10b57cec5SDimitry Andric //===- BlockExtractor.cpp - Extracts blocks into their own functions ------===// 20b57cec5SDimitry Andric // 30b57cec5SDimitry Andric // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 40b57cec5SDimitry Andric // See https://llvm.org/LICENSE.txt for license information. 50b57cec5SDimitry Andric // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 60b57cec5SDimitry Andric // 70b57cec5SDimitry Andric //===----------------------------------------------------------------------===// 80b57cec5SDimitry Andric // 90b57cec5SDimitry Andric // This pass extracts the specified basic blocks from the module into their 100b57cec5SDimitry Andric // own functions. 110b57cec5SDimitry Andric // 120b57cec5SDimitry Andric //===----------------------------------------------------------------------===// 130b57cec5SDimitry Andric 14e8d8bef9SDimitry Andric #include "llvm/Transforms/IPO/BlockExtractor.h" 150b57cec5SDimitry Andric #include "llvm/ADT/STLExtras.h" 160b57cec5SDimitry Andric #include "llvm/ADT/Statistic.h" 170b57cec5SDimitry Andric #include "llvm/IR/Instructions.h" 180b57cec5SDimitry Andric #include "llvm/IR/Module.h" 19e8d8bef9SDimitry Andric #include "llvm/IR/PassManager.h" 200b57cec5SDimitry Andric #include "llvm/Support/CommandLine.h" 210b57cec5SDimitry Andric #include "llvm/Support/Debug.h" 220b57cec5SDimitry Andric #include "llvm/Support/MemoryBuffer.h" 230b57cec5SDimitry Andric #include "llvm/Transforms/IPO.h" 240b57cec5SDimitry Andric #include "llvm/Transforms/Utils/BasicBlockUtils.h" 250b57cec5SDimitry Andric #include "llvm/Transforms/Utils/CodeExtractor.h" 260b57cec5SDimitry Andric 270b57cec5SDimitry Andric using namespace llvm; 280b57cec5SDimitry Andric 290b57cec5SDimitry Andric #define DEBUG_TYPE "block-extractor" 300b57cec5SDimitry Andric 310b57cec5SDimitry Andric STATISTIC(NumExtracted, "Number of basic blocks extracted"); 320b57cec5SDimitry Andric 330b57cec5SDimitry Andric static cl::opt<std::string> BlockExtractorFile( 340b57cec5SDimitry Andric "extract-blocks-file", cl::value_desc("filename"), 350b57cec5SDimitry Andric cl::desc("A file containing list of basic blocks to extract"), cl::Hidden); 360b57cec5SDimitry Andric 37fe6060f1SDimitry Andric static cl::opt<bool> 38fe6060f1SDimitry Andric BlockExtractorEraseFuncs("extract-blocks-erase-funcs", 390b57cec5SDimitry Andric cl::desc("Erase the existing functions"), 400b57cec5SDimitry Andric cl::Hidden); 410b57cec5SDimitry Andric namespace { 42e8d8bef9SDimitry Andric class BlockExtractor { 43e8d8bef9SDimitry Andric public: 44e8d8bef9SDimitry Andric BlockExtractor(bool EraseFunctions) : EraseFunctions(EraseFunctions) {} 45e8d8bef9SDimitry Andric bool runOnModule(Module &M); 46bdd1243dSDimitry Andric void 47bdd1243dSDimitry Andric init(const std::vector<std::vector<BasicBlock *>> &GroupsOfBlocksToExtract) { 48bdd1243dSDimitry Andric GroupsOfBlocks = GroupsOfBlocksToExtract; 490b57cec5SDimitry Andric if (!BlockExtractorFile.empty()) 500b57cec5SDimitry Andric loadFile(); 510b57cec5SDimitry Andric } 520b57cec5SDimitry Andric 53e8d8bef9SDimitry Andric private: 54bdd1243dSDimitry Andric std::vector<std::vector<BasicBlock *>> GroupsOfBlocks; 55e8d8bef9SDimitry Andric bool EraseFunctions; 56e8d8bef9SDimitry Andric /// Map a function name to groups of blocks. 