[Bug debug/46893] New: [4.5/4.6 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[Bug debug/46893] New: [4.5/4.6 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46893

           Summary: [4.5/4.6 Regression] ICE: in trunc_int_for_mode, at
                    explow.c:56 with -O -g
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: debug
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: zsojka@seznam.cz
              Host: x86_64-pc-linux-gnu
            Target: x86_64-pc-linux-gnu


Created attachment 22714
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22714
reduced testcase (from gcc.dg/980526-1.c)

Compiler output:
$ gcc -O -g pr46893.c        
pr46893.c: In function 'foo':
pr46893.c:12:1: internal compiler error: in trunc_int_for_mode, at explow.c:56
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

(gdb) bt
#0  fancy_abort (file=0x1103c60 "/mnt/svn/gcc-trunk/gcc/explow.c", line=56,
function=0x1103db0 "trunc_int_for_mode")
    at /mnt/svn/gcc-trunk/gcc/diagnostic.c:892
#1  0x000000000068c5ef in trunc_int_for_mode (c=<value optimized out>,
mode=<value optimized out>)
    at /mnt/svn/gcc-trunk/gcc/explow.c:56
#2  0x0000000000671df9 in gen_int_mode (c=<value optimized out>, mode=<value
optimized out>)
    at /mnt/svn/gcc-trunk/gcc/emit-rtl.c:426
#3  0x00000000008b7a76 in simplify_const_unary_operation (code=<value optimized
out>, mode=DFmode, op=0x7ffff5ccf390, 
    op_mode=<value optimized out>) at
/mnt/svn/gcc-trunk/gcc/simplify-rtx.c:1377
#4  0x00000000008b3c50 in simplify_unary_operation (code=NEG, mode=DFmode,
op=0x7ffff5ccf390, op_mode=DFmode)
    at /mnt/svn/gcc-trunk/gcc/simplify-rtx.c:522
#5  0x0000000000608246 in cselib_expand_value_rtx_1 (orig=0x7ffff5ccf3b0,
evd=<value optimized out>, 
    max_depth=<value optimized out>) at /mnt/svn/gcc-trunk/gcc/cselib.c:1475
#6  0x000000000060964f in cselib_expand_value_rtx_cb (orig=<value optimized
out>, regs_active=<value optimized out>, 
    max_depth=<value optimized out>, cb=<value optimized out>, data=<value
optimized out>)
    at /mnt/svn/gcc-trunk/gcc/cselib.c:1226
#7  0x0000000000a7bc22 in vt_expand_loc_callback (x=0x17e1e30, regs=0x17a44c0,
max_depth=8, data=0x7fffffffd370)
    at /mnt/svn/gcc-trunk/gcc/var-tracking.c:7008
#8  0x0000000000607fc2 in cselib_expand_value_rtx_1 (orig=0x17e1e30,
evd=0x7fffffffd340, max_depth=8)
    at /mnt/svn/gcc-trunk/gcc/cselib.c:1379
#9  0x000000000060964f in cselib_expand_value_rtx_cb (orig=<value optimized
out>, regs_active=<value optimized out>, 
    max_depth=<value optimized out>, cb=<value optimized out>, data=<value
optimized out>)
    at /mnt/svn/gcc-trunk/gcc/cselib.c:1226
#10 0x0000000000a68f8b in vt_expand_loc (loc=<value optimized out>, vars=<value
optimized out>)
    at /mnt/svn/gcc-trunk/gcc/var-tracking.c:7071
#11 0x0000000000a77e72 in vt_expand_loc (varp=0x17bc340, data=<value optimized
out>)
    at /mnt/svn/gcc-trunk/gcc/var-tracking.c:7065
#12 emit_note_insn_var_location (varp=0x17bc340, data=<value optimized out>) at
/mnt/svn/gcc-trunk/gcc/var-tracking.c:7182
#13 0x00000000010a6028 in htab_traverse_noresize (htab=<value optimized out>, 
    callback=0xa777e0 <emit_note_insn_var_location>, info=0x7fffffffd590) at
/mnt/svn/gcc-trunk/libiberty/hashtab.c:784
#14 0x0000000000a6f74d in emit_notes_for_changes (insn=0x7ffff7ee8798,
where=EMIT_NOTE_AFTER_INSN, 
    vars=<value optimized out>) at /mnt/svn/gcc-trunk/gcc/var-tracking.c:7531
#15 0x0000000000a756bd in emit_notes_in_bb () at
/mnt/svn/gcc-trunk/gcc/var-tracking.c:7930
#16 vt_emit_notes () at /mnt/svn/gcc-trunk/gcc/var-tracking.c:7997
#17 0x0000000000a7b713 in variable_tracking_main_1 () at
/mnt/svn/gcc-trunk/gcc/var-tracking.c:8609
#18 variable_tracking_main () at /mnt/svn/gcc-trunk/gcc/var-tracking.c:8623
#19 0x00000000007ec63f in execute_one_pass (pass=0x16366a0) at
/mnt/svn/gcc-trunk/gcc/passes.c:1553
#20 0x00000000007ec905 in execute_pass_list (pass=0x16366a0) at
/mnt/svn/gcc-trunk/gcc/passes.c:1608
#21 0x00000000007ec917 in execute_pass_list (pass=0x16331a0) at
/mnt/svn/gcc-trunk/gcc/passes.c:1609
#22 0x00000000007ec917 in execute_pass_list (pass=0x1633200) at
/mnt/svn/gcc-trunk/gcc/passes.c:1609
#23 0x000000000092d156 in tree_rest_of_compilation (fndecl=0x7ffff5cc3f00) at
/mnt/svn/gcc-trunk/gcc/tree-optimize.c:422
#24 0x0000000000af2962 in cgraph_expand_function (node=0x7ffff5cc8160) at
/mnt/svn/gcc-trunk/gcc/cgraphunit.c:1508
#25 0x0000000000af4f3a in cgraph_expand_all_functions () at
/mnt/svn/gcc-trunk/gcc/cgraphunit.c:1567
#26 cgraph_optimize () at /mnt/svn/gcc-trunk/gcc/cgraphunit.c:1827
#27 0x0000000000af54ba in cgraph_finalize_compilation_unit () at
/mnt/svn/gcc-trunk/gcc/cgraphunit.c:1031
#28 0x0000000000508cdc in c_write_global_declarations () at
/mnt/svn/gcc-trunk/gcc/c-decl.c:9845
#29 0x00000000008d69d6 in compile_file (argc=14, argv=0x7fffffffdb28) at
/mnt/svn/gcc-trunk/gcc/toplev.c:591
#30 do_compile (argc=14, argv=0x7fffffffdb28) at
/mnt/svn/gcc-trunk/gcc/toplev.c:1874
#31 toplev_main (argc=14, argv=0x7fffffffdb28) at
/mnt/svn/gcc-trunk/gcc/toplev.c:1937
#32 0x00007ffff6586bbd in __libc_start_main () from /lib/libc.so.6
#33 0x00000000004ef6b9 in _start ()


