From e027afd8f73652440f801da8eafbe057c3f052cb Mon Sep 17 00:00:00 2001 From: Nikita Popov Date: Fri, 28 Oct 2022 14:08:34 +0200 Subject: [PATCH] [InstCombine] Convert test to opaque pointers (NFC) With some manual fixup for different alloca types and variable name clashes. --- .../Transforms/InstCombine/memcpy-from-global.ll | 232 +++++++++------------ 1 file changed, 102 insertions(+), 130 deletions(-) diff --git a/llvm/test/Transforms/InstCombine/memcpy-from-global.ll b/llvm/test/Transforms/InstCombine/memcpy-from-global.ll index ae36531..4e1c88f 100644 --- a/llvm/test/Transforms/InstCombine/memcpy-from-global.ll +++ b/llvm/test/Transforms/InstCombine/memcpy-from-global.ll @@ -1,7 +1,7 @@ ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; RUN: opt < %s -passes=instcombine -S | FileCheck %s target datalayout = "E-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64" -@C.0.1248 = internal constant [128 x float] [ float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00 ], align 32 ; <[128 x float]*> [#uses=1] +@C.0.1248 = internal constant [128 x float] [ float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00 ], align 32 ; [#uses=1] define float @test1(i32 %hash, float %x, float %y, float %z, float %w) { ; CHECK-LABEL: @test1( @@ -9,64 +9,63 @@ define float @test1(i32 %hash, float %x, float %y, float %z, float %w) { ; CHECK-NEXT: [[TMP3:%.*]] = shl i32 [[HASH:%.*]], 2 ; CHECK-NEXT: [[TMP5:%.*]] = and i32 [[TMP3]], 124 ; CHECK-NEXT: [[TMP0:%.*]] = zext i32 [[TMP5]] to i64 -; CHECK-NEXT: [[TMP753:%.*]] = getelementptr [128 x float], [128 x float]* @C.0.1248, i64 0, i64 [[TMP0]] -; CHECK-NEXT: [[TMP9:%.*]] = load float, float* [[TMP753]], align 16 +; CHECK-NEXT: [[TMP753:%.*]] = getelementptr [128 x float], ptr @C.0.1248, i64 0, i64 [[TMP0]] +; CHECK-NEXT: [[TMP9:%.*]] = load float, ptr [[TMP753]], align 16 ; CHECK-NEXT: [[TMP11:%.*]] = fmul float [[TMP9]], [[X:%.*]] ; CHECK-NEXT: [[TMP13:%.*]] = fadd float [[TMP11]], 0.000000e+00 ; CHECK-NEXT: [[TMP17_SUM52:%.*]] = or i32 [[TMP5]], 1 ; CHECK-NEXT: [[TMP1:%.*]] = zext i32 [[TMP17_SUM52]] to i64 -; CHECK-NEXT: [[TMP1851:%.*]] = getelementptr [128 x float], [128 x float]* @C.0.1248, i64 0, i64 [[TMP1]] -; CHECK-NEXT: [[TMP19:%.*]] = load float, float* [[TMP1851]], align 4 +; CHECK-NEXT: [[TMP1851:%.*]] = getelementptr [128 x float], ptr @C.0.1248, i64 0, i64 [[TMP1]] +; CHECK-NEXT: [[TMP19:%.*]] = load float, ptr [[TMP1851]], align 4 ; CHECK-NEXT: [[TMP21:%.*]] = fmul float [[TMP19]], [[Y:%.*]] ; CHECK-NEXT: [[TMP23:%.*]] = fadd float [[TMP21]], [[TMP13]] ; CHECK-NEXT: [[TMP27_SUM50:%.*]] = or i32 [[TMP5]], 2 ; CHECK-NEXT: [[TMP2:%.*]] = zext i32 [[TMP27_SUM50]] to i64 -; CHECK-NEXT: [[TMP2849:%.*]] = getelementptr [128 x float], [128 x float]* @C.0.1248, i64 0, i64 [[TMP2]] -; CHECK-NEXT: [[TMP29:%.*]] = load float, float* [[TMP2849]], align 8 +; CHECK-NEXT: [[TMP2849:%.*]] = getelementptr [128 x float], ptr @C.0.1248, i64 0, i64 [[TMP2]] +; CHECK-NEXT: [[TMP29:%.*]] = load float, ptr [[TMP2849]], align 8 ; CHECK-NEXT: [[TMP31:%.*]] = fmul float [[TMP29]], [[Z:%.*]] ; CHECK-NEXT: [[TMP33:%.*]] = fadd float [[TMP31]], [[TMP23]] ; CHECK-NEXT: [[TMP37_SUM48:%.*]] = or i32 [[TMP5]], 3 ; CHECK-NEXT: [[TMP3:%.