The number of variables with Hungarian notation (l_ptr, n_ptr etc.)
has been significantly reduced over the last couple of years.

We now root out the last traces of this practice.
There are no functional changes in this commit.
Reviewed-by: NYing Xue <ying.xue@windriver.com>
Signed-off-by: NJon Maloy <jon.maloy@ericsson.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We move the definition of struct tipc_link from link.h to link.c in
order to minimize its exposure to the rest of the code.

When needed, we define new functions to make it possible for external
entities to access and set data in the link.

Apart from the above, there are no functional changes.
Reviewed-by: NYing Xue <ying.xue@windriver.com>
Signed-off-by: NJon Maloy <jon.maloy@ericsson.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In our effort to have less code and include dependencies between
entities such as node, link and bearer, we try to narrow down
the exposed interface towards the node as much as possible.

In this commit, we move the definition of struct tipc_node, along
with many of its associated function declarations, from node.h to
node.c. We also move some function definitions from link.c and
name_distr.c to node.c, since they access fields in struct tipc_node
that should not be externally visible. The moved functions are renamed
according to new location, and made static whenever possible.

There are no functional changes in this commit.
Reviewed-by: NYing Xue <ying.xue@windriver.com>
Signed-off-by: NJon Maloy <jon.maloy@ericsson.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

According to the node FSM a node in state SELF_UP_PEER_UP cannot
change state inside a lock context, except when a TUNNEL_PROTOCOL
(SYNCH or FAILOVER) packet arrives. However, the node's individual
links may still change state.

Since each link now is protected by its own spinlock, we finally have
the conditions in place to convert the node spinlock to an rwlock_t.
If the node state and arriving packet type are rigth, we can let the
link directly receive the packet under protection of its own spinlock
and the node lock in read mode. In all other cases we use the node
lock in write mode. This enables full concurrent execution between
parallel links during steady-state traffic situations, i.e., 99+ %
of the time.

This commit implements this change.
Reviewed-by: NYing Xue <ying.xue@windriver.com>
Signed-off-by: NJon Maloy <jon.maloy@ericsson.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

As a preparation to allow parallel links to work more independently
from each other we introduce a per-link spinlock, to be stored in the
struct nodes's link entry area. Since the node lock still is a regular
spinlock there is no increase in parallellism at this stage.
Reviewed-by: NYing Xue <ying.xue@windriver.com>
Signed-off-by: NJon Maloy <jon.maloy@ericsson.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The file name_distr.c currently contains three functions,
named_cluster_distribute(), tipc_publ_subcscribe() and
tipc_publ_unsubscribe() that all directly access fields in
struct tipc_node. We want to eliminate such dependencies, so
we move those functions to the file node.c and rename them to
tipc_node_broadcast(), tipc_node_subscribe() and tipc_node_unsubscribe()
respectively.
Reviewed-by: NYing Xue <ying.xue@windriver.com>
Signed-off-by: NJon Maloy <jon.maloy@ericsson.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The function tipc_node_check_state() contains the core logics
for handling link synchronization and failover. For this reason,
it is important to keep it as comprehensible as possible.

In this commit, we make three small cleanups.

1) If the node is in state SELF_DOWN_PEER_LEAVING and the received
   packet confirms that the peer has lost contact, there will be no
   further action in this function. To make this clearer, we return
   from the function directly after the state change.

2) Since commit 0f8b8e28 ("tipc: eliminate risk of stalled
   link synchronization") only the logically first TUNNEL_PROTO/SYNCH
   packet can alter the link state and set the synch point,
   independently of arrival order. Hence, there is not any longer any
   need to adjust the synch value in case such packets arrive in
   disorder. We remove this adjustment.

3) It is the intention that any message arriving on any of the links
   may trig a check for and possible termination of a node SYNCH state.
   A redundant and unnoticed check for tipc_link_is_synching() obviously
   beats this purpose, with the effect that only packets arriving on the
   synching link may currently end the synch state. We remove this check.
   This change will further shorten the synchronization period between
   parallel links.
Reviewed-by: NYing Xue <ying.xue@windriver.com>
Signed-off-by: NJon Maloy <jon.maloy@ericsson.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In commit 5cbb28a4 ("tipc: linearize arriving NAME_DISTR
and LINK_PROTO buffers") we added linearization of NAME_DISTRIBUTOR,
LINK_PROTOCOL/RESET and LINK_PROTOCOL/ACTIVATE to the function
tipc_udp_recv(). The location of the change was selected in order
to make the commit easily appliable to 'net' and 'stable'.

We now move this linearization to where it should be done, in the
functions tipc_named_rcv() and tipc_link_proto_rcv() respectively.
Reviewed-by: NYing Xue <ying.xue@windriver.com>
Signed-off-by: NJon Maloy <jon.maloy@ericsson.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

NAPI drivers no longer need to observe a particular protocol
to benefit from busy polling (CONFIG_NET_RX_BUSY_POLL=y)

napi_hash_add() and napi_hash_del() are automatically called
from core networking stack, respectively from
netif_napi_add() and netif_napi_del()

This patch depends on free_netdev() and netif_napi_del() being
called from process context, which seems to be the norm.

Drivers might still prefer to call napi_hash_del() on their
own, since they might combine all the rcu grace periods into
a single one, knowing their NAPI structures lifetime, while
core networking stack has no idea of a possible combining.

Once this patch proves to not bring serious regressions,
we will cleanup drivers to either remove napi_hash_del()
or provide appropriate rcu grace periods combining.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

napi_hash_del() will soon be used from both drivers (if they want)
or core networking stack.

Callers are responsibles to ensure an RCU grace period is respected
before freeing napi structure : napi_hash_del() can signal if
this RCU grace period is needed or not.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We do not often add/delete a napi context.
Moving napi_hash[] into read mostly section avoids potential false sharing.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

netif_tx_napi_add() is a variant of netif_napi_add()

It should be used by drivers that use a napi structure
to exclusively poll TX.

We do not want to add this kind of napi in napi_hash[] in following
patches, adding generic busy polling to all NAPI drivers.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We would like to automatically provide busy polling support
to all NAPI drivers, without them having to implement anything.

skb_mark_napi_id() can be called from napi_gro_receive() and
napi_get_frags().

Few drivers are still calling skb_mark_napi_id() because
they use netif_receive_skb(). They should eventually call
napi_gro_receive() instead. I will leave this to drivers
maintainers.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Instead of having to implement complex ndo_busy_poll() method,
drivers can simply rely on NAPI poll logic.

Busy polling gains are mainly coming from polling itself,
not on exact details on how we poll the device.

ndo_busy_poll() if implemented can avoid touching
napi state, but it adds extra synchronization between
normal napi->poll() and busy poll handler, slowing down
the common path (non busy polling) with extra atomic operations.
In practice few drivers ever got busy poll because of the complexity.

We could go one step further, and make busy polling
available for all NAPI drivers, but this would require
that all netif_napi_del() calls are done in process context
so that we can call synchronize_rcu().
Full audit would be required.

Before this is done, a driver still needs to call :

- skb_mark_napi_id() for each skb provided to the stack.
- napi_hash_add() and napi_hash_del() to allocate a napi_id per napi struct.
- Make sure RCU grace period is respected after napi_hash_del() before
  memory containing napi structure is freed.

Followup patch implements busy poll for mlx5 driver as an example.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Instead of blocking BH in whole sk_busy_loop(), block them
only around ->ndo_busy_poll() calls.

This has many benefits.

1) allow tunneled traffic to use busy poll as well as native traffic.
   Tunnels handlers usually call netif_rx() and depend on net_rx_action()
   being run (from sofirq handler)

2) allow RFS/RPS being used (sending IPI to other cpus if needed)

3) use the 'lets burn cpu cycles' budget to do useful work
   (like TX completions, timers, RCU callbacks...)

4) reduce BH latencies, making busy poll a better citizen.

Tested:

Tested with SIT tunnel

lpaa5:~# echo 0 >/proc/sys/net/core/busy_read
lpaa5:~# ./netperf -H 2002:af6:786::1 -t TCP_RR
MIGRATED TCP REQUEST/RESPONSE TEST from ::0 (::) port 0 AF_INET6 to 2002:af6:786::1 () port 0 AF_INET6 : first burst 0
Local /Remote
Socket Size   Request  Resp.   Elapsed  Trans.
Send   Recv   Size     Size    Time     Rate
bytes  Bytes  bytes    bytes   secs.    per sec

16384  87380  1        1       10.00    37373.93
16384  87380

Now enable busy poll on both hosts

lpaa5:~# echo 70 >/proc/sys/net/core/busy_read
lpaa6:~# echo 70 >/proc/sys/net/core/busy_read

lpaa5:~# ./netperf -H 2002:af6:786::1 -t TCP_RR
MIGRATED TCP REQUEST/RESPONSE TEST from ::0 (::) port 0 AF_INET6 to 2002:af6:786::1 () port 0 AF_INET6 : first burst 0
Local /Remote
Socket Size   Request  Resp.   Elapsed  Trans.
Send   Recv   Size     Size    Time     Rate
bytes  Bytes  bytes    bytes   secs.    per sec

16384  87380  1        1       10.00    58314.77
16384  87380
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

There is really little gain from inlining this big function.
We'll soon make it even bigger in following patches.

This means we no longer need to export napi_by_id()
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

skb->sender_cpu and skb->napi_id share a common storage,
and we had various bugs about this.

We had to call skb_sender_cpu_clear() in some places to
not leave a prior skb->napi_id and fool netdev_pick_tx()

As suggested by Alexei, we could split the space so that
these errors can not happen.

0 value being reserved as the common (not initialized) value,
let's reserve [1 .. NR_CPUS] range for valid sender_cpu,
and [NR_CPUS+1 .. ~0U] for valid napi_id.

This will allow proper busy polling support over tunnels.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Suggested-by: NAlexei Starovoitov <ast@kernel.org>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Replace printk calls with preferred unconditional log method calls to keep
kernel messages clean.

Added newline to "too small MTU" message.
Signed-off-by: NBastian Stender <bst@pengutronix.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This reverts commit 00ee5927 ("net: fix __netdev_update_features return
on ndo_set_features failure")
and adds a comment explaining why it's okay to return a value other than
0 upon error. Some drivers might actually change flags and return an
error so it's better to fire a spurious notification rather than miss
these.

CC: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

While possibly in future we don't necessarily need to use
sk_buff_head.lock this is a rather larger change, as it affects the
af_unix fd garbage collector, diag and socket cleanups. This is too much
for a stable patch.

For the time being grab sk_buff_head.lock without disabling bh and irqs,
so don't use locked skb_queue_tail.

Fixes: 869e7c62 ("net: af_unix: implement stream sendpage support")
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Reported-by: NEric Dumazet <edumazet@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix the following warning:

  CC      net/core/rtnetlink.o
net/core/rtnetlink.c: In function ‘rtnl_fill_ifinfo’:
net/core/rtnetlink.c:1308:1: warning: the frame size of 2864 bytes is larger than 2048 bytes [-Wframe-larger-than=]
 }
 ^
by splitting up the huge rtnl_fill_ifinfo into some smaller ones, so we
don't have the huge frame allocations at the same time.

Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

1. new skb only need dst and ip address(v4 or v6).
2. skb_copy may need high order pages, which is very rare on long running server.
Signed-off-by: NJunwei Zhang <linggao.zjw@alibaba-inc.com>
Signed-off-by: NMartin Zhang <martinbj2008@gmail.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use PAGE_ALIGNED(...) instead of open-coding it.
Signed-off-by: NTobias Klauser <tklauser@distanz.ch>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

rb->frames_per_block is an unsigned int, thus can never be negative.

Also fix spacing in the calculation of frames_per_block.
Signed-off-by: NTobias Klauser <tklauser@distanz.ch>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When a vlan is configured with REORDER_HEADER set to 0, the vlan
header is put back into the packet and makes it appear that
the vlan header is still there even after it's been processed.
This posses a problem for bridge and macvlan ports.  The packets
passed to those device may be forwarded and at the time of the
forward, vlan headers end up being unexpectedly present.

With the patch, we make sure that we do not put the vlan header
back (when REORDER_HEADER is 0) if a bridge or macvlan has
been configured on top of the vlan device.
Signed-off-by: NVladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When we have multiple stacked vlan devices all of which have
turned off REORDER_HEADER flag, the untag operation does not
locate the ethernet addresses correctly for nested vlans.
The reason is that in case of REORDER_HEADER flag being off,
the outer vlan headers are put back and the mac_len is adjusted
to account for the presense of the header.  Then, the subsequent
untag operation, for the next level vlan, always use VLAN_ETH_HLEN
to locate the begining of the ethernet header and that ends up
being a multiple of 4 bytes short of the actuall beginning
of the mac header (the multiple depending on the how many vlan
encapsulations ethere are).

As a reslult, if there are multiple levles of vlan devices
with REODER_HEADER being off, the recevied packets end up
being dropped.

To solve this, we use skb->mac_len as the offset.  The value
is always set on receive path and starts out as a ETH_HLEN.
The value is also updated when the vlan header manupations occur
so we know it will be correct.
Signed-off-by: NVladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

the OUTMCAST stat is double incremented, getting bumped once in the mcast code
itself, and again in the common ip output path.  Remove the mcast bump, as its
not needed

Validated by the reporter, with good results
Signed-off-by: NNeil Horman <nhorman@tuxdriver.com>
Reported-by: NClaus Jensen <claus.jensen@microsemi.com>
CC: Claus Jensen <claus.jensen@microsemi.com>
CC: David Miller <davem@davemloft.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

A recent flaw in the netdev feature setting resulted in warnings
like this one from VLAN interfaces:

 WARNING: CPU: 1 PID: 4975 at net/core/dev.c:2419 skb_warn_bad_offload+0xbc/0xcb()
 : caps=(0x00000000001b5820, 0x00000000001b5829) len=2782 data_len=0 gso_size=1348 gso_type=16 ip_summed=3

The ":" is supposed to be preceded by a driver name, but in this
case it is an empty string since the device has no parent.

There are many types of network devices without a parent. The
anonymous warnings for these devices can be hard to debug.  Log
the network device name instead in these cases to assist further
debugging.

This is mostly similar to how __netdev_printk() handles orphan
devices.
Signed-off-by: NBjørn Mork <bjorn@mork.no>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In case multiple writes to a unix stream socket race we could end up in a
situation where we pre-allocate a new skb for use in unix_stream_sendpage
but have to free it again in the locked section because another skb
has been appended meanwhile, which we must use. Accidentally we didn't
clear the pointer after consuming it and so we touched freed memory
while appending it to the sk_receive_queue. So, clear the pointer after
consuming the skb.

This bug has been found with syzkaller
(http://github.com/google/syzkaller) by Dmitry Vyukov.

Fixes: 869e7c62 ("net: af_unix: implement stream sendpage support")
Reported-by: NDmitry Vyukov <dvyukov@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Sending ICMP packets with raw sockets ends up in the SNMP counters
logging the type as the first byte of the IPv4 header rather than
the ICMP header. This is fixed by adding the IP Header Length to
the casting into a icmphdr struct.
Signed-off-by: NBen Cartwright-Cox <ben@benjojo.co.uk>
Acked-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If ndo_set_features fails __netdev_update_features() will return -1 but
this is wrong because it is expected to return 0 if no features were
changed (see netdev_update_features()), which will cause a netdev
notifier to be called without any actual changes. Fix this by returning
0 if ndo_set_features fails.

Fixes: 6cb6a27c ("net: Call netdev_features_change() from netdev_update_features()")
CC: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When __netdev_update_features() was updated to ensure some features are
disabled on new lower devices, an error was introduced for devices which
don't have the ndo_set_features() method set. Before we'll just set the
new features, but now we return an error and don't set them. Fix this by
returning the old behaviour and setting err to 0 when ndo_set_features
is not present.

Fixes: e7868a85 ("net/core: ensure features get disabled on new lower devs")
CC: Jarod Wilson <jarod@redhat.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Ido Schimmel <idosch@mellanox.com>
CC: Sander Eikelenboom <linux@eikelenboom.it>
CC: Andy Gospodarek <gospo@cumulusnetworks.com>
CC: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Reviewed-by: NJiri Pirko <jiri@mellanox.com>
Reviewed-by: NAndy Gospodarek <gospo@cumulusnetworks.com>
Reviewed-by: NJarod Wilson <jarod@redhat.com>
Tested-by: NFlorian Fainelli <f.fainelli@gmail.com>
Tested-by: NDave Young <dyoung@redhat.com>
Tested-by: NChristian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When NET_SWITCHDEV=n, switchdev_port_attr_set simply returns EOPNOTSUPP.
In this case we should not emit errors and warnings to the kernel log.
Reported-by: NSander Eikelenboom <linux@eikelenboom.it>
Tested-by: NChristian Borntraeger <borntraeger@de.ibm.com>
Fixes: 0bc05d58 ("switchdev: allow caller to explicitly request
attr_set as deferred")
Fixes: 6ac311ae ("Adding switchdev ageing notification on port
bridged")
Signed-off-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

SYNACK packets might be attached to request sockets.

