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/LLVMContext.h"
28 #include "llvm/Module.h"
29 #include "llvm/Pass.h"
30 #include "llvm/Analysis/DebugInfo.h"
31 #include "llvm/ValueSymbolTable.h"
32 #include "llvm/TypeSymbolTable.h"
33 #include "llvm/Transforms/Utils/Local.h"
34 #include "llvm/Support/Compiler.h"
35 #include "llvm/ADT/SmallPtrSet.h"
36 using namespace llvm;
37 
38 namespace {
39   class VISIBILITY_HIDDEN StripSymbols : public ModulePass {
40     bool OnlyDebugInfo;
41   public:
42     static char ID; // Pass identification, replacement for typeid
43     explicit StripSymbols(bool ODI = false)
44       : ModulePass(&ID), OnlyDebugInfo(ODI) {}
45 
46     virtual bool runOnModule(Module &M);
47 
48     virtual void getAnalysisUsage(AnalysisUsage &AU) const {
49       AU.setPreservesAll();
50     }
51   };
52 
53   class VISIBILITY_HIDDEN StripNonDebugSymbols : public ModulePass {
54   public:
55     static char ID; // Pass identification, replacement for typeid
56     explicit StripNonDebugSymbols()
57       : ModulePass(&ID) {}
58 
59     virtual bool runOnModule(Module &M);
60 
61     virtual void getAnalysisUsage(AnalysisUsage &AU) const {
62       AU.setPreservesAll();
63     }
64   };
65 
66   class VISIBILITY_HIDDEN StripDebugDeclare : public ModulePass {
67   public:
68     static char ID; // Pass identification, replacement for typeid
69     explicit StripDebugDeclare()
70       : ModulePass(&ID) {}
71 
72     virtual bool runOnModule(Module &M);
73 
74     virtual void getAnalysisUsage(AnalysisUsage &AU) const {
75       AU.setPreservesAll();
76     }
77   };
78 }
79 
80 char StripSymbols::ID = 0;
81 static RegisterPass<StripSymbols>
82 X("strip", "Strip all symbols from a module");
83 
84 ModulePass *llvm::createStripSymbolsPass(bool OnlyDebugInfo) {
85   return new StripSymbols(OnlyDebugInfo);
86 }
87 
88 char StripNonDebugSymbols::ID = 0;
89 static RegisterPass<StripNonDebugSymbols>
90 Y("strip-nondebug", "Strip all symbols, except dbg symbols, from a module");
91 
92 ModulePass *llvm::createStripNonDebugSymbolsPass() {
93   return new StripNonDebugSymbols();
94 }
95 
96 char StripDebugDeclare::ID = 0;
97 static RegisterPass<StripDebugDeclare>
98 Z("strip-debug-declare", "Strip all llvm.dbg.declare intrinsics");
99 
100 ModulePass *llvm::createStripDebugDeclarePass() {
101   return new StripDebugDeclare();
102 }
103 
104 /// OnlyUsedBy - Return true if V is only used by Usr.
105 static bool OnlyUsedBy(Value *V, Value *Usr) {
106   for(Value::use_iterator I = V->use_begin(), E = V->use_end(); I != E; ++I) {
107     User *U = *I;
108     if (U != Usr)
109       return false;
110   }
111   return true;
112 }
113 
114 static void RemoveDeadConstant(Constant *C) {
115   assert(C->use_empty() && "Constant is not dead!");
116   SmallPtrSet<Constant *, 4> Operands;
117   for (unsigned i = 0, e = C->getNumOperands(); i != e; ++i)
118     if (isa<DerivedType>(C->getOperand(i)->getType()) &&
119         OnlyUsedBy(C->getOperand(i), C))
120       Operands.insert(C->getOperand(i));
121   if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
122     if (!GV->hasLocalLinkage()) return;   // Don't delete non static globals.
123     GV->eraseFromParent();
124   }
125   else if (!isa<Function>(C))
126     if (isa<CompositeType>(C->getType()))
127       C->destroyConstant();
128 
129   // If the constant referenced anything, see if we can delete it as well.
130   for (SmallPtrSet<Constant *, 4>::iterator OI = Operands.begin(),
131          OE = Operands.end(); OI != OE; ++OI)
132     RemoveDeadConstant(*OI);
133 }
134 
135 // Strip the symbol table of its names.
