[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20171104170548.vkkgp363k2piyufx@earth>
Date: Sat, 4 Nov 2017 18:05:48 +0100
From: Sebastian Reichel <sebastian.reichel@...labora.co.uk>
To: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc: Andrey Smirnov <andrew.smirnov@...il.com>,
linux-kernel@...r.kernel.org, linux-serial@...r.kernel.org,
Rob Herring <robh@...nel.org>, cphealy@...il.com,
Guenter Roeck <linux@...ck-us.net>,
Lucas Stach <l.stach@...gutronix.de>,
Nikita Yushchenko <nikita.yoush@...entembedded.com>,
Lee Jones <lee.jones@...aro.org>, Pavel Machek <pavel@....cz>,
Andy Shevchenko <andy.shevchenko@...il.com>,
Johan Hovold <johan@...nel.org>
Subject: Re: [PATCH v10 1/5] serdev: Make .remove in struct
serdev_device_driver optional
Hi Greg,
On Sat, Nov 04, 2017 at 12:24:31PM +0100, Greg Kroah-Hartman wrote:
> On Tue, Oct 31, 2017 at 09:36:52AM -0700, Andrey Smirnov wrote:
> > Using devres infrastructure it is possible to write a serdev driver
> > that doesn't have any code that needs to be called as a part of
> > .remove. Add code to make .remove optional.
>
> What about manual unbind from userspace through sysfs? You need to have
> a remove function. All drivers need that, to not have it is pretty lazy :)
Resources,that have been requested via devres are free'd by the
devres framework after the driver specific remove function. This
also works for manual unbind.
If *all* driver resources are allocated using devres, you will end
up with an empty remove function. In that case it makes sense to
remove it completly. This is what quite a few (mainline) i2c and
spi drivers actually do.
-- Sebastian
Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)
Powered by blists - more mailing lists