systemd/5009-cgroup-Restrict-effective-limits-with-global-resourc.patch
Franck Bui ebea3f9a92 - Add patches that implement [jsc#PED-5659]
5003-cgroup-rename-TasksMax-structure-to-CGroupTasksMax.patch
  5004-bus-print-properties-ignore-CGROUP_LIMIT_MAX-for-Mem.patch
  5005-bus-print-properties-prettify-more-unset-properties.patch
  5006-cgroup-Add-EffectiveMemoryMax-EffectiveMemoryHigh-an.patch
  5007-test-Convert-rlimit-test-to-subtest-of-generic-limit.patch
  5008-test-Add-effective-cgroup-limits-testing.patch
  5009-cgroup-Restrict-effective-limits-with-global-resourc.patch
  5010-cgroup-Rename-effective-limits-internal-table.patch
  They are temporarily put in quarantine to get broader testing but should be
  eventually moved to the git repo.

OBS-URL: https://build.opensuse.org/package/show/Base:System/systemd?expand=0&rev=1485
2024-01-23 13:52:53 +00:00

57 lines
2.6 KiB
Diff

From edfd70ab2c5490131fcf5a4348e31c19251d3479 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Michal=20Koutn=C3=BD?= <mkoutny@suse.com>
Date: Tue, 23 Jan 2024 11:30:20 +0100
Subject: [PATCH 5009/5011] cgroup: Restrict effective limits with global
resource provision
Global resource (whole system or root cg's (e.g. in a container)) is
also a well-defined limit for memory and tasks, take it into account
when calculating effective limits.
(cherry picked from commit 93f8e88d23bd383b5134f32c1e2ee315ac3a38c8)
[mkoutny: fixes jsc#PED-5659]
---
man/systemd.resource-control.xml | 2 +-
src/core/cgroup.c | 11 +++++++++++
2 files changed, 12 insertions(+), 1 deletion(-)
diff --git a/man/systemd.resource-control.xml b/man/systemd.resource-control.xml
index 5b7900af87..ded4722e7a 100644
--- a/man/systemd.resource-control.xml
+++ b/man/systemd.resource-control.xml
@@ -422,7 +422,7 @@ CPUWeight=20 DisableControllers=cpu / \
<literal>memory.max</literal> control group attribute. For details about this control group attribute, see
<ulink url="https://docs.kernel.org/admin-guide/cgroup-v2.html#memory-interface-files">Memory Interface Files</ulink>.
The effective configuration is reported as <varname>EffectiveMemoryMax=</varname> (the value is
- the most stringent limit of the unit and parent slices).</para>
+ the most stringent limit of the unit and parent slices and it is capped by physical memory).</para>
<para>While <varname>StartupMemoryMax=</varname> applies to the startup and shutdown phases of the system,
<varname>MemoryMax=</varname> applies to normal runtime of the system, and if the former is not set also to
diff --git a/src/core/cgroup.c b/src/core/cgroup.c
index 0842036559..a53d90486c 100644
--- a/src/core/cgroup.c
+++ b/src/core/cgroup.c
@@ -4006,6 +4006,17 @@ static uint64_t unit_get_effective_limit_one(Unit *u, CGroupLimitType type) {
assert(u);
assert(UNIT_HAS_CGROUP_CONTEXT(u));
+ if (unit_has_name(u, SPECIAL_ROOT_SLICE))
+ switch (type) {
+ case CGROUP_LIMIT_MEMORY_MAX:
+ case CGROUP_LIMIT_MEMORY_HIGH:
+ return physical_memory();
+ case CGROUP_LIMIT_TASKS_MAX:
+ return system_tasks_max();
+ default:
+ assert_not_reached();
+ }
+
cc = unit_get_cgroup_context(u);
switch (type) {
/* Note: on legacy/hybrid hierarchies memory_max stays CGROUP_LIMIT_MAX unless configured
--
2.35.3