When page del from buddy and need expand, it will account free_pages in
zone's migratetype.
The current way is to subtract the page number of the current order when
deleting, and then add it back when expanding.
This is unnecessary, as when migrating the same type, we can directly
record the difference between the high-order pages and the expand added,
and then subtract it directly.
This patch merge that, only when del and expand done, then account
free_pages.
Link: https://lkml.kernel.org/r/20240826064048.187790-1-link@vivo.com
Signed-off-by: Huan Yang <link@vivo.com>
Reviewed-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
This is an attempt to mitigate the issue of running out of memory when THP
is always enabled. During runtime whenever a THP is being faulted in
(__do_huge_pmd_anonymous_page) or collapsed by khugepaged
(collapse_huge_page), the THP is added to _deferred_list. Whenever memory
reclaim happens in linux, the kernel runs the deferred_split shrinker
which goes through the _deferred_list.
If the folio was partially mapped, the shrinker attempts to split it. If
the folio is not partially mapped, the shrinker checks if the THP was
underused, i.e. how many of the base 4K pages of the entire THP were
zero-filled. If this number goes above a certain threshold (decided by
/sys/kernel/mm/transparent_hugepage/khugepaged/max_ptes_none), the
shrinker will attempt to split that THP. Then at remap time, the pages
that were zero-filled are mapped to the shared zeropage, hence saving
memory.
Link: https://lkml.kernel.org/r/20240830100438.3623486-6-usamaarif642@gmail.com
Signed-off-by: Usama Arif <usamaarif642@gmail.com>
Suggested-by: Rik van Riel <riel@surriel.com>
Co-authored-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Alexander Zhu <alexlzhu@fb.com>
Cc: Barry Song <baohua@kernel.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Kairui Song <ryncsn@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Nico Pache <npache@redhat.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Shuang Zhai <zhais@google.com>
Cc: Yu Zhao <yuzhao@google.com>
Cc: Shuang Zhai <szhai2@cs.rochester.edu>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "mm: split underused THPs", v5.
The current upstream default policy for THP is always. However, Meta uses
madvise in production as the current THP=always policy vastly
overprovisions THPs in sparsely accessed memory areas, resulting in
excessive memory pressure and premature OOM killing. Using madvise +
relying on khugepaged has certain drawbacks over THP=always. Using
madvise hints mean THPs aren't "transparent" and require userspace
changes. Waiting for khugepaged to scan memory and collapse pages into
THP can be slow and unpredictable in terms of performance (i.e. you dont
know when the collapse will happen), while production environments require
predictable performance. If there is enough memory available, its better
for both performance and predictability to have a THP from fault time,
i.e. THP=always rather than wait for khugepaged to collapse it, and deal
with sparsely populated THPs when the system is running out of memory.
This patch series is an attempt to mitigate the issue of running out of
memory when THP is always enabled. During runtime whenever a THP is being
faulted in or collapsed by khugepaged, the THP is added to a list.
Whenever memory reclaim happens, the kernel runs the deferred_split
shrinker which goes through the list and checks if the THP was underused,
i.e. how many of the base 4K pages of the entire THP were zero-filled.
If this number goes above a certain threshold, the shrinker will attempt
to split that THP. Then at remap time, the pages that were zero-filled
are mapped to the shared zeropage, hence saving memory. This method
avoids the downside of wasting memory in areas where THP is sparsely
filled when THP is always enabled, while still providing the upside THPs
like reduced TLB misses without having to use madvise.
Meta production workloads that were CPU bound (>99% CPU utilzation) were
tested with THP shrinker. The results after 2 hours are as follows:
| THP=madvise | THP=always | THP=always
| | | + shrinker series
| | | + max_ptes_none=409
-----------------------------------------------------------------------------
Performance improvement | - | +1.8% | +1.7%
(over THP=madvise) | | |
-----------------------------------------------------------------------------
Memory usage | 54.6G | 58.8G (+7.7%) | 55.9G (+2.4%)
-----------------------------------------------------------------------------
max_ptes_none=409 means that any THP that has more than 409 out of 512
(80%) zero filled filled pages will be split.
