[RFC PATCH bpf-next 02/12] bpf: Make struct_ops tasks_rcu grace period optional

[Martin KaFai Lau <martin.lau@linux.dev>]

From: Martin KaFai Lau <martin.lau@kernel.org>

bpf_struct_ops_map_free() currently waits for both a regular RCU grace
period and a tasks RCU grace period for every struct_ops map through
synchronize_rcu_mult(call_rcu, call_rcu_tasks).

A regular RCU grace period is still required for all struct_ops maps
because the struct_ops trampoline ksyms requires a rcu grace period
(take a look at the list_del_rcu in __bpf_ksym_del).
Add a map_free_pre_rcu() callback so the struct_ops map can remove
ksyms before bpf_map_put() wait for the regular rcu grace period.

The tasks RCU grace period is only needed by tcp_congestion_ops.
Add free_after_tasks_rcu_gp only to struct bpf_struct_ops instead
of the bpf_map.

When CONFIG_TASKS_RCU=n, synchronize_rcu_tasks() is the same as
synchronize_rcu(). Since all struct_ops maps now complete a regular RCU
grace period before bpf_struct_ops_map_free() runs, skip the extra
synchronize_rcu_tasks() call in this case.

This cleanup prepares for a later patch that needs to support
free_after_mult_rcu_gp.

Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
---
 include/linux/bpf.h         |  7 +++++++
 kernel/bpf/bpf_struct_ops.c | 31 +++++++++++++------------------
 kernel/bpf/syscall.c        |  3 +++
 net/ipv4/bpf_tcp_ca.c       | 16 ++++++++++++++++
 4 files changed, 39 insertions(+), 18 deletions(-)

diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index 1b28cacc3075..a276eada19c4 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -86,6 +86,7 @@ struct bpf_map_ops {
 	struct bpf_map *(*map_alloc)(union bpf_attr *attr);
 	void (*map_release)(struct bpf_map *map, struct file *map_file);
 	void (*map_free)(struct bpf_map *map);
+	void (*map_free_pre_rcu)(struct bpf_map *map);
 	int (*map_get_next_key)(struct bpf_map *map, void *key, void *next_key);
 	void (*map_release_uref)(struct bpf_map *map);
 	void *(*map_lookup_elem_sys_only)(struct bpf_map *map, void *key);
@@ -1992,6 +1993,11 @@ struct btf_member;
  *	   unloaded while in use.
  * @name: The name of the struct bpf_struct_ops object.
  * @func_models: Func models
+ * @free_after_tasks_rcu_gp: Set to true if it needs the bpf core to wait for
+ *                           a tasks_rcu gp before freeing the struct_ops map
+ *                           and its progs. It is unnecessary if the @unreg
+ *                           has waited for the correct rcu gp or the @unreg
+ *                           has ensured all struct_ops prog has finished running.
  */
 struct bpf_struct_ops {
 	const struct bpf_verifier_ops *verifier_ops;
@@ -2010,6 +2016,7 @@ struct bpf_struct_ops {
 	struct module *owner;
 	const char *name;
 	struct btf_func_model func_models[BPF_STRUCT_OPS_MAX_NR_MEMBERS];
+	bool free_after_tasks_rcu_gp;
 };
 
 /* Every member of a struct_ops type has an instance even a member is not
diff --git a/kernel/bpf/bpf_struct_ops.c b/kernel/bpf/bpf_struct_ops.c
index 08791180d71d..28eab24ef0ed 100644
--- a/kernel/bpf/bpf_struct_ops.c
+++ b/kernel/bpf/bpf_struct_ops.c
@@ -983,9 +983,18 @@ static void __bpf_struct_ops_map_free(struct bpf_map *map)
 	bpf_map_area_free(st_map);
 }
 
+static void bpf_struct_ops_map_free_pre_rcu(struct bpf_map *map)
+{
+	struct bpf_struct_ops_map *st_map = (struct bpf_struct_ops_map *)map;
+
+	bpf_struct_ops_map_del_ksyms(st_map);
+}
+
 static void bpf_struct_ops_map_free(struct bpf_map *map)
 {
 	struct bpf_struct_ops_map *st_map = (struct bpf_struct_ops_map *)map;
+	struct bpf_struct_ops *st_ops = st_map->st_ops_desc->st_ops;
+	bool tasks_rcu = st_ops->free_after_tasks_rcu_gp;
 
