Merge branch 'bpf-add-support-for-local-percpu-kptr'

Yonghong Song says:

====================
bpf: Add support for local percpu kptr

Patch set [1] implemented cgroup local storage BPF_MAP_TYPE_CGRP_STORAGE
similar to sk/task/inode local storage and old BPF_MAP_TYPE_CGROUP_STORAGE
map is marked as deprecated since old BPF_MAP_TYPE_CGROUP_STORAGE map can
only work with current cgroup.

Similarly, the existing BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE map
is a percpu version of BPF_MAP_TYPE_CGROUP_STORAGE and only works
with current cgroup. But there is no replacement which can work
with arbitrary cgroup.

This patch set solved this problem but adding support for local
percpu kptr. The map value can have a percpu kptr field which holds
a bpf prog allocated percpu data. The below is an example,

  struct percpu_val_t {
    ... fields ...
  }

  struct map_value_t {
    struct percpu_val_t __percpu_kptr *percpu_data_ptr;
  }

In the above, 'map_value_t' is the map value type for a
BPF_MAP_TYPE_CGRP_STORAGE map. User can access 'percpu_data_ptr'
and then read/write percpu data. This covers BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE
and more. So BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE map type
is marked as deprecated.

In additional, local percpu kptr supports the same map type
as other kptrs including hash, lru_hash, array, sk/inode/task/cgrp
local storage. Currently, percpu data structure does not support
non-scalars or special fields (e.g., bpf_spin_lock, bpf_rb_root, etc.).
They can be supported in the future if there exist use cases.

Please for individual patches for details.

  [1] https://lore.kernel.org/all/20221026042835.672317-1-yhs@fb.com/

Changelog:
  v2 -> v3:
    - fix libbpf_str test failure.
  v1 -> v2:
    - does not support special fields in percpu data structure.
    - rename __percpu attr to __percpu_kptr attr.
    - rename BPF_KPTR_PERCPU_REF to BPF_KPTR_PERCPU.
    - better code to handle bpf_{this,per}_cpu_ptr() helpers.
    - add more negative tests.
    - fix a bpftool related test failure.
====================

Link: https://lore.kernel.org/r/20230827152729.1995219-1-yonghong.song@linux.devSigned-off-by: Alexei Starovoitov <ast@kernel.org>

include/linux/bpf.h

@@ -55,8 +55,8 @@ struct cgroup;
 extern struct idr btf_idr;
 extern spinlock_t btf_idr_lock;
 extern struct kobject *btf_kobj;
-extern struct bpf_mem_alloc bpf_global_ma;
-extern bool bpf_global_ma_set;
+extern struct bpf_mem_alloc bpf_global_ma, bpf_global_percpu_ma;
+extern bool bpf_global_ma_set, bpf_global_percpu_ma_set;
 
 typedef u64 (*bpf_callback_t)(u64, u64, u64, u64, u64);
 typedef int (*bpf_iter_init_seq_priv_t)(void *private_data,
@@ -180,14 +180,15 @@ enum btf_field_type {
 	BPF_TIMER      = (1 << 1),
 	BPF_KPTR_UNREF = (1 << 2),
 	BPF_KPTR_REF   = (1 << 3),
-	BPF_KPTR       = BPF_KPTR_UNREF | BPF_KPTR_REF,
-	BPF_LIST_HEAD  = (1 << 4),
-	BPF_LIST_NODE  = (1 << 5),
-	BPF_RB_ROOT    = (1 << 6),
-	BPF_RB_NODE    = (1 << 7),
+	BPF_KPTR_PERCPU = (1 << 4),
+	BPF_KPTR       = BPF_KPTR_UNREF | BPF_KPTR_REF | BPF_KPTR_PERCPU,
+	BPF_LIST_HEAD  = (1 << 5),
+	BPF_LIST_NODE  = (1 << 6),
+	BPF_RB_ROOT    = (1 << 7),
+	BPF_RB_NODE    = (1 << 8),
 	BPF_GRAPH_NODE_OR_ROOT = BPF_LIST_NODE | BPF_LIST_HEAD |
 				 BPF_RB_NODE | BPF_RB_ROOT,
-	BPF_REFCOUNT   = (1 << 8),
+	BPF_REFCOUNT   = (1 << 9),
 };
 
 typedef void (*btf_dtor_kfunc_t)(void *);
@@ -300,6 +301,8 @@ static inline const char *btf_field_type_name(enum btf_field_type type)
 	case BPF_KPTR_UNREF:
 	case BPF_KPTR_REF:
 		return "kptr";
+	case BPF_KPTR_PERCPU:
+		return "percpu_kptr";
 	case BPF_LIST_HEAD:
 		return "bpf_list_head";
 	case BPF_LIST_NODE:
@@ -325,6 +328,7 @@ static inline u32 btf_field_type_size(enum btf_field_type type)
 		return sizeof(struct bpf_timer);
 	case BPF_KPTR_UNREF:
 	case BPF_KPTR_REF:
+	case BPF_KPTR_PERCPU:
 		return sizeof(u64);
 	case BPF_LIST_HEAD:
 		return sizeof(struct bpf_list_head);
@@ -351,6 +355,7 @@ static inline u32 btf_field_type_align(enum btf_field_type type)
 		return __alignof__(struct bpf_timer);
 	case BPF_KPTR_UNREF:
 	case BPF_KPTR_REF:
+	case BPF_KPTR_PERCPU:
 		return __alignof__(u64);
 	case BPF_LIST_HEAD:
 		return __alignof__(struct bpf_list_head);
@@ -389,6 +394,7 @@ static inline void bpf_obj_init_field(const struct btf_field *field, void *addr)
 	case BPF_TIMER:
 	case BPF_KPTR_UNREF:
 	case BPF_KPTR_REF:
+	case BPF_KPTR_PERCPU:
 		break;
 	default:
 		WARN_ON_ONCE(1);

include/linux/bpf_verifier.h

@@ -480,6 +480,7 @@ struct bpf_insn_aux_data {
 	bool zext_dst; /* this insn zero extends dst reg */
 	bool storage_get_func_atomic; /* bpf_*_storage_get() with atomic memory alloc */
 	bool is_iter_next; /* bpf_iter_<type>_next() kfunc call */
+	bool call_with_percpu_alloc_ptr; /* {this,per}_cpu_ptr() with prog percpu alloc */
 	u8 alu_state; /* used in combination with alu_limit */
 
 	/* below fields are initialized once */

include/uapi/linux/bpf.h

@@ -932,7 +932,14 @@ enum bpf_map_type {
 	 */
 	BPF_MAP_TYPE_CGROUP_STORAGE = BPF_MAP_TYPE_CGROUP_STORAGE_DEPRECATED,
 	BPF_MAP_TYPE_REUSEPORT_SOCKARRAY,
-	BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE,
+	BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE_DEPRECATED,
+	/* BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE is available to bpf programs
+	 * attaching to a cgroup. The new mechanism (BPF_MAP_TYPE_CGRP_STORAGE +
+	 * local percpu kptr) supports all BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE
+	 * functionality and more. So mark * BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE
+	 * deprecated.
+	 */
+	BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE = BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE_DEPRECATED,
 	BPF_MAP_TYPE_QUEUE,
 	BPF_MAP_TYPE_STACK,
 	BPF_MAP_TYPE_SK_STORAGE,

kernel/bpf/btf.c

@@ -3293,6 +3293,8 @@ static int btf_find_kptr(const struct btf *btf, const struct btf_type *t,
 		type = BPF_KPTR_UNREF;
 	else if (!strcmp("kptr", __btf_name_by_offset(btf, t->name_off)))
 		type = BPF_KPTR_REF;
+	else if (!strcmp("percpu_kptr", __btf_name_by_offset(btf, t->name_off)))
+		type = BPF_KPTR_PERCPU;
 	else
 		return -EINVAL;
 
