openwrt/staging/blogic.git
8 years agonet: Add net-device param to the get offloaded stats ndo
Or Gerlitz [Tue, 22 Nov 2016 21:09:54 +0000 (23:09 +0200)]
net: Add net-device param to the get offloaded stats ndo

Some drivers would need to check few internal matters for
that. To be used in downstream mlx5 commit.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'phy-broadcom-wirespeed-downshift-support'
David S. Miller [Thu, 24 Nov 2016 20:46:03 +0000 (15:46 -0500)]
Merge branch 'phy-broadcom-wirespeed-downshift-support'

Florian Fainelli says:

====================
net: phy: broadcom: Wirespeed/downshift support

This patch series adds support for the Broadcom Wirespeed, aka
downsfhit feature utilizing the recently added ethtool PHY tunables.

Tested with two Gigabit link partners with a 4-wire cable having only
2 pairs connected.

Last patch in the series is a fix that was required for testing, which
should make it to -stable, which I can submit separate against net if
you prefer David.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: bcm_sf2: Ensure we re-negotiate EEE during after link change
Florian Fainelli [Tue, 22 Nov 2016 19:40:58 +0000 (11:40 -0800)]
net: dsa: bcm_sf2: Ensure we re-negotiate EEE during after link change

In case the link change and EEE is enabled or disabled, always try to
re-negotiate this with the link partner.

Fixes: 450b05c15f9c ("net: dsa: bcm_sf2: add support for controlling EEE")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: phy: bcm7xxx: Add support for downshift/Wirespeed
Florian Fainelli [Tue, 22 Nov 2016 19:40:57 +0000 (11:40 -0800)]
net: phy: bcm7xxx: Add support for downshift/Wirespeed

Add support for configuring the downshift/Wirespeed enable/disable
toggles and specify a link retry value ranging from 1 to 9. Since the
integrated BCM7xxx have issues when wirespeed is enabled and EEE is also
enabled, we do disable EEE if wirespeed is enabled.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: phy: broadcom: Allow enabling or disabling of EEE
Florian Fainelli [Tue, 22 Nov 2016 19:40:56 +0000 (11:40 -0800)]
net: phy: broadcom: Allow enabling or disabling of EEE

In preparation for adding support for Wirespeed/downshift, we need to
change bcm_phy_eee_enable() to allow enabling or disabling EEE, so make
the function take an extra enable/disable boolean parameter and rename
it to illustrate it sets EEE, not necessarily just enables it.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: phy: broadcom: Add support code for downshift/Wirespeed
Florian Fainelli [Tue, 22 Nov 2016 19:40:55 +0000 (11:40 -0800)]
net: phy: broadcom: Add support code for downshift/Wirespeed

Broadcom's Wirespeed feature allows us to configure how auto-negotiation
should behave with fewer working pairs of wires on a cable. Add support
code for retrieving and setting such downshift counters using the
recently added ethtool downshift tunables.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: phy: broadcom: Move bcm54xx_auxctl_{read, write} to common library
Florian Fainelli [Tue, 22 Nov 2016 19:40:54 +0000 (11:40 -0800)]
net: phy: broadcom: Move bcm54xx_auxctl_{read, write} to common library

We are going to need these functions to implement support for Broadcom
Wirespeed, aka downshift.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotcp: enhance tcp_collapse_retrans() with skb_shift()
Eric Dumazet [Tue, 15 Nov 2016 20:51:50 +0000 (12:51 -0800)]
tcp: enhance tcp_collapse_retrans() with skb_shift()

In commit 2331ccc5b323 ("tcp: enhance tcp collapsing"),
we made a first step allowing copying right skb to left skb head.

Since all skbs in socket write queue are headless (but possibly the very
first one), this strategy often does not work.

This patch extends tcp_collapse_retrans() to perform frag shifting,
thanks to skb_shift() helper.

This helper needs to not BUG on non headless skbs, as callers are ok
with that.

Tested:

Following packetdrill test now passes :

0.000 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3
   +0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0
   +0 bind(3, ..., ...) = 0
   +0 listen(3, 1) = 0

   +0 < S 0:0(0) win 32792 <mss 1460,sackOK,nop,nop,nop,wscale 8>
   +0 > S. 0:0(0) ack 1 <mss 1460,nop,nop,sackOK,nop,wscale 8>
+.100 < . 1:1(0) ack 1 win 257
   +0 accept(3, ..., ...) = 4

   +0 setsockopt(4, SOL_TCP, TCP_NODELAY, [1], 4) = 0
   +0 write(4, ..., 200) = 200
   +0 > P. 1:201(200) ack 1
+.001 write(4, ..., 200) = 200
   +0 > P. 201:401(200) ack 1
+.001 write(4, ..., 200) = 200
   +0 > P. 401:601(200) ack 1
+.001 write(4, ..., 200) = 200
   +0 > P. 601:801(200) ack 1
+.001 write(4, ..., 200) = 200
   +0 > P. 801:1001(200) ack 1
+.001 write(4, ..., 100) = 100
   +0 > P. 1001:1101(100) ack 1
+.001 write(4, ..., 100) = 100
   +0 > P. 1101:1201(100) ack 1
+.001 write(4, ..., 100) = 100
   +0 > P. 1201:1301(100) ack 1
+.001 write(4, ..., 100) = 100
   +0 > P. 1301:1401(100) ack 1

+.099 < . 1:1(0) ack 201 win 257
+.001 < . 1:1(0) ack 201 win 257 <nop,nop,sack 1001:1401>
   +0 > P. 201:1001(800) ack 1

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: add MV88E6097 switch
Stefan Eichenberger [Tue, 22 Nov 2016 16:47:21 +0000 (17:47 +0100)]
net: dsa: mv88e6xxx: add MV88E6097 switch

Add support for the MV88E6097 switch. The change was tested on an Armada
based platform with a MV88E6097 switch.

Signed-off-by: Stefan Eichenberger <stefan.eichenberger@netmodule.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/phy: add trace events for mdio accesses
Uwe Kleine-König [Tue, 22 Nov 2016 15:47:11 +0000 (16:47 +0100)]
net/phy: add trace events for mdio accesses

Make it possible to generate trace events for mdio read and write accesses.

Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoVSOCK: add loopback to virtio_transport
Stefan Hajnoczi [Mon, 21 Nov 2016 13:56:31 +0000 (13:56 +0000)]
VSOCK: add loopback to virtio_transport

The VMware VMCI transport supports loopback inside virtual machines.
This patch implements loopback for virtio-vsock.

Flow control is handled by the virtio-vsock protocol as usual.  The
sending process stops transmitting on a connection when the peer's
receive buffer space is exhausted.

Cathy Avery <cavery@redhat.com> noticed this difference between VMCI and
virtio-vsock when a test case using loopback failed.  Although loopback
isn't the main point of AF_VSOCK, it is useful for testing and
virtio-vsock must match VMCI semantics so that userspace programs run
regardless of the underlying transport.

My understanding is that loopback is not supported on the host side with
VMCI.  Follow that by implementing it only in the guest driver, not the
vhost host driver.

Cc: Jorgen Hansen <jhansen@vmware.com>
Reported-by: Cathy Avery <cavery@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
David S. Miller [Tue, 22 Nov 2016 16:29:28 +0000 (11:29 -0500)]
Merge git://git./linux/kernel/git/davem/net

All conflicts were simple overlapping changes except perhaps
for the Thunder driver.

That driver has a change_mtu method explicitly for sending
a message to the hardware.  If that fails it returns an
error.

Normally a driver doesn't need an ndo_change_mtu method becuase those
are usually just range changes, which are now handled generically.
But since this extra operation is needed in the Thunder driver, it has
to stay.

However, if the message send fails we have to restore the original
MTU before the change because the entire call chain expects that if
an error is thrown by ndo_change_mtu then the MTU did not change.
Therefore code is added to nicvf_change_mtu to remember the original
MTU, and to restore it upon nicvf_update_hw_max_frs() failue.

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agomarvell: mark mvneta and mvpp2 32-bit only
Arnd Bergmann [Tue, 22 Nov 2016 14:21:22 +0000 (15:21 +0100)]
marvell: mark mvneta and mvpp2 32-bit only

Both of these drivers won't work on 64-bit architectures unless they
are redesigned, since they store a virtual address pointer in a 32-bit
field of the descriptors:

drivers/net/ethernet/marvell/mvneta_bm.c: In function 'mvneta_bm_construct':
drivers/net/ethernet/marvell/mvneta_bm.c:103:16: error: cast from pointer to integer of different size [-Werror=pointer-to-int-cast]
drivers/net/ethernet/marvell/mvpp2.c: In function 'mvpp2_prs_vlan_init':
drivers/net/ethernet/marvell/mvpp2.c:2563:32: error: large integer implicitly truncated to unsigned type [-Werror=overflow]

This limits the COMPILE_TEST option for the two drivers again to
only build them on 32-bit. This seems nicer than shutting up the
warnings, in case we ever actually want to use them on 64-bit,
as the warnings indicate which parts of the driver are currently
broken there.

Fixes: a0627f776a45 ("net: marvell: Allow drivers to be built with COMPILE_TEST")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'mlxsw-thermal-zone'
David S. Miller [Tue, 22 Nov 2016 15:04:19 +0000 (10:04 -0500)]
Merge branch 'mlxsw-thermal-zone'

Jiri Pirko says:

====================
mlxsw: core: Implement thermal zone

Implement thermal zone for mlxsw based HW.
The first patch is just a register dependency for the second patch.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agomlxsw: core: Implement thermal zone
Ivan Vecera [Tue, 22 Nov 2016 10:24:13 +0000 (11:24 +0100)]
mlxsw: core: Implement thermal zone

Implement thermal zone for mlxsw based HW. It uses temperature sensor
provided by ASIC (the same as mlxsw hwmon interface) to report current
temp to thermal core. The ASIC's PWM is then used to control speed
of system fans registered as cooling devices.

