1*4ba319b5SDimitry Andric //===- BlockExtractor.cpp - Extracts blocks into their own functions ------===//
2*4ba319b5SDimitry Andric //
3*4ba319b5SDimitry Andric //                     The LLVM Compiler Infrastructure
4*4ba319b5SDimitry Andric //
5*4ba319b5SDimitry Andric // This file is distributed under the University of Illinois Open Source
6*4ba319b5SDimitry Andric // License. See LICENSE.TXT for details.
7*4ba319b5SDimitry Andric //
8*4ba319b5SDimitry Andric //===----------------------------------------------------------------------===//
9*4ba319b5SDimitry Andric //
10*4ba319b5SDimitry Andric // This pass extracts the specified basic blocks from the module into their
11*4ba319b5SDimitry Andric // own functions.
12*4ba319b5SDimitry Andric //
13*4ba319b5SDimitry Andric //===----------------------------------------------------------------------===//
14*4ba319b5SDimitry Andric 
15*4ba319b5SDimitry Andric #include "llvm/ADT/STLExtras.h"
16*4ba319b5SDimitry Andric #include "llvm/ADT/Statistic.h"
17*4ba319b5SDimitry Andric #include "llvm/IR/Instructions.h"
18*4ba319b5SDimitry Andric #include "llvm/IR/Module.h"
19*4ba319b5SDimitry Andric #include "llvm/Pass.h"
20*4ba319b5SDimitry Andric #include "llvm/Support/CommandLine.h"
21*4ba319b5SDimitry Andric #include "llvm/Support/Debug.h"
22*4ba319b5SDimitry Andric #include "llvm/Support/MemoryBuffer.h"
23*4ba319b5SDimitry Andric #include "llvm/Transforms/IPO.h"
24*4ba319b5SDimitry Andric #include "llvm/Transforms/Utils/BasicBlockUtils.h"
25*4ba319b5SDimitry Andric #include "llvm/Transforms/Utils/CodeExtractor.h"
26*4ba319b5SDimitry Andric using namespace llvm;
27*4ba319b5SDimitry Andric 
28*4ba319b5SDimitry Andric #define DEBUG_TYPE "block-extractor"
29*4ba319b5SDimitry Andric 
30*4ba319b5SDimitry Andric STATISTIC(NumExtracted, "Number of basic blocks extracted");
31*4ba319b5SDimitry Andric 
32*4ba319b5SDimitry Andric static cl::opt<std::string> BlockExtractorFile(
33*4ba319b5SDimitry Andric     "extract-blocks-file", cl::value_desc("filename"),
34*4ba319b5SDimitry Andric     cl::desc("A file containing list of basic blocks to extract"), cl::Hidden);
35*4ba319b5SDimitry Andric 
36*4ba319b5SDimitry Andric cl::opt<bool> BlockExtractorEraseFuncs("extract-blocks-erase-funcs",
37*4ba319b5SDimitry Andric                                        cl::desc("Erase the existing functions"),
38*4ba319b5SDimitry Andric                                        cl::Hidden);
39*4ba319b5SDimitry Andric 
40*4ba319b5SDimitry Andric namespace {
41*4ba319b5SDimitry Andric class BlockExtractor : public ModulePass {
42*4ba319b5SDimitry Andric   SmallVector<BasicBlock *, 16> Blocks;
43*4ba319b5SDimitry Andric   bool EraseFunctions;
44*4ba319b5SDimitry Andric   SmallVector<std::pair<std::string, std::string>, 32> BlocksByName;
45*4ba319b5SDimitry Andric 
46*4ba319b5SDimitry Andric public:
47*4ba319b5SDimitry Andric   static char ID;
BlockExtractor(const SmallVectorImpl<BasicBlock * > & BlocksToExtract,bool EraseFunctions)48*4ba319b5SDimitry Andric   BlockExtractor(const SmallVectorImpl<BasicBlock *> &BlocksToExtract,
49*4ba319b5SDimitry Andric                  bool EraseFunctions)
50*4ba319b5SDimitry Andric       : ModulePass(ID), Blocks(BlocksToExtract.begin(), BlocksToExtract.end()),
51*4ba319b5SDimitry Andric         EraseFunctions(EraseFunctions) {
52*4ba319b5SDimitry Andric     if (!BlockExtractorFile.empty())
53*4ba319b5SDimitry Andric       loadFile();
54*4ba319b5SDimitry Andric   }
BlockExtractor()55*4ba319b5SDimitry Andric   BlockExtractor() : BlockExtractor(SmallVector<BasicBlock *, 0>(), false) {}
56*4ba319b5SDimitry Andric   bool runOnModule(Module &M) override;
57*4ba319b5SDimitry Andric 
58*4ba319b5SDimitry Andric private:
59*4ba319b5SDimitry Andric   void loadFile();
60*4ba319b5SDimitry Andric   void splitLandingPadPreds(Function &F);
61*4ba319b5SDimitry Andric };
62*4ba319b5SDimitry Andric } // end anonymous namespace
