; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=riscv32 -mattr=+d,+zfh,+zvfh,+v,+m -target-abi=ilp32d \
; RUN:     -verify-machineinstrs < %s | FileCheck %s
; RUN: llc -mtriple=riscv64 -mattr=+d,+zfh,+zvfh,+v,+m -target-abi=lp64d \
; RUN:     -verify-machineinstrs < %s | FileCheck %s

declare <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x i1>, i32)
declare <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double>, <vscale x 1 x i1>, i32)
declare <vscale x 1 x double> @llvm.vp.fmul.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x i1>, i32)

; (-N0 * -N1) + N2 --> (N0 * N1) + N2
define <vscale x 1 x double> @test1(<vscale x 1 x double> %a, <vscale x 1 x double> %b, <vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: test1:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT:    vfmadd.vv v9, v8, v10, v0.t
; CHECK-NEXT:    vmv.v.v v8, v9
; CHECK-NEXT:    ret
  %nega = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %a, <vscale x 1 x i1> %m, i32 %evl)
  %negb = call <vscale x 1 x double> @llvm.vp.fneg.nxv1f64(<vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
  %v = call <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %nega, <vscale x 1 x double> %negb, <vscale x 1 x double> %c, <vscale x 1 x i1> %m, i32 %evl)
  ret <vscale x 1 x double> %v
}

; (fma x, c1, (fmul x, c2)) -> (fmul x, c1+c2)
define <vscale x 1 x double> @test2(<vscale x 1 x double> %a, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: test2:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI1_0)
; CHECK-NEXT:    addi a1, a1, %lo(.LCPI1_0)
; CHECK-NEXT:    vsetvli a2, zero, e64, m1, ta, ma
; CHECK-NEXT:    vlse64.v v9, (a1), zero
; CHECK-NEXT:    lui a1, %hi(.LCPI1_1)
; CHECK-NEXT:    fld fa5, %lo(.LCPI1_1)(a1)
; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT:    vfadd.vf v9, v9, fa5, v0.t
; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
; CHECK-NEXT:    ret
  %elt.head1 = insertelement <vscale x 1 x double> poison, double 2.0, i32 0
  %c1 = shufflevector <vscale x 1 x double> %elt.head1, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
  %t = call <vscale x 1 x double> @llvm.vp.fmul.nxv1f64(<vscale x 1 x double> %a, <vscale x 1 x double> %c1, <vscale x 1 x i1> %m, i32 %evl)
  %elt.head2 = insertelement <vscale x 1 x double> poison, double 4.0, i32 0
  %c2 = shufflevector <vscale x 1 x double> %elt.head2, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
  %v = call fast <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %a, <vscale x 1 x double> %c2, <vscale x 1 x double> %t, <vscale x 1 x i1> %m, i32 %evl)
  ret <vscale x 1 x double> %v
}

; (fma (fmul x, c1), c2, y) -> (fma x, c1*c2, y)
define <vscale x 1 x double> @test3(<vscale x 1 x double> %a, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
; CHECK-LABEL: test3:
; CHECK:       # %bb.0:
; CHECK-NEXT:    lui a1, %hi(.LCPI2_0)
; CHECK-NEXT:    addi a1, a1, %lo(.LCPI2_0)
; CHECK-NEXT:    vsetvli a2, zero, e64, m1, ta, ma
; CHECK-NEXT:    vlse64.v v10, (a1), zero
; CHECK-NEXT:    lui a1, %hi(.LCPI2_1)
; CHECK-NEXT:    fld fa5, %lo(.LCPI2_1)(a1)
; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, ma
; CHECK-NEXT:    vfmul.vf v10, v10, fa5, v0.t
; CHECK-NEXT:    vfmadd.vv v10, v8, v9, v0.t
; CHECK-NEXT:    vmv.v.v v8, v10
; CHECK-NEXT:    ret
  %elt.head1 = insertelement <vscale x 1 x double> poison, double 2.0, i32 0
  %c1 = shufflevector <vscale x 1 x double> %elt.head1, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
  %t = call <vscale x 1 x double> @llvm.vp.fmul.nxv1f64(<vscale x 1 x double> %a, <vscale x 1 x double> %c1, <vscale x 1 x i1> %m, i32 %evl)
  %elt.head2 = insertelement <vscale x 1 x double> poison, double 4.0, i32 0
  %c2 = shufflevector <vscale x 1 x double> %elt.head2, <vscale x 1 x double> poison, <vscale x 1 x i32> zeroinitializer
  %v = call fast <vscale x 1 x double> @llvm.vp.fma.nxv1f64(<vscale x 1 x double> %t, <vscale x 1 x double> %c2, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
  ret <vscale x 1 x double> %v
}