Call set_pte_range() once per contiguous range of the folio instead of
once per page. This batches the updates to mm counters and the rmap.
With a will-it-scale.page_fault3 like app (change file write fault testing
to read fault testing. Trying to upstream it to will-it-scale at [1]) got
15% performance gain on a 48C/96T Cascade Lake test box with 96 processes
running against xfs.
Perf data collected before/after the change:
18.73%--page_add_file_rmap
|
--11.60%--__mod_lruvec_page_state
|
|--7.40%--__mod_memcg_lruvec_state
| |
| --5.58%--cgroup_rstat_updated
|
--2.53%--__mod_lruvec_state
|
--1.48%--__mod_node_page_state
9.93%--page_add_file_rmap_range
|
--2.67%--__mod_lruvec_page_state
|
|--1.95%--__mod_memcg_lruvec_state
| |
| --1.57%--cgroup_rstat_updated
|
--0.61%--__mod_lruvec_state
|
--0.54%--__mod_node_page_state
The running time of __mode_lruvec_page_state() is reduced about 9%.
[1]: https://github.com/antonblanchard/will-it-scale/pull/37
Link: https://lkml.kernel.org/r/20230802151406.3735276-38-willy@infradead.org
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Add PFN_PTE_SHIFT, update_mmu_cache_range(), flush_dcache_folio()
and flush_icache_pages().
Change the PG_dc_clean flag from being per-page to per-folio (which means
it cannot always be set as we don't know that all pages in this folio were
cleaned). Enhance the internal flush routines to take the number of pages
to flush.
Link: https://lkml.kernel.org/r/20230802151406.3735276-9-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Vineet Gupta <vgupta@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
This is the folio equivalent of page_mapping_file(), but rename it to make
it clear that it's very different from page_file_mapping().
Theoretically, there's nothing flush-only about it, but there are no other
users today, and I doubt there will be; it's almost always more useful to
know the swapfile's mapping or the swapcache's mapping.
Link: https://lkml.kernel.org/r/20230802151406.3735276-5-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "New page table range API", v6.
This patchset changes the API used by the MM to set up page table entries.
The four APIs are:
set_ptes(mm, addr, ptep, pte, nr)
update_mmu_cache_range(vma, addr, ptep, nr)
flush_dcache_folio(folio)
flush_icache_pages(vma, page, nr)
flush_dcache_folio() isn't technically new, but no architecture
implemented it, so I've done that for them. The old APIs remain around
but are mostly implemented by calling the new interfaces.
The new APIs are based around setting up N page table entries at once.
The N entries belong to the same PMD, the same folio and the same VMA, so
ptep++ is a legitimate operation, and locking is taken care of for you.
Some architectures can do a better job of it than just a loop, but I have
hesitated to make too deep a change to architectures I don't understand
well.
One thing I have changed in every architecture is that PG_arch_1 is now a
per-folio bit instead of a per-page bit when used for dcache clean/dirty
tracking. This was something that would have to happen eventually, and it
makes sense to do it now rather than iterate over every page involved in a
cache flush and figure out if it needs to happen.
The point of all this is better performance, and Fengwei Yin has measured
improvement on x86. I suspect you'll see improvement on your architecture
too. Try the new will-it-scale test mentioned here:
https://lore.kernel.org/linux-mm/20230206140639.538867-5-fengwei.yin@intel.com/
You'll need to run it on an XFS filesystem and have
CONFIG_TRANSPARENT_HUGEPAGE set.
This patchset is the basis for much of the anonymous large folio work
being done by Ryan, so it's received quite a lot of testing over the last
few months.
This patch (of 38):
Determine if a value lies within a range more efficiently (subtraction +
comparison vs two comparisons and an AND). It also has useful (under some
circumstances) behaviour if the range exceeds the maximum value of the
type. Convert all the conflicting definitions of in_range() within the
kernel; some can use the generic definition while others need their own
definition.
Link: https://lkml.kernel.org/r/20230802151406.3735276-1-willy@infradead.org
Link: https://lkml.kernel.org/r/20230802151406.3735276-2-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Currently, memcg uses rstat to maintain aggregated hierarchical stats.
Counters are maintained for hierarchical stats at each memcg. Rstat
tracks which cgroups have updates on which cpus to keep those counters
fresh on the read-side.
Non-hierarchical stats are currently not covered by rstat. Their per-cpu
counters are summed up on every read, which is expensive. The original
implementation did the same. At some point before rstat, non-hierarchical
aggregated counters were introduced by commit a983b5ebee ("mm:
memcontrol: fix excessive complexity in memory.stat reporting"). However,
those counters were updated on the performance critical write-side, which
caused regressions, so they were later removed by commit 815744d751
("mm: memcontrol: don't batch updates of local VM stats and events"). See
[1] for more detailed history.
Kernel versions in between a983b5ebee & 815744d751 (a year and a half)
enjoyed cheap reads of non-hierarchical stats, specifically on cgroup v1.
When moving to more recent kernels, a performance regression for reading
non-hierarchical stats is observed.
Now that we have rstat, we know exactly which percpu counters have updates
for each stat. We can maintain non-hierarchical counters again, making
reads much more efficient, without affecting the performance critical
write-side. Hence, add non-hierarchical (i.e local) counters for the
stats, and extend rstat flushing to keep those up-to-date.
A caveat is that we now need a stats flush before reading
local/non-hierarchical stats through {memcg/lruvec}_page_state_local() or
memcg_events_local(), where we previously only needed a flush to read
hierarchical stats. Most contexts reading non-hierarchical stats are
already doing a flush, add a flush to the only missing context in
count_shadow_nodes().
With this patch, reading memory.stat from 1000 memcgs is 3x faster on a
machine with 256 cpus on cgroup v1:
# for i in $(seq 1000); do mkdir /sys/fs/cgroup/memory/cg$i; done
# time cat /sys/fs/cgroup/memory/cg*/memory.stat > /dev/null
real 0m0.125s
user 0m0.005s
sys 0m0.120s
After:
real 0m0.032s
user 0m0.005s
sys 0m0.027s
To make sure there are no regressions on cgroup v2, I ran an artificial
reclaim/refault stress test [2] that creates (NR_CPUS * 2) cgroups,
assigns them limits, runs a worker process in each cgroup that allocates
tmpfs memory equal to quadruple the limit (to invoke reclaim
continuously), and then reads back the entire file (to invoke refaults).
All workers are run in parallel, and zram is used as a swapping backend.
Both reclaim and refault have conditional stats flushing. I ran this on a
machine with 112 cpus, once on mm-unstable, and once on mm-unstable with
this patch reverted.
(1) A few runs without this patch:
# time ./stress_reclaim_refault.sh
real 0m9.949s
user 0m0.496s
sys 14m44.974s
# time ./stress_reclaim_refault.sh
real 0m10.049s
user 0m0.486s
sys 14m55.791s
# time ./stress_reclaim_refault.sh
real 0m9.984s
user 0m0.481s
sys 14m53.841s
(2) A few runs with this patch:
# time ./stress_reclaim_refault.sh
real 0m9.885s
user 0m0.486s
sys 14m48.753s
# time ./stress_reclaim_refault.sh
real 0m9.903s
user 0m0.495s
sys 14m48.339s
# time ./stress_reclaim_refault.sh
real 0m9.861s
user 0m0.507s
sys 14m49.317s
No regressions are observed with this patch. There is actually a very
slight improvement. If I have to guess, maybe it's because we avoid
the percpu loop in count_shadow_nodes() when calling
lruvec_page_state_local(), but I could not prove this using perf, it's
probably in the noise.
[1] https://lore.kernel.org/lkml/20230725201811.GA1231514@cmpxchg.org/
[2] https://lore.kernel.org/lkml/CAJD7tkb17x=qwoO37uxyYXLEUVp15BQKR+Xfh7Sg9Hx-wTQ_=w@mail.gmail.com/
Link: https://lkml.kernel.org/r/20230803185046.1385770-1-yosryahmed@google.com
Link: https://lkml.kernel.org/r/20230726153223.821757-2-yosryahmed@google.com
Signed-off-by: Yosry Ahmed <yosryahmed@google.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Roman Gushchin <roman.gushchin@linux.dev>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Shakeel Butt <shakeelb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "Per-VMA lock support for swap and userfaults", v7.
When per-VMA locks were introduced in [1] several types of page faults
would still fall back to mmap_lock to keep the patchset simple. Among
them are swap and userfault pages. The main reason for skipping those
cases was the fact that mmap_lock could be dropped while handling these
faults and that required additional logic to be implemented. Implement
the mechanism to allow per-VMA locks to be dropped for these cases.
First, change handle_mm_fault to drop per-VMA locks when returning
VM_FAULT_RETRY or VM_FAULT_COMPLETED to be consistent with the way
mmap_lock is handled. Then change folio_lock_or_retry to accept vm_fault
and return vm_fault_t which simplifies later patches. Finally allow swap
and uffd page faults to be handled under per-VMA locks by dropping per-VMA
and retrying, the same way it's done under mmap_lock. Naturally, once VMA
lock is dropped that VMA should be assumed unstable and can't be used.
This patch (of 6):
Commit [1] introduced IO polling support duding swapin to reduce swap read
latency for block devices that can be polled. However later commit [2]
removed polling support. Therefore it seems safe to remove do_poll
parameter in read_swap_cache_async and always call swap_readpage with
synchronous=false waiting for IO completion in folio_lock_or_retry.
[1] commit 23955622ff ("swap: add block io poll in swapin path")
[2] commit 9650b453a3 ("block: ignore RWF_HIPRI hint for sync dio")
Link: https://lkml.kernel.org/r/20230630211957.1341547-1-surenb@google.com
Link: https://lkml.kernel.org/r/20230630211957.1341547-2-surenb@google.com
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Suggested-by: "Huang, Ying" <ying.huang@intel.com>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Hillf Danton <hdanton@sina.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Laurent Dufour <ldufour@linux.ibm.com>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michel Lespinasse <michel@lespinasse.org>
Cc: Minchan Kim <minchan@google.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Punit Agrawal <punit.agrawal@bytedance.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Jann Horn demonstrated how userfaultfd ioctl UFFDIO_COPY into a private
shmem mapping can add valid PTEs to page table collapse_pte_mapped_thp()
thought it had emptied: page lock on the huge page is enough to protect
against WP faults (which find the PTE has been cleared), but not enough to
protect against userfaultfd. "BUG: Bad rss-counter state" followed.
retract_page_tables() protects against this by checking !vma->anon_vma;
but we know that MADV_COLLAPSE needs to be able to work on private shmem
mappings, even those with an anon_vma prepared for another part of the
mapping; and we know that MADV_COLLAPSE needs to work on shared shmem
mappings which are userfaultfd_armed(). Whether it needs to work on
private shmem mappings which are userfaultfd_armed(), I'm not so sure: but
assume that it does.
Just for this case, take the pmd_lock() two steps earlier: not because it
gives any protection against this case itself, but because ptlock nests
inside it, and it's the dropping of ptlock which let the bug in. In other
cases, continue to minimize the pmd_lock() hold time.
Link: https://lkml.kernel.org/r/4d31abf5-56c0-9f3d-d12f-c9317936691@google.com
Fixes: 1043173eb5 ("mm/khugepaged: collapse_pte_mapped_thp() with mmap_read_lock()")
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Jann Horn <jannh@google.com>
Closes: https://lore.kernel.org/linux-mm/CAG48ez0FxiRC4d3VTu_a9h=rg5FW-kYD5Rg5xo_RDBM0LTTqZQ@mail.gmail.com/
Acked-by: Peter Xu <peterx@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
