Merge branch 'bpf-per-cpu-cgroup-storage'

Roman Gushchin says:

====================
This patchset implements per-cpu cgroup local storage and provides
an example how per-cpu and shared cgroup local storage can be used
for efficient accounting of network traffic.

v4->v3:
  1) incorporated Alexei's feedback

v3->v2:
  1) incorporated Song's feedback
  2) rebased on top of current bpf-next

v2->v1:
  1) added a selftest implementing network counters
  2) added a missing free() in cgroup local storage selftest
====================
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

include/linux/bpf-cgroup.h

@@ -2,6 +2,7 @@
 #ifndef _BPF_CGROUP_H
 #define _BPF_CGROUP_H
 
+#include <linux/bpf.h>
 #include <linux/errno.h>
 #include <linux/jump_label.h>
 #include <linux/percpu.h>
@@ -22,7 +23,11 @@ struct bpf_cgroup_storage;
 extern struct static_key_false cgroup_bpf_enabled_key;
 #define cgroup_bpf_enabled static_branch_unlikely(&cgroup_bpf_enabled_key)
 
-DECLARE_PER_CPU(void*, bpf_cgroup_storage);
+DECLARE_PER_CPU(struct bpf_cgroup_storage*,
+		bpf_cgroup_storage[MAX_BPF_CGROUP_STORAGE_TYPE]);
+
+#define for_each_cgroup_storage_type(stype) \
+	for (stype = 0; stype < MAX_BPF_CGROUP_STORAGE_TYPE; stype++)
 
 struct bpf_cgroup_storage_map;
 
@@ -32,7 +37,10 @@ struct bpf_storage_buffer {
 };
 
 struct bpf_cgroup_storage {
+	union {
 		struct bpf_storage_buffer *buf;
+		void __percpu *percpu_buf;
+	};
 	struct bpf_cgroup_storage_map *map;
 	struct bpf_cgroup_storage_key key;
 	struct list_head list;
@@ -43,7 +51,7 @@ struct bpf_cgroup_storage {
 struct bpf_prog_list {
 	struct list_head node;
 	struct bpf_prog *prog;
-	struct bpf_cgroup_storage *storage;
+	struct bpf_cgroup_storage *storage[MAX_BPF_CGROUP_STORAGE_TYPE];
 };
 
 struct bpf_prog_array;
@@ -101,18 +109,26 @@ int __cgroup_bpf_run_filter_sock_ops(struct sock *sk,
 int __cgroup_bpf_check_dev_permission(short dev_type, u32 major, u32 minor,
 				      short access, enum bpf_attach_type type);
 
-static inline void bpf_cgroup_storage_set(struct bpf_cgroup_storage *storage)
+static inline enum bpf_cgroup_storage_type cgroup_storage_type(
+	struct bpf_map *map)
 {
-	struct bpf_storage_buffer *buf;
+	if (map->map_type == BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE)
+		return BPF_CGROUP_STORAGE_PERCPU;
 
-	if (!storage)
-		return;
+	return BPF_CGROUP_STORAGE_SHARED;
+}
+
+static inline void bpf_cgroup_storage_set(struct bpf_cgroup_storage
+					  *storage[MAX_BPF_CGROUP_STORAGE_TYPE])
+{
+	enum bpf_cgroup_storage_type stype;
 
-	buf = READ_ONCE(storage->buf);
-	this_cpu_write(bpf_cgroup_storage, &buf->data[0]);
+	for_each_cgroup_storage_type(stype)
+		this_cpu_write(bpf_cgroup_storage[stype], storage[stype]);
 }
 
-struct bpf_cgroup_storage *bpf_cgroup_storage_alloc(struct bpf_prog *prog);
+struct bpf_cgroup_storage *bpf_cgroup_storage_alloc(struct bpf_prog *prog,
+					enum bpf_cgroup_storage_type stype);
 void bpf_cgroup_storage_free(struct bpf_cgroup_storage *storage);
 void bpf_cgroup_storage_link(struct bpf_cgroup_storage *storage,
 			     struct cgroup *cgroup,
@@ -121,6 +137,10 @@ void bpf_cgroup_storage_unlink(struct bpf_cgroup_storage *storage);
 int bpf_cgroup_storage_assign(struct bpf_prog *prog, struct bpf_map *map);
 void bpf_cgroup_storage_release(struct bpf_prog *prog, struct bpf_map *map);
 
+int bpf_percpu_cgroup_storage_copy(struct bpf_map *map, void *key, void *value);
+int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
+				     void *value, u64 flags);
+
 /* Wrappers for __cgroup_bpf_run_filter_skb() guarded by cgroup_bpf_enabled. */
 #define BPF_CGROUP_RUN_PROG_INET_INGRESS(sk, skb)			      \
 ({									      \
@@ -265,15 +285,24 @@ static inline int cgroup_bpf_prog_query(const union bpf_attr *attr,
 	return -EINVAL;
 }
 
-static inline void bpf_cgroup_storage_set(struct bpf_cgroup_storage *storage) {}
+static inline void bpf_cgroup_storage_set(
+	struct bpf_cgroup_storage *storage[MAX_BPF_CGROUP_STORAGE_TYPE]) {}
 static inline int bpf_cgroup_storage_assign(struct bpf_prog *prog,
 					    struct bpf_map *map) { return 0; }
 static inline void bpf_cgroup_storage_release(struct bpf_prog *prog,
 					      struct bpf_map *map) {}
 static inline struct bpf_cgroup_storage *bpf_cgroup_storage_alloc(
-	struct bpf_prog *prog) { return 0; }
+	struct bpf_prog *prog, enum bpf_cgroup_storage_type stype) { return 0; }
 static inline void bpf_cgroup_storage_free(
 	struct bpf_cgroup_storage *storage) {}
+static inline int bpf_percpu_cgroup_storage_copy(struct bpf_map *map, void *key,
+						 void *value) {
+	return 0;
+}
+static inline int bpf_percpu_cgroup_storage_update(struct bpf_map *map,
+					void *key, void *value, u64 flags) {
+	return 0;
+}
 
 #define cgroup_bpf_enabled (0)
 #define BPF_CGROUP_PRE_CONNECT_ENABLED(sk) (0)
@@ -293,6 +322,8 @@ static inline void bpf_cgroup_storage_free(
 #define BPF_CGROUP_RUN_PROG_SOCK_OPS(sock_ops) ({ 0; })
 #define BPF_CGROUP_RUN_PROG_DEVICE_CGROUP(type,major,minor,access) ({ 0; })
 
+#define for_each_cgroup_storage_type(stype) for (; false; )
+
 #endif /* CONFIG_CGROUP_BPF */
 
 #endif /* _BPF_CGROUP_H */

include/linux/bpf.h

@@ -272,6 +272,14 @@ struct bpf_prog_offload {
 	u32			jited_len;
 };
 
+enum bpf_cgroup_storage_type {
+	BPF_CGROUP_STORAGE_SHARED,
+	BPF_CGROUP_STORAGE_PERCPU,
+	__BPF_CGROUP_STORAGE_MAX
+};
+
+#define MAX_BPF_CGROUP_STORAGE_TYPE __BPF_CGROUP_STORAGE_MAX
+
 struct bpf_prog_aux {
 	atomic_t refcnt;
 	u32 used_map_cnt;
@@ -289,7 +297,7 @@ struct bpf_prog_aux {
 	struct bpf_prog *prog;
 	struct user_struct *user;
 	u64 load_time; /* ns since boottime */
-	struct bpf_map *cgroup_storage;
+	struct bpf_map *cgroup_storage[MAX_BPF_CGROUP_STORAGE_TYPE];
 	char name[BPF_OBJ_NAME_LEN];
 #ifdef CONFIG_SECURITY
 	void *security;
@@ -358,7 +366,7 @@ int bpf_prog_test_run_skb(struct bpf_prog *prog, const union bpf_attr *kattr,
  */
 struct bpf_prog_array_item {
 	struct bpf_prog *prog;
-	struct bpf_cgroup_storage *cgroup_storage;
+	struct bpf_cgroup_storage *cgroup_storage[MAX_BPF_CGROUP_STORAGE_TYPE];
 };
 
 struct bpf_prog_array {

include/linux/bpf_types.h

@@ -43,6 +43,7 @@ BPF_MAP_TYPE(BPF_MAP_TYPE_CGROUP_ARRAY, cgroup_array_map_ops)
 #endif
 #ifdef CONFIG_CGROUP_BPF
 BPF_MAP_TYPE(BPF_MAP_TYPE_CGROUP_STORAGE, cgroup_storage_map_ops)
+BPF_MAP_TYPE(BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE, cgroup_storage_map_ops)
 #endif
 BPF_MAP_TYPE(BPF_MAP_TYPE_HASH, htab_map_ops)
 BPF_MAP_TYPE(BPF_MAP_TYPE_PERCPU_HASH, htab_percpu_map_ops)

include/uapi/linux/bpf.h

@@ -127,6 +127,7 @@ enum bpf_map_type {
 	BPF_MAP_TYPE_SOCKHASH,
 	BPF_MAP_TYPE_CGROUP_STORAGE,
 	BPF_MAP_TYPE_REUSEPORT_SOCKARRAY,
+	BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE,
 };
 
 enum bpf_prog_type {

kernel/bpf/cgroup.c

@@ -25,6 +25,7 @@ EXPORT_SYMBOL(cgroup_bpf_enabled_key);
  */
 void cgroup_bpf_put(struct cgroup *cgrp)
 {
+	enum bpf_cgroup_storage_type stype;
 	unsigned int type;
 
 	for (type = 0; type < ARRAY_SIZE(cgrp->bpf.progs); type++) {
@@ -34,8 +35,10 @@ void cgroup_bpf_put(struct cgroup *cgrp)
 		list_for_each_entry_safe(pl, tmp, progs, node) {
 			list_del(&pl->node);
 			bpf_prog_put(pl->prog);
-			bpf_cgroup_storage_unlink(pl->storage);
-			bpf_cgroup_storage_free(pl->storage);
+			for_each_cgroup_storage_type(stype) {
+				bpf_cgroup_storage_unlink(pl->storage[stype]);
+				bpf_cgroup_storage_free(pl->storage[stype]);
+			}
 			kfree(pl);
 			static_branch_dec(&cgroup_bpf_enabled_key);
 		}
@@ -97,6 +100,7 @@ static int compute_effective_progs(struct cgroup *cgrp,
 				   enum bpf_attach_type type,
 				   struct bpf_prog_array __rcu **array)
 {
+	enum bpf_cgroup_storage_type stype;
 	struct bpf_prog_array *progs;
 	struct bpf_prog_list *pl;
 	struct cgroup *p = cgrp;
@@ -125,7 +129,9 @@ static int compute_effective_progs(struct cgroup *cgrp,
 				continue;
 
 			progs->items[cnt].prog = pl->prog;
-			progs->items[cnt].cgroup_storage = pl->storage;
+			for_each_cgroup_storage_type(stype)
+				progs->items[cnt].cgroup_storage[stype] =
+					pl->storage[stype];
 			cnt++;
 		}
 	} while ((p = cgroup_parent(p)));
@@ -232,7 +238,9 @@ int __cgroup_bpf_attach(struct cgroup *cgrp, struct bpf_prog *prog,
 {
 	struct list_head *progs = &cgrp->bpf.progs[type];
 	struct bpf_prog *old_prog = NULL;
-	struct bpf_cgroup_storage *storage, *old_storage = NULL;
+	struct bpf_cgroup_storage *storage[MAX_BPF_CGROUP_STORAGE_TYPE],
+		*old_storage[MAX_BPF_CGROUP_STORAGE_TYPE] = {NULL};
+	enum bpf_cgroup_storage_type stype;
 	struct bpf_prog_list *pl;
 	bool pl_was_allocated;
 	int err;
@@ -254,34 +262,44 @@ int __cgroup_bpf_attach(struct cgroup *cgrp, struct bpf_prog *prog,
 	if (prog_list_length(progs) >= BPF_CGROUP_MAX_PROGS)
 		return -E2BIG;
 
-	storage = bpf_cgroup_storage_alloc(prog);
-	if (IS_ERR(storage))
+	for_each_cgroup_storage_type(stype) {
+		storage[stype] = bpf_cgroup_storage_alloc(prog, stype);
+		if (IS_ERR(storage[stype])) {
+			storage[stype] = NULL;
+			for_each_cgroup_storage_type(stype)
+				bpf_cgroup_storage_free(storage[stype]);
 			return -ENOMEM;
+		}
+	}
 
 	if (flags & BPF_F_ALLOW_MULTI) {
 		list_for_each_entry(pl, progs, node) {
 			if (pl->prog == prog) {
 				/* disallow attaching the same prog twice */
-				bpf_cgroup_storage_free(storage);
+				for_each_cgroup_storage_type(stype)
+					bpf_cgroup_storage_free(storage[stype]);
 				return -EINVAL;
 			}
 		}
 
 		pl = kmalloc(sizeof(*pl), GFP_KERNEL);
 		if (!pl) {
-			bpf_cgroup_storage_free(storage);
+			for_each_cgroup_storage_type(stype)
+				bpf_cgroup_storage_free(storage[stype]);
 			return -ENOMEM;
 		}
 
 		pl_was_allocated = true;
 		pl->prog = prog;
-		pl->storage = storage;
+		for_each_cgroup_storage_type(stype)
+			pl->storage[stype] = storage[stype];
 		list_add_tail(&pl->node, progs);
 	} else {
 		if (list_empty(progs)) {
 			pl = kmalloc(sizeof(*pl), GFP_KERNEL);
 			if (!pl) {
-				bpf_cgroup_storage_free(storage);
+				for_each_cgroup_storage_type(stype)
+					bpf_cgroup_storage_free(storage[stype]);
 				return -ENOMEM;
 			}
 			pl_was_allocated = true;
@@ -289,12 +307,15 @@ int __cgroup_bpf_attach(struct cgroup *cgrp, struct bpf_prog *prog,
 		} else {
 			pl = list_first_entry(progs, typeof(*pl), node);
 			old_prog = pl->prog;
-			old_storage = pl->storage;
-			bpf_cgroup_storage_unlink(old_storage);
+			for_each_cgroup_storage_type(stype) {
+				old_storage[stype] = pl->storage[stype];
+				bpf_cgroup_storage_unlink(old_storage[stype]);
+			}
 			pl_was_allocated = false;
 		}
 		pl->prog = prog;
-		pl->storage = storage;
+		for_each_cgroup_storage_type(stype)
+			pl->storage[stype] = storage[stype];
 	}
 
 	cgrp->bpf.flags[type] = flags;
@@ -304,21 +325,27 @@ int __cgroup_bpf_attach(struct cgroup *cgrp, struct bpf_prog *prog,
 		goto cleanup;
 
 	static_branch_inc(&cgroup_bpf_enabled_key);
-	if (old_storage)
-		bpf_cgroup_storage_free(old_storage);
+	for_each_cgroup_storage_type(stype) {
+		if (!old_storage[stype])
+			continue;
+		bpf_cgroup_storage_free(old_storage[stype]);
+	}
 	if (old_prog) {
 		bpf_prog_put(old_prog);
 		static_branch_dec(&cgroup_bpf_enabled_key);
 	}
-	bpf_cgroup_storage_link(storage, cgrp, type);
+	for_each_cgroup_storage_type(stype)
+		bpf_cgroup_storage_link(storage[stype], cgrp, type);
 	return 0;
 
 cleanup:
 	/* and cleanup the prog list */
 	pl->prog = old_prog;
-	bpf_cgroup_storage_free(pl->storage);
-	pl->storage = old_storage;
-	bpf_cgroup_storage_link(old_storage, cgrp, type);
+	for_each_cgroup_storage_type(stype) {
+		bpf_cgroup_storage_free(pl->storage[stype]);
+		pl->storage[stype] = old_storage[stype];
+		bpf_cgroup_storage_link(old_storage[stype], cgrp, type);
+	}
 	if (pl_was_allocated) {
 		list_del(&pl->node);
 		kfree(pl);
@@ -339,6 +366,7 @@ int __cgroup_bpf_detach(struct cgroup *cgrp, struct bpf_prog *prog,
 			enum bpf_attach_type type, u32 unused_flags)
 {
 	struct list_head *progs = &cgrp->bpf.progs[type];
+	enum bpf_cgroup_storage_type stype;
 	u32 flags = cgrp->bpf.flags[type];
 	struct bpf_prog *old_prog = NULL;
 	struct bpf_prog_list *pl;
@@ -385,8 +413,10 @@ int __cgroup_bpf_detach(struct cgroup *cgrp, struct bpf_prog *prog,
 
 	/* now can actually delete it from this cgroup list */
 	list_del(&pl->node);
-	bpf_cgroup_storage_unlink(pl->storage);
-	bpf_cgroup_storage_free(pl->storage);
+	for_each_cgroup_storage_type(stype) {
+		bpf_cgroup_storage_unlink(pl->storage[stype]);
+		bpf_cgroup_storage_free(pl->storage[stype]);
+	}
 	kfree(pl);
 	if (list_empty(progs))
 		/* last program was detached, reset flags to zero */

kernel/bpf/helpers.c

@@ -194,16 +194,28 @@ const struct bpf_func_proto bpf_get_current_cgroup_id_proto = {
 	.ret_type	= RET_INTEGER,
 };
 
-DECLARE_PER_CPU(void*, bpf_cgroup_storage);
+#ifdef CONFIG_CGROUP_BPF
+DECLARE_PER_CPU(struct bpf_cgroup_storage*,
+		bpf_cgroup_storage[MAX_BPF_CGROUP_STORAGE_TYPE]);
 
 BPF_CALL_2(bpf_get_local_storage, struct bpf_map *, map, u64, flags)
 {
-	/* map and flags arguments are not used now,
-	 * but provide an ability to extend the API
-	 * for other types of local storages.
-	 * verifier checks that their values are correct.
+	/* flags argument is not used now,
+	 * but provides an ability to extend the API.
+	 * verifier checks that its value is correct.
 	 */
-	return (unsigned long) this_cpu_read(bpf_cgroup_storage);
+	enum bpf_cgroup_storage_type stype = cgroup_storage_type(map);
+	struct bpf_cgroup_storage *storage;
+	void *ptr;
+
+	storage = this_cpu_read(bpf_cgroup_storage[stype]);
+
+	if (stype == BPF_CGROUP_STORAGE_SHARED)
+		ptr = &READ_ONCE(storage->buf)->data[0];
+	else
+		ptr = this_cpu_ptr(storage->percpu_buf);
+
+	return (unsigned long)ptr;
 }
 
 const struct bpf_func_proto bpf_get_local_storage_proto = {
@@ -214,3 +226,4 @@ const struct bpf_func_proto bpf_get_local_storage_proto = {
 	.arg2_type	= ARG_ANYTHING,
 };
 #endif
+#endif

kernel/bpf/local_storage.c

@@ -7,7 +7,8 @@
 #include <linux/rbtree.h>
 #include <linux/slab.h>
 
-DEFINE_PER_CPU(void*, bpf_cgroup_storage);
+DEFINE_PER_CPU(struct bpf_cgroup_storage*,
+	       bpf_cgroup_storage[MAX_BPF_CGROUP_STORAGE_TYPE]);
 
 #ifdef CONFIG_CGROUP_BPF
 
@@ -151,6 +152,71 @@ static int cgroup_storage_update_elem(struct bpf_map *map, void *_key,
 	return 0;
 }
 
+int bpf_percpu_cgroup_storage_copy(struct bpf_map *_map, void *_key,
+				   void *value)
+{
+	struct bpf_cgroup_storage_map *map = map_to_storage(_map);
+	struct bpf_cgroup_storage_key *key = _key;
+	struct bpf_cgroup_storage *storage;
+	int cpu, off = 0;
+	u32 size;
+
+	rcu_read_lock();
+	storage = cgroup_storage_lookup(map, key, false);
+	if (!storage) {
+		rcu_read_unlock();
+		return -ENOENT;
+	}
+
+	/* per_cpu areas are zero-filled and bpf programs can only
+	 * access 'value_size' of them, so copying rounded areas
+	 * will not leak any kernel data
+	 */
+	size = round_up(_map->value_size, 8);
+	for_each_possible_cpu(cpu) {
+		bpf_long_memcpy(value + off,
+				per_cpu_ptr(storage->percpu_buf, cpu), size);
+		off += size;
+	}
+	rcu_read_unlock();
+	return 0;
+}
+
+int bpf_percpu_cgroup_storage_update(struct bpf_map *_map, void *_key,
+				     void *value, u64 map_flags)
+{
+	struct bpf_cgroup_storage_map *map = map_to_storage(_map);
+	struct bpf_cgroup_storage_key *key = _key;
+	struct bpf_cgroup_storage *storage;
+	int cpu, off = 0;
+	u32 size;
+
+	if (map_flags != BPF_ANY && map_flags != BPF_EXIST)
+		return -EINVAL;
+
+	rcu_read_lock();
+	storage = cgroup_storage_lookup(map, key, false);
+	if (!storage) {
+		rcu_read_unlock();
+		return -ENOENT;
+	}
+
+	/* the user space will provide round_up(value_size, 8) bytes that
+	 * will be copied into per-cpu area. bpf programs can only access
+	 * value_size of it. During lookup the same extra bytes will be
+	 * returned or zeros which were zero-filled by percpu_alloc,
+	 * so no kernel data leaks possible
+	 */
+	size = round_up(_map->value_size, 8);
+	for_each_possible_cpu(cpu) {
+		bpf_long_memcpy(per_cpu_ptr(storage->percpu_buf, cpu),
+				value + off, size);
+		off += size;
+	}
+	rcu_read_unlock();
+	return 0;
+}
+
 static int cgroup_storage_get_next_key(struct bpf_map *_map, void *_key,
 				       void *_next_key)
 {
@@ -251,6 +317,7 @@ const struct bpf_map_ops cgroup_storage_map_ops = {
 
 int bpf_cgroup_storage_assign(struct bpf_prog *prog, struct bpf_map *_map)
 {
+	enum bpf_cgroup_storage_type stype = cgroup_storage_type(_map);
 	struct bpf_cgroup_storage_map *map = map_to_storage(_map);
 	int ret = -EBUSY;
 
@@ -258,11 +325,12 @@ int bpf_cgroup_storage_assign(struct bpf_prog *prog, struct bpf_map *_map)
 
 	if (map->prog && map->prog != prog)
 		goto unlock;
-	if (prog->aux->cgroup_storage && prog->aux->cgroup_storage != _map)
+	if (prog->aux->cgroup_storage[stype] &&
+	    prog->aux->cgroup_storage[stype] != _map)
 		goto unlock;
 
 	map->prog = prog;
-	prog->aux->cgroup_storage = _map;
+	prog->aux->cgroup_storage[stype] = _map;
 	ret = 0;
 unlock:
 	spin_unlock_bh(&map->lock);
@@ -272,70 +340,117 @@ int bpf_cgroup_storage_assign(struct bpf_prog *prog, struct bpf_map *_map)
 
 void bpf_cgroup_storage_release(struct bpf_prog *prog, struct bpf_map *_map)
 {
+	enum bpf_cgroup_storage_type stype = cgroup_storage_type(_map);
 	struct bpf_cgroup_storage_map *map = map_to_storage(_map);
 
 	spin_lock_bh(&map->lock);
 	if (map->prog == prog) {
-		WARN_ON(prog->aux->cgroup_storage != _map);
+		WARN_ON(prog->aux->cgroup_storage[stype] != _map);
 		map->prog = NULL;
-		prog->aux->cgroup_storage = NULL;
+		prog->aux->cgroup_storage[stype] = NULL;
 	}
 	spin_unlock_bh(&map->lock);
 }
 
-struct bpf_cgroup_storage *bpf_cgroup_storage_alloc(struct bpf_prog *prog)
+static size_t bpf_cgroup_storage_calculate_size(struct bpf_map *map, u32 *pages)
+{
+	size_t size;
+
+	if (cgroup_storage_type(map) == BPF_CGROUP_STORAGE_SHARED) {
+		size = sizeof(struct bpf_storage_buffer) + map->value_size;
+		*pages = round_up(sizeof(struct bpf_cgroup_storage) + size,
+				  PAGE_SIZE) >> PAGE_SHIFT;
+	} else {
+		size = map->value_size;
+		*pages = round_up(round_up(size, 8) * num_possible_cpus(),
+				  PAGE_SIZE) >> PAGE_SHIFT;
+	}
+
+	return size;
+}
+
+struct bpf_cgroup_storage *bpf_cgroup_storage_alloc(struct bpf_prog *prog,
+					enum bpf_cgroup_storage_type stype)
 {
 	struct bpf_cgroup_storage *storage;
 	struct bpf_map *map;
+	gfp_t flags;
+	size_t size;
 	u32 pages;
 
-	map = prog->aux->cgroup_storage;
+	map = prog->aux->cgroup_storage[stype];
 	if (!map)
 		return NULL;
 
-	pages = round_up(sizeof(struct bpf_cgroup_storage) +
-			 sizeof(struct bpf_storage_buffer) +
-			 map->value_size, PAGE_SIZE) >> PAGE_SHIFT;
+	size = bpf_cgroup_storage_calculate_size(map, &pages);
+
 	if (bpf_map_charge_memlock(map, pages))
 		return ERR_PTR(-EPERM);
 
 	storage = kmalloc_node(sizeof(struct bpf_cgroup_storage),
 			       __GFP_ZERO | GFP_USER, map->numa_node);
-	if (!storage) {
-		bpf_map_uncharge_memlock(map, pages);
-		return ERR_PTR(-ENOMEM);
+	if (!storage)
+		goto enomem;
+
+	flags = __GFP_ZERO | GFP_USER;
+
+	if (stype == BPF_CGROUP_STORAGE_SHARED) {
+		storage->buf = kmalloc_node(size, flags, map->numa_node);
+		if (!storage->buf)
+			goto enomem;
+	} else {
+		storage->percpu_buf = __alloc_percpu_gfp(size, 8, flags);
+		if (!storage->percpu_buf)
+			goto enomem;
 	}
 
-	storage->buf = kmalloc_node(sizeof(struct bpf_storage_buffer) +
-				    map->value_size, __GFP_ZERO | GFP_USER,
-				    map->numa_node);
-	if (!storage->buf) {
+	storage->map = (struct bpf_cgroup_storage_map *)map;
+
+	return storage;
+
+enomem:
 	bpf_map_uncharge_memlock(map, pages);
 	kfree(storage);
 	return ERR_PTR(-ENOMEM);
-	}
+}
 
-	storage->map = (struct bpf_cgroup_storage_map *)map;
+static void free_shared_cgroup_storage_rcu(struct rcu_head *rcu)
+{
+	struct bpf_cgroup_storage *storage =
+		container_of(rcu, struct bpf_cgroup_storage, rcu);
 
-	return storage;
+	kfree(storage->buf);
+	kfree(storage);
+}
+
+static void free_percpu_cgroup_storage_rcu(struct rcu_head *rcu)
+{
+	struct bpf_cgroup_storage *storage =
+		container_of(rcu, struct bpf_cgroup_storage, rcu);
+
+	free_percpu(storage->percpu_buf);
+	kfree(storage);
 }
 
 void bpf_cgroup_storage_free(struct bpf_cgroup_storage *storage)
 {
-	u32 pages;
+	enum bpf_cgroup_storage_type stype;
 	struct bpf_map *map;
+	u32 pages;
 
 	if (!storage)
 		return;
 
 	map = &storage->map->map;
-	pages = round_up(sizeof(struct bpf_cgroup_storage) +
-			 sizeof(struct bpf_storage_buffer) +
-			 map->value_size, PAGE_SIZE) >> PAGE_SHIFT;
+
+	bpf_cgroup_storage_calculate_size(map, &pages);
 	bpf_map_uncharge_memlock(map, pages);
 
-	kfree_rcu(storage->buf, rcu);
-	kfree_rcu(storage, rcu);
+	stype = cgroup_storage_type(map);
+	if (stype == BPF_CGROUP_STORAGE_SHARED)
+		call_rcu(&storage->rcu, free_shared_cgroup_storage_rcu);
+	else
+		call_rcu(&storage->rcu, free_percpu_cgroup_storage_rcu);
 }
 
 void bpf_cgroup_storage_link(struct bpf_cgroup_storage *storage,

kernel/bpf/map_in_map.c

@@ -24,7 +24,8 @@ struct bpf_map *bpf_map_meta_alloc(int inner_map_ufd)
 	 * in the verifier is not enough.
 	 */
 	if (inner_map->map_type == BPF_MAP_TYPE_PROG_ARRAY ||
-	    inner_map->map_type == BPF_MAP_TYPE_CGROUP_STORAGE) {
+	    inner_map->map_type == BPF_MAP_TYPE_CGROUP_STORAGE ||
+	    inner_map->map_type == BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE) {
 		fdput(f);
 		return ERR_PTR(-ENOTSUPP);
 	}

kernel/bpf/syscall.c

@@ -686,7 +686,8 @@ static int map_lookup_elem(union bpf_attr *attr)
 
 	if (map->map_type == BPF_MAP_TYPE_PERCPU_HASH ||
 	    map->map_type == BPF_MAP_TYPE_LRU_PERCPU_HASH ||
-	    map->map_type == BPF_MAP_TYPE_PERCPU_ARRAY)
+	    map->map_type == BPF_MAP_TYPE_PERCPU_ARRAY ||
+	    map->map_type == BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE)
 		value_size = round_up(map->value_size, 8) * num_possible_cpus();
 	else if (IS_FD_MAP(map))
 		value_size = sizeof(u32);
@@ -705,6 +706,8 @@ static int map_lookup_elem(union bpf_attr *attr)
 		err = bpf_percpu_hash_copy(map, key, value);
 	} else if (map->map_type == BPF_MAP_TYPE_PERCPU_ARRAY) {
 		err = bpf_percpu_array_copy(map, key, value);
+	} else if (map->map_type == BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE) {
+		err = bpf_percpu_cgroup_storage_copy(map, key, value);
 	} else if (map->map_type == BPF_MAP_TYPE_STACK_TRACE) {
 		err = bpf_stackmap_copy(map, key, value);
 	} else if (IS_FD_ARRAY(map)) {
@@ -774,7 +777,8 @@ static int map_update_elem(union bpf_attr *attr)
 
 	if (map->map_type == BPF_MAP_TYPE_PERCPU_HASH ||
 	    map->map_type == BPF_MAP_TYPE_LRU_PERCPU_HASH ||
-	    map->map_type == BPF_MAP_TYPE_PERCPU_ARRAY)
+	    map->map_type == BPF_MAP_TYPE_PERCPU_ARRAY ||
+	    map->map_type == BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE)
 		value_size = round_up(map->value_size, 8) * num_possible_cpus();
 	else
 		value_size = map->value_size;
@@ -809,6 +813,9 @@ static int map_update_elem(union bpf_attr *attr)
 		err = bpf_percpu_hash_update(map, key, value, attr->flags);
 	} else if (map->map_type == BPF_MAP_TYPE_PERCPU_ARRAY) {
 		err = bpf_percpu_array_update(map, key, value, attr->flags);
+	} else if (map->map_type == BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE) {
+		err = bpf_percpu_cgroup_storage_update(map, key, value,
+						       attr->flags);
 	} else if (IS_FD_ARRAY(map)) {
 		rcu_read_lock();
 		err = bpf_fd_array_map_update_elem(map, f.file, key, value,
@@ -988,10 +995,15 @@ static int find_prog_type(enum bpf_prog_type type, struct bpf_prog *prog)
 /* drop refcnt on maps used by eBPF program and free auxilary data */
 static void free_used_maps(struct bpf_prog_aux *aux)
 {
+	enum bpf_cgroup_storage_type stype;
 	int i;
 
-	if (aux->cgroup_storage)
-		bpf_cgroup_storage_release(aux->prog, aux->cgroup_storage);
+	for_each_cgroup_storage_type(stype) {
+		if (!aux->cgroup_storage[stype])
+			continue;
+		bpf_cgroup_storage_release(aux->prog,
+					   aux->cgroup_storage[stype]);
+	}
 
 	for (i = 0; i < aux->used_map_cnt; i++)
 		bpf_map_put(aux->used_maps[i]);

kernel/bpf/verifier.c

@@ -2074,6 +2074,7 @@ static int check_map_func_compatibility(struct bpf_verifier_env *env,
 			goto error;
 		break;
 	case BPF_MAP_TYPE_CGROUP_STORAGE:
+	case BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE:
 		if (func_id != BPF_FUNC_get_local_storage)
 			goto error;
 		break;
@@ -2164,7 +2165,8 @@ static int check_map_func_compatibility(struct bpf_verifier_env *env,
 			goto error;
 		break;
 	case BPF_FUNC_get_local_storage:
-		if (map->map_type != BPF_MAP_TYPE_CGROUP_STORAGE)
+		if (map->map_type != BPF_MAP_TYPE_CGROUP_STORAGE &&
+		    map->map_type != BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE)
 			goto error;
 		break;
 	case BPF_FUNC_sk_select_reuseport:
@@ -5049,6 +5051,12 @@ static int check_map_prog_compatibility(struct bpf_verifier_env *env,
 	return 0;
 }
 
+static bool bpf_map_is_cgroup_storage(struct bpf_map *map)
+{
+	return (map->map_type == BPF_MAP_TYPE_CGROUP_STORAGE ||
+		map->map_type == BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE);
+}
+
 /* look for pseudo eBPF instructions that access map FDs and
  * replace them with actual map pointers
  */
@@ -5139,10 +5147,9 @@ static int replace_map_fd_with_map_ptr(struct bpf_verifier_env *env)
 			}
 			env->used_maps[env->used_map_cnt++] = map;
 
-			if (map->map_type == BPF_MAP_TYPE_CGROUP_STORAGE &&
+			if (bpf_map_is_cgroup_storage(map) &&
 			    bpf_cgroup_storage_assign(env->prog, map)) {
-				verbose(env,
-					"only one cgroup storage is allowed\n");
+				verbose(env, "only one cgroup storage of each type is allowed\n");
 				fdput(f);
 				return -EBUSY;
 			}
@@ -5171,11 +5178,15 @@ static int replace_map_fd_with_map_ptr(struct bpf_verifier_env *env)
 /* drop refcnt of maps used by the rejected program */
 static void release_maps(struct bpf_verifier_env *env)
 {
+	enum bpf_cgroup_storage_type stype;
 	int i;
 
-	if (env->prog->aux->cgroup_storage)
+	for_each_cgroup_storage_type(stype) {
+		if (!env->prog->aux->cgroup_storage[stype])
+			continue;
 		bpf_cgroup_storage_release(env->prog,
-					   env->prog->aux->cgroup_storage);
+			env->prog->aux->cgroup_storage[stype]);
+	}
 
 	for (i = 0; i < env->used_map_cnt; i++)
 		bpf_map_put(env->used_maps[i]);

net/bpf/test_run.c

@@ -12,7 +12,7 @@
 #include <linux/sched/signal.h>
 
 static __always_inline u32 bpf_test_run_one(struct bpf_prog *prog, void *ctx,
-					    struct bpf_cgroup_storage *storage)
+		struct bpf_cgroup_storage *storage[MAX_BPF_CGROUP_STORAGE_TYPE])
 {
 	u32 ret;
 
@@ -28,13 +28,20 @@ static __always_inline u32 bpf_test_run_one(struct bpf_prog *prog, void *ctx,
 
 static u32 bpf_test_run(struct bpf_prog *prog, void *ctx, u32 repeat, u32 *time)
 {
-	struct bpf_cgroup_storage *storage = NULL;
+	struct bpf_cgroup_storage *storage[MAX_BPF_CGROUP_STORAGE_TYPE] = { 0 };
+	enum bpf_cgroup_storage_type stype;
 	u64 time_start, time_spent = 0;
 	u32 ret = 0, i;
 
-	storage = bpf_cgroup_storage_alloc(prog);
-	if (IS_ERR(storage))
-		return PTR_ERR(storage);
+	for_each_cgroup_storage_type(stype) {
+		storage[stype] = bpf_cgroup_storage_alloc(prog, stype);
+		if (IS_ERR(storage[stype])) {
+			storage[stype] = NULL;
+			for_each_cgroup_storage_type(stype)
+				bpf_cgroup_storage_free(storage[stype]);
+			return -ENOMEM;
+		}
+	}
 
 	if (!repeat)
 		repeat = 1;
@@ -53,7 +60,8 @@ static u32 bpf_test_run(struct bpf_prog *prog, void *ctx, u32 repeat, u32 *time)
 	do_div(time_spent, repeat);
 	*time = time_spent > U32_MAX ? U32_MAX : (u32)time_spent;
 
-	bpf_cgroup_storage_free(storage);
+	for_each_cgroup_storage_type(stype)
+		bpf_cgroup_storage_free(storage[stype]);
 
 	return ret;
 }

samples/bpf/test_cgrp2_attach2.c

@@ -209,7 +209,7 @@ static int map_fd = -1;
 
 static int prog_load_cnt(int verdict, int val)
 {
-	int cgroup_storage_fd;
+	int cgroup_storage_fd, percpu_cgroup_storage_fd;
 
 	if (map_fd < 0)
 		map_fd = bpf_create_map(BPF_MAP_TYPE_ARRAY, 4, 8, 1, 0);
@@ -225,6 +225,14 @@ static int prog_load_cnt(int verdict, int val)
 		return -1;
 	}
 
+	percpu_cgroup_storage_fd = bpf_create_map(
+		BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE,
+		sizeof(struct bpf_cgroup_storage_key), 8, 0, 0);
+	if (percpu_cgroup_storage_fd < 0) {
+		printf("failed to create map '%s'\n", strerror(errno));
+		return -1;
+	}
+
 	struct bpf_insn prog[] = {
 		BPF_MOV32_IMM(BPF_REG_0, 0),
 		BPF_STX_MEM(BPF_W, BPF_REG_10, BPF_REG_0, -4), /* *(u32 *)(fp - 4) = r0 */
@@ -235,11 +243,20 @@ static int prog_load_cnt(int verdict, int val)
 		BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, 0, 2),
 		BPF_MOV64_IMM(BPF_REG_1, val), /* r1 = 1 */
 		BPF_RAW_INSN(BPF_STX | BPF_XADD | BPF_DW, BPF_REG_0, BPF_REG_1, 0, 0), /* xadd r0 += r1 */
+
 		BPF_LD_MAP_FD(BPF_REG_1, cgroup_storage_fd),
 		BPF_MOV64_IMM(BPF_REG_2, 0),
 		BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0, BPF_FUNC_get_local_storage),
 		BPF_MOV64_IMM(BPF_REG_1, val),
 		BPF_RAW_INSN(BPF_STX | BPF_XADD | BPF_W, BPF_REG_0, BPF_REG_1, 0, 0),
+
+		BPF_LD_MAP_FD(BPF_REG_1, percpu_cgroup_storage_fd),
+		BPF_MOV64_IMM(BPF_REG_2, 0),
+		BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0, BPF_FUNC_get_local_storage),
+		BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_0, 0),
+		BPF_ALU64_IMM(BPF_ADD, BPF_REG_3, 0x1),
+		BPF_STX_MEM(BPF_W, BPF_REG_0, BPF_REG_3, 0),
+
 		BPF_MOV64_IMM(BPF_REG_0, verdict), /* r0 = verdict */
 		BPF_EXIT_INSN(),
 	};

tools/bpf/bpftool/map.c

@@ -72,13 +72,15 @@ static const char * const map_type_name[] = {
 	[BPF_MAP_TYPE_SOCKHASH]		= "sockhash",
 	[BPF_MAP_TYPE_CGROUP_STORAGE]	= "cgroup_storage",
 	[BPF_MAP_TYPE_REUSEPORT_SOCKARRAY] = "reuseport_sockarray",
+	[BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE]	= "percpu_cgroup_storage",
 };
 
 static bool map_is_per_cpu(__u32 type)
 {
 	return type == BPF_MAP_TYPE_PERCPU_HASH ||
 	       type == BPF_MAP_TYPE_PERCPU_ARRAY ||
-	       type == BPF_MAP_TYPE_LRU_PERCPU_HASH;
+	       type == BPF_MAP_TYPE_LRU_PERCPU_HASH ||
+	       type == BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE;
 }
 
 static bool map_is_map_of_maps(__u32 type)

tools/include/uapi/linux/bpf.h

@@ -127,6 +127,7 @@ enum bpf_map_type {
 	BPF_MAP_TYPE_SOCKHASH,
 	BPF_MAP_TYPE_CGROUP_STORAGE,
 	BPF_MAP_TYPE_REUSEPORT_SOCKARRAY,
+	BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE,
 };
 
 enum bpf_prog_type {

tools/testing/selftests/bpf/Makefile

@@ -23,7 +23,8 @@ $(TEST_CUSTOM_PROGS): $(OUTPUT)/%: %.c
 TEST_GEN_PROGS = test_verifier test_tag test_maps test_lru_map test_lpm_map test_progs \
 	test_align test_verifier_log test_dev_cgroup test_tcpbpf_user \
 	test_sock test_btf test_sockmap test_lirc_mode2_user get_cgroup_id_user \
-	test_socket_cookie test_cgroup_storage test_select_reuseport test_section_names
+	test_socket_cookie test_cgroup_storage test_select_reuseport test_section_names \
+	test_netcnt
 
