-//=-- InstrProf.cpp - Instrumented profiling format support -----------------=//
+//===- InstrProf.cpp - Instrumented profiling format support --------------===//
//
// The LLVM Compiler Infrastructure
//
//===----------------------------------------------------------------------===//
#include "llvm/ProfileData/InstrProf.h"
+#include "llvm/ADT/ArrayRef.h"
+#include "llvm/ADT/SmallString.h"
+#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/StringExtras.h"
+#include "llvm/ADT/StringRef.h"
+#include "llvm/ADT/Triple.h"
+#include "llvm/IR/Constant.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/Function.h"
+#include "llvm/IR/GlobalValue.h"
#include "llvm/IR/GlobalVariable.h"
+#include "llvm/IR/Instruction.h"
+#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/MDBuilder.h"
+#include "llvm/IR/Metadata.h"
#include "llvm/IR/Module.h"
+#include "llvm/IR/Type.h"
+#include "llvm/Support/Casting.h"
+#include "llvm/Support/CommandLine.h"
+#include "llvm/Support/Compiler.h"
#include "llvm/Support/Compression.h"
+#include "llvm/Support/Endian.h"
+#include "llvm/Support/Error.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/LEB128.h"
#include "llvm/Support/ManagedStatic.h"
+#include "llvm/Support/MathExtras.h"
+#include "llvm/Support/Path.h"
+#include "llvm/Support/SwapByteOrder.h"
+#include <algorithm>
+#include <cassert>
+#include <cstddef>
+#include <cstdint>
+#include <cstring>
+#include <memory>
+#include <string>
+#include <system_error>
+#include <utility>
+#include <vector>
using namespace llvm;
+static cl::opt<bool> StaticFuncFullModulePrefix(
+ "static-func-full-module-prefix", cl::init(true), cl::Hidden,
+ cl::desc("Use full module build paths in the profile counter names for "
+ "static functions."));
+
+// This option is tailored to users that have different top-level directory in
+// profile-gen and profile-use compilation. Users need to specific the number
+// of levels to strip. A value larger than the number of directories in the
+// source file will strip all the directory names and only leave the basename.
+//
+// Note current ThinLTO module importing for the indirect-calls assumes
+// the source directory name not being stripped. A non-zero option value here
+// can potentially prevent some inter-module indirect-call-promotions.
+static cl::opt<unsigned> StaticFuncStripDirNamePrefix(
+ "static-func-strip-dirname-prefix", cl::init(0), cl::Hidden,
+ cl::desc("Strip specified level of directory name from source path in "
+ "the profile counter name for static functions."));
+
+static std::string getInstrProfErrString(instrprof_error Err) {
+ switch (Err) {
+ case instrprof_error::success:
+ return "Success";
+ case instrprof_error::eof:
+ return "End of File";
+ case instrprof_error::unrecognized_format:
+ return "Unrecognized instrumentation profile encoding format";
+ case instrprof_error::bad_magic:
+ return "Invalid instrumentation profile data (bad magic)";
+ case instrprof_error::bad_header:
+ return "Invalid instrumentation profile data (file header is corrupt)";
+ case instrprof_error::unsupported_version:
+ return "Unsupported instrumentation profile format version";
+ case instrprof_error::unsupported_hash_type:
+ return "Unsupported instrumentation profile hash type";
+ case instrprof_error::too_large:
+ return "Too much profile data";
+ case instrprof_error::truncated:
+ return "Truncated profile data";
+ case instrprof_error::malformed:
+ return "Malformed instrumentation profile data";
+ case instrprof_error::unknown_function:
+ return "No profile data available for function";
+ case instrprof_error::hash_mismatch:
+ return "Function control flow change detected (hash mismatch)";
+ case instrprof_error::count_mismatch:
+ return "Function basic block count change detected (counter mismatch)";
+ case instrprof_error::counter_overflow:
+ return "Counter overflow";
+ case instrprof_error::value_site_count_mismatch:
+ return "Function value site count change detected (counter mismatch)";
+ case instrprof_error::compress_failed:
+ return "Failed to compress data (zlib)";
+ case instrprof_error::uncompress_failed:
+ return "Failed to uncompress data (zlib)";
+ case instrprof_error::empty_raw_profile:
+ return "Empty raw profile file";
+ case instrprof_error::zlib_unavailable:
+ return "Profile uses zlib compression but the profile reader was built without zlib support";
+ }
+ llvm_unreachable("A value of instrprof_error has no message.");
+}
+
namespace {
+
+// FIXME: This class is only here to support the transition to llvm::Error. It
+// will be removed once this transition is complete. Clients should prefer to
+// deal with the Error value directly, rather than converting to error_code.
class InstrProfErrorCategoryType : public std::error_category {
- const char *name() const LLVM_NOEXCEPT override { return "llvm.instrprof"; }
+ const char *name() const noexcept override { return "llvm.instrprof"; }
+
std::string message(int IE) const override {
- instrprof_error E = static_cast<instrprof_error>(IE);
- switch (E) {
- case instrprof_error::success:
- return "Success";
- case instrprof_error::eof:
- return "End of File";
- case instrprof_error::unrecognized_format:
- return "Unrecognized instrumentation profile encoding format";
- case instrprof_error::bad_magic:
- return "Invalid instrumentation profile data (bad magic)";
- case instrprof_error::bad_header:
- return "Invalid instrumentation profile data (file header is corrupt)";
- case instrprof_error::unsupported_version:
- return "Unsupported instrumentation profile format version";
- case instrprof_error::unsupported_hash_type:
- return "Unsupported instrumentation profile hash type";
- case instrprof_error::too_large:
- return "Too much profile data";
- case instrprof_error::truncated:
- return "Truncated profile data";
- case instrprof_error::malformed:
- return "Malformed instrumentation profile data";
- case instrprof_error::unknown_function:
- return "No profile data available for function";
- case instrprof_error::hash_mismatch:
- return "Function control flow change detected (hash mismatch)";
- case instrprof_error::count_mismatch:
- return "Function basic block count change detected (counter mismatch)";
- case instrprof_error::counter_overflow:
- return "Counter overflow";
- case instrprof_error::value_site_count_mismatch:
- return "Function value site count change detected (counter mismatch)";
- }
- llvm_unreachable("A value of instrprof_error has no message.");
+ return getInstrProfErrString(static_cast<instrprof_error>(IE));
}
};
+
} // end anonymous namespace
static ManagedStatic<InstrProfErrorCategoryType> ErrorCategory;
return *ErrorCategory;
}
+namespace {
+
+const char *InstrProfSectNameCommon[] = {
+#define INSTR_PROF_SECT_ENTRY(Kind, SectNameCommon, SectNameCoff, Prefix) \
+ SectNameCommon,
+#include "llvm/ProfileData/InstrProfData.inc"
+};
+
+const char *InstrProfSectNameCoff[] = {
+#define INSTR_PROF_SECT_ENTRY(Kind, SectNameCommon, SectNameCoff, Prefix) \
+ SectNameCoff,
+#include "llvm/ProfileData/InstrProfData.inc"
+};
+
+const char *InstrProfSectNamePrefix[] = {
+#define INSTR_PROF_SECT_ENTRY(Kind, SectNameCommon, SectNameCoff, Prefix) \
+ Prefix,
+#include "llvm/ProfileData/InstrProfData.inc"
+};
+
+} // namespace
+
namespace llvm {
+std::string getInstrProfSectionName(InstrProfSectKind IPSK,
+ Triple::ObjectFormatType OF,
+ bool AddSegmentInfo) {
+ std::string SectName;
+
+ if (OF == Triple::MachO && AddSegmentInfo)
+ SectName = InstrProfSectNamePrefix[IPSK];
+
+ if (OF == Triple::COFF)
+ SectName += InstrProfSectNameCoff[IPSK];
+ else
+ SectName += InstrProfSectNameCommon[IPSK];
+
+ if (OF == Triple::MachO && IPSK == IPSK_data && AddSegmentInfo)
+ SectName += ",regular,live_support";
+
+ return SectName;
+}
+
+void SoftInstrProfErrors::addError(instrprof_error IE) {
+ if (IE == instrprof_error::success)
+ return;
+
+ if (FirstError == instrprof_error::success)
+ FirstError = IE;
+
+ switch (IE) {
+ case instrprof_error::hash_mismatch:
+ ++NumHashMismatches;
+ break;
+ case instrprof_error::count_mismatch:
+ ++NumCountMismatches;
+ break;
+ case instrprof_error::counter_overflow:
+ ++NumCounterOverflows;
+ break;
+ case instrprof_error::value_site_count_mismatch:
+ ++NumValueSiteCountMismatches;
+ break;
+ default:
+ llvm_unreachable("Not a soft error");
+ }
+}
+
+std::string InstrProfError::message() const {
+ return getInstrProfErrString(Err);
+}
+
+char InstrProfError::ID = 0;
+
std::string getPGOFuncName(StringRef RawFuncName,
GlobalValue::LinkageTypes Linkage,
StringRef FileName,
uint64_t Version LLVM_ATTRIBUTE_UNUSED) {
+ return GlobalValue::getGlobalIdentifier(RawFuncName, Linkage, FileName);
+}
+
+// Strip NumPrefix level of directory name from PathNameStr. If the number of
+// directory separators is less than NumPrefix, strip all the directories and
+// leave base file name only.
+static StringRef stripDirPrefix(StringRef PathNameStr, uint32_t NumPrefix) {
+ uint32_t Count = NumPrefix;
+ uint32_t Pos = 0, LastPos = 0;
+ for (auto & CI : PathNameStr) {
+ ++Pos;
+ if (llvm::sys::path::is_separator(CI)) {
+ LastPos = Pos;
+ --Count;
+ }
+ if (Count == 0)
+ break;
+ }
+ return PathNameStr.substr(LastPos);
+}
+
+// Return the PGOFuncName. This function has some special handling when called
+// in LTO optimization. The following only applies when calling in LTO passes
+// (when \c InLTO is true): LTO's internalization privatizes many global linkage
+// symbols. This happens after value profile annotation, but those internal
+// linkage functions should not have a source prefix.
+// Additionally, for ThinLTO mode, exported internal functions are promoted
+// and renamed. We need to ensure that the original internal PGO name is
+// used when computing the GUID that is compared against the profiled GUIDs.
+// To differentiate compiler generated internal symbols from original ones,
+// PGOFuncName meta data are created and attached to the original internal
+// symbols in the value profile annotation step
+// (PGOUseFunc::annotateIndirectCallSites). If a symbol does not have the meta
+// data, its original linkage must be non-internal.
+std::string getPGOFuncName(const Function &F, bool InLTO, uint64_t Version) {
+ if (!InLTO) {
+ StringRef FileName = (StaticFuncFullModulePrefix
+ ? F.getParent()->getName()
+ : sys::path::filename(F.getParent()->getName()));
+ if (StaticFuncFullModulePrefix && StaticFuncStripDirNamePrefix != 0)
+ FileName = stripDirPrefix(FileName, StaticFuncStripDirNamePrefix);
+ return getPGOFuncName(F.getName(), F.getLinkage(), FileName, Version);
+ }
- // Function names may be prefixed with a binary '1' to indicate
- // that the backend should not modify the symbols due to any platform
- // naming convention. Do not include that '1' in the PGO profile name.
- if (RawFuncName[0] == '\1')
- RawFuncName = RawFuncName.substr(1);
-
- std::string FuncName = RawFuncName;
- if (llvm::GlobalValue::isLocalLinkage(Linkage)) {
- // For local symbols, prepend the main file name to distinguish them.
- // Do not include the full path in the file name since there's no guarantee
- // that it will stay the same, e.g., if the files are checked out from
- // version control in different locations.
- if (FileName.empty())
- FuncName = FuncName.insert(0, "<unknown>:");
- else
- FuncName = FuncName.insert(0, FileName.str() + ":");
+ // In LTO mode (when InLTO is true), first check if there is a meta data.
+ if (MDNode *MD = getPGOFuncNameMetadata(F)) {
+ StringRef S = cast<MDString>(MD->getOperand(0))->getString();
+ return S.str();
}
- return FuncName;
-}
-std::string getPGOFuncName(const Function &F, uint64_t Version) {
- return getPGOFuncName(F.getName(), F.getLinkage(), F.getParent()->getName(),
- Version);
+ // If there is no meta data, the function must be a global before the value
+ // profile annotation pass. Its current linkage may be internal if it is
+ // internalized in LTO mode.
+ return getPGOFuncName(F.getName(), GlobalValue::ExternalLinkage, "");
}
StringRef getFuncNameWithoutPrefix(StringRef PGOFuncName, StringRef FileName) {
// \p FuncName is the string used as profile lookup key for the function. A
// symbol is created to hold the name. Return the legalized symbol name.
-static std::string getPGOFuncNameVarName(StringRef FuncName,
- GlobalValue::LinkageTypes Linkage) {
+std::string getPGOFuncNameVarName(StringRef FuncName,
+ GlobalValue::LinkageTypes Linkage) {
std::string VarName = getInstrProfNameVarPrefix();
VarName += FuncName;
return VarName;
// Now fix up illegal chars in local VarName that may upset the assembler.
- const char *InvalidChars = "-:<>\"'";
+ const char *InvalidChars = "-:<>/\"'";
size_t found = VarName.find_first_of(InvalidChars);
while (found != std::string::npos) {
VarName[found] = '_';
GlobalVariable *createPGOFuncNameVar(Module &M,
GlobalValue::LinkageTypes Linkage,
- StringRef FuncName) {
-
+ StringRef PGOFuncName) {
// We generally want to match the function's linkage, but available_externally
// and extern_weak both have the wrong semantics, and anything that doesn't
// need to link across compilation units doesn't need to be visible at all.
Linkage == GlobalValue::ExternalLinkage)
Linkage = GlobalValue::PrivateLinkage;
- auto *Value = ConstantDataArray::getString(M.getContext(), FuncName, false);
+ auto *Value =
+ ConstantDataArray::getString(M.getContext(), PGOFuncName, false);
auto FuncNameVar =
new GlobalVariable(M, Value->getType(), true, Linkage, Value,
- getPGOFuncNameVarName(FuncName, Linkage));
+ getPGOFuncNameVarName(PGOFuncName, Linkage));
// Hide the symbol so that we correctly get a copy for each executable.
if (!GlobalValue::isLocalLinkage(FuncNameVar->getLinkage()))
return FuncNameVar;
}
-GlobalVariable *createPGOFuncNameVar(Function &F, StringRef FuncName) {
- return createPGOFuncNameVar(*F.getParent(), F.getLinkage(), FuncName);
+GlobalVariable *createPGOFuncNameVar(Function &F, StringRef PGOFuncName) {
+ return createPGOFuncNameVar(*F.getParent(), F.getLinkage(), PGOFuncName);
}
-void InstrProfSymtab::create(const Module &M) {
- for (const Function &F : M)
- addFuncName(getPGOFuncName(F));
+Error InstrProfSymtab::create(Module &M, bool InLTO) {
+ for (Function &F : M) {
+ // Function may not have a name: like using asm("") to overwrite the name.
+ // Ignore in this case.
+ if (!F.hasName())
+ continue;
+ const std::string &PGOFuncName = getPGOFuncName(F, InLTO);
+ if (Error E = addFuncName(PGOFuncName))
+ return E;
+ 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);
+ if (Error E = addFuncName(OtherFuncName))
+ return E;
+ MD5FuncMap.emplace_back(Function::getGUID(OtherFuncName), &F);
+ }
+ }
+ }
+ Sorted = false;
+ finalizeSymtab();
+ return Error::success();
+}
+uint64_t InstrProfSymtab::getFunctionHashFromAddress(uint64_t Address) {
finalizeSymtab();
+ auto Result =
+ std::lower_bound(AddrToMD5Map.begin(), AddrToMD5Map.end(), Address,
+ [](const std::pair<uint64_t, uint64_t> &LHS,
+ uint64_t RHS) { return LHS.first < RHS; });
+ // Raw function pointer collected by value profiler may be from
+ // external functions that are not instrumented. They won't have
+ // mapping data to be used by the deserializer. Force the value to
+ // be 0 in this case.
