net: allow unbound socket for packets in VRF when tcp_l3mdev_accept set
Subsystems (2)
PATCH 1/1 net: allow unbound socket for packets in VRF when tcp_l3mdev_accept set
The commit 3c82a21f4320 ("net: allow binding socket in a VRF when there's an unbound socket") changed the inet socket lookup to avoid packets in a VRF from matching an unbound socket. This is to ensure the necessary isolation between the default and other VRFs for routing and forwarding. VRF-unaware processes running in the default VRF cannot access another VRF and have to be run with 'ip vrf exec <vrf>'. This is to be expected with tcp_l3mdev_accept disabled, but could be reallowed when this sysctl option is enabled. So instead of directly checking dif and sdif in inet[6]_match, here call inet_sk_bound_dev_eq(). This allows a match on unbound socket for non-zero sdif i.e. for packets in a VRF, if tcp_l3mdev_accept is enabled.
---
Nettest results for VRF testing remain unchanged:
$ diff nettest-baseline-502c6f8cedcc.txt nettest-fix.txt $ tail -3 nettest-fix.txt Tests passed: 869 Tests failed: 5
Disclaimer: While I do not think that there should be any noticeable socket throughput degradation due to these changes, I am unable to carry out any performance tests for this, nor provide any follow-up support if there is any such degradation.
--- include/net/inet6_hashtables.h | 7 +++---- include/net/inet_hashtables.h | 19 +++---------------- include/net/inet_sock.h | 11 +++++++++++ 3 files changed, 17 insertions(+), 20 deletions(-)
include/net/inet6_hashtables.h +3 -4
@@ -110,8 +110,6 @@ static inline bool inet6_match(struct net *net, const struct sock *sk,const __portpair ports, const int dif, const int sdif) { - int bound_dev_if; - if (!net_eq(sock_net(sk), net) || sk->sk_family != AF_INET6 || sk->sk_portpair != ports ||
@@ -119,8 +117,9 @@ static inline bool inet6_match(struct net *net, const struct sock *sk, !ipv6_addr_equal(&sk->sk_v6_rcv_saddr, daddr))return false; - bound_dev_if =READ_ONCE (sk->sk_bound_dev_if); - return bound_dev_if == dif || bound_dev_if == sdif; +/* READ_ONCE() paired with WRITE_ONCE() in sock_bindtoindex_locked() */ +return inet_sk_bound_dev_eq (net,READ_ONCE (sk->sk_bound_dev_if), dif, + sdif); }#endif /* IS_ENABLED(CONFIG_IPV6) */
include/net/inet_hashtables.h +3 -16
@@ -175,17 +175,6 @@ static inline void inet_ehash_locks_free(struct inet_hashinfo *hashinfo) hashinfo->ehash_locks =NULL ; } -static inline bool inet_sk_bound_dev_eq (struct net *net,int bound_dev_if, -int dif,int sdif) -{ -#if IS_ENABLED (CONFIG_NET_L3_MASTER_DEV ) -return inet_bound_dev_eq (!!READ_ONCE (net->ipv4.sysctl_tcp_l3mdev_accept), - bound_dev_if, dif, sdif); -#else -return inet_bound_dev_eq (true, bound_dev_if, dif, sdif); -#endif -} -struct inet_bind_bucket * inet_bind_bucket_create(struct kmem_cache *cachep, struct net *net, struct inet_bind_hashbucket *head,
@@ -271,16 +260,14 @@ static inline bool inet_match(struct net *net, const struct sock *sk, const __addrpair cookie, const __portpair ports, int dif, int sdif) { - int bound_dev_if; - if (!net_eq(sock_net(sk), net) || sk->sk_portpair != ports || sk->sk_addrpair != cookie) return false; -/* Paired with WRITE_ONCE() from sock_bindtoindex_locked() */ - bound_dev_if =READ_ONCE (sk->sk_bound_dev_if); -return bound_dev_if == dif || bound_dev_if == sdif; +/* READ_ONCE() paired with WRITE_ONCE() in sock_bindtoindex_locked() */ +return inet_sk_bound_dev_eq (net,READ_ONCE (sk->sk_bound_dev_if), dif, + sdif); }/* Sockets in TCP_CLOSE state are _always_ taken out of the hash, so we need
include/net/inet_sock.h +11 -0
@@ -149,6 +149,17 @@ static inline bool inet_bound_dev_eq(bool l3mdev_accept, int bound_dev_if,return bound_dev_if == dif || bound_dev_if == sdif; } +static inline bool inet_sk_bound_dev_eq (struct net *net,int bound_dev_if, +int dif,int sdif) +{ +#if IS_ENABLED (CONFIG_NET_L3_MASTER_DEV ) +return inet_bound_dev_eq (!!READ_ONCE (net->ipv4.sysctl_tcp_l3mdev_accept), + bound_dev_if, dif, sdif); +#else +return inet_bound_dev_eq (true, bound_dev_if, dif, sdif); +#endif +} +struct inet_cork {unsigned int flags;__be32 addr;
Signature · 1 line
2.20.1
[ cc Andy who reported a similar issue ]
Quoted 124 lines · Mike Manning
The commit
3c82a21f4320("net: allow binding socket in a VRF when there's an unbound socket") changed the inet socket lookup to avoid packets in a VRF from matching an unbound socket. This is to ensure the necessary isolation between the default and other VRFs for routing and forwarding. VRF-unaware processes running in the default VRF cannot access another VRF and have to be run with 'ip vrf exec <vrf>'. This is to be expected withtcp_l3mdev_acceptdisabled, but could be reallowed when this sysctl option is enabled. So instead of directly checking dif and sdif in inet[6]_match, here callinet_sk_bound_dev_eq(). This allows a match on unbound socket for non-zero sdif i.e. for packets in a VRF, iftcp_l3mdev_acceptis enabled.Fixes 3c82a21f4320 ("net: allow binding socket in a VRF when there's an unbound socket")Signed-off-by Mike Manning <[email protected]>---
Nettest results for VRF testing remain unchanged:
$ diff nettest-baseline-502c6f8cedcc.txt nettest-fix.txt $ tail -3 nettest-fix.txt Tests passed: 869 Tests failed: 5Disclaimer: While I do not think that there should be any noticeable socket throughput degradation due to these changes, I am unable to carry out any performance tests for this, nor provide any follow-up support if there is any such degradation.
--- include/net/inet6_hashtables.h | 7 +++---- include/net/inet_hashtables.h | 19 +++---------------- include/net/inet_sock.h | 11 +++++++++++ 3 files changed, 17 insertions(+), 20 deletions(-)include/net/inet6_hashtables.h +3
-4@@ -110,8 +110,6 @@ static inline bool inet6_match(struct net *net, const struct sock *sk,const __portpair ports, const int dif, const int sdif) { - int bound_dev_if; - if (!net_eq(sock_net(sk), net) || sk->sk_family != AF_INET6 || sk->sk_portpair != ports ||@@ -119,8 +117,9 @@ static inline bool inet6_match(struct net *net, const struct sock *sk, !ipv6_addr_equal(&sk->sk_v6_rcv_saddr, daddr))return false; - bound_dev_if =READ_ONCE (sk->sk_bound_dev_if); - return bound_dev_if == dif || bound_dev_if == sdif; +/* READ_ONCE() paired with WRITE_ONCE() in sock_bindtoindex_locked() */ +return inet_sk_bound_dev_eq (net,READ_ONCE (sk->sk_bound_dev_if), dif, + sdif); }#endif /* IS_ENABLED(CONFIG_IPV6) */ include/net/inet_hashtables.h +3
-16@@ -175,17 +175,6 @@ static inline void inet_ehash_locks_free(struct inet_hashinfo *hashinfo) hashinfo->ehash_locks =NULL ; } -static inline bool inet_sk_bound_dev_eq (struct net *net,int bound_dev_if, -int dif,int sdif) -{ -#if IS_ENABLED (CONFIG_NET_L3_MASTER_DEV ) -return inet_bound_dev_eq (!!READ_ONCE (net->ipv4.sysctl_tcp_l3mdev_accept), - bound_dev_if, dif, sdif); -#else -return inet_bound_dev_eq (true, bound_dev_if, dif, sdif); -#endif -} -struct inet_bind_bucket * inet_bind_bucket_create(struct kmem_cache *cachep, struct net *net, struct inet_bind_hashbucket *head,@@ -271,16 +260,14 @@ static inline bool inet_match(struct net *net, const struct sock *sk, const __addrpair cookie, const __portpair ports, int dif, int sdif) { - int bound_dev_if; - if (!net_eq(sock_net(sk), net) || sk->sk_portpair != ports || sk->sk_addrpair != cookie) return false; -/* Paired with WRITE_ONCE() from sock_bindtoindex_locked() */ - bound_dev_if =READ_ONCE (sk->sk_bound_dev_if); -return bound_dev_if == dif || bound_dev_if == sdif; +/* READ_ONCE() paired with WRITE_ONCE() in sock_bindtoindex_locked() */ +return inet_sk_bound_dev_eq (net,READ_ONCE (sk->sk_bound_dev_if), dif, + sdif); }/* Sockets in TCP_CLOSE state are _always_ taken out of the hash, so we need include/net/inet_sock.h +11