@@ -3457,6 +3459,7 @@ static int btf_find_struct_field(const struct btf *btf,
 			break;
 		case BPF_KPTR_UNREF:
 		case BPF_KPTR_REF:
+		case BPF_KPTR_PERCPU:
 			ret = btf_find_kptr(btf, member_type, off, sz,
 					    idx < info_cnt ? &info[idx] : &tmp);
 			if (ret < 0)
@@ -3523,6 +3526,7 @@ static int btf_find_datasec_var(const struct btf *btf, const struct btf_type *t,
 			break;
 		case BPF_KPTR_UNREF:
 		case BPF_KPTR_REF:
+		case BPF_KPTR_PERCPU:
 			ret = btf_find_kptr(btf, var_type, off, sz,
 					    idx < info_cnt ? &info[idx] : &tmp);
 			if (ret < 0)
@@ -3783,6 +3787,7 @@ struct btf_record *btf_parse_fields(const struct btf *btf, const struct btf_type
 			break;
 		case BPF_KPTR_UNREF:
 		case BPF_KPTR_REF:
+		case BPF_KPTR_PERCPU:
 			ret = btf_parse_kptr(btf, &rec->fields[i], &info_arr[i]);
 			if (ret < 0)
 				goto end;

kernel/bpf/core.c

@@ -64,8 +64,8 @@
 #define OFF	insn->off
 #define IMM	insn->imm
 
-struct bpf_mem_alloc bpf_global_ma;
-bool bpf_global_ma_set;
+struct bpf_mem_alloc bpf_global_ma, bpf_global_percpu_ma;
+bool bpf_global_ma_set, bpf_global_percpu_ma_set;
 
 /* No hurry in this branch
  *
@@ -2921,7 +2921,9 @@ static int __init bpf_global_ma_init(void)
 
 	ret = bpf_mem_alloc_init(&bpf_global_ma, 0, false);
 	bpf_global_ma_set = !ret;
-	return ret;
+	ret = bpf_mem_alloc_init(&bpf_global_percpu_ma, 0, true);
+	bpf_global_percpu_ma_set = !ret;
+	return !bpf_global_ma_set || !bpf_global_percpu_ma_set;
 }
 late_initcall(bpf_global_ma_init);
 #endif

kernel/bpf/helpers.c

@@ -1902,6 +1902,14 @@ __bpf_kfunc void *bpf_obj_new_impl(u64 local_type_id__k, void *meta__ign)
 	return p;
 }
 
+__bpf_kfunc void *bpf_percpu_obj_new_impl(u64 local_type_id__k, void *meta__ign)
+{
+	u64 size = local_type_id__k;
+
+	/* The verifier has ensured that meta__ign must be NULL */
+	return bpf_mem_alloc(&bpf_global_percpu_ma, size);
+}
+
 /* Must be called under migrate_disable(), as required by bpf_mem_free */
 void __bpf_obj_drop_impl(void *p, const struct btf_record *rec)
 {
@@ -1930,6 +1938,12 @@ __bpf_kfunc void bpf_obj_drop_impl(void *p__alloc, void *meta__ign)
 	__bpf_obj_drop_impl(p, meta ? meta->record : NULL);
 }
 
+__bpf_kfunc void bpf_percpu_obj_drop_impl(void *p__alloc, void *meta__ign)
+{
+	/* The verifier has ensured that meta__ign must be NULL */
+	bpf_mem_free_rcu(&bpf_global_percpu_ma, p__alloc);
+}
+
 __bpf_kfunc void *bpf_refcount_acquire_impl(void *p__refcounted_kptr, void *meta__ign)
 {
 	struct btf_struct_meta *meta = meta__ign;
@@ -2442,7 +2456,9 @@ BTF_SET8_START(generic_btf_ids)
 BTF_ID_FLAGS(func, crash_kexec, KF_DESTRUCTIVE)
 #endif
 BTF_ID_FLAGS(func, bpf_obj_new_impl, KF_ACQUIRE | KF_RET_NULL)
+BTF_ID_FLAGS(func, bpf_percpu_obj_new_impl, KF_ACQUIRE | KF_RET_NULL)
 BTF_ID_FLAGS(func, bpf_obj_drop_impl, KF_RELEASE)
+BTF_ID_FLAGS(func, bpf_percpu_obj_drop_impl, KF_RELEASE)
 BTF_ID_FLAGS(func, bpf_refcount_acquire_impl, KF_ACQUIRE | KF_RET_NULL)
 BTF_ID_FLAGS(func, bpf_list_push_front_impl)
 BTF_ID_FLAGS(func, bpf_list_push_back_impl)

kernel/bpf/memalloc.c

@@ -499,15 +499,16 @@ int bpf_mem_alloc_init(struct bpf_mem_alloc *ma, int size, bool percpu)
 	struct obj_cgroup *objcg = NULL;
 	int cpu, i, unit_size, percpu_size = 0;
 
+	/* room for llist_node and per-cpu pointer */
+	if (percpu)
+		percpu_size = LLIST_NODE_SZ + sizeof(void *);
+
 	if (size) {
 		pc = __alloc_percpu_gfp(sizeof(*pc), 8, GFP_KERNEL);
 		if (!pc)
 			return -ENOMEM;
 
-		if (percpu)
-			/* room for llist_node and per-cpu pointer */
-			percpu_size = LLIST_NODE_SZ + sizeof(void *);
-		else
+		if (!percpu)
 			size += LLIST_NODE_SZ; /* room for llist_node */
 		unit_size = size;
 
@@ -527,10 +528,6 @@ int bpf_mem_alloc_init(struct bpf_mem_alloc *ma, int size, bool percpu)
 		return 0;
 	}
 
-	/* size == 0 && percpu is an invalid combination */
-	if (WARN_ON_ONCE(percpu))
-		return -EINVAL;
-
 	pcc = __alloc_percpu_gfp(sizeof(*cc), 8, GFP_KERNEL);
 	if (!pcc)
 		return -ENOMEM;
@@ -543,6 +540,7 @@ int bpf_mem_alloc_init(struct bpf_mem_alloc *ma, int size, bool percpu)
 			c = &cc->cache[i];
 			c->unit_size = sizes[i];
 			c->objcg = objcg;
+			c->percpu_size = percpu_size;
 			c->tgt = c;
 			prefill_mem_cache(c, cpu);
 		}

kernel/bpf/syscall.c

@@ -514,6 +514,7 @@ void btf_record_free(struct btf_record *rec)
 		switch (rec->fields[i].type) {
 		case BPF_KPTR_UNREF:
 		case BPF_KPTR_REF:
+		case BPF_KPTR_PERCPU:
 			if (rec->fields[i].kptr.module)
 				module_put(rec->fields[i].kptr.module);
 			btf_put(rec->fields[i].kptr.btf);
@@ -560,6 +561,7 @@ struct btf_record *btf_record_dup(const struct btf_record *rec)
 		switch (fields[i].type) {
 		case BPF_KPTR_UNREF:
 		case BPF_KPTR_REF:
+		case BPF_KPTR_PERCPU:
 			btf_get(fields[i].kptr.btf);
 			if (fields[i].kptr.module && !try_module_get(fields[i].kptr.module)) {
 				ret = -ENXIO;
@@ -650,6 +652,7 @@ void bpf_obj_free_fields(const struct btf_record *rec, void *obj)
 			WRITE_ONCE(*(u64 *)field_ptr, 0);
 			break;
 		case BPF_KPTR_REF:
+		case BPF_KPTR_PERCPU:
 			xchgd_field = (void *)xchg((unsigned long *)field_ptr, 0);
 			if (!xchgd_field)
 				break;
@@ -1045,6 +1048,7 @@ static int map_check_btf(struct bpf_map *map, const struct btf *btf,
 				break;
 			case BPF_KPTR_UNREF:
 			case BPF_KPTR_REF:
+			case BPF_KPTR_PERCPU:
 			case BPF_REFCOUNT:
 				if (map->map_type != BPF_MAP_TYPE_HASH &&
 				    map->map_type != BPF_MAP_TYPE_PERCPU_HASH &&

kernel/bpf/verifier.c

@@ -304,7 +304,7 @@ struct bpf_kfunc_call_arg_meta {
 	/* arg_{btf,btf_id,owning_ref} are used by kfunc-specific handling,
 	 * generally to pass info about user-defined local kptr types to later
 	 * verification logic
-	 *   bpf_obj_drop
+	 *   bpf_obj_drop/bpf_percpu_obj_drop
 	 *     Record the local kptr type to be drop'd
 	 *   bpf_refcount_acquire (via KF_ARG_PTR_TO_REFCOUNTED_KPTR arg type)
 	 *     Record the local kptr type to be refcount_incr'd and use
@@ -5001,6 +5001,8 @@ static int map_kptr_match_type(struct bpf_verifier_env *env,
 			perm_flags |= PTR_UNTRUSTED;
 	} else {
 		perm_flags = PTR_MAYBE_NULL | MEM_ALLOC;
+		if (kptr_field->type == BPF_KPTR_PERCPU)
+			perm_flags |= MEM_PERCPU;
 	}
 
 	if (base_type(reg->type) != PTR_TO_BTF_ID || (type_flag(reg->type) & ~perm_flags))
@@ -5044,7 +5046,7 @@ static int map_kptr_match_type(struct bpf_verifier_env *env,
 	 */
 	if (!btf_struct_ids_match(&env->log, reg->btf, reg->btf_id, reg->off,
 				  kptr_field->kptr.btf, kptr_field->kptr.btf_id,
-				  kptr_field->type == BPF_KPTR_REF))
+				  kptr_field->type != BPF_KPTR_UNREF))
 		goto bad_type;
 	return 0;
 bad_type:
@@ -5088,7 +5090,18 @@ static bool rcu_safe_kptr(const struct btf_field *field)
 {
 	const struct btf_field_kptr *kptr = &field->kptr;
 
-	return field->type == BPF_KPTR_REF && rcu_protected_object(kptr->btf, kptr->btf_id);
+	return field->type == BPF_KPTR_PERCPU ||
+	       (field->type == BPF_KPTR_REF && rcu_protected_object(kptr->btf, kptr->btf_id));
+}
+
+static u32 btf_ld_kptr_type(struct bpf_verifier_env *env, struct btf_field *kptr_field)
+{
+	if (rcu_safe_kptr(kptr_field) && in_rcu_cs(env)) {
+		if (kptr_field->type != BPF_KPTR_PERCPU)
+			return PTR_MAYBE_NULL | MEM_RCU;
+		return PTR_MAYBE_NULL | MEM_RCU | MEM_PERCPU;
+	}
+	return PTR_MAYBE_NULL | PTR_UNTRUSTED;
 }
 
 static int check_map_kptr_access(struct bpf_verifier_env *env, u32 regno,
@@ -5114,7 +5127,8 @@ static int check_map_kptr_access(struct bpf_verifier_env *env, u32 regno,
 	/* We only allow loading referenced kptr, since it will be marked as
 	 * untrusted, similar to unreferenced kptr.
 	 */
-	if (class != BPF_LDX && kptr_field->type == BPF_KPTR_REF) {
+	if (class != BPF_LDX &&
+	    (kptr_field->type == BPF_KPTR_REF || kptr_field->type == BPF_KPTR_PERCPU)) {
 		verbose(env, "store to referenced kptr disallowed\n");
 		return -EACCES;
 	}
@@ -5125,10 +5139,7 @@ static int check_map_kptr_access(struct bpf_verifier_env *env, u32 regno,
 		 * value from map as PTR_TO_BTF_ID, with the correct type.
 		 */
 		mark_btf_ld_reg(env, cur_regs(env), value_regno, PTR_TO_BTF_ID, kptr_field->kptr.btf,
-				kptr_field->kptr.btf_id,
-				rcu_safe_kptr(kptr_field) && in_rcu_cs(env) ?
-				PTR_MAYBE_NULL | MEM_RCU :
-				PTR_MAYBE_NULL | PTR_UNTRUSTED);
+				kptr_field->kptr.btf_id, btf_ld_kptr_type(env, kptr_field));
 		/* For mark_ptr_or_null_reg */
 		val_reg->id = ++env->id_gen;
 	} else if (class == BPF_STX) {
@@ -5182,6 +5193,7 @@ static int check_map_access(struct bpf_verifier_env *env, u32 regno,
 			switch (field->type) {
 			case BPF_KPTR_UNREF:
 			case BPF_KPTR_REF:
+			case BPF_KPTR_PERCPU:
 				if (src != ACCESS_DIRECT) {
 					verbose(env, "kptr cannot be accessed indirectly by helper\n");
 					return -EACCES;
@@ -6209,7 +6221,7 @@ static int check_ptr_to_btf_access(struct bpf_verifier_env *env,
 		}
 
 		if (type_is_alloc(reg->type) && !type_is_non_owning_ref(reg->type) &&
-		    !reg->ref_obj_id) {
+		    !(reg->type & MEM_RCU) && !reg->ref_obj_id) {
 			verbose(env, "verifier internal error: ref_obj_id for allocated object must be non-zero\n");
 			return -EFAULT;
 		}
@@ -7320,7 +7332,7 @@ static int process_kptr_func(struct bpf_verifier_env *env, int regno,
 		verbose(env, "off=%d doesn't point to kptr\n", kptr_off);
 		return -EACCES;
 	}
-	if (kptr_field->type != BPF_KPTR_REF) {
+	if (kptr_field->type != BPF_KPTR_REF && kptr_field->type != BPF_KPTR_PERCPU) {
 		verbose(env, "off=%d kptr isn't referenced kptr\n", kptr_off);
 		return -EACCES;
 	}
@@ -7753,6 +7765,7 @@ static const struct bpf_reg_types btf_ptr_types = {
 static const struct bpf_reg_types percpu_btf_ptr_types = {
 	.types = {
 		PTR_TO_BTF_ID | MEM_PERCPU,
+		PTR_TO_BTF_ID | MEM_PERCPU | MEM_RCU,
 		PTR_TO_BTF_ID | MEM_PERCPU | PTR_TRUSTED,
 	}
 };
@@ -7831,8 +7844,10 @@ static int check_reg_type(struct bpf_verifier_env *env, u32 regno,
 	if (base_type(arg_type) == ARG_PTR_TO_MEM)
 		type &= ~DYNPTR_TYPE_FLAG_MASK;
 
-	if (meta->func_id == BPF_FUNC_kptr_xchg && type_is_alloc(type))
+	if (meta->func_id == BPF_FUNC_kptr_xchg && type_is_alloc(type)) {
 		type &= ~MEM_ALLOC;
+		type &= ~MEM_PERCPU;
+	}
 
 	for (i = 0; i < ARRAY_SIZE(compatible->types); i++) {
 		expected = compatible->types[i];
@@ -7915,6 +7930,7 @@ static int check_reg_type(struct bpf_verifier_env *env, u32 regno,
 		break;
 	}
 	case PTR_TO_BTF_ID | MEM_ALLOC:
+	case PTR_TO_BTF_ID | MEM_PERCPU | MEM_ALLOC:
 		if (meta->func_id != BPF_FUNC_spin_lock && meta->func_id != BPF_FUNC_spin_unlock &&
 		    meta->func_id != BPF_FUNC_kptr_xchg) {
 			verbose(env, "verifier internal error: unimplemented handling of MEM_ALLOC\n");
@@ -7926,6 +7942,7 @@ static int check_reg_type(struct bpf_verifier_env *env, u32 regno,
 		}
 		break;
 	case PTR_TO_BTF_ID | MEM_PERCPU:
+	case PTR_TO_BTF_ID | MEM_PERCPU | MEM_RCU:
 	case PTR_TO_BTF_ID | MEM_PERCPU | PTR_TRUSTED:
 		/* Handled by helper specific checks */
 		break;
@@ -9532,6 +9549,7 @@ static int check_helper_call(struct bpf_verifier_env *env, struct bpf_insn *insn
 			     int *insn_idx_p)
 {
 	enum bpf_prog_type prog_type = resolve_prog_type(env->prog);
+	bool returns_cpu_specific_alloc_ptr = false;
 	const struct bpf_func_proto *fn = NULL;
 	enum bpf_return_type ret_type;
 	enum bpf_type_flag ret_flag;
@@ -9642,6 +9660,26 @@ static int check_helper_call(struct bpf_verifier_env *env, struct bpf_insn *insn
 				return -EFAULT;
 			}
 			err = unmark_stack_slots_dynptr(env, &regs[meta.release_regno]);
+		} else if (func_id == BPF_FUNC_kptr_xchg && meta.ref_obj_id) {
+			u32 ref_obj_id = meta.ref_obj_id;
+			bool in_rcu = in_rcu_cs(env);
+			struct bpf_func_state *state;
+			struct bpf_reg_state *reg;
+
+			err = release_reference_state(cur_func(env), ref_obj_id);
+			if (!err) {
+				bpf_for_each_reg_in_vstate(env->cur_state, state, reg, ({
+					if (reg->ref_obj_id == ref_obj_id) {
+						if (in_rcu && (reg->type & MEM_ALLOC) && (reg->type & MEM_PERCPU)) {
+							reg->ref_obj_id = 0;
+							reg->type &= ~MEM_ALLOC;
+							reg->type |= MEM_RCU;
+						} else {
+							mark_reg_invalid(env, reg);
+						}
+					}
+				}));
+			}
 		} else if (meta.ref_obj_id) {
 			err = release_reference(env, meta.ref_obj_id);
 		} else if (register_is_null(&regs[meta.release_regno])) {
@@ -9770,6 +9808,23 @@ static int check_helper_call(struct bpf_verifier_env *env, struct bpf_insn *insn
 
