maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

Rewrite ICF using parallel utilities 

Summary: Rewrite ICF using parallel utilities

(cherry picked from FBD16472975)
This commit is contained in:
laith sakka 2019-07-24 17:13:15 -07:00 committed by Maksim Panchenko
parent 8d5854ef09
commit 5084534699
1 changed files with 27 additions and 51 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

78
bolt/src/Passes/IdenticalCodeFolding.cpp
View File

@ -331,55 +331,26 @@ void IdenticalCodeFolding::runOnFunctions(BinaryContext &BC) {
std::atomic<uint64_t> CallsSavedEstimate{0};
std::atomic<uint64_t> NumFoldedLastIteration{0};
CongruentBucketsMap CongruentBuckets;
std::unique_ptr<ThreadPool> ThPool;
if (!opts::NoThreads)
ThPool = std::make_unique<ThreadPool>(opts::ThreadCount);
// Hash all the functions
auto hashFunctions = [&]() {
NamedRegionTimer HashFunctionsTimer("hashing", "hashing", "ICF breakdown",
"ICF breakdown", opts::TimeICF);
ParallelUtilities::WorkFuncTy WorkFun = [&](BinaryFunction &BF) {
// Make sure indices are in-order.
BF.updateLayoutIndices();
// Perform hashing for a block of functions
auto hashBlock =
[&](std::map<uint64_t, BinaryFunction>::iterator BlockBegin,
std::map<uint64_t, BinaryFunction>::iterator BlockEnd) {
Timer T("hash block", "hash block");
DEBUG(T.startTimer());
for (auto It = BlockBegin; It != BlockEnd; ++It) {
auto &BF = It->second;
if (!this->shouldOptimize(BF))
continue;
// Make sure indices are in-order.
BF.updateLayoutIndices();
// Pre-compute hash before pushing into hashtable.
BF.hash(/*Recompute=*/true, opts::UseDFS);
}
DEBUG(T.stopTimer());
// Pre-compute hash before pushing into hashtable.
BF.hash(/*Recompute=*/true, opts::UseDFS);
};
if (opts::NoThreads) {
hashBlock(BC.getBinaryFunctions().begin(), BC.getBinaryFunctions().end());
return;
}
ParallelUtilities::PredicateTy SkipFunc = [&](const BinaryFunction &BF) {
return !shouldOptimize(BF);
};
const unsigned BlockSize = OriginalFunctionCount / (2 * opts::ThreadCount);
unsigned Counter = 0;
auto BlockBegin = BC.getBinaryFunctions().begin();
for (auto It = BC.getBinaryFunctions().begin();
It != BC.getBinaryFunctions().end(); ++It, ++Counter) {
if (Counter >= BlockSize) {
ThPool->async(hashBlock, BlockBegin, std::next(It));
BlockBegin = std::next(It);
Counter = 0;
}
}
ThPool->async(hashBlock, BlockBegin, BC.getBinaryFunctions().end());
ThPool->wait();
ParallelUtilities::runOnEachFunction(
BC, ParallelUtilities::SchedulingPolicy::SP_TRIVIAL, WorkFun, SkipFunc,
"hashFunctions", /*ForceSequential*/ false, 2);
};
// Creates buckets with congruent functions - functions that potentially
@ -405,8 +376,12 @@ void IdenticalCodeFolding::runOnFunctions(BinaryContext &BC) {
Timer SinglePass("single fold pass", "single fold pass");
DEBUG(SinglePass.startTimer());
// Perform the work for a single congruent list
auto performFoldingForItem = [&](std::set<BinaryFunction *> &Candidates) {
ThreadPool *ThPool;
if (!opts::NoThreads)
ThPool = &ParallelUtilities::getThreadPool();
// Fold identical functions within a single congruent bucket
auto procesSingleBucket = [&](std::set<BinaryFunction *> &Candidates) {
Timer T("folding single congruent list", "folding single congruent list");
DEBUG(T.startTimer());
@ -426,7 +401,8 @@ void IdenticalCodeFolding::runOnFunctions(BinaryContext &BC) {
// different options.
std::stable_sort(Twins.begin(), Twins.end(),
[](const BinaryFunction *A, const BinaryFunction *B) {
return A->getFunctionNumber() < B->getFunctionNumber();
return A->getFunctionNumber() <
B->getFunctionNumber();
});
BinaryFunction *ParentBF = Twins[0];
@ -457,21 +433,21 @@ void IdenticalCodeFolding::runOnFunctions(BinaryContext &BC) {
DEBUG(T.stopTimer());
};
// Create a task for each congruent list
// Create a task for each congruent bucket
for (auto &Entry : CongruentBuckets) {
auto &Candidates = Entry.second;
if (Candidates.size() < 2)
auto &Bucket = Entry.second;
if (Bucket.size() < 2)
continue;
if (opts::NoThreads)
performFoldingForItem(Candidates);
procesSingleBucket(Bucket);
else
ThPool->async(performFoldingForItem, std::ref(Candidates));
ThPool->async(procesSingleBucket, std::ref(Bucket));
}
if (opts::NoThreads)
return;
ThPool->wait();
if (!opts::NoThreads)
ThPool->wait();
DEBUG(SinglePass.stopTimer());
};