Merge tag 'cgroup-for-5.20' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup

Pull cgroup updates from Tejun Heo:
 "Several core optimizations:

   - threadgroup_rwsem write locking is skipped when configuring
     controllers in empty subtrees.

     Combined with CLONE_INTO_CGROUP, this allows the common static
     usage pattern to not grab threadgroup_rwsem at all (glibc still
     doesn't seem ready for CLONE_INTO_CGROUP unfortunately).

   - threadgroup_rwsem used to be put into non-percpu mode by default
     due to latency concerns in specific use cases. There's no reason
     for everyone else to pay for it. Make the behavior optional.

   - psi no longer allocates memory when disabled.

  ... along with some code cleanups"

* tag 'cgroup-for-5.20' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
  cgroup: Skip subtree root in cgroup_update_dfl_csses()
  cgroup: remove "no" prefixed mount options
  cgroup: Make !percpu threadgroup_rwsem operations optional
  cgroup: Add "no" prefixed mount options
  cgroup: Elide write-locking threadgroup_rwsem when updating csses on an empty subtree
  cgroup.c: remove redundant check for mixable cgroup in cgroup_migrate_vet_dst
  cgroup.c: add helper __cset_cgroup_from_root to cleanup duplicated codes
  psi: dont alloc memory for psi by default

Author: torvalds

Documentation/admin-guide/cgroup-v2.rst

@@ -184,6 +184,14 @@ cgroup v2 currently supports the following mount options.
 	ignored on non-init namespace mounts.  Please refer to the
 	Delegation section for details.
 
+  favordynmods
+        Reduce the latencies of dynamic cgroup modifications such as
+        task migrations and controller on/offs at the cost of making
+        hot path operations such as forks and exits more expensive.
+        The static usage pattern of creating a cgroup, enabling
+        controllers, and then seeding it with CLONE_INTO_CGROUP is
+        not affected by this option.
+
   memory_localevents
         Only populate memory.events with data for the current cgroup,
         and not any subtrees. This is legacy behaviour, the default

include/linux/cgroup-defs.h

@@ -88,20 +88,33 @@ enum {
 	 */
 	CGRP_ROOT_NS_DELEGATE	= (1 << 3),
 
+	/*
+	 * Reduce latencies on dynamic cgroup modifications such as task
+	 * migrations and controller on/offs by disabling percpu operation on
+	 * cgroup_threadgroup_rwsem. This makes hot path operations such as
+	 * forks and exits into the slow path and more expensive.
+	 *
+	 * The static usage pattern of creating a cgroup, enabling controllers,
+	 * and then seeding it with CLONE_INTO_CGROUP doesn't require write
+	 * locking cgroup_threadgroup_rwsem and thus doesn't benefit from
+	 * favordynmod.
+	 */
+	CGRP_ROOT_FAVOR_DYNMODS = (1 << 4),
+
 	/*
 	 * Enable cpuset controller in v1 cgroup to use v2 behavior.
 	 */
-	CGRP_ROOT_CPUSET_V2_MODE = (1 << 4),
+	CGRP_ROOT_CPUSET_V2_MODE = (1 << 16),
 
 	/*
 	 * Enable legacy local memory.events.
 	 */
-	CGRP_ROOT_MEMORY_LOCAL_EVENTS = (1 << 5),
+	CGRP_ROOT_MEMORY_LOCAL_EVENTS = (1 << 17),
 
 	/*
 	 * Enable recursive subtree protection
 	 */
-	CGRP_ROOT_MEMORY_RECURSIVE_PROT = (1 << 6),
+	CGRP_ROOT_MEMORY_RECURSIVE_PROT = (1 << 18),
 };
 
 /* cftype->flags */
@@ -480,7 +493,7 @@ struct cgroup {
 	struct work_struct release_agent_work;
 
 	/* used to track pressure stalls */
-	struct psi_group psi;
+	struct psi_group *psi;
 
