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 || strncmp(V->getNameStart(), "llvm.dbg", 8))
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 void findUsedValues(Module &M,
161                     SmallPtrSet<const GlobalValue*, 8>& llvmUsedValues) {
162   if (GlobalVariable *LLVMUsed = M.getGlobalVariable("llvm.used")) {
163     llvmUsedValues.insert(LLVMUsed);
164     // Collect values that are preserved as per explicit request.
165     // llvm.used is used to list these values.
166     if (ConstantArray *Inits =
167         dyn_cast<ConstantArray>(LLVMUsed->getInitializer())) {
168       for (unsigned i = 0, e = Inits->getNumOperands(); i != e; ++i) {
169         if (GlobalValue *GV = dyn_cast<GlobalValue>(Inits->getOperand(i)))
170           llvmUsedValues.insert(GV);
171         else if (ConstantExpr *CE =
172                  dyn_cast<ConstantExpr>(Inits->getOperand(i)))
173           if (CE->getOpcode() == Instruction::BitCast)
174             if (GlobalValue *GV = dyn_cast<GlobalValue>(CE->getOperand(0)))
175               llvmUsedValues.insert(GV);
176       }
177     }
178   }
179 }
180 
181 /// StripSymbolNames - Strip symbol names.
182 bool StripSymbolNames(Module &M, bool PreserveDbgInfo) {
183 
184   SmallPtrSet<const GlobalValue*, 8> llvmUsedValues;
185   findUsedValues(M, llvmUsedValues);
186 
187   for (Module::global_iterator I = M.global_begin(), E = M.global_end();
188        I != E; ++I) {
189     if (I->hasLocalLinkage() && llvmUsedValues.count(I) == 0)
190       if (!PreserveDbgInfo || strncmp(I->getNameStart(), "llvm.dbg", 8))
191         I->setName("");     // Internal symbols can't participate in linkage
192   }
193 
194   for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I) {
195     if (I->hasLocalLinkage() && llvmUsedValues.count(I) == 0)
196       if (!PreserveDbgInfo || strncmp(I->getNameStart(), "llvm.dbg", 8))
197         I->setName("");     // Internal symbols can't participate in linkage
198     StripSymtab(I->getValueSymbolTable(), PreserveDbgInfo);
199   }
200 
201   // Remove all names from types.
202   StripTypeSymtab(M.getTypeSymbolTable(), PreserveDbgInfo);
203 
204   return true;
205 }
206 
207 // StripDebugInfo - Strip debug info in the module if it exists.
208 // To do this, we remove llvm.dbg.func.start, llvm.dbg.stoppoint, and
209 // llvm.dbg.region.end calls, and any globals they point to if now dead.
210 bool StripDebugInfo(Module &M) {
211 
212   SmallPtrSet<const GlobalValue*, 8> llvmUsedValues;
213   findUsedValues(M, llvmUsedValues);
214 
215   SmallVector<GlobalVariable *, 2> CUs;
216   SmallVector<GlobalVariable *, 4> GVs;
217   SmallVector<GlobalVariable *, 4> SPs;
218   CollectDebugInfoAnchors(M, CUs, GVs, SPs);
219   // These anchors use LinkOnce linkage so that the optimizer does not
220   // remove them accidently. Set InternalLinkage for all these debug
221   // info anchors.
222   for (SmallVector<GlobalVariable *, 2>::iterator I = CUs.begin(),
223          E = CUs.end(); I != E; ++I)
224     (*I)->setLinkage(GlobalValue::InternalLinkage);
225   for (SmallVector<GlobalVariable *, 4>::iterator I = GVs.begin(),
226          E = GVs.end(); I != E; ++I)
227     (*I)->setLinkage(GlobalValue::InternalLinkage);
228   for (SmallVector<GlobalVariable *, 4>::iterator I = SPs.begin(),
229          E = SPs.end(); I != E; ++I)
230     (*I)->setLinkage(GlobalValue::InternalLinkage);
231 
232 
233  // Delete all dbg variables.
