Some network drivers use a non default hard_header_len

Transmitted skb should take into account dev->hard_header_len, or risk
crashes or expensive reallocations.

In the case of aoe, lets reserve MAX_HEADER bytes.

David reported a crash in defxx driver, solved by this patch.
Reported-by: David Oostdyk <daveo@ll.mit.edu>
Tested-by: David Oostdyk <daveo@ll.mit.edu>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Ed Cashin <ecashin@coraid.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Marc Kleine-Budde says:

====================
here's a patch series for net for the v3.9 release cycle. Fengguang Wu found
two problems with the sja1000 drivers:

A macro in the SH architecture collides with one in the sja1000 driver. I
created a minimal patch suited for stable, only changing this particular
define. (Once net is merged back to net-next, I'll post a patch to uniformly
use a SJA1000_ prefix for the sja100 private defines.) If you prefer, I can
squash both patches together.

Fengguang further noticed that the peak pcmcia driver will not compile on archs
without ioport support. I created a patch to limit the driver to archs which
select HAS_IOPORT in Kconfig.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Jeff Kirsher says:

====================
This series contains updates to e1000, ixgb and e1000e for Christoph.

Christoph provides 3 patches to resolve missing dma_error_call's to
provided Intel drivers which did not have this fix.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Steffen Klassert says:

====================
1) Initialize the satype field in key_notify_policy_flush(),
   this was left uninitialized. From Nicolas Dichtel.

2) The sequence number difference for replay notifications
   was misscalculated on ESN sequence number wrap. We need
   a separate replay notify function for esn.

3) Fix an off by one in the esn replay notify function.
   From Mathias Krause.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit 184b8904 ("tg3: Use VPD fw version
when present") introduced VPD parsing that contained a potential length
overflow.

Limit the hardware's reported firmware string length (max 255 bytes) to
stay inside the driver's firmware string length (32 bytes). On overflow,
truncate the formatted firmware string instead of potentially overwriting
portions of the tg3 struct.

http://cansecwest.com/slides/2013/PrivateCore%20CSW%202013.pdfSigned-off-by: Kees Cook <keescook@chromium.org>
Reported-by: Oded Horovitz <oded@privatecore.com>
Reported-by: Brad Spengler <spender@grsecurity.net>
Cc: stable@vger.kernel.org
Cc: Matt Carlson <mcarlson@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

It seems that commit

commit 292f1c7f
Author: Jiri Pirko <jiri@resnulli.us>
Date:   Tue Feb 12 00:12:03 2013 +0000

    sch: make htb_rate_cfg and functions around that generic

adds little regression.

Before:

# tc qdisc add dev eth0 root handle 1: htb default ffff
# tc class add dev eth0 classid 1:ffff htb rate 5Gbit
# tc -s class show dev eth0
class htb 1:ffff root prio 0 rate 5000Mbit ceil 5000Mbit burst 625b cburst
625b
 Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
 rate 0bit 0pps backlog 0b 0p requeues 0
 lended: 0 borrowed: 0 giants: 0
 tokens: 31 ctokens: 31

After:

# tc qdisc add dev eth0 root handle 1: htb default ffff
# tc class add dev eth0 classid 1:ffff htb rate 5Gbit
# tc -s class show dev eth0
class htb 1:ffff root prio 0 rate 1544Mbit ceil 1544Mbit burst 625b cburst
625b
 Sent 5073 bytes 41 pkt (dropped 0, overlimits 0 requeues 0)
 rate 1976bit 2pps backlog 0b 0p requeues 0
 lended: 41 borrowed: 0 giants: 0
 tokens: 1802 ctokens: 1802

This probably due to lost u64 cast of rate parameter in
psched_ratecfg_precompute() (net/sched/sch_generic.c).
Signed-off-by: Sergey Popovich <popovich_sergei@mail.ru>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

To restart tx queue use netif_wake_queue() intead of netif_start_queue()
so that net schedule will restart transmission immediately which will
increase network performance while doing huge data transfers.
Reported-by: Dan Franke <dan.franke@schneider-electric.com>
Suggested-by: Sriramakrishnan A G <srk@ti.com>
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

To restart tx queue use netif_wake_queue() intead of netif_start_queue()
so that net schedule will restart transmission immediately which will
increase network performance while doing huge data transfers.
Reported-by: Dan Franke <dan.franke@schneider-electric.com>
Suggested-by: Sriramakrishnan A G <srk@ti.com>
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Fix to return a negative error code from the error handling case
instead of 0(possible overwrite to 0 by ops->fill_xstats call),
as returned elsewhere in this function.
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>

Without this patch
1. boot with nfs (no_console_suspend)
2. echo mem >/sys/power/state
3. wakeup by wakesource
4. print "eth0: tx queue full"

This fix above problem by reinit bd queue at restart function
Signed-off-by: Frank Li <Frank.Li@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Pull IOMMU fixes from Joerg Roedel:
 "Here are some fixes which have collected since Linux v3.9-rc1.

  The most important one fixes a long-standing regressen which make
  re-hotplugged devices unusable when AMD IOMMU is used.

  The other patches fix build issues (build regression on OMAP and a
  section mismatch).  One patch just removes a duplicate header include."

* tag 'iommu-fixes-v3.9-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
  iommu/amd: Make sure dma_ops are set for hotplug devices
  x86, io_apic: remove duplicated include from irq_remapping.c
  iommu: OMAP: build only on OMAP2+
  amd_iommu_init: remove __init from amd_iommu_erratum_746_workaround

Commit 06ae43f3 ("Don't bother with redoing rw_verify_area() from
default_file_splice_from()") lost the checks to test existence of the
write/aio_write methods.  My apologies ;-/

Eventually, we want that in fs/splice.c side of things (no point
repeating it for every buffer, after all), but for now this is the
obvious minimal fix.
Reported-by: Dave Jones <davej@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

This patch limits the PEAK PCAN-PC Card driver to systems with ioports. Fixes a
compile time breakage on SH:

    drivers/net/can/sja1000/peak_pcmcia.c:626:2: error:
        implicit declaration of function 'ioport_unmap'
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Cc: Paul Mundt <lethal@linux-sh.org>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

Thias patch fixes a define conflict between the SH architecture and the sja1000
driver:

    drivers/net/can/sja1000/sja1000.h:59:0: warning:
        "REG_SR" redefined [enabled by default]
    arch/sh/include/asm/ptrace_32.h:25:0: note:
         this is the location of the previous definition

A SJA1000_ prefix is added to the offending sja1000 define only, to make a
minimal patch suited for stable. A later patch will add a SJA1000_ prefix to
all defines in sja1000.h.

Cc: linux-stable <stable@vger.kernel.org>
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

After dma_map_page, dma_mapping_error must be called. It seems safe to
not free the skb/page allocated in this function, as the skb/page can be
reused later.
Signed-off-by: Christoph Paasch <christoph.paasch@uclouvain.be>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

After dma_map_single, dma_mapping_error must be called. It seems safe to
not free the skb allocated in this function, as the skb can be reused
later.

Additionally this patch fixes one coding-style error.
Signed-off-by: Christoph Paasch <christoph.paasch@uclouvain.be>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

After dma_map_single, dma_mapping_error must be called.
Signed-off-by: Christoph Paasch <christoph.paasch@uclouvain.be>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

There is a bug introduced with commit 27c2127 that causes
devices which are hot unplugged and then hot-replugged to
not have per-device dma_ops set. This causes these devices
to not function correctly. Fixed with this patch.