Tested revisions:
r167621 - crash
4.5 r166509 - crash
4.4 r166509 - OK

[Bug debug/46893] [4.5/4.6 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46893

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2010.12.15 01:24:18
                 CC|                            |aoliva at gcc dot gnu.org
   Target Milestone|---                         |4.6.0
     Ever Confirmed|0                           |1

--- Comment #1 from H.J. Lu <hjl.tools at gmail dot com> 2010-12-15 01:24:18 UTC ---
It is caused by revision 151362:

http://gcc.gnu.org/ml/gcc-cvs/2009-09/msg00108.html

[Bug debug/46893] [4.5/4.6 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46893

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

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

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> 2010-12-15 10:17:15 UTC ---
Created attachment 22761
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22761
gcc46-pr46893.patch

Untested fix.

[Bug debug/46893] [4.5/4.6 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[aoliva at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46893

Alexandre Oliva <aoliva at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|jakub at gcc dot gnu.org    |aoliva at gcc dot gnu.org

--- Comment #3 from Alexandre Oliva <aoliva at gcc dot gnu.org> 2010-12-16 04:43:31 UTC ---
Mine

[Bug debug/46893] [4.5/4.6 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[aoliva at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46893

Alexandre Oliva <aoliva at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #22761|0                           |1
        is obsolete|                            |

--- Comment #4 from Alexandre Oliva <aoliva at gcc dot gnu.org> 2010-12-16 04:49:23 UTC ---
Created attachment 22779
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22779
Improved patch

Jakub, your patch would work for VIEW_CONVERT_EXPR, but not for other
field-access operations that may reach that point.