 	/* st_ops->owner was acquired during map_alloc to implicitly holds
 	 * the btf's refcnt. The acquire was only done when btf_is_module()
@@ -996,24 +1005,8 @@ static void bpf_struct_ops_map_free(struct bpf_map *map)
 
 	bpf_struct_ops_map_dissoc_progs(st_map);
 
-	bpf_struct_ops_map_del_ksyms(st_map);
-
-	/* The struct_ops's function may switch to another struct_ops.
-	 *
-	 * For example, bpf_tcp_cc_x->init() may switch to
-	 * another tcp_cc_y by calling
-	 * setsockopt(TCP_CONGESTION, "tcp_cc_y").
-	 * During the switch,  bpf_struct_ops_put(tcp_cc_x) is called
-	 * and its refcount may reach 0 which then free its
-	 * trampoline image while tcp_cc_x is still running.
-	 *
-	 * A vanilla rcu gp is to wait for all bpf-tcp-cc prog
-	 * to finish. bpf-tcp-cc prog is non sleepable.
-	 * A rcu_tasks gp is to wait for the last few insn
-	 * in the tramopline image to finish before releasing
-	 * the trampoline image.
-	 */
-	synchronize_rcu_mult(call_rcu, call_rcu_tasks);
+	if (tasks_rcu && IS_ENABLED(CONFIG_TASKS_RCU))
+		synchronize_rcu_tasks();
 
 	__bpf_struct_ops_map_free(map);
 }
@@ -1122,6 +1115,7 @@ static struct bpf_map *bpf_struct_ops_map_alloc(union bpf_attr *attr)
 
 	mutex_init(&st_map->lock);
 	bpf_map_init_from_attr(map, attr);
+	map->free_after_rcu_gp = true;
 
 	return map;
 
@@ -1154,6 +1148,7 @@ const struct bpf_map_ops bpf_struct_ops_map_ops = {
 	.map_alloc_check = bpf_struct_ops_map_alloc_check,
 	.map_alloc = bpf_struct_ops_map_alloc,
 	.map_free = bpf_struct_ops_map_free,
+	.map_free_pre_rcu = bpf_struct_ops_map_free_pre_rcu,
 	.map_get_next_key = bpf_struct_ops_map_get_next_key,
 	.map_lookup_elem = bpf_struct_ops_map_lookup_elem,
 	.map_delete_elem = bpf_struct_ops_map_delete_elem,
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 6600e126fbfb..d0e8e9c8c888 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -950,6 +950,9 @@ void bpf_map_put(struct bpf_map *map)
 		/* bpf_map_free_id() must be called first */
 		bpf_map_free_id(map);
 
+		if (map->ops->map_free_pre_rcu)
+			map->ops->map_free_pre_rcu(map);
+
 		WARN_ON_ONCE(atomic64_read(&map->sleepable_refcnt));
 		/* RCU tasks trace grace period implies RCU grace period. */
 		if (READ_ONCE(map->free_after_mult_rcu_gp))
diff --git a/net/ipv4/bpf_tcp_ca.c b/net/ipv4/bpf_tcp_ca.c
index 791e15063237..e224ecafbd69 100644
--- a/net/ipv4/bpf_tcp_ca.c
+++ b/net/ipv4/bpf_tcp_ca.c
@@ -339,6 +339,22 @@ static struct bpf_struct_ops bpf_tcp_congestion_ops = {
 	.validate = bpf_tcp_ca_validate,
 	.name = "tcp_congestion_ops",
 	.cfi_stubs = &__bpf_ops_tcp_congestion_ops,
+	/* The struct_ops's function may switch to another struct_ops.
+	 *
+	 * For example, bpf_tcp_cc_x->init() may switch to
+	 * another tcp_cc_y by calling
+	 * setsockopt(TCP_CONGESTION, "tcp_cc_y").
+	 * During the switch,  bpf_struct_ops_put(tcp_cc_x) is called
+	 * and its refcount may reach 0 which then free its
+	 * trampoline image while tcp_cc_x is still running.
+	 *
+	 * A vanilla rcu gp is to wait for all bpf-tcp-cc prog
+	 * to finish. bpf-tcp-cc prog is non sleepable.
+	 * A rcu_tasks gp is to wait for the last few insn
+	 * in the tramopline image to finish before releasing
+	 * the trampoline image.
+	 */
+	.free_after_tasks_rcu_gp = true,
 	.owner = THIS_MODULE,
 };
 
-- 
2.53.0-Meta