1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 < %s | FileCheck %s -check-prefix=MUBUF
3 ; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -amdgpu-enable-flat-scratch < %s | FileCheck %s -check-prefix=FLATSCR
5 ; Make sure there's no assertion from passing a 0 alignment value
6 define void @memcpy_fixed_align(i8 addrspace(5)* %dst, i8 addrspace(1)* %src) {
7 ; MUBUF-LABEL: memcpy_fixed_align:
9 ; MUBUF-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
10 ; MUBUF-NEXT: global_load_dword v0, v[1:2], off offset:36
11 ; MUBUF-NEXT: global_load_dword v11, v[1:2], off offset:32
12 ; MUBUF-NEXT: global_load_dwordx4 v[3:6], v[1:2], off offset:16
13 ; MUBUF-NEXT: global_load_dwordx4 v[7:10], v[1:2], off
14 ; MUBUF-NEXT: s_waitcnt vmcnt(3)
15 ; MUBUF-NEXT: buffer_store_dword v0, off, s[0:3], s32 offset:36
16 ; MUBUF-NEXT: s_waitcnt vmcnt(3)
17 ; MUBUF-NEXT: buffer_store_dword v11, off, s[0:3], s32 offset:32
18 ; MUBUF-NEXT: s_waitcnt vmcnt(3)
19 ; MUBUF-NEXT: buffer_store_dword v6, off, s[0:3], s32 offset:28
20 ; MUBUF-NEXT: buffer_store_dword v5, off, s[0:3], s32 offset:24
21 ; MUBUF-NEXT: buffer_store_dword v4, off, s[0:3], s32 offset:20
22 ; MUBUF-NEXT: buffer_store_dword v3, off, s[0:3], s32 offset:16
23 ; MUBUF-NEXT: s_waitcnt vmcnt(6)
24 ; MUBUF-NEXT: buffer_store_dword v10, off, s[0:3], s32 offset:12
25 ; MUBUF-NEXT: buffer_store_dword v9, off, s[0:3], s32 offset:8
26 ; MUBUF-NEXT: buffer_store_dword v8, off, s[0:3], s32 offset:4
27 ; MUBUF-NEXT: buffer_store_dword v7, off, s[0:3], s32
28 ; MUBUF-NEXT: s_waitcnt vmcnt(0)
29 ; MUBUF-NEXT: s_setpc_b64 s[30:31]
31 ; FLATSCR-LABEL: memcpy_fixed_align:
33 ; FLATSCR-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
34 ; FLATSCR-NEXT: global_load_dwordx2 v[11:12], v[1:2], off offset:32
35 ; FLATSCR-NEXT: global_load_dwordx4 v[3:6], v[1:2], off offset:16
36 ; FLATSCR-NEXT: global_load_dwordx4 v[7:10], v[1:2], off
37 ; FLATSCR-NEXT: s_waitcnt vmcnt(2)
38 ; FLATSCR-NEXT: scratch_store_dwordx2 off, v[11:12], s32 offset:32
39 ; FLATSCR-NEXT: s_waitcnt vmcnt(2)
40 ; FLATSCR-NEXT: scratch_store_dwordx4 off, v[3:6], s32 offset:16
41 ; FLATSCR-NEXT: s_waitcnt vmcnt(2)
42 ; FLATSCR-NEXT: scratch_store_dwordx4 off, v[7:10], s32
43 ; FLATSCR-NEXT: s_waitcnt vmcnt(0)
44 ; FLATSCR-NEXT: s_setpc_b64 s[30:31]
45 %alloca = alloca [40 x i8], addrspace(5)
46 %cast = bitcast [40 x i8] addrspace(5)* %alloca to i8 addrspace(5)*
47 call void @llvm.memcpy.p5i8.p1i8.i64(i8 addrspace(5)* align 4 dereferenceable(40) %cast, i8 addrspace(1)* align 4 dereferenceable(40) %src, i64 40, i1 false)
51 declare void @llvm.memcpy.p5i8.p1i8.i64(i8 addrspace(5)* noalias nocapture writeonly, i8 addrspace(1)* noalias nocapture readonly, i64, i1 immarg) #0
53 attributes #0 = { argmemonly nounwind willreturn }