[RISCV] Use vmv.s.x instead of vfmv.s.f when the floating point scalar is 0.

[jacquesguan]

Summary

Use integer vector scalar move instruction when move 0 to avoid add a integer-float move instruction.

[craig.topper]

LGTM

llvm/lib/Target/RISCV/RISCVInstrInfoVVLPatterns.td

 
 } // Predicates = [HasVInstructions]
 
 let Predicates = [HasVInstructionsAnyF] in {
 
 // 17.2. Floating-Point Scalar Move Instructions
 foreach vti = AllFloatVectors in {
   def : Pat<(vti.Vector (riscv_vfmv_s_f_vl (vti.Vector vti.RegClass:$merge),
+                                           (vti.Scalar (fpimm0)),
+                                           VLOpFrag)),
+            (!cast<Instruction>("PseudoVMV_S_X_"#vti.LMul.MX)
+                vti.RegClass:$merge, X0, GPR:$vl, vti.Log2SEW)>;
+  def : Pat<(vti.Vector (riscv_vfmv_s_f_vl (vti.Vector vti.RegClass:$merge),
                                            vti.ScalarRegClass:$rs1,
                                            VLOpFrag)),
             (!cast<Instruction>("PseudoVFMV_S_"#vti.ScalarSuffix#"_"#vti.LMul.MX)
                 vti.RegClass:$merge,
                 (vti.Scalar vti.ScalarRegClass:$rs1), GPR:$vl, vti.Log2SEW)>;
   defvar ivti = GetIntVTypeInfo<vti>.Vti;
   def : Pat<(vti.Vector (riscv_vrgather_vv_vl vti.RegClass:$rs2,
                                               (ivti.Vector vti.RegClass:$rs1),

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-fp-buildvec.ll

 
 define void @buildvec_dominant0_v2f32(<2 x float>* %x) {
 ; CHECK-LABEL: buildvec_dominant0_v2f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    lui a1, %hi(.LCPI2_0)
 ; CHECK-NEXT:    addi a1, a1, %lo(.LCPI2_0)
 ; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
 ; CHECK-NEXT:    vlse32.v v8, (a1), zero
-; CHECK-NEXT:    fmv.w.x ft0, zero
 ; CHECK-NEXT:    vsetvli zero, zero, e32, mf2, tu, mu
-; CHECK-NEXT:    vfmv.s.f v8, ft0
+; CHECK-NEXT:    vmv.s.x v8, zero
 ; CHECK-NEXT:    vse32.v v8, (a0)
 ; CHECK-NEXT:    ret
   store <2 x float> <float 0.0, float 1.0>, <2 x float>* %x
   ret void
 }
 
 ; We don't want to lower this to the insertion of two scalar elements as above,
 ; as each would require their own load from the constant pool.

 
 define void @buildvec_dominant0_v4f32(<4 x float>* %x) {
 ; CHECK-LABEL: buildvec_dominant0_v4f32:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    lui a1, %hi(.LCPI4_0)
 ; CHECK-NEXT:    addi a1, a1, %lo(.LCPI4_0)
 ; CHECK-NEXT:    vlse32.v v8, (a1), zero
-; CHECK-NEXT:    fmv.w.x ft0, zero
-; CHECK-NEXT:    vfmv.s.f v9, ft0
+; CHECK-NEXT:    vmv.s.x v9, zero
 ; CHECK-NEXT:    vsetivli zero, 3, e32, m1, tu, mu
 ; CHECK-NEXT:    vslideup.vi v8, v9, 2
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vse32.v v8, (a0)
 ; CHECK-NEXT:    ret
   store <4 x float> <float 2.0, float 2.0, float 0.0, float 2.0>, <4 x float>* %x
   ret void
 }
 
 define void @buildvec_dominant1_v4f32(<4 x float>* %x, float %f) {
 ; CHECK-LABEL: buildvec_dominant1_v4f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    fmv.w.x ft0, zero
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
-; CHECK-NEXT:    vfmv.s.f v8, ft0
+; CHECK-NEXT:    vmv.s.x v8, zero
 ; CHECK-NEXT:    vfmv.v.f v9, fa0
 ; CHECK-NEXT:    vsetivli zero, 2, e32, m1, tu, mu
 ; CHECK-NEXT:    vslideup.vi v9, v8, 1
 ; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
 ; CHECK-NEXT:    vse32.v v9, (a0)
 ; CHECK-NEXT:    ret
   %v0 = insertelement <4 x float> undef, float %f, i32 0
   %v1 = insertelement <4 x float> %v0, float 0.0, i32 1