From 435bf0d3f99a164df7e8c30428cef266b91d1d3b Mon Sep 17 00:00:00 2001 From: John Fastabend Date: Wed, 18 Oct 2017 07:10:15 -0700 Subject: bpf: enforce TCP only support for sockmap Only TCP sockets have been tested and at the moment the state change callback only handles TCP sockets. This adds a check to ensure that sockets actually being added are TCP sockets. For net-next we can consider UDP support. Signed-off-by: John Fastabend Acked-by: Daniel Borkmann Acked-by: Alexei Starovoitov Signed-off-by: David S. Miller --- kernel/bpf/sockmap.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'kernel/bpf/sockmap.c') diff --git a/kernel/bpf/sockmap.c b/kernel/bpf/sockmap.c index 6424ce0e4969..c68899d5b246 100644 --- a/kernel/bpf/sockmap.c +++ b/kernel/bpf/sockmap.c @@ -840,6 +840,12 @@ static int sock_map_update_elem(struct bpf_map *map, return -EINVAL; } + if (skops.sk->sk_type != SOCK_STREAM || + skops.sk->sk_protocol != IPPROTO_TCP) { + fput(socket->file); + return -EOPNOTSUPP; + } + err = sock_map_ctx_update_elem(&skops, map, key, flags); fput(socket->file); return err; -- cgit v1.2.3 From 34f79502bbcfab659b8729da68b5e387f96eb4c1 Mon Sep 17 00:00:00 2001 From: John Fastabend Date: Wed, 18 Oct 2017 07:10:36 -0700 Subject: bpf: avoid preempt enable/disable in sockmap using tcp_skb_cb region SK_SKB BPF programs are run from the socket/tcp context but early in the stack before much of the TCP metadata is needed in tcp_skb_cb. So we can use some unused fields to place BPF metadata needed for SK_SKB programs when implementing the redirect function. This allows us to drop the preempt disable logic. It does however require an API change so sk_redirect_map() has been updated to additionally provide ctx_ptr to skb. Note, we do however continue to disable/enable preemption around actual BPF program running to account for map updates. Signed-off-by: John Fastabend Acked-by: Daniel Borkmann Acked-by: Alexei Starovoitov Signed-off-by: David S. Miller --- kernel/bpf/sockmap.c | 19 +++++++++---------- 1 file changed, 9 insertions(+), 10 deletions(-) (limited to 'kernel/bpf/sockmap.c') diff --git a/kernel/bpf/sockmap.c b/kernel/bpf/sockmap.c index c68899d5b246..beaabb21c3a3 100644 --- a/kernel/bpf/sockmap.c +++ b/kernel/bpf/sockmap.c @@ -39,6 +39,7 @@ #include #include #include +#include struct bpf_stab { struct bpf_map map; @@ -101,9 +102,16 @@ static int smap_verdict_func(struct smap_psock *psock, struct sk_buff *skb) return SK_DROP; skb_orphan(skb); + /* We need to ensure that BPF metadata for maps is also cleared + * when we orphan the skb so that we don't have the possibility + * to reference a stale map. + */ + TCP_SKB_CB(skb)->bpf.map = NULL; skb->sk = psock->sock; bpf_compute_data_end(skb); + preempt_disable(); rc = (*prog->bpf_func)(skb, prog->insnsi); + preempt_enable(); skb->sk = NULL; return rc; @@ -114,17 +122,10 @@ static void smap_do_verdict(struct smap_psock *psock, struct sk_buff *skb) struct sock *sk; int rc; - /* Because we use per cpu values to feed input from sock redirect - * in BPF program to do_sk_redirect_map() call we need to ensure we - * are not preempted. RCU read lock is not sufficient in this case - * with CONFIG_PREEMPT_RCU enabled so we must be explicit here. - */ - preempt_disable(); rc = smap_verdict_func(psock, skb); switch (rc) { case SK_REDIRECT: - sk = do_sk_redirect_map(); - preempt_enable(); + sk = do_sk_redirect_map(skb); if (likely(sk)) { struct smap_psock *peer = smap_psock_sk(sk); @@ -141,8 +142,6 @@ static void smap_do_verdict(struct smap_psock *psock, struct sk_buff *skb) /* Fall through and free skb otherwise */ case SK_DROP: default: - if (rc != SK_REDIRECT) - preempt_enable(); kfree_skb(skb); } } -- cgit v1.2.3 From fb50df8d32283cd95932a182a46a10070c4a8832 Mon Sep 17 00:00:00 2001 From: John Fastabend Date: Wed, 18 Oct 2017 07:11:22 -0700 Subject: bpf: require CAP_NET_ADMIN when using sockmap maps Restrict sockmap to CAP_NET_ADMIN. Signed-off-by: John Fastabend Acked-by: Daniel Borkmann Acked-by: Alexei Starovoitov Signed-off-by: David S. Miller --- kernel/bpf/sockmap.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel/bpf/sockmap.c') diff --git a/kernel/bpf/sockmap.c b/kernel/bpf/sockmap.c index beaabb21c3a3..2b6eb35ae5d3 100644 --- a/kernel/bpf/sockmap.c +++ b/kernel/bpf/sockmap.c @@ -486,6 +486,9 @@ static struct bpf_map *sock_map_alloc(union bpf_attr *attr) int err = -EINVAL; u64 cost; + if (!capable(CAP_NET_ADMIN)) + return ERR_PTR(-EPERM); + /* check sanity of attributes */ if (attr->max_entries == 0 || attr->key_size != 4 || attr->value_size != 4 || attr->map_flags & ~BPF_F_NUMA_NODE) -- cgit v1.2.3 From 8108a77515126f6db4374e8593956e20430307c0 Mon Sep 17 00:00:00 2001 From: John Fastabend Date: Fri, 27 Oct 2017 09:45:34 -0700 Subject: bpf: bpf_compute_data uses incorrect cb structure SK_SKB program types use bpf_compute_data to store the end of the packet data. However, bpf_compute_data assumes the cb is stored in the qdisc layer format. But, for SK_SKB this is the wrong layer of the stack for this type. It happens to work (sort of!) because in most cases nothing happens to be overwritten today. This is very fragile and error prone. Fortunately, we have another hole in tcp_skb_cb we can use so lets put the data_end value there. Note, SK_SKB program types do not use data_meta, they are failed by sk_skb_is_valid_access(). Signed-off-by: John Fastabend Acked-by: Alexei Starovoitov Signed-off-by: David S. Miller --- kernel/bpf/sockmap.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) (limited to 'kernel/bpf/sockmap.c') diff --git a/kernel/bpf/sockmap.c b/kernel/bpf/sockmap.c index 2b6eb35ae5d3..6778fb773934 100644 --- a/kernel/bpf/sockmap.c +++ b/kernel/bpf/sockmap.c @@ -93,6 +93,14 @@ static inline struct smap_psock *smap_psock_sk(const struct sock *sk) return rcu_dereference_sk_user_data(sk); } +/* compute the linear packet data range [data, data_end) for skb when + * sk_skb type programs are in use. + */ +static inline void bpf_compute_data_end_sk_skb(struct sk_buff *skb) +{ + TCP_SKB_CB(skb)->bpf.data_end = skb->data + skb_headlen(skb); +} + static int smap_verdict_func(struct smap_psock *psock, struct sk_buff *skb) { struct bpf_prog *prog = READ_ONCE(psock->bpf_verdict); @@ -108,7 +116,7 @@ static int smap_verdict_func(struct smap_psock *psock, struct sk_buff *skb) */ TCP_SKB_CB(skb)->bpf.map = NULL; skb->sk = psock->sock; - bpf_compute_data_end(skb); + bpf_compute_data_end_sk_skb(skb); preempt_disable(); rc = (*prog->bpf_func)(skb, prog->insnsi); preempt_enable(); @@ -368,7 +376,7 @@ static int smap_parse_func_strparser(struct strparser *strp, * any socket yet. */ skb->sk = psock->sock; - bpf_compute_data_end(skb); + bpf_compute_data_end_sk_skb(skb); rc = (*prog->bpf_func)(skb, prog->insnsi); skb->sk = NULL; rcu_read_unlock(); -- cgit v1.2.3 From bfa640757e9378c2f26867e723f1287e94f5a7ad Mon Sep 17 00:00:00 2001 From: John Fastabend Date: Fri, 27 Oct 2017 09:45:53 -0700 Subject: bpf: rename sk_actions to