1500465 - Need APIs to measure touch scrolling input latency (Non-WebRender call path) on desktop and GeckoView

Status: RESOLVED FIXED

(Core :: Panning and Zooming, enhancement, P3)

Description

[Markus Stange [:mstange]]

The performance team wants to add level 1 telemetry probes for touch scrolling latency, most importantly on Windows and Android/GeckoView. We may also want to have a Talos test that measures this latency.

We will need to have APIs to get the time that passes between the OS touch event and the composite that presents this touch move to the screen.

Comment 1

[Kartikaya Gupta (email:[e-mail])]

The touch events already have a timestamp on them (if not we can add one), so those can be used to track the time from when the OS gives us the event to the point where APZ uses that event to mutate the scroll position. However past that point there's a many-to-one collapsing where multiple touch events can have their effect show up on the screen in a single composite. So we'd somehow have to track all the touch events that were processed by APZ between samplings, and hand that bag of information to the compositor at sample time (or just expose it via an API to be read after a sampling). Then when the compositor presents the frame it can record the telemetry probes for those touch events.

This assumes we only care about touch events that are eventually processed by APZ, and not touch events that preventDefault()'d but then drive some JS scrolling. That would be much harder to track.

Comment 2

[Christina Cheung]

Attachment: Bug 1500465 - APIs to measure touch scrolling input latency on desktop and GeckoView r=mstange

Comment 3

[Christina Cheung]

Attachment: Bug 1500465 - move all mIncoming to mSampled r=mstange

Depends on D34912

Comment 4

[Christina Cheung]

Attachment: Bug 1500465 - simplified the APZController r=mstange

Depends on D34913

Comment 5

[Christina Cheung]

Attachment: Bug 1500465 - Removed ScrollPayload and added feedback from reviewers in refactoring and formatting the code. r=mstange

Comment 6

[Christina Cheung]

Attachment: Bug 1500465 - Removed ScrollPayload, removed trimIncomingFrameScrolls, and added feedback from reviewers in refactoring and formatting the code. r=mstange

Comment 7

[Christina Cheung]

Attachment: Bug 1500465 - Measure touch scroll latency and connect to telemetry.

Comment 8

[Christina Cheung]

Attachment: Bug 1500465 - Refactored touch scroll latency and connect to telemetry. r=kats,mstange

Depends on D36246

Comment 9

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/40010303c474
Measure touch scroll latency and connect to telemetry. r=kats

Comment 10

[Andrei Ciure[:aciure]]

https://hg.mozilla.org/mozilla-central/rev/40010303c474

Comment 11

[Kartikaya Gupta (email:[e-mail])]

Given the OOM crashes in both automation and affecting real-world users, I'd like to get this backed out unless a fix is forthcoming today. Anybody have objections?

Comment 12

[Christina Cheung]

(In reply to Kartikaya Gupta (email:[e-mail]) (away 17-Jul-2019 to Feb-2020) from comment #11)

Given the OOM crashes in both automation and affecting real-world users, I'd like to get this backed out unless a fix is forthcoming today. Anybody have objections?

Yes, we will need to ask for a back out.

Comment 13

[Pulsebot]

Backout by [e-mail]:
https://hg.mozilla.org/mozilla-central/rev/c4b20b18bf97
Backed out changeset 40010303c474 on request of christina_ for causing OOM crashes in automation. a=backout

Comment 14

[Kartikaya Gupta (email:[e-mail])]

Thanks!

Comment 15

[Sean Feng [:sefeng]]

Taking it to address the last review comments.

Comment 16

[Sean Feng [:sefeng]]

Attachment: data_collection_review_1500465.txt

Comment 17

[Tim Smith (inactive) 👨‍🔬 [:tdsmith]]

Comment on attachment 9113058 [details]
data_collection_review_1500465.txt

Thanks for flagging this for review.

DATA COLLECTION REVIEW RESPONSE:

    Is there or will there be documentation that describes the schema for the ultimate data set available publicly, complete and accurate?

Yes. This collection is Telemetry so is documented in its definitions file [Histograms.json](https://hg.mozilla.org/mozilla-central/file/tip/toolkit/components/telemetry/Histograms.json) and the [Probe Dictionary](https://telemetry.mozilla.org/probe-dictionary/).

    Is there a control mechanism that allows the user to turn the data collection on and off?

Yes. This collection is Telemetry so can be controlled through Firefox's Preferences.

    If the request is for permanent data collection, is there someone who will monitor the data over time?

Yes, :sefeng is responsible.

    Using the category system of data types on the Mozilla wiki, what collection type of data do the requested measurements fall under?

Category 1, Technical.

    Is the data collection request for default-on or default-off?

Default on for all channels.

    Does the instrumentation include the addition of any new identifiers?

No.

    Is the data collection covered by the existing Firefox privacy notice?

Yes.

    Does there need to be a check-in in the future to determine whether to renew the data?

No. This collection is permanent.

Comment 18

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/283c7696cf3a
Measure touch scroll latency and connect to telemetry. r=mstange,botond

Comment 19

[Narcis Beleuzu [:NarcisB]]

https://hg.mozilla.org/mozilla-central/rev/283c7696cf3a