[Bug c++/45875] New: ice in gimple_fold_obj_type_ref_known_binfo with -O2

[Bug c++/45875] New: ice in gimple_fold_obj_type_ref_known_binfo with -O2

[dcb314 at hotmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

           Summary: ice in gimple_fold_obj_type_ref_known_binfo with -O2
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: dcb314@hotmail.com


I just tried to compile package ddd-3.3.12 with the latest 4.6 snapshot
20101002 on a Fedora Linux box. The compiler said

settings.C: At global scope:
settings.C:4009:1: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

valgrind said

==2738== Invalid read of size 2
==2738==    at 0xD04E9D: gimple_fold_obj_type_ref_known_binfo
(gimple-fold.c:144
4)
==2738==    by 0xB190D1: propagate_info_to_inlined_callees (ipa-prop.c:1527)
==2738==    by 0xB13CA5: cgraph_decide_inlining (ipa-inline.c:1248)
==2738==    by 0x87040E: execute_one_pass (passes.c:1569)
==2738==    by 0x870AE9: execute_ipa_pass_list (passes.c:1936)
==2738==    by 0xB09A87: cgraph_optimize (cgraphunit.c:1733)
==2738==    by 0xB09CC9: cgraph_finalize_compilation_unit (cgraphunit.c:1020)
==2738==    by 0x54F55C: cp_write_global_declarations (decl2.c:3928)
==2738==    by 0x91041B: toplev_main (toplev.c:951)
==2738==    by 0x325FC1EC5C: (below main) (in /lib64/libc-2.12.so)
==2738==  Address 0x0 is not stack'd, malloc'd or (recently) free'd
==2738== 

Preprocessed source code attached. Flag -O2 required.

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hubicka at gcc dot gnu.org
          Component|c++                         |tree-optimization
   Target Milestone|---                         |4.6.0
            Summary|ice in                      |[4.6 Regression] ice in
                   |gimple_fold_obj_type_ref_kn |gimple_fold_obj_type_ref_kn
                   |own_binfo with -O2          |own_binfo with -O2

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

--- Comment #2 from Richard Guenther <rguenth at gcc dot gnu.org> 2010-10-03 19:59:11 UTC ---
Btw, please don't strip CPP line notes from preprocessed source, that makes
un-including impossible.

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[dcb314 at hotmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

--- Comment #3 from dcb <dcb314 at hotmail dot com> 2010-10-03 20:22:39 UTC ---
(In reply to comment #2)
> Btw, please don't strip CPP line notes from preprocessed source, that makes
> un-including impossible.

Two questions

1. What is un-including, please ?

2. So instead of the grep -v "^#" I normally do
to strip pre-processor lines, what would you like instead ?

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

--- Comment #4 from Richard Guenther <rguenth at gcc dot gnu.org> 2010-10-03 20:39:34 UTC ---
(In reply to comment #3)
> (In reply to comment #2)
> > Btw, please don't strip CPP line notes from preprocessed source, that makes
> > un-including impossible.
> 
> Two questions
> 
> 1. What is un-including, please ?

un-include is a script that can selectively replace system header contents
with proper #include directives.  That allows for example to modify a
testcase to build with slightly different OS / multilib combinations.

> 2. So instead of the grep -v "^#" I normally do
> to strip pre-processor lines, what would you like instead ?

Nothing.  Keep it just like GCC produces it.

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[hjl.tools at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

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

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

--- Comment #5 from H.J. Lu <hjl.tools at gmail dot com> 2010-10-04 01:02:17 UTC ---
It is caused by revision 159507:

http://gcc.gnu.org/ml/gcc-cvs/2010-05/msg00559.html

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[jamborm at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

Martin Jambor <jamborm at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2010.10.13 14:48:12
     Ever Confirmed|0                           |1

--- Comment #7 from Martin Jambor <jamborm at gcc dot gnu.org> 2010-10-13 14:48:12 UTC ---
Mine.

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[dcb314 at hotmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

--- Comment #8 from dcb <dcb314 at hotmail dot com> 2010-10-13 22:14:03 UTC ---
(In reply to comment #6)
> This testcase was reduced from testsuite/g++.dg/abi/covariant3.C, but is seems
> to be the same problem.

So no one makes sure that the standard gcc/g++ testsuite
compiles without valgrind errors at all times ?

I would have thought that would be a useful sanity
check and some useful work for someone.

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

--- Comment #9 from Andrew Pinski <pinskia at gcc dot gnu.org> 2010-10-13 22:16:56 UTC ---
(In reply to comment #8)> 
> So no one makes sure that the standard gcc/g++ testsuite
> compiles without valgrind errors at all times ?

No what happens is that g++.dg/abi testsuite is not run at -O1 but only at -O0
so of course this was not detected in the testsuite even with valgrind running
:).

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[jamborm at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

Martin Jambor <jamborm at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |zsojka at seznam dot cz

--- Comment #10 from Martin Jambor <jamborm at gcc dot gnu.org> 2010-10-14 18:51:52 UTC ---
*** Bug 45921 has been marked as a duplicate of this bug. ***

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[zsojka at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

--- Comment #11 from Zdenek Sojka <zsojka at seznam dot cz> 2010-10-14 22:49:12 UTC ---
Created attachment 22045
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22045
reduced testcase from comment #1

Reduced from the original testcase

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[jamborm at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

--- Comment #12 from Martin Jambor <jamborm at gcc dot gnu.org> 2010-10-15 12:15:32 UTC ---
(In reply to comment #11)
> Created attachment 22045 [details]
> reduced testcase from comment #1
> 
> Reduced from the original testcase

Hm, this must a be a different bug than the problem with he testcase
in comment #6.  I have just posted a fix to the latter:

http://gcc.gnu.org/ml/gcc-patches/2010-10/msg01328.html

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[jamborm at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

--- Comment #13 from Martin Jambor <jamborm at gcc dot gnu.org> 2010-10-21 13:06:20 UTC ---
I have just posted a patch fixing the original issue reported in this
bug to the mailing list:

http://gcc.gnu.org/ml/gcc-patches/2010-10/msg01822.html

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[jamborm at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

--- Comment #14 from Martin Jambor <jamborm at gcc dot gnu.org> 2010-10-21 14:35:05 UTC ---
Author: jamborm
Date: Thu Oct 21 14:34:58 2010
New Revision: 165780

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=165780
Log:
2010-10-21  Martin Jambor  <mjambor@suse.cz>

    PR tree-optimization/45875
    * tree.c (get_binfo_at_offset): Remove initial zero offset test.

    * testsuite/g++.dg/ipa/pr45875.C: New test.


Added:
    trunk/gcc/testsuite/g++.dg/ipa/pr45875.C
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/tree.c

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

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

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

--- Comment #15 from Jakub Jelinek <jakub at gcc dot gnu.org> 2010-10-26 03:51:52 UTC ---
Fixed.

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[jamborm at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

Martin Jambor <jamborm at gcc dot gnu.org> changed:

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

--- Comment #16 from Martin Jambor <jamborm at gcc dot gnu.org> 2010-11-01 16:00:32 UTC ---
No, it is not, it happened that two different bug were reported in
this PR.  On the other hand, I have just submitted a patch for the
remaining issue:

http://gcc.gnu.org/ml/gcc-patches/2010-11/msg00043.html

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[jamborm at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

--- Comment #17 from Martin Jambor <jamborm at gcc dot gnu.org> 2010-11-02 15:06:28 UTC ---
Author: jamborm
Date: Tue Nov  2 15:06:21 2010
New Revision: 166190

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=166190
Log:
2010-11-02  Martin Jambor  <mjambor@suse.cz>

    PR tree-optimization/45875
    * gimple-fold.c (get_first_base_binfo_with_virtuals): Removed.
    (gimple_get_relevant_ref_binfo): Detect primary bases according to
    their field offset.

    * testsuite/g++.dg/torture/pr45875.C: New test.


Added:
    trunk/gcc/testsuite/g++.dg/torture/pr45875.C
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/gimple-fold.c
    trunk/gcc/testsuite/ChangeLog

[Bug tree-optimization/45875] [4.6 Regression] ice in gimple_fold_obj_type_ref_known_binfo with -O2

[jamborm at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45875

Martin Jambor <jamborm at gcc dot gnu.org> changed:

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

--- Comment #18 from Martin Jambor <jamborm at gcc dot gnu.org> 2010-11-02 15:25:02 UTC ---
Now it is fixed.