llvm-extract.cpp revision 2cccc6220cc351b52d2cd2d0b7139502e854b68d 
1//===- llvm-extract.cpp - LLVM function extraction utility ----------------===//
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// This utility changes the input module to only contain a single function,
11// which is primarily used for debugging transformations.
12//
13//===----------------------------------------------------------------------===//
14
15#include "llvm/IR/LLVMContext.h"
16#include "llvm/ADT/SetVector.h"
17#include "llvm/ADT/SmallPtrSet.h"
18#include "llvm/Assembly/PrintModulePass.h"
19#include "llvm/Bitcode/ReaderWriter.h"
20#include "llvm/IR/DataLayout.h"
21#include "llvm/IR/Module.h"
22#include "llvm/IRReader/IRReader.h"
23#include "llvm/PassManager.h"
24#include "llvm/Support/CommandLine.h"
25#include "llvm/Support/ManagedStatic.h"
26#include "llvm/Support/PrettyStackTrace.h"
27#include "llvm/Support/Regex.h"
28#include "llvm/Support/Signals.h"
29#include "llvm/Support/SourceMgr.h"
30#include "llvm/Support/SystemUtils.h"
31#include "llvm/Support/ToolOutputFile.h"
32#include "llvm/Transforms/IPO.h"
33#include <memory>
34using namespace llvm;
35
36// InputFilename - The filename to read from.
37static cl::opt<std::string>
38InputFilename(cl::Positional, cl::desc("<input bitcode file>"),
39              cl::init("-"), cl::value_desc("filename"));
40
41static cl::opt<std::string>
42OutputFilename("o", cl::desc("Specify output filename"),
43               cl::value_desc("filename"), cl::init("-"));
44
45static cl::opt<bool>
46Force("f", cl::desc("Enable binary output on terminals"));
47
48static cl::opt<bool>
49DeleteFn("delete", cl::desc("Delete specified Globals from Module"));
50
51// ExtractFuncs - The functions to extract from the module.
52static cl::list<std::string>
53ExtractFuncs("func", cl::desc("Specify function to extract"),
54             cl::ZeroOrMore, cl::value_desc("function"));
55
56// ExtractRegExpFuncs - The functions, matched via regular expression, to
57// extract from the module.
58static cl::list<std::string>
59ExtractRegExpFuncs("rfunc", cl::desc("Specify function(s) to extract using a "
60                                     "regular expression"),
61                   cl::ZeroOrMore, cl::value_desc("rfunction"));
62
63// ExtractAlias - The alias to extract from the module.
64static cl::list<std::string>
65ExtractAliases("alias", cl::desc("Specify alias to extract"),
66               cl::ZeroOrMore, cl::value_desc("alias"));
67
68
69// ExtractRegExpAliases - The aliases, matched via regular expression, to
70// extract from the module.
71static cl::list<std::string>
72ExtractRegExpAliases("ralias", cl::desc("Specify alias(es) to extract using a "
73                                        "regular expression"),
74                     cl::ZeroOrMore, cl::value_desc("ralias"));
75
76// ExtractGlobals - The globals to extract from the module.
77static cl::list<std::string>
78ExtractGlobals("glob", cl::desc("Specify global to extract"),
79               cl::ZeroOrMore, cl::value_desc("global"));
80
81// ExtractRegExpGlobals - The globals, matched via regular expression, to
82// extract from the module...
83static cl::list<std::string>
84ExtractRegExpGlobals("rglob", cl::desc("Specify global(s) to extract using a "
85                                       "regular expression"),
86                     cl::ZeroOrMore, cl::value_desc("rglobal"));
87
88static cl::opt<bool>
89OutputAssembly("S",
90               cl::desc("Write output as LLVM assembly"), cl::Hidden);
91
92int main(int argc, char **argv) {
93  // Print a stack trace if we signal out.
94  sys::PrintStackTraceOnErrorSignal();
95  PrettyStackTraceProgram X(argc, argv);
96
97  LLVMContext &Context = getGlobalContext();
98  llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
99  cl::ParseCommandLineOptions(argc, argv, "llvm extractor\n");
100
101  // Use lazy loading, since we only care about selected global values.
102  SMDiagnostic Err;
103  OwningPtr<Module> M;
104  M.reset(getLazyIRFileModule(InputFilename, Err, Context));
105
106  if (M.get() == 0) {
107    Err.print(argv[0], errs());
108    return 1;
109  }
110
111  // Use SetVector to avoid duplicates.
112  SetVector<GlobalValue *> GVs;
113
114  // Figure out which aliases we should extract.
115  for (size_t i = 0, e = ExtractAliases.size(); i != e; ++i) {
116    GlobalAlias *GA = M->getNamedAlias(ExtractAliases[i]);
117    if (!GA) {
118      errs() << argv[0] << ": program doesn't contain alias named '"
119             << ExtractAliases[i] << "'!\n";
120      return 1;
121    }
122    GVs.insert(GA);
123  }
124
125  // Extract aliases via regular expression matching.
126  for (size_t i = 0, e = ExtractRegExpAliases.size(); i != e; ++i) {
127    std::string Error;
128    Regex RegEx(ExtractRegExpAliases[i]);
129    if (!RegEx.isValid(Error)) {
130      errs() << argv[0] << ": '" << ExtractRegExpAliases[i] << "' "
131        "invalid regex: " << Error;
132    }
133    bool match = false;
134    for (Module::alias_iterator GA = M->alias_begin(), E = M->alias_end();
135         GA != E; GA++) {
136      if (RegEx.match(GA->getName())) {
137        GVs.insert(&*GA);
138        match = true;
139      }
140    }
141    if (!match) {
142      errs() << argv[0] << ": program doesn't contain global named '"
143             << ExtractRegExpAliases[i] << "'!\n";
144      return 1;
145    }
146  }
147
148  // Figure out which globals we should extract.
