Xen devel - Oct 2007 - unregister_xenbus_watch() handling of xenwatch_mutex

2.6.23 acquires the mutex before canceling pending events, whereas
linux-2.6.18-xen acquires and releases the mutex back-to-back after
doing the cancellation. It would seem to me that the former method is
more robust, but otoh I''m wondering why this change then
wasn''t also
made in the 2.6.18 tree.

Thanks, Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

That''s a question for Jeremy. I think the 2.6.18 version is equally
correct
and only needs one if statement. Maybe hoisting the lock acquisition gave
Jeremy a warmer fuzzy feeling? :-)

 -- Keir

On 18/10/07 12:38, "Jan Beulich" <jbeulich@novell.com> wrote:
> 2.6.23 acquires the mutex before canceling pending events, whereas
> linux-2.6.18-xen acquires and releases the mutex back-to-back after
> doing the cancellation. It would seem to me that the former method is
> more robust, but otoh I''m wondering why this change then
wasn''t also
> made in the 2.6.18 tree.
> 
> Thanks, Jan
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

Keir Fraser wrote:
> That''s a question for Jeremy. I think the 2.6.18 version is
equally correct
> and only needs one if statement. Maybe hoisting the lock acquisition gave
> Jeremy a warmer fuzzy feeling? :-)
>   
I don''t remember changing this specifically.  It might have been in
response to a review comment, or maybe xen-unstable changed since I
grabbed the code...

    J

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel