> > > > ../../../../arch/hpcarm/dev/sed_saip.c:90: #error "define btop, ptob."
> > > > 
> > > > Here is the code:
> > > > 
> > > > #if defined __mips__ || defined __sh__ || defined __arm__
> > > > #define __BTOP(x)               ((paddr_t)(x) >> PGSHIFT)
> > > > #define __PTOB(x)               ((paddr_t)(x) << PGSHIFT)
> > > > #else 
> > > > #error "define btop, ptob."
> > > > #endif
> > > 
> > > Hmm, is it OK for code to check against compiler provided macros (see
> > > below) instead of relying on MACHINE (or whatever kernel-provided value
> > > appropriate)?
> > 
> > What about just killing this #if? This code will not ever be built for
> > mips or sh, it is an arm only file.

>Hmm, right. Following patch does the job (the same change goes for
>ipaq_lcd.c, for consistency). Now I notice ipaq_lcd.c does not get
>built, I'm going to have a look... JORNADA720 built fine.

[patch snipped]

Why not use bus_space_mmap() ?

The file is under hpcarm, but the SED1356 is used in a fair number
of embedded designs. It would be useful IMHO to have a mi driver
for this with hpc and hpcarm attachments.

Robert Swindells