On Thu, Jan 06, 2011 at 03:49:11PM -0500, Christoph Hellwig wrote: > On Tue, Jan 04, 2011 at 08:52:31PM +1100, Nick Piggin wrote: > > OK I missed that part about not requiring dirty metadata to be written, > > just currently ongoing async operations. But then I don't understand how > > it would be used by nfsd, how does nfsd start some async operation on > > the inode metadata such that ->commit_metadata would do anything useful > > for it? > > NFSD calls various inode operations (create/mkdir/mknod/link/symlink/ > rename/unlink/rmdir/setattr) and then requires those operations to be > on disk before completing the request to the client, but it does not > require other dirty state to be written (data, unlogged size > or timestamp updates). Take a look at the XFS implementation: it just > checks if the inode is still pinned (that is in the in-memory log, but > not commited to disk) and if so forces the log up to the log buffer > that contains the last changes to the inode. OK, I don't exactly see why a sync_inode with appropriate flag could not solve that problem. I'll take a bit more look through nfs and xfs. Thanks... -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
