Re: [RFC 2/2] PM: Lock all devices during suspend/hibernate

["Rafael J. Wysocki" <rjw@sisk.pl>]

On Wednesday, 1 August 2007 19:58, Alan Stern wrote:
> On Wed, 1 Aug 2007, Rafael J. Wysocki wrote:
> 
> > On Wednesday, 1 August 2007 16:11, Alan Stern wrote:
> > > On Wed, 1 Aug 2007, Rafael J. Wysocki wrote:
> > > 
> > > > > +static DECLARE_RWSEM(device_registration_rwsem);
> > > > 
> > > > Is it only intended for device registration, or can it be used in some other
> > > > code paths too?
> > > 
> > > The patch uses it only for registration, since that's where it is 
> > > needed.  I suppose it could be used on other paths as well, if anybody 
> > > wanted to make something else mutually exclusive with suspend.  
> > > (Although then it probably ought to be renamed...)
> > 
> > That was exactly my thought.  Why don't we give it a more general name from
> > the start?
> 
> Like pm_suspend_rwsem?  And change the accessor functions to 
> pm_block_suspend() and pm_allow_suspend()?  (Although it applies to 
> quiescing as well as suspending.)

I'd prefer pm_sleep_rwsem and, respectively, pm_sleep_lock() and
pm_sleep_unlock(), so that it corresponds to the new configuration option
(CONFIG_PM_SLEEP == (CONFIG_SUSPEND || CONFIG_HIBERNATION)).

Greetings,
Rafael


-- 
"Premature optimization is the root of all evil." - Donald Knuth