[llvm] [BOLT]Fix profile quality reporting for small binaries (PR #130810)
via llvm-commits
llvm-commits at lists.llvm.org
Thu Apr 3 08:01:59 PDT 2025
https://github.com/ShatianWang updated https://github.com/llvm/llvm-project/pull/130810
>From 39112aed99fb6342d2e821896fa35485569ab7e2 Mon Sep 17 00:00:00 2001
From: Shatian Wang <shatian at meta.com>
Date: Tue, 11 Mar 2025 09:05:08 -0700
Subject: [PATCH 1/2] Report 0.00% profile quality gaps under empty eligible
functions / blocks
---
bolt/lib/Passes/ProfileQualityStats.cpp | 13 +++++--
.../profile-quality-reporting-small-binary.s | 35 +++++++++++++++++++
2 files changed, 45 insertions(+), 3 deletions(-)
create mode 100644 bolt/test/X86/profile-quality-reporting-small-binary.s
diff --git a/bolt/lib/Passes/ProfileQualityStats.cpp b/bolt/lib/Passes/ProfileQualityStats.cpp
index 332c78da8a1e3..61d67b4f9068e 100644
--- a/bolt/lib/Passes/ProfileQualityStats.cpp
+++ b/bolt/lib/Passes/ProfileQualityStats.cpp
@@ -157,8 +157,10 @@ void printCFGContinuityStats(raw_ostream &OS,
FractionECUnreachables.push_back(FractionECUnreachable);
}
- if (FractionECUnreachables.empty())
+ if (FractionECUnreachables.empty()) {
+ OS << "function CFG discontinuity 0.00%; ";
return;
+ }
llvm::sort(FractionECUnreachables);
const int Rank = int(FractionECUnreachables.size() *
@@ -251,8 +253,10 @@ void printCallGraphFlowConservationStats(
}
}
- if (CallGraphGaps.empty())
+ if (CallGraphGaps.empty()) {
+ OS << "call graph flow conservation gap 0.00%; ";
return;
+ }
llvm::sort(CallGraphGaps);
const int Rank =
@@ -340,8 +344,11 @@ void printCFGFlowConservationStats(raw_ostream &OS,
}
}
- if (CFGGapsWeightedAvg.empty())
+ if (CFGGapsWeightedAvg.empty()) {
+ OS << "CFG flow conservation gap 0.00% (weighted) 0.00% (worst)\n";
return;
+ }
+
llvm::sort(CFGGapsWeightedAvg);
const int RankWA = int(CFGGapsWeightedAvg.size() *
opts::PercentileForProfileQualityCheck / 100);
diff --git a/bolt/test/X86/profile-quality-reporting-small-binary.s b/bolt/test/X86/profile-quality-reporting-small-binary.s
new file mode 100644
index 0000000000000..603d5c3218bc3
--- /dev/null
+++ b/bolt/test/X86/profile-quality-reporting-small-binary.s
@@ -0,0 +1,35 @@
+## Test that BOLT-INFO is correctly formatted after profile quality reporting for
+## a small binary.
+
+# RUN: llvm-mc --filetype=obj --triple x86_64-unknown-unknown %s -o %t.o
+# RUN: link_fdata %s %t.o %t.fdata
+# RUN: llvm-strip --strip-unneeded %t.o
+# RUN: %clang %cflags %t.o -o %t.exe -Wl,-q
+# RUN: llvm-bolt %t.exe -o %t.bolt --data=%t.fdata \
+# RUN: 2>&1 | FileCheck %s
+
+# CHECK: BOLT-INFO: profile quality metrics for the hottest 2 functions (reporting top 5% values): function CFG discontinuity 0.00%; call graph flow conservation gap 0.00%; CFG flow conservation gap 0.00% (weighted) 0.00% (worst)
+# CHECK-NEXT: BOLT-INFO:
+
+ .text
+ .globl func
+ .type func, @function
+func:
+ pushq %rbp
+ ret
+LLfunc_end:
+ .size func, LLfunc_end-func
+
+
+ .globl main
+ .type main, @function
+main:
+ pushq %rbp
+ movq %rsp, %rbp
+LLmain_func:
+ call func
+# FDATA: 1 main #LLmain_func# 1 func 0 0 500
+ movl $4, %edi
+ retq
+.Lmain_end:
+ .size main, .Lmain_end-main
>From 8121b86517884ef5fd8ae4f6595365577f0aeb34 Mon Sep 17 00:00:00 2001
From: Shatian Wang <shatian at meta.com>
Date: Wed, 19 Mar 2025 12:36:31 -0700
Subject: [PATCH 2/2] Fixed format; added EH stats; skipped selected blocks in
CFG flow conservation score calculation
---
bolt/lib/Passes/ProfileQualityStats.cpp | 181 +++++++++++++-----
.../profile-quality-reporting-small-binary.s | 2 +-
bolt/test/X86/profile-quality-reporting.test | 2 +-
3 files changed, 137 insertions(+), 48 deletions(-)
diff --git a/bolt/lib/Passes/ProfileQualityStats.cpp b/bolt/lib/Passes/ProfileQualityStats.cpp
index 61d67b4f9068e..c19ba76d10c93 100644
--- a/bolt/lib/Passes/ProfileQualityStats.cpp
+++ b/bolt/lib/Passes/ProfileQualityStats.cpp
@@ -91,8 +91,12 @@ void printCFGContinuityStats(raw_ostream &OS,
std::vector<double> FractionECUnreachables;
for (const BinaryFunction *Function : Functions) {
- if (Function->size() <= 1)
+ if (Function->size() <= 1) {
+ NumUnreachables.push_back(0);
+ SumECUnreachables.push_back(0);
+ FractionECUnreachables.push_back(0.0);
continue;
+ }
// Compute the sum of all BB execution counts (ECs).
size_t NumPosECBBs = 0;
@@ -157,11 +161,6 @@ void printCFGContinuityStats(raw_ostream &OS,
FractionECUnreachables.push_back(FractionECUnreachable);
}
- if (FractionECUnreachables.empty()) {
- OS << "function CFG discontinuity 0.00%; ";
- return;
- }
-
llvm::sort(FractionECUnreachables);
const int Rank = int(FractionECUnreachables.size() *
opts::PercentileForProfileQualityCheck / 100);
@@ -189,8 +188,10 @@ void printCallGraphFlowConservationStats(
std::vector<double> CallGraphGaps;
for (const BinaryFunction *Function : Functions) {
- if (Function->size() <= 1 || !Function->isSimple())
+ if (Function->size() <= 1 || !Function->isSimple()) {
+ CallGraphGaps.push_back(0.0);
continue;
+ }
const uint64_t FunctionNum = Function->getFunctionNumber();
std::vector<uint64_t> &IncomingFlows =
@@ -253,11 +254,6 @@ void printCallGraphFlowConservationStats(
}
}
- if (CallGraphGaps.empty()) {
- OS << "call graph flow conservation gap 0.00%; ";
- return;
- }
-
llvm::sort(CallGraphGaps);
const int Rank =
int(CallGraphGaps.size() * opts::PercentileForProfileQualityCheck / 100);
@@ -269,7 +265,7 @@ void printCallGraphFlowConservationStats(
}
}
-void printCFGFlowConservationStats(raw_ostream &OS,
+void printCFGFlowConservationStats(const BinaryContext &BC, raw_ostream &OS,
iterator_range<function_iterator> &Functions,
FlowInfo &TotalFlowMap) {
std::vector<double> CFGGapsWeightedAvg;
@@ -279,8 +275,12 @@ void printCFGFlowConservationStats(raw_ostream &OS,
// reporting the distribution of worst gaps.
const uint16_t MinBlockCount = 500;
for (const BinaryFunction *Function : Functions) {
- if (Function->size() <= 1 || !Function->isSimple())
+ if (Function->size() <= 1 || !Function->isSimple()) {
+ CFGGapsWeightedAvg.push_back(0.0);
+ CFGGapsWorst.push_back(0.0);
+ CFGGapsWorstAbs.push_back(0);
continue;
+ }
const uint64_t FunctionNum = Function->getFunctionNumber();
std::vector<uint64_t> &MaxCountMaps =
@@ -299,12 +299,38 @@ void printCFGFlowConservationStats(raw_ostream &OS,
if (BB.isEntryPoint() || BB.succ_size() == 0)
continue;
+ if (BB.getKnownExecutionCount() == 0 || BB.getNumNonPseudos() == 0)
+ continue;
+
+ // We don't consider blocks that is a landing pad or has a
+ // positive-execution-count landing pad
+ if (BB.isLandingPad())
+ continue;
+
+ bool HasPosECLP = false;
+ for (const BinaryBasicBlock *LP : BB.landing_pads()) {
+ if (LP->getKnownExecutionCount() > 0) {
+ HasPosECLP = true;
+ break;
+ }
+ }
+ if (HasPosECLP)
+ continue;
+
+ // We don't consider blocks that end with a recursive call instruction
+ const MCInst *Inst = BB.getLastNonPseudoInstr();
+ if (BC.MIB->isCall(*Inst)) {
+ const MCSymbol *DstSym = BC.MIB->getTargetSymbol(*Inst);
+ const BinaryFunction *DstFunc =
+ DstSym ? BC.getFunctionForSymbol(DstSym) : nullptr;
+ if (DstFunc == Function)
+ continue;
+ }
+
const uint64_t Max = MaxCountMaps[BB.getLayoutIndex()];
const uint64_t Min = MinCountMaps[BB.getLayoutIndex()];
const double Gap = 1 - (double)Min / Max;
double Weight = BB.getKnownExecutionCount() * BB.getNumNonPseudos();
- if (Weight == 0)
- continue;
// We use log to prevent the stats from being dominated by extremely hot
// blocks
Weight = log(Weight);
@@ -320,33 +346,27 @@ void printCFGFlowConservationStats(raw_ostream &OS,
BBWorstGapAbs = &BB;
}
}
- if (WeightSum > 0) {
- const double WeightedGap = WeightedGapSum / WeightSum;
- if (opts::Verbosity >= 2 && (WeightedGap >= 0.1 || WorstGap >= 0.9)) {
- OS << "Nontrivial CFG gap observed in function "
- << Function->getPrintName() << "\n"
- << "Weighted gap: " << formatv("{0:P}", WeightedGap) << "\n";
- if (BBWorstGap)
- OS << "Worst gap: " << formatv("{0:P}", WorstGap)
- << " at BB with input offset: 0x"
- << Twine::utohexstr(BBWorstGap->getInputOffset()) << "\n";
- if (BBWorstGapAbs)
- OS << "Worst gap (absolute value): " << WorstGapAbs << " at BB with "
- << "input offset 0x"
- << Twine::utohexstr(BBWorstGapAbs->getInputOffset()) << "\n";
- if (opts::Verbosity >= 3)
- Function->dump();
- }
-
- CFGGapsWeightedAvg.push_back(WeightedGap);
- CFGGapsWorst.push_back(WorstGap);
- CFGGapsWorstAbs.push_back(WorstGapAbs);
+ double WeightedGap = WeightedGapSum;
+ if (WeightSum > 0)
+ WeightedGap /= WeightSum;
+ if (opts::Verbosity >= 2 && WorstGap >= 0.9) {
+ OS << "Nontrivial CFG gap observed in function "
+ << Function->getPrintName() << "\n"
+ << "Weighted gap: " << formatv("{0:P}", WeightedGap) << "\n";
+ if (BBWorstGap)
+ OS << "Worst gap: " << formatv("{0:P}", WorstGap)
+ << " at BB with input offset: 0x"
+ << Twine::utohexstr(BBWorstGap->getInputOffset()) << "\n";
+ if (BBWorstGapAbs)
+ OS << "Worst gap (absolute value): " << WorstGapAbs << " at BB with "
+ << "input offset 0x"
+ << Twine::utohexstr(BBWorstGapAbs->getInputOffset()) << "\n";
+ if (opts::Verbosity >= 3)
+ Function->dump();
}
- }
-
- if (CFGGapsWeightedAvg.empty()) {
- OS << "CFG flow conservation gap 0.00% (weighted) 0.00% (worst)\n";
- return;
