The VM tries to balance reclaim pressure between anon and file so as to
reduce the amount of IO incurred due to the memory shortage. It already
counts refaults and swapins, but in addition it should also count
writepage calls during reclaim.
For swap, this is obvious: it's IO that wouldn't have occurred if the
anonymous memory hadn't been under memory pressure. From a relative
balancing point of view this makes sense as well: even if anon is cold and
reclaimable, a cache that isn't thrashing may have equally cold pages that
don't require IO to reclaim.
For file writeback, it's trickier: some of the reclaim writepage IO would
have likely occurred anyway due to dirty expiration. But not all of it -
premature writeback reduces batching and generates additional writes.
Since the flushers are already woken up by the time the VM starts writing
cache pages one by one, let's assume that we'e likely causing writes that
wouldn't have happened without memory pressure. In addition, the per-page
cost of IO would have probably been much cheaper if written in larger
batches from the flusher thread rather than the single-page-writes from
kswapd.
For our purposes - getting the trend right to accelerate convergence on a
stable state that doesn't require paging at all - this is sufficiently
accurate. If we later wanted to optimize for sustained thrashing, we can
still refine the measurements.
Count all writepage calls from kswapd as IO cost toward the LRU that the
page belongs to.
Why do this dynamically? Don't we know in advance that anon pages require
IO to reclaim, and so could build in a static bias?
First, scanning is not the same as reclaiming. If all the anon pages are
referenced, we may not swap for a while just because we're scanning the
anon list. During this time, however, it's important that we age
anonymous memory and the page cache at the same rate so that their
hot-cold gradients are comparable. Everything else being equal, we still
want to reclaim the coldest memory overall.
Second, we keep copies in swap unless the page changes. If there is
swap-backed data that's mostly read (tmpfs file) and has been swapped out
before, we can reclaim it without incurring additional IO.
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Link: http://lkml.kernel.org/r/20200520232525.798933-14-hannes@cmpxchg.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Since the LRUs were split into anon and file lists, the VM has been
balancing between page cache and anonymous pages based on per-list ratios
of scanned vs. rotated pages. In most cases that tips page reclaim
towards the list that is easier to reclaim and has the fewest actively
used pages, but there are a few problems with it:
1. Refaults and LRU rotations are weighted the same way, even though
one costs IO and the other costs a bit of CPU.
2. The less we scan an LRU list based on already observed rotations,
the more we increase the sampling interval for new references, and
rotations become even more likely on that list. This can enter a
death spiral in which we stop looking at one list completely until
the other one is all but annihilated by page reclaim.
Since commit a528910e12 ("mm: thrash detection-based file cache sizing")
we have refault detection for the page cache. Along with swapin events,
they are good indicators of when the file or anon list, respectively, is
too small for its workingset and needs to grow.
For example, if the page cache is thrashing, the cache pages need more
time in memory, while there may be colder pages on the anonymous list.
Likewise, if swapped pages are faulting back in, it indicates that we
reclaim anonymous pages too aggressively and should back off.
Replace LRU rotations with refaults and swapins as the basis for relative
reclaim cost of the two LRUs. This will have the VM target list balances
that incur the least amount of IO on aggregate.
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Link: http://lkml.kernel.org/r/20200520232525.798933-12-hannes@cmpxchg.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We activate cache refaults with reuse distances in pages smaller than the
size of the total cache. This allows new pages with competitive access
frequencies to establish themselves, as well as challenge and potentially
displace pages on the active list that have gone cold.
However, that assumes that active cache can only replace other active
cache in a competition for the hottest memory. This is not a great
default assumption. The page cache might be thrashing while there are
enough completely cold and unused anonymous pages sitting around that we'd
only have to write to swap once to stop all IO from the cache.
Activate cache refaults when their reuse distance in pages is smaller than
the total userspace workingset, including anonymous pages.
Reclaim can still decide how to balance pressure among the two LRUs
depending on the IO situation. Rotational drives will prefer avoiding
random IO from swap and go harder after cache. But fundamentally, hot
cache should be able to compete with anon pages for a place in RAM.
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Link: http://lkml.kernel.org/r/20200520232525.798933-6-hannes@cmpxchg.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We use refault information to determine whether the cache workingset is
stable or transitioning, and dynamically adjust the inactive:active file
LRU ratio so as to maximize protection from one-off cache during stable
periods, and minimize IO during transitions.
With cgroups and their nested LRU lists, we currently don't do this
correctly. While recursive cgroup reclaim establishes a relative LRU
order among the pages of all involved cgroups, refaults only affect the
local LRU order in the cgroup in which they are occuring. As a result,
cache transitions can take longer in a cgrouped system as the active pages
of sibling cgroups aren't challenged when they should be.
[ Right now, this is somewhat theoretical, because the siblings, under
continued regular reclaim pressure, should eventually run out of
inactive pages - and since inactive:active *size* balancing is also
done on a cgroup-local level, we will challenge the active pages
eventually in most cases. But the next patch will move that relative
size enforcement to the reclaim root as well, and then this patch
here will be necessary to propagate refault pressure to siblings. ]
This patch moves refault detection to the root of reclaim. Instead of
remembering the cgroup owner of an evicted page, remember the cgroup that
caused the reclaim to happen. When refaults later occur, they'll
correctly influence the cross-cgroup LRU order that reclaim follows.
I.e. if global reclaim kicked out pages in some subgroup A/B/C, the
refault of those pages will challenge the global LRU order, and not just
the local order down inside C.
[hannes@cmpxchg.org: use page_memcg() instead of another lookup]
Link: http://lkml.kernel.org/r/20191115160722.GA309754@cmpxchg.org
Link: http://lkml.kernel.org/r/20191107205334.158354-3-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Rik van Riel <riel@surriel.com>
Cc: Shakeel Butt <shakeelb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
There is a per-memcg lruvec and a NUMA node lruvec. Which one is being
used is somewhat confusing right now, and it's easy to make mistakes -
especially when it comes to global reclaim.
How it works: when memory cgroups are enabled, we always use the
root_mem_cgroup's per-node lruvecs. When memory cgroups are not compiled
in or disabled at runtime, we use pgdat->lruvec.
Document that in a comment.
Due to the way the reclaim code is generalized, all lookups use the
mem_cgroup_lruvec() helper function, and nobody should have to find the
right lruvec manually right now. But to avoid future mistakes, rename the
pgdat->lruvec member to pgdat->__lruvec and delete the convenience wrapper
that suggests it's a commonly accessed member.
While in this area, swap the mem_cgroup_lruvec() argument order. The name
suggests a memcg operation, yet it takes a pgdat first and a memcg second.
I have to double take every time I call this. Fix that.
Link: http://lkml.kernel.org/r/20191022144803.302233-3-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Cc: Roman Gushchin <guro@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Memcg counters for shadow nodes are broken because the memcg pointer is
obtained in a wrong way. The following approach is used:
virt_to_page(xa_node)->mem_cgroup
Since commit 4d96ba3530 ("mm: memcg/slab: stop setting
page->mem_cgroup pointer for slab pages") page->mem_cgroup pointer isn't
set for slab pages, so memcg_from_slab_page() should be used instead.
Also I doubt that it ever worked correctly: virt_to_head_page() should
be used instead of virt_to_page(). Otherwise objects residing on tail
pages are not accounted, because only the head page contains a valid
mem_cgroup pointer. That was a case since the introduction of these
counters by the commit 68d48e6a2d ("mm: workingset: add vmstat counter
for shadow nodes").
Link: http://lkml.kernel.org/r/20190801233532.138743-1-guro@fb.com
Fixes: 4d96ba3530 ("mm: memcg/slab: stop setting page->mem_cgroup pointer for slab pages")
Signed-off-by: Roman Gushchin <guro@fb.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "mm: memcontrol: memory.stat cost & correctness".
The cgroup memory.stat file holds recursive statistics for the entire
subtree. The current implementation does this tree walk on-demand
whenever the file is read. This is giving us problems in production.
1. The cost of aggregating the statistics on-demand is high. A lot of
system service cgroups are mostly idle and their stats don't change
between reads, yet we always have to check them. There are also always
some lazily-dying cgroups sitting around that are pinned by a handful
of remaining page cache; the same applies to them.
In an application that periodically monitors memory.stat in our
fleet, we have seen the aggregation consume up to 5% CPU time.
2. When cgroups die and disappear from the cgroup tree, so do their
accumulated vm events. The result is that the event counters at
higher-level cgroups can go backwards and confuse some of our
automation, let alone people looking at the graphs over time.
To address both issues, this patch series changes the stat
implementation to spill counts upwards when the counters change.
The upward spilling is batched using the existing per-cpu cache. In a
sparse file stress test with 5 level cgroup nesting, the additional cost
of the flushing was negligible (a little under 1% of CPU at 100% CPU
utilization, compared to the 5% of reading memory.stat during regular
operation).
This patch (of 4):
memcg_page_state(), lruvec_page_state(), memcg_sum_events() are
currently returning the state of the local memcg or lruvec, not the
recursive state.
In practice there is a demand for both versions, although the callers
that want the recursive counts currently sum them up by hand.
Per default, cgroups are considered recursive entities and generally we
expect more users of the recursive counters, with the local counts being
special cases. To reflect that in the name, add a _local suffix to the
current implementations.
The following patch will re-incarnate these functions with recursive
semantics, but with an O(1) implementation.
[hannes@cmpxchg.org: fix bisection hole]
Link: http://lkml.kernel.org/r/20190417160347.GC23013@cmpxchg.org
Link: http://lkml.kernel.org/r/20190412151507.2769-2-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Michal Hocko <mhocko@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mem_cgroup_node_nr_lru_pages() is just a convenience wrapper around
lruvec_page_state() that takes bitmasks of lru indexes and aggregates the
counts for those.
