[PATCH] D105192: [AMDGPU] Optimize VGPR LiveRange in waterfall loops

[Ruiling, Song via Phabricator via llvm-commits]

ruiling added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SIOptimizeVGPRLiveRange.cpp:490
+    // Replace uses in Loop block
+    if (UseBlock == Loop && UseMI != PHI.getInstr())
+      O.setReg(NewReg);
----------------
nit: Can we do register replacement before inserting the phi instruction? then we don't need the check `UseMI != PHI.getInstr()`.


================
Comment at: llvm/lib/Target/AMDGPU/SIOptimizeVGPRLiveRange.cpp:494
+
+  LiveVariables::VarInfo &OldVarInfo = LV->getVarInfo(Reg);
+
----------------
ruiling wrote:
> I still think we need to update the LiveVariables::VarInfo for the `NewReg` to make sure LiveVariables analysis being updated. We just need to update the `Kills` to the last use in the waterfall loop. as the waterfall loop is pretty small, I think we can just reverse iterate to find the last use. I am not sure if there is any better way?
You also need to call `addRegisterKilled` to mark the last use/uses of NewReg as killed.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D105192/new/

https://reviews.llvm.org/D105192