[Bug bootstrap/58388] New: LTO profiledbootstrap fails in stage feedback for tree-ssa-structalias.c with "internal compiler error: in try_make_edge_direct_simple_call, at ipa-prop.c:2606"

[Bug bootstrap/58388] New: LTO profiledbootstrap fails in stage feedback for tree-ssa-structalias.c with "internal compiler error: in try_make_edge_direct_simple_call, at ipa-prop.c:2606"

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58388

            Bug ID: 58388
           Summary: LTO profiledbootstrap fails in stage feedback for
                    tree-ssa-structalias.c with "internal compiler error:
                    in try_make_edge_direct_simple_call, at
                    ipa-prop.c:2606"
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: bootstrap
          Assignee: unassigned at gcc dot gnu.org
          Reporter: ubizjak at gmail dot com
                CC: hubicka at gcc dot gnu.org

LTO profiled bootstrap of gcc, configured --with-build-config=bootstrap-lto
--enable-languages=c,c++,fortran fails [1] with:

../../gcc-svn/trunk/gcc/tree-ssa-structalias.c:7433:1: internal compiler error:
in try_make_edge_direct_simple_call, at ipa-prop.c:2606
 }
 ^
0xcd3f85 try_make_edge_direct_simple_call
    ../../gcc-svn/trunk/gcc/ipa-prop.c:2603
0xcd3f85 update_indirect_edges_after_inlining
    ../../gcc-svn/trunk/gcc/ipa-prop.c:2704
0xcd3f85 propagate_info_to_inlined_callees
    ../../gcc-svn/trunk/gcc/ipa-prop.c:2771
0xcd4a0b ipa_propagate_indirect_call_infos(cgraph_edge*, vec<cgraph_edge*,
va_heap, vl_ptr>*)
    ../../gcc-svn/trunk/gcc/ipa-prop.c:2933
0x16def84 inline_call(cgraph_edge*, bool, vec<cgraph_edge*, va_heap, vl_ptr>*,
int*, bool)
    ../../gcc-svn/trunk/gcc/ipa-inline-transform.c:271
0x16dc620 inline_small_functions
    ../../gcc-svn/trunk/gcc/ipa-inline.c:1769
0x16dcd5f ipa_inline
    ../../gcc-svn/trunk/gcc/ipa-inline.c:2010
0x16dcd5f execute
    ../../gcc-svn/trunk/gcc/ipa-inline.c:2380

[1] http://gcc.gnu.org/ml/gcc-regression/2013-09/msg00344.html

[Bug bootstrap/58388] LTO profiledbootstrap fails in stage feedback for tree-ssa-structalias.c with "internal compiler error: in try_make_edge_direct_simple_call, at ipa-prop.c:2606"

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58388

Jan Hubicka <hubicka at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mjambor at suse dot cz

--- Comment #1 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
Martin,
this looks bit more like yours at first glance.  I will try to reproduce it.

[Bug bootstrap/58388] [4.9 Regression] LTO profiledbootstrap fails in stage feedback for tree-ssa-structalias.c with "internal compiler error: in try_make_edge_direct_simple_call, at ipa-prop.c:2606"

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58388

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.9.0
            Summary|LTO profiledbootstrap fails |[4.9 Regression] LTO
                   |in stage feedback for       |profiledbootstrap fails in
                   |tree-ssa-structalias.c with |stage feedback for
                   |"internal compiler error:   |tree-ssa-structalias.c with
                   |in                          |"internal compiler error:
                   |try_make_edge_direct_simple |in
                   |_call, at ipa-prop.c:2606"  |try_make_edge_direct_simple
                   |                            |_call, at ipa-prop.c:2606"

[Bug bootstrap/58388] [4.9 Regression] LTO profiledbootstrap fails in stage feedback for tree-ssa-structalias.c with "internal compiler error: in try_make_edge_direct_simple_call, at ipa-prop.c:2606"

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58388

Jan Hubicka <hubicka at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2013-09-12
     Ever confirmed|0                           |1

--- Comment #2 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
Martin,
I think the problem is just the assert itself:
      gcc_checking_assert (cs->callee
                           && (jfunc->type != IPA_JF_CONST
                               || !cgraph_node_for_jfunc (jfunc)
                               || cs->callee == cgraph_node_for_jfunc
(jfunc)));

Since have possibly resolved speculation, cs->callee may point to a clone
(either ipa-cp or inline created).  Shall we just drop the test or weaken it if
the edge was previously speculative?