Cc: stable@vger.kernel.org
Reported-by: Andreas Degert <andreas.degert@googlemail.com>
Signed-off-by: Joerg Roedel <joro@8bytes.org>

Pull vfs fixes from Al Viro:
 "stable fodder; assorted deadlock fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  vt: synchronize_rcu() under spinlock is not nice...
  Nest rename_lock inside vfsmount_lock
  Don't bother with redoing rw_verify_area() from default_file_splice_from()

vcs_poll_data_free() calls unregister_vt_notifier(), which calls
atomic_notifier_chain_unregister(), which calls synchronize_rcu().
Do it *after* we'd dropped ->f_lock.

Cc: stable@vger.kernel.org (all kernels since 2.6.37)
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

... lest we get livelocks between path_is_under() and d_path() and friends.

The thing is, wrt fairness lglocks are more similar to rwsems than to rwlocks;
it is possible to have thread B spin on attempt to take lock shared while thread
A is already holding it shared, if B is on lower-numbered CPU than A and there's
a thread C spinning on attempt to take the same lock exclusive.

As the result, we need consistent ordering between vfsmount_lock (lglock) and
rename_lock (seq_lock), even though everything that takes both is going to take
vfsmount_lock only shared.
Spotted-by: Brad Spengler <spender@grsecurity.net>
Cc: stable@vger.kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

Pull networking fixes from David Miller:

 1) Always increment IPV4 ID field in encapsulated GSO packets, even
    when DF is set.  Regression fix from Pravin B Shelar.

 2) Fix per-net subsystem initialization in netfilter conntrack,
    otherwise we may access dynamically allocated memory before it is
    actually allocated.  From Gao Feng.

 3) Fix DMA buffer lengths in iwl3945 driver, from Stanislaw Gruszka.

 4) Fix race between submission of sync vs async commands in mwifiex
    driver, from Amitkumar Karwar.

 5) Add missing cancel of command timer in mwifiex driver, from Bing
    Zhao.

 6) Missing SKB free in rtlwifi USB driver, from Jussi Kivilinna.

 7) Thermal layer tries to use a genetlink multicast string that is
    longer than the 16 character limit.  Fix it and add a BUG check to
    prevent this kind of thing from happening in the future.

 From Masatake YAMATO.

 8) Fix many bugs in the handling of the teardown of L2TP connections,
    UDP encapsulation instances, and sockets.  From Tom Parkin.

 9) Missing socket release in IRDA, from Kees Cook.

10) Fix fec driver modular build, from Fabio Estevam.

11) Erroneous use of kfree() instead of free_netdev() in lantiq_etop,
    from Wei Yongjun.

12) Fix bugs in handling of queue numbers and steering rules in mlx4
    driver, from Moshe Lazer, Hadar Hen Zion, and Or Gerlitz.

13) Some FOO_DIAG_MAX constants were defined off by one, fix from Andrey
    Vagin.

14) TCP segmentation deferral is unintentionally done too strongly,
    breaking ACK clocking.  Fix from Eric Dumazet.

15) net_enable_timestamp() can legitimately be invoked from software
    interrupts, and in a way that is safe, so remove the WARN_ON().
    Also from Eric Dumazet.

16) Fix use after free in VLANs, from Cong Wang.

17) Fix TCP slow start retransmit storms after SACK reneging, from
    Yuchung Cheng.

18) Unix socket release should mark a socket dead before NULL'ing out
    sock->sk, otherwise we can race.  Fix from Paul Moore.

19) IPV6 addrconf code can try to free static memory, from Hong Zhiguo.

20) Fix register mis-programming, NULL pointer derefs, and wrong PHC
    clock frequency in IGB driver.  From Lior LevyAlex Williamson, Jiri
    Benc, and Jeff Kirsher.

