From 42b7061f1f0a2c52c8302a25ec656f990236b933 Mon Sep 17 00:00:00 2001 From: Nikita Popov Date: Wed, 24 Jan 2024 11:20:16 +0100 Subject: [PATCH] [LAA] Drop alias scope metadata that is not valid across iterations (#79161) LAA currently adds memory locations with their original AATags to AST. However, scoped alias AATags may be valid only within one loop iteration, while LAA reasons across iterations. Fix this by determining which alias scopes are defined inside the loop, and drop AATags that reference these scopes. Fixes https://github.com/llvm/llvm-project/issues/79137. (cherry picked from commit cd7ea4ea657ea41b42fcbd0e6b33faa46608d18e) --- llvm/lib/Analysis/LoopAccessAnalysis.cpp | 53 ++++++++++-- .../LoopAccessAnalysis/noalias-scope-decl.ll | 11 ++- .../X86/loop-vectorizer-noalias.ll | 83 ++----------------- 3 files changed, 63 insertions(+), 84 deletions(-) diff --git a/llvm/lib/Analysis/LoopAccessAnalysis.cpp b/llvm/lib/Analysis/LoopAccessAnalysis.cpp index 7e67c90152829d..dd6b88fee415a7 100644 --- a/llvm/lib/Analysis/LoopAccessAnalysis.cpp +++ b/llvm/lib/Analysis/LoopAccessAnalysis.cpp @@ -657,16 +657,18 @@ class AccessAnalysis { AccessAnalysis(Loop *TheLoop, AAResults *AA, LoopInfo *LI, MemoryDepChecker::DepCandidates &DA, - PredicatedScalarEvolution &PSE) - : TheLoop(TheLoop), BAA(*AA), AST(BAA), LI(LI), DepCands(DA), PSE(PSE) { + PredicatedScalarEvolution &PSE, + SmallPtrSetImpl &LoopAliasScopes) + : TheLoop(TheLoop), BAA(*AA), AST(BAA), LI(LI), DepCands(DA), PSE(PSE), + LoopAliasScopes(LoopAliasScopes) { // We're analyzing dependences across loop iterations. BAA.enableCrossIterationMode(); } /// Register a load and whether it is only read from. void addLoad(MemoryLocation &Loc, Type *AccessTy, bool IsReadOnly) { - Value *Ptr = const_cast(Loc.Ptr); - AST.add(Loc.getWithNewSize(LocationSize::beforeOrAfterPointer())); + Value *Ptr = const_cast(Loc.Ptr); + AST.add(adjustLoc(Loc)); Accesses[MemAccessInfo(Ptr, false)].insert(AccessTy); if (IsReadOnly) ReadOnlyPtr.insert(Ptr); @@ -674,8 +676,8 @@ class AccessAnalysis { /// Register a store. void addStore(MemoryLocation &Loc, Type *AccessTy) { - Value *Ptr = const_cast(Loc.Ptr); - AST.add(Loc.getWithNewSize(LocationSize::beforeOrAfterPointer())); + Value *Ptr = const_cast(Loc.Ptr); + AST.add(adjustLoc(Loc)); Accesses[MemAccessInfo(Ptr, true)].insert(AccessTy); } @@ -731,6 +733,32 @@ class AccessAnalysis { private: typedef MapVector> PtrAccessMap; + /// Adjust the MemoryLocation so that it represents accesses to this + /// location across all iterations, rather than a single one. + MemoryLocation adjustLoc(MemoryLocation Loc) const { + // The accessed location varies within the loop, but remains within the + // underlying object. + Loc.Size = LocationSize::beforeOrAfterPointer(); + Loc.AATags.Scope = adjustAliasScopeList(Loc.AATags.Scope); + Loc.AATags.NoAlias = adjustAliasScopeList(Loc.AATags.NoAlias); + return Loc; + } + + /// Drop alias scopes that are only valid within a single loop iteration. + MDNode *adjustAliasScopeList(MDNode *ScopeList) const { + if (!ScopeList) + return nullptr; + + // For the sake of simplicity, drop the whole scope list if any scope is + // iteration-local. + if (any_of(ScopeList->operands(), [&](Metadata *Scope) { + return LoopAliasScopes.contains(cast(Scope)); + })) + return nullptr; + + return ScopeList; + } + /// Go over all memory access and check whether runtime pointer checks /// are needed and build sets of dependency check candidates. void processMemAccesses(); @@ -775,6 +803,10 @@ class AccessAnalysis { PredicatedScalarEvolution &PSE; DenseMap> UnderlyingObjects; + + /// Alias scopes that are declared inside the loop, and as such not valid + /// across iterations. + SmallPtrSetImpl &LoopAliasScopes; }; } // end anonymous namespace @@ -2283,6 +2315,7 @@ void LoopAccessInfo::analyzeLoop(AAResults *AA, LoopInfo *LI, // Holds the Load and Store instructions. SmallVector Loads; SmallVector Stores; + SmallPtrSet LoopAliasScopes; // Holds all the different accesses in the loop. unsigned NumReads = 0; @@ -2326,6 +2359,11 @@ void LoopAccessInfo::analyzeLoop(AAResults *AA, LoopInfo *LI, if (HasComplexMemInst) continue; + // Record alias scopes defined inside the loop. + if (auto *Decl = dyn_cast(&I)) + for (Metadata *Op : Decl->getScopeList()->operands()) + LoopAliasScopes.insert(cast(Op)); + // Many math library functions read the rounding mode. We will only // vectorize a loop if it contains known function calls that don't set // the flag. Therefore, it is safe to ignore this read from memory. @@ -2407,7 +2445,8 @@ void LoopAccessInfo::analyzeLoop(AAResults *AA, LoopInfo *LI, } MemoryDepChecker::DepCandidates DependentAccesses; - AccessAnalysis Accesses(TheLoop, AA, LI, DependentAccesses, *PSE); + AccessAnalysis Accesses(TheLoop, AA, LI, DependentAccesses, *PSE, + LoopAliasScopes); // Holds the analyzed pointers. We don't want to call getUnderlyingObjects // multiple times on the same object. If the ptr is accessed twice, once diff --git a/llvm/test/Analysis/LoopAccessAnalysis/noalias-scope-decl.ll b/llvm/test/Analysis/LoopAccessAnalysis/noalias-scope-decl.ll index 98bb5f99a40a1e..fb296f5089422d 100644 --- a/llvm/test/Analysis/LoopAccessAnalysis/noalias-scope-decl.ll +++ b/llvm/test/Analysis/LoopAccessAnalysis/noalias-scope-decl.ll @@ -7,8 +7,17 @@ define void @test_scope_in_loop(ptr %arg, i64 %num) { ; CHECK-LABEL: 'test_scope_in_loop' ; CHECK-NEXT: loop: -; CHECK-NEXT: Memory dependences are safe +; CHECK-NEXT: Report: unsafe dependent memory operations in loop. Use #pragma clang loop distribute(enable) to allow loop distribution to attempt to isolate the offending operations into a