[Bug bootstrap/58388] [4.9 Regression] LTO profiledbootstrap fails in stage feedback for tree-ssa-structalias.c with "internal compiler error: in try_make_edge_direct_simple_call, at ipa-prop.c:2606"

[jamborm at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58388

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

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

--- Comment #3 from Martin Jambor <jamborm at gcc dot gnu.org> ---
(In reply to Jan Hubicka from comment #2)
> Martin,
> I think the problem is just the assert itself:
>       gcc_checking_assert (cs->callee
>                            && (jfunc->type != IPA_JF_CONST
>                                || !cgraph_node_for_jfunc (jfunc)
>                                || cs->callee == cgraph_node_for_jfunc
> (jfunc)));
> 
> Since have possibly resolved speculation, cs->callee may point to a clone
> (either ipa-cp or inline created).  Shall we just drop the test or weaken it
> if the edge was previously speculative?

It took me some time to reproduce this in a way that would give me
enough debug info to look at stuff.  Nevertheless, indeed that is the
case, cs->callee is an inline clone.  I'll add (cs != ie) to the
disjunctions in the assert.

[Bug bootstrap/58388] [4.9 Regression] LTO profiledbootstrap fails in stage feedback for tree-ssa-structalias.c with "internal compiler error: in try_make_edge_direct_simple_call, at ipa-prop.c:2606"

[hubicka at ucw dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58388

--- Comment #4 from Jan Hubicka <hubicka at ucw dot cz> ---
> It took me some time to reproduce this in a way that would give me
> enough debug info to look at stuff.  Nevertheless, indeed that is the

I usually debug these crashes by simply running stage1 compiler when
stagefeedback crashes.

> case, cs->callee is an inline clone.  I'll add (cs != ie) to the
> disjunctions in the assert.
That sounds OK.

[Bug bootstrap/58388] [4.9 Regression] LTO profiledbootstrap fails in stage feedback for tree-ssa-structalias.c with "internal compiler error: in try_make_edge_direct_simple_call, at ipa-prop.c:2606"

[jamborm at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58388

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-p
                   |                            |atches/2013-09/msg01010.htm
                   |                            |l

--- Comment #5 from Martin Jambor <jamborm at gcc dot gnu.org> ---
I have posted a patch to the mailing list:

http://gcc.gnu.org/ml/gcc-patches/2013-09/msg01010.html

[Bug bootstrap/58388] [4.9 Regression] LTO profiledbootstrap fails in stage feedback for tree-ssa-structalias.c with "internal compiler error: in try_make_edge_direct_simple_call, at ipa-prop.c:2606"

[jamborm at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58388

--- Comment #6 from Martin Jambor <jamborm at gcc dot gnu.org> ---
Author: jamborm
Date: Fri Sep 13 12:04:54 2013
New Revision: 202563

URL: http://gcc.gnu.org/viewcvs?rev=202563&root=gcc&view=rev
Log:
2013-09-13  Martin Jambor  <mjambor@suse.cz>

    PR bootstrap/58388
    * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
    the assert if the edge was a speculative one.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/ipa-prop.c

[Bug bootstrap/58388] [4.9 Regression] LTO profiledbootstrap fails in stage feedback for tree-ssa-structalias.c with "internal compiler error: in try_make_edge_direct_simple_call, at ipa-prop.c:2606"

[jamborm at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58388

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

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

--- Comment #7 from Martin Jambor <jamborm at gcc dot gnu.org> ---
Fixed.