On Fri 06-08-10 06:39:29, Wu Fengguang wrote:
> On Fri, Aug 06, 2010 at 01:00:16AM +0800, Jan Kara wrote:
> > I'm just afraid that in some
> > pathological cases this could result in bad writeback pattern - like if
> > there is some process which manages to dirty just a few pages while we are
> > doing writeout, this looping could result in writing just a few pages in
> > each round which is bad for fragmentation etc.
>
> Such inodes will be redirty_tail()ed here:
>
> if (mapping_tagged(mapping, PAGECACHE_TAG_DIRTY)) {
> /*
> * We didn't write back all the pages. nfs_writepages()
> * sometimes bales out without doing anything.
> */
> inode->i_state |= I_DIRTY_PAGES;
> if (wbc->nr_to_write <= 0) {
> /*
> * slice used up: queue for next turn
> */
> requeue_io(inode);
> } else {
> /*
> * Writeback blocked by something other than
> * congestion. Delay the inode for some time to
> * avoid spinning on the CPU (100% iowait)
> * retrying writeback of the dirty page/inode
> * that cannot be performed immediately.
> */
> redirty_tail(inode);
> }
Yes. And then, when there are no inodes in b_more_io, they get queued
again for writeback. So for non-background WB_SYNC_NONE writeback we can
just write a few pages over and over again... Oh, ok we won't because of
my start_time fix I suppose. Maybe a comment about this by the nr_to_write
< MAX_WRITEBACK_PAGES check would be good.
> > Actually, this comment probably also applies to your patch where you
> > change the queueing logic in writeback_single_inode(), doesn't it?
>
> Can you elaborate?
Sorry, my comment only applies to this particular patch. In your change
to writeback_single_inode() you requeue_io() only if nr_to_write <= 0.
Honza
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
