From: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
To: Steven Rostedt <rostedt@goodmis.org>
Cc: linux-kernel@vger.kernel.org, mingo@elte.hu,
laijs@cn.fujitsu.com, dipankar@in.ibm.com,
akpm@linux-foundation.org, mathieu.desnoyers@efficios.com,
josh@joshtriplett.org, niv@us.ibm.com, tglx@linutronix.de,
peterz@infradead.org, Valdis.Kletnieks@vt.edu,
dhowells@redhat.com, edumazet@google.com, darren@dvhart.com,
fweisbec@gmail.com, sbw@mit.edu, torvalds@linux-foundation.org
Subject: Re: [PATCH RFC ticketlock] Auto-queued ticketlock
Date: Tue, 11 Jun 2013 08:00:32 -0700 [thread overview]
Message-ID: <20130611150032.GA1762@linux.vnet.ibm.com> (raw)
In-Reply-To: <20130611095655.GA5146@linux.vnet.ibm.com>
On Tue, Jun 11, 2013 at 02:56:55AM -0700, Paul E. McKenney wrote:
> On Mon, Jun 10, 2013 at 08:44:40PM -0400, Steven Rostedt wrote:
> > On Sun, 2013-06-09 at 12:36 -0700, Paul E. McKenney wrote:
[ . . . ]
> > > +bool tkt_q_do_spin(arch_spinlock_t *asp, struct __raw_tickets inc)
> > > +{
> > > + struct tkt_q **oldtail;
> > > + struct tkt_q tq;
> > > + struct tkt_q_head *tqhp;
> > > +
> > > + /*
> > > + * Ensure that accesses to queue header happen after sensing
> > > + * the lock's have-queue bit.
> > > + */
> > > + smp_mb(); /* See above block comment. */
> > > +
> > > + /* If there no longer is a queue, leave. */
> > > + tqhp = tkt_q_find_head(asp);
> > > + if (tqhp == NULL)
> > > + return false;
> > > +
> > > + /* Initialize our queue element. */
> > > + tq.cpu = raw_smp_processor_id();
> > > + tq.tail = inc.tail;
> > > + tq.next = NULL;
> > > +
> > > + /* Check to see if we already hold the lock. */
> > > + if (ACCESS_ONCE(tqhp->head_tkt) == inc.tail) {
> > > + /* The last holder left before queue formed, we hold lock. */
> > > + tqhp->head_tkt = -1;
> > > + return true;
> > > + }
> > > +
> > > + /* Add our element to the tail of the queue. */
> > > + oldtail = xchg(&tqhp->spin_tail, &tq.next);
> >
> > Boy this is tricky code! I thought I found a race window here, but as I
> > went to write my email saying "Gotcha!" I found that it wasn't a race
> > after all. But as I went though the effort of writing this, I figured I
> > would send this out as documentation for others to see. Hmm, I wonder if
> > we can use this email to add more comments. Anyway, here's what I
> > thought was wrong ;-)
>
> If you didn't know any better, you might even think that I had done
> something like this before. ;-)
>
> > OK, I originally thought there was a race window here. Let's say that an
> > NMI hit right here, and it happens to be a big one, where lots of things
> > can happen on other CPUs right now.
> >
> > The scenario is that there's just one item on the queue, which is
> > waiting for the lock to be released, and is spinning below in the:
> >
> > while (ACCESS_ONCE(tq.cpu) != -1)
> > cpu_relax();
> >
> > And then the lock is released, where in tkt_q_do_wake() the following is
> > called:
> >
> > ACCESS_ONCE(tqp->cpu) = -1;
> >
> > Now the old queued task is released. But it's tq->next hasn't been set
> > yet, and is still NULL. It leaves by doing:
> >
> > ACCESS_ONCE(tqhp->spin) = tq.next;
> > return true;
> >
> > All before this task gets to set *oldtail to &tq. But, I then looked
> > below...
> >
> >
> > > + ACCESS_ONCE(*oldtail) = &tq;
> > > +
> > > + /* Spin until handoff. */
> > > + while (ACCESS_ONCE(tq.cpu) != -1)
> > > + cpu_relax();
> > > +
> > > + /*
> > > + * Remove our element from the queue. If the queue is now empty,
> > > + * update carefully so that the next acquisition will queue itself
> > > + * at the head of the list.
> > > + */
> > > + if (tq.next == NULL) {
> >
> > This checks for that scenario.
>
> Yep!
>
> > As if the old task were to come out
> > spinning, the problem would only be if it was the last one on the list,
> > and its tq.next was NULL. But if that was the case, then we set spin to
> > NULL and do the next trick, where I thought I gotcha again...
