[AArch64][GlobalISel] Walk through G_TRUNC in getTestBitReg

When you encounter a G_TRUNC, you are moving from a larger type to a smaller
type.

Asking for the i-th bit on a larger value is the same as asking for the i-th
bit on a smaller value.

So, we should always be able to walk through G_TRUNC when computing the bit
for a TB(N)Z.

Differential Revision: https://reviews.llvm.org/D73748

Author: Jessica Paquette

llvm/lib/Target/AArch64/AArch64InstructionSelector.cpp

@@ -998,7 +998,11 @@ static Register getTestBitReg(Register Reg, MachineRegisterInfo &MRI) {
     Register NextReg;
 
     // (tbz (any_ext x), b) -> (tbz x, b) if we don't use the extended bits.
-    if (Opc == TargetOpcode::G_ANYEXT || Opc == TargetOpcode::G_ZEXT)
+    //
+    // (tbz (trunc x), b) -> (tbz x, b) is always safe, because the bit number
+    // on the truncated x is the same as the bit number on x.
+    if (Opc == TargetOpcode::G_ANYEXT || Opc == TargetOpcode::G_ZEXT ||
+        Opc == TargetOpcode::G_TRUNC)
       NextReg = MI->getOperand(1).getReg();
 
     // Did we find something worth folding?

llvm/test/CodeGen/AArch64/GlobalISel/opt-fold-trunc-tbz-tbnz.mir

@@ -0,0 +1,35 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -mtriple aarch64-unknown-unknown -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s
+...
+---
+name:            fold_trunc
+alignment:       4
+legalized:       true
+regBankSelected: true
+tracksRegLiveness: true
+body:             |
+  ; CHECK-LABEL: name: fold_trunc
+  ; CHECK: bb.0:
+  ; CHECK:   successors: %bb.0(0x40000000), %bb.1(0x40000000)
+  ; CHECK:   liveins: $x0
+  ; CHECK:   %copy:gpr64all = COPY $x0
+  ; CHECK:   [[COPY:%[0-9]+]]:gpr32all = COPY %copy.sub_32
+  ; CHECK:   [[COPY1:%[0-9]+]]:gpr32 = COPY [[COPY]]
+  ; CHECK:   TBNZW [[COPY1]], 3, %bb.1
+  ; CHECK:   B %bb.0
+  ; CHECK: bb.1:
+  ; CHECK:   RET_ReallyLR
+  bb.0:
+    successors: %bb.0, %bb.1
+    liveins: $x0
+    %copy:gpr(s64) = COPY $x0
+    %bit:gpr(s32) = G_CONSTANT i32 8
+    %zero:gpr(s32) = G_CONSTANT i32 0
+    %fold_me:gpr(s32) = G_TRUNC %copy(s64)
+    %and:gpr(s32) = G_AND %fold_me, %bit
+    %cmp:gpr(s32) = G_ICMP intpred(ne), %and(s32), %zero
+    %cmp_trunc:gpr(s1) = G_TRUNC %cmp(s32)
+    G_BRCOND %cmp_trunc(s1), %bb.1
+    G_BR %bb.0
+  bb.1:
+    RET_ReallyLR