llvm/test/CodeGen/RISCV/half-arith.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=riscv32 -mattr=+zfh -verify-machineinstrs \
; RUN:   -target-abi ilp32f < %s | FileCheck -check-prefix=CHECKIZFH %s
; RUN: llc -mtriple=riscv64 -mattr=+zfh -verify-machineinstrs \
; RUN:   -target-abi lp64f < %s | FileCheck -check-prefix=CHECKIZFH %s
; RUN: llc -mtriple=riscv32 -verify-machineinstrs < %s \
; RUN:   | FileCheck -check-prefix=RV32I %s
; RUN: llc -mtriple=riscv64 -verify-machineinstrs < %s \
; RUN:   | FileCheck -check-prefix=RV64I %s
; RUN: llc -mtriple=riscv32 -mattr=+zfhmin -verify-machineinstrs \
; RUN:   -target-abi ilp32f < %s | FileCheck -check-prefixes=CHECKIZFHMIN,CHECK-RV32-FSGNJ %s
; RUN: llc -mtriple=riscv64 -mattr=+zfhmin -verify-machineinstrs \
; RUN:   -target-abi lp64f < %s | FileCheck --check-prefixes=CHECKIZFHMIN,CHECK-RV64-FSGNJ %s

; These tests are each targeted at a particular RISC-V FPU instruction.
; Compares and conversions can be found in half-fcmp.ll and half-convert.ll
; respectively. Some other half-*.ll files in this folder exercise LLVM IR
; instructions that don't directly match a RISC-V instruction.

define half @fadd_s(half %a, half %b) nounwind {
; CHECKIZFH-LABEL: fadd_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fadd.h fa0, fa0, fa1
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fadd_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -16
; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a1
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s2, a1, -1
; RV32I-NEXT:    and a0, a0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s1
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 16
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fadd_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -32
; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a1
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s2, a1, -1
; RV64I-NEXT:    and a0, a0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s1
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 32
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fadd_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa0
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %1 = fadd half %a, %b
  ret half %1
}

define half @fsub_s(half %a, half %b) nounwind {
; CHECKIZFH-LABEL: fsub_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fsub.h fa0, fa0, fa1
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fsub_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -16
; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a1
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s2, a1, -1
; RV32I-NEXT:    and a0, a0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s1
; RV32I-NEXT:    call __subsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 16
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fsub_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -32
; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a1
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s2, a1, -1
; RV64I-NEXT:    and a0, a0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s1
; RV64I-NEXT:    call __subsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 32
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fsub_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa0
; CHECKIZFHMIN-NEXT:    fsub.s ft0, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %1 = fsub half %a, %b
  ret half %1
}

define half @fmul_s(half %a, half %b) nounwind {
; CHECKIZFH-LABEL: fmul_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmul.h fa0, fa0, fa1
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fmul_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -16
; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a1
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s2, a1, -1
; RV32I-NEXT:    and a0, a0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s1
; RV32I-NEXT:    call __mulsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 16
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fmul_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -32
; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a1
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s2, a1, -1
; RV64I-NEXT:    and a0, a0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s1
; RV64I-NEXT:    call __mulsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 32
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fmul_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa0
; CHECKIZFHMIN-NEXT:    fmul.s ft0, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %1 = fmul half %a, %b
  ret half %1
}

define half @fdiv_s(half %a, half %b) nounwind {
; CHECKIZFH-LABEL: fdiv_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fdiv.h fa0, fa0, fa1
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fdiv_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -16
; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a1
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s2, a1, -1
; RV32I-NEXT:    and a0, a0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s1
; RV32I-NEXT:    call __divsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 16
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fdiv_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -32
; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a1
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s2, a1, -1
; RV64I-NEXT:    and a0, a0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s1
; RV64I-NEXT:    call __divsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 32
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fdiv_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa0
; CHECKIZFHMIN-NEXT:    fdiv.s ft0, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %1 = fdiv half %a, %b
  ret half %1
}

declare half @llvm.sqrt.f16(half)

define half @fsqrt_s(half %a) nounwind {
; CHECKIZFH-LABEL: fsqrt_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fsqrt.h fa0, fa0
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fsqrt_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -16
; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    slli a0, a0, 16
; RV32I-NEXT:    srli a0, a0, 16
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    call sqrtf@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 16
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fsqrt_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -16
; RV64I-NEXT:    sd ra, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    slli a0, a0, 48
; RV64I-NEXT:    srli a0, a0, 48
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    call sqrtf@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 16
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fsqrt_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa0
; CHECKIZFHMIN-NEXT:    fsqrt.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %1 = call half @llvm.sqrt.f16(half %a)
  ret half %1
}

declare half @llvm.copysign.f16(half, half)