*]] = zext i32 [[TMP37_SUM48]] to i64 -; CHECK-NEXT: [[TMP3847:%.*]] = getelementptr [128 x float], [128 x float]* @C.0.1248, i64 0, i64 [[TMP3]] -; CHECK-NEXT: [[TMP39:%.*]] = load float, float* [[TMP3847]], align 4 +; CHECK-NEXT: [[TMP3847:%.*]] = getelementptr [128 x float], ptr @C.0.1248, i64 0, i64 [[TMP3]] +; CHECK-NEXT: [[TMP39:%.*]] = load float, ptr [[TMP3847]], align 4 ; CHECK-NEXT: [[TMP41:%.*]] = fmul float [[TMP39]], [[W:%.*]] ; CHECK-NEXT: [[TMP43:%.*]] = fadd float [[TMP41]], [[TMP33]] ; CHECK-NEXT: ret float [[TMP43]] ; entry: - %lookupTable = alloca [128 x float], align 16 ; <[128 x float]*> [#uses=5] - %lookupTable1 = bitcast [128 x float]* %lookupTable to i8* ; [#uses=1] - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 %lookupTable1, i8* align 16 bitcast ([128 x float]* @C.0.1248 to i8*), i64 512, i1 false) + %lookupTable = alloca [128 x float], align 16 ; [#uses=5] + call void @llvm.memcpy.p0.p0.i64(ptr align 16 %lookupTable, ptr align 16 @C.0.1248, i64 512, i1 false) %tmp3 = shl i32 %hash, 2 ; [#uses=1] %tmp5 = and i32 %tmp3, 124 ; [#uses=4] - %tmp753 = getelementptr [128 x float], [128 x float]* %lookupTable, i32 0, i32 %tmp5 ; [#uses=1] - %tmp9 = load float, float* %tmp753 ; [#uses=1] + %tmp753 = getelementptr [128 x float], ptr %lookupTable, i32 0, i32 %tmp5 ; [#uses=1] + %tmp9 = load float, ptr %tmp753 ; [#uses=1] %tmp11 = fmul float %tmp9, %x ; [#uses=1] %tmp13 = fadd float %tmp11, 0.000000e+00 ; [#uses=1] %tmp17.sum52 = or i32 %tmp5, 1 ; [#uses=1] - %tmp1851 = getelementptr [128 x float], [128 x float]* %lookupTable, i32 0, i32 %tmp17.sum52 ; [#uses=1] - %tmp19 = load float, float* %tmp1851 ; [#uses=1] + %tmp1851 = getelementptr [128 x float], ptr %lookupTable, i32 0, i32 %tmp17.sum52 ; [#uses=1] + %tmp19 = load float, ptr %tmp1851 ; [#uses=1] %tmp21 = fmul float %tmp19, %y ; [#uses=1] %tmp23 = fadd float %tmp21, %tmp13 ; [#uses=1] %tmp27.sum50 = or i32 %tmp5, 2 ; [#uses=1] - %tmp2849 = getelementptr [128 x float], [128 x float]* %lookupTable, i32 0, i32 %tmp27.sum50 ; [#uses=1] - %tmp29 = load float, float* %tmp2849 ; [#uses=1] + %tmp2849 = getelementptr [128 x float], ptr %lookupTable, i32 0, i32 %tmp27.sum50 ; [#uses=1] + %tmp29 = load float, ptr %tmp2849 ; [#uses=1] %tmp31 = fmul float %tmp29, %z ; [#uses=1] %tmp33 = fadd float %tmp31, %tmp23 ; [#uses=1] %tmp37.sum48 = or i32 %tmp5, 3 ; [#uses=1] - %tmp3847 = getelementptr [128 x float], [128 x float]* %lookupTable, i32 0, i32 %tmp37.sum48 ; [#uses=1] - %tmp39 = load float, float* %tmp3847 ; [#uses=1] + %tmp3847 = getelementptr [128 x float], ptr %lookupTable, i32 0, i32 %tmp37.sum48 ; [#uses=1] + %tmp39 = load float, ptr %tmp3847 ; [#uses=1] %tmp41 = fmul float %tmp39, %w ; [#uses=1] %tmp43 = fadd float %tmp41, %tmp33 ; [#uses=1] ret float %tmp43 } -declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i1) nounwind -declare void @llvm.memcpy.p1i8.p0i8.i64(i8 addrspace(1)* nocapture, i8* nocapture, i64, i1) nounwind -declare void @llvm.memcpy.p0i8.p1i8.i64(i8* nocapture, i8 addrspace(1)* nocapture, i64, i1) nounwind -declare void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* nocapture, i8 addrspace(1)* nocapture, i64, i1) nounwind +declare void @llvm.memcpy.p0.p0.i64(ptr nocapture, ptr