https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91555
Bug ID: 91555
Summary: [9.2 regression] Optimizer bug
Product: gcc
Version: 9.2.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: skunk at iskunk dot org
Target Milestone: ---
Created attachment 46761
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=46761&action=edit
Quasi-minimal test program
I have encountered what appears to be a bug in the optimizer of GCC 9.2.0. (I
am unclear as to whether it is the tree optimizer or RTL optimizer that is at
issue; please recategorize as appropriate.)
The problem occurs inside some timezone-handling code that is part of a larger
project. I have extracted the code into a program that may not be a minimal
test case, but nevertheless exhibits the problem.
This incorrect optimization has been observed in x86-64 builds on both Linux
and Solaris.
Correct result:
$ gcc -O1 gcc9-opt-bug.c -o bug
$ ./bug
result: t = 1009489458 (correct)
Wrong result:
$ gcc -O2 gcc9-opt-bug.c -o bug
$ ./bug
WRONG 13
result: t = 18446744073709551615
Workaround using "volatile" keyword:
$ gcc -O2 -DUSE_VOLATILE gcc9-opt-bug.c -o bug
$ ./bug
result: t = 1009489458 (correct)
This code was correctly optimized by GCC 4.8.1, so this appears to be a
regression.
