STRs:

1. Open https://jsfiddle.net/jib1/dy6rh1nw/ and allow camera.

Expected result:

• 3 videos: small (160x120), medium (320x240), and large (640x480).

Actual result:

• 3 videos: large (640x480), medium (320x240), and large (640x480).

Regression range:
51:10.64 INFO: Newest known good nightly: 2018-09-18
51:10.64 INFO: Oldest known bad nightly: 2018-11-09
51:10.64 INFO: To resume, run:
51:10.64 INFO: /usr/local/bin/mozregression --repo=mozilla-central --good=2018-09-18 --bad=2018-11-09

Run the above command to see good and bad result.

The range is wide because it crashes in the range 09/19 - 11/08.

The crash, e.g. bp-20d1bfab-a4da-47b2-aa47-a55b70190111, began in bug 1253499 (narrowed down):
12:36.57 INFO: No more inbound revisions, bisection finished.
12:36.57 INFO: Last good revision: c4e3131d033e72be1b0b8ed70c9860657ef0c37f
12:36.57 INFO: First bad revision: 38ff8f0f79f3f75e4c341c564e8fc7967fe9e051
12:36.57 INFO: Pushlog:
https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=c4e3131d033e72be1b0b8ed70c9860657ef0c37f&tochange=38ff8f0f79f3f75e4c341c564e8fc7967fe9e051

The crash stopped again with bug 1376873 (narrowed down):
6:47.55 INFO: Narrowed inbound regression window from [faf1cfd8, 1f47cb09] (3 builds) to [faf1cfd8, e83c311e] (2 builds) (~1 steps left)
6:47.55 INFO: No more inbound revisions, bisection finished.
6:47.55 INFO: First good revision: e83c311e5293902be4db4ecea17cff87c633f7cf
6:47.55 INFO: Last bad revision: faf1cfd83dd5a3359edfd6d7e66a7123a84bb031
6:47.55 INFO: Pushlog:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=faf1cfd83dd5a3359edfd6d7e66a7123a84bb031&tochange=e83c311e5293902be4db4ecea17cff87c633f7cf

I'm blocking both bugs because I'm not confident bug 1376873 fixes the crash introduced by bug 1253499 so much as avoid it.

Could this be why Google Meet has low outgoing frame rate from my machine (bug 1515205)?

[Tracking Requested - why for this release]: Might be cause of bug 1515205.

Note that the fiddle crashes in release (64.0 and 64.0.2)—as it's in the inner crashing regression-range between bug 1253499 and bug 1376873, so I wouldn't call it unaffected. Marking as wontfix. We can change that if we decide to open a separate bug for the crash.

Thank you for investigating this. Unfortunately this is not the cause of Bug 1515205, I see the expected resolutions in VideoStreamFactory::CreateEncoderStreams for the three simulcast streams on Meet and incorrect resolutions with the supplied fiddle.

On further investigation, it appears that the lowest resolution is not working correctly because of a typo in the fiddle:

At line 17:

{rid: "lo", maxBitrate: 150000, scaleResolutionDownBy: 4},

But at line 78:

'a=simulcast: recv rid=hi;mid;low\r\n';

If I change line 17 to "low", I see the expected resolutions.

It's unfortunate that this crashes on release, but since the crash is not present on Beta or Nightly, I'm not sure how much time we want to spend investigating this at the moment. I'm going to retitle the bug and clear the tracking flags.

Firefox 65 and 66 are not affected by this crash. I missed resetting the affected flags when I reset the tracking flags.

Hey Jan,
Can you still reproduce this or should we close it?
I can't reproduce the cras following your steps on a Macbook with MacOS 11.6.

It just crashed for me bp-50a5a97f-4715-4277-a3a5-cd8b80211104 in Nightly.

That one is bug 1738931.

In the process of migrating remaining bugs to the new severity system, the severity for this bug cannot be automatically determined. Please retriage this bug using the new severity system.

This is not crashing anymore.