+ CFGGapsWeightedAvg.push_back(WeightedGap);
+ CFGGapsWorst.push_back(WorstGap);
+ CFGGapsWorstAbs.push_back(WorstGapAbs);
}
llvm::sort(CFGGapsWeightedAvg);
@@ -355,7 +375,7 @@ void printCFGFlowConservationStats(raw_ostream &OS,
llvm::sort(CFGGapsWorst);
const int RankW =
int(CFGGapsWorst.size() * opts::PercentileForProfileQualityCheck / 100);
- OS << formatv("CFG flow conservation gap {0:P} (weighted) {1:P} (worst)\n",
+ OS << formatv("CFG flow conservation gap {0:P} (weighted) {1:P} (worst); ",
CFGGapsWeightedAvg[RankWA], CFGGapsWorst[RankW]);
if (opts::Verbosity >= 1) {
OS << "distribution of weighted CFG flow conservation gaps\n";
@@ -372,6 +392,74 @@ void printCFGFlowConservationStats(raw_ostream &OS,
}
}
+void printExceptionHandlingStats(const BinaryContext &BC, raw_ostream &OS,
+ iterator_range<function_iterator> &Functions) {
+ std::vector<double> LPCountFractionsOfTotalBBEC;
+ std::vector<double> LPCountFractionsOfTotalInvokeEC;
+ for (const BinaryFunction *Function : Functions) {
+ size_t LPECSum = 0;
+ size_t BBECSum = 0;
+ size_t InvokeECSum = 0;
+ for (BinaryBasicBlock &BB : *Function) {
+ const size_t BBEC = BB.getKnownExecutionCount();
+ BBECSum += BBEC;
+ if (BB.isLandingPad())
+ LPECSum += BBEC;
+ for (const MCInst &Inst : BB) {
+ if (!BC.MIB->isCall(Inst))
+ continue;
+ if (BC.MIB->isInvoke(Inst)) {
+ const std::optional<MCPlus::MCLandingPad> EHInfo =
+ BC.MIB->getEHInfo(Inst);
+ if (EHInfo->first)
+ InvokeECSum += BBEC;
+ }
+ }
+ }
+ // We only consider functions with at least MinLPECSum counts in landing
+ // pads to avoid false positives due to sampling noise
+ const uint16_t MinLPECSum = 50;
+ if (LPECSum <= MinLPECSum) {
+ LPCountFractionsOfTotalBBEC.push_back(0.0);
+ LPCountFractionsOfTotalInvokeEC.push_back(0.0);
+ continue;
+ }
+ const double FracTotalBBEC = (double)LPECSum / BBECSum;
+ const double FracTotalInvokeEC = (double)LPECSum / InvokeECSum;
+ LPCountFractionsOfTotalBBEC.push_back(FracTotalBBEC);
+ LPCountFractionsOfTotalInvokeEC.push_back(FracTotalInvokeEC);
+
+ if (opts::Verbosity >= 2 && FracTotalInvokeEC >= 0.05) {
+ OS << "Non-trivial usage of exception handling observed in function "
+ << Function->getPrintName() << "\n"
+ << formatv(
+ "Fraction of total InvokeEC that goes to landing pads: {0:P}\n",
+ FracTotalInvokeEC);
+ if (opts::Verbosity >= 3)
+ Function->dump();
+ }
+ }
+
+ llvm::sort(LPCountFractionsOfTotalBBEC);
+ const int RankBBEC = int(LPCountFractionsOfTotalBBEC.size() *
+ opts::PercentileForProfileQualityCheck / 100);
+ llvm::sort(LPCountFractionsOfTotalInvokeEC);
+ const int RankInvoke = int(LPCountFractionsOfTotalInvokeEC.size() *
+ opts::PercentileForProfileQualityCheck / 100);
+ OS << formatv("exception handling usage {0:P} (of total BBEC) {1:P} (of "
+ "total InvokeEC)\n",
+ LPCountFractionsOfTotalBBEC[RankBBEC],
+ LPCountFractionsOfTotalInvokeEC[RankInvoke]);
+ if (opts::Verbosity >= 1) {
+ OS << "distribution of exception handling usage as a fraction of total "
+ "BBEC of each function\n";
+ printDistribution(OS, LPCountFractionsOfTotalBBEC, /*Fraction=*/true);
+ OS << "distribution of exception handling usage as a fraction of total "
+ "InvokeEC of each function\n";
+ printDistribution(OS, LPCountFractionsOfTotalInvokeEC, /*Fraction=*/true);
+ }
+}
+
void computeFlowMappings(const BinaryContext &BC, FlowInfo &TotalFlowMap) {
// Increment block inflow and outflow with CFG jump counts.