Use skb_to_full_sk() helper to avoid illegal accesses to
inet_sk(skb->sk)

Fixes: ca6fb065 ("tcp: attach SYNACK messages to request sockets instead of listener")
Signed-off-by: NEric Dumazet <edumazet@google.com>
Reported-by: NSander Eikelenboom <linux@eikelenboom.it>
Acked-by: NJulian Anastasov <ja@ssi.bg>
Acked-by: NSimon Horman <horms@verge.net.au>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Some functions access TCP sockets without holding a lock and
might output non consistent data, depending on compiler and or
architecture.

tcp_diag_get_info(), tcp_get_info(), tcp_poll(), get_tcp4_sock() ...

Introduce sk_state_load() and sk_state_store() to fix the issues,
and more clearly document where this lack of locking is happening.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

now sctp auth cannot work well when setting a hmacid manually, which
is caused by that we didn't use the network order for hmacid, so fix
it by adding the transformation in sctp_auth_ep_set_hmacs.

even we set hmacid with the network order in userspace, it still
can't work, because of this condition in sctp_auth_ep_set_hmacs():

		if (id > SCTP_AUTH_HMAC_ID_MAX)
			return -EOPNOTSUPP;

so this wasn't working before and thus it won't break compatibility.

Fixes: 65b07e5d ("[SCTP]: API updates to suport SCTP-AUTH extensions.")
Signed-off-by: NXin Long <lucien.xin@gmail.com>
Signed-off-by: NMarcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Acked-by: NNeil Horman <nhorman@tuxdriver.com>
Acked-by: NVlad Yasevich <vyasevich@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Since it's introduction in commit 69e3c75f ("net: TX_RING and
packet mmap"), TX_RING could be used from SOCK_DGRAM and SOCK_RAW
side. When used with SOCK_DGRAM only, the size_max > dev->mtu +
reserve check should have reserve as 0, but currently, this is
unconditionally set (in it's original form as dev->hard_header_len).

I think this is not correct since tpacket_fill_skb() would then
take dev->mtu and dev->hard_header_len into account for SOCK_DGRAM,
the extra VLAN_HLEN could be possible in both cases. Presumably, the
reserve code was copied from packet_snd(), but later on missed the
check. Make it similar as we have it in packet_snd().

Fixes: 69e3c75f ("net: TX_RING and packet mmap")
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NWillem de Bruijn <willemb@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In case no struct sockaddr_ll has been passed to packet
socket's sendmsg() when doing a TX_RING flush run, then
skb->protocol is set to po->num instead, which is the protocol
passed via socket(2)/bind(2).

Applications only xmitting can go the path of allocating the
socket as socket(PF_PACKET, <mode>, 0) and do a bind(2) on the
TX_RING with sll_protocol of 0. That way, register_prot_hook()
is neither called on creation nor on bind time, which saves
cycles when there's no interest in capturing anyway.

That leaves us however with po->num 0 instead and therefore
the TX_RING flush run sets skb->protocol to 0 as well. Eric
reported that this leads to problems when using tools like
trafgen over bonding device. I.e. the bonding's hash function
could invoke the kernel's flow dissector, which depends on
skb->protocol being properly set. In the current situation, all
the traffic is then directed to a single slave.

Fix it up by inferring skb->protocol from the Ethernet header
when not set and we have ARPHRD_ETHER device type. This is only
done in case of SOCK_RAW and where we have a dev->hard_header_len
length. In case of ARPHRD_ETHER devices, this is guaranteed to
cover ETH_HLEN, and therefore being accessed on the skb after
the skb_store_bits().
Reported-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NWillem de Bruijn <willemb@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Packet sockets can be used by various net devices and are not
really restricted to ARPHRD_ETHER device types. However, when
currently checking for the extra 4 bytes that can be transmitted
in VLAN case, our assumption is that we generally probe on
ARPHRD_ETHER devices. Therefore, before looking into Ethernet
header, check the device type first.

This also fixes the issue where non-ARPHRD_ETHER devices could
have no dev->hard_header_len in TX_RING SOCK_RAW case, and thus
the check would test unfilled linear part of the skb (instead
of non-linear).

Fixes: 57f89bfa ("network: Allow af_packet to transmit +4 bytes for VLAN packets.")
Fixes: 52f1454f ("packet: allow to transmit +4 byte in TX_RING slot for VLAN case")
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NWillem de Bruijn <willemb@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We concluded that the skb_probe_transport_header() should better be
called unconditionally. Avoiding the call into the flow dissector has
also not really much to do with the direct xmit mode.

While it seems that only virtio_net code makes use of GSO from non
RX/TX ring packet socket paths, we should probe for a transport header
nevertheless before they hit devices.

Reference: http://thread.gmane.org/gmane.linux.network/386173/Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>