[llvm-branch-commits] [llvm] [AMDGPU] Don't spill an SGPR while SCC is live in frame index lowering (PR #224080)

[Domenic Nutile via llvm-branch-commits]

================
@@ -166,3 +166,129 @@ body:             |
 
     S_ENDPGM 0, implicit $vcc
 ...
+
+---
+# SCC is live *into* the scalar frame index user: S_CMP_EQ_U32 produces it and
+# S_CSELECT_B32 reads it. With no SGPR free to hold the V_READFIRSTLANE_B32
----------------
saxlungs wrote:

Made a new pre-commit PR in the stack here: https://github.com/llvm/llvm-project/pull/225196

https://github.com/llvm/llvm-project/pull/224080