Roman Gushchin discovered that udp4_lib_lookup2() was not reloading
first item in the rcu protected list, in case the loop was restarted.

This produced soft lockups as in https://lkml.org/lkml/2013/4/16/37

rcu_dereference(X)/ACCESS_ONCE(X) seem to not work as intended if X is
ptr->field :

In some cases, gcc caches the value or ptr->field in a register.

Use a barrier() to disallow such caching, as documented in
Documentation/atomic_ops.txt line 114

Thanks a lot to Roman for providing analysis and numerous patches.
Diagnosed-by: NRoman Gushchin <klamm@yandex-team.ru>
Signed-off-by: NEric Dumazet <edumazet@google.com>
Reported-by: NBoris Zhmurov <zhmurov@yandex-team.ru>
Signed-off-by: NRoman Gushchin <klamm@yandex-team.ru>
Acked-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Since the recent addition of 8021AD, we need to set the new field vlan_proto in
sk_buff. Otherwise, it will trigger BUG() call in vlan_proto_idx().

This patch fixes the problem.
Signed-off-by: NHaiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: NK. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In case the port list dump does not fit into one skb currently the
dump would start over again. Fix this by continue from the last dumped
port.

Introduced by commit d90f889e (team: handle sending port list in the
same way option list is sent)
Signed-off-by: NJiri Pirko <jiri@resnulli.us>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

non-rcu variant of list_first_or_null_rcu
Signed-off-by: NJiri Pirko <jiri@resnulli.us>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The SELinux labeled IPsec code was improperly handling its reference
counting, dropping a reference on a delete operation instead of on a
free/release operation.
Reported-by: NOndrej Moris <omoris@redhat.com>
Signed-off-by: NPaul Moore <pmoore@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In some cases after deleting a policy from the SPD the policy would
remain in the dst/flow/route cache for an extended period of time
which caused problems for SELinux as its dynamic network access
controls key off of the number of XFRM policy and state entries.
This patch corrects this problem by forcing a XFRM garbage collection
whenever a policy is sucessfully removed.
Reported-by: NOndrej Moris <omoris@redhat.com>
Signed-off-by: NPaul Moore <pmoore@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

udp6 over GRE tunnel does not work after to GRE tso changes. GRE
tso handler passes inner packet but keeps track of outer header
start in SKB_GSO_CB(skb)->mac_offset.  udp6 fragment need to
take care of outer header, which start at the mac_offset, while
adding fragment header.
This bug is introduced by commit 68c33163 (GRE: Add TCP
segmentation offload for GRE).
Reported-by: NDmitry Kravkov <dkravkov@gmail.com>
Signed-off-by: NPravin B Shelar <pshelar@nicira.com>
Tested-by: NDmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The dev_mc_sync_multiple function is currently calling
__hw_addr_sync, and not __hw_addr_sync_multiple.  This will result in
addresses only being synced to the first device from the set.

	Corrected by calling the _multiple variant.
Signed-off-by: NJay Vosburgh <fubar@us.ibm.com>
Reviewed-by: NVlad Yasevich <vyasevic@redhat.com>
Tested-by: NShawn Bohrer <sbohrer@rgmadvisors.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently, __hw_addr_sync_one is called in a loop by
__hw_addr_sync_multiple to sync each of a "from" device's hw addresses
to a "to" device.  __hw_addr_sync_one calls __hw_addr_add_ex to attempt
to add each address.  __hw_addr_add_ex is called with global=false, and
sync=true.

	__hw_addr_add_ex checks to see if the new address matches an
address already on the list.  If so, it tests global and sync.  In this
case, sync=true, and it then checks if the address is already synced,
and if so, returns 0.

	This 0 return causes __hw_addr_sync_one to increment the sync_cnt
and refcount for the "from" list's address entry, even though the address
is already synced and has a reference and sync_cnt.  This will cause
the sync_cnt and refcount to increment without bound every time an
addresses is added to the "from" device and synced to the "to" device.

	The fix here has two parts:

	First, when __hw_addr_add_ex finds the address already exists
and is synced, return -EEXIST instead of 0.

	Second, __hw_addr_sync_one checks the error return for -EEXIST,
and if so, it (a) does not add a refcount/sync_cnt, and (b) returns 0
itself so that __hw_addr_sync_multiple will not return an error.
Signed-off-by: NJay Vosburgh <fubar@us.ibm.com>
Reviewed-by: NVlad Yasevich <vyasevic@redhat.com>
Tested-by: NShawn Bohrer <sbohrer@rgmadvisors.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When an address is added to a subordinate interface (the "to"
list), the address entry in the "from" list is not marked "synced" as
the entry added to the "to" list is.

	When performing the unsync operation (e.g., dev_mc_unsync),
__hw_addr_unsync_one calls __hw_addr_del_entry with the "synced"
parameter set to true for the case when the address reference is being
released from the "from" list.  This causes a test inside to fail,
with the result being that the reference count on the "from" address
is not properly decremeted and the address on the "from" list will
never be freed.

	Correct this by having __hw_addr_unsync_one call the
__hw_addr_del_entry function with the "sync" flag set to false for the
"remove from the from list" case.
Signed-off-by: NJay Vosburgh <fubar@us.ibm.com>
Reviewed-by: NVlad Yasevich <vyasevic@redhat.com>
Tested-by: NShawn Bohrer <sbohrer@rgmadvisors.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The sync_cnt field is not being initialized, which can result
in arbitrary values in the field.  Fixed by initializing it to zero.
Signed-off-by: NJay Vosburgh <fubar@us.ibm.com>
Reviewed-by: NVlad Yasevich <vyasevic@redhat.com>
Tested-by: NShawn Bohrer <sbohrer@rgmadvisors.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This stat is not relevant in IPv6, there is no checksum in IPv6 header.
Just leave a comment to explain the hole.
Signed-off-by: NNicolas Dichtel <nicolas.dichtel@6wind.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

John W. Linville says:

====================
Please pull this batch of fixes intended for the 3.10 stream...

Regarding the NFC bits, Samuel says:

"This is the first batch of NFC fixes for 3.10, and it contains:

- 3 fixes for the NFC MEI support:
        * We now depend on the correct Kconfig symbol.
        * We register an MEI event callback whenever we enable an NFC device,
          otherwise we fail to read anything after an enable/disable cycle.
        * We only disable an MEI device from its disable mey_phy_ops,
          preventing useless consecutive disable calls.

- An NFC Makefile cleanup, as I forgot to remove a commented out line when
  moving the LLCP code to the NFC top level directory."

As for the mac80211 bits, Johannes says:

"This time I have a fix from Stanislaw for a stupid mistake I made in the
auth/assoc timeout changes, a fix from Felix for 64-bit traffic counters
and one from Helmut for address mask handling in mac80211. I also have a
few fixes myself for four different crashes reported by a few people."

And Johannes says this about the iwlwifi bit:

"This fixes a brown paper-bag bug that we really should've caught in
review. More details in the changelog for the fix."

On top of that...

Arend van Spriel and Hante Meuleman cooperate to send a series of AP
and P2P mode fixes for brcmfmac.

Gabor Juhos corrects a register offset for AR9550, avoiding a bus
error.

Dan Carpenter provides a fixup to some dmesg output in the atmel
driver.

And, finally...

Felix Fietkau not only gives us a trio of small AR934x fixes, but
also refactors the ath9k aggregation session start/stop handling
(using the generic mac80211 support) in order to avoid a deadlock.

Please let me know if there are problems!
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

During a PCI EEH/AER error recovery flow, if the device did not successfully
restart, the error_detected() hook may be called a second time with a
"perm_failure" state. This patch skips over driver cleanup for the second
invocation of the callback.

Also, Lancer error recovery code is fixed-up to handle these changes.
Signed-off-by: NKalesh AP <kalesh.purayil@emulex.com>
Signed-off-by: NSomnath kotur <somnath.kotur@emulex.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

HW does not compute L4 checksum for IP Fragmented packets.
Signed-off-by: NKalesh AP <kalesh.purayil@emulex.com>
Signed-off-by: NSomnath Kotur <somnath.kotur@emulex.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Pablo Neira Ayuso says:

====================
The following patchset contains Netfilter/IPVS fixes for 3.10-rc3,
they are:

* fix xt_addrtype with IPv6, from Florian Westphal. This required
  a new hook for IPv6 functions in the netfilter core to avoid
  hard dependencies with the ipv6 subsystem when this match is
  only used for IPv4.

* fix connection reuse case in IPVS. Currently, if an reused
  connection are directed to the same server. If that server is
  down, those connection would fail. Therefore, clear the
  connection and choose a new server among the available ones.

* fix possible non-nul terminated string sent to user-space if
  ipt_ULOG is used as the default netfilter logging stub, from
  Chen Gang.

* fix mark logging of IPv6 packets in xt_LOG, from Michal Kubecek.
  This bug has been there since 2.6.26.

* Fix breakage ip_vs_sh due to incorrect structure layout for
  RCU, from Jan Beulich.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

kfree_rcu() requires offsetof(..., rcu_head) < 4096, which can
get violated with a sufficiently high CONFIG_IP_VS_SH_TAB_BITS.
Signed-off-by: NJan Beulich <jbeulich@suse.com>
Signed-off-by: NSimon Horman <horms@verge.net.au>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless into for-davem

In dump_ipv6_packet(), the "recurse" parameter is zero only if
dumping contents of a packet embedded into an ICMPv6 error
message. Therefore we want to log packet mark if recurse is
non-zero, not when it is zero.
Signed-off-by: NMichal Kubecek <mkubecek@suse.cz>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

We currently allow changing the mq flag (IFF_MULTI_QUEUE) for a persistent
device. This will result a mismatch between the number the queues in netdev and
tuntap. This is because we only allocate a 1q netdevice when IFF_MULTI_QUEUE was
not specified, so when we set the IFF_MULTI_QUEUE and try to attach more queues
later, netif_set_real_num_tx_queues() may fail which result a single queue
netdevice with multiple sockets attached.

Solve this by disallowing changing the mq flag for persistent device.

Bug was introduced by commit edfb6a14
(tuntap: reduce memory using of queues).
Reported-by: NSriram Narasimhan <sriram.narasimhan@hp.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The three arrays of strings: af_family_key_strings,
af_family_slock_key_strings and af_family_clock_key_strings have not
VSOCK's string
Signed-off-by: NFederico Vaga <federico.vaga@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

To: linux-kernel@vger.kernel.org
Cc: x86@kernel.org, trinity@vger.kernel.org, Andy Lutomirski <luto@amacapital.net>, netdev@vger.kernel.org, "David S.
	Miller" <davem@davemloft.net>
Subject: [PATCH 5/5] net: Block MSG_CMSG_COMPAT in send(m)msg and recv(m)msg

MSG_CMSG_COMPAT is (AFAIK) not intended to be part of the API --
it's a hack that steals a bit to indicate to other networking code
that a compat entry was used.  So don't allow it from a non-compat
syscall.

This prevents an oops when running this code:

int main()
{
	int s;
	struct sockaddr_in addr;
	struct msghdr *hdr;

	char *highpage = mmap((void*)(TASK_SIZE_MAX - 4096), 4096,
	                      PROT_READ | PROT_WRITE,
	                      MAP_PRIVATE | MAP_ANONYMOUS | MAP_FIXED, -1, 0);
	if (highpage == MAP_FAILED)
		err(1, "mmap");

	s = socket(AF_INET, SOCK_DGRAM, IPPROTO_UDP);
	if (s == -1)
		err(1, "socket");

