[PATCH] D105467: [AMDGPU] Mark waterfall loops as SI_WATERFALL_LOOP

[Jay Foad via Phabricator via llvm-commits]

foad added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/SILowerControlFlow.cpp:426
+
+  MachineInstr *Branch =
+      BuildMI(MBB, MI, DL, TII->get(AMDGPU::S_CBRANCH_EXECNZ))
----------------
Can you use MachineInstr::setDesc to mutate MI into an S_CBRANCH_EXECNZ, instead of creating a new instruction and erasing the old one?


================
Comment at: llvm/lib/Target/AMDGPU/SILowerI1Copies.cpp:174
             MI.getOpcode() == AMDGPU::SI_ELSE ||
+            MI.getOpcode() == AMDGPU::SI_WATERFALL_LOOP ||
             MI.getOpcode() == AMDGPU::SI_LOOP) {
----------------
Is this a change in behaviour? Previously we used an S_CBRANCH_EXECNZ which would not set Divergent here.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D105467/new/

https://reviews.llvm.org/D105467