57e8d8bef9SDimitry Andric SmallVector<std::pair<std::string, SmallVector<std::string, 4>>, 4> 58e8d8bef9SDimitry Andric BlocksByName; 59e8d8bef9SDimitry Andric 60e8d8bef9SDimitry Andric void loadFile(); 61e8d8bef9SDimitry Andric void splitLandingPadPreds(Function &F); 62e8d8bef9SDimitry Andric }; 63e8d8bef9SDimitry Andric 640b57cec5SDimitry Andric } // end anonymous namespace 650b57cec5SDimitry Andric 660b57cec5SDimitry Andric /// Gets all of the blocks specified in the input file. 670b57cec5SDimitry Andric void BlockExtractor::loadFile() { 680b57cec5SDimitry Andric auto ErrOrBuf = MemoryBuffer::getFile(BlockExtractorFile); 690b57cec5SDimitry Andric if (ErrOrBuf.getError()) 700b57cec5SDimitry Andric report_fatal_error("BlockExtractor couldn't load the file."); 710b57cec5SDimitry Andric // Read the file. 720b57cec5SDimitry Andric auto &Buf = *ErrOrBuf; 730b57cec5SDimitry Andric SmallVector<StringRef, 16> Lines; 740b57cec5SDimitry Andric Buf->getBuffer().split(Lines, '\n', /*MaxSplit=*/-1, 750b57cec5SDimitry Andric /*KeepEmpty=*/false); 760b57cec5SDimitry Andric for (const auto &Line : Lines) { 770b57cec5SDimitry Andric SmallVector<StringRef, 4> LineSplit; 780b57cec5SDimitry Andric Line.split(LineSplit, ' ', /*MaxSplit=*/-1, 790b57cec5SDimitry Andric /*KeepEmpty=*/false); 800b57cec5SDimitry Andric if (LineSplit.empty()) 810b57cec5SDimitry Andric continue; 828bcb0991SDimitry Andric if (LineSplit.size()!=2) 8381ad6265SDimitry Andric report_fatal_error("Invalid line format, expecting lines like: 'funcname bb1[;bb2..]'", 8481ad6265SDimitry Andric /*GenCrashDiag=*/false); 850b57cec5SDimitry Andric SmallVector<StringRef, 4> BBNames; 860b57cec5SDimitry Andric LineSplit[1].split(BBNames, ';', /*MaxSplit=*/-1, 870b57cec5SDimitry Andric /*KeepEmpty=*/false); 880b57cec5SDimitry Andric if (BBNames.empty()) 890b57cec5SDimitry Andric report_fatal_error("Missing bbs name"); 905ffd83dbSDimitry Andric BlocksByName.push_back( 915ffd83dbSDimitry Andric {std::string(LineSplit[0]), {BBNames.begin(), BBNames.end()}}); 920b57cec5SDimitry Andric } 930b57cec5SDimitry Andric } 940b57cec5SDimitry Andric 950b57cec5SDimitry Andric /// Extracts the landing pads to make sure all of them have only one 960b57cec5SDimitry Andric /// predecessor. 970b57cec5SDimitry Andric void BlockExtractor::splitLandingPadPreds(Function &F) { 980b57cec5SDimitry Andric for (BasicBlock &BB : F) { 990b57cec5SDimitry Andric for (Instruction &I : BB) { 1000b57cec5SDimitry Andric if (!isa<InvokeInst>(&I)) 1010b57cec5SDimitry Andric continue; 1020b57cec5SDimitry Andric InvokeInst *II = cast<InvokeInst>(&I); 1030b57cec5SDimitry Andric BasicBlock *Parent = II->getParent(); 1040b57cec5SDimitry Andric BasicBlock *LPad = II->getUnwindDest(); 1050b57cec5SDimitry Andric 1060b57cec5SDimitry Andric // Look through the landing pad's predecessors. If one of them ends in an 1070b57cec5SDimitry Andric // 'invoke', then we want to split the landing pad. 1080b57cec5SDimitry Andric bool Split = false; 109bdd1243dSDimitry Andric for (auto *PredBB : predecessors(LPad)) { 1100b57cec5SDimitry Andric if (PredBB->isLandingPad() && PredBB != Parent && 1110b57cec5SDimitry Andric isa<InvokeInst>(Parent->getTerminator())) { 1120b57cec5SDimitry Andric Split = true; 1130b57cec5SDimitry Andric break; 1140b57cec5SDimitry Andric } 1150b57cec5SDimitry Andric } 1160b57cec5SDimitry Andric 1170b57cec5SDimitry Andric if (!Split) 1180b57cec5SDimitry Andric continue; 1190b57cec5SDimitry Andric 1200b57cec5SDimitry Andric SmallVector<BasicBlock *, 2> NewBBs; 1210b57cec5SDimitry Andric SplitLandingPadPredecessors(LPad, Parent, ".1", ".2", NewBBs); 1220b57cec5SDimitry Andric } 1230b57cec5SDimitry Andric } 1240b57cec5SDimitry Andric } 1250b57cec5SDimitry Andric 1260b57cec5SDimitry Andric bool BlockExtractor::runOnModule(Module &M) { 1270b57cec5SDimitry Andric bool Changed = false; 1280b57cec5SDimitry Andric 1290b57cec5SDimitry Andric // Get all the functions. 1300b57cec5SDimitry Andric SmallVector<Function *, 4> Functions; 1310b57cec5SDimitry Andric for (Function &F : M) { 1320b57cec5SDimitry Andric splitLandingPadPreds(F); 1330b57cec5SDimitry Andric Functions.push_back(&F); 1340b57cec5SDimitry Andric } 1350b57cec5SDimitry Andric 1360b57cec5SDimitry Andric // Get all the blocks specified in the input file. 1370b57cec5SDimitry Andric unsigned NextGroupIdx = GroupsOfBlocks.size(); 1380b57cec5SDimitry Andric GroupsOfBlocks.resize(NextGroupIdx + BlocksByName.size()); 1390b57cec5SDimitry Andric for (const auto &BInfo : BlocksByName) { 1400b57cec5SDimitry Andric Function *F = M.getFunction(BInfo.first); 1410b57cec5SDimitry Andric if (!F) 14281ad6265SDimitry Andric report_fatal_error("Invalid function name specified in the input file", 14381ad6265SDimitry Andric /*GenCrashDiag=*/false); 1440b57cec5SDimitry Andric for (const auto &BBInfo : BInfo.second) { 145*0fca6ea1SDimitry Andric auto Res = llvm::find_if( 146*0fca6ea1SDimitry Andric *F, [&](const BasicBlock &BB) { return BB.getName() == BBInfo; }); 1470b57cec5SDimitry Andric if (Res == F->end()) 14881ad6265SDimitry Andric report_fatal_error("Invalid block name specified in the input file", 14981ad6265SDimitry Andric /*GenCrashDiag=*/false); 1500b57cec5SDimitry Andric GroupsOfBlocks[NextGroupIdx].push_back(&*Res); 1510b57cec5SDimitry Andric } 1520b57cec5SDimitry Andric ++NextGroupIdx; 1530b57cec5SDimitry Andric } 1540b57cec5SDimitry Andric 1550b57cec5SDimitry Andric // Extract each group of basic blocks. 