Re: [Bug 14564] capture-example sleeping function called from invalid context at arch/x86/mm/fault.c

Re: [Bug 14564] capture-example sleeping function called from invalid context at arch/x86/mm/fault.c

[Sean]

bugzilla-daemon@bugzilla.kernel.org wrote:
> http://bugzilla.kernel.org/show_bug.cgi?id=14564
>
>
> Jean-Francois Moine <moinejf@free.fr> changed:
>
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>                  CC|                            |moinejf@free.fr
>
>
>
>
> --- Comment #22 from Jean-Francois Moine <moinejf@free.fr>  2010-01-03 07:02:45 ---
> Hello Sean,
>
> Sorry to be a bit late. Looking at the dmesg, I found that the gspca version
> was 2.7.0. May you upgrade your linux media stuff from LinuxTv.org and check if
> this problem still occurs?
>
> Jef
>   
Jef,

I upgraded to the latest v4l-dvb from http://linuxtv.org/hg/v4l-dvb, 
made the kernel modules, made the v4l libraries, and recompiled 
capture-example.c. Gspca now shows 2.8.0. The error still persists. Alan 
Stern's latest patch to ohci-q.c traps the error. I think it is an issue 
with the cpu or usb controller on the Vortex86SX SoC.

Sean

Re: [Bug 14564] capture-example sleeping function called from invalid context at arch/x86/mm/fault.c

[Alan Stern]

On Mon, 4 Jan 2010, Sean wrote:

> Jef,
> 
> I upgraded to the latest v4l-dvb from http://linuxtv.org/hg/v4l-dvb, 
> made the kernel modules, made the v4l libraries, and recompiled 
> capture-example.c. Gspca now shows 2.8.0. The error still persists. Alan 
> Stern's latest patch to ohci-q.c traps the error. I think it is an issue 
> with the cpu or usb controller on the Vortex86SX SoC.

The CPU is more likely than the USB controller.  The erroneous pointer 
values we see are virtual addresses, whereas the USB controller would 
probably write a physical (or DMA) address if it was malfunctioning.

Or it could be some bizarre timing problem with the memory bus, or 
something else equally obscure.  You didn't mention before that this 
was a SoC rather than an ordinary PC.

Alan Stern