 	/* used to store eBPF programs */
 	struct cgroup_bpf bpf;

include/linux/cgroup.h

@@ -674,7 +674,7 @@ static inline void pr_cont_cgroup_path(struct cgroup *cgrp)
 
 static inline struct psi_group *cgroup_psi(struct cgroup *cgrp)
 {
-	return &cgrp->psi;
+	return cgrp->psi;
 }
 
 bool cgroup_psi_enabled(void);

init/Kconfig

@@ -945,6 +945,16 @@ if CGROUPS
 config PAGE_COUNTER
 	bool
 
+config CGROUP_FAVOR_DYNMODS
+        bool "Favor dynamic modification latency reduction by default"
+        help
+          This option enables the "favordynmods" mount option by default
+          which reduces the latencies of dynamic cgroup modifications such
+          as task migrations and controller on/offs at the cost of making
+          hot path operations such as forks and exits more expensive.
+
+          Say N if unsure.
+
 config MEMCG
 	bool "Memory controller"
 	select PAGE_COUNTER

kernel/cgroup/cgroup-internal.h

@@ -233,6 +233,7 @@ void cgroup_kn_unlock(struct kernfs_node *kn);
 int cgroup_path_ns_locked(struct cgroup *cgrp, char *buf, size_t buflen,
 			  struct cgroup_namespace *ns);
 
+void cgroup_favor_dynmods(struct cgroup_root *root, bool favor);
 void cgroup_free_root(struct cgroup_root *root);
 void init_cgroup_root(struct cgroup_fs_context *ctx);
 int cgroup_setup_root(struct cgroup_root *root, u16 ss_mask);

kernel/cgroup/cgroup-v1.c

@@ -875,6 +875,8 @@ static int cgroup1_show_options(struct seq_file *seq, struct kernfs_root *kf_roo
 		seq_puts(seq, ",xattr");
 	if (root->flags & CGRP_ROOT_CPUSET_V2_MODE)
 		seq_puts(seq, ",cpuset_v2_mode");
+	if (root->flags & CGRP_ROOT_FAVOR_DYNMODS)
+		seq_puts(seq, ",favordynmods");
 
 	spin_lock(&release_agent_path_lock);
 	if (strlen(root->release_agent_path))
@@ -898,6 +900,8 @@ enum cgroup1_param {
 	Opt_noprefix,
 	Opt_release_agent,
 	Opt_xattr,
+	Opt_favordynmods,
+	Opt_nofavordynmods,
 };
 
 const struct fs_parameter_spec cgroup1_fs_parameters[] = {
@@ -909,6 +913,8 @@ const struct fs_parameter_spec cgroup1_fs_parameters[] = {
 	fsparam_flag  ("noprefix",	Opt_noprefix),
 	fsparam_string("release_agent",	Opt_release_agent),
 	fsparam_flag  ("xattr",		Opt_xattr),
+	fsparam_flag  ("favordynmods",	Opt_favordynmods),
+	fsparam_flag  ("nofavordynmods", Opt_nofavordynmods),
 	{}
 };
 
@@ -960,6 +966,12 @@ int cgroup1_parse_param(struct fs_context *fc, struct fs_parameter *param)
 	case Opt_xattr:
 		ctx->flags |= CGRP_ROOT_XATTR;
 		break;
+	case Opt_favordynmods:
+		ctx->flags |= CGRP_ROOT_FAVOR_DYNMODS;
+		break;
+	case Opt_nofavordynmods:
+		ctx->flags &= ~CGRP_ROOT_FAVOR_DYNMODS;
+		break;
 	case Opt_release_agent:
 		/* Specifying two release agents is forbidden */
 		if (ctx->release_agent)
@@ -1211,8 +1223,11 @@ static int cgroup1_root_to_use(struct fs_context *fc)
 	init_cgroup_root(ctx);
 
 	ret = cgroup_setup_root(root, ctx->subsys_mask);
-	if (ret)
+	if (!ret)
+		cgroup_favor_dynmods(root, ctx->flags & CGRP_ROOT_FAVOR_DYNMODS);
+	else
 		cgroup_free_root(root);
+
 	return ret;
 }