ohci_hcd leak ed

ohci_hcd leak ed

[DiP]

Hi all

When I take my two sonix-based webcams to my box, and started vlc on both I got

<code>
Aug 19 02:14:45 mixer kernel: [  558.384176] ohci_hcd 0000:00:0b.0: leak ed f53dd580 (#81) state 2
[repeated many times...]
Aug 19 02:14:45 mixer kernel: [  558.406651] gspca: no transfer endpoint found
</code>

That with standard archlinux kernel 3.0
Linux mixer 3.0-ARCH #1 SMP PREEMPT Sat Aug 6 16:49:00 CEST 2011 i686 Mobile
AMD Sempron(tm) Processor 3800+ AuthenticAMD GNU/Linux

Also I got almost the same with my other box (kernel is hand-compiled)
Linux di 3.0.1-pf-di #10 SMP PREEMPT Wed Aug 17 19:23:37 EEST 2011 x86_64 Intel
(R) Core(TM)2 Duo CPU E8200 @ 2.66GHz GenuineIntel GNU/Linux

Aug 18 23:36:37 di kernel: gspca: no transfer endpoint found
Aug 18 23:36:38 di kernel: gspca: no transfer endpoint found
Aug 18 23:37:22 di kernel: gspca: no transfer endpoint found
Aug 18 23:37:23 di kernel: gspca: no transfer endpoint found

With that (hand-compiled kernel)
Linux belu4ka 2.6.38-pf7-nusq #1 SMP PREEMPT Thu May 5 11:27:48 EEST 2011 i686 Mobile AMD Sempron(tm) Processor 3800+ AuthenticAMD GNU/Linux

I got
Aug 19 00:30:13 belu4ka kernel: [  353.140330] gspca: bandwidth not wide enough - trying again
Aug 19 00:30:13 belu4ka kernel: [  353.409184] ohci_hcd 0000:00:0b.0: leak ed f560c400 (#81) state 2

modules:
gspca_sonixb           10863  0 
gspca_main             19539  1 gspca_sonixb

webcams:
idVendor=0c45, idProduct=60b0
idVendor=0c45, idProduct=6029

The problem persists only when I make vlc show me both cams in the same time.
Video from the first camera started still displayed, but the second cam is switched off.

I wonder if somebody tell me if I should bugreport that and where to do that?
I didn`t found anything similar to my problem in linux-usb, linux-video and linux-kernel lists?

//D

ohci_hcd leak ed

[DiP]

I found where it is.

It seems to appear on my desktop box when I connect cams to the same bus.
Two connectors next to each other.

But on the notebook it does matter. Any bus, any port.

Problem still alive with a kernel 
3.0.2-pf-di #1 SMP PREEMPT Fri Aug 19
00:21:22 EEST 2011 x86_64 Intel(R) Core(TM)2 Duo CPU E8200 @ 2.66GHz
GenuineIntel GNU/Linux

//D

ohci_hcd leak ed

[Jonathan Neuschäfer]

On Fri, Aug 19, 2011 at 12:53:46AM +0300, DiP wrote:
> I found where it is.
> 
> It seems to appear on my desktop box when I connect cams to the same bus.
> Two connectors next to each other.
> 
> But on the notebook it does matter. Any bus, any port.
> 
> Problem still alive with a kernel 
> 3.0.2-pf-di #1 SMP PREEMPT Fri Aug 19
> 00:21:22 EEST 2011 x86_64 Intel(R) Core(TM)2 Duo CPU E8200 @ 2.66GHz
> GenuineIntel GNU/Linux

Sounds like a problem worth reporting on the relevant mailing lists. If
you have the time and patience, bisecting the (possible) bug may be useful
(see http://www.kernel.org/pub/software/scm/git/docs/git-bisect.html).

