2019-05-19 13:08:55 +01:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2011-10-16 02:01:52 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2016-01-22 15:10:40 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2017-02-08 18:51:30 +01:00
|
|
|
|
2006-06-23 02:04:16 -07:00
|
|
|
|
2006-01-11 12:17:46 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 17:04:11 +09:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2022-01-21 22:10:46 -08:00
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-05-13 17:21:04 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2007-10-18 14:47:32 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2006-03-24 03:16:04 -08:00
|
|
|
|
mm: memcontrol: rewrite charge API
These patches rework memcg charge lifetime to integrate more naturally
with the lifetime of user pages. This drastically simplifies the code and
reduces charging and uncharging overhead. The most expensive part of
charging and uncharging is the page_cgroup bit spinlock, which is removed
entirely after this series.
Here are the top-10 profile entries of a stress test that reads a 128G
sparse file on a freshly booted box, without even a dedicated cgroup (i.e.
executing in the root memcg). Before:
15.36% cat [kernel.kallsyms] [k] copy_user_generic_string
13.31% cat [kernel.kallsyms] [k] memset
11.48% cat [kernel.kallsyms] [k] do_mpage_readpage
4.23% cat [kernel.kallsyms] [k] get_page_from_freelist
2.38% cat [kernel.kallsyms] [k] put_page
2.32% cat [kernel.kallsyms] [k] __mem_cgroup_commit_charge
2.18% kswapd0 [kernel.kallsyms] [k] __mem_cgroup_uncharge_common
1.92% kswapd0 [kernel.kallsyms] [k] shrink_page_list
1.86% cat [kernel.kallsyms] [k] __radix_tree_lookup
1.62% cat [kernel.kallsyms] [k] __pagevec_lru_add_fn
After:
15.67% cat [kernel.kallsyms] [k] copy_user_generic_string
13.48% cat [kernel.kallsyms] [k] memset
11.42% cat [kernel.kallsyms] [k] do_mpage_readpage
3.98% cat [kernel.kallsyms] [k] get_page_from_freelist
2.46% cat [kernel.kallsyms] [k] put_page
2.13% kswapd0 [kernel.kallsyms] [k] shrink_page_list
1.88% cat [kernel.kallsyms] [k] __radix_tree_lookup
1.67% cat [kernel.kallsyms] [k] __pagevec_lru_add_fn
1.39% kswapd0 [kernel.kallsyms] [k] free_pcppages_bulk
1.30% cat [kernel.kallsyms] [k] kfree
As you can see, the memcg footprint has shrunk quite a bit.
text data bss dec hex filename
37970 9892 400 48262 bc86 mm/memcontrol.o.old
35239 9892 400 45531 b1db mm/memcontrol.o
This patch (of 4):
The memcg charge API charges pages before they are rmapped - i.e. have an
actual "type" - and so every callsite needs its own set of charge and
uncharge functions to know what type is being operated on. Worse,
uncharge has to happen from a context that is still type-specific, rather
than at the end of the page's lifetime with exclusive access, and so
requires a lot of synchronization.
Rewrite the charge API to provide a generic set of try_charge(),
commit_charge() and cancel_charge() transaction operations, much like
what's currently done for swap-in:
mem_cgroup_try_charge() attempts to reserve a charge, reclaiming
pages from the memcg if necessary.
mem_cgroup_commit_charge() commits the page to the charge once it
has a valid page->mapping and PageAnon() reliably tells the type.
mem_cgroup_cancel_charge() aborts the transaction.
This reduces the charge API and enables subsequent patches to
drastically simplify uncharging.
As pages need to be committed after rmap is established but before they
are added to the LRU, page_add_new_anon_rmap() must stop doing LRU
additions again. Revive lru_cache_add_active_or_unevictable().
[hughd@google.com: fix shmem_unuse]
[hughd@google.com: Add comments on the private use of -EAGAIN]
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vladimir Davydov <vdavydov@parallels.com>
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-08-08 14:19:20 -07:00
|
|
|
|
2008-02-07 00:13:53 -08:00
|
|
|
|
2017-11-15 17:37:41 -08:00
|
|
|
|
2014-04-07 15:37:19 -07:00
|
|
|
|
2018-10-26 15:06:08 -07:00
|
|
|
|
psi: pressure stall information for CPU, memory, and IO
When systems are overcommitted and resources become contended, it's hard
to tell exactly the impact this has on workload productivity, or how close
the system is to lockups and OOM kills. In particular, when machines work
multiple jobs concurrently, the impact of overcommit in terms of latency
and throughput on the individual job can be enormous.
In order to maximize hardware utilization without sacrificing individual
job health or risk complete machine lockups, this patch implements a way
to quantify resource pressure in the system.
A kernel built with CONFIG_PSI=y creates files in /proc/pressure/ that
expose the percentage of time the system is stalled on CPU, memory, or IO,
respectively. Stall states are aggregate versions of the per-task delay
accounting delays:
cpu: some tasks are runnable but not executing on a CPU
memory: tasks are reclaiming, or waiting for swapin or thrashing cache
io: tasks are waiting for io completions
These percentages of walltime can be thought of as pressure percentages,
and they give a general sense of system health and productivity loss
incurred by resource overcommit. They can also indicate when the system
is approaching lockup scenarios and OOMs.
To do this, psi keeps track of the task states associated with each CPU
and samples the time they spend in stall states. Every 2 seconds, the
samples are averaged across CPUs - weighted by the CPUs' non-idle time to
eliminate artifacts from unused CPUs - and translated into percentages of
walltime. A running average of those percentages is maintained over 10s,
1m, and 5m periods (similar to the loadaverage).
[hannes@cmpxchg.org: doc fixlet, per Randy]
Link: http://lkml.kernel.org/r/20180828205625.GA14030@cmpxchg.org
[hannes@cmpxchg.org: code optimization]
Link: http://lkml.kernel.org/r/20180907175015.GA8479@cmpxchg.org
[hannes@cmpxchg.org: rename psi_clock() to psi_update_work(), per Peter]
Link: http://lkml.kernel.org/r/20180907145404.GB11088@cmpxchg.org
[hannes@cmpxchg.org: fix build]
Link: http://lkml.kernel.org/r/20180913014222.GA2370@cmpxchg.org
Link: http://lkml.kernel.org/r/20180828172258.3185-9-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Daniel Drake <drake@endlessm.com>
Tested-by: Suren Baghdasaryan <surenb@google.com>
Cc: Christopher Lameter <cl@linux.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Johannes Weiner <jweiner@fb.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Enderborg <peter.enderborg@sony.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vinayak Menon <vinmenon@codeaurora.org>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-10-26 15:06:27 -07:00
|
|
|
|
2019-10-18 20:20:20 -07:00
|
|
|
|
2020-08-06 23:19:55 -07:00
|
|
|
|
2022-01-21 22:10:46 -08:00
|
|
|
|
2023-02-14 15:01:42 +00:00
|
|
|
|
|
|
|
|
|
2023-12-15 15:51:54 -05:00
|
|
|
|
mm: allow read-ahead with IOCB_NOWAIT set
Readahead support for IOCB_NOWAIT was introduced in commit 2e85abf053b9
("mm: allow read-ahead with IOCB_NOWAIT set"). However, this
implementation broke the semantics of IOCB_NOWAIT by potentially causing
it to wait on I/O during memory reclamation. This behavior was later
modified in commit efa8480a8316 ("fs: RWF_NOWAIT should imply IOCB_NOIO").
To resolve the blocking issue during memory reclamation, we can use
memalloc_noio_{save,restore} to ensure non-blocking behavior. This change
restores the original functionality, allowing preadv2(IOCB_NOWAIT) to
trigger readahead if the file content is not present in the page cache.
While this process may trigger direct memory reclamation, the
__GFP_NORETRY flag is set in the readahead GFP flags, ensuring it won't
block.
A use case for this change is when we want to trigger readahead in the
preadv2(2) syscall if the file cache is absent, but without waiting for
certain filesystem locks, like xfs_ilock. A simple example is as follows:
retry:
if (preadv2(fd, iovec, cnt, offset, RWF_NOWAIT) < 0) {
do_other_work();
goto retry;
}
Link: https://lore.gnuweeb.org/io-uring/20200624164127.GP21350@casper.infradead.org/
Link: https://lkml.kernel.org/r/20240820022639.89562-1-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Christian Brauner <brauner@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-08-20 10:26:39 +08:00
|
|
|
|
2025-01-11 15:07:37 +08:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2021-02-10 11:15:11 +00:00
|
|
|
|
2006-03-22 00:08:33 -08:00
|
|
|
|
|
|
|
|
|
2013-04-29 15:06:10 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2009-08-17 19:52:36 +02:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-12-12 16:54:24 -08:00
|
|
|
|
2022-02-09 20:22:12 +00:00
|
|
|
|
[PATCH] swap: swap_lock replace list+device
The idea of a swap_device_lock per device, and a swap_list_lock over them all,
is appealing; but in practice almost every holder of swap_device_lock must
already hold swap_list_lock, which defeats the purpose of the split.
The only exceptions have been swap_duplicate, valid_swaphandles and an
untrodden path in try_to_unuse (plus a few places added in this series).
valid_swaphandles doesn't show up high in profiles, but swap_duplicate does
demand attention. However, with the hold time in get_swap_pages so much
reduced, I've not yet found a load and set of swap device priorities to show
even swap_duplicate benefitting from the split. Certainly the split is mere
overhead in the common case of a single swap device.
So, replace swap_list_lock and swap_device_lock by spinlock_t swap_lock
(generally we seem to prefer an _ in the name, and not hide in a macro).
If someone can show a regression in swap_duplicate, then probably we should
add a hashlock for the swap_map entries alone (shorts being anatomic), so as
to help the case of the single swap device too.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 15:54:41 -07:00
|
|
|
|
2018-04-10 16:36:56 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-04-12 15:50:21 +02:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-06-08 21:33:54 -07:00
|
|
|
|
2014-12-12 16:54:24 -08:00
|
|
|
|
2005-10-29 18:16:41 -07:00
|
|
|
|
2018-04-10 16:36:56 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-06-08 21:33:54 -07:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-04-12 15:50:21 +02:00
|
|
|
|
2021-08-02 13:44:20 +02:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2011-04-21 18:19:44 -06:00
|
|
|
|
2011-03-22 22:23:41 +11:00
|
|
|
|
2018-04-10 16:36:56 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2014-12-12 16:54:24 -08:00
|
|
|
|
2023-01-20 11:26:49 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2005-10-29 18:16:41 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2005-10-29 18:16:41 -07:00
|
|
|
|
[PATCH] swap: swap_lock replace list+device
The idea of a swap_device_lock per device, and a swap_list_lock over them all,
is appealing; but in practice almost every holder of swap_device_lock must
already hold swap_list_lock, which defeats the purpose of the split.
The only exceptions have been swap_duplicate, valid_swaphandles and an
untrodden path in try_to_unuse (plus a few places added in this series).
valid_swaphandles doesn't show up high in profiles, but swap_duplicate does
demand attention. However, with the hold time in get_swap_pages so much
reduced, I've not yet found a load and set of swap device priorities to show
even swap_duplicate benefitting from the split. Certainly the split is mere
overhead in the common case of a single swap device.
So, replace swap_list_lock and swap_device_lock by spinlock_t swap_lock
(generally we seem to prefer an _ in the name, and not hide in a macro).
If someone can show a regression in swap_duplicate, then probably we should
add a hashlock for the swap_map entries alone (shorts being anatomic), so as
to help the case of the single swap device too.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-09-03 15:54:41 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2018-04-10 16:36:56 -07:00
|
|
|
|
2024-08-02 17:55:23 +02:00
|
|
|
|
2024-08-26 14:58:13 +08:00
|
|
|
|
2023-12-20 23:44:56 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2011-04-21 18:19:44 -06:00
|
|
|
|
2011-03-22 22:23:36 +11:00
|
|
|
|
2022-02-09 20:22:12 +00:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2017-11-21 09:17:59 -05:00
|
|
|
|
2021-05-08 00:35:49 -04:00
|
|
|
|
2014-04-03 14:47:49 -07:00
|
|
|
|
2021-05-08 00:35:49 -04:00
|
|
|
|
|
|
|
|
|
2016-12-12 16:43:17 -08:00
|
|
|
|
2017-11-21 09:17:59 -05:00
|
|
|
|
2016-12-12 16:43:17 -08:00
|
|
|
|
2023-09-26 12:20:17 -07:00
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:49 -07:00
|
|
|
|
2021-05-08 00:35:49 -04:00
|
|
|
|
mm: keep page cache radix tree nodes in check
Previously, page cache radix tree nodes were freed after reclaim emptied
out their page pointers. But now reclaim stores shadow entries in their
place, which are only reclaimed when the inodes themselves are
reclaimed. This is problematic for bigger files that are still in use
after they have a significant amount of their cache reclaimed, without
any of those pages actually refaulting. The shadow entries will just
sit there and waste memory. In the worst case, the shadow entries will
accumulate until the machine runs out of memory.
To get this under control, the VM will track radix tree nodes
exclusively containing shadow entries on a per-NUMA node list. Per-NUMA
rather than global because we expect the radix tree nodes themselves to
be allocated node-locally and we want to reduce cross-node references of
otherwise independent cache workloads. A simple shrinker will then
reclaim these nodes on memory pressure.
A few things need to be stored in the radix tree node to implement the
shadow node LRU and allow tree deletions coming from the list:
1. There is no index available that would describe the reverse path
from the node up to the tree root, which is needed to perform a
deletion. To solve this, encode in each node its offset inside the
parent. This can be stored in the unused upper bits of the same
member that stores the node's height at no extra space cost.
2. The number of shadow entries needs to be counted in addition to the
regular entries, to quickly detect when the node is ready to go to
the shadow node LRU list. The current entry count is an unsigned
int but the maximum number of entries is 64, so a shadow counter
can easily be stored in the unused upper bits.
3. Tree modification needs tree lock and tree root, which are located
in the address space, so store an address_space backpointer in the
node. The parent pointer of the node is in a union with the 2-word
rcu_head, so the backpointer comes at no extra cost as well.
4. The node needs to be linked to an LRU list, which requires a list
head inside the node. This does increase the size of the node, but
it does not change the number of objects that fit into a slab page.
[akpm@linux-foundation.org: export the right function]
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Rik van Riel <riel@redhat.com>
Reviewed-by: Minchan Kim <minchan@kernel.org>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Bob Liu <bob.liu@oracle.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jan Kara <jack@suse.cz>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Luigi Semenzato <semenzato@google.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Metin Doslu <metin@citusdata.com>
Cc: Michel Lespinasse <walken@google.com>
Cc: Ozgun Erdogan <ozgun@citusdata.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Roman Gushchin <klamm@yandex-team.ru>
Cc: Ryan Mallon <rmallon@gmail.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-04-03 14:47:56 -07:00
|
|
|
|
2017-11-21 09:17:59 -05:00
|
|
|
|
|
|
|
|
|
mm: filemap: don't plant shadow entries without radix tree node
When the underflow checks were added to workingset_node_shadow_dec(),
they triggered immediately:
kernel BUG at ./include/linux/swap.h:276!
invalid opcode: 0000 [#1] SMP
Modules linked in: isofs usb_storage fuse xt_CHECKSUM ipt_MASQUERADE nf_nat_masquerade_ipv4 tun nf_conntrack_netbios_ns nf_conntrack_broadcast ip6t_REJECT nf_reject_ipv6
soundcore wmi acpi_als pinctrl_sunrisepoint kfifo_buf tpm_tis industrialio acpi_pad pinctrl_intel tpm_tis_core tpm nfsd auth_rpcgss nfs_acl lockd grace sunrpc dm_crypt
CPU: 0 PID: 20929 Comm: blkid Not tainted 4.8.0-rc8-00087-gbe67d60ba944 #1
Hardware name: System manufacturer System Product Name/Z170-K, BIOS 1803 05/06/2016
task: ffff8faa93ecd940 task.stack: ffff8faa7f478000
RIP: page_cache_tree_insert+0xf1/0x100
Call Trace:
__add_to_page_cache_locked+0x12e/0x270
add_to_page_cache_lru+0x4e/0xe0
mpage_readpages+0x112/0x1d0
blkdev_readpages+0x1d/0x20
__do_page_cache_readahead+0x1ad/0x290
force_page_cache_readahead+0xaa/0x100
page_cache_sync_readahead+0x3f/0x50
generic_file_read_iter+0x5af/0x740
blkdev_read_iter+0x35/0x40
__vfs_read+0xe1/0x130
vfs_read+0x96/0x130
SyS_read+0x55/0xc0
entry_SYSCALL_64_fastpath+0x13/0x8f
Code: 03 00 48 8b 5d d8 65 48 33 1c 25 28 00 00 00 44 89 e8 75 19 48 83 c4 18 5b 41 5c 41 5d 41 5e 5d c3 0f 0b 41 bd ef ff ff ff eb d7 <0f> 0b e8 88 68 ef ff 0f 1f 84 00
RIP page_cache_tree_insert+0xf1/0x100
This is a long-standing bug in the way shadow entries are accounted in
the radix tree nodes. The shrinker needs to know when radix tree nodes
contain only shadow entries, no pages, so node->count is split in half
to count shadows in the upper bits and pages in the lower bits.
Unfortunately, the radix tree implementation doesn't know of this and
assumes all entries are in node->count. When there is a shadow entry
directly in root->rnode and the tree is later extended, the radix tree
implementation will copy that entry into the new node and and bump its
node->count, i.e. increases the page count bits. Once the shadow gets
removed and we subtract from the upper counter, node->count underflows
and triggers the warning. Afterwards, without node->count reaching 0
again, the radix tree node is leaked.
Limit shadow entries to when we have actual radix tree nodes and can
count them properly. That means we lose the ability to detect refaults
from files that had only the first page faulted in at eviction time.
Fixes: 449dd6984d0e ("mm: keep page cache radix tree nodes in check")
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reported-and-tested-by: Linus Torvalds <torvalds@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-04 22:02:08 +02:00
|
|
|
|
2021-05-08 00:35:49 -04:00
|
|
|
|
2025-05-14 19:15:07 +01:00
|
|
|
|
mm: filemap: don't plant shadow entries without radix tree node
When the underflow checks were added to workingset_node_shadow_dec(),
they triggered immediately:
kernel BUG at ./include/linux/swap.h:276!
invalid opcode: 0000 [#1] SMP
Modules linked in: isofs usb_storage fuse xt_CHECKSUM ipt_MASQUERADE nf_nat_masquerade_ipv4 tun nf_conntrack_netbios_ns nf_conntrack_broadcast ip6t_REJECT nf_reject_ipv6
soundcore wmi acpi_als pinctrl_sunrisepoint kfifo_buf tpm_tis industrialio acpi_pad pinctrl_intel tpm_tis_core tpm nfsd auth_rpcgss nfs_acl lockd grace sunrpc dm_crypt
CPU: 0 PID: 20929 Comm: blkid Not tainted 4.8.0-rc8-00087-gbe67d60ba944 #1
Hardware name: System manufacturer System Product Name/Z170-K, BIOS 1803 05/06/2016
task: ffff8faa93ecd940 task.stack: ffff8faa7f478000
RIP: page_cache_tree_insert+0xf1/0x100
Call Trace:
__add_to_page_cache_locked+0x12e/0x270
add_to_page_cache_lru+0x4e/0xe0
mpage_readpages+0x112/0x1d0
blkdev_readpages+0x1d/0x20
__do_page_cache_readahead+0x1ad/0x290
force_page_cache_readahead+0xaa/0x100
page_cache_sync_readahead+0x3f/0x50
generic_file_read_iter+0x5af/0x740
blkdev_read_iter+0x35/0x40
__vfs_read+0xe1/0x130
vfs_read+0x96/0x130
SyS_read+0x55/0xc0
entry_SYSCALL_64_fastpath+0x13/0x8f
Code: 03 00 48 8b 5d d8 65 48 33 1c 25 28 00 00 00 44 89 e8 75 19 48 83 c4 18 5b 41 5c 41 5d 41 5e 5d c3 0f 0b 41 bd ef ff ff ff eb d7 <0f> 0b e8 88 68 ef ff 0f 1f 84 00
RIP page_cache_tree_insert+0xf1/0x100
This is a long-standing bug in the way shadow entries are accounted in
the radix tree nodes. The shrinker needs to know when radix tree nodes
contain only shadow entries, no pages, so node->count is split in half
to count shadows in the upper bits and pages in the lower bits.
Unfortunately, the radix tree implementation doesn't know of this and
assumes all entries are in node->count. When there is a shadow entry
directly in root->rnode and the tree is later extended, the radix tree
implementation will copy that entry into the new node and and bump its
node->count, i.e. increases the page count bits. Once the shadow gets
removed and we subtract from the upper counter, node->count underflows
and triggers the warning. Afterwards, without node->count reaching 0
again, the radix tree node is leaked.
Limit shadow entries to when we have actual radix tree nodes and can
count them properly. That means we lose the ability to detect refaults
from files that had only the first page faulted in at eviction time.
Fixes: 449dd6984d0e ("mm: keep page cache radix tree nodes in check")
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reported-and-tested-by: Linus Torvalds <torvalds@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-04 22:02:08 +02:00
|
|
|
|
2014-04-03 14:47:49 -07:00
|
|
|
|
|
|
|
|
|
2021-05-08 20:04:05 -04:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-05-08 20:04:05 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-05-08 20:04:05 -04:00
|
|
|
|
|
|
|
|
|
mm: __delete_from_page_cache show Bad page if mapped
Commit e1534ae95004 ("mm: differentiate page_mapped() from
page_mapcount() for compound pages") changed the famous
BUG_ON(page_mapped(page)) in __delete_from_page_cache() to
VM_BUG_ON_PAGE(page_mapped(page)): which gives us more info when
CONFIG_DEBUG_VM=y, but nothing at all when not.
Although it has not usually been very helpul, being hit long after the
error in question, we do need to know if it actually happens on users'
systems; but reinstating a crash there is likely to be opposed :)
In the non-debug case, pr_alert("BUG: Bad page cache") plus dump_page(),
dump_stack(), add_taint() - I don't really believe LOCKDEP_NOW_UNRELIABLE,
but that seems to be the standard procedure now. Move that, or the
VM_BUG_ON_PAGE(), up before the deletion from tree: so that the
unNULLified page->mapping gives a little more information.
If the inode is being evicted (rather than truncated), it won't have any
vmas left, so it's safe(ish) to assume that the raised mapcount is
erroneous, and we can discount it from page_count to avoid leaking the
page (I'm less worried by leaking the occasional 4kB, than losing a
potential 2MB page with each 4kB page leaked).
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-09 14:08:07 -08:00
|
|
|
|
2021-05-08 20:04:05 -04:00
|
|
|
|
|
|
|
|
|
mm: __delete_from_page_cache show Bad page if mapped
Commit e1534ae95004 ("mm: differentiate page_mapped() from
page_mapcount() for compound pages") changed the famous
BUG_ON(page_mapped(page)) in __delete_from_page_cache() to
VM_BUG_ON_PAGE(page_mapped(page)): which gives us more info when
CONFIG_DEBUG_VM=y, but nothing at all when not.
Although it has not usually been very helpul, being hit long after the
error in question, we do need to know if it actually happens on users'
systems; but reinstating a crash there is likely to be opposed :)
In the non-debug case, pr_alert("BUG: Bad page cache") plus dump_page(),
dump_stack(), add_taint() - I don't really believe LOCKDEP_NOW_UNRELIABLE,
but that seems to be the standard procedure now. Move that, or the
VM_BUG_ON_PAGE(), up before the deletion from tree: so that the
unNULLified page->mapping gives a little more information.
If the inode is being evicted (rather than truncated), it won't have any
vmas left, so it's safe(ish) to assume that the raised mapcount is
erroneous, and we can discount it from page_count to avoid leaking the
page (I'm less worried by leaking the occasional 4kB, than losing a
potential 2MB page with each 4kB page leaked).
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-09 14:08:07 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-03-24 18:09:52 -07:00
|
|
|
|
2024-04-09 21:22:56 +02:00
|
|
|
|
2022-03-24 18:09:52 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-05-29 13:19:03 +02:00
|
|
|
|
2022-03-24 18:09:52 -07:00
|
|
|
|
|
|
|
|
|
mm: __delete_from_page_cache show Bad page if mapped
Commit e1534ae95004 ("mm: differentiate page_mapped() from
page_mapcount() for compound pages") changed the famous
BUG_ON(page_mapped(page)) in __delete_from_page_cache() to
VM_BUG_ON_PAGE(page_mapped(page)): which gives us more info when
CONFIG_DEBUG_VM=y, but nothing at all when not.
Although it has not usually been very helpul, being hit long after the
error in question, we do need to know if it actually happens on users'
systems; but reinstating a crash there is likely to be opposed :)
In the non-debug case, pr_alert("BUG: Bad page cache") plus dump_page(),
dump_stack(), add_taint() - I don't really believe LOCKDEP_NOW_UNRELIABLE,
but that seems to be the standard procedure now. Move that, or the
VM_BUG_ON_PAGE(), up before the deletion from tree: so that the
unNULLified page->mapping gives a little more information.
If the inode is being evicted (rather than truncated), it won't have any
vmas left, so it's safe(ish) to assume that the raised mapcount is
erroneous, and we can discount it from page_count to avoid leaking the
page (I'm less worried by leaking the occasional 4kB, than losing a
potential 2MB page with each 4kB page leaked).
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-09 14:08:07 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-08 20:04:05 -04:00
|
|
|
|
|
|
|
|
|
2017-11-15 17:37:29 -08:00
|
|
|
|
2017-07-10 15:47:35 -07:00
|
|
|
|
2021-05-08 20:04:05 -04:00
|
|
|
|
2017-11-15 17:37:29 -08:00
|
|
|
|
2021-05-08 20:04:05 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-09-23 15:38:03 -07:00
|
|
|
|
2016-07-26 15:26:18 -07:00
|
|
|
|
2025-08-21 14:55:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-11-15 17:37:29 -08:00
|
|
|
|
|
|
|
|
|
2021-05-08 20:04:05 -04:00
|
|
|
|
|
|
|
|
|
2022-03-24 18:13:59 -07:00
|
|
|
|
2017-11-15 17:37:29 -08:00
|
|
|
|
2022-03-24 18:13:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-08 20:04:05 -04:00
|
|
|
|
|
|
|
|
|
2017-11-15 17:37:29 -08:00
|
|
|
|
|
|
|
|
|
2022-03-24 18:13:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-11-15 17:37:29 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2018-04-10 16:36:56 -07:00
|
|
|
|
2017-11-15 17:37:29 -08:00
|
|
|
|
2021-05-09 09:33:42 -04:00
|
|
|
|
2017-11-15 17:37:29 -08:00
|
|
|
|
2021-05-09 09:33:42 -04:00
|
|
|
|
2017-11-15 17:37:29 -08:00
|
|
|
|
2021-07-23 09:29:46 -04:00
|
|
|
|
2021-05-08 20:04:05 -04:00
|
|
|
|
2021-05-08 00:35:49 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2021-07-28 15:52:34 -04:00
|
|
|
|
2017-11-15 17:37:18 -08:00
|
|
|
|
2022-05-01 07:35:31 -04:00
|
|
|
|
2017-11-15 17:37:18 -08:00
|
|
|
|
2022-05-01 07:35:31 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-11-15 17:37:18 -08:00
|
|
|
|
2025-02-13 13:56:12 +08:00
|
|
|
|
2017-11-15 17:37:18 -08:00
|
|
|
|
|
|
|
|
|
2011-03-22 16:32:43 -07:00
|
|
|
|
2021-05-09 09:33:42 -04:00
|
|
|
|
|
|
|
|
|
2011-03-22 16:32:43 -07:00
|
|
|
|
2021-05-09 09:33:42 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2011-03-22 16:32:43 -07:00
|
|
|
|
2021-05-09 09:33:42 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-05-09 09:33:42 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-05-09 09:33:42 -04:00
|
|
|
|
vfs: keep inodes with page cache off the inode shrinker LRU
Historically (pre-2.5), the inode shrinker used to reclaim only empty
inodes and skip over those that still contained page cache. This caused
problems on highmem hosts: struct inode could put fill lowmem zones
before the cache was getting reclaimed in the highmem zones.
To address this, the inode shrinker started to strip page cache to
facilitate reclaiming lowmem. However, this comes with its own set of
problems: the shrinkers may drop actively used page cache just because
the inodes are not currently open or dirty - think working with a large
git tree. It further doesn't respect cgroup memory protection settings
and can cause priority inversions between containers.
Nowadays, the page cache also holds non-resident info for evicted cache
pages in order to detect refaults. We've come to rely heavily on this
data inside reclaim for protecting the cache workingset and driving swap
behavior. We also use it to quantify and report workload health through
psi. The latter in turn is used for fleet health monitoring, as well as
driving automated memory sizing of workloads and containers, proactive
reclaim and memory offloading schemes.
The consequences of dropping page cache prematurely is that we're seeing
subtle and not-so-subtle failures in all of the above-mentioned
scenarios, with the workload generally entering unexpected thrashing
states while losing the ability to reliably detect it.
To fix this on non-highmem systems at least, going back to rotating
inodes on the LRU isn't feasible. We've tried (commit a76cf1a474d7
("mm: don't reclaim inodes with many attached pages")) and failed
(commit 69056ee6a8a3 ("Revert "mm: don't reclaim inodes with many
attached pages"")).
The issue is mostly that shrinker pools attract pressure based on their
size, and when objects get skipped the shrinkers remember this as
deferred reclaim work. This accumulates excessive pressure on the
remaining inodes, and we can quickly eat into heavily used ones, or
dirty ones that require IO to reclaim, when there potentially is plenty
of cold, clean cache around still.
Instead, this patch keeps populated inodes off the inode LRU in the
first place - just like an open file or dirty state would. An otherwise
clean and unused inode then gets queued when the last cache entry
disappears. This solves the problem without reintroducing the reclaim
issues, and generally is a bit more scalable than having to wade through
potentially hundreds of thousands of busy inodes.
Locking is a bit tricky because the locks protecting the inode state
(i_lock) and the inode LRU (lru_list.lock) don't nest inside the
irq-safe page cache lock (i_pages.xa_lock). Page cache deletions are
serialized through i_lock, taken before the i_pages lock, to make sure
depopulated inodes are queued reliably. Additions may race with
deletions, but we'll check again in the shrinker. If additions race
with the shrinker itself, we're protected by the i_lock: if find_inode()
or iput() win, the shrinker will bail on the elevated i_count or
I_REFERENCED; if the shrinker wins and goes ahead with the inode, it
will set I_FREEING and inhibit further igets(), which will cause the
other side to create a new instance of the inode instead.
Link: https://lkml.kernel.org/r/20210614211904.14420-4-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Roman Gushchin <guro@fb.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-11-08 18:31:24 -08:00
|
|
|
|
2021-09-02 14:53:18 -07:00
|
|
|
|
2021-05-09 09:33:42 -04:00
|
|
|
|
2021-09-02 14:53:18 -07:00
|
|
|
|
vfs: keep inodes with page cache off the inode shrinker LRU
Historically (pre-2.5), the inode shrinker used to reclaim only empty
inodes and skip over those that still contained page cache. This caused
problems on highmem hosts: struct inode could put fill lowmem zones
before the cache was getting reclaimed in the highmem zones.
To address this, the inode shrinker started to strip page cache to
facilitate reclaiming lowmem. However, this comes with its own set of
problems: the shrinkers may drop actively used page cache just because
the inodes are not currently open or dirty - think working with a large
git tree. It further doesn't respect cgroup memory protection settings
and can cause priority inversions between containers.
Nowadays, the page cache also holds non-resident info for evicted cache
pages in order to detect refaults. We've come to rely heavily on this
data inside reclaim for protecting the cache workingset and driving swap
behavior. We also use it to quantify and report workload health through
psi. The latter in turn is used for fleet health monitoring, as well as
driving automated memory sizing of workloads and containers, proactive
reclaim and memory offloading schemes.
The consequences of dropping page cache prematurely is that we're seeing
subtle and not-so-subtle failures in all of the above-mentioned
scenarios, with the workload generally entering unexpected thrashing
states while losing the ability to reliably detect it.
To fix this on non-highmem systems at least, going back to rotating
inodes on the LRU isn't feasible. We've tried (commit a76cf1a474d7
("mm: don't reclaim inodes with many attached pages")) and failed
(commit 69056ee6a8a3 ("Revert "mm: don't reclaim inodes with many
attached pages"")).
The issue is mostly that shrinker pools attract pressure based on their
size, and when objects get skipped the shrinkers remember this as
deferred reclaim work. This accumulates excessive pressure on the
remaining inodes, and we can quickly eat into heavily used ones, or
dirty ones that require IO to reclaim, when there potentially is plenty
of cold, clean cache around still.
Instead, this patch keeps populated inodes off the inode LRU in the
first place - just like an open file or dirty state would. An otherwise
clean and unused inode then gets queued when the last cache entry
disappears. This solves the problem without reintroducing the reclaim
issues, and generally is a bit more scalable than having to wade through
potentially hundreds of thousands of busy inodes.
Locking is a bit tricky because the locks protecting the inode state
(i_lock) and the inode LRU (lru_list.lock) don't nest inside the
irq-safe page cache lock (i_pages.xa_lock). Page cache deletions are
serialized through i_lock, taken before the i_pages lock, to make sure
depopulated inodes are queued reliably. Additions may race with
deletions, but we'll check again in the shrinker. If additions race
with the shrinker itself, we're protected by the i_lock: if find_inode()
or iput() win, the shrinker will bail on the elevated i_count or
I_REFERENCED; if the shrinker wins and goes ahead with the inode, it
will set I_FREEING and inhibit further igets(), which will cause the
other side to create a new instance of the inode instead.
Link: https://lkml.kernel.org/r/20210614211904.14420-4-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Roman Gushchin <guro@fb.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-11-08 18:31:24 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2010-12-01 13:35:19 -05:00
|
|
|
|
2021-05-09 09:33:42 -04:00
|
|
|
|
2011-03-22 16:30:53 -07:00
|
|
|
|
|
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
2018-04-10 16:36:56 -07:00
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
2017-12-04 03:59:45 -05:00
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
2020-06-27 22:19:08 -04:00
|
|
|
|
2019-09-23 15:34:52 -07:00
|
|
|
|
2021-03-12 23:13:46 -05:00
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
2017-12-04 03:59:45 -05:00
|
|
|
|
2021-03-12 23:13:46 -05:00
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
2019-09-23 15:34:52 -07:00
|
|
|
|
|
|
|
|
|
2021-03-12 23:13:46 -05:00
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
2019-09-23 15:34:52 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
2021-03-12 23:13:46 -05:00
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
2019-09-23 15:34:52 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-12 23:13:46 -05:00
|
|
|
|
2019-09-23 15:34:52 -07:00
|
|
|
|
2020-06-27 22:19:08 -04:00
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
2019-09-23 15:34:52 -07:00
|
|
|
|
2020-06-27 22:19:08 -04:00
|
|
|
|
2017-12-04 03:59:45 -05:00
|
|
|
|
2020-06-27 22:19:08 -04:00
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
|
|
|
|
|
vfs: keep inodes with page cache off the inode shrinker LRU
Historically (pre-2.5), the inode shrinker used to reclaim only empty
inodes and skip over those that still contained page cache. This caused
problems on highmem hosts: struct inode could put fill lowmem zones
before the cache was getting reclaimed in the highmem zones.
To address this, the inode shrinker started to strip page cache to
facilitate reclaiming lowmem. However, this comes with its own set of
problems: the shrinkers may drop actively used page cache just because
the inodes are not currently open or dirty - think working with a large
git tree. It further doesn't respect cgroup memory protection settings
and can cause priority inversions between containers.
Nowadays, the page cache also holds non-resident info for evicted cache
pages in order to detect refaults. We've come to rely heavily on this
data inside reclaim for protecting the cache workingset and driving swap
behavior. We also use it to quantify and report workload health through
psi. The latter in turn is used for fleet health monitoring, as well as
driving automated memory sizing of workloads and containers, proactive
reclaim and memory offloading schemes.
The consequences of dropping page cache prematurely is that we're seeing
subtle and not-so-subtle failures in all of the above-mentioned
scenarios, with the workload generally entering unexpected thrashing
states while losing the ability to reliably detect it.
To fix this on non-highmem systems at least, going back to rotating
inodes on the LRU isn't feasible. We've tried (commit a76cf1a474d7
("mm: don't reclaim inodes with many attached pages")) and failed
(commit 69056ee6a8a3 ("Revert "mm: don't reclaim inodes with many
attached pages"")).
The issue is mostly that shrinker pools attract pressure based on their
size, and when objects get skipped the shrinkers remember this as
deferred reclaim work. This accumulates excessive pressure on the
remaining inodes, and we can quickly eat into heavily used ones, or
dirty ones that require IO to reclaim, when there potentially is plenty
of cold, clean cache around still.
Instead, this patch keeps populated inodes off the inode LRU in the
first place - just like an open file or dirty state would. An otherwise
clean and unused inode then gets queued when the last cache entry
disappears. This solves the problem without reintroducing the reclaim
issues, and generally is a bit more scalable than having to wade through
potentially hundreds of thousands of busy inodes.
Locking is a bit tricky because the locks protecting the inode state
(i_lock) and the inode LRU (lru_list.lock) don't nest inside the
irq-safe page cache lock (i_pages.xa_lock). Page cache deletions are
serialized through i_lock, taken before the i_pages lock, to make sure
depopulated inodes are queued reliably. Additions may race with
deletions, but we'll check again in the shrinker. If additions race
with the shrinker itself, we're protected by the i_lock: if find_inode()
or iput() win, the shrinker will bail on the elevated i_count or
I_REFERENCED; if the shrinker wins and goes ahead with the inode, it
will set I_FREEING and inhibit further igets(), which will cause the
other side to create a new instance of the inode instead.
Link: https://lkml.kernel.org/r/20210614211904.14420-4-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Roman Gushchin <guro@fb.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-11-08 18:31:24 -08:00
|
|
|
|
2021-09-02 14:53:18 -07:00
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
|
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
2021-07-23 09:29:46 -04:00
|
|
|
|
|
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
2021-09-02 14:53:18 -07:00
|
|
|
|
vfs: keep inodes with page cache off the inode shrinker LRU
Historically (pre-2.5), the inode shrinker used to reclaim only empty
inodes and skip over those that still contained page cache. This caused
problems on highmem hosts: struct inode could put fill lowmem zones
before the cache was getting reclaimed in the highmem zones.
To address this, the inode shrinker started to strip page cache to
facilitate reclaiming lowmem. However, this comes with its own set of
problems: the shrinkers may drop actively used page cache just because
the inodes are not currently open or dirty - think working with a large
git tree. It further doesn't respect cgroup memory protection settings
and can cause priority inversions between containers.
Nowadays, the page cache also holds non-resident info for evicted cache
pages in order to detect refaults. We've come to rely heavily on this
data inside reclaim for protecting the cache workingset and driving swap
behavior. We also use it to quantify and report workload health through
psi. The latter in turn is used for fleet health monitoring, as well as
driving automated memory sizing of workloads and containers, proactive
reclaim and memory offloading schemes.
The consequences of dropping page cache prematurely is that we're seeing
subtle and not-so-subtle failures in all of the above-mentioned
scenarios, with the workload generally entering unexpected thrashing
states while losing the ability to reliably detect it.
To fix this on non-highmem systems at least, going back to rotating
inodes on the LRU isn't feasible. We've tried (commit a76cf1a474d7
("mm: don't reclaim inodes with many attached pages")) and failed
(commit 69056ee6a8a3 ("Revert "mm: don't reclaim inodes with many
attached pages"")).
The issue is mostly that shrinker pools attract pressure based on their
size, and when objects get skipped the shrinkers remember this as
deferred reclaim work. This accumulates excessive pressure on the
remaining inodes, and we can quickly eat into heavily used ones, or
dirty ones that require IO to reclaim, when there potentially is plenty
of cold, clean cache around still.
Instead, this patch keeps populated inodes off the inode LRU in the
first place - just like an open file or dirty state would. An otherwise
clean and unused inode then gets queued when the last cache entry
disappears. This solves the problem without reintroducing the reclaim
issues, and generally is a bit more scalable than having to wade through
potentially hundreds of thousands of busy inodes.
Locking is a bit tricky because the locks protecting the inode state
(i_lock) and the inode LRU (lru_list.lock) don't nest inside the
irq-safe page cache lock (i_pages.xa_lock). Page cache deletions are
serialized through i_lock, taken before the i_pages lock, to make sure
depopulated inodes are queued reliably. Additions may race with
deletions, but we'll check again in the shrinker. If additions race
with the shrinker itself, we're protected by the i_lock: if find_inode()
or iput() win, the shrinker will bail on the elevated i_count or
I_REFERENCED; if the shrinker wins and goes ahead with the inode, it
will set I_FREEING and inhibit further igets(), which will cause the
other side to create a new instance of the inode instead.
Link: https://lkml.kernel.org/r/20210614211904.14420-4-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Roman Gushchin <guro@fb.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-11-08 18:31:24 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
|
|
|
|
|
2017-11-15 17:37:33 -08:00
|
|
|
|
|
|
|
|
|
2016-07-29 14:10:57 +02:00
|
|
|
|
2013-04-29 15:08:42 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-05-22 11:54:16 -07:00
|
|
|
|
|
|
|
|
|
2013-04-29 15:08:42 -07:00
|
|
|
|
2014-05-22 11:54:16 -07:00
|
|
|
|
|
|
|
|
|
2013-04-29 15:08:42 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2016-07-29 14:10:57 +02:00
|
|
|
|
2013-04-29 15:08:42 -07:00
|
|
|
|
2017-07-06 07:02:22 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-07-14 14:47:22 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2005-05-01 08:59:26 -07:00
|
|
|
|
|
|
|
|
|
2006-03-24 03:17:45 -08:00
|
|
|
|
2005-05-01 08:59:26 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
[PATCH] fadvise(): write commands
Add two new linux-specific fadvise extensions():
LINUX_FADV_ASYNC_WRITE: start async writeout of any dirty pages between file
offsets `offset' and `offset+len'. Any pages which are currently under
writeout are skipped, whether or not they are dirty.
LINUX_FADV_WRITE_WAIT: wait upon writeout of any dirty pages between file
offsets `offset' and `offset+len'.
By combining these two operations the application may do several things:
LINUX_FADV_ASYNC_WRITE: push some or all of the dirty pages at the disk.
LINUX_FADV_WRITE_WAIT, LINUX_FADV_ASYNC_WRITE: push all of the currently dirty
pages at the disk.
LINUX_FADV_WRITE_WAIT, LINUX_FADV_ASYNC_WRITE, LINUX_FADV_WRITE_WAIT: push all
of the currently dirty pages at the disk, wait until they have been written.
It should be noted that none of these operations write out the file's
metadata. So unless the application is strictly performing overwrites of
already-instantiated disk blocks, there are no guarantees here that the data
will be available after a crash.
To complete this suite of operations I guess we should have a "sync file
metadata only" operation. This gives applications access to all the building
blocks needed for all sorts of sync operations. But sync-metadata doesn't fit
well with the fadvise() interface. Probably it should be a new syscall:
sys_fmetadatasync().
The patch also diddles with the meaning of `endbyte' in sys_fadvise64_64().
It is made to represent that last affected byte in the file (ie: it is
inclusive). Generally, all these byterange and pagerange functions are
inclusive so we can easily represent EOF with -1.
As Ulrich notes, these two functions are somewhat abusive of the fadvise()
concept, which appears to be "set the future policy for this fd".
But these commands are a perfect fit with the fadvise() impementation, and
several of the existing fadvise() commands are synchronous and don't affect
future policy either. I think we can live with the slight incongruity.
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-03-24 03:18:04 -08:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm: write_cache_pages integrity fix
In write_cache_pages, nr_to_write is heeded even for data-integrity syncs,
so the function will return success after writing out nr_to_write pages,
even if that was not sufficient to guarantee data integrity.
The callers tend to set it to values that could break data interity
semantics easily in practice. For example, nr_to_write can be set to
mapping->nr_pages * 2, however if a file has a single, dirty page, then
fsync is called, subsequent pages might be concurrently added and dirtied,
then write_cache_pages might writeout two of these newly dirty pages,
while not writing out the old page that should have been written out.
Fix this by ignoring nr_to_write if it is a data integrity sync.
This is a data integrity bug.
The reason this has been done in the past is to avoid stalling sync
operations behind page dirtiers.
"If a file has one dirty page at offset 1000000000000000 then someone
does an fsync() and someone else gets in first and starts madly writing
pages at offset 0, we want to write that page at 1000000000000000.
Somehow."
What we do today is return success after an arbitrary amount of pages are
written, whether or not we have provided the data-integrity semantics that
the caller has asked for. Even this doesn't actually fix all stall cases
completely: in the above situation, if the file has a huge number of pages
in pagecache (but not dirty), then mapping->nrpages is going to be huge,
even if pages are being dirtied.
This change does indeed make the possibility of long stalls lager, and
that's not a good thing, but lying about data integrity is even worse. We
have to either perform the sync, or return -ELINUXISLAME so at least the
caller knows what has happened.
There are subsequent competing approaches in the works to solve the stall
problems properly, without compromising data integrity.
Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Dave Chinner <david@fromorbit.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2009-01-06 14:39:08 -08:00
|
|
|
|
[PATCH] writeback: fix range handling
When a writeback_control's `start' and `end' fields are used to
indicate a one-byte-range starting at file offset zero, the required
values of .start=0,.end=0 mean that the ->writepages() implementation
has no way of telling that it is being asked to perform a range
request. Because we're currently overloading (start == 0 && end == 0)
to mean "this is not a write-a-range request".
To make all this sane, the patch changes range of writeback_control.
So caller does: If it is calling ->writepages() to write pages, it
sets range (range_start/end or range_cyclic) always.
And if range_cyclic is true, ->writepages() thinks the range is
cyclic, otherwise it just uses range_start and range_end.
This patch does,
- Add LLONG_MAX, LLONG_MIN, ULLONG_MAX to include/linux/kernel.h
-1 is usually ok for range_end (type is long long). But, if someone did,
range_end += val; range_end is "val - 1"
u64val = range_end >> bits; u64val is "~(0ULL)"
or something, they are wrong. So, this adds LLONG_MAX to avoid nasty
things, and uses LLONG_MAX for range_end.
- All callers of ->writepages() sets range_start/end or range_cyclic.
- Fix updates of ->writeback_index. It seems already bit strange.
If it starts at 0 and ended by check of nr_to_write, this last
index may reduce chance to scan end of file. So, this updates
->writeback_index only if range_cyclic is true or whole-file is
scanned.
Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Cc: Nathan Scott <nathans@sgi.com>
Cc: Anton Altaparmakov <aia21@cantab.net>
Cc: Steven French <sfrench@us.ibm.com>
Cc: "Vladimir V. Saveliev" <vs@namesys.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 02:03:26 -07:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2021-07-14 14:47:22 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
[PATCH] writeback: fix range handling
When a writeback_control's `start' and `end' fields are used to
indicate a one-byte-range starting at file offset zero, the required
values of .start=0,.end=0 mean that the ->writepages() implementation
has no way of telling that it is being asked to perform a range
request. Because we're currently overloading (start == 0 && end == 0)
to mean "this is not a write-a-range request".
To make all this sane, the patch changes range of writeback_control.
So caller does: If it is calling ->writepages() to write pages, it
sets range (range_start/end or range_cyclic) always.
And if range_cyclic is true, ->writepages() thinks the range is
cyclic, otherwise it just uses range_start and range_end.
This patch does,
- Add LLONG_MAX, LLONG_MIN, ULLONG_MAX to include/linux/kernel.h
-1 is usually ok for range_end (type is long long). But, if someone did,
range_end += val; range_end is "val - 1"
u64val = range_end >> bits; u64val is "~(0ULL)"
or something, they are wrong. So, this adds LLONG_MAX to avoid nasty
things, and uses LLONG_MAX for range_end.
- All callers of ->writepages() sets range_start/end or range_cyclic.
- Fix updates of ->writeback_index. It seems already bit strange.
If it starts at 0 and ended by check of nr_to_write, this last
index may reduce chance to scan end of file. So, this updates
->writeback_index only if range_cyclic is true or whole-file is
scanned.
Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Cc: Nathan Scott <nathans@sgi.com>
Cc: Anton Altaparmakov <aia21@cantab.net>
Cc: Steven French <sfrench@us.ibm.com>
Cc: "Vladimir V. Saveliev" <vs@namesys.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-23 02:03:26 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2008-07-11 19:27:31 -04:00
|
|
|
|
[PATCH] fadvise(): write commands
Add two new linux-specific fadvise extensions():
LINUX_FADV_ASYNC_WRITE: start async writeout of any dirty pages between file
offsets `offset' and `offset+len'. Any pages which are currently under
writeout are skipped, whether or not they are dirty.
LINUX_FADV_WRITE_WAIT: wait upon writeout of any dirty pages between file
offsets `offset' and `offset+len'.
By combining these two operations the application may do several things:
LINUX_FADV_ASYNC_WRITE: push some or all of the dirty pages at the disk.
LINUX_FADV_WRITE_WAIT, LINUX_FADV_ASYNC_WRITE: push all of the currently dirty
pages at the disk.
LINUX_FADV_WRITE_WAIT, LINUX_FADV_ASYNC_WRITE, LINUX_FADV_WRITE_WAIT: push all
of the currently dirty pages at the disk, wait until they have been written.
It should be noted that none of these operations write out the file's
metadata. So unless the application is strictly performing overwrites of
already-instantiated disk blocks, there are no guarantees here that the data
will be available after a crash.
To complete this suite of operations I guess we should have a "sync file
metadata only" operation. This gives applications access to all the building
blocks needed for all sorts of sync operations. But sync-metadata doesn't fit
well with the fadvise() interface. Probably it should be a new syscall:
sys_fmetadatasync().
The patch also diddles with the meaning of `endbyte' in sys_fadvise64_64().
It is made to represent that last affected byte in the file (ie: it is
inclusive). Generally, all these byterange and pagerange functions are
inclusive so we can easily represent EOF with -1.
As Ulrich notes, these two functions are somewhat abusive of the fadvise()
concept, which appears to be "set the future policy for this fd".
But these commands are a perfect fit with the fadvise() impementation, and
several of the existing fadvise() commands are synchronous and don't affect
future policy either. I think we can live with the slight incongruity.
Cc: Michael Kerrisk <mtk-manpages@gmx.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-03-24 03:18:04 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2008-07-11 19:27:31 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2024-12-20 08:47:48 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-02-18 20:02:08 +08:00
|
|
|
|
2024-12-20 08:47:48 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-06-20 07:05:41 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-06-20 07:05:41 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-16 19:39:40 +00:00
|
|
|
|
2018-01-16 06:26:49 -05:00
|
|
|
|
|
|
|
|
|
2017-06-20 07:05:41 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2018-01-16 06:26:49 -05:00
|
|
|
|
|
|
|
|
|
2023-01-16 19:39:40 +00:00
|
|
|
|
|
|
|
|
|
2018-01-16 06:26:49 -05:00
|
|
|
|
|
|
|
|
|
2023-01-16 19:39:40 +00:00
|
|
|
|
2018-01-16 06:26:49 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-06-20 07:05:41 -05:00
|
|
|
|
2023-01-16 19:39:40 +00:00
|
|
|
|
2017-06-20 07:05:41 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-07-06 07:02:24 -04:00
|
|
|
|
2015-11-05 18:47:23 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros
PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time
ago with promise that one day it will be possible to implement page
cache with bigger chunks than PAGE_SIZE.
This promise never materialized. And unlikely will.
We have many places where PAGE_CACHE_SIZE assumed to be equal to
PAGE_SIZE. And it's constant source of confusion on whether
PAGE_CACHE_* or PAGE_* constant should be used in a particular case,
especially on the border between fs and mm.
Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much
breakage to be doable.
Let's stop pretending that pages in page cache are special. They are
not.
The changes are pretty straight-forward:
- <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>;
- PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN};
- page_cache_get() -> get_page();
- page_cache_release() -> put_page();
This patch contains automated changes generated with coccinelle using
script below. For some reason, coccinelle doesn't patch header files.
I've called spatch for them manually.
The only adjustment after coccinelle is revert of changes to
PAGE_CAHCE_ALIGN definition: we are going to drop it later.
There are few places in the code where coccinelle didn't reach. I'll
fix them manually in a separate patch. Comments and documentation also
will be addressed with the separate patch.
virtual patch
@@
expression E;
@@
- E << (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
expression E;
@@
- E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT)
+ E
@@
@@
- PAGE_CACHE_SHIFT
+ PAGE_SHIFT
@@
@@
- PAGE_CACHE_SIZE
+ PAGE_SIZE
@@
@@
- PAGE_CACHE_MASK
+ PAGE_MASK
@@
expression E;
@@
- PAGE_CACHE_ALIGN(E)
+ PAGE_ALIGN(E)
@@
expression E;
@@
- page_cache_get(E)
+ get_page(E)
@@
expression E;
@@
- page_cache_release(E)
+ put_page(E)
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 15:29:47 +03:00
|
|
|
|
|
|
|
|
|
2023-01-04 13:14:28 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2017-11-15 17:35:05 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2023-01-04 13:14:28 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-11-15 17:35:05 -08:00
|
|
|
|
|
|
|
|
|
2023-01-04 13:14:28 -08:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-01-04 13:14:28 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-01-04 13:14:28 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2015-11-05 18:47:23 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
2015-11-05 18:47:23 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-07-06 07:02:24 -04:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2009-08-17 19:30:27 +02:00
|
|
|
|
|
|
|
|
|
2019-06-20 17:05:37 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-07-28 07:24:43 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
2017-07-28 07:24:43 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2009-08-17 19:30:27 +02:00
|
|
|
|
2015-11-05 18:47:23 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
2015-11-05 18:47:23 -08:00
|
|
|
|
2017-07-06 07:02:22 -04:00
|
|
|
|
2015-11-05 18:47:23 -08:00
|
|
|
|
2017-07-31 10:29:38 -04:00
|
|
|
|
2017-07-06 07:02:22 -04:00
|
|
|
|
2015-11-05 18:47:23 -08:00
|
|
|
|
2017-07-06 07:02:22 -04:00
|
|
|
|
2015-11-05 18:47:23 -08:00
|
|
|
|
2019-09-23 15:34:48 -07:00
|
|
|
|
2017-07-26 10:21:11 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2019-09-23 15:34:48 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2021-10-28 08:47:05 -06:00
|
|
|
|
|
|
|
|
|
2021-11-05 13:37:13 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-09-05 14:45:57 -07:00
|
|
|
|
2021-11-05 13:37:13 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-09-05 14:45:57 -07:00
|
|
|
|
|
|
|
|
|
2021-11-05 13:37:13 -07:00
|
|
|
|
2022-09-05 14:45:57 -07:00
|
|
|
|
2021-11-05 13:37:13 -07:00
|
|
|
|
2022-09-05 14:45:57 -07:00
|
|
|
|
|
|
|
|
|
2021-11-05 13:37:13 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-09-05 14:45:57 -07:00
|
|
|
|
mm: provide filemap_range_needs_writeback() helper
Patch series "Improve IOCB_NOWAIT O_DIRECT reads", v3.
An internal workload complained because it was using too much CPU, and
when I took a look, we had a lot of io_uring workers going to town.
For an async buffered read like workload, I am normally expecting _zero_
offloads to a worker thread, but this one had tons of them. I'd drop
caches and things would look good again, but then a minute later we'd
regress back to using workers. Turns out that every minute something
was reading parts of the device, which would add page cache for that
inode. I put patches like these in for our kernel, and the problem was
solved.
Don't -EAGAIN IOCB_NOWAIT dio reads just because we have page cache
entries for the given range. This causes unnecessary work from the
callers side, when the IO could have been issued totally fine without
blocking on writeback when there is none.
This patch (of 3):
For O_DIRECT reads/writes, we check if we need to issue a call to
filemap_write_and_wait_range() to issue and/or wait for writeback for any
page in the given range. The existing mechanism just checks for a page in
the range, which is suboptimal for IOCB_NOWAIT as we'll fallback to the
slow path (and needing retry) if there's just a clean page cache page in
the range.
Provide filemap_range_needs_writeback() which tries a little harder to
check if we actually need to issue and/or wait for writeback in the range.
Link: https://lkml.kernel.org/r/20210224164455.1096727-1-axboe@kernel.dk
Link: https://lkml.kernel.org/r/20210224164455.1096727-2-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-04-29 22:55:18 -07:00
|
|
|
|
2021-10-28 08:47:05 -06:00
|
|
|
|
mm: provide filemap_range_needs_writeback() helper
Patch series "Improve IOCB_NOWAIT O_DIRECT reads", v3.
An internal workload complained because it was using too much CPU, and
when I took a look, we had a lot of io_uring workers going to town.
For an async buffered read like workload, I am normally expecting _zero_
offloads to a worker thread, but this one had tons of them. I'd drop
caches and things would look good again, but then a minute later we'd
regress back to using workers. Turns out that every minute something
was reading parts of the device, which would add page cache for that
inode. I put patches like these in for our kernel, and the problem was
solved.
Don't -EAGAIN IOCB_NOWAIT dio reads just because we have page cache
entries for the given range. This causes unnecessary work from the
callers side, when the IO could have been issued totally fine without
blocking on writeback when there is none.
This patch (of 3):
For O_DIRECT reads/writes, we check if we need to issue a call to
filemap_write_and_wait_range() to issue and/or wait for writeback for any
page in the given range. The existing mechanism just checks for a page in
the range, which is suboptimal for IOCB_NOWAIT as we'll fallback to the
slow path (and needing retry) if there's just a clean page cache page in
the range.
Provide filemap_range_needs_writeback() which tries a little harder to
check if we actually need to issue and/or wait for writeback in the range.
Link: https://lkml.kernel.org/r/20210224164455.1096727-1-axboe@kernel.dk
Link: https://lkml.kernel.org/r/20210224164455.1096727-2-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-04-29 22:55:18 -07:00
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-03-24 03:17:45 -08:00
|
|
|
|
|
|
|
|
|
2017-03-30 17:11:36 -03:00
|
|
|
|
2006-03-24 03:17:45 -08:00
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
2006-03-24 03:17:45 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-06-27 21:23:51 +08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
filemap: skip write and wait if end offset precedes start
Patch series "filemap: skip write and wait if end offset precedes start",
v2.
A fix for the odd write and wait behavior described in the patch 1 commit
log. Technically patch 1 could simply remove the check rather than lift
it into the callers, but this seemed a bit more user friendly to me.
Patch 2 is appended after observation that fadvise() interacted poorly
with the v1 patch. This is no longer a problem with v2, making patch 2
purely a cleanup.
This series survived both fstests and ltp regression runs without
observable problems. I had (end < start) warning checks in each relevant
function, with fadvise() being the only caller that triggered them. That
said, I dropped the warnings after testing because there seemed to much
potential for noise from the various other callers.
This patch (of 2):
A call to file[map]_write_and_wait_range() with an end offset that
precedes the start offset but happens to land in the same page can trigger
writeback submission but fails to wait on the submitted page. Writeback
submission occurs because __filemap_fdatawrite_range() passes both offsets
down into write_cache_pages(), which rounds down to page indexes before it
starts processing writeback. However, __filemap_fdatawait_range()
immediately returns if the byte-granular end offset precedes the start
offset.
This behavior was observed in the form of unpredictable latency from a
frequent write and wait call with incorrect parameters. The behavior gave
the impression that the fdatawait path might occasionally fail to wait on
writeback, but further investigation showed the latency was from
write_cache_pages() waiting on writeback state to clear for a page already
under writeback. Therefore, this indicated that fdatawait actually never
waits on writeback in this particular situation.
The byte granular check in __filemap_fdatawait_range() goes all the way
back to the old wait_on_page_writeback() helper. It originally used page
offsets and so would have waited in this problematic case. That changed
to byte granularity file offsets in commit 94004ed726f3 ("kill
wait_on_page_writeback_range"), which subtly changed this behavior. The
check itself has become somewhat redundant since the error checking code
that used to follow the wait loop (at the time of the aforementioned
commit) has now been removed and lifted into the higher level callers.
Therefore, we can restore historical fdatawait behavior by simply removing
the check. Since the current fdatawait behavior has been in place for
quite some time and is consistent with other interfaces that use file
offsets, instead lift the check into the file[map]_write_and_wait_range()
helpers to provide consistent behavior between the write and wait.
Link: https://lkml.kernel.org/r/20221128155632.3950447-1-bfoster@redhat.com
Link: https://lkml.kernel.org/r/20221128155632.3950447-2-bfoster@redhat.com
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-11-28 10:56:31 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-07-26 10:21:11 -04:00
|
|
|
|
[PATCH] Fix and add EXPORT_SYMBOL(filemap_write_and_wait)
This patch add EXPORT_SYMBOL(filemap_write_and_wait) and use it.
See mm/filemap.c:
And changes the filemap_write_and_wait() and filemap_write_and_wait_range().
Current filemap_write_and_wait() doesn't wait if filemap_fdatawrite()
returns error. However, even if filemap_fdatawrite() returned an
error, it may have submitted the partially data pages to the device.
(e.g. in the case of -ENOSPC)
<quotation>
Andrew Morton writes,
If filemap_fdatawrite() returns an error, this might be due to some
I/O problem: dead disk, unplugged cable, etc. Given the generally
crappy quality of the kernel's handling of such exceptions, there's a
good chance that the filemap_fdatawait() will get stuck in D state
forever.
</quotation>
So, this patch doesn't wait if filemap_fdatawrite() returns the -EIO.
Trond, could you please review the nfs part? Especially I'm not sure,
nfs must use the "filemap_fdatawrite(inode->i_mapping) == 0", or not.
Acked-by: Trond Myklebust <trond.myklebust@fys.uio.no>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-08 01:02:14 -08:00
|
|
|
|
|
|
|
|
|
2020-01-30 22:12:07 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-06-27 21:23:51 +08:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2022-06-27 21:23:51 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
[PATCH] Fix and add EXPORT_SYMBOL(filemap_write_and_wait)
This patch add EXPORT_SYMBOL(filemap_write_and_wait) and use it.
See mm/filemap.c:
And changes the filemap_write_and_wait() and filemap_write_and_wait_range().
Current filemap_write_and_wait() doesn't wait if filemap_fdatawrite()
returns error. However, even if filemap_fdatawrite() returned an
error, it may have submitted the partially data pages to the device.
(e.g. in the case of -ENOSPC)
<quotation>
Andrew Morton writes,
If filemap_fdatawrite() returns an error, this might be due to some
I/O problem: dead disk, unplugged cable, etc. Given the generally
crappy quality of the kernel's handling of such exceptions, there's a
good chance that the filemap_fdatawait() will get stuck in D state
forever.
</quotation>
So, this patch doesn't wait if filemap_fdatawrite() returns the -EIO.
Trond, could you please review the nfs part? Especially I'm not sure,
nfs must use the "filemap_fdatawrite(inode->i_mapping) == 0", or not.
Acked-by: Trond Myklebust <trond.myklebust@fys.uio.no>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-08 01:02:14 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2009-04-15 13:22:37 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
fs: new infrastructure for writeback error handling and reporting
Most filesystems currently use mapping_set_error and
filemap_check_errors for setting and reporting/clearing writeback errors
at the mapping level. filemap_check_errors is indirectly called from
most of the filemap_fdatawait_* functions and from
filemap_write_and_wait*. These functions are called from all sorts of
contexts to wait on writeback to finish -- e.g. mostly in fsync, but
also in truncate calls, getattr, etc.
The non-fsync callers are problematic. We should be reporting writeback
errors during fsync, but many places spread over the tree clear out
errors before they can be properly reported, or report errors at
nonsensical times.
If I get -EIO on a stat() call, there is no reason for me to assume that
it is because some previous writeback failed. The fact that it also
clears out the error such that a subsequent fsync returns 0 is a bug,
and a nasty one since that's potentially silent data corruption.
This patch adds a small bit of new infrastructure for setting and
reporting errors during address_space writeback. While the above was my
original impetus for adding this, I think it's also the case that
current fsync semantics are just problematic for userland. Most
applications that call fsync do so to ensure that the data they wrote
has hit the backing store.
In the case where there are multiple writers to the file at the same
time, this is really hard to determine. The first one to call fsync will
see any stored error, and the rest get back 0. The processes with open
fds may not be associated with one another in any way. They could even
be in different containers, so ensuring coordination between all fsync
callers is not really an option.
One way to remedy this would be to track what file descriptor was used
to dirty the file, but that's rather cumbersome and would likely be
slow. However, there is a simpler way to improve the semantics here
without incurring too much overhead.
This set adds an errseq_t to struct address_space, and a corresponding
one is added to struct file. Writeback errors are recorded in the
mapping's errseq_t, and the one in struct file is used as the "since"
value.
This changes the semantics of the Linux fsync implementation such that
applications can now use it to determine whether there were any
writeback errors since fsync(fd) was last called (or since the file was
opened in the case of fsync having never been called).
Note that those writeback errors may have occurred when writing data
that was dirtied via an entirely different fd, but that's the case now
with the current mapping_set_error/filemap_check_error infrastructure.
This will at least prevent you from getting a false report of success.
The new behavior is still consistent with the POSIX spec, and is more
reliable for application developers. This patch just adds some basic
infrastructure for doing this, and ensures that the f_wb_err "cursor"
is properly set when a file is opened. Later patches will change the
existing code to use this new infrastructure for reporting errors at
fsync time.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
2017-07-06 07:02:25 -04:00
|
|
|
|
|
|
|
|
|
2017-07-24 06:22:15 -04:00
|
|
|
|
fs: new infrastructure for writeback error handling and reporting
Most filesystems currently use mapping_set_error and
filemap_check_errors for setting and reporting/clearing writeback errors
at the mapping level. filemap_check_errors is indirectly called from
most of the filemap_fdatawait_* functions and from
filemap_write_and_wait*. These functions are called from all sorts of
contexts to wait on writeback to finish -- e.g. mostly in fsync, but
also in truncate calls, getattr, etc.
The non-fsync callers are problematic. We should be reporting writeback
errors during fsync, but many places spread over the tree clear out
errors before they can be properly reported, or report errors at
nonsensical times.
If I get -EIO on a stat() call, there is no reason for me to assume that
it is because some previous writeback failed. The fact that it also
clears out the error such that a subsequent fsync returns 0 is a bug,
and a nasty one since that's potentially silent data corruption.
This patch adds a small bit of new infrastructure for setting and
reporting errors during address_space writeback. While the above was my
original impetus for adding this, I think it's also the case that
current fsync semantics are just problematic for userland. Most
applications that call fsync do so to ensure that the data they wrote
has hit the backing store.
In the case where there are multiple writers to the file at the same
time, this is really hard to determine. The first one to call fsync will
see any stored error, and the rest get back 0. The processes with open
fds may not be associated with one another in any way. They could even
be in different containers, so ensuring coordination between all fsync
callers is not really an option.
One way to remedy this would be to track what file descriptor was used
to dirty the file, but that's rather cumbersome and would likely be
slow. However, there is a simpler way to improve the semantics here
without incurring too much overhead.
This set adds an errseq_t to struct address_space, and a corresponding
one is added to struct file. Writeback errors are recorded in the
mapping's errseq_t, and the one in struct file is used as the "since"
value.
This changes the semantics of the Linux fsync implementation such that
applications can now use it to determine whether there were any
writeback errors since fsync(fd) was last called (or since the file was
opened in the case of fsync having never been called).
Note that those writeback errors may have occurred when writing data
that was dirtied via an entirely different fd, but that's the case now
with the current mapping_set_error/filemap_check_error infrastructure.
This will at least prevent you from getting a false report of success.
The new behavior is still consistent with the POSIX spec, and is more
reliable for application developers. This patch just adds some basic
infrastructure for doing this, and ensures that the f_wb_err "cursor"
is properly set when a file is opened. Later patches will change the
existing code to use this new infrastructure for reporting errors at
fsync time.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
2017-07-06 07:02:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
fs: new infrastructure for writeback error handling and reporting
Most filesystems currently use mapping_set_error and
filemap_check_errors for setting and reporting/clearing writeback errors
at the mapping level. filemap_check_errors is indirectly called from
most of the filemap_fdatawait_* functions and from
filemap_write_and_wait*. These functions are called from all sorts of
contexts to wait on writeback to finish -- e.g. mostly in fsync, but
also in truncate calls, getattr, etc.
The non-fsync callers are problematic. We should be reporting writeback
errors during fsync, but many places spread over the tree clear out
errors before they can be properly reported, or report errors at
nonsensical times.
If I get -EIO on a stat() call, there is no reason for me to assume that
it is because some previous writeback failed. The fact that it also
clears out the error such that a subsequent fsync returns 0 is a bug,
and a nasty one since that's potentially silent data corruption.
This patch adds a small bit of new infrastructure for setting and
reporting errors during address_space writeback. While the above was my
original impetus for adding this, I think it's also the case that
current fsync semantics are just problematic for userland. Most
applications that call fsync do so to ensure that the data they wrote
has hit the backing store.
In the case where there are multiple writers to the file at the same
time, this is really hard to determine. The first one to call fsync will
see any stored error, and the rest get back 0. The processes with open
fds may not be associated with one another in any way. They could even
be in different containers, so ensuring coordination between all fsync
callers is not really an option.
One way to remedy this would be to track what file descriptor was used
to dirty the file, but that's rather cumbersome and would likely be
slow. However, there is a simpler way to improve the semantics here
without incurring too much overhead.
This set adds an errseq_t to struct address_space, and a corresponding
one is added to struct file. Writeback errors are recorded in the
mapping's errseq_t, and the one in struct file is used as the "since"
value.
This changes the semantics of the Linux fsync implementation such that
applications can now use it to determine whether there were any
writeback errors since fsync(fd) was last called (or since the file was
opened in the case of fsync having never been called).
Note that those writeback errors may have occurred when writing data
that was dirtied via an entirely different fd, but that's the case now
with the current mapping_set_error/filemap_check_error infrastructure.
This will at least prevent you from getting a false report of success.
The new behavior is still consistent with the POSIX spec, and is more
reliable for application developers. This patch just adds some basic
infrastructure for doing this, and ensures that the f_wb_err "cursor"
is properly set when a file is opened. Later patches will change the
existing code to use this new infrastructure for reporting errors at
fsync time.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
2017-07-06 07:02:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-10-03 16:15:25 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
fs: new infrastructure for writeback error handling and reporting
Most filesystems currently use mapping_set_error and
filemap_check_errors for setting and reporting/clearing writeback errors
at the mapping level. filemap_check_errors is indirectly called from
most of the filemap_fdatawait_* functions and from
filemap_write_and_wait*. These functions are called from all sorts of
contexts to wait on writeback to finish -- e.g. mostly in fsync, but
also in truncate calls, getattr, etc.
The non-fsync callers are problematic. We should be reporting writeback
errors during fsync, but many places spread over the tree clear out
errors before they can be properly reported, or report errors at
nonsensical times.
If I get -EIO on a stat() call, there is no reason for me to assume that
it is because some previous writeback failed. The fact that it also
clears out the error such that a subsequent fsync returns 0 is a bug,
and a nasty one since that's potentially silent data corruption.
This patch adds a small bit of new infrastructure for setting and
reporting errors during address_space writeback. While the above was my
original impetus for adding this, I think it's also the case that
current fsync semantics are just problematic for userland. Most
applications that call fsync do so to ensure that the data they wrote
has hit the backing store.
In the case where there are multiple writers to the file at the same
time, this is really hard to determine. The first one to call fsync will
see any stored error, and the rest get back 0. The processes with open
fds may not be associated with one another in any way. They could even
be in different containers, so ensuring coordination between all fsync
callers is not really an option.
One way to remedy this would be to track what file descriptor was used
to dirty the file, but that's rather cumbersome and would likely be
slow. However, there is a simpler way to improve the semantics here
without incurring too much overhead.
This set adds an errseq_t to struct address_space, and a corresponding
one is added to struct file. Writeback errors are recorded in the
mapping's errseq_t, and the one in struct file is used as the "since"
value.
This changes the semantics of the Linux fsync implementation such that
applications can now use it to determine whether there were any
writeback errors since fsync(fd) was last called (or since the file was
opened in the case of fsync having never been called).
Note that those writeback errors may have occurred when writing data
that was dirtied via an entirely different fd, but that's the case now
with the current mapping_set_error/filemap_check_error infrastructure.
This will at least prevent you from getting a false report of success.
The new behavior is still consistent with the POSIX spec, and is more
reliable for application developers. This patch just adds some basic
infrastructure for doing this, and ensures that the f_wb_err "cursor"
is properly set when a file is opened. Later patches will change the
existing code to use this new infrastructure for reporting errors at
fsync time.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
2017-07-06 07:02:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
fs: new infrastructure for writeback error handling and reporting
Most filesystems currently use mapping_set_error and
filemap_check_errors for setting and reporting/clearing writeback errors
at the mapping level. filemap_check_errors is indirectly called from
most of the filemap_fdatawait_* functions and from
filemap_write_and_wait*. These functions are called from all sorts of
contexts to wait on writeback to finish -- e.g. mostly in fsync, but
also in truncate calls, getattr, etc.
The non-fsync callers are problematic. We should be reporting writeback
errors during fsync, but many places spread over the tree clear out
errors before they can be properly reported, or report errors at
nonsensical times.
If I get -EIO on a stat() call, there is no reason for me to assume that
it is because some previous writeback failed. The fact that it also
clears out the error such that a subsequent fsync returns 0 is a bug,
and a nasty one since that's potentially silent data corruption.
This patch adds a small bit of new infrastructure for setting and
reporting errors during address_space writeback. While the above was my
original impetus for adding this, I think it's also the case that
current fsync semantics are just problematic for userland. Most
applications that call fsync do so to ensure that the data they wrote
has hit the backing store.
In the case where there are multiple writers to the file at the same
time, this is really hard to determine. The first one to call fsync will
see any stored error, and the rest get back 0. The processes with open
fds may not be associated with one another in any way. They could even
be in different containers, so ensuring coordination between all fsync
callers is not really an option.
One way to remedy this would be to track what file descriptor was used
to dirty the file, but that's rather cumbersome and would likely be
slow. However, there is a simpler way to improve the semantics here
without incurring too much overhead.
This set adds an errseq_t to struct address_space, and a corresponding
one is added to struct file. Writeback errors are recorded in the
mapping's errseq_t, and the one in struct file is used as the "since"
value.
This changes the semantics of the Linux fsync implementation such that
applications can now use it to determine whether there were any
writeback errors since fsync(fd) was last called (or since the file was
opened in the case of fsync having never been called).
Note that those writeback errors may have occurred when writing data
that was dirtied via an entirely different fd, but that's the case now
with the current mapping_set_error/filemap_check_error infrastructure.
This will at least prevent you from getting a false report of success.
The new behavior is still consistent with the POSIX spec, and is more
reliable for application developers. This patch just adds some basic
infrastructure for doing this, and ensures that the f_wb_err "cursor"
is properly set when a file is opened. Later patches will change the
existing code to use this new infrastructure for reporting errors at
fsync time.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
2017-07-06 07:02:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
filemap: skip write and wait if end offset precedes start
Patch series "filemap: skip write and wait if end offset precedes start",
v2.
A fix for the odd write and wait behavior described in the patch 1 commit
log. Technically patch 1 could simply remove the check rather than lift
it into the callers, but this seemed a bit more user friendly to me.
Patch 2 is appended after observation that fadvise() interacted poorly
with the v1 patch. This is no longer a problem with v2, making patch 2
purely a cleanup.
This series survived both fstests and ltp regression runs without
observable problems. I had (end < start) warning checks in each relevant
function, with fadvise() being the only caller that triggered them. That
said, I dropped the warnings after testing because there seemed to much
potential for noise from the various other callers.
This patch (of 2):
A call to file[map]_write_and_wait_range() with an end offset that
precedes the start offset but happens to land in the same page can trigger
writeback submission but fails to wait on the submitted page. Writeback
submission occurs because __filemap_fdatawrite_range() passes both offsets
down into write_cache_pages(), which rounds down to page indexes before it
starts processing writeback. However, __filemap_fdatawait_range()
immediately returns if the byte-granular end offset precedes the start
offset.
This behavior was observed in the form of unpredictable latency from a
frequent write and wait call with incorrect parameters. The behavior gave
the impression that the fdatawait path might occasionally fail to wait on
writeback, but further investigation showed the latency was from
write_cache_pages() waiting on writeback state to clear for a page already
under writeback. Therefore, this indicated that fdatawait actually never
waits on writeback in this particular situation.
The byte granular check in __filemap_fdatawait_range() goes all the way
back to the old wait_on_page_writeback() helper. It originally used page
offsets and so would have waited in this problematic case. That changed
to byte granularity file offsets in commit 94004ed726f3 ("kill
wait_on_page_writeback_range"), which subtly changed this behavior. The
check itself has become somewhat redundant since the error checking code
that used to follow the wait loop (at the time of the aforementioned
commit) has now been removed and lifted into the higher level callers.
Therefore, we can restore historical fdatawait behavior by simply removing
the check. Since the current fdatawait behavior has been in place for
quite some time and is consistent with other interfaces that use file
offsets, instead lift the check into the file[map]_write_and_wait_range()
helpers to provide consistent behavior between the write and wait.
Link: https://lkml.kernel.org/r/20221128155632.3950447-1-bfoster@redhat.com
Link: https://lkml.kernel.org/r/20221128155632.3950447-2-bfoster@redhat.com
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2022-11-28 10:56:31 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-07-26 10:21:11 -04:00
|
|
|
|
fs: new infrastructure for writeback error handling and reporting
Most filesystems currently use mapping_set_error and
filemap_check_errors for setting and reporting/clearing writeback errors
at the mapping level. filemap_check_errors is indirectly called from
most of the filemap_fdatawait_* functions and from
filemap_write_and_wait*. These functions are called from all sorts of
contexts to wait on writeback to finish -- e.g. mostly in fsync, but
also in truncate calls, getattr, etc.
The non-fsync callers are problematic. We should be reporting writeback
errors during fsync, but many places spread over the tree clear out
errors before they can be properly reported, or report errors at
nonsensical times.
If I get -EIO on a stat() call, there is no reason for me to assume that
it is because some previous writeback failed. The fact that it also
clears out the error such that a subsequent fsync returns 0 is a bug,
and a nasty one since that's potentially silent data corruption.
This patch adds a small bit of new infrastructure for setting and
reporting errors during address_space writeback. While the above was my
original impetus for adding this, I think it's also the case that
current fsync semantics are just problematic for userland. Most
applications that call fsync do so to ensure that the data they wrote
has hit the backing store.
In the case where there are multiple writers to the file at the same
time, this is really hard to determine. The first one to call fsync will
see any stored error, and the rest get back 0. The processes with open
fds may not be associated with one another in any way. They could even
be in different containers, so ensuring coordination between all fsync
callers is not really an option.
One way to remedy this would be to track what file descriptor was used
to dirty the file, but that's rather cumbersome and would likely be
slow. However, there is a simpler way to improve the semantics here
without incurring too much overhead.
This set adds an errseq_t to struct address_space, and a corresponding
one is added to struct file. Writeback errors are recorded in the
mapping's errseq_t, and the one in struct file is used as the "since"
value.
This changes the semantics of the Linux fsync implementation such that
applications can now use it to determine whether there were any
writeback errors since fsync(fd) was last called (or since the file was
opened in the case of fsync having never been called).
Note that those writeback errors may have occurred when writing data
that was dirtied via an entirely different fd, but that's the case now
with the current mapping_set_error/filemap_check_error infrastructure.
This will at least prevent you from getting a false report of success.
The new behavior is still consistent with the POSIX spec, and is more
reliable for application developers. This patch just adds some basic
infrastructure for doing this, and ensures that the f_wb_err "cursor"
is properly set when a file is opened. Later patches will change the
existing code to use this new infrastructure for reporting errors at
fsync time.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
2017-07-06 07:02:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2011-03-22 16:30:52 -07:00
|
|
|
|
2022-11-01 10:53:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2011-03-22 16:30:52 -07:00
|
|
|
|
|
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
2011-03-22 16:30:52 -07:00
|
|
|
|
2022-11-01 10:53:22 -07:00
|
|
|
|
2011-03-22 16:30:52 -07:00
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
2022-05-01 07:35:31 -04:00
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
|
|
|
|
|
2011-03-22 16:30:52 -07:00
|
|
|
|
2022-11-01 10:53:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2011-03-22 16:30:52 -07:00
|
|
|
|
2022-11-01 10:53:22 -07:00
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
|
|
|
|
|
2011-03-22 16:30:52 -07:00
|
|
|
|
2023-10-06 11:46:27 -07:00
|
|
|
|
2020-06-03 16:01:54 -07:00
|
|
|
|
2021-09-02 14:53:18 -07:00
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
2015-06-24 16:57:24 -07:00
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
|
|
|
|
|
2022-11-01 10:53:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-09-02 14:53:18 -07:00
|
|
|
|
2022-05-01 07:35:31 -04:00
|
|
|
|
2022-11-01 10:53:22 -07:00
|
|
|
|
|
|
|
|
|
2011-03-22 16:30:52 -07:00
|
|
|
|
2022-11-01 10:53:22 -07:00
|
|
|
|
2011-03-22 16:30:52 -07:00
|
|
|
|
2020-12-08 08:56:28 -05:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
mm/filemap: use xas_try_split() in __filemap_add_folio()
Patch series "Minimize xa_node allocation during xarry split", v3.
When splitting a multi-index entry in XArray from order-n to order-m,
existing xas_split_alloc()+xas_split() approach requires 2^(n %
XA_CHUNK_SHIFT) xa_node allocations. But its callers,
__filemap_add_folio() and shmem_split_large_entry(), use at most 1
xa_node. To minimize xa_node allocation and remove the limitation of no
split from order-12 (or above) to order-0 (or anything between 0 and
5)[1], xas_try_split() was added[2], which allocates (n / XA_CHUNK_SHIFT -
m / XA_CHUNK_SHIFT) xa_node. It is used for non-uniform folio split, but
can be used by __filemap_add_folio() and shmem_split_large_entry().
xas_split_alloc() and xas_split() split an order-9 to order-0:
---------------------------------
| | | | | | | | |
| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| | | | | | | | |
---------------------------------
| | | |
------- --- --- -------
| | ... | |
V V V V
----------- ----------- ----------- -----------
| xa_node | | xa_node | ... | xa_node | | xa_node |
----------- ----------- ----------- -----------
xas_try_split() splits an order-9 to order-0:
---------------------------------
| | | | | | | | |
| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| | | | | | | | |
---------------------------------
|
|
V
-----------
| xa_node |
-----------
xas_try_split() is designed to be called iteratively with n = m + 1.
xas_try_split_mini_order() is added to minmize the number of calls to
xas_try_split() by telling the caller the next minimal order to split to
instead of n - 1. Splitting order-n to order-m when m= l * XA_CHUNK_SHIFT
does not require xa_node allocation and requires 1 xa_node when n=l *
XA_CHUNK_SHIFT and m = n - 1, so it is OK to use xas_try_split() with n >
m + 1 when no new xa_node is needed.
xfstests quick group test passed on xfs and tmpfs.
[1] https://lore.kernel.org/linux-mm/Z6YX3RznGLUD07Ao@casper.infradead.org/
[2] https://lore.kernel.org/linux-mm/20250226210032.2044041-1-ziy@nvidia.com/
This patch (of 2):
During __filemap_add_folio(), a shadow entry is covering n slots and a
folio covers m slots with m < n is to be added. Instead of splitting all
n slots, only the m slots covered by the folio need to be split and the
remaining n-m shadow entries can be retained with orders ranging from m to
n-1. This method only requires
(n/XA_CHUNK_SHIFT) - (m/XA_CHUNK_SHIFT)
new xa_nodes instead of
(n % XA_CHUNK_SHIFT) * ((n/XA_CHUNK_SHIFT) - (m/XA_CHUNK_SHIFT))
new xa_nodes, compared to the original xas_split_alloc() + xas_split()
one. For example, to insert an order-0 folio when an order-9 shadow entry
is present (assuming XA_CHUNK_SHIFT is 6), 1 xa_node is needed instead of
8.
xas_try_split_min_order() is introduced to reduce the number of calls to
xas_try_split() during split.
Link: https://lkml.kernel.org/r/20250314222113.711703-1-ziy@nvidia.com
Link: https://lkml.kernel.org/r/20250314222113.711703-2-ziy@nvidia.com
Signed-off-by: Zi Yan <ziy@nvidia.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kairui Song <kasong@tencent.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Mattew Wilcox <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Kirill A. Shuemov <kirill.shutemov@linux.intel.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yang Shi <yang@os.amperecomputing.com>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-03-14 18:21:12 -04:00
|
|
|
|
mm/filemap: optimize filemap folio adding
Instead of doing multiple tree walks, do one optimism range check with
lock hold, and exit if raced with another insertion. If a shadow exists,
check it with a new xas_get_order helper before releasing the lock to
avoid redundant tree walks for getting its order.
Drop the lock and do the allocation only if a split is needed.
In the best case, it only need to walk the tree once. If it needs to
alloc and split, 3 walks are issued (One for first ranged conflict check
and order retrieving, one for the second check after allocation, one for
the insert after split).
Testing with 4K pages, in an 8G cgroup, with 16G brd as block device:
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap --rw=randread --time_based \
--ramp_time=30s --runtime=5m --group_reporting
Before:
bw ( MiB/s): min= 1027, max= 3520, per=100.00%, avg=2445.02, stdev=18.90, samples=8691
iops : min=263001, max=901288, avg=625924.36, stdev=4837.28, samples=8691
After (+7.3%):
bw ( MiB/s): min= 493, max= 3947, per=100.00%, avg=2625.56, stdev=25.74, samples=8651
iops : min=126454, max=1010681, avg=672142.61, stdev=6590.48, samples=8651
Test result with THP (do a THP randread then switch to 4K page in hope it
issues a lot of splitting):
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap -thp=1 --readonly \
--rw=randread --time_based --ramp_time=30s --runtime=10m \
--group_reporting
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap \
--rw=randread --time_based --runtime=5s --group_reporting
Before:
bw ( KiB/s): min= 4141, max=14202, per=100.00%, avg=7935.51, stdev=96.85, samples=18976
iops : min= 1029, max= 3548, avg=1979.52, stdev=24.23, samples=18976·
READ: bw=4545B/s (4545B/s), 4545B/s-4545B/s (4545B/s-4545B/s), io=64.0KiB (65.5kB), run=14419-14419msec
After (+12.5%):
bw ( KiB/s): min= 4611, max=15370, per=100.00%, avg=8928.74, stdev=105.17, samples=19146
iops : min= 1151, max= 3842, avg=2231.27, stdev=26.29, samples=19146
READ: bw=4635B/s (4635B/s), 4635B/s-4635B/s (4635B/s-4635B/s), io=64.0KiB (65.5kB), run=14137-14137msec
The performance is better for both 4K (+7.5%) and THP (+12.5%) cached read.
Link: https://lkml.kernel.org/r/20240415171857.19244-5-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-16 01:18:56 +08:00
|
|
|
|
2024-04-16 01:18:54 +08:00
|
|
|
|
mm/filemap: use xas_try_split() in __filemap_add_folio()
Patch series "Minimize xa_node allocation during xarry split", v3.
When splitting a multi-index entry in XArray from order-n to order-m,
existing xas_split_alloc()+xas_split() approach requires 2^(n %
XA_CHUNK_SHIFT) xa_node allocations. But its callers,
__filemap_add_folio() and shmem_split_large_entry(), use at most 1
xa_node. To minimize xa_node allocation and remove the limitation of no
split from order-12 (or above) to order-0 (or anything between 0 and
5)[1], xas_try_split() was added[2], which allocates (n / XA_CHUNK_SHIFT -
m / XA_CHUNK_SHIFT) xa_node. It is used for non-uniform folio split, but
can be used by __filemap_add_folio() and shmem_split_large_entry().
xas_split_alloc() and xas_split() split an order-9 to order-0:
---------------------------------
| | | | | | | | |
| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| | | | | | | | |
---------------------------------
| | | |
------- --- --- -------
| | ... | |
V V V V
----------- ----------- ----------- -----------
| xa_node | | xa_node | ... | xa_node | | xa_node |
----------- ----------- ----------- -----------
xas_try_split() splits an order-9 to order-0:
---------------------------------
| | | | | | | | |
| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| | | | | | | | |
---------------------------------
|
|
V
-----------
| xa_node |
-----------
xas_try_split() is designed to be called iteratively with n = m + 1.
xas_try_split_mini_order() is added to minmize the number of calls to
xas_try_split() by telling the caller the next minimal order to split to
instead of n - 1. Splitting order-n to order-m when m= l * XA_CHUNK_SHIFT
does not require xa_node allocation and requires 1 xa_node when n=l *
XA_CHUNK_SHIFT and m = n - 1, so it is OK to use xas_try_split() with n >
m + 1 when no new xa_node is needed.
xfstests quick group test passed on xfs and tmpfs.
[1] https://lore.kernel.org/linux-mm/Z6YX3RznGLUD07Ao@casper.infradead.org/
[2] https://lore.kernel.org/linux-mm/20250226210032.2044041-1-ziy@nvidia.com/
This patch (of 2):
During __filemap_add_folio(), a shadow entry is covering n slots and a
folio covers m slots with m < n is to be added. Instead of splitting all
n slots, only the m slots covered by the folio need to be split and the
remaining n-m shadow entries can be retained with orders ranging from m to
n-1. This method only requires
(n/XA_CHUNK_SHIFT) - (m/XA_CHUNK_SHIFT)
new xa_nodes instead of
(n % XA_CHUNK_SHIFT) * ((n/XA_CHUNK_SHIFT) - (m/XA_CHUNK_SHIFT))
new xa_nodes, compared to the original xas_split_alloc() + xas_split()
one. For example, to insert an order-0 folio when an order-9 shadow entry
is present (assuming XA_CHUNK_SHIFT is 6), 1 xa_node is needed instead of
8.
xas_try_split_min_order() is introduced to reduce the number of calls to
xas_try_split() during split.
Link: https://lkml.kernel.org/r/20250314222113.711703-1-ziy@nvidia.com
Link: https://lkml.kernel.org/r/20250314222113.711703-2-ziy@nvidia.com
Signed-off-by: Zi Yan <ziy@nvidia.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kairui Song <kasong@tencent.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Mattew Wilcox <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Kirill A. Shuemov <kirill.shutemov@linux.intel.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yang Shi <yang@os.amperecomputing.com>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-03-14 18:21:12 -04:00
|
|
|
|
2008-07-25 19:45:30 -07:00
|
|
|
|
2020-12-08 08:56:28 -05:00
|
|
|
|
|
|
|
|
|
2024-08-22 15:50:10 +02:00
|
|
|
|
|
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
2008-07-25 19:45:30 -07:00
|
|
|
|
2023-09-26 12:20:17 -07:00
|
|
|
|
mm/filemap: optimize filemap folio adding
Instead of doing multiple tree walks, do one optimism range check with
lock hold, and exit if raced with another insertion. If a shadow exists,
check it with a new xas_get_order helper before releasing the lock to
avoid redundant tree walks for getting its order.
Drop the lock and do the allocation only if a split is needed.
In the best case, it only need to walk the tree once. If it needs to
alloc and split, 3 walks are issued (One for first ranged conflict check
and order retrieving, one for the second check after allocation, one for
the insert after split).
Testing with 4K pages, in an 8G cgroup, with 16G brd as block device:
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap --rw=randread --time_based \
--ramp_time=30s --runtime=5m --group_reporting
Before:
bw ( MiB/s): min= 1027, max= 3520, per=100.00%, avg=2445.02, stdev=18.90, samples=8691
iops : min=263001, max=901288, avg=625924.36, stdev=4837.28, samples=8691
After (+7.3%):
bw ( MiB/s): min= 493, max= 3947, per=100.00%, avg=2625.56, stdev=25.74, samples=8651
iops : min=126454, max=1010681, avg=672142.61, stdev=6590.48, samples=8651
Test result with THP (do a THP randread then switch to 4K page in hope it
issues a lot of splitting):
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap -thp=1 --readonly \
--rw=randread --time_based --ramp_time=30s --runtime=10m \
--group_reporting
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap \
--rw=randread --time_based --runtime=5s --group_reporting
Before:
bw ( KiB/s): min= 4141, max=14202, per=100.00%, avg=7935.51, stdev=96.85, samples=18976
iops : min= 1029, max= 3548, avg=1979.52, stdev=24.23, samples=18976·
READ: bw=4545B/s (4545B/s), 4545B/s-4545B/s (4545B/s-4545B/s), io=64.0KiB (65.5kB), run=14419-14419msec
After (+12.5%):
bw ( KiB/s): min= 4611, max=15370, per=100.00%, avg=8928.74, stdev=105.17, samples=19146
iops : min= 1151, max= 3842, avg=2231.27, stdev=26.29, samples=19146
READ: bw=4635B/s (4635B/s), 4635B/s-4635B/s (4635B/s-4635B/s), io=64.0KiB (65.5kB), run=14137-14137msec
The performance is better for both 4K (+7.5%) and THP (+12.5%) cached read.
Link: https://lkml.kernel.org/r/20240415171857.19244-5-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-16 01:18:56 +08:00
|
|
|
|
2023-09-26 12:20:17 -07:00
|
|
|
|
|
|
|
|
|
2020-10-15 20:05:20 -07:00
|
|
|
|
2019-09-05 14:03:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-10-15 20:05:20 -07:00
|
|
|
|
mm/filemap: optimize filemap folio adding
Instead of doing multiple tree walks, do one optimism range check with
lock hold, and exit if raced with another insertion. If a shadow exists,
check it with a new xas_get_order helper before releasing the lock to
avoid redundant tree walks for getting its order.
Drop the lock and do the allocation only if a split is needed.
In the best case, it only need to walk the tree once. If it needs to
alloc and split, 3 walks are issued (One for first ranged conflict check
and order retrieving, one for the second check after allocation, one for
the insert after split).
Testing with 4K pages, in an 8G cgroup, with 16G brd as block device:
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap --rw=randread --time_based \
--ramp_time=30s --runtime=5m --group_reporting
Before:
bw ( MiB/s): min= 1027, max= 3520, per=100.00%, avg=2445.02, stdev=18.90, samples=8691
iops : min=263001, max=901288, avg=625924.36, stdev=4837.28, samples=8691
After (+7.3%):
bw ( MiB/s): min= 493, max= 3947, per=100.00%, avg=2625.56, stdev=25.74, samples=8651
iops : min=126454, max=1010681, avg=672142.61, stdev=6590.48, samples=8651
Test result with THP (do a THP randread then switch to 4K page in hope it
issues a lot of splitting):
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap -thp=1 --readonly \
--rw=randread --time_based --ramp_time=30s --runtime=10m \
--group_reporting
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap \
--rw=randread --time_based --runtime=5s --group_reporting
Before:
bw ( KiB/s): min= 4141, max=14202, per=100.00%, avg=7935.51, stdev=96.85, samples=18976
iops : min= 1029, max= 3548, avg=1979.52, stdev=24.23, samples=18976·
READ: bw=4545B/s (4545B/s), 4545B/s-4545B/s (4545B/s-4545B/s), io=64.0KiB (65.5kB), run=14419-14419msec
After (+12.5%):
bw ( KiB/s): min= 4611, max=15370, per=100.00%, avg=8928.74, stdev=105.17, samples=19146
iops : min= 1151, max= 3842, avg=2231.27, stdev=26.29, samples=19146
READ: bw=4635B/s (4635B/s), 4635B/s-4635B/s (4635B/s-4635B/s), io=64.0KiB (65.5kB), run=14137-14137msec
The performance is better for both 4K (+7.5%) and THP (+12.5%) cached read.
Link: https://lkml.kernel.org/r/20240415171857.19244-5-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-16 01:18:56 +08:00
|
|
|
|
mm/filemap: use xas_try_split() in __filemap_add_folio()
Patch series "Minimize xa_node allocation during xarry split", v3.
When splitting a multi-index entry in XArray from order-n to order-m,
existing xas_split_alloc()+xas_split() approach requires 2^(n %
XA_CHUNK_SHIFT) xa_node allocations. But its callers,
__filemap_add_folio() and shmem_split_large_entry(), use at most 1
xa_node. To minimize xa_node allocation and remove the limitation of no
split from order-12 (or above) to order-0 (or anything between 0 and
5)[1], xas_try_split() was added[2], which allocates (n / XA_CHUNK_SHIFT -
m / XA_CHUNK_SHIFT) xa_node. It is used for non-uniform folio split, but
can be used by __filemap_add_folio() and shmem_split_large_entry().
xas_split_alloc() and xas_split() split an order-9 to order-0:
---------------------------------
| | | | | | | | |
| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| | | | | | | | |
---------------------------------
| | | |
------- --- --- -------
| | ... | |
V V V V
----------- ----------- ----------- -----------
| xa_node | | xa_node | ... | xa_node | | xa_node |
----------- ----------- ----------- -----------
xas_try_split() splits an order-9 to order-0:
---------------------------------
| | | | | | | | |
| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| | | | | | | | |
---------------------------------
|
|
V
-----------
| xa_node |
-----------
xas_try_split() is designed to be called iteratively with n = m + 1.
xas_try_split_mini_order() is added to minmize the number of calls to
xas_try_split() by telling the caller the next minimal order to split to
instead of n - 1. Splitting order-n to order-m when m= l * XA_CHUNK_SHIFT
does not require xa_node allocation and requires 1 xa_node when n=l *
XA_CHUNK_SHIFT and m = n - 1, so it is OK to use xas_try_split() with n >
m + 1 when no new xa_node is needed.
xfstests quick group test passed on xfs and tmpfs.
[1] https://lore.kernel.org/linux-mm/Z6YX3RznGLUD07Ao@casper.infradead.org/
[2] https://lore.kernel.org/linux-mm/20250226210032.2044041-1-ziy@nvidia.com/
This patch (of 2):
During __filemap_add_folio(), a shadow entry is covering n slots and a
folio covers m slots with m < n is to be added. Instead of splitting all
n slots, only the m slots covered by the folio need to be split and the
remaining n-m shadow entries can be retained with orders ranging from m to
n-1. This method only requires
(n/XA_CHUNK_SHIFT) - (m/XA_CHUNK_SHIFT)
new xa_nodes instead of
(n % XA_CHUNK_SHIFT) * ((n/XA_CHUNK_SHIFT) - (m/XA_CHUNK_SHIFT))
new xa_nodes, compared to the original xas_split_alloc() + xas_split()
one. For example, to insert an order-0 folio when an order-9 shadow entry
is present (assuming XA_CHUNK_SHIFT is 6), 1 xa_node is needed instead of
8.
xas_try_split_min_order() is introduced to reduce the number of calls to
xas_try_split() during split.
Link: https://lkml.kernel.org/r/20250314222113.711703-1-ziy@nvidia.com
Link: https://lkml.kernel.org/r/20250314222113.711703-2-ziy@nvidia.com
Signed-off-by: Zi Yan <ziy@nvidia.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kairui Song <kasong@tencent.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Mattew Wilcox <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Kirill A. Shuemov <kirill.shutemov@linux.intel.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yang Shi <yang@os.amperecomputing.com>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-03-14 18:21:12 -04:00
|
|
|
|
2020-10-15 20:05:20 -07:00
|
|
|
|
|
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
2020-10-15 20:05:20 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm/filemap: optimize filemap folio adding
Instead of doing multiple tree walks, do one optimism range check with
lock hold, and exit if raced with another insertion. If a shadow exists,
check it with a new xas_get_order helper before releasing the lock to
avoid redundant tree walks for getting its order.
Drop the lock and do the allocation only if a split is needed.
In the best case, it only need to walk the tree once. If it needs to
alloc and split, 3 walks are issued (One for first ranged conflict check
and order retrieving, one for the second check after allocation, one for
the insert after split).
Testing with 4K pages, in an 8G cgroup, with 16G brd as block device:
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap --rw=randread --time_based \
--ramp_time=30s --runtime=5m --group_reporting
Before:
bw ( MiB/s): min= 1027, max= 3520, per=100.00%, avg=2445.02, stdev=18.90, samples=8691
iops : min=263001, max=901288, avg=625924.36, stdev=4837.28, samples=8691
After (+7.3%):
bw ( MiB/s): min= 493, max= 3947, per=100.00%, avg=2625.56, stdev=25.74, samples=8651
iops : min=126454, max=1010681, avg=672142.61, stdev=6590.48, samples=8651
Test result with THP (do a THP randread then switch to 4K page in hope it
issues a lot of splitting):
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap -thp=1 --readonly \
--rw=randread --time_based --ramp_time=30s --runtime=10m \
--group_reporting
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap \
--rw=randread --time_based --runtime=5s --group_reporting
Before:
bw ( KiB/s): min= 4141, max=14202, per=100.00%, avg=7935.51, stdev=96.85, samples=18976
iops : min= 1029, max= 3548, avg=1979.52, stdev=24.23, samples=18976·
READ: bw=4545B/s (4545B/s), 4545B/s-4545B/s (4545B/s-4545B/s), io=64.0KiB (65.5kB), run=14419-14419msec
After (+12.5%):
bw ( KiB/s): min= 4611, max=15370, per=100.00%, avg=8928.74, stdev=105.17, samples=19146
iops : min= 1151, max= 3842, avg=2231.27, stdev=26.29, samples=19146
READ: bw=4635B/s (4635B/s), 4635B/s-4635B/s (4635B/s-4635B/s), io=64.0KiB (65.5kB), run=14137-14137msec
The performance is better for both 4K (+7.5%) and THP (+12.5%) cached read.
Link: https://lkml.kernel.org/r/20240415171857.19244-5-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-16 01:18:56 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-10-15 20:05:20 -07:00
|
|
|
|
mm/filemap: use xas_try_split() in __filemap_add_folio()
Patch series "Minimize xa_node allocation during xarry split", v3.
When splitting a multi-index entry in XArray from order-n to order-m,
existing xas_split_alloc()+xas_split() approach requires 2^(n %
XA_CHUNK_SHIFT) xa_node allocations. But its callers,
__filemap_add_folio() and shmem_split_large_entry(), use at most 1
xa_node. To minimize xa_node allocation and remove the limitation of no
split from order-12 (or above) to order-0 (or anything between 0 and
5)[1], xas_try_split() was added[2], which allocates (n / XA_CHUNK_SHIFT -
m / XA_CHUNK_SHIFT) xa_node. It is used for non-uniform folio split, but
can be used by __filemap_add_folio() and shmem_split_large_entry().
xas_split_alloc() and xas_split() split an order-9 to order-0:
---------------------------------
| | | | | | | | |
| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| | | | | | | | |
---------------------------------
| | | |
------- --- --- -------
| | ... | |
V V V V
----------- ----------- ----------- -----------
| xa_node | | xa_node | ... | xa_node | | xa_node |
----------- ----------- ----------- -----------
xas_try_split() splits an order-9 to order-0:
---------------------------------
| | | | | | | | |
| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| | | | | | | | |
---------------------------------
|
|
V
-----------
| xa_node |
-----------
xas_try_split() is designed to be called iteratively with n = m + 1.
xas_try_split_mini_order() is added to minmize the number of calls to
xas_try_split() by telling the caller the next minimal order to split to
instead of n - 1. Splitting order-n to order-m when m= l * XA_CHUNK_SHIFT
does not require xa_node allocation and requires 1 xa_node when n=l *
XA_CHUNK_SHIFT and m = n - 1, so it is OK to use xas_try_split() with n >
m + 1 when no new xa_node is needed.
xfstests quick group test passed on xfs and tmpfs.
[1] https://lore.kernel.org/linux-mm/Z6YX3RznGLUD07Ao@casper.infradead.org/
[2] https://lore.kernel.org/linux-mm/20250226210032.2044041-1-ziy@nvidia.com/
This patch (of 2):
During __filemap_add_folio(), a shadow entry is covering n slots and a
folio covers m slots with m < n is to be added. Instead of splitting all
n slots, only the m slots covered by the folio need to be split and the
remaining n-m shadow entries can be retained with orders ranging from m to
n-1. This method only requires
(n/XA_CHUNK_SHIFT) - (m/XA_CHUNK_SHIFT)
new xa_nodes instead of
(n % XA_CHUNK_SHIFT) * ((n/XA_CHUNK_SHIFT) - (m/XA_CHUNK_SHIFT))
new xa_nodes, compared to the original xas_split_alloc() + xas_split()
one. For example, to insert an order-0 folio when an order-9 shadow entry
is present (assuming XA_CHUNK_SHIFT is 6), 1 xa_node is needed instead of
8.
xas_try_split_min_order() is introduced to reduce the number of calls to
xas_try_split() during split.
Link: https://lkml.kernel.org/r/20250314222113.711703-1-ziy@nvidia.com
Link: https://lkml.kernel.org/r/20250314222113.711703-2-ziy@nvidia.com
Signed-off-by: Zi Yan <ziy@nvidia.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kairui Song <kasong@tencent.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Mattew Wilcox <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Kirill A. Shuemov <kirill.shutemov@linux.intel.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yang Shi <yang@os.amperecomputing.com>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-03-14 18:21:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-09-05 14:03:12 -04:00
|
|
|
|
|
|
|
|
|
mm/filemap: use xas_try_split() in __filemap_add_folio()
Patch series "Minimize xa_node allocation during xarry split", v3.
When splitting a multi-index entry in XArray from order-n to order-m,
existing xas_split_alloc()+xas_split() approach requires 2^(n %
XA_CHUNK_SHIFT) xa_node allocations. But its callers,
__filemap_add_folio() and shmem_split_large_entry(), use at most 1
xa_node. To minimize xa_node allocation and remove the limitation of no
split from order-12 (or above) to order-0 (or anything between 0 and
5)[1], xas_try_split() was added[2], which allocates (n / XA_CHUNK_SHIFT -
m / XA_CHUNK_SHIFT) xa_node. It is used for non-uniform folio split, but
can be used by __filemap_add_folio() and shmem_split_large_entry().
xas_split_alloc() and xas_split() split an order-9 to order-0:
---------------------------------
| | | | | | | | |
| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| | | | | | | | |
---------------------------------
| | | |
------- --- --- -------
| | ... | |
V V V V
----------- ----------- ----------- -----------
| xa_node | | xa_node | ... | xa_node | | xa_node |
----------- ----------- ----------- -----------
xas_try_split() splits an order-9 to order-0:
---------------------------------
| | | | | | | | |
| 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 |
| | | | | | | | |
---------------------------------
|
|
V
-----------
| xa_node |
-----------
xas_try_split() is designed to be called iteratively with n = m + 1.
xas_try_split_mini_order() is added to minmize the number of calls to
xas_try_split() by telling the caller the next minimal order to split to
instead of n - 1. Splitting order-n to order-m when m= l * XA_CHUNK_SHIFT
does not require xa_node allocation and requires 1 xa_node when n=l *
XA_CHUNK_SHIFT and m = n - 1, so it is OK to use xas_try_split() with n >
m + 1 when no new xa_node is needed.
xfstests quick group test passed on xfs and tmpfs.
[1] https://lore.kernel.org/linux-mm/Z6YX3RznGLUD07Ao@casper.infradead.org/
[2] https://lore.kernel.org/linux-mm/20250226210032.2044041-1-ziy@nvidia.com/
This patch (of 2):
During __filemap_add_folio(), a shadow entry is covering n slots and a
folio covers m slots with m < n is to be added. Instead of splitting all
n slots, only the m slots covered by the folio need to be split and the
remaining n-m shadow entries can be retained with orders ranging from m to
n-1. This method only requires
(n/XA_CHUNK_SHIFT) - (m/XA_CHUNK_SHIFT)
new xa_nodes instead of
(n % XA_CHUNK_SHIFT) * ((n/XA_CHUNK_SHIFT) - (m/XA_CHUNK_SHIFT))
new xa_nodes, compared to the original xas_split_alloc() + xas_split()
one. For example, to insert an order-0 folio when an order-9 shadow entry
is present (assuming XA_CHUNK_SHIFT is 6), 1 xa_node is needed instead of
8.
xas_try_split_min_order() is introduced to reduce the number of calls to
xas_try_split() during split.
Link: https://lkml.kernel.org/r/20250314222113.711703-1-ziy@nvidia.com
Link: https://lkml.kernel.org/r/20250314222113.711703-2-ziy@nvidia.com
Signed-off-by: Zi Yan <ziy@nvidia.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kairui Song <kasong@tencent.com>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Mattew Wilcox <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Kirill A. Shuemov <kirill.shutemov@linux.intel.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Yang Shi <yang@os.amperecomputing.com>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-03-14 18:21:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm/filemap: optimize filemap folio adding
Instead of doing multiple tree walks, do one optimism range check with
lock hold, and exit if raced with another insertion. If a shadow exists,
check it with a new xas_get_order helper before releasing the lock to
avoid redundant tree walks for getting its order.
Drop the lock and do the allocation only if a split is needed.
In the best case, it only need to walk the tree once. If it needs to
alloc and split, 3 walks are issued (One for first ranged conflict check
and order retrieving, one for the second check after allocation, one for
the insert after split).
Testing with 4K pages, in an 8G cgroup, with 16G brd as block device:
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap --rw=randread --time_based \
--ramp_time=30s --runtime=5m --group_reporting
Before:
bw ( MiB/s): min= 1027, max= 3520, per=100.00%, avg=2445.02, stdev=18.90, samples=8691
iops : min=263001, max=901288, avg=625924.36, stdev=4837.28, samples=8691
After (+7.3%):
bw ( MiB/s): min= 493, max= 3947, per=100.00%, avg=2625.56, stdev=25.74, samples=8651
iops : min=126454, max=1010681, avg=672142.61, stdev=6590.48, samples=8651
Test result with THP (do a THP randread then switch to 4K page in hope it
issues a lot of splitting):
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap -thp=1 --readonly \
--rw=randread --time_based --ramp_time=30s --runtime=10m \
--group_reporting
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap \
--rw=randread --time_based --runtime=5s --group_reporting
Before:
bw ( KiB/s): min= 4141, max=14202, per=100.00%, avg=7935.51, stdev=96.85, samples=18976
iops : min= 1029, max= 3548, avg=1979.52, stdev=24.23, samples=18976·
READ: bw=4545B/s (4545B/s), 4545B/s-4545B/s (4545B/s-4545B/s), io=64.0KiB (65.5kB), run=14419-14419msec
After (+12.5%):
bw ( KiB/s): min= 4611, max=15370, per=100.00%, avg=8928.74, stdev=105.17, samples=19146
iops : min= 1151, max= 3842, avg=2231.27, stdev=26.29, samples=19146
READ: bw=4635B/s (4635B/s), 4635B/s-4635B/s (4635B/s-4635B/s), io=64.0KiB (65.5kB), run=14137-14137msec
The performance is better for both 4K (+7.5%) and THP (+12.5%) cached read.
Link: https://lkml.kernel.org/r/20240415171857.19244-5-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-16 01:18:56 +08:00
|
|
|
|
2020-10-15 20:05:20 -07:00
|
|
|
|
|
|
|
|
|
mm/filemap: optimize filemap folio adding
Instead of doing multiple tree walks, do one optimism range check with
lock hold, and exit if raced with another insertion. If a shadow exists,
check it with a new xas_get_order helper before releasing the lock to
avoid redundant tree walks for getting its order.
Drop the lock and do the allocation only if a split is needed.
In the best case, it only need to walk the tree once. If it needs to
alloc and split, 3 walks are issued (One for first ranged conflict check
and order retrieving, one for the second check after allocation, one for
the insert after split).
Testing with 4K pages, in an 8G cgroup, with 16G brd as block device:
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap --rw=randread --time_based \
--ramp_time=30s --runtime=5m --group_reporting
Before:
bw ( MiB/s): min= 1027, max= 3520, per=100.00%, avg=2445.02, stdev=18.90, samples=8691
iops : min=263001, max=901288, avg=625924.36, stdev=4837.28, samples=8691
After (+7.3%):
bw ( MiB/s): min= 493, max= 3947, per=100.00%, avg=2625.56, stdev=25.74, samples=8651
iops : min=126454, max=1010681, avg=672142.61, stdev=6590.48, samples=8651
Test result with THP (do a THP randread then switch to 4K page in hope it
issues a lot of splitting):
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap -thp=1 --readonly \
--rw=randread --time_based --ramp_time=30s --runtime=10m \
--group_reporting
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap \
--rw=randread --time_based --runtime=5s --group_reporting
Before:
bw ( KiB/s): min= 4141, max=14202, per=100.00%, avg=7935.51, stdev=96.85, samples=18976
iops : min= 1029, max= 3548, avg=1979.52, stdev=24.23, samples=18976·
READ: bw=4545B/s (4545B/s), 4545B/s-4545B/s (4545B/s-4545B/s), io=64.0KiB (65.5kB), run=14419-14419msec
After (+12.5%):
bw ( KiB/s): min= 4611, max=15370, per=100.00%, avg=8928.74, stdev=105.17, samples=19146
iops : min= 1151, max= 3842, avg=2231.27, stdev=26.29, samples=19146
READ: bw=4635B/s (4635B/s), 4635B/s-4635B/s (4635B/s-4635B/s), io=64.0KiB (65.5kB), run=14137-14137msec
The performance is better for both 4K (+7.5%) and THP (+12.5%) cached read.
Link: https://lkml.kernel.org/r/20240415171857.19244-5-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-16 01:18:56 +08:00
|
|
|
|
|
|
|
|
|
2020-10-15 20:05:20 -07:00
|
|
|
|
|
|
|
|
|
2020-12-08 08:56:28 -05:00
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-09-05 14:03:12 -04:00
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
|
|
|
|
|
2019-09-05 14:03:12 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm/filemap: optimize filemap folio adding
Instead of doing multiple tree walks, do one optimism range check with
lock hold, and exit if raced with another insertion. If a shadow exists,
check it with a new xas_get_order helper before releasing the lock to
avoid redundant tree walks for getting its order.
Drop the lock and do the allocation only if a split is needed.
In the best case, it only need to walk the tree once. If it needs to
alloc and split, 3 walks are issued (One for first ranged conflict check
and order retrieving, one for the second check after allocation, one for
the insert after split).
Testing with 4K pages, in an 8G cgroup, with 16G brd as block device:
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap --rw=randread --time_based \
--ramp_time=30s --runtime=5m --group_reporting
Before:
bw ( MiB/s): min= 1027, max= 3520, per=100.00%, avg=2445.02, stdev=18.90, samples=8691
iops : min=263001, max=901288, avg=625924.36, stdev=4837.28, samples=8691
After (+7.3%):
bw ( MiB/s): min= 493, max= 3947, per=100.00%, avg=2625.56, stdev=25.74, samples=8651
iops : min=126454, max=1010681, avg=672142.61, stdev=6590.48, samples=8651
Test result with THP (do a THP randread then switch to 4K page in hope it
issues a lot of splitting):
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap -thp=1 --readonly \
--rw=randread --time_based --ramp_time=30s --runtime=10m \
--group_reporting
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap \
--rw=randread --time_based --runtime=5s --group_reporting
Before:
bw ( KiB/s): min= 4141, max=14202, per=100.00%, avg=7935.51, stdev=96.85, samples=18976
iops : min= 1029, max= 3548, avg=1979.52, stdev=24.23, samples=18976·
READ: bw=4545B/s (4545B/s), 4545B/s-4545B/s (4545B/s-4545B/s), io=64.0KiB (65.5kB), run=14419-14419msec
After (+12.5%):
bw ( KiB/s): min= 4611, max=15370, per=100.00%, avg=8928.74, stdev=105.17, samples=19146
iops : min= 1151, max= 3842, avg=2231.27, stdev=26.29, samples=19146
READ: bw=4635B/s (4635B/s), 4635B/s-4635B/s (4635B/s-4635B/s), io=64.0KiB (65.5kB), run=14137-14137msec
The performance is better for both 4K (+7.5%) and THP (+12.5%) cached read.
Link: https://lkml.kernel.org/r/20240415171857.19244-5-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-16 01:18:56 +08:00
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
|
|
|
|
|
mm/filemap: optimize filemap folio adding
Instead of doing multiple tree walks, do one optimism range check with
lock hold, and exit if raced with another insertion. If a shadow exists,
check it with a new xas_get_order helper before releasing the lock to
avoid redundant tree walks for getting its order.
Drop the lock and do the allocation only if a split is needed.
In the best case, it only need to walk the tree once. If it needs to
alloc and split, 3 walks are issued (One for first ranged conflict check
and order retrieving, one for the second check after allocation, one for
the insert after split).
Testing with 4K pages, in an 8G cgroup, with 16G brd as block device:
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap --rw=randread --time_based \
--ramp_time=30s --runtime=5m --group_reporting
Before:
bw ( MiB/s): min= 1027, max= 3520, per=100.00%, avg=2445.02, stdev=18.90, samples=8691
iops : min=263001, max=901288, avg=625924.36, stdev=4837.28, samples=8691
After (+7.3%):
bw ( MiB/s): min= 493, max= 3947, per=100.00%, avg=2625.56, stdev=25.74, samples=8651
iops : min=126454, max=1010681, avg=672142.61, stdev=6590.48, samples=8651
Test result with THP (do a THP randread then switch to 4K page in hope it
issues a lot of splitting):
echo 3 > /proc/sys/vm/drop_caches
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap -thp=1 --readonly \
--rw=randread --time_based --ramp_time=30s --runtime=10m \
--group_reporting
fio -name=cached --numjobs=16 --filename=/mnt/test.img \
--buffered=1 --ioengine=mmap \
--rw=randread --time_based --runtime=5s --group_reporting
Before:
bw ( KiB/s): min= 4141, max=14202, per=100.00%, avg=7935.51, stdev=96.85, samples=18976
iops : min= 1029, max= 3548, avg=1979.52, stdev=24.23, samples=18976·
READ: bw=4545B/s (4545B/s), 4545B/s-4545B/s (4545B/s-4545B/s), io=64.0KiB (65.5kB), run=14419-14419msec
After (+12.5%):
bw ( KiB/s): min= 4611, max=15370, per=100.00%, avg=8928.74, stdev=105.17, samples=19146
iops : min= 1151, max= 3842, avg=2231.27, stdev=26.29, samples=19146
READ: bw=4635B/s (4635B/s), 4635B/s-4635B/s (4635B/s-4635B/s), io=64.0KiB (65.5kB), run=14137-14137msec
The performance is better for both 4K (+7.5%) and THP (+12.5%) cached read.
Link: https://lkml.kernel.org/r/20240415171857.19244-5-ryncsn@gmail.com
Signed-off-by: Kairui Song <kasong@tencent.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-04-16 01:18:56 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
2019-09-05 14:03:12 -04:00
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
2015-06-24 16:57:24 -07:00
|
|
|
|
2021-07-23 09:29:46 -04:00
|
|
|
|
2013-09-12 15:13:59 -07:00
|
|
|
|
2017-11-17 10:01:45 -05:00
|
|
|
|
2020-12-08 08:56:28 -05:00
|
|
|
|
2025-05-14 19:15:07 +01:00
|
|
|
|
2019-09-05 14:03:12 -04:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-12-08 08:56:28 -05:00
|
|
|
|
2014-04-03 14:47:51 -07:00
|
|
|
|
2020-12-08 08:56:28 -05:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2014-04-03 14:47:51 -07:00
|
|
|
|
2008-10-18 20:26:32 -07:00
|
|
|
|
mm/filemap: add AS_KERNEL_FILE
Patch series "introduce kernel file mapped folios", v4.
Btrfs currently tracks its metadata pages in the page cache, using a fake
inode (fs_info->btree_inode) with offsets corresponding to where the
metadata is stored in the filesystem's full logical address space.
A consequence of this is that when btrfs uses filemap_add_folio(), this
usage is charged to the cgroup of whichever task happens to be running at
the time. These folios don't belong to any particular user cgroup, so I
don't think it makes much sense for them to be charged in that way. Some
negative consequences as a result:
- A task can be holding some important btrfs locks, then need to lookup
some metadata and go into reclaim, extending the duration it holds
that lock for, and unfairly pushing its own reclaim pain onto other
cgroups.
- If that cgroup goes into reclaim, it might reclaim these folios a
different non-reclaiming cgroup might need soon. This is naturally
offset by LRU reclaim, but still.
We have two options for how to manage such file pages:
1. charge them to the root cgroup.
2. don't charge them to any cgroup at all.
2. breaks the invariant that every mapped page has a cgroup. This is
workable, but unnecessarily risky. Therefore, go with 1.
A very similar proposal to use the root cgroup was previously made by Qu,
where he eventually proposed the idea of setting it per address_space.
This makes good sense for the btrfs use case, as the behavior should apply
to all use of the address_space, not select allocations. I.e., if someone
adds another filemap_add_folio() call using btrfs's btree_inode, we would
almost certainly want to account that to the root cgroup as well.
This patch (of 3):
Add the flag AS_KERNEL_FILE to the address_space to indicate that this
mapping's memory is exempt from the usual memcg accounting.
[boris@bur.io: fix CONFIG_MEMCG build for AS_KERNEL_FILE]
Link: https://lkml.kernel.org/r/6de59ddeec81b5c294d337c001ba0061631d4ec6.1755816635.git.boris@bur.io
Link: https://lore.kernel.org/linux-mm/b5fef5372ae454a7b6da4f2f75c427aeab6a07d6.1727498749.git.wqu@suse.com/
Link: https://lkml.kernel.org/r/f09c4e2c90351d4cb30a1969f7a863b9238bd291.1755812945.git.boris@bur.io
Signed-off-by: Boris Burkov <boris@bur.io>
Suggested-by: Qu Wenruo <wqu@suse.com>
Suggested-by: Shakeel Butt <shakeel.butt@linux.dev>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-08-21 14:55:35 -07:00
|
|
|
|
|
|
|
|
|
2008-10-18 20:26:32 -07:00
|
|
|
|
mm/filemap: add AS_KERNEL_FILE
Patch series "introduce kernel file mapped folios", v4.
Btrfs currently tracks its metadata pages in the page cache, using a fake
inode (fs_info->btree_inode) with offsets corresponding to where the
metadata is stored in the filesystem's full logical address space.
A consequence of this is that when btrfs uses filemap_add_folio(), this
usage is charged to the cgroup of whichever task happens to be running at
the time. These folios don't belong to any particular user cgroup, so I
don't think it makes much sense for them to be charged in that way. Some
negative consequences as a result:
- A task can be holding some important btrfs locks, then need to lookup
some metadata and go into reclaim, extending the duration it holds
that lock for, and unfairly pushing its own reclaim pain onto other
cgroups.
- If that cgroup goes into reclaim, it might reclaim these folios a
different non-reclaiming cgroup might need soon. This is naturally
offset by LRU reclaim, but still.
We have two options for how to manage such file pages:
1. charge them to the root cgroup.
2. don't charge them to any cgroup at all.
2. breaks the invariant that every mapped page has a cgroup. This is
workable, but unnecessarily risky. Therefore, go with 1.
A very similar proposal to use the root cgroup was previously made by Qu,
where he eventually proposed the idea of setting it per address_space.
This makes good sense for the btrfs use case, as the behavior should apply
to all use of the address_space, not select allocations. I.e., if someone
adds another filemap_add_folio() call using btrfs's btree_inode, we would
almost certainly want to account that to the root cgroup as well.
This patch (of 3):
Add the flag AS_KERNEL_FILE to the address_space to indicate that this
mapping's memory is exempt from the usual memcg accounting.
[boris@bur.io: fix CONFIG_MEMCG build for AS_KERNEL_FILE]
Link: https://lkml.kernel.org/r/6de59ddeec81b5c294d337c001ba0061631d4ec6.1755816635.git.boris@bur.io
Link: https://lore.kernel.org/linux-mm/b5fef5372ae454a7b6da4f2f75c427aeab6a07d6.1727498749.git.wqu@suse.com/
Link: https://lkml.kernel.org/r/f09c4e2c90351d4cb30a1969f7a863b9238bd291.1755812945.git.boris@bur.io
Signed-off-by: Boris Burkov <boris@bur.io>
Suggested-by: Qu Wenruo <wqu@suse.com>
Suggested-by: Shakeel Butt <shakeel.butt@linux.dev>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-08-21 14:55:35 -07:00
|
|
|
|
|
|
|
|
|
2024-04-16 01:18:54 +08:00
|
|
|
|
mm/filemap: add AS_KERNEL_FILE
Patch series "introduce kernel file mapped folios", v4.
Btrfs currently tracks its metadata pages in the page cache, using a fake
inode (fs_info->btree_inode) with offsets corresponding to where the
metadata is stored in the filesystem's full logical address space.
A consequence of this is that when btrfs uses filemap_add_folio(), this
usage is charged to the cgroup of whichever task happens to be running at
the time. These folios don't belong to any particular user cgroup, so I
don't think it makes much sense for them to be charged in that way. Some
negative consequences as a result:
- A task can be holding some important btrfs locks, then need to lookup
some metadata and go into reclaim, extending the duration it holds
that lock for, and unfairly pushing its own reclaim pain onto other
cgroups.
- If that cgroup goes into reclaim, it might reclaim these folios a
different non-reclaiming cgroup might need soon. This is naturally
offset by LRU reclaim, but still.
We have two options for how to manage such file pages:
1. charge them to the root cgroup.
2. don't charge them to any cgroup at all.
2. breaks the invariant that every mapped page has a cgroup. This is
workable, but unnecessarily risky. Therefore, go with 1.
A very similar proposal to use the root cgroup was previously made by Qu,
where he eventually proposed the idea of setting it per address_space.
This makes good sense for the btrfs use case, as the behavior should apply
to all use of the address_space, not select allocations. I.e., if someone
adds another filemap_add_folio() call using btrfs's btree_inode, we would
almost certainly want to account that to the root cgroup as well.
This patch (of 3):
Add the flag AS_KERNEL_FILE to the address_space to indicate that this
mapping's memory is exempt from the usual memcg accounting.
[boris@bur.io: fix CONFIG_MEMCG build for AS_KERNEL_FILE]
Link: https://lkml.kernel.org/r/6de59ddeec81b5c294d337c001ba0061631d4ec6.1755816635.git.boris@bur.io
Link: https://lore.kernel.org/linux-mm/b5fef5372ae454a7b6da4f2f75c427aeab6a07d6.1727498749.git.wqu@suse.com/
Link: https://lkml.kernel.org/r/f09c4e2c90351d4cb30a1969f7a863b9238bd291.1755812945.git.boris@bur.io
Signed-off-by: Boris Burkov <boris@bur.io>
Suggested-by: Qu Wenruo <wqu@suse.com>
Suggested-by: Shakeel Butt <shakeel.butt@linux.dev>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-08-21 14:55:35 -07:00
|
|
|
|
|
|
|
|
|
2024-04-16 01:18:54 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-08 08:56:28 -05:00
|
|
|
|
|
|
|
|
|
2024-04-16 01:18:54 +08:00
|
|
|
|
|
|
|
|
|
2020-12-08 08:56:28 -05:00
|
|
|
|
2024-04-16 01:18:54 +08:00
|
|
|
|
2014-04-03 14:47:51 -07:00
|
|
|
|
2020-12-08 08:56:28 -05:00
|
|
|
|
2014-04-03 14:47:51 -07:00
|
|
|
|
2020-12-08 08:56:28 -05:00
|
|
|
|
|
|
|
|
|
2016-05-20 16:56:25 -07:00
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:51 -07:00
|
|
|
|
2020-12-08 08:56:28 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-08-21 14:55:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:51 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2020-12-08 08:56:28 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2006-03-24 03:16:04 -08:00
|
|
|
|
2024-03-21 09:36:40 -07:00
|
|
|
|
2006-03-24 03:16:04 -08:00
|
|
|
|
2010-05-24 14:32:08 -07:00
|
|
|
|
2020-12-15 23:11:07 -05:00
|
|
|
|
2010-05-24 14:32:08 -07:00
|
|
|
|
2006-03-24 03:16:04 -08:00
|
|
|
|
cpuset: mm: reduce large amounts of memory barrier related damage v3
Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.
[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths. This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32. The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.
For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.
This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side. This is much cheaper on some architectures, including x86. The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.
While updating the nodemask, a check is made to see if a false failure
is a risk. If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.
In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%. The
actual results were
3.3.0-rc3 3.3.0-rc3
rc3-vanilla nobarrier-v2r1
Clients 1 UserTime 0.07 ( 0.00%) 0.08 (-14.19%)
Clients 2 UserTime 0.07 ( 0.00%) 0.07 ( 2.72%)
Clients 4 UserTime 0.08 ( 0.00%) 0.07 ( 3.29%)
Clients 1 SysTime 0.70 ( 0.00%) 0.65 ( 6.65%)
Clients 2 SysTime 0.85 ( 0.00%) 0.82 ( 3.65%)
Clients 4 SysTime 1.41 ( 0.00%) 1.41 ( 0.32%)
Clients 1 WallTime 0.77 ( 0.00%) 0.74 ( 4.19%)
Clients 2 WallTime 0.47 ( 0.00%) 0.45 ( 3.73%)
Clients 4 WallTime 0.38 ( 0.00%) 0.37 ( 1.58%)
Clients 1 Flt/sec/cpu 497620.28 ( 0.00%) 520294.53 ( 4.56%)
Clients 2 Flt/sec/cpu 414639.05 ( 0.00%) 429882.01 ( 3.68%)
Clients 4 Flt/sec/cpu 257959.16 ( 0.00%) 258761.48 ( 0.31%)
Clients 1 Flt/sec 495161.39 ( 0.00%) 517292.87 ( 4.47%)
Clients 2 Flt/sec 820325.95 ( 0.00%) 850289.77 ( 3.65%)
Clients 4 Flt/sec 1020068.93 ( 0.00%) 1022674.06 ( 0.26%)
MMTests Statistics: duration
Sys Time Running Test (seconds) 135.68 132.17
User+Sys Time Running Test (seconds) 164.2 160.13
Total Elapsed Time (seconds) 123.46 120.87
The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected). The
actual number of page faults is noticeably improved.
For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.
To test the actual bug the commit fixed I opened two terminals. The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data. In a second window, the nodemask of the
cpuset was continually randomised in a loop.
Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 16:34:11 -07:00
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:24 -07:00
|
|
|
|
cpuset: mm: reduce large amounts of memory barrier related damage v3
Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.
[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths. This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32. The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.
For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.
This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side. This is much cheaper on some architectures, including x86. The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.
While updating the nodemask, a check is made to see if a false failure
is a risk. If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.
In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%. The
actual results were
3.3.0-rc3 3.3.0-rc3
rc3-vanilla nobarrier-v2r1
Clients 1 UserTime 0.07 ( 0.00%) 0.08 (-14.19%)
Clients 2 UserTime 0.07 ( 0.00%) 0.07 ( 2.72%)
Clients 4 UserTime 0.08 ( 0.00%) 0.07 ( 3.29%)
Clients 1 SysTime 0.70 ( 0.00%) 0.65 ( 6.65%)
Clients 2 SysTime 0.85 ( 0.00%) 0.82 ( 3.65%)
Clients 4 SysTime 1.41 ( 0.00%) 1.41 ( 0.32%)
Clients 1 WallTime 0.77 ( 0.00%) 0.74 ( 4.19%)
Clients 2 WallTime 0.47 ( 0.00%) 0.45 ( 3.73%)
Clients 4 WallTime 0.38 ( 0.00%) 0.37 ( 1.58%)
Clients 1 Flt/sec/cpu 497620.28 ( 0.00%) 520294.53 ( 4.56%)
Clients 2 Flt/sec/cpu 414639.05 ( 0.00%) 429882.01 ( 3.68%)
Clients 4 Flt/sec/cpu 257959.16 ( 0.00%) 258761.48 ( 0.31%)
Clients 1 Flt/sec 495161.39 ( 0.00%) 517292.87 ( 4.47%)
Clients 2 Flt/sec 820325.95 ( 0.00%) 850289.77 ( 3.65%)
Clients 4 Flt/sec 1020068.93 ( 0.00%) 1022674.06 ( 0.26%)
MMTests Statistics: duration
Sys Time Running Test (seconds) 135.68 132.17
User+Sys Time Running Test (seconds) 164.2 160.13
Total Elapsed Time (seconds) 123.46 120.87
The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected). The
actual number of page faults is noticeably improved.
For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.
To test the actual bug the commit fixed I opened two terminals. The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data. In a second window, the nodemask of the
cpuset was continually randomised in a loop.
Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 16:34:11 -07:00
|
|
|
|
2024-05-31 13:53:50 -07:00
|
|
|
|
2020-12-15 23:11:07 -05:00
|
|
|
|
cpuset: mm: reduce large amounts of memory barrier related damage v3
Commit c0ff7453bb5c ("cpuset,mm: fix no node to alloc memory when
changing cpuset's mems") wins a super prize for the largest number of
memory barriers entered into fast paths for one commit.
[get|put]_mems_allowed is incredibly heavy with pairs of full memory
barriers inserted into a number of hot paths. This was detected while
investigating at large page allocator slowdown introduced some time
after 2.6.32. The largest portion of this overhead was shown by
oprofile to be at an mfence introduced by this commit into the page
allocator hot path.
For extra style points, the commit introduced the use of yield() in an
implementation of what looks like a spinning mutex.
This patch replaces the full memory barriers on both read and write
sides with a sequence counter with just read barriers on the fast path
side. This is much cheaper on some architectures, including x86. The
main bulk of the patch is the retry logic if the nodemask changes in a
manner that can cause a false failure.
While updating the nodemask, a check is made to see if a false failure
is a risk. If it is, the sequence number gets bumped and parallel
allocators will briefly stall while the nodemask update takes place.
In a page fault test microbenchmark, oprofile samples from
__alloc_pages_nodemask went from 4.53% of all samples to 1.15%. The
actual results were
3.3.0-rc3 3.3.0-rc3
rc3-vanilla nobarrier-v2r1
Clients 1 UserTime 0.07 ( 0.00%) 0.08 (-14.19%)
Clients 2 UserTime 0.07 ( 0.00%) 0.07 ( 2.72%)
Clients 4 UserTime 0.08 ( 0.00%) 0.07 ( 3.29%)
Clients 1 SysTime 0.70 ( 0.00%) 0.65 ( 6.65%)
Clients 2 SysTime 0.85 ( 0.00%) 0.82 ( 3.65%)
Clients 4 SysTime 1.41 ( 0.00%) 1.41 ( 0.32%)
Clients 1 WallTime 0.77 ( 0.00%) 0.74 ( 4.19%)
Clients 2 WallTime 0.47 ( 0.00%) 0.45 ( 3.73%)
Clients 4 WallTime 0.38 ( 0.00%) 0.37 ( 1.58%)
Clients 1 Flt/sec/cpu 497620.28 ( 0.00%) 520294.53 ( 4.56%)
Clients 2 Flt/sec/cpu 414639.05 ( 0.00%) 429882.01 ( 3.68%)
Clients 4 Flt/sec/cpu 257959.16 ( 0.00%) 258761.48 ( 0.31%)
Clients 1 Flt/sec 495161.39 ( 0.00%) 517292.87 ( 4.47%)
Clients 2 Flt/sec 820325.95 ( 0.00%) 850289.77 ( 3.65%)
Clients 4 Flt/sec 1020068.93 ( 0.00%) 1022674.06 ( 0.26%)
MMTests Statistics: duration
Sys Time Running Test (seconds) 135.68 132.17
User+Sys Time Running Test (seconds) 164.2 160.13
Total Elapsed Time (seconds) 123.46 120.87
The overall improvement is small but the System CPU time is much
improved and roughly in correlation to what oprofile reported (these
performance figures are without profiling so skew is expected). The
actual number of page faults is noticeably improved.
For benchmarks like kernel builds, the overall benefit is marginal but
the system CPU time is slightly reduced.
To test the actual bug the commit fixed I opened two terminals. The
first ran within a cpuset and continually ran a small program that
faulted 100M of anonymous data. In a second window, the nodemask of the
cpuset was continually randomised in a loop.
Without the commit, the program would fail every so often (usually
within 10 seconds) and obviously with the commit everything worked fine.
With this patch applied, it also worked fine so the fix should be
functionally equivalent.
Signed-off-by: Mel Gorman <mgorman@suse.de>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-03-21 16:34:11 -07:00
|
|
|
|
2020-12-15 23:11:07 -05:00
|
|
|
|
2006-03-24 03:16:04 -08:00
|
|
|
|
2024-03-21 09:36:40 -07:00
|
|
|
|
2006-03-24 03:16:04 -08:00
|
|
|
|
2024-03-21 09:36:40 -07:00
|
|
|
|
2006-03-24 03:16:04 -08:00
|
|
|
|
|
|
|
|
|
2021-05-24 13:02:30 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
|
|
|
|
|
2021-01-16 11:22:14 -05:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2021-01-16 11:22:14 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-01-16 11:22:14 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2025-01-11 15:07:37 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2021-01-16 11:22:14 -05:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
|
|
|
|
|
2025-01-11 15:07:37 +08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
2020-09-20 10:38:47 -07:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-09-20 10:38:47 -07:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-06-20 12:06:13 +02:00
|
|
|
|
2011-05-24 17:11:29 -07:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-08-03 13:01:22 -07:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
Minor page waitqueue cleanups
Tim Chen and Kan Liang have been battling a customer load that shows
extremely long page wakeup lists. The cause seems to be constant NUMA
migration of a hot page that is shared across a lot of threads, but the
actual root cause for the exact behavior has not been found.
Tim has a patch that batches the wait list traversal at wakeup time, so
that we at least don't get long uninterruptible cases where we traverse
and wake up thousands of processes and get nasty latency spikes. That
is likely 4.14 material, but we're still discussing the page waitqueue
specific parts of it.
In the meantime, I've tried to look at making the page wait queues less
expensive, and failing miserably. If you have thousands of threads
waiting for the same page, it will be painful. We'll need to try to
figure out the NUMA balancing issue some day, in addition to avoiding
the excessive spinlock hold times.
That said, having tried to rewrite the page wait queues, I can at least
fix up some of the braindamage in the current situation. In particular:
(a) we don't want to continue walking the page wait list if the bit
we're waiting for already got set again (which seems to be one of
the patterns of the bad load). That makes no progress and just
causes pointless cache pollution chasing the pointers.
(b) we don't want to put the non-locking waiters always on the front of
the queue, and the locking waiters always on the back. Not only is
that unfair, it means that we wake up thousands of reading threads
that will just end up being blocked by the writer later anyway.
Also add a comment about the layout of 'struct wait_page_key' - there is
an external user of it in the cachefiles code that means that it has to
match the layout of 'struct wait_bit_key' in the two first members. It
so happens to match, because 'struct page *' and 'unsigned long *' end
up having the same values simply because the page flags are the first
member in struct page.
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Christopher Lameter <cl@linux.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-08-27 13:55:12 -07:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
2025-08-05 18:22:51 +01:00
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
2025-08-05 18:22:51 +01:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
2011-05-24 17:11:29 -07:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-07-23 12:33:41 -07:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
2011-05-24 17:11:29 -07:00
|
|
|
|
|
|
|
|
|
2021-01-15 17:14:48 -05:00
|
|
|
|
2014-09-25 13:55:19 +10:00
|
|
|
|
2021-01-16 11:22:14 -05:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
|
|
|
|
|
2014-09-25 13:55:19 +10:00
|
|
|
|
2021-01-16 11:22:14 -05:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-10 04:58:28 +01:00
|
|
|
|
2017-08-25 09:13:55 -07:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2022-03-24 18:09:49 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2022-03-24 18:09:49 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2022-03-24 18:09:49 -07:00
|
|
|
|
2021-01-15 17:14:48 -05:00
|
|
|
|
2022-03-24 18:09:49 -07:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
|
|
|
|
|
2017-02-22 15:44:41 -08:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-01 19:38:25 -05:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
|
|
|
|
|
2021-08-16 23:36:31 -04:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
|
|
|
|
|
2021-08-16 23:36:31 -04:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-08-05 18:22:51 +01:00
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
2025-08-05 18:22:51 +01:00
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
|
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
|
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2021-01-16 11:22:14 -05:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2017-06-20 12:06:13 +02:00
|
|
|
|
2018-10-26 15:06:08 -07:00
|
|
|
|
psi: pressure stall information for CPU, memory, and IO
When systems are overcommitted and resources become contended, it's hard
to tell exactly the impact this has on workload productivity, or how close
the system is to lockups and OOM kills. In particular, when machines work
multiple jobs concurrently, the impact of overcommit in terms of latency
and throughput on the individual job can be enormous.
In order to maximize hardware utilization without sacrificing individual
job health or risk complete machine lockups, this patch implements a way
to quantify resource pressure in the system.
A kernel built with CONFIG_PSI=y creates files in /proc/pressure/ that
expose the percentage of time the system is stalled on CPU, memory, or IO,
respectively. Stall states are aggregate versions of the per-task delay
accounting delays:
cpu: some tasks are runnable but not executing on a CPU
memory: tasks are reclaiming, or waiting for swapin or thrashing cache
io: tasks are waiting for io completions
These percentages of walltime can be thought of as pressure percentages,
and they give a general sense of system health and productivity loss
incurred by resource overcommit. They can also indicate when the system
is approaching lockup scenarios and OOMs.
To do this, psi keeps track of the task states associated with each CPU
and samples the time they spend in stall states. Every 2 seconds, the
samples are averaged across CPUs - weighted by the CPUs' non-idle time to
eliminate artifacts from unused CPUs - and translated into percentages of
walltime. A running average of those percentages is maintained over 10s,
1m, and 5m periods (similar to the loadaverage).
[hannes@cmpxchg.org: doc fixlet, per Randy]
Link: http://lkml.kernel.org/r/20180828205625.GA14030@cmpxchg.org
[hannes@cmpxchg.org: code optimization]
Link: http://lkml.kernel.org/r/20180907175015.GA8479@cmpxchg.org
[hannes@cmpxchg.org: rename psi_clock() to psi_update_work(), per Peter]
Link: http://lkml.kernel.org/r/20180907145404.GB11088@cmpxchg.org
[hannes@cmpxchg.org: fix build]
Link: http://lkml.kernel.org/r/20180913014222.GA2370@cmpxchg.org
Link: http://lkml.kernel.org/r/20180828172258.3185-9-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Daniel Drake <drake@endlessm.com>
Tested-by: Suren Baghdasaryan <surenb@google.com>
Cc: Christopher Lameter <cl@linux.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Johannes Weiner <jweiner@fb.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Enderborg <peter.enderborg@sony.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vinayak Menon <vinmenon@codeaurora.org>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-10-26 15:06:27 -07:00
|
|
|
|
2022-08-15 07:11:35 +00:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
psi: pressure stall information for CPU, memory, and IO
When systems are overcommitted and resources become contended, it's hard
to tell exactly the impact this has on workload productivity, or how close
the system is to lockups and OOM kills. In particular, when machines work
multiple jobs concurrently, the impact of overcommit in terms of latency
and throughput on the individual job can be enormous.
In order to maximize hardware utilization without sacrificing individual
job health or risk complete machine lockups, this patch implements a way
to quantify resource pressure in the system.
A kernel built with CONFIG_PSI=y creates files in /proc/pressure/ that
expose the percentage of time the system is stalled on CPU, memory, or IO,
respectively. Stall states are aggregate versions of the per-task delay
accounting delays:
cpu: some tasks are runnable but not executing on a CPU
memory: tasks are reclaiming, or waiting for swapin or thrashing cache
io: tasks are waiting for io completions
These percentages of walltime can be thought of as pressure percentages,
and they give a general sense of system health and productivity loss
incurred by resource overcommit. They can also indicate when the system
is approaching lockup scenarios and OOMs.
To do this, psi keeps track of the task states associated with each CPU
and samples the time they spend in stall states. Every 2 seconds, the
samples are averaged across CPUs - weighted by the CPUs' non-idle time to
eliminate artifacts from unused CPUs - and translated into percentages of
walltime. A running average of those percentages is maintained over 10s,
1m, and 5m periods (similar to the loadaverage).
[hannes@cmpxchg.org: doc fixlet, per Randy]
Link: http://lkml.kernel.org/r/20180828205625.GA14030@cmpxchg.org
[hannes@cmpxchg.org: code optimization]
Link: http://lkml.kernel.org/r/20180907175015.GA8479@cmpxchg.org
[hannes@cmpxchg.org: rename psi_clock() to psi_update_work(), per Peter]
Link: http://lkml.kernel.org/r/20180907145404.GB11088@cmpxchg.org
[hannes@cmpxchg.org: fix build]
Link: http://lkml.kernel.org/r/20180913014222.GA2370@cmpxchg.org
Link: http://lkml.kernel.org/r/20180828172258.3185-9-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Daniel Drake <drake@endlessm.com>
Tested-by: Suren Baghdasaryan <surenb@google.com>
Cc: Christopher Lameter <cl@linux.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Johannes Weiner <jweiner@fb.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Enderborg <peter.enderborg@sony.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vinayak Menon <vinmenon@codeaurora.org>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-10-26 15:06:27 -07:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
2022-08-15 07:11:35 +00:00
|
|
|
|
psi: pressure stall information for CPU, memory, and IO
When systems are overcommitted and resources become contended, it's hard
to tell exactly the impact this has on workload productivity, or how close
the system is to lockups and OOM kills. In particular, when machines work
multiple jobs concurrently, the impact of overcommit in terms of latency
and throughput on the individual job can be enormous.
In order to maximize hardware utilization without sacrificing individual
job health or risk complete machine lockups, this patch implements a way
to quantify resource pressure in the system.
A kernel built with CONFIG_PSI=y creates files in /proc/pressure/ that
expose the percentage of time the system is stalled on CPU, memory, or IO,
respectively. Stall states are aggregate versions of the per-task delay
accounting delays:
cpu: some tasks are runnable but not executing on a CPU
memory: tasks are reclaiming, or waiting for swapin or thrashing cache
io: tasks are waiting for io completions
These percentages of walltime can be thought of as pressure percentages,
and they give a general sense of system health and productivity loss
incurred by resource overcommit. They can also indicate when the system
is approaching lockup scenarios and OOMs.
To do this, psi keeps track of the task states associated with each CPU
and samples the time they spend in stall states. Every 2 seconds, the
samples are averaged across CPUs - weighted by the CPUs' non-idle time to
eliminate artifacts from unused CPUs - and translated into percentages of
walltime. A running average of those percentages is maintained over 10s,
1m, and 5m periods (similar to the loadaverage).
[hannes@cmpxchg.org: doc fixlet, per Randy]
Link: http://lkml.kernel.org/r/20180828205625.GA14030@cmpxchg.org
[hannes@cmpxchg.org: code optimization]
Link: http://lkml.kernel.org/r/20180907175015.GA8479@cmpxchg.org
[hannes@cmpxchg.org: rename psi_clock() to psi_update_work(), per Peter]
Link: http://lkml.kernel.org/r/20180907145404.GB11088@cmpxchg.org
[hannes@cmpxchg.org: fix build]
Link: http://lkml.kernel.org/r/20180913014222.GA2370@cmpxchg.org
Link: http://lkml.kernel.org/r/20180828172258.3185-9-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Daniel Drake <drake@endlessm.com>
Tested-by: Suren Baghdasaryan <surenb@google.com>
Cc: Christopher Lameter <cl@linux.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Johannes Weiner <jweiner@fb.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Enderborg <peter.enderborg@sony.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vinayak Menon <vinmenon@codeaurora.org>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-10-26 15:06:27 -07:00
|
|
|
|
2018-10-26 15:06:08 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
|
|
|
|
|
2021-01-16 11:22:14 -05:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
|
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
|
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
|
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
|
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
|
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
|
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-08-27 16:25:09 -07:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
|
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
|
|
|
|
|
psi: pressure stall information for CPU, memory, and IO
When systems are overcommitted and resources become contended, it's hard
to tell exactly the impact this has on workload productivity, or how close
the system is to lockups and OOM kills. In particular, when machines work
multiple jobs concurrently, the impact of overcommit in terms of latency
and throughput on the individual job can be enormous.
In order to maximize hardware utilization without sacrificing individual
job health or risk complete machine lockups, this patch implements a way
to quantify resource pressure in the system.
A kernel built with CONFIG_PSI=y creates files in /proc/pressure/ that
expose the percentage of time the system is stalled on CPU, memory, or IO,
respectively. Stall states are aggregate versions of the per-task delay
accounting delays:
cpu: some tasks are runnable but not executing on a CPU
memory: tasks are reclaiming, or waiting for swapin or thrashing cache
io: tasks are waiting for io completions
These percentages of walltime can be thought of as pressure percentages,
and they give a general sense of system health and productivity loss
incurred by resource overcommit. They can also indicate when the system
is approaching lockup scenarios and OOMs.
To do this, psi keeps track of the task states associated with each CPU
and samples the time they spend in stall states. Every 2 seconds, the
samples are averaged across CPUs - weighted by the CPUs' non-idle time to
eliminate artifacts from unused CPUs - and translated into percentages of
walltime. A running average of those percentages is maintained over 10s,
1m, and 5m periods (similar to the loadaverage).
[hannes@cmpxchg.org: doc fixlet, per Randy]
Link: http://lkml.kernel.org/r/20180828205625.GA14030@cmpxchg.org
[hannes@cmpxchg.org: code optimization]
Link: http://lkml.kernel.org/r/20180907175015.GA8479@cmpxchg.org
[hannes@cmpxchg.org: rename psi_clock() to psi_update_work(), per Peter]
Link: http://lkml.kernel.org/r/20180907145404.GB11088@cmpxchg.org
[hannes@cmpxchg.org: fix build]
Link: http://lkml.kernel.org/r/20180913014222.GA2370@cmpxchg.org
Link: http://lkml.kernel.org/r/20180828172258.3185-9-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Daniel Drake <drake@endlessm.com>
Tested-by: Suren Baghdasaryan <surenb@google.com>
Cc: Christopher Lameter <cl@linux.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Johannes Weiner <jweiner@fb.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Enderborg <peter.enderborg@sony.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vinayak Menon <vinmenon@codeaurora.org>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-10-26 15:06:27 -07:00
|
|
|
|
2022-08-15 07:11:35 +00:00
|
|
|
|
psi: pressure stall information for CPU, memory, and IO
When systems are overcommitted and resources become contended, it's hard
to tell exactly the impact this has on workload productivity, or how close
the system is to lockups and OOM kills. In particular, when machines work
multiple jobs concurrently, the impact of overcommit in terms of latency
and throughput on the individual job can be enormous.
In order to maximize hardware utilization without sacrificing individual
job health or risk complete machine lockups, this patch implements a way
to quantify resource pressure in the system.
A kernel built with CONFIG_PSI=y creates files in /proc/pressure/ that
expose the percentage of time the system is stalled on CPU, memory, or IO,
respectively. Stall states are aggregate versions of the per-task delay
accounting delays:
cpu: some tasks are runnable but not executing on a CPU
memory: tasks are reclaiming, or waiting for swapin or thrashing cache
io: tasks are waiting for io completions
These percentages of walltime can be thought of as pressure percentages,
and they give a general sense of system health and productivity loss
incurred by resource overcommit. They can also indicate when the system
is approaching lockup scenarios and OOMs.
To do this, psi keeps track of the task states associated with each CPU
and samples the time they spend in stall states. Every 2 seconds, the
samples are averaged across CPUs - weighted by the CPUs' non-idle time to
eliminate artifacts from unused CPUs - and translated into percentages of
walltime. A running average of those percentages is maintained over 10s,
1m, and 5m periods (similar to the loadaverage).
[hannes@cmpxchg.org: doc fixlet, per Randy]
Link: http://lkml.kernel.org/r/20180828205625.GA14030@cmpxchg.org
[hannes@cmpxchg.org: code optimization]
Link: http://lkml.kernel.org/r/20180907175015.GA8479@cmpxchg.org
[hannes@cmpxchg.org: rename psi_clock() to psi_update_work(), per Peter]
Link: http://lkml.kernel.org/r/20180907145404.GB11088@cmpxchg.org
[hannes@cmpxchg.org: fix build]
Link: http://lkml.kernel.org/r/20180913014222.GA2370@cmpxchg.org
Link: http://lkml.kernel.org/r/20180828172258.3185-9-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Daniel Drake <drake@endlessm.com>
Tested-by: Suren Baghdasaryan <surenb@google.com>
Cc: Christopher Lameter <cl@linux.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Johannes Weiner <jweiner@fb.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Enderborg <peter.enderborg@sony.com>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vinayak Menon <vinmenon@codeaurora.org>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-10-26 15:06:27 -07:00
|
|
|
|
|
|
|
|
|
2018-10-26 15:06:08 -07:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
mm: allow a controlled amount of unfairness in the page lock
Commit 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic") made
the page locking entirely fair, in that if a waiter came in while the
lock was held, the lock would be transferred to the lockers strictly in
order.
That was intended to finally get rid of the long-reported watchdog
failures that involved the page lock under extreme load, where a process
could end up waiting essentially forever, as other page lockers stole
the lock from under it.
It also improved some benchmarks, but it ended up causing huge
performance regressions on others, simply because fair lock behavior
doesn't end up giving out the lock as aggressively, causing better
worst-case latency, but potentially much worse average latencies and
throughput.
Instead of reverting that change entirely, this introduces a controlled
amount of unfairness, with a sysctl knob to tune it if somebody needs
to. But the default value should hopefully be good for any normal load,
allowing a few rounds of lock stealing, but enforcing the strict
ordering before the lock has been stolen too many times.
There is also a hint from Matthieu Baerts that the fair page coloring
may end up exposing an ABBA deadlock that is hidden by the usual
optimistic lock stealing, and while the unfairness doesn't fix the
fundamental issue (and I'm still looking at that), it avoids it in
practice.
The amount of unfairness can be modified by writing a new value to the
'sysctl_page_lock_unfairness' variable (default value of 5, exposed
through /proc/sys/vm/page_lock_unfairness), but that is hopefully
something we'd use mainly for debugging rather than being necessary for
any deep system tuning.
This whole issue has exposed just how critical the page lock can be, and
how contended it gets under certain locks. And the main contention
doesn't really seem to be anything related to IO (which was the origin
of this lock), but for things like just verifying that the page file
mapping is stable while faulting in the page into a page table.
Link: https://lore.kernel.org/linux-fsdevel/ed8442fd-6f54-dd84-cd4a-941e8b7ee603@MichaelLarabel.com/
Link: https://www.phoronix.com/scan.php?page=article&item=linux-50-59&num=1
Link: https://lore.kernel.org/linux-fsdevel/c560a38d-8313-51fb-b1ec-e904bd8836bc@tessares.net/
Reported-and-tested-by: Michael Larabel <Michael@michaellarabel.com>
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Chris Mason <clm@fb.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-09-13 14:05:35 -07:00
|
|
|
|
|
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2020-07-23 10:16:49 -07:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
|
|
|
|
|
2022-01-21 22:10:46 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-02-18 05:52:01 +00:00
|
|
|
|
2022-01-21 22:10:46 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-08 18:08:20 -07:00
|
|
|
|
2022-01-21 22:10:46 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-08 18:08:20 -07:00
|
|
|
|
|
|
|
|
|
2022-01-21 22:10:46 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-08-15 07:11:35 +00:00
|
|
|
|
2022-01-21 22:10:46 -08:00
|
|
|
|
2024-01-11 15:24:20 +00:00
|
|
|
|
2022-01-21 22:10:46 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-08-15 07:11:35 +00:00
|
|
|
|
2022-01-21 22:10:46 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-08 18:08:20 -07:00
|
|
|
|
2022-01-21 22:10:46 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-08-15 07:11:35 +00:00
|
|
|
|
2022-01-21 22:10:46 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
2016-12-25 13:00:30 +10:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
2014-09-25 13:55:19 +10:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
2014-09-25 13:55:19 +10:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
2021-08-16 23:36:31 -04:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:02 -08:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
2021-08-16 23:36:31 -04:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
2021-08-16 23:36:31 -04:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
2021-08-16 23:36:31 -04:00
|
|
|
|
2021-02-24 12:02:02 -08:00
|
|
|
|
2021-08-16 23:36:31 -04:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
2022-09-14 10:17:38 +08:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
2021-08-16 23:36:31 -04:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-12-07 15:44:35 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-12-07 15:44:35 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-12-07 15:44:35 -05:00
|
|
|
|
mm: optimize PageWaiters bit use for unlock_page()
In commit 62906027091f ("mm: add PageWaiters indicating tasks are
waiting for a page bit") Nick Piggin made our page locking no longer
unconditionally touch the hashed page waitqueue, which not only helps
performance in general, but is particularly helpful on NUMA machines
where the hashed wait queues can bounce around a lot.
However, the "clear lock bit atomically and then test the waiters bit"
sequence turns out to be much more expensive than it needs to be,
because you get a nasty stall when trying to access the same word that
just got updated atomically.
On architectures where locking is done with LL/SC, this would be trivial
to fix with a new primitive that clears one bit and tests another
atomically, but that ends up not working on x86, where the only atomic
operations that return the result end up being cmpxchg and xadd. The
atomic bit operations return the old value of the same bit we changed,
not the value of an unrelated bit.
On x86, we could put the lock bit in the high bit of the byte, and use
"xadd" with that bit (where the overflow ends up not touching other
bits), and look at the other bits of the result. However, an even
simpler model is to just use a regular atomic "and" to clear the lock
bit, and then the sign bit in eflags will indicate the resulting state
of the unrelated bit #7.
So by moving the PageWaiters bit up to bit #7, we can atomically clear
the lock bit and test the waiters bit on x86 too. And architectures
with LL/SC (which is all the usual RISC suspects), the particular bit
doesn't matter, so they are fine with this approach too.
This avoids the extra access to the same atomic word, and thus avoids
the costly stall at page unlock time.
The only downside is that the interface ends up being a bit odd and
specialized: clear a bit in a byte, and test the sign bit. Nick doesn't
love the resulting name of the new primitive, but I'd rather make the
name be descriptive and very clear about the limitation imposed by
trying to work across all relevant architectures than make it be some
generic thing that doesn't make the odd semantics explicit.
So this introduces the new architecture primitive
clear_bit_unlock_is_negative_byte();
and adds the trivial implementation for x86. We have a generic
non-optimized fallback (that just does a "clear_bit()"+"test_bit(7)"
combination) which can be overridden by any architecture that can do
better. According to Nick, Power has the same hickup x86 has, for
example, but some other architectures may not even care.
All these optimizations mean that my page locking stress-test (which is
just executing a lot of small short-lived shell scripts: "make test" in
the git source tree) no longer makes our page locking look horribly bad.
Before all these optimizations, just the unlock_page() costs were just
over 3% of all CPU overhead on "make test". After this, it's down to
0.66%, so just a quarter of the cost it used to be.
(The difference on NUMA is bigger, but there this micro-optimization is
likely less noticeable, since the big issue on NUMA was not the accesses
to 'struct page', but the waitqueue accesses that were already removed
by Nick's earlier commit).
Acked-by: Nick Piggin <npiggin@gmail.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Bob Peterson <rpeterso@redhat.com>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Andrew Lutomirski <luto@kernel.org>
Cc: Andreas Gruenbacher <agruenba@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-12-27 11:40:38 -08:00
|
|
|
|
2020-12-07 15:44:35 -05:00
|
|
|
|
|
|
|
|
|
2023-10-04 17:53:15 +01:00
|
|
|
|
2021-01-15 17:14:48 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-12-07 15:44:35 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-10-04 17:53:03 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-04 17:53:15 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-01-10 16:32:57 +00:00
|
|
|
|
2023-10-04 17:53:15 +01:00
|
|
|
|
2023-10-04 17:53:03 +01:00
|
|
|
|
2023-10-04 17:53:15 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-10-04 17:53:03 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
2021-04-22 22:58:32 -04:00
|
|
|
|
|
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
2021-04-22 22:58:32 -04:00
|
|
|
|
|
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
2021-04-22 22:58:32 -04:00
|
|
|
|
|
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
|
|
|
|
|
2021-04-22 22:58:32 -04:00
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
2021-01-15 17:14:48 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
2021-04-22 22:58:32 -04:00
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
|
|
|
|
|
2021-04-22 22:58:32 -04:00
|
|
|
|
|
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
2024-03-19 11:13:26 +00:00
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
2021-04-22 22:58:32 -04:00
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
|
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
2021-04-22 22:58:32 -04:00
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
|
|
|
|
|
2021-04-22 22:58:32 -04:00
|
|
|
|
|
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
2024-03-19 11:13:26 +00:00
|
|
|
|
|
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-22 22:58:32 -04:00
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
|
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-22 22:58:32 -04:00
|
|
|
|
2020-02-10 10:00:21 +00:00
|
|
|
|
2025-05-27 07:28:52 -06:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-05-27 07:28:56 -06:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-05-27 07:28:52 -06:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-12-20 08:47:47 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-09-06 12:48:14 -04:00
|
|
|
|
2024-12-20 08:47:47 -07:00
|
|
|
|
2025-05-27 07:28:56 -06:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-12-20 08:47:47 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-05-27 07:28:52 -06:00
|
|
|
|
2024-12-20 08:47:47 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-09-06 12:48:14 -04:00
|
|
|
|
2024-12-20 08:47:47 -07:00
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2025-09-06 12:48:15 -04:00
|
|
|
|
2021-03-03 15:21:55 -05:00
|
|
|
|
2023-10-04 17:53:16 +01:00
|
|
|
|
|
|
|
|
|
2025-09-06 12:48:15 -04:00
|
|
|
|
2023-10-04 17:53:16 +01:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2025-09-06 12:48:15 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-10-04 17:53:16 +01:00
|
|
|
|
|
|
|
|
|
2014-06-04 16:10:34 -07:00
|
|
|
|
2021-03-03 15:21:55 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-06-04 16:10:34 -07:00
|
|
|
|
2021-03-03 15:21:55 -05:00
|
|
|
|
|
|
|
|
|
2020-12-08 01:25:39 -05:00
|
|
|
|
2014-06-04 16:10:34 -07:00
|
|
|
|
2008-04-28 02:12:38 -07:00
|
|
|
|
2025-09-06 12:48:15 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm: fix VM_BUG_ON(PageTail) and BUG_ON(PageWriteback)
Twice now, when exercising ext4 looped on shmem huge pages, I have crashed
on the PF_ONLY_HEAD check inside PageWaiters(): ext4_finish_bio() calling
end_page_writeback() calling wake_up_page() on tail of a shmem huge page,
no longer an ext4 page at all.
