Patch series "hugetlb: add demote/split page functionality", v4.
The concurrent use of multiple hugetlb page sizes on a single system is
becoming more common. One of the reasons is better TLB support for
gigantic page sizes on x86 hardware. In addition, hugetlb pages are
being used to back VMs in hosting environments.
When using hugetlb pages to back VMs, it is often desirable to
preallocate hugetlb pools. This avoids the delay and uncertainty of
allocating hugetlb pages at VM startup. In addition, preallocating huge
pages minimizes the issue of memory fragmentation that increases the
longer the system is up and running.
In such environments, a combination of larger and smaller hugetlb pages
are preallocated in anticipation of backing VMs of various sizes. Over
time, the preallocated pool of smaller hugetlb pages may become depleted
while larger hugetlb pages still remain. In such situations, it is
desirable to convert larger hugetlb pages to smaller hugetlb pages.
Converting larger to smaller hugetlb pages can be accomplished today by
first freeing the larger page to the buddy allocator and then allocating
the smaller pages. For example, to convert 50 GB pages on x86:
gb_pages=`cat .../hugepages-1048576kB/nr_hugepages`
m2_pages=`cat .../hugepages-2048kB/nr_hugepages`
echo $(($gb_pages - 50)) > .../hugepages-1048576kB/nr_hugepages
echo $(($m2_pages + 25600)) > .../hugepages-2048kB/nr_hugepages
On an idle system this operation is fairly reliable and results are as
expected. The number of 2MB pages is increased as expected and the time
of the operation is a second or two.
However, when there is activity on the system the following issues
arise:
1) This process can take quite some time, especially if allocation of
the smaller pages is not immediate and requires migration/compaction.
2) There is no guarantee that the total size of smaller pages allocated
will match the size of the larger page which was freed. This is
because the area freed by the larger page could quickly be
fragmented.
In a test environment with a load that continually fills the page cache
with clean pages, results such as the following can be observed:
Unexpected number of 2MB pages allocated: Expected 25600, have 19944
real 0m42.092s
user 0m0.008s
sys 0m41.467s
To address these issues, introduce the concept of hugetlb page demotion.
Demotion provides a means of 'in place' splitting of a hugetlb page to
pages of a smaller size. This avoids freeing pages to buddy and then
trying to allocate from buddy.
Page demotion is controlled via sysfs files that reside in the per-hugetlb
page size and per node directories.
- demote_size
Target page size for demotion, a smaller huge page size. File
can be written to chose a smaller huge page size if multiple are
available.
- demote
Writable number of hugetlb pages to be demoted
To demote 50 GB huge pages, one would:
cat .../hugepages-1048576kB/free_hugepages /* optional, verify free pages */
cat .../hugepages-1048576kB/demote_size /* optional, verify target size */
echo 50 > .../hugepages-1048576kB/demote
Only hugetlb pages which are free at the time of the request can be
demoted. Demotion does not add to the complexity of surplus pages and
honors reserved huge pages. Therefore, when a value is written to the
sysfs demote file, that value is only the maximum number of pages which
will be demoted. It is possible fewer will actually be demoted. The
recently introduced per-hstate mutex is used to synchronize demote
operations with other operations that modify hugetlb pools.
Real world use cases
--------------------
The above scenario describes a real world use case where hugetlb pages
are used to back VMs on x86. Both issues of long allocation times and
not necessarily getting the expected number of smaller huge pages after
a free and allocate cycle have been experienced. The occurrence of
these issues is dependent on other activity within the host and can not
be predicted.
This patch (of 5):
Two new sysfs files are added to demote hugtlb pages. These files are
both per-hugetlb page size and per node. Files are:
demote_size - The size in Kb that pages are demoted to. (read-write)
demote - The number of huge pages to demote. (write-only)
By default, demote_size is the next smallest huge page size. Valid huge
page sizes less than huge page size may be written to this file. When
huge pages are demoted, they are demoted to this size.
Writing a value to demote will result in an attempt to demote that
number of hugetlb pages to an appropriate number of demote_size pages.
NOTE: Demote interfaces are only provided for huge page sizes if there
is a smaller target demote huge page size. For example, on x86 1GB huge
pages will have demote interfaces. 2MB huge pages will not have demote
interfaces.
This patch does not provide full demote functionality. It only provides
the sysfs interfaces.
It also provides documentation for the new interfaces.
[mike.kravetz@oracle.com: n_mask initialization does not need to be protected by the mutex]
Link: https://lkml.kernel.org/r/0530e4ef-2492-5186-f919-5db68edea654@oracle.com
Link: https://lkml.kernel.org/r/20211007181918.136982-2-mike.kravetz@oracle.com
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Cc: David Hildenbrand <david@redhat.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Zi Yan <ziy@nvidia.com>
Cc: Muchun Song <songmuchun@bytedance.com>
Cc: Naoya Horiguchi <naoya.horiguchi@linux.dev>
Cc: David Rientjes <rientjes@google.com>
Cc: "Aneesh Kumar K . V" <aneesh.kumar@linux.ibm.com>
Cc: Nghia Le <nghialm78@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
There was a report that starting an Ubuntu in docker while using cpuset
to bind it to movable nodes (a node only has movable zone, like a node
for hotplug or a Persistent Memory node in normal usage) will fail due
to memory allocation failure, and then OOM is involved and many other
innocent processes got killed.
