[PATCH 0/2] rcar-csi2: Update how DT is traversed and parsed

[PATCH 0/2] rcar-csi2: Update how DT is traversed and parsed

[Niklas Söderlund]

Hi,

This series touches up how the R-Car CSI-2 driver traverse and parse DT 
device nodes. Patch 1/2  switches to only use the fwnode interface and 
2/2 specifies the bus type used for parsing.

This is done a preparation to later modify the rcar-vin driver in a 
similar fashion to be able to drop the use of the 
v4l2_async_notifier_parse_fwnode_endpoints_by_port() helper which I 
understand is marked for deprecation. Therefore comments on the over all 
solution of parsing DT here would be appreciated.

Niklas Söderlund (2):
  rcar-csi2: Switch to using fwnode instead of OF
  rcar-csi2: Set bus type when parsing fwnode

 drivers/media/platform/rcar-vin/rcar-csi2.c | 18 ++++++++++--------
 1 file changed, 10 insertions(+), 8 deletions(-)

-- 
2.28.0

[PATCH 1/2] rcar-csi2: Switch to using fwnode instead of OF

[Niklas Söderlund]

Use the fwnode_graph_get_endpoint_by_id() interface instead of
of_graph_get_endpoint_by_regs() to fetch the fwnode. This saves
translating between a device_node and fwnode_handle.

Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
---
 drivers/media/platform/rcar-vin/rcar-csi2.c | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/drivers/media/platform/rcar-vin/rcar-csi2.c b/drivers/media/platform/rcar-vin/rcar-csi2.c
index 511cd4984777ad99..23e89ef2429d310a 100644
--- a/drivers/media/platform/rcar-vin/rcar-csi2.c
+++ b/drivers/media/platform/rcar-vin/rcar-csi2.c
@@ -873,31 +873,31 @@ static int rcsi2_parse_dt(struct rcar_csi2 *priv)
 {
 	struct v4l2_async_subdev *asd;
 	struct fwnode_handle *fwnode;
-	struct device_node *ep;
+	struct fwnode_handle *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);
+	ep = fwnode_graph_get_endpoint_by_id(dev_fwnode(priv->dev), 0, 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);
+	ret = v4l2_fwnode_endpoint_parse(ep, &v4l2_ep);
 	if (ret) {
 		dev_err(priv->dev, "Could not parse v4l2 endpoint\n");
-		of_node_put(ep);
+		fwnode_handle_put(ep);
 		return -EINVAL;
 	}
 
 	ret = rcsi2_parse_v4l2(priv, &v4l2_ep);
 	if (ret) {
-		of_node_put(ep);
+		fwnode_handle_put(ep);
 		return ret;
 	}
 
-	fwnode = fwnode_graph_get_remote_endpoint(of_fwnode_handle(ep));
-	of_node_put(ep);
+	fwnode = fwnode_graph_get_remote_endpoint(ep);
+	fwnode_handle_put(ep);
 
 	dev_dbg(priv->dev, "Found '%pOF'\n", to_of_node(fwnode));
 
-- 
2.28.0

[PATCH 2/2] rcar-csi2: Set bus type when parsing fwnode

[Niklas Söderlund]

The only supported bus for the R-Car CSI-2 driver is CSI-2 DPHY, specify
this before parsing the fwnode.

Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
---
 drivers/media/platform/rcar-vin/rcar-csi2.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/media/platform/rcar-vin/rcar-csi2.c b/drivers/media/platform/rcar-vin/rcar-csi2.c
index 23e89ef2429d310a..b2e58f51b94fccd7 100644
--- a/drivers/media/platform/rcar-vin/rcar-csi2.c
+++ b/drivers/media/platform/rcar-vin/rcar-csi2.c
@@ -874,7 +874,9 @@ static int rcsi2_parse_dt(struct rcar_csi2 *priv)
 	struct v4l2_async_subdev *asd;
 	struct fwnode_handle *fwnode;
 	struct fwnode_handle *ep;
-	struct v4l2_fwnode_endpoint v4l2_ep = { .bus_type = 0 };
+	struct v4l2_fwnode_endpoint v4l2_ep = {
+		.bus_type = V4L2_MBUS_CSI2_DPHY
+	};
 	int ret;
 
 	ep = fwnode_graph_get_endpoint_by_id(dev_fwnode(priv->dev), 0, 0, 0);
-- 
2.28.0

Re: [PATCH 1/2] rcar-csi2: Switch to using fwnode instead of OF

[Jacopo Mondi]