I'm trying this patch. Only the first hunk is needed to fix the bug.  The last
hunk to simplify-rtx might be useful to detect other similar bugs, and the
other simplify-rtx hunks ensure the operations that are only appropriate for
int scalars are not applied to other types.

I haven't tested this yet.

[Bug debug/46893] [4.5/4.6 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46893

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

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

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> 2010-12-16 08:02:21 UTC ---
I agree the simplify-rtx.c change makes sense, but disagree on the cfgexpand.c
side.  In theory the VCE doesn't have to be the outermost expression, it can be
embedded in another handled component or another VCE, etc.
We expand tem to op0 and that only ever considers tem's mode, not mode1 or
whatever.  Thus, if op0 is a CONST_INT and thus has no mode, we should look at
tem's mode, which is where it would get its mode from if it had any.

I'm fine with limiting that just to CONST_INTs if you want.  But it also needs
to be done in both places, not just one.

[Bug debug/46893] [4.5/4.6 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46893

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.6.0                       |4.5.3

[Bug debug/46893] [4.5/4.6 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[aoliva at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46893

Alexandre Oliva <aoliva at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|aoliva at gcc dot gnu.org   |jakub at gcc dot gnu.org

--- Comment #6 from Alexandre Oliva <aoliva at gcc dot gnu.org> 2010-12-16 17:10:52 UTC ---
Sorry, I was thinking nonsense.  Jakub, your patch is good, go for it.

The other changes I made, even in as much as they make sense on their own, were
a result of the misunderstanding that led me down the wrong path, and to
suspect your patch's correctness  Now I see they're probably not useful, so I
withdraw them.

[Bug debug/46893] [4.5/4.6 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46893

--- Comment #7 from Jakub Jelinek <jakub at gcc dot gnu.org> 2010-12-16 17:14:35 UTC ---
Author: jakub
Date: Thu Dec 16 17:14:30 2010
New Revision: 167949

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=167949
Log:
    PR debug/46893
    * cfgexpand.c (expand_debug_expr): If GET_MODE (op0) is VOIDmode,
    use TYPE_MODE (TREE_TYPE (tem)) instead of mode1.

    * gcc.dg/pr46893.c: New test.

Added:
    trunk/gcc/testsuite/gcc.dg/pr46893.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/cfgexpand.c
    trunk/gcc/testsuite/ChangeLog

[Bug debug/46893] [4.5 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46893

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |4.6.0
            Summary|[4.5/4.6 Regression] ICE:   |[4.5 Regression] ICE: in
                   |in trunc_int_for_mode, at   |trunc_int_for_mode, at
                   |explow.c:56 with -O -g      |explow.c:56 with -O -g
      Known to fail|4.6.0                       |

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> 2010-12-16 19:46:54 UTC ---
Fixed on the trunk so far.

[Bug debug/46893] [4.5 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46893

--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-01-16 20:20:24 UTC ---
Author: jakub
Date: Sun Jan 16 20:20:21 2011
New Revision: 168865

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168865
Log:
    Backport from mainline
    2010-12-16  Jakub Jelinek  <jakub@redhat.com>

    PR debug/46893
    * cfgexpand.c (expand_debug_expr): If GET_MODE (op0) is VOIDmode,
    use TYPE_MODE (TREE_TYPE (tem)) instead of mode1.

    * gcc.dg/pr46893.c: New test.

Added:
    branches/gcc-4_5-branch/gcc/testsuite/gcc.dg/pr46893.c
Modified:
    branches/gcc-4_5-branch/gcc/ChangeLog
    branches/gcc-4_5-branch/gcc/cfgexpand.c
    branches/gcc-4_5-branch/gcc/testsuite/ChangeLog

[Bug debug/46893] [4.5 Regression] ICE: in trunc_int_for_mode, at explow.c:56 with -O -g

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46893

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

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

--- Comment #10 from Jakub Jelinek <jakub at gcc dot gnu.org> 2011-01-17 08:07:12 UTC ---
Fixed.