1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py 2; RUN: llc -mtriple=armv7-none-linux-eabi < %s | FileCheck %s 3 4define void @vld1x2(i8* %v4, i32 %v2) { 5; CHECK-LABEL: vld1x2: 6; CHECK: @ %bb.0: 7; CHECK-NEXT: mov r0, #0 8; CHECK-NEXT: .LBB0_1: @ %.preheader 9; CHECK-NEXT: @ =>This Inner Loop Header: Depth=1 10; CHECK-NEXT: cmp r0, #0 11; CHECK-NEXT: bne .LBB0_1 12; CHECK-NEXT: @ %bb.2: @ %.loopexit 13; CHECK-NEXT: @ in Loop: Header=BB0_1 Depth=1 14; CHECK-NEXT: vst3.8 {d16, d17, d18}, [r0] 15; CHECK-NEXT: b .LBB0_1 16 br label %.preheader 17 18.preheader: ; preds = %.preheader, %3 19 %v5 = icmp eq i8* %v4, undef 20 br i1 %v5, label %.loopexit, label %.preheader 21 22.loopexit: ; preds = %.preheader 23 %v6 = tail call { <8 x i8>, <8 x i8> } @llvm.arm.neon.vld1x2.v8i8.p0i8(i8* %v4) 24 %v7 = getelementptr inbounds i8, i8* %v4, i32 %v2 25 %v8 = tail call { <8 x i8>, <8 x i8> } @llvm.arm.neon.vld1x2.v8i8.p0i8(i8* %v7) 26 tail call void @llvm.arm.neon.vst3.p0i8.v8i8(i8* undef, <8 x i8> undef, <8 x i8> undef, <8 x i8> undef, i32 1) 27 br label %.preheader 28} 29 30define void @vld1x3(i8* %v4, i32 %v2) { 31; CHECK-LABEL: vld1x3: 32; CHECK: @ %bb.0: 33; CHECK-NEXT: mov r0, #0 34; CHECK-NEXT: .LBB1_1: @ %.preheader 35; CHECK-NEXT: @ =>This Inner Loop Header: Depth=1 36; CHECK-NEXT: cmp r0, #0 37; CHECK-NEXT: bne .LBB1_1 38; CHECK-NEXT: @ %bb.2: @ %.loopexit 39; CHECK-NEXT: @ in Loop: Header=BB1_1 Depth=1 40; CHECK-NEXT: vst3.8 {d16, d17, d18}, [r0] 41; CHECK-NEXT: b .LBB1_1 42 br label %.preheader 43 44.preheader: ; preds = %.preheader, %3 45 %v5 = icmp eq i8* %v4, undef 46 br i1 %v5, label %.loopexit, label %.preheader 47 48.loopexit: ; preds = %.preheader 49 %v6 = tail call { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm.neon.vld1x3.v8i8.p0i8(i8* %v4) 50 %v7 = getelementptr inbounds i8, i8* %v4, i32 %v2 51 %v8 = tail call { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm.neon.vld1x3.v8i8.p0i8(i8* %v7) 52 tail call void @llvm.arm.neon.vst3.p0i8.v8i8(i8* undef, <8 x i8> undef, <8 x i8> undef, <8 x i8> undef, i32 1) 53 br label %.preheader 54} 55 56define void @vld1x4(i8* %v4, i32 %v2) { 57; CHECK-LABEL: vld1x4: 58; CHECK: @ %bb.0: 59; CHECK-NEXT: mov r0, #0 60; CHECK-NEXT: .LBB2_1: @ %.preheader 61; CHECK-NEXT: @ =>This Inner Loop Header: Depth=1 62; CHECK-NEXT: cmp r0, #0 63; CHECK-NEXT: bne .LBB2_1 64; CHECK-NEXT: @ %bb.2: @ %.loopexit 65; CHECK-NEXT: @ in Loop: Header=BB2_1 Depth=1 66; CHECK-NEXT: vst3.8 {d16, d17, d18}, [r0] 67; CHECK-NEXT: b .LBB2_1 68 br label %.preheader 69 70.preheader: ; preds = %.preheader, %3 71 %v5 = icmp eq i8* %v4, undef 72 br i1 %v5, label %.loopexit, label %.preheader 73 74.loopexit: ; preds = %.preheader 75 %v6 = tail call { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm.neon.vld1x4.v8i8.p0i8(i8* %v4) 76 %v7 = getelementptr inbounds i8, i8* %v4, i32 %v2 77 %v8 = tail call { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm.neon.vld1x4.v8i8.p0i8(i8* %v7) 78 tail call void @llvm.arm.neon.vst3.p0i8.v8i8(i8* undef, <8 x i8> undef, <8 x i8> undef, <8 x i8> undef, i32 1) 79 br label %.preheader 80} 81 82declare void @llvm.arm.neon.vst3.p0i8.v8i8(i8*, <8 x i8>, <8 x i8>, <8 x i8>, i32) 83declare { <8 x i8>, <8 x i8> } @llvm.arm.neon.vld1x2.v8i8.p0i8(i8*) 84declare { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm.neon.vld1x3.v8i8.p0i8(i8*) 85declare { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm.neon.vld1x4.v8i8.p0i8(i8*) 86