Re: [PATCH 3/3] WorkStruct: Use direct assignment rather than cmpxchg()

[Andrew Morton <akpm@osdl.org>]

On Thu, 07 Dec 2006 15:31:43 +0000
David Howells <dhowells@redhat.com> wrote:

> Use direct assignment rather than cmpxchg() as the latter is unavailable and
> unimplementable on some platforms and is actually unnecessary.
> 
> The use of cmpxchg() was to guard against two possibilities, neither of which
> can actually occur:
> 
>  (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, 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.
> 
> If that's a problem, then a bitops-proofed assignment will be required -
> equivalent to atomic_set() vs other atomic_xxx() ops.
> 

I don't understand, as usual.

afacit in all (but one) cases we do

	if (!test_and_set_bit(WORK_STRUCT_PENDING, &work->management)) {
		...
		set_wq_data(work, wq);
		...
		<now do stuff which makes it possible for run_workqueue()
		 to get a look at the new work>
	}

cancel_delayed_work() looks OK too.

The possible exception is schedule_on_each_cpu() which is being lazy, but
looks fixable.


So...  afaict in all the places where there can be a concurrent
set_wq_data() and test_and_set_bit(), WORK_STRUCT_PENDING is reliably set,
and we can assume (and ensure) that a failing test_and_set_bit() will not
write to the affected word at all.

What am I missing?