Add current prio to proc state, track prio changes

[deliverable/lttng-analyses.git] / lttnganalyses / cli / command.py

# The MIT License (MIT)
#
# Copyright (C) 2015 - Julien Desfossez <jdesfossez@efficios.com>
#               2015 - Philippe Proulx <pproulx@efficios.com>
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in
# all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.

from ..core import analysis
from ..linuxautomaton import automaton
from .. import __version__
from . import progressbar
from ..linuxautomaton import common
from .. import _version
from babeltrace import TraceCollection
import argparse
import sys
import subprocess
import json
import re
from . import mi


class Command:
    _MI_BASE_TAGS = ['linux-kernel', 'lttng-analyses']
    _MI_AUTHORS = [
        'Julien Desfossez',
        'Antoine Busque',
        'Philippe Proulx',
    ]
    _MI_URL = 'https://github.com/lttng/lttng-analyses'

    def __init__(self, mi_mode=False):
        self._analysis = None
        self._analysis_conf = None
        self._args = None
        self._handles = None
        self._traces = None
        self._ticks = 0
        self._mi_mode = mi_mode
        self._create_automaton()
        self._mi_setup()

    @property
    def mi_mode(self):
        return self._mi_mode

    def run(self):
        try:
            self._parse_args()
            self._open_trace()
            self._create_analysis()
            self._run_analysis()
            self._close_trace()
        except KeyboardInterrupt:
            sys.exit(0)

    def _error(self, msg, exit_code=1):
        try:
            import termcolor

            msg = termcolor.colored(msg, 'red', attrs=['bold'])
        except:
            pass

        print(msg, file=sys.stderr)
        sys.exit(exit_code)

    def _gen_error(self, msg, exit_code=1):
        self._error('Error: {}'.format(msg), exit_code)

    def _cmdline_error(self, msg, exit_code=1):
        self._error('Command line error: {}'.format(msg), exit_code)

    def _print(self, msg):
        if not self._mi_mode:
            print(msg)

    def _mi_create_result_table(self, table_class_name, begin, end,
                                subtitle=None):
        return mi.ResultTable(self._mi_table_classes[table_class_name],
                              begin, end, subtitle)

    def _mi_setup(self):
        self._mi_table_classes = {}

        for tc_tuple in self._MI_TABLE_CLASSES:
            table_class = mi.TableClass(tc_tuple[0], tc_tuple[1], tc_tuple[2])
            self._mi_table_classes[table_class.name] = table_class

        self._mi_clear_result_tables()

    def _mi_print_metadata(self):
        tags = self._MI_BASE_TAGS + self._MI_TAGS
        infos = mi.get_metadata(version=self._MI_VERSION, title=self._MI_TITLE,
                                description=self._MI_DESCRIPTION,
                                authors=self._MI_AUTHORS, url=self._MI_URL,
                                tags=tags,
                                table_classes=self._mi_table_classes.values())
        print(json.dumps(infos))

    def _mi_append_result_table(self, result_table):
        if not result_table or not result_table.rows:
            return

        tc_name = result_table.table_class.name
        self._mi_get_result_tables(tc_name).append(result_table)

    def _mi_append_result_tables(self, result_tables):
        if not result_tables:
            return

        for result_table in result_tables:
            self._mi_append_result_table(result_table)

    def _mi_clear_result_tables(self):
        self._result_tables = {}

    def _mi_get_result_tables(self, table_class_name):
        if table_class_name not in self._result_tables:
            self._result_tables[table_class_name] = []

        return self._result_tables[table_class_name]

    def _mi_print(self):
        results = []

        for result_tables in self._result_tables.values():
            for result_table in result_tables:
                results.append(result_table.to_native_object())

        obj = {
            'results': results,
        }

        print(json.dumps(obj))

    def _create_summary_result_tables(self):
        pass

    def _open_trace(self):
        traces = TraceCollection()
        handles = traces.add_traces_recursive(self._args.path, 'ctf')
        if handles == {}:
            self._gen_error('Failed to open ' + self._args.path, -1)
        self._handles = handles
        self._traces = traces
        self._process_date_args()
        if not self._args.skip_validation:
            self._check_lost_events()

    def _close_trace(self):
        for handle in self._handles.values():
            self._traces.remove_trace(handle)

    def _check_lost_events(self):
        self._print('Checking the trace for lost events...')
        try:
            subprocess.check_output('babeltrace "%s"' % self._args.path,
                                    shell=True)
        except subprocess.CalledProcessError:
            self._gen_error('Cannot run babeltrace on the trace, cannot verify'
                            ' if events were lost during the trace recording')

    def _pre_analysis(self):
        pass

    def _post_analysis(self):
        if not self._mi_mode:
            return

        if self._ticks > 1:
            self._create_summary_result_tables()

        self._mi_print()