 		break;
 	}
+	case BPF_FUNC_per_cpu_ptr:
+	case BPF_FUNC_this_cpu_ptr:
+	{
+		struct bpf_reg_state *reg = &regs[BPF_REG_1];
+		const struct btf_type *type;
+
+		if (reg->type & MEM_RCU) {
+			type = btf_type_by_id(reg->btf, reg->btf_id);
+			if (!type || !btf_type_is_struct(type)) {
+				verbose(env, "Helper has invalid btf/btf_id in R1\n");
+				return -EFAULT;
+			}
+			returns_cpu_specific_alloc_ptr = true;
+			env->insn_aux_data[insn_idx].call_with_percpu_alloc_ptr = true;
+		}
+		break;
+	}
 	case BPF_FUNC_user_ringbuf_drain:
 		err = __check_func_call(env, insn, insn_idx_p, meta.subprogno,
 					set_user_ringbuf_callback_state);
@@ -9858,6 +9913,9 @@ static int check_helper_call(struct bpf_verifier_env *env, struct bpf_insn *insn
 			}
 			regs[BPF_REG_0].type = PTR_TO_MEM | ret_flag;
 			regs[BPF_REG_0].mem_size = tsize;
+		} else {
+			if (returns_cpu_specific_alloc_ptr) {
+				regs[BPF_REG_0].type = PTR_TO_BTF_ID | MEM_ALLOC | MEM_RCU;
 			} else {
 				/* MEM_RDONLY may be carried from ret_flag, but it
 				 * doesn't apply on PTR_TO_BTF_ID. Fold it, otherwise
@@ -9865,8 +9923,9 @@ static int check_helper_call(struct bpf_verifier_env *env, struct bpf_insn *insn
 				 * check_mem_access().
 				 */
 				ret_flag &= ~MEM_RDONLY;
-
 				regs[BPF_REG_0].type = PTR_TO_BTF_ID | ret_flag;
+			}
+
 			regs[BPF_REG_0].btf = meta.ret_btf;
 			regs[BPF_REG_0].btf_id = meta.ret_btf_id;
 		}
@@ -9882,8 +9941,11 @@ static int check_helper_call(struct bpf_verifier_env *env, struct bpf_insn *insn
 		if (func_id == BPF_FUNC_kptr_xchg) {
 			ret_btf = meta.kptr_field->kptr.btf;
 			ret_btf_id = meta.kptr_field->kptr.btf_id;
-			if (!btf_is_kernel(ret_btf))
+			if (!btf_is_kernel(ret_btf)) {
 				regs[BPF_REG_0].type |= MEM_ALLOC;
+				if (meta.kptr_field->type == BPF_KPTR_PERCPU)
+					regs[BPF_REG_0].type |= MEM_PERCPU;
+			}
 		} else {
 			if (fn->ret_btf_id == BPF_PTR_POISON) {
 				verbose(env, "verifier internal error:");
@@ -10268,6 +10330,8 @@ enum special_kfunc_type {
 	KF_bpf_dynptr_slice,
 	KF_bpf_dynptr_slice_rdwr,
 	KF_bpf_dynptr_clone,
+	KF_bpf_percpu_obj_new_impl,
+	KF_bpf_percpu_obj_drop_impl,
 };
 
 BTF_SET_START(special_kfunc_set)
@@ -10288,6 +10352,8 @@ BTF_ID(func, bpf_dynptr_from_xdp)
 BTF_ID(func, bpf_dynptr_slice)
 BTF_ID(func, bpf_dynptr_slice_rdwr)
 BTF_ID(func, bpf_dynptr_clone)
+BTF_ID(func, bpf_percpu_obj_new_impl)
+BTF_ID(func, bpf_percpu_obj_drop_impl)
 BTF_SET_END(special_kfunc_set)
 
 BTF_ID_LIST(special_kfunc_list)
@@ -10310,6 +10376,8 @@ BTF_ID(func, bpf_dynptr_from_xdp)
 BTF_ID(func, bpf_dynptr_slice)
 BTF_ID(func, bpf_dynptr_slice_rdwr)
 BTF_ID(func, bpf_dynptr_clone)
+BTF_ID(func, bpf_percpu_obj_new_impl)
+BTF_ID(func, bpf_percpu_obj_drop_impl)
 
 static bool is_kfunc_ret_null(struct bpf_kfunc_call_arg_meta *meta)
 {
@@ -11004,7 +11072,17 @@ static int check_kfunc_args(struct bpf_verifier_env *env, struct bpf_kfunc_call_
 			}
 			break;
 		case KF_ARG_PTR_TO_ALLOC_BTF_ID:
-			if (reg->type != (PTR_TO_BTF_ID | MEM_ALLOC)) {
+			if (reg->type == (PTR_TO_BTF_ID | MEM_ALLOC)) {
+				if (meta->func_id != special_kfunc_list[KF_bpf_obj_drop_impl]) {
+					verbose(env, "arg#%d expected for bpf_obj_drop_impl()\n", i);
+					return -EINVAL;
+				}
+			} else if (reg->type == (PTR_TO_BTF_ID | MEM_ALLOC | MEM_PERCPU)) {
+				if (meta->func_id != special_kfunc_list[KF_bpf_percpu_obj_drop_impl]) {
+					verbose(env, "arg#%d expected for bpf_percpu_obj_drop_impl()\n", i);
+					return -EINVAL;
+				}
+			} else {
 				verbose(env, "arg#%d expected pointer to allocated object\n", i);
 				return -EINVAL;
 			}
@@ -11012,8 +11090,7 @@ static int check_kfunc_args(struct bpf_verifier_env *env, struct bpf_kfunc_call_
 				verbose(env, "allocated object must be referenced\n");
 				return -EINVAL;
 			}
-			if (meta->btf == btf_vmlinux &&
-			    meta->func_id == special_kfunc_list[KF_bpf_obj_drop_impl]) {
+			if (meta->btf == btf_vmlinux) {
 				meta->arg_btf = reg->btf;
 				meta->arg_btf_id = reg->btf_id;
 			}
@@ -11413,6 +11490,7 @@ static int check_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
 		/* Only exception is bpf_obj_new_impl */
 		if (meta.btf != btf_vmlinux ||
 		    (meta.func_id != special_kfunc_list[KF_bpf_obj_new_impl] &&
+		     meta.func_id != special_kfunc_list[KF_bpf_percpu_obj_new_impl] &&
 		     meta.func_id != special_kfunc_list[KF_bpf_refcount_acquire_impl])) {
 			verbose(env, "acquire kernel function does not return PTR_TO_BTF_ID\n");
 			return -EINVAL;
@@ -11426,11 +11504,16 @@ static int check_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
 		ptr_type = btf_type_skip_modifiers(desc_btf, t->type, &ptr_type_id);
 
 		if (meta.btf == btf_vmlinux && btf_id_set_contains(&special_kfunc_set, meta.func_id)) {
-			if (meta.func_id == special_kfunc_list[KF_bpf_obj_new_impl]) {
+			if (meta.func_id == special_kfunc_list[KF_bpf_obj_new_impl] ||
+			    meta.func_id == special_kfunc_list[KF_bpf_percpu_obj_new_impl]) {
+				struct btf_struct_meta *struct_meta;
 				struct btf *ret_btf;
 				u32 ret_btf_id;
 
-				if (unlikely(!bpf_global_ma_set))
+				if (meta.func_id == special_kfunc_list[KF_bpf_obj_new_impl] && !bpf_global_ma_set)
+					return -ENOMEM;
+
+				if (meta.func_id == special_kfunc_list[KF_bpf_percpu_obj_new_impl] && !bpf_global_percpu_ma_set)
 					return -ENOMEM;
 
 				if (((u64)(u32)meta.arg_constant.value) != meta.arg_constant.value) {
@@ -11443,24 +11526,38 @@ static int check_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
 
 				/* This may be NULL due to user not supplying a BTF */
 				if (!ret_btf) {
-					verbose(env, "bpf_obj_new requires prog BTF\n");
+					verbose(env, "bpf_obj_new/bpf_percpu_obj_new requires prog BTF\n");
 					return -EINVAL;
 				}
 
 				ret_t = btf_type_by_id(ret_btf, ret_btf_id);
 				if (!ret_t || !__btf_type_is_struct(ret_t)) {
-					verbose(env, "bpf_obj_new type ID argument must be of a struct\n");
+					verbose(env, "bpf_obj_new/bpf_percpu_obj_new type ID argument must be of a struct\n");
+					return -EINVAL;
+				}
+
+				struct_meta = btf_find_struct_meta(ret_btf, ret_btf_id);
+				if (meta.func_id == special_kfunc_list[KF_bpf_percpu_obj_new_impl]) {
+					if (!__btf_type_is_scalar_struct(env, ret_btf, ret_t, 0)) {
+						verbose(env, "bpf_percpu_obj_new type ID argument must be of a struct of scalars\n");
+						return -EINVAL;
+					}
+
+					if (struct_meta) {
+						verbose(env, "bpf_percpu_obj_new type ID argument must not contain special fields\n");
 						return -EINVAL;
 					}
+				}
 
 				mark_reg_known_zero(env, regs, BPF_REG_0);
 				regs[BPF_REG_0].type = PTR_TO_BTF_ID | MEM_ALLOC;
 				regs[BPF_REG_0].btf = ret_btf;
 				regs[BPF_REG_0].btf_id = ret_btf_id;
+				if (meta.func_id == special_kfunc_list[KF_bpf_percpu_obj_new_impl])
+					regs[BPF_REG_0].type |= MEM_PERCPU;
 
 				insn_aux->obj_new_size = ret_t->size;
-				insn_aux->kptr_struct_meta =
-					btf_find_struct_meta(ret_btf, ret_btf_id);
+				insn_aux->kptr_struct_meta = struct_meta;
 			} else if (meta.func_id == special_kfunc_list[KF_bpf_refcount_acquire_impl]) {
 				mark_reg_known_zero(env, regs, BPF_REG_0);
 				regs[BPF_REG_0].type = PTR_TO_BTF_ID | MEM_ALLOC;
@@ -11597,7 +11694,8 @@ static int check_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
 			regs[BPF_REG_0].id = ++env->id_gen;
 	} else if (btf_type_is_void(t)) {
 		if (meta.btf == btf_vmlinux && btf_id_set_contains(&special_kfunc_set, meta.func_id)) {
-			if (meta.func_id == special_kfunc_list[KF_bpf_obj_drop_impl]) {
+			if (meta.func_id == special_kfunc_list[KF_bpf_obj_drop_impl] ||
+			    meta.func_id == special_kfunc_list[KF_bpf_percpu_obj_drop_impl]) {
 				insn_aux->kptr_struct_meta =
 					btf_find_struct_meta(meta.arg_btf,
 							     meta.arg_btf_id);
@@ -18266,21 +18364,35 @@ static int fixup_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
 		insn->imm = BPF_CALL_IMM(desc->addr);
 	if (insn->off)
 		return 0;
-	if (desc->func_id == special_kfunc_list[KF_bpf_obj_new_impl]) {
+	if (desc->func_id == special_kfunc_list[KF_bpf_obj_new_impl] ||
+	    desc->func_id == special_kfunc_list[KF_bpf_percpu_obj_new_impl]) {
 		struct btf_struct_meta *kptr_struct_meta = env->insn_aux_data[insn_idx].kptr_struct_meta;
 		struct bpf_insn addr[2] = { BPF_LD_IMM64(BPF_REG_2, (long)kptr_struct_meta) };
 		u64 obj_new_size = env->insn_aux_data[insn_idx].obj_new_size;
 
+		if (desc->func_id == special_kfunc_list[KF_bpf_percpu_obj_new_impl] && kptr_struct_meta) {
+			verbose(env, "verifier internal error: NULL kptr_struct_meta expected at insn_idx %d\n",
+				insn_idx);
+			return -EFAULT;
+		}
+
 		insn_buf[0] = BPF_MOV64_IMM(BPF_REG_1, obj_new_size);
 		insn_buf[1] = addr[0];
 		insn_buf[2] = addr[1];
 		insn_buf[3] = *insn;
 		*cnt = 4;
 	} else if (desc->func_id == special_kfunc_list[KF_bpf_obj_drop_impl] ||
+		   desc->func_id == special_kfunc_list[KF_bpf_percpu_obj_drop_impl] ||
 		   desc->func_id == special_kfunc_list[KF_bpf_refcount_acquire_impl]) {
 		struct btf_struct_meta *kptr_struct_meta = env->insn_aux_data[insn_idx].kptr_struct_meta;
 		struct bpf_insn addr[2] = { BPF_LD_IMM64(BPF_REG_2, (long)kptr_struct_meta) };
 
+		if (desc->func_id == special_kfunc_list[KF_bpf_percpu_obj_drop_impl] && kptr_struct_meta) {
+			verbose(env, "verifier internal error: NULL kptr_struct_meta expected at insn_idx %d\n",
+				insn_idx);
+			return -EFAULT;
+		}
+
 		if (desc->func_id == special_kfunc_list[KF_bpf_refcount_acquire_impl] &&
 		    !kptr_struct_meta) {
 			verbose(env, "verifier internal error: kptr_struct_meta expected at insn_idx %d\n",
@@ -18608,6 +18720,25 @@ static int do_misc_fixups(struct bpf_verifier_env *env)
 			goto patch_call_imm;
 		}
 
+		/* bpf_per_cpu_ptr() and bpf_this_cpu_ptr() */
+		if (env->insn_aux_data[i + delta].call_with_percpu_alloc_ptr) {
+			/* patch with 'r1 = *(u64 *)(r1 + 0)' since for percpu data,
+			 * bpf_mem_alloc() returns a ptr to the percpu data ptr.
+			 */
+			insn_buf[0] = BPF_LDX_MEM(BPF_DW, BPF_REG_1, BPF_REG_1, 0);
+			insn_buf[1] = *insn;
+			cnt = 2;
+
+			new_prog = bpf_patch_insn_data(env, i + delta, insn_buf, cnt);
+			if (!new_prog)
+				return -ENOMEM;
+
+			delta += cnt - 1;
+			env->prog = prog = new_prog;
+			insn = new_prog->insnsi + i + delta;
+			goto patch_call_imm;
+		}
+
 		/* BPF_EMIT_CALL() assumptions in some of the map_gen_lookup
 		 * and other inlining handlers are currently limited to 64 bit
 		 * only.

tools/include/uapi/linux/bpf.h

@@ -932,7 +932,14 @@ enum bpf_map_type {
 	 */
 	BPF_MAP_TYPE_CGROUP_STORAGE = BPF_MAP_TYPE_CGROUP_STORAGE_DEPRECATED,
 	BPF_MAP_TYPE_REUSEPORT_SOCKARRAY,
-	BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE,
+	BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE_DEPRECATED,
+	/* BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE is available to bpf programs
+	 * attaching to a cgroup. The new mechanism (BPF_MAP_TYPE_CGRP_STORAGE +
+	 * local percpu kptr) supports all BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE
+	 * functionality and more. So mark * BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE
+	 * deprecated.
+	 */
+	BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE = BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE_DEPRECATED,
 	BPF_MAP_TYPE_QUEUE,
 	BPF_MAP_TYPE_STACK,
 	BPF_MAP_TYPE_SK_STORAGE,

tools/lib/bpf/bpf_helpers.h

@@ -181,6 +181,7 @@ enum libbpf_tristate {
 #define __ksym __attribute__((section(".ksyms")))
 #define __kptr_untrusted __attribute__((btf_type_tag("kptr_untrusted")))
 #define __kptr __attribute__((btf_type_tag("kptr")))
+#define __percpu_kptr __attribute__((btf_type_tag("percpu_kptr")))
 
 #define bpf_ksym_exists(sym) ({									\
 	_Static_assert(!__builtin_constant_p(!!sym), #sym " should be marked as __weak");	\

tools/testing/selftests/bpf/bpf_experimental.h

@@ -131,4 +131,35 @@ extern int bpf_rbtree_add_impl(struct bpf_rb_root *root, struct bpf_rb_node *nod
  */
 extern struct bpf_rb_node *bpf_rbtree_first(struct bpf_rb_root *root) __ksym;
 
+/* Description
+ *	Allocates a percpu object of the type represented by 'local_type_id' in
+ *	program BTF. User may use the bpf_core_type_id_local macro to pass the
+ *	type ID of a struct in program BTF.
+ *
+ *	The 'local_type_id' parameter must be a known constant.
+ *	The 'meta' parameter is rewritten by the verifier, no need for BPF
+ *	program to set it.
+ * Returns
+ *	A pointer to a percpu object of the type corresponding to the passed in
+ *	'local_type_id', or NULL on failure.
+ */
+extern void *bpf_percpu_obj_new_impl(__u64 local_type_id, void *meta) __ksym;
+
+/* Convenience macro to wrap over bpf_percpu_obj_new_impl */
+#define bpf_percpu_obj_new(type) ((type __percpu_kptr *)bpf_percpu_obj_new_impl(bpf_core_type_id_local(type), NULL))
+
+/* Description
+ *	Free an allocated percpu object. All fields of the object that require
+ *	destruction will be destructed before the storage is freed.
+ *
+ *	The 'meta' parameter is rewritten by the verifier, no need for BPF
+ *	program to set it.
+ * Returns
+ *	Void.
+ */
+extern void bpf_percpu_obj_drop_impl(void *kptr, void *meta) __ksym;
+
+/* Convenience macro to wrap over bpf_obj_drop_impl */
+#define bpf_percpu_obj_drop(kptr) bpf_percpu_obj_drop_impl(kptr, NULL)
+
 #endif

tools/testing/selftests/bpf/prog_tests/libbpf_str.c

@@ -142,10 +142,14 @@ static void test_libbpf_bpf_map_type_str(void)
 		/* Special case for map_type_name BPF_MAP_TYPE_CGROUP_STORAGE_DEPRECATED
 		 * where it and BPF_MAP_TYPE_CGROUP_STORAGE have the same enum value
 		 * (map_type). For this enum value, libbpf_bpf_map_type_str() picks
-		 * BPF_MAP_TYPE_CGROUP_STORAGE.
+		 * BPF_MAP_TYPE_CGROUP_STORAGE. The same for
+		 * BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE_DEPRECATED and
+		 * BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE.
 		 */
 		if (strcmp(map_type_name, "BPF_MAP_TYPE_CGROUP_STORAGE_DEPRECATED") == 0)
 			continue;
+		if (strcmp(map_type_name, "BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE_DEPRECATED") == 0)
+			continue;
 
 		ASSERT_STREQ(buf, map_type_name, "exp_str_value");
 	}

tools/testing/selftests/bpf/prog_tests/linked_list.c

@@ -65,8 +65,8 @@ static struct {
 	{ "map_compat_raw_tp", "tracing progs cannot use bpf_{list_head,rb_root} yet" },
 	{ "map_compat_raw_tp_w", "tracing progs cannot use bpf_{list_head,rb_root} yet" },
 	{ "obj_type_id_oor", "local type ID argument must be in range [0, U32_MAX]" },
-	{ "obj_new_no_composite", "bpf_obj_new type ID argument must be of a struct" },
-	{ "obj_new_no_struct", "bpf_obj_new type ID argument must be of a struct" },
+	{ "obj_new_no_composite", "bpf_obj_new/bpf_percpu_obj_new type ID argument must be of a struct" },
+	{ "obj_new_no_struct", "bpf_obj_new/bpf_percpu_obj_new type ID argument must be of a struct" },
 	{ "obj_drop_non_zero_off", "R1 must have zero offset when passed to release func" },
 	{ "new_null_ret", "R0 invalid mem access 'ptr_or_null_'" },
 	{ "obj_new_acq", "Unreleased reference id=" },

tools/testing/selftests/bpf/prog_tests/percpu_alloc.c

+// SPDX-License-Identifier: GPL-2.0
+#include <test_progs.h>
+#include "percpu_alloc_array.skel.h"
+#include "percpu_alloc_cgrp_local_storage.skel.h"
+#include "percpu_alloc_fail.skel.h"
+
+static void test_array(void)
+{
+	struct percpu_alloc_array *skel;
+	int err, prog_fd;
+	LIBBPF_OPTS(bpf_test_run_opts, topts);
+
+	skel = percpu_alloc_array__open();
+	if (!ASSERT_OK_PTR(skel, "percpu_alloc_array__open"))
+		return;
+
+	bpf_program__set_autoload(skel->progs.test_array_map_1, true);
+	bpf_program__set_autoload(skel->progs.test_array_map_2, true);
+	bpf_program__set_autoload(skel->progs.test_array_map_3, true);
+	bpf_program__set_autoload(skel->progs.test_array_map_4, true);
+
+	skel->rodata->nr_cpus = libbpf_num_possible_cpus();
+
+	err = percpu_alloc_array__load(skel);
+	if (!ASSERT_OK(err, "percpu_alloc_array__load"))
+		goto out;
+
+	err = percpu_alloc_array__attach(skel);
+	if (!ASSERT_OK(err, "percpu_alloc_array__attach"))
+		goto out;
+
+	prog_fd = bpf_program__fd(skel->progs.test_array_map_1);
+	err = bpf_prog_test_run_opts(prog_fd, &topts);
+	ASSERT_OK(err, "test_run array_map 1-4");
+	ASSERT_EQ(topts.retval, 0, "test_run array_map 1-4");
+	ASSERT_EQ(skel->bss->cpu0_field_d, 2, "cpu0_field_d");
+	ASSERT_EQ(skel->bss->sum_field_c, 1, "sum_field_c");
+out:
+	percpu_alloc_array__destroy(skel);
+}
+
+static void test_array_sleepable(void)
+{
+	struct percpu_alloc_array *skel;
+	int err, prog_fd;
+	LIBBPF_OPTS(bpf_test_run_opts, topts);
+
+	skel = percpu_alloc_array__open();
+	if (!ASSERT_OK_PTR(skel, "percpu_alloc__open"))
+		return;
+
+	bpf_program__set_autoload(skel->progs.test_array_map_10, true);
+
+	skel->rodata->nr_cpus = libbpf_num_possible_cpus();
+
+	err = percpu_alloc_array__load(skel);
+	if (!ASSERT_OK(err, "percpu_alloc_array__load"))
+		goto out;
+
+	err = percpu_alloc_array__attach(skel);
+	if (!ASSERT_OK(err, "percpu_alloc_array__attach"))
+		goto out;
+
+	prog_fd = bpf_program__fd(skel->progs.test_array_map_10);
+	err = bpf_prog_test_run_opts(prog_fd, &topts);
+	ASSERT_OK(err, "test_run array_map_10");
+	ASSERT_EQ(topts.retval, 0, "test_run array_map_10");
+	ASSERT_EQ(skel->bss->cpu0_field_d, 2, "cpu0_field_d");
+	ASSERT_EQ(skel->bss->sum_field_c, 1, "sum_field_c");
+out:
+	percpu_alloc_array__destroy(skel);
+}
+
+static void test_cgrp_local_storage(void)
+{
+	struct percpu_alloc_cgrp_local_storage *skel;
+	int err, cgroup_fd, prog_fd;
+	LIBBPF_OPTS(bpf_test_run_opts, topts);
+
+	cgroup_fd = test__join_cgroup("/percpu_alloc");
+	if (!ASSERT_GE(cgroup_fd, 0, "join_cgroup /percpu_alloc"))
+		return;
+
+	skel = percpu_alloc_cgrp_local_storage__open();
+	if (!ASSERT_OK_PTR(skel, "percpu_alloc_cgrp_local_storage__open"))
+		goto close_fd;
+
+	skel->rodata->nr_cpus = libbpf_num_possible_cpus();
+
+	err = percpu_alloc_cgrp_local_storage__load(skel);
+	if (!ASSERT_OK(err, "percpu_alloc_cgrp_local_storage__load"))
+		goto destroy_skel;
+
+	err = percpu_alloc_cgrp_local_storage__attach(skel);
+	if (!ASSERT_OK(err, "percpu_alloc_cgrp_local_storage__attach"))
+		goto destroy_skel;
+
+	prog_fd = bpf_program__fd(skel->progs.test_cgrp_local_storage_1);
+	err = bpf_prog_test_run_opts(prog_fd, &topts);
+	ASSERT_OK(err, "test_run cgrp_local_storage 1-3");
+	ASSERT_EQ(topts.retval, 0, "test_run cgrp_local_storage 1-3");
+	ASSERT_EQ(skel->bss->cpu0_field_d, 2, "cpu0_field_d");
+	ASSERT_EQ(skel->bss->sum_field_c, 1, "sum_field_c");
+
+destroy_skel:
+	percpu_alloc_cgrp_local_storage__destroy(skel);
+close_fd:
+	close(cgroup_fd);
+}
+
+static void test_failure(void) {
+	RUN_TESTS(percpu_alloc_fail);
+}
+
+void test_percpu_alloc(void)
+{
+	if (test__start_subtest("array"))
+		test_array();
+	if (test__start_subtest("array_sleepable"))
+		test_array_sleepable();
+	if (test__start_subtest("cgrp_local_storage"))
+		test_cgrp_local_storage();
+	if (test__start_subtest("failure_tests"))
+		test_failure();
+}

tools/testing/selftests/bpf/progs/percpu_alloc_array.c

+#include "bpf_experimental.h"
+
+struct val_t {
+	long b, c, d;
+};
+
+struct elem {
+	long sum;
+	struct val_t __percpu_kptr *pc;
+};
+
+struct {
+	__uint(type, BPF_MAP_TYPE_ARRAY);
+	__uint(max_entries, 1);
+	__type(key, int);
+	__type(value, struct elem);
+} array SEC(".maps");
+
+void bpf_rcu_read_lock(void) __ksym;
+void bpf_rcu_read_unlock(void) __ksym;
+
+const volatile int nr_cpus;
+
+/* Initialize the percpu object */
+SEC("?fentry/bpf_fentry_test1")
+int BPF_PROG(test_array_map_1)
+{
+	struct val_t __percpu_kptr *p;
+	struct elem *e;
+	int index = 0;
+
+	e = bpf_map_lookup_elem(&array, &index);
+	if (!e)
+		return 0;
+
+	p = bpf_percpu_obj_new(struct val_t);
+	if (!p)
+		return 0;
+
+	p = bpf_kptr_xchg(&e->pc, p);
+	if (p)
+		bpf_percpu_obj_drop(p);
+
+	return 0;
+}
+
+/* Update percpu data */
+SEC("?fentry/bpf_fentry_test2")
+int BPF_PROG(test_array_map_2)
+{
+	struct val_t __percpu_kptr *p;
+	struct val_t *v;
+	struct elem *e;
+	int index = 0;
+
+	e = bpf_map_lookup_elem(&array, &index);
+	if (!e)
+		return 0;
+
+	p = e->pc;
+	if (!p)
+		return 0;
+
+	v = bpf_per_cpu_ptr(p, 0);
+	if (!v)
+		return 0;
+	v->c = 1;
+	v->d = 2;
+
+	return 0;
+}
+
+int cpu0_field_d, sum_field_c;
+
+/* Summarize percpu data */
+SEC("?fentry/bpf_fentry_test3")
+int BPF_PROG(test_array_map_3)
+{
+	struct val_t __percpu_kptr *p;
+	int i, index = 0;
+	struct val_t *v;
+	struct elem *e;
+
+	e = bpf_map_lookup_elem(&array, &index);
+	if (!e)
+		return 0;
+
+	p = e->pc;
+	if (!p)
+		return 0;
+
+	bpf_for(i, 0, nr_cpus) {
+		v = bpf_per_cpu_ptr(p, i);
+		if (v) {
+			if (i == 0)
+				cpu0_field_d = v->d;
+			sum_field_c += v->c;
+		}
+	}
+
+	return 0;
+}
+
+/* Explicitly free allocated percpu data */
+SEC("?fentry/bpf_fentry_test4")
+int BPF_PROG(test_array_map_4)
+{
+	struct val_t __percpu_kptr *p;
+	struct elem *e;
+	int index = 0;
+
+	e = bpf_map_lookup_elem(&array, &index);
+	if (!e)
+		return 0;
+
+	/* delete */
+	p = bpf_kptr_xchg(&e->pc, NULL);
+	if (p) {
+		bpf_percpu_obj_drop(p);
+	}
+
+	return 0;
+}
+
+SEC("?fentry.s/bpf_fentry_test1")
+int BPF_PROG(test_array_map_10)
+{
+	struct val_t __percpu_kptr *p, *p1;
+	int i, index = 0;
+	struct val_t *v;
+	struct elem *e;
+
+	e = bpf_map_lookup_elem(&array, &index);
+	if (!e)
+		return 0;
+
+	bpf_rcu_read_lock();
+	p = e->pc;
+	if (!p) {
+		p = bpf_percpu_obj_new(struct val_t);
+		if (!p)
+			goto out;
+
+		p1 = bpf_kptr_xchg(&e->pc, p);
+		if (p1) {
+			/* race condition */
+			bpf_percpu_obj_drop(p1);
+		}
+	}
+
+	v = bpf_this_cpu_ptr(p);
+	v->c = 3;
+	v = bpf_this_cpu_ptr(p);
+	v->c = 0;
+
+	v = bpf_per_cpu_ptr(p, 0);
+	if (!v)
+		goto out;
+	v->c = 1;
+	v->d = 2;
+
+	/* delete */
+	p1 = bpf_kptr_xchg(&e->pc, NULL);
+	if (!p1)
+		goto out;
+
+	bpf_for(i, 0, nr_cpus) {
+		v = bpf_per_cpu_ptr(p, i);
+		if (v) {
+			if (i == 0)
+				cpu0_field_d = v->d;
+			sum_field_c += v->c;
+		}
+	}
+
+	/* finally release p */
+	bpf_percpu_obj_drop(p1);
+out:
+	bpf_rcu_read_unlock();
+	return 0;
+}
+
+char _license[] SEC("license") = "GPL";

