Re: [RFC PATCH 0/19] A kernel-level configfs enabled generic target engine for Linux v2.6.32

["Nicholas A. Bellinger" <nab@linux-iscsi.org>]

On Fri, 2009-09-11 at 20:17 -0700, Daniel Walker wrote:
> On Fri, 2009-09-11 at 20:08 -0700, Nicholas A. Bellinger wrote:
> 
> 
> > 
> > Ok, I will have a look at converting the HBA semaphore to a mutex in
> > patch 5, and use some single use waitqueues in patch 16.  Btw, the ones
> > using init_MUTEX_LOCKED() in the latter source file are being used when
> > shutting down target_core_mod allocated kernel threads to sychronize
> > between the stopping processing threads and configfs process context
> > that waitqueues would work better for.  Thanks for pointing this out!
> 
> You could also use completions in patch 16 if the waitqueues turn out to
> be not what your looking for. Completion can sometimes be a find&replace
> substitute for the locked semaphores.

Sorry, include/linux/complete.h is what I was thinking here too..

> 
> 
> > I was thinking about going through the typedefs with a sed chainsaw, but
> > I decided against for this TCM patch series because it makes keeping
> > backward ports to <= v2.6.31 slightly easier to maintain for me for the
> > current users of LIO kernel code.
> > 
> > However, I have been starting to not use typedef for the newest ALUA and
> > pieces of PR code, and I would be happy to have another look at typedef
> > style comments on this once the other larger interest items are on the
> > way to being resolved.
> 
> It's up to you, but you should schedule the typedefs for removal at some
> point.
> 

Ok, I will do that.

Many thanks for your most valuable of time,

--nab

> Daniel
>