1 //===- FunctionImport.cpp - ThinLTO Summary-based Function Import ---------===//
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 file implements Function import based on summaries.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "llvm/Transforms/IPO/FunctionImport.h"
15 
16 #include "llvm/ADT/StringSet.h"
17 #include "llvm/IR/AutoUpgrade.h"
18 #include "llvm/IR/DiagnosticPrinter.h"
19 #include "llvm/IR/IntrinsicInst.h"
20 #include "llvm/IR/Module.h"
21 #include "llvm/IRReader/IRReader.h"
22 #include "llvm/Linker/Linker.h"
23 #include "llvm/Object/FunctionIndexObjectFile.h"
24 #include "llvm/Support/CommandLine.h"
25 #include "llvm/Support/Debug.h"
26 #include "llvm/Support/SourceMgr.h"
27 using namespace llvm;
28 
29 #define DEBUG_TYPE "function-import"
30 
31 // Load lazily a module from \p FileName in \p Context.
32 static std::unique_ptr<Module> loadFile(const std::string &FileName,
33                                         LLVMContext &Context) {
34   SMDiagnostic Err;
35   DEBUG(dbgs() << "Loading '" << FileName << "'\n");
36   std::unique_ptr<Module> Result = getLazyIRFileModule(FileName, Err, Context);
37   if (!Result) {
38     Err.print("function-import", errs());
39     return nullptr;
40   }
41 
42   Result->materializeMetadata();
43   UpgradeDebugInfo(*Result);
44 
45   return Result;
46 }
47 
48 // Get a Module for \p FileName from the cache, or load it lazily.
49 Module &FunctionImporter::getOrLoadModule(StringRef FileName) {
50   auto &Module = ModuleMap[FileName];
51   if (!Module)
52     Module = loadFile(FileName, Context);
53   return *Module;
54 }
55 
56 /// Walk through the instructions in \p F looking for external
57 /// calls not already in the \p CalledFunctions set. If any are
58 /// found they are added to the \p Worklist for importing.
59 static void findExternalCalls(const Function &F, StringSet<> &CalledFunctions,
60                               SmallVector<StringRef, 64> &Worklist) {
61   for (auto &BB : F) {
62     for (auto &I : BB) {
63       if (isa<CallInst>(I)) {
64         DEBUG(dbgs() << "Found a call: '" << I << "'\n");
65         auto CalledFunction = cast<CallInst>(I).getCalledFunction();
66         // Insert any new external calls that have not already been
67         // added to set/worklist.
68         if (CalledFunction && CalledFunction->hasName() &&
69             CalledFunction->isDeclaration() &&
70             !CalledFunctions.count(CalledFunction->getName())) {
71           CalledFunctions.insert(CalledFunction->getName());
72           Worklist.push_back(CalledFunction->getName());
73         }
74       }
75     }
76   }
77 }
78 
79 // Automatically import functions in Module \p M based on the summaries index.
80 //
81 // The current implementation imports every called functions that exists in the
82 // summaries index.
83 bool FunctionImporter::importFunctions(Module &M) {
84   assert(&Context == &M.getContext());
85 
86   bool Changed = false;
87 
88   /// First step is collecting the called external functions.
89   StringSet<> CalledFunctions;
90   SmallVector<StringRef, 64> Worklist;
91   for (auto &F : M) {
92     if (F.isDeclaration() || F.hasFnAttribute(Attribute::OptimizeNone))
93       continue;
94     findExternalCalls(F, CalledFunctions, Worklist);
95   }
96 
97   /// Second step: for every call to an external function, try to import it.
98 
99   // Linker that will be used for importing function
100   Linker L(&M, DiagnosticHandler);
101 
102   while (!Worklist.empty()) {
103     auto CalledFunctionName = Worklist.pop_back_val();
104     DEBUG(dbgs() << "Process import for " << CalledFunctionName << "\n");
105 
106     // Try to get a summary for this function call.
107     auto InfoList = Index.findFunctionInfoList(CalledFunctionName);
108     if (InfoList == Index.end()) {
109       DEBUG(dbgs() << "No summary for " << CalledFunctionName
110                    << " Ignoring.\n");
111       continue;
112     }
113     assert(!InfoList->second.empty() && "No summary, error at import?");
114 
115     // Comdat can have multiple entries, FIXME: what do we do with them?
116     auto &Info = InfoList->second[0];
117     assert(Info && "Nullptr in list, error importing summaries?\n");
118 
119     auto *Summary = Info->functionSummary();
120     if (!Summary) {
121       // FIXME: in case we are lazyloading summaries, we can do it now.
122       dbgs() << "Missing summary for  " << CalledFunctionName
123              << ", error at import?\n";
124       llvm_unreachable("Missing summary");
125     }
126 
127     //
128     // No profitability notion right now, just import all the time...
129     //
130 
131     // Get the module path from the summary.
132     auto FileName = Summary->modulePath();
133     DEBUG(dbgs() << "Importing " << CalledFunctionName << " from " << FileName
134                  << "\n");
135 
136     // Get the module for the import (potentially from the cache).
137     auto &Module = getOrLoadModule(FileName);
138 
139     // The function that we will import!
