llvm-project/llvm/lib/Analysis/MemoryDependenceAnalysis.cpp

596 lines
23 KiB
C++
Raw Normal View History

//===- MemoryDependenceAnalysis.cpp - Mem Deps Implementation --*- C++ -*-===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file implements an analysis that determines, for a given memory
// operation, what preceding memory operations it depends on. It builds on
// alias analysis information, and tries to provide a lazy, caching interface to
// a common kind of alias information query.
//
//===----------------------------------------------------------------------===//
#define DEBUG_TYPE "memdep"
#include "llvm/Analysis/MemoryDependenceAnalysis.h"
#include "llvm/Constants.h"
#include "llvm/Instructions.h"
#include "llvm/Function.h"
#include "llvm/Analysis/AliasAnalysis.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/Support/CFG.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Target/TargetData.h"
using namespace llvm;
// Control the calculation of non-local dependencies by only examining the
// predecessors if the basic block has less than X amount (50 by default).
static cl::opt<int>
PredLimit("nonlocaldep-threshold", cl::Hidden, cl::init(50),
cl::desc("Control the calculation of non-local"
"dependencies (default = 50)"));
STATISTIC(NumCacheNonlocal, "Number of cached non-local responses");
STATISTIC(NumUncacheNonlocal, "Number of uncached non-local responses");
char MemoryDependenceAnalysis::ID = 0;
// Register this pass...
static RegisterPass<MemoryDependenceAnalysis> X("memdep",
"Memory Dependence Analysis", false, true);
/// verifyRemoved - Verify that the specified instruction does not occur
/// in our internal data structures.
void MemoryDependenceAnalysis::verifyRemoved(Instruction *D) const {
for (LocalDepMapType::const_iterator I = LocalDeps.begin(),
E = LocalDeps.end(); I != E; ++I) {
assert(I->first != D && "Inst occurs in data structures");
assert(I->second.getPointer() != D &&
"Inst occurs in data structures");
}
for (nonLocalDepMapType::const_iterator I = depGraphNonLocal.begin(),
E = depGraphNonLocal.end(); I != E; ++I) {
assert(I->first != D && "Inst occurs in data structures");
for (DenseMap<BasicBlock*, DepResultTy>::iterator II = I->second.begin(),
EE = I->second.end(); II != EE; ++II)
assert(II->second.getPointer() != D && "Inst occurs in data structures");
}
for (reverseDepMapType::const_iterator I = reverseDep.begin(),
E = reverseDep.end(); I != E; ++I)
for (SmallPtrSet<Instruction*, 4>::const_iterator II = I->second.begin(),
EE = I->second.end(); II != EE; ++II)
assert(*II != D && "Inst occurs in data structures");
for (reverseDepMapType::const_iterator I = reverseDepNonLocal.begin(),
E = reverseDepNonLocal.end();
I != E; ++I)
for (SmallPtrSet<Instruction*, 4>::const_iterator II = I->second.begin(),
EE = I->second.end(); II != EE; ++II)
assert(*II != D && "Inst occurs in data structures");
}
/// getAnalysisUsage - Does not modify anything. It uses Alias Analysis.
///
void MemoryDependenceAnalysis::getAnalysisUsage(AnalysisUsage &AU) const {
AU.setPreservesAll();
AU.addRequiredTransitive<AliasAnalysis>();
AU.addRequiredTransitive<TargetData>();
}
/// getCallSiteDependency - Private helper for finding the local dependencies
/// of a call site.
MemDepResult MemoryDependenceAnalysis::
getCallSiteDependency(CallSite C, Instruction *start, BasicBlock *block) {
DepResultTy &cachedResult = LocalDeps[C.getInstruction()];
AliasAnalysis &AA = getAnalysis<AliasAnalysis>();
TargetData &TD = getAnalysis<TargetData>();
BasicBlock::iterator blockBegin = C.getInstruction()->getParent()->begin();
BasicBlock::iterator QI = C.getInstruction();
2008-09-12 03:41:10 +08:00
// If the starting point was specified, use it
if (start) {
QI = start;
blockBegin = start->getParent()->begin();
// If the starting point wasn't specified, but the block was, use it
} else if (!start && block) {
QI = block->end();
blockBegin = block->begin();
}
// Walk backwards through the block, looking for dependencies
while (QI != blockBegin) {
--QI;
// If this inst is a memory op, get the pointer it accessed
Value* pointer = 0;
uint64_t pointerSize = 0;
if (StoreInst* S = dyn_cast<StoreInst>(QI)) {
pointer = S->getPointerOperand();
Executive summary: getTypeSize -> getTypeStoreSize / getABITypeSize. The meaning of getTypeSize was not clear - clarifying it is important now that we have x86 long double and arbitrary precision integers. The issue with long double is that it requires 80 bits, and this is not a multiple of its alignment. This gives a primitive type for which getTypeSize differed from getABITypeSize. For arbitrary precision integers it is even worse: there is the minimum number of bits needed to hold the type (eg: 36 for an i36), the maximum number of bits that will be overwriten when storing the type (40 bits for i36) and the ABI size (i.e. the storage size rounded up to a multiple of the alignment; 64 bits for i36). This patch removes getTypeSize (not really - it is still there but deprecated to allow for a gradual transition). Instead there is: (1) getTypeSizeInBits - a number of bits that suffices to hold all values of the type. For a primitive type, this is the minimum number of bits. For an i36 this is 36 bits. For x86 long double it is 80. This corresponds to gcc's TYPE_PRECISION. (2) getTypeStoreSizeInBits - the maximum number of bits that is written when storing the type (or read when reading it). For an i36 this is 40 bits, for an x86 long double it is 80 bits. This is the size alias analysis is interested in (getTypeStoreSize returns the number of bytes). There doesn't seem to be anything corresponding to this in gcc. (3) getABITypeSizeInBits - this is getTypeStoreSizeInBits rounded up to a multiple of the alignment. For an i36 this is 64, for an x86 long double this is 96 or 128 depending on the OS. This is the spacing between consecutive elements when you form an array out of this type (getABITypeSize returns the number of bytes). This is TYPE_SIZE in gcc. Since successive elements in a SequentialType (arrays, pointers and vectors) need to be aligned, the spacing between them will be given by getABITypeSize. This means that the size of an array is the length times the getABITypeSize. It also means that GEP computations need to use getABITypeSize when computing offsets. Furthermore, if an alloca allocates several elements at once then these too need to be aligned, so the size of the alloca has to be the number of elements multiplied by getABITypeSize. Logically speaking this doesn't have to be the case when allocating just one element, but it is simpler to also use getABITypeSize in this case. So alloca's and mallocs should use getABITypeSize. Finally, since gcc's only notion of size is that given by getABITypeSize, if you want to output assembler etc the same as gcc then getABITypeSize is the size you want. Since a store will overwrite no more than getTypeStoreSize bytes, and a read will read no more than that many bytes, this is the notion of size appropriate for alias analysis calculations. In this patch I have corrected all type size uses except some of those in ScalarReplAggregates, lib/Codegen, lib/Target (the hard cases). I will get around to auditing these too at some point, but I could do with some help. Finally, I made one change which I think wise but others might consider pointless and suboptimal: in an unpacked struct the amount of space allocated for a field is now given by the ABI size rather than getTypeStoreSize. I did this because every other place that reserves memory for a type (eg: alloca) now uses getABITypeSize, and I didn't want to make an exception for unpacked structs, i.e. I did it to make things more uniform. This only effects structs containing long doubles and arbitrary precision integers. If someone wants to pack these types more tightly they can always use a packed struct. llvm-svn: 43620
2007-11-02 04:53:16 +08:00
pointerSize = TD.getTypeStoreSize(S->getOperand(0)->getType());
} else if (AllocationInst* AI = dyn_cast<AllocationInst>(QI)) {
pointer = AI;
if (ConstantInt* C = dyn_cast<ConstantInt>(AI->getArraySize()))
pointerSize = C->getZExtValue() *
Executive summary: getTypeSize -> getTypeStoreSize / getABITypeSize. The meaning of getTypeSize was not clear - clarifying it is important now that we have x86 long double and arbitrary precision integers. The issue with long double is that it requires 80 bits, and this is not a multiple of its alignment. This gives a primitive type for which getTypeSize differed from getABITypeSize. For arbitrary precision integers it is even worse: there is the minimum number of bits needed to hold the type (eg: 36 for an i36), the maximum number of bits that will be overwriten when storing the type (40 bits for i36) and the ABI size (i.e. the storage size rounded up to a multiple of the alignment; 64 bits for i36). This patch removes getTypeSize (not really - it is still there but deprecated to allow for a gradual transition). Instead there is: (1) getTypeSizeInBits - a number of bits that suffices to hold all values of the type. For a primitive type, this is the minimum number of bits. For an i36 this is 36 bits. For x86 long double it is 80. This corresponds to gcc's TYPE_PRECISION. (2) getTypeStoreSizeInBits - the maximum number of bits that is written when storing the type (or read when reading it). For an i36 this is 40 bits, for an x86 long double it is 80 bits. This is the size alias analysis is interested in (getTypeStoreSize returns the number of bytes). There doesn't seem to be anything corresponding to this in gcc. (3) getABITypeSizeInBits - this is getTypeStoreSizeInBits rounded up to a multiple of the alignment. For an i36 this is 64, for an x86 long double this is 96 or 128 depending on the OS. This is the spacing between consecutive elements when you form an array out of this type (getABITypeSize returns the number of bytes). This is TYPE_SIZE in gcc. Since successive elements in a SequentialType (arrays, pointers and vectors) need to be aligned, the spacing between them will be given by getABITypeSize. This means that the size of an array is the length times the getABITypeSize. It also means that GEP computations need to use getABITypeSize when computing offsets. Furthermore, if an alloca allocates several elements at once then these too need to be aligned, so the size of the alloca has to be the number of elements multiplied by getABITypeSize. Logically speaking this doesn't have to be the case when allocating just one element, but it is simpler to also use getABITypeSize in this case. So alloca's and mallocs should use getABITypeSize. Finally, since gcc's only notion of size is that given by getABITypeSize, if you want to output assembler etc the same as gcc then getABITypeSize is the size you want. Since a store will overwrite no more than getTypeStoreSize bytes, and a read will read no more than that many bytes, this is the notion of size appropriate for alias analysis calculations. In this patch I have corrected all type size uses except some of those in ScalarReplAggregates, lib/Codegen, lib/Target (the hard cases). I will get around to auditing these too at some point, but I could do with some help. Finally, I made one change which I think wise but others might consider pointless and suboptimal: in an unpacked struct the amount of space allocated for a field is now given by the ABI size rather than getTypeStoreSize. I did this because every other place that reserves memory for a type (eg: alloca) now uses getABITypeSize, and I didn't want to make an exception for unpacked structs, i.e. I did it to make things more uniform. This only effects structs containing long doubles and arbitrary precision integers. If someone wants to pack these types more tightly they can always use a packed struct. llvm-svn: 43620
2007-11-02 04:53:16 +08:00
TD.getABITypeSize(AI->getAllocatedType());
else
pointerSize = ~0UL;
} else if (VAArgInst* V = dyn_cast<VAArgInst>(QI)) {
pointer = V->getOperand(0);
Executive summary: getTypeSize -> getTypeStoreSize / getABITypeSize. The meaning of getTypeSize was not clear - clarifying it is important now that we have x86 long double and arbitrary precision integers. The issue with long double is that it requires 80 bits, and this is not a multiple of its alignment. This gives a primitive type for which getTypeSize differed from getABITypeSize. For arbitrary precision integers it is even worse: there is the minimum number of bits needed to hold the type (eg: 36 for an i36), the maximum number of bits that will be overwriten when storing the type (40 bits for i36) and the ABI size (i.e. the storage size rounded up to a multiple of the alignment; 64 bits for i36). This patch removes getTypeSize (not really - it is still there but deprecated to allow for a gradual transition). Instead there is: (1) getTypeSizeInBits - a number of bits that suffices to hold all values of the type. For a primitive type, this is the minimum number of bits. For an i36 this is 36 bits. For x86 long double it is 80. This corresponds to gcc's TYPE_PRECISION. (2) getTypeStoreSizeInBits - the maximum number of bits that is written when storing the type (or read when reading it). For an i36 this is 40 bits, for an x86 long double it is 80 bits. This is the size alias analysis is interested in (getTypeStoreSize returns the number of bytes). There doesn't seem to be anything corresponding to this in gcc. (3) getABITypeSizeInBits - this is getTypeStoreSizeInBits rounded up to a multiple of the alignment. For an i36 this is 64, for an x86 long double this is 96 or 128 depending on the OS. This is the spacing between consecutive elements when you form an array out of this type (getABITypeSize