Signed-off-by: Ivan Vecera <cera@cera.cz>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agomlxsw: reg: Add Management Fan Speed Limit register
Jiri Pirko [Tue, 22 Nov 2016 10:24:12 +0000 (11:24 +0100)]
mlxsw: reg: Add Management Fan Speed Limit register

The MFSL register is used to configure the fan speed event / interrupt
notification mechanism. Fan speed threshold are defined for both
under-speed and over-speed.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'mv88e6390-initial-support'
David S. Miller [Tue, 22 Nov 2016 14:55:32 +0000 (09:55 -0500)]
Merge branch 'mv88e6390-initial-support'

Andrew Lunn says:

====================
Start adding support for mv88e6390

This is the first patchset implementing support for the mv88e6390
family.  This is a new generation of switch devices and has numerous
incompatible changes to the registers. These patches allow the switch
to the detected during probe, and makes the statistics unit work.

These patches are insufficient to make the mv88e6390 functional. More
patches will follow.

v2:
  Move stats code into global1
  Change DT compatible string to mv88e6190
  Fixed mv88e6351 stats which v1 had broken
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Move g1 stats code in global1.[ch]
Andrew Lunn [Mon, 21 Nov 2016 22:27:05 +0000 (23:27 +0100)]
net: dsa: mv88e6xxx: Move g1 stats code in global1.[ch]

Move the stats functions which access global 1 registers into
global1.c.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Implement mv88e6390 get_stats
Andrew Lunn [Mon, 21 Nov 2016 22:27:04 +0000 (23:27 +0100)]
net: dsa: mv88e6xxx: Implement mv88e6390 get_stats

The mv88e6390 uses a different bit to select between bank0 and bank1
of the statistics. So implement an ops function for this, and pass the
selector bit to the generic stats read function. Also, the histogram
selection has moved for the mv88e6390, so abstract its selection as
well.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Add stats_get_stats to ops structure
Andrew Lunn [Mon, 21 Nov 2016 22:27:03 +0000 (23:27 +0100)]
net: dsa: mv88e6xxx: Add stats_get_stats to ops structure

Different families have different sets of statistics. Abstract this
using a stats_get_stats op. The mv88e6390 needs a different
implementation, which will be added later.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Add stats_get_sset_count|string to ops structure
Andrew Lunn [Mon, 21 Nov 2016 22:27:02 +0000 (23:27 +0100)]
net: dsa: mv88e6xxx: Add stats_get_sset_count|string to ops structure

Different families have different sets of statistics. Abstract this
using a stats_get_sset_count and stats_get_strings op. Each stat has a
bitmap, and the ops implementer uses a bit map mask to count the
statistics which apply for the family, or return the list of strings.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
v2:
  Rename functions to avoid _ prefix.
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Add mv88e6390 statistics unit init
Andrew Lunn [Mon, 21 Nov 2016 22:27:01 +0000 (23:27 +0100)]
net: dsa: mv88e6xxx: Add mv88e6390 statistics unit init

The statistics unit on the mv88e6390 needs the histogram mode to be
configured in a different register compared to other devices. Add an
ops to do this.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
v2:
  Rename to mv88e6390_g1_stats_set_histogram
  Move into global1.c
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Add mv88e6390 stats snapshot operation
Andrew Lunn [Mon, 21 Nov 2016 22:27:00 +0000 (23:27 +0100)]
net: dsa: mv88e6xxx: Add mv88e6390 stats snapshot operation

The MV88E6390 has a control register for what the histogram statistics
actually contain. This means the stat_snapshot method should not set
this information. So implement the 6390 stats_snapshot function without
these bits.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Add comment about family a device belongs to
Andrew Lunn [Mon, 21 Nov 2016 22:26:59 +0000 (23:26 +0100)]
net: dsa: mv88e6xxx: Add comment about family a device belongs to

Knowing the family of device belongs to helps with picking the ops
implementation which is appropriate to the device. So add a comment to
each structure of ops.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Abstract stats_snapshot into ops structure
Andrew Lunn [Mon, 21 Nov 2016 22:26:58 +0000 (23:26 +0100)]
net: dsa: mv88e6xxx: Abstract stats_snapshot into ops structure

Taking a stats snapshot differs between same families. Abstract this
into an ops member. At the same time, move the code into global1.[ch],
since the registers are in the global1 range.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Add the mv88e6390 family
Andrew Lunn [Mon, 21 Nov 2016 22:26:57 +0000 (23:26 +0100)]
net: dsa: mv88e6xxx: Add the mv88e6390 family

With the devices added to the tables, the probe will recognize the
switch. This however is not sufficient to make it work properly, other
changes are needed because of incompatibilities.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Fix unused variable warning by using variable
Andrew Lunn [Mon, 21 Nov 2016 22:26:56 +0000 (23:26 +0100)]
net: dsa: mv88e6xxx: Fix unused variable warning by using variable

_mv88e6xxx_stats_wait() did not check the return value from
 mv88e6xxx_g1_read(), so the compiler complained about set but unused
 err.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Take switch out of reset before probe
Andrew Lunn [Mon, 21 Nov 2016 22:26:55 +0000 (23:26 +0100)]
net: dsa: mv88e6xxx: Take switch out of reset before probe

The switch needs to be taken out of reset before we can read its ID
register on the MDIO bus.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Mon, 21 Nov 2016 23:27:41 +0000 (15:27 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jmorris/linux-security

Pull apparmor bugfix from James Morris:
 "This has a fix for a policy replacement bug that is fairly serious for
  apache mod_apparmor users, as it results in the wrong policy being
  applied on an network facing service"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/linux-security:
  apparmor: fix change_hat not finding hat after policy replacement

8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc
Linus Torvalds [Mon, 21 Nov 2016 21:56:17 +0000 (13:56 -0800)]
Merge git://git./linux/kernel/git/davem/sparc

Pull sparc fixes from David Miller:

 1) With modern networking cards we can run out of 32-bit DMA space, so
    support 64-bit DMA addressing when possible on sparc64. From Dave
    Tushar.

 2) Some signal frame validation checks are inverted on sparc32, fix
    from Andreas Larsson.

 3) Lockdep tables can get too large in some circumstances on sparc64,
    add a way to adjust the size a bit. From Babu Moger.

 4) Fix NUMA node probing on some sun4v systems, from Thomas Tai.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc:
  sparc: drop duplicate header scatterlist.h
  lockdep: Limit static allocations if PROVE_LOCKING_SMALL is defined
  config: Adding the new config parameter CONFIG_PROVE_LOCKING_SMALL for sparc
  sunbmac: Fix compiler warning
  sunqe: Fix compiler warnings
  sparc64: Enable 64-bit DMA
  sparc64: Enable sun4v dma ops to use IOMMU v2 APIs
  sparc64: Bind PCIe devices to use IOMMU v2 service
  sparc64: Initialize iommu_map_table and iommu_pool
  sparc64: Add ATU (new IOMMU) support
  sparc64: Add FORCE_MAX_ZONEORDER and default to 13
  sparc64: fix compile warning section mismatch in find_node()
  sparc32: Fix inverted invalid_frame_pointer checks on sigreturns
  sparc64: Fix find_node warning if numa node cannot be found

8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Mon, 21 Nov 2016 21:26:28 +0000 (13:26 -0800)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Clear congestion control state when changing algorithms on an
    existing socket, from Florian Westphal.

 2) Fix register bit values in altr_tse_pcs portion of stmmac driver,
    from Jia Jie Ho.

 3) Fix PTP handling in stammc driver for GMAC4, from Giuseppe
    CAVALLARO.

 4) Fix udplite multicast delivery handling, it ignores the udp_table
    parameter passed into the lookups, from Pablo Neira Ayuso.

 5) Synchronize the space estimated by rtnl_vfinfo_size and the space
    actually used by rtnl_fill_vfinfo. From Sabrina Dubroca.

 6) Fix memory leak in fib_info when splitting nodes, from Alexander
    Duyck.

 7) If a driver does a napi_hash_del() explicitily and not via
    netif_napi_del(), it must perform RCU synchronization as needed. Fix
    this in virtio-net and bnxt drivers, from Eric Dumazet.

 8) Likewise, it is not necessary to invoke napi_hash_del() is we are
    also doing neif_napi_del() in the same code path. Remove such calls
    from be2net and cxgb4 drivers, also from Eric Dumazet.

 9) Don't allocate an ID in peernet2id_alloc() if the netns is dead,
    from WANG Cong.

10) Fix OF node and device struct leaks in of_mdio, from Johan Hovold.

11) We cannot cache routes in ip6_tunnel when using inherited traffic
    classes, from Paolo Abeni.

12) Fix several crashes and leaks in cpsw driver, from Johan Hovold.

13) Splice operations cannot use freezable blocking calls in AF_UNIX,
    from WANG Cong.

14) Link dump filtering by master device and kind support added an error
    in loop index updates during the dump if we actually do filter, fix
    from Zhang Shengju.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (59 commits)
  tcp: zero ca_priv area when switching cc algorithms
  net: l2tp: Treat NET_XMIT_CN as success in l2tp_eth_dev_xmit
  ethernet: stmmac: make DWMAC_STM32 depend on it's associated SoC
  tipc: eliminate obsolete socket locking policy description
  rtnl: fix the loop index update error in rtnl_dump_ifinfo()
  l2tp: fix racy SOCK_ZAPPED flag check in l2tp_ip{,6}_bind()
  net: macb: add check for dma mapping error in start_xmit()
  rtnetlink: fix FDB size computation
  netns: fix get_net_ns_by_fd(int pid) typo
  af_unix: conditionally use freezable blocking calls in read
  net: ethernet: ti: cpsw: fix fixed-link phy probe deferral
  net: ethernet: ti: cpsw: add missing sanity check
  net: ethernet: ti: cpsw: fix secondary-emac probe error path
  net: ethernet: ti: cpsw: fix of_node and phydev leaks
  net: ethernet: ti: cpsw: fix deferred probe
  net: ethernet: ti: cpsw: fix mdio device reference leak
  net: ethernet: ti: cpsw: fix bad register access in probe error path
  net: sky2: Fix shutdown crash
  cfg80211: limit scan results cache size
  net sched filters: pass netlink message flags in event notification
  ...

