[Bug tree-optimization/81529] [8 Regression] ICE in vectorizable_induction, at tree-vect-loop.c:6613

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81529

--- Comment #4 from Richard Biener  ---
Author: rguenth
Date: Tue Jul 25 13:22:31 2017
New Revision: 250517

URL: https://gcc.gnu.org/viewcvs?rev=250517=gcc=rev
Log:
2017-07-25  Richard Biener  

PR tree-optimization/81529
* tree-vect-stmts.c (process_use): Disregard live induction PHIs
when optimizing backedge uses.

* gfortran.dg/pr81529.f90: New testcase.

Added:
trunk/gcc/testsuite/gfortran.dg/pr81529.f90
Modified:
trunk/gcc/ChangeLog
trunk/gcc/testsuite/ChangeLog
trunk/gcc/tree-vect-stmts.c

[Bug tree-optimization/81529] [8 Regression] ICE in vectorizable_induction, at tree-vect-loop.c:6613

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81529

Richard Biener  changed:

   What|Removed |Added

 Status|ASSIGNED|RESOLVED
 Resolution|--- |FIXED

--- Comment #3 from Richard Biener  ---
Fixed.

[Bug tree-optimization/81529] [8 Regression] ICE in vectorizable_induction, at tree-vect-loop.c:6613

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81529

--- Comment #2 from Richard Biener  ---
Ok, so we have an induction that's only used outside of the loop (well,
-fno-tree-scev-cprop ...) and this causes the VF to be reset to 1 as the
induction
computation is vect_unused_in_scope (should be used_by_induction I guess).
Caused by another change of mine I think.

Testing patch.

[Bug tree-optimization/81529] [8 Regression] ICE in vectorizable_induction, at tree-vect-loop.c:6613

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81529

Richard Biener  changed:

   What|Removed |Added

 Status|UNCONFIRMED |ASSIGNED
   Last reconfirmed||2017-07-25
Version|7.0 |8.0
   Assignee|unassigned at gcc dot gnu.org  |rguenth at gcc dot 
gnu.org
   Target Milestone|--- |8.0
 Ever confirmed|0   |1

--- Comment #1 from Richard Biener  ---
Mine.