; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc --mtriple=loongarch64 -mattr=+d,-lsx      < %s | FileCheck %s --check-prefix=LA64
; RUN: llc --mtriple=loongarch32 -mattr=+d           < %s | FileCheck %s --check-prefix=LA32
; RUN: llc --mtriple=loongarch64 -mattr=+d,+lsx      < %s | FileCheck %s --check-prefix=LSX
; RUN: llc --mtriple=loongarch64 -mattr=+d,+lasx     < %s | FileCheck %s --check-prefix=LASX

; ISSUE-186645 - Merge 64-bit operations as 128-bit operations while noimplicitfloat
; attribute is set are not legal. However, we can generate a paired 64-bit loads and
; stores, without using floating point registers.

define void @memcpy16_noimplicitfloat(ptr %p0, ptr %p1) noimplicitfloat {
; LA64-LABEL: memcpy16_noimplicitfloat:
; LA64:       # %bb.0:
; LA64-NEXT:    ld.d $a2, $a1, 8
; LA64-NEXT:    st.d $a2, $a0, 8
; LA64-NEXT:    ld.d $a1, $a1, 0
; LA64-NEXT:    st.d $a1, $a0, 0
; LA64-NEXT:    ret
;
; LA32-LABEL: memcpy16_noimplicitfloat:
; LA32:       # %bb.0:
; LA32-NEXT:    ld.w $a2, $a1, 12
; LA32-NEXT:    st.w $a2, $a0, 12
; LA32-NEXT:    ld.w $a2, $a1, 8
; LA32-NEXT:    st.w $a2, $a0, 8
; LA32-NEXT:    ld.w $a2, $a1, 4
; LA32-NEXT:    st.w $a2, $a0, 4
; LA32-NEXT:    ld.w $a1, $a1, 0
; LA32-NEXT:    st.w $a1, $a0, 0
; LA32-NEXT:    ret
;
; LSX-LABEL: memcpy16_noimplicitfloat:
; LSX:       # %bb.0:
; LSX-NEXT:    ld.d $a2, $a1, 8
; LSX-NEXT:    st.d $a2, $a0, 8
; LSX-NEXT:    ld.d $a1, $a1, 0
; LSX-NEXT:    st.d $a1, $a0, 0
; LSX-NEXT:    ret
;
; LASX-LABEL: memcpy16_noimplicitfloat:
; LASX:       # %bb.0:
; LASX-NEXT:    ld.d $a2, $a1, 8
; LASX-NEXT:    st.d $a2, $a0, 8
; LASX-NEXT:    ld.d $a1, $a1, 0
; LASX-NEXT:    st.d $a1, $a0, 0
; LASX-NEXT:    ret
    call void @llvm.memcpy.p0.p0.i64(ptr align 4 %p0, ptr align 4 %p1, i64 16, i1 false)
    ret void
}

define void @memcpy16_aligned(ptr %p0, ptr %p1) {
; LA64-LABEL: memcpy16_aligned:
; LA64:       # %bb.0:
; LA64-NEXT:    ld.d $a2, $a1, 8
; LA64-NEXT:    st.d $a2, $a0, 8
; LA64-NEXT:    ld.d $a1, $a1, 0
; LA64-NEXT:    st.d $a1, $a0, 0
; LA64-NEXT:    ret
;
; LA32-LABEL: memcpy16_aligned:
; LA32:       # %bb.0:
; LA32-NEXT:    ld.w $a2, $a1, 12
; LA32-NEXT:    st.w $a2, $a0, 12
; LA32-NEXT:    ld.w $a2, $a1, 8
; LA32-NEXT:    st.w $a2, $a0, 8
; LA32-NEXT:    ld.w $a2, $a1, 4
; LA32-NEXT:    st.w $a2, $a0, 4
; LA32-NEXT:    ld.w $a1, $a1, 0
; LA32-NEXT:    st.w $a1, $a0, 0
; LA32-NEXT:    ret
;
; LSX-LABEL: memcpy16_aligned:
; LSX:       # %bb.0:
; LSX-NEXT:    vld $vr0, $a1, 0
; LSX-NEXT:    vst $vr0, $a0, 0
; LSX-NEXT:    ret
;
; LASX-LABEL: memcpy16_aligned:
; LASX:       # %bb.0:
; LASX-NEXT:    vld $vr0, $a1, 0
; LASX-NEXT:    vst $vr0, $a0, 0
; LASX-NEXT:    ret
    call void @llvm.memcpy.p0.p0.i64(ptr align 4 %p0, ptr align 4 %p1, i64 16, i1 false)
    ret void
}

define void @memcpy32_aligned(ptr %p0, ptr %p1) {
; LA64-LABEL: memcpy32_aligned:
; LA64:       # %bb.0:
; LA64-NEXT:    ld.d $a2, $a1, 24
; LA64-NEXT:    st.d $a2, $a0, 24
; LA64-NEXT:    ld.d $a2, $a1, 16
; LA64-NEXT:    st.d $a2, $a0, 16
; LA64-NEXT:    ld.d $a2, $a1, 8
; LA64-NEXT:    st.d $a2, $a0, 8
; LA64-NEXT:    ld.d $a1, $a1, 0
; LA64-NEXT:    st.d $a1, $a0, 0
; LA64-NEXT:    ret
;
; LA32-LABEL: memcpy32_aligned:
; LA32:       # %bb.0:
; LA32-NEXT:    ld.w $a2, $a1, 28
; LA32-NEXT:    st.w $a2, $a0, 28
; LA32-NEXT:    ld.w $a2, $a1, 24
; LA32-NEXT:    st.w $a2, $a0, 24
; LA32-NEXT:    ld.w $a2, $a1, 20
; LA32-NEXT:    st.w $a2, $a0, 20
; LA32-NEXT:    ld.w $a2, $a1, 16
; LA32-NEXT:    st.w $a2, $a0, 16
; LA32-NEXT:    ld.w $a2, $a1, 12
; LA32-NEXT:    st.w $a2, $a0, 12
; LA32-NEXT:    ld.w $a2, $a1, 8
; LA32-NEXT:    st.w $a2, $a0, 8
; LA32-NEXT:    ld.w $a2, $a1, 4
; LA32-NEXT:    st.w $a2, $a0, 4
; LA32-NEXT:    ld.w $a1, $a1, 0
; LA32-NEXT:    st.w $a1, $a0, 0
; LA32-NEXT:    ret
;
; LSX-LABEL: memcpy32_aligned:
; LSX:       # %bb.0:
; LSX-NEXT:    vld $vr0, $a1, 16
; LSX-NEXT:    vst $vr0, $a0, 16
; LSX-NEXT:    vld $vr0, $a1, 0
; LSX-NEXT:    vst $vr0, $a0, 0
; LSX-NEXT:    ret
;
; LASX-LABEL: memcpy32_aligned:
; LASX:       # %bb.0:
; LASX-NEXT:    xvld $xr0, $a1, 0
; LASX-NEXT:    xvst $xr0, $a0, 0
; LASX-NEXT:    ret
    call void @llvm.memcpy.p0.p0.i64(ptr align 4 %p0, ptr align 4 %p1, i64 32, i1 false)
    ret void
}

declare void @llvm.memcpy.p0.p0.i64(ptr, ptr, i64, i1)