1 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
2 ; RUN: opt -S -memcpyopt < %s | FileCheck %s --check-prefixes=CHECK,LIBCALLS
3 ; RUN: opt -S -memcpyopt -mtriple=amdgcn-- < %s | FileCheck %s --check-prefixes=CHECK,NO-LIBCALLS
4 ; RUN: opt -S -memcpyopt -mtriple=amdgcn-- -enable-memcpyopt-without-libcalls < %s \
5 ; RUN: | FileCheck %s --check-prefixes=CHECK,LIBCALLS
7 ; REQUIRES: amdgpu-registered-target
9 define void @dont_create_memset(ptr %p) {
10 ; LIBCALLS-LABEL: @dont_create_memset(
11 ; LIBCALLS-NEXT: [[P1:%.*]] = getelementptr i32, ptr [[P:%.*]], i64 1
12 ; LIBCALLS-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[P]], i64 2
13 ; LIBCALLS-NEXT: [[P3:%.*]] = getelementptr i32, ptr [[P]], i64 3
14 ; LIBCALLS-NEXT: call void @llvm.memset.p0.i64(ptr align 4 [[P]], i8 0, i64 16, i1 false)
15 ; LIBCALLS-NEXT: ret void
17 ; NO-LIBCALLS-LABEL: @dont_create_memset(
18 ; NO-LIBCALLS-NEXT: store i32 0, ptr [[P:%.*]], align 4
19 ; NO-LIBCALLS-NEXT: [[P1:%.*]] = getelementptr i32, ptr [[P]], i64 1
20 ; NO-LIBCALLS-NEXT: store i32 0, ptr [[P1]], align 4
21 ; NO-LIBCALLS-NEXT: [[P2:%.*]] = getelementptr i32, ptr [[P]], i64 2
22 ; NO-LIBCALLS-NEXT: store i32 0, ptr [[P2]], align 4
23 ; NO-LIBCALLS-NEXT: [[P3:%.*]] = getelementptr i32, ptr [[P]], i64 3
24 ; NO-LIBCALLS-NEXT: store i32 0, ptr [[P3]], align 4
25 ; NO-LIBCALLS-NEXT: ret void
28 %p1 = getelementptr i32, ptr %p, i64 1
30 %p2 = getelementptr i32, ptr %p, i64 2
32 %p3 = getelementptr i32, ptr %p, i64 3
39 define void @dont_create_memcpy(ptr %p1, ptr %p2) {
40 ; LIBCALLS-LABEL: @dont_create_memcpy(
41 ; LIBCALLS-NEXT: call void @llvm.memmove.p0.p0.i64(ptr align 4 [[P2:%.*]], ptr align 4 [[P1:%.*]], i64 8, i1 false)
42 ; LIBCALLS-NEXT: ret void
44 ; NO-LIBCALLS-LABEL: @dont_create_memcpy(
45 ; NO-LIBCALLS-NEXT: [[V:%.*]] = load [[TY:%.*]], ptr [[P1:%.*]], align 4
46 ; NO-LIBCALLS-NEXT: store [[TY]] [[V]], ptr [[P2:%.*]], align 4
47 ; NO-LIBCALLS-NEXT: ret void
49 %v = load %ty, ptr %p1
54 define void @forward_memcpy(ptr noalias %p1, ptr noalias %p2, ptr noalias %p3) {
55 ; CHECK-LABEL: @forward_memcpy(
56 ; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr [[P2:%.*]], ptr [[P1:%.*]], i64 16, i1 false)
57 ; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr [[P3:%.*]], ptr [[P1]], i64 16, i1 false)
58 ; CHECK-NEXT: ret void
60 call void @llvm.memcpy.p0.p0.i64(ptr %p2, ptr %p1, i64 16, i1 false)
61 call void @llvm.memcpy.p0.p0.i64(ptr %p3, ptr %p2, i64 16, i1 false)
65 declare void @llvm.memcpy.p0.p0.i64(ptr, ptr, i64, i1)