nocapture, i64, i1) nounwind +declare void @llvm.memcpy.p1.p0.i64(ptr addrspace(1) nocapture, ptr nocapture, i64, i1) nounwind +declare void @llvm.memcpy.p0.p1.i64(ptr nocapture, ptr addrspace(1) nocapture, i64, i1) nounwind +declare void @llvm.memcpy.p1.p1.i64(ptr addrspace(1) nocapture, ptr addrspace(1) nocapture, i64, i1) nounwind %T = type { i8, [123 x i8] } %U = type { i32, i32, i32, i32, i32 } @@ -77,63 +76,56 @@ declare void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* nocapture, i8 addrspace define void @test2() { ; CHECK-LABEL: @test2( -; CHECK-NEXT: [[B1:%.*]] = alloca [124 x i8], align 8 -; CHECK-NEXT: [[B1_SUB:%.*]] = getelementptr inbounds [124 x i8], [124 x i8]* [[B1]], i64 0, i64 0 -; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* noundef nonnull align 8 dereferenceable(124) [[B1_SUB]], i8* noundef nonnull align 16 dereferenceable(124) getelementptr inbounds ([[T:%.*]], %T* @G, i64 0, i32 0), i64 124, i1 false) -; CHECK-NEXT: call void @bar(i8* nonnull [[B1_SUB]]) +; CHECK-NEXT: [[B:%.*]] = alloca [[T:%.*]], align 8 +; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 8 dereferenceable(124) [[B]], ptr noundef nonnull align 16 dereferenceable(124) @G, i64 124, i1 false) +; CHECK-NEXT: call void @bar(ptr nonnull [[B]]) ; CHECK-NEXT: ret void ; %A = alloca %T %B = alloca %T - %a = bitcast %T* %A to i8* - %b = bitcast %T* %B to i8* ; %A alloca is deleted ; use @G instead of %A - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%T* @G to i8*), i64 124, i1 false) - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %b, i8* align 4 %a, i64 124, i1 false) - call void @bar(i8* %b) + call void @llvm.memcpy.p0.p0.i64(ptr align 4 %A, ptr align 4 @G, i64 124, i1 false) + call void @llvm.memcpy.p0.p0.i64(ptr align 4 %B, ptr align 4 %A, i64 124, i1 false) + call void @bar(ptr %B) ret void } define void @test2_no_null_opt() #0 { ; CHECK-LABEL: @test2_no_null_opt( -; CHECK-NEXT: [[B1:%.*]] = alloca [124 x i8], align 8 -; CHECK-NEXT: [[B1_SUB:%.*]] = getelementptr inbounds [124 x i8], [124 x i8]* [[B1]], i64 0, i64 0 -; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* noundef align 8 dereferenceable(124) [[B1_SUB]], i8* noundef nonnull align 16 dereferenceable(124) getelementptr inbounds ([[T:%.*]], %T* @G, i64 0, i32 0), i64 124, i1 false) -; CHECK-NEXT: call void @bar(i8* [[B1_SUB]]) +; CHECK-NEXT: [[B:%.*]] = alloca [[T:%.*]], align 8 +; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 8 dereferenceable(124) [[B]], ptr noundef nonnull align 16 dereferenceable(124) @G, i64 124, i1 false) +; CHECK-NEXT: call void @bar(ptr nonnull [[B]]) ; CHECK-NEXT: ret void ; %A = alloca %T %B = alloca %T - %a = bitcast %T* %A to i8* - %b = bitcast %T* %B to i8* ; %A alloca is deleted ; use @G instead of %A - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%T* @G to i8*), i64 124, i1 false) - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %b, i8* align 4 %a, i64 124, i1 false) - call void @bar(i8* %b) + call void @llvm.memcpy.p0.p0.i64(ptr align 4 %A, ptr align 4 @G, i64 124, i1 false) + call void @llvm.memcpy.p0.p0.i64(ptr align 4 %B, ptr align 4 %A, i64 124, i1 false) + call void @bar(ptr %B) ret void } define void @test2_addrspacecast() { ; CHECK-LABEL: @test2_addrspacecast( -; CHECK-NEXT: [[B1:%.