Re: [RFC PATCH v0] Add tw5864 driver

[Hans Verkuil]

On 03/11/2016 09:40 AM, Andrey Utkin wrote:
> On Fri, 11 Mar 2016 09:00:18 +0100
> Hans Verkuil <hverkuil@xxxxxxxxx> wrote:
>> The reason is likely to be the tw5864_queue_setup function which has
>> not been updated to handle CREATE_BUFS support correctly. It should
>> look like this:
>>
>> static int tw5864_queue_setup(struct vb2_queue *q,
>> 			      unsigned int *num_buffers,
>> 			      unsigned int *num_planes, unsigned int
>> sizes[], void *alloc_ctxs[])
>> {
>> 	struct tw5864_input *dev = vb2_get_drv_priv(q);
>>
>> 	if (q->num_buffers + *num_buffers < 12)
>> 		*num_buffers = 12 - q->num_buffers;
>>
>> 	alloc_ctxs[0] = dev->alloc_ctx;
>> 	if (*num_planes)
>> 		return sizes[0] < H264_VLC_BUF_SIZE ? -EINVAL : 0;
>>
>> 	sizes[0] = H264_VLC_BUF_SIZE;
>> 	*num_planes = 1;
>>
>> 	return 0;
>> }
> 
> Thanks for suggestion, but now the failure looks this way:
> 
> Streaming ioctls:
>         test read/write: OK
>                 fail: v4l2-test-buffers.cpp(297): g_field() == V4L2_FIELD_ANY

While userspace may specify FIELD_ANY when setting a format, the driver should
always map that to a specific field setting and should never return FIELD_ANY
back to userspace.

In this case, the 'field' field of the v4l2_buffer struct has FIELD_ANY which
means it is not set correctly (or at all) in the driver.

It's a common mistake, which is why v4l2-compliance tests for it :-)

Regards,

	Hans

>                 fail: v4l2-test-buffers.cpp(703): buf.check(q, last_seq)
>                 fail: v4l2-test-buffers.cpp(976): captureBufs(node, q, m2m_q, frame_count, false)
>         test MMAP: FAIL
> 
> Will check that later. If you have any suggestions, I would be very
> grateful.
> --
> 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
>