define half @fsgnj_s(half %a, half %b) nounwind {
; CHECKIZFH-LABEL: fsgnj_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fsgnj.h fa0, fa0, fa1
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fsgnj_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    lui a2, 1048568
; RV32I-NEXT:    and a1, a1, a2
; RV32I-NEXT:    slli a0, a0, 17
; RV32I-NEXT:    srli a0, a0, 17
; RV32I-NEXT:    or a0, a0, a1
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fsgnj_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    lui a2, 1048568
; RV64I-NEXT:    and a1, a1, a2
; RV64I-NEXT:    slli a0, a0, 49
; RV64I-NEXT:    srli a0, a0, 49
; RV64I-NEXT:    or a0, a0, a1
; RV64I-NEXT:    ret
;
; CHECK-RV32-FSGNJ-LABEL: fsgnj_s:
; CHECK-RV32-FSGNJ:       # %bb.0:
; CHECK-RV32-FSGNJ-NEXT:    addi sp, sp, -16
; CHECK-RV32-FSGNJ-NEXT:    fsh fa1, 12(sp)
; CHECK-RV32-FSGNJ-NEXT:    fsh fa0, 8(sp)
; CHECK-RV32-FSGNJ-NEXT:    lbu a0, 13(sp)
; CHECK-RV32-FSGNJ-NEXT:    lbu a1, 9(sp)
; CHECK-RV32-FSGNJ-NEXT:    andi a0, a0, 128
; CHECK-RV32-FSGNJ-NEXT:    andi a1, a1, 127
; CHECK-RV32-FSGNJ-NEXT:    or a0, a1, a0
; CHECK-RV32-FSGNJ-NEXT:    sb a0, 9(sp)
; CHECK-RV32-FSGNJ-NEXT:    flh fa0, 8(sp)
; CHECK-RV32-FSGNJ-NEXT:    addi sp, sp, 16
; CHECK-RV32-FSGNJ-NEXT:    ret
;
; CHECK-RV64-FSGNJ-LABEL: fsgnj_s:
; CHECK-RV64-FSGNJ:       # %bb.0:
; CHECK-RV64-FSGNJ-NEXT:    addi sp, sp, -16
; CHECK-RV64-FSGNJ-NEXT:    fsh fa1, 8(sp)
; CHECK-RV64-FSGNJ-NEXT:    fsh fa0, 0(sp)
; CHECK-RV64-FSGNJ-NEXT:    lbu a0, 9(sp)
; CHECK-RV64-FSGNJ-NEXT:    lbu a1, 1(sp)
; CHECK-RV64-FSGNJ-NEXT:    andi a0, a0, 128
; CHECK-RV64-FSGNJ-NEXT:    andi a1, a1, 127
; CHECK-RV64-FSGNJ-NEXT:    or a0, a1, a0
; CHECK-RV64-FSGNJ-NEXT:    sb a0, 1(sp)
; CHECK-RV64-FSGNJ-NEXT:    flh fa0, 0(sp)
; CHECK-RV64-FSGNJ-NEXT:    addi sp, sp, 16
; CHECK-RV64-FSGNJ-NEXT:    ret
; CHECKFSGNJ-LABEL: fsgnj_s:
; CHECKFSGNJ:       # %bb.0:
; CHECKFSGNJ-NEXT:    addi sp, sp, -16
; CHECKFSGNJ-NEXT:    fsh fa1, 12(sp)
; CHECKFSGNJ-NEXT:    fsh fa0, 8(sp)
; CHECKFSGNJ-NEXT:    lbu a0, 13(sp)
; CHECKFSGNJ-NEXT:    lbu a1, 9(sp)
; CHECKFSGNJ-NEXT:    andi a0, a0, 128
; CHECKFSGNJ-NEXT:    andi a1, a1, 127
; CHECKFSGNJ-NEXT:    or a0, a1, a0
; CHECKFSGNJ-NEXT:    sb a0, 9(sp)
; CHECKFSGNJ-NEXT:    flh fa0, 8(sp)
; CHECKFSGNJ-NEXT:    addi sp, sp, 16
; CHECKFSGNJ-NEXT:    ret
; CHECK64FSGNJ-LABEL: fsgnj_s:
; CHECK64FSGNJ:       # %bb.0:
; CHECK64FSGNJ-NEXT:    addi sp, sp, -16
; CHECK64FSGNJ-NEXT:    fsh fa1, 8(sp)
; CHECK64FSGNJ-NEXT:    fsh fa0, 0(sp)
; CHECK64FSGNJ-NEXT:    lbu a0, 9(sp)
; CHECK64FSGNJ-NEXT:    lbu a1, 1(sp)
; CHECK64FSGNJ-NEXT:    andi a0, a0, 128
; CHECK64FSGNJ-NEXT:    andi a1, a1, 127
; CHECK64FSGNJ-NEXT:    or a0, a1, a0
; CHECK64FSGNJ-NEXT:    sb a0, 1(sp)
; CHECK64FSGNJ-NEXT:    flh fa0, 0(sp)
; CHECK64FSGNJ-NEXT:    addi sp, sp, 16
; CHECK64FSGNJ-NEXT:    ret
  %1 = call half @llvm.copysign.f16(half %a, half %b)
  ret half %1
}

; This function performs extra work to ensure that
; DAGCombiner::visitBITCAST doesn't replace the fneg with an xor.
define i32 @fneg_s(half %a, half %b) nounwind {
; CHECKIZFH-LABEL: fneg_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fadd.h ft0, fa0, fa0
; CHECKIZFH-NEXT:    fneg.h ft1, ft0
; CHECKIZFH-NEXT:    feq.h a0, ft0, ft1
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fneg_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -16
; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s1, a1, -1
; RV32I-NEXT:    and a0, a0, s1
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    and a0, a0, s1
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    lui a0, 524288
; RV32I-NEXT:    xor a0, s0, a0
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    and a0, a0, s1
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s0
; RV32I-NEXT:    call __eqsf2@plt
; RV32I-NEXT:    seqz a0, a0
; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 16
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fneg_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -32
; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s1, a1, -1
; RV64I-NEXT:    and a0, a0, s1
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    and a0, a0, s1
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    lui a0, 524288
; RV64I-NEXT:    xor a0, s0, a0
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    and a0, a0, s1
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s0
; RV64I-NEXT:    call __eqsf2@plt
; RV64I-NEXT:    seqz a0, a0
; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 32
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fneg_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa0
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fneg.s ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, ft1
; CHECKIZFHMIN-NEXT:    feq.s a0, ft0, ft1
; CHECKIZFHMIN-NEXT:    ret
  %1 = fadd half %a, %a
  %2 = fneg half %1
  %3 = fcmp oeq half %1, %2
  %4 = zext i1 %3 to i32
  ret i32 %4
}

