Re: [PATCH v5 bpf-next 1/3] bpf: add lookup_and_delete_elem support to hashtab

[Yonghong Song <yhs@fb.com>]

On 4/16/21 2:58 AM, Denis Salopek wrote:
> Extend the existing bpf_map_lookup_and_delete_elem() functionality to
> hashtab map types, in addition to stacks and queues.
> Create a new hashtab bpf_map_ops function that does lookup and deletion
> of the element under the same bucket lock and add the created map_ops to
> bpf.h.
> 
> Cc: Juraj Vijtiuk <juraj.vijtiuk@sartura.hr>
> Cc: Luka Oreskovic <luka.oreskovic@sartura.hr>
> Cc: Luka Perkov <luka.perkov@sartura.hr>
> Cc: Yonghong Song <yhs@fb.com>
> Cc: Andrii Nakryiko <andrii.nakryiko@gmail.com>
> Cc: Daniel Borkmann <daniel@iogearbox.net>
> Signed-off-by: Denis Salopek <denis.salopek@sartura.hr>
> ---
> v2: Add functionality for LRU/per-CPU, add test_progs tests.
> v3: Add bpf_map_lookup_and_delete_elem_flags() and enable BPF_F_LOCK
> flag, change CHECKs to ASSERT_OKs, initialize variables to 0.
> v4: Fix the return value for unsupported map types.
> v5: Split patch to 3 patches. Extend BPF_MAP_LOOKUP_AND_DELETE_ELEM
> documentation with this changes.
> ---
>   include/linux/bpf.h            |  2 +
>   include/uapi/linux/bpf.h       | 13 +++++
>   kernel/bpf/hashtab.c           | 99 ++++++++++++++++++++++++++++++++++
>   kernel/bpf/syscall.c           | 33 ++++++++++--
>   tools/include/uapi/linux/bpf.h | 13 +++++
>   5 files changed, 156 insertions(+), 4 deletions(-)
> 
> diff --git a/include/linux/bpf.h b/include/linux/bpf.h
> index ff8cd68c01b3..d39fe682799e 100644
> --- a/include/linux/bpf.h
> +++ b/include/linux/bpf.h
> @@ -68,6 +68,8 @@ struct bpf_map_ops {
>   	void *(*map_lookup_elem_sys_only)(struct bpf_map *map, void *key);
>   	int (*map_lookup_batch)(struct bpf_map *map, const union bpf_attr *attr,
>   				union bpf_attr __user *uattr);
> +	int (*map_lookup_and_delete_elem)(struct bpf_map *map, void *key,
> +					  void *value, u64 flags);
>   	int (*map_lookup_and_delete_batch)(struct bpf_map *map,
>   					   const union bpf_attr *attr,
>   					   union bpf_attr __user *uattr);
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index df164a44bb41..f30cabe02814 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -527,6 +527,15 @@ union bpf_iter_link_info {
>    *		Look up an element with the given *key* in the map referred to
>    *		by the file descriptor *fd*, and if found, delete the element.
>    *
> + *		For **BPF_MAP_TYPE_QUEUE** and **BPF_MAP_TYPE_STACK** map
> + *		types, the *flags* argument needs to be set to 0, but for other
> + *		map types, it may be specified as:
> + *
> + *		**BPF_F_LOCK**
> + *			Look up and delete the value of a spin-locked map
> + *			without returning the lock. This must be specified if
> + *			the elements contain a spinlock.
> + *
>    *		The **BPF_MAP_TYPE_QUEUE** and **BPF_MAP_TYPE_STACK** map types
>    *		implement this command as a "pop" operation, deleting the top
>    *		element rather than one corresponding to *key*.
> @@ -536,6 +545,10 @@ union bpf_iter_link_info {
>    *		This command is only valid for the following map types:
>    *		* **BPF_MAP_TYPE_QUEUE**
>    *		* **BPF_MAP_TYPE_STACK**
> + *		* **BPF_MAP_TYPE_HASH**
> + *		* **BPF_MAP_TYPE_PERCPU_HASH**
> + *		* **BPF_MAP_TYPE_LRU_HASH**
> + *		* **BPF_MAP_TYPE_LRU_PERCPU_HASH**
>    *
>    *	Return
>    *		Returns zero on success. On error, -1 is returned and *errno*
> diff --git a/kernel/bpf/hashtab.c b/kernel/bpf/hashtab.c
> index d7ebb12ffffc..5e57503d4706 100644
> --- a/kernel/bpf/hashtab.c
> +++ b/kernel/bpf/hashtab.c
> @@ -1401,6 +1401,101 @@ static void htab_map_seq_show_elem(struct bpf_map *map, void *key,
>   	rcu_read_unlock();
>   }
>   
> +static int __htab_map_lookup_and_delete_elem(struct bpf_map *map, void *key,
> +					     void *value, bool is_lru_map,
> +					     bool is_percpu, u64 flags)
> +{
> +	struct bpf_htab *htab = container_of(map, struct bpf_htab, map);
> +	struct hlist_nulls_head *head;
> +	unsigned long bflags;
> +	struct htab_elem *l;
> +	u32 hash, key_size;
> +	struct bucket *b;
> +	int ret;
> +
> +	if ((flags & ~BPF_F_LOCK) ||
> +	    ((flags & BPF_F_LOCK) && !map_value_has_spin_lock(map)))
> +		return -EINVAL;

We don't need to check here. It has been checked in
map_lookup_and_delete_elem().

> +
> +	key_size = map->key_size;
> +
> +	hash = htab_map_hash(key, key_size, htab->hashrnd);
> +	b = __select_bucket(htab, hash);
> +	head = &b->head;
> +
> +	ret = htab_lock_bucket(htab, b, hash, &bflags);
> +	if (ret)
> +		return ret;
> +
> +	l = lookup_elem_raw(head, hash, key, key_size);
> +	if (l) {
> +		if (is_percpu) {
> +			u32 roundup_value_size = round_up(map->value_size, 8);
> +			void __percpu *pptr;
> +			int off = 0, cpu;
> +
> +			pptr = htab_elem_get_ptr(l, key_size);
> +			for_each_possible_cpu(cpu) {
> +				bpf_long_memcpy(value + off,
> +						per_cpu_ptr(pptr, cpu),
> +						roundup_value_size);
> +				off += roundup_value_size;
> +			}
> +		} else {
> +			if (flags & BPF_F_LOCK)
> +				copy_map_value_locked(map, value, l->key +
> +						      round_up(key_size, 8),
> +						      true);
> +			else
> +				copy_map_value(map, value, l->key +
> +					       round_up(key_size, 8));

You can have a common declaration like below in the beginning of the block.
	u32 roundup_key_size = round_up(map->key_size, 8);
and use roundup_key_size in copy_map_value_locked() and
copy_map_value().

> +			check_and_init_map_lock(map, value);
> +		}
> +
> +		hlist_nulls_del_rcu(&l->hash_node);
> +		if (!is_lru_map)
> +			free_htab_elem(htab, l);
> +	} else
> +		ret = -ENOENT;

Probably it is more readable if you write above like
	if (!l) {
		ret = -ENOENT;
	} else {
		...
	}