        addr.sin_family = AF_INET;
        addr.sin_port = htons(1);
        addr.sin_addr.s_addr = htonl(INADDR_LOOPBACK);
	if (connect(s, (struct sockaddr*)&addr, sizeof(addr)) != 0)
		err(1, "connect");

	void *evil = highpage + 4096 - COMPAT_MSGHDR_SIZE;
	printf("Evil address is %p\n", evil);

	if (syscall(__NR_sendmmsg, s, evil, 1, MSG_CMSG_COMPAT) < 0)
		err(1, "sendmmsg");

	return 0;
}

Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: NAndy Lutomirski <luto@amacapital.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Accessing the OTP memory on AR9950 causes a data bus
like this:

  Data bus error, epc == 801f7774, ra == 801f7774
  Oops[#1]:
  CPU: 0 PID: 1 Comm: swapper Not tainted 3.10.0-rc3 #592
  task: 87c28000 ti: 87c22000 task.ti: 87c22000
  $ 0   : 00000000 00000061 deadc0de 00000000
  $ 4   : b8115f18 00015f18 00000007 00000004
  $ 8   : 00000001 7c7c3c7c 7c7c7c7c 7c7c7c7c
  $12   : 7c7c3c7c 80320a68 00000000 7c7c7c3c
  $16   : 87cd8010 00015f18 00000007 00000000
  $20   : 00000064 00000004 87c23c7c 8035210c
  $24   : 00000000 801f3674
  $28   : 87c22000 87c23b48 00000001 801f7774
  Hi    : 00000000
  Lo    : 00000064
  epc   : 801f7774 ath9k_hw_wait+0x58/0xb0
      Not tainted
  ra    : 801f7774 ath9k_hw_wait+0x58/0xb0
  Status: 1000cc03 KERNEL EXL IE
  Cause : 4080801c
  PrId  : 00019750 (MIPS 74Kc)
  Modules linked in:
  Process swapper (pid: 1, threadinfo=87c22000, task=87c28000, ts=00000000)
  Stack : 0000000f 00000061 00002710 8006240c 00000001 87cd8010 87c23bb0 87cd8010
          00000000 00000004 00000003 80210c7c 000000b3 67fa8000 0000032a 000006fe
          000003e8 00000002 00000028 87c23bf0 000003ff 80210d24 803e5630 80210e28
          00000000 00000007 87cd8010 00007044 00000004 00000061 000003ff 000001ff
          87c26000 87cd8010 00000220 87cd8bb8 80210000 8020fcf4 87c22000 87c23c08
          ...
  Call Trace:
  [<801f7774>] ath9k_hw_wait+0x58/0xb0
  [<80210c7c>] ar9300_otp_read_word+0x80/0xd4
  [<80210d24>] ar9300_read_otp+0x54/0xb0
  [<8020fcf4>] ar9300_check_eeprom_header+0x1c/0x40
  [<80210fe4>] ath9k_hw_ar9300_fill_eeprom+0x118/0x39c
  [<80206650>] ath9k_hw_eeprom_init+0x74/0xb4
  [<801f96d0>] ath9k_hw_init+0x7ec/0x96c
  [<801e65ec>] ath9k_init_device+0x340/0x758
  [<801f35d0>] ath_ahb_probe+0x21c/0x2c0
  [<801c041c>] driver_probe_device+0xc0/0x1e4
  [<801c05ac>] __driver_attach+0x6c/0xa4
  [<801bea08>] bus_for_each_dev+0x64/0xa8
  [<801bfa40>] bus_add_driver+0xcc/0x24c
  [<801c0954>] driver_register+0xbc/0x17c
  [<803f8fc0>] ath9k_init+0x5c/0x88
  [<800608fc>] do_one_initcall+0xec/0x1a0
  [<803e6a68>] kernel_init_freeable+0x13c/0x200
  [<80309cdc>] kernel_init+0x1c/0xe4
  [<80062450>] ret_from_kernel_thread+0x10/0x18

On the AR9550, the OTP registers are located at
the same address as on the AR9340. Use the correct
values to avoid the error.

Cc: stable@vger.kernel.org  # 3.6+
Signed-off-by: NGabor Juhos <juhosg@openwrt.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

For p2p client mode powersave mode should be kept disabled. It is
working but inefficient. In general p2p links do no benefit from this
mode, because these links are setup temporarily to transfer data.
Reviewed-by: NArend Van Spriel <arend@broadcom.com>
Signed-off-by: NHante Meuleman <meuleman@broadcom.com>
Signed-off-by: NArend van Spriel <arend@broadcom.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Multi channel support was disabled. This patch will enable it and
configure the P2P GO on the correct frequency when multi channel
is used.
Reviewed-by: NArend Van Spriel <arend@broadcom.com>
Signed-off-by: NHante Meuleman <meuleman@broadcom.com>
Signed-off-by: NArend van Spriel <arend@broadcom.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Upon deleting a P2P_CLIENT/GO interface the vif and consequently
the wdev is freed before the net_device is actually being unregistered
but cfg80211 still needs to access the wdev. Using destructor field
to free the net_device and vif.
Reviewed-by: NHante Meuleman <meuleman@broadcom.com>
Signed-off-by: NArend van Spriel <arend@broadcom.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

When registration fails the net device is no longer needed. Free
the net device and remove reference to private data from the
driver.
Reviewed-by: NHante Meuleman <meuleman@broadcom.com>
Signed-off-by: NArend van Spriel <arend@broadcom.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Pass the struct brcmf_cfg80211_info instance instead of obtaining
through vif itself using vif->wdev. This is needed as the netdev
associated with this vif is already unregistered.
Reviewed-by: NHante Meuleman <meuleman@broadcom.com>
Signed-off-by: NArend van Spriel <arend@broadcom.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

The firmware requires that on p2p setup when net interfaces
are created or updated that they start initially with the same
channel as the channel in use for the current connection
(if any). If none exists take default channel 11.
Reviewed-by: NArend Van Spriel <arend@broadcom.com>
Reviewed-by: NFranky (Zhenhui) Lin <frankyl@broadcom.com>
Reviewed-by: NPieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: NHante Meuleman <meuleman@broadcom.com>
Signed-off-by: NArend van Spriel <arend@broadcom.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

ARP offloading should only be used in STA or P2P client mode. It
is currently configured once at init. When being configured for AP
ARP offloading should be turned off and when AP mode is left it can
be turned back on.

Cc: stable@vger.kernel.org
Reviewed-by: NArend Van Spriel <arend@broadcom.com>
Signed-off-by: NHante Meuleman <meuleman@broadcom.com>
Signed-off-by: NArend van Spriel <arend@broadcom.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Unlike ipv4_redirect() and ipv4_sk_redirect(), ip_do_redirect()
doesn't call __build_flow_key() directly but via
ip_rt_build_flow_key() wrapper. This leads to __build_flow_key()
getting pointer to IPv4 header of the ICMP redirect packet
rather than pointer to the embedded IPv4 header of the packet
initiating the redirect.

As a result, handling of ICMP redirects initiated by TCP packets
is broken. Issue was introduced by

	4895c771 ("ipv4: Add FIB nexthop exceptions.")
Signed-off-by: NMichal Kubecek <mkubecek@suse.cz>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The phy_init_eee has to exit with an error when the
local device and its link partner both do not support EEE.
So this patch fixes a problem when verify this.
Signed-off-by: NGiuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Expire cached connection for new TCP/SCTP connection if real
server is down. Otherwise, IPVS uses the dead server for the
reused connection, instead of a new working one.
Signed-off-by: NGrzegorz Lyczba <grzegorz.lyczba@gmail.com>
Acked-by: NHans Schillstrom <hans@schillstrom.com>
Acked-by: NJulian Anastasov <ja@ssi.bg>
Signed-off-by: NSimon Horman <horms@verge.net.au>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

In 63b77bf4

	iwlwifi: dvm: don't send zeroed LQ cmd

I tried to avoid to send zeroed LQ cmd, but I made a (very)
stupid mistake in the memcmp.
Since this patch has been ported to stable, the fix should
go to stable too.