136 //
137 static void StripSymtab(ValueSymbolTable &ST, bool PreserveDbgInfo) {
138   for (ValueSymbolTable::iterator VI = ST.begin(), VE = ST.end(); VI != VE; ) {
139     Value *V = VI->getValue();
140     ++VI;
141     if (!isa<GlobalValue>(V) || cast<GlobalValue>(V)->hasLocalLinkage()) {
142       if (!PreserveDbgInfo || !V->getName().startswith("llvm.dbg"))
143         // Set name to "", removing from symbol table!
144         V->setName("");
145     }
146   }
147 }
148 
149 // Strip the symbol table of its names.
150 static void StripTypeSymtab(TypeSymbolTable &ST, bool PreserveDbgInfo) {
151   for (TypeSymbolTable::iterator TI = ST.begin(), E = ST.end(); TI != E; ) {
152     if (PreserveDbgInfo && strncmp(TI->first.c_str(), "llvm.dbg", 8) == 0)
153       ++TI;
154     else
155       ST.remove(TI++);
156   }
157 }
158 
159 /// Find values that are marked as llvm.used.
160 static void findUsedValues(GlobalVariable *LLVMUsed,
161                            SmallPtrSet<const GlobalValue*, 8> &UsedValues) {
162   if (LLVMUsed == 0) return;
163   UsedValues.insert(LLVMUsed);
164 
165   ConstantArray *Inits = dyn_cast<ConstantArray>(LLVMUsed->getInitializer());
166   if (Inits == 0) return;
167 
168   for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i)
169     if (GlobalValue *GV =
170           dyn_cast<GlobalValue>(Inits->getOperand(i)->stripPointerCasts()))
171       UsedValues.insert(GV);
172 }
173 
174 /// StripSymbolNames - Strip symbol names.
175 bool StripSymbolNames(Module &M, bool PreserveDbgInfo) {
176 
177   SmallPtrSet<const GlobalValue*, 8> llvmUsedValues;
178   findUsedValues(M.getGlobalVariable("llvm.used"), llvmUsedValues);
179   findUsedValues(M.getGlobalVariable("llvm.compiler.used"), llvmUsedValues);
180 
181   for (Module::global_iterator I = M.global_begin(), E = M.global_end();
182        I != E; ++I) {
183     if (I->hasLocalLinkage() && llvmUsedValues.count(I) == 0)
184       if (!PreserveDbgInfo || !I->getName().startswith("llvm.dbg"))
185         I->setName("");     // Internal symbols can't participate in linkage
186   }
187 
188   for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I) {
189     if (I->hasLocalLinkage() && llvmUsedValues.count(I) == 0)
190       if (!PreserveDbgInfo || !I->getName().startswith("llvm.dbg"))
191         I->setName("");     // Internal symbols can't participate in linkage
192     StripSymtab(I->getValueSymbolTable(), PreserveDbgInfo);
193   }
194 
195   // Remove all names from types.
196   StripTypeSymtab(M.getTypeSymbolTable(), PreserveDbgInfo);
197 
198   return true;
199 }
200 
201 // StripDebugInfo - Strip debug info in the module if it exists.
202 // To do this, we remove llvm.dbg.func.start, llvm.dbg.stoppoint, and
203 // llvm.dbg.region.end calls, and any globals they point to if now dead.
204 bool StripDebugInfo(Module &M) {
205 
206   SmallPtrSet<const GlobalValue*, 8> llvmUsedValues;
207   findUsedValues(M.getGlobalVariable("llvm.used"), llvmUsedValues);
208   findUsedValues(M.getGlobalVariable("llvm.compiler.used"), llvmUsedValues);
209 
210   SmallVector<GlobalVariable *, 2> CUs;
211   SmallVector<GlobalVariable *, 4> GVs;
212   SmallVector<GlobalVariable *, 4> SPs;
213   CollectDebugInfoAnchors(M, CUs, GVs, SPs);
214   // These anchors use LinkOnce linkage so that the optimizer does not
215   // remove them accidently. Set InternalLinkage for all these debug
216   // info anchors.
