sfrench/cifs-2.6.git
5 years agotipc: use standard write_lock & unlock functions when creating node
Jon Maloy [Thu, 11 Apr 2019 19:56:28 +0000 (21:56 +0200)]
tipc: use standard write_lock & unlock functions when creating node

In the function tipc_node_create() we protect the peer capability field
by using the node rw_lock. However, we access the lock directly instead
of using the dedicated functions for this, as we do everywhere else in
node.c. This cosmetic spot is fixed here.

Fixes: 40999f11ce67 ("tipc: make link capability update thread safe")
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: sched: flower: use correct ht function to prevent duplicates
Vlad Buslov [Thu, 11 Apr 2019 16:12:20 +0000 (19:12 +0300)]
net: sched: flower: use correct ht function to prevent duplicates

Implementation of function rhashtable_insert_fast() check if its internal
helper function __rhashtable_insert_fast() returns non-NULL pointer and
seemingly return -EEXIST in such case. However, since
__rhashtable_insert_fast() is called with NULL key pointer, it never
actually checks for duplicates, which means that -EEXIST is never returned
to the user. Use rhashtable_lookup_insert_fast() hash table API instead. In
order to verify that it works as expected and prevent the problem from
happening in future, extend tc-tests with new test that verifies that no
new filters with existing key can be inserted to flower classifier.

Fixes: 1f17f7742eeb ("net: sched: flower: insert filter to ht before offloading it to hw")
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonetns: read NETNSA_NSID as s32 attribute in rtnl_net_getid()
Guillaume Nault [Thu, 11 Apr 2019 14:45:57 +0000 (16:45 +0200)]
netns: read NETNSA_NSID as s32 attribute in rtnl_net_getid()

NETNSA_NSID is signed. Use nla_get_s32() to avoid confusion.

Signed-off-by: Guillaume Nault <gnault@redhat.com>
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: fou: remove redundant code in gue_udp_recv
Lorenzo Bianconi [Tue, 9 Apr 2019 10:03:07 +0000 (12:03 +0200)]
net: fou: remove redundant code in gue_udp_recv

Remove not useful protocol version check in gue_udp_recv since just
gue version 0 can hit that code. Moreover remove duplicated hdrlen
computation

Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agofou: correct spelling of encapsulation
Simon Horman [Tue, 9 Apr 2019 07:59:07 +0000 (09:59 +0200)]
fou: correct spelling of encapsulation

Correct spelling of encapsulation.
Found by inspection.

Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'net-sched-taprio-fix-picos_per_byte-miscalculation'
David S. Miller [Thu, 11 Apr 2019 02:27:43 +0000 (19:27 -0700)]
Merge branch 'net-sched-taprio-fix-picos_per_byte-miscalculation'

Leandro Dorileo says:

====================
net/sched: taprio: fix picos_per_byte miscalculation

This set fixes miscalculations based on invalid link speed values.

Changes in v6:
 + Avoid locking a spinlock while calling __ethtool_get_link_ksettings()
   (suggested by: Cong Wang);

Changes in v5:
 + Don't iterate over all the net_device maintained list (suggested by: Florian Fainelli);

Changes in v4:
 + converted pr_info calls to netdev_dbg (suggested by: Florian Fainelli);

Changes in v3:
 + yet pr_info() format warnings;

Changes in v2:
 + fixed pr_info() format both on cbs and taprio patches;
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet/sched: cbs: fix port_rate miscalculation
Leandro Dorileo [Mon, 8 Apr 2019 17:12:18 +0000 (10:12 -0700)]
net/sched: cbs: fix port_rate miscalculation

The Credit Based Shaper heavily depends on link speed to calculate
the scheduling credits, we can't properly calculate the credits if the
device has failed to report the link speed.

In that case we can't dequeue packets assuming a wrong port rate that will
result into an inconsistent credit distribution.

This patch makes sure we fail to dequeue case:

1) __ethtool_get_link_ksettings() reports error or 2) the ethernet driver
failed to set the ksettings' speed value (setting link speed to
SPEED_UNKNOWN).

Additionally we properly re calculate the port rate whenever the link speed
is changed.

Fixes: 3d0bd028ffb4a ("net/sched: Add support for HW offloading for CBS")
Signed-off-by: Leandro Dorileo <leandro.maciel.dorileo@intel.com>
Reviewed-by: Vedang Patel <vedang.patel@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet/sched: taprio: fix picos_per_byte miscalculation
Leandro Dorileo [Mon, 8 Apr 2019 17:12:17 +0000 (10:12 -0700)]
net/sched: taprio: fix picos_per_byte miscalculation

The Time Aware Priority Scheduler is heavily dependent to link speed,
it relies on it to calculate transmission bytes per cycle, we can't
properly calculate the so called budget if the device has failed
to report the link speed.

