__list_lru_walk_one() acquires nlru spin lock (nlru->lock) for longer
duration if there are more number of items in the lru list. As per the
current code, it can hold the spin lock for upto maximum UINT_MAX
entries at a time. So if there are more number of items in the lru
list, then "BUG: spinlock lockup suspected" is observed in the below
path:
spin_bug+0x90
do_raw_spin_lock+0xfc
_raw_spin_lock+0x28
list_lru_add+0x28
dput+0x1c8
path_put+0x20
terminate_walk+0x3c
path_lookupat+0x100
filename_lookup+0x6c
user_path_at_empty+0x54
SyS_faccessat+0xd0
el0_svc_naked+0x24
This nlru->lock is acquired by another CPU in this path -
d_lru_shrink_move+0x34
dentry_lru_isolate_shrink+0x48
__list_lru_walk_one.isra.10+0x94
list_lru_walk_node+0x40
shrink_dcache_sb+0x60
do_remount_sb+0xbc
do_emergency_remount+0xb0
process_one_work+0x228
worker_thread+0x2e0
kthread+0xf4
ret_from_fork+0x10
Fix this lockup by reducing the number of entries to be shrinked from
the lru list to 1024 at once. Also, add cond_resched() before
processing the lru list again.
Link: http://marc.info/?t=149722864900001&r=1&w=2
Link: http://lkml.kernel.org/r/1498707575-2472-1-git-send-email-stummala@codeaurora.org
Signed-off-by: Sahitya Tummala <stummala@codeaurora.org>
Suggested-by: Jan Kara <jack@suse.cz>
Suggested-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Alexander Polakov <apolyakov@beget.ru>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
list_lru_count_node() iterates over all memcgs to get the total number of
entries on the node but it can race with memcg_drain_all_list_lrus(),
which migrates the entries from a dead cgroup to another. This can return
incorrect number of entries from list_lru_count_node().
Fix this by keeping track of entries per node and simply return it in
list_lru_count_node().
Link: http://lkml.kernel.org/r/1498707555-30525-1-git-send-email-stummala@codeaurora.org
Signed-off-by: Sahitya Tummala <stummala@codeaurora.org>
Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Alexander Polakov <apolyakov@beget.ru>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
expand_stack(vma) fails if address < stack_guard_gap even if there is no
vma->vm_prev. I don't think this makes sense, and we didn't do this
before the recent commit 1be7107fbe ("mm: larger stack guard gap,
between vmas").
We do not need a gap in this case, any address is fine as long as
security_mmap_addr() doesn't object.
This also simplifies the code, we know that address >= prev->vm_end and
thus underflow is not possible.
Link: http://lkml.kernel.org/r/20170628175258.GA24881@redhat.com
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 1be7107fbe ("mm: larger stack guard gap, between vmas") has
introduced a regression in some rust and Java environments which are
trying to implement their own stack guard page. They are punching a new
MAP_FIXED mapping inside the existing stack Vma.
This will confuse expand_{downwards,upwards} into thinking that the
stack expansion would in fact get us too close to an existing non-stack
vma which is a correct behavior wrt safety. It is a real regression on
the other hand.
Let's work around the problem by considering PROT_NONE mapping as a part
of the stack. This is a gros hack but overflowing to such a mapping
would trap anyway an we only can hope that usespace knows what it is
doing and handle it propely.
Fixes: 1be7107fbe ("mm: larger stack guard gap, between vmas")
Link: http://lkml.kernel.org/r/20170705182849.GA18027@dhcp22.suse.cz
Signed-off-by: Michal Hocko <mhocko@suse.com>
Debugged-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Ben Hutchings <ben@decadent.org.uk>
Cc: Willy Tarreau <w@1wt.eu>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
presently pages in the balloon device have random value, and these pages
will be scanned by ksmd on the host. They usually cannot be merged.
Enqueue zero pages will resolve this problem.
Link: http://lkml.kernel.org/r/1498698637-26389-1-git-send-email-zhenwei.pi@youruncloud.com
Signed-off-by: zhenwei.pi <zhenwei.pi@youruncloud.com>
Cc: Gioh Kim <gi-oh.kim@profitbricks.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Cc: Rafael Aquini <aquini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The align_offset parameter is used by bitmap_find_next_zero_area_off()
to represent the offset of map's base from the previous alignment
boundary; the function ensures that the returned index, plus the
align_offset, honors the specified align_mask.
The logic introduced by commit b5be83e308 ("mm: cma: align to physical
address, not CMA region position") has the cma driver calculate the
offset to the *next* alignment boundary. In most cases, the base
alignment is greater than that specified when making allocations,
resulting in a zero offset whether we align up or down. In the example
given with the commit, the base alignment (8MB) was half the requested
alignment (16MB) so the math also happened to work since the offset is
8MB in both directions. However, when requesting allocations with an
alignment greater than twice that of the base, the returned index would
not be correctly aligned.
Also, the align_order arguments of cma_bitmap_aligned_mask() and
cma_bitmap_aligned_offset() should not be negative so the argument type
was made unsigned.
Fixes: b5be83e308 ("mm: cma: align to physical address, not CMA region position")
Link: http://lkml.kernel.org/r/20170628170742.2895-1-opendmb@gmail.com
Signed-off-by: Angus Clark <angus@angusclark.org>
Signed-off-by: Doug Berger <opendmb@gmail.com>
Acked-by: Gregory Fong <gregory.0xf0@gmail.com>
Cc: Doug Berger <opendmb@gmail.com>
Cc: Angus Clark <angus@angusclark.org>
Cc: Laura Abbott <labbott@redhat.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Lucas Stach <l.stach@pengutronix.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Shiraz Hashim <shashim@codeaurora.org>
Cc: Jaewon Kim <jaewon31.kim@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
__remove_zone() sets up up zone_type, but never uses it for anything.
This does not cause a warning, due to the (necessary) use of
-Wno-unused-but-set-variable. However, it's noise, so just delete it.
Link: http://lkml.kernel.org/r/20170624043421.24465-2-jhubbard@nvidia.com
Signed-off-by: John Hubbard <jhubbard@nvidia.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
It seems that there are still people using 32b kernels which a lot of
memory and the IO tend to suck a lot for them by default. Mostly
because writers are throttled too when the lowmem is used. We have
highmem_is_dirtyable to work around that issue but it seems we never
bothered to document it. Let's do it now, finally.
Link: http://lkml.kernel.org/r/20170626093200.18958-1-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Alkis Georgopoulos <alkisg@gmail.com>
Cc: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
wb_stat_sum() disables interrupts and calls __wb_stat_sum() which
eventually calls __percpu_counter_sum(). However, the percpu routine is
already irq-safe. Simplify the code a bit by making wb_stat_sum()
directly call percpu_counter_sum_positive() and not disable interrupts.
Also remove the now-uneeded __wb_stat_sum() which was just a wrapper
over percpu_counter_sum_positive().
Link: http://lkml.kernel.org/r/1498230681-29103-1-git-send-email-nborisov@suse.com
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Jan Kara <jack@suse.cz>
Cc: Jens Axboe <axboe@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently pg_data_t is just a struct which describes a NUMA node memory
layout. Let's keep the comment simple and remove ambiguity.
Link: http://lkml.kernel.org/r/1498220534-22717-1-git-send-email-nborisov@suse.com
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
get_cpu_var() disables preemption and returns the per-CPU version of the
variable. Disabling preemption is useful to ensure atomic access to the
variable within the critical section.
In this case however, after the per-CPU version of the variable is
obtained the ->free_lock is acquired. For that reason it seems the raw
accessor could be used. It only seems that ->slots_ret should be
retested (because with disabled preemption this variable can not be set
to NULL otherwise).
This popped up during PREEMPT-RT testing because it tries to take
spinlocks in a preempt disabled section. In RT, spinlocks can sleep.
Link: http://lkml.kernel.org/r/20170623114755.2ebxdysacvgxzott@linutronix.de
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ying Huang <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Since current_order starts as MAX_ORDER-1 and is then only decremented,
the second half of the loop condition seems superfluous. However, if
order is 0, we may decrement current_order past 0, making it UINT_MAX.
This is obviously too subtle ([1], [2]).
Since we need to add some comment anyway, change the two variables to
signed, making the counting-down for loop look more familiar, and
apparently also making gcc generate slightly smaller code.
[1] https://lkml.org/lkml/2016/6/20/493
[2] https://lkml.org/lkml/2017/6/19/345
[akpm@linux-foundation.org: fix up reject fixupping]
Link: http://lkml.kernel.org/r/20170621185529.2265-1-linux@rasmusvillemoes.dk
Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Reported-by: Hao Lee <haolee.swjtu@gmail.com>
Acked-by: Wei Yang <weiyang@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
__register_one_node() initializes local parameters "p_node" & "parent"
for register_node().
But, register_node() does not use them.
Remove the related code of "parent" node, cleanup __register_one_node()
and register_node().
Link: http://lkml.kernel.org/r/1498013846-20149-1-git-send-email-douly.fnst@cn.fujitsu.com
Signed-off-by: Dou Liyang <douly.fnst@cn.fujitsu.com>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
pagetypeinfo_showmixedcount_print is found to take a lot of time to
complete and it does this holding the zone lock and disabling
interrupts. In some cases it is found to take more than a second (On a
2.4GHz,8Gb RAM,arm64 cpu).
Avoid taking the zone lock similar to what is done by read_page_owner,
which means possibility of inaccurate results.
Link: http://lkml.kernel.org/r/1498045643-12257-1-git-send-email-vinmenon@codeaurora.org
Signed-off-by: Vinayak Menon <vinmenon@codeaurora.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: zhongjiang <zhongjiang@huawei.com>
Cc: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Sudip Mukherjee <sudipm.mukherjee@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: David Rientjes <rientjes@google.com>
Cc: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
new_page is yet another duplication of the migration callback which has
to handle hugetlb migration specially. We can safely use the generic
new_page_nodemask for the same purpose.
Please note that gigantic hugetlb pages do not need any special handling
because alloc_huge_page_nodemask will make sure to check pages in all
per node pools. The reason this was done previously was that
alloc_huge_page_node treated NO_NUMA_NODE and a specific node
differently and so alloc_huge_page_node(nid) would check on this
specific node.
Link: http://lkml.kernel.org/r/20170622193034.28972-4-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Reported-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Tested-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
alloc_huge_page_nodemask tries to allocate from any numa node in the
allowed node mask starting from lower numa nodes. This might lead to
filling up those low NUMA nodes while others are not used. We can
reduce this risk by introducing a concept of the preferred node similar
to what we have in the regular page allocator. We will start allocating
from the preferred nid and then iterate over all allowed nodes in the
zonelist order until we try them all.
This is mimicing the page allocator logic except it operates on per-node
mempools. dequeue_huge_page_vma already does this so distill the
zonelist logic into a more generic dequeue_huge_page_nodemask and use it
in alloc_huge_page_nodemask.
This will allow us to use proper per numa distance fallback also for
alloc_huge_page_node which can use alloc_huge_page_nodemask now and we
can get rid of alloc_huge_page_node helper which doesn't have any user
anymore.
Link: http://lkml.kernel.org/r/20170622193034.28972-3-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Tested-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "mm, hugetlb: allow proper node fallback dequeue".
While working on a hugetlb migration issue addressed in a separate
patchset[1] I have noticed that the hugetlb allocations from the
preallocated pool are quite subotimal.
[1] //lkml.kernel.org/r/20170608074553.22152-1-mhocko@kernel.org
There is no fallback mechanism implemented and no notion of preferred
node. I have tried to work around it but Vlastimil was right to push
back for a more robust solution. It seems that such a solution is to
reuse zonelist approach we use for the page alloctor.
This series has 3 patches. The first one tries to make hugetlb
allocation layers more clear. The second one implements the zonelist
hugetlb pool allocation and introduces a preferred node semantic which
is used by the migration callbacks. The last patch is a clean up.
This patch (of 3):
Hugetlb allocation path for fresh huge pages is unnecessarily complex
and it mixes different interfaces between layers.
__alloc_buddy_huge_page is the central place to perform a new
allocation. It checks for the hugetlb overcommit and then relies on
__hugetlb_alloc_buddy_huge_page to invoke the page allocator. This is
all good except that __alloc_buddy_huge_page pushes vma and address down
the callchain and so __hugetlb_alloc_buddy_huge_page has to deal with
two different allocation modes - one for memory policy and other node
specific (or to make it more obscure node non-specific) requests.
