linux-stable.git/mm/memcontrol.c, branch linux-3.13.y

memcg: reparent charges of children before processing parent

2014-03-24T04:44:20+00:00

commit 4fb1a86fb5e4209a7d4426d4e586c58e9edc74ac upstream.

Sometimes the cleanup after memcg hierarchy testing gets stuck in
mem_cgroup_reparent_charges(), unable to bring non-kmem usage down to 0.

There may turn out to be several causes, but a major cause is this: the
workitem to offline parent can get run before workitem to offline child;
parent's mem_cgroup_reparent_charges() circles around waiting for the
child's pages to be reparented to its lrus, but it's holding
cgroup_mutex which prevents the child from reaching its
mem_cgroup_reparent_charges().

Further testing showed that an ordered workqueue for cgroup_destroy_wq
is not always good enough: percpu_ref_kill_and_confirm's call_rcu_sched
stage on the way can mess up the order before reaching the workqueue.

Instead, when offlining a memcg, call mem_cgroup_reparent_charges() on
all its children (and grandchildren, in the correct order) to have their
charges reparented first.

Fixes: e5fca243abae ("cgroup: use a dedicated workqueue for cgroup destruction")
Signed-off-by: Filipe Brandenburger 
Signed-off-by: Hugh Dickins 
Reviewed-by: Tejun Heo 
Acked-by: Michal Hocko 
Cc: Johannes Weiner 
Cc: 	[v3.10+]
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

memcg: fix endless loop in __mem_cgroup_iter_next()

2014-03-24T04:43:59+00:00

commit ce48225fe3b1b0d1fc9fceb96ac3d8a879e45114 upstream.

Commit 0eef615665ed ("memcg: fix css reference leak and endless loop in
mem_cgroup_iter") got the interaction with the commit a few before it
d8ad30559715 ("mm/memcg: iteration skip memcgs not yet fully
initialized") slightly wrong, and we didn't notice at the time.

It's elusive, and harder to get than the original, but for a couple of
days before rc1, I several times saw a endless loop similar to that
supposedly being fixed.

This time it was a tighter loop in __mem_cgroup_iter_next(): because we
can get here when our root has already been offlined, and the ordering
of conditions was such that we then just cycled around forever.

Fixes: 0eef615665ed ("memcg: fix css reference leak and endless loop in mem_cgroup_iter").
Signed-off-by: Hugh Dickins 
Acked-by: Michal Hocko 
Cc: Johannes Weiner 
Cc: Greg Thelen 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

memcg: fix css reference leak and endless loop in mem_cgroup_iter

2014-02-13T21:55:29+00:00

commit 0eef615665ede1e0d603ea9ecca88c1da6f02234 upstream.

Commit 19f39402864e ("memcg: simplify mem_cgroup_iter") has reorganized
mem_cgroup_iter code in order to simplify it.  A part of that change was
dropping an optimization which didn't call css_tryget on the root of the
walked tree.  The patch however didn't change the css_put part in
mem_cgroup_iter which excludes root.

This wasn't an issue at the time because __mem_cgroup_iter_next bailed
out for root early without taking a reference as cgroup iterators
(css_next_descendant_pre) didn't visit root themselves.

Nevertheless cgroup iterators have been reworked to visit root by commit
bd8815a6d802 ("cgroup: make css_for_each_descendant() and friends
include the origin css in the iteration") when the root bypass have been
dropped in __mem_cgroup_iter_next.  This means that css_put is not
called for root and so css along with mem_cgroup and other cgroup
internal object tied by css lifetime are never freed.

Fix the issue by reintroducing root check in __mem_cgroup_iter_next and
do not take css reference for it.

This reference counting magic protects us also from another issue, an
endless loop reported by Hugh Dickins when reclaim races with root
removal and css_tryget called by iterator internally would fail.  There
would be no other nodes to visit so __mem_cgroup_iter_next would return
NULL and mem_cgroup_iter would interpret it as "start looping from root
again" and so mem_cgroup_iter would loop forever internally.

Signed-off-by: Michal Hocko 
Reported-by: Hugh Dickins 
Tested-by: Hugh Dickins 
Cc: Johannes Weiner 
Cc: Greg Thelen 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

memcg: fix endless loop caused by mem_cgroup_iter

2014-02-13T21:55:29+00:00

commit ecc736fc3c71c411a9d201d8588c9e7e049e5d8c upstream.

Hugh has reported an endless loop when the hardlimit reclaim sees the
same group all the time.  This might happen when the reclaim races with
the memcg removal.

shrink_zone
                                                [rmdir root]
  mem_cgroup_iter(root, NULL, reclaim)
    // prev = NULL
    rcu_read_lock()
    mem_cgroup_iter_load
      last_visited = iter->last_visited   // gets root || NULL
      css_tryget(last_visited)            // failed
      last_visited = NULL                 [1]
    memcg = root = __mem_cgroup_iter_next(root, NULL)
    mem_cgroup_iter_update
      iter->last_visited = root;
    reclaim->generation = iter->generation

 mem_cgroup_iter(root, root, reclaim)
   // prev = root
   rcu_read_lock
    mem_cgroup_iter_load
      last_visited = iter->last_visited   // gets root
      css_tryget(last_visited)            // failed
    [1]

The issue seemed to be introduced by commit 5f5781619718 ("memcg: relax
memcg iter caching") which has replaced unconditional css_get/css_put by
css_tryget/css_put for the cached iterator.

This patch fixes the issue by skipping css_tryget on the root of the
tree walk in mem_cgroup_iter_load and symmetrically doesn't release it
in mem_cgroup_iter_update.

