[Bug tree-optimization/46880] New: [4.6 Regression] g++.dg/other/pr40446.C FAILs with custom flags

[Bug tree-optimization/46880] New: [4.6 Regression] g++.dg/other/pr40446.C FAILs with custom flags

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

           Summary: [4.6 Regression] g++.dg/other/pr40446.C FAILs with
                    custom flags
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: zsojka@seznam.cz
              Host: x86_64-pc-linux-gnu
            Target: x86_64-pc-linux-gnu


Created attachment 22702
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22702
reduced testcase

Compiler output:
$ g++ -O -fgcse -fstack-protector-all -funroll-loops -fschedule-insns
-fsched-pressure pr40446.ii
$ ./a.out 
Aborted

Tested revisions:
r167643 - fail
r167556 - fail
r165699 - OK

[Bug tree-optimization/46880] [4.6 Regression] g++.dg/other/pr40446.C FAILs with custom flags

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.6.0

[Bug tree-optimization/46880] [4.6 Regression] g++.dg/other/pr40446.C FAILs with custom flags

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

--- Comment #1 from Zdenek Sojka <zsojka at seznam dot cz> 2010-12-18 21:33:03 UTC ---
Maybe better set of flags:

$ g++ -O -fgcse -fstack-protector-all -fschedule-insns -fsched-pressure -fweb
pr46880.C 
$ ./a.out 
Aborted

_Z3foov:
...
    xorpd    xmm1, xmm1    # tmp65
    movsd    xmm0, QWORD PTR .LC1[rip]    # tmp68,
...
    shufpd    xmm0, xmm1, 1    # tmp66, tmp65
...

.LC1:
    .long    0
    .long    1072693248

is wrong, it loads 1.0 to lower 64bits of xmm0, and shufpd can't solve the
situation with given values in xmm0 and xmm1, when xmm0 is output:
it just moves xmm0[0] or xmm0[1] to xmm0[0], and xmm1[0] or xmm1[1] to xmm0[1]

either
    xorpd    xmm0, xmm0    # tmp65
    movsd    xmm1, QWORD PTR .LC1[rip]    # tmp68,
    shufpd    xmm0, xmm1, 1    # tmp66, tmp65
or do it the same as with -fno-web:
    xorpd    xmm0, xmm0    # tmp65
    movsd    xmm1, QWORD PTR .LC1[rip]    # tmp67,
    unpcklpd    xmm0, xmm1    # tmp65, tmp67

[Bug target/46880] [4.4/4.5/4.6 Regression] generating of shufpd is broken

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

--- Comment #2 from Zdenek Sojka <zsojka at seznam dot cz> 2010-12-18 22:05:03 UTC ---
Created attachment 22816
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22816
better testcase

$ gcc -O pr46880-2.c 
$ ./a.out 
Aborted

(will probably fail with strict aliasing)

4.4.5, 4.5.2, 4.6.0 fails, 4.3.5 doesn't generate shufpd

[Bug target/46880] [4.4/4.5/4.6 Regression] generating of shufpd is broken

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

H.J. Lu <hjl.tools at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2010.12.18 23:16:58
                 CC|                            |vmakarov at redhat dot com
     Ever Confirmed|0                           |1

--- Comment #3 from H.J. Lu <hjl.tools at gmail dot com> 2010-12-18 23:16:58 UTC ---
This is caused by revision 167519:

http://gcc.gnu.org/ml/gcc-cvs/2010-12/msg00200.html

[Bug target/46880] [4.4/4.5/4.6 Regression] generating of shufpd is broken

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
                 CC|                            |jakub at gcc dot gnu.org
         AssignedTo|unassigned at gcc dot       |jakub at gcc dot gnu.org
                   |gnu.org                     |

[Bug target/46880] [4.4/4.5/4.6 Regression] generating of shufpd is broken

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> 2010-12-21 14:45:08 UTC ---
Created attachment 22836
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22836
gcc46-pr46880.patch

Untested fix.  In both of these insns, for the shufpd $2 alternative %2 operand
uses constraint 0 while %1 operand uses constraint x.  So, using shuftpd $2,
%2, %0 looks wrong, because that will never look at the other operand,
shuftpd $2, %0, %0 is IMHO a nop.

[Bug target/46880] [4.4/4.5/4.6 Regression] generating of shufpd is broken

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> 2010-12-21 22:37:30 UTC ---
Author: jakub
Date: Tue Dec 21 22:37:23 2010
New Revision: 168135

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168135
Log:
    PR target/46880
    * config/i386/sse.md (sse2_loadlpd, sse2_movsd): Fix shufpd source
    operand.

    * gcc.target/i386/pr46880.c: New test.

Added:
    trunk/gcc/testsuite/gcc.target/i386/pr46880.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/i386/sse.md
    trunk/gcc/testsuite/ChangeLog

[Bug target/46880] [4.4/4.5 Regression] generating of shufpd is broken

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |4.6.0
            Summary|[4.4/4.5/4.6 Regression]    |[4.4/4.5 Regression]
                   |generating of shufpd is     |generating of shufpd is
                   |broken                      |broken
      Known to fail|4.6.0                       |

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> 2010-12-21 22:39:26 UTC ---
Fixed on the trunk so far.

[Bug target/46880] [4.4/4.5 Regression] generating of shufpd is broken

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-01-16 20:25:04 UTC ---
Author: jakub
Date: Sun Jan 16 20:25:01 2011
New Revision: 168868

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168868
Log:
    Backport from mainline
    2010-12-21  Jakub Jelinek  <jakub@redhat.com>

    PR target/46880
    * config/i386/sse.md (sse2_loadlpd, sse2_movsd): Fix shufpd source
    operand.

    * gcc.target/i386/pr46880.c: New test.

Added:
    branches/gcc-4_5-branch/gcc/testsuite/gcc.target/i386/pr46880.c
Modified:
    branches/gcc-4_5-branch/gcc/ChangeLog
    branches/gcc-4_5-branch/gcc/config/i386/sse.md
    branches/gcc-4_5-branch/gcc/testsuite/ChangeLog

[Bug target/46880] [4.4/4.5 Regression] generating of shufpd is broken

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-01-16 22:57:44 UTC ---
Author: jakub
Date: Sun Jan 16 22:57:42 2011
New Revision: 168880

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168880
Log:
    Backport from mainline
    2010-12-21  Jakub Jelinek  <jakub@redhat.com>

    PR target/46880
    * config/i386/sse.md (sse2_loadlpd, sse2_movsd): Fix shufpd source
    operand.

    * gcc.target/i386/pr46880.c: New test.

Added:
    branches/gcc-4_4-branch/gcc/testsuite/gcc.target/i386/pr46880.c
Modified:
    branches/gcc-4_4-branch/gcc/ChangeLog
    branches/gcc-4_4-branch/gcc/config/i386/sse.md
    branches/gcc-4_4-branch/gcc/testsuite/ChangeLog

[Bug target/46880] [4.4/4.5 Regression] generating of shufpd is broken

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED

--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-01-17 08:01:17 UTC ---
Fixed.

[Bug target/46880] [4.4/4.5 Regression] generating of shufpd is broken

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46880

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |4.4.6, 4.5.3
   Target Milestone|4.6.0                       |4.4.6
      Known to fail|4.4.6                       |4.4.5