bpf: add queue and stack maps

Queue/stack maps implement a FIFO/LIFO data storage for ebpf programs.
These maps support peek, pop and push operations that are exposed to eBPF
programs through the new bpf_map[peek/pop/push] helpers.  Those operations
are exposed to userspace applications through the already existing
syscalls in the following way:

BPF_MAP_LOOKUP_ELEM            -> peek
BPF_MAP_LOOKUP_AND_DELETE_ELEM -> pop
BPF_MAP_UPDATE_ELEM            -> push

Queue/stack maps are implemented using a buffer, tail and head indexes,
hence BPF_F_NO_PREALLOC is not supported.

As opposite to other maps, queue and stack do not use RCU for protecting
maps values, the bpf_map[peek/pop] have a ARG_PTR_TO_UNINIT_MAP_VALUE
argument that is a pointer to a memory zone where to save the value of a
map.  Basically the same as ARG_PTR_TO_UNINIT_MEM, but the size has not
be passed as an extra argument.

Our main motivation for implementing queue/stack maps was to keep track
of a pool of elements, like network ports in a SNAT, however we forsee
other use cases, like for exampling saving last N kernel events in a map
and then analysing from userspace.

Signed-off-by: Mauricio Vasquez B <mauricio.vasquez@polito.it>
Acked-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

6 files changed, 359 insertions, 2 deletions

diff --git a/kernel/bpf/Makefile b/kernel/bpf/Makefile
index ff8262626b8f..4c2fa3ac56f6 100644
--- a/kernel/bpf/Makefile
+++ b/kernel/bpf/Makefile
@@ -3,7 +3,7 @@ obj-y := core.o
 
 obj-$(CONFIG_BPF_SYSCALL) += syscall.o verifier.o inode.o helpers.o tnum.o
 obj-$(CONFIG_BPF_SYSCALL) += hashtab.o arraymap.o percpu_freelist.o bpf_lru_list.o lpm_trie.o map_in_map.o
-obj-$(CONFIG_BPF_SYSCALL) += local_storage.o
+obj-$(CONFIG_BPF_SYSCALL) += local_storage.o queue_stack_maps.o
 obj-$(CONFIG_BPF_SYSCALL) += disasm.o
 obj-$(CONFIG_BPF_SYSCALL) += btf.o
 ifeq ($(CONFIG_NET),y)
diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c
index defcf4df6d91..7c7eeea8cffc 100644
--- a/kernel/bpf/core.c
+++ b/kernel/bpf/core.c
@@ -1783,6 +1783,9 @@ BPF_CALL_0(bpf_user_rnd_u32)
 const struct bpf_func_proto bpf_map_lookup_elem_proto __weak;
 const struct bpf_func_proto bpf_map_update_elem_proto __weak;
 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_get_prandom_u32_proto __weak;
 const struct bpf_func_proto bpf_get_smp_processor_id_proto __weak;
diff --git a/kernel/bpf/helpers.c b/kernel/bpf/helpers.c
index 6502115e8f55..ab0d5e3f9892 100644
--- a/kernel/bpf/helpers.c
+++ b/kernel/bpf/helpers.c
@@ -76,6 +76,49 @@ const struct bpf_func_proto bpf_map_delete_elem_proto = {
 	.arg2_type	= ARG_PTR_TO_MAP_KEY,
 };
 
+BPF_CALL_3(bpf_map_push_elem, struct bpf_map *, map, void *, value, u64, flags)
+{
+	return map->ops->map_push_elem(map, value, flags);
+}
+
+const struct bpf_func_proto bpf_map_push_elem_proto = {
+	.func		= bpf_map_push_elem,
+	.gpl_only	= false,
+	.pkt_access	= true,
+	.ret_type	= RET_INTEGER,
+	.arg1_type	= ARG_CONST_MAP_PTR,
+	.arg2_type	= ARG_PTR_TO_MAP_VALUE,
+	.arg3_type	= ARG_ANYTHING,
+};
+
+BPF_CALL_2(bpf_map_pop_elem, struct bpf_map *, map, void *, value)
+{
+	return map->ops->map_pop_elem(map, value);
+}
+
+const struct bpf_func_proto bpf_map_pop_elem_proto = {
+	.func		= bpf_map_pop_elem,
+	.gpl_only	= false,
+	.pkt_access	= true,
+	.ret_type	= RET_INTEGER,
+	.arg1_type	= ARG_CONST_MAP_PTR,
+	.arg2_type	= ARG_PTR_TO_UNINIT_MAP_VALUE,
+};
+
+BPF_CALL_2(bpf_map_peek_elem, struct bpf_map *, map, void *, value)
+{
+	return map->ops->map_peek_elem(map, value);
+}
+
+const struct bpf_func_proto bpf_map_peek_elem_proto = {
+	.func		= bpf_map_pop_elem,
+	.gpl_only	= false,
+	.pkt_access	= true,
+	.ret_type	= RET_INTEGER,
+	.arg1_type	= ARG_CONST_MAP_PTR,
+	.arg2_type	= ARG_PTR_TO_UNINIT_MAP_VALUE,
+};
+
 const struct bpf_func_proto bpf_get_prandom_u32_proto = {
 	.func		= bpf_user_rnd_u32,
 	.gpl_only	= false,
diff --git a/kernel/bpf/queue_stack_maps.c b/kernel/bpf/queue_stack_maps.c
new file mode 100644
index 000000000000..12a93fb37449
--- /dev/null
+++ b/kernel/bpf/queue_stack_maps.c
@@ -0,0 +1,288 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * queue_stack_maps.c: BPF queue and stack maps
+ *
+ * Copyright (c) 2018 Politecnico di Torino
+ */
+#include <linux/bpf.h>
+#include <linux/list.h>
+#include <linux/slab.h>
+#include "percpu_freelist.h"
+
+#define QUEUE_STACK_CREATE_FLAG_MASK \
+	(BPF_F_NUMA_NODE | BPF_F_RDONLY | BPF_F_WRONLY)
+
+
+struct bpf_queue_stack {
+	struct bpf_map map;
+	raw_spinlock_t lock;
+	u32 head, tail;
+	u32 size; /* max_entries + 1 */
+
+	char elements[0] __aligned(8);
+};
+
+static struct bpf_queue_stack *bpf_queue_stack(struct bpf_map *map)
+{
+	return container_of(map, struct bpf_queue_stack, map);
+}
+
+static bool queue_stack_map_is_empty(struct bpf_queue_stack *qs)
+{
+	return qs->head == qs->tail;
+}
+
+static bool queue_stack_map_is_full(struct bpf_queue_stack *qs)
+{
+	u32 head = qs->head + 1;
+
+	if (unlikely(head >= qs->size))
+		head = 0;
+
+	return head == qs->tail;
+}
+
+/* Called from syscall */
+static int queue_stack_map_alloc_check(union bpf_attr *attr)
+{
+	/* check sanity of attributes */
+	if (attr->max_entries == 0 || attr->key_size != 0 ||
+	    attr->map_flags & ~QUEUE_STACK_CREATE_FLAG_MASK)
+		return -EINVAL;
+
+	if (attr->value_size > KMALLOC_MAX_SIZE)
+		/* if value_size is bigger, the user space won't be able to
+		 * access the elements.
+		 */
+		return -E2BIG;
+
+	return 0;
+}
+
+static struct bpf_map *queue_stack_map_alloc(union bpf_attr *attr)
+{
+	int ret, numa_node = bpf_map_attr_numa_node(attr);
+	struct bpf_queue_stack *qs;
+	u32 size, value_size;
+	u64 queue_size, cost;
+
+	size = attr->max_entries + 1;
+	value_size = attr->value_size;
+
+	queue_size = sizeof(*qs) + (u64) value_size * size;
+
+	cost = queue_size;
+	if (cost >= U32_MAX - PAGE_SIZE)
+		return ERR_PTR(-E2BIG);
+
+	cost = round_up(cost, PAGE_SIZE) >> PAGE_SHIFT;
+
+	ret = bpf_map_precharge_memlock(cost);
+	if (ret < 0)
+		return ERR_PTR(ret);
+
+	qs = bpf_map_area_alloc(queue_size, numa_node);
+	if (!qs)
+		return ERR_PTR(-ENOMEM);
+
+	memset(qs, 0, sizeof(*qs));
+
+	bpf_map_init_from_attr(&qs->map, attr);
+
+	qs->map.pages = cost;
+	qs->size = size;
+
+	raw_spin_lock_init(&qs->lock);
+
+	return &qs->map;
+}
+
+/* Called when map->refcnt goes to zero, either from workqueue or from syscall */
+static void queue_stack_map_free(struct bpf_map *map)
+{
+	struct bpf_queue_stack *qs = bpf_queue_stack(map);
+
+	/* at this point bpf_prog->aux->refcnt == 0 and this map->refcnt == 0,
+	 * so the programs (can be more than one that used this map) were
+	 * disconnected from events. Wait for outstanding critical sections in
+	 * these programs to complete
+	 */
+	synchronize_rcu();
+
+	bpf_map_area_free(qs);
+}
+
+static int __queue_map_get(struct bpf_map *map, void *value, bool delete)
+{
+	struct bpf_queue_stack *qs = bpf_queue_stack(map);
+	unsigned long flags;
+	int err = 0;
+	void *ptr;
+
+	raw_spin_lock_irqsave(&qs->lock, flags);
+
+	if (queue_stack_map_is_empty(qs)) {
+		err = -ENOENT;
+		goto out;
+	}
+
+	ptr = &qs->elements[qs->tail * qs->map.value_size];
+	memcpy(value, ptr, qs->map.value_size);
+
+	if (delete) {
+		if (unlikely(++qs->tail >= qs->size))
+			qs->tail = 0;
+	}
+
+out:
+	raw_spin_unlock_irqrestore(&qs->lock, flags);
+	return err;
+}
+
+
+static int __stack_map_get(struct bpf_map *map, void *value, bool delete)
+{
+	struct bpf_queue_stack *qs = bpf_queue_stack(map);
+	unsigned long flags;
+	int err = 0;
+	void *ptr;
+	u32 index;
+
+	raw_spin_lock_irqsave(&qs->lock, flags);
+
+	if (queue_stack_map_is_empty(qs)) {
+		err = -ENOENT;
+		goto out;
+	}
+
+	index = qs->head - 1;
+	if (unlikely(index >= qs->size))
+		index = qs->size - 1;
+
+	ptr = &qs->elements[index * qs->map.value_size];
+	memcpy(value, ptr, qs->map.value_size);
+
+	if (delete)
+		qs->head = index;
+
+out:
+	raw_spin_unlock_irqrestore(&qs->lock, flags);
+	return err;
+}
+
+/* Called from syscall or from eBPF program */
+static int queue_map_peek_elem(struct bpf_map *map, void *value)
+{
+	return __queue_map_get(map, value, false);
+}
+
+/* Called from syscall or from eBPF program */
+static int stack_map_peek_elem(struct bpf_map *map, void *value)
+{
+	return __stack_map_get(map, value, false);
+}
+
+/* Called from syscall or from eBPF program */
+static int queue_map_pop_elem(struct bpf_map *map, void *value)
+{
+	return __queue_map_get(map, value, true);
+}
+
+/* Called from syscall or from eBPF program */
+static int stack_map_pop_elem(struct bpf_map *map, void *value)
+{
+	return __stack_map_get(map, value, true);
+}
+
+/* Called from syscall or from eBPF program */
+static int queue_stack_map_push_elem(struct bpf_map *map, void *value,
+				     u64 flags)
+{
+	struct bpf_queue_stack *qs = bpf_queue_stack(map);
+	unsigned long irq_flags;
+	int err = 0;
+	void *dst;
+
+	/* BPF_EXIST is used to force making room for a new element in case the
+	 * map is full
+	 */
+	bool replace = (flags & BPF_EXIST);
+
+	/* Check supported flags for queue and stack maps */
+	if (flags & BPF_NOEXIST || flags > BPF_EXIST)
+		return -EINVAL;
+
+	raw_spin_lock_irqsave(&qs->lock, irq_flags);
+
+	if (queue_stack_map_is_full(qs)) {
+		if (!replace) {
+			err = -E2BIG;
+			goto out;
+		}
+		/* advance tail pointer to overwrite oldest element */
+		if (unlikely(++qs->tail >= qs->size))
+			qs->tail = 0;
+	}
+
+	dst = &qs->elements[qs->head * qs->map.value_size];
+	memcpy(dst, value, qs->map.value_size);
+
+	if (unlikely(++qs->head >= qs->size))
+		qs->head = 0;
+
+out:
+	raw_spin_unlock_irqrestore(&qs->lock, irq_flags);
+	return err;
+}
+
+/* Called from syscall or from eBPF program */
+static void *queue_stack_map_lookup_elem(struct bpf_map *map, void *key)
+{
+	return NULL;
+}
+
+/* Called from syscall or from eBPF program */
+static int queue_stack_map_update_elem(struct bpf_map *map, void *key,
+				       void *value, u64 flags)
+{
+	return -EINVAL;
+}
+
+/* Called from syscall or from eBPF program */
+static int queue_stack_map_delete_elem(struct bpf_map *map, void *key)
+{
+	return -EINVAL;
+}
+
+/* Called from syscall */
+static int queue_stack_map_get_next_key(struct bpf_map *map, void *key,
+					void *next_key)
+{
+	return -EINVAL;
+}
+
+const struct bpf_map_ops queue_map_ops = {
+	.map_alloc_check = queue_stack_map_alloc_check,
+	.map_alloc = queue_stack_map_alloc,
+	.map_free = queue_stack_map_free,
+	.map_lookup_elem = queue_stack_map_lookup_elem,
+	.map_update_elem = queue_stack_map_update_elem,
+	.map_delete_elem = queue_stack_map_delete_elem,
+	.map_push_elem = queue_stack_map_push_elem,
+	.map_pop_elem = queue_map_pop_elem,
+	.map_peek_elem = queue_map_peek_elem,
+	.map_get_next_key = queue_stack_map_get_next_key,
+};
+
+const struct bpf_map_ops stack_map_ops = {
+	.map_alloc_check = queue_stack_map_alloc_check,
+	.map_alloc = queue_stack_map_alloc,
+	.map_free = queue_stack_map_free,
+	.map_lookup_elem = queue_stack_map_lookup_elem,
+	.map_update_elem = queue_stack_map_update_elem,
+	.map_delete_elem = queue_stack_map_delete_elem,
+	.map_push_elem = queue_stack_map_push_elem,
+	.map_pop_elem = stack_map_pop_elem,
+	.map_peek_elem = stack_map_peek_elem,
+	.map_get_next_key = queue_stack_map_get_next_key,
+};
diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 78d9dd95e25f..1617407f9ee5 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -727,6 +727,9 @@ static int map_lookup_elem(union bpf_attr *attr)
 		err = bpf_fd_htab_map_lookup_elem(map, key, value);
 	} else if (map->map_type == BPF_MAP_TYPE_REUSEPORT_SOCKARRAY) {
 		err = bpf_fd_reuseport_array_lookup_elem(map, key, value);
+	} else if (map->map_type == BPF_MAP_TYPE_QUEUE ||
+		   map->map_type == BPF_MAP_TYPE_STACK) {
+		err = map->ops->map_peek_elem(map, value);
 	} else {
 		rcu_read_lock();
 		ptr = map->ops->map_lookup_elem(map, key);
@@ -857,6 +860,9 @@ static int map_update_elem(union bpf_attr *attr)
 		/* rcu_read_lock() is not needed */
 		err = bpf_fd_reuseport_array_update_elem(map, key, value,
 							 attr->flags);
+	} else if (map->map_type == BPF_MAP_TYPE_QUEUE ||
+		   map->map_type == BPF_MAP_TYPE_STACK) {
+		err = map->ops->map_push_elem(map, value, attr->flags);
 	} else {
 		rcu_read_lock();
 		err = map->ops->map_update_elem(map, key, value, attr->flags);
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index d84c91ac3b70..7d6d9cf9ebd5 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -2324,6 +2324,13 @@ static int check_map_func_compatibility(struct bpf_verifier_env *env,
 		if (func_id != BPF_FUNC_sk_select_reuseport)
 			goto error;
 		break;
+	case BPF_MAP_TYPE_QUEUE:
+	case BPF_MAP_TYPE_STACK:
+		if (func_id != BPF_FUNC_map_peek_elem &&
+		    func_id != BPF_FUNC_map_pop_elem &&
+		    func_id != BPF_FUNC_map_push_elem)
+			goto error;
+		break;
 	default:
 		break;
 	}
@@ -2380,6 +2387,13 @@ static int check_map_func_compatibility(struct bpf_verifier_env *env,
 		if (map->map_type != BPF_MAP_TYPE_REUSEPORT_SOCKARRAY)
 			goto error;
 		break;
+	case BPF_FUNC_map_peek_elem:
+	case BPF_FUNC_map_pop_elem:
+	case BPF_FUNC_map_push_elem:
+		if (map->map_type != BPF_MAP_TYPE_QUEUE &&
+		    map->map_type != BPF_MAP_TYPE_STACK)
+			goto error;
+		break;
 	default:
 		break;
 	}
@@ -2675,7 +2689,10 @@ record_func_map(struct bpf_verifier_env *env, struct bpf_call_arg_meta *meta,
 	if (func_id != BPF_FUNC_tail_call &&
 	    func_id != BPF_FUNC_map_lookup_elem &&
 	    func_id != BPF_FUNC_map_update_elem &&
-	    func_id != BPF_FUNC_map_delete_elem)
+	    func_id != BPF_FUNC_map_delete_elem &&
+	    func_id != BPF_FUNC_map_push_elem &&
+	    func_id != BPF_FUNC_map_pop_elem &&
+	    func_id != BPF_FUNC_map_peek_elem)
 		return 0;
 
 	if (meta->map_ptr == NULL) {