Replace callsites where the bitmask is simple enough with direct
lruvec_page_state() calls.
This removes the last extern user of mem_cgroup_node_nr_lru_pages(), so
make that function private again, too.
Link: http://lkml.kernel.org/r/20190228163020.24100-5-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Roman Gushchin <guro@fb.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
workingset_eviction() doesn't use and never did use the @mapping
argument. Remove it.
Link: http://lkml.kernel.org/r/20190228083329.31892-1-aryabinin@virtuozzo.com
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Rik van Riel <riel@surriel.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: William Kucharski <william.kucharski@oracle.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
totalram_pages and totalhigh_pages are made static inline function.
Main motivation was that managed_page_count_lock handling was complicating
things. It was discussed in length here,
https://lore.kernel.org/patchwork/patch/995739/#1181785 So it seemes
better to remove the lock and convert variables to atomic, with preventing
poteintial store-to-read tearing as a bonus.
[akpm@linux-foundation.org: coding style fixes]
Link: http://lkml.kernel.org/r/1542090790-21750-4-git-send-email-arunks@codeaurora.org
Signed-off-by: Arun KS <arunks@codeaurora.org>
Suggested-by: Michal Hocko <mhocko@suse.com>
Suggested-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Reviewed-by: Pavel Tatashin <pasha.tatashin@soleen.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: David Hildenbrand <david@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull XArray conversion from Matthew Wilcox:
"The XArray provides an improved interface to the radix tree data
structure, providing locking as part of the API, specifying GFP flags
at allocation time, eliminating preloading, less re-walking the tree,
more efficient iterations and not exposing RCU-protected pointers to
its users.
This patch set
1. Introduces the XArray implementation
2. Converts the pagecache to use it
3. Converts memremap to use it
The page cache is the most complex and important user of the radix
tree, so converting it was most important. Converting the memremap
code removes the only other user of the multiorder code, which allows
us to remove the radix tree code that supported it.
I have 40+ followup patches to convert many other users of the radix
tree over to the XArray, but I'd like to get this part in first. The
other conversions haven't been in linux-next and aren't suitable for
applying yet, but you can see them in the xarray-conv branch if you're
interested"
* 'xarray' of git://git.infradead.org/users/willy/linux-dax: (90 commits)
radix tree: Remove multiorder support
radix tree test: Convert multiorder tests to XArray
radix tree tests: Convert item_delete_rcu to XArray
radix tree tests: Convert item_kill_tree to XArray
radix tree tests: Move item_insert_order
radix tree test suite: Remove multiorder benchmarking
radix tree test suite: Remove __item_insert
memremap: Convert to XArray
xarray: Add range store functionality
xarray: Move multiorder_check to in-kernel tests
xarray: Move multiorder_shrink to kernel tests
xarray: Move multiorder account test in-kernel
radix tree test suite: Convert iteration test to XArray
radix tree test suite: Convert tag_tagged_items to XArray
radix tree: Remove radix_tree_clear_tags
radix tree: Remove radix_tree_maybe_preload_order
radix tree: Remove split/join code
radix tree: Remove radix_tree_update_node_t
page cache: Finish XArray conversion
dax: Convert page fault handlers to XArray
...
The page cache and most shrinkable slab caches hold data that has been
read from disk, but there are some caches that only cache CPU work, such
as the dentry and inode caches of procfs and sysfs, as well as the subset
of radix tree nodes that track non-resident page cache.
Currently, all these are shrunk at the same rate: using DEFAULT_SEEKS for
the shrinker's seeks setting tells the reclaim algorithm that for every
two page cache pages scanned it should scan one slab object.
This is a bogus setting. A virtual inode that required no IO to create is
not twice as valuable as a page cache page; shadow cache entries with
eviction distances beyond the size of memory aren't either.
In most cases, the behavior in practice is still fine. Such virtual
caches don't tend to grow and assert themselves aggressively, and usually
get picked up before they cause problems. But there are scenarios where
that's not true.
Our database workloads suffer from two of those. For one, their file
workingset is several times bigger than available memory, which has the
kernel aggressively create shadow page cache entries for the non-resident
parts of it. The workingset code does tell the VM that most of these are
expendable, but the VM ends up balancing them 2:1 to cache pages as per
the seeks setting. This is a huge waste of memory.
These workloads also deal with tens of thousands of open files and use
/proc for introspection, which ends up growing the proc_inode_cache to
absurdly large sizes - again at the cost of valuable cache space, which
isn't a reasonable trade-off, given that proc inodes can be re-created
without involving the disk.
This patch implements a "zero-seek" setting for shrinkers that results in
a target ratio of 0:1 between their objects and IO-backed caches. This
allows such virtual caches to grow when memory is available (they do
cache/avoid CPU work after all), but effectively disables them as soon as
IO-backed objects are under pressure.
It then switches the shrinkers for procfs and sysfs metadata, as well as
excess page cache shadow nodes, to the new zero-seek setting.
Link: http://lkml.kernel.org/r/20181009184732.762-5-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reported-by: Domas Mituzas <dmituzas@fb.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Rik van Riel <riel@surriel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Make it easier to catch bugs in the shadow node shrinker by adding a
counter for the shadow nodes in circulation.
[akpm@linux-foundation.org: assert that irqs are disabled, for __inc_lruvec_page_state()]
[akpm@linux-foundation.org: s/WARN_ON_ONCE/VM_WARN_ON_ONCE/, per Johannes]
Link: http://lkml.kernel.org/r/20181009184732.762-4-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
No need to use the preemption-safe lruvec state function inside the
reclaim region that has irqs disabled.
Link: http://lkml.kernel.org/r/20181009184732.762-3-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Rik van Riel <riel@surriel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Refaults happen during transitions between workingsets as well as in-place
thrashing. Knowing the difference between the two has a range of
applications, including measuring the impact of memory shortage on the
system performance, as well as the ability to smarter balance pressure
between the filesystem cache and the swap-backed workingset.
During workingset transitions, inactive cache refaults and pushes out
established active cache. When that active cache isn't stale, however,
and also ends up refaulting, that's bonafide thrashing.
Introduce a new page flag that tells on eviction whether the page has been
active or not in its lifetime. This bit is then stored in the shadow
entry, to classify refaults as transitioning or thrashing.
How many page->flags does this leave us with on 32-bit?
20 bits are always page flags
21 if you have an MMU
23 with the zone bits for DMA, Normal, HighMem, Movable
29 with the sparsemem section bits
30 if PAE is enabled
31 with this patch.
So on 32-bit PAE, that leaves 1 bit for distinguishing two NUMA nodes. If
that's not enough, the system can switch to discontigmem and re-gain the 6
or 7 sparsemem section bits.
Link: http://lkml.kernel.org/r/20180828172258.3185-3-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>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "psi: pressure stall information for CPU, memory, and IO", v4.
Overview
PSI reports the overall wallclock time in which the tasks in a system (or
cgroup) wait for (contended) hardware resources.
This helps users understand the resource pressure their workloads are
under, which allows them to rootcause and fix throughput and latency
problems caused by overcommitting, underprovisioning, suboptimal job
placement in a grid; as well as anticipate major disruptions like OOM.
Real-world applications
We're using the data collected by PSI (and its previous incarnation,
memdelay) quite extensively at Facebook, and with several success stories.
One usecase is avoiding OOM hangs/livelocks. The reason these happen is
because the OOM killer is triggered by reclaim not being able to free
pages, but with fast flash devices there is *always* some clean and
uptodate cache to reclaim; the OOM killer never kicks in, even as tasks
spend 90% of the time thrashing the cache pages of their own executables.
There is no situation where this ever makes sense in practice. We wrote a
<100 line POC python script to monitor memory pressure and kill stuff way
before such pathological thrashing leads to full system losses that would
require forcible hard resets.
We've since extended and deployed this code into other places to guarantee
latency and throughput SLAs, since they're usually violated way before the
kernel OOM killer would ever kick in.
It is available here: https://github.com/facebookincubator/oomd
Eventually we probably want to trigger the in-kernel OOM killer based on
extreme sustained pressure as well, so that Linux can avoid memory
livelocks - which technically aren't deadlocks, but to the user
indistinguishable from them - out of the box. We'd continue using OOMD as
the first line of defense to ensure workload health and implement complex
kill policies that are beyond the scope of the kernel.
We also use PSI memory pressure for loadshedding. Our batch job
infrastructure used to use heuristics based on various VM stats to
anticipate OOM situations, with lackluster success. We switched it to PSI
and managed to anticipate and avoid OOM kills and lockups fairly reliably.
The reduction of OOM outages in the worker pool raised the pool's
aggregate productivity, and we were able to switch that service to smaller
machines.
Lastly, we use cgroups to isolate a machine's main workload from
maintenance crap like package upgrades, logging, configuration, as well as
to prevent multiple workloads on a machine from stepping on each others'
toes. We were not able to configure this properly without the pressure
metrics; we would see latency or bandwidth drops, but it would often be
hard to impossible to rootcause it post-mortem.
We now log and graph pressure for the containers in our fleet and can
trivially link latency spikes and throughput drops to shortages of
specific resources after the fact, and fix the job config/scheduling.
PSI has also received testing, feedback, and feature requests from Android
and EndlessOS for the purpose of low-latency OOM killing, to intervene in
pressure situations before the UI starts hanging.
How do you use this feature?