In that case we can't dequeue packets assuming a wrong budget.
This patch makes sure we fail to dequeue case:

1) __ethtool_get_link_ksettings() reports error or 2) the ethernet
driver failed to set the ksettings' speed value (setting link speed
to SPEED_UNKNOWN).

Additionally we re calculate the budget whenever the link speed is
changed.

Fixes: 5a781ccbd19e4 ("tc: Add support for configuring the taprio scheduler")
Signed-off-by: Leandro Dorileo <leandro.maciel.dorileo@intel.com>
Reviewed-by: Vedang Patel <vedang.patel@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: strparser: fix comment
Jakub Kicinski [Wed, 10 Apr 2019 20:18:57 +0000 (13:18 -0700)]
net: strparser: fix comment

Fix comment.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Handle RTA_GATEWAY set to 0
David Ahern [Wed, 10 Apr 2019 17:05:51 +0000 (10:05 -0700)]
ipv4: Handle RTA_GATEWAY set to 0

Govindarajulu reported a regression with Network Manager which sends an
RTA_GATEWAY attribute with the address set to 0. Fixup the handling of
RTA_GATEWAY to only set fc_gw_family if the gateway address is actually
set.

Fixes: f35b794b3b405 ("ipv4: Prepare fib_config for IPv6 gateway")
Reported-by: Govindarajulu Varadarajan <govind.varadar@gmail.com>
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'net-sched-move-back-qlen-to-per-CPU-accounting'
David S. Miller [Wed, 10 Apr 2019 19:20:46 +0000 (12:20 -0700)]
Merge branch 'net-sched-move-back-qlen-to-per-CPU-accounting'

Paolo Abeni says:

====================
net: sched: move back qlen to per CPU accounting

The commit 46b1c18f9deb ("net: sched: put back q.qlen into a single location")
introduced some measurable regression in the contended scenarios for
lock qdisc.

As Eric suggested we could replace q.qlen access with calls to qdisc_is_empty()
in the datapath and revert the above commit. The TC subsystem updates
qdisc->is_empty in a somewhat loose way: notably 'is_empty' is set only when
the qdisc dequeue() calls return a NULL ptr. That is, the invocation after
the last packet is dequeued.

The above is good enough for BYPASS implementation - the only downside is that
we end up avoiding the optimization for a very small time-frame - but will
break hard things when internal structures consistency for classful qdisc
relies on child qdisc_is_empty().

A more strict 'is_empty' update adds a relevant complexity to its life-cycle, so
this series takes a different approach: we allow lockless qdisc to switch from
per CPU accounting to global stats accounting when the NOLOCK bit is cleared.
Since most pieces of infrastructure are already in place, this requires very
little changes to the pfifo_fast qdisc, and any later NOLOCK qdisc can hook
there with little effort - no need to maintain two different implementations.

The first 2 patches removes direct qlen access from non core TC code, the 3rd
and 4th patches place and use the infrastructure to allow stats account
switching and the 5th patch is the actual revert.

 v1 -> v2:
  - fixed build issues
  - more descriptive commit message for patch 5/5
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoRevert: "net: sched: put back q.qlen into a single location"
Paolo Abeni [Wed, 10 Apr 2019 12:32:41 +0000 (14:32 +0200)]
Revert: "net: sched: put back q.qlen into a single location"

This revert commit 46b1c18f9deb ("net: sched: put back q.qlen into
a single location").
After the previous patch, when a NOLOCK qdisc is enslaved to a
locking qdisc it switches to global stats accounting. As a consequence,
when a classful qdisc accesses directly a child qdisc's qlen, such
qdisc is not doing per CPU accounting and qlen value is consistent.

In the control path nobody uses directly qlen since commit
e5f0e8f8e45 ("net: sched: introduce and use qdisc tree flush/purge
helpers"), so we can remove the contented atomic ops from the
datapath.

v1 -> v2:
 - complete the qdisc_qstats_atomic_qlen_dec() ->
   qdisc_qstats_cpu_qlen_dec() replacement, fix build issue
 - more descriptive commit message

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: sched: when clearing NOLOCK, clear TCQ_F_CPUSTATS, too
Paolo Abeni [Wed, 10 Apr 2019 12:32:40 +0000 (14:32 +0200)]
net: sched: when clearing NOLOCK, clear TCQ_F_CPUSTATS, too

Since stats updating is always consistent with TCQ_F_CPUSTATS flag,
we can disable it at qdisc creation time flipping such bit.

In my experiments, if the NOLOCK flag is cleared, per CPU stats
accounting does not give any measurable performance gain, but it
waste some memory.

Let's clear TCQ_F_CPUSTATS together with NOLOCK, when enslaving
a NOLOCK qdisc to 'lock' one.

