This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug tree-optimization/46032] openmp inhibits loop vectorization
- From: "rguenther at suse dot de" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Wed, 27 May 2015 08:43:08 +0000
- Subject: [Bug tree-optimization/46032] openmp inhibits loop vectorization
- Auto-submitted: auto-generated
- References: <bug-46032-4 at http dot gcc dot gnu dot org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46032
--- Comment #19 from rguenther at suse dot de <rguenther at suse dot de> ---
On Wed, 27 May 2015, jakub at gcc dot gnu.org wrote:
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46032
>
> --- Comment #18 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
> The *.omp_fn.* functions indeed, while they necessarily have to be addressable,
> because that is how they are passed to the libgomp entrypoints, are never
> called by anything but the libgomp runtime. For GOMP_parallel*, they are only
> called before the GOMP_parallel* function exits, for GOMP_task* they could be
> called at some later point.
Ok, so this just means that IPA PTA would need to handle those specially
(and thus the OMP functions should be marked specially in the cgraph
node). Not that I think IPA PTA is anywhere near production ready
(or I have time to fix it up properly...). Just testing the addressable
fix now.