returns the number of bytes). This is TYPE_SIZE in gcc. Since successive elements in a SequentialType (arrays, pointers and vectors) need to be aligned, the spacing between them will be given by getABITypeSize. This means that the size of an array is the length times the getABITypeSize. It also means that GEP computations need to use getABITypeSize when computing offsets. Furthermore, if an alloca allocates several elements at once then these too need to be aligned, so the size of the alloca has to be the number of elements multiplied by getABITypeSize. Logically speaking this doesn't have to be the case when allocating just one element, but it is simpler to also use getABITypeSize in this case. So alloca's and mallocs should use getABITypeSize. Finally, since gcc's only notion of size is that given by getABITypeSize, if you want to output assembler etc the same as gcc then getABITypeSize is the size you want. Since a store will overwrite no more than getTypeStoreSize bytes, and a read will read no more than that many bytes, this is the notion of size appropriate for alias analysis calculations. In this patch I have corrected all type size uses except some of those in ScalarReplAggregates, lib/Codegen, lib/Target (the hard cases). I will get around to auditing these too at some point, but I could do with some help. Finally, I made one change which I think wise but others might consider pointless and suboptimal: in an unpacked struct the amount of space allocated for a field is now given by the ABI size rather than getTypeStoreSize. I did this because every other place that reserves memory for a type (eg: alloca) now uses getABITypeSize, and I didn't want to make an exception for unpacked structs, i.e. I did it to make things more uniform. This only effects structs containing long doubles and arbitrary precision integers. If someone wants to pack these types more tightly they can always use a packed struct. llvm-svn: 43620
2007-11-02 04:53:16 +08:00
pointerSize = TD.getTypeStoreSize(V->getType());
} else if (FreeInst* F = dyn_cast<FreeInst>(QI)) {
pointer = F->getPointerOperand();
// FreeInsts erase the entire structure
pointerSize = ~0UL;
} else if (CallSite::get(QI).getInstruction() != 0) {
AliasAnalysis::ModRefBehavior result =
AA.getModRefBehavior(CallSite::get(QI));
if (result != AliasAnalysis::DoesNotAccessMemory) {
if (!start && !block) {
cachedResult = DepResultTy(QI, Normal);
reverseDep[QI].insert(C.getInstruction());
}
return MemDepResult::get(QI);
} else {
continue;
}
} else
continue;
if (AA.getModRefInfo(C, pointer, pointerSize) != AliasAnalysis::NoModRef) {
if (!start && !block) {
cachedResult = DepResultTy(QI, Normal);
reverseDep[QI].insert(C.getInstruction());
}
return MemDepResult::get(QI);
}
}
// No dependence found
cachedResult = DepResultTy(0, NonLocal);
return MemDepResult::getNonLocal();
}
/// nonLocalHelper - Private helper used to calculate non-local dependencies
/// by doing DFS on the predecessors of a block to find its dependencies.
void MemoryDependenceAnalysis::nonLocalHelper(Instruction* query,
BasicBlock* block,
DenseMap<BasicBlock*, DepResultTy> &resp) {
// Set of blocks that we've already visited in our DFS
SmallPtrSet<BasicBlock*, 4> visited;
// If we're updating a dirtied cache entry, we don't need to reprocess
// already computed entries.
for (DenseMap<BasicBlock*, DepResultTy>::iterator I = resp.begin(),
E = resp.end(); I != E; ++I)
if (I->second.getInt() != Dirty)
visited.insert(I->first);
// Current stack of the DFS
SmallVector<BasicBlock*, 4> stack;
for (pred_iterator PI = pred_begin(block), PE = pred_end(block);
PI != PE; ++PI)
stack.push_back(*PI);
// Do a basic DFS
while (!stack.empty()) {
BasicBlock* BB = stack.back();
// If we've already visited this block, no need to revist
if (visited.count(BB)) {
stack.pop_back();
continue;
}
// If we find a new block with a local dependency for query,
// then we insert the new dependency and backtrack.
if (BB != block) {
visited.insert(BB);
MemDepResult localDep = getDependency(query, 0, BB);
if (!localDep.isNonLocal()) {
resp.insert(std::make_pair(BB, ConvFromResult(localDep)));
stack.pop_back();
continue;
}
// If we re-encounter the starting block, we still need to search it
// because there might be a dependency in the starting block AFTER
// the position of the query. This is necessary to get loops right.
} else if (BB == block) {
visited.insert(BB);
MemDepResult localDep = getDependency(query, 0, BB);
if (localDep.getInst() != query)
resp.insert(std::make_pair(BB, ConvFromResult(localDep)));
stack.pop_back();
continue;
}
// If we didn't find anything, recurse on the precessors of this block
// Only do this for blocks with a small number of predecessors.
bool predOnStack = false;
bool inserted = false;
if (std::distance(pred_begin(BB), pred_end(BB)) <= PredLimit) {
for (pred_iterator PI = pred_begin(BB), PE = pred_end(BB);
PI != PE; ++PI)
if (!visited.count(*PI)) {
stack.push_back(*PI);
inserted = true;
} else
predOnStack = true;
}
// If we inserted a new predecessor, then we'll come back to this block
if (inserted)
continue;
// If we didn't insert because we have no predecessors, then this
// query has no dependency at all.
else if (!inserted && !predOnStack) {
resp.insert(std::make_pair(BB, DepResultTy(0, None)));
// If we didn't insert because our predecessors are already on the stack,
// then we might still have a dependency, but it will be discovered during
// backtracking.
} else if (!inserted && predOnStack){
resp.insert(std::make_pair(BB, DepResultTy(0, NonLocal)));
}
stack.pop_back();
}
}
/// getNonLocalDependency - Fills the passed-in map with the non-local
/// dependencies of the queries. The map will contain NonLocal for
/// blocks between the query and its dependencies.
void MemoryDependenceAnalysis::getNonLocalDependency(Instruction* query,
DenseMap<BasicBlock*, MemDepResult> &resp) {
if (depGraphNonLocal.count(query)) {
DenseMap<BasicBlock*, DepResultTy> &cached = depGraphNonLocal[query];
NumCacheNonlocal++;
SmallVector<BasicBlock*, 4> dirtied;
for (DenseMap<BasicBlock*, DepResultTy>::iterator I = cached.begin(),
E = cached.end(); I != E; ++I)
if (I->second.getInt() == Dirty)
dirtied.push_back(I->first);
for (SmallVector<BasicBlock*, 4>::iterator I = dirtied.begin(),
E = dirtied.end(); I != E; ++I) {
MemDepResult localDep = getDependency(query, 0, *I);
if (!localDep.isNonLocal())
cached[*I] = ConvFromResult(localDep);
else {
cached.erase(*I);
nonLocalHelper(query, *I, cached);
}
}
// Update the reverse non-local dependency cache.
for (DenseMap<BasicBlock*, DepResultTy>::iterator I = cached.begin(),
E = cached.end(); I != E; ++I) {
if (Instruction *Inst = I->second.getPointer())
reverseDepNonLocal[Inst].insert(query);
resp[I->first] = ConvToResult(I->second);
}
return;
}
NumUncacheNonlocal++;
// If not, go ahead and search for non-local deps.