-0@@ -149,6 +149,17 @@ static inline bool inet_bound_dev_eq(bool l3mdev_accept, int bound_dev_if,return bound_dev_if == dif || bound_dev_if == sdif; } +static inline bool inet_sk_bound_dev_eq (struct net *net,int bound_dev_if, +int dif,int sdif) +{ +#if IS_ENABLED (CONFIG_NET_L3_MASTER_DEV ) +return inet_bound_dev_eq (!!READ_ONCE (net->ipv4.sysctl_tcp_l3mdev_accept), + bound_dev_if, dif, sdif); +#else +return inet_bound_dev_eq (true, bound_dev_if, dif, sdif); +#endif +} +struct inet_cork {unsigned int flags;__be32 addr;
Quoted 29 lines · Mike Manning
The commit
3c82a21f4320("net: allow binding socket in a VRF when there's an unbound socket") changed the inet socket lookup to avoid packets in a VRF from matching an unbound socket. This is to ensure the necessary isolation between the default and other VRFs for routing and forwarding. VRF-unaware processes running in the default VRF cannot access another VRF and have to be run with 'ip vrf exec <vrf>'. This is to be expected withtcp_l3mdev_acceptdisabled, but could be reallowed when this sysctl option is enabled. So instead of directly checking dif and sdif in inet[6]_match, here callinet_sk_bound_dev_eq(). This allows a match on unbound socket for non-zero sdif i.e. for packets in a VRF, iftcp_l3mdev_acceptis enabled.Fixes 3c82a21f4320 ("net: allow binding socket in a VRF when there's an unbound socket")Signed-off-by Mike Manning <[email protected]>---
Nettest results for VRF testing remain unchanged:
$ diff nettest-baseline-502c6f8cedcc.txt nettest-fix.txt $ tail -3 nettest-fix.txt Tests passed: 869 Tests failed: 5Disclaimer: While I do not think that there should be any noticeable socket throughput degradation due to these changes, I am unable to carry out any performance tests for this, nor provide any follow-up support if there is any such degradation.
Quoted 6 lines
--- include/net/inet6_hashtables.h | 7 +++---- include/net/inet_hashtables.h | 19 +++---------------- include/net/inet_sock.h | 11 +++++++++++ 3 files changed, 17 insertions(+), 20 deletions(-)
This patch was applied to netdev/net.git (master) by David S. Miller <[email protected]>:
Hello:
This patch was applied to netdev/net.git (master)
by David S. Miller <[email protected]>:
Quoted 12 lines · On Mon, 25 Jul 2022 19:14:42 +0100 you wrote:
The commit
3c82a21f4320("net: allow binding socket in a VRF when there's an unbound socket") changed the inet socket lookup to avoid packets in a VRF from matching an unbound socket. This is to ensure the necessary isolation between the default and other VRFs for routing and forwarding. VRF-unaware processes running in the default VRF cannot access another VRF and have to be run with 'ip vrf exec <vrf>'. This is to be expected withtcp_l3mdev_acceptdisabled, but could be reallowed when this sysctl option is enabled. So instead of directly checking dif and sdif in inet[6]_match, here callinet_sk_bound_dev_eq(). This allows a match on unbound socket for non-zero sdif i.e. for packets in a VRF, iftcp_l3mdev_acceptis enabled.
Quoted 1 line
[...]
Here is the summary with links:
- net: allow unbound socket for packets in VRF when tcp_l3mdev_accept set
https://git.kernel.org/netdev/net/c/944fd1aeacb6
You are awesome, thank you!
Signature · 2 lines
Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html