Upstream fix for CVE-2017-7186 (Upstream rev 1688)

Fix Unicode property crash for 32-bit characters greater than 0x10ffff.

diff --git a/pcre_internal.h b/pcre_internal.h
index 2923b29f82deccd1c63bda1eb726e52f04f88e11..154d3f62a38ddee59f70d7d686d4ad05fb31a216 100644 (file)
--- a/pcre_internal.h
+++ b/pcre_internal.h
@@ -2772,6 +2772,9 @@ extern const pcre_uint8  PRIV(ucd_stage1)[];
 extern const pcre_uint16 PRIV(ucd_stage2)[];
 extern const pcre_uint32 PRIV(ucp_gentype)[];
 extern const pcre_uint32 PRIV(ucp_gbtable)[];
+#ifdef COMPILE_PCRE32
+extern const ucd_record  PRIV(dummy_ucd_record)[];
+#endif
 #ifdef SUPPORT_JIT
 extern const int         PRIV(ucp_typerange)[];
 #endif
@@ -2780,9 +2783,15 @@ extern const int         PRIV(ucp_typerange)[];
 /* UCD access macros */
 
 #define UCD_BLOCK_SIZE 128
-#define GET_UCD(ch) (PRIV(ucd_records) + \
+#define REAL_GET_UCD(ch) (PRIV(ucd_records) + \
         PRIV(ucd_stage2)[PRIV(ucd_stage1)[(int)(ch) / UCD_BLOCK_SIZE] * \
         UCD_BLOCK_SIZE + (int)(ch) % UCD_BLOCK_SIZE])
+        
+#ifdef COMPILE_PCRE32
+#define GET_UCD(ch) ((ch > 0x10ffff)? PRIV(dummy_ucd_record) : REAL_GET_UCD(ch))
+#else
+#define GET_UCD(ch) REAL_GET_UCD(ch)
+#endif 
 
 #define UCD_CHARTYPE(ch)    GET_UCD(ch)->chartype
 #define UCD_SCRIPT(ch)      GET_UCD(ch)->script

diff --git a/pcre_ucd.c b/pcre_ucd.c
index 69c4fd42c34b2dd449779699856f2e183d90af99..f22f826c4c245106b13114f81277cc4324fdb62a 100644 (file)
--- a/pcre_ucd.c
+++ b/pcre_ucd.c
@@ -38,6 +38,20 @@ const pcre_uint16 PRIV(ucd_stage2)[] = {0};
 const pcre_uint32 PRIV(ucd_caseless_sets)[] = {0};
 #else
 
+/* If the 32-bit library is run in non-32-bit mode, character values
+greater than 0x10ffff may be encountered. For these we set up a
+special record. */
+
+#ifdef COMPILE_PCRE32
+const ucd_record PRIV(dummy_ucd_record)[] = {{
+  ucp_Common,    /* script */
+  ucp_Cn,        /* type unassigned */
+  ucp_gbOther,   /* grapheme break property */
+  0,             /* case set */
+  0,             /* other case */
+  }};
+#endif
+
 /* When recompiling tables with a new Unicode version, please check the
 types in this structure definition from pcre_internal.h (the actual
 field names will be different):