public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug target/50813] New: gcc.dg/torture/vshuf-{v4di,v8si}.c fail on AVX target
@ 2011-10-20 19:43 ubizjak at gmail dot com
2011-10-20 19:46 ` [Bug target/50813] " ubizjak at gmail dot com
` (3 more replies)
0 siblings, 4 replies; 5+ messages in thread
From: ubizjak at gmail dot com @ 2011-10-20 19:43 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50813
Bug #: 50813
Summary: gcc.dg/torture/vshuf-{v4di,v8si}.c fail on AVX target
Classification: Unclassified
Product: gcc
Version: 4.7.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
AssignedTo: unassigned@gcc.gnu.org
ReportedBy: ubizjak@gmail.com
CC: jakub@gcc.gnu.org, rth@gcc.gnu.org
Target: x86-avx
Recent failure on AVX targets [1].
These two testcases try to generate avx2_permv2ti insn, which is disabled on
non-AVX2 targets:
(define_insn "avx2_permv2ti"
[(set (match_operand:V4DI 0 "register_operand" "=x")
(unspec:V4DI
[(match_operand:V4DI 1 "register_operand" "x")
(match_operand:V4DI 2 "nonimmediate_operand" "xm")
(match_operand:SI 3 "const_0_to_255_operand" "n")]
UNSPEC_VPERMTI))]
"TARGET_AVX2"
"vperm2i128\t{%3, %2, %1, %0|%0, %1, %2, %3}"
[(set_attr "type" "sselog")
(set_attr "prefix" "vex")
(set_attr "mode" "OI")])
The pattern is generated through ix86_vectorize_builtin_vec_perm_ok via:
Breakpoint 1, gen_avx2_permv2ti (operand0=0x7ffff15862e0,
operand1=0x7ffff15862a0, operand2=0x7ffff15862c0, operand3=0x7ffff1997670)
at insn-emit.c:14480
14480 {
(gdb) bt
#0 gen_avx2_permv2ti (operand0=0x7ffff15862e0, operand1=0x7ffff15862a0,
operand2=0x7ffff15862c0, operand3=0x7ffff1997670) at insn-emit.c:14480
#1 0x0000000000adca70 in expand_vec_perm_even_odd_1 (d=0x7fffffffdda0, odd=0)
at /home/uros/gcc-svn/trunk/gcc/config/i386/i386.c:36132
#2 0x0000000000adcdda in expand_vec_perm_even_odd (d=0x7fffffffdda0)
at /home/uros/gcc-svn/trunk/gcc/config/i386/i386.c:36201
#3 0x0000000000add743 in ix86_expand_vec_perm_builtin_1 (d=0x7fffffffdda0)
at /home/uros/gcc-svn/trunk/gcc/config/i386/i386.c:36456
#4 0x0000000000adecdc in ix86_vectorize_builtin_vec_perm_ok (
vec_type=0x7ffff1ac65e8, mask=0x7ffff1b132a0)
at /home/uros/gcc-svn/trunk/gcc/config/i386/i386.c:36776
#5 0x00000000007c89a3 in can_vec_perm_expr_p (type=0x7ffff1ac65e8,
sel=0x7ffff1b132a0) at /home/uros/gcc-svn/trunk/gcc/optabs.c:6717
[1] http://gcc.gnu.org/ml/gcc-testresults/2011-10/msg02319.html
^ permalink raw reply [flat|nested] 5+ messages in thread
* [Bug target/50813] gcc.dg/torture/vshuf-{v4di,v8si}.c fail on AVX target
2011-10-20 19:43 [Bug target/50813] New: gcc.dg/torture/vshuf-{v4di,v8si}.c fail on AVX target ubizjak at gmail dot com
@ 2011-10-20 19:46 ` ubizjak at gmail dot com
2011-10-20 19:53 ` jakub at gcc dot gnu.org
` (2 subsequent siblings)
3 siblings, 0 replies; 5+ messages in thread
From: ubizjak at gmail dot com @ 2011-10-20 19:46 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50813
Uros Bizjak <ubizjak at gmail dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2011-10-20
Ever Confirmed|0 |1
--- Comment #1 from Uros Bizjak <ubizjak at gmail dot com> 2011-10-20 19:45:44 UTC ---
The failure manifests as:
In file included from vshuf-v4di.c:14:0:
vshuf-main.inc: In function ‘test_16’:
vshuf-main.inc:28:1: error: unrecognizable insn:
(insn 36 39 37 3 (set (reg:V4DI 99)
(unspec:V4DI [
(reg:V4DI 66 [ a.8 ])
(reg:V4DI 98)
(const_int 32 [0x20])
] UNSPEC_VPERMTI)) vshuf-main.inc:28 -1
(nil))
vshuf-main.inc:28:1: internal compiler error: in extract_insn, at recog.c:2137
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
-O2 -mavx
^ permalink raw reply [flat|nested] 5+ messages in thread
* [Bug target/50813] gcc.dg/torture/vshuf-{v4di,v8si}.c fail on AVX target
2011-10-20 19:43 [Bug target/50813] New: gcc.dg/torture/vshuf-{v4di,v8si}.c fail on AVX target ubizjak at gmail dot com
2011-10-20 19:46 ` [Bug target/50813] " ubizjak at gmail dot com
@ 2011-10-20 19:53 ` jakub at gcc dot gnu.org
2011-10-21 19:33 ` jakub at gcc dot gnu.org
2012-01-20 11:29 ` ubizjak at gmail dot com
3 siblings, 0 replies; 5+ messages in thread
From: jakub at gcc dot gnu.org @ 2011-10-20 19:53 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50813
--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-10-20 19:52:45 UTC ---
Strange, wonder how I've missed this.