> +
> +	htab_unlock_bucket(htab, b, hash, bflags);
> +
> +	if (is_lru_map && l)
> +		bpf_lru_push_free(&htab->lru, &l->lru_node);
> +
> +	return ret;
> +}
> +
> +static int htab_map_lookup_and_delete_elem(struct bpf_map *map, void *key,
> +					   void *value, u64 flags)
> +{
> +	return __htab_map_lookup_and_delete_elem(map, key, value, false, false,
> +						 flags);
> +}
> +
> +static int htab_percpu_map_lookup_and_delete_elem(struct bpf_map *map,
> +						  void *key, void *value,
> +						  u64 flags)
> +{
> +	return __htab_map_lookup_and_delete_elem(map, key, value, false, true,
> +						 flags);
> +}
> +
> +static int htab_lru_map_lookup_and_delete_elem(struct bpf_map *map, void *key,
> +					       void *value, u64 flags)
> +{
> +	return __htab_map_lookup_and_delete_elem(map, key, value, true, false,
> +						 flags);
> +}
> +
> +static int htab_lru_percpu_map_lookup_and_delete_elem(struct bpf_map *map,
> +						      void *key, void *value,
> +						      u64 flags)
> +{
> +	return __htab_map_lookup_and_delete_elem(map, key, value, true, true,
> +						 flags);
> +}
> +
>   static int
>   __htab_map_lookup_and_delete_batch(struct bpf_map *map,
>   				   const union bpf_attr *attr,
> @@ -1934,6 +2029,7 @@ const struct bpf_map_ops htab_map_ops = {
>   	.map_free = htab_map_free,
>   	.map_get_next_key = htab_map_get_next_key,
>   	.map_lookup_elem = htab_map_lookup_elem,
> +	.map_lookup_and_delete_elem = htab_map_lookup_and_delete_elem,
>   	.map_update_elem = htab_map_update_elem,
>   	.map_delete_elem = htab_map_delete_elem,
>   	.map_gen_lookup = htab_map_gen_lookup,
> @@ -1954,6 +2050,7 @@ const struct bpf_map_ops htab_lru_map_ops = {
>   	.map_free = htab_map_free,
>   	.map_get_next_key = htab_map_get_next_key,
>   	.map_lookup_elem = htab_lru_map_lookup_elem,
> +	.map_lookup_and_delete_elem = htab_lru_map_lookup_and_delete_elem,
>   	.map_lookup_elem_sys_only = htab_lru_map_lookup_elem_sys,
>   	.map_update_elem = htab_lru_map_update_elem,
>   	.map_delete_elem = htab_lru_map_delete_elem,
> @@ -2077,6 +2174,7 @@ const struct bpf_map_ops htab_percpu_map_ops = {
>   	.map_free = htab_map_free,
>   	.map_get_next_key = htab_map_get_next_key,
>   	.map_lookup_elem = htab_percpu_map_lookup_elem,
> +	.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_seq_show_elem = htab_percpu_map_seq_show_elem,
> @@ -2096,6 +2194,7 @@ const struct bpf_map_ops htab_lru_percpu_map_ops = {
>   	.map_free = htab_map_free,
>   	.map_get_next_key = htab_map_get_next_key,
>   	.map_lookup_elem = htab_lru_percpu_map_lookup_elem,
> +	.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_seq_show_elem = htab_percpu_map_seq_show_elem,
> diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
> index fd495190115e..78f6312d9bdb 100644
> --- a/kernel/bpf/syscall.c
> +++ b/kernel/bpf/syscall.c
> @@ -1468,7 +1468,7 @@ int generic_map_lookup_batch(struct bpf_map *map,
>   	return err;
>   }
>   
> -#define BPF_MAP_LOOKUP_AND_DELETE_ELEM_LAST_FIELD value
> +#define BPF_MAP_LOOKUP_AND_DELETE_ELEM_LAST_FIELD flags
>   
>   static int map_lookup_and_delete_elem(union bpf_attr *attr)
>   {
> @@ -1484,6 +1484,9 @@ static int map_lookup_and_delete_elem(union bpf_attr *attr)
>   	if (CHECK_ATTR(BPF_MAP_LOOKUP_AND_DELETE_ELEM))
>   		return -EINVAL;
>   
> +	if (attr->flags & ~BPF_F_LOCK)
> +		return -EINVAL;
> +
>   	f = fdget(ufd);
>   	map = __bpf_map_get(f);
>   	if (IS_ERR(map))
> @@ -1494,24 +1497,46 @@ static int map_lookup_and_delete_elem(union bpf_attr *attr)
>   		goto err_put;
>   	}
>   
> +	if (attr->flags && (map->map_type == BPF_MAP_TYPE_QUEUE ||
> +	    map->map_type == BPF_MAP_TYPE_STACK)) {

Need better code alignment.

	if (attr->flags &&
	    (map->map_type == BPF_MAP_TYPE_QUEUE ||
	     map->map_type == BPF_MAP_TYPE_STACK)) {
		...
	}

> +		err = -EINVAL;
> +		goto err_put;
> +	}
> +
> +	if ((attr->flags & BPF_F_LOCK) &&
> +	    !map_value_has_spin_lock(map)) {
> +		err = -EINVAL;
> +		goto err_put;
> +	}
> +
>   	key = __bpf_copy_key(ukey, map->key_size);
>   	if (IS_ERR(key)) {
>   		err = PTR_ERR(key);
>   		goto err_put;
>   	}
>   
> -	value_size = map->value_size;
> +	value_size = bpf_map_value_size(map);
>   
>   	err = -ENOMEM;
>   	value = kmalloc(value_size, GFP_USER | __GFP_NOWARN);
>   	if (!value)
>   		goto free_key;
>   
> +	err = -ENOTSUPP;
>   	if (map->map_type == BPF_MAP_TYPE_QUEUE ||
>   	    map->map_type == BPF_MAP_TYPE_STACK) {
>   		err = map->ops->map_pop_elem(map, value);
> -	} else {
> -		err = -ENOTSUPP;
> +	} else if (map->map_type == BPF_MAP_TYPE_HASH ||
> +		   map->map_type == BPF_MAP_TYPE_PERCPU_HASH ||
> +		   map->map_type == BPF_MAP_TYPE_LRU_HASH ||
> +		   map->map_type == BPF_MAP_TYPE_LRU_PERCPU_HASH) {
> +		if (!bpf_map_is_dev_bound(map)) {
> +			bpf_disable_instrumentation();
> +			rcu_read_lock();
> +			err = map->ops->map_lookup_and_delete_elem(map, key, value, attr->flags);
> +			rcu_read_unlock();
> +			bpf_enable_instrumentation();
> +		}
>   	}
>   
>   	if (err)
[...]