[RISCV] Add intrinsics for vfmv.f.s and vfmv.s.f

Similar to D93365, but for floating point. No need for special ISD opcodes
though. We can directly isel these from intrinsics. I had to use anyfloat_ty
instead of anyvector_ty in the intrinsics to make LLVMVectorElementType not
crash when imported into the -gen-dag-isel tablegen backend.

Differential Revision: https://reviews.llvm.org/D93426

Author: topperc

llvm/include/llvm/IR/IntrinsicsRISCV.td

@@ -349,4 +349,12 @@ let TargetPrefix = "riscv" in {
                                     [IntrNoMem]>, RISCVVIntrinsic {
     let ExtendOperand = 2;
   }
+
+  def int_riscv_vfmv_f_s : Intrinsic<[LLVMVectorElementType<0>],
+                                     [llvm_anyfloat_ty],
+                                     [IntrNoMem]>, RISCVVIntrinsic;
+  def int_riscv_vfmv_s_f : Intrinsic<[llvm_anyfloat_ty],
+                                     [LLVMMatchType<0>, LLVMVectorElementType<0>,
+                                      llvm_anyint_ty],
+                                     [IntrNoMem]>, RISCVVIntrinsic;
 } // TargetPrefix = "riscv"

llvm/lib/Target/RISCV/RISCVInstrInfoV.td

@@ -980,8 +980,9 @@ let hasSideEffects = 0, mayLoad = 0, mayStore = 0, vm = 1 in {
 // Floating-Point Scalar Move Instructions
 def VFMV_F_S : RVInstV<0b010000, 0b00000, OPFVV, (outs FPR32:$vd),
                       (ins VR:$vs2), "vfmv.f.s", "$vd, $vs2">;
-def VFMV_S_F : RVInstV2<0b010000, 0b00000, OPFVF, (outs VR:$vd),
-                      (ins FPR32:$rs1), "vfmv.s.f", "$vd, $rs1">;
+let Constraints = "$vd = $vd_wb" in
+def VFMV_S_F : RVInstV2<0b010000, 0b00000, OPFVF, (outs VR:$vd_wb),
+                      (ins VR:$vd, FPR32:$rs1), "vfmv.s.f", "$vd, $rs1">;
 
 } // hasSideEffects = 0, mayLoad = 0, mayStore = 0, vm = 1
 } // Predicates = [HasStdExtV, HasStdExtF]

llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td

@@ -1370,12 +1370,37 @@ let mayLoad = 0, mayStore = 0, hasSideEffects = 0, usesCustomInserter = 1,
           Constraints = "$rd = $rs1" in
       def PseudoVMV_S_X # "_" # m.MX: Pseudo<(outs m.vrclass:$rd),
                                              (ins m.vrclass:$rs1, GPR:$rs2,
-                                             GPR:$vl, ixlenimm:$sew),
+                                                  GPR:$vl, ixlenimm:$sew),
                                              []>, RISCVVPseudo;
     }
   }
 }
+} // Predicates = [HasStdExtV]
+
+//===----------------------------------------------------------------------===//
+// 17.2. Floating-Point Scalar Move Instructions
+//===----------------------------------------------------------------------===//
+
+let Predicates = [HasStdExtV, HasStdExtF] in {
+let mayLoad = 0, mayStore = 0, hasSideEffects = 0, usesCustomInserter = 1,
+    Uses = [VL, VTYPE] in {
+  foreach m = MxList.m in {
+    let VLMul = m.value in {
+      let SEWIndex = 2, BaseInstr = VFMV_F_S in
+      def PseudoVFMV_F_S # "_" # m.MX : Pseudo<(outs FPR32:$rd),
+                                               (ins m.vrclass:$rs2,
+                                                    ixlenimm:$sew),
+                                               []>, RISCVVPseudo;
+      let VLIndex = 3, SEWIndex = 4, BaseInstr = VFMV_S_F,
+          Constraints = "$rd = $rs1" in
+      def PseudoVFMV_S_F # "_" # m.MX : Pseudo<(outs m.vrclass:$rd),
+                                               (ins m.vrclass:$rs1, FPR32:$rs2,
+                                                    GPR:$vl, ixlenimm:$sew),
+                                               []>, RISCVVPseudo;
+    }
+  }
 }
+} // Predicates = [HasStdExtV, HasStdExtF]
 
 //===----------------------------------------------------------------------===//
 // Patterns.
@@ -1557,3 +1582,34 @@ foreach vti = AllIntegerVectors in {
              (vti.Vector $rs1), $rs2, (NoX0 GPR:$vl), vti.SEW)>;
 }
 } // Predicates = [HasStdExtV]
+
+//===----------------------------------------------------------------------===//
+// 17.2. Floating-Point Scalar Move Instructions
+//===----------------------------------------------------------------------===//
+
+let Predicates = [HasStdExtV, HasStdExtF] in {
+foreach fvti = AllFloatVectors in {
+  defvar instr = !cast<Instruction>("PseudoVFMV_F_S_" # fvti.LMul.MX);
+  def : Pat<(fvti.Scalar (int_riscv_vfmv_f_s (fvti.Vector fvti.RegClass:$rs2))),
+             // Floating point instructions with a scalar result will always
+             // generate the result in a register of class FPR32. When dealing
+             // with the f64 variant of a pattern we need to promote the FPR32
+             // subregister generated by the instruction to the FPR64 base
+             // register expected by the type in the pattern
+             !cond(!eq(!cast<string>(fvti.ScalarRegClass),
+                       !cast<string>(FPR64)):
+                      (SUBREG_TO_REG (i32 -1),
+                                     (instr $rs2, fvti.SEW), sub_32),
+                   !eq(!cast<string>(fvti.ScalarRegClass),
+                       !cast<string>(FPR16)):
+                      (EXTRACT_SUBREG (instr $rs2, fvti.SEW), sub_16),
+                   !eq(1, 1):
+                      (instr $rs2, fvti.SEW))>;
+
+  def : Pat<(fvti.Vector (int_riscv_vfmv_s_f (fvti.Vector fvti.RegClass:$rs1),
+                         (fvti.Scalar fvti.ScalarRegClass:$rs2), GPR:$vl)),
+            (!cast<Instruction>("PseudoVFMV_S_F_" # fvti.LMul.MX)
+             (fvti.Vector $rs1), ToFPR32<fvti.Scalar, fvti.ScalarRegClass, "rs2">.ret,
+             (NoX0 GPR:$vl), fvti.SEW)>;
+}
+} // Predicates = [HasStdExtV, HasStdExtF]