Seen on 4.0 "Thu Mar 17 06:05:21 UTC 2005" and trunk of "Thu Mar 17 15:06:06 UTC
2005" at -O2. The test-case in the first attachment has this code snippet:
vec += (base[i+1] - base[i]);
The expression at the right hand of that line is compiled at -O2 into:
move.d [$r11+],$r9
sub.d [$r11-4],$r9
Tsk tsk. Some part of GCC doesn't adjust expressions when postincrements are
involved.
I don't see this bug with 3.2.1, so I call it a regression, though local changes
might affect the actual regressionness. It is unknown whether it's also a
regression for 3.4 and 3.3.
--
Summary: [4.1, 4.0 regression] Mishandling of postincrement
causes bzip2 miscompilation.
Product: gcc
Version: 4.0.0
Status: UNCONFIRMED
Keywords: wrong-code
Severity: normal
Priority: P2
Component: rtl-optimization
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: hp at gcc dot gnu dot org
CC: gcc-bugs at gcc dot gnu dot org
GCC target triplet: cris-*-*
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=20527
