Fix a crash when finding a Unicode property for a character with a code point greater than 0x10ffff

This commit is contained in:
Petr Písař 2017-02-27 15:08:39 +01:00
parent ec2d617244
commit c36d2c5e3d
2 changed files with 81 additions and 1 deletions

View File

@ -0,0 +1,72 @@
From 8037f71d03b3cd8919248f38448a0a2d3715c18c Mon Sep 17 00:00:00 2001
From: ph10 <ph10@2f5784b3-3f2a-0410-8824-cb99058d5e15>
Date: Fri, 24 Feb 2017 17:30:30 +0000
Subject: [PATCH] Fix Unicode property crash for 32-bit characters greater than
0x10ffff.
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
git-svn-id: svn://vcs.exim.org/pcre/code/trunk@1688 2f5784b3-3f2a-0410-8824-cb99058d5e15
Petr Písař: Ported to 8.40.
diff --git a/pcre_internal.h b/pcre_internal.h
index 2923b29..154d3f6 100644
--- 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 69c4fd4..f22f826 100644
--- 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):
--
2.7.4

View File

@ -55,6 +55,10 @@ Patch7: pcre-8.40-Make-pcretest-check-size-of-O-argument.patch
# Document pcretest input cannot contain binary zeroes, upstream bug #2045,
# in upstream after 8.40
Patch8: pcre-8.40-Minor-doc-update.patch
# Fix a crash when finding a Unicode property for a character with a code
# point greater than 0x10ffff in UTF-32 library while UTF mode is disabled,
# upstream bug #2052, in upstream after 8.40
Patch9: pcre-8.40-Fix-Unicode-property-crash-for-32-bit-characters-gre.patch
BuildRequires: readline-devel
BuildRequires: autoconf
BuildRequires: automake
@ -151,6 +155,7 @@ Utilities demonstrating PCRE capabilities like pcregrep or pcretest.
%patch6 -p1
%patch7 -p1
%patch8 -p1
%patch9 -p1
# Because of rpath patch
libtoolize --copy --force
autoreconf -vif
@ -247,10 +252,13 @@ make %{?_smp_mflags} check VERBOSE=yes
%{_mandir}/man1/pcretest.*
%changelog
* Fri Feb 24 2017 Petr Pisar <ppisar@redhat.com> - 8.40-5
* Mon Feb 27 2017 Petr Pisar <ppisar@redhat.com> - 8.40-5
- Fix a crash in pcretest when \O directive was supplied with too big number
(upstream bug #2044)
- Document pcretest input cannot contain binary zeroes (upstream bug #2045)
- Fix a crash when finding a Unicode property for a character with a code
point greater than 0x10ffff in UTF-32 library while UTF mode is disabled
(upstream bug #2052)
* Thu Feb 23 2017 Petr Pisar <ppisar@redhat.com> - 8.40-4
- Fix a crash in pcretest when printing non-ASCII characters