A kernel with CONFIG_PSI=y will create a /proc/pressure directory with 3
files: cpu, memory, and io. If using cgroup2, cgroups will also have
cpu.pressure, memory.pressure and io.pressure files, which simply
aggregate task stalls at the cgroup level instead of system-wide.
The cpu file contains one line:
some avg10=2.04 avg60=0.75 avg300=0.40 total=157656722
The averages give the percentage of walltime in which one or more tasks
are delayed on the runqueue while another task has the CPU. They're
recent averages over 10s, 1m, 5m windows, so you can tell short term
trends from long term ones, similarly to the load average.
The total= value gives the absolute stall time in microseconds. This
allows detecting latency spikes that might be too short to sway the
running averages. It also allows custom time averaging in case the
10s/1m/5m windows aren't adequate for the usecase (or are too coarse with
future hardware).
What to make of this "some" metric? If CPU utilization is at 100% and CPU
pressure is 0, it means the system is perfectly utilized, with one
runnable thread per CPU and nobody waiting. At two or more runnable tasks
per CPU, the system is 100% overcommitted and the pressure average will
indicate as much. From a utilization perspective this is a great state of
course: no CPU cycles are being wasted, even when 50% of the threads were
to go idle (as most workloads do vary). From the perspective of the
individual job it's not great, however, and they would do better with more
resources. Depending on what your priority and options are, raised "some"
numbers may or may not require action.
The memory file contains two lines:
some avg10=70.24 avg60=68.52 avg300=69.91 total=3559632828
full avg10=57.59 avg60=58.06 avg300=60.38 total=3300487258
The some line is the same as for cpu, the time in which at least one task
is stalled on the resource. In the case of memory, this includes waiting
on swap-in, page cache refaults and page reclaim.
The full line, however, indicates time in which *nobody* is using the CPU
productively due to pressure: all non-idle tasks are waiting for memory in
one form or another. Significant time spent in there is a good trigger
for killing things, moving jobs to other machines, or dropping incoming
requests, since neither the jobs nor the machine overall are making too
much headway.
The io file is similar to memory. Because the block layer doesn't have a
concept of hardware contention right now (how much longer is my IO request
taking due to other tasks?), it reports CPU potential lost on all IO
delays, not just the potential lost due to competition.
FAQ
Q: How is PSI's CPU component different from the load average?
A: There are several quirks in the load average that make it hard to
impossible to tell how overcommitted the CPU really is.
1. The load average is reported as a raw number of active tasks.
You need to know how many CPUs there are in the system, how many
CPUs the workload is allowed to use, then think about what the
proportion between load and the number of CPUs mean for the
tasks trying to run.
PSI reports the percentage of wallclock time in which tasks are
waiting for a CPU to run on. It doesn't matter how many CPUs are
present or usable. The number always tells the quality of life
of tasks in the system or in a particular cgroup.
2. The shortest averaging window is 1m, which is extremely coarse,
and it's sampled in 5s intervals. A *lot* can happen on a CPU in
5 seconds. This *may* be able to identify persistent long-term
trends and very clear and obvious overloads, but it's unusable
for latency spikes and more subtle overutilization.
PSI's shortest window is 10s. It also exports the cumulative
stall times (in microseconds) of synchronously recorded events.
3. On Linux, the load average for historical reasons includes all
TASK_UNINTERRUPTIBLE tasks. This gives a broader sense of how
busy the system is, but on the flipside it doesn't distinguish
whether tasks are likely to contend over the CPU or IO - which
obviously requires very different interventions from a sys admin
or a job scheduler.
PSI reports independent metrics for CPU and IO. You can tell
which resource is making the tasks wait, but in conjunction
still see how overloaded the system is overall.
Q: What's the cost / performance impact of this feature?
A: PSI's primary cost is in the scheduler, in particular task wakeups
and sleeps.
I benchmarked this code using Facebook's two most scheduling
sensitive workloads: memcache and webserver. They handle a ton of
small requests - lots of wakeups and sleeps with little actual work
in between - so they tend to be canaries for scheduler regressions.
In the tests, the boxes were handling live traffic over the course
of several hours. Half the machines, the control, ran with
CONFIG_PSI=n.
For memcache I used eight machines total. They're 2-socket, 14
core, 56 thread boxes. The test runs for half the test period,
flips the test and control kernels on the hardware to rule out HW
factors, DC location etc., then runs the other half of the test.
For the webservers, I used 32 machines total. They're single
socket, 16 core, 32 thread machines.
During the memcache test, CPU load was nopsi=78.05% psi=78.98% in
the first half and nopsi=77.52% psi=78.25%, so PSI added between
0.7 and 0.9 percentage points to the CPU load, a difference of
about 1%.
UPDATE: I re-ran this test with the v3 version of this patch set
and the CPU utilization was equivalent between test and control.
UPDATE: v4 is on par with v3.
As far as end-to-end request latency from the client perspective
goes, we don't sample those finely enough to capture the requests
going to those particular machines during the test, but we know the
p50 turnaround time in this workload is 54us, and perf bench sched
pipe on those machines show nopsi=5.232666 us/op and psi=5.587347
us/op, so this doesn't add much here either.
The profile for the pipe benchmark shows:
0.87% sched-pipe [kernel.vmlinux] [k] psi_group_change
0.83% perf.real [kernel.vmlinux] [k] psi_group_change
0.82% perf.real [kernel.vmlinux] [k] psi_task_change
0.58% sched-pipe [kernel.vmlinux] [k] psi_task_change
The webserver load is running inside 4 nested cgroup levels. The
CPU load with both nopsi and psi kernels was indistinguishable at
81%.
For comparison, we had to disable the cgroup cpu controller on the
webservers because it added 4 percentage points to the CPU% during
this same exact test.
Versions of this accounting code now run on 80% of our fleet. None
of our workloads have reported regressions during the rollout.
Daniel Drake said:
: I just retested the latest version at
: http://git.cmpxchg.org/cgit.cgi/linux-psi.git (Linux 4.18) and the results
: are great.
:
: Test setup:
: Endless OS
: GeminiLake N4200 low end laptop
: 2GB RAM
: swap (and zram swap) disabled
:
: Baseline test: open a handful of large-ish apps and several website
: tabs in Google Chrome.
:
: Results: after a couple of minutes, system is excessively thrashing, mouse
: cursor can barely be moved, UI is not responding to mouse clicks, so it's
: impractical to recover from this situation as an ordinary user
:
: Add my simple killer:
: https://gist.github.com/dsd/a8988bf0b81a6163475988120fe8d9cd
:
: Results: when the thrashing causes the UI to become sluggish, the killer
: steps in and kills something (usually a chrome tab), and the system
: remains usable. I repeatedly opened more apps and more websites over a 15
: minute period but I wasn't able to get the system to a point of UI
: unresponsiveness.
Suren said:
: Backported to 4.9 and retested on ARMv8 8 code system running Android.
: Signals behave as expected reacting to memory pressure, no jumps in
: "total" counters that would indicate an overflow/underflow issues. Nicely
: done!
This patch (of 9):
If we keep just enough refault information to match the *current* page
cache during reclaim time, we could lose a lot of events when there is
only a temporary spike in non-cache memory consumption that pushes out all
the cache. Once cache comes back, we won't see those refaults. They
might not be actionable for LRU aging, but we want to know about them for
measuring memory pressure.
[hannes@cmpxchg.org: switch to NUMA-aware lru and slab counters]
Link: http://lkml.kernel.org/r/20181009184732.762-2-hannes@cmpxchg.org
Link: http://lkml.kernel.org/r/20180828172258.3185-2-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <jweiner@fb.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Rik van Riel <riel@surriel.com>
Tested-by: Daniel Drake <drake@endlessm.com>
Tested-by: Suren Baghdasaryan <surenb@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vinayak Menon <vinmenon@codeaurora.org>
Cc: Christopher Lameter <cl@linux.com>
Cc: Peter Enderborg <peter.enderborg@sony.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We construct an XA_STATE and use it to delete the node with
xas_store() rather than adding a special function for this unique
use case. Includes a test that simulates this usage for the
test suite.
Signed-off-by: Matthew Wilcox <willy@infradead.org>
This is a direct replacement for struct radix_tree_node. A couple of
struct members have changed name, so convert those. Use a #define so
that radix tree users continue to work without change.
Signed-off-by: Matthew Wilcox <willy@infradead.org>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Introduce xarray value entries and tagged pointers to replace radix
tree exceptional entries. This is a slight change in encoding to allow
the use of an extra bit (we can now store BITS_PER_LONG - 1 bits in a
value entry). It is also a change in emphasis; exceptional entries are
intimidating and different. As the comment explains, you can choose
to store values or pointers in the xarray and they are both first-class
citizens.
Signed-off-by: Matthew Wilcox <willy@infradead.org>
Reviewed-by: Josef Bacik <jbacik@fb.com>
Provide list_lru_shrink_walk_irq() and let it behave like
list_lru_walk_one() except that it locks the spinlock with
spin_lock_irq(). This is used by scan_shadow_nodes() because its lock
nests within the i_pages lock which is acquired with IRQ. This change
allows to use proper locking promitives instead hand crafted
lock_irq_disable() plus spin_lock().
There is no EXPORT_SYMBOL provided because the current user is in-kernel
only.
Add list_lru_shrink_walk_irq() which acquires the spinlock with the
proper locking primitives.
Link: http://lkml.kernel.org/r/20180716111921.5365-5-bigeasy@linutronix.de
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We need to distinguish the situations when shrinker has very small
amount of objects (see vfs_pressure_ratio() called from
super_cache_count()), and when it has no objects at all. Currently, in
the both of these cases, shrinker::count_objects() returns 0.
