11dc6a8d1SPeter Collingbourne //===- SplitModule.cpp - Split a module into partitions -------------------===//
21dc6a8d1SPeter Collingbourne //
31dc6a8d1SPeter Collingbourne //                     The LLVM Compiler Infrastructure
41dc6a8d1SPeter Collingbourne //
51dc6a8d1SPeter Collingbourne // This file is distributed under the University of Illinois Open Source
61dc6a8d1SPeter Collingbourne // License. See LICENSE.TXT for details.
71dc6a8d1SPeter Collingbourne //
81dc6a8d1SPeter Collingbourne //===----------------------------------------------------------------------===//
91dc6a8d1SPeter Collingbourne //
101dc6a8d1SPeter Collingbourne // This file defines the function llvm::SplitModule, which splits a module
111dc6a8d1SPeter Collingbourne // into multiple linkable partitions. It can be used to implement parallel code
121dc6a8d1SPeter Collingbourne // generation for link-time optimization.
131dc6a8d1SPeter Collingbourne //
141dc6a8d1SPeter Collingbourne //===----------------------------------------------------------------------===//
151dc6a8d1SPeter Collingbourne 
161dc6a8d1SPeter Collingbourne #include "llvm/Transforms/Utils/SplitModule.h"
175adb96ccSEugene Zelenko #include "llvm/ADT/DenseMap.h"
18d19d4d30SSergei Larin #include "llvm/ADT/EquivalenceClasses.h"
195adb96ccSEugene Zelenko #include "llvm/ADT/SmallPtrSet.h"
205adb96ccSEugene Zelenko #include "llvm/ADT/SmallVector.h"
215adb96ccSEugene Zelenko #include "llvm/ADT/StringRef.h"
225adb96ccSEugene Zelenko #include "llvm/IR/Comdat.h"
235adb96ccSEugene Zelenko #include "llvm/IR/Constant.h"
24f74f091eSEvgeniy Stepanov #include "llvm/IR/Constants.h"
251dc6a8d1SPeter Collingbourne #include "llvm/IR/Function.h"
261dc6a8d1SPeter Collingbourne #include "llvm/IR/GlobalAlias.h"
271dc6a8d1SPeter Collingbourne #include "llvm/IR/GlobalObject.h"
285adb96ccSEugene Zelenko #include "llvm/IR/GlobalIndirectSymbol.h"
291dc6a8d1SPeter Collingbourne #include "llvm/IR/GlobalValue.h"
305adb96ccSEugene Zelenko #include "llvm/IR/GlobalVariable.h"
315adb96ccSEugene Zelenko #include "llvm/IR/Instruction.h"
321dc6a8d1SPeter Collingbourne #include "llvm/IR/Module.h"
335adb96ccSEugene Zelenko #include "llvm/IR/User.h"
345adb96ccSEugene Zelenko #include "llvm/IR/Value.h"
355adb96ccSEugene Zelenko #include "llvm/Support/Casting.h"
36d19d4d30SSergei Larin #include "llvm/Support/Debug.h"
375adb96ccSEugene Zelenko #include "llvm/Support/ErrorHandling.h"
381dc6a8d1SPeter Collingbourne #include "llvm/Support/MD5.h"
391dc6a8d1SPeter Collingbourne #include "llvm/Support/raw_ostream.h"
401dc6a8d1SPeter Collingbourne #include "llvm/Transforms/Utils/Cloning.h"
415adb96ccSEugene Zelenko #include "llvm/Transforms/Utils/ValueMapper.h"
425adb96ccSEugene Zelenko #include <algorithm>
435adb96ccSEugene Zelenko #include <cassert>
445adb96ccSEugene Zelenko #include <iterator>
455adb96ccSEugene Zelenko #include <memory>
46d19d4d30SSergei Larin #include <queue>
475adb96ccSEugene Zelenko #include <utility>
485adb96ccSEugene Zelenko #include <vector>
491dc6a8d1SPeter Collingbourne 
501dc6a8d1SPeter Collingbourne using namespace llvm;
511dc6a8d1SPeter Collingbourne 
525adb96ccSEugene Zelenko #define DEBUG_TYPE "split-module"
535adb96ccSEugene Zelenko 
54d19d4d30SSergei Larin namespace {
555adb96ccSEugene Zelenko 
565adb96ccSEugene Zelenko using ClusterMapType = EquivalenceClasses<const GlobalValue *>;
575adb96ccSEugene Zelenko using ComdatMembersType = DenseMap<const Comdat *, const GlobalValue *>;
585adb96ccSEugene Zelenko using ClusterIDMapType = DenseMap<const GlobalValue *, unsigned>;
595adb96ccSEugene Zelenko 
605adb96ccSEugene Zelenko } // end anonymous namespace
