1; RUN: opt < %s -msan -msan-check-access-address=0 -S | FileCheck %s 2 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-S128" 4target triple = "x86_64-unknown-linux-gnu" 5 6declare <4 x float> @llvm.x86.sse.cmp.ss(<4 x float>, <4 x float>, i8) nounwind readnone 7declare <4 x float> @llvm.x86.sse.cmp.ps(<4 x float>, <4 x float>, i8) nounwind readnone 8declare <2 x double> @llvm.x86.sse2.cmp.sd(<2 x double>, <2 x double>, i8) nounwind readnone 9declare i32 @llvm.x86.sse.comineq.ss(<4 x float>, <4 x float>) nounwind readnone 10declare i32 @llvm.x86.sse2.ucomilt.sd(<2 x double>, <2 x double>) nounwind readnone 11 12 13define <4 x float> @test_sse_cmp_ss(<4 x float> %a, <4 x float> %b) sanitize_memory { 14entry: 15 %0 = tail call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a, <4 x float> %b, i8 4) 16 ret <4 x float> %0 17} 18 19; CHECK-LABEL: @test_sse_cmp_ss 20; CHECK: %[[A:.*]] = or <4 x i32> 21; CHECK: %[[B:.*]] = extractelement <4 x i32> %[[A]], i64 0 22; CHECK: %[[C:.*]] = icmp ne i32 %[[B]], 0 23; CHECK: %[[D:.*]] = sext i1 %[[C]] to i128 24; CHECK: %[[E:.*]] = bitcast i128 %[[D]] to <4 x i32> 25; CHECK: store <4 x i32> %[[E]] 26 27 28define <4 x float> @test_sse_cmp_ps(<4 x float> %a, <4 x float> %b) sanitize_memory { 29entry: 30 %0 = tail call <4 x float> @llvm.x86.sse.cmp.ps(<4 x float> %a, <4 x float> %b, i8 4) 31 ret <4 x float> %0 32} 33 34; CHECK-LABEL: @test_sse_cmp_ps 35; CHECK: %[[A:.*]] = or <4 x i32> 36; CHECK: %[[B:.*]] = icmp ne <4 x i32> %[[A]], zeroinitializer 37; CHECK: %[[C:.*]] = sext <4 x i1> %[[B]] to <4 x i32> 38; CHECK: store <4 x i32> %[[C]] 39 40 41define <2 x double> @test_sse2_cmp_sd(<2 x double> %a, <2 x double> %b) sanitize_memory { 42entry: 43 %0 = tail call <2 x double> @llvm.x86.sse2.cmp.sd(<2 x double> %a, <2 x double> %b, i8 4) 44 ret <2 x double> %0 45} 46 47; CHECK-LABEL: @test_sse2_cmp_sd 48; CHECK: %[[A:.*]] = or <2 x i64> 49; CHECK: %[[B:.*]] = extractelement <2 x i64> %[[A]], i64 0 50; CHECK: %[[C:.*]] = icmp ne i64 %[[B]], 0 51; CHECK: %[[D:.*]] = sext i1 %[[C]] to i128 52; CHECK: %[[E:.*]] = bitcast i128 %[[D]] to <2 x i64> 53; CHECK: store <2 x i64> %[[E]] 54 55 56define i32 @test_sse_comineq_ss(<4 x float> %a, <4 x float> %b) sanitize_memory { 57entry: 58 %0 = tail call i32 @llvm.x86.sse.comineq.ss(<4 x float> %a, <4 x float> %b) 59 ret i32 %0 60} 61 62; CHECK-LABEL: @test_sse_comineq_ss 63; CHECK: %[[A:.*]] = or <4 x i32> 64; CHECK: %[[B:.*]] = extractelement <4 x i32> %[[A]], i64 0 65; CHECK: %[[C:.*]] = icmp ne i32 %[[B]], 0 66; CHECK: %[[D:.*]] = sext i1 %[[C]] to i32 67; CHECK: store i32 %[[D]] 68 69 70define i32 @test_sse2_ucomilt_sd(<2 x double> %a, <2 x double> %b) sanitize_memory { 71entry: 72 %0 = tail call i32 @llvm.x86.sse2.ucomilt.sd(<2 x double> %a, <2 x double> %b) 73 ret i32 %0 74} 75 76; CHECK-LABEL: @test_sse2_ucomilt_sd 77; CHECK: %[[A:.*]] = or <2 x i64> 78; CHECK: %[[B:.*]] = extractelement <2 x i64> %[[A]], i64 0 79; CHECK: %[[C:.*]] = icmp ne i64 %[[B]], 0 80; CHECK: %[[D:.*]] = sext i1 %[[C]] to i32 81; CHECK: store i32 %[[D]] 82