To test out the patches, the below commands without the shrinker will
invoke OOM killer immediately and kill stress, but will not fail with the
shrinker:
echo 450 > /sys/kernel/mm/transparent_hugepage/khugepaged/max_ptes_none
mkdir /sys/fs/cgroup/test
echo $$ > /sys/fs/cgroup/test/cgroup.procs
echo 20M > /sys/fs/cgroup/test/memory.max
echo 0 > /sys/fs/cgroup/test/memory.swap.max
# allocate twice memory.max for each stress worker and touch 40/512 of
# each THP, i.e. vm-stride 50K.
# With the shrinker, max_ptes_none of 470 and below won't invoke OOM
# killer.
# Without the shrinker, OOM killer is invoked immediately irrespective
# of max_ptes_none value and kills stress.
stress --vm 1 --vm-bytes 40M --vm-stride 50K
This patch (of 5):
Here being unused means containing only zeros and inaccessible to
userspace. When splitting an isolated thp under reclaim or migration, the
unused subpages can be mapped to the shared zeropage, hence saving memory.
This is particularly helpful when the internal fragmentation of a thp is
high, i.e. it has many untouched subpages.
This is also a prerequisite for THP low utilization shrinker which will be
introduced in later patches, where underutilized THPs are split, and the
zero-filled pages are freed saving memory.
Link: https://lkml.kernel.org/r/20240830100438.3623486-1-usamaarif642@gmail.com
Link: https://lkml.kernel.org/r/20240830100438.3623486-3-usamaarif642@gmail.com
Signed-off-by: Yu Zhao <yuzhao@google.com>
Signed-off-by: Usama Arif <usamaarif642@gmail.com>
Tested-by: Shuang Zhai <zhais@google.com>
Cc: Alexander Zhu <alexlzhu@fb.com>
Cc: Barry Song <baohua@kernel.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Kairui Song <ryncsn@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Nico Pache <npache@redhat.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Shuang Zhai <szhai2@cs.rochester.edu>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "mm/vdpa: correct misuse of non-direct-reclaim __GFP_NOFAIL
and improve related doc and warn", v4.
__GFP_NOFAIL carries the semantics of never failing, so its callers do not
check the return value:
%__GFP_NOFAIL: The VM implementation _must_ retry infinitely: the caller
cannot handle allocation failures. The allocation could block
indefinitely but will never return with failure. Testing for
failure is pointless.
However, __GFP_NOFAIL can sometimes fail if it exceeds size limits or is
used with GFP_ATOMIC/GFP_NOWAIT in a non-sleepable context. This patchset
handles illegal using __GFP_NOFAIL together with GFP_ATOMIC lacking
__GFP_DIRECT_RECLAIM(without this, we can't do anything to reclaim memory
to satisfy the nofail requirement) and improve related document and
warnings.
The proper size limits for __GFP_NOFAIL will be handled separately after
more discussions.
This patch (of 3):
mm doesn't support non-blockable __GFP_NOFAIL allocation. Because
persisting in providing __GFP_NOFAIL services for non-block users who
cannot perform direct memory reclaim may only result in an endless busy
loop.
Therefore, in such cases, the current mm-core may directly return a NULL
pointer:
static inline struct page *
__alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
struct alloc_context *ac)
{
...
if (gfp_mask & __GFP_NOFAIL) {
/*
* All existing users of the __GFP_NOFAIL are blockable, so warn
* of any new users that actually require GFP_NOWAIT
*/
if (WARN_ON_ONCE_GFP(!can_direct_reclaim, gfp_mask))
goto fail;
...
}
...
fail:
warn_alloc(gfp_mask, ac->nodemask,
"page allocation failure: order:%u", order);
got_pg:
return page;
}
Unfortuantely, vpda does that nofail allocation under non-sleepable lock.
A possible way to fix that is to move the pages allocation out of the lock
into the caller, but having to allocate a huge number of pages and
auxiliary page array seems to be problematic as well per Tetsuon: " You
should implement proper error handling instead of using __GFP_NOFAIL if
count can become large."
So I chose another way, which does not release kernel bounce pages when
user tries to register userspace bounce pages. Then we can avoid
allocating in paths where failure is not expected.(e.g in the release).
We pay this for more memory usage as we don't release kernel bounce pages
but further optimizations could be done on top.
