From: Hugh Dickins <hughd@google.com>
To: Simon Jeons <simon.jeons@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>,
Petr Holasek <pholasek@redhat.com>,
Andrea Arcangeli <aarcange@redhat.com>,
Izik Eidus <izik.eidus@ravellosystems.com>,
linux-kernel@vger.kernel.org, linux-mm@kvack.org
Subject: Re: [PATCH 6/11] ksm: remove old stable nodes more thoroughly
Date: Sun, 27 Jan 2013 15:05:46 -0800 (PST) [thread overview]
Message-ID: <alpine.LNX.2.00.1301271451130.17495@eggly.anvils> (raw)
In-Reply-To: <1359262556.4159.23.camel@kernel>
On Sat, 26 Jan 2013, Simon Jeons wrote:
> On Fri, 2013-01-25 at 18:01 -0800, Hugh Dickins wrote:
> > Switching merge_across_nodes after running KSM is liable to oops on stale
> > nodes still left over from the previous stable tree. It's not something
> > that people will often want to do, but it would be lame to demand a reboot
> > when they're trying to determine which merge_across_nodes setting is best.
> >
> > How can this happen? We only permit switching merge_across_nodes when
> > pages_shared is 0, and usually set run 2 to force that beforehand, which
> > ought to unmerge everything: yet oopses still occur when you then run 1.
> >
> > Three causes:
> >
> > 1. The old stable tree (built according to the inverse merge_across_nodes)
> > has not been fully torn down. A stable node lingers until get_ksm_page()
> > notices that the page it references no longer references it: but the page
> > is not necessarily freed as soon as expected, particularly when swapcache.
> >
>
> When can this happen?
Whenever there's an additional reference to the page, beyond those for
its ptes in userspace - swapcache for example, or pinned by get_user_pages.
That delays its being freed (arriving at the "page->mapping = NULL;"
in free_pages_prepare()). Or it might simply be sitting in a pagevec,
waiting for that to be filled up, to be freed as part of a batch.
>
> > Fix this with a pass through the old stable tree, applying get_ksm_page()
> > to each of the remaining nodes (most found stale and removed immediately),
> > with forced removal of any left over. Unless the page is still mapped:
> > I've not seen that case, it shouldn't occur, but better to WARN_ON_ONCE
> > and EBUSY than BUG.
> >
> > 2. __ksm_enter() has a nice little optimization, to insert the new mm
> > just behind ksmd's cursor, so there's a full pass for it to stabilize
> > (or be removed) before ksmd addresses it. Nice when ksmd is running,
> > but not so nice when we're trying to unmerge all mms: we were missing
> > those mms forked and inserted behind the unmerge cursor. Easily fixed
> > by inserting at the end when KSM_RUN_UNMERGE.
>
> mms forked will be unmerged just after ksmd's cursor since they're
> inserted behind it, why will be missing?
unmerge_and_remove_all_rmap_items() makes one pass through the list
from start to finish: insert behind the cursor and it will be missed.
next prev parent reply other threads:[~2013-01-27 23:05 UTC|newest]
Thread overview: 69+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-01-26 1:53 [PATCH 0/11] ksm: NUMA trees and page migration Hugh Dickins
2013-01-26 1:54 ` [PATCH 1/11] ksm: allow trees per NUMA node Hugh Dickins
2013-01-27 1:14 ` Simon Jeons
2013-01-27 2:54 ` Hugh Dickins
2013-01-27 3:16 ` Simon Jeons
2013-01-27 21:55 ` Hugh Dickins
2013-01-28 23:03 ` Andrew Morton
2013-01-29 1:17 ` Hugh Dickins
2013-01-28 23:08 ` Andrew Morton
2013-01-29 1:38 ` Hugh Dickins
2013-02-05 16:41 ` Mel Gorman
2013-02-07 23:57 ` Hugh Dickins
2013-01-26 1:56 ` [PATCH 2/11] ksm: add sysfs ABI Documentation Hugh Dickins
2013-01-26 1:58 ` [PATCH 3/11] ksm: trivial tidyups Hugh Dickins
2013-01-28 23:11 ` Andrew Morton
2013-01-29 1:44 ` Hugh Dickins
2013-01-26 1:59 ` [PATCH 4/11] ksm: reorganize ksm_check_stable_tree Hugh Dickins
2013-02-05 16:48 ` Mel Gorman
2013-02-08 0:07 ` Hugh Dickins
2013-02-14 11:30 ` Mel Gorman
2013-01-26 2:00 ` [PATCH 5/11] ksm: get_ksm_page locked Hugh Dickins
2013-01-27 2:36 ` Simon Jeons
2013-01-27 22:08 ` Hugh Dickins
2013-01-28 0:36 ` Simon Jeons
2013-01-28 3:35 ` Hugh Dickins
2013-01-27 2:48 ` Simon Jeons
2013-01-27 22:10 ` Hugh Dickins
2013-02-05 17:18 ` Mel Gorman
2013-02-08 0:33 ` Hugh Dickins
2013-02-14 11:34 ` Mel Gorman
2013-01-26 2:01 ` [PATCH 6/11] ksm: remove old stable nodes more thoroughly Hugh Dickins
2013-01-27 4:55 ` Simon Jeons
2013-01-27 23:05 ` Hugh Dickins [this message]
2013-01-28 1:42 ` Simon Jeons
2013-01-28 4:14 ` Hugh Dickins
2013-01-28 2:12 ` Simon Jeons
2013-01-28 4:19 ` Hugh Dickins
2013-01-28 6:36 ` Simon Jeons
2013-01-28 23:44 ` Andrew Morton
2013-01-29 2:03 ` Hugh Dickins
2013-02-05 17:55 ` Mel Gorman
2013-02-08 19:33 ` Hugh Dickins
2013-02-14 11:58 ` Mel Gorman
2013-02-14 22:19 ` Hugh Dickins
2013-01-26 2:03 ` [PATCH 7/11] ksm: make KSM page migration possible Hugh Dickins
2013-01-27 5:47 ` Simon Jeons
2013-01-27 23:12 ` Hugh Dickins
2013-01-28 0:41 ` Simon Jeons
2013-01-28 3:44 ` Hugh Dickins
2013-02-05 19:11 ` Mel Gorman
2013-02-08 20:52 ` Hugh Dickins
2013-01-26 2:05 ` [PATCH 8/11] ksm: make !merge_across_nodes migration safe Hugh Dickins
2013-01-27 8:49 ` Simon Jeons
2013-01-27 23:25 ` Hugh Dickins
2013-01-28 3:44 ` Simon Jeons
2013-01-26 2:06 ` [PATCH 9/11] ksm: enable KSM page migration Hugh Dickins
2013-01-26 2:07 ` [PATCH 10/11] mm: remove offlining arg to migrate_pages Hugh Dickins
2013-01-26 2:10 ` [PATCH 11/11] ksm: stop hotremove lockdep warning Hugh Dickins
2013-01-27 6:23 ` Simon Jeons
2013-01-27 23:35 ` Hugh Dickins
2013-02-08 18:45 ` Gerald Schaefer
2013-02-11 22:13 ` Hugh Dickins
2013-01-28 23:54 ` [PATCH 0/11] ksm: NUMA trees and page migration Andrew Morton
2013-01-29 0:49 ` Izik Eidus
2013-01-29 2:26 ` Izik Eidus
2013-01-29 16:51 ` Andrea Arcangeli
2013-01-31 0:05 ` Ric Mason
2013-01-29 1:07 ` Hugh Dickins
2013-01-29 10:45 ` Gleb Natapov
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=alpine.LNX.2.00.1301271451130.17495@eggly.anvils \
--to=hughd@google.com \
--cc=aarcange@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=izik.eidus@ravellosystems.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=pholasek@redhat.com \
--cc=simon.jeons@gmail.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;
as well as URLs for NNTP newsgroup(s).