On Thu, Mar 29, 2012 at 03:07:59PM -0400, Christoph Hellwig wrote: > I don't like this solution. What speaks against building up the page > array after the first buffer lookup failed, but before linking the > buffer into the rbtree? That's the the inode cache and all the VFS > caches do. That could be done, I think. I always wondered why it was done after insert, but never knew the reason so I didn't change the logic. I'll respin it and see what happens. Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
