From mboxrd@z Thu Jan 1 00:00:00 1970 From: Greg Kroah-Hartman Subject: Re: [PATCH 00/13] leds: fix attribute-creation races Date: Thu, 26 Jun 2014 19:25:06 -0400 Message-ID: <20140626232506.GA26949@kroah.com> References: <1403716136-32694-1-git-send-email-johan@kernel.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: Sender: linux-leds-owner@vger.kernel.org To: Bryan Wu Cc: Johan Hovold , Richard Purdie , Dmitry Torokhov , linux-input@vger.kernel.org, lkml , Linux LED Subsystem , Janne Kanniainen , Jiri Kosina , =?iso-8859-1?Q?Bj=F8rn?= Mork List-Id: linux-input@vger.kernel.org On Wed, Jun 25, 2014 at 03:46:19PM -0700, Bryan Wu wrote: > On Wed, Jun 25, 2014 at 10:08 AM, Johan Hovold wrote: > > LED-class device attributes should be created using the attribute groups > > of struct device, rather than be added manually after the device has > > been registered in order to avoid racing with userspace. > > > > The first patch adds an attribute groups field to struct led_classdev, > > which is used during registration to add device attributes in a > > race-free manner. > > > > The remaining patches convert the current racy LED-device attribute > > creations to use the new facility. Hopefully my grep-patterns have > > caught most instances. Note that this also tends to simplify probe error > > paths. > > > > The lp55xx-common failed to remove the attribute group it is currently > > creating, something which is fixed separately. > > > > Note that this series include one driver from drivers/input/keyboard. > > > > The individual led-driver patches have been compile tested only. > > > > Thanks a lot for driving this. I will applied this patchset into my > -devel branch. After I got the Ack from Input guys, I will apply it to > my for-next branch then. Series looks good to me, Johan, thanks for doing this work. greg k-h