On Thu, Apr 28, 2011 at 08:55:46AM -0400, Christoph Hellwig wrote: > Replace our own handcrafted I/O clustering with the generic write_cache_pages > helper. While the old code would add any additional page in an existing > mapping, the new code iterates over the pages, either adding them to > a previous mapping if it fits or otherwise gets a new one. > > This has survived xfsqa for small and 4k blocksize on x86. I've had a quick look at the patches, and I can't see anything obvious that jumps out at me with a big red flag. It look slike a very neat optimisation and simplification. I'll do some testing on them before doing a more robust review, though.... Cheers, Dave. -- Dave Chinner david@xxxxxxxxxxxxx _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs