From: Eugene V. Lyubimkin <jackyf@debian.org>
Subject: Fix a DoS in Unicode processing [CVE-2009-3626]
Bug-Debian: http://bugs.debian.org/552291
Bug: http://rt.perl.org/rt3/Public/Bug/Display.html?id=69973
Origin: upstream, http://perl5.git.perl.org/perl.git/commit/0abd0d78a73da1c4d13b1c700526b7e5d03b32d4.

Resolves segmentation fault in some tricky tainted non-UTF-8 matches.

Signed-off-by: Eugene V. Lyubimkin <jackyf@debian.org>

---
 ext/re/t/regop.t |   12 ++++++------
 regcomp.c        |   17 +++++++++++------
 regexec.c        |    9 ++-------
 3 files changed, 19 insertions(+), 19 deletions(-)

diff --git a/ext/re/t/regop.t b/ext/re/t/regop.t
index 7fe7b20..f111b91 100755
--- a/ext/re/t/regop.t
+++ b/ext/re/t/regop.t
@@ -233,12 +233,12 @@ anchored "ABC" at 0
 #Freeing REx: "(\\.COM|\\.EXE|\\.BAT|\\.CMD|\\.VBS|\\.VBE|\\.JS|\\.JSE|\\."......
 %MATCHED%
 floating ""$ at 3..4 (checking floating)
-1:1[1] 3:2[1] 5:2[64] 45:83[1] 47:84[1] 48:85[0]
-stclass EXACTF <.> minlen 3
-Found floating substr ""$ at offset 30...
-Does not contradict STCLASS...
-Guessed: match at offset 26
-Matching stclass EXACTF <.> against ".exe"
+#1:1[1] 3:2[1] 5:2[64] 45:83[1] 47:84[1] 48:85[0]
+#stclass EXACTF <.> minlen 3
+#Found floating substr ""$ at offset 30...
+#Does not contradict STCLASS...
+#Guessed: match at offset 26
+#Matching stclass EXACTF <.> against ".exe"
 ---
 #Compiling REx "[q]"
 #size 12 nodes Got 100 bytes for offset annotations.
diff --git a/regcomp.c b/regcomp.c
index 49e69b2..b7fb032 100644
--- a/regcomp.c
+++ b/regcomp.c
@@ -2820,13 +2820,18 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                                 }
                             } else {
 /* 
-    Currently we assume that the trie can handle unicode and ascii
-    matches fold cased matches. If this proves true then the following
-    define will prevent tries in this situation. 
-    
-    #define TRIE_TYPE_IS_SAFE (UTF || optype==EXACT)
-*/
+    Currently we do not believe that the trie logic can
+    handle case insensitive matching properly when the
+    pattern is not unicode (thus forcing unicode semantics).
+
+    If/when this is fixed the following define can be swapped
+    in below to fully enable trie logic.
+
 #define TRIE_TYPE_IS_SAFE 1
+
+*/
+#define TRIE_TYPE_IS_SAFE (UTF || optype==EXACT)
+
                                 if ( last && TRIE_TYPE_IS_SAFE ) {
                                     make_trie( pRExC_state, 
                                             startbranch, first, cur, tail, count, 
diff --git a/regexec.c b/regexec.c
index 7a42c4f..32994de 100644
--- a/regexec.c
+++ b/regexec.c
@@ -1006,16 +1006,15 @@ Perl_re_intuit_start(pTHX_ REGEXP * const prog, SV *sv, char *strpos,
 
 #define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len,  \
 uvc, charid, foldlen, foldbuf, uniflags) STMT_START {                       \
-    UV uvc_unfolded = 0;						    \
     switch (trie_type) {                                                    \
     case trie_utf8_fold:                                                    \
 	if ( foldlen>0 ) {                                                  \
-	    uvc_unfolded = uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \
+	    uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \
 	    foldlen -= len;                                                 \
 	    uscan += len;                                                   \
 	    len=0;                                                          \
 	} else {                                                            \
-	    uvc_unfolded = uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
+	    uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \
 	    uvc = to_uni_fold( uvc, foldbuf, &foldlen );                    \
 	    foldlen -= UNISKIP( uvc );                                      \
 	    uscan = foldbuf + UNISKIP( uvc );                               \
@@ -1041,7 +1040,6 @@ uvc, charid, foldlen, foldbuf, uniflags) STMT_START {                       \
 	uvc = (UV)*uc;                                                      \
 	len = 1;                                                            \
     }                                                                       \
-									    \
     if (uvc < 256) {                                                        \
 	charid = trie->charmap[ uvc ];                                      \
     }                                                                       \
@@ -1054,9 +1052,6 @@ uvc, charid, foldlen, foldbuf, uniflags) STMT_START {                       \
 		charid = (U16)SvIV(*svpp);                                  \
 	}                                                                   \
     }                                                                       \
-    if (!charid && trie_type == trie_utf8_fold && !UTF) {		    \
-	charid = trie->charmap[uvc_unfolded];			    	    \
-    }								    	    \
 } STMT_END
 
 #define REXEC_FBC_EXACTISH_CHECK(CoNd)                 \
-- 
tg: (daf8b46..) fixes/trie-logic-match (depends on: upstream)