separate loop +; CHECK-NEXT: Backward loop carried data dependence. ; CHECK-NEXT: Dependences: +; CHECK-NEXT: Backward: +; CHECK-NEXT: %load.prev = load i8, ptr %prev.ptr, align 1, !alias.scope !0, !noalias !3 -> +; CHECK-NEXT: store i8 %add, ptr %cur.ptr, align 1, !alias.scope !3 +; CHECK-EMPTY: +; CHECK-NEXT: Forward: +; CHECK-NEXT: %load.cur = load i8, ptr %cur.ptr, align 1, !alias.scope !3 -> +; CHECK-NEXT: store i8 %add, ptr %cur.ptr, align 1, !alias.scope !3 +; CHECK-EMPTY: ; CHECK-NEXT: Run-time memory checks: ; CHECK-NEXT: Grouped accesses: ; CHECK-EMPTY: diff --git a/llvm/test/Transforms/PhaseOrdering/X86/loop-vectorizer-noalias.ll b/llvm/test/Transforms/PhaseOrdering/X86/loop-vectorizer-noalias.ll index 846787f721ba7e..5c85c0d21f59f7 100644 --- a/llvm/test/Transforms/PhaseOrdering/X86/loop-vectorizer-noalias.ll +++ b/llvm/test/Transforms/PhaseOrdering/X86/loop-vectorizer-noalias.ll @@ -14,89 +14,25 @@ entry: } ; This loop should not get vectorized. -; FIXME: This is a miscompile. define void @accsum(ptr noundef %vals, i64 noundef %num) #0 { ; CHECK-LABEL: define void @accsum( ; CHECK-SAME: ptr nocapture noundef [[VALS:%.*]], i64 noundef [[NUM:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { ; CHECK-NEXT: entry: ; CHECK-NEXT: [[CMP1:%.*]] = icmp ugt i64 [[NUM]], 1 -; CHECK-NEXT: br i1 [[CMP1]], label [[ITER_CHECK:%.*]], label [[FOR_END:%.*]] -; CHECK: iter.check: -; CHECK-NEXT: [[TMP0:%.*]] = add i64 [[NUM]], -1 -; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[NUM]], 9 -; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[FOR_BODY_PREHEADER:%.*]], label [[VECTOR_MAIN_LOOP_ITER_CHECK:%.*]] -; CHECK: vector.main.loop.iter.check: -; CHECK-NEXT: [[MIN_ITERS_CHECK3:%.*]] = icmp ult i64 [[NUM]], 33 -; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK3]], label [[VEC_EPILOG_PH:%.*]], label [[VECTOR_PH:%.*]] -; CHECK: vector.ph: -; CHECK-NEXT: [[N_VEC:%.*]] = and i64 [[TMP0]], -32 -; CHECK-NEXT: br label [[VECTOR_BODY:%.*]] -; CHECK: vector.body: -; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ] -; CHECK-NEXT: [[OFFSET_IDX:%.*]] = or disjoint i64 [[INDEX]], 1 -; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[VALS]], i64 [[OFFSET_IDX]] -; CHECK-NEXT: [[TMP2:%.*]] = getelementptr i8, ptr [[TMP1]], i64 -1 -; CHECK-NEXT: tail call void @llvm.experimental.noalias.scope.decl(metadata [[META0:![0-9]+]]) -; CHECK-NEXT: tail call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]]) -; CHECK-NEXT: [[TMP3:%.*]] = getelementptr i8, ptr [[TMP1]], i64 15 -; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i8>, ptr [[TMP2]], align 1, !alias.scope [[META3]], !noalias [[META0]] -; CHECK-NEXT: [[WIDE_LOAD4:%.*]] = load <16 x i8>, ptr [[TMP3]], align 1, !alias.scope [[META3]], !noalias [[META0]] -; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i64 16 -; CHECK-NEXT: [[WIDE_LOAD5:%.