1dc40be75SVolkan Keles //===- BlockExtractor.cpp - Extracts blocks into their own functions ------===//
2dc40be75SVolkan Keles //
3dc40be75SVolkan Keles //                     The LLVM Compiler Infrastructure
4dc40be75SVolkan Keles //
5dc40be75SVolkan Keles // This file is distributed under the University of Illinois Open Source
6dc40be75SVolkan Keles // License. See LICENSE.TXT for details.
7dc40be75SVolkan Keles //
8dc40be75SVolkan Keles //===----------------------------------------------------------------------===//
9dc40be75SVolkan Keles //
10dc40be75SVolkan Keles // This pass extracts the specified basic blocks from the module into their
11dc40be75SVolkan Keles // own functions.
12dc40be75SVolkan Keles //
13dc40be75SVolkan Keles //===----------------------------------------------------------------------===//
14dc40be75SVolkan Keles 
15dc40be75SVolkan Keles #include "llvm/ADT/STLExtras.h"
16dc40be75SVolkan Keles #include "llvm/ADT/Statistic.h"
17dc40be75SVolkan Keles #include "llvm/IR/Instructions.h"
18dc40be75SVolkan Keles #include "llvm/IR/Module.h"
19dc40be75SVolkan Keles #include "llvm/Pass.h"
20dc40be75SVolkan Keles #include "llvm/Support/CommandLine.h"
21dc40be75SVolkan Keles #include "llvm/Support/Debug.h"
22dc40be75SVolkan Keles #include "llvm/Support/MemoryBuffer.h"
23dc40be75SVolkan Keles #include "llvm/Transforms/IPO.h"
24dc40be75SVolkan Keles #include "llvm/Transforms/Utils/BasicBlockUtils.h"
25dc40be75SVolkan Keles #include "llvm/Transforms/Utils/CodeExtractor.h"
26dc40be75SVolkan Keles using namespace llvm;
27dc40be75SVolkan Keles 
28dc40be75SVolkan Keles #define DEBUG_TYPE "block-extractor"
29dc40be75SVolkan Keles 
30dc40be75SVolkan Keles STATISTIC(NumExtracted, "Number of basic blocks extracted");
31dc40be75SVolkan Keles 
32dc40be75SVolkan Keles static cl::opt<std::string> BlockExtractorFile(
33dc40be75SVolkan Keles     "extract-blocks-file", cl::value_desc("filename"),
34dc40be75SVolkan Keles     cl::desc("A file containing list of basic blocks to extract"), cl::Hidden);
35dc40be75SVolkan Keles 
36dc40be75SVolkan Keles cl::opt<bool> BlockExtractorEraseFuncs("extract-blocks-erase-funcs",
37dc40be75SVolkan Keles                                        cl::desc("Erase the existing functions"),
38dc40be75SVolkan Keles                                        cl::Hidden);
39dc40be75SVolkan Keles 
40dc40be75SVolkan Keles namespace {
41dc40be75SVolkan Keles class BlockExtractor : public ModulePass {
42dc40be75SVolkan Keles   SmallVector<BasicBlock *, 16> Blocks;
43dc40be75SVolkan Keles   bool EraseFunctions;
44dc40be75SVolkan Keles   SmallVector<std::pair<std::string, std::string>, 32> BlocksByName;
45dc40be75SVolkan Keles 
46dc40be75SVolkan Keles public:
47dc40be75SVolkan Keles   static char ID;
48dc40be75SVolkan Keles   BlockExtractor(const SmallVectorImpl<BasicBlock *> &BlocksToExtract,
49dc40be75SVolkan Keles                  bool EraseFunctions)
50dc40be75SVolkan Keles       : ModulePass(ID), Blocks(BlocksToExtract.begin(), BlocksToExtract.end()),
51dc40be75SVolkan Keles         EraseFunctions(EraseFunctions) {
52dc40be75SVolkan Keles     if (!BlockExtractorFile.empty())
53dc40be75SVolkan Keles       loadFile();
54dc40be75SVolkan Keles   }
55dc40be75SVolkan Keles   BlockExtractor() : BlockExtractor(SmallVector<BasicBlock *, 0>(), false) {}
56dc40be75SVolkan Keles   bool runOnModule(Module &M) override;
57dc40be75SVolkan Keles 
58dc40be75SVolkan Keles private:
59dc40be75SVolkan Keles   void loadFile();
60dc40be75SVolkan Keles   void splitLandingPadPreds(Function &F);
61dc40be75SVolkan Keles };
62dc40be75SVolkan Keles } // end anonymous namespace
63dc40be75SVolkan Keles 
64dc40be75SVolkan Keles char BlockExtractor::ID = 0;
65dc40be75SVolkan Keles INITIALIZE_PASS(BlockExtractor, "extract-blocks",
66dc40be75SVolkan Keles                 "Extract basic blocks from module", false, false)
67dc40be75SVolkan Keles 
68dc40be75SVolkan Keles ModulePass *llvm::createBlockExtractorPass() { return new BlockExtractor(); }
69dc40be75SVolkan Keles ModulePass *llvm::createBlockExtractorPass(
70dc40be75SVolkan Keles     const SmallVectorImpl<BasicBlock *> &BlocksToExtract, bool EraseFunctions) {
71dc40be75SVolkan Keles   return new BlockExtractor(BlocksToExtract, EraseFunctions);
72dc40be75SVolkan Keles }
73dc40be75SVolkan Keles 
74dc40be75SVolkan Keles /// Gets all of the blocks specified in the input file.
