net: bpf: fix request_sock leak in filter.c
Subsystems (3)
PATCH 1/1 net: bpf: fix request_sock leak in filter.c
A customer reported a request_socket leak in a Calico cloud environment. We found that a BPF program was doing a socket lookup with takes a refcnt on the socket and that it was finding the request_socket but returning the parent LISTEN socket via sk_to_full_sk() without decrementing the child request socket 1st, resulting in request_sock slab object leak. This patch retains the existing behaviour of returning full socks to the caller but it also decrements the child request_socket if one is present before doing so to prevent the leak.
Thanks to Curtis Taylor for all the help in diagnosing and testing this. And thanks to Antoine Tenart for the reproducer and patch input.
--- net/core/filter.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-)
net/core/filter.c +8 -6
@@ -6514,13 +6514,14 @@ __bpf_sk_lookup(struct sk_buff *skb, struct bpf_sock_tuple *tuple, u32 len, {struct sock *sk =__bpf_skc_lookup (skb, tuple, len, caller_net, ifindex, proto, netns_id, flags); +struct sock *sk1 = sk;if (sk) { sk =sk_to_full_sk (sk); -if (!sk_fullsock (sk)) { -sock_gen_put (sk); +if (!sk_fullsock (sk1)) +sock_gen_put (sk1); +if (!sk_fullsock (sk))return NULL ; - } }return sk;
@@ -6551,13 +6552,14 @@ bpf_sk_lookup(struct sk_buff *skb, struct bpf_sock_tuple *tuple, u32 len, {struct sock *sk =bpf_skc_lookup (skb, tuple, len, proto, netns_id, flags); +struct sock *sk1 = sk;if (sk) { sk =sk_to_full_sk (sk); -if (!sk_fullsock (sk)) { -sock_gen_put (sk); +if (!sk_fullsock (sk1)) +sock_gen_put (sk1); +if (!sk_fullsock (sk))return NULL ; - } }return sk;
Signature · 1 line
2.31.1
Hi Jon,
This patch is targeted at the networking subsystem, as such (see the "NETWORKING [GENERAL]" section in MAINTAINERS), you should send it to [email protected] and to the networking maintainers (David, Jakub, Paolo & Eric).
This also fixes an issue and should be targeted at [net] instead of [net-next]. Because of this you'll also need a Fixes: tag.
Quoting Jon Maxwell (2022-06-07 03:38:44)
Quoted 8 lines
A customer reported a
request_socketleak in a Calico cloud environment. We found that a BPF program was doing a socket lookup with takes a refcnt on the socket and that it was finding therequest_socketbut returning the parent LISTEN socket viask_to_full_sk()without decrementing the child request socket 1st, resulting inrequest_sockslab object leak. This patch retains the existing behaviour of returning full socks to the caller but it also decrements the childrequest_socketif one is present before doing so to prevent the leak.
Quoted 6 lines
Thanks to Curtis Taylor for all the help in diagnosing and testing this. And thanks to Antoine Tenart for the reproducer and patch input.
Signed-off-by Jon Maxwell <[email protected]>Tested-by Curtis Taylor <[email protected]>Co-developed-by Antoine Tenart <[email protected]>
You need to put my SoB here when using the above tag. You'll also need to put your SoB at the end of all the above tags instead of the top.
net/core/filter.c __bpf_sk_lookup(struct sk_buff *skb, struct bpf_sock_tuple *tuple, u32 len, · line 6521
Quoted 10 lines
net/core/filter.c +1
-2@@ -6514,13 +6514,14 @@ __bpf_sk_lookup(struct sk_buff *skb, struct bpf_sock_tuple *tuple, u32 len, {struct sock *sk =__bpf_skc_lookup (skb, tuple, len, caller_net, ifindex, proto, netns_id, flags); +struct sock *sk1 = sk;if (sk) { sk =sk_to_full_sk (sk); -if (!sk_fullsock (sk)) { -sock_gen_put (sk);
I'd suggest to add a comment here to explain why sock_gen_put is called on the original sk.