This just screams for a reorganization.
This patch pulls out all the vma specific handling up to
__alloc_buddy_huge_page_with_mpol where it belongs.
__alloc_buddy_huge_page will get nodemask argument and
__hugetlb_alloc_buddy_huge_page will become a trivial wrapper over the
page allocator.
In short:
__alloc_buddy_huge_page_with_mpol - memory policy handling
__alloc_buddy_huge_page - overcommit handling and accounting
__hugetlb_alloc_buddy_huge_page - page allocator layer
Also note that __hugetlb_alloc_buddy_huge_page and its cpuset retry loop
is not really needed because the page allocator already handles the
cpusets update.
Finally __hugetlb_alloc_buddy_huge_page had a special case for node
specific allocations (when no policy is applied and there is a node
given). This has relied on __GFP_THISNODE to not fallback to a different
node. alloc_huge_page_node is the only caller which relies on this
behavior so move the __GFP_THISNODE there.
Not only does this remove quite some code it also should make those
layers easier to follow and clear wrt responsibilities.
Link: http://lkml.kernel.org/r/20170622193034.28972-2-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Tested-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
During the debugging of the problem described in
https://lkml.org/lkml/2017/5/17/542 and fixed by Tetsuo Handa in
https://lkml.org/lkml/2017/5/19/383 , I've found that the existing debug
output is not really useful to understand issues related to the oom
reaper.
So, I assume, that adding some tracepoints might help with debugging of
similar issues.
Trace the following events:
1) a process is marked as an oom victim,
2) a process is added to the oom reaper list,
3) the oom reaper starts reaping process's mm,
4) the oom reaper finished reaping,
5) the oom reaper skips reaping.
How it works in practice? Below is an example which show how the problem
mentioned above can be found: one process is added twice to the
oom_reaper list:
$ cd /sys/kernel/debug/tracing
$ echo "oom:mark_victim" > set_event
$ echo "oom:wake_reaper" >> set_event
$ echo "oom:skip_task_reaping" >> set_event
$ echo "oom:start_task_reaping" >> set_event
$ echo "oom:finish_task_reaping" >> set_event
$ cat trace_pipe
allocate-502 [001] .... 91.836405: mark_victim: pid=502
allocate-502 [001] .N.. 91.837356: wake_reaper: pid=502
allocate-502 [000] .N.. 91.871149: wake_reaper: pid=502
oom_reaper-23 [000] .... 91.871177: start_task_reaping: pid=502
oom_reaper-23 [000] .N.. 91.879511: finish_task_reaping: pid=502
oom_reaper-23 [000] .... 91.879580: skip_task_reaping: pid=502
Link: http://lkml.kernel.org/r/20170530185231.GA13412@castle
Signed-off-by: Roman Gushchin <guro@fb.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
MADV_FREE is identical to MADV_DONTNEED from the point of view of uffd
monitor. The monitor has to stop handling #PF events in the range being
freed. We are reusing userfaultfd_remove callback along with the logic
required to re-get and re-validate the VMA which may change or disappear
because userfaultfd_remove releases mmap_sem.
Link: http://lkml.kernel.org/r/1497876311-18615-1-git-send-email-rppt@linux.vnet.ibm.com
Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The condition checking for THP straddling end of invalidated range is
wrong - it checks 'index' against 'end' but 'index' has been already
advanced to point to the end of THP and thus the condition can never be
true. As a result THP straddling 'end' has been fully invalidated.
Given the nature of invalidate_mapping_pages(), this could be only
performance issue. In fact, we are lucky the condition is wrong because
if it was ever true, we'd leave locked page behind.
Fix the condition checking for THP straddling 'end' and also properly
unlock the page. Also update the comment before the condition to
explain why we decide not to invalidate the page as it was not clear to
me and I had to ask Kirill.
Link: http://lkml.kernel.org/r/20170619124723.21656-1-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The hugetlb code has its own function to report human-readable sizes.
Convert it to use the shared string_get_size() function. This will lead
to a minor difference in user visible output (MiB/GiB instead of MB/GB),
but some would argue that's desirable anyway.
Link: http://lkml.kernel.org/r/20170606190350.GA20010@bombadil.infradead.org
Signed-off-by: Matthew Wilcox <mawilcox@microsoft.com>
Cc: Liam R. Howlett <Liam.Howlett@Oracle.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Cc: zhong jiang <zhongjiang@huawei.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Alice has reported the following UBSAN splat:
UBSAN: Undefined behaviour in mm/memcontrol.c:661:17
signed integer overflow:
-2147483644 - 2147483525 cannot be represented in type 'long int'
CPU: 1 PID: 11758 Comm: mybibtex2filena Tainted: P O 4.9.25-gentoo #4
Hardware name: XXXXXX, BIOS YYYYYY
Call Trace:
dump_stack+0x59/0x87
ubsan_epilogue+0xe/0x40
handle_overflow+0xbb/0xf0
__ubsan_handle_sub_overflow+0x12/0x20
memcg_check_events.isra.36+0x223/0x360
mem_cgroup_commit_charge+0x55/0x140
wp_page_copy+0x34e/0xb80
do_wp_page+0x1e6/0x1300
handle_mm_fault+0x88b/0x1990
__do_page_fault+0x2de/0x8a0
do_page_fault+0x1a/0x20
error_code+0x67/0x6c
The reason is that we subtract two signed types. Let's fix this by
truly mimicing time_after and cast the result of the subtraction.
Link: http://lkml.kernel.org/r/20170616150057.GQ30580@dhcp22.suse.cz
Signed-off-by: Michal Hocko <mhocko@suse.com>
Reported-by: Alice Ferrazzi <alicef@gentoo.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
A few hugetlb allocators loop while calling the page allocator and can
potentially prevent rescheduling if the page allocator slowpath is not
utilized.
Conditionally schedule when large numbers of hugepages can be allocated.
Anshuman:
"Fixes a task which was getting hung while writing like 10000 hugepages
(16MB on POWER8) into /proc/sys/vm/nr_hugepages."
Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1706091535300.66176@chino.kir.corp.google.com
Signed-off-by: David Rientjes <rientjes@google.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Tested-by: Anshuman Khandual <khandual@linux.vnet.ibm.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 394e31d2ce ("mem-hotplug: alloc new page from a nearest
neighbor node when mem-offline") has duplicated a large part of
alloc_migrate_target with some hotplug specific special casing.
