1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py 2; RUN: llc < %s | FileCheck %s 3target 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" 4target triple = "x86_64-pc-linux" 5 6 7%0 = type <{ i64, i64, %1, %1, [21 x %2] }> 8%1 = type <{ i64, i64, i64 }> 9%2 = type <{ i32, i32, i8 addrspace(2)* }> 10%3 = type { i8*, i8*, i8*, i8*, i32 } 11%4 = type <{ %5*, i8*, i32, i32, [4 x i64], [4 x i64], [4 x i64], [4 x i64], [4 x i64] }> 12%5 = type <{ void (i32)*, i8*, i32 (i8*, ...)* }> 13 14define void @foo(i8* nocapture %_stubArgs) nounwind { 15; CHECK-LABEL: foo: 16; CHECK: # %bb.0: # %entry 17; CHECK-NEXT: subq $152, %rsp 18; CHECK-NEXT: movq 48(%rdi), %rax 19; CHECK-NEXT: movl 64(%rdi), %ecx 20; CHECK-NEXT: movl $200, %esi 21; CHECK-NEXT: addl 68(%rdi), %esi 22; CHECK-NEXT: imull $46, %ecx, %edx 23; CHECK-NEXT: addq %rsi, %rdx 24; CHECK-NEXT: shlq $4, %rdx 25; CHECK-NEXT: imull $47, %ecx, %ecx 26; CHECK-NEXT: addq %rsi, %rcx 27; CHECK-NEXT: shlq $4, %rcx 28; CHECK-NEXT: cmpl $0, (%rdi) 29; CHECK-NEXT: jne .LBB0_1 30; CHECK-NEXT: # %bb.2: # %entry 31; CHECK-NEXT: xorps %xmm0, %xmm0 32; CHECK-NEXT: jmp .LBB0_3 33; CHECK-NEXT: .LBB0_1: 34; CHECK-NEXT: movaps (%rax,%rdx), %xmm0 35; CHECK-NEXT: .LBB0_3: # %entry 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 44; CHECK-NEXT: retq 45entry: 46 %i0 = alloca i8*, align 8 47 %i2 = alloca i8*, align 8 48 %b.i = alloca [16 x <2 x double>], align 16 49 %conv = bitcast i8* %_stubArgs to i32* 50 %tmp1 = load i32, i32* %conv, align 4 51 %ptr8 = getelementptr i8, i8* %_stubArgs, i64 16 52 %i4 = bitcast i8* %ptr8 to <2 x double>* 53 %ptr20 = getelementptr i8, i8* %_stubArgs, i64 48 54 %i7 = bitcast i8* %ptr20 to <2 x double> addrspace(1)** 55 %tmp21 = load <2 x double> addrspace(1)*, <2 x double> addrspace(1)** %i7, align 8 56 %ptr28 = getelementptr i8, i8* %_stubArgs, i64 64 57 %i9 = bitcast i8* %ptr28 to i32* 58 %tmp29 = load i32, i32* %i9, align 4 59 %ptr32 = getelementptr i8, i8* %_stubArgs, i64 68 60 %i10 = bitcast i8* %ptr32 to i32* 61 %tmp33 = load i32, i32* %i10, align 4 62 %tmp17.i = mul i32 10, 20 63 %tmp19.i = add i32 %tmp17.i, %tmp33 64 %conv21.i = zext i32 %tmp19.i to i64 65 %tmp6.i = and i32 42, -32 66 %tmp42.i = add i32 %tmp6.i, 17 67 %tmp44.i = insertelement <2 x i32> undef, i32 %tmp42.i, i32 1 68 %tmp96676677.i = or i32 17, -4 69 %ptr4438.i = getelementptr inbounds [16 x <2 x double>], [16 x <2 x double>]* %b.i, i64 0, i64 0 70 %arrayidx4506.i = getelementptr [16 x <2 x double>], [16 x <2 x double>]* %b.i, i64 0, i64 4 71 %tmp52.i = insertelement <2 x i32> %tmp44.i, i32 0, i32 0 72 %tmp78.i = extractelement <2 x i32> %tmp44.i, i32 1 73 %tmp97.i = add i32 %tmp78.i, %tmp96676677.i 74 %tmp99.i = insertelement <2 x i32> %tmp52.i, i32 %tmp97.i, i32 1 75 %tmp154.i = extractelement <2 x i32> %tmp99.i, i32 1 76 %tmp156.i = extractelement <2 x i32> %tmp52.i, i32 0 77 %tmp158.i = urem i32 %tmp156.i, %tmp1 78 %i38 = mul i32 %tmp154.i, %tmp29 79 %i39 = add i32 %tmp158.i, %i38 80 %conv160.i = zext i32 %i39 to i64 81 %tmp22.sum652.i = add i64 %conv160.i, %conv21.i 82 %arrayidx161.i = getelementptr <2 x double>, <2 x double> addrspace(1)* %tmp21, i64 %tmp22.sum652.i 83 %tmp162.i = load <2 x double>, <2 x double> addrspace(1)* %arrayidx161.i, align 16 84 %tmp222.i = add i32 %tmp154.i, 1 85 %i43 = mul i32 %tmp222.i, %tmp29 86 %i44 = add i32 %tmp158.i, %i43 87 %conv228.i = zext i32 %i44 to i64 88 %tmp22.sum656.i = add i64 %conv228.i, %conv21.i 89 %arrayidx229.i = getelementptr <2 x double>, <2 x double> addrspace(1)* %tmp21, i64 %tmp22.sum656.i 90 %tmp230.i = load <2 x double>, <2 x double> addrspace(1)* %arrayidx229.i, align 16 91 %cmp432.i = icmp ult i32 %tmp156.i, %tmp1 92 93; %shl.i should not be sinked below the compare. 94 95 %cond.i = select i1 %cmp432.i, <2 x double> %tmp162.i, <2 x double> zeroinitializer 96 store <2 x double> %cond.i, <2 x double>* %ptr4438.i, align 16 97 %cond448.i = select i1 %cmp432.i, <2 x double> %tmp230.i, <2 x double> zeroinitializer 98 store <2 x double> %cond448.i, <2 x double>* %arrayidx4506.i, align 16 99 ret void 100} 101 102 103 104