From mboxrd@z Thu Jan  1 00:00:00 1970
From: "Cousson, Benoit" <b-cousson@ti.com>
Subject: Re: [PATCHv3 3/17] dmtimer: add omap2420 hwmod database
Date: Mon, 4 Oct 2010 09:50:17 +0200
Message-ID: <4CA98739.9040309@ti.com>
References: <1285059116-26366-1-git-send-email-tarun.kanti@ti.com> <4CA4FC22.9030806@ti.com> <alpine.DEB.2.00.1010011623020.22941@utopia.booyaka.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="ISO-8859-1"; format=flowed
Content-Transfer-Encoding: 7bit
Return-path: <linux-omap-owner@vger.kernel.org>
Received: from devils.ext.ti.com ([198.47.26.153]:47411 "EHLO
	devils.ext.ti.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1753136Ab0JDHu0 (ORCPT
	<rfc822;linux-omap@vger.kernel.org>); Mon, 4 Oct 2010 03:50:26 -0400
In-Reply-To: <alpine.DEB.2.00.1010011623020.22941@utopia.booyaka.com>
Sender: linux-omap-owner@vger.kernel.org
List-Id: linux-omap@vger.kernel.org
To: Paul Walmsley <paul@pwsan.com>
Cc: "DebBarma, Tarun Kanti" <tarun.kanti@ti.com>, "linux-omap@vger.kernel.org" <linux-omap@vger.kernel.org>, "Gopinath, Thara" <thara@ti.com>, "Basak, Partha" <p-basak2@ti.com>, Kevin Hilman <khilman@deeprootsystems.com>, Tony Lindgren <tony@atomide.com>

Hi Paul,

On 10/2/2010 12:25 AM, Paul Walmsley wrote:
> On Thu, 30 Sep 2010, Cousson, Benoit wrote:
>
>> On 9/21/2010 10:51 AM, DebBarma, Tarun Kanti wrote:
>>
>>>    #include "omap_hwmod_common_data.h"
>>>
>>>    #include "prm-regbits-24xx.h"
>>> @@ -121,6 +123,614 @@ static struct omap_hwmod omap2420_l4_wkup_hwmod = {
>>>           .omap_chip      = OMAP_CHIP_INIT(CHIP_IS_OMAP2420),
>>>           .flags          = HWMOD_NO_IDLEST,
>>>    };
>>> +/* Timer Common */
>>> +static char *timer_clk_src_names[] = {
>>> +       "sys_ck",
>>> +       "func_32k_ck",
>>> +       "alt_ck",
>>> +       NULL,
>>> +};
>>
>> I have an issue with that, because this is a pure duplication of the clock_sel
>> information already contained in the clock data:
>>
>> static const struct clksel omap24xx_gpt_clksel[] = {
>>          { .parent =&func_32k_ck, .rates = gpt_32k_rates },
>> 	{ .parent =&sys_ck,      .rates = gpt_sys_rates },
>>        	{ .parent =&alt_ck,      .rates = gpt_alt_rates },
>> 	{ .parent = NULL },
>> };
>>
>> And duplicating the same information somewhere else is most of the time a bad
>> idea.
>
> Yep, there's no way that info should be in the hwmod data, in the current
> setup.  It belongs in the clkdev tables.  Example below.
>
>> That being said... I don't really know how to handle that properly :-)
>>
>> We have to find a better way to select the proper source clock in a soc
>> independent way.
>>
>> Maybe Paul will have some idea?
>
> Here's how it's done:
>
>     http://marc.info/?l=linux-omap&m=128596931017785&w=2
>
> and
>
>     http://marc.info/?l=linux-omap&m=128596931417805&w=2

The famous clock alias... I don't know why but I always forgot that 
solution each time I have such concern:-(
This is indeed the very clean and cool way to do that clock selection.
We can even remove this #define to identified them and use the clock 
string name directly.

Thanks Paul,
Benoit