It can be reproduced with command:
$ docker run -it --rm --cpuset-mems 4 ubuntu:latest bash -c "grep Mems_allowed /proc/self/status"
(where node 4 is a movable node)
runc:[2:INIT] invoked oom-killer: gfp_mask=0x500cc2(GFP_HIGHUSER|__GFP_ACCOUNT), order=0, oom_score_adj=0
CPU: 8 PID: 8291 Comm: runc:[2:INIT] Tainted: G W I E 5.8.2-0.g71b519a-default #1 openSUSE Tumbleweed (unreleased)
Hardware name: Dell Inc. PowerEdge R640/0PHYDR, BIOS 2.6.4 04/09/2020
Call Trace:
dump_stack+0x6b/0x88
dump_header+0x4a/0x1e2
oom_kill_process.cold+0xb/0x10
out_of_memory.part.0+0xaf/0x230
out_of_memory+0x3d/0x80
__alloc_pages_slowpath.constprop.0+0x954/0xa20
__alloc_pages_nodemask+0x2d3/0x300
pipe_write+0x322/0x590
new_sync_write+0x196/0x1b0
vfs_write+0x1c3/0x1f0
ksys_write+0xa7/0xe0
do_syscall_64+0x52/0xd0
entry_SYSCALL_64_after_hwframe+0x44/0xa9
Mem-Info:
active_anon:392832 inactive_anon:182 isolated_anon:0
active_file:68130 inactive_file:151527 isolated_file:0
unevictable:2701 dirty:0 writeback:7
slab_reclaimable:51418 slab_unreclaimable:116300
mapped:45825 shmem:735 pagetables:2540 bounce:0
free:159849484 free_pcp:73 free_cma:0
Node 4 active_anon:1448kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB mapped:0kB dirty:0kB writeback:0kB shmem:0kB shmem_thp: 0kB shmem_pmdmapped: 0kB anon_thp: 0kB writeback_tmp:0kB all_unreclaimable? no
Node 4 Movable free:130021408kB min:9140kB low:139160kB high:269180kB reserved_highatomic:0KB active_anon:1448kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB writepending:0kB present:130023424kB managed:130023424kB mlocked:0kB kernel_stack:0kB pagetables:0kB bounce:0kB free_pcp:292kB local_pcp:84kB free_cma:0kB
lowmem_reserve[]: 0 0 0 0 0
Node 4 Movable: 1*4kB (M) 0*8kB 0*16kB 1*32kB (M) 0*64kB 0*128kB 1*256kB (M) 1*512kB (M) 1*1024kB (M) 0*2048kB 31743*4096kB (M) = 130021156kB
oom-kill:constraint=CONSTRAINT_CPUSET,nodemask=(null),cpuset=docker-9976a269caec812c134fa317f27487ee36e1129beba7278a463dd53e5fb9997b.scope,mems_allowed=4,global_oom,task_memcg=/system.slice/containerd.service,task=containerd,pid=4100,uid=0
Out of memory: Killed process 4100 (containerd) total-vm:4077036kB, anon-rss:51184kB, file-rss:26016kB, shmem-rss:0kB, UID:0 pgtables:676kB oom_score_adj:0
oom_reaper: reaped process 8248 (docker), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
oom_reaper: reaped process 2054 (node_exporter), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
oom_reaper: reaped process 1452 (systemd-journal), now anon-rss:0kB, file-rss:8564kB, shmem-rss:4kB
oom_reaper: reaped process 2146 (munin-node), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
oom_reaper: reaped process 8291 (runc:[2:INIT]), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
The reason is that in this case, the target cpuset nodes only have
movable zone, while the creation of an OS in docker sometimes needs to
allocate memory in non-movable zones (dma/dma32/normal) like
GFP_HIGHUSER, and the cpuset limit forbids the allocation, then
out-of-memory killing is involved even when normal nodes and movable
nodes both have many free memory.
The OOM killer cannot help to resolve the situation as there is no
usable memory for the request in the cpuset scope. The only reasonable
measure to take is to fail the allocation right away and have the caller
to deal with it.
So add a check for cases like this in the slowpath of allocation, and
bail out early returning NULL for the allocation.
As page allocation is one of the hottest path in kernel, this check will
hurt all users with sane cpuset configuration, add a static branch check
and detect the abnormal config in cpuset memory binding setup so that
the extra check cost in page allocation is not paid by everyone.
[thanks to Micho Hocko and David Rientjes for suggesting not handling
it inside OOM code, adding cpuset check, refining comments]
Link: https://lkml.kernel.org/r/1632481657-68112-1-git-send-email-feng.tang@intel.com
Signed-off-by: Feng Tang <feng.tang@intel.com>
Suggested-by: Michal Hocko <mhocko@suse.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Zefan Li <lizefan.x@bytedance.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
GCC and Clang can use the "alloc_size" attribute to better inform the
results of __builtin_object_size() (for compile-time constant values).
