[PATCH 1/2] Introduce device_platform_driver() macro

[Sascha Hauer]

On Tue, Feb 12, 2013 at 10:57:09AM +0400, Alexander Shiyan wrote:
> Hello.
> 
> > On Fri, Feb 08, 2013 at 10:07:01PM +0400, Alexander Shiyan wrote:
> > > device_platform_driver() - Helper macro for drivers that don't do
> > > anything special in module registration. This eliminates a lot of
> > > boilerplate. Driver registration will called on device_initcall.
> > > 
> > > Signed-off-by: Alexander Shiyan <shc_work at mail.ru>
> > 
> > Good move ;)
> > 
> > > ---
> > >  include/driver.h | 12 ++++++++++++
> > >  1 file changed, 12 insertions(+)
> > > 
> > > diff --git a/include/driver.h b/include/driver.h
> > > index 31f5d69..d8ac66c 100644
> > > --- a/include/driver.h
> > > +++ b/include/driver.h
> > > @@ -390,6 +390,18 @@ extern struct list_head bus_list;
> > >  extern struct bus_type platform_bus;
> > >  
> > >  int platform_driver_register(struct driver_d *drv);
> > > +
> > > +/* device_platform_driver() - Helper macro for drivers that don't do
> > > + * anything special in module registration. This eliminates a lot of
> > > + * boilerplate. Each module may only use this macro once.
> > > + */
> > > +#define device_platform_driver(drv)				\
> > > +	static int __init device_platform_driver_init(void)	\
> > > +	{							\
> > > +		return platform_driver_register(&drv);		\
> > > +	}							\
> > > +	device_initcall(device_platform_driver_init)
> > > +
> > 
> > Can we use drv##_init as name for the function instead? This way the
> > driver name is in the function name and objdumps and backtraces would
> > have a bit more context.
> 
> OK. Maybe we should rename the names of driver_d structures to better
> understand what is what is?
> In this case, the procedure in the dump would look like;
> __initcall_platform_bus0
> __initcall_cfi_driver10
> __initcall_smc911x_device10
> and so on. I.e. explicitly specify a function (bus, driver, device).
> I can do it in a separate patch if it looks OK.

I think this is not necessary. I just grepped through the existing
driver structs and most of them seem to have a meaningful (at least
unique) name. I just don't wanted to end up with tons of functions
having the same name.

> 
> Besides, as far as I know currently we are not using __init section.
> Can we define this one in the linker script and release this memory after
> initialization is complete?

What would you do then with the memory afterwards? The tlsf allocator
supports memory pools, so in theory we could use it, but adding support
for this to barebox would probably eat up more space in the binary than
you could safe with the few initcalls.

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |