mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Jamie Lokier <jamie@shareable.org>
To: Ingo Molnar <mingo@redhat.com>
Cc: Hugh Dickins <hugh@veritas.com>,
	Rusty Russell <rusty@rustcorp.com.au>,
	Andrew Morton <akpm@osdl.org>,
	linux-kernel@vger.kernel.org, Linus Torvalds <torvalds@osdl.org>
Subject: Re: [PATCH 2] Little fixes to previous futex patch
Date: Sun, 7 Sep 2003 14:00:17 +0100	[thread overview]
Message-ID: <20030907130017.GA19977@mail.jlokier.co.uk> (raw)
In-Reply-To: <Pine.LNX.4.44.0309070322310.17404-100000@devserv.devel.redhat.com>

Ingo Molnar wrote:
> btw., regarding this fix:
> 
>   ChangeSet@1.1179.2.5, 2003-09-06 12:28:20-07:00, hugh@veritas.com
>     [PATCH] Fix futex hashing bugs
> 
> why dont we do this:
> 
>                         } else {
>                                 /* Make sure to stop if key1 == key2 */
> 				if (head1 == head2)
> 					break;
>                                 list_add_tail(i, head2);
>                                 this->key = key2;
>                                 if (ret - nr_wake >= nr_requeue)
>                                         break;
>                         }
> 
> instead of the current:
> 
>                         } else {
>                                 list_add_tail(i, head2);
>                                 this->key = key2;
>                                 if (ret - nr_wake >= nr_requeue)
>                                         break;
>                                 /* Make sure to stop if key1 == key2 */
>                                 if (head1 == head2 && head1 != next)
>                                         head1 = i;
>                         }
> 
> what's the point in requeueing once, and then exiting the loop by changing
> the loop exit condition variable?

Hugh's patch is clever and subtle.  It doesn't exit the loop; the
loop continues from "next".

What it does is change the end condition so that the loop stops just
before the first requeued futex.  Let's call that one REQUEUED1.

If are other futexes to requeue, they after inserted after REQUEUED1
(because head2 wasn't changed), yet the end condition _isn't_ changed
when this happens, because now head1 != head2.

This causes the correct number of futexes to be requeued at the end of
the wait list.

> You are trying to avoid the lockup but the first one ought to be the
> most straightforward way to do it.

Hugh's patch returns the correct retval _and_ requeues the correct
number of waiters to the end of the queue.  And it does it without
fancy code.

Remember that the waiter order is visible to userspace - it's used by
"fair" operations, so it's appropriate that requeuing a futex to
itself moves nr_requeues waiters to the end of the queue, just like it
does when it requeues to a different futex.

If the code to handle that were complicated, I'd vote for dropping it.
But Hugh's patch does exactly the right thing in a simple way.  Lovely!

-- Jamie

  parent reply	other threads:[~2003-09-07 13:00 UTC|newest]

