This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug testsuite/50185] New: [4.7 Regression] FAIL: gcc.target/i386/avx2-vmovmskb-2.c scan-assembler vmovmskb on x86_64-apple-darwin10
- From: "dominiq at lps dot ens.fr" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Thu, 25 Aug 2011 09:35:25 +0000
- Subject: [Bug testsuite/50185] New: [4.7 Regression] FAIL: gcc.target/i386/avx2-vmovmskb-2.c scan-assembler vmovmskb on x86_64-apple-darwin10
- Auto-submitted: auto-generated
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50185
Bug #: 50185
Summary: [4.7 Regression] FAIL:
gcc.target/i386/avx2-vmovmskb-2.c scan-assembler
vmovmskb on x86_64-apple-darwin10
Classification: Unclassified
Product: gcc
Version: 4.7.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: testsuite
AssignedTo: unassigned@gcc.gnu.org
ReportedBy: dominiq@lps.ens.fr
CC: hjl@gcc.gnu.org, kirill.yukhin@intel.com
Host: x86_64-apple-darwin10
Target: x86_64-apple-darwin10
Build: x86_64-apple-darwin10
In the tests introduced in revision 178006, scan-assembler vmovmskb fails for
gcc.target/i386/avx2-vmovmskb-2.c on x86_64-apple-darwin10. The assembly is
.text
.align 4,0x90
_do_test:
LFB1005:
pushq %rbp
LCFI0:
xorl %eax, %eax
movl $1, %esi
LCFI1:
movq %rsp, %rbp
LCFI2:
andq $-32, %rsp
subq $32, %rsp
vmovdqa LC0(%rip), %ymm0
vmovdqa %ymm0, (%rsp)
jmp L3
.align 4,0x90
L2:
addq $1, %rax
cmpq $32, %rax
je L7
L3:
cmpb $0, (%rsp,%rax)
movl %eax, %ecx
jns L2
movl %esi, %edi
addq $1, %rax
sall %cl, %edi
orl %edi, %edx
cmpq $32, %rax
jne L3
L7:
vmovdqa LC1(%rip), %ymm0
vpmovmskb %ymm0, %eax
cmpl %edx, %eax
jne L8
leave
LCFI3:
vzeroupper
ret
L8:
LCFI4:
vzeroupper
call _abort
...
So there is no vmovmskb, but three vmovdqa. Any idea why?