Merge "Track SurfaceFlinger main thread slices" into main (da5e98d6) · Commits · e / os / android_frameworks_base

packages/SystemUI/metrics/perfetto/metrics_specs/Android.bp

+8 −2

Original line number	Diff line number	Diff line
		@@ -22,6 +22,7 @@ filegroup {
		name: "sysui_metrics_artifacts",
		srcs: [
		"android_bitmap_metric.textproto",
		"android_sf_critical_work_main_thread.textproto",
		],
		}

		@@ -30,6 +31,11 @@ prebuilt_etc {
		src: "android_bitmap_metric.textproto",
		}

		prebuilt_etc {
		name: "android_sf_critical_work_main_thread.textproto",
		src: "android_sf_critical_work_main_thread.textproto",
		}

		python_library_host {
		name: "metrics-tests-utils",
		srcs: ["tests/utils/*.py"],
		@@ -47,8 +53,8 @@ python_test_host {
		],
		data: [
		":trace_processor_shell_prebuilt",
		"android_bitmap_metric.textproto",
		"tests/data/android_bitmap_metric_output.txt",
		"*.textproto",
		"tests/data/*.txt",
		],
		test_options: {
		unit_test: true,

packages/SystemUI/metrics/perfetto/metrics_specs/android_sf_critical_work_main_thread.textproto

0 → 100644

+103 −0

Original line number	Diff line number	Diff line
		# Copyright (C) 2025 The Android Open Source Project
		#
		# Licensed under the Apache License, Version 2.0 (the "License");
		# you may not use this file except in compliance with the License.
		# You may obtain a copy of the License at
		#
		# http://www.apache.org/licenses/LICENSE-2.0
		#
		# Unless required by applicable law or agreed to in writing, software
		# distributed under the License is distributed on an "AS IS" BASIS,
		# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
		# See the License for the specific language governing permissions and
		# limitations under the License.

		# The Surfaceflinger process consists of key slices: commit, composition, and postComposition,
		# Refresh Rate Selection and Transaction Handling in the CriticalWorkload track.
		# This template spec tracks metrics related to duration and count for these slices.
		# The metrics are cuj-scoped.

		# SQL query to retrieve details about various CUJs, including their unique IDs, names, timestamps
		# and their durations.
		query: {
		id: "cujs"
		table: {
		table_name: "android_jank_latency_cujs"
		module_name: "android.cujs.sysui_cujs"
		column_names: "cuj_name"
		column_names: "ts"
		column_names: "dur"
		column_names: "cuj_id"
		}
		select_columns: { column_name: "cuj_name" }
		select_columns: { column_name: "dur" }
		select_columns: { column_name: "cuj_id" alias: "id" }
		select_columns: { column_name: "ts" }
		}

		# SQL query that joins the slice and track table of the standard SQL library. Subsequently, the
		# results are filtered based on the track name, process name, and specific slices of interest.

		query: {
		id: "sf_critical_work_slices"
		simple_slices: {
		track_name_glob: "CriticalWorkload"
		process_name_glob: "/system/bin/surfaceflinger"
		}
		# filtering slices by slice name since few instantaneous events are also tracked in CriticalWorkload
		# track and they appear as slices in the stdlib slice table.
		filters: {
		column_name: "slice_name"
		op: EQUAL
		string_rhs: "Commit"
		string_rhs: "Composition"
		string_rhs: "Post Composition"
		string_rhs: "Transaction Handling"
		string_rhs: "Refresh Rate Selection"
		}
		select_columns: { column_name: "dur" }
		select_columns: { column_name: "slice_name" }
		select_columns: { column_name: "id" }
		select_columns: { column_name: "ts" }
		}

		# Define the TraceMetricV2TemplateSpec to generate these aggregates as 5 metrics with cuj_name
		# and slice_name as dimensions.
		# The spec uses interval_intersect to ensure the metrics are CUJ scoped.
		metric_template_spec: {
		id_prefix: "android_sf_critical_work_main_thread_cuj"
		dimensions: "cuj_name"
		dimensions: "slice_name"
		value_columns: "max_dur"
		value_columns: "avg_dur"
		value_columns: "count"
		query: {
		interval_intersect: {
		base: {
		inner_query_id: "sf_critical_work_slices"
		}
		interval_intersect: {
		inner_query_id: "cujs"
		}
		}
		group_by: {
		column_names: "slice_name"
		column_names: "cuj_name"
		aggregates: {
		column_name: "dur"
		op: MAX
		result_column_name: "max_dur"
		}
		aggregates: {
		column_name: "dur"
		op: MEAN
		result_column_name: "avg_dur"
		}
		aggregates: {
		column_name: "slice_name"
		op: COUNT
		result_column_name: "count"
		}
		}
		}
		}

packages/SystemUI/metrics/perfetto/metrics_specs/tests/data/android_sf_critical_work_main_thread_output.txt

0 → 100644

+195 −0

Original line number	Diff line number	Diff line
		metric_bundles {
		row {
		values {
		double_value: 12000000.0
		}
		values {
		double_value: 7000000.0
		}
		values {
		double_value: 2.0
		}
		dimension {
		string_value: "BACK_PANEL_ARROW"
		}
		dimension {
		string_value: "Commit"
		}
		}
		row {
		values {
		double_value: 4000000.0
		}
		values {
		double_value: 3500000.0
		}
		values {
		double_value: 2.0
		}
		dimension {
		string_value: "BACK_PANEL_ARROW"
		}
		dimension {
		string_value: "Composition"
		}
		}
		row {
		values {
		double_value: 3000000.0
		}
		values {
		double_value: 3000000.0
		}
		values {
		double_value: 1.0
		}
		dimension {
		string_value: "BACK_PANEL_ARROW"
		}
		dimension {
		string_value: "Post Composition"
		}
		}
		row {
		values {
		double_value: 3000000.0
		}
		values {
		double_value: 3000000.0
		}
		values {
		double_value: 1.0
		}
		dimension {
		string_value: "BACK_PANEL_ARROW"
		}
		dimension {
		string_value: "Refresh Rate Selection"
		}
		}
		row {
		values {
		double_value: 5000000.0
		}
		values {
		double_value: 5000000.0
		}
		values {
		double_value: 1.0
		}
		dimension {
		string_value: "BACK_PANEL_ARROW"
		}
		dimension {
		string_value: "Transaction Handling"
		}
		}
		specs {
		id: "android_sf_critical_work_main_thread_cuj_max_dur"
		dimensions: "cuj_name"
		dimensions: "slice_name"
		value: "max_dur"
		query {
		interval_intersect {
		base {
		inner_query_id: "sf_critical_work_slices"
		}
		interval_intersect {
		inner_query_id: "cujs"
		}
		}
		group_by {
		column_names: "slice_name"
		column_names: "cuj_name"
		aggregates {
		column_name: "dur"
		op: MAX
		result_column_name: "max_dur"
		}
		aggregates {
		column_name: "dur"
		op: MEAN
		result_column_name: "avg_dur"
		}
		aggregates {
		column_name: "slice_name"
		op: COUNT
		result_column_name: "count"
		}
		}
		}
		bundle_id: "android_sf_critical_work_main_thread_cuj"
		}
		specs {
		id: "android_sf_critical_work_main_thread_cuj_avg_dur"
		dimensions: "cuj_name"
		dimensions: "slice_name"
		value: "avg_dur"
		query {
		interval_intersect {
		base {
		inner_query_id: "sf_critical_work_slices"
		}
		interval_intersect {
		inner_query_id: "cujs"
		}
		}
		group_by {
		column_names: "slice_name"
		column_names: "cuj_name"
		aggregates {
		column_name: "dur"
		op: MAX
		result_column_name: "max_dur"
		}
		aggregates {
		column_name: "dur"
		op: MEAN
		result_column_name: "avg_dur"
		}
		aggregates {
		column_name: "slice_name"
		op: COUNT
		result_column_name: "count"
		}
		}
		}
		bundle_id: "android_sf_critical_work_main_thread_cuj"
		}
		specs {
		id: "android_sf_critical_work_main_thread_cuj_count"
		dimensions: "cuj_name"
		dimensions: "slice_name"
		value: "count"
		query {
		interval_intersect {
		base {
		inner_query_id: "sf_critical_work_slices"
		}
		interval_intersect {
		inner_query_id: "cujs"
		}
		}
		group_by {
		column_names: "slice_name"
		column_names: "cuj_name"
		aggregates {
		column_name: "dur"
		op: MAX
		result_column_name: "max_dur"
		}
		aggregates {
		column_name: "dur"
		op: MEAN
		result_column_name: "avg_dur"
		}
		aggregates {
		column_name: "slice_name"
		op: COUNT
		result_column_name: "count"
		}
		}
		}
		bundle_id: "android_sf_critical_work_main_thread_cuj"
		}
		}

packages/SystemUI/metrics/perfetto/metrics_specs/tests/metrics_v2_test.py

+13 −0

Original line number	Diff line number	Diff line
		@@ -14,6 +14,7 @@
		# limitations under the License.

		from metrics_specs.tests.utils import android_bitmap_metric_trace
		from metrics_specs.tests.utils import android_sf_critical_work_main_thread_trace
		from metrics_specs.tests.utils import test_helper
		import unittest

		@@ -34,5 +35,17 @@ class MetricsV2Test(unittest.TestCase):
		]
		)

		def test_android_android_sf_critical_work_main_thread_metric(self):
		self.helper.verify_metric(
		spec_file="android_sf_critical_work_main_thread.textproto",
		trace_proto_bytes = android_sf_critical_work_main_thread_trace.get_proto(),
		expected_output_file = "android_sf_critical_work_main_thread_output.txt",
		metric_ids = [
		"android_sf_critical_work_main_thread_cuj_max_dur",
		"android_sf_critical_work_main_thread_cuj_avg_dur",
		"android_sf_critical_work_main_thread_cuj_count",
		]
		)

		if __name__ == '__main__':
		unittest.main()

packages/SystemUI/metrics/perfetto/metrics_specs/tests/utils/android_sf_critical_work_main_thread_trace.py

0 → 100755

+396 −0

Original line number	Diff line number	Diff line
		#!/usr/bin/env python3
		# Copyright (C) 2025 The Android Open Source Project
		#
		# Licensed under the Apache License, Version 2.0 (the "License");
		# you may not use this file except in compliance with the License.
		# You may obtain a copy of the License at
		#
		# http://www.apache.org/licenses/LICENSE-2.0
		#
		# Unless required by applicable law or agreed to in writing, software
		# distributed under the License is distributed on an "AS IS" BASIS,
		# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
		# See the License for the specific language governing permissions and
		# limitations under the License.

		from metrics_specs.tests.utils import trace_proto_builder
		from perfetto.protos.perfetto.trace.perfetto_trace_pb2 import Trace

		SYSUI_PID = 5000
		SYSUI_UI_TID = 5020
		SF_PID = 6000
		SF_TID = 6020
		RANDOM_PROCESS_PID = 7000
		RANDOM_PROCESS_TID = 7020

		SYSUI_PACKAGE = "com.android.systemui"
		SF_PACKAGE = "/system/bin/surfaceflinger"
		RANDOM_PROCESS_PACKAGE = "random_process"

		SYSUI_UID = 10001
		SF_UID = 10002
		RANDOM_PROCESS_UID = 10003
		SYSUI_RTID = 1555
		SF_RTID = 1655
		RANDOM_PROCESS_RTID = 1755

		FIRST_CUJ = "J<BACK_PANEL_ARROW>"

		TRACK_NAME_1 = "CriticalWorkload"
		TRACK_ID_1 = 400
		TRACK_NAME_2 = "random_track"
		TRACK_ID_2 = 500
		TRACK_ID_3 = 600

		COMMIT_SLICE = "Commit"
		COMPOSITION_SLICE = "Composition"
		POST_COMPOSITION_SLICE = "Post Composition"
		TRANSACTIONAL_HANDLING_SLICE = "Transaction Handling"
		REFRESH_RATE_SELECTION_SLICE = "Refresh Rate Selection"

		LAYER_1 = "TX - first_layer#0"

		def setup_trace():
		trace = trace_proto_builder.TraceProtoBuilder(Trace())
		add_process(trace, package_name=SYSUI_PACKAGE, uid=SYSUI_UID, pid=SYSUI_PID)
		add_process(trace, package_name=SF_PACKAGE, uid=SF_UID, pid=SF_PID)
		trace.add_ftrace_packet(cpu=0)
		return trace

		def add_slices_and_track(trace):
		# Add CriticalWorkload and a random track to the SF process
		trace.add_process_track_descriptor(
		process_track=TRACK_ID_1,
		track_name=TRACK_NAME_1,
		pid=SF_PID,
		process_name=SF_PACKAGE,
		)
		trace.add_process_track_descriptor(
		process_track=TRACK_ID_2,
		track_name=TRACK_NAME_2,
		pid=SF_PID,
		process_name=SF_PACKAGE,
		)