; This function performs extra work to ensure that
; DAGCombiner::visitBITCAST doesn't replace the fneg with an xor.
define half @fsgnjn_s(half %a, half %b) nounwind {
; CHECKIZFH-LABEL: fsgnjn_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fadd.h ft0, fa0, fa1
; CHECKIZFH-NEXT:    fsgnjn.h fa0, fa0, ft0
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fsgnjn_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a1
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    lui a0, 16
; RV32I-NEXT:    addi s3, a0, -1
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s2
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    lui a1, 524288
; RV32I-NEXT:    xor a0, a0, a1
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lui a1, 1048568
; RV32I-NEXT:    and a0, a0, a1
; RV32I-NEXT:    slli s1, s1, 17
; RV32I-NEXT:    srli s1, s1, 17
; RV32I-NEXT:    or a0, s1, a0
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fsgnjn_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a1
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    lui a0, 16
; RV64I-NEXT:    addiw s3, a0, -1
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s2
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    lui a1, 524288
; RV64I-NEXT:    xor a0, a0, a1
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    lui a1, 1048568
; RV64I-NEXT:    and a0, a0, a1
; RV64I-NEXT:    slli s1, s1, 49
; RV64I-NEXT:    srli s1, s1, 49
; RV64I-NEXT:    or a0, s1, a0
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECK-RV32-FSGNJ-LABEL: fsgnjn_s:
; CHECK-RV32-FSGNJ:       # %bb.0:
; CHECK-RV32-FSGNJ-NEXT:    addi sp, sp, -16
; CHECK-RV32-FSGNJ-NEXT:    fcvt.s.h ft0, fa1
; CHECK-RV32-FSGNJ-NEXT:    fcvt.s.h ft1, fa0
; CHECK-RV32-FSGNJ-NEXT:    fadd.s ft0, ft1, ft0
; CHECK-RV32-FSGNJ-NEXT:    fcvt.h.s ft0, ft0
; CHECK-RV32-FSGNJ-NEXT:    fcvt.s.h ft0, ft0
; CHECK-RV32-FSGNJ-NEXT:    fneg.s ft0, ft0
; CHECK-RV32-FSGNJ-NEXT:    fcvt.h.s ft0, ft0
; CHECK-RV32-FSGNJ-NEXT:    fsh fa0, 8(sp)
; CHECK-RV32-FSGNJ-NEXT:    fsh ft0, 12(sp)
; CHECK-RV32-FSGNJ-NEXT:    lbu a0, 9(sp)
; CHECK-RV32-FSGNJ-NEXT:    lbu a1, 13(sp)
; CHECK-RV32-FSGNJ-NEXT:    andi a0, a0, 127
; CHECK-RV32-FSGNJ-NEXT:    andi a1, a1, 128
; CHECK-RV32-FSGNJ-NEXT:    or a0, a0, a1
; CHECK-RV32-FSGNJ-NEXT:    sb a0, 9(sp)
; CHECK-RV32-FSGNJ-NEXT:    flh fa0, 8(sp)
; CHECK-RV32-FSGNJ-NEXT:    addi sp, sp, 16
; CHECK-RV32-FSGNJ-NEXT:    ret
;
; CHECK-RV64-FSGNJ-LABEL: fsgnjn_s:
; CHECK-RV64-FSGNJ:       # %bb.0:
; CHECK-RV64-FSGNJ-NEXT:    addi sp, sp, -16
; CHECK-RV64-FSGNJ-NEXT:    fcvt.s.h ft0, fa1
; CHECK-RV64-FSGNJ-NEXT:    fcvt.s.h ft1, fa0
; CHECK-RV64-FSGNJ-NEXT:    fadd.s ft0, ft1, ft0
; CHECK-RV64-FSGNJ-NEXT:    fcvt.h.s ft0, ft0
; CHECK-RV64-FSGNJ-NEXT:    fcvt.s.h ft0, ft0
; CHECK-RV64-FSGNJ-NEXT:    fneg.s ft0, ft0
; CHECK-RV64-FSGNJ-NEXT:    fcvt.h.s ft0, ft0
; CHECK-RV64-FSGNJ-NEXT:    fsh fa0, 0(sp)
; CHECK-RV64-FSGNJ-NEXT:    fsh ft0, 8(sp)
; CHECK-RV64-FSGNJ-NEXT:    lbu a0, 1(sp)
; CHECK-RV64-FSGNJ-NEXT:    lbu a1, 9(sp)
; CHECK-RV64-FSGNJ-NEXT:    andi a0, a0, 127
; CHECK-RV64-FSGNJ-NEXT:    andi a1, a1, 128
; CHECK-RV64-FSGNJ-NEXT:    or a0, a0, a1
; CHECK-RV64-FSGNJ-NEXT:    sb a0, 1(sp)
; CHECK-RV64-FSGNJ-NEXT:    flh fa0, 0(sp)
; CHECK-RV64-FSGNJ-NEXT:    addi sp, sp, 16
; CHECK-RV64-FSGNJ-NEXT:    ret
; CHECKFSGNJ-LABEL: fsgnjn_s:
; CHECKFSGNJ:       # %bb.0:
; CHECKFSGNJ-NEXT:    addi sp, sp, -16
; CHECKFSGNJ-NEXT:    fcvt.s.h ft0, fa1
; CHECKFSGNJ-NEXT:    fcvt.s.h ft1, fa0
; CHECKFSGNJ-NEXT:    fadd.s ft0, ft1, ft0
; CHECKFSGNJ-NEXT:    fcvt.h.s ft0, ft0
; CHECKFSGNJ-NEXT:    fcvt.s.h ft0, ft0
; CHECKFSGNJ-NEXT:    fneg.s ft0, ft0
; CHECKFSGNJ-NEXT:    fcvt.h.s ft0, ft0
; CHECKFSGNJ-NEXT:    fsh fa0, 8(sp)
; CHECKFSGNJ-NEXT:    fsh ft0, 12(sp)
; CHECKFSGNJ-NEXT:    lbu a0, 9(sp)
; CHECKFSGNJ-NEXT:    lbu a1, 13(sp)
; CHECKFSGNJ-NEXT:    andi a0, a0, 127
; CHECKFSGNJ-NEXT:    andi a1, a1, 128
; CHECKFSGNJ-NEXT:    or a0, a0, a1
; CHECKFSGNJ-NEXT:    sb a0, 9(sp)
; CHECKFSGNJ-NEXT:    flh fa0, 8(sp)
; CHECKFSGNJ-NEXT:    addi sp, sp, 16
; CHECKFSGNJ-NEXT:    ret
; CHECK64FSGNJ-LABEL: fsgnjn_s:
; CHECK64FSGNJ:       # %bb.0:
; CHECK64FSGNJ-NEXT:    addi sp, sp, -16
; CHECK64FSGNJ-NEXT:    fcvt.s.h ft0, fa1
; CHECK64FSGNJ-NEXT:    fcvt.s.h ft1, fa0
; CHECK64FSGNJ-NEXT:    fadd.s ft0, ft1, ft0
; CHECK64FSGNJ-NEXT:    fcvt.h.s ft0, ft0
; CHECK64FSGNJ-NEXT:    fcvt.s.h ft0, ft0
; CHECK64FSGNJ-NEXT:    fneg.s ft0, ft0
; CHECK64FSGNJ-NEXT:    fcvt.h.s ft0, ft0
; CHECK64FSGNJ-NEXT:    fsh fa0, 0(sp)
; CHECK64FSGNJ-NEXT:    fsh ft0, 8(sp)
; CHECK64FSGNJ-NEXT:    lbu a0, 1(sp)
; CHECK64FSGNJ-NEXT:    lbu a1, 9(sp)
; CHECK64FSGNJ-NEXT:    andi a0, a0, 127
; CHECK64FSGNJ-NEXT:    andi a1, a1, 128
; CHECK64FSGNJ-NEXT:    or a0, a0, a1
; CHECK64FSGNJ-NEXT:    sb a0, 1(sp)
; CHECK64FSGNJ-NEXT:    flh fa0, 0(sp)
; CHECK64FSGNJ-NEXT:    addi sp, sp, 16
; CHECK64FSGNJ-NEXT:    ret
  %1 = fadd half %a, %b
  %2 = fneg half %1
  %3 = call half @llvm.copysign.f16(half %a, half %2)
  ret half %3
}

declare half @llvm.fabs.f16(half)

; This function performs extra work to ensure that
; DAGCombiner::visitBITCAST doesn't replace the fabs with an and.
define half @fabs_s(half %a, half %b) nounwind {
; CHECKIZFH-LABEL: fabs_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fadd.h ft0, fa0, fa1
; CHECKIZFH-NEXT:    fabs.h ft1, ft0
; CHECKIZFH-NEXT:    fadd.h fa0, ft1, ft0
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fabs_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -16
; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a1
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s2, a1, -1
; RV32I-NEXT:    and a0, a0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s1
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    and a0, a0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    slli a0, a0, 1
; RV32I-NEXT:    srli a0, a0, 1
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    and a0, a0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, s0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 16
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fabs_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -32
; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a1
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s2, a1, -1
; RV64I-NEXT:    and a0, a0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s1
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    and a0, a0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    slli a0, a0, 33
; RV64I-NEXT:    srli a0, a0, 33
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    and a0, a0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, s0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 32
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fabs_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa0
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fabs.s ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, ft1
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %1 = fadd half %a, %b
  %2 = call half @llvm.fabs.f16(half %1)
  %3 = fadd half %2, %1
  ret half %3
}

declare half @llvm.minnum.f16(half, half)

define half @fmin_s(half %a, half %b) nounwind {
; CHECKIZFH-LABEL: fmin_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmin.h fa0, fa0, fa1
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fmin_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -16
; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a1
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s2, a1, -1
; RV32I-NEXT:    and a0, a0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s1
; RV32I-NEXT:    call fminf@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 16
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fmin_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -32
; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a1
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s2, a1, -1
; RV64I-NEXT:    and a0, a0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s1
; RV64I-NEXT:    call fminf@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 32
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fmin_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa0
; CHECKIZFHMIN-NEXT:    fmin.s ft0, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %1 = call half @llvm.minnum.f16(half %a, half %b)
  ret half %1
}

declare half @llvm.maxnum.f16(half, half)

define half @fmax_s(half %a, half %b) nounwind {
; CHECKIZFH-LABEL: fmax_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmax.h fa0, fa0, fa1
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fmax_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -16
; RV32I-NEXT:    sw ra, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 8(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 4(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 0(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a1
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s2, a1, -1
; RV32I-NEXT:    and a0, a0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s2
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s1
; RV32I-NEXT:    call fmaxf@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 8(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 4(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 0(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 16
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fmax_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -32
; RV64I-NEXT:    sd ra, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 0(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a1
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s2, a1, -1
; RV64I-NEXT:    and a0, a0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s2
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s1
; RV64I-NEXT:    call fmaxf@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 0(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 32
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fmax_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa0
; CHECKIZFHMIN-NEXT:    fmax.s ft0, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %1 = call half @llvm.maxnum.f16(half %a, half %b)
  ret half %1
}

declare half @llvm.fma.f16(half, half, half)

