aboutsummaryrefslogtreecommitdiffstats
path: root/net
diff options
context:
space:
mode:
authorStephen Hemminger <shemminger@vyatta.com>2011-01-11 17:54:42 -0500
committerPablo Neira Ayuso <pablo@netfilter.org>2011-01-11 17:54:42 -0500
commit13ee6ac579574a2a95e982b19920fd2495dce8cd (patch)
treeb1cfcae34d29ce1494254d814fc92e23a426ef71 /net
parent83723d60717f8da0f53f91cf42a845ed56c09662 (diff)
netfilter: fix race in conntrack between dump_table and destroy
The netlink interface to dump the connection tracking table has a race when entries are deleted at the same time. A customer reported a crash and the backtrace showed thatctnetlink_dump_table was running while a conntrack entry was being destroyed. (see https://bugzilla.vyatta.com/show_bug.cgi?id=6402). According to RCU documentation, when using hlist_nulls the reader must handle the case of seeing a deleted entry and not proceed further down the linked list. The old code would continue which caused the scan to walk into the free list. This patch uses locking (rather than RCU) for this operation which is guaranteed safe, and no longer requires getting reference while doing dump operation. Signed-off-by: Stephen Hemminger <shemminger@vyatta.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Diffstat (limited to 'net')
-rw-r--r--net/netfilter/nf_conntrack_netlink.c14
1 files changed, 5 insertions, 9 deletions
diff --git a/net/netfilter/nf_conntrack_netlink.c b/net/netfilter/nf_conntrack_netlink.c
index 746140264b2d..5cb8d3027b18 100644
--- a/net/netfilter/nf_conntrack_netlink.c
+++ b/net/netfilter/nf_conntrack_netlink.c
@@ -645,25 +645,23 @@ ctnetlink_dump_table(struct sk_buff *skb, struct netlink_callback *cb)
645 struct nfgenmsg *nfmsg = nlmsg_data(cb->nlh); 645 struct nfgenmsg *nfmsg = nlmsg_data(cb->nlh);
646 u_int8_t l3proto = nfmsg->nfgen_family; 646 u_int8_t l3proto = nfmsg->nfgen_family;
647 647
648 rcu_read_lock(); 648 spin_lock_bh(&nf_conntrack_lock);
649 last = (struct nf_conn *)cb->args[1]; 649 last = (struct nf_conn *)cb->args[1];
650 for (; cb->args[0] < net->ct.htable_size; cb->args[0]++) { 650 for (; cb->args[0] < net->ct.htable_size; cb->args[0]++) {
651restart: 651restart:
652 hlist_nulls_for_each_entry_rcu(h, n, &net->ct.hash[cb->args[0]], 652 hlist_nulls_for_each_entry(h, n, &net->ct.hash[cb->args[0]],
653 hnnode) { 653 hnnode) {
654 if (NF_CT_DIRECTION(h) != IP_CT_DIR_ORIGINAL) 654 if (NF_CT_DIRECTION(h) != IP_CT_DIR_ORIGINAL)
655 continue; 655 continue;
656 ct = nf_ct_tuplehash_to_ctrack(h); 656 ct = nf_ct_tuplehash_to_ctrack(h);
657 if (!atomic_inc_not_zero(&ct->ct_general.use))
658 continue;
659 /* Dump entries of a given L3 protocol number. 657 /* Dump entries of a given L3 protocol number.
660 * If it is not specified, ie. l3proto == 0, 658 * If it is not specified, ie. l3proto == 0,
661 * then dump everything. */ 659 * then dump everything. */
662 if (l3proto && nf_ct_l3num(ct) != l3proto) 660 if (l3proto && nf_ct_l3num(ct) != l3proto)
663 goto releasect; 661 continue;
664 if (cb->args[1]) { 662 if (cb->args[1]) {
665 if (ct != last) 663 if (ct != last)
666 goto releasect; 664 continue;
667 cb->args[1] = 0; 665 cb->args[1] = 0;
668 } 666 }
669 if (ctnetlink_fill_info(skb, NETLINK_CB(cb->skb).pid, 667 if (ctnetlink_fill_info(skb, NETLINK_CB(cb->skb).pid,
@@ -681,8 +679,6 @@ restart:
681 if (acct) 679 if (acct)
682 memset(acct, 0, sizeof(struct nf_conn_counter[IP_CT_DIR_MAX])); 680 memset(acct, 0, sizeof(struct nf_conn_counter[IP_CT_DIR_MAX]));
683 } 681 }
684releasect:
685 nf_ct_put(ct);
686 } 682 }
687 if (cb->args[1]) { 683 if (cb->args[1]) {
688 cb->args[1] = 0; 684 cb->args[1] = 0;
@@ -690,7 +686,7 @@ releasect:
690 } 686 }
691 } 687 }
692out: 688out:
693 rcu_read_unlock(); 689 spin_unlock_bh(&nf_conntrack_lock);
694 if (last) 690 if (last)
695 nf_ct_put(last); 691 nf_ct_put(last);
696 692