#include "llvm/Analysis/LoopAccessAnalysis.h"
#include "llvm/ADT/APInt.h"
#include "llvm/ADT/DenseMap.h"
#include "llvm/ADT/DepthFirstIterator.h"
#include "llvm/ADT/EquivalenceClasses.h"
#include "llvm/ADT/PointerIntPair.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/ADT/SetVector.h"
#include "llvm/ADT/SmallPtrSet.h"
#include "llvm/ADT/SmallSet.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/iterator_range.h"
#include "llvm/Analysis/AliasAnalysis.h"
#include "llvm/Analysis/AliasSetTracker.h"
#include "llvm/Analysis/LoopAnalysisManager.h"
#include "llvm/Analysis/LoopInfo.h"
#include "llvm/Analysis/LoopIterator.h"
#include "llvm/Analysis/MemoryLocation.h"
#include "llvm/Analysis/OptimizationRemarkEmitter.h"
#include "llvm/Analysis/ScalarEvolution.h"
#include "llvm/Analysis/ScalarEvolutionExpressions.h"
#include "llvm/Analysis/TargetLibraryInfo.h"
#include "llvm/Analysis/ValueTracking.h"
#include "llvm/Analysis/VectorUtils.h"
#include "llvm/IR/BasicBlock.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/DebugLoc.h"
#include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/DiagnosticInfo.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/InstrTypes.h"
#include "llvm/IR/Instruction.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/Operator.h"
#include "llvm/IR/PassManager.h"
#include "llvm/IR/PatternMatch.h"
#include "llvm/IR/Type.h"
#include "llvm/IR/Value.h"
#include "llvm/IR/ValueHandle.h"
#include "llvm/InitializePasses.h"
#include "llvm/Pass.h"
#include "llvm/Support/Casting.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/raw_ostream.h"
#include <algorithm>
#include <cassert>
#include <cstdint>
#include <iterator>
#include <utility>
#include <vector>

Namespaces
	llvm
	This is an optimization pass for GlobalISel generic memory operations.

Macros
#define	DEBUG_TYPE "loop-accesses"

#define	LAA_NAME "loop-accesses"

Functions
static const SCEV *	getMinFromExprs (const SCEV I, const SCEV J, ScalarEvolution *SE)
	Compare `I` and `J` and return the minimum. More...

static bool	hasComputableBounds (PredicatedScalarEvolution &PSE, Value Ptr, const SCEV PtrScev, Loop *L, bool Assume)
	Check whether a pointer can participate in a runtime bounds check. More...

static bool	isNoWrap (PredicatedScalarEvolution &PSE, const ValueToValueMap &Strides, Value Ptr, Type AccessTy, Loop *L)
	Check whether a pointer address cannot wrap. More...

static void	visitPointers (Value StartPtr, const Loop &InnermostLoop, function_ref< void(Value )> AddPointer)

static void	findForkedSCEVs (ScalarEvolution SE, const Loop L, Value Ptr, SmallVectorImpl< PointerIntPair< const SCEV , 1, bool >> &ScevList, unsigned Depth)

static SmallVector< PointerIntPair< const SCEV *, 1, bool > >	findForkedPointer (PredicatedScalarEvolution &PSE, const ValueToValueMap &StridesMap, Value Ptr, const Loop L)

static bool	isInBoundsGep (Value *Ptr)

static bool	isNoWrapAddRec (Value Ptr, const SCEVAddRecExpr AR, PredicatedScalarEvolution &PSE, const Loop *L)
	Return true if an AddRec pointer `Ptr` is unsigned non-wrapping, i.e. More...

static bool	isSafeDependenceDistance (const DataLayout &DL, ScalarEvolution &SE, const SCEV &BackedgeTakenCount, const SCEV &Dist, uint64_t Stride, uint64_t TypeByteSize)
	Given a dependence-distance `Dist` between two memory accesses, that have the same stride whose absolute value is given in `Stride`, and that have the same type size `TypeByteSize`, in a loop whose takenCount is `BackedgeTakenCount`, check if it is possible to prove statically that the dependence distance is larger than the range that the accesses will travel through the execution of the loop. More...

static bool	areStridedAccessesIndependent (uint64_t Distance, uint64_t Stride, uint64_t TypeByteSize)
	Check the dependence for two accesses with the same stride `Stride`. More...

