lib/Linker/LinkArchives.cpp

   1 //===- lib/Linker/LinkArchives.cpp - Link LLVM objects and libraries ------===//
   2 //
   3 //                     The LLVM Compiler Infrastructure
   4 //
   5 // This file was developed by the LLVM research group and is distributed under
   6 // the University of Illinois Open Source License. See LICENSE.TXT for details.
   7 //
   8 //===----------------------------------------------------------------------===//
   9 //
  10 // This file contains routines to handle linking together LLVM bytecode files,
  11 // and to handle annoying things like static libraries.
  12 //
  13 //===----------------------------------------------------------------------===//
  14
  15 #include "llvm/Linker.h"
  16 #include "llvm/Module.h"
  17 #include "llvm/ModuleProvider.h"
  18 #include "llvm/ADT/SetOperations.h"
  19 #include "llvm/Bytecode/Reader.h"
  20 #include "llvm/Bytecode/Archive.h"
  21 #include "llvm/Config/config.h"
  22 #include <memory>
  23 #include <set>
  24 using namespace llvm;
  25
  26 /// GetAllDefinedSymbols - Modifies its parameter DefinedSymbols to contain the
  27 /// name of each externally-visible symbol defined in M.
  28 ///
  29 static void
  30 GetAllDefinedSymbols(Module *M, std::set<std::string> &DefinedSymbols) {
  31   for (Module::iterator I = M->begin(), E = M->end(); I != E; ++I)
  32     if (I->hasName() && !I->isExternal() && !I->hasInternalLinkage())
  33       DefinedSymbols.insert(I->getName());
  34   for (Module::global_iterator I = M->global_begin(), E = M->global_end();
  35        I != E; ++I)
  36     if (I->hasName() && !I->isExternal() && !I->hasInternalLinkage())
  37       DefinedSymbols.insert(I->getName());
  38 }
  39
  40 /// GetAllUndefinedSymbols - calculates the set of undefined symbols that still
  41 /// exist in an LLVM module. This is a bit tricky because there may be two
  42 /// symbols with the same name but different LLVM types that will be resolved to
  43 /// each other but aren't currently (thus we need to treat it as resolved).
  44 ///
  45 /// Inputs:
  46 ///  M - The module in which to find undefined symbols.
  47 ///
  48 /// Outputs:
  49 ///  UndefinedSymbols - A set of C++ strings containing the name of all
  50 ///                     undefined symbols.
  51 ///
  52 static void
  53 GetAllUndefinedSymbols(Module *M, std::set<std::string> &UndefinedSymbols) {
  54   std::set<std::string> DefinedSymbols;
  55   UndefinedSymbols.clear();
  56
  57   // If the program doesn't define a main, try pulling one in from a .a file.
  58   // This is needed for programs where the main function is defined in an
  59   // archive, such f2c'd programs.
  60   Function *Main = M->getMainFunction();
  61   if (Main == 0 || Main->isExternal())
  62     UndefinedSymbols.insert("main");
  63
  64   for (Module::iterator I = M->begin(), E = M->end(); I != E; ++I)
  65     if (I->hasName()) {
  66       if (I->isExternal())
  67         UndefinedSymbols.insert(I->getName());
  68       else if (!I->hasInternalLinkage())
  69         DefinedSymbols.insert(I->getName());
  70     }
  71   for (Module::global_iterator I = M->global_begin(), E = M->global_end();
  72        I != E; ++I)
  73     if (I->hasName()) {
  74       if (I->isExternal())
  75         UndefinedSymbols.insert(I->getName());
  76       else if (!I->hasInternalLinkage())
  77         DefinedSymbols.insert(I->getName());
  78     }
  79
  80   // Prune out any defined symbols from the undefined symbols set...
  81   for (std::set<std::string>::iterator I = UndefinedSymbols.begin();
  82        I != UndefinedSymbols.end(); )
  83     if (DefinedSymbols.count(*I))
  84       UndefinedSymbols.erase(I++);  // This symbol really is defined!
  85     else
  86       ++I; // Keep this symbol in the undefined symbols list
  87 }
  88
  89 /// LinkInArchive - opens an archive library and link in all objects which
  90 /// provide symbols that are currently undefined.
  91 ///
  92 /// Inputs:
  93 ///  Filename - The pathname of the archive.
  94 ///
  95 /// Return Value:
  96 ///  TRUE  - An error occurred.
  97 ///  FALSE - No errors.
  98 bool
  99 Linker::LinkInArchive(const sys::Path &Filename) {
 100
 101   // Make sure this is an archive file we're dealing with
 102   if (!Filename.isArchive())
 103     return error("File '" + Filename.toString() + "' is not an archive.");
 104
 105   // Open the archive file
 106   verbose("Linking archive file '" + Filename.toString() + "'");
 107
 108   // Find all of the symbols currently undefined in the bytecode program.
 109   // If all the symbols are defined, the program is complete, and there is
 110   // no reason to link in any archive files.
 111   std::set<std::string> UndefinedSymbols;
 112   GetAllUndefinedSymbols(Composite, UndefinedSymbols);
 113
 114   if (UndefinedSymbols.empty()) {
 115     verbose("No symbols undefined, skipping library '" +
 116             Filename.toString() + "'");
 117     return false;  // No need to link anything in!
 118   }
 119
 120   std::string ErrMsg;
 121   std::auto_ptr<Archive> AutoArch (
 122     Archive::OpenAndLoadSymbols(Filename,&ErrMsg));
 123
 124   Archive* arch = AutoArch.get();
 125
 126   if (!arch)
 127     return error("Cannot read archive '" + Filename.toString() +
 128                  "': " + ErrMsg);
 129
 130   // Save a set of symbols that are not defined by the archive. Since we're
 131   // entering a loop, there's no point searching for these multiple times. This
 132   // variable is used to "set_subtract" from the set of undefined symbols.
 133   std::set<std::string> NotDefinedByArchive;
 134
 135   // While we are linking in object files, loop.
 136   while (true) {
 137
 138     // Find the modules we need to link into the target module
 139     std::set<ModuleProvider*> Modules;
 140     arch->findModulesDefiningSymbols(UndefinedSymbols, Modules);
 141
 142     // If we didn't find any more modules to link this time, we are done
 143     // searching this archive.
 144     if (Modules.empty())
 145       break;
 146
 147     // Any symbols remaining in UndefinedSymbols after
 148     // findModulesDefiningSymbols are ones that the archive does not define. So
 149     // we add them to the NotDefinedByArchive variable now.
 150     NotDefinedByArchive.insert(UndefinedSymbols.begin(),
 151         UndefinedSymbols.end());
 152
 153     // Loop over all the ModuleProviders that we got back from the archive
 154     for (std::set<ModuleProvider*>::iterator I=Modules.begin(), E=Modules.end();
 155          I != E; ++I) {
 156
 157       // Get the module we must link in.
 158       std::auto_ptr<Module> AutoModule( (*I)->releaseModule() );
 159       Module* aModule = AutoModule.get();
 160
 161       verbose("  Linking in module: " + aModule->getModuleIdentifier());
 162
 163       // Link it in
 164       if (LinkInModule(aModule))
 165         return error("Cannot link in module '" +
 166                      aModule->getModuleIdentifier() + "': " + Error);
 167     }
 168
 169     // Get the undefined symbols from the aggregate module. This recomputes the
 170     // symbols we still need after the new modules have been linked in.
 171     GetAllUndefinedSymbols(Composite, UndefinedSymbols);
 172
 173     // At this point we have two sets of undefined symbols: UndefinedSymbols
 174     // which holds the undefined symbols from all the modules, and
 175     // NotDefinedByArchive which holds symbols we know the archive doesn't
 176     // define. There's no point searching for symbols that we won't find in the
 177     // archive so we subtract these sets.
 178     set_subtract(UndefinedSymbols, NotDefinedByArchive);
 179
 180     // If there's no symbols left, no point in continuing to search the
 181     // archive.
 182     if (UndefinedSymbols.empty())
 183       break;
 184   }
 185
 186   return false;
 187 }