I guess easiest would be probably just to
if (!TARGET_AVX2)
{
struct expand_vec_perm_d d_copy = *d;
d_copy.target = gen_lowpart (V{4DF,8SF}mode, d->target);
d_copy.op0 = gen_lowpart (V{4DF,8SF}mode, d->op0);
d_copy.op1 = gen_lowpart (V{4DF,8SF}mode, d->op1);
return expand_vec_perm_even_odd_1 (&d_copy);
}
for V{4DI,8SI}mode. Or if (!TARGET_AVX2) return false;.
^ permalink raw reply [flat|nested] 5+ messages in thread
* [Bug target/50813] gcc.dg/torture/vshuf-{v4di,v8si}.c fail on AVX target
2011-10-20 19:43 [Bug target/50813] New: gcc.dg/torture/vshuf-{v4di,v8si}.c fail on AVX target ubizjak at gmail dot com
2011-10-20 19:46 ` [Bug target/50813] " ubizjak at gmail dot com
2011-10-20 19:53 ` jakub at gcc dot gnu.org
@ 2011-10-21 19:33 ` jakub at gcc dot gnu.org
2012-01-20 11:29 ` ubizjak at gmail dot com
3 siblings, 0 replies; 5+ messages in thread
From: jakub at gcc dot gnu.org @ 2011-10-21 19:33 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50813
--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-10-21 19:32:23 UTC ---
Author: jakub
Date: Fri Oct 21 19:32:20 2011
New Revision: 180311
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=180311
Log:
PR target/50813
* config/i386/i386.c (expand_vec_perm_even_odd_1): Handle
V4DImode and V8SImode for !TARGET_AVX2.
* gcc.dg/torture/vshuf-32.inc: Add broadcast permutation
from element other than first and reverse permutation.
* gcc.dg/torture/vshuf-16.inc: Likewise.
* gcc.dg/torture/vshuf-8.inc: Likewise.
* gcc.dg/torture/vshuf-4.inc: Likewise.
Modified:
trunk/gcc/ChangeLog
trunk/gcc/config/i386/i386.c
trunk/gcc/testsuite/ChangeLog
trunk/gcc/testsuite/gcc.dg/torture/vshuf-16.inc
trunk/gcc/testsuite/gcc.dg/torture/vshuf-32.inc
trunk/gcc/testsuite/gcc.dg/torture/vshuf-4.inc
trunk/gcc/testsuite/gcc.dg/torture/vshuf-8.inc
^ permalink raw reply [flat|nested] 5+ messages in thread
* [Bug target/50813] gcc.dg/torture/vshuf-{v4di,v8si}.c fail on AVX target
2011-10-20 19:43 [Bug target/50813] New: gcc.dg/torture/vshuf-{v4di,v8si}.c fail on AVX target ubizjak at gmail dot com
` (2 preceding siblings ...)
2011-10-21 19:33 ` jakub at gcc dot gnu.org
@ 2012-01-20 11:29 ` ubizjak at gmail dot com
3 siblings, 0 replies; 5+ messages in thread
From: ubizjak at gmail dot com @ 2012-01-20 11:29 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50813
Uros Bizjak <ubizjak at gmail dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |RESOLVED
Resolution| |FIXED
Target Milestone|--- |4.7.0
--- Comment #4 from Uros Bizjak <ubizjak at gmail dot com> 2012-01-20 10:41:52 UTC ---
Fixed.
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2012-01-20 10:42 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-10-20 19:43 [Bug target/50813] New: gcc.dg/torture/vshuf-{v4di,v8si}.c fail on AVX target ubizjak at gmail dot com
2011-10-20 19:46 ` [Bug target/50813] " ubizjak at gmail dot com
2011-10-20 19:53 ` jakub at gcc dot gnu.org
2011-10-21 19:33 ` jakub at gcc dot gnu.org
2012-01-20 11:29 ` ubizjak at gmail dot com
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).