*]] = alloca [124 x i8], align 8 -; CHECK-NEXT: [[B1_SUB:%.*]] = getelementptr inbounds [124 x i8], [124 x i8]* [[B1]], i64 0, i64 0 -; CHECK-NEXT: [[B:%.*]] = addrspacecast i8* [[B1_SUB]] to i8 addrspace(1)* -; CHECK-NEXT: call void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* noundef align 4 dereferenceable(124) [[B]], i8 addrspace(1)* noundef align 4 dereferenceable(124) addrspacecast (i8* getelementptr inbounds ([[T:%.*]], %T* @G, i64 0, i32 0) to i8 addrspace(1)*), i64 124, i1 false) -; CHECK-NEXT: call void @bar_as1(i8 addrspace(1)* [[B]]) +; CHECK-NEXT: [[B:%.*]] = alloca [[T:%.*]], align 8 +; CHECK-NEXT: [[B_CAST:%.*]] = addrspacecast ptr [[B]] to ptr addrspace(1) +; CHECK-NEXT: call void @llvm.memcpy.p1.p1.i64(ptr addrspace(1) noundef align 4 dereferenceable(124) [[B_CAST]], ptr addrspace(1) noundef align 4 dereferenceable(124) addrspacecast (ptr @G to ptr addrspace(1)), i64 124, i1 false) +; CHECK-NEXT: call void @bar_as1(ptr addrspace(1) [[B_CAST]]) ; CHECK-NEXT: ret void ; %A = alloca %T %B = alloca %T - %a = addrspacecast %T* %A to i8 addrspace(1)* - %b = addrspacecast %T* %B to i8 addrspace(1)* + %a.cast = addrspacecast ptr %A to ptr addrspace(1) + %b.cast = addrspacecast ptr %B to ptr addrspace(1) ; %A alloca is deleted @@ -141,108 +133,100 @@ define void @test2_addrspacecast() { ; cast into the alloca doesn't work for the addrspacecast yet. ; use @G instead of %A - call void @llvm.memcpy.p1i8.p0i8.i64(i8 addrspace(1)* align 4 %a, i8* align 4 bitcast (%T* @G to i8*), i64 124, i1 false) - call void @llvm.memcpy.p1i8.p1i8.i64(i8 addrspace(1)* align 4 %b, i8 addrspace(1)* align 4 %a, i64 124, i1 false) - call void @bar_as1(i8 addrspace(1)* %b) + call void @llvm.memcpy.p1.p0.i64(ptr addrspace(1) align 4 %a.cast, ptr align 4 @G, i64 124, i1 false) + call void @llvm.memcpy.p1.p1.i64(ptr addrspace(1) align 4 %b.cast, ptr addrspace(1) align 4 %a.cast, i64 124, i1 false) + call void @bar_as1(ptr addrspace(1) %b.cast) ret void } -declare void @bar(i8*) -declare void @bar_as1(i8 addrspace(1)*) +declare void @bar(ptr) +declare void @bar_as1(ptr addrspace(1)) ;; Should be able to eliminate the alloca. define void @test3() { ; CHECK-LABEL: @test3( -; CHECK-NEXT: call void @bar(i8* nonnull getelementptr inbounds ([[T:%.*]], %T* @G, i64 0, i32 0)) #[[ATTR3:[0-9]+]] +; CHECK-NEXT: call void @bar(ptr nonnull @G) #[[ATTR3:[0-9]+]] ; CHECK-NEXT: ret void ; %A = alloca %T - %a = bitcast %T* %A to i8* - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%T* @G to i8*), i64 124, i1 false) - call void @bar(i8* %a) readonly + call void @llvm.memcpy.p0.p0.i64(ptr align 4 %A, ptr align 4 @G, i64 124, i1 false) + call void @bar(ptr %A) readonly ret void } define void @test3_addrspacecast() { ; CHECK-LABEL: @test3_addrspacecast( -; CHECK-NEXT: call void @bar(i8* nonnull getelementptr inbounds ([[T:%.*]], %T* @G, i64 0, i32 0)) #[[ATTR3]] +; CHECK-NEXT: call void @bar(ptr nonnull @G) #[[ATTR3]] ; CHECK-NEXT: ret void ; %A = alloca %T - %a = bitcast %T* %A to i8* - call void @llvm.memcpy.p0i8.p1i8.i64(i8* align 4 %a, i8 addrspace(1)* align 4 addrspacecast (%T* @G to i8 addrspace(1)*), i64 124, i1 false) - call void @bar(i8* %a) readonly + call void @llvm.memcpy.p0.p1.i64(ptr align 4 %A, ptr addrspace(1) align 4 addrspacecast (ptr @G to ptr addrspace(1)), i64 124, i1 false) + call void @bar(ptr %A) readonly ret void } define void @test4() { ; CHECK-LABEL: @test4( -; CHECK-NEXT: call void @baz(i8* nonnull byval(i8) getelementptr inbounds ([[T:%.