+ if (Result != AddrToMD5Map.end() && Result->first == Address)
+ return (uint64_t)Result->second;
+ return 0;
}
-int collectPGOFuncNameStrings(const std::vector<std::string> &NameStrs,
- bool doCompression, std::string &Result) {
+Error collectPGOFuncNameStrings(ArrayRef<std::string> NameStrs,
+ bool doCompression, std::string &Result) {
+ assert(!NameStrs.empty() && "No name data to emit");
+
uint8_t Header[16], *P = Header;
std::string UncompressedNameStrings =
- join(NameStrs.begin(), NameStrs.end(), StringRef(" "));
+ join(NameStrs.begin(), NameStrs.end(), getInstrProfNameSeparator());
+
+ assert(StringRef(UncompressedNameStrings)
+ .count(getInstrProfNameSeparator()) == (NameStrs.size() - 1) &&
+ "PGO name is invalid (contains separator token)");
unsigned EncLen = encodeULEB128(UncompressedNameStrings.length(), P);
P += EncLen;
- auto WriteStringToResult = [&](size_t CompressedLen,
- const std::string &InputStr) {
+ auto WriteStringToResult = [&](size_t CompressedLen, StringRef InputStr) {
EncLen = encodeULEB128(CompressedLen, P);
P += EncLen;
char *HeaderStr = reinterpret_cast<char *>(&Header[0]);
unsigned HeaderLen = P - &Header[0];
Result.append(HeaderStr, HeaderLen);
Result += InputStr;
- return 0;
+ return Error::success();
};
- if (!doCompression)
+ if (!doCompression) {
return WriteStringToResult(0, UncompressedNameStrings);
+ }
- SmallVector<char, 128> CompressedNameStrings;
- zlib::Status Success =
- zlib::compress(StringRef(UncompressedNameStrings), CompressedNameStrings,
- zlib::BestSizeCompression);
-
- if (Success != zlib::StatusOK)
- return 1;
+ SmallString<128> CompressedNameStrings;
+ Error E = zlib::compress(StringRef(UncompressedNameStrings),
+ CompressedNameStrings, zlib::BestSizeCompression);
+ if (E) {
+ consumeError(std::move(E));
+ return make_error<InstrProfError>(instrprof_error::compress_failed);
+ }
- return WriteStringToResult(
- CompressedNameStrings.size(),
- std::string(CompressedNameStrings.data(), CompressedNameStrings.size()));
+ return WriteStringToResult(CompressedNameStrings.size(),
+ CompressedNameStrings);
}
-StringRef getPGOFuncNameInitializer(GlobalVariable *NameVar) {
+StringRef getPGOFuncNameVarInitializer(GlobalVariable *NameVar) {
auto *Arr = cast<ConstantDataArray>(NameVar->getInitializer());
StringRef NameStr =
Arr->isCString() ? Arr->getAsCString() : Arr->getAsString();
return NameStr;
}
-int collectPGOFuncNameStrings(const std::vector<GlobalVariable *> &NameVars,
- std::string &Result, bool doCompression) {
+Error collectPGOFuncNameStrings(ArrayRef<GlobalVariable *> NameVars,
+ std::string &Result, bool doCompression) {
std::vector<std::string> NameStrs;
for (auto *NameVar : NameVars) {
- NameStrs.push_back(getPGOFuncNameInitializer(NameVar));
+ NameStrs.push_back(getPGOFuncNameVarInitializer(NameVar));
}
return collectPGOFuncNameStrings(
NameStrs, zlib::isAvailable() && doCompression, Result);
}
-int readPGOFuncNameStrings(StringRef NameStrings, InstrProfSymtab &Symtab) {
+Error readPGOFuncNameStrings(StringRef NameStrings, InstrProfSymtab &Symtab) {
const uint8_t *P = reinterpret_cast<const uint8_t *>(NameStrings.data());
const uint8_t *EndP = reinterpret_cast<const uint8_t *>(NameStrings.data() +
NameStrings.size());
SmallString<128> UncompressedNameStrings;
StringRef NameStrings;
if (isCompressed) {
+ if (!llvm::zlib::isAvailable())
+ return make_error<InstrProfError>(instrprof_error::zlib_unavailable);
+
StringRef CompressedNameStrings(reinterpret_cast<const char *>(P),
CompressedSize);
- if (zlib::uncompress(CompressedNameStrings, UncompressedNameStrings,
- UncompressedSize) != zlib::StatusOK)
- return 1;
+ if (Error E =
+ zlib::uncompress(CompressedNameStrings, UncompressedNameStrings,
+ UncompressedSize)) {
+ consumeError(std::move(E));
+ return make_error<InstrProfError>(instrprof_error::uncompress_failed);
+ }
P += CompressedSize;
NameStrings = StringRef(UncompressedNameStrings.data(),
UncompressedNameStrings.size());
}
// Now parse the name strings.
