From mboxrd@z Thu Jan  1 00:00:00 1970
From: John.Linn@xilinx.com (John Linn)
Date: Mon, 7 Feb 2011 08:44:11 -0700
Subject: [PATCH 2/4] ARM: Xilinx: Adding Xilinx board support
In-Reply-To: <20110205161853.GC29411@n2100.arm.linux.org.uk>
References: <1296922124-23379-1-git-send-email-john.linn@xilinx.com>
	<1296922124-23379-2-git-send-email-john.linn@xilinx.com>
	<eee52c43-76da-4e47-9f40-4a2d7ee963b6@VA3EHSMHS032.ehs.local>
	<20110205161853.GC29411@n2100.arm.linux.org.uk>
Message-ID: <b3e8c7d4-4db6-41a7-aa06-67c8e327e2aa@VA3EHSMHS005.ehs.local>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

> -----Original Message-----
> From: Russell King - ARM Linux [mailto:linux at arm.linux.org.uk]
> Sent: Saturday, February 05, 2011 9:19 AM
> To: John Linn
> Cc: linux-arm-kernel at lists.infradead.org; kernel at vger.kernel.org;
catalin.marinas at arm.com;
> glikely at secretlab.ca
> Subject: Re: [PATCH 2/4] ARM: Xilinx: Adding Xilinx board support
> 
> On Sat, Feb 05, 2011 at 09:08:42AM -0700, John Linn wrote:
> > +void __init irq_init(void)
> > +{
> > +	gic_cpu_base_addr = (void __iomem *)SCU_GIC_CPU_BASE;
> 
> Just pass this address into gic_init - there's no need to initialize
> this variable yourself anymore.
> 
> > +	gic_init(0, 29, (void __iomem *)SCU_GIC_DIST_BASE,
gic_cpu_base_addr);
> ...
> > +static struct map_desc io_desc[] __initdata = {
> ...
> > +#ifdef CONFIG_DEBUG_LL
> > +	{
> > +		.virtual	= UART0_BASE,
> > +		.pfn		= __phys_to_pfn(UART0_BASE),
> > +		.length		= SZ_4K,
> > +		.type		= MT_DEVICE,
> > +	},
> > +#endif
> 
> You shouldn't need to do this if your addruart macro in
mach/debug-macros.S
> is correctly returning the virtual and physical addresses in rp and
rv.

Hi Russell,

It's interesting as I'm looking at this further.  Sure enough it looks
to me like any 
early mappings in the MMU get destroyed.  Paging_init calls
devicemaps_init which says 
it destroys the early debug mappings.  Maybe I'm misinterpreting that?

I notice that the mach-tegra platform does have a mapping in the table
for iotable_init
but it's not conditional on the DEBUG_LL and it covers the APB which
includes the UART 
for early debug.

Sorry for my slowness, but it still seems like the early MMU mappings go
away. What am I 
not seeing that I should be?

Thanks, appreciate your patience and help
John




This email and any attachments are intended for the sole use of the named recipient(s) and contain(s) confidential information that may be proprietary, privileged or copyrighted under applicable law. If you are not the intended recipient, do not read, copy, or forward this email message or any attachments. Delete this email message and any attachments immediately.