From: Dave Chinner <david@fromorbit.com>
To: Christoph Hellwig <hch@infradead.org>
Cc: xfs@oss.sgi.com
Subject: Re: [PATCH 16/18] xfs: serialise inode reclaim within an AG
Date: Mon, 27 Sep 2010 10:56:13 +1000 [thread overview]
Message-ID: <20100927005613.GA5665@dastard> (raw)
In-Reply-To: <20100925234929.GB23975@infradead.org>
On Sat, Sep 25, 2010 at 07:49:29PM -0400, Christoph Hellwig wrote:
> I really don't like the way the "trylock" variable is overloaded here.
> Just add a new skipped variable for restarting the scan and otherwise
> use (flags & SYNC_TRYLOCK) directly.
>
> > + int trylock = !!(flags & SYNC_TRYLOCK);
> >
> > +restart:
> > ag = 0;
> > while ((pag = xfs_perag_get_tag(mp, ag, XFS_ICI_RECLAIM_TAG))) {
> > unsigned long first_index = 0;
> > @@ -837,6 +839,17 @@ xfs_reclaim_inodes_ag(
> >
> > ag = pag->pag_agno + 1;
> >
> > + if (!mutex_trylock(&pag->pag_ici_reclaim_lock)) {
> > + if (trylock) {
> > + trylock++;
> > + continue;
> > + }
> > + mutex_lock(&pag->pag_ici_reclaim_lock);
> > + }
> > +
> > + if (trylock)
> > + first_index = pag->pag_ici_reclaim_cursor;
>
> Also this could be made more clear by:
>
> if (flags & SYNC_TRYLOCK) {
> if (!mutex_trylock(&pag->pag_ici_reclaim_lock)) {
> skipped++;
> continue;
> }
>
> first_index = pag->pag_ici_reclaim_cursor;
> } else {
> mutex_lock(&pag->pag_ici_reclaim_lock);
> }
I'll leave the trylock variable, but add a skipped variable. The
trylock two-pass algorithm (where the second pass goes into full
blocking reclaim mode) is intentional to ensure that we block
shrinker calls when there are more shrinkers than AGs rather than
spinning just trying to get per-ag reclaim locks. Regardless, I'll
clean up the code like you suggest because it is neater.
>
> > +
> > do {
> > struct xfs_inode *batch[XFS_LOOKUP_BATCH];
> > int i;
> > @@ -889,8 +902,19 @@ xfs_reclaim_inodes_ag(
> >
> > } while (nr_found && !done && *nr_to_scan > 0);
> >
> > + pag->pag_ici_reclaim_cursor = (done || !trylock) ? 0 : first_index;
>
> if ((flags & SYNC_TRYLOCK) && !done)
> pag->pag_ici_reclaim_cursor = first_index;
> else
> pag->pag_ici_reclaim_cursor = 0;
>
> > + /*
> > + * if we skipped any AG, and we still have scan count remaining, do
> > + * another pass this time waiting on the reclaim locks.
> > + */
> > + if (trylock > 1 && *nr_to_scan) {
> > + trylock = 0;
> > + goto restart;
> > + }
>
> In addition to waiting on the lock this also ignores the reclaim cursor.
Fixed the comment to indicate this is intentіonal.
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
next prev parent reply other threads:[~2010-09-27 0:55 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-09-24 12:30 [PATCH 0/18] xfs: metadata scalability V3 Dave Chinner
2010-09-24 12:30 ` [PATCH 01/18] xfs: force background CIL push under sustained load Dave Chinner
2010-09-24 12:31 ` [PATCH 02/18] xfs: reduce the number of CIL lock round trips during commit Dave Chinner
2010-09-24 12:31 ` [PATCH 03/18] xfs: remove debug assert for per-ag reference counting Dave Chinner
2010-09-24 12:31 ` [PATCH 04/18] xfs: lockless per-ag lookups Dave Chinner
2010-09-24 12:31 ` [PATCH 05/18] xfs: don't use vfs writeback for pure metadata modifications Dave Chinner
2010-09-25 23:42 ` Christoph Hellwig
2010-09-27 1:09 ` Dave Chinner
2010-09-24 12:31 ` [PATCH 06/18] xfs: rename xfs_buf_get_nodaddr to be more appropriate Dave Chinner
2010-09-24 12:31 ` [PATCH 07/18] xfs: introduced uncached buffer read primitve Dave Chinner
2010-09-24 12:31 ` [PATCH 08/18] xfs: store xfs_mount in the buftarg instead of in the xfs_buf Dave Chinner
2010-09-24 12:31 ` [PATCH 09/18] xfs: kill XBF_FS_MANAGED buffers Dave Chinner
2010-09-24 12:31 ` [PATCH 10/18] xfs: use unhashed buffers for size checks Dave Chinner
2010-09-24 12:31 ` [PATCH 11/18] xfs: remove buftarg hash for external devices Dave Chinner
2010-09-24 12:31 ` [PATCH 12/18] xfs: split inode AG walking into separate code for reclaim Dave Chinner
2010-09-24 12:31 ` [PATCH 13/18] xfs: split out inode walk inode grabbing Dave Chinner
2010-09-25 16:31 ` Christoph Hellwig
2010-09-24 12:31 ` [PATCH 14/18] xfs: implement batched inode lookups for AG walking Dave Chinner
2010-09-25 16:32 ` Christoph Hellwig
2010-09-24 12:31 ` [PATCH 15/18] xfs: batch inode reclaim lookup Dave Chinner
2010-09-24 12:31 ` [PATCH 16/18] xfs: serialise inode reclaim within an AG Dave Chinner
2010-09-25 23:49 ` Christoph Hellwig
2010-09-27 0:56 ` Dave Chinner [this message]
2010-09-24 12:31 ` [PATCH 17/18] xfs: convert buffer cache hash to rbtree Dave Chinner
2010-09-24 12:31 ` [PATCH 18/18] xfs: pack xfs_buf structure more tightly Dave Chinner
-- strict thread matches above, loose matches on Subject: below --
2010-09-27 1:47 [PATCH 0/18] xfs: metadata scalability V4 Dave Chinner
2010-09-27 1:47 ` [PATCH 16/18] xfs: serialise inode reclaim within an AG Dave Chinner
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20100927005613.GA5665@dastard \
--to=david@fromorbit.com \
--cc=hch@infradead.org \
--cc=xfs@oss.sgi.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox