1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py 2; RUN: llc < %s -mtriple=x86_64-unknown | FileCheck %s -check-prefix=X64 3; RUN: llc < %s -mtriple=i686-unknown | FileCheck %s -check-prefix=X86 4 5%struct.SA = type { i32 , i32 , i32 , i32 , i32}; 6 7define void @foo(ptr nocapture %ctx, i32 %n) local_unnamed_addr #0 { 8; X64-LABEL: foo: 9; X64: # %bb.0: # %entry 10; X64-NEXT: movl 16(%rdi), %eax 11; X64-NEXT: movl (%rdi), %ecx 12; X64-NEXT: addl %eax, %ecx 13; X64-NEXT: addl %eax, %ecx 14; X64-NEXT: addl %eax, %ecx 15; X64-NEXT: leal (%rcx,%rax), %edx 16; X64-NEXT: leal 1(%rax,%rcx), %ecx 17; X64-NEXT: movl %ecx, 12(%rdi) 18; X64-NEXT: leal 1(%rax,%rdx), %eax 19; X64-NEXT: movl %eax, 16(%rdi) 20; X64-NEXT: retq 21; 22; X86-LABEL: foo: 23; X86: # %bb.0: # %entry 24; X86-NEXT: pushl %esi 25; X86-NEXT: .cfi_def_cfa_offset 8 26; X86-NEXT: .cfi_offset %esi, -8 27; X86-NEXT: movl {{[0-9]+}}(%esp), %eax 28; X86-NEXT: movl 16(%eax), %ecx 29; X86-NEXT: movl (%eax), %edx 30; X86-NEXT: addl %ecx, %edx 31; X86-NEXT: addl %ecx, %edx 32; X86-NEXT: addl %ecx, %edx 33; X86-NEXT: leal 1(%ecx,%edx), %esi 34; X86-NEXT: addl %ecx, %edx 35; X86-NEXT: movl %esi, 12(%eax) 36; X86-NEXT: leal 1(%ecx,%edx), %ecx 37; X86-NEXT: movl %ecx, 16(%eax) 38; X86-NEXT: popl %esi 39; X86-NEXT: .cfi_def_cfa_offset 4 40; X86-NEXT: retl 41 entry: 42 %0 = load i32, ptr %ctx, align 8 43 %h3 = getelementptr inbounds %struct.SA, ptr %ctx, i64 0, i32 3 44 %h4 = getelementptr inbounds %struct.SA, ptr %ctx, i64 0, i32 4 45 %1 = load i32, ptr %h4, align 8 46 %add = add i32 %0 , 1 47 %add1 = add i32 %add, %1 48 %add2 = add i32 %add1, %1 49 %add3 = add i32 %add2, %1 50 %add4 = add i32 %add3, %1 51 store i32 %add4, ptr %h3, align 4 52 %add29 = add i32 %add4, %1 53 store i32 %add29, ptr %h4, align 8 54 ret void 55} 56 57 58 59define void @foo_loop(ptr nocapture %ctx, i32 %n) local_unnamed_addr #0 { 60; X64-LABEL: foo_loop: 61; X64: # %bb.0: # %entry 62; X64-NEXT: .p2align 4 63; X64-NEXT: .LBB1_1: # %loop 64; X64-NEXT: # =>This Inner Loop Header: Depth=1 65; X64-NEXT: movl (%rdi), %ecx 66; X64-NEXT: movl 16(%rdi), %eax 67; X64-NEXT: leal 1(%rcx,%rax), %edx 68; X64-NEXT: movl %edx, 12(%rdi) 69; X64-NEXT: decl %esi 70; X64-NEXT: jne .LBB1_1 71; X64-NEXT: # %bb.2: # %exit 72; X64-NEXT: addl %eax, %ecx 73; X64-NEXT: leal 1(%rax,%rcx), %ecx 74; X64-NEXT: leal (%rax,%rax), %edx 75; X64-NEXT: addl %eax, %edx 76; X64-NEXT: addl %edx, %ecx 77; X64-NEXT: addl %edx, %ecx 78; X64-NEXT: movl %ecx, 16(%rdi) 79; X64-NEXT: retq 80; 81; X86-LABEL: foo_loop: 82; X86: # %bb.0: # %entry 83; X86-NEXT: pushl %edi 84; X86-NEXT: .cfi_def_cfa_offset 8 85; X86-NEXT: pushl %esi 86; X86-NEXT: .cfi_def_cfa_offset 12 87; X86-NEXT: .cfi_offset %esi, -12 88; X86-NEXT: .cfi_offset %edi, -8 89; X86-NEXT: movl {{[0-9]+}}(%esp), %edx 90; X86-NEXT: movl {{[0-9]+}}(%esp), %eax 91; X86-NEXT: .p2align 4 92; X86-NEXT: .LBB1_1: # %loop 93; X86-NEXT: # =>This Inner Loop Header: Depth=1 94; X86-NEXT: movl (%eax), %esi 95; X86-NEXT: movl 16(%eax), %ecx 96; X86-NEXT: leal 1(%esi,%ecx), %edi 97; X86-NEXT: movl %edi, 12(%eax) 98; X86-NEXT: decl %edx 99; X86-NEXT: jne .LBB1_1 100; X86-NEXT: # %bb.2: # %exit 101; X86-NEXT: addl %ecx, %esi 102; X86-NEXT: leal 1(%ecx,%esi), %edx 103; X86-NEXT: leal (%ecx,%ecx), %esi 104; X86-NEXT: addl %ecx, %esi 105; X86-NEXT: addl %esi, %edx 106; X86-NEXT: addl %esi, %edx 107; X86-NEXT: movl %edx, 16(%eax) 108; X86-NEXT: popl %esi 109; X86-NEXT: .cfi_def_cfa_offset 8 110; X86-NEXT: popl %edi 111; X86-NEXT: .cfi_def_cfa_offset 4 112; X86-NEXT: retl 113 entry: 114 br label %loop 115 116 loop: 117 %iter = phi i32 [%n ,%entry ] ,[ %iter.ctr ,%loop] 118 %0 = load i32, ptr %ctx, align 8 119 %h3 = getelementptr inbounds %struct.SA, ptr %ctx, i64 0, i32 3 120 %h4 = getelementptr inbounds %struct.SA, ptr %ctx, i64 0, i32 4 121 %1 = load i32, ptr %h4, align 8 122 %add = add i32 %0, 1 123 %add4 = add i32 %add, %1 124 store i32 %add4, ptr %h3, align 4 125 %add291 = add i32 %add4, %1 126 %add292 = add i32 %add291, %1 127 %add293 = add i32 %add292, %1 128 %add294 = add i32 %add293, %1 129 %add295 = add i32 %add294, %1 130 %add296 = add i32 %add295, %1 131 %add29 = add i32 %add296, %1 132 %iter.ctr = sub i32 %iter , 1 133 %res = icmp ne i32 %iter.ctr , 0 134 br i1 %res , label %loop , label %exit 135 136 exit: 137 store i32 %add29, ptr %h4, align 8 138 ret void 139} 140