75dc40be75SVolkan Keles void BlockExtractor::loadFile() {
76dc40be75SVolkan Keles   auto ErrOrBuf = MemoryBuffer::getFile(BlockExtractorFile);
77*ebf34ea3SVolkan Keles   if (ErrOrBuf.getError())
78dc40be75SVolkan Keles     report_fatal_error("BlockExtractor couldn't load the file.");
79dc40be75SVolkan Keles   // Read the file.
80dc40be75SVolkan Keles   auto &Buf = *ErrOrBuf;
81dc40be75SVolkan Keles   SmallVector<StringRef, 16> Lines;
82dc40be75SVolkan Keles   Buf->getBuffer().split(Lines, '\n', /*MaxSplit=*/-1,
83dc40be75SVolkan Keles                          /*KeepEmpty=*/false);
84dc40be75SVolkan Keles   for (const auto &Line : Lines) {
85dc40be75SVolkan Keles     auto FBPair = Line.split(' ');
86dc40be75SVolkan Keles     BlocksByName.push_back({FBPair.first, FBPair.second});
87dc40be75SVolkan Keles   }
88dc40be75SVolkan Keles }
89dc40be75SVolkan Keles 
90dc40be75SVolkan Keles /// Extracts the landing pads to make sure all of them have only one
91dc40be75SVolkan Keles /// predecessor.
92dc40be75SVolkan Keles void BlockExtractor::splitLandingPadPreds(Function &F) {
93dc40be75SVolkan Keles   for (BasicBlock &BB : F) {
94dc40be75SVolkan Keles     for (Instruction &I : BB) {
95dc40be75SVolkan Keles       if (!isa<InvokeInst>(&I))
96dc40be75SVolkan Keles         continue;
97dc40be75SVolkan Keles       InvokeInst *II = cast<InvokeInst>(&I);
98dc40be75SVolkan Keles       BasicBlock *Parent = II->getParent();
99dc40be75SVolkan Keles       BasicBlock *LPad = II->getUnwindDest();
100dc40be75SVolkan Keles 
101dc40be75SVolkan Keles       // Look through the landing pad's predecessors. If one of them ends in an
102dc40be75SVolkan Keles       // 'invoke', then we want to split the landing pad.
103dc40be75SVolkan Keles       bool Split = false;
104dc40be75SVolkan Keles       for (auto PredBB : predecessors(LPad)) {
105dc40be75SVolkan Keles         if (PredBB->isLandingPad() && PredBB != Parent &&
106dc40be75SVolkan Keles             isa<InvokeInst>(Parent->getTerminator())) {
107dc40be75SVolkan Keles           Split = true;
108dc40be75SVolkan Keles           break;
109dc40be75SVolkan Keles         }
110dc40be75SVolkan Keles       }
111dc40be75SVolkan Keles 
112dc40be75SVolkan Keles       if (!Split)
113dc40be75SVolkan Keles         continue;
114dc40be75SVolkan Keles 
115dc40be75SVolkan Keles       SmallVector<BasicBlock *, 2> NewBBs;
116dc40be75SVolkan Keles       SplitLandingPadPredecessors(LPad, Parent, ".1", ".2", NewBBs);
117dc40be75SVolkan Keles     }
118dc40be75SVolkan Keles   }
119dc40be75SVolkan Keles }
120dc40be75SVolkan Keles 
121dc40be75SVolkan Keles bool BlockExtractor::runOnModule(Module &M) {
122dc40be75SVolkan Keles 
123dc40be75SVolkan Keles   bool Changed = false;
124dc40be75SVolkan Keles 
125dc40be75SVolkan Keles   // Get all the functions.
