StripSymbols.cpp revision 619acdc63ab0a47d125dca0591285c8ac4c9ed20 
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/Analysis/DebugInfo.h"
30#include "llvm/ValueSymbolTable.h"
31#include "llvm/TypeSymbolTable.h"
32#include "llvm/Transforms/Utils/Local.h"
33#include "llvm/ADT/SmallPtrSet.h"
34using namespace llvm;
35
36namespace {
37  class StripSymbols : public ModulePass {
38    bool OnlyDebugInfo;
39  public:
40    static char ID; // Pass identification, replacement for typeid
41    explicit StripSymbols(bool ODI = false)
42      : ModulePass(&ID), OnlyDebugInfo(ODI) {}
43
44    virtual bool runOnModule(Module &M);
45
46    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
47      AU.setPreservesAll();
48    }
49  };
50
51  class StripNonDebugSymbols : public ModulePass {
52  public:
53    static char ID; // Pass identification, replacement for typeid
54    explicit StripNonDebugSymbols()
55      : ModulePass(&ID) {}
56
57    virtual bool runOnModule(Module &M);
58
59    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
60      AU.setPreservesAll();
61    }
62  };
63
64  class StripDebugDeclare : public ModulePass {
65  public:
66    static char ID; // Pass identification, replacement for typeid
67    explicit StripDebugDeclare()
68      : ModulePass(&ID) {}
69
70    virtual bool runOnModule(Module &M);
71
72    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
73      AU.setPreservesAll();
74    }
75  };
76
77  class StripDeadDebugInfo : public ModulePass {
78  public:
79    static char ID; // Pass identification, replacement for typeid
80    explicit StripDeadDebugInfo()
81      : ModulePass(&ID) {}
82
83    virtual bool runOnModule(Module &M);
84
85    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
86      AU.setPreservesAll();
87    }
88  };
89}
90
91char StripSymbols::ID = 0;
92INITIALIZE_PASS(StripSymbols, "strip",
93                "Strip all symbols from a module", false, false);
94
95ModulePass *llvm::createStripSymbolsPass(bool OnlyDebugInfo) {
96  return new StripSymbols(OnlyDebugInfo);
97}
98
99char StripNonDebugSymbols::ID = 0;
100INITIALIZE_PASS(StripNonDebugSymbols, "strip-nondebug",
101                "Strip all symbols, except dbg symbols, from a module",
102                false, false);
103
104ModulePass *llvm::createStripNonDebugSymbolsPass() {
105  return new StripNonDebugSymbols();
106}
107
108char StripDebugDeclare::ID = 0;
109INITIALIZE_PASS(StripDebugDeclare, "strip-debug-declare",
110                "Strip all llvm.dbg.declare intrinsics", false, false);
111
112ModulePass *llvm::createStripDebugDeclarePass() {
113  return new StripDebugDeclare();
114}
115
116char StripDeadDebugInfo::ID = 0;
117INITIALIZE_PASS(StripDeadDebugInfo, "strip-dead-debug-info",
118                "Strip debug info for unused symbols", false, false);
119
120ModulePass *llvm::createStripDeadDebugInfoPass() {
121  return new StripDeadDebugInfo();
122}
123
124/// OnlyUsedBy - Return true if V is only used by Usr.
125static bool OnlyUsedBy(Value *V, Value *Usr) {
126  for(Value::use_iterator I = V->use_begin(), E = V->use_end(); I != E; ++I) {
127    User *U = *I;
128    if (U != Usr)
129      return false;
130  }
131  return true;
132}
133
134static void RemoveDeadConstant(Constant *C) {
135  assert(C->use_empty() && "Constant is not dead!");
136  SmallPtrSet<Constant*, 4> Operands;
137  for (unsigned i = 0, e = C->getNumOperands(); i != e; ++i)
138    if (isa<DerivedType>(C->getOperand(i)->getType()) &&
139        OnlyUsedBy(C->getOperand(i), C))
140      Operands.insert(cast<Constant>(C->getOperand(i)));
141  if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
142    if (!GV->hasLocalLinkage()) return;   // Don't delete non static globals.
143    GV->eraseFromParent();
144  }
145  else if (!isa<Function>(C))
146    if (isa<CompositeType>(C->getType()))
147      C->destroyConstant();
148
149  // If the constant referenced anything, see if we can delete it as well.
150  for (SmallPtrSet<Constant*, 4>::iterator OI = Operands.begin(),
151         OE = Operands.end(); OI != OE; ++OI)
152    RemoveDeadConstant(*OI);
153}
154
155// Strip the symbol table of its names.
156//
157static void StripSymtab(ValueSymbolTable &ST, bool PreserveDbgInfo) {
158  for (ValueSymbolTable::iterator VI = ST.begin(), VE = ST.end(); VI != VE; ) {
159    Value *V = VI->getValue();
160    ++VI;
161    if (!isa<GlobalValue>(V) || cast<GlobalValue>(V)->hasLocalLinkage()) {
162      if (!PreserveDbgInfo || !V->getName().startswith("llvm.dbg"))
163        // Set name to "", removing from symbol table!
164        V->setName("");
165    }
166  }
167}
168
169// Strip the symbol table of its names.
170static void StripTypeSymtab(TypeSymbolTable &ST, bool PreserveDbgInfo) {
171  for (TypeSymbolTable::iterator TI = ST.begin(), E = ST.end(); TI != E; ) {
172    if (PreserveDbgInfo && StringRef(TI->first).startswith("llvm.dbg"))
173      ++TI;
174    else
175      ST.remove(TI++);
176  }
177}
178
179/// Find values that are marked as llvm.used.
180static void findUsedValues(GlobalVariable *LLVMUsed,
181                           SmallPtrSet<const GlobalValue*, 8> &UsedValues) {
182  if (LLVMUsed == 0) return;
183  UsedValues.insert(LLVMUsed);
184
185  ConstantArray *Inits = dyn_cast<ConstantArray>(LLVMUsed->getInitializer());
186  if (Inits == 0) return;
187
188  for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i)
189    if (GlobalValue *GV =
190          dyn_cast<GlobalValue>(Inits->getOperand(i)->stripPointerCasts()))
191      UsedValues.insert(GV);
192}
193
194/// StripSymbolNames - Strip symbol names.
195static bool StripSymbolNames(Module &M, bool PreserveDbgInfo) {
196
197  SmallPtrSet<const GlobalValue*, 8> llvmUsedValues;
198  findUsedValues(M.getGlobalVariable("llvm.used"), llvmUsedValues);
199  findUsedValues(M.getGlobalVariable("llvm.compiler.used"), llvmUsedValues);
200
201  for (Module::global_iterator I = M.global_begin(), E = M.global_end();
202       I != E; ++I) {
203    if (I->hasLocalLinkage() && llvmUsedValues.count(I) == 0)
204      if (!PreserveDbgInfo || !I->getName().startswith("llvm.dbg"))
205        I->setName("");     // Internal symbols can't participate in linkage
206  }
207
208  for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I) {
209    if (I->hasLocalLinkage() && llvmUsedValues.count(I) == 0)
210      if (!PreserveDbgInfo || !I->getName().startswith("llvm.dbg"))
211        I->setName("");     // Internal symbols can't participate in linkage
212    StripSymtab(I->getValueSymbolTable(), PreserveDbgInfo);
213  }
214
215  // Remove all names from types.
216  StripTypeSymtab(M.getTypeSymbolTable(), PreserveDbgInfo);
217
218  return true;
219}
220
221// StripDebugInfo - Strip debug info in the module if it exists.
222// To do this, we remove llvm.dbg.func.start, llvm.dbg.stoppoint, and
223// llvm.dbg.region.end calls, and any globals they point to if now dead.
224static bool StripDebugInfo(Module &M) {
225
226  bool Changed = false;
227
228  // Remove all of the calls to the debugger intrinsics, and remove them from
229  // the module.
230  if (Function *Declare = M.getFunction("llvm.dbg.declare")) {
231    while (!Declare->use_empty()) {
232      CallInst *CI = cast<CallInst>(Declare->use_back());
233      CI->eraseFromParent();
234    }
235    Declare->eraseFromParent();
236    Changed = true;
237  }
238
239  if (Function *DbgVal = M.getFunction("llvm.dbg.value")) {
240    while (!DbgVal->use_empty()) {
241      CallInst *CI = cast<CallInst>(DbgVal->use_back());
242      CI->eraseFromParent();
243    }
244    DbgVal->eraseFromParent();
245    Changed = true;
246  }
247
248  for (Module::named_metadata_iterator NMI = M.named_metadata_begin(),
249         NME = M.named_metadata_end(); NMI != NME;) {
250    NamedMDNode *NMD = NMI;
251    ++NMI;
252    if (NMD->getName().startswith("llvm.dbg.")) {
253      NMD->eraseFromParent();
254      Changed = true;
255    }
256  }
257
258  for (Module::iterator MI = M.begin(), ME = M.end(); MI != ME; ++MI)
259    for (Function::iterator FI = MI->begin(), FE = MI->end(); FI != FE;
260         ++FI)
261      for (BasicBlock::iterator BI = FI->begin(), BE = FI->end(); BI != BE;
262           ++BI) {
263        if (!BI->getDebugLoc().isUnknown()) {
264          Changed = true;
265          BI->setDebugLoc(DebugLoc());
266        }
267      }
268
269  return Changed;
270}
271
272bool StripSymbols::runOnModule(Module &M) {
273  bool Changed = false;
274  Changed |= StripDebugInfo(M);
275  if (!OnlyDebugInfo)
276    Changed |= StripSymbolNames(M, false);
277  return Changed;
278}
279
280bool StripNonDebugSymbols::runOnModule(Module &M) {
281  return StripSymbolNames(M, true);
282}
283
284bool StripDebugDeclare::runOnModule(Module &M) {
285