DenseMap<BasicBlock*, DepResultTy> &cached = depGraphNonLocal[query];
nonLocalHelper(query, query->getParent(), cached);
// Update the non-local dependency cache
for (DenseMap<BasicBlock*, DepResultTy>::iterator I = cached.begin(),
E = cached.end(); I != E; ++I) {
// FIXME: Merge with the code above!
if (Instruction *Inst = I->second.getPointer())
reverseDepNonLocal[Inst].insert(query);
resp[I->first] = ConvToResult(I->second);
}
}
/// getDependency - Return the instruction on which a memory operation
/// depends. The local parameter indicates if the query should only
/// evaluate dependencies within the same basic block.
/// FIXME: ELIMINATE START/BLOCK and make the caching happen in a higher level
/// METHOD.
MemDepResult MemoryDependenceAnalysis::getDependency(Instruction *query,
Instruction *start,
BasicBlock *block) {
// Start looking for dependencies with the queried inst
BasicBlock::iterator QI = query;
// Check for a cached result
// FIXME: why do this when Block or Start is specified??
DepResultTy &cachedResult = LocalDeps[query];
if (start)
QI = start;
else if (block)
QI = block->end();
else if (cachedResult.getInt() != Dirty) {
// If we have a _confirmed_ cached entry, return it.
return ConvToResult(cachedResult);
} else if (Instruction *Inst = cachedResult.getPointer()) {
// If we have an unconfirmed cached entry, we can start our search from it.
QI = Inst;
}
AliasAnalysis& AA = getAnalysis<AliasAnalysis>();
TargetData& TD = getAnalysis<TargetData>();
// Get the pointer value for which dependence will be determined
Value* dependee = 0;
uint64_t dependeeSize = 0;
bool queryIsVolatile = false;
if (StoreInst* S = dyn_cast<StoreInst>(query)) {
dependee = S->getPointerOperand();
Executive summary: getTypeSize -> getTypeStoreSize / getABITypeSize. The meaning of getTypeSize was not clear - clarifying it is important now that we have x86 long double and arbitrary precision integers. The issue with long double is that it requires 80 bits, and this is not a multiple of its alignment. This gives a primitive type for which getTypeSize differed from getABITypeSize. For arbitrary precision integers it is even worse: there is the minimum number of bits needed to hold the type (eg: 36 for an i36), the maximum number of bits that will be overwriten when storing the type (40 bits for i36) and the ABI size (i.e. the storage size rounded up to a multiple of the alignment; 64 bits for i36). This patch removes getTypeSize (not really - it is still there but deprecated to allow for a gradual transition). Instead there is: (1) getTypeSizeInBits - a number of bits that suffices to hold all values of the type. For a primitive type, this is the minimum number of bits. For an i36 this is 36 bits. For x86 long double it is 80. This corresponds to gcc's TYPE_PRECISION. (2) getTypeStoreSizeInBits - the maximum number of bits that is written when storing the type (or read when reading it). For an i36 this is 40 bits, for an x86 long double it is 80 bits. This is the size alias analysis is interested in (getTypeStoreSize returns the number of bytes). There doesn't seem to be anything corresponding to this in gcc. (3) getABITypeSizeInBits - this is getTypeStoreSizeInBits rounded up to a multiple of the alignment. For an i36 this is 64, for an x86 long double this is 96 or 128 depending on the OS. This is the spacing between consecutive elements when you form an array out of this type (getABITypeSize returns the number of bytes). This is TYPE_SIZE in gcc. Since successive elements in a SequentialType (arrays, pointers and vectors) need to be aligned, the spacing between them will be given by getABITypeSize. This means that the size of an array is the length times the getABITypeSize. It also means that GEP computations need to use getABITypeSize when computing offsets. Furthermore, if an alloca allocates several elements at once then these too need to be aligned, so the size of the alloca has to be the number of elements multiplied by getABITypeSize. Logically speaking this doesn't have to be the case when allocating just one element, but it is simpler to also use getABITypeSize in this case. So alloca's and mallocs should use getABITypeSize. Finally, since gcc's only notion of size is that given by getABITypeSize, if you want to output assembler etc the same as gcc then getABITypeSize is the size you want. Since a store will overwrite no more than getTypeStoreSize bytes, and a read will read no more than that many bytes, this is the notion of size appropriate for alias analysis calculations. In this patch I have corrected all type size uses except some of those in ScalarReplAggregates, lib/Codegen, lib/Target (the hard cases). I will get around to auditing these too at some point, but I could do with some help. Finally, I made one change which I think wise but others might consider pointless and suboptimal: in an unpacked struct the amount of space allocated for a field is now given by the ABI size rather than getTypeStoreSize. I did this because every other place that reserves memory for a type (eg: alloca) now uses getABITypeSize, and I didn't want to make an exception for unpacked structs, i.e. I did it to make things more uniform. This only effects structs containing long doubles and arbitrary precision integers. If someone wants to pack these types more tightly they can always use a packed struct. llvm-svn: 43620
2007-11-02 04:53:16 +08:00
dependeeSize = TD.getTypeStoreSize(S->getOperand(0)->getType());
queryIsVolatile = S->isVolatile();
} else if (LoadInst* L = dyn_cast<LoadInst>(query)) {
dependee = L->getPointerOperand();
Executive summary: getTypeSize -> getTypeStoreSize / getABITypeSize. The meaning of getTypeSize was not clear - clarifying it is important now that we have x86 long double and arbitrary precision integers. The issue with long double is that it requires 80 bits, and this is not a multiple of its alignment. This gives a primitive type for which getTypeSize differed from getABITypeSize. For arbitrary precision integers it is even worse: there is the minimum number of bits needed to hold the type (eg: 36 for an i36), the maximum number of bits that will be overwriten when storing the type (40 bits for i36) and the ABI size (i.e. the storage size rounded up to a multiple of the alignment; 64 bits for i36). This patch removes getTypeSize (not really - it is still there but deprecated to allow for a gradual transition). Instead there is: (1) getTypeSizeInBits - a number of bits that suffices to hold all values of the type. For a primitive type, this is the minimum number of bits. For an i36 this is 36 bits. For x86 long double it is 80. This corresponds to gcc's TYPE_PRECISION. (2) getTypeStoreSizeInBits - the maximum number of bits that is written when storing the type (or read when reading it). For an i36 this is 40 bits, for an x86 long double it is 80 bits. This is the size alias analysis is interested in (getTypeStoreSize returns the number of bytes). There doesn't seem to be anything corresponding to this in gcc. (3) getABITypeSizeInBits - this is getTypeStoreSizeInBits rounded up to a multiple of the alignment. For an i36 this is 64, for an x86 long double this is 96 or 128 depending on the OS. This is the spacing between consecutive elements when you form an array out of this type (getABITypeSize returns the number of bytes). This is TYPE_SIZE in gcc. Since successive elements in a SequentialType (arrays, pointers and vectors) need to be aligned, the spacing between them will be given by getABITypeSize. This means that the size of an array is the length times the getABITypeSize. It also means that GEP computations need to use getABITypeSize when computing offsets. Furthermore, if an alloca allocates several elements at once then these too need to be aligned, so the size of the alloca has