*]], %T* @G, i64 0, i32 0)) +; CHECK-NEXT: call void @baz(ptr nonnull byval(i8) @G) ; CHECK-NEXT: ret void ; %A = alloca %T - %a = bitcast %T* %A to i8* - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%T* @G to i8*), i64 124, i1 false) - call void @baz(i8* byval(i8) %a) + call void @llvm.memcpy.p0.p0.i64(ptr align 4 %A, ptr align 4 @G, i64 124, i1 false) + call void @baz(ptr byval(i8) %A) ret void } -declare void @llvm.lifetime.start.p0i8(i64, i8*) +declare void @llvm.lifetime.start.p0(i64, ptr) define void @test5() { ; CHECK-LABEL: @test5( -; CHECK-NEXT: call void @baz(i8* nonnull byval(i8) getelementptr inbounds ([[T:%.*]], %T* @G, i64 0, i32 0)) +; CHECK-NEXT: call void @baz(ptr nonnull byval(i8) @G) ; CHECK-NEXT: ret void ; %A = alloca %T - %a = bitcast %T* %A to i8* - call void @llvm.lifetime.start.p0i8(i64 -1, i8* %a) - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%T* @G to i8*), i64 124, i1 false) - call void @baz(i8* byval(i8) %a) + call void @llvm.lifetime.start.p0(i64 -1, ptr %A) + call void @llvm.memcpy.p0.p0.i64(ptr align 4 %A, ptr align 4 @G, i64 124, i1 false) + call void @baz(ptr byval(i8) %A) ret void } -declare void @baz(i8* byval(i8)) +declare void @baz(ptr byval(i8)) define void @test6() { ; CHECK-LABEL: @test6( -; CHECK-NEXT: call void @bar(i8* nonnull bitcast ([2 x %U]* @H to i8*)) #[[ATTR3]] +; CHECK-NEXT: call void @bar(ptr nonnull @H) #[[ATTR3]] ; CHECK-NEXT: ret void ; %A = alloca %U, align 16 - %a = bitcast %U* %A to i8* - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 %a, i8* align 16 bitcast ([2 x %U]* @H to i8*), i64 20, i1 false) - call void @bar(i8* %a) readonly + call void @llvm.memcpy.p0.p0.i64(ptr align 16 %A, ptr align 16 @H, i64 20, i1 false) + call void @bar(ptr %A) readonly ret void } define void @test7() { ; CHECK-LABEL: @test7( -; CHECK-NEXT: call void @bar(i8* nonnull bitcast ([2 x %U]* @H to i8*)) #[[ATTR3]] +; CHECK-NEXT: call void @bar(ptr nonnull @H) #[[ATTR3]] ; CHECK-NEXT: ret void ; %A = alloca %U, align 16 - %a = bitcast %U* %A to i8* - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%U* getelementptr ([2 x %U], [2 x %U]* @H, i64 0, i32 0) to i8*), i64 20, i1 false) - call void @bar(i8* %a) readonly + call void @llvm.memcpy.p0.p0.i64(ptr align 4 %A, ptr align 4 @H, i64 20, i1 false) + call void @bar(ptr %A) readonly ret void } define void @test8() { ; CHECK-LABEL: @test8( ; CHECK-NEXT: [[AL:%.*]] = alloca [[U:%.*]], align 16 -; CHECK-NEXT: [[A:%.*]] = bitcast %U* [[AL]] to i8* -; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* noundef nonnull align 16 dereferenceable(20) [[A]], i8* noundef nonnull align 4 dereferenceable(20) bitcast (%U* getelementptr inbounds ([2 x %U], [2 x %U]* @H, i64 0, i64 1) to i8*), i64 20, i1 false) -; CHECK-NEXT: call void @bar(i8* nonnull [[A]]) #[[ATTR3]] +; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 16 dereferenceable(20) [[AL]], ptr noundef nonnull align 4 dereferenceable(20) getelementptr inbounds ([2 x %U], ptr @H, i64 0, i64 1), i64 20, i1 false) +; CHECK-NEXT: call void @bar(ptr nonnull [[AL]]) #[[ATTR3]] ; CHECK-NEXT: ret void ; %al = alloca %U, align 16 - %a = bitcast %U* %al to i8* - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%U* getelementptr ([2 x %U], [2 x %U]* @H, i64 0, i32 1) to i8*), i64 20, i1 false) - call void @bar(i8* %a) readonly + call void @llvm.memcpy.p0.p0.i64(ptr align 4 %al, ptr align 4 getelementptr ([2 x %U], ptr @H, i64 0, i32 1), i64 20, i1 false) + call void @bar(ptr %al) readonly ret void } @@ -250,39 +234,35 @@ define void @test8() { define void @test8_addrspacecast() { ; CHECK-LABEL: @test8_addrspacecast( ; CHECK-NEXT: [[AL:%.*]] = alloca [[U:%.*]], align 16 -; CHECK-NEXT: [[A:%.*]] = bitcast %U* [[AL]] to i8* -; CHECK-NEXT: call void @llvm.memcpy.p0i8.p1i8.i64(i8* noundef nonnull align 16 dereferenceable(20) [[A]], i8 addrspace(1)* noundef align 4 dereferenceable(20) addrspacecast (i8* bitcast (%U* getelementptr inbounds ([2 x %U], [2 x %U]* @H, i64 0, i64 1) to i8*) to i8 addrspace(1)*), i64 20, i1 false) -; CHECK-NEXT: call void @bar(i8* nonnull [[A]]) #[[ATTR3]] +; CHECK-NEXT: call void @llvm.memcpy.p0.p1.i64(ptr noundef nonnull align 16 dereferenceable(20) [[AL]], ptr addrspace(1) noundef align 4 dereferenceable(20) addrspacecast (ptr getelementptr inbounds ([2 x %U], ptr @H, i64 0, i64 1) to ptr addrspace(1)), i64 20, i1 false) +; CHECK-NEXT: call void @bar(ptr nonnull [[AL]]) #[[ATTR3]] ; CHECK-NEXT: ret void ; %Al = alloca %U, align 16 - %a = bitcast %U* %Al to i8* - call void @llvm.memcpy.p0i8.p1i8.i64(i8* align 4 %a, i8 addrspace(1)* align 4 addrspacecast (%U* getelementptr ([2 x %U], [2 x %U]* @H, i64 0, i32 1) to i8 addrspace(1)*), i64 20, i1 false) - call void @bar(i8* %a) readonly + call void @llvm.memcpy.p0.p1.i64(ptr align 4 %Al, ptr addrspace(1) align 4 addrspacecast (ptr getelementptr ([2 x %U], ptr @H, i64 0, i32 1) to ptr addrspace(1)), i64 20, i1 false) + call void @bar(ptr %Al) readonly ret void } define void @test9() { ; CHECK-LABEL: @test9( -; CHECK-NEXT: call void @bar(i8* nonnull bitcast (%U* getelementptr inbounds ([2 x %U], [2 x %U]* @H, i64 0, i64 1) to i8*)) #[[ATTR3]] +; CHECK-NEXT: call void @bar(ptr nonnull getelementptr inbounds ([2 x %U], ptr @H, i64 0, i64 1)) #[[ATTR3]] ; CHECK-NEXT: ret void ; %A = alloca %U, align 4 - %a = bitcast %U* %A to i8* - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 %a, i8* align 4 bitcast (%U* getelementptr ([2 x %U], [2 x %U]* @H, i64 0, i32 1) to i8*), i64 20, i1 false) - call void @bar(i8* %a) readonly + call void @llvm.memcpy.p0.p0.i64(ptr align 4 %A, ptr align 4 getelementptr ([2 x %U], ptr @H, i64 0, i32 1), i64 20, i1 false) + call void @bar(ptr %A) readonly ret void } define void @test9_addrspacecast() { ; CHECK-LABEL: @test9_addrspacecast( -; CHECK-NEXT: call void @bar(i8* nonnull bitcast (%U* getelementptr inbounds ([2 x %U], [2 x %U]* @H, i64 0, i64 1) to i8*)) #[[ATTR3]] +; CHECK-NEXT: call void @bar(ptr nonnull getelementptr inbounds ([2 x %U], ptr @H, i64 0, i64 1)) #[[ATTR3]] ; CHECK-NEXT: ret void ; %A = alloca %U, align 4 - %a = bitcast %U* %A to i8* - call void @llvm.memcpy.p0i8.p1i8.i64(i8* align 