Revision: 635
http://www.exim.org/viewvc/pcre2?view=rev&revision=635
Author: ph10
Date: 2016-12-31 13:35:31 +0000 (Sat, 31 Dec 2016)
Log Message:
-----------
Make the test for over-complication while auto-possessifying bite sooner.
Modified Paths:
--------------
code/trunk/ChangeLog
code/trunk/src/pcre2_auto_possess.c
code/trunk/testdata/testinput15
code/trunk/testdata/testoutput15
Modified: code/trunk/ChangeLog
===================================================================
--- code/trunk/ChangeLog 2016-12-29 16:29:05 UTC (rev 634)
+++ code/trunk/ChangeLog 2016-12-31 13:35:31 UTC (rev 635)
@@ -272,7 +272,12 @@
41. A minor change to pcre2grep: colour reset is now "<esc>[0m" instead of
"<esc>[00m".
+42. The limit in the auto-possessification code that was intended to catch
+overly-complicated patterns and not spend too much time auto-possessifying was
+being reset too often, resulting in very long compile times for some patterns.
+Now such patterns are no longer completely auto-possessified.
+
Version 10.22 29-July-2016
--------------------------
Modified: code/trunk/src/pcre2_auto_possess.c
===================================================================
--- code/trunk/src/pcre2_auto_possess.c 2016-12-29 16:29:05 UTC (rev 634)
+++ code/trunk/src/pcre2_auto_possess.c 2016-12-31 13:35:31 UTC (rev 635)
@@ -1046,8 +1046,10 @@
/* Replaces single character iterations with their possessive alternatives
if appropriate. This function modifies the compiled opcode! Hitting a
-non-existant opcode may indicate a bug in PCRE2, but it can also be caused if a
-bad UTF string was compiled with PCRE2_NO_UTF_CHECK.
+non-existent opcode may indicate a bug in PCRE2, but it can also be caused if a
+bad UTF string was compiled with PCRE2_NO_UTF_CHECK. The rec_limit catches
+overly complicated or large patterns. In these cases, the check just stops,
+leaving the remainder of the pattern unpossessified.
Arguments:
code points to start of the byte code
@@ -1065,7 +1067,7 @@
PCRE2_SPTR end;
PCRE2_UCHAR *repeat_opcode;
uint32_t list[8];
-int rec_limit;
+int rec_limit = 10000;
for (;;)
{
@@ -1080,7 +1082,6 @@
get_chr_property_list(code, utf, cb->fcc, list) : NULL;
list[1] = c == OP_STAR || c == OP_PLUS || c == OP_QUERY || c == OP_UPTO;
- rec_limit = 1000;
if (end != NULL && compare_opcodes(end, utf, cb, list, end, &rec_limit))
{
switch(c)
@@ -1137,7 +1138,6 @@
list[1] = (c & 1) == 0;
- rec_limit = 1000;
if (compare_opcodes(end, utf, cb, list, end, &rec_limit))
{
switch (c)
Modified: code/trunk/testdata/testinput15
===================================================================
--- code/trunk/testdata/testinput15 2016-12-29 16:29:05 UTC (rev 634)
+++ code/trunk/testdata/testinput15 2016-12-31 13:35:31 UTC (rev 635)
@@ -160,4 +160,9 @@
/(*NO_AUTO_POSSESS)\w+(?C1)/BI
abc\=callout_fail=1
+# This test breaks the JIT stack limit
+
+/(|]+){2,2452}/
+ (|]+){2,2452}
+
# End of testinput15
Modified: code/trunk/testdata/testoutput15
===================================================================
--- code/trunk/testdata/testoutput15 2016-12-29 16:29:05 UTC (rev 634)
+++ code/trunk/testdata/testoutput15 2016-12-31 13:35:31 UTC (rev 635)
@@ -380,4 +380,11 @@
1 ^^
No match
+# This test breaks the JIT stack limit
+
+/(|]+){2,2452}/
+ (|]+){2,2452}
+ 0:
+ 1:
+
# End of testinput15