to be the number of elements multiplied by getABITypeSize. Logically speaking this doesn't have to be the case when allocating just one element, but it is simpler to also use getABITypeSize in this case. So alloca's and mallocs should use getABITypeSize. Finally, since gcc's only notion of size is that given by getABITypeSize, if you want to output assembler etc the same as gcc then getABITypeSize is the size you want. Since a store will overwrite no more than getTypeStoreSize bytes, and a read will read no more than that many bytes, this is the notion of size appropriate for alias analysis calculations. In this patch I have corrected all type size uses except some of those in ScalarReplAggregates, lib/Codegen, lib/Target (the hard cases). I will get around to auditing these too at some point, but I could do with some help. Finally, I made one change which I think wise but others might consider pointless and suboptimal: in an unpacked struct the amount of space allocated for a field is now given by the ABI size rather than getTypeStoreSize. I did this because every other place that reserves memory for a type (eg: alloca) now uses getABITypeSize, and I didn't want to make an exception for unpacked structs, i.e. I did it to make things more uniform. This only effects structs containing long doubles and arbitrary precision integers. If someone wants to pack these types more tightly they can always use a packed struct. llvm-svn: 43620
2007-11-02 04:53:16 +08:00
dependeeSize = TD.getTypeStoreSize(L->getType());
queryIsVolatile = L->isVolatile();
} else if (VAArgInst* V = dyn_cast<VAArgInst>(query)) {
dependee = V->getOperand(0);
Executive summary: getTypeSize -> getTypeStoreSize / getABITypeSize. The meaning of getTypeSize was not clear - clarifying it is important now that we have x86 long double and arbitrary precision integers. The issue with long double is that it requires 80 bits, and this is not a multiple of its alignment. This gives a primitive type for which getTypeSize differed from getABITypeSize. For arbitrary precision integers it is even worse: there is the minimum number of bits needed to hold the type (eg: 36 for an i36), the maximum number of bits that will be overwriten when storing the type (40 bits for i36) and the ABI size (i.e. the storage size rounded up to a multiple of the alignment; 64 bits for i36). This patch removes getTypeSize (not really - it is still there but deprecated to allow for a gradual transition). Instead there is: (1) getTypeSizeInBits - a number of bits that suffices to hold all values of the type. For a primitive type, this is the minimum number of bits. For an i36 this is 36 bits. For x86 long double it is 80. This corresponds to gcc's TYPE_PRECISION. (2) getTypeStoreSizeInBits - the maximum number of bits that is written when storing the type (or read when reading it). For an i36 this is 40 bits, for an x86 long double it is 80 bits. This is the size alias analysis is interested in (getTypeStoreSize returns the number of bytes). There doesn't seem to be anything corresponding to this in gcc. (3) getABITypeSizeInBits - this is getTypeStoreSizeInBits rounded up to a multiple of the alignment. For an i36 this is 64, for an x86 long double this is 96 or 128 depending on the OS. This is the spacing between consecutive elements when you form an array out of this type (getABITypeSize returns the number of bytes). This is TYPE_SIZE in gcc. Since successive elements in a SequentialType (arrays, pointers and vectors) need to be aligned, the spacing between them will be given by getABITypeSize. This means that the size of an array is the length times the getABITypeSize. It also means that GEP computations need to use getABITypeSize when computing offsets. Furthermore, if an alloca allocates several elements at once then these too need to be aligned, so the size of the alloca has to be the number of elements multiplied by getABITypeSize. Logically speaking this doesn't have to be the case when allocating just one element, but it is simpler to also use getABITypeSize in this case. So alloca's and mallocs should use getABITypeSize. Finally, since gcc's only notion of size is that given by getABITypeSize, if you want to output assembler etc the same as gcc then getABITypeSize is the size you want. Since a store will overwrite no more than getTypeStoreSize bytes, and a read will read no more than that many bytes, this is the notion of size appropriate for alias analysis calculations. In this patch I have corrected all type size uses except some of those in ScalarReplAggregates, lib/Codegen, lib/Target (the hard cases). I will get around to auditing these too at some point, but I could do with some help. Finally, I made one change which I think wise but others might consider pointless and suboptimal: in an unpacked struct the amount of space allocated for a field is now given by the ABI size rather than getTypeStoreSize. I did this because every other place that reserves memory for a type (eg: alloca) now uses getABITypeSize, and I didn't want to make an exception for unpacked structs, i.e. I did it to make things more uniform. This only effects structs containing long doubles and arbitrary precision integers. If someone wants to pack these types more tightly they can always use a packed struct. llvm-svn: 43620
2007-11-02 04:53:16 +08:00
dependeeSize = TD.getTypeStoreSize(V->getType());
} else if (FreeInst* F = dyn_cast<FreeInst>(query)) {
dependee = F->getPointerOperand();
// FreeInsts erase the entire structure, not just a field
dependeeSize = ~0UL;
} else if (CallSite::get(query).getInstruction() != 0)
return getCallSiteDependency(CallSite::get(query), start, block);
else if (isa<AllocationInst>(query))
return MemDepResult::getNone();
else
return MemDepResult::getNone();
BasicBlock::iterator blockBegin = block ? block->begin()
: query->getParent()->begin();
// Walk backwards through the basic block, looking for dependencies
while (QI != blockBegin) {
--QI;
// If this inst is a memory op, get the pointer it accessed
Value* pointer = 0;
uint64_t pointerSize = 0;
if (StoreInst* S = dyn_cast<StoreInst>(QI)) {
// All volatile loads/stores depend on each other
if (queryIsVolatile && S->isVolatile()) {
if (!start && !block) {
cachedResult = DepResultTy(S, Normal);
reverseDep[S].insert(query);
}
return MemDepResult::get(S);
}
pointer = S->getPointerOperand();
Executive summary: getTypeSize -> getTypeStoreSize / getABITypeSize. The meaning of getTypeSize was not clear - clarifying it is important now that we have x86 long double and arbitrary precision integers. The issue with long double is that it requires 80 bits, and this is not a multiple of its alignment. This gives a primitive type for which getTypeSize differed from getABITypeSize. For arbitrary precision integers it is even worse: there is the minimum number of bits needed to hold the type (eg: 36 for an i36), the maximum number of bits that will be overwriten when storing the type (40 bits for i36) and the ABI size (i.e. the storage size rounded up to a multiple of the alignment; 64 bits for i36). This patch removes getTypeSize (not really - it is still there but deprecated to allow for a gradual transition). Instead there is: (1) getTypeSizeInBits - a number of bits that suffices to hold all values of the type. For a primitive type, this is the minimum number of bits. For an i36 this is 36 bits. For x86 long double it is 80. This corresponds to gcc's TYPE_PRECISION. (2) getTypeStoreSizeInBits - the maximum number of bits that is written when storing the type (or read when reading it). For an i36 this is 40 bits, for an x86 long double it is 80 bits. This is the size alias analysis is interested in (getTypeStoreSize returns the number of bytes). There doesn't seem to be anything corresponding to this in