149  for (size_t i = 0, e = ExtractGlobals.size(); i != e; ++i) {
150    GlobalValue *GV = M->getNamedGlobal(ExtractGlobals[i]);
151    if (!GV) {
152      errs() << argv[0] << ": program doesn't contain global named '"
153             << ExtractGlobals[i] << "'!\n";
154      return 1;
155    }
156    GVs.insert(GV);
157  }
158
159  // Extract globals via regular expression matching.
160  for (size_t i = 0, e = ExtractRegExpGlobals.size(); i != e; ++i) {
161    std::string Error;
162    Regex RegEx(ExtractRegExpGlobals[i]);
163    if (!RegEx.isValid(Error)) {
164      errs() << argv[0] << ": '" << ExtractRegExpGlobals[i] << "' "
165        "invalid regex: " << Error;
166    }
167    bool match = false;
168    for (Module::global_iterator GV = M->global_begin(),
169           E = M->global_end(); GV != E; GV++) {
170      if (RegEx.match(GV->getName())) {
171        GVs.insert(&*GV);
172        match = true;
173      }
174    }
175    if (!match) {
176      errs() << argv[0] << ": program doesn't contain global named '"
177             << ExtractRegExpGlobals[i] << "'!\n";
178      return 1;
179    }
180  }
181
182  // Figure out which functions we should extract.
183  for (size_t i = 0, e = ExtractFuncs.size(); i != e; ++i) {
184    GlobalValue *GV = M->getFunction(ExtractFuncs[i]);
185    if (!GV) {
186      errs() << argv[0] << ": program doesn't contain function named '"
187             << ExtractFuncs[i] << "'!\n";
188      return 1;
189    }
190    GVs.insert(GV);
191  }
192  // Extract functions via regular expression matching.
193  for (size_t i = 0, e = ExtractRegExpFuncs.size(); i != e; ++i) {
194    std::string Error;
195    StringRef RegExStr = ExtractRegExpFuncs[i];
196    Regex RegEx(RegExStr);
197    if (!RegEx.isValid(Error)) {
198      errs() << argv[0] << ": '" << ExtractRegExpFuncs[i] << "' "
199        "invalid regex: " << Error;
200    }
201    bool match = false;
202    for (Module::iterator F = M->begin(), E = M->end(); F != E;
203         F++) {
204      if (RegEx.match(F->getName())) {
205        GVs.insert(&*F);
206        match = true;
207      }
208    }
209    if (!match) {
210      errs() << argv[0] << ": program doesn't contain global named '"
211             << ExtractRegExpFuncs[i] << "'!\n";
212      return 1;
213    }
214  }
215
216  // Materialize requisite global values.
217  if (!DeleteFn)
218    for (size_t i = 0, e = GVs.size(); i != e; ++i) {
219      GlobalValue *GV = GVs[i];
220      if (GV->isMaterializable()) {
221        std::string ErrInfo;
222        if (GV->Materialize(&ErrInfo)) {
223          errs() << argv[0] << ": error reading input: " << ErrInfo << "\n";
224          return 1;
225        }
226      }
227    }
228  else {
229    // Deleting. Materialize every GV that's *not* in GVs.
230    SmallPtrSet<GlobalValue *, 8> GVSet(GVs.begin(), GVs.end());
231    for (Module::global_iterator I = M->global_begin(), E = M->global_end();
232         I != E; ++I) {
233      GlobalVariable *G = I;
234      if (!GVSet.count(G) && G->isMaterializable()) {
235        std::string ErrInfo;
236        if (G->Materialize(&ErrInfo)) {
237          errs() << argv[0] << ": error reading input: " << ErrInfo << "\n";
238          return 1;
239        }
240      }
241    }
242    for (Module::iterator I = M->begin(), E = M->end(); I != E; ++I) {
243      Function *F = I;
244      if (!GVSet.count(F) && F->isMaterializable()) {
245        std::string ErrInfo;
246        if (F->Materialize(&ErrInfo)) {
247          errs() << argv[0] << ": error reading input: " << ErrInfo << "\n";
248          return 1;
249        }
250      }
251    }
252  }
253
254  // In addition to deleting all other functions, we also want to spiff it
255  // up a little bit.  Do this now.
256  PassManager Passes;
257  Passes.add(new DataLayout(M.get())); // Use correct DataLayout
258
259  std::vector<GlobalValue*> Gvs(GVs.begin(), GVs.end());
260
261  Passes.add(createGVExtractionPass(Gvs, DeleteFn));
262  if (!DeleteFn)
263    Passes.add(createGlobalDCEPass());           // Delete unreachable globals
264  Passes.add(createStripDeadDebugInfoPass());    // Remove dead debug info
265  Passes.add(createStripDeadPrototypesPass());   // Remove dead func decls
266
267  std::string ErrorInfo;
268  tool_output_file Out(OutputFilename.c_str(), ErrorInfo, sys::fs::F_Binary);
269  if (!ErrorInfo.empty()) {
270    errs() << ErrorInfo << '\n';
271    return 1;
272  }
273
274  if (OutputAssembly)
275    Passes.add(createPrintModulePass(&Out.os()));
276  else if (Force || !CheckBitcodeOutputToConsole(Out.os(), true))
277    Passes.add(createBitcodeWriterPass(Out.os()));
278
279  Passes.run(*M.get());
280
281  // Declare success.
282  Out.keep();
283
284  return 0;
285}
286