On Wednesday 12 August 2009 22:31:11 Karicheri, Muralidharan wrote:
> Hans,
>
> Thanks for reviewing this. I have some questions though against your
> comments. Please see below for details..
>
> Murali Karicheri
> Software Design Engineer
> Texas Instruments Inc.
> Germantown, MD 20874
> email: [email protected]
>
> >First of all I've reviewed the other patches in this series and they look
> >OK
> >to me. So you can add
> >Reviewed-by: Hans Verkuil <[email protected]>
> >for patches 1, 2, 3 and 5.
> >
> Will do.
>
> >> + ret = vpif_check_format(ch, &common->fmt.fmt.pix, 0);
> >> + if (ret)
> >> + return ret;
> >> +
> >> + /* Enable streamon on the sub device */
> >> + ret = v4l2_device_call_until_err(&vpif_obj.v4l2_dev,
> >> + vpif_obj.sd[ch->curr_sd_index]->grp_id,
> >> + video, s_stream, 1);
> >> +
> >> + if (ret && (ret != -ENOIOCTLCMD)) {
> >> + vpif_dbg(1, debug, "stream on failed in subdev\n");
> >> + return ret;
> >> + }
> >
> >I strongly recommend that instead of calling the subdev for the current
> >input
> >indirectly using v4l2_device_call_until_err() you use the v4l2_subdev
> >pointer
> >associated with the current input directly. So typically when the input is
> >changed you update a ch->curr_sd pointer to the associated struct
> >v4l2_subdev.
> >
> >After that you can just call v4l2_subdev_call(ch->sd, video, s_stream, 1).
> >Much more concise.
> >
> >In addition the v4l2_device_call_until_err() macro will replace the
> >ENOIOCTLCMD
> >error by 0. The idea behind this macro is that you have an unknown number
> >of
> >subdevs (>= 0) that might be interested in this command, but if no one is,
> >then
> >that's fine too. So in the code above the extra check to see whether the
> >return code was -ENOIOCTLCMD is not needed in the case of
> >v4l2_device_call_until_err(). But it is needed if you switch to
> >v4l2_subdev_call().
> >
> So in short what you are suggesting is to replace all instances of
> v4l2_device_call_until_err() with v4l2_subdev_call() since after input
> selection we know exactly which sub device to direct the application
> request to.
Yes. I think that is the best approach here.
> >> + .fops = &vpif_fops,
> >> + .minor = -1,
> >> + .ioctl_ops = &vpif_ioctl_ops,
> >> + .current_norm = V4L2_STD_625_50,
> >
> >No need to set current_norm since it's overridden by g_std.
> >
> You mean s_std() right?
Hmm, that was a bad sentence of mine. What I meant is that if you define the
g_std function, then the current_norm will not be used by the v4l2-ioctl code.
If you do not give it a g_std function, then the v4l2-ioctl code will return
current_norm instead. So you either have a g_std function and do not setup
current_norm, or you use current_norm and omit g_std.
Personally I think that current_norm handling only confuses people and all
drivers should just make a g_std function.
>
> >Note: I've just found a bug in the default handling of VIDIOC_G_PARM in
> >v4l2-ioctl.c since that uses current_norm even when g_std is defined.
> >I will make a fix for that. As a general remark I have to say that I
> >never liked that v4l2-ioctl has default handling for g_std. It's a
> >dangerous
> >construction that will definitely fail whenever you have both video and vbi
> >device nodes.
> >
> Ok. Understood.
>
> So I will make the next set of patches with the changes suggested by you and
> It would be ready for merge to your tree as well as to v4l-dvb linux-next
> tree (through your pull request to Mauro)
Excellent news!
Regards,
Hans
>
> Thanks and regards,
>
> Murali
>
>
--
Hans Verkuil - video4linux developer - sponsored by TANDBERG Telecom
_______________________________________________
Davinci-linux-open-source mailing list
[email protected]
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
