aboutsummaryrefslogtreecommitdiffstats
path: root/net/ipv4
diff options
context:
space:
mode:
authorHarald Welte <laforge@netfilter.org>2005-09-23 02:46:57 -0400
committerDavid S. Miller <davem@davemloft.net>2005-09-23 02:46:57 -0400
commit1dfbab59498d6f227c91988bab6c71af049a5333 (patch)
tree6b20409a232ebe8c37f16d06b3fbcde6bec8f328 /net/ipv4
parenta82b748930fce0dab22c64075c38c830ae116904 (diff)
[NETFILTER] Fix conntrack event cache deadlock/oops
This patch fixes a number of bugs. It cannot be reasonably split up in multiple fixes, since all bugs interact with each other and affect the same function: Bug #1: The event cache code cannot be called while a lock is held. Therefore, the call to ip_conntrack_event_cache() within ip_ct_refresh_acct() needs to be moved outside of the locked section. This fixes a number of 2.6.14-rcX oops and deadlock reports. Bug #2: We used to call ct_add_counters() for unconfirmed connections without holding a lock. Since the add operations are not atomic, we could race with another CPU. Bug #3: ip_ct_refresh_acct() lost REFRESH events in some cases where refresh (and the corresponding event) are desired, but no accounting shall be performed. Both, evenst and accounting implicitly depended on the skb parameter bein non-null. We now re-introduce a non-accounting "ip_ct_refresh()" variant to explicitly state the desired behaviour. Signed-off-by: Harald Welte <laforge@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/ipv4')
-rw-r--r--net/ipv4/netfilter/ip_conntrack_amanda.c2
-rw-r--r--net/ipv4/netfilter/ip_conntrack_core.c49
-rw-r--r--net/ipv4/netfilter/ip_conntrack_helper_pptp.c1
-rw-r--r--net/ipv4/netfilter/ip_conntrack_netbios_ns.c2
-rw-r--r--net/ipv4/netfilter/ip_conntrack_standalone.c2
5 files changed, 28 insertions, 28 deletions
diff --git a/net/ipv4/netfilter/ip_conntrack_amanda.c b/net/ipv4/netfilter/ip_conntrack_amanda.c
index dc20881004bc..fa3f914117ec 100644
--- a/net/ipv4/netfilter/ip_conntrack_amanda.c
+++ b/net/ipv4/netfilter/ip_conntrack_amanda.c
@@ -65,7 +65,7 @@ static int help(struct sk_buff **pskb,
65 65
66 /* increase the UDP timeout of the master connection as replies from 66 /* increase the UDP timeout of the master connection as replies from
67 * Amanda clients to the server can be quite delayed */ 67 * Amanda clients to the server can be quite delayed */
68 ip_ct_refresh_acct(ct, ctinfo, NULL, master_timeout * HZ); 68 ip_ct_refresh(ct, *pskb, master_timeout * HZ);
69 69
70 /* No data? */ 70 /* No data? */
71 dataoff = (*pskb)->nh.iph->ihl*4 + sizeof(struct udphdr); 71 dataoff = (*pskb)->nh.iph->ihl*4 + sizeof(struct udphdr);
diff --git a/net/ipv4/netfilter/ip_conntrack_core.c b/net/ipv4/netfilter/ip_conntrack_core.c
index c1f82e0c81cf..ea65dd3e517a 100644
--- a/net/ipv4/netfilter/ip_conntrack_core.c
+++ b/net/ipv4/netfilter/ip_conntrack_core.c
@@ -1112,45 +1112,46 @@ void ip_conntrack_helper_unregister(struct ip_conntrack_helper *me)
1112 synchronize_net(); 1112 synchronize_net();
1113} 1113}
1114 1114
1115static inline void ct_add_counters(struct ip_conntrack *ct, 1115/* Refresh conntrack for this many jiffies and do accounting if do_acct is 1 */
1116 enum ip_conntrack_info ctinfo, 1116void __ip_ct_refresh_acct(struct ip_conntrack *ct,
1117 const struct sk_buff *skb)
1118{
1119#ifdef CONFIG_IP_NF_CT_ACCT
1120 if (skb) {
1121 ct->counters[CTINFO2DIR(ctinfo)].packets++;
1122 ct->counters[CTINFO2DIR(ctinfo)].bytes +=
1123 ntohs(skb->nh.iph->tot_len);
1124 }
1125#endif
1126}
1127
1128/* Refresh conntrack for this many jiffies and do accounting (if skb != NULL) */
1129void ip_ct_refresh_acct(struct ip_conntrack *ct,
1130 enum ip_conntrack_info ctinfo, 1117 enum ip_conntrack_info ctinfo,
1131 const struct sk_buff *skb, 1118 const struct sk_buff *skb,
1132 unsigned long extra_jiffies) 1119 unsigned long extra_jiffies,
1120 int do_acct)
1133{ 1121{
1122 int do_event = 0;
1123
1134 IP_NF_ASSERT(ct->timeout.data == (unsigned long)ct); 1124 IP_NF_ASSERT(ct->timeout.data == (unsigned long)ct);
1125 IP_NF_ASSERT(skb);