define half @fmadd_s(half %a, half %b, half %c) nounwind {
; CHECKIZFH-LABEL: fmadd_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmadd.h fa0, fa0, fa1, fa2
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fmadd_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a2
; RV32I-NEXT:    mv s1, a1
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s3, a1, -1
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a2, a0
; RV32I-NEXT:    mv a0, s2
; RV32I-NEXT:    mv a1, s1
; RV32I-NEXT:    call fmaf@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fmadd_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a2
; RV64I-NEXT:    mv s1, a1
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s3, a1, -1
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a2, a0
; RV64I-NEXT:    mv a0, s2
; RV64I-NEXT:    mv a1, s1
; RV64I-NEXT:    call fmaf@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fmadd_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa2
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa0
; CHECKIZFHMIN-NEXT:    fmadd.s ft0, ft2, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %1 = call half @llvm.fma.f16(half %a, half %b, half %c)
  ret half %1
}

define half @fmsub_s(half %a, half %b, half %c) nounwind {
; CHECKIZFH-LABEL: fmsub_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmv.h.x ft0, zero
; CHECKIZFH-NEXT:    fadd.h ft0, fa2, ft0
; CHECKIZFH-NEXT:    fmsub.h fa0, fa0, fa1, ft0
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fmsub_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a1
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    lui a0, 16
; RV32I-NEXT:    addi s3, a0, -1
; RV32I-NEXT:    and a0, a2, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    lui a1, 524288
; RV32I-NEXT:    xor a0, a0, a1
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    and a0, s2, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a2, a0
; RV32I-NEXT:    mv a0, s1
; RV32I-NEXT:    mv a1, s0
; RV32I-NEXT:    call fmaf@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fmsub_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a1
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    lui a0, 16
; RV64I-NEXT:    addiw s3, a0, -1
; RV64I-NEXT:    and a0, a2, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    lui a1, 524288
; RV64I-NEXT:    xor a0, a0, a1
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    and a0, s2, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a2, a0
; RV64I-NEXT:    mv a0, s1
; RV64I-NEXT:    mv a1, s0
; RV64I-NEXT:    call fmaf@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fmsub_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa2
; CHECKIZFHMIN-NEXT:    fmv.w.x ft1, zero
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fneg.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa0
; CHECKIZFHMIN-NEXT:    fmadd.s ft0, ft2, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %c_ = fadd half 0.0, %c ; avoid negation using xor
  %negc = fsub half -0.0, %c_
  %1 = call half @llvm.fma.f16(half %a, half %b, half %negc)
  ret half %1
}

define half @fnmadd_s(half %a, half %b, half %c) nounwind {
; CHECKIZFH-LABEL: fnmadd_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmv.h.x ft0, zero
; CHECKIZFH-NEXT:    fadd.h ft1, fa0, ft0
; CHECKIZFH-NEXT:    fadd.h ft0, fa2, ft0
; CHECKIZFH-NEXT:    fnmadd.h fa0, ft1, fa1, ft0
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fnmadd_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s1, a2
; RV32I-NEXT:    mv s0, a1
; RV32I-NEXT:    lui s3, 16
; RV32I-NEXT:    addi s3, s3, -1
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s2, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    lui s4, 524288
; RV32I-NEXT:    xor a0, a0, s4
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    xor a0, a0, s4
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    and a0, s2, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a2, a0
; RV32I-NEXT:    mv a0, s2
; RV32I-NEXT:    mv a1, s0
; RV32I-NEXT:    call fmaf@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fnmadd_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s4, 0(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s1, a2
; RV64I-NEXT:    mv s0, a1
; RV64I-NEXT:    lui s3, 16
; RV64I-NEXT:    addiw s3, s3, -1
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s2, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    lui s4, 524288
; RV64I-NEXT:    xor a0, a0, s4
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    xor a0, a0, s4
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    and a0, s2, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a2, a0
; RV64I-NEXT:    mv a0, s2
; RV64I-NEXT:    mv a1, s0
; RV64I-NEXT:    call fmaf@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s4, 0(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fnmadd_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa0
; CHECKIZFHMIN-NEXT:    fmv.w.x ft1, zero
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa2
; CHECKIZFHMIN-NEXT:    fadd.s ft1, ft2, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fneg.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, ft1
; CHECKIZFHMIN-NEXT:    fneg.s ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa1
; CHECKIZFHMIN-NEXT:    fmadd.s ft0, ft0, ft2, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %a_ = fadd half 0.0, %a
  %c_ = fadd half 0.0, %c
  %nega = fsub half -0.0, %a_
  %negc = fsub half -0.0, %c_
  %1 = call half @llvm.fma.f16(half %nega, half %b, half %negc)
  ret half %1
}