This fixes https://bugzilla.kernel.org/show_bug.cgi?id=58341

Cc: stable@vger.kernel.org
Reported-by: NHinnerk van Bruinehsen <h.v.bruinehsen@fu-berlin.de>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

When changing the MAC address of a single vif mac80211 will check if
the new address fits into the address mask specified by the driver.
This only needs to be done when using multiple BSSIDs. Hence, check
the new address only against all other vifs.

Also fix the MAC address assignment on new interfaces if the user
changed the address of a vif such that perm_addr is not covered by
addr_mask anymore.

Resolves:
https://bugzilla.kernel.org/show_bug.cgi?id=57371Signed-off-by: NHelmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: NJakub Kicinski <kubakici@wp.pl>
Reported-by: NAlessandro Lannocca <alessandro.lannocca@gmail.com>
Cc: Alessandro Lannocca <alessandro.lannocca@gmail.com>
Cc: Bruno Randolf <br1@thinktube.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Since Eric's commit efe117ab ("Speedup ieee80211_remove_interfaces")
there's a bug in mac80211 when it unregisters with AP_VLAN interfaces
up. If the AP_VLAN interface was registered after the AP it belongs
to (which is the typical case) and then we get into this code path,
unregister_netdevice_many() will crash because it isn't prepared to
deal with interfaces being closed in the middle of it. Exactly this
happens though, because we iterate the list, find the AP master this
AP_VLAN belongs to and dev_close() the dependent VLANs. After this,
unregister_netdevice_many() won't pick up the fact that the AP_VLAN
is already down and will do it again, causing a crash.

Cc: stable@vger.kernel.org [2.6.33+]
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

A lot of code in mac80211 assumes that the hw queues are
set up correctly for all interfaces (except for monitor)
but this isn't true for AP_VLAN interfaces. Fix this by
copying the AP master configuration when an AP VLAN is
brought up, after this the AP interface can't change its
configuration any more and needs to be brought down to
change it, which also forces AP_VLAN interfaces down, so
just copying in open() is sufficient.
Reported-by: NJouni Malinen <j@w1.fi>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Daniel Petre reported crashes in icmp_dst_unreach() with following call
graph:

#3 [ffff88003fc03938] __stack_chk_fail at ffffffff81037f77
#4 [ffff88003fc03948] icmp_send at ffffffff814d5fec
#5 [ffff88003fc03ae8] ipv4_link_failure at ffffffff814a1795
#6 [ffff88003fc03af8] ipgre_tunnel_xmit at ffffffff814e7965
#7 [ffff88003fc03b78] dev_hard_start_xmit at ffffffff8146e032
#8 [ffff88003fc03bc8] sch_direct_xmit at ffffffff81487d66
#9 [ffff88003fc03c08] __qdisc_run at ffffffff81487efd
#10 [ffff88003fc03c48] dev_queue_xmit at ffffffff8146e5a7
#11 [ffff88003fc03c88] ip_finish_output at ffffffff814ab596

Daniel found a similar problem mentioned in
 http://lkml.indiana.edu/hypermail/linux/kernel/1007.0/00961.html

And indeed this is the root cause : skb->cb[] contains data fooling IP
stack.

We must clear IPCB in ip_tunnel_xmit() sooner in case dst_link_failure()
is called. Or else skb->cb[] might contain garbage from GSO segmentation
layer.

A similar fix was tested on linux-3.9, but gre code was refactored in
linux-3.10. I'll send patches for stable kernels as well.

Many thanks to Daniel for providing reports, patches and testing !
Reported-by: NDaniel Petre <daniel.petre@rcs-rds.ro>
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>