Hi Niklas,

On Mon, Sep 14, 2020 at 11:50:10PM +0200, Niklas Söderlund wrote:
> Use the fwnode_graph_get_endpoint_by_id() interface instead of
> of_graph_get_endpoint_by_regs() to fetch the fwnode. This saves
> translating between a device_node and fwnode_handle.
>
> Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>

Looks good
Reviewed-by: Jacopo Mondi <jacopo+renesas@jmondi.org>

Thanks
  j

> ---
>  drivers/media/platform/rcar-vin/rcar-csi2.c | 14 +++++++-------
>  1 file changed, 7 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/media/platform/rcar-vin/rcar-csi2.c b/drivers/media/platform/rcar-vin/rcar-csi2.c
> index 511cd4984777ad99..23e89ef2429d310a 100644
> --- a/drivers/media/platform/rcar-vin/rcar-csi2.c
> +++ b/drivers/media/platform/rcar-vin/rcar-csi2.c
> @@ -873,31 +873,31 @@ static int rcsi2_parse_dt(struct rcar_csi2 *priv)
>  {
>  	struct v4l2_async_subdev *asd;
>  	struct fwnode_handle *fwnode;
> -	struct device_node *ep;
> +	struct fwnode_handle *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);
> +	ep = fwnode_graph_get_endpoint_by_id(dev_fwnode(priv->dev), 0, 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);
> +	ret = v4l2_fwnode_endpoint_parse(ep, &v4l2_ep);
>  	if (ret) {
>  		dev_err(priv->dev, "Could not parse v4l2 endpoint\n");
> -		of_node_put(ep);
> +		fwnode_handle_put(ep);
>  		return -EINVAL;
>  	}
>
>  	ret = rcsi2_parse_v4l2(priv, &v4l2_ep);
>  	if (ret) {
> -		of_node_put(ep);
> +		fwnode_handle_put(ep);
>  		return ret;
>  	}
>
> -	fwnode = fwnode_graph_get_remote_endpoint(of_fwnode_handle(ep));
> -	of_node_put(ep);
> +	fwnode = fwnode_graph_get_remote_endpoint(ep);
> +	fwnode_handle_put(ep);
>
>  	dev_dbg(priv->dev, "Found '%pOF'\n", to_of_node(fwnode));
>
> --
> 2.28.0
>

Re: [PATCH 2/2] rcar-csi2: Set bus type when parsing fwnode

[Jacopo Mondi]

Hi Niklas,

On Mon, Sep 14, 2020 at 11:50:11PM +0200, Niklas Söderlund wrote:
> The only supported bus for the R-Car CSI-2 driver is CSI-2 DPHY, specify
> this before parsing the fwnode.
>
> Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
> ---
>  drivers/media/platform/rcar-vin/rcar-csi2.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/media/platform/rcar-vin/rcar-csi2.c b/drivers/media/platform/rcar-vin/rcar-csi2.c
> index 23e89ef2429d310a..b2e58f51b94fccd7 100644
> --- a/drivers/media/platform/rcar-vin/rcar-csi2.c
> +++ b/drivers/media/platform/rcar-vin/rcar-csi2.c
> @@ -874,7 +874,9 @@ static int rcsi2_parse_dt(struct rcar_csi2 *priv)
>  	struct v4l2_async_subdev *asd;
>  	struct fwnode_handle *fwnode;
>  	struct fwnode_handle *ep;
> -	struct v4l2_fwnode_endpoint v4l2_ep = { .bus_type = 0 };
> +	struct v4l2_fwnode_endpoint v4l2_ep = {
> +		.bus_type = V4L2_MBUS_CSI2_DPHY
> +	};

I would also take the occasion to make bus-type mandatory in
bindings as v4l2_fwnode_endpoint_parse() will fail only if it detect a
mismatch between bus_type and "bus-type".

For older DTS we won't detect mismatches, but that's not worse than
what we have today.

In case you update bindings I would update the error message in the
v4l2_fwnode_endpoint_parse() failure path to report the mismatch.