Clang can additionally use alloc_size to inform the results of
__builtin_dynamic_object_size() (for run-time values).
Because GCC sees the frequent use of struct_size() as an allocator size
argument, and notices it can return SIZE_MAX (the overflow indication),
it complains about these call sites overflowing (since SIZE_MAX is
greater than the default -Walloc-size-larger-than=PTRDIFF_MAX). This
isn't helpful since we already know a SIZE_MAX will be caught at
run-time (this was an intentional design). To deal with this, we must
disable this check as it is both a false positive and redundant. (Clang
does not have this warning option.)
Unfortunately, just checking the -Wno-alloc-size-larger-than is not
sufficient to make the __alloc_size attribute behave correctly under
older GCC versions. The attribute itself must be disabled in those
situations too, as there appears to be no way to reliably silence the
SIZE_MAX constant expression cases for GCC versions less than 9.1:
In file included from ./include/linux/resource_ext.h:11,
from ./include/linux/pci.h:40,
from drivers/net/ethernet/intel/ixgbe/ixgbe.h:9,
from drivers/net/ethernet/intel/ixgbe/ixgbe_lib.c:4:
In function 'kmalloc_node',
inlined from 'ixgbe_alloc_q_vector' at ./include/linux/slab.h:743:9:
./include/linux/slab.h:618:9: error: argument 1 value '18446744073709551615' exceeds maximum object size 9223372036854775807 [-Werror=alloc-size-larger-than=]
return __kmalloc_node(size, flags, node);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
./include/linux/slab.h: In function 'ixgbe_alloc_q_vector':
./include/linux/slab.h:455:7: note: in a call to allocation function '__kmalloc_node' declared here
void *__kmalloc_node(size_t size, gfp_t flags, int node) __assume_slab_alignment __malloc;
^~~~~~~~~~~~~~
Specifically:
'-Wno-alloc-size-larger-than' is not correctly handled by GCC < 9.1
https://godbolt.org/z/hqsfG7q84 (doesn't disable)
https://godbolt.org/z/P9jdrPTYh (doesn't admit to not knowing about option)
https://godbolt.org/z/465TPMWKb (only warns when other warnings appear)
'-Walloc-size-larger-than=18446744073709551615' is not handled by GCC < 8.2
https://godbolt.org/z/73hh1EPxz (ignores numeric value)
Since anything marked with __alloc_size would also qualify for marking
with __malloc, just include __malloc along with it to avoid redundant
markings. (Suggested by Linus Torvalds.)
Finally, make sure checkpatch.pl doesn't get confused about finding the
__alloc_size attribute on functions. (Thanks to Joe Perches.)
Link: https://lkml.kernel.org/r/20210930222704.2631604-3-keescook@chromium.org
Signed-off-by: Kees Cook <keescook@chromium.org>
Tested-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Andy Whitcroft <apw@canonical.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Daniel Micay <danielmicay@gmail.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Dennis Zhou <dennis@kernel.org>
Cc: Dwaipayan Ray <dwaipayanray1@gmail.com>
Cc: Joe Perches <joe@perches.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Alexandre Bounine <alex.bou9@gmail.com>
Cc: Gustavo A. R. Silva <gustavoars@kernel.org>
Cc: Ira Weiny <ira.weiny@intel.com>
Cc: Jing Xiangfeng <jingxiangfeng@huawei.com>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: kernel test robot <lkp@intel.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Miguel Ojeda <ojeda@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Souptick Joarder <jrdr.linux@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Patch series "stackdepot, kasan, workqueue: Avoid expanding stackdepot
slabs when holding raw_spin_lock", v2.
Shuah Khan reported [1]:
| When CONFIG_PROVE_RAW_LOCK_NESTING=y and CONFIG_KASAN are enabled,
| kasan_record_aux_stack() runs into "BUG: Invalid wait context" when
| it tries to allocate memory attempting to acquire spinlock in page
| allocation code while holding workqueue pool raw_spinlock.
|
| There are several instances of this problem when block layer tries
| to __queue_work(). Call trace from one of these instances is below:
|
| kblockd_mod_delayed_work_on()
| mod_delayed_work_on()
| __queue_delayed_work()
| __queue_work() (rcu_read_lock, raw_spin_lock pool->lock held)
| insert_work()
| kasan_record_aux_stack()
| kasan_save_stack()
| stack_depot_save()
| alloc_pages()
| __alloc_pages()
| get_page_from_freelist()
| rm_queue()
| rm_queue_pcplist()
| local_lock_irqsave(&pagesets.lock, flags);
| [ BUG: Invalid wait context triggered ]
PROVE_RAW_LOCK_NESTING is pointing out that (on RT kernels) the locking
rules are being violated. More generally, memory is being allocated
from a non-preemptive context (raw_spin_lock'd c-s) where it is not
allowed.
