commit caf6912f3f upstream.
We're not factoring in the start of the file for where to write and
read the swapfile, which leads to very unfortunate side effects of
writing where we should not be...
Fixes: dd6bd0d9c7 ("swap: use bdev_read_page() / bdev_write_page()")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Cc: Anthony Iliopoulos <ailiop@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
swap_type_of is used for two entirely different purposes:
(1) check what swap type a given device/offset corresponds to
(2) find the first available swap device that can be written to
Mixing both in a single function creates an unreadable mess. Create two
separate functions instead, and switch both to pass a dev_t instead of
a struct block_device to further simplify the code.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
In current implementation, newly created or swap-in anonymous page is
started on active list. Growing active list results in rebalancing
active/inactive list so old pages on active list are demoted to inactive
list. Hence, the page on active list isn't protected at all.
Following is an example of this situation.
Assume that 50 hot pages on active list. Numbers denote the number of
pages on active/inactive list (active | inactive).
1. 50 hot pages on active list
50(h) | 0
2. workload: 50 newly created (used-once) pages
50(uo) | 50(h)
3. workload: another 50 newly created (used-once) pages
50(uo) | 50(uo), swap-out 50(h)
This patch tries to fix this issue. Like as file LRU, newly created or
swap-in anonymous pages will be inserted to the inactive list. They are
promoted to active list if enough reference happens. This simple
modification changes the above example as following.
1. 50 hot pages on active list
50(h) | 0
2. workload: 50 newly created (used-once) pages
50(h) | 50(uo)
3. workload: another 50 newly created (used-once) pages
50(h) | 50(uo), swap-out 50(uo)
As you can see, hot pages on active list would be protected.
Note that, this implementation has a drawback that the page cannot be
promoted and will be swapped-out if re-access interval is greater than the
size of inactive list but less than the size of total(active+inactive).
To solve this potential issue, following patch will apply workingset
detection similar to the one that's already applied to file LRU.
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hugh Dickins <hughd@google.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Link: http://lkml.kernel.org/r/1595490560-15117-3-git-send-email-iamjoonsoo.kim@lge.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The VM tries to balance reclaim pressure between anon and file so as to
reduce the amount of IO incurred due to the memory shortage. It already
counts refaults and swapins, but in addition it should also count
writepage calls during reclaim.
For swap, this is obvious: it's IO that wouldn't have occurred if the
anonymous memory hadn't been under memory pressure. From a relative
balancing point of view this makes sense as well: even if anon is cold and
reclaimable, a cache that isn't thrashing may have equally cold pages that
don't require IO to reclaim.
For file writeback, it's trickier: some of the reclaim writepage IO would
have likely occurred anyway due to dirty expiration. But not all of it -
premature writeback reduces batching and generates additional writes.
Since the flushers are already woken up by the time the VM starts writing
cache pages one by one, let's assume that we'e likely causing writes that
wouldn't have happened without memory pressure. In addition, the per-page
cost of IO would have probably been much cheaper if written in larger
batches from the flusher thread rather than the single-page-writes from
kswapd.
For our purposes - getting the trend right to accelerate convergence on a
stable state that doesn't require paging at all - this is sufficiently
accurate. If we later wanted to optimize for sustained thrashing, we can
still refine the measurements.
Count all writepage calls from kswapd as IO cost toward the LRU that the
page belongs to.
Why do this dynamically? Don't we know in advance that anon pages require
IO to reclaim, and so could build in a static bias?
First, scanning is not the same as reclaiming. If all the anon pages are
referenced, we may not swap for a while just because we're scanning the
anon list. During this time, however, it's important that we age
anonymous memory and the page cache at the same rate so that their
hot-cold gradients are comparable. Everything else being equal, we still
want to reclaim the coldest memory overall.
Second, we keep copies in swap unless the page changes. If there is
swap-backed data that's mostly read (tmpfs file) and has been swapped out
before, we can reclaim it without incurring additional IO.
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Link: http://lkml.kernel.org/r/20200520232525.798933-14-hannes@cmpxchg.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Since the LRUs were split into anon and file lists, the VM has been
balancing between page cache and anonymous pages based on per-list ratios
of scanned vs. rotated pages. In most cases that tips page reclaim
towards the list that is easier to reclaim and has the fewest actively
used pages, but there are a few problems with it:
1. Refaults and LRU rotations are weighted the same way, even though
one costs IO and the other costs a bit of CPU.
2. The less we scan an LRU list based on already observed rotations,
the more we increase the sampling interval for new references, and
rotations become even more likely on that list. This can enter a
death spiral in which we stop looking at one list completely until
the other one is all but annihilated by page reclaim.
Since commit a528910e12 ("mm: thrash detection-based file cache sizing")
we have refault detection for the page cache. Along with swapin events,
they are good indicators of when the file or anon list, respectively, is
too small for its workingset and needs to grow.
For example, if the page cache is thrashing, the cache pages need more
time in memory, while there may be colder pages on the anonymous list.
Likewise, if swapped pages are faulting back in, it indicates that we
reclaim anonymous pages too aggressively and should back off.
Replace LRU rotations with refaults and swapins as the basis for relative
reclaim cost of the two LRUs. This will have the VM target list balances
that incur the least amount of IO on aggregate.
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Link: http://lkml.kernel.org/r/20200520232525.798933-12-hannes@cmpxchg.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently, scan pressure between the anon and file LRU lists is balanced
based on a mixture of reclaim efficiency and a somewhat vague notion of
"value" of having certain pages in memory over others. That concept of
value is problematic, because it has caused us to count any event that
remotely makes one LRU list more or less preferrable for reclaim, even
when these events are not directly comparable and impose very different
costs on the system. One example is referenced file pages that we still
deactivate and referenced anonymous pages that we actually rotate back to
the head of the list.
There is also conceptual overlap with the LRU algorithm itself. By
rotating recently used pages instead of reclaiming them, the algorithm
already biases the applied scan pressure based on page value. Thus, when
rebalancing scan pressure due to rotations, we should think of reclaim
cost, and leave assessing the page value to the LRU algorithm.
Lastly, considering both value-increasing as well as value-decreasing
events can sometimes cause the same type of event to be counted twice,
i.e. how rotating a page increases the LRU value, while reclaiming it
succesfully decreases the value. In itself this will balance out fine,
but it quietly skews the impact of events that are only recorded once.
The abstract metric of "value", the murky relationship with the LRU
algorithm, and accounting both negative and positive events make the
current pressure balancing model hard to reason about and modify.
This patch switches to a balancing model of accounting the concrete,
actually observed cost of reclaiming one LRU over another. For now, that
cost includes pages that are scanned but rotated back to the list head.
Subsequent patches will add consideration for IO caused by refaulting of
recently evicted pages.
Replace struct zone_reclaim_stat with two cost counters in the lruvec, and
make everything that affects cost go through a new lru_note_cost()
function.
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Link: http://lkml.kernel.org/r/20200520232525.798933-9-hannes@cmpxchg.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
They're the same function, and for the purpose of all callers they are
equivalent to lru_cache_add().
[akpm@linux-foundation.org: fix it for local_lock changes]
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Rik van Riel <riel@surriel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Link: http://lkml.kernel.org/r/20200520232525.798933-5-hannes@cmpxchg.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The cgroup swaprate throttling is about matching new anon allocations to
the rate of available IO when that is being throttled. It's the io
controller hooking into the VM, rather than a memory controller thing.
Rename mem_cgroup_throttle_swaprate() to cgroup_throttle_swaprate(), and
drop the @memcg argument which is only used to check whether the preceding
page charge has succeeded and the fault is proceeding.
We could decouple the call from mem_cgroup_try_charge() here as well, but
that would cause unnecessary churn: the following patches convert all
callsites to a new charge API and we'll decouple as we go along.
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Alex Shi <alex.shi@linux.alibaba.com>
Reviewed-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Roman Gushchin <guro@fb.com>
Cc: Balbir Singh <bsingharora@gmail.com>
Link: http://lkml.kernel.org/r/20200508183105.225460-5-hannes@cmpxchg.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Merge updates from Andrew Morton:
"A few little subsystems and a start of a lot of MM patches.
Subsystems affected by this patch series: squashfs, ocfs2, parisc,
vfs. With mm subsystems: slab-generic, slub, debug, pagecache, gup,
swap, memcg, pagemap, memory-failure, vmalloc, kasan"
* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (128 commits)
kasan: move kasan_report() into report.c
mm/mm_init.c: report kasan-tag information stored in page->flags
ubsan: entirely disable alignment checks under UBSAN_TRAP
kasan: fix clang compilation warning due to stack protector
x86/mm: remove vmalloc faulting
mm: remove vmalloc_sync_(un)mappings()
x86/mm/32: implement arch_sync_kernel_mappings()
x86/mm/64: implement arch_sync_kernel_mappings()
mm/ioremap: track which page-table levels were modified
mm/vmalloc: track which page-table levels were modified
mm: add functions to track page directory modifications
s390: use __vmalloc_node in stack_alloc
powerpc: use __vmalloc_node in alloc_vm_stack
arm64: use __vmalloc_node in arch_alloc_vmap_stack
mm: remove vmalloc_user_node_flags
mm: switch the test_vmalloc module to use __vmalloc_node
mm: remove __vmalloc_node_flags_caller
mm: remove both instances of __vmalloc_node_flags
mm: remove the prot argument to __vmalloc_node
mm: remove the pgprot argument to __vmalloc
...