The patch itself is good
Reviewed-by: Jacopo Mondi <jacopo+renesas@jmondi.org>

Thanks
  j


>  	int ret;
>
>  	ep = fwnode_graph_get_endpoint_by_id(dev_fwnode(priv->dev), 0, 0, 0);
> --
> 2.28.0
>

Re: [PATCH 2/2] rcar-csi2: Set bus type when parsing fwnode

[Sakari Ailus]

Hi Jacopo,

On Tue, Sep 15, 2020 at 01:27:52PM +0200, Jacopo Mondi wrote:
> Hi Niklas,
> 
> On Mon, Sep 14, 2020 at 11:50:11PM +0200, Niklas Söderlund wrote:
> > The only supported bus for the R-Car CSI-2 driver is CSI-2 DPHY, specify
> > this before parsing the fwnode.
> >
> > Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
> > ---
> >  drivers/media/platform/rcar-vin/rcar-csi2.c | 4 +++-
> >  1 file changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/media/platform/rcar-vin/rcar-csi2.c b/drivers/media/platform/rcar-vin/rcar-csi2.c
> > index 23e89ef2429d310a..b2e58f51b94fccd7 100644
> > --- a/drivers/media/platform/rcar-vin/rcar-csi2.c
> > +++ b/drivers/media/platform/rcar-vin/rcar-csi2.c
> > @@ -874,7 +874,9 @@ static int rcsi2_parse_dt(struct rcar_csi2 *priv)
> >  	struct v4l2_async_subdev *asd;
> >  	struct fwnode_handle *fwnode;
> >  	struct fwnode_handle *ep;
> > -	struct v4l2_fwnode_endpoint v4l2_ep = { .bus_type = 0 };
> > +	struct v4l2_fwnode_endpoint v4l2_ep = {
> > +		.bus_type = V4L2_MBUS_CSI2_DPHY
> > +	};
> 
> I would also take the occasion to make bus-type mandatory in
> bindings as v4l2_fwnode_endpoint_parse() will fail only if it detect a
> mismatch between bus_type and "bus-type".

You don't really need bus-type property if the hardware supports a single
type. Then you can, as above, parse the endpoint with that type set by the
caller.

-- 
Sakari Ailus

Re: [PATCH 2/2] rcar-csi2: Set bus type when parsing fwnode

[Jacopo Mondi]

Hi Sakari,

On Sat, Oct 03, 2020 at 12:21:03AM +0300, Sakari Ailus wrote:
> Hi Jacopo,
>
> On Tue, Sep 15, 2020 at 01:27:52PM +0200, Jacopo Mondi wrote:
> > Hi Niklas,
> >
> > On Mon, Sep 14, 2020 at 11:50:11PM +0200, Niklas Söderlund wrote:
> > > The only supported bus for the R-Car CSI-2 driver is CSI-2 DPHY, specify
> > > this before parsing the fwnode.
> > >
> > > Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
> > > ---
> > >  drivers/media/platform/rcar-vin/rcar-csi2.c | 4 +++-
> > >  1 file changed, 3 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/media/platform/rcar-vin/rcar-csi2.c b/drivers/media/platform/rcar-vin/rcar-csi2.c
> > > index 23e89ef2429d310a..b2e58f51b94fccd7 100644
> > > --- a/drivers/media/platform/rcar-vin/rcar-csi2.c
> > > +++ b/drivers/media/platform/rcar-vin/rcar-csi2.c
> > > @@ -874,7 +874,9 @@ static int rcsi2_parse_dt(struct rcar_csi2 *priv)
> > >  	struct v4l2_async_subdev *asd;
> > >  	struct fwnode_handle *fwnode;
> > >  	struct fwnode_handle *ep;
> > > -	struct v4l2_fwnode_endpoint v4l2_ep = { .bus_type = 0 };
> > > +	struct v4l2_fwnode_endpoint v4l2_ep = {
> > > +		.bus_type = V4L2_MBUS_CSI2_DPHY
> > > +	};
> >
> > I would also take the occasion to make bus-type mandatory in
> > bindings as v4l2_fwnode_endpoint_parse() will fail only if it detect a
> > mismatch between bus_type and "bus-type".
>
> You don't really need bus-type property if the hardware supports a single
> type. Then you can, as above, parse the endpoint with that type set by the
> caller.

Ok, that's a bit confusing as if there's no bus-type property no
bus mismatch could ever be detected, not at run-time by the v4l2-fwnode
framework, nor by DTS validation. Of course, the chances that a DTS for a
device that only supports CSI-2 specifies (in example) parallel bus
properties are quite low, so I'm fine with the way things are.

>
> --
> Sakari Ailus