1506699 - Fennec UI shuts down too early when dumping captured profile on shutdown

Status: RESOLVED FIXED

(Firefox for Android Graveyard :: General, enhancement)

Description

[Jan Henning [:JanH]]

When explicitly quitting, the Android UI normally only waits for sanitizing to finish before shutting down as well, regardless of what Gecko might still be doing.

We already check for MOZ_PROFILER_STARTUP=... [1], so we might as well check for MOZ_PROFILER_SHUTDOWN and integrate jchens patch into our code base
https://gist.github.com/anonymous/158d3cf764f7e07655c23ec5d866db29

That is force the Quit button to be always visible and don't immediately finish the activity when exiting - instead, wait for Gecko to completely exit before quitting the Android UI, too.

[1] https://dxr.mozilla.org/mozilla-central/rev/237e4c0633fda8e227b2ab3ab57e417c980a2811/mobile/android/base/java/org/mozilla/gecko/GeckoApp.java#930

Comment 1

[Jan Henning [:JanH]]

Attachment: Bug 1506699 - Make dumping a profile to disk easier when using MOZ_PROFILER_SHUTDOWN. r?snorp

We're already have some special handling in the Android UI when detecting
MOZ_PROFILER_STARTUP, so we might as well handle MOZ_PROFILER_SHUTDOWN, too.

This means two things:
1. In order to actually quit the app, the "Quit" button needs to be enabled.
2. We need to wait until we're sure that the profile has been dumped to disk
   before finishing the activity (and thereby possibly killing our process).

Comment 2

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/8711ee626573
Make dumping a profile to disk easier when using MOZ_PROFILER_SHUTDOWN. r=snorp

Comment 3

[Noemi Erli[:noemi_erli]]

https://hg.mozilla.org/mozilla-central/rev/8711ee626573

Comment 4

[Chris Peterson [:cpeterson]]

Thanks for the fix, Jan. I'm marking 64=wontfix because we don't need to uplift this fix to Fennec 64 Beta. Developers usually profile Nightly builds, not Beta builds.

Comment 5

[Jan Henning [:JanH]]

True enough, although this could also be helpful if we ever got a repeat performance of something like the Safebrowsing hangs we've had, where the true cause remained a mystery for quite some time. But there's no pressing need right now, so it's fine.