1; RUN: opt < %s -basicaa -dse -S | FileCheck %s
2target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
3
4%struct.vec2 = type { <4 x i32>, <4 x i32> }
5%struct.vec2plusi = type { <4 x i32>, <4 x i32>, i32 }
6
7@glob1 = global %struct.vec2 zeroinitializer, align 16
8@glob2 = global %struct.vec2plusi zeroinitializer, align 16
9
10define void @write24to28(i32* nocapture %p) nounwind uwtable ssp {
11; CHECK-LABEL: @write24to28(
12entry:
13  %arrayidx0 = getelementptr inbounds i32, i32* %p, i64 1
14  %p3 = bitcast i32* %arrayidx0 to i8*
15; CHECK: call void @llvm.memset.p0i8.i64(i8* %p3, i8 0, i64 24, i32 4, i1 false)
16  call void @llvm.memset.p0i8.i64(i8* %p3, i8 0, i64 28, i32 4, i1 false)
17  %arrayidx1 = getelementptr inbounds i32, i32* %p, i64 7
18  store i32 1, i32* %arrayidx1, align 4
19  ret void
20}
21
22define void @write28to32(i32* nocapture %p) nounwind uwtable ssp {
23; CHECK-LABEL: @write28to32(
24entry:
25  %p3 = bitcast i32* %p to i8*
26; CHECK: call void @llvm.memset.p0i8.i64(i8* %p3, i8 0, i64 28, i32 4, i1 false)
27  call void @llvm.memset.p0i8.i64(i8* %p3, i8 0, i64 32, i32 4, i1 false)
28  %arrayidx1 = getelementptr inbounds i32, i32* %p, i64 7
29  store i32 1, i32* %arrayidx1, align 4
30  ret void
31}
32
33define void @dontwrite28to32memset(i32* nocapture %p) nounwind uwtable ssp {
34; CHECK-LABEL: @dontwrite28to32memset(
35entry:
36  %p3 = bitcast i32* %p to i8*
37; CHECK: call void @llvm.memset.p0i8.i64(i8* %p3, i8 0, i64 32, i32 16, i1 false)
38  call void @llvm.memset.p0i8.i64(i8* %p3, i8 0, i64 32, i32 16, i1 false)
39  %arrayidx1 = getelementptr inbounds i32, i32* %p, i64 7
40  store i32 1, i32* %arrayidx1, align 4
41  ret void
42}
43
44define void @write32to36(%struct.vec2plusi* nocapture %p) nounwind uwtable ssp {
45; CHECK-LABEL: @write32to36(
46entry:
47  %0 = bitcast %struct.vec2plusi* %p to i8*
48; CHECK: tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* %0, i8* bitcast (%struct.vec2plusi* @glob2 to i8*), i64 32, i32 16, i1 false)
49  tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* %0, i8* bitcast (%struct.vec2plusi* @glob2 to i8*), i64 36, i32 16, i1 false)
50  %c = getelementptr inbounds %struct.vec2plusi, %struct.vec2plusi* %p, i64 0, i32 2
51  store i32 1, i32* %c, align 4
52  ret void
53}
54
55define void @write16to32(%struct.vec2* nocapture %p) nounwind uwtable ssp {
56; CHECK-LABEL: @write16to32(
57entry:
58  %0 = bitcast %struct.vec2* %p to i8*
59; CHECK: tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* %0, i8* bitcast (%struct.vec2* @glob1 to i8*), i64 16, i32 16, i1 false)
60  tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* %0, i8* bitcast (%struct.vec2* @glob1 to i8*), i64 32, i32 16, i1 false)
61  %c = getelementptr inbounds %struct.vec2, %struct.vec2* %p, i64 0, i32 1
62  store <4 x i32> <i32 1, i32 2, i32 3, i32 4>, <4 x i32>* %c, align 4
63  ret void
64}
65
66define void @dontwrite28to32memcpy(%struct.vec2* nocapture %p) nounwind uwtable ssp {
67; CHECK-LABEL: @dontwrite28to32memcpy(
68entry:
69  %0 = bitcast %struct.vec2* %p to i8*
70; CHECK: tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* %0, i8* bitcast (%struct.vec2* @glob1 to i8*), i64 32, i32 16, i1 false)
71  tail call void @llvm.memcpy.p0i8.p0i8.i64(i8* %0, i8* bitcast (%struct.vec2* @glob1 to i8*), i64 32, i32 16, i1 false)
72  %arrayidx1 = getelementptr inbounds %struct.vec2, %struct.vec2* %p, i64 0, i32 0, i64 7
73  store i32 1, i32* %arrayidx1, align 4
74  ret void
75}
76
77declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i32, i1) nounwind
78declare void @llvm.memset.p0i8.i64(i8* nocapture, i8, i64, i32, i1) nounwind
79
80%struct.trapframe = type { i64, i64, i64 }
81
82; bugzilla 11455 - make sure negative GEP's don't break this optimisation
83; CHECK-LABEL: @cpu_lwp_fork(
84define void @cpu_lwp_fork(%struct.trapframe* %md_regs, i64 %pcb_rsp0) nounwind uwtable noinline ssp {
85entry:
86  %0 = inttoptr i64 %pcb_rsp0 to %struct.trapframe*
87  %add.ptr = getelementptr inbounds %struct.trapframe, %struct.trapframe* %0, i64 -1
88  %1 = bitcast %struct.trapframe* %add.ptr to i8*
89  %2 = bitcast %struct.trapframe* %md_regs to i8*
90; CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* %1, i8* %2, i64 24, i32 1, i1 false)
91  call void @llvm.memcpy.p0i8.p0i8.i64(i8* %1, i8* %2, i64 24, i32 1, i1 false)
92  %tf_trapno = getelementptr inbounds %struct.trapframe, %struct.trapframe* %0, i64 -1, i32 1
93  store i64 3, i64* %tf_trapno, align 8
94  ret void
95}
96