From: Eric Dumazet <eduma...@google.com> Kirill found that recently added synchronize_rcu() call in ip6mr_sk_done() was slowing down netns dismantle and posted a patch to use it only if the socket was found.
I instead suggested to get rid of this call, and use instead SOCK_RCU_FREE We might later change IPv4 side to use the same technique and unify both stacks. IPv4 does not use synchronize_rcu() but has a call_rcu() that could be replaced by SOCK_RCU_FREE. Tested: time for i in {1..1000}; do unshare -n /bin/false;done Before : real 7m18.911s After : real 10.187s Fixes: 8571ab479a6e ("ip6mr: Make mroute_sk rcu-based") Signed-off-by: Eric Dumazet <eduma...@google.com> Reported-by: Kirill Tkhai <ktk...@virtuozzo.com> Cc: Yuval Mintz <yuv...@mellanox.com> --- net/ipv6/ip6mr.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/net/ipv6/ip6mr.c b/net/ipv6/ip6mr.c index 2a38f9de45d399fafc9f4bcc662b44be17279e51..7345bd6c4b7dda39c0d73d542e9ca9a5366542ff 100644 --- a/net/ipv6/ip6mr.c +++ b/net/ipv6/ip6mr.c @@ -1443,6 +1443,7 @@ static int ip6mr_sk_init(struct mr_table *mrt, struct sock *sk) err = -EADDRINUSE; } else { rcu_assign_pointer(mrt->mroute_sk, sk); + sock_set_flag(sk, SOCK_RCU_FREE); net->ipv6.devconf_all->mc_forwarding++; } write_unlock_bh(&mrt_lock); @@ -1472,6 +1473,10 @@ int ip6mr_sk_done(struct sock *sk) if (sk == rtnl_dereference(mrt->mroute_sk)) { write_lock_bh(&mrt_lock); RCU_INIT_POINTER(mrt->mroute_sk, NULL); + /* Note that mroute_sk had SOCK_RCU_FREE set, + * so the RCU grace period before sk freeing + * is guaranteed by sk_destruct() + */ net->ipv6.devconf_all->mc_forwarding--; write_unlock_bh(&mrt_lock); inet6_netconf_notify_devconf(net, RTM_NEWNETCONF, @@ -1485,7 +1490,6 @@ int ip6mr_sk_done(struct sock *sk) } } rtnl_unlock(); - synchronize_rcu(); return err; }