Signed-off-by: Michal Hocko 
Reported-by: Hugh Dickins 
Tested-by: Hugh Dickins 
Cc: Johannes Weiner 
Cc: Greg Thelen 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

mm/memcg: iteration skip memcgs not yet fully initialized

2014-02-13T21:55:28+00:00

commit d8ad30559715ce97afb7d1a93a12fd90e8fff312 upstream.

It is surprising that the mem_cgroup iterator can return memcgs which
have not yet been fully initialized.  By accident (or trial and error?)
this appears not to present an actual problem; but it may be better to
prevent such surprises, by skipping memcgs not yet online.

Signed-off-by: Hugh Dickins 
Cc: Tejun Heo 
Acked-by: Michal Hocko 
Cc: Johannes Weiner 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

memcg: fix memcg_size() calculation

2014-01-02T22:40:30+00:00

The mem_cgroup structure contains nr_node_ids pointers to
mem_cgroup_per_node objects, not the objects themselves.

Signed-off-by: Vladimir Davydov 
Acked-by: Michal Hocko 
Cc: Glauber Costa 
Cc: Johannes Weiner 
Cc: Balbir Singh 
Cc: KAMEZAWA Hiroyuki 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm: memcg: do not allow task about to OOM kill to bypass the limit

2013-12-13T02:19:26+00:00

Commit 4942642080ea ("mm: memcg: handle non-error OOM situations more
gracefully") allowed tasks that already entered a memcg OOM condition to
bypass the memcg limit on subsequent allocation attempts hoping this
would expedite finishing the page fault and executing the kill.

David Rientjes is worried that this breaks memcg isolation guarantees
and since there is no evidence that the bypass actually speeds up fault
processing just change it so that these subsequent charge attempts fail
outright.  The notable exception being __GFP_NOFAIL charges which are
required to bypass the limit regardless.

Signed-off-by: Johannes Weiner 
Reported-by: David Rientjes 
Acked-by: Michal Hocko 
Acked-bt: David Rientjes 
Cc: 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm: memcg: fix race condition between memcg teardown and swapin

2013-12-13T02:19:26+00:00

There is a race condition between a memcg being torn down and a swapin
triggered from a different memcg of a page that was recorded to belong
to the exiting memcg on swapout (with CONFIG_MEMCG_SWAP extension).  The
result is unreclaimable pages pointing to dead memcgs, which can lead to
anything from endless loops in later memcg teardown (the page is charged
to all hierarchical parents but is not on any LRU list) or crashes from
following the dangling memcg pointer.

Memcgs with tasks in them can not be torn down and usually charges don't
show up in memcgs without tasks.  Swapin with the CONFIG_MEMCG_SWAP
extension is the notable exception because it charges the cgroup that
was recorded as owner during swapout, which may be empty and in the
process of being torn down when a task in another memcg triggers the
swapin:

  teardown:                 swapin:

                            lookup_swap_cgroup_id()
                            rcu_read_lock()
                            mem_cgroup_lookup()
                            css_tryget()
                            rcu_read_unlock()
  disable css_tryget()
  call_rcu()
    offline_css()
      reparent_charges()
                            res_counter_charge() (hierarchical!)
                            css_put()
                              css_free()
                            pc->mem_cgroup = dead memcg
                            add page to dead lru

Add a final reparenting step into css_free() to make sure any such raced
charges are moved out of the memcg before it's finally freed.

In the longer term it would be cleaner to have the css_tryget() and the
res_counter charge under the same RCU lock section so that the charge
reparenting is deferred until the last charge whose tryget succeeded is
visible.  But this will require more invasive changes that will be
harder to evaluate and backport into stable, so better defer them to a
separate change set.

Signed-off-by: Johannes Weiner 
Acked-by: Michal Hocko 
Cc: David Rientjes 
Cc: 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm: memcg: do not declare OOM from __GFP_NOFAIL allocations

2013-12-13T02:19:26+00:00

Commit 84235de394d9 ("fs: buffer: move allocation failure loop into the
allocator") started recognizing __GFP_NOFAIL in memory cgroups but
forgot to disable the OOM killer.

Any task that does not fail allocation will also not enter the OOM
completion path.  So don't declare an OOM state in this case or it'll be
leaked and the task be able to bypass the limit until the next
userspace-triggered page fault cleans up the OOM state.

Reported-by: William Dauchy 
Signed-off-by: Johannes Weiner 
Acked-by: Michal Hocko 
Cc: David Rientjes 
Cc: 	[3.12.x]
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

mm, thp: change pmd_trans_huge_lock() to return taken lock

2013-11-15T00:32:14+00:00

With split ptlock it's important to know which lock
pmd_trans_huge_lock() took.  This patch adds one more parameter to the
function to return the lock.

In most places migration to new api is trivial.  Exception is
move_huge_pmd(): we need to take two locks if pmd tables are different.

Signed-off-by: Naoya Horiguchi 
Signed-off-by: Kirill A. Shutemov 
Tested-by: Alex Thorlton 
Cc: Ingo Molnar 
Cc: "Eric W . Biederman" 
Cc: "Paul E . McKenney" 
Cc: Al Viro 
Cc: Andi Kleen 
Cc: Andrea Arcangeli 
Cc: Dave Hansen 
Cc: Dave Jones 
Cc: David Howells 
Cc: Frederic Weisbecker 
Cc: Johannes Weiner 
Cc: Kees Cook 
Cc: Mel Gorman 
Cc: Michael Kerrisk 
Cc: Oleg Nesterov 
Cc: Peter Zijlstra 
Cc: Rik van Riel 
Cc: Robin Holt 
Cc: Sedat Dilek 
Cc: Srikar Dronamraju 
Cc: Thomas Gleixner 
Cc: Hugh Dickins 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds