[Openmp-commits] [PATCH] D87259: [OpenMPOpt][SplitMemTransfer] Implementation of the real issue and wait api functions.

Joseph Huber via Phabricator via Openmp-commits openmp-commits at lists.llvm.org
Tue Sep 29 09:11:25 PDT 2020

jhuber6 added a comment.

In D87259#2301065 <https://reviews.llvm.org/D87259#2301065>, @grokos wrote:

> I think these API functions should also include the source location pointer from https://reviews.llvm.org/D87946. We need to consider renaming the `*_issue` and `*_wait` functions to extend the `*_loc` API the aforementioned patch is introducing. E.g. after D87946 <https://reviews.llvm.org/D87946> the "current" `data begin` API function will be `__tgt_target_data_begin_mapper_loc`, so this patch should extend that name as `__tgt_target_data_begin_mapper_loc_issue` and `__tgt_target_data_begin_mapper_loc_wait`. Because both patches make changes to the API, I think it's better to wait until the former patch has been committed.

We've also been talking about extending the API to optionally include declaration names in the mappers so we can refer to mapped variables by name in error and information messages. If we're breaking the API it's probably best to do it all at once so I'm not sure how soon we'll be able to push D87946 <https://reviews.llvm.org/D87946>.

  rG LLVM Github Monorepo



More information about the Openmp-commits mailing list