On Sat, Oct 22, 2011 at 04:32:24AM -0700, Nicholas A. Bellinger wrote: > On Tue, 2011-10-18 at 06:57 -0400, Christoph Hellwig wrote: > > plain text document attachment (target-optimize-single-sgls) > > If we only have a single task per command (which at least in my testing > > is the by far most common case) we do not have to allocate a new per-task > > S/G list but can reuse the one from the command. > > > > Signed-off-by: Christoph Hellwig <hch@xxxxxx> > > > > A very nice optimization for the single task case, but you would be > surprised how many uber fast flash backends end up using max_sectors=128 > and use task_count > 1 for normal large block operation.. :) The default 128 max_sector limit in Linux is utterly stupid - already for any server / HPC workload given that we ask the fs to build small bios just to merge them at the request layer. It's even more so at for a target box. But there's a small issue in the target code as well - we're not really limited to max_sectors per task, but to max_sectors * number of bios (speaking of the iblock case). -- To unsubscribe from this list: send the line "unsubscribe target-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
