net: dont call jump_label_dec from irq context
Eric Dumazet [Mon, 28 Nov 2011 11:16:50 +0000 (11:16 +0000)]
Igor Maravic reported an error caused by jump_label_dec() being called
from IRQ context :

 BUG: sleeping function called from invalid context at kernel/mutex.c:271
 in_atomic(): 1, irqs_disabled(): 0, pid: 0, name: swapper
 1 lock held by swapper/0:
  #0:  (&n->timer){+.-...}, at: [<ffffffff8107ce90>] call_timer_fn+0x0/0x340
 Pid: 0, comm: swapper Not tainted 3.2.0-rc2-net-next-mpls+ #1
Call Trace:
 <IRQ>  [<ffffffff8104f417>] __might_sleep+0x137/0x1f0
 [<ffffffff816b9a2f>] mutex_lock_nested+0x2f/0x370
 [<ffffffff810a89fd>] ? trace_hardirqs_off+0xd/0x10
 [<ffffffff8109a37f>] ? local_clock+0x6f/0x80
 [<ffffffff810a90a5>] ? lock_release_holdtime.part.22+0x15/0x1a0
 [<ffffffff81557929>] ? sock_def_write_space+0x59/0x160
 [<ffffffff815e936e>] ? arp_error_report+0x3e/0x90
 [<ffffffff810969cd>] atomic_dec_and_mutex_lock+0x5d/0x80
 [<ffffffff8112fc1d>] jump_label_dec+0x1d/0x50
 [<ffffffff81566525>] net_disable_timestamp+0x15/0x20
 [<ffffffff81557a75>] sock_disable_timestamp+0x45/0x50
 [<ffffffff81557b00>] __sk_free+0x80/0x200
 [<ffffffff815578d0>] ? sk_send_sigurg+0x70/0x70
 [<ffffffff815e936e>] ? arp_error_report+0x3e/0x90
 [<ffffffff81557cba>] sock_wfree+0x3a/0x70
 [<ffffffff8155c2b0>] skb_release_head_state+0x70/0x120
 [<ffffffff8155c0b6>] __kfree_skb+0x16/0x30
 [<ffffffff8155c119>] kfree_skb+0x49/0x170
 [<ffffffff815e936e>] arp_error_report+0x3e/0x90
 [<ffffffff81575bd9>] neigh_invalidate+0x89/0xc0
 [<ffffffff81578dbe>] neigh_timer_handler+0x9e/0x2a0
 [<ffffffff81578d20>] ? neigh_update+0x640/0x640
 [<ffffffff81073558>] __do_softirq+0xc8/0x3a0

Since jump_label_{inc|dec} must be called from process context only,
we must defer jump_label_dec() if net_disable_timestamp() is called
from interrupt context.

Reported-by: Igor Maravic <igorm@etf.rs>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/core/dev.c
net/ipv4/netfilter/ip_queue.c
net/ipv6/netfilter/ip6_queue.c

index 962e3de..c7ef6c5 100644 (file)
@@ -1441,15 +1441,38 @@ int call_netdevice_notifiers(unsigned long val, struct net_device *dev)
 EXPORT_SYMBOL(call_netdevice_notifiers);
 
 static struct jump_label_key netstamp_needed __read_mostly;
+#ifdef HAVE_JUMP_LABEL
+/* We are not allowed to call jump_label_dec() from irq context
+ * If net_disable_timestamp() is called from irq context, defer the
+ * jump_label_dec() calls.
+ */
+static atomic_t netstamp_needed_deferred;
+#endif
 
 void net_enable_timestamp(void)
 {
+#ifdef HAVE_JUMP_LABEL
+       int deferred = atomic_xchg(&netstamp_needed_deferred, 0);
+
+       if (deferred) {
+               while (--deferred)
+                       jump_label_dec(&netstamp_needed);
+               return;
+       }
+#endif
+       WARN_ON(in_interrupt());
        jump_label_inc(&netstamp_needed);
 }
 EXPORT_SYMBOL(net_enable_timestamp);
 
 void net_disable_timestamp(void)
 {
+#ifdef HAVE_JUMP_LABEL
+       if (in_interrupt()) {
+               atomic_inc(&netstamp_needed_deferred);
+               return;
+       }
+#endif
        jump_label_dec(&netstamp_needed);
 }
 EXPORT_SYMBOL(net_disable_timestamp);
index e59aabd..a057fe6 100644 (file)
@@ -404,6 +404,7 @@ __ipq_rcv_skb(struct sk_buff *skb)
        int status, type, pid, flags;
        unsigned int nlmsglen, skblen;
        struct nlmsghdr *nlh;
+       bool enable_timestamp = false;
 
        skblen = skb->len;
        if (skblen < sizeof(*nlh))
@@ -441,12 +442,13 @@ __ipq_rcv_skb(struct sk_buff *skb)
                        RCV_SKB_FAIL(-EBUSY);
                }
        } else {
-               net_enable_timestamp();
+               enable_timestamp = true;
                peer_pid = pid;
        }
 
        spin_unlock_bh(&queue_lock);
-
+       if (enable_timestamp)
+               net_enable_timestamp();
        status = ipq_receive_peer(NLMSG_DATA(nlh), type,
                                  nlmsglen - NLMSG_LENGTH(0));
        if (status < 0)
index e63c397..fb80a23 100644 (file)
@@ -405,6 +405,7 @@ __ipq_rcv_skb(struct sk_buff *skb)
        int status, type, pid, flags;
        unsigned int nlmsglen, skblen;
        struct nlmsghdr *nlh;
+       bool enable_timestamp = false;
 
        skblen = skb->len;
        if (skblen < sizeof(*nlh))
@@ -442,11 +443,13 @@ __ipq_rcv_skb(struct sk_buff *skb)
                        RCV_SKB_FAIL(-EBUSY);
                }
        } else {
-               net_enable_timestamp();
+               enable_timestamp = true;
                peer_pid = pid;
        }
 
        spin_unlock_bh(&queue_lock);
+       if (enable_timestamp)
+               net_enable_timestamp();
 
        status = ipq_receive_peer(NLMSG_DATA(nlh), type,
                                  nlmsglen - NLMSG_LENGTH(0));