In some swap scalability test, it is found that there are heavy lock
contention on swap cache even if we have split one swap cache radix tree
per swap device to one swap cache radix tree every 64 MB trunk in commit
4b3ef9daa4 ("mm/swap: split swap cache into 64MB trunks").
The reason is as follow. After the swap device becomes fragmented so
that there's no free swap cluster, the swap device will be scanned
linearly to find the free swap slots. swap_info_struct->cluster_next is
the next scanning base that is shared by all CPUs. So nearby free swap
slots will be allocated for different CPUs. The probability for
multiple CPUs to operate on the same 64 MB trunk is high. This causes
the lock contention on the swap cache.
To solve the issue, in this patch, for SSD swap device, a percpu version
next scanning base (cluster_next_cpu) is added. Every CPU will use its
own per-cpu next scanning base. And after finishing scanning a 64MB
trunk, the per-cpu scanning base will be changed to the beginning of
another randomly selected 64MB trunk. In this way, the probability for
multiple CPUs to operate on the same 64 MB trunk is reduced greatly.
Thus the lock contention is reduced too. For HDD, because sequential
access is more important for IO performance, the original shared next
scanning base is used.
To test the patch, we have run 16-process pmbench memory benchmark on a
2-socket server machine with 48 cores. One ram disk is configured as the
swap device per socket. The pmbench working-set size is much larger than
the available memory so that swapping is triggered. The memory read/write
ratio is 80/20 and the accessing pattern is random. In the original
implementation, the lock contention on the swap cache is heavy. The perf
profiling data of the lock contention code path is as following,
_raw_spin_lock_irq.add_to_swap_cache.add_to_swap.shrink_page_list: 7.91
_raw_spin_lock_irqsave.__remove_mapping.shrink_page_list: 7.11
_raw_spin_lock.swapcache_free_entries.free_swap_slot.__swap_entry_free: 2.51
_raw_spin_lock_irqsave.swap_cgroup_record.mem_cgroup_uncharge_swap: 1.66
_raw_spin_lock_irq.shrink_inactive_list.shrink_lruvec.shrink_node: 1.29
_raw_spin_lock.free_pcppages_bulk.drain_pages_zone.drain_pages: 1.03
_raw_spin_lock_irq.shrink_active_list.shrink_lruvec.shrink_node: 0.93
After applying this patch, it becomes,
_raw_spin_lock.swapcache_free_entries.free_swap_slot.__swap_entry_free: 3.58
_raw_spin_lock_irq.shrink_inactive_list.shrink_lruvec.shrink_node: 2.3
_raw_spin_lock_irqsave.swap_cgroup_record.mem_cgroup_uncharge_swap: 2.26
_raw_spin_lock_irq.shrink_active_list.shrink_lruvec.shrink_node: 1.8
_raw_spin_lock.free_pcppages_bulk.drain_pages_zone.drain_pages: 1.19
The lock contention on the swap cache is almost eliminated.
And the pmbench score increases 18.5%. The swapin throughput increases
18.7% from 2.96 GB/s to 3.51 GB/s. While the swapout throughput increases
18.5% from 2.99 GB/s to 3.54 GB/s.
We need really fast disk to show the benefit. I have tried this on 2
Intel P3600 NVMe disks. The performance improvement is only about 1%.
The improvement should be better on the faster disks, such as Intel Optane
disk.
[ying.huang@intel.com: fix cluster_next_cpu allocation and freeing, per Daniel]
Link: http://lkml.kernel.org/r/20200525002648.336325-1-ying.huang@intel.com
[ying.huang@intel.com: v4]
Link: http://lkml.kernel.org/r/20200529010840.928819-1-ying.huang@intel.com
Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Hugh Dickins <hughd@google.com>
Link: http://lkml.kernel.org/r/20200520031502.175659-1-ying.huang@intel.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
swap_info_struct->swap_map[] encodes some flag and count. And to
do some condition check, it also introduces some special values.
Currently those macros are defined with some magic order, which makes
audience hard to understand the exact meaning.
This patch split those macros into three categories:
flag
special value for first swap_map
special value for continued swap_map
May this help audiences a little.
[akpm@linux-foundation.org: tweak capitalization in comments]
Signed-off-by: Wei Yang <richard.weiyang@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20200501015259.32237-1-richard.weiyang@gmail.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The various struct pagevec per CPU variables are protected by disabling
either preemption or interrupts across the critical sections. Inside
these sections spinlocks have to be acquired.
These spinlocks are regular spinlock_t types which are converted to
"sleeping" spinlocks on PREEMPT_RT enabled kernels. Obviously sleeping
locks cannot be acquired in preemption or interrupt disabled sections.
local locks provide a trivial way to substitute preempt and interrupt
disable instances. On a non PREEMPT_RT enabled kernel local_lock() maps
to preempt_disable() and local_lock_irq() to local_irq_disable().
Create lru_rotate_pvecs containing the pagevec and the locallock.
Create lru_pvecs containing the remaining pagevecs and the locallock.
Add lru_add_drain_cpu_zone() which is used from compact_zone() to avoid
exporting the pvec structure.
Change the relevant call sites to acquire these locks instead of using
preempt_disable() / get_cpu() / get_cpu_var() and local_irq_disable() /
local_irq_save().
There is neither a functional change nor a change in the generated
binary code for non PREEMPT_RT enabled non-debug kernels.
When lockdep is enabled local locks have lockdep maps embedded. These
allow lockdep to validate the protections, i.e. inappropriate usage of a
preemption only protected sections would result in a lockdep warning
while the same problem would not be noticed with a plain
preempt_disable() based protection.
local locks also improve readability as they provide a named scope for
the protections while preempt/interrupt disable are opaque scopeless.
Finally local locks allow PREEMPT_RT to substitute them with real
locking primitives to ensure the correctness of operation in a fully
preemptible kernel.
[ bigeasy: Adopted to use local_lock ]
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20200527201119.1692513-4-bigeasy@linutronix.de
The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.
Also, notice that, dynamic memory allocations won't be affected by
this change:
"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]
This issue was found with the help of Coccinelle.
[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 7649773293 ("cxgb3/l2t: Fix undefined behaviour")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
When backporting commit 9c4e6b1a70 ("mm, mlock, vmscan: no more skipping
pagevecs") to our 4.9 kernel, our test bench noticed around 10% down with
a couple of vm-scalability's test cases (lru-file-readonce,
lru-file-readtwice and lru-file-mmap-read). I didn't see that much down
on my VM (32c-64g-2nodes). It might be caused by the test configuration,
which is 32c-256g with NUMA disabled and the tests were run in root memcg,
so the tests actually stress only one inactive and active lru. It sounds
not very usual in mordern production environment.
That commit did two major changes:
1. Call page_evictable()
2. Use smp_mb to force the PG_lru set visible
It looks they contribute the most overhead. The page_evictable() is a
function which does function prologue and epilogue, and that was used by
page reclaim path only. However, lru add is a very hot path, so it sounds
better to make it inline. However, it calls page_mapping() which is not
inlined either, but the disassemble shows it doesn't do push and pop
operations and it sounds not very straightforward to inline it.
Other than this, it sounds smp_mb() is not necessary for x86 since
SetPageLRU is atomic which enforces memory barrier already, replace it
with smp_mb__after_atomic() in the following patch.
With the two fixes applied, the tests can get back around 5% on that test
bench and get back normal on my VM. Since the test bench configuration is
not that usual and I also saw around 6% up on the latest upstream, so it
sounds good enough IMHO.
The below is test data (lru-file-readtwice throughput) against the v5.6-rc4:
mainline w/ inline fix
150MB 154MB
With this patch the throughput gets 2.67% up. The data with using
smp_mb__after_atomic() is showed in the following patch.
Shakeel Butt did the below test:
On a real machine with limiting the 'dd' on a single node and reading 100
GiB sparse file (less than a single node). Just ran a single instance to
not cause the lru lock contention. The cmdline used is "dd if=file-100GiB
of=/dev/null bs=4k". Ran the cmd 10 times with drop_caches in between and
measured the time it took.
Without patch: 56.64143 +- 0.672 sec
With patches: 56.10 +- 0.21 sec
[akpm@linux-foundation.org: move page_evictable() to internal.h]
Fixes: 9c4e6b1a70 ("mm, mlock, vmscan: no more skipping pagevecs")
Signed-off-by: Yang Shi <yang.shi@linux.alibaba.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Tested-by: Shakeel Butt <shakeelb@google.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Link: http://lkml.kernel.org/r/1584500541-46817-1-git-send-email-yang.shi@linux.alibaba.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
