On Thu, Mar 05, 2015 at 11:33:55AM -0800, Tariq Saeed wrote:
>
> On 03/04/2015 03:57 PM, Mark Fasheh wrote:
> > Ok, so the problem is that we're not holding a cluster lock across acl
> Yes, but in addition to that, the inode lock in ocfs2_setattr MUST not
> be dropped till
> the exit since all operations,mode change, size change etc,
> in that function plus any changes to acls are atomic from user's point
> of view.
That's fine, my primary concern is that we're not locking ACL calls _at
all_. The fix you submitted doesn't really adress that, it just fixes one
isolated case, hence my description of a more complete solution :)
--Mark
--
Mark Fasheh
_______________________________________________
Ocfs2-devel mailing list
Ocfs2-devel@oss.oracle.com
https://oss.oracle.com/mailman/listinfo/ocfs2-devel
