summaryrefslogtreecommitdiff
path: root/lib/Transforms/IPO
diff options
context:
space:
mode:
Diffstat (limited to 'lib/Transforms/IPO')
-rw-r--r--lib/Transforms/IPO/GlobalOpt.cpp10
-rw-r--r--lib/Transforms/IPO/LowerTypeTests.cpp2
-rw-r--r--lib/Transforms/IPO/PartialInlining.cpp16
-rw-r--r--lib/Transforms/IPO/PassManagerBuilder.cpp9
-rw-r--r--lib/Transforms/IPO/SampleProfile.cpp26
5 files changed, 38 insertions, 25 deletions
diff --git a/lib/Transforms/IPO/GlobalOpt.cpp b/lib/Transforms/IPO/GlobalOpt.cpp
index 12090bff381a..4bb2984e3b47 100644
--- a/lib/Transforms/IPO/GlobalOpt.cpp
+++ b/lib/Transforms/IPO/GlobalOpt.cpp
@@ -448,9 +448,13 @@ static void transferSRADebugInfo(GlobalVariable *GV, GlobalVariable *NGV,
for (auto *GVE : GVs) {
DIVariable *Var = GVE->getVariable();
DIExpression *Expr = GVE->getExpression();
- if (NumElements > 1)
- Expr = DIExpression::createFragmentExpression(Expr, FragmentOffsetInBits,
- FragmentSizeInBits);
+ if (NumElements > 1) {
+ if (auto E = DIExpression::createFragmentExpression(
+ Expr, FragmentOffsetInBits, FragmentSizeInBits))
+ Expr = *E;
+ else
+ return;
+ }
auto *NGVE = DIGlobalVariableExpression::get(GVE->getContext(), Var, Expr);
NGV->addDebugInfo(NGVE);
}
diff --git a/lib/Transforms/IPO/LowerTypeTests.cpp b/lib/Transforms/IPO/LowerTypeTests.cpp
index 9fa5ed9ab2b8..6cef866b7b84 100644
--- a/lib/Transforms/IPO/LowerTypeTests.cpp
+++ b/lib/Transforms/IPO/LowerTypeTests.cpp
@@ -1401,7 +1401,7 @@ void LowerTypeTestsModule::buildBitSetsFromFunctionsNative(
FAlias->takeName(F);
if (FAlias->hasName())
F->setName(FAlias->getName() + ".cfi");
- F->replaceAllUsesWith(FAlias);
+ F->replaceUsesExceptBlockAddr(FAlias);
}
if (!F->isDeclarationForLinker())
F->setLinkage(GlobalValue::InternalLinkage);
diff --git a/lib/Transforms/IPO/PartialInlining.cpp b/lib/Transforms/IPO/PartialInlining.cpp
index b5267f75e417..c47d8b78df30 100644
--- a/lib/Transforms/IPO/PartialInlining.cpp
+++ b/lib/Transforms/IPO/PartialInlining.cpp
@@ -931,15 +931,17 @@ bool PartialInlinerImpl::tryPartialInline(FunctionCloner &Cloner) {
if (!shouldPartialInline(CS, Cloner, WeightedRcost, ORE))
continue;
- ORE.emit([&]() {
- return OptimizationRemark(DEBUG_TYPE, "PartiallyInlined",
- CS.getInstruction())
- << ore::NV("Callee", Cloner.OrigFunc) << " partially inlined into "
- << ore::NV("Caller", CS.getCaller());
- });
+ // Construct remark before doing the inlining, as after successful inlining
+ // the callsite is removed.
+ OptimizationRemark OR(DEBUG_TYPE, "PartiallyInlined", CS.getInstruction());
+ OR << ore::NV("Callee", Cloner.OrigFunc) << " partially inlined into "
+ << ore::NV("Caller", CS.getCaller());
InlineFunctionInfo IFI(nullptr, GetAssumptionCache, PSI);
- InlineFunction(CS, IFI);
+ if (!InlineFunction(CS, IFI))
+ continue;
+
+ ORE.emit(OR);
// Now update the entry count:
if (CalleeEntryCountV && CallSiteToProfCountMap.count(User)) {
diff --git a/lib/Transforms/IPO/PassManagerBuilder.cpp b/lib/Transforms/IPO/PassManagerBuilder.cpp
index 828eb5eee297..5d3736655093 100644
--- a/lib/Transforms/IPO/PassManagerBuilder.cpp
+++ b/lib/Transforms/IPO/PassManagerBuilder.cpp
@@ -467,6 +467,9 @@ void PassManagerBuilder::populateModulePassManager(
addExtensionsToPM(EP_ModuleOptimizerEarly, MPM);
+ if (OptLevel > 2)
+ MPM.add(createCallSiteSplittingPass());
+
MPM.add(createIPSCCPPass()); // IP SCCP
MPM.add(createCalledValuePropagationPass());
MPM.add(createGlobalOptimizerPass()); // Optimize out global vars
@@ -545,6 +548,9 @@ void PassManagerBuilder::populateModulePassManager(
// unrolling/vectorization/... now. We'll first run the inliner + CGSCC passes
// during ThinLTO and perform the rest of the optimizations afterward.
if (PrepareForThinLTO) {
+ // Ensure we perform any last passes, but do so before renaming anonymous
+ // globals in case the passes add any.
+ addExtensionsToPM(EP_OptimizerLast, MPM);
// Rename anon globals to be able to export them in the summary.
MPM.add(createNameAnonGlobalPass());
return;
@@ -703,6 +709,9 @@ void PassManagerBuilder::addLTOOptimizationPasses(legacy::PassManagerBase &PM) {
PM.add(createInferFunctionAttrsLegacyPass());
if (OptLevel > 1) {
+ // Split call-site with more constrained arguments.
+ PM.add(createCallSiteSplittingPass());
+
// Indirect call promotion. This should promote all the targets that are
// left by the earlier promotion pass that promotes intra-module targets.
// This two-step promotion is to save the compile time. For LTO, it should
diff --git a/lib/Transforms/IPO/SampleProfile.cpp b/lib/Transforms/IPO/SampleProfile.cpp
index 34414f96ccae..8930e9b2b957 100644
--- a/lib/Transforms/IPO/SampleProfile.cpp
+++ b/lib/Transforms/IPO/SampleProfile.cpp
@@ -1182,24 +1182,20 @@ void SampleProfileLoader::buildEdges(Function &F) {
}
}
-/// Sorts the CallTargetMap \p M by count in descending order and stores the
-/// sorted result in \p Sorted. Returns the total counts.
-static uint64_t SortCallTargets(SmallVector<InstrProfValueData, 2> &Sorted,
- const SampleRecord::CallTargetMap &M) {
- Sorted.clear();
- uint64_t Sum = 0;
- for (auto I = M.begin(); I != M.end(); ++I) {
- Sum += I->getValue();
- Sorted.push_back({Function::getGUID(I->getKey()), I->getValue()});
- }
- std::sort(Sorted.begin(), Sorted.end(),
+/// Returns the sorted CallTargetMap \p M by count in descending order.
+static SmallVector<InstrProfValueData, 2> SortCallTargets(
+ const SampleRecord::CallTargetMap &M) {
+ SmallVector<InstrProfValueData, 2> R;
+ for (auto I = M.begin(); I != M.end(); ++I)
+ R.push_back({Function::getGUID(I->getKey()), I->getValue()});
+ std::sort(R.begin(), R.end(),
[](const InstrProfValueData &L, const InstrProfValueData &R) {
if (L.Count == R.Count)
return L.Value > R.Value;
else
return L.Count > R.Count;
});
- return Sum;
+ return R;
}
/// \brief Propagate weights into edges
@@ -1292,8 +1288,10 @@ void SampleProfileLoader::propagateWeights(Function &F) {
auto T = FS->findCallTargetMapAt(LineOffset, Discriminator);
if (!T || T.get().empty())
continue;
- SmallVector<InstrProfValueData, 2> SortedCallTargets;
- uint64_t Sum = SortCallTargets(SortedCallTargets, T.get());
+ SmallVector<InstrProfValueData, 2> SortedCallTargets =
+ SortCallTargets(T.get());
+ uint64_t Sum;
+ findIndirectCallFunctionSamples(I, Sum);
annotateValueSite(*I.getParent()->getParent()->getParent(), I,
SortedCallTargets, Sum, IPVK_IndirectCallTarget,
SortedCallTargets.size());
View Lorry Controller Status