The patch introduces new SHRINK_EMPTY return value, which will be used
for "no objects at all" case. It's is a refactoring mostly, as
SHRINK_EMPTY is replaced by 0 by all callers of do_shrink_slab() in this
patch, and all the magic will happen in further.
Link: http://lkml.kernel.org/r/153063069574.1818.11037751256699341813.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Tested-by: Shakeel Butt <shakeelb@google.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <jbacik@fb.com>
Cc: Li RongQing <lirongqing@baidu.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Matthias Kaehlcke <mka@chromium.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Philippe Ombredanne <pombredanne@nexb.com>
Cc: Roman Gushchin <guro@fb.com>
Cc: Sahitya Tummala <stummala@codeaurora.org>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Waiman Long <longman@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Add list_lru::shrinker_id field and populate it by registered shrinker
id.
This will be used to set correct bit in memcg shrinkers map by lru code
in next patches, after there appeared the first related to memcg element
in list_lru.
Link: http://lkml.kernel.org/r/153063059758.1818.14866596416857717800.stgit@localhost.localdomain
Signed-off-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Tested-by: Shakeel Butt <shakeelb@google.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <jbacik@fb.com>
Cc: Li RongQing <lirongqing@baidu.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Matthias Kaehlcke <mka@chromium.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Philippe Ombredanne <pombredanne@nexb.com>
Cc: Roman Gushchin <guro@fb.com>
Cc: Sahitya Tummala <stummala@codeaurora.org>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Waiman Long <longman@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
shadow_lru_isolate() disables interrupts and acquires a lock. It could
use spin_lock_irq() instead. It also uses local_irq_enable() while it
could use spin_unlock_irq()/xa_unlock_irq().
Use proper suffix for lock/unlock in order to enable/disable interrupts
during release/acquire of a lock.
Link: http://lkml.kernel.org/r/20180622151221.28167-3-bigeasy@linutronix.de
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Kirill Tkhai <ktkhai@virtuozzo.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "mm: use irq locking suffix instead local_irq_disable()".
A small series which avoids using local_irq_disable()/local_irq_enable()
but instead does spin_lock_irq()/spin_unlock_irq() so it is within the
context of the lock which it belongs to. Patch #1 is a cleanup where
local_irq_.*() remained after the lock was removed.
This patch (of 2):
In 0c7c1bed7e ("mm: make counting of list_lru_one::nr_items lockless")
the
spin_lock(&nlru->lock);
statement was replaced with
rcu_read_lock();
in __list_lru_count_one(). The comment in count_shadow_nodes() says
that the local_irq_disable() is required because the lock must be
acquired with disabled interrupts and (spin_lock()) does not do so.
Since the lock is replaced with rcu_read_lock() the local_irq_disable()
is no longer needed. The code path is
list_lru_shrink_count()
-> list_lru_count_one()
-> __list_lru_count_one()
-> rcu_read_lock()
-> list_lru_from_memcg_idx()
-> rcu_read_unlock()
Remove the local_irq_disable() statement.
Link: http://lkml.kernel.org/r/20180622151221.28167-2-bigeasy@linutronix.de
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Kirill Tkhai <ktkhai@virtuozzo.com>
Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Remove the address_space ->tree_lock and use the xa_lock newly added to
the radix_tree_root. Rename the address_space ->page_tree to ->i_pages,
since we don't really care that it's a tree.
[willy@infradead.org: fix nds32, fs/dax.c]
Link: http://lkml.kernel.org/r/20180406145415.GB20605@bombadil.infradead.orgLink: http://lkml.kernel.org/r/20180313132639.17387-9-willy@infradead.org
Signed-off-by: Matthew Wilcox <mawilcox@microsoft.com>
Acked-by: Jeff Layton <jlayton@redhat.com>
Cc: Darrick J. Wong <darrick.wong@oracle.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
During truncation, the mapping has already been checked for shmem and
dax so it's known that workingset_update_node is required.
This patch avoids the checks on mapping for each page being truncated.
In all other cases, a lookup helper is used to determine if
workingset_update_node() needs to be called. The one danger is that the
API is slightly harder to use as calling workingset_update_node directly
without checking for dax or shmem mappings could lead to surprises.
However, the API rarely needs to be used and hopefully the comment is
enough to give people the hint.
sparsetruncate (tiny)
4.14.0-rc4 4.14.0-rc4
oneirq-v1r1 pickhelper-v1r1
Min Time 141.00 ( 0.00%) 140.00 ( 0.71%)
1st-qrtle Time 142.00 ( 0.00%) 141.00 ( 0.70%)
2nd-qrtle Time 142.00 ( 0.00%) 142.00 ( 0.00%)
3rd-qrtle Time 143.00 ( 0.00%) 143.00 ( 0.00%)
Max-90% Time 144.00 ( 0.00%) 144.00 ( 0.00%)
Max-95% Time 147.00 ( 0.00%) 145.00 ( 1.36%)
Max-99% Time 195.00 ( 0.00%) 191.00 ( 2.05%)
Max Time 230.00 ( 0.00%) 205.00 ( 10.87%)
Amean Time 144.37 ( 0.00%) 143.82 ( 0.38%)
Stddev Time 10.44 ( 0.00%) 9.00 ( 13.74%)
Coeff Time 7.23 ( 0.00%) 6.26 ( 13.41%)
Best99%Amean Time 143.72 ( 0.00%) 143.34 ( 0.26%)
Best95%Amean Time 142.37 ( 0.00%) 142.00 ( 0.26%)
Best90%Amean Time 142.19 ( 0.00%) 141.85 ( 0.24%)
Best75%Amean Time 141.92 ( 0.00%) 141.58 ( 0.24%)
Best50%Amean Time 141.69 ( 0.00%) 141.31 ( 0.27%)
Best25%Amean Time 141.38 ( 0.00%) 140.97 ( 0.29%)
As you'd expect, the gain is marginal but it can be detected. The
differences in bonnie are all within the noise which is not surprising
given the impact on the microbenchmark.
radix_tree_update_node_t is a callback for some radix operations that
optionally passes in a private field. The only user of the callback is
workingset_update_node and as it no longer requires a mapping, the
private field is removed.
Link: http://lkml.kernel.org/r/20171018075952.10627-3-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Many source files in the tree are missing licensing information, which
makes it harder for compliance tools to determine the correct license.
By default all files without license information are under the default
license of the kernel, which is GPL version 2.
Update the files which contain no license information with the 'GPL-2.0'
SPDX license identifier. The SPDX identifier is a legally binding
shorthand, which can be used instead of the full boiler plate text.
This patch is based on work done by Thomas Gleixner and Kate Stewart and
Philippe Ombredanne.
How this work was done:
Patches were generated and checked against linux-4.14-rc6 for a subset of
the use cases:
- file had no licensing information it it.
- file was a */uapi/* one with no licensing information in it,
- file was a */uapi/* one with existing licensing information,
Further patches will be generated in subsequent months to fix up cases
where non-standard license headers were used, and references to license
had to be inferred by heuristics based on keywords.
The analysis to determine which SPDX License Identifier to be applied to
a file was done in a spreadsheet of side by side results from of the
output of two independent scanners (ScanCode & Windriver) producing SPDX
tag:value files created by Philippe Ombredanne. Philippe prepared the
base worksheet, and did an initial spot review of a few 1000 files.
The 4.13 kernel was the starting point of the analysis with 60,537 files
assessed. Kate Stewart did a file by file comparison of the scanner
results in the spreadsheet to determine which SPDX license identifier(s)
to be applied to the file. She confirmed any determination that was not
immediately clear with lawyers working with the Linux Foundation.
Criteria used to select files for SPDX license identifier tagging was:
- Files considered eligible had to be source code files.
- Make and config files were included as candidates if they contained >5
lines of source
- File already had some variant of a license header in it (even if <5
lines).
All documentation files were explicitly excluded.
The following heuristics were used to determine which SPDX license
identifiers to apply.
- when both scanners couldn't find any license traces, file was
considered to have no license information in it, and the top level
COPYING file license applied.
For non */uapi/* files that summary was:
SPDX license identifier # files
---------------------------------------------------|-------
GPL-2.0 11139
and resulted in the first patch in this series.
If that file was a */uapi/* path one, it was "GPL-2.0 WITH
Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was:
SPDX license identifier # files
---------------------------------------------------|-------
GPL-2.0 WITH Linux-syscall-note 930
and resulted in the second patch in this series.
- if a file had some form of licensing information in it, and was one
of the */uapi/* ones, it was denoted with the Linux-syscall-note if
any GPL family license was found in the file or had no licensing in
it (per prior point). Results summary:
SPDX license identifier # files
---------------------------------------------------|------
GPL-2.0 WITH Linux-syscall-note 270
GPL-2.0+ WITH Linux-syscall-note 169
((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21
((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17
LGPL-2.1+ WITH Linux-syscall-note 15
GPL-1.0+ WITH Linux-syscall-note 14
((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5
LGPL-2.0+ WITH Linux-syscall-note 4
LGPL-2.1 WITH Linux-syscall-note 3
((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3
((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1
and that resulted in the third patch in this series.
- when the two scanners agreed on the detected license(s), that became
the concluded license(s).
- when there was disagreement between the two scanners (one detected a
license but the other didn't, or they both detected different
licenses) a manual inspection of the file occurred.
- In most cases a manual inspection of the information in the file
resulted in a clear resolution of the license that should apply (and
which scanner probably needed to revisit its heuristics).
- When it was not immediately clear, the license identifier was
confirmed with lawyers working with the Linux Foundation.
- If there was any question as to the appropriate license identifier,
the file was flagged for further research and to be revisited later
in time.