define half @fnmadd_s_2(half %a, half %b, half %c) nounwind {
; CHECKIZFH-LABEL: fnmadd_s_2:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmv.h.x ft0, zero
; CHECKIZFH-NEXT:    fadd.h ft1, fa1, ft0
; CHECKIZFH-NEXT:    fadd.h ft0, fa2, ft0
; CHECKIZFH-NEXT:    fnmadd.h fa0, ft1, fa0, ft0
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fnmadd_s_2:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s4, 8(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s1, a2
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    lui s3, 16
; RV32I-NEXT:    addi s3, s3, -1
; RV32I-NEXT:    and a0, a1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s2, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    lui s4, 524288
; RV32I-NEXT:    xor a0, a0, s4
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    xor a0, a0, s4
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    and a0, s2, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a2, a0
; RV32I-NEXT:    mv a0, s0
; RV32I-NEXT:    mv a1, s2
; RV32I-NEXT:    call fmaf@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s4, 8(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fnmadd_s_2:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s4, 0(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s1, a2
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    lui s3, 16
; RV64I-NEXT:    addiw s3, s3, -1
; RV64I-NEXT:    and a0, a1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s2, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    lui s4, 524288
; RV64I-NEXT:    xor a0, a0, s4
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    xor a0, a0, s4
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    and a0, s2, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a2, a0
; RV64I-NEXT:    mv a0, s0
; RV64I-NEXT:    mv a1, s2
; RV64I-NEXT:    call fmaf@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s4, 0(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fnmadd_s_2:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa1
; CHECKIZFHMIN-NEXT:    fmv.w.x ft1, zero
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa2
; CHECKIZFHMIN-NEXT:    fadd.s ft1, ft2, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fneg.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, ft1
; CHECKIZFHMIN-NEXT:    fneg.s ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa0
; CHECKIZFHMIN-NEXT:    fmadd.s ft0, ft2, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %b_ = fadd half 0.0, %b
  %c_ = fadd half 0.0, %c
  %negb = fsub half -0.0, %b_
  %negc = fsub half -0.0, %c_
  %1 = call half @llvm.fma.f16(half %a, half %negb, half %negc)
  ret half %1
}

define half @fnmadd_s_3(half %a, half %b, half %c) nounwind {
; RV32IZFH-LABEL: fnmadd_s_3:
; RV32IZFH:       # %bb.0:
; RV32IZFH-NEXT:    fmadd.h ft0, fa0, fa1, fa2
; RV32IZFH-NEXT:    fneg.h fa0, ft0
; RV32IZFH-NEXT:    ret
;
; RV64IZFH-LABEL: fnmadd_s_3:
; RV64IZFH:       # %bb.0:
; RV64IZFH-NEXT:    fmadd.h ft0, fa0, fa1, fa2
; RV64IZFH-NEXT:    fneg.h fa0, ft0
; RV64IZFH-NEXT:    ret
;
; CHECKIZFH-LABEL: fnmadd_s_3:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmadd.h ft0, fa0, fa1, fa2
; CHECKIZFH-NEXT:    fneg.h fa0, ft0
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fnmadd_s_3:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a2
; RV32I-NEXT:    mv s1, a1
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s3, a1, -1
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a2, a0
; RV32I-NEXT:    mv a0, s2
; RV32I-NEXT:    mv a1, s1
; RV32I-NEXT:    call fmaf@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lui a1, 1048568
; RV32I-NEXT:    xor a0, a0, a1
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fnmadd_s_3:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a2
; RV64I-NEXT:    mv s1, a1
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s3, a1, -1
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a2, a0
; RV64I-NEXT:    mv a0, s2
; RV64I-NEXT:    mv a1, s1
; RV64I-NEXT:    call fmaf@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    lui a1, 1048568
; RV64I-NEXT:    xor a0, a0, a1
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fnmadd_s_3:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa2
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa0
; CHECKIZFHMIN-NEXT:    fmadd.s ft0, ft2, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fneg.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %1 = call half @llvm.fma.f16(half %a, half %b, half %c)
  %neg = fneg half %1
  ret half %neg
}


define half @fnmadd_nsz(half %a, half %b, half %c) nounwind {
; RV32IZFH-LABEL: fnmadd_nsz:
; RV32IZFH:       # %bb.0:
; RV32IZFH-NEXT:    fnmadd.h fa0, fa0, fa1, fa2
; RV32IZFH-NEXT:    ret
;
; RV64IZFH-LABEL: fnmadd_nsz:
; RV64IZFH:       # %bb.0:
; RV64IZFH-NEXT:    fnmadd.h fa0, fa0, fa1, fa2
; RV64IZFH-NEXT:    ret
;
; CHECKIZFH-LABEL: fnmadd_nsz:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fnmadd.h fa0, fa0, fa1, fa2
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fnmadd_nsz:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a2
; RV32I-NEXT:    mv s1, a1
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s3, a1, -1
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a2, a0
; RV32I-NEXT:    mv a0, s2
; RV32I-NEXT:    mv a1, s1
; RV32I-NEXT:    call fmaf@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lui a1, 1048568
; RV32I-NEXT:    xor a0, a0, a1
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fnmadd_nsz:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a2
; RV64I-NEXT:    mv s1, a1
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s3, a1, -1
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a2, a0
; RV64I-NEXT:    mv a0, s2
; RV64I-NEXT:    mv a1, s1
; RV64I-NEXT:    call fmaf@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    lui a1, 1048568
; RV64I-NEXT:    xor a0, a0, a1
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fnmadd_nsz:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa2
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa0
; CHECKIZFHMIN-NEXT:    fmadd.s ft0, ft2, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fneg.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %1 = call nsz half @llvm.fma.f16(half %a, half %b, half %c)
  %neg = fneg nsz half %1
  ret half %neg
}

define half @fnmsub_s(half %a, half %b, half %c) nounwind {
; CHECKIZFH-LABEL: fnmsub_s:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmv.h.x ft0, zero
; CHECKIZFH-NEXT:    fadd.h ft0, fa0, ft0
; CHECKIZFH-NEXT:    fnmsub.h fa0, ft0, fa1, fa2
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fnmsub_s:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a2
; RV32I-NEXT:    mv s1, a1
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s3, a1, -1
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    lui a1, 524288
; RV32I-NEXT:    xor a0, a0, a1
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    and a0, s2, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, s1
; RV32I-NEXT:    mv a2, s0
; RV32I-NEXT:    call fmaf@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fnmsub_s:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a2
; RV64I-NEXT:    mv s1, a1
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s3, a1, -1
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    lui a1, 524288
; RV64I-NEXT:    xor a0, a0, a1
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    and a0, s2, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, s1
; RV64I-NEXT:    mv a2, s0
; RV64I-NEXT:    call fmaf@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fnmsub_s:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa0
; CHECKIZFHMIN-NEXT:    fmv.w.x ft1, zero
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fneg.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa2
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa1
; CHECKIZFHMIN-NEXT:    fmadd.s ft0, ft0, ft2, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %a_ = fadd half 0.0, %a
  %nega = fsub half -0.0, %a_
  %1 = call half @llvm.fma.f16(half %nega, half %b, half %c)
  ret half %1
}

