Subject: Re: [PATCH v3 1/5] reset: add reset-simple to unify socfpga, stm32,
 sunxi, and zx2967
To: =?UTF-8?Q?Andreas_F=c3=a4rber?= <afaerber@suse.de>,
        Philipp Zabel <p.zabel@pengutronix.de>
Cc: linux-kernel@vger.kernel.org, Alexandru Gagniuc <alex.g@adaptrum.com>,
        Maxime Coquelin <mcoquelin.stm32@gmail.com>,
        Alexandre Torgue <alexandre.torgue@st.com>,
        Maxime Ripard <maxime.ripard@free-electrons.com>,
        Chen-Yu Tsai <wens@csie.org>, Baoyou Xie <baoyou.xie@linaro.org>,
        Eugeniy Paltsev <Eugeniy.Paltsev@synopsys.com>,
        Steffen Trumtrar <s.trumtrar@pengutronix.de>,
        Dinh Nguyen <dinguyen@kernel.org>,
        linux-arm-kernel@lists.infradead.org, kernel@pengutronix.de
References: <20170816094701.30678-1-p.zabel@pengutronix.de>
 <20170816094701.30678-2-p.zabel@pengutronix.de>
 <51db563d-d6e1-18f1-01c1-3fdcf8269773@arm.com>
 <d6b20df2-12f7-68b1-f7de-0c7f205eba13@suse.de>
 <1502896293.7596.7.camel@pengutronix.de>
 <706d596c-a444-dcc9-a210-02d3564f3e83@suse.de>
From: Andre Przywara <andre.przywara@arm.com>
Message-ID: <ac9d55ae-46f4-6f5c-d2d2-146e4af85250@arm.com>
Date: Wed, 16 Aug 2017 17:46:17 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101
 Thunderbird/52.2.1
MIME-Version: 1.0
In-Reply-To: <706d596c-a444-dcc9-a210-02d3564f3e83@suse.de>
Content-Type: text/plain; charset=utf-8
Content-Language: en-GB
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2289
Lines: 58

Hi,

On 16/08/17 17:41, Andreas Färber wrote:
> Am 16.08.2017 um 17:11 schrieb Philipp Zabel:
>> On Wed, 2017-08-16 at 14:12 +0200, Andreas Färber wrote:
>>> Am 16.08.2017 um 13:30 schrieb Andre Przywara:
>>>> On 16/08/17 10:46, Philipp Zabel wrote:
>>>>> +/**
>>>>> + * struct reset_simple_devdata - simple reset controller properties
>>>>> + * @active_low: if true, bits are cleared to assert the reset. Otherwise, bits
>>>>> + *              are set to assert the reset.
>>>>> + */
>>>>> +struct reset_simple_devdata {
>>>>>>>> +	bool active_low;
>>>>> +};
>>>>> +
>>>>> +static const struct reset_simple_devdata reset_simple_active_low = {
>>>>>>>> +	.active_low = true,
>>>>> +};
>>>>> +
>>>>> +static const struct of_device_id reset_simple_dt_ids[] = {
>>>>>>>> +	{ .compatible = "allwinner,sun6i-a31-clock-reset",
>>>>> +		.data = &reset_simple_active_low },
>>>>
>>>> Can we have a additional generic compatible string here? New users of
>>>> this driver then wouldn't need to explicitly enter their name into the
>>>> driver, but could just use the generic name as a fallback. This would
>>>> enable the driver without any Linux code change just by adding a DT node.
>>>>
>>>> 	compatible = "nexell,s5p6818-reset", "simple-reset";
>>>>
>>>> Whenever we need a quirk (now or in the future), we can add the specific
>>>> name into this structure along with the required workarounds.
>>>
>>> Same question about binding here. However the way it is done today, we
>>> would also need some optional active-low property then or two different
>>> compatible strings, as this is currently controlled via the DT matches.
>>
>> I'd like to decouple this from the issue at hand, which is de-
>> duplicating simple reset code without device tree changes.
>>
>> I'll make a separate suggestion for a simple binding on top of this
>> series.
> 
> Okay, I'll continue using my own driver then, until it's clear how
> exactly I'm supposed to piggy-back on this new driver.

It's not very nice, but if you can't wait, what about using:

	compatible = "your_vendor,your-reset",
		     "allwinner,sun6i-a31-clock-reset";

for now, then later adding the new generic compatible string at the end,
once we agreed on the naming?
That should give you support for all ranges of kernels.

Cheers,
Andre.