[v3,5/5] media: i2c: alvium: fix req_fr check into alvium_s_frame_interval()

  Actually req_fr check into alvium_s_frame_interval() is wrong.
In particular req_fr can't be >=max and <= min at the same time.
Fix this using clamp and remove dft_fr parameter from
alvium_get_frame_interval() not more used.

Signed-off-by: Tommaso Merciai <tomm.merciai@gmail.com>
---
 drivers/media/i2c/alvium-csi2.c | 12 ++++--------
 1 file changed, 4 insertions(+), 8 deletions(-)
  

Hi Tommaso,

Thank you for the patch.

On Wed, Dec 20, 2023 at 01:40:23PM +0100, Tommaso Merciai wrote:
> Actually req_fr check into alvium_s_frame_interval() is wrong.
> In particular req_fr can't be >=max and <= min at the same time.
> Fix this using clamp and remove dft_fr parameter from
> alvium_get_frame_interval() not more used.

The commit message should have explained why clamping is better than
picking a default value, as that's a functional change. If you propose
an updated commit message in a reply, I think Sakari can update the
patch when applying the series to his tree, there's no need for a v4.

> Signed-off-by: Tommaso Merciai <tomm.merciai@gmail.com>

Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>

> ---
>  drivers/media/i2c/alvium-csi2.c | 12 ++++--------
>  1 file changed, 4 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/media/i2c/alvium-csi2.c b/drivers/media/i2c/alvium-csi2.c
> index 240bf991105e..01111a00902d 100644
> --- a/drivers/media/i2c/alvium-csi2.c
> +++ b/drivers/media/i2c/alvium-csi2.c
> @@ -1171,12 +1171,10 @@ static int alvium_set_bayer_pattern(struct alvium_dev *alvium,
>  }
>  
>  static int alvium_get_frame_interval(struct alvium_dev *alvium,
> -				     u64 *dft_fr, u64 *min_fr, u64 *max_fr)
> +				     u64 *min_fr, u64 *max_fr)
>  {
>  	int ret = 0;
>  
> -	alvium_read(alvium, REG_BCRM_ACQUISITION_FRAME_RATE_RW,
> -		    dft_fr, &ret);
>  	alvium_read(alvium, REG_BCRM_ACQUISITION_FRAME_RATE_MIN_R,
>  		    min_fr, &ret);
>  	alvium_read(alvium, REG_BCRM_ACQUISITION_FRAME_RATE_MAX_R,
> @@ -1647,7 +1645,7 @@ static int alvium_s_frame_interval(struct v4l2_subdev *sd,
>  {
>  	struct alvium_dev *alvium = sd_to_alvium(sd);
>  	struct device *dev = &alvium->i2c_client->dev;
> -	u64 req_fr, dft_fr, min_fr, max_fr;
> +	u64 req_fr, min_fr, max_fr;
>  	struct v4l2_fract *interval;
>  	int ret;
>  
> @@ -1657,7 +1655,7 @@ static int alvium_s_frame_interval(struct v4l2_subdev *sd,
>  	if (fi->interval.denominator == 0)
>  		return -EINVAL;
>  
> -	ret = alvium_get_frame_interval(alvium, &dft_fr, &min_fr, &max_fr);
> +	ret = alvium_get_frame_interval(alvium, &min_fr, &max_fr);
>  	if (ret) {
>  		dev_err(dev, "Fail to get frame interval\n");
>  		return ret;
> @@ -1670,9 +1668,7 @@ static int alvium_s_frame_interval(struct v4l2_subdev *sd,
>  
>  	req_fr = (u64)((fi->interval.denominator * USEC_PER_SEC) /
>  		       fi->interval.numerator);
> -
> -	if (req_fr >= max_fr && req_fr <= min_fr)
> -		req_fr = dft_fr;
> +	req_fr = clamp(req_fr, min_fr, max_fr);
>  
>  	interval = v4l2_subdev_state_get_interval(sd_state, 0);
>
  

Hi Laurent,

On Wed, Dec 20, 2023 at 03:02:36PM +0200, Laurent Pinchart wrote:
> Hi Tommaso,
> 
> Thank you for the patch.
> 
> On Wed, Dec 20, 2023 at 01:40:23PM +0100, Tommaso Merciai wrote:
> > Actually req_fr check into alvium_s_frame_interval() is wrong.
> > In particular req_fr can't be >=max and <= min at the same time.
> > Fix this using clamp and remove dft_fr parameter from
> > alvium_get_frame_interval() not more used.
> 
> The commit message should have explained why clamping is better than
> picking a default value, as that's a functional change. If you propose
> an updated commit message in a reply, I think Sakari can update the
> patch when applying the series to his tree, there's no need for a v4.

What about:

Actually req_fr check into alvium_s_frame_interval() is wrong.
In particular req_fr can't be >=max and <= min at the same time.
Fix this using clamp and remove dft_fr parameter from
alvium_get_frame_interval() not more used.

Clamp function make sure that if the setted value exceeds the limits is
replaced with min_fr/max_fr instead of setting the value readed back
from the hw.

What do you think?

Thanks & Regards,
Tommaso

> 
> > Signed-off-by: Tommaso Merciai <tomm.merciai@gmail.com>
> 
> Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> 
> > ---
> >  drivers/media/i2c/alvium-csi2.c | 12 ++++--------
> >  1 file changed, 4 insertions(+), 8 deletions(-)
> > 
> > diff --git a/drivers/media/i2c/alvium-csi2.c b/drivers/media/i2c/alvium-csi2.c
> > index 240bf991105e..01111a00902d 100644
> > --- a/drivers/media/i2c/alvium-csi2.c
> > +++ b/drivers/media/i2c/alvium-csi2.c
> > @@ -1171,12 +1171,10 @@ static int alvium_set_bayer_pattern(struct alvium_dev *alvium,
> >  }
> >  
> >  static int alvium_get_frame_interval(struct alvium_dev *alvium,
> > -				     u64 *dft_fr, u64 *min_fr, u64 *max_fr)
> > +				     u64 *min_fr, u64 *max_fr)
> >  {
> >  	int ret = 0;
> >  
> > -	alvium_read(alvium, REG_BCRM_ACQUISITION_FRAME_RATE_RW,
> > -		    dft_fr, &ret);
> >  	alvium_read(alvium, REG_BCRM_ACQUISITION_FRAME_RATE_MIN_R,
> >  		    min_fr, &ret);
> >  	alvium_read(alvium, REG_BCRM_ACQUISITION_FRAME_RATE_MAX_R,
> > @@ -1647,7 +1645,7 @@ static int alvium_s_frame_interval(struct v4l2_subdev *sd,
> >  {
> >  	struct alvium_dev *alvium = sd_to_alvium(sd);
> >  	struct device *dev = &alvium->i2c_client->dev;
> > -	u64 req_fr, dft_fr, min_fr, max_fr;
> > +	u64 req_fr, min_fr, max_fr;
> >  	struct v4l2_fract *interval;
> >  	int ret;
> >  
> > @@ -1657,7 +1655,7 @@ static int alvium_s_frame_interval(struct v4l2_subdev *sd,
> >  	if (fi->interval.denominator == 0)
> >  		return -EINVAL;
> >  
> > -	ret = alvium_get_frame_interval(alvium, &dft_fr, &min_fr, &max_fr);
> > +	ret = alvium_get_frame_interval(alvium, &min_fr, &max_fr);
> >  	if (ret) {
> >  		dev_err(dev, "Fail to get frame interval\n");
> >  		return ret;
> > @@ -1670,9 +1668,7 @@ static int alvium_s_frame_interval(struct v4l2_subdev *sd,
> >  
> >  	req_fr = (u64)((fi->interval.denominator * USEC_PER_SEC) /
> >  		       fi->interval.numerator);
> > -
> > -	if (req_fr >= max_fr && req_fr <= min_fr)
> > -		req_fr = dft_fr;
> > +	req_fr = clamp(req_fr, min_fr, max_fr);
> >  
> >  	interval = v4l2_subdev_state_get_interval(sd_state, 0);
> >  
> 
> -- 
> Regards,
> 
> Laurent Pinchart
  

Hi Tommaso,

On Wed, Dec 20, 2023 at 02:52:53PM +0100, Tommaso Merciai wrote:
> Hi Laurent,
> 
> On Wed, Dec 20, 2023 at 03:02:36PM +0200, Laurent Pinchart wrote:
> > Hi Tommaso,
> > 
> > Thank you for the patch.
> > 
> > On Wed, Dec 20, 2023 at 01:40:23PM +0100, Tommaso Merciai wrote:
> > > Actually req_fr check into alvium_s_frame_interval() is wrong.
> > > In particular req_fr can't be >=max and <= min at the same time.
> > > Fix this using clamp and remove dft_fr parameter from
> > > alvium_get_frame_interval() not more used.
> > 
> > The commit message should have explained why clamping is better than
> > picking a default value, as that's a functional change. If you propose
> > an updated commit message in a reply, I think Sakari can update the
> > patch when applying the series to his tree, there's no need for a v4.
> 
> What about:
> 
> Actually req_fr check into alvium_s_frame_interval() is wrong.
> In particular req_fr can't be >=max and <= min at the same time.
> Fix this using clamp and remove dft_fr parameter from
> alvium_get_frame_interval() not more used.
> 
> Clamp function make sure that if the setted value exceeds the limits is
> replaced with min_fr/max_fr instead of setting the value readed back
> from the hw.
> 
> What do you think?

I used this, hopefully it's ok:

media: i2c: alvium: fix req_fr check in alvium_s_frame_interval()

req_fr check in alvium_s_frame_interval() is incorrect. In particular
req_fr can't be >=max and <= min at the same time. Ensure the requested
frame rate remains within the supported range between min_fr and max_fr by
clamping it.

Also remove the unused dft_fr argument of alvium_get_frame_interval().
  

Hi Sakari,
Sorry for delay.

On Wed, Feb 14, 2024 at 11:14:35AM +0000, Sakari Ailus wrote:
> Hi Tommaso,
> 
> On Wed, Dec 20, 2023 at 02:52:53PM +0100, Tommaso Merciai wrote:
> > Hi Laurent,
> > 
> > On Wed, Dec 20, 2023 at 03:02:36PM +0200, Laurent Pinchart wrote:
> > > Hi Tommaso,
> > > 
> > > Thank you for the patch.
> > > 
> > > On Wed, Dec 20, 2023 at 01:40:23PM +0100, Tommaso Merciai wrote:
> > > > Actually req_fr check into alvium_s_frame_interval() is wrong.
> > > > In particular req_fr can't be >=max and <= min at the same time.
> > > > Fix this using clamp and remove dft_fr parameter from
> > > > alvium_get_frame_interval() not more used.
> > > 
> > > The commit message should have explained why clamping is better than
> > > picking a default value, as that's a functional change. If you propose
> > > an updated commit message in a reply, I think Sakari can update the
> > > patch when applying the series to his tree, there's no need for a v4.
> > 
> > What about:
> > 
> > Actually req_fr check into alvium_s_frame_interval() is wrong.
> > In particular req_fr can't be >=max and <= min at the same time.
> > Fix this using clamp and remove dft_fr parameter from
> > alvium_get_frame_interval() not more used.
> > 
> > Clamp function make sure that if the setted value exceeds the limits is
> > replaced with min_fr/max_fr instead of setting the value readed back
> > from the hw.
> > 
> > What do you think?
> 
> I used this, hopefully it's ok:
> 
> media: i2c: alvium: fix req_fr check in alvium_s_frame_interval()
> 
> req_fr check in alvium_s_frame_interval() is incorrect. In particular
> req_fr can't be >=max and <= min at the same time. Ensure the requested
> frame rate remains within the supported range between min_fr and max_fr by
> clamping it.
> 
> Also remove the unused dft_fr argument of alvium_get_frame_interval().

Looks good to me! :)
Thanks for this.

Regards,
Tommaso


> 
> -- 
> Kind regards,
> 
> Sakari Ailus