SPI related oops

SPI related oops

[Arnaud Patard]

Hi,

Since some time ago I was seeing a reboot as long as my n770 was
booting recent kernels from the omap git. Now, I found out that the
crash is due to a oops as soon as udev plays with the uevent file of the
spi master. 
After some debugging, I found that the following commit is the culprit : 
http://source.mvista.com/git/gitweb.cgi?p=linux-omap-2.6.git;a=commit;h=3bd0f6943520e459659d10f3282285e43d3990f1

Reverting it allows my kernel to boot but I think it's not a real
fix. Has someone seen a similar issue and has a fix for it ?

[ Also, I'm wondering why the spi_busnum_to_master() function exists
because grepping the current omap git tree gives 0 user... ]

Regards,
Arnaud

Re: SPI related oops

[David Brownell]

On Thursday 18 January 2007 11:44 am, Arnaud Patard wrote:
> Hi,
> 
> Since some time ago I was seeing a reboot as long as my n770 was
> booting recent kernels from the omap git. Now, I found out that the
> crash is due to a oops as soon as udev plays with the uevent file of the
> spi master. 
> After some debugging, I found that the following commit is the culprit : 
> http://source.mvista.com/git/gitweb.cgi?p=linux-omap-2.6.git;a=commit;h=3bd0f6943520e459659d10f3282285e43d3990f1
> 
> Reverting it allows my kernel to boot but I think it's not a real
> fix. Has someone seen a similar issue and has a fix for it ?

Yeah, that patch shouldn't have been merged.  Someone posted a
working patch on LKML not long ago, but I'd prefer a simpler fix.

 
> [ Also, I'm wondering why the spi_busnum_to_master() function exists
> because grepping the current omap git tree gives 0 user... ]

It exists to support non-integrated SPI adapters.  Consider a PCI
plug-in card with a couple SPI interfaces, or one accessed through
a USB link to a microcontroller.

- Dave

Re: SPI related oops

[Arnaud Patard]

David Brownell <david-b@pacbell.net> writes:

> On Thursday 18 January 2007 11:44 am, Arnaud Patard wrote:
>> Hi,
>> 
>> Since some time ago I was seeing a reboot as long as my n770 was
>> booting recent kernels from the omap git. Now, I found out that the
>> crash is due to a oops as soon as udev plays with the uevent file of the
>> spi master. 
>> After some debugging, I found that the following commit is the culprit : 
>> http://source.mvista.com/git/gitweb.cgi?p=linux-omap-2.6.git;a=commit;h=3bd0f6943520e459659d10f3282285e43d3990f1
>> 
>> Reverting it allows my kernel to boot but I think it's not a real
>> fix. Has someone seen a similar issue and has a fix for it ?
>
> Yeah, that patch shouldn't have been merged.  Someone posted a
> working patch on LKML not long ago, but I'd prefer a simpler fix.

Ok. Happy to see that it's not a local issue.

>
>  
>> [ Also, I'm wondering why the spi_busnum_to_master() function exists
>> because grepping the current omap git tree gives 0 user... ]
>
> It exists to support non-integrated SPI adapters.  Consider a PCI
> plug-in card with a couple SPI interfaces, or one accessed through
> a USB link to a microcontroller.

Thanks for the explanation

Arnaud