Page MenuHomePhabricator

[CUDA] Fix order of memcpy arguments in __shfl_*(<64-bit type>).
ClosedPublic

Authored by tra on Jan 22 2020, 3:14 PM.

Details

Reviewers
jlebar
timshen
Summary

Wrong argument order resulted in broken shfl ops for 64-bit types.

Event Timeline

tra created this revision.Jan 22 2020, 3:14 PM
Herald added a project: Restricted Project. · View Herald TranscriptJan 22 2020, 3:14 PM
timshen accepted this revision.Jan 23 2020, 12:44 PM

What's the test situation for these headers?

This revision is now accepted and ready to land.Jan 23 2020, 12:44 PM
tra added a comment.Jan 23 2020, 1:25 PM

What's the test situation for these headers?

They are not testable in clang/llvm tree as they need CUDA SDK headers. I should be able to add some tests in the test-suite.

hans added a comment.Jan 24 2020, 10:29 AM
In D73237#1837077, @tra wrote:

Yes, go ahead and "git cherry-pick -x" it and push to the branch.