linux-toradex.git/mm, branch v3.10.39

mm: hugetlb: fix softlockup when a large number of hugepages are freed.

2014-05-06T14:55:32+00:00

commit 55f67141a8927b2be3e51840da37b8a2320143ed upstream.

When I decrease the value of nr_hugepage in procfs a lot, softlockup
happens.  It is because there is no chance of context switch during this
process.

On the other hand, when I allocate a large number of hugepages, there is
some chance of context switch.  Hence softlockup doesn't happen during
this process.  So it's necessary to add the context switch in the
freeing process as same as allocating process to avoid softlockup.

When I freed 12 TB hugapages with kernel-2.6.32-358.el6, the freeing
process occupied a CPU over 150 seconds and following softlockup message
appeared twice or more.

$ echo 6000000 > /proc/sys/vm/nr_hugepages
$ cat /proc/sys/vm/nr_hugepages
6000000
$ grep ^Huge /proc/meminfo
HugePages_Total:   6000000
HugePages_Free:    6000000
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:       2048 kB
$ echo 0 > /proc/sys/vm/nr_hugepages

BUG: soft lockup - CPU#16 stuck for 67s! [sh:12883] ...
Pid: 12883, comm: sh Not tainted 2.6.32-358.el6.x86_64 #1
Call Trace:
  free_pool_huge_page+0xb8/0xd0
  set_max_huge_pages+0x128/0x190
  hugetlb_sysctl_handler_common+0x113/0x140
  hugetlb_sysctl_handler+0x1e/0x20
  proc_sys_call_handler+0x97/0xd0
  proc_sys_write+0x14/0x20
  vfs_write+0xb8/0x1a0
  sys_write+0x51/0x90
  __audit_syscall_exit+0x265/0x290
  system_call_fastpath+0x16/0x1b

I have not confirmed this problem with upstream kernels because I am not
able to prepare the machine equipped with 12TB memory now.  However I
confirmed that the amount of decreasing hugepages was directly
proportional to the amount of required time.

I measured required times on a smaller machine.  It showed 130-145
hugepages decreased in a millisecond.

  Amount of decreasing     Required time      Decreasing rate
  hugepages                     (msec)         (pages/msec)
  ------------------------------------------------------------
  10,000 pages == 20GB         70 -  74          135-142
  30,000 pages == 60GB        208 - 229          131-144

It means decrement of 6TB hugepages will trigger softlockup with the
default threshold 20sec, in this decreasing rate.

Signed-off-by: Masayoshi Mizuma 
Cc: Joonsoo Kim 
Cc: Michal Hocko 
Cc: Wanpeng Li 
Cc: Aneesh Kumar 
Cc: KOSAKI Motohiro 
Cc: Naoya Horiguchi 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

mm: try_to_unmap_cluster() should lock_page() before mlocking

2014-05-06T14:55:32+00:00

commit 57e68e9cd65b4b8eb4045a1e0d0746458502554c upstream.

A BUG_ON(!PageLocked) was triggered in mlock_vma_page() by Sasha Levin
fuzzing with trinity.  The call site try_to_unmap_cluster() does not lock
the pages other than its check_page parameter (which is already locked).

The BUG_ON in mlock_vma_page() is not documented and its purpose is
somewhat unclear, but apparently it serializes against page migration,
which could otherwise fail to transfer the PG_mlocked flag.  This would
not be fatal, as the page would be eventually encountered again, but
NR_MLOCK accounting would become distorted nevertheless.  This patch adds
a comment to the BUG_ON in mlock_vma_page() and munlock_vma_page() to that
effect.

The call site try_to_unmap_cluster() is fixed so that for page !=
check_page, trylock_page() is attempted (to avoid possible deadlocks as we
already have check_page locked) and mlock_vma_page() is performed only
upon success.  If the page lock cannot be obtained, the page is left
without PG_mlocked, which is again not a problem in the whole unevictable
memory design.

Signed-off-by: Vlastimil Babka 
Signed-off-by: Bob Liu 
Reported-by: Sasha Levin 
Cc: Wanpeng Li 
Cc: Michel Lespinasse 
Cc: KOSAKI Motohiro 
Acked-by: Rik van Riel 
Cc: David Rientjes 
Cc: Mel Gorman 
Cc: Hugh Dickins 
Cc: Joonsoo Kim 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

bdi: avoid oops on device removal

2014-04-27T00:15:35+00:00

commit 5acda9d12dcf1ad0d9a5a2a7c646de3472fa7555 upstream.

After commit 839a8e8660b6 ("writeback: replace custom worker pool
implementation with unbound workqueue") when device is removed while we
are writing to it we crash in bdi_writeback_workfn() ->
set_worker_desc() because bdi->dev is NULL.

This can happen because even though bdi_unregister() cancels all pending
flushing work, nothing really prevents new ones from being queued from
balance_dirty_pages() or other places.

Fix the problem by clearing BDI_registered bit in bdi_unregister() and
checking it before scheduling of any flushing work.

Fixes: 839a8e8660b6777e7fe4e80af1a048aebe2b5977

Reviewed-by: Tejun Heo 
Signed-off-by: Jan Kara 
Cc: Derek Basehore 
Cc: Jens Axboe 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

backing_dev: fix hung task on sync

2014-04-27T00:15:35+00:00

commit 6ca738d60c563d5c6cf6253ee4b8e76fa77b2b9e upstream.

bdi_wakeup_thread_delayed() used the mod_delayed_work() function to
schedule work to writeback dirty inodes.  The problem with this is that
it can delay work that is scheduled for immediate execution, such as the
work from sync_inodes_sb().  This can happen since mod_delayed_work()
can now steal work from a work_queue.  This fixes the problem by using
queue_delayed_work() instead.  This is a regression caused by commit
839a8e8660b6 ("writeback: replace custom worker pool implementation with
unbound workqueue").

The reason that this causes a problem is that laptop-mode will change
the delay, dirty_writeback_centisecs, to 60000 (10 minutes) by default.
In the case that bdi_wakeup_thread_delayed() races with
sync_inodes_sb(), sync will be stopped for 10 minutes and trigger a hung
task.  Even if dirty_writeback_centisecs is not long enough to cause a
hung task, we still don't want to delay sync for that long.

We fix the problem by using queue_delayed_work() when we want to
schedule writeback sometime in future.  This function doesn't change the
timer if it is already armed.

For the same reason, we also change bdi_writeback_workfn() to
immediately queue the work again in the case that the work_list is not
empty.  The same problem can happen if the sync work is run on the
rescue worker.

[jack@suse.cz: update changelog, add comment, use bdi_wakeup_thread_delayed()]
Signed-off-by: Derek Basehore 
Reviewed-by: Jan Kara 
Cc: Alexander Viro 
Reviewed-by: Tejun Heo 
Cc: Greg Kroah-Hartman 
Cc: "Darrick J. Wong" 
Cc: Derek Basehore 
Cc: Kees Cook 
Cc: Benson Leung 
Cc: Sonny Rao 
Cc: Luigi Semenzato 
Cc: Jens Axboe 
Cc: Dave Chinner 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

mm: close PageTail race

2014-04-03T19:01:05+00:00

commit 668f9abbd4334e6c29fa8acd71635c4f9101caa7 upstream.

Commit bf6bddf1924e ("mm: introduce compaction and migration for
ballooned pages") introduces page_count(page) into memory compaction
which dereferences page->first_page if PageTail(page).

This results in a very rare NULL pointer dereference on the
aforementioned page_count(page).  Indeed, anything that does
compound_head(), including page_count() is susceptible to racing with
prep_compound_page() and seeing a NULL or dangling page->first_page
pointer.

This patch uses Andrea's implementation of compound_trans_head() that
deals with such a race and makes it the default compound_head()
implementation.  This includes a read memory barrier that ensures that
if PageTail(head) is true that we return a head page that is neither
NULL nor dangling.  The patch then adds a store memory barrier to
prep_compound_page() to ensure page->first_page is set.

This is the safest way to ensure we see the head page that we are
expecting, PageTail(page) is already in the unlikely() path and the
memory barriers are unfortunately required.

Hugetlbfs is the exception, we don't enforce a store memory barrier
during init since no race is possible.

Signed-off-by: David Rientjes 
Cc: Holger Kiehl 
Cc: Christoph Lameter 
Cc: Rafael Aquini 
Cc: Vlastimil Babka 
Cc: Michal Hocko 
Cc: Mel Gorman 
Cc: Andrea Arcangeli 
Cc: Rik van Riel 
Cc: "Kirill A. Shutemov" 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

memcg: reparent charges of children before processing parent

2014-03-24T04:38:20+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.

[The version for 3.10.34 (or perhaps now 3.10.35) is this below.
Yes, more differences, and the old mem_cgroup_reparent_charges line
is intentionally left in for 3.10 whereas it was removed for 3.12+:
that's because the css/cgroup iterator changed in between, it used
not to supply the root of the subtree, but nowadays it does - Hugh]


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 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

mm/compaction: break out of loop on !PageBuddy in isolate_freepages_block

2014-03-24T04:38:18+00:00

commit 2af120bc040c5ebcda156df6be6a66610ab6957f upstream.

