CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
212 lines
8.2 KiB
C++
212 lines
8.2 KiB
C++
#include "llvm/Analysis/StaticDataProfileInfo.h"
|
|
#include "llvm/Analysis/ProfileSummaryInfo.h"
|
|
#include "llvm/IR/Constant.h"
|
|
#include "llvm/IR/Constants.h"
|
|
#include "llvm/IR/GlobalVariable.h"
|
|
#include "llvm/IR/Module.h"
|
|
#include "llvm/InitializePasses.h"
|
|
#include "llvm/ProfileData/InstrProf.h"
|
|
|
|
#define DEBUG_TYPE "static-data-profile-info"
|
|
|
|
using namespace llvm;
|
|
|
|
namespace llvm {
|
|
// FIXME: This option is added for incremental rollout purposes.
|
|
// After the option, string literal partitioning should be implied by
|
|
// AnnotateStaticDataSectionPrefix in MemProfUse.cpp and this option should be
|
|
// cleaned up.
|
|
cl::opt<bool> AnnotateStringLiteralSectionPrefix(
|
|
"memprof-annotate-string-literal-section-prefix", cl::init(false),
|
|
cl::Hidden,
|
|
cl::desc("If true, annotate the string literal data section prefix"));
|
|
namespace memprof {
|
|
// Returns true iff the global variable has custom section either by
|
|
// __attribute__((section("name")))
|
|
// (https://clang.llvm.org/docs/AttributeReference.html#section-declspec-allocate)
|
|
// or #pragma clang section directives
|
|
// (https://clang.llvm.org/docs/LanguageExtensions.html#specifying-section-names-for-global-objects-pragma-clang-section).
|
|
static bool hasExplicitSectionName(const GlobalVariable &GVar) {
|
|
if (GVar.hasSection())
|
|
return true;
|
|
|
|
auto Attrs = GVar.getAttributes();
|
|
if (Attrs.hasAttribute("bss-section") || Attrs.hasAttribute("data-section") ||
|
|
Attrs.hasAttribute("relro-section") ||
|
|
Attrs.hasAttribute("rodata-section"))
|
|
return true;
|
|
return false;
|
|
}
|
|
|
|
AnnotationKind getAnnotationKind(const GlobalVariable &GV) {
|
|
if (GV.isDeclarationForLinker())
|
|
return AnnotationKind::DeclForLinker;
|
|
// Skip 'llvm.'-prefixed global variables conservatively because they are
|
|
// often handled specially,
|
|
StringRef Name = GV.getName();
|
|
if (Name.starts_with("llvm."))
|
|
return AnnotationKind::ReservedName;
|
|
// Respect user-specified custom data sections.
|
|
if (hasExplicitSectionName(GV))
|
|
return AnnotationKind::ExplicitSection;
|
|
return AnnotationKind::AnnotationOK;
|
|
}
|
|
|
|
bool IsAnnotationOK(const GlobalVariable &GV) {
|
|
return getAnnotationKind(GV) == AnnotationKind::AnnotationOK;
|
|
}
|
|
} // namespace memprof
|
|
} // namespace llvm
|
|
|
|
void StaticDataProfileInfo::addConstantProfileCount(
|
|
const Constant *C, std::optional<uint64_t> Count) {
|
|
if (!Count) {
|
|
ConstantWithoutCounts.insert(C);
|
|
return;
|
|
}
|
|
uint64_t &OriginalCount = ConstantProfileCounts[C];
|
|
OriginalCount = llvm::SaturatingAdd(*Count, OriginalCount);
|
|
// Clamp the count to getInstrMaxCountValue. InstrFDO reserves a few
|
|
// large values for special use.
|
|
if (OriginalCount > getInstrMaxCountValue())
|
|
OriginalCount = getInstrMaxCountValue();
|
|
}
|
|
|
|
StaticDataProfileInfo::StaticDataHotness
|
|
StaticDataProfileInfo::getConstantHotnessUsingProfileCount(
|
|
const Constant *C, const ProfileSummaryInfo *PSI, uint64_t Count) const {
|
|
// The accummulated counter shows the constant is hot. Return enum 'hot'
|
|
// whether this variable is seen by unprofiled functions or not.
|
|
if (PSI->isHotCount(Count))
|
|
return StaticDataHotness::Hot;
|
|
// The constant is not hot, and seen by unprofiled functions. We don't want to
|
|
// assign it to unlikely sections, even if the counter says 'cold'. So return
|
|
// enum 'LukewarmOrUnknown'.
|
|
if (ConstantWithoutCounts.count(C))
|
|
return StaticDataHotness::LukewarmOrUnknown;
|
|
// The accummulated counter shows the constant is cold so return enum 'cold'.
