Re: [patch 04/16] I/O driver for 8250-compatible UARTs

[Tom Rini]

On Wed, Aug 31, 2005 at 10:03:34PM +0100, Russell King wrote:
> On Wed, Aug 31, 2005 at 01:10:39PM -0700, Tom Rini wrote:
> > On Wed, Aug 31, 2005 at 01:38:52PM -0600, Bjorn Helgaas wrote:
> > > On Monday 29 August 2005 10:09 am, Tom Rini wrote:
> > > >  linux-2.6.13-trini/drivers/serial/kgdb_8250.c  |  594 +++++++++++++++++++++
> > > 
> > > The existing stuff in drivers/serial is named "8250_*"; is
> > > there a reason you're using "kgdb_8250" rather than "8250_kgdb"?
> > 
> > All the other kgdb stuff tends to be prefixed, not suffixed.  But I
> > don't really care either way.
> 
> I'd prefer it was 8250_kgdb.c actually - that keeps it along side the
> other 8250 files.

Will do.

> > > > +	switch (CURRENTPORT.iotype) {
> > > > +	case UPIO_MEM:
> > > > +		if (CURRENTPORT.mapbase)
> > > > +			kgdb8250_needs_request_mem_region = 1;
> > > > +		if (CURRENTPORT.flags & UPF_IOREMAP) {
> > > > +			CURRENTPORT.membase = ioport_map(CURRENTPORT.mapbase,
> > > > +						      8 << KGDB8250_REG_SHIFT);
> > > 
> > > Shouldn't this be ioremap instead of ioport_map?
> > 
> > If I remember right from the testing, no.  Or if my memory is wrong and
> > that's retorihcal, sure.
> 
> ioport_map() is supposed to be used to map the IO range for the ioread/
> iowrite operations.  IOW, it takes something compatible with inb() and
> friends and converts it to something compatible with ioread8() and
> friends.
> 
> It does not take a MMIO cookie, so the code above appears to be
> conceptually wrong.
> 

So it's luck (or another mapping I didn't see elsewhere) that this
worked, and it should still be ioremap(...) to use with ioread/write8
later on in the code?

-- 
Tom Rini
http://gate.crashing.org/~trini/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/