aboutsummaryrefslogtreecommitdiff
path: root/llvm/lib/Transforms/IPO
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/lib/Transforms/IPO')
-rw-r--r--llvm/lib/Transforms/IPO/FunctionImport.cpp4
-rw-r--r--llvm/lib/Transforms/IPO/FunctionSpecialization.cpp11
-rw-r--r--llvm/lib/Transforms/IPO/GlobalOpt.cpp16
-rw-r--r--llvm/lib/Transforms/IPO/MemProfContextDisambiguation.cpp5
-rw-r--r--llvm/lib/Transforms/IPO/SampleProfile.cpp11
-rw-r--r--llvm/lib/Transforms/IPO/SampleProfileMatcher.cpp4
-rw-r--r--llvm/lib/Transforms/IPO/WholeProgramDevirt.cpp4
7 files changed, 38 insertions, 17 deletions
diff --git a/llvm/lib/Transforms/IPO/FunctionImport.cpp b/llvm/lib/Transforms/IPO/FunctionImport.cpp
index 83aa7de..28ee444 100644
--- a/llvm/lib/Transforms/IPO/FunctionImport.cpp
+++ b/llvm/lib/Transforms/IPO/FunctionImport.cpp
@@ -72,6 +72,7 @@ STATISTIC(NumImportedModules, "Number of modules imported from");
STATISTIC(NumDeadSymbols, "Number of dead stripped symbols in index");
STATISTIC(NumLiveSymbols, "Number of live symbols in index");
+namespace llvm {
cl::opt<bool>
ForceImportAll("force-import-all", cl::init(false), cl::Hidden,
cl::desc("Import functions with noinline attribute"));
@@ -185,9 +186,8 @@ static cl::opt<bool> CtxprofMoveRootsToOwnModule(
extern cl::list<GlobalValue::GUID> MoveSymbolGUID;
-namespace llvm {
extern cl::opt<bool> EnableMemProfContextDisambiguation;
-}
+} // end namespace llvm
// Load lazily a module from \p FileName in \p Context.
static std::unique_ptr<Module> loadFile(const std::string &FileName,
diff --git a/llvm/lib/Transforms/IPO/FunctionSpecialization.cpp b/llvm/lib/Transforms/IPO/FunctionSpecialization.cpp
index 4f53738..150a2dc 100644
--- a/llvm/lib/Transforms/IPO/FunctionSpecialization.cpp
+++ b/llvm/lib/Transforms/IPO/FunctionSpecialization.cpp
@@ -28,10 +28,13 @@ using namespace llvm;
STATISTIC(NumSpecsCreated, "Number of specializations created");
+namespace llvm {
+
static cl::opt<bool> ForceSpecialization(
- "force-specialization", cl::init(false), cl::Hidden, cl::desc(
- "Force function specialization for every call site with a constant "
- "argument"));
+ "force-specialization", cl::init(false), cl::Hidden,
+ cl::desc(
+ "Force function specialization for every call site with a constant "
+ "argument"));
static cl::opt<unsigned> MaxClones(
"funcspec-max-clones", cl::init(3), cl::Hidden, cl::desc(
@@ -91,6 +94,8 @@ static cl::opt<bool> SpecializeLiteralConstant(
extern cl::opt<bool> ProfcheckDisableMetadataFixes;
+} // end namespace llvm
+
bool InstCostVisitor::canEliminateSuccessor(BasicBlock *BB,
BasicBlock *Succ) const {
unsigned I = 0;
diff --git a/llvm/lib/Transforms/IPO/GlobalOpt.cpp b/llvm/lib/Transforms/IPO/GlobalOpt.cpp
index f88d51f..99c4982 100644
--- a/llvm/lib/Transforms/IPO/GlobalOpt.cpp
+++ b/llvm/lib/Transforms/IPO/GlobalOpt.cpp
@@ -1680,7 +1680,9 @@ processGlobal(GlobalValue &GV,
/// FastCC.
static void ChangeCalleesToFastCall(Function *F) {
for (User *U : F->users())
- cast<CallBase>(U)->setCallingConv(CallingConv::Fast);
+ if (auto *Call = dyn_cast<CallBase>(U))
+ if (Call->getCalledOperand() == F)
+ Call->setCallingConv(CallingConv::Fast);
}
static AttributeList StripAttr(LLVMContext &C, AttributeList Attrs,
@@ -1766,10 +1768,12 @@ isValidCandidateForColdCC(Function &F,
return false;
for (User *U : F.users()) {
- CallBase &CB = cast<CallBase>(*U);
- Function *CallerFunc = CB.getParent()->getParent();
+ CallBase *CB = dyn_cast<CallBase>(U);
+ if (!CB || CB->getCalledOperand() != &F)
+ continue;
+ Function *CallerFunc = CB->getParent()->getParent();
BlockFrequencyInfo &CallerBFI = GetBFI(*CallerFunc);
- if (!isColdCallSite(CB, CallerBFI))
+ if (!isColdCallSite(*CB, CallerBFI))
return false;
if (!llvm::is_contained(AllCallsCold, CallerFunc))
return false;
@@ -1779,7 +1783,9 @@ isValidCandidateForColdCC(Function &F,
static void changeCallSitesToColdCC(Function *F) {
for (User *U : F->users())
- cast<CallBase>(U)->setCallingConv(CallingConv::Cold);
+ if (auto *Call = dyn_cast<CallBase>(U))
+ if (Call->getCalledOperand() == F)
+ Call->setCallingConv(CallingConv::Cold);
}
// This function iterates over all the call instructions in the input Function
diff --git a/llvm/lib/Transforms/IPO/MemProfContextDisambiguation.cpp b/llvm/lib/Transforms/IPO/MemProfContextDisambiguation.cpp
index 15f4d76..ddb95a4 100644
--- a/llvm/lib/Transforms/IPO/MemProfContextDisambiguation.cpp
+++ b/llvm/lib/Transforms/IPO/MemProfContextDisambiguation.cpp
@@ -214,11 +214,12 @@ static cl::opt<bool> MemProfRequireDefinitionForPromotion(
"memprof-require-definition-for-promotion", cl::init(false), cl::Hidden,
cl::desc(
"Require target function definition when promoting indirect calls"));
-} // namespace llvm
extern cl::opt<bool> MemProfReportHintedSizes;
extern cl::opt<unsigned> MinClonedColdBytePercent;
+} // namespace llvm
+
namespace {
/// CRTP base for graphs built from either IR or ThinLTO summary index.
///
@@ -3980,7 +3981,6 @@ void CallsiteContextGraph<DerivedCCG, FuncTy, CallTy>::identifyClones(
void ModuleCallsiteContextGraph::updateAllocationCall(
CallInfo &Call, AllocationType AllocType) {
std::string AllocTypeString = getAllocTypeAttributeString(AllocType);
- removeAnyExistingAmbiguousAttribute(cast<CallBase>(Call.call()));
auto A = llvm::Attribute::get(Call.call()->getFunction()->getContext(),
"memprof", AllocTypeString);
cast<CallBase>(Call.call())->addFnAttr(A);
@@ -5642,7 +5642,6 @@ bool MemProfContextDisambiguation::applyImport(Module &M) {
// clone J-1 (J==0 is the original clone and does not have a VMaps
// entry).
CBClone = cast<CallBase>((*VMaps[J - 1])[CB]);
- removeAnyExistingAmbiguousAttribute(CBClone);
CBClone->addFnAttr(A);
ORE.emit(OptimizationRemark(DEBUG_TYPE, "MemprofAttribute", CBClone)
<< ore::NV("AllocationCall", CBClone) << " in clone "
diff --git a/llvm/lib/Transforms/IPO/SampleProfile.cpp b/llvm/lib/Transforms/IPO/SampleProfile.cpp
index 5bc7e34..e39e311 100644
--- a/llvm/lib/Transforms/IPO/SampleProfile.cpp
+++ b/llvm/lib/Transforms/IPO/SampleProfile.cpp
@@ -116,6 +116,8 @@ STATISTIC(
NumCSInlinedHitGrowthLimit,
"Number of functions with FDO inline stopped due to growth size limit");
+namespace llvm {
+
// Command line option to specify the file to read samples from. This is
// mainly used for debugging.
static cl::opt<std::string> SampleProfileFile(
@@ -198,7 +200,6 @@ static cl::opt<bool> DisableSampleLoaderInlining(
"pass, and merge (or scale) profiles (as configured by "
"--sample-profile-merge-inlinee)."));
-namespace llvm {
cl::opt<bool>
SortProfiledSCC("sort-profiled-scc-member", cl::init(true), cl::Hidden,
cl::desc("Sort profiled recursion by edge weights."));
@@ -1664,8 +1665,9 @@ void SampleProfileLoader::generateMDProfMetadata(Function &F) {
else if (OverwriteExistingWeights)
I.setMetadata(LLVMContext::MD_prof, nullptr);
} else if (!isa<IntrinsicInst>(&I)) {
- setBranchWeights(I, {static_cast<uint32_t>(BlockWeights[BB])},
- /*IsExpected=*/false);
+ setBranchWeights(
+ I, ArrayRef<uint32_t>{static_cast<uint32_t>(BlockWeights[BB])},
+ /*IsExpected=*/false);
}
}
} else if (OverwriteExistingWeights || ProfileSampleBlockAccurate) {
@@ -1676,7 +1678,8 @@ void SampleProfileLoader::generateMDProfMetadata(Function &F) {
if (cast<CallBase>(I).isIndirectCall()) {
I.setMetadata(LLVMContext::MD_prof, nullptr);
} else {
- setBranchWeights(I, {uint32_t(0)}, /*IsExpected=*/false);
+ setBranchWeights(I, ArrayRef<uint32_t>{uint32_t(0)},
+ /*IsExpected=*/false);
}
}
}
diff --git a/llvm/lib/Transforms/IPO/SampleProfileMatcher.cpp b/llvm/lib/Transforms/IPO/SampleProfileMatcher.cpp
index 093a39e..70b8614 100644
--- a/llvm/lib/Transforms/IPO/SampleProfileMatcher.cpp
+++ b/llvm/lib/Transforms/IPO/SampleProfileMatcher.cpp
@@ -23,6 +23,8 @@ using namespace sampleprof;
#define DEBUG_TYPE "sample-profile-matcher"
+namespace llvm {
+
static cl::opt<unsigned> FuncProfileSimilarityThreshold(
"func-profile-similarity-threshold", cl::Hidden, cl::init(80),
cl::desc("Consider a profile matches a function if the similarity of their "
@@ -55,6 +57,8 @@ static cl::opt<unsigned> SalvageStaleProfileMaxCallsites(
cl::desc("The maximum number of callsites in a function, above which stale "
"profile matching will be skipped."));
+} // end namespace llvm
+
void SampleProfileMatcher::findIRAnchors(const Function &F,
AnchorMap &IRAnchors) const {
// For inlined code, recover the original callsite and callee by finding the
diff --git a/llvm/lib/Transforms/IPO/WholeProgramDevirt.cpp b/llvm/lib/Transforms/IPO/WholeProgramDevirt.cpp
index 09bffa7..ac41fdd 100644
--- a/llvm/lib/Transforms/IPO/WholeProgramDevirt.cpp
+++ b/llvm/lib/Transforms/IPO/WholeProgramDevirt.cpp
@@ -120,6 +120,8 @@ STATISTIC(NumVirtConstProp1Bit,
"Number of 1 bit virtual constant propagations");
STATISTIC(NumVirtConstProp, "Number of virtual constant propagations");
+namespace llvm {
+
static cl::opt<PassSummaryAction> ClSummaryAction(
"wholeprogramdevirt-summary-action",
cl::desc("What to do with the summary when running this pass"),
@@ -175,6 +177,8 @@ static cl::list<std::string>
extern cl::opt<bool> ProfcheckDisableMetadataFixes;
+} // end namespace llvm
+
/// With Clang, a pure virtual class's deleting destructor is emitted as a
/// `llvm.trap` intrinsic followed by an unreachable IR instruction. In the
/// context of whole program devirtualization, the deleting destructor of a pure
generated by cgit v1.1 at 2025-10-02 21:50:42 +0000