[Bug regression/57313] New: [4.9 Regression]: gcc.dg/binop-xor3.c scan-tree-dump-times

[Bug regression/57313] New: [4.9 Regression]: gcc.dg/binop-xor3.c scan-tree-dump-times

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57313

            Bug ID: 57313
           Summary: [4.9 Regression]: gcc.dg/binop-xor3.c
                    scan-tree-dump-times
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P3
         Component: regression
          Assignee: unassigned at gcc dot gnu.org
          Reporter: hp at gcc dot gnu.org
                CC: glisse at gcc dot gnu.org
              Host: x86_64-unknown-linux-gnu
            Target: cris-axis-elf

Created attachment 30139
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=30139&action=edit
binop-xor3.c.165t.optimized

This test previously passed, now it fails.
A patch in the revision range (last_known_working:first_known_failing)
198890:198899
exposed or caused this regression.  Since then it fails as follows:

Running /tmp/hpautotest-gcc0/gcc/gcc/testsuite/gcc.dg/dg.exp ...
FAIL: gcc.dg/binop-xor3.c scan-tree-dump-times optimized "\\^" 1

The files binop-xor3.s and binop-xor3.c.165t.optimized are attached.

Author of the most suspect patch (having touched the actual test-case and
related code) in revision range CC:ed.

Apparently, cris-elf is just one of many targets for which this test now fails;
I guess half of them regressed but x86_64 and i686 (and ia64, sh4, spu), as
seen on gcc-testresults@ since then: s390-ibm-linux-gnu
<http://gcc.gnu.org/ml/gcc-testresults/2013-05/msg01643.html>,
m68k-unknown-linux-gnu
<http://gcc.gnu.org/ml/gcc-testresults/2013-05/msg01593.html>,
powerpc-ibm-aix7.1.0.0
<http://gcc.gnu.org/ml/gcc-testresults/2013-05/msg01567.html>,
powerpc64-unknown-linux-gnu
<http://gcc.gnu.org/ml/gcc-testresults/2013-05/msg01709.html>.

[Bug regression/57313] [4.9 Regression]: gcc.dg/binop-xor3.c scan-tree-dump-times

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57313

--- Comment #1 from Hans-Peter Nilsson <hp at gcc dot gnu.org> ---
Created attachment 30140
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=30140&action=edit
binop-xor3.s

[Bug regression/57313] [4.9 Regression]: gcc.dg/binop-xor3.c scan-tree-dump-times

[glisse at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57313

--- Comment #2 from Marc Glisse <glisse at gcc dot gnu.org> ---
It was xfailed before. It started passing as a side effect of r198893 so I
removed the xfail, but apparently it now depends on the platform :-(

What do you get with -O -fdump-tree-original?

[Bug regression/57313] [4.9 Regression]: gcc.dg/binop-xor3.c scan-tree-dump-times

[glisse at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57313

--- Comment #3 from Marc Glisse <glisse at gcc dot gnu.org> ---
(In reply to Marc Glisse from comment #2)
> What do you get with -O -fdump-tree-original?

I tried, and the transformation doesn't happen at all (I was wondering if it
might be undone by another pass later).

See http://gcc.gnu.org/ml/gcc-patches/2013-05/msg00981.html for more
discussion.

[Bug testsuite/57313] [4.9 Regression]: gcc.dg/binop-xor3.c scan-tree-dump-times

[glisse at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57313

Marc Glisse <glisse at gcc dot gnu.org> changed:

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

--- Comment #4 from Marc Glisse <glisse at gcc dot gnu.org> ---
r199020 | glisse | 2013-05-17 17:31:02 +0200 (Fri, 17 May 2013) | 6 lines

2013-05-17  Marc Glisse  <marc.glisse@inria.fr>

        PR testsuite/57313
        * gcc.dg/binop-xor3.c: Restrict to platforms known to work (x86).