Re: [PATCH 1/3] media: Provide a helper for setting bus_info field

[Sakari Ailus <sakari.ailus@linux.intel.com>]

Hi Hans,

On Tue, Jan 25, 2022 at 01:54:45PM +0100, Hans Verkuil wrote:
> Hi Sakari,
> 
> On 22/01/2022 17:36, Sakari Ailus wrote:
> > The bus_info or a similar field exists in a lot of structs, yet drivers
> > tend to set the value of that field by themselves in a determinable way.
> > Thus provide a helper for doing this. To be used in subsequent patches.
> > 
> > Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
> > ---
> >  include/media/media-device.h | 30 +++++++++++++++++++++++++++---
> >  1 file changed, 27 insertions(+), 3 deletions(-)
> > 
> > diff --git a/include/media/media-device.h b/include/media/media-device.h
> > index 1345e6da688a..9f0458068196 100644
> > --- a/include/media/media-device.h
> > +++ b/include/media/media-device.h
> > @@ -13,12 +13,13 @@
> >  
> >  #include <linux/list.h>
> >  #include <linux/mutex.h>
> > +#include <linux/pci.h>
> > +#include <linux/platform_device.h>
> >  
> >  #include <media/media-devnode.h>
> >  #include <media/media-entity.h>
> >  
> >  struct ida;
> > -struct device;
> >  struct media_device;
> >  
> >  /**
> > @@ -181,8 +182,7 @@ struct media_device {
> >  	atomic_t request_id;
> >  };
> >  
> > -/* We don't need to include pci.h or usb.h here */
> > -struct pci_dev;
> > +/* We don't need to include usb.h here */
> >  struct usb_device;
> >  
> >  #ifdef CONFIG_MEDIA_CONTROLLER
> > @@ -496,4 +496,28 @@ static inline void __media_device_usb_init(struct media_device *mdev,
> >  #define media_device_usb_init(mdev, udev, name) \
> >  	__media_device_usb_init(mdev, udev, name, KBUILD_MODNAME)
> >  
> > +static inline void
> > +__media_set_bus_info(char *bus_info, size_t bus_info_size, struct device *dev)
> > +{
> > +	if (!dev || *bus_info)
> > +		return;
> > +
> > +	if (dev_is_platform(dev))
> > +		snprintf(bus_info, bus_info_size, "platform:%s", dev_name(dev));
> > +	else if (dev_is_pci(dev))
> > +		snprintf(bus_info, bus_info_size, "PCI:%s", dev_name(dev));
> > +}
> > +
> > +/**
> > + * media_set_bus_info() - Conditionally set bus_info
> > + *
> > + * @bus_info:	Variable where to write the bus info (char array)
> > + * @dev:	Related struct device
> > + *
> > + * Sets bus information based on device conditionally, if the first character of
> > + * &bus_info is not '\0' and dev is non-NULL.
> > + */
> > +#define media_set_bus_info(bus_info, dev) \
> > +	__media_set_bus_info(bus_info, sizeof(bus_info), dev)
> 
> Wouldn't it be simpler to make two #defines:
> 
> #define media_set_bus_info(mdev, dev) \
> 	__media_set_bus_info(mdev->bus_info, sizeof(mdev->bus_info), dev)
> 
> and:
> 
> #define v4l2_cap_set_bus_info(cap, dev) \
> 	__media_set_bus_info(cap->bus_info, sizeof(cap->bus_info), dev)
> 
> That way the sizeof() always works correctly.
> 
> This could also be static inlines to have better type checking, of course.
> 
> Another option is:
> 
> #define media_set_bus_info(s, dev) \
> 	__media_set_bus_info((s)->bus_info, sizeof((s)->bus_info), dev)
> 
> That's more generic, but it does make the assumption that the struct s
> has a field bus_info. Which is a reasonable assumption IMHO.
> 
> I do like the idea of this series.

Thanks for the comments.

I prefer Laurent's suggestion of removing the macro and simply using
sizeof() in the caller. Note that there will be a very small number of
these calls and so far none in the drivers (nor there should be any).

-- 
Kind regards,

Sakari Ailus