TotalFlowMapTy &TotalIncomingFlows = TotalFlowMap.TotalIncomingFlows;
@@ -526,8 +614,8 @@ void printAll(BinaryContext &BC, FunctionListType &ValidFunctions,
100 - opts::PercentileForProfileQualityCheck);
printCFGContinuityStats(BC.outs(), Functions);
printCallGraphFlowConservationStats(BC.outs(), Functions, TotalFlowMap);
- printCFGFlowConservationStats(BC.outs(), Functions, TotalFlowMap);
-
+ printCFGFlowConservationStats(BC, BC.outs(), Functions, TotalFlowMap);
+ printExceptionHandlingStats(BC, BC.outs(), Functions);
// Print more detailed bucketed stats if requested.
if (opts::Verbosity >= 1 && RealNumTopFunctions >= 5) {
const size_t PerBucketSize = RealNumTopFunctions / 5;
@@ -557,7 +645,8 @@ void printAll(BinaryContext &BC, FunctionListType &ValidFunctions,
MaxFunctionExecutionCount);
printCFGContinuityStats(BC.outs(), Functions);
printCallGraphFlowConservationStats(BC.outs(), Functions, TotalFlowMap);
- printCFGFlowConservationStats(BC.outs(), Functions, TotalFlowMap);
+ printCFGFlowConservationStats(BC, BC.outs(), Functions, TotalFlowMap);
+ printExceptionHandlingStats(BC, BC.outs(), Functions);
}
}
}
diff --git a/bolt/test/X86/profile-quality-reporting-small-binary.s b/bolt/test/X86/profile-quality-reporting-small-binary.s
index 603d5c3218bc3..2b147c5eca81e 100644
--- a/bolt/test/X86/profile-quality-reporting-small-binary.s
+++ b/bolt/test/X86/profile-quality-reporting-small-binary.s
@@ -8,7 +8,7 @@
# RUN: llvm-bolt %t.exe -o %t.bolt --data=%t.fdata \
# RUN: 2>&1 | FileCheck %s
-# CHECK: BOLT-INFO: profile quality metrics for the hottest 2 functions (reporting top 5% values): function CFG discontinuity 0.00%; call graph flow conservation gap 0.00%; CFG flow conservation gap 0.00% (weighted) 0.00% (worst)
+# CHECK: BOLT-INFO: profile quality metrics for the hottest 2 functions (reporting top 5% values): function CFG discontinuity 0.00%; call graph flow conservation gap 0.00%; CFG flow conservation gap 0.00% (weighted) 0.00% (worst); exception handling usage 0.00% (of total BBEC) 0.00% (of total InvokeEC)
# CHECK-NEXT: BOLT-INFO:
.text
diff --git a/bolt/test/X86/profile-quality-reporting.test b/bolt/test/X86/profile-quality-reporting.test
index 2e15a6b245afa..210d3e10a3890 100644
--- a/bolt/test/X86/profile-quality-reporting.test
+++ b/bolt/test/X86/profile-quality-reporting.test
@@ -1,4 +1,4 @@
## Check profile quality stats reporting
RUN: yaml2obj %p/Inputs/blarge_new.yaml &> %t.exe
RUN: llvm-bolt %t.exe -o %t.out --pa -p %p/Inputs/blarge_new.preagg.txt | FileCheck %s
-CHECK: profile quality metrics for the hottest 5 functions (reporting top 5% values): function CFG discontinuity 100.00%; call graph flow conservation gap 60.00%; CFG flow conservation gap 45.53% (weighted) 96.87% (worst)
+CHECK: profile quality metrics for the hottest 5 functions (reporting top 5% values): function CFG discontinuity 100.00%; call graph flow conservation gap 60.00%; CFG flow conservation gap 45.53% (weighted) 96.87% (worst); exception handling usage 0.00% (of total BBEC) 0.00% (of total InvokeEC)
More information about the llvm-commits
mailing list