[PATCH 5.15 1/3] cgroup/cpuset: Skip spread flags update on v2

[PATCH 5.15 1/3] cgroup/cpuset: Skip spread flags update on v2

[Waiman Long]

commit 18f9a4d47527772515ad6cbdac796422566e6440 upstream.

Cpuset v2 has no spread flags to set. So we can skip spread
flags update if cpuset v2 is being used. Also change the name to
cpuset_update_task_spread_flags() to indicate that there are multiple
spread flags.

Signed-off-by: Waiman Long <longman@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
---
 kernel/cgroup/cpuset.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
index 428820bf141d..a7bbe277ae66 100644
--- a/kernel/cgroup/cpuset.c
+++ b/kernel/cgroup/cpuset.c
@@ -450,11 +450,15 @@ static void guarantee_online_mems(struct cpuset *cs, nodemask_t *pmask)
 /*
  * update task's spread flag if cpuset's page/slab spread flag is set
  *
- * Call with callback_lock or cpuset_rwsem held.
+ * Call with callback_lock or cpuset_rwsem held. The check can be skipped
+ * if on default hierarchy.
  */
-static void cpuset_update_task_spread_flag(struct cpuset *cs,
+static void cpuset_update_task_spread_flags(struct cpuset *cs,
 					struct task_struct *tsk)
 {
+	if (cgroup_subsys_on_dfl(cpuset_cgrp_subsys))
+		return;
+
 	if (is_spread_page(cs))
 		task_set_spread_page(tsk);
 	else
@@ -1941,7 +1945,7 @@ static void update_tasks_flags(struct cpuset *cs)
 
 	css_task_iter_start(&cs->css, 0, &it);
 	while ((task = css_task_iter_next(&it)))
-		cpuset_update_task_spread_flag(cs, task);
+		cpuset_update_task_spread_flags(cs, task);
 	css_task_iter_end(&it);
 }
 
@@ -2270,7 +2274,7 @@ static void cpuset_attach(struct cgroup_taskset *tset)
 		WARN_ON_ONCE(set_cpus_allowed_ptr(task, cpus_attach));
 
 		cpuset_change_task_nodemask(task, &cpuset_attach_nodemask_to);
-		cpuset_update_task_spread_flag(cs, task);
+		cpuset_update_task_spread_flags(cs, task);
 	}
 
 	/*
-- 
2.31.1

[PATCH 5.15 2/3] cgroup/cpuset: Make cpuset_fork() handle CLONE_INTO_CGROUP properly

[Waiman Long]

commit 42a11bf5c5436e91b040aeb04063be1710bb9f9c upstream.

By default, the clone(2) syscall spawn a child process into the same
cgroup as its parent. With the use of the CLONE_INTO_CGROUP flag
introduced by commit ef2c41cf38a7 ("clone3: allow spawning processes
into cgroups"), the child will be spawned into a different cgroup which
is somewhat similar to writing the child's tid into "cgroup.threads".

The current cpuset_fork() method does not properly handle the
CLONE_INTO_CGROUP case where the cpuset of the child may be different
from that of its parent.  Update the cpuset_fork() method to treat the
CLONE_INTO_CGROUP case similar to cpuset_attach().

Since the newly cloned task has not been running yet, its actual
memory usage isn't known. So it is not necessary to make change to mm
in cpuset_fork().

Fixes: ef2c41cf38a7 ("clone3: allow spawning processes into cgroups")
Reported-by: Giuseppe Scrivano <gscrivan@redhat.com>
Signed-off-by: Waiman Long <longman@redhat.com>
Cc: stable@vger.kernel.org # v5.7+
Signed-off-by: Tejun Heo <tj@kernel.org>
---
 kernel/cgroup/cpuset.c | 62 ++++++++++++++++++++++++++++--------------
 1 file changed, 42 insertions(+), 20 deletions(-)

diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
index a7bbe277ae66..3a7690450a3b 100644
--- a/kernel/cgroup/cpuset.c
+++ b/kernel/cgroup/cpuset.c
@@ -2238,16 +2238,33 @@ static void cpuset_cancel_attach(struct cgroup_taskset *tset)
 }
 
 /*
- * Protected by cpuset_rwsem.  cpus_attach is used only by cpuset_attach()
+ * Protected by cpuset_rwsem. cpus_attach is used only by cpuset_attach_task()
  * but we can't allocate it dynamically there.  Define it global and
  * allocate from cpuset_init().
  */
 static cpumask_var_t cpus_attach;
+static nodemask_t cpuset_attach_nodemask_to;
+
+static void cpuset_attach_task(struct cpuset *cs, struct task_struct *task)
+{
+	percpu_rwsem_assert_held(&cpuset_rwsem);
+
+	if (cs != &top_cpuset)
+		guarantee_online_cpus(task, cpus_attach);
+	else
+		cpumask_copy(cpus_attach, task_cpu_possible_mask(task));
+	/*
+	 * can_attach beforehand should guarantee that this doesn't
+	 * fail.  TODO: have a better way to handle failure here
+	 */
+	WARN_ON_ONCE(set_cpus_allowed_ptr(task, cpus_attach));
+
+	cpuset_change_task_nodemask(task, &cpuset_attach_nodemask_to);
+	cpuset_update_task_spread_flags(cs, task);
+}
 
 static void cpuset_attach(struct cgroup_taskset *tset)
 {
-	/* static buf protected by cpuset_rwsem */
-	static nodemask_t cpuset_attach_nodemask_to;
 	struct task_struct *task;
 	struct task_struct *leader;
 	struct cgroup_subsys_state *css;
@@ -2262,20 +2279,8 @@ static void cpuset_attach(struct cgroup_taskset *tset)
 
 	guarantee_online_mems(cs, &cpuset_attach_nodemask_to);
 
