StripSymbols.cpp revision a168fc98dedfc8cac01c34f84b699fe5f48ad76d 
1//===- StripSymbols.cpp - Strip symbols and debug info from a module ------===//
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// The StripSymbols transformation implements code stripping. Specifically, it
11// can delete:
12//
13//   * names for virtual registers
14//   * symbols for internal globals and functions
15//   * debug information
16//
17// Note that this transformation makes code much less readable, so it should
18// only be used in situations where the 'strip' utility would be used, such as
19// reducing code size or making it harder to reverse engineer code.
20//
21//===----------------------------------------------------------------------===//
22
23#include "llvm/Transforms/IPO.h"
24#include "llvm/Constants.h"
25#include "llvm/DerivedTypes.h"
26#include "llvm/Instructions.h"
27#include "llvm/Module.h"
28#include "llvm/Pass.h"
29#include "llvm/ValueSymbolTable.h"
30#include "llvm/TypeSymbolTable.h"
31#include "llvm/Support/Compiler.h"
32#include "llvm/ADT/SmallPtrSet.h"
33using namespace llvm;
34
35namespace {
36  class VISIBILITY_HIDDEN StripSymbols : public ModulePass {
37    bool OnlyDebugInfo;
38  public:
39    static char ID; // Pass identification, replacement for typeid
40    explicit StripSymbols(bool ODI = false)
41      : ModulePass(&ID), OnlyDebugInfo(ODI) {}
42
43    virtual bool runOnModule(Module &M);
44
45    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
46      AU.setPreservesAll();
47    }
48  };
49
50  class VISIBILITY_HIDDEN StripNonDebugSymbols : public ModulePass {
51  public:
52    static char ID; // Pass identification, replacement for typeid
53    explicit StripNonDebugSymbols()
54      : ModulePass(&ID) {}
55
56    virtual bool runOnModule(Module &M);
57
58    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
59      AU.setPreservesAll();
60    }
61  };
62}
63
64char StripSymbols::ID = 0;
65static RegisterPass<StripSymbols>
66X("strip", "Strip all symbols from a module");
67
68ModulePass *llvm::createStripSymbolsPass(bool OnlyDebugInfo) {
69  return new StripSymbols(OnlyDebugInfo);
70}
71
72char StripNonDebugSymbols::ID = 0;
73static RegisterPass<StripNonDebugSymbols>
74Y("strip-nondebug", "Strip all symbols, except dbg symbols, from a module");
75
76ModulePass *llvm::createStripNonDebugSymbolsPass() {
77  return new StripNonDebugSymbols();
78}
79
80/// OnlyUsedBy - Return true if V is only used by Usr.
81static bool OnlyUsedBy(Value *V, Value *Usr) {
82  for(Value::use_iterator I = V->use_begin(), E = V->use_end(); I != E; ++I) {
83    User *U = *I;
84    if (U != Usr)
85      return false;
86  }
87  return true;
88}
89
90static void RemoveDeadConstant(Constant *C) {
91  assert(C->use_empty() && "Constant is not dead!");
92  SmallPtrSet<Constant *, 4> Operands;
93  for (unsigned i = 0, e = C->getNumOperands(); i != e; ++i)
94    if (isa<DerivedType>(C->getOperand(i)->getType()) &&
95        OnlyUsedBy(C->getOperand(i), C))
96      Operands.insert(C->getOperand(i));
97  if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
98    if (!GV->hasLocalLinkage()) return;   // Don't delete non static globals.
99    GV->eraseFromParent();
100  }
101  else if (!isa<Function>(C))
102    if (isa<CompositeType>(C->getType()))
103      C->destroyConstant();
104
105  // If the constant referenced anything, see if we can delete it as well.
106  for (SmallPtrSet<Constant *, 4>::iterator OI = Operands.begin(),
107         OE = Operands.end(); OI != OE; ++OI)
108    RemoveDeadConstant(*OI);
109}
110
111// Strip the symbol table of its names.
112//
113static void StripSymtab(ValueSymbolTable &ST, bool PreserveDbgInfo) {
114  for (ValueSymbolTable::iterator VI = ST.begin(), VE = ST.end(); VI != VE; ) {
115    Value *V = VI->getValue();
116    ++VI;
117    if (!isa<GlobalValue>(V) || cast<GlobalValue>(V)->hasLocalLinkage()) {
118      if (!PreserveDbgInfo || strncmp(V->getNameStart(), "llvm.dbg", 8))
119        // Set name to "", removing from symbol table!