define half @fnmsub_s_2(half %a, half %b, half %c) nounwind {
; CHECKIZFH-LABEL: fnmsub_s_2:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmv.h.x ft0, zero
; CHECKIZFH-NEXT:    fadd.h ft0, fa1, ft0
; CHECKIZFH-NEXT:    fnmsub.h fa0, ft0, fa0, fa2
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fnmsub_s_2:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a2
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    lui a0, 16
; RV32I-NEXT:    addi s3, a0, -1
; RV32I-NEXT:    and a0, a1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    lui a1, 524288
; RV32I-NEXT:    xor a0, a0, a1
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    and a0, s2, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s1
; RV32I-NEXT:    mv a2, s0
; RV32I-NEXT:    call fmaf@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fnmsub_s_2:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a2
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    lui a0, 16
; RV64I-NEXT:    addiw s3, a0, -1
; RV64I-NEXT:    and a0, a1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    lui a1, 524288
; RV64I-NEXT:    xor a0, a0, a1
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    and a0, s2, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s1
; RV64I-NEXT:    mv a2, s0
; RV64I-NEXT:    call fmaf@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fnmsub_s_2:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa1
; CHECKIZFHMIN-NEXT:    fmv.w.x ft1, zero
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fneg.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa2
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa0
; CHECKIZFHMIN-NEXT:    fmadd.s ft0, ft2, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %b_ = fadd half 0.0, %b
  %negb = fsub half -0.0, %b_
  %1 = call half @llvm.fma.f16(half %a, half %negb, half %c)
  ret half %1
}

define half @fmadd_s_contract(half %a, half %b, half %c) nounwind {
; CHECKIZFH-LABEL: fmadd_s_contract:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmadd.h fa0, fa0, fa1, fa2
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fmadd_s_contract:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a2
; RV32I-NEXT:    mv s1, a1
; RV32I-NEXT:    lui a1, 16
; RV32I-NEXT:    addi s3, a1, -1
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s2
; RV32I-NEXT:    call __mulsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, s0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fmadd_s_contract:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a2
; RV64I-NEXT:    mv s1, a1
; RV64I-NEXT:    lui a1, 16
; RV64I-NEXT:    addiw s3, a1, -1
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s2
; RV64I-NEXT:    call __mulsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, s0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fmadd_s_contract:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa0
; CHECKIZFHMIN-NEXT:    fmul.s ft0, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa2
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %1 = fmul contract half %a, %b
  %2 = fadd contract half %1, %c
  ret half %2
}

define half @fmsub_s_contract(half %a, half %b, half %c) nounwind {
; CHECKIZFH-LABEL: fmsub_s_contract:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmv.h.x ft0, zero
; CHECKIZFH-NEXT:    fadd.h ft0, fa2, ft0
; CHECKIZFH-NEXT:    fmsub.h fa0, fa0, fa1, ft0
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fmsub_s_contract:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a1
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    lui a0, 16
; RV32I-NEXT:    addi s3, a0, -1
; RV32I-NEXT:    and a0, a2, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s1
; RV32I-NEXT:    call __mulsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    and a0, s2, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s0
; RV32I-NEXT:    call __subsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fmsub_s_contract:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a1
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    lui a0, 16
; RV64I-NEXT:    addiw s3, a0, -1
; RV64I-NEXT:    and a0, a2, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s1
; RV64I-NEXT:    call __mulsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    and a0, s2, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s0
; RV64I-NEXT:    call __subsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fmsub_s_contract:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa2
; CHECKIZFHMIN-NEXT:    fmv.w.x ft1, zero
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa0
; CHECKIZFHMIN-NEXT:    fmul.s ft1, ft2, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, ft1
; CHECKIZFHMIN-NEXT:    fsub.s ft0, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %c_ = fadd half 0.0, %c ; avoid negation using xor
  %1 = fmul contract half %a, %b
  %2 = fsub contract half %1, %c_
  ret half %2
}

define half @fnmadd_s_contract(half %a, half %b, half %c) nounwind {
; CHECKIZFH-LABEL: fnmadd_s_contract:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmv.h.x ft0, zero
; CHECKIZFH-NEXT:    fadd.h ft1, fa0, ft0
; CHECKIZFH-NEXT:    fadd.h ft2, fa1, ft0
; CHECKIZFH-NEXT:    fadd.h ft0, fa2, ft0
; CHECKIZFH-NEXT:    fnmadd.h fa0, ft1, ft2, ft0
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fnmadd_s_contract:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a2
; RV32I-NEXT:    mv s1, a1
; RV32I-NEXT:    lui s3, 16
; RV32I-NEXT:    addi s3, s3, -1
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    and a0, s2, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s2
; RV32I-NEXT:    call __mulsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    lui a1, 524288
; RV32I-NEXT:    xor a0, a0, a1
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, s0
; RV32I-NEXT:    call __subsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fnmadd_s_contract:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a2
; RV64I-NEXT:    mv s1, a1
; RV64I-NEXT:    lui s3, 16
; RV64I-NEXT:    addiw s3, s3, -1
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    and a0, s2, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s2
; RV64I-NEXT:    call __mulsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    lui a1, 524288
; RV64I-NEXT:    xor a0, a0, a1
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, s0
; RV64I-NEXT:    call __subsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fnmadd_s_contract:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa0
; CHECKIZFHMIN-NEXT:    fmv.w.x ft1, zero
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa1
; CHECKIZFHMIN-NEXT:    fadd.s ft2, ft2, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft2, ft2
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft3, fa2
; CHECKIZFHMIN-NEXT:    fadd.s ft1, ft3, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, ft2
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fmul.s ft0, ft0, ft2
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fneg.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, ft1
; CHECKIZFHMIN-NEXT:    fsub.s ft0, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %a_ = fadd half 0.0, %a ; avoid negation using xor
  %b_ = fadd half 0.0, %b ; avoid negation using xor
  %c_ = fadd half 0.0, %c ; avoid negation using xor
  %1 = fmul contract half %a_, %b_
  %2 = fneg half %1
  %3 = fsub contract half %2, %c_
  ret half %3
}

