[PATCHSET v3] writeback, memcg: Implement foreign inode flushing

From: Tejun Heo <tj@kernel.org>
To: axboe@kernel.dk, jack@suse.cz, hannes@cmpxchg.org,
	mhocko@kernel.org, vdavydov.dev@gmail.com
Cc: cgroups@vger.kernel.org, linux-mm@kvack.org,
	linux-block@vger.kernel.org, linux-kernel@vger.kernel.org,
	kernel-team@fb.com, guro@fb.com, akpm@linux-foundation.org
Subject: [PATCHSET v3] writeback, memcg: Implement foreign inode flushing
Date: Mon, 26 Aug 2019 09:06:51 -0700	[thread overview]
Message-ID: <20190826160656.870307-1-tj@kernel.org> (raw)

Hello,

Changes from v1[1]:

* More comments explaining the parameters.

* 0003-writeback-Separate-out-wb_get_lookup-from-wb_get_create.patch
  added and avoid spuriously creating missing wbs for foreign
  flushing.

Changes from v2[2]:

* Added livelock avoidance and applied other smaller changes suggested
  by Jan.

There's an inherent mismatch between memcg and writeback.  The former
trackes ownership per-page while the latter per-inode.  This was a
deliberate design decision because honoring per-page ownership in the
writeback path is complicated, may lead to higher CPU and IO overheads
and deemed unnecessary given that write-sharing an inode across
different cgroups isn't a common use-case.

Combined with inode majority-writer ownership switching, this works
well enough in most cases but there are some pathological cases.  For
example, let's say there are two cgroups A and B which keep writing to
different but confined parts of the same inode.  B owns the inode and
A's memory is limited far below B's.  A's dirty ratio can rise enough
to trigger balance_dirty_pages() sleeps but B's can be low enough to
avoid triggering background writeback.  A will be slowed down without
a way to make writeback of the dirty pages happen.

This patchset implements foreign dirty recording and foreign mechanism
so that when a memcg encounters a condition as above it can trigger
flushes on bdi_writebacks which can clean its pages.  Please see the
last patch for more details.

This patchset contains the following four patches.

 0001-writeback-Generalize-and-expose-wb_completion.patch
 0002-bdi-Add-bdi-id.patch
 0003-writeback-Separate-out-wb_get_lookup-from-wb_get_create.patch
 0004-writeback-memcg-Implement-cgroup_writeback_by_id.patch
 0005-writeback-memcg-Implement-foreign-dirty-flushing.patch

0001-0004 are prep patches which expose wb_completion and implement
bdi->id and flushing by bdi and memcg IDs.

0005 implements foreign inode flushing.

Thanks.  diffstat follows.

 fs/fs-writeback.c                |  130 ++++++++++++++++++++++++++++---------
 include/linux/backing-dev-defs.h |   23 ++++++
 include/linux/backing-dev.h      |    5 +
 include/linux/memcontrol.h       |   39 +++++++++++
 include/linux/writeback.h        |    2 
 mm/backing-dev.c                 |  120 +++++++++++++++++++++++++++++-----
 mm/memcontrol.c                  |  134 +++++++++++++++++++++++++++++++++++++++
 mm/page-writeback.c              |    4 +
 8 files changed, 404 insertions(+), 53 deletions(-)

--
tejun

[1] http://lkml.kernel.org/r/20190803140155.181190-1-tj@kernel.org
[2] http://lkml.kenrel.org/r/20190815195619.GA2263813@devbig004.ftw2.facebook.com