Quoted 9 lines
net/core/filter.c +3
-1+if (!sk_fullsock (sk1)) +sock_gen_put (sk1); +if (!sk_fullsock (sk))return NULL ; - } }return sk;
net/core/filter.c bpf_sk_lookup(struct sk_buff *skb, struct bpf_sock_tuple *tuple, u32 len, · line 6558
Quoted 9 lines
net/core/filter.c +1
-2@@ -6551,13 +6552,14 @@ bpf_sk_lookup(struct sk_buff *skb, struct bpf_sock_tuple *tuple, u32 len, {struct sock *sk =bpf_skc_lookup (skb, tuple, len, proto, netns_id, flags); +struct sock *sk1 = sk;if (sk) { sk =sk_to_full_sk (sk); -if (!sk_fullsock (sk)) { -sock_gen_put (sk);
Ditto.
net/core/filter.c __bpf_sk_lookup(struct sk_buff *skb, struct bpf_sock_tuple *tuple, u32 len, · line 6527
Quoted 8 lines
net/core/filter.c +3
-1+if (!sk_fullsock (sk1)) +sock_gen_put (sk1); +if (!sk_fullsock (sk))return NULL ; - } }return sk;
Thanks!
Antoine
Quoted 72 lines · Antoine Tenart
Hi Jon,
This patch is targeted at the networking subsystem, as such (see the "NETWORKING [GENERAL]" section in
MAINTAINERS), you should send it to [email protected] and to the networking maintainers (David, Jakub, Paolo & Eric).This also fixes an issue and should be targeted at [net] instead of [net-next]. Because of this you'll also need a Fixes: tag.
Quoting Jon Maxwell (2022-06-07 03:38:44)
Quoted 8 lines
A customer reported a
request_socketleak in a Calico cloud environment. We found that a BPF program was doing a socket lookup with takes a refcnt on the socket and that it was finding therequest_socketbut returning the parent LISTEN socket viask_to_full_sk()without decrementing the child request socket 1st, resulting inrequest_sockslab object leak. This patch retains the existing behaviour of returning full socks to the caller but it also decrements the childrequest_socketif one is present before doing so to prevent the leak.Quoted 6 lines
Thanks to Curtis Taylor for all the help in diagnosing and testing this. And thanks to Antoine Tenart for the reproducer and patch input.
Signed-off-by Jon Maxwell <[email protected]>Tested-by Curtis Taylor <[email protected]>Co-developed-by Antoine Tenart <[email protected]>You need to put my SoB here when using the above tag. You'll also need to put your SoB at the end of all the above tags instead of the top.
net/core/filter.c __bpf_sk_lookup(struct sk_buff *skb, struct bpf_sock_tuple *tuple, u32 len, · line 6521
Quoted 10 lines
net/core/filter.c +1
-2@@ -6514,13 +6514,14 @@ __bpf_sk_lookup(struct sk_buff *skb, struct bpf_sock_tuple *tuple, u32 len, {struct sock *sk =__bpf_skc_lookup (skb, tuple, len, caller_net, ifindex, proto, netns_id, flags); +struct sock *sk1 = sk;if (sk) { sk =sk_to_full_sk (sk); -if (!sk_fullsock (sk)) { -sock_gen_put (sk);I'd suggest to add a comment here to explain why
sock_gen_putis called on the original sk.Quoted 9 lines
net/core/filter.c +3
-1+if (!sk_fullsock (sk1)) +sock_gen_put (sk1); +if (!sk_fullsock (sk))return NULL ; - } }return sk;net/core/filter.c bpf_sk_lookup(struct sk_buff *skb, struct bpf_sock_tuple *tuple, u32 len, · line 6558
Quoted 9 lines
net/core/filter.c +1
-2@@ -6551,13 +6552,14 @@ bpf_sk_lookup(struct sk_buff *skb, struct bpf_sock_tuple *tuple, u32 len, {struct sock *sk =bpf_skc_lookup (skb, tuple, len, proto, netns_id, flags); +struct sock *sk1 = sk;if (sk) { sk =sk_to_full_sk (sk); -if (!sk_fullsock (sk)) { -sock_gen_put (sk);Ditto.
Quoted 7 lines
net/core/filter.c +3
-1+if (!sk_fullsock (sk1)) +sock_gen_put (sk1); +if (!sk_fullsock (sk))return NULL ; - } }
Quoted 4 lines
Review bpf_sk_lookup(struct sk_buff *skb, struct bpf_sock_tuple *tuple, u32 len, · line 6565
Quoted 1 line
(quoted diff) +0
-0return sk;Thanks!
Antoine
Thanks Antoine, ack I'll do that and resubmit.