On 11/16/19 12:55 AM, Niklas Söderlund wrote:
> Add a video device capability flag to indicate that its inputs and/or
> outputs are controlled by the Media Controller instead of the V4L2 API.
> When this flag is set, ioctl for enum inputs and outputs are
> automatically enabled and programmed to call a helper function.
>
> Signed-off-by: Helen Koike <helen.koike@xxxxxxxxxxxxx>
> Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@xxxxxxxxxxxx>
> ---
> .../media/uapi/v4l/vidioc-querycap.rst | 3 ++
> .../media/videodev2.h.rst.exceptions | 1 +
> drivers/media/v4l2-core/v4l2-dev.c | 10 ++++--
> drivers/media/v4l2-core/v4l2-ioctl.c | 36 +++++++++++++++++++
> include/uapi/linux/videodev2.h | 2 ++
> 5 files changed, 50 insertions(+), 2 deletions(-)
>
> diff --git a/Documentation/media/uapi/v4l/vidioc-querycap.rst b/Documentation/media/uapi/v4l/vidioc-querycap.rst
> index 5f9930195d624c73..8b621ecb906afe96 100644
> --- a/Documentation/media/uapi/v4l/vidioc-querycap.rst
> +++ b/Documentation/media/uapi/v4l/vidioc-querycap.rst
> @@ -264,6 +264,9 @@ specification the ioctl returns an ``EINVAL`` error code.
> * - ``V4L2_CAP_TOUCH``
> - 0x10000000
> - This is a touch device.
> + * - ``V4L2_CAP_IO_MC``
> + - 0x20000000
> + - The inputs and/or outputs of this device are controlled by the Media Controller see :ref:`media_controller`.
'see' should start a new sentence. So: ... Controller. See ...
I also think this needs a more detailed description. At minimum this should explain that there is only one
input and/or output as seen from userspace and it depends on the topology from/to which actual video source the
video device captures/outputs its video.
And be very precise when you are talking about inputs/outputs: are those the inputs/outputs as seen from
userspace, or the inputs/outputs to the video I/O entity?
> * - ``V4L2_CAP_DEVICE_CAPS``
> - 0x80000000
> - The driver fills the ``device_caps`` field. This capability can
> diff --git a/Documentation/media/videodev2.h.rst.exceptions b/Documentation/media/videodev2.h.rst.exceptions
> index cb6ccf91776e6b56..a625fb90e3a989a7 100644
> --- a/Documentation/media/videodev2.h.rst.exceptions
> +++ b/Documentation/media/videodev2.h.rst.exceptions
> @@ -176,6 +176,7 @@ replace define V4L2_CAP_STREAMING device-capabilities
> replace define V4L2_CAP_META_OUTPUT device-capabilities
> replace define V4L2_CAP_DEVICE_CAPS device-capabilities
> replace define V4L2_CAP_TOUCH device-capabilities
> +replace define V4L2_CAP_IO_MC device-capabilities
>
> # V4L2 pix flags
> replace define V4L2_PIX_FMT_PRIV_MAGIC :c:type:`v4l2_pix_format`
> diff --git a/drivers/media/v4l2-core/v4l2-dev.c b/drivers/media/v4l2-core/v4l2-dev.c
> index 4293df8d664f70b3..f0f5fcaa618c7be8 100644
> --- a/drivers/media/v4l2-core/v4l2-dev.c
> +++ b/drivers/media/v4l2-core/v4l2-dev.c
> @@ -725,7 +725,6 @@ static void determine_valid_ioctls(struct video_device *vdev)
> SET_VALID_IOCTL(ops, VIDIOC_G_STD, vidioc_g_std);
> if (is_rx) {
> SET_VALID_IOCTL(ops, VIDIOC_QUERYSTD, vidioc_querystd);
> - SET_VALID_IOCTL(ops, VIDIOC_ENUMINPUT, vidioc_enum_input);
> set_bit(_IOC_NR(VIDIOC_G_INPUT), valid_ioctls);
> set_bit(_IOC_NR(VIDIOC_S_INPUT), valid_ioctls);
> SET_VALID_IOCTL(ops, VIDIOC_ENUMAUDIO, vidioc_enumaudio);
> @@ -733,14 +732,21 @@ static void determine_valid_ioctls(struct video_device *vdev)
> SET_VALID_IOCTL(ops, VIDIOC_S_AUDIO, vidioc_s_audio);
> SET_VALID_IOCTL(ops, VIDIOC_QUERY_DV_TIMINGS, vidioc_query_dv_timings);
> SET_VALID_IOCTL(ops, VIDIOC_S_EDID, vidioc_s_edid);
> + if (vdev->device_caps & V4L2_CAP_IO_MC)
Add an is_io_mc bool at the top and use that.
> + set_bit(_IOC_NR(VIDIOC_ENUMINPUT), valid_ioctls);
> + else
> + SET_VALID_IOCTL(ops, VIDIOC_ENUMINPUT, vidioc_enum_input);
Ah, now I see why you did things the way you did in patch 1.
This is confusing in this series: for once I think it is better if patches 1 and 4 are
combined in a single patch.
In any case, this means that the G/S_IN/OUTPUT ioctls should only be valid if:
- CAP_IO_MC is set
- or the vidioc_g/s_in/output op is set by the driver
- or the vidioc_enum_in/output op is set by the driver
I've been wondering: now that we know what the device caps are we can do a much
better job of validating if the driver supports the right ioctls.
We've been relying on v4l2-compliance for that, but part of the v4l2-compliance
validation can be done here instead, issuing a WARN_ON if the driver forgets to
implement something that it should have implemented.
Something for a future patch, though.
> }
> if (is_tx) {
> - SET_VALID_IOCTL(ops, VIDIOC_ENUMOUTPUT, vidioc_enum_output);
> set_bit(_IOC_NR(VIDIOC_G_OUTPUT), valid_ioctls);
> set_bit(_IOC_NR(VIDIOC_S_OUTPUT), valid_ioctls);
> SET_VALID_IOCTL(ops, VIDIOC_ENUMAUDOUT, vidioc_enumaudout);
> SET_VALID_IOCTL(ops, VIDIOC_G_AUDOUT, vidioc_g_audout);
> SET_VALID_IOCTL(ops, VIDIOC_S_AUDOUT, vidioc_s_audout);
> + if (vdev->device_caps & V4L2_CAP_IO_MC)
> + set_bit(_IOC_NR(VIDIOC_ENUMOUTPUT), valid_ioctls);
> + else
> + SET_VALID_IOCTL(ops, VIDIOC_ENUMOUTPUT, vidioc_enum_output);
> }
> if (ops->vidioc_g_parm || ops->vidioc_g_std)
> set_bit(_IOC_NR(VIDIOC_G_PARM), valid_ioctls);
> diff --git a/drivers/media/v4l2-core/v4l2-ioctl.c b/drivers/media/v4l2-core/v4l2-ioctl.c
> index 4a461de28677c5a8..a1c048c8fe6eff2f 100644
> --- a/drivers/media/v4l2-core/v4l2-ioctl.c
> +++ b/drivers/media/v4l2-core/v4l2-ioctl.c
> @@ -1085,6 +1085,36 @@ static int v4l_querycap(const struct v4l2_ioctl_ops *ops,
> return ret;
> }
>
> +static int v4l2_ioctl_enum_input_mc(struct file *file, void *priv,
> + struct v4l2_input *i)
> +{
> + struct video_device *vfd = video_devdata(file);
> +
> + if (i->index > 0)
Just say: if (i->index)
> + return -EINVAL;
> +
> + memset(i, 0, sizeof(*i));
Not needed, everything after index is already zeroed.
> + strlcpy(i->name, vfd->name, sizeof(i->name));
strscpy
> + i->type = V4L2_INPUT_TYPE_CAMERA;
> +
> + return 0;
> +}
> +
> +static int v4l2_ioctl_enum_output_mc(struct file *file, void *priv,
> + struct v4l2_output *o)
> +{
> + struct video_device *vfd = video_devdata(file);
> +
> + if (o->index > 0)
> + return -EINVAL;
> +
> + memset(o, 0, sizeof(*o));
> + strlcpy(o->name, vfd->name, sizeof(o->name));
Same comments as for v4l2_ioctl_enum_input_mc.
> + o->type = V4L2_OUTPUT_TYPE_ANALOG;
> +
> + return 0;
> +}
> +
> static int v4l2_ioctl_g_single_input(struct file *file, void *priv, unsigned int *i)
> {
> *i = 0;
> @@ -1181,6 +1211,9 @@ static int v4l_enuminput(const struct v4l2_ioctl_ops *ops,
> if (is_valid_ioctl(vfd, VIDIOC_S_STD))
> p->capabilities |= V4L2_IN_CAP_STD;
>
> + if (vfd->device_caps & V4L2_CAP_IO_MC)
> + return v4l2_ioctl_enum_input_mc(file, fh, p);
I think you can incorporate the code of v4l2_ioctl_enum_input_mc
here instead of creating a separate function. It's a matter of taste, though.
> +
> return ops->vidioc_enum_input(file, fh, p);
> }
>
> @@ -1199,6 +1232,9 @@ static int v4l_enumoutput(const struct v4l2_ioctl_ops *ops,
> if (is_valid_ioctl(vfd, VIDIOC_S_STD))
> p->capabilities |= V4L2_OUT_CAP_STD;
>
> + if (vfd->device_caps & V4L2_CAP_IO_MC)
> + return v4l2_ioctl_enum_output_mc(file, fh, p);
If you DO keep separate functions, then it is better to move v4l2_ioctl_enum_output_mc
to just before this function, so the two are close together.
> +
> return ops->vidioc_enum_output(file, fh, p);
> }
>
> diff --git a/include/uapi/linux/videodev2.h b/include/uapi/linux/videodev2.h
> index 04481c717fee75c4..c97b55bc19eb892a 100644
> --- a/include/uapi/linux/videodev2.h
> +++ b/include/uapi/linux/videodev2.h
> @@ -487,6 +487,8 @@ struct v4l2_capability {
>
> #define V4L2_CAP_TOUCH 0x10000000 /* Is a touch device */
>
> +#define V4L2_CAP_IO_MC 0x20000000 /* Is input/output controlled by the media controller */
> +
> #define V4L2_CAP_DEVICE_CAPS 0x80000000 /* sets device capabilities field */
>
> /*
>
Regards,
Hans