To properly fix this, we must prevent stackdepot from replenishing its
"stack slab" pool if memory allocations cannot be done in the current
context: it's a bug to use either GFP_ATOMIC nor GFP_NOWAIT in certain
non-preemptive contexts, including raw_spin_locks (see gfp.h and commit
ab00db216c).
The only downside is that saving a stack trace may fail if: stackdepot
runs out of space AND the same stack trace has not been recorded before.
I expect this to be unlikely, and a simple experiment (boot the kernel)
didn't result in any failure to record stack trace from insert_work().
The series includes a few minor fixes to stackdepot that I noticed in
preparing the series. It then introduces __stack_depot_save(), which
exposes the option to force stackdepot to not allocate any memory.
Finally, KASAN is changed to use the new stackdepot interface and
provide kasan_record_aux_stack_noalloc(), which is then used by
workqueue code.
[1] https://lkml.kernel.org/r/20210902200134.25603-1-skhan@linuxfoundation.org
This patch (of 6):
<linux/stackdepot.h> refers to gfp_t, but doesn't include gfp.h.
Fix it by including <linux/gfp.h>.
Link: https://lkml.kernel.org/r/20210913112609.2651084-1-elver@google.com
Link: https://lkml.kernel.org/r/20210913112609.2651084-2-elver@google.com
Signed-off-by: Marco Elver <elver@google.com>
Tested-by: Shuah Khan <skhan@linuxfoundation.org>
Acked-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Lai Jiangshan <jiangshanlai@gmail.com>
Cc: Walter Wu <walter-zh.wu@mediatek.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Vijayanand Jitta <vjitta@codeaurora.org>
Cc: Vinayak Menon <vinmenon@codeaurora.org>
Cc: "Gustavo A. R. Silva" <gustavoars@kernel.org>
Cc: Taras Madan <tarasmadan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
With CONFIG_SLUB_CPU_PARTIAL enabled, SLUB keeps a percpu list of
partial slabs that can be promoted to cpu slab when the previous one is
depleted, without accessing the shared partial list. A slab can be
added to this list by 1) refill of an empty list from get_partial_node()
- once we really have to access the shared partial list, we acquire
multiple slabs to amortize the cost of locking, and 2) first free to a
previously full slab - instead of putting the slab on a shared partial
list, we can more cheaply freeze it and put it on the per-cpu list.
To control how large a percpu partial list can grow for a kmem cache,
set_cpu_partial() calculates a target number of free objects on each
cpu's percpu partial list, and this can be also set by the sysfs file
cpu_partial.
However, the tracking of actual number of objects is imprecise, in order
to limit overhead from cpu X freeing an objects to a slab on percpu
partial list of cpu Y. Basically, the percpu partial slabs form a
single linked list, and when we add a new slab to the list with current
head "oldpage", we set in the struct page of the slab we're adding:
page->pages = oldpage->pages + 1; // this is precise
page->pobjects = oldpage->pobjects + (page->objects - page->inuse);
page->next = oldpage;
Thus the real number of free objects in the slab (objects - inuse) is
only determined at the moment of adding the slab to the percpu partial
list, and further freeing doesn't update the pobjects counter nor
propagate it to the current list head. As Jann reports [1], this can
easily lead to large inaccuracies, where the target number of objects
(up to 30 by default) can translate to the same number of (empty) slab
pages on the list. In case 2) above, we put a slab with 1 free object
on the list, thus only increase page->pobjects by 1, even if there are
subsequent frees on the same slab. Jann has noticed this in practice
and so did we [2] when investigating significant increase of kmemcg
usage after switching from SLAB to SLUB.
While this is no longer a problem in kmemcg context thanks to the
accounting rewrite in 5.9, the memory waste is still not ideal and it's
questionable whether it makes sense to perform free object count based
control when object counts can easily become so much inaccurate. So
this patch converts the accounting to be based on number of pages only
(which is precise) and removes the page->pobjects field completely.
This is also ultimately simpler.
To retain the existing set_cpu_partial() heuristic, first calculate the
target number of objects as previously, but then convert it to target
number of pages by assuming the pages will be half-filled on average.
This assumption might obviously also be inaccurate in practice, but
cannot degrade to actual number of pages being equal to the target
number of objects.
We could also skip the intermediate step with target number of objects
and rewrite the heuristic in terms of pages. However we still have the
sysfs file cpu_partial which uses number of objects and could break
existing users if it suddenly becomes number of pages, so this patch
doesn't do that.
In practice, after this patch the heuristics limit the size of percpu
partial list up to 2 pages. In case of a reported regression (which
would mean some workload has benefited from the previous imprecise
object based counting), we can tune the heuristics to get a better
compromise within the new scheme, while still avoid the unexpectedly
long percpu partial lists.
[1] https://lore.kernel.org/linux-mm/CAG48ez2Qx5K1Cab-m8BdSibp6wLTip6ro4=-umR7BLsEgjEYzA@mail.gmail.com/
[2] https://lore.kernel.org/all/2f0f46e8-2535-410a-1859-e9cfa4e57c18@suse.cz/
==========
Evaluation
==========
Mel was kind enough to run v1 through mmtests machinery for netperf
(localhost) and hackbench and, for most significant results see below.
So there are some apparent regressions, especially with hackbench, which
I think ultimately boils down to having shorter percpu partial lists on
average and some benchmarks benefiting from longer ones. Monitoring
slab usage also indicated less memory usage by slab. Based on that, the
following patch will bump the defaults to allow longer percpu partial
lists than after this patch.
However the goal is certainly not such that we would limit the percpu
partial lists to 30 pages just because previously a specific alloc/free
pattern could lead to the limit of 30 objects translate to a limit to 30
pages - that would make little sense. This is a correctness patch, and
if a workload benefits from larger lists, the sysfs tuning knobs are
still there to allow that.
Netperf
2-socket Intel(R) Xeon(R) Gold 5218R CPU @ 2.10GHz (20 cores, 40 threads per socket), 384GB RAM
TCP-RR:
hmean before 127045.79 after 121092.94 (-4.69%, worse)
stddev before 2634.37 after 1254.08
UDP-RR:
hmean before 166985.45 after 160668.94 ( -3.78%, worse)
stddev before 4059.69 after 1943.63
2-socket Intel(R) Xeon(R) CPU E5-2698 v4 @ 2.20GHz (20 cores, 40 threads per socket), 512GB RAM
TCP-RR:
hmean before 84173.25 after 76914.72 ( -8.62%, worse)
UDP-RR:
hmean before 93571.12 after 96428.69 ( 3.05%, better)
stddev before 23118.54 after 16828.14
2-socket Intel(R) Xeon(R) CPU E5-2670 v3 @ 2.30GHz (12 cores, 24 threads per socket), 64GB RAM
TCP-RR:
hmean before 49984.92 after 48922.27 ( -2.13%, worse)
stddev before 6248.15 after 4740.51
UDP-RR:
hmean before 61854.31 after 68761.81 ( 11.17%, better)
stddev before 4093.54 after 5898.91
other machines - within 2%
Hackbench
(results before and after the patch, negative % means worse)
2-socket AMD EPYC 7713 (64 cores, 128 threads per core), 256GB RAM
hackbench-process-sockets
Amean 1 0.5380 0.5583 ( -3.78%)
Amean 4 0.7510 0.8150 ( -8.52%)
Amean 7 0.7930 0.9533 ( -20.22%)
Amean 12 0.7853 1.1313 ( -44.06%)
Amean 21 1.1520 1.4993 ( -30.15%)
Amean 30 1.6223 1.9237 ( -18.57%)
Amean 48 2.6767 2.9903 ( -11.72%)
Amean 79 4.0257 5.1150 ( -27.06%)
Amean 110 5.5193 7.4720 ( -35.38%)
Amean 141 7.2207 9.9840 ( -38.27%)
Amean 172 8.4770 12.1963 ( -43.88%)
Amean 203 9.6473 14.3137 ( -48.37%)
Amean 234 11.3960 18.7917 ( -64.90%)
Amean 265 13.9627 22.4607 ( -60.86%)
Amean 296 14.9163 26.0483 ( -74.63%)
hackbench-thread-sockets
Amean 1 0.5597 0.5877 ( -5.00%)
Amean 4 0.7913 0.8960 ( -13.23%)
Amean 7 0.8190 1.0017 ( -22.30%)
Amean 12 0.9560 1.1727 ( -22.66%)
Amean 21 1.7587 1.5660 ( 10.96%)
Amean 30 2.4477 1.9807 ( 19.08%)
Amean 48 3.4573 3.0630 ( 11.41%)
Amean 79 4.7903 5.1733 ( -8.00%)
Amean 110 6.1370 7.4220 ( -20.94%)
Amean 141 7.5777 9.2617 ( -22.22%)
Amean 172 9.2280 11.0907 ( -20.18%)
Amean 203 10.2793 13.3470 ( -29.84%)
Amean 234 11.2410 17.1070 ( -52.18%)
Amean 265 12.5970 23.3323 ( -85.22%)
