From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Morton Date: Mon, 10 Jun 2013 23:31:34 +0000 Subject: Re: [PATCH] backlight: add CONFIG_PM_SLEEP to suspend/resume functions Message-Id: <20130610163134.1c4124abba64c5f7099cb4ff@linux-foundation.org> List-Id: References: <000201ce631f$d4ad0d50$7e0727f0$@samsung.com> <1944274.r5ID26zhWd@wuerfel> In-Reply-To: <1944274.r5ID26zhWd@wuerfel> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Arnd Bergmann Cc: Jingoo Han , 'Shuah Khan' , linux-fbdev@vger.kernel.org, linux-kernel@vger.kernel.org, shuahkhan@gmail.com, rpurdie@rpsys.net, FlorianSchandinat@gmx.de, plagnioj@jcrosoft.com, tomi.valkeinen@ti.com, rafael.j.wysocki@intel.com On Fri, 07 Jun 2013 12:02:31 +0200 Arnd Bergmann wrote: > On Friday 07 June 2013 10:39:20 Jingoo Han wrote: > > Add CONFIG_PM_SLEEP to suspend/resume functions to fix the following > > build warning when CONFIG_PM_SLEEP is not selected. This is because > > sleep PM callbacks defined by SIMPLE_DEV_PM_OPS are only used when > > the CONFIG_PM_SLEEP is enabled. > > > > drivers/video/backlight/backlight.c:211:12: warning: 'backlight_suspend' defined but not used [-Wunused-function] > > drivers/video/backlight/backlight.c:225:12: warning: 'backlight_resume' defined but not used [-Wunused-function] > > > > Signed-off-by: Jingoo Han > > --- > > drivers/video/backlight/backlight.c | 2 ++ > > 1 file changed, 2 insertions(+) > > Your patch looks ok, but I find it extremely annoying to have new warnings > like this one come up every single day in linux-next. It really shouldn't > be this hard to use a macro called SIMPLE_DEV_PM_OPS() correctly. > > Below is an implementation of SIMPLE_DEV_PM_OPS and UNIVERSAL_DEV_PM_OPS > that avoids this issue by introducing an unused reference to the suspend > and resume functions. gcc is smart enough to leave out that unused code > by itself, and it would actually improve compile-time coverage to have > something like this, besides being harder to misuse. > > This would be a better approach if we didn't already have all the "#ifdef > CONFIG_PM_SLEEP" in place that hide the functions now. Unfortunately we > already have over 300 uses of SIMPLE_DEV_PM_OPS/UNIVERSAL_DEV_PM_OPS > in the kernel today, so removing all the #ifdef atomically without > creating more build errors is rather hard to do. > > Maybe someone has an idea how to extend my approach so it works with > and without the #ifdef, to let us transition to a situation that no > longer needs them. You could create new macros, and add a checkpatch rule to remind people to not use the old ones. Then people can migrate over from the old macros at a leisurely pace. The problem will be in thinking up decent names for the new macros.