Hi,
I have noticed a possible bug over the past ten days that pretty soon makes it impossible to do anything related to sound and then necessitates a reboot (complete with hard power button press). Here is what happened the last time (an hour ago) when I was forced to reboot:
I got:
After Stopping session c1 of User "xyz"
(*) A stop job is running on session c1 for user "xyz"
(+) A stop job is running for Manage Sound Card State (restore & state)
After Stopped Login service, I get
(-) A stop job is running for User Manager for UID yyyy
(+) A stop job is running for Manage Sound Card State (restore & state)
After Stopped Authorization Manager, I get
(+) A stop job is running for Manage Sound Card State (restore & state)
After about 2 mins, (*) and (-) get done (separately), and after about 6 min (+) gets done.
However, I get a message:
Waiting for process: alsactl, wireplumber, chromium-browse (not browser), pipewire
and after a while (because I have already wasted close to 30 minutes if you add the time it took me to figure out the need to do a reboot), I give up and do a hard reboot.
I wonder what the problem is. Note that I did shut down chromium-browser and firefox before reboot. I rebooted into the 5.16 kernel ten days ago. I am on a Dell XPS13 and do nightly updates. I also hibernate every night and usually do not reboot for weeks: I can not recall why I decided to reboot 10 days ago, but it must have been some oddity. My guess is that I was hibernating and waking for at least 20 days before that.
Any suggestions? Let me know what additional information I can provide. If it is indeed a bug, what should I file under?
Many thanks and best wishes, Ranjan
On Sun, 20 Feb 2022 09:49:38 -0600 Ranjan Maitra mlmaitra@gmx.com wrote:
Hi,
I have noticed a possible bug over the past ten days that pretty soon makes it impossible to do anything related to sound and then necessitates a reboot (complete with hard power button press). Here
[snip]
I wonder what the problem is. Note that I did shut down chromium-browser and firefox before reboot. I rebooted into the 5.16 kernel ten days ago. I am on a Dell XPS13 and do nightly updates. I also hibernate every night and usually do not reboot for weeks: I can not recall why I decided to reboot 10 days ago, but it must have been some oddity. My guess is that I was hibernating and waking for at least 20 days before that.
I don't think hibernating has anything to do with the problem.
Any suggestions? Let me know what additional information I can provide. If it is indeed a bug, what should I file under?
I *think* it might be that one of the browsers is not closing a sound callback or poll that it has set up. Pipewire or wireplumber try to close it during shutdown, but it is orphaned, so no longer active. A zombie! Someone had that problem with a discord connection recently on the mailing list. Open the ticket against wireplumber, putting all the information you have in this email. Again, I *think*, if my speculation is correct, that wireplumber should, under certain circumstances, ignore such links during shutdown.
The browser *should* close such links when the tab is closed or on shutdown, so it is the party really at fault, but having wireplumber ignore such links during shutdown is a quick and dirty workaround. You could experiment to see which browser is causing the issue and open a bugzilla against that one if you can figure it out. But, that seems to be completely antithetical to your work process.
This is all guesses and speculation, so could be completely wrong.
Hi Stan,
Thanks for the response.
[snip]
I wonder what the problem is. Note that I did shut down chromium-browser and firefox before reboot. I rebooted into the 5.16 kernel ten days ago. I am on a Dell XPS13 and do nightly updates. I also hibernate every night and usually do not reboot for weeks: I can not recall why I decided to reboot 10 days ago, but it must have been some oddity. My guess is that I was hibernating and waking for at least 20 days before that.
I don't think hibernating has anything to do with the problem.
I do not think so either, thanks.
Any suggestions? Let me know what additional information I can provide. If it is indeed a bug, what should I file under?
I *think* it might be that one of the browsers is not closing a sound callback or poll that it has set up. Pipewire or wireplumber try to close it during shutdown, but it is orphaned, so no longer active. A zombie! Someone had that problem with a discord connection recently on the mailing list. Open the ticket against wireplumber, putting all the information you have in this email. Again, I *think*, if my speculation is correct, that wireplumber should, under certain circumstances, ignore such links during shutdown.
Thank you, I will do this.
Question: is there more output I can provide from somewhere?
The browser *should* close such links when the tab is closed or on shutdown, so it is the party really at fault, but having wireplumber ignore such links during shutdown is a quick and dirty workaround. You could experiment to see which browser is causing the issue and open a bugzilla against that one if you can figure it out. But, that seems to be completely antithetical to your work process.
Right, initially, I was thinking it was zoom related but that is largely because zoom uses audio. So, I doubt that it is related to the browser strictly speaking.
