diff options
author | LeiZhou-97 <lei.zhou@intel.com> | 2023-06-13 18:59:29 +0800 |
---|---|---|
committer | Tejun Heo <tj@kernel.org> | 2023-06-21 10:43:29 -1000 |
commit | e973dfe9299441ad263efedf094897fbb31e6a87 (patch) | |
tree | f82e03504488762181fbca61573c7e47ee4e4942 /Documentation/gpu/drm-uapi.rst | |
parent | 36de5f303ca1bd6fce74815ef17ef3d8ff8737b5 (diff) |
cgroup/misc: Expose misc.current on cgroup v2 root
Hello,
This patch is to expose misc.current on cgroup v2 root for tracking
how much of the resource has been consumed in total on the system.
Most of the cloud infrastucture use cgroup to fetch the host
information for scheduling purpose.
Currently, the misc controller can be used by Intel TDX HKIDs and
AMD SEV ASIDs, which are both used for creating encrypted VMs.
Intel TDX and AMD SEV are mostly be used by the cloud providers
for providing confidential VMs.
In actual use of a server, these confidential VMs may be launched
in different ways. For the cloud solution, there are kubvirt and
coco (tracked by kubepods.slice); on host, they can be booted
directly through qemu by end user (tracked by user.slice), etc.
In this complex environment, when wanting to know how many resource
is used in total it has to iterate through all existing slices to
get the value of each misc.current and add them up to calculate
the total number of consumed keys.
So exposing misc.current to root cgroup tends to give much easier
when calculates how much resource has been used in total, which
helps to schedule and count resources for the cloud infrastucture.
Signed-off-by: LeiZhou-97 <lei.zhou@intel.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Diffstat (limited to 'Documentation/gpu/drm-uapi.rst')
0 files changed, 0 insertions, 0 deletions