Merge branch 'Introduce access remote cpu elem support in BPF percpu map'

Feng zhou says:

====================

From: Feng Zhou <zhoufeng.zf@bytedance.com>

Trace some functions, such as enqueue_task_fair, need to access the
corresponding cpu, not the current cpu, and bpf_map_lookup_elem percpu map
cannot do it. So add bpf_map_lookup_percpu_elem to accomplish it for
percpu_array_map, percpu_hash_map, lru_percpu_hash_map.

v1->v2: Addressed comments from Alexei Starovoitov.
- add a selftest for bpf_map_lookup_percpu_elem.
====================

Signed-off-by: Alexei Starovoitov <ast@kernel.org>

include/linux/bpf.h

@@ -89,6 +89,7 @@ struct bpf_map_ops {
 	int (*map_push_elem)(struct bpf_map *map, void *value, u64 flags);
 	int (*map_pop_elem)(struct bpf_map *map, void *value);
 	int (*map_peek_elem)(struct bpf_map *map, void *value);
+	void *(*map_lookup_percpu_elem)(struct bpf_map *map, void *key, u32 cpu);
 
 	/* funcs called by prog_array and perf_event_array map */
 	void *(*map_fd_get_ptr)(struct bpf_map *map, struct file *map_file,
@@ -2184,6 +2185,7 @@ extern const struct bpf_func_proto bpf_map_delete_elem_proto;
 extern const struct bpf_func_proto bpf_map_push_elem_proto;
 extern const struct bpf_func_proto bpf_map_pop_elem_proto;
 extern const struct bpf_func_proto bpf_map_peek_elem_proto;
+extern const struct bpf_func_proto bpf_map_lookup_percpu_elem_proto;
 
 extern const struct bpf_func_proto bpf_get_prandom_u32_proto;
 extern const struct bpf_func_proto bpf_get_smp_processor_id_proto;

include/uapi/linux/bpf.h

@@ -5164,6 +5164,14 @@ union bpf_attr {
  *		if not NULL, is a reference which must be released using its
  *		corresponding release function, or moved into a BPF map before
  *		program exit.
+ *
+ * void *bpf_map_lookup_percpu_elem(struct bpf_map *map, const void *key, u32 cpu)
+ * 	Description
+ * 		Perform a lookup in *percpu map* for an entry associated to
+ * 		*key* on *cpu*.
+ * 	Return
+ * 		Map value associated to *key* on *cpu*, or **NULL** if no entry
+ * 		was found or *cpu* is invalid.
  */
 #define __BPF_FUNC_MAPPER(FN)		\
 	FN(unspec),			\
@@ -5361,6 +5369,7 @@ union bpf_attr {
 	FN(skb_set_tstamp),		\
 	FN(ima_file_hash),		\
 	FN(kptr_xchg),			\
+	FN(map_lookup_percpu_elem),     \
 	/* */
 
 /* integer value in 'imm' field of BPF_CALL instruction selects which helper

kernel/bpf/arraymap.c

@@ -243,6 +243,20 @@ static void *percpu_array_map_lookup_elem(struct bpf_map *map, void *key)
 	return this_cpu_ptr(array->pptrs[index & array->index_mask]);
 }
 
+static void *percpu_array_map_lookup_percpu_elem(struct bpf_map *map, void *key, u32 cpu)
+{
+	struct bpf_array *array = container_of(map, struct bpf_array, map);
+	u32 index = *(u32 *)key;
+
+	if (cpu >= nr_cpu_ids)
+		return NULL;
+
+	if (unlikely(index >= array->map.max_entries))
+		return NULL;
+
+	return per_cpu_ptr(array->pptrs[index & array->index_mask], cpu);
+}
+
 int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value)
 {
 	struct bpf_array *array = container_of(map, struct bpf_array, map);
@@ -725,6 +739,7 @@ const struct bpf_map_ops percpu_array_map_ops = {
 	.map_lookup_elem = percpu_array_map_lookup_elem,
 	.map_update_elem = array_map_update_elem,
 	.map_delete_elem = array_map_delete_elem,
+	.map_lookup_percpu_elem = percpu_array_map_lookup_percpu_elem,
 	.map_seq_show_elem = percpu_array_map_seq_show_elem,
 	.map_check_btf = array_map_check_btf,
 	.map_lookup_batch = generic_map_lookup_batch,

kernel/bpf/core.c

@@ -2619,6 +2619,7 @@ const struct bpf_func_proto bpf_map_delete_elem_proto __weak;
 const struct bpf_func_proto bpf_map_push_elem_proto __weak;
 const struct bpf_func_proto bpf_map_pop_elem_proto __weak;
 const struct bpf_func_proto bpf_map_peek_elem_proto __weak;
+const struct bpf_func_proto bpf_map_lookup_percpu_elem_proto __weak;
 const struct bpf_func_proto bpf_spin_lock_proto __weak;
 const struct bpf_func_proto bpf_spin_unlock_proto __weak;
 const struct bpf_func_proto bpf_jiffies64_proto __weak;

kernel/bpf/hashtab.c

@@ -2199,6 +2199,20 @@ static void *htab_percpu_map_lookup_elem(struct bpf_map *map, void *key)
 		return NULL;
 }
 
