From mboxrd@z Thu Jan  1 00:00:00 1970
From: Ilya Yanok <yanok@emcraft.com>
Date: Wed, 11 Feb 2009 22:40:42 +0300
Subject: [U-Boot] [PATCH 2/3] dnet: driver for Dave DNET ethernet
	controller
In-Reply-To: <20090205213201.5A0EA8322908@gemini.denx.de>
References: <1233589490-14293-1-git-send-email-yanok@emcraft.com>
	<1233589490-14293-3-git-send-email-yanok@emcraft.com>
	<20090202202116.CF0BC8322908@gemini.denx.de>
	<498A5F2F.1030603@emcraft.com>
	<20090205213201.5A0EA8322908@gemini.denx.de>
Message-ID: <499329BA.5040807@emcraft.com>
List-Id: <u-boot.lists.denx.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
To: u-boot@lists.denx.de

Dear Wolfgang,

Wolfgang Denk wrote:
>> nothing magical in offsets actually... we usually have offsets in specs
>> not C structures...
>>     
>
> Just to name one reason (but a very important one):  using  a  struct
> allows for strict type checking by the compiler. If you use address +
> offset  you  will  not  get  any  help from the compiler when you for
> example read 8 bit from a register that must be accessed as a 32  bit
> entity.
>   

But this works only if we use "memory-style" IO access, if we use
read/write accessor functions that info we be lost as all pointers we be
silently converted into void*... And using "memory-style" IO access is
strongly discouraged in the kernel...
BTW, what is the general rule for U-Boot? I can see that most accesses
are done "memory-style" but I can see accessor functions used here and
there...

Regards, Ilya.