With slot cache gone, clean up the allocation helpers even more.
folio_alloc_swap will be the only entry for allocation and adding the
folio to swap cache (except suspend), making it opposite of
folio_free_swap.
Link: https://lkml.kernel.org/r/20250313165935.63303-8-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Matthew Wilcow (Oracle) <willy@infradead.org>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Slot cache is no longer needed now, removing it and all related code.
- vm-scalability with: `usemem --init-time -O -y -x -R -31 1G`,
12G memory cgroup using simulated pmem as SWAP (32G pmem, 32 CPUs),
16 test runs for each case, measuring the total throughput:
Before (KB/s) (stdev) After (KB/s) (stdev)
Random (4K): 424907.60 (24410.78) 414745.92 (34554.78)
Random (64K): 163308.82 (11635.72) 167314.50 (18434.99)
Sequential (4K, !-R): 6150056.79 (103205.90) 6321469.06 (115878.16)
The performance changes are below noise level.
- Build linux kernel with make -j96, using 4K folio with 1.5G memory
cgroup limit and 64K folio with 2G memory cgroup limit, on top of tmpfs,
12 test runs, measuring the system time:
Before (s) (stdev) After (s) (stdev)
make -j96 (4K): 6445.69 (61.95) 6408.80 (69.46)
make -j96 (64K): 6841.71 (409.04) 6437.99 (435.55)
Similar to above, 64k mTHP case showed a slight improvement.
Link: https://lkml.kernel.org/r/20250313165935.63303-7-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Matthew Wilcow (Oracle) <willy@infradead.org>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Current allocation workflow first traverses the plist with a global lock
held, after choosing a device, it uses the percpu cluster on that swap
device. This commit moves the percpu cluster variable out of being tied
to individual swap devices, making it a global percpu variable, and will
be used directly for allocation as a fast path.
The global percpu cluster variable will never point to a HDD device, and
allocations on a HDD device are still globally serialized.
This improves the allocator performance and prepares for removal of the
slot cache in later commits. There shouldn't be much observable behavior
change, except one thing: this changes how swap device allocation rotation
works.
Currently, each allocation will rotate the plist, and because of the
existence of slot cache (one order 0 allocation usually returns 64
entries), swap devices of the same priority are rotated for every 64 order
0 entries consumed. High order allocations are different, they will
bypass the slot cache, and so swap device is rotated for every 16K, 32K,
or up to 2M allocation.
The rotation rule was never clearly defined or documented, it was changed
several times without mentioning.
After this commit, and once slot cache is gone in later commits, swap
device rotation will happen for every consumed cluster. Ideally non-HDD
devices will be rotated if 2M space has been consumed for each order.
Fragmented clusters will rotate the device faster, which seems OK. HDD
devices is rotated for every allocation regardless of the allocation
order, which should be OK too and trivial.
This commit also slightly changes allocation behaviour for slot cache.
The new added cluster allocation fast path may allocate entries from
different device to the slot cache, this is not observable from user
space, only impact performance very slightly, and slot cache will be just
gone in next commit, so this can be ignored.
Link: https://lkml.kernel.org/r/20250313165935.63303-6-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Matthew Wilcow (Oracle) <willy@infradead.org>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The counter update before allocation design was useful to avoid
unnecessary scan when device is full, so it will abort early if the
counter indicates the device is full. But that is an uncommon case, and
now scanning of a full device is very fast, so the up-front update is not
helpful any more.
Remove it and simplify the slot allocation logic.
Link: https://lkml.kernel.org/r/20250313165935.63303-5-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Matthew Wilcow (Oracle) <willy@infradead.org>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
This flag exists temporarily to allow the allocator to bypass the slot
cache during freeing, so reclaiming one slot will free the slot
immediately.
But now we have already removed slot cache usage on freeing, so this flag
has no effect now.
Link: https://lkml.kernel.org/r/20250313165935.63303-3-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Matthew Wilcow (Oracle) <willy@infradead.org>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "mm, swap: remove swap slot cache", v3.
Slot cache was initially introduced by commit 67afa38e012e ("mm/swap: add
cache for swap slots allocation") to reduce the lock contention of
si->lock.
Previous series "mm, swap: rework of swap allocator locks" [1] removed
swap slot cache for freeing path as freeing path no longer touches
si->lock in most cased. Allocation path also have slight to none
contention on si->lock since that series, but slot cache still helps to
reduce other overheads, like counters and the plist.
This series removes the slot cache from allocation path too, by using the
cluster as allocation fast path and also reduce other overheads.
Now slot cache is completely gone, the code is much simplified without
obvious feature or performance change, also clean up related workaround.
Also this should avoid other potential issues, e.g. the long pinning of
swap slots: swap slot cache pins swap slots with HAS_CACHE, causing
reclaim or allocation fail to use these slots on scanning.
The only behavior change is the swap device allocation rotation mechanism,
as explained in the patch "mm, swap: use percpu cluster as allocation fast
path".
Test results are looking good after deleting the swap slot cache:
- vm-scalability with: `usemem --init-time -O -y -x -R -31 1G`,
12G memory cgroup using simulated pmem as SWAP (32G pmem, 32 CPUs),
16 test runs for each case, measuring the total throughput:
Before (KB/s) (stdev) After (KB/s) (stdev)
Random (4K): 424907.60 (24410.78) 414745.92 (34554.78)
Random (64K): 163308.82 (11635.72) 167314.50 (18434.99)
Sequential (4K, !-R): 6150056.79 (103205.90) 6321469.06 (115878.16)
- Build linux kernel with make -j96, using 4K folio with 1.5G memory
cgroup limit and 64K folio with 2G memory cgroup limit, on top of tmpfs,
12 test runs, measuring the system time:
Before (s) (stdev) After (s) (stdev)
make -j96 (4K): 6445.69 (61.95) 6408.80 (69.46)
make -j96 (64K): 6841.71 (409.04) 6437.99 (435.55)
The performance is unchanged, slightly better in some cases.
[1] https://lore.kernel.org/linux-mm/20250113175732.48099-1-ryncsn@gmail.com/
This patch (of 7):
Swap allocator will do swap cache reclaim to recycle HAS_CACHE slots for
allocation. It initiates the reclaim from the offset to be reclaimed and
looks up the corresponding folio. The lookup process is lockless, so it's
possible the folio will be removed from the swap cache and given a
different swap entry before the reclaim locks the folio. If it happens,
the reclaim will end up reclaiming an irrelevant folio, and return wrong
return value.
This shouldn't cause any problem with correctness or stability, but it is
indeed confusing and unexpected, and will increase fragmentation, decrease
performance.
Fix this by checking whether the folio is still pointing to the offset the
allocator want to reclaim before reclaiming it.
Link: https://lkml.kernel.org/r/20250313165935.63303-1-ryncsn@gmail.com
Link: https://lkml.kernel.org/r/20250313165935.63303-2-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kairui Song <kasong@tencent.com>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Matthew Wilcow (Oracle) <willy@infradead.org>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
swap_reclaim_full_clusters() has no return value now, just remove the
stale comment which says swap_reclaim_full_clusters() wil return a bool
value.
Link: https://lkml.kernel.org/r/20250222160850.505274-7-shikemeng@huaweicloud.com
Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
Cc: Kairui Song <ryncsn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
We will add si back to plist in swap_usage_sub(), just correct the wrong
comment which says we will remove si from plist in swap_usage_sub().
Link: https://lkml.kernel.org/r/20250222160850.505274-6-shikemeng@huaweicloud.com
Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
Cc: Kairui Song <ryncsn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Before alloc from a cluster, we will aqcuire cluster's lock and make sure
it is usable by cluster_is_usable(), so there is no need to set
SWAP_MAP_BAD for cluster to be discarded.
Link: https://lkml.kernel.org/r/20250222160850.505274-5-shikemeng@huaweicloud.com
Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
Reviewed-by: Kairui Song <kasong@tencent.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
It's only called in scan_swap_map_slots().
And also remove the stale code comment in scan_swap_map_slots() because
it's not fit for the current cluster allocation mechanism.
Link: https://lkml.kernel.org/r/20250205092721.9395-13-bhe@redhat.com
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: Kairui Song <ryncsn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Since commit eb085574a752 ("mm, swap: fix race between swapoff and some
swap operations"), the non_swap_entry() checking has been taken off from
function __swap_duplicate(). Hence, in the kernel-doc comment, the line
'swp_entry is migration entry -> EINVAL' is obsolete. Remove that line to
avoid misleading people.
Link: https://lkml.kernel.org/r/20250205092721.9395-12-bhe@redhat.com
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: Kairui Song <ryncsn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The new function name can reflect the real behaviour of the function more
clearly and more accurately. And the renaming avoids the confusion
between swap_swapcount() and swp_swapcount().
Link: https://lkml.kernel.org/r/20250205092721.9395-11-bhe@redhat.com
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: Kairui Song <ryncsn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
In free_swap_and_cache_nr(), invocation of get_swap_device() has done the
checking if it's a swap entry. So remove the redundant checking here.
Link: https://lkml.kernel.org/r/20250205092721.9395-10-bhe@redhat.com
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: Kairui Song <ryncsn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
In the last 'for' loop inside setup_clusters(), using two local variable
'k' and 'j' are obvisouly redundant. Using 'j' is enough and simpler.
And also move macro SWAP_CLUSTER_COLS close to its only user
setup_clusters().
Link: https://lkml.kernel.org/r/20250205092721.9395-8-bhe@redhat.com
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: Kairui Song <ryncsn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Now, swap_count_continued() has two callers, __swap_duplicate() and
__swap_entry_free_locked(), the relevant code comment is stale. Update it
to reflect the current situation.
[bhe@redhat.com: v2]
Link: https://lkml.kernel.org/r/Z6V0/UvG1fvkQ4t/@fedora
Link: https://lkml.kernel.org/r/20250205092721.9395-7-bhe@redhat.com
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: Kairui Song <ryncsn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
There are two predicates in the name of swap_is_has_cache() which is
confusing. Renaming it to remove the confusion and can better reflect its
functionality.
Link: https://lkml.kernel.org/r/20250205092721.9395-6-bhe@redhat.com
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: Kairui Song <ryncsn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Since ci->lock has been taken when isolating cluster from
si->free_clusters or taking si->percpu_cluster->next[order], it's
unnecessary to scan and check the cluster range availability if i'ts empty
cluster, and this can accelerate the huge page swapping.
Link: https://lkml.kernel.org/r/20250205092721.9395-5-bhe@redhat.com
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: Kairui Song <ryncsn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
It doesn't make sense to have a zero value of shift. Remove it to avoid
confusion.
Link: https://lkml.kernel.org/r/20250205092721.9395-4-bhe@redhat.com
Signed-off-by: Baoquan He <bhe@redhat.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: Kairui Song <ryncsn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
If allocation is racy with swapoff, we may call free_cluster for cluster
already in free list and trigger BUG_ON() as following:
Allocation Swapoff
cluster_alloc_swap_entry
...
/* may get a free cluster with offset */
offset = xxx;
if (offset)
ci = lock_cluster(si, offset);
...
del_from_avail_list(p, true);
si->flags &= ~SWP_WRITEOK;
alloc_swap_scan_cluster(si, ci, ...)
...
/* failed to alloc entry from free entry */
if (!cluster_alloc_range(...))
break;
...
/* add back a free cluster */
relocate_cluster(si, ci);
if (!ci->count)
free_cluster(si, ci);
VM_BUG_ON(ci->flags == CLUSTER_FLAG_FREE);
To prevent the BUG_ON(), call free_cluster() for free cluster to move the
cluster to tail of list.
Check cluster is not free before calling free_cluster() in
relocate_cluster() to avoid BUG_ON().
Link: https://lkml.kernel.org/r/20250222160850.505274-4-shikemeng@huaweicloud.com
Fixes: 3b644773eefd ("mm, swap: reduce contention on device lock")
Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
Reviewed-by: Kairui Song <kasong@tencent.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Use correct step in loop to wait all clusters in wait_for_allocation().
If we miss some cluster in wait_for_allocation(), use after free may occur
as follows:
shmem_writepage swapoff
folio_alloc_swap
get_swap_pages
scan_swap_map_slots
cluster_alloc_swap_entry
alloc_swap_scan_cluster
cluster_alloc_range
/* SWP_WRITEOK is valid */
if (!(si->flags & SWP_WRITEOK))
...
del_from_avail_list(p, true);
...
/* miss the cluster in shmem_writepage */
wait_for_allocation()
...
try_to_unuse()
memset(si->swap_map + start, usage, nr_pages);
swap_range_alloc(si, nr_pages);
ci->count += nr_pages;
/* return a valid entry */
...
exit_swap_address_space(p->type);
...
...
add_to_swap_cache
/* dereference swap_address_space(entry) which is NULL */
xas_lock_irq(&xas);
Link: https://lkml.kernel.org/r/20250222160850.505274-3-shikemeng@huaweicloud.com
Fixes: 9a0ddeb79880 ("mm, swap: hold a reference during scan and cleanup flag usage")
Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
Reviewed-by: Kairui Song <kasong@tencent.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
If no swap cache is reclaimed, cluster taken off from full_clusters list
will not be put in any list and we can't reclaime HAS_CACHE slots
efficiently. Do relocate_cluster for such cluster to avoid inefficiency.
Link: https://lkml.kernel.org/r/20250224113910.522439-1-shikemeng@huaweicloud.com
Fixes: 3b644773eefd ("mm, swap: reduce contention on device lock")
Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
Reviewed-by: Kairui Song <kasong@tencent.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Add a NULL check on the return value of swp_swap_info in __swap_duplicate
to prevent crashes caused by NULL pointer dereference.
The reason why swp_swap_info() returns NULL is unclear; it may be due
to CPU cache issues or DDR bit flips. The probability of this issue is
very small - it has been observed to occur approximately 1 in 500,000
times per week. The stack info we encountered is as follows:
Unable to handle kernel NULL pointer dereference at virtual address
0000000000000058
[RB/E]rb_sreason_str_set: sreason_str set null_pointer
Mem abort info:
ESR = 0x0000000096000005
EC = 0x25: DABT (current EL), IL = 32 bits
SET = 0, FnV = 0
EA = 0, S1PTW = 0
FSC = 0x05: level 1 translation fault
Data abort info:
ISV = 0, ISS = 0x00000005, ISS2 = 0x00000000
CM = 0, WnR = 0, TnD = 0, TagAccess = 0
GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
user pgtable: 4k pages, 39-bit VAs, pgdp=00000008a80e5000
[0000000000000058] pgd=0000000000000000, p4d=0000000000000000,
pud=0000000000000000
Internal error: Oops: 0000000096000005 [#1] PREEMPT SMP
Skip md ftrace buffer dump for: 0x1609e0
...
pc : swap_duplicate+0x44/0x164
lr : copy_page_range+0x508/0x1e78
sp : ffffffc0f2a699e0
x29: ffffffc0f2a699e0 x28: ffffff8a5b28d388 x27: ffffff8b06603388
x26: ffffffdf7291fe70 x25: 0000000000000006 x24: 0000000000100073
x23: 00000000002d2d2f x22: 0000000000000008 x21: 0000000000000000
x20: 00000000002d2d2f x19: 18000000002d2d2f x18: ffffffdf726faec0
x17: 0000000000000000 x16: 0010000000000001 x15: 0040000000000001
x14: 0400000000000001 x13: ff7ffffffffffb7f x12: ffeffffffffffbff
x11: ffffff8a5c7e1898 x10: 0000000000000018 x9 : 0000000000000006
x8 : 1800000000000000 x7 : 0000000000000000 x6 : ffffff8057c01f10
x5 : 000000000000a318 x4 : 0000000000000000 x3 : 0000000000000000
x2 : 0000006daf200000 x1 : 0000000000000001 x0 : 18000000002d2d2f
Call trace:
swap_duplicate+0x44/0x164
copy_page_range+0x508/0x1e78
copy_process+0x1278/0x21cc
kernel_clone+0x90/0x438
__arm64_sys_clone+0x5c/0x8c
invoke_syscall+0x58/0x110
do_el0_svc+0x8c/0xe0
el0_svc+0x38/0x9c
el0t_64_sync_handler+0x44/0xec
el0t_64_sync+0x1a8/0x1ac
Code: 9139c35a 71006f3f 54000568 f8797b55 (f9402ea8)
---[ end trace 0000000000000000 ]---
Kernel panic - not syncing: Oops: Fatal exception
SMP: stopping secondary CPUs
The patch seems to only provide a workaround, but there are no more
effective software solutions to handle the bit flips problem. This path
will change the issue from a system crash to a process exception, thereby
reducing the impact on the entire machine.
akpm: this is probably a kernel bug, but this patch keeps the system
running and doesn't reduce that bug's debuggability.
Link: https://lkml.kernel.org/r/e223b0e6ba2f4924984b1917cc717bd5@honor.com
Signed-off-by: gao xu <gaoxu2@honor.com>
Reviewed-by: Barry Song <baohua@kernel.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Yosry Ahmed <yosry.ahmed@linux.dev>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
There is a code error that will cause the swap entry allocator to reclaim
and check the whole cluster with an unexpected tail offset instead of the
part that needs to be reclaimed. This may cause corruption of the swap
map, so fix it.
Link: https://lkml.kernel.org/r/20250130115131.37777-1-ryncsn@gmail.com
Fixes: 3b644773eefd ("mm, swap: reduce contention on device lock")
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Chris Li <chrisl@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The slot cache for freeing path is mostly for reducing the overhead of
si->lock. As we have basically eliminated the si->lock usage for freeing
path, it can be removed.
This helps simplify the code, and avoids swap entries from being hold in
cache upon freeing. The delayed freeing of entries have been causing
trouble for further optimizations for zswap [1] and in theory will also
cause more fragmentation, and extra overhead.
Test with build linux kernel showed both performance and fragmentation is
better without the cache:
tiem make -j96 / 768M memcg, 4K pages, 10G ZRAM, avg of 4 test run::
Before:
Sys time: 36047.78, Real time: 472.43
After: (-7.6% sys time, -7.3% real time)
Sys time: 33314.76, Real time: 437.67
time make -j96 / 1152M memcg, 64K mTHP, 10G ZRAM, avg of 4 test run:
Before:
Sys time: 46859.04, Real time: 562.63
hugepages-64kB/stats/swpout: 1783392
hugepages-64kB/stats/swpout_fallback: 240875
After: (-23.3% sys time, -21.3% real time)
Sys time: 35958.87, Real time: 442.69
hugepages-64kB/stats/swpout: 1866267
hugepages-64kB/stats/swpout_fallback: 158330
Sequential SWAP should be also slightly faster, tests didn't show a
measurable difference though, at least no regression:
Swapin 4G zero page on ZRAM (time in us):
Before (avg. 1923756)
1912391 1927023 1927957 1916527 1918263 1914284 1934753 1940813 1921791
After (avg. 1922290):
1919101 1925743 1916810 1917007 1923930 1935152 1917403 1923549 1921913
Link: https://lore.kernel.org/all/CAMgjq7ACohT_uerSz8E_994ZZCv709Zor+43hdmesW_59W1BWw@mail.gmail.com/[1]
Link: https://lkml.kernel.org/r/20250113175732.48099-14-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Suggested-by: Chris Li <chrisl@kernel.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Non-rotational devices (SSD / ZRAM) can tolerate fragmentation, so the
goal of the SWAP allocator is to avoid contention for clusters. It uses a
per-CPU cluster design, and each CPU will use a different cluster as much
as possible.
However, HDDs are very sensitive to fragmentation, contention is trivial
in comparison. Therefore, we use one global cluster instead. This
ensures that each order will be written to the same cluster as much as
possible, which helps make the I/O more continuous.
This ensures that the performance of the cluster allocator is as good as
that of the old allocator. Tests after this commit compared to those
before this series:
Tested using 'make -j32' with tinyconfig, a 1G memcg limit, and HDD swap:
make -j32 with tinyconfig, using 1G memcg limit and HDD swap:
Before this series:
114.44user 29.11system 39:42.90elapsed 6%CPU (0avgtext+0avgdata 157284maxresident)k
2901232inputs+0outputs (238877major+4227640minor)pagefaults
After this commit:
113.90user 23.81system 38:11.77elapsed 6%CPU (0avgtext+0avgdata 157260maxresident)k
2548728inputs+0outputs (235471major+4238110minor)pagefaults
[ryncsn@gmail.com: check kmalloc() return in setup_clusters]
Link: https://lkml.kernel.org/r/CAMgjq7Au+o04ckHyT=iU-wVx9az=t0B-ZiC5E0bDqNrAtNOP-g@mail.gmail.com
Link: https://lkml.kernel.org/r/20250113175732.48099-13-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Suggested-by: Chris Li <chrisl@kernel.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
It's a common operation to retrieve the cluster info from offset,
introduce a helper for this.
Link: https://lkml.kernel.org/r/20250113175732.48099-12-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Suggested-by: Chris Li <chrisl@kernel.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Instead of using a returning argument, we can simply store the next
cluster offset to the fixed percpu location, which reduce the stack usage
and simplify the function:
Object size:
./scripts/bloat-o-meter mm/swapfile.o mm/swapfile.o.new
add/remove: 0/0 grow/shrink: 0/2 up/down: 0/-271 (-271)
Function old new delta
get_swap_pages 2847 2733 -114
alloc_swap_scan_cluster 894 737 -157
Total: Before=30833, After=30562, chg -0.88%
Stack usage:
Before:
swapfile.c:1190:5:get_swap_pages 240 static
After:
swapfile.c:1185:5:get_swap_pages 216 static
Link: https://lkml.kernel.org/r/20250113175732.48099-11-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chis Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Currently, swap locking is mainly composed of two locks: the cluster lock
(ci->lock) and the device lock (si->lock).
The cluster lock is much more fine-grained, so it is best to use ci->lock
instead of si->lock as much as possible.
We have cleaned up other hard dependencies on si->lock. Following the new
cluster allocator design, most operations don't need to touch si->lock at
all. In practice, we only need to take si->lock when moving clusters
between lists.
To achieve this, this commit reworks the locking pattern of all si->lock
and ci->lock users, eliminates all usage of ci->lock inside si->lock, and
introduces a new design to avoid touching si->lock unless needed.
For minimal contention and easier understanding of the system, two ideas
are introduced with the corresponding helpers: isolation and relocation.
- Clusters will be `isolated` from the list when iterating the list
to search for an allocatable cluster.
This ensures other CPUs won't walk into the same cluster easily,
and it releases si->lock after acquiring ci->lock, providing the
only place that handles the inversion of two locks, and avoids
contention.
Iterating the cluster list almost always moves the cluster
(free -> nonfull, nonfull -> frag, frag -> frag tail), but it
doesn't know where the cluster should be moved to until scanning
is done. So keeping the cluster off-list is a good option with
low overhead.
