1f22ef01cSRoman Divacky //===-- Internalize.cpp - Mark functions internal -------------------------===//
2f22ef01cSRoman Divacky //
3f22ef01cSRoman Divacky //                     The LLVM Compiler Infrastructure
4f22ef01cSRoman Divacky //
5f22ef01cSRoman Divacky // This file is distributed under the University of Illinois Open Source
6f22ef01cSRoman Divacky // License. See LICENSE.TXT for details.
7f22ef01cSRoman Divacky //
8f22ef01cSRoman Divacky //===----------------------------------------------------------------------===//
9f22ef01cSRoman Divacky //
10f22ef01cSRoman Divacky // This pass loops over all of the functions in the input module, looking for a
11f22ef01cSRoman Divacky // main function.  If a main function is found, all other functions and all
12f22ef01cSRoman Divacky // global variables with initializers are marked as internal.
13f22ef01cSRoman Divacky //
14f22ef01cSRoman Divacky //===----------------------------------------------------------------------===//
15f22ef01cSRoman Divacky 
16f22ef01cSRoman Divacky #define DEBUG_TYPE "internalize"
17f22ef01cSRoman Divacky #include "llvm/Analysis/CallGraph.h"
18f22ef01cSRoman Divacky #include "llvm/Transforms/IPO.h"
19f22ef01cSRoman Divacky #include "llvm/Pass.h"
20f22ef01cSRoman Divacky #include "llvm/Module.h"
21f22ef01cSRoman Divacky #include "llvm/Support/CommandLine.h"
22f22ef01cSRoman Divacky #include "llvm/Support/Debug.h"
23f22ef01cSRoman Divacky #include "llvm/Support/raw_ostream.h"
24f22ef01cSRoman Divacky #include "llvm/ADT/Statistic.h"
25f22ef01cSRoman Divacky #include <fstream>
26f22ef01cSRoman Divacky #include <set>
27f22ef01cSRoman Divacky using namespace llvm;
28f22ef01cSRoman Divacky 
29f22ef01cSRoman Divacky STATISTIC(NumAliases  , "Number of aliases internalized");
30f22ef01cSRoman Divacky STATISTIC(NumFunctions, "Number of functions internalized");
31f22ef01cSRoman Divacky STATISTIC(NumGlobals  , "Number of global vars internalized");
32f22ef01cSRoman Divacky 
33f22ef01cSRoman Divacky // APIFile - A file which contains a list of symbols that should not be marked
34f22ef01cSRoman Divacky // external.
35f22ef01cSRoman Divacky static cl::opt<std::string>
36f22ef01cSRoman Divacky APIFile("internalize-public-api-file", cl::value_desc("filename"),
37f22ef01cSRoman Divacky         cl::desc("A file containing list of symbol names to preserve"));
38f22ef01cSRoman Divacky 
39f22ef01cSRoman Divacky // APIList - A list of symbols that should not be marked internal.
40f22ef01cSRoman Divacky static cl::list<std::string>
41f22ef01cSRoman Divacky APIList("internalize-public-api-list", cl::value_desc("list"),
42f22ef01cSRoman Divacky         cl::desc("A list of symbol names to preserve"),
43f22ef01cSRoman Divacky         cl::CommaSeparated);
44f22ef01cSRoman Divacky 
45f22ef01cSRoman Divacky namespace {
46f22ef01cSRoman Divacky   class InternalizePass : public ModulePass {
47f22ef01cSRoman Divacky     std::set<std::string> ExternalNames;
48f22ef01cSRoman Divacky     /// If no api symbols were specified and a main function is defined,
49f22ef01cSRoman Divacky     /// assume the main function is the only API
50f22ef01cSRoman Divacky     bool AllButMain;
51f22ef01cSRoman Divacky   public:
52f22ef01cSRoman Divacky     static char ID; // Pass identification, replacement for typeid
53f22ef01cSRoman Divacky     explicit InternalizePass(bool AllButMain = true);
54f22ef01cSRoman Divacky     explicit InternalizePass(const std::vector <const char *>& exportList);
55f22ef01cSRoman Divacky     void LoadFile(const char *Filename);
56f22ef01cSRoman Divacky     virtual bool runOnModule(Module &M);
57f22ef01cSRoman Divacky 
58f22ef01cSRoman Divacky     virtual void getAnalysisUsage(AnalysisUsage &AU) const {
59f22ef01cSRoman Divacky       AU.setPreservesCFG();
60f22ef01cSRoman Divacky       AU.addPreserved<CallGraph>();
61f22ef01cSRoman Divacky     }
62f22ef01cSRoman Divacky   };
63f22ef01cSRoman Divacky } // end anonymous namespace
64f22ef01cSRoman Divacky 
65f22ef01cSRoman Divacky char InternalizePass::ID = 0;
66e580952dSDimitry Andric INITIALIZE_PASS(InternalizePass, "internalize",
672754fe60SDimitry Andric                 "Internalize Global Symbols", false, false)
68f22ef01cSRoman Divacky 
69f22ef01cSRoman Divacky InternalizePass::InternalizePass(bool AllButMain)
70e580952dSDimitry Andric   : ModulePass(ID), AllButMain(AllButMain){
712754fe60SDimitry Andric   initializeInternalizePassPass(*PassRegistry::getPassRegistry());
72f22ef01cSRoman Divacky   if (!APIFile.empty())           // If a filename is specified, use it.
