* Re: [PATCH 2/2] rcar-csi2: Use V4L2 async helpers to create the notifier
2020-08-07 11:16 ` [PATCH 2/2] rcar-csi2: Use V4L2 async helpers to create the notifier Niklas Söderlund
@ 2020-08-10 9:36 ` Jacopo Mondi
2020-08-11 22:42 ` Laurent Pinchart
2020-08-18 7:51 ` Sakari Ailus
2 siblings, 0 replies; 7+ messages in thread
From: Jacopo Mondi @ 2020-08-10 9:36 UTC (permalink / raw)
To: Niklas Söderlund; +Cc: Sakari Ailus, linux-media, linux-renesas-soc
Hi Niklas,
On Fri, Aug 07, 2020 at 01:16:19PM +0200, Niklas Söderlund wrote:
> The V4L2 async framework helpers now populates the async notifier with
helpers -> populate
> endpoint matching information and there is no need to do this manually
"with endpoint matching information" sounds weird.
What about
The V4L2 async framework helpers have now moved to match async
subdevices matching on endpoints. There is not need anymore to do this
manually...
> in the R-Car CSI-2 driver, switch to using the provided helper.
>
> Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
> ---
> drivers/media/platform/rcar-vin/rcar-csi2.c | 48 +++++----------------
> 1 file changed, 11 insertions(+), 37 deletions(-)
>
> diff --git a/drivers/media/platform/rcar-vin/rcar-csi2.c b/drivers/media/platform/rcar-vin/rcar-csi2.c
> index c6cc4f473a077899..f0067ff21d5d9d33 100644
> --- a/drivers/media/platform/rcar-vin/rcar-csi2.c
> +++ b/drivers/media/platform/rcar-vin/rcar-csi2.c
> @@ -362,7 +362,6 @@ struct rcar_csi2 {
> struct media_pad pads[NR_OF_RCAR_CSI2_PAD];
>
> struct v4l2_async_notifier notifier;
> - struct v4l2_async_subdev asd;
> struct v4l2_subdev *remote;
>
> struct v4l2_mbus_framefmt mf;
> @@ -774,9 +773,11 @@ static const struct v4l2_async_notifier_operations rcar_csi2_notify_ops = {
> .unbind = rcsi2_notify_unbind,
> };
>
> -static int rcsi2_parse_v4l2(struct rcar_csi2 *priv,
> - struct v4l2_fwnode_endpoint *vep)
> +static int rcar_csi2_parse_v4l2(struct device *dev,
> + struct v4l2_fwnode_endpoint *vep,
> + struct v4l2_async_subdev *asd)
> {
> + struct rcar_csi2 *priv = dev_get_drvdata(dev);
> unsigned int i;
>
> /* Only port 0 endpoint 0 is valid. */
> @@ -806,53 +807,26 @@ static int rcsi2_parse_v4l2(struct rcar_csi2 *priv,
> }
> }
>
> + dev_dbg(priv->dev, "Found '%pOF'\n", to_of_node(asd->match.fwnode));
> +
> return 0;
> }
>
> static int rcsi2_parse_dt(struct rcar_csi2 *priv)
> {
> - struct device_node *ep;
> - struct v4l2_fwnode_endpoint v4l2_ep = { .bus_type = 0 };
> int ret;
>
> - ep = of_graph_get_endpoint_by_regs(priv->dev->of_node, 0, 0);
> - if (!ep) {
> - dev_err(priv->dev, "Not connected to subdevice\n");
> - return -EINVAL;
> - }
> -
> - ret = v4l2_fwnode_endpoint_parse(of_fwnode_handle(ep), &v4l2_ep);
> - if (ret) {
> - dev_err(priv->dev, "Could not parse v4l2 endpoint\n");
> - of_node_put(ep);
> - return -EINVAL;
> - }
> -
> - ret = rcsi2_parse_v4l2(priv, &v4l2_ep);
> - if (ret) {
> - of_node_put(ep);
> - return ret;
> - }
> -
> - priv->asd.match.fwnode =
> - fwnode_graph_get_remote_endpoint(of_fwnode_handle(ep));
> - priv->asd.match_type = V4L2_ASYNC_MATCH_FWNODE;
> -
> - of_node_put(ep);
> -
> v4l2_async_notifier_init(&priv->notifier);
>
> - ret = v4l2_async_notifier_add_subdev(&priv->notifier, &priv->asd);
> - if (ret) {
> - fwnode_handle_put(priv->asd.match.fwnode);
> + ret = v4l2_async_notifier_parse_fwnode_endpoints_by_port(
> + priv->dev, &priv->notifier,
> + sizeof(struct v4l2_async_subdev), 0,
> + rcar_csi2_parse_v4l2);
> + if (ret)
I have not really followed the whole call chain, but I presume
v4l2_async_notifier_parse_fwnode_endpoints_by_port() could fail after
having added the async subdev ? Is it worth calling notifier_cleanup()
if we fail here ?
This apart, the patch looks really nice, finally we're not the black
sheeps anymore!
Reviewed-by: Jacopo Mondi <jacopo+renesas@jmondi.org>
Thanks
j
> return ret;
> - }
>
> priv->notifier.ops = &rcar_csi2_notify_ops;
>
> - dev_dbg(priv->dev, "Found '%pOF'\n",
> - to_of_node(priv->asd.match.fwnode));
> -
> ret = v4l2_async_subdev_notifier_register(&priv->subdev,
> &priv->notifier);
> if (ret)
> --
> 2.28.0
>
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [PATCH 2/2] rcar-csi2: Use V4L2 async helpers to create the notifier
2020-08-07 11:16 ` [PATCH 2/2] rcar-csi2: Use V4L2 async helpers to create the notifier Niklas Söderlund
2020-08-10 9:36 ` Jacopo Mondi
@ 2020-08-11 22:42 ` Laurent Pinchart
2020-08-18 7:51 ` Sakari Ailus
2 siblings, 0 replies; 7+ messages in thread
From: Laurent Pinchart @ 2020-08-11 22:42 UTC (permalink / raw)
To: Niklas Söderlund; +Cc: Sakari Ailus, linux-media, linux-renesas-soc
Hi Niklas,
Thank you for the patch.
On Fri, Aug 07, 2020 at 01:16:19PM +0200, Niklas Söderlund wrote:
> The V4L2 async framework helpers now populates the async notifier with
> endpoint matching information and there is no need to do this manually
> in the R-Car CSI-2 driver, switch to using the provided helper.
>
> Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
> ---
> drivers/media/platform/rcar-vin/rcar-csi2.c | 48 +++++----------------
> 1 file changed, 11 insertions(+), 37 deletions(-)
>
> diff --git a/drivers/media/platform/rcar-vin/rcar-csi2.c b/drivers/media/platform/rcar-vin/rcar-csi2.c
> index c6cc4f473a077899..f0067ff21d5d9d33 100644
> --- a/drivers/media/platform/rcar-vin/rcar-csi2.c
> +++ b/drivers/media/platform/rcar-vin/rcar-csi2.c
> @@ -362,7 +362,6 @@ struct rcar_csi2 {
> struct media_pad pads[NR_OF_RCAR_CSI2_PAD];
>
> struct v4l2_async_notifier notifier;
> - struct v4l2_async_subdev asd;
> struct v4l2_subdev *remote;
>
> struct v4l2_mbus_framefmt mf;
> @@ -774,9 +773,11 @@ static const struct v4l2_async_notifier_operations rcar_csi2_notify_ops = {
> .unbind = rcsi2_notify_unbind,
> };
>
> -static int rcsi2_parse_v4l2(struct rcar_csi2 *priv,
> - struct v4l2_fwnode_endpoint *vep)
> +static int rcar_csi2_parse_v4l2(struct device *dev,
> + struct v4l2_fwnode_endpoint *vep,
> + struct v4l2_async_subdev *asd)
> {
> + struct rcar_csi2 *priv = dev_get_drvdata(dev);
> unsigned int i;
>
> /* Only port 0 endpoint 0 is valid. */
> @@ -806,53 +807,26 @@ static int rcsi2_parse_v4l2(struct rcar_csi2 *priv,
> }
> }
>
> + dev_dbg(priv->dev, "Found '%pOF'\n", to_of_node(asd->match.fwnode));
> +
> return 0;
> }
>
> static int rcsi2_parse_dt(struct rcar_csi2 *priv)
> {
> - struct device_node *ep;
> - struct v4l2_fwnode_endpoint v4l2_ep = { .bus_type = 0 };
> int ret;
>
> - ep = of_graph_get_endpoint_by_regs(priv->dev->of_node, 0, 0);
> - if (!ep) {
> - dev_err(priv->dev, "Not connected to subdevice\n");
> - return -EINVAL;
> - }
> -
> - ret = v4l2_fwnode_endpoint_parse(of_fwnode_handle(ep), &v4l2_ep);
> - if (ret) {
> - dev_err(priv->dev, "Could not parse v4l2 endpoint\n");
> - of_node_put(ep);
> - return -EINVAL;
> - }
> -
> - ret = rcsi2_parse_v4l2(priv, &v4l2_ep);
> - if (ret) {
> - of_node_put(ep);
> - return ret;
> - }
> -
> - priv->asd.match.fwnode =
> - fwnode_graph_get_remote_endpoint(of_fwnode_handle(ep));
> - priv->asd.match_type = V4L2_ASYNC_MATCH_FWNODE;
> -
> - of_node_put(ep);
> -
> v4l2_async_notifier_init(&priv->notifier);
>
> - ret = v4l2_async_notifier_add_subdev(&priv->notifier, &priv->asd);
> - if (ret) {
> - fwnode_handle_put(priv->asd.match.fwnode);
> + ret = v4l2_async_notifier_parse_fwnode_endpoints_by_port(
> + priv->dev, &priv->notifier,
> + sizeof(struct v4l2_async_subdev), 0,
> + rcar_csi2_parse_v4l2);
It seems a bit expensive to go through a complex logic of iterating over
all endpoints when rcar_csi2_parse_v4l2() will only accept port 0,
endpoint 0, but there's a clear benefit from not having to maintain this
code in the driver.
With Jacopo's comments addressed,
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
If anyone has time to spend on it, I think
__v4l2_async_notifier_parse_fwnode_ep() should be duplicated in
v4l2_async_notifier_parse_fwnode_endpoints() and
v4l2_async_notifier_parse_fwnode_endpoints_by_port(), with the latter
only iterating over the endpoints of the requested port.
I also wonder if the !is_available check needs to be inside the loop:
fwnode_graph_for_each_endpoint(dev_fwnode(dev), fwnode) {
struct fwnode_handle *dev_fwnode;
bool is_available;
dev_fwnode = fwnode_graph_get_port_parent(fwnode);
is_available = fwnode_device_is_available(dev_fwnode);
fwnode_handle_put(dev_fwnode);
if (!is_available)
continue;
...
}
Isn't dev_fwnode == dev_fwnode(dev) and thus constant ?
> + if (ret)
> return ret;
> - }
>
> priv->notifier.ops = &rcar_csi2_notify_ops;
>
> - dev_dbg(priv->dev, "Found '%pOF'\n",
> - to_of_node(priv->asd.match.fwnode));
> -
> ret = v4l2_async_subdev_notifier_register(&priv->subdev,
> &priv->notifier);
> if (ret)
--
Regards,
Laurent Pinchart
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [PATCH 2/2] rcar-csi2: Use V4L2 async helpers to create the notifier
2020-08-07 11:16 ` [PATCH 2/2] rcar-csi2: Use V4L2 async helpers to create the notifier Niklas Söderlund
2020-08-10 9:36 ` Jacopo Mondi
2020-08-11 22:42 ` Laurent Pinchart
@ 2020-08-18 7:51 ` Sakari Ailus
2 siblings, 0 replies; 7+ messages in thread
From: Sakari Ailus @ 2020-08-18 7:51 UTC (permalink / raw)
To: Niklas Söderlund; +Cc: linux-media, linux-renesas-soc
Hi Niklas,
On Fri, Aug 07, 2020 at 01:16:19PM +0200, Niklas Söderlund wrote:
> The V4L2 async framework helpers now populates the async notifier with
> endpoint matching information and there is no need to do this manually
> in the R-Car CSI-2 driver, switch to using the provided helper.
