1567278 - YouTube playback not working

Status: RESOLVED DUPLICATE of bug 1566540

(Core :: Audio/Video: Playback, defect, P2)

Description

[richardf9661]

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.14; rv:68.0) Gecko/20100101 Firefox/68.0

Steps to reproduce:

Going to https://youtube.com, and selecting any video, the video will not play. It will usually give an error stating to Restart the Device.

Actual results:

The video will continue to load, but will never play even after clicking pause/play.

Expected results:

The video should have played as normal in any other version of Firefox (including the current version of Quantum).

Comment 1

[Andrea Guglielmi]

I was writing on mozillaZine forum for the same problem.
I tried with a fresh, new profile with no luck but as I had to use Terminal to invoke Profile Manager I was able to see a lot of errors like:

[Child 880, MediaDecoderStateMachine #1] WARNING: Decoder=12ee8d200 Decode error: NS_ERROR_DOM_MEDIA_DECODE_ERR (0x806e0004): file /builds/worker/workspace/build/src/dom/media/MediaDecoderStateMachine.cpp, line 3305

OSX Catalina - 10.15 Beta (19A512f)
Firefox Nightly - Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:70.0) Gecko/20100101 Firefox/70.0 ID:20190718095458

Comment 2

[richardf9661]

(In reply to Andrea Guglielmi from comment #1)

I was writing on mozillaZine forum for the same problem.
I tried with a fresh, new profile with no luck but as I had to use Terminal to invoke Profile Manager I was able to see a lot of errors like:

[Child 880, MediaDecoderStateMachine #1] WARNING: Decoder=12ee8d200 Decode error: NS_ERROR_DOM_MEDIA_DECODE_ERR (0x806e0004): file /builds/worker/workspace/build/src/dom/media/MediaDecoderStateMachine.cpp, line 3305

OSX Catalina - 10.15 Beta (19A512f)
Firefox Nightly - Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:70.0) Gecko/20100101 Firefox/70.0 ID:20190718095458

I'm on the same version of macOS... This was also occurring in Dev 3... I'm quite certain it's an issue with Firefox Nightly alone and not a macOS setting/issue.

Comment 3

[Jean-Yves Avenard [:jya]]

Working for my with 19A487

Comment 4

[bl.lebrun]

On Windows 10, Youtube playback is busted for 4k, will start to play the video but then begins to drop every frame thereafter. Works for 1080p and 1440p however. CPU usage also jumps higher than it normally should as well when it attempts to playback 4k content.

Comment 5

[Jean-Yves Avenard [:jya]]

(In reply to bl.lebrun from comment #4)

On Windows 10, Youtube playback is busted for 4k, will start to play the video but then begins to drop every frame thereafter. Works for 1080p and 1440p however. CPU usage also jumps higher than it normally should as well when it attempts to playback 4k content.

So your machine is too slow to play 4K content, don't select it.

Your graphic card probably doesn't have vp9 hardware support.

In any case, this is totally irrelevant to this bug which is about YT on macos

Comment 6

[bl.lebrun]

(In reply to Jean-Yves Avenard [:jya] from comment #5)

(In reply to bl.lebrun from comment #4)

On Windows 10, Youtube playback is busted for 4k, will start to play the video but then begins to drop every frame thereafter. Works for 1080p and 1440p however. CPU usage also jumps higher than it normally should as well when it attempts to playback 4k content.

So your machine is too slow to play 4K content, don't select it.

Your graphic card probably doesn't have vp9 hardware support.

In any case, this is totally irrelevant to this bug which is about YT on macos

I'm on a gtx 1080 ti and an i7 4790k, 4k shouldn't be an issue and it was never an issue

Comment 7

[richardf9661]

(In reply to bl.lebrun from comment #6)

(In reply to Jean-Yves Avenard [:jya] from comment #5)

(In reply to bl.lebrun from comment #4)

On Windows 10, Youtube playback is busted for 4k, will start to play the video but then begins to drop every frame thereafter. Works for 1080p and 1440p however. CPU usage also jumps higher than it normally should as well when it attempts to playback 4k content.

So your machine is too slow to play 4K content, don't select it.

Your graphic card probably doesn't have vp9 hardware support.

In any case, this is totally irrelevant to this bug which is about YT on macos

I'm on a gtx 1080 ti and an i7 4790k, 4k shouldn't be an issue and it was never an issue

Regardless of whether or not it's the system's fault or not, you're at least able to play YouTube, correct? People on a Mac and the latest versions of Firefox Nightly are unable to play any YouTube video, in any quality.

Comment 8

[tgn-ff]

The issue is, actually, even worse. If you try to open a new tab which will open in a new process (eg. by setting dom.ipc.processCount to -1), then the browser itself becomes unresponsive. Even worse than that, when the Nightly from 2018-01-01 becomes unresponsive, it's impossible to interact with the X anymore. I needed to kill Firefox from a TTY.

It seems to be a split testing from youtube. I could reproduce for some time yesterday then it stopped until today.

I tried to kill the main and content processes with a SIGABRT. The crash reports don't seem useful, but in any case, here they are:
https://crash-stats.mozilla.org/report/index/5a1d6f48-958c-4b2f-938b-ec05c0190718
https://crash-stats.mozilla.org/report/index/2ebf41f8-46c5-4f76-9dca-9e1c00190718
https://crash-stats.mozilla.org/report/index/a9066d08-21c5-4064-8181-99a920190718
https://crash-stats.mozilla.org/report/index/8d6d8622-1ded-45db-a70b-45c190190718
https://crash-stats.mozilla.org/report/index/d429b5c7-55c8-431e-8b91-272c80190718
https://crash-stats.mozilla.org/report/index/ee4804ee-f356-4c91-94c6-73c6b0190718

I, also, tried to find a regression range, since I found out that the Nightly of 2017-01-01 was good, but mozregression didn't find "artifacts", therefore the pushlog isn't short:

8:43.14 INFO: Got as far as we can go bisecting nightlies...
8:43.14 INFO: Last good revision: c616a6fd5e4b20cca139fcdd3957682afaa862b9 (2017-11-13)
8:43.14 INFO: First bad revision: f0c0fb9182d695081edf170d8e3bcb8164f2c96a (2017-11-14)
8:43.14 INFO: Pushlog:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=c616a6fd5e4b20cca139fcdd3957682afaa862b9&tochange=f0c0fb9182d695081edf170d8e3bcb8164f2c96a

Comment 9

[tgn-ff]

I forgot to mention that this is not a mac-specific issue. I'm on Linux.

Comment 10

[richardf9661]

(In reply to tgn-ff from comment #9)

I forgot to mention that this is not a mac-specific issue. I'm on Linux.
I'm glad it's not Mac-specific because when it is, it tends to take much longer to resolve the issue compared to when it's spread across different systems.

Comment 11

[Asif Youssuff]

Is this still broken, because I can't currently reproduce this on Linux?

Comment 12

[tgn-ff]

richardf9661, can you verify that Firefox freezes for you too? Set dom.ipc.processCount to -1 in about:config, restart Firefox, then open a video page. After it starts loading and doesn't play, open a new tab and visit, say, mozilla.org. At this point the browser becomes unresponsive for me. What about you?

@Asif
For me it's an issue that sometimes reproduces constantly and sometimes (like right now) not at all.

Comment 13

[richardf9661]

(In reply to Asif Youssuff from comment #11)

Is this still broken, because I can't currently reproduce this on Linux?

It is as of July 19, 2019 at 6:27EST.

Comment 14

[richardf9661]

(In reply to tgn-ff from comment #12)

richardf9661, can you verify that Firefox freezes for you too? Set dom.ipc.processCount to -1 in about:config, restart Firefox, then open a video page. After it starts loading and doesn't play, open a new tab and visit, say, mozilla.org. At this point the browser becomes unresponsive for me. What about you?

@Asif
For me it's an issue that sometimes reproduces constantly and sometimes (like right now) not at all.

This issue is not occurring on my system, I set it to -1, it was at 8 before, and I Force Quit Firefox Nightly. After reopening it, I went to https://youtube.com, selected a random video, waited a few seconds for it to load. I then opened a new tab (mozilla.org as you had recommended) and it's still working fine. It's currently set to -1 right now, but I am going to be resetting it/setting it back to 8.