		# Add slices within cuj duration to the CriticalWorkload track in SF process
		trace.add_track_event_slice(
		name=COMMIT_SLICE,
		ts=27_000_000,
		dur=12_000_000,
		track=TRACK_ID_1,
		trusted_sequence_id=100,
		)
		trace.add_track_event_slice(
		name=TRANSACTIONAL_HANDLING_SLICE,
		ts=27_500_000,
		dur=5_000_000,
		track=TRACK_ID_1,
		trusted_sequence_id=100,
		)
		trace.add_track_event_slice(
		name=REFRESH_RATE_SELECTION_SLICE,
		ts=34_000_000,
		dur=3_000_000,
		track=TRACK_ID_1,
		trusted_sequence_id=100,
		)
		trace.add_track_event_slice(
		name=COMPOSITION_SLICE,
		ts=39_000_000,
		dur=4_000_000,
		track=TRACK_ID_1,
		trusted_sequence_id=100,
		)
		trace.add_track_event_slice(
		name=POST_COMPOSITION_SLICE,
		ts=43_000_000,
		dur=3_000_000,
		track=TRACK_ID_1,
		trusted_sequence_id=100,
		)

		trace.add_track_event_slice(
		name=COMMIT_SLICE,
		ts=84_000_000,
		dur=2_000_000,
		track=TRACK_ID_1,
		trusted_sequence_id=100,
		)
		trace.add_track_event_slice(
		name=COMPOSITION_SLICE,
		ts=86_000_000,
		dur=3_000_000,
		track=TRACK_ID_1,
		trusted_sequence_id=100,
		)

		# Add Post Composition slice outside the CUJ duration ( should not be tracked )
		trace.add_track_event_slice(
		name=POST_COMPOSITION_SLICE,
		ts=89_000_000,
		dur=3_000_000,
		track=TRACK_ID_1,
		trusted_sequence_id=100,
		)

		# Add slices to the random track ( should not be tracked)
		trace.add_track_event_slice(
		name=COMMIT_SLICE,
		ts=27_000_000,
		dur=12_000_000,
		track=TRACK_ID_2,
		trusted_sequence_id=100,
		)

		# Add CriticalWorkload track to the random process
		trace.add_process_track_descriptor(
		process_track=TRACK_ID_3,
		track_name=TRACK_NAME_1,
		pid=RANDOM_PROCESS_PID,
		process_name=RANDOM_PROCESS_PACKAGE,
		)

		# Add slices within cuj duration to the CriticalWorkload track in random process ( should not be tracked )
		trace.add_track_event_slice(
		name=COMMIT_SLICE,
		ts=27_000_000,
		dur=12_000_000,
		track=TRACK_ID_3,
		trusted_sequence_id=100,
		)
		return trace


		def add_process(trace, package_name, uid, pid):
		trace.add_package_list(ts=0, name=package_name, uid=uid, version_code=1)
		trace.add_process(pid=pid, ppid=pid, cmdline=package_name, uid=uid)
		trace.add_thread(tid=pid, tgid=pid, cmdline="MainThread", name="MainThread")

		def add_expected_surface_frame_events(trace, ts, dur, token, pid):
		trace.add_expected_surface_frame_start_event(
		ts=ts,
		cookie=100000 + token,
		token=token,
		display_frame_token=100 + token,
		pid=pid,
		layer_name="",
		)
		trace.add_frame_end_event(ts=ts + dur, cookie=100000 + token)

		def add_actual_surface_frame_events(trace, ts, dur, token, layer, pid):
		cookie = token + 1
		trace.add_actual_surface_frame_start_event(
		ts=ts,
		cookie=100002 + cookie,
		token=token,
		display_frame_token=token + 100,
		pid=pid,
		present_type=1,
		on_time_finish=1,
		gpu_composition=0,
		jank_type=1,
		prediction_type=3,
		layer_name=layer,
		)
		trace.add_frame_end_event(ts=ts + dur, cookie=100002 + cookie)

		def add_cuj(trace, cuj_name):
		# Add 2 CUJs in the trace with the specified cuj_name.

		trace.add_async_atrace_for_thread(
		ts=25_000_000,
		ts_end=77_000_000,
		buf=cuj_name,
		pid=SYSUI_PID,
		tid=SYSUI_UI_TID,
		)

		trace.add_async_atrace_for_thread(
		ts=83_000_000,
		ts_end=102_000_000,
		buf=cuj_name,
		pid=SYSUI_PID,
		tid=SYSUI_UI_TID,
		)

		trace.add_atrace_instant(
		ts=25_000_001,
		buf=cuj_name + "#UIThread",
		pid=SYSUI_PID,
		tid=SYSUI_UI_TID,
		)

		trace.add_atrace_instant(
		ts=83_000_001,
		buf=cuj_name + "#UIThread",
		pid=SYSUI_PID,
		tid=SYSUI_UI_TID,
		)

		trace.add_atrace_instant_for_track(
		ts=25_000_001,
		buf="FT#beginVsync#20",
		pid=SYSUI_PID,
		tid=SYSUI_UI_TID,
		track_name=cuj_name,
		)

