[Bug middle-end/46404] New: [4.6 Regression] ICE: definition in block 5 does not dominate use in block 32 with -O -fgraphite-identity -fno-tree-scev-cprop

[Bug middle-end/46404] New: [4.6 Regression] ICE: definition in block 5 does not dominate use in block 32 with -O -fgraphite-identity -fno-tree-scev-cprop

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46404

           Summary: [4.6 Regression] ICE: definition in block 5 does not
                    dominate use in block 32 with -O -fgraphite-identity
                    -fno-tree-scev-cprop
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: zsojka@seznam.cz
                CC: spop@gcc.gnu.org
              Host: x86_64-pc-linux-gnu
            Target: x86_64-pc-linux-gnu


Created attachment 22360
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22360
reduced testcase

Compiler output:
$ gcc -O -fgraphite-identity -fno-tree-scev-cprop testcase.c
testcase.c: In function 'foo':
testcase.c:4:1: error: definition in block 5 does not dominate use in block 32
for SSA_NAME: i_6 in statement:
D.2701_26 = i_6 > 0;
testcase.c:4:1: internal compiler error: verify_ssa failed
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

Tested revisions:
r166509 - crash
r163636 - crash
r161659 - OK
4.5 r166509 - OK

[Bug middle-end/46404] [4.6 Regression] ICE: definition in block 5 does not dominate use in block 32 with -O -fgraphite-identity -fno-tree-scev-cprop

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46404

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2010.11.10 12:12:06
   Target Milestone|---                         |4.6.0
     Ever Confirmed|0                           |1

--- Comment #1 from H.J. Lu <hjl.tools at gmail dot com> 2010-11-10 12:12:06 UTC ---
It is caused by revision 163118:

http://gcc.gnu.org/ml/gcc-cvs/2010-08/msg00329.html

[Bug middle-end/46404] [4.6 Regression] ICE: definition in block 5 does not dominate use in block 32 with -O -fgraphite-identity -fno-tree-scev-cprop

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46404

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1

[Bug middle-end/46404] [4.6 Regression] ICE: definition in block 5 does not dominate use in block 32 with -O -fgraphite-identity -fno-tree-scev-cprop

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46404

Sebastian Pop <spop at gcc dot gnu.org> changed:

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

--- Comment #2 from Sebastian Pop <spop at gcc dot gnu.org> 2010-12-15 17:38:37 UTC ---
I cannot reproduce this error on trunk r167844.
I am running a git bisect to identify which patch fixed this PR.

[Bug middle-end/46404] [4.6 Regression] ICE: definition in block 5 does not dominate use in block 32 with -O -fgraphite-identity -fno-tree-scev-cprop

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46404

--- Comment #3 from Zdenek Sojka <zsojka at seznam dot cz> 2010-12-15 17:52:09 UTC ---
I can't reproduce it neither with the reduced testcase, but the original one
still fails (I am sorry for not mentioning it was reduced from
gcc.dg/graphite/pr42284.c):

$ gcc -O -fgraphite-identity -fno-tree-scev-cprop pr42284.i 
/mnt/svn/gcc-trunk/gcc/testsuite/gcc.dg/graphite/pr42284.c: In function
'inflate_fixed':
/mnt/svn/gcc-trunk/gcc/testsuite/gcc.dg/graphite/pr42284.c:11:1: error:
definition in block 5 does not dominate use in block 59
for SSA_NAME: i_34 in statement:
D.2738_52 = 256 - i_34;
/mnt/svn/gcc-trunk/gcc/testsuite/gcc.dg/graphite/pr42284.c:11:1: internal
compiler error: verify_ssa failed
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

[Bug middle-end/46404] [4.6 Regression] ICE: definition in block 5 does not dominate use in block 32 with -O -fgraphite-identity -fno-tree-scev-cprop

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46404

--- Comment #4 from Sebastian Pop <spop at gcc dot gnu.org> 2010-12-15 17:55:26 UTC ---
Ok, I will have a look at that one as well.
Thanks for pointing out that this error was not fixed, but probably hidden.

[Bug middle-end/46404] [4.6 Regression] ICE: definition in block 5 does not dominate use in block 32 with -O -fgraphite-identity -fno-tree-scev-cprop

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46404

--- Comment #5 from Sebastian Pop <spop at gcc dot gnu.org> 2010-12-15 20:44:41 UTC ---
This commit fixed the reduced testcase:

 Fix PR46845: handle scop parameters before scev analysis.

    2010-12-14  Sebastian Pop  <sebastian.pop@amd.com>

        PR tree-optimization/46845
        * sese.c (scalar_evolution_in_region): Handle scop parameters
        before scev analysis.

        * gcc.dg/graphite/id-pr46845.c: New.

    git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@167844
138bc75d-0d04-0410-961f-82ee72b054a4

[Bug middle-end/46404] [4.6 Regression] ICE: definition in block 5 does not dominate use in block 32 with -O -fgraphite-identity -fno-tree-scev-cprop

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46404

--- Comment #6 from Sebastian Pop <spop at gcc dot gnu.org> 2010-12-15 22:01:04 UTC ---
Author: spop
Date: Wed Dec 15 22:00:59 2010
New Revision: 167876

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=167876
Log:
Add testcase for PR46404.

2010-12-15  Sebastian Pop  <sebastian.pop@amd.com>

    PR tree-optimization/46404
    * gcc.dg/graphite/pr46404.c: New.

Added:
    trunk/gcc/testsuite/gcc.dg/graphite/pr46404.c
Modified:
    trunk/gcc/testsuite/ChangeLog

[Bug middle-end/46404] [4.6 Regression] ICE: definition in block 5 does not dominate use in block 32 with -O -fgraphite-identity -fno-tree-scev-cprop

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46404

--- Comment #7 from Sebastian Pop <spop at gcc dot gnu.org> 2010-12-16 20:05:23 UTC ---
Author: spop
Date: Thu Dec 16 20:05:18 2010
New Revision: 167954

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=167954
Log:
Fix PR46404: call scev_reset to reset the number of iterations as well.

2010-12-16  Sebastian Pop  <sebastian.pop@amd.com>

    PR tree-optimization/46404
    * graphite-clast-to-gimple.c (gloog): Call scev_reset.

    * gcc.dg/graphite/pr46404-1.c: New.

Added:
    trunk/gcc/testsuite/gcc.dg/graphite/pr46404-1.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/graphite-clast-to-gimple.c
    trunk/gcc/testsuite/ChangeLog

[Bug middle-end/46404] [4.6 Regression] ICE: definition in block 5 does not dominate use in block 32 with -O -fgraphite-identity -fno-tree-scev-cprop

[spop at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46404

Sebastian Pop <spop at gcc dot gnu.org> changed:

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

--- Comment #8 from Sebastian Pop <spop at gcc dot gnu.org> 2010-12-16 20:06:47 UTC ---
Fixed.