We received several reports of bad page state when freeing CMA pages
previously allocated with alloc_contig_range:

    BUG: Bad page state in process Binder_A  pfn:63202
    page:d21130b0 count:0 mapcount:1 mapping:  (null) index:0x7dfbf
    page flags: 0x40080068(uptodate|lru|active|swapbacked)

Based on the page state, it looks like the page was still in use.  The
page flags do not make sense for the use case though.  Further debugging
showed that despite alloc_contig_range returning success, at least one
page in the range still remained in the buddy allocator.

There is an issue with isolate_freepages_block.  In strict mode (which
CMA uses), if any pages in the range cannot be isolated,
isolate_freepages_block should return failure 0.  The current check
keeps track of the total number of isolated pages and compares against
the size of the range:

        if (strict && nr_strict_required > total_isolated)
                total_isolated = 0;

After taking the zone lock, if one of the pages in the range is not in
the buddy allocator, we continue through the loop and do not increment
total_isolated.  If in the last iteration of the loop we isolate more
than one page (e.g.  last page needed is a higher order page), the check
for total_isolated may pass and we fail to detect that a page was
skipped.  The fix is to bail out if the loop immediately if we are in
strict mode.  There's no benfit to continuing anyway since we need all
pages to be isolated.  Additionally, drop the error checking based on
nr_strict_required and just check the pfn ranges.  This matches with
what isolate_freepages_range does.

Signed-off-by: Laura Abbott 
Acked-by: Minchan Kim 
Cc: Mel Gorman 
Acked-by: Vlastimil Babka 
Cc: Joonsoo Kim 
Acked-by: Bartlomiej Zolnierkiewicz 
Acked-by: Michal Nazarewicz 
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

2014-03-07T05:30:05+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 
Cc: 	[3.10+]
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

mm/memory-failure.c: move refcount only in !MF_COUNT_INCREASED

2014-02-22T20:41:26+00:00

commit 8d547ff4ac5927245e0833ac18528f939da0ee0e upstream.

mce-test detected a test failure when injecting error to a thp tail
page.  This is because we take page refcount of the tail page in
madvise_hwpoison() while the fix in commit a3e0f9e47d5e
("mm/memory-failure.c: transfer page count from head page to tail page
after split thp") assumes that we always take refcount on the head page.

When a real memory error happens we take refcount on the head page where
memory_failure() is called without MF_COUNT_INCREASED set, so it seems
to me that testing memory error on thp tail page using madvise makes
little sense.

This patch cancels moving refcount in !MF_COUNT_INCREASED for valid
testing.

[akpm@linux-foundation.org: s/&&/&/]
Signed-off-by: Naoya Horiguchi 
Cc: Andi Kleen 
Cc: Wanpeng Li 
Cc: Chen Gong 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Greg Kroah-Hartman

mm: fix process accidentally killed by mce because of huge page migration

2014-02-20T19:06:12+00:00

Based on c8721bbbdd36382de51cd6b7a56322e0acca2414 upstream, but only the
bugfix portion pulled out.

Hi Naoya or Greg,

We found a bug in 3.10.x.
The problem is that we accidentally have a hwpoisoned hugepage in free
hugepage list. It could happend in the the following scenario:

        process A                           process B

  migrate_huge_page
  put_page (old hugepage)
    linked to free hugepage list
                                     hugetlb_fault
                                       hugetlb_no_page
                                         alloc_huge_page
                                           dequeue_huge_page_vma
                                             dequeue_huge_page_node
                                               (steal hwpoisoned hugepage)
  set_page_hwpoison_huge_page
  dequeue_hwpoisoned_huge_page
    (fail to dequeue)

I tested this bug, one process keeps allocating huge page, and I 
use sysfs interface to soft offline a huge page, then received:
"MCE: Killing UCP:2717 due to hardware memory corruption fault at 8200034"

Upstream kernel is free from this bug because of these two commits:

f15bdfa802bfa5eb6b4b5a241b97ec9fa1204a35
mm/memory-failure.c: fix memory leak in successful soft offlining

c8721bbbdd36382de51cd6b7a56322e0acca2414
mm: memory-hotplug: enable memory hotplug to handle hugepage

The first one, although the problem is about memory leak, this patch
moves unset_migratetype_isolate(), which is important to avoid the race.
The latter is not a bug fix and it's too big, so I rewrite a small one.

The following patch can fix this bug.(please apply f15bdfa802bf first)

Signed-off-by: Xishi Qiu 
Reviewed-by: Naoya Horiguchi 
Signed-off-by: Greg Kroah-Hartman