140     GlobalValue *SGV = Module.getNamedValue(CalledFunctionName);
141     StringRef ImportFunctionName = CalledFunctionName;
142     if (!SGV) {
143       // Might be local in source Module, promoted/renamed in dest Module M.
144       std::pair<StringRef, StringRef> Split =
145           CalledFunctionName.split(".llvm.");
146       SGV = Module.getNamedValue(Split.first);
147 #ifndef NDEBUG
148       // Assert that Split.second is module id
149       uint64_t ModuleId;
150       assert(!Split.second.getAsInteger(10, ModuleId));
151       assert(ModuleId == Index.getModuleId(FileName));
152 #endif
153     }
154     Function *F = dyn_cast<Function>(SGV);
155     if (!F && isa<GlobalAlias>(SGV)) {
156       auto *SGA = dyn_cast<GlobalAlias>(SGV);
157       F = dyn_cast<Function>(SGA->getBaseObject());
158       ImportFunctionName = F->getName();
159     }
160     if (!F) {
161       errs() << "Can't load function '" << CalledFunctionName << "' in Module '"
162              << FileName << "', error in the summary?\n";
163       llvm_unreachable("Can't load function in Module");
164     }
165 
166     // We cannot import weak_any functions/aliases without possibly affecting
167     // the order they are seen and selected by the linker, changing program
168     // semantics.
169     if (SGV->hasWeakAnyLinkage()) {
170       DEBUG(dbgs() << "Ignoring import request for weak-any "
171                    << (isa<Function>(SGV) ? "function " : "alias ")
172                    << CalledFunctionName << " from " << FileName << "\n");
173       continue;
174     }
175 
176     // Link in the specified function.
177     if (L.linkInModule(&Module, Linker::Flags::None, &Index, F))
178       report_fatal_error("Function Import: link error");
179 
180     // Process the newly imported function and add callees to the worklist.
181     GlobalValue *NewGV = M.getNamedValue(ImportFunctionName);
182     assert(NewGV);
183     Function *NewF = dyn_cast<Function>(NewGV);
184     assert(NewF);
185     findExternalCalls(*NewF, CalledFunctions, Worklist);
186 
187     Changed = true;
188   }
189   return Changed;
190 }
191 
192 /// Summary file to use for function importing when using -function-import from
193 /// the command line.
194 static cl::opt<std::string>
195     SummaryFile("summary-file",
196                 cl::desc("The summary file to use for function importing."));
197 
198 static void diagnosticHandler(const DiagnosticInfo &DI) {
199   raw_ostream &OS = errs();
200   DiagnosticPrinterRawOStream DP(OS);
201   DI.print(DP);
202   OS << '\n';
203 }
204 
205 /// Parse the function index out of an IR file and return the function
206 /// index object if found, or nullptr if not.
207 static std::unique_ptr<FunctionInfoIndex>
208 getFunctionIndexForFile(StringRef Path, std::string &Error,
209                         DiagnosticHandlerFunction DiagnosticHandler) {
210   std::unique_ptr<MemoryBuffer> Buffer;
211   ErrorOr<std::unique_ptr<MemoryBuffer>> BufferOrErr =
212       MemoryBuffer::getFile(Path);
213   if (std::error_code EC = BufferOrErr.getError()) {
214     Error = EC.message();
215     return nullptr;
216   }
217   Buffer = std::move(BufferOrErr.get());
218   ErrorOr<std::unique_ptr<object::FunctionIndexObjectFile>> ObjOrErr =
219       object::FunctionIndexObjectFile::create(Buffer->getMemBufferRef(),
220                                               DiagnosticHandler);
221   if (std::error_code EC = ObjOrErr.getError()) {
222     Error = EC.message();
223     return nullptr;
224   }
225   return (*ObjOrErr)->takeIndex();
226 }
227 
228 /// Pass that performs cross-module function import provided a summary file.
229 class FunctionImportPass : public ModulePass {
230 
231 public:
232   /// Pass identification, replacement for typeid
233   static char ID;
234 
235   explicit FunctionImportPass() : ModulePass(ID) {}
236 
237   bool runOnModule(Module &M) override {
238     if (SummaryFile.empty()) {
239       report_fatal_error("error: -function-import requires -summary-file\n");
240     }
241     std::string Error;
242     std::unique_ptr<FunctionInfoIndex> Index =
243         getFunctionIndexForFile(SummaryFile, Error, diagnosticHandler);
244     if (!Index) {
245       errs() << "Error loading file '" << SummaryFile << "': " << Error << "\n";
246       return false;
247     }
248 
249     // Perform the import now.
250     FunctionImporter Importer(M.getContext(), *Index, diagnosticHandler);
251     return Importer.importFunctions(M);
252 
253     return false;
254   }
255 };
256 
257 char FunctionImportPass::ID = 0;
258 INITIALIZE_PASS_BEGIN(FunctionImportPass, "function-import",
259                       "Summary Based Function Import", false, false)
260 INITIALIZE_PASS_END(FunctionImportPass, "function-import",
261                     "Summary Based Function Import", false, false)
262 
263 namespace llvm {
264 Pass *createFunctionImportPass() { return new FunctionImportPass(); }
265 }
266