1801301 - The "activityAfterIdle" message is wrongly triggered after the PC wakes up from "Sleep" or "Hibernation"

Status: RESOLVED WONTFIX

(Firefox :: Messaging System, defect, P1)

Description

[Marius Coman [:mcoman], Ecosystem QA]

Attachment: sleep-hibernate issue.mp4

[Affected versions]:

• Firefox Beta 108.0b3 - Build ID: 20221117185908

[Affected Platforms]:

• Windows 7 x64

[Prerequisites]:

• Have the "Remote Settings Devtools" add-on installed from here.

[Steps to reproduce]:

1. Open the browser from the prerequisites.
2. Click the "Remote Settings Devtools" toolbar button from the right part of the toolbar.
3. Change the environment to "Stage" and restart the browser.
4. Put the PC to "Sleep".
5. Wake up the PC and observe the behavior.

[Expected result]:

• The "activityAfterIdle" message is NOT triggered.

[Actual result]:

• The "activityAfterIdle" message is triggered.

[Additional Notes]:

• This issue is also reproducible if the PC was put to "Hibernate" instead of "Sleep".
• Attached is a screen recording of the issue:

Comment 1

[Shane Hughes [:aminomancer]]

Thanks for recording this Marius. Have you been able to reproduce this on a non-virtual machine?

Comment 2

[Shane Hughes [:aminomancer]]

Maybe it's firing an "idle" event immediately after the wake event. Or maybe the sleep and wake events are just not firing at all. I would like to work with you or another QA engineer on the patch for this if that's possible. I have a couple ideas.

Note: If the sleep/wake events are actually firing, we might set a flag here to disable the idle callback and set a timeout to re-enable it. So that idle events can't happen within a few event loops of waking.

Otherwise, we might need to find some other way to identify when the OS is put to sleep.

Comment 3

[Marius Coman [:mcoman], Ecosystem QA]

(In reply to Shane Hughes [:aminomancer] from comment #1)

Thanks for recording this Marius. Have you been able to reproduce this on a non-virtual machine?

Hi, Shane! Yes, the issue is reproducible on a non-virtual machine.

Comment 4

[Shane Hughes [:aminomancer]]

Attachment: Bug 1801301 - Fix idle message trigger firing after wake. r=barret

Add a 1s grace period after waking from sleep in which notifications
from the idle service will be ignored. Break out the idle trigger tests
because we've reached the max statements per function limit.

Comment 5

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/dcc7f67f4426
Fix idle message trigger firing after wake. r=barret

Comment 6

[Shane Hughes [:aminomancer]]

Comment on attachment 9305741 [details]
Bug 1801301 - Fix idle message trigger firing after wake. r=barret

Beta/Release Uplift Approval Request

• User impact if declined: Planned messages that will use the idle trigger would have to be restricted from Windows users until 109, making the data collected less useful since it's the largest cohort by OS.
• Is this code covered by automated tests?: Yes
• Has the fix been verified in Nightly?: No
• Needs manual test from QE?: No
• If yes, steps to reproduce:
• List of other uplifts needed: Bug 1802333
• Risk to taking this patch: Low
• Why is the change risky/not risky? (and alternatives if risky): The idle trigger is not currently used by any messages, local or remote, so the affected code is never executed in normal operation. It's covered by automated tests, which are reliably passing on try. When we do add messages that would cause the affected code to be executed, it will be within narrow experimental treatment branches, so any negative consequences would be confined to a small group of opted-in users.
• String changes made/needed:
• Is Android affected?: No

Comment 7

[Natalia Csoregi [:nataliaCs]]

https://hg.mozilla.org/mozilla-central/rev/dcc7f67f4426

Comment 8

[Dianna Smith [:diannaS]]

Comment on attachment 9305741 [details]
Bug 1801301 - Fix idle message trigger firing after wake. r=barret

Approved for 108.0rc1

Comment 9

[Dianna Smith [:diannaS]]

https://hg.mozilla.org/releases/mozilla-beta/rev/1a68bcdacbe0

Comment 10

[Shane Hughes [:aminomancer]]

These patches did not fix the underlying issue, which will be tracked in bug 1802865, so we're changing the tracking status to reflect that