From mboxrd@z Thu Jan  1 00:00:00 1970
From: Aneesh V <aneesh@ti.com>
Date: Sat, 25 Jun 2011 17:35:44 +0530
Subject: [U-Boot] [PATCH v2 13/22] omap4: add clock support
In-Reply-To: <20110517214432.F3D9F1491B0A@gemini.denx.de>
References: <1298893591-17636-1-git-send-email-aneesh@ti.com>
	<1305472900-4004-14-git-send-email-aneesh@ti.com>
	<20110515190029.7179C1491B06@gemini.denx.de>
	<4DD27893.5000606@ti.com>
	<20110517214432.F3D9F1491B0A@gemini.denx.de>
Message-ID: <4E05CF18.4020800@ti.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,

On Wednesday 18 May 2011 03:14 AM, Wolfgang Denk wrote:
> Dear Aneesh V,
>
[snip ...]
>>> NAK.  We do not accept base address plus offset notation.  Please
>>> declare C structs instead.
>>>
>>
>> Ok. will do.
>>
>> Again just curious, what's the reasoning behind this policy? Is it just
>> aesthetics or something more?
>
> It's more.  When using base+offset, the compiler will know nothing
> about the data type, so it canot warn you if you use a 32 bit accessor
> on a 8 bit register, or vice versa.  When you pass a struct element,
> it can check type information.

Is this really valid in our context? readl()/writel() and friends just
cast the address to the target type pointer. I tried writeb() on my 32
bit register represented using the struct approach and the compiler
didn't give any warning (as expected)!

best regards,
Aneesh