Collect unclean shutdown debug information

components/browser_watcher/postmortem_report_collector.cc

+// Copyright 2016 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "components/browser_watcher/postmortem_report_collector.h"
+
+#include <utility>
+
+#include "base/debug/activity_analyzer.h"
+#include "base/files/file_enumerator.h"
+#include "base/files/file_util.h"
+#include "base/logging.h"
+#include "base/path_service.h"
+#include "components/browser_watcher/postmortem_minidump_writer.h"
+#include "third_party/crashpad/crashpad/client/settings.h"
+#include "third_party/crashpad/crashpad/util/misc/uuid.h"
+
+using base::FilePath;
+
+namespace browser_watcher {
+
+using base::debug::GlobalActivityAnalyzer;
+using base::debug::ThreadActivityAnalyzer;
+using crashpad::CrashReportDatabase;
+
+namespace {
+
+// An object that deletes a file when it goes out of scope.
+class FileDeleter {
+ public:
+  explicit FileDeleter(FilePath path) : file_path_(path) {}
+  ~FileDeleter() {
+    if (!base::DeleteFile(file_path_, false))
+      LOG(ERROR) << "Failed to delete " << file_path_.value();
+  }
+
+ private:
+  FilePath file_path_;
+};
+
+}  // namespace
+
+PostmortemReportCollector::PostmortemReportCollector(
+    const FilePath& debug_dir,
+    const FilePath::StringType& debug_file_pattern,
+    std::unique_ptr<crashpad::CrashReportDatabase> report_database)
+    : debug_state_dir_(debug_dir),
+      debug_file_pattern_(debug_file_pattern),
+      report_database_(std::move(report_database)) {
+  DCHECK_NE(true, debug_dir.empty());
+  DCHECK_NE(true, debug_file_pattern.empty());
+  DCHECK_NE(nullptr, report_database_.get());
+}
+
+// TODO(manzagop): throttling and graceful handling of too much data.
+void PostmortemReportCollector::CollectAndSubmitForUpload(
+    const std::set<FilePath>& excluded_debug_files) {
+  DCHECK_NE(nullptr, report_database_.get());
+
+  // Collect the list of files to harvest.
+  std::vector<FilePath> debug_files =
+      GetDebugStateFilePaths(excluded_debug_files);
+
+  // Determine the crashpad client id.
+  crashpad::UUID client_id;
+  crashpad::Settings* settings = report_database_->GetSettings();
+  if (settings) {
+    // If GetSettings() or GetClientID() fails client_id will be left at its
+    // default value, all zeroes, which is appropriate.
+    settings->GetClientID(&client_id);
+  }
+
+  // Note: the code below involves two notions of report:
+  //   - crashpad reports
+  //   - chrome internal state reports, which are reported wrapped inside a
+  //     crashpad report
+  for (const FilePath& file : debug_files) {
+    FileDeleter debug_file_deleter(file);

[Sigurður Ásgeirsson, 2016/08/11 17:44:15]

The intent here seems to be to make at most one reporting attempt per found
file.
However, there's a problem with "stuck" files, such as files where for whatever
reason this process doesn't have delete permission, or the file's already open
by someone else, etc.

In order to avoid reporting a "stuck" file for perpetuity, I'd condition the
report on being able to delete the file. That way a "stuck" file is never
reported.
I think under-reporting is the lesser evil there?

[manzagop (departed), 2016/08/12 21:23:22]

Good point. Done.

+
+    // Collect the data from the debug file to a proto.
+    std::unique_ptr<StabilityReport> report_proto = Collect(file);
+    if (!report_proto) {
+      // The file was empty, or there was an error collecting the data. Detailed
+      // logging happens within the Collect function.
+      continue;
+    }
+
+    // Prepare a crashpad report.
+    CrashReportDatabase::NewReport* new_report;
+    CrashReportDatabase::OperationStatus database_status =
+        report_database_->PrepareNewCrashReport(&new_report);
+    if (database_status != CrashReportDatabase::kNoError) {
+      // Note that we delete the report even in this case, to prevent report
+      // accumulation.
+      LOG(ERROR) << "PrepareNewCrashReport failed";
+      continue;
+    }
+    CrashReportDatabase::CallErrorWritingCrashReport
+        call_error_writing_crash_report(report_database_.get(), new_report);
+
+    // Write the report to a minidump.
+    if (!WriteReportToMinidump(*report_proto, client_id, new_report->uuid,
+                               reinterpret_cast<FILE*>(new_report->handle))) {
+      // Note that we delete the report even in this case, to prevent report
+      // accumulation. Detailed logging happens within the function.
+      continue;
+    }
+
+    // Finalize the report wrt the report database.
+    call_error_writing_crash_report.Disarm();
+    crashpad::UUID unused_report_id;
+    database_status = report_database_->FinishedWritingCrashReport(
+        new_report, &unused_report_id);
+    if (database_status != CrashReportDatabase::kNoError) {
+      // Note that we delete the report even in this case, to prevent report
+      // accumulation.
+      LOG(ERROR) << "FinishedWritingCrashReport failed";
+      continue;
+    }
+  }
+}
+
+PostmortemReportCollector::PostmortemReportCollector() {}
+
+std::vector<FilePath> PostmortemReportCollector::GetDebugStateFilePaths(
+    const std::set<FilePath>& excluded_debug_files) {
+  std::vector<FilePath> paths;
+  base::FileEnumerator enumerator(debug_state_dir_, false /* recursive */,
+                                  base::FileEnumerator::FILES,
+                                  debug_file_pattern_);
+  FilePath path;
+  for (path = enumerator.Next(); !path.empty(); path = enumerator.Next()) {
+    if (excluded_debug_files.find(path) == excluded_debug_files.end())
+      paths.push_back(path);
+  }
+  return paths;
+}
+
+std::unique_ptr<StabilityReport> PostmortemReportCollector::Collect(
+    const base::FilePath& debug_state_file) {
+  // Create a global analyzer.
+  std::unique_ptr<GlobalActivityAnalyzer> global_analyzer =
+      GlobalActivityAnalyzer::CreateWithFile(debug_state_file);
+  if (!global_analyzer)
+    return nullptr;
+
+  // Early exit if there is no data.
+  ThreadActivityAnalyzer* thread_analyzer = global_analyzer->GetFirstAnalyzer();
+  if (!thread_analyzer) {
+    // No data. This case happens in the case of a clean exit.
+    return nullptr;
+  }
+
+  // Iterate through the thread analyzers, fleshing out the report.
+  std::unique_ptr<StabilityReport> report(new StabilityReport());
+  ProcessState* process_state = report->add_process_states();
+
+  while (thread_analyzer) {
+    // Only valid analyzers are expected.
+    DCHECK(thread_analyzer->IsValid());
+
+    ThreadState* thread_state = process_state->add_threads();
+    thread_state->set_thread_name(thread_analyzer->GetThreadName());
+
+    // TODO(manzagop): flesh this out.
+
+    thread_analyzer = global_analyzer->GetNextAnalyzer();
+  }
+
+  return report;
+}
+
+bool PostmortemReportCollector::WriteReportToMinidump(
+    const StabilityReport& report,
+    const crashpad::UUID& client_id,
+    const crashpad::UUID& report_id,
+    FILE* minidump_file) {
+  PostmortemMinidumpWriter writer(minidump_file);
+  return writer.WriteDump(report, client_id, report_id);
+}
+
+}  // namespace browser_watcher