linux/net, branch v3.10.4

vlan: fix a race in egress prio management

2013-07-28T23:30:05Z

[ Upstream commit 3e3aac497513c669e1c62c71e1d552ea85c1d974 ] egress_priority_map[] hash table updates are protected by rtnl, and we never remove elements until device is dismantled. We have to make sure that before inserting an new element in hash table, all its fields are committed to memory or else another cpu could find corrupt values and crash. Signed-off-by: Eric Dumazet Cc: Patrick McHardy Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman

vlan: mask vlan prio bits

2013-07-28T23:30:05Z

[ Upstream commit d4b812dea4a236f729526facf97df1a9d18e191c ] In commit 48cc32d38a52d0b68f91a171a8d00531edc6a46e ("vlan: don't deliver frames for unknown vlans to protocols") Florian made sure we set pkt_type to PACKET_OTHERHOST if the vlan id is set and we could find a vlan device for this particular id. But we also have a problem if prio bits are set. Steinar reported an issue on a router receiving IPv6 frames with a vlan tag of 4000 (id 0, prio 2), and tunneled into a sit device, because skb->vlan_tci is set. Forwarded frame is completely corrupted : We can see (8100:4000) being inserted in the middle of IPv6 source address : 16:48:00.780413 IP6 2001:16d8:8100:4000:ee1c:0:9d9:bc87 > 9f94:4d95:2001:67c:29f4::: ICMP6, unknown icmp6 type (0), length 64 0x0000: 0000 0029 8000 c7c3 7103 0001 a0ae e651 0x0010: 0000 0000 ccce 0b00 0000 0000 1011 1213 0x0020: 1415 1617 1819 1a1b 1c1d 1e1f 2021 2223 0x0030: 2425 2627 2829 2a2b 2c2d 2e2f 3031 3233 It seems we are not really ready to properly cope with this right now. We can probably do better in future kernels : vlan_get_ingress_priority() should be a netdev property instead of a per vlan_dev one. For stable kernels, lets clear vlan_tci to fix the bugs. Reported-by: Steinar H. Gunderson Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman

pkt_sched: sch_qfq: remove a source of high packet delay/jitter

2013-07-28T23:30:04Z

[ Upstream commit 87f40dd6ce7042caca0b3b557e8923127f51f902 ] QFQ+ inherits from QFQ a design choice that may cause a high packet delay/jitter and a severe short-term unfairness. As QFQ, QFQ+ uses a special quantity, the system virtual time, to track the service provided by the ideal system it approximates. When a packet is dequeued, this quantity must be incremented by the size of the packet, divided by the sum of the weights of the aggregates waiting to be served. Tracking this sum correctly is a non-trivial task, because, to preserve tight service guarantees, the decrement of this sum must be delayed in a special way [1]: this sum can be decremented only after that its value would decrease also in the ideal system approximated by QFQ+. For efficiency, QFQ+ keeps track only of the 'instantaneous' weight sum, increased and decreased immediately as the weight of an aggregate changes, and as an aggregate is created or destroyed (which, in its turn, happens as a consequence of some class being created/destroyed/changed). However, to avoid the problems caused to service guarantees by these immediate decreases, QFQ+ increments the system virtual time using the maximum value allowed for the weight sum, 2^10, in place of the dynamic, instantaneous value. The instantaneous value of the weight sum is used only to check whether a request of weight increase or a class creation can be satisfied. Unfortunately, the problems caused by this choice are worse than the temporary degradation of the service guarantees that may occur, when a class is changed or destroyed, if the instantaneous value of the weight sum was used to update the system virtual time. In fact, the fraction of the link bandwidth guaranteed by QFQ+ to each aggregate is equal to the ratio between the weight of the aggregate and the sum of the weights of the competing aggregates. The packet delay guaranteed to the aggregate is instead inversely proportional to the guaranteed bandwidth. By using the maximum possible value, and not the actual value of the weight sum, QFQ+ provides each aggregate with the worst possible service guarantees, and not with service guarantees related to the actual set of competing aggregates. To see the consequences of this fact, consider the following simple example. Suppose that only the following aggregates are backlogged, i.e., that only the classes in the following aggregates have packets to transmit: one aggregate with weight 10, say A, and ten aggregates with weight 1, say B1, B2, ..., B10. In particular, suppose that these aggregates are always backlogged. Given the weight distribution, the smoothest and fairest service order would be: A B1 A B2 A B3 A B4 A B5 A B6 A B7 A B8 A B9 A B10 A B1 A B2 ... QFQ+ would provide exactly this optimal service if it used the actual value for the weight sum instead of the maximum possible value, i.e., 11 instead of 2^10. In contrast, since QFQ+ uses the latter value, it serves aggregates as follows (easy to prove and to reproduce experimentally): A B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 A A A A A A A A A A B1 B2 ... B10 A A ... By replacing 10 with N in the above example, and by increasing N, one can increase at will the maximum packet delay and the jitter experienced by the classes in aggregate A. This patch addresses this issue by just using the above 'instantaneous' value of the weight sum, instead of the maximum possible value, when updating the system virtual time. After the instantaneous weight sum is decreased, QFQ+ may deviate from the ideal service for a time interval in the order of the time to serve one maximum-size packet for each backlogged class. The worst-case extent of the deviation exhibited by QFQ+ during this time interval [1] is basically the same as of the deviation described above (but, without this patch, QFQ+ suffers from such a deviation all the time). Finally, this patch modifies the comment to the function qfq_slot_insert, to make it coherent with the fact that the weight sum used by QFQ+ can now be lower than the maximum possible value. [1] P. Valente, "Extending WF2Q+ to support a dynamic traffic mix", Proceedings of AAA-IDEA'05, June 2005. Signed-off-by: Paolo Valente Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman

ipv4: set transport header earlier

2013-07-28T23:30:04Z

[ Upstream commit 21d1196a35f5686c4323e42a62fdb4b23b0ab4a3 ] commit 45f00f99d6e ("ipv4: tcp: clean up tcp_v4_early_demux()") added a performance regression for non GRO traffic, basically disabling IP early demux. IPv6 stack resets transport header in ip6_rcv() before calling IP early demux in ip6_rcv_finish(), while IPv4 does this only in ip_local_deliver_finish(), _after_ IP early demux. GRO traffic happened to enable IP early demux because transport header is also set in inet_gro_receive() Instead of reverting the faulty commit, we can make IPv4/IPv6 behave the same : transport_header should be set in ip_rcv() instead of ip_local_deliver_finish() ip_local_deliver_finish() can also use skb_network_header_len() which is faster than ip_hdrlen() Signed-off-by: Eric Dumazet Cc: Neal Cardwell Cc: Tom Herbert Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman

ipv6: only static routes qualify for equal cost multipathing

2013-07-28T23:30:02Z

[ Upstream commit 307f2fb95e9b96b3577916e73d92e104f8f26494 ] Static routes in this case are non-expiring routes which did not get configured by autoconf or by icmpv6 redirects. To make sure we actually get an ecmp route while searching for the first one in this fib6_node's leafs, also make sure it matches the ecmp route assumptions. v2: a) Removed RTF_EXPIRE check in dst.from chain. The check of RTF_ADDRCONF already ensures that this route, even if added again without RTF_EXPIRES (in case of a RA announcement with infinite timeout), does not cause the rt6i_nsiblings logic to go wrong if a later RA updates the expiration time later. v3: a) Allow RTF_EXPIRES routes to enter the ecmp route set. We have to do so, because an pmtu event could update the RTF_EXPIRES flag and we would not count this route, if another route joins this set. We now filter only for RTF_GATEWAY|RTF_ADDRCONF|RTF_DYNAMIC, which are flags that don't get changed after rt6_info construction. Cc: Nicolas Dichtel Signed-off-by: Hannes Frederic Sowa Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman

gre: Fix MTU sizing check for gretap tunnels

2013-07-28T23:30:02Z

[ Upstream commit 8c91e162e058bb91b7766f26f4d5823a21941026 ] This change fixes an MTU sizing issue seen with gretap tunnels when non-gso packets are sent from the interface. In my case I was able to reproduce the issue by simply sending a ping of 1421 bytes with the gretap interface created on a device with a standard 1500 mtu. This fix is based on the fact that the tunnel mtu is already adjusted by dev->hard_header_len so it would make sense that any packets being compared against that mtu should also be adjusted by hard_header_len and the tunnel header instead of just the tunnel header. Signed-off-by: Alexander Duyck Reported-by: Cong Wang Acked-by: Eric Dumazet Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman

ipv6: fix route selection if kernel is not compiled with CONFIG_IPV6_ROUTER_PREF

2013-07-28T23:30:00Z

[ Upstream commit afc154e978de1eb11c555bc8bcec1552f75ebc43 ] This is a follow-up patch to 3630d40067a21d4dfbadc6002bb469ce26ac5d52 ("ipv6: rt6_check_neigh should successfully verify neigh if no NUD information are available"). Since the removal of rt->n in rt6_info we can end up with a dst == NULL in rt6_check_neigh. In case the kernel is not compiled with CONFIG_IPV6_ROUTER_PREF we should also select a route with unkown NUD state but we must not avoid doing round robin selection on routes with the same target. So introduce and pass down a boolean ``do_rr'' to indicate when we should update rt->rr_ptr. As soon as no route is valid we do backtracking and do a lookup on a higher level in the fib trie. v2: a) Improved rt6_check_neigh logic (no need to create neighbour there) and documented return values. v3: a) Introduce enum rt6_nud_state to get rid of the magic numbers (thanks to David Miller). b) Update and shorten commit message a bit to actualy reflect the source. Reported-by: Pierre Emeriaud Cc: YOSHIFUJI Hideaki Signed-off-by: Hannes Frederic Sowa Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman

9p: fix off by one causing access violations and memory corruption

2013-07-28T23:29:58Z

[ Upstream commit 110ecd69a9feea82a152bbf9b12aba57e6396883 ] p9_release_pages() would attempt to dereference one value past the end of pages[]. This would cause the following crashes: [ 6293.171817] BUG: unable to handle kernel paging request at ffff8807c96f3000 [ 6293.174146] IP: [] p9_release_pages+0x3b/0x60 [ 6293.176447] PGD 79c5067 PUD 82c1e3067 PMD 82c197067 PTE 80000007c96f3060 [ 6293.180060] Oops: 0000 [#1] PREEMPT SMP DEBUG_PAGEALLOC [ 6293.180060] Modules linked in: [ 6293.180060] CPU: 62 PID: 174043 Comm: modprobe Tainted: G W 3.10.0-next-20130710-sasha #3954 [ 6293.180060] task: ffff8807b803b000 ti: ffff880787dde000 task.ti: ffff880787dde000 [ 6293.180060] RIP: 0010:[] [] p9_release_pages+0x3b/0x60 [ 6293.214316] RSP: 0000:ffff880787ddfc28 EFLAGS: 00010202 [ 6293.214316] RAX: 0000000000000001 RBX: ffff8807c96f2ff8 RCX: 0000000000000000 [ 6293.222017] RDX: ffff8807b803b000 RSI: 0000000000000001 RDI: ffffea001c7e3d40 [ 6293.222017] RBP: ffff880787ddfc48 R08: 0000000000000000 R09: 0000000000000000 [ 6293.222017] R10: 0000000000000001 R11: 0000000000000000 R12: 0000000000000001 [ 6293.222017] R13: 0000000000000001 R14: ffff8807cc50c070 R15: ffff8807cc50c070 [ 6293.222017] FS: 00007f572641d700(0000) GS:ffff8807f3600000(0000) knlGS:0000000000000000 [ 6293.256784] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 6293.256784] CR2: ffff8807c96f3000 CR3: 00000007c8e81000 CR4: 00000000000006e0 [ 6293.256784] Stack: [ 6293.256784] ffff880787ddfcc8 ffff880787ddfcc8 0000000000000000 ffff880787ddfcc8 [ 6293.256784] ffff880787ddfd48 ffffffff84128be8 ffff880700000002 0000000000000001 [ 6293.256784] ffff8807b803b000 ffff880787ddfce0 0000100000000000 0000000000000000 [ 6293.256784] Call Trace: [ 6293.256784] [] p9_virtio_zc_request+0x598/0x630 [ 6293.256784] [] ? wake_up_bit+0x40/0x40 [ 6293.256784] [] p9_client_zc_rpc+0x111/0x3a0 [ 6293.256784] [] ? sched_clock_cpu+0x108/0x120 [ 6293.256784] [] p9_client_read+0xe1/0x2c0 [ 6293.256784] [] v9fs_file_read+0x90/0xc0 [ 6293.256784] [] vfs_read+0xc3/0x130 [ 6293.256784] [] ? trace_hardirqs_on+0xd/0x10 [ 6293.256784] [] SyS_read+0x62/0xa0 [ 6293.256784] [] tracesys+0xdd/0xe2 [ 6293.256784] Code: 66 90 48 89 fb 41 89 f5 48 8b 3f 48 85 ff 74 29 85 f6 74 25 45 31 e4 66 0f 1f 84 00 00 00 00 00 e8 eb 14 12 fd 41 ff c4 49 63 c4 <48> 8b 3c c3 48 85 ff 74 05 45 39 e5 75 e7 48 83 c4 08 5b 41 5c [ 6293.256784] RIP [] p9_release_pages+0x3b/0x60 [ 6293.256784] RSP [ 6293.256784] CR2: ffff8807c96f3000 [ 6293.256784] ---[ end trace 50822ee72cd360fc ]--- Signed-off-by: Sasha Levin Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman

ipv6: in case of link failure remove route directly instead of letting it expire

2013-07-28T23:29:58Z

[ Upstream commit 1eb4f758286884e7566627164bca4c4a16952a83 ] We could end up expiring a route which is part of an ecmp route set. Doing so would invalidate the rt->rt6i_nsiblings calculations and could provoke the following panic: [ 80.144667] ------------[ cut here ]------------ [ 80.145172] kernel BUG at net/ipv6/ip6_fib.c:733! [ 80.145172] invalid opcode: 0000 [#1] SMP [ 80.145172] Modules linked in: 8021q nf_conntrack_netbios_ns nf_conntrack_broadcast ipt_MASQUERADE ip6table_mangle ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 iptable_nat nf_nat_ipv4 nf_nat iptable_mangle nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack ebtable_filter ebtables ip6table_filter ip6_tables +snd_hda_intel snd_hda_codec snd_hwdep snd_seq snd_seq_device snd_pcm snd_page_alloc snd_timer virtio_balloon snd soundcore i2c_piix4 i2c_core virtio_net virtio_blk [ 80.145172] CPU: 1 PID: 786 Comm: ping6 Not tainted 3.10.0+ #118 [ 80.145172] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011 [ 80.145172] task: ffff880117fa0000 ti: ffff880118770000 task.ti: ffff880118770000 [ 80.145172] RIP: 0010:[] [] fib6_add+0x75d/0x830 [ 80.145172] RSP: 0018:ffff880118771798 EFLAGS: 00010202 [ 80.145172] RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffff88011350e480 [ 80.145172] RDX: ffff88011350e238 RSI: 0000000000000004 RDI: ffff88011350f738 [ 80.145172] RBP: ffff880118771848 R08: ffff880117903280 R09: 0000000000000001 [ 80.145172] R10: 0000000000000000 R11: 0000000000000000 R12: ffff88011350f680 [ 80.145172] R13: ffff880117903280 R14: ffff880118771890 R15: ffff88011350ef90 [ 80.145172] FS: 00007f02b5127740(0000) GS:ffff88011fd00000(0000) knlGS:0000000000000000 [ 80.145172] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b [ 80.145172] CR2: 00007f981322a000 CR3: 00000001181b1000 CR4: 00000000000006e0 [ 80.145172] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 80.145172] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 80.145172] Stack: [ 80.145172] 0000000000000001 ffff880100000000 ffff880100000000 ffff880117903280 [ 80.145172] 0000000000000000 ffff880119a4cf00 0000000000000400 00000000000007fa [ 80.145172] 0000000000000000 0000000000000000 0000000000000000 ffff88011350f680 [ 80.145172] Call Trace: [ 80.145172] [] ? rt6_bind_peer+0x4b/0x90 [ 80.145172] [] __ip6_ins_rt+0x45/0x70 [ 80.145172] [] ip6_ins_rt+0x35/0x40 [ 80.145172] [] ip6_pol_route.isra.44+0x3a4/0x4b0 [ 80.145172] [] ip6_pol_route_output+0x2a/0x30 [ 80.145172] [] fib6_rule_action+0xd7/0x210 [ 80.145172] [] ? ip6_pol_route_input+0x30/0x30 [ 80.145172] [] fib_rules_lookup+0xc6/0x140 [ 80.145172] [] fib6_rule_lookup+0x44/0x80 [ 80.145172] [] ? ip6_pol_route_input+0x30/0x30 [ 80.145172] [] ip6_route_output+0x73/0xb0 [ 80.145172] [] ip6_dst_lookup_tail+0x2c3/0x2e0 [ 80.145172] [] ? list_del+0x11/0x40 [ 80.145172] [] ? remove_wait_queue+0x3c/0x50 [ 80.145172] [] ip6_dst_lookup_flow+0x3d/0xa0 [ 80.145172] [] rawv6_sendmsg+0x267/0xc20 [ 80.145172] [] inet_sendmsg+0x63/0xb0 [ 80.145172] [] ? selinux_socket_sendmsg+0x23/0x30 [ 80.145172] [] sock_sendmsg+0xa6/0xd0 [ 80.145172] [] SYSC_sendto+0x128/0x180 [ 80.145172] [] ? update_curr+0xec/0x170 [ 80.145172] [] ? kvm_clock_get_cycles+0x9/0x10 [ 80.145172] [] ? __getnstimeofday+0x3e/0xd0 [ 80.145172] [] SyS_sendto+0xe/0x10 [ 80.145172] [] system_call_fastpath+0x16/0x1b [ 80.145172] Code: fe ff ff 41 f6 45 2a 06 0f 85 ca fe ff ff 49 8b 7e 08 4c 89 ee e8 94 ef ff ff e9 b9 fe ff ff 48 8b 82 28 05 00 00 e9 01 ff ff ff <0f> 0b 49 8b 54 24 30 0d 00 00 40 00 89 83 14 01 00 00 48 89 53 [ 80.145172] RIP [] fib6_add+0x75d/0x830 [ 80.145172] RSP [ 80.387413] ---[ end trace 02f20b7a8b81ed95 ]--- [ 80.390154] Kernel panic - not syncing: Fatal exception in interrupt Signed-off-by: Hannes Frederic Sowa Cc: Nicolas Dichtel Cc: YOSHIFUJI Hideaki Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman

ipv6: rt6_check_neigh should successfully verify neigh if no NUD information are available

2013-07-28T23:29:50Z

[ Upstream commit 3630d40067a21d4dfbadc6002bb469ce26ac5d52 ] After the removal of rt->n we do not create a neighbour entry at route insertion time (rt6_bind_neighbour is gone). As long as no neighbour is created because of "useful traffic" we skip this routing entry because rt6_check_neigh cannot pick up a valid neighbour (neigh == NULL) and thus returns false. This change was introduced by commit 887c95cc1da53f66a5890fdeab13414613010097 ("ipv6: Complete neighbour entry removal from dst_entry.") To quote RFC4191: "If the host has no information about the router's reachability, then the host assumes the router is reachable." and also: "A host MUST NOT probe a router's reachability in the absence of useful traffic that the host would have sent to the router if it were reachable." So, just assume the router is reachable and let's rt6_probe do the rest. We don't need to create a neighbour on route insertion time. If we don't compile with CONFIG_IPV6_ROUTER_PREF (RFC4191 support) a neighbour is only valid if its nud_state is NUD_VALID. I did not find any references that we should probe the router on route insertion time via the other RFCs. So skip this route in that case. v2: a) use IS_ENABLED instead of #ifdefs (thanks to Sergei Shtylyov) Reported-by: Pierre Emeriaud Cc: YOSHIFUJI Hideaki Signed-off-by: Hannes Frederic Sowa Signed-off-by: David S. Miller Signed-off-by: Greg Kroah-Hartman