linux/include
SeongJae Park 0b9b241406 inet: frags: batch fqdir destroy works
On a few of our systems, I found frequent 'unshare(CLONE_NEWNET)' calls
make the number of active slab objects including 'sock_inode_cache' type
rapidly and continuously increase.  As a result, memory pressure occurs.

In more detail, I made an artificial reproducer that resembles the
workload that we found the problem and reproduce the problem faster.  It
merely repeats 'unshare(CLONE_NEWNET)' 50,000 times in a loop.  It takes
about 2 minutes.  On 40 CPU cores / 70GB DRAM machine, the available
memory continuously reduced in a fast speed (about 120MB per second,
15GB in total within the 2 minutes).  Note that the issue don't
reproduce on every machine.  On my 6 CPU cores machine, the problem
didn't reproduce.

'cleanup_net()' and 'fqdir_work_fn()' are functions that deallocate the
relevant memory objects.  They are asynchronously invoked by the work
queues and internally use 'rcu_barrier()' to ensure safe destructions.
'cleanup_net()' works in a batched maneer in a single thread worker,
while 'fqdir_work_fn()' works for each 'fqdir_exit()' call in the
'system_wq'.  Therefore, 'fqdir_work_fn()' called frequently under the
workload and made the contention for 'rcu_barrier()' high.  In more
detail, the global mutex, 'rcu_state.barrier_mutex' became the
bottleneck.

This commit avoids such contention by doing the 'rcu_barrier()' and
subsequent lightweight works in a batched manner, as similar to that of
'cleanup_net()'.  The fqdir hashtable destruction, which is done before
the 'rcu_barrier()', is still allowed to run in parallel for fast
processing, but this commit makes it to use a dedicated work queue
instead of the 'system_wq', to make sure that the number of threads is
bounded.

Signed-off-by: SeongJae Park <sjpark@amazon.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20201211112405.31158-1-sjpark@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2020-12-12 15:08:54 -08:00
..
acpi pci-v5.10-changes 2020-10-22 12:41:00 -07:00
asm-generic Merge branch 'for-5.10-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/dennis/percpu 2020-11-15 08:57:19 -08:00
clocksource
crypto
drm drm: drm_print.h: fix kernel-doc markups 2020-10-27 11:21:39 +01:00
dt-bindings dt-bindings: firmware: add IMX_SC_R_CAN(x) macro for CAN 2020-11-20 12:06:45 +01:00
keys rxrpc: Don't leak the service-side session key to userspace 2020-11-23 18:09:29 +00:00
kunit kunit: fix display of failed expectations for strings 2020-11-10 13:45:15 -07:00
kvm ARM: 2020-10-23 11:17:56 -07:00
linux Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec-next 2020-12-12 12:28:42 -08:00
math-emu
media ARM: SoC platform updates 2020-10-24 10:33:08 -07:00
memory
misc
net inet: frags: batch fqdir destroy works 2020-12-12 15:08:54 -08:00
pcmcia
ras mm,hwpoison: introduce MF_MSG_UNSPLIT_THP 2020-10-16 11:11:17 -07:00
rdma net: don't include ethtool.h from netdevice.h 2020-11-23 17:27:04 -08:00
scsi scsi: libiscsi: Fix NOP race condition 2020-11-16 22:32:50 -05:00
soc Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net 2020-12-11 22:29:38 -08:00
sound ASoC: Fixes for v5.11 2020-11-19 19:56:29 +01:00
target
trace Merge https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next 2020-12-04 07:48:12 -08:00
uapi A new set of wireless changes: 2020-12-12 10:07:56 -08:00
vdso
video gpu: ipu-v3: remove unused functions 2020-10-26 10:42:38 +01:00
xen xen: branch for v5.10-rc1c 2020-10-25 10:55:35 -07:00