Further simplify the UpdateCorpusDatabase for single test. PiperOrigin-RevId: 938858039
diff --git a/centipede/centipede_interface.cc b/centipede/centipede_interface.cc index 0131ab6..e07ca4b 100644 --- a/centipede/centipede_interface.cc +++ b/centipede/centipede_interface.cc
@@ -350,12 +350,12 @@ PeriodicAction::ZeroDelayConstInterval(absl::Seconds(15))}; } -int UpdateCorpusDatabaseForFuzzTests( - Environment env, CentipedeCallbacksFactory& callbacks_factory) { - FUZZTEST_LOG(INFO) - << "Starting the update of the corpus database for fuzz tests:" - << "\nBinary: " << env.binary - << "\nCorpus database: " << env.fuzztest_corpus_database; +int UpdateCorpusDatabase(Environment env, + CentipedeCallbacksFactory& callbacks_factory) { + FUZZTEST_LOG(INFO) << "Starting the update of the corpus database for:" + << "\nFuzz test: " << env.test_name + << "\nBinary: " << env.binary + << "\nCorpus database: " << env.fuzztest_corpus_database; // Step 1: Preliminary set up of test sharding, binary info, etc. const auto [test_shard_index, total_test_shards] = SetUpTestSharding(); @@ -378,10 +378,6 @@ return stamp; }(); - std::vector<std::string> fuzz_tests_to_run = {env.test_name}; - FUZZTEST_LOG(INFO) << "Fuzz tests to run: " - << absl::StrJoin(fuzz_tests_to_run, ", "); - const bool is_workdir_specified = !env.workdir.empty(); // When env.workdir is empty, the full workdir paths will be formed by // appending the fuzz test names to the base workdir path. We use different @@ -405,200 +401,190 @@ int exit_code = EXIT_SUCCESS; - // Step 2: Iterate over the fuzz tests and run them. - const std::string binary = env.binary; - for (int i = 0; i < fuzz_tests_to_run.size(); ++i) { - // Clean up previous stop requests. stop_time will be set later. - ClearEarlyStopRequestAndSetStopTime(/*stop_time=*/absl::InfiniteFuture()); + // Step 2: Run the fuzz test. - if (!is_workdir_specified) { - env.workdir = base_workdir_path / fuzz_tests_to_run[i]; - } - const auto execution_id_path = - (base_workdir_path / - absl::StrCat(fuzz_tests_to_run[i], ".execution_id")) - .string(); + // Clean up previous stop requests. stop_time will be set later. + ClearEarlyStopRequestAndSetStopTime(/*stop_time=*/absl::InfiniteFuture()); - bool is_resuming = false; - if (!is_workdir_specified && !env.fuzztest_execution_id.empty()) { - // Use the execution IDs to resume or skip tests. - const bool execution_id_matched = [&] { - if (!RemotePathExists(execution_id_path)) return false; - FUZZTEST_CHECK(!RemotePathIsDirectory(execution_id_path)); - std::string prev_execution_id; - FUZZTEST_CHECK_OK( - RemoteFileGetContents(execution_id_path, prev_execution_id)); - return prev_execution_id == env.fuzztest_execution_id; - }(); - if (execution_id_matched) { - // If execution IDs match but the previous coverage is missing, it means - // the test was previously finished, and we skip running for the test. - if (!RemotePathExists(WorkDir{env}.CoverageDirPath())) { - FUZZTEST_LOG(INFO) - << "Skipping running the fuzz test " << fuzz_tests_to_run[i]; - continue; - } - // If execution IDs match and the previous coverage exists, it means - // the same workflow got interrupted when running the test. So we resume - // the test. - is_resuming = true; - FUZZTEST_LOG(INFO) << "Resuming running the fuzz test " - << fuzz_tests_to_run[i]; - } else { - // If the execution IDs mismatch, we start a new run. - is_resuming = false; - FUZZTEST_LOG(INFO) << "Starting a new run of the fuzz test " - << fuzz_tests_to_run[i]; + if (!is_workdir_specified) { + env.workdir = base_workdir_path / env.test_name; + } + const auto execution_id_path = + (base_workdir_path / absl::StrCat(env.test_name, ".execution_id")) + .string(); + + bool is_resuming = false; + if (!is_workdir_specified && !env.fuzztest_execution_id.empty()) { + // Use the execution IDs to resume or skip tests. + const bool execution_id_matched = [&] { + if (!RemotePathExists(execution_id_path)) return false; + FUZZTEST_CHECK(!RemotePathIsDirectory(execution_id_path)); + std::string prev_execution_id; + FUZZTEST_CHECK_OK( + RemoteFileGetContents(execution_id_path, prev_execution_id)); + return prev_execution_id == env.fuzztest_execution_id; + }(); + if (execution_id_matched) { + // If execution IDs match but the previous coverage is missing, it means + // the test was previously finished, and we skip running for the test. + if (!RemotePathExists(WorkDir{env}.CoverageDirPath())) { + FUZZTEST_LOG(INFO) << "Skipping running the fuzz test " + << env.test_name; + return exit_code; } + // If execution IDs match and the previous coverage exists, it means + // the same workflow got interrupted when running the test. So we resume + // the test. + is_resuming = true; + FUZZTEST_LOG(INFO) << "Resuming running the fuzz test " << env.test_name; + } else { + // If the execution IDs mismatch, we start a new run. + is_resuming = false; + FUZZTEST_LOG(INFO) << "Starting a new run of the fuzz test " + << env.test_name; } - if (RemotePathExists(env.workdir) && !is_resuming) { - // This could be a workdir from a failed run that used a different version - // of the binary. We delete it so that we don't have to deal with - // the assumptions under which it is safe to reuse an old workdir. + } + + if (RemotePathExists(env.workdir) && !is_resuming) { + // This could be a workdir from a failed run that used a different version + // of the binary. We delete it so that we don't have to deal with + // the assumptions under which it is safe to reuse an old workdir. + FUZZTEST_CHECK_OK(RemotePathDelete(env.workdir, /*recursively=*/true)); + } + const WorkDir workdir{env}; + FUZZTEST_CHECK_OK(RemoteMkdir( + workdir.CoverageDirPath())); // Implicitly creates the workdir + + // Updating execution ID must be after creating the coverage dir. Otherwise + // if it fails to create coverage dir after updating execution ID, next + // attempt would skip this test. + if (!is_workdir_specified && !env.fuzztest_execution_id.empty() && + !is_resuming) { + FUZZTEST_CHECK_OK( + RemoteFileSetContents(execution_id_path, env.fuzztest_execution_id)); + } + + absl::Cleanup clean_up_workdir = [is_workdir_specified, &env] { + if (!is_workdir_specified && !EarlyStopRequested()) { FUZZTEST_CHECK_OK(RemotePathDelete(env.workdir, /*recursively=*/true)); } - const WorkDir workdir{env}; - FUZZTEST_CHECK_OK(RemoteMkdir( - workdir.CoverageDirPath())); // Implicitly creates the workdir + }; - // Updating execution ID must be after creating the coverage dir. Otherwise - // if it fails to create coverage dir after updating execution ID, next - // attempt would skip this test. - if (!is_workdir_specified && !env.fuzztest_execution_id.empty() && - !is_resuming) { - FUZZTEST_CHECK_OK( - RemoteFileSetContents(execution_id_path, env.fuzztest_execution_id)); + const std::filesystem::path fuzztest_db_path = + corpus_database_path / env.test_name; + const std::filesystem::path regression_dir = fuzztest_db_path / "regression"; + const std::filesystem::path coverage_dir = fuzztest_db_path / "coverage"; + + // Seed the fuzzing session with the latest coverage corpus and regression + // inputs from the previous fuzzing session. + if (!is_resuming) { + FUZZTEST_CHECK_OK(GenerateSeedCorpusFromConfig( + GetSeedCorpusConfig( + env, regression_dir.c_str(), + env.fuzztest_replay_coverage_inputs ? coverage_dir.c_str() : ""), + env.binary_name, env.binary_hash)) + << "while generating the seed corpus"; + } + + absl::Duration time_limit = env.fuzztest_time_limit_per_test; + absl::Duration time_spent = absl::ZeroDuration(); + const std::string fuzzing_time_file = + std::filesystem::path(env.workdir) / "fuzzing_time"; + if (is_resuming && RemotePathExists(fuzzing_time_file)) { + time_spent = ReadFuzzingTime(fuzzing_time_file); + time_limit = std::max(time_limit - time_spent, absl::ZeroDuration()); + } + is_resuming = false; + + if (EarlyStopRequested()) { + if (ExitCode() != EXIT_SUCCESS) { + exit_code = ExitCode(); + FUZZTEST_LOG(ERROR) << "Early stop requested for test " << env.test_name + << " with failure exit code " << exit_code; + } else { + FUZZTEST_LOG(INFO) << "Skipping test " << env.test_name + << " due to early stop requested without failure."; } + return exit_code; + } - absl::Cleanup clean_up_workdir = [is_workdir_specified, &env] { - if (!is_workdir_specified && !EarlyStopRequested()) { - FUZZTEST_CHECK_OK(RemotePathDelete(env.workdir, /*recursively=*/true)); - } - }; + FUZZTEST_LOG(INFO) << (env.fuzztest_only_replay ? "Replaying " : "Fuzzing ") + << env.test_name << " for " << time_limit + << "\n\tTest binary: " << env.binary; - const std::filesystem::path fuzztest_db_path = - corpus_database_path / fuzz_tests_to_run[i]; - const std::filesystem::path regression_dir = - fuzztest_db_path / "regression"; - const std::filesystem::path coverage_dir = fuzztest_db_path / "coverage"; + const absl::Time start_time = absl::Now(); + ClearEarlyStopRequestAndSetStopTime(/*stop_time=*/start_time + time_limit); + PeriodicAction record_fuzzing_time = + RecordFuzzingTime(fuzzing_time_file, start_time - time_spent); + Fuzz(env, binary_info, pcs_file_path, callbacks_factory); + record_fuzzing_time.Nudge(); + record_fuzzing_time.Stop(); - // Seed the fuzzing session with the latest coverage corpus and regression - // inputs from the previous fuzzing session. - if (!is_resuming) { - FUZZTEST_CHECK_OK(GenerateSeedCorpusFromConfig( - GetSeedCorpusConfig( - env, regression_dir.c_str(), - env.fuzztest_replay_coverage_inputs ? coverage_dir.c_str() : ""), - env.binary_name, env.binary_hash)) - << "while generating the seed corpus"; + if (!stats_root_path.empty()) { + const auto stats_dir = stats_root_path / env.test_name; + FUZZTEST_CHECK_OK(RemoteMkdir(stats_dir.c_str())); + FUZZTEST_CHECK_OK(RemoteFileRename( + workdir.FuzzingStatsPath(), + (stats_dir / absl::StrCat("fuzzing_stats_", execution_stamp)).c_str())); + } + + if (EarlyStopRequested()) { + if (ExitCode() != EXIT_SUCCESS) { + exit_code = ExitCode(); + FUZZTEST_LOG(ERROR) << "Early stop requested for test " << env.test_name + << " with failure exit code " << exit_code; + } else { + FUZZTEST_LOG(INFO) << "Skip updating corpus database due to early stop " + "requested without failure."; } + return exit_code; + } + // The test time limit does not apply for the rest of the steps. + ClearEarlyStopRequestAndSetStopTime(/*stop_time=*/absl::InfiniteFuture()); - absl::Duration time_limit = env.fuzztest_time_limit_per_test; - absl::Duration time_spent = absl::ZeroDuration(); - const std::string fuzzing_time_file = - std::filesystem::path(env.workdir) / "fuzzing_time"; - if (is_resuming && RemotePathExists(fuzzing_time_file)) { - time_spent = ReadFuzzingTime(fuzzing_time_file); - time_limit = std::max(time_limit - time_spent, absl::ZeroDuration()); + // TODO(xinhaoyuan): Have a separate flag to skip corpus updating instead + // of checking whether workdir is specified or not. + const bool skip_corpus_db_update = + env.fuzztest_only_replay || is_workdir_specified; + if (skip_corpus_db_update && !env.report_crash_summary) return exit_code; + + // Deduplicate and optionally update the crashing inputs. + CrashSummary crash_summary{env.fuzztest_binary_identifier, env.test_name}; + const absl::flat_hash_map<std::string, CrashDetails> crashes_by_signature = + GetCrashesFromWorkdir(workdir, env.total_shards); + if (skip_corpus_db_update) { + // Just report the crashes. + FUZZTEST_CHECK(env.report_crash_summary); + for (const auto& [crash_signature, crash_details] : crashes_by_signature) { + crash_summary.AddCrash({/*id=*/crash_details.input_signature, + /*category=*/crash_details.description, + crash_signature, crash_details.description}); } - is_resuming = false; + crash_summary.Report(&std::cerr); + return exit_code; + } + OrganizeCrashingInputs(regression_dir, fuzztest_db_path / "crashing", env, + callbacks_factory, crashes_by_signature, + crash_summary); + if (env.report_crash_summary) crash_summary.Report(&std::cerr); - if (EarlyStopRequested()) { - if (ExitCode() != EXIT_SUCCESS) { - exit_code = ExitCode(); - FUZZTEST_LOG(ERROR) - << "Early stop requested for test " << fuzz_tests_to_run[i] - << " with failure exit code " << exit_code; - } else { - FUZZTEST_LOG(INFO) << "Skipping test " << fuzz_tests_to_run[i] - << " due to early stop requested without failure."; - } - continue; - } - - FUZZTEST_LOG(INFO) << (env.fuzztest_only_replay ? "Replaying " : "Fuzzing ") - << fuzz_tests_to_run[i] << " for " << time_limit - << "\n\tTest binary: " << env.binary; - - const