Add to the split module utility an SCC based method which allows not to globalize any local variables.
Summary:
Currently llvm::SplitModule as the first step globalizes all local objects, which might not be desirable in some scenarios.
This change adds a new flag to llvm::SplitModule that uses SCC approach to search for a balanced partition without the need to externalize symbols.
Such partition might not be possible or fully balanced for a given number of partitions, and is a function of the module properties (global/local dependencies within the module).
Joint development Tobias Edler von Koch (tobias@codeaurora.org) and Sergei Larin (slarin@codeaurora.org)
Subscribers: llvm-commits, joker.eph
Differential Revision: http://reviews.llvm.org/D16124
llvm-svn: 258083
diff --git a/llvm/lib/Transforms/Utils/SplitModule.cpp b/llvm/lib/Transforms/Utils/SplitModule.cpp
old mode 100644
new mode 100755
index ad6b782..b3d0f52
--- a/llvm/lib/Transforms/Utils/SplitModule.cpp
+++ b/llvm/lib/Transforms/Utils/SplitModule.cpp
@@ -13,19 +13,166 @@
//
//===----------------------------------------------------------------------===//
+#define DEBUG_TYPE "split-module"
+
#include "llvm/Transforms/Utils/SplitModule.h"
+#include "llvm/ADT/EquivalenceClasses.h"
#include "llvm/ADT/Hashing.h"
+#include "llvm/ADT/MapVector.h"
+#include "llvm/ADT/SetVector.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/GlobalAlias.h"
#include "llvm/IR/GlobalObject.h"
#include "llvm/IR/GlobalValue.h"
#include "llvm/IR/Module.h"
+#include "llvm/Support/CommandLine.h"
+#include "llvm/Support/Debug.h"
#include "llvm/Support/MD5.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Transforms/Utils/Cloning.h"
+#include <queue>
using namespace llvm;
+namespace {
+typedef EquivalenceClasses<const GlobalValue *> ClusterMapType;
+typedef DenseMap<const Comdat *, const GlobalValue *> ComdatMembersType;
+typedef DenseMap<const GlobalValue *, unsigned> ClusterIDMapType;
+}
+
+static void addNonConstUser(ClusterMapType &GVtoClusterMap,
+ const GlobalValue *GV, const User *U) {
+ assert((!isa<Constant>(U) || isa<GlobalValue>(U)) && "Bad user");
+
+ if (const Instruction *I = dyn_cast<Instruction>(U)) {
+ const GlobalValue *F = I->getParent()->getParent();
+ GVtoClusterMap.unionSets(GV, F);
+ } else if (isa<GlobalAlias>(U) || isa<Function>(U) ||
+ isa<GlobalVariable>(U)) {
+ GVtoClusterMap.unionSets(GV, cast<GlobalValue>(U));
+ } else {
+ llvm_unreachable("Underimplemented use case");
+ }
+}
+
+// Find partitions for module in the way that no locals need to be
+// globalized.
+// Try to balance pack those partitions into N files since this roughly equals
+// thread balancing for the backend codegen step.
+static void findPartitions(Module *M, ClusterIDMapType &ClusterIDMap,
+ unsigned N) {
+ // At this point module should have the proper mix of globals and locals.
+ // As we attempt to partition this module, we must not change any
+ // locals to globals.
+
+ DEBUG(dbgs() << "Partition module with (" << M->size() << ")functions\n");
+ ClusterMapType GVtoClusterMap;
+ ComdatMembersType ComdatMembers;
+
+ auto recordGVSet = [&GVtoClusterMap, &ComdatMembers](GlobalValue &GV) {
+ if (GV.isDeclaration())
+ return;
+
+ if (!GV.hasName())
+ GV.setName("__llvmsplit_unnamed");
+
+ // Comdat groups must not be partitioned. For comdat groups that contain
+ // locals, record all their members here so we can keep them together.
+ // Comdat groups that only contain external globals are already handled by
+ // the MD5-based partitioning.
+ if (const Comdat *C = GV.getComdat()) {
+ auto &Member = ComdatMembers[C];
+ if (Member)
+ GVtoClusterMap.unionSets(Member, &GV);
+ else
+ Member = &GV;
+ }
+
+ // Further only iterate over local GVs.
+ if (!GV.hasLocalLinkage())
+ return;
+
+ for (auto *U : GV.users()) {
+ SmallVector<const User *, 4> Worklist;
+ Worklist.push_back(U);
+ while (!Worklist.empty()) {
+ const User *UU = Worklist.pop_back_val();
+ // For each constant that is not a GV (a pure const) recurse.
+ if (isa<Constant>(UU) && !isa<GlobalValue>(UU)) {
+ Worklist.append(UU->user_begin(), UU->user_end());
+ continue;
+ }
+ addNonConstUser(GVtoClusterMap, &GV, UU);
+ }
+ }
+ };
+
+ std::for_each(M->begin(), M->end(), recordGVSet);
+ std::for_each(M->global_begin(), M->global_end(), recordGVSet);
+ std::for_each(M->alias_begin(), M->alias_end(), recordGVSet);
+
+ // Assigned all GVs to merged clusters while balancing number of objects in
+ // each.
