Linux-mm Archive on lore.kernel.org
 help / color / Atom feed
From: Jan Kara <jack@suse.cz>
To: Mel Gorman <mgorman@suse.de>
Cc: Jan Kara <jack@suse.cz>, linux-mm@kvack.org, mhocko@suse.cz
Subject: Re: [PATCH 1/6] mm: migration: Factor out code to compute expected number of page references
Date: Fri, 14 Dec 2018 16:53:11 +0100
Message-ID: <20181214155311.GG8896@quack2.suse.cz> (raw)
In-Reply-To: <20181214151045.GG28934@suse.de>

On Fri 14-12-18 15:10:46, Mel Gorman wrote:
> On Tue, Dec 11, 2018 at 06:21:38PM +0100, Jan Kara wrote:
> > Factor out function to compute number of expected page references in
> > migrate_page_move_mapping(). Note that we move hpage_nr_pages() and
> > page_has_private() checks from under xas_lock_irq() however this is safe
> > since we hold page lock.
> > 
> > Signed-off-by: Jan Kara <jack@suse.cz>
> > ---
> >  mm/migrate.c | 27 +++++++++++++++++----------
> >  1 file changed, 17 insertions(+), 10 deletions(-)
> > 
> > diff --git a/mm/migrate.c b/mm/migrate.c
> > index f7e4bfdc13b7..789c7bc90a0c 100644
> > --- a/mm/migrate.c
> > +++ b/mm/migrate.c
> > @@ -428,6 +428,22 @@ static inline bool buffer_migrate_lock_buffers(struct buffer_head *head,
> >  }
> >  #endif /* CONFIG_BLOCK */
> >  
> > +static int expected_page_refs(struct page *page)
> > +{
> > +	int expected_count = 1;
> > +
> > +	/*
> > +	 * Device public or private pages have an extra refcount as they are
> > +	 * ZONE_DEVICE pages.
> > +	 */
> > +	expected_count += is_device_private_page(page);
> > +	expected_count += is_device_public_page(page);
> > +	if (page->mapping)
> > +		expected_count += hpage_nr_pages(page) + page_has_private(page);
> > +
> > +	return expected_count;
> > +}
> > +
> 
> I noticed during testing that THP allocation success rates under the
> mmtests configuration global-dhp__workload_thpscale-madvhugepage-xfs were
> terrible with massive latencies introduced somewhere in the series. I
> haven't tried chasing it down as it's relatively late but this block
> looked odd and I missed it the first time.

Interesting. I've run config-global-dhp__workload_thpscale and that didn't
show anything strange. But the numbers were fluctuating a lot both with and
without my patches applied. I'll have a look if I can reproduce this
sometime next week and look what could be causing the delays.

> This page->mapping test is relevant for the "Anonymous page without
> mapping" check but I think it's wrong. An anonymous page without mapping
> doesn't have a NULL mapping, it sets PAGE_MAPPING_ANON and the field can
> be special in other ways. I think you meant to use page_mapping(page)
> here, not page->mapping?

Yes, that's a bug. It should have been page_mapping(page). Thanks for
catching this.

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

  reply index

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-12-11 17:21 mm: migrate: Fix page migration stalls for blkdev pages Jan Kara
2018-12-11 17:21 ` [PATCH 1/6] mm: migration: Factor out code to compute expected number of page references Jan Kara
2018-12-13 13:05   ` Mel Gorman
2018-12-14 15:10   ` Mel Gorman
2018-12-14 15:53     ` Jan Kara [this message]
2018-12-14 16:24       ` Mel Gorman
2018-12-17 13:11         ` Jan Kara
2018-12-11 17:21 ` [PATCH 2/6] mm: migrate: Lock buffers before migrate_page_move_mapping() Jan Kara
2018-12-13 14:19   ` Mel Gorman
2018-12-11 17:21 ` [PATCH 3/6] mm: migrate: Move migrate_page_lock_buffers() Jan Kara
2018-12-13 14:57   ` Mel Gorman
2018-12-11 17:21 ` [PATCH 4/6] mm: migrate: Provide buffer_migrate_page_norefs() Jan Kara
2018-12-13 15:34   ` Mel Gorman
2018-12-14  4:53   ` Andrew Morton
2018-12-14  9:26     ` Jan Kara
2018-12-11 17:21 ` [PATCH 5/6] blkdev: Avoid migration stalls for blkdev pages Jan Kara
2018-12-13 15:35   ` Mel Gorman
2018-12-11 17:21 ` [PATCH 6/6] mm: migrate: Drop unused argument of migrate_page_move_mapping() Jan Kara
2018-12-13 15:35   ` Mel Gorman
2018-12-13 16:17     ` Jan Kara
2018-12-17 13:17       ` Jan Kara

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=20181214155311.GG8896@quack2.suse.cz \
    --to=jack@suse.cz \
    --cc=linux-mm@kvack.org \
    --cc=mgorman@suse.de \
    --cc=mhocko@suse.cz \
    /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

Linux-mm Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-mm/0 linux-mm/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-mm linux-mm/ https://lore.kernel.org/linux-mm \
		linux-mm@kvack.org
	public-inbox-index linux-mm

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kvack.linux-mm


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git