[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <9d5788ab-4576-08c3-e347-9348aba3804a@metafoo.de>
Date: Tue, 21 Dec 2021 15:12:05 +0100
From: Lars-Peter Clausen <lars@...afoo.de>
To: Uwe Kleine-König <u.kleine-koenig@...gutronix.de>
Cc: Alexandre Torgue <alexandre.torgue@...s.st.com>,
David Lechner <david@...hnology.com>,
Fabrice Gasnier <fabrice.gasnier@...s.st.com>,
Jarkko Nikula <jarkko.nikula@...ux.intel.com>,
Kamel Bouhara <kamel.bouhara@...tlin.com>,
Maxime Coquelin <mcoquelin.stm32@...il.com>,
Oleksij Rempel <linux@...pel-privat.de>,
Patrick Havelange <patrick.havelange@...ensium.com>,
Syed Nayyar Waris <syednwaris@...il.com>,
William Breathitt Gray <vilhelm.gray@...il.com>,
linux-arm-kernel@...ts.infradead.org, linux-iio@...r.kernel.org,
linux-kernel@...r.kernel.org,
linux-stm32@...md-mailman.stormreply.com
Subject: Re: [PATCH 0/8] counter: Remove struct counter_device::priv
On 12/21/21 2:22 PM, Uwe Kleine-König wrote:
> On Tue, Dec 21, 2021 at 01:04:50PM +0100, Lars-Peter Clausen wrote:
>> On 12/21/21 12:35 PM, Uwe Kleine-König wrote:
>>> On Tue, Dec 21, 2021 at 12:12:12PM +0100, Lars-Peter Clausen wrote:
>>>> On 12/21/21 11:45 AM, Uwe Kleine-König wrote:
>>>>> similar to patch
>>>>> https://lore.kernel.org/r/4bde7cbd9e43a5909208102094444219d3154466.1640072891.git.vilhelm.gray@gmail.com
>>>>> the usage of struct counter_device::priv can be replaced by
>>>>> container_of which improves type safety and code size.
>>>>>
>>>>> This series depends on above patch, converts the remaining drivers and
>>>>> finally drops struct counter_device::priv.
>>>> Not sure if this is such a good idea. struct counter_device should not be
>>>> embedded in the drivers state struct in the first place.
>>> Just to mention it: My patch series didn't change this, this was already
>>> broken before.
>> I know, but this series has to be reverted when the framework is fixed.
> All drivers have to be touched. With my patch series you have to modify
> one function in each driver, without my patch you have touch nearly
> every function.
>
I'm not so sure. I don't see how you have to modify every function.
You'd keep using priv to get a pointer to the state struct.
That said having a centralized function in each driver to get the state
struct from the counter device doesn't hurt either.
Powered by blists - more mailing lists