serial_ns16550 driver question

[Sascha Hauer]

On Thu, Jul 21, 2011 at 05:12:24AM +0200, Jean-Christophe PLAGNIOL-VILLARD wrote:
> On 22:31 Wed 20 Jul     , Sascha Hauer wrote:
> > On Wed, Jul 20, 2011 at 11:51:04PM +0400, Antony Pavlov wrote:
> > > Hi!
> > > 
> > > In include/ns16550.h we have:
> > > 
> > > struct NS16550_plat {
> > >         unsigned int clock;
> > >         unsigned char f_caps;
> > >         /**
> > >          * register read access capability
> > >          */
> > >         unsigned int (*reg_read) (unsigned long base, unsigned char reg_offset);
> > >         /**
> > >          * register write access capability
> > >          */
> > >         void (*reg_write) (unsigned int val, unsigned long base,
> > >                                     unsigned char reg_offset);
> > > };
> > > 
> > > Why reg_read and reg_write's argument base has type unsigned long?
> > > IMHO pointer type (void * or char *)  is more natural.
> > 
> > You're right, I also wondered about this recently when looking at a
> > patch on the list. I would also rather see void __iomem *. The problem
> > is that this driver is also used on X86 which uses inb/outb which take
> > an integer argument.
> > The Linux driver works around this by having a membase and a iobase
> > field along with different register accessors. While being cleaner
> > I'm unsure if we want to go this way.
> 
> we can do it via resoure and the flasgs IORESOURCE_MEM_8BIT,
> IORESOURCE_MEM_168BITi etc..

Yes, but this does not solve the problem with different accessor
functions (and their argument type).

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 |