[Bug c/52611] New: ia64-hp-hpux11.31, internal compiler error: verify_gimple failed

[Bug c/52611] New: ia64-hp-hpux11.31, internal compiler error: verify_gimple failed

[pda at freeshell dot org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52611

             Bug #: 52611
           Summary: ia64-hp-hpux11.31, internal compiler error:
                    verify_gimple failed
    Classification: Unclassified
           Product: gcc
           Version: 4.6.3
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: pda@freeshell.org


Created attachment 26910
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=26910
ffi.i from src/ia64/ffi.c

This isn't my code, but from the libffi package.  Attached is the preprocessor
output, and the complete error output looks like

depbase=`echo src/ia64/ffi.lo | sed 's|[^/]*$|.deps/&|;s|\.lo$||'`;\
/bin/sh ./libtool --tag=CC   --mode=compile gcc -DHAVE_CONFIG_H -I. -I..  -I.
-I../include -Iinclude -I../src  -g  -fno-strict-aliasing -fwrapv  -save-temps
-Wall -fexceptions -MT src/ia64/ffi.lo -MD -MP -MF $depbase.Tpo -c -o
src/ia64/ffi.lo ../src/ia64/ffi.c &&\
mv -f $depbase.Tpo $depbase.Plo
libtool: compile:  gcc -DHAVE_CONFIG_H -I. -I.. -I. -I../include -Iinclude
-I../src -g -fno-strict-aliasing -fwrapv -save-temps -Wall -fexceptions -MT
src/ia64/ffi.lo -MD -MP -MF src/ia64/.deps/ffi.Tpo -c ../src/ia64/ffi.c  -fPIC
-DPIC -o src/ia64/.libs/ffi.o
../src/ia64/ffi.c: In function 'ffi_prep_closure_loc':
../src/ia64/ffi.c:409:1: error: invalid types in nop conversion
long long unsigned int
void *
D.3647 = (long long unsigned int) codeloc;

../src/ia64/ffi.c:409:1: internal compiler error: verify_gimple failed
Please submit a full bug report,
with preprocessed source if appropriate.

[Bug c/52611] ia64-hp-hpux11.31, internal compiler error: verify_gimple failed

[steven at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52611

Steven Bosscher <steven at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-03-18
                 CC|                            |steven at gcc dot gnu.org
     Ever Confirmed|0                           |1

--- Comment #1 from Steven Bosscher <steven at gcc dot gnu.org> 2012-03-18 15:53:27 UTC ---
Compiled with checking enabled, obviously. With release checking the test case
does not ICE, but I haven't checked the correctness of the resulting code.

[Bug c/52611] ia64-hp-hpux11.31, internal compiler error: verify_gimple failed

[steven at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52611

Steven Bosscher <steven at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |4.8.0
      Known to fail|                            |4.6.3

--- Comment #2 from Steven Bosscher <steven at gcc dot gnu.org> 2012-03-18 16:13:31 UTC ---
Reduced test case:

$ ./xgcc --version
xgcc (GCC) 4.6.3
Copyright (C) 2011 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

$
$
$ ./xgcc -B. -S -v pr52611.i 
Reading specs from ./specs
COLLECT_GCC=./xgcc
COLLECT_LTO_WRAPPER=./lto-wrapper
Target: ia64-hp-hpux11.31
Configured with: ../gcc-4.6.3/configure --with-mpfr=/opt/cfarm/mpfr-2.3.1
--with-gmp=/opt/cfarm/gmp-4.2.4 --with-mpc=/opt/cfarm/mpc-0.8
--enable-languages=c --disable-libmudflap --disable-libssp --disable-libitm
--disable-lto --disable-bootstrap --target=ia64-hp-hpux11.31 --enable-checking
Thread model: posix
gcc version 4.6.3 (GCC) 
COLLECT_GCC_OPTIONS='-B' '.' '-S' '-v'
 ./cc1 -fpreprocessed pr52611.i -quiet -dumpbase pr52611.i -auxbase pr52611
-version -o pr52611.s
GNU C (GCC) version 4.6.3 (ia64-hp-hpux11.31)
        compiled by GNU C version 4.3.2, GMP version 4.2.4, MPFR version 2.3.1,
MPC version 0.8
GGC heuristics: --param ggc-min-expand=30 --param ggc-min-heapsize=4096
GNU C (GCC) version 4.6.3 (ia64-hp-hpux11.31)
        compiled by GNU C version 4.3.2, GMP version 4.2.4, MPFR version 2.3.1,
MPC version 0.8
GGC heuristics: --param ggc-min-expand=30 --param ggc-min-heapsize=4096
Compiler executable checksum: b14f6f9b99e71d9e506f7502606725db
pr52611.i: In function ‘foo’:
pr52611.i:5:1: error: invalid types in nop conversion
UINT64
void *
D.1263 = (UINT64) x;

