https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68915
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |ASSIGNED
Last reconfirmed| |2015-12-16
CC| |vries at gcc dot gnu.org
Assignee|unassigned at gcc dot gnu.org |rguenth at gcc dot
gnu.org
Ever confirmed|0 |1
--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed from the dump.
Rainer, the dump says
/vol/gcc/src/hg/trunk/local/gcc/testsuite/gcc.dg/vect/pr46032.c:40:33: note:
not vectorized: no vectype for stmt: # VUSE <.MEM_4>
_19 = *pretmp_30[idx_3];
scalar_type: unsigned intD.4
/vol/gcc/src/hg/trunk/local/gcc/testsuite/gcc.dg/vect/pr46032.c:40:33: note:
bad data references.
but that's "impossible" from
/* { dg-require-effective-target vect_int } */
which either isn't in effect for some weird reason (dg-options overriding sth?)
or is confused by optimize attribute usage.
I suppose changing dg-options to dg-additional-options "-fopenmp -fipa-pta"
might work, will commit that.
That would have been the correct thing to do when moving the testcases to
gcc.dg/vect/ with
2015-12-01 Tom de Vries <t...@codesourcery.com>
* gcc.dg/pr46032.c: Move to ...
* gcc.dg/vect/pr46032.c: here. Add dg-require-effective-target
vect_int.
* gcc.dg/pr46032-2.c: Move to ...
* gcc.dg/gomp/pr46032-2.c: ... here.
Drop dg-require-effective-target fopenmp.
* gcc.dg/pr46032-3.c: Move to ...
* gcc.dg/gomp/pr46032-3.c: ... here.
Drop dg-require-effective-target fopenmp.
