This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Dataflow miscompiles soft-fp/floatuntitf.c on x86_64
- From: Uros Bizjak <ubizjak at gmail dot com>
- To: GCC <gcc at gcc dot gnu dot org>
- Cc: Kenneth Zadeck <zadeck at naturalbridge dot com>
- Date: Wed, 23 May 2007 19:34:12 +0200
- Subject: Dataflow miscompiles soft-fp/floatuntitf.c on x86_64
Hello!
Latest dataflow branch segfaults for:
FAIL: gcc.dg/torture/fp-int-convert-float128-timode.c -O0 execution test
FAIL: gcc.dg/torture/fp-int-convert-float128-timode.c -O1 execution test
FAIL: gcc.dg/torture/fp-int-convert-float128-timode.c -O2 execution test
FAIL: gcc.dg/torture/fp-int-convert-float128-timode.c -O3
-fomit-frame-pointer execution test
FAIL: gcc.dg/torture/fp-int-convert-float128-timode.c -O3 -g execution
test
FAIL: gcc.dg/torture/fp-int-convert-float128-timode.c -Os execution test
(The tests were run on FC6, x86_64-pc-linux-gnu).
This is due to segfault in soft-fp/floatuntitf.c at __floatuntitf+202,
where %rsp is not aligned to 16 bytes for movdqa:
0x0000000000402c10 <__floatuntitf+192>: mov %ax,0xfffffffffffffffe(%rsp)
0x0000000000402c15 <__floatuntitf+197>: andb
$0x7f,0xffffffffffffffff(%rsp)
0x0000000000402c1a <__floatuntitf+202>: movdqa
0xfffffffffffffff0(%rsp),%xmm0 <<<
0x0000000000402c20 <__floatuntitf+208>: pop %rbx
0x0000000000402c21 <__floatuntitf+209>: pop %rbp
0x0000000000402c22 <__floatuntitf+210>: retq
Uros.