pr52611.i:5:1: internal compiler error: verify_gimple failed
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.
$
$
$ cat pr52611.i 
typedef void *PTR64 __attribute__((mode(DI)));
typedef unsigned int UINT64 __attribute__((__mode__(__DI__)));

UINT64
foo (void *x)
{
  return (UINT64)(PTR64)x;
}

I cannot reproduce this with trunk of today (r185505).

[Bug c/52611] ia64-hp-hpux11.31, internal compiler error: verify_gimple failed

[steven at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52611

--- Comment #3 from Steven Bosscher <steven at gcc dot gnu.org> 2012-03-18 16:21:54 UTC ---
The problem happens very early on, during gimplification:

#1  0x00000000009e7546 in verify_gimple_assign_unary (stmt=0x7ffff7fb5f00) at
../../gcc-4.6.3/gcc/tree-cfg.c:3241
3241                error ("invalid types in nop conversion");
(gdb) l
3236
3237            /* Otherwise assert we are converting between types of the
3238               same kind.  */
3239            if (INTEGRAL_TYPE_P (lhs_type) != INTEGRAL_TYPE_P (rhs1_type))
3240              {
3241                error ("invalid types in nop conversion");
3242                debug_generic_expr (lhs_type);
3243                debug_generic_expr (rhs1_type);
3244                return true;
3245              }
(gdb) p debug_tree(lhs_type)
 <integer_type 0x7ffff7faa348 UINT64 unsigned DI
    size <integer_cst 0x7ffff7ec7a00 type <integer_type 0x7ffff7ee20a8
bit_size_type> constant 64>
    unit size <integer_cst 0x7ffff7ec7a28 type <integer_type 0x7ffff7ee2000
long unsigned int> constant 8>
    align 64 symtab 0 alias set -1 canonical type 0x7ffff7ee27e0 precision 64
min <integer_cst 0x7ffff7ec7a50 0> max <integer_cst 0x7ffff7ec79d8
18446744073709551615> context <translation_unit_decl 0x7ffff7eefda8 D.1262>>
$7 = void
(gdb) p debug_tree(rhs1_type)
 <pointer_type 0x7ffff7ee2f18
    type <void_type 0x7ffff7ee2e70 void VOID
        align 8 symtab 0 alias set -1 canonical type 0x7ffff7ee2e70
        pointer_to_this <pointer_type 0x7ffff7faa1f8>>
    public unsigned SI
    size <integer_cst 0x7ffff7ec7898 type <integer_type 0x7ffff7ee20a8
bit_size_type> constant 32>
    unit size <integer_cst 0x7ffff7ec75a0 type <integer_type 0x7ffff7ee2000
long unsigned int> constant 4>
    align 32 symtab 0 alias set -1 canonical type 0x7ffff7ee2f18
    pointer_to_this <pointer_type 0x7ffff7ef8348> reference_to_this
<reference_type 0x7ffff7ef6f18>>
$8 = void
(gdb) bt 10
#0  error (gmsgid=0xfe0890 "invalid types in nop conversion") at
../../gcc-4.6.3/gcc/diagnostic.c:751
#1  0x00000000009e7546 in verify_gimple_assign_unary (stmt=0x7ffff7fb5f00) at
../../gcc-4.6.3/gcc/tree-cfg.c:3241
#2  0x00000000009e9627 in verify_gimple_assign (stmt=0x7ffff7fb5f00) at
../../gcc-4.6.3/gcc/tree-cfg.c:3827
#3  0x00000000009e9c43 in verify_types_in_gimple_stmt (stmt=0x7ffff7fb5f00) at
../../gcc-4.6.3/gcc/tree-cfg.c:3984
#4  0x00000000009e9ea2 in verify_types_in_gimple_seq_2 (stmts=0x7ffff7fb4420)
at ../../gcc-4.6.3/gcc/tree-cfg.c:4084
#5  0x00000000009ea096 in verify_types_in_gimple_seq (stmts=0x7ffff7fb4420) at
../../gcc-4.6.3/gcc/tree-cfg.c:4101
#6  0x0000000000820ec5 in gimplify_body (body_p=0x7ffff7fa9998,
fndecl=0x7ffff7fa9900, do_parms=1 '\001') at
../../gcc-4.6.3/gcc/gimplify.c:7781
#7  0x000000000082155e in gimplify_function_tree (fndecl=0x7ffff7fa9900) at
../../gcc-4.6.3/gcc/gimplify.c:7870
#8  0x0000000000cb1eef in cgraph_analyze_function (node=0x7ffff7fc3000) at
../../gcc-4.6.3/gcc/cgraphunit.c:797
#9  0x0000000000cb2a90 in cgraph_analyze_functions () at
../../gcc-4.6.3/gcc/cgraphunit.c:983
(More stack frames follow...)
(gdb) 