In total, over 70 hours of logged manual review was done on the
spreadsheet to determine the SPDX license identifiers to apply to the
source files by Kate, Philippe, Thomas and, in some cases, confirmation
by lawyers working with the Linux Foundation.
Kate also obtained a third independent scan of the 4.13 code base from
FOSSology, and compared selected files where the other two scanners
disagreed against that SPDX file, to see if there was new insights. The
Windriver scanner is based on an older version of FOSSology in part, so
they are related.
Thomas did random spot checks in about 500 files from the spreadsheets
for the uapi headers and agreed with SPDX license identifier in the
files he inspected. For the non-uapi files Thomas did random spot checks
in about 15000 files.
In initial set of patches against 4.14-rc6, 3 files were found to have
copy/paste license identifier errors, and have been fixed to reflect the
correct identifier.
Additionally Philippe spent 10 hours this week doing a detailed manual
inspection and review of the 12,461 patched files from the initial patch
version early this week with:
- a full scancode scan run, collecting the matched texts, detected
license ids and scores
- reviewing anything where there was a license detected (about 500+
files) to ensure that the applied SPDX license was correct
- reviewing anything where there was no detection but the patch license
was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied
SPDX license was correct
This produced a worksheet with 20 files needing minor correction. This
worksheet was then exported into 3 different .csv files for the
different types of files to be modified.
These .csv files were then reviewed by Greg. Thomas wrote a script to
parse the csv files and add the proper SPDX tag to the file, in the
format that the file expected. This script was further refined by Greg
based on the output to detect more types of files automatically and to
distinguish between header and source .c files (which need different
comment types.) Finally Greg ran the script using the .csv files to
generate the patches.
Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org>
Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
lruvecs are at the intersection of the NUMA node and memcg, which is the
scope for most paging activity.
Introduce a convenient accounting infrastructure that maintains
statistics per node, per memcg, and the lruvec itself.
Then convert over accounting sites for statistics that are already
tracked in both nodes and memcgs and can be easily switched.
[hannes@cmpxchg.org: fix crash in the new cgroup stat keeping code]
Link: http://lkml.kernel.org/r/20170531171450.GA10481@cmpxchg.org
[hannes@cmpxchg.org: don't track uncharged pages at all
Link: http://lkml.kernel.org/r/20170605175254.GA8547@cmpxchg.org
[hannes@cmpxchg.org: add missing free_percpu()]
Link: http://lkml.kernel.org/r/20170605175354.GB8547@cmpxchg.org
[linux@roeck-us.net: hexagon: fix build error caused by include file order]
Link: http://lkml.kernel.org/r/20170617153721.GA4382@roeck-us.net
Link: http://lkml.kernel.org/r/20170530181724.27197-6-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The memory controllers stat function names are awkwardly long and
arbitrarily different from the zone and node stat functions.
The current interface is named:
mem_cgroup_read_stat()
mem_cgroup_update_stat()
mem_cgroup_inc_stat()
mem_cgroup_dec_stat()
mem_cgroup_update_page_stat()
mem_cgroup_inc_page_stat()
mem_cgroup_dec_page_stat()
This patch renames it to match the corresponding node stat functions:
memcg_page_state() [node_page_state()]
mod_memcg_state() [mod_node_state()]
inc_memcg_state() [inc_node_state()]
dec_memcg_state() [dec_node_state()]
mod_memcg_page_state() [mod_node_page_state()]
inc_memcg_page_state() [inc_node_page_state()]
dec_memcg_page_state() [dec_node_page_state()]
Link: http://lkml.kernel.org/r/20170404220148.28338-4-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The current duplication is a high-maintenance mess, and it's painful to
add new items or query memcg state from the rest of the VM.
This increases the size of the stat array marginally, but we should aim
to track all these stats on a per-cgroup level anyway.
Link: http://lkml.kernel.org/r/20170404220148.28338-3-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: Michal Hocko <mhocko@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Since commit 59dc76b0d4 ("mm: vmscan: reduce size of inactive file
list") we noticed bigger IO spikes during changes in cache access
patterns.
The patch in question shrunk the inactive list size to leave more room
for the current workingset in the presence of streaming IO. However,
workingset transitions that previously happened on the inactive list are
now pushed out of memory and incur more refaults to complete.
This patch disables active list protection when refaults are being
observed. This accelerates workingset transitions, and allows more of
the new set to establish itself from memory, without eating into the
ability to protect the established workingset during stable periods.
The workloads that were measurably affected for us were hit pretty bad
by it, with refault/majfault rates doubling and tripling during cache
transitions, and the machines sustaining half-hour periods of 100% IO
utilization, where they'd previously have sub-minute peaks at 60-90%.
Stateful services that handle user data tend to be more conservative
with kernel upgrades. As a result we hit most page cache issues with
some delay, as was the case here.
The severity seemed to warrant a stable tag.
Fixes: 59dc76b0d4 ("mm: vmscan: reduce size of inactive file list")
Link: http://lkml.kernel.org/r/20170404220052.27593-1-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: <stable@vger.kernel.org> [4.7+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 0a6b76dd23 ("mm: workingset: make shadow node shrinker memcg
aware") enabled cgroup-awareness in the shadow node shrinker, but forgot
to also enable cgroup-awareness in the list_lru the shadow nodes sit on.
Consequently, all shadow nodes are sitting on a global (per-NUMA node)
list, while the shrinker applies the limits according to the amount of
cache in the cgroup its shrinking. The result is excessive pressure on
the shadow nodes from cgroups that have very little cache.
Enable memcg-mode on the shadow node LRUs, such that per-cgroup limits
are applied to per-cgroup lists.
Fixes: 0a6b76dd23 ("mm: workingset: make shadow node shrinker memcg aware")
Link: http://lkml.kernel.org/r/20170322005320.8165-1-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vladimir Davydov <vdavydov@tarantool.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: <stable@vger.kernel.org> [4.6+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull IDR rewrite from Matthew Wilcox:
"The most significant part of the following is the patch to rewrite the
IDR & IDA to be clients of the radix tree. But there's much more,
including an enhancement of the IDA to be significantly more space
efficient, an IDR & IDA test suite, some improvements to the IDR API
(and driver changes to take advantage of those improvements), several
improvements to the radix tree test suite and RCU annotations.
The IDR & IDA rewrite had a good spin in linux-next and Andrew's tree
for most of the last cycle. Coupled with the IDR test suite, I feel
pretty confident that any remaining bugs are quite hard to hit. 0-day
did a great job of watching my git tree and pointing out problems; as
it hit them, I added new test-cases to be sure not to be caught the
same way twice"
Willy goes on to expand a bit on the IDR rewrite rationale:
"The radix tree and the IDR use very similar data structures.
Merging the two codebases lets us share the memory allocation pools,
and results in a net deletion of 500 lines of code. It also opens up
the possibility of exposing more of the features of the radix tree to
users of the IDR (and I have some interesting patches along those
lines waiting for 4.12)
It also shrinks the size of the 'struct idr' from 40 bytes to 24 which
will shrink a fair few data structures that embed an IDR"
* 'idr-4.11' of git://git.infradead.org/users/willy/linux-dax: (32 commits)
radix tree test suite: Add config option for map shift
idr: Add missing __rcu annotations
radix-tree: Fix __rcu annotations
radix-tree: Add rcu_dereference and rcu_assign_pointer calls
radix tree test suite: Run iteration tests for longer
radix tree test suite: Fix split/join memory leaks
radix tree test suite: Fix leaks in regression2.c
radix tree test suite: Fix leaky tests
radix tree test suite: Enable address sanitizer
radix_tree_iter_resume: Fix out of bounds error
radix-tree: Store a pointer to the root in each node
radix-tree: Chain preallocated nodes through ->parent
radix tree test suite: Dial down verbosity with -v
radix tree test suite: Introduce kmalloc_verbose
idr: Return the deleted entry from idr_remove
radix tree test suite: Build separate binaries for some tests
ida: Use exceptional entries for small IDAs
ida: Move ida_bitmap to a percpu variable
Reimplement IDR and IDA using the radix tree
radix-tree: Add radix_tree_iter_delete
...
Remove the prototypes for shmem_mapping() and shmem_zero_setup() from
linux/mm.h, since they are already provided in linux/shmem_fs.h. But
shmem_fs.h must then provide the inline stub for shmem_mapping() when
CONFIG_SHMEM is not set, and a few more cfiles now need to #include it.
Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1702081658250.1549@eggly.anvils
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
lruvec_lru_size returns the full size of the LRU list while we sometimes
need a value reduced only to eligible zones (e.g. for lowmem requests).
inactive_list_is_low is one such user. Later patches will add more of
them. Add a new parameter to lruvec_lru_size and allow it filter out
zones which are not eligible for the given context.
Link: http://lkml.kernel.org/r/20170117103702.28542-2-mhocko@kernel.org
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Acked-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Instead of having this mysterious private_data in each radix_tree_node,
store a pointer to the root, which can be useful for debugging. This also
relieves the mm code from the duty of updating it.
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Matthew Wilcox <mawilcox@microsoft.com>
Several people report seeing warnings about inconsistent radix tree
nodes followed by crashes in the workingset code, which all looked like
use-after-free access from the shadow node shrinker.
