1651230 - Need test coverage for attempting to play an EME video with an uninstalled codec triggering a GMP update check

Status: NEW

(Core :: Audio/Video, enhancement, P3)

Description

[:Gijs (he/him)]

We accidentally regressed this in bug 1642465, and we only noticed because some performance measurements were accidentally depending on this (but probably shouldn't).

We should have an actual test for this.

Comment 1

[Bryce Seager van Dyk [:bryce] (he/him) - Not reading bugmail]

We could possibly do this with hacking around clearkey and having testing prefs/mocks/etc. Ideally we'd be able to test Widevine in automation. Historically I thought we'd avoided network access in tests, but it appears we're at least somewhat flexible on that.

Comment 2

[Greg Mierzwinski [:sparky]]

(In reply to Bryce Seager van Dyk (:bryce) from comment #1)

We could possibly do this with hacking around clearkey and having testing prefs/mocks/etc. Ideally we'd be able to test Widevine in automation. Historically I thought we'd avoided network access in tests, but it appears we're at least somewhat flexible on that.

In perf testing, we're a little more flexible on this now in some cases - I'm not sure if the same is true for functional tests. The youtube playback tests, for instance, need about 15Gb of media data so we have to host the test on our own servers.

Comment 3

[Chris Peterson [:cpeterson]]

Unassigning bugs assigned to Bryce because he no longer works at Mozilla.

Comment 4

[Rob Wu [:robwu]]

While looking at bug 2043803, I noticed some questionable logic in checkForUpdates at https://searchfox.org/firefox-main/rev/e28b34ab33dbf49364999070168cbb7e11e8e5bd/toolkit/mozapps/extensions/internal/GMPProvider.sys.mjs#557-574

There are two callers of checkForUpdates:

• media.eme.enabled preference handler (source) kicks off update check after 10 seconds.
• EMEVideo:CDMMissing observer (source) kicking off update check immediately.

The implementation returns early if an update check is pending. But if the 10-second timer was activated, the presumably high-priority 0-second timer is not immediately scheduled, resulting in an up to 10 second delay before the GMP update check actually runs.