SmallVector<StringRef, 0> Names;
- NameStrings.split(Names, ' ');
+ NameStrings.split(Names, getInstrProfNameSeparator());
for (StringRef &Name : Names)
- Symtab.addFuncName(Name);
+ if (Error E = Symtab.addFuncName(Name))
+ return E;
while (P < EndP && *P == 0)
P++;
}
- Symtab.finalizeSymtab();
- return 0;
+ return Error::success();
}
-instrprof_error InstrProfValueSiteRecord::merge(InstrProfValueSiteRecord &Input,
- uint64_t Weight) {
+void InstrProfValueSiteRecord::merge(InstrProfValueSiteRecord &Input,
+ uint64_t Weight,
+ function_ref<void(instrprof_error)> Warn) {
this->sortByTargetValues();
Input.sortByTargetValues();
auto I = ValueData.begin();
auto IE = ValueData.end();
- instrprof_error Result = instrprof_error::success;
for (auto J = Input.ValueData.begin(), JE = Input.ValueData.end(); J != JE;
++J) {
while (I != IE && I->Value < J->Value)
bool Overflowed;
I->Count = SaturatingMultiplyAdd(J->Count, Weight, I->Count, &Overflowed);
if (Overflowed)
- Result = instrprof_error::counter_overflow;
+ Warn(instrprof_error::counter_overflow);
++I;
continue;
}
ValueData.insert(I, *J);
}
- return Result;
}
-instrprof_error InstrProfValueSiteRecord::scale(uint64_t Weight) {
- instrprof_error Result = instrprof_error::success;
+void InstrProfValueSiteRecord::scale(uint64_t Weight,
+ function_ref<void(instrprof_error)> Warn) {
for (auto I = ValueData.begin(), IE = ValueData.end(); I != IE; ++I) {
bool Overflowed;
I->Count = SaturatingMultiply(I->Count, Weight, &Overflowed);
if (Overflowed)
- Result = instrprof_error::counter_overflow;
+ Warn(instrprof_error::counter_overflow);
}
- return Result;
}
// Merge Value Profile data from Src record to this record for ValueKind.
// Scale merged value counts by \p Weight.
-instrprof_error InstrProfRecord::mergeValueProfData(uint32_t ValueKind,
- InstrProfRecord &Src,
- uint64_t Weight) {
+void InstrProfRecord::mergeValueProfData(
+ uint32_t ValueKind, InstrProfRecord &Src, uint64_t Weight,
+ function_ref<void(instrprof_error)> Warn) {
uint32_t ThisNumValueSites = getNumValueSites(ValueKind);
uint32_t OtherNumValueSites = Src.getNumValueSites(ValueKind);
- if (ThisNumValueSites != OtherNumValueSites)
- return instrprof_error::value_site_count_mismatch;
+ if (ThisNumValueSites != OtherNumValueSites) {
+ Warn(instrprof_error::value_site_count_mismatch);
+ return;
+ }
+ if (!ThisNumValueSites)
+ return;
std::vector<InstrProfValueSiteRecord> &ThisSiteRecords =
- getValueSitesForKind(ValueKind);
- std::vector<InstrProfValueSiteRecord> &OtherSiteRecords =
+ getOrCreateValueSitesForKind(ValueKind);
+ MutableArrayRef<InstrProfValueSiteRecord> OtherSiteRecords =
Src.getValueSitesForKind(ValueKind);
- instrprof_error Result = instrprof_error::success;
for (uint32_t I = 0; I < ThisNumValueSites; I++)
- MergeResult(Result, ThisSiteRecords[I].merge(OtherSiteRecords[I], Weight));
- return Result;
+ ThisSiteRecords[I].merge(OtherSiteRecords[I], Weight, Warn);
}
-instrprof_error InstrProfRecord::merge(InstrProfRecord &Other,
- uint64_t Weight) {
+void InstrProfRecord::merge(InstrProfRecord &Other, uint64_t Weight,
+ function_ref<void(instrprof_error)> Warn) {
// If the number of counters doesn't match we either have bad data
// or a hash collision.
- if (Counts.size() != Other.Counts.size())
- return instrprof_error::count_mismatch;
-
- instrprof_error Result = instrprof_error::success;
+ if (Counts.size() != Other.Counts.size()) {
+ Warn(instrprof_error::count_mismatch);
+ return;
+ }
for (size_t I = 0, E = Other.Counts.size(); I < E; ++I) {
bool Overflowed;
Counts[I] =
SaturatingMultiplyAdd(Other.Counts[I], Weight, Counts[I], &Overflowed);
if (Overflowed)
- Result = instrprof_error::counter_overflow;
+ Warn(instrprof_error::counter_overflow);
}
for (uint32_t Kind = IPVK_First; Kind <= IPVK_Last; ++Kind)
- MergeResult(Result, mergeValueProfData(Kind, Other, Weight));
-
- return Result;
+ mergeValueProfData(Kind, Other, Weight, Warn);
}
-instrprof_error InstrProfRecord::scaleValueProfData(uint32_t ValueKind,
- uint64_t Weight) {
- uint32_t ThisNumValueSites = getNumValueSites(ValueKind);
- std::vector<InstrProfValueSiteRecord> &ThisSiteRecords =
- getValueSitesForKind(ValueKind);
- instrprof_error Result = instrprof_error::success;
- for (uint32_t I = 0; I < ThisNumValueSites; I++)
- MergeResult(Result, ThisSiteRecords[I].scale(Weight));
- return Result;
+void InstrProfRecord::scaleValueProfData(
+ uint32_t ValueKind, uint64_t Weight,
+ function_ref<void(instrprof_error)> Warn) {
+ for (auto &R : getValueSitesForKind(ValueKind))
+ R.scale(Weight, Warn);
}
-instrprof_error InstrProfRecord::scale(uint64_t Weight) {
- instrprof_error Result = instrprof_error::success;
+void InstrProfRecord::scale(uint64_t Weight,
+ function_ref<void(instrprof_error)> Warn) {
for (auto &Count : this->Counts) {
bool Overflowed;
Count = SaturatingMultiply(Count, Weight, &Overflowed);
- if (Overflowed && Result == instrprof_error::success) {
- Result = instrprof_error::counter_overflow;
- }
+ if (Overflowed)
+ Warn(instrprof_error::counter_overflow);
}
for (uint32_t Kind = IPVK_First; Kind <= IPVK_Last; ++Kind)
- MergeResult(Result, scaleValueProfData(Kind, Weight));
-
- return Result;
+ scaleValueProfData(Kind, Weight, Warn);
}
// Map indirect call target name hash to name string.