This is all guesses and speculation, so could be completely wrong.
Thanks again, other suggestions would be welcome.
Best wishes, Ranjan
On Sun, 20 Feb 2022 11:50:03 -0600 Ranjan Maitra mlmaitra@gmx.com wrote:
Hi Stan,
I *think* it might be that one of the browsers is not closing a sound callback or poll that it has set up. Pipewire or wireplumber try to close it during shutdown, but it is orphaned, so no longer active. A zombie! Someone had that problem with a discord connection recently on the mailing list. Open the ticket against wireplumber, putting all the information you have in this email. Again, I *think*, if my speculation is correct, that wireplumber should, under certain circumstances, ignore such links during shutdown.
Thank you, I will do this.
Question: is there more output I can provide from somewhere?
The only think I can think of is the journal. You could look at the messages from the long running shutdown and see if there are any related to this. When I shutdown, if there is something that is sleeping or suspended, I see a message on the console that it is waiting for the offending process. That might make it into the journal, and give a pointer to the offender.
On Sun, 2022-02-20 at 09:49 -0600, Ranjan Maitra wrote:
Hi,
I have noticed a possible bug over the past ten days that pretty soon makes it impossible to do anything related to sound and then necessitates a reboot (complete with hard power button press). Here is what happened the last time (an hour ago) when I was forced to reboot:
I got:
After Stopping session c1 of User "xyz"
(*) A stop job is running on session c1 for user "xyz"
(+) A stop job is running for Manage Sound Card State (restore & state)
After Stopped Login service, I get
(-) A stop job is running for User Manager for UID yyyy
(+) A stop job is running for Manage Sound Card State (restore & state)
After Stopped Authorization Manager, I get
(+) A stop job is running for Manage Sound Card State (restore & state)
After about 2 mins, (*) and (-) get done (separately), and after about 6 min (+) gets done.
However, I get a message:
Waiting for process: alsactl, wireplumber, chromium-browse (not browser), pipewire
and after a while (because I have already wasted close to 30 minutes if you add the time it took me to figure out the need to do a reboot), I give up and do a hard reboot.
I wonder what the problem is. Note that I did shut down chromium- browser and firefox before reboot. I rebooted into the 5.16 kernel ten days ago. I am on a Dell XPS13 and do nightly updates. I also hibernate every night and usually do not reboot for weeks: I can not recall why I decided to reboot 10 days ago, but it must have been some oddity. My guess is that I was hibernating and waking for at least 20 days before that.
Any suggestions? Let me know what additional information I can provide. If it is indeed a bug, what should I file under?
Many thanks and best wishes, Ranjan _______________________________________________
Hi,
Not sure is it the same bug, but recently I had issues with never kernel's breaking audio completely, wireplumber gets stuck in Uninterruptible sleep (D state), no audio at all, only option is to hard reboot.
This is one of the bugs, but there are links to pipewire bug, kernel bug ...[1]
For me last working kernel in 5.16.x series is kernel-5.16.7- 200.fc35.x86_64.
Some people mention suspend2ram/sleep, but for me it is broken from first boot, sleep is completely broken, since it's not possible to suspend that wireplumber process. Also other people have issues with USB audio, but in my case it's integrated audio.
Not sure if it helps, but possibly related.
Regards, Branko
[1] https://gitlab.freedesktop.org/pipewire/wireplumber/-/issues/186
On Sun Feb20'22 07:11:11PM, Branko Grubić wrote:
From: Branko Grubić bitlord0xff@gmail.com Date: Sun, 20 Feb 2022 19:11:11 +0100 To: Community support for Fedora users users@lists.fedoraproject.org Reply-To: Community support for Fedora users users@lists.fedoraproject.org Subject: Re: audio-related bug(?) on Fedora 35
On Sun, 2022-02-20 at 09:49 -0600, Ranjan Maitra wrote:
Hi,
I have noticed a possible bug over the past ten days that pretty soon makes it impossible to do anything related to sound and then necessitates a reboot (complete with hard power button press). Here is what happened the last time (an hour ago) when I was forced to reboot:
I got:
After Stopping session c1 of User "xyz"
(*) A stop job is running on session c1 for user "xyz"
(+) A stop job is running for Manage Sound Card State (restore & state)
After Stopped Login service, I get
(-) A stop job is running for User Manager for UID yyyy
(+) A stop job is running for Manage Sound Card State (restore & state)
After Stopped Authorization Manager, I get
(+) A stop job is running for Manage Sound Card State (restore & state)
After about 2 mins, (*) and (-) get done (separately), and after about 6 min (+) gets done.
However, I get a message:
Waiting for process: alsactl, wireplumber, chromium-browse (not browser), pipewire
and after a while (because I have already wasted close to 30 minutes if you add the time it took me to figure out the need to do a reboot), I give up and do a hard reboot.
I wonder what the problem is. Note that I did shut down chromium- browser and firefox before reboot. I rebooted into the 5.16 kernel ten days ago. I am on a Dell XPS13 and do nightly updates. I also hibernate every night and usually do not reboot for weeks: I can not recall why I decided to reboot 10 days ago, but it must have been some oddity. My guess is that I was hibernating and waking for at least 20 days before that.
Any suggestions? Let me know what additional information I can provide. If it is indeed a bug, what should I file under?
Many thanks and best wishes, Ranjan _______________________________________________
Hi,
Not sure is it the same bug, but recently I had issues with never kernel's breaking audio completely, wireplumber gets stuck in Uninterruptible sleep (D state), no audio at all, only option is to hard reboot.
This is one of the bugs, but there are links to pipewire bug, kernel bug ...[1]
For me last working kernel in 5.16.x series is kernel-5.16.7- 200.fc35.x86_64.
Some people mention suspend2ram/sleep, but for me it is broken from first boot, sleep is completely broken, since it's not possible to suspend that wireplumber process. Also other people have issues with USB audio, but in my case it's integrated audio.
Not sure if it helps, but possibly related.
Regards, Branko
[1] https://gitlab.freedesktop.org/pipewire/wireplumber/-/issues/186
I see, thank you. I think I first booted into 5.16 kernel (and indeed beyond 5.15.9 kernel) about ten days ago so this seems plausible. I have now booted into my oldest kernel (5.16.7) and will let you know if the problem persists.
FWIW, I also have that Logitech H390 USB COMPUTER HEADSET plugged in. It often drops in and out in zoom calls, and I have not related it previously, but it is possible that the first time this happens, problems start from there. But hard reboot "solves" the problem.
The problem with me is not no sound after suspend, but sound itself starts being unusable after a while (of using sound).
Oh, I now recall what the original issue was: hibernate was not working, in the sense that it was coming back on without hibernating. But it is from a week ago and not easily reconstructed now.
If I were to: sudo dnf downgrade pipewire as suggested then would that not get upgraded at the next nightly update. Perhaps I do the downgrade after every update.
Btw, this one is exactly what is happening: https://ask.fedoraproject.org/t/fedora-35-not-working-well-after-wakeup/2012... but better documented than my post here.
Thanks again! Ranjan
On Sun Feb20'22 07:11:11PM, Branko Grubić wrote:
From: Branko Grubić bitlord0xff@gmail.com Date: Sun, 20 Feb 2022 19:11:11 +0100 To: Community support for Fedora users users@lists.fedoraproject.org Reply-To: Community support for Fedora users users@lists.fedoraproject.org Subject: Re: audio-related bug(?) on Fedora 35
On Sun, 2022-02-20 at 09:49 -0600, Ranjan Maitra wrote:
Hi,
I have noticed a possible bug over the past ten days that pretty soon makes it impossible to do anything related to sound and then necessitates a reboot (complete with hard power button press). Here is what happened the last time (an hour ago) when I was forced to reboot:
I got:
After Stopping session c1 of User "xyz"
(*) A stop job is running on session c1 for user "xyz"
(+) A stop job is running for Manage Sound Card State (restore & state)
After Stopped Login service, I get
(-) A stop job is running for User Manager for UID yyyy
(+) A stop job is running for Manage Sound Card State (restore & state)
After Stopped Authorization Manager, I get
(+) A stop job is running for Manage Sound Card State (restore & state)
After about 2 mins, (*) and (-) get done (separately), and after about 6 min (+) gets done.
However, I get a message:
Waiting for process: alsactl, wireplumber, chromium-browse (not browser), pipewire
and after a while (because I have already wasted close to 30 minutes if you add the time it took me to figure out the need to do a reboot), I give up and do a hard reboot.
I wonder what the problem is. Note that I did shut down chromium- browser and firefox before reboot. I rebooted into the 5.16 kernel ten days ago. I am on a Dell XPS13 and do nightly updates. I also hibernate every night and usually do not reboot for weeks: I can not recall why I decided to reboot 10 days ago, but it must have been some oddity. My guess is that I was hibernating and waking for at least 20 days before that.
Any suggestions? Let me know what additional information I can provide. If it is indeed a bug, what should I file under?
