From 40a1227ea845a37ab197dd1caffb60b047fa36b1 Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Wed, 8 Aug 2018 01:01:21 -0700 Subject: [PATCH 1/9] tcp: Avoid TCP syncookie rejected by SO_REUSEPORT socket Although the actual cookie check "__cookie_v[46]_check()" does not involve sk specific info, it checks whether the sk has recent synq overflow event in "tcp_synq_no_recent_overflow()". The tcp_sk(sk)->rx_opt.ts_recent_stamp is updated every second when it has sent out a syncookie (through "tcp_synq_overflow()"). The above per sk "recent synq overflow event timestamp" works well for non SO_REUSEPORT use case. However, it may cause random connection request reject/discard when SO_REUSEPORT is used with syncookie because it fails the "tcp_synq_no_recent_overflow()" test. When SO_REUSEPORT is used, it usually has multiple listening socks serving TCP connection requests destinated to the same local IP:PORT. There are cases that the TCP-ACK-COOKIE may not be received by the same sk that sent out the syncookie. For example, if reuse->socks[] began with {sk0, sk1}, 1) sk1 sent out syncookies and tcp_sk(sk1)->rx_opt.ts_recent_stamp was updated. 2) the reuse->socks[] became {sk1, sk2} later. e.g. sk0 was first closed and then sk2 was added. Here, sk2 does not have ts_recent_stamp set. There are other ordering that will trigger the similar situation below but the idea is the same. 3) When the TCP-ACK-COOKIE comes back, sk2 was selected. "tcp_synq_no_recent_overflow(sk2)" returns true. In this case, all syncookies sent by sk1 will be handled (and rejected) by sk2 while sk1 is still alive. The userspace may create and remove listening SO_REUSEPORT sockets as it sees fit. E.g. Adding new thread (and SO_REUSEPORT sock) to handle incoming requests, old process stopping and new process starting...etc. With or without SO_ATTACH_REUSEPORT_[CB]BPF, the sockets leaving and joining a reuseport group makes picking the same sk to check the syncookie very difficult (if not impossible). The later patches will allow bpf prog more flexibility in deciding where a sk should be located in a bpf map and selecting a particular SO_REUSEPORT sock as it sees fit. e.g. Without closing any sock, replace the whole bpf reuseport_array in one map_update() by using map-in-map. Getting the syncookie check working smoothly across socks in the same "reuse->socks[]" is important. A partial solution is to set the newly added sk's ts_recent_stamp to the max ts_recent_stamp of a reuseport group but that will require to iterate through reuse->socks[] OR pessimistically set it to "now - TCP_SYNCOOKIE_VALID" when a sk is joining a reuseport group. However, neither of them will solve the existing sk getting moved around the reuse->socks[] and that sk may not have ts_recent_stamp updated, unlikely under continuous synflood but not impossible. This patch opts to treat the reuseport group as a whole when considering the last synq overflow timestamp since they are serving the same IP:PORT from the userspace (and BPF program) perspective. "synq_overflow_ts" is added to "struct sock_reuseport". The tcp_synq_overflow() and tcp_synq_no_recent_overflow() will update/check reuse->synq_overflow_ts if the sk is in a reuseport group. Similar to the reuseport decision in __inet_lookup_listener(), both sk->sk_reuseport and sk->sk_reuseport_cb are tested for SO_REUSEPORT usage. Update on "synq_overflow_ts" happens at roughly once every second. A synflood test was done with a 16 rx-queues and 16 reuseport sockets. No meaningful performance change is observed. Before and after the change is ~9Mpps in IPv4. Cc: Eric Dumazet Signed-off-by: Martin KaFai Lau Acked-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann --- include/net/sock_reuseport.h | 4 ++++ include/net/tcp.h | 30 ++++++++++++++++++++++++++++-- net/core/sock_reuseport.c | 1 + 3 files changed, 33 insertions(+), 2 deletions(-) diff --git a/include/net/sock_reuseport.h b/include/net/sock_reuseport.h index 0054b3a9b923..6bef7a0052f2 100644 --- a/include/net/sock_reuseport.h +++ b/include/net/sock_reuseport.h @@ -12,6 +12,10 @@ struct sock_reuseport { u16 max_socks; /* length of socks */ u16 num_socks; /* elements in socks */ + /* The last synq overflow event timestamp of this + * reuse->socks[] group. + */ + unsigned int synq_overflow_ts; struct bpf_prog __rcu *prog; /* optional BPF sock selector */ struct sock *socks[0]; /* array of sock pointers */ }; diff --git a/include/net/tcp.h b/include/net/tcp.h index d769dc20359b..d196901c9dba 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -36,6 +36,7 @@ #include #include #include +#include #include #include #include @@ -473,9 +474,22 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb); */ static inline void tcp_synq_overflow(const struct sock *sk) { - unsigned int last_overflow = tcp_sk(sk)->rx_opt.ts_recent_stamp; + unsigned int last_overflow; unsigned int now = jiffies; + if (sk->sk_reuseport) { + struct sock_reuseport *reuse; + + reuse = rcu_dereference(sk->sk_reuseport_cb); + if (likely(reuse)) { + last_overflow = READ_ONCE(reuse->synq_overflow_ts); + if (time_after32(now, last_overflow + HZ)) + WRITE_ONCE(reuse->synq_overflow_ts, now); + return; + } + } + + last_overflow = tcp_sk(sk)->rx_opt.ts_recent_stamp; if (time_after32(now, last_overflow + HZ)) tcp_sk(sk)->rx_opt.ts_recent_stamp = now; } @@ -483,9 +497,21 @@ static inline void tcp_synq_overflow(const struct sock *sk) /* syncookies: no recent synqueue overflow on this listening socket? */ static inline bool tcp_synq_no_recent_overflow(const struct sock *sk) { - unsigned int last_overflow = tcp_sk(sk)->rx_opt.ts_recent_stamp; + unsigned int last_overflow; unsigned int now = jiffies; + if (sk->sk_reuseport) { + struct sock_reuseport *reuse; + + reuse = rcu_dereference(sk->sk_reuseport_cb); + if (likely(reuse)) { + last_overflow = READ_ONCE(reuse->synq_overflow_ts); + return time_after32(now, last_overflow + + TCP_SYNCOOKIE_VALID); + } + } + + last_overflow = tcp_sk(sk)->rx_opt.ts_recent_stamp; return time_after32(now, last_overflow + TCP_SYNCOOKIE_VALID); } diff --git a/net/core/sock_reuseport.c b/net/core/sock_reuseport.c index 064acb04be0f..3f188fad0162 100644 --- a/net/core/sock_reuseport.c +++ b/net/core/sock_reuseport.c @@ -81,6 +81,7 @@ static struct sock_reuseport *reuseport_grow(struct sock_reuseport *reuse) memcpy(more_reuse->socks, reuse->socks, reuse->num_socks * sizeof(struct sock *)); + more_reuse->synq_overflow_ts = READ_ONCE(reuse->synq_overflow_ts); for (i = 0; i < reuse->num_socks; ++i) rcu_assign_pointer(reuse->socks[i]->sk_reuseport_cb, From 736b46027eb4a4c602d3b8b93d2f48c9facbd915 Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Wed, 8 Aug 2018 01:01:22 -0700 Subject: [PATCH 2/9] net: Add ID (if needed) to sock_reuseport and expose reuseport_lock A later patch will introduce a BPF_MAP_TYPE_REUSEPORT_ARRAY which allows a SO_REUSEPORT sk to be added to a bpf map. When a sk is removed from reuse->socks[], it also needs to be removed from the bpf map. Also, when adding a sk to a bpf map, the bpf map needs to ensure it is indeed in a reuse->socks[]. Hence, reuseport_lock is needed by the bpf map to ensure its map_update_elem() and map_delete_elem() operations are in-sync with the reuse->socks[]. The BPF_MAP_TYPE_REUSEPORT_ARRAY map will only acquire the reuseport_lock after ensuring the adding sk is already in a reuseport group (i.e. reuse->socks[]). The map_lookup_elem() will be lockless. This patch also adds an ID to sock_reuseport. A later patch will introduce BPF_PROG_TYPE_SK_REUSEPORT which allows a bpf prog to select a sk from a bpf map. It is inflexible to statically enforce a bpf map can only contain the sk belonging to a particular reuse->socks[] (i.e. same IP:PORT) during the bpf verification time. For example, think about the the map-in-map situation where the inner map can be dynamically changed in runtime and the outer map may have inner maps belonging to different reuseport groups. Hence, when the bpf prog (in the new BPF_PROG_TYPE_SK_REUSEPORT type) selects a sk, this selected sk has to be checked to ensure it belongs to the requesting reuseport group (i.e. the group serving that IP:PORT). The "sk->sk_reuseport_cb" pointer cannot be used for this checking purpose because the pointer value will change after reuseport_grow(). Instead of saving all checking conditions like the ones preced calling "reuseport_add_sock()" and compare them everytime a bpf_prog is run, a 32bits ID is introduced to survive the reuseport_grow(). The ID is only acquired if any of the reuse->socks[] is added to the newly introduced "BPF_MAP_TYPE_REUSEPORT_ARRAY" map. If "BPF_MAP_TYPE_REUSEPORT_ARRAY" is not used, the changes in this patch is a no-op. Signed-off-by: Martin KaFai Lau Acked-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann --- include/net/sock_reuseport.h | 6 ++++++ net/core/sock_reuseport.c | 27 ++++++++++++++++++++++++++- 2 files changed, 32 insertions(+), 1 deletion(-) diff --git a/include/net/sock_reuseport.h b/include/net/sock_reuseport.h index 6bef7a0052f2..e1a7681856f7 100644 --- a/include/net/sock_reuseport.h +++ b/include/net/sock_reuseport.h @@ -5,8 +5,11 @@ #include #include #include +#include #include +extern spinlock_t reuseport_lock; + struct sock_reuseport { struct rcu_head rcu; @@ -16,6 +19,8 @@ struct sock_reuseport { * reuse->socks[] group. */ unsigned int synq_overflow_ts; + /* ID stays the same even after the size of socks[] grows. */ + unsigned int reuseport_id; struct bpf_prog __rcu *prog; /* optional BPF sock selector */ struct sock *socks[0]; /* array of sock pointers */ }; @@ -29,5 +34,6 @@ extern struct sock *reuseport_select_sock(struct sock *sk, int hdr_len); extern struct bpf_prog *reuseport_attach_prog(struct sock *sk, struct bpf_prog *prog); +int reuseport_get_id(struct sock_reuseport *reuse); #endif /* _SOCK_REUSEPORT_H */ diff --git a/net/core/sock_reuseport.c b/net/core/sock_reuseport.c index 3f188fad0162..cf2e4d305af9 100644 --- a/net/core/sock_reuseport.c +++ b/net/core/sock_reuseport.c @@ -8,11 +8,33 @@ #include #include +#include #include #define INIT_SOCKS 128 -static DEFINE_SPINLOCK(reuseport_lock); +DEFINE_SPINLOCK(reuseport_lock); + +#define REUSEPORT_MIN_ID 1 +static DEFINE_IDA(reuseport_ida); + +int reuseport_get_id(struct sock_reuseport *reuse) +{ + int id; + + if (reuse->reuseport_id) + return reuse->reuseport_id; + + id = ida_simple_get(&reuseport_ida, REUSEPORT_MIN_ID, 0, + /* Called under reuseport_lock */ + GFP_ATOMIC); + if (id < 0) + return id; + + reuse->reuseport_id = id; + + return reuse->reuseport_id; +} static struct sock_reuseport *__reuseport_alloc(unsigned int max_socks) { @@ -78,6 +100,7 @@ static struct sock_reuseport *reuseport_grow(struct sock_reuseport *reuse) more_reuse->max_socks = more_socks_size; more_reuse->num_socks = reuse->num_socks; more_reuse->prog = reuse->prog; + more_reuse->reuseport_id = reuse->reuseport_id; memcpy(more_reuse->socks, reuse->socks, reuse->num_socks * sizeof(struct sock *)); @@ -102,6 +125,8 @@ static void reuseport_free_rcu(struct rcu_head *head) reuse = container_of(head, struct sock_reuseport, rcu); if (reuse->prog) bpf_prog_destroy(reuse->prog); + if (reuse->reuseport_id) + ida_simple_remove(&reuseport_ida, reuse->reuseport_id); kfree(reuse); } From 5dc4c4b7d4e8115e7cde96a030f98cb3ab2e458c Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Wed, 8 Aug 2018 01:01:24 -0700 Subject: [PATCH 3/9] bpf: Introduce BPF_MAP_TYPE_REUSEPORT_SOCKARRAY This patch introduces a new map type BPF_MAP_TYPE_REUSEPORT_SOCKARRAY. To unleash the full potential of a bpf prog, it is essential for the userspace to be capable of directly setting up a bpf map which can then be consumed by the bpf prog to make decision. In this case, decide which SO_REUSEPORT sk to serve the incoming request. By adding BPF_MAP_TYPE_REUSEPORT_SOCKARRAY, the userspace has total control and visibility on where a SO_REUSEPORT sk should be located in a bpf map. The later patch will introduce BPF_PROG_TYPE_SK_REUSEPORT such that the bpf prog can directly select a sk from the bpf map. That will raise the programmability of the bpf prog attached to a reuseport group (a group of sk serving the same IP:PORT). For example, in UDP, the bpf prog can peek into the payload (e.g. through the "data" pointer introduced in the later patch) to learn the application level's connection information and then decide which sk to pick from a bpf map. The userspace can tightly couple the sk's location in a bpf map with the application logic in generating the UDP payload's connection information. This connection info contact/API stays within the userspace. Also, when used with map-in-map, the userspace can switch the old-server-process's inner map to a new-server-process's inner map in one call "bpf_map_update_elem(outer_map, &index, &new_reuseport_array)". The bpf prog will then direct incoming requests to the new process instead of the old process. The old process can finish draining the pending requests (e.g. by "accept()") before closing the old-fds. [Note that deleting a fd from a bpf map does not necessary mean the fd is closed] During map_update_elem(), Only SO_REUSEPORT sk (i.e. which has already been added to a reuse->socks[]) can be used. That means a SO_REUSEPORT sk that is "bind()" for UDP or "bind()+listen()" for TCP. These conditions are ensured in "reuseport_array_update_check()". A SO_REUSEPORT sk can only be added once to a map (i.e. the same sk cannot be added twice even to the same map). SO_REUSEPORT already allows