Amean 296 17.1540 24.2857 ( -41.57%)
2-socket Intel(R) Xeon(R) Gold 5218R CPU @ 2.10GHz (20 cores, 40 threads
per socket), 384GB RAM
hackbench-process-sockets
Amean 1 0.5760 0.4793 ( 16.78%)
Amean 4 0.9430 0.9707 ( -2.93%)
Amean 7 1.5517 1.8843 ( -21.44%)
Amean 12 2.4903 2.7267 ( -9.49%)
Amean 21 3.9560 4.2877 ( -8.38%)
Amean 30 5.4613 5.8343 ( -6.83%)
Amean 48 8.5337 9.2937 ( -8.91%)
Amean 79 14.0670 15.2630 ( -8.50%)
Amean 110 19.2253 21.2467 ( -10.51%)
Amean 141 23.7557 25.8550 ( -8.84%)
Amean 172 28.4407 29.7603 ( -4.64%)
Amean 203 33.3407 33.9927 ( -1.96%)
Amean 234 38.3633 39.1150 ( -1.96%)
Amean 265 43.4420 43.8470 ( -0.93%)
Amean 296 48.3680 48.9300 ( -1.16%)
hackbench-thread-sockets
Amean 1 0.6080 0.6493 ( -6.80%)
Amean 4 1.0000 1.0513 ( -5.13%)
Amean 7 1.6607 2.0260 ( -22.00%)
Amean 12 2.7637 2.9273 ( -5.92%)
Amean 21 5.0613 4.5153 ( 10.79%)
Amean 30 6.3340 6.1140 ( 3.47%)
Amean 48 9.0567 9.5577 ( -5.53%)
Amean 79 14.5657 15.7983 ( -8.46%)
Amean 110 19.6213 21.6333 ( -10.25%)
Amean 141 24.1563 26.2697 ( -8.75%)
Amean 172 28.9687 30.2187 ( -4.32%)
Amean 203 33.9763 34.6970 ( -2.12%)
Amean 234 38.8647 39.3207 ( -1.17%)
Amean 265 44.0813 44.1507 ( -0.16%)
Amean 296 49.2040 49.4330 ( -0.47%)
2-socket Intel(R) Xeon(R) CPU E5-2698 v4 @ 2.20GHz (20 cores, 40 threads
per socket), 512GB RAM
hackbench-process-sockets
Amean 1 0.5027 0.5017 ( 0.20%)
Amean 4 1.1053 1.2033 ( -8.87%)
Amean 7 1.8760 2.1820 ( -16.31%)
Amean 12 2.9053 3.1810 ( -9.49%)
Amean 21 4.6777 4.9920 ( -6.72%)
Amean 30 6.5180 6.7827 ( -4.06%)
Amean 48 10.0710 10.5227 ( -4.48%)
Amean 79 16.4250 17.5053 ( -6.58%)
Amean 110 22.6203 24.4617 ( -8.14%)
Amean 141 28.0967 31.0363 ( -10.46%)
Amean 172 34.4030 36.9233 ( -7.33%)
Amean 203 40.5933 43.0850 ( -6.14%)
Amean 234 46.6477 48.7220 ( -4.45%)
Amean 265 53.0530 53.9597 ( -1.71%)
Amean 296 59.2760 59.9213 ( -1.09%)
hackbench-thread-sockets
Amean 1 0.5363 0.5330 ( 0.62%)
Amean 4 1.1647 1.2157 ( -4.38%)
Amean 7 1.9237 2.2833 ( -18.70%)
Amean 12 2.9943 3.3110 ( -10.58%)
Amean 21 4.9987 5.1880 ( -3.79%)
Amean 30 6.7583 7.0043 ( -3.64%)
Amean 48 10.4547 10.8353 ( -3.64%)
Amean 79 16.6707 17.6790 ( -6.05%)
Amean 110 22.8207 24.4403 ( -7.10%)
Amean 141 28.7090 31.0533 ( -8.17%)
Amean 172 34.9387 36.8260 ( -5.40%)
Amean 203 41.1567 43.0450 ( -4.59%)
Amean 234 47.3790 48.5307 ( -2.43%)
Amean 265 53.9543 54.6987 ( -1.38%)
Amean 296 60.0820 60.2163 ( -0.22%)
1-socket Intel(R) Xeon(R) CPU E3-1240 v5 @ 3.50GHz (4 cores, 8 threads),
32 GB RAM
hackbench-process-sockets
Amean 1 1.4760 1.5773 ( -6.87%)
Amean 3 3.9370 4.0910 ( -3.91%)
Amean 5 6.6797 6.9357 ( -3.83%)
Amean 7 9.3367 9.7150 ( -4.05%)
Amean 12 15.7627 16.1400 ( -2.39%)
Amean 18 23.5360 23.6890 ( -0.65%)
Amean 24 31.0663 31.3137 ( -0.80%)
Amean 30 38.7283 39.0037 ( -0.71%)
Amean 32 41.3417 41.6097 ( -0.65%)
hackbench-thread-sockets
Amean 1 1.5250 1.6043 ( -5.20%)
Amean 3 4.0897 4.2603 ( -4.17%)
Amean 5 6.7760 7.0933 ( -4.68%)
Amean 7 9.4817 9.9157 ( -4.58%)
Amean 12 15.9610 16.3937 ( -2.71%)
Amean 18 23.9543 24.3417 ( -1.62%)
Amean 24 31.4400 31.7217 ( -0.90%)
Amean 30 39.2457 39.5467 ( -0.77%)
Amean 32 41.8267 42.1230 ( -0.71%)
2-socket Intel(R) Xeon(R) CPU E5-2670 v3 @ 2.30GHz (12 cores, 24 threads