Use stats update helper inside pfifo_fast, to cope correctly with
TCQ_F_CPUSTATS flag change.

As a side effect, q.qlen value for any child qdiscs is always
consistent for all lock classfull qdiscs.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: sched: always do stats accounting according to TCQ_F_CPUSTATS
Paolo Abeni [Wed, 10 Apr 2019 12:32:39 +0000 (14:32 +0200)]
net: sched: always do stats accounting according to TCQ_F_CPUSTATS

The core sched implementation checks independently for NOLOCK flag
to acquire/release the root spin lock and for qdisc_is_percpu_stats()
to account per CPU values in many places.

This change update the last few places checking the TCQ_F_NOLOCK to
do per CPU stats accounting according to qdisc_is_percpu_stats()
value.

The above allows to clean dev_requeue_skb() implementation a bit
and makes stats update always consistent with a single flag.

v1 -> v2:
 - do not move qdisc_is_empty definition, fix build issue

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: sched: prefer qdisc_is_empty() over direct qlen access
Paolo Abeni [Wed, 10 Apr 2019 12:32:38 +0000 (14:32 +0200)]
net: sched: prefer qdisc_is_empty() over direct qlen access

When checking for root qdisc queue length, do not access directly q.qlen.
In the following patches we will move back qlen accounting to per CPU
values for NOLOCK qdiscs.

Instead, prefer the qdisc_is_empty() helper usage.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: caif: avoid using qdisc_qlen()
Paolo Abeni [Wed, 10 Apr 2019 12:32:37 +0000 (14:32 +0200)]
net: caif: avoid using qdisc_qlen()

Such helper does not cope correctly with NOLOCK qdiscs.
In the following patches we will move back qlen to per CPU
values for such qdiscs, so qdisc_qlen_sum() is not an option,
too.
Instead, use qlen only for lock qdiscs, and always set
flow off for NOLOCK qdiscs with a not empty tx queue.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: phy: remove unnecessary callback settings in C45 drivers
Heiner Kallweit [Sun, 7 Apr 2019 10:11:35 +0000 (12:11 +0200)]
net: phy: remove unnecessary callback settings in C45 drivers

genphy_c45_aneg_done() is used by phylib as fallback for c45 PHY's if
callback aneg_done isn't defined. So we don't have to set this
explicitly. Same for genphy_c45_pma_read_abilities().

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'netdevsim-small-spring-cleanup'
David S. Miller [Tue, 9 Apr 2019 18:16:59 +0000 (11:16 -0700)]
Merge branch 'netdevsim-small-spring-cleanup'

Jiri Pirko says:

====================
netdevsim: small spring cleanup

Nothing serious, just cosmetics.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonetdevsim: assume CONFIG_NET_DEVLINK is always enabled
Jiri Pirko [Tue, 9 Apr 2019 15:14:54 +0000 (17:14 +0200)]
netdevsim: assume CONFIG_NET_DEVLINK is always enabled

Since commit f6b19b354d50 ("net: devlink: select NET_DEVLINK
from drivers") adds implicit select of NET_DEVLINK for
netdevsim, the code does not have to deal with the case
when CONFIG_NET_DEVLINK is not enabled. So remove the ifcase.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonetdevsim: let net core to free netdevsim netdev
Jiri Pirko [Tue, 9 Apr 2019 15:14:53 +0000 (17:14 +0200)]
netdevsim: let net core to free netdevsim netdev

No need to free it ourselves, just set the "needs_free_netdev" flag
and leave the work to net core.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonetdevsim: remove nsim_dellink() implementation
Jiri Pirko [Tue, 9 Apr 2019 15:14:52 +0000 (17:14 +0200)]
netdevsim: remove nsim_dellink() implementation

Remove nsim_dellink() implementation. The rtnetlink code sets the
dellink op to unregister_netdevice_queue(), so this is not needed.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
David S. Miller [Tue, 9 Apr 2019 06:39:36 +0000 (23:39 -0700)]
Merge git://git./linux/kernel/git/davem/net

5 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Tue, 9 Apr 2019 03:10:46 +0000 (17:10 -1000)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Off by one and bounds checking fixes in NFC, from Dan Carpenter.

 2) There have been many weird regressions in r8169 since we turned ASPM
    support on, some are still not understood nor completely resolved.
    Let's turn this back off for now. From Heiner Kallweit.

 3) Signess fixes for ethtool speed value handling, from Michael
    Zhivich.

 4) Handle timestamps properly in macb driver, from Paul Thomas.

 5) Two erspan fixes, it's the usual "skb ->data potentially reallocated
    and we're holding a stale protocol header pointer". From Lorenzo
    Bianconi.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
  bnxt_en: Reset device on RX buffer errors.
  bnxt_en: Improve RX consumer index validity check.
  net: macb driver, check for SKBTX_HW_TSTAMP
  qlogic: qlcnic: fix use of SPEED_UNKNOWN ethtool constant
  broadcom: tg3: fix use of SPEED_UNKNOWN ethtool constant
  ethtool: avoid signed-unsigned comparison in ethtool_validate_speed()
  net: ip6_gre: fix possible use-after-free in ip6erspan_rcv
  net: ip_gre: fix possible use-after-free in erspan_rcv
  r8169: disable ASPM again
  MAINTAINERS: ieee802154: update documentation file pattern
  net: vrf: Fix ping failed when vrf mtu is set to 0
  selftests: add a tc matchall test case
  nfc: nci: Potential off by one in ->pipes[] array
  NFC: nci: Add some bounds checking in nci_hci_cmd_received()

5 years agoMerge branch 'fixes-v5.1' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Tue, 9 Apr 2019 03:06:43 +0000 (17:06 -1000)]
Merge branch 'fixes-v5.1' of git://git./linux/kernel/git/jmorris/linux-security

Pull TPM fixes from James Morris:
 "From Jarkko: These are critical fixes for v5.1. Contains also couple
  of new selftests for v5.1 features (partial reads in /dev/tpm0)"

* 'fixes-v5.1' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/linux-security:
  selftests/tpm2: Open tpm dev in unbuffered mode
  selftests/tpm2: Extend tests to cover partial reads
  KEYS: trusted: fix -Wvarags warning
  tpm: Fix the type of the return value in calc_tpm2_event_size()
  KEYS: trusted: allow trusted.ko to initialize w/o a TPM
  tpm: fix an invalid condition in tpm_common_poll
  tpm: turn on TPM on suspend for TPM 1.x

5 years agoMerge tag 'xtensa-20190408' of git://github.com/jcmvbkbc/linux-xtensa
Linus Torvalds [Tue, 9 Apr 2019 03:04:42 +0000 (17:04 -1000)]
Merge tag 'xtensa-20190408' of git://github.com/jcmvbkbc/linux-xtensa

Pull xtensa fixes from Max Filippov:

 - fix syscall number passed to trace_sys_exit

 - fix syscall number initialization in start_thread

 - fix level interpretation in the return_address

 - fix format string warning in init_pmd

* tag 'xtensa-20190408' of git://github.com/jcmvbkbc/linux-xtensa:
  xtensa: fix format string warning in init_pmd
  xtensa: fix return_address
  xtensa: fix initialization of pt_regs::syscall in start_thread
  xtensa: use actual syscall number in do_syscall_trace_leave

5 years agoMerge branch 'bnxt_en-fixes'
David S. Miller [Mon, 8 Apr 2019 23:39:41 +0000 (16:39 -0700)]
Merge branch 'bnxt_en-fixes'

Michael Chan says:

====================
bnxt_en: 2 bug fixes.

The first patch prevents possible driver crash if we get a bad RX index
from the hardware.  The second patch resets the device when the hardware
reports buffer error to recover from the error.

Please queue these for -stable also.  Thanks.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnxt_en: Reset device on RX buffer errors.
Michael Chan [Mon, 8 Apr 2019 21:39:55 +0000 (17:39 -0400)]
bnxt_en: Reset device on RX buffer errors.

If the RX completion indicates RX buffers errors, the RX ring will be
disabled by firmware and no packets will be received on that ring from
that point on.  Recover by resetting the device.

Fixes: c0c050c58d84 ("bnxt_en: New Broadcom ethernet driver.")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnxt_en: Improve RX consumer index validity check.
Michael Chan [Mon, 8 Apr 2019 21:39:54 +0000 (17:39 -0400)]
bnxt_en: Improve RX consumer index validity check.

There is logic to check that the RX/TPA consumer index is the expected
index to work around a hardware problem.  However, the potentially bad
consumer index is first used to index into an array to reference an entry.
This can potentially crash if the bad consumer index is beyond legal
range.  Improve the logic to use the consumer index for dereferencing
after the validity check and log an error message.

Fixes: fa7e28127a5a ("bnxt_en: Add workaround to detect bad opaque in rx completion (part 2)")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: macb driver, check for SKBTX_HW_TSTAMP
Paul Thomas [Mon, 8 Apr 2019 19:37:54 +0000 (15:37 -0400)]
net: macb driver, check for SKBTX_HW_TSTAMP

Make sure SKBTX_HW_TSTAMP (i.e. SOF_TIMESTAMPING_TX_HARDWARE) has been
enabled for this skb. It does fix the issue where normal socks that
aren't expecting a timestamp will not wake up on select, but when a
user does want a SOF_TIMESTAMPING_TX_HARDWARE it does work.

Signed-off-by: Paul Thomas <pthomas8589@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoethtool: thunder_bgx: use ethtool.h constants for speed and duplex
Michael Zhivich [Mon, 8 Apr 2019 19:00:46 +0000 (15:00 -0400)]
ethtool: thunder_bgx: use ethtool.h constants for speed and duplex

Use constants provided by ethtool.h for speed and duplex values
instead of raw integer constants to increase code readability.

thunder_bgx already uses SPEED_UNKNOWN and DUPLEX_UNKNOWN constants,
also provided by ethtool.h.

Signed-off-by: Michael Zhivich <mzhivich@akamai.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: phy: fix setting autoneg_complete in genphy_update_link
Heiner Kallweit [Mon, 8 Apr 2019 17:28:28 +0000 (19:28 +0200)]
net: phy: fix setting autoneg_complete in genphy_update_link

The original patch didn't set phydev->autoneg_complete in one exit path.
Fix this.

Fixes: 4950c2ba49cc ("net: phy: fix autoneg mismatch case in genphy_read_status")
Reported-by: Simon Horman <horms@verge.net.au>
Tested-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'ethtool-fix-use-of-SPEED_UNKNOWN-constant'
David S. Miller [Mon, 8 Apr 2019 23:30:43 +0000 (16:30 -0700)]
Merge branch 'ethtool-fix-use-of-SPEED_UNKNOWN-constant'

Michael Zhivich says:

====================
ethtool: fix use of SPEED_UNKNOWN constant

This patch series addresses 2 related issues:

1. ethtool_validate_speed() triggers a "signed-unsigned comparison"
warning due to type difference of SPEED_UNKNOWN constant (int)
and argument to ethtool_validate_speed (__u32).

2. some drivers use u16 storage for SPEED_UNKNOWN constant,
resulting in value truncation and thus failure to test against
SPEED_UNKNOWN correctly.

This revised series addresses several feedback comments:
- split up the patch in to series
- do not unnecessarily change drivers that use "int" storage
  for speed values
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoqlogic: qlcnic: fix use of SPEED_UNKNOWN ethtool constant
Michael Zhivich [Mon, 8 Apr 2019 14:48:47 +0000 (10:48 -0400)]
qlogic: qlcnic: fix use of SPEED_UNKNOWN ethtool constant

qlcnic driver uses u16 to store SPEED_UKNOWN ethtool constant,
which is defined as -1, resulting in value truncation and
thus incorrect test results against SPEED_UNKNOWN.

For example, the following test will print "False":

    u16 speed = SPEED_UNKNOWN;

    if (speed == SPEED_UNKNOWN)
        printf("True");
    else
        printf("False");

Change storage of speed to use u32 to avoid this issue.

Signed-off-by: Michael Zhivich <mzhivich@akamai.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobroadcom: tg3: fix use of SPEED_UNKNOWN ethtool constant
Michael Zhivich [Mon, 8 Apr 2019 14:48:46 +0000 (10:48 -0400)]
broadcom: tg3: fix use of SPEED_UNKNOWN ethtool constant

tg3 driver uses u16 to store SPEED_UKNOWN ethtool constant,
which is defined as -1, resulting in value truncation and
thus incorrect test results against SPEED_UNKNOWN.

For example, the following test will print "False":

u16 speed = SPEED_UNKNOWN;

if (speed == SPEED_UNKNOWN)
    printf("True");
else
    printf("False");

Change storage of speed to use u32 to avoid this issue.

Signed-off-by: Michael Zhivich <mzhivich@akamai.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoethtool: avoid signed-unsigned comparison in ethtool_validate_speed()
Michael Zhivich [Mon, 8 Apr 2019 14:48:45 +0000 (10:48 -0400)]
ethtool: avoid signed-unsigned comparison in ethtool_validate_speed()

When building C++ userspace code that includes ethtool.h
with "-Werror -Wall", g++ complains about signed-unsigned comparison in
ethtool_validate_speed() due to definition of SPEED_UNKNOWN as -1.

Explicitly cast SPEED_UNKNOWN to __u32 to match type of
ethtool_validate_speed() argument.

Signed-off-by: Michael Zhivich <mzhivich@akamai.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'mlxsw-Add-support-for-devlink-info-command'
David S. Miller [Mon, 8 Apr 2019 23:28:01 +0000 (16:28 -0700)]
Merge branch 'mlxsw-Add-support-for-devlink-info-command'

Ido Schimmel says:

====================
mlxsw: Add support for devlink info command

Shalom says:

Expose the following ASIC information via devlink info command:
  - Driver name
  - Hardware revision
  - Firmware PSID
  - Running firmware version

Standard output example:
  $ devlink dev info pci/0000:03:00.0
  pci/0000:03:00.0:
    versions:
        fixed:
          hw.revision A0
          fw.psid MT_2750110033
        running:
          fw.version 13.1910.622

