Internalize.cpp revision 5de97242323586a5e4399a26f69831bb7edc2156 
1//===-- Internalize.cpp - Mark functions internal -------------------------===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This pass loops over all of the functions in the input module, looking for a
11// main function.  If a main function is found, all other functions and all
12// global variables with initializers are marked as internal.
13//
14//===----------------------------------------------------------------------===//
15
16#define DEBUG_TYPE "internalize"
17#include "llvm/Analysis/CallGraph.h"
18#include "llvm/Transforms/IPO.h"
19#include "llvm/Pass.h"
20#include "llvm/Module.h"
21#include "llvm/Support/CommandLine.h"
22#include "llvm/Support/Compiler.h"
23#include "llvm/Support/Debug.h"
24#include "llvm/ADT/Statistic.h"
25#include <fstream>
26#include <set>
27using namespace llvm;
28
29STATISTIC(NumFunctions, "Number of functions internalized");
30STATISTIC(NumGlobals  , "Number of global vars internalized");
31
32// APIFile - A file which contains a list of symbols that should not be marked
33// external.
34static cl::opt<std::string>
35APIFile("internalize-public-api-file", cl::value_desc("filename"),
36        cl::desc("A file containing list of symbol names to preserve"));
37
38// APIList - A list of symbols that should not be marked internal.
39static cl::list<std::string>
40APIList("internalize-public-api-list", cl::value_desc("list"),
41        cl::desc("A list of symbol names to preserve"),
42        cl::CommaSeparated);
43
44namespace {
45  class VISIBILITY_HIDDEN InternalizePass : public ModulePass {
46    std::set<std::string> ExternalNames;
47    /// If no api symbols were specified and a main function is defined,
48    /// assume the main function is the only API
49    bool AllButMain;
50  public:
51    static char ID; // Pass identification, replacement for typeid
52    explicit InternalizePass(bool AllButMain = true);
53    explicit InternalizePass(const std::vector <const char *>& exportList);
54    void LoadFile(const char *Filename);
55    virtual bool runOnModule(Module &M);
56
57    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
58      AU.setPreservesCFG();
59      AU.addPreserved<CallGraph>();
60    }
61  };
62} // end anonymous namespace
63
64char InternalizePass::ID = 0;
65static RegisterPass<InternalizePass>
66X("internalize", "Internalize Global Symbols");
67
68InternalizePass::InternalizePass(bool AllButMain)
69  : ModulePass(&ID), AllButMain(AllButMain){
70  if (!APIFile.empty())           // If a filename is specified, use it.
71    LoadFile(APIFile.c_str());
72  if (!APIList.empty())           // If a list is specified, use it as well.
73    ExternalNames.insert(APIList.begin(), APIList.end());
74}
75
76InternalizePass::InternalizePass(const std::vector<const char *>&exportList)
77  : ModulePass(&ID), AllButMain(false){
78  for(std::vector<const char *>::const_iterator itr = exportList.begin();
79        itr != exportList.end(); itr++) {
80    ExternalNames.insert(*itr);
81  }
82}
83
84void InternalizePass::LoadFile(const char *Filename) {
85  // Load the APIFile...
86  std::ifstream In(Filename);
87  if (!In.good()) {
88    cerr << "WARNING: Internalize couldn't load file '" << Filename
89         << "'! Continuing as if it's empty.\n";
90    return; // Just continue as if the file were empty
91  }
92  while (In) {
93    std::string Symbol;
94    In >> Symbol;
95    if (!Symbol.empty())
96      ExternalNames.insert(Symbol);
97  }
98}
99
100bool InternalizePass::runOnModule(Module &M) {
101  if (ExternalNames.empty()) {
102    // Return if we're not in 'all but main' mode and have no external api
103    if (!AllButMain)
104      return false;
105    // If no list or file of symbols was specified, check to see if there is a
106    // "main" symbol defined in the module.  If so, use it, otherwise do not
107    // internalize the module, it must be a library or something.
108    //
109    Function *MainFunc = M.getFunction("main");
110    if (MainFunc == 0 || MainFunc->isDeclaration())
111      return false;  // No main found, must be a library...
112
113    // Preserve main, internalize all else.
114    ExternalNames.insert(MainFunc->getName());
115  }
116
117  bool Changed = false;
118
119  // Mark all functions not in the api as internal.
120  for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I)
121    if (!I->isDeclaration() &&         // Function must be defined here
122        !I->hasInternalLinkage() &&  // Can't already have internal linkage
123        !ExternalNames.count(I->getName())) {// Not marked to keep external?
124      I->setLinkage(GlobalValue::InternalLinkage);
125      Changed = true;
126      ++NumFunctions;
127      DOUT << "Internalizing func " << I->getName() << "\n";
128    }
129
130  // Never internalize the llvm.used symbol.  It is used to implement
131  // attribute((used)).
132  ExternalNames.insert("llvm.used");
133
134  // Never internalize anchors used by the machine module info, else the info
135  // won't find them.  (see MachineModuleInfo.)
136  ExternalNames.insert("llvm.dbg.compile_units");
137  ExternalNames.insert("llvm.dbg.global_variables");
138  ExternalNames.insert("llvm.dbg.subprograms");
139  ExternalNames.insert("llvm.global_ctors");
140  ExternalNames.insert("llvm.global_dtors");
141  ExternalNames.insert("llvm.noinline");
142  ExternalNames.insert("llvm.global.annotations");
143
144  // Mark all global variables with initializers that are not in the api as
145  // internal as well.
146  for (Module::global_iterator I = M.global_begin(), E = M.global_end();
147       I != E; ++I)
148    if (!I->isDeclaration() && !I->hasInternalLinkage() &&
149        !ExternalNames.count(I->getName())) {
150      I->setLinkage(GlobalValue::InternalLinkage);
151      Changed = true;
152      ++NumGlobals;
153      DOUT << "Internalized gvar " << I->getName() << "\n";
154    }
155
156  return Changed;
157}
158
159ModulePass *llvm::createInternalizePass(bool AllButMain) {
160  return new InternalizePass(AllButMain);
161}
162
163ModulePass *llvm::createInternalizePass(const std::vector <const char *> &el) {
164  return new InternalizePass(el);
165}
166