234   for (Module::global_iterator I = M.global_begin(), E = M.global_end();
235        I != E; ++I) {
236     GlobalVariable *GV = dyn_cast<GlobalVariable>(I);
237     if (!GV) continue;
238     if (!GV->use_empty() && llvmUsedValues.count(I) == 0) {
239       if (strncmp(GV->getNameStart(), "llvm.dbg", 8) == 0) {
240         GV->replaceAllUsesWith(M.getContext().getUndef(GV->getType()));
241       }
242     }
243   }
244 
245   Function *FuncStart = M.getFunction("llvm.dbg.func.start");
246   Function *StopPoint = M.getFunction("llvm.dbg.stoppoint");
247   Function *RegionStart = M.getFunction("llvm.dbg.region.start");
248   Function *RegionEnd = M.getFunction("llvm.dbg.region.end");
249   Function *Declare = M.getFunction("llvm.dbg.declare");
250 
251   std::vector<Constant*> DeadConstants;
252 
253   // Remove all of the calls to the debugger intrinsics, and remove them from
254   // the module.
255   if (FuncStart) {
256     while (!FuncStart->use_empty()) {
257       CallInst *CI = cast<CallInst>(FuncStart->use_back());
258       Value *Arg = CI->getOperand(1);
259       assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
260       CI->eraseFromParent();
261       if (Arg->use_empty())
262         if (Constant *C = dyn_cast<Constant>(Arg))
263           DeadConstants.push_back(C);
264     }
265     FuncStart->eraseFromParent();
266   }
267   if (StopPoint) {
268     while (!StopPoint->use_empty()) {
269       CallInst *CI = cast<CallInst>(StopPoint->use_back());
270       Value *Arg = CI->getOperand(3);
271       assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
272       CI->eraseFromParent();
273       if (Arg->use_empty())
274         if (Constant *C = dyn_cast<Constant>(Arg))
275           DeadConstants.push_back(C);
276     }
277     StopPoint->eraseFromParent();
278   }
279   if (RegionStart) {
280     while (!RegionStart->use_empty()) {
281       CallInst *CI = cast<CallInst>(RegionStart->use_back());
282       Value *Arg = CI->getOperand(1);
283       assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
284       CI->eraseFromParent();
285       if (Arg->use_empty())
286         if (Constant *C = dyn_cast<Constant>(Arg))
287           DeadConstants.push_back(C);
288     }
289     RegionStart->eraseFromParent();
290   }
291   if (RegionEnd) {
292     while (!RegionEnd->use_empty()) {
293       CallInst *CI = cast<CallInst>(RegionEnd->use_back());
294       Value *Arg = CI->getOperand(1);
295       assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
296       CI->eraseFromParent();
297       if (Arg->use_empty())
298         if (Constant *C = dyn_cast<Constant>(Arg))
299           DeadConstants.push_back(C);
300     }
301     RegionEnd->eraseFromParent();
302   }
303   if (Declare) {
304     while (!Declare->use_empty()) {
305       CallInst *CI = cast<CallInst>(Declare->use_back());
306       Value *Arg1 = CI->getOperand(1);
307       Value *Arg2 = CI->getOperand(2);
308       assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
309       CI->eraseFromParent();
310       if (Arg1->use_empty()) {
311         if (Constant *C = dyn_cast<Constant>(Arg1))
312           DeadConstants.push_back(C);
313         else
314           RecursivelyDeleteTriviallyDeadInstructions(Arg1);
315       }
316       if (Arg2->use_empty())
317         if (Constant *C = dyn_cast<Constant>(Arg2))
318           DeadConstants.push_back(C);
319     }
320     Declare->eraseFromParent();
321   }
322 
323   // llvm.dbg.compile_units and llvm.dbg.subprograms are marked as linkonce
324   // but since we are removing all debug information, make them internal now.
325   // FIXME: Use private linkage maybe?