Pretty JSON example:
  $ devlink -jp dev info pci/0000:03:00.0
  {
      "info": {
          "pci/0000:03:00.0": {
              "versions": {
                  "fixed": {
                      "hw.revision": "A0",
                      "fw.psid": "MT_2750110033"
                  },
                  "running": {
                      "fw.version": "13.1910.622"
                  }
              }
          }
      }
  }

v2:
  - Add fw.psid documentation
  - Add driver name
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoDocumentation: networking: devlink-info-versions: Add fw.psid
Shalom Toledo [Mon, 8 Apr 2019 06:59:36 +0000 (06:59 +0000)]
Documentation: networking: devlink-info-versions: Add fw.psid

Add firmware parameter id (fw.psid).

Signed-off-by: Shalom Toledo <shalomt@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agomlxsw: core: Add support for devlink info command
Shalom Toledo [Mon, 8 Apr 2019 06:59:35 +0000 (06:59 +0000)]
mlxsw: core: Add support for devlink info command

Expose the following ASIC information via devlink info command:
  - Driver name
  - Hardware revision
  - Firmware PSID
  - Running firmware version

Standard output example:
  $ devlink dev info pci/0000:03:00.0
  pci/0000:03:00.0:
    driver mlxsw_spectrum
    versions:
        fixed:
          hw.revision A0
          fw.psid MT_2750110033
        running:
          fw.version 13.1910.622

Signed-off-by: Shalom Toledo <shalomt@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agomlxsw: reg: Add MGIR register
Shalom Toledo [Mon, 8 Apr 2019 06:59:34 +0000 (06:59 +0000)]
mlxsw: reg: Add MGIR register

Add MGIR register. MGIR, Management General Information Register, allows
software to query the hardware and firmware general information.

Signed-off-by: Shalom Toledo <shalomt@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoqmi_wwan: Add quirk for Quectel dynamic config
Kristian Evensen [Sun, 7 Apr 2019 13:39:09 +0000 (15:39 +0200)]
qmi_wwan: Add quirk for Quectel dynamic config

Most, if not all, Quectel devices use dynamic interface numbers, and
users are able to change the USB configuration at will. Matching on for
example interface number is therefore not possible.

Instead, the QMI device can be identified by looking at the interface
class, subclass and protocol (all 0xff), as well as the number of
endpoints. The reason we need to look at the number of endpoints, is
that the diagnostic port interface has the same class, subclass and
protocol as QMI. However, the diagnostic port only has two endpoints,
while QMI has three.

Until now, we have identified the QMI device by combining a match on
class, subclass and protocol, with a call to the function
quectel_diag_detect(). In quectel_diag_detect(), we check if the number
of endpoints matches for known Quectel vendor/product ids.

Adding new vendor/product ids to quectel_diag_detect() is not a good
long-term solution. This commit replaces the function with a quirk, and
applies the quirk to affected Quectel devices that I have been able to
test the change with (EP06, EM12 and EC25). If the quirk is set and the
number of endpoints equal two, we return from qmi_wwan_probe() with
-ENODEV.

Signed-off-by: Kristian Evensen <kristian.evensen@gmail.com>
Acked-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: phy: replace genphy_10g_driver with genphy_c45_driver
Heiner Kallweit [Sun, 7 Apr 2019 11:55:01 +0000 (13:55 +0200)]
net: phy: replace genphy_10g_driver with genphy_c45_driver

Recently a number of generic functions for Clause 45 PHY's has been
added. So let's replace the old very limited genphy_10g_driver with a
genphy_c45_driver. This driver isn't limited to 10G, however it's worth
to be noted that Clause 45 doesn't cover 1000Base-T. For using
1000Base-T with a Clause 45 PHY a dedicated PHY driver using vendor
registers is needed.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'erspan-use-after-free'
David S. Miller [Mon, 8 Apr 2019 23:16:47 +0000 (16:16 -0700)]
Merge branch 'erspan-use-after-free'

Lorenzo Bianconi says:

====================
fix possible use-after-free in erspan_v{4,6}

Similar to what I did in commit bb9bd814ebf0 ("ipv6: sit: reset ip
header pointer in ipip6_rcv"), fix possible use-after-free in
erspan_rcv and ip6erspan_rcv extracting tunnel metadata since the
packet can be 'uncloned' running __iptunnel_pull_header
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: ip6_gre: fix possible use-after-free in ip6erspan_rcv
Lorenzo Bianconi [Sat, 6 Apr 2019 15:16:53 +0000 (17:16 +0200)]
net: ip6_gre: fix possible use-after-free in ip6erspan_rcv

erspan_v6 tunnels run __iptunnel_pull_header on received skbs to remove
erspan header. This can determine a possible use-after-free accessing
pkt_md pointer in ip6erspan_rcv since the packet will be 'uncloned'
running pskb_expand_head if it is a cloned gso skb (e.g if the packet has
been sent though a veth device). Fix it resetting pkt_md pointer after
__iptunnel_pull_header

Fixes: 1d7e2ed22f8d ("net: erspan: refactor existing erspan code")
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: ip_gre: fix possible use-after-free in erspan_rcv
Lorenzo Bianconi [Sat, 6 Apr 2019 15:16:52 +0000 (17:16 +0200)]
net: ip_gre: fix possible use-after-free in erspan_rcv

erspan tunnels run __iptunnel_pull_header on received skbs to remove
gre and erspan headers. This can determine a possible use-after-free
accessing pkt_md pointer in erspan_rcv since the packet will be 'uncloned'
running pskb_expand_head if it is a cloned gso skb (e.g if the packet has
been sent though a veth device). Fix it resetting pkt_md pointer after
__iptunnel_pull_header

Fixes: 1d7e2ed22f8d ("net: erspan: refactor existing erspan code")
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoselftests/tpm2: Open tpm dev in unbuffered mode
Tadeusz Struk [Tue, 12 Feb 2019 23:42:05 +0000 (15:42 -0800)]
selftests/tpm2: Open tpm dev in unbuffered mode

In order to have control over how many bytes are read or written
the device needs to be opened in unbuffered mode.

Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Reviewed-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Tested-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: James Morris <james.morris@microsoft.com>
5 years agoselftests/tpm2: Extend tests to cover partial reads
Tadeusz Struk [Tue, 12 Feb 2019 23:42:10 +0000 (15:42 -0800)]
selftests/tpm2: Extend tests to cover partial reads

Three new tests added:
1. Send get random cmd, read header in 1st read, read the rest in second
   read - expect success
2. Send get random cmd, read only part of the response, send another
   get random command, read the response - expect success
3. Send get random cmd followed by another get random cmd, without
   reading the first response - expect the second cmd to fail with -EBUSY

Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Reviewed-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Tested-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: James Morris <james.morris@microsoft.com>
5 years agoKEYS: trusted: fix -Wvarags warning
ndesaulniers@google.com [Mon, 22 Oct 2018 23:43:57 +0000 (16:43 -0700)]
KEYS: trusted: fix -Wvarags warning

Fixes the warning reported by Clang:
security/keys/trusted.c:146:17: warning: passing an object that
undergoes default
      argument promotion to 'va_start' has undefined behavior [-Wvarargs]
        va_start(argp, h3);
                       ^
security/keys/trusted.c:126:37: note: parameter of type 'unsigned
char' is declared here
unsigned char *h2, unsigned char h3, ...)
                               ^
Specifically, it seems that both the C90 (4.8.1.1) and C11 (7.16.1.4)
standards explicitly call this out as undefined behavior:

The parameter parmN is the identifier of the rightmost parameter in
the variable parameter list in the function definition (the one just
before the ...). If the parameter parmN is declared with ... or with a
type that is not compatible with the type that results after
application of the default argument promotions, the behavior is
undefined.

Link: https://github.com/ClangBuiltLinux/linux/issues/41
Link: https://www.eskimo.com/~scs/cclass/int/sx11c.html
Suggested-by: David Laight <David.Laight@aculab.com>
Suggested-by: Denis Kenzior <denkenz@gmail.com>
Suggested-by: James Bottomley <jejb@linux.vnet.ibm.com>
Suggested-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Tested-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: James Morris <james.morris@microsoft.com>
5 years agotpm: Fix the type of the return value in calc_tpm2_event_size()
Yue Haibing [Wed, 20 Feb 2019 08:25:38 +0000 (16:25 +0800)]
tpm: Fix the type of the return value in calc_tpm2_event_size()

calc_tpm2_event_size() has an invalid signature because
it returns a 'size_t' where as its signature says that
it returns 'int'.

Cc: <stable@vger.kernel.org>
Fixes: 4d23cc323cdb ("tpm: add securityfs support for TPM 2.0 firmware event log")
Suggested-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: Yue Haibing <yuehaibing@huawei.com>
Reviewed-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: James Morris <james.morris@microsoft.com>
5 years agoKEYS: trusted: allow trusted.ko to initialize w/o a TPM
Jarkko Sakkinen [Mon, 25 Mar 2019 14:43:10 +0000 (16:43 +0200)]
KEYS: trusted: allow trusted.ko to initialize w/o a TPM

Allow trusted.ko to initialize w/o a TPM. This commit also adds checks
to the exported functions to fail when a TPM is not available.