Note that void* seems to be of SImode, but the test case casts "void *x" to
PTR64 which is a DImode pointer. Front end bug?

[Bug c/52611] ia64-hp-hpux11.31, internal compiler error: verify_gimple failed

[steven at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52611

--- Comment #4 from Steven Bosscher <steven at gcc dot gnu.org> 2012-03-18 16:32:15 UTC ---
I don't think we should drop a cast that changes a machine mode:

--- tree-ssa.c.orig     2012-03-18 17:32:32.000000000 +0100
+++ tree-ssa.c  2012-03-18 17:30:40.000000000 +0100
@@ -1264,6 +1264,10 @@ useless_type_conversion_p (tree outer_ty
   if (POINTER_TYPE_P (inner_type)
       && POINTER_TYPE_P (outer_type))
     {
+      /* Do not lose casts between pointers in different machine modes.  */
+      if (TYPE_MODE (inner_type) != TYPE_MODE (outer_type))
+       return false;
+
       /* Do not lose casts between pointers to different address spaces.  */
       if (TYPE_ADDR_SPACE (TREE_TYPE (outer_type))
          != TYPE_ADDR_SPACE (TREE_TYPE (inner_type)))

GCC 4.7 and GCC 4.8 have a similar check in place.

[Bug c/52611] ia64-hp-hpux11.31, internal compiler error: verify_gimple failed

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52611

--- Comment #5 from Richard Guenther <rguenth at gcc dot gnu.org> 2012-03-19 10:53:23 UTC ---
(In reply to comment #4)
> I don't think we should drop a cast that changes a machine mode:
> 
> --- tree-ssa.c.orig     2012-03-18 17:32:32.000000000 +0100
> +++ tree-ssa.c  2012-03-18 17:30:40.000000000 +0100
> @@ -1264,6 +1264,10 @@ useless_type_conversion_p (tree outer_ty
>    if (POINTER_TYPE_P (inner_type)
>        && POINTER_TYPE_P (outer_type))
>      {
> +      /* Do not lose casts between pointers in different machine modes.  */
> +      if (TYPE_MODE (inner_type) != TYPE_MODE (outer_type))
> +       return false;
> +
>        /* Do not lose casts between pointers to different address spaces.  */
>        if (TYPE_ADDR_SPACE (TREE_TYPE (outer_type))
>           != TYPE_ADDR_SPACE (TREE_TYPE (inner_type)))
> 
> GCC 4.7 and GCC 4.8 have a similar check in place.

Yes, the patch that introduced this can/should be backported.

[Bug tree-optimization/52611] ia64-hp-hpux11.31, internal compiler error: verify_gimple failed

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=52611

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.7.0
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED

--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Fixed in GCC 4.7.0 by r0-114570 (which removes the special case for void*).