When an ICMP ICMP_FRAG_NEEDED (or ICMPV6_PKT_TOOBIG) message finds a
LISTEN socket, and this socket is currently owned by the user, we
set TCP_MTU_REDUCED_DEFERRED flag in listener tsq_flags.

This is bad because if we clone the parent before it had a chance to
clear the flag, the child inherits the tsq_flags value, and next
tcp_release_cb() on the child will decrement sk_refcnt.

Result is that we might free a live TCP socket, as reported by
Dormando.

IPv4: Attempt to release TCP socket in state 1

Fix this issue by testing sk_state against TCP_LISTEN early, so that we
set TCP_MTU_REDUCED_DEFERRED on appropriate sockets (not a LISTEN one)

This bug was introduced in commit 563d34d0
(tcp: dont drop MTU reduction indications)
Reported-by: Ndormando <dormando@rydia.net>
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The UPDATE_QSTAT function introduced on February 15, 2012
in commit 1355b704 "bnx2x: consistent statistics after
internal driver reload" incorrectly fails to handle overflow
during addition of the lower 32-bit field of a stat.

This bug is present since 3.4-rc1 and should thus be considered
a candidate for stable 3.4+ releases.

Google-Bug-Id: 8374428
Signed-off-by: NMaciej Żenczykowski <maze@google.com>
Cc: Mintz Yuval <yuvalmin@broadcom.com>
Acked-by: NEilon Greenstein <eilong@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

ubuf info allocator uses guest controlled head as an index,
so a malicious guest could put the same head entry in the ring twice,
and we will get two callbacks on the same value.
To fix use upend_idx which is guaranteed to be unique.
Reported-by: NRusty Russell <rusty@rustcorp.com.au>
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Cc: stable@kernel.org
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Most of the support was already there.  The only thing that was missing
was the call to set the flag.  Add this call.
Signed-off-by: NVlad Yasevich <vyasevic@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If the hardware initialization fails in bnx2x_nic_load() after adding
napi objects, they would not be deleted. A subsequent attempt to unload
the bnx2x module detects a corruption in the napi list.

Add the missing napi deletion to the error path.
Signed-off-by: NMichal Schmidt <mschmidt@redhat.com>
Acked-by: NDmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

John W. Linville says:

====================
On the NFC bits, Samuel says:

"With this one we have:

- A fix for properly decreasing socket ack log.
- A timer and works cleanup upon NFC device removal.
- A monitoroing socket cleanup round from llcp_socket_release.
- A proper error report to pending sockets upon NFC device removal."

Regarding the Bluetooth bits, Gustavo says:

"I have these two patches for 3.9, these add support for two more devices to
the bluetooth drivers."

Along with those, we have a few wireless driver fixes...

Bing Zhao provides an mwifiex to prevent an out-of-bounds memory
access.

John Crispin offers a Kconfig fix to enable some otherwise dead code
in rt2x00.  The correct symbols were added in -rc1 through a different
tree, but the symbols for enabling the wireless driver didn't match.

Larry Finger brings an rtlwifi fix for a scheduling while atomic bug,
and another fix for a reassociation problem caused by failing to
clear the BSSID after a disconnect.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix which was done in the following commit in cpsw driver has
to be taken forward to davinci emac driver as well.

commit d35162f8
Author: Daniel Mack <zonque@gmail.com>
Date:   Tue Mar 12 06:31:19 2013 +0000

    net: ethernet: cpsw: fix usage of cpdma_check_free_tx_desc()

    Commit fae50823 ("net: ethernet: davinci_cpdma: Add boundary for rx
    and tx descriptors") introduced a function to check the current
    allocation state of tx packets. The return value is taken into account
    to stop the netqork queue on the adapter in case there are no free
    slots.

    However, cpdma_check_free_tx_desc() returns 'true' if there is room in
    the bitmap, not 'false', so the usage of the function is wrong.
Reported-by: NPrabhakar Lad <prabhakar.csengg@gmail.com>
Tested-by: NPrabhakar Lad <prabhakar.csengg@gmail.com>
Signed-off-by: NMugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The automated ARM build tests have shown that the tulip de4x5 driver
uses the old-style virt_to_bus() interface on some architectures.

