1505685 - Receiver not registered: HeadSetStateReceiver, PackageAddedReceiver

Status: RESOLVED FIXED

(Firefox for Android Graveyard :: General, defect)

Description

[Petru-Mugurel Lingurar [:petru]]

As a continuation of bug 1501648 this should be investigated more to find the culprit and fix it.

Comment 1

[Petru-Mugurel Lingurar [:petru]]

Attachment: Bug 1505685 - Part 1 - Register PackageAddedReceiver on the main thread; r?JanH

Speculative fix for avoiding the situation in which the initialization of Mma
and registering of the PackageAddedReceiver continues on the background thread
after BrowserActivity is destroyed and calls for unregistering the receiver.
By registering the receiver on the main thread we have the benefit of
sequentially execution which would guarantee that the registration is done
before trying to unregister it.

Comment 2

[Petru-Mugurel Lingurar [:petru]]

Attachment: Bug 1505685 - Part 2 - Cleanup - remove second SwitchBoard.loadConfig(); r?JanH

Executing SwitchBoard.loadConfig() on a background thread is the main purpose
of the AsyncConfigLoader so executing it immediately again in the same thread
is redundant.

Depends on D12031

Comment 3

[Petru-Mugurel Lingurar [:petru]]

Attachment: Bug 1505685 - Part 3 - Register HeadSetStateReceiver on the main thread; r?JanH

Speculative fix.

initialize() is called by Gecko on it's background thread after it started.
The value for the "dom.audiochannel.mediaControl" preference also comes on a
background thread.
This both can end up registering the HeadSetStateReceiver while unregistering it
happens only on the main thread.
The patch tries to avoid the race condition by making sure Receiver's
registering and unregistering is done serially on the main thread.

Depends on D12032

Comment 4

[Petru-Mugurel Lingurar [:petru]]

Attachment: Bug 1505685 - Part 2 - Cleanup - remove second SwitchBoard.loadConfig(); r?JanH

Executing SwitchBoard.loadConfig() on a background thread is the main purpose
of the AsyncConfigLoader so executing it immediately again in the same thread
is redundant.

Depends on D12031

Depends on D12031

Comment 5

[Petru-Mugurel Lingurar [:petru]]

Attachment: Bug 1505685 - Part 3 - Register HeadSetStateReceiver on the main thread; r?JanH

Speculative fix.

initialize() is called by Gecko on it's background thread after it started.
The value for the "dom.audiochannel.mediaControl" preference also comes on a
background thread.
This both can end up registering the HeadSetStateReceiver while unregistering it
happens only on the main thread.
The patch tries to avoid the race condition by making sure Receiver's
registering and unregistering is done serially on the main thread.

Depends on D12032

Depends on D12305

Comment 6

[Petru-Mugurel Lingurar [:petru]]

The new differentials apparently were caused because "moz-phab don't yet support reordering/merging/adding/deleting commits".
So after updating patches in a stack they should be submitted manually with arcanist, at least until bug 1481539 is resolved.

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/9344087c8d8e
Part 1 - Register PackageAddedReceiver on the main thread; r=JanH
https://hg.mozilla.org/integration/mozilla-inbound/rev/a46710b7723f
Part 2 - Cleanup - remove second SwitchBoard.loadConfig(); r=JanH
https://hg.mozilla.org/integration/mozilla-inbound/rev/6e2a67a9a9c9
Part 3  - Register HeadSetStateReceiver on the main thread; r=JanH

Comment 8

[Cosmin Sabou [:CosminS]]

Backed out for android checkstyle failures on GeckoMediaControlAgent.

Push with failures: https://treeherder.mozilla.org/#/jobs?repo=mozilla-inbound&resultStatus=success%2Ctestfailed%2Cbusted%2Cexception&fromchange=6e2a67a9a9c9b5d15e9231716f869fc522dfa99b&tochange=86d8ead0b9a51d9cd0f780e0c279f8a86d2b187e&searchStr=android%2C4.0%2Capi16%2B%2Copt%2Candroid%2Cgradle%2Ctests%2Cbuild-android-checkstyle%2Fopt%2Ca%28checkstyle%29&selectedJob=213094199

Failure log: https://treeherder.mozilla.org/logviewer.html#?job_id=213094199&repo=mozilla-inbound&lineNumber=2976

Backout link: https://hg.mozilla.org/integration/mozilla-inbound/rev/86d8ead0b9a51d9cd0f780e0c279f8a86d2b187e

Comment 9

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/cbe820dfb9ad
Part 1 - Register PackageAddedReceiver on the main thread; r=JanH
https://hg.mozilla.org/integration/mozilla-inbound/rev/2a75f31dcd5f
Part 2 - Cleanup - remove second SwitchBoard.loadConfig(); r=JanH
https://hg.mozilla.org/integration/mozilla-inbound/rev/be4f02491218
Part 3 - Register HeadSetStateReceiver on the main thread; r=JanH

Comment 10

[Cristina Coroiu [:ccoroiu]]

https://hg.mozilla.org/mozilla-central/rev/cbe820dfb9ad
https://hg.mozilla.org/mozilla-central/rev/2a75f31dcd5f
https://hg.mozilla.org/mozilla-central/rev/be4f02491218

Comment 11

[Jan Henning [:JanH]]

Fixed on 64 through bug 1501648, which disabled the diagnostic crashes on Beta.