platform_system_core/metrics/metrics_daemon.h
Sonny Rao 4b8aebb0f4 Add additional stats for swapping from /proc/vmstat
This adds swap-in and swap-out rate calculation on the same intervals
as major page fault collection.

BUG=chromium:261965
TEST=observe Platform.SwapIn{Long,Short} and
Platform.SwapOut{Long,Short} appear in histograms

Change-Id: Ifcdba1088cdff355b8e132145ac79635b185663b
Reviewed-on: https://gerrit.chromium.org/gerrit/64162
Reviewed-by: Luigi Semenzato <semenzato@chromium.org>
Commit-Queue: Sonny Rao <sonnyrao@chromium.org>
Tested-by: Sonny Rao <sonnyrao@chromium.org>
2013-08-01 18:15:49 -07:00

428 lines
16 KiB
C++

// Copyright (c) 2010 The Chromium OS Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
#ifndef METRICS_DAEMON_H_
#define METRICS_DAEMON_H_
#include <dbus/dbus.h>
#include <glib.h>
#include <map>
#include <base/file_path.h>
#include <base/memory/scoped_ptr.h>
#include <base/time.h>
#include <gtest/gtest_prod.h> // for FRIEND_TEST
#include "metrics_library.h"
namespace chromeos_metrics {
class FrequencyCounter;
class TaggedCounter;
class TaggedCounterReporter;
}
class MetricsDaemon {
public:
MetricsDaemon();
~MetricsDaemon();
// Initializes.
void Init(bool testing, MetricsLibraryInterface* metrics_lib,
const std::string& diskstats_path,
const std::string& vmstats_path,
const std::string& cpuinfo_max_freq_path,
const std::string& scaling_max_freq_path);
// Does all the work. If |run_as_daemon| is true, daemonizes by
// forking.
void Run(bool run_as_daemon);
private:
friend class MetricsDaemonTest;
FRIEND_TEST(MetricsDaemonTest, CheckSystemCrash);
FRIEND_TEST(MetricsDaemonTest, ComputeEpochNoCurrent);
FRIEND_TEST(MetricsDaemonTest, ComputeEpochNoLast);
FRIEND_TEST(MetricsDaemonTest, GetHistogramPath);
FRIEND_TEST(MetricsDaemonTest, IsNewEpoch);
FRIEND_TEST(MetricsDaemonTest, LookupPowerState);
FRIEND_TEST(MetricsDaemonTest, LookupScreenSaverState);
FRIEND_TEST(MetricsDaemonTest, LookupSessionState);
FRIEND_TEST(MetricsDaemonTest, MessageFilter);
FRIEND_TEST(MetricsDaemonTest, ParseVmStats);
FRIEND_TEST(MetricsDaemonTest, PowerStateChanged);
FRIEND_TEST(MetricsDaemonTest, ProcessKernelCrash);
FRIEND_TEST(MetricsDaemonTest, ProcessMeminfo);
FRIEND_TEST(MetricsDaemonTest, ProcessMeminfo2);
FRIEND_TEST(MetricsDaemonTest, ProcessUncleanShutdown);
FRIEND_TEST(MetricsDaemonTest, ProcessUserCrash);
FRIEND_TEST(MetricsDaemonTest, ReportCrashesDailyFrequency);
FRIEND_TEST(MetricsDaemonTest, ReadFreqToInt);
FRIEND_TEST(MetricsDaemonTest, ReportDailyUse);
FRIEND_TEST(MetricsDaemonTest, ReportDiskStats);
FRIEND_TEST(MetricsDaemonTest, ReportKernelCrashInterval);
FRIEND_TEST(MetricsDaemonTest, ReportUncleanShutdownInterval);
FRIEND_TEST(MetricsDaemonTest, ReportUserCrashInterval);
FRIEND_TEST(MetricsDaemonTest, ScreenSaverStateChanged);
FRIEND_TEST(MetricsDaemonTest, SendMetric);
FRIEND_TEST(MetricsDaemonTest, SendCpuThrottleMetrics);
FRIEND_TEST(MetricsDaemonTest, SessionStateChanged);
FRIEND_TEST(MetricsDaemonTest, SetUserActiveState);
FRIEND_TEST(MetricsDaemonTest, SetUserActiveStateTimeJump);
// The power states (see power_states.h).
enum PowerState {
kUnknownPowerState = -1, // Initial/unknown power state.