61d19d4d30SSergei Larin 
62d19d4d30SSergei Larin static void addNonConstUser(ClusterMapType &GVtoClusterMap,
63d19d4d30SSergei Larin                             const GlobalValue *GV, const User *U) {
64d19d4d30SSergei Larin   assert((!isa<Constant>(U) || isa<GlobalValue>(U)) && "Bad user");
65d19d4d30SSergei Larin 
66d19d4d30SSergei Larin   if (const Instruction *I = dyn_cast<Instruction>(U)) {
67d19d4d30SSergei Larin     const GlobalValue *F = I->getParent()->getParent();
68d19d4d30SSergei Larin     GVtoClusterMap.unionSets(GV, F);
6995549497SDavid Majnemer   } else if (isa<GlobalIndirectSymbol>(U) || isa<Function>(U) ||
70d19d4d30SSergei Larin              isa<GlobalVariable>(U)) {
71d19d4d30SSergei Larin     GVtoClusterMap.unionSets(GV, cast<GlobalValue>(U));
72d19d4d30SSergei Larin   } else {
73d19d4d30SSergei Larin     llvm_unreachable("Underimplemented use case");
74d19d4d30SSergei Larin   }
75d19d4d30SSergei Larin }
76d19d4d30SSergei Larin 
77f74f091eSEvgeniy Stepanov // Adds all GlobalValue users of V to the same cluster as GV.
78f74f091eSEvgeniy Stepanov static void addAllGlobalValueUsers(ClusterMapType &GVtoClusterMap,
79f74f091eSEvgeniy Stepanov                                    const GlobalValue *GV, const Value *V) {
80f74f091eSEvgeniy Stepanov   for (auto *U : V->users()) {
81f74f091eSEvgeniy Stepanov     SmallVector<const User *, 4> Worklist;
82f74f091eSEvgeniy Stepanov     Worklist.push_back(U);
83f74f091eSEvgeniy Stepanov     while (!Worklist.empty()) {
84f74f091eSEvgeniy Stepanov       const User *UU = Worklist.pop_back_val();
85f74f091eSEvgeniy Stepanov       // For each constant that is not a GV (a pure const) recurse.
86f74f091eSEvgeniy Stepanov       if (isa<Constant>(UU) && !isa<GlobalValue>(UU)) {
87f74f091eSEvgeniy Stepanov         Worklist.append(UU->user_begin(), UU->user_end());
88f74f091eSEvgeniy Stepanov         continue;
89f74f091eSEvgeniy Stepanov       }
90f74f091eSEvgeniy Stepanov       addNonConstUser(GVtoClusterMap, GV, UU);
91f74f091eSEvgeniy Stepanov     }
92f74f091eSEvgeniy Stepanov   }
93f74f091eSEvgeniy Stepanov }
94f74f091eSEvgeniy Stepanov 
95d19d4d30SSergei Larin // Find partitions for module in the way that no locals need to be
96d19d4d30SSergei Larin // globalized.
97d19d4d30SSergei Larin // Try to balance pack those partitions into N files since this roughly equals
98d19d4d30SSergei Larin // thread balancing for the backend codegen step.
99d19d4d30SSergei Larin static void findPartitions(Module *M, ClusterIDMapType &ClusterIDMap,
100d19d4d30SSergei Larin                            unsigned N) {
101d19d4d30SSergei Larin   // At this point module should have the proper mix of globals and locals.
102d19d4d30SSergei Larin   // As we attempt to partition this module, we must not change any
103d19d4d30SSergei Larin   // locals to globals.
104d19d4d30SSergei Larin   DEBUG(dbgs() << "Partition module with (" << M->size() << ")functions\n");
105d19d4d30SSergei Larin   ClusterMapType GVtoClusterMap;
106d19d4d30SSergei Larin   ComdatMembersType ComdatMembers;
107d19d4d30SSergei Larin 
108d19d4d30SSergei Larin   auto recordGVSet = [&GVtoClusterMap, &ComdatMembers](GlobalValue &GV) {
109d19d4d30SSergei Larin     if (GV.isDeclaration())
110d19d4d30SSergei Larin       return;
111d19d4d30SSergei Larin 
112d19d4d30SSergei Larin     if (!GV.hasName())
113d19d4d30SSergei Larin       GV.setName("__llvmsplit_unnamed");
114d19d4d30SSergei Larin 
115d19d4d30SSergei Larin     // Comdat groups must not be partitioned. For comdat groups that contain
116d19d4d30SSergei Larin     // locals, record all their members here so we can keep them together.
