[lkml]   [2010]   [Jun]   [15]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: [PATCH 11/12] vmscan: Write out dirty pages in batch
On Mon, Jun 14, 2010 at 04:21:43PM -0700, Andrew Morton wrote:
> Yup.
> But then, this all really should be done at the block layer so other
> io-submitting-paths can benefit from it.
> IOW, maybe "the sort queue is the submission queue" wasn't a good idea.

Even if has not effect on the actual I/O patters it has a massive
benefit for the filesystem. When probing delalloc/unwritten space at
least XFS does try to convert a larger extent forward from the index,
but doesn't bother to go backwards. By providing the trivial sort here
we make life a lot easier for the filesystem.

In addition to that we do get better I/O patters especially with short
queues as smart writepage implementatons will also submit the next few
pages, which is essentially free given how the storage works.
This means we already have a page cleaned before we might even submit it
without sorting.

 \ /
  Last update: 2010-06-15 12:59    [W:0.087 / U:5.612 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site