126dc40be75SVolkan Keles   SmallVector<Function *, 4> Functions;
127dc40be75SVolkan Keles   for (Function &F : M) {
128dc40be75SVolkan Keles     splitLandingPadPreds(F);
129dc40be75SVolkan Keles     Functions.push_back(&F);
130dc40be75SVolkan Keles   }
131dc40be75SVolkan Keles 
132dc40be75SVolkan Keles   // Get all the blocks specified in the input file.
133dc40be75SVolkan Keles   for (const auto &BInfo : BlocksByName) {
134dc40be75SVolkan Keles     Function *F = M.getFunction(BInfo.first);
135dc40be75SVolkan Keles     if (!F)
136dc40be75SVolkan Keles       report_fatal_error("Invalid function name specified in the input file");
137dc40be75SVolkan Keles     auto Res = llvm::find_if(*F, [&](const BasicBlock &BB) {
138dc40be75SVolkan Keles       return BB.getName().equals(BInfo.second);
139dc40be75SVolkan Keles     });
140dc40be75SVolkan Keles     if (Res == F->end())
141dc40be75SVolkan Keles       report_fatal_error("Invalid block name specified in the input file");
142dc40be75SVolkan Keles     Blocks.push_back(&*Res);
143dc40be75SVolkan Keles   }
144dc40be75SVolkan Keles 
145dc40be75SVolkan Keles   // Extract basic blocks.
146dc40be75SVolkan Keles   for (BasicBlock *BB : Blocks) {
147dc40be75SVolkan Keles     // Check if the module contains BB.
148dc40be75SVolkan Keles     if (BB->getParent()->getParent() != &M)
149dc40be75SVolkan Keles       report_fatal_error("Invalid basic block");
150dc40be75SVolkan Keles     DEBUG(dbgs() << "BlockExtractor: Extracting " << BB->getParent()->getName()
151dc40be75SVolkan Keles                  << ":" << BB->getName() << "\n");
152dc40be75SVolkan Keles     SmallVector<BasicBlock *, 2> BlocksToExtractVec;
153dc40be75SVolkan Keles     BlocksToExtractVec.push_back(BB);
154dc40be75SVolkan Keles     if (const InvokeInst *II = dyn_cast<InvokeInst>(BB->getTerminator()))
155dc40be75SVolkan Keles       BlocksToExtractVec.push_back(II->getUnwindDest());
156dc40be75SVolkan Keles     CodeExtractor(BlocksToExtractVec).extractCodeRegion();
157dc40be75SVolkan Keles     ++NumExtracted;
158dc40be75SVolkan Keles     Changed = true;
159dc40be75SVolkan Keles   }
160dc40be75SVolkan Keles 
161dc40be75SVolkan Keles   // Erase the functions.
162dc40be75SVolkan Keles   if (EraseFunctions || BlockExtractorEraseFuncs) {
163dc40be75SVolkan Keles     for (Function *F : Functions) {
164dc40be75SVolkan Keles       DEBUG(dbgs() << "BlockExtractor: Deleting " << F->getName() << "\n");
165dc40be75SVolkan Keles       F->eraseFromParent();
166dc40be75SVolkan Keles     }
167dc40be75SVolkan Keles     // Set linkage as ExternalLinkage to avoid erasing unreachable functions.
168dc40be75SVolkan Keles     for (Function &F : M)
169dc40be75SVolkan Keles       F.setLinkage(GlobalValue::ExternalLinkage);
170dc40be75SVolkan Keles     Changed = true;
171dc40be75SVolkan Keles   }
172dc40be75SVolkan Keles 
173dc40be75SVolkan Keles   return Changed;
174dc40be75SVolkan Keles }
175