117d19d4d30SSergei Larin     // Comdat groups that only contain external globals are already handled by
118d19d4d30SSergei Larin     // the MD5-based partitioning.
119d19d4d30SSergei Larin     if (const Comdat *C = GV.getComdat()) {
120d19d4d30SSergei Larin       auto &Member = ComdatMembers[C];
121d19d4d30SSergei Larin       if (Member)
122d19d4d30SSergei Larin         GVtoClusterMap.unionSets(Member, &GV);
123d19d4d30SSergei Larin       else
124d19d4d30SSergei Larin         Member = &GV;
125d19d4d30SSergei Larin     }
126d19d4d30SSergei Larin 
127427f570cSSergei Larin     // For aliases we should not separate them from their aliasees regardless
128427f570cSSergei Larin     // of linkage.
12995549497SDavid Majnemer     if (auto *GIS = dyn_cast<GlobalIndirectSymbol>(&GV)) {
13095549497SDavid Majnemer       if (const GlobalObject *Base = GIS->getBaseObject())
131427f570cSSergei Larin         GVtoClusterMap.unionSets(&GV, Base);
132427f570cSSergei Larin     }
133427f570cSSergei Larin 
134f74f091eSEvgeniy Stepanov     if (const Function *F = dyn_cast<Function>(&GV)) {
135f74f091eSEvgeniy Stepanov       for (const BasicBlock &BB : *F) {
136f74f091eSEvgeniy Stepanov         BlockAddress *BA = BlockAddress::lookup(&BB);
137f74f091eSEvgeniy Stepanov         if (!BA || !BA->isConstantUsed())
138d19d4d30SSergei Larin           continue;
139f74f091eSEvgeniy Stepanov         addAllGlobalValueUsers(GVtoClusterMap, F, BA);
140d19d4d30SSergei Larin       }
141d19d4d30SSergei Larin     }
142f74f091eSEvgeniy Stepanov 
143f74f091eSEvgeniy Stepanov     if (GV.hasLocalLinkage())
144f74f091eSEvgeniy Stepanov       addAllGlobalValueUsers(GVtoClusterMap, &GV, &GV);
145d19d4d30SSergei Larin   };
146d19d4d30SSergei Larin 
147*ecf0e952SAaron Ballman   llvm::for_each(M->functions(), recordGVSet);
148*ecf0e952SAaron Ballman   llvm::for_each(M->globals(), recordGVSet);
149*ecf0e952SAaron Ballman   llvm::for_each(M->aliases(), recordGVSet);
150d19d4d30SSergei Larin 
151d19d4d30SSergei Larin   // Assigned all GVs to merged clusters while balancing number of objects in
152d19d4d30SSergei Larin   // each.
153d19d4d30SSergei Larin   auto CompareClusters = [](const std::pair<unsigned, unsigned> &a,
154d19d4d30SSergei Larin                             const std::pair<unsigned, unsigned> &b) {
155d19d4d30SSergei Larin     if (a.second || b.second)
156d19d4d30SSergei Larin       return a.second > b.second;
157d19d4d30SSergei Larin     else
158d19d4d30SSergei Larin       return a.first > b.first;
159d19d4d30SSergei Larin   };
160d19d4d30SSergei Larin 
161d19d4d30SSergei Larin   std::priority_queue<std::pair<unsigned, unsigned>,
162d19d4d30SSergei Larin                       std::vector<std::pair<unsigned, unsigned>>,
163d19d4d30SSergei Larin                       decltype(CompareClusters)>
164d19d4d30SSergei Larin       BalancinQueue(CompareClusters);
165d19d4d30SSergei Larin   // Pre-populate priority queue with N slot blanks.
166d19d4d30SSergei Larin   for (unsigned i = 0; i < N; ++i)
167d19d4d30SSergei Larin     BalancinQueue.push(std::make_pair(i, 0));
168d19d4d30SSergei Larin 
1695adb96ccSEugene Zelenko   using SortType = std::pair<unsigned, ClusterMapType::iterator>;
1705adb96ccSEugene Zelenko 
171d19d4d30SSergei Larin   SmallVector<SortType, 64> Sets;
172b30f2f51SMatthias Braun   SmallPtrSet<const GlobalValue *, 32> Visited;
173d19d4d30SSergei Larin 
174d19d4d30SSergei Larin   // To guarantee determinism, we have to sort SCC according to size.