absl::Time start_time = absl::Now(); - ClearEarlyStopRequestAndSetStopTime(/*stop_time=*/start_time + time_limit); - PeriodicAction record_fuzzing_time = - RecordFuzzingTime(fuzzing_time_file, start_time - time_spent); - Fuzz(env, binary_info, pcs_file_path, callbacks_factory); - record_fuzzing_time.Nudge(); - record_fuzzing_time.Stop(); - - if (!stats_root_path.empty()) { - const auto stats_dir = stats_root_path / fuzz_tests_to_run[i]; - FUZZTEST_CHECK_OK(RemoteMkdir(stats_dir.c_str())); - FUZZTEST_CHECK_OK(RemoteFileRename( - workdir.FuzzingStatsPath(), - (stats_dir / absl::StrCat("fuzzing_stats_", execution_stamp)) - .c_str())); - } - - if (EarlyStopRequested()) { - if (ExitCode() != EXIT_SUCCESS) { - exit_code = ExitCode(); - FUZZTEST_LOG(ERROR) - << "Early stop requested for test " << fuzz_tests_to_run[i] - << " with failure exit code " << exit_code; - } else { - FUZZTEST_LOG(INFO) << "Skip updating corpus database due to early stop " - "requested without failure."; - } - continue; - } - // The test time limit does not apply for the rest of the steps. - ClearEarlyStopRequestAndSetStopTime(/*stop_time=*/absl::InfiniteFuture()); - - // TODO(xinhaoyuan): Have a separate flag to skip corpus updating instead - // of checking whether workdir is specified or not. - const bool skip_corpus_db_update = - env.fuzztest_only_replay || is_workdir_specified; - if (skip_corpus_db_update && !env.report_crash_summary) continue; - - // Deduplicate and optionally update the crashing inputs. - CrashSummary crash_summary{env.fuzztest_binary_identifier, - fuzz_tests_to_run[i]}; - const absl::flat_hash_map<std::string, CrashDetails> crashes_by_signature = - GetCrashesFromWorkdir(workdir, env.total_shards); - if (skip_corpus_db_update) { - // Just report the crashes. - FUZZTEST_CHECK(env.report_crash_summary); - for (const auto& [crash_signature, crash_details] : - crashes_by_signature) { - crash_summary.AddCrash({/*id=*/crash_details.input_signature, - /*category=*/crash_details.description, - crash_signature, crash_details.description}); - } - crash_summary.Report(&std::cerr); - continue; - } - OrganizeCrashingInputs(regression_dir, fuzztest_db_path / "crashing", env, - callbacks_factory, crashes_by_signature, - crash_summary); - if (env.report_crash_summary) crash_summary.Report(&std::cerr); - - // Distill and store the coverage corpus. - Distill(env); - if (RemotePathExists(coverage_dir.c_str())) { - // In the future, we will store k latest coverage corpora for some k, but - // for now we only keep the latest one. - FUZZTEST_CHECK_OK( - RemotePathDelete(coverage_dir.c_str(), /*recursively=*/true)); - } - FUZZTEST_CHECK_OK(RemoteMkdir(coverage_dir.c_str())); - std::vector<std::string> distilled_corpus_files; + // Distill and store the coverage corpus. + Distill(env); + if (RemotePathExists(coverage_dir.c_str())) { + // In the future, we will store k latest coverage corpora for some k, but + // for now we only keep the latest one. FUZZTEST_CHECK_OK( - RemoteGlobMatch(workdir.DistilledCorpusFilePaths().AllShardsGlob(), - distilled_corpus_files)); - for (const std::string& corpus_file : distilled_corpus_files) { - const std::string file_name = - std::filesystem::path(corpus_file).filename(); - FUZZTEST_CHECK_OK( - RemoteFileRename(corpus_file, (coverage_dir / file_name).c_str())); - } + RemotePathDelete(coverage_dir.c_str(), /*recursively=*/true)); + } + FUZZTEST_CHECK_OK(RemoteMkdir(coverage_dir.c_str())); + std::vector<std::string> distilled_corpus_files; + FUZZTEST_CHECK_OK( + RemoteGlobMatch(workdir.DistilledCorpusFilePaths().AllShardsGlob(), + distilled_corpus_files)); + for (const std::string& corpus_file : distilled_corpus_files) { + const std::string file_name = std::filesystem::path(corpus_file).filename(); + FUZZTEST_CHECK_OK( + RemoteFileRename(corpus_file, (coverage_dir / file_name).c_str())); } return exit_code; @@ -815,7 +801,7 @@ FUZZTEST_CHECK(updated_env.fuzztest_time_limit_per_test >= absl::Seconds(1)) << "Time limit per fuzz test must be at least 1 second."; - return UpdateCorpusDatabaseForFuzzTests(updated_env, callbacks_factory); + return UpdateCorpusDatabase(updated_env, callbacks_factory); } }