align with bpf infrastructure Recent additions to support multiple programs in cgroups impose a strict requirement, "all yes is yes, any no is no". To enforce this the infrastructure requires the 'no' return code, SK_DROP in this case, to be 0. To apply these rules to SK_SKB program types the sk_actions return codes need to be adjusted. This fix adds SK_PASS and makes 'SK_DROP = 0'. Finally, remove SK_ABORTED to remove any chance that the API may allow aborted program flows to be passed up the stack. This would be incorrect behavior and allow programs to break existing policies. Signed-off-by: John Fastabend Acked-by: Alexei Starovoitov Signed-off-by: David S. Miller --- kernel/bpf/sockmap.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'kernel/bpf/sockmap.c') diff --git a/kernel/bpf/sockmap.c b/kernel/bpf/sockmap.c index 6778fb773934..66f00a2b27f4 100644 --- a/kernel/bpf/sockmap.c +++ b/kernel/bpf/sockmap.c @@ -122,7 +122,8 @@ static int smap_verdict_func(struct smap_psock *psock, struct sk_buff *skb) preempt_enable(); skb->sk = NULL; - return rc; + return rc == SK_PASS ? + (TCP_SKB_CB(skb)->bpf.map ? SK_REDIRECT : SK_PASS) : SK_DROP; } static void smap_do_verdict(struct smap_psock *psock, struct sk_buff *skb) -- cgit v1.2.3 From 04686ef299db5ff299bfc5a4504b189e46842078 Mon Sep 17 00:00:00 2001 From: John Fastabend Date: Tue, 31 Oct 2017 19:17:31 -0700 Subject: bpf: remove SK_REDIRECT from UAPI Now that SK_REDIRECT is no longer a valid return code. Remove it from the UAPI completely. Then do a namespace remapping internal to sockmap so SK_REDIRECT is no longer externally visible. Patchs primary change is to do a namechange from SK_REDIRECT to __SK_REDIRECT Reported-by: Alexei Starovoitov Signed-off-by: John Fastabend Signed-off-by: David S. Miller --- kernel/bpf/sockmap.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-) (limited to 'kernel/bpf/sockmap.c') diff --git a/kernel/bpf/sockmap.c b/kernel/bpf/sockmap.c index 66f00a2b27f4..dbd7b322a86b 100644 --- a/kernel/bpf/sockmap.c +++ b/kernel/bpf/sockmap.c @@ -101,13 +101,19 @@ static inline void bpf_compute_data_end_sk_skb(struct sk_buff *skb) TCP_SKB_CB(skb)->bpf.data_end = skb->data + skb_headlen(skb); } +enum __sk_action { + __SK_DROP = 0, + __SK_PASS, + __SK_REDIRECT, +}; + static int smap_verdict_func(struct smap_psock *psock, struct sk_buff *skb) { struct bpf_prog *prog = READ_ONCE(psock->bpf_verdict); int rc; if (unlikely(!prog)) - return SK_DROP; + return __SK_DROP; skb_orphan(skb); /* We need to ensure that BPF metadata for maps is also cleared @@ -122,8 +128,10 @@ static int smap_verdict_func(struct smap_psock *psock, struct sk_buff *skb) preempt_enable(); skb->sk = NULL; + /* Moving return codes from UAPI namespace into internal namespace */ return rc == SK_PASS ? - (TCP_SKB_CB(skb)->bpf.map ? SK_REDIRECT : SK_PASS) : SK_DROP; + (TCP_SKB_CB(skb)->bpf.map ? __SK_REDIRECT : __SK_PASS) : + __SK_DROP; } static void smap_do_verdict(struct smap_psock *psock, struct sk_buff *skb) @@ -133,7 +141,7 @@ static void smap_do_verdict(struct smap_psock *psock, struct sk_buff *skb) rc = smap_verdict_func(psock, skb); switch (rc) { - case SK_REDIRECT: + case __SK_REDIRECT: sk = do_sk_redirect_map(skb); if (likely(sk)) { struct smap_psock *peer = smap_psock_sk(sk); @@ -149,7 +157,7 @@ static void smap_do_verdict(struct smap_psock *psock, struct sk_buff *skb) } } /* Fall through and free skb otherwise */ - case SK_DROP: + case __SK_DROP: default: kfree_skb(skb); } -- cgit v1.2.3