fix some leaks in espintcp
Subsystems (3)
cover fix some leaks in espintcp
kmemleak spotted a few leaks that have been here since the beginning.
Sabrina Dubroca (2):
espintcp: fix skb leaks
espintcp: remove encap socket caching to avoid reference leak
include/net/xfrm.h | 1 - net/ipv4/esp4.c | 53 ++++++------------------------------------- net/ipv6/esp6.c | 53 ++++++------------------------------------- net/xfrm/espintcp.c | 4 +++- net/xfrm/xfrm_state.c | 3 --- 5 files changed, 17 insertions(+), 97 deletions(-)
Signature · 1 line
2.49.0
PATCH 1/2 espintcp: fix skb leaks
A few error paths are missing a kfree_skb.
--- net/ipv4/esp4.c | 4 +++- net/ipv6/esp6.c | 4 +++- net/xfrm/espintcp.c | 4 +++- 3 files changed, 9 insertions(+), 3 deletions(-)
net/ipv4/esp4.c +3 -1
@@ -199,8 +199,10 @@ static int esp_output_tcp_finish(struct xfrm_state *x, struct sk_buff *skb) sk =esp_find_tcp_sk (x); err =PTR_ERR_OR_ZERO (sk); -if (err) +if (err) { +kfree_skb (skb); goto out; + }bh_lock_sock (sk);if (sock_owned_by_user (sk))
net/ipv6/esp6.c +3 -1
@@ -216,8 +216,10 @@ static int esp_output_tcp_finish(struct xfrm_state *x, struct sk_buff *skb) sk =esp6_find_tcp_sk (x); err =PTR_ERR_OR_ZERO (sk); -if (err) +if (err) { +kfree_skb (skb); goto out; + }bh_lock_sock (sk);if (sock_owned_by_user (sk))
net/xfrm/espintcp.c +3 -1
@@ -171,8 +171,10 @@ int espintcp_queue_out(struct sock *sk, struct sk_buff *skb)struct espintcp_ctx *ctx =espintcp_getctx (sk);if (skb_queue_len (&ctx->out_queue) >= -READ_ONCE (net_hotdata.max_backlog)) +READ_ONCE (net_hotdata.max_backlog)) { +kfree_skb (skb);return -ENOBUFS ; + }__skb_queue_tail (&ctx->out_queue, skb);
Signature · 1 line
2.49.0
PATCH 2/2 espintcp: remove encap socket caching to avoid reference leak
In reply to Sabrina DubrocaThe current scheme for caching the encap socket can lead to reference leaks when we try to delete the netns.
The reference chain is: xfrm_state -> enacp_sk -> netns
Since the encap socket is a userspace socket, it holds a reference on the netns. If we delete the espintcp state (through flush or individual delete) before removing the netns, the reference on the socket is dropped and the netns is correctly deleted. Otherwise, the netns may not be reachable anymore (if all processes within the ns have terminated), so we cannot delete the xfrm state to drop its reference on the socket.
This patch results in a small (~2% in my tests) performance regression.
A GC-type mechanism could be added for the socket cache, to clear references if the state hasn't been used "recently", but it's a lot more complex than just not caching the socket.
--- include/net/xfrm.h | 1 - net/ipv4/esp4.c | 49 ++++--------------------------------------- net/ipv6/esp6.c | 49 ++++--------------------------------------- net/xfrm/xfrm_state.c | 3 --- 4 files changed, 8 insertions(+), 94 deletions(-)
include/net/xfrm.h +0 -1
@@ -236,7 +236,6 @@ struct xfrm_state {/* Data for encapsulator */ struct xfrm_encap_tmpl *encap; -struct sock __rcu *encap_sk;/* NAT keepalive */ u32 nat_keepalive_interval;/* seconds */
net/ipv4/esp4.c +4 -45
83 lines not on this page · raw message
net/ipv6/esp6.c +4 -45
83 lines not on this page · raw message
net/xfrm/xfrm_state.c +0 -3
@@ -840,9 +840,6 @@ int __xfrm_state_delete(struct xfrm_state *x)xfrm_nat_keepalive_state_updated (x);spin_unlock (&net->xfrm.xfrm_state_lock); -if (x->encap_sk) -sock_put (rcu_dereference_raw (x->encap_sk)); -xfrm_dev_state_delete (x);/* All xfrm_state objects are created by xfrm_state_alloc.
Signature · 1 line
2.49.0
Re: [PATCH ipsec 1/2] espintcp: fix skb leaks
In reply to Sabrina DubrocaQuoted 4 lines · Sabrina Dubroca
A few error paths are missing a
kfree_skb.Fixes e27cca96cd68 ("xfrm: add espintcp (RFC 8229)")Signed-off-by Sabrina Dubroca <[email protected]>
Re: [PATCH ipsec 2/2] espintcp: remove encap socket caching to avoid reference leak
In reply to Sabrina DubrocaQuoted 13 lines · Sabrina Dubroca
The current scheme for caching the encap socket can lead to reference leaks when we try to delete the netns.
The reference chain is:
xfrm_state->enacp_sk-> netnsSince the encap socket is a userspace socket, it holds a reference on the netns. If we delete the espintcp state (through flush or individual delete) before removing the netns, the reference on the socket is dropped and the netns is correctly deleted. Otherwise, the netns may not be reachable anymore (if all processes within the ns have terminated), so we cannot delete the xfrm state to drop its reference on the socket.
Quoted 6 lines
This patch results in a small (~2% in my tests) performance regression.
A GC-type mechanism could be added for the socket cache, to clear references if the state hasn't been used "recently", but it's a lot more complex than just not caching the socket.
Less is more :)
Quoted 2 lines
Fixes e27cca96cd68 ("xfrm: add espintcp (RFC 8229)")Signed-off-by Sabrina Dubroca <[email protected]>
Steffen Klassert
maintainerQuoted 5 lines · Sabrina Dubroca
kmemleak spotted a few leaks that have been here since the beginning.
Sabrina Dubroca (2):
espintcp: fix skb leaks
espintcp: remove encap socket caching to avoid reference leak
Series applied, thanks Sabrina!