[Bug middle-end/50754] New: [4.7 Regression] ICE in expand_debug_expr, at cfgexpand.c:3341

[Bug middle-end/50754] New: [4.7 Regression] ICE in expand_debug_expr, at cfgexpand.c:3341

[Joost.VandeVondele at pci dot uzh.ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50754

             Bug #: 50754
           Summary: [4.7 Regression] ICE in expand_debug_expr, at
                    cfgexpand.c:3341
    Classification: Unclassified
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: Joost.VandeVondele@pci.uzh.ch


somewhere in the last 3 days the following regression appeared in trunk:

gfortran -march=core2 -mcx16 -msahf -mno-movbe -mno-aes -mno-pclmul -mno-popcnt
-mno-abm -mno-lwp -mno-fma -mno-fma4 -mno-xop -mno-bmi -mno-bmi2 -mno-tbm
-mno-avx -mno-avx2 -mno-sse4.2 -mno-sse4.1 -mno-lzcnt --param l1-cache-size=32
--param l1-cache-line-size=64 --param l2-cache-size=4096 -mtune=core2 -g -O3
-ffast-math -funroll-loops -ftree-vectorize -ffree-form  bug.f90

 <vec_perm_expr 0x7fcf671e41f8
    type <vector_type 0x7fcf67110540
        type <real_type 0x7fcf670aef18 real(kind=8) DF
            size <integer_cst 0x7fcf6709eec0 constant 64>
            unit size <integer_cst 0x7fcf6709eee0 constant 8>
            align 64 symtab 0 alias set 2 canonical type 0x7fcf670aef18
precision 64
            pointer_to_this <pointer_type 0x7fcf670bc150>>
        V2DF
        size <integer_cst 0x7fcf670b1400 constant 128>
        unit size <integer_cst 0x7fcf670b1420 constant 16>
        align 128 symtab 0 alias set 2 canonical type 0x7fcf67110540 nunits 2
        pointer_to_this <pointer_type 0x7fcf67117c78>>

    arg 0 <ssa_name 0x7fcf671d8cd0 type <vector_type 0x7fcf67110540>
        visited var <var_decl 0x7fcf671df1e0 vect_var_.27>def_stmt
vect_var_.27_126 = MEM[(real(kind=8)[3] *)&respos + 8B];

        version 126> arg 1 <ssa_name 0x7fcf671d8cd0>
    arg 2 <vector_cst 0x7fcf671cc5a0
        type <vector_type 0x7fcf671c1c78 type <integer_type 0x7fcf670ae7e0>
            unsigned V2DI size <integer_cst 0x7fcf670b1400 128> unit size
<integer_cst 0x7fcf670b1420 16>
            align 128 symtab 0 alias set -1 canonical type 0x7fcf671c1c78
nunits 2>
        constant
        elt0:  <integer_cst 0x7fcf671c6400 constant 1>
        elt1:  <integer_cst 0x7fcf670b1300 constant 0>>
    bug.f90:9:0>
bug.f90: In function ‘calcbox’:
bug.f90:4:0: internal compiler error: in expand_debug_expr, at cfgexpand.c:3341

> cat bug.f90
MODULE gauss_colloc
  INTEGER, PARAMETER :: dp=8
CONTAINS
FUNCTION calcBox() RESULT(res)
    REAL(dp) :: cci0, cci1, cci2, delta_i, m(0:2,0:2), maxr2, r_0, sqDi
    REAL(dp), DIMENSION(0:2)    :: l, resPos
    r_0=0.0_dp
    DO i=0,2
        r_0=r_0-0.5*resPos(2-i)*l(i)
    END DO
    cci0 = -((-4.0_dp * m(2,2) * r_0 * m(1,1) + &
        m(2,2) * l(1) ** 2 + l(2) ** 2 * m(1,1) &
        - 2.0_dp * l(1) * m(1,2) * l(2) + 4.0_dp * r_0 * m(1,2) ** 2) &
        / (m(2,2) * m(1,1) - m(1,2) ** 2)) / 4.0_dp-maxr2
    delta_i=cci1*cci1-4.0_dp*cci2*cci0
    IF (delta_i<0.0_dp) THEN
        imin=fullShift(2)+CEILING((-cci1-sqDi)/(2.0_dp*cci2))
    END IF
END FUNCTION
END MODULE

[Bug middle-end/50754] [4.7 Regression] ICE in expand_debug_expr, at cfgexpand.c:3341

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50754

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011-10-17
     Ever Confirmed|0                           |1

--- Comment #1 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2011-10-17 09:38:03 UTC ---
Confirmed on x86_64-apple-darwin10 revision 180062 with '-O2 -ftree-vectorize
-ffast-math -g', slightly reduced (valid?) test