The problem is that PageWriteback is not accompanied by a page reference
(as the NOTE at the end of test_clear_page_writeback() acknowledges): as
soon as TestClearPageWriteback has been done, that page could be removed
from page cache, freed, and reused for something else by the time that
wake_up_page() is reached.
https://lore.kernel.org/linux-mm/20200827122019.GC14765@casper.infradead.org/
Matthew Wilcox suggested avoiding or weakening the PageWaiters() tail
check; but I'm paranoid about even looking at an unreferenced struct page,
lest its memory might itself have already been reused or hotremoved (and
wake_up_page_bit() may modify that memory with its ClearPageWaiters()).
Then on crashing a second time, realized there's a stronger reason against
that approach. If my testing just occasionally crashes on that check,
when the page is reused for part of a compound page, wouldn't it be much
more common for the page to get reused as an order-0 page before reaching
wake_up_page()? And on rare occasions, might that reused page already be
marked PageWriteback by its new user, and already be waited upon? What
would that look like?
It would look like BUG_ON(PageWriteback) after wait_on_page_writeback()
in write_cache_pages() (though I have never seen that crash myself).
Matthew Wilcox explaining this to himself:
"page is allocated, added to page cache, dirtied, writeback starts,
--- thread A ---
filesystem calls end_page_writeback()
test_clear_page_writeback()
--- context switch to thread B ---
truncate_inode_pages_range() finds the page, it doesn't have writeback set,
we delete it from the page cache. Page gets reallocated, dirtied, writeback
starts again. Then we call write_cache_pages(), see
PageWriteback() set, call wait_on_page_writeback()
--- context switch back to thread A ---
wake_up_page(page, PG_writeback);
... thread B is woken, but because the wakeup was for the old use of
the page, PageWriteback is still set.
Devious"
And prior to 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic")
this would have been much less likely: before that, wake_page_function()'s
non-exclusive case would stop walking and not wake if it found Writeback
already set again; whereas now the non-exclusive case proceeds to wake.
I have not thought of a fix that does not add a little overhead: the
simplest fix is for end_page_writeback() to get_page() before calling
test_clear_page_writeback(), then put_page() after wake_up_page().
Was there a chance of missed wakeups before, since a page freed before
reaching wake_up_page() would have PageWaiters cleared? I think not,
because each waiter does hold a reference on the page. This bug comes
when the old use of the page, the one we do TestClearPageWriteback on,
had *no* waiters, so no additional page reference beyond the page cache
(and whoever racily freed it). The reuse of the page has a waiter
holding a reference, and its own PageWriteback set; but the belated
wake_up_page() has woken the reuse to hit that BUG_ON(PageWriteback).
Reported-by: syzbot+3622cea378100f45d59f@syzkaller.appspotmail.com
Reported-by: Qian Cai <cai@lca.pw>
Fixes: 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic")
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org # v5.8+
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-11-24 08:46:43 -08:00
|
|
|
|
2021-03-03 15:21:55 -05:00
|
|
|
|
mm: fix VM_BUG_ON(PageTail) and BUG_ON(PageWriteback)
Twice now, when exercising ext4 looped on shmem huge pages, I have crashed
on the PF_ONLY_HEAD check inside PageWaiters(): ext4_finish_bio() calling
end_page_writeback() calling wake_up_page() on tail of a shmem huge page,
no longer an ext4 page at all.
The problem is that PageWriteback is not accompanied by a page reference
(as the NOTE at the end of test_clear_page_writeback() acknowledges): as
soon as TestClearPageWriteback has been done, that page could be removed
from page cache, freed, and reused for something else by the time that
wake_up_page() is reached.
https://lore.kernel.org/linux-mm/20200827122019.GC14765@casper.infradead.org/
Matthew Wilcox suggested avoiding or weakening the PageWaiters() tail
check; but I'm paranoid about even looking at an unreferenced struct page,
lest its memory might itself have already been reused or hotremoved (and
wake_up_page_bit() may modify that memory with its ClearPageWaiters()).
Then on crashing a second time, realized there's a stronger reason against
that approach. If my testing just occasionally crashes on that check,
when the page is reused for part of a compound page, wouldn't it be much
more common for the page to get reused as an order-0 page before reaching
wake_up_page()? And on rare occasions, might that reused page already be
marked PageWriteback by its new user, and already be waited upon? What
would that look like?
It would look like BUG_ON(PageWriteback) after wait_on_page_writeback()
in write_cache_pages() (though I have never seen that crash myself).
Matthew Wilcox explaining this to himself:
"page is allocated, added to page cache, dirtied, writeback starts,
--- thread A ---
filesystem calls end_page_writeback()
test_clear_page_writeback()
--- context switch to thread B ---
truncate_inode_pages_range() finds the page, it doesn't have writeback set,
we delete it from the page cache. Page gets reallocated, dirtied, writeback
starts again. Then we call write_cache_pages(), see
PageWriteback() set, call wait_on_page_writeback()
--- context switch back to thread A ---
wake_up_page(page, PG_writeback);
... thread B is woken, but because the wakeup was for the old use of
the page, PageWriteback is still set.
Devious"
And prior to 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic")
this would have been much less likely: before that, wake_page_function()'s
non-exclusive case would stop walking and not wake if it found Writeback
already set again; whereas now the non-exclusive case proceeds to wake.
I have not thought of a fix that does not add a little overhead: the
simplest fix is for end_page_writeback() to get_page() before calling
test_clear_page_writeback(), then put_page() after wake_up_page().
Was there a chance of missed wakeups before, since a page freed before
reaching wake_up_page() would have PageWaiters cleared? I think not,
because each waiter does hold a reference on the page. This bug comes
when the old use of the page, the one we do TestClearPageWriteback on,
had *no* waiters, so no additional page reference beyond the page cache
(and whoever racily freed it). The reuse of the page has a waiter
holding a reference, and its own PageWriteback set; but the belated
wake_up_page() has woken the reuse to hit that BUG_ON(PageWriteback).
Reported-by: syzbot+3622cea378100f45d59f@syzkaller.appspotmail.com
Reported-by: Qian Cai <cai@lca.pw>
Fixes: 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic")
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org # v5.8+
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-11-24 08:46:43 -08:00
|
|
|
|
2021-03-03 15:21:55 -05:00
|
|
|
|
2023-10-04 17:53:17 +01:00
|
|
|
|
mm: fix VM_BUG_ON(PageTail) and BUG_ON(PageWriteback)
Twice now, when exercising ext4 looped on shmem huge pages, I have crashed
on the PF_ONLY_HEAD check inside PageWaiters(): ext4_finish_bio() calling
end_page_writeback() calling wake_up_page() on tail of a shmem huge page,
no longer an ext4 page at all.
The problem is that PageWriteback is not accompanied by a page reference
(as the NOTE at the end of test_clear_page_writeback() acknowledges): as
soon as TestClearPageWriteback has been done, that page could be removed
from page cache, freed, and reused for something else by the time that
wake_up_page() is reached.
https://lore.kernel.org/linux-mm/20200827122019.GC14765@casper.infradead.org/
Matthew Wilcox suggested avoiding or weakening the PageWaiters() tail
check; but I'm paranoid about even looking at an unreferenced struct page,
lest its memory might itself have already been reused or hotremoved (and
wake_up_page_bit() may modify that memory with its ClearPageWaiters()).
Then on crashing a second time, realized there's a stronger reason against
that approach. If my testing just occasionally crashes on that check,
when the page is reused for part of a compound page, wouldn't it be much
more common for the page to get reused as an order-0 page before reaching
wake_up_page()? And on rare occasions, might that reused page already be
marked PageWriteback by its new user, and already be waited upon? What
would that look like?
It would look like BUG_ON(PageWriteback) after wait_on_page_writeback()
in write_cache_pages() (though I have never seen that crash myself).
Matthew Wilcox explaining this to himself:
"page is allocated, added to page cache, dirtied, writeback starts,
--- thread A ---
filesystem calls end_page_writeback()
test_clear_page_writeback()
--- context switch to thread B ---
truncate_inode_pages_range() finds the page, it doesn't have writeback set,
we delete it from the page cache. Page gets reallocated, dirtied, writeback
starts again. Then we call write_cache_pages(), see
PageWriteback() set, call wait_on_page_writeback()
--- context switch back to thread A ---
wake_up_page(page, PG_writeback);
... thread B is woken, but because the wakeup was for the old use of
the page, PageWriteback is still set.
Devious"
And prior to 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic")
this would have been much less likely: before that, wake_page_function()'s
non-exclusive case would stop walking and not wake if it found Writeback
already set again; whereas now the non-exclusive case proceeds to wake.
I have not thought of a fix that does not add a little overhead: the
simplest fix is for end_page_writeback() to get_page() before calling
test_clear_page_writeback(), then put_page() after wake_up_page().
Was there a chance of missed wakeups before, since a page freed before
reaching wake_up_page() would have PageWaiters cleared? I think not,
because each waiter does hold a reference on the page. This bug comes
when the old use of the page, the one we do TestClearPageWriteback on,
had *no* waiters, so no additional page reference beyond the page cache
(and whoever racily freed it). The reuse of the page has a waiter
holding a reference, and its own PageWriteback set; but the belated
wake_up_page() has woken the reuse to hit that BUG_ON(PageWriteback).
Reported-by: syzbot+3622cea378100f45d59f@syzkaller.appspotmail.com
Reported-by: Qian Cai <cai@lca.pw>
Fixes: 2a9127fcf229 ("mm: rewrite wait_on_page_bit_common() logic")
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org # v5.8+
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-11-24 08:46:43 -08:00
|
|
|
|
2021-03-03 15:21:55 -05:00
|
|
|
|
2025-09-06 12:48:15 -04:00
|
|
|
|
2025-09-06 12:48:14 -04:00
|
|
|
|
2021-03-03 15:21:55 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-03-03 15:21:55 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2021-03-01 19:38:25 -05:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-03-01 19:38:25 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-03-01 19:38:25 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-12-08 00:07:31 -05:00
|
|
|
|
2007-12-06 11:18:49 -05:00
|
|
|
|
2021-03-04 12:02:54 -05:00
|
|
|
|
mm: put_and_wait_on_page_locked() while page is migrated
Waiting on a page migration entry has used wait_on_page_locked() all along
since 2006: but you cannot safely wait_on_page_locked() without holding a
reference to the page, and that extra reference is enough to make
migrate_page_move_mapping() fail with -EAGAIN, when a racing task faults
on the entry before migrate_page_move_mapping() gets there.
And that failure is retried nine times, amplifying the pain when trying to
migrate a popular page. With a single persistent faulter, migration
sometimes succeeds; with two or three concurrent faulters, success becomes
much less likely (and the more the page was mapped, the worse the overhead
of unmapping and remapping it on each try).
This is especially a problem for memory offlining, where the outer level
retries forever (or until terminated from userspace), because a heavy
refault workload can trigger an endless loop of migration failures.
wait_on_page_locked() is the wrong tool for the job.
David Herrmann (but was he the first?) noticed this issue in 2014:
https://marc.info/?l=linux-mm&m=140110465608116&w=2
Tim Chen started a thread in August 2017 which appears relevant:
https://marc.info/?l=linux-mm&m=150275941014915&w=2 where Kan Liang went
on to implicate __migration_entry_wait():
https://marc.info/?l=linux-mm&m=150300268411980&w=2 and the thread ended
up with the v4.14 commits: 2554db916586 ("sched/wait: Break up long wake
list walk") 11a19c7b099f ("sched/wait: Introduce wakeup boomark in
wake_up_page_bit")
Baoquan He reported "Memory hotplug softlock issue" 14 November 2018:
https://marc.info/?l=linux-mm&m=154217936431300&w=2
We have all assumed that it is essential to hold a page reference while
waiting on a page lock: partly to guarantee that there is still a struct
page when MEMORY_HOTREMOVE is configured, but also to protect against
reuse of the struct page going to someone who then holds the page locked
indefinitely, when the waiter can reasonably expect timely unlocking.
But in fact, so long as wait_on_page_bit_common() does the put_page(), and
is careful not to rely on struct page contents thereafter, there is no
need to hold a reference to the page while waiting on it. That does mean
that this case cannot go back through the loop: but that's fine for the
page migration case, and even if used more widely, is limited by the "Stop
walking if it's locked" optimization in wake_page_function().
Add interface put_and_wait_on_page_locked() to do this, using "behavior"
enum in place of "lock" arg to wait_on_page_bit_common() to implement it.
No interruptible or killable variant needed yet, but they might follow: I
have a vague notion that reporting -EINTR should take precedence over
return from wait_on_page_bit_common() without knowing the page state, so
arrange it accordingly - but that may be nothing but pedantic.
__migration_entry_wait() still has to take a brief reference to the page,
prior to calling put_and_wait_on_page_locked(): but now that it is dropped
before waiting, the chance of impeding page migration is very much
reduced. Should we perhaps disable preemption across this?
shrink_page_list()'s __ClearPageLocked(): that was a surprise! This
survived a lot of testing before that showed up. PageWaiters may have
been set by wait_on_page_bit_common(), and the reference dropped, just
before shrink_page_list() succeeds in freezing its last page reference: in
such a case, unlock_page() must be used. Follow the suggestion from
Michal Hocko, just revert a978d6f52106 ("mm: unlockless reclaim") now:
that optimization predates PageWaiters, and won't buy much these days; but
we can reinstate it for the !PageWaiters case if anyone notices.
It does raise the question: should vmscan.c's is_page_cache_freeable() and
__remove_mapping() now treat a PageWaiters page as if an extra reference
were held? Perhaps, but I don't think it matters much, since
shrink_page_list() already had to win its trylock_page(), so waiters are
not very common there: I noticed no difference when trying the bigger
change, and it's surely not needed while put_and_wait_on_page_locked() is
only used for page migration.
[willy@infradead.org: add put_and_wait_on_page_locked() kerneldoc]
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1811261121330.1116@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Baoquan He <bhe@redhat.com>
Tested-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: David Herrmann <dh.herrmann@gmail.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Nick Piggin <npiggin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2018-12-28 00:36:14 -08:00
|
|
|
|
2007-12-06 11:18:49 -05:00
|
|
|
|
2020-12-08 00:07:31 -05:00
|
|
|
|
2007-12-06 11:18:49 -05:00
|
|
|
|
2020-12-30 17:58:40 -05:00
|
|
|
|
2020-05-22 09:12:09 -06:00
|
|
|
|
2021-01-16 11:22:14 -05:00
|
|
|
|
2023-12-05 10:29:54 +08:00
|
|
|
|
2021-02-24 12:02:09 -08:00
|
|
|
|
2021-01-16 11:22:14 -05:00
|
|
|
|
2021-02-24 12:02:09 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-30 17:58:40 -05:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:09 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-05-22 09:12:09 -06:00
|
|
|
|
|
|
|
|
|
2014-08-06 16:07:24 -07:00
|
|
|
|
|
|
|
|
|
2023-06-30 14:19:55 -07:00
|
|
|
|
|
|
|
|
|
2023-06-30 14:19:56 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-08-06 16:07:24 -07:00
|
|
|
|
2023-06-30 14:19:55 -07:00
|
|
|
|
2023-06-30 14:19:56 -07:00
|
|
|
|
2014-08-06 16:07:24 -07:00
|
|
|
|
2023-06-30 14:19:55 -07:00
|
|
|
|
2010-10-26 14:21:57 -07:00
|
|
|
|
2023-06-30 14:19:55 -07:00
|
|
|
|
|
|
|
|
|
2020-04-01 21:08:45 -07:00
|
|
|
|
2011-05-24 17:11:30 -07:00
|
|
|
|
2023-06-30 14:19:56 -07:00
|
|
|
|
|
|
|
|
|
2011-05-24 17:11:30 -07:00
|
|
|
|
|
|
|
|
|
2023-06-30 14:19:55 -07:00
|
|
|
|
2011-05-24 17:11:30 -07:00
|
|
|
|
2023-06-30 14:19:56 -07:00
|
|
|
|
2011-05-24 17:11:30 -07:00
|
|
|
|
2021-03-04 10:21:02 -05:00
|
|
|
|
2011-05-24 17:11:30 -07:00
|
|
|
|
2021-03-04 10:21:02 -05:00
|
|
|
|
2023-06-30 14:19:55 -07:00
|
|
|
|
2020-12-14 19:05:02 -08:00
|
|
|
|
|
|
|
|
|
2021-03-18 21:39:45 -04:00
|
|
|
|
2011-05-24 17:11:30 -07:00
|
|
|
|
2020-12-08 00:07:31 -05:00
|
|
|
|
2020-12-14 19:05:02 -08:00
|
|
|
|
2023-06-30 14:19:56 -07:00
|
|
|
|
2023-06-30 14:19:55 -07:00
|
|
|
|
2020-12-14 19:05:02 -08:00
|
|
|
|
|
|
|
|
|
2020-12-08 00:07:31 -05:00
|
|
|
|
2010-10-26 14:21:57 -07:00
|
|
|
|
2020-12-14 19:05:02 -08:00
|
|
|
|
2023-06-30 14:19:55 -07:00
|
|
|
|
2010-10-26 14:21:57 -07:00
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
|
|
|
|
|
2023-06-21 14:24:02 -07:00
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
|
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
2025-06-05 13:49:35 +08:00
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2025-06-05 13:49:35 +08:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
|
|
|
|
|
2024-06-25 12:18:52 +02:00
|
|
|
|
2023-06-21 14:24:02 -07:00
|
|
|
|
2024-06-25 12:18:52 +02:00
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
|
|
|
|
|
2024-06-25 12:18:52 +02:00
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
|
|
|
|
|
2019-05-13 17:21:29 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
|
|
|
|
|
2023-06-21 14:24:02 -07:00
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
|
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-21 14:24:02 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
|
|
|
|
|
2023-06-21 14:24:02 -07:00
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2017-11-21 14:07:06 -05:00
|
|
|
|
2014-04-03 14:47:44 -07:00
|
|
|
|
2021-05-10 16:33:22 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-04-02 14:55:16 +02:00
|
|
|
|
2021-05-10 16:33:22 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:36 -08:00
|
|
|
|
2023-03-07 15:34:05 +01:00
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2020-10-13 16:51:34 -07:00
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
2020-12-15 23:22:38 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2020-12-15 23:22:38 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-03-07 15:34:05 +01:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-10-13 16:51:34 -07:00
|
|
|
|
2020-12-15 23:22:38 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2008-07-25 19:45:31 -07:00
|
|
|
|
|
|
|
|
|
2018-05-16 16:12:50 -04:00
|
|
|
|
2020-12-15 23:22:38 -05:00
|
|
|
|
|
|
|
|
|
2018-05-16 16:12:50 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-15 23:22:38 -05:00
|
|
|
|
2018-05-16 16:12:50 -04:00
|
|
|
|
2016-07-26 15:26:04 -07:00
|
|
|
|
2024-06-25 13:53:50 -07:00
|
|
|
|
2018-05-16 16:12:50 -04:00
|
|
|
|
2016-07-26 15:26:04 -07:00
|
|
|
|
2020-12-15 23:22:38 -05:00
|
|
|
|
|
|
|
|
|
2018-05-16 16:12:50 -04:00
|
|
|
|
2008-07-25 19:45:31 -07:00
|
|
|
|
2010-11-11 14:05:19 -08:00
|
|
|
|
2008-07-25 19:45:31 -07:00
|
|
|
|
|
|
|
|
|
2020-12-15 23:22:38 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
2020-04-01 21:05:07 -07:00
|
|
|
|
|
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-04-01 21:05:07 -07:00
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
2020-04-01 21:05:07 -07:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-05-26 16:43:23 -04:00
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
2023-03-07 15:34:10 +01:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
2023-05-26 16:43:23 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-03-07 15:34:05 +01:00
|
|
|
|
2023-03-07 15:34:09 +01:00
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
|
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
|
|
|
|
|
2023-03-07 15:34:10 +01:00
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
|
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
|
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
|
|
|
|
|
2018-12-28 00:37:35 -08:00
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
2020-08-06 23:19:55 -07:00
|
|
|
|
|
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
|
|
|
|
|
2020-08-06 23:19:55 -07:00
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
2020-12-24 12:55:56 -05:00
|
|
|
|
|
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
2024-08-22 15:50:10 +02:00
|
|
|
|
|
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
2024-08-22 15:50:10 +02:00
|
|
|
|
2023-05-19 16:10:37 -04:00
|
|
|
|
2020-09-24 08:51:40 +02:00
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
2014-12-29 20:30:35 +01:00
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
2022-07-01 14:04:43 -06:00
|
|
|
|
|
|
|
|
|
2025-08-12 21:52:25 +08:00
|
|
|
|
2022-07-01 14:04:43 -06:00
|
|
|
|
filemap: kill page_cache_read usage in filemap_fault
Patch series "drop the mmap_sem when doing IO in the fault path", v6.
Now that we have proper isolation in place with cgroups2 we have started
going through and fixing the various priority inversions. Most are all
gone now, but this one is sort of weird since it's not necessarily a
priority inversion that happens within the kernel, but rather because of
something userspace does.
We have giant applications that we want to protect, and parts of these
giant applications do things like watch the system state to determine how
healthy the box is for load balancing and such. This involves running
'ps' or other such utilities. These utilities will often walk
/proc/<pid>/whatever, and these files can sometimes need to
down_read(&task->mmap_sem). Not usually a big deal, but we noticed when
we are stress testing that sometimes our protected application has latency
spikes trying to get the mmap_sem for tasks that are in lower priority
cgroups.
This is because any down_write() on a semaphore essentially turns it into
a mutex, so even if we currently have it held for reading, any new readers
will not be allowed on to keep from starving the writer. This is fine,
except a lower priority task could be stuck doing IO because it has been
throttled to the point that its IO is taking much longer than normal. But
because a higher priority group depends on this completing it is now stuck
behind lower priority work.
In order to avoid this particular priority inversion we want to use the
existing retry mechanism to stop from holding the mmap_sem at all if we
are going to do IO. This already exists in the read case sort of, but
needed to be extended for more than just grabbing the page lock. With
io.latency we throttle at submit_bio() time, so the readahead stuff can
block and even page_cache_read can block, so all these paths need to have
the mmap_sem dropped.
The other big thing is ->page_mkwrite. btrfs is particularly shitty here
because we have to reserve space for the dirty page, which can be a very
expensive operation. We use the same retry method as the read path, and
simply cache the page and verify the page is still setup properly the next
pass through ->page_mkwrite().
I've tested these patches with xfstests and there are no regressions.
This patch (of 3):
If we do not have a page at filemap_fault time we'll do this weird forced
page_cache_read thing to populate the page, and then drop it again and
loop around and find it. This makes for 2 ways we can read a page in
filemap_fault, and it's not really needed. Instead add a FGP_FOR_MMAP
flag so that pagecache_get_page() will return a unlocked page that's in
pagecache. Then use the normal page locking and readpage logic already in
filemap_fault. This simplifies the no page in page cache case
significantly.
[akpm@linux-foundation.org: fix comment text]
[josef@toxicpanda.com: don't unlock null page in FGP_FOR_MMAP case]
Link: http://lkml.kernel.org/r/20190312201742.22935-1-josef@toxicpanda.com
Link: http://lkml.kernel.org/r/20181211173801.29535-2-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:14 -07:00
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
|
|
|
|
|
2024-08-22 15:50:09 +02:00
|
|
|
|
|
|
|
|
|
2023-05-19 16:10:37 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
2023-05-19 16:10:37 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-08-22 15:50:10 +02:00
|
|
|
|
2023-05-19 16:10:37 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-12-20 08:47:50 -07:00
|
|
|
|
|
|
|
|
|
2023-05-19 16:10:37 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
|
|
|
|
|
2024-08-22 15:50:10 +02:00
|
|
|
|
filemap: kill page_cache_read usage in filemap_fault
Patch series "drop the mmap_sem when doing IO in the fault path", v6.
Now that we have proper isolation in place with cgroups2 we have started
going through and fixing the various priority inversions. Most are all
gone now, but this one is sort of weird since it's not necessarily a
priority inversion that happens within the kernel, but rather because of
something userspace does.
We have giant applications that we want to protect, and parts of these
giant applications do things like watch the system state to determine how
healthy the box is for load balancing and such. This involves running
'ps' or other such utilities. These utilities will often walk
/proc/<pid>/whatever, and these files can sometimes need to
down_read(&task->mmap_sem). Not usually a big deal, but we noticed when
we are stress testing that sometimes our protected application has latency
spikes trying to get the mmap_sem for tasks that are in lower priority
cgroups.
This is because any down_write() on a semaphore essentially turns it into
a mutex, so even if we currently have it held for reading, any new readers
will not be allowed on to keep from starving the writer. This is fine,
except a lower priority task could be stuck doing IO because it has been
throttled to the point that its IO is taking much longer than normal. But
because a higher priority group depends on this completing it is now stuck
behind lower priority work.
In order to avoid this particular priority inversion we want to use the
existing retry mechanism to stop from holding the mmap_sem at all if we
are going to do IO. This already exists in the read case sort of, but
needed to be extended for more than just grabbing the page lock. With
io.latency we throttle at submit_bio() time, so the readahead stuff can
block and even page_cache_read can block, so all these paths need to have
the mmap_sem dropped.
The other big thing is ->page_mkwrite. btrfs is particularly shitty here
because we have to reserve space for the dirty page, which can be a very
expensive operation. We use the same retry method as the read path, and
simply cache the page and verify the page is still setup properly the next
pass through ->page_mkwrite().
I've tested these patches with xfstests and there are no regressions.
This patch (of 3):
If we do not have a page at filemap_fault time we'll do this weird forced
page_cache_read thing to populate the page, and then drop it again and
loop around and find it. This makes for 2 ways we can read a page in
filemap_fault, and it's not really needed. Instead add a FGP_FOR_MMAP
flag so that pagecache_get_page() will return a unlocked page that's in
pagecache. Then use the normal page locking and readpage logic already in
filemap_fault. This simplifies the no page in page cache case
significantly.
[akpm@linux-foundation.org: fix comment text]
[josef@toxicpanda.com: don't unlock null page in FGP_FOR_MMAP case]
Link: http://lkml.kernel.org/r/20190312201742.22935-1-josef@toxicpanda.com
Link: http://lkml.kernel.org/r/20181211173801.29535-2-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:14 -07:00
|
|
|
|
2023-05-19 16:10:37 -04:00
|
|
|
|
|
|
|
|
|
mm: fix error handling in __filemap_get_folio() with FGP_NOWAIT
original report:
https://lore.kernel.org/all/CAKhLTr1UL3ePTpYjXOx2AJfNk8Ku2EdcEfu+CH1sf3Asr=B-Dw@mail.gmail.com/T/
When doing buffered writes with FGP_NOWAIT, under memory pressure, the
system returned ENOMEM despite there being plenty of available memory, to
be reclaimed from page cache. The user space used io_uring interface,
which in turn submits I/O with FGP_NOWAIT (the fast path).
retsnoop pointed to iomap_get_folio:
00:34:16.180612 -> 00:34:16.180651 TID/PID 253786/253721
(reactor-1/combined_tests):
entry_SYSCALL_64_after_hwframe+0x76
do_syscall_64+0x82
__do_sys_io_uring_enter+0x265
io_submit_sqes+0x209
io_issue_sqe+0x5b
io_write+0xdd
xfs_file_buffered_write+0x84
iomap_file_buffered_write+0x1a6
32us [-ENOMEM] iomap_write_begin+0x408
iter=&{.inode=0xffff8c67aa031138,.len=4096,.flags=33,.iomap={.addr=0xffffffffffffffff,.length=4096,.type=1,.flags=3,.bdev=0x…
pos=0 len=4096 foliop=0xffffb32c296b7b80
! 4us [-ENOMEM] iomap_get_folio
iter=&{.inode=0xffff8c67aa031138,.len=4096,.flags=33,.iomap={.addr=0xffffffffffffffff,.length=4096,.type=1,.flags=3,.bdev=0x…
pos=0 len=4096
This is likely a regression caused by 66dabbb65d67 ("mm: return an ERR_PTR
from __filemap_get_folio"), which moved error handling from
io_map_get_folio() to __filemap_get_folio(), but broke FGP_NOWAIT
handling, so ENOMEM is being escaped to user space. Had it correctly
returned -EAGAIN with NOWAIT, either io_uring or user space itself would
be able to retry the request.
It's not enough to patch io_uring since the iomap interface is the one
responsible for it, and pwritev2(RWF_NOWAIT) and AIO interfaces must
return the proper error too.
The patch was tested with scylladb test suite (its original reproducer),
and the tests all pass now when memory is pressured.
Link: https://lkml.kernel.org/r/20250224143700.23035-1-raphaelsc@scylladb.com
Fixes: 66dabbb65d67 ("mm: return an ERR_PTR from __filemap_get_folio")
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Cc: "Darrick J. Wong" <djwong@kernel.org>
Cc: Matthew Wilcow (Oracle) <willy@infradead.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-02-24 11:37:00 -03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-19 16:10:37 -04:00
|
|
|
|
mm: fix error handling in __filemap_get_folio() with FGP_NOWAIT
original report:
https://lore.kernel.org/all/CAKhLTr1UL3ePTpYjXOx2AJfNk8Ku2EdcEfu+CH1sf3Asr=B-Dw@mail.gmail.com/T/
When doing buffered writes with FGP_NOWAIT, under memory pressure, the
system returned ENOMEM despite there being plenty of available memory, to
be reclaimed from page cache. The user space used io_uring interface,
which in turn submits I/O with FGP_NOWAIT (the fast path).
retsnoop pointed to iomap_get_folio:
00:34:16.180612 -> 00:34:16.180651 TID/PID 253786/253721
(reactor-1/combined_tests):
entry_SYSCALL_64_after_hwframe+0x76
do_syscall_64+0x82
__do_sys_io_uring_enter+0x265
io_submit_sqes+0x209
io_issue_sqe+0x5b
io_write+0xdd
xfs_file_buffered_write+0x84
iomap_file_buffered_write+0x1a6
32us [-ENOMEM] iomap_write_begin+0x408
iter=&{.inode=0xffff8c67aa031138,.len=4096,.flags=33,.iomap={.addr=0xffffffffffffffff,.length=4096,.type=1,.flags=3,.bdev=0x…
pos=0 len=4096 foliop=0xffffb32c296b7b80
! 4us [-ENOMEM] iomap_get_folio
iter=&{.inode=0xffff8c67aa031138,.len=4096,.flags=33,.iomap={.addr=0xffffffffffffffff,.length=4096,.type=1,.flags=3,.bdev=0x…
pos=0 len=4096
This is likely a regression caused by 66dabbb65d67 ("mm: return an ERR_PTR
from __filemap_get_folio"), which moved error handling from
io_map_get_folio() to __filemap_get_folio(), but broke FGP_NOWAIT
handling, so ENOMEM is being escaped to user space. Had it correctly
returned -EAGAIN with NOWAIT, either io_uring or user space itself would
be able to retry the request.
It's not enough to patch io_uring since the iomap interface is the one
responsible for it, and pwritev2(RWF_NOWAIT) and AIO interfaces must
return the proper error too.
The patch was tested with scylladb test suite (its original reproducer),
and the tests all pass now when memory is pressured.
Link: https://lkml.kernel.org/r/20250224143700.23035-1-raphaelsc@scylladb.com
Fixes: 66dabbb65d67 ("mm: return an ERR_PTR from __filemap_get_folio")
Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Cc: "Darrick J. Wong" <djwong@kernel.org>
Cc: Matthew Wilcow (Oracle) <willy@infradead.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-02-24 11:37:00 -03:00
|
|
|
|
filemap: kill page_cache_read usage in filemap_fault
Patch series "drop the mmap_sem when doing IO in the fault path", v6.
Now that we have proper isolation in place with cgroups2 we have started
going through and fixing the various priority inversions. Most are all
gone now, but this one is sort of weird since it's not necessarily a
priority inversion that happens within the kernel, but rather because of
something userspace does.
We have giant applications that we want to protect, and parts of these
giant applications do things like watch the system state to determine how
healthy the box is for load balancing and such. This involves running
'ps' or other such utilities. These utilities will often walk
/proc/<pid>/whatever, and these files can sometimes need to
down_read(&task->mmap_sem). Not usually a big deal, but we noticed when
we are stress testing that sometimes our protected application has latency
spikes trying to get the mmap_sem for tasks that are in lower priority
cgroups.
This is because any down_write() on a semaphore essentially turns it into
a mutex, so even if we currently have it held for reading, any new readers
will not be allowed on to keep from starving the writer. This is fine,
except a lower priority task could be stuck doing IO because it has been
throttled to the point that its IO is taking much longer than normal. But
because a higher priority group depends on this completing it is now stuck
behind lower priority work.
In order to avoid this particular priority inversion we want to use the
existing retry mechanism to stop from holding the mmap_sem at all if we
are going to do IO. This already exists in the read case sort of, but
needed to be extended for more than just grabbing the page lock. With
io.latency we throttle at submit_bio() time, so the readahead stuff can
block and even page_cache_read can block, so all these paths need to have
the mmap_sem dropped.
The other big thing is ->page_mkwrite. btrfs is particularly shitty here
because we have to reserve space for the dirty page, which can be a very
expensive operation. We use the same retry method as the read path, and
simply cache the page and verify the page is still setup properly the next
pass through ->page_mkwrite().
I've tested these patches with xfstests and there are no regressions.
This patch (of 3):
If we do not have a page at filemap_fault time we'll do this weird forced
page_cache_read thing to populate the page, and then drop it again and
loop around and find it. This makes for 2 ways we can read a page in
filemap_fault, and it's not really needed. Instead add a FGP_FOR_MMAP
flag so that pagecache_get_page() will return a unlocked page that's in
pagecache. Then use the normal page locking and readpage logic already in
filemap_fault. This simplifies the no page in page cache case
significantly.
[akpm@linux-foundation.org: fix comment text]
[josef@toxicpanda.com: don't unlock null page in FGP_FOR_MMAP case]
Link: http://lkml.kernel.org/r/20190312201742.22935-1-josef@toxicpanda.com
Link: http://lkml.kernel.org/r/20181211173801.29535-2-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:14 -07:00
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
|
|
|
|
|
filemap: kill page_cache_read usage in filemap_fault
Patch series "drop the mmap_sem when doing IO in the fault path", v6.
Now that we have proper isolation in place with cgroups2 we have started
going through and fixing the various priority inversions. Most are all
gone now, but this one is sort of weird since it's not necessarily a
priority inversion that happens within the kernel, but rather because of
something userspace does.
We have giant applications that we want to protect, and parts of these
giant applications do things like watch the system state to determine how
healthy the box is for load balancing and such. This involves running
'ps' or other such utilities. These utilities will often walk
/proc/<pid>/whatever, and these files can sometimes need to
down_read(&task->mmap_sem). Not usually a big deal, but we noticed when
we are stress testing that sometimes our protected application has latency
spikes trying to get the mmap_sem for tasks that are in lower priority
cgroups.
This is because any down_write() on a semaphore essentially turns it into
a mutex, so even if we currently have it held for reading, any new readers
will not be allowed on to keep from starving the writer. This is fine,
except a lower priority task could be stuck doing IO because it has been
throttled to the point that its IO is taking much longer than normal. But
because a higher priority group depends on this completing it is now stuck
behind lower priority work.
In order to avoid this particular priority inversion we want to use the
existing retry mechanism to stop from holding the mmap_sem at all if we
are going to do IO. This already exists in the read case sort of, but
needed to be extended for more than just grabbing the page lock. With
io.latency we throttle at submit_bio() time, so the readahead stuff can
block and even page_cache_read can block, so all these paths need to have
the mmap_sem dropped.
The other big thing is ->page_mkwrite. btrfs is particularly shitty here
because we have to reserve space for the dirty page, which can be a very
expensive operation. We use the same retry method as the read path, and
simply cache the page and verify the page is still setup properly the next
pass through ->page_mkwrite().
I've tested these patches with xfstests and there are no regressions.
This patch (of 3):
If we do not have a page at filemap_fault time we'll do this weird forced
page_cache_read thing to populate the page, and then drop it again and
loop around and find it. This makes for 2 ways we can read a page in
filemap_fault, and it's not really needed. Instead add a FGP_FOR_MMAP
flag so that pagecache_get_page() will return a unlocked page that's in
pagecache. Then use the normal page locking and readpage logic already in
filemap_fault. This simplifies the no page in page cache case
significantly.
[akpm@linux-foundation.org: fix comment text]
[josef@toxicpanda.com: don't unlock null page in FGP_FOR_MMAP case]
Link: http://lkml.kernel.org/r/20190312201742.22935-1-josef@toxicpanda.com
Link: http://lkml.kernel.org/r/20181211173801.29535-2-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:14 -07:00
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
2023-03-07 15:34:10 +01:00
|
|
|
|
|
|
|
|
|
2024-12-20 08:47:50 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-03-08 11:45:35 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:44 -08:00
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:44 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:44 -08:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:44 -08:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:44 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:44 -08:00
|
|
|
|
2024-06-25 13:53:50 -07:00
|
|
|
|
2021-02-25 17:15:44 -08:00
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:44 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:44 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-25 17:16:00 -08:00
|
|
|
|
2020-09-01 23:17:50 -04:00
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:16:11 -08:00
|
|
|
|
2020-09-01 23:17:50 -04:00
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
2020-09-01 23:17:50 -04:00
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
2020-09-01 23:17:50 -04:00
|
|
|
|
2014-05-06 12:50:05 -07:00
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
2020-09-01 23:17:50 -04:00
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
2022-10-17 09:18:00 -07:00
|
|
|
|
2020-09-01 23:17:50 -04:00
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
2022-10-17 09:18:00 -07:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2020-09-01 23:17:50 -04:00
|
|
|
|
|
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:16:11 -08:00
|
|
|
|
2022-10-17 09:18:00 -07:00
|
|
|
|
2024-08-12 15:42:05 +08:00
|
|
|
|
2022-10-17 09:18:00 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-09-26 12:20:17 -07:00
|
|
|
|
2022-10-17 09:18:00 -07:00
|
|
|
|
2024-08-12 15:42:05 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-10-17 09:18:00 -07:00
|
|
|
|
2024-08-12 15:42:05 +08:00
|
|
|
|
|
|
|
|
|
2020-09-01 23:17:50 -04:00
|
|
|
|
2014-04-03 14:47:46 -07:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-10-17 09:17:59 -07:00
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
2022-10-17 09:17:59 -07:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2024-08-12 15:42:05 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2024-08-12 15:42:05 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
2024-08-12 15:42:05 +08:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
|
|
|
|
|
2024-08-12 15:42:05 +08:00
|
|
|
|
2024-09-06 16:05:12 -07:00
|
|
|
|
2024-08-12 15:42:05 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
2024-08-12 15:42:05 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
2020-06-27 22:19:08 -04:00
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-07 14:15:07 -05:00
|
|
|
|
2021-02-25 17:15:56 -08:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2022-06-03 15:30:25 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2017-09-06 16:21:21 -07:00
|
|
|
|
2022-06-03 15:30:25 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2022-06-03 15:30:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2022-06-03 15:30:25 -04:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2022-06-03 15:30:25 -04:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-10-06 13:01:20 +02:00
|
|
|
|
2022-06-03 15:30:25 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-04-27 08:46:01 +02:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
2006-04-27 08:46:01 +02:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-04-27 08:46:01 +02:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-04-27 08:46:01 +02:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
|
|
|
|
|
2006-04-27 08:46:01 +02:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-04-27 08:46:01 +02:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
|
|
|
|
|
2021-03-06 16:38:38 -05:00
|
|
|
|
2008-07-25 19:45:31 -07:00
|
|
|
|
|
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-06 16:38:38 -05:00
|
|
|
|
2018-05-16 18:00:33 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-06 16:38:38 -05:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
2006-04-27 08:46:01 +02:00
|
|
|
|
2024-09-03 07:25:17 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-06-25 13:53:50 -07:00
|
|
|
|
2018-05-16 18:00:33 -04:00
|
|
|
|
2016-07-26 15:26:04 -07:00
|
|
|
|
2021-03-06 16:38:38 -05:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
2008-07-25 19:45:31 -07:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-06-27 22:19:08 -04:00
|
|
|
|
2025-04-03 16:54:17 -07:00
|
|
|
|
2018-05-16 18:00:33 -04:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
2021-03-06 16:38:38 -05:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
2018-05-16 18:00:33 -04:00
|
|
|
|
|
|
|
|
|
2006-04-27 08:46:01 +02:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-11-07 10:46:34 +08:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
|
|
|
|
|
2008-07-25 19:45:31 -07:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
2006-04-27 08:46:01 +02:00
|
|
|
|
2022-08-23 17:40:17 -07:00
|
|
|
|
2006-04-27 08:46:01 +02:00
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2023-01-04 13:14:27 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2023-10-06 13:01:20 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
2023-01-04 13:14:27 -08:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-01-04 13:14:27 -08:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-01-04 13:14:27 -08:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2008-07-25 19:45:31 -07:00
|
|
|
|
|
|
|
|
|
2023-01-04 13:14:27 -08:00
|
|
|
|
2018-05-16 18:12:54 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-04 13:14:27 -08:00
|
|
|
|
2018-05-16 18:12:54 -04:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2014-05-06 12:50:05 -07:00
|
|
|
|
2023-01-04 13:14:27 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-11-15 17:34:33 -08:00
|
|
|
|
|
|
|
|
|
2008-07-25 19:45:31 -07:00
|
|
|
|
2017-11-15 17:34:33 -08:00
|
|
|
|
2023-01-04 13:14:27 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-11-15 17:34:33 -08:00
|
|
|
|
|
|
|
|
|
2023-01-04 13:14:27 -08:00
|
|
|
|
2017-11-15 17:34:33 -08:00
|
|
|
|
2023-01-04 13:14:27 -08:00
|
|
|
|
2017-11-15 17:34:33 -08:00
|
|
|
|
2008-07-25 19:45:31 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-01-04 13:14:27 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-01-04 13:14:27 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
[PATCH] readahead: backoff on I/O error
Backoff readahead size exponentially on I/O error.
Michael Tokarev <mjt@tls.msk.ru> described the problem as:
[QUOTE]
Suppose there's a CD-rom with a scratch/etc, one sector is unreadable.
In order to "fix" it, one have to read it and write to another CD-rom,
or something.. or just ignore the error (if it's just a skip in a video
stream). Let's assume the unreadable block is number U.
But current behavior is just insane. An application requests block
number N, which is before U. Kernel tries to read-ahead blocks N..U.
Cdrom drive tries to read it, re-read it.. for some time. Finally,
when all the N..U-1 blocks are read, kernel returns block number N
(as requested) to an application, successefully.
Now an app requests block number N+1, and kernel tries to read
blocks N+1..U+1. Retrying again as in previous step.
And so on, up to when an app requests block number U-1. And when,
finally, it requests block U, it receives read error.
So, kernel currentry tries to re-read the same failing block as
many times as the current readahead value (256 (times?) by default).
This whole process already killed my cdrom drive (I posted about it
to LKML several months ago) - literally, the drive has fried, and
does not work anymore. Ofcourse that problem was a bug in firmware
(or whatever) of the drive *too*, but.. main problem with that is
current readahead logic as described above.
[/QUOTE]
Which was confirmed by Jens Axboe <axboe@suse.de>:
[QUOTE]
For ide-cd, it tends do only end the first part of the request on a
medium error. So you may see a lot of repeats :/
[/QUOTE]
With this patch, retries are expected to be reduced from, say, 256, to 5.
[akpm@osdl.org: cleanups]
Signed-off-by: Wu Fengguang <wfg@mail.ustc.edu.cn>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 05:48:43 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-04-01 21:04:50 -07:00
|
|
|
|
[PATCH] readahead: backoff on I/O error
Backoff readahead size exponentially on I/O error.
Michael Tokarev <mjt@tls.msk.ru> described the problem as:
[QUOTE]
Suppose there's a CD-rom with a scratch/etc, one sector is unreadable.
In order to "fix" it, one have to read it and write to another CD-rom,
or something.. or just ignore the error (if it's just a skip in a video
stream). Let's assume the unreadable block is number U.
But current behavior is just insane. An application requests block
number N, which is before U. Kernel tries to read-ahead blocks N..U.
Cdrom drive tries to read it, re-read it.. for some time. Finally,
when all the N..U-1 blocks are read, kernel returns block number N
(as requested) to an application, successefully.
Now an app requests block number N+1, and kernel tries to read
blocks N+1..U+1. Retrying again as in previous step.
And so on, up to when an app requests block number U-1. And when,
finally, it requests block U, it receives read error.
So, kernel currentry tries to re-read the same failing block as
many times as the current readahead value (256 (times?) by default).
This whole process already killed my cdrom drive (I posted about it
to LKML several months ago) - literally, the drive has fried, and
does not work anymore. Ofcourse that problem was a bug in firmware
(or whatever) of the drive *too*, but.. main problem with that is
current readahead logic as described above.
[/QUOTE]
Which was confirmed by Jens Axboe <axboe@suse.de>:
[QUOTE]
For ide-cd, it tends do only end the first part of the request on a
medium error. So you may see a lot of repeats :/
[/QUOTE]
With this patch, retries are expected to be reduced from, say, 256, to 5.
[akpm@osdl.org: cleanups]
Signed-off-by: Wu Fengguang <wfg@mail.ustc.edu.cn>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-25 05:48:43 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
|
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
|
|
|
|
|
2021-03-05 10:29:41 -05:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
|
|
|
|
|
2021-03-05 10:29:41 -05:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2021-03-05 10:29:41 -05:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2022-06-17 20:00:17 -04:00
|
|
|
|
|
|
|
|
|
2024-06-25 13:53:50 -07:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
|
|
|
|
|
2021-03-05 10:29:41 -05:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2021-03-05 10:29:41 -05:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2021-03-05 10:29:41 -05:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2023-06-27 10:43:49 -07:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2021-03-05 10:29:41 -05:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-05-12 17:37:01 -04:00
|
|
|
|
2021-03-10 10:19:30 -05:00
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
2022-09-15 10:41:56 +01:00
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-09-15 10:41:56 +01:00
|
|
|
|
|
|
|
|
|
2022-05-12 17:37:01 -04:00
|
|
|
|
2022-09-15 10:41:56 +01:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:15 -08:00
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
2021-03-10 10:19:30 -05:00
|
|
|
|
2021-02-24 12:02:15 -08:00
|
|
|
|
|
|
|
|
|
2021-03-10 10:19:30 -05:00
|
|
|
|
2021-02-24 12:02:38 -08:00
|
|
|
|
2022-05-12 17:37:01 -04:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:38 -08:00
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:28 -08:00
|
|
|
|
2023-02-08 18:18:17 +00:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:28 -08:00
|
|
|
|
2021-03-10 11:04:19 -05:00
|
|
|
|
2021-02-24 12:02:28 -08:00
|
|
|
|
|
|
|
|
|
2023-02-08 18:18:17 +00:00
|
|
|
|
2021-02-24 12:02:28 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-10 11:04:19 -05:00
|
|
|
|
2021-02-24 12:02:28 -08:00
|
|
|
|
|
|
|
|
|
2021-03-10 11:04:19 -05:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:28 -08:00
|
|
|
|
|
|
|
|
|
2021-03-10 11:04:19 -05:00
|
|
|
|
2021-02-24 12:02:28 -08:00
|
|
|
|
|
|
|
|
|
2025-07-28 16:39:51 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-02-09 20:21:27 +00:00
|
|
|
|
2021-02-24 12:02:28 -08:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:22 -08:00
|
|
|
|
2023-02-08 18:18:17 +00:00
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-30 17:58:40 -05:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2021-02-24 12:02:25 -08:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2021-02-24 12:02:25 -08:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2021-08-16 23:36:31 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:22 -08:00
|
|
|
|
2021-02-24 12:02:05 -08:00
|
|
|
|
2020-12-30 17:58:40 -05:00
|
|
|
|
2021-02-24 12:02:25 -08:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
|
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2020-12-30 17:58:40 -05:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
2021-02-24 12:02:28 -08:00
|
|
|
|
2023-02-08 18:18:17 +00:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:28 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-05-12 17:37:01 -04:00
|
|
|
|
|
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2021-02-24 12:02:28 -08:00
|
|
|
|
2020-12-30 17:58:40 -05:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-30 17:58:40 -05:00
|
|
|
|
2021-02-24 12:02:28 -08:00
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
|
|
|
|
|
mm/filemap: change filemap_create_folio() to take a struct kiocb
Patch series "Uncached buffered IO", v8.
5 years ago I posted patches adding support for RWF_UNCACHED, as a way to
do buffered IO that isn't page cache persistent. The approach back then
was to have private pages for IO, and then get rid of them once IO was
done. But that then runs into all the issues that O_DIRECT has, in terms
of synchronizing with the page cache.
So here's a new approach to the same concent, but using the page cache as
synchronization. Due to excessive bike shedding on the naming, this is
now named RWF_DONTCACHE, and is less special in that it's just page cache
IO, except it prunes the ranges once IO is completed.
Why do this, you may ask? The tldr is that device speeds are only getting
faster, while reclaim is not. Doing normal buffered IO can be very
unpredictable, and suck up a lot of resources on the reclaim side. This
leads people to use O_DIRECT as a work-around, which has its own set of
restrictions in terms of size, offset, and length of IO. It's also
inherently synchronous, and now you need async IO as well. While the
latter isn't necessarily a big problem as we have good options available
there, it also should not be a requirement when all you want to do is read
or write some data without caching.
Even on desktop type systems, a normal NVMe device can fill the entire
page cache in seconds. On the big system I used for testing, there's a
lot more RAM, but also a lot more devices. As can be seen in some of the
results in the following patches, you can still fill RAM in seconds even
when there's 1TB of it. Hence this problem isn't solely a "big
hyperscaler system" issue, it's common across the board.
Common for both reads and writes with RWF_DONTCACHE is that they use the
page cache for IO. Reads work just like a normal buffered read would,
with the only exception being that the touched ranges will get pruned
after data has been copied. For writes, the ranges will get writeback
kicked off before the syscall returns, and then writeback completion will
prune the range. Hence writes aren't synchronous, and it's easy to
pipeline writes using RWF_DONTCACHE. Folios that aren't instantiated by
RWF_DONTCACHE IO are left untouched. This means you that uncached IO will
take advantage of the page cache for uptodate data, but not leave anything
it instantiated/created in cache.
File systems need to support this. This patchset adds support for the
generic read path, which covers file systems like ext4. Patches exist to
add support for iomap/XFS and btrfs as well, which sit on top of this
series. If RWF_DONTCACHE IO is attempted on a file system that doesn't
support it, -EOPNOTSUPP is returned. Hence the user can rely on it either
working as designed, or flagging and error if that's not the case. The
intent here is to give the application a sensible fallback path - eg, it
may fall back to O_DIRECT if appropriate, or just live with the fact that
uncached IO isn't available and do normal buffered IO.
Adding "support" to other file systems should be trivial, most of the time
just a one-liner adding FOP_DONTCACHE to the fop_flags in the
file_operations struct, if the file system is using either iomap or the
generic filemap helpers for reading and writing.
Performance results are in patch 8 for reads, and you can find the write
side results in the XFS patch adding support for DONTCACHE writes for XFS:
https://git.kernel.dk/cgit/linux/commit/?h=buffered-uncached-fs.10&id=257e92de795fdff7d7e256501e024fac6da6a7f4
with the tldr being that I see about a 65% improvement in performance for
both, with fully predictable IO times. CPU reduction is substantial as
well, with no kswapd activity at all for reclaim when using uncached IO.
Using it from applications is trivial - just set RWF_DONTCACHE for the
read or write, using pwritev2(2) or preadv2(2). For io_uring, same thing,
just set RWF_DONTCACHE in sqe->rw_flags for a buffered read/write
operation. And that's it.
Patches 1..7 are just prep patches, and should have no functional changes
at all. Patch 8 adds support for the filemap path for RWF_DONTCACHE
reads, and patches 9..12 are just prep patches for supporting the write
side of uncached writes. In the below mentioned branch, there are then
patches to adopt uncached reads and writes for xfs, btrfs, and ext4. The
latter currently relies on bit of a hack for passing whether this is an
uncached write or not through ->write_begin(), which can hopefully go away
once ext4 adopts iomap for buffered writes. I say this is a hack as it's
not the prettiest way to do it, however it is fully solid and will work
just fine.
Passes full xfstests and fsx overnight runs, no issues observed. That
includes the vm running the testing also using RWF_DONTCACHE on the host.
I'll post fsstress and fsx patches for RWF_DONTCACHE separately. As far
as I'm concerned, no further work needs doing here.
And git tree for the patches is here:
https://git.kernel.dk/cgit/linux/log/?h=buffered-uncached.10
with the file system patches on top adding support for xfs/btrfs/ext4
here:
https://git.kernel.dk/cgit/linux/log/?h=buffered-uncached-fs.10
This patch (of 12):
Rather than pass in both the file and position directly from the kiocb,
just take a struct kiocb instead. With the kiocb being passed in, skip
passing in the address_space separately as well. While doing so, move the
ki_flags checking into filemap_create_folio() as well. In preparation for
actually needing the kiocb in the function.
No functional changes in this patch.
Link: https://lkml.kernel.org/r/20241220154831.1086649-1-axboe@kernel.dk
Link: https://lkml.kernel.org/r/20241220154831.1086649-2-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:39 -07:00
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
mm/filemap: change filemap_create_folio() to take a struct kiocb
Patch series "Uncached buffered IO", v8.
5 years ago I posted patches adding support for RWF_UNCACHED, as a way to
do buffered IO that isn't page cache persistent. The approach back then
was to have private pages for IO, and then get rid of them once IO was
done. But that then runs into all the issues that O_DIRECT has, in terms
of synchronizing with the page cache.
So here's a new approach to the same concent, but using the page cache as
synchronization. Due to excessive bike shedding on the naming, this is
now named RWF_DONTCACHE, and is less special in that it's just page cache
IO, except it prunes the ranges once IO is completed.
Why do this, you may ask? The tldr is that device speeds are only getting
faster, while reclaim is not. Doing normal buffered IO can be very
unpredictable, and suck up a lot of resources on the reclaim side. This
leads people to use O_DIRECT as a work-around, which has its own set of
restrictions in terms of size, offset, and length of IO. It's also
inherently synchronous, and now you need async IO as well. While the
latter isn't necessarily a big problem as we have good options available
there, it also should not be a requirement when all you want to do is read
or write some data without caching.
Even on desktop type systems, a normal NVMe device can fill the entire
page cache in seconds. On the big system I used for testing, there's a
lot more RAM, but also a lot more devices. As can be seen in some of the
results in the following patches, you can still fill RAM in seconds even
when there's 1TB of it. Hence this problem isn't solely a "big
hyperscaler system" issue, it's common across the board.
Common for both reads and writes with RWF_DONTCACHE is that they use the
page cache for IO. Reads work just like a normal buffered read would,
with the only exception being that the touched ranges will get pruned
after data has been copied. For writes, the ranges will get writeback
kicked off before the syscall returns, and then writeback completion will
prune the range. Hence writes aren't synchronous, and it's easy to
pipeline writes using RWF_DONTCACHE. Folios that aren't instantiated by
RWF_DONTCACHE IO are left untouched. This means you that uncached IO will
take advantage of the page cache for uptodate data, but not leave anything
it instantiated/created in cache.
File systems need to support this. This patchset adds support for the
generic read path, which covers file systems like ext4. Patches exist to
add support for iomap/XFS and btrfs as well, which sit on top of this
series. If RWF_DONTCACHE IO is attempted on a file system that doesn't
support it, -EOPNOTSUPP is returned. Hence the user can rely on it either
working as designed, or flagging and error if that's not the case. The
intent here is to give the application a sensible fallback path - eg, it
may fall back to O_DIRECT if appropriate, or just live with the fact that
uncached IO isn't available and do normal buffered IO.
Adding "support" to other file systems should be trivial, most of the time
just a one-liner adding FOP_DONTCACHE to the fop_flags in the
file_operations struct, if the file system is using either iomap or the
generic filemap helpers for reading and writing.
Performance results are in patch 8 for reads, and you can find the write
side results in the XFS patch adding support for DONTCACHE writes for XFS:
https://git.kernel.dk/cgit/linux/commit/?h=buffered-uncached-fs.10&id=257e92de795fdff7d7e256501e024fac6da6a7f4
with the tldr being that I see about a 65% improvement in performance for
both, with fully predictable IO times. CPU reduction is substantial as
well, with no kswapd activity at all for reclaim when using uncached IO.
Using it from applications is trivial - just set RWF_DONTCACHE for the
read or write, using pwritev2(2) or preadv2(2). For io_uring, same thing,
just set RWF_DONTCACHE in sqe->rw_flags for a buffered read/write
operation. And that's it.
Patches 1..7 are just prep patches, and should have no functional changes
at all. Patch 8 adds support for the filemap path for RWF_DONTCACHE
reads, and patches 9..12 are just prep patches for supporting the write
side of uncached writes. In the below mentioned branch, there are then
patches to adopt uncached reads and writes for xfs, btrfs, and ext4. The
latter currently relies on bit of a hack for passing whether this is an
uncached write or not through ->write_begin(), which can hopefully go away
once ext4 adopts iomap for buffered writes. I say this is a hack as it's
not the prettiest way to do it, however it is fully solid and will work
just fine.
Passes full xfstests and fsx overnight runs, no issues observed. That
includes the vm running the testing also using RWF_DONTCACHE on the host.
I'll post fsstress and fsx patches for RWF_DONTCACHE separately. As far
as I'm concerned, no further work needs doing here.
And git tree for the patches is here:
https://git.kernel.dk/cgit/linux/log/?h=buffered-uncached.10
with the file system patches on top adding support for xfs/btrfs/ext4
here:
https://git.kernel.dk/cgit/linux/log/?h=buffered-uncached-fs.10
This patch (of 12):
Rather than pass in both the file and position directly from the kiocb,
just take a struct kiocb instead. With the kiocb being passed in, skip
passing in the address_space separately as well. While doing so, move the
ki_flags checking into filemap_create_folio() as well. In preparation for
actually needing the kiocb in the function.
No functional changes in this patch.
Link: https://lkml.kernel.org/r/20241220154831.1086649-1-axboe@kernel.dk
Link: https://lkml.kernel.org/r/20241220154831.1086649-2-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:39 -07:00
|
|
|
|
2021-03-10 10:34:00 -05:00
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
2024-08-22 15:50:10 +02:00
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
mm/filemap: change filemap_create_folio() to take a struct kiocb
Patch series "Uncached buffered IO", v8.
5 years ago I posted patches adding support for RWF_UNCACHED, as a way to
do buffered IO that isn't page cache persistent. The approach back then
was to have private pages for IO, and then get rid of them once IO was
done. But that then runs into all the issues that O_DIRECT has, in terms
of synchronizing with the page cache.
So here's a new approach to the same concent, but using the page cache as
synchronization. Due to excessive bike shedding on the naming, this is
now named RWF_DONTCACHE, and is less special in that it's just page cache
IO, except it prunes the ranges once IO is completed.
Why do this, you may ask? The tldr is that device speeds are only getting
faster, while reclaim is not. Doing normal buffered IO can be very
unpredictable, and suck up a lot of resources on the reclaim side. This
leads people to use O_DIRECT as a work-around, which has its own set of
restrictions in terms of size, offset, and length of IO. It's also
inherently synchronous, and now you need async IO as well. While the
latter isn't necessarily a big problem as we have good options available
there, it also should not be a requirement when all you want to do is read
or write some data without caching.
Even on desktop type systems, a normal NVMe device can fill the entire
page cache in seconds. On the big system I used for testing, there's a
lot more RAM, but also a lot more devices. As can be seen in some of the
results in the following patches, you can still fill RAM in seconds even
when there's 1TB of it. Hence this problem isn't solely a "big
hyperscaler system" issue, it's common across the board.
Common for both reads and writes with RWF_DONTCACHE is that they use the
page cache for IO. Reads work just like a normal buffered read would,
with the only exception being that the touched ranges will get pruned
after data has been copied. For writes, the ranges will get writeback
kicked off before the syscall returns, and then writeback completion will
prune the range. Hence writes aren't synchronous, and it's easy to
pipeline writes using RWF_DONTCACHE. Folios that aren't instantiated by
RWF_DONTCACHE IO are left untouched. This means you that uncached IO will
take advantage of the page cache for uptodate data, but not leave anything
it instantiated/created in cache.
File systems need to support this. This patchset adds support for the
generic read path, which covers file systems like ext4. Patches exist to
add support for iomap/XFS and btrfs as well, which sit on top of this
series. If RWF_DONTCACHE IO is attempted on a file system that doesn't
support it, -EOPNOTSUPP is returned. Hence the user can rely on it either
working as designed, or flagging and error if that's not the case. The
intent here is to give the application a sensible fallback path - eg, it
may fall back to O_DIRECT if appropriate, or just live with the fact that
uncached IO isn't available and do normal buffered IO.
Adding "support" to other file systems should be trivial, most of the time
just a one-liner adding FOP_DONTCACHE to the fop_flags in the
file_operations struct, if the file system is using either iomap or the
generic filemap helpers for reading and writing.
Performance results are in patch 8 for reads, and you can find the write
side results in the XFS patch adding support for DONTCACHE writes for XFS:
https://git.kernel.dk/cgit/linux/commit/?h=buffered-uncached-fs.10&id=257e92de795fdff7d7e256501e024fac6da6a7f4
with the tldr being that I see about a 65% improvement in performance for
both, with fully predictable IO times. CPU reduction is substantial as
well, with no kswapd activity at all for reclaim when using uncached IO.
Using it from applications is trivial - just set RWF_DONTCACHE for the
read or write, using pwritev2(2) or preadv2(2). For io_uring, same thing,
just set RWF_DONTCACHE in sqe->rw_flags for a buffered read/write
operation. And that's it.
Patches 1..7 are just prep patches, and should have no functional changes
at all. Patch 8 adds support for the filemap path for RWF_DONTCACHE
reads, and patches 9..12 are just prep patches for supporting the write
side of uncached writes. In the below mentioned branch, there are then
patches to adopt uncached reads and writes for xfs, btrfs, and ext4. The
latter currently relies on bit of a hack for passing whether this is an
uncached write or not through ->write_begin(), which can hopefully go away
once ext4 adopts iomap for buffered writes. I say this is a hack as it's
not the prettiest way to do it, however it is fully solid and will work
just fine.
Passes full xfstests and fsx overnight runs, no issues observed. That
includes the vm running the testing also using RWF_DONTCACHE on the host.
I'll post fsstress and fsx patches for RWF_DONTCACHE separately. As far
as I'm concerned, no further work needs doing here.
And git tree for the patches is here:
https://git.kernel.dk/cgit/linux/log/?h=buffered-uncached.10
with the file system patches on top adding support for xfs/btrfs/ext4
here:
https://git.kernel.dk/cgit/linux/log/?h=buffered-uncached-fs.10
This patch (of 12):
Rather than pass in both the file and position directly from the kiocb,
just take a struct kiocb instead. With the kiocb being passed in, skip
passing in the address_space separately as well. While doing so, move the
ki_flags checking into filemap_create_folio() as well. In preparation for
actually needing the kiocb in the function.
No functional changes in this patch.
Link: https://lkml.kernel.org/r/20241220154831.1086649-1-axboe@kernel.dk
Link: https://lkml.kernel.org/r/20241220154831.1086649-2-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:39 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-08-22 15:50:10 +02:00
|
|
|
|
2021-03-10 10:34:00 -05:00
|
|
|
|
2021-02-24 12:02:18 -08:00
|
|
|
|
mm/filemap: add read support for RWF_DONTCACHE
Add RWF_DONTCACHE as a read operation flag, which means that any data read
wil be removed from the page cache upon completion. Uses the page cache
to synchronize, and simply prunes folios that were instantiated when the
operation completes. While it would be possible to use private pages for
this, using the page cache as synchronization is handy for a variety of
reasons:
1) No special truncate magic is needed
2) Async buffered reads need some place to serialize, using the page
cache is a lot easier than writing extra code for this
3) The pruning cost is pretty reasonable
and the code to support this is much simpler as a result.
You can think of uncached buffered IO as being the much more attractive
cousin of O_DIRECT - it has none of the restrictions of O_DIRECT. Yes, it
will copy the data, but unlike regular buffered IO, it doesn't run into
the unpredictability of the page cache in terms of reclaim. As an
example, on a test box with 32 drives, reading them with buffered IO looks
as follows:
Reading bs 65536, uncached 0
1s: 145945MB/sec
2s: 158067MB/sec
3s: 157007MB/sec
4s: 148622MB/sec
5s: 118824MB/sec
6s: 70494MB/sec
7s: 41754MB/sec
8s: 90811MB/sec
9s: 92204MB/sec
10s: 95178MB/sec
11s: 95488MB/sec
12s: 95552MB/sec
13s: 96275MB/sec
where it's quite easy to see where the page cache filled up, and
performance went from good to erratic, and finally settles at a much
lower rate. Looking at top while this is ongoing, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7535 root 20 0 267004 0 0 S 3199 0.0 8:40.65 uncached
3326 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd4
3327 root 20 0 0 0 0 R 100.0 0.0 0:17.22 kswapd5
3328 root 20 0 0 0 0 R 100.0 0.0 0:13.29 kswapd6
3332 root 20 0 0 0 0 R 100.0 0.0 0:11.11 kswapd10
3339 root 20 0 0 0 0 R 100.0 0.0 0:16.25 kswapd17
3348 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd26
3343 root 20 0 0 0 0 R 100.0 0.0 0:16.30 kswapd21
3344 root 20 0 0 0 0 R 100.0 0.0 0:11.92 kswapd22
3349 root 20 0 0 0 0 R 100.0 0.0 0:16.28 kswapd27
3352 root 20 0 0 0 0 R 99.7 0.0 0:11.89 kswapd30
3353 root 20 0 0 0 0 R 96.7 0.0 0:16.04 kswapd31
3329 root 20 0 0 0 0 R 96.4 0.0 0:11.41 kswapd7
3345 root 20 0 0 0 0 R 96.4 0.0 0:13.40 kswapd23
3330 root 20 0 0 0 0 S 91.1 0.0 0:08.28 kswapd8
3350 root 20 0 0 0 0 S 86.8 0.0 0:11.13 kswapd28
3325 root 20 0 0 0 0 S 76.3 0.0 0:07.43 kswapd3
3341 root 20 0 0 0 0 S 74.7 0.0 0:08.85 kswapd19
3334 root 20 0 0 0 0 S 71.7 0.0 0:10.04 kswapd12
3351 root 20 0 0 0 0 R 60.5 0.0 0:09.59 kswapd29
3323 root 20 0 0 0 0 R 57.6 0.0 0:11.50 kswapd1
[...]
which is just showing a partial list of the 32 kswapd threads that are
running mostly full tilt, burning ~28 full CPU cores.
If the same test case is run with RWF_DONTCACHE set for the buffered read,
the output looks as follows:
Reading bs 65536, uncached 0
1s: 153144MB/sec
2s: 156760MB/sec
3s: 158110MB/sec
4s: 158009MB/sec
5s: 158043MB/sec
6s: 157638MB/sec
7s: 157999MB/sec
8s: 158024MB/sec
9s: 157764MB/sec
10s: 157477MB/sec
11s: 157417MB/sec
12s: 157455MB/sec
13s: 157233MB/sec
14s: 156692MB/sec
which is just chugging along at ~155GB/sec of read performance. Looking
at top, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7961 root 20 0 267004 0 0 S 3180 0.0 5:37.95 uncached
8024 axboe 20 0 14292 4096 0 R 1.0 0.0 0:00.13 top
where just the test app is using CPU, no reclaim is taking place outside
of the main thread. Not only is performance 65% better, it's also using
half the CPU to do it.
Link: https://lkml.kernel.org/r/20241220154831.1086649-9-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:46 -07:00
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2021-03-10 10:34:00 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-03-23 21:29:04 -04:00
|
|
|
|
2021-03-10 10:34:00 -05:00
|
|
|
|
|
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
mm/filemap: change filemap_create_folio() to take a struct kiocb
Patch series "Uncached buffered IO", v8.
5 years ago I posted patches adding support for RWF_UNCACHED, as a way to
do buffered IO that isn't page cache persistent. The approach back then
was to have private pages for IO, and then get rid of them once IO was
done. But that then runs into all the issues that O_DIRECT has, in terms
of synchronizing with the page cache.
So here's a new approach to the same concent, but using the page cache as
synchronization. Due to excessive bike shedding on the naming, this is
now named RWF_DONTCACHE, and is less special in that it's just page cache
IO, except it prunes the ranges once IO is completed.
Why do this, you may ask? The tldr is that device speeds are only getting
faster, while reclaim is not. Doing normal buffered IO can be very
unpredictable, and suck up a lot of resources on the reclaim side. This
leads people to use O_DIRECT as a work-around, which has its own set of
restrictions in terms of size, offset, and length of IO. It's also
inherently synchronous, and now you need async IO as well. While the
latter isn't necessarily a big problem as we have good options available
there, it also should not be a requirement when all you want to do is read
or write some data without caching.
Even on desktop type systems, a normal NVMe device can fill the entire
page cache in seconds. On the big system I used for testing, there's a
lot more RAM, but also a lot more devices. As can be seen in some of the
results in the following patches, you can still fill RAM in seconds even
when there's 1TB of it. Hence this problem isn't solely a "big
hyperscaler system" issue, it's common across the board.
Common for both reads and writes with RWF_DONTCACHE is that they use the
page cache for IO. Reads work just like a normal buffered read would,
with the only exception being that the touched ranges will get pruned
after data has been copied. For writes, the ranges will get writeback
kicked off before the syscall returns, and then writeback completion will
prune the range. Hence writes aren't synchronous, and it's easy to
pipeline writes using RWF_DONTCACHE. Folios that aren't instantiated by
RWF_DONTCACHE IO are left untouched. This means you that uncached IO will
take advantage of the page cache for uptodate data, but not leave anything
it instantiated/created in cache.
File systems need to support this. This patchset adds support for the
generic read path, which covers file systems like ext4. Patches exist to
add support for iomap/XFS and btrfs as well, which sit on top of this
series. If RWF_DONTCACHE IO is attempted on a file system that doesn't
support it, -EOPNOTSUPP is returned. Hence the user can rely on it either
working as designed, or flagging and error if that's not the case. The
intent here is to give the application a sensible fallback path - eg, it
may fall back to O_DIRECT if appropriate, or just live with the fact that
uncached IO isn't available and do normal buffered IO.
Adding "support" to other file systems should be trivial, most of the time
just a one-liner adding FOP_DONTCACHE to the fop_flags in the
file_operations struct, if the file system is using either iomap or the
generic filemap helpers for reading and writing.
Performance results are in patch 8 for reads, and you can find the write
side results in the XFS patch adding support for DONTCACHE writes for XFS:
https://git.kernel.dk/cgit/linux/commit/?h=buffered-uncached-fs.10&id=257e92de795fdff7d7e256501e024fac6da6a7f4
with the tldr being that I see about a 65% improvement in performance for
both, with fully predictable IO times. CPU reduction is substantial as
well, with no kswapd activity at all for reclaim when using uncached IO.
Using it from applications is trivial - just set RWF_DONTCACHE for the
read or write, using pwritev2(2) or preadv2(2). For io_uring, same thing,
just set RWF_DONTCACHE in sqe->rw_flags for a buffered read/write
operation. And that's it.
Patches 1..7 are just prep patches, and should have no functional changes
at all. Patch 8 adds support for the filemap path for RWF_DONTCACHE
reads, and patches 9..12 are just prep patches for supporting the write
side of uncached writes. In the below mentioned branch, there are then
patches to adopt uncached reads and writes for xfs, btrfs, and ext4. The
latter currently relies on bit of a hack for passing whether this is an
uncached write or not through ->write_begin(), which can hopefully go away
once ext4 adopts iomap for buffered writes. I say this is a hack as it's
not the prettiest way to do it, however it is fully solid and will work
just fine.
Passes full xfstests and fsx overnight runs, no issues observed. That
includes the vm running the testing also using RWF_DONTCACHE on the host.
I'll post fsstress and fsx patches for RWF_DONTCACHE separately. As far
as I'm concerned, no further work needs doing here.
And git tree for the patches is here:
https://git.kernel.dk/cgit/linux/log/?h=buffered-uncached.10
with the file system patches on top adding support for xfs/btrfs/ext4
here:
https://git.kernel.dk/cgit/linux/log/?h=buffered-uncached-fs.10
This patch (of 12):
Rather than pass in both the file and position directly from the kiocb,
just take a struct kiocb instead. With the kiocb being passed in, skip
passing in the address_space separately as well. While doing so, move the
ki_flags checking into filemap_create_folio() as well. In preparation for
actually needing the kiocb in the function.
No functional changes in this patch.
Link: https://lkml.kernel.org/r/20241220154831.1086649-1-axboe@kernel.dk
Link: https://lkml.kernel.org/r/20241220154831.1086649-2-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:39 -07:00
|
|
|
|
2021-03-10 10:34:00 -05:00
|
|
|
|
2021-02-24 12:02:18 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm/filemap: change filemap_create_folio() to take a struct kiocb
Patch series "Uncached buffered IO", v8.
5 years ago I posted patches adding support for RWF_UNCACHED, as a way to
do buffered IO that isn't page cache persistent. The approach back then
was to have private pages for IO, and then get rid of them once IO was
done. But that then runs into all the issues that O_DIRECT has, in terms
of synchronizing with the page cache.
So here's a new approach to the same concent, but using the page cache as
synchronization. Due to excessive bike shedding on the naming, this is
now named RWF_DONTCACHE, and is less special in that it's just page cache
IO, except it prunes the ranges once IO is completed.
Why do this, you may ask? The tldr is that device speeds are only getting
faster, while reclaim is not. Doing normal buffered IO can be very
unpredictable, and suck up a lot of resources on the reclaim side. This
leads people to use O_DIRECT as a work-around, which has its own set of
restrictions in terms of size, offset, and length of IO. It's also
inherently synchronous, and now you need async IO as well. While the
latter isn't necessarily a big problem as we have good options available
there, it also should not be a requirement when all you want to do is read
or write some data without caching.
Even on desktop type systems, a normal NVMe device can fill the entire
page cache in seconds. On the big system I used for testing, there's a
lot more RAM, but also a lot more devices. As can be seen in some of the
results in the following patches, you can still fill RAM in seconds even
when there's 1TB of it. Hence this problem isn't solely a "big
hyperscaler system" issue, it's common across the board.
Common for both reads and writes with RWF_DONTCACHE is that they use the
page cache for IO. Reads work just like a normal buffered read would,
with the only exception being that the touched ranges will get pruned
after data has been copied. For writes, the ranges will get writeback
kicked off before the syscall returns, and then writeback completion will
prune the range. Hence writes aren't synchronous, and it's easy to
pipeline writes using RWF_DONTCACHE. Folios that aren't instantiated by
RWF_DONTCACHE IO are left untouched. This means you that uncached IO will
take advantage of the page cache for uptodate data, but not leave anything
it instantiated/created in cache.
File systems need to support this. This patchset adds support for the
generic read path, which covers file systems like ext4. Patches exist to
add support for iomap/XFS and btrfs as well, which sit on top of this
series. If RWF_DONTCACHE IO is attempted on a file system that doesn't
support it, -EOPNOTSUPP is returned. Hence the user can rely on it either
working as designed, or flagging and error if that's not the case. The
intent here is to give the application a sensible fallback path - eg, it
may fall back to O_DIRECT if appropriate, or just live with the fact that
uncached IO isn't available and do normal buffered IO.
Adding "support" to other file systems should be trivial, most of the time
just a one-liner adding FOP_DONTCACHE to the fop_flags in the
file_operations struct, if the file system is using either iomap or the
generic filemap helpers for reading and writing.
Performance results are in patch 8 for reads, and you can find the write
side results in the XFS patch adding support for DONTCACHE writes for XFS:
https://git.kernel.dk/cgit/linux/commit/?h=buffered-uncached-fs.10&id=257e92de795fdff7d7e256501e024fac6da6a7f4
with the tldr being that I see about a 65% improvement in performance for
both, with fully predictable IO times. CPU reduction is substantial as
well, with no kswapd activity at all for reclaim when using uncached IO.
Using it from applications is trivial - just set RWF_DONTCACHE for the
read or write, using pwritev2(2) or preadv2(2). For io_uring, same thing,
just set RWF_DONTCACHE in sqe->rw_flags for a buffered read/write
operation. And that's it.
Patches 1..7 are just prep patches, and should have no functional changes
at all. Patch 8 adds support for the filemap path for RWF_DONTCACHE
reads, and patches 9..12 are just prep patches for supporting the write
side of uncached writes. In the below mentioned branch, there are then
patches to adopt uncached reads and writes for xfs, btrfs, and ext4. The
latter currently relies on bit of a hack for passing whether this is an
uncached write or not through ->write_begin(), which can hopefully go away
once ext4 adopts iomap for buffered writes. I say this is a hack as it's
not the prettiest way to do it, however it is fully solid and will work
just fine.
Passes full xfstests and fsx overnight runs, no issues observed. That
includes the vm running the testing also using RWF_DONTCACHE on the host.
I'll post fsstress and fsx patches for RWF_DONTCACHE separately. As far
as I'm concerned, no further work needs doing here.
And git tree for the patches is here:
https://git.kernel.dk/cgit/linux/log/?h=buffered-uncached.10
with the file system patches on top adding support for xfs/btrfs/ext4
here:
https://git.kernel.dk/cgit/linux/log/?h=buffered-uncached-fs.10
This patch (of 12):
Rather than pass in both the file and position directly from the kiocb,
just take a struct kiocb instead. With the kiocb being passed in, skip
passing in the address_space separately as well. While doing so, move the
ki_flags checking into filemap_create_folio() as well. In preparation for
actually needing the kiocb in the function.
No functional changes in this patch.
Link: https://lkml.kernel.org/r/20241220154831.1086649-1-axboe@kernel.dk
Link: https://lkml.kernel.org/r/20241220154831.1086649-2-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:39 -07:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:18 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2021-02-24 12:02:18 -08:00
|
|
|
|
|
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2021-03-10 10:34:00 -05:00
|
|
|
|
2021-02-24 12:02:18 -08:00
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:32 -08:00
|
|
|
|
2021-03-10 14:01:22 -05:00
|
|
|
|
2021-02-24 12:02:32 -08:00
|
|
|
|
|
|
|
|
|
2021-03-10 14:01:22 -05:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:32 -08:00
|
|
|
|
|
|
|
|
|
mm/filemap: add read support for RWF_DONTCACHE
Add RWF_DONTCACHE as a read operation flag, which means that any data read
wil be removed from the page cache upon completion. Uses the page cache
to synchronize, and simply prunes folios that were instantiated when the
operation completes. While it would be possible to use private pages for
this, using the page cache as synchronization is handy for a variety of
reasons:
1) No special truncate magic is needed
2) Async buffered reads need some place to serialize, using the page
cache is a lot easier than writing extra code for this
3) The pruning cost is pretty reasonable
and the code to support this is much simpler as a result.
You can think of uncached buffered IO as being the much more attractive
cousin of O_DIRECT - it has none of the restrictions of O_DIRECT. Yes, it
will copy the data, but unlike regular buffered IO, it doesn't run into
the unpredictability of the page cache in terms of reclaim. As an
example, on a test box with 32 drives, reading them with buffered IO looks
as follows:
Reading bs 65536, uncached 0
1s: 145945MB/sec
2s: 158067MB/sec
3s: 157007MB/sec
4s: 148622MB/sec
5s: 118824MB/sec
6s: 70494MB/sec
7s: 41754MB/sec
8s: 90811MB/sec
9s: 92204MB/sec
10s: 95178MB/sec
11s: 95488MB/sec
12s: 95552MB/sec
13s: 96275MB/sec
where it's quite easy to see where the page cache filled up, and
performance went from good to erratic, and finally settles at a much
lower rate. Looking at top while this is ongoing, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7535 root 20 0 267004 0 0 S 3199 0.0 8:40.65 uncached
3326 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd4
3327 root 20 0 0 0 0 R 100.0 0.0 0:17.22 kswapd5
3328 root 20 0 0 0 0 R 100.0 0.0 0:13.29 kswapd6
3332 root 20 0 0 0 0 R 100.0 0.0 0:11.11 kswapd10
3339 root 20 0 0 0 0 R 100.0 0.0 0:16.25 kswapd17
3348 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd26
3343 root 20 0 0 0 0 R 100.0 0.0 0:16.30 kswapd21
3344 root 20 0 0 0 0 R 100.0 0.0 0:11.92 kswapd22
3349 root 20 0 0 0 0 R 100.0 0.0 0:16.28 kswapd27
3352 root 20 0 0 0 0 R 99.7 0.0 0:11.89 kswapd30
3353 root 20 0 0 0 0 R 96.7 0.0 0:16.04 kswapd31
3329 root 20 0 0 0 0 R 96.4 0.0 0:11.41 kswapd7
3345 root 20 0 0 0 0 R 96.4 0.0 0:13.40 kswapd23
3330 root 20 0 0 0 0 S 91.1 0.0 0:08.28 kswapd8
3350 root 20 0 0 0 0 S 86.8 0.0 0:11.13 kswapd28
3325 root 20 0 0 0 0 S 76.3 0.0 0:07.43 kswapd3
3341 root 20 0 0 0 0 S 74.7 0.0 0:08.85 kswapd19
3334 root 20 0 0 0 0 S 71.7 0.0 0:10.04 kswapd12
3351 root 20 0 0 0 0 R 60.5 0.0 0:09.59 kswapd29
3323 root 20 0 0 0 0 R 57.6 0.0 0:11.50 kswapd1
[...]
which is just showing a partial list of the 32 kswapd threads that are
running mostly full tilt, burning ~28 full CPU cores.
If the same test case is run with RWF_DONTCACHE set for the buffered read,
the output looks as follows:
Reading bs 65536, uncached 0
1s: 153144MB/sec
2s: 156760MB/sec
3s: 158110MB/sec
4s: 158009MB/sec
5s: 158043MB/sec
6s: 157638MB/sec
7s: 157999MB/sec
8s: 158024MB/sec
9s: 157764MB/sec
10s: 157477MB/sec
11s: 157417MB/sec
12s: 157455MB/sec
13s: 157233MB/sec
14s: 156692MB/sec
which is just chugging along at ~155GB/sec of read performance. Looking
at top, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7961 root 20 0 267004 0 0 S 3180 0.0 5:37.95 uncached
8024 axboe 20 0 14292 4096 0 R 1.0 0.0 0:00.13 top
where just the test app is using CPU, no reclaim is taking place outside
of the main thread. Not only is performance 65% better, it's also using
half the CPU to do it.
Link: https://lkml.kernel.org/r/20241220154831.1086649-9-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:46 -07:00
|
|
|
|
|
|
|
|
|
2021-03-10 14:01:22 -05:00
|
|
|
|
2021-02-24 12:02:32 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-08 18:18:17 +00:00
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2021-03-10 14:01:22 -05:00
|
|
|
|
mm: allow read-ahead with IOCB_NOWAIT set
Readahead support for IOCB_NOWAIT was introduced in commit 2e85abf053b9
("mm: allow read-ahead with IOCB_NOWAIT set"). However, this
implementation broke the semantics of IOCB_NOWAIT by potentially causing
it to wait on I/O during memory reclamation. This behavior was later
modified in commit efa8480a8316 ("fs: RWF_NOWAIT should imply IOCB_NOIO").
To resolve the blocking issue during memory reclamation, we can use
memalloc_noio_{save,restore} to ensure non-blocking behavior. This change
restores the original functionality, allowing preadv2(IOCB_NOWAIT) to
trigger readahead if the file content is not present in the page cache.
While this process may trigger direct memory reclamation, the
__GFP_NORETRY flag is set in the readahead GFP flags, ensuring it won't
block.
A use case for this change is when we want to trigger readahead in the
preadv2(2) syscall if the file cache is absent, but without waiting for
certain filesystem locks, like xfs_ilock. A simple example is as follows:
retry:
if (preadv2(fd, iovec, cnt, offset, RWF_NOWAIT) < 0) {
do_other_work();
goto retry;
}
Link: https://lore.gnuweeb.org/io-uring/20200624164127.GP21350@casper.infradead.org/
Link: https://lkml.kernel.org/r/20240820022639.89562-1-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Christian Brauner <brauner@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-08-20 10:26:39 +08:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
mm/filemap: align last_index to folio size
On XFS systems with pagesize=4K, blocksize=16K, and
CONFIG_TRANSPARENT_HUGEPAGE enabled, We observed the following readahead
behaviors:
# echo 3 > /proc/sys/vm/drop_caches
# dd if=test of=/dev/null bs=64k count=1
# ./tools/mm/page-types -r -L -f /mnt/xfs/test
foffset offset flags
0 136d4c __RU_l_________H______t_________________F_1
1 136d4d __RU_l__________T_____t_________________F_1
2 136d4e __RU_l__________T_____t_________________F_1
3 136d4f __RU_l__________T_____t_________________F_1
...
c 136bb8 __RU_l_________H______t_________________F_1
d 136bb9 __RU_l__________T_____t_________________F_1
e 136bba __RU_l__________T_____t_________________F_1
f 136bbb __RU_l__________T_____t_________________F_1 <-- first read
10 13c2cc ___U_l_________H______t______________I__F_1 <-- readahead flag
11 13c2cd ___U_l__________T_____t______________I__F_1
12 13c2ce ___U_l__________T_____t______________I__F_1
13 13c2cf ___U_l__________T_____t______________I__F_1
...
1c 1405d4 ___U_l_________H______t_________________F_1
1d 1405d5 ___U_l__________T_____t_________________F_1
1e 1405d6 ___U_l__________T_____t_________________F_1
1f 1405d7 ___U_l__________T_____t_________________F_1
[ra_size = 32, req_count = 16, async_size = 16]
# echo 3 > /proc/sys/vm/drop_caches
# dd if=test of=/dev/null bs=60k count=1
# ./page-types -r -L -f /mnt/xfs/test
foffset offset flags
0 136048 __RU_l_________H______t_________________F_1
...
c 110a40 __RU_l_________H______t_________________F_1
d 110a41 __RU_l__________T_____t_________________F_1
e 110a42 __RU_l__________T_____t_________________F_1 <-- first read
f 110a43 __RU_l__________T_____t_________________F_1 <-- first readahead flag
10 13e7a8 ___U_l_________H______t_________________F_1
...
20 137a00 ___U_l_________H______t_______P______I__F_1 <-- second readahead flag (20 - 2f)
21 137a01 ___U_l__________T_____t_______P______I__F_1
...
3f 10d4af ___U_l__________T_____t_______P_________F_1
[first readahead: ra_size = 32, req_count = 15, async_size = 17]
When reading 64k data (same for 61-63k range, where last_index is
page-aligned in filemap_get_pages()), 128k readahead is triggered via
page_cache_sync_ra() and the PG_readahead flag is set on the next folio
(the one containing 0x10 page).
When reading 60k data, 128k readahead is also triggered via
page_cache_sync_ra(). However, in this case the readahead flag is set on
the 0xf page. Although the requested read size (req_count) is 60k, the
actual read will be aligned to folio size (64k), which triggers the
readahead flag and initiates asynchronous readahead via
page_cache_async_ra(). This results in two readahead operations totaling
256k.
The root cause is that when the requested size is smaller than the actual
read size (due to folio alignment), it triggers asynchronous readahead.
By changing last_index alignment from page size to folio size, we ensure
the requested size matches the actual read size, preventing the case where
a single read operation triggers two readahead operations.
After applying the patch:
# echo 3 > /proc/sys/vm/drop_caches
# dd if=test of=/dev/null bs=60k count=1
# ./page-types -r -L -f /mnt/xfs/test
foffset offset flags
0 136d4c __RU_l_________H______t_________________F_1
1 136d4d __RU_l__________T_____t_________________F_1
2 136d4e __RU_l__________T_____t_________________F_1
3 136d4f __RU_l__________T_____t_________________F_1
...
c 136bb8 __RU_l_________H______t_________________F_1
d 136bb9 __RU_l__________T_____t_________________F_1
e 136bba __RU_l__________T_____t_________________F_1 <-- first read
f 136bbb __RU_l__________T_____t_________________F_1
10 13c2cc ___U_l_________H______t______________I__F_1 <-- readahead flag
11 13c2cd ___U_l__________T_____t______________I__F_1
12 13c2ce ___U_l__________T_____t______________I__F_1
13 13c2cf ___U_l__________T_____t______________I__F_1
...
1c 1405d4 ___U_l_________H______t_________________F_1
1d 1405d5 ___U_l__________T_____t_________________F_1
1e 1405d6 ___U_l__________T_____t_________________F_1
1f 1405d7 ___U_l__________T_____t_________________F_1
[ra_size = 32, req_count = 16, async_size = 16]
The same phenomenon will occur when reading from 49k to 64k. Set the
readahead flag to the next folio.
Because the minimum order of folio in address_space equals the block size
(at least in xfs and bcachefs that already support bs > ps), having
request_count aligned to block size will not cause overread.
[klarasmodin@gmail.com: fix overflow on 32-bit]
Link: https://lkml.kernel.org/r/yru7qf5gvyzccq5ohhpylvxug5lr5tf54omspbjh4sm6pcdb2r@fpjgj2pxw7va
[akpm@linux-foundation.org: update it for Max's constification efforts]
Link: https://lkml.kernel.org/r/20250711055509.91587-1-youling.tang@linux.dev
Co-developed-by: Chi Zhiling <chizhiling@kylinos.cn>
Signed-off-by: Chi Zhiling <chizhiling@kylinos.cn>
Signed-off-by: Youling Tang <tangyouling@kylinos.cn>
Signed-off-by: Klara Modin <klarasmodin@gmail.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Youling Tang <youling.tang@linux.dev>
Cc: David Hildenbrand <david@redhat.com>
Cc: Klara Modin <klarasmodin@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-07-11 13:55:09 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:35 -08:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-08 10:24:00 +08:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2024-12-20 08:47:40 -07:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:35 -08:00
|
|
|
|
|
|
|
|
|
mm: allow read-ahead with IOCB_NOWAIT set
Readahead support for IOCB_NOWAIT was introduced in commit 2e85abf053b9
("mm: allow read-ahead with IOCB_NOWAIT set"). However, this
implementation broke the semantics of IOCB_NOWAIT by potentially causing
it to wait on I/O during memory reclamation. This behavior was later
modified in commit efa8480a8316 ("fs: RWF_NOWAIT should imply IOCB_NOIO").
To resolve the blocking issue during memory reclamation, we can use
memalloc_noio_{save,restore} to ensure non-blocking behavior. This change
restores the original functionality, allowing preadv2(IOCB_NOWAIT) to
trigger readahead if the file content is not present in the page cache.
While this process may trigger direct memory reclamation, the
__GFP_NORETRY flag is set in the readahead GFP flags, ensuring it won't
block.
A use case for this change is when we want to trigger readahead in the
preadv2(2) syscall if the file cache is absent, but without waiting for
certain filesystem locks, like xfs_ilock. A simple example is as follows:
retry:
if (preadv2(fd, iovec, cnt, offset, RWF_NOWAIT) < 0) {
do_other_work();
goto retry;
}
Link: https://lore.gnuweeb.org/io-uring/20200624164127.GP21350@casper.infradead.org/
Link: https://lkml.kernel.org/r/20240820022639.89562-1-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Christian Brauner <brauner@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-08-20 10:26:39 +08:00
|
|
|
|
|
|
|
|
|
mm/filemap: add read support for RWF_DONTCACHE
Add RWF_DONTCACHE as a read operation flag, which means that any data read
wil be removed from the page cache upon completion. Uses the page cache
to synchronize, and simply prunes folios that were instantiated when the
operation completes. While it would be possible to use private pages for
this, using the page cache as synchronization is handy for a variety of
reasons:
1) No special truncate magic is needed
2) Async buffered reads need some place to serialize, using the page
cache is a lot easier than writing extra code for this
3) The pruning cost is pretty reasonable
and the code to support this is much simpler as a result.
You can think of uncached buffered IO as being the much more attractive
cousin of O_DIRECT - it has none of the restrictions of O_DIRECT. Yes, it
will copy the data, but unlike regular buffered IO, it doesn't run into
the unpredictability of the page cache in terms of reclaim. As an
example, on a test box with 32 drives, reading them with buffered IO looks
as follows:
Reading bs 65536, uncached 0
1s: 145945MB/sec
2s: 158067MB/sec
3s: 157007MB/sec
4s: 148622MB/sec
5s: 118824MB/sec
6s: 70494MB/sec
7s: 41754MB/sec
8s: 90811MB/sec
9s: 92204MB/sec
10s: 95178MB/sec
11s: 95488MB/sec
12s: 95552MB/sec
13s: 96275MB/sec
where it's quite easy to see where the page cache filled up, and
performance went from good to erratic, and finally settles at a much
lower rate. Looking at top while this is ongoing, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7535 root 20 0 267004 0 0 S 3199 0.0 8:40.65 uncached
3326 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd4
3327 root 20 0 0 0 0 R 100.0 0.0 0:17.22 kswapd5
3328 root 20 0 0 0 0 R 100.0 0.0 0:13.29 kswapd6
3332 root 20 0 0 0 0 R 100.0 0.0 0:11.11 kswapd10
3339 root 20 0 0 0 0 R 100.0 0.0 0:16.25 kswapd17
3348 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd26
3343 root 20 0 0 0 0 R 100.0 0.0 0:16.30 kswapd21
3344 root 20 0 0 0 0 R 100.0 0.0 0:11.92 kswapd22
3349 root 20 0 0 0 0 R 100.0 0.0 0:16.28 kswapd27
3352 root 20 0 0 0 0 R 99.7 0.0 0:11.89 kswapd30
3353 root 20 0 0 0 0 R 96.7 0.0 0:16.04 kswapd31
3329 root 20 0 0 0 0 R 96.4 0.0 0:11.41 kswapd7
3345 root 20 0 0 0 0 R 96.4 0.0 0:13.40 kswapd23
3330 root 20 0 0 0 0 S 91.1 0.0 0:08.28 kswapd8
3350 root 20 0 0 0 0 S 86.8 0.0 0:11.13 kswapd28
3325 root 20 0 0 0 0 S 76.3 0.0 0:07.43 kswapd3
3341 root 20 0 0 0 0 S 74.7 0.0 0:08.85 kswapd19
3334 root 20 0 0 0 0 S 71.7 0.0 0:10.04 kswapd12
3351 root 20 0 0 0 0 R 60.5 0.0 0:09.59 kswapd29
3323 root 20 0 0 0 0 R 57.6 0.0 0:11.50 kswapd1
[...]
which is just showing a partial list of the 32 kswapd threads that are
running mostly full tilt, burning ~28 full CPU cores.
If the same test case is run with RWF_DONTCACHE set for the buffered read,
the output looks as follows:
Reading bs 65536, uncached 0
1s: 153144MB/sec
2s: 156760MB/sec
3s: 158110MB/sec
4s: 158009MB/sec
5s: 158043MB/sec
6s: 157638MB/sec
7s: 157999MB/sec
8s: 158024MB/sec
9s: 157764MB/sec
10s: 157477MB/sec
11s: 157417MB/sec
12s: 157455MB/sec
13s: 157233MB/sec
14s: 156692MB/sec
which is just chugging along at ~155GB/sec of read performance. Looking
at top, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7961 root 20 0 267004 0 0 S 3180 0.0 5:37.95 uncached
8024 axboe 20 0 14292 4096 0 R 1.0 0.0 0:00.13 top
where just the test app is using CPU, no reclaim is taking place outside
of the main thread. Not only is performance 65% better, it's also using
half the CPU to do it.
Link: https://lkml.kernel.org/r/20241220154831.1086649-9-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:46 -07:00
|
|
|
|
|
|
|
|
|
2024-12-20 08:47:40 -07:00
|
|
|
|
mm: allow read-ahead with IOCB_NOWAIT set
Readahead support for IOCB_NOWAIT was introduced in commit 2e85abf053b9
("mm: allow read-ahead with IOCB_NOWAIT set"). However, this
implementation broke the semantics of IOCB_NOWAIT by potentially causing
it to wait on I/O during memory reclamation. This behavior was later
modified in commit efa8480a8316 ("fs: RWF_NOWAIT should imply IOCB_NOIO").
To resolve the blocking issue during memory reclamation, we can use
memalloc_noio_{save,restore} to ensure non-blocking behavior. This change
restores the original functionality, allowing preadv2(IOCB_NOWAIT) to
trigger readahead if the file content is not present in the page cache.
While this process may trigger direct memory reclamation, the
__GFP_NORETRY flag is set in the readahead GFP flags, ensuring it won't
block.
A use case for this change is when we want to trigger readahead in the
preadv2(2) syscall if the file cache is absent, but without waiting for
certain filesystem locks, like xfs_ilock. A simple example is as follows:
retry:
if (preadv2(fd, iovec, cnt, offset, RWF_NOWAIT) < 0) {
do_other_work();
goto retry;
}
Link: https://lore.gnuweeb.org/io-uring/20200624164127.GP21350@casper.infradead.org/
Link: https://lkml.kernel.org/r/20240820022639.89562-1-laoar.shao@gmail.com
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Christian Brauner <brauner@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-08-20 10:26:39 +08:00
|
|
|
|
|
|
|
|
|
2023-02-08 10:24:00 +08:00
|
|
|
|
2021-02-24 12:02:35 -08:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
mm/filemap: change filemap_create_folio() to take a struct kiocb
Patch series "Uncached buffered IO", v8.
5 years ago I posted patches adding support for RWF_UNCACHED, as a way to
do buffered IO that isn't page cache persistent. The approach back then
was to have private pages for IO, and then get rid of them once IO was
done. But that then runs into all the issues that O_DIRECT has, in terms
of synchronizing with the page cache.
So here's a new approach to the same concent, but using the page cache as
synchronization. Due to excessive bike shedding on the naming, this is
now named RWF_DONTCACHE, and is less special in that it's just page cache
IO, except it prunes the ranges once IO is completed.
Why do this, you may ask? The tldr is that device speeds are only getting
faster, while reclaim is not. Doing normal buffered IO can be very
unpredictable, and suck up a lot of resources on the reclaim side. This
leads people to use O_DIRECT as a work-around, which has its own set of
restrictions in terms of size, offset, and length of IO. It's also
inherently synchronous, and now you need async IO as well. While the
latter isn't necessarily a big problem as we have good options available
there, it also should not be a requirement when all you want to do is read
or write some data without caching.
Even on desktop type systems, a normal NVMe device can fill the entire
page cache in seconds. On the big system I used for testing, there's a
lot more RAM, but also a lot more devices. As can be seen in some of the
results in the following patches, you can still fill RAM in seconds even
when there's 1TB of it. Hence this problem isn't solely a "big
hyperscaler system" issue, it's common across the board.
Common for both reads and writes with RWF_DONTCACHE is that they use the
page cache for IO. Reads work just like a normal buffered read would,
with the only exception being that the touched ranges will get pruned
after data has been copied. For writes, the ranges will get writeback
kicked off before the syscall returns, and then writeback completion will
prune the range. Hence writes aren't synchronous, and it's easy to
pipeline writes using RWF_DONTCACHE. Folios that aren't instantiated by
RWF_DONTCACHE IO are left untouched. This means you that uncached IO will
take advantage of the page cache for uptodate data, but not leave anything
it instantiated/created in cache.
File systems need to support this. This patchset adds support for the
generic read path, which covers file systems like ext4. Patches exist to
add support for iomap/XFS and btrfs as well, which sit on top of this
series. If RWF_DONTCACHE IO is attempted on a file system that doesn't
support it, -EOPNOTSUPP is returned. Hence the user can rely on it either
working as designed, or flagging and error if that's not the case. The
intent here is to give the application a sensible fallback path - eg, it
may fall back to O_DIRECT if appropriate, or just live with the fact that
uncached IO isn't available and do normal buffered IO.
Adding "support" to other file systems should be trivial, most of the time
just a one-liner adding FOP_DONTCACHE to the fop_flags in the
file_operations struct, if the file system is using either iomap or the
generic filemap helpers for reading and writing.
Performance results are in patch 8 for reads, and you can find the write
side results in the XFS patch adding support for DONTCACHE writes for XFS:
https://git.kernel.dk/cgit/linux/commit/?h=buffered-uncached-fs.10&id=257e92de795fdff7d7e256501e024fac6da6a7f4
with the tldr being that I see about a 65% improvement in performance for
both, with fully predictable IO times. CPU reduction is substantial as
well, with no kswapd activity at all for reclaim when using uncached IO.
Using it from applications is trivial - just set RWF_DONTCACHE for the
read or write, using pwritev2(2) or preadv2(2). For io_uring, same thing,
just set RWF_DONTCACHE in sqe->rw_flags for a buffered read/write
operation. And that's it.
Patches 1..7 are just prep patches, and should have no functional changes
at all. Patch 8 adds support for the filemap path for RWF_DONTCACHE
reads, and patches 9..12 are just prep patches for supporting the write
side of uncached writes. In the below mentioned branch, there are then
patches to adopt uncached reads and writes for xfs, btrfs, and ext4. The
latter currently relies on bit of a hack for passing whether this is an
uncached write or not through ->write_begin(), which can hopefully go away
once ext4 adopts iomap for buffered writes. I say this is a hack as it's
not the prettiest way to do it, however it is fully solid and will work
just fine.
Passes full xfstests and fsx overnight runs, no issues observed. That
includes the vm running the testing also using RWF_DONTCACHE on the host.
I'll post fsstress and fsx patches for RWF_DONTCACHE separately. As far
as I'm concerned, no further work needs doing here.
And git tree for the patches is here:
https://git.kernel.dk/cgit/linux/log/?h=buffered-uncached.10
with the file system patches on top adding support for xfs/btrfs/ext4
here:
https://git.kernel.dk/cgit/linux/log/?h=buffered-uncached-fs.10
This patch (of 12):
Rather than pass in both the file and position directly from the kiocb,
just take a struct kiocb instead. With the kiocb being passed in, skip
passing in the address_space separately as well. While doing so, move the
ki_flags checking into filemap_create_folio() as well. In preparation for
actually needing the kiocb in the function.
No functional changes in this patch.
Link: https://lkml.kernel.org/r/20241220154831.1086649-1-axboe@kernel.dk
Link: https://lkml.kernel.org/r/20241220154831.1086649-2-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Reviewed-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:39 -07:00
|
|
|
|
2021-02-24 12:02:18 -08:00
|
|
|
|
2021-02-24 12:02:35 -08:00
|
|
|
|
2021-02-24 12:02:18 -08:00
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2021-03-10 14:01:22 -05:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:35 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-10 14:01:22 -05:00
|
|
|
|
2025-07-28 16:39:52 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-08 18:18:17 +00:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:35 -08:00
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
|
|
|
|
|
2024-09-03 10:21:00 +00:00
|
|
|
|
2021-02-24 12:02:35 -08:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2021-02-24 12:02:35 -08:00
|
|
|
|
2021-03-10 14:01:22 -05:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2021-02-24 12:01:55 -08:00
|
|
|
|
2021-02-24 12:02:22 -08:00
|
|
|
|
2021-02-24 12:02:35 -08:00
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
|
|
|
|
|
2022-06-10 14:44:41 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-05-27 07:28:53 -06:00
|
|
|
|
mm/filemap: add read support for RWF_DONTCACHE
Add RWF_DONTCACHE as a read operation flag, which means that any data read
wil be removed from the page cache upon completion. Uses the page cache
to synchronize, and simply prunes folios that were instantiated when the
operation completes. While it would be possible to use private pages for
this, using the page cache as synchronization is handy for a variety of
reasons:
1) No special truncate magic is needed
2) Async buffered reads need some place to serialize, using the page
cache is a lot easier than writing extra code for this
3) The pruning cost is pretty reasonable
and the code to support this is much simpler as a result.
You can think of uncached buffered IO as being the much more attractive
cousin of O_DIRECT - it has none of the restrictions of O_DIRECT. Yes, it
will copy the data, but unlike regular buffered IO, it doesn't run into
the unpredictability of the page cache in terms of reclaim. As an
example, on a test box with 32 drives, reading them with buffered IO looks
as follows:
Reading bs 65536, uncached 0
1s: 145945MB/sec
2s: 158067MB/sec
3s: 157007MB/sec
4s: 148622MB/sec
5s: 118824MB/sec
6s: 70494MB/sec
7s: 41754MB/sec
8s: 90811MB/sec
9s: 92204MB/sec
10s: 95178MB/sec
11s: 95488MB/sec
12s: 95552MB/sec
13s: 96275MB/sec
where it's quite easy to see where the page cache filled up, and
performance went from good to erratic, and finally settles at a much
lower rate. Looking at top while this is ongoing, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7535 root 20 0 267004 0 0 S 3199 0.0 8:40.65 uncached
3326 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd4
3327 root 20 0 0 0 0 R 100.0 0.0 0:17.22 kswapd5
3328 root 20 0 0 0 0 R 100.0 0.0 0:13.29 kswapd6
3332 root 20 0 0 0 0 R 100.0 0.0 0:11.11 kswapd10
3339 root 20 0 0 0 0 R 100.0 0.0 0:16.25 kswapd17
3348 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd26
3343 root 20 0 0 0 0 R 100.0 0.0 0:16.30 kswapd21
3344 root 20 0 0 0 0 R 100.0 0.0 0:11.92 kswapd22
3349 root 20 0 0 0 0 R 100.0 0.0 0:16.28 kswapd27
3352 root 20 0 0 0 0 R 99.7 0.0 0:11.89 kswapd30
3353 root 20 0 0 0 0 R 96.7 0.0 0:16.04 kswapd31
3329 root 20 0 0 0 0 R 96.4 0.0 0:11.41 kswapd7
3345 root 20 0 0 0 0 R 96.4 0.0 0:13.40 kswapd23
3330 root 20 0 0 0 0 S 91.1 0.0 0:08.28 kswapd8
3350 root 20 0 0 0 0 S 86.8 0.0 0:11.13 kswapd28
3325 root 20 0 0 0 0 S 76.3 0.0 0:07.43 kswapd3
3341 root 20 0 0 0 0 S 74.7 0.0 0:08.85 kswapd19
3334 root 20 0 0 0 0 S 71.7 0.0 0:10.04 kswapd12
3351 root 20 0 0 0 0 R 60.5 0.0 0:09.59 kswapd29
3323 root 20 0 0 0 0 R 57.6 0.0 0:11.50 kswapd1
[...]
which is just showing a partial list of the 32 kswapd threads that are
running mostly full tilt, burning ~28 full CPU cores.
If the same test case is run with RWF_DONTCACHE set for the buffered read,
the output looks as follows:
Reading bs 65536, uncached 0
1s: 153144MB/sec
2s: 156760MB/sec
3s: 158110MB/sec
4s: 158009MB/sec
5s: 158043MB/sec
6s: 157638MB/sec
7s: 157999MB/sec
8s: 158024MB/sec
9s: 157764MB/sec
10s: 157477MB/sec
11s: 157417MB/sec
12s: 157455MB/sec
13s: 157233MB/sec
14s: 156692MB/sec
which is just chugging along at ~155GB/sec of read performance. Looking
at top, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7961 root 20 0 267004 0 0 S 3180 0.0 5:37.95 uncached
8024 axboe 20 0 14292 4096 0 R 1.0 0.0 0:00.13 top
where just the test app is using CPU, no reclaim is taking place outside
of the main thread. Not only is performance 65% better, it's also using
half the CPU to do it.
Link: https://lkml.kernel.org/r/20241220154831.1086649-9-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:46 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-05-27 07:28:56 -06:00
|
|
|
|
mm/filemap: add read support for RWF_DONTCACHE
Add RWF_DONTCACHE as a read operation flag, which means that any data read
wil be removed from the page cache upon completion. Uses the page cache
to synchronize, and simply prunes folios that were instantiated when the
operation completes. While it would be possible to use private pages for
this, using the page cache as synchronization is handy for a variety of
reasons:
1) No special truncate magic is needed
2) Async buffered reads need some place to serialize, using the page
cache is a lot easier than writing extra code for this
3) The pruning cost is pretty reasonable
and the code to support this is much simpler as a result.
You can think of uncached buffered IO as being the much more attractive
cousin of O_DIRECT - it has none of the restrictions of O_DIRECT. Yes, it
will copy the data, but unlike regular buffered IO, it doesn't run into
the unpredictability of the page cache in terms of reclaim. As an
example, on a test box with 32 drives, reading them with buffered IO looks
as follows:
Reading bs 65536, uncached 0
1s: 145945MB/sec
2s: 158067MB/sec
3s: 157007MB/sec
4s: 148622MB/sec
5s: 118824MB/sec
6s: 70494MB/sec
7s: 41754MB/sec
8s: 90811MB/sec
9s: 92204MB/sec
10s: 95178MB/sec
11s: 95488MB/sec
12s: 95552MB/sec
13s: 96275MB/sec
where it's quite easy to see where the page cache filled up, and
performance went from good to erratic, and finally settles at a much
lower rate. Looking at top while this is ongoing, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7535 root 20 0 267004 0 0 S 3199 0.0 8:40.65 uncached
3326 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd4
3327 root 20 0 0 0 0 R 100.0 0.0 0:17.22 kswapd5
3328 root 20 0 0 0 0 R 100.0 0.0 0:13.29 kswapd6
3332 root 20 0 0 0 0 R 100.0 0.0 0:11.11 kswapd10
3339 root 20 0 0 0 0 R 100.0 0.0 0:16.25 kswapd17
3348 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd26
3343 root 20 0 0 0 0 R 100.0 0.0 0:16.30 kswapd21
3344 root 20 0 0 0 0 R 100.0 0.0 0:11.92 kswapd22
3349 root 20 0 0 0 0 R 100.0 0.0 0:16.28 kswapd27
3352 root 20 0 0 0 0 R 99.7 0.0 0:11.89 kswapd30
3353 root 20 0 0 0 0 R 96.7 0.0 0:16.04 kswapd31
3329 root 20 0 0 0 0 R 96.4 0.0 0:11.41 kswapd7
3345 root 20 0 0 0 0 R 96.4 0.0 0:13.40 kswapd23
3330 root 20 0 0 0 0 S 91.1 0.0 0:08.28 kswapd8
3350 root 20 0 0 0 0 S 86.8 0.0 0:11.13 kswapd28
3325 root 20 0 0 0 0 S 76.3 0.0 0:07.43 kswapd3
3341 root 20 0 0 0 0 S 74.7 0.0 0:08.85 kswapd19
3334 root 20 0 0 0 0 S 71.7 0.0 0:10.04 kswapd12
3351 root 20 0 0 0 0 R 60.5 0.0 0:09.59 kswapd29
3323 root 20 0 0 0 0 R 57.6 0.0 0:11.50 kswapd1
[...]
which is just showing a partial list of the 32 kswapd threads that are
running mostly full tilt, burning ~28 full CPU cores.
If the same test case is run with RWF_DONTCACHE set for the buffered read,
the output looks as follows:
Reading bs 65536, uncached 0
1s: 153144MB/sec
2s: 156760MB/sec
3s: 158110MB/sec
4s: 158009MB/sec
5s: 158043MB/sec
6s: 157638MB/sec
7s: 157999MB/sec
8s: 158024MB/sec
9s: 157764MB/sec
10s: 157477MB/sec
11s: 157417MB/sec
12s: 157455MB/sec
13s: 157233MB/sec
14s: 156692MB/sec
which is just chugging along at ~155GB/sec of read performance. Looking
at top, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7961 root 20 0 267004 0 0 S 3180 0.0 5:37.95 uncached
8024 axboe 20 0 14292 4096 0 R 1.0 0.0 0:00.13 top
where just the test app is using CPU, no reclaim is taking place outside
of the main thread. Not only is performance 65% better, it's also using
half the CPU to do it.
Link: https://lkml.kernel.org/r/20241220154831.1086649-9-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:46 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2021-02-24 12:02:42 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2021-02-24 12:02:42 -08:00
|
|
|
|
2022-04-29 11:53:28 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-02-24 12:02:42 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-02-24 12:02:42 -08:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2017-08-29 16:13:18 +02:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
2008-02-08 04:21:24 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2021-02-24 12:01:55 -08:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
|
|
|
|
|
mm/filemap.c: fix update prev_pos after one read request done
ra->prev_pos tracks the last visited byte in the previous read request.
It is used to check whether it is sequential read in ondemand_readahead
and thus affects the readahead window.
After commit 06c0444290ce ("mm/filemap.c: generic_file_buffered_read() now
uses find_get_pages_contig"), update logic of prev_pos is changed. It
updates prev_pos after each return from filemap_get_pages(). But the read
request from user may be not fully completed at this point. The updated
prev_pos impacts the subsequent readahead window.
The real problem is performance drop of fsck_msdos between linux-5.4 and
linux-5.15(also linux-6.4). Comparing to linux-5.4,It spends about 110%
time and read 140% pages. The read pattern of fsck_msdos is not fully
sequential.
Simplified read pattern of fsck_msdos likes below:
1.read at page offset 0xa,size 0x1000
2.read at other page offset like 0x20,size 0x1000
3.read at page offset 0xa,size 0x4000
4.read at page offset 0xe,size 0x1000
Here is the read status on linux-6.4:
1.after read at page offset 0xa,size 0x1000
->page ofs 0xa go into pagecache
2.after read at page offset 0x20,size 0x1000
->page ofs 0x20 go into pagecache
3.read at page offset 0xa,size 0x4000
->filemap_get_pages read ofs 0xa from pagecache and returns
->prev_pos is updated to 0xb and goto next loop
->filemap_get_pages tends to read ofs 0xb,size 0x3000
->initial_readahead case in ondemand_readahead since prev_pos is
the same as request ofs.
->read 8 pages while async size is 5 pages
(PageReadahead flag at page 0xe)
4.read at page offset 0xe,size 0x1000
->hit page 0xe with PageReadahead flag set,double the ra_size.
read 16 pages while async size is 16 pages
Now it reads 24 pages while actually uses 5 pages
on linux-5.4:
1.the same as 6.4
2.the same as 6.4
3.read at page offset 0xa,size 0x4000
->read ofs 0xa from pagecache
->read ofs 0xb,size 0x3000 using page_cache_sync_readahead
read 3 pages
->prev_pos is updated to 0xd before generic_file_buffered_read
returns
4.read at page offset 0xe,size 0x1000
->initial_readahead case in ondemand_readahead since
request ofs-prev_pos==1
->read 4 pages while async size is 3 pages
Now it reads 7 pages while actually uses 5 pages.
In above demo, the initial_readahead case is triggered by offset of user
request on linux-5.4. While it may be triggered by update logic of
prev_pos on linux-6.4.
To fix the performance drop, update prev_pos after finishing one read
request.
Link: https://lkml.kernel.org/r/20230628110220.120134-1-haibo.li@mediatek.com
Signed-off-by: Haibo Li <haibo.li@mediatek.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-28 19:02:20 +08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2024-09-13 16:04:18 -04:00
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
2016-12-14 12:45:25 -08:00
|
|
|
|
2020-12-18 04:07:11 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-09-13 13:57:04 -04:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
vfs,mm: fix a dead loop in truncate_inode_pages_range()
We triggered a deadloop in truncate_inode_pages_range() on 32 bits
architecture with the test case bellow:
...
fd = open();
write(fd, buf, 4096);
preadv64(fd, &iovec, 1, 0xffffffff000);
ftruncate(fd, 0);
...
Then ftruncate() will not return forever.
The filesystem used in this case is ubifs, but it can be triggered on
many other filesystems.
When preadv64() is called with offset=0xffffffff000, a page with
index=0xffffffff will be added to the radix tree of ->mapping. Then
this page can be found in ->mapping with pagevec_lookup(). After that,
truncate_inode_pages_range(), which is called in ftruncate(), will fall
into an infinite loop:
- find a page with index=0xffffffff, since index>=end, this page won't
be truncated
- index++, and index become 0
- the page with index=0xffffffff will be found again
The data type of index is unsigned long, so index won't overflow to 0 on
64 bits architecture in this case, and the dead loop won't happen.
Since truncate_inode_pages_range() is executed with holding lock of
inode->i_rwsem, any operation related with this lock will be blocked,
and a hung task will happen, e.g.:
INFO: task truncate_test:3364 blocked for more than 120 seconds.
...
call_rwsem_down_write_failed+0x17/0x30
generic_file_write_iter+0x32/0x1c0
ubifs_write_iter+0xcc/0x170
__vfs_write+0xc4/0x120
vfs_write+0xb2/0x1b0
SyS_write+0x46/0xa0
The page with index=0xffffffff added to ->mapping is useless. Fix this
by checking the read position before allocating pages.
Link: http://lkml.kernel.org/r/1475151010-40166-1-git-send-email-fangwei1@huawei.com
Signed-off-by: Wei Fang <fangwei1@huawei.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-10-07 17:01:52 -07:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2017-02-03 13:13:29 -08:00
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
2021-02-24 12:02:42 -08:00
|
|
|
|
2020-12-14 19:04:52 -08:00
|
|
|
|
|
|
|
|
|
2021-11-05 13:36:49 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-22 14:50:17 +01:00
|
|
|
|
2021-02-24 12:01:55 -08:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-06-10 14:44:41 -04:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
|
|
|
|
|
mm/filemap.c: fix update prev_pos after one read request done
ra->prev_pos tracks the last visited byte in the previous read request.
It is used to check whether it is sequential read in ondemand_readahead
and thus affects the readahead window.
After commit 06c0444290ce ("mm/filemap.c: generic_file_buffered_read() now
uses find_get_pages_contig"), update logic of prev_pos is changed. It
updates prev_pos after each return from filemap_get_pages(). But the read
request from user may be not fully completed at this point. The updated
prev_pos impacts the subsequent readahead window.
The real problem is performance drop of fsck_msdos between linux-5.4 and
linux-5.15(also linux-6.4). Comparing to linux-5.4,It spends about 110%
time and read 140% pages. The read pattern of fsck_msdos is not fully
sequential.
Simplified read pattern of fsck_msdos likes below:
1.read at page offset 0xa,size 0x1000
2.read at other page offset like 0x20,size 0x1000
3.read at page offset 0xa,size 0x4000
4.read at page offset 0xe,size 0x1000
Here is the read status on linux-6.4:
1.after read at page offset 0xa,size 0x1000
->page ofs 0xa go into pagecache
2.after read at page offset 0x20,size 0x1000
->page ofs 0x20 go into pagecache
3.read at page offset 0xa,size 0x4000
->filemap_get_pages read ofs 0xa from pagecache and returns
->prev_pos is updated to 0xb and goto next loop
->filemap_get_pages tends to read ofs 0xb,size 0x3000
->initial_readahead case in ondemand_readahead since prev_pos is
the same as request ofs.
->read 8 pages while async size is 5 pages
(PageReadahead flag at page 0xe)
4.read at page offset 0xe,size 0x1000
->hit page 0xe with PageReadahead flag set,double the ra_size.
read 16 pages while async size is 16 pages
Now it reads 24 pages while actually uses 5 pages
on linux-5.4:
1.the same as 6.4
2.the same as 6.4
3.read at page offset 0xa,size 0x4000
->read ofs 0xa from pagecache
->read ofs 0xb,size 0x3000 using page_cache_sync_readahead
read 3 pages
->prev_pos is updated to 0xd before generic_file_buffered_read
returns
4.read at page offset 0xe,size 0x1000
->initial_readahead case in ondemand_readahead since
request ofs-prev_pos==1
->read 4 pages while async size is 3 pages
Now it reads 7 pages while actually uses 5 pages.
In above demo, the initial_readahead case is triggered by offset of user
request on linux-5.4. While it may be triggered by update logic of
prev_pos on linux-6.4.
To fix the performance drop, update prev_pos after finishing one read
request.
Link: https://lkml.kernel.org/r/20230628110220.120134-1-haibo.li@mediatek.com
Signed-off-by: Haibo Li <haibo.li@mediatek.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-28 19:02:20 +08:00
|
|
|
|
|
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
|
|
|
|
|
2021-10-31 22:22:19 -04:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2021-10-31 22:22:19 -04:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
2021-10-31 22:22:19 -04:00
|
|
|
|
2021-02-24 12:01:59 -08:00
|
|
|
|
|
|
|
|
|
2021-10-31 22:22:19 -04:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
2021-10-31 22:22:19 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
2021-10-31 22:22:19 -04:00
|
|
|
|
|
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
2021-10-31 22:22:19 -04:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
2021-02-24 12:02:42 -08:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
mm/filemap.c: fix update prev_pos after one read request done
ra->prev_pos tracks the last visited byte in the previous read request.
It is used to check whether it is sequential read in ondemand_readahead
and thus affects the readahead window.
After commit 06c0444290ce ("mm/filemap.c: generic_file_buffered_read() now
uses find_get_pages_contig"), update logic of prev_pos is changed. It
updates prev_pos after each return from filemap_get_pages(). But the read
request from user may be not fully completed at this point. The updated
prev_pos impacts the subsequent readahead window.
The real problem is performance drop of fsck_msdos between linux-5.4 and
linux-5.15(also linux-6.4). Comparing to linux-5.4,It spends about 110%
time and read 140% pages. The read pattern of fsck_msdos is not fully
sequential.
Simplified read pattern of fsck_msdos likes below:
1.read at page offset 0xa,size 0x1000
2.read at other page offset like 0x20,size 0x1000
3.read at page offset 0xa,size 0x4000
4.read at page offset 0xe,size 0x1000
Here is the read status on linux-6.4:
1.after read at page offset 0xa,size 0x1000
->page ofs 0xa go into pagecache
2.after read at page offset 0x20,size 0x1000
->page ofs 0x20 go into pagecache
3.read at page offset 0xa,size 0x4000
->filemap_get_pages read ofs 0xa from pagecache and returns
->prev_pos is updated to 0xb and goto next loop
->filemap_get_pages tends to read ofs 0xb,size 0x3000
->initial_readahead case in ondemand_readahead since prev_pos is
the same as request ofs.
->read 8 pages while async size is 5 pages
(PageReadahead flag at page 0xe)
4.read at page offset 0xe,size 0x1000
->hit page 0xe with PageReadahead flag set,double the ra_size.
read 16 pages while async size is 16 pages
Now it reads 24 pages while actually uses 5 pages
on linux-5.4:
1.the same as 6.4
2.the same as 6.4
3.read at page offset 0xa,size 0x4000
->read ofs 0xa from pagecache
->read ofs 0xb,size 0x3000 using page_cache_sync_readahead
read 3 pages
->prev_pos is updated to 0xd before generic_file_buffered_read
returns
4.read at page offset 0xe,size 0x1000
->initial_readahead case in ondemand_readahead since
request ofs-prev_pos==1
->read 4 pages while async size is 3 pages
Now it reads 7 pages while actually uses 5 pages.
In above demo, the initial_readahead case is triggered by offset of user
request on linux-5.4. While it may be triggered by update logic of
prev_pos on linux-6.4.
To fix the performance drop, update prev_pos after finishing one read
request.
Link: https://lkml.kernel.org/r/20230628110220.120134-1-haibo.li@mediatek.com
Signed-off-by: Haibo Li <haibo.li@mediatek.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-28 19:02:20 +08:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
mm/filemap: add read support for RWF_DONTCACHE
Add RWF_DONTCACHE as a read operation flag, which means that any data read
wil be removed from the page cache upon completion. Uses the page cache
to synchronize, and simply prunes folios that were instantiated when the
operation completes. While it would be possible to use private pages for
this, using the page cache as synchronization is handy for a variety of
reasons:
1) No special truncate magic is needed
2) Async buffered reads need some place to serialize, using the page
cache is a lot easier than writing extra code for this
3) The pruning cost is pretty reasonable
and the code to support this is much simpler as a result.
You can think of uncached buffered IO as being the much more attractive
cousin of O_DIRECT - it has none of the restrictions of O_DIRECT. Yes, it
will copy the data, but unlike regular buffered IO, it doesn't run into
the unpredictability of the page cache in terms of reclaim. As an
example, on a test box with 32 drives, reading them with buffered IO looks
as follows:
Reading bs 65536, uncached 0
1s: 145945MB/sec
2s: 158067MB/sec
3s: 157007MB/sec
4s: 148622MB/sec
5s: 118824MB/sec
6s: 70494MB/sec
7s: 41754MB/sec
8s: 90811MB/sec
9s: 92204MB/sec
10s: 95178MB/sec
11s: 95488MB/sec
12s: 95552MB/sec
13s: 96275MB/sec
where it's quite easy to see where the page cache filled up, and
performance went from good to erratic, and finally settles at a much
lower rate. Looking at top while this is ongoing, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7535 root 20 0 267004 0 0 S 3199 0.0 8:40.65 uncached
3326 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd4
3327 root 20 0 0 0 0 R 100.0 0.0 0:17.22 kswapd5
3328 root 20 0 0 0 0 R 100.0 0.0 0:13.29 kswapd6
3332 root 20 0 0 0 0 R 100.0 0.0 0:11.11 kswapd10
3339 root 20 0 0 0 0 R 100.0 0.0 0:16.25 kswapd17
3348 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd26
3343 root 20 0 0 0 0 R 100.0 0.0 0:16.30 kswapd21
3344 root 20 0 0 0 0 R 100.0 0.0 0:11.92 kswapd22
3349 root 20 0 0 0 0 R 100.0 0.0 0:16.28 kswapd27
3352 root 20 0 0 0 0 R 99.7 0.0 0:11.89 kswapd30
3353 root 20 0 0 0 0 R 96.7 0.0 0:16.04 kswapd31
3329 root 20 0 0 0 0 R 96.4 0.0 0:11.41 kswapd7
3345 root 20 0 0 0 0 R 96.4 0.0 0:13.40 kswapd23
3330 root 20 0 0 0 0 S 91.1 0.0 0:08.28 kswapd8
3350 root 20 0 0 0 0 S 86.8 0.0 0:11.13 kswapd28
3325 root 20 0 0 0 0 S 76.3 0.0 0:07.43 kswapd3
3341 root 20 0 0 0 0 S 74.7 0.0 0:08.85 kswapd19
3334 root 20 0 0 0 0 S 71.7 0.0 0:10.04 kswapd12
3351 root 20 0 0 0 0 R 60.5 0.0 0:09.59 kswapd29
3323 root 20 0 0 0 0 R 57.6 0.0 0:11.50 kswapd1
[...]
which is just showing a partial list of the 32 kswapd threads that are
running mostly full tilt, burning ~28 full CPU cores.
If the same test case is run with RWF_DONTCACHE set for the buffered read,
the output looks as follows:
Reading bs 65536, uncached 0
1s: 153144MB/sec
2s: 156760MB/sec
3s: 158110MB/sec
4s: 158009MB/sec
5s: 158043MB/sec
6s: 157638MB/sec
7s: 157999MB/sec
8s: 158024MB/sec
9s: 157764MB/sec
10s: 157477MB/sec
11s: 157417MB/sec
12s: 157455MB/sec
13s: 157233MB/sec
14s: 156692MB/sec
which is just chugging along at ~155GB/sec of read performance. Looking
at top, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7961 root 20 0 267004 0 0 S 3180 0.0 5:37.95 uncached
8024 axboe 20 0 14292 4096 0 R 1.0 0.0 0:00.13 top
where just the test app is using CPU, no reclaim is taking place outside
of the main thread. Not only is performance 65% better, it's also using
half the CPU to do it.
Link: https://lkml.kernel.org/r/20241220154831.1086649-9-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:46 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-05-27 07:28:53 -06:00
|
|
|
|
mm/filemap: add read support for RWF_DONTCACHE
Add RWF_DONTCACHE as a read operation flag, which means that any data read
wil be removed from the page cache upon completion. Uses the page cache
to synchronize, and simply prunes folios that were instantiated when the
operation completes. While it would be possible to use private pages for
this, using the page cache as synchronization is handy for a variety of
reasons:
1) No special truncate magic is needed
2) Async buffered reads need some place to serialize, using the page
cache is a lot easier than writing extra code for this
3) The pruning cost is pretty reasonable
and the code to support this is much simpler as a result.
You can think of uncached buffered IO as being the much more attractive
cousin of O_DIRECT - it has none of the restrictions of O_DIRECT. Yes, it
will copy the data, but unlike regular buffered IO, it doesn't run into
the unpredictability of the page cache in terms of reclaim. As an
example, on a test box with 32 drives, reading them with buffered IO looks
as follows:
Reading bs 65536, uncached 0
1s: 145945MB/sec
2s: 158067MB/sec
3s: 157007MB/sec
4s: 148622MB/sec
5s: 118824MB/sec
6s: 70494MB/sec
7s: 41754MB/sec
8s: 90811MB/sec
9s: 92204MB/sec
10s: 95178MB/sec
11s: 95488MB/sec
12s: 95552MB/sec
13s: 96275MB/sec
where it's quite easy to see where the page cache filled up, and
performance went from good to erratic, and finally settles at a much
lower rate. Looking at top while this is ongoing, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7535 root 20 0 267004 0 0 S 3199 0.0 8:40.65 uncached
3326 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd4
3327 root 20 0 0 0 0 R 100.0 0.0 0:17.22 kswapd5
3328 root 20 0 0 0 0 R 100.0 0.0 0:13.29 kswapd6
3332 root 20 0 0 0 0 R 100.0 0.0 0:11.11 kswapd10
3339 root 20 0 0 0 0 R 100.0 0.0 0:16.25 kswapd17
3348 root 20 0 0 0 0 R 100.0 0.0 0:16.40 kswapd26
3343 root 20 0 0 0 0 R 100.0 0.0 0:16.30 kswapd21
3344 root 20 0 0 0 0 R 100.0 0.0 0:11.92 kswapd22
3349 root 20 0 0 0 0 R 100.0 0.0 0:16.28 kswapd27
3352 root 20 0 0 0 0 R 99.7 0.0 0:11.89 kswapd30
3353 root 20 0 0 0 0 R 96.7 0.0 0:16.04 kswapd31
3329 root 20 0 0 0 0 R 96.4 0.0 0:11.41 kswapd7
3345 root 20 0 0 0 0 R 96.4 0.0 0:13.40 kswapd23
3330 root 20 0 0 0 0 S 91.1 0.0 0:08.28 kswapd8
3350 root 20 0 0 0 0 S 86.8 0.0 0:11.13 kswapd28
3325 root 20 0 0 0 0 S 76.3 0.0 0:07.43 kswapd3
3341 root 20 0 0 0 0 S 74.7 0.0 0:08.85 kswapd19
3334 root 20 0 0 0 0 S 71.7 0.0 0:10.04 kswapd12
3351 root 20 0 0 0 0 R 60.5 0.0 0:09.59 kswapd29
3323 root 20 0 0 0 0 R 57.6 0.0 0:11.50 kswapd1
[...]
which is just showing a partial list of the 32 kswapd threads that are
running mostly full tilt, burning ~28 full CPU cores.
If the same test case is run with RWF_DONTCACHE set for the buffered read,
the output looks as follows:
Reading bs 65536, uncached 0
1s: 153144MB/sec
2s: 156760MB/sec
3s: 158110MB/sec
4s: 158009MB/sec
5s: 158043MB/sec
6s: 157638MB/sec
7s: 157999MB/sec
8s: 158024MB/sec
9s: 157764MB/sec
10s: 157477MB/sec
11s: 157417MB/sec
12s: 157455MB/sec
13s: 157233MB/sec
14s: 156692MB/sec
which is just chugging along at ~155GB/sec of read performance. Looking
at top, we see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
7961 root 20 0 267004 0 0 S 3180 0.0 5:37.95 uncached
8024 axboe 20 0 14292 4096 0 R 1.0 0.0 0:00.13 top
where just the test app is using CPU, no reclaim is taking place outside
of the main thread. Not only is performance 65% better, it's also using
half the CPU to do it.
Link: https://lkml.kernel.org/r/20241220154831.1086649-9-axboe@kernel.dk
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Chris Mason <clm@meta.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-12-20 08:47:46 -07:00
|
|
|
|
|
|
|
|
|
2021-12-06 15:25:33 -05:00
|
|
|
|
2020-12-14 19:04:56 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2008-10-15 22:01:13 -07:00
|
|
|
|
mm/filemap.c: fix update prev_pos after one read request done
ra->prev_pos tracks the last visited byte in the previous read request.
It is used to check whether it is sequential read in ondemand_readahead
and thus affects the readahead window.
After commit 06c0444290ce ("mm/filemap.c: generic_file_buffered_read() now
uses find_get_pages_contig"), update logic of prev_pos is changed. It
updates prev_pos after each return from filemap_get_pages(). But the read
request from user may be not fully completed at this point. The updated
prev_pos impacts the subsequent readahead window.
The real problem is performance drop of fsck_msdos between linux-5.4 and
linux-5.15(also linux-6.4). Comparing to linux-5.4,It spends about 110%
time and read 140% pages. The read pattern of fsck_msdos is not fully
sequential.
Simplified read pattern of fsck_msdos likes below:
1.read at page offset 0xa,size 0x1000
2.read at other page offset like 0x20,size 0x1000
3.read at page offset 0xa,size 0x4000
4.read at page offset 0xe,size 0x1000
Here is the read status on linux-6.4:
1.after read at page offset 0xa,size 0x1000
->page ofs 0xa go into pagecache
2.after read at page offset 0x20,size 0x1000
->page ofs 0x20 go into pagecache
3.read at page offset 0xa,size 0x4000
->filemap_get_pages read ofs 0xa from pagecache and returns
->prev_pos is updated to 0xb and goto next loop
->filemap_get_pages tends to read ofs 0xb,size 0x3000
->initial_readahead case in ondemand_readahead since prev_pos is
the same as request ofs.
->read 8 pages while async size is 5 pages
(PageReadahead flag at page 0xe)
4.read at page offset 0xe,size 0x1000
->hit page 0xe with PageReadahead flag set,double the ra_size.
read 16 pages while async size is 16 pages
Now it reads 24 pages while actually uses 5 pages
on linux-5.4:
1.the same as 6.4
2.the same as 6.4
3.read at page offset 0xa,size 0x4000
->read ofs 0xa from pagecache
->read ofs 0xb,size 0x3000 using page_cache_sync_readahead
read 3 pages
->prev_pos is updated to 0xd before generic_file_buffered_read
returns
4.read at page offset 0xe,size 0x1000
->initial_readahead case in ondemand_readahead since
request ofs-prev_pos==1
->read 4 pages while async size is 3 pages
Now it reads 7 pages while actually uses 5 pages.
In above demo, the initial_readahead case is triggered by offset of user
request on linux-5.4. While it may be triggered by update logic of
prev_pos on linux-6.4.
To fix the performance drop, update prev_pos after finishing one read
request.
Link: https://lkml.kernel.org/r/20230628110220.120134-1-haibo.li@mediatek.com
Signed-off-by: Haibo Li <haibo.li@mediatek.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-28 19:02:20 +08:00
|
|
|
|
2021-02-24 12:02:42 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-02-24 12:02:42 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-06-01 16:58:56 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
netfs: Implement unbuffered/DIO read support
Implement support for unbuffered and DIO reads in the netfs library,
utilising the existing read helper code to do block splitting and
individual queuing. The code also handles extraction of the destination
buffer from the supplied iterator, allowing async unbuffered reads to take
place.
The read will be split up according to the rsize setting and, if supplied,
the ->clamp_length() method. Note that the next subrequest will be issued
as soon as issue_op returns, without waiting for previous ones to finish.
The network filesystem needs to pause or handle queuing them if it doesn't
want to fire them all at the server simultaneously.
Once all the subrequests have finished, the state will be assessed and the
amount of data to be indicated as having being obtained will be
determined. As the subrequests may finish in any order, if an intermediate
subrequest is short, any further subrequests may be copied into the buffer
and then abandoned.
In the future, this will also take care of doing an unbuffered read from
encrypted content, with the decryption being done by the library.
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Jeff Layton <jlayton@kernel.org>
cc: linux-cachefs@redhat.com
cc: linux-fsdevel@vger.kernel.org
cc: linux-mm@kvack.org
2022-01-14 17:39:55 +00:00
|
|
|
|
2023-06-01 16:58:56 +02:00
|
|
|
|
2024-09-11 17:34:39 +01:00
|
|
|
|
|
|
|
|
|
2023-06-01 16:58:57 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-09-11 17:34:39 +01:00
|
|
|
|
2023-06-01 16:58:57 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-09-11 17:34:39 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-02-21 11:38:17 +00:00
|
|
|
|
2023-06-01 16:58:57 +02:00
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2014-04-04 14:20:57 -04:00
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2014-04-04 14:20:57 -04:00
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2014-04-04 14:20:57 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2019-11-21 23:25:07 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
2019-11-21 23:25:07 +00:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2014-03-05 22:53:04 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
mm/filemap: generic_file_read_iter(): check for zero reads unconditionally
If
- generic_file_read_iter() gets called with a zero read length,
- the read offset is at a page boundary,
- IOCB_DIRECT is not set
- and the page in question hasn't made it into the page cache yet,
then do_generic_file_read() will trigger a readahead with a req_size hint
of zero.
Since roundup_pow_of_two(0) is undefined, UBSAN reports
UBSAN: Undefined behaviour in include/linux/log2.h:63:13
shift exponent 64 is too large for 64-bit type 'long unsigned int'
CPU: 3 PID: 1017 Comm: sa1 Tainted: G L 4.5.0-next-20160318+ #14
[...]
Call Trace:
[...]
[<ffffffff813ef61a>] ondemand_readahead+0x3aa/0x3d0
[<ffffffff813ef61a>] ? ondemand_readahead+0x3aa/0x3d0
[<ffffffff813c73bd>] ? find_get_entry+0x2d/0x210
[<ffffffff813ef9c3>] page_cache_sync_readahead+0x63/0xa0
[<ffffffff813cc04d>] do_generic_file_read+0x80d/0xf90
[<ffffffff813cc955>] generic_file_read_iter+0x185/0x420
[...]
[<ffffffff81510b06>] __vfs_read+0x256/0x3d0
[...]
when get_init_ra_size() gets called from ondemand_readahead().
The net effect is that the initial readahead size is arch dependent for
requested read lengths of zero: for example, since
1UL << (sizeof(unsigned long) * 8)
evaluates to 1 on x86 while its result is 0 on ARMv7, the initial readahead
size becomes 4 on the former and 0 on the latter.
What's more, whether or not the file access timestamp is updated for zero
length reads is decided differently for the two cases of IOCB_DIRECT
being set or cleared: in the first case, generic_file_read_iter()
explicitly skips updating that timestamp while in the latter case, it is
always updated through the call to do_generic_file_read().
According to POSIX, zero length reads "do not modify the last data access
timestamp" and thus, the IOCB_DIRECT behaviour is POSIXly correct.
Let generic_file_read_iter() unconditionally check the requested read
length at its entry and return immediately with success if it is zero.
Signed-off-by: Nicolai Stange <nicstange@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-25 14:22:14 -07:00
|
|
|
|
2017-08-29 16:13:18 +02:00
|
|
|
|
mm/filemap: generic_file_read_iter(): check for zero reads unconditionally
If
- generic_file_read_iter() gets called with a zero read length,
- the read offset is at a page boundary,
- IOCB_DIRECT is not set
- and the page in question hasn't made it into the page cache yet,
then do_generic_file_read() will trigger a readahead with a req_size hint
of zero.
Since roundup_pow_of_two(0) is undefined, UBSAN reports
UBSAN: Undefined behaviour in include/linux/log2.h:63:13
shift exponent 64 is too large for 64-bit type 'long unsigned int'
CPU: 3 PID: 1017 Comm: sa1 Tainted: G L 4.5.0-next-20160318+ #14
[...]
Call Trace:
[...]
[<ffffffff813ef61a>] ondemand_readahead+0x3aa/0x3d0
[<ffffffff813ef61a>] ? ondemand_readahead+0x3aa/0x3d0
[<ffffffff813c73bd>] ? find_get_entry+0x2d/0x210
[<ffffffff813ef9c3>] page_cache_sync_readahead+0x63/0xa0
[<ffffffff813cc04d>] do_generic_file_read+0x80d/0xf90
[<ffffffff813cc955>] generic_file_read_iter+0x185/0x420
[...]
[<ffffffff81510b06>] __vfs_read+0x256/0x3d0
[...]
when get_init_ra_size() gets called from ondemand_readahead().
The net effect is that the initial readahead size is arch dependent for
requested read lengths of zero: for example, since
1UL << (sizeof(unsigned long) * 8)
evaluates to 1 on x86 while its result is 0 on ARMv7, the initial readahead
size becomes 4 on the former and 0 on the latter.
What's more, whether or not the file access timestamp is updated for zero
length reads is decided differently for the two cases of IOCB_DIRECT
being set or cleared: in the first case, generic_file_read_iter()
explicitly skips updating that timestamp while in the latter case, it is
always updated through the call to do_generic_file_read().
According to POSIX, zero length reads "do not modify the last data access
timestamp" and thus, the IOCB_DIRECT behaviour is POSIXly correct.
Let generic_file_read_iter() unconditionally check the requested read
length at its entry and return immediately with success if it is zero.
Signed-off-by: Nicolai Stange <nicstange@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-03-25 14:22:14 -07:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:45 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2015-04-09 13:52:01 -04:00
|
|
|
|
2017-08-29 16:13:18 +02:00
|
|
|
|
2014-03-05 22:53:04 -05:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-06-01 16:58:56 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2016-10-03 09:48:08 +11:00
|
|
|
|
|
|
|
|
|
2017-04-13 14:13:36 -04:00
|
|
|
|
2016-10-10 13:26:27 -04:00
|
|
|
|
2016-04-07 08:51:55 -07:00
|
|
|
|
2017-04-13 14:13:36 -04:00
|
|
|
|
Fix race when checking i_size on direct i/o read
So far I've had one ACK for this, and no other comments. So I think it
is probably time to send this via some suitable tree. I'm guessing that
the vfs tree would be the most appropriate route, but not sure that
there is one at the moment (don't see anything recent at kernel.org)
so in that case I think -mm is the "back up plan". Al, please let me
know if you will take this?
Steve.
---------------------
Following on from the "Re: [PATCH v3] vfs: fix a bug when we do some dio
reads with append dio writes" thread on linux-fsdevel, this patch is my
current version of the fix proposed as option (b) in that thread.
Removing the i_size test from the direct i/o read path at vfs level
means that filesystems now have to deal with requests which are beyond
i_size themselves. These I've divided into three sets:
a) Those with "no op" ->direct_IO (9p, cifs, ceph)
These are obviously not going to be an issue
b) Those with "home brew" ->direct_IO (nfs, fuse)
I've been told that NFS should not have any problem with the larger
i_size, however I've added an extra test to FUSE to duplicate the
original behaviour just to be on the safe side.
c) Those using __blockdev_direct_IO()
These call through to ->get_block() which should deal with the EOF
condition correctly. I've verified that with GFS2 and I believe that
Zheng has verified it for ext4. I've also run the test on XFS and it
passes both before and after this change.
The part of the patch in filemap.c looks a lot larger than it really is
- there are only two lines of real change. The rest is just indentation
of the contained code.
There remains a test of i_size though, which was added for btrfs. It
doesn't cause the other filesystems a problem as the test is performed
after ->direct_IO has been called. It is possible that there is a race
that does matter to btrfs, however this patch doesn't change that, so
its still an overall improvement.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Reported-by: Zheng Liu <gnehzuil.liu@gmail.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Dave Chinner <david@fromorbit.com>
Acked-by: Miklos Szeredi <miklos@szeredi.hu>
Cc: Chris Mason <clm@fb.com>
Cc: Josef Bacik <jbacik@fb.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2014-01-24 14:42:22 +00:00
|
|
|
|
2021-02-24 12:01:45 -08:00
|
|
|
|
|
|
|
|
|
2010-05-23 11:00:54 -04:00
|
|
|
|
Fix race when checking i_size on direct i/o read
So far I've had one ACK for this, and no other comments. So I think it
is probably time to send this via some suitable tree. I'm guessing that
the vfs tree would be the most appropriate route, but not sure that
there is one at the moment (don't see anything recent at kernel.org)
so in that case I think -mm is the "back up plan". Al, please let me
know if you will take this?
Steve.
---------------------
Following on from the "Re: [PATCH v3] vfs: fix a bug when we do some dio
reads with append dio writes" thread on linux-fsdevel, this patch is my
current version of the fix proposed as option (b) in that thread.
Removing the i_size test from the direct i/o read path at vfs level
means that filesystems now have to deal with requests which are beyond
i_size themselves. These I've divided into three sets:
a) Those with "no op" ->direct_IO (9p, cifs, ceph)
These are obviously not going to be an issue
b) Those with "home brew" ->direct_IO (nfs, fuse)
I've been told that NFS should not have any problem with the larger
i_size, however I've added an extra test to FUSE to duplicate the
original behaviour just to be on the safe side.
c) Those using __blockdev_direct_IO()
These call through to ->get_block() which should deal with the EOF
condition correctly. I've verified that with GFS2 and I believe that
Zheng has verified it for ext4. I've also run the test on XFS and it
passes both before and after this change.
The part of the patch in filemap.c looks a lot larger than it really is
- there are only two lines of real change. The rest is just indentation
of the contained code.
There remains a test of i_size though, which was added for btrfs. It
doesn't cause the other filesystems a problem as the test is performed
after ->direct_IO has been called. It is possible that there is a race
that does matter to btrfs, however this patch doesn't change that, so
its still an overall improvement.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Reported-by: Zheng Liu <gnehzuil.liu@gmail.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Dave Chinner <david@fromorbit.com>
Acked-by: Miklos Szeredi <miklos@szeredi.hu>
Cc: Chris Mason <clm@fb.com>
Cc: Josef Bacik <jbacik@fb.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2014-01-24 14:42:22 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2015-02-16 15:58:53 -08:00
|
|
|
|
|
|
|
|
|
Fix race when checking i_size on direct i/o read
So far I've had one ACK for this, and no other comments. So I think it
is probably time to send this via some suitable tree. I'm guessing that
the vfs tree would be the most appropriate route, but not sure that
there is one at the moment (don't see anything recent at kernel.org)
so in that case I think -mm is the "back up plan". Al, please let me
know if you will take this?
Steve.
---------------------
Following on from the "Re: [PATCH v3] vfs: fix a bug when we do some dio
reads with append dio writes" thread on linux-fsdevel, this patch is my
current version of the fix proposed as option (b) in that thread.
Removing the i_size test from the direct i/o read path at vfs level
means that filesystems now have to deal with requests which are beyond
i_size themselves. These I've divided into three sets:
a) Those with "no op" ->direct_IO (9p, cifs, ceph)
These are obviously not going to be an issue
b) Those with "home brew" ->direct_IO (nfs, fuse)
I've been told that NFS should not have any problem with the larger
i_size, however I've added an extra test to FUSE to duplicate the
original behaviour just to be on the safe side.
c) Those using __blockdev_direct_IO()
These call through to ->get_block() which should deal with the EOF
condition correctly. I've verified that with GFS2 and I believe that
Zheng has verified it for ext4. I've also run the test on XFS and it
passes both before and after this change.
The part of the patch in filemap.c looks a lot larger than it really is
- there are only two lines of real change. The rest is just indentation
of the contained code.
There remains a test of i_size though, which was added for btrfs. It
doesn't cause the other filesystems a problem as the test is performed
after ->direct_IO has been called. It is possible that there is a race
that does matter to btrfs, however this patch doesn't change that, so
its still an overall improvement.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Reported-by: Zheng Liu <gnehzuil.liu@gmail.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Dave Chinner <david@fromorbit.com>
Acked-by: Miklos Szeredi <miklos@szeredi.hu>
Cc: Chris Mason <clm@fb.com>
Cc: Josef Bacik <jbacik@fb.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2014-01-24 14:42:22 +00:00
|
|
|
|
2021-11-05 13:37:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:45 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2021-02-24 12:02:45 -08:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2014-03-05 22:53:04 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-02-14 15:01:42 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
fs/pipe: add simpler helpers for common cases
The fix to atomically read the pipe head and tail state when not holding
the pipe mutex has caused a number of headaches due to the size change
of the involved types.
It turns out that we don't have _that_ many places that access these
fields directly and were affected, but we have more than we strictly
should have, because our low-level helper functions have been designed
to have intimate knowledge of how the pipes work.
And as a result, that random noise of direct 'pipe->head' and
'pipe->tail' accesses makes it harder to pinpoint any actual potential
problem spots remaining.
For example, we didn't have a "is the pipe full" helper function, but
instead had a "given these pipe buffer indexes and this pipe size, is
the pipe full". That's because some low-level pipe code does actually
want that much more complicated interface.
But most other places literally just want a "is the pipe full" helper,
and not having it meant that those places ended up being unnecessarily
much too aware of this all.
It would have been much better if only the very core pipe code that
cared had been the one aware of this all.
So let's fix it - better late than never. This just introduces the
trivial wrappers for "is this pipe full or empty" and to get how many
pipe buffers are used, so that instead of writing
if (pipe_full(pipe->head, pipe->tail, pipe->max_usage))
the places that literally just want to know if a pipe is full can just
say
if (pipe_is_full(pipe))
instead. The existing trivial cases were converted with a 'sed' script.
This cuts down on the places that access pipe->head and pipe->tail
directly outside of the pipe code (and core splice code) quite a lot.
The splice code in particular still revels in doing the direct low-level
accesses, and the fuse fuse_dev_splice_write() code also seems a bit
unnecessarily eager to go very low-level, but it's at least a bit better
than it used to be.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2025-03-06 18:25:35 -10:00
|
|
|
|
2023-02-14 15:01:42 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-22 14:50:18 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-14 15:01:42 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-22 14:49:49 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-14 15:01:42 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
fs/pipe: add simpler helpers for common cases
The fix to atomically read the pipe head and tail state when not holding
the pipe mutex has caused a number of headaches due to the size change
of the involved types.
It turns out that we don't have _that_ many places that access these
fields directly and were affected, but we have more than we strictly
should have, because our low-level helper functions have been designed
to have intimate knowledge of how the pipes work.
And as a result, that random noise of direct 'pipe->head' and
'pipe->tail' accesses makes it harder to pinpoint any actual potential
problem spots remaining.
For example, we didn't have a "is the pipe full" helper function, but
instead had a "given these pipe buffer indexes and this pipe size, is
the pipe full". That's because some low-level pipe code does actually
want that much more complicated interface.
But most other places literally just want a "is the pipe full" helper,
and not having it meant that those places ended up being unnecessarily
much too aware of this all.
It would have been much better if only the very core pipe code that
cared had been the one aware of this all.
So let's fix it - better late than never. This just introduces the
trivial wrappers for "is this pipe full or empty" and to get how many
pipe buffers are used, so that instead of writing
if (pipe_full(pipe->head, pipe->tail, pipe->max_usage))
the places that literally just want to know if a pipe is full can just
say
if (pipe_is_full(pipe))
instead. The existing trivial cases were converted with a 'sed' script.
This cuts down on the places that access pipe->head and pipe->tail
directly outside of the pipe code (and core splice code) quite a lot.
The splice code in particular still revels in doing the direct low-level
accesses, and the fuse fuse_dev_splice_write() code also seems a bit
unnecessarily eager to go very low-level, but it's at least a bit better
than it used to be.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2025-03-06 18:25:35 -10:00
|
|
|
|
2023-02-14 15:01:42 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-22 14:49:48 +01:00
|
|
|
|
2023-02-14 15:01:42 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-22 14:49:48 +01:00
|
|
|
|
2023-02-14 15:01:42 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
fs/pipe: add simpler helpers for common cases
The fix to atomically read the pipe head and tail state when not holding
the pipe mutex has caused a number of headaches due to the size change
of the involved types.
It turns out that we don't have _that_ many places that access these
fields directly and were affected, but we have more than we strictly
should have, because our low-level helper functions have been designed
to have intimate knowledge of how the pipes work.
And as a result, that random noise of direct 'pipe->head' and
'pipe->tail' accesses makes it harder to pinpoint any actual potential
problem spots remaining.
For example, we didn't have a "is the pipe full" helper function, but
instead had a "given these pipe buffer indexes and this pipe size, is
the pipe full". That's because some low-level pipe code does actually
want that much more complicated interface.
But most other places literally just want a "is the pipe full" helper,
and not having it meant that those places ended up being unnecessarily
much too aware of this all.
It would have been much better if only the very core pipe code that
cared had been the one aware of this all.
So let's fix it - better late than never. This just introduces the
trivial wrappers for "is this pipe full or empty" and to get how many
pipe buffers are used, so that instead of writing
if (pipe_full(pipe->head, pipe->tail, pipe->max_usage))
the places that literally just want to know if a pipe is full can just
say
if (pipe_is_full(pipe))
instead. The existing trivial cases were converted with a 'sed' script.
This cuts down on the places that access pipe->head and pipe->tail
directly outside of the pipe code (and core splice code) quite a lot.
The splice code in particular still revels in doing the direct low-level
accesses, and the fuse fuse_dev_splice_write() code also seems a bit
unnecessarily eager to go very low-level, but it's at least a bit better
than it used to be.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2025-03-06 18:25:35 -10:00
|
|
|
|
2023-02-14 15:01:42 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-15 08:00:31 +00:00
|
|
|
|
2023-02-14 15:01:42 +00:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:52 -08:00
|
|
|
|
2021-02-25 17:15:48 -08:00
|
|
|
|
2021-02-25 17:15:52 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:52 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:52 -08:00
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:52 -08:00
|
|
|
|
|
|
|
|
|
2022-02-09 20:21:27 +00:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:52 -08:00
|
|
|
|
2025-01-02 11:04:11 -08:00
|
|
|
|
2021-02-25 17:15:52 -08:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:52 -08:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:52 -08:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:48 -08:00
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:48 -08:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2024-09-06 16:05:12 -07:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:48 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-06 18:06:47 -07:00
|
|
|
|
2021-02-25 17:15:48 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm/filemap: fix mapping_seek_hole_data on THP & 32-bit
No problem on 64-bit, or without huge pages, but xfstests generic/285
and other SEEK_HOLE/SEEK_DATA tests have regressed on huge tmpfs, and on
32-bit architectures, with the new mapping_seek_hole_data(). Several
different bugs turned out to need fixing.
u64 cast to stop losing bits when converting unsigned long to loff_t
(and let's use shifts throughout, rather than mixed with * and /).
Use round_up() when advancing pos, to stop assuming that pos was already
THP-aligned when advancing it by THP-size. (This use of round_up()
assumes that any THP has THP-aligned index: true at present and true
going forward, but could be recoded to avoid the assumption.)
Use xas_set() when iterating away from a THP, so that xa_index stays in
synch with start, instead of drifting away to return bogus offset.
Check start against end to avoid wrapping 32-bit xa_index to 0 (and to
handle these additional cases, seek_data or not, it's easier to break
the loop than goto: so rearrange exit from the function).
[hughd@google.com: remove unneeded u64 casts, per Matthew]
Link: https://lkml.kernel.org/r/alpine.LSU.2.11.2104221347240.1170@eggly.anvils
Link: https://lkml.kernel.org/r/alpine.LSU.2.11.2104211737410.3299@eggly.anvils
Fixes: 41139aa4c3a3 ("mm/filemap: add mapping_seek_hole_data")
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Dave Chinner <dchinner@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: William Kucharski <william.kucharski@oracle.com>
Cc: Yang Shi <yang.shi@linux.alibaba.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-04-23 14:29:00 -07:00
|
|
|
|
2021-02-25 17:15:48 -08:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:48 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
mm/filemap: fix mapping_seek_hole_data on THP & 32-bit
No problem on 64-bit, or without huge pages, but xfstests generic/285
and other SEEK_HOLE/SEEK_DATA tests have regressed on huge tmpfs, and on
32-bit architectures, with the new mapping_seek_hole_data(). Several
different bugs turned out to need fixing.
u64 cast to stop losing bits when converting unsigned long to loff_t
(and let's use shifts throughout, rather than mixed with * and /).
Use round_up() when advancing pos, to stop assuming that pos was already
THP-aligned when advancing it by THP-size. (This use of round_up()
assumes that any THP has THP-aligned index: true at present and true
going forward, but could be recoded to avoid the assumption.)
Use xas_set() when iterating away from a THP, so that xa_index stays in
synch with start, instead of drifting away to return bogus offset.
Check start against end to avoid wrapping 32-bit xa_index to 0 (and to
handle these additional cases, seek_data or not, it's easier to break
the loop than goto: so rearrange exit from the function).
[hughd@google.com: remove unneeded u64 casts, per Matthew]
Link: https://lkml.kernel.org/r/alpine.LSU.2.11.2104221347240.1170@eggly.anvils
Link: https://lkml.kernel.org/r/alpine.LSU.2.11.2104211737410.3299@eggly.anvils
Fixes: 41139aa4c3a3 ("mm/filemap: add mapping_seek_hole_data")
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Dave Chinner <dchinner@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: William Kucharski <william.kucharski@oracle.com>
Cc: Yang Shi <yang.shi@linux.alibaba.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-04-23 14:29:00 -07:00
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
2021-02-25 17:15:48 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:52 -08:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:48 -08:00
|
|
|
|
mm/filemap: fix mapping_seek_hole_data on THP & 32-bit
No problem on 64-bit, or without huge pages, but xfstests generic/285
and other SEEK_HOLE/SEEK_DATA tests have regressed on huge tmpfs, and on
32-bit architectures, with the new mapping_seek_hole_data(). Several
different bugs turned out to need fixing.
u64 cast to stop losing bits when converting unsigned long to loff_t
(and let's use shifts throughout, rather than mixed with * and /).
Use round_up() when advancing pos, to stop assuming that pos was already
THP-aligned when advancing it by THP-size. (This use of round_up()
assumes that any THP has THP-aligned index: true at present and true
going forward, but could be recoded to avoid the assumption.)
Use xas_set() when iterating away from a THP, so that xa_index stays in
synch with start, instead of drifting away to return bogus offset.
Check start against end to avoid wrapping 32-bit xa_index to 0 (and to
handle these additional cases, seek_data or not, it's easier to break
the loop than goto: so rearrange exit from the function).
[hughd@google.com: remove unneeded u64 casts, per Matthew]
Link: https://lkml.kernel.org/r/alpine.LSU.2.11.2104221347240.1170@eggly.anvils
Link: https://lkml.kernel.org/r/alpine.LSU.2.11.2104211737410.3299@eggly.anvils
Fixes: 41139aa4c3a3 ("mm/filemap: add mapping_seek_hole_data")
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Dave Chinner <dchinner@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: William Kucharski <william.kucharski@oracle.com>
Cc: Yang Shi <yang.shi@linux.alibaba.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-04-23 14:29:00 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:48 -08:00
|
|
|
|
|
|
|
|
|
mm/filemap: fix mapping_seek_hole_data on THP & 32-bit
No problem on 64-bit, or without huge pages, but xfstests generic/285
and other SEEK_HOLE/SEEK_DATA tests have regressed on huge tmpfs, and on
32-bit architectures, with the new mapping_seek_hole_data(). Several
different bugs turned out to need fixing.
u64 cast to stop losing bits when converting unsigned long to loff_t
(and let's use shifts throughout, rather than mixed with * and /).
Use round_up() when advancing pos, to stop assuming that pos was already
THP-aligned when advancing it by THP-size. (This use of round_up()
assumes that any THP has THP-aligned index: true at present and true
going forward, but could be recoded to avoid the assumption.)
Use xas_set() when iterating away from a THP, so that xa_index stays in
synch with start, instead of drifting away to return bogus offset.
Check start against end to avoid wrapping 32-bit xa_index to 0 (and to
handle these additional cases, seek_data or not, it's easier to break
the loop than goto: so rearrange exit from the function).
[hughd@google.com: remove unneeded u64 casts, per Matthew]
Link: https://lkml.kernel.org/r/alpine.LSU.2.11.2104221347240.1170@eggly.anvils
Link: https://lkml.kernel.org/r/alpine.LSU.2.11.2104211737410.3299@eggly.anvils
Fixes: 41139aa4c3a3 ("mm/filemap: add mapping_seek_hole_data")
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Dave Chinner <dchinner@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: William Kucharski <william.kucharski@oracle.com>
Cc: Yang Shi <yang.shi@linux.alibaba.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-04-23 14:29:00 -07:00
|
|
|
|
2021-02-25 17:15:48 -08:00
|
|
|
|
|
|
|
|
|
2020-12-17 00:12:26 -05:00
|
|
|
|
|
|
|
|
|
2021-02-25 17:15:48 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
2021-03-01 19:38:25 -05:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
2019-03-15 11:26:07 -07:00
|
|
|
|
|
|
|
|
|
mm: make lock_folio_maybe_drop_mmap() VMA lock aware
Patch series "Handle more faults under the VMA lock", v2.
At this point, we're handling the majority of file-backed page faults
under the VMA lock, using the ->map_pages entry point. This patch set
attempts to expand that for the following siutations:
- We have to do a read. This could be because we've hit the point in
the readahead window where we need to kick off the next readahead,
or because the page is simply not present in cache.
- We're handling a write fault. Most applications don't do I/O by writes
to shared mmaps for very good reasons, but some do, and it'd be nice
to not make that slow unnecessarily.
- We're doing a COW of a private mapping (both PTE already present
and PTE not-present). These are two different codepaths and I handle
both of them in this patch set.
There is no support in this patch set for drivers to mark themselves as
being VMA lock friendly; they could implement the ->map_pages
vm_operation, but if they do, they would be the first. This is probably
something we want to change at some point in the future, and I've marked
where to make that change in the code.
There is very little performance change in the benchmarks we've run;
mostly because the vast majority of page faults are handled through the
other paths. I still think this patch series is useful for workloads that
may take these paths more often, and just for cleaning up the fault path
in general (it's now clearer why we have to retry in these cases).
This patch (of 6):
Drop the VMA lock instead of the mmap_lock if that's the one which
is held.
Link: https://lkml.kernel.org/r/20231006195318.4087158-1-willy@infradead.org
Link: https://lkml.kernel.org/r/20231006195318.4087158-2-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-10-06 20:53:13 +01:00
|
|
|
|
2019-03-15 11:26:07 -07:00
|
|
|
|
|
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-08 00:07:31 -05:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
mm: make lock_folio_maybe_drop_mmap() VMA lock aware
Patch series "Handle more faults under the VMA lock", v2.
At this point, we're handling the majority of file-backed page faults
under the VMA lock, using the ->map_pages entry point. This patch set
attempts to expand that for the following siutations:
- We have to do a read. This could be because we've hit the point in
the readahead window where we need to kick off the next readahead,
or because the page is simply not present in cache.
- We're handling a write fault. Most applications don't do I/O by writes
to shared mmaps for very good reasons, but some do, and it'd be nice
to not make that slow unnecessarily.
- We're doing a COW of a private mapping (both PTE already present
and PTE not-present). These are two different codepaths and I handle
both of them in this patch set.
There is no support in this patch set for drivers to mark themselves as
being VMA lock friendly; they could implement the ->map_pages
vm_operation, but if they do, they would be the first. This is probably
something we want to change at some point in the future, and I've marked
where to make that change in the code.
There is very little performance change in the benchmarks we've run;
mostly because the vast majority of page faults are handled through the
other paths. I still think this patch series is useful for workloads that
may take these paths more often, and just for cleaning up the fault path
in general (it's now clearer why we have to retry in these cases).
This patch (of 6):
Drop the VMA lock instead of the mmap_lock if that's the one which
is held.
Link: https://lkml.kernel.org/r/20231006195318.4087158-1-willy@infradead.org
Link: https://lkml.kernel.org/r/20231006195318.4087158-2-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-10-06 20:53:13 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
mm: make lock_folio_maybe_drop_mmap() VMA lock aware
Patch series "Handle more faults under the VMA lock", v2.
At this point, we're handling the majority of file-backed page faults
under the VMA lock, using the ->map_pages entry point. This patch set
attempts to expand that for the following siutations:
- We have to do a read. This could be because we've hit the point in
the readahead window where we need to kick off the next readahead,
or because the page is simply not present in cache.
- We're handling a write fault. Most applications don't do I/O by writes
to shared mmaps for very good reasons, but some do, and it'd be nice
to not make that slow unnecessarily.
- We're doing a COW of a private mapping (both PTE already present
and PTE not-present). These are two different codepaths and I handle
both of them in this patch set.
There is no support in this patch set for drivers to mark themselves as
being VMA lock friendly; they could implement the ->map_pages
vm_operation, but if they do, they would be the first. This is probably
something we want to change at some point in the future, and I've marked
where to make that change in the code.
There is very little performance change in the benchmarks we've run;
mostly because the vast majority of page faults are handled through the
other paths. I still think this patch series is useful for workloads that
may take these paths more often, and just for cleaning up the fault path
in general (it's now clearer why we have to retry in these cases).
This patch (of 6):
Drop the VMA lock instead of the mmap_lock if that's the one which
is held.
Link: https://lkml.kernel.org/r/20231006195318.4087158-1-willy@infradead.org
Link: https://lkml.kernel.org/r/20231006195318.4087158-2-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-10-06 20:53:13 +01:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-01 19:38:25 -05:00
|
|
|
|
|
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
2019-03-13 11:44:18 -07:00
|
|
|
|
|
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
2021-04-07 21:18:55 +01:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2025-06-18 20:42:53 +01:00
|
|
|
|
2025-06-09 10:27:25 +01:00
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
2021-07-24 23:37:13 -04:00
|
|
|
|
|
|
|
|
|
2024-06-27 10:39:51 +10:00
|
|
|
|
2021-07-24 23:37:13 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-05-25 14:23:45 -04:00
|
|
|
|
2021-07-24 23:37:13 -04:00
|
|
|
|
|
|
|
|
|
2025-06-09 10:27:26 +01:00
|
|
|
|
|
|
|
|
|
2021-07-24 23:37:13 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm/filemap: allow arch to request folio size for exec memory
Change the readahead config so that if it is being requested for an
executable mapping, do a synchronous read into a set of folios with an
arch-specified order and in a naturally aligned manner. We no longer
center the read on the faulting page but simply align it down to the
previous natural boundary. Additionally, we don't bother with an
asynchronous part.
On arm64 if memory is physically contiguous and naturally aligned to the
"contpte" size, we can use contpte mappings, which improves utilization of
the TLB. When paired with the "multi-size THP" feature, this works well
to reduce dTLB pressure. However iTLB pressure is still high due to
executable mappings having a low likelihood of being in the required folio
size and mapping alignment, even when the filesystem supports readahead
into large folios (e.g. XFS).
The reason for the low likelihood is that the current readahead algorithm
starts with an order-0 folio and increases the folio order by 2 every time
the readahead mark is hit. But most executable memory tends to be
accessed randomly and so the readahead mark is rarely hit and most
executable folios remain order-0.
So let's special-case the read(ahead) logic for executable mappings. The
trade-off is performance improvement (due to more efficient storage of the
translations in iTLB) vs potential for making reclaim more difficult (due
to the folios being larger so if a part of the folio is hot the whole
thing is considered hot). But executable memory is a small portion of the
overall system memory so I doubt this will even register from a reclaim
perspective.
I've chosen 64K folio size for arm64 which benefits both the 4K and 16K
base page size configs. Crucially the same amount of data is still read
(usually 128K) so I'm not expecting any read amplification issues. I
don't anticipate any write amplification because text is always RO.
Note that the text region of an ELF file could be populated into the page
cache for other reasons than taking a fault in a mmapped area. The most
common case is due to the loader read()ing the header which can be shared
with the beginning of text. So some text will still remain in small
folios, but this simple, best effort change provides good performance
improvements as is.
Confine this special-case approach to the bounds of the VMA. This
prevents wasting memory for any padding that might exist in the file
between sections. Previously the padding would have been contained in
order-0 folios and would be easy to reclaim. But now it would be part of
a larger folio so more difficult to reclaim. Solve this by simply not
reading it into memory in the first place.
Benchmarking
============
The below shows pgbench and redis benchmarks on Graviton3 arm64 system.
First, confirmation that this patch causes more text to be contained in
64K folios:
+----------------------+---------------+---------------+---------------+
| File-backed folios by| system boot | pgbench | redis |
| size as percentage of+-------+-------+-------+-------+-------+-------+
| all mapped text mem |before | after |before | after |before | after |
+======================+=======+=======+=======+=======+=======+=======+
| base-page-4kB | 78% | 30% | 78% | 11% | 73% | 14% |
| thp-aligned-8kB | 1% | 0% | 0% | 0% | 1% | 0% |
| thp-aligned-16kB | 17% | 4% | 17% | 3% | 20% | 4% |
| thp-aligned-32kB | 1% | 1% | 1% | 2% | 1% | 1% |
| thp-aligned-64kB | 3% | 63% | 3% | 81% | 4% | 77% |
| thp-aligned-128kB | 0% | 1% | 1% | 1% | 1% | 2% |
| thp-unaligned-64kB | 0% | 0% | 0% | 1% | 0% | 1% |
| thp-unaligned-128kB | 0% | 1% | 0% | 0% | 0% | 0% |
| thp-partial | 0% | 0% | 0% | 1% | 0% | 1% |
+----------------------+-------+-------+-------+-------+-------+-------+
| cont-aligned-64kB | 4% | 65% | 4% | 83% | 6% | 79% |
+----------------------+-------+-------+-------+-------+-------+-------+
The above shows that for both workloads (each isolated with cgroups) as
well as the general system state after boot, the amount of text backed by
4K and 16K folios reduces and the amount backed by 64K folios increases
significantly. And the amount of text that is contpte-mapped
significantly increases (see last row).
And this is reflected in performance improvement. "(I)" indicates a
statistically significant improvement. Note TPS and Reqs/sec are rates so
bigger is better, ms is time so smaller is better:
+-------------+-------------------------------------------+------------+
| Benchmark | Result Class | Improvemnt |
+=============+===========================================+============+
| pts/pgbench | Scale: 1 Clients: 1 RO (TPS) | (I) 3.47% |
| | Scale: 1 Clients: 1 RO - Latency (ms) | -2.88% |
| | Scale: 1 Clients: 250 RO (TPS) | (I) 5.02% |
| | Scale: 1 Clients: 250 RO - Latency (ms) | (I) -4.79% |
| | Scale: 1 Clients: 1000 RO (TPS) | (I) 6.16% |
| | Scale: 1 Clients: 1000 RO - Latency (ms) | (I) -5.82% |
| | Scale: 100 Clients: 1 RO (TPS) | 2.51% |
| | Scale: 100 Clients: 1 RO - Latency (ms) | -3.51% |
| | Scale: 100 Clients: 250 RO (TPS) | (I) 4.75% |
| | Scale: 100 Clients: 250 RO - Latency (ms) | (I) -4.44% |
| | Scale: 100 Clients: 1000 RO (TPS) | (I) 6.34% |
| | Scale: 100 Clients: 1000 RO - Latency (ms)| (I) -5.95% |
+-------------+-------------------------------------------+------------+
| pts/redis | Test: GET Connections: 50 (Reqs/sec) | (I) 3.20% |
| | Test: GET Connections: 1000 (Reqs/sec) | (I) 2.55% |
| | Test: LPOP Connections: 50 (Reqs/sec) | (I) 4.59% |
| | Test: LPOP Connections: 1000 (Reqs/sec) | (I) 4.81% |
| | Test: LPUSH Connections: 50 (Reqs/sec) | (I) 5.31% |
| | Test: LPUSH Connections: 1000 (Reqs/sec) | (I) 4.36% |
| | Test: SADD Connections: 50 (Reqs/sec) | (I) 2.64% |
| | Test: SADD Connections: 1000 (Reqs/sec) | (I) 4.15% |
| | Test: SET Connections: 50 (Reqs/sec) | (I) 3.11% |
| | Test: SET Connections: 1000 (Reqs/sec) | (I) 3.36% |
+-------------+-------------------------------------------+------------+
[ryan.roberts@arm.com: fix use-after-free]
Link: https://lkml.kernel.org/r/ea7f9da7-9a9f-4b85-9d0a-35b320f5ed25@arm.com
[ryan.roberts@arm.com: use the vma_pages() helper instead of open-coding]
Link: https://lkml.kernel.org/r/0e0f674b-3b7e-494f-ae7a-fc9dbb98dad4@arm.com
Link: https://lkml.kernel.org/r/20250609092729.274960-6-ryan.roberts@arm.com
Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Will Deacon <will@kernel.org>
Cc: Chaitanya S Prakash <chaitanyas.prakash@arm.com>
Cc: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-06-09 10:27:27 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2011-05-24 17:12:28 -07:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
2022-05-25 14:23:45 -04:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2021-04-07 21:18:55 +01:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
|
|
|
|
|
2011-05-24 17:12:29 -07:00
|
|
|
|
2020-08-14 17:31:27 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-08-14 17:31:27 -07:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
mm/filemap: allow arch to request folio size for exec memory
Change the readahead config so that if it is being requested for an
executable mapping, do a synchronous read into a set of folios with an
arch-specified order and in a naturally aligned manner. We no longer
center the read on the faulting page but simply align it down to the
previous natural boundary. Additionally, we don't bother with an
asynchronous part.
On arm64 if memory is physically contiguous and naturally aligned to the
"contpte" size, we can use contpte mappings, which improves utilization of
the TLB. When paired with the "multi-size THP" feature, this works well
to reduce dTLB pressure. However iTLB pressure is still high due to
executable mappings having a low likelihood of being in the required folio
size and mapping alignment, even when the filesystem supports readahead
into large folios (e.g. XFS).
The reason for the low likelihood is that the current readahead algorithm
starts with an order-0 folio and increases the folio order by 2 every time
the readahead mark is hit. But most executable memory tends to be
accessed randomly and so the readahead mark is rarely hit and most
executable folios remain order-0.
So let's special-case the read(ahead) logic for executable mappings. The
trade-off is performance improvement (due to more efficient storage of the
translations in iTLB) vs potential for making reclaim more difficult (due
to the folios being larger so if a part of the folio is hot the whole
thing is considered hot). But executable memory is a small portion of the
overall system memory so I doubt this will even register from a reclaim
perspective.
I've chosen 64K folio size for arm64 which benefits both the 4K and 16K
base page size configs. Crucially the same amount of data is still read
(usually 128K) so I'm not expecting any read amplification issues. I
don't anticipate any write amplification because text is always RO.
Note that the text region of an ELF file could be populated into the page
cache for other reasons than taking a fault in a mmapped area. The most
common case is due to the loader read()ing the header which can be shared
with the beginning of text. So some text will still remain in small
folios, but this simple, best effort change provides good performance
improvements as is.
Confine this special-case approach to the bounds of the VMA. This
prevents wasting memory for any padding that might exist in the file
between sections. Previously the padding would have been contained in
order-0 folios and would be easy to reclaim. But now it would be part of
a larger folio so more difficult to reclaim. Solve this by simply not
reading it into memory in the first place.
Benchmarking
============
The below shows pgbench and redis benchmarks on Graviton3 arm64 system.
First, confirmation that this patch causes more text to be contained in
64K folios:
+----------------------+---------------+---------------+---------------+
| File-backed folios by| system boot | pgbench | redis |
| size as percentage of+-------+-------+-------+-------+-------+-------+
| all mapped text mem |before | after |before | after |before | after |
+======================+=======+=======+=======+=======+=======+=======+
| base-page-4kB | 78% | 30% | 78% | 11% | 73% | 14% |
| thp-aligned-8kB | 1% | 0% | 0% | 0% | 1% | 0% |
| thp-aligned-16kB | 17% | 4% | 17% | 3% | 20% | 4% |
| thp-aligned-32kB | 1% | 1% | 1% | 2% | 1% | 1% |
| thp-aligned-64kB | 3% | 63% | 3% | 81% | 4% | 77% |
| thp-aligned-128kB | 0% | 1% | 1% | 1% | 1% | 2% |
| thp-unaligned-64kB | 0% | 0% | 0% | 1% | 0% | 1% |
| thp-unaligned-128kB | 0% | 1% | 0% | 0% | 0% | 0% |
| thp-partial | 0% | 0% | 0% | 1% | 0% | 1% |
+----------------------+-------+-------+-------+-------+-------+-------+
| cont-aligned-64kB | 4% | 65% | 4% | 83% | 6% | 79% |
+----------------------+-------+-------+-------+-------+-------+-------+
The above shows that for both workloads (each isolated with cgroups) as
well as the general system state after boot, the amount of text backed by
4K and 16K folios reduces and the amount backed by 64K folios increases
significantly. And the amount of text that is contpte-mapped
significantly increases (see last row).
And this is reflected in performance improvement. "(I)" indicates a
statistically significant improvement. Note TPS and Reqs/sec are rates so
bigger is better, ms is time so smaller is better:
+-------------+-------------------------------------------+------------+
| Benchmark | Result Class | Improvemnt |
+=============+===========================================+============+
| pts/pgbench | Scale: 1 Clients: 1 RO (TPS) | (I) 3.47% |
| | Scale: 1 Clients: 1 RO - Latency (ms) | -2.88% |
| | Scale: 1 Clients: 250 RO (TPS) | (I) 5.02% |
| | Scale: 1 Clients: 250 RO - Latency (ms) | (I) -4.79% |
| | Scale: 1 Clients: 1000 RO (TPS) | (I) 6.16% |
| | Scale: 1 Clients: 1000 RO - Latency (ms) | (I) -5.82% |
| | Scale: 100 Clients: 1 RO (TPS) | 2.51% |
| | Scale: 100 Clients: 1 RO - Latency (ms) | -3.51% |
| | Scale: 100 Clients: 250 RO (TPS) | (I) 4.75% |
| | Scale: 100 Clients: 250 RO - Latency (ms) | (I) -4.44% |
| | Scale: 100 Clients: 1000 RO (TPS) | (I) 6.34% |
| | Scale: 100 Clients: 1000 RO - Latency (ms)| (I) -5.95% |
+-------------+-------------------------------------------+------------+
| pts/redis | Test: GET Connections: 50 (Reqs/sec) | (I) 3.20% |
| | Test: GET Connections: 1000 (Reqs/sec) | (I) 2.55% |
| | Test: LPOP Connections: 50 (Reqs/sec) | (I) 4.59% |
| | Test: LPOP Connections: 1000 (Reqs/sec) | (I) 4.81% |
| | Test: LPUSH Connections: 50 (Reqs/sec) | (I) 5.31% |
| | Test: LPUSH Connections: 1000 (Reqs/sec) | (I) 4.36% |
| | Test: SADD Connections: 50 (Reqs/sec) | (I) 2.64% |
| | Test: SADD Connections: 1000 (Reqs/sec) | (I) 4.15% |
| | Test: SET Connections: 50 (Reqs/sec) | (I) 3.11% |
| | Test: SET Connections: 1000 (Reqs/sec) | (I) 3.36% |
+-------------+-------------------------------------------+------------+
[ryan.roberts@arm.com: fix use-after-free]
Link: https://lkml.kernel.org/r/ea7f9da7-9a9f-4b85-9d0a-35b320f5ed25@arm.com
[ryan.roberts@arm.com: use the vma_pages() helper instead of open-coding]
Link: https://lkml.kernel.org/r/0e0f674b-3b7e-494f-ae7a-fc9dbb98dad4@arm.com
Link: https://lkml.kernel.org/r/20250609092729.274960-6-ryan.roberts@arm.com
Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Will Deacon <will@kernel.org>
Cc: Chaitanya S Prakash <chaitanyas.prakash@arm.com>
Cc: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-06-09 10:27:27 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2020-10-15 20:06:31 -07:00
|
|
|
|
2025-06-09 10:27:26 +01:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2020-06-08 21:33:54 -07:00
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2021-07-29 14:57:01 -04:00
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
2019-03-13 11:44:18 -07:00
|
|
|
|
|
|
|
|
|
2021-07-29 14:57:01 -04:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2025-06-09 10:27:25 +01:00
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
|
|
|
|
|
2020-04-01 21:04:40 -07:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2021-07-29 14:57:01 -04:00
|
|
|
|
2025-08-15 11:32:24 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-07-29 14:57:01 -04:00
|
|
|
|
|
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2021-07-29 14:57:01 -04:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
|
|
|
|
|
filemap: avoid unnecessary major faults in filemap_fault()
A major fault occurred when using mlockall(MCL_CURRENT | MCL_FUTURE) in
application, which leading to an unexpected issue[1].
This is caused by temporarily cleared PTE during a read+clear/modify/write
update of the PTE, eg, do_numa_page()/change_pte_range().
For the data segment of the user-mode program, the global variable area is
a private mapping. After the pagecache is loaded, the private anonymous
page is generated after the COW is triggered. Mlockall can lock COW pages
(anonymous pages), but the original file pages cannot be locked and may be
reclaimed. If the global variable (private anon page) is accessed when
vmf->pte is zeroed in numa fault, a file page fault will be triggered. At
this time, the original private file page may have been reclaimed. If the
page cache is not available at this time, a major fault will be triggered
and the file will be read, causing additional overhead.
This issue affects our traffic analysis service. The inbound traffic is
heavy. If a major fault occurs, the I/O schedule is triggered and the
original I/O is suspended. Generally, the I/O schedule is 0.7 ms. If
other applications are operating disks, the system needs to wait for more
than 10 ms. However, the inbound traffic is heavy and the NIC buffer is
small. As a result, packet loss occurs. But the traffic analysis service
can't tolerate packet loss.
Fix this by holding PTL and rechecking the PTE in filemap_fault() before
triggering a major fault. We do this check only if vma is VM_LOCKED to
reduce the performance impact in common scenarios.
In our product environment, there were 7 major faults every 12 hours.
After the patch is applied, no major fault have been triggered.
Testing file page read and write page fault performance in ext4 and
ramdisk using will-it-scale[2] on a x86 physical machine. The data is the
average change compared with the mainline after the patch is applied. The
test results are within the range of fluctuation. We do this check only
if vma is VM_LOCKED, therefore, no performance regressions is caused for
most common cases.
The test results are as follows:
processes processes_idle threads threads_idle
ext4 private file write: 0.22% 0.26% 1.21% -0.15%
ext4 private file read: 0.03% 1.00% 1.39% 0.34%
ext4 shared file write: -0.50% -0.02% -0.14% -0.02%
ramdisk private file write: 0.07% 0.02% 0.53% 0.04%
ramdisk private file read: 0.01% 1.60% -0.32% -0.02%
[1] https://lore.kernel.org/linux-mm/9e62fd9a-bee0-52bf-50a7-498fa17434ee@huawei.com/
[2] https://github.com/antonblanchard/will-it-scale/
Link: https://lkml.kernel.org/r/20240306083809.1236634-1-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Suggested-by: "Huang, Ying" <ying.huang@intel.com>
Suggested-by: David Hildenbrand <david@redhat.com>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-03-06 16:38:09 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-09-26 14:46:16 +08:00
|
|
|
|
|
|
|
|
|
filemap: avoid unnecessary major faults in filemap_fault()
A major fault occurred when using mlockall(MCL_CURRENT | MCL_FUTURE) in
application, which leading to an unexpected issue[1].
This is caused by temporarily cleared PTE during a read+clear/modify/write
update of the PTE, eg, do_numa_page()/change_pte_range().
For the data segment of the user-mode program, the global variable area is
a private mapping. After the pagecache is loaded, the private anonymous
page is generated after the COW is triggered. Mlockall can lock COW pages
(anonymous pages), but the original file pages cannot be locked and may be
reclaimed. If the global variable (private anon page) is accessed when
vmf->pte is zeroed in numa fault, a file page fault will be triggered. At
this time, the original private file page may have been reclaimed. If the
page cache is not available at this time, a major fault will be triggered
and the file will be read, causing additional overhead.
This issue affects our traffic analysis service. The inbound traffic is
heavy. If a major fault occurs, the I/O schedule is triggered and the
original I/O is suspended. Generally, the I/O schedule is 0.7 ms. If
other applications are operating disks, the system needs to wait for more
than 10 ms. However, the inbound traffic is heavy and the NIC buffer is
small. As a result, packet loss occurs. But the traffic analysis service
can't tolerate packet loss.
Fix this by holding PTL and rechecking the PTE in filemap_fault() before
triggering a major fault. We do this check only if vma is VM_LOCKED to
reduce the performance impact in common scenarios.
In our product environment, there were 7 major faults every 12 hours.
After the patch is applied, no major fault have been triggered.
Testing file page read and write page fault performance in ext4 and
ramdisk using will-it-scale[2] on a x86 physical machine. The data is the
average change compared with the mainline after the patch is applied. The
test results are within the range of fluctuation. We do this check only
if vma is VM_LOCKED, therefore, no performance regressions is caused for
most common cases.
The test results are as follows:
processes processes_idle threads threads_idle
ext4 private file write: 0.22% 0.26% 1.21% -0.15%
ext4 private file read: 0.03% 1.00% 1.39% 0.34%
ext4 shared file write: -0.50% -0.02% -0.14% -0.02%
ramdisk private file write: 0.07% 0.02% 0.53% 0.04%
ramdisk private file read: 0.01% 1.60% -0.32% -0.02%
[1] https://lore.kernel.org/linux-mm/9e62fd9a-bee0-52bf-50a7-498fa17434ee@huawei.com/
[2] https://github.com/antonblanchard/will-it-scale/
Link: https://lkml.kernel.org/r/20240306083809.1236634-1-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Suggested-by: "Huang, Ying" <ying.huang@intel.com>
Suggested-by: David Hildenbrand <david@redhat.com>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-03-06 16:38:09 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2007-07-19 01:46:59 -07:00
|
|
|
|
2007-07-19 01:47:03 -07:00
|
|
|
|
2006-06-23 02:03:49 -07:00
|
|
|
|
2007-07-19 01:46:59 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-08-06 16:07:24 -07:00
|
|
|
|
2020-06-08 21:33:54 -07:00
|
|
|
|
2014-08-06 16:07:24 -07:00
|
|
|
|
2020-06-08 21:33:54 -07:00
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
2014-08-06 16:07:24 -07:00
|
|
|
|
2020-06-08 21:33:54 -07:00
|
|
|
|
2014-08-06 16:07:24 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2018-06-07 17:08:00 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2017-02-24 14:56:41 -08:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
|
|
|
|
|
2018-06-07 17:08:00 -07:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
|
|
|
|
|
2007-10-31 09:19:46 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
filemap: add trace events for get_pages, map_pages, and fault
To allow precise tracking of page caches accessed, add new tracepoints
that trigger when a process actually accesses them.
The ureadahead program used by ChromeOS traces the disk access of programs
as they start up at boot up. It uses mincore(2) or the
'mm_filemap_add_to_page_cache' trace event to accomplish this. It stores
this information in a "pack" file and on subsequent boots, it will read
the pack file and call readahead(2) on the information so that disk
storage can be loaded into RAM before the applications actually need it.
A problem we see is that due to the kernel's readahead algorithm that can
aggressively pull in more data than needed (to try and accomplish the same
goal) and this data is also recorded. The end result is that the pack
file contains a lot of pages on disk that are never actually used.
Calling readahead(2) on these unused pages can slow down the system boot
up times.
To solve this, add 3 new trace events, get_pages, map_pages, and fault.
These will be used to trace the pages are not only pulled in from disk,
but are actually used by the application. Only those pages will be stored
in the pack file, and this helps out the performance of boot up.
With the combination of these 3 new trace events and
mm_filemap_add_to_page_cache, we observed a reduction in the pack file by
7.3% - 20% on ChromeOS varying by device.
Link: https://lkml.kernel.org/r/20240813100312.3930505-1-takayas@chromium.org
Signed-off-by: Takaya Saeki <takayas@chromium.org>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Reviewed-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Cc: Junichi Uekawa <uekawa@chromium.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-08-13 10:03:12 +00:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2013-10-16 13:46:59 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
2023-03-07 15:34:10 +01:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2021-07-29 14:57:01 -04:00
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
filemap: avoid unnecessary major faults in filemap_fault()
A major fault occurred when using mlockall(MCL_CURRENT | MCL_FUTURE) in
application, which leading to an unexpected issue[1].
This is caused by temporarily cleared PTE during a read+clear/modify/write
update of the PTE, eg, do_numa_page()/change_pte_range().
For the data segment of the user-mode program, the global variable area is
a private mapping. After the pagecache is loaded, the private anonymous
page is generated after the COW is triggered. Mlockall can lock COW pages
(anonymous pages), but the original file pages cannot be locked and may be
reclaimed. If the global variable (private anon page) is accessed when
vmf->pte is zeroed in numa fault, a file page fault will be triggered. At
this time, the original private file page may have been reclaimed. If the
page cache is not available at this time, a major fault will be triggered
and the file will be read, causing additional overhead.
This issue affects our traffic analysis service. The inbound traffic is
heavy. If a major fault occurs, the I/O schedule is triggered and the
original I/O is suspended. Generally, the I/O schedule is 0.7 ms. If
other applications are operating disks, the system needs to wait for more
than 10 ms. However, the inbound traffic is heavy and the NIC buffer is
small. As a result, packet loss occurs. But the traffic analysis service
can't tolerate packet loss.
Fix this by holding PTL and rechecking the PTE in filemap_fault() before
triggering a major fault. We do this check only if vma is VM_LOCKED to
reduce the performance impact in common scenarios.
In our product environment, there were 7 major faults every 12 hours.
After the patch is applied, no major fault have been triggered.
Testing file page read and write page fault performance in ext4 and
ramdisk using will-it-scale[2] on a x86 physical machine. The data is the
average change compared with the mainline after the patch is applied. The
test results are within the range of fluctuation. We do this check only
if vma is VM_LOCKED, therefore, no performance regressions is caused for
most common cases.
The test results are as follows:
processes processes_idle threads threads_idle
ext4 private file write: 0.22% 0.26% 1.21% -0.15%
ext4 private file read: 0.03% 1.00% 1.39% 0.34%
ext4 shared file write: -0.50% -0.02% -0.14% -0.02%
ramdisk private file write: 0.07% 0.02% 0.53% 0.04%
ramdisk private file read: 0.01% 1.60% -0.32% -0.02%
[1] https://lore.kernel.org/linux-mm/9e62fd9a-bee0-52bf-50a7-498fa17434ee@huawei.com/
[2] https://github.com/antonblanchard/will-it-scale/
Link: https://lkml.kernel.org/r/20240306083809.1236634-1-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Suggested-by: "Huang, Ying" <ying.huang@intel.com>
Suggested-by: David Hildenbrand <david@redhat.com>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-03-06 16:38:09 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
|
|
|
|
|
2017-07-06 15:40:25 -07:00
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
filemap: kill page_cache_read usage in filemap_fault
Patch series "drop the mmap_sem when doing IO in the fault path", v6.
Now that we have proper isolation in place with cgroups2 we have started
going through and fixing the various priority inversions. Most are all
gone now, but this one is sort of weird since it's not necessarily a
priority inversion that happens within the kernel, but rather because of
something userspace does.
We have giant applications that we want to protect, and parts of these
giant applications do things like watch the system state to determine how
healthy the box is for load balancing and such. This involves running
'ps' or other such utilities. These utilities will often walk
/proc/<pid>/whatever, and these files can sometimes need to
down_read(&task->mmap_sem). Not usually a big deal, but we noticed when
we are stress testing that sometimes our protected application has latency
spikes trying to get the mmap_sem for tasks that are in lower priority
cgroups.
This is because any down_write() on a semaphore essentially turns it into
a mutex, so even if we currently have it held for reading, any new readers
will not be allowed on to keep from starving the writer. This is fine,
except a lower priority task could be stuck doing IO because it has been
throttled to the point that its IO is taking much longer than normal. But
because a higher priority group depends on this completing it is now stuck
behind lower priority work.
In order to avoid this particular priority inversion we want to use the
existing retry mechanism to stop from holding the mmap_sem at all if we
are going to do IO. This already exists in the read case sort of, but
needed to be extended for more than just grabbing the page lock. With
io.latency we throttle at submit_bio() time, so the readahead stuff can
block and even page_cache_read can block, so all these paths need to have
the mmap_sem dropped.
The other big thing is ->page_mkwrite. btrfs is particularly shitty here
because we have to reserve space for the dirty page, which can be a very
expensive operation. We use the same retry method as the read path, and
simply cache the page and verify the page is still setup properly the next
pass through ->page_mkwrite().
I've tested these patches with xfstests and there are no regressions.
This patch (of 3):
If we do not have a page at filemap_fault time we'll do this weird forced
page_cache_read thing to populate the page, and then drop it again and
loop around and find it. This makes for 2 ways we can read a page in
filemap_fault, and it's not really needed. Instead add a FGP_FOR_MMAP
flag so that pagecache_get_page() will return a unlocked page that's in
pagecache. Then use the normal page locking and readpage logic already in
filemap_fault. This simplifies the no page in page cache case
significantly.
[akpm@linux-foundation.org: fix comment text]
[josef@toxicpanda.com: don't unlock null page in FGP_FOR_MMAP case]
Link: http://lkml.kernel.org/r/20190312201742.22935-1-josef@toxicpanda.com
Link: http://lkml.kernel.org/r/20181211173801.29535-2-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:14 -07:00
|
|
|
|
|
|
|
|
|
2023-03-07 15:34:10 +01:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2020-04-01 21:04:53 -07:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2010-10-26 14:21:56 -07:00
|
|
|
|
|
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2010-10-26 14:21:56 -07:00
|
|
|
|
|
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
2010-10-26 14:21:56 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
mm/filemap: clarify filemap_fault() comments for not uptodate case
The existing comments in filemap_fault() suggest that, after either a
minor fault has occurred and filemap_get_folio() found a folio in the page
cache, or a major fault arose and __filemap_get_folio(FGP_CREATE...) did
the job (having relied on do_sync_mmap_readahead() or filemap_read_folio()
to read in the folio), the only possible reason it could not be uptodate
is because of an error.
This is not so, as if, for instance, the fault occurred within a VMA which
had the VM_RAND_READ flag set (via madvise() with the MADV_RANDOM flag
specified), this would cause even synchronous readahead to fail to read in
the folio.
I confirmed this by dropping page caches and faulting in memory
madvise()'d this way, observing that this code path was reached on each
occasion.
Clarify the comments to include this case, and additionally update the
comment recently added around the invalidate lock logic to make it clear
the comment explicitly refers to the minor fault case.
In addition, while we're here, refer to folios rather than pages.
[lstoakes@gmail.com: correct identation as per Christopher's feedback]
Link: https://lkml.kernel.org/r/2c7014c0-6343-4e76-8697-3f84f54350bd@lucifer.local
Link: https://lkml.kernel.org/r/20230930231029.88196-1-lstoakes@gmail.com
Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-10-01 00:10:29 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
mm/filemap: clarify filemap_fault() comments for not uptodate case
The existing comments in filemap_fault() suggest that, after either a
minor fault has occurred and filemap_get_folio() found a folio in the page
cache, or a major fault arose and __filemap_get_folio(FGP_CREATE...) did
the job (having relied on do_sync_mmap_readahead() or filemap_read_folio()
to read in the folio), the only possible reason it could not be uptodate
is because of an error.
This is not so, as if, for instance, the fault occurred within a VMA which
had the VM_RAND_READ flag set (via madvise() with the MADV_RANDOM flag
specified), this would cause even synchronous readahead to fail to read in
the folio.
I confirmed this by dropping page caches and faulting in memory
madvise()'d this way, observing that this code path was reached on each
occasion.
Clarify the comments to include this case, and additionally update the
comment recently added around the invalidate lock logic to make it clear
the comment explicitly refers to the minor fault case.
In addition, while we're here, refer to folios rather than pages.
[lstoakes@gmail.com: correct identation as per Christopher's feedback]
Link: https://lkml.kernel.org/r/2c7014c0-6343-4e76-8697-3f84f54350bd@lucifer.local
Link: https://lkml.kernel.org/r/20230930231029.88196-1-lstoakes@gmail.com
Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-10-01 00:10:29 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
|
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
mm/filemap: clarify filemap_fault() comments for not uptodate case
The existing comments in filemap_fault() suggest that, after either a
minor fault has occurred and filemap_get_folio() found a folio in the page
cache, or a major fault arose and __filemap_get_folio(FGP_CREATE...) did
the job (having relied on do_sync_mmap_readahead() or filemap_read_folio()
to read in the folio), the only possible reason it could not be uptodate
is because of an error.
This is not so, as if, for instance, the fault occurred within a VMA which
had the VM_RAND_READ flag set (via madvise() with the MADV_RANDOM flag
specified), this would cause even synchronous readahead to fail to read in
the folio.
I confirmed this by dropping page caches and faulting in memory
madvise()'d this way, observing that this code path was reached on each
occasion.
Clarify the comments to include this case, and additionally update the
comment recently added around the invalidate lock logic to make it clear
the comment explicitly refers to the minor fault case.
In addition, while we're here, refer to folios rather than pages.
[lstoakes@gmail.com: correct identation as per Christopher's feedback]
Link: https://lkml.kernel.org/r/2c7014c0-6343-4e76-8697-3f84f54350bd@lucifer.local
Link: https://lkml.kernel.org/r/20230930231029.88196-1-lstoakes@gmail.com
Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-10-01 00:10:29 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2020-06-08 21:33:54 -07:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2009-06-16 15:31:25 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2007-10-31 09:19:46 -07:00
|
|
|
|
mm: fix fault vs invalidate race for linear mappings
Fix the race between invalidate_inode_pages and do_no_page.
Andrea Arcangeli identified a subtle race between invalidation of pages from
pagecache with userspace mappings, and do_no_page.
The issue is that invalidation has to shoot down all mappings to the page,
before it can be discarded from the pagecache. Between shooting down ptes to
a particular page, and actually dropping the struct page from the pagecache,
do_no_page from any process might fault on that page and establish a new
mapping to the page just before it gets discarded from the pagecache.
The most common case where such invalidation is used is in file truncation.
This case was catered for by doing a sort of open-coded seqlock between the
file's i_size, and its truncate_count.
Truncation will decrease i_size, then increment truncate_count before
unmapping userspace pages; do_no_page will read truncate_count, then find the
page if it is within i_size, and then check truncate_count under the page
table lock and back out and retry if it had subsequently been changed (ptl
will serialise against unmapping, and ensure a potentially updated
truncate_count is actually visible).
Complexity and documentation issues aside, the locking protocol fails in the
case where we would like to invalidate pagecache inside i_size. do_no_page
can come in anytime and filemap_nopage is not aware of the invalidation in
progress (as it is when it is outside i_size). The end result is that
dangling (->mapping == NULL) pages that appear to be from a particular file
may be mapped into userspace with nonsense data. Valid mappings to the same
place will see a different page.
Andrea implemented two working fixes, one using a real seqlock, another using
a page->flags bit. He also proposed using the page lock in do_no_page, but
that was initially considered too heavyweight. However, it is not a global or
per-file lock, and the page cacheline is modified in do_no_page to increment
_count and _mapcount anyway, so a further modification should not be a large
performance hit. Scalability is not an issue.
This patch implements this latter approach. ->nopage implementations return
with the page locked if it is possible for their underlying file to be
invalidated (in that case, they must set a special vm_flags bit to indicate
so). do_no_page only unlocks the page after setting up the mapping
completely. invalidation is excluded because it holds the page lock during
invalidation of each page (and ensures that the page is not mapped while
holding the lock).
This also allows significant simplifications in do_no_page, because we have
the page locked in the right place in the pagecache from the start.
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-07-19 01:46:57 -07:00
|
|
|
|
|
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
2007-07-19 01:47:05 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
2022-05-12 17:37:01 -04:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
mm: fix fault vs invalidate race for linear mappings
Fix the race between invalidate_inode_pages and do_no_page.
Andrea Arcangeli identified a subtle race between invalidation of pages from
pagecache with userspace mappings, and do_no_page.
The issue is that invalidation has to shoot down all mappings to the page,
before it can be discarded from the pagecache. Between shooting down ptes to
a particular page, and actually dropping the struct page from the pagecache,
do_no_page from any process might fault on that page and establish a new
mapping to the page just before it gets discarded from the pagecache.
The most common case where such invalidation is used is in file truncation.
This case was catered for by doing a sort of open-coded seqlock between the
file's i_size, and its truncate_count.
Truncation will decrease i_size, then increment truncate_count before
unmapping userspace pages; do_no_page will read truncate_count, then find the
page if it is within i_size, and then check truncate_count under the page
table lock and back out and retry if it had subsequently been changed (ptl
will serialise against unmapping, and ensure a potentially updated
truncate_count is actually visible).
Complexity and documentation issues aside, the locking protocol fails in the
case where we would like to invalidate pagecache inside i_size. do_no_page
can come in anytime and filemap_nopage is not aware of the invalidation in
progress (as it is when it is outside i_size). The end result is that
dangling (->mapping == NULL) pages that appear to be from a particular file
may be mapped into userspace with nonsense data. Valid mappings to the same
place will see a different page.
Andrea implemented two working fixes, one using a real seqlock, another using
a page->flags bit. He also proposed using the page lock in do_no_page, but
that was initially considered too heavyweight. However, it is not a global or
per-file lock, and the page cacheline is modified in do_no_page to increment
_count and _mapcount anyway, so a further modification should not be a large
performance hit. Scalability is not an issue.
This patch implements this latter approach. ->nopage implementations return
with the page locked if it is possible for their underlying file to be
invalidated (in that case, they must set a special vm_flags bit to indicate
so). do_no_page only unlocks the page after setting up the mapping
completely. invalidation is excluded because it holds the page lock during
invalidation of each page (and ensures that the page is not mapped while
holding the lock).
This also allows significant simplifications in do_no_page, because we have
the page locked in the right place in the pagecache from the start.
Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2007-07-19 01:46:57 -07:00
|
|
|
|
|
|
|
|
|
2005-12-15 14:28:17 -08:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2007-07-19 01:47:03 -07:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-06-08 21:33:54 -07:00
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-05-06 17:04:14 +01:00
|
|
|
|
2021-03-10 10:46:41 -05:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
filemap: drop the mmap_sem for all blocking operations
Currently we only drop the mmap_sem if there is contention on the page
lock. The idea is that we issue readahead and then go to lock the page
while it is under IO and we want to not hold the mmap_sem during the IO.
The problem with this is the assumption that the readahead does anything.
In the case that the box is under extreme memory or IO pressure we may end
up not reading anything at all for readahead, which means we will end up
reading in the page under the mmap_sem.
Even if the readahead does something, it could get throttled because of io
pressure on the system and the process is in a lower priority cgroup.
Holding the mmap_sem while doing IO is problematic because it can cause
system-wide priority inversions. Consider some large company that does a
lot of web traffic. This large company has load balancing logic in it's
core web server, cause some engineer thought this was a brilliant plan.
This load balancing logic gets statistics from /proc about the system,
which trip over processes mmap_sem for various reasons. Now the web
server application is in a protected cgroup, but these other processes may
not be, and if they are being throttled while their mmap_sem is held we'll
stall, and cause this nice death spiral.
Instead rework filemap fault path to drop the mmap sem at any point that
we may do IO or block for an extended period of time. This includes while
issuing readahead, locking the page, or needing to call ->readpage because
readahead did not occur. Then once we have a fully uptodate page we can
return with VM_FAULT_RETRY and come back again to find our nicely in-cache
page that was gotten outside of the mmap_sem.
This patch also adds a new helper for locking the page with the mmap_sem
dropped. This doesn't make sense currently as generally speaking if the
page is already locked it'll have been read in (unless there was an error)
before it was unlocked. However a forthcoming patchset will change this
with the ability to abort read-ahead bio's if necessary, making it more
likely that we could contend for a page lock and still have a not uptodate
page. This allows us to deal with this case by grabbing the lock and
issuing the IO without the mmap_sem held, and then returning
VM_FAULT_RETRY to come back around.
[josef@toxicpanda.com: v6]
Link: http://lkml.kernel.org/r/20181212152757.10017-1-josef@toxicpanda.com
[kirill@shutemov.name: fix race in filemap_fault()]
Link: http://lkml.kernel.org/r/20181228235106.okk3oastsnpxusxs@kshutemo-mobl1
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/20181211173801.29535-4-josef@toxicpanda.com
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Tested-by: syzbot+b437b5a429d680cf2217@syzkaller.appspotmail.com
Cc: Dave Chinner <david@fromorbit.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2019-03-13 11:44:22 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2007-07-19 01:46:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-16 19:39:39 +00:00
|
|
|
|
|
|
|
|
|
2014-04-07 15:37:19 -07:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-16 19:39:39 +00:00
|
|
|
|
|
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-01-16 19:39:39 +00:00
|
|
|
|
|
|
|
|
|
2025-05-12 10:57:12 +08:00
|
|
|
|
mm: filemap: coding style cleanup for filemap_map_pmd()
Patch series "Solve silent data loss caused by poisoned page cache (shmem/tmpfs)", v5.
When discussing the patch that splits page cache THP in order to offline
the poisoned page, Noaya mentioned there is a bigger problem [1] that
prevents this from working since the page cache page will be truncated
if uncorrectable errors happen. By looking this deeper it turns out
this approach (truncating poisoned page) may incur silent data loss for
all non-readonly filesystems if the page is dirty. It may be worse for
in-memory filesystem, e.g. shmem/tmpfs since the data blocks are
actually gone.
To solve this problem we could keep the poisoned dirty page in page
cache then notify the users on any later access, e.g. page fault,
read/write, etc. The clean page could be truncated as is since they can
be reread from disk later on.
The consequence is the filesystems may find poisoned page and manipulate
it as healthy page since all the filesystems actually don't check if the
page is poisoned or not in all the relevant paths except page fault. In
general, we need make the filesystems be aware of poisoned page before
we could keep the poisoned page in page cache in order to solve the data
loss problem.
To make filesystems be aware of poisoned page we should consider:
- The page should be not written back: clearing dirty flag could
prevent from writeback.
- The page should not be dropped (it shows as a clean page) by drop
caches or other callers: the refcount pin from hwpoison could prevent
from invalidating (called by cache drop, inode cache shrinking, etc),
but it doesn't avoid invalidation in DIO path.
- The page should be able to get truncated/hole punched/unlinked: it
works as it is.
- Notify users when the page is accessed, e.g. read/write, page fault
and other paths (compression, encryption, etc).
The scope of the last one is huge since almost all filesystems need do
it once a page is returned from page cache lookup. There are a couple
of options to do it:
1. Check hwpoison flag for every path, the most straightforward way.
2. Return NULL for poisoned page from page cache lookup, the most
callsites check if NULL is returned, this should have least work I
think. But the error handling in filesystems just return -ENOMEM,
the error code will incur confusion to the users obviously.
3. To improve #2, we could return error pointer, e.g. ERR_PTR(-EIO),
but this will involve significant amount of code change as well
since all the paths need check if the pointer is ERR or not just
like option #1.
I did prototypes for both #1 and #3, but it seems #3 may require more
changes than #1. For #3 ERR_PTR will be returned so all the callers
need to check the return value otherwise invalid pointer may be
dereferenced, but not all callers really care about the content of the
page, for example, partial truncate which just sets the truncated range
in one page to 0. So for such paths it needs additional modification if
ERR_PTR is returned. And if the callers have their own way to handle
the problematic pages we need to add a new FGP flag to tell FGP
functions to return the pointer to the page.
It may happen very rarely, but once it happens the consequence (data
corruption) could be very bad and it is very hard to debug. It seems
this problem had been slightly discussed before, but seems no action was
taken at that time. [2]
As the aforementioned investigation, it needs huge amount of work to
solve the potential data loss for all filesystems. But it is much
easier for in-memory filesystems and such filesystems actually suffer
more than others since even the data blocks are gone due to truncating.
So this patchset starts from shmem/tmpfs by taking option #1.
TODO:
* The unpoison has been broken since commit 0ed950d1f281 ("mm,hwpoison: make
get_hwpoison_page() call get_any_page()"), and this patch series make
refcount check for unpoisoning shmem page fail.
* Expand to other filesystems. But I haven't heard feedback from filesystem
developers yet.
Patch breakdown:
Patch #1: cleanup, depended by patch #2
Patch #2: fix THP with hwpoisoned subpage(s) PMD map bug
Patch #3: coding style cleanup
Patch #4: refactor and preparation.
Patch #5: keep the poisoned page in page cache and handle such case for all
the paths.
Patch #6: the previous patches unblock page cache THP split, so this patch
add page cache THP split support.
This patch (of 4):
A minor cleanup to the indent.
Link: https://lkml.kernel.org/r/20211020210755.23964-1-shy828301@gmail.com
Link: https://lkml.kernel.org/r/20211020210755.23964-4-shy828301@gmail.com
Signed-off-by: Yang Shi <shy828301@gmail.com>
Reviewed-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Peter Xu <peterx@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-11-05 13:41:04 -07:00
|
|
|
|
|
|
|
|
|
2023-01-16 19:39:39 +00:00
|
|
|
|
mm: filemap: coding style cleanup for filemap_map_pmd()
Patch series "Solve silent data loss caused by poisoned page cache (shmem/tmpfs)", v5.
When discussing the patch that splits page cache THP in order to offline
the poisoned page, Noaya mentioned there is a bigger problem [1] that
prevents this from working since the page cache page will be truncated
if uncorrectable errors happen. By looking this deeper it turns out
this approach (truncating poisoned page) may incur silent data loss for
all non-readonly filesystems if the page is dirty. It may be worse for
in-memory filesystem, e.g. shmem/tmpfs since the data blocks are
actually gone.
To solve this problem we could keep the poisoned dirty page in page
cache then notify the users on any later access, e.g. page fault,
read/write, etc. The clean page could be truncated as is since they can
be reread from disk later on.
The consequence is the filesystems may find poisoned page and manipulate
it as healthy page since all the filesystems actually don't check if the
page is poisoned or not in all the relevant paths except page fault. In
general, we need make the filesystems be aware of poisoned page before
we could keep the poisoned page in page cache in order to solve the data
loss problem.
To make filesystems be aware of poisoned page we should consider:
- The page should be not written back: clearing dirty flag could
prevent from writeback.
- The page should not be dropped (it shows as a clean page) by drop
caches or other callers: the refcount pin from hwpoison could prevent
from invalidating (called by cache drop, inode cache shrinking, etc),
but it doesn't avoid invalidation in DIO path.
- The page should be able to get truncated/hole punched/unlinked: it
works as it is.
- Notify users when the page is accessed, e.g. read/write, page fault
and other paths (compression, encryption, etc).
The scope of the last one is huge since almost all filesystems need do
it once a page is returned from page cache lookup. There are a couple
of options to do it:
1. Check hwpoison flag for every path, the most straightforward way.
2. Return NULL for poisoned page from page cache lookup, the most
callsites check if NULL is returned, this should have least work I
think. But the error handling in filesystems just return -ENOMEM,
the error code will incur confusion to the users obviously.
3. To improve #2, we could return error pointer, e.g. ERR_PTR(-EIO),
but this will involve significant amount of code change as well
since all the paths need check if the pointer is ERR or not just
like option #1.
I did prototypes for both #1 and #3, but it seems #3 may require more
changes than #1. For #3 ERR_PTR will be returned so all the callers
need to check the return value otherwise invalid pointer may be
dereferenced, but not all callers really care about the content of the
page, for example, partial truncate which just sets the truncated range
in one page to 0. So for such paths it needs additional modification if
ERR_PTR is returned. And if the callers have their own way to handle
the problematic pages we need to add a new FGP flag to tell FGP
functions to return the pointer to the page.
It may happen very rarely, but once it happens the consequence (data
corruption) could be very bad and it is very hard to debug. It seems
this problem had been slightly discussed before, but seems no action was
taken at that time. [2]
As the aforementioned investigation, it needs huge amount of work to
solve the potential data loss for all filesystems. But it is much
easier for in-memory filesystems and such filesystems actually suffer
more than others since even the data blocks are gone due to truncating.
So this patchset starts from shmem/tmpfs by taking option #1.
TODO:
* The unpoison has been broken since commit 0ed950d1f281 ("mm,hwpoison: make
get_hwpoison_page() call get_any_page()"), and this patch series make
refcount check for unpoisoning shmem page fail.
* Expand to other filesystems. But I haven't heard feedback from filesystem
developers yet.
Patch breakdown:
Patch #1: cleanup, depended by patch #2
Patch #2: fix THP with hwpoisoned subpage(s) PMD map bug
Patch #3: coding style cleanup
Patch #4: refactor and preparation.
Patch #5: keep the poisoned page in page cache and handle such case for all
the paths.
Patch #6: the previous patches unblock page cache THP split, so this patch
add page cache THP split support.
This patch (of 4):
A minor cleanup to the indent.
Link: https://lkml.kernel.org/r/20211020210755.23964-1-shy828301@gmail.com
Link: https://lkml.kernel.org/r/20211020210755.23964-4-shy828301@gmail.com
Signed-off-by: Yang Shi <shy828301@gmail.com>
Reviewed-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Peter Xu <peterx@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-11-05 13:41:04 -07:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm: fix oops when filemap_map_pmd() without prealloc_pte
syzbot reports oops in lockdep's __lock_acquire(), called from
__pte_offset_map_lock() called from filemap_map_pages(); or when I run the
repro, the oops comes in pmd_install(), called from filemap_map_pmd()
called from filemap_map_pages(), just before the __pte_offset_map_lock().
The problem is that filemap_map_pmd() has been assuming that when it finds
pmd_none(), a page table has already been prepared in prealloc_pte; and
indeed do_fault_around() has been careful to preallocate one there, when
it finds pmd_none(): but what if *pmd became none in between?
My 6.6 mods in mm/khugepaged.c, avoiding mmap_lock for write, have made it
easy for *pmd to be cleared while servicing a page fault; but even before
those, a huge *pmd might be zapped while a fault is serviced.
