While testing how linux behaves on SYNFLOOD attack on multiqueue device
(ixgbe), I found that SYNACK messages were dropped at Qdisc level
because we send them all on a single queue.

Obvious choice is to reflect incoming SYN packet @queue_mapping to
SYNACK packet.

Under stress, my machine could only send 25.000 SYNACK per second (for
200.000 incoming SYN per second). NIC : ixgbe with 16 rx/tx queues.

After patch, not a single SYNACK is dropped.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Hans Schillstrom <hans.schillstrom@ericsson.com>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Another problem on SYNFLOOD/DDOS attack is the inetpeer cache getting
larger and larger, using lots of memory and cpu time.

tcp_v4_send_synack()
->inet_csk_route_req()
 ->ip_route_output_flow()
  ->rt_set_nexthop()
   ->rt_init_metrics()
    ->inet_getpeer( create = true)

This is a side effect of commit a4daad6b (net: Pre-COW metrics for
TCP) added in 2.6.39

Possible solution :

Instruct inet_csk_route_req() to remove FLOWI_FLAG_PRECOW_METRICS

Before patch :

# grep peer /proc/slabinfo
inet_peer_cache   4175430 4175430    192   42    2 : tunables    0    0    0 : slabdata  99415  99415      0

Samples: 41K of event 'cycles', Event count (approx.): 30716565122
+  20,24%      ksoftirqd/0  [kernel.kallsyms]           [k] inet_getpeer
+   8,19%      ksoftirqd/0  [kernel.kallsyms]           [k] peer_avl_rebalance.isra.1
+   4,81%      ksoftirqd/0  [kernel.kallsyms]           [k] sha_transform
+   3,64%      ksoftirqd/0  [kernel.kallsyms]           [k] fib_table_lookup
+   2,36%      ksoftirqd/0  [ixgbe]                     [k] ixgbe_poll
+   2,16%      ksoftirqd/0  [kernel.kallsyms]           [k] __ip_route_output_key
+   2,11%      ksoftirqd/0  [kernel.kallsyms]           [k] kernel_map_pages
+   2,11%      ksoftirqd/0  [kernel.kallsyms]           [k] ip_route_input_common
+   2,01%      ksoftirqd/0  [kernel.kallsyms]           [k] __inet_lookup_established
+   1,83%      ksoftirqd/0  [kernel.kallsyms]           [k] md5_transform
+   1,75%      ksoftirqd/0  [kernel.kallsyms]           [k] check_leaf.isra.9
+   1,49%      ksoftirqd/0  [kernel.kallsyms]           [k] ipt_do_table
+   1,46%      ksoftirqd/0  [kernel.kallsyms]           [k] hrtimer_interrupt
+   1,45%      ksoftirqd/0  [kernel.kallsyms]           [k] kmem_cache_alloc
+   1,29%      ksoftirqd/0  [kernel.kallsyms]           [k] inet_csk_search_req
+   1,29%      ksoftirqd/0  [kernel.kallsyms]           [k] __netif_receive_skb
+   1,16%      ksoftirqd/0  [kernel.kallsyms]           [k] copy_user_generic_string
+   1,15%      ksoftirqd/0  [kernel.kallsyms]           [k] kmem_cache_free
+   1,02%      ksoftirqd/0  [kernel.kallsyms]           [k] tcp_make_synack
+   0,93%      ksoftirqd/0  [kernel.kallsyms]           [k] _raw_spin_lock_bh
+   0,87%      ksoftirqd/0  [kernel.kallsyms]           [k] __call_rcu
+   0,84%      ksoftirqd/0  [kernel.kallsyms]           [k] rt_garbage_collect
+   0,84%      ksoftirqd/0  [kernel.kallsyms]           [k] fib_rules_lookup
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Hans Schillstrom <hans.schillstrom@ericsson.com>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently, we terminate the eeprom access through clearing the CS by:

RTL_W8 (Cfg9346, ~EE_CS); or writeb (~EE_CS, ee_addr);

This would left the eeprom into "Config. Register Write Enable:"
state which is not expcted as the highest two bits were set to
0x11 ( expected is the "Normal" mode (0x00)). Solving this by write
0x0 instead of ~EE_CS when terminating the eeprom access.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently, we enable the receiver before setting the ring address which could
lead the card DMA into unexpected areas. Solving this by set the ring address
before enabling the receiver.

btw. I find and test this in qemu as I didn't have a 8139cp card in hand. please
review it carefully.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When NetLabel is not enabled, e.g. CONFIG_NETLABEL=n, and the system
receives a CIPSO tagged packet it is dropped (cipso_v4_validate()
returns non-zero).  In most cases this is the correct and desired
behavior, however, in the case where we are simply forwarding the
traffic, e.g. acting as a network bridge, this becomes a problem.

This patch fixes the forwarding problem by providing the basic CIPSO
validation code directly in ip_options_compile() without the need for
the NetLabel or CIPSO code.  The new validation code can not perform
any of the CIPSO option label/value verification that
cipso_v4_validate() does, but it can verify the basic CIPSO option
format.

The behavior when NetLabel is enabled is unchanged.
Signed-off-by: Paul Moore <pmoore@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We need to validate the number of pages consumed by data_len, otherwise frags
array could be overflowed by userspace. So this patch validate data_len and
return -EMSGSIZE when data_len may occupies more frags than MAX_SKB_FRAGS.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

dql->num_queued could change while processing dql_completed().
To provide consistent calculation, added an on stack variable.
Signed-off-by: Hiroaki SHIMODA <shimoda.hiroaki@gmail.com>
Cc: Tom Herbert <therbert@google.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Denys Fedoryshchenko <denys@visp.net.lb>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When below pattern is observed,

                                               TIME
       dql_queued()         dql_completed()     |
      a) initial state                          |
                                                |
      b) X bytes queued                         V

      c) Y bytes queued
                           d) X bytes completed
      e) Z bytes queued
                           f) Y bytes completed

a) dql->limit has already some value and there is no in-flight packet.
b) X bytes queued.
c) Y bytes queued and excess limit.
d) X bytes completed and dql->prev_ovlimit is set and also
   dql->prev_num_queued is set Y.
e) Z bytes queued.
f) Y bytes completed. inprogress and prev_inprogress are true.

At f), according to the comment, all_prev_completed becomes
true and limit should be increased. But POSDIFF() ignores
(completed == dql->prev_num_queued) case, so limit is decreased.
Signed-off-by: Hiroaki SHIMODA <shimoda.hiroaki@gmail.com>
Cc: Tom Herbert <therbert@google.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Denys Fedoryshchenko <denys@visp.net.lb>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

POSDIFF() fails to take into account integer overflow case.
Signed-off-by: Hiroaki SHIMODA <shimoda.hiroaki@gmail.com>
Cc: Tom Herbert <therbert@google.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Denys Fedoryshchenko <denys@visp.net.lb>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The "!mlx4_is_slave" is totally confusing.  Fix with
constant MLX4_CMD_NATIVE, which is the intended behavior.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The range check was performed after using the port number.

Reverse this to prevent a potential array overflow.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

- pass the following parameters:
  - firmware version (added QUERY_FW paravirtualization for that)

  - disable Blueflame on slaves. KVM disables write combining on guests,
    and we get better performance without BF in this case. (This requires
    QUERY_DEV_CAP paravirtualization, also in this commit)

  - max qp rdma as destination

- get rid of a chunk of "if (0)" dead code
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Port is used as an array index before we know if that is proper.

For example, in the catas event case, port is zero; however,
the port index should lie in the range (1..2).

Fix this by using 'port' only in the events where it is of interest.

Test for port out of range in the default (unhandled event) case,
and do not output a message if it is not an ethernet port.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In SRIOV mode, the number of EQs used when computing the total ICM size
was incorrect.

To fix this, we do the following:
1. We add a new structure to mlx4_dev, mlx4_phys_caps, to contain physical HCA
   capabilities.  The PPF uses the phys capabilities when it computes things
   like ICM size.

   The dev_caps structure will then contain the paravirtualized values, making
   bookkeeping much easier in SRIOV mode. We add a structure rather than a
   single parameter because there will be other fields in the phys_caps.

   The first field we add to the mlx4_phys_caps structure is num_phys_eqs.

2. In INIT_HCA, when running in SRIOV mode, the "log_num_eqs" parameter
   passed to the FW is the number of EQs per VF/PF; each function (PF or VF)
   has this number of EQs available.

   However, the total number of EQs which must be allowed for in the ICM is
   (1 << log_num_eqs) * (#VFs + #PFs).  Rather than compute this quantity,
   we allocate ICM space for 1024 EQs (which is the device maximum
   number of EQs, and which is the value we place in the mlx4_phys_caps structure).

   For INIT_HCA, however, we use the per-function number of EQs as described
   above.
Signed-off-by: Marcel Apfelbaum <marcela@dev.mellanox.co.il>
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Ths fixes the comparison in the FLR (Function Level Reset) event case.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Pull battery updates from Anton Vorontsov:
 "A bunch of fixes for v3.5, nothing extraordinary."

* tag 'for-v3.5' of git://git.infradead.org/battery-2.6: (27 commits)
  smb347-charger: Include missing <linux/err.h>
  smb347-charger: Clean up battery attributes
  max17042_battery: Add support for max17047/50 chip
  sbs-battery.c: Capacity attr = remaining relative capacity
  isp1704_charger: Use after free on probe error
  ds2781_battery: Use DS2781_PARAM_EEPROM_SIZE and DS2781_USER_EEPROM_SIZE
  power_supply: Fix a typo in BATTERY_DS2781 Kconfig entry
  charger-manager: Provide cm_notify_event function for in-kernel use
  charger-manager: Poll battery health in normal state
  smb347-charger: Convert to regmap API
  smb347-charger: Move IRQ enabling to the end of probe
  smb347-charger: Rename few functions to match better what they are doing
  smb347-charger: Convert to use module_i2c_driver()
  smb347_charger: Cleanup power supply registration code in probe
  ab8500: Clean up probe routines
  ab8500_fg: Harden platform data check
  ab8500_btemp: Harden platform data check
  ab8500_charger: Harden platform data check
  MAINTAINERS: Fix 'F' entry for the power supply class
  max17042_battery: Handle irq request failure case
  ...

Pull two small kvm fixes from Avi Kivity:
 "A build fix for non-kvm archs and a transparent hugepage refcount
  bugfix on hosts with 4M pages."

* git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: Export asm-generic/kvm_para.h
  KVM: MMU: fix huge page adapted on non-PAE host

Pull final round of SCSI updates from James Bottomley:
 "This is primarily another round of driver updates (bnx2fc, qla2xxx,
  qla4xxx) including the target mode driver for qla2xxx.  We've also got
  a couple of regression fixes (async scanning, broken this merge window
  and a fix to a long standing break in the scsi_wait_scan module)."

* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (45 commits)
  [SCSI] fix scsi_wait_scan
  [SCSI] fix async probe regression
  [SCSI] be2iscsi: fix dma free size mismatch regression
  [SCSI] qla4xxx: Update driver version to 5.02.00-k17
  [SCSI] qla4xxx: Capture minidump for ISP82XX on firmware failure
  [SCSI] qla4xxx: Add change_queue_depth API support
  [SCSI] qla4xxx: Fix clear ddb mbx command failure issue.
  [SCSI] qla4xxx: Fix kernel panic during discovery logout.
  [SCSI] qla4xxx: Correct early completion of pending mbox.
  [SCSI] fcoe, bnx2fc, libfcoe: SW FCoE and bnx2fc use FCoE Syfs
  [SCSI] libfcoe: Add fcoe_sysfs
  [SCSI] bnx2fc: Allocate fcoe_ctlr with bnx2fc_interface, not as a member
  [SCSI] fcoe: Allocate fcoe_ctlr with fcoe_interface, not as a member
  [SCSI] Fix dm-multipath starvation when scsi host is busy
  [SCSI] ufs: fix potential NULL pointer dereferencing error in ufshcd_prove.
  [SCSI] qla2xxx: don't free pool that wasn't allocated
  [SCSI] mptfusion: unlock on error in mpt_config()
  [SCSI] tcm_qla2xxx: Add >= 24xx series fabric module for target-core
  [SCSI] qla2xxx: Add LLD target-mode infrastructure for >= 24xx series
  [SCSI] Revert "qla2xxx: During loopdown perform Diagnostic loopback."
  ...

Pull follow-up block updates from Jens Axboe:
 "Includes the fix from Eric to resolve the infinite loop in looking up
  or creating an IO context for an exiting task.

  Also a bunch of mtip32xx patches.  Fairly trivial stuff.  It's fairly
  new though, but no point in keeping it out of the tree until 3.6 imho.
  I've thrown it through basic testing locally as well, works fine.  The
  one contentious part is the patch that Greg complained about (sysfs
  file with multiple values, should be a debugfs file), which is
  perfectly valid, but not a regression from what the file contains now.
  That will be fixed up separately."

* 'for-linus' of git://git.kernel.dk/linux-block:
  block: avoid infinite loop in get_task_io_context()
  mtip32xx: Changes to sysfs entries
  mtip32xx: Convert macro definitions for flag bits to enum
  mtip32xx: minor performance tweak
  mtip32xx: Fix to support more than one sector in exec_drive_command()
  mtip32xx: Use plain spinlock for 'cmd_issue_lock'
  mtip32xx: Set block queue boundary variables
  mtip32xx: Fix to handle TFE for PIO(IOCTL/internal) commands
  mtip32xx: Change HDIO_GET_IDENTITY to return stored data
  mtip32xx: Set custom timeouts for PIO commands
  mtip32xx: fix clearing an incorrect register in mtip_init_port

tty_lock_pair() do the right thing to avoid deadlocks, but should
instruct LOCKDEP of this to avoid a splat.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Use the same mechanism as the block devices are using, but move the
helper functions from fs/direct-io.c into fs/inode.c to remove the
dependency on CONFIG_BLOCK.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Fred Isaman <iisaman@netapp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Pull mce cleanup from Tony Luck:
 "One more mce cleanup before the 3.5 merge window closes"

* tag 'please-pull-mce' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
  x86/mce: Cleanup timer mess

Pull s390 patches from Heiko Carstens:
 "A couple of s390 patches for the 3.5 merge window.  Just a collection
  of bug fixes and cleanups."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/uaccess: fix access_ok compile warnings
  s390/cmpxchg: select HAVE_CMPXCHG_LOCAL option
  s390/cmpxchg: fix sign extension bugs
  s390/cmpxchg: fix 1 and 2 byte memory accesses
  s390/cmpxchg: fix compile warnings specific to s390
  s390/cmpxchg: add missing memory barrier to cmpxchg64
  s390/cpu: remove cpu "capabilities" sysfs attribute
  s390/kernel: Fix smp_call_ipl_cpu() for offline CPUs
  s390/kernel: Introduce memcpy_absolute() function
  s390/headers: replace __s390x__ with CONFIG_64BIT where possible
  s390/headers: remove #ifdef __KERNEL__ from not exported headers
  s390/irq: split irq stats for cpu-measurement alert facilities
  s390/kexec: Move early_pgm_check_handler() to text section
  s390/kdump: Use real mode for PSW restart and kexec
  s390/kdump: Account /sys/kernel/kexec_crash_size changes in OS info
  s390/kernel: Remove OS info init function call and diag 308 for kdump

Pull misc parisc updates from James Bottomley:
 "This is a couple of updates to complete our fixes and one to fix a
  compile failure caused during the merge window.  Additionally, we now
  switch to the generic strncopy_from_user."