> >
> >
> > > +
> > > + /* Mark the queue empty. */
> > > + tqhp->spin = NULL;
> > > +
> > > + /* Try to point the tail back at the head. */
> > > + if (cmpxchg(&tqhp->spin_tail,
> > > + &tq.next,
> > > + &tqhp->spin) == &tq.next)
> >
> > Here, I was thinking, oh wait, what happens if this is called right
> > before the xchg() above. Then we would set spin_tail but not update the
> > old tq.next. But wait! look at what we assign spin_tail to. It's the
> > address of spin, which would be what oldtail would point to above, and
> > then above would set spin to the new tq!
>
> Yep again!
>
> > OK, I haven't found a issue here yet, but youss are beiing trickssy! We
> > don't like trickssy, and we must find precccciouss!!!
> >
> >
> > This code is starting to make me look like Gollum :-p
>
> Hmmm... The time and effort to do this might almost have been worthwhile
> just to accomplish that! ;-)
>
> But yes, this would need better comments, design documentation, or
> maybe both.
And for whatever it might be worth, here is an attempted upgrade for
comments.
First, I upgrade the comment for the xchg() that does the enqueue:
/*
* Add our element to the tail of the queue. Note that if the
* queue is empty, the ->spin_tail pointer will reference
* the queue's head pointer, namely ->spin.
*/
oldtail = xchg(&tqhp->spin_tail, &tq.next);
ACCESS_ONCE(*oldtail) = &tq;
Next, I upgrade the comment for the dequeue operation:
/*
* Remove our element from the queue. If the queue is now empty,
* update carefully so that the next acquisition will enqueue itself
* at the head of the list. Of course, the next enqueue operation
* might be happening concurrently, and this code needs to handle all
* of the possible combinations, keeping in mind that the enqueue
* operation happens in two stages: (1) update the tail pointer and
* (2) update the predecessor's ->next pointer. With this in mind,
* the following code needs to deal with three scenarios:
*
* 1. tq is the last entry. In this case, we use cmpxchg to
* point the list tail back to the list head (->spin). If
* the cmpxchg fails, that indicates that we are instead
* in scenario 2 below. If the cmpxchg succeeds, the next
* enqueue operation's tail-pointer exchange will enqueue
* the next element at the queue head, because the ->spin_tail
* pointer now references the queue head.
*
* 2. tq is the last entry, and the next entry has updated the
* tail pointer but has not yet updated tq.next. In this
* case, tq.next is NULL, the cmpxchg will fail, and the
* code will wait for the enqueue to complete before completing
* removal of tq from the list.
*
* 3. tq is not the last pointer. In this case, tq.next is non-NULL,
* so the following code simply removes tq from the list.
*/
if (tq.next == NULL) {
/* Mark the queue empty. */
tqhp->spin = NULL;
/* Try to point the tail back at the head. */
if (cmpxchg(&tqhp->spin_tail,
&tq.next,
&tqhp->spin) == &tq.next)
return true; /* Succeeded, queue is now empty. */
/* Failed, if needed, wait for the enqueue to complete. */
while (tq.next == NULL)
cpu_relax();
/* The following code will repair the head. */
}
smp_mb(); /* Force ordering between handoff and critical section. */
/*
* Advance list-head pointer. This same task will be the next to
* access this when releasing the lock, so no need for a memory
* barrier after the following assignment.
*/
ACCESS_ONCE(tqhp->spin) = tq.next;
return true;
}
Thanx, Paul
next prev parent reply other threads:[~2013-06-11 15:15 UTC|newest]
Thread overview: 96+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-06-09 19:36 Paul E. McKenney
2013-06-10 20:47 ` Steven Rostedt
2013-06-10 20:57 ` Paul E. McKenney
2013-06-10 21:01 ` Thomas Gleixner
2013-06-10 21:15 ` Paul E. McKenney
2013-06-10 21:08 ` Steven Rostedt
2013-06-10 21:30 ` Paul E. McKenney