8 years agonet: ieee802154: constify ieee802154_ops structures
Bhumika Goyal [Mon, 21 Nov 2016 20:30:14 +0000 (02:00 +0530)]
net: ieee802154: constify ieee802154_ops structures

Declare the structure ieee802154_ops as const as it is only passed as an
argument to the function  ieee802154_alloc_hw. This argument is of type
const struct ieee802154_ops *, so ieee80254_ops structures having this
property can be declared as const.
Done using Coccinelle:

@r1 disable optional_qualifier @
identifier i;
position p;
@@
static struct ieee802154_ops i@p = {...};

@ok1@
identifier r1.i;
position p;
expression e1;
@@
ieee802154_alloc_hw(e1,&i@p)

@bad@
position p!={r1.p,ok1.p};
identifier r1.i;
@@
i@p

@depends on !bad disable optional_qualifier@
identifier r1.i;
@@
static
+const
struct ieee802154_ops  i={...};

@depends on !bad disable optional_qualifier@
identifier r1.i;
@@
+const
struct ieee802154_ops  i;

The before and after size details of the affected files are:

   text    data     bss     dec     hex filename
   8669    1176      16    9861    2685 drivers/net/ieee802154/adf7242.o
   8805    1048      16    9869    268d drivers/net/ieee802154/adf7242.o

   text    data     bss     dec     hex filename
   7211    2296      32    9539    2543 drivers/net/ieee802154/atusb.o
   7339    2160      32    9531    253b drivers/net/ieee802154/atusb.o

Signed-off-by: Bhumika Goyal <bhumirks@gmail.com>
Acked-by: Stefan Schmidt <stefan@osg.samsung.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'geneve-lwt-efficiency'
David S. Miller [Mon, 21 Nov 2016 19:05:50 +0000 (14:05 -0500)]
Merge branch 'geneve-lwt-efficiency'

Pravin B Shelar says:

====================
geneve: Use LWT more effectively.

Following patch series make use of geneve LWT code path for
geneve netdev type of device.
This allows us to simplify geneve module without changing any
functionality.

v2-v3:
Rebase against latest net-next.

v1-v2:
Fix warning reported by kbuild test robot.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agogeneve: Optimize geneve device lookup.
pravin shelar [Mon, 21 Nov 2016 19:03:01 +0000 (11:03 -0800)]
geneve: Optimize geneve device lookup.

Rather than comparing 64-bit tunnel-id, compare tunnel vni
which is 24-bit id. This also save conversion from vni
to tunnel id on each tunnel packet receive.

Signed-off-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agogeneve: Remove redundant socket checks.
pravin shelar [Mon, 21 Nov 2016 19:03:00 +0000 (11:03 -0800)]
geneve: Remove redundant socket checks.

Geneve already has check for device socket in route
lookup function. So no need to check it in xmit
function.

Signed-off-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agogeneve: Merge ipv4 and ipv6 geneve_build_skb()
pravin shelar [Mon, 21 Nov 2016 19:02:59 +0000 (11:02 -0800)]
geneve: Merge ipv4 and ipv6 geneve_build_skb()

There are minimal difference in building Geneve header
between ipv4 and ipv6 geneve tunnels. Following patch
refactors code to unify it.

Signed-off-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agogeneve: Unify LWT and netdev handling.
pravin shelar [Mon, 21 Nov 2016 19:02:58 +0000 (11:02 -0800)]
geneve: Unify LWT and netdev handling.

Current geneve implementation has two separate cases to handle.
1. netdev xmit
2. LWT xmit.

In case of netdev, geneve configuration is stored in various
struct geneve_dev members. For example geneve_addr, ttl, tos,
label, flags, dst_cache, etc. For LWT ip_tunnel_info is passed
to the device in ip_tunnel_info.

Following patch uses ip_tunnel_info struct to store almost all
of configuration of a geneve netdevice. This allows us to unify
most of geneve driver code around ip_tunnel_info struct.
This dramatically simplify geneve code, since it does not
need to handle two different configuration cases. Removes
duplicate code, single code path can handle either type
of geneve devices.

Signed-off-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'tcp-cong-undo_cwnd-mandatory'
David S. Miller [Mon, 21 Nov 2016 18:20:17 +0000 (13:20 -0500)]
Merge branch 'tcp-cong-undo_cwnd-mandatory'

Florian Westphal says:

====================
tcp: make undo_cwnd mandatory for congestion modules

highspeed, illinois, scalable, veno and yeah congestion control algorithms
don't provide a 'cwnd_undo' function.  This makes the stack default to a
'reno undo' which doubles cwnd.  However, the ssthresh implementation of
these algorithms do not halve the slowstart threshold. This causes similar
issue as the one fixed for dctcp in ce6dd23329b1e ("dctcp: avoid bogus
doubling of cwnd after loss").

In light of this it seems better to remove the fallback and make undo_cwnd
mandatory.

First patch fixes those spots where reno undo seems incorrect by providing
.cwnd_undo functions, second patch removes the fallback.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotcp: make undo_cwnd mandatory for congestion modules
Florian Westphal [Mon, 21 Nov 2016 13:18:38 +0000 (14:18 +0100)]
tcp: make undo_cwnd mandatory for congestion modules

The undo_cwnd fallback in the stack doubles cwnd based on ssthresh,
which un-does reno halving behaviour.

It seems more appropriate to let congctl algorithms pair .ssthresh
and .undo_cwnd properly. Add a 'tcp_reno_undo_cwnd' function and wire it
up for all congestion algorithms that used to rely on the fallback.

Cc: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotcp: add cwnd_undo functions to various tcp cc algorithms
Florian Westphal [Mon, 21 Nov 2016 13:18:37 +0000 (14:18 +0100)]
tcp: add cwnd_undo functions to various tcp cc algorithms

congestion control algorithms that do not halve cwnd in their .ssthresh
should provide a .cwnd_undo rather than rely on current fallback which
assumes reno halving (and thus doubles the cwnd).

All of these do 'something else' in their .ssthresh implementation, thus
store the cwnd on loss and provide .undo_cwnd to restore it again.

A followup patch will remove the fallback and all algorithms will
need to provide a .cwnd_undo function.

Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'bridge-igmpv3-mldv2-support'
David S. Miller [Mon, 21 Nov 2016 18:16:59 +0000 (13:16 -0500)]
Merge branch 'bridge-igmpv3-mldv2-support'

Nikolay Aleksandrov says:

====================
bridge: add support for IGMPv3 and MLDv2 querier

This patch-set adds support for IGMPv3 and MLDv2 querier in the bridge.
Two new options which can be toggled via netlink and sysfs are added that
control the version per-bridge:
 multicast_igmp_version - default 2, can be set to 3
 multicast_mld_version - default 1, can be set to 2 (this option is
                         disabled if CONFIG_IPV6=n)

Note that the names do not include "querier", I think that these options
can be re-used later as more IGMPv3 support is added to the bridge so we
can avoid adding more options to switch between v2 and v3 behaviour.

The set uses the already existing br_ip{4,6}_multicast_alloc_query
functions and adds the appropriate header based on the chosen version.

For the initial support I have removed the compatibility implementation
(RFC3376 sec 7.3.1, 7.3.2; RFC3810 sec 8.3.1, 8.3.2), because there are
some details that we need to sort out.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobridge: mcast: add MLDv2 querier support
Nikolay Aleksandrov [Mon, 21 Nov 2016 12:03:25 +0000 (13:03 +0100)]
bridge: mcast: add MLDv2 querier support

This patch adds basic support for MLDv2 queries, the default is MLDv1
as before. A new multicast option - multicast_mld_version, adds the
ability to change it between 1 and 2 via netlink and sysfs.
The MLD option is disabled if CONFIG_IPV6 is disabled.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobridge: mcast: add IGMPv3 query support
Nikolay Aleksandrov [Mon, 21 Nov 2016 12:03:24 +0000 (13:03 +0100)]
bridge: mcast: add IGMPv3 query support

This patch adds basic support for IGMPv3 queries, the default is IGMPv2
as before. A new multicast option - multicast_igmp_version, adds the
ability to change it between 2 and 3 via netlink and sysfs. The option
struct member is in a 4 byte hole in net_bridge.

There also a few minor style adjustments in br_multicast_new_group and
br_multicast_add_group.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotcp: zero ca_priv area when switching cc algorithms
Florian Westphal [Mon, 21 Nov 2016 09:08:37 +0000 (10:08 +0100)]
tcp: zero ca_priv area when switching cc algorithms

We need to zero out the private data area when application switches
connection to different algorithm (TCP_CONGESTION setsockopt).

When congestion ops get assigned at connect time everything is already
zeroed because sk_alloc uses GFP_ZERO flag.  But in the setsockopt case
this contains whatever previous cc placed there.

Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: l2tp: Treat NET_XMIT_CN as success in l2tp_eth_dev_xmit
Gao Feng [Mon, 21 Nov 2016 00:56:21 +0000 (08:56 +0800)]
net: l2tp: Treat NET_XMIT_CN as success in l2tp_eth_dev_xmit

The tc could return NET_XMIT_CN as one congestion notification, but
it does not mean the packe is lost. Other modules like ipvlan,
macvlan, and others treat NET_XMIT_CN as success too.
So l2tp_eth_dev_xmit should add the NET_XMIT_CN check.

Signed-off-by: Gao Feng <gfree.wind@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agodriver: macvlan: Remove duplicated IFF_UP condition check in macvlan_forward_source
Gao Feng [Mon, 21 Nov 2016 00:26:38 +0000 (08:26 +0800)]
driver: macvlan: Remove duplicated IFF_UP condition check in macvlan_forward_source

The function macvlan_forward_source_one has already checked the flag
IFF_UP, so needn't check it outside in macvlan_forward_source too.

Signed-off-by: Gao Feng <gfree.wind@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoethernet: stmmac: make DWMAC_STM32 depend on it's associated SoC
Peter Robinson [Sun, 20 Nov 2016 17:22:38 +0000 (17:22 +0000)]
ethernet: stmmac: make DWMAC_STM32 depend on it's associated SoC

There's not much point, except compile test, enabling the stmmac
platform drivers unless the STM32 SoC is enabled. It's not
useful without it.

Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agomlx4: avoid unnecessary dirtying of critical fields
Eric Dumazet [Sun, 20 Nov 2016 17:24:36 +0000 (09:24 -0800)]
mlx4: avoid unnecessary dirtying of critical fields

While stressing a 40Gbit mlx4 NIC with busy polling, I found false
sharing in mlx4 driver that can be easily avoided.

This patch brings an additional 7 % performance improvement in UDP_RR
workload.

1) If we received no frame during one mlx4_en_process_rx_cq()
   invocation, no need to call mlx4_cq_set_ci() and/or dirty ring->cons

2) Do not refill rx buffers if we have plenty of them.
   This avoids false sharing and allows some bulk/batch optimizations.
   Page allocator and its locks will thank us.

Finally, mlx4_en_poll_rx_cq() should not return 0 if it determined
cpu handling NIC IRQ should be changed. We should return budget-1
instead, to not fool net_rx_action() and its netdev_budget.

v2: keep AVG_PERF_COUNTER(... polled) even if polled is 0

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobnx2: use READ_ONCE() instead of barrier()
Eric Dumazet [Sat, 19 Nov 2016 22:57:11 +0000 (14:57 -0800)]
bnx2: use READ_ONCE() instead of barrier()

barrier() is a big hammer compared to READ_ONCE(),
and requires comments explaining what is protected.

READ_ONCE() is more precise and compiler should generate
better overall code.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoudp: avoid one cache line miss in recvmsg()
Eric Dumazet [Sat, 19 Nov 2016 01:18:03 +0000 (17:18 -0800)]
udp: avoid one cache line miss in recvmsg()

UDP_SKB_CB(skb)->partial_cov is located at offset 66 in skb,
requesting a cold cache line being read in cpu cache.

We can avoid this cache line miss for UDP sockets,
as partial_cov has a meaning only for UDPLite.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'mlx5-bpf-refcnt-fixes'
David S. Miller [Mon, 21 Nov 2016 16:25:59 +0000 (11:25 -0500)]
Merge branch 'mlx5-bpf-refcnt-fixes'

Daniel Borkmann says:

====================
Couple of BPF refcount fixes for mlx5

Various mlx5 bugs on eBPF refcount handling found during review.
Last patch in series adds a __must_check to BPF helpers to make
sure we won't run into it again w/o compiler complaining first.

v2 -> v3:

 - Just reworked patch 2/4 so we don't need bpf_prog_sub().
 - Rebased, rest as is.

v1 -> v2:

 - After discussion with Alexei, we agreed upon rebasing the
   patches against net-next.
 - Since net-next, I've also added the __must_check to enforce
   future users to check for errors.
 - Fixed up commit message #2.
 - Simplify assignment from patch #1 based on Saeed's feedback
   on previous set.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobpf: add __must_check attributes to refcount manipulating helpers
Daniel Borkmann [Sat, 19 Nov 2016 00:45:03 +0000 (01:45 +0100)]
bpf: add __must_check attributes to refcount manipulating helpers

Helpers like bpf_prog_add(), bpf_prog_inc(), bpf_map_inc() can fail
with an error, so make sure the caller properly checks their return
value and not just ignores it, which could worst-case lead to use
after free.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobpf, mlx5: drop priv->xdp_prog reference on netdev cleanup
Daniel Borkmann [Sat, 19 Nov 2016 00:45:02 +0000 (01:45 +0100)]
bpf, mlx5: drop priv->xdp_prog reference on netdev cleanup

mlx5e_xdp_set() is currently the only place where we drop reference on the
prog sitting in priv->xdp_prog when it's exchanged by a new one. We also
need to make sure that we eventually release that reference, for example,
in case the netdev is dismantled, otherwise we leak the program.

Fixes: 86994156c736 ("net/mlx5e: XDP fast RX drop bpf programs support")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobpf, mlx5: fix various refcount issues in mlx5e_xdp_set
Daniel Borkmann [Sat, 19 Nov 2016 00:45:01 +0000 (01:45 +0100)]
bpf, mlx5: fix various refcount issues in mlx5e_xdp_set

There are multiple issues in mlx5e_xdp_set():

1) The batched bpf_prog_add() is currently not checked for errors. When
   doing so, it should be done at an earlier point in time to makes sure
   that we cannot fail anymore at the time we want to set the program for
   each channel. The batched refs short-cut can only be performed when we
   don't need to perform a reset for changing the rq type and the device
   was in opened state. In case the device was not in opened state, then
   the next mlx5e_open_locked() will aquire the refs from the control prog
   via mlx5e_create_rq(), same when we need to perform a reset.

2) When swapping the priv->xdp_prog, then no extra reference count must be
   taken since we got that from call path via dev_change_xdp_fd() already.
   Otherwise, we'd never be able to release the program. Also, bpf_prog_add()
   without checking the return code could fail.

Fixes: 86994156c736 ("net/mlx5e: XDP fast RX drop bpf programs support")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobpf, mlx5: fix mlx5e_create_rq taking reference on prog
Daniel Borkmann [Sat, 19 Nov 2016 00:45:00 +0000 (01:45 +0100)]
bpf, mlx5: fix mlx5e_create_rq taking reference on prog

In mlx5e_create_rq(), when creating a new queue, we call bpf_prog_add() but
without checking the return value. bpf_prog_add() can fail since 92117d8443bc
("bpf: fix refcnt overflow"), so we really must check it. Take the reference
right when we assign it to the rq from priv->xdp_prog, and just drop the
reference on error path. Destruction in mlx5e_destroy_rq() looks good, though.

Fixes: 86994156c736 ("net/mlx5e: XDP fast RX drop bpf programs support")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoapparmor: fix change_hat not finding hat after policy replacement
John Johansen [Thu, 1 Sep 2016 04:10:06 +0000 (21:10 -0700)]
apparmor: fix change_hat not finding hat after policy replacement

After a policy replacement, the task cred may be out of date and need
to be updated. However change_hat is using the stale profiles from
the out of date cred resulting in either: a stale profile being applied
or, incorrect failure when searching for a hat profile as it has been
migrated to the new parent profile.

Fixes: 01e2b670aa898a39259bc85c78e3d74820f4d3b6 (failure to find hat)
Fixes: 898127c34ec03291c86f4ff3856d79e9e18952bc (stale policy being applied)
Bugzilla: https://bugzilla.suse.com/show_bug.cgi?id=1000287
Cc: stable@vger.kernel.org
Signed-off-by: John Johansen <john.johansen@canonical.com>
Signed-off-by: James Morris <james.l.morris@oracle.com>
8 years agoMerge branch 'mV88e6xxx-interrupt-fixes'
David S. Miller [Mon, 21 Nov 2016 02:16:14 +0000 (21:16 -0500)]
Merge branch 'mV88e6xxx-interrupt-fixes'

Andrew Lunn says:

====================
Fixes for the MV88e6xxx interrupt code

The interrupt code was never tested with a board who's probing
resulted in an -EPROBE_DEFFERED. So the clean up paths never got
tested. I now do have -EPROBE_DEFFERED, and things break badly during
cleanup. These are the fixes.

This is fixing code in net-next.

v2:
Fix typo pointed out by David Miller
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Hold the mutex while freeing g1 interrupts
Andrew Lunn [Sun, 20 Nov 2016 19:14:19 +0000 (20:14 +0100)]
net: dsa: mv88e6xxx: Hold the mutex while freeing g1 interrupts

Freeing interrupts requires switch register access to mask the
interrupts. Hence we must hold the register mutex.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Fix releasing for the global2 interrupts
Andrew Lunn [Sun, 20 Nov 2016 19:14:18 +0000 (20:14 +0100)]
net: dsa: mv88e6xxx: Fix releasing for the global2 interrupts

It is not possible to use devm_request_threaded_irq() because we have
two stacked interrupt controllers in one device. The lower interrupt
controller cannot be removed until the upper is fully removed. This
happens too late with the devm API, resulting in error messages about
removing a domain while there is still an active interrupt. Swap to
using request_threaded_irq() and manage the release of the interrupt
manually.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Fix cleanup on error for g1 interrupt setup
Andrew Lunn [Sun, 20 Nov 2016 19:14:17 +0000 (20:14 +0100)]
net: dsa: mv88e6xxx: Fix cleanup on error for g1 interrupt setup

On error, remask the interrupts, release all maps, and remove the
domain. This cannot be done using the mv88e6xxx_g1_irq_free() because
some of these actions are not idempotent.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Mask g1 interrupts and free interrupt
Andrew Lunn [Sun, 20 Nov 2016 19:14:16 +0000 (20:14 +0100)]
net: dsa: mv88e6xxx: Mask g1 interrupts and free interrupt

Fix the g1 interrupt free code such that is masks any further
interrupts, and then releases the interrupt.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Fix unconditional irq freeing
Andrew Lunn [Sun, 20 Nov 2016 19:14:15 +0000 (20:14 +0100)]
net: dsa: mv88e6xxx: Fix unconditional irq freeing

Trying to remove an IRQ domain that was not created results in an
Opps. Add the necessary checks that the irqs were created before
freeing them.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: mv88e6xxx: Fix typos when removing g1 interrupts
Andrew Lunn [Sun, 20 Nov 2016 19:14:14 +0000 (20:14 +0100)]
net: dsa: mv88e6xxx: Fix typos when removing g1 interrupts

Simple typos, s/g2/g1/

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoLinux 4.9-rc6
Linus Torvalds [Sun, 20 Nov 2016 21:52:19 +0000 (13:52 -0800)]
Linux 4.9-rc6

8 years agoMerge branch 'fixes' of git://git.armlinux.org.uk/~rmk/linux-arm
Linus Torvalds [Sun, 20 Nov 2016 18:27:39 +0000 (10:27 -0800)]
Merge branch 'fixes' of git://git.armlinux.org.uk/~rmk/linux-arm

Pull ARM fixes from Russell King:
 "A few more ARM fixes:

   - the assembly backtrace code suffers problems with the new printk()
     implementation which assumes that kernel messages without KERN_CONT
     should have newlines inserted between them. Fix this.
   - fix a section naming error - ".init.text" rather than ".text.init"
   - preallocate DMA debug memory at core_initcall() time rather than
     fs_initcall(), as we have some core drivers that need to use DMA
     mapping - and that triggers a kernel warning from the DMA debug
     code.
   - fix XIP kernels after the ro_after_init changes made this data
     permanently read-only"

* 'fixes' of git://git.armlinux.org.uk/~rmk/linux-arm:
  ARM: Fix XIP kernels
  ARM: 8628/1: dma-mapping: preallocate DMA-debug hash tables in core_initcall
  ARM: 8624/1: proc-v7m.S: fix init section name
  ARM: fix backtrace

8 years agotipc: eliminate obsolete socket locking policy description
Jon Paul Maloy [Sat, 19 Nov 2016 19:47:07 +0000 (14:47 -0500)]
tipc: eliminate obsolete socket locking policy description

The comment block in socket.c describing the locking policy is
obsolete, and does not reflect current reality. We remove it in this
commit.

Since the current locking policy is much simpler and follows a
mainstream approach, we see no need to add a new description.

Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agortnl: fix the loop index update error in rtnl_dump_ifinfo()
Zhang Shengju [Sat, 19 Nov 2016 15:28:32 +0000 (23:28 +0800)]
rtnl: fix the loop index update error in rtnl_dump_ifinfo()

If the link is filtered out, loop index should also be updated. If not,
loop index will not be correct.

Fixes: dc599f76c22b0 ("net: Add support for filtering link dump by master device and kind")
Signed-off-by: Zhang Shengju <zhangshengju@cmss.chinamobile.com>
Acked-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: fix bogus cast in skb_pagelen() and use unsigned variables
Alexey Dobriyan [Sat, 19 Nov 2016 01:08:08 +0000 (04:08 +0300)]
net: fix bogus cast in skb_pagelen() and use unsigned variables

1) cast to "int" is unnecessary:
   u8 will be promoted to int before decrementing,
   small positive numbers fit into "int", so their values won't be changed
   during promotion.

   Once everything is int including loop counters, signedness doesn't
   matter: 32-bit operations will stay 32-bit operations.

   But! Someone tried to make this loop smart by making everything of
   the same type apparently in an attempt to optimise it.
   Do the optimization, just differently.
   Do the cast where it matters. :^)

2) frag size is unsigned entity and sum of fragments sizes is also
   unsigned.

Make everything unsigned, leave no MOVSX instruction behind.

add/remove: 0/0 grow/shrink: 0/3 up/down: 0/-4 (-4)
function                                     old     new   delta
skb_cow_data                                 835     834      -1
ip_do_fragment                              2549    2548      -1
ip6_fragment                                3130    3128      -2
Total: Before=154865032, After=154865028, chg -0.00%

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonetlink: smaller nla_attr_minlen table
Alexey Dobriyan [Sat, 19 Nov 2016 00:59:07 +0000 (03:59 +0300)]
netlink: smaller nla_attr_minlen table

Length of a netlink attribute may be u16 but lengths of basic attributes
are much smaller, so small we can save 16 bytes of .rodata and pocket
change inside .text.

16-bit is worse on x86-64 than 8-bit because of operand size override prefix.

add/remove: 0/0 grow/shrink: 0/3 up/down: 0/-19 (-19)
function                                     old     new   delta
validate_nla                                 418     417      -1
nla_policy_len                                66      64      -2
nla_attr_minlen                               32      16     -16
Total: Before=154865051, After=154865032, chg -0.00%

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonetlink: use "unsigned int" in nla_next()
Alexey Dobriyan [Sat, 19 Nov 2016 00:54:35 +0000 (03:54 +0300)]
netlink: use "unsigned int" in nla_next()

->nla_len is unsigned entity (it's length after all) and u16,
thus it can't overflow when being aligned into int/unsigned int.

(nlmsg_next has the same code, but I didn't yet convince myself
it is correct to do so).

There is pointer arithmetic in this function and offset being
unsigned is better:

add/remove: 0/0 grow/shrink: 1/64 up/down: 5/-309 (-304)
function                                     old     new   delta
nl80211_set_wiphy                           1444    1449      +5
team_nl_cmd_options_set                      997     995      -2
tcf_em_tree_validate                         872     870      -2
switchdev_port_bridge_setlink                352     350      -2
switchdev_port_br_afspec                     312     310      -2
rtm_to_fib_config                            428     426      -2
qla4xxx_sysfs_ddb_set_param                 2193    2191      -2
qla4xxx_iface_set_param                     4470    4468      -2
ovs_nla_free_flow_actions                    152     150      -2
output_userspace                             518     516      -2
...
nl80211_set_reg                              654     649      -5
validate_scan_freqs                          148     142      -6
validate_linkmsg                             288     282      -6
nl80211_parse_connkeys                       489     483      -6
nlattr_set                                   231     224      -7
nf_tables_delsetelem                         267     260      -7
do_setlink                                  3416    3408      -8
netlbl_cipsov4_add_std                      1672    1659     -13
nl80211_parse_sched_scan                    2902    2888     -14
nl80211_trigger_scan                        1738    1720     -18
do_execute_actions                          2821    2738     -83
Total: Before=154865355, After=154865051, chg -0.00%

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: make struct napi_alloc_cache::skb_count unsigned int
Alexey Dobriyan [Sat, 19 Nov 2016 00:47:56 +0000 (03:47 +0300)]
net: make struct napi_alloc_cache::skb_count unsigned int

size_t is way too much for an integer not exceeding 64.

Space savings: 10 bytes!

add/remove: 0/0 grow/shrink: 0/3 up/down: 0/-10 (-10)
function                                     old     new   delta
napi_consume_skb                             165     163      -2
__kfree_skb_flush                             56      53      -3
__kfree_skb_defer                             97      92      -5
Total: Before=154865639, After=154865629, chg -0.00%

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agol2tp: fix racy SOCK_ZAPPED flag check in l2tp_ip{,6}_bind()
Guillaume Nault [Fri, 18 Nov 2016 21:13:00 +0000 (22:13 +0100)]
l2tp: fix racy SOCK_ZAPPED flag check in l2tp_ip{,6}_bind()

Lock socket before checking the SOCK_ZAPPED flag in l2tp_ip6_bind().
Without lock, a concurrent call could modify the socket flags between
the sock_flag(sk, SOCK_ZAPPED) test and the lock_sock() call. This way,
a socket could be inserted twice in l2tp_ip6_bind_table. Releasing it
would then leave a stale pointer there, generating use-after-free
errors when walking through the list or modifying adjacent entries.

BUG: KASAN: use-after-free in l2tp_ip6_close+0x22e/0x290 at addr ffff8800081b0ed8
Write of size 8 by task syz-executor/10987
CPU: 0 PID: 10987 Comm: syz-executor Not tainted 4.8.0+ #39
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.8.2-0-g33fbe13 by qemu-project.org 04/01/2014
 ffff880031d97838 ffffffff829f835b ffff88001b5a1640 ffff8800081b0ec0
 ffff8800081b15a0 ffff8800081b6d20 ffff880031d97860 ffffffff8174d3cc
 ffff880031d978f0 ffff8800081b0e80 ffff88001b5a1640 ffff880031d978e0
Call Trace:
 [<ffffffff829f835b>] dump_stack+0xb3/0x118 lib/dump_stack.c:15
 [<ffffffff8174d3cc>] kasan_object_err+0x1c/0x70 mm/kasan/report.c:156
 [<     inline     >] print_address_description mm/kasan/report.c:194
 [<ffffffff8174d666>] kasan_report_error+0x1f6/0x4d0 mm/kasan/report.c:283
 [<     inline     >] kasan_report mm/kasan/report.c:303
 [<ffffffff8174db7e>] __asan_report_store8_noabort+0x3e/0x40 mm/kasan/report.c:329
 [<     inline     >] __write_once_size ./include/linux/compiler.h:249
 [<     inline     >] __hlist_del ./include/linux/list.h:622
 [<     inline     >] hlist_del_init ./include/linux/list.h:637
 [<ffffffff8579047e>] l2tp_ip6_close+0x22e/0x290 net/l2tp/l2tp_ip6.c:239
 [<ffffffff850b2dfd>] inet_release+0xed/0x1c0 net/ipv4/af_inet.c:415
 [<ffffffff851dc5a0>] inet6_release+0x50/0x70 net/ipv6/af_inet6.c:422
 [<ffffffff84c4581d>] sock_release+0x8d/0x1d0 net/socket.c:570
 [<ffffffff84c45976>] sock_close+0x16/0x20 net/socket.c:1017
 [<ffffffff817a108c>] __fput+0x28c/0x780 fs/file_table.c:208
 [<ffffffff817a1605>] ____fput+0x15/0x20 fs/file_table.c:244
 [<ffffffff813774f9>] task_work_run+0xf9/0x170
 [<ffffffff81324aae>] do_exit+0x85e/0x2a00
 [<ffffffff81326dc8>] do_group_exit+0x108/0x330
 [<ffffffff81348cf7>] get_signal+0x617/0x17a0 kernel/signal.c:2307
 [<ffffffff811b49af>] do_signal+0x7f/0x18f0
 [<ffffffff810039bf>] exit_to_usermode_loop+0xbf/0x150 arch/x86/entry/common.c:156
 [<     inline     >] prepare_exit_to_usermode arch/x86/entry/common.c:190
 [<ffffffff81006060>] syscall_return_slowpath+0x1a0/0x1e0 arch/x86/entry/common.c:259
 [<ffffffff85e4d726>] entry_SYSCALL_64_fastpath+0xc4/0xc6
Object at ffff8800081b0ec0, in cache L2TP/IPv6 size: 1448
Allocated:
PID = 10987
 [ 1116.897025] [<ffffffff811ddcb6>] save_stack_trace+0x16/0x20
 [ 1116.897025] [<ffffffff8174c736>] save_stack+0x46/0xd0
 [ 1116.897025] [<ffffffff8174c9ad>] kasan_kmalloc+0xad/0xe0
 [ 1116.897025] [<ffffffff8174cee2>] kasan_slab_alloc+0x12/0x20
 [ 1116.897025] [<     inline     >] slab_post_alloc_hook mm/slab.h:417
 [ 1116.897025] [<     inline     >] slab_alloc_node mm/slub.c:2708
 [ 1116.897025] [<     inline     >] slab_alloc mm/slub.c:2716
 [ 1116.897025] [<ffffffff817476a8>] kmem_cache_alloc+0xc8/0x2b0 mm/slub.c:2721
 [ 1116.897025] [<ffffffff84c4f6a9>] sk_prot_alloc+0x69/0x2b0 net/core/sock.c:1326
 [ 1116.897025] [<ffffffff84c58ac8>] sk_alloc+0x38/0xae0 net/core/sock.c:1388
 [ 1116.897025] [<ffffffff851ddf67>] inet6_create+0x2d7/0x1000 net/ipv6/af_inet6.c:182
 [ 1116.897025] [<ffffffff84c4af7b>] __sock_create+0x37b/0x640 net/socket.c:1153
 [ 1116.897025] [<     inline     >] sock_create net/socket.c:1193
 [ 1116.897025] [<     inline     >] SYSC_socket net/socket.c:1223
 [ 1116.897025] [<ffffffff84c4b46f>] SyS_socket+0xef/0x1b0 net/socket.c:1203
 [ 1116.897025] [<ffffffff85e4d685>] entry_SYSCALL_64_fastpath+0x23/0xc6
Freed:
PID = 10987
 [ 1116.897025] [<ffffffff811ddcb6>] save_stack_trace+0x16/0x20
 [ 1116.897025] [<ffffffff8174c736>] save_stack+0x46/0xd0
 [ 1116.897025] [<ffffffff8174cf61>] kasan_slab_free+0x71/0xb0
 [ 1116.897025] [<     inline     >] slab_free_hook mm/slub.c:1352
 [ 1116.897025] [<     inline     >] slab_free_freelist_hook mm/slub.c:1374
 [ 1116.897025] [<     inline     >] slab_free mm/slub.c:2951
 [ 1116.897025] [<ffffffff81748b28>] kmem_cache_free+0xc8/0x330 mm/slub.c:2973
 [ 1116.897025] [<     inline     >] sk_prot_free net/core/sock.c:1369
 [ 1116.897025] [<ffffffff84c541eb>] __sk_destruct+0x32b/0x4f0 net/core/sock.c:1444
 [ 1116.897025] [<ffffffff84c5aca4>] sk_destruct+0x44/0x80 net/core/sock.c:1452
 [ 1116.897025] [<ffffffff84c5ad33>] __sk_free+0x53/0x220 net/core/sock.c:1460
 [ 1116.897025] [<ffffffff84c5af23>] sk_free+0x23/0x30 net/core/sock.c:1471
 [ 1116.897025] [<ffffffff84c5cb6c>] sk_common_release+0x28c/0x3e0 ./include/net/sock.h:1589
 [ 1116.897025] [<ffffffff8579044e>] l2tp_ip6_close+0x1fe/0x290 net/l2tp/l2tp_ip6.c:243
 [ 1116.897025] [<ffffffff850b2dfd>] inet_release+0xed/0x1c0 net/ipv4/af_inet.c:415
 [ 1116.897025] [<ffffffff851dc5a0>] inet6_release+0x50/0x70 net/ipv6/af_inet6.c:422
 [ 1116.897025] [<ffffffff84c4581d>] sock_release+0x8d/0x1d0 net/socket.c:570
 [ 1116.897025] [<ffffffff84c45976>] sock_close+0x16/0x20 net/socket.c:1017
 [ 1116.897025] [<ffffffff817a108c>] __fput+0x28c/0x780 fs/file_table.c:208
 [ 1116.897025] [<ffffffff817a1605>] ____fput+0x15/0x20 fs/file_table.c:244
 [ 1116.897025] [<ffffffff813774f9>] task_work_run+0xf9/0x170
 [ 1116.897025] [<ffffffff81324aae>] do_exit+0x85e/0x2a00
 [ 1116.897025] [<ffffffff81326dc8>] do_group_exit+0x108/0x330
 [ 1116.897025] [<ffffffff81348cf7>] get_signal+0x617/0x17a0 kernel/signal.c:2307
 [ 1116.897025] [<ffffffff811b49af>] do_signal+0x7f/0x18f0
 [ 1116.897025] [<ffffffff810039bf>] exit_to_usermode_loop+0xbf/0x150 arch/x86/entry/common.c:156
 [ 1116.897025] [<     inline     >] prepare_exit_to_usermode arch/x86/entry/common.c:190
 [ 1116.897025] [<ffffffff81006060>] syscall_return_slowpath+0x1a0/0x1e0 arch/x86/entry/common.c:259
 [ 1116.897025] [<ffffffff85e4d726>] entry_SYSCALL_64_fastpath+0xc4/0xc6
Memory state around the buggy address:
 ffff8800081b0d80: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
 ffff8800081b0e00: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
>ffff8800081b0e80: fc fc fc fc fc fc fc fc fb fb fb fb fb fb fb fb
                                                    ^
 ffff8800081b0f00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
 ffff8800081b0f80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb

==================================================================

The same issue exists with l2tp_ip_bind() and l2tp_ip_bind_table.

Fixes: c51ce49735c1 ("l2tp: fix oops in L2TP IP sockets for connect() AF_UNSPEC case")
Reported-by: Baozeng Ding <sploving1@gmail.com>
Reported-by: Andrey Konovalov <andreyknvl@google.com>
Tested-by: Baozeng Ding <sploving1@gmail.com>
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge tag 'armsoc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Sun, 20 Nov 2016 02:40:47 +0000 (18:40 -0800)]
Merge tag 'armsoc-fixes' of git://git./linux/kernel/git/arm/arm-soc

Pull ARM SoC fixes from Olof Johansson:
 "Again a set of smaller fixes across several platforms (OMAP, Marvell,
  Allwinner, i.MX, etc).

  A handful of typo fixes and smaller missing contents from device
  trees, with some tweaks to OMAP mach files to deal with CPU feature
  print misformatting, potential NULL ptr dereference and one setup
  issue with UARTs"

* tag 'armsoc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  ipmi/bt-bmc: change compatible node to 'aspeed, ast2400-ibt-bmc'
  ARM: dts: STiH410-b2260: Fix typo in spi0 chipselect definition
  ARM: dts: omap5: board-common: fix wrong SMPS6 (VDD-DDR3) voltage
  ARM: omap3: Add missing memory node in SOM-LV
  arm64: dts: marvell: add unique identifiers for Armada A8k SPI controllers
  arm64: dts: marvell: fix clocksource for CP110 slave SPI0
  arm64: dts: marvell: Fix typo in label name on Armada 37xx
  ASoC: omap-abe-twl6040: fix typo in bindings documentation
  dts: omap5: board-common: enable twl6040 headset jack detection
  dts: omap5: board-common: add phandle to reference Palmas gpadc
  ARM: OMAP2+: avoid NULL pointer dereference
  ARM: OMAP2+: PRM: initialize en_uart4_mask and grpsel_uart4_mask
  ARM: dts: omap3: Fix memory node in Torpedo board
  ARM: AM43XX: Select OMAP_INTERCONNECT in Kconfig
  ARM: OMAP3: Fix formatting of features printed
  ARM: dts: imx53-qsb: Fix regulator constraints
  ARM: dts: sun8i: fix the pinmux for UART1

8 years agoMerge tag 'ext4_for_stable' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso...
Linus Torvalds [Sun, 20 Nov 2016 02:33:50 +0000 (18:33 -0800)]
Merge tag 'ext4_for_stable' of git://git./linux/kernel/git/tytso/ext4

Pull ext4 fixes from Ted Ts'o:
 "A security fix (so a maliciously corrupted file system image won't
  panic the kernel) and some fixes for CONFIG_VMAP_STACK"

* tag 'ext4_for_stable' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: sanity check the block and cluster size at mount time
  fscrypto: don't use on-stack buffer for key derivation
  fscrypto: don't use on-stack buffer for filename encryption

8 years agoext4: sanity check the block and cluster size at mount time
Theodore Ts'o [Fri, 18 Nov 2016 18:00:24 +0000 (13:00 -0500)]
ext4: sanity check the block and cluster size at mount time

If the block size or cluster size is insane, reject the mount.  This
is important for security reasons (although we shouldn't be just
depending on this check).

Ref: http://www.securityfocus.com/archive/1/539661
Ref: https://bugzilla.redhat.com/show_bug.cgi?id=1332506
Reported-by: Borislav Petkov <bp@alien8.de>
Reported-by: Nikolay Borisov <kernel@kyup.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Cc: stable@vger.kernel.org
8 years agofscrypto: don't use on-stack buffer for key derivation
Eric Biggers [Mon, 14 Nov 2016 01:41:09 +0000 (20:41 -0500)]
fscrypto: don't use on-stack buffer for key derivation

With the new (in 4.9) option to use a virtually-mapped stack
(CONFIG_VMAP_STACK), stack buffers cannot be used as input/output for
the scatterlist crypto API because they may not be directly mappable to
struct page.  get_crypt_info() was using a stack buffer to hold the
output from the encryption operation used to derive the per-file key.
Fix it by using a heap buffer.

