All of lore.kernel.org
 help / color / mirror / Atom feed
From: Brian Foster <bfoster@redhat.com>
To: "Darrick J. Wong" <darrick.wong@oracle.com>
Cc: linux-xfs@vger.kernel.org
Subject: Re: [PATCH v3 3/4] xfs: randomly fall back to near mode lookup algorithm in debug mode
Date: Mon, 19 Aug 2019 14:19:22 -0400	[thread overview]
Message-ID: <20190819181922.GE2875@bfoster> (raw)
In-Reply-To: <20190817013703.GB752159@magnolia>

On Fri, Aug 16, 2019 at 06:37:03PM -0700, Darrick J. Wong wrote:
> On Thu, Aug 15, 2019 at 08:55:37AM -0400, Brian Foster wrote:
> > The last block scan is the dominant near mode allocation algorithm
> > for a newer filesystem with fewer, large free extents. Add debug
> > mode logic to randomly fall back to lookup mode to improve
> > regression test coverage.
> 
> How about just using an errortag since the new sysfs interface lets
> testcases / admins control the frequency?
> 

We could do that, but my understanding of the equivalent logic in the
current algorithm is that we want broad coverage of both near mode
sub-algorithms across the entire suite of tests. Hence we randomly drop
allocations into either algorithm when DEBUG mode is enabled. IIRC, we
do something similar with sparse inodes (i.e., randomly allocate sparse
inode chunks even when unnecessary) so the functionality isn't only
covered by targeted tests.

Do we have the ability to have always on error tags as such? I thought
we had default frequency values for each tag, but I thought they still
had to be explicitly enabled. If that's the case, I'm sure we could come
up with such an on-by-default mechanism and perhaps switch over these
remaining DEBUG mode hacks, but that's a follow up thing IMO..

Brian

> --D
> 
> > Signed-off-by: Brian Foster <bfoster@redhat.com>
> > ---
> >  fs/xfs/libxfs/xfs_alloc.c | 8 +++++++-
> >  1 file changed, 7 insertions(+), 1 deletion(-)
> > 
> > diff --git a/fs/xfs/libxfs/xfs_alloc.c b/fs/xfs/libxfs/xfs_alloc.c
> > index 7753b61ba532..d550aa5597bf 100644
> > --- a/fs/xfs/libxfs/xfs_alloc.c
> > +++ b/fs/xfs/libxfs/xfs_alloc.c
> > @@ -1266,6 +1266,7 @@ xfs_alloc_ag_vextent_near(
> >  	int			i;
> >  	xfs_agblock_t		bno;
> >  	xfs_extlen_t		len;
> > +	bool			lastblock;
> >  
> >  	/* handle unitialized agbno range so caller doesn't have to */
> >  	if (!args->min_agbno && !args->max_agbno)
> > @@ -1291,7 +1292,12 @@ xfs_alloc_ag_vextent_near(
> >  	 * Otherwise run the optimized lookup search algorithm from the current
> >  	 * location to the end of the tree.
> >  	 */
> > -	if (xfs_btree_islastblock(acur.cnt, 0)) {
> > +	lastblock = xfs_btree_islastblock(acur.cnt, 0);
> > +#ifdef DEBUG
> > +	if (lastblock)
> > +		lastblock = prandom_u32() & 1;
> > +#endif
> > +	if (lastblock) {
> >  		int	j;
> >  
> >  		trace_xfs_alloc_cur_lastblock(args);
> > -- 
> > 2.20.1
> > 

  reply	other threads:[~2019-08-19 18:19 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-15 12:55 [PATCH v3 0/4] xfs: rework near mode extent allocation Brian Foster
2019-08-15 12:55 ` [PATCH v3 1/4] xfs: track active state of allocation btree cursors Brian Foster
2019-08-17  0:51   ` Darrick J. Wong
2019-08-19 18:03     ` Brian Foster
2019-08-15 12:55 ` [PATCH v3 2/4] xfs: use locality optimized cntbt lookups for near mode allocations Brian Foster
2019-08-17  1:34   ` Darrick J. Wong
2019-08-19 18:06     ` Brian Foster
2019-08-15 12:55 ` [PATCH v3 3/4] xfs: randomly fall back to near mode lookup algorithm in debug mode Brian Foster
2019-08-17  1:37   ` Darrick J. Wong
2019-08-19 18:19     ` Brian Foster [this message]
2019-08-15 12:55 ` [PATCH v3 4/4] xfs: refactor successful AG allocation accounting code Brian Foster
2019-08-17  0:28   ` Darrick J. Wong

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=20190819181922.GE2875@bfoster \
    --to=bfoster@redhat.com \
    --cc=darrick.wong@oracle.com \
    --cc=linux-xfs@vger.kernel.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.