Index: include/llvm/IR/ModuleSummaryIndex.h =================================================================== --- include/llvm/IR/ModuleSummaryIndex.h +++ include/llvm/IR/ModuleSummaryIndex.h @@ -162,7 +162,7 @@ protected: /// GlobalValueSummary constructor. GlobalValueSummary(SummaryKind K, GVFlags Flags, std::vector Refs) - : Kind(K), Flags(Flags), RefEdgeList(std::move(Refs)) {} + : Kind(K), Flags(Flags), OriginalName(0), RefEdgeList(std::move(Refs)) {} public: virtual ~GlobalValueSummary() = default; @@ -528,6 +528,10 @@ // FIXME: Add bitcode read/write support for this field. std::map TypeIdMap; + /// Mapping from original ID to GUID. If original ID can map to multiple + /// GUIDs, it will be mapped to 0. + std::map OidGuidMap; + // YAML I/O support. friend yaml::MappingTraits; @@ -555,9 +559,17 @@ return GlobalValueMap.find(ValueGUID); } + /// Return the GUID for \p OriginalId in the OidGuidMap. + GlobalValue::GUID getGUIDFromOriginalID(GlobalValue::GUID OriginalID) const { + const auto I = OidGuidMap.find(OriginalID); + return I == OidGuidMap.end() ? 0 : I->second; + } + /// Add a global value summary for a value of the given name. void addGlobalValueSummary(StringRef ValueName, std::unique_ptr Summary) { + addOriginalName(GlobalValue::getGUID(ValueName), + Summary->getOriginalName()); GlobalValueMap[GlobalValue::getGUID(ValueName)].push_back( std::move(Summary)); } @@ -565,9 +577,21 @@ /// Add a global value summary for a value of the given GUID. void addGlobalValueSummary(GlobalValue::GUID ValueGUID, std::unique_ptr Summary) { + addOriginalName(ValueGUID, Summary->getOriginalName()); GlobalValueMap[ValueGUID].push_back(std::move(Summary)); } + /// Add an original name for the value of the given GUID. + void addOriginalName(GlobalValue::GUID ValueGUID, + GlobalValue::GUID OrigGUID) { + if (OrigGUID == 0 || ValueGUID == OrigGUID) + return; + if (OidGuidMap.count(OrigGUID) && OidGuidMap[OrigGUID] != ValueGUID) + OidGuidMap[OrigGUID] = 0; + else + OidGuidMap[OrigGUID] = ValueGUID; + } + /// Find the summary for global \p GUID in module \p ModuleId, or nullptr if /// not found. GlobalValueSummary *findSummaryInModule(GlobalValue::GUID ValueGUID, Index: include/llvm/ProfileData/SampleProfReader.h =================================================================== --- include/llvm/ProfileData/SampleProfReader.h +++ include/llvm/ProfileData/SampleProfReader.h @@ -283,7 +283,12 @@ /// \brief Return the samples collected for function \p F. FunctionSamples *getSamplesFor(const Function &F) { - return &Profiles[F.getName()]; + // The function name may have been updated by adding suffix. In sample + // profile, the function names are all stripped, so we need to strip + // the function name suffix before matching with profile. + if (Profiles.count(F.getName().split('.').first)) + return &Profiles[(F.getName().split('.').first)]; + return nullptr; } /// \brief Return all the profiles. Index: lib/Bitcode/Reader/BitcodeReader.cpp =================================================================== --- lib/Bitcode/Reader/BitcodeReader.cpp +++ lib/Bitcode/Reader/BitcodeReader.cpp @@ -4846,6 +4846,7 @@ // Keep around the last seen summary to be used when we see an optional // "OriginalName" attachement. GlobalValueSummary *LastSeenSummary = nullptr; + GlobalValue::GUID LastSeenGUID = 0; bool Combined = false; // We can expect to see any number of type ID information records before @@ -5014,6 +5015,7 @@ PendingTypeTestAssumeConstVCalls.clear(); PendingTypeCheckedLoadConstVCalls.clear(); LastSeenSummary = FS.get(); + LastSeenGUID = GUID; FS->setModulePath(ModuleIdMap[ModuleId]); TheIndex.addGlobalValueSummary(GUID, std::move(FS)); Combined = true; @@ -5040,6 +5042,7 @@ AS->setAliasee(AliaseeInModule); GlobalValue::GUID GUID = getGUIDFromValueId(ValueID).first; + LastSeenGUID = GUID; TheIndex.addGlobalValueSummary(GUID, std::move(AS)); Combined = true; break; @@ -5056,6 +5059,7 @@ LastSeenSummary = FS.get(); FS->setModulePath(ModuleIdMap[ModuleId]); GlobalValue::GUID GUID = getGUIDFromValueId(ValueID).first; + LastSeenGUID = GUID; TheIndex.addGlobalValueSummary(GUID, std::move(FS)); Combined = true; break; @@ -5066,8 +5070,10 @@ if (!LastSeenSummary) return error("Name attachment that does not follow a combined record"); LastSeenSummary->setOriginalName(OriginalName); + TheIndex.addOriginalName(LastSeenGUID, OriginalName); // Reset the LastSeenSummary LastSeenSummary = nullptr; + LastSeenGUID = 0; break; } case bitc::FS_TYPE_TESTS: { Index: lib/Bitcode/Writer/BitcodeWriter.cpp =================================================================== --- lib/Bitcode/Writer/BitcodeWriter.cpp +++ lib/Bitcode/Writer/BitcodeWriter.cpp @@ -3708,9 +3708,16 @@ for (auto &EI : FS->calls()) { // If this GUID doesn't have a value id, it doesn't have a function // summary and we don't need to record any calls to it. - if (!hasValueId(EI.first.getGUID())) - continue; - NameVals.push_back(getValueId(EI.first.getGUID())); + GlobalValue::GUID GUID = EI.first.getGUID(); + if (!hasValueId(GUID)) { + // For SamplePGO, the indirect call targets for local functions will + // have its original name annotated in profile. We try to find the + // corresponding PGOFuncName as the GUID. + GUID = Index.getGUIDFromOriginalID(GUID); + if (GUID == 0 || !hasValueId(GUID)) + continue; + } + NameVals.push_back(getValueId(GUID)); if (HasProfileData) NameVals.push_back(static_cast(EI.second.Hotness)); } Index: lib/ProfileData/InstrProf.cpp =================================================================== --- lib/ProfileData/InstrProf.cpp +++ lib/ProfileData/InstrProf.cpp @@ -298,6 +298,17 @@ const std::string &PGOFuncName = getPGOFuncName(F, InLTO); addFuncName(PGOFuncName); MD5FuncMap.emplace_back(Function::getGUID(PGOFuncName), &F); + // In ThinLTO, local function may have been promoted to global and have + // suffix added to the function name. We need to add the stripped function + // name to the symbol table so that we can find a match from profile. + if (InLTO) { + auto pos = PGOFuncName.find('.'); + if (pos != std::string::npos) { + const std::string &OtherFuncName = PGOFuncName.substr(0, pos); + addFuncName(OtherFuncName); + MD5FuncMap.emplace_back(Function::getGUID(OtherFuncName), &F); + } + } } finalizeSymtab(); Index: lib/Transforms/IPO/FunctionImport.cpp =================================================================== --- lib/Transforms/IPO/FunctionImport.cpp +++ lib/Transforms/IPO/FunctionImport.cpp @@ -197,6 +197,15 @@ auto GUID = Edge.first.getGUID(); DEBUG(dbgs() << " edge -> " << GUID << " Threshold:" << Threshold << "\n"); + if (Index.findGlobalValueSummaryList(GUID) == Index.end()) { + // For SamplePGO, the indirect call targets for local functions will + // have its original name annotated in profile. We try to find the + // corresponding PGOFuncName as the GUID. + GUID = Index.getGUIDFromOriginalID(GUID); + if (GUID == 0) + continue; + } + if (DefinedGVSummaries.count(GUID)) { DEBUG(dbgs() << "ignored! Target already in destination module.\n"); continue; Index: lib/Transforms/IPO/SampleProfile.cpp =================================================================== --- lib/Transforms/IPO/SampleProfile.cpp +++ lib/Transforms/IPO/SampleProfile.cpp @@ -1383,7 +1383,7 @@ bool SampleProfileLoader::runOnFunction(Function &F) { F.setEntryCount(0); Samples = Reader->getSamplesFor(F); - if (!Samples->empty()) + if (Samples && !Samples->empty()) return emitAnnotations(F); return false; } Index: test/Transforms/PGOProfile/Inputs/thinlto_samplepgo_icp.ll =================================================================== --- /dev/null +++ test/Transforms/PGOProfile/Inputs/thinlto_samplepgo_icp.ll @@ -0,0 +1,27 @@ +target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-unknown-linux-gnu" + +@fptr = external local_unnamed_addr global void ()*, align 8 + +; Function Attrs: norecurse nounwind uwtable +define void @_Z6updatei(i32 %i) local_unnamed_addr #0 { +entry: + store void ()* @_ZL3foov, void ()** @fptr, align 8 + ret void +} + +; Function Attrs: norecurse nounwind readnone uwtable +define internal void @_ZL3foov() #1 { +entry: + ret void +} + +!llvm.dbg.cu = !