Re: timer: permit statically-declared work with deferrable timers

[Phil Carmody <ext-phil.2.carmody@nokia.com>]

On 30/09/10 22:08 +0200, ext Thomas Gleixner wrote:
> On Thu, 30 Sep 2010, Phil Carmody wrote:
> 
> > 
> > Arjen, Thomas,
> > 
> > This patch is an enabler which hopefully can lead to simplification of code 
> > elsewhere. For example, it would turn Artem's patch I refer to in the commit
> > message (8eab945c5616fc984e97b922d6a2559be93f39a1) into just:
> > 
> > -static DECLARE_DELAYED_WORK(cache_cleaner, do_cache_clean);
> > +static DECLARE_DEFERRED_WORK(cache_cleaner, do_cache_clean);
> > 
> > rather than the creation of a __init helper that required touching 3 files.
> > 
> > I'm prepared to follow up with such simplifying patches if it's considered
> > worthwhile.
> 
> If it simplies stuff, no objections from my side. The patch looks
> reasonable enough.
> 
> How many (ab)users will it clean up ?

Not a large number; absolute tops - a dozen. I've not investigated any apart from
the couple that Artem addressed initially.

And it's OK, the clients weren't the _ab_users. If anything we're the abusers for 
using sleight-of-hand to hide flags inside pointer values. (Which the C standard
and GCC tried hard to stop us doing. Not hard enough, though ;-) .)

My main intention is to make future patches like Artem's trivial. Change the 
type - change one line. Life's too short to headscratch and fight compilers.

Cheers,
Phil