Comment 15

[tgn-ff]

Someone with edit rights, please hide my comments and the replies to them on this bug. It seems I'm hitting a very different, Linux (possibly systemd) only bug. For the record, it happens if you forcefully close the current session with loginctl kill-session.

Comment 16

[Adrian London]

Same issue. MacOS Catalina dev beta 4, FF Nightly 70 (2019-07-19).

Youtube videos don't play. Videos from any other (non-youtube) source work fine. It used to work a few days ago but I'm not totally sure whether that was 2 or 3 days ago. I've just downloaded Nightly 2019-07-15 and youtube videos work fine. Back to 2019-07-19 and nothing. Tried safe-mode and a new profile.

As expected, other browsers don't have this issue. I don't have any way to test a different O/S version.

Comment 17

[Jean-Yves Avenard [:jya]]

(In reply to Adrian London from comment #16)

Same issue. MacOS Catalina dev beta 4, FF Nightly 70 (2019-07-19).

Youtube videos don't play. Videos from any other (non-youtube) source work fine. It used to work a few days ago but I'm not totally sure whether that was 2 or 3 days ago. I've just downloaded Nightly 2019-07-15 and youtube videos work fine. Back to 2019-07-19 and nothing. Tried safe-mode and a new profile.

As expected, other browsers don't have this issue. I don't have any way to test a different O/S version.

Could you moz-regression to determine which changed caused that regression? Thank you
https://mozilla.github.io/mozregression/install.html

Comment 18

[Adrian London]

(In reply to Jean-Yves Avenard [:jya] from comment #17)

Could you moz-regression to determine which changed caused that regression? Thank you
https://mozilla.github.io/mozregression/install.html

That was an interesting tool! After about 20 firefox version downloads .....
https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=5671bf0f8a214d690b4754aab8e6304102c1b791&tochange=368e9ae19115fce103b410d423d78076d59db7e7

12:22.70 INFO: Narrowed inbound regression window from [5671bf0f, 3c6a1d3f] (3 builds) to [5671bf0f, 368e9ae1] (2 builds) (~1 steps left)
12:22.70 INFO: No more inbound revisions, bisection finished.
12:22.70 INFO: Last good revision: 5671bf0f8a214d690b4754aab8e6304102c1b791
12:22.70 INFO: First bad revision: 368e9ae19115fce103b410d423d78076d59db7e7

So, Michael Froman, what do you have against Youtube? :-)

Comment 19

[Andrea Guglielmi]

I was trying to find regression downloading versions from http://ftp.mozilla.org/pub/firefox/nightly/2019/07/ and found:

Good: 2019-07-14-21-40-27-mozilla-central/firefox-70.0a1.en-US.mac.dmg
NO good: 2019-07-15-09-40-48-mozilla-central/firefox-70.0a1.en-US.mac.dmg

Then I wanted to try moz-regression tool for the first time and found the same result as Adrian did:
https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=5671bf0f8a214d690b4754aab8e6304102c1b791&tochange=368e9ae19115fce103b410d423d78076d59db7e7

Hope it will get fixed soon.

Comment 20

[Adrian London]

Hi Andrea, and anyone else interested ...

The mozregression tool is a really good one. After it isolated the change you can open the html link and click through to see the source code. I'm no developer (well, I dabble) but it's easy to see that the change relates to OPUS decoding. Specifically RDD (whatever that is!). Clear code changes and comments from the developer made that deduction easy.

So ... there's a simple fix. Disable it. about:config and toggle "media.rdd-opus.enabled" to false.

Comment 21

[Andrea Guglielmi]

(In reply to Adrian London from comment #20)

So ... there's a simple fix. Disable it. about:config and toggle "media.rdd-opus.enabled" to false.
Maybe you're not a developer but...you did it!!!
I didn't even think that could be an option to disable Rdd-Opus, just tried and it works like a charm, thank you.

Now let the "real" developers dig it a bit :)

Ciao!

Comment 22

[Andrea Guglielmi]

Sorry, I eat up a blank line and it looks like a single quote.
Thank you Adrian for the tip.

