[Bug tree-optimization/51356] New: [4.7 Regression] FAIL: gfortran.dg/anyallcount_1.f90 -O3 (internal compiler error) on powerpc-apple-darwin9 with -m32

[Bug tree-optimization/51356] New: [4.7 Regression] FAIL: gfortran.dg/anyallcount_1.f90 -O3 (internal compiler error) on powerpc-apple-darwin9 with -m32

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51356

             Bug #: 51356
           Summary: [4.7 Regression] FAIL: gfortran.dg/anyallcount_1.f90
                    -O3  (internal compiler error) on
                    powerpc-apple-darwin9 with -m32
    Classification: Unclassified
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: dominiq@lps.ens.fr
                CC: iains@gcc.gnu.org, irar@gcc.gnu.org, jakub@gcc.gnu.org
              Host: powerpc-apple-darwin9
            Target: powerpc-apple-darwin9
             Build: powerpc-apple-darwin9


Between revisions 181105 (OK) and 181301, the following tests have started to
fail on powerpc-apple-darwin9 with -m32:

FAIL: gfortran.dg/anyallcount_1.f90  -O3 -fomit-frame-pointer  (internal
compiler error)
FAIL: gfortran.dg/anyallcount_1.f90  -O3 -fomit-frame-pointer  (test for excess
errors)
WARNING: gfortran.dg/anyallcount_1.f90  -O3 -fomit-frame-pointer  compilation
failed to produce executable
FAIL: gfortran.dg/anyallcount_1.f90  -O3 -fomit-frame-pointer -funroll-loops 
(internal compiler error)
FAIL: gfortran.dg/anyallcount_1.f90  -O3 -fomit-frame-pointer -funroll-loops 
(test for excess errors)
WARNING: gfortran.dg/anyallcount_1.f90  -O3 -fomit-frame-pointer -funroll-loops
 compilation failed to produce executable
FAIL: gfortran.dg/anyallcount_1.f90  -O3 -fomit-frame-pointer
-funroll-all-loops -finline-functions  (internal compiler error)
FAIL: gfortran.dg/anyallcount_1.f90  -O3 -fomit-frame-pointer
-funroll-all-loops -finline-functions  (test for excess errors)
WARNING: gfortran.dg/anyallcount_1.f90  -O3 -fomit-frame-pointer
-funroll-all-loops -finline-functions  compilation failed to produce executable
FAIL: gfortran.dg/anyallcount_1.f90  -O3 -g  (internal compiler error)
FAIL: gfortran.dg/anyallcount_1.f90  -O3 -g  (test for excess errors)
WARNING: gfortran.dg/anyallcount_1.f90  -O3 -g  compilation failed to produce
executable

This can be reproduced with '-O2 -ftree-vectorize' on the following reduced
test:

program main
   implicit none
   character(len=*), parameter :: f='(3L1)'
   character(len=*), parameter :: g='(3I1)'
   real, dimension(3,3) :: a
   logical(kind=8), dimension(3,3) :: m8
   character(len=3) :: res
   data a /-1.0, -2.0, -3.0, 2.0, 1.0, -2.1, 1.0, 2.0, 3.0 /

   m8 = a > 0

   write (unit=res,fmt=f) any(m8,dim=1)
   if (res /= 'FTT') call abort
   write (unit=res,fmt=f) any(m8,dim=2)
   if (res /= 'TTT') call abort

   write (unit=res,fmt=f) all(m8,dim=1)
   if (res /= 'FFT') call abort
   write (unit=res,fmt=f) all(m8,dim=2)
   if (res /= 'FFF') call abort

   write (unit=res,fmt=g) count(m8,dim=1)
   if (res /= '023') call abort

   write (unit=res,fmt=g) count(m8,dim=2)
   if (res /= '221') call abort

end program main

[karma] /Users/dominiq% gfc -O2 -ftree-vectorize anyallcount_1_red.f90
anyallcount_1_red.f90: In function 'MAIN__':
anyallcount_1_red.f90:2:0: internal compiler error: Bus error