another sk to be created for the same IP:PORT. There is no need to re-create a similar usage in the BPF side. When a SO_REUSEPORT is deleted from the "reuse->socks[]" (e.g. "close()"), it will notify the bpf map to remove it from the map also. It is done through "bpf_sk_reuseport_detach()" and it will only be called if >=1 of the "reuse->sock[]" has ever been added to a bpf map. The map_update()/map_delete() has to be in-sync with the "reuse->socks[]". Hence, the same "reuseport_lock" used by "reuse->socks[]" has to be used here also. Care has been taken to ensure the lock is only acquired when the adding sk passes some strict tests. and freeing the map does not require the reuseport_lock. The reuseport_array will also support lookup from the syscall side. It will return a sock_gen_cookie(). The sock_gen_cookie() is on-demand (i.e. a sk's cookie is not generated until the very first map_lookup_elem()). The lookup cookie is 64bits but it goes against the logical userspace expectation on 32bits sizeof(fd) (and as other fd based bpf maps do also). It may catch user in surprise if we enforce value_size=8 while userspace still pass a 32bits fd during update. Supporting different value_size between lookup and update seems unintuitive also. We also need to consider what if other existing fd based maps want to return 64bits value from syscall's lookup in the future. Hence, reuseport_array supports both value_size 4 and 8, and assuming user will usually use value_size=4. The syscall's lookup will return ENOSPC on value_size=4. It will will only return 64bits value from sock_gen_cookie() when user consciously choose value_size=8 (as a signal that lookup is desired) which then requires a 64bits value in both lookup and update. Signed-off-by: Martin KaFai Lau Acked-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann --- include/linux/bpf.h | 28 +++ include/linux/bpf_types.h | 3 + include/uapi/linux/bpf.h | 1 + kernel/bpf/Makefile | 3 + kernel/bpf/arraymap.c | 2 +- kernel/bpf/reuseport_array.c | 363 +++++++++++++++++++++++++++++++++++ kernel/bpf/syscall.c | 6 + net/core/sock_reuseport.c | 8 + 8 files changed, 413 insertions(+), 1 deletion(-) create mode 100644 kernel/bpf/reuseport_array.c diff --git a/include/linux/bpf.h b/include/linux/bpf.h index cd8790d2c6ed..db11662faea6 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -524,6 +524,7 @@ static inline int bpf_map_attr_numa_node(const union bpf_attr *attr) } struct bpf_prog *bpf_prog_get_type_path(const char *name, enum bpf_prog_type type); +int array_map_alloc_check(union bpf_attr *attr); #else /* !CONFIG_BPF_SYSCALL */ static inline struct bpf_prog *bpf_prog_get(u32 ufd) @@ -769,6 +770,33 @@ static inline void __xsk_map_flush(struct bpf_map *map) } #endif +#if defined(CONFIG_INET) && defined(CONFIG_BPF_SYSCALL) +void bpf_sk_reuseport_detach(struct sock *sk); +int bpf_fd_reuseport_array_lookup_elem(struct bpf_map *map, void *key, + void *value); +int bpf_fd_reuseport_array_update_elem(struct bpf_map *map, void *key, + void *value, u64 map_flags); +#else +static inline void bpf_sk_reuseport_detach(struct sock *sk) +{ +} + +#ifdef CONFIG_BPF_SYSCALL +static inline int bpf_fd_reuseport_array_lookup_elem(struct bpf_map *map, + void *key, void *value) +{ + return -EOPNOTSUPP; +} + +static inline int bpf_fd_reuseport_array_update_elem(struct bpf_map *map, + void *key, void *value, + u64 map_flags) +{ + return -EOPNOTSUPP; +} +#endif /* CONFIG_BPF_SYSCALL */ +#endif /* defined(CONFIG_INET) && defined(CONFIG_BPF_SYSCALL) */ + /* verifier prototypes for helper functions called from eBPF programs */ extern const struct bpf_func_proto bpf_map_lookup_elem_proto; extern const struct bpf_func_proto bpf_map_update_elem_proto; diff --git a/include/linux/bpf_types.h b/include/linux/bpf_types.h index add08be53b6f..14fd6c02d258 100644 --- a/include/linux/bpf_types.h +++ b/include/linux/bpf_types.h @@ -60,4 +60,7 @@ BPF_MAP_TYPE(BPF_MAP_TYPE_CPUMAP, cpu_map_ops) #if defined(CONFIG_XDP_SOCKETS) BPF_MAP_TYPE(BPF_MAP_TYPE_XSKMAP, xsk_map_ops) #endif +#ifdef CONFIG_INET +BPF_MAP_TYPE(BPF_MAP_TYPE_REUSEPORT_SOCKARRAY, reuseport_array_ops) +#endif #endif diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index dd5758dc35d3..40f584bc7da0 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -126,6 +126,7 @@ enum bpf_map_type { BPF_MAP_TYPE_XSKMAP, BPF_MAP_TYPE_SOCKHASH, BPF_MAP_TYPE_CGROUP_STORAGE, + BPF_MAP_TYPE_REUSEPORT_SOCKARRAY, }; enum bpf_prog_type { diff --git a/kernel/bpf/Makefile b/kernel/bpf/Makefile index e8906cbad81f..0488b8258321 100644 --- a/kernel/bpf/Makefile +++ b/kernel/bpf/Makefile @@ -23,3 +23,6 @@ ifeq ($(CONFIG_PERF_EVENTS),y) obj-$(CONFIG_BPF_SYSCALL) += stackmap.o endif obj-$(CONFIG_CGROUP_BPF) += cgroup.o +ifeq ($(CONFIG_INET),y) +obj-$(CONFIG_BPF_SYSCALL) += reuseport_array.o +endif diff --git a/kernel/bpf/arraymap.c b/kernel/bpf/arraymap.c index 2aa55d030c77..f6ca3e712831 100644 --- a/kernel/bpf/arraymap.c +++ b/kernel/bpf/arraymap.c @@ -54,7 +54,7 @@ static int bpf_array_alloc_percpu(struct bpf_array *array) } /* Called from syscall */ -static int array_map_alloc_check(union bpf_attr *attr) +int array_map_alloc_check(union bpf_attr *attr) { bool percpu = attr->map_type == BPF_MAP_TYPE_PERCPU_ARRAY; int numa_node = bpf_map_attr_numa_node(attr); diff --git a/kernel/bpf/reuseport_array.c b/kernel/bpf/reuseport_array.c new file mode 100644 index 000000000000..18e225de80ff --- /dev/null +++ b/kernel/bpf/reuseport_array.c @@ -0,0 +1,363 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (c) 2018 Facebook + */ +#include +#include +#include +#include + +struct reuseport_array { + struct bpf_map map; + struct sock __rcu *ptrs[]; +}; + +static struct reuseport_array *reuseport_array(struct bpf_map *map) +{ + return (struct reuseport_array *)map; +} + +/* The caller must hold the reuseport_lock */ +void bpf_sk_reuseport_detach(struct sock *sk) +{ + struct sock __rcu **socks; + + write_lock_bh(&sk->sk_callback_lock); + socks = sk->sk_user_data; + if (socks) { + WRITE_ONCE(sk->sk_user_data, NULL); + /* + * Do not move this NULL assignment outside of + * sk->sk_callback_lock because there is + * a race with reuseport_array_free() + * which does not hold the reuseport_lock. + */ + RCU_INIT_POINTER(*socks, NULL); + } + write_unlock_bh(&sk->sk_callback_lock); +} + +static int reuseport_array_alloc_check(union bpf_attr *attr) +{ + if (attr->value_size != sizeof(u32) && + attr->value_size != sizeof(u64)) + return -EINVAL; + + return array_map_alloc_check(attr); +} + +static void *reuseport_array_lookup_elem(struct bpf_map *map, void *key) +{ + struct reuseport_array *array = reuseport_array(map); + u32 index = *(u32 *)key; + + if (unlikely(index >= array->map.max_entries)) + return NULL; + + return rcu_dereference(array->ptrs[index]); +} + +/* Called from syscall only */ +static int reuseport_array_delete_elem(struct bpf_map *map, void *key) +{ + struct reuseport_array *array = reuseport_array(map); + u32 index = *(u32 *)key; + struct sock *sk; + int err; + + if (index >= map->max_entries) + return -E2BIG; + + if (!rcu_access_pointer(array->ptrs[index])) + return -ENOENT; + + spin_lock_bh(&reuseport_lock); + + sk = rcu_dereference_protected(array->ptrs[index], + lockdep_is_held(&reuseport_lock)); + if (sk) { + write_lock_bh(&sk->sk_callback_lock); + WRITE_ONCE(sk->sk_user_data, NULL); + RCU_INIT_POINTER(array->ptrs[index], NULL); + write_unlock_bh(&sk->sk_callback_lock); + err = 0; + } else { + err = -ENOENT; + } + + spin_unlock_bh(&reuseport_lock); + + return err; +} + +static void reuseport_array_free(struct bpf_map *map) +{ + struct reuseport_array *array = reuseport_array(map); + struct sock *sk; + u32 i; + + synchronize_rcu(); + + /* + * ops->map_*_elem() will not be able to access this + * array now. Hence, this function only races with + * bpf_sk_reuseport_detach() which was triggerred by + * close() or disconnect(). + * + * This function and bpf_sk_reuseport_detach() are + * both removing sk from "array". Who removes it + * first does not matter. + * + * The only concern here is bpf_sk_reuseport_detach() + * may access "array" which is being freed here. + * bpf_sk_reuseport_detach() access this "array" + * through sk->sk_user_data _and_ with sk->sk_callback_lock + * held which is enough because this "array" is not freed + * until all sk->sk_user_data has stopped referencing this "array". + * + * Hence, due to the above, taking "reuseport_lock" is not + * needed here. + */ + + /* + * Since reuseport_lock is not taken, sk is accessed under + * rcu_read_lock() + */ + rcu_read_lock(); + for (i = 0; i < map->max_entries; i++) { + sk = rcu_dereference(array->ptrs[i]); + if (sk) { + write_lock_bh(&sk->sk_callback_lock); + /* + * No need for WRITE_ONCE(). At this point, + * no one is reading it without taking the + * sk->sk_callback_lock. + */ + sk->sk_user_data = NULL; + write_unlock_bh(&sk->sk_callback_lock); + RCU_INIT_POINTER(array->ptrs[i], NULL); + } + } + rcu_read_unlock(); + + /* + * Once reaching here, all sk->sk_user_data is not + * referenceing this "array". "array" can be freed now. + */ + bpf_map_area_free(array); +} + +static struct bpf_map *reuseport_array_alloc(union bpf_attr *attr) +{ + int err, numa_node = bpf_map_attr_numa_node(attr); + struct reuseport_array *array; + u64 cost, array_size; + + if (!capable(CAP_SYS_ADMIN)) + return ERR_PTR(-EPERM); + + array_size = sizeof(*array); + array_size += (u64)attr->max_entries * sizeof(struct sock *); + + /* make sure there is no u32 overflow later in round_up() */ + cost = array_size; + if (cost >= U32_MAX - PAGE_SIZE) + return ERR_PTR(-ENOMEM); + cost = round_up(cost, PAGE_SIZE) >> PAGE_SHIFT; + + err = bpf_map_precharge_memlock(cost); + if (err) + return ERR_PTR(err); + + /* allocate all map elements and zero-initialize them */ + array = bpf_map_area_alloc(array_size, numa_node); + if (!array) + return ERR_PTR(-ENOMEM); + + /* copy mandatory map attributes */ + bpf_map_init_from_attr(&array->map, attr); + array->map.pages = cost; + + return &array->map; +} + +int bpf_fd_reuseport_array_lookup_elem(struct bpf_map *map, void *key, + void *value) +{ + struct sock *sk; + int err; + + if (map->value_size != sizeof(u64)) + return -ENOSPC; + + rcu_read_lock(); + sk = reuseport_array_lookup_elem(map, key); + if (sk) { + *(u64 *)value = sock_gen_cookie(sk); + err = 0; + } else { + err = -ENOENT; + } + rcu_read_unlock(); + + return err; +} + +static int +reuseport_array_update_check(const struct reuseport_array *array, + const struct sock *nsk, + const struct sock *osk, + const struct sock_reuseport *nsk_reuse, + u32 map_flags) +{ + if (osk && map_flags == BPF_NOEXIST) + return -EEXIST; + + if (!osk && map_flags == BPF_EXIST) + return -ENOENT; + + if (nsk->sk_protocol != IPPROTO_UDP && nsk->sk_protocol != IPPROTO_TCP) + return -ENOTSUPP; + + if (nsk->sk_family != AF_INET && nsk->sk_family != AF_INET6) + return -ENOTSUPP; + + if (nsk->sk_type != SOCK_STREAM && nsk->sk_type != SOCK_DGRAM) + return -ENOTSUPP; + + /* + * sk must be hashed (i.e. listening in the TCP case or binded + * in the UDP case) and + * it must also be a SO_REUSEPORT sk (i.e. reuse cannot be NULL). + * + * Also, sk will be used in bpf helper that is protected by + * rcu_read_lock(). + */ + if (!sock_flag(nsk, SOCK_RCU_FREE) || !sk_hashed(nsk) || !nsk_reuse) + return -EINVAL; + + /* READ_ONCE because the sk->sk_callback_lock may not be held here */ + if (READ_ONCE(nsk->sk_user_data)) + return -EBUSY; + + return 0; +} + +/* + * Called from syscall only. + * The "nsk" in the fd refcnt. + * The "osk" and "reuse" are protected by reuseport_lock. + */ +int bpf_fd_reuseport_array_update_elem(struct bpf_map *map, void *key, + void *value, u64 map_flags) +{ + struct reuseport_array *array = reuseport_array(map); + struct sock *free_osk = NULL, *osk, *nsk; + struct sock_reuseport *reuse; + u32 index = *(u32 *)key; + struct socket *socket; + int err, fd; + + if (map_flags > BPF_EXIST) + return -EINVAL; + + if (index >= map->max_entries) + return -E2BIG; + + if (map->value_size == sizeof(u64)) { + u64 fd64 = *(u64 *)value; + + if (fd64 > S32_MAX) + return -EINVAL; + fd = fd64; + } else { + fd = *(int *)value; + } + + socket = sockfd_lookup(fd, &err); + if (!socket) + return err; + + nsk = socket->sk; + if (!nsk) { + err = -EINVAL; + goto put_file; + } + + /* Quick checks before taking reuseport_lock */ + err = reuseport_array_update_check(array, nsk, + rcu_access_pointer(array->ptrs[index]), + rcu_access_pointer(nsk->sk_reuseport_cb), + map_flags); + if (err) + goto put_file; + + spin_lock_bh(&reuseport_lock); + /* + * Some of the checks only need reuseport_lock + * but it is done under sk_callback_lock also + * for simplicity reason. + */ + write_lock_bh(&nsk->sk_callback_lock); + + osk = rcu_dereference_protected(array->ptrs[index], + lockdep_is_held(&reuseport_lock)); + reuse = rcu_dereference_protected(nsk->sk_reuseport_cb, + lockdep_is_held(&reuseport_lock)); + err = reuseport_array_update_check(array, nsk, osk, reuse, map_flags); + if (err) + goto put_file_unlock; + + /* Ensure reuse->reuseport_id is set */ + err = reuseport_get_id(reuse); + if (err < 0) + goto put_file_unlock; + + WRITE_ONCE(nsk->sk_user_data, &array->ptrs[index]); + rcu_assign_pointer(array->ptrs[index], nsk); + free_osk = osk; + err = 0; + +put_file_unlock: + write_unlock_bh(&nsk->sk_callback_lock); + + if (free_osk) { + write_lock_bh(&free_osk->sk_callback_lock); + WRITE_ONCE(free_osk->sk_user_data, NULL); + write_unlock_bh(&free_osk->sk_callback_lock); + } + + spin_unlock_bh(&reuseport_lock); +put_file: + fput(socket->file); + return err; +} + +/* Called from syscall */ +static int reuseport_array_get_next_key(struct bpf_map *map, void *key, + void *next_key) +{ + struct reuseport_array *array = reuseport_array(map); + u32 index = key ? *(u32 *)key : U32_MAX; + u32 *next = (u32 *)next_key; + + if (index >= array->map.max_entries) { + *next = 0; + return 0; + } + + if (index == array->map.max_entries - 1) + return -ENOENT; + + *next = index + 1; + return 0; +} + +const struct bpf_map_ops reuseport_array_ops = { + .map_alloc_check = reuseport_array_alloc_check, + .map_alloc = reuseport_array_alloc, + .map_free = reuseport_array_free, + .map_lookup_elem = reuseport_array_lookup_elem, + .map_get_next_key = reuseport_array_get_next_key, + .map_delete_elem = reuseport_array_delete_elem, +}; diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c index 5af4e9e2722d..57f4d076141b 100644 --- a/kernel/bpf/syscall.c +++ b/kernel/bpf/syscall.c @@ -684,6 +684,8 @@ static int map_lookup_elem(union bpf_attr *attr) err = bpf_fd_array_map_lookup_elem(map, key, value); } else if (IS_FD_HASH(map)) { 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 { rcu_read_lock(); ptr = map->ops->map_lookup_elem(map, key); @@ -790,6 +792,10 @@ static int map_update_elem(union bpf_attr *attr) err = bpf_fd_htab_map_update_elem(map, f.file, key, value, attr->flags); rcu_read_unlock(); + } else if (map->map_type == BPF_MAP_TYPE_REUSEPORT_SOCKARRAY) { + /* rcu_read_lock() is not needed */ + err = bpf_fd_reuseport_array_update_elem(map, key, value, + attr->flags); } else { rcu_read_lock(); err = map->ops->map_update_elem(map, key, value, attr->flags); diff --git a/net/core/sock_reuseport.c b/net/core/sock_reuseport.c index cf2e4d305af9..8235f2439816 100644 --- a/net/core/sock_reuseport.c +++ b/net/core/sock_reuseport.c @@ -186,6 +186,14 @@ void reuseport_detach_sock(struct sock *sk) spin_lock_bh(&reuseport_lock); reuse = rcu_dereference_protected(sk->sk_reuseport_cb, lockdep_is_held(&reuseport_lock)); + + /* At least one of the sk in this reuseport group is added to + * a bpf map. Notify the bpf side. The bpf map logic will + * remove the sk if it is indeed added to a bpf map. + */ + if (reuse->reuseport_id) + bpf_sk_reuseport_detach(sk); + rcu_assign_pointer(sk->sk_reuseport_cb, NULL); for (i = 0; i < reuse->num_socks; i++) { From 2dbb9b9e6df67d444fbe425c7f6014858d337adf Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Wed, 8 Aug 2018 01:01:25 -0700 Subject: [PATCH 4/9] bpf: Introduce BPF_PROG_TYPE_SK_REUSEPORT This patch adds a BPF_PROG_TYPE_SK_REUSEPORT which can select a SO_REUSEPORT sk from a BPF_MAP_TYPE_REUSEPORT_ARRAY. Like other non SK_FILTER/CGROUP_SKB program, it requires CAP_SYS_ADMIN. BPF_PROG_TYPE_SK_REUSEPORT introduces "struct sk_reuseport_kern" to store the bpf context instead of using the skb->cb[48]. At the SO_REUSEPORT sk lookup time, it is in the middle of transiting from a lower layer (ipv4/ipv6) to a upper layer (udp/tcp). At this point, it is not always clear where the bpf context can be appended in the skb->cb[48] to avoid saving-and-restoring cb[]. Even putting aside the difference between ipv4-vs-ipv6 and udp-vs-tcp. It is not clear if the lower layer is only ipv4 and ipv6 in the future and will it not touch the cb[] again before transiting to the upper layer. For example, in udp_gro_receive(), it uses the 48 byte NAPI_GRO_CB instead of IP[6]CB and it may still modify the cb[] after calling the udp[46]_lib_lookup_skb(). Because of the above reason, if sk->cb is used for the bpf ctx, saving-and-restoring is needed and likely the whole 48 bytes cb[] has to be saved and restored. Instead of saving, setting and restoring the cb[], this patch opts to create a new "struct sk_reuseport_kern" and setting the needed values in there. The new BPF_PROG_TYPE_SK_REUSEPORT and "struct sk_reuseport_(kern|md)" will serve all ipv4/ipv6 + udp/tcp combinations. There is no protocol specific usage at this point and it is also inline with the current sock_reuseport.c implementation (i.e. no protocol specific requirement). In "struct sk_reuseport_md", this patch exposes data/data_end/len with semantic similar to other existing usages. Together with "bpf_skb_load_bytes()" and "bpf_skb_load_bytes_relative()", the bpf prog can peek anywhere in the skb. The "bind_inany" tells the bpf prog that the reuseport group is bind-ed to a local INANY address which cannot be learned from skb. The new "bind_inany" is added to "struct sock_reuseport" which will be used when running the new "BPF_PROG_TYPE_SK_REUSEPORT" bpf prog in order to avoid repeating the "bind INANY" test on "sk_v6_rcv_saddr/sk->sk_rcv_saddr" every time a bpf prog is run. It can only be properly initialized when a "sk->sk_reuseport" enabled sk is adding to a hashtable (i.e. during "reuseport_alloc()" and "reuseport_add_sock()"). The new "sk_select_reuseport()" is the main helper that the bpf prog will use to select a SO_REUSEPORT sk. It is the only function that can use the new BPF_MAP_TYPE_REUSEPORT_ARRAY. As mentioned in the earlier patch, the validity of a selected sk is checked in run time in "sk_select_reuseport()". Doing the check in verification time is difficult and inflexible (consider the map-in-map use case). The runtime check is to compare the selected sk's reuseport_id with the reuseport_id that we want. This helper will return -EXXX if the selected sk cannot serve the incoming request (e.g. reuseport_id not match). The bpf prog can decide if it wants to do SK_DROP as its discretion. When the bpf prog returns SK_PASS, the kernel will check if a valid sk has been selected (i.e. "reuse_kern->selected_sk != NULL"). If it does , it will use the selected sk. If not, the kernel will select one from "reuse->socks[]" (as before this patch). The SK_DROP and SK_PASS handling logic will be in the next patch. Signed-off-by: Martin KaFai Lau Acked-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann --- include/linux/bpf_types.h | 3 + include/linux/filter.h | 15 ++ include/net/addrconf.h | 1 + include/net/sock_reuseport.h | 6 +- include/uapi/linux/bpf.h | 36 ++++- kernel/bpf/verifier.c | 9 ++ net/core/filter.c | 269 +++++++++++++++++++++++++++++++- net/core/sock_reuseport.c | 20 ++- net/ipv4/inet_connection_sock.c | 9 ++ net/ipv4/inet_hashtables.c | 5 +- net/ipv4/udp.c | 5 +- 11 files changed, 365 insertions(+), 13 deletions(-) diff --git a/include/linux/bpf_types.h b/include/linux/bpf_types.h index 14fd6c02d258..cd26c090e7c0 100644 --- a/include/linux/bpf_types.h +++ b/include/linux/bpf_types.h @@ -29,6 +29,9 @@ BPF_PROG_TYPE(BPF_PROG_TYPE_CGROUP_DEVICE, cg_dev) #ifdef CONFIG_BPF_LIRC_MODE2 BPF_PROG_TYPE(BPF_PROG_TYPE_LIRC_MODE2, lirc_mode2) #endif +#ifdef CONFIG_INET +BPF_PROG_TYPE(BPF_PROG_TYPE_SK_REUSEPORT, sk_reuseport) +#endif BPF_MAP_TYPE(BPF_MAP_TYPE_ARRAY, array_map_ops) BPF_MAP_TYPE(BPF_MAP_TYPE_PERCPU_ARRAY, percpu_array_map_ops) diff --git a/include/linux/filter.h b/include/linux/filter.h index 2b072dab32c0..70e9d57677fe 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -32,6 +32,7 @@ struct seccomp_data; struct bpf_prog_aux; struct xdp_rxq_info; struct xdp_buff; +struct sock_reuseport; /* ArgX, context and stack frame pointer register positions. Note, * Arg1, Arg2, Arg3, etc are used as argument mappings of function @@ -833,6 +834,20 @@ void bpf_warn_invalid_xdp_action(u32 act); struct sock *do_sk_redirect_map(struct sk_buff *skb); struct sock *do_msg_redirect_map(struct sk_msg_buff *md); +#ifdef CONFIG_INET +struct sock *bpf_run_sk_reuseport(struct sock_reuseport *reuse, struct sock *sk, + struct bpf_prog *prog, struct sk_buff *skb, + u32 hash); +#else +static inline struct sock * +bpf_run_sk_reuseport(struct sock_reuseport *reuse, struct sock *sk, + struct bpf_prog *prog, struct sk_buff *skb, + u32 hash) +{ + return NULL; +} +#endif + #ifdef CONFIG_BPF_JIT extern int bpf_jit_enable; extern int bpf_jit_harden; diff --git a/include/net/addrconf.h b/include/net/addrconf.h index 5f43f7a70fe6..6def0351bcc3 100644 --- a/include/net/addrconf.h +++ b/include/net/addrconf.h @@ -108,6 +108,7 @@ int ipv6_get_lladdr(struct net_device *dev, struct in6_addr *addr, u32 banned_flags); bool inet_rcv_saddr_equal(const struct sock *sk, const struct sock *sk2, bool match_wildcard); +bool inet_rcv_saddr_any(const struct sock *sk); void addrconf_join_solict(struct net_device *dev, const struct in6_addr *addr); void addrconf_leave_solict(struct inet6_dev *idev, const struct in6_addr *addr); diff --git a/include/net/sock_reuseport.h b/include/net/sock_reuseport.h index e1a7681856f7..73b569556be6 100644 --- a/include/net/sock_reuseport.h +++ b/include/net/sock_reuseport.h @@ -21,12 +21,14 @@ struct sock_reuseport { unsigned int synq_overflow_ts; /* ID stays the same even after the size of socks[] grows. */ unsigned int reuseport_id; + bool bind_inany; struct bpf_prog __rcu *prog; /* optional BPF sock selector */ struct sock *socks[0]; /* array of sock pointers */ }; -extern int reuseport_alloc(struct sock *sk); -extern int reuseport_add_sock(struct sock *sk, struct sock *sk2); +extern int reuseport_alloc(struct sock *sk, bool bind_inany); +extern int reuseport_add_sock(struct sock *sk, struct sock *sk2, + bool bind_inany); extern void reuseport_detach_sock(struct sock *sk); extern struct sock *reuseport_select_sock(struct sock *sk, u32 hash, diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index 40f584bc7da0..3102a2a23c31 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -151,6 +151,7 @@ enum bpf_prog_type { BPF_PROG_TYPE_CGROUP_SOCK_ADDR, BPF_PROG_TYPE_LWT_SEG6LOCAL, BPF_PROG_TYPE_LIRC_MODE2, + BPF_PROG_TYPE_SK_REUSEPORT, }; enum bpf_attach_type { @@ -2114,6 +2115,14 @@ union bpf_attr { * the shared data. * Return * Pointer to the local storage area. + * + * int bpf_sk_select_reuseport(struct sk_reuseport_md *reuse, struct bpf_map *map, void *key, u64 flags) + * Description + * Select a SO_REUSEPORT sk from a BPF_MAP_TYPE_REUSEPORT_ARRAY map + * It checks the selected sk is matching the incoming + * request in the skb. + * Return + * 0 on success, or a negative error in case of failure. */ #define __BPF_FUNC_MAPPER(FN) \ FN(unspec), \ @@ -2197,7 +2206,8 @@ union bpf_attr { FN(rc_keydown), \ FN(skb_cgroup_id), \ FN(get_current_cgroup_id), \ - FN(get_local_storage), + FN(get_local_storage), \ + FN(sk_select_reuseport), /* integer value in 'imm' field of BPF_CALL instruction selects which helper * function eBPF program intends to call @@ -2414,6 +2424,30 @@ struct sk_msg_md { __u32 local_port; /* stored in host byte order */ }; +struct sk_reuseport_md { + /* + * Start of directly accessible data. It begins from + * the tcp/udp header. + */ + void *data; + void *data_end; /* End of directly accessible data */ + /* + * Total length of packet (starting from the tcp/udp header). + * Note that the directly accessible bytes (data_end - data) + * could be less than this "len". Those bytes could be + * indirectly read by a helper "bpf_skb_load_bytes()". + */ + __u32 len; + /* + * Eth protocol in the mac header (network byte order). e.g. + * ETH_P_IP(0x0800) and ETH_P_IPV6(0x86DD) + */ + __u32 eth_protocol; + __u32 ip_protocol; /* IP protocol. e.g. IPPROTO_TCP, IPPROTO_UDP */ + __u32 bind_inany; /* Is sock bound to an INANY address? */ + __u32 hash; /* A hash of the packet 4 tuples */ +}; + #define BPF_TAG_SIZE 8 struct bpf_prog_info { diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c index 587468a9c37d..ca90679a7fe5 100644 --- a/kernel/bpf/verifier.c +++ b/kernel/bpf/verifier.c @@ -1310,6 +1310,7 @@ static bool may_access_direct_pkt_data(struct bpf_verifier_env *env, case BPF_PROG_TYPE_LWT_IN: case BPF_PROG_TYPE_LWT_OUT: case BPF_PROG_TYPE_LWT_SEG6LOCAL: + case BPF_PROG_TYPE_SK_REUSEPORT: /* dst_input() and dst_output() can't write for now */ if (t == BPF_WRITE) return false; @@ -2166,6 +2167,10 @@ static int check_map_func_compatibility(struct bpf_verifier_env *env, func_id != BPF_FUNC_msg_redirect_hash) goto error; break; + case BPF_MAP_TYPE_REUSEPORT_SOCKARRAY: + if (func_id != BPF_FUNC_sk_select_reuseport) + goto error; + break; default: break; } @@ -2217,6 +2222,10 @@ static int check_map_func_compatibility(struct bpf_verifier_env *env, if (map->map_type != BPF_MAP_TYPE_CGROUP_STORAGE) goto error; break; + case BPF_FUNC_sk_select_reuseport: + if (map->map_type != BPF_MAP_TYPE_REUSEPORT_SOCKARRAY) + goto error; + break; default: break; } diff --git a/net/core/filter.c b/net/core/filter.c index 2de7dd9f2a57..142595b4e0d1 100644 --- a/net/core/filter.c +++ b/net/core/filter.c @@ -1462,7 +1462,7 @@ static int __reuseport_attach_prog(struct bpf_prog *prog, struct sock *sk) return -ENOMEM; if (sk_unhashed(sk) && sk->sk_reuseport) { - err = reuseport_alloc(sk); + err = reuseport_alloc(sk, false); if (err) return err; } else if (!rcu_access_pointer(sk->sk_reuseport_cb)) { @@ -7013,3 +7013,270 @@ out: release_sock(sk); return ret; } + +#ifdef CONFIG_INET +struct sk_reuseport_kern { + struct sk_buff *skb; + struct sock *sk; + struct sock *selected_sk; + void *data_end; + u32 hash; + u32 reuseport_id; + bool bind_inany; +}; + +static void bpf_init_reuseport_kern(struct sk_reuseport_kern *reuse_kern, + struct sock_reuseport *reuse, + struct sock *sk, struct sk_buff *skb, + u32 hash) +{ + reuse_kern->skb = skb; + reuse_kern->sk = sk; + reuse_kern->selected_sk = NULL; + reuse_kern->data_end = skb->data + skb_headlen(skb); + reuse_kern->hash = hash; + reuse_kern->reuseport_id = reuse->reuseport_id; + reuse_kern->bind_inany = reuse->bind_inany; +} + +struct sock *bpf_run_sk_reuseport(struct sock_reuseport *reuse, struct sock *sk, + struct bpf_prog *prog, struct sk_buff *skb, + u32 hash) +{ + struct sk_reuseport_kern reuse_kern; + enum sk_action action; + + bpf_init_reuseport_kern(&reuse_kern, reuse, sk, skb, hash); + action = BPF_PROG_RUN(prog, &reuse_kern); + + if (action == SK_PASS) + return reuse_kern.selected_sk; + else + return ERR_PTR(-ECONNREFUSED); +} + +BPF_CALL_4(sk_select_reuseport, struct sk_reuseport_kern *, reuse_kern, + struct bpf_map *, map, void *, key, u32, flags) +{ + struct sock_reuseport *reuse; + struct sock *selected_sk; + + selected_sk = map->ops->map_lookup_elem(map, key); + if (!selected_sk) + return -ENOENT; + + reuse = rcu_dereference(selected_sk->sk_reuseport_cb); + if (!reuse) + /* selected_sk is unhashed (e.g. by close()) after the + * above map_lookup_elem(). Treat selected_sk has already + * been removed from the map. + */ + return -ENOENT; + + if (unlikely(reuse->reuseport_id != reuse_kern->reuseport_id)) { + struct sock *sk; + + if (unlikely(!reuse_kern->reuseport_id)) + /* There is a small race between adding the + * sk to the map and setting the + * reuse_kern->reuseport_id. + * Treat it as the sk has not been added to + * the bpf map yet. + */ + return -ENOENT; + + sk = reuse_kern->sk; + if (sk->sk_protocol != selected_sk->sk_protocol) + return -EPROTOTYPE; + else if (sk->sk_family != selected_sk->sk_family) + return -EAFNOSUPPORT; + + /* Catch all. Likely bound to a different sockaddr. */ + return -EBADFD; + } + + reuse_kern->selected_sk = selected_sk; + + return 0; +} + +static const struct bpf_func_proto sk_select_reuseport_proto = { + .func = sk_select_reuseport, + .gpl_only = false, + .ret_type = RET_INTEGER, + .arg1_type = ARG_PTR_TO_CTX, + .arg2_type = ARG_CONST_MAP_PTR, + .arg3_type = ARG_PTR_TO_MAP_KEY, + .arg4_type = ARG_ANYTHING, +}; + +BPF_CALL_4(sk_reuseport_load_bytes, + const struct sk_reuseport_kern *, reuse_kern, u32, offset, + void *, to, u32, len) +{ + return ____bpf_skb_load_bytes(reuse_kern->skb, offset, to, len); +} + +static const struct bpf_func_proto sk_reuseport_load_bytes_proto = { + .func = sk_reuseport_load_bytes, + .gpl_only = false, + .ret_type = RET_INTEGER, + .arg1_type = ARG_PTR_TO_CTX, + .arg2_type = ARG_ANYTHING, + .arg3_type = ARG_PTR_TO_UNINIT_MEM, + .arg4_type = ARG_CONST_SIZE, +}; + +BPF_CALL_5(sk_reuseport_load_bytes_relative, + const struct sk_reuseport_kern *, reuse_kern, u32, offset, + void *, to, u32, len, u32, start_header) +{ + return ____bpf_skb_load_bytes_relative(reuse_kern->skb, offset, to, + len, start_header); +} + +static const struct bpf_func_proto sk_reuseport_load_bytes_relative_proto = { + .func = sk_reuseport_load_bytes_relative, + .gpl_only = false, + .ret_type = RET_INTEGER, + .arg1_type = ARG_PTR_TO_CTX, + .arg2_type = ARG_ANYTHING, + .arg3_type = ARG_PTR_TO_UNINIT_MEM, + .arg4_type = ARG_CONST_SIZE, + .arg5_type = ARG_ANYTHING, +}; + +static const struct bpf_func_proto * +sk_reuseport_func_proto(enum bpf_func_id func_id, + const struct bpf_prog *prog) +{ + switch (func_id) { + case BPF_FUNC_sk_select_reuseport: + return &sk_select_reuseport_proto; + case BPF_FUNC_skb_load_bytes: + return &sk_reuseport_load_bytes_proto; + case BPF_FUNC_skb_load_bytes_relative: + return &sk_reuseport_load_bytes_relative_proto; + default: + return bpf_base_func_proto(func_id); + } +} + +static bool +sk_reuseport_is_valid_access(int off, int size, + enum bpf_access_type type, + const struct bpf_prog *prog, + struct bpf_insn_access_aux *info) +{ + const u32 size_default = sizeof(__u32); + + if (off < 0 || off >= sizeof(struct sk_reuseport_md) || + off % size || type != BPF_READ) + return false; + + switch (off) { + case offsetof(struct sk_reuseport_md, data): + info->reg_type = PTR_TO_PACKET; + return size == sizeof(__u64); + + case offsetof(struct sk_reuseport_md, data_end): + info->reg_type = PTR_TO_PACKET_END; + return size == sizeof(__u64); + + case offsetof(struct sk_reuseport_md, hash): + return size == size_default; + + /* Fields that allow narrowing */ + case offsetof(struct sk_reuseport_md, eth_protocol): + if (size < FIELD_SIZEOF(struct sk_buff, protocol)) + return false; + case offsetof(struct sk_reuseport_md, ip_protocol): + case offsetof(struct sk_reuseport_md, bind_inany): + case offsetof(struct sk_reuseport_md, len): + bpf_ctx_record_field_size(info, size_default); + return bpf_ctx_narrow_access_ok(off, size, size_default); + + default: + return false; + } +} + +#define SK_REUSEPORT_LOAD_FIELD(F) ({ \ + *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct sk_reuseport_kern, F), \ + si->dst_reg, si->src_reg, \ + bpf_target_off(struct sk_reuseport_kern, F, \ + FIELD_SIZEOF(struct sk_reuseport_kern, F), \ + target_size)); \ + }) + +#define SK_REUSEPORT_LOAD_SKB_FIELD(SKB_FIELD) \ + SOCK_ADDR_LOAD_NESTED_FIELD(struct sk_reuseport_kern, \ + struct sk_buff, \ + skb, \ + SKB_FIELD) + +#define SK_REUSEPORT_LOAD_SK_FIELD_SIZE_OFF(SK_FIELD, BPF_SIZE, EXTRA_OFF) \ + SOCK_ADDR_LOAD_NESTED_FIELD_SIZE_OFF(struct sk_reuseport_kern, \ + struct sock, \ + sk, \ + SK_FIELD, BPF_SIZE, EXTRA_OFF) + +static u32 sk_reuseport_convert_ctx_access(enum bpf_access_type type, + const struct bpf_insn *si, + struct bpf_insn *insn_buf, + struct bpf_prog *prog, + u32 *target_size) +{ + struct bpf_insn *insn = insn_buf; + + switch (si->off) { + case offsetof(struct sk_reuseport_md, data): + SK_REUSEPORT_LOAD_SKB_FIELD(data); + break; + + case offsetof(struct sk_reuseport_md, len): + SK_REUSEPORT_LOAD_SKB_FIELD(len); + break; + + case offsetof(struct sk_reuseport_md, eth_protocol): + SK_REUSEPORT_LOAD_SKB_FIELD(protocol); + break; + + case offsetof(struct sk_reuseport_md, ip_protocol): + BUILD_BUG_ON(hweight_long(SK_FL_PROTO_MASK) != BITS_PER_BYTE); + SK_REUSEPORT_LOAD_SK_FIELD_SIZE_OFF(__sk_flags_offset, + BPF_W, 0); + *insn++ = BPF_ALU32_IMM(BPF_AND, si->dst_reg, SK_FL_PROTO_MASK); + *insn++ = BPF_ALU32_IMM(BPF_RSH, si->dst_reg, + SK_FL_PROTO_SHIFT); + /* SK_FL_PROTO_MASK and SK_FL_PROTO_SHIFT are endian + * aware. No further narrowing or masking is needed. + */ + *target_size = 1; + break; + + case offsetof(struct sk_reuseport_md, data_end): + SK_REUSEPORT_LOAD_FIELD(data_end); + break; + + case offsetof(struct sk_reuseport_md, hash): + SK_REUSEPORT_LOAD_FIELD(hash); + break; + + case offsetof(struct sk_reuseport_md, bind_inany): + SK_REUSEPORT_LOAD_FIELD(bind_inany); + break; + } + + return insn - insn_buf; +} + +const struct bpf_verifier_ops sk_reuseport_verifier_ops = { + .get_func_proto = sk_reuseport_func_proto, + .is_valid_access = sk_reuseport_is_valid_access, + .convert_ctx_access = sk_reuseport_convert_ctx_access, +}; + +const struct bpf_prog_ops sk_reuseport_prog_ops = { +}; +#endif /* CONFIG_INET */ diff --git a/net/core/sock_reuseport.c b/net/core/sock_reuseport.c index 8235f2439816..d260167f5f77 100644 --- a/net/core/sock_reuseport.c +++ b/net/core/sock_reuseport.c @@ -51,7 +51,7 @@ static struct sock_reuseport *__reuseport_alloc(unsigned int max_socks) return reuse; } -int reuseport_alloc(struct sock *sk) +int reuseport_alloc(struct sock *sk, bool bind_inany) { struct sock_reuseport *reuse; @@ -63,9 +63,17 @@ int reuseport_alloc(struct sock *sk) /* Allocation attempts can occur concurrently via the setsockopt path * and the bind/hash path. Nothing to do when we lose the race. */ - if (rcu_dereference_protected(sk->sk_reuseport_cb, - lockdep_is_held(&reuseport_lock))) + reuse = rcu_dereference_protected(sk->sk_reuseport_cb, + lockdep_is_held(&reuseport_lock)); + if (reuse) { + /* Only set reuse->bind_inany if the bind_inany is true. + * Otherwise, it will overwrite the reuse->bind_inany + * which was set by the bind/hash path. + */ + if (bind_inany) + reuse->bind_inany = bind_inany; goto out; + } reuse = __reuseport_alloc(INIT_SOCKS); if (!reuse) { @@ -75,6 +83,7 @@ int reuseport_alloc(struct sock *sk) reuse->socks[0] = sk; reuse->num_socks = 1; + reuse->bind_inany = bind_inany; rcu_assign_pointer(sk->sk_reuseport_cb, reuse); out: @@ -101,6 +110,7 @@ static struct sock_reuseport *reuseport_grow(struct sock_reuseport *reuse) more_reuse->num_socks = reuse->num_socks; more_reuse->prog = reuse->prog; more_reuse->reuseport_id = reuse->reuseport_id; + more_reuse->bind_inany = reuse->bind_inany; memcpy(more_reuse->socks, reuse->socks, reuse->num_socks * sizeof(struct sock *)); @@ -136,12 +146,12 @@ static void reuseport_free_rcu(struct rcu_head *head) * @sk2: Socket belonging to the existing reuseport group. * May return ENOMEM and not add socket to group under memory pressure. */ -int reuseport_add_sock(struct sock *sk, struct sock *sk2) +int reuseport_add_sock(struct sock *sk, struct sock *sk2, bool bind_inany) { struct sock_reuseport *old_reuse, *reuse; if (!rcu_access_pointer(sk2->sk_reuseport_cb)) { - int err = reuseport_alloc(sk2); + int err = reuseport_alloc(sk2, bind_inany); if (err) return err; diff --git a/net/ipv4/inet_connection_sock.c b/net/ipv4/inet_connection_sock.c index 33a88e045efd..dfd5009f96ef 100644 --- a/net/ipv4/inet_connection_sock.c +++ b/net/ipv4/inet_connection_sock.c @@ -107,6 +107,15 @@ bool inet_rcv_saddr_equal(const struct sock *sk, const struct sock *sk2, } EXPORT_SYMBOL(inet_rcv_saddr_equal); +bool inet_rcv_saddr_any(const struct sock *sk) +{ +#if IS_ENABLED(CONFIG_IPV6) + if (sk->sk_family == AF_INET6) + return ipv6_addr_any(&sk->sk_v6_rcv_saddr); +#endif + return !sk->sk_rcv_saddr; +} + void inet_get_local_port_range(struct net *net, int *low, int *high) { unsigned int seq; diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c index 3647167c8fa3..370e24463fb7 100644 --- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -567,10 +567,11 @@ static int inet_reuseport_add_sock(struct sock *sk, inet_csk(sk2)->icsk_bind_hash == tb && sk2->sk_reuseport && uid_eq(uid, sock_i_uid(sk2)) && inet_rcv_saddr_equal(sk, sk2, false)) - return reuseport_add_sock(sk, sk2); + return reuseport_add_sock(sk, sk2, + inet_rcv_saddr_any(sk)); } - return reuseport_alloc(sk); + return reuseport_alloc(sk, inet_rcv_saddr_any(sk)); } int __inet_hash(struct sock *sk, struct sock *osk) diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index 060e841dde40..038dd7909051 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -221,11 +221,12 @@ static int udp_reuseport_add_sock(struct sock *sk, struct udp_hslot *hslot) (sk2->sk_bound_dev_if == sk->sk_bound_dev_if) && sk2->sk_reuseport && uid_eq(uid, sock_i_uid(sk2)) && inet_rcv_saddr_equal(sk, sk2, false)) { - return reuseport_add_sock(sk, sk2); + return reuseport_add_sock(sk, sk2, + inet_rcv_saddr_any(sk)); } } - return reuseport_alloc(sk); + return reuseport_alloc(sk, inet_rcv_saddr_any(sk)); } /** From 8217ca653ec601246832d562207bc24bdf652d2f Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Wed, 8 Aug 2018 01:01:26 -0700 Subject: [PATCH 5/9] bpf: Enable BPF_PROG_TYPE_SK_REUSEPORT bpf prog in reuseport selection This patch allows a BPF_PROG_TYPE_SK_REUSEPORT bpf prog to select a SO_REUSEPORT sk from a BPF_MAP_TYPE_REUSEPORT_ARRAY introduced in the earlier patch. "bpf_run_sk_reuseport()" will return -ECONNREFUSED when the BPF_PROG_TYPE_SK_REUSEPORT prog returns SK_DROP. The callers, in inet[6]_hashtable.c and ipv[46]/udp.c, are modified to handle this case and return NULL immediately instead of continuing the sk search from its hashtable. It re-uses the existing SO_ATTACH_REUSEPORT_EBPF setsockopt to attach BPF_PROG_TYPE_SK_REUSEPORT. The "sk_reuseport_attach_bpf()" will check if the attaching bpf prog is in the new SK_REUSEPORT or the existing SOCKET_FILTER type and then check different things accordingly. One level of "__reuseport_attach_prog()" call is removed. The "sk_unhashed() && ..." and "sk->sk_reuseport_cb" tests are pushed back to "reuseport_attach_prog()" in sock_reuseport.c. sock_reuseport.c seems to have more knowledge on those test requirements than filter.c. In "reuseport_attach_prog()", after new_prog is attached to reuse->prog, the old_prog (if any) is also directly freed instead of returning the old_prog to the caller and asking the caller to free. The sysctl_optmem_max check is moved back to the "sk_reuseport_attach_filter()" and "sk_reuseport_attach_bpf()". As of other bpf prog types, the new BPF_PROG_TYPE_SK_REUSEPORT is only bounded by the usual "bpf_prog_charge_memlock()" during load time instead of bounded by both bpf_prog_charge_memlock and sysctl_optmem_max. Signed-off-by: Martin KaFai Lau Acked-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann --- include/linux/filter.h | 1 + include/net/sock_reuseport.h | 3 +- net/core/filter.c | 89 +++++++++++++++++++++--------------- net/core/sock_reuseport.c | 36 +++++++++++---- net/ipv4/inet_hashtables.c | 14 ++++-- net/ipv4/udp.c | 4 ++ net/ipv6/inet6_hashtables.c | 14 ++++-- net/ipv6/udp.c | 4 ++ 8 files changed, 107 insertions(+), 58 deletions(-) diff --git a/include/linux/filter.h b/include/linux/filter.h index 70e9d57677fe..5d565c50bcb2 100644 --- a/include/linux/filter.h +++ b/include/linux/filter.h @@ -753,6 +753,7 @@ int sk_attach_filter(struct sock_fprog *fprog, struct sock *sk); int sk_attach_bpf(u32 ufd, struct sock *sk); int sk_reuseport_attach_filter(struct sock_fprog *fprog, struct sock *sk); int sk_reuseport_attach_bpf(u32 ufd, struct sock *sk); +void sk_reuseport_prog_free(struct bpf_prog *prog); int sk_detach_filter(struct sock *sk); int sk_get_filter(struct sock *sk, struct sock_filter __user *filter, unsigned int len); diff --git a/include/net/sock_reuseport.h b/include/net/sock_reuseport.h index 73b569556be6..8a5f70c7cdf2 100644 --- a/include/net/sock_reuseport.h +++ b/include/net/sock_reuseport.h @@ -34,8 +34,7 @@ extern struct sock *reuseport_select_sock(struct sock *sk, u32 hash, struct sk_buff *skb, int hdr_len); -extern struct bpf_prog *reuseport_attach_prog(struct sock *sk, - struct bpf_prog *prog); +extern int reuseport_attach_prog(struct sock *sk, struct bpf_prog *prog); int reuseport_get_id(struct sock_reuseport *reuse); #endif /* _SOCK_REUSEPORT_H */ diff --git a/net/core/filter.c b/net/core/filter.c index 142595b4e0d1..22906b31d43f 100644 --- a/net/core/filter.c +++ b/net/core/filter.c @@ -1453,30 +1453,6 @@ static int __sk_attach_prog(struct bpf_prog *prog, struct sock *sk) return 0; } -static int __reuseport_attach_prog(struct bpf_prog *prog, struct sock *sk) -{ - struct bpf_prog *old_prog; - int err; - - if (bpf_prog_size(prog->len) > sysctl_optmem_max) - return -ENOMEM; - - if (sk_unhashed(sk) && sk->sk_reuseport) { - err = reuseport_alloc(sk, false); - if (err) - return err; - } else if (!rcu_access_pointer(sk->sk_reuseport_cb)) { - /* The socket wasn't bound with SO_REUSEPORT */ - return -EINVAL; - } - - old_prog = reuseport_attach_prog(sk, prog); - if (old_prog) - bpf_prog_destroy(old_prog); - - return 0; -} - static struct bpf_prog *__get_filter(struct sock_fprog *fprog, struct sock *sk) { @@ -1550,13 +1526,15 @@ int sk_reuseport_attach_filter(struct sock_fprog *fprog, struct sock *sk) if (IS_ERR(prog)) return PTR_ERR(prog); - err = __reuseport_attach_prog(prog, sk); - if (err < 0) { - __bpf_prog_release(prog); - return err; - } + if (bpf_prog_size(prog->len) > sysctl_optmem_max) + err = -ENOMEM; + else + err = reuseport_attach_prog(sk, prog); - return 0; + if (err) + __bpf_prog_release(prog); + + return err; } static struct bpf_prog *__get_bpf(u32 ufd, struct sock *sk) @@ -1586,19 +1564,58 @@ int sk_attach_bpf(u32 ufd, struct sock *sk) int sk_reuseport_attach_bpf(u32 ufd, struct sock *sk) { - struct bpf_prog *prog = __get_bpf(ufd, sk); + struct bpf_prog *prog; int err; + if (sock_flag(sk, SOCK_FILTER_LOCKED)) + return -EPERM; + + prog = bpf_prog_get_type(ufd, BPF_PROG_TYPE_SOCKET_FILTER); + if (IS_ERR(prog) && PTR_ERR(prog) == -EINVAL) + prog = bpf_prog_get_type(ufd, BPF_PROG_TYPE_SK_REUSEPORT); if (IS_ERR(prog)) return PTR_ERR(prog); - err = __reuseport_attach_prog(prog, sk); - if (err < 0) { - bpf_prog_put(prog); - return err; + if (prog->type == BPF_PROG_TYPE_SK_REUSEPORT) { + /* Like other non BPF_PROG_TYPE_SOCKET_FILTER + * bpf prog (e.g. sockmap). It depends on the + * limitation imposed by bpf_prog_load(). + * Hence, sysctl_optmem_max is not checked. + */ + if ((sk->sk_type != SOCK_STREAM && + sk->sk_type != SOCK_DGRAM) || + (sk->sk_protocol != IPPROTO_UDP && + sk->sk_protocol != IPPROTO_TCP) || + (sk->sk_family != AF_INET && + sk->sk_family != AF_INET6)) { + err = -ENOTSUPP; + goto err_prog_put; + } + } else { + /* BPF_PROG_TYPE_SOCKET_FILTER */ + if (bpf_prog_size(prog->len) > sysctl_optmem_max) { + err = -ENOMEM; + goto err_prog_put; + } } - return 0; + err = reuseport_attach_prog(sk, prog); +err_prog_put: + if (err) + bpf_prog_put(prog); + + return err; +} + +void sk_reuseport_prog_free(struct bpf_prog *prog) +{ + if (!prog) + return; + + if (prog->type == BPF_PROG_TYPE_SK_REUSEPORT) + bpf_prog_put(prog); + else + bpf_prog_destroy(prog); } struct bpf_scratchpad { diff --git a/net/core/sock_reuseport.c b/net/core/sock_reuseport.c index d260167f5f77..ba5cba56f574 100644 --- a/net/core/sock_reuseport.c +++ b/net/core/sock_reuseport.c @@ -9,6 +9,7 @@ #include #include #include +#include #include #define INIT_SOCKS 128 @@ -133,8 +134,7 @@ static void reuseport_free_rcu(struct rcu_head *head) struct sock_reuseport *reuse; reuse = container_of(head, struct sock_reuseport, rcu); - if (reuse->prog) - bpf_prog_destroy(reuse->prog); + sk_reuseport_prog_free(rcu_dereference_protected(reuse->prog, 1)); if (reuse->reuseport_id) ida_simple_remove(&reuseport_ida, reuse->reuseport_id); kfree(reuse); @@ -219,9 +219,9 @@ void reuseport_detach_sock(struct sock *sk) } EXPORT_SYMBOL(reuseport_detach_sock); -static struct sock *run_bpf(struct sock_reuseport *reuse, u16 socks, - struct bpf_prog *prog, struct sk_buff *skb, - int hdr_len) +static struct sock *run_bpf_filter(struct sock_reuseport *reuse, u16 socks, + struct bpf_prog *prog, struct sk_buff *skb, + int hdr_len) { struct sk_buff *nskb = NULL; u32 index; @@ -282,9 +282,15 @@ struct sock *reuseport_select_sock(struct sock *sk, /* paired with smp_wmb() in reuseport_add_sock() */ smp_rmb(); - if (prog && skb) - sk2 = run_bpf(reuse, socks, prog, skb, hdr_len); + if (!prog || !skb) + goto select_by_hash; + if (prog->type == BPF_PROG_TYPE_SK_REUSEPORT) + sk2 = bpf_run_sk_reuseport(reuse, sk, prog, skb, hash); + else + sk2 = run_bpf_filter(reuse, socks, prog, skb, hdr_len); + +select_by_hash: /* no bpf or invalid bpf result: fall back to hash usage */ if (!sk2) sk2 = reuse->socks[reciprocal_scale(hash, socks)]; @@ -296,12 +302,21 @@ out: } EXPORT_SYMBOL(reuseport_select_sock); -struct bpf_prog * -reuseport_attach_prog(struct sock *sk, struct bpf_prog *prog) +int reuseport_attach_prog(struct sock *sk, struct bpf_prog *prog) { struct sock_reuseport *reuse; struct bpf_prog *old_prog; + if (sk_unhashed(sk) && sk->sk_reuseport) { + int err = reuseport_alloc(sk, false); + + if (err) + return err; + } else if (!rcu_access_pointer(sk->sk_reuseport_cb)) { + /* The socket wasn't bound with SO_REUSEPORT */ + return -EINVAL; + } + spin_lock_bh(&reuseport_lock); reuse = rcu_dereference_protected(sk->sk_reuseport_cb, lockdep_is_held(&reuseport_lock)); @@ -310,6 +325,7 @@ reuseport_attach_prog(struct sock *sk, struct bpf_prog *prog) rcu_assign_pointer(reuse->prog, prog); spin_unlock_bh(&reuseport_lock); - return old_prog; + sk_reuseport_prog_free(old_prog); + return 0; } EXPORT_SYMBOL(reuseport_attach_prog); diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c index 370e24463fb7..f5c9ef2586de 100644 --- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -328,7 +328,7 @@ struct sock *__inet_lookup_listener(struct net *net, saddr, sport, daddr, hnum, dif, sdif); if (result) - return result; + goto done; /* Lookup lhash2 with INADDR_ANY */ @@ -337,9 +337,10 @@ struct sock *__inet_lookup_listener(struct net *net, if (ilb2->count > ilb->count) goto port_lookup; - return inet_lhash2_lookup(net, ilb2, skb, doff, - saddr, sport, daddr, hnum, - dif, sdif); + result = inet_lhash2_lookup(net, ilb2, skb, doff, + saddr, sport, daddr, hnum, + dif, sdif); + goto done; port_lookup: sk_for_each_rcu(sk, &ilb->head) { @@ -352,12 +353,15 @@ port_lookup: result = reuseport_select_sock(sk, phash, skb, doff); if (result) - return result; + goto done; } result = sk; hiscore = score; } } +done: + if (unlikely(IS_ERR(result))) + return NULL; return result; } EXPORT_SYMBOL_GPL(__inet_lookup_listener); diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index 038dd7909051..f4e35b2ff8b8 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -499,6 +499,8 @@ struct sock *__udp4_lib_lookup(struct net *net, __be32 saddr, daddr, hnum, dif, sdif, exact_dif, hslot2, skb); } + if (unlikely(IS_ERR(result))) + return NULL; return result; } begin: @@ -513,6 +515,8 @@ begin: saddr, sport); result = reuseport_select_sock(sk, hash, skb, sizeof(struct udphdr)); + if (unlikely(IS_ERR(result))) + return NULL; if (result) return result; } diff --git a/net/ipv6/inet6_hashtables.c b/net/ipv6/inet6_hashtables.c index 595ad408dba0..3d7c7460a0c5 100644 --- a/net/ipv6/inet6_hashtables.c +++ b/net/ipv6/inet6_hashtables.c @@ -191,7 +191,7 @@ struct sock *inet6_lookup_listener(struct net *net, saddr, sport, daddr, hnum, dif, sdif); if (result) - return result; + goto done; /* Lookup lhash2 with in6addr_any */ @@ -200,9 +200,10 @@ struct sock *inet6_lookup_listener(struct net *net, if (ilb2->count > ilb->count) goto port_lookup; - return inet6_lhash2_lookup(net, ilb2, skb, doff, - saddr, sport, daddr, hnum, - dif, sdif); + result = inet6_lhash2_lookup(net, ilb2, skb, doff, + saddr, sport, daddr, hnum, + dif, sdif); + goto done; port_lookup: sk_for_each(sk, &ilb->head) { @@ -214,12 +215,15 @@ port_lookup: result = reuseport_select_sock(sk, phash, skb, doff); if (result) - return result; + goto done; } result = sk; hiscore = score; } } +done: + if (unlikely(IS_ERR(result))) + return NULL; return result; } EXPORT_SYMBOL_GPL(inet6_lookup_listener); diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c index f6b96956a8ed..83f4c77c79d8 100644 --- a/net/ipv6/udp.c +++ b/net/ipv6/udp.c @@ -235,6 +235,8 @@ struct sock *__udp6_lib_lookup(struct net *net, exact_dif, hslot2, skb); } + if (unlikely(IS_ERR(result))) + return NULL; return result; } begin: @@ -249,6 +251,8 @@ begin: saddr, sport); result = reuseport_select_sock(sk, hash, skb, sizeof(struct udphdr)); + if (unlikely(IS_ERR(result))) + return NULL; if (result) return result; } From aa5f0c96cc7b96a678779055b2ff4cd9dabd8ba7 Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Wed, 8 Aug 2018 01:01:27 -0700 Subject: [PATCH 6/9] bpf: Refactor ARRAY_SIZE macro to bpf_util.h This patch refactors the ARRAY_SIZE macro to bpf_util.h. Signed-off-by: Martin KaFai Lau Acked-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann --- tools/testing/selftests/bpf/bpf_util.h | 4 ++++ tools/testing/selftests/bpf/test_align.c | 5 +---- tools/testing/selftests/bpf/test_btf.c | 5 +---- tools/testing/selftests/bpf/test_sock.c | 5 +---- tools/testing/selftests/bpf/test_sock_addr.c | 5 +---- tools/testing/selftests/bpf/test_verifier.c | 5 +---- 6 files changed, 9 insertions(+), 20 deletions(-) diff --git a/tools/testing/selftests/bpf/bpf_util.h b/tools/testing/selftests/bpf/bpf_util.h index d0811b3d6a6f..315a44fa32af 100644 --- a/tools/testing/selftests/bpf/bpf_util.h +++ b/tools/testing/selftests/bpf/bpf_util.h @@ -44,4 +44,8 @@ static inline unsigned int bpf_num_possible_cpus(void) name[bpf_num_possible_cpus()] #define bpf_percpu(name, cpu) name[(cpu)].v +#ifndef ARRAY_SIZE +# define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0])) +#endif + #endif /* __BPF_UTIL__ */ diff --git a/tools/testing/selftests/bpf/test_align.c b/tools/testing/selftests/bpf/test_align.c index 6b1b302310fe..5f377ec53f2f 100644 --- a/tools/testing/selftests/bpf/test_align.c +++ b/tools/testing/selftests/bpf/test_align.c @@ -18,10 +18,7 @@ #include "../../../include/linux/filter.h" #include "bpf_rlimit.h" - -#ifndef ARRAY_SIZE -# define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0])) -#endif +#include "bpf_util.h" #define MAX_INSNS 512 #define MAX_MATCHES 16 diff --git a/tools/testing/selftests/bpf/test_btf.c b/tools/testing/selftests/bpf/test_btf.c index 7fa8c800c540..6b5cfeb7a9cc 100644 --- a/tools/testing/selftests/bpf/test_btf.c +++ b/tools/testing/selftests/bpf/test_btf.c @@ -19,6 +19,7 @@ #include #include "bpf_rlimit.h" +#include "bpf_util.h" static uint32_t pass_cnt; static uint32_t error_cnt; @@ -93,10 +94,6 @@ static int __base_pr(const char *format, ...) #define MAX_NR_RAW_TYPES 1024 #define BTF_LOG_BUF_SIZE 65535 -#ifndef ARRAY_SIZE -# define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0])) -#endif - static struct args { unsigned int raw_test_num; unsigned int file_test_num; diff --git a/tools/testing/selftests/bpf/test_sock.c b/tools/testing/selftests/bpf/test_sock.c index f4d99fabc56d..b8ebe2f58074 100644 --- a/tools/testing/selftests/bpf/test_sock.c +++ b/tools/testing/selftests/bpf/test_sock.c @@ -14,10 +14,7 @@ #include "cgroup_helpers.h" #include "bpf_rlimit.h" - -#ifndef ARRAY_SIZE -# define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0])) -#endif +#include "bpf_util.h" #define CG_PATH "/foo" #define MAX_INSNS 512 diff --git a/tools/testing/selftests/bpf/test_sock_addr.c b/tools/testing/selftests/bpf/test_sock_addr.c index 2e45c92d1111..aeeb76a54d63 100644 --- a/tools/testing/selftests/bpf/test_sock_addr.c +++ b/tools/testing/selftests/bpf/test_sock_addr.c @@ -20,15 +20,12 @@ #include "cgroup_helpers.h" #include "bpf_rlimit.h" +#include "bpf_util.h" #ifndef ENOTSUPP # define ENOTSUPP 524 #endif -#ifndef ARRAY_SIZE -# define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0])) -#endif - #define CG_PATH "/foo" #define CONNECT4_PROG_PATH "./connect4_prog.o" #define CONNECT6_PROG_PATH "./connect6_prog.o" diff --git a/tools/testing/selftests/bpf/test_verifier.c b/tools/testing/selftests/bpf/test_verifier.c index 452cf5c6c784..67c412d19c09 100644 --- a/tools/testing/selftests/bpf/test_verifier.c +++ b/tools/testing/selftests/bpf/test_verifier.c @@ -42,12 +42,9 @@ #endif #include "bpf_rlimit.h" #include "bpf_rand.h" +#include "bpf_util.h" #include "../../../include/linux/filter.h" -#ifndef ARRAY_SIZE -# define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0])) -#endif - #define MAX_INSNS BPF_MAXINSNS #define MAX_FIXUPS 8 #define MAX_NR_MAPS 8 From 3bd43a8c91cba0493e507ed7baefa9b5613c28a9 Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Wed, 8 Aug 2018 01:01:29 -0700 Subject: [PATCH 7/9] bpf: Sync bpf.h uapi to tools/ This patch sync include/uapi/linux/bpf.h to tools/include/uapi/linux/ Signed-off-by: Martin KaFai Lau Acked-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann --- tools/include/uapi/linux/bpf.h | 37 +++++++++++++++++++++++++++++++++- 1 file changed, 36 insertions(+), 1 deletion(-) diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h index dd5758dc35d3..3102a2a23c31 100644 --- a/tools/include/uapi/linux/bpf.h +++ b/tools/include/uapi/linux/bpf.h @@ -126,6 +126,7 @@ enum bpf_map_type { BPF_MAP_TYPE_XSKMAP, BPF_MAP_TYPE_SOCKHASH, BPF_MAP_TYPE_CGROUP_STORAGE, + BPF_MAP_TYPE_REUSEPORT_SOCKARRAY, }; enum bpf_prog_type { @@ -150,6 +151,7 @@ enum bpf_prog_type { BPF_PROG_TYPE_CGROUP_SOCK_ADDR, BPF_PROG_TYPE_LWT_SEG6LOCAL, BPF_PROG_TYPE_LIRC_MODE2, + BPF_PROG_TYPE_SK_REUSEPORT, }; enum bpf_attach_type { @@ -2113,6 +2115,14 @@ union bpf_attr { * the shared data. * Return * Pointer to the local storage area. + * + * int bpf_sk_select_reuseport(struct sk_reuseport_md *reuse, struct bpf_map *map, void *key, u64 flags) + * Description + * Select a SO_REUSEPORT sk from a BPF_MAP_TYPE_REUSEPORT_ARRAY map + * It checks the selected sk is matching the incoming + * request in the skb. + * Return + * 0 on success, or a negative error in case of failure. */ #define __BPF_FUNC_MAPPER(FN) \ FN(unspec), \ @@ -2196,7 +2206,8 @@ union bpf_attr { FN(rc_keydown), \ FN(skb_cgroup_id), \ FN(get_current_cgroup_id), \ - FN(get_local_storage), + FN(get_local_storage), \ + FN(sk_select_reuseport), /* integer value in 'imm' field of BPF_CALL instruction selects which helper * function eBPF program intends to call @@ -2413,6 +2424,30 @@ struct sk_msg_md { __u32 local_port; /* stored in host byte order */ }; +struct sk_reuseport_md { + /* + * Start of directly accessible data. It begins from + * the tcp/udp header. + */ + void *data; + void *data_end; /* End of directly accessible data */ + /* + * Total length of packet (starting from the tcp/udp header). + * Note that the directly accessible bytes (data_end - data) + * could be less than this "len". Those bytes could be + * indirectly read by a helper "bpf_skb_load_bytes()". + */ + __u32 len; + /* + * Eth protocol in the mac header (network byte order). e.g. + * ETH_P_IP(0x0800) and ETH_P_IPV6(0x86DD) + */ + __u32 eth_protocol; + __u32 ip_protocol; /* IP protocol. e.g. IPPROTO_TCP, IPPROTO_UDP */ + __u32 bind_inany; /* Is sock bound to an INANY address? */ + __u32 hash; /* A hash of the packet 4 tuples */ +}; + #define BPF_TAG_SIZE 8 struct bpf_prog_info { From 6bc8529c414f931ce0acef3099b015cf2f5c4291 Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Wed, 8 Aug 2018 01:01:30 -0700 Subject: [PATCH 8/9] bpf: test BPF_MAP_TYPE_REUSEPORT_SOCKARRAY This patch adds tests for the new BPF_MAP_TYPE_REUSEPORT_SOCKARRAY. Signed-off-by: Martin KaFai Lau Acked-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann --- tools/lib/bpf/libbpf.c | 1 + tools/testing/selftests/bpf/test_maps.c | 262 +++++++++++++++++++++++- 2 files changed, 262 insertions(+), 1 deletion(-) diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c index 40211b51427a..2abd0f112627 100644 --- a/tools/lib/bpf/libbpf.c +++ b/tools/lib/bpf/libbpf.c @@ -1501,6 +1501,7 @@ static bool bpf_prog_type__needs_kver(enum bpf_prog_type type) case BPF_PROG_TYPE_SK_MSG: case BPF_PROG_TYPE_CGROUP_SOCK_ADDR: case BPF_PROG_TYPE_LIRC_MODE2: + case BPF_PROG_TYPE_SK_REUSEPORT: return false; case BPF_PROG_TYPE_UNSPEC: case BPF_PROG_TYPE_KPROBE: diff --git a/tools/testing/selftests/bpf/test_maps.c b/tools/testing/selftests/bpf/test_maps.c index 6c253343a6f9..4b7c74f5faa7 100644 --- a/tools/testing/selftests/bpf/test_maps.c +++ b/tools/testing/selftests/bpf/test_maps.c @@ -17,7 +17,8 @@ #include #include - +#include +#include #include #include @@ -26,8 +27,21 @@ #include "bpf_util.h" #include "bpf_rlimit.h" +#ifndef ENOTSUPP +#define ENOTSUPP 524 +#endif + static int map_flags; +#define CHECK(condition, tag, format...) ({ \ + int __ret = !!(condition); \ + if (__ret) { \ + printf("%s(%d):FAIL:%s ", __func__, __LINE__, tag); \ + printf(format); \ + exit(-1); \ + } \ +}) + static void test_hashmap(int task, void *data) { long long key, next_key, first_key, value; @@ -1150,6 +1164,250 @@ static void test_map_wronly(void) assert(bpf_map_get_next_key(fd, &key, &value) == -1 && errno == EPERM); } +static void prepare_reuseport_grp(int type, int map_fd, + __s64 *fds64, __u64 *sk_cookies, + unsigned int n) +{ + socklen_t optlen, addrlen; + struct sockaddr_in6 s6; + const __u32 index0 = 0; + const int optval = 1; + unsigned int i; + u64 sk_cookie; + __s64 fd64; + int err; + + s6.sin6_family = AF_INET6; + s6.sin6_addr = in6addr_any; + s6.sin6_port = 0; + addrlen = sizeof(s6); + optlen = sizeof(sk_cookie); + + for (i = 0; i < n; i++) { + fd64 = socket(AF_INET6, type, 0); + CHECK(fd64 == -1, "socket()", + "sock_type:%d fd64:%lld errno:%d\n", + type, fd64, errno); + + err = setsockopt(fd64, SOL_SOCKET, SO_REUSEPORT, + &optval, sizeof(optval)); + CHECK(err == -1, "setsockopt(SO_REUSEEPORT)", + "err:%d errno:%d\n", err, errno); + + /* reuseport_array does not allow unbound sk */ + err = bpf_map_update_elem(map_fd, &index0, &fd64, + BPF_ANY); + CHECK(err != -1 || errno != EINVAL, + "reuseport array update unbound sk", + "sock_type:%d err:%d errno:%d\n", + type, err, errno); + + err = bind(fd64, (struct sockaddr *)&s6, sizeof(s6)); + CHECK(err == -1, "bind()", + "sock_type:%d err:%d errno:%d\n", type, err, errno); + + if (i == 0) { + err = getsockname(fd64, (struct sockaddr *)&s6, + &addrlen); + CHECK(err == -1, "getsockname()", + "sock_type:%d err:%d errno:%d\n", + type, err, errno); + } + + err = getsockopt(fd64, SOL_SOCKET, SO_COOKIE, &sk_cookie, + &optlen); + CHECK(err == -1, "getsockopt(SO_COOKIE)", + "sock_type:%d err:%d errno:%d\n", type, err, errno); + + if (type == SOCK_STREAM) { + /* + * reuseport_array does not allow + * non-listening tcp sk. + */ + err = bpf_map_update_elem(map_fd, &index0, &fd64, + BPF_ANY); + CHECK(err != -1 || errno != EINVAL, + "reuseport array update non-listening sk", + "sock_type:%d err:%d errno:%d\n", + type, err, errno); + err = listen(fd64, 0); + CHECK(err == -1, "listen()", + "sock_type:%d, err:%d errno:%d\n", + type, err, errno); + } + + fds64[i] = fd64; + sk_cookies[i] = sk_cookie; + } +} + +static void test_reuseport_array(void) +{ +#define REUSEPORT_FD_IDX(err, last) ({ (err) ? last : !last; }) + + const __u32 array_size = 4, index0 = 0, index3 = 3; + int types[2] = { SOCK_STREAM, SOCK_DGRAM }, type; + __u64 grpa_cookies[2], sk_cookie, map_cookie; + __s64 grpa_fds64[2] = { -1, -1 }, fd64 = -1; + const __u32 bad_index = array_size; + int map_fd, err, t, f; + __u32 fds_idx = 0; + int fd; + + map_fd = bpf_create_map(BPF_MAP_TYPE_REUSEPORT_SOCKARRAY, + sizeof(__u32), sizeof(__u64), array_size, 0); + CHECK(map_fd == -1, "reuseport array create", + "map_fd:%d, errno:%d\n", map_fd, errno); + + /* Test lookup/update/delete with invalid index */ + err = bpf_map_delete_elem(map_fd, &bad_index); + CHECK(err != -1 || errno != E2BIG, "reuseport array del >=max_entries", + "err:%d errno:%d\n", err, errno); + + err = bpf_map_update_elem(map_fd, &bad_index, &fd64, BPF_ANY); + CHECK(err != -1 || errno != E2BIG, + "reuseport array update >=max_entries", + "err:%d errno:%d\n", err, errno); + + err = bpf_map_lookup_elem(map_fd, &bad_index, &map_cookie); + CHECK(err != -1 || errno != ENOENT, + "reuseport array update >=max_entries", + "err:%d errno:%d\n", err, errno); + + /* Test lookup/delete non existence elem */ + err = bpf_map_lookup_elem(map_fd, &index3, &map_cookie); + CHECK(err != -1 || errno != ENOENT, + "reuseport array lookup not-exist elem", + "err:%d errno:%d\n", err, errno); + err = bpf_map_delete_elem(map_fd, &index3); + CHECK(err != -1 || errno != ENOENT, + "reuseport array del not-exist elem", + "err:%d errno:%d\n", err, errno); + + for (t = 0; t < ARRAY_SIZE(types); t++) { + type = types[t]; + + prepare_reuseport_grp(type, map_fd, grpa_fds64, + grpa_cookies, ARRAY_SIZE(grpa_fds64)); + + /* Test BPF_* update flags */ + /* BPF_EXIST failure case */ + err = bpf_map_update_elem(map_fd, &index3, &grpa_fds64[fds_idx], + BPF_EXIST); + CHECK(err != -1 || errno != ENOENT, + "reuseport array update empty elem BPF_EXIST", + "sock_type:%d err:%d errno:%d\n", + type, err, errno); + fds_idx = REUSEPORT_FD_IDX(err, fds_idx); + + /* BPF_NOEXIST success case */ + err = bpf_map_update_elem(map_fd, &index3, &grpa_fds64[fds_idx], + BPF_NOEXIST); + CHECK(err == -1, + "reuseport array update empty elem BPF_NOEXIST", + "sock_type:%d err:%d errno:%d\n", + type, err, errno); + fds_idx = REUSEPORT_FD_IDX(err, fds_idx); + + /* BPF_EXIST success case. */ + err = bpf_map_update_elem(map_fd, &index3, &grpa_fds64[fds_idx], + BPF_EXIST); + CHECK(err == -1, + "reuseport array update same elem BPF_EXIST", + "sock_type:%d err:%d errno:%d\n", type, err, errno); + fds_idx = REUSEPORT_FD_IDX(err, fds_idx); + + /* BPF_NOEXIST failure case */ + err = bpf_map_update_elem(map_fd, &index3, &grpa_fds64[fds_idx], + BPF_NOEXIST); + CHECK(err != -1 || errno != EEXIST, + "reuseport array update non-empty elem BPF_NOEXIST", + "sock_type:%d err:%d errno:%d\n", + type, err, errno); + fds_idx = REUSEPORT_FD_IDX(err, fds_idx); + + /* BPF_ANY case (always succeed) */ + err = bpf_map_update_elem(map_fd, &index3, &grpa_fds64[fds_idx], + BPF_ANY); + CHECK(err == -1, + "reuseport array update same sk with BPF_ANY", + "sock_type:%d err:%d errno:%d\n", type, err, errno); + + fd64 = grpa_fds64[fds_idx]; + sk_cookie = grpa_cookies[fds_idx]; + + /* The same sk cannot be added to reuseport_array twice */ + err = bpf_map_update_elem(map_fd, &index3, &fd64, BPF_ANY); + CHECK(err != -1 || errno != EBUSY, + "reuseport array update same sk with same index", + "sock_type:%d err:%d errno:%d\n", + type, err, errno); + + err = bpf_map_update_elem(map_fd, &index0, &fd64, BPF_ANY); + CHECK(err != -1 || errno != EBUSY, + "reuseport array update same sk with different index", + "sock_type:%d err:%d errno:%d\n", + type, err, errno); + + /* Test delete elem */ + err = bpf_map_delete_elem(map_fd, &index3); + CHECK(err == -1, "reuseport array delete sk", + "sock_type:%d err:%d errno:%d\n", + type, err, errno); + + /* Add it back with BPF_NOEXIST */ + err = bpf_map_update_elem(map_fd, &index3, &fd64, BPF_NOEXIST); + CHECK(err == -1, + "reuseport array re-add with BPF_NOEXIST after del", + "sock_type:%d err:%d errno:%d\n", type, err, errno); + + /* Test cookie */ + err = bpf_map_lookup_elem(map_fd, &index3, &map_cookie); + CHECK(err == -1 || sk_cookie != map_cookie, + "reuseport array lookup re-added sk", + "sock_type:%d err:%d errno:%d sk_cookie:0x%llx map_cookie:0x%llxn", + type, err, errno, sk_cookie, map_cookie); + + /* Test elem removed by close() */ + for (f = 0; f < ARRAY_SIZE(grpa_fds64); f++) + close(grpa_fds64[f]); + err = bpf_map_lookup_elem(map_fd, &index3, &map_cookie); + CHECK(err != -1 || errno != ENOENT, + "reuseport array lookup after close()", + "sock_type:%d err:%d errno:%d\n", + type, err, errno); + } + + /* Test SOCK_RAW */ + fd64 = socket(AF_INET6, SOCK_RAW, IPPROTO_UDP); + CHECK(fd64 == -1, "socket(SOCK_RAW)", "err:%d errno:%d\n", + err, errno); + err = bpf_map_update_elem(map_fd, &index3, &fd64, BPF_NOEXIST); + CHECK(err != -1 || errno != ENOTSUPP, "reuseport array update SOCK_RAW", + "err:%d errno:%d\n", err, errno); + close(fd64); + + /* Close the 64 bit value map */ + close(map_fd); + + /* Test 32 bit fd */ + map_fd = bpf_create_map(BPF_MAP_TYPE_REUSEPORT_SOCKARRAY, + sizeof(__u32), sizeof(__u32), array_size, 0); + CHECK(map_fd == -1, "reuseport array create", + "map_fd:%d, errno:%d\n", map_fd, errno); + prepare_reuseport_grp(SOCK_STREAM, map_fd, &fd64, &sk_cookie, 1); + fd = fd64; + err = bpf_map_update_elem(map_fd, &index3, &fd, BPF_NOEXIST); + CHECK(err == -1, "reuseport array update 32 bit fd", + "err:%d errno:%d\n", err, errno); + err = bpf_map_lookup_elem(map_fd, &index3, &map_cookie); + CHECK(err != -1 || errno != ENOSPC, + "reuseport array lookup 32 bit fd", + "err:%d errno:%d\n", err, errno); + close(fd); + close(map_fd); +} + static void run_all_tests(void) { test_hashmap(0, NULL); @@ -1170,6 +1428,8 @@ static void run_all_tests(void) test_map_rdonly(); test_map_wronly(); + + test_reuseport_array(); } int main(void) From 91134d849a0e8fbc70b8607d280e0d325dcaf7bb Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Wed, 8 Aug 2018 01:01:31 -0700 Subject: [PATCH 9/9] bpf: Test BPF_PROG_TYPE_SK_REUSEPORT This patch add tests for the new BPF_PROG_TYPE_SK_REUSEPORT. The tests cover: - IPv4/IPv6 + TCP/UDP - TCP syncookie - TCP fastopen - Cases when the bpf_sk_select_reuseport() returning errors - Cases when the bpf prog returns SK_DROP - Values from sk_reuseport_md - outer_map => reuseport_array The test depends on commit 3eee1f75f2b9 ("bpf: fix bpf_skb_load_bytes_relative pkt length check") Signed-off-by: Martin KaFai Lau Acked-by: Alexei Starovoitov Signed-off-by: Daniel Borkmann --- tools/lib/bpf/bpf.c | 1 + tools/lib/bpf/bpf.h | 1 + tools/testing/selftests/bpf/Makefile | 4 +- tools/testing/selftests/bpf/bpf_helpers.h | 4 + .../selftests/bpf/test_select_reuseport.c | 688 ++++++++++++++++++ .../bpf/test_select_reuseport_common.h | 36 + .../bpf/test_select_reuseport_kern.c | 180 +++++ 7 files changed, 912 insertions(+), 2 deletions(-) create mode 100644 tools/testing/selftests/bpf/test_select_reuseport.c create mode 100644 tools/testing/selftests/bpf/test_select_reuseport_common.h create mode 100644 tools/testing/selftests/bpf/test_select_reuseport_kern.c diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c index 9ddc89dae962..60aa4ca8b2c5 100644 --- a/tools/lib/bpf/bpf.c +++ b/tools/lib/bpf/bpf.c @@ -92,6 +92,7 @@ int bpf_create_map_xattr(const struct bpf_create_map_attr *create_attr) attr.btf_key_type_id = create_attr->btf_key_type_id; attr.btf_value_type_id = create_attr->btf_value_type_id; attr.map_ifindex = create_attr->map_ifindex; + attr.inner_map_fd = create_attr->inner_map_fd; return sys_bpf(BPF_MAP_CREATE, &attr, sizeof(attr)); } diff --git a/tools/lib/bpf/bpf.h