[Openmp-commits] [PATCH] D142512: [OpenMP][libomptarget] Fix mapping/prelock.cpp test

[Carlo Bertolli via Phabricator via Openmp-commits]

carlo.bertolli added a comment.

This test works correctly with the old plugin. A locked pointer is not a device pointer. It is a host pointer that has been pinned in memory and that has been made accessible by ROCr to the device agent that will be assigned the data transfers implementing the map clause. In abstract programming terms, the second feature is only an AMD one and it does not need to have a corresponding functionality on other targets.

I am guessing this is a problem only in the next gen plugin, which needs fixing. Leads for potential issues:

- Is the relevant device agent passed in to the hsa_amd_memory_async_copy?
- Is the device agent passed in to the hsa_amd_memory_lock function?

You could also have a look at my patch for the old plugin as another potential lead:
https://reviews.llvm.org/D139208


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D142512/new/

https://reviews.llvm.org/D142512