		trace.add_atrace_instant_for_track(
		ts=25_000_010,
		buf="FT#layerId#0",
		pid=SYSUI_PID,
		tid=SYSUI_UI_TID,
		track_name=cuj_name,
		)

		trace.add_atrace_instant_for_track(
		ts=76_000_001,
		buf="FT#endVsync#30",
		pid=SYSUI_PID,
		tid=SYSUI_UI_TID,
		track_name=cuj_name,
		)

		trace.add_atrace_instant_for_track(
		ts=83_000_001,
		buf="FT#beginVsync#65",
		pid=SYSUI_PID,
		tid=SYSUI_UI_TID,
		track_name=cuj_name,
		)

		trace.add_atrace_instant_for_track(
		ts=83_000_010,
		buf="FT#layerId#0",
		pid=SYSUI_PID,
		tid=SYSUI_UI_TID,
		track_name=cuj_name,
		)

		trace.add_atrace_instant_for_track(
		ts=101_000_001,
		buf="FT#endVsync#70",
		pid=SYSUI_PID,
		tid=SYSUI_UI_TID,
		track_name=cuj_name,
		)

		# Add Choreographer#doFrame slices within CUJ boundary.
		trace.add_frame(
		vsync=20,
		ts_do_frame=26_000_000,
		ts_end_do_frame=32_000_000,
		tid=SYSUI_UI_TID,
		pid=SYSUI_PID,
		)

		trace.add_atrace_for_thread(
		ts=27_000_000,
		ts_end=28_000_000,
		buf="DrawFrames 20",
		tid=SYSUI_RTID,
		pid=SYSUI_PID,
		)

		trace.add_frame(
		vsync=22,
		ts_do_frame=43_000_000,
		ts_end_do_frame=49_000_000,
		tid=SYSUI_UI_TID,
		pid=SYSUI_PID,
		)

		trace.add_atrace_for_thread(
		ts=44_000_000,
		ts_end=45_000_000,
		buf="DrawFrames 22",
		tid=SYSUI_RTID,
		pid=SYSUI_PID,
		)

		trace.add_frame(
		vsync=24,
		ts_do_frame=60_000_000,
		ts_end_do_frame=65_000_000,
		tid=SYSUI_UI_TID,
		pid=SYSUI_PID,
		)

		trace.add_atrace_for_thread(
		ts=61_000_000,
		ts_end=62_000_000,
		buf="DrawFrames 24",
		tid=SYSUI_RTID,
		pid=SYSUI_PID,
		)

		trace.add_frame(
		vsync=65,
		ts_do_frame=84_000_000,
		ts_end_do_frame=89_000_000,
		tid=SYSUI_UI_TID,
		pid=SYSUI_PID,
		)

		trace.add_atrace_for_thread(
		ts=85_000_000,
		ts_end=86_000_000,
		buf="DrawFrames 65",
		tid=SYSUI_RTID,
		pid=SYSUI_PID,
		)

		# Add expected and actual frames.
		add_expected_surface_frame_events(
		trace, ts=27_000_000, dur=16_000_000, token=20, pid=SYSUI_PID
		)
		add_actual_surface_frame_events(
		trace,
		ts=27_000_000,
		dur=7_000_000,
		token=20,
		layer=LAYER_1,
		pid=SYSUI_PID,
		)

		add_expected_surface_frame_events(
		trace, ts=44_000_000, dur=16_000_000, token=22, pid=SYSUI_PID
		)
		add_actual_surface_frame_events(
		trace,
		ts=44_000_000,
		dur=7_000_000,
		token=22,
		layer=LAYER_1,
		pid=SYSUI_PID,
		)

		add_expected_surface_frame_events(
		trace, ts=61_000_000, dur=16_000_000, token=24, pid=SYSUI_PID
		)
		add_actual_surface_frame_events(
		trace,
		ts=61_000_000,
		dur=6_000_000,
		token=24,
		layer=LAYER_1,
		pid=SYSUI_PID,
		)

		add_expected_surface_frame_events(
		trace, ts=84_000_000, dur=10_000_000, token=65, pid=SYSUI_PID
		)
		add_actual_surface_frame_events(
		trace,
		ts=84_000_000,
		dur=6_000_000,
		token=65,
		layer=LAYER_1,
		pid=SYSUI_PID,
		)

		def get_proto():
		trace = setup_trace()
		add_cuj(trace, FIRST_CUJ)
		builder = add_slices_and_track(trace)
		return builder.trace.SerializeToString()