b/tools/lib/bpf/bpf.h index 0639a30a457d..6f38164b2618 100644 --- a/tools/lib/bpf/bpf.h +++ b/tools/lib/bpf/bpf.h @@ -39,6 +39,7 @@ struct bpf_create_map_attr { __u32 btf_key_type_id; __u32 btf_value_type_id; __u32 map_ifindex; + __u32 inner_map_fd; }; int bpf_create_map_xattr(const struct bpf_create_map_attr *create_attr); diff --git a/tools/testing/selftests/bpf/Makefile b/tools/testing/selftests/bpf/Makefile index 17a7a5818ee1..daed162043c2 100644 --- a/tools/testing/selftests/bpf/Makefile +++ b/tools/testing/selftests/bpf/Makefile @@ -23,7 +23,7 @@ $(TEST_CUSTOM_PROGS): $(OUTPUT)/%: %.c TEST_GEN_PROGS = test_verifier test_tag test_maps test_lru_map test_lpm_map test_progs \ test_align test_verifier_log test_dev_cgroup test_tcpbpf_user \ test_sock test_btf test_sockmap test_lirc_mode2_user get_cgroup_id_user \ - test_socket_cookie test_cgroup_storage + test_socket_cookie test_cgroup_storage test_select_reuseport TEST_GEN_FILES = test_pkt_access.o test_xdp.o test_l4lb.o test_tcp_estats.o test_obj_id.o \ test_pkt_md_access.o test_xdp_redirect.o test_xdp_meta.o sockmap_parse_prog.o \ @@ -34,7 +34,7 @@ TEST_GEN_FILES = test_pkt_access.o test_xdp.o test_l4lb.o test_tcp_estats.o test test_btf_haskv.o test_btf_nokv.o test_sockmap_kern.o test_tunnel_kern.o \ test_get_stack_rawtp.o test_sockmap_kern.o test_sockhash_kern.o \ test_lwt_seg6local.o sendmsg4_prog.o sendmsg6_prog.o test_lirc_mode2_kern.o \ - get_cgroup_id_kern.o socket_cookie_prog.o + get_cgroup_id_kern.o socket_cookie_prog.o test_select_reuseport_kern.o # Order correspond to 'make run_tests' order TEST_PROGS := test_kmod.sh \ diff --git a/tools/testing/selftests/bpf/bpf_helpers.h b/tools/testing/selftests/bpf/bpf_helpers.h index 9ba1c72d7cf5..5c32266c2c38 100644 --- a/tools/testing/selftests/bpf/bpf_helpers.h +++ b/tools/testing/selftests/bpf/bpf_helpers.h @@ -111,6 +111,8 @@ static int (*bpf_xdp_adjust_tail)(void *ctx, int offset) = static int (*bpf_skb_get_xfrm_state)(void *ctx, int index, void *state, int size, int flags) = (void *) BPF_FUNC_skb_get_xfrm_state; +static int (*bpf_sk_select_reuseport)(void *ctx, void *map, void *key, __u32 flags) = + (void *) BPF_FUNC_sk_select_reuseport; static int (*bpf_get_stack)(void *ctx, void *buf, int size, int flags) = (void *) BPF_FUNC_get_stack; static int (*bpf_fib_lookup)(void *ctx, struct bpf_fib_lookup *params, @@ -173,6 +175,8 @@ struct bpf_map_def { static int (*bpf_skb_load_bytes)(void *ctx, int off, void *to, int len) = (void *) BPF_FUNC_skb_load_bytes; +static int (*bpf_skb_load_bytes_relative)(void *ctx, int off, void *to, int len, __u32 start_header) = + (void *) BPF_FUNC_skb_load_bytes_relative; static int (*bpf_skb_store_bytes)(void *ctx, int off, void *from, int len, int flags) = (void *) BPF_FUNC_skb_store_bytes; static int (*bpf_l3_csum_replace)(void *ctx, int off, int from, int to, int flags) = diff --git a/tools/testing/selftests/bpf/test_select_reuseport.c b/tools/testing/selftests/bpf/test_select_reuseport.c new file mode 100644 index 000000000000..75646d9b34aa --- /dev/null +++ b/tools/testing/selftests/bpf/test_select_reuseport.c @@ -0,0 +1,688 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright (c) 2018 Facebook */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "bpf_rlimit.h" +#include "bpf_util.h" +#include "test_select_reuseport_common.h" + +#define MIN_TCPHDR_LEN 20 +#define UDPHDR_LEN 8 + +#define TCP_SYNCOOKIE_SYSCTL "/proc/sys/net/ipv4/tcp_syncookies" +#define TCP_FO_SYSCTL "/proc/sys/net/ipv4/tcp_fastopen" +#define REUSEPORT_ARRAY_SIZE 32 + +static int result_map, tmp_index_ovr_map, linum_map, data_check_map; +static enum result expected_results[NR_RESULTS]; +static int sk_fds[REUSEPORT_ARRAY_SIZE]; +static int reuseport_array, outer_map; +static int select_by_skb_data_prog; +static int saved_tcp_syncookie; +static struct bpf_object *obj; +static int saved_tcp_fo; +static __u32 index_zero; +static int epfd; + +static union sa46 { + struct sockaddr_in6 v6; + struct sockaddr_in v4; + sa_family_t family; +} srv_sa; + +#define CHECK(condition, tag, format...) ({ \ + int __ret = !!(condition); \ + if (__ret) { \ + printf("%s(%d):FAIL:%s ", __func__, __LINE__, tag); \ + printf(format); \ + exit(-1); \ + } \ +}) + +static void create_maps(void) +{ + struct bpf_create_map_attr attr = {}; + + /* Creating reuseport_array */ + attr.name = "reuseport_array"; + attr.map_type = BPF_MAP_TYPE_REUSEPORT_SOCKARRAY; + attr.key_size = sizeof(__u32); + attr.value_size = sizeof(__u32); + attr.max_entries = REUSEPORT_ARRAY_SIZE; + + reuseport_array = bpf_create_map_xattr(&attr); + CHECK(reuseport_array == -1, "creating reuseport_array", + "reuseport_array:%d errno:%d\n", reuseport_array, errno); + + /* Creating outer_map */ + attr.name = "outer_map"; + attr.map_type = BPF_MAP_TYPE_ARRAY_OF_MAPS; + attr.key_size = sizeof(__u32); + attr.value_size = sizeof(__u32); + attr.max_entries = 1; + attr.inner_map_fd = reuseport_array; + outer_map = bpf_create_map_xattr(&attr); + CHECK(outer_map == -1, "creating outer_map", + "outer_map:%d errno:%d\n", outer_map, errno); +} + +static void prepare_bpf_obj(void) +{ + struct bpf_program *prog; + struct bpf_map *map; + int err; + struct bpf_object_open_attr attr = { + .file = "test_select_reuseport_kern.o", + .prog_type = BPF_PROG_TYPE_SK_REUSEPORT, + }; + + obj = bpf_object__open_xattr(&attr); + CHECK(IS_ERR_OR_NULL(obj), "open test_select_reuseport_kern.o", + "obj:%p PTR_ERR(obj):%ld\n", obj, PTR_ERR(obj)); + + prog = bpf_program__next(NULL, obj); + CHECK(!prog, "get first bpf_program", "!prog\n"); + bpf_program__set_type(prog, attr.prog_type); + + map = bpf_object__find_map_by_name(obj, "outer_map"); + CHECK(!map, "find outer_map", "!map\n"); + err = bpf_map__reuse_fd(map, outer_map); + CHECK(err, "reuse outer_map", "err:%d\n", err); + + err = bpf_object__load(obj); + CHECK(err, "load bpf_object", "err:%d\n", err); + + select_by_skb_data_prog = bpf_program__fd(prog); + CHECK(select_by_skb_data_prog == -1, "get prog fd", + "select_by_skb_data_prog:%d\n", select_by_skb_data_prog); + + map = bpf_object__find_map_by_name(obj, "result_map"); + CHECK(!map, "find result_map", "!map\n"); + result_map = bpf_map__fd(map); + CHECK(result_map == -1, "get result_map fd", + "result_map:%d\n", result_map); + + map = bpf_object__find_map_by_name(obj, "tmp_index_ovr_map"); + CHECK(!map, "find tmp_index_ovr_map", "!map\n"); + tmp_index_ovr_map = bpf_map__fd(map); + CHECK(tmp_index_ovr_map == -1, "get tmp_index_ovr_map fd", + "tmp_index_ovr_map:%d\n", tmp_index_ovr_map); + + map = bpf_object__find_map_by_name(obj, "linum_map"); + CHECK(!map, "find linum_map", "!map\n"); + linum_map = bpf_map__fd(map); + CHECK(linum_map == -1, "get linum_map fd", + "linum_map:%d\n", linum_map); + + map = bpf_object__find_map_by_name(obj, "data_check_map"); + CHECK(!map, "find data_check_map", "!map\n"); + data_check_map = bpf_map__fd(map); + CHECK(data_check_map == -1, "get data_check_map fd", + "data_check_map:%d\n", data_check_map); +} + +static void sa46_init_loopback(union sa46 *sa, sa_family_t family) +{ + memset(sa, 0, sizeof(*sa)); + sa->family = family; + if (sa->family == AF_INET6) + sa->v6.sin6_addr = in6addr_loopback; + else + sa->v4.sin_addr.s_addr = htonl(INADDR_LOOPBACK); +} + +static void sa46_init_inany(union sa46 *sa, sa_family_t family) +{ + memset(sa, 0, sizeof(*sa)); + sa->family = family; + if (sa->family == AF_INET6) + sa->v6.sin6_addr = in6addr_any; + else + sa->v4.sin_addr.s_addr = INADDR_ANY; +} + +static int read_int_sysctl(const char *sysctl) +{ + char buf[16]; + int fd, ret; + + fd = open(sysctl, 0); + CHECK(fd == -1, "open(sysctl)", "sysctl:%s fd:%d errno:%d\n", + sysctl, fd, errno); + + ret = read(fd, buf, sizeof(buf)); + CHECK(ret <= 0, "read(sysctl)", "sysctl:%s ret:%d errno:%d\n", + sysctl, ret, errno); + close(fd); + + return atoi(buf); +} + +static void write_int_sysctl(const char *sysctl, int v) +{ + int fd, ret, size; + char buf[16]; + + fd = open(sysctl, O_RDWR); + CHECK(fd == -1, "open(sysctl)", "sysctl:%s fd:%d errno:%d\n", + sysctl, fd, errno); + + size = snprintf(buf, sizeof(buf), "%d", v); + ret = write(fd, buf, size); + CHECK(ret != size, "write(sysctl)", + "sysctl:%s ret:%d size:%d errno:%d\n", sysctl, ret, size, errno); + close(fd); +} + +static void restore_sysctls(void) +{ + write_int_sysctl(TCP_FO_SYSCTL, saved_tcp_fo); + write_int_sysctl(TCP_SYNCOOKIE_SYSCTL, saved_tcp_syncookie); +} + +static void enable_fastopen(void) +{ + int fo; + + fo = read_int_sysctl(TCP_FO_SYSCTL); + write_int_sysctl(TCP_FO_SYSCTL, fo | 7); +} + +static void enable_syncookie(void) +{ + write_int_sysctl(TCP_SYNCOOKIE_SYSCTL, 2); +} + +static void disable_syncookie(void) +{ + write_int_sysctl(TCP_SYNCOOKIE_SYSCTL, 0); +} + +static __u32 get_linum(void) +{ + __u32 linum; + int err; + + err = bpf_map_lookup_elem(linum_map, &index_zero, &linum); + CHECK(err == -1, "lookup_elem(linum_map)", "err:%d errno:%d\n", + err, errno); + + return linum; +} + +static void check_data(int type, sa_family_t family, const struct cmd *cmd, + int cli_fd) +{ + struct data_check expected = {}, result; + union sa46 cli_sa; + socklen_t addrlen; + int err; + + addrlen = sizeof(cli_sa); + err = getsockname(cli_fd, (struct sockaddr *)&cli_sa, + &addrlen); + CHECK(err == -1, "getsockname(cli_fd)", "err:%d errno:%d\n", + err, errno); + + err = bpf_map_lookup_elem(data_check_map, &index_zero, &result); + CHECK(err == -1, "lookup_elem(data_check_map)", "err:%d errno:%d\n", + err, errno); + + if (type == SOCK_STREAM) { + expected.len = MIN_TCPHDR_LEN; + expected.ip_protocol = IPPROTO_TCP; + } else { + expected.len = UDPHDR_LEN; + expected.ip_protocol = IPPROTO_UDP; + } + + if (family == AF_INET6) { + expected.eth_protocol = htons(ETH_P_IPV6); + expected.bind_inany = !srv_sa.v6.sin6_addr.s6_addr32[3] && + !srv_sa.v6.sin6_addr.s6_addr32[2] && + !srv_sa.v6.sin6_addr.s6_addr32[1] && + !srv_sa.v6.sin6_addr.s6_addr32[0]; + + memcpy(&expected.skb_addrs[0], cli_sa.v6.sin6_addr.s6_addr32, + sizeof(cli_sa.v6.sin6_addr)); + memcpy(&expected.skb_addrs[4], &in6addr_loopback, + sizeof(in6addr_loopback)); + expected.skb_ports[0] = cli_sa.v6.sin6_port; + expected.skb_ports[1] = srv_sa.v6.sin6_port; + } else { + expected.eth_protocol = htons(ETH_P_IP); + expected.bind_inany = !srv_sa.v4.sin_addr.s_addr; + + expected.skb_addrs[0] = cli_sa.v4.sin_addr.s_addr; + expected.skb_addrs[1] = htonl(INADDR_LOOPBACK); + expected.skb_ports[0] = cli_sa.v4.sin_port; + expected.skb_ports[1] = srv_sa.v4.sin_port; + } + + if (memcmp(&result, &expected, offsetof(struct data_check, + equal_check_end))) { + printf("unexpected data_check\n"); + printf(" result: (0x%x, %u, %u)\n", + result.eth_protocol, result.ip_protocol, + result.bind_inany); + printf("expected: (0x%x, %u, %u)\n", + expected.eth_protocol, expected.ip_protocol, + expected.bind_inany); + CHECK(1, "data_check result != expected", + "bpf_prog_linum:%u\n", get_linum()); + } + + CHECK(!result.hash, "data_check result.hash empty", + "result.hash:%u", result.hash); + + expected.len += cmd ? sizeof(*cmd) : 0; + if (type == SOCK_STREAM) + CHECK(expected.len > result.len, "expected.len > result.len", + "expected.len:%u result.len:%u bpf_prog_linum:%u\n", + expected.len, result.len, get_linum()); + else + CHECK(expected.len != result.len, "expected.len != result.len", + "expected.len:%u result.len:%u bpf_prog_linum:%u\n", + expected.len, result.len, get_linum()); +} + +static void check_results(void) +{ + __u32 results[NR_RESULTS]; + __u32 i, broken = 0; + int err; + + for (i = 0; i < NR_RESULTS; i++) { + err = bpf_map_lookup_elem(result_map, &i, &results[i]); + CHECK(err == -1, "lookup_elem(result_map)", + "i:%u err:%d errno:%d\n", i, err, errno); + } + + for (i = 0; i < NR_RESULTS; i++) { + if (results[i] != expected_results[i]) { + broken = i; + break; + } + } + + if (i == NR_RESULTS) + return; + + printf("unexpected result\n"); + printf(" result: ["); + printf("%u", results[0]); + for (i = 1; i < NR_RESULTS; i++) + printf(", %u", results[i]); + printf("]\n"); + + printf("expected: ["); + printf("%u", expected_results[0]); + for (i = 1; i < NR_RESULTS; i++) + printf(", %u", expected_results[i]); + printf("]\n"); + + CHECK(expected_results[broken] != results[broken], + "unexpected result", + "expected_results[%u] != results[%u] bpf_prog_linum:%u\n", + broken, broken, get_linum()); +} + +static int send_data(int type, sa_family_t family, void *data, size_t len, + enum result expected) +{ + union sa46 cli_sa; + int fd, err; + + fd = socket(family, type, 0); + CHECK(fd == -1, "socket()", "fd:%d errno:%d\n", fd, errno); + + sa46_init_loopback(&cli_sa, family); + err = bind(fd, (struct sockaddr *)&cli_sa, sizeof(cli_sa)); + CHECK(fd == -1, "bind(cli_sa)", "err:%d errno:%d\n", err, errno); + + err = sendto(fd, data, len, MSG_FASTOPEN, (struct sockaddr *)&srv_sa, + sizeof(srv_sa)); + CHECK(err != len && expected >= PASS, + "sendto()", "family:%u err:%d errno:%d expected:%d\n", + family, err, errno, expected); + + return fd; +} + +static void do_test(int type, sa_family_t family, struct cmd *cmd, + enum result expected) +{ + int nev, srv_fd, cli_fd; + struct epoll_event ev; + struct cmd rcv_cmd; + ssize_t nread; + + cli_fd = send_data(type, family, cmd, cmd ? sizeof(*cmd) : 0, + expected); + nev = epoll_wait(epfd, &ev, 1, expected >= PASS ? 