    def _run_analysis(self):
        self._pre_analysis()
        progressbar.progressbar_setup(self)

        for event in self._traces.events:
            progressbar.progressbar_update(self)
            self._analysis.process_event(event)
            if self._analysis.ended:
                break
            self._automaton.process_event(event)

        progressbar.progressbar_finish(self)
        self._analysis.end()
        self._post_analysis()

    def _print_date(self, begin_ns, end_ns):
        date = 'Timerange: [%s, %s]' % (
            common.ns_to_hour_nsec(begin_ns, gmt=self._args.gmt,
                                   multi_day=True),
            common.ns_to_hour_nsec(end_ns, gmt=self._args.gmt,
                                   multi_day=True))
        self._print(date)

    def _get_uniform_freq_values(self, durations):
        if self._args.uniform_step is not None:
            return (self._args.uniform_min, self._args.uniform_max,
                    self._args.uniform_step)

        if self._args.min is not None:
            self._args.uniform_min = self._args.min
        else:
            self._args.uniform_min = min(durations)
        if self._args.max is not None:
            self._args.uniform_max = self._args.max
        else:
            self._args.uniform_max = max(durations)

        # ns to µs
        self._args.uniform_min /= 1000
        self._args.uniform_max /= 1000
        self._args.uniform_step = (
            (self._args.uniform_max - self._args.uniform_min) /
            self._args.freq_resolution
        )

        return self._args.uniform_min, self._args.uniform_max, \
            self._args.uniform_step

    def _validate_transform_common_args(self, args):
        refresh_period_ns = None
        if args.refresh is not None:
            try:
                refresh_period_ns = common.duration_str_to_ns(args.refresh)
            except ValueError as e:
                self._cmdline_error(str(e))

        self._analysis_conf = analysis.AnalysisConfig()
        self._analysis_conf.refresh_period = refresh_period_ns
        self._analysis_conf.period_begin_ev_name = args.period_begin
        self._analysis_conf.period_end_ev_name = args.period_end
        self._analysis_conf.period_key_fields = args.period_key.split(',')
        if args.cpu:
            self._analysis_conf.cpu_list = args.cpu.split(',')
            self._analysis_conf.cpu_list = [int(cpu) for cpu in
                                            self._analysis_conf.cpu_list]

        # convert min/max args from µs to ns, if needed
        if hasattr(args, 'min') and args.min is not None:
            args.min *= 1000
            self._analysis_conf.min_duration = args.min
        if hasattr(args, 'max') and args.max is not None:
            args.max *= 1000
            self._analysis_conf.max_duration = args.max

        if hasattr(args, 'procname'):
            if args.procname:
                self._analysis_conf.proc_list = args.procname.split(',')

        if hasattr(args, 'tid'):
            if args.tid:
                self._analysis_conf.tid_list = args.tid.split(',')
                self._analysis_conf.tid_list = [int(tid) for tid in
                                                self._analysis_conf.tid_list]

        if hasattr(args, 'freq'):
            args.uniform_min = None
            args.uniform_max = None
            args.uniform_step = None

            if args.freq_series:
                # implies uniform buckets
                args.freq_uniform = True

        if self._mi_mode:
            # force no progress in MI mode
            args.no_progress = True

            # print MI metadata if required
            if args.metadata:
                self._mi_print_metadata()
                sys.exit(0)

        # validate path argument (required at this point)
        if not args.path:
            self._cmdline_error('Please specify a trace path')

        if type(args.path) is list:
            args.path = args.path[0]

    def _validate_transform_args(self, args):
        pass

    def _parse_args(self):
        ap = argparse.ArgumentParser(description=self._DESC)

        # common arguments
        ap.add_argument('-r', '--refresh', type=str,
                        help='Refresh period, with optional units suffix '
                        '(default units: s)')
        ap.add_argument('--gmt', action='store_true',
                        help='Manipulate timestamps based on GMT instead '
                             'of local time')
        ap.add_argument('--skip-validation', action='store_true',
                        help='Skip the trace validation')
        ap.add_argument('--begin', type=str, help='start time: '
                                                  'hh:mm:ss[.nnnnnnnnn]')
        ap.add_argument('--end', type=str, help='end time: '
                                                'hh:mm:ss[.nnnnnnnnn]')
        ap.add_argument('--period-begin', type=str,
                        help='Analysis period start marker event name')
        ap.add_argument('--period-end', type=str,
                        help='Analysis period end marker event name '
                        '(requires --period-begin)')
        ap.add_argument('--period-key', type=str, default='cpu_id',
                        help='Optional, list of event field names used to '
                        'match period markers (default: cpu_id)')
        ap.add_argument('--cpu', type=str,
                        help='Filter the results only for this list of '
                        'CPU IDs')
        ap.add_argument('--timerange', type=str, help='time range: '
                                                      '[begin,end]')
        ap.add_argument('-V', '--version', action='version',
                        version='LTTng Analyses v' + __version__)

        # MI mode-dependent arguments
        if self._mi_mode:
            ap.add_argument('--metadata', action='store_true',
                            help='Show analysis\'s metadata')
            ap.add_argument('path', metavar='<path/to/trace>',
                            help='trace path', nargs='*')
        else:
            ap.add_argument('--no-progress', action='store_true',
                            help='Don\'t display the progress bar')
            ap.add_argument('path', metavar='<path/to/trace>',
                            help='trace path')

        # Used to add command-specific args
        self._add_arguments(ap)

        args = ap.parse_args()
        self._validate_transform_common_args(args)
        self._validate_transform_args(args)
        self._args = args

    @staticmethod
    def _add_proc_filter_args(ap):
        ap.add_argument('--procname', type=str,
                        help='Filter the results only for this list of '
                        'process names')
        ap.add_argument('--tid', type=str,
                        help='Filter the results only for this list of TIDs')

    @staticmethod
    def _add_min_max_args(ap):
        ap.add_argument('--min', type=float,
                        help='Filter out durations shorter than min usec')
        ap.add_argument('--max', type=float,
                        help='Filter out durations longer than max usec')

    @staticmethod
    def _add_freq_args(ap, help=None):
        if not help:
            help = 'Output the frequency distribution'

        ap.add_argument('--freq', action='store_true', help=help)
        ap.add_argument('--freq-resolution', type=int, default=20,
                        help='Frequency distribution resolution '
                        '(default 20)')
        ap.add_argument('--freq-uniform', action='store_true',
                        help='Use a uniform resolution across distributions')
        ap.add_argument('--freq-series', action='store_true',
                        help='Consolidate frequency distribution histogram '
                        'as a single one')

    @staticmethod
    def _add_log_args(ap, help=None):
        if not help:
            help = 'Output the events in chronological order'

        ap.add_argument('--log', action='store_true', help=help)

    @staticmethod
    def _add_top_args(ap, help=None):
        if not help:
            help = 'Output the top results'

        ap.add_argument('--limit', type=int, default=10,
                        help='Limit to top X (default = 10)')
        ap.add_argument('--top', action='store_true', help=help)

    @staticmethod
    def _add_stats_args(ap, help=None):
        if not help:
            help = 'Output statistics'

        ap.add_argument('--stats', action='store_true', help=help)

    def _add_arguments(self, ap):
        pass

    def _process_date_args(self):
        def date_to_epoch_nsec(date):
            ts = common.date_to_epoch_nsec(self._handles, date, self._args.gmt)
            if ts is None:
                self._cmdline_error('Invalid date format: "{}"'.format(date))

            return ts

        self._args.multi_day = common.is_multi_day_trace_collection(
            self._handles)
        begin_ts = None
        end_ts = None

        if self._args.timerange:
            begin_ts, end_ts = common.extract_timerange(self._handles,
                                                        self._args.timerange,
                                                        self._args.gmt)
            if None in [begin_ts, end_ts]:
                self._cmdline_error(
                    'Invalid time format: "{}"'.format(self._args.timerange))
        else:
            if self._args.begin:
                begin_ts = date_to_epoch_nsec(self._args.begin)
            if self._args.end:
                end_ts = date_to_epoch_nsec(self._args.end)

                # We have to check if timestamp_begin is None, which
                # it always is in older versions of babeltrace. In
                # that case, the test is simply skipped and an invalid
                # --end value will cause an empty analysis
                if self._traces.timestamp_begin is not None and \
                   end_ts < self._traces.timestamp_begin:
                    self._cmdline_error(
                        '--end timestamp before beginning of trace')

        self._analysis_conf.begin_ts = begin_ts
        self._analysis_conf.end_ts = end_ts

    def _create_analysis(self):
        notification_cbs = {
            analysis.Analysis.TICK_CB: self._analysis_tick_cb
        }

        self._analysis = self._ANALYSIS_CLASS(self.state, self._analysis_conf)
        self._analysis.register_notification_cbs(notification_cbs)

    def _create_automaton(self):
        self._automaton = automaton.Automaton()
        self.state = self._automaton.state

    def _analysis_tick_cb(self, **kwargs):
        begin_ns = kwargs['begin_ns']
        end_ns = kwargs['end_ns']

        self._analysis_tick(begin_ns, end_ns)
        self._ticks += 1

    def _analysis_tick(self, begin_ns, end_ns):
        raise NotImplementedError()


# create MI version
_cmd_version = _version.get_versions()['version']
_version_match = re.match(r'(\d+)\.(\d+)\.(\d+)(.*)', _cmd_version)
Command._MI_VERSION = [
    int(_version_match.group(1)),
    int(_version_match.group(2)),
    int(_version_match.group(3)),
    _version_match.group(4),
]