From 38b9342ee62e9a1a868654ba6619de9da251bd57 Mon Sep 17 00:00:00 2001 From: Vishal Chourasia Date: Fri, 6 Jun 2025 09:40:05 +0530 Subject: Documentation: cgroup: add section explaining controller availability MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Add "Availability" section to Control Group v2 docs. It describes the meaning of a controller being available in a cgroup, complementing the existing "Enabling and Disabling" section. This update improves the clarity of cgroup controller management by explicitly distinguishing between: 1. Availability – when a controller is supported by the kernel and listed in "cgroup.controllers", making its interface files accessible in the cgroup's directory. 2. Enabling – when a controller is enabled via explicitly writing the name of the controller to "cgroup.subtree_control" to control distribution of resource across the cgroup's immediate children. As an example, consider /sys/fs/cgroup # cat cgroup.controllers cpuset cpu io memory hugetlb pids misc /sys/fs/cgroup # cat cgroup.subtree_control # No controllers enabled by default /sys/fs/cgroup # echo +cpu +memory > cgroup.subtree_control # enabling "cpu" and "memory" /sys/fs/cgroup # cat cgroup.subtree_control cpu memory # cpu and memory enabled in /sys/fs/cgroup /sys/fs/cgroup # mkdir foo_cgrp /sys/fs/cgroup # cd foo_cgrp/ /sys/fs/cgroup/foo_cgrp # cat cgroup.controllers cpu memory # cpu and memory available in 'foo_cgrp' /sys/fs/cgroup/foo_cgrp # cat cgroup.subtree_control # empty by default /sys/fs/cgroup/foo_cgrp # ls cgroup.controllers cpu.max.burst memory.numa_stat cgroup.events cpu.pressure memory.oom.group cgroup.freeze cpu.stat memory.peak cgroup.kill cpu.stat.local memory.pressure cgroup.max.depth cpu.weight memory.reclaim cgroup.max.descendants cpu.weight.nice memory.stat cgroup.pressure io.pressure memory.swap.current cgroup.procs memory.current memory.swap.events cgroup.stat memory.events memory.swap.high cgroup.subtree_control memory.events.local memory.swap.max cgroup.threads memory.high memory.swap.peak cgroup.type memory.low memory.zswap.current cpu.idle memory.max memory.zswap.max cpu.max memory.min memory.zswap.writeback In this example, "cpu" and "memory" are enabled in the root cgroup, making them available in "foo_cgrp". This exposes the corresponding interface files in "foo_cgrp/", allowing resource control of processes in that cgroup. However, these controllers are not yet enabled in "foo_cgrp" itself. Once a controller is available in a cgroup it can be used to resource control processes of the cgroup. Acked-by: Michal Koutný Reviewed-by: Bagas Sanjaya Signed-off-by: Vishal Chourasia Signed-off-by: Tejun Heo --- Documentation/admin-guide/cgroup-v2.rst | 9 +++++++++ 1 file changed, 9 insertions(+) (limited to 'Documentation/admin-guide') diff --git a/Documentation/admin-guide/cgroup-v2.rst b/Documentation/admin-guide/cgroup-v2.rst index 0cc35a14afbe..31acc64e656f 100644 --- a/Documentation/admin-guide/cgroup-v2.rst +++ b/Documentation/admin-guide/cgroup-v2.rst @@ -435,6 +435,15 @@ both cgroups. Controlling Controllers ----------------------- +Availablity +~~~~~~~~~~~ + +A controller is available in a cgroup when it is supported by the kernel (i.e., +compiled in, not disabled and not attached to a v1 hierarchy) and listed in the +"cgroup.controllers" file. Availability means the controller's interface files +are exposed in the cgroup’s directory, allowing the distribution of the target +resource to be observed or controlled within that cgroup. + Enabling and Disabling ~~~~~~~~~~~~~~~~~~~~~~ -- cgit v1.2.3 From 646faf36d7271c597497ca547a59912fcab49be9 Mon Sep 17 00:00:00 2001 From: Michal Koutný Date: Fri, 18 Jul 2025 11:18:54 +0200 Subject: cgroup: Add compatibility option for content of /proc/cgroups MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit /proc/cgroups lists only v1 controllers by default, however, this is only enforced since the commit af000ce85293b ("cgroup: Do not report unavailable v1 controllers in /proc/cgroups") and there is software in the wild that uses content of /proc/cgroups to decide on availability of v2 (sic) controllers. Add a boottime param that can bring back the previous behavior for setups where the check in the software cannot be changed and it causes e.g. unintended OOMs. Also, this patch takes out cgrp_v1_visible from cgroup1_subsys_absent() guard since it's only important to check which hierarchy (v1 vs v2) the subsys is attached to. This has no effect on the printed message but the code is cleaner since cgrp_v1_visible is really about mounted hierarchies, not the content of /proc/cgroups. Link: https://lore.kernel.org/r/b26b60b7d0d2a5ecfd2f3c45f95f32922ed24686.camel@decadent.org.uk Fixes: af000ce85293b ("cgroup: Do not report unavailable v1 controllers in /proc/cgroups") Fixes: a0ab1453226d8 ("cgroup: Print message when /proc/cgroups is read on v2-only system") Signed-off-by: Michal Koutný Signed-off-by: Tejun Heo --- Documentation/admin-guide/kernel-parameters.txt | 8 ++++++++ kernel/cgroup/cgroup-v1.c | 14 ++++++++++++-- 2 files changed, 20 insertions(+), 2 deletions(-) (limited to 'Documentation/admin-guide') diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt index a3ea40b22fb9..60fd3f84272b 100644 --- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -630,6 +630,14 @@ named mounts. Specifying both "all" and "named" disables all v1 hierarchies. + cgroup_v1_proc= [KNL] Show also missing controllers in /proc/cgroups + Format: { "true" | "false" } + /proc/cgroups lists only v1 controllers by default. + This compatibility option enables listing also v2 + controllers (whose v1 code is not compiled!), so that + semi-legacy software can check this file to decide + about usage of v2 (sic) controllers. + cgroup_favordynmods= [KNL] Enable or Disable favordynmods. Format: { "true" | "false" } Defaults to the value of CONFIG_CGROUP_FAVOR_DYNMODS. diff --git a/kernel/cgroup/cgroup-v1.c b/kernel/cgroup/cgroup-v1.c index fa24c032ed6f..2a4a387f867a 100644 --- a/kernel/cgroup/cgroup-v1.c +++ b/kernel/cgroup/cgroup-v1.c @@ -32,6 +32,9 @@ static u16 cgroup_no_v1_mask; /* disable named v1 mounts */ static bool cgroup_no_v1_named; +/* Show unavailable controllers in /proc/cgroups */ +static bool proc_show_all; + /* * pidlist destructions need to be flushed on cgroup destruction. Use a * separate workqueue as flush domain. @@ -683,10 +686,11 @@ int proc_cgroupstats_show(struct seq_file *m, void *v) */ for_each_subsys(ss, i) { - if (cgroup1_subsys_absent(ss)) - continue; cgrp_v1_visible |= ss->root != &cgrp_dfl_root; + if (!proc_show_all && cgroup1_subsys_absent(ss)) + continue; + seq_printf(m, "%s\t%d\t%d\t%d\n", ss->legacy_name, ss->root->hierarchy_id, atomic_read(&ss->root->nr_cgrps), @@ -1359,3 +1363,9 @@ static int __init cgroup_no_v1(char *str) return 1; } __setup("cgroup_no_v1=", cgroup_no_v1); + +static int __init cgroup_v1_proc(char *str) +{ + return (kstrtobool(str, &proc_show_all) == 0); +} +__setup("cgroup_v1_proc=", cgroup_v1_proc); -- cgit v1.2.3