175d19d4d30SSergei Larin   // When size is the same, use leader's name.
176d19d4d30SSergei Larin   for (ClusterMapType::iterator I = GVtoClusterMap.begin(),
177d19d4d30SSergei Larin                                 E = GVtoClusterMap.end(); I != E; ++I)
178d19d4d30SSergei Larin     if (I->isLeader())
179d19d4d30SSergei Larin       Sets.push_back(
180d19d4d30SSergei Larin           std::make_pair(std::distance(GVtoClusterMap.member_begin(I),
181d19d4d30SSergei Larin                                        GVtoClusterMap.member_end()), I));
182d19d4d30SSergei Larin 
183d19d4d30SSergei Larin   std::sort(Sets.begin(), Sets.end(), [](const SortType &a, const SortType &b) {
184d19d4d30SSergei Larin     if (a.first == b.first)
185d19d4d30SSergei Larin       return a.second->getData()->getName() > b.second->getData()->getName();
186d19d4d30SSergei Larin     else
187d19d4d30SSergei Larin       return a.first > b.first;
188d19d4d30SSergei Larin   });
189d19d4d30SSergei Larin 
190d19d4d30SSergei Larin   for (auto &I : Sets) {
191d19d4d30SSergei Larin     unsigned CurrentClusterID = BalancinQueue.top().first;
192d19d4d30SSergei Larin     unsigned CurrentClusterSize = BalancinQueue.top().second;
193d19d4d30SSergei Larin     BalancinQueue.pop();
194d19d4d30SSergei Larin 
195d19d4d30SSergei Larin     DEBUG(dbgs() << "Root[" << CurrentClusterID << "] cluster_size(" << I.first
196d19d4d30SSergei Larin                  << ") ----> " << I.second->getData()->getName() << "\n");
197d19d4d30SSergei Larin 
198d19d4d30SSergei Larin     for (ClusterMapType::member_iterator MI =
199d19d4d30SSergei Larin              GVtoClusterMap.findLeader(I.second);
200d19d4d30SSergei Larin          MI != GVtoClusterMap.member_end(); ++MI) {
201d19d4d30SSergei Larin       if (!Visited.insert(*MI).second)
202d19d4d30SSergei Larin         continue;
203d19d4d30SSergei Larin       DEBUG(dbgs() << "----> " << (*MI)->getName()
204d19d4d30SSergei Larin                    << ((*MI)->hasLocalLinkage() ? " l " : " e ") << "\n");
205d19d4d30SSergei Larin       Visited.insert(*MI);
206d19d4d30SSergei Larin       ClusterIDMap[*MI] = CurrentClusterID;
207d19d4d30SSergei Larin       CurrentClusterSize++;
208d19d4d30SSergei Larin     }
209d19d4d30SSergei Larin     // Add this set size to the number of entries in this cluster.
210d19d4d30SSergei Larin     BalancinQueue.push(std::make_pair(CurrentClusterID, CurrentClusterSize));
211d19d4d30SSergei Larin   }
212d19d4d30SSergei Larin }
213d19d4d30SSergei Larin 
2141dc6a8d1SPeter Collingbourne static void externalize(GlobalValue *GV) {
2151dc6a8d1SPeter Collingbourne   if (GV->hasLocalLinkage()) {
2161dc6a8d1SPeter Collingbourne     GV->setLinkage(GlobalValue::ExternalLinkage);
2171dc6a8d1SPeter Collingbourne     GV->setVisibility(GlobalValue::HiddenVisibility);
2181dc6a8d1SPeter Collingbourne   }
2191dc6a8d1SPeter Collingbourne 
2201dc6a8d1SPeter Collingbourne   // Unnamed entities must be named consistently between modules. setName will
2211dc6a8d1SPeter Collingbourne   // give a distinct name to each such entity.
2221dc6a8d1SPeter Collingbourne   if (!GV->hasName())
2231dc6a8d1SPeter Collingbourne     GV->setName("__llvmsplit_unnamed");
2241dc6a8d1SPeter Collingbourne }
2251dc6a8d1SPeter Collingbourne 
2261dc6a8d1SPeter Collingbourne // Returns whether GV should be in partition (0-based) I of N.