[v-songbaohua@oppo.com: Refine the changelog]
Link: https://lkml.kernel.org/r/20240830202823.21478-1-21cnbao@gmail.com
Link: https://lkml.kernel.org/r/20240830202823.21478-2-21cnbao@gmail.com
Fixes: 6c77ed2288 ("vduse: Support using userspace pages as bounce buffer")
Signed-off-by: Barry Song <v-songbaohua@oppo.com>
Reviewed-by: Xie Yongji <xieyongji@bytedance.com>
Tested-by: Xie Yongji <xieyongji@bytedance.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Hailong.Liu <hailong.liu@oppo.com>
Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Uladzislau Rezki (Sony) <urezki@gmail.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yafang Shao <laoar.shao@gmail.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: "Eugenio Pérez" <eperezma@redhat.com>
Cc: Kees Cook <kees@kernel.org>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Maxime Coquelin <maxime.coquelin@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The mutex array pointer shares a cacheline with the spinlock:
ffffffff84187480 B hugetlb_fault_mutex_table
ffffffff84187488 B hugetlb_lock
This is because the former is annotated with a macro forcing cacheline
alignment. I suspect it was meant to be the variant which on top of it
makes sure the object does not share the cacheline with anyone.
Since array pointer itself is de facto read-only such an annotation does
not make sense there anyway. Instead mark it __ro_after_init along with
the size var.
Do however move the spinlock out of the way.
[akpm@linux-foundation.org: move section directives to the end of the definitions, per convention]
[akpm@linux-foundation.org: DEFINE_SPINLOCK doesn't permit section modifiers at end-of-definition]
Link: https://lkml.kernel.org/r/20240828160704.1425767-1-mjguzik@gmail.com
Signed-off-by: Mateusz Guzik <mjguzik@gmail.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Muchun Song <muchun.song@linux.dev>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Although shmem_get_folio_gfp() is correctly putting inodes on the
shrinklist according to the folio size, shmem_unused_huge_shrink() was
still dealing with that shrinklist in terms of HPAGE_PMD_SIZE.
Generalize that; and to handle the mixture of sizes more sensibly,
shmem_alloc_and_add_folio() give it a number of pages to be freed
(approximate: no need to minimize that with an exact calculation) instead
of a number of inodes to split.
[akpm@linux-foundation.org: comment tweak, per David]
Link: https://lkml.kernel.org/r/d8c40850-6774-7a93-1e2c-8d941683b260@google.com
Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
There has been a long-standing and very minor off-by-one, where
shmem_get_folio_gfp() decides if a large folio extends beyond i_size far
enough to leave a page or more for freeing later under pressure.
This is not something needed for stable: but it will be proportionately
more significant as support for smaller large folios is added, and is best
fixed before duplicating the check in other places.
Link: https://lkml.kernel.org/r/d8e75079-af2d-8519-56df-6be1dccc247a@google.com
Fixes: 779750d20b ("shmem: split huge pages beyond i_size under memory pressure")
Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
This reverts commit 0742cadf5e4c ("mm/damon/lru_sort: adjust local
variable to dynamic allocation").
The commit was introduced to avoid unnecessary usage of stack memory for
per-scheme region priorities histogram buffer. The fix is nice, but the
point of the fix looks not very clear if the commit message is not read
together. That's mainly because the buffer is a private field, which
means it is hidden from the DAMON API users. That's not the fault of the
fix but the underlying data structure.
Now the per-scheme histogram buffer is gone, so the problem that the
commit was fixing is also removed. The use of kmemdup() has no more point
but just making the code bit difficult to understand. Revert the fix.
Link: https://lkml.kernel.org/r/20240826042323.87025-5-sj@kernel.org
Signed-off-by: SeongJae Park <sj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "replace per-quota region priorities histogram buffer with
per-context one".
Each DAMOS quota (struct damos_quota) maintains a histogram for total
regions size per its prioritization score. DAMOS calcultes minimum
prioritization score of regions that are ok to apply the DAMOS action to
while respecting the quota. The histogram is constructed only for the
calculation of the minimum score in damos_adjust_quota() for each quota
which called by kdamond_fn().
Hence, there is no real reason to have per-quota histogram. Only
per-kdamond histogram is needed, since parallel kdamonds could have races
otherwise. The current implementation is only wasting the memory, and can
easily cause unintended stack usage[1].
So, introducing a per-kdamond histogram and replacing the per-quota one
with it would be the right solution for the issue. However, supporting
multiple DAMON contexts per kdamond is still an ongoing work[2] without a
clear estimated time of arrival. Meanwhile, per-context histogram could
be an effective and straightforward solution having no blocker. Let's fix
the problem first in the way.
This patch (of 4):
Introduce per-context buffer for region priority scores-total size
histogram. Same to the per-quota one (->histogram of struct damos_quota),
the new buffer is hidden from DAMON API users by being defined as a
private field of DAMON context structure. It is dynamically allocated and
de-allocated at the beginning and ending of the execution of the kdamond
by kdamond_fn() itself.
[1] commit 0742cadf5e4c ("mm/damon/lru_sort: adjust local variable to dynamic allocation")
[2] https://lore.kernel.org/20240531122320.909060-1-yorha.op@gmail.com
Link: https://lkml.kernel.org/r/20240826042323.87025-1-sj@kernel.org
Link: https://lkml.kernel.org/r/20240826042323.87025-2-sj@kernel.org
Signed-off-by: SeongJae Park <sj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "mm: count the number of anonymous THPs per size", v4.
Knowing the number of transparent anon THPs in the system is crucial
for performance analysis. It helps in understanding the ratio and
distribution of THPs versus small folios throughout the system.
Additionally, partial unmapping by userspace can lead to significant waste
of THPs over time and increase memory reclamation pressure. We need this
information for comprehensive system tuning.
This patch (of 2):
Let's track for each anonymous THP size, how many of them are currently
allocated. We'll track the complete lifespan of an anon THP, starting
when it becomes an anon THP ("large anon folio") (->mapping gets set),
until it gets freed (->mapping gets cleared).
Introduce a new "nr_anon" counter per THP size and adjust the
corresponding counter in the following cases:
* We allocate a new THP and call folio_add_new_anon_rmap() to map
it the first time and turn it into an anon THP.
* We split an anon THP into multiple smaller ones.
* We migrate an anon THP, when we prepare the destination.
* We free an anon THP back to the buddy.
Note that AnonPages in /proc/meminfo currently tracks the total number of
*mapped* anonymous *pages*, and therefore has slightly different
semantics. In the future, we might also want to track "nr_anon_mapped"
for each THP size, which might be helpful when comparing it to the number
of allocated anon THPs (long-term pinning, stuck in swapcache, memory
leaks, ...).
Further note that for now, we only track anon THPs after they got their
->mapping set, for example via folio_add_new_anon_rmap(). If we would
allocate some in the swapcache, they will only show up in the statistics
for now after they have been mapped to user space the first time, where we
call folio_add_new_anon_rmap().
[akpm@linux-foundation.org: documentation fixups, per David]
Link: https://lkml.kernel.org/r/3e8add35-e26b-443b-8a04-1078f4bc78f6@redhat.com
Link: https://lkml.kernel.org/r/20240824010441.21308-1-21cnbao@gmail.com
Link: https://lkml.kernel.org/r/20240824010441.21308-2-21cnbao@gmail.com
Signed-off-by: Barry Song <v-songbaohua@oppo.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: Chuanhua Han <hanchuanhua@oppo.com>
Cc: Kairui Song <kasong@tencent.com>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Lance Yang <ioworker0@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Shuai Yuan <yuanshuai@oppo.com>
Cc: Usama Arif <usamaarif642@gmail.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Previously we had a situation where shmem mTHP controls and stats were not
exposed for some supported sizes and were exposed for some unsupported
sizes. So let's clean that up.
Anon mTHP can support all large orders [2, PMD_ORDER]. But shmem can
support all large orders [1, MAX_PAGECACHE_ORDER]. However, per-size
shmem controls and stats were previously being exposed for all the anon
mTHP orders, meaning order-1 was not present, and for arm64 64K base
pages, orders 12 and 13 were exposed but were not supported internally.
Tidy this all up by defining ctrl and stats attribute groups for anon and
file separately. Anon ctrl and stats groups are populated for all orders
in THP_ORDERS_ALL_ANON and file ctrl and stats groups are populated for
all orders in THP_ORDERS_ALL_FILE_DEFAULT.
Additionally, create "any" ctrl and stats attribute groups which are
populated for all orders in (THP_ORDERS_ALL_ANON |
THP_ORDERS_ALL_FILE_DEFAULT). swpout stats use this since they apply to
anon and shmem.
The side-effect of all this is that different hugepage-*kB directories
contain different sets of controls and stats, depending on which memory
types support that size. This approach is preferred over the alternative,
which is to populate dummy controls and stats for memory types that do not
support a given size.
[ryan.roberts@arm.com: file pages and shmem can also be split]
Link: https://lkml.kernel.org/r/f7ced14c-8bc5-405f-bee7-94f63980f525@arm.comLink: https://lkml.kernel.org/r/20240808111849.651867-3-ryan.roberts@arm.com
Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
Tested-by: Barry Song <baohua@kernel.org>
Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Gavin Shan <gshan@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Lance Yang <ioworker0@gmail.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