63*4ba319b5SDimitry Andric 
64*4ba319b5SDimitry Andric char BlockExtractor::ID = 0;
65*4ba319b5SDimitry Andric INITIALIZE_PASS(BlockExtractor, "extract-blocks",
66*4ba319b5SDimitry Andric                 "Extract basic blocks from module", false, false)
67*4ba319b5SDimitry Andric 
createBlockExtractorPass()68*4ba319b5SDimitry Andric ModulePass *llvm::createBlockExtractorPass() { return new BlockExtractor(); }
createBlockExtractorPass(const SmallVectorImpl<BasicBlock * > & BlocksToExtract,bool EraseFunctions)69*4ba319b5SDimitry Andric ModulePass *llvm::createBlockExtractorPass(
70*4ba319b5SDimitry Andric     const SmallVectorImpl<BasicBlock *> &BlocksToExtract, bool EraseFunctions) {
71*4ba319b5SDimitry Andric   return new BlockExtractor(BlocksToExtract, EraseFunctions);
72*4ba319b5SDimitry Andric }
73*4ba319b5SDimitry Andric 
74*4ba319b5SDimitry Andric /// Gets all of the blocks specified in the input file.
loadFile()75*4ba319b5SDimitry Andric void BlockExtractor::loadFile() {
76*4ba319b5SDimitry Andric   auto ErrOrBuf = MemoryBuffer::getFile(BlockExtractorFile);
77*4ba319b5SDimitry Andric   if (ErrOrBuf.getError())
78*4ba319b5SDimitry Andric     report_fatal_error("BlockExtractor couldn't load the file.");
79*4ba319b5SDimitry Andric   // Read the file.
80*4ba319b5SDimitry Andric   auto &Buf = *ErrOrBuf;
81*4ba319b5SDimitry Andric   SmallVector<StringRef, 16> Lines;
82*4ba319b5SDimitry Andric   Buf->getBuffer().split(Lines, '\n', /*MaxSplit=*/-1,
83*4ba319b5SDimitry Andric                          /*KeepEmpty=*/false);
84*4ba319b5SDimitry Andric   for (const auto &Line : Lines) {
85*4ba319b5SDimitry Andric     auto FBPair = Line.split(' ');
86*4ba319b5SDimitry Andric     BlocksByName.push_back({FBPair.first, FBPair.second});
87*4ba319b5SDimitry Andric   }
88*4ba319b5SDimitry Andric }
89*4ba319b5SDimitry Andric 
90*4ba319b5SDimitry Andric /// Extracts the landing pads to make sure all of them have only one
91*4ba319b5SDimitry Andric /// predecessor.
splitLandingPadPreds(Function & F)92*4ba319b5SDimitry Andric void BlockExtractor::splitLandingPadPreds(Function &F) {
93*4ba319b5SDimitry Andric   for (BasicBlock &BB : F) {
94*4ba319b5SDimitry Andric     for (Instruction &I : BB) {
95*4ba319b5SDimitry Andric       if (!isa<InvokeInst>(&I))
96*4ba319b5SDimitry Andric         continue;
97*4ba319b5SDimitry Andric       InvokeInst *II = cast<InvokeInst>(&I);
98*4ba319b5SDimitry Andric       BasicBlock *Parent = II->getParent();
99*4ba319b5SDimitry Andric       BasicBlock *LPad = II->getUnwindDest();
100*4ba319b5SDimitry Andric 
101*4ba319b5SDimitry Andric       // Look through the landing pad's predecessors. If one of them ends in an
102*4ba319b5SDimitry Andric       // 'invoke', then we want to split the landing pad.
103*4ba319b5SDimitry Andric       bool Split = false;
104*4ba319b5SDimitry Andric       for (auto PredBB : predecessors(LPad)) {
105*4ba319b5SDimitry Andric         if (PredBB->isLandingPad() && PredBB != Parent &&
106*4ba319b5SDimitry Andric             isa<InvokeInst>(Parent->getTerminator())) {
107*4ba319b5SDimitry Andric           Split = true;
108*4ba319b5SDimitry Andric           break;
109*4ba319b5SDimitry Andric         }
110*4ba319b5SDimitry Andric       }
111*4ba319b5SDimitry Andric 
112*4ba319b5SDimitry Andric       if (!Split)
113*4ba319b5SDimitry Andric         continue;
114*4ba319b5SDimitry Andric 
115*4ba319b5SDimitry Andric       SmallVector<BasicBlock *, 2> NewBBs;
116*4ba319b5SDimitry Andric       SplitLandingPadPredecessors(LPad, Parent, ".1", ".2", NewBBs);
117*4ba319b5SDimitry Andric     }
118*4ba319b5SDimitry Andric   }
119*4ba319b5SDimitry Andric }
120*4ba319b5SDimitry Andric 
runOnModule(Module & M)121*4ba319b5SDimitry Andric bool BlockExtractor::runOnModule(Module &M) {
122*4ba319b5SDimitry Andric 
123*4ba319b5SDimitry Andric   bool Changed = false;
124*4ba319b5SDimitry Andric 
125*4ba319b5SDimitry Andric   // Get all the functions.
