[dm-devel] [PATCH] dm-thinp: report correct optimal I/O size

Joe Thornber thornber at redhat.com
Thu Apr 28 08:47:33 UTC 2011


On Thu, 2011-04-28 at 04:42 -0400, Christoph Hellwig wrote:
> On Wed, Apr 27, 2011 at 06:25:25PM +0100, Joe Thornber wrote:
> > tc->block_size is in sectors (you are passing sectors on the target
> > line?).
> > 
> > What's probably happening here is we should be doing:
> > 
> > blk_limits_io_opt(limits, min(<some theoretical max>, tc->block_size << SECTOR_SHIFT));
> 
> Yes, that should do it.  I don't even think we need the max, the optimum
> I/O size is a 32-bit value and we'll reach the limit of the possible
> block sizes much earlier.

I'm tempted to just say min(16M, tc->block_size << SECTOR_SHIFT).  Does
this sound reasonable to you?





More information about the dm-devel mailing list