To be more precise it tried to enfore the allocation from a different
node than the original page. As a result the two function diverged in
their shared logic, e.g. the hugetlb allocation strategy.
Let's unify the two and express different NUMA requirements by the given
nodemask. new_node_page will simply exclude the node it doesn't care
about and alloc_migrate_target will use all the available nodes.
alloc_migrate_target will then learn to migrate hugetlb pages more
sanely and use preallocated pool when possible.
Please note that alloc_migrate_target used to call alloc_page resp.
alloc_pages_current so the memory policy of the current context which is
quite strange when we consider that it is used in the context of
alloc_contig_range which just tries to migrate pages which stand in the
way.
Link: http://lkml.kernel.org/r/20170608074553.22152-4-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Xishi Qiu <qiuxishi@huawei.com>
Cc: zhong jiang <zhongjiang@huawei.com>
Cc: Joonsoo Kim <js1304@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
new_node_page will try to use the origin's next NUMA node as the
migration destination for hugetlb pages. If such a node doesn't have
any preallocated pool it falls back to __alloc_buddy_huge_page_no_mpol
to allocate a surplus page instead. This is quite subotpimal for any
configuration when hugetlb pages are no distributed to all NUMA nodes
evenly. Say we have a hotplugable node 4 and spare hugetlb pages are
node 0
/sys/devices/system/node/node0/hugepages/hugepages-2048kB/nr_hugepages:10000
/sys/devices/system/node/node1/hugepages/hugepages-2048kB/nr_hugepages:0
/sys/devices/system/node/node2/hugepages/hugepages-2048kB/nr_hugepages:0
/sys/devices/system/node/node3/hugepages/hugepages-2048kB/nr_hugepages:0
/sys/devices/system/node/node4/hugepages/hugepages-2048kB/nr_hugepages:10000
/sys/devices/system/node/node5/hugepages/hugepages-2048kB/nr_hugepages:0
/sys/devices/system/node/node6/hugepages/hugepages-2048kB/nr_hugepages:0
/sys/devices/system/node/node7/hugepages/hugepages-2048kB/nr_hugepages:0
Now we consume the whole pool on node 4 and try to offline this node.
All the allocated pages should be moved to node0 which has enough
preallocated pages to hold them. With the current implementation
offlining very likely fails because hugetlb allocations during runtime
are much less reliable.
Fix this by reusing the nodemask which excludes migration source and try
to find a first node which has a page in the preallocated pool first and
fall back to __alloc_buddy_huge_page_no_mpol only when the whole pool is
consumed.
[akpm@linux-foundation.org: remove bogus arg from alloc_huge_page_nodemask() stub]
Link: http://lkml.kernel.org/r/20170608074553.22152-3-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Xishi Qiu <qiuxishi@huawei.com>
Cc: zhong jiang <zhongjiang@huawei.com>
Cc: Joonsoo Kim <js1304@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
new_node_page tries to allocate the target page on a different NUMA node
than the source page. This makes sense in most cases during the hotplug
because we are likely to offline the whole numa node. But there are
cases where there are no other nodes to fallback (e.g. when offlining
parts of the only existing node) and we have to fallback to allocating
from the source node. The current code does that but it can be
simplified by checking the nmask and updating it before we even try to
allocate rather than special casing it.
This patch shouldn't introduce any functional change.
Link: http://lkml.kernel.org/r/20170608074553.22152-2-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Xishi Qiu <qiuxishi@huawei.com>
Cc: zhong jiang <zhongjiang@huawei.com>
Cc: Joonsoo Kim <js1304@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
movable_node kernel parameter allows making hotpluggable NUMA nodes to
put all the hotplugable memory into movable zone which allows more or
less reliable memory hotremove. At least this is the case for the NUMA
nodes present during the boot (see find_zone_movable_pfns_for_nodes).
This is not the case for the memory hotplug, though.
echo online > /sys/devices/system/memory/memoryXYZ/state
will default to a kernel zone (usually ZONE_NORMAL) unless the
particular memblock is already in the movable zone range which is not
the case normally when onlining the memory from the udev rule context
for a freshly hotadded NUMA node. The only option currently is to have
a special udev rule to echo online_movable to all memblocks belonging to
such a node which is rather clumsy. Not to mention this is inconsistent
as well because what ended up in the movable zone during the boot will
end up in a kernel zone after hotremove & hotadd without special care.
It would be nice to reuse memblock_is_hotpluggable but the runtime
hotplug doesn't have that information available because the boot and
hotplug paths are not shared and it would be really non trivial to make
them use the same code path because the runtime hotplug doesn't play
with the memblock allocator at all.
Teach move_pfn_range that MMOP_ONLINE_KEEP can use the movable zone if
movable_node is enabled and the range doesn't overlap with the existing
normal zone. This should provide a reasonable default onlining
strategy.
Strictly speaking the semantic is not identical with the boot time
initialization because find_zone_movable_pfns_for_nodes covers only the
hotplugable range as described by the BIOS/FW. From my experience this
is usually a full node though (except for Node0 which is special and
never goes away completely). If this turns out to be a problem in the
real life we can tweak the code to store hotplug flag into memblocks but
let's keep this simple now.
Link: http://lkml.kernel.org/r/20170612111227.GI7476@dhcp22.suse.cz
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Reza Arbab <arbab@linux.vnet.ibm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Yasuaki Ishimatsu <yasu.isimatu@gmail.com>
Cc: <qiuxishi@huawei.com>
Cc: Kani Toshimitsu <toshi.kani@hpe.com>
Cc: <slaoub@gmail.com>
Cc: Joonsoo Kim <js1304@gmail.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Daniel Kiper <daniel.kiper@oracle.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When migrating a transparent hugepage, migrate_misplaced_transhuge_page
guards itself against a concurrent fastgup of the page by checking that
the page count is equal to 2 before and after installing the new pmd.
If the page count changes, then the pmd is reverted back to the original
entry, however there is a small window where the new (possibly writable)
pmd is installed and the underlying page could be written by userspace.
Restoring the old pmd could therefore result in loss of data.
This patch fixes the problem by freezing the page count whilst updating
the page tables, which protects against a concurrent fastgup without the
need to restore the old pmd in the failure case (since the page count
can no longer change under our feet).
