re: apne

[<cah...@po...>]

On Mon, 23 Dec 2002, Geert Uytterhoeven wrote:
> On Mon, 23 Dec 2002, [iso-8859-2] Krystian Bac=B3awski wrote:
> > On Mon, 23 Dec 2002, Geert Uytterhoeven wrote:
> > > On Mon, 23 Dec 2002, [iso-8859-2] Krystian Bac=B3awski wrote:
> > > > As far as I know apne driver in current version of kernel makes oop=
s.
> > > >=20
> > > > I found out that apne driver in 2.4.20 kernel was modified by someo=
ne who
> > > > knew nothing about amiga (he/she cut out a lot of code with GAYLE s=
ymbols
> > > > and replaced readb/writeb with inb/outb). I replaced apne.c from cu=
rrent
> > > > cvs tree with old one from 2.4.18-cvs. As for me it works.
> > >=20
> > > PCMCIA is some sort of ISA, so inb() and friends should be OK.
> > > Most probably the ISA changes in the m68k-specific part of the tree w=
eren't
> > > ported to APUS.
> >=20
> > Of course should be ok, but someone also cut GAYLE_OFFSET and friends. =
Is
> > amiga pcmcia driver has anything to do with ISA ? Amiga 1200 doesn't ha=
ve
> > ISA. PCMCIA is serviced by GAYLE chip.
>=20
> Well, but PCMCIA is just some form of ISA. In the Linux/m68k kernels with=
 Gayle
> PCMCIA support, inb() and friends work on the PCMCIA bus.
>=20
> Please take a look at include/asm-m68k/io.h.

Hmm... Yes, writeb/outb, readb/inb in asm-m68k/io.h are the same but
those in asm-ppc/io.h are different:

#define inb(port) in_8((u8 *)((port)+_IO_BASE))
#define outb(val,port) out_8((u8 *)((port)+_IO_BASE), (val))
#define readb(addr) in_8((volatile u8 *)(addr))
#define writeb(b,addr) out_8((volatile u8*)(addr),(b))

As the _IO_BASE is defined as zero in/read out/write are almost the same,
but what with "volatile" definition ?

I'll rework driver so that inb/outb won't be replaced by readb/writeb.

Regards
Krystian Baclawski

=0D=0A=0D=0A***************r-e-k-l-a-m-a**************=0D=0A=0D=0AMasz do=B6=E6 p=B3acenia prowizji bankowi ?=0D=0AmBank - za=B3=F3=BF konto=0D=0Ahttp://epieniadze.onet.pl/mbank