per socket), 64GB RAM
hackbench-process-sockets
Amean 1 1.0347 1.0880 ( -5.15%)
Amean 4 1.7267 1.8527 ( -7.30%)
Amean 7 2.6707 2.8110 ( -5.25%)
Amean 12 4.1617 4.3383 ( -4.25%)
Amean 21 7.0070 7.2600 ( -3.61%)
Amean 30 9.9187 10.2397 ( -3.24%)
Amean 48 15.6710 16.3923 ( -4.60%)
Amean 79 24.7743 26.1247 ( -5.45%)
Amean 110 34.3000 35.9307 ( -4.75%)
Amean 141 44.2043 44.8010 ( -1.35%)
Amean 172 54.2430 54.7260 ( -0.89%)
Amean 192 60.6557 60.9777 ( -0.53%)
hackbench-thread-sockets
Amean 1 1.0610 1.1353 ( -7.01%)
Amean 4 1.7543 1.9140 ( -9.10%)
Amean 7 2.7840 2.9573 ( -6.23%)
Amean 12 4.3813 4.4937 ( -2.56%)
Amean 21 7.3460 7.5350 ( -2.57%)
Amean 30 10.2313 10.5190 ( -2.81%)
Amean 48 15.9700 16.5940 ( -3.91%)
Amean 79 25.3973 26.6637 ( -4.99%)
Amean 110 35.1087 36.4797 ( -3.91%)
Amean 141 45.8220 46.3053 ( -1.05%)
Amean 172 55.4917 55.7320 ( -0.43%)
Amean 192 62.7490 62.5410 ( 0.33%)
Link: https://lkml.kernel.org/r/20211012134651.11258-1-vbabka@suse.cz
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reported-by: Jann Horn <jannh@google.com>
Cc: Roman Gushchin <guro@fb.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch is to fix an out-of-bound access issue when jit-ing the
bpf_pseudo_func insn (i.e. ld_imm64 with src_reg == BPF_PSEUDO_FUNC)
In jit_subprog(), it currently reuses the subprog index cached in
insn[1].imm. This subprog index is an index into a few array related
to subprogs. For example, in jit_subprog(), it is an index to the newly
allocated 'struct bpf_prog **func' array.
The subprog index was cached in insn[1].imm after add_subprog(). However,
this could become outdated (and too big in this case) if some subprogs
are completely removed during dead code elimination (in
adjust_subprog_starts_after_remove). The cached index in insn[1].imm
is not updated accordingly and causing out-of-bound issue in the later
jit_subprog().
Unlike bpf_pseudo_'func' insn, the current bpf_pseudo_'call' insn
is handling the DCE properly by calling find_subprog(insn->imm) to
figure out the index instead of caching the subprog index.
The existing bpf_adj_branches() will adjust the insn->imm
whenever insn is added or removed.
Instead of having two ways handling subprog index,
this patch is to make bpf_pseudo_func works more like
bpf_pseudo_call.
First change is to stop caching the subprog index result
in insn[1].imm after add_subprog(). The verification
process will use find_subprog(insn->imm) to figure
out the subprog index.
Second change is in bpf_adj_branches() and have it to
adjust the insn->imm for the bpf_pseudo_func insn also
whenever insn is added or removed.
Third change is in jit_subprog(). Like the bpf_pseudo_call handling,
bpf_pseudo_func temporarily stores the find_subprog() result
in insn->off. It is fine because the prog's insn has been finalized
at this point. insn->off will be reset back to 0 later to avoid
confusing the userspace prog dump tool.
Fixes: 69c087ba62 ("bpf: Add bpf_for_each_map_elem() helper")
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20211106014014.651018-1-kafai@fb.com
For creating fattrs with the label field already allocated for us. I
also update nfs_free_fattr() to free the label in the end.
