[Bug tree-optimization/54022] New: [4.8 regression] ICE in hoist_edge_and_branch_if_true

[Bug tree-optimization/54022] New: [4.8 regression] ICE in hoist_edge_and_branch_if_true

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54022

             Bug #: 54022
           Summary: [4.8 regression] ICE in hoist_edge_and_branch_if_true
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: ebotcazou@gcc.gnu.org


Created attachment 27826
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=27826
Reduced testcase

Visible on x86-64/Linux at least at -O2:

eric@hermes:~/build/gcc/native> gcc/xgcc -Bgcc -S discr2.ads -O2
+===========================GNAT BUG DETECTED==============================+
| 4.8.0 20120716 (experimental) [trunk revision 189525] (x86_64-suse-linux) GCC
error:|
| in hoist_edge_and_branch_if_true, at tree-switch-conversion.c:80         |
| Error detected around discr2.ads:6:1|

The testcase can be added to the testsuite as gnat.dg/specs/discr2.ads.

[Bug tree-optimization/54022] [4.8 regression] ICE in hoist_edge_and_branch_if_true

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54022

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |stevenb.gcc at gmail dot
                   |                            |com
   Target Milestone|---                         |4.8.0

[Bug tree-optimization/54022] [4.8 regression] ICE in hoist_edge_and_branch_if_true

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54022

--- Comment #1 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2012-07-19 10:21:20 UTC ---
Something peculiar here is that the default and another case jumps to the same
basic block:

  switch (D.4089_3) <default: L0, case 0 ... 7: <L3>, case 8: <L0>>

[...]

  # R86s_8 = PHI <0(2), 0(3), 0(6), 1(14)>
L0:
<L3>:
  D.4087_7 = D.4084_1 ^ 1;
  D.4086_9 = D.4087_7 & R86s_8;
  if (D.4086_9 != 0)
    goto <bb 8>;
  else
    goto <bb 9>;

[Bug tree-optimization/54022] [4.8 regression] ICE in hoist_edge_and_branch_if_true

[steven at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54022

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-07-19
                 CC|                            |steven at gcc dot gnu.org
     Ever Confirmed|0                           |1

--- Comment #2 from Steven Bosscher <steven at gcc dot gnu.org> 2012-07-19 11:04:32 UTC ---
(In reply to comment #1)
> Something peculiar here is that the default and another case jumps to the same
> basic block:

Just like in PR53887 so this is very peculiar -- I thought I had already fixed
this.  Will look into it tonight.

[Bug tree-optimization/54022] [4.8 regression] ICE in hoist_edge_and_branch_if_true

[steven at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54022

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

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

--- Comment #3 from Steven Bosscher <steven at gcc dot gnu.org> 2012-07-22 21:29:42 UTC ---
.

*** This bug has been marked as a duplicate of bug 53881 ***