[PATCH v2 0/2] ARM: davinci: misc. VPIF cleanups

[PATCH v2 0/2] ARM: davinci: misc. VPIF cleanups

[Kevin Hilman]

This are the platforms-specific cleanups needed to go along with the
VPIF cleanups for raw capture support[1], which is now merged in the
media tree.

[1] https://marc.info/?l=linux-arm-kernel&m=149643932427173

Changes since v1:
- fix additional warning in board-dm646x-evm.c
- add DT maintiners in CC for patch 2

Kevin Hilman (2):
  ARM: davinci: pdata-quirks: fix const warnings
  ARM: dts: da850-evm: drop VPIF endpoints

 arch/arm/boot/dts/da850-evm.dts          | 21 ---------------------
 arch/arm/mach-davinci/board-dm646x-evm.c |  4 ++--
 arch/arm/mach-davinci/pdata-quirks.c     |  4 ++--
 3 files changed, 4 insertions(+), 25 deletions(-)

-- 
2.9.3

[PATCH v2 1/2] ARM: davinci: pdata-quirks: fix const warnings

[Kevin Hilman]

After VPIF was converted to enable getting subdevs from DT, the
pdata is no longer const, so remove these to avoid compiler warnings.

Signed-off-by: Kevin Hilman <khilman@baylibre.com>
---
 arch/arm/mach-davinci/board-dm646x-evm.c | 4 ++--
 arch/arm/mach-davinci/pdata-quirks.c     | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/arch/arm/mach-davinci/board-dm646x-evm.c b/arch/arm/mach-davinci/board-dm646x-evm.c
index cb176826d1cb..055e947a6a39 100644
--- a/arch/arm/mach-davinci/board-dm646x-evm.c
+++ b/arch/arm/mach-davinci/board-dm646x-evm.c
@@ -641,7 +641,7 @@ static struct vpif_subdev_info vpif_capture_sdev_info[] = {
 	},
 };
 
-static const struct vpif_input dm6467_ch0_inputs[] = {
+static struct vpif_input dm6467_ch0_inputs[] = {
 	{
 		.input = {
 			.index = 0,
@@ -656,7 +656,7 @@ static const struct vpif_input dm6467_ch0_inputs[] = {
 	},
 };
 
-static const struct vpif_input dm6467_ch1_inputs[] = {
+static struct vpif_input dm6467_ch1_inputs[] = {
        {
 		.input = {
 			.index = 0,
diff --git a/arch/arm/mach-davinci/pdata-quirks.c b/arch/arm/mach-davinci/pdata-quirks.c
index 329f5402ad1d..4858b1cdf31b 100644
--- a/arch/arm/mach-davinci/pdata-quirks.c
+++ b/arch/arm/mach-davinci/pdata-quirks.c
@@ -33,7 +33,7 @@ static struct tvp514x_platform_data tvp5146_pdata = {
 
 #define TVP514X_STD_ALL (V4L2_STD_NTSC | V4L2_STD_PAL)
 
-static const struct vpif_input da850_ch0_inputs[] = {
+static struct vpif_input da850_ch0_inputs[] = {
 	{
 		.input = {
 			.index = 0,
@@ -48,7 +48,7 @@ static const struct vpif_input da850_ch0_inputs[] = {
 	},
 };
 
-static const struct vpif_input da850_ch1_inputs[] = {
+static struct vpif_input da850_ch1_inputs[] = {
 	{
 		.input = {
 			.index = 0,
-- 
2.9.3

[PATCH v2 2/2] ARM: dts: da850-evm: drop VPIF endpoints

[Kevin Hilman]

Drop the unused endpoints.  They should only be used when there is
an actual remote-endpoint connected.

Signed-off-by: Kevin Hilman <khilman@baylibre.com>
---
 arch/arm/boot/dts/da850-evm.dts | 21 ---------------------
 1 file changed, 21 deletions(-)

diff --git a/arch/arm/boot/dts/da850-evm.dts b/arch/arm/boot/dts/da850-evm.dts
index a423e8ebfb37..67e72bc72e80 100644
--- a/arch/arm/boot/dts/da850-evm.dts
+++ b/arch/arm/boot/dts/da850-evm.dts
@@ -301,25 +301,4 @@
 	pinctrl-names = "default";
 	pinctrl-0 = <&vpif_capture_pins>, <&vpif_display_pins>;
 	status = "okay";
-
-	/* VPIF capture port */
-	port at 0 {
-		vpif_input_ch0: endpoint at 0 {
-			reg = <0>;
-			bus-width = <8>;
-		};
-
-		vpif_input_ch1: endpoint at 1 {
-			reg = <1>;
-			bus-width = <8>;
-			data-shift = <8>;
-		};
-	};
-
-	/* VPIF display port */
-	port at 1 {
-		vpif_output_ch0: endpoint {
-			bus-width = <8>;
-		};
-	};
 };
-- 
2.9.3

