[llvm-branch-commits] [llvm] [InlineSpiller][AMDGPU] Implement subreg reload during RA spill (PR #175002)

[Matt Arsenault via llvm-branch-commits]

================
@@ -3915,6 +3924,11 @@ SIRegisterInfo::getRegClassForSizeOnBank(unsigned Size,
   }
 }
 
+const TargetRegisterClass *
+SIRegisterInfo::getConstrainedRegClass(const TargetRegisterClass *RC) const {
+  return getProperlyAlignedRC(RC);
+}
----------------
arsenm wrote:

Remove this. This is spreading an AMDGPU hack 

https://github.com/llvm/llvm-project/pull/175002