[Bug translation/67892] New: Wrong code at -O1 and above

[Bug translation/67892] New: Wrong code at -O1 and above

[sikkins at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67892

            Bug ID: 67892
           Summary: Wrong code at -O1 and above
           Product: gcc
           Version: 5.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: translation
          Assignee: unassigned at gcc dot gnu.org
          Reporter: sikkins at hotmail dot com
  Target Milestone: ---

Created attachment 36461
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=36461&action=edit
test case

gcc-5 emits wrong code for the attached test case when called with option -O1
and above. However, adding option -fno-tree-dominator-opts fixes the issue.

gcc versions tested: 5.1.0 and 5.2.0
Targets: x86_64-linux-gnu / x86_64-w64-mingw32

[Bug translation/67892] [5/6 Regression] Wrong code at -O1 and above

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67892

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
      Known to work|                            |4.9.3
           Keywords|                            |wrong-code
   Last reconfirmed|                            |2015-10-08
                 CC|                            |law at gcc dot gnu.org
     Ever confirmed|0                           |1
            Summary|Wrong code at -O1 and above |[5/6 Regression] Wrong code
                   |                            |at -O1 and above
   Target Milestone|---                         |5.3

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed.  Sounds like a FSM threading bug, confirmed on trunk as well.

[Bug translation/67892] [5/6 Regression] Wrong code at -O1 and above

[law at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67892

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |law at redhat dot com

--- Comment #2 from Jeffrey A. Law <law at redhat dot com> ---
Not the FSM bits, but the slightly older bits in the traditional threader which
allow threading after traversing backedges.  I think I see what's happening,
but I need to dig a bit more to be sure I've got the real problem.

[Bug tree-optimization/67892] [5/6 Regression] Wrong code at -O1 and above

[law at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67892

--- Comment #3 from Jeffrey A. Law <law at gcc dot gnu.org> ---
Author: law
Date: Thu Oct 29 16:20:06 2015
New Revision: 229538

URL: https://gcc.gnu.org/viewcvs?rev=229538&root=gcc&view=rev
Log:
[PATCH][PR tree-optimization/67892] Use FSM threader to handle backedges

        PR tree-optimization/67892
        * tree-ssa-threadedge.c (simplify_controL_stmt_condition): Fix typo
        in comment.
        (thread_through_normal_block): If we have seen a backedge, then
        do nothing.  No longer call find_jump_threads_backwards here.
        (thread_across_edge): Use find_jump_threads_backwards to find
        jump threads if the old style threader was not successful.
        * tree-ssa-threadbackward.c (get_gimple_control_stmt): Use
        gsi_last_nondebug_bb.  Return NULL if the block does not end
        with a control statement.
        (find_jump_threads_backwards): Setup code moved here from
        tree-ssa-threadedge.c::thread_through_normal_block.  Accept
        single edge argument instead of name & block.
        * tree-ssa-threadbackward.h (find_jump_threads_backwards): Update
        prototype.

        PR tree-optimization/67892
        * gcc.dg/tree-ssa/pr21417: Update expected output.
        * gcc.dg/tree-ssa/ssa-dom-thread-2b.c: Likewise.

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gcc.dg/tree-ssa/pr21417.c
    trunk/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-2b.c
    trunk/gcc/tree-ssa-threadbackward.c
    trunk/gcc/tree-ssa-threadbackward.h
    trunk/gcc/tree-ssa-threadedge.c

[Bug tree-optimization/67892] [5/6 Regression] Wrong code at -O1 and above

[law at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67892

Jeffrey A. Law <law at redhat dot com> changed:

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

--- Comment #4 from Jeffrey A. Law <law at redhat dot com> ---
Fixed on trunk.