GlobalISel: Add G_FMAD instruction

llvm-svn: 371254

Author: arsenm

llvm/include/llvm/CodeGen/GlobalISel/MachineIRBuilder.h

@@ -1372,6 +1372,13 @@ class MachineIRBuilder {
     return buildInstr(TargetOpcode::G_FMA, {Dst}, {Src0, Src1, Src2});
   }
 
+  /// Build and insert \p Res = G_FMAD \p Op0, \p Op1, \p Op2
+  MachineInstrBuilder buildFMAD(const DstOp &Dst, const SrcOp &Src0,
+                                const SrcOp &Src1, const SrcOp &Src2,
+                                Optional<unsigned> Flags = None) {
+    return buildInstr(TargetOpcode::G_FMAD, {Dst}, {Src0, Src1, Src2}, Flags);
+  }
+
   /// Build and insert \p Res = G_FNEG \p Op0
   MachineInstrBuilder buildFNeg(const DstOp &Dst, const SrcOp &Src0) {
     return buildInstr(TargetOpcode::G_FNEG, {Dst}, {Src0});

llvm/include/llvm/Support/TargetOpcodes.def

@@ -439,6 +439,9 @@ HANDLE_TARGET_OPCODE(G_FMUL)
 /// Generic FMA multiplication. Behaves like llvm fma intrinsic
 HANDLE_TARGET_OPCODE(G_FMA)
 
+/// Generic FP multiply and add. Behaves as separate fmul and fadd.
+HANDLE_TARGET_OPCODE(G_FMAD)
+
 /// Generic FP division.
 HANDLE_TARGET_OPCODE(G_FDIV)
 

llvm/include/llvm/Target/GenericOpcodes.td

@@ -624,6 +624,15 @@ def G_FMA : GenericInstruction {
   let isCommutable = 0;
 }
 
+/// Generic FP multiply and add. Perform a * b + c, while getting the
+/// same result as the separately rounded operations, unlike G_FMA.
+def G_FMAD : GenericInstruction {
+  let OutOperandList = (outs type0:$dst);
+  let InOperandList = (ins type0:$src1, type0:$src2, type0:$src3);
+  let hasSideEffects = 0;
+  let isCommutable = 0;
+}
+
 // Generic FP division.
 def G_FDIV : GenericInstruction {
   let OutOperandList = (outs type0:$dst);

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

@@ -77,6 +77,7 @@ def : GINodeEquiv<G_UITOFP, uint_to_fp>;
 def : GINodeEquiv<G_FADD, fadd>;
 def : GINodeEquiv<G_FSUB, fsub>;
 def : GINodeEquiv<G_FMA, fma>;
+def : GINodeEquiv<G_FMAD, fmad>;
 def : GINodeEquiv<G_FMUL, fmul>;
 def : GINodeEquiv<G_FDIV, fdiv>;
 def : GINodeEquiv<G_FREM, frem>;

llvm/lib/CodeGen/GlobalISel/LegalizerHelper.cpp

@@ -1753,6 +1753,7 @@ LegalizerHelper::widenScalar(MachineInstr &MI, unsigned TypeIdx, LLT WideTy) {
   case TargetOpcode::G_FMUL:
   case TargetOpcode::G_FSUB:
   case TargetOpcode::G_FMA:
+  case TargetOpcode::G_FMAD:
   case TargetOpcode::G_FNEG:
   case TargetOpcode::G_FABS:
   case TargetOpcode::G_FCANONICALIZE:
@@ -2828,6 +2829,7 @@ LegalizerHelper::fewerElementsVector(MachineInstr &MI, unsigned TypeIdx,
   case G_FDIV:
   case G_FREM:
   case G_FMA:
+  case G_FMAD:
   case G_FPOW:
   case G_FEXP:
   case G_FEXP2:

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

@@ -312,6 +312,9 @@
 # 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_FMAD (opcode 116): 1 type index, 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_FDIV (opcode {{[0-9]+}}): 1 type index, 0 imm indices
 # DEBUG-NEXT: .. opcode {{[0-9]+}} is aliased to {{[0-9]+}}
 # DEBUG-NEXT: .. the first uncovered type index: 1, OK

llvm/unittests/CodeGen/GlobalISel/MachineIRBuilderTest.cpp

@@ -130,6 +130,8 @@ TEST_F(GISelMITest, TestBuildFPInsts) {
   B.buildFAdd(S64, Copies[0], Copies[1]);
   B.buildFSub(S64, Copies[0], Copies[1]);
   B.buildFMA(S64, Copies[0], Copies[1], Copies[2]);
+  B.buildFMAD(S64, Copies[0], Copies[1], Copies[2]);
+  B.buildFMAD(S64, Copies[0], Copies[1], Copies[2], MachineInstr::FmNoNans);
   B.buildFNeg(S64, Copies[0]);
   B.buildFAbs(S64, Copies[0]);
   B.buildFCopysign(S64, Copies[0], Copies[1]);
@@ -141,6 +143,8 @@ TEST_F(GISelMITest, TestBuildFPInsts) {
   ; CHECK: [[FADD:%[0-9]+]]:_(s64) = G_FADD [[COPY0]]:_, [[COPY1]]:_
   ; CHECK: [[FSUB:%[0-9]+]]:_(s64) = G_FSUB [[COPY0]]:_, [[COPY1]]:_
   ; CHECK: [[FMA:%[0-9]+]]:_(s64) = G_FMA [[COPY0]]:_, [[COPY1]]:_, [[COPY2]]:_
+  ; CHECK: [[FMAD0:%[0-9]+]]:_(s64) = G_FMAD [[COPY0]]:_, [[COPY1]]:_, [[COPY2]]:_
+  ; CHECK: [[FMAD1:%[0-9]+]]:_(s64) = nnan G_FMAD [[COPY0]]:_, [[COPY1]]:_, [[COPY2]]:_
   ; CHECK: [[FNEG:%[0-9]+]]:_(s64) = G_FNEG [[COPY0]]:_
   ; CHECK: [[FABS:%[0-9]+]]:_(s64) = G_FABS [[COPY0]]:_
   ; CHECK: [[FCOPYSIGN:%[0-9]+]]:_(s64) = G_FCOPYSIGN [[COPY0]]:_, [[COPY1]]:_