Link: http://lkml.kernel.org/r/1497349722-6731-4-git-send-email-will.deacon@arm.com
Signed-off-by: Will Deacon <will.deacon@arm.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Steve Capper <steve.capper@arm.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
page_ref_freeze and page_ref_unfreeze are designed to be used as a pair,
wrapping a critical section where struct pages can be modified without
having to worry about consistency for a concurrent fast-GUP.
Whilst page_ref_freeze has full barrier semantics due to its use of
atomic_cmpxchg, page_ref_unfreeze is implemented using atomic_set, which
doesn't provide any barrier semantics and allows the operation to be
reordered with respect to page modifications in the critical section.
This patch ensures that page_ref_unfreeze is ordered after any critical
section updates, by invoking smp_mb() prior to the atomic_set.
Link: http://lkml.kernel.org/r/1497349722-6731-3-git-send-email-will.deacon@arm.com
Signed-off-by: Will Deacon <will.deacon@arm.com>
Acked-by: Steve Capper <steve.capper@arm.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The madvise policy for transparent huge pages is meant to avoid unwanted
allocations of transparent huge pages. It allows a policy of disabling
the extra memory pressure and effort to arrange for a huge page when it
is not needed.
DAX by definition never incurs this overhead since it is statically
allocated. The policy choice makes even less sense for device-dax which
tries to guarantee a given tlb-fault size. Specifically, the following
setting:
echo never > /sys/kernel/mm/transparent_hugepage/enabled
...violates that guarantee and silently disables all device-dax
instances with a 2M or 1G alignment. So, let's avoid that non-obvious
side effect by force enabling thp for dax mappings in all cases.
It is worth noting that the reason this uses vma_is_dax(), and the
resulting header include changes, is that previous attempts to add a
VM_DAX flag were NAKd.
Link: http://lkml.kernel.org/r/149739531127.20686.15813586620597484283.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Christoph Hellwig <hch@lst.de>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Turn the macro into a static inline and rewrite the condition checks for
better readability in preparation for adding another condition.
[ross.zwisler@linux.intel.com: fix logic to make conversion equivalent]
[akpm@linux-foundation.org: resolve vs mm-make-pr_set_thp_disable-immediately-active.patch]
[akpm@linux-foundation.org: include coredump.h for MMF_DISABLE_THP]
Link: http://lkml.kernel.org/r/149739530612.20686.14760671150202647861.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Acked-by: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
After enabling CONFIG_TRACE_ENUM_MAP_FILE (which will soon be renamed to
CONFIG_TRACE_EVAL_MAP_FILE), I am able to examine the enums that have
been evaluated:
# cat /sys/kernel/debug/tracing/enum_map
(which will soon be renamed to eval_map)
And it showed some interesting results:
[..]
ZONE_MOVABLE 3 (oom)
ZONE_NORMAL 2 (oom)
ZONE_DMA32 1 (oom)
ZONE_DMA 0 (oom)
3 3 (oom)
2 2 (oom)
1 1 (oom)
COMPACT_PRIO_ASYNC 2 (oom)
COMPACT_PRIO_SYNC_LIGHT 1 (oom)
COMPACT_PRIO_SYNC_FULL 0 (oom)
[..]
ZONE_DMA 0 (vmscan)
3 3 (vmscan)
2 2 (vmscan)
1 1 (vmscan)
COMPACT_PRIO_ASYNC 2 (vmscan)
[..]
ZONE_DMA 0 (kmem)
3 3 (kmem)
2 2 (kmem)
1 1 (kmem)
COMPACT_PRIO_ASYNC 2 (kmem)
[..]
ZONE_DMA 0 (compaction)
3 3 (compaction)
2 2 (compaction)
1 1 (compaction)
COMPACT_PRIO_ASYNC 2 (compaction)
[..]
The name within the parenthesis are the trace systems that the enum/eval
maps are associated with. When there's a number evaluated to another
number, that tells me that the TRACE_DEFINE_ENUM() was used on a #define
and not an enum. As #defines get converted normally, they are not needed
to be evaluated.
Each of the above trace systems with the number to number evaluation
included the file include/trace/events/mmflags.h which has:
/* High-level compaction status feedback */
#define COMPACTION_FAILED 1
#define COMPACTION_WITHDRAWN 2
#define COMPACTION_PROGRESS 3
[..]
#define COMPACTION_FEEDBACK \
EM(COMPACTION_FAILED, "failed") \
EM(COMPACTION_WITHDRAWN, "withdrawn") \
EMe(COMPACTION_PROGRESS, "progress")
Which is still needed for the __print_symbolic() usage in the
trace_event. But it is not needed to be evaluated.
Removing the evaluation part removes the unnecessary evaluations of
numbers to numbers.
Link: http://lkml.kernel.org/r/20170615074944.7be9a647@gandalf.local.home
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When the user specifies too many hugepages or an invalid
default_hugepagesz the communication to the user is implicit in the
allocation message. This patch adds a warning when the desired page
count is not allocated and prints an error when the default_hugepagesz
is invalid on boot.
During boot hugepages will allocate until there is a fraction of the
hugepage size left. That is, we allocate until either the request is
satisfied or memory for the pages is exhausted. When memory for the
pages is exhausted, it will most likely lead to the system failing with
the OOM manager not finding enough (or anything) to kill (unless you're
using really big hugepages in the order of 100s of MB or in the GBs).
The user will most likely see the OOM messages much later in the boot
sequence than the implicitly stated message. Worse yet, you may even
get an OOM for each processor which causes many pages of OOMs on modern
systems. Although these messages will be printed earlier than the OOM
messages, at least giving the user errors and warnings will highlight
the configuration as an issue. I'm trying to point the user in the
right direction by providing a more robust statement of what is failing.
During the sysctl or echo command, the user can check the results much
easier than if the system hangs during boot and the scenario of having
nothing to OOM for kernel memory is highly unlikely.
Mike said:
"Before sending out this patch, I asked Liam off list why he was doing
it. Was it something he just thought would be useful? Or, was there
some type of user situation/need. He said that he had been called in
to assist on several occasions when a system OOMed during boot. In
almost all of these situations, the user had grossly misconfigured
huge pages.