Dave Jones managed to reproduce the issue with a debug patch applied,
which confirmed that the radix tree shrinking indeed frees shadow nodes
while they are still linked to the shadow LRU:
WARNING: CPU: 2 PID: 53 at lib/radix-tree.c:643 delete_node+0x1e4/0x200
CPU: 2 PID: 53 Comm: kswapd0 Not tainted 4.10.0-rc2-think+ #3
Call Trace:
delete_node+0x1e4/0x200
__radix_tree_delete_node+0xd/0x10
shadow_lru_isolate+0xe6/0x220
__list_lru_walk_one.isra.4+0x9b/0x190
list_lru_walk_one+0x23/0x30
scan_shadow_nodes+0x2e/0x40
shrink_slab.part.44+0x23d/0x5d0
shrink_node+0x22c/0x330
kswapd+0x392/0x8f0
This is the WARN_ON_ONCE(!list_empty(&node->private_list)) placed in the
inlined radix_tree_shrink().
The problem is with 14b468791f ("mm: workingset: move shadow entry
tracking to radix tree exceptional tracking"), which passes an update
callback into the radix tree to link and unlink shadow leaf nodes when
tree entries change, but forgot to pass the callback when reclaiming a
shadow node.
While the reclaimed shadow node itself is unlinked by the shrinker, its
deletion from the tree can cause the left-most leaf node in the tree to
be shrunk. If that happens to be a shadow node as well, we don't unlink
it from the LRU as we should.
Consider this tree, where the s are shadow entries:
root->rnode
|
[0 n]
| |
[s ] [sssss]
Now the shadow node shrinker reclaims the rightmost leaf node through
the shadow node LRU:
root->rnode
|
[0 ]
|
[s ]
Because the parent of the deleted node is the first level below the
root and has only one child in the left-most slot, the intermediate
level is shrunk and the node containing the single shadow is put in
its place:
root->rnode
|
[s ]
The shrinker again sees a single left-most slot in a first level node
and thus decides to store the shadow in root->rnode directly and free
the node - which is a leaf node on the shadow node LRU.
root->rnode
|
s
Without the update callback, the freed node remains on the shadow LRU,
where it causes later shrinker runs to crash.
Pass the node updater callback into __radix_tree_delete_node() in case
the deletion causes the left-most branch in the tree to collapse too.
Also add warnings when linked nodes are freed right away, rather than
wait for the use-after-free when the list is scanned much later.
Fixes: 14b468791f ("mm: workingset: move shadow entry tracking to radix tree exceptional tracking")
Reported-by: Dave Chinner <david@fromorbit.com>
Reported-by: Hugh Dickins <hughd@google.com>
Reported-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-and-tested-by: Dave Jones <davej@codemonkey.org.uk>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Chris Leech <cleech@redhat.com>
Cc: Lee Duncan <lduncan@suse.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Matthew Wilcox <mawilcox@linuxonhyperv.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Since commit 59dc76b0d4 ("mm: vmscan: reduce size of inactive file
list") the size of the active file list is no longer limited to half of
memory. Increase the shadow node limit accordingly to avoid throwing
out shadow entries that might still result in eligible refaults.
The exact size of the active list now depends on the overall size of the
page cache, but converges toward taking up most of the space:
In mm/vmscan.c::inactive_list_is_low(),
* total target max
* memory ratio inactive
* -------------------------------------
* 10MB 1 5MB
* 100MB 1 50MB
* 1GB 3 250MB
* 10GB 10 0.9GB
* 100GB 31 3GB
* 1TB 101 10GB
* 10TB 320 32GB
It would be possible to apply the same precise ratios when determining
the limit for radix tree nodes containing shadow entries, but since it
is merely an approximation of the oldest refault distances in the wild
and the code also makes assumptions about the node population density,
keep it simple and always target the full cache size.
While at it, clarify the comment and the formula for memory footprint.
Link: http://lkml.kernel.org/r/20161117214701.29000-1-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently, we track the shadow entries in the page cache in the upper
bits of the radix_tree_node->count, behind the back of the radix tree
implementation. Because the radix tree code has no awareness of them,
we rely on random subtleties throughout the implementation (such as the
node->count != 1 check in the shrinking code, which is meant to exclude
multi-entry nodes but also happens to skip nodes with only one shadow
entry, as that's accounted in the upper bits). This is error prone and
has, in fact, caused the bug fixed in d3798ae8c6 ("mm: filemap: don't
plant shadow entries without radix tree node").
To remove these subtleties, this patch moves shadow entry tracking from
the upper bits of node->count to the existing counter for exceptional
entries. node->count goes back to being a simple counter of valid
entries in the tree node and can be shrunk to a single byte.
This vastly simplifies the page cache code. All accounting happens
natively inside the radix tree implementation, and maintaining the LRU
linkage of shadow nodes is consolidated into a single function in the
workingset code that is called for leaf nodes affected by a change in
the page cache tree.
This also removes the last user of the __radix_delete_node() return
value. Eliminate it.
Link: http://lkml.kernel.org/r/20161117193211.GE23430@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Matthew Wilcox <mawilcox@linuxonhyperv.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
When the shadow page shrinker tries to reclaim a radix tree node but
finds it in an unexpected state - it should contain no pages, and
non-zero shadow entries - there is no need to kill the executing task or
even the entire system. Warn about the invalid state, then leave that
tree node be. Simply don't put it back on the shadow LRU for future
reclaim and move on.
Link: http://lkml.kernel.org/r/20161117191138.22769-4-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Matthew Wilcox <mawilcox@linuxonhyperv.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Antonio reports the following crash when using fuse under memory pressure:
kernel BUG at /build/linux-a2WvEb/linux-4.4.0/mm/workingset.c:346!
invalid opcode: 0000 [#1] SMP
Modules linked in: all of them
CPU: 2 PID: 63 Comm: kswapd0 Not tainted 4.4.0-36-generic #55-Ubuntu
Hardware name: System manufacturer System Product Name/P8H67-M PRO, BIOS 3904 04/27/2013
task: ffff88040cae6040 ti: ffff880407488000 task.ti: ffff880407488000
RIP: shadow_lru_isolate+0x181/0x190
Call Trace:
__list_lru_walk_one.isra.3+0x8f/0x130
list_lru_walk_one+0x23/0x30
scan_shadow_nodes+0x34/0x50
shrink_slab.part.40+0x1ed/0x3d0
shrink_zone+0x2ca/0x2e0
kswapd+0x51e/0x990
kthread+0xd8/0xf0
ret_from_fork+0x3f/0x70
which corresponds to the following sanity check in the shadow node
tracking:
BUG_ON(node->count & RADIX_TREE_COUNT_MASK);
The workingset code tracks radix tree nodes that exclusively contain
shadow entries of evicted pages in them, and this (somewhat obscure)
line checks whether there are real pages left that would interfere with
reclaim of the radix tree node under memory pressure.
While discussing ways how fuse might sneak pages into the radix tree
past the workingset code, Miklos pointed to replace_page_cache_page(),
and indeed there is a problem there: it properly accounts for the old
page being removed - __delete_from_page_cache() does that - but then
does a raw raw radix_tree_insert(), not accounting for the replacement
page. Eventually the page count bits in node->count underflow while
leaving the node incorrectly linked to the shadow node LRU.
To address this, make sure replace_page_cache_page() uses the tracked
page insertion code, page_cache_tree_insert(). This fixes the page
accounting and makes sure page-containing nodes are properly unlinked
from the shadow node LRU again.
Also, make the sanity checks a bit less obscure by using the helpers for
checking the number of pages and shadows in a radix tree node.
Fixes: 449dd6984d ("mm: keep page cache radix tree nodes in check")
Link: http://lkml.kernel.org/r/20160919155822.29498-1-hannes@cmpxchg.org
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reported-by: Antonio SJ Musumeci <trapexit@spawn.link>
Debugged-by: Miklos Szeredi <miklos@szeredi.hu>
Cc: <stable@vger.kernel.org> [3.15+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Working set and refault detection is still zone-based, fix it.
Link: http://lkml.kernel.org/r/1467970510-21195-16-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Memcg needs adjustment after moving LRUs to the node. Limits are
tracked per memcg but the soft-limit excess is tracked per zone. As
global page reclaim is based on the node, it is easy to imagine a
situation where a zone soft limit is exceeded even though the memcg
limit is fine.
This patch moves the soft limit tree the node. Technically, all the
variable names should also change but people are already familiar by the
meaning of "mz" even if "mn" would be a more appropriate name now.
Link: http://lkml.kernel.org/r/1467970510-21195-15-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Earlier patches focused on having direct reclaim and kswapd use data
that is node-centric for reclaiming but shrink_node() itself still uses
too much zone information. This patch removes unnecessary zone-based
information with the most important decision being whether to continue
reclaim or not. Some memcg APIs are adjusted as a result even though
memcg itself still uses some zone information.
[mgorman@techsingularity.net: optimization]
Link: http://lkml.kernel.org/r/1468588165-12461-2-git-send-email-mgorman@techsingularity.net
Link: http://lkml.kernel.org/r/1467970510-21195-14-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This moves the LRU lists from the zone to the node and related data such
as counters, tracing, congestion tracking and writeback tracking.
Unfortunately, due to reclaim and compaction retry logic, it is
necessary to account for the number of LRU pages on both zone and node
logic. Most reclaim logic is based on the node counters but the retry
logic uses the zone counters which do not distinguish inactive and
active sizes. It would be possible to leave the LRU counters on a
per-zone basis but it's a heavier calculation across multiple cache
lines that is much more frequent than the retry checks.
Other than the LRU counters, this is mostly a mechanical patch but note
that it introduces a number of anomalies. For example, the scans are
per-zone but using per-node counters. We also mark a node as congested
when a zone is congested. This causes weird problems that are fixed
later but is easier to review.