The off-list time window of a cluster is also minimal. In the worst
case, one CPU will return the cluster after scanning the 512 entries
on it, which we used to busy wait with a spin lock.
This is done with the new helper `isolate_lock_cluster`.
- Clusters will be `relocated` after allocation or freeing, according
to their usage count and status.
Allocations no longer hold si->lock now, and may drop ci->lock for
reclaim, so the cluster could be moved to any location while no lock
is held. Besides, isolation clears all flags when it takes the
cluster off the list (the flags must be in sync with the list status,
so cluster users don't need to touch si->lock for checking its list
status). So the cluster has to be relocated to the right list
according to its usage after allocation or freeing.
Relocation is optional, if the cluster flags indicate it's already
on the right list, it will skip touching the list or si->lock.
This is done with `relocate_cluster` after allocation or with
`[partial_]free_cluster` after freeing.
This handled usage of all kinds of clusters in a clean way.
Scanning and allocation by iterating the cluster list is handled by
"isolate - <scan / allocate> - relocate".
Scanning and allocation of per-CPU clusters will only involve
"<scan / allocate> - relocate", as it knows which cluster to lock
and use.
Freeing will only involve "relocate".
Each CPU will keep using its per-CPU cluster until the 512 entries
are all consumed. Freeing also has to free 512 entries to trigger
cluster movement in the best case, so si->lock is rarely touched.
Testing with building the Linux kernel with defconfig showed huge
improvement:
tiem make -j96 / 768M memcg, 4K pages, 10G ZRAM, on Intel 8255C:
Before:
Sys time: 73578.30, Real time: 864.05
After: (-50.7% sys time, -44.8% real time)
Sys time: 36227.49, Real time: 476.66
time make -j96 / 1152M memcg, 64K mTHP, 10G ZRAM, on Intel 8255C:
(avg of 4 test run)
Before:
Sys time: 74044.85, Real time: 846.51
hugepages-64kB/stats/swpout: 1735216
hugepages-64kB/stats/swpout_fallback: 430333
After: (-40.4% sys time, -37.1% real time)
Sys time: 44160.56, Real time: 532.07
hugepages-64kB/stats/swpout: 1786288
hugepages-64kB/stats/swpout_fallback: 243384
time make -j32 / 512M memcg, 4K pages, 5G ZRAM, on AMD 7K62:
Before:
Sys time: 8098.21, Real time: 401.3
After: (-22.6% sys time, -12.8% real time )
Sys time: 6265.02, Real time: 349.83
The allocation success rate also slightly improved as we sanitized the
usage of clusters with new defined helpers, previously dropping
si->lock or ci->lock during scan will cause cluster order shuffle.
Link: https://lkml.kernel.org/r/20250113175732.48099-10-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Suggested-by: Chris Li <chrisl@kernel.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Currently, we are only using flags to indicate which list the cluster is
on. Using one bit for each list type might be a waste, as the list type
grows, we will consume too many bits. Additionally, the current mixed
usage of '&' and '==' is a bit confusing.
Make it clean by using an enum to define all possible cluster statuses.
Only an off-list cluster will have the NONE (0) flag. And use a wrapper
to annotate and sanitize all flag settings and list movements.
Link: https://lkml.kernel.org/r/20250113175732.48099-9-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Suggested-by: Chris Li <chrisl@kernel.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The flag SWP_SCANNING was used as an indicator of whether a device is
being scanned for allocation, and prevents swapoff. Combined with
SWP_WRITEOK, they work as a set of barriers for a clean swapoff:
1. Swapoff clears SWP_WRITEOK, allocation requests will see
~SWP_WRITEOK and abort as it's serialized by si->lock.
2. Swapoff unuses all allocated entries.
3. Swapoff waits for SWP_SCANNING flag to be cleared, so ongoing
allocations will stop, preventing UAF.
4. Now swapoff can free everything safely.
This will make the allocation path have a hard dependency on si->lock.
Allocation always have to acquire si->lock first for setting SWP_SCANNING
and checking SWP_WRITEOK.
This commit removes this flag, and just uses the existing per-CPU refcount
instead to prevent UAF in step 3, which serves well for such usage without
dependency on si->lock, and scales very well too. Just hold a reference
during the whole scan and allocation process. Swapoff will kill and wait
for the counter.
And for preventing any allocation from happening after step 1 so the unuse
in step 2 can ensure all slots are free, swapoff will acquire the ci->lock
of each cluster one by one to ensure all allocations see ~SWP_WRITEOK and
abort.
This way these dependences on si->lock are gone. And worth noting we
can't kill the refcount as the first step for swapoff as the unuse process
have to acquire the refcount.
Link: https://lkml.kernel.org/r/20250113175732.48099-8-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chis Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
When the swap device is full (inuse_pages == pages), it should be removed
from the allocation available plist. If any slot is freed, the swap
device should be added back to the plist. Additionally, during swapon or
swapoff, the swap device is forcefully added or removed.
Currently, the condition (inuse_pages == pages) is checked after every
counter update, then remove or add the device accordingly. This is
serialized by si->lock.
This commit decouples it from the protection of si->lock and reworked
plist removal and adding, making it possible to get rid of the hard
dependency on si->lock in allocation path in later commits.
To achieve this, simply using another lock is not an optimal approach, as
the overhead is observable for a hot counter, and may cause complex
locking issues. Thus, this commit manages to make it a lock-free atomic
operation, by embedding the plist state into the second highest bit of the
atomic counter.
Simply making the counter an atomic will not work, if the update and plist
status check are not performed atomically, we may miss an addition or
removal. With the embedded info we can update the counter and check the
plist status with single atomic operations, and avoid any extra overheads:
If the counter is full (inuse_pages == pages) and the off-list bit is
unset, we attempt to remove it from the plist. If the counter is not full
(inuse_pages != pages) and the off-list bit is set, we attempt to add it
to the plist. Removing, adding and bit update is serialized with a lock,
which is a cold path. Ordinary counter updates will be lock-free.
Link: https://lkml.kernel.org/r/20250113175732.48099-7-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chis Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Remove highest_bit and lowest_bit. After the HDD allocation path has been
removed, the only purpose of these two fields is to determine whether the
device is full or not, which can instead be determined by checking the
inuse_pages.
Link: https://lkml.kernel.org/r/20250113175732.48099-6-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chis Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cluster lock (ci->lock) was introduced to reduce contention for certain
operations. Using cluster lock for HDD is not helpful as HDD have a poor
performance, so locking isn't the bottleneck. But having different set of
locks for HDD / non-HDD prevents further rework of device lock (si->lock).
This commit just changed all lock_cluster_or_swap_info to lock_cluster,
which is a safe and straight conversion since cluster info is always
allocated now, also removed all cluster_info related checks.
Link: https://lkml.kernel.org/r/20250113175732.48099-5-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Suggested-by: Chris Li <chrisl@kernel.org>
Reviewed-by: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
We are currently using different swap allocation algorithm for HDD and
non-HDD. This leads to the existence of a different set of locks, and the
code path is heavily bloated, causing difficulties for further
optimization and maintenance.
This commit removes all HDD swap allocation and related dead code, and
uses the cluster allocation algorithm instead.
The performance may drop temporarily, but this should be negligible: The
main advantage of the legacy HDD allocation algorithm is that it tends to
use continuous slots, but swap device gets fragmented quickly anyway, and
the attempt to use continuous slots will fail easily.
This commit also enables mTHP swap on HDD, which is expected to be
beneficial, and following commits will adapt and optimize the cluster
allocator for HDD.
Link: https://lkml.kernel.org/r/20250113175732.48099-4-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Suggested-by: Chris Li <chrisl@kernel.org>
Suggested-by: "Huang, Ying" <ying.huang@linux.alibaba.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The name of the function is confusing, and the code is much easier to
follow after folding, also rename the confusing naming "p" to more
meaningful "si".
Link: https://lkml.kernel.org/r/20250113175732.48099-3-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chis Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "mm, swap: rework of swap allocator locks", v4.
This series greatly improved swap performance by reworking the locking
design and simplify a lot of code path. Test showed a up to 400%
vm-scalability improvement with pmem as SWAP, and up to 37% reduce of
kernel compile real time with ZRAM as SWAP (up to 60% improvement in
system time).
This is part of the new swap allocator discussed during the "Swap
Abstraction" discussion at LSF/MM 2024, and "mTHP and swap allocator"
discussion at LPC 2024.
This is a follow up of previous swap cluster allocator series:
https://lore.kernel.org/linux-mm/20240730-swap-allocator-v5-0-cb9c148b9297@kernel.org/
Also enables further optimizations which will come later.
Previous series introduced a fully cluster based allocator, this series
completely get rid of the old allocator and makes the new allocator avoid
touching the si->lock unless needed. This bring huge performance gain and
get rid of slot cache for freeing path.
Currently, swap locking is mainly composed of two locks, cluster lock
(ci->lock) and device lock (si->lock). The device lock is widely used to
protect many things, causing it to be the main bottleneck for SWAP.
Cluster lock is much more fine-grained, so it will be best to use ci->lock
instead of si->lock as much as possible.
`perf lock' indicates this issue clearly. Doing linux kernel build using
tmpfs and ZRAM with limited memory (make -j64 with 1G memcg and 4k pages),
result of "perf lock contention -ab sleep 3" shows:
contended total wait max wait avg wait type caller
34948 53.63 s 7.11 ms 1.53 ms spinlock free_swap_and_cache_nr+0x350
16569 40.05 s 6.45 ms 2.42 ms spinlock get_swap_pages+0x231
11191 28.41 s 7.03 ms 2.54 ms spinlock swapcache_free_entries+0x59
4147 22.78 s 122.66 ms 5.49 ms spinlock page_vma_mapped_walk+0x6f3
4595 7.17 s 6.79 ms 1.56 ms spinlock swapcache_free_entries+0x59
406027 2.74 s 2.59 ms 6.74 us spinlock list_lru_add+0x39
...snip...
The top 5 caller are all users of si->lock, total wait time sums to
several minutes in the 3 seconds time window.
Following the new allocator design, many operation doesn't need to touch
si->lock at all. We only need to take si->lock when doing operations
across multiple clusters (changing the cluster list). So ideally
allocator should always take ci->lock first, then take si->lock only if
needed. But due to historical reasons, ci->lock is used inside si->lock
critical section, causing lock inversion if we simply try to acquire
si->lock after acquiring ci->lock.
This series audited all si->lock usage, clean up legacy codes, eliminate
usage of si->lock as much as possible by introducing new designs based on
the new cluster allocator.
Old HDD allocation codes are removed, cluster allocator is adapted with
small changes for HDD usage, test is looking OK.
And this also removed slot cache for freeing path. The performance is
even better without it now, and this enables other clean up and
optimizations as discussed before:
https://lore.kernel.org/all/CAMgjq7ACohT_uerSz8E_994ZZCv709Zor+43hdmesW_59W1BWw@mail.gmail.com/
After this series, lock contention on si->lock is nearly unobservable
with `perf lock` with the same test above:
contended total wait max wait avg wait type caller
... snip ...
52 127.12 us 3.82 us 2.44 us spinlock move_cluster+0x2c
56 120.77 us 12.41 us 2.16 us spinlock move_cluster+0x2c
... snip ...
10 21.96 us 2.78 us 2.20 us spinlock isolate_lock_cluster+0x20
... snip ...
9 19.27 us 2.70 us 2.14 us spinlock move_cluster+0x2c
... snip ...
5 11.07 us 2.70 us 2.21 us spinlock isolate_lock_cluster+0x20
`move_cluster' and `isolate_lock_cluster' (two new introduced helper) are
basically the only users of si->lock now, performance gain is huge, and
LOC is reduced.
Tests Results:
vm-scalability
==============
Running `usemem --init-time -O -y -x -R -31 1G` from vm-scalability in a
12G memory cgroup using simulated pmem as SWAP backend (32G pmem, 32
CPUs).
Using 4K folio by default, 64k mTHP and sequential access (!-R) results
are also provided. 6 test runs for each case, Total Throughput:
Test Before (KB/s) (stdev) After (KB/s) (stdev) Delta
---------------------------------------------------------------------------
Random (4K): 69937.11 (16449.77) 369816.17 (24476.68) +428.78%
Random (64k): 123442.83 (13207.51) 216379.00 (25024.83) +75.28%
Sequential (4K): 6313909.83 (148856.12) 6419860.66 (183563.38) +1.7%
Sequential access will cause lower stress for the allocator so the gain is
limited, but with random access (which is much closer to real workloads)
the performance gain is huge.
Build kernel with defconfig on tmpfs with ZRAM
==============================================
Below results shows a test matrix using different memory cgroup limit and
job numbets, and scaled up progressive for a intuitive result. Done on a
48c96t system.
6 test run for each case, it can be seen clearly that as concurrent job
number goes higher the performance gain is higher, but even -j6 is showing
slight improvement.
make -j<NR> | System Time (seconds) | Total Time (seconds)
(NR / Mem / ZRAM) | (Before / After / Delta) | (Before / After / Delta)
With 4k pages only:
6 / 192M / 3G | 1533 / 1522 / -0.7% | 1420 / 1414 / -0.3%
12 / 256M / 4G | 2275 / 2226 / -2.2% | 758 / 742 / -2.1%
24 / 384M / 5G | 3596 / 3154 / -12.3% | 476 / 422 / -11.3%
48 / 768M / 7G | 8159 / 3605 / -55.8% | 330 / 221 / -33.0%
96 / 1.5G / 10G | 18541 / 6462 / -65.1% | 283 / 180 / -36.4%
With 64k mTHP:
24 / 512M / 5G | 3585 / 3469 / -3.2% | 293 / 290 / -0.1%
48 / 1G / 7G | 8173 / 3607 / -55.9% | 251 / 158 / -37.0%
96 / 2G / 10G | 16305 / 7791 / -52.2% | 226 / 144 / -36.3%
The fragmentation are reduced too:
With: make -j96 / 1152M memcg, 64K mTHP:
(avg of 4 test run)
Before:
hugepages-64kB/stats/swpout: 1696184
hugepages-64kB/stats/swpout_fallback: 414318
After: (-63.2% mTHP swapout failure)
hugepages-64kB/stats/swpout: 1866267
hugepages-64kB/stats/swpout_fallback: 158330
There is a up to 65.1% improvement in sys time for build kernel test,
and lower fragmentation rate.
Build kernel with tinyconfig on tmpfs with HDD as swap:
=======================================================
This test is similar to above, but HDD test is very noisy and slow, the
deviation is huge, so just use tinyconfig instead and take the median test
result of 3 test run, which looks OK:
Before this series:
114.44user 29.11system 39:42.90elapsed 6%CPU
2901232inputs+0outputs (238877major+4227640minor)pagefaults
After this commit:
113.90user 23.81system 38:11.77elapsed 6%CPU
2548728inputs+0outputs (235471major+4238110minor)pagefaults
Single thread SWAP:
===================
Sequential SWAP should also be slightly faster as we removed a lot of
unnecessary parts. Test using micro benchmark for swapout/in 4G
zero memory using ZRAM, 10 test runs:
Swapout Before (avg. 3359304):
3353796 3358551 3371305 3356043 3367524 3355303 3355924 3354513 3360776
Swapin Before (avg. 1928698):
1920283 1927183 1934105 1921373 1926562 1938261 1927726 1928636 1934155
Swapout After (avg. 3347511, -0.4%):
3337863 3347948 3355235 3339081 3333134 3353006 3354917 3346055 3360359
Swapin After (avg. 1922290, -0.3%):
1919101 1925743 1916810 1917007 1923930 1935152 1917403 1923549 1921913
The gain is limited at noise level but seems slightly better.