Pass *	llvm::createLAAPass ()

Variables
static cl::opt< unsigned, true >	VectorizationFactor ("force-vector-width", cl::Hidden, cl::desc("Sets the SIMD width. Zero is autoselect."), cl::location(VectorizerParams::VectorizationFactor))

static cl::opt< unsigned, true >	VectorizationInterleave ("force-vector-interleave", cl::Hidden, cl::desc("Sets the vectorization interleave count. " "Zero is autoselect."), cl::location(VectorizerParams::VectorizationInterleave))

static cl::opt< unsigned, true >	RuntimeMemoryCheckThreshold ("runtime-memory-check-threshold", cl::Hidden, cl::desc("When performing memory disambiguation checks at runtime do not " "generate more than this number of comparisons (default = 8)."), cl::location(VectorizerParams::RuntimeMemoryCheckThreshold), cl::init(8))

static cl::opt< unsigned >	MemoryCheckMergeThreshold ("memory-check-merge-threshold", cl::Hidden, cl::desc("Maximum number of comparisons done when trying to merge " "runtime memory checks. (default = 100)"), cl::init(100))
	The maximum iterations used to merge memory checks. More...

static cl::opt< unsigned >	MaxDependences ("max-dependences", cl::Hidden, cl::desc("Maximum number of dependences collected by " "loop-access analysis (default = 100)"), cl::init(100))
	We collect dependences up to this threshold. More...

static cl::opt< bool >	EnableMemAccessVersioning ("enable-mem-access-versioning", cl::init(true), cl::Hidden, cl::desc("Enable symbolic stride memory access versioning"))
	This enables versioning on the strides of symbolically striding memory accesses in code like the following. More...

static cl::opt< bool >	EnableForwardingConflictDetection ("store-to-load-forwarding-conflict-detection", cl::Hidden, cl::desc("Enable conflict detection in loop-access analysis"), cl::init(true))
	Enable store-to-load forwarding conflict detection. More...

static cl::opt< unsigned >	MaxForkedSCEVDepth ("max-forked-scev-depth", cl::Hidden, cl::desc("Maximum recursion depth when finding forked SCEVs (default = 5)"), cl::init(5))

static const char	laa_name [] = "Loop Access Analysis"

Macro Definition Documentation

◆ DEBUG_TYPE

#define DEBUG_TYPE "loop-accesses"

Definition at line 72 of file LoopAccessAnalysis.cpp.

◆ LAA_NAME

#define LAA_NAME "loop-accesses"

Definition at line 2717 of file LoopAccessAnalysis.cpp.

Function Documentation

◆ areStridedAccessesIndependent()

static bool areStridedAccessesIndependent	(	uint64_t	Distance,
		uint64_t	Stride,
		uint64_t	TypeByteSize
	)

static

Check the dependence for two accesses with the same stride Stride.

Distance is the positive distance and TypeByteSize is type size in bytes.

Returns: true if they are independent.

Definition at line 1799 of file LoopAccessAnalysis.cpp.

References assert().

◆ findForkedPointer()

static SmallVector<PointerIntPair<const SCEV *, 1, bool> > findForkedPointer	(	PredicatedScalarEvolution &	PSE,
		const ValueToValueMap &	StridesMap,
		Value *	Ptr,
		const Loop *	L
	)

static

Definition at line 953 of file LoopAccessAnalysis.cpp.

References assert(), llvm::dbgs(), findForkedSCEVs(), llvm::PredicatedScalarEvolution::getSE(), llvm::ScalarEvolution::isLoopInvariant(), llvm::ScalarEvolution::isSCEVable(), LLVM_DEBUG, MaxForkedSCEVDepth, Ptr, and llvm::replaceSymbolicStrideSCEV().

◆ findForkedSCEVs()

static void findForkedSCEVs	(	ScalarEvolution *	SE,
		const Loop *	L,
		Value *	Ptr,
		SmallVectorImpl< PointerIntPair< const SCEV *, 1, bool >> &	ScevList,
		unsigned	Depth
	)

static

◆ getMinFromExprs()

static const SCEV* getMinFromExprs	(	const SCEV *	I,
		const SCEV *	J,
		ScalarEvolution *	SE
	)

static

Compare I and J and return the minimum.

Return nullptr in case we couldn't find an answer.

Definition at line 369 of file LoopAccessAnalysis.cpp.

References llvm::ScalarEvolution::getMinusSCEV(), and I.

◆ hasComputableBounds()

static bool hasComputableBounds	(	PredicatedScalarEvolution &	PSE,
		Value *	Ptr,
		const SCEV *	PtrScev,
		Loop *	L,
		bool	Assume
	)

static

Check whether a pointer can participate in a runtime bounds check.

If Assume, try harder to prove that we can compute the bounds of Ptr by adding run-time checks (overflow checks) if necessary.

Definition at line 742 of file LoopAccessAnalysis.cpp.

References llvm::PredicatedScalarEvolution::getAsAddRec(), llvm::PredicatedScalarEvolution::getSE(), llvm::SCEVAddRecExpr::isAffine(), llvm::ScalarEvolution::isLoopInvariant(), and Ptr.

◆ isInBoundsGep()

static bool isInBoundsGep ( Value * Ptr )

static

Definition at line 1314 of file LoopAccessAnalysis.cpp.

References GEP, and Ptr.

Referenced by llvm::getPtrStride().

◆ isNoWrap()

static bool isNoWrap	(	PredicatedScalarEvolution &	PSE,
		const ValueToValueMap &	Strides,
		Value *	Ptr,
		Type *	AccessTy,
		Loop *	L
	)

static

Check whether a pointer address cannot wrap.

Definition at line 760 of file LoopAccessAnalysis.cpp.

References llvm::getPtrStride(), llvm::PredicatedScalarEvolution::getSCEV(), llvm::PredicatedScalarEvolution::getSE(), llvm::PredicatedScalarEvolution::hasNoOverflow(), llvm::SCEVWrapPredicate::IncrementNUSW, llvm::ScalarEvolution::isLoopInvariant(), and Ptr.

◆ isNoWrapAddRec()

static bool isNoWrapAddRec	(	Value *	Ptr,
		const SCEVAddRecExpr *	AR,
		PredicatedScalarEvolution &	PSE,
		const Loop *	L
	)

static

Return true if an AddRec pointer Ptr is unsigned non-wrapping, i.e.

monotonically increasing/decreasing.

Definition at line 1322 of file LoopAccessAnalysis.cpp.

References llvm::SCEV::FlagNSW, GEP, llvm::SCEVNAryExpr::getNoWrapFlags(), llvm::PredicatedScalarEvolution::getSCEV(), Index, llvm::SCEV::NoWrapMask, and Ptr.

Referenced by llvm::getPtrStride().

◆ isSafeDependenceDistance()

static bool isSafeDependenceDistance	(	const DataLayout &	DL,
		ScalarEvolution &	SE,
		const SCEV &	BackedgeTakenCount,
		const SCEV &	Dist,
		uint64_t	Stride,
		uint64_t	TypeByteSize
	)

static

Given a dependence-distance Dist between two memory accesses, that have the same stride whose absolute value is given in Stride, and that have the same type size TypeByteSize, in a loop whose takenCount is BackedgeTakenCount, check if it is possible to prove statically that the dependence distance is larger than the range that the accesses will travel through the execution of the loop.

If so, return true; false otherwise. This is useful for example in loops such as the following (PR31098): for (i = 0; i < D; ++i) { = out[i]; out[i+D] = }

Definition at line 1739 of file LoopAccessAnalysis.cpp.

References DL, llvm::ScalarEvolution::getConstant(), llvm::ScalarEvolution::getMinusSCEV(), llvm::ScalarEvolution::getMulExpr(), llvm::ScalarEvolution::getNegativeSCEV(), llvm::ScalarEvolution::getNoopOrSignExtend(), llvm::SCEV::getType(), llvm::ScalarEvolution::getZeroExtendExpr(), and llvm::ScalarEvolution::isKnownPositive().

◆ visitPointers()

static void visitPointers	(	Value *	StartPtr,
		const Loop &	InnermostLoop,
		function_ref< void(Value *)>	AddPointer
	)

static

Definition at line 774 of file LoopAccessAnalysis.cpp.

