diff options
| -rw-r--r-- | include/linux/bpf.h | 59 | ||||
| -rw-r--r-- | include/linux/netdevice.h | 6 | ||||
| -rw-r--r-- | include/uapi/linux/bpf.h | 1 | ||||
| -rw-r--r-- | kernel/bpf/offload.c | 188 | ||||
| -rw-r--r-- | kernel/bpf/syscall.c | 44 | ||||
| -rw-r--r-- | kernel/bpf/verifier.c | 7 | ||||
| -rw-r--r-- | tools/include/uapi/linux/bpf.h | 1 |
7 files changed, 293 insertions, 13 deletions
diff --git a/include/linux/bpf.h b/include/linux/bpf.h index 9fff1ace1d8e..5c2c104dc2c5 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -74,6 +74,33 @@ struct bpf_map { char name[BPF_OBJ_NAME_LEN]; }; +struct bpf_offloaded_map; + +struct bpf_map_dev_ops { + int (*map_get_next_key)(struct bpf_offloaded_map *map, + void *key, void *next_key); + int (*map_lookup_elem)(struct bpf_offloaded_map *map, + void *key, void *value); + int (*map_update_elem)(struct bpf_offloaded_map *map, + void *key, void *value, u64 flags); + int (*map_delete_elem)(struct bpf_offloaded_map *map, void *key); +}; + +struct bpf_offloaded_map { + struct bpf_map map; + struct net_device *netdev; + const struct bpf_map_dev_ops *dev_ops; + void *dev_priv; + struct list_head offloads; +}; + +static inline struct bpf_offloaded_map *map_to_offmap(struct bpf_map *map) +{ + return container_of(map, struct bpf_offloaded_map, map); +} + +extern const struct bpf_map_ops bpf_map_offload_ops; + /* function argument constraints */ enum bpf_arg_type { ARG_DONTCARE = 0, /* unused argument in helper function */ @@ -369,6 +396,7 @@ int __bpf_prog_charge(struct user_struct *user, u32 pages); void __bpf_prog_uncharge(struct user_struct *user, u32 pages); void bpf_prog_free_id(struct bpf_prog *prog, bool do_idr_lock); +void bpf_map_free_id(struct bpf_map *map, bool do_idr_lock); struct bpf_map *bpf_map_get_with_uref(u32 ufd); struct bpf_map *__bpf_map_get(struct fd f); @@ -556,6 +584,15 @@ void bpf_prog_offload_destroy(struct bpf_prog *prog); int bpf_prog_offload_info_fill(struct bpf_prog_info *info, struct bpf_prog *prog); +int bpf_map_offload_lookup_elem(struct bpf_map *map, void *key, void *value); +int bpf_map_offload_update_elem(struct bpf_map *map, + void *key, void *value, u64 flags); +int bpf_map_offload_delete_elem(struct bpf_map *map, void *key); +int bpf_map_offload_get_next_key(struct bpf_map *map, + void *key, void *next_key); + +bool bpf_offload_dev_match(struct bpf_prog *prog, struct bpf_map *map); + #if defined(CONFIG_NET) && defined(CONFIG_BPF_SYSCALL) int bpf_prog_offload_init(struct bpf_prog *prog, union bpf_attr *attr); @@ -563,6 +600,14 @@ static inline bool bpf_prog_is_dev_bound(struct bpf_prog_aux *aux) { return aux->offload_requested; } + +static inline bool bpf_map_is_dev_bound(struct bpf_map *map) +{ + return unlikely(map->ops == &bpf_map_offload_ops); +} + +struct bpf_map *bpf_map_offload_map_alloc(union bpf_attr *attr); +void bpf_map_offload_map_free(struct bpf_map *map); #else static inline int bpf_prog_offload_init(struct bpf_prog *prog, union bpf_attr *attr) @@ -574,6 +619,20 @@ static inline bool bpf_prog_is_dev_bound(struct bpf_prog_aux *aux) { return false; } + +static inline bool bpf_map_is_dev_bound(struct bpf_map *map) +{ + return false; +} + +static inline struct bpf_map *bpf_map_offload_map_alloc(union bpf_attr *attr) +{ + return ERR_PTR(-EOPNOTSUPP); +} + +static inline void bpf_map_offload_map_free(struct bpf_map *map) +{ +} #endif /* CONFIG_NET && CONFIG_BPF_SYSCALL */ #if defined(CONFIG_STREAM_PARSER) && defined(CONFIG_BPF_SYSCALL) && defined(CONFIG_INET) diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index