217   for (SmallVector<GlobalVariable *, 2>::iterator I = CUs.begin(),
218          E = CUs.end(); I != E; ++I)
219     (*I)->setLinkage(GlobalValue::InternalLinkage);
220   for (SmallVector<GlobalVariable *, 4>::iterator I = GVs.begin(),
221          E = GVs.end(); I != E; ++I)
222     (*I)->setLinkage(GlobalValue::InternalLinkage);
223   for (SmallVector<GlobalVariable *, 4>::iterator I = SPs.begin(),
224          E = SPs.end(); I != E; ++I)
225     (*I)->setLinkage(GlobalValue::InternalLinkage);
226 
227 
228  // Delete all dbg variables.
229   for (Module::global_iterator I = M.global_begin(), E = M.global_end();
230        I != E; ++I) {
231     GlobalVariable *GV = dyn_cast<GlobalVariable>(I);
232     if (!GV) continue;
233     if (!GV->use_empty() && llvmUsedValues.count(I) == 0) {
234       if (GV->getName().startswith("llvm.dbg")) {
235         GV->replaceAllUsesWith(UndefValue::get(GV->getType()));
236       }
237     }
238   }
239 
240   Function *FuncStart = M.getFunction("llvm.dbg.func.start");
241   Function *StopPoint = M.getFunction("llvm.dbg.stoppoint");
242   Function *RegionStart = M.getFunction("llvm.dbg.region.start");
243   Function *RegionEnd = M.getFunction("llvm.dbg.region.end");
244   Function *Declare = M.getFunction("llvm.dbg.declare");
245 
246   std::vector<Constant*> DeadConstants;
247 
248   // Remove all of the calls to the debugger intrinsics, and remove them from
249   // the module.
250   if (FuncStart) {
251     while (!FuncStart->use_empty()) {
252       CallInst *CI = cast<CallInst>(FuncStart->use_back());
253       Value *Arg = CI->getOperand(1);
254       assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
255       CI->eraseFromParent();
256       if (Arg->use_empty())
257         if (Constant *C = dyn_cast<Constant>(Arg))
258           DeadConstants.push_back(C);
259     }
260     FuncStart->eraseFromParent();
261   }
262   if (StopPoint) {
263     while (!StopPoint->use_empty()) {
264       CallInst *CI = cast<CallInst>(StopPoint->use_back());
265       Value *Arg = CI->getOperand(3);
266       assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
267       CI->eraseFromParent();
268       if (Arg->use_empty())
269         if (Constant *C = dyn_cast<Constant>(Arg))
270           DeadConstants.push_back(C);
271     }
272     StopPoint->eraseFromParent();
273   }
274   if (RegionStart) {
275     while (!RegionStart->use_empty()) {
276       CallInst *CI = cast<CallInst>(RegionStart->use_back());
277       Value *Arg = CI->getOperand(1);
278       assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
279       CI->eraseFromParent();
280       if (Arg->use_empty())
281         if (Constant *C = dyn_cast<Constant>(Arg))
282           DeadConstants.push_back(C);
283     }
284     RegionStart->eraseFromParent();
285   }
286   if (RegionEnd) {
287     while (!RegionEnd->use_empty()) {
288       CallInst *CI = cast<CallInst>(RegionEnd->use_back());
289       Value *Arg = CI->getOperand(1);
290       assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
291       CI->eraseFromParent();
292       if (Arg->use_empty())
293         if (Constant *C = dyn_cast<Constant>(Arg))
294           DeadConstants.push_back(C);
295     }
296     RegionEnd->eraseFromParent();
297   }
298   if (Declare) {
299     while (!Declare->use_empty()) {
300       CallInst *CI = cast<CallInst>(Declare->use_back());
301       Value *Arg1 = CI->getOperand(1);
302       Value *Arg2 = CI->getOperand(2);
303       assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
304       CI->eraseFromParent();
305       if (Arg1->use_empty()) {
306         if (Constant *C = dyn_cast<Constant>(Arg1))
307           DeadConstants.push_back(C);
308         else
309           RecursivelyDeleteTriviallyDeadInstructions(Arg1);
310       }
311       if (Arg2->use_empty())
312         if (Constant *C = dyn_cast<Constant>(Arg2))
313           DeadConstants.push_back(C);
314     }
315     Declare->eraseFromParent();
316   }
317 
318   // llvm.dbg.compile_units and llvm.dbg.subprograms are marked as linkonce
319   // but since we are removing all debug information, make them internal now.