Many thanks and best wishes, Ranjan _______________________________________________
Hi,
Not sure is it the same bug, but recently I had issues with never kernel's breaking audio completely, wireplumber gets stuck in Uninterruptible sleep (D state), no audio at all, only option is to hard reboot.
This is one of the bugs, but there are links to pipewire bug, kernel bug ...[1]
For me last working kernel in 5.16.x series is kernel-5.16.7- 200.fc35.x86_64.
Some people mention suspend2ram/sleep, but for me it is broken from first boot, sleep is completely broken, since it's not possible to suspend that wireplumber process. Also other people have issues with USB audio, but in my case it's integrated audio.
Not sure if it helps, but possibly related.
Regards, Branko
[1] https://gitlab.freedesktop.org/pipewire/wireplumber/-/issues/186
Actually, another suggestion in one of those links is to take out alsa-utils and that is what i have now done with kernel 5.16.9. I will report if that has any benefit. However, the volume appears lower.
Ranjan
On Sun Feb20'22 01:08:29PM, Ranjan Maitra wrote:
From: Ranjan Maitra mlmaitra@gmx.com Date: Sun, 20 Feb 2022 13:08:29 -0600 To: Community support for Fedora users users@lists.fedoraproject.org Reply-To: Community support for Fedora users users@lists.fedoraproject.org Subject: Re: audio-related bug(?) on Fedora 35
On Sun Feb20'22 07:11:11PM, Branko Grubić wrote:
From: Branko Grubić bitlord0xff@gmail.com Date: Sun, 20 Feb 2022 19:11:11 +0100 To: Community support for Fedora users users@lists.fedoraproject.org Reply-To: Community support for Fedora users users@lists.fedoraproject.org Subject: Re: audio-related bug(?) on Fedora 35
On Sun, 2022-02-20 at 09:49 -0600, Ranjan Maitra wrote:
Hi,
I have noticed a possible bug over the past ten days that pretty soon makes it impossible to do anything related to sound and then necessitates a reboot (complete with hard power button press). Here is what happened the last time (an hour ago) when I was forced to reboot:
I got:
After Stopping session c1 of User "xyz"
(*) A stop job is running on session c1 for user "xyz"
(+) A stop job is running for Manage Sound Card State (restore & state)
After Stopped Login service, I get
(-) A stop job is running for User Manager for UID yyyy
(+) A stop job is running for Manage Sound Card State (restore & state)
After Stopped Authorization Manager, I get
(+) A stop job is running for Manage Sound Card State (restore & state)
After about 2 mins, (*) and (-) get done (separately), and after about 6 min (+) gets done.
However, I get a message:
Waiting for process: alsactl, wireplumber, chromium-browse (not browser), pipewire
and after a while (because I have already wasted close to 30 minutes if you add the time it took me to figure out the need to do a reboot), I give up and do a hard reboot.
I wonder what the problem is. Note that I did shut down chromium- browser and firefox before reboot. I rebooted into the 5.16 kernel ten days ago. I am on a Dell XPS13 and do nightly updates. I also hibernate every night and usually do not reboot for weeks: I can not recall why I decided to reboot 10 days ago, but it must have been some oddity. My guess is that I was hibernating and waking for at least 20 days before that.
Any suggestions? Let me know what additional information I can provide. If it is indeed a bug, what should I file under?
Many thanks and best wishes, Ranjan _______________________________________________
Hi,
Not sure is it the same bug, but recently I had issues with never kernel's breaking audio completely, wireplumber gets stuck in Uninterruptible sleep (D state), no audio at all, only option is to hard reboot.
This is one of the bugs, but there are links to pipewire bug, kernel bug ...[1]
For me last working kernel in 5.16.x series is kernel-5.16.7- 200.fc35.x86_64.
Some people mention suspend2ram/sleep, but for me it is broken from first boot, sleep is completely broken, since it's not possible to suspend that wireplumber process. Also other people have issues with USB audio, but in my case it's integrated audio.
Not sure if it helps, but possibly related.
Regards, Branko
[1] https://gitlab.freedesktop.org/pipewire/wireplumber/-/issues/186
Actually, another suggestion in one of those links is to take out alsa-utils and that is what i have now done with kernel 5.16.9. I will report if that has any benefit. However, the volume appears lower.
So, removing alsa-utils and using kernel 5.16.9 has no effect in the problem. The only difference is that the process that hangs is now (only) wireplumber. Now trying kernel 5.16.7 (having brought alsa-utils back).