DB users want to pre-allocate just the right amount of huge pages, but
sometimes they can be really off. In such situations, the huge page
init code just allocates as many huge pages as it can and reports the
number allocated. There is no indication that it quit allocating
because it ran out of memory. Of course, a user could compare the
number in the message to what they requested on the command line to
determine if they got all the huge pages they requested. The thought
was that it would be useful to at least flag this situation. That way,
the user might be able to better relate the huge page allocation
failure to the OOM.
I'm not sure if the e-mail discussion made it obvious that this is
something he has seen on several occasions.
I see Michal's point that this will only flag the situation where
someone configures huge pages very badly. And, a more extensive look
at the situation of misconfiguring huge pages might be in order. But,
this has happened on several occasions which led to the creation of
this patch"
[akpm@linux-foundation.org: reposition memfmt() to avoid forward declaration]
Link: http://lkml.kernel.org/r/20170603005413.10380-1-Liam.Howlett@Oracle.com
Signed-off-by: Liam R. Howlett <Liam.Howlett@Oracle.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Cc: zhongjiang <zhongjiang@huawei.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
While activating a CMA area we check to make sure that all the PFNs in
the range are inside the same zone. This is a requirement for
alloc_contig_range() to work. Any CMA area failing the check is
disabled for good. This happens silently right now making all future
cma_alloc() allocations failure inevitable.
Here we add an error message stating that the CMA area could not be
activated which makes it easier to explain any future cma_alloc()
failures on it. While in there, change the bail out goto label from
'err' to 'not_in_zone' which makes more sense.
Link: http://lkml.kernel.org/r/20170605023729.26303-1-khandual@linux.vnet.ibm.com
Signed-off-by: Anshuman Khandual <khandual@linux.vnet.ibm.com>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When ioremap a 67112960 bytes vm_area with the vmallocinfo:
[..]
0xec79b000-0xec7fa000 389120 ftl_add_mtd+0x4d0/0x754 pages=94 vmalloc
0xec800000-0xecbe1000 4067328 kbox_proc_mem_write+0x104/0x1c4 phys=8b520000 ioremap
we get the result:
0xf1000000-0xf5001000 67112960 devm_ioremap+0x38/0x7c phys=40000000 ioremap
For the align for ioremap must be less than '1 << IOREMAP_MAX_ORDER':
if (flags & VM_IOREMAP)
align = 1ul << clamp_t(int, get_count_order_long(size),
PAGE_SHIFT, IOREMAP_MAX_ORDER);
So it makes idiot like me a litte puzzled why this was a jump the
vm_area from 0xec800000-0xecbe1000 to 0xf1000000-0xf5001000, and leaving
0xed000000-0xf1000000 as a big hole.
This patch is to show all of vm_area, including vmas which are freeing
but still in the vmap_area_list, to make it more clear about why we will
get 0xf1000000-0xf5001000 in the above case. And we will get a
vmallocinfo like:
[..]
0xec79b000-0xec7fa000 389120 ftl_add_mtd+0x4d0/0x754 pages=94 vmalloc
0xec800000-0xecbe1000 4067328 kbox_proc_mem_write+0x104/0x1c4 phys=8b520000 ioremap
[..]
0xece7c000-0xece7e000 8192 unpurged vm_area
0xece7e000-0xece83000 20480 vm_map_ram
0xf0099000-0xf00aa000 69632 vm_map_ram
after this patch.
Link: http://lkml.kernel.org/r/1496649682-20710-1-git-send-email-xieyisheng1@huawei.com
Signed-off-by: Yisheng Xie <xieyisheng1@huawei.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: zijun_hu <zijun_hu@htc.com>
Cc: "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Hanjun Guo <guohanjun@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Make @root exclusive in mem_cgroup_low; it is never considered low when
looked at directly and is not checked when traversing the tree. In
effect, @root is handled identically to how root_mem_cgroup was
previously handled by mem_cgroup_low.
If @root is not excluded from the checks, a cgroup underneath @root will
never be considered low during targeted reclaim of @root, e.g. due to
memory.current > memory.high, unless @root is misconfigured to have
memory.low > memory.high.
Excluding @root enables using memory.low to prioritize memory usage
between cgroups within a subtree of the hierarchy that is limited by
memory.high or memory.max, e.g. when ROOT owns @root's controls but
delegates the @root directory to a USER so that USER can create and
administer children of @root.
For example, given cgroup A with children B and C:
A
/ \
B C
and
1. A/memory.current > A/memory.high
2. A/B/memory.current < A/B/memory.low
3. A/C/memory.current >= A/C/memory.low
As 'A' is high, i.e. triggers reclaim from 'A', and 'B' is low, we
should reclaim from 'C' until 'A' is no longer high or until we can no
longer reclaim from 'C'. If 'A', i.e. @root, isn't excluded by
mem_cgroup_low when reclaming from 'A', then 'B' won't be considered low
and we will reclaim indiscriminately from both 'B' and 'C'.
Here is the test I used to confirm the bug and the patch.