This bug could most easily be observed in a CONFIG_DEBUG_SG kernel
because this allowed the BUG in sg_set_buf() to be triggered.

Cc: stable@vger.kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
8 years agofscrypto: don't use on-stack buffer for filename encryption
Eric Biggers [Mon, 14 Nov 2016 01:35:52 +0000 (20:35 -0500)]
fscrypto: don't use on-stack buffer for filename encryption

With the new (in 4.9) option to use a virtually-mapped stack
(CONFIG_VMAP_STACK), stack buffers cannot be used as input/output for
the scatterlist crypto API because they may not be directly mappable to
struct page.  For short filenames, fname_encrypt() was encrypting a
stack buffer holding the padded filename.  Fix it by encrypting the
filename in-place in the output buffer, thereby making the temporary
buffer unnecessary.

This bug could most easily be observed in a CONFIG_DEBUG_SG kernel
because this allowed the BUG in sg_set_buf() to be triggered.

Cc: stable@vger.kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
8 years agoMerge branch 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Sat, 19 Nov 2016 21:35:09 +0000 (13:35 -0800)]
Merge branch 'i2c/for-current' of git://git./linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:
 "Some I2C driver bugfixes (and one documentation fix)"

* 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: i2c-mux-pca954x: fix deselect enabling for device-tree
  i2c: digicolor: use clk_disable_unprepare instead of clk_unprepare
  i2c: mux: fix up dependencies
  i2c: Documentation: i2c-topology: fix minor whitespace nit
  i2c: mux: demux-pinctrl: make drivers with no pinctrl work again

8 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Sat, 19 Nov 2016 21:31:40 +0000 (13:31 -0800)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull KVM fixes from Radim Krčmář:
 "ARM:
   - Fix handling of the 32bit cycle counter
   - Fix cycle counter filtering

  x86:
   - Fix a race leading to double unregistering of user notifiers
   - Amend oversight in kvm_arch_set_irq that turned Hyper-V code dead
   - Use SRCU around kvm_lapic_set_vapic_addr
   - Avoid recursive flushing of asynchronous page faults
   - Do not rely on deferred update in KVM_GET_CLOCK, which fixes #GP
   - Let userspace know that KVM_GET_CLOCK is useful with master clock;
     4.9 changed the return value to better match the guest clock, but
     didn't provide means to let guests take advantage of it"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  kvm: x86: merge kvm_arch_set_irq and kvm_arch_set_irq_inatomic
  KVM: x86: fix missed SRCU usage in kvm_lapic_set_vapic_addr
  KVM: async_pf: avoid recursive flushing of work items
  kvm: kvmclock: let KVM_GET_CLOCK return whether the master clock is in use
  KVM: Disable irq while unregistering user notifier
  KVM: x86: do not go through vcpu in __get_kvmclock_ns
  KVM: arm64: Fix the issues when guest PMCCFILTR is configured
  arm64: KVM: pmu: Fix AArch32 cycle counter access

8 years agoi2c: i2c-mux-pca954x: fix deselect enabling for device-tree
Alex Hemme [Sat, 19 Nov 2016 09:48:38 +0000 (10:48 +0100)]
i2c: i2c-mux-pca954x: fix deselect enabling for device-tree

Deselect functionality can be ignored for device-trees with
"i2c-mux-idle-disconnect" entries if no platform_data is available.
By enabling the deselect functionality outside the platform_data
block the logic works as it did in previous kernels.

Fixes: 7fcac9807175 ("i2c: i2c-mux-pca954x: convert to use an explicit i2c mux core")
Cc: <stable@vger.kernel.org> # v4.7+
Signed-off-by: Alex Hemme <ahemme@cisco.com>
Signed-off-by: Ziyang Wu <ziywu@cisco.com>
[touched up a few minor issues /peda]
Signed-off-by: Peter Rosin <peda@axentia.se>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
8 years agoMerge tag 'powerpc-4.9-5' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc...
Linus Torvalds [Sat, 19 Nov 2016 19:21:59 +0000 (11:21 -0800)]
Merge tag 'powerpc-4.9-5' of git://git./linux/kernel/git/powerpc/linux

Pull powerpc fixes from Michael Ellerman:
 "Fixes marked for stable:
   - fix system reset interrupt winkle wakeups
   - fix setting of AIL in hypervisor mode

  Fixes for code merged this cycle:
   - fix exception vector build with 2.23 era binutils
   - fix missing update of HID register on secondary CPUs

  Other:
   - fix missing pr_cont()s
   - invalidate ERAT on tlbiel for POWER9 DD1"

* tag 'powerpc-4.9-5' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/mm: Fix missing update of HID register on secondary CPUs
  powerpc/mm/radix: Invalidate ERAT on tlbiel for POWER9 DD1
  powerpc/64: Fix setting of AIL in hypervisor mode
  powerpc/oops: Fix missing pr_cont()s in instruction dump
  powerpc/oops: Fix missing pr_cont()s in show_regs()
  powerpc/oops: Fix missing pr_cont()s in print_msr_bits() et. al.
  powerpc/oops: Fix missing pr_cont()s in show_stack()
  powerpc: Fix exception vector build with 2.23 era binutils
  powerpc/64s: Fix system reset interrupt winkle wakeups

8 years agoMerge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Sat, 19 Nov 2016 19:15:45 +0000 (11:15 -0800)]
Merge branch 'linus' of git://git./linux/kernel/git/herbert/crypto-2.6

Pull crypto fixes from Herbert Xu:
 "This fixes the following issues:

   - Compiler warning in caam driver that was the last one remaining

   - Do not register aes-xts in caam drivers on unsupported platforms

   - Regression in algif_hash interface that may lead to an oops"

* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: algif_hash - Fix NULL hash crash with shash
  crypto: caam - fix type mismatch warning
  crypto: caam - do not register AES-XTS mode on LP units

8 years agoMerge tag 'leds_4.9-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/j.anaszewsk...
Linus Torvalds [Sat, 19 Nov 2016 19:09:28 +0000 (11:09 -0800)]
Merge tag 'leds_4.9-rc6' of git://git./linux/kernel/git/j.anaszewski/linux-leds

Pull LED subsystem update from Jacek Anaszewski:
 "I'd like to announce a new co-maintainer - Pavel Machek"

* tag 'leds_4.9-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/j.anaszewski/linux-leds:
  MAINTAINERS: Add LED subsystem co-maintainer

8 years agoMerge tag 'dmaengine-fix-4.9-rc6' of git://git.infradead.org/users/vkoul/slave-dma
Linus Torvalds [Sat, 19 Nov 2016 19:05:47 +0000 (11:05 -0800)]
Merge tag 'dmaengine-fix-4.9-rc6' of git://git.infradead.org/users/vkoul/slave-dma

Pull dmaengine fixes from Vinod Koul:
 "Some driver fixes which we pending in my tree:

   - return error code fix in edma driver
   - Kconfig fix for genric allocator in mmp_tdma
   - fix uninitialized value in sun6i
   - Runtime pm fixes for cppi"

* tag 'dmaengine-fix-4.9-rc6' of git://git.infradead.org/users/vkoul/slave-dma:
  dmaengine: cppi41: More PM runtime fixes
  dmaengine: cpp41: Fix handling of error path
  dmaengine: cppi41: Fix unpaired pm runtime when only a USB hub is connected
  dmaengine: cppi41: Fix list not empty warning on module removal
  dmaengine: sun6i: fix the uninitialized value for v_lli
  dmaengine: mmp_tdma: add missing select GENERIC_ALLOCATOR in Kconfig
  dmaengine: edma: Fix error return code in edma_alloc_chan_resources()

8 years agokvm: x86: merge kvm_arch_set_irq and kvm_arch_set_irq_inatomic
Paolo Bonzini [Thu, 17 Nov 2016 14:55:47 +0000 (15:55 +0100)]
kvm: x86: merge kvm_arch_set_irq and kvm_arch_set_irq_inatomic

kvm_arch_set_irq is unused since commit b97e6de9c96.  Merge
its functionality with kvm_arch_set_irq_inatomic.

Reported-by: Jiang Biao <jiang.biao2@zte.com.cn>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
8 years agoKVM: x86: fix missed SRCU usage in kvm_lapic_set_vapic_addr
Paolo Bonzini [Thu, 17 Nov 2016 14:55:46 +0000 (15:55 +0100)]
KVM: x86: fix missed SRCU usage in kvm_lapic_set_vapic_addr

Reported by syzkaller:

    [ INFO: suspicious RCU usage. ]
    4.9.0-rc4+ #47 Not tainted
    -------------------------------
    ./include/linux/kvm_host.h:536 suspicious rcu_dereference_check() usage!

    stack backtrace:
    CPU: 1 PID: 6679 Comm: syz-executor Not tainted 4.9.0-rc4+ #47
    Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
     ffff880039e2f6d0 ffffffff81c2e46b ffff88003e3a5b40 0000000000000000
     0000000000000001 ffffffff83215600 ffff880039e2f700 ffffffff81334ea9
     ffffc9000730b000 0000000000000004 ffff88003c4f8420 ffff88003d3f8000
    Call Trace:
     [<     inline     >] __dump_stack lib/dump_stack.c:15
     [<ffffffff81c2e46b>] dump_stack+0xb3/0x118 lib/dump_stack.c:51
     [<ffffffff81334ea9>] lockdep_rcu_suspicious+0x139/0x180 kernel/locking/lockdep.c:4445
     [<     inline     >] __kvm_memslots include/linux/kvm_host.h:534
     [<     inline     >] kvm_memslots include/linux/kvm_host.h:541
     [<ffffffff8105d6ae>] kvm_gfn_to_hva_cache_init+0xa1e/0xce0 virt/kvm/kvm_main.c:1941
     [<ffffffff8112685d>] kvm_lapic_set_vapic_addr+0xed/0x140 arch/x86/kvm/lapic.c:2217