21) skb->ip_summed logic in pch_gbe driver is reversed, breaking packet
    forwarding.  Fix from Veaceslav Falico.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (65 commits)
  ipv4: Fix ip-header identification for gso packets.
  bonding: remove already created master sysfs link on failure
  af_unix: dont send SCM_CREDENTIAL when dest socket is NULL
  pch_gbe: fix ip_summed checksum reporting on rx
  igb: fix PHC stopping on max freq
  igb: make sensor info static
  igb: SR-IOV init reordering
  igb: Fix null pointer dereference
  igb: fix i350 anti spoofing config
  ixgbevf: don't release the soft entries
  ipv6: fix bad free of addrconf_init_net
  unix: fix a race condition in unix_release()
  tcp: undo spurious timeout after SACK reneging
  bnx2x: fix assignment of signed expression to unsigned variable
  bridge: fix crash when set mac address of br interface
  8021q: fix a potential use-after-free
  net: remove a WARN_ON() in net_enable_timestamp()
  tcp: preserve ACK clocking in TSO
  net: fix *_DIAG_MAX constants
  net/mlx4_core: Disallow releasing VF QPs which have steering rules
  ...

Pull NFS client bugfixes from Trond Myklebust:
 - Fix an NFSv4 idmapper regression
 - Fix an Oops in the pNFS blocks client
 - Fix up various issues with pNFS layoutcommit
 - Ensure correct read ordering of variables in
   rpc_wake_up_task_queue_locked

* tag 'nfs-for-3.9-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  SUNRPC: Add barriers to ensure read ordering in rpc_wake_up_task_queue_locked
  NFSv4.1: Add a helper pnfs_commit_and_return_layout
  NFSv4.1: Always clear the NFS_INO_LAYOUTCOMMIT in layoutreturn
  NFSv4.1: Fix a race in pNFS layoutcommit
  pnfs-block: removing DM device maybe cause oops when call dev_remove
  NFSv4: Fix the string length returned by the idmapper

ip-header id needs to be incremented even if IP_DF flag is set.
This behaviour was changed in commit 490ab081
(IP_GRE: Fix IP-Identification).

Following patch fixes it so that identification is always
incremented.
Reported-by: Cong Wang <amwang@redhat.com>
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

If slave sysfs symlink failes to be created - we end up without removing
the master sysfs symlink. Remove it in case of failure.
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

SCM_SCREDENTIALS should apply to write() syscalls only either source or destination
socket asserted SOCK_PASSCRED. The original implememtation in maybe_add_creds is wrong,
and breaks several LSB testcases ( i.e. /tset/LSB.os/netowkr/recvfrom/T.recvfrom).
Origionally-authored-by: Karel Srot <ksrot@redhat.com>
Signed-off-by: Ding Tianhong <dingtianhong@huawei.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Jeff Kirsher says:

====================
This series contains updates to ixgbevf and igb.

The ixgbevf calls to pci_disable_msix() and to free the msix_entries
memory should not occur if device open fails.  Instead they should be
called during device driver removal to balance with the call to
pci_enable_msix() and the call to allocate msix_entries memory
during the device probe and driver load.

The remaining 4 of 5 igb patches are simple 1-3 line patches to fix
several issues such as possible null pointer dereference, PHC stopping
on max frequency, make sensor info static and SR-IOV initialization
reordering.

The remaining igb patch to fix anti-spoofing config fixes a problem
in i350 where anti spoofing configuration was written into a wrong
register.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

skb->ip_summed should be CHECKSUM_UNNECESSARY when the driver reports that
checksums were correct and CHECKSUM_NONE in any other case. They're
currently placed vice versa, which breaks the forwarding scenario. Fix it
by placing them as described above.
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Remove duplicated include.
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Joerg Roedel <joro@8bytes.org>

For 82576 MAC type, max_adj is reported as 1000000000 ppb. However, if
this value is passed to igb_ptp_adjfreq_82576, incvalue overflows out of
INCVALUE_82576_MASK, resulting in setting of zero TIMINCA.incvalue, stopping
the PHC (instead of going at twice the nominal speed).