gcc. (3) getABITypeSizeInBits - this is getTypeStoreSizeInBits rounded up to a multiple of the alignment. For an i36 this is 64, for an x86 long double this is 96 or 128 depending on the OS. This is the spacing between consecutive elements when you form an array out of this type (getABITypeSize returns the number of bytes). This is TYPE_SIZE in gcc. Since successive elements in a SequentialType (arrays, pointers and vectors) need to be aligned, the spacing between them will be given by getABITypeSize. This means that the size of an array is the length times the getABITypeSize. It also means that GEP computations need to use getABITypeSize when computing offsets. Furthermore, if an alloca allocates several elements at once then these too need to be aligned, so the size of the alloca has to be the number of elements multiplied by getABITypeSize. Logically speaking this doesn't have to be the case when allocating just one element, but it is simpler to also use getABITypeSize in this case. So alloca's and mallocs should use getABITypeSize. Finally, since gcc's only notion of size is that given by getABITypeSize, if you want to output assembler etc the same as gcc then getABITypeSize is the size you want. Since a store will overwrite no more than getTypeStoreSize bytes, and a read will read no more than that many bytes, this is the notion of size appropriate for alias analysis calculations. In this patch I have corrected all type size uses except some of those in ScalarReplAggregates, lib/Codegen, lib/Target (the hard cases). I will get around to auditing these too at some point, but I could do with some help. Finally, I made one change which I think wise but others might consider pointless and suboptimal: in an unpacked struct the amount of space allocated for a field is now given by the ABI size rather than getTypeStoreSize. I did this because every other place that reserves memory for a type (eg: alloca) now uses getABITypeSize, and I didn't want to make an exception for unpacked structs, i.e. I did it to make things more uniform. This only effects structs containing long doubles and arbitrary precision integers. If someone wants to pack these types more tightly they can always use a packed struct. llvm-svn: 43620
2007-11-02 04:53:16 +08:00
pointerSize = TD.getTypeStoreSize(S->getOperand(0)->getType());
} else if (LoadInst* L = dyn_cast<LoadInst>(QI)) {
// All volatile loads/stores depend on each other
if (queryIsVolatile && L->isVolatile()) {
if (!start && !block) {
cachedResult = DepResultTy(L, Normal);
reverseDep[L].insert(query);
}
return MemDepResult::get(L);
}
pointer = L->getPointerOperand();
Executive summary: getTypeSize -> getTypeStoreSize / getABITypeSize. The meaning of getTypeSize was not clear - clarifying it is important now that we have x86 long double and arbitrary precision integers. The issue with long double is that it requires 80 bits, and this is not a multiple of its alignment. This gives a primitive type for which getTypeSize differed from getABITypeSize. For arbitrary precision integers it is even worse: there is the minimum number of bits needed to hold the type (eg: 36 for an i36), the maximum number of bits that will be overwriten when storing the type (40 bits for i36) and the ABI size (i.e. the storage size rounded up to a multiple of the alignment; 64 bits for i36). This patch removes getTypeSize (not really - it is still there but deprecated to allow for a gradual transition). Instead there is: (1) getTypeSizeInBits - a number of bits that suffices to hold all values of the type. For a primitive type, this is the minimum number of bits. For an i36 this is 36 bits. For x86 long double it is 80. This corresponds to gcc's TYPE_PRECISION. (2) getTypeStoreSizeInBits - the maximum number of bits that is written when storing the type (or read when reading it). For an i36 this is 40 bits, for an x86 long double it is 80 bits. This is the size alias analysis is interested in (getTypeStoreSize returns the number of bytes). There doesn't seem to be anything corresponding to this in gcc. (3) getABITypeSizeInBits - this is getTypeStoreSizeInBits rounded up to a multiple of the alignment. For an i36 this is 64, for an x86 long double this is 96 or 128 depending on the OS. This is the spacing between consecutive elements when you form an array out of this type (getABITypeSize returns the number of bytes). This is TYPE_SIZE in gcc. Since successive elements in a SequentialType (arrays, pointers and vectors) need to be aligned, the spacing between them will be given by getABITypeSize. This means that the size of an array is the length times the getABITypeSize. It also means that GEP computations need to use getABITypeSize when computing offsets. Furthermore, if an alloca allocates several elements at once then these too need to be aligned, so the size of the alloca has to be the number of elements multiplied by getABITypeSize. Logically speaking this doesn't have to be the case when allocating just one element, but it is simpler to also use getABITypeSize in this case. So alloca's and mallocs should use getABITypeSize. Finally, since gcc's only notion of size is that given by getABITypeSize, if you want to output assembler etc the same as gcc then getABITypeSize is the size you want. Since a store will overwrite no more than getTypeStoreSize bytes, and a read will read no more than that many bytes, this is the notion of size appropriate for alias analysis calculations. In this patch I have corrected all type size uses except some of those in ScalarReplAggregates, lib/Codegen, lib/Target (the hard cases). I will get around to auditing these too at some point, but I could do with some help. Finally, I made one change which I think wise but others might consider pointless and suboptimal: in an unpacked struct the amount of space allocated for a field is now given by the ABI size rather than getTypeStoreSize. I did this because every other place that reserves memory for a type (eg: alloca) now uses getABITypeSize, and I didn't want to make an exception for unpacked structs, i.e. I did it to make things more uniform. This only effects structs containing long doubles and arbitrary precision integers. If someone wants to pack these types more tightly they can always use a packed struct. llvm-svn: 43620
2007-11-02 04:53:16 +08:00
pointerSize = TD.getTypeStoreSize(L->getType());
} else if (AllocationInst* AI = dyn_cast<AllocationInst>(QI)) {
pointer = AI;
if (ConstantInt* C = dyn_cast<ConstantInt>(AI->getArraySize()))
pointerSize = C->getZExtValue() *
Executive summary: getTypeSize -> getTypeStoreSize / getABITypeSize. The meaning of getTypeSize was not clear - clarifying it is important now that we have x86 long double and arbitrary precision integers. The issue with long double is that it requires 80 bits, and this is not a multiple of its alignment. This gives a primitive type for which getTypeSize differed from getABITypeSize. For arbitrary precision integers it is even worse: there is the minimum number of bits needed to hold the type (eg: 36 for an i36), the maximum number of bits that will be overwriten when storing the type (40 bits for i36) and the ABI size (i.e. the storage size rounded up to a multiple of the alignment; 64 bits for i36). This patch removes getTypeSize (not really - it is still there but deprecated to allow for a gradual transition). Instead there is: (1) getTypeSizeInBits - a number of bits that suffices to hold all values of the type. For a primitive type, this is the minimum number of bits. For an i36 this is 36 bits. For x86 long double it is 80. This corresponds to gcc's TYPE_PRECISION. (2) getTypeStoreSizeInBits - the maximum number of bits that is written when storing the type (or read when reading it). For an i36 this is 40 bits, for an x86 long double it is 80 bits. This