[all-commits] [llvm/llvm-project] 2695e2: [OpenMP][CUDA] Fix the issue that P2P memcpy doesn...

[Shilei Tian via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 2695e23ad96ba2e6417cae89eb51f2b6ba5b7666
      https://github.com/llvm/llvm-project/commit/2695e23ad96ba2e6417cae89eb51f2b6ba5b7666
  Author: Shilei Tian <i at tianshilei.me>
  Date:   2022-06-28 (Tue, 28 Jun 2022)

  Changed paths:
    M openmp/libomptarget/plugins/cuda/src/rtl.cpp

  Log Message:
  -----------
  [OpenMP][CUDA] Fix the issue that P2P memcpy doesn't work

This patch fixes the issue that P2P memcpy doesn't work. The root cause is we didn't set current context when calling the API function. In addition, a matrix to track the states of each pair of devices is also added such that we only need to query and configure the device once.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D122764