GlobalISel: support same-size casts on AArch64.

Mostly Ahmed's work again, I'm just sprucing things up slightly before
committing.

llvm-svn: 283952

llvm/lib/Target/AArch64/AArch64InstructionSelector.cpp

@@ -41,6 +41,32 @@ AArch64InstructionSelector::AArch64InstructionSelector(
   : InstructionSelector(), TM(TM), STI(STI), TII(*STI.getInstrInfo()),
       TRI(*STI.getRegisterInfo()), RBI(RBI) {}
 
+// FIXME: This should be target-independent, inferred from the types declared
+// for each class in the bank.
+static const TargetRegisterClass *
+getRegClassForTypeOnBank(LLT Ty, const RegisterBank &RB,
+                         const RegisterBankInfo &RBI) {
+  if (RB.getID() == AArch64::GPRRegBankID) {
+    if (Ty.getSizeInBits() <= 32)
+      return &AArch64::GPR32RegClass;
+    if (Ty.getSizeInBits() == 64)
+      return &AArch64::GPR64RegClass;
+    return nullptr;
+  }
+
+  if (RB.getID() == AArch64::FPRRegBankID) {
+    if (Ty.getSizeInBits() == 32)
+      return &AArch64::FPR32RegClass;
+    if (Ty.getSizeInBits() == 64)
+      return &AArch64::FPR64RegClass;
+    if (Ty.getSizeInBits() == 128)
+      return &AArch64::FPR128RegClass;
+    return nullptr;
+  }
+
+  return nullptr;
+}
+
 /// Check whether \p I is a currently unsupported binary operation:
 /// - it has an unsized type
 /// - an operand is not a vreg
@@ -493,6 +519,42 @@ bool AArch64InstructionSelector::select(MachineInstr &I) const {
     I.eraseFromParent();
     return true;
   }
+
+  case TargetOpcode::G_INTTOPTR:
+  case TargetOpcode::G_PTRTOINT:
+  case TargetOpcode::G_BITCAST: {
+    const LLT DstTy = MRI.getType(I.getOperand(0).getReg());
+    const LLT SrcTy = MRI.getType(I.getOperand(1).getReg());
+
+    const unsigned DstReg = I.getOperand(0).getReg();
+    const unsigned SrcReg = I.getOperand(1).getReg();
+
+    const RegisterBank &DstRB = *RBI.getRegBank(DstReg, MRI, TRI);
+    const RegisterBank &SrcRB = *RBI.getRegBank(SrcReg, MRI, TRI);
+
+    const TargetRegisterClass *DstRC =
+        getRegClassForTypeOnBank(DstTy, DstRB, RBI);
+    if (!DstRC)
+      return false;
+
+    const TargetRegisterClass *SrcRC =
+        getRegClassForTypeOnBank(SrcTy, SrcRB, RBI);
+    if (!SrcRC)
+      return false;
+
+    if (!RBI.constrainGenericRegister(SrcReg, *SrcRC, MRI) ||
+        !RBI.constrainGenericRegister(DstReg, *DstRC, MRI)) {
+      DEBUG(dbgs() << "Failed to constrain G_BITCAST\n");
+      return false;
+    }
+
+    BuildMI(MBB, I, I.getDebugLoc(), TII.get(AArch64::COPY))
+        .addDef(DstReg)
+        .addUse(SrcReg);
+
+    I.eraseFromParent();
+    return true;
+  }
   }
 
   return false;

llvm/lib/Target/AArch64/AArch64MachineLegalizer.cpp

@@ -176,5 +176,18 @@ AArch64MachineLegalizer::AArch64MachineLegalizer() {
   setAction({G_INTTOPTR, 0, p0}, Legal);
   setAction({G_INTTOPTR, 1, s64}, Legal);
 
+  for (auto Ty : {s1, s8, s16, s32, s64}) {
+    setAction({G_BITCAST, 0, Ty}, Legal);
+    setAction({G_BITCAST, 1, Ty}, Legal);
+  }
+
+  for (int EltSize = 8; EltSize <= 64; EltSize *= 2) {
+    setAction({G_BITCAST, 0, LLT::vector(128/EltSize, EltSize)}, Legal);
+    if (EltSize == 64)
+      continue;
+
+    setAction({G_BITCAST, 0, LLT::vector(64/EltSize, EltSize)}, Legal);
+  }
+
   computeTables();
 }

llvm/test/CodeGen/AArch64/GlobalISel/arm64-instructionselect.mir

@@ -86,6 +86,8 @@
   define void @anyext_gpr() { ret void }
   define void @zext_gpr() { ret void }
   define void @sext_gpr() { ret void }
+
+  define void @casts() { ret void }
 ...
 
 ---
@@ -1423,3 +1425,32 @@ body:             |
     %2(s8) = COPY %w0
     %3(s32) = G_SEXT %2
 ...
+
+---
+# CHECK-LABEL: name: casts
+name:            casts
+legalized:       true
+regBankSelected: true
+
+# CHECK:      registers:
+# CHECK-NEXT:  - { id: 0, class: gpr64 }
+# CHECK-NEXT:  - { id: 1, class: fpr64 }
+registers:
+  - { id: 0, class: gpr }
+  - { id: 1, class: fpr }
+  - { id: 2, class: gpr }
+  - { id: 3, class: gpr }
+
+# CHECK:  body:
+# CHECK:    %0 = COPY %x0
+# CHECK:    %1 = COPY %0
+# CHECK:    %2 = COPY %0
+# CHECK:    %3 = COPY %2
+body:             |
+  bb.0:
+    liveins: %w0
+    %0(s64) = COPY %x0
+    %1(<8 x s8>) = G_BITCAST %0(s64)
+    %2(p0) = G_INTTOPTR %0
+    %3(s64) = G_PTRTOINT %2
+...