1908317 - Firefox 128ESR no longer plays videos on Solaris - version 115ESR was ok

Status: RESOLVED FIXED

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

Description

[Petr Sumbera]

Steps to reproduce:

Firefox 128ESR no longer plays videos on Solaris - version 115ESR was ok.

Last working version was 116.01a. Where my old binary from 2023-06-19 (0d3d276273c4) is ok and binary from 2023-06-20 (d8571e33cdd0) is bad.

It just says: "No video with supported format and MIME type found".

Tested with: https://tekeye.uk/html/html5-video-test-page
(youtube doesn't work too)

Unfortunately above hg hashes (hg id) are no longer recognized?!!!

Comment 1

[Petr Sumbera]

Mike, do you please have any idea why old hg id hashes are no longer valid with current source repo? Or possibly what might have been the change which caused this regression?

Comment 2

[BugBot [:suhaib / :marco]]

The Bugbug bot thinks this bug should belong to the 'Core::Audio/Video: Playback' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Comment 3

[Mike Hommey [:glandium]]

Mike, do you please have any idea why old hg id hashes are no longer valid with current source repo?

The real question would be where did you get them. Are you keeping your source in Mercurial and is it patched? Because neither of the ids you gave exist in our repos AFAICT (and no, repos haven't changed).

Comment 4

[Petr Sumbera]

(In reply to Mike Hommey [:glandium] from comment #3)

The real question would be where did you get them. Are you keeping your source in Mercurial and is it patched? Because neither of the ids you gave exist in our repos AFAICT (and no, repos haven't changed).

Thank you!. This was my bad I do have there extra patches (commits)...

Comment 5

[Petr Sumbera]

I was able to bisect it to like this.

The first bad revision is:
changeset: 668485:e608dfe11fc7
user: Alexandre Lissy <[e-mail]>
date: Tue Jun 20 10:23:23 2023 -0700
description:
Bug 1835804 - Completely block from doing audio decoding on Content and RDD r=alwu

Comment 6

[:gerard-majax]

(In reply to Petr Sumbera from comment #5)

I was able to bisect it to like this.

The first bad revision is:
changeset: 668485:e608dfe11fc7
user: Alexandre Lissy <[e-mail]>
date: Tue Jun 20 10:23:23 2023 -0700
description:
Bug 1835804 - Completely block from doing audio decoding on Content and RDD r=alwu

Yes well, this has been here for a while. Are you sure Utility process is working correctly? Unfortunately, given the status of Solaris, I cannot debug that.

Comment 7

[Petr Sumbera]

Yes well, this has been here for a while. Are you sure Utility process is working correctly? Unfortunately, given the status of Solaris, I cannot debug that.

What is Utility process? And how can I verify it's working correctly? And yes I know I realized this late...

Comment 8

[:gerard-majax]

(In reply to Petr Sumbera from comment #7)

Yes well, this has been here for a while. Are you sure Utility process is working correctly? Unfortunately, given the status of Solaris, I cannot debug that.

What is Utility process? And how can I verify it's working correctly? And yes I know I realized this late...

I dont get it, have you not hit the problem on release since last year ?
Utility is a new set of processes, one of them holding audio decoding.

Comment 9

[:gerard-majax]

Check about:processes to verify, but I am wondering if all we need is not just enabling in https://searchfox.org/mozilla-central/rev/b220e40ff2ee3d10ce68e07d8a8a577d5558e2a2/modules/libpref/init/StaticPrefList.yaml#10462-10479 ?

Comment 10

[Petr Sumbera]

(In reply to :gerard-majax from comment #8)

(In reply to Petr Sumbera from comment #7)

Yes well, this has been here for a while. Are you sure Utility process is working correctly? Unfortunately, given the status of Solaris, I cannot debug that.

What is Utility process? And how can I verify it's working correctly? And yes I know I realized this late...

I dont get it, have you not hit the problem on release since last year ?