4 %a, i8 addrspace(1)* align 4 addrspacecast (%U* getelementptr ([2 x %U], [2 x %U]* @H, i64 0, i32 1) to i8 addrspace(1)*), i64 20, i1 false) - call void @bar(i8* %a) readonly + call void @llvm.memcpy.p0.p1.i64(ptr align 4 %A, ptr addrspace(1) align 4 addrspacecast (ptr getelementptr ([2 x %U], ptr @H, i64 0, i32 1) to ptr addrspace(1)), i64 20, i1 false) + call void @bar(ptr %A) readonly ret void } @@ -294,17 +274,14 @@ define void @test9_small_global() { ; CHECK-LABEL: @test9_small_global( ; CHECK-NEXT: entry: ; CHECK-NEXT: [[CC:%.*]] = alloca [1000000 x i8], align 16 -; CHECK-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [1000000 x i8], [1000000 x i8]* [[CC]], i64 0, i64 0 -; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* noundef nonnull align 16 dereferenceable(3) [[ARRAYDECAY]], i8* noundef nonnull align 16 dereferenceable(3) getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i64 0, i64 0), i64 3, i1 false) -; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* noundef nonnull align 16 dereferenceable(1000000) getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i64 0, i64 0), i8* noundef nonnull align 16 dereferenceable(1000000) [[ARRAYDECAY]], i64 1000000, i1 false) +; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 16 dereferenceable(3) [[CC]], ptr noundef nonnull align 16 dereferenceable(3) @_ZL3KKK, i64 3, i1 false) +; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 16 dereferenceable(1000000) @bbb, ptr noundef nonnull align 16 dereferenceable(1000000) [[CC]], i64 1000000, i1 false) ; CHECK-NEXT: ret void ; entry: %cc = alloca [1000000 x i8], align 16 - %cc.0..sroa_idx = getelementptr inbounds [1000000 x i8], [1000000 x i8]* %cc, i64 0, i64 0 - %arraydecay = getelementptr inbounds [1000000 x i8], [1000000 x i8]* %cc, i32 0, i32 0 - call void @llvm.memcpy.p0i8.p0i8.i64(i8* %arraydecay, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i32 0, i32 0), i64 3, i1 false) - call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 16 getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i32 0, i32 0), i8* align 16 %arraydecay, i64 1000000, i1 false) + call void @llvm.memcpy.p0.p0.i64(ptr %cc, ptr @_ZL3KKK, i64 3, i1 false) + call void @llvm.memcpy.p0.p0.i64(ptr align 16 @bbb, ptr align 16 %cc, i64 1000000, i1 false) ret void } @@ -312,15 +289,13 @@ entry: define void @test10_same_global() { ; CHECK-LABEL: @test10_same_global( ; CHECK-NEXT: entry: -; CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* noundef nonnull align 16 dereferenceable(3) getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i64 0, i64 0), i8* noundef nonnull align 16 dereferenceable(3) getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i64 0, i64 0), i64 3, i1 false) +; CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr noundef nonnull align 16 dereferenceable(3) @bbb, ptr noundef nonnull align 16 dereferenceable(3) @_ZL3KKK, i64 3, i1 false) ; CHECK-NEXT: ret void ; entry: %cc = alloca [3 x i8], align 1 - %cc.0..sroa_idx = getelementptr inbounds [3 x i8], [3 x i8]* %cc, i64 0, i64 0 - %arraydecay = getelementptr inbounds [3 x i8], [3 x i8]* %cc, i32 0, i32 0 - call void @llvm.memcpy.p0i8.p0i8.i64(i8* %arraydecay, i8* getelementptr