From 87151b8689d890dfb495081f7be9b9e257f7a2df Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 10 Apr 2012 20:08:39 +0000 Subject: net: allow pskb_expand_head() to get maximum tailroom Marc Merlin reported many order-1 allocations failures in TX path on its wireless setup, that dont make any sense with MTU=1500 network, and non SG capable hardware. Turns out part of the problem comes from pskb_expand_head() not using ksize() to get exact head size given by kmalloc(). Doing the same thing than __alloc_skb() allows more tailroom in skb and can prevent future reallocations. As a bonus, struct skb_shared_info becomes cache line aligned. Reported-by: Marc MERLIN Tested-by: Marc MERLIN Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/core/skbuff.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'net/core') diff --git a/net/core/skbuff.c b/net/core/skbuff.c index baf8d281152c..e59840010d45 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -952,9 +952,11 @@ int pskb_expand_head(struct sk_buff *skb, int nhead, int ntail, goto adjust_others; } - data = kmalloc(size + sizeof(struct skb_shared_info), gfp_mask); + data = kmalloc(size + SKB_DATA_ALIGN(sizeof(struct skb_shared_info)), + gfp_mask); if (!data) goto nodata; + size = SKB_WITH_OVERHEAD(ksize(data)); /* Copy only real data... and, alas, header. This should be * optimized for the cases when header is void. -- cgit v1.2.2 From 7d3d43dab4e978d8d9ad1acf8af15c9b1c4b0f0f Mon Sep 17 00:00:00 2001 From: "Eric W. Biederman" Date: Fri, 6 Apr 2012 15:33:35 +0000 Subject: net: In unregister_netdevice_notifier unregister the netdevices. We already synthesize events in register_netdevice_notifier and synthesizing events in unregister_netdevice_notifier allows to us remove the need for special case cleanup code. This change should be safe as it adds no new cases for existing callers of unregiser_netdevice_notifier to handle. Signed-off-by: Eric W. Biederman Signed-off-by: David S. Miller --- net/core/dev.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) (limited to 'net/core') diff --git a/net/core/dev.c b/net/core/dev.c index c25d453b2803..9bb8f87c4cda 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -1409,14 +1409,34 @@ EXPORT_SYMBOL(register_netdevice_notifier); * register_netdevice_notifier(). The notifier is unlinked into the * kernel structures and may then be reused. A negative errno code * is returned on a failure. + * + * After unregistering unregister and down device events are synthesized + * for all devices on the device list to the removed notifier to remove + * the need for special case cleanup code. */ int unregister_netdevice_notifier(struct notifier_block *nb) { + struct net_device *dev; + struct net *net; int err; rtnl_lock(); err = raw_notifier_chain_unregister(&netdev_chain, nb); + if (err) + goto unlock; + + for_each_net(net) { + for_each_netdev(net, dev) { + if (dev->flags & IFF_UP) { + nb->notifier_call(nb, NETDEV_GOING_DOWN, dev); + nb->notifier_call(nb, NETDEV_DOWN, dev); + } + nb->notifier_call(nb, NETDEV_UNREGISTER, dev); + nb->notifier_call(nb, NETDEV_UNREGISTER_BATCH, dev); + } + } +unlock: rtnl_unlock(); return err; } -- cgit v1.2.2 From b922934d017f1cc831b017913ed7d1a56c558b43 Mon Sep 17 00:00:00 2001 From: Julian Anastasov Date: Mon, 16 Apr 2012 04:43:15 +0000 Subject: netns: do not leak net_generic data on failed init ops_init should free the net_generic data on init failure and __register_pernet_operations should not call ops_free when NET_NS is not enabled. Signed-off-by: Julian Anastasov Reviewed-by: "Eric W. Biederman" Signed-off-by: David S. Miller --- net/core/net_namespace.c | 33 ++++++++++++++++++--------------- 1 file changed, 18 insertions(+), 15 deletions(-) (limited to 'net/core') diff --git a/net/core/net_namespace.c b/net/core/net_namespace.c index 0e950fda9a0a..31a5ae51a45c 100644 --- a/net/core/net_namespace.c +++ b/net/core/net_namespace.c @@ -83,21 +83,29 @@ assign: static int ops_init(const struct pernet_operations *ops, struct net *net) { - int err; + int err = -ENOMEM; + void *data = NULL; + if (ops->id && ops->size) { - void *data = kzalloc(ops->size, GFP_KERNEL); + data = kzalloc(ops->size, GFP_KERNEL); if (!data) - return -ENOMEM; + goto out; err = net_assign_generic(net, *ops->id, data); - if (err) { - kfree(data); - return err; - } + if (err) + goto cleanup; } + err = 0; if (ops->init) - return ops->init(net); - return 0; + err = ops->init(net); + if (!err) + return 0; + +cleanup: + kfree(data); + +out: + return err; } static void ops_free(const struct pernet_operations *ops, struct net *net) @@ -448,12 +456,7 @@ static void __unregister_pernet_operations(struct pernet_operations *ops) static int __register_pernet_operations(struct list_head *list, struct pernet_operations *ops) { - int err = 0; - err = ops_init(ops, &init_net); - if (err) - ops_free(ops, &init_net); - return err; - + return ops_init(ops, &init_net); } static void __unregister_pernet_operations(struct pernet_operations *ops) -- cgit v1.2.2 From bbe362be5368b9f531b95a4a9b502ae2832e1dac Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 19 Apr 2012 07:16:21 +0000 Subject: drop_monitor: allow more events per second It seems there is a logic error in trace_drop_common(), since we store only 64 drops, even if they are from same location. This fix is a one liner, but we probably need more work to avoid useless atomic dec/inc Now I can watch 1 Mpps drops through dropwatch... Signed-off-by: Eric Dumazet Cc: Neil Horman Acked-by: Neil Horman Signed-off-by: David S. Miller --- net/core/drop_monitor.c | 1 + 1 file changed, 1 insertion(+) (limited to 'net/core') diff --git a/net/core/drop_monitor.c b/net/core/drop_monitor.c index 7f36b38e060f..5c3c81a609e5 100644 --- a/net/core/drop_monitor.c +++ b/net/core/drop_monitor.c @@ -150,6 +150,7 @@ static void trace_drop_common(struct sk_buff *skb, void *location) for (i = 0; i < msg->entries; i++) { if (!memcmp(&location, msg->points[i].pc, sizeof(void *))) { msg->points[i].count++; + atomic_inc(&data->dm_hit_count); goto out; } } -- cgit v1.2.2