[llvm] [AMDGPU] Allocate i1 argument to SGPRs (PR #72461)

[Matt Arsenault via llvm-commits]

================
@@ -715,6 +715,14 @@ bool SILowerI1Copies::lowerCopiesToI1() {
       assert(!MI.getOperand(1).getSubReg());
 
       if (!SrcReg.isVirtual() || (!isLaneMaskReg(SrcReg) && !isVreg1(SrcReg))) {
+        if (!SrcReg.isVirtual() &&
+            TII->getRegisterInfo().getRegSizeInBits(SrcReg, *MRI) == 64) {
+          // When calling convention allocates SGPR for i1, for GPUs with
+          // wavefront size 64, i1 return value is put in 64b SGPR.
+          assert(ST->isWave64());
+          continue;
+        }
+
----------------
arsenm wrote:

The virtual register copies are getting emitted backwards, yes.

You could hack around this in LowerFormalArguments by manually copying to a virtual register with the correct class, and then emitting a copy from that virtual register. 

https://github.com/llvm/llvm-project/pull/72461