+static void *htab_percpu_map_lookup_percpu_elem(struct bpf_map *map, void *key, u32 cpu)
+{
+	struct htab_elem *l;
+
+	if (cpu >= nr_cpu_ids)
+		return NULL;
+
+	l = __htab_map_lookup_elem(map, key);
+	if (l)
+		return per_cpu_ptr(htab_elem_get_ptr(l, map->key_size), cpu);
+	else
+		return NULL;
+}
+
 static void *htab_lru_percpu_map_lookup_elem(struct bpf_map *map, void *key)
 {
 	struct htab_elem *l = __htab_map_lookup_elem(map, key);
@@ -2211,6 +2225,22 @@ static void *htab_lru_percpu_map_lookup_elem(struct bpf_map *map, void *key)
 	return NULL;
 }
 
+static void *htab_lru_percpu_map_lookup_percpu_elem(struct bpf_map *map, void *key, u32 cpu)
+{
+	struct htab_elem *l;
+
+	if (cpu >= nr_cpu_ids)
+		return NULL;
+
+	l = __htab_map_lookup_elem(map, key);
+	if (l) {
+		bpf_lru_node_set_ref(&l->lru_node);
+		return per_cpu_ptr(htab_elem_get_ptr(l, map->key_size), cpu);
+	}
+
+	return NULL;
+}
+
 int bpf_percpu_hash_copy(struct bpf_map *map, void *key, void *value)
 {
 	struct htab_elem *l;
@@ -2300,6 +2330,7 @@ const struct bpf_map_ops htab_percpu_map_ops = {
 	.map_lookup_and_delete_elem = htab_percpu_map_lookup_and_delete_elem,
 	.map_update_elem = htab_percpu_map_update_elem,
 	.map_delete_elem = htab_map_delete_elem,
+	.map_lookup_percpu_elem = htab_percpu_map_lookup_percpu_elem,
 	.map_seq_show_elem = htab_percpu_map_seq_show_elem,
 	.map_set_for_each_callback_args = map_set_for_each_callback_args,
 	.map_for_each_callback = bpf_for_each_hash_elem,
@@ -2318,6 +2349,7 @@ const struct bpf_map_ops htab_lru_percpu_map_ops = {
 	.map_lookup_and_delete_elem = htab_lru_percpu_map_lookup_and_delete_elem,
 	.map_update_elem = htab_lru_percpu_map_update_elem,
 	.map_delete_elem = htab_lru_map_delete_elem,
+	.map_lookup_percpu_elem = htab_lru_percpu_map_lookup_percpu_elem,
 	.map_seq_show_elem = htab_percpu_map_seq_show_elem,
 	.map_set_for_each_callback_args = map_set_for_each_callback_args,
 	.map_for_each_callback = bpf_for_each_hash_elem,

kernel/bpf/helpers.c

@@ -119,6 +119,22 @@ const struct bpf_func_proto bpf_map_peek_elem_proto = {
 	.arg2_type	= ARG_PTR_TO_UNINIT_MAP_VALUE,
 };
 
+BPF_CALL_3(bpf_map_lookup_percpu_elem, struct bpf_map *, map, void *, key, u32, cpu)
+{
+	WARN_ON_ONCE(!rcu_read_lock_held() && !rcu_read_lock_bh_held());
+	return (unsigned long) map->ops->map_lookup_percpu_elem(map, key, cpu);
+}
+
+const struct bpf_func_proto bpf_map_lookup_percpu_elem_proto = {
+	.func		= bpf_map_lookup_percpu_elem,
+	.gpl_only	= false,
+	.pkt_access	= true,
+	.ret_type	= RET_PTR_TO_MAP_VALUE_OR_NULL,
+	.arg1_type	= ARG_CONST_MAP_PTR,
+	.arg2_type	= ARG_PTR_TO_MAP_KEY,
+	.arg3_type	= ARG_ANYTHING,
+};
+
 const struct bpf_func_proto bpf_get_prandom_u32_proto = {
 	.func		= bpf_user_rnd_u32,
 	.gpl_only	= false,
@@ -1420,6 +1436,8 @@ bpf_base_func_proto(enum bpf_func_id func_id)
 		return &bpf_map_pop_elem_proto;
 	case BPF_FUNC_map_peek_elem:
 		return &bpf_map_peek_elem_proto;
+	case BPF_FUNC_map_lookup_percpu_elem:
+		return &bpf_map_lookup_percpu_elem_proto;
 	case BPF_FUNC_get_prandom_u32:
 		return &bpf_get_prandom_u32_proto;
 	case BPF_FUNC_get_smp_processor_id:

kernel/bpf/verifier.c

@@ -6137,6 +6137,12 @@ static int check_map_func_compatibility(struct bpf_verifier_env *env,
 		    map->map_type != BPF_MAP_TYPE_BLOOM_FILTER)
 			goto error;
 		break;
+	case BPF_FUNC_map_lookup_percpu_elem:
+		if (map->map_type != BPF_MAP_TYPE_PERCPU_ARRAY &&
+		    map->map_type != BPF_MAP_TYPE_PERCPU_HASH &&
+		    map->map_type != BPF_MAP_TYPE_LRU_PERCPU_HASH)
+			goto error;
+		break;
 	case BPF_FUNC_sk_storage_get:
 	case BPF_FUNC_sk_storage_delete:
 		if (map->map_type != BPF_MAP_TYPE_SK_STORAGE)
@@ -6750,7 +6756,8 @@ record_func_map(struct bpf_verifier_env *env, struct bpf_call_arg_meta *meta,
 	    func_id != BPF_FUNC_map_pop_elem &&
 	    func_id != BPF_FUNC_map_peek_elem &&
 	    func_id != BPF_FUNC_for_each_map_elem &&
-	    func_id != BPF_FUNC_redirect_map)
+	    func_id != BPF_FUNC_redirect_map &&
+	    func_id != BPF_FUNC_map_lookup_percpu_elem)
 		return 0;
 
 	if (map == NULL) {
@@ -13810,7 +13817,8 @@ static int do_misc_fixups(struct bpf_verifier_env *env)
 		     insn->imm == BPF_FUNC_map_pop_elem    ||
 		     insn->imm == BPF_FUNC_map_peek_elem   ||
 		     insn->imm == BPF_FUNC_redirect_map    ||
-		     insn->imm == BPF_FUNC_for_each_map_elem)) {
+		     insn->imm == BPF_FUNC_for_each_map_elem ||
+		     insn->imm == BPF_FUNC_map_lookup_percpu_elem)) {
 			aux = &env->insn_aux_data[i + delta];
 			if (bpf_map_ptr_poisoned(aux))
 				goto patch_call_imm;
@@ -13859,6 +13867,8 @@ static int do_misc_fixups(struct bpf_verifier_env *env)
 					      bpf_callback_t callback_fn,
 					      void *callback_ctx,
 					      u64 flags))NULL));
+			BUILD_BUG_ON(!__same_type(ops->map_lookup_percpu_elem,
+				     (void *(*)(struct bpf_map *map, void *key, u32 cpu))NULL));
 
 patch_map_ops_generic:
 			switch (insn->imm) {
@@ -13886,6 +13896,9 @@ patch_map_ops_generic:
 			case BPF_FUNC_for_each_map_elem:
 				insn->imm = BPF_CALL_IMM(ops->map_for_each_callback);
 				continue;
+			case BPF_FUNC_map_lookup_percpu_elem:
+				insn->imm = BPF_CALL_IMM(ops->map_lookup_percpu_elem);
+				continue;
 			}
 
 			goto patch_call_imm;

kernel/trace/bpf_trace.c

@@ -1197,6 +1197,8 @@ bpf_tracing_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_map_pop_elem_proto;
 	case BPF_FUNC_map_peek_elem:
 		return &bpf_map_peek_elem_proto;
+	case BPF_FUNC_map_lookup_percpu_elem:
+		return &bpf_map_lookup_percpu_elem_proto;
 	case BPF_FUNC_ktime_get_ns:
 		return &bpf_ktime_get_ns_proto;
 	case BPF_FUNC_ktime_get_boot_ns:

tools/include/uapi/linux/bpf.h

@@ -5164,6 +5164,14 @@ union bpf_attr {
  *		if not NULL, is a reference which must be released using its
  *		corresponding release function, or moved into a BPF map before
  *		program exit.
+ *
+ * void *bpf_map_lookup_percpu_elem(struct bpf_map *map, const void *key, u32 cpu)
+ * 	Description
+ * 		Perform a lookup in *percpu map* for an entry associated to
+ * 		*key* on *cpu*.
+ * 	Return
+ * 		Map value associated to *key* on *cpu*, or **NULL** if no entry
+ * 		was found or *cpu* is invalid.
  */
 #define __BPF_FUNC_MAPPER(FN)		\
 	FN(unspec),			\
@@ -5361,6 +5369,7 @@ union bpf_attr {
 	FN(skb_set_tstamp),		\
 	FN(ima_file_hash),		\
 	FN(kptr_xchg),			\
+	FN(map_lookup_percpu_elem),     \
 	/* */
 