In the event that there is excessive overhead on 32-bit systems due to
the nodes being on LRU then there are two potential solutions
1. Long-term isolation of highmem pages when reclaim is lowmem
When pages are skipped, they are immediately added back onto the LRU
list. If lowmem reclaim persisted for long periods of time, the same
highmem pages get continually scanned. The idea would be that lowmem
keeps those pages on a separate list until a reclaim for highmem pages
arrives that splices the highmem pages back onto the LRU. It potentially
could be implemented similar to the UNEVICTABLE list.
That would reduce the skip rate with the potential corner case is that
highmem pages have to be scanned and reclaimed to free lowmem slab pages.
2. Linear scan lowmem pages if the initial LRU shrink fails
This will break LRU ordering but may be preferable and faster during
memory pressure than skipping LRU pages.
Link: http://lkml.kernel.org/r/1467970510-21195-4-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patchset: "Move LRU page reclaim from zones to nodes v9"
This series moves LRUs from the zones to the node. While this is a
current rebase, the test results were based on mmotm as of June 23rd.
Conceptually, this series is simple but there are a lot of details.
Some of the broad motivations for this are;
1. The residency of a page partially depends on what zone the page was
allocated from. This is partially combatted by the fair zone allocation
policy but that is a partial solution that introduces overhead in the
page allocator paths.
2. Currently, reclaim on node 0 behaves slightly different to node 1. For
example, direct reclaim scans in zonelist order and reclaims even if
the zone is over the high watermark regardless of the age of pages
in that LRU. Kswapd on the other hand starts reclaim on the highest
unbalanced zone. A difference in distribution of file/anon pages due
to when they were allocated results can result in a difference in
again. While the fair zone allocation policy mitigates some of the
problems here, the page reclaim results on a multi-zone node will
always be different to a single-zone node.
it was scheduled on as a result.
3. kswapd and the page allocator scan zones in the opposite order to
avoid interfering with each other but it's sensitive to timing. This
mitigates the page allocator using pages that were allocated very recently
in the ideal case but it's sensitive to timing. When kswapd is allocating
from lower zones then it's great but during the rebalancing of the highest
zone, the page allocator and kswapd interfere with each other. It's worse
if the highest zone is small and difficult to balance.
4. slab shrinkers are node-based which makes it harder to identify the exact
relationship between slab reclaim and LRU reclaim.
The reason we have zone-based reclaim is that we used to have
large highmem zones in common configurations and it was necessary
to quickly find ZONE_NORMAL pages for reclaim. Today, this is much
less of a concern as machines with lots of memory will (or should) use
64-bit kernels. Combinations of 32-bit hardware and 64-bit hardware are
rare. Machines that do use highmem should have relatively low highmem:lowmem
ratios than we worried about in the past.
Conceptually, moving to node LRUs should be easier to understand. The
page allocator plays fewer tricks to game reclaim and reclaim behaves
similarly on all nodes.
The series has been tested on a 16 core UMA machine and a 2-socket 48
core NUMA machine. The UMA results are presented in most cases as the NUMA
machine behaved similarly.
pagealloc
---------
This is a microbenchmark that shows the benefit of removing the fair zone
allocation policy. It was tested uip to order-4 but only orders 0 and 1 are
shown as the other orders were comparable.
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
Min total-odr0-1 490.00 ( 0.00%) 457.00 ( 6.73%)
Min total-odr0-2 347.00 ( 0.00%) 329.00 ( 5.19%)
Min total-odr0-4 288.00 ( 0.00%) 273.00 ( 5.21%)
Min total-odr0-8 251.00 ( 0.00%) 239.00 ( 4.78%)
Min total-odr0-16 234.00 ( 0.00%) 222.00 ( 5.13%)
Min total-odr0-32 223.00 ( 0.00%) 211.00 ( 5.38%)
Min total-odr0-64 217.00 ( 0.00%) 208.00 ( 4.15%)
Min total-odr0-128 214.00 ( 0.00%) 204.00 ( 4.67%)
Min total-odr0-256 250.00 ( 0.00%) 230.00 ( 8.00%)
Min total-odr0-512 271.00 ( 0.00%) 269.00 ( 0.74%)
Min total-odr0-1024 291.00 ( 0.00%) 282.00 ( 3.09%)
Min total-odr0-2048 303.00 ( 0.00%) 296.00 ( 2.31%)
Min total-odr0-4096 311.00 ( 0.00%) 309.00 ( 0.64%)
Min total-odr0-8192 316.00 ( 0.00%) 314.00 ( 0.63%)
Min total-odr0-16384 317.00 ( 0.00%) 315.00 ( 0.63%)
Min total-odr1-1 742.00 ( 0.00%) 712.00 ( 4.04%)
Min total-odr1-2 562.00 ( 0.00%) 530.00 ( 5.69%)
Min total-odr1-4 457.00 ( 0.00%) 433.00 ( 5.25%)
Min total-odr1-8 411.00 ( 0.00%) 381.00 ( 7.30%)
Min total-odr1-16 381.00 ( 0.00%) 356.00 ( 6.56%)
Min total-odr1-32 372.00 ( 0.00%) 346.00 ( 6.99%)
Min total-odr1-64 372.00 ( 0.00%) 343.00 ( 7.80%)
Min total-odr1-128 375.00 ( 0.00%) 351.00 ( 6.40%)
Min total-odr1-256 379.00 ( 0.00%) 351.00 ( 7.39%)
Min total-odr1-512 385.00 ( 0.00%) 355.00 ( 7.79%)
Min total-odr1-1024 386.00 ( 0.00%) 358.00 ( 7.25%)
Min total-odr1-2048 390.00 ( 0.00%) 362.00 ( 7.18%)
Min total-odr1-4096 390.00 ( 0.00%) 362.00 ( 7.18%)
Min total-odr1-8192 388.00 ( 0.00%) 363.00 ( 6.44%)
This shows a steady improvement throughout. The primary benefit is from
reduced system CPU usage which is obvious from the overall times;
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
User 189.19 191.80
System 2604.45 2533.56
Elapsed 2855.30 2786.39
The vmstats also showed that the fair zone allocation policy was definitely
removed as can be seen here;
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v8
DMA32 allocs 28794729769 0
Normal allocs 48432501431 77227309877
Movable allocs 0 0
tiobench on ext4
----------------
tiobench is a benchmark that artifically benefits if old pages remain resident
while new pages get reclaimed. The fair zone allocation policy mitigates this
problem so pages age fairly. While the benchmark has problems, it is important
that tiobench performance remains constant as it implies that page aging
problems that the fair zone allocation policy fixes are not re-introduced.
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
Min PotentialReadSpeed 89.65 ( 0.00%) 90.21 ( 0.62%)
Min SeqRead-MB/sec-1 82.68 ( 0.00%) 82.01 ( -0.81%)
Min SeqRead-MB/sec-2 72.76 ( 0.00%) 72.07 ( -0.95%)
Min SeqRead-MB/sec-4 75.13 ( 0.00%) 74.92 ( -0.28%)
Min SeqRead-MB/sec-8 64.91 ( 0.00%) 65.19 ( 0.43%)
Min SeqRead-MB/sec-16 62.24 ( 0.00%) 62.22 ( -0.03%)
Min RandRead-MB/sec-1 0.88 ( 0.00%) 0.88 ( 0.00%)
Min RandRead-MB/sec-2 0.95 ( 0.00%) 0.92 ( -3.16%)
Min RandRead-MB/sec-4 1.43 ( 0.00%) 1.34 ( -6.29%)
Min RandRead-MB/sec-8 1.61 ( 0.00%) 1.60 ( -0.62%)
Min RandRead-MB/sec-16 1.80 ( 0.00%) 1.90 ( 5.56%)
Min SeqWrite-MB/sec-1 76.41 ( 0.00%) 76.85 ( 0.58%)
Min SeqWrite-MB/sec-2 74.11 ( 0.00%) 73.54 ( -0.77%)
Min SeqWrite-MB/sec-4 80.05 ( 0.00%) 80.13 ( 0.10%)
Min SeqWrite-MB/sec-8 72.88 ( 0.00%) 73.20 ( 0.44%)
Min SeqWrite-MB/sec-16 75.91 ( 0.00%) 76.44 ( 0.70%)
Min RandWrite-MB/sec-1 1.18 ( 0.00%) 1.14 ( -3.39%)
Min RandWrite-MB/sec-2 1.02 ( 0.00%) 1.03 ( 0.98%)
Min RandWrite-MB/sec-4 1.05 ( 0.00%) 0.98 ( -6.67%)
Min RandWrite-MB/sec-8 0.89 ( 0.00%) 0.92 ( 3.37%)
Min RandWrite-MB/sec-16 0.92 ( 0.00%) 0.93 ( 1.09%)
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 approx-v9
User 645.72 525.90
System 403.85 331.75
Elapsed 6795.36 6783.67
This shows that the series has little or not impact on tiobench which is
desirable and a reduction in system CPU usage. It indicates that the fair
zone allocation policy was removed in a manner that didn't reintroduce
one class of page aging bug. There were only minor differences in overall
reclaim activity
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
Minor Faults 645838 647465
Major Faults 573 640
Swap Ins 0 0
Swap Outs 0 0
DMA allocs 0 0
DMA32 allocs 46041453 44190646
Normal allocs 78053072 79887245
Movable allocs 0 0
Allocation stalls 24 67
Stall zone DMA 0 0
Stall zone DMA32 0 0
Stall zone Normal 0 2
Stall zone HighMem 0 0
Stall zone Movable 0 65
Direct pages scanned 10969 30609
Kswapd pages scanned 93375144 93492094
Kswapd pages reclaimed 93372243 93489370
Direct pages reclaimed 10969 30609
Kswapd efficiency 99% 99%
Kswapd velocity 13741.015 13781.934
Direct efficiency 100% 100%
Direct velocity 1.614 4.512
Percentage direct scans 0% 0%
kswapd activity was roughly comparable. There were differences in direct
reclaim activity but negligible in the context of the overall workload
(velocity of 4 pages per second with the patches applied, 1.6 pages per
second in the baseline kernel).
pgbench read-only large configuration on ext4
---------------------------------------------
pgbench is a database benchmark that can be sensitive to page reclaim
decisions. This also checks if removing the fair zone allocation policy
is safe
pgbench Transactions
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v8
Hmean 1 188.26 ( 0.00%) 189.78 ( 0.81%)
Hmean 5 330.66 ( 0.00%) 328.69 ( -0.59%)
Hmean 12 370.32 ( 0.00%) 380.72 ( 2.81%)
Hmean 21 368.89 ( 0.00%) 369.00 ( 0.03%)
Hmean 30 382.14 ( 0.00%) 360.89 ( -5.56%)
Hmean 32 428.87 ( 0.00%) 432.96 ( 0.95%)
Negligible differences again. As with tiobench, overall reclaim activity
was comparable.
bonnie++ on ext4
----------------
No interesting performance difference, negligible differences on reclaim
stats.
paralleldd on ext4
------------------
This workload uses varying numbers of dd instances to read large amounts of
data from disk.