-	cgroup_taskset_for_each(task, css, tset) {
-		if (cs != &top_cpuset)
-			guarantee_online_cpus(task, cpus_attach);
-		else
-			cpumask_copy(cpus_attach, task_cpu_possible_mask(task));
-		/*
-		 * can_attach beforehand should guarantee that this doesn't
-		 * fail.  TODO: have a better way to handle failure here
-		 */
-		WARN_ON_ONCE(set_cpus_allowed_ptr(task, cpus_attach));
-
-		cpuset_change_task_nodemask(task, &cpuset_attach_nodemask_to);
-		cpuset_update_task_spread_flags(cs, task);
-	}
+	cgroup_taskset_for_each(task, css, tset)
+		cpuset_attach_task(cs, task);
 
 	/*
 	 * Change mm for all threadgroup leaders. This is expensive and may
@@ -2948,11 +2953,28 @@ static void cpuset_bind(struct cgroup_subsys_state *root_css)
  */
 static void cpuset_fork(struct task_struct *task)
 {
-	if (task_css_is_root(task, cpuset_cgrp_id))
+	struct cpuset *cs;
+	bool same_cs;
+
+	rcu_read_lock();
+	cs = task_cs(task);
+	same_cs = (cs == task_cs(current));
+	rcu_read_unlock();
+
+	if (same_cs) {
+		if (cs == &top_cpuset)
+			return;
+
+		set_cpus_allowed_ptr(task, current->cpus_ptr);
+		task->mems_allowed = current->mems_allowed;
 		return;
+	}
 
-	set_cpus_allowed_ptr(task, current->cpus_ptr);
-	task->mems_allowed = current->mems_allowed;
+	/* CLONE_INTO_CGROUP */
+	percpu_down_write(&cpuset_rwsem);
+	guarantee_online_mems(cs, &cpuset_attach_nodemask_to);
+	cpuset_attach_task(cs, task);
+	percpu_up_write(&cpuset_rwsem);
 }
 
 struct cgroup_subsys cpuset_cgrp_subsys = {
-- 
2.31.1

[PATCH 5.15 3/3] cgroup/cpuset: Add cpuset_can_fork() and cpuset_cancel_fork() methods

[Waiman Long]

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@suse.com>
Signed-off-by: Waiman Long <longman@redhat.com>
Cc: stable@vger.kernel.org # v5.7+
Signed-off-by: Tejun Heo <tj@kernel.org>
---
 kernel/cgroup/cpuset.c | 88 +++++++++++++++++++++++++++++++++++++++---
 1 file changed, 83 insertions(+), 5 deletions(-)

diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
index 3a7690450a3b..58a515b4cbc0 100644
--- a/kernel/cgroup/cpuset.c
+++ b/kernel/cgroup/cpuset.c
@@ -2186,6 +2186,18 @@ static int fmeter_getrate(struct fmeter *fmp)
 
 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)
 {
@@ -2200,10 +2212,8 @@ static int cpuset_can_attach(struct cgroup_taskset *tset)
 
 	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) {
@@ -2220,7 +2230,6 @@ static int cpuset_can_attach(struct cgroup_taskset *tset)
 	 * changes which zero cpus/mems_allowed.
 	 */
 	cs->attach_in_progress++;
-	ret = 0;
 out_unlock:
 	percpu_up_write(&cpuset_rwsem);
 	return ret;
@@ -2946,6 +2955,68 @@ static void cpuset_bind(struct cgroup_subsys_state *root_css)
 	percpu_up_write(&cpuset_rwsem);
 }
 
+/*
+ * 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
@@ -2974,6 +3045,11 @@ static void cpuset_fork(struct task_struct *task)
 	percpu_down_write(&cpuset_rwsem);
 	guarantee_online_mems(cs, &cpuset_attach_nodemask_to);
 	cpuset_attach_task(cs, task);
+
+	cs->attach_in_progress--;
+	if (!cs->attach_in_progress)
+		wake_up(&cpuset_attach_wq);
+
 	percpu_up_write(&cpuset_rwsem);
 }
 
@@ -2987,6 +3063,8 @@ 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,
-- 
2.31.1

Re: [PATCH 5.15 1/3] cgroup/cpuset: Skip spread flags update on v2

[Greg Kroah-Hartman]

On Mon, Apr 17, 2023 at 11:36:45AM -0400, Waiman Long wrote:
> commit 18f9a4d47527772515ad6cbdac796422566e6440 upstream.
> 
> Cpuset v2 has no spread flags to set. So we can skip spread
> flags update if cpuset v2 is being used. Also change the name to
> cpuset_update_task_spread_flags() to indicate that there are multiple
> spread flags.
> 
> Signed-off-by: Waiman Long <longman@redhat.com>
> Signed-off-by: Tejun Heo <tj@kernel.org>
> ---
>  kernel/cgroup/cpuset.c | 12 ++++++++----
>  1 file changed, 8 insertions(+), 4 deletions(-)
> 

Same as the 5.10.y series, I've dropped all of these due to build
errors.

thanks,

greg k-h