Re: [PATCH] memory ordering in __kfifo primitives

["Paul E. McKenney" <paulmck@us.ibm.com>]

On Thu, Aug 10, 2006 at 01:48:02AM -0400, Mike Christie wrote:
> Paul E. McKenney wrote:
> > OK, it appears that we are even.  I forgot to attach the promised
> > analysis of the callers to __kfifo_put() and __kfifo_get(), and
> > the open-iscsi@googlegroups.com email address listed as maintainer
> > in drivers/scsi/libiscsi.c bounces complaining that, as a non-member,
> > I am not allowed to send it email.  ;-)
> 
> Sorry about that. I do not have any control over the email list. I will
> change the maintainer info entry to indicate that users should just send
> mail to me or linux-scsi.

Sounds good!

> > Anyway, this time the analysis really is attached, sorry for my confusion!
> 
> We have change the code a little since the analysis was made. But, it
> does not really matter much now. I am fine with us just grabbing the
> session lock or xmitmitex (I will send a patch and test it as well) if
> your barrier patch is not accepted. We grab the session lock in the fast
> path now so there is not much benefit left for us.

I am happy to go either way -- the patch with the memory barriers
(which does have the side-effect of slowing down kfifo_get() and
kfifo_put(), by the way), or a patch removing the comments saying
that it is OK to invoke __kfifo_get() and __kfifo_put() without
locking.

Any other thoughts on which is better?  (1) the memory barriers or
(2) requiring the caller hold appropriate locks across calls to
__kfifo_get() and __kfifo_put()?

							Thanx, Paul