[patch,openacc] Generate sequential loop for OpenACC loop directive inside kernels

[Cesar Philippidis]

On 09/20/2018 10:14 AM, Cesar Philippidis wrote:
> As Chung-Lin noted here
> <https://gcc.gnu.org/ml/gcc-patches/2015-06/msg01079.html>:
> 
>   This patch adjusts omp-low.c:expand_omp_for_generic() to expand to a
>   "sequential" loop form (without the OMP runtime calls), used for loop
>   directives inside OpenACC kernels constructs. Tom mentions that this
>   allows the kernels parallelization to work when '#pragma acc loop'
>   makes the front-ends create OMP_FOR, which the loop analysis phases
>   don't understand.
> 
> I bootstrapped and regtested it on x86_64 Linux with nvptx offloading.
> Is this patch OK for trunk?

I forgot to mention how that patch depends on the
omp_target_base_pointers_restrict_p functionality from omp lowering that
I removed back in June when I added support for the OpenACC 2.5 data
clause semantics. It turned out that I was too aggressive when I was
removing unused code. That's because, at least initially, there was no
test cases that exercised that functionality in trunk until Chung-Lin's
kernels patch goes in.

Anyway, this patch is specifically required to get
kernels-acc-loop-reduction.c working.

Is this OK for trunk? I bootstrapped and regression tested it on x86_64
Linux with nvptx offloading.

Thanks,
Cesar
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Reintroduce-omp_target_base_pointers_restrict_p.patch
Type: text/x-patch
Size: 5204 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20180920/69e20cb4/attachment.bin>