1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s | FileCheck %s
3 target 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"
4 target triple = "x86_64-pc-linux"
7 %0 = type <{ i64, i64, %1, %1, [21 x %2] }>
8 %1 = type <{ i64, i64, i64 }>
9 %2 = type <{ i32, i32, ptr addrspace(2) }>
10 %3 = type { ptr, ptr, ptr, ptr, i32 }
11 %4 = type <{ ptr, ptr, i32, i32, [4 x i64], [4 x i64], [4 x i64], [4 x i64], [4 x i64] }>
12 %5 = type <{ ptr, ptr, ptr }>
14 define void @foo(ptr nocapture %_stubArgs) nounwind {
16 ; CHECK: # %bb.0: # %entry
17 ; CHECK-NEXT: movq 48(%rdi), %rax
18 ; CHECK-NEXT: movl 64(%rdi), %ecx
19 ; CHECK-NEXT: movl $200, %esi
20 ; CHECK-NEXT: addl 68(%rdi), %esi
21 ; CHECK-NEXT: imull $46, %ecx, %edx
22 ; CHECK-NEXT: addq %rsi, %rdx
23 ; CHECK-NEXT: shlq $4, %rdx
24 ; CHECK-NEXT: imull $47, %ecx, %ecx
25 ; CHECK-NEXT: addq %rsi, %rcx
26 ; CHECK-NEXT: shlq $4, %rcx
27 ; CHECK-NEXT: cmpl $0, (%rdi)
28 ; CHECK-NEXT: jne .LBB0_1
29 ; CHECK-NEXT: # %bb.2: # %entry
30 ; CHECK-NEXT: xorps %xmm0, %xmm0
31 ; CHECK-NEXT: jmp .LBB0_3
32 ; CHECK-NEXT: .LBB0_1:
33 ; CHECK-NEXT: movaps (%rax,%rdx), %xmm0
34 ; CHECK-NEXT: .LBB0_3: # %entry
35 ; CHECK-NEXT: leaq -{{[0-9]+}}(%rsp), %rsp
36 ; CHECK-NEXT: movaps (%rax,%rcx), %xmm1
37 ; CHECK-NEXT: movaps %xmm0, -{{[0-9]+}}(%rsp)
38 ; CHECK-NEXT: jne .LBB0_5
39 ; CHECK-NEXT: # %bb.4: # %entry
40 ; CHECK-NEXT: xorps %xmm1, %xmm1
41 ; CHECK-NEXT: .LBB0_5: # %entry
42 ; CHECK-NEXT: movaps %xmm1, -{{[0-9]+}}(%rsp)
43 ; CHECK-NEXT: addq $152, %rsp
46 %i0 = alloca ptr, align 8
47 %i2 = alloca ptr, align 8
48 %b.i = alloca [16 x <2 x double>], align 16
49 %tmp1 = load i32, ptr %_stubArgs, align 4
50 %ptr8 = getelementptr i8, ptr %_stubArgs, i64 16
51 %ptr20 = getelementptr i8, ptr %_stubArgs, i64 48
52 %tmp21 = load ptr addrspace(1), ptr %ptr20, align 8
53 %ptr28 = getelementptr i8, ptr %_stubArgs, i64 64
54 %tmp29 = load i32, ptr %ptr28, align 4
55 %ptr32 = getelementptr i8, ptr %_stubArgs, i64 68
56 %tmp33 = load i32, ptr %ptr32, align 4
57 %tmp17.i = mul i32 10, 20
58 %tmp19.i = add i32 %tmp17.i, %tmp33
59 %conv21.i = zext i32 %tmp19.i to i64
60 %tmp6.i = and i32 42, -32
61 %tmp42.i = add i32 %tmp6.i, 17
62 %tmp44.i = insertelement <2 x i32> undef, i32 %tmp42.i, i32 1
63 %tmp96676677.i = or i32 17, -4
64 %arrayidx4506.i = getelementptr [16 x <2 x double>], ptr %b.i, i64 0, i64 4
65 %tmp52.i = insertelement <2 x i32> %tmp44.i, i32 0, i32 0
66 %tmp78.i = extractelement <2 x i32> %tmp44.i, i32 1
67 %tmp97.i = add i32 %tmp78.i, %tmp96676677.i
68 %tmp99.i = insertelement <2 x i32> %tmp52.i, i32 %tmp97.i, i32 1
69 %tmp154.i = extractelement <2 x i32> %tmp99.i, i32 1
70 %tmp156.i = extractelement <2 x i32> %tmp52.i, i32 0
71 %tmp158.i = urem i32 %tmp156.i, %tmp1
72 %i38 = mul i32 %tmp154.i, %tmp29
73 %i39 = add i32 %tmp158.i, %i38
74 %conv160.i = zext i32 %i39 to i64
75 %tmp22.sum652.i = add i64 %conv160.i, %conv21.i
76 %arrayidx161.i = getelementptr <2 x double>, ptr addrspace(1) %tmp21, i64 %tmp22.sum652.i
77 %tmp162.i = load <2 x double>, ptr addrspace(1) %arrayidx161.i, align 16
78 %tmp222.i = add i32 %tmp154.i, 1
79 %i43 = mul i32 %tmp222.i, %tmp29
80 %i44 = add i32 %tmp158.i, %i43
81 %conv228.i = zext i32 %i44 to i64
82 %tmp22.sum656.i = add i64 %conv228.i, %conv21.i
83 %arrayidx229.i = getelementptr <2 x double>, ptr addrspace(1) %tmp21, i64 %tmp22.sum656.i
84 %tmp230.i = load <2 x double>, ptr addrspace(1) %arrayidx229.i, align 16
85 %cmp432.i = icmp ult i32 %tmp156.i, %tmp1
87 ; %shl.i should not be sinked below the compare.
89 %cond.i = select i1 %cmp432.i, <2 x double> %tmp162.i, <2 x double> zeroinitializer
90 store <2 x double> %cond.i, ptr %b.i, align 16
91 %cond448.i = select i1 %cmp432.i, <2 x double> %tmp230.i, <2 x double> zeroinitializer
92 store <2 x double> %cond448.i, ptr %arrayidx4506.i, align 16