Thanks, Ranjan
On Sun Feb20'22 04:45:18PM, Ranjan Maitra wrote:
From: Ranjan Maitra mlmaitra@gmx.com Date: Sun, 20 Feb 2022 16:45:18 -0600 To: Community support for Fedora users users@lists.fedoraproject.org Reply-To: Community support for Fedora users users@lists.fedoraproject.org Subject: Re: audio-related bug(?) on Fedora 35
On Sun Feb20'22 01:08:29PM, Ranjan Maitra wrote:
From: Ranjan Maitra mlmaitra@gmx.com Date: Sun, 20 Feb 2022 13:08:29 -0600 To: Community support for Fedora users users@lists.fedoraproject.org Reply-To: Community support for Fedora users users@lists.fedoraproject.org Subject: Re: audio-related bug(?) on Fedora 35
On Sun Feb20'22 07:11:11PM, Branko Grubić wrote:
From: Branko Grubić bitlord0xff@gmail.com Date: Sun, 20 Feb 2022 19:11:11 +0100 To: Community support for Fedora users users@lists.fedoraproject.org Reply-To: Community support for Fedora users users@lists.fedoraproject.org Subject: Re: audio-related bug(?) on Fedora 35
On Sun, 2022-02-20 at 09:49 -0600, Ranjan Maitra wrote:
Hi,
I have noticed a possible bug over the past ten days that pretty soon makes it impossible to do anything related to sound and then necessitates a reboot (complete with hard power button press). Here is what happened the last time (an hour ago) when I was forced to reboot:
I got:
After Stopping session c1 of User "xyz"
(*) A stop job is running on session c1 for user "xyz"
(+) A stop job is running for Manage Sound Card State (restore & state)
After Stopped Login service, I get
(-) A stop job is running for User Manager for UID yyyy
(+) A stop job is running for Manage Sound Card State (restore & state)
After Stopped Authorization Manager, I get
(+) A stop job is running for Manage Sound Card State (restore & state)
After about 2 mins, (*) and (-) get done (separately), and after about 6 min (+) gets done.
However, I get a message:
Waiting for process: alsactl, wireplumber, chromium-browse (not browser), pipewire
and after a while (because I have already wasted close to 30 minutes if you add the time it took me to figure out the need to do a reboot), I give up and do a hard reboot.
I wonder what the problem is. Note that I did shut down chromium- browser and firefox before reboot. I rebooted into the 5.16 kernel ten days ago. I am on a Dell XPS13 and do nightly updates. I also hibernate every night and usually do not reboot for weeks: I can not recall why I decided to reboot 10 days ago, but it must have been some oddity. My guess is that I was hibernating and waking for at least 20 days before that.
Any suggestions? Let me know what additional information I can provide. If it is indeed a bug, what should I file under?
Many thanks and best wishes, Ranjan _______________________________________________
Hi,
Not sure is it the same bug, but recently I had issues with never kernel's breaking audio completely, wireplumber gets stuck in Uninterruptible sleep (D state), no audio at all, only option is to hard reboot.
This is one of the bugs, but there are links to pipewire bug, kernel bug ...[1]
For me last working kernel in 5.16.x series is kernel-5.16.7- 200.fc35.x86_64.
Some people mention suspend2ram/sleep, but for me it is broken from first boot, sleep is completely broken, since it's not possible to suspend that wireplumber process. Also other people have issues with USB audio, but in my case it's integrated audio.
Not sure if it helps, but possibly related.
Regards, Branko
[1] https://gitlab.freedesktop.org/pipewire/wireplumber/-/issues/186
Actually, another suggestion in one of those links is to take out alsa-utils and that is what i have now done with kernel 5.16.9. I will report if that has any benefit. However, the volume appears lower.
So, removing alsa-utils and using kernel 5.16.9 has no effect in the problem. The only difference is that the process that hangs is now (only) wireplumber. Now trying kernel 5.16.7 (having brought alsa-utils back).
With kernel 5.16.7, I did not initially have the above problem, but then, the machine would not hibernate. Would try to do so, shut down nm-applet and then reconnect (without it even going down -- screen shutting off at all). I had this problem earlier too.
Upon doing a reboot, I got the above behavior in toto (except no chromium-browse and no pipewire in the Waiting for process system-shutdown, etc).
Now I have done dnf downgrade my kernel (and was taken to 5.14.10). So far, it has done two rounds of sound, and hibernated and come back once. I will observe a bit more and report on how things go with this kernel.
Many thanks and best, Ranjan
-
Branko Grubić -
Ranjan Maitra -
stan