tools/testing/selftests/bpf/progs/percpu_alloc_cgrp_local_storage.c

+#include "bpf_experimental.h"
+
+struct val_t {
+	long b, c, d;
+};
+
+struct elem {
+	long sum;
+	struct val_t __percpu_kptr *pc;
+};
+
+struct {
+	__uint(type, BPF_MAP_TYPE_CGRP_STORAGE);
+	__uint(map_flags, BPF_F_NO_PREALLOC);
+	__type(key, int);
+	__type(value, struct elem);
+} cgrp SEC(".maps");
+
+const volatile int nr_cpus;
+
+/* Initialize the percpu object */
+SEC("fentry/bpf_fentry_test1")
+int BPF_PROG(test_cgrp_local_storage_1)
+{
+	struct task_struct *task;
+	struct val_t __percpu_kptr *p;
+	struct elem *e;
+
+	task = bpf_get_current_task_btf();
+	e = bpf_cgrp_storage_get(&cgrp, task->cgroups->dfl_cgrp, 0,
+				 BPF_LOCAL_STORAGE_GET_F_CREATE);
+	if (!e)
+		return 0;
+
+	p = bpf_percpu_obj_new(struct val_t);
+	if (!p)
+		return 0;
+
+	p = bpf_kptr_xchg(&e->pc, p);
+	if (p)
+		bpf_percpu_obj_drop(p);
+
+	return 0;
+}
+
+/* Percpu data collection */
+SEC("fentry/bpf_fentry_test2")
+int BPF_PROG(test_cgrp_local_storage_2)
+{
+	struct task_struct *task;
+	struct val_t __percpu_kptr *p;
+	struct val_t *v;
+	struct elem *e;
+
+	task = bpf_get_current_task_btf();
+	e = bpf_cgrp_storage_get(&cgrp, task->cgroups->dfl_cgrp, 0, 0);
+	if (!e)
+		return 0;
+
+	p = e->pc;
+	if (!p)
+		return 0;
+
+	v = bpf_per_cpu_ptr(p, 0);
+	if (!v)
+		return 0;
+	v->c = 1;
+	v->d = 2;
+	return 0;
+}
+
+int cpu0_field_d, sum_field_c;
+
+/* Summarize percpu data collection */
+SEC("fentry/bpf_fentry_test3")
+int BPF_PROG(test_cgrp_local_storage_3)
+{
+	struct task_struct *task;
+	struct val_t __percpu_kptr *p;
+	struct val_t *v;
+	struct elem *e;
+	int i;
+
+	task = bpf_get_current_task_btf();
+	e = bpf_cgrp_storage_get(&cgrp, task->cgroups->dfl_cgrp, 0, 0);
+	if (!e)
+		return 0;
+
+	p = e->pc;
+	if (!p)
+		return 0;
+
+	bpf_for(i, 0, nr_cpus) {
+		v = bpf_per_cpu_ptr(p, i);
+		if (v) {
+			if (i == 0)
+				cpu0_field_d = v->d;
+			sum_field_c += v->c;
+		}
+	}
+
+	return 0;
+}
+
+char _license[] SEC("license") = "GPL";

tools/testing/selftests/bpf/progs/percpu_alloc_fail.c

+#include "bpf_experimental.h"
+#include "bpf_misc.h"
+
+struct val_t {
+	long b, c, d;
+};
+
+struct val2_t {
+	long b;
+};
+
+struct val_with_ptr_t {
+	char *p;
+};
+
+struct val_with_rb_root_t {
+	struct bpf_spin_lock lock;
+};
+
+struct elem {
+	long sum;
+	struct val_t __percpu_kptr *pc;
+};
+
+struct {
+	__uint(type, BPF_MAP_TYPE_ARRAY);
+	__uint(max_entries, 1);
+	__type(key, int);
+	__type(value, struct elem);
+} array SEC(".maps");
+
+long ret;
+
+SEC("?fentry/bpf_fentry_test1")
+__failure __msg("store to referenced kptr disallowed")
+int BPF_PROG(test_array_map_1)
+{
+	struct val_t __percpu_kptr *p;
+	struct elem *e;
+	int index = 0;
+
+	e = bpf_map_lookup_elem(&array, &index);
+	if (!e)
+		return 0;
+
+	p = bpf_percpu_obj_new(struct val_t);
+	if (!p)
+		return 0;
+
+	p = bpf_kptr_xchg(&e->pc, p);
+	if (p)
+		bpf_percpu_obj_drop(p);
+
+	e->pc = (struct val_t __percpu_kptr *)ret;
+	return 0;
+}
+
+SEC("?fentry/bpf_fentry_test1")
+__failure __msg("invalid kptr access, R2 type=percpu_ptr_val2_t expected=ptr_val_t")
+int BPF_PROG(test_array_map_2)
+{
+	struct val2_t __percpu_kptr *p2;
+	struct val_t __percpu_kptr *p;
+	struct elem *e;
+	int index = 0;
+
+	e = bpf_map_lookup_elem(&array, &index);
+	if (!e)
+		return 0;
+
+	p2 = bpf_percpu_obj_new(struct val2_t);
+	if (!p2)
+		return 0;
+
+	p = bpf_kptr_xchg(&e->pc, p2);
+	if (p)
+		bpf_percpu_obj_drop(p);
+
+	return 0;
+}
+
+SEC("?fentry.s/bpf_fentry_test1")
+__failure __msg("R1 type=scalar expected=percpu_ptr_, percpu_rcu_ptr_, percpu_trusted_ptr_")
+int BPF_PROG(test_array_map_3)
+{
+	struct val_t __percpu_kptr *p, *p1;
+	struct val_t *v;
+	struct elem *e;
+	int index = 0;
+
+	e = bpf_map_lookup_elem(&array, &index);
+	if (!e)
+		return 0;
+
+	p = bpf_percpu_obj_new(struct val_t);
+	if (!p)
+		return 0;
+
+	p1 = bpf_kptr_xchg(&e->pc, p);
+	if (p1)
+		bpf_percpu_obj_drop(p1);
+
+	v = bpf_this_cpu_ptr(p);
+	ret = v->b;
+	return 0;
+}
+
+SEC("?fentry.s/bpf_fentry_test1")
+__failure __msg("arg#0 expected for bpf_percpu_obj_drop_impl()")
+int BPF_PROG(test_array_map_4)
+{
+	struct val_t __percpu_kptr *p;
+
+	p = bpf_percpu_obj_new(struct val_t);
+	if (!p)
+		return 0;
+
+	bpf_obj_drop(p);
+	return 0;
+}
+
+SEC("?fentry.s/bpf_fentry_test1")
+__failure __msg("arg#0 expected for bpf_obj_drop_impl()")
+int BPF_PROG(test_array_map_5)
+{
+	struct val_t *p;
+
+	p = bpf_obj_new(struct val_t);
+	if (!p)
+		return 0;
+
+	bpf_percpu_obj_drop(p);
+	return 0;
+}
+
+SEC("?fentry.s/bpf_fentry_test1")
+__failure __msg("bpf_percpu_obj_new type ID argument must be of a struct of scalars")
+int BPF_PROG(test_array_map_6)
+{
+	struct val_with_ptr_t __percpu_kptr *p;
+
+	p = bpf_percpu_obj_new(struct val_with_ptr_t);
+	if (!p)
+		return 0;
+
+	bpf_percpu_obj_drop(p);
+	return 0;
+}
+
+SEC("?fentry.s/bpf_fentry_test1")
+__failure __msg("bpf_percpu_obj_new type ID argument must not contain special fields")
+int BPF_PROG(test_array_map_7)
+{
+	struct val_with_rb_root_t __percpu_kptr *p;
+
+	p = bpf_percpu_obj_new(struct val_with_rb_root_t);
+	if (!p)
+		return 0;
+
+	bpf_percpu_obj_drop(p);
+	return 0;
+}
+
+char _license[] SEC("license") = "GPL";

tools/testing/selftests/bpf/test_bpftool_synctypes.py

@@ -509,6 +509,15 @@ def main():
     source_map_types.remove('cgroup_storage_deprecated')
     source_map_types.add('cgroup_storage')
 
+    # The same applied to BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE_DEPRECATED and
+    # BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE which share the same enum value
+    # and source_map_types picks
+    # BPF_MAP_TYPE_PERCPU_CGROUP_STORAGE_DEPRECATED/percpu_cgroup_storage_deprecated.
+    # Replace 'percpu_cgroup_storage_deprecated' with 'percpu_cgroup_storage'
+    # so it aligns with what `bpftool map help` shows.
+    source_map_types.remove('percpu_cgroup_storage_deprecated')
+    source_map_types.add('percpu_cgroup_storage')
+
     help_map_types = map_info.get_map_help()
     help_map_options = map_info.get_options()
     map_info.close()