73f22ef01cSRoman Divacky     LoadFile(APIFile.c_str());
74f22ef01cSRoman Divacky   if (!APIList.empty())           // If a list is specified, use it as well.
75f22ef01cSRoman Divacky     ExternalNames.insert(APIList.begin(), APIList.end());
76f22ef01cSRoman Divacky }
77f22ef01cSRoman Divacky 
78f22ef01cSRoman Divacky InternalizePass::InternalizePass(const std::vector<const char *>&exportList)
79e580952dSDimitry Andric   : ModulePass(ID), AllButMain(false){
802754fe60SDimitry Andric   initializeInternalizePassPass(*PassRegistry::getPassRegistry());
81f22ef01cSRoman Divacky   for(std::vector<const char *>::const_iterator itr = exportList.begin();
82f22ef01cSRoman Divacky         itr != exportList.end(); itr++) {
83f22ef01cSRoman Divacky     ExternalNames.insert(*itr);
84f22ef01cSRoman Divacky   }
85f22ef01cSRoman Divacky }
86f22ef01cSRoman Divacky 
87f22ef01cSRoman Divacky void InternalizePass::LoadFile(const char *Filename) {
88f22ef01cSRoman Divacky   // Load the APIFile...
89f22ef01cSRoman Divacky   std::ifstream In(Filename);
90f22ef01cSRoman Divacky   if (!In.good()) {
91f22ef01cSRoman Divacky     errs() << "WARNING: Internalize couldn't load file '" << Filename
92f22ef01cSRoman Divacky          << "'! Continuing as if it's empty.\n";
93f22ef01cSRoman Divacky     return; // Just continue as if the file were empty
94f22ef01cSRoman Divacky   }
95f22ef01cSRoman Divacky   while (In) {
96f22ef01cSRoman Divacky     std::string Symbol;
97f22ef01cSRoman Divacky     In >> Symbol;
98f22ef01cSRoman Divacky     if (!Symbol.empty())
99f22ef01cSRoman Divacky       ExternalNames.insert(Symbol);
100f22ef01cSRoman Divacky   }
101f22ef01cSRoman Divacky }
102f22ef01cSRoman Divacky 
103f22ef01cSRoman Divacky bool InternalizePass::runOnModule(Module &M) {
104f22ef01cSRoman Divacky   CallGraph *CG = getAnalysisIfAvailable<CallGraph>();
105f22ef01cSRoman Divacky   CallGraphNode *ExternalNode = CG ? CG->getExternalCallingNode() : 0;
106f22ef01cSRoman Divacky 
107f22ef01cSRoman Divacky   if (ExternalNames.empty()) {
108f22ef01cSRoman Divacky     // Return if we're not in 'all but main' mode and have no external api
109f22ef01cSRoman Divacky     if (!AllButMain)
110f22ef01cSRoman Divacky       return false;
111f22ef01cSRoman Divacky     // If no list or file of symbols was specified, check to see if there is a
112f22ef01cSRoman Divacky     // "main" symbol defined in the module.  If so, use it, otherwise do not
113f22ef01cSRoman Divacky     // internalize the module, it must be a library or something.
114f22ef01cSRoman Divacky     //
115f22ef01cSRoman Divacky     Function *MainFunc = M.getFunction("main");
116f22ef01cSRoman Divacky     if (MainFunc == 0 || MainFunc->isDeclaration())
117f22ef01cSRoman Divacky       return false;  // No main found, must be a library...
118f22ef01cSRoman Divacky 
119f22ef01cSRoman Divacky     // Preserve main, internalize all else.
120f22ef01cSRoman Divacky     ExternalNames.insert(MainFunc->getName());
121f22ef01cSRoman Divacky   }
122f22ef01cSRoman Divacky 
123f22ef01cSRoman Divacky   bool Changed = false;
124f22ef01cSRoman Divacky 
125dff0c46cSDimitry Andric   // Never internalize functions which code-gen might insert.
126cb4dff85SDimitry Andric   // FIXME: We should probably add this (and the __stack_chk_guard) via some
127cb4dff85SDimitry Andric   // type of call-back in CodeGen.
128dff0c46cSDimitry Andric   ExternalNames.insert("__stack_chk_fail");
129dff0c46cSDimitry Andric 
130f22ef01cSRoman Divacky   // Mark all functions not in the api as internal.
131f22ef01cSRoman Divacky   // FIXME: maybe use private linkage?
132f22ef01cSRoman Divacky   for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I)
133f22ef01cSRoman Divacky     if (!I->isDeclaration() &&         // Function must be defined here
1343b0f4066SDimitry Andric         // Available externally is really just a "declaration with a body".