20:00:55@sjchrist-vm ? ~ $ cat ~/.bin/memcg_low_test
#!/bin/bash
x62mb=$((62<<20))
x66mb=$((66<<20))
x94mb=$((94<<20))
x98mb=$((98<<20))
setup() {
set -e
if [[ -n $DEBUG ]]; then
set -x
fi
trap teardown EXIT HUP INT TERM
if [[ ! -e /mnt/1gb.swap ]]; then
sudo fallocate -l 1G /mnt/1gb.swap > /dev/null
sudo mkswap /mnt/1gb.swap > /dev/null
fi
if ! swapon --show=NAME | grep -q "/mnt/1gb.swap"; then
sudo swapon /mnt/1gb.swap
fi
if [[ ! -e /cgroup/cgroup.controllers ]]; then
sudo mount -t cgroup2 none /cgroup
fi
grep -q memory /cgroup/cgroup.controllers
sudo sh -c "echo '+memory' > /cgroup/cgroup.subtree_control"
sudo mkdir /cgroup/A && sudo chown $USER:$USER /cgroup/A
sudo sh -c "echo '+memory' > /cgroup/A/cgroup.subtree_control"
sudo sh -c "echo '96m' > /cgroup/A/memory.high"
mkdir /cgroup/A/0
mkdir /cgroup/A/1
echo 64m > /cgroup/A/0/memory.low
}
teardown() {
set +e
trap - EXIT HUP INT TERM
if [[ -z $1 ]]; then
printf "\n"
printf "%0.s*" {1..35}
printf "\nFAILED!\n\n"
tail /cgroup/A/**/memory.current
printf "%0.s*" {1..35}
printf "\n\n"
fi
ps | grep stress | tr -s ' ' | cut -f 2 -d ' ' | xargs -I % kill %
sleep 2
if [[ -e /cgroup/A/0 ]]; then
rmdir /cgroup/A/0
fi
if [[ -e /cgroup/A/1 ]]; then
rmdir /cgroup/A/1
fi
if [[ -e /cgroup/A ]]; then
sudo rmdir /cgroup/A
fi
}
stress_test() {
sudo sh -c "echo $$ > /cgroup/A/$1/cgroup.procs"
stress --vm 1 --vm-bytes 64M --vm-keep > /dev/null &
sudo sh -c "echo $$ > /cgroup/A/$2/cgroup.procs"
stress --vm 1 --vm-bytes 64M --vm-keep > /dev/null &
sudo sh -c "echo $$ > /cgroup/cgroup.procs"
sleep 1
# A/0 should be consuming more memory than A/1
[[ $(cat /cgroup/A/0/memory.current) -ge $(cat /cgroup/A/1/memory.current) ]]
# A/0 should be consuming ~64mb
[[ $(cat /cgroup/A/0/memory.current) -ge $x62mb ]] && [[ $(cat /cgroup/A/0/memory.current) -le $x66mb ]]
# A should cumulatively be consuming ~96mb
[[ $(cat /cgroup/A/memory.current) -ge $x94mb ]] && [[ $(cat /cgroup/A/memory.current) -le $x98mb ]]
# Stop the stressors
ps | grep stress | tr -s ' ' | cut -f 2 -d ' ' | xargs -I % kill %
}
teardown 1
setup
for ((i=1;i<=$1;i++)); do
printf "ITERATION $i of $1 - stress_test 0 1"
stress_test 0 1
printf "\x1b[2K\r"
printf "ITERATION $i of $1 - stress_test 1 0"
stress_test 1 0
printf "\x1b[2K\r"
printf "ITERATION $i of $1 - PASSED\n"
done
teardown 1
echo PASSED!
20:11:26@sjchrist-vm ? ~ $ memcg_low_test 10
Link: http://lkml.kernel.org/r/1496434412-21005-1-git-send-email-sean.j.christopherson@intel.com
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Acked-by: Balbir Singh <bsingharora@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
PR_SET_THP_DISABLE has a rather subtle semantic. It doesn't affect any
existing mapping because it only updated mm->def_flags which is a
template for new mappings.
The mappings created after prctl(PR_SET_THP_DISABLE) have VM_NOHUGEPAGE
flag set. This can be quite surprising for all those applications which
do not do prctl(); fork() & exec() and want to control their own THP
behavior.
Another usecase when the immediate semantic of the prctl might be useful
is a combination of pre- and post-copy migration of containers with
CRIU. In this case CRIU populates a part of a memory region with data
that was saved during the pre-copy stage. Afterwards, the region is
registered with userfaultfd and CRIU expects to get page faults for the
parts of the region that were not yet populated. However, khugepaged
collapses the pages and the expected page faults do not occur.
In more general case, the prctl(PR_SET_THP_DISABLE) could be used as a
temporary mechanism for enabling/disabling THP process wide.
Implementation wise, a new MMF_DISABLE_THP flag is added. This flag is
tested when decision whether to use huge pages is taken either during
page fault of at the time of THP collapse.
It should be noted, that the new implementation makes PR_SET_THP_DISABLE
master override to any per-VMA setting, which was not the case
previously.
Fixes: a0715cc226 ("mm, thp: add VM_INIT_DEF_MASK and PRCTL_THP_DISABLE")
Link: http://lkml.kernel.org/r/1496415802-30944-1-git-send-email-rppt@linux.vnet.ibm.com
Signed-off-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Pavel Emelyanov <xemul@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
By default, vmpressure events are not pass-through, i.e. they propagate
up through the memcg hierarchy until an event notifier is found for any
threshold level.
This presents a difficulty when a thread waiting on a read(2) for a
vmpressure event cannot distinguish between local memory pressure and
memory pressure in a descendant memcg, especially when that thread may
not control the memcg hierarchy.
Consider a user-controlled child memcg with a smaller limit than a
top-level memcg controlled by the "Activity Manager" specified in
Documentation/cgroup-v1/memory.txt. It may register for memory pressure
notification for descendant memcgs to make a policy decision: oom kill a
low priority job, increase the limit, decrease other limits, etc. If it
registers for memory pressure notification on the top-level memcg, it
currently cannot distinguish between memory pressure in its own memcg or
a descendant memcg, which is user-controlled.
Conversely, if a user registers for memory pressure notification on
their own descendant memcg, the Activity Manager does not receive any
pressure notification for that child memcg hierarchy. Vmpressure events
are not received for ancestor memcgs if the memcg experiencing pressure
have notifiers registered, perhaps outside the knowledge of the thread
waiting on read(2) at the top level.
Both of these are consequences of vmpressure notification not being
pass-through.
This implements a pass-through behavior for vmpressure events. When
writing to control.event_control, vmpressure event handlers may
optionally specify a mode. There are two new modes:
- "hierarchy": always propagate memory pressure events up the hierarchy
regardless if descendant memcgs have their own notifiers registered,
and
- "local": only receive notifications when the memcg for which the
event is registered experiences memory pressure.
Of course, processes may register for one notification of "low,local",
for example, and another for "low".
If no mode is specified, the current behavior is maintained for
backwards compatibility.
See the change to Documentation/cgroup-v1/memory.txt for full
specification.
[dan.carpenter@oracle.com: free the same pointer we allocated]
Link: http://lkml.kernel.org/r/20170613191820.GA20003@elgon.mountain
Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1705311421320.8946@chino.kir.corp.google.com
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Anton Vorontsov <anton@enomsg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
dequeue_hwpoisoned_huge_page() is no longer used, so let's remove it.
