[Bug tree-optimization/86927] [8/9 Regression] Gcc miscompiles at -O3 on valid code

2018-08-27 Thread rguenth at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86927

--- Comment #6 from Richard Biener  ---
Author: rguenth
Date: Mon Aug 27 14:17:03 2018
New Revision: 263888

URL: https://gcc.gnu.org/viewcvs?rev=263888&root=gcc&view=rev
Log:
2018-08-27  Richard Biener  

PR tree-optimization/86927
* tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
use const cond reduction code.

* gcc.dg/vect/pr86927.c: New testcase.

Added:
trunk/gcc/testsuite/gcc.dg/vect/pr86927.c
Modified:
trunk/gcc/ChangeLog
trunk/gcc/testsuite/ChangeLog
trunk/gcc/tree-vect-loop.c

[Bug tree-optimization/86927] [8/9 Regression] Gcc miscompiles at -O3 on valid code

2018-08-27 Thread rguenth at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86927

--- Comment #5 from Richard Biener  ---
Created attachment 44607
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=44607&action=edit
patch

I am testing the attached.

[Bug tree-optimization/86927] [8/9 Regression] Gcc miscompiles at -O3 on valid code

2018-08-17 Thread rguenth at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86927

Richard Biener  changed:

   What|Removed |Added

   Priority|P3  |P2
 Status|NEW |ASSIGNED
   Assignee|unassigned at gcc dot gnu.org  |rguenth at gcc dot 
gnu.org

--- Comment #4 from Richard Biener  ---
Mine then.

[Bug tree-optimization/86927] [8/9 Regression] Gcc miscompiles at -O3 on valid code

2018-08-13 Thread marxin at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86927

--- Comment #3 from Martin Liška  ---
Then it ends in Richi's revision r249553. Sounds like culprit.

[Bug tree-optimization/86927] [8/9 Regression] Gcc miscompiles at -O3 on valid code

2018-08-13 Thread amonakov at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86927

Alexander Monakov  changed:

   What|Removed |Added

 CC||amonakov at gcc dot gnu.org

--- Comment #2 from Alexander Monakov  ---
Martin, can you try bisecting with -fvect-cost-model=unlimited ?

May be caused by vectorization, reduction created for 'c' looks wrong. gcc-7
did not try to vectorize this.

[Bug tree-optimization/86927] [8/9 Regression] Gcc miscompiles at -O3 on valid code

2018-08-13 Thread marxin at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86927

Martin Liška  changed:

   What|Removed |Added

 Status|UNCONFIRMED |NEW
  Known to work||7.3.0
Version|unknown |8.2.0
   Keywords||wrong-code
   Last reconfirmed||2018-08-13
 CC||hubicka at gcc dot gnu.org,
   ||marxin at gcc dot gnu.org,
   ||rguenth at gcc dot gnu.org
 Ever confirmed|0   |1
Summary|Gcc miscompiles at -O3 on   |[8/9 Regression] Gcc
   |valid code  |miscompiles at -O3 on valid
   ||code
   Target Milestone|--- |8.3
  Known to fail||8.2.0, 9.0

--- Comment #1 from Martin Liška  ---
I see it starting from r255268, but it's probably a latent issue.