This patch (of 13):
Direct reclaim can skip the whole folio after reclaimed a set of folio
based slots. Also simplify the code for allocation, reduce indention.
Link: https://lkml.kernel.org/r/20250113175732.48099-1-ryncsn@gmail.com
Link: https://lkml.kernel.org/r/20250113175732.48099-2-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chis Li <chrisl@kernel.org> (Google)
Cc: "Huang, Ying" <ying.huang@linux.alibaba.com>
Cc: Hugh Dickens <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
There are two flags used to synchronize allocation and scanning with
swapoff: SWP_WRITEOK and SWP_SCANNING.
SWP_WRITEOK: Swapoff will first unset this flag, at this point any further
swap allocation or scanning on this device should just abort so no more
new entries will be referencing this device. Swapoff will then unuse all
existing swap entries.
SWP_SCANNING: This flag is set when device is being scanned. Swapoff will
wait for all scanner to stop before the final release of the swap device
structures to avoid UAF. Note this flag is the highest used bit of
si->flags so it could be added up arithmetically, if there are multiple
scanner.
commit 5f843a9a3a1e ("mm: swap: separate SSD allocation from
scan_swap_map_slots()") ignored SWP_SCANNING and SWP_WRITEOK flags while
separating cluster allocation path from the old allocation path. Add the
flags back to fix swapoff race. The race is hard to trigger as si->lock
prevents most parallel operations, but si->lock could be dropped for
reclaim or discard. This issue is found during code review.
This commit fixes this problem. For SWP_SCANNING, Just like before, set
the flag before scan and remove it afterwards.
For SWP_WRITEOK, there are several places where si->lock could be dropped,
it will be error-prone and make the code hard to follow if we try to cover
these places one by one. So just do one check before the real allocation,
which is also very similar like before. With new cluster allocator it may
waste a bit of time iterating the clusters but won't take long, and
swapoff is not performance sensitive.
Link: https://lkml.kernel.org/r/20241112083414.78174-1-ryncsn@gmail.com
Fixes: 5f843a9a3a1e ("mm: swap: separate SSD allocation from scan_swap_map_slots()")
Reported-by: "Huang, Ying" <ying.huang@intel.com>
Closes: https://lore.kernel.org/linux-mm/87a5es3f1f.fsf@yhuang6-desk2.ccr.corp.intel.com/
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
When running low on usable slots, cluster allocator will try to reclaim
the full clusters aggressively to reclaim HAS_CACHE slots. This
guarantees that as long as there are any usable slots, HAS_CACHE or not,
the swap device will be usable and workload won't go OOM early.
Before the cluster allocator, swap allocator fails easily if device is
filled up with reclaimable HAS_CACHE slots. Which can be easily
reproduced with following simple program:
#include <stdio.h>
#include <string.h>
#include <linux/mman.h>
#include <sys/mman.h>
#define SIZE 8192UL * 1024UL * 1024UL
int main(int argc, char **argv) {
long tmp;
char *p = mmap(NULL, SIZE, PROT_READ | PROT_WRITE,
MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
memset(p, 0, SIZE);
madvise(p, SIZE, MADV_PAGEOUT);
for (unsigned long i = 0; i < SIZE; ++i)
tmp += p[i];
getchar(); /* Pause */
return 0;
}
Setup an 8G non ramdisk swap, the first run of the program will swapout 8G
ram successfully. But run same program again after the first run paused,
the second run can't swapout all 8G memory as now half of the swap device
is pinned by HAS_CACHE. There was a random scan in the old allocator that
may reclaim part of the HAS_CACHE by luck, but it's unreliable.
The new allocator's added reclaim of full clusters when device is low on
usable slots. But when multiple CPUs are seeing the device is low on
usable slots at the same time, they ran into a thundering herd problem.
This is an observable problem on large machine with mass parallel
workload, as full cluster reclaim is slower on large swap device and
higher number of CPUs will also make things worse.
Testing using a 128G ZRAM on a 48c96t system. When the swap device is
very close to full (eg. 124G / 128G), running build linux kernel with
make -j96 in a 1G memory cgroup will hung (not a softlockup though)
spinning in full cluster reclaim for about ~5min before go OOM.
To solve this, split the full reclaim into two parts:
- Instead of do a synchronous aggressively reclaim when device is low,
do only one aggressively reclaim when device is strictly full with a
kworker. This still ensures in worst case the device won't be unusable
because of HAS_CACHE slots.
- To avoid allocation (especially higher order) suffer from HAS_CACHE
filling up clusters and kworker not responsive enough, do one synchronous
scan every time the free list is drained, and only scan one cluster. This
is kind of similar to the random reclaim before, keeps the full clusters
rotated and has a minimal latency. This should provide a fair reclaim
strategy suitable for most workloads.
Link: https://lkml.kernel.org/r/20241022175512.10398-1-ryncsn@gmail.com
Fixes: 2cacbdfdee65 ("mm: swap: add a adaptive full cluster cache reclaim")
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
I got a bad pud error and lost a 1GB HugeTLB when calling swapoff. The
problem can be reproduced by the following steps:
1. Allocate an anonymous 1GB HugeTLB and some other anonymous memory.
2. Swapout the above anonymous memory.
3. run swapoff and we will get a bad pud error in kernel message:
mm/pgtable-generic.c:42: bad pud 00000000743d215d(84000001400000e7)
We can tell that pud_clear_bad is called by pud_none_or_clear_bad in
unuse_pud_range() by ftrace. And therefore the HugeTLB pages will never
be freed because we lost it from page table. We can skip HugeTLB pages
for unuse_vma to fix it.
Link: https://lkml.kernel.org/r/20241015014521.570237-1-liushixin2@huawei.com
Fixes: 0fe6e20b9c4c ("hugetlb, rmap: add reverse mapping for hugepage")
Signed-off-by: Liu Shixin <liushixin2@huawei.com>
Acked-by: Muchun Song <muchun.song@linux.dev>
Cc: Naoya Horiguchi <nao.horiguchi@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
A report [1] was uploaded from syzbot.
In the previous commit 862590ac3708 ("mm: swap: allow cache reclaim to
skip slot cache"), the __try_to_reclaim_swap() function reads offset and
folio->entry from folio without folio_lock protection.
In the currently reported KCSAN log, it is assumed that the actual
data-race will not occur because the calltrace that does WRITE already
obtains the folio_lock and then writes.
However, the existing __try_to_reclaim_swap() function was already
implemented to perform reads under folio_lock protection [1], and there is
a risk of a data-race occurring through a function other than the one
shown in the KCSAN log.
Therefore, I think it is appropriate to change
read operations for folio to be performed under folio_lock.
[1]
==================================================================
BUG: KCSAN: data-race in __delete_from_swap_cache / __try_to_reclaim_swap
write to 0xffffea0004c90328 of 8 bytes by task 5186 on cpu 0:
__delete_from_swap_cache+0x1f0/0x290 mm/swap_state.c:163
delete_from_swap_cache+0x72/0xe0 mm/swap_state.c:243
folio_free_swap+0x1d8/0x1f0 mm/swapfile.c:1850
free_swap_cache mm/swap_state.c:293 [inline]
free_pages_and_swap_cache+0x1fc/0x410 mm/swap_state.c:325
__tlb_batch_free_encoded_pages mm/mmu_gather.c:136 [inline]
tlb_batch_pages_flush mm/mmu_gather.c:149 [inline]
tlb_flush_mmu_free mm/mmu_gather.c:366 [inline]
tlb_flush_mmu+0x2cf/0x440 mm/mmu_gather.c:373
zap_pte_range mm/memory.c:1700 [inline]
zap_pmd_range mm/memory.c:1739 [inline]
zap_pud_range mm/memory.c:1768 [inline]
zap_p4d_range mm/memory.c:1789 [inline]
unmap_page_range+0x1f3c/0x22d0 mm/memory.c:1810
unmap_single_vma+0x142/0x1d0 mm/memory.c:1856
unmap_vmas+0x18d/0x2b0 mm/memory.c:1900
exit_mmap+0x18a/0x690 mm/mmap.c:1864
__mmput+0x28/0x1b0 kernel/fork.c:1347
mmput+0x4c/0x60 kernel/fork.c:1369
exit_mm+0xe4/0x190 kernel/exit.c:571
do_exit+0x55e/0x17f0 kernel/exit.c:926
do_group_exit+0x102/0x150 kernel/exit.c:1088
get_signal+0xf2a/0x1070 kernel/signal.c:2917
arch_do_signal_or_restart+0x95/0x4b0 arch/x86/kernel/signal.c:337
exit_to_user_mode_loop kernel/entry/common.c:111 [inline]
exit_to_user_mode_prepare include/linux/entry-common.h:328 [inline]
__syscall_exit_to_user_mode_work kernel/entry/common.c:207 [inline]
syscall_exit_to_user_mode+0x59/0x130 kernel/entry/common.c:218
do_syscall_64+0xd6/0x1c0 arch/x86/entry/common.c:89
entry_SYSCALL_64_after_hwframe+0x77/0x7f
read to 0xffffea0004c90328 of 8 bytes by task 5189 on cpu 1:
__try_to_reclaim_swap+0x9d/0x510 mm/swapfile.c:198
free_swap_and_cache_nr+0x45d/0x8a0 mm/swapfile.c:1915
zap_pte_range mm/memory.c:1656 [inline]
zap_pmd_range mm/memory.c:1739 [inline]
zap_pud_range mm/memory.c:1768 [inline]
zap_p4d_range mm/memory.c:1789 [inline]
unmap_page_range+0xcf8/0x22d0 mm/memory.c:1810
unmap_single_vma+0x142/0x1d0 mm/memory.c:1856
unmap_vmas+0x18d/0x2b0 mm/memory.c:1900
exit_mmap+0x18a/0x690 mm/mmap.c:1864
__mmput+0x28/0x1b0 kernel/fork.c:1347
mmput+0x4c/0x60 kernel/fork.c:1369
exit_mm+0xe4/0x190 kernel/exit.c:571
do_exit+0x55e/0x17f0 kernel/exit.c:926
__do_sys_exit kernel/exit.c:1055 [inline]
__se_sys_exit kernel/exit.c:1053 [inline]
__x64_sys_exit+0x1f/0x20 kernel/exit.c:1053
x64_sys_call+0x2d46/0x2d60 arch/x86/include/generated/asm/syscalls_64.h:61
do_syscall_x64 arch/x86/entry/common.c:52 [inline]
do_syscall_64+0xc9/0x1c0 arch/x86/entry/common.c:83
entry_SYSCALL_64_after_hwframe+0x77/0x7f
value changed: 0x0000000000000242 -> 0x0000000000000000
Link: https://lkml.kernel.org/r/20241007070623.23340-1-aha310510@gmail.com
Reported-by: syzbot+fa43f1b63e3aa6f66329@syzkaller.appspotmail.com
Fixes: 862590ac3708 ("mm: swap: allow cache reclaim to skip slot cache")
Signed-off-by: Jeongjun Park <aha310510@gmail.com>
Acked-by: Chris Li <chrisl@kernel.org>
Reviewed-by: Kairui Song <kasong@tencent.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Retrieve a folio from the page cache rather than a page. Saves a couple
of conversions between page & folio.
Link: https://lkml.kernel.org/r/20240826202138.3804238-1-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "mm: store zero pages to be swapped out in a bitmap", v8.
As shown in the patch series that introduced the zswap same-filled
optimization [1], 10-20% of the pages stored in zswap are same-filled.
This is also observed across Meta's server fleet. By using VM counters in
swap_writepage (not included in this patchseries) it was found that less
than 1% of the same-filled pages to be swapped out are non-zero pages.
For conventional swap setup (without zswap), rather than reading/writing
these pages to flash resulting in increased I/O and flash wear, a bitmap
can be used to mark these pages as zero at write time, and the pages can
be filled at read time if the bit corresponding to the page is set.
When using zswap with swap, this also means that a zswap_entry does not
need to be allocated for zero filled pages resulting in memory savings
which would offset the memory used for the bitmap.
A similar attempt was made earlier in [2] where zswap would only track
zero-filled pages instead of same-filled. This patchseries adds
zero-filled pages optimization to swap (hence it can be used even if zswap
is disabled) and removes the same-filled code from zswap (as only 1% of
the same-filled pages are non-zero), simplifying code.
[1] https://lore.kernel.org/all/20171018104832epcms5p1b2232e2236258de3d03d1344dde9fce0@epcms5p1/
[2] https://lore.kernel.org/lkml/20240325235018.2028408-1-yosryahmed@google.com/
This patch (of 2):
Approximately 10-20% of pages to be swapped out are zero pages [1].
Rather than reading/writing these pages to flash resulting
in increased I/O and flash wear, a bitmap can be used to mark these
pages as zero at write time, and the pages can be filled at
read time if the bit corresponding to the page is set.
With this patch, NVMe writes in Meta server fleet decreased
by almost 10% with conventional swap setup (zswap disabled).
[1] https://lore.kernel.org/all/20171018104832epcms5p1b2232e2236258de3d03d1344dde9fce0@epcms5p1/
Link: https://lkml.kernel.org/r/20240823190545.979059-1-usamaarif642@gmail.com
Link: https://lkml.kernel.org/r/20240823190545.979059-2-usamaarif642@gmail.com
Signed-off-by: Usama Arif <usamaarif642@gmail.com>
Reviewed-by: Chengming Zhou <chengming.zhou@linux.dev>
Reviewed-by: Yosry Ahmed <yosryahmed@google.com>
Reviewed-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Usama Arif <usamaarif642@gmail.com>
Cc: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
We've been noticing a trend of significant lock contention in the swap
subsystem as core counts have been increasing in our fleet. It turns out
that our swapfiles on btrfs on flash were in fact using the old swap code
for rotational storage.
This turns out to be a detection issue in the swapon sequence: btrfs sets
si->bdev during swap activation, which currently happens *after* swapon's
SSD detection and cluster setup. Thus, none of the SSD optimizations and
cluster lock splitting are enabled for btrfs swap.
Rearrange the swapon sequence so that filesystem activation happens
*before* determining swap behavior based on the backing device.
Afterwards, the nonrotational drive is detected correctly:
- Adding 2097148k swap on /mnt/swapfile. Priority:-3 extents:1 across:2097148k
+ Adding 2097148k swap on /mnt/swapfile. Priority:-3 extents:1 across:2097148k SS
Link: https://lkml.kernel.org/r/20240822112707.351844-1-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "support large folio swap-out and swap-in for shmem", v5.
Shmem will support large folio allocation [1] [2] to get a better
performance, however, the memory reclaim still splits the precious large
folios when trying to swap-out shmem, which may lead to the memory
fragmentation issue and can not take advantage of the large folio for
shmeme.
Moreover, the swap code already supports for swapping out large folio
without split, and large folio swap-in[3] series is queued into
mm-unstable branch. Hence this patch set also supports the large folio
swap-out and swap-in for shmem.
This patch (of 9):
To support shmem large folio swap operations, add a new parameter to
swap_shmem_alloc() that allows batch SWAP_MAP_SHMEM flag setting for shmem
swap entries.
While we are at it, using folio_nr_pages() to get the number of pages of
the folio as a preparation.
Link: https://lkml.kernel.org/r/cover.1723434324.git.baolin.wang@linux.alibaba.com
Link: https://lkml.kernel.org/r/99f64115d04b285e009580eb177352c57119ffd0.1723434324.git.baolin.wang@linux.alibaba.com
Signed-off-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Reviewed-by: Barry Song <baohua@kernel.org>
Cc: Chris Li <chrisl@kernel.org>
Cc: Daniel Gomez <da.gomez@samsung.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Lance Yang <ioworker0@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Pankaj Raghav <p.raghav@samsung.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yang Shi <shy828301@gmail.com>
Cc: Zi Yan <ziy@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Zhiguo reported that swap release could be a serious bottleneck during
process exits[1]. With mTHP, we have the opportunity to batch free swaps.
Thanks to the work of Chris and Kairui[2], I was able to achieve this
optimization with minimal code changes by building on their efforts.
If swap_count is 1, which is likely true as most anon memory are private,
we can free all contiguous swap slots all together.
Ran the below test program for measuring the bandwidth of munmap
using zRAM and 64KiB mTHP:
#include <sys/mman.h>
#include <sys/time.h>
#include <stdlib.h>
unsigned long long tv_to_ms(struct timeval tv)
{
return tv.tv_sec * 1000 + tv.tv_usec / 1000;
}
main()
{
struct timeval tv_b, tv_e;
int i;
#define SIZE 1024*1024*1024
void *p = mmap(NULL, SIZE, PROT_READ | PROT_WRITE,
MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
if (!p) {
perror("fail to get memory");
exit(-1);
}
madvise(p, SIZE, MADV_HUGEPAGE);
memset(p, 0x11, SIZE); /* write to get mem */
madvise(p, SIZE, MADV_PAGEOUT);
gettimeofday(&tv_b, NULL);
munmap(p, SIZE);
gettimeofday(&tv_e, NULL);
printf("munmap in bandwidth: %ld bytes/ms\n",
SIZE/(tv_to_ms(tv_e) - tv_to_ms(tv_b)));
}
The result is as below (munmap bandwidth):
mm-unstable mm-unstable-with-patch
round1 21053761 63161283
round2 21053761 63161283
round3 21053761 63161283
round4 20648881 67108864
round5 20648881 67108864
munmap bandwidth becomes 3X faster.
[1] https://lore.kernel.org/linux-mm/20240731133318.527-1-justinjiang@vivo.com/
[2] https://lore.kernel.org/linux-mm/20240730-swap-allocator-v5-0-cb9c148b9297@kernel.org/
[v-songbaohua@oppo.com: check all swaps belong to same swap_cgroup in swap_pte_batch()]
Link: https://lkml.kernel.org/r/20240815215308.55233-1-21cnbao@gmail.com
[hughd@google.com: add mem_cgroup_disabled() check]
Link: https://lkml.kernel.org/r/33f34a88-0130-5444-9b84-93198eeb50e7@google.com
[21cnbao@gmail.com: add missing zswap_invalidate()]
Link: https://lkml.kernel.org/r/20240821054921.43468-1-21cnbao@gmail.com
Link: https://lkml.kernel.org/r/20240807215859.57491-3-21cnbao@gmail.com
Signed-off-by: Barry Song <v-songbaohua@oppo.com>
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Kairui Song <kasong@tencent.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Barry Song <baohua@kernel.org>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "mm: batch free swaps for zap_pte_range()", v3.
Batch free swap slots for zap_pte_range(), making munmap three times
faster when the page table entries are filled with swap entries to
be freed. This is likely another advantage of using mTHP.
This patch (of 3):
"p" means "pointer to something", rename it to a more meaningful
identifier - "si". We also have a case with the name "sis", rename it to
"si" as well.
Link: https://lkml.kernel.org/r/20240807215859.57491-1-21cnbao@gmail.com
Link: https://lkml.kernel.org/r/20240807215859.57491-2-21cnbao@gmail.com
Signed-off-by: Barry Song <v-songbaohua@oppo.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kairui Song <kasong@tencent.com>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Zhiguo Jiang <justinjiang@vivo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Link all full cluster with one full list, and reclaim from it when the
allocation have ran out of all usable clusters.
There are many reason a folio can end up being in the swap cache while
having no swap count reference. So the best way to search for such slots
is still by iterating the swap clusters.
With the list as an LRU, iterating from the oldest cluster and keep them
rotating is a very doable and clean way to free up potentially not inuse
clusters.
When any allocation failure, try reclaim and rotate only one cluster.
This is adaptive for high order allocations they can tolerate fallback.
So this avoids latency, and give the full cluster list an fair chance to
get reclaimed. It release the usage stress for the fallback order 0
allocation or following up high order allocation.
If the swap device is getting very full, reclaim more aggresively to
ensure no OOM will happen. This ensures order 0 heavy workload won't go
OOM as order 0 won't fail if any cluster still have any space.
[ryncsn@gmail.com: fix discard of full cluster]
Link: https://lkml.kernel.org/r/CAMgjq7CWwK75_2Zi5P40K08pk9iqOcuWKL6khu=x4Yg_nXaQag@mail.gmail.com
Link: https://lkml.kernel.org/r/20240730-swap-allocator-v5-9-cb9c148b9297@kernel.org
Signed-off-by: Kairui Song <kasong@tencent.com>
Reported-by: Barry Song <21cnbao@gmail.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Kairui Song <ryncsn@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
This commit implements reclaim during scan for cluster allocator.
Cluster scanning were unable to reuse SWAP_HAS_CACHE slots, which could
result in low allocation success rate or early OOM.
So to ensure maximum allocation success rate, integrate reclaiming with
scanning. If found a range of suitable swap slots but fragmented due to
HAS_CACHE, just try to reclaim the slots.
Link: https://lkml.kernel.org/r/20240730-swap-allocator-v5-8-cb9c148b9297@kernel.org
Signed-off-by: Kairui Song <kasong@tencent.com>
Reported-by: Barry Song <21cnbao@gmail.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Now swap cluster allocator arranges the clusters in LRU style, so the
"cold" cluster stay at the head of nonfull lists are the ones that were
used for allocation long time ago and still partially occupied. So if
allocator can't find enough contiguous slots to satisfy an high order
allocation, it's unlikely there will be slot being free on them to satisfy
the allocation, at least in a short period.
As a result, nonfull cluster scanning will waste time repeatly scanning
the unusable head of the list.
Also, multiple CPUs could content on the same head cluster of nonfull
list. Unlike free clusters which are removed from the list when any CPU
starts using it, nonfull cluster stays on the head.
So introduce a new list frag list, all scanned nonfull clusters will be
moved to this list. Both for avoiding repeated scanning and contention.
Frag list is still used as fallback for allocations, so if one CPU failed
to allocate one order of slots, it can still steal other CPU's clusters.
And order 0 will favor the fragmented clusters to better protect nonfull
clusters
If any slots on a fragment list are being freed, move the fragment list
back to nonfull list indicating it worth another scan on the cluster.
Compared to scan upon freeing a slot, this keep the scanning lazy and save
some CPU if there are still other clusters to use.
It may seems unneccessay to keep the fragmented cluster on list at all if
they can't be used for specific order allocation. But this will start to
make sense once reclaim dring scanning is ready.
Link: https://lkml.kernel.org/r/20240730-swap-allocator-v5-7-cb9c148b9297@kernel.org
Signed-off-by: Kairui Song <kasong@tencent.com>
Reported-by: Barry Song <21cnbao@gmail.com>
Cc: Chris Li <chrisl@kernel.org>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kalesh Singh <kaleshsingh@google.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
