Re: [RFC][PATCH 0/7]: Fix for unsafe notifier chain

[Andrew Morton <akpm@osdl.org>]

Alan Stern <stern@rowland.harvard.edu> wrote:
>
> On Tue, 13 Dec 2005, Keith Owens wrote:
> 
> > On Thu, 8 Dec 2005 14:53:56 -0500 (EST), 
> > Alan Stern <stern@rowland.harvard.edu> wrote:
> > >The code below defines three new data structures: atomic_notifier_head,
> > >blocking_notifier_head, and raw_notifier_head.  The first two correspond
> > >to what we had in the earlier patch, and raw_notifier_head is almost the
> > >same as the current implementation, with no locking or protection at all.  
> > 
> > Acked-By: Keith Owens <kaos@sgi.com>
> > 
> > I do not care how this problem is fixed, I am happy with any solution that
> > 
> > (a) stops notify chains being racy and
> > (b) allows users of notify_die() to be safely unloaded.
> 
> Andrew, I've been waiting to hear back about this.

Was subconciously hoping it'd go away, I guess.

> Was that latest
> proposal (three separate types of notifier chains, each with its own API,
> one of them being completely raw) acceptable?

Yes, it looks sane enough.