[PATCH] D106498: AMDGPU: Treat IMPLICIT_DEF like a constant lanemask source

[Matt Arsenault via Phabricator via llvm-commits]

arsenm created this revision.
arsenm added reviewers: sameerds, nhaehnle, foad, ruiling.
Herald added subscribers: kerbowa, hiraditya, t-tye, tpr, dstuttard, yaxunl, jvesely, kzhuravl.
arsenm requested review of this revision.
Herald added a subscriber: wdng.
Herald added a project: LLVM.

This is partially a workaround. SILowerI1Copies does not understand
unstructured loops. This would result in inserting instructions to
merge a mask register in the same block where it was defined in an
unstructured loop. This is also a small code improvement.


https://reviews.llvm.org/D106498

Files:
  llvm/lib/Target/AMDGPU/SILowerI1Copies.cpp
  llvm/test/CodeGen/AMDGPU/loop_break.ll
  llvm/test/CodeGen/AMDGPU/lower-i1-copies-implicit-def-unstructured-loop.mir
  llvm/test/CodeGen/AMDGPU/sgpr-control-flow.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D106498.360601.patch
Type: text/x-patch
Size: 13591 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20210721/3fbe4ae0/attachment-0001.bin>