* tag 'parisc-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/parisc-2.6:
  [PARISC] update parisc to use generic strncpy_from_user()
  [PARISC] Fix parisc compile failure after smp: Add task_struct argument to __cpu_up()
  [PARISC] fix TLB fault path on PA2.0 narrow systems
  [PARISC] fix boot failure on 32-bit systems caused by branch stubs placed before .text

Pull additional x86 fixes from Peter Anvin.

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, amd, xen: Avoid NULL pointer paravirt references
  x86, mtrr: Fix a type overflow in range_to_mtrr func
  x86, realmode: Unbreak the ia64 build of drivers/acpi/sleep.c
  x86/mm/pat: Improve scaling of pat_pagerange_is_ram()
  x86: hpet: Fix copy-and-paste mistake in earlier change
  x86/mce: Fix 32-bit build
  x86/bitops: Move BIT_64() for a wider use

Pull late-merged development and first fixes for arm-soc from Olof Johansson:
 "This branch contains a few development patches for Samsung and
  Versatile Express that were submitted to arm-soc near the beginning of
  the merge window.  We picked them up with the agreement that they
  would need to sit in linux-next for a while, and now they have.

  There are also two fixes:
   - One long-standing build breakage on ixp4xx due to missing gpiolib
     dependencies.
   - The other is for some gpio device tree changes needed on lpc32xx."

* tag 'devel-late' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  ARM: LPC32xx: Adjust dts files to gpio dt binding
  ixp4xx: fix compilation by adding gpiolib support
  ARM: vexpress: Remove twice included header files
  ARM: vexpress: Device Tree updates
  ARM: EXYNOS: Support suspend and resume for EXYNOS5250
  ARM: EXYNOS: Add Clock register list for save and restore
  ARM: EXYNOS: Add PMU table for EXYNOS5250
  ARM: EXYNOS: Rename of function for pm.c
  ARM: EXYNOS: Remove GIC save & restore function
  ARM: dts: Add node for interrupt combiner controller on EXYNOS5250
  ARM: S3C24XX: add support for second irq set of S3C2416
  ARM: S3C64XX: use timekeeping wrapper on cpuidle
  ARM: S3C64XX: declare the states with the new api on cpuidle
  ARM: S3C64XX: Hook up carrier class modules on Cragganmore
  ARM: S3C64XX: Initial hookup for Bells module on Cragganmore

Pull Sparc updates from David S. Miller:

 1) Remove the idiotic situation wherein Leon was a special case in all
    of the TLB/cache handling code.

    The worst side effect of this bogosity is that you couldn't build a
    kernel with Leon support enabled (to get better build coverage), and
    test boot it on a non-LEON cpu.

    Leon is, in all core respects, programatically identical to the
    32-bit SRMMU.  Except that they put the TLB registers in a different
    alternate address space location.

    Through code patching (for fast paths) and run time checks, this
    issue is now a thing of the past.

    From Sam Ravnborg.

 2) There was a mis-merge of arch/sparc/Kconfig for one of the
    clockevents changes that went in, causing 32-bit sparc to start
    failing to build.

    I merged in your tree to get those clockevents changes (and added a
    note to the merge commit) then added Stephen Rothwell's fix for the
    merge error.

 3) Software quad floating point emulation was not working properly on
    more recent Niagara chips, because the way the situation is reported
    by the cpu has changed.

    Nobody noticed because gcc emits calls to software emulation
    routines in glibc.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc: (25 commits)
  sparc: fix sparc64 build due to leon.h inclusion
  sparc32: remove unused variable in head_32.S
  sparc32,leon: fix leon bootup
  sparc32: Export leon_dma_ops to modules.
  sparc32: support leon + sun in dma_make_coherent()
  sparc32,leon: always support leon in ioport
  sparc32,leon: always include leon_pmc in build
  sparc32: refactor cpu_idle()
  sparc32: srmmu_probe now knows about leon too
  sparc32: drop LEON hack for ASI_M_MMUREGS
  sparc32: introduce run-time patching of srmmu access functions
  sparc32: introduce support for run-time patching for all shared assembler code
  sparc32,leon: fix section mismatch warning
  sparc32,leon: always include leon_smp + leon_mm in build
  sparc32,leon: always include leon_kernel in build
  sparc32,leon: clean up leon.h
  sparc32: handle leon in cpu.c
  sparc32: handle leon in irq_32.c
  sparc32: add support for run-time patching of leon/sun single instructions
  sparc32: introduce sparc32_start_kernel called from head_32.S
  ...

