From 9979a55a833883242e3a29f3596676edd7199c46 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 22 Mar 2013 14:38:28 +0000 Subject: net: remove a WARN_ON() in net_enable_timestamp() The WARN_ON(in_interrupt()) in net_enable_timestamp() can get false positive, in socket clone path, run from softirq context : [ 3641.624425] WARNING: at net/core/dev.c:1532 net_enable_timestamp+0x7b/0x80() [ 3641.668811] Call Trace: [ 3641.671254] [] warn_slowpath_common+0x87/0xc0 [ 3641.677871] [] warn_slowpath_null+0x1a/0x20 [ 3641.683683] [] net_enable_timestamp+0x7b/0x80 [ 3641.689668] [] sk_clone_lock+0x425/0x450 [ 3641.695222] [] inet_csk_clone_lock+0x16/0x170 [ 3641.701213] [] tcp_create_openreq_child+0x29/0x820 [ 3641.707663] [] ? ipt_do_table+0x222/0x670 [ 3641.713354] [] tcp_v4_syn_recv_sock+0xab/0x3d0 [ 3641.719425] [] tcp_check_req+0x3da/0x530 [ 3641.724979] [] ? inet_hashinfo_init+0x60/0x80 [ 3641.730964] [] ? tcp_v4_rcv+0x79f/0xbe0 [ 3641.736430] [] tcp_v4_do_rcv+0x38d/0x4f0 [ 3641.741985] [] tcp_v4_rcv+0xa7a/0xbe0 Its safe at this point because the parent socket owns a reference on the netstamp_needed, so we cant have a 0 -> 1 transition, which requires to lock a mutex. Instead of refining the check, lets remove it, as all known callers are safe. If it ever changes in the future, static_key_slow_inc() will complain anyway. Reported-by: Laurent Chavey Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/core/dev.c | 1 - 1 file changed, 1 deletion(-) (limited to 'net/core/dev.c') diff --git a/net/core/dev.c b/net/core/dev.c index d540ced1f6c6..b13e5c766c11 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -1545,7 +1545,6 @@ void net_enable_timestamp(void) return; } #endif - WARN_ON(in_interrupt()); static_key_slow_inc(&netstamp_needed); } EXPORT_SYMBOL(net_enable_timestamp); -- cgit v1.2.1 From a561cf7edf9863198bfccecfc5cfe26d951ebd20 Mon Sep 17 00:00:00 2001 From: Shmulik Ladkani Date: Wed, 27 Mar 2013 23:13:26 +0000 Subject: net: core: Remove redundant call to 'nf_reset' in 'dev_forward_skb' 'nf_reset' is called just prior calling 'netif_rx'. No need to call it twice. Reported-by: Igor Michailov Signed-off-by: Shmulik Ladkani Acked-by: Eric Dumazet Signed-off-by: David S. Miller --- net/core/dev.c | 1 - 1 file changed, 1 deletion(-) (limited to 'net/core/dev.c') diff --git a/net/core/dev.c b/net/core/dev.c index b13e5c766c11..6591440cc03d 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -1624,7 +1624,6 @@ int dev_forward_skb(struct net_device *dev, struct sk_buff *skb) } skb_orphan(skb); - nf_reset(skb); if (unlikely(!is_skb_forwardable(dev, skb))) { atomic_long_inc(&dev->rx_dropped); -- cgit v1.2.1 From 00cfec37484761a44a3b6f4675a54caa618210ae Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 29 Mar 2013 03:01:22 +0000 Subject: net: add a synchronize_net() in netdev_rx_handler_unregister() commit 35d48903e97819 (bonding: fix rx_handler locking) added a race in bonding driver, reported by Steven Rostedt who did a very good diagnosis : I'm currently debugging a crash in an old 3.0-rt kernel that one of our customers is seeing. The bug happens with a stress test that loads and unloads the bonding module in a loop (I don't know all the details as I'm not the one that is directly interacting with the customer). But the bug looks to be something that may still be present and possibly present in mainline too. It will just be much harder to trigger it in mainline. In -rt, interrupts are threads, and can schedule in and out just like any other thread. Note, mainline now supports interrupt threads so this may be easily reproducible in mainline as well. I don't have the ability to tell the customer to try mainline or other kernels, so my hands are somewhat tied to what I can do. But according to a core dump, I tracked down that the eth irq thread crashed in bond_handle_frame() here: slave = bond_slave_get_rcu(skb->dev); bond = slave->bond; <--- BUG the slave returned was NULL and accessing slave->bond caused a NULL pointer dereference. Looking at the code that unregisters the handler: void netdev_rx_handler_unregister(struct net_device *dev) { ASSERT_RTNL(); RCU_INIT_POINTER(dev->rx_handler, NULL); RCU_INIT_POINTER(dev->rx_handler_data, NULL); } Which is basically: dev->rx_handler = NULL; dev->rx_handler_data = NULL; And looking at __netif_receive_skb() we have: rx_handler = rcu_dereference(skb->dev->rx_handler); if (rx_handler) { if (pt_prev) { ret = deliver_skb(skb, pt_prev, orig_dev); pt_prev = NULL; } switch (rx_handler(&skb)) { My question to all of you is, what stops this interrupt from happening while the bonding module is unloading? What happens if the interrupt triggers and we have this: CPU0 CPU1 ---- ---- rx_handler = skb->dev->rx_handler netdev_rx_handler_unregister() { dev->rx_handler = NULL; dev->rx_handler_data = NULL; rx_handler() bond_handle_frame() { slave = skb->dev->rx_handler; bond = slave->bond; <-- NULL pointer dereference!!! What protection am I missing in the bond release handler that would prevent the above from happening? We can fix bug this in two ways. First is adding a test in bond_handle_frame() and others to check if rx_handler_data is NULL. A second way is adding a synchronize_net() in netdev_rx_handler_unregister() to make sure that a rcu protected reader has the guarantee to see a non NULL rx_handler_data. The second way is better as it avoids an extra test in fast path. Reported-by: Steven Rostedt Signed-off-by: Eric Dumazet Cc: Jiri Pirko Cc: Paul E. McKenney Acked-by: Steven Rostedt Reviewed-by: Paul E. McKenney Signed-off-by: David S. Miller --- net/core/dev.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'net/core/dev.c') diff --git a/net/core/dev.c b/net/core/dev.c index 6591440cc03d..13e6447f0398 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -3313,6 +3313,7 @@ int netdev_rx_handler_register(struct net_device *dev, if (dev->rx_handler) return -EBUSY; + /* Note: rx_handler_data must be set before rx_handler */ rcu_assign_pointer(dev->rx_handler_data, rx_handler_data); rcu_assign_pointer(dev->rx_handler, rx_handler); @@ -3333,6 +3334,11 @@ void netdev_rx_handler_unregister(struct net_device *dev) ASSERT_RTNL(); RCU_INIT_POINTER(dev->rx_handler, NULL); + /* a reader seeing a non NULL rx_handler in a rcu_read_lock() + * section has a guarantee to see a non NULL rx_handler_data + * as well. + */ + synchronize_net(); RCU_INIT_POINTER(dev->rx_handler_data, NULL); } EXPORT_SYMBOL_GPL(netdev_rx_handler_unregister); -- cgit v1.2.1 From 124dff01afbdbff251f0385beca84ba1b9adda68 Mon Sep 17 00:00:00 2001 From: Patrick McHardy Date: Fri, 5 Apr 2013 20:42:05 +0200 Subject: netfilter: don't reset nf_trace in nf_reset() Commit 130549fe ("netfilter: reset nf_trace in nf_reset") added code to reset nf_trace in nf_reset(). This is wrong and unnecessary. nf_reset() is used in the following cases: - when passing packets up the the socket layer, at which point we want to release all netfilter references that might keep modules pinned while the packet is queued. nf_trace doesn't matter anymore at this point. - when encapsulating or decapsulating IPsec packets. We want to continue tracing these packets after IPsec processing. - when passing packets through virtual network devices. Only devices on that encapsulate in IPv4/v6 matter since otherwise nf_trace is not used anymore. Its not entirely clear whether those packets should be traced after that, however we've always done that. - when passing packets through virtual network devices that make the packet cross network namespace boundaries. This is the only cases where we clearly want to reset nf_trace and is also what the original patch intended to fix. Add a new function nf_reset_trace() and use it in dev_forward_skb() to fix this properly. Signed-off-by: Patrick McHardy Signed-off-by: David S. Miller --- net/core/dev.c | 1 + 1 file changed, 1 insertion(+) (limited to 'net/core/dev.c') diff --git a/net/core/dev.c b/net/core/dev.c index 13e6447f0398..e7d68ed8aafe 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -1639,6 +1639,7 @@ int dev_forward_skb(struct net_device *dev, struct sk_buff *skb) skb->mark = 0; secpath_reset(skb); nf_reset(skb); + nf_reset_trace(skb); return netif_rx(skb); } EXPORT_SYMBOL_GPL(dev_forward_skb); -- cgit v1.2.1