Re: [PATCH] MIPS: IP28: Correct IO_BASE in mach-ip28/spaces.h for proper ioremap

[Thomas Bogendoerfer <tsbogend@alpha.franken.de>]

On Tue, Aug 19, 2014 at 10:00:34AM +0200, Ralf Baechle wrote:
> On Mon, Aug 18, 2014 at 10:55:02PM -0400, Joshua Kinard wrote:
> 
> > --- a/arch/mips/include/asm/mach-ip28/spaces.h
> > +++ b/arch/mips/include/asm/mach-ip28/spaces.h
> > @@ -18,7 +18,7 @@
> >  #define PHYS_OFFSET	_AC(0x20000000, UL)
> > 
> >  #define UNCAC_BASE	_AC(0xc0000000, UL)     /* 0xa0000000 + PHYS_OFFSET */
> > -#define IO_BASE		UNCAC_BASE
> > +#define IO_BASE		_AC(0x9000000000000000, UL)
> > 
> >  #include <asm/mach-generic/spaces.h>
> 
> I think the real culprit is not the definition of IO_BASE but of
> UNCAC_BASE.  0xc0000000UL is KSEG2 for a 32 bit kernel - but for a 64 bit
> kernel UNCAC_BASE should be defined as _AC(0x9000000000000000, UL).

that was my first thought as well. I just wondered whether it's necessary
to reflect PHY_OFFSET != 0... nevertheless UNCAC_BASE is not usuable
for memory access on IP28 at all. At least not without poking at the
memory controller for slower access cycles, which the current kernel
avoids totaly.

> Which are the defaults in <asm/mach-generic/spaces.h> so just deleting
> both UNCAC_BASE and IO_BASE from mach-ip28/spaces.h should fix things?

I'll give a spin later.

Thomas.

-- 
Crap can work. Given enough thrust pigs will fly, but it's not necessarily a
good idea.                                                [ RFC1925, 2.3 ]