We do use ESR versions only. So I haven't noticed it. I do build nightly builds but with very limited testing...

Comment 11

[Mike Hommey [:glandium]]

This sounds very similar to bug 1886654: the lack of correlation between what the prefs say and what the code expects.

Comment 12

[:gerard-majax]

(In reply to Petr Sumbera from comment #10)

(In reply to :gerard-majax from comment #8)

(In reply to Petr Sumbera from comment #7)

Yes well, this has been here for a while. Are you sure Utility process is working correctly? Unfortunately, given the status of Solaris, I cannot debug that.

What is Utility process? And how can I verify it's working correctly? And yes I know I realized this late...

I dont get it, have you not hit the problem on release since last year ?

We do use ESR versions only. So I haven't noticed it. I do build nightly builds but with very limited testing...

Ok I was assuming you tested / used more the nightlies. I guess just switch the pref in about:config ? And at build time in the yaml file, once you have checked, we can land a patch. It worked without much complains on OpenBSD as well as FreeBSD (checked myself there) so hopefully Solaris should be easy as well

Comment 13

[Petr Sumbera]

Unfortunately enabling media.utility-process.enabled (which was indeed false) didn't help. Tested on first bad commit with about:configchange and with 128.0 release and modification to modules/libpref/init/StaticPrefList.yam.

Comment 14

[:gerard-majax]

(In reply to Petr Sumbera from comment #13)

Unfortunately enabling media.utility-process.enabled (which was indeed false) didn't help. Tested on first bad commit with about:configchange and with 128.0 release and modification to modules/libpref/init/StaticPrefList.yam.

You need to verify if utility process is working. Try MOZ_LOG=utilityproc:5 to maybe start collecting data? Can you verify by just playing a simple audio file? Can you make sure you run that on a debug build ?

Comment 15

[Petr Sumbera]

(In reply to Petr Sumbera from comment #13)

Unfortunately enabling media.utility-process.enabled (which was indeed false) didn't help. Tested on first bad commit with about:configchange and with 128.0 release and modification to modules/libpref/init/StaticPrefList.yam.

Sorry. about:config change on 128 worked...

Comment 16

[Petr Sumbera]

Sorry. about:config change on 128 worked...

I wonder why following didn't worked:

--- firefox-128.0/modules/libpref/init/StaticPrefList.yaml
+++ firefox-128.0/modules/libpref/init/StaticPrefList.yam
@@ -10124,6 +10124,8 @@
   value: true
 #elif defined(XP_OPENBSD)
   value: true
+#elif defined(XP_SOLARIS)
+  value: true
 #else
   value: false
 #endif

Maybe XP_SOLARIS need to be defined for yaml somewhere...

Comment 17

[:gerard-majax]

(In reply to Petr Sumbera from comment #16)

Sorry. about:config change on 128 worked...

I wonder why following didn't worked:

--- firefox-128.0/modules/libpref/init/StaticPrefList.yaml
+++ firefox-128.0/modules/libpref/init/StaticPrefList.yam
@@ -10124,6 +10124,8 @@
   value: true
 #elif defined(XP_OPENBSD)
   value: true
+#elif defined(XP_SOLARIS)
+  value: true
 #else
   value: false
 #endif

Maybe XP_SOLARIS need to be defined for yaml somewhere...

is it working elsewhere? At least it's a relief if it works and we just need to fix the define ...

Comment 18

[:gerard-majax]

it's not the C++ preprocessor so maybe you have to hack it there: https://searchfox.org/mozilla-central/rev/b220e40ff2ee3d10ce68e07d8a8a577d5558e2a2/modules/libpref/init/StaticPrefList.yaml#123

Comment 19

[Mike Hommey [:glandium]]

It should be defined via https://searchfox.org/mozilla-central/rev/b220e40ff2ee3d10ce68e07d8a8a577d5558e2a2/modules/libpref/init/generate_static_pref_list.py#419

Comment 20

[Petr Sumbera]

Sorry again. I was enabling media.rdd-process.enabled instead. So I expect XP_SOLARIS is working. I'm will check it...

