GlobalISel: Translate llvm.fshl/llvm.fshr

Author: arsenm

llvm/include/llvm/Support/TargetOpcodes.def

@@ -385,6 +385,12 @@ HANDLE_TARGET_OPCODE(G_LSHR)
 // Generic arithmetic right-shift
 HANDLE_TARGET_OPCODE(G_ASHR)
 
+// Generic funnel left shift
+HANDLE_TARGET_OPCODE(G_FSHL)
+
+// Generic funnel right shift
+HANDLE_TARGET_OPCODE(G_FSHR)
+
 /// Generic integer-base comparison, also applicable to vectors of integers.
 HANDLE_TARGET_OPCODE(G_ICMP)
 

llvm/include/llvm/Target/GenericOpcodes.td

@@ -308,6 +308,22 @@ def G_ASHR : GenericInstruction {
   let hasSideEffects = 0;
 }
 
+/// Funnel 'double' shifts take 3 operands, 2 inputs and the shift amount.
+/// fshl(X,Y,Z): (X << (Z % bitwidth)) | (Y >> (bitwidth - (Z % bitwidth)))
+def G_FSHL : GenericInstruction {
+  let OutOperandList = (outs type0:$dst);
+  let InOperandList = (ins type0:$src1, type0:$src2, type1:$src3);
+  let hasSideEffects = 0;
+}
+
+/// Funnel 'double' shifts take 3 operands, 2 inputs and the shift amount.
+/// fshr(X,Y,Z): (X << (bitwidth - (Z % bitwidth))) | (Y >> (Z % bitwidth))
+def G_FSHR : GenericInstruction {
+  let OutOperandList = (outs type0:$dst);
+  let InOperandList = (ins type0:$src1, type0:$src2, type1:$src3);
+  let hasSideEffects = 0;
+}
+
 // Generic integer comparison.
 def G_ICMP : GenericInstruction {
   let OutOperandList = (outs type0:$dst);

llvm/include/llvm/Target/GlobalISel/SelectionDAGCompat.td

@@ -93,6 +93,8 @@ def : GINodeEquiv<G_INTRINSIC_W_SIDE_EFFECTS, intrinsic_w_chain>;
 def : GINodeEquiv<G_BR, br>;
 def : GINodeEquiv<G_BSWAP, bswap>;
 def : GINodeEquiv<G_BITREVERSE, bitreverse>;
+def : GINodeEquiv<G_FSHL, fshl>;
+def : GINodeEquiv<G_FSHR, fshr>;
 def : GINodeEquiv<G_CTLZ, ctlz>;
 def : GINodeEquiv<G_CTTZ, cttz>;
 def : GINodeEquiv<G_CTLZ_ZERO_UNDEF, ctlz_zero_undef>;

llvm/lib/CodeGen/GlobalISel/IRTranslator.cpp

@@ -1220,8 +1220,12 @@ unsigned IRTranslator::getSimpleIntrinsicOpcode(Intrinsic::ID ID) {
       break;
     case Intrinsic::bswap:
       return TargetOpcode::G_BSWAP;
-  case Intrinsic::bitreverse:
+    case Intrinsic::bitreverse:
       return TargetOpcode::G_BITREVERSE;
+    case Intrinsic::fshl:
+      return TargetOpcode::G_FSHL;
+    case Intrinsic::fshr:
+      return TargetOpcode::G_FSHR;
     case Intrinsic::ceil:
       return TargetOpcode::G_FCEIL;
     case Intrinsic::cos:

llvm/test/CodeGen/AArch64/GlobalISel/arm64-irtranslator.ll

@@ -1396,6 +1396,30 @@ define i32 @test_bitreverse_intrinsic(i32 %a) {
   ret i32 %res
 }
 
+declare i32 @llvm.fshl.i32(i32, i32, i32)
+define i32 @test_fshl_intrinsic(i32 %a, i32 %b, i32 %c) {
+; CHECK-LABEL: name: test_fshl_intrinsic
+; CHECK: [[A:%[0-9]+]]:_(s32) = COPY $w0
+; CHECK: [[B:%[0-9]+]]:_(s32) = COPY $w1
+; CHECK: [[C:%[0-9]+]]:_(s32) = COPY $w2
+; CHECK: [[RES:%[0-9]+]]:_(s32) = G_FSHL [[A]], [[B]], [[C]]
+; CHECK: $w0 = COPY [[RES]]
+  %res = call i32 @llvm.fshl.i32(i32 %a, i32 %b, i32 %c)
+  ret i32 %res
+}
+
+declare i32 @llvm.fshr.i32(i32, i32, i32)
+define i32 @test_fshr_intrinsic(i32 %a, i32 %b, i32 %c) {
+; CHECK-LABEL: name: test_fshr_intrinsic
+; CHECK: [[A:%[0-9]+]]:_(s32) = COPY $w0
+; CHECK: [[B:%[0-9]+]]:_(s32) = COPY $w1
+; CHECK: [[C:%[0-9]+]]:_(s32) = COPY $w2
+; CHECK: [[RES:%[0-9]+]]:_(s32) = G_FSHR [[A]], [[B]], [[C]]
+; CHECK: $w0 = COPY [[RES]]
+  %res = call i32 @llvm.fshr.i32(i32 %a, i32 %b, i32 %c)
+  ret i32 %res
+}
+
 declare void @llvm.lifetime.start.p0i8(i64, i8*)
 declare void @llvm.lifetime.end.p0i8(i64, i8*)
 define void @test_lifetime_intrin() {

llvm/test/CodeGen/AArch64/GlobalISel/legalizer-info-validation.mir

@@ -263,6 +263,12 @@
 # DEBUG-NEXT: .. opcode {{[0-9]+}} is aliased to {{[0-9]+}}
 # DEBUG-NEXT: .. type index coverage check SKIPPED: user-defined predicate detected
 # DEBUG-NEXT: .. imm index coverage check SKIPPED: user-defined predicate detected
+# DEBUG-NEXT: G_FSHL (opcode {{[0-9]+}}): 2 type indices, 0 imm indices
+# DEBUG-NEXT: .. type index coverage check SKIPPED: no rules defined
+# DEBUG-NEXT: .. imm index coverage check SKIPPED: no rules defined
+# DEBUG-NEXT: G_FSHR (opcode {{[0-9]+}}): 2 type indices, 0 imm indices
+# DEBUG-NEXT: .. type index coverage check SKIPPED: no rules defined
+# DEBUG-NEXT: .. imm index coverage check SKIPPED: no rules defined
 # DEBUG-NEXT: G_ICMP (opcode {{[0-9]+}}): 2 type indices, 0 imm indices
 # DEBUG-NEXT: .. type index coverage check SKIPPED: user-defined predicate detected
 # DEBUG-NEXT: .. imm index coverage check SKIPPED: user-defined predicate detected