inbounds ([3 x i8], [3 x i8]* @_ZL3KKK, i32 0, i32 0), i64 3, i1 false) - call void @llvm.memcpy.p0i8.p0i8.i64(i8* getelementptr inbounds ([1000000 x i8], [1000000 x i8]* @bbb, i32 0, i32 0), i8* %arraydecay, i64 3, i1 false) + call void @llvm.memcpy.p0.p0.i64(ptr %cc, ptr @_ZL3KKK, i64 3, i1 false) + call void @llvm.memcpy.p0.p0.i64(ptr @bbb, ptr %cc, i64 3, i1 false) ret void } @@ -328,18 +303,17 @@ entry: define float @test11(i64 %i) { ; CHECK-LABEL: @test11( ; CHECK-NEXT: entry: -; CHECK-NEXT: [[G:%.*]] = getelementptr [4 x float], [4 x float] addrspace(1)* @I, i64 0, i64 [[I:%.*]] -; CHECK-NEXT: [[R:%.*]] = load float, float addrspace(1)* [[G]], align 4 +; CHECK-NEXT: [[G:%.*]] = getelementptr [4 x float], ptr addrspace(1) @I, i64 0, i64 [[I:%.*]] +; CHECK-NEXT: [[R:%.*]] = load float, ptr addrspace(1) [[G]], align 4 ; CHECK-NEXT: ret float [[R]] ; entry: %a = alloca [4 x float], align 4 - %b = bitcast [4 x float]* %a to i8* - call void @llvm.lifetime.start.p0i8(i64 16, i8* %b) - call void @llvm.memcpy.p0i8.p1i8.i64(i8* align 4 %b, i8 addrspace(1)* align 4 bitcast ([4 x float] addrspace(1)* @I to i8 addrspace(1)*), i64 16, i1 false) - %g = getelementptr inbounds [4 x float], [4 x float]* %a, i64 0, i64 %i - %r = load float, float* %g, align 4 + call void @llvm.lifetime.start.p0(i64 16, ptr %a) + call void @llvm.memcpy.p0.p1.i64(ptr align 4 %a, ptr addrspace(1) align 4 @I, i64 16, i1 false) + %g = getelementptr inbounds [4 x float], ptr %a, i64 0, i64 %i + %r = load float, ptr %g, align 4 ret float %r } @@ -348,21 +322,19 @@ define float @test11_volatile(i64 %i) { ; CHECK-LABEL: @test11_volatile( ; CHECK-NEXT: entry: ; CHECK-NEXT: [[A:%.*]] = alloca [4 x float], align 4 -; CHECK-NEXT: [[B:%.*]] = bitcast [4 x float]* [[A]] to i8* -; CHECK-NEXT: call void @llvm.lifetime.start.p0i8(i64 16, i8* nonnull [[B]]) -; CHECK-NEXT: call void @llvm.memcpy.p0i8.p1i8.i64(i8* align 4 [[B]], i8 addrspace(1)* align 4 bitcast ([4 x float] addrspace(1)* @I to i8 addrspace(1)*), i64 16, i1 true) -; CHECK-NEXT: [[G:%.*]] = getelementptr inbounds [4 x float], [4 x float]* [[A]], i64 0, i64 [[I:%.*]] -; CHECK-NEXT: [[R:%.*]] = load float, float* [[G]], align 4 +; CHECK-NEXT: call void @llvm.lifetime.start.p0(i64 16, ptr nonnull [[A]]) +; CHECK-NEXT: call void @llvm.memcpy.p0.p1.i64(ptr align 4 [[A]], ptr addrspace(1) align 4 @I, i64 16, i1 true) +; CHECK-NEXT: [[G:%.*]] = getelementptr inbounds [4 x float], ptr [[A]], i64 0, i64 [[I:%.*]] +; CHECK-NEXT: [[R:%.*]] = load float, ptr [[G]], align 4 ; CHECK-NEXT: ret float [[R]] ; entry: %a = alloca [4 x float], align 4 - %b = bitcast [4 x float]* %a to i8* - call void @llvm.lifetime.start.p0i8(i64 16, i8* %b) - call void @llvm.memcpy.p0i8.p1i8.i64(i8* align 4 %b, i8 addrspace(1)* align 4 bitcast ([4 x float] addrspace(1)* @I to i8 addrspace(1)*), i64 16, i1 true) - %g = getelementptr inbounds [4 x float], [4 x float]* %a, i64 0, i64 %i - %r = load float, float* %g, align 4 + call void @llvm.lifetime.start.p0(i64 16, ptr %a) + call void @llvm.memcpy.p0.p1.i64(ptr align 4 %a, ptr addrspace(1) align 4 @I, i64 16, i1 true) + %g = getelementptr inbounds [4 x float], ptr %a, i64 0, i64 %i + %r = load float, ptr %g, align 4 ret float %r } -- 2.7.4