[llvm-branch-commits] [llvm] AMDGPU/GlobalISel: add RegBankLegalize rules for extends and trunc (PR #132383)

[Petar Avramovic via llvm-branch-commits]

================
@@ -292,13 +311,23 @@ void RegBankLegalizeHelper::lower(MachineInstr &MI,
   case Ext32To64: {
     const RegisterBank *RB = MRI.getRegBank(MI.getOperand(0).getReg());
     MachineInstrBuilder Hi;
-
-    if (MI.getOpcode() == AMDGPU::G_ZEXT) {
+    switch (MI.getOpcode()) {
+    case AMDGPU::G_ZEXT: {
       Hi = B.buildConstant({RB, S32}, 0);
-    } else {
+      break;
+    }
+    case AMDGPU::G_SEXT: {
       // Replicate sign bit from 32-bit extended part.
       auto ShiftAmt = B.buildConstant({RB, S32}, 31);
       Hi = B.buildAShr({RB, S32}, MI.getOperand(1).getReg(), ShiftAmt);
+      break;
+    }
+    case AMDGPU::G_ANYEXT: {
+      Hi = B.buildUndef({RB, S32});
----------------
petar-avramovic wrote:

this one https://github.com/llvm/llvm-project/pull/127825 ?

https://github.com/llvm/llvm-project/pull/132383