#define STATE(name, capname) kPowerState ## capname,
#include "power_states.h"
kNumberPowerStates
};
// The user session states (see session_states.h).
enum SessionState {
kUnknownSessionState = -1, // Initial/unknown user session state.
#define STATE(name, capname) kSessionState ## capname,
#include "session_states.h"
kNumberSessionStates
};
// State for disk stats collector callback.
enum StatsState {
kStatsShort, // short wait before short interval collection
kStatsLong, // final wait before new collection
};
// Data record for aggregating daily usage.
class UseRecord {
public:
UseRecord() : day_(0), seconds_(0) {}
int day_;
int seconds_;
};
// Type of scale to use for meminfo histograms. For most of them we use
// percent of total RAM, but for some we use absolute numbers, usually in
// megabytes, on a log scale from 0 to 4000, and 0 to 8000 for compressed
// swap (since it can be larger than total RAM).
enum MeminfoOp {
kMeminfoOp_HistPercent = 0,
kMeminfoOp_HistLog,
kMeminfoOp_SwapTotal,
kMeminfoOp_SwapFree,
};
// Record for retrieving and reporting values from /proc/meminfo.
struct MeminfoRecord {
const char* name; // print name
const char* match; // string to match in output of /proc/meminfo
MeminfoOp op; // histogram scale selector, or other operator
int value; // value from /proc/meminfo
};
// Record for retrieving and reporting values from /proc/vmstat
struct VmstatRecord {
uint64_t page_faults_; // major faults
uint64_t swap_in_; // pages swapped in
uint64_t swap_out_; // pages swapped out
};
typedef std::map<std::string, chromeos_metrics::FrequencyCounter*>
FrequencyCounters;
// Metric parameters.
static const char kMetricAnyCrashesDailyName[];
static const char kMetricAnyCrashesWeeklyName[];
static const char kMetricCrashFrequencyBuckets;
static const char kMetricCrashFrequencyMax;
static const char kMetricCrashFrequencyMin;
static const int kMetricCrashIntervalBuckets;
static const int kMetricCrashIntervalMax;
static const int kMetricCrashIntervalMin;
static const int kMetricDailyUseTimeBuckets;
static const int kMetricDailyUseTimeMax;
static const int kMetricDailyUseTimeMin;
static const char kMetricDailyUseTimeName[];
static const char kMetricKernelCrashesDailyName[];
static const char kMetricKernelCrashesWeeklyName[];
static const char kMetricKernelCrashIntervalName[];
static const char kMetricsPath[];
static const char kMetricUncleanShutdownIntervalName[];
static const char kMetricUncleanShutdownsDailyName[];
static const char kMetricUncleanShutdownsWeeklyName[];
static const char kMetricUserCrashesDailyName[];
static const char kMetricUserCrashesWeeklyName[];
static const char kMetricUserCrashIntervalName[];
static const char kMetricReadSectorsLongName[];
static const char kMetricReadSectorsShortName[];
static const char kMetricWriteSectorsLongName[];
static const char kMetricWriteSectorsShortName[];
static const char kMetricPageFaultsShortName[];
static const char kMetricPageFaultsLongName[];
static const char kMetricSwapInLongName[];
static const char kMetricSwapInShortName[];
static const char kMetricSwapOutLongName[];
static const char kMetricSwapOutShortName[];
static const char kMetricScaledCpuFrequencyName[];
static const int kMetricStatsShortInterval;
static const int kMetricStatsLongInterval;
static const int kMetricMeminfoInterval;
static const int kMetricSectorsIOMax;
static const int kMetricSectorsBuckets;
static const int kMetricPageFaultsMax;
static const int kMetricPageFaultsBuckets;
static const char kMetricsDiskStatsPath[];
static const char kMetricsVmStatsPath[];
// Array of power states.
static const char* kPowerStates_[kNumberPowerStates];
// Array of user session states.
static const char* kSessionStates_[kNumberSessionStates];
// Returns the active time since boot (uptime minus sleep time) in seconds.
double GetActiveTime();
// Clears and deletes the data contained in frequency_counters_.
void DeleteFrequencyCounters();
// Configures the given crash interval reporter.
void ConfigureCrashIntervalReporter(
const char* histogram_name,
scoped_ptr<chromeos_metrics::TaggedCounterReporter>* reporter);
// Configures the given frequency counter reporter.
void ConfigureCrashFrequencyReporter(const char* histogram_name);
// Returns file path to persistent file for generating given histogram.