 TEST_GEN_FILES = test_pkt_access.o test_xdp.o test_l4lb.o test_tcp_estats.o test_obj_id.o \
 	test_pkt_md_access.o test_xdp_redirect.o test_xdp_meta.o sockmap_parse_prog.o     \
@@ -35,7 +36,7 @@ TEST_GEN_FILES = test_pkt_access.o test_xdp.o test_l4lb.o test_tcp_estats.o test
 	test_get_stack_rawtp.o test_sockmap_kern.o test_sockhash_kern.o \
 	test_lwt_seg6local.o sendmsg4_prog.o sendmsg6_prog.o test_lirc_mode2_kern.o \
 	get_cgroup_id_kern.o socket_cookie_prog.o test_select_reuseport_kern.o \
-	test_skb_cgroup_id_kern.o bpf_flow.o
+	test_skb_cgroup_id_kern.o bpf_flow.o netcnt_prog.o
 
 # Order correspond to 'make run_tests' order
 TEST_PROGS := test_kmod.sh \
@@ -72,6 +73,7 @@ $(OUTPUT)/test_tcpbpf_user: cgroup_helpers.c
 $(OUTPUT)/test_progs: trace_helpers.c
 $(OUTPUT)/get_cgroup_id_user: cgroup_helpers.c
 $(OUTPUT)/test_cgroup_storage: cgroup_helpers.c
+$(OUTPUT)/test_netcnt: cgroup_helpers.c
 
 .PHONY: force
 

tools/testing/selftests/bpf/netcnt_common.h

+// SPDX-License-Identifier: GPL-2.0
+#ifndef __NETCNT_COMMON_H
+#define __NETCNT_COMMON_H
+
+#include <linux/types.h>
+
+#define MAX_PERCPU_PACKETS 32
+
+struct percpu_net_cnt {
+	__u64 packets;
+	__u64 bytes;
+
+	__u64 prev_ts;
+
+	__u64 prev_packets;
+	__u64 prev_bytes;
+};
+
+struct net_cnt {
+	__u64 packets;
+	__u64 bytes;
+};
+
+#endif

tools/testing/selftests/bpf/netcnt_prog.c

+// SPDX-License-Identifier: GPL-2.0
+#include <linux/bpf.h>
+#include <linux/version.h>
+
+#include "bpf_helpers.h"
+#include "netcnt_common.h"
+
+#define MAX_BPS	(3 * 1024 * 1024)
+
+#define REFRESH_TIME_NS	100000000
+#define NS_PER_SEC	1000000000
+
+struct bpf_map_def SEC("maps") percpu_netcnt = {
+	.type = BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE,
+	.key_size = sizeof(struct bpf_cgroup_storage_key),
+	.value_size = sizeof(struct percpu_net_cnt),
+};
+
+struct bpf_map_def SEC("maps") netcnt = {
+	.type = BPF_MAP_TYPE_CGROUP_STORAGE,
+	.key_size = sizeof(struct bpf_cgroup_storage_key),
+	.value_size = sizeof(struct net_cnt),
+};
+
+SEC("cgroup/skb")
+int bpf_nextcnt(struct __sk_buff *skb)
+{
+	struct percpu_net_cnt *percpu_cnt;
+	char fmt[] = "%d %llu %llu\n";
+	struct net_cnt *cnt;
+	__u64 ts, dt;
+	int ret;
+
+	cnt = bpf_get_local_storage(&netcnt, 0);
+	percpu_cnt = bpf_get_local_storage(&percpu_netcnt, 0);
+
+	percpu_cnt->packets++;
+	percpu_cnt->bytes += skb->len;
+
+	if (percpu_cnt->packets > MAX_PERCPU_PACKETS) {
+		__sync_fetch_and_add(&cnt->packets,
+				     percpu_cnt->packets);
+		percpu_cnt->packets = 0;
+
+		__sync_fetch_and_add(&cnt->bytes,
+				     percpu_cnt->bytes);
+		percpu_cnt->bytes = 0;
+	}
+
+	ts = bpf_ktime_get_ns();
+	dt = ts - percpu_cnt->prev_ts;
+
+	dt *= MAX_BPS;
+	dt /= NS_PER_SEC;
+
+	if (cnt->bytes + percpu_cnt->bytes - percpu_cnt->prev_bytes < dt)
+		ret = 1;
+	else
+		ret = 0;
+
+	if (dt > REFRESH_TIME_NS) {
+		percpu_cnt->prev_ts = ts;
+		percpu_cnt->prev_packets = cnt->packets;
+		percpu_cnt->prev_bytes = cnt->bytes;
+	}
+
+	return !!ret;
+}
+
+char _license[] SEC("license") = "GPL";
+__u32 _version SEC("version") = LINUX_VERSION_CODE;

tools/testing/selftests/bpf/test_cgroup_storage.c

@@ -4,6 +4,7 @@
 #include <linux/filter.h>
 #include <stdio.h>
 #include <stdlib.h>
+#include <sys/sysinfo.h>
 
 #include "bpf_rlimit.h"
 #include "cgroup_helpers.h"
@@ -15,6 +16,14 @@ char bpf_log_buf[BPF_LOG_BUF_SIZE];
 int main(int argc, char **argv)
 {
 	struct bpf_insn prog[] = {
+		BPF_LD_MAP_FD(BPF_REG_1, 0), /* percpu map fd */
+		BPF_MOV64_IMM(BPF_REG_2, 0), /* flags, not used */
+		BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
+			     BPF_FUNC_get_local_storage),
+		BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_0, 0),
+		BPF_ALU64_IMM(BPF_ADD, BPF_REG_3, 0x1),
+		BPF_STX_MEM(BPF_W, BPF_REG_0, BPF_REG_3, 0),
+
 		BPF_LD_MAP_FD(BPF_REG_1, 0), /* map fd */
 		BPF_MOV64_IMM(BPF_REG_2, 0), /* flags, not used */
 		BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
@@ -28,9 +37,18 @@ int main(int argc, char **argv)
 	};
 	size_t insns_cnt = sizeof(prog) / sizeof(struct bpf_insn);
 	int error = EXIT_FAILURE;
-	int map_fd, prog_fd, cgroup_fd;
+	int map_fd, percpu_map_fd, prog_fd, cgroup_fd;
 	struct bpf_cgroup_storage_key key;
 	unsigned long long value;
+	unsigned long long *percpu_value;
+	int cpu, nproc;
+
+	nproc = get_nprocs_conf();
+	percpu_value = malloc(sizeof(*percpu_value) * nproc);
+	if (!percpu_value) {
+		printf("Not enough memory for per-cpu area (%d cpus)\n", nproc);
+		goto err;
+	}
 
 	map_fd = bpf_create_map(BPF_MAP_TYPE_CGROUP_STORAGE, sizeof(key),
 				sizeof(value), 0, 0);
@@ -39,7 +57,15 @@ int main(int argc, char **argv)
 		goto out;
 	}
 
