[PATCH/RFC] v4l: vsp1: Change pixel count at scale-up setting

[PATCH/RFC] v4l: vsp1: Change pixel count at scale-up setting

[Yoshihiro Kaneko]

From: Atsushi Akatsuka <atsushi.akatsuka.vb@hitachi.com>

This commit sets AMD bit of VI6_UDSn_CTRL register,
and modifies scaling formula to fit AMD bit.

Signed-off-by: Atsushi Akatsuka <atsushi.akatsuka.vb@hitachi.com>
Signed-off-by: Hiroki Negishi <hiroki.negishi.zr@hitachi-solutions.com>
Signed-off-by: Yoshihiro Kaneko <ykaneko0929@gmail.com>
---

This patch is based on the master branch of linuxtv.org/media_tree.git.

 drivers/media/platform/vsp1/vsp1_uds.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/media/platform/vsp1/vsp1_uds.c b/drivers/media/platform/vsp1/vsp1_uds.c
index ccc8243..e7a046d 100644
--- a/drivers/media/platform/vsp1/vsp1_uds.c
+++ b/drivers/media/platform/vsp1/vsp1_uds.c
@@ -64,10 +64,10 @@ static unsigned int uds_output_size(unsigned int input, unsigned int ratio)
 		mp = ratio / 4096;
 		mp = mp < 4 ? 1 : (mp < 8 ? 2 : 4);
 
-		return (input - 1) / mp * mp * 4096 / ratio + 1;
+		return input / mp * mp * 4096 / ratio;
 	} else {
 		/* Up-scaling */
-		return (input - 1) * 4096 / ratio + 1;
+		return input * 4096 / ratio;
 	}
 }
 
@@ -145,7 +145,8 @@ static int uds_s_stream(struct v4l2_subdev *subdev, int enable)
 
 	vsp1_uds_write(uds, VI6_UDS_CTRL,
 		       (uds->scale_alpha ? VI6_UDS_CTRL_AON : 0) |
-		       (multitap ? VI6_UDS_CTRL_BC : 0));
+		       (multitap ? VI6_UDS_CTRL_BC : 0) |
+		       VI6_UDS_CTRL_AMD);
 
 	vsp1_uds_write(uds, VI6_UDS_PASS_BWIDTH,
 		       (uds_passband_width(hscale)
-- 
1.9.1

Re: [PATCH/RFC] v4l: vsp1: Change pixel count at scale-up setting

[Laurent Pinchart]

Hi Kaneko-san,

Thank you for the patch.

On Monday 15 June 2015 02:29:14 Yoshihiro Kaneko wrote:
> From: Atsushi Akatsuka <atsushi.akatsuka.vb@hitachi.com>
> 
> This commit sets AMD bit of VI6_UDSn_CTRL register,
> and modifies scaling formula to fit AMD bit.

What's the rationale for that ? What are the side effects of setting the AMD 
bit ? Will it change anything beside the scaling factor computation formula ?

> Signed-off-by: Atsushi Akatsuka <atsushi.akatsuka.vb@hitachi.com>
> Signed-off-by: Hiroki Negishi <hiroki.negishi.zr@hitachi-solutions.com>
> Signed-off-by: Yoshihiro Kaneko <ykaneko0929@gmail.com>
> ---
> 
> This patch is based on the master branch of linuxtv.org/media_tree.git.
> 
>  drivers/media/platform/vsp1/vsp1_uds.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/media/platform/vsp1/vsp1_uds.c
> b/drivers/media/platform/vsp1/vsp1_uds.c index ccc8243..e7a046d 100644
> --- a/drivers/media/platform/vsp1/vsp1_uds.c
> +++ b/drivers/media/platform/vsp1/vsp1_uds.c
> @@ -64,10 +64,10 @@ static unsigned int uds_output_size(unsigned int input,
> unsigned int ratio) mp = ratio / 4096;
>  		mp = mp < 4 ? 1 : (mp < 8 ? 2 : 4);
> 
> -		return (input - 1) / mp * mp * 4096 / ratio + 1;
> +		return input / mp * mp * 4096 / ratio;

According to the datasheet I have access to the AMD bit only influences the 
scale-up case.

>  	} else {
>  		/* Up-scaling */
> -		return (input - 1) * 4096 / ratio + 1;
> +		return input * 4096 / ratio;
>  	}
>  }
> 
> @@ -145,7 +145,8 @@ static int uds_s_stream(struct v4l2_subdev *subdev, int
> enable)
> 
>  	vsp1_uds_write(uds, VI6_UDS_CTRL,
>  		       (uds->scale_alpha ? VI6_UDS_CTRL_AON : 0) |
> -		       (multitap ? VI6_UDS_CTRL_BC : 0));
> +		       (multitap ? VI6_UDS_CTRL_BC : 0) |
> +		       VI6_UDS_CTRL_AMD);
> 
>  	vsp1_uds_write(uds, VI6_UDS_PASS_BWIDTH,
>  		       (uds_passband_width(hscale)

-- 
Regards,

Laurent Pinchart