resolve comments
diff --git a/llvm/test/tools/llvm-profgen/afdo-with-vtable-pie.test b/llvm/test/tools/llvm-profgen/afdo-with-vtable-pie.test
index b51cf33..56f888d 100644
--- a/llvm/test/tools/llvm-profgen/afdo-with-vtable-pie.test
+++ b/llvm/test/tools/llvm-profgen/afdo-with-vtable-pie.test
@@ -1,6 +1,5 @@
-
RUN: llvm-profgen --perfscript=%p/Inputs/pie-lbr-perf.script \
-RUN: --data-access-profile=%p/Inputs/pie-dap-perf.txt \
+RUN: --data-access-perftrace=%p/Inputs/pie-dap-perf.txt \
RUN: --binary=%p/Inputs/dap-pie.bin --format=text --pid=1725662 \
RUN: -ignore-stack-samples -use-dwarf-correlation -o %t.afdo
diff --git a/llvm/test/tools/llvm-profgen/afdo-with-vtable.test b/llvm/test/tools/llvm-profgen/afdo-with-vtable.test
index 14d743c..65b601e4 100644
--- a/llvm/test/tools/llvm-profgen/afdo-with-vtable.test
+++ b/llvm/test/tools/llvm-profgen/afdo-with-vtable.test
@@ -1,4 +1,4 @@
-RUN: llvm-profgen --perfscript=%p/Inputs/lbr-perf-for-dap.script --data-access-profile=%p/Inputs/dap-perf-trace.txt \
+RUN: llvm-profgen --perfscript=%p/Inputs/lbr-perf-for-dap.script --data-access-perftrace=%p/Inputs/dap-perf-trace.txt \
RUN: --binary=%p/Inputs/dap.bin --format=text --pid=3446532 \
RUN: -ignore-stack-samples -use-dwarf-correlation -o %t.afdo
diff --git a/llvm/tools/llvm-profgen/ProfileGenerator.cpp b/llvm/tools/llvm-profgen/ProfileGenerator.cpp
index ef66160..41ed191 100644
--- a/llvm/tools/llvm-profgen/ProfileGenerator.cpp
+++ b/llvm/tools/llvm-profgen/ProfileGenerator.cpp
@@ -539,22 +539,7 @@
populateBodySamplesForAllFunctions(SC.RangeCounter);
// Fill in boundary sample counts as well as call site samples for calls
populateBoundarySamplesForAllFunctions(SC.BranchCounter);
-
- // For each instruction with vtable accesses, get its symbolized inline
- // stack, and add the vtable counters to the function samples.
- for (const auto &[IpData, Count] : SC.DataAccessCounter) {
- uint64_t InstAddr = IpData.first;
- const SampleContextFrameVector &FrameVec =
- Binary->getCachedFrameLocationStack(InstAddr, false);
- if (!FrameVec.empty()) {
- FunctionSamples &FunctionProfile =
- getLeafProfileAndAddTotalSamples(FrameVec, 0);
- LineLocation Loc(
- FrameVec.back().Location.LineOffset,
- getBaseDiscriminator(FrameVec.back().Location.Discriminator));
- FunctionProfile.addTypeSamplesAt(Loc, FunctionId(IpData.second), Count);
- }
- }
+ populateTypeSamplesForAllFunctions(SC.DataAccessCounter);
updateFunctionSamples();
}
@@ -621,8 +606,8 @@
getLeafProfileAndAddTotalSamples(FrameVec, 0);
FunctionProfile.addCalledTargetSamples(
FrameVec.back().Location.LineOffset,
- FrameVec.back().Location.Discriminator,
- FunctionId(CalleeName), Count);
+ FrameVec.back().Location.Discriminator, FunctionId(CalleeName),
+ Count);
}
}
}
@@ -645,8 +630,7 @@
getBaseDiscriminator(FrameVec[I - 1].Location.Discriminator));
FunctionSamplesMap &SamplesMap =
FunctionProfile->functionSamplesAt(Callsite);
- auto Ret =
- SamplesMap.emplace(FrameVec[I].Func, FunctionSamples());
+ auto Ret = SamplesMap.emplace(FrameVec[I].Func, FunctionSamples());
if (Ret.second) {
SampleContext Context(FrameVec[I].Func);
Ret.first->second.setContext(Context);
@@ -761,6 +745,26 @@
}
}
+void ProfileGenerator::populateTypeSamplesForAllFunctions(
+ const DataAccessSample &DataAccessSamples) {
+ // For each instruction with vtable accesses, get its symbolized inline
+ // stack, and add the vtable counters to the function samples.
+ for (const auto &[IpData, Count] : DataAccessSamples) {
+ uint64_t InstAddr = IpData.first;
+ const SampleContextFrameVector &FrameVec =
+ Binary->getCachedFrameLocationStack(InstAddr,
+ /* UseProbeDiscriminator=*/false);
+ if (!FrameVec.empty()) {
+ FunctionSamples &FunctionProfile =
+ getLeafProfileAndAddTotalSamples(FrameVec, 0);
+ LineLocation Loc(
+ FrameVec.back().Location.LineOffset,
+ getBaseDiscriminator(FrameVec.back().Location.Discriminator));
+ FunctionProfile.addTypeSamplesAt(Loc, FunctionId(IpData.second), Count);
+ }
+ }
+}
+
void ProfileGeneratorBase::calculateBodySamplesAndSize(
const FunctionSamples &FSamples, uint64_t &TotalBodySamples,
uint64_t &FuncBodySize) {
diff --git a/llvm/tools/llvm-profgen/ProfileGenerator.h b/llvm/tools/llvm-profgen/ProfileGenerator.h
index 5e361285..bafdc18 100644
--- a/llvm/tools/llvm-profgen/ProfileGenerator.h
+++ b/llvm/tools/llvm-profgen/ProfileGenerator.h
@@ -182,6 +182,8 @@
populateBodySamplesWithProbesForAllFunctions(const RangeSample &RangeCounter);
void populateBoundarySamplesWithProbesForAllFunctions(
const BranchSample &BranchCounters);
+ void
+ populateTypeSamplesForAllFunctions(const DataAccessSample &DataAccessSamples);
void postProcessProfiles();
void trimColdProfiles(const SampleProfileMap &Profiles,
uint64_t ColdCntThreshold);
diff --git a/llvm/tools/llvm-profgen/llvm-profgen.cpp b/llvm/tools/llvm-profgen/llvm-profgen.cpp
index 7924844..d36e29e 100644
--- a/llvm/tools/llvm-profgen/llvm-profgen.cpp
+++ b/llvm/tools/llvm-profgen/llvm-profgen.cpp
@@ -68,8 +68,9 @@
cl::cat(ProfGenCategory));
static cl::opt<std::string> DataAccessProfileFilename(
- "data-access-profile", cl::value_desc("data-access-profile"),
- cl::desc("Path to the data access profile to be generated."),
+ "data-access-perftrace", cl::value_desc("data-access-perftrace"),
+ cl::desc("File path of a Linux perf raw trace (generated by `perf report "
+ "-D`) consisting of memory access events."),
cl::cat(ProfGenCategory));
extern cl::opt<bool> ShowDisassemblyOnly;