326   if (Constant *C = M.getNamedGlobal("llvm.dbg.compile_units"))
327     if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
328       GV->setLinkage(GlobalValue::InternalLinkage);
329 
330   if (Constant *C = M.getNamedGlobal("llvm.dbg.subprograms"))
331     if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
332       GV->setLinkage(GlobalValue::InternalLinkage);
333 
334   if (Constant *C = M.getNamedGlobal("llvm.dbg.global_variables"))
335     if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C))
336       GV->setLinkage(GlobalValue::InternalLinkage);
337 
338   // Delete all dbg variables.
339   for (Module::global_iterator I = M.global_begin(), E = M.global_end();
340        I != E; ++I) {
341     GlobalVariable *GV = dyn_cast<GlobalVariable>(I);
342     if (!GV) continue;
343     if (GV->use_empty() && llvmUsedValues.count(I) == 0
344         && (!GV->hasSection()
345             || strcmp(GV->getSection().c_str(), "llvm.metadata") == 0))
346       DeadConstants.push_back(GV);
347   }
348 
349   if (DeadConstants.empty())
350     return false;
351 
352   // Delete any internal globals that were only used by the debugger intrinsics.
353   while (!DeadConstants.empty()) {
354     Constant *C = DeadConstants.back();
355     DeadConstants.pop_back();
356     if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
357       if (GV->hasLocalLinkage())
358         RemoveDeadConstant(GV);
359     }
360     else
361       RemoveDeadConstant(C);
362   }
363 
364   // Remove all llvm.dbg types.
365   TypeSymbolTable &ST = M.getTypeSymbolTable();
366   for (TypeSymbolTable::iterator TI = ST.begin(), TE = ST.end(); TI != TE; ) {
367     if (!strncmp(TI->first.c_str(), "llvm.dbg.", 9))
368       ST.remove(TI++);
369     else
370       ++TI;
371   }
372 
373   return true;
374 }
375 
376 bool StripSymbols::runOnModule(Module &M) {
377   bool Changed = false;
378   Changed |= StripDebugInfo(M);
379   if (!OnlyDebugInfo)
380     Changed |= StripSymbolNames(M, false);
381   return Changed;
382 }
383 
384 bool StripNonDebugSymbols::runOnModule(Module &M) {
385   return StripSymbolNames(M, true);
386 }
387 
388 bool StripDebugDeclare::runOnModule(Module &M) {
389 
390   Function *Declare = M.getFunction("llvm.dbg.declare");
391   std::vector<Constant*> DeadConstants;
392 
393   if (Declare) {
394     while (!Declare->use_empty()) {
395       CallInst *CI = cast<CallInst>(Declare->use_back());
396       Value *Arg1 = CI->getOperand(1);
397       Value *Arg2 = CI->getOperand(2);
398       assert(CI->use_empty() && "llvm.dbg intrinsic should have void result");
399       CI->eraseFromParent();
400       if (Arg1->use_empty()) {
401         if (Constant *C = dyn_cast<Constant>(Arg1))
402           DeadConstants.push_back(C);
403         else
404           RecursivelyDeleteTriviallyDeadInstructions(Arg1);
405       }
406       if (Arg2->use_empty())
407         if (Constant *C = dyn_cast<Constant>(Arg2))
408           DeadConstants.push_back(C);
409     }
410     Declare->eraseFromParent();
411   }
412 
413   // Delete all llvm.dbg.global_variables.
414   for (Module::global_iterator I = M.global_begin(), E = M.global_end();
415        I != E; ++I) {
416     GlobalVariable *GV = dyn_cast<GlobalVariable>(I);
417     if (!GV) continue;
418     if (GV->use_empty() && GV->hasName()
419         && strncmp(GV->getNameStart(), "llvm.dbg.global_variable", 24) == 0)
420       DeadConstants.push_back(GV);
421   }
422 
423   while (!DeadConstants.empty()) {
424     Constant *C = DeadConstants.back();
425     DeadConstants.pop_back();
426     if (GlobalVariable *GV = dyn_cast<GlobalVariable>(C)) {
427       if (GV->hasLocalLinkage())
428         RemoveDeadConstant(GV);
429     }
430     else
431       RemoveDeadConstant(C);
432   }
433 
434   return true;
435 }
436