*]] = load <16 x i8>, ptr [[TMP1]], align 1, !alias.scope [[META0]], !noalias [[META3]] -; CHECK-NEXT: [[WIDE_LOAD6:%.*]] = load <16 x i8>, ptr [[TMP4]], align 1, !alias.scope [[META0]], !noalias [[META3]] -; CHECK-NEXT: [[TMP5:%.*]] = add <16 x i8> [[WIDE_LOAD5]], [[WIDE_LOAD]] -; CHECK-NEXT: [[TMP6:%.*]] = add <16 x i8> [[WIDE_LOAD6]], [[WIDE_LOAD4]] -; CHECK-NEXT: store <16 x i8> [[TMP5]], ptr [[TMP1]], align 1, !alias.scope [[META0]], !noalias [[META3]] -; CHECK-NEXT: store <16 x i8> [[TMP6]], ptr [[TMP4]], align 1, !alias.scope [[META0]], !noalias [[META3]] -; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 32 -; CHECK-NEXT: [[TMP7:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] -; CHECK-NEXT: br i1 [[TMP7]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP5:![0-9]+]] -; CHECK: middle.block: -; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]] -; CHECK-NEXT: br i1 [[CMP_N]], label [[FOR_END]], label [[VEC_EPILOG_ITER_CHECK:%.*]] -; CHECK: vec.epilog.iter.check: -; CHECK-NEXT: [[IND_END9:%.*]] = or disjoint i64 [[N_VEC]], 1 -; CHECK-NEXT: [[N_VEC_REMAINING:%.*]] = and i64 [[TMP0]], 24 -; CHECK-NEXT: [[MIN_EPILOG_ITERS_CHECK:%.*]] = icmp eq i64 [[N_VEC_REMAINING]], 0 -; CHECK-NEXT: br i1 [[MIN_EPILOG_ITERS_CHECK]], label [[FOR_BODY_PREHEADER]], label [[VEC_EPILOG_PH]] -; CHECK: vec.epilog.ph: -; CHECK-NEXT: [[VEC_EPILOG_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], [[VEC_EPILOG_ITER_CHECK]] ], [ 0, [[VECTOR_MAIN_LOOP_ITER_CHECK]] ] -; CHECK-NEXT: [[N_VEC8:%.*]] = and i64 [[TMP0]], -8 -; CHECK-NEXT: [[IND_END:%.*]] = or disjoint i64 [[N_VEC8]], 1 -; CHECK-NEXT: br label [[VEC_EPILOG_VECTOR_BODY:%.*]] -; CHECK: vec.epilog.vector.body: -; CHECK-NEXT: [[INDEX11:%.*]] = phi i64 [ [[VEC_EPILOG_RESUME_VAL]], [[VEC_EPILOG_PH]] ], [ [[INDEX_NEXT15:%.*]], [[VEC_EPILOG_VECTOR_BODY]] ] -; CHECK-NEXT: [[OFFSET_IDX12:%.*]] = or disjoint i64 [[INDEX11]], 1 -; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds i8, ptr [[VALS]], i64 [[OFFSET_IDX12]] -; CHECK-NEXT: [[TMP9:%.*]] = getelementptr i8, ptr [[TMP8]], i64 -1 -; CHECK-NEXT: tail call void @llvm.experimental.noalias.scope.decl(metadata [[META0]]) -; CHECK-NEXT: tail call void @llvm.experimental.noalias.scope.decl(metadata [[META3]]) -; CHECK-NEXT: [[WIDE_LOAD13:%.*]] = load <8 x i8>, ptr [[TMP9]], align 1, !alias.scope [[META3]], !noalias [[META0]] -; CHECK-NEXT: [[WIDE_LOAD14:%.*]] = load <8 x i8>, ptr [[TMP8]], align 1, !alias.scope [[META0]], !noalias [[META3]] -; CHECK-NEXT: [[TMP10:%.*]] = add <8 x i8> [[WIDE_LOAD14]], [[WIDE_LOAD13]] -; CHECK-NEXT: store <8 x i8> [[TMP10]], ptr [[TMP8]], align 1, !alias.scope [[META0]], !noalias [[META3]] -; CHECK-NEXT: [[INDEX_NEXT15]] = add nuw i64 [[INDEX11]], 8 -; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT15]], [[N_VEC8]] -; CHECK-NEXT: br i1 [[TMP11]], label [[VEC_EPILOG_MIDDLE_BLOCK:%.