is the size alias analysis is interested in (getTypeStoreSize returns the number of bytes). There doesn't seem to be anything corresponding to this in gcc. (3) getABITypeSizeInBits - this is getTypeStoreSizeInBits rounded up to a multiple of the alignment. For an i36 this is 64, for an x86 long double this is 96 or 128 depending on the OS. This is the spacing between consecutive elements when you form an array out of this type (getABITypeSize returns the number of bytes). This is TYPE_SIZE in gcc. Since successive elements in a SequentialType (arrays, pointers and vectors) need to be aligned, the spacing between them will be given by getABITypeSize. This means that the size of an array is the length times the getABITypeSize. It also means that GEP computations need to use getABITypeSize when computing offsets. Furthermore, if an alloca allocates several elements at once then these too need to be aligned, so the size of the alloca has to be the number of elements multiplied by getABITypeSize. Logically speaking this doesn't have to be the case when allocating just one element, but it is simpler to also use getABITypeSize in this case. So alloca's and mallocs should use getABITypeSize. Finally, since gcc's only notion of size is that given by getABITypeSize, if you want to output assembler etc the same as gcc then getABITypeSize is the size you want. Since a store will overwrite no more than getTypeStoreSize bytes, and a read will read no more than that many bytes, this is the notion of size appropriate for alias analysis calculations. In this patch I have corrected all type size uses except some of those in ScalarReplAggregates, lib/Codegen, lib/Target (the hard cases). I will get around to auditing these too at some point, but I could do with some help. Finally, I made one change which I think wise but others might consider pointless and suboptimal: in an unpacked struct the amount of space allocated for a field is now given by the ABI size rather than getTypeStoreSize. I did this because every other place that reserves memory for a type (eg: alloca) now uses getABITypeSize, and I didn't want to make an exception for unpacked structs, i.e. I did it to make things more uniform. This only effects structs containing long doubles and arbitrary precision integers. If someone wants to pack these types more tightly they can always use a packed struct. llvm-svn: 43620
2007-11-02 04:53:16 +08:00
TD.getABITypeSize(AI->getAllocatedType());
else
pointerSize = ~0UL;
} else if (VAArgInst* V = dyn_cast<VAArgInst>(QI)) {
pointer = V->getOperand(0);
Executive summary: getTypeSize -> getTypeStoreSize / getABITypeSize. The meaning of getTypeSize was not clear - clarifying it is important now that we have x86 long double and arbitrary precision integers. The issue with long double is that it requires 80 bits, and this is not a multiple of its alignment. This gives a primitive type for which getTypeSize differed from getABITypeSize. For arbitrary precision integers it is even worse: there is the minimum number of bits needed to hold the type (eg: 36 for an i36), the maximum number of bits that will be overwriten when storing the type (40 bits for i36) and the ABI size (i.e. the storage size rounded up to a multiple of the alignment; 64 bits for i36). This patch removes getTypeSize (not really - it is still there but deprecated to allow for a gradual transition). Instead there is: (1) getTypeSizeInBits - a number of bits that suffices to hold all values of the type. For a primitive type, this is the minimum number of bits. For an i36 this is 36 bits. For x86 long double it is 80. This corresponds to gcc's TYPE_PRECISION. (2) getTypeStoreSizeInBits - the maximum number of bits that is written when storing the type (or read when reading it). For an i36 this is 40 bits, for an x86 long double it is 80 bits. This is the size alias analysis is interested in (getTypeStoreSize returns the number of bytes). There doesn't seem to be anything corresponding to this in gcc. (3) getABITypeSizeInBits - this is getTypeStoreSizeInBits rounded up to a multiple of the alignment. For an i36 this is 64, for an x86 long double this is 96 or 128 depending on the OS. This is the spacing between consecutive elements when you form an array out of this type (getABITypeSize returns the number of bytes). This is TYPE_SIZE in gcc. Since successive elements in a SequentialType (arrays, pointers and vectors) need to be aligned, the spacing between them will be given by getABITypeSize. This means that the size of an array is the length times the getABITypeSize. It also means that GEP computations need to use getABITypeSize when computing offsets. Furthermore, if an alloca allocates several elements at once then these too need to be aligned, so the size of the alloca has to be the number of elements multiplied by getABITypeSize. Logically speaking this doesn't have to be the case when allocating just one element, but it is simpler to also use getABITypeSize in this case. So alloca's and mallocs should use getABITypeSize. Finally, since gcc's only notion of size is that given by getABITypeSize, if you want to output assembler etc the same as gcc then getABITypeSize is the size you want. Since a store will overwrite no more than getTypeStoreSize bytes, and a read will read no more than that many bytes, this is the notion of size appropriate for alias analysis calculations. In this patch I have corrected all type size uses except some of those in ScalarReplAggregates, lib/Codegen, lib/Target (the hard cases). I will get around to auditing these too at some point, but I could do with some help. Finally, I made one change which I think wise but others might consider pointless and suboptimal: in an unpacked struct the amount of space allocated for a field is now given by the ABI size rather than getTypeStoreSize. I did this because every other place that reserves memory for a type (eg: alloca) now uses getABITypeSize, and I didn't want to make an exception for unpacked structs, i.e. I did it to make things more uniform. This only effects structs containing long doubles and arbitrary precision integers. If someone wants to pack these types more tightly they can always use a packed struct. llvm-svn: 43620
2007-11-02 04:53:16 +08:00
pointerSize = TD.getTypeStoreSize(V->getType());
} else if (FreeInst* F = dyn_cast<FreeInst>(QI)) {
pointer = F->getPointerOperand();
// FreeInsts erase the entire structure
pointerSize = ~0UL;
} else if (CallSite::get(QI).getInstruction() != 0) {
// Call insts need special handling. Check if they can modify our pointer
AliasAnalysis::ModRefResult MR = AA.getModRefInfo(CallSite::get(QI),
dependee, dependeeSize);
if (MR != AliasAnalysis::NoModRef) {
// Loads don't depend on read-only calls
if (isa<LoadInst>(query) && MR == AliasAnalysis::Ref)
continue;
if (!start && !block) {
cachedResult = DepResultTy(QI, Normal);
reverseDep[QI].insert(query);
}
return MemDepResult::get(QI);
} else {
continue;
}
}
// If we found a pointer, check if it could be the same as our pointer
if (pointer) {
AliasAnalysis::AliasResult R = AA.alias(pointer, pointerSize,
dependee, dependeeSize);
if (R != AliasAnalysis::NoAlias) {
// May-alias loads don't depend on each other
if (isa<LoadInst>(query) && isa<LoadInst>(QI) &&
R == AliasAnalysis::MayAlias)
continue;
if (!start && !block) {
cachedResult = DepResultTy(QI, Normal);
reverseDep[QI].insert(query);
}
return MemDepResult::get(QI);
}
}
}
// If we found nothing, return the non-local flag
if (!start && !block)
cachedResult = DepResultTy(0, NonLocal);
return MemDepResult::getNonLocal();
}
/// dropInstruction - Remove an instruction from the analysis, making
/// absolutely conservative assumptions when updating the cache. This is
/// useful, for example when an instruction is changed rather than removed.
void MemoryDependenceAnalysis::dropInstruction(Instruction* drop) {
LocalDepMapType::iterator depGraphEntry = LocalDeps.find(drop);
if (depGraphEntry != LocalDeps.end())
if (Instruction *Inst = depGraphEntry->second.getPointer())
reverseDep[Inst].erase(drop);
// Drop dependency information for things that depended on this instr
SmallPtrSet<Instruction*, 4>& set = reverseDep[drop];
for (SmallPtrSet<Instruction*, 4>::iterator I = set.begin(), E = set.end();
I != E; ++I)
LocalDeps.erase(*I);
LocalDeps.erase(drop);
reverseDep.erase(drop);
for (DenseMap<BasicBlock*, DepResultTy>::iterator DI =
depGraphNonLocal[drop].begin(), DE = depGraphNonLocal[drop].end();
DI != DE; ++DI)
if (Instruction *Inst = DI->second.getPointer())
reverseDepNonLocal[Inst].erase(drop);
if (reverseDepNonLocal.count(drop)) {
SmallPtrSet<Instruction*, 4>& set =
reverseDepNonLocal[drop];
for (SmallPtrSet<Instruction*, 4>::iterator I = set.begin(), E = set.end();
I != E; ++I)
for (DenseMap<BasicBlock*, DepResultTy>::iterator DI =
depGraphNonLocal[*I].begin(), DE = depGraphNonLocal[*I].end();
DI != DE; ++DI)
if (DI->second == DepResultTy(drop, Normal))
// FIXME: Why not remember the old insertion point??
DI->second = DepResultTy(0, Dirty);
}
reverseDepNonLocal.erase(drop);
depGraphNonLocal.erase(drop);
}
/// removeInstruction - Remove an instruction from the dependence analysis,
/// updating the dependence of instructions that previously depended on it.
/// This method attempts to keep the cache coherent using the reverse map.
void MemoryDependenceAnalysis::removeInstruction(Instruction *RemInst) {
// Walk through the Non-local dependencies, removing this one as the value
// for any cached queries.
for (DenseMap<BasicBlock*, DepResultTy>::iterator DI =
depGraphNonLocal[RemInst].begin(), DE = depGraphNonLocal[RemInst].end();
DI != DE; ++DI)
if (Instruction *Inst = DI->second.getPointer())
reverseDepNonLocal[Inst].erase(RemInst);
// Shortly after this, we will look for things that depend on RemInst. In
// order to update these, we'll need a new dependency to base them on. We
// could completely delete any entries that depend on this, but it is better
// to make a more accurate approximation where possible. Compute that better
// approximation if we can.
DepResultTy NewDependency;
// If we have a cached local dependence query for this instruction, remove it.
//
LocalDepMapType::iterator LocalDepEntry = LocalDeps.find(RemInst);
if (LocalDepEntry != LocalDeps.end()) {
DepResultTy LocalDep = LocalDepEntry->second;
// Remove this local dependency info.
LocalDeps.erase(LocalDepEntry);
// Remove us from DepInst's reverse set now that the local dep info is gone.
if (Instruction *Inst = LocalDep.getPointer())
reverseDep[Inst].erase(RemInst);
// If we have unconfirmed info, don't trust it.
if (LocalDep.getInt() != Dirty) {
// If we have a confirmed non-local flag, use it.
if (LocalDep.getInt() == NonLocal || LocalDep.getInt() == None) {
// The only time this dependency is confirmed is if it is non-local.
NewDependency = LocalDep;
} else {
// If we have dep info for RemInst, set them to it.
Instruction *NDI = next(BasicBlock::iterator(LocalDep.getPointer()));
if (NDI != RemInst) // Don't use RemInst for the new dependency!
NewDependency = DepResultTy(NDI, Dirty);
}
}
}
// If we don't already have a local dependency answer for this instruction,
// use the immediate successor of RemInst. We use the successor because
// getDependence starts by checking the immediate predecessor of what is in
// the cache.
if (NewDependency == DepResultTy(0, Dirty))
NewDependency = DepResultTy(next(BasicBlock::iterator(RemInst)), Dirty);
// Loop over all of the things that depend on the instruction we're removing.
//
reverseDepMapType::iterator ReverseDepIt = reverseDep.find(RemInst);
if (ReverseDepIt != reverseDep.end()) {
SmallPtrSet<Instruction*, 4> &ReverseDeps = ReverseDepIt->second;
for (SmallPtrSet<Instruction*, 4>::iterator I = ReverseDeps.begin(),
E = ReverseDeps.end(); I != E; ++I) {
Instruction *InstDependingOnRemInst = *I;
// If we thought the instruction depended on itself (possible for
// unconfirmed dependencies) ignore the update.
if (InstDependingOnRemInst == RemInst) continue;
// Insert the new dependencies.
LocalDeps[InstDependingOnRemInst] = NewDependency;
// If our NewDependency is an instruction, make sure to remember that new
// things depend on it.
if (Instruction *Inst = NewDependency.getPointer())
reverseDep[Inst].insert(InstDependingOnRemInst);
}
reverseDep.erase(RemInst);
}
ReverseDepIt = reverseDepNonLocal.find(RemInst);
if (ReverseDepIt != reverseDepNonLocal.end()) {
SmallPtrSet<Instruction*, 4>& set = ReverseDepIt->second;
for (SmallPtrSet<Instruction*, 4>::iterator I = set.begin(), E = set.end();
I != E; ++I)
for (DenseMap<BasicBlock*, DepResultTy>::iterator DI =
depGraphNonLocal[*I].begin(), DE = depGraphNonLocal[*I].end();
DI != DE; ++DI)
if (DI->second == DepResultTy(RemInst, Normal))
// FIXME: Why not remember the old insertion point??
DI->second = DepResultTy(0, Dirty);
reverseDepNonLocal.erase(ReverseDepIt);
}
depGraphNonLocal.erase(RemInst);
getAnalysis<AliasAnalysis>().deleteValue(RemInst);
DEBUG(verifyRemoved(RemInst));
}