HTH,
	Jonathan Neusch?fer

ohci_hcd leak ed

[DiP]

On Fri, 19 Aug 2011 21:42:09 +0200
Jonathan Neusch?fer <j.neuschaefer@gmx.net> wrote:

> On Fri, Aug 19, 2011 at 12:53:46AM +0300, DiP wrote:
> > I found where it is.
> > 
> > It seems to appear on my desktop box when I connect cams to the same bus.
> > Two connectors next to each other.
> > 
> > But on the notebook it does matter. Any bus, any port.
> > 
> > Problem still alive with a kernel 
> > 3.0.2-pf-di #1 SMP PREEMPT Fri Aug 19
> > 00:21:22 EEST 2011 x86_64 Intel(R) Core(TM)2 Duo CPU E8200 @ 2.66GHz
> > GenuineIntel GNU/Linux
> 
> Sounds like a problem worth reporting on the relevant mailing lists. If
> you have the time and patience, bisecting the (possible) bug may be useful
> (see http://www.kernel.org/pub/software/scm/git/docs/git-bisect.html).
> 

Sorry, can not get what to deal with that bisect. I wonder if you could explain
me what to do, or point me to the more explanatory articles/wiki etc?

> HTH,
> 	Jonathan Neusch?fer
> 

ohci_hcd leak ed

[DiP]

On Fri, 19 Aug 2011 21:42:09 +0200
Jonathan Neusch?fer <j.neuschaefer@gmx.net> wrote:

> On Fri, Aug 19, 2011 at 12:53:46AM +0300, DiP wrote:
> > I found where it is.
> > 
> > It seems to appear on my desktop box when I connect cams to the same bus.
> > Two connectors next to each other.
> > 
> > But on the notebook it does matter. Any bus, any port.
> > 
> > Problem still alive with a kernel 
> > 3.0.2-pf-di #1 SMP PREEMPT Fri Aug 19
> > 00:21:22 EEST 2011 x86_64 Intel(R) Core(TM)2 Duo CPU E8200 @ 2.66GHz
> > GenuineIntel GNU/Linux
> 
> Sounds like a problem worth reporting on the relevant mailing lists. If
> you have the time and patience, bisecting the (possible) bug may be useful
> (see http://www.kernel.org/pub/software/scm/git/docs/git-bisect.html).
> 
I can drive trought versions, but maybe it`s too hard for me to understand
kernel driver code - that`s what I mean. Also I don`t understand how must look
the correct one ))

Is there some technology to get the prix with no experience with kernel C
programming?

//D

ohci_hcd leak ed

[Jonathan Neuschäfer]

On Fri, Aug 19, 2011 at 11:16:18PM +0300, DiP wrote:
> On Fri, 19 Aug 2011 21:42:09 +0200
> Jonathan Neusch?fer <j.neuschaefer@gmx.net> wrote:
> 
> > On Fri, Aug 19, 2011 at 12:53:46AM +0300, DiP wrote:
> > > I found where it is.
> > > 
> > > It seems to appear on my desktop box when I connect cams to the same bus.
> > > Two connectors next to each other.
> > > 
> > > But on the notebook it does matter. Any bus, any port.
> > > 
> > > Problem still alive with a kernel 
> > > 3.0.2-pf-di #1 SMP PREEMPT Fri Aug 19
> > > 00:21:22 EEST 2011 x86_64 Intel(R) Core(TM)2 Duo CPU E8200 @ 2.66GHz
> > > GenuineIntel GNU/Linux
> > 
> > Sounds like a problem worth reporting on the relevant mailing lists. If
> > you have the time and patience, bisecting the (possible) bug may be useful
> > (see http://www.kernel.org/pub/software/scm/git/docs/git-bisect.html).
> > 
> 
> Sorry, can not get what to deal with that bisect. I wonder if you could explain
> me what to do, or point me to the more explanatory articles/wiki etc?

Git bisect is actually quite easy to use. First you have to find a "good"
version (i.e. git commit). A version is good if the bug isn't there.
Then you start the bisection:
$ git bisect start bad-version good-version

Git bisect will check out a version for you to test. Now test this
version and tell git bisect whether you experience the bug or not.
$ git bisect good  # if the bug is there
$ git bisect bad   # if it's not there

Repeat this precess until git bisect good/bad prints a message of the form:

    02f8c6aee8df3cdc935e9bdd4f2d020306035dbe is the first bad commit.
    commit 02f8c6aee8df3cdc935e9bdd4f2d020306035dbe
    Author: Linus Torvalds <torvalds@linux-foundation.org>
    Date:   Thu Jul 21 19:17:23 2011 -0700

        Linux 3.0

    :100644 100644 60d91f7... 6a5bdad... M  Makefile


Hope this helps,
	Jonathan Neusch?fer