FilePath GetHistogramPath(const char* histogram_name);
// Creates the event loop and enters it.
void Loop();
// D-Bus filter callback.
static DBusHandlerResult MessageFilter(DBusConnection* connection,
DBusMessage* message,
void* user_data);
// Processes power state change.
void PowerStateChanged(const char* state_name, base::Time now);
// Given the state name, returns the state id.
PowerState LookupPowerState(const char* state_name);
// Processes user session state change.
void SessionStateChanged(const char* state_name, base::Time now);
// Given the state name, returns the state id.
SessionState LookupSessionState(const char* state_name);
// Updates the user-active state to |active| and logs the usage data
// since the last update. If the user has just become active,
// reschedule the daily use monitor for more frequent updates --
// this is followed by an exponential back-off (see UseMonitor).
// While in active use, this method should be called at intervals no
// longer than kUseMonitorIntervalMax otherwise new use time will be
// discarded.
void SetUserActiveState(bool active, base::Time now);
// Updates the daily usage file, if necessary, by adding |seconds|
// of active use to the |day| since Epoch. If there's usage data for
// day in the past in the usage file, that data is sent to UMA and
// removed from the file. If there's already usage data for |day| in
// the usage file, the |seconds| are accumulated.
void LogDailyUseRecord(int day, int seconds);
// Updates the active use time and logs time between user-space
// process crashes.
void ProcessUserCrash();
// Updates the active use time and logs time between kernel crashes.
void ProcessKernelCrash();
// Updates the active use time and logs time between unclean shutdowns.
void ProcessUncleanShutdown();
// Checks if a kernel crash has been detected and returns true if
// so. The method assumes that a kernel crash has happened if
// |crash_file| exists. It removes the file immediately if it
// exists, so it must not be called more than once.
bool CheckSystemCrash(const std::string& crash_file);
// Callbacks for the daily use monitor. The daily use monitor uses
// LogDailyUseRecord to aggregate current usage data and send it to
// UMA, if necessary. It also reschedules itself using an
// exponentially bigger interval (up to a certain maximum) -- so
// usage is monitored less frequently with longer active use.
static gboolean UseMonitorStatic(gpointer data);
bool UseMonitor();
// Schedules or reschedules a daily use monitor for |interval|
// seconds from now. |backoff| mode is used by the use monitor to
// reschedule itself. If there's a monitor scheduled already and
// |backoff| is false, unschedules it first. Doesn't schedule a
// monitor for more than kUseMonitorIntervalMax seconds in the
// future (see metrics_daemon.cc). Returns true if a new use monitor
// was scheduled, false otherwise (note that if |backoff| is false a
// new use monitor will always be scheduled).
bool ScheduleUseMonitor(int interval, bool backoff);
// Unschedules a scheduled use monitor, if any.
void UnscheduleUseMonitor();
// Report daily use through UMA.
static void ReportDailyUse(void* handle, int tag, int count);
// Sends a regular (exponential) histogram sample to Chrome for
// transport to UMA. See MetricsLibrary::SendToUMA in
// metrics_library.h for a description of the arguments.
void SendMetric(const std::string& name, int sample,
int min, int max, int nbuckets);
// Sends a linear histogram sample to Chrome for transport to UMA. See
// MetricsLibrary::SendToUMA in metrics_library.h for a description of the
// arguments.
void SendLinearMetric(const std::string& name, int sample,
int max, int nbuckets);
// Initializes vm and disk stats reporting.
void StatsReporterInit();
// Schedules a callback for the next vm and disk stats collection.
void ScheduleStatsCallback(int wait);
// Reads cumulative disk statistics from sysfs. Returns true for success.
bool DiskStatsReadStats(long int* read_sectors, long int* write_sectors);
// Reads cumulative vm statistics from procfs. Returns true for success.
bool VmStatsReadStats(struct VmstatRecord* stats);
// Parse cumulative vm statistics from a C string. Returns true for success.
bool VmStatsParseStats(const char* stats, struct VmstatRecord* record);
// Reports disk and vm statistics (static version for glib). Arguments are a
// glib artifact.
static gboolean StatsCallbackStatic(void* handle);
// Reports disk and vm statistics.
void StatsCallback();
// Schedules meminfo collection callback.
void ScheduleMeminfoCallback(int wait);
// Reports memory statistics (static version for glib). Argument is a glib
// artifact.
static gboolean MeminfoCallbackStatic(void* handle);
// Reports memory statistics. Returns false on failure.
bool MeminfoCallback();
// Parses content of /proc/meminfo and sends fields of interest to UMA.