2013-06-10 21:35 ` Eric Dumazet
2013-06-10 21:54 ` Paul E. McKenney
2013-06-10 23:02 ` Steven Rostedt
2013-06-11 0:22 ` Paul E. McKenney
2013-06-11 0:44 ` Steven Rostedt
2013-06-11 0:51 ` Linus Torvalds
2013-06-11 7:53 ` Lai Jiangshan
2013-06-11 10:14 ` Paul E. McKenney
2013-06-11 15:22 ` Steven Rostedt
2013-06-11 16:45 ` Paul E. McKenney
2013-06-11 10:06 ` Paul E. McKenney
2013-06-11 17:53 ` Davidlohr Bueso
2013-06-11 18:05 ` Paul E. McKenney
2013-06-11 18:10 ` Steven Rostedt
2013-06-11 18:14 ` Davidlohr Bueso
2013-06-11 18:46 ` Paul E. McKenney
2013-06-12 17:50 ` Davidlohr Bueso
2013-06-12 18:15 ` Linus Torvalds
2013-06-12 20:03 ` Davidlohr Bueso
2013-06-12 20:26 ` Linus Torvalds
2013-06-12 20:40 ` Davidlohr Bueso
2013-06-12 21:06 ` Raymond Jennings
2013-06-12 23:32 ` Al Viro
2013-06-13 0:01 ` Linus Torvalds
2013-06-13 0:20 ` Al Viro
2013-06-13 0:38 ` Linus Torvalds
2013-06-13 0:49 ` Al Viro
2013-06-13 0:59 ` Linus Torvalds
2013-06-14 15:00 ` Waiman Long
2013-06-14 15:37 ` Linus Torvalds
2013-06-14 18:17 ` Waiman Long
2013-06-15 1:26 ` Benjamin Herrenschmidt
2013-06-15 3:36 ` Waiman Long
2013-06-12 20:37 ` Linus Torvalds
2013-06-12 18:18 ` Steven Rostedt
2013-06-11 9:56 ` Paul E. McKenney
2013-06-11 15:00 ` Paul E. McKenney [this message]
2013-06-11 1:04 ` Steven Rostedt
2013-06-11 9:52 ` Paul E. McKenney
2013-06-11 14:48 ` Lai Jiangshan
2013-06-11 15:10 ` Lai Jiangshan
2013-06-11 16:48 ` Paul E. McKenney
2013-06-11 17:17 ` Linus Torvalds
2013-06-11 17:30 ` Paul E. McKenney
2013-06-11 16:21 ` Paul E. McKenney
2013-06-11 15:57 ` Waiman Long
2013-06-11 16:20 ` Steven Rostedt
2013-06-11 16:43 ` Paul E. McKenney
2013-06-11 17:13 ` Steven Rostedt
2013-06-11 17:43 ` Paul E. McKenney
2013-06-11 17:35 ` Waiman Long
2013-06-11 16:36 ` Paul E. McKenney
2013-06-11 17:01 ` Steven Rostedt
2013-06-11 17:16 ` Paul E. McKenney
2013-06-11 18:41 ` Waiman Long
2013-06-11 18:54 ` Davidlohr Bueso
2013-06-11 19:49 ` Paul E. McKenney
2013-06-11 20:09 ` Steven Rostedt
2013-06-11 20:32 ` Paul E. McKenney
2013-06-11 20:53 ` Steven Rostedt
2013-06-11 20:25 ` Jason Low
2013-06-11 20:36 ` Paul E. McKenney
2013-06-11 20:56 ` Steven Rostedt
2013-06-11 21:09 ` Paul E. McKenney
2013-06-12 1:19 ` Lai Jiangshan
2013-06-12 1:58 ` Steven Rostedt
2013-06-12 10:12 ` Paul E. McKenney
2013-06-12 11:06 ` Lai Jiangshan
2013-06-12 14:21 ` Paul E. McKenney
2013-06-12 14:15 ` Lai Jiangshan
2013-06-12 14:44 ` Paul E. McKenney
2013-06-11 17:02 ` [PATCH RFC ticketlock] v2 " Paul E. McKenney
2013-06-11 17:35 ` Linus Torvalds
2013-06-11 17:49 ` Paul E. McKenney
2013-06-11 17:36 ` Steven Rostedt
2013-06-11 17:52 ` Paul E. McKenney
2013-06-12 15:40 ` [PATCH RFC ticketlock] v3 " Paul E. McKenney
2013-06-12 16:13 ` Lai Jiangshan
2013-06-12 16:59 ` Paul E. McKenney
2013-06-13 2:55 ` Lai Jiangshan
2013-06-13 15:22 ` Paul E. McKenney
2013-06-13 23:25 ` Lai Jiangshan
2013-06-13 23:57 ` Paul E. McKenney
2013-06-14 1:28 ` Lai Jiangshan
2013-06-14 23:49 ` Paul E. McKenney
2013-06-14 7:12 ` Lai Jiangshan
2013-06-14 23:46 ` Paul E. McKenney
[not found] ` <CAC4Lta3dpTDc19rXLVQkZrxbu8AJL+Foc6ocAktUAozCpk2-Mg@mail.gmail.com>
2013-07-01 9:19 ` Raghavendra KT
2013-07-02 5:56 ` Paul E. McKenney
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20130611150032.GA1762@linux.vnet.ibm.com \
--to=paulmck@linux.vnet.ibm.com \
--cc=Valdis.Kletnieks@vt.edu \
--cc=akpm@linux-foundation.org \
--cc=darren@dvhart.com \
--cc=dhowells@redhat.com \
--cc=dipankar@in.ibm.com \
--cc=edumazet@google.com \
--cc=fweisbec@gmail.com \
--cc=josh@joshtriplett.org \
--cc=laijs@cn.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mathieu.desnoyers@efficios.com \
--cc=mingo@elte.hu \
--cc=niv@us.ibm.com \
--cc=peterz@infradead.org \
--cc=rostedt@goodmis.org \
--cc=sbw@mit.edu \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
Powered by JetHome