Thread overview: 102+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2003-08-26  3:12 [PATCH 2/2] Futex non-page-pinning fix Rusty Russell
2003-08-26  4:06 ` Andrew Morton
2003-08-26  5:30   ` Ingo Molnar
2003-08-26  5:50     ` Andrew Morton
2003-08-26  5:58       ` Ingo Molnar
2003-08-26  6:14         ` Andrew Morton
2003-08-26  6:36           ` Ingo Molnar
2003-08-26  7:02             ` Andrew Morton
2003-08-26  7:56               ` Arjan van de Ven
2003-08-26  8:08                 ` Muli Ben-Yehuda
2003-08-26  8:11                   ` Arjan van de Ven
2003-08-26  8:25                   ` Andrew Morton
2003-08-26  9:02                     ` Muli Ben-Yehuda
2003-08-26 10:38                   ` William Lee Irwin III
2003-08-26 10:44                     ` Andrew Morton
2003-08-26 10:45                       ` William Lee Irwin III
2003-08-26 17:29                         ` Andrew Morton
2003-08-26 19:35                           ` William Lee Irwin III
2003-08-27  5:17   ` Rusty Russell
2003-08-27  6:20     ` Andrew Morton
2003-08-28  0:47       ` Rusty Russell
2003-08-28  8:21         ` Andrew Morton
2003-08-29  3:46           ` Rusty Russell
2003-08-29  4:17             ` Andrew Morton
2003-08-30  7:49               ` Rusty Russell
2003-09-01  0:35             ` Jamie Lokier
2003-09-01  4:11               ` Rusty Russell
2003-09-01 20:57                 ` Hugh Dickins
2003-09-02  3:12                   ` Rusty Russell
2003-09-02  6:51                     ` Jamie Lokier
2003-09-02 16:14                       ` Hugh Dickins
2003-09-02 19:54                         ` Jamie Lokier
2003-09-02 20:15                           ` Andrew Morton
2003-09-02 21:20                             ` Jamie Lokier
2003-09-03  2:40                         ` Rusty Russell
2003-09-03  7:36                           ` [PATCH] Alternate futex non-page-pinning and COW fix Jamie Lokier
2003-09-03 11:19                             ` Hugh Dickins
2003-09-03 14:38                               ` Jamie Lokier
2003-09-03 17:39                               ` Jamie Lokier
2003-09-03 17:55                                 ` Linus Torvalds
2003-09-03 18:06                                   ` Hugh Dickins
2003-09-03 18:19                                     ` Linus Torvalds
2003-09-03 18:43                                       ` Hugh Dickins
2003-09-03 19:05                                         ` Linus Torvalds
2003-09-03 19:40                                           ` Hugh Dickins
2003-09-03 20:04                                             ` Linus Torvalds
2003-09-04  2:43                                           ` Rusty Russell
2003-09-04  8:28                                             ` Linus Torvalds
2003-09-04 12:20                                               ` Hugh Dickins
2003-09-04 15:40                                                 ` Linus Torvalds
2003-09-04 16:55                                                   ` Hugh Dickins
2003-09-04 18:38                                           ` Jamie Lokier
2003-09-04 18:46                                             ` Linus Torvalds
2003-09-04 20:04                                               ` Jamie Lokier
2003-09-04 21:49                                                 ` Linus Torvalds
2003-09-04 22:10                                                   ` Jamie Lokier
2003-09-04 17:16                                   ` Jamie Lokier
2003-09-04 17:38                                     ` Linus Torvalds
2003-09-04 18:42                                       ` Linus Torvalds
2003-09-04 18:42                                         ` Linus Torvalds
2003-09-05  3:55                                     ` Rusty Russell
2003-09-05 17:55                                       ` Jamie Lokier
2003-09-04 17:26                                   ` Jamie Lokier
2003-09-04  1:30                               ` Rusty Russell
2003-09-04 21:00                                 ` Jamie Lokier
2003-09-05  5:19                                   ` Rusty Russell
2003-09-05 20:54                                     ` Jamie Lokier
2003-09-07  6:45                                       ` Rusty Russell
2003-09-07 13:20                                         ` Jamie Lokier
2003-09-08  1:49                                           ` Rusty Russell
2003-09-08  9:44                                             ` Jamie Lokier
2003-09-03 14:40                             ` [PATCH 2] Little fixes to previous futex patch Jamie Lokier
2003-09-04 16:45                               ` Hugh Dickins
2003-09-04 17:59                                 ` Jamie Lokier
2003-09-04 18:35                                   ` Hugh Dickins
2003-09-04 20:11                                     ` Jamie Lokier
2003-09-04 21:36                                       ` Hugh Dickins
2003-09-04 21:58                                         ` Jamie Lokier
2003-09-07  7:23                                         ` Ingo Molnar
2003-09-07 12:27                                           ` Hugh Dickins
2003-09-07 15:03                                             ` Jamie Lokier
2003-09-08  1:56                                             ` Rusty Russell
2003-09-07 13:00                                           ` Jamie Lokier [this message]
2003-09-08  3:32                                             ` Ingo Molnar
2003-09-08  9:33                                               ` Jamie Lokier
2003-09-08  9:57                                                 ` Ingo Molnar
2003-09-05  4:56                                   ` Rusty Russell
2003-09-03 15:34                             ` [PATCH] Alternate futex non-page-pinning and COW fix Andrew Morton
2003-09-03 17:16                               ` Jamie Lokier
2003-09-04  1:35                             ` Rusty Russell
2003-09-04 17:35                               ` Jamie Lokier
2003-09-03  0:14                       ` [PATCH 2/2] Futex non-page-pinning fix Rusty Russell
2003-09-03  1:16                         ` Andrew Morton
2003-09-03  1:54                           ` Dave Hansen
2003-09-03  2:54                             ` Andrew Morton
2003-09-02  3:23                   ` Hugh Dickins
2003-09-02 23:58                     ` Rusty Russell
2003-08-27  8:37     ` Hugh Dickins
2003-08-27  8:56       ` William Lee Irwin III
2003-08-27 10:38       ` Andrew Morton
2003-08-27 10:57         ` Hugh Dickins
2003-08-28  8:03       ` Rusty Russell

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=20030907130017.GA19977@mail.jlokier.co.uk \
    --to=jamie@shareable.org \
    --cc=akpm@osdl.org \
    --cc=hugh@veritas.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@redhat.com \
    --cc=rusty@rustcorp.com.au \
    --cc=torvalds@osdl.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

all inboxes | Powered by JetHome®