On Mon, Feb 15, 2010 at 10:04:52PM +0100, Andi Kleen wrote: > Christoph Hellwig <hch at infradead.org> writes: > > > This series implements a real fdatasync, that is one that is not > > simply identical to fdatasync, but one that skips logging the inode > > > ... to fsync? Yes, Chris Wedgwood also pointed this out privately already.