+ auto CompareClusters = [](const std::pair<unsigned, unsigned> &a,
+ const std::pair<unsigned, unsigned> &b) {
+ if (a.second || b.second)
+ return a.second > b.second;
+ else
+ return a.first > b.first;
+ };
+
+ std::priority_queue<std::pair<unsigned, unsigned>,
+ std::vector<std::pair<unsigned, unsigned>>,
+ decltype(CompareClusters)>
+ BalancinQueue(CompareClusters);
+ // Pre-populate priority queue with N slot blanks.
+ for (unsigned i = 0; i < N; ++i)
+ BalancinQueue.push(std::make_pair(i, 0));
+
+ typedef std::pair<unsigned, ClusterMapType::iterator> SortType;
+ SmallVector<SortType, 64> Sets;
+ SmallPtrSet<const GlobalValue *, 64> Visited;
+
+ // To guarantee determinism, we have to sort SCC according to size.
+ // When size is the same, use leader's name.
+ for (ClusterMapType::iterator I = GVtoClusterMap.begin(),
+ E = GVtoClusterMap.end(); I != E; ++I)
+ if (I->isLeader())
+ Sets.push_back(
+ std::make_pair(std::distance(GVtoClusterMap.member_begin(I),
+ GVtoClusterMap.member_end()), I));
+
+ std::sort(Sets.begin(), Sets.end(), [](const SortType &a, const SortType &b) {
+ if (a.first == b.first)
+ return a.second->getData()->getName() > b.second->getData()->getName();
+ else
+ return a.first > b.first;
+ });
+
+ for (auto &I : Sets) {
+ unsigned CurrentClusterID = BalancinQueue.top().first;
+ unsigned CurrentClusterSize = BalancinQueue.top().second;
+ BalancinQueue.pop();
+
+ DEBUG(dbgs() << "Root[" << CurrentClusterID << "] cluster_size(" << I.first
+ << ") ----> " << I.second->getData()->getName() << "\n");
+
+ for (ClusterMapType::member_iterator MI =
+ GVtoClusterMap.findLeader(I.second);
+ MI != GVtoClusterMap.member_end(); ++MI) {
+ if (!Visited.insert(*MI).second)
+ continue;
+ DEBUG(dbgs() << "----> " << (*MI)->getName()
+ << ((*MI)->hasLocalLinkage() ? " l " : " e ") << "\n");
+ Visited.insert(*MI);
+ ClusterIDMap[*MI] = CurrentClusterID;
+ CurrentClusterSize++;
+ }
+ // Add this set size to the number of entries in this cluster.
+ BalancinQueue.push(std::make_pair(CurrentClusterID, CurrentClusterSize));
+ }
+}
+
static void externalize(GlobalValue *GV) {
if (GV->hasLocalLinkage()) {
GV->setLinkage(GlobalValue::ExternalLinkage);
@@ -62,24 +209,46 @@
void llvm::SplitModule(
std::unique_ptr<Module> M, unsigned N,
- std::function<void(std::unique_ptr<Module> MPart)> ModuleCallback) {
- for (Function &F : *M)
- externalize(&F);
- for (GlobalVariable &GV : M->globals())
- externalize(&GV);
- for (GlobalAlias &GA : M->aliases())
- externalize(&GA);
+ std::function<void(std::unique_ptr<Module> MPart)> ModuleCallback,
+ bool PreserveLocals) {
+ if (!PreserveLocals) {
+ for (Function &F : *M)
+ externalize(&F);
+ for (GlobalVariable &GV : M->globals())
+ externalize(&GV);
+ for (GlobalAlias &GA : M->aliases())
+ externalize(&GA);
- // FIXME: We should be able to reuse M as the last partition instead of
- // cloning it.
- for (unsigned I = 0; I != N; ++I) {
- ValueToValueMapTy VMap;
- std::unique_ptr<Module> MPart(
- CloneModule(M.get(), VMap, [=](const GlobalValue *GV) {
- return isInPartition(GV, I, N);
- }));
- if (I != 0)
- MPart->setModuleInlineAsm("");
- ModuleCallback(std::move(MPart));
+ // FIXME: We should be able to reuse M as the last partition instead of
+ // cloning it.
+ for (unsigned I = 0; I != N; ++I) {
+ ValueToValueMapTy VMap;
+ std::unique_ptr<Module> MPart(
+ CloneModule(M.get(), VMap, [=](const GlobalValue *GV) {
+ return isInPartition(GV, I, N);
+ }));
+ if (I != 0)
+ MPart->setModuleInlineAsm("");
+ ModuleCallback(std::move(MPart));
+ }
+ } else {
+ // This performs splitting without a need for externalization, which might not
+ // always be possible.
+ ClusterIDMapType ClusterIDMap;
+ findPartitions(M.get(), ClusterIDMap, N);
+
+ for (unsigned I = 0; I < N; ++I) {
+ ValueToValueMapTy VMap;
+ std::unique_ptr<Module> MPart(
+ CloneModule(M.get(), VMap, [&](const GlobalValue *GV) {
+ if (ClusterIDMap.count(GV))
+ return (ClusterIDMap[GV] == I);
+ else
+ return isInPartition(GV, I, N);
+ }));
+ if (I != 0)
+ MPart->setModuleInlineAsm("");
+ ModuleCallback(std::move(MPart));
+ }
}
}