Alpha, Sparc and PowerPC did not hit this problem, because they
use a different code path, and most other architectures actually
do provide VIRT_TO_BUS.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Cc: Grant Grundler <grundler@parisc-linux.org>
Cc: netdev@vger.kernel.org
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Disabling CONFIG_VIRT_TO_BUS on ARM showed that the hisax netjet
driver depends on this deprecated functionality but is not
marked so in Kconfig.

Rather than adding ARM to the already long list of architectures
that this driver is broken on, this patch adds 'depends on
VIRT_TO_BUS' and removes the dependency on !SPARC, which is
also implied by that.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Cc: Karsten Keil <isdn@linux-pingi.de>
Cc: netdev@vger.kernel.org
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit bd329e12 ("net: cdc_ncm: do not bind to NCM compatible MBIM devices")
introduced a new policy, preferring MBIM for dual NCM/MBIM functions if
the cdc_mbim driver was enabled.  This caused a regression for users
wanting to use NCM.

Devices implementing NCM backwards compatibility according to section
3.2 of the MBIM v1.0 specification allow either NCM or MBIM on a single
USB function, using different altsettings.  The cdc_ncm and cdc_mbim
drivers will both probe such functions, and must agree on a common
policy for selecting either MBIM or NCM.  Until now, this policy has
been set at build time based on CONFIG_USB_NET_CDC_MBIM.

Use a module parameter to set the system policy at runtime, allowing the
user to prefer NCM on systems with the cdc_mbim driver.

Cc: Greg Suarez <gsuarez@smithmicro.com>
Cc: Alexey Orishko <alexey.orishko@stericsson.com>
Reported-by: NGeir Haatveit <nospam@haatveit.nu>
Reported-by: NTommi Kyntola <kynde@ts.ray.fi>
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=54791Signed-off-by: NBjørn Mork <bjorn@mork.no>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Range/validity checks on rta_type in rtnetlink_rcv_msg() do
not account for flags that may be set.  This causes the function
to return -EINVAL when flags are set on the type (for example
NLA_F_NESTED).
Signed-off-by: NVlad Yasevich <vyasevic@redhat.com>
Acked-by: NThomas Graf <tgraf@suug.ch>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This reverts commit 412ed947.

The commit is wrong as tiph points to the outer IPv4 header which is
installed at ipgre_header() and not the inner one which is protocol dependant.

This commit broke succesfully opennhrp which use PF_PACKET socket with
ETH_P_NHRP protocol. Additionally ssl_addr is set to the link-layer
IPv4 address. This address is written by ipgre_header() to the skb
earlier, and this is the IPv4 header tiph should point to - regardless
of the inner protocol payload.
Signed-off-by: NTimo Teräs <timo.teras@iki.fi>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

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

When neighbour table is full, dst_neigh_lookup/dst_neigh_lookup_skb will return
-ENOBUFS which is absolutely non zero, while all the code in kernel which use
above functions assume failure only on zero return which will cause panic. (for
example: : https://bugzilla.kernel.org/show_bug.cgi?id=54731).

This patch corrects above error with smallest changes to kernel source code and
also correct two return value check missing bugs in drivers/infiniband/hw/cxgb4/cm.c

Tested on my x86_64 SMP machine
Reported-by: NZhouyi Zhou <zhouzhouyi@gmail.com>
Tested-by: NZhouyi Zhou <zhouzhouyi@gmail.com>
Signed-off-by: NZhouyi Zhou <zhouzhouyi@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The Kconfig file help information incorrectly mentions that the
SMSC LAN75xx config option is for SMSC LAN95xx devices.
Signed-off-by: NRobert de Vries <rhdv@xs4all.nl>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Jesse Gross says:

====================
A few different bug fixes, including several for issues with userspace
communication that have gone unnoticed up until now.  These are intended
for net/3.9.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit dc975382 introduces napi support
but never calls napi_disable. This will generate a kernel oops
(kernel BUG at include/linux/netdevice.h:473!) every time, when
ndo_stop is called followed by ndo_start.
Add the missing napi_diable call.
Signed-off-by: NGeorg Hofmann <georg@hofmannsweb.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Proviously we would only restart the FEC when PHY link or duplex state
changed. PHY does not always bring down the link for speed changes, in
which case we would not detect any change and keep FEC running.

Switching link speed without restarting the FEC results in the FEC being
stuck in an indefinite state, generating error conditions for every
packet.
Signed-off-by: NLucas Stach <l.stach@pengutronix.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Hi.

I'm trying to send big chunks of memory from application address space via
TCP socket using vmsplice + splice like this

   mem = mmap(128Mb);
   vmsplice(pipe[1], mem); /* splice memory into pipe */
   splice(pipe[0], tcp_socket); /* send it into network */

When I'm lucky and a huge page splices into the pipe and then into the socket
_and_ client and server ends of the TCP connection are on the same host,
communicating via lo, the whole connection gets stuck! The sending queue
becomes full and app stops writing/splicing more into it, but the receiving
queue remains empty, and that's why.

The __skb_fill_page_desc observes a tail page of a huge page and erroneously
propagates its page->pfmemalloc value onto socket (the pfmemalloc on tail pages
contain garbage). Then this skb->pfmemalloc leaks through lo and due to the

    tcp_v4_rcv
    sk_filter
        if (skb->pfmemalloc && !sock_flag(sk, SOCK_MEMALLOC)) /* true */
            return -ENOMEM
        goto release_and_discard;

no packets reach the socket. Even TCP re-transmits are dropped by this, as skb
cloning clones the pfmemalloc flag as well.

That said, here's the proper page->pfmemalloc propagation onto socket: we
must check the huge-page's head page only, other pages' pfmemalloc and mapping
values do not contain what is expected in this place. However, I'm not sure
whether this fix is _complete_, since pfmemalloc propagation via lo also
oesn't look great.

Both, bit propagation from page to skb and this check in sk_filter, were
introduced by c48a11c7 (netvm: propagate page->pfmemalloc to skb), in v3.5 so
Mel and stable@ are in Cc.
Signed-off-by: NPavel Emelyanov <xemul@parallels.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Acked-by: NMel Gorman <mgorman@suse.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Chrome OS team reported a crash on a Pixel ChromeBook in TCP stack :

https://code.google.com/p/chromium/issues/detail?id=182056

commit a21d4572 (tcp: avoid order-1 allocations on wifi and tx
path) did a poor choice adding an 'avail_size' field to skb, while
what we really needed was a 'reserved_tailroom' one.

It would have avoided commit 22b4a4f2 (tcp: fix retransmit of
partially acked frames) and this commit.

Crash occurs because skb_split() is not aware of the 'avail_size'
management (and should not be aware)
Signed-off-by: NEric Dumazet <edumazet@google.com>
Reported-by: NMukesh Agrawal <quiche@chromium.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The driver was failing to clear the BSSID when a disconnect happened. That
prevented a reconnection. This problem is reported at
https://bugzilla.redhat.com/show_bug.cgi?id=789605,
https://bugzilla.redhat.com/show_bug.cgi?id=866786,
https://bugzilla.redhat.com/show_bug.cgi?id=906734, and
https://bugzilla.kernel.org/show_bug.cgi?id=46171.

Thanks to Jussi Kivilinna for making the critical observation
that led to the solution.
Reported-by: NJussi Kivilinna <jussi.kivilinna@iki.fi>
Tested-by: NJussi Kivilinna <jussi.kivilinna@iki.fi>
Tested-by: NAlessandro Lannocca <alessandro.lannocca@gmail.com>
Signed-off-by: NLarry Finger <Larry.Finger@lwfinger.net>
Cc: Stable <stable@vger.kernel.org>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Since v3.9-rc1 the kernel has basic support for Ralink WiSoC. The config symbols
are named slightly different than before. Fix the rt2x00 to match the new
symbols.

The commit causing this breakage is:
commit ae2b5bb6
Author: John Crispin <blogic@openwrt.org>
Date:   Sun Jan 20 22:05:30 2013 +0100
MIPS: ralink: adds Kbuild files
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Acked-by: NGertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

commit bd877e48 ("net: qmi_wwan: use a single bind function for
all device types") made Gobi 1K devices fail probing.

Using the number of endpoints in the default altsetting to decide
whether the function use one or two interfaces is wrong.  Other
altsettings may provide more endpoints.

With Gobi 1K devices, USB interface #3's altsetting is 0 by default, but
altsetting 0 only provides one interrupt endpoint and is not sufficent
for QMI.  Altsetting 1 provides all 3 endpoints required for qmi_wwan
and works with QMI. Gobi 1K layout for intf#3 is:

    Interface Descriptor:  255/255/255
      bInterfaceNumber        3
      bAlternateSetting       0
      Endpoint Descriptor:  Interrupt IN
    Interface Descriptor:  255/255/255
      bInterfaceNumber        3
      bAlternateSetting       1
      Endpoint Descriptor:  Interrupt IN
      Endpoint Descriptor:  Bulk IN
      Endpoint Descriptor:  Bulk OUT

Prior to commit bd877e48, we would call usbnet_get_endpoints
before giving up finding enough endpoints. Removing the early
endpoint number test and the strict functional descriptor
requirement allow qmi_wwan_bind to continue until
usbnet_get_endpoints has made the final attempt to collect
endpoints.  This restores the behaviour from before commit
bd877e48 without losing the added benefit of using a single bind
function.

The driver has always required a CDC Union functional descriptor
for two-interface functions. Using the existence of this
descriptor to detect two-interface functions is the logically
correct method.
Reported-by: NDan Williams <dcbw@redhat.com>
Signed-off-by: NBjørn Mork <bjorn@mork.no>
Tested-by: NDan Williams <dcbw@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

a long time ago by the commit

  commit 93456b6d
  Author: Denis V. Lunev <den@openvz.org>
  Date:   Thu Jan 10 03:23:38 2008 -0800

    [IPV4]: Unify access to the routing tables.

the defenition of FIB_HASH_TABLE size has obtained wrong dependency:
it should depend upon CONFIG_IP_MULTIPLE_TABLES (as was in the original
code) but it was depended from CONFIG_IP_ROUTE_MULTIPATH

This patch returns the situation to the original state.

The problem was spotted by Tingwei Liu.
Signed-off-by: NDenis V. Lunev <den@openvz.org>
CC: Tingwei Liu <tingw.liu@gmail.com>
CC: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

sctp_assoc_lookup_tsn() function searchs which transport a certain TSN
was sent on, if not found in the active_path transport, then go search
all the other transports in the peer's transport_addr_list, however, we
should continue to the next entry rather than break the loop when meet
the active_path transport.
Signed-off-by: NXufeng Zhang <xufeng.zhang@windriver.com>
Acked-by: NNeil Horman <nhorman@tuxdriver.com>
Acked-by: NVlad Yasevich <vyasevich@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When SCTP is done processing a duplicate cookie chunk, it tries
to delete a newly created association.  For that, it has to set
the right association for the side-effect processing to work.
However, when it uses the SCTP_CMD_NEW_ASOC command, that performs
more work then really needed (like hashing the associationa and
assigning it an id) and there is no point to do that only to
delete the association as a next step.  In fact, it also creates
an impossible condition where an association may be found by
the getsockopt() call, and that association is empty.  This
causes a crash in some sctp getsockopts.

The solution is rather simple.  We simply use SCTP_CMD_SET_ASOC
command that doesn't have all the overhead and does exactly
what we need.
Reported-by: NKarl Heiss <kheiss@gmail.com>
Tested-by: NKarl Heiss <kheiss@gmail.com>
CC: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: NVlad Yasevich <vyasevich@gmail.com>
Acked-by: NNeil Horman <nhorman@tuxdriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit d13ba512 ("tg3: Remove
SPEED_UNKNOWN checks") cleaned up the autoneg advertisement by
removing some dead code. One effect of this change was that the
advertisement register would not be updated if autoneg is turned off.

This exposed a bug on the 5715 device w.r.t linking. The 5715 defaults
to advertise only 10Mb Full duplex. But with autoneg disabled, it needs
the configured speed enabled in the advertisement register to link up.

This patch adds the work around to advertise all speeds on the 5715 when
autoneg is disabled.
Reported-by: NMarcin Miotk <marcinmiotk81@gmail.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NNithin Nayak Sujir <nsujir@broadcom.com>
Signed-off-by: NMichael Chan <mchan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Ben Hutchings says:

====================
Just the one bug fix I mentioned before, but it's a pretty important one
as it can cause silent data corruption or IOMMU page faults.

This would be suitable for stable and should apply cleanly to all the
3.x.y branches.  I'm still working through testing of larger sets of
fixes.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

bond_update_speed_duplex() might sleep while calling underlying slave's
routines. Move it out of atomic context in bond_enslave() and remove it
from bond_miimon_commit() - it was introduced by commit 546add79, however
when the slave interfaces go up/change state it's their responsibility to
fire NETDEV_UP/NETDEV_CHANGE events so that bonding can properly update
their speed.

