Re: [PATCH v4 1/4] media: v4l: Support passing sub-device argument to v4l2_get_link_freq()

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Jacopo,

Thanks for the review.

On Tue, Apr 30, 2024 at 09:01:09AM +0200, Jacopo Mondi wrote:
> Hi Sakari
> 
> On Mon, Apr 29, 2024 at 10:08:49PM +0300, Sakari Ailus wrote:
> > v4l2_get_link_freq() accepts a V4L2 control handler for now, but it needs
> > to take struct v4l2_subdev argument in order to obtain the link frequency

This should have been media_pad actually. I'll update the text for v5 (same
in the subject).

> > using get_mbus_config() pad op. Prepare for this by allowing struct
> > v4l2_subdev as well.
> >
> > Signed-off-by: Sakari Ailus <sakari.ailus@xxxxxxxxxxxxxxx>
> 
> Nice! I need exactly something like this to avoid pre-computing a lot
> a frequencies in a driver I'm working on.
> 
> > ---
> >  drivers/media/v4l2-core/v4l2-common.c | 19 ++++++++++++++++---
> >  include/media/v4l2-common.h           | 14 +++++++++++---
> >  2 files changed, 27 insertions(+), 6 deletions(-)
> >
> > diff --git a/drivers/media/v4l2-core/v4l2-common.c b/drivers/media/v4l2-core/v4l2-common.c
> > index 4165c815faef..7f69b5a025fa 100644
> > --- a/drivers/media/v4l2-core/v4l2-common.c
> > +++ b/drivers/media/v4l2-core/v4l2-common.c
> > @@ -464,8 +464,8 @@ int v4l2_fill_pixfmt(struct v4l2_pix_format *pixfmt, u32 pixelformat,
> >  }
> >  EXPORT_SYMBOL_GPL(v4l2_fill_pixfmt);
> >
> > -s64 v4l2_get_link_freq(struct v4l2_ctrl_handler *handler, unsigned int mul,
> > -		       unsigned int div)
> > +s64 __v4l2_get_link_freq_ctrl(struct v4l2_ctrl_handler *handler,
> > +			      unsigned int mul, unsigned int div)
> >  {
> >  	struct v4l2_ctrl *ctrl;
> >  	s64 freq;
> > @@ -500,7 +500,20 @@ s64 v4l2_get_link_freq(struct v4l2_ctrl_handler *handler, unsigned int mul,
> >
> >  	return freq > 0 ? freq : -EINVAL;
> >  }
> > -EXPORT_SYMBOL_GPL(v4l2_get_link_freq);
> > +EXPORT_SYMBOL_GPL(__v4l2_get_link_freq_ctrl);
> > +
> > +s64 __v4l2_get_link_freq_pad(struct media_pad *pad, unsigned int mul,
> > +			     unsigned int div)
> > +{
> > +	struct v4l2_subdev *sd;
> > +
> > +	sd = media_entity_to_v4l2_subdev(pad->entity);
> > +	if (!sd)
> > +		return -ENODEV;
> > +
> > +	return __v4l2_get_link_freq_ctrl(sd->ctrl_handler, mul, div);
> > +}
> > +EXPORT_SYMBOL_GPL(__v4l2_get_link_freq_pad);
> >
> >  /*
> >   * Simplify a fraction using a simple continued fraction decomposition. The
> > diff --git a/include/media/v4l2-common.h b/include/media/v4l2-common.h
> > index 63ad36f04f72..d7115cd61a38 100644
> > --- a/include/media/v4l2-common.h
> > +++ b/include/media/v4l2-common.h
> > @@ -525,7 +525,8 @@ int v4l2_fill_pixfmt_mp(struct v4l2_pix_format_mplane *pixfmt, u32 pixelformat,
> >  /**
> >   * v4l2_get_link_freq - Get link rate from transmitter
> >   *
> > - * @handler: The transmitter's control handler
> > + * @pad: The transmitter's media pad (or control handler for compatibility
> > + *	 reasons, don't use in new code)
> >   * @mul: The multiplier between pixel rate and link frequency. Bits per pixel on
> >   *	 D-PHY, samples per clock on parallel. 0 otherwise.
> >   * @div: The divisor between pixel rate and link frequency. Number of data lanes
> > @@ -541,8 +542,15 @@ int v4l2_fill_pixfmt_mp(struct v4l2_pix_format_mplane *pixfmt, u32 pixelformat,
> >   * * %-ENOENT: Link frequency or pixel rate control not found
> >   * * %-EINVAL: Invalid link frequency value
> >   */
> > -s64 v4l2_get_link_freq(struct v4l2_ctrl_handler *handler, unsigned int mul,
> > -		       unsigned int div);
> > +#define v4l2_get_link_freq(pad, mul, div)				\
> > +	_Generic(pad,							\
> > +		 struct media_pad *: __v4l2_get_link_freq_pad,		\
> > +		 struct v4l2_ctrl_handler *: __v4l2_get_link_freq_ctrl)	\
> > +	(pad, mul, div)
> 
> That's nice and elegant, but given that I count 10 drivers in mainline
> using this function, wouldn't it be better to make them use a pad and
> remove the ctrl_handler version completely to avoid it being used in
> future ?

It'd be nice to add a function that takes the CSI-2 configuration from the
endpoint as well, doing this manually in drivers is awkward and a lot of
extra code. I'm all for this but I'd do this separately, after which we can
remove the above macro.

Note that for the IVSC driver this is a bugfix.

> 
> > +s64 __v4l2_get_link_freq_pad(struct media_pad *pad, unsigned int mul,
> > +			     unsigned int div);
> > +s64 __v4l2_get_link_freq_ctrl(struct v4l2_ctrl_handler *handler,
> > +			      unsigned int mul, unsigned int div);
> >
> >  void v4l2_simplify_fraction(u32 *numerator, u32 *denominator,
> >  		unsigned int n_terms, unsigned int threshold);

-- 
Kind regards,

Sakari Ailus




[Index of Archives]     [Linux Input]     [Video for Linux]     [Gstreamer Embedded]     [Mplayer Users]     [Linux USB Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]

  Powered by Linux