Re: [PATCH v2 2/4] media: i2c: alvium: removal of fr field

[Laurent Pinchart <laurent.pinchart@ideasonboard.com>]

On Wed, Dec 20, 2023 at 01:03:12PM +0100, Tommaso Merciai wrote:
> On Wed, Dec 20, 2023 at 01:29:02PM +0200, Laurent Pinchart wrote:
> > On Wed, Dec 20, 2023 at 12:19:23PM +0100, Tommaso Merciai wrote:
> > > On Wed, Dec 20, 2023 at 12:06:43PM +0200, Laurent Pinchart wrote:
> > > > On Wed, Dec 20, 2023 at 11:02:51AM +0100, Tommaso Merciai wrote:
> > > > > On Wed, Dec 20, 2023 at 11:13:09AM +0200, Laurent Pinchart wrote:
> > > > > > Hi Tommaso,
> > > > > > 
> > > > > > Thank you for the patch.
> > > > > > 
> > > > > > Use the imperative in the subject line:
> > > > > > 
> > > > > > media: i2c: alvium: Remove the fr field of the alvium_dev structure
> > > > > > 
> > > > > > On Wed, Dec 20, 2023 at 09:56:07AM +0100, Tommaso Merciai wrote:
> > > > > > > The fr (frame rate) field of the alvium_dev structure is
> > > > > > > only used to pass result from alvium_set_frame_interval() to
> > > > > > > alvium_set_frame_rate() that writes this info into the hw reg.
> > > > > > > Replace them with function parameter.
> > > > > > 
> > > > > > Replace it with a function parameter.
> > > > > 
> > > > > Thanks I'll fix this in v3.
> > > > > 
> > > > > > > 
> > > > > > > Signed-off-by: Tommaso Merciai <tomm.merciai@gmail.com>
> > > > > > > ---
> > > > > > >  drivers/media/i2c/alvium-csi2.c | 24 ++++++++++++------------
> > > > > > >  drivers/media/i2c/alvium-csi2.h |  1 -
> > > > > > >  2 files changed, 12 insertions(+), 13 deletions(-)
> > > > > > > 
> > > > > > > diff --git a/drivers/media/i2c/alvium-csi2.c b/drivers/media/i2c/alvium-csi2.c
> > > > > > > index 0dcd69bf9f92..a9ff6cc97cff 100644
> > > > > > > --- a/drivers/media/i2c/alvium-csi2.c
> > > > > > > +++ b/drivers/media/i2c/alvium-csi2.c
> > > > > > > @@ -1185,19 +1185,19 @@ static int alvium_get_frame_interval(struct alvium_dev *alvium,
> > > > > > >  	return ret;
> > > > > > >  }
> > > > > > >  
> > > > > > > -static int alvium_set_frame_rate(struct alvium_dev *alvium)
> > > > > > > +static int alvium_set_frame_rate(struct alvium_dev *alvium, u64 fr)
> > > > > > >  {
> > > > > > >  	struct device *dev = &alvium->i2c_client->dev;
> > > > > > >  	int ret;
> > > > > > >  
> > > > > > >  	ret = alvium_write_hshake(alvium, REG_BCRM_ACQUISITION_FRAME_RATE_RW,
> > > > > > > -				  alvium->fr);
> > > > > > > +				  fr);
> > > > > > >  	if (ret) {
> > > > > > >  		dev_err(dev, "Fail to set frame rate lanes reg\n");
> > > > > > >  		return ret;
> > > > > > >  	}
> > > > > > >  
> > > > > > > -	dev_dbg(dev, "set frame rate: %llu us\n", alvium->fr);
> > > > > > > +	dev_dbg(dev, "set frame rate: %llu us\n", fr);
> > > > > > >  
> > > > > > >  	return 0;
> > > > > > >  }
> > > > > > > @@ -1661,10 +1661,11 @@ static int alvium_g_frame_interval(struct v4l2_subdev *sd,
> > > > > > >  }
> > > > > > >  
> > > > > > >  static int alvium_set_frame_interval(struct alvium_dev *alvium,
> > > > > > > -				     struct v4l2_subdev_frame_interval *fi)
> > > > > > > +				     struct v4l2_subdev_frame_interval *fi,
> > > > > > > +				     u64 *req_fr)
> > > > > > >  {
> > > > > > >  	struct device *dev = &alvium->i2c_client->dev;
> > > > > > > -	u64 req_fr, dft_fr, min_fr, max_fr;
> > > > > > > +	u64 dft_fr, min_fr, max_fr;
> > > > > > >  	int ret;
> > > > > > >  
> > > > > > >  	if (fi->interval.denominator == 0)
> > > > > > > @@ -1681,13 +1682,12 @@ static int alvium_set_frame_interval(struct alvium_dev *alvium,
> > > > > > >  	dev_dbg(dev, "fi->interval.denominator = %d\n",
> > > > > > >  		fi->interval.denominator);
> > > > > > >  
> > > > > > > -	req_fr = (u64)((fi->interval.denominator * USEC_PER_SEC) /
> > > > > > > +	*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;
> > > > > > > +	if (*req_fr >= max_fr && *req_fr <= min_fr)
> > > > > > > +		*req_fr = dft_fr;
> > > > > > 
> > > > > > Shouldn't we clamp the value to [min, max] instead of using the default
> > > > > > if it's out of range ? Something like
> > > > > > 
> > > > > > 	*req_fr = clamp(*req_fr, min_fr, max_fr)
> > > > > > 
> > > > > > This makes me realize that the current code is wrong, req_fr can't be >=
> > > > > > max and <= min at the same time. You probably meant || instead of &&.
> > > > > > 
> > > > > > This should be fixed in a separate patch.
> > > > > 
> > > > > If this is ok for you, after this series I can put a patch with || fix
> > > > > instead of clamping, because if we clamp dft_fr is not used any more.
> > > > > After if you agree I will work on clamping.
> > > > > Thanks for the catch! :)
> > > > 
> > > > It's fine to fix this on top of the series, but I don't see why you
> > > > would need to first use ||. You can call clamp() and remove dft_fr.
> > > 
> > > I'm just thinking out loud eh :)
> > > 
> > > Maybe in the future we need to expose fr infos to the user to play with
> > > that. But we are writing for now, then we can replan to readd dft_fr
> > > read later.
> > > 
> > > I think this is what your are suggesting:
> > > 
> > > +++ 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);
> > > 
> > > right?
> > 
> > Yes this looks good to me.
> 
> Just an info.
> Can I proceed to send v3 of this series and then the fix or better to
> wait?