I've tested it on all combinations of ifup/ifdown, autoneg/speed/duplex
changes, remote-controlled and local, on (not) MII-based cards. All changes
are visible.
Signed-off-by: NVeaceslav Falico <vfalico@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit fae50823 ("net: ethernet: davinci_cpdma: Add boundary for rx
and tx descriptors") introduced a function to check the current
allocation state of tx packets. The return value is taken into account
to stop the netqork queue on the adapter in case there are no free
slots.

However, cpdma_check_free_tx_desc() returns 'true' if there is room in
the bitmap, not 'false', so the usage of the function is wrong.
Signed-off-by: NDaniel Mack <zonque@gmail.com>
Cc: Mugunthan V N <mugunthanvnm@ti.com>
Reported-by: NSven Neumann <s.neumann@raumfeld.com>
Reported-by: NAndreas Fenkart <andreas.fenkart@streamunlimited.com>
Tested-by: NMugunthan V N <mugunthanvnm@ti.com>
Acked-by: NMugunthan V N <mugunthanvnm@ti.com>
Tested-by: NAndreas Fenkart <andreas.fenkart@streamunlimited.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Using TX push when notifying the NIC of multiple new descriptors in
the ring will very occasionally cause the TX DMA engine to re-use an
old descriptor.  This can result in a duplicated or partly duplicated
packet (new headers with old data), or an IOMMU page fault.  This does
not happen when the pushed descriptor is the only one written.

TX push also provides little latency benefit when a packet requires
more than one descriptor.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

[ Bug added added in commit 05e8ef4a (net: factor out
  skb_mac_gso_segment() from skb_gso_segment() ) ]

move vlan_depth out of while loop, or else vlan_depth always is ETH_HLEN,
can not be increased, and lead to infinite loop when frame has two vlan headers.
Signed-off-by: NLi RongQing <roy.qing.li@gmail.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

  k is u32 which never < 0, need type cast, or cause issue.
Signed-off-by: NChen Gang <gang.chen@asianux.com>
Acked-by: NRussell King <rmk+kernel@arm.linux.org.uk>
Acked-by: NMircea Gherzan <mgherzan@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

__netpoll_cleanup() is called in netconsole_netdev_event() while holding a
spinlock. Release/acquire the spinlock before/after it and restart the
loop. Also, disable the netconsole completely, because we won't have chance
after the restart of the loop, and might end up in a situation where
nt->enabled == 1 and nt->np.dev == NULL.
Signed-off-by: NVeaceslav Falico <vfalico@redhat.com>
Acked-by: NNeil Horman <nhorman@tuxdriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The bridge multicast fast leave feature was added sufficient space
was not reserved in the netlink message. This means the flag may be
lost in netlink events and results of queries.

Found by observation while looking up some netlink stuff for discussion with Vlad.
Problem introduced by commit c2d3babf
Author: David S. Miller <davem@davemloft.net>
Date:   Wed Dec 5 16:24:45 2012 -0500

    bridge: implement multicast fast leave
Signed-off-by: NStephen Hemminger <stephen@networkplumber.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Included changes ares:
- fix packet parsing routine to avoid to read beyond the packet boundary
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This is needed in order to detect if the timestamp option appears
more than once in a packet, to remove the option if the packet is
fragmented, etc. My previous change neglected to store the option
location when the router addresses were prespecified and Pointer >
Length. But now the option location is also stored when Flag is an
unrecognized value, to ensure these option handling behaviors are
still performed.
Signed-off-by: NDavid Ward <david.ward@ll.mit.edu>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

New scheme calls for 3rd party VPD at offset 0x0 and Chelsio VPD at offset
0x400 of the function.  If no 3rd party VPD is present, then a copy of
Chelsio's VPD will be at offset 0x0 to keep in line with PCI spec which
requires the VPD to be present at offset 0x0.
Signed-off-by: NSantosh Rastapur <santosh@chelsio.com>
Signed-off-by: NVipul Pandya <vipul@chelsio.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

batadv_iv_ogm_process() accesses the packet using the tt_num_changes
attribute regardless of the real packet len (assuming the length check
was done before). Therefore a length check is needed to avoid reading
random memory.
Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>
Signed-off-by: NAntonio Quartulli <ordex@autistici.org>