[PATCH] D55568: AMDGPU: Don't peel of the offset if the resulting base could possibly be negative in Indirect addressing.

[Changpeng Fang via Phabricator via llvm-commits]

cfang created this revision.
cfang added reviewers: arsenm, msearles, rampitec.
Herald added subscribers: t-tye, tpr, dstuttard, yaxunl, nhaehnle, wdng, jvesely, kzhuravl.

Don't peel of the offset if the resulting base could possibly be negative in Indirect addressing. 
This is because the M0 field is of unsigned.

This patch achieves the similar goal as https://reviews.llvm.org/D55241, but keeps the optimization
if the base is known unsigned.


https://reviews.llvm.org/D55568

Files:
  lib/Target/AMDGPU/AMDGPUISelDAGToDAG.cpp
  test/CodeGen/AMDGPU/indirect-addressing-si-gfx9.ll
  test/CodeGen/AMDGPU/indirect-addressing-si-pregfx9.ll
  test/CodeGen/AMDGPU/indirect-addressing-si.ll

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D55568.177755.patch
Type: text/x-patch
Size: 11374 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20181211/68be64e9/attachment.bin>