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
- Make of_phandle_args_to_fwspec() generally available (Marc Zyngier)
- Allow matching of interrupt-maps local to interrupt controller or PCI
device (Marc Zyngier)
- Add Apple SoC (e.g., M1) PCIe host controller driver, which enables
access to USB type-A, Ethernet, Wi-Fi, Bluetooth devices; these require
additional drivers of their own (Alyssa Rosenzweig)
- Add apple INTx, per-port, and MSI interrupt support (Marc Zyngier)
- Configure apple Requester-ID-to-Stream-ID mapper for IOMMU (DART) support
(Marc Zyngier)
* pci/host/apple:
PCI: apple: Configure RID to SID mapper on device addition
iommu/dart: Exclude MSI doorbell from PCIe device IOVA range
PCI: apple: Implement MSI support
PCI: apple: Add INTx and per-port interrupt support
PCI: apple: Set up reference clocks when probing
PCI: apple: Add initial hardware bring-up
PCI: of: Allow matching of an interrupt-map local to a PCI device
of/irq: Allow matching of an interrupt-map local to an interrupt controller
irqdomain: Make of_phandle_args_to_fwspec() generally available
- Define macros for PCI_EXP_DEVCTL_PAYLOAD_* (Pali Rohár)
- Set Max Payload Size to 512 bytes per Marvell spec (Pali Rohár)
- Downgrade PIO Response Status messages to debug level (Marek Behún)
- Preserve CRS SV (Config Request Retry Software Visibility) bit in
emulated Root Control register (Pali Rohár)
- Fix issue in configuring reference clock (Pali Rohár)
- Don't clear status bits for masked interrupts (Pali Rohár)
- Don't mask unused interrupts (Pali Rohár)
- Avoid code repetition in advk_pcie_rd_conf() (Marek Behún)
- Retry config accesses on CRS response (Pali Rohár)
- Simplify emulated Root Capabilities initialization (Pali Rohár)
- Fix several link training issues (Pali Rohár)
- Fix link-up checking via LTSSM (Pali Rohár)
- Fix reporting of Data Link Layer Link Active (Pali Rohár)
- Fix emulation of W1C bits (Marek Behún)
- Fix MSI domain .alloc() method to return zero on success (Marek Behún)
- Read entire 16-bit MSI vector in MSI handler, not just low 8 bits (Marek
Behún)
- Clear Root Port I/O Space, Memory Space, and Bus Master Enable bits at
startup; PCI core will set those as necessary (Pali Rohár)
- When operating as a Root Port, set class code to "PCI Bridge" instead of
the default "Mass Storage Controller" (Pali Rohár)
- Add emulation for PCI_BRIDGE_CTL_BUS_RESET since aardvark doesn't
implement this per spec (Pali Rohár)
- Add emulation of option ROM BAR since aardvark doesn't implement this per
spec (Pali Rohár)
* remotes/lorenzo/pci/aardvark:
PCI: aardvark: Fix support for PCI_ROM_ADDRESS1 on emulated bridge
PCI: aardvark: Fix support for PCI_BRIDGE_CTL_BUS_RESET on emulated bridge
PCI: aardvark: Set PCI Bridge Class Code to PCI Bridge
PCI: aardvark: Fix support for bus mastering and PCI_COMMAND on emulated bridge
PCI: aardvark: Read all 16-bits from PCIE_MSI_PAYLOAD_REG
PCI: aardvark: Fix return value of MSI domain .alloc() method
PCI: pci-bridge-emul: Fix emulation of W1C bits
PCI: aardvark: Fix reporting Data Link Layer Link Active
PCI: aardvark: Fix checking for link up via LTSSM state
PCI: aardvark: Fix link training
PCI: aardvark: Simplify initialization of rootcap on virtual bridge
PCI: aardvark: Implement re-issuing config requests on CRS response
PCI: aardvark: Deduplicate code in advk_pcie_rd_conf()
PCI: aardvark: Do not unmask unused interrupts
PCI: aardvark: Do not clear status bits of masked interrupts
PCI: aardvark: Fix configuring Reference clock
PCI: aardvark: Fix preserving PCI_EXP_RTCTL_CRSSVE flag on emulated bridge
PCI: aardvark: Don't spam about PIO Response Status
PCI: aardvark: Fix PCIe Max Payload Size setting
PCI: Add PCI_EXP_DEVCTL_PAYLOAD_* macros
- Tidy setup-irq.c comments (Pranay Sanghai)
- Fix misspellings (Krzysztof Wilczyński)
- Fix sprintf(), sscanf() format mismatches (Krzysztof Wilczyński)
- Tidy cpqphp code formatting (Krzysztof Wilczyński)
- Remove unused pci_pool wrappers, which have been replaced by dma_pool
(Cai Huoqing)
- Remove a redundant initialization in __pci_reset_function_locked() (Colin
Ian King)
- Use 'unsigned int' instead of 'unsigned' (Krzysztof Wilczyński)
- Update PCI subsystem information in MAINTAINERS (Krzysztof Wilczyński)
- Include generic <linux/> headers instead of <asm/> for cpqphp and vmd
(Krzysztof Wilczyński)
* pci/misc:
PCI: vmd: Drop redundant includes of <asm/device.h>, <asm/msi.h>
PCI: cpqphp: Use <linux/io.h> instead of <asm/io.h>
MAINTAINERS: Update PCI subsystem information
PCI: Prefer 'unsigned int' over bare 'unsigned'
PCI: Remove redundant 'rc' initialization
PCI: Remove unused pci_pool wrappers
PCI: cpqphp: Format if-statement code block correctly
PCI: Use unsigned to match sscanf("%x") in pci_dev_str_match_path()
PCI: hv: Remove unnecessary use of %hx
PCI: Correct misspelled and remove duplicated words
PCI: Tidy comments
- Add pci_read_vpd_any(), pci_write_vpd_any() to access VPD at arbitrary
offsets (Heiner Kallweit)
- Use VPD API to replace custom code in cxgb3 driver (Heiner Kallweit)
* pci/vpd:
cxgb3: Remove seeprom_write and use VPD API
cxgb3: Use VPD API in t3_seeprom_wp()
cxgb3: Remove t3_seeprom_read and use VPD API
PCI/VPD: Use pci_read_vpd_any() in pci_vpd_size()
PCI/VPD: Add pci_read/write_vpd_any()