define half @fnmsub_s_contract(half %a, half %b, half %c) nounwind {
; CHECKIZFH-LABEL: fnmsub_s_contract:
; CHECKIZFH:       # %bb.0:
; CHECKIZFH-NEXT:    fmv.h.x ft0, zero
; CHECKIZFH-NEXT:    fadd.h ft1, fa0, ft0
; CHECKIZFH-NEXT:    fadd.h ft0, fa1, ft0
; CHECKIZFH-NEXT:    fnmsub.h fa0, ft1, ft0, fa2
; CHECKIZFH-NEXT:    ret
;
; RV32I-LABEL: fnmsub_s_contract:
; RV32I:       # %bb.0:
; RV32I-NEXT:    addi sp, sp, -32
; RV32I-NEXT:    sw ra, 28(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s0, 24(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s1, 20(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s2, 16(sp) # 4-byte Folded Spill
; RV32I-NEXT:    sw s3, 12(sp) # 4-byte Folded Spill
; RV32I-NEXT:    mv s0, a2
; RV32I-NEXT:    mv s1, a1
; RV32I-NEXT:    lui s3, 16
; RV32I-NEXT:    addi s3, s3, -1
; RV32I-NEXT:    and a0, a0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    li a1, 0
; RV32I-NEXT:    call __addsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s2, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s2, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s2
; RV32I-NEXT:    call __mulsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    mv s1, a0
; RV32I-NEXT:    and a0, s0, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv s0, a0
; RV32I-NEXT:    and a0, s1, s3
; RV32I-NEXT:    call __extendhfsf2@plt
; RV32I-NEXT:    mv a1, a0
; RV32I-NEXT:    mv a0, s0
; RV32I-NEXT:    call __subsf3@plt
; RV32I-NEXT:    call __truncsfhf2@plt
; RV32I-NEXT:    lw ra, 28(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s0, 24(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s1, 20(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s2, 16(sp) # 4-byte Folded Reload
; RV32I-NEXT:    lw s3, 12(sp) # 4-byte Folded Reload
; RV32I-NEXT:    addi sp, sp, 32
; RV32I-NEXT:    ret
;
; RV64I-LABEL: fnmsub_s_contract:
; RV64I:       # %bb.0:
; RV64I-NEXT:    addi sp, sp, -48
; RV64I-NEXT:    sd ra, 40(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s0, 32(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s1, 24(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s2, 16(sp) # 8-byte Folded Spill
; RV64I-NEXT:    sd s3, 8(sp) # 8-byte Folded Spill
; RV64I-NEXT:    mv s0, a2
; RV64I-NEXT:    mv s1, a1
; RV64I-NEXT:    lui s3, 16
; RV64I-NEXT:    addiw s3, s3, -1
; RV64I-NEXT:    and a0, a0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    li a1, 0
; RV64I-NEXT:    call __addsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s2, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s2, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s2
; RV64I-NEXT:    call __mulsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    mv s1, a0
; RV64I-NEXT:    and a0, s0, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv s0, a0
; RV64I-NEXT:    and a0, s1, s3
; RV64I-NEXT:    call __extendhfsf2@plt
; RV64I-NEXT:    mv a1, a0
; RV64I-NEXT:    mv a0, s0
; RV64I-NEXT:    call __subsf3@plt
; RV64I-NEXT:    call __truncsfhf2@plt
; RV64I-NEXT:    ld ra, 40(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s0, 32(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s1, 24(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s2, 16(sp) # 8-byte Folded Reload
; RV64I-NEXT:    ld s3, 8(sp) # 8-byte Folded Reload
; RV64I-NEXT:    addi sp, sp, 48
; RV64I-NEXT:    ret
;
; CHECKIZFHMIN-LABEL: fnmsub_s_contract:
; CHECKIZFHMIN:       # %bb.0:
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, fa0
; CHECKIZFHMIN-NEXT:    fmv.w.x ft1, zero
; CHECKIZFHMIN-NEXT:    fadd.s ft0, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft2, fa1
; CHECKIZFHMIN-NEXT:    fadd.s ft1, ft2, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, ft1
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fmul.s ft0, ft0, ft1
; CHECKIZFHMIN-NEXT:    fcvt.h.s ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft0, ft0
; CHECKIZFHMIN-NEXT:    fcvt.s.h ft1, fa2
; CHECKIZFHMIN-NEXT:    fsub.s ft0, ft1, ft0
; CHECKIZFHMIN-NEXT:    fcvt.h.s fa0, ft0
; CHECKIZFHMIN-NEXT:    ret
  %a_ = fadd half 0.0, %a ; avoid negation using xor
  %b_ = fadd half 0.0, %b ; avoid negation using xor
  %1 = fmul contract half %a_, %b_
  %2 = fsub contract half %c, %1
  ret half %2
}