[PATCH] D36831: [AMDGPU] Transform __read_pipe_* and __write_pipe_*

[Valery Pykhtin via Phabricator via llvm-commits]

vpykhtin added a comment.

Splitting AMDGPULibFunc in two classes looks a huge overkill. How about modifying AMDGPULibFunc::parse so it could accept unmangled names and just return an enum id for the function (using some fast lookup approach)? Type info for such functions can be left unpopulated and supposed to be handled by the client (as in fold_read_write_pipe).


https://reviews.llvm.org/D36831