[PATCH v2 2/2] ARM: dts: da850-evm: drop VPIF endpoints

[David Lechner]

On 06/09/2017 12:21 PM, Kevin Hilman wrote:
> Drop the unused endpoints.  They should only be used when there is
> an actual remote-endpoint connected.
> 

Would it make sense to implement the status property here instead of 
removing the nodes completely? If other property values are the same for 
all users, then it seems like making all users have to dig up the docs 
and copy/paste is not as nice as just overriding the status property. 
But, if users have to change other values as well, then it makes sense 
to just delete these.


 > -	/* VPIF capture port */
 > -	port at 0 {

-->		status = "disabled";

 > -		vpif_input_ch0: endpoint at 0 {
 > -			reg = <0>;
 > -			bus-width = <8>;
 > -		};
 > -
 > -		vpif_input_ch1: endpoint at 1 {
 > -			reg = <1>;
 > -			bus-width = <8>;
 > -			data-shift = <8>;
 > -		};
 > -	};
 > -
 > -	/* VPIF display port */
 > -	port at 1 {

-->		status = "disabled";

 > -		vpif_output_ch0: endpoint {
 > -			bus-width = <8>;
 > -		};
 > -	};

> Signed-off-by: Kevin Hilman <khilman@baylibre.com>
> ---
>   arch/arm/boot/dts/da850-evm.dts | 21 ---------------------
>   1 file changed, 21 deletions(-)
> 
> diff --git a/arch/arm/boot/dts/da850-evm.dts b/arch/arm/boot/dts/da850-evm.dts
> index a423e8ebfb37..67e72bc72e80 100644
> --- a/arch/arm/boot/dts/da850-evm.dts
> +++ b/arch/arm/boot/dts/da850-evm.dts
> @@ -301,25 +301,4 @@
>   	pinctrl-names = "default";
>   	pinctrl-0 = <&vpif_capture_pins>, <&vpif_display_pins>;
>   	status = "okay";
> -
> -	/* VPIF capture port */
> -	port at 0 {
> -		vpif_input_ch0: endpoint at 0 {
> -			reg = <0>;
> -			bus-width = <8>;
> -		};
> -
> -		vpif_input_ch1: endpoint at 1 {
> -			reg = <1>;
> -			bus-width = <8>;
> -			data-shift = <8>;
> -		};
> -	};
> -
> -	/* VPIF display port */
> -	port at 1 {
> -		vpif_output_ch0: endpoint {
> -			bus-width = <8>;
> -		};
> -	};
>   };
> 

[PATCH v2 2/2] ARM: dts: da850-evm: drop VPIF endpoints

[Kevin Hilman]

Hi David,

On Fri, Jun 9, 2017 at 2:17 PM, David Lechner <david@lechnology.com> wrote:
> On 06/09/2017 12:21 PM, Kevin Hilman wrote:
>>
>> Drop the unused endpoints.  They should only be used when there is
>> an actual remote-endpoint connected.
>>
>
> Would it make sense to implement the status property here instead of
> removing the nodes completely? If other property values are the same for all
> users, then it seems like making all users have to dig up the docs and
> copy/paste is not as nice as just overriding the status property. But, if
> users have to change other values as well, then it makes sense to just
> delete these.

I considered that, but in reality, when adding an endpoint for VPIF
input, you'll not only need to add the remote-endpoint, but the
bus_width will depend on the remote-endpoint as well (e.g. tvp514x for
composite only uses 8-bit bus, but for raw-camera capture, it can use
10 or 12.)

Therefore, I think it best to delete them entirely.

Thanks for the review,

Kevin

[PATCH v2 1/2] ARM: davinci: pdata-quirks: fix const warnings

[Sekhar Nori]

On Friday 09 June 2017 10:51 PM, Kevin Hilman wrote:
> After VPIF was converted to enable getting subdevs from DT, the
> pdata is no longer const, so remove these to avoid compiler warnings.
> 
> Signed-off-by: Kevin Hilman <khilman@baylibre.com>

Applied to v4.13/soc. I dropped the "pdata-quirks" prefix in subject
line since the patch is fixing more that that.

Thanks,
Sekhar