Comment 23

[Jean-Yves Avenard [:jya]]

I can reproduce the issue with 10.15 beta build 19A512f

Comment 24

[Jean-Yves Avenard [:jya]]

Are you using 10.15 beta or 10.14? I can't reproduce on latest 10.14

Comment 25

[Michael Froman [:mjf]]

(In reply to Jean-Yves Avenard [:jya] from comment #23)

I can reproduce the issue with 10.15 beta build 19A512f

This seems like a possible duplication of Bug 1566540, which Haik is looking at as a sandboxing bug with 10.15 Beta.

Comment 26

[richardf9661]

(In reply to Jean-Yves Avenard [:jya] from comment #24)

Are you using 10.15 beta or 10.14? I can't reproduce on latest 10.14

It's occurring for me on 10.15 Beta... both Developer versions 3 and 4.

Comment 27

[richardf9661]

(In reply to Michael Froman [:mjf] from comment #25)

(In reply to Jean-Yves Avenard [:jya] from comment #23)

I can reproduce the issue with 10.15 beta build 19A512f

This seems like a possible duplication of Bug 1566540, which Haik is looking at as a sandboxing bug with 10.15 Beta.

Whatever the case may be, the solution that fixes bug 1566540, also fixes bug 1567278. The fix is the following: "On Nightly, setting media.rdd-opus.enabled to false and restarting avoided the crash for me on macOS 10.15 Beta 4."

Comment 28

[Paul Adenot (:padenot)]

Michael, I'm assigning to you, make sure to dupe or close as appropriate when you know more about this, thanks!

Comment 29

[Michael Froman [:mjf]]

(In reply to richardf9661 from comment #27)

(In reply to Michael Froman [:mjf] from comment #25)

(In reply to Jean-Yves Avenard [:jya] from comment #23)

I can reproduce the issue with 10.15 beta build 19A512f

This seems like a possible duplication of Bug 1566540, which Haik is looking at as a sandboxing bug with 10.15 Beta.

Whatever the case may be, the solution that fixes bug 1566540, also fixes bug 1567278. The fix is the following: "On Nightly, setting media.rdd-opus.enabled to false and restarting avoided the crash for me on macOS 10.15 Beta 4."

@richardf9661, would you mind trying again on Nightly (with media.rdd-opus.enabled set to true) now that Bug 1566540 has landed? I don't have a 10.15 machine that I can test this on.

Comment 30

[Adrian London]

I'm not Richard, but can confirm that youtube videos now play even when media.rdd-opus.enabled is set back to true.

Comment 31

[Andrea Guglielmi]

Confirmed.
Reset media.rdd-opus.enabled to true, restarted FF and Youtube works.

Comment 32

[novenary]

I'm on Linux and youtube playback is broken for me on both of my machines (still broken as of 2019-07-26-01-36-54). Setting media.rdd-opus.enabled to false fixes it.

Comment 33

[Michael Froman [:mjf]]

(In reply to Dan Elkouby from comment #32)

I'm on Linux and youtube playback is broken for me on both of my machines (still broken as of 2019-07-26-01-36-54). Setting media.rdd-opus.enabled to false fixes it.

@Dan - this bug was mostly dealing with macOS sandboxing issues, but there is another open bug for Linux, Bug 1567389, that is similar but is caused by Opus and intializing cubeb-alsa backends. Also, so I'm clear, what version of Firefox are you running, Release, Beta, or Nightly?

Comment 35

[novenary]

(In reply to Michael Froman [:mjf] from comment #33)

@Dan - this bug was mostly dealing with macOS sandboxing issues, but there is another open bug for Linux, Bug 1567389, that is similar but is caused by Opus and intializing cubeb-alsa backends. Also, so I'm clear, what version of Firefox are you running, Release, Beta, or Nightly?

Sorry, I must have overlooked that when searching for similar issues. I will follow that other bug. I'm on nightly which was recently broken, stable is fine.

Comment 36

[Michael Froman [:mjf]]

(In reply to Dan Elkouby from comment #35)

(In reply to Michael Froman [:mjf] from comment #33)

@Dan - this bug was mostly dealing with macOS sandboxing issues, but there is another open bug for Linux, Bug 1567389, that is similar but is caused by Opus and intializing cubeb-alsa backends. Also, so I'm clear, what version of Firefox are you running, Release, Beta, or Nightly?

Sorry, I must have overlooked that when searching for similar issues. I will follow that other bug. I'm on nightly which was recently broken, stable is fine.

@Dan - thanks for the follow-up info!