kern/141696: [rum] [panic] rum(4)+ vimage = kernel panic

[Bjoern A. Zeeb]

On Thu, 17 Dec 2009, Julian Elischer wrote:

Hi Julian,

> Venture37 wrote:
>> The following reply was made to PR kern/141696; it has been noted by GNATS.
>> 
>> From: Venture37 <venture37 at geeklan.co.uk>
>> To: bug-followup at FreeBSD.org, venture37 at geeklan.co.uk
>> Cc:  Subject: Re: kern/141696: [rum] [panic] rum(4)+ vimage = kernel panic
>> Date: Thu, 17 Dec 2009 15:25:28 +0000
>>
>>  Photo of the trace output
>>  http://img64.imageshack.us/i/img1517.jpg/
>
> Marko, et al.
>
> looking at this, the crash is in
>
> ifindex_alloc_locked()
> from:
> if_alloc()
> rum_attach()
> device_attach()
> device_probe_and_attach()
> usb_probe_and_attach_sub()
> usb_probe_and_attach()
> uhub_explore()
> usb_bus_explore()
> usb_process()
>
> the question is "where should we load the current vnet?
>
> is it up to the driver (if_rum) or should device_attach assume that the 
> resulting device MIGHT be a network device.

Different question - why is it a problem for rum/usb whne it's not a
problem for all the other physical interfaces like bge, bce, em, ....
or have we just not noticed what happens if we have them as module and
attach at run-time?

/bz

-- 
Bjoern A. Zeeb         It will not break if you know what you are doing.