uint64_t InstrProfRecord::remapValue(uint64_t Value, uint32_t ValueKind,
- ValueMapType *ValueMap) {
- if (!ValueMap)
+ InstrProfSymtab *SymTab) {
+ if (!SymTab)
return Value;
- switch (ValueKind) {
- case IPVK_IndirectCallTarget: {
- auto Result =
- std::lower_bound(ValueMap->begin(), ValueMap->end(), Value,
- [](const std::pair<uint64_t, uint64_t> &LHS,
- uint64_t RHS) { return LHS.first < RHS; });
- if (Result != ValueMap->end())
- Value = (uint64_t)Result->second;
- break;
- }
- }
+
+ if (ValueKind == IPVK_IndirectCallTarget)
+ return SymTab->getFunctionHashFromAddress(Value);
+
return Value;
}
void InstrProfRecord::addValueData(uint32_t ValueKind, uint32_t Site,
InstrProfValueData *VData, uint32_t N,
- ValueMapType *ValueMap) {
+ InstrProfSymtab *ValueMap) {
for (uint32_t I = 0; I < N; I++) {
VData[I].Value = remapValue(VData[I].Value, ValueKind, ValueMap);
}
std::vector<InstrProfValueSiteRecord> &ValueSites =
- getValueSitesForKind(ValueKind);
+ getOrCreateValueSitesForKind(ValueKind);
if (N == 0)
- ValueSites.push_back(InstrProfValueSiteRecord());
+ ValueSites.emplace_back();
else
ValueSites.emplace_back(VData, VData + N);
}
#include "llvm/ProfileData/InstrProfData.inc"
/*!
- * \brief ValueProfRecordClosure Interface implementation for InstrProfRecord
+ * ValueProfRecordClosure Interface implementation for InstrProfRecord
* class. These C wrappers are used as adaptors so that C++ code can be
* invoked as callbacks.
*/
void getValueForSiteInstrProf(const void *R, InstrProfValueData *Dst,
uint32_t K, uint32_t S) {
reinterpret_cast<const InstrProfRecord *>(R)->getValueForSite(Dst, K, S);
- return;
}
ValueProfData *allocValueProfDataInstrProf(size_t TotalSizeInBytes) {
// Wrapper implementation using the closure mechanism.
uint32_t ValueProfData::getSize(const InstrProfRecord &Record) {
- InstrProfRecordClosure.Record = &Record;
- return getValueProfDataSize(&InstrProfRecordClosure);
+ auto Closure = InstrProfRecordClosure;
+ Closure.Record = &Record;
+ return getValueProfDataSize(&Closure);
}
// Wrapper implementation using the closure mechanism.
}
void ValueProfRecord::deserializeTo(InstrProfRecord &Record,
- InstrProfRecord::ValueMapType *VMap) {
+ InstrProfSymtab *SymTab) {
Record.reserveSites(Kind, NumValueSites);
InstrProfValueData *ValueData = getValueProfRecordValueData(this);
for (uint64_t VSite = 0; VSite < NumValueSites; ++VSite) {
uint8_t ValueDataCount = this->SiteCountArray[VSite];
- Record.addValueData(Kind, VSite, ValueData, ValueDataCount, VMap);
+ Record.addValueData(Kind, VSite, ValueData, ValueDataCount, SymTab);
ValueData += ValueDataCount;
}
}
void ValueProfRecord::swapBytes(support::endianness Old,
support::endianness New) {
using namespace support;
+
if (Old == New)
return;
}
void ValueProfData::deserializeTo(InstrProfRecord &Record,
- InstrProfRecord::ValueMapType *VMap) {
+ InstrProfSymtab *SymTab) {
if (NumValueKinds == 0)
return;
ValueProfRecord *VR = getFirstValueProfRecord(this);
for (uint32_t K = 0; K < NumValueKinds; K++) {
- VR->deserializeTo(Record, VMap);
+ VR->deserializeTo(Record, SymTab);
VR = getValueProfRecordNext(VR);
}
}
template <class T>
static T swapToHostOrder(const unsigned char *&D, support::endianness Orig) {
using namespace support;
+
if (Orig == little)
return endian::readNext<T, little, unaligned>(D);
else
ValueProfData());
}
-instrprof_error ValueProfData::checkIntegrity() {
+Error ValueProfData::checkIntegrity() {
if (NumValueKinds > IPVK_Last + 1)
- return instrprof_error::malformed;
+ return make_error<InstrProfError>(instrprof_error::malformed);
// Total size needs to be mulltiple of quadword size.