 /* integer value in 'imm' field of BPF_CALL instruction selects which helper

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

@@ -0,0 +1,46 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2022 Bytedance
+
+#include <test_progs.h>
+
+#include "test_map_lookup_percpu_elem.skel.h"
+
+#define TEST_VALUE  1
+
+void test_map_lookup_percpu_elem(void)
+{
+	struct test_map_lookup_percpu_elem *skel;
+	int key = 0, ret;
+	int nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
+	int *buf;
+
+	buf = (int *)malloc(nr_cpus*sizeof(int));
+	if (!ASSERT_OK_PTR(buf, "malloc"))
+		return;
+	memset(buf, 0, nr_cpus*sizeof(int));
+	buf[0] = TEST_VALUE;
+
+	skel = test_map_lookup_percpu_elem__open_and_load();
+	if (!ASSERT_OK_PTR(skel, "test_map_lookup_percpu_elem__open_and_load"))
+		return;
+	ret = test_map_lookup_percpu_elem__attach(skel);
+	ASSERT_OK(ret, "test_map_lookup_percpu_elem__attach");
+
+	ret = bpf_map_update_elem(bpf_map__fd(skel->maps.percpu_array_map), &key, buf, 0);
+	ASSERT_OK(ret, "percpu_array_map update");
+
+	ret = bpf_map_update_elem(bpf_map__fd(skel->maps.percpu_hash_map), &key, buf, 0);
+	ASSERT_OK(ret, "percpu_hash_map update");
+
+	ret = bpf_map_update_elem(bpf_map__fd(skel->maps.percpu_lru_hash_map), &key, buf, 0);
+	ASSERT_OK(ret, "percpu_lru_hash_map update");
+
+	syscall(__NR_getuid);
+
+	ret = skel->bss->percpu_array_elem_val == TEST_VALUE &&
+	      skel->bss->percpu_hash_elem_val == TEST_VALUE &&
+	      skel->bss->percpu_lru_hash_elem_val == TEST_VALUE;
+	ASSERT_OK(!ret, "bpf_map_lookup_percpu_elem success");
+
+	test_map_lookup_percpu_elem__destroy(skel);
+}

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

@@ -0,0 +1,54 @@
+// SPDX-License-Identifier: GPL-2.0
+// Copyright (c) 2022 Bytedance
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+
+int percpu_array_elem_val = 0;
+int percpu_hash_elem_val = 0;
+int percpu_lru_hash_elem_val = 0;
+
+struct {
+	__uint(type, BPF_MAP_TYPE_PERCPU_ARRAY);
+	__uint(max_entries, 1);
+	__type(key, __u32);
+	__type(value, __u32);
+} percpu_array_map SEC(".maps");
+
+struct {
+	__uint(type, BPF_MAP_TYPE_PERCPU_HASH);
+	__uint(max_entries, 1);
+	__type(key, __u32);
+	__type(value, __u32);
+} percpu_hash_map SEC(".maps");
+
+struct {
+	__uint(type, BPF_MAP_TYPE_LRU_PERCPU_HASH);
+	__uint(max_entries, 1);
+	__type(key, __u32);
+	__type(value, __u32);
+} percpu_lru_hash_map SEC(".maps");
+
+SEC("tp/syscalls/sys_enter_getuid")
+int sysenter_getuid(const void *ctx)
+{
+	__u32 key = 0;
+	__u32 cpu = 0;
+	__u32 *value;
+
+	value = bpf_map_lookup_percpu_elem(&percpu_array_map, &key, cpu);
+	if (value)
+		percpu_array_elem_val = *value;
+
+	value = bpf_map_lookup_percpu_elem(&percpu_hash_map, &key, cpu);
+	if (value)
+		percpu_hash_elem_val = *value;
+
+	value = bpf_map_lookup_percpu_elem(&percpu_lru_hash_map, &key, cpu);
+	if (value)
+		percpu_lru_hash_elem_val = *value;
+
+	return 0;
+}
+
+char _license[] SEC("license") = "GPL";