For me you can send v3. Bonus points if you include the above fix in v3
as a patch at the end :-) I think we would then be ready to merge the
whole series.

> > > > > > >  
> > > > > > > -	alvium->fr = req_fr;
> > > > > > >  	alvium->frame_interval.numerator = fi->interval.numerator;
> > > > > > >  	alvium->frame_interval.denominator = fi->interval.denominator;
> > > > > > >  
> > > > > > > @@ -1699,6 +1699,7 @@ static int alvium_s_frame_interval(struct v4l2_subdev *sd,
> > > > > > >  				   struct v4l2_subdev_frame_interval *fi)
> > > > > > >  {
> > > > > > >  	struct alvium_dev *alvium = sd_to_alvium(sd);
> > > > > > > +	u64 req_fr = ALVIUM_DEFAULT_FR_HZ;
> > > > > > 
> > > > > > Do you need to initialize the variable ? It doesn't seem to be required.
> > > > > 
> > > > > Really not, it's just to maintain the logic of alvium->fr. I will drop
> > > > > this in v3, thanks!
> > > > > 
> > > > > > With these small issues fixed,
> > > > > > 
> > > > > > Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
> > > > > > 
> > > > > > >  	int ret;
> > > > > > >  
> > > > > > >  	/*
> > > > > > > @@ -1711,9 +1712,9 @@ static int alvium_s_frame_interval(struct v4l2_subdev *sd,
> > > > > > >  	if (alvium->streaming)
> > > > > > >  		return -EBUSY;
> > > > > > >  
> > > > > > > -	ret = alvium_set_frame_interval(alvium, fi);
> > > > > > > +	ret = alvium_set_frame_interval(alvium, fi, &req_fr);
> > > > > > >  	if (!ret)
> > > > > > > -		ret = alvium_set_frame_rate(alvium);
> > > > > > > +		ret = alvium_set_frame_rate(alvium, req_fr);
> > > > > > >  
> > > > > > >  	return ret;
> > > > > > >  }
> > > > > > > @@ -2273,7 +2274,6 @@ static int alvium_subdev_init(struct alvium_dev *alvium)
> > > > > > >  	/* Setup initial frame interval*/
> > > > > > >  	alvium->frame_interval.numerator = 1;
> > > > > > >  	alvium->frame_interval.denominator = ALVIUM_DEFAULT_FR_HZ;
> > > > > > > -	alvium->fr = ALVIUM_DEFAULT_FR_HZ;
> > > > > > >  
> > > > > > >  	/* Setup the initial mode */
> > > > > > >  	alvium->mode.fmt = alvium_csi2_default_fmt;
> > > > > > > diff --git a/drivers/media/i2c/alvium-csi2.h b/drivers/media/i2c/alvium-csi2.h
> > > > > > > index 17f0bbbd1839..80066ac25047 100644
> > > > > > > --- a/drivers/media/i2c/alvium-csi2.h
> > > > > > > +++ b/drivers/media/i2c/alvium-csi2.h
> > > > > > > @@ -443,7 +443,6 @@ struct alvium_dev {
> > > > > > >  
> > > > > > >  	struct alvium_mode mode;
> > > > > > >  	struct v4l2_fract frame_interval;
> > > > > > > -	u64 fr;
> > > > > > >  
> > > > > > >  	u8 h_sup_csi_lanes;
> > > > > > >  	u64 link_freq;

-- 
Regards,

Laurent Pinchart