320   // FIXME: Use private linkage maybe?
321   if (Constant *C = M.getNamedGlobal("llvm.dbg.compile_units"))
322     if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
323       GV->setLinkage(GlobalValue::InternalLinkage);
324 
325   if (Constant *C = M.getNamedGlobal("llvm.dbg.subprograms"))
326     if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
327       GV->setLinkage(GlobalValue::InternalLinkage);
328 
329   if (Constant *C = M.getNamedGlobal("llvm.dbg.global_variables"))
330     if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
331       GV->setLinkage(GlobalValue::InternalLinkage);
332 
333   // Delete all dbg variables.
334   for (Module::global_iterator I = M.global_begin(), E = M.global_end();
335        I != E; ++I) {
336     GlobalVariable *GV = dyn_cast<GlobalVariable>(I);
337     if (!GV) continue;
338     if (GV->use_empty() && llvmUsedValues.count(I) == 0
339         && (!GV->hasSection()
340             || strcmp(GV->getSection().c_str(), "llvm.metadata") == 0))
341       DeadConstants.push_back(GV);
342   }
343 
344   if (DeadConstants.empty())
345     return false;
346 
347   // Delete any internal globals that were only used by the debugger intrinsics.
348   while (!DeadConstants.empty()) {
349     Constant *C = DeadConstants.back();
350     DeadConstants.pop_back();
351     if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
352       if (GV->hasLocalLinkage())
353         RemoveDeadConstant(GV);
354     }
355     else
356       RemoveDeadConstant(C);
357   }
358 
359   // Remove all llvm.dbg types.
360   TypeSymbolTable &ST = M.getTypeSymbolTable();
361   for (TypeSymbolTable::iterator TI = ST.begin(), TE = ST.end(); TI != TE; ) {
362     if (!strncmp(TI->first.c_str(), "llvm.dbg.", 9))
363       ST.remove(TI++);
364     else
365       ++TI;
366   }
367 
368   return true;
369 }
370 
371 bool StripSymbols::runOnModule(Module &M) {
372   bool Changed = false;
373   Changed |= StripDebugInfo(M);
374   if (!OnlyDebugInfo)
375     Changed |= StripSymbolNames(M, false);
376   return Changed;
377 }
378 
379 bool StripNonDebugSymbols::runOnModule(Module &M) {
380   return StripSymbolNames(M, true);
381 }
382 
383 bool StripDebugDeclare::runOnModule(Module &M) {
384 
385   Function *Declare = M.getFunction("llvm.dbg.declare");
386   std::vector<Constant*> DeadConstants;
387 
388   if (Declare) {
389     while (!Declare->use_empty()) {
390       CallInst *CI = cast<CallInst>(Declare->use_back());
391       Value *Arg1 = CI->getOperand(1);
392       Value *Arg2 = CI->getOperand(2);
393       assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
394       CI->eraseFromParent();
395       if (Arg1->use_empty()) {
396         if (Constant *C = dyn_cast<Constant>(Arg1))
397           DeadConstants.push_back(C);
398         else
399           RecursivelyDeleteTriviallyDeadInstructions(Arg1);
400       }
401       if (Arg2->use_empty())
402         if (Constant *C = dyn_cast<Constant>(Arg2))
403           DeadConstants.push_back(C);
404     }
405     Declare->eraseFromParent();
406   }
407 
408   // Delete all llvm.dbg.global_variables.
409   for (Module::global_iterator I = M.global_begin(), E = M.global_end();
410        I != E; ++I) {
411     GlobalVariable *GV = dyn_cast<GlobalVariable>(I);
412     if (!GV) continue;
413     if (GV->use_empty() && GV->getName().startswith("llvm.dbg.global_variable"))
414       DeadConstants.push_back(GV);
415   }
416 
417   while (!DeadConstants.empty()) {
418     Constant *C = DeadConstants.back();
419     DeadConstants.pop_back();
420     if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
421       if (GV->hasLocalLinkage())
422         RemoveDeadConstant(GV);
423     }
424     else
425       RemoveDeadConstant(C);
426   }
427 
428   return true;
429 }
430