*]], label [[VEC_EPILOG_VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]] -; CHECK: vec.epilog.middle.block: -; CHECK-NEXT: [[CMP_N10:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC8]] -; CHECK-NEXT: br i1 [[CMP_N10]], label [[FOR_END]], label [[FOR_BODY_PREHEADER]] +; CHECK-NEXT: br i1 [[CMP1]], label [[FOR_BODY_PREHEADER:%.*]], label [[FOR_END:%.*]] ; CHECK: for.body.preheader: -; CHECK-NEXT: [[I_02_PH:%.*]] = phi i64 [ 1, [[ITER_CHECK]] ], [ [[IND_END9]], [[VEC_EPILOG_ITER_CHECK]] ], [ [[IND_END]], [[VEC_EPILOG_MIDDLE_BLOCK]] ] +; CHECK-NEXT: [[LOAD_INITIAL:%.*]] = load i8, ptr [[VALS]], align 1 ; CHECK-NEXT: br label [[FOR_BODY:%.*]] ; CHECK: for.body: -; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[I_02_PH]], [[FOR_BODY_PREHEADER]] ] +; CHECK-NEXT: [[STORE_FORWARDED:%.*]] = phi i8 [ [[LOAD_INITIAL]], [[FOR_BODY_PREHEADER]] ], [ [[ADD_I:%.*]], [[FOR_BODY]] ] +; CHECK-NEXT: [[I_02:%.*]] = phi i64 [ 1, [[FOR_BODY_PREHEADER]] ], [ [[INC:%.*]], [[FOR_BODY]] ] ; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[VALS]], i64 [[I_02]] -; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr i8, ptr [[ARRAYIDX]], i64 -1 -; CHECK-NEXT: tail call void @llvm.experimental.noalias.scope.decl(metadata [[META0]]) -; CHECK-NEXT: tail call void @llvm.experimental.noalias.scope.decl(metadata [[META3]]) -; CHECK-NEXT: [[TMP12:%.*]] = load i8, ptr [[ARRAYIDX1]], align 1, !alias.scope [[META3]], !noalias [[META0]] -; CHECK-NEXT: [[TMP13:%.*]] = load i8, ptr [[ARRAYIDX]], align 1, !alias.scope [[META0]], !noalias [[META3]] -; CHECK-NEXT: [[ADD_I:%.*]] = add i8 [[TMP13]], [[TMP12]] +; CHECK-NEXT: [[TMP0:%.*]] = load i8, ptr [[ARRAYIDX]], align 1, !alias.scope [[META0:![0-9]+]], !noalias [[META3:![0-9]+]] +; CHECK-NEXT: [[ADD_I]] = add i8 [[TMP0]], [[STORE_FORWARDED]] ; CHECK-NEXT: store i8 [[ADD_I]], ptr [[ARRAYIDX]], align 1, !alias.scope [[META0]], !noalias [[META3]] ; CHECK-NEXT: [[INC]] = add nuw i64 [[I_02]], 1 ; CHECK-NEXT: [[EXITCOND_NOT:%.*]] = icmp eq i64 [[INC]], [[NUM]] -; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]] +; CHECK-NEXT: br i1 [[EXITCOND_NOT]], label [[FOR_END]], label [[FOR_BODY]] ; CHECK: for.end: ; CHECK-NEXT: ret void ; @@ -139,9 +75,4 @@ attributes #0 = { "target-features"="+cmov,+cx8,+fxsr,+mmx,+sse,+sse2,+x87"} ; CHECK: [[META2]] = distinct !{[[META2]], !"acc"} ; CHECK: [[META3]] = !{[[META4:![0-9]+]]} ; CHECK: [[META4]] = distinct !{[[META4]], [[META2]], !"acc: %prev"} -; CHECK: [[LOOP5]] = distinct !{[[LOOP5]], [[META6:![0-9]+]], [[META7:![0-9]+]]} -; CHECK: [[META6]] = !{!"llvm.loop.isvectorized", i32 1} -; CHECK: [[META7]] = !{!"llvm.loop.unroll.runtime.disable"} -; CHECK: [[LOOP8]] = distinct !{[[LOOP8]], [[META6]], [[META7]]} -; CHECK: [[LOOP9]] = distinct !{[[LOOP9]], [[META7]], [[META6]]} ;.