From mboxrd@z Thu Jan  1 00:00:00 1970
From: dinh.linux@gmail.com (Dinh Nguyen)
Date: Wed, 30 Apr 2014 17:06:28 -0500
Subject: [PATCH] arm: socfpga: reserve the region at start of phys mem
In-Reply-To: <20140430173220.GZ5546@joshc.qualcomm.com>
References: <1392744897-6384-1-git-send-email-bigeasy@linutronix.de>
 <20140430161717.GY5546@joshc.qualcomm.com> <53612C70.6000109@linutronix.de>
 <20140430173220.GZ5546@joshc.qualcomm.com>
Message-ID: <536173E4.2010404@gmail.com>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

Hi Sebastian,

On 4/30/14 12:32 PM, Josh Cartwright wrote:
> On Wed, Apr 30, 2014 at 07:01:36PM +0200, Sebastian Andrzej Siewior wrote:
>> On 04/30/2014 06:17 PM, Josh Cartwright wrote:
>>> Is there a particular reason why you aren't describing this reservation
>>> in devicetree using a /memreserve/ node?
>> Because the kernel is using memory at a specific address it did not
>> allocate. If the memory address where the second CPU come up could be
>> set then this node wouldn't be required and a simple kmalloc() would do
>> it, too.
> I understand the need to reserve the area, but questioning the mechanism
> used to do so.  How is the socfpga case different from Highbank, for
> example, which makes use of /memreserve/ for what appears to be a very
> similar purpose (see arch/arm/boot/dts/highbank.dts):
>
> 	/* First 4KB has pen for secondary cores. */
> 	/memreserve/ 0x00000000 0x0001000;
>
I apologize but it looks like I accidently dropped this patch. But in
light of Josh's
comments, I think doing the /memreserve/ is the correct way.

Dinh