ef7b348e8498..0b3ab42d50fe 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -804,6 +804,8 @@ enum bpf_netdev_command { BPF_OFFLOAD_VERIFIER_PREP, BPF_OFFLOAD_TRANSLATE, BPF_OFFLOAD_DESTROY, + BPF_OFFLOAD_MAP_ALLOC, + BPF_OFFLOAD_MAP_FREE, }; struct bpf_prog_offload_ops; @@ -834,6 +836,10 @@ struct netdev_bpf { struct { struct bpf_prog *prog; } offload; + /* BPF_OFFLOAD_MAP_ALLOC, BPF_OFFLOAD_MAP_FREE */ + struct { + struct bpf_offloaded_map *offmap; + }; }; }; diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index 395d261948de..7c2259e8bc54 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -245,6 +245,7 @@ union bpf_attr { * BPF_F_NUMA_NODE is set). */ char map_name[BPF_OBJ_NAME_LEN]; + __u32 map_ifindex; /* ifindex of netdev to create on */ }; struct { /* anonymous struct used by BPF_MAP_*_ELEM commands */ diff --git a/kernel/bpf/offload.c b/kernel/bpf/offload.c index cdd1e19a668b..453785fa1881 100644 --- a/kernel/bpf/offload.c +++ b/kernel/bpf/offload.c @@ -24,11 +24,13 @@ #include <linux/rtnetlink.h> #include <linux/rwsem.h> -/* Protects bpf_prog_offload_devs and offload members of all progs. +/* Protects bpf_prog_offload_devs, bpf_map_offload_devs and offload members + * of all progs. * RTNL lock cannot be taken when holding this lock. */ static DECLARE_RWSEM(bpf_devs_lock); static LIST_HEAD(bpf_prog_offload_devs); +static LIST_HEAD(bpf_map_offload_devs); static int bpf_dev_offload_check(struct net_device *netdev) { @@ -250,11 +252,186 @@ int bpf_prog_offload_info_fill(struct bpf_prog_info *info, const struct bpf_prog_ops bpf_offload_prog_ops = { }; +static int bpf_map_offload_ndo(struct bpf_offloaded_map *offmap, + enum bpf_netdev_command cmd) +{ + struct netdev_bpf data = {}; + struct net_device *netdev; + + ASSERT_RTNL(); + + data.command = cmd; + data.offmap = offmap; + /* Caller must make sure netdev is valid */ + netdev = offmap->netdev; + + return netdev->netdev_ops->ndo_bpf(netdev, &data); +} + +struct bpf_map *bpf_map_offload_map_alloc(union bpf_attr *attr) +{ + struct net *net = current->nsproxy->net_ns; + struct bpf_offloaded_map *offmap; + int err; + + if (!capable(CAP_SYS_ADMIN)) + return ERR_PTR(-EPERM); + if (attr->map_type != BPF_MAP_TYPE_HASH) + return ERR_PTR(-EINVAL); + + offmap = kzalloc(sizeof(*offmap), GFP_USER); + if (!offmap) + return ERR_PTR(-ENOMEM); + + bpf_map_init_from_attr(&offmap->map, attr); + + rtnl_lock(); + down_write(&bpf_devs_lock); + offmap->netdev = __dev_get_by_index(net, attr->map_ifindex); + err = bpf_dev_offload_check(offmap->netdev); + if (err) + goto err_unlock; + + err = bpf_map_offload_ndo(offmap, BPF_OFFLOAD_MAP_ALLOC); + if (err) + goto err_unlock; + + list_add_tail(&offmap->offloads, &bpf_map_offload_devs); + up_write(&bpf_devs_lock); + rtnl_unlock(); + + return &offmap->map; + +err_unlock: + up_write(&bpf_devs_lock); + rtnl_unlock(); + kfree(offmap); + return ERR_PTR(err); +} + +static void __bpf_map_offload_destroy(struct bpf_offloaded_map *offmap) +{ + WARN_ON(bpf_map_offload_ndo(offmap, BPF_OFFLOAD_MAP_FREE)); + /* Make sure BPF_MAP_GET_NEXT_ID can't find this dead map */ + bpf_map_free_id(&offmap->map, true); + list_del_init(&offmap->offloads); + offmap->netdev = NULL; +} + +void bpf_map_offload_map_free(struct bpf_map *map) +{ + struct bpf_offloaded_map *offmap = map_to_offmap(map); + + rtnl_lock(); + down_write(&bpf_devs_lock); + if (offmap->netdev) + __bpf_map_offload_destroy(offmap); + up_write(&bpf_devs_lock); + rtnl_unlock(); + + kfree(offmap); +} + +int bpf_map_offload_lookup_elem(struct bpf_map *map, void *key, void *value) +{ + struct bpf_offloaded_map *offmap = map_to_offmap(map); + int ret = -ENODEV; + + down_read(&bpf_devs_lock); + if (offmap->netdev) + ret = offmap->dev_ops->map_lookup_elem(offmap, key, value); + up_read(&bpf_devs_lock); + + return ret; +} + +int bpf_map_offload_update_elem(struct bpf_map *map, + void *key, void *value, u64 flags) +{ + struct bpf_offloaded_map *offmap = map_to_offmap(map); + int ret = -ENODEV; + + if (unlikely(flags > BPF_EXIST)) + return -EINVAL; + + down_read(&bpf_devs_lock); + if (offmap->netdev) + ret = offmap->dev_ops->map_update_elem(offmap, key, value, + flags); + up_read(&bpf_devs_lock); + + return ret; +} + +int bpf_map_offload_delete_elem(struct bpf_map *map, void *key) +{ + struct bpf_offloaded_map *offmap = map_to_offmap(map); + int ret = -ENODEV; + + down_read(&bpf_devs_lock); + if (offmap->netdev) + ret = offmap->dev_ops->map_delete_elem(offmap, key); + up_read(&bpf_devs_lock); + + return ret; +} + +int bpf_map_offload_get_next_key(struct bpf_map *map, void *key, void *next_key) +{ + struct bpf_offloaded_map *offmap = map_to_offmap(map); + int ret = -ENODEV; + + down_read(&bpf_devs_lock); + if (offmap->netdev) + ret = offmap->dev_ops->map_get_next_key(offmap, key, next_key); + up_read(&bpf_devs_lock); + + return ret; +} + +bool bpf_offload_dev_match(struct bpf_prog *prog, struct bpf_map *map) +{ + struct bpf_offloaded_map *offmap; + struct bpf_prog_offload *offload; + bool ret; + + if (!!bpf_prog_is_dev_bound(prog->aux) != !!bpf_map_is_dev_bound(map)) + return false; + if (!bpf_prog_is_dev_bound(prog->aux)) + return true; + + down_read(&bpf_devs_lock); + offload = prog->aux->offload; + offmap = map_to_offmap(map); + + ret = offload && offload->netdev == offmap->netdev; + up_read(&bpf_devs_lock); + + return ret; +} + +static void bpf_offload_orphan_all_progs(struct net_device *netdev) +{ + struct bpf_prog_offload *offload, *tmp; + + list_for_each_entry_safe(offload, tmp, &bpf_prog_offload_devs, offloads) + if (offload->netdev == netdev) + __bpf_prog_offload_destroy(offload->prog); +} + +static void bpf_offload_orphan_all_maps(struct net_device *netdev) +{ + struct bpf_offloaded_map *offmap, *tmp; + + list_for_each_entry_safe(offmap, tmp, &bpf_map_offload_devs, offloads) + if (offmap->netdev == netdev) + __bpf_map_offload_destroy(offmap); +} + static int bpf_offload_notification(struct notifier_block *notifier, ulong event, void *ptr) { struct net_device *netdev = netdev_notifier_info_to_dev(ptr); - struct bpf_prog_offload *offload, *tmp; ASSERT_RTNL(); @@ -265,11 +442,8 @@ static int bpf_offload_notification(struct notifier_block *notifier, break; down_write(&bpf_devs_lock); - list_for_each_entry_safe(offload, tmp, &bpf_prog_offload_devs, - offloads) { - if (offload->netdev == netdev) - __bpf_prog_offload_destroy(offload->prog); - } + bpf_offload_orphan_all_progs(netdev); + bpf_offload_orphan_all_maps(netdev); up_write(&bpf_devs_lock); break; default: diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index a3f726bb42ea..c691b9e972e3 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -94,6 +94,11 @@ static int check_uarg_tail_zero(void __user *uaddr, return 0; } +const struct bpf_map_ops bpf_map_offload_ops = { + .map_alloc = bpf_map_offload_map_alloc, + .map_free = bpf_map_offload_map_free, +}; + static struct bpf_map *find_and_alloc_map(union bpf_attr *attr) { const struct bpf_map_ops *ops; @@ -111,6 +116,8 @@ static struct bpf_map *find_and_alloc_map(union bpf_attr *attr) if (err) return ERR_PTR(err); } + if (attr->map_ifindex) + ops = &bpf_map_offload_ops; map = ops->map_alloc(attr); if (IS_ERR(map)) return map; @@ -208,16 +215,25 @@ static int bpf_map_alloc_id(struct bpf_map *map) return id > 0 ? 0 : id; } -static void bpf_map_free_id(struct bpf_map *map, bool do_idr_lock) +void bpf_map_free_id(struct bpf_map *map, bool do_idr_lock) { unsigned long flags; + /* Offloaded maps are removed from the IDR store when their device + * disappears - even if someone holds an fd to them they are unusable, + * the memory is gone, all ops will fail; they are simply waiting for + * refcnt to drop to be freed. + */ + if (!map->id) + return; + if (do_idr_lock) spin_lock_irqsave(&map_idr_lock, flags); else __acquire(&map_idr_lock); idr_remove(&map_idr, map->id); + map->id = 0; if (do_idr_lock) spin_unlock_irqrestore(&map_idr_lock, flags); @@ -397,7 +413,7 @@ static int bpf_obj_name_cpy(char *dst, const char *src) return 0; } -#define BPF_MAP_CREATE_LAST_FIELD map_name +#define BPF_MAP_CREATE_LAST_FIELD map_ifindex /* called via syscall */ static int map_create(union bpf_attr *attr) { @@ -585,8 +601,10 @@ static int map_lookup_elem(union bpf_attr *attr) if (!value) goto free_key; - if (map->map_type == BPF_MAP_TYPE_PERCPU_HASH || - map->map_type == BPF_MAP_TYPE_LRU_PERCPU_HASH) { + if (bpf_map_is_dev_bound(map)) { + err = bpf_map_offload_lookup_elem(map, key, value); + } else if (map->map_type == BPF_MAP_TYPE_PERCPU_HASH || + map->map_type == BPF_MAP_TYPE_LRU_PERCPU_HASH) { 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); @@ -673,7 +691,10 @@ static int map_update_elem(union bpf_attr *attr) goto free_value; /* Need to create a kthread, thus must support schedule */ - if (map->map_type == BPF_MAP_TYPE_CPUMAP) { + if (bpf_map_is_dev_bound(map)) { + err = bpf_map_offload_update_elem(map, key, value, attr->flags); + goto out; + } else if (map->map_type == BPF_MAP_TYPE_CPUMAP) { err = map->ops->map_update_elem(map, key, value, attr->flags); goto out; } @@ -750,6 +771,11 @@ static int map_delete_elem(union bpf_attr *attr) goto err_put; } + if (bpf_map_is_dev_bound(map)) { + err = bpf_map_offload_delete_elem(map, key); + goto out; + } + preempt_disable(); __this_cpu_inc(bpf_prog_active); rcu_read_lock(); @@ -757,7 +783,7 @@ static int map_delete_elem(union bpf_attr *attr) rcu_read_unlock(); __this_cpu_dec(bpf_prog_active); preempt_enable(); - +out: if (!err) trace_bpf_map_delete_elem(map, ufd, key); kfree(key); @@ -807,9 +833,15 @@ static int map_get_next_key(union bpf_attr *attr) if (!next_key) goto free_key; + if (bpf_map_is_dev_bound(map)) { + err = bpf_map_offload_get_next_key(map, key, next_key); + goto out; + } + rcu_read_lock(); err = map->ops->map_get_next_key(map, key, next_key); rcu_read_unlock(); +out: if (err) goto free_next_key; diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index 48b61caa94cb..ceabb394d2dc 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -4816,6 +4816,13 @@ static int check_map_prog_compatibility(struct bpf_verifier_env *env, return -EINVAL; } } + + if ((bpf_prog_is_dev_bound(prog->aux) || bpf_map_is_dev_bound(map)) && + !bpf_offload_dev_match(prog, map)) { + verbose(env, "offload device mismatch between prog and map\n"); + return -EINVAL; + } + return 0; } diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h index 4e8c60acfa32..69f96af4a569 100644 --- a/tools/include/uapi/linux/bpf.h +++ b/tools/include/uapi/linux/bpf.h @@ -245,6 +245,7 @@ union bpf_attr { * BPF_F_NUMA_NODE is set). */ char map_name[BPF_OBJ_NAME_LEN]; + __u32 map_ifindex; /* ifindex of netdev to create on */ }; struct { /* anonymous struct used by BPF_MAP_*_ELEM commands */ |