Add a library for aggregating branch frequencies from perf data

PiperOrigin-RevId: 700388882

propeller/BUILD

@@ -595,6 +595,26 @@ cc_library(
     ],
 )
 
+cc_library(
+    name = "perf_branch_frequencies_aggregator",
+    srcs = ["perf_branch_frequencies_aggregator.cc"],
+    hdrs = ["perf_branch_frequencies_aggregator.h"],
+    deps = [
+        ":binary_content",
+        ":branch_frequencies",
+        ":branch_frequencies_aggregator",
+        ":perf_data_provider",
+        ":perfdata_reader",
+        ":propeller_options_cc_proto",
+        ":propeller_statistics",
+        ":resolve_mmap_name",
+        ":status_macros",
+        "@abseil-cpp//absl/log",
+        "@abseil-cpp//absl/status",
+        "@abseil-cpp//absl/status:statusor",
+    ],
+)
+
 ########################
 #  Tests & Test Utils  #
 ########################
@@ -1048,3 +1068,26 @@ cc_test(
         "@com_google_googletest//:gtest_main",
     ],
 )
+
+cc_test(
+    name = "perf_branch_frequencies_aggregator_test",
+    srcs = ["perf_branch_frequencies_aggregator_test.cc"],
+    data = [
+        "//propeller/testdata:sample.arm.bin",
+    ],
+    deps = [
+        ":binary_content",
+        ":branch_frequencies",
+        ":file_perf_data_provider",
+        ":perf_branch_frequencies_aggregator",
+        ":perf_data_provider",
+        ":propeller_options_cc_proto",
+        ":propeller_statistics",
+        ":status_testing_macros",
+        "@abseil-cpp//absl/status",
+        "@abseil-cpp//absl/status:status_matchers",
+        "@abseil-cpp//absl/status:statusor",
+        "@abseil-cpp//absl/strings",
+        "@com_google_googletest//:gtest_main",
+    ],
+)

propeller/CMakeLists.txt

@@ -43,6 +43,7 @@ add_library(propeller_lib OBJECT
   node_chain_builder.cc
   node_chain.cc
   perfdata_reader.cc
+  perf_branch_frequencies_aggregator.cc
   program_cfg_builder.cc
   program_cfg.cc
   propeller_statistics.cc
@@ -83,6 +84,7 @@ propeller_generate_tests(
     lazy_evaluator_test.cc
     lbr_branch_aggregator_test.cc
     perfdata_reader_test.cc
+    perf_branch_frequencies_aggregator_test.cc
     propeller_statistics_test.cc
     proto_branch_frequencies_aggregator_test.cc
     spe_tid_pid_provider_test.cc

propeller/perf_branch_frequencies_aggregator.cc

@@ -0,0 +1,58 @@
+#include "propeller/perf_branch_frequencies_aggregator.h"
+
+#include <optional>
+#include <string>
+#include <utility>
+
+#include "absl/log/log.h"
+#include "absl/status/status.h"
+#include "absl/status/statusor.h"
+#include "propeller/binary_content.h"
+#include "propeller/branch_frequencies.h"
+#include "propeller/perf_data_provider.h"
+#include "propeller/perfdata_reader.h"
+#include "propeller/propeller_options.pb.h"
+#include "propeller/propeller_statistics.h"
+#include "propeller/resolve_mmap_name.h"
+#include "propeller/status_macros.h"
+
+namespace propeller {
+
+absl::StatusOr<BranchFrequencies>
+PerfBranchFrequenciesAggregator::AggregateBranchFrequencies(
+    const PropellerOptions &options, const BinaryContent &binary_content,
+    PropellerStats &stats) {
+  PropellerStats::ProfileStats &profile_stats = stats.profile_stats;
+  BranchFrequencies frequencies;
+
+  while (true) {
+    ASSIGN_OR_RETURN(std::optional<PerfDataProvider::BufferHandle> perf_data,
+                     perf_data_provider_->GetNext());
+    if (!perf_data.has_value()) break;
+
+    const std::string description = perf_data->description;
+    LOG(INFO) << "Parsing " << description << " ...";
+    absl::StatusOr<PerfDataReader> perf_data_reader = BuildPerfDataReader(
+        std::move(*perf_data), &binary_content, ResolveMmapName(options));
+    if (!perf_data_reader.ok()) {
+      LOG(WARNING) << "Skipped profile " << description << ": "
+                   << perf_data_reader.status();
+      continue;
+    }
+
+    profile_stats.binary_mmap_num += perf_data_reader->binary_mmaps().size();
+    ++profile_stats.perf_file_parsed;
+    RETURN_IF_ERROR(perf_data_reader->AggregateSpe(frequencies));
+  }
+  profile_stats.br_counters_accumulated +=
+      frequencies.GetNumberOfTakenBranchCounters();
+  if (profile_stats.br_counters_accumulated <= 100)
+    LOG(WARNING) << "Too few branch records in perf data.";
+  if (profile_stats.perf_file_parsed == 0) {
+    return absl::FailedPreconditionError(
+        "No perf file is parsed, cannot proceed.");
+  }
+  return frequencies;
+}
+
+}  // namespace propeller

propeller/perf_branch_frequencies_aggregator.h

@@ -0,0 +1,48 @@
+#ifndef PROPELLER_PERF_BRANCH_FREQUENCIES_AGGREGATOR_H_
+#define PROPELLER_PERF_BRANCH_FREQUENCIES_AGGREGATOR_H_
+
+#include <memory>
+#include <utility>
+
+#include "absl/status/statusor.h"
+#include "propeller/binary_content.h"
+#include "propeller/branch_frequencies.h"
+#include "propeller/branch_frequencies_aggregator.h"
+#include "propeller/perf_data_provider.h"
+#include "propeller/propeller_options.pb.h"
+#include "propeller/propeller_statistics.h"
+
+namespace propeller {
+// `PerfBranchFrequenciesAggregator` is an implementation of
+// `BranchFrequenciesAggregator` that builds `BranchFrequencies` from perf data
+// containing SPE entries. The perf data can come from any `PerfDataProvider`,
+// such as from a file or mock.
+class PerfBranchFrequenciesAggregator : public BranchFrequenciesAggregator {
+ public:
+  explicit PerfBranchFrequenciesAggregator(
+      std::unique_ptr<PerfDataProvider> perf_data_provider)
+      : perf_data_provider_(std::move(perf_data_provider)) {}
+
+  // PerfBranchFrequenciesAggregator is move-only.
+  PerfBranchFrequenciesAggregator(PerfBranchFrequenciesAggregator&&) = default;
+  PerfBranchFrequenciesAggregator& operator=(
+      PerfBranchFrequenciesAggregator&&) = default;
+  PerfBranchFrequenciesAggregator(const PerfBranchFrequenciesAggregator&) =
+      delete;
+  PerfBranchFrequenciesAggregator& operator=(
+      const PerfBranchFrequenciesAggregator&) = delete;
+
+  // Aggregates branch frequencies from perf data, may return an `absl::Status`
+  // if the perf data can't be successfully parsed and aggregated (it doesn't
+  // exist, is malformed, etc.).
+  absl::StatusOr<BranchFrequencies> AggregateBranchFrequencies(
+      const PropellerOptions& options, const BinaryContent& binary_content,
+      PropellerStats& stats) override;
+
+ private:
+  std::unique_ptr<PerfDataProvider> perf_data_provider_;
+};
+
+}  // namespace propeller
+
+#endif  // PROPELLER_PERF_BRANCH_FREQUENCIES_AGGREGATOR_H_

propeller/perf_branch_frequencies_aggregator_test.cc

@@ -0,0 +1,64 @@
+#include "propeller/perf_branch_frequencies_aggregator.h"
+
+#include <memory>
+#include <optional>
+#include <string>
+#include <string_view>
+#include <utility>
+#include <vector>
+
+#include "propeller/status_testing_macros.h"
+#include "gmock/gmock.h"
+#include "gtest/gtest.h"
+#include "absl/status/status.h"
+#include "absl/status/statusor.h"
+#include "absl/status/status_matchers.h"
+#include "absl/strings/str_cat.h"
+#include "propeller/binary_content.h"
+#include "propeller/branch_frequencies.h"
+#include "propeller/file_perf_data_provider.h"
+#include "propeller/perf_data_provider.h"
+#include "propeller/propeller_options.pb.h"
+#include "propeller/propeller_statistics.h"
+
+namespace propeller {
+namespace {
+using ::testing::AllOf;
+using ::testing::Field;
+using ::testing::SizeIs;
+using ::absl_testing::IsOkAndHolds;
+using ::absl_testing::StatusIs;
+
+class MockPerfDataProvider : public PerfDataProvider {
+ public:
+  MOCK_METHOD(absl::StatusOr<std::optional<PerfDataProvider::BufferHandle>>,
+              GetNext, ());
+  MOCK_METHOD(absl::StatusOr<std::vector<PerfDataProvider::BufferHandle>>,
+              GetAllAvailableOrNext, ());
+};
+
+static constexpr std::string_view kTestDataDir =
+    "propeller/testdata/";
+
+TEST(PerfBranchFrequenciesAggregator, FailsIfNoPerfData) {
+  auto mock_perf_data_provider = std::make_unique<MockPerfDataProvider>();
+
+  // PerfDataProvider::BufferHandle is not copyable, so we can't use
+  // `testing::Return` here.
+  EXPECT_CALL(*mock_perf_data_provider, GetNext()).WillRepeatedly([]() {
+    return absl::InvalidArgumentError("No perf data");
+  });
+  EXPECT_CALL(*mock_perf_data_provider, GetAllAvailableOrNext())
+      .WillRepeatedly(
+          []() { return absl::InvalidArgumentError("No perf data"); });
+
+  PropellerStats stats;
+  EXPECT_THAT(
+      PerfBranchFrequenciesAggregator(std::move(mock_perf_data_provider))
+          .AggregateBranchFrequencies(PropellerOptions(), BinaryContent(),
+                                      stats),
+      StatusIs(absl::StatusCode::kInvalidArgument));
+}
+
+}  // namespace
+}  // namespace propeller