Pull networking changes from David S. Miller:

 1) Fix IPSEC header length calculation for transport mode in ESP.  The
    issue is whether to do the calculation before or after alignment.
    Fix from Benjamin Poirier.

 2) Fix regression in IPV6 IPSEC fragment length calculations, from Gao
    Feng.  This is another transport vs tunnel mode issue.

 3) Handle AF_UNSPEC connect()s properly in L2TP to avoid OOPSes.  Fix
    from James Chapman.

 4) Fix USB ASIX driver's reception of full sized VLAN packets, from
    Eric Dumazet.

 5) Allow drop monitor (and, more generically, all generic netlink
    protocols) to be automatically loaded as a module.  From Neil
    Horman.

Fix up trivial conflict in Documentation/feature-removal-schedule.txt
due to new entries added next to each other at the end. As usual.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (38 commits)
  net/smsc911x: Repair broken failure paths
  virtio-net: remove useless disable on freeze
  netdevice: Update netif_dbg for CONFIG_DYNAMIC_DEBUG
  drop_monitor: Add module alias to enable automatic module loading
  genetlink: Build a generic netlink family module alias
  net: add MODULE_ALIAS_NET_PF_PROTO_NAME
  r6040: Do a Proper deinit at errorpath and also when driver unloads (calling r6040_remove_one)
  r6040: disable pci device if the subsequent calls (after pci_enable_device) fails
  skb: avoid unnecessary reallocations in __skb_cow
  net: sh_eth: fix the rxdesc pointer when rx descriptor empty happens
  asix: allow full size 8021Q frames to be received
  rds_rdma: don't assume infiniband device is PCI
  l2tp: fix oops in L2TP IP sockets for connect() AF_UNSPEC case
  mac80211: fix ADDBA declined after suspend with wowlan
  wlcore: fix undefined symbols when CONFIG_PM is not defined
  mac80211: fix flag check for QoS NOACK frames
  ath9k_hw: apply internal regulator settings on AR933x
  ath9k_hw: update AR933x initvals to fix issues with high power devices
  ath9k: fix a use-after-free-bug when ath_tx_setup_buffer() fails
  ath9k: stop rx dma before stopping tx
  ...

Calling get_task_io_context() on a exiting task which isn't %current can
loop forever. This triggers at boot time on my dev machine.

BUG: soft lockup - CPU#3 stuck for 22s ! [mountall.1603]

Fix this by making create_task_io_context() returns -EBUSY in this case
to break the loop.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Alan Cox <alan@linux.intel.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Signed-off-by: James Bottomley <JBottomley@Parallels.com>

* Formatted the output of 'registers' entry
* Added "Commands in Q' to output of 'registers' entry
* Added a new entry 'flags'
Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Convert macro definitions for flags bits to enum
Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

When checking for command completions if the register value is zero, proceed
to next register.
Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Fix to support more than one sector in exec_drive_command().
Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

'cmd_issue_lock' is for only acquiring a free slot, and it is not used
in interrupt context. So replaced irq version with non-irq version of spinlock.
Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Set the following block queue boundary variables
	* max_hw_sectors
	* max_segment_size
Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>

Removed setting of q->nr_requests.
Signed-off-by: Jens Axboe <axboe@kernel.dk>

If a PIO (IOCTL/internal) command resulted in TFE, signal the wait event or break out of polling.
Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

For the ioctl command HDIO_GET_IDENTITY, return the stored copy of IDENTIFY
DATA instead of sending the command to the device - similar to libata.
Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

This change sets custom timeouts depending on PIO command.
Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Fix clearing an incorrect register in mtip_init_port
Signed-off-by: Asai Thambi S P <asamymuthupa@micron.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>