The backtrace I get is

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_PROTECTION_FAILURE at address: 0x0000000c
0x0085eaac in vect_analyze_data_refs_alignment (loop_vinfo=<value temporarily
unavailable, due to optimizations>, bb_vinfo=0x0) at tree-vectorizer.h:333
333              && (loop->inner == (gimple_bb (stmt))->loop_father));
(gdb) bt
#0  0x0085eaac in vect_analyze_data_refs_alignment (loop_vinfo=<value
temporarily unavailable, due to optimizations>, bb_vinfo=0x0) at
tree-vectorizer.h:333
#1  0x0085ea68 in vect_analyze_data_refs_alignment (loop_vinfo=<value
temporarily unavailable, due to optimizations>, bb_vinfo=0x0) at
../../work/gcc/tree-vect-data-refs.c:806

The 'Bus error' is triggered by any of the WRITE statements.

[Bug tree-optimization/51356] [4.7 Regression] FAIL: gfortran.dg/anyallcount_1.f90 -O3 (internal compiler error) on powerpc-apple-darwin9 with -m32

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51356

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|powerpc-apple-darwin9       |powerpc-*-*
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011-11-30
               Host|powerpc-apple-darwin9       |
     Ever Confirmed|0                           |1
              Build|powerpc-apple-darwin9       |

[Bug tree-optimization/51356] [4.7 Regression] FAIL: gfortran.dg/anyallcount_1.f90 -O3 (internal compiler error) on powerpc-apple-darwin9 with -m32

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51356

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-11-30 17:28:47 UTC ---
Created attachment 25956
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25956
gcc47-pr51356.patch

The problem is that PowerPC returns a TImode vector from
get_vectype_for_scalar_type of DImode boolean/integer type.
vect_recog_bool_pattern sees a store of a DImode boolean, attempts to optimize
it but doesn't just create the pattern stmts etc., but also already changes
DR_STMT.  Later on in vect_pattern_recog_1
if (VECTOR_MODE_P (TYPE_MODE (type_in)))
is unfortunately not true (while type_in is a vector, it doesn't have vector
mode) and thus the caller thinks type_in is a scalar type, but
get_vectype_for_scalar_type on it fails, therefore vect_mark_pattern_stmts
isn't called, yet DR_STMT has been already adjusted.

This patch fixes it by giving up in that case already in
vect_recog_bool_pattern.  Alternative would be to check for VECTOR_TYPE in
vect_pattern_recog_1 instead of testing for vector mode?  Not sure how useful
that would be though, I doubt we could actually vectorize it anyway.

[Bug tree-optimization/51356] [4.7 Regression] FAIL: gfortran.dg/anyallcount_1.f90 -O3 (internal compiler error) on powerpc-apple-darwin9 with -m32

[iains at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51356

--- Comment #2 from Iain Sandoe <iains at gcc dot gnu.org> 2011-12-01 08:31:57 UTC ---
works for me on top of 181853 (powerpc-darwin9).

[Bug tree-optimization/51356] [4.7 Regression] FAIL: gfortran.dg/anyallcount_1.f90 -O3 (internal compiler error) on powerpc-apple-darwin9 with -m32

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51356

--- Comment #3 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2011-12-01 14:57:13 UTC ---
> works for me on top of 181853 (powerpc-darwin9).

For me too on top of 181782 (powerpc-darwin9, regtested gcc and gfortran) and
181863 (x86_64-apple-darwin10, full regtest).

> ... I doubt we could actually vectorize it anyway.

The test was not vectorized before the patch nor after it.

[Bug tree-optimization/51356] [4.7 Regression] FAIL: gfortran.dg/anyallcount_1.f90 -O3 (internal compiler error) on powerpc-apple-darwin9 with -m32

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51356

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-12-01 19:14:06 UTC ---
Author: jakub
Date: Thu Dec  1 19:14:03 2011
New Revision: 181891

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=181891
Log:
    PR tree-optimization/51356
    * tree-vect-patterns.c (vect_recog_bool_pattern): Give up if
    vectype doesn't have VECTOR_MODE_P.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/tree-vect-patterns.c

[Bug tree-optimization/51356] [4.7 Regression] FAIL: gfortran.dg/anyallcount_1.f90 -O3 (internal compiler error) on powerpc-apple-darwin9 with -m32

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51356

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

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

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-12-01 19:16:53 UTC ---
Fixed.