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

Matt Arsenault via llvm-commits llvm-commits at lists.llvm.org
Mon Dec 4 18:18:10 PST 2023


================
@@ -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:

But contextually these shouldn't be getting processed. What is the MIR at the point of the assert?

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


More information about the llvm-commits mailing list