286  Function *Declare = M.getFunction("llvm.dbg.declare");
287  std::vector<Constant*> DeadConstants;
288
289  if (Declare) {
290    while (!Declare->use_empty()) {
291      CallInst *CI = cast<CallInst>(Declare->use_back());
292      Value *Arg1 = CI->getArgOperand(0);
293      Value *Arg2 = CI->getArgOperand(1);
294      assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
295      CI->eraseFromParent();
296      if (Arg1->use_empty()) {
297        if (Constant *C = dyn_cast<Constant>(Arg1))
298          DeadConstants.push_back(C);
299        else
300          RecursivelyDeleteTriviallyDeadInstructions(Arg1);
301      }
302      if (Arg2->use_empty())
303        if (Constant *C = dyn_cast<Constant>(Arg2))
304          DeadConstants.push_back(C);
305    }
306    Declare->eraseFromParent();
307  }
308
309  while (!DeadConstants.empty()) {
310    Constant *C = DeadConstants.back();
311    DeadConstants.pop_back();
312    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
313      if (GV->hasLocalLinkage())
314        RemoveDeadConstant(GV);
315    } else
316      RemoveDeadConstant(C);
317  }
318
319  return true;
320}
321
322/// getRealLinkageName - If special LLVM prefix that is used to inform the asm
323/// printer to not emit usual symbol prefix before the symbol name is used then
324/// return linkage name after skipping this special LLVM prefix.
325static StringRef getRealLinkageName(StringRef LinkageName) {
326  char One = '\1';
327  if (LinkageName.startswith(StringRef(&One, 1)))
328    return LinkageName.substr(1);
329  return LinkageName;
330}
331
332bool StripDeadDebugInfo::runOnModule(Module &M) {
333  bool Changed = false;
334
335  // Debugging infomration is encoded in llvm IR using metadata. This is designed
336  // such a way that debug info for symbols preserved even if symbols are
337  // optimized away by the optimizer. This special pass removes debug info for
338  // such symbols.
339
340  // llvm.dbg.gv keeps track of debug info for global variables.
341  if (NamedMDNode *NMD = M.getNamedMetadata("llvm.dbg.gv")) {
342    SmallVector<MDNode *, 8> MDs;
343    for (unsigned i = 0, e = NMD->getNumOperands(); i != e; ++i)
344      if (DIGlobalVariable(NMD->getOperand(i)).Verify())
345        MDs.push_back(NMD->getOperand(i));
346      else
347        Changed = true;
348    NMD->eraseFromParent();
349    NMD = NULL;
350
351    for (SmallVector<MDNode *, 8>::iterator I = MDs.begin(),
352           E = MDs.end(); I != E; ++I) {
353      if (M.getGlobalVariable(DIGlobalVariable(*I).getGlobal()->getName(),
354                              true)) {
355        if (!NMD)
356          NMD = M.getOrInsertNamedMetadata("llvm.dbg.gv");
357        NMD->addOperand(*I);
358      }
359      else
360        Changed = true;
361    }
362  }
363
364  // llvm.dbg.sp keeps track of debug info for subprograms.
365  if (NamedMDNode *NMD = M.getNamedMetadata("llvm.dbg.sp")) {
366    SmallVector<MDNode *, 8> MDs;
367    for (unsigned i = 0, e = NMD->getNumOperands(); i != e; ++i)
368      if (DISubprogram(NMD->getOperand(i)).Verify())
369        MDs.push_back(NMD->getOperand(i));
370      else
371        Changed = true;
372    NMD->eraseFromParent();
373    NMD = NULL;
374
375    for (SmallVector<MDNode *, 8>::iterator I = MDs.begin(),
376           E = MDs.end(); I != E; ++I) {
377      bool FnIsLive = false;
378      if (Function *F = DISubprogram(*I).getFunction())
379        if (M.getFunction(F->getName()))
380          FnIsLive = true;
381      if (FnIsLive) {
382          if (!NMD)
383            NMD = M.getOrInsertNamedMetadata("llvm.dbg.sp");
384          NMD->addOperand(*I);
385      } else {
386        // Remove llvm.dbg.lv.fnname named mdnode which may have been used
387        // to hold debug info for dead function's local variables.
388        StringRef FName = DISubprogram(*I).getLinkageName();
389        if (FName.empty())
390          FName = DISubprogram(*I).getName();
391        if (NamedMDNode *LVNMD =
392            M.getNamedMetadata(Twine("llvm.dbg.lv.",
393                                     getRealLinkageName(FName))))
394          LVNMD->eraseFromParent();
395      }
396    }
397  }
398
399  return Changed;
400}
401