2271dc6a8d1SPeter Collingbourne static bool isInPartition(const GlobalValue *GV, unsigned I, unsigned N) {
22895549497SDavid Majnemer   if (auto *GIS = dyn_cast<GlobalIndirectSymbol>(GV))
22995549497SDavid Majnemer     if (const GlobalObject *Base = GIS->getBaseObject())
2301dc6a8d1SPeter Collingbourne       GV = Base;
2311dc6a8d1SPeter Collingbourne 
2321dc6a8d1SPeter Collingbourne   StringRef Name;
2331dc6a8d1SPeter Collingbourne   if (const Comdat *C = GV->getComdat())
2341dc6a8d1SPeter Collingbourne     Name = C->getName();
2351dc6a8d1SPeter Collingbourne   else
2361dc6a8d1SPeter Collingbourne     Name = GV->getName();
2371dc6a8d1SPeter Collingbourne 
2381dc6a8d1SPeter Collingbourne   // Partition by MD5 hash. We only need a few bits for evenness as the number
2391dc6a8d1SPeter Collingbourne   // of partitions will generally be in the 1-2 figure range; the low 16 bits
2401dc6a8d1SPeter Collingbourne   // are enough.
2411dc6a8d1SPeter Collingbourne   MD5 H;
2421dc6a8d1SPeter Collingbourne   MD5::MD5Result R;
2431dc6a8d1SPeter Collingbourne   H.update(Name);
2441dc6a8d1SPeter Collingbourne   H.final(R);
2451dc6a8d1SPeter Collingbourne   return (R[0] | (R[1] << 8)) % N == I;
2461dc6a8d1SPeter Collingbourne }
2471dc6a8d1SPeter Collingbourne 
2481dc6a8d1SPeter Collingbourne void llvm::SplitModule(
2491dc6a8d1SPeter Collingbourne     std::unique_ptr<Module> M, unsigned N,
250d3f4c05aSBenjamin Kramer     function_ref<void(std::unique_ptr<Module> MPart)> ModuleCallback,
251d19d4d30SSergei Larin     bool PreserveLocals) {
252d19d4d30SSergei Larin   if (!PreserveLocals) {
2531dc6a8d1SPeter Collingbourne     for (Function &F : *M)
2541dc6a8d1SPeter Collingbourne       externalize(&F);
2551dc6a8d1SPeter Collingbourne     for (GlobalVariable &GV : M->globals())
2561dc6a8d1SPeter Collingbourne       externalize(&GV);
2571dc6a8d1SPeter Collingbourne     for (GlobalAlias &GA : M->aliases())
2581dc6a8d1SPeter Collingbourne       externalize(&GA);
25995549497SDavid Majnemer     for (GlobalIFunc &GIF : M->ifuncs())
26095549497SDavid Majnemer       externalize(&GIF);
2611dc6a8d1SPeter Collingbourne   }
262f74f091eSEvgeniy Stepanov 
263d19d4d30SSergei Larin   // This performs splitting without a need for externalization, which might not
264d19d4d30SSergei Larin   // always be possible.
265d19d4d30SSergei Larin   ClusterIDMapType ClusterIDMap;
266d19d4d30SSergei Larin   findPartitions(M.get(), ClusterIDMap, N);
267d19d4d30SSergei Larin 
268f74f091eSEvgeniy Stepanov   // FIXME: We should be able to reuse M as the last partition instead of
269f74f091eSEvgeniy Stepanov   // cloning it.
270d19d4d30SSergei Larin   for (unsigned I = 0; I < N; ++I) {
271d19d4d30SSergei Larin     ValueToValueMapTy VMap;
272d19d4d30SSergei Larin     std::unique_ptr<Module> MPart(
273d19d4d30SSergei Larin         CloneModule(M.get(), VMap, [&](const GlobalValue *GV) {
274d19d4d30SSergei Larin           if (ClusterIDMap.count(GV))
275d19d4d30SSergei Larin             return (ClusterIDMap[GV] == I);
276d19d4d30SSergei Larin           else
277d19d4d30SSergei Larin             return isInPartition(GV, I, N);
278d19d4d30SSergei Larin         }));
279d19d4d30SSergei Larin     if (I != 0)
280d19d4d30SSergei Larin       MPart->setModuleInlineAsm("");
281d19d4d30SSergei Larin     ModuleCallback(std::move(MPart));
282d19d4d30SSergei Larin   }
283d19d4d30SSergei Larin }
284