[PATCH] D114652: [AMDGPU] Move call clobbered return address registers s[30:31] to callee saved range

[Venkata Ramanaiah Nalamothu via Phabricator via llvm-commits]

RamNalamothu marked 2 inline comments as done.
RamNalamothu added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SIFrameLowering.cpp:1351
+  // does not use CSR list, the clobbering of return address by function calls
+  // (D117243) or otherwise (D120922) is ignored/not seen by the IPRA's register
+  // usage collection. This will ensure save/restore of return address happens
----------------
arsenm wrote:
> Referring to phabricator reviews in a comment is weird
I have seen this in many places of clang/llvm and in fact in AMDGPU's source/tests as well. But if you say this is not a recommended practice, I will remove those references while landing the patch.


================
Comment at: llvm/lib/Target/AMDGPU/SIFrameLowering.cpp:1359
+  if (!MFI->isEntryFunction() &&
+      (FrameInfo.hasCalls() || MRI.isPhysRegModified(RAReg_sub0) ||
+       MRI.isPhysRegModified(RAReg_sub1))) {
----------------
arsenm wrote:
> Can just directly check isPhysRegModified(RetAddrReg)
Fixed. Thanks.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D114652/new/

https://reviews.llvm.org/D114652