Tomi,
On 10/3/22 7:32 PM, Tomi Valkeinen wrote:
> Hi,
>
> On 29/09/2022 09:48, Bingbu Cao wrote:
>>
>>
>> On 8/31/22 10:13 PM, Tomi Valkeinen wrote:
>>> V4L2_FRAME_DESC_ENTRY_MAX is currently set to 4. In theory it's possible
>>> to have an arbitrary amount of streams in a single pad, so preferably
>>> there should be no hardcoded maximum number.
>>>
>>> However, I believe a reasonable max is 8, which would cover a CSI-2 pad
>>> with 4 streams of pixel data and 4 streams of metadata.
>>>
>>> Signed-off-by: Tomi Valkeinen <tomi.valkeinen@xxxxxxxxxxxxxxxx>
>>> Reviewed-by: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx>
>>> Reviewed-by: Hans Verkuil <hverkuil-cisco@xxxxxxxxx>
>>> Reviewed-by: Jacopo Mondi <jacopo+renesas@xxxxxxxxxx>
>>> ---
>>> include/media/v4l2-subdev.h | 6 +++++-
>>> 1 file changed, 5 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/include/media/v4l2-subdev.h b/include/media/v4l2-subdev.h
>>> index 9689f38a0af1..3797b99bb408 100644
>>> --- a/include/media/v4l2-subdev.h
>>> +++ b/include/media/v4l2-subdev.h
>>> @@ -358,7 +358,11 @@ struct v4l2_mbus_frame_desc_entry {
>>> } bus;
>>> };
>>> -#define V4L2_FRAME_DESC_ENTRY_MAX 4
>>> + /*
>>> + * If this number is too small, it should be dropped altogether and the
>>> + * API switched to a dynamic number of frame descriptor entries.
>>> + */
>>> +#define V4L2_FRAME_DESC_ENTRY_MAX 8
>>
>> The number 8 here is still not enough I think, CSI2 specification already
>> extended the VC identifier to be at most 5 bits, which support a max of
>> 32 VCs.
>
> Well, 8 frame desc entries is not enough for the "old" CSI2 either as there can be a lot of data-types, each a separate stream ("stream" as defined by this series) which needs a frame desc.
>
>> Considering the metadata, the number should be larger, it looks like that
>> we have to switch using dynamic number?
>
> Do we have a current use case which needs more than 8 streams per pad? If not, I'd stay away from this for the time being. This can be changed later.
Yes, we have a use case which run at maximum 6 streams + 4 meta-data.
>
>> BTW, does this change break the uAPI?
>
> No, the frame desc is fully internal to the kernel.
Got it, thanks.
>
> Tomi
--
Best regards,
Bingbu Cao
