1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
3 ; RUN: llc < %s -mattr=+sse3,+sse4.1 -mcpu=penryn -stats 2>&1 | grep "6 machinelicm"
4 ; RUN: llc < %s -mattr=+sse3,+sse4.1 -mcpu=penryn | FileCheck %s
8 target triple = "x86_64-apple-darwin10.0"
9 %struct.Key = type { i64 }
10 %struct.__Rec = type opaque
11 %struct.__vv = type { }
13 define %struct.__vv* @t(%struct.Key* %desc, i64 %p) nounwind ssp {
15 ; CHECK: ## %bb.0: ## %entry
16 ; CHECK-NEXT: pushq %r14
17 ; CHECK-NEXT: pushq %rbx
18 ; CHECK-NEXT: pushq %rax
19 ; CHECK-NEXT: movq %rsi, %r14
20 ; CHECK-NEXT: movq %rdi, %rbx
21 ; CHECK-NEXT: orq $2097152, %r14 ## imm = 0x200000
22 ; CHECK-NEXT: andl $15728640, %r14d ## imm = 0xF00000
23 ; CHECK-NEXT: .p2align 4, 0x90
24 ; CHECK-NEXT: LBB0_1: ## %bb4
25 ; CHECK-NEXT: ## =>This Inner Loop Header: Depth=1
26 ; CHECK-NEXT: xorl %eax, %eax
27 ; CHECK-NEXT: callq _xxGetOffsetForCode
28 ; CHECK-NEXT: movq %rbx, %rdi
29 ; CHECK-NEXT: xorl %esi, %esi
30 ; CHECK-NEXT: xorl %eax, %eax
31 ; CHECK-NEXT: callq _xxCalculateMidType
32 ; CHECK-NEXT: cmpl $1, %eax
33 ; CHECK-NEXT: jne LBB0_1
34 ; CHECK-NEXT: ## %bb.2: ## %bb26
35 ; CHECK-NEXT: ## in Loop: Header=BB0_1 Depth=1
36 ; CHECK-NEXT: cmpq $1048576, %r14 ## imm = 0x100000
37 ; CHECK-NEXT: jne LBB0_1
38 ; CHECK-NEXT: ## %bb.3: ## %bb.i
39 ; CHECK-NEXT: ## in Loop: Header=BB0_1 Depth=1
40 ; CHECK-NEXT: movl 0, %eax
41 ; CHECK-NEXT: xorps %xmm0, %xmm0
42 ; CHECK-NEXT: cvtsi2ss %rax, %xmm0
43 ; CHECK-NEXT: movl 4, %eax
44 ; CHECK-NEXT: xorps %xmm1, %xmm1
45 ; CHECK-NEXT: cvtsi2ss %rax, %xmm1
46 ; CHECK-NEXT: movl 8, %eax
47 ; CHECK-NEXT: xorps %xmm2, %xmm2
48 ; CHECK-NEXT: cvtsi2ss %rax, %xmm2
49 ; CHECK-NEXT: insertps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[2,3]
50 ; CHECK-NEXT: insertps {{.*#+}} xmm0 = xmm0[0,1],xmm2[0],xmm0[3]
51 ; CHECK-NEXT: movaps %xmm0, 0
52 ; CHECK-NEXT: jmp LBB0_1
56 bb4: ; preds = %bb.i, %bb26, %bb4, %entry
58 %0 = call i32 (...) @xxGetOffsetForCode(i32 undef) nounwind ; <i32> [#uses=0]
59 %ins = or i64 %p, 2097152 ; <i64> [#uses=1]
60 %1 = call i32 (...) @xxCalculateMidType(%struct.Key* %desc, i32 0) nounwind ; <i32> [#uses=1]
61 %cond = icmp eq i32 %1, 1 ; <i1> [#uses=1]
62 br i1 %cond, label %bb26, label %bb4
65 %2 = and i64 %ins, 15728640 ; <i64> [#uses=1]
66 %cond.i = icmp eq i64 %2, 1048576 ; <i1> [#uses=1]
67 br i1 %cond.i, label %bb.i, label %bb4
70 %3 = load i32, i32* null, align 4 ; <i32> [#uses=1]
71 %4 = uitofp i32 %3 to float ; <float> [#uses=1]
72 %.sum13.i = add i64 0, 4 ; <i64> [#uses=1]
73 %5 = getelementptr i8, i8* null, i64 %.sum13.i ; <i8*> [#uses=1]
74 %6 = bitcast i8* %5 to i32* ; <i32*> [#uses=1]
75 %7 = load i32, i32* %6, align 4 ; <i32> [#uses=1]
76 %8 = uitofp i32 %7 to float ; <float> [#uses=1]
77 %.sum.i = add i64 0, 8 ; <i64> [#uses=1]
78 %9 = getelementptr i8, i8* null, i64 %.sum.i ; <i8*> [#uses=1]
79 %10 = bitcast i8* %9 to i32* ; <i32*> [#uses=1]
80 %11 = load i32, i32* %10, align 4 ; <i32> [#uses=1]
81 %12 = uitofp i32 %11 to float ; <float> [#uses=1]
82 %13 = insertelement <4 x float> undef, float %4, i32 0 ; <<4 x float>> [#uses=1]
83 %14 = insertelement <4 x float> %13, float %8, i32 1 ; <<4 x float>> [#uses=1]
84 %15 = insertelement <4 x float> %14, float %12, i32 2 ; <<4 x float>> [#uses=1]
85 store <4 x float> %15, <4 x float>* null, align 16
89 declare i32 @xxGetOffsetForCode(...)
91 declare i32 @xxCalculateMidType(...)