mbox series

[v1,0/2] memcg, vmpressure: expose vmpressure controls

Message ID 20200414171840.22053-1-lmoiseichuk@magicleap.com (mailing list archive)
Headers show
Series memcg, vmpressure: expose vmpressure controls | expand

Message

svc_lmoiseichuk@magicleap.com April 14, 2020, 5:18 p.m. UTC
From: Leonid Moiseichuk <lmoiseichuk@magicleap.com>

Small tweak to populate vmpressure parameters to userspace without
any built-in logic change.

The vmpressure is used actively (e.g. on Android) to track mm stress.
vmpressure parameters selected empiricaly quite long time ago and not
always suitable for modern memory configurations. Modern 8 GB devices
starts triggering medium threshold when about 3 GB memory not used,
and situation with 12 and 16 GB devices even worse.

Testing performed:

* Build kernel for x86-64 and aarch64 (Tegra X2 SoC)
* Booted and checked that all properties published
* Tuned settings and observed adequate response
* sysbench and memory bubble application (memsize from lmbench)
  used to verify reactions

Change Log:

  v1:

  => updated commit with 8 GB numbers per Michal Hocko request

  v0:

  => initial implementation
  => tested on x86-64 and aarch64

Leonid Moiseichuk (2):
  memcg, vmpressure: expose vmpressure controls
  memcg, vmpressure: expose vmpressure controls

 .../admin-guide/cgroup-v1/memory.rst          |  12 +-
 include/linux/vmpressure.h                    |  35 ++++++
 mm/memcontrol.c                               | 113 ++++++++++++++++++
 mm/vmpressure.c                               | 101 +++++++---------
 4 files changed, 200 insertions(+), 61 deletions(-)

Comments

Michal Hocko April 14, 2020, 6:35 p.m. UTC | #1
On Tue 14-04-20 13:18:38, svc_lmoiseichuk@magicleap.com wrote:
> From: Leonid Moiseichuk <lmoiseichuk@magicleap.com>
> 
> Small tweak to populate vmpressure parameters to userspace without
> any built-in logic change.
> 
> The vmpressure is used actively (e.g. on Android) to track mm stress.
> vmpressure parameters selected empiricaly quite long time ago and not
> always suitable for modern memory configurations. Modern 8 GB devices
> starts triggering medium threshold when about 3 GB memory not used,
> and situation with 12 and 16 GB devices even worse.

I am sorry but this doesn't answer questions I've had for the previous
version of the patch. Please do not post newer versions until there is a
consensusn on the approach based on the review feedback. In order to not
fragment the discussion, let's continue in the original email thread
http://lkml.kernel.org/r/20200413215750.7239-1-lmoiseichuk@magicleap.com