Post new topic   Reply to topic
View previous topic Printable version Log in to check your private messages View next topic
Author Message
dkadiogluOffline
15 Post subject: [solved] possible regression with kernel 2.6.37  PostPosted: 05.01.2011, 10:43



Joined: 2011-01-05
Posts: 16

Status: Offline
Hi,

@slh
as I was told in IRC I'm posting my problem with my bluetooth input devices and kernel 2.6.37 here. They work until I don't touch them for some seconds and my bluetooth device autosuspends. Then dmesg shows the following:

      Code:
btusb 5-1:1.0: no reset_resume for driver btusb?
[ 1250.747352] btusb 5-1:1.1: no reset_resume for driver btusb?


One time I got the following trace: http://pastebin.com/nuP5jGFw

Can you please look into this, slh?

Thanks in advance!

dkadioglu


Last edited by dkadioglu on 25.02.2011, 10:06; edited 1 time in total
 
 View user's profile Send private message  
Reply with quote Back to top
dkadiogluOffline
Post subject: RE: possible regression with kernel 2.6.37  PostPosted: 05.01.2011, 22:02



Joined: 2011-01-05
Posts: 16

Status: Offline
I investigated the problem further and I'm almost sure that the autosuspend functionality of usbcore is the culprit. I deactivated it with usbcore.autosuspend=-1 appended to the kernel command line and now everything works as it should! No more errors in dmesg and no more hangs of my bluetooth input devices.

Anyway, slh, I hope you find the time to look into it as deactivating autosuspend completely is only a workaround. If you need further information, please ask.
 
 View user's profile Send private message  
Reply with quote Back to top
slhOffline
Post subject: RE: possible regression with kernel 2.6.37  PostPosted: 05.01.2011, 22:34



Joined: 2010-08-25
Posts: 747

Status: Offline
It would be interesting to determine at which exact step that bug occurs, while I can't really test bluetooth in practice, I can't confirm any problem while plugging in an USB bluetooth stick and issuing "hcitool scan" (I can't test anything else, as I don't use any bluetooth devices myself). It might help to raise this issue, with its call trace, at bugzilla.kernel.org, where someone more familiar with bluetooth might take a look.

USB autosuspend has been enabled by pretty much all distributions for years, which means in practice that the non-suspending code paths are not tested thoroughly anymore. Given that the last time I tried disabling it (because in the beginning it uncovered bugs in many drivers not expecting to be suspended), I received multiple breakage reports within hours, I'm not going to disable USB_SUSPEND again.
 
 View user's profile Send private message  
Reply with quote Back to top
dkadiogluOffline
Post subject: RE: possible regression with kernel 2.6.37  PostPosted: 05.01.2011, 23:55



Joined: 2011-01-05
Posts: 16

Status: Offline
The weird thing is, that my bluetooth devices work without problem until I don't touch them for some seconds. When I touch them again they come up or not (I can't reproduce that by now). Most times they don't work anymore and in dmesg I can find the above posted error concerning no reset_resume for btusb.

I've posted it on bugzilla.kernel.org: https://bugzilla.kernel.org/show_bug.cgi?id=26182
Hopefully sb. can help me with that.

Thanks for now slh!
 
 View user's profile Send private message  
Reply with quote Back to top
dkadiogluOffline
Post subject: RE: possible regression with kernel 2.6.37  PostPosted: 01.02.2011, 13:36



Joined: 2011-01-05
Posts: 16

Status: Offline
Hi again,

my problems concerning the usability of my bluetooth devices got even worse -> https://bugzilla.kernel.org/show_bug.cgi?id=26182

By now I've got no answer from anyone regarding my bug report. Any idea what else I can do?

Thanks and Regards
 
 View user's profile Send private message  
Reply with quote Back to top
dkadiogluOffline
Post subject: RE: possible regression with kernel 2.6.37  PostPosted: 24.02.2011, 21:29



Joined: 2011-01-05
Posts: 16

Status: Offline
Hi slh,

could you please add the following commit to the next kernel update: http://git.kernel.org/?p=linux/kernel/g ... b56d88ac03
According to the replies in the bug report, this should fix my problem with btusb.

Thanks in advance.
Regards
 
 View user's profile Send private message  
Reply with quote Back to top
dkadiogluOffline
Post subject: RE: possible regression with kernel 2.6.37  PostPosted: 25.02.2011, 10:05



Joined: 2011-01-05
Posts: 16

Status: Offline
Thanks for integrating the fix in the latest update! This fixed the problem for me!

Regards
 
 View user's profile Send private message  
Reply with quote Back to top
Display posts from previous:     
Jump to:  
All times are GMT - 12 Hours
Post new topic   Reply to topic
View previous topic Printable version Log in to check your private messages View next topic
Powered by Zafenio