xref: /llvm-project/llvm/test/CodeGen/AArch64/ldst-paired-aliasing.ll (revision 6e54fccede402c9ed0e8038aa258a99c5a2773e5)
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2; RUN: llc -mcpu cortex-a53 < %s | FileCheck %s
3target datalayout = "e-m:e-i64:64-i128:128-n8:16:32:64-S128"
4target triple = "aarch64--linux-gnu"
5
6declare void @f(ptr, ptr)
7declare void @f2(ptr, ptr)
8declare void @_Z5setupv()
9declare void @llvm.memset.p0.i64(ptr nocapture, i8, i64, i1) #3
10
11define i32 @main() local_unnamed_addr #1 {
12; Make sure the stores happen in the correct order (the exact instructions could change).
13; CHECK-LABEL: main:
14; CHECK:       // %bb.0: // %for.body.lr.ph.i.i.i.i.i.i63
15; CHECK-NEXT:    sub sp, sp, #112
16; CHECK-NEXT:    str x30, [sp, #96] // 8-byte Folded Spill
17; CHECK-NEXT:    .cfi_def_cfa_offset 112
18; CHECK-NEXT:    .cfi_offset w30, -16
19; CHECK-NEXT:    bl _Z5setupv
20; CHECK-NEXT:    movi v0.4s, #1
21; CHECK-NEXT:    mov w9, #1
22; CHECK-NEXT:    add x0, sp, #48
23; CHECK-NEXT:    mov x1, sp
24; CHECK-NEXT:    str xzr, [sp, #80]
25; CHECK-NEXT:    str w9, [sp, #80]
26; CHECK-NEXT:    stp q0, q0, [sp, #48]
27; CHECK-NEXT:    ldr w8, [sp, #48]
28; CHECK-NEXT:    cmp w8, #1
29; CHECK-NEXT:    b.ne .LBB0_2
30; CHECK-NEXT:  // %bb.1: // %for.inc
31; CHECK-NEXT:    bl f
32; CHECK-NEXT:    b .LBB0_3
33; CHECK-NEXT:  .LBB0_2: // %if.then
34; CHECK-NEXT:    bl f2
35; CHECK-NEXT:  .LBB0_3: // %common.ret
36; CHECK-NEXT:    ldr x30, [sp, #96] // 8-byte Folded Reload
37; CHECK-NEXT:    mov w0, wzr
38; CHECK-NEXT:    add sp, sp, #112
39; CHECK-NEXT:    ret
40
41
42for.body.lr.ph.i.i.i.i.i.i63:
43  %b1 = alloca [10 x i32], align 16
44  %b2 = alloca [10 x i32], align 16
45  tail call void @_Z5setupv()
46  %x2 = getelementptr inbounds [10 x i32], ptr %b1, i64 0, i64 6
47  call void @llvm.memset.p0.i64(ptr align 8 %x2, i8 0, i64 16, i1 false)
48  store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr %b1, align 16
49  %incdec.ptr.i7.i.i.i.i.i.i64.3 = getelementptr inbounds [10 x i32], ptr %b1, i64 0, i64 4
50  store <4 x i32> <i32 1, i32 1, i32 1, i32 1>, ptr %incdec.ptr.i7.i.i.i.i.i.i64.3, align 16
51  %incdec.ptr.i7.i.i.i.i.i.i64.7 = getelementptr inbounds [10 x i32], ptr %b1, i64 0, i64 8
52  store i32 1, ptr %incdec.ptr.i7.i.i.i.i.i.i64.7, align 16
53  %x6 = load i32, ptr %b1, align 16
54  %cmp6 = icmp eq i32 %x6, 1
55  br i1 %cmp6, label %for.inc, label %if.then
56
57for.inc:
58  call void @f(ptr %b1, ptr %b2)
59  ret i32 0
60
61if.then:
62  call void @f2(ptr %b1, ptr %b2)
63  ret i32 0
64}
65