Link: http://lkml.kernel.org/r/1496305019-5493-9-git-send-email-n-horiguchi@ah.jp.nec.com
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently me_huge_page() relies on dequeue_hwpoisoned_huge_page() to
keep the error hugepage away from the system, which is OK but not good
enough because the hugepage still has a refcount and unpoison doesn't
work on the error hugepage (PageHWPoison flags are cleared but pages are
still leaked.) And there's "wasting health subpages" issue too. This
patch reworks on me_huge_page() to solve these issues.
For hugetlb file, recently we have truncating code so let's use it in
hugetlbfs specific ->error_remove_page().
For anonymous hugepage, it's helpful to dissolve the error page after
freeing it into free hugepage list. Migration entry and PageHWPoison in
the head page prevent the access to it.
TODO: dissolve_free_huge_page() can fail but we don't considered it yet.
It's not critical (and at least no worse that now) because in such case
the error hugepage just stays in free hugepage list without being
dissolved. By virtue of PageHWPoison in head page, it's never allocated
to processes.
[akpm@linux-foundation.org: fix unused var warnings]
Fixes: 23a003bfd2 ("mm/madvise: pass return code of memory_failure() to userspace")
Link: http://lkml.kernel.org/r/20170417055948.GM31394@yexl-desktop
Link: http://lkml.kernel.org/r/1496305019-5493-8-git-send-email-n-horiguchi@ah.jp.nec.com
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
memory_failure() is a big function and hard to maintain. Handling
hugetlb- and non-hugetlb- case in a single function is not good, so this
patch separates PageHuge() branch into a new function, which saves many
PageHuge() check.
Link: http://lkml.kernel.org/r/1496305019-5493-7-git-send-email-n-horiguchi@ah.jp.nec.com
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Now we have code to rescue most of healthy pages from a hwpoisoned
hugepage. So let's apply it to soft_offline_free_page too.
Link: http://lkml.kernel.org/r/1496305019-5493-6-git-send-email-n-horiguchi@ah.jp.nec.com
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently hugepage migrated by soft-offline (i.e. due to correctable
memory errors) is contained as a hugepage, which means many non-error
pages in it are unreusable, i.e. wasted.
This patch solves this issue by dissolving source hugepages into buddy.
As done in previous patch, PageHWPoison is set only on a head page of
the error hugepage. Then in dissoliving we move the PageHWPoison flag
to the raw error page so that all healthy subpages return back to buddy.
[arnd@arndb.de: fix warnings: replace some macros with inline functions]
Link: http://lkml.kernel.org/r/20170609102544.2947326-1-arnd@arndb.de
Link: http://lkml.kernel.org/r/1496305019-5493-5-git-send-email-n-horiguchi@ah.jp.nec.com
Signed-off-by: Anshuman Khandual <khandual@linux.vnet.ibm.com>
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We'd like to narrow down the error region in memory error on hugetlb
pages. However, currently we set PageHWPoison flags on all subpages in
the error hugepage and add # of subpages to num_hwpoison_pages, which
doesn't fit our purpose.
So this patch changes the behavior and we only set PageHWPoison on the
head page then increase num_hwpoison_pages only by 1. This is a
preparation for narrow-down part which comes in later patches.
Link: http://lkml.kernel.org/r/1496305019-5493-4-git-send-email-n-horiguchi@ah.jp.nec.com
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Cc: Anshuman Khandual <khandual@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We avoid calling __mod_node_page_state(NR_FILE_PAGES) for hugetlb page
now, but it's not enough because later code doesn't handle hugetlb
properly. Actually in our testing, WARN_ON_ONCE(PageDirty(page)) at the
end of this function fires for hugetlb, which makes no sense. So we
should return immediately for hugetlb pages.
Link: http://lkml.kernel.org/r/1496305019-5493-3-git-send-email-n-horiguchi@ah.jp.nec.com
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Cc: Anshuman Khandual <khandual@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "mm: hwpoison: fixlet for hugetlb migration".
This patchset updates the hwpoison/hugetlb code to address 2 reported
issues.
One is madvise(MADV_HWPOISON) failure reported by Intel's lkp robot (see
http://lkml.kernel.org/r/20170417055948.GM31394@yexl-desktop.) First
half was already fixed in mainline, and another half about hugetlb cases
are solved in this series.
Another issue is "narrow-down error affected region into a single 4kB
page instead of a whole hugetlb page" issue, which was tried by Anshuman
(http://lkml.kernel.org/r/20170420110627.12307-1-khandual@linux.vnet.ibm.com)
and I updated it to apply it more widely.
This patch (of 9):
We no longer use MIGRATE_ISOLATE to prevent reuse of hwpoison hugepages
as we did before. So current dequeue_huge_page_node() doesn't work as
intended because it still uses is_migrate_isolate_page() for this check.
This patch fixes it with PageHWPoison flag.
Link: http://lkml.kernel.org/r/1496305019-5493-2-git-send-email-n-horiguchi@ah.jp.nec.com
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Cc: Anshuman Khandual <khandual@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
To install a buffer_head into the cpu's LRU queue, bh_lru_install()
would construct a new copy of the queue and then memcpy it over the real
queue. But it's easily possible to do the update in-place, which is
faster and simpler. Some work can also be skipped if the buffer_head
was already in the queue.
As a microbenchmark I timed how long it takes to run sb_getblk()
10,000,000 times alternating between BH_LRU_SIZE + 1 blocks.
Effectively, this benchmarks looking up buffer_heads that are in the
page cache but not in the LRU:
Before this patch: 1.758s
After this patch: 1.653s
This patch also removes about 350 bytes of compiled code (on x86_64),
partly due to removal of the memcpy() which was being inlined+unrolled.
Link: http://lkml.kernel.org/r/20161229193445.1913-1-ebiggers3@gmail.com
Signed-off-by: Eric Biggers <ebiggers@google.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Sahitya Tummala
Oleg Nesterov
Michal Hocko
zhenwei.pi
Doug Berger
John Hubbard
Nikolay Borisov
Sebastian Andrzej Siewior
Rasmus Villemoes
Vasily Averin
Dou Liyang
Vinayak Menon
Roman Gushchin
Mike Rapoport
Jan Kara
Matthew Wilcox
David Rientjes
Michal Hocko
Andy Shevchenko
Will Deacon
Dan Williams
Steven Rostedt (VMware)
Liam R. Howlett
Anshuman Khandual
Yisheng Xie
Sean Christopherson
Naoya Horiguchi
Eric Biggers