Sender: Guenter Roeck <groeck7@gmail.com>
Message-ID: <ec4259d3-a7bf-d296-7805-aac3f7480b93@roeck-us.net>
Date:   Tue, 14 Mar 2023 09:42:49 -0700
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.7.1
Subject: Re: [PATCH v2] watchdog: ebc-c384_wdt: Migrate to the regmap API
Content-Language: en-US
To:     William Breathitt Gray <william.gray@linaro.org>,
        Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc:     Wim Van Sebroeck <wim@linux-watchdog.org>,
        linux-watchdog@vger.kernel.org, linux-kernel@vger.kernel.org,
        Paul Demetrotion <pdemetrotion@winsystems.com>,
        techsupport@winsystems.com
References: <20230314152916.185939-1-william.gray@linaro.org>
 <ZBCX0iVwYD4uVxwM@smile.fi.intel.com> <ZBChWA6aGLPZ/2vG@fedora>
From:   Guenter Roeck <linux@roeck-us.net>
In-Reply-To: <ZBChWA6aGLPZ/2vG@fedora>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 3/14/23 09:31, William Breathitt Gray wrote:
> On Tue, Mar 14, 2023 at 05:50:42PM +0200, Andy Shevchenko wrote:
>> On Tue, Mar 14, 2023 at 11:29:16AM -0400, William Breathitt Gray wrote:
>>> The regmap API supports IO port accessors so we can take advantage of
>>> regmap abstractions rather than handling access to the device registers
>>> directly in the driver.
>>
>> ...
>>
>>>   - Utilize watchdog_set_drvdata() and watchdog_get_drvdata()
>>
>> I'm wondering why you can't use dev_get_regmap() instead.
> 
> We can set `wdd->parent = dev` in ebc_c384_wdt_probe(), and then use
> `dev_get_regmap(wdev->parent)` to retrieve the regmap. The only downside
> I see if perhaps the added latency a call to devres_find(), whereas
> using watchdog_get_drvdata() is just a pointer dereference.
> 
> I'm indifferent to either choice, so if Guenter or Wim have a preference
> here I'll follow their decision.
> 

I am not inclined to accept a patch which calls dev_get_regmap() more
than once. It is not just added latency, it is unnecessarily executing
a lot of code. Maybe that call is abused nowadays, and/or maybe people do not
care about wasting CPU cycles anymore, but that is not its intended use case.

>>
>>>   - Map watchdog control registers based on offset 0x1 and adjust regmap
>>>     configurations accordingly; offset 0x0 is unused in this driver so we
>>>     should avoid unnecessary exposure of it
>>
>> I'm wondering what bad could happen if you expose it.
> 
> The WINSYSTEMS EBC-C384 documentation I have does not specify what
> offset 0x0 does (nor offsets 0x3-0x4), so I don't know if there are side
> effects to reading those addresses. Really, I'm just avoiding the hassle
> of writing an explicit precious registers table for those offsets by not
> exposing them at all.
> 

Counter questions:
- What would be the purpose of exposing register addresses if they are not needed ?
- What bad can happen by _not_ exposing those register addresses ?

Guenter