On Wed, Sep 21, 2011 at 7:56 AM, Thomas Petazzoni
<thomas.petazzoni@xxxxxxxxxxxxxxxxxx> wrote:
> Hello,
>
> On an x86 platform, we have managed to use a Hauppauge USB Live 2
> capture device with the cx231xx on a 3.0 kernel with the patch at [1].
> Things work nicely.
>
> However, using a similar 3.0 kernel with the exact same device on an
> ARM platform (BeagleBoard-XM), starting a V4L application to capture
> the video immediately triggers the following BUG_ON in
> arch/arm/mm/dma-mapping.c:
>
> 429 void ___dma_single_cpu_to_dev(const void *kaddr, size_t size,
> 430 enum dma_data_direction dir)
> 431 {
> 432 unsigned long paddr;
> 433
> 434 BUG_ON(!virt_addr_valid(kaddr) || !virt_addr_valid(kaddr + size - 1));
>
> This problem looks similar to the problem fixed on the gspca driver by:
>
> commit 882787ff8fdeb0be790547ee9b22b281095e95da
> Author: Jason Wang <jason77.wang@xxxxxxxxx>
> Date: Fri Sep 3 06:57:19 2010 -0300
>
> V4L/DVB: gspca - main: Fix a crash of some webcams on ARM arch
>
> When plugging some webcams on ARM, the system crashes.
> This is because we alloc buffer for an urb through usb_buffer_alloc,
> the alloced buffer is already in DMA coherent region, so we should
> set the flag of this urb to URB_NO_TRANSFER_DMA_MAP, otherwise when
> we submit this urb, the hcd core will handle this address as an
> non-DMA address and call dma_map_single/sg to map it. On arm
> architecture, dma_map_single a DMA coherent address will be catched
> by a BUG_ON().
>
> Signed-off-by: Jason Wang <jason77.wang@xxxxxxxxx>
> Signed-off-by: Jean-François Moine <moinejf@xxxxxxx>
> Cc: stable@xxxxxxxxxx
> Signed-off-by: Mauro Carvalho Chehab <mchehab@xxxxxxxxxx>
>
> I guess the cx231xx driver is trying to DMA-map buffers whose location
> is not appropriate for DMA-mapping, because they are already in an DMA
> coherent region. Is the fix just to add the same
> URB_NO_TRANSFER_DMA_MAP to the urb->transfer_flags ? Or is it something
> completely different ?
Hi Thomas,
I ran into the same issue on em28xx in the past (which is what those
parts of cx231xx are based on). Yes, just adding
URB_NO_TRANSFER_DMA_MAP should result in it starting to work. Please
try that out, and assuming it works feel free to submit a patch which
can be included upstream.
Regards,
Devin
--
Devin J. Heitmueller - Kernel Labs
http://www.kernellabs.com
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
