On Mon, Feb 23, 2015 at 12:51:09PM -0800, Christoph Hellwig wrote: > This looks correct, but is there a good (performance) reason against > simply unconditionally flushing and waiting? No idea. All I am concerned about is correctness - getting the partial block flushed in the case that the on-disk size is the same as the in-memory size is the fix needed here, otherwise the behaviour should be unchanged. Given that I'm not sure what the effect of an unconditional flush is going to be, I'm not going to mix such a change with an otherwise obvious data corruption fix that we need to backport to other kernels. If you have the time to determine there is no performance impact from an unconditional flush, then I'll happily take the change. ;) Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