120        V->setName("");
121    }
122  }
123}
124
125// Strip the symbol table of its names.
126static void StripTypeSymtab(TypeSymbolTable &ST, bool PreserveDbgInfo) {
127  for (TypeSymbolTable::iterator TI = ST.begin(), E = ST.end(); TI != E; ) {
128    if (PreserveDbgInfo && strncmp(TI->first.c_str(), "llvm.dbg", 8) == 0)
129      ++TI;
130    else
131      ST.remove(TI++);
132  }
133}
134
135/// Find values that are marked as llvm.used.
136void findUsedValues(Module &M,
137                    SmallPtrSet<const GlobalValue*, 8>& llvmUsedValues) {
138  if (GlobalVariable *LLVMUsed = M.getGlobalVariable("llvm.used")) {
139    llvmUsedValues.insert(LLVMUsed);
140    // Collect values that are preserved as per explicit request.
141    // llvm.used is used to list these values.
142    if (ConstantArray *Inits =
143        dyn_cast<ConstantArray>(LLVMUsed->getInitializer())) {
144      for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i) {
145        if (GlobalValue *GV = dyn_cast<GlobalValue>(Inits->getOperand(i)))
146          llvmUsedValues.insert(GV);
147        else if (ConstantExpr *CE =
148                 dyn_cast<ConstantExpr>(Inits->getOperand(i)))
149          if (CE->getOpcode() == Instruction::BitCast)
150            if (GlobalValue *GV = dyn_cast<GlobalValue>(CE->getOperand(0)))
151              llvmUsedValues.insert(GV);
152      }
153    }
154  }
155}
156
157/// StripSymbolNames - Strip symbol names.
158bool StripSymbolNames(Module &M, bool PreserveDbgInfo) {
159
160  SmallPtrSet<const GlobalValue*, 8> llvmUsedValues;
161  findUsedValues(M, llvmUsedValues);
162
163  for (Module::global_iterator I = M.global_begin(), E = M.global_end();
164       I != E; ++I) {
165    if (I->hasLocalLinkage() && llvmUsedValues.count(I) == 0)
166      if (!PreserveDbgInfo || strncmp(I->getNameStart(), "llvm.dbg", 8))
167        I->setName("");     // Internal symbols can't participate in linkage
168  }
169
170  for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I) {
171    if (I->hasLocalLinkage() && llvmUsedValues.count(I) == 0)
172      if (!PreserveDbgInfo || strncmp(I->getNameStart(), "llvm.dbg", 8))
173        I->setName("");     // Internal symbols can't participate in linkage
174    StripSymtab(I->getValueSymbolTable(), PreserveDbgInfo);
175  }
176
177  // Remove all names from types.
178  StripTypeSymtab(M.getTypeSymbolTable(), PreserveDbgInfo);
179
180  return true;
181}
182
183// StripDebugInfo - Strip debug info in the module if it exists.
184// To do this, we remove llvm.dbg.func.start, llvm.dbg.stoppoint, and
185// llvm.dbg.region.end calls, and any globals they point to if now dead.
186bool StripDebugInfo(Module &M) {
187
188  Function *FuncStart = M.getFunction("llvm.dbg.func.start");
189  Function *StopPoint = M.getFunction("llvm.dbg.stoppoint");
190  Function *RegionStart = M.getFunction("llvm.dbg.region.start");
191  Function *RegionEnd = M.getFunction("llvm.dbg.region.end");
192  Function *Declare = M.getFunction("llvm.dbg.declare");
193
194  std::vector<Constant*> DeadConstants;
195
196  // Remove all of the calls to the debugger intrinsics, and remove them from
197  // the module.
198  if (FuncStart) {
199    while (!FuncStart->use_empty()) {
200      CallInst *CI = cast<CallInst>(FuncStart->use_back());
201      Value *Arg = CI->getOperand(1);
202      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
203      CI->eraseFromParent();
204      if (Arg->use_empty())
205        if (Constant *C = dyn_cast<Constant>(Arg))
206          DeadConstants.push_back(C);
207    }
208    FuncStart->eraseFromParent();
209  }
210  if (StopPoint) {
211    while (!StopPoint->use_empty()) {
212      CallInst *CI = cast<CallInst>(StopPoint->use_back());
213      Value *Arg = CI->getOperand(3);
214      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
215      CI->eraseFromParent();
216      if (Arg->use_empty())
217        if (Constant *C = dyn_cast<Constant>(Arg))
218          DeadConstants.push_back(C);
219    }
220    StopPoint->eraseFromParent();
221  }
222  if (RegionStart) {
223    while (!RegionStart->use_empty()) {
224      CallInst *CI = cast<CallInst>(RegionStart->use_back());
225      Value *Arg = CI->getOperand(1);
226      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
227      CI->eraseFromParent();
228      if (Arg->use_empty())
229        if (Constant *C = dyn_cast<Constant>(Arg))
230          DeadConstants.push_back(C);
231    }
232    RegionStart->eraseFromParent();
233  }
234  if (RegionEnd) {
235    while (!RegionEnd->use_empty()) {
236      CallInst *CI = cast<CallInst>(RegionEnd->use_back());
237      Value *Arg = CI->getOperand(1);
238      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
239      CI->eraseFromParent();
240      if (Arg->use_empty())
241        if (Constant *C = dyn_cast<Constant>(Arg))
242          DeadConstants.push_back(C);
243    }
244    RegionEnd->eraseFromParent();
245  }
246  if (Declare) {
247    while (!Declare->use_empty()) {
248      CallInst *CI = cast<CallInst>(Declare->use_back());
249      Value *Arg1 = CI->getOperand(1);
250      Value *Arg2 = CI->getOperand(2);
251      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
252      CI->eraseFromParent();
253      if (Arg1->use_empty()) {
254        if (Constant *C = dyn_cast<Constant>(Arg1))
255          DeadConstants.push_back(C);
256        if (Instruction *I = dyn_cast<Instruction>(Arg1))
257          I->eraseFromParent();
258      }
259      if (Arg2->use_empty())
260        if (Constant *C = dyn_cast<Constant>(Arg2))
261          DeadConstants.push_back(C);
262    }
263    Declare->eraseFromParent();
264  }
265
266  SmallPtrSet<const GlobalValue*, 8> llvmUsedValues;
267  findUsedValues(M, llvmUsedValues);
268
269  // llvm.dbg.compile_units and llvm.dbg.subprograms are marked as linkonce
270  // but since we are removing all debug information, make them internal now.
271  // FIXME: Use private linkage maybe?
272  if (Constant *C = M.getNamedGlobal("llvm.dbg.compile_units"))
273    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
274      GV->setLinkage(GlobalValue::InternalLinkage);
275
276  if (Constant *C = M.getNamedGlobal("llvm.dbg.subprograms"))
277    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
278      GV->setLinkage(GlobalValue::InternalLinkage);
279
280  if (Constant *C = M.getNamedGlobal("llvm.dbg.global_variables"))
281    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
282      GV->setLinkage(GlobalValue::InternalLinkage);
283
284  // Delete all dbg variables.
285  for (Module::global_iterator I = M.global_begin(), E = M.global_end();
286       I != E; ++I) {
287    GlobalVariable *GV = dyn_cast<GlobalVariable>(I);
288    if (!GV) continue;
289    if (GV->use_empty() && llvmUsedValues.count(I) == 0
290        && (!GV->hasSection()
291            || strcmp(GV->getSection().c_str(), "llvm.metadata") == 0))
292      DeadConstants.push_back(GV);
293  }
294
295  if (DeadConstants.empty())
296    return false;
297
298  // Delete any internal globals that were only used by the debugger intrinsics.
299  while (!DeadConstants.empty()) {
300    Constant *C = DeadConstants.back();
301    DeadConstants.pop_back();
302    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
303      if (GV->hasLocalLinkage())
304        RemoveDeadConstant(GV);
305    }
306    else
307      RemoveDeadConstant(C);
308  }
309
310  // Remove all llvm.dbg types.
311  TypeSymbolTable &ST = M.getTypeSymbolTable();
312  for (TypeSymbolTable::iterator TI = ST.begin(), TE = ST.end(); TI != TE; ) {
313    if (!strncmp(TI->first.c_str(), "llvm.dbg.", 9))
314      ST.remove(TI++);
315    else
316      ++TI;
317  }
318
319  return true;
320}
321
322bool StripSymbols::runOnModule(Module &M) {
323  bool Changed = false;
324  Changed |= StripDebugInfo(M);
325  if (!OnlyDebugInfo)
326    Changed |= StripSymbolNames(M, false);
327  return Changed;
328}
329
330bool StripNonDebugSymbols::runOnModule(Module &M) {
331  return StripSymbolNames(M, true);
332}
333