1560b57cec5SDimitry Andric for (auto &BBs : GroupsOfBlocks) { 1570b57cec5SDimitry Andric SmallVector<BasicBlock *, 32> BlocksToExtractVec; 1580b57cec5SDimitry Andric for (BasicBlock *BB : BBs) { 1590b57cec5SDimitry Andric // Check if the module contains BB. 1600b57cec5SDimitry Andric if (BB->getParent()->getParent() != &M) 16181ad6265SDimitry Andric report_fatal_error("Invalid basic block", /*GenCrashDiag=*/false); 1620b57cec5SDimitry Andric LLVM_DEBUG(dbgs() << "BlockExtractor: Extracting " 1630b57cec5SDimitry Andric << BB->getParent()->getName() << ":" << BB->getName() 1640b57cec5SDimitry Andric << "\n"); 1650b57cec5SDimitry Andric BlocksToExtractVec.push_back(BB); 1660b57cec5SDimitry Andric if (const InvokeInst *II = dyn_cast<InvokeInst>(BB->getTerminator())) 1670b57cec5SDimitry Andric BlocksToExtractVec.push_back(II->getUnwindDest()); 1680b57cec5SDimitry Andric ++NumExtracted; 1690b57cec5SDimitry Andric Changed = true; 1700b57cec5SDimitry Andric } 1718bcb0991SDimitry Andric CodeExtractorAnalysisCache CEAC(*BBs[0]->getParent()); 1728bcb0991SDimitry Andric Function *F = CodeExtractor(BlocksToExtractVec).extractCodeRegion(CEAC); 1730b57cec5SDimitry Andric if (F) 1740b57cec5SDimitry Andric LLVM_DEBUG(dbgs() << "Extracted group '" << (*BBs.begin())->getName() 1750b57cec5SDimitry Andric << "' in: " << F->getName() << '\n'); 1760b57cec5SDimitry Andric else 1770b57cec5SDimitry Andric LLVM_DEBUG(dbgs() << "Failed to extract for group '" 1780b57cec5SDimitry Andric << (*BBs.begin())->getName() << "'\n"); 1790b57cec5SDimitry Andric } 1800b57cec5SDimitry Andric 1810b57cec5SDimitry Andric // Erase the functions. 1820b57cec5SDimitry Andric if (EraseFunctions || BlockExtractorEraseFuncs) { 1830b57cec5SDimitry Andric for (Function *F : Functions) { 1840b57cec5SDimitry Andric LLVM_DEBUG(dbgs() << "BlockExtractor: Trying to delete " << F->getName() 1850b57cec5SDimitry Andric << "\n"); 1860b57cec5SDimitry Andric F->deleteBody(); 1870b57cec5SDimitry Andric } 1880b57cec5SDimitry Andric // Set linkage as ExternalLinkage to avoid erasing unreachable functions. 1890b57cec5SDimitry Andric for (Function &F : M) 1900b57cec5SDimitry Andric F.setLinkage(GlobalValue::ExternalLinkage); 1910b57cec5SDimitry Andric Changed = true; 1920b57cec5SDimitry Andric } 1930b57cec5SDimitry Andric 1940b57cec5SDimitry Andric return Changed; 1950b57cec5SDimitry Andric } 196e8d8bef9SDimitry Andric 197bdd1243dSDimitry Andric BlockExtractorPass::BlockExtractorPass( 198bdd1243dSDimitry Andric std::vector<std::vector<BasicBlock *>> &&GroupsOfBlocks, 199bdd1243dSDimitry Andric bool EraseFunctions) 200bdd1243dSDimitry Andric : GroupsOfBlocks(GroupsOfBlocks), EraseFunctions(EraseFunctions) {} 201e8d8bef9SDimitry Andric 202e8d8bef9SDimitry Andric PreservedAnalyses BlockExtractorPass::run(Module &M, 203e8d8bef9SDimitry Andric ModuleAnalysisManager &AM) { 204bdd1243dSDimitry Andric BlockExtractor BE(EraseFunctions); 205bdd1243dSDimitry Andric BE.init(GroupsOfBlocks); 206e8d8bef9SDimitry Andric return BE.runOnModule(M) ? PreservedAnalyses::none() 207e8d8bef9SDimitry Andric : PreservedAnalyses::all(); 208e8d8bef9SDimitry Andric } 209