{!0} +!llvm.module.flags = !{!3} +!llvm.ident = !{!31} + +!0 = distinct !DICompileUnit(language: DW_LANG_C_plus_plus, file: !1, producer: "clang version 5.0.0 (trunk 297016)", isOptimized: true, runtimeVersion: 0, emissionKind: NoDebug, enums: !2) +!1 = !DIFile(filename: "b.cc", directory: "/ssd/llvm/abc/small") +!2 = !{} +!3 = !{i32 2, !"Debug Info Version", i32 3} +!31 = !{!"clang version 5.0.0 (trunk 297016)"} Index: test/Transforms/PGOProfile/thinlto_samplepgo_icp.ll =================================================================== --- /dev/null +++ test/Transforms/PGOProfile/thinlto_samplepgo_icp.ll @@ -0,0 +1,63 @@ +; Do setup work for all below tests: generate bitcode and combined index +; RUN: opt -module-summary %s -o %t.bc +; RUN: opt -module-summary %p/Inputs/thinlto_samplepgo_icp.ll -o %t2.bc +; RUN: llvm-lto -thinlto -o %t3 %t.bc %t2.bc + +; Checks if calls to static target functions are properly imported and promoted +; by ICP. Note that the GUID in the profile is from the oroginal name. +; RUN: opt -function-import -summary-file %t3.thinlto.bc %t.bc -o %t4.bc -print-imports 2>&1 | FileCheck %s --check-prefix=IMPORTS +; IMPORTS: Import _ZL3foov.llvm.0 +; RUN: opt %t4.bc -icp-lto -pgo-icall-prom -S -icp-count-threshold=1 | FileCheck %s --check-prefix=ICALL-PROM + +target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" +target triple = "x86_64-unknown-linux-gnu" + +@fptr = local_unnamed_addr global void ()* null, align 8 + +; Function Attrs: norecurse uwtable +define i32 @main() local_unnamed_addr #0 !prof !34 { +entry: + %0 = load void ()*, void ()** @fptr, align 8 +; ICALL-PROM: br i1 %{{[0-9]+}}, label %if.true.direct_targ, label %if.false.orig_indirect + tail call void %0(), !prof !40 + ret i32 0 +} + +!llvm.dbg.cu = !{!0} +!llvm.module.flags = !{!3,!4} +!llvm.ident = !{!31} + +!0 = distinct !DICompileUnit(language: DW_LANG_C_plus_plus, file: !1, producer: "clang version 5.0.0 (trunk 297016)", isOptimized: true, runtimeVersion: 0, emissionKind: NoDebug, enums: !2) +!1 = !DIFile(filename: "main.cc", directory: ".") +!2 = !{} +!3 = !{i32 2, !"Debug Info Version", i32 3} +!4 = !{i32 1, !"ProfileSummary", !5} +!5 = !{!6, !7, !8, !9, !10, !11, !12, !13} +!6 = !{!"ProfileFormat", !"SampleProfile"} +!7 = !{!"TotalCount", i64 3003} +!8 = !{!"MaxCount", i64 3000} +!9 = !{!"MaxInternalCount", i64 0} +!10 = !{!"MaxFunctionCount", i64 0} +!11 = !{!"NumCounts", i64 3} +!12 = !{!"NumFunctions", i64 1} +!13 = !{!"DetailedSummary", !14} +!14 = !{!15, !16, !17, !18, !19, !20, !20, !21, !21, !22, !23, !24, !25, !26, !27, !28, !29, !30} +!15 = !{i32 10000, i64 3000, i32 1} +!16 = !{i32 100000, i64 3000, i32 1} +!17 = !{i32 200000, i64 3000, i32 1} +!18 = !{i32 300000, i64 3000, i32 1} +!19 = !{i32 400000, i64 3000, i32 1} +!20 = !{i32 500000, i64 3000, i32 1} +!21 = !{i32 600000, i64 3000, i32 1} +!22 = !{i32 700000, i64 3000, i32 1} +!23 = !{i32 800000, i64 3000, i32 1} +!24 = !{i32 900000, i64 3000, i32 1} +!25 = !{i32 950000, i64 3000, i32 1} +!26 = !{i32 990000, i64 3000, i32 1} +!27 = !{i32 999000, i64 3000, i32 1} +!28 = !{i32 999900, i64 2, i32 2} +!29 = !{i32 999990, i64 2, i32 2} +!30 = !{i32 999999, i64 2, i32 2} +!31 = !{!"clang version 5.0.0 (trunk 297016)"} +!34 = !{!"function_entry_count", i64 1} +!40 = !{!"VP", i32 0, i64 3000, i64 -8789629626369651636, i64 3000}