Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 23 Jan 2020 12:02:40 +0000
From:   Charles Keepax <ckeepax@opensource.cirrus.com>
To:     Mark Brown <broonie@kernel.org>
CC:     <lee.jones@linaro.org>, <lgirdwood@gmail.com>,
        <linux-kernel@vger.kernel.org>, <patches@opensource.cirrus.com>
Subject: Re: [PATCH RESEND 1/2] regulator: arizona-ldo1: Improve handling of
 regulator unbinding
Message-ID: <20200123120240.GD4098@ediswmail.ad.cirrus.com>
References: <20200122110842.10702-1-ckeepax@opensource.cirrus.com>
 <20200122131149.GE3833@sirena.org.uk>
 <20200123092639.GC4098@ediswmail.ad.cirrus.com>
 <20200123114805.GA5440@sirena.org.uk>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Disposition: inline
In-Reply-To: <20200123114805.GA5440@sirena.org.uk>
User-Agent: Mutt/1.5.21 (2010-09-15)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Jan 23, 2020 at 11:48:05AM +0000, Mark Brown wrote:
> On Thu, Jan 23, 2020 at 09:26:39AM +0000, Charles Keepax wrote:
> 
> > 3) We could look at doing something in regmap IRQ to change when
> > it does PM runtime calls, it is regmap doing runtime gets when
> > drivers remove IRQs that causes the issue. But my accessment was
> > that what regmap is doing makes perfect sense, so I don't think
> > this is a good approach.
> 
> Why do you even care about the errors?  It's not like this device is
> going to get removed in a production system and the primary IRQ will be
> disabled when the core is removed, this is just something that happens
> during development isn't it?

We do have certain customers who test unbinding the driver and
complain when it throws errors. Admittedly, you are correct that
it is a little bit of a stretch to imagine a situation where this
is a massive problem in production. Best I can offer is one of
our CODECs gets into a laptop and someone wants to unbind/bind
the driver, to clear some issue or something.

Its been a while since I was looking at this issue, so I will
double check to see if the issue is purely one of error messages
or if it actually can cause problems (memory leaks or issues
rebinding). But on a personal note I would be somewhat happier if
we could come up with some acceptable way to make the driver
unbind cleanly.

I am more than happy to do the leg work if we really don't like
this solution. Do either you or Lee have any thoughts on my
selective MFD remove helpers? That seemed like the most promising
alternative solution to me.

Thanks,
Charles