The difference in symptomatic stack traces comes from the "memory model"
in use: pmd_install() uses pmd_populate() uses page_to_pfn(): in some
models that is strict, and will oops on the NULL prealloc_pte; in other
models, it will construct a bogus value to be populated into *pmd, then
__pte_offset_map_lock() oops when trying to access split ptlock pointer
(or some other symptom in normal case of ptlock embedded not pointer).
Link: https://lore.kernel.org/linux-mm/20231115065506.19780-1-jose.pekkarinen@foxhound.fi/
Link: https://lkml.kernel.org/r/6ed0c50c-78ef-0719-b3c5-60c0c010431c@google.com
Fixes: f9ce0be71d1f ("mm: Cleanup faultaround and finish_fault() codepaths")
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-and-tested-by: syzbot+89edd67979b52675ddec@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/linux-mm/0000000000005e44550608a0806c@google.com/
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Jann Horn <jannh@google.com>,
Cc: José Pekkarinen <jose.pekkarinen@foxhound.fi>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: <stable@vger.kernel.org> [5.12+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-11-17 00:49:18 -08:00
|
|
|
|
2021-11-05 13:38:38 -07:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
|
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-12 23:33:43 -05:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2021-03-12 23:33:43 -05:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2021-03-12 23:33:43 -05:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2024-06-25 13:53:50 -07:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2024-11-29 13:53:03 +01:00
|
|
|
|
|
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2021-03-12 23:33:43 -05:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2021-03-12 23:33:43 -05:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2021-03-12 23:33:43 -05:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2021-03-12 23:33:43 -05:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2021-03-12 23:33:43 -05:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-03-12 23:46:45 -05:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2021-03-12 23:33:43 -05:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2021-03-12 23:33:43 -05:00
|
|
|
|
|
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-09-14 21:47:41 +08:00
|
|
|
|
mm/memory: do not populate page table entries beyond i_size
Patch series "Fix SIGBUS semantics with large folios", v3.
Accessing memory within a VMA, but beyond i_size rounded up to the next
page size, is supposed to generate SIGBUS.
Darrick reported[1] an xfstests regression in v6.18-rc1. generic/749
failed due to missing SIGBUS. This was caused by my recent changes that
try to fault in the whole folio where possible:
19773df031bc ("mm/fault: try to map the entire file folio in finish_fault()")
357b92761d94 ("mm/filemap: map entire large folio faultaround")
These changes did not consider i_size when setting up PTEs, leading to
xfstest breakage.
However, the problem has been present in the kernel for a long time -
since huge tmpfs was introduced in 2016. The kernel happily maps
PMD-sized folios as PMD without checking i_size. And huge=always tmpfs
allocates PMD-size folios on any writes.
I considered this corner case when I implemented a large tmpfs, and my
conclusion was that no one in their right mind should rely on receiving a
SIGBUS signal when accessing beyond i_size. I cannot imagine how it could
be useful for the workload.
But apparently filesystem folks care a lot about preserving strict SIGBUS
semantics.
Generic/749 was introduced last year with reference to POSIX, but no real
workloads were mentioned. It also acknowledged the tmpfs deviation from
the test case.
POSIX indeed says[3]:
References within the address range starting at pa and
continuing for len bytes to whole pages following the end of an
object shall result in delivery of a SIGBUS signal.
The patchset fixes the regression introduced by recent changes as well as
more subtle SIGBUS breakage due to split failure on truncation.
This patch (of 2):
Accesses within VMA, but beyond i_size rounded up to PAGE_SIZE are
supposed to generate SIGBUS.
Recent changes attempted to fault in full folio where possible. They did
not respect i_size, which led to populating PTEs beyond i_size and
breaking SIGBUS semantics.
Darrick reported generic/749 breakage because of this.
However, the problem existed before the recent changes. With huge=always
tmpfs, any write to a file leads to PMD-size allocation. Following the
fault-in of the folio will install PMD mapping regardless of i_size.
Fix filemap_map_pages() and finish_fault() to not install:
- PTEs beyond i_size;
- PMD mappings across i_size;
Make an exception for shmem/tmpfs that for long time intentionally
mapped with PMDs across i_size.
Link: https://lkml.kernel.org/r/20251027115636.82382-1-kirill@shutemov.name
Link: https://lkml.kernel.org/r/20251027115636.82382-2-kirill@shutemov.name
Signed-off-by: Kiryl Shutsemau <kas@kernel.org>
Fixes: 6795801366da ("xfs: Support large folios")
Reported-by: "Darrick J. Wong" <djwong@kernel.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-10-27 11:56:35 +00:00
|
|
|
|
|
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2025-09-04 21:27:37 +08:00
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
|
|
|
|
|
2023-08-02 16:14:05 +01:00
|
|
|
|
|
|
|
|
|
2025-09-23 12:07:10 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm/memory: do not populate page table entries beyond i_size
Patch series "Fix SIGBUS semantics with large folios", v3.
Accessing memory within a VMA, but beyond i_size rounded up to the next
page size, is supposed to generate SIGBUS.
Darrick reported[1] an xfstests regression in v6.18-rc1. generic/749
failed due to missing SIGBUS. This was caused by my recent changes that
try to fault in the whole folio where possible:
19773df031bc ("mm/fault: try to map the entire file folio in finish_fault()")
357b92761d94 ("mm/filemap: map entire large folio faultaround")
These changes did not consider i_size when setting up PTEs, leading to
xfstest breakage.
However, the problem has been present in the kernel for a long time -
since huge tmpfs was introduced in 2016. The kernel happily maps
PMD-sized folios as PMD without checking i_size. And huge=always tmpfs
allocates PMD-size folios on any writes.
I considered this corner case when I implemented a large tmpfs, and my
conclusion was that no one in their right mind should rely on receiving a
SIGBUS signal when accessing beyond i_size. I cannot imagine how it could
be useful for the workload.
But apparently filesystem folks care a lot about preserving strict SIGBUS
semantics.
Generic/749 was introduced last year with reference to POSIX, but no real
workloads were mentioned. It also acknowledged the tmpfs deviation from
the test case.
POSIX indeed says[3]:
References within the address range starting at pa and
continuing for len bytes to whole pages following the end of an
object shall result in delivery of a SIGBUS signal.
The patchset fixes the regression introduced by recent changes as well as
more subtle SIGBUS breakage due to split failure on truncation.
This patch (of 2):
Accesses within VMA, but beyond i_size rounded up to PAGE_SIZE are
supposed to generate SIGBUS.
Recent changes attempted to fault in full folio where possible. They did
not respect i_size, which led to populating PTEs beyond i_size and
breaking SIGBUS semantics.
Darrick reported generic/749 breakage because of this.
However, the problem existed before the recent changes. With huge=always
tmpfs, any write to a file leads to PMD-size allocation. Following the
fault-in of the folio will install PMD mapping regardless of i_size.
Fix filemap_map_pages() and finish_fault() to not install:
- PTEs beyond i_size;
- PMD mappings across i_size;
Make an exception for shmem/tmpfs that for long time intentionally
mapped with PMDs across i_size.
Link: https://lkml.kernel.org/r/20251027115636.82382-1-kirill@shutemov.name
Link: https://lkml.kernel.org/r/20251027115636.82382-2-kirill@shutemov.name
Signed-off-by: Kiryl Shutsemau <kas@kernel.org>
Fixes: 6795801366da ("xfs: Support large folios")
Reported-by: "Darrick J. Wong" <djwong@kernel.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-10-27 11:56:35 +00:00
|
|
|
|
2025-09-23 12:07:10 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
2023-08-02 16:14:05 +01:00
|
|
|
|
|
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
2024-03-22 17:35:55 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-11-14 15:49:45 +00:00
|
|
|
|
2023-08-02 16:14:05 +01:00
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
2023-08-02 16:14:05 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-04-12 14:47:51 +08:00
|
|
|
|
2025-09-04 21:27:37 +08:00
|
|
|
|
|
|
|
|
|
2023-09-20 04:53:35 +01:00
|
|
|
|
2023-08-02 16:14:05 +01:00
|
|
|
|
|
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
2023-08-02 16:14:05 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-04-12 14:47:51 +08:00
|
|
|
|
2025-09-04 21:27:37 +08:00
|
|
|
|
|
|
|
|
|
2023-09-20 04:53:35 +01:00
|
|
|
|
2023-08-02 16:14:05 +01:00
|
|
|
|
|
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
2023-08-02 16:14:05 +01:00
|
|
|
|
2025-09-04 21:27:37 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-09-14 21:47:41 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-06-09 10:27:25 +01:00
|
|
|
|
2023-09-14 21:47:41 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-09-04 21:27:37 +08:00
|
|
|
|
2023-09-14 21:47:41 +08:00
|
|
|
|
2024-03-22 17:35:55 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-09-14 21:47:41 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-09-04 21:27:37 +08:00
|
|
|
|
2023-09-14 21:47:41 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-04-12 14:47:51 +08:00
|
|
|
|
2025-09-04 21:27:37 +08:00
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
2025-09-04 21:27:37 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-04-07 15:37:19 -07:00
|
|
|
|
2024-08-22 15:50:13 +02:00
|
|
|
|
2021-01-14 15:24:19 +00:00
|
|
|
|
2018-05-17 00:08:30 -04:00
|
|
|
|
2021-03-12 23:46:45 -05:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2024-04-12 14:47:51 +08:00
|
|
|
|
2025-06-09 10:27:25 +01:00
|
|
|
|
|
|
|
|
|
mm/memory: do not populate page table entries beyond i_size
Patch series "Fix SIGBUS semantics with large folios", v3.
Accessing memory within a VMA, but beyond i_size rounded up to the next
page size, is supposed to generate SIGBUS.
Darrick reported[1] an xfstests regression in v6.18-rc1. generic/749
failed due to missing SIGBUS. This was caused by my recent changes that
try to fault in the whole folio where possible:
19773df031bc ("mm/fault: try to map the entire file folio in finish_fault()")
357b92761d94 ("mm/filemap: map entire large folio faultaround")
These changes did not consider i_size when setting up PTEs, leading to
xfstest breakage.
However, the problem has been present in the kernel for a long time -
since huge tmpfs was introduced in 2016. The kernel happily maps
PMD-sized folios as PMD without checking i_size. And huge=always tmpfs
allocates PMD-size folios on any writes.
I considered this corner case when I implemented a large tmpfs, and my
conclusion was that no one in their right mind should rely on receiving a
SIGBUS signal when accessing beyond i_size. I cannot imagine how it could
be useful for the workload.
But apparently filesystem folks care a lot about preserving strict SIGBUS
semantics.
Generic/749 was introduced last year with reference to POSIX, but no real
workloads were mentioned. It also acknowledged the tmpfs deviation from
the test case.
POSIX indeed says[3]:
References within the address range starting at pa and
continuing for len bytes to whole pages following the end of an
object shall result in delivery of a SIGBUS signal.
The patchset fixes the regression introduced by recent changes as well as
more subtle SIGBUS breakage due to split failure on truncation.
This patch (of 2):
Accesses within VMA, but beyond i_size rounded up to PAGE_SIZE are
supposed to generate SIGBUS.
Recent changes attempted to fault in full folio where possible. They did
not respect i_size, which led to populating PTEs beyond i_size and
breaking SIGBUS semantics.
Darrick reported generic/749 breakage because of this.
However, the problem existed before the recent changes. With huge=always
tmpfs, any write to a file leads to PMD-size allocation. Following the
fault-in of the folio will install PMD mapping regardless of i_size.
Fix filemap_map_pages() and finish_fault() to not install:
- PTEs beyond i_size;
- PMD mappings across i_size;
Make an exception for shmem/tmpfs that for long time intentionally
mapped with PMDs across i_size.
Link: https://lkml.kernel.org/r/20251027115636.82382-1-kirill@shutemov.name
Link: https://lkml.kernel.org/r/20251027115636.82382-2-kirill@shutemov.name
Signed-off-by: Kiryl Shutsemau <kas@kernel.org>
Fixes: 6795801366da ("xfs: Support large folios")
Reported-by: "Darrick J. Wong" <djwong@kernel.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-10-27 11:56:35 +00:00
|
|
|
|
2014-04-07 15:37:19 -07:00
|
|
|
|
|
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
2021-03-12 23:46:45 -05:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2014-04-07 15:37:19 -07:00
|
|
|
|
mm/memory: do not populate page table entries beyond i_size
Patch series "Fix SIGBUS semantics with large folios", v3.
Accessing memory within a VMA, but beyond i_size rounded up to the next
page size, is supposed to generate SIGBUS.
Darrick reported[1] an xfstests regression in v6.18-rc1. generic/749
failed due to missing SIGBUS. This was caused by my recent changes that
try to fault in the whole folio where possible:
19773df031bc ("mm/fault: try to map the entire file folio in finish_fault()")
357b92761d94 ("mm/filemap: map entire large folio faultaround")
These changes did not consider i_size when setting up PTEs, leading to
xfstest breakage.
However, the problem has been present in the kernel for a long time -
since huge tmpfs was introduced in 2016. The kernel happily maps
PMD-sized folios as PMD without checking i_size. And huge=always tmpfs
allocates PMD-size folios on any writes.
I considered this corner case when I implemented a large tmpfs, and my
conclusion was that no one in their right mind should rely on receiving a
SIGBUS signal when accessing beyond i_size. I cannot imagine how it could
be useful for the workload.
But apparently filesystem folks care a lot about preserving strict SIGBUS
semantics.
Generic/749 was introduced last year with reference to POSIX, but no real
workloads were mentioned. It also acknowledged the tmpfs deviation from
the test case.
POSIX indeed says[3]:
References within the address range starting at pa and
continuing for len bytes to whole pages following the end of an
object shall result in delivery of a SIGBUS signal.
The patchset fixes the regression introduced by recent changes as well as
more subtle SIGBUS breakage due to split failure on truncation.
This patch (of 2):
Accesses within VMA, but beyond i_size rounded up to PAGE_SIZE are
supposed to generate SIGBUS.
Recent changes attempted to fault in full folio where possible. They did
not respect i_size, which led to populating PTEs beyond i_size and
breaking SIGBUS semantics.
Darrick reported generic/749 breakage because of this.
However, the problem existed before the recent changes. With huge=always
tmpfs, any write to a file leads to PMD-size allocation. Following the
fault-in of the folio will install PMD mapping regardless of i_size.
Fix filemap_map_pages() and finish_fault() to not install:
- PTEs beyond i_size;
- PMD mappings across i_size;
Make an exception for shmem/tmpfs that for long time intentionally
mapped with PMDs across i_size.
Link: https://lkml.kernel.org/r/20251027115636.82382-1-kirill@shutemov.name
Link: https://lkml.kernel.org/r/20251027115636.82382-2-kirill@shutemov.name
Signed-off-by: Kiryl Shutsemau <kas@kernel.org>
Fixes: 6795801366da ("xfs: Support large folios")
Reported-by: "Darrick J. Wong" <djwong@kernel.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-10-27 11:56:35 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-04-07 15:37:19 -07:00
|
|
|
|
2021-01-14 15:24:19 +00:00
|
|
|
|
|
|
|
|
|
2023-06-08 18:11:29 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-04-12 14:47:51 +08:00
|
|
|
|
|
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
2016-07-26 15:25:23 -07:00
|
|
|
|
2021-01-14 15:24:19 +00:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2018-05-17 00:08:30 -04:00
|
|
|
|
2023-09-21 16:15:35 +08:00
|
|
|
|
2023-08-02 16:14:02 +01:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2023-09-14 21:47:41 +08:00
|
|
|
|
|
|
|
|
|
2024-04-12 14:47:51 +08:00
|
|
|
|
2023-09-14 21:47:41 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm/memory: do not populate page table entries beyond i_size
Patch series "Fix SIGBUS semantics with large folios", v3.
Accessing memory within a VMA, but beyond i_size rounded up to the next
page size, is supposed to generate SIGBUS.
Darrick reported[1] an xfstests regression in v6.18-rc1. generic/749
failed due to missing SIGBUS. This was caused by my recent changes that
try to fault in the whole folio where possible:
19773df031bc ("mm/fault: try to map the entire file folio in finish_fault()")
357b92761d94 ("mm/filemap: map entire large folio faultaround")
These changes did not consider i_size when setting up PTEs, leading to
xfstest breakage.
However, the problem has been present in the kernel for a long time -
since huge tmpfs was introduced in 2016. The kernel happily maps
PMD-sized folios as PMD without checking i_size. And huge=always tmpfs
allocates PMD-size folios on any writes.
I considered this corner case when I implemented a large tmpfs, and my
conclusion was that no one in their right mind should rely on receiving a
SIGBUS signal when accessing beyond i_size. I cannot imagine how it could
be useful for the workload.
But apparently filesystem folks care a lot about preserving strict SIGBUS
semantics.
Generic/749 was introduced last year with reference to POSIX, but no real
workloads were mentioned. It also acknowledged the tmpfs deviation from
the test case.
POSIX indeed says[3]:
References within the address range starting at pa and
continuing for len bytes to whole pages following the end of an
object shall result in delivery of a SIGBUS signal.
The patchset fixes the regression introduced by recent changes as well as
more subtle SIGBUS breakage due to split failure on truncation.
This patch (of 2):
Accesses within VMA, but beyond i_size rounded up to PAGE_SIZE are
supposed to generate SIGBUS.
Recent changes attempted to fault in full folio where possible. They did
not respect i_size, which led to populating PTEs beyond i_size and
breaking SIGBUS semantics.
Darrick reported generic/749 breakage because of this.
However, the problem existed before the recent changes. With huge=always
tmpfs, any write to a file leads to PMD-size allocation. Following the
fault-in of the folio will install PMD mapping regardless of i_size.
Fix filemap_map_pages() and finish_fault() to not install:
- PTEs beyond i_size;
- PMD mappings across i_size;
Make an exception for shmem/tmpfs that for long time intentionally
mapped with PMDs across i_size.
Link: https://lkml.kernel.org/r/20251027115636.82382-1-kirill@shutemov.name
Link: https://lkml.kernel.org/r/20251027115636.82382-2-kirill@shutemov.name
Signed-off-by: Kiryl Shutsemau <kas@kernel.org>
Fixes: 6795801366da ("xfs: Support large folios")
Reported-by: "Darrick J. Wong" <djwong@kernel.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Dave Chinner <david@fromorbit.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Liam Howlett <liam.howlett@oracle.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Mike Rapoport <rppt@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2025-10-27 11:56:35 +00:00
|
|
|
|
|
|
|
|
|
2020-11-24 18:48:26 +00:00
|
|
|
|
2021-03-12 23:46:45 -05:00
|
|
|
|
2023-09-14 21:47:41 +08:00
|
|
|
|
2024-04-12 14:47:51 +08:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
filemap: add trace events for get_pages, map_pages, and fault
To allow precise tracking of page caches accessed, add new tracepoints
that trigger when a process actually accesses them.
The ureadahead program used by ChromeOS traces the disk access of programs
as they start up at boot up. It uses mincore(2) or the
'mm_filemap_add_to_page_cache' trace event to accomplish this. It stores
this information in a "pack" file and on subsequent boots, it will read
the pack file and call readahead(2) on the information so that disk
storage can be loaded into RAM before the applications actually need it.
A problem we see is that due to the kernel's readahead algorithm that can
aggressively pull in more data than needed (to try and accomplish the same
goal) and this data is also recorded. The end result is that the pack
file contains a lot of pages on disk that are never actually used.
Calling readahead(2) on these unused pages can slow down the system boot
up times.
To solve this, add 3 new trace events, get_pages, map_pages, and fault.
These will be used to trace the pages are not only pulled in from disk,
but are actually used by the application. Only those pages will be stored
in the pack file, and this helps out the performance of boot up.
With the combination of these 3 new trace events and
mm_filemap_add_to_page_cache, we observed a reduction in the pack file by
7.3% - 20% on ChromeOS varying by device.
Link: https://lkml.kernel.org/r/20240813100312.3930505-1-takayas@chromium.org
Signed-off-by: Takaya Saeki <takayas@chromium.org>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Reviewed-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Cc: Junichi Uekawa <uekawa@chromium.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2024-08-13 10:03:12 +00:00
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2014-04-07 15:37:19 -07:00
|
|
|
|
2023-09-14 21:47:41 +08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-19 15:19:23 +03:00
|
|
|
|
2014-04-07 15:37:19 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2018-06-07 17:08:00 -07:00
|
|
|
|
2012-06-12 16:20:29 +02:00
|
|
|
|
2020-11-16 14:33:37 +01:00
|
|
|
|
2021-03-12 23:57:44 -05:00
|
|
|
|
2018-06-07 17:08:00 -07:00
|
|
|
|
2012-06-12 16:20:29 +02:00
|
|
|
|
2020-11-16 14:33:37 +01:00
|
|
|
|
2017-02-24 14:56:41 -08:00
|
|
|
|
2021-03-12 23:57:44 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2012-06-12 16:20:29 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2012-06-12 16:20:37 +02:00
|
|
|
|
2021-03-12 23:57:44 -05:00
|
|
|
|
2012-06-12 16:20:37 +02:00
|
|
|
|
2021-03-12 23:57:44 -05:00
|
|
|
|
2012-06-12 16:20:37 +02:00
|
|
|
|
2021-03-12 23:57:44 -05:00
|
|
|
|
|
|
|
|
|
2012-06-12 16:20:29 +02:00
|
|
|
|
2020-11-16 14:33:37 +01:00
|
|
|
|
2012-06-12 16:20:29 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2009-09-27 22:29:37 +04:00
|
|
|
|
2007-07-19 01:46:59 -07:00
|
|
|
|
2014-04-07 15:37:19 -07:00
|
|
|
|
2012-06-12 16:20:29 +02:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-04 18:40:12 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-04-29 11:53:28 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-06-16 20:33:26 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm: drop the assumption that VM_SHARED always implies writable
Patch series "permit write-sealed memfd read-only shared mappings", v4.
The man page for fcntl() describing memfd file seals states the following
about F_SEAL_WRITE:-
Furthermore, trying to create new shared, writable memory-mappings via
mmap(2) will also fail with EPERM.
With emphasis on 'writable'. In turns out in fact that currently the
kernel simply disallows all new shared memory mappings for a memfd with
F_SEAL_WRITE applied, rendering this documentation inaccurate.
This matters because users are therefore unable to obtain a shared mapping
to a memfd after write sealing altogether, which limits their usefulness.
This was reported in the discussion thread [1] originating from a bug
report [2].
This is a product of both using the struct address_space->i_mmap_writable
atomic counter to determine whether writing may be permitted, and the
kernel adjusting this counter when any VM_SHARED mapping is performed and
more generally implicitly assuming VM_SHARED implies writable.
It seems sensible that we should only update this mapping if VM_MAYWRITE
is specified, i.e. whether it is possible that this mapping could at any
point be written to.
If we do so then all we need to do to permit write seals to function as
documented is to clear VM_MAYWRITE when mapping read-only. It turns out
this functionality already exists for F_SEAL_FUTURE_WRITE - we can
therefore simply adapt this logic to do the same for F_SEAL_WRITE.
We then hit a chicken and egg situation in mmap_region() where the check
for VM_MAYWRITE occurs before we are able to clear this flag. To work
around this, perform this check after we invoke call_mmap(), with careful
consideration of error paths.
Thanks to Andy Lutomirski for the suggestion!
[1]:https://lore.kernel.org/all/20230324133646.16101dfa666f253c4715d965@linux-foundation.org/
[2]:https://bugzilla.kernel.org/show_bug.cgi?id=217238
This patch (of 3):
There is a general assumption that VMAs with the VM_SHARED flag set are
writable. If the VM_MAYWRITE flag is not set, then this is simply not the
case.
Update those checks which affect the struct address_space->i_mmap_writable
field to explicitly test for this by introducing
[vma_]is_shared_maywrite() helper functions.
This remains entirely conservative, as the lack of VM_MAYWRITE guarantees
that the VMA cannot be written to.
Link: https://lkml.kernel.org/r/cover.1697116581.git.lstoakes@gmail.com
Link: https://lkml.kernel.org/r/d978aefefa83ec42d18dfa964ad180dbcde34795.1697116581.git.lstoakes@gmail.com
Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
Suggested-by: Andy Lutomirski <luto@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Muchun Song <muchun.song@linux.dev>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-10-12 18:04:28 +01:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-06-16 20:33:26 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2018-10-26 15:04:03 -07:00
|
|
|
|
2018-04-13 15:35:27 -07:00
|
|
|
|
2018-10-26 15:04:03 -07:00
|
|
|
|
2018-04-13 15:35:27 -07:00
|
|
|
|
2021-05-04 18:40:12 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-06-16 20:33:26 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-04 18:40:12 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-06-16 20:33:26 +01:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2018-04-13 15:35:27 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2025-06-16 20:33:26 +01:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2025-06-16 20:33:26 +01:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
2022-05-01 21:39:29 -04:00
|
|
|
|
2014-04-03 14:48:18 -07:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2022-05-08 15:07:11 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
2023-03-07 15:34:10 +01:00
|
|
|
|
2024-08-22 15:50:10 +02:00
|
|
|
|
|
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
2007-10-16 01:24:57 -07:00
|
|
|
|
2024-08-22 15:50:10 +02:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
2007-10-16 01:24:57 -07:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
2007-10-16 01:24:57 -07:00
|
|
|
|
|
|
|
|
|
2017-12-04 04:02:00 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2016-03-15 14:55:36 -07:00
|
|
|
|
2022-05-12 17:12:21 -04:00
|
|
|
|
2016-03-15 14:55:36 -07:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2021-12-23 15:17:28 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2016-03-15 14:55:39 -07:00
|
|
|
|
2021-12-23 15:17:28 -05:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2016-03-15 14:55:36 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2016-03-15 14:55:39 -07:00
|
|
|
|
|
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
mm/filemap.c: clear page error before actual read
Mount failure issue happens under the scenario: Application forked dozens
of threads to mount the same number of cramfs images separately in docker,
but several mounts failed with high probability. Mount failed due to the
checking result of the page(read from the superblock of loop dev) is not
uptodate after wait_on_page_locked(page) returned in function cramfs_read:
wait_on_page_locked(page);
if (!PageUptodate(page)) {
...
}
The reason of the checking result of the page not uptodate: systemd-udevd
read the loopX dev before mount, because the status of loopX is Lo_unbound
at this time, so loop_make_request directly trigger the calling of io_end
handler end_buffer_async_read, which called SetPageError(page). So It
caused the page can't be set to uptodate in function
end_buffer_async_read:
if(page_uptodate && !PageError(page)) {
SetPageUptodate(page);
}
Then mount operation is performed, it used the same page which is just
accessed by systemd-udevd above, Because this page is not uptodate, it
will launch a actual read via submit_bh, then wait on this page by calling
wait_on_page_locked(page). When the I/O of the page done, io_end handler
end_buffer_async_read is called, because no one cleared the page
error(during the whole read path of mount), which is caused by
systemd-udevd reading, so this page is still in "PageError" status, which
can't be set to uptodate in function end_buffer_async_read, then caused
mount failure.
But sometimes mount succeed even through systemd-udeved read loopX dev
just before, The reason is systemd-udevd launched other loopX read just
between step 3.1 and 3.2, the steps as below:
1, loopX dev default status is Lo_unbound;
2, systemd-udved read loopX dev (page is set to PageError);
3, mount operation
1) set loopX status to Lo_bound;
==>systemd-udevd read loopX dev<==
2) read loopX dev(page has no error)
3) mount succeed
As the loopX dev status is set to Lo_bound after step 3.1, so the other
loopX dev read by systemd-udevd will go through the whole I/O stack, part
of the call trace as below:
SYS_read
vfs_read
do_sync_read
blkdev_aio_read
generic_file_aio_read
do_generic_file_read:
ClearPageError(page);
mapping->a_ops->readpage(filp, page);
here, mapping->a_ops->readpage() is blkdev_readpage. In latest kernel,
some function name changed, the call trace as below:
blkdev_read_iter
generic_file_read_iter
generic_file_buffered_read:
/*
* A previous I/O error may have been due to temporary
* failures, eg. mutipath errors.
* Pg_error will be set again if readpage fails.
*/
ClearPageError(page);
/* Start the actual read. The read will unlock the page*/
error=mapping->a_ops->readpage(flip, page);
We can see ClearPageError(page) is called before the actual read,
then the read in step 3.2 succeed.
This patch is to add the calling of ClearPageError just before the actual
read of read path of cramfs mount. Without the patch, the call trace as
below when performing cramfs mount:
do_mount
cramfs_read
cramfs_blkdev_read
read_cache_page
do_read_cache_page:
filler(data, page);
or
mapping->a_ops->readpage(data, page);
With the patch, the call trace as below when performing mount:
do_mount
cramfs_read
cramfs_blkdev_read
read_cache_page:
do_read_cache_page:
ClearPageError(page); <== new add
filler(data, page);
or
mapping->a_ops->readpage(data, page);
With the patch, mount operation trigger the calling of
ClearPageError(page) before the actual read, the page has no error if no
additional page error happen when I/O done.
Signed-off-by: Xianting Tian <xianting_tian@126.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Jan Kara <jack@suse.cz>
Cc: <yubin@h3c.com>
Link: http://lkml.kernel.org/r/1583318844-22971-1-git-send-email-xianting_tian@126.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-04-01 21:04:47 -07:00
|
|
|
|
2022-05-12 17:12:21 -04:00
|
|
|
|
2022-05-12 17:37:01 -04:00
|
|
|
|
2022-05-12 17:47:06 -04:00
|
|
|
|
2022-05-12 17:12:21 -04:00
|
|
|
|
2022-05-12 17:47:06 -04:00
|
|
|
|
|
|
|
|
|
2022-05-12 17:12:21 -04:00
|
|
|
|
|
|
|
|
|
2016-03-15 14:55:36 -07:00
|
|
|
|
2007-05-09 13:42:20 +01:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
|
|
|
|
|
2007-05-06 14:49:04 -07:00
|
|
|
|
2010-01-27 09:20:03 -08:00
|
|
|
|
|
|
|
|
|
2022-05-01 21:39:29 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2010-01-27 09:20:03 -08:00
|
|
|
|
2022-05-01 21:39:29 -04:00
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
2022-05-01 21:39:29 -04:00
|
|
|
|
|
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
2022-05-01 21:39:29 -04:00
|
|
|
|
|
|
|
|
|
2010-01-27 09:20:03 -08:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
2022-05-01 21:39:29 -04:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
2022-05-01 21:39:29 -04:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-02-06 16:25:19 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
2022-05-01 21:39:29 -04:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-05-01 21:39:29 -04:00
|
|
|
|
2020-12-16 11:45:30 -05:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2014-04-03 14:48:18 -07:00
|
|
|
|
2022-05-01 21:39:29 -04:00
|
|
|
|
2010-01-27 09:20:03 -08:00
|
|
|
|
2022-05-01 21:39:29 -04:00
|
|
|
|
2019-07-11 20:55:17 -07:00
|
|
|
|
2010-01-27 09:20:03 -08:00
|
|
|
|
2014-04-03 14:48:18 -07:00
|
|
|
|
2010-01-27 09:20:03 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2011-12-21 11:05:48 -06:00
|
|
|
|
2010-01-27 09:20:03 -08:00
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
2021-01-28 19:19:45 +01:00
|
|
|
|
|
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
2010-01-27 09:20:03 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-07-11 20:55:20 -07:00
|
|
|
|
2010-01-27 09:20:03 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-11-30 17:49:44 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-01 16:58:58 +02:00
|
|
|
|
2019-11-30 17:49:44 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2020-11-16 14:33:37 +01:00
|
|
|
|
2019-11-30 17:49:44 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-01 16:58:58 +02:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-06-01 16:58:58 +02:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-06-01 16:58:58 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2008-07-23 21:27:04 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2016-04-07 08:51:56 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-06-01 16:58:58 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2008-07-23 21:27:04 -07:00
|
|
|
|
fs: fix data invalidation in the cleancache during direct IO
Patch series "Properly invalidate data in the cleancache", v2.
We've noticed that after direct IO write, buffered read sometimes gets
stale data which is coming from the cleancache. The reason for this is
that some direct write hooks call call invalidate_inode_pages2[_range]()
conditionally iff mapping->nrpages is not zero, so we may not invalidate
data in the cleancache.
Another odd thing is that we check only for ->nrpages and don't check
for ->nrexceptional, but invalidate_inode_pages2[_range] also
invalidates exceptional entries as well. So we invalidate exceptional
entries only if ->nrpages != 0? This doesn't feel right.
- Patch 1 fixes direct IO writes by removing ->nrpages check.
- Patch 2 fixes similar case in invalidate_bdev().
Note: I only fixed conditional cleancache_invalidate_inode() here.
Do we also need to add ->nrexceptional check in into invalidate_bdev()?
- Patches 3-4: some optimizations.
This patch (of 4):
Some direct IO write fs hooks call invalidate_inode_pages2[_range]()
conditionally iff mapping->nrpages is not zero. This can't be right,
because invalidate_inode_pages2[_range]() also invalidate data in the
cleancache via cleancache_invalidate_inode() call. So if page cache is
empty but there is some data in the cleancache, buffered read after
direct IO write would get stale data from the cleancache.
Also it doesn't feel right to check only for ->nrpages because
invalidate_inode_pages2[_range] invalidates exceptional entries as well.
Fix this by calling invalidate_inode_pages2[_range]() regardless of
nrpages state.
Note: nfs,cifs,9p doesn't need similar fix because the never call
cleancache_get_page() (nor directly, nor via mpage_readpage[s]()), so
they are not affected by this bug.
Fixes: c515e1fd361c ("mm/fs: add hooks to support cleancache")
Link: http://lkml.kernel.org/r/20170424164135.22350-2-aryabinin@virtuozzo.com
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Alexey Kuznetsov <kuznet@virtuozzo.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Nikolay Borisov <n.borisov.lkml@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-03 14:55:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-01 16:58:57 +02:00
|
|
|
|
fs: fix data invalidation in the cleancache during direct IO
Patch series "Properly invalidate data in the cleancache", v2.
We've noticed that after direct IO write, buffered read sometimes gets
stale data which is coming from the cleancache. The reason for this is
that some direct write hooks call call invalidate_inode_pages2[_range]()
conditionally iff mapping->nrpages is not zero, so we may not invalidate
data in the cleancache.
Another odd thing is that we check only for ->nrpages and don't check
for ->nrexceptional, but invalidate_inode_pages2[_range] also
invalidates exceptional entries as well. So we invalidate exceptional
entries only if ->nrpages != 0? This doesn't feel right.
- Patch 1 fixes direct IO writes by removing ->nrpages check.
- Patch 2 fixes similar case in invalidate_bdev().
Note: I only fixed conditional cleancache_invalidate_inode() here.
Do we also need to add ->nrexceptional check in into invalidate_bdev()?
- Patches 3-4: some optimizations.
This patch (of 4):
Some direct IO write fs hooks call invalidate_inode_pages2[_range]()
conditionally iff mapping->nrpages is not zero. This can't be right,
because invalidate_inode_pages2[_range]() also invalidate data in the
cleancache via cleancache_invalidate_inode() call. So if page cache is
empty but there is some data in the cleancache, buffered read after
direct IO write would get stale data from the cleancache.
Also it doesn't feel right to check only for ->nrpages because
invalidate_inode_pages2[_range] invalidates exceptional entries as well.
Fix this by calling invalidate_inode_pages2[_range]() regardless of
nrpages state.
Note: nfs,cifs,9p doesn't need similar fix because the never call
cleancache_get_page() (nor directly, nor via mpage_readpage[s]()), so
they are not affected by this bug.
Fixes: c515e1fd361c ("mm/fs: add hooks to support cleancache")
Link: http://lkml.kernel.org/r/20170424164135.22350-2-aryabinin@virtuozzo.com
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Alexey Kuznetsov <kuznet@virtuozzo.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Nikolay Borisov <n.borisov.lkml@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-05-03 14:55:59 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-01 16:58:58 +02:00
|
|
|
|
2008-07-23 21:27:04 -07:00
|
|
|
|
|
|
|
|
|
2017-04-13 14:10:15 -04:00
|
|
|
|
2008-07-23 21:27:04 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2017-09-21 08:16:29 -06:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2019-11-30 17:49:41 -08:00
|
|
|
|
|
|
|
|
|
2019-11-30 17:49:47 -08:00
|
|
|
|
|
|
|
|
|
2019-11-30 17:49:41 -08:00
|
|
|
|
2008-07-23 21:27:04 -07:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-06-01 16:58:58 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2010-10-26 14:21:58 -07:00
|
|
|
|
2017-04-13 14:10:15 -04:00
|
|
|
|
2010-10-26 14:21:58 -07:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2014-02-11 20:58:20 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2021-02-24 12:01:45 -08:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2022-02-19 23:19:49 -05:00
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
2022-02-19 23:19:49 -05:00
|
|
|
|
|
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
|
|
|
|
|
2024-05-27 18:36:08 +02:00
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
|
|
|
|
|
2007-10-16 01:25:03 -07:00
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
2024-05-27 18:36:08 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
2022-09-15 17:04:16 +02:00
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
2024-05-27 18:36:08 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
mm: make sendfile(2) killable
Currently a simple program below issues a sendfile(2) system call which
takes about 62 days to complete in my test KVM instance.
int fd;
off_t off = 0;
fd = open("file", O_RDWR | O_TRUNC | O_SYNC | O_CREAT, 0644);
ftruncate(fd, 2);
lseek(fd, 0, SEEK_END);
sendfile(fd, fd, &off, 0xfffffff);
Now you should not ask kernel to do a stupid stuff like copying 256MB in
2-byte chunks and call fsync(2) after each chunk but if you do, sysadmin
should have a way to stop you.
We actually do have a check for fatal_signal_pending() in
generic_perform_write() which triggers in this path however because we
always succeed in writing something before the check is done, we return
value > 0 from generic_perform_write() and thus the information about
signal gets lost.
Fix the problem by doing the signal check before writing anything. That
way generic_perform_write() returns -EINTR, the error gets propagated up
and the sendfile loop terminates early.
Signed-off-by: Jan Kara <jack@suse.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2015-10-22 13:32:21 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-07-16 09:36:06 +00:00
|
|
|
|
2024-07-15 14:24:01 -04:00
|
|
|
|
2014-06-04 16:10:31 -07:00
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
|
|
|
|
|
2024-05-27 18:36:08 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
mm: flush dcache before writing into page to avoid alias
The cache alias problem will happen if the changes of user shared mapping
is not flushed before copying, then user and kernel mapping may be mapped
into two different cache line, it is impossible to guarantee the coherence
after iov_iter_copy_from_user_atomic. So the right steps should be:
flush_dcache_page(page);
kmap_atomic(page);
write to page;
kunmap_atomic(page);
flush_dcache_page(page);
More precisely, we might create two new APIs flush_dcache_user_page and
flush_dcache_kern_page to replace the two flush_dcache_page accordingly.
Here is a snippet tested on omap2430 with VIPT cache, and I think it is
not ARM-specific:
int val = 0x11111111;
fd = open("abc", O_RDWR);
addr = mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_SHARED, fd, 0);
*(addr+0) = 0x44444444;
tmp = *(addr+0);
*(addr+1) = 0x77777777;
write(fd, &val, sizeof(int));
close(fd);
The results are not always 0x11111111 0x77777777 at the beginning as expected. Sometimes we see 0x44444444 0x77777777.
Signed-off-by: Anfei <anfei.zhou@gmail.com>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: Miklos Szeredi <miklos@szeredi.hu>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: <linux-arch@vger.kernel.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-02-02 13:44:02 -08:00
|
|
|
|
2024-05-27 18:36:08 +02:00
|
|
|
|
2015-10-07 08:32:38 +01:00
|
|
|
|
2025-02-28 12:37:22 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-05-27 18:36:08 +02:00
|
|
|
|
|
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
2025-07-16 09:36:06 +00:00
|
|
|
|
2024-07-10 15:45:32 -04:00
|
|
|
|
2021-04-30 10:26:41 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
|
|
|
|
|
2021-05-31 00:32:44 -04:00
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
2021-05-31 00:32:44 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
2024-05-27 18:36:08 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-05-31 00:32:44 -04:00
|
|
|
|
2024-05-27 18:36:08 +02:00
|
|
|
|
|
|
|
|
|
2025-02-28 12:37:22 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-05-27 18:36:08 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2023-06-01 16:58:55 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2007-10-16 01:25:01 -07:00
|
|
|
|
2014-02-11 21:34:08 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2009-08-17 18:10:06 +02:00
|
|
|
|
2014-04-03 03:17:43 -04:00
|
|
|
|
2009-08-17 18:10:06 +02:00
|
|
|
|
2014-04-03 03:17:43 -04:00
|
|
|
|
2009-08-17 18:10:06 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-12 15:50:21 +02:00
|
|
|
|
2009-08-17 18:10:06 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2021-04-12 15:50:21 +02:00
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2009-08-17 18:10:06 +02:00
|
|
|
|
2014-04-03 03:17:43 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2021-05-04 18:40:12 -07:00
|
|
|
|
2023-06-01 16:59:01 +02:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-06-01 16:59:01 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-06-01 16:59:01 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-10-19 23:28:13 -07:00
|
|
|
|
2015-04-09 13:52:01 -04:00
|
|
|
|
2023-06-01 16:59:01 +02:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2015-02-16 15:58:53 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2023-06-01 16:59:01 +02:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-10-19 23:28:13 -07:00
|
|
|
|
2023-06-01 16:59:01 +02:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2014-04-03 03:17:43 -04:00
|
|
|
|
2009-08-17 18:10:06 +02:00
|
|
|
|
|
|
|
|
|
2014-04-03 03:17:43 -04:00
|
|
|
|
2009-08-17 18:10:06 +02:00
|
|
|
|
2014-04-03 03:17:43 -04:00
|
|
|
|
2009-08-17 18:10:06 +02:00
|
|
|
|
2014-04-03 03:17:43 -04:00
|
|
|
|
2009-08-17 18:10:06 +02:00
|
|
|
|
2021-04-12 15:50:21 +02:00
|
|
|
|
2019-03-05 15:48:42 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2009-08-17 18:10:06 +02:00
|
|
|
|
2014-04-03 03:17:43 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2009-08-17 19:52:36 +02:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2016-01-22 15:40:57 -05:00
|
|
|
|
2015-04-09 12:55:47 -04:00
|
|
|
|
|
|
|
|
|
2015-04-07 11:28:12 -04:00
|
|
|
|
2016-01-22 15:40:57 -05:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2016-04-07 08:52:01 -07:00
|
|
|
|
|
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
|
|
|
|
|
2014-04-03 03:17:43 -04:00
|
|
|
|
2005-04-16 15:20:36 -07:00
|
|
|
|
2006-08-29 19:05:54 +01:00
|
|
|
|
2021-07-28 15:14:48 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-08-29 19:05:54 +01:00
|
|
|
|
2021-07-28 15:14:48 -04:00
|
|
|
|
|
|
|
|
|
2006-08-29 19:05:54 +01:00
|
|
|
|
2021-07-28 15:14:48 -04:00
|
|
|
|
|
|
|
|
|
2009-04-03 16:42:36 +01:00
|
|
|
|
2021-07-28 15:14:48 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2006-08-29 19:05:54 +01:00
|
|
|
|
2021-07-28 15:14:48 -04:00
|
|
|
|
2006-08-29 19:05:54 +01:00
|
|
|
|
2021-07-28 15:14:48 -04:00
|
|
|
|
2006-08-29 19:05:54 +01:00
|
|
|
|
2021-07-28 15:14:48 -04:00
|
|
|
|
2006-08-29 19:05:54 +01:00
|
|
|
|
2021-07-28 15:14:48 -04:00
|
|
|
|
mm: merge folio_has_private()/filemap_release_folio() call pairs
Patch series "mm, netfs, fscache: Stop read optimisation when folio
removed from pagecache", v7.
This fixes an optimisation in fscache whereby we don't read from the cache
for a particular file until we know that there's data there that we don't
have in the pagecache. The problem is that I'm no longer using PG_fscache
(aka PG_private_2) to indicate that the page is cached and so I don't get
a notification when a cached page is dropped from the pagecache.
The first patch merges some folio_has_private() and
filemap_release_folio() pairs and introduces a helper,
folio_needs_release(), to indicate if a release is required.
The second patch is the actual fix. Following Willy's suggestions[1], it
adds an AS_RELEASE_ALWAYS flag to an address_space that will make
filemap_release_folio() always call ->release_folio(), even if
PG_private/PG_private_2 aren't set. folio_needs_release() is altered to
add a check for this.
This patch (of 2):
Make filemap_release_folio() check folio_has_private(). Then, in most
cases, where a call to folio_has_private() is immediately followed by a
call to filemap_release_folio(), we can get rid of the test in the pair.
There are a couple of sites in mm/vscan.c that this can't so easily be
done. In shrink_folio_list(), there are actually three cases (something
different is done for incompletely invalidated buffers), but
filemap_release_folio() elides two of them.
In shrink_active_list(), we don't have have the folio lock yet, so the
check allows us to avoid locking the page unnecessarily.
A wrapper function to check if a folio needs release is provided for those
places that still need to do it in the mm/ directory. This will acquire
additional parts to the condition in a future patch.
After this, the only remaining caller of folio_has_private() outside of
mm/ is a check in fuse.
Link: https://lkml.kernel.org/r/20230628104852.3391651-1-dhowells@redhat.com
Link: https://lkml.kernel.org/r/20230628104852.3391651-2-dhowells@redhat.com
Reported-by: Rohith Surabattula <rohiths.msft@gmail.com>
Suggested-by: Matthew Wilcox <willy@infradead.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Steve French <sfrench@samba.org>
Cc: Shyam Prasad N <nspmangalore@gmail.com>
Cc: Rohith Surabattula <rohiths.msft@gmail.com>
Cc: Dave Wysochanski <dwysocha@redhat.com>
Cc: Dominique Martinet <asmadeus@codewreck.org>
Cc: Ilya Dryomov <idryomov@gmail.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: Xiubo Li <xiubli@redhat.com>
Cc: Jingbo Xu <jefflexu@linux.alibaba.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-06-28 11:48:51 +01:00
|
|
|
|
|
|
|
|
|
2021-07-28 15:14:48 -04:00
|
|
|
|
|
|
|
|
|
2006-08-29 19:05:54 +01:00
|
|
|
|
2022-04-29 17:00:05 -04:00
|
|
|
|
|
|
|
|
|
2022-05-01 01:08:08 -04:00
|
|
|
|
2006-08-29 19:05:54 +01:00
|
|
|
|
2021-07-28 15:14:48 -04:00
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
mm: Provide a means of invalidation without using launder_folio
Implement a replacement for launder_folio. The key feature of
invalidate_inode_pages2() is that it locks each folio individually, unmaps
it to prevent mmap'd accesses interfering and calls the ->launder_folio()
address_space op to flush it. This has problems: firstly, each folio is
written individually as one or more small writes; secondly, adjacent folios
cannot be added so easily into the laundry; thirdly, it's yet another op to
implement.
Instead, use the invalidate lock to cause anyone wanting to add a folio to
the inode to wait, then unmap all the folios if we have mmaps, then,
conditionally, use ->writepages() to flush any dirty data back and then
discard all pages.
The invalidate lock prevents ->read_iter(), ->write_iter() and faulting
through mmap all from adding pages for the duration.
This is then used from netfslib to handle the flusing in unbuffered and
direct writes.
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Matthew Wilcox <willy@infradead.org>
cc: Miklos Szeredi <miklos@szeredi.hu>
cc: Trond Myklebust <trond.myklebust@hammerspace.com>
cc: Christoph Hellwig <hch@lst.de>
cc: Andrew Morton <akpm@linux-foundation.org>
cc: Alexander Viro <viro@zeniv.linux.org.uk>
cc: Christian Brauner <brauner@kernel.org>
cc: Jeff Layton <jlayton@kernel.org>
cc: linux-mm@kvack.org
cc: linux-fsdevel@vger.kernel.org
cc: netfs@lists.linux.dev
cc: v9fs@lists.linux.dev
cc: linux-afs@lists.infradead.org
cc: ceph-devel@vger.kernel.org
cc: linux-cifs@vger.kernel.org
cc: linux-nfs@vger.kernel.org
cc: devel@lists.orangefs.org
2024-03-27 08:51:38 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-08-28 22:02:45 +01:00
|
|
|
|
mm: Provide a means of invalidation without using launder_folio
Implement a replacement for launder_folio. The key feature of
invalidate_inode_pages2() is that it locks each folio individually, unmaps
it to prevent mmap'd accesses interfering and calls the ->launder_folio()
address_space op to flush it. This has problems: firstly, each folio is
written individually as one or more small writes; secondly, adjacent folios
cannot be added so easily into the laundry; thirdly, it's yet another op to
implement.
Instead, use the invalidate lock to cause anyone wanting to add a folio to
the inode to wait, then unmap all the folios if we have mmaps, then,
conditionally, use ->writepages() to flush any dirty data back and then
discard all pages.
The invalidate lock prevents ->read_iter(), ->write_iter() and faulting
through mmap all from adding pages for the duration.
This is then used from netfslib to handle the flusing in unbuffered and
direct writes.
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Matthew Wilcox <willy@infradead.org>
cc: Miklos Szeredi <miklos@szeredi.hu>
cc: Trond Myklebust <trond.myklebust@hammerspace.com>
cc: Christoph Hellwig <hch@lst.de>
cc: Andrew Morton <akpm@linux-foundation.org>
cc: Alexander Viro <viro@zeniv.linux.org.uk>
cc: Christian Brauner <brauner@kernel.org>
cc: Jeff Layton <jlayton@kernel.org>
cc: linux-mm@kvack.org
cc: linux-fsdevel@vger.kernel.org
cc: netfs@lists.linux.dev
cc: v9fs@lists.linux.dev
cc: linux-afs@lists.infradead.org
cc: ceph-devel@vger.kernel.org
cc: linux-cifs@vger.kernel.org
cc: linux-nfs@vger.kernel.org
cc: devel@lists.orangefs.org
2024-03-27 08:51:38 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-06-27 13:17:37 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
2024-02-19 19:01:21 -08:00
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-02-19 19:01:21 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-09-06 16:05:12 -07:00
|
|
|
|
2024-02-19 19:01:21 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-03-15 05:55:56 -04:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-09-17 00:00:53 +08:00
|
|
|
|
2024-03-15 05:55:56 -04:00
|
|
|
|
|
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
2024-06-27 13:17:37 -07:00
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-02-19 19:01:21 -08:00
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
2024-02-19 19:01:21 -08:00
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2025-01-21 09:27:22 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-07-22 21:59:30 -04:00
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-07-22 21:59:30 -04:00
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-07-22 21:59:30 -04:00
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-07-22 21:59:30 -04:00
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
2025-01-21 09:27:22 -08:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-07-22 21:59:30 -04:00
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2024-05-31 14:12:01 -04:00
|
|
|
|
cachestat: implement cachestat syscall
There is currently no good way to query the page cache state of large file
sets and directory trees. There is mincore(), but it scales poorly: the
kernel writes out a lot of bitmap data that userspace has to aggregate,
when the user really doesn not care about per-page information in that
case. The user also needs to mmap and unmap each file as it goes along,
which can be quite slow as well.
Some use cases where this information could come in handy:
* Allowing database to decide whether to perform an index scan or
direct table queries based on the in-memory cache state of the
index.
* Visibility into the writeback algorithm, for performance issues
diagnostic.
* Workload-aware writeback pacing: estimating IO fulfilled by page
cache (and IO to be done) within a range of a file, allowing for
more frequent syncing when and where there is IO capacity, and
batching when there is not.
* Computing memory usage of large files/directory trees, analogous to
the du tool for disk usage.
More information about these use cases could be found in the following
thread:
https://lore.kernel.org/lkml/20230315170934.GA97793@cmpxchg.org/
This patch implements a new syscall that queries cache state of a file and
summarizes the number of cached pages, number of dirty pages, number of
pages marked for writeback, number of (recently) evicted pages, etc. in a
given range. Currently, the syscall is only wired in for x86
architecture.
NAME
cachestat - query the page cache statistics of a file.
SYNOPSIS
#include <sys/mman.h>
struct cachestat_range {
__u64 off;
__u64 len;
};
struct cachestat {
__u64 nr_cache;
__u64 nr_dirty;
__u64 nr_writeback;
__u64 nr_evicted;
__u64 nr_recently_evicted;
};
int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
struct cachestat *cstat, unsigned int flags);
DESCRIPTION
cachestat() queries the number of cached pages, number of dirty
pages, number of pages marked for writeback, number of evicted
pages, number of recently evicted pages, in the bytes range given by
`off` and `len`.
An evicted page is a page that is previously in the page cache but
has been evicted since. A page is recently evicted if its last
eviction was recent enough that its reentry to the cache would
indicate that it is actively being used by the system, and that
there is memory pressure on the system.
These values are returned in a cachestat struct, whose address is
given by the `cstat` argument.
The `off` and `len` arguments must be non-negative integers. If
`len` > 0, the queried range is [`off`, `off` + `len`]. If `len` ==
0, we will query in the range from `off` to the end of the file.
The `flags` argument is unused for now, but is included for future
extensibility. User should pass 0 (i.e no flag specified).
Currently, hugetlbfs is not supported.
Because the status of a page can change after cachestat() checks it
but before it returns to the application, the returned values may
contain stale information.
RETURN VALUE
On success, cachestat returns 0. On error, -1 is returned, and errno
is set to indicate the error.
ERRORS
EFAULT cstat or cstat_args points to an invalid address.
EINVAL invalid flags.
EBADF invalid file descriptor.
EOPNOTSUPP file descriptor is of a hugetlbfs file
[nphamcs@gmail.com: replace rounddown logic with the existing helper]
Link: https://lkml.kernel.org/r/20230504022044.3675469-1-nphamcs@gmail.com
Link: https://lkml.kernel.org/r/20230503013608.2431726-3-nphamcs@gmail.com
Signed-off-by: Nhat Pham <nphamcs@gmail.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Brian Foster <bfoster@redhat.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
2023-05-02 18:36:07 -07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
