From: Waiman Long <longman@xxxxxxxxxx>
commit eee87853794187f6adbe19533ed79c8b44b36a91 upstream.
In the case of CLONE_INTO_CGROUP, not all cpusets are ready to accept
new tasks. It is too late to check that in cpuset_fork(). So we need
to add the cpuset_can_fork() and cpuset_cancel_fork() methods to
pre-check it before we can allow attachment to a different cpuset.
We also need to set the attach_in_progress flag to alert other code
that a new task is going to be added to the cpuset.
Fixes: ef2c41cf38a7 ("clone3: allow spawning processes into cgroups")
Suggested-by: Michal Koutný <mkoutny@xxxxxxxx>
Signed-off-by: Waiman Long <longman@xxxxxxxxxx>
Cc: stable@xxxxxxxxxxxxxxx # v5.7+
Signed-off-by: Tejun Heo <tj@xxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
kernel/cgroup/cpuset.c | 88 ++++++++++++++++++++++++++++++++++++++++++++++---
1 file changed, 83 insertions(+), 5 deletions(-)
--- a/kernel/cgroup/cpuset.c
+++ b/kernel/cgroup/cpuset.c
@@ -2151,6 +2151,18 @@ static int fmeter_getrate(struct fmeter
static struct cpuset *cpuset_attach_old_cs;
+/*
+ * Check to see if a cpuset can accept a new task
+ * For v1, cpus_allowed and mems_allowed can't be empty.
+ */
+static int cpuset_can_attach_check(struct cpuset *cs)
+{
+ if (!is_in_v2_mode() &&
+ (cpumask_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed)))
+ return -ENOSPC;
+ return 0;
+}
+
/* Called by cgroups to determine if a cpuset is usable; cpuset_rwsem held */
static int cpuset_can_attach(struct cgroup_taskset *tset)
{
@@ -2165,10 +2177,8 @@ static int cpuset_can_attach(struct cgro
percpu_down_write(&cpuset_rwsem);
- /* allow moving tasks into an empty cpuset if on default hierarchy */
- ret = -ENOSPC;
- if (!is_in_v2_mode() &&
- (cpumask_empty(cs->cpus_allowed) || nodes_empty(cs->mems_allowed)))
+ ret = cpuset_can_attach_check(cs);
+ if (ret)
goto out_unlock;
cgroup_taskset_for_each(task, css, tset) {
@@ -2185,7 +2195,6 @@ static int cpuset_can_attach(struct cgro
* changes which zero cpus/mems_allowed.
*/
cs->attach_in_progress++;
- ret = 0;
out_unlock:
percpu_up_write(&cpuset_rwsem);
return ret;
@@ -2913,6 +2922,68 @@ static void cpuset_bind(struct cgroup_su
}
/*
+ * In case the child is cloned into a cpuset different from its parent,
+ * additional checks are done to see if the move is allowed.
+ */
+static int cpuset_can_fork(struct task_struct *task, struct css_set *cset)
+{
+ struct cpuset *cs = css_cs(cset->subsys[cpuset_cgrp_id]);
+ bool same_cs;
+ int ret;
+
+ rcu_read_lock();
+ same_cs = (cs == task_cs(current));
+ rcu_read_unlock();
+
+ if (same_cs)
+ return 0;
+
+ lockdep_assert_held(&cgroup_mutex);
+ percpu_down_write(&cpuset_rwsem);
+
+ /* Check to see if task is allowed in the cpuset */
+ ret = cpuset_can_attach_check(cs);
+ if (ret)
+ goto out_unlock;
+
+ ret = task_can_attach(task, cs->effective_cpus);
+ if (ret)
+ goto out_unlock;
+
+ ret = security_task_setscheduler(task);
+ if (ret)
+ goto out_unlock;
+
+ /*
+ * Mark attach is in progress. This makes validate_change() fail
+ * changes which zero cpus/mems_allowed.
+ */
+ cs->attach_in_progress++;
+out_unlock:
+ percpu_up_write(&cpuset_rwsem);
+ return ret;
+}
+
+static void cpuset_cancel_fork(struct task_struct *task, struct css_set *cset)
+{
+ struct cpuset *cs = css_cs(cset->subsys[cpuset_cgrp_id]);
+ bool same_cs;
+
+ rcu_read_lock();
+ same_cs = (cs == task_cs(current));
+ rcu_read_unlock();
+
+ if (same_cs)
+ return;
+
+ percpu_down_write(&cpuset_rwsem);
+ cs->attach_in_progress--;
+ if (!cs->attach_in_progress)
+ wake_up(&cpuset_attach_wq);
+ percpu_up_write(&cpuset_rwsem);
+}
+
+/*
* Make sure the new task conform to the current state of its parent,
* which could have been changed by cpuset just after it inherits the
* state from the parent and before it sits on the cgroup's task list.
@@ -2946,6 +3017,11 @@ static void cpuset_fork(struct task_stru
else
guarantee_online_cpus(cs, cpus_attach);
cpuset_attach_task(cs, task);
+
+ cs->attach_in_progress--;
+ if (!cs->attach_in_progress)
+ wake_up(&cpuset_attach_wq);
+
percpu_up_write(&cpuset_rwsem);
}
@@ -2959,6 +3035,8 @@ struct cgroup_subsys cpuset_cgrp_subsys
.attach = cpuset_attach,
.post_attach = cpuset_post_attach,
.bind = cpuset_bind,
+ .can_fork = cpuset_can_fork,
+ .cancel_fork = cpuset_cancel_fork,
.fork = cpuset_fork,
.legacy_cftypes = legacy_files,
.dfl_cftypes = dfl_files,