Fix the advertised max_adj value to the largest value hardware can handle.
As there is no min_adj value available (-max_adj is used instead), this will
also prevent stopping the clock intentionally. It's probably not a big deal,
other igb MAC types don't support stopping the clock, either.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Acked-by: Matthew Vick <matthew.vick@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Trivial sparse warning.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

igb is ineffective at setting a lower total VFs because:

int pci_sriov_set_totalvfs(struct pci_dev *dev, u16 numvfs)
{
        ...
        /* Shouldn't change if VFs already enabled */
        if (dev->sriov->ctrl & PCI_SRIOV_CTRL_VFE)
                return -EBUSY;

Swap init ordering.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Greg Rose <gregory.v.rose@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

The max_vfs= option has always been self limiting to the number of VFs
supported by the device.  fa44f2f1 added SR-IOV configuration via
sysfs, but in the process broke this self correction factor.  The
failing path is:

igb_probe
  igb_sw_init
    if (max_vfs > 7) {
        adapter->vfs_allocated_count = 7;
    ...
    igb_probe_vfs
    igb_enable_sriov(, max_vfs)
      if (num_vfs > 7) {
        err = -EPERM;
        ...

This leaves vfs_allocated_count = 7 and vf_data = NULL, so we bomb out
when igb_probe finally calls igb_reset.  It seems like a really bad
idea, and somewhat pointless, to set vfs_allocated_count separate from
vf_data, but limiting max_vfs is enough to avoid the null pointer.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Greg Rose <gregory.v.rose@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Fix a problem in i350 where anti spoofing configuration was written into a
wrong register.
Signed-off-by: Lior Levy <lior.levy@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

When the ixgbevf driver is opened the request to allocate MSIX irq
vectors may fail.  In that case the driver will call ixgbevf_down()
which will call ixgbevf_irq_disable() to clear the HW interrupt
registers and calls synchronize_irq() using the msix_entries pointer in
the adapter structure.  However, when the function to request the MSIX
irq vectors failed it had already freed the msix_entries which causes
an OOPs from using the NULL pointer in synchronize_irq().

The calls to pci_disable_msix() and to free the msix_entries memory
should not occur if device open fails.  Instead they should be called
during device driver removal to balance with the call to
pci_enable_msix() and the call to allocate msix_entries memory
during the device probe and driver load.
Signed-off-by: Li Xun <xunleer.li@huawei.com>
Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Tested-by: Sibai Li <sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Pull timer fix from Thomas Gleixner:
 "A single bugfix which prevents that a non functional timer device is
  selected to provide the fallback device, which is supposed to serve
  timer interrupts on behalf of non functional devices ..."

* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  clockevents: Don't allow dummy broadcast timers

Signed-off-by: Hong Zhiguo <honkiko@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

As reported by Jan, and others over the past few years, there is a
race condition caused by unix_release setting the sock->sk pointer
to NULL before properly marking the socket as dead/orphaned.  This
can cause a problem with the LSM hook security_unix_may_send() if
there is another socket attempting to write to this partially
released socket in between when sock->sk is set to NULL and it is
marked as dead/orphaned.  This patch fixes this by only setting
sock->sk to NULL after the socket has been marked as dead; I also
take the opportunity to make unix_release_sock() a void function
as it only ever returned 0/success.

Dave, I think this one should go on the -stable pile.

Special thanks to Jan for coming up with a reproducer for this
problem.
Reported-by: Jan Stancek <jan.stancek@gmail.com>
Signed-off-by: Paul Moore <pmoore@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Pull infiniband/rdma fixes from Roland Dreier:
 "Small batch of InfiniBand/RDMA fixes for 3.9:

   - Fix for TX lockup in IPoIB
   - QLogic -> Intel update for qib driver
   - Small static checker fix for qib
   - Fix error path return value in cxgb4"

* tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IB/qib: change QLogic to Intel
  IB/ipath: Silence a static checker warning
  IPoIB: Fix send lockup due to missed TX completion
  RDMA/cxgb4: Fix error return code in create_qp()