-	prog[0].imm = map_fd;
+	percpu_map_fd = bpf_create_map(BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE,
+				       sizeof(key), sizeof(value), 0, 0);
+	if (percpu_map_fd < 0) {
+		printf("Failed to create map: %s\n", strerror(errno));
+		goto out;
+	}
+
+	prog[0].imm = percpu_map_fd;
+	prog[7].imm = map_fd;
 	prog_fd = bpf_load_program(BPF_PROG_TYPE_CGROUP_SKB,
 				   prog, insns_cnt, "GPL", 0,
 				   bpf_log_buf, BPF_LOG_BUF_SIZE);
@@ -77,7 +103,15 @@ int main(int argc, char **argv)
 	}
 
 	if (bpf_map_lookup_elem(map_fd, &key, &value)) {
-		printf("Failed to lookup cgroup storage\n");
+		printf("Failed to lookup cgroup storage 0\n");
+		goto err;
+	}
+
+	for (cpu = 0; cpu < nproc; cpu++)
+		percpu_value[cpu] = 1000;
+
+	if (bpf_map_update_elem(percpu_map_fd, &key, percpu_value, 0)) {
+		printf("Failed to update the data in the cgroup storage\n");
 		goto err;
 	}
 
@@ -120,11 +154,31 @@ int main(int argc, char **argv)
 		goto err;
 	}
 
+	/* Check the final value of the counter in the percpu local storage */
+
+	for (cpu = 0; cpu < nproc; cpu++)
+		percpu_value[cpu] = 0;
+
+	if (bpf_map_lookup_elem(percpu_map_fd, &key, percpu_value)) {
+		printf("Failed to lookup the per-cpu cgroup storage\n");
+		goto err;
+	}
+
+	value = 0;
+	for (cpu = 0; cpu < nproc; cpu++)
+		value += percpu_value[cpu];
+
+	if (value != nproc * 1000 + 6) {
+		printf("Unexpected data in the per-cpu cgroup storage\n");
+		goto err;
+	}
+
 	error = 0;
 	printf("test_cgroup_storage:PASS\n");
 
 err:
 	cleanup_cgroup_environment();
+	free(percpu_value);
 
 out:
 	return error;

tools/testing/selftests/bpf/test_netcnt.c

+// SPDX-License-Identifier: GPL-2.0
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <errno.h>
+#include <assert.h>
+#include <sys/sysinfo.h>
+#include <sys/time.h>
+
+#include <linux/bpf.h>
+#include <bpf/bpf.h>
+#include <bpf/libbpf.h>
+
+#include "cgroup_helpers.h"
+#include "bpf_rlimit.h"
+#include "netcnt_common.h"
+
+#define BPF_PROG "./netcnt_prog.o"
+#define TEST_CGROUP "/test-network-counters/"
+
+static int bpf_find_map(const char *test, struct bpf_object *obj,
+			const char *name)
+{
+	struct bpf_map *map;
+
+	map = bpf_object__find_map_by_name(obj, name);
+	if (!map) {
+		printf("%s:FAIL:map '%s' not found\n", test, name);
+		return -1;
+	}
+	return bpf_map__fd(map);
+}
+
+int main(int argc, char **argv)
+{
+	struct percpu_net_cnt *percpu_netcnt;
+	struct bpf_cgroup_storage_key key;
+	int map_fd, percpu_map_fd;
+	int error = EXIT_FAILURE;
+	struct net_cnt netcnt;
+	struct bpf_object *obj;
+	int prog_fd, cgroup_fd;
+	unsigned long packets;
+	unsigned long bytes;
+	int cpu, nproc;
+	__u32 prog_cnt;
+
+	nproc = get_nprocs_conf();
+	percpu_netcnt = malloc(sizeof(*percpu_netcnt) * nproc);
+	if (!percpu_netcnt) {
+		printf("Not enough memory for per-cpu area (%d cpus)\n", nproc);
+		goto err;
+	}
+
+	if (bpf_prog_load(BPF_PROG, BPF_PROG_TYPE_CGROUP_SKB,
+			  &obj, &prog_fd)) {
+		printf("Failed to load bpf program\n");
+		goto out;
+	}
+
+	if (setup_cgroup_environment()) {
+		printf("Failed to load bpf program\n");
+		goto err;
+	}
+
+	/* Create a cgroup, get fd, and join it */
+	cgroup_fd = create_and_get_cgroup(TEST_CGROUP);
+	if (!cgroup_fd) {
+		printf("Failed to create test cgroup\n");
+		goto err;
+	}
+
+	if (join_cgroup(TEST_CGROUP)) {
+		printf("Failed to join cgroup\n");
+		goto err;
+	}
+
+	/* Attach bpf program */
+	if (bpf_prog_attach(prog_fd, cgroup_fd, BPF_CGROUP_INET_EGRESS, 0)) {
+		printf("Failed to attach bpf program");
+		goto err;
+	}
+
+	assert(system("ping localhost -6 -c 10000 -f -q > /dev/null") == 0);
+
+	if (bpf_prog_query(cgroup_fd, BPF_CGROUP_INET_EGRESS, 0, NULL, NULL,
+			   &prog_cnt)) {
+		printf("Failed to query attached programs");
+		goto err;
+	}
+
+	map_fd = bpf_find_map(__func__, obj, "netcnt");
+	if (map_fd < 0) {
+		printf("Failed to find bpf map with net counters");
+		goto err;
+	}
+
+	percpu_map_fd = bpf_find_map(__func__, obj, "percpu_netcnt");
+	if (percpu_map_fd < 0) {
+		printf("Failed to find bpf map with percpu net counters");
+		goto err;
+	}
+
+	if (bpf_map_get_next_key(map_fd, NULL, &key)) {
+		printf("Failed to get key in cgroup storage\n");
+		goto err;
+	}
+
+	if (bpf_map_lookup_elem(map_fd, &key, &netcnt)) {
+		printf("Failed to lookup cgroup storage\n");
+		goto err;
+	}
+
+	if (bpf_map_lookup_elem(percpu_map_fd, &key, &percpu_netcnt[0])) {
+		printf("Failed to lookup percpu cgroup storage\n");
+		goto err;
+	}
+
+	/* Some packets can be still in per-cpu cache, but not more than
+	 * MAX_PERCPU_PACKETS.
+	 */
+	packets = netcnt.packets;
+	bytes = netcnt.bytes;
+	for (cpu = 0; cpu < nproc; cpu++) {
+		if (percpu_netcnt[cpu].packets > MAX_PERCPU_PACKETS) {
+			printf("Unexpected percpu value: %llu\n",
+			       percpu_netcnt[cpu].packets);
+			goto err;
+		}
+
+		packets += percpu_netcnt[cpu].packets;
+		bytes += percpu_netcnt[cpu].bytes;
+	}
+
+	/* No packets should be lost */
+	if (packets != 10000) {
+		printf("Unexpected packet count: %lu\n", packets);
+		goto err;
+	}
+
+	/* Let's check that bytes counter matches the number of packets
+	 * multiplied by the size of ipv6 ICMP packet.
+	 */
+	if (bytes != packets * 104) {
+		printf("Unexpected bytes count: %lu\n", bytes);
+		goto err;
+	}
+
+	error = 0;
+	printf("test_netcnt:PASS\n");
+
+err:
+	cleanup_cgroup_environment();
+	free(percpu_netcnt);
+
+out:
+	return error;
+}

tools/testing/selftests/bpf/test_verifier.c

@@ -68,6 +68,7 @@ struct bpf_test {
 	int fixup_prog2[MAX_FIXUPS];
 	int fixup_map_in_map[MAX_FIXUPS];
 	int fixup_cgroup_storage[MAX_FIXUPS];
+	int fixup_percpu_cgroup_storage[MAX_FIXUPS];
 	const char *errstr;
 	const char *errstr_unpriv;
 	uint32_t retval;
@@ -4676,7 +4677,7 @@ static struct bpf_test tests[] = {
 		.prog_type = BPF_PROG_TYPE_CGROUP_SKB,
 	},
 	{
-		"invalid per-cgroup storage access 3",
+		"invalid cgroup storage access 3",
 		.insns = {
 			BPF_MOV64_IMM(BPF_REG_2, 0),
 			BPF_LD_MAP_FD(BPF_REG_1, 0),
@@ -4743,6 +4744,121 @@ static struct bpf_test tests[] = {
 		.errstr = "get_local_storage() doesn't support non-zero flags",
 		.prog_type = BPF_PROG_TYPE_CGROUP_SKB,
 	},
+	{
+		"valid per-cpu cgroup storage access",
+		.insns = {
+			BPF_MOV64_IMM(BPF_REG_2, 0),
+			BPF_LD_MAP_FD(BPF_REG_1, 0),
+			BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
+				     BPF_FUNC_get_local_storage),
+			BPF_LDX_MEM(BPF_W, BPF_REG_1, BPF_REG_0, 0),
+			BPF_MOV64_REG(BPF_REG_0, BPF_REG_1),
+			BPF_ALU64_IMM(BPF_AND, BPF_REG_0, 1),
+			BPF_EXIT_INSN(),
+		},
+		.fixup_percpu_cgroup_storage = { 1 },
+		.result = ACCEPT,
+		.prog_type = BPF_PROG_TYPE_CGROUP_SKB,
+	},
+	{
+		"invalid per-cpu cgroup storage access 1",
+		.insns = {
+			BPF_MOV64_IMM(BPF_REG_2, 0),
+			BPF_LD_MAP_FD(BPF_REG_1, 0),
+			BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
+				     BPF_FUNC_get_local_storage),
+			BPF_LDX_MEM(BPF_W, BPF_REG_1, BPF_REG_0, 0),
+			BPF_MOV64_REG(BPF_REG_0, BPF_REG_1),
+			BPF_ALU64_IMM(BPF_AND, BPF_REG_0, 1),
+			BPF_EXIT_INSN(),
+		},
+		.fixup_map1 = { 1 },
+		.result = REJECT,
+		.errstr = "cannot pass map_type 1 into func bpf_get_local_storage",
+		.prog_type = BPF_PROG_TYPE_CGROUP_SKB,
+	},
+	{
+		"invalid per-cpu cgroup storage access 2",
+		.insns = {
+			BPF_MOV64_IMM(BPF_REG_2, 0),
+			BPF_LD_MAP_FD(BPF_REG_1, 1),
+			BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
+				     BPF_FUNC_get_local_storage),
+			BPF_ALU64_IMM(BPF_AND, BPF_REG_0, 1),
+			BPF_EXIT_INSN(),
+		},
+		.result = REJECT,
+		.errstr = "fd 1 is not pointing to valid bpf_map",
+		.prog_type = BPF_PROG_TYPE_CGROUP_SKB,
+	},
+	{
+		"invalid per-cpu cgroup storage access 3",
+		.insns = {
+			BPF_MOV64_IMM(BPF_REG_2, 0),
+			BPF_LD_MAP_FD(BPF_REG_1, 0),
+			BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
+				     BPF_FUNC_get_local_storage),
+			BPF_LDX_MEM(BPF_W, BPF_REG_1, BPF_REG_0, 256),
+			BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, 1),
+			BPF_MOV64_IMM(BPF_REG_0, 0),
+			BPF_EXIT_INSN(),
+		},
+		.fixup_percpu_cgroup_storage = { 1 },
+		.result = REJECT,
+		.errstr = "invalid access to map value, value_size=64 off=256 size=4",
+		.prog_type = BPF_PROG_TYPE_CGROUP_SKB,
+	},
+	{
+		"invalid per-cpu cgroup storage access 4",
+		.insns = {
+			BPF_MOV64_IMM(BPF_REG_2, 0),
+			BPF_LD_MAP_FD(BPF_REG_1, 0),
+			BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
+				     BPF_FUNC_get_local_storage),
+			BPF_LDX_MEM(BPF_W, BPF_REG_1, BPF_REG_0, -2),
+			BPF_MOV64_REG(BPF_REG_0, BPF_REG_1),
+			BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, 1),
+			BPF_EXIT_INSN(),
+		},
+		.fixup_cgroup_storage = { 1 },
+		.result = REJECT,
+		.errstr = "invalid access to map value, value_size=64 off=-2 size=4",
+		.prog_type = BPF_PROG_TYPE_CGROUP_SKB,
+	},
+	{
+		"invalid per-cpu cgroup storage access 5",
+		.insns = {
+			BPF_MOV64_IMM(BPF_REG_2, 7),
+			BPF_LD_MAP_FD(BPF_REG_1, 0),
+			BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
+				     BPF_FUNC_get_local_storage),
+			BPF_LDX_MEM(BPF_W, BPF_REG_1, BPF_REG_0, 0),
+			BPF_MOV64_REG(BPF_REG_0, BPF_REG_1),
+			BPF_ALU64_IMM(BPF_AND, BPF_REG_0, 1),
+			BPF_EXIT_INSN(),
+		},
+		.fixup_percpu_cgroup_storage = { 1 },
+		.result = REJECT,
+		.errstr = "get_local_storage() doesn't support non-zero flags",
+		.prog_type = BPF_PROG_TYPE_CGROUP_SKB,
+	},
+	{
+		"invalid per-cpu cgroup storage access 6",
+		.insns = {
+			BPF_MOV64_REG(BPF_REG_2, BPF_REG_1),
+			BPF_LD_MAP_FD(BPF_REG_1, 0),
+			BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0,
+				     BPF_FUNC_get_local_storage),
+			BPF_LDX_MEM(BPF_W, BPF_REG_1, BPF_REG_0, 0),
+			BPF_MOV64_REG(BPF_REG_0, BPF_REG_1),
+			BPF_ALU64_IMM(BPF_AND, BPF_REG_0, 1),
+			BPF_EXIT_INSN(),
+		},
+		.fixup_percpu_cgroup_storage = { 1 },
+		.result = REJECT,
+		.errstr = "get_local_storage() doesn't support non-zero flags",
+		.prog_type = BPF_PROG_TYPE_CGROUP_SKB,
+	},
 	{
 		"multiple registers share map_lookup_elem result",
 		.insns = {
@@ -12615,15 +12731,17 @@ static int create_map_in_map(void)
 	return outer_map_fd;
 }
 
-static int create_cgroup_storage(void)
+static int create_cgroup_storage(bool percpu)
 {
+	enum bpf_map_type type = percpu ? BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE :
+		BPF_MAP_TYPE_CGROUP_STORAGE;
 	int fd;
 
-	fd = bpf_create_map(BPF_MAP_TYPE_CGROUP_STORAGE,
-			    sizeof(struct bpf_cgroup_storage_key),
+	fd = bpf_create_map(type, sizeof(struct bpf_cgroup_storage_key),
 			    TEST_DATA_LEN, 0, 0);
 	if (fd < 0)
-		printf("Failed to create array '%s'!\n", strerror(errno));
+		printf("Failed to create cgroup storage '%s'!\n",
+		       strerror(errno));
 
 	return fd;
 }
@@ -12641,6 +12759,7 @@ static void do_test_fixup(struct bpf_test *test, struct bpf_insn *prog,
 	int *fixup_prog2 = test->fixup_prog2;
 	int *fixup_map_in_map = test->fixup_map_in_map;
 	int *fixup_cgroup_storage = test->fixup_cgroup_storage;
+	int *fixup_percpu_cgroup_storage = test->fixup_percpu_cgroup_storage;
 
 	if (test->fill_helper)
 		test->fill_helper(test);
@@ -12710,12 +12829,20 @@ static void do_test_fixup(struct bpf_test *test, struct bpf_insn *prog,
 	}
 
 	if (*fixup_cgroup_storage) {
-		map_fds[7] = create_cgroup_storage();
+		map_fds[7] = create_cgroup_storage(false);
 		do {
 			prog[*fixup_cgroup_storage].imm = map_fds[7];
 			fixup_cgroup_storage++;
 		} while (*fixup_cgroup_storage);
 	}
+
+	if (*fixup_percpu_cgroup_storage) {
+		map_fds[8] = create_cgroup_storage(true);
+		do {
+			prog[*fixup_percpu_cgroup_storage].imm = map_fds[8];
+			fixup_percpu_cgroup_storage++;
+		} while (*fixup_percpu_cgroup_storage);
+	}
 }
 
 static void do_test_single(struct bpf_test *test, bool unpriv,