Re: [PATCH RFC ticketlock] Auto-queued ticketlock

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

On Mon, Jun 10, 2013 at 07:02:56PM -0400, Steven Rostedt wrote:
> On Sun, 2013-06-09 at 12:36 -0700, Paul E. McKenney wrote:
> 
> > +/*
> > + * Return a pointer to the queue header associated with the specified lock,
> > + * or return NULL if there is no queue for the lock or if the lock's queue
> > + * is in transition.
> > + */
> > +static struct tkt_q_head *tkt_q_find_head(arch_spinlock_t *asp)
> 
> BTW, what does "asp" mean? arch_spinlock?

"arch_spinlock pointer", but yes.  Or I suppose a millenia-late warning
to Cleopatra.

>                                            If so, can we just call it
> "lock" and be consistent with all the other spinlock calls in the
> kernel. Because, I keep thinking this has something to do with Microsoft
> dynamic web pages.

Fair enough!

							Thanx, Paul

> -- Steve
> 
> > +{
> > +	int i;
> > +	int start;
> > +
> > +	start = i = tkt_q_hash(asp);
> > +	do
> > +		if (tkt_q_heads[i].ref == asp)
> > +			return &tkt_q_heads[i];
> > +	while ((i = tkt_q_next_slot(i)) != start);
> > +	return NULL;
> > +}
> > +
> 
>