Comment 21

[Petr Sumbera]

Attachment: Bug1908317.patch

Ok, actually both media.rdd-process.enabled and media.utility-process.enabled need to be set to true so that I can play videos.

Therefore I have set also all media.rdd-* as BSDs do.

Can you please integrate it? Ideally also into 128 branch.

Thank you both!

Comment 22

[:gerard-majax]

Can you submit it via Phab? I can approve and ask for uplift

Comment 23

[:gerard-majax]

Attachment: Bug 1908317 - Add XP_SOLARIS to media-related prefs r?padenot!

Comment 24

[:gerard-majax]

Comment on attachment 9414053 [details]
Bug 1908317 - Add XP_SOLARIS to media-related prefs r?padenot!

ESR Uplift Approval Request

• If this is not a sec:{high,crit} bug, please state case for ESR consideration: Solaris maintainer mostly targets ESR
• User impact if declined: No media playback working on Solaris
• Fix Landed on Version: 130
• Risk to taking this patch: Low
• Why is the change risky/not risky? (and alternatives if risky): Only flips prefs on Solaris, tested by Solaris maintainer

Comment 25

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/8c746156d7c5
Add XP_SOLARIS to media-related prefs r=padenot

Comment 26

[Petr Sumbera]

Thank you very much for fixing this. I just wonder whether setting media.audioFocus.management for Solaris is intentional?

Comment 27

[:gerard-majax]

(In reply to Petr Sumbera from comment #26)

Thank you very much for fixing this. I just wonder whether setting media.audioFocus.management for Solaris is intentional?

I took your patch as-is, I made no change at all

Comment 28

[Iulian Moraru]

https://hg.mozilla.org/mozilla-central/rev/8c746156d7c5

Comment 29

[BugBot [:suhaib / :marco]]

Set release status flags based on info from the regressing bug 1835804

Comment 30

[Alastor Wu [:alwu]]

(In reply to :gerard-majax from comment #27)

(In reply to Petr Sumbera from comment #26)

Thank you very much for fixing this. I just wonder whether setting media.audioFocus.management for Solaris is intentional?

I took your patch as-is, I made no change at all

That shouldn't be enabled on Solaris. That is only enabled by default on Android.

Comment 31

[Norisz Fay [:noriszfay]]

Backed out as requested by developer

Backout link

Comment 32

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/7d083d75388a
Add XP_SOLARIS to media-related prefs r=padenot

Comment 33

[Pulsebot]

Backout by [e-mail]:
https://hg.mozilla.org/mozilla-central/rev/17630c12ac58
Backed out changeset 8c746156d7c5 as requested by developer CLOSED TREE

Comment 34

[Norisz Fay [:noriszfay]]

The backout was merged to central and then merged back to autoland. As a result of this, there's a chance this has to be relanded on autoland in order for it to get fixed. Here is the central revision for the backout: https://hg.mozilla.org/mozilla-central/rev/17630c12ac58

Comment 35

[:gerard-majax]

re-landed from phab, are we good?

Comment 36

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/9aa5f8358ff3
Add XP_SOLARIS to media-related prefs r=padenot

Comment 37

[Norisz Fay [:noriszfay]]

Yes, that is fine

Comment 38

[Serban Stanca [:SerbanS]]

https://hg.mozilla.org/mozilla-central/rev/7d083d75388a
https://hg.mozilla.org/mozilla-central/rev/9aa5f8358ff3

Comment 39

[Donal Meehan [:dmeehan]]

Comment on attachment 9414053 [details]
Bug 1908317 - Add XP_SOLARIS to media-related prefs r?padenot!

Approved for 128.1esr.

Comment 40

[Pulsebot]

https://hg.mozilla.org/releases/mozilla-esr128/rev/15725075b76e