[AMDGPU] Combine and x, (sext cc from i1) => select cc, x, 0

[rampitec]

Summary

Also factored out function to check if a boolean is an already
deserialized value which does not require v_cndmask_b32 to be
loaded. Added binary logical operators to its check.

[rampitec]

Fixed operand order

[rampitec]

Renamed function.

[vpykhtin]

LGTM.

lib/Target/AMDGPU/SIISelLowering.cpp

const SIInstrInfo *TII = getSubtarget()->getInstrInfo();
     // anyway. Avoid creating the harder to understand 64-bit immediate
     // materialization.
     return splitBinaryBitConstantOpImpl(DCI, SL, Opc, LHS, ValLo, ValHi);
   }
 
   return SDValue();
 }
 
+// Returns true if argument is a boolean value which is not serialized into
+// memory or argument and does not require v_cmdmask_b32 to be deserialized.
+static bool isBoolSGPR(SDValue V) {
+  if (V.getValueType() != MVT::i1)
+    return false;
+  switch (V.getOpcode()) {
+  default: break;
+  case ISD::SETCC:
+  case ISD::AND:
+  case ISD::OR:
+  case ISD::XOR:
+  case AMDGPUISD::FP_CLASS:
+    return true;
+  }
+  return false;
+}
+
 SDValue SITargetLowering::performAndCombine(SDNode *N,
                                             DAGCombinerInfo &DCI) const {
   if (DCI.isBeforeLegalize())
     return SDValue();
 
   SelectionDAG &DAG = DCI.DAG;
   EVT VT = N->getValueType(0);
   SDValue LHS = N->getOperand(0);
if (LCC == ISD::SETO) {
 
         SDLoc DL(N);
         return DAG.getNode(AMDGPUISD::FP_CLASS, DL, MVT::i1,
                            X, DAG.getConstant(Mask, DL, MVT::i32));
       }
     }
   }
 
+  if (VT == MVT::i32 &&
+      (RHS.getOpcode() == ISD::SIGN_EXTEND || LHS.getOpcode() == ISD::SIGN_EXTEND)) {
+    // and x, (sext cc from i1) => select cc, x, 0
+    if (RHS.getOpcode() != ISD::SIGN_EXTEND)
+      std::swap(LHS, RHS);
+    if (isBoolSGPR(RHS.getOperand(0)))
+      return DAG.getSelect(SDLoc(N), MVT::i32, RHS.getOperand(0),
+                           LHS, DAG.getConstant(0, SDLoc(N), MVT::i32));
+  }
+
   return SDValue();
 }
 
 SDValue SITargetLowering::performOrCombine(SDNode *N,
                                            DAGCombinerInfo &DCI) const {
   SelectionDAG &DAG = DCI.DAG;
   SDValue LHS = N->getOperand(0);
   SDValue RHS = N->getOperand(1);
SDValue SITargetLowering::performAddCombine(SDNode *N,
 
   Opc = RHS.getOpcode();
   switch (Opc) {
   default: break;
   case ISD::ZERO_EXTEND:
   case ISD::SIGN_EXTEND:
   case ISD::ANY_EXTEND: {
     auto Cond = RHS.getOperand(0);
-    if (Cond.getOpcode() != ISD::SETCC &&
-        Cond.getOpcode() != AMDGPUISD::FP_CLASS)
+    if (!isBoolSGPR(Cond))
       break;
     SDVTList VTList = DAG.getVTList(MVT::i32, MVT::i1);
     SDValue Args[] = { LHS, DAG.getConstant(0, SL, MVT::i32), Cond };
     Opc = (Opc == ISD::SIGN_EXTEND) ? ISD::SUBCARRY : ISD::ADDCARRY;
     return DAG.getNode(Opc, SL, VTList, Args);
   }
   case ISD::ADDCARRY: {
     // add x, (addcarry y, 0, cc) => addcarry x, y, cc

test/CodeGen/AMDGPU/combine-and-sext-bool.ll

+; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+
+; GCN-LABEL: {{^}}and_i1_sext_bool:
+; GCN: v_cmp_{{gt|le}}_u32_e{{32|64}} [[CC:[^,]+]], v{{[0-9]+}}, v{{[0-9]+}}
+; GCN: v_cndmask_b32_e{{32|64}} [[VAL:v[0-9]+]], 0, v{{[0-9]+}}, [[CC]]
+; GCN: store_dword {{.*}}[[VAL]]
+; GCN-NOT: v_cndmask_b32_e64 v{{[0-9]+}}, {{0|-1}}, {{0|-1}}
+; GCN-NOT: v_and_b32_e32
+
+define amdgpu_kernel void @and_i1_sext_bool(i32 addrspace(1)* nocapture %arg) {
+bb:
+  %x = tail call i32 @llvm.amdgcn.workitem.id.x()
+  %y = tail call i32 @llvm.amdgcn.workitem.id.y()
+  %gep = getelementptr inbounds i32, i32 addrspace(1)* %arg, i32 %x
+  %v = load i32, i32 addrspace(1)* %gep, align 4
+  %cmp = icmp ugt i32 %x, %y
+  %ext = sext i1 %cmp to i32
+  %and = and i32 %v, %ext
+  store i32 %and, i32 addrspace(1)* %gep, align 4
+  ret void
+}
+
+declare i32 @llvm.amdgcn.workitem.id.x() #0
+
+declare i32 @llvm.amdgcn.workitem.id.y() #0
+
+attributes #0 = { nounwind readnone speculatable }

test/CodeGen/AMDGPU/combine-cond-add-sub.ll

bb:
   %v = load i32, i32 addrspace(1)* %gep, align 4
   %cmp = tail call zeroext i1 @llvm.amdgcn.class.f32(float %x, i32 608)
   %ext = sext i1 %cmp to i32
   %add = add i32 %v, %ext
   store i32 %add, i32 addrspace(1)* %gep, align 4
   ret void
 }
 
+; GCN-LABEL: {{^}}add_and:
+; GCN: s_and_b64 [[CC:[^,]+]],
+; GCN: v_addc_u32_e{{32|64}} v{{[0-9]+}}, {{[^,]+}}, 0, v{{[0-9]+}}, [[CC]]
+; GCN-NOT: v_cndmask
+
+define amdgpu_kernel void @add_and(i32 addrspace(1)* nocapture %arg) {
+bb:
+  %x = tail call i32 @llvm.amdgcn.workitem.id.x()
+  %y = tail call i32 @llvm.amdgcn.workitem.id.y()
+  %gep = getelementptr inbounds i32, i32 addrspace(1)* %arg, i32 %x
+  %v = load i32, i32 addrspace(1)* %gep, align 4
+  %cmp1 = icmp ugt i32 %x, %y
+  %cmp2 = icmp ugt i32 %x, 1
+  %cmp = and i1 %cmp1, %cmp2
+  %ext = zext i1 %cmp to i32
+  %add = add i32 %v, %ext
+  store i32 %add, i32 addrspace(1)* %gep, align 4
+  ret void
+}
+
 declare i1 @llvm.amdgcn.class.f32(float, i32) #0
 
 declare i32 @llvm.amdgcn.workitem.id.x() #0
 
 declare i32 @llvm.amdgcn.workitem.id.y() #0
 
 attributes #0 = { nounwind readnone speculatable }