References llvm::LoopBase< BlockT, LoopT >::contains(), llvm::LoopBase< BlockT, LoopT >::getHeader(), llvm::SmallPtrSetImpl< PtrType >::insert(), llvm::SmallVectorImpl< T >::pop_back_val(), and Ptr.

Referenced by llvm::MemoryDepChecker::addAccess().

Variable Documentation

◆ EnableForwardingConflictDetection

cl::opt<bool> EnableForwardingConflictDetection("store-to-load-forwarding-conflict-detection", cl::Hidden, cl::desc("Enable conflict detection in loop-access analysis"), cl::init(true))

static

Enable store-to-load forwarding conflict detection.

This option can be disabled for correctness testing.

◆ EnableMemAccessVersioning

cl::opt<bool> EnableMemAccessVersioning("enable-mem-access-versioning", cl::init(true), cl::Hidden, cl::desc("Enable symbolic stride memory access versioning"))

static

This enables versioning on the strides of symbolically striding memory accesses in code like the following.

for (i = 0; i < N; ++i) A[i * Stride1] += B[i * Stride2] ...

Will be roughly translated to if (Stride1 == 1 && Stride2 == 1) { for (i = 0; i < N; i+=4) A[i:i+3] += ... } else ...

◆ laa_name

const char laa_name[] = "Loop Access Analysis"

static

Definition at line 2716 of file LoopAccessAnalysis.cpp.

◆ MaxDependences

cl::opt<unsigned> MaxDependences("max-dependences", cl::Hidden, cl::desc("Maximum number of dependences collected by " "loop-access analysis (default = 100)"), cl::init(100))

static

We collect dependences up to this threshold.

Referenced by llvm::MemoryDepChecker::areDepsSafe().

◆ MaxForkedSCEVDepth

cl::opt<unsigned> MaxForkedSCEVDepth("max-forked-scev-depth", cl::Hidden, cl::desc("Maximum recursion depth when finding forked SCEVs (default = 5)"), cl::init(5))

static

Referenced by findForkedPointer().

◆ MemoryCheckMergeThreshold

cl::opt<unsigned> MemoryCheckMergeThreshold("memory-check-merge-threshold", cl::Hidden, cl::desc("Maximum number of comparisons done when trying to merge " "runtime memory checks. (default = 100)"), cl::init(100))

static

The maximum iterations used to merge memory checks.

◆ RuntimeMemoryCheckThreshold

cl::opt<unsigned, true> RuntimeMemoryCheckThreshold("runtime-memory-check-threshold", cl::Hidden, cl::desc("When performing memory disambiguation checks at runtime do not " "generate more than this number of comparisons (default = 8)."), cl::location(VectorizerParams::RuntimeMemoryCheckThreshold), cl::init(8))

static

◆ VectorizationFactor

cl::opt<unsigned, true> VectorizationFactor("force-vector-width", cl::Hidden, cl::desc("Sets the SIMD width. Zero is autoselect."), cl::location(VectorizerParams::VectorizationFactor))

static

◆ VectorizationInterleave

cl::opt<unsigned, true> VectorizationInterleave("force-vector-interleave", cl::Hidden, cl::desc("Sets the vectorization interleave count. " "Zero is autoselect."), cl::location( VectorizerParams::VectorizationInterleave))

static

Referenced by llvm::VectorizerParams::isInterleaveForced().

Namespaces

Macros

Functions

Variables

Macro Definition Documentation

◆ DEBUG_TYPE

◆ LAA_NAME

Function Documentation

◆ areStridedAccessesIndependent()

◆ findForkedPointer()

◆ findForkedSCEVs()

◆ getMinFromExprs()

◆ hasComputableBounds()

◆ isInBoundsGep()

◆ isNoWrap()

◆ isNoWrapAddRec()

◆ isSafeDependenceDistance()

◆ visitPointers()

Variable Documentation

◆ EnableForwardingConflictDetection

◆ EnableMemAccessVersioning

◆ laa_name

◆ MaxDependences

◆ MaxForkedSCEVDepth

◆ MemoryCheckMergeThreshold

◆ RuntimeMemoryCheckThreshold

◆ VectorizationFactor

◆ VectorizationInterleave