Hi,

On Wed, Dec 14, 2011 at 11:34 PM, Sakari Ailus <[email protected]> wrote:

>> +     case VIDIOC_G_FD_RESULT:
>> +     {
>> +             struct v4l2_fd_result *fr = arg;
>> +
>> +             if (!ops->vidioc_g_fd_result)
>> +                     break;
>> +
>> +             ret = ops->vidioc_g_fd_result(file, fh, fr);
>> +
>> +             dbgarg(cmd, "index=%d", fr->buf_index);
>> +             break;
>> +     }
>> +     case VIDIOC_G_FD_COUNT:
>> +     {
>> +             struct v4l2_fd_count *fc = arg;
>> +
>> +             if (!ops->vidioc_g_fd_count)
>> +                     break;
>> +
>> +             ret = ops->vidioc_g_fd_count(file, fh, fc);
>> +
>> +             dbgarg(cmd, "index=%d", fc->buf_index);
>> +             break;
>> +     }
>
> The patch description tells these ioctls may be called between... what? I'd

In fact, these ioctls should be called after return from poll.

> think such information could be better provided as events.

Yes, I still think so, but the length of returned data is variant, also
event has the 64 byte length's limitation.

> How is face detection enabled or disabled?

Currently, streaming on will trigger detection enabling, and streaming off
will trigger detection disabling.

>
> Could you detect other objects than faces?

The -v2 has been extended to detect objects, not limited to faces.

> Would events be large enough to deliver you the necessary data? We could

Looks like event(64 bytes) is not large enough to deliver the data.

> also consider delivering the information as a data structure on a separate
> plane.

Could you let me know how to do it?

>> +/**
>> + * struct v4l2_fd_result - VIDIOC_G_FD_RESULT argument
>> + * @buf_index:       entry, index of v4l2_buffer for face detection
>> + * @face_cnt:        return, how many faces detected from the @buf_index
>> + * @fd:              return, result of faces' detection
>> + */
>> +struct v4l2_fd_result {
>> +     __u32   buf_index;
>> +     __u32   face_cnt;
>> +     __u32   reserved[6];
>> +     struct v4l2_fd_detection *fd;
>
> Aligning structure sizes to a power of two is considered to be a good
> practice.

There are many changes on these structure in -v2.


thanks,
--
Ming Lei
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to