5 : 0); + CHECK((nev <= 0 && expected >= PASS) || + (nev > 0 && expected < PASS), + "nev <> expected", + "nev:%d expected:%d type:%d family:%d data:(%d, %d)\n", + nev, expected, type, family, + cmd ? cmd->reuseport_index : -1, + cmd ? cmd->pass_on_failure : -1); + check_results(); + check_data(type, family, cmd, cli_fd); + + if (expected < PASS) + return; + + CHECK(expected != PASS_ERR_SK_SELECT_REUSEPORT && + cmd->reuseport_index != ev.data.u32, + "check cmd->reuseport_index", + "cmd:(%u, %u) ev.data.u32:%u\n", + cmd->pass_on_failure, cmd->reuseport_index, ev.data.u32); + + srv_fd = sk_fds[ev.data.u32]; + if (type == SOCK_STREAM) { + int new_fd = accept(srv_fd, NULL, 0); + + CHECK(new_fd == -1, "accept(srv_fd)", + "ev.data.u32:%u new_fd:%d errno:%d\n", + ev.data.u32, new_fd, errno); + + nread = recv(new_fd, &rcv_cmd, sizeof(rcv_cmd), MSG_DONTWAIT); + CHECK(nread != sizeof(rcv_cmd), + "recv(new_fd)", + "ev.data.u32:%u nread:%zd sizeof(rcv_cmd):%zu errno:%d\n", + ev.data.u32, nread, sizeof(rcv_cmd), errno); + + close(new_fd); + } else { + nread = recv(srv_fd, &rcv_cmd, sizeof(rcv_cmd), MSG_DONTWAIT); + CHECK(nread != sizeof(rcv_cmd), + "recv(sk_fds)", + "ev.data.u32:%u nread:%zd sizeof(rcv_cmd):%zu errno:%d\n", + ev.data.u32, nread, sizeof(rcv_cmd), errno); + } + + close(cli_fd); +} + +static void test_err_inner_map(int type, sa_family_t family) +{ + struct cmd cmd = { + .reuseport_index = 0, + .pass_on_failure = 0, + }; + + printf("%s: ", __func__); + expected_results[DROP_ERR_INNER_MAP]++; + do_test(type, family, &cmd, DROP_ERR_INNER_MAP); + printf("OK\n"); +} + +static void test_err_skb_data(int type, sa_family_t family) +{ + printf("%s: ", __func__); + expected_results[DROP_ERR_SKB_DATA]++; + do_test(type, family, NULL, DROP_ERR_SKB_DATA); + printf("OK\n"); +} + +static void test_err_sk_select_port(int type, sa_family_t family) +{ + struct cmd cmd = { + .reuseport_index = REUSEPORT_ARRAY_SIZE, + .pass_on_failure = 0, + }; + + printf("%s: ", __func__); + expected_results[DROP_ERR_SK_SELECT_REUSEPORT]++; + do_test(type, family, &cmd, DROP_ERR_SK_SELECT_REUSEPORT); + printf("OK\n"); +} + +static void test_pass(int type, sa_family_t family) +{ + struct cmd cmd; + int i; + + printf("%s: ", __func__); + cmd.pass_on_failure = 0; + for (i = 0; i < REUSEPORT_ARRAY_SIZE; i++) { + expected_results[PASS]++; + cmd.reuseport_index = i; + do_test(type, family, &cmd, PASS); + } + printf("OK\n"); +} + +static void test_syncookie(int type, sa_family_t family) +{ + int err, tmp_index = 1; + struct cmd cmd = { + .reuseport_index = 0, + .pass_on_failure = 0, + }; + + if (type != SOCK_STREAM) + return; + + printf("%s: ", __func__); + /* + * +1 for TCP-SYN and + * +1 for the TCP-ACK (ack the syncookie) + */ + expected_results[PASS] += 2; + enable_syncookie(); + /* + * Simulate TCP-SYN and TCP-ACK are handled by two different sk: + * TCP-SYN: select sk_fds[tmp_index = 1] tmp_index is from the + * tmp_index_ovr_map + * TCP-ACK: select sk_fds[reuseport_index = 0] reuseport_index + * is from the cmd.reuseport_index + */ + err = bpf_map_update_elem(tmp_index_ovr_map, &index_zero, + &tmp_index, BPF_ANY); + CHECK(err == -1, "update_elem(tmp_index_ovr_map, 0, 1)", + "err:%d errno:%d\n", err, errno); + do_test(type, family, &cmd, PASS); + err = bpf_map_lookup_elem(tmp_index_ovr_map, &index_zero, + &tmp_index); + CHECK(err == -1 || tmp_index != -1, + "lookup_elem(tmp_index_ovr_map)", + "err:%d errno:%d tmp_index:%d\n", + err, errno, tmp_index); + disable_syncookie(); + printf("OK\n"); +} + +static void test_pass_on_err(int type, sa_family_t family) +{ + struct cmd cmd = { + .reuseport_index = REUSEPORT_ARRAY_SIZE, + .pass_on_failure = 1, + }; + + printf("%s: ", __func__); + expected_results[PASS_ERR_SK_SELECT_REUSEPORT] += 1; + do_test(type, family, &cmd, PASS_ERR_SK_SELECT_REUSEPORT); + printf("OK\n"); +} + +static void prepare_sk_fds(int type, sa_family_t family, bool inany) +{ + const int first = REUSEPORT_ARRAY_SIZE - 1; + int i, err, optval = 1; + struct epoll_event ev; + socklen_t addrlen; + + if (inany) + sa46_init_inany(&srv_sa, family); + else + sa46_init_loopback(&srv_sa, family); + addrlen = sizeof(srv_sa); + + /* + * The sk_fds[] is filled from the back such that the order + * is exactly opposite to the (struct sock_reuseport *)reuse->socks[]. + */ + for (i = first; i >= 0; i--) { + sk_fds[i] = socket(family, type, 0); + CHECK(sk_fds[i] == -1, "socket()", "sk_fds[%d]:%d errno:%d\n", + i, sk_fds[i], errno); + err = setsockopt(sk_fds[i], SOL_SOCKET, SO_REUSEPORT, + &optval, sizeof(optval)); + CHECK(err == -1, "setsockopt(SO_REUSEPORT)", + "sk_fds[%d] err:%d errno:%d\n", + i, err, errno); + + if (i == first) { + err = setsockopt(sk_fds[i], SOL_SOCKET, + SO_ATTACH_REUSEPORT_EBPF, + &select_by_skb_data_prog, + sizeof(select_by_skb_data_prog)); + CHECK(err == -1, "setsockopt(SO_ATTACH_REUEPORT_EBPF)", + "err:%d errno:%d\n", err, errno); + } + + err = bind(sk_fds[i], (struct sockaddr *)&srv_sa, addrlen); + CHECK(err == -1, "bind()", "sk_fds[%d] err:%d errno:%d\n", + i, err, errno); + + if (type == SOCK_STREAM) { + err = listen(sk_fds[i], 10); + CHECK(err == -1, "listen()", + "sk_fds[%d] err:%d errno:%d\n", + i, err, errno); + } + + err = bpf_map_update_elem(reuseport_array, &i, &sk_fds[i], + BPF_NOEXIST); + CHECK(err == -1, "update_elem(reuseport_array)", + "sk_fds[%d] err:%d errno:%d\n", i, err, errno); + + if (i == first) { + socklen_t addrlen = sizeof(srv_sa); + + err = getsockname(sk_fds[i], (struct sockaddr *)&srv_sa, + &addrlen); + CHECK(err == -1, "getsockname()", + "sk_fds[%d] err:%d errno:%d\n", i, err, errno); + } + } + + epfd = epoll_create(1); + CHECK(epfd == -1, "epoll_create(1)", + "epfd:%d errno:%d\n", epfd, errno); + + ev.events = EPOLLIN; + for (i = 0; i < REUSEPORT_ARRAY_SIZE; i++) { + ev.data.u32 = i; + err = epoll_ctl(epfd, EPOLL_CTL_ADD, sk_fds[i], &ev); + CHECK(err, "epoll_ctl(EPOLL_CTL_ADD)", "sk_fds[%d]\n", i); + } +} + +static void setup_per_test(int type, unsigned short family, bool inany) +{ + int ovr = -1, err; + + prepare_sk_fds(type, family, inany); + err = bpf_map_update_elem(tmp_index_ovr_map, &index_zero, &ovr, + BPF_ANY); + CHECK(err == -1, "update_elem(tmp_index_ovr_map, 0, -1)", + "err:%d errno:%d\n", err, errno); +} + +static void cleanup_per_test(void) +{ + int i, err; + + for (i = 0; i < REUSEPORT_ARRAY_SIZE; i++) + close(sk_fds[i]); + close(epfd); + + err = bpf_map_delete_elem(outer_map, &index_zero); + CHECK(err == -1, "delete_elem(outer_map)", + "err:%d errno:%d\n", err, errno); +} + +static void cleanup(void) +{ + close(outer_map); + close(reuseport_array); + bpf_object__close(obj); +} + +static void test_all(void) +{ + /* Extra SOCK_STREAM to test bind_inany==true */ + const int types[] = { SOCK_STREAM, SOCK_DGRAM, SOCK_STREAM }; + const char * const type_strings[] = { "TCP", "UDP", "TCP" }; + const char * const family_strings[] = { "IPv6", "IPv4" }; + const unsigned short families[] = { AF_INET6, AF_INET }; + const bool bind_inany[] = { false, false, true }; + int t, f, err; + + for (f = 0; f < ARRAY_SIZE(families); f++) { + unsigned short family = families[f]; + + for (t = 0; t < ARRAY_SIZE(types); t++) { + bool inany = bind_inany[t]; + int type = types[t]; + + printf("######## %s/%s %s ########\n", + family_strings[f], type_strings[t], + inany ? " INANY " : "LOOPBACK"); + + setup_per_test(type, family, inany); + + test_err_inner_map(type, family); + + /* Install reuseport_array to the outer_map */ + err = bpf_map_update_elem(outer_map, &index_zero, + &reuseport_array, BPF_ANY); + CHECK(err == -1, "update_elem(outer_map)", + "err:%d errno:%d\n", err, errno); + + test_err_skb_data(type, family); + test_err_sk_select_port(type, family); + test_pass(type, family); + test_syncookie(type, family); + test_pass_on_err(type, family); + + cleanup_per_test(); + printf("\n"); + } + } +} + +int main(int argc, const char **argv) +{ + create_maps(); + prepare_bpf_obj(); + saved_tcp_fo = read_int_sysctl(TCP_FO_SYSCTL); + saved_tcp_syncookie = read_int_sysctl(TCP_SYNCOOKIE_SYSCTL); + enable_fastopen(); + disable_syncookie(); + atexit(restore_sysctls); + + test_all(); + + cleanup(); + return 0; +} diff --git a/tools/testing/selftests/bpf/test_select_reuseport_common.h b/tools/testing/selftests/bpf/test_select_reuseport_common.h new file mode 100644 index 000000000000..08eb2a9f145f --- /dev/null +++ b/tools/testing/selftests/bpf/test_select_reuseport_common.h @@ -0,0 +1,36 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* Copyright (c) 2018 Facebook */ + +#ifndef __TEST_SELECT_REUSEPORT_COMMON_H +#define __TEST_SELECT_REUSEPORT_COMMON_H + +#include + +enum result { + DROP_ERR_INNER_MAP, + DROP_ERR_SKB_DATA, + DROP_ERR_SK_SELECT_REUSEPORT, + DROP_MISC, + PASS, + PASS_ERR_SK_SELECT_REUSEPORT, + NR_RESULTS, +}; + +struct cmd { + __u32 reuseport_index; + __u32 pass_on_failure; +}; + +struct data_check { + __u32 ip_protocol; + __u32 skb_addrs[8]; + __u16 skb_ports[2]; + __u16 eth_protocol; + __u8 bind_inany; + __u8 equal_check_end[0]; + + __u32 len; + __u32 hash; +}; + +#endif diff --git a/tools/testing/selftests/bpf/test_select_reuseport_kern.c b/tools/testing/selftests/bpf/test_select_reuseport_kern.c new file mode 100644 index 000000000000..5b54ec637ada --- /dev/null +++ b/tools/testing/selftests/bpf/test_select_reuseport_kern.c @@ -0,0 +1,180 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright (c) 2018 Facebook */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "bpf_endian.h" +#include "bpf_helpers.h" +#include "test_select_reuseport_common.h" + +int _version SEC("version") = 1; + +#ifndef offsetof +#define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER) +#endif + +struct bpf_map_def SEC("maps") outer_map = { + .type = BPF_MAP_TYPE_ARRAY_OF_MAPS, + .key_size = sizeof(__u32), + .value_size = sizeof(__u32), + .max_entries = 1, +}; + +struct bpf_map_def SEC("maps") result_map = { + .type = BPF_MAP_TYPE_ARRAY, + .key_size = sizeof(__u32), + .value_size = sizeof(__u32), + .max_entries = NR_RESULTS, +}; + +struct bpf_map_def SEC("maps") tmp_index_ovr_map = { + .type = BPF_MAP_TYPE_ARRAY, + .key_size = sizeof(__u32), + .value_size = sizeof(int), + .max_entries = 1, +}; + +struct bpf_map_def SEC("maps") linum_map = { + .type = BPF_MAP_TYPE_ARRAY, + .key_size = sizeof(__u32), + .value_size = sizeof(__u32), + .max_entries = 1, +}; + +struct bpf_map_def SEC("maps") data_check_map = { + .type = BPF_MAP_TYPE_ARRAY, + .key_size = sizeof(__u32), + .value_size = sizeof(struct data_check), + .max_entries = 1, +}; + +#define GOTO_DONE(_result) ({ \ + result = (_result); \ + linum = __LINE__; \ + goto done; \ +}) + +SEC("select_by_skb_data") +int _select_by_skb_data(struct sk_reuseport_md *reuse_md) +{ + __u32 linum, index = 0, flags = 0, index_zero = 0; + __u32 *result_cnt, *linum_value; + struct data_check data_check = {}; + struct cmd *cmd, cmd_copy; + void *data, *data_end; + void *reuseport_array; + enum result result; + int *index_ovr; + int err; + + data = reuse_md->data; + data_end = reuse_md->data_end; + data_check.len = reuse_md->len; + data_check.eth_protocol = reuse_md->eth_protocol; + data_check.ip_protocol = reuse_md->ip_protocol; + data_check.hash = reuse_md->hash; + data_check.bind_inany = reuse_md->bind_inany; + if (data_check.eth_protocol == bpf_htons(ETH_P_IP)) { + if (bpf_skb_load_bytes_relative(reuse_md, + offsetof(struct iphdr, saddr), + data_check.skb_addrs, 8, + BPF_HDR_START_NET)) + GOTO_DONE(DROP_MISC); + } else { + if (bpf_skb_load_bytes_relative(reuse_md, + offsetof(struct ipv6hdr, saddr), + data_check.skb_addrs, 32, + BPF_HDR_START_NET)) + GOTO_DONE(DROP_MISC); + } + + /* + * The ip_protocol could be a compile time decision + * if the bpf_prog.o is dedicated to either TCP or + * UDP. + * + * Otherwise, reuse_md->ip_protocol or + * the protocol field in the iphdr can be used. + */ + if (data_check.ip_protocol == IPPROTO_TCP) { + struct tcphdr *th = data; + + if (th + 1 > data_end) + GOTO_DONE(DROP_MISC); + + data_check.skb_ports[0] = th->source; + data_check.skb_ports[1] = th->dest; + + if ((th->doff << 2) + sizeof(*cmd) > data_check.len) + GOTO_DONE(DROP_ERR_SKB_DATA); + if (bpf_skb_load_bytes(reuse_md, th->doff << 2, &cmd_copy, + sizeof(cmd_copy))) + GOTO_DONE(DROP_MISC); + cmd = &cmd_copy; + } else if (data_check.ip_protocol == IPPROTO_UDP) { + struct udphdr *uh = data; + + if (uh + 1 > data_end) + GOTO_DONE(DROP_MISC); + + data_check.skb_ports[0] = uh->source; + data_check.skb_ports[1] = uh->dest; + + if (sizeof(struct udphdr) + sizeof(*cmd) > data_check.len) + GOTO_DONE(DROP_ERR_SKB_DATA); + if (data + sizeof(struct udphdr) + sizeof(*cmd) > data_end) { + if (bpf_skb_load_bytes(reuse_md, sizeof(struct udphdr), + &cmd_copy, sizeof(cmd_copy))) + GOTO_DONE(DROP_MISC); + cmd = &cmd_copy; + } else { + cmd = data + sizeof(struct udphdr); + } + } else { + GOTO_DONE(DROP_MISC); + } + + reuseport_array = bpf_map_lookup_elem(&outer_map, &index_zero); + if (!reuseport_array) + GOTO_DONE(DROP_ERR_INNER_MAP); + + index = cmd->reuseport_index; + index_ovr = bpf_map_lookup_elem(&tmp_index_ovr_map, &index_zero); + if (!index_ovr) + GOTO_DONE(DROP_MISC); + + if (*index_ovr != -1) { + index = *index_ovr; + *index_ovr = -1; + } + err = bpf_sk_select_reuseport(reuse_md, reuseport_array, &index, + flags); + if (!err) + GOTO_DONE(PASS); + + if (cmd->pass_on_failure) + GOTO_DONE(PASS_ERR_SK_SELECT_REUSEPORT); + else + GOTO_DONE(DROP_ERR_SK_SELECT_REUSEPORT); + +done: + result_cnt = bpf_map_lookup_elem(&result_map, &result); + if (!result_cnt) + return SK_DROP; + + bpf_map_update_elem(&linum_map, &index_zero, &linum, BPF_ANY); + bpf_map_update_elem(&data_check_map, &index_zero, &data_check, BPF_ANY); + + (*result_cnt)++; + return result < PASS ? SK_DROP : SK_PASS; +} + +char _license[] SEC("license") = "GPL";