MODULE gauss_colloc
  INTEGER, PARAMETER :: dp=8
CONTAINS
FUNCTION calcBox() RESULT(res)
implicit none
    integer :: i, imin
    REAL(dp) :: r_0, res
    REAL(dp), DIMENSION(0:2)    :: l, resPos
    l = 1.0
    resPos = [1.0,2.0,3.0]
    r_0=0.0_dp
    DO i=0,2
        r_0=r_0-0.5*resPos(2-i)*l(i)
    END DO
    imin =0
    IF (r_0<0.0_dp) THEN
        imin=1
    END IF
    res=imin
END FUNCTION
END MODULE

[Bug middle-end/50754] [4.7 Regression] ICE in expand_debug_expr, at cfgexpand.c:3341

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50754

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at redhat dot com

--- Comment #2 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2011-10-17 15:45:12 UTC ---
Revision 179960 is OK and the ICE does not show for compilers configured with
--enable-checking=release.

[Bug middle-end/50754] [4.7 Regression] ICE in expand_debug_expr, at cfgexpand.c:3341

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50754

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

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

--- Comment #3 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2011-10-18 13:28:48 UTC ---
This is due to revision 180047

Author:    rth
Date:    Sat Oct 15 19:36:50 2011 UTC (2 days, 17 hours ago)
Changed paths:    5
Log Message:    
Use VEC_PERM_EXPR in the vectorizer.

        * tree-vect-slp.c: Include langhooks.h.
        (vect_create_mask_and_perm): Emit VEC_PERM_EXPR, not a builtin.
        (vect_transform_slp_perm_load): Use can_vec_perm_expr_p.  Simplify
        mask creation for VEC_PERM_EXPR.
        * tree-vect-stmts.c (perm_mask_for_reverse): Return the mask,
        not the builtin.
        (reverse_vec_elements): Emit VEC_PERM_EXPR not a builtin.
        * Makefile.in (tree-vect-slp.o): Update dependency.
        * optabs.c (can_vec_perm_expr_p): Allow NULL as unknown constant.

Note that the fix in revision 80100

Author:    rth
Date:    Mon Oct 17 17:02:05 2011 UTC (20 hours, 24 minutes ago)
Changed paths:    2
Log Message:    
PR 50746
        * optabs.c (expand_vec_perm_expr): Fix indexing error.

does solve the problem.

[Bug middle-end/50754] [4.7 Regression] ICE in expand_debug_expr, at cfgexpand.c:3341

[Joost.VandeVondele at pci dot uzh.ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50754

Joost VandeVondele <Joost.VandeVondele at pci dot uzh.ch> changed:

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

--- Comment #4 from Joost VandeVondele <Joost.VandeVondele at pci dot uzh.ch> 2011-10-18 19:19:09 UTC ---
OK.. the latest version passes here as well. Let's close as fixed

[Bug middle-end/50754] [4.7 Regression] ICE in expand_debug_expr, at cfgexpand.c:3341

[Joost.VandeVondele at pci dot uzh.ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50754

Joost VandeVondele <Joost.VandeVondele at pci dot uzh.ch> changed:

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

--- Comment #5 from Joost VandeVondele <Joost.VandeVondele at pci dot uzh.ch> 2011-10-18 19:22:04 UTC ---
No, not quite. With the original commandline I still see a failure here for 
gcc version 4.7.0 20111018 (experimental) [trunk revision 180161] (GCC)

[Bug middle-end/50754] [4.7 Regression] ICE in expand_debug_expr, at cfgexpand.c:3341

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50754

--- Comment #6 from Dominique d'Humieres <dominiq at lps dot ens.fr> 2011-10-18 19:43:39 UTC ---
> does solve the problem.

Oops!-( I ment "does not solve the problem": when the fingers don't follow the
brain;-).

[Bug middle-end/50754] [4.7 Regression] ICE in expand_debug_expr, at cfgexpand.c:3341

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50754

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

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

--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-10-18 22:03:28 UTC ---
Created attachment 25550
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25550
gcc47-pr50754.patch

Untested fix.

[Bug middle-end/50754] [4.7 Regression] ICE in expand_debug_expr, at cfgexpand.c:3341

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50754

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-10-19 19:31:25 UTC ---
Author: jakub
Date: Wed Oct 19 19:31:14 2011
New Revision: 180201

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=180201
Log:
    PR middle-end/50754
    * cfgexpand.c (expand_debug_expr): Handle WIDEN_LSHIFT_EXPR, ignore
    VEC_PERM_EXPR.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/cfgexpand.c

[Bug middle-end/50754] [4.7 Regression] ICE in expand_debug_expr, at cfgexpand.c:3341

[Joost.VandeVondele at pci dot uzh.ch]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50754

Joost VandeVondele <Joost.VandeVondele at pci dot uzh.ch> changed:

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

--- Comment #9 from Joost VandeVondele <Joost.VandeVondele at pci dot uzh.ch> 2011-10-21 05:43:11 UTC ---
this is fixed now in trunk, I assume this can be closed.