1353b0f4066SDimitry Andric         !I->hasAvailableExternallyLinkage() &&
136f22ef01cSRoman Divacky         !I->hasLocalLinkage() &&  // Can't already have internal linkage
137f22ef01cSRoman Divacky         !ExternalNames.count(I->getName())) {// Not marked to keep external?
138f22ef01cSRoman Divacky       I->setLinkage(GlobalValue::InternalLinkage);
139f22ef01cSRoman Divacky       // Remove a callgraph edge from the external node to this function.
140f22ef01cSRoman Divacky       if (ExternalNode) ExternalNode->removeOneAbstractEdgeTo((*CG)[I]);
141f22ef01cSRoman Divacky       Changed = true;
142f22ef01cSRoman Divacky       ++NumFunctions;
143f22ef01cSRoman Divacky       DEBUG(dbgs() << "Internalizing func " << I->getName() << "\n");
144f22ef01cSRoman Divacky     }
145f22ef01cSRoman Divacky 
146f22ef01cSRoman Divacky   // Never internalize the llvm.used symbol.  It is used to implement
147f22ef01cSRoman Divacky   // attribute((used)).
148f22ef01cSRoman Divacky   // FIXME: Shouldn't this just filter on llvm.metadata section??
149f22ef01cSRoman Divacky   ExternalNames.insert("llvm.used");
150f22ef01cSRoman Divacky   ExternalNames.insert("llvm.compiler.used");
151f22ef01cSRoman Divacky 
152f22ef01cSRoman Divacky   // Never internalize anchors used by the machine module info, else the info
153f22ef01cSRoman Divacky   // won't find them.  (see MachineModuleInfo.)
154f22ef01cSRoman Divacky   ExternalNames.insert("llvm.global_ctors");
155f22ef01cSRoman Divacky   ExternalNames.insert("llvm.global_dtors");
156f22ef01cSRoman Divacky   ExternalNames.insert("llvm.global.annotations");
157f22ef01cSRoman Divacky 
158dff0c46cSDimitry Andric   // Never internalize symbols code-gen inserts.
159dff0c46cSDimitry Andric   ExternalNames.insert("__stack_chk_guard");
160dff0c46cSDimitry Andric 
161f22ef01cSRoman Divacky   // Mark all global variables with initializers that are not in the api as
162f22ef01cSRoman Divacky   // internal as well.
163f22ef01cSRoman Divacky   // FIXME: maybe use private linkage?
164f22ef01cSRoman Divacky   for (Module::global_iterator I = M.global_begin(), E = M.global_end();
165f22ef01cSRoman Divacky        I != E; ++I)
166f22ef01cSRoman Divacky     if (!I->isDeclaration() && !I->hasLocalLinkage() &&
167f22ef01cSRoman Divacky         // Available externally is really just a "declaration with a body".
168f22ef01cSRoman Divacky         !I->hasAvailableExternallyLinkage() &&
169f22ef01cSRoman Divacky         !ExternalNames.count(I->getName())) {
170f22ef01cSRoman Divacky       I->setLinkage(GlobalValue::InternalLinkage);
171f22ef01cSRoman Divacky       Changed = true;
172f22ef01cSRoman Divacky       ++NumGlobals;
173f22ef01cSRoman Divacky       DEBUG(dbgs() << "Internalized gvar " << I->getName() << "\n");
174f22ef01cSRoman Divacky     }
175f22ef01cSRoman Divacky 
176f22ef01cSRoman Divacky   // Mark all aliases that are not in the api as internal as well.
177f22ef01cSRoman Divacky   for (Module::alias_iterator I = M.alias_begin(), E = M.alias_end();
178f22ef01cSRoman Divacky        I != E; ++I)
179f22ef01cSRoman Divacky     if (!I->isDeclaration() && !I->hasInternalLinkage() &&
180f22ef01cSRoman Divacky         // Available externally is really just a "declaration with a body".
181f22ef01cSRoman Divacky         !I->hasAvailableExternallyLinkage() &&
182f22ef01cSRoman Divacky         !ExternalNames.count(I->getName())) {
183f22ef01cSRoman Divacky       I->setLinkage(GlobalValue::InternalLinkage);
184f22ef01cSRoman Divacky       Changed = true;
185f22ef01cSRoman Divacky       ++NumAliases;
186f22ef01cSRoman Divacky       DEBUG(dbgs() << "Internalized alias " << I->getName() << "\n");
187f22ef01cSRoman Divacky     }
188f22ef01cSRoman Divacky 
189f22ef01cSRoman Divacky   return Changed;
190f22ef01cSRoman Divacky }
191f22ef01cSRoman Divacky 
192f22ef01cSRoman Divacky ModulePass *llvm::createInternalizePass(bool AllButMain) {
193f22ef01cSRoman Divacky   return new InternalizePass(AllButMain);
194f22ef01cSRoman Divacky }
195f22ef01cSRoman Divacky 
196f22ef01cSRoman Divacky ModulePass *llvm::createInternalizePass(const std::vector <const char *> &el) {
197f22ef01cSRoman Divacky   return new InternalizePass(el);
198f22ef01cSRoman Divacky }
199