[Bug target/56864] New: [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[Bug target/56864] New: [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[schwab@linux-m68k.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

             Bug #: 56864
           Summary: [4.9 regression] FAIL:
                    gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c
                    scan-tree-dump-times vect "vectorized 1 loops" 0
    Classification: Unclassified
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: schwab@linux-m68k.org
            Target: powerpc-*-*


Created attachment 29818
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29818
costmodel-vect-76b.c.111t.vect

spawn /daten/gcc/gcc-20130406/Build/gcc/xgcc
-B/daten/gcc/gcc-20130406/Build/gcc/
/daten/gcc/gcc-20130406/gcc/testsuite/gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c
-fno-diagnostics-show-caret -O2 -ftree-vectorize -fvect-cost-model -fno-common
-maltivec -fdump-tree-vect-details -lm -m32 -o ./costmodel-vect-76b.exe
PASS: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c (test for excess errors)
Setting LD_LIBRARY_PATH to
:/daten/gcc/gcc-20130406/Build/gcc:/daten/gcc/gcc-20130406/Build/gcc/64:/daten/gcc/gcc-20130406/Build/powerpc64-linux/./libsanitizer/asan/.libs::/daten/gcc/gcc-20130406/Build/gcc:/daten/gcc/gcc-20130406/Build/gcc/64:/daten/gcc/gcc-20130406/Build/powerpc64-linux/./libsanitizer/asan/.libs:/daten/gcc/gcc-20130406/Build/powerpc64-linux/libstdc++-v3/src/.libs:/daten/gcc/gcc-20130406/Build/powerpc64-linux/libmudflap/.libs:/daten/gcc/gcc-20130406/Build/powerpc64-linux/libsanitizer/.libs:/daten/gcc/gcc-20130406/Build/powerpc64-linux/libssp/.libs:/daten/gcc/gcc-20130406/Build/powerpc64-linux/libgomp/.libs:/daten/gcc/gcc-20130406/Build/powerpc64-linux/libitm/.libs:/daten/gcc/gcc-20130406/Build/powerpc64-linux/libatomic/.libs:/daten/gcc/gcc-20130406/Build/./gcc:/daten/gcc/gcc-20130406/Build/./prev-gcc
spawn [open ...]
PASS: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c execution test
FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect
"vectorized 1 loops" 0
FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect
"vectorization not profitable" 1

Only fails for -m32.

Configured with: ../configure --prefix=/usr --build=powerpc64-linux
--enable-checking=release --enable-shared --with-system-zlib
--with-cpu=default32 --with-cpu-64=power4 --enable-secureplt
--with-long-double-128

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

--- Comment #1 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2013-04-07 14:56:42 UTC ---
The tests pass for me on costmodel-vect-76b.c at r196686.

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[schwab@linux-m68k.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

--- Comment #2 from Andreas Schwab <schwab@linux-m68k.org> 2013-04-07 15:26:32 UTC ---
The test was run with r197541.

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

--- Comment #3 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2013-04-07 15:31:13 UTC ---
> > The tests pass for me on costmodel-vect-76b.c at r196686.
>
> The test was run with r197541.

I did a copy&paste error, the tests pass for me at r197531.

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[schwab@linux-m68k.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

--- Comment #4 from Andreas Schwab <schwab@linux-m68k.org> 2013-04-07 16:30:10 UTC ---
Neither r196686 nor r197531 are revisions on the trunk.

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

Dominique d'Humieres <dominiq at lps dot ens.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |steven at gcc dot gnu.org

--- Comment #5 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2013-04-07 17:33:12 UTC ---
> Neither r196686 nor r197531 are revisions on the trunk.

Well, I record the revision number given by svn:

Updated to revision xxxxxx

The corresponding actual revision for trunk may be some number below, e.g.,
197530 for 197531, or 196683 for 196686. If you know a simple way to get the
last actual revision for trunk (branch), I'll be glad to use it.

Note that between r197531 and r197541, the only changes I can find are r197533
and r197534. Both are supposed to have been tested on
powerpc64-unknown-linux-gnu (see
http://gcc.gnu.org/ml/gcc-patches/2013-04/msg00263.html and
http://gcc.gnu.org/ml/gcc-patches/2013-04/msg00333.html), but may be not with
-m32. Since there is no division in the test, I suspect that if there is a
regression in the 197531 to 197541 range, it is likely r197533, hence CCed
Steven.

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.9.0

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[wschmidt at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

Bill Schmidt <wschmidt at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |wschmidt at gcc dot gnu.org

--- Comment #6 from Bill Schmidt <wschmidt at gcc dot gnu.org> 2013-05-01 17:49:00 UTC ---
I can't confirm this today, either.  The test passes with r198500.  Andreas, do
you still see a problem with the current trunk?

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[schwab@linux-m68k.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

--- Comment #7 from Andreas Schwab <schwab@linux-m68k.org> 2013-05-01 20:07:45 UTC ---
Still failing.

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[wschmidt at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

--- Comment #8 from Bill Schmidt <wschmidt at gcc dot gnu.org> 2013-05-01 20:13:35 UTC ---
If possible, please check whether this began failing with r196872.  That commit
looks suspicious for at least one other test.  I'm stabbing in the dark since I
can't reproduce this one.

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[schwab@linux-m68k.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

--- Comment #9 from Andreas Schwab <schwab@linux-m68k.org> 2013-05-02 10:37:53 UTC ---
r196872 it is.

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[wschmidt at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

Bill Schmidt <wschmidt at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rguenth at gcc dot gnu.org

--- Comment #10 from Bill Schmidt <wschmidt at gcc dot gnu.org> 2013-05-02 15:55:07 UTC ---
Well, I still can't find the magic to reproduce this myself, but our overnight
tester shows the failure
(http://gcc.gnu.org/ml/gcc-testresults/2013-05/msg00094.html).

Richi, this is another test that started failing with the same commit as
PR56865.  In this case, it appears we are now vectorizing a loop that we
formerly expected not to.  Are you able to reproduce this?

Andreas, would you be able to provide -fdump-tree-vect-details information for
r196871 and r196872 to make it easier to debug the problem?

Thanks,
Bill

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[schwab@linux-m68k.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

--- Comment #11 from Andreas Schwab <schwab@linux-m68k.org> 2013-05-02 16:19:57 UTC ---
Created attachment 30003
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=30003
r196871 dump

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[schwab@linux-m68k.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

--- Comment #12 from Andreas Schwab <schwab@linux-m68k.org> 2013-05-02 16:36:18 UTC ---
Created attachment 30005
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=30005
r196872 dump

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[dje at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

David Edelsohn <dje at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2013-07-23
                 CC|                            |dje at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #13 from David Edelsohn <dje at gcc dot gnu.org> ---
confirmed.

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1
           Assignee|unassigned at gcc dot gnu.org      |rguenth at gcc dot gnu.org

[Bug target/56864] [4.9 regression] FAIL: gcc.dg/vect/costmodel/ppc/costmodel-vect-76b.c scan-tree-dump-times vect "vectorized 1 loops" 0

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56864

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

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

--- Comment #15 from Richard Biener <rguenth at gcc dot gnu.org> ---
Doesn't fail anymore according to

http://gcc.gnu.org/ml/gcc-testresults/2013-10/msg01946.html

and according to my cross compiler.  Also with -fvect-cost-model=dynamic
(the testsuite uses -O2 and thus now -fvect-cost-model=cheap - something
you may want to change).