if (TotalSize % sizeof(uint64_t))
- return instrprof_error::malformed;
+ return make_error<InstrProfError>(instrprof_error::malformed);
ValueProfRecord *VR = getFirstValueProfRecord(this);
for (uint32_t K = 0; K < this->NumValueKinds; K++) {
if (VR->Kind > IPVK_Last)
- return instrprof_error::malformed;
+ return make_error<InstrProfError>(instrprof_error::malformed);
VR = getValueProfRecordNext(VR);
if ((char *)VR - (char *)this > (ptrdiff_t)TotalSize)
- return instrprof_error::malformed;
+ return make_error<InstrProfError>(instrprof_error::malformed);
}
- return instrprof_error::success;
+ return Error::success();
}
-ErrorOr<std::unique_ptr<ValueProfData>>
+Expected<std::unique_ptr<ValueProfData>>
ValueProfData::getValueProfData(const unsigned char *D,
const unsigned char *const BufferEnd,
support::endianness Endianness) {
using namespace support;
+
if (D + sizeof(ValueProfData) > BufferEnd)
- return instrprof_error::truncated;
+ return make_error<InstrProfError>(instrprof_error::truncated);
const unsigned char *Header = D;
uint32_t TotalSize = swapToHostOrder<uint32_t>(Header, Endianness);
if (D + TotalSize > BufferEnd)
- return instrprof_error::too_large;
+ return make_error<InstrProfError>(instrprof_error::too_large);
std::unique_ptr<ValueProfData> VPD = allocValueProfData(TotalSize);
memcpy(VPD.get(), D, TotalSize);
// Byte swap.
VPD->swapBytesToHost(Endianness);
- instrprof_error EC = VPD->checkIntegrity();
- if (EC != instrprof_error::success)
- return EC;
+ Error E = VPD->checkIntegrity();
+ if (E)
+ return std::move(E);
return std::move(VPD);
}
void ValueProfData::swapBytesToHost(support::endianness Endianness) {
using namespace support;
+
if (Endianness == getHostEndianness())
return;
void ValueProfData::swapBytesFromHost(support::endianness Endianness) {
using namespace support;
+
if (Endianness == getHostEndianness())
return;
void annotateValueSite(Module &M, Instruction &Inst,
const InstrProfRecord &InstrProfR,
- InstrProfValueKind ValueKind, uint32_t SiteIdx) {
+ InstrProfValueKind ValueKind, uint32_t SiteIdx,
+ uint32_t MaxMDCount) {
uint32_t NV = InstrProfR.getNumValueDataForSite(ValueKind, SiteIdx);
+ if (!NV)
+ return;
uint64_t Sum = 0;
std::unique_ptr<InstrProfValueData[]> VD =
InstrProfR.getValueForSite(ValueKind, SiteIdx, &Sum);
+ ArrayRef<InstrProfValueData> VDs(VD.get(), NV);
+ annotateValueSite(M, Inst, VDs, Sum, ValueKind, MaxMDCount);
+}
+
+void annotateValueSite(Module &M, Instruction &Inst,
+ ArrayRef<InstrProfValueData> VDs,
+ uint64_t Sum, InstrProfValueKind ValueKind,
+ uint32_t MaxMDCount) {
LLVMContext &Ctx = M.getContext();
MDBuilder MDHelper(Ctx);
SmallVector<Metadata *, 3> Vals;
MDHelper.createConstant(ConstantInt::get(Type::getInt64Ty(Ctx), Sum)));
// Value Profile Data
- uint32_t MDCount = 3;
- for (uint32_t I = 0; I < NV; ++I) {
+ uint32_t MDCount = MaxMDCount;
+ for (auto &VD : VDs) {
Vals.push_back(MDHelper.createConstant(
- ConstantInt::get(Type::getInt64Ty(Ctx), VD[I].Value)));
+ ConstantInt::get(Type::getInt64Ty(Ctx), VD.Value)));
Vals.push_back(MDHelper.createConstant(
- ConstantInt::get(Type::getInt64Ty(Ctx), VD[I].Count)));
+ ConstantInt::get(Type::getInt64Ty(Ctx), VD.Count)));
if (--MDCount == 0)
break;
}
return true;
}
-// The argument to this method is a vector of cutoff percentages and the return
-// value is a vector of (Cutoff, MinBlockCount, NumBlocks) triplets.
-void ProfileSummary::computeDetailedSummary() {
- if (DetailedSummaryCutoffs.empty())
+MDNode *getPGOFuncNameMetadata(const Function &F) {
+ return F.getMetadata(getPGOFuncNameMetadataName());
+}
+
+void createPGOFuncNameMetadata(Function &F, StringRef PGOFuncName) {
+ // Only for internal linkage functions.
+ if (PGOFuncName == F.getName())
+ return;
+ // Don't create duplicated meta-data.