|
|
if (PSI->isColdCount(Count))
|
|
return StaticDataHotness::Cold;
|
|
|
|
return StaticDataHotness::LukewarmOrUnknown;
|
|
}
|
|
|
|
StaticDataProfileInfo::StaticDataHotness
|
|
StaticDataProfileInfo::getSectionHotnessUsingDataAccessProfile(
|
|
std::optional<StringRef> MaybeSectionPrefix) const {
|
|
if (!MaybeSectionPrefix)
|
|
return StaticDataHotness::LukewarmOrUnknown;
|
|
StringRef Prefix = *MaybeSectionPrefix;
|
|
assert((Prefix == "hot" || Prefix == "unlikely") &&
|
|
"Expect section_prefix to be one of hot or unlikely");
|
|
return Prefix == "hot" ? StaticDataHotness::Hot : StaticDataHotness::Cold;
|
|
}
|
|
|
|
StringRef StaticDataProfileInfo::hotnessToStr(StaticDataHotness Hotness) const {
|
|
switch (Hotness) {
|
|
case StaticDataHotness::Cold:
|
|
return "unlikely";
|
|
case StaticDataHotness::Hot:
|
|
return "hot";
|
|
default:
|
|
return "";
|
|
}
|
|
}
|
|
|
|
std::optional<uint64_t>
|
|
StaticDataProfileInfo::getConstantProfileCount(const Constant *C) const {
|
|
auto I = ConstantProfileCounts.find(C);
|
|
if (I == ConstantProfileCounts.end())
|
|
return std::nullopt;
|
|
return I->second;
|
|
}
|
|
|
|
StringRef StaticDataProfileInfo::getConstantSectionPrefix(
|
|
const Constant *C, const ProfileSummaryInfo *PSI) const {
|
|
std::optional<uint64_t> Count = getConstantProfileCount(C);
|
|
|
|
#ifndef NDEBUG
|
|
auto DbgPrintPrefix = [](StringRef Prefix) {
|
|
return Prefix.empty() ? "<empty>" : Prefix;
|
|
};
|
|
#endif
|
|
|
|
if (EnableDataAccessProf) {
|
|
// Both data access profiles and PGO counters are available. Use the
|
|
// hotter one to be conservative. Basically, we want the non-unlikely
|
|
// sections to have max coverage of accessed symbols and meanwhile can
|
|
// tolerant some cold symbols in it, and the unlikely section variant to not
|
|
// have potentially hot symbols if possible, to avoid the penalty of access
|
|
// cold pages.
|
|
if (const GlobalVariable *GV = dyn_cast<GlobalVariable>(C);
|
|
GV && llvm::memprof::IsAnnotationOK(*GV) &&
|
|
(AnnotateStringLiteralSectionPrefix ||
|
|
!GV->getName().starts_with(".str"))) {
|
|
// Note a global var is covered by data access profiles iff the
|
|
// symbol name is preserved in the symbol table; most notably, a string
|
|
// literal with private linkage (e.g., those not externalized by ThinLTO
|
|
// and with insignificant address) won't have an entry in the symbol
|
|
// table (unless there is another string with identical content that
|
|
// gets a symbol table entry). For the private-linkage string literals,
|
|
// their hotness will be at least lukewarm (i.e., empty prefix).
|
|
auto HotnessFromDataAccessProf =
|
|
getSectionHotnessUsingDataAccessProfile(GV->getSectionPrefix());
|
|
|
|
if (!Count) {
|
|
StringRef Prefix = hotnessToStr(HotnessFromDataAccessProf);
|
|
LLVM_DEBUG(dbgs() << GV->getName() << " has section prefix "
|
|
<< DbgPrintPrefix(Prefix)
|
|
<< ", solely from data access profiles\n");
|
|
return Prefix;
|
|
}
|
|
|
|
auto HotnessFromPGO = getConstantHotnessUsingProfileCount(C, PSI, *Count);
|
|
StaticDataHotness GlobalVarHotness = StaticDataHotness::LukewarmOrUnknown;
|
|
if (HotnessFromDataAccessProf == StaticDataHotness::Hot ||
|
|
HotnessFromPGO == StaticDataHotness::Hot) {
|
|
GlobalVarHotness = StaticDataHotness::Hot;
|
|
} else if (HotnessFromDataAccessProf ==
|
|
StaticDataHotness::LukewarmOrUnknown ||
|
|
HotnessFromPGO == StaticDataHotness::LukewarmOrUnknown) {
|
|
GlobalVarHotness = StaticDataHotness::LukewarmOrUnknown;
|
|
} else {
|
|
GlobalVarHotness = StaticDataHotness::Cold;
|
|
}
|
|
StringRef Prefix = hotnessToStr(GlobalVarHotness);
|
|
LLVM_DEBUG(
|
|
dbgs() << GV->getName() << " has section prefix "
|
|
<< DbgPrintPrefix(Prefix)
|
|
<< ", the max from data access profiles as "
|
|
<< DbgPrintPrefix(hotnessToStr(HotnessFromDataAccessProf))
|
|
<< " and PGO counters as "
|
|
<< DbgPrintPrefix(hotnessToStr(HotnessFromPGO)) << "\n");
|
|
return Prefix;
|
|
}
|
|
}
|
|
if (!Count)
|
|
return "";
|
|
return hotnessToStr(getConstantHotnessUsingProfileCount(C, PSI, *Count));
|
|
}
|
|
|
|
bool StaticDataProfileInfoWrapperPass::doInitialization(Module &M) {
|
|
bool EnableDataAccessProf = false;
|
|
if (auto *MD = mdconst::extract_or_null<ConstantInt>(
|
|
M.getModuleFlag("EnableDataAccessProf")))
|
|
EnableDataAccessProf = MD->getZExtValue();
|
|
Info.reset(new StaticDataProfileInfo(EnableDataAccessProf));
|
|
return false;
|
|
}
|
|
|
|
bool StaticDataProfileInfoWrapperPass::doFinalization(Module &M) {
|
|
Info.reset();
|
|
return false;
|
|
}
|
|
|
|
INITIALIZE_PASS(StaticDataProfileInfoWrapperPass, "static-data-profile-info",
|
|
"Static Data Profile Info", false, true)
|
|
|
|
StaticDataProfileInfoWrapperPass::StaticDataProfileInfoWrapperPass()
|
|
: ImmutablePass(ID) {}
|
|
|
|
char StaticDataProfileInfoWrapperPass::ID = 0;
|