[AE] XBMC constantly hogging audio device
#16
(2013-01-07, 04:00)DDDamian Wrote: Just to weigh in - yes, we did release the ALSA device in earlier builds, but it was reported to cause issues for some as ALSA didn't like being started and stopped frequently. The issue was specific to ALSA - the other platform where we needed to release the device was the Windows WASAPI exclusive mode, and that platform is now fine.
Thanks for this statement.

Why not let the user configure it with the streamsilence setting? There are systems out there where starting and stopping ALSA devices does not cause any issues (including my HTPC Wink). In the short-term this setting might help you guys in releasing Frodo, as you can still offer suspend/resume support for ALSA sinks. In the long-term I guess this can be resolved in a clean and generic way.
Reply
#17
@vfred Because its broken?

uNi
Reply
#18
I just found this thread after searching the forums for this issue. I have created a Track ticket (#13879) about this issue two weeks ago and thought you guys might not be aware of the problem, but after reading through this thread I noticed that the XBMC developers luckily are aware of it.

Now from a user-perspective I see a few things that could be improved communication-wise in the future when dealing with such things:
  • Add critical issues such as this one as release blockers on Github. This way users can see that this is being worked on and might stop asking stupid questions. And people like me would not start opening Trac tickets, which would probably be a huge time-saver for you devs out there.
  • Another thing you might consider is not bumping RC release numbers if there are critical issues and known regressions. Just a thought. Right now the release schedule looks like the next release is 12.0 final. Since davilla stated that this needs to be fixed in 12.0 you are giving users very little time to actually verify that the problem has been solved. Yes, you could release an additional RC which contains the fix, but this could have been avoided if this known issue would have been set as a release blocker for an earlier RC or beta.

Now please don't get me wrong. I think the XBMC team is doing an awesome job, creating a very cool piece of software. This is just a suggestion from an average user, trying to help you guys in making the XBMC community even cooler.
Main: CPU: Intel Core i7-4790K, GFX: AMD RX480 8GB OC, Debian Linux
HTPC: CPU: Intel Core i5-3475S, GFX: Intel HD4000, Gentoo Linux (Kernel 4-stable), Kodi with 4K@30Hz and HD audio, AVR: Denon X6300H with 7.1.4 setup, TV: Samsung UE55HU7590@SEK3500
Fun: Microsoft XBox, with XBMC :)
Reply
#19
Fair suggestions. Trying to get each platform to follow expected behaviour has been the issue, especially for Linux and Android, and the work really needs to be done in the sinks to keep the engine as generic as possible. That said, devs for those platforms have been working hard on this as it's been identified as a blocker.

There's a catch-22 on Linux especially: many users need audio to suspend to release it for external players or to reduce CPU usage when idle, some need it to "stream silence" so their receivers don't drop the input, and in many ALSA/hardware configs ALSA does not seem to play nicely in it's own pause routines, nor being stopped and restarted. It caused audio to crap out on some systems. Turning that feature off lead to this issue of hogging the device.
System: XBMC HTPC with HDMI WASAPI & AudioEngine - Denon  AVR-3808CI  - Denon DVD-5900 Universal Player  - Denon DCM-27 CD-Changer
- Sony BDP-S580 Blu-Ray  - X-Box 360  - Android tablet wireless remote - 7.1 Streem/Axiom/Velodyne Surround System
If I have been able to help feel free to add to my reputation +/- below - thanks!
Reply
#20
FYI

2032 (GH issue)
2074 (GH issue)
http://forum.xbmc.org/showthread.php?tid=147938
http://forum.xbmc.org/showthread.php?tid=150963
http://forum.xbmc.org/showthread.php?tid=148428
http://forum.xbmc.org/showthread.php?tid=136140
http://forum.xbmc.org/showthread.php?tid=150791
http://forum.xbmc.org/showthread.php?pid=1266056

I've openend ticket for listing dmix devices which would be a workaround for these issues.
http://trac.xbmc.org/ticket/13966

Let's hope someone finds the time to fix it. I'm still searching for it, time that is.
Reply
#21
@Falcon1: You are right that dmix can work around a few issues in some cases. I have a 7.1 speaker setup for instance and dmix would not really help here as long as #13195 is not fixed either, as the number of channels cannot be detected automatically by dmix.

One issue which would definitely remain with dmix is the higher power usage, which is still a pretty major thing for 24/7 HTPCs. Right now it looks like there are still a lot of regressions an the ALSA-side. Sad
Main: CPU: Intel Core i7-4790K, GFX: AMD RX480 8GB OC, Debian Linux
HTPC: CPU: Intel Core i5-3475S, GFX: Intel HD4000, Gentoo Linux (Kernel 4-stable), Kodi with 4K@30Hz and HD audio, AVR: Denon X6300H with 7.1.4 setup, TV: Samsung UE55HU7590@SEK3500
Fun: Microsoft XBox, with XBMC :)
Reply
#22
(2013-01-06, 19:26)davilla Wrote: it is a release blocker.
signed "The Release Manager"
I just installed 12.0 stable and XBMC is still blocking the ALSA device. Are there any options we need to set in order to get suspend/resume support for the ALSA sinks?
Reply
#23
(2013-01-30, 10:24)vfred Wrote:
(2013-01-06, 19:26)davilla Wrote: it is a release blocker.
signed "The Release Manager"
I just installed 12.0 stable and XBMC is still blocking the ALSA device. Are there any options we need to set in order to get suspend/resume support for the ALSA sinks?

Same problem. Seems that linux has no priority. At least there should be option. Around new year it was working with a git snapshot. But seems the fix was reverted shortly after commit.

Reply
#24
A fix is a valid fix, if it fixes something without introducing regressions.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
#25
Yeah and release blocker apparently is not a release blocker.

mythtv can do HD audio via alsa without continuously blocking the alsa device, different codebase I realise, but it shows it is possible.
If I have helped you or increased your knowledge, click the 'thumbs up' button to give thanks :) (People with less than 20 posts won't see the "thumbs up" button.)
Reply
#26
It's possible to release the audio device by changing the output device in XBMC (Settings->System->Audio Output->Audio output device), so if there is some way to script this then an external player could release it when playing and reset it after.

In Eden it was possible to change the audio device using python and the HTTP API but this has been deprecated and the JSON-RPC API does not appear to support changing this setting.

Any other ideas?
Reply
#27
@nickr:
Patch is welcome. There can be done a whole lot, if you have the time to do so.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
#28
I would also like to see this fixed, or at least get an option to tell XBMC to release it when not playing anything. I use a Wimp (music streaming service similar to Spotify), and prefer to start it in a x2go session from another pc, so i dont have to log out of XBMC, log into the desktop to play music, and then back into XBMC again when im done playing music in Wimp.
Reply
#29
I just compiled the latest Frodo-branch from git and it looks like this issue has finally been fixed. I'm looking forward to a great 12.1 release, thanks everybody. But wait - what happened to the 12.1 milestone on Github?
Main: CPU: Intel Core i7-4790K, GFX: AMD RX480 8GB OC, Debian Linux
HTPC: CPU: Intel Core i5-3475S, GFX: Intel HD4000, Gentoo Linux (Kernel 4-stable), Kodi with 4K@30Hz and HD audio, AVR: Denon X6300H with 7.1.4 setup, TV: Samsung UE55HU7590@SEK3500
Fun: Microsoft XBox, with XBMC :)
Reply

Logout Mark Read Team Forum Stats Members Help
[AE] XBMC constantly hogging audio device1