+ if (getPGOFuncNameMetadata(F))
return;
- auto Iter = CountFrequencies.begin();
- auto End = CountFrequencies.end();
- std::sort(DetailedSummaryCutoffs.begin(), DetailedSummaryCutoffs.end());
-
- uint32_t BlocksSeen = 0;
- uint64_t CurrSum = 0, Count = 0;
-
- for (uint32_t Cutoff : DetailedSummaryCutoffs) {
- assert(Cutoff <= 999999);
- APInt Temp(128, TotalCount);
- APInt N(128, Cutoff);
- APInt D(128, ProfileSummary::Scale);
- Temp *= N;
- Temp = Temp.sdiv(D);
- uint64_t DesiredCount = Temp.getZExtValue();
- assert(DesiredCount <= TotalCount);
- while (CurrSum < DesiredCount && Iter != End) {
- Count = Iter->first;
- uint32_t Freq = Iter->second;
- CurrSum += (Count * Freq);
- BlocksSeen += Freq;
- Iter++;
- }
- assert(CurrSum >= DesiredCount);
- ProfileSummaryEntry PSE = {Cutoff, Count, BlocksSeen};
- DetailedSummary.push_back(PSE);
+ LLVMContext &C = F.getContext();
+ MDNode *N = MDNode::get(C, MDString::get(C, PGOFuncName));
+ F.setMetadata(getPGOFuncNameMetadataName(), N);
+}
+
+bool needsComdatForCounter(const Function &F, const Module &M) {
+ if (F.hasComdat())
+ return true;
+
+ if (!Triple(M.getTargetTriple()).supportsCOMDAT())
+ return false;
+
+ // See createPGOFuncNameVar for more details. To avoid link errors, profile
+ // counters for function with available_externally linkage needs to be changed
+ // to linkonce linkage. On ELF based systems, this leads to weak symbols to be
+ // created. Without using comdat, duplicate entries won't be removed by the
+ // linker leading to increased data segement size and raw profile size. Even
+ // worse, since the referenced counter from profile per-function data object
+ // will be resolved to the common strong definition, the profile counts for
+ // available_externally functions will end up being duplicated in raw profile
+ // data. This can result in distorted profile as the counts of those dups
+ // will be accumulated by the profile merger.
+ GlobalValue::LinkageTypes Linkage = F.getLinkage();
+ if (Linkage != GlobalValue::ExternalWeakLinkage &&
+ Linkage != GlobalValue::AvailableExternallyLinkage)
+ return false;
+
+ return true;
+}
+
+// Check if INSTR_PROF_RAW_VERSION_VAR is defined.
+bool isIRPGOFlagSet(const Module *M) {
+ auto IRInstrVar =
+ M->getNamedGlobal(INSTR_PROF_QUOTE(INSTR_PROF_RAW_VERSION_VAR));
+ if (!IRInstrVar || IRInstrVar->isDeclaration() ||
+ IRInstrVar->hasLocalLinkage())
+ return false;
+
+ // Check if the flag is set.
+ if (!IRInstrVar->hasInitializer())
+ return false;
+
+ const Constant *InitVal = IRInstrVar->getInitializer();
+ if (!InitVal)
+ return false;
+
+ return (dyn_cast<ConstantInt>(InitVal)->getZExtValue() &
+ VARIANT_MASK_IR_PROF) != 0;
+}
+
+// Check if we can safely rename this Comdat function.
+bool canRenameComdatFunc(const Function &F, bool CheckAddressTaken) {
+ if (F.getName().empty())
+ return false;
+ if (!needsComdatForCounter(F, *(F.getParent())))
+ return false;
+ // Unsafe to rename the address-taken function (which can be used in
+ // function comparison).
+ if (CheckAddressTaken && F.hasAddressTaken())
+ return false;
+ // Only safe to do if this function may be discarded if it is not used
+ // in the compilation unit.
+ if (!GlobalValue::isDiscardableIfUnused(F.getLinkage()))
+ return false;
+
+ // For AvailableExternallyLinkage functions.
+ if (!F.hasComdat()) {
+ assert(F.getLinkage() == GlobalValue::AvailableExternallyLinkage);
+ return true;
}
+ return true;
}
-ProfileSummary::ProfileSummary(const IndexedInstrProf::Summary &S)
- : TotalCount(S.get(IndexedInstrProf::Summary::TotalBlockCount)),
- MaxBlockCount(S.get(IndexedInstrProf::Summary::MaxBlockCount)),
- MaxInternalBlockCount(
- S.get(IndexedInstrProf::Summary::MaxInternalBlockCount)),
- MaxFunctionCount(S.get(IndexedInstrProf::Summary::MaxFunctionCount)),
- NumBlocks(S.get(IndexedInstrProf::Summary::TotalNumBlocks)),
- NumFunctions(S.get(IndexedInstrProf::Summary::TotalNumFunctions)) {
- for (unsigned I = 0; I < S.NumCutoffEntries; I++) {
- const IndexedInstrProf::Summary::Entry &Ent = S.getEntry(I);
- DetailedSummary.emplace_back((uint32_t)Ent.Cutoff, Ent.MinBlockCount,
- Ent.NumBlocks);
+// Parse the value profile options.
+void getMemOPSizeRangeFromOption(StringRef MemOPSizeRange, int64_t &RangeStart,
+ int64_t &RangeLast) {
+ static const int64_t DefaultMemOPSizeRangeStart = 0;
+ static const int64_t DefaultMemOPSizeRangeLast = 8;
+ RangeStart = DefaultMemOPSizeRangeStart;
+ RangeLast = DefaultMemOPSizeRangeLast;
+
+ if (!MemOPSizeRange.empty()) {
+ auto Pos = MemOPSizeRange.find(':');
+ if (Pos != std::string::npos) {
+ if (Pos > 0)
+ MemOPSizeRange.substr(0, Pos).getAsInteger(10, RangeStart);
+ if (Pos < MemOPSizeRange.size() - 1)
+ MemOPSizeRange.substr(Pos + 1).getAsInteger(10, RangeLast);
+ } else
+ MemOPSizeRange.getAsInteger(10, RangeLast);
}
+ assert(RangeLast >= RangeStart);
}
} // end namespace llvm
OSDN Git Service
