LKML Archive on
help / color / mirror / Atom feed
From: Andrew Morton <>
To: David Howells <>
Cc: David Miller <>,,
Subject: Re: cmpxchg() in kernel/workqueue.c breaks things
Date: Thu, 7 Dec 2006 03:28:17 -0800	[thread overview]
Message-ID: <> (raw)
In-Reply-To: <>

On Thu, 07 Dec 2006 11:03:49 +0000
David Howells <> wrote:

> David Miller <> wrote:
> > David, you have to fix the locking scheme used in kernel/workqueue.c,
> > you absolutely cannot assume that cmpxchg() is available on all
> > platforms.  This breaks the build on the platforms that don't
> > have such an instruction, and no it cannot emulated.
> Yeah, I've figured that one out.  Also, having considered things last night, I
> think the use of cmpxchg() is unnecessary.
> I was trying to handle against two possibilities:
>  (1) The pending flag may have been unset or may be cleared.  However, given
>      where it's called, the pending flag is _always_ set.  I don't think it
>      can be unset whilst we're in set_wq_data().
>      Once the work is enqueued to be actually run, the only way off the queue
>      is for it to be actually run.
>      If it's a delayed work item, then the bit can't be cleared by the timer
>      because we haven't started the timer yet.  Also, the pending bit can't be
>      cleared by cancelling the delayed work _until_ the work item has had its
>      timer started.
>  (2) The workqueue pointer might change.  This can only happen in two cases:
>      (a) The work item has just been queued to actually run, and so we're
>          protected by the appropriate workqueue spinlock.
>      (b) A delayed work item is being queued, and so the timer hasn't been
>      	 started yet, and so no one else knows about the work item or can
>      	 access it (the pending bit protects us).
>      Besides, set_wq_data() _sets_ the workqueue pointer unconditionally, so
>      it can be assigned instead.
> So, I think replacing the set_wq_data() with a straight assignment would be
> okay in most cases.  The problem is where we end up tangling with
> test_and_set_bit() emulated using spinlocks, and even then it's not a problem
> _provided_ test_and_set_bit() doesn't attempt to modify the word if the bit
> was set.

I don't see why the 2.6.19 logic needed changing.

a) Nobody should be freeing the work_struct itself without running
   flush_scheduled_work() and

b) even if the work_struct _did_ get freed, the callback function won't
   care, because there's nothing in that work_struct which it's interested

  reply	other threads:[~2006-12-07 11:29 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-12-07  8:09 David Miller
2006-12-07  8:23 ` Al Viro
2006-12-07 11:03 ` David Howells
2006-12-07 11:28   ` Andrew Morton [this message]
2006-12-07 11:46   ` David Howells
2006-12-07 16:35     ` Andrew Morton
2006-12-07 13:24 ` David Howells
2006-12-07 14:53 ` Russell King

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:

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \ \ \ \ \ \ \
    --subject='Re: cmpxchg() in kernel/workqueue.c breaks things' \

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).