This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH][2/3][PR65458] Mark omp thread functions as parallelized
- From: Jakub Jelinek <jakub at redhat dot com>
- To: Tom de Vries <Tom_deVries at mentor dot com>
- Cc: Richard Biener <rguenther at suse dot de>, GCC Patches <gcc-patches at gcc dot gnu dot org>
- Date: Fri, 20 Mar 2015 13:31:55 +0100
- Subject: Re: [PATCH][2/3][PR65458] Mark omp thread functions as parallelized
- Authentication-results: sourceware.org; auth=none
- References: <5509B42F dot 1040707 at mentor dot com> <20150318172540 dot GO1746 at tucnak dot redhat dot com> <550AACA9 dot 60207 at mentor dot com> <20150319111127 dot GT1746 at tucnak dot redhat dot com> <550AB288 dot 8020107 at mentor dot com> <20150319112924 dot GV1746 at tucnak dot redhat dot com> <550C0667 dot 4030504 at mentor dot com> <20150320115754 dot GE1746 at tucnak dot redhat dot com> <550C12F8 dot 4000205 at mentor dot com>
- Reply-to: Jakub Jelinek <jakub at redhat dot com>
On Fri, Mar 20, 2015 at 01:30:48PM +0100, Tom de Vries wrote:
> On 20-03-15 12:57, Jakub Jelinek wrote:
> >>@@ -5569,6 +5569,7 @@ expand_omp_taskreg (struct omp_region *region)
> >>> /* Inform the callgraph about the new function. */
> >>> DECL_STRUCT_FUNCTION (child_fn)->curr_properties = cfun->curr_properties;
> >>> cgraph_node::add_new_function (child_fn, true);
> >>>+ cgraph_node::get (child_fn)->parallelized_function = 1;
> >IMHO you want to do this in create_omp_child_function instead,
>
> That way, the patch would no longer work for parloops. The current location
> of the fix is triggered by both parloops, and the omp-in-source processing.
>
> >that way you
> >handle it not just for the parallel region, but also e.g. for the task copy
> >functions etc.
>
> I propose to handle task copy like this:
> ...
> diff --git a/gcc/omp-low.c b/gcc/omp-low.c
> index 9be39b7..f8b5f85 100644
> --- a/gcc/omp-low.c
> +++ b/gcc/omp-low.c
> @@ -1564,6 +1564,7 @@ finalize_task_copyfn (gomp_task *task_stmt)
>
> /* Inform the callgraph about the new function. */
> cgraph_node::add_new_function (child_fn, false);
> + cgraph_node::get (child_fn)->parallelized_function = 1;
> }
>
> /* Destroy a omp_context data structures. Called through the splay tree
> ...
>
> OK if retesting succeeds?
Ok.
Jakub