Re: Recent patch for videobuf causing a crash to my driver

[Mauro Carvalho Chehab <mchehab@redhat.com>]

Em 22-06-2012 05:50, Laurent Pinchart escreveu:
> Hi Hans,
> 
> On Friday 22 June 2012 09:50:44 Hans Verkuil wrote:
>> On 22/06/12 05:39, Prabhakar Lad wrote:
>>> Hi Federico,
>>>
>>> Recent patch from you (commit id a8f3c203e19b702fa5e8e83a9b6fb3c5a6d1cce4)
>>> which added cached buffer support to videobuf dma contig, is causing my
>>> driver to crash.
>>> Has this patch being tested for 'uncached' buffers ? If I replace this
>>> mapping logic with remap_pfn_range() my driver works without any crash.
>>>
>>> Or is that I am missing somewhere ?
>>
>> No, I had the same problem this week with vpif_capture. Since I was running
>> an unusual setup (a 3.0 kernel with the media subsystem patched to 3.5-rc1)
>> I didn't know whether it was caused by a mismatch between 3.0 and a 3.5
>> media subsystem.
>>
>> I intended to investigate this next week, but now it is clear that it is
>> this patch that is causing the problem.
> 
> Time to port the driver to videobuf2 ? ;-)

The regression needs to be fixed anyway, and send to stable. A patch that converts
it to VB2 won't met stable requirements.

Regards,
Mauro