialize.3

@@ -1,4 +1,4 @@
-.TH PCRE2SERIALIZE 3 "03 November 2015" "PCRE2 10.21"
+.TH PCRE2SERIALIZE 3 "24 May 2016" "PCRE2 10.22"
 .SH NAME
 PCRE2 - Perl-compatible regular expressions (revised API)
 .SH "SAVING AND RE-USING PRECOMPILED PCRE2 PATTERNS"
@@ -30,6 +30,16 @@ PCRE2's 16-bit library cannot be reloaded on a 64-bit system, nor can they be
 reloaded using the 8-bit library.
 .
 .
+.SH "SECURITY CONCERNS"
+.rs
+.sp
+The facility for saving and restoring compiled patterns is intended for use 
+within individual applications. As such, the data supplied to
+\fBpcre2_serialize_decode()\fP is expected to be trusted data, not data from
+arbitrary external sources. There is only some simple consistency checking, not 
+complete validation of what is being re-loaded.
+.
+.
 .SH "SAVING COMPILED PATTERNS"
 .rs
 .sp
@@ -129,11 +139,12 @@ is filled with those that fit, and the remainder are ignored. The yield of the
 function is the number of decoded patterns, or one of the following negative
 error codes:
 .sp
-  PCRE2_ERROR_BADDATA   second argument is zero or less
+  PCRE2_ERROR_BADDATA    second argument is zero or less
-  PCRE2_ERROR_BADMAGIC  mismatch of id bytes in the data
+  PCRE2_ERROR_BADMAGIC   mismatch of id bytes in the data
-  PCRE2_ERROR_BADMODE   mismatch of variable unit size or PCRE2 version
+  PCRE2_ERROR_BADMODE    mismatch of code unit size or PCRE2 version
-  PCRE2_ERROR_MEMORY    memory allocation failed
+  PCRE2_ERROR_BADSERIALIZEDDATA  other sanity check failure
-  PCRE2_ERROR_NULL      first or third argument is NULL
+  PCRE2_ERROR_MEMORY     memory allocation failed
+  PCRE2_ERROR_NULL       first or third argument is NULL
 .sp
 PCRE2_ERROR_BADMAGIC may mean that the data is corrupt, or that it was compiled
 on a system with different endianness.
@@ -170,6 +181,6 @@ Cambridge, England.
 .rs
 .sp
 .nf
-Last updated: 03 November 2015
+Last updated: 24 May 2016
-Copyright (c) 1997-2015 University of Cambridge.
+Copyright (c) 1997-2016 University of Cambridge.
 .fi

src/pcre2.h

@@ -245,6 +245,7 @@ numbers must not be changed. */
 #define PCRE2_ERROR_BADSUBSTITUTION   (-59)
 #define PCRE2_ERROR_BADSUBSPATTERN    (-60)
 #define PCRE2_ERROR_TOOMANYREPLACE    (-61)
+#define PCRE2_ERROR_BADSERIALIZEDDATA (-62)
 
 /* Request types for pcre2_pattern_info() */
 

src/pcre2.h.in

@@ -245,6 +245,7 @@ numbers must not be changed. */
 #define PCRE2_ERROR_BADSUBSTITUTION   (-59)
 #define PCRE2_ERROR_BADSUBSPATTERN    (-60)
 #define PCRE2_ERROR_TOOMANYREPLACE    (-61)
+#define PCRE2_ERROR_BADSERIALIZEDDATA (-62)
 
 /* Request types for pcre2_pattern_info() */
 

src/pcre2_error.c

@@ -252,6 +252,7 @@ static const unsigned char match_error_texts[] =
   /* 60 */
   "match with end before start is not supported\0"
   "too many replacements (more than INT_MAX)\0"
+  "bad serialized data\0" 
   ;
 
 

src/pcre2_serialize.c

@@ -158,6 +158,7 @@ int32_t i, j;
 
 if (data == NULL || codes == NULL) return PCRE2_ERROR_NULL;
 if (number_of_codes <= 0) return PCRE2_ERROR_BADDATA;
+if (data->number_of_codes <= 0) return PCRE2_ERROR_BADSERIALIZEDDATA;
 if (data->magic != SERIALIZED_DATA_MAGIC) return PCRE2_ERROR_BADMAGIC;
 if (data->version != SERIALIZED_DATA_VERSION) return PCRE2_ERROR_BADMODE;
 if (data->config != SERIALIZED_DATA_CONFIG) return PCRE2_ERROR_BADMODE;
@@ -188,6 +189,8 @@ for (i = 0; i < number_of_codes; i++)
   CODE_BLOCKSIZE_TYPE blocksize;
   memcpy(&blocksize, src_bytes + offsetof(pcre2_real_code, blocksize),
     sizeof(CODE_BLOCKSIZE_TYPE));
+  if (blocksize <= sizeof(pcre2_real_code))
+    return PCRE2_ERROR_BADSERIALIZEDDATA;
 
   /* The allocator provided by gcontext replaces the original one. */
 
@@ -208,6 +211,10 @@ for (i = 0; i < number_of_codes; i++)
 
   memcpy(((uint8_t *)dst_re) + sizeof(pcre2_memctl),
     src_bytes + sizeof(pcre2_memctl), blocksize - sizeof(pcre2_memctl));
+  if (dst_re->magic_number != MAGIC_NUMBER ||
+      dst_re->name_entry_size > MAX_NAME_SIZE + IMM2_SIZE + 1 ||
+      dst_re->name_count > MAX_NAME_COUNT)
+    return PCRE2_ERROR_BADSERIALIZEDDATA;
 
   /* At the moment only one table is supported. */