From mboxrd@z Thu Jan  1 00:00:00 1970
From: linux@arm.linux.org.uk (Russell King - ARM Linux)
Date: Thu, 22 Jul 2010 10:46:43 +0100
Subject: [RFC,PATCH 3/3] arm: remove machine_desc.io_pg_offst and .phys_io
In-Reply-To: <1279703355.27276.100.camel@pororo.lan>
References: <1279073377.400469.233833413341.0.gpush@pororo>
	<1279073377.404006.432977736300.3.gpush@pororo>
	<AANLkTilsIR36GNlbpZHt3FB5Dp2msedtUMFRSlZ-N-w9@mail.gmail.com>
	<20100720100448.GB32702@n2100.arm.linux.org.uk>
	<1279621292.27276.19.camel@pororo.lan>
	<1279629158.27276.29.camel@pororo.lan>
	<20100721085024.GD6009@n2100.arm.linux.org.uk>
	<1279703355.27276.100.camel@pororo.lan>
Message-ID: <20100722094643.GB31293@n2100.arm.linux.org.uk>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

On Wed, Jul 21, 2010 at 11:09:15AM +0200, Jeremy Kerr wrote:
> Russell,
> 
> > It's extremely unlikely to have a UART at address 0 (because that's by
> > default where the CPU jumps to on reset.)  That's not to say it could
> > never happen - it's possible to hard-wire the CPU into 'hivec' mode
> > where it'll instead jump to 0xffff0000.  In that case, if you're sane
> > you'd put your SDRAM at 0x00000000.
> 
> OK, the patch I have just posted is a bit of a hack that does the 0x0
> check; happy to do this a different way.
> 
> So, the simplest solution seems to be to make the uart mapping code
> dependent on !DEBUG_ICEDCC (but leave the code there to add the mappings
> for ARCH_NETWINDER, ARCH_CATS and ARCH_RPC). Would you like a new patch
> for this?

I think the last patch you posted is fine as-is.  Compiling out the
setup code for the DEBUG_ICEDCC would be an optimization which could
be done though...  If you think it's appropriate, please do so.

Meanwhile, I'm happy enough to merge this patch (or the optimized one)
and the follow-on patch which kills off the initializers.

Out of interest, have the OMAP platforms been compile tested with this?