1126
1127 write_lock_bh(&ip_conntrack_lock);
1135 1128
1136 /* If not in hash table, timer will not be active yet */ 1129 /* If not in hash table, timer will not be active yet */
1137 if (!is_confirmed(ct)) { 1130 if (!is_confirmed(ct)) {
1138 ct->timeout.expires = extra_jiffies; 1131 ct->timeout.expires = extra_jiffies;
1139 ct_add_counters(ct, ctinfo, skb); 1132 do_event = 1;
1140 } else { 1133 } else {
1141 write_lock_bh(&ip_conntrack_lock);
1142 /* Need del_timer for race avoidance (may already be dying). */ 1134 /* Need del_timer for race avoidance (may already be dying). */
1143 if (del_timer(&ct->timeout)) { 1135 if (del_timer(&ct->timeout)) {
1144 ct->timeout.expires = jiffies + extra_jiffies; 1136 ct->timeout.expires = jiffies + extra_jiffies;
1145 add_timer(&ct->timeout); 1137 add_timer(&ct->timeout);
1146 /* FIXME: We loose some REFRESH events if this function 1138 do_event = 1;
1147 * is called without an skb. I'll fix this later -HW */
1148 if (skb)
1149 ip_conntrack_event_cache(IPCT_REFRESH, skb);
1150 } 1139 }
1151 ct_add_counters(ct, ctinfo, skb);
1152 write_unlock_bh(&ip_conntrack_lock);
1153 } 1140 }
1141
1142#ifdef CONFIG_IP_NF_CT_ACCT
1143 if (do_acct) {
1144 ct->counters[CTINFO2DIR(ctinfo)].packets++;
1145 ct->counters[CTINFO2DIR(ctinfo)].bytes +=
1146 ntohs(skb->nh.iph->tot_len);
1147 }
1148#endif
1149
1150 write_unlock_bh(&ip_conntrack_lock);
1151
1152 /* must be unlocked when calling event cache */
1153 if (do_event)
1154 ip_conntrack_event_cache(IPCT_REFRESH, skb);
1154} 1155}
1155 1156
1156#if defined(CONFIG_IP_NF_CONNTRACK_NETLINK) || \ 1157#if defined(CONFIG_IP_NF_CONNTRACK_NETLINK) || \
diff --git a/net/ipv4/netfilter/ip_conntrack_helper_pptp.c b/net/ipv4/netfilter/ip_conntrack_helper_pptp.c
index 8236ee0fb090..926a6684643d 100644
--- a/net/ipv4/netfilter/ip_conntrack_helper_pptp.c
+++ b/net/ipv4/netfilter/ip_conntrack_helper_pptp.c
@@ -172,7 +172,6 @@ static int destroy_sibling_or_exp(const struct ip_conntrack_tuple *t)
172 DEBUGP("setting timeout of conntrack %p to 0\n", sibling); 172 DEBUGP("setting timeout of conntrack %p to 0\n", sibling);
173 sibling->proto.gre.timeout = 0; 173 sibling->proto.gre.timeout = 0;
174 sibling->proto.gre.stream_timeout = 0; 174 sibling->proto.gre.stream_timeout = 0;
175 /* refresh_acct will not modify counters if skb == NULL */
176 if (del_timer(&sibling->timeout)) 175 if (del_timer(&sibling->timeout))
177 sibling->timeout.function((unsigned long)sibling); 176 sibling->timeout.function((unsigned long)sibling);
178 ip_conntrack_put(sibling); 177 ip_conntrack_put(sibling);
diff --git a/net/ipv4/netfilter/ip_conntrack_netbios_ns.c b/net/ipv4/netfilter/ip_conntrack_netbios_ns.c
index 71ef19d126d0..577bac22dcc6 100644
--- a/net/ipv4/netfilter/ip_conntrack_netbios_ns.c
+++ b/net/ipv4/netfilter/ip_conntrack_netbios_ns.c
@@ -91,7 +91,7 @@ static int help(struct sk_buff **pskb,
91 ip_conntrack_expect_related(exp); 91 ip_conntrack_expect_related(exp);
92 ip_conntrack_expect_put(exp); 92 ip_conntrack_expect_put(exp);
93 93
94 ip_ct_refresh_acct(ct, ctinfo, NULL, timeout * HZ); 94 ip_ct_refresh(ct, *pskb, timeout * HZ);
95out: 95out:
96 return NF_ACCEPT; 96 return NF_ACCEPT;
97} 97}
diff --git a/net/ipv4/netfilter/ip_conntrack_standalone.c b/net/ipv4/netfilter/ip_conntrack_standalone.c
index d3c7808010ec..dd476b191f4b 100644
--- a/net/ipv4/netfilter/ip_conntrack_standalone.c
+++ b/net/ipv4/netfilter/ip_conntrack_standalone.c
@@ -989,7 +989,7 @@ EXPORT_SYMBOL(need_ip_conntrack);
989EXPORT_SYMBOL(ip_conntrack_helper_register); 989EXPORT_SYMBOL(ip_conntrack_helper_register);
990EXPORT_SYMBOL(ip_conntrack_helper_unregister); 990EXPORT_SYMBOL(ip_conntrack_helper_unregister);
991EXPORT_SYMBOL(ip_ct_iterate_cleanup); 991EXPORT_SYMBOL(ip_ct_iterate_cleanup);
992EXPORT_SYMBOL(ip_ct_refresh_acct); 992EXPORT_SYMBOL(__ip_ct_refresh_acct);
993 993
994EXPORT_SYMBOL(ip_conntrack_expect_alloc); 994EXPORT_SYMBOL(ip_conntrack_expect_alloc);
995EXPORT_SYMBOL(ip_conntrack_expect_put); 995EXPORT_SYMBOL(ip_conntrack_expect_put);