>
> Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
> ---
> drivers/media/platform/rcar-vin/rcar-csi2.c | 48 +++++----------------
> 1 file changed, 11 insertions(+), 37 deletions(-)
>
> diff --git a/drivers/media/platform/rcar-vin/rcar-csi2.c b/drivers/media/platform/rcar-vin/rcar-csi2.c
> index c6cc4f473a077899..f0067ff21d5d9d33 100644
> --- a/drivers/media/platform/rcar-vin/rcar-csi2.c
> +++ b/drivers/media/platform/rcar-vin/rcar-csi2.c
> @@ -362,7 +362,6 @@ struct rcar_csi2 {
> struct media_pad pads[NR_OF_RCAR_CSI2_PAD];
>
> struct v4l2_async_notifier notifier;
> - struct v4l2_async_subdev asd;
> struct v4l2_subdev *remote;
>
> struct v4l2_mbus_framefmt mf;
> @@ -774,9 +773,11 @@ static const struct v4l2_async_notifier_operations rcar_csi2_notify_ops = {
> .unbind = rcsi2_notify_unbind,
> };
>
> -static int rcsi2_parse_v4l2(struct rcar_csi2 *priv,
> - struct v4l2_fwnode_endpoint *vep)
> +static int rcar_csi2_parse_v4l2(struct device *dev,
> + struct v4l2_fwnode_endpoint *vep,
> + struct v4l2_async_subdev *asd)
> {
> + struct rcar_csi2 *priv = dev_get_drvdata(dev);
> unsigned int i;
>
> /* Only port 0 endpoint 0 is valid. */
> @@ -806,53 +807,26 @@ static int rcsi2_parse_v4l2(struct rcar_csi2 *priv,
> }
> }
>
> + dev_dbg(priv->dev, "Found '%pOF'\n", to_of_node(asd->match.fwnode));
> +
> return 0;
> }
>
> static int rcsi2_parse_dt(struct rcar_csi2 *priv)
> {
> - struct device_node *ep;
> - struct v4l2_fwnode_endpoint v4l2_ep = { .bus_type = 0 };
> int ret;
>
> - ep = of_graph_get_endpoint_by_regs(priv->dev->of_node, 0, 0);
> - if (!ep) {
> - dev_err(priv->dev, "Not connected to subdevice\n");
> - return -EINVAL;
> - }
> -
> - ret = v4l2_fwnode_endpoint_parse(of_fwnode_handle(ep), &v4l2_ep);
> - if (ret) {
> - dev_err(priv->dev, "Could not parse v4l2 endpoint\n");
> - of_node_put(ep);
> - return -EINVAL;
> - }
> -
> - ret = rcsi2_parse_v4l2(priv, &v4l2_ep);
> - if (ret) {
> - of_node_put(ep);
> - return ret;
> - }
> -
> - priv->asd.match.fwnode =
> - fwnode_graph_get_remote_endpoint(of_fwnode_handle(ep));
> - priv->asd.match_type = V4L2_ASYNC_MATCH_FWNODE;
> -
> - of_node_put(ep);
> -
> v4l2_async_notifier_init(&priv->notifier);
>
> - ret = v4l2_async_notifier_add_subdev(&priv->notifier, &priv->asd);
> - if (ret) {
> - fwnode_handle_put(priv->asd.match.fwnode);
> + ret = v4l2_async_notifier_parse_fwnode_endpoints_by_port(
> + priv->dev, &priv->notifier,
> + sizeof(struct v4l2_async_subdev), 0,
> + rcar_csi2_parse_v4l2);
Please don't. We'd rather like to remove that helper that necessitates the
use of deprecated behaviour. This driver is the last user left actually.
Could you see an example in drivers/media/pci/intel/ipu3/ipu3-cio2.c on how
to make this more condensed?
There's more code left than with that helper as there are multiple things
done here, and a few of these things interact at least in the general case:
- setting binding defaults (missing here),
- parsing the DT as well as storing the configuration you need (should you
not use the V4L2 EP as such),
- getting the remote and
- registration of the subdev in the notifier.
You should also set the bus_type field if you know that in advance --- I
believe you do.
> + if (ret)
> return ret;
> - }
>
> priv->notifier.ops = &rcar_csi2_notify_ops;
>
> - dev_dbg(priv->dev, "Found '%pOF'\n",
> - to_of_node(priv->asd.match.fwnode));
> -
> ret = v4l2_async_subdev_notifier_register(&priv->subdev,
> &priv->notifier);
> if (ret)
--
Kind regards,
Sakari Ailus
^ permalink raw reply [flat|nested] 7+ messages in thread