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v9
Amean Elapsd-1 186.04 ( 0.00%) 189.41 ( -1.82%)
Amean Elapsd-3 192.27 ( 0.00%) 191.38 ( 0.46%)
Amean Elapsd-5 185.21 ( 0.00%) 182.75 ( 1.33%)
Amean Elapsd-7 183.71 ( 0.00%) 182.11 ( 0.87%)
Amean Elapsd-12 180.96 ( 0.00%) 181.58 ( -0.35%)
Amean Elapsd-16 181.36 ( 0.00%) 183.72 ( -1.30%)
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v9
User 1548.01 1552.44
System 8609.71 8515.08
Elapsed 3587.10 3594.54
There is little or no change in performance but some drop in system CPU usage.
4.7.0-rc3 4.7.0-rc3
mmotm-20160623 nodelru-v9
Minor Faults 362662 367360
Major Faults 1204 1143
Swap Ins 22 0
Swap Outs 2855 1029
DMA allocs 0 0
DMA32 allocs 31409797 28837521
Normal allocs 46611853 49231282
Movable allocs 0 0
Direct pages scanned 0 0
Kswapd pages scanned 40845270 40869088
Kswapd pages reclaimed 40830976 40855294
Direct pages reclaimed 0 0
Kswapd efficiency 99% 99%
Kswapd velocity 11386.711 11369.769
Direct efficiency 100% 100%
Direct velocity 0.000 0.000
Percentage direct scans 0% 0%
Page writes by reclaim 2855 1029
Page writes file 0 0
Page writes anon 2855 1029
Page reclaim immediate 771 1628
Sector Reads 293312636 293536360
Sector Writes 18213568 18186480
Page rescued immediate 0 0
Slabs scanned 128257 132747
Direct inode steals 181 56
Kswapd inode steals 59 1131
It basically shows that kswapd was active at roughly the same rate in
both kernels. There was also comparable slab scanning activity and direct
reclaim was avoided in both cases. There appears to be a large difference
in numbers of inodes reclaimed but the workload has few active inodes and
is likely a timing artifact.
stutter
-------
stutter simulates a simple workload. One part uses a lot of anonymous
memory, a second measures mmap latency and a third copies a large file.
The primary metric is checking for mmap latency.
stutter
4.7.0-rc4 4.7.0-rc4
mmotm-20160623 nodelru-v8
Min mmap 16.6283 ( 0.00%) 13.4258 ( 19.26%)
1st-qrtle mmap 54.7570 ( 0.00%) 34.9121 ( 36.24%)
2nd-qrtle mmap 57.3163 ( 0.00%) 46.1147 ( 19.54%)
3rd-qrtle mmap 58.9976 ( 0.00%) 47.1882 ( 20.02%)
Max-90% mmap 59.7433 ( 0.00%) 47.4453 ( 20.58%)
Max-93% mmap 60.1298 ( 0.00%) 47.6037 ( 20.83%)
Max-95% mmap 73.4112 ( 0.00%) 82.8719 (-12.89%)
Max-99% mmap 92.8542 ( 0.00%) 88.8870 ( 4.27%)
Max mmap 1440.6569 ( 0.00%) 121.4201 ( 91.57%)
Mean mmap 59.3493 ( 0.00%) 42.2991 ( 28.73%)
Best99%Mean mmap 57.2121 ( 0.00%) 41.8207 ( 26.90%)
Best95%Mean mmap 55.9113 ( 0.00%) 39.9620 ( 28.53%)
Best90%Mean mmap 55.6199 ( 0.00%) 39.3124 ( 29.32%)
Best50%Mean mmap 53.2183 ( 0.00%) 33.1307 ( 37.75%)
Best10%Mean mmap 45.9842 ( 0.00%) 20.4040 ( 55.63%)
Best5%Mean mmap 43.2256 ( 0.00%) 17.9654 ( 58.44%)
Best1%Mean mmap 32.9388 ( 0.00%) 16.6875 ( 49.34%)
This shows a number of improvements with the worst-case outlier greatly
improved.
Some of the vmstats are interesting
4.7.0-rc4 4.7.0-rc4
mmotm-20160623nodelru-v8
Swap Ins 163 502
Swap Outs 0 0
DMA allocs 0 0
DMA32 allocs 618719206 1381662383
Normal allocs 891235743 564138421
Movable allocs 0 0
Allocation stalls 2603 1
Direct pages scanned 216787 2
Kswapd pages scanned 50719775 41778378
Kswapd pages reclaimed 41541765 41777639
Direct pages reclaimed 209159 0
Kswapd efficiency 81% 99%
Kswapd velocity 16859.554 14329.059
Direct efficiency 96% 0%
Direct velocity 72.061 0.001
Percentage direct scans 0% 0%
Page writes by reclaim 6215049 0
Page writes file 6215049 0
Page writes anon 0 0
Page reclaim immediate 70673 90
Sector Reads 81940800 81680456
Sector Writes 100158984 98816036
Page rescued immediate 0 0
Slabs scanned 1366954 22683
While this is not guaranteed in all cases, this particular test showed
a large reduction in direct reclaim activity. It's also worth noting
that no page writes were issued from reclaim context.
This series is not without its hazards. There are at least three areas
that I'm concerned with even though I could not reproduce any problems in
that area.
1. Reclaim/compaction is going to be affected because the amount of reclaim is
no longer targetted at a specific zone. Compaction works on a per-zone basis
so there is no guarantee that reclaiming a few THP's worth page pages will
have a positive impact on compaction success rates.
2. The Slab/LRU reclaim ratio is affected because the frequency the shrinkers
are called is now different. This may or may not be a problem but if it
is, it'll be because shrinkers are not called enough and some balancing
is required.
3. The anon/file reclaim ratio may be affected. Pages about to be dirtied are
distributed between zones and the fair zone allocation policy used to do
something very similar for anon. The distribution is now different but not
necessarily in any way that matters but it's still worth bearing in mind.
VM statistic counters for reclaim decisions are zone-based. If the kernel
is to reclaim on a per-node basis then we need to track per-node
statistics but there is no infrastructure for that. The most notable
change is that the old node_page_state is renamed to
sum_zone_node_page_state. The new node_page_state takes a pglist_data and
uses per-node stats but none exist yet. There is some renaming such as
vm_stat to vm_zone_stat and the addition of vm_node_stat and the renaming
of mod_state to mod_zone_state. Otherwise, this is mostly a mechanical
patch with no functional change. There is a lot of similarity between the
node and zone helpers which is unfortunate but there was no obvious way of
reusing the code and maintaining type safety.
Link: http://lkml.kernel.org/r/1467970510-21195-2-git-send-email-mgorman@techsingularity.net
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Rik van Riel <riel@surriel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 23047a96d7 ("mm: workingset: per-cgroup cache thrash
detection") added a page->mem_cgroup lookup to the cache eviction,
refault, and activation paths, as well as locking to the activation
path, and the vm-scalability tests showed a regression of -23%.
While the test in question is an artificial worst-case scenario that
doesn't occur in real workloads - reading two sparse files in parallel
at full CPU speed just to hammer the LRU paths - there is still some
optimizations that can be done in those paths.
Inline the lookup functions to eliminate calls. Also, page->mem_cgroup
doesn't need to be stabilized when counting an activation; we merely
need to hold the RCU lock to prevent the memcg from being freed.
This cuts down on overhead quite a bit:
23047a96d7 063f6715e77a7be5770d6081fe
---------------- --------------------------
%stddev %change %stddev
\ | \
21621405 +- 0% +11.3% 24069657 +- 2% vm-scalability.throughput
[linux@roeck-us.net: drop unnecessary include file]
[hannes@cmpxchg.org: add WARN_ON_ONCE()s]
Link: http://lkml.kernel.org/r/20160707194024.GA26580@cmpxchg.org
Link: http://lkml.kernel.org/r/20160624175101.GA3024@cmpxchg.org
Reported-by: Ye Xiaolong <xiaolong.ye@intel.com>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Vladimir Davydov <vdavydov@virtuozzo.com>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 612e44939c ("mm: workingset: eviction buckets for bigmem/lowbit
machines") added a printk without a log level. Quieten it by using
pr_info().
Link: http://lkml.kernel.org/r/1466982072-29836-2-git-send-email-anton@ozlabs.org
Signed-off-by: Anton Blanchard <anton@samba.org>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Johannes Weiner