126*4ba319b5SDimitry Andric   SmallVector<Function *, 4> Functions;
127*4ba319b5SDimitry Andric   for (Function &F : M) {
128*4ba319b5SDimitry Andric     splitLandingPadPreds(F);
129*4ba319b5SDimitry Andric     Functions.push_back(&F);
130*4ba319b5SDimitry Andric   }
131*4ba319b5SDimitry Andric 
132*4ba319b5SDimitry Andric   // Get all the blocks specified in the input file.
133*4ba319b5SDimitry Andric   for (const auto &BInfo : BlocksByName) {
134*4ba319b5SDimitry Andric     Function *F = M.getFunction(BInfo.first);
135*4ba319b5SDimitry Andric     if (!F)
136*4ba319b5SDimitry Andric       report_fatal_error("Invalid function name specified in the input file");
137*4ba319b5SDimitry Andric     auto Res = llvm::find_if(*F, [&](const BasicBlock &BB) {
138*4ba319b5SDimitry Andric       return BB.getName().equals(BInfo.second);
139*4ba319b5SDimitry Andric     });
140*4ba319b5SDimitry Andric     if (Res == F->end())
141*4ba319b5SDimitry Andric       report_fatal_error("Invalid block name specified in the input file");
142*4ba319b5SDimitry Andric     Blocks.push_back(&*Res);
143*4ba319b5SDimitry Andric   }
144*4ba319b5SDimitry Andric 
145*4ba319b5SDimitry Andric   // Extract basic blocks.
146*4ba319b5SDimitry Andric   for (BasicBlock *BB : Blocks) {
147*4ba319b5SDimitry Andric     // Check if the module contains BB.
148*4ba319b5SDimitry Andric     if (BB->getParent()->getParent() != &M)
149*4ba319b5SDimitry Andric       report_fatal_error("Invalid basic block");
150*4ba319b5SDimitry Andric     LLVM_DEBUG(dbgs() << "BlockExtractor: Extracting "
151*4ba319b5SDimitry Andric                       << BB->getParent()->getName() << ":" << BB->getName()
152*4ba319b5SDimitry Andric                       << "\n");
153*4ba319b5SDimitry Andric     SmallVector<BasicBlock *, 2> BlocksToExtractVec;
154*4ba319b5SDimitry Andric     BlocksToExtractVec.push_back(BB);
155*4ba319b5SDimitry Andric     if (const InvokeInst *II = dyn_cast<InvokeInst>(BB->getTerminator()))
156*4ba319b5SDimitry Andric       BlocksToExtractVec.push_back(II->getUnwindDest());
157*4ba319b5SDimitry Andric     CodeExtractor(BlocksToExtractVec).extractCodeRegion();
158*4ba319b5SDimitry Andric     ++NumExtracted;
159*4ba319b5SDimitry Andric     Changed = true;
160*4ba319b5SDimitry Andric   }
161*4ba319b5SDimitry Andric 
162*4ba319b5SDimitry Andric   // Erase the functions.
163*4ba319b5SDimitry Andric   if (EraseFunctions || BlockExtractorEraseFuncs) {
164*4ba319b5SDimitry Andric     for (Function *F : Functions) {
165*4ba319b5SDimitry Andric       LLVM_DEBUG(dbgs() << "BlockExtractor: Trying to delete " << F->getName()
166*4ba319b5SDimitry Andric                         << "\n");
167*4ba319b5SDimitry Andric       F->deleteBody();
168*4ba319b5SDimitry Andric     }
169*4ba319b5SDimitry Andric     // Set linkage as ExternalLinkage to avoid erasing unreachable functions.
170*4ba319b5SDimitry Andric     for (Function &F : M)
171*4ba319b5SDimitry Andric       F.setLinkage(GlobalValue::ExternalLinkage);
172*4ba319b5SDimitry Andric     Changed = true;
173*4ba319b5SDimitry Andric   }
174*4ba319b5SDimitry Andric 
175*4ba319b5SDimitry Andric   return Changed;
176*4ba319b5SDimitry Andric }
177