Reported-by: Dmitry Vyukov <dvyukov@google.com>
Fixes: fda4e2e85589191b123d31cdc21fd33ee70f50fd
Cc: Andrew Honig <ahonig@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
8 years agoKVM: async_pf: avoid recursive flushing of work items
Paolo Bonzini [Thu, 17 Nov 2016 14:55:45 +0000 (15:55 +0100)]
KVM: async_pf: avoid recursive flushing of work items

This was reported by syzkaller:

    [ INFO: possible recursive locking detected ]
    4.9.0-rc4+ #49 Not tainted
    ---------------------------------------------
    kworker/2:1/5658 is trying to acquire lock:
     ([ 1644.769018] (&work->work)
    [<     inline     >] list_empty include/linux/compiler.h:243
    [<ffffffff8128dd60>] flush_work+0x0/0x660 kernel/workqueue.c:1511

    but task is already holding lock:
     ([ 1644.769018] (&work->work)
    [<ffffffff812916ab>] process_one_work+0x94b/0x1900 kernel/workqueue.c:2093

    stack backtrace:
    CPU: 2 PID: 5658 Comm: kworker/2:1 Not tainted 4.9.0-rc4+ #49
    Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
    Workqueue: events async_pf_execute
     ffff8800676ff630 ffffffff81c2e46b ffffffff8485b930 ffff88006b1fc480
     0000000000000000 ffffffff8485b930 ffff8800676ff7e0 ffffffff81339b27
     ffff8800676ff7e8 0000000000000046 ffff88006b1fcce8 ffff88006b1fccf0
    Call Trace:
    ...
    [<ffffffff8128ddf3>] flush_work+0x93/0x660 kernel/workqueue.c:2846
    [<ffffffff812954ea>] __cancel_work_timer+0x17a/0x410 kernel/workqueue.c:2916
    [<ffffffff81295797>] cancel_work_sync+0x17/0x20 kernel/workqueue.c:2951
    [<ffffffff81073037>] kvm_clear_async_pf_completion_queue+0xd7/0x400 virt/kvm/async_pf.c:126
    [<     inline     >] kvm_free_vcpus arch/x86/kvm/x86.c:7841
    [<ffffffff810b728d>] kvm_arch_destroy_vm+0x23d/0x620 arch/x86/kvm/x86.c:7946
    [<     inline     >] kvm_destroy_vm virt/kvm/kvm_main.c:731
    [<ffffffff8105914e>] kvm_put_kvm+0x40e/0x790 virt/kvm/kvm_main.c:752
    [<ffffffff81072b3d>] async_pf_execute+0x23d/0x4f0 virt/kvm/async_pf.c:111
    [<ffffffff8129175c>] process_one_work+0x9fc/0x1900 kernel/workqueue.c:2096
    [<ffffffff8129274f>] worker_thread+0xef/0x1480 kernel/workqueue.c:2230
    [<ffffffff812a5a94>] kthread+0x244/0x2d0 kernel/kthread.c:209
    [<ffffffff831f102a>] ret_from_fork+0x2a/0x40 arch/x86/entry/entry_64.S:433

The reason is that kvm_put_kvm is causing the destruction of the VM, but
the page fault is still on the ->queue list.  The ->queue list is owned
by the VCPU, not by the work items, so we cannot just add list_del to
the work item.

Instead, use work->vcpu to note async page faults that have been resolved
and will be processed through the done list.  There is no need to flush
those.

Cc: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
8 years agokvm: kvmclock: let KVM_GET_CLOCK return whether the master clock is in use
Paolo Bonzini [Wed, 9 Nov 2016 16:48:15 +0000 (17:48 +0100)]
kvm: kvmclock: let KVM_GET_CLOCK return whether the master clock is in use

Userspace can read the exact value of kvmclock by reading the TSC
and fetching the timekeeping parameters out of guest memory.  This
however is brittle and not necessary anymore with KVM 4.11.  Provide
a mechanism that lets userspace know if the new KVM_GET_CLOCK
semantics are in effect, and---since we are at it---if the clock
is stable across all VCPUs.

Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
8 years agoKVM: Disable irq while unregistering user notifier
Ignacio Alvarado [Fri, 4 Nov 2016 19:15:55 +0000 (12:15 -0700)]
KVM: Disable irq while unregistering user notifier

Function user_notifier_unregister should be called only once for each
registered user notifier.

Function kvm_arch_hardware_disable can be executed from an IPI context
which could cause a race condition with a VCPU returning to user mode
and attempting to unregister the notifier.

Signed-off-by: Ignacio Alvarado <ikalvarado@google.com>
Cc: stable@vger.kernel.org
Fixes: 18863bdd60f8 ("KVM: x86 shared msr infrastructure")
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
8 years agoKVM: x86: do not go through vcpu in __get_kvmclock_ns
Paolo Bonzini [Wed, 16 Nov 2016 17:31:30 +0000 (18:31 +0100)]
KVM: x86: do not go through vcpu in __get_kvmclock_ns

Going through the first VCPU is wrong if you follow a KVM_SET_CLOCK with
a KVM_GET_CLOCK immediately after, without letting the VCPU run and
call kvm_guest_time_update.

To fix this, compute the kvmclock value ourselves, using the master
clock (tsc, nsec) pair as the base and the host CPU frequency as
the scale.

Reported-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
8 years agoMerge tag 'kvm-arm-for-4.9-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git...
Radim Krčmář [Sat, 19 Nov 2016 17:02:07 +0000 (18:02 +0100)]
Merge tag 'kvm-arm-for-4.9-rc6' of git://git./linux/kernel/git/kvmarm/kvmarm

KVM/ARM updates for v4.9-rc6

- Fix handling of the 32bit cycle counter
- Fix cycle counter filtering

8 years agoMerge tag 'batadv-next-for-davem-20161119' of git://git.open-mesh.org/linux-merge
David S. Miller [Sat, 19 Nov 2016 16:13:05 +0000 (11:13 -0500)]
Merge tag 'batadv-next-for-davem-20161119' of git://git.open-mesh.org/linux-merge

Simon Wunderlich says:

====================
This feature patchset includes the following changes:

 - 6 patches adding functionality to detect a WiFi interface under
   other virtual interfaces, like VLANs. They introduce a cache for
   the detected the WiFi configuration to avoid RTNL locking in
   critical sections. Patches have been prepared by Marek Lindner
   and Sven Eckelmann

 - Enable automatic module loading for genl requests, by Sven Eckelmann

 - Fix a potential race condition on interface removal. This is not
   happening very often in practice, but requires bigger changes to fix,
   so we are sending this to net-next. By Linus Luessing
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge tag 'batadv-net-for-davem-20161119' of git://git.open-mesh.org/linux-merge
David S. Miller [Sat, 19 Nov 2016 16:11:52 +0000 (11:11 -0500)]
Merge tag 'batadv-net-for-davem-20161119' of git://git.open-mesh.org/linux-merge

Simon Wunderlich says:

====================
Here are two batman-adv bugfix patches:

 - Revert a splat on disabling interface which created another problem,
   by Sven Eckelmann

 - Fix error handling when the primary interface disappears during a
   throughput meter test, by Sven Eckelmann
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agosparc: drop duplicate header scatterlist.h
Geliang Tang [Fri, 18 Nov 2016 14:21:17 +0000 (22:21 +0800)]
sparc: drop duplicate header scatterlist.h

Drop duplicate header scatterlist.h from iommu_common.h.

Signed-off-by: Geliang Tang <geliangtang@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoaf_packet: Use virtio_net_hdr_from_skb() directly.
Jarno Rajahalme [Fri, 18 Nov 2016 23:40:42 +0000 (15:40 -0800)]
af_packet: Use virtio_net_hdr_from_skb() directly.

Remove static function __packet_rcv_vnet(), which only called
virtio_net_hdr_from_skb() and BUG()ged out if an error code was
returned.  Instead, call virtio_net_hdr_from_skb() from the former
call sites of __packet_rcv_vnet() and actually use the error handling
code that is already there.

Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoaf_packet: Use virtio_net_hdr_to_skb().
Jarno Rajahalme [Fri, 18 Nov 2016 23:40:41 +0000 (15:40 -0800)]
af_packet: Use virtio_net_hdr_to_skb().

Use the common virtio_net_hdr_to_skb() instead of open coding it.
Other call sites were changed by commit fd2a0437dc, but this one was
missed, maybe because it is split in two parts of the source code.

Interim comparisons of 'vnet_hdr->gso_type' still work as both the
vnet_hdr and skb notion of gso_type is zero when there is no gso.

Fixes: fd2a0437dc ("virtio_net: introduce virtio_net_hdr_{from,to}_skb")
Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agovirtio_net: Do not clear memory for struct virtio_net_hdr twice.
Jarno Rajahalme [Fri, 18 Nov 2016 23:40:40 +0000 (15:40 -0800)]
virtio_net: Do not clear memory for struct virtio_net_hdr twice.

virtio_net_hdr_from_skb() clears the memory for the header, so there
is no point for the callers to do the same.

Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agovirtio_net.h: Fix comment.
Jarno Rajahalme [Fri, 18 Nov 2016 23:40:39 +0000 (15:40 -0800)]
virtio_net.h: Fix comment.

Fix incorrent comment after the final #endif.

Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agovirtio_net: Simplify call sites for virtio_net_hdr_{from, to}_skb().
Jarno Rajahalme [Fri, 18 Nov 2016 23:40:38 +0000 (15:40 -0800)]
virtio_net: Simplify call sites for virtio_net_hdr_{from, to}_skb().

No point storing the return value of virtio_net_hdr_to_skb() or
virtio_net_hdr_from_skb() to a variable when the value is used only
once as a boolean in an immediately following if statement.

Signed-off-by: Jarno Rajahalme <jarno@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: macb: add check for dma mapping error in start_xmit()
Alexey Khoroshilov [Fri, 18 Nov 2016 22:40:10 +0000 (01:40 +0300)]
net: macb: add check for dma mapping error in start_xmit()

at91ether_start_xmit() does not check for dma mapping errors.

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>