[Bug middle-end/48836] New: internal compiler error: in execute_todo, at passes.c:1261

[Bug middle-end/48836] New: internal compiler error: in execute_todo, at passes.c:1261

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48836

           Summary: internal compiler error: in execute_todo, at
                    passes.c:1261
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: hubicka@gcc.gnu.org


Created attachment 24152
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=24152
testcase.

jh@evans:/abuild/jh/build-mozilla-new11-lto-noelfhackO3/extensions/spellcheck/src>
/abuild/jh/trunk-install/bin/g++ -O3 ~/mozInlineSpellWordUtil.ii  -flto
-fno-strict-aliasing  -S -g -fno-exceptions
/abuild/jh/mozilla-central2/mozilla-central/extensions/spellcheck/src/mozInlineSpellWordUtil.cpp:
In member function '_ZNK14WordSplitState17ClassifyCharacterEii.constprop.25':
/abuild/jh/mozilla-central2/mozilla-central/extensions/spellcheck/src/mozInlineSpellWordUtil.cpp:863:1:
internal compiler error: in execute_todo, at passes.c:1261
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://gcc.gnu.org/bugs.html> for instructions.

[Bug middle-end/48836] internal compiler error: in execute_todo, at passes.c:1261

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48836

--- Comment #1 from Jan Hubicka <hubicka at gcc dot gnu.org> 2011-04-30 16:32:30 UTC ---
The reason why we conclude updating is needed is redirecting of callees:
#0  0x00000000006abdd4 in bitmap_element_link (head=0x7ffff539b760,
bit=Unhandled dwarf expression opcode 0xf3
) at ../../gcc/bitmap.c:431
#1  bitmap_set_bit (head=0x7ffff539b760, bit=Unhandled dwarf expression opcode
0xf3
) at ../../gcc/bitmap.c:653
#2  0x0000000000a7225a in finalize_ssa_defs (stmt=0x7ffff349a720) at
../../gcc/tree-ssa-operands.c:463
#3  finalize_ssa_stmt_operands (stmt=0x7ffff349a720) at
../../gcc/tree-ssa-operands.c:566
#4  build_ssa_operands (stmt=0x7ffff349a720) at
../../gcc/tree-ssa-operands.c:1078
#5  update_stmt_operands (stmt=0x7ffff349a720) at
../../gcc/tree-ssa-operands.c:1137
#6  0x0000000000b9fe63 in cgraph_redirect_edge_call_stmt_to_callee
(e=0x7ffff30f1548) at ../../gcc/cgraphunit.c:2209
#7  0x0000000000bb47d3 in inline_transform (node=Unhandled dwarf expression
opcode 0xf3
) at ../../gcc/ipa-inline-transform.c:305


We are probably right that update_ssa TODO should be dropped with aliasing on.

This patch fixes the problem, but I am not sure it is the best approach around.
Index: ipa-inline-transform.c
===================================================================
--- ipa-inline-transform.c      (revision 173216)
+++ ipa-inline-transform.c      (working copy)
@@ -45,6 +45,7 @@
 #include "ipa-prop.h"
 #include "ipa-inline.h"
 #include "tree-inline.h"
+#include "tree-pass.h"

 int ncalls_inlined;
 int nfunctions_inlined;
@@ -305,6 +306,9 @@
       cgraph_redirect_edge_call_stmt_to_callee (e);
       if (!e->inline_failed || warn_inline)
         inline_p = true;
+      /* Redirecting edges might lead to a need for vops to be recomputed.  */
+      if (need_ssa_update_p (cfun))
+       todo |= TODO_update_ssa_only_virtuals;
     }

   if (inline_p)

[Bug middle-end/48836] internal compiler error: in execute_todo, at passes.c:1261

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48836

--- Comment #2 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-05-02 10:26:20 UTC ---
+      /* Redirecting edges might lead to a need for vops to be recomputed.  */
+      if (need_ssa_update_p (cfun))

you can omit that check, the updater won't do anything if there is nothing
to do.

[Bug middle-end/48836] internal compiler error: in execute_todo, at passes.c:1261

[hubicka at ucw dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48836

--- Comment #3 from Jan Hubicka <hubicka at ucw dot cz> 2011-05-02 11:37:22 UTC ---
> +      /* Redirecting edges might lead to a need for vops to be recomputed.  */
> +      if (need_ssa_update_p (cfun))
> 
> you can omit that check, the updater won't do anything if there is nothing
> to do.

Yep, I worked that out in meantime. So should I re-check with that change and
commit?
Thanks,
Honza

[Bug middle-end/48836] internal compiler error: in execute_todo, at passes.c:1261

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48836

--- Comment #4 from Jan Hubicka <hubicka at gcc dot gnu.org> 2011-06-12 18:08:55 UTC ---
Author: hubicka
Date: Sun Jun 12 18:08:52 2011
New Revision: 174970

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=174970
Log:
    PR middle-end/48836
    * ipa-inline-transform.c: Include tree-pass.h
    (inline_transform): Set TODO_update_ssa_only_virtuals.
    * Makefile.in (ipa-inline-transform.o): Add tree-pass.h.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/Makefile.in
    trunk/gcc/ipa-inline-transform.c

[Bug middle-end/48836] internal compiler error: in execute_todo, at passes.c:1261

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48836

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

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

--- Comment #5 from Jan Hubicka <hubicka at gcc dot gnu.org> 2011-06-12 18:15:04 UTC ---
Fixed.