// Returns false on errors. |meminfo_raw| contains the content of
// /proc/meminfo.
bool ProcessMeminfo(const std::string& meminfo_raw);
// Parses meminfo data from |meminfo_raw|. |fields| is a vector containing
// the fields of interest. The order of the fields must be the same in which
// /proc/meminfo prints them. The result of parsing fields[i] is placed in
// fields[i].value.
bool FillMeminfo(const std::string& meminfo_raw,
std::vector<MeminfoRecord>* fields);
// Schedule a memory use callback. |new_callback| is true when this callback
// is scheduled for the first time. When |new_callback| is false,
// |time_elapsed| is the active (non-sleep) time that has passed between now
// and the original callback scheduling time. We use it to reschedule a
// callback that fired too early because we slept.
void ScheduleMemuseCallback(bool new_callback, double time_elapsed);
// Static wrapper for MemuseCallback. Always returns false.
static gboolean MemuseCallbackStatic(void* handle);
// Calls MemuseCallbackWork, and possibly schedules next callback, if enough
// active time has passed. Otherwise reschedules itself to simulate active
// time callbacks (i.e. wall clock time minus sleep time).
void MemuseCallback();
// Reads /proc/meminfo and sends total anonymous memory usage to UMA.
bool MemuseCallbackWork();
// Parses meminfo data and sends it to UMA.
bool ProcessMemuse(const std::string& meminfo_raw);
// Sends stats for thermal CPU throttling.
void SendCpuThrottleMetrics();
// Reads an integer CPU frequency value from sysfs.
bool ReadFreqToInt(const std::string& sysfs_file_name, int* value);
// Test mode.
bool testing_;
// The metrics library handle.
MetricsLibraryInterface* metrics_lib_;
// Timestamps last network state update. This timestamp is used to
// sample the time from the network going online to going offline so
// TimeTicks ensures a monotonically increasing TimeDelta.
base::TimeTicks network_state_last_;
// Current power state.
PowerState power_state_;
// Current user session state.
SessionState session_state_;
// Is the user currently active: power is on, user session has
// started, screen is not locked.
bool user_active_;
// Timestamps last user active update. Active use time is aggregated
// each day before sending to UMA so using time since the epoch as
// the timestamp.
base::Time user_active_last_;
// Daily active use time in seconds.
scoped_ptr<chromeos_metrics::TaggedCounter> daily_use_;
// Active use time between user-space process crashes.
scoped_ptr<chromeos_metrics::TaggedCounterReporter> user_crash_interval_;
// Active use time between kernel crashes.
scoped_ptr<chromeos_metrics::TaggedCounterReporter> kernel_crash_interval_;
// Active use time between unclean shutdowns crashes.
scoped_ptr<chromeos_metrics::TaggedCounterReporter>
unclean_shutdown_interval_;
// Map of all frequency counters, to simplify flushing them.
FrequencyCounters frequency_counters_;
// Sleep period until the next daily usage aggregation performed by
// the daily use monitor (see ScheduleUseMonitor).
int usemon_interval_;
// Scheduled daily use monitor source (see ScheduleUseMonitor).
GSource* usemon_source_;
// Time of initial scheduling of memuse callback
double memuse_initial_time_;
// Selects the wait time for the next memory use callback.
unsigned int memuse_interval_index_;
// Contain the most recent disk and vm cumulative stats.
long int read_sectors_;
long int write_sectors_;
struct VmstatRecord vmstats_;
StatsState stats_state_;
double stats_initial_time_;
std::string diskstats_path_;
std::string vmstats_path_;
std::string scaling_max_freq_path_;
std::string cpuinfo_max_freq_path_;
};
#endif // METRICS_DAEMON_H_