Fixes: 240730437deb ("KEYS: trusted: explicitly use tpm_chip structure...")
Cc: James Morris <jmorris@namei.org>
Reported-by: Dan Williams <dan.j.williams@intel.com>
Tested-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: James Morris <james.morris@microsoft.com>
5 years agotpm: fix an invalid condition in tpm_common_poll
Tadeusz Struk [Wed, 27 Mar 2019 18:32:38 +0000 (11:32 -0700)]
tpm: fix an invalid condition in tpm_common_poll

The poll condition should only check response_length,
because reads should only be issued if there is data to read.
The response_read flag only prevents double writes.
The problem was that the write set the response_read to false,
enqued a tpm job, and returned. Then application called poll
which checked the response_read flag and returned EPOLLIN.
Then the application called read, but got nothing.
After all that the async_work kicked in.
Added also mutex_lock around the poll check to prevent
other possible race conditions.

Fixes: 9488585b21bef0df12 ("tpm: add support for partial reads")
Reported-by: Mantas MikulÄ—nas <grawity@gmail.com>
Tested-by: Mantas MikulÄ—nas <grawity@gmail.com>
Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Reviewed-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Signed-off-by: James Morris <james.morris@microsoft.com>
5 years agotpm: turn on TPM on suspend for TPM 1.x
Jarkko Sakkinen [Fri, 22 Mar 2019 10:51:20 +0000 (12:51 +0200)]
tpm: turn on TPM on suspend for TPM 1.x

tpm_chip_start/stop() should be also called for TPM 1.x devices on
suspend. Add that functionality back. Do not lock the chip because
it is unnecessary as there are no multiple threads using it when
doing the suspend.

Fixes: a3fbfae82b4c ("tpm: take TPM chip power gating out of tpm_transmit()")
Reported-by: Paul Zimmerman <pauldzim@gmail.com>
Signed-off-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>
Tested-by: Domenico Andreoli <domenico.andreoli@linux.com>
Signed-off-by: James Morris <james.morris@microsoft.com>
5 years agoMerge branch 'hns3-next'
David S. Miller [Mon, 8 Apr 2019 22:30:49 +0000 (15:30 -0700)]
Merge branch 'hns3-next'

Huazhong Tan says:

====================
code optimizations & bugfixes for HNS3 driver

This patchset includes bugfixes and code optimizations for
the HNS3 ethernet controller driver.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns3: fix keep_alive_timer not stop problem
Huazhong Tan [Sat, 6 Apr 2019 07:43:36 +0000 (15:43 +0800)]
net: hns3: fix keep_alive_timer not stop problem

When hclgevf_client_start() fails or VF driver unloaded, there is
nobody to disable keep_alive_timer.

So this patch fixes them.

Fixes: a6d818e31d08 ("net: hns3: Add vport alive state checking support")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns3: remove resetting check in hclgevf_reset_task_schedule
Huazhong Tan [Sat, 6 Apr 2019 07:43:35 +0000 (15:43 +0800)]
net: hns3: remove resetting check in hclgevf_reset_task_schedule

The checking of HCLGEVF_STATE_RST_HANDLING flag in the
hclgevf_reset_task_schedule() will make some scheduling of
reset pending fail. This flag will be checked in the
hclgevf_reset_service_task(), it is unnecessary to check it
in the hclgevf_reset_task_schedule(). So this patch removes it.

Fixes: 35a1e50343bd ("net: hns3: Add VF Reset Service Task to support event handling")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns3: add error handler for initializing command queue
Huazhong Tan [Sat, 6 Apr 2019 07:43:34 +0000 (15:43 +0800)]
net: hns3: add error handler for initializing command queue

This patch adds error handler for the failure of command queue
initialization both PF and VF.

Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns3: stop mailbox handling when command queue need re-init
Huazhong Tan [Sat, 6 Apr 2019 07:43:33 +0000 (15:43 +0800)]
net: hns3: stop mailbox handling when command queue need re-init

If the command queue needs re-initialization, the mailbox handling
task should do nothing, otherwise this task will just get some error
print.

Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns3: handle pending reset while reset fail
Huazhong Tan [Sat, 6 Apr 2019 07:43:32 +0000 (15:43 +0800)]
net: hns3: handle pending reset while reset fail

The ongoing lower-level reset will fail when there is a higher-level
reset occurs, so the error handler should deal with this situation.

Fixes: 6a5f6fa382f3 ("net: hns3: add error handler for hclgevf_reset()")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns3: do not request reset when hardware resetting
Huazhong Tan [Sat, 6 Apr 2019 07:43:31 +0000 (15:43 +0800)]
net: hns3: do not request reset when hardware resetting

When hardware reset does not finish, the driver should not
request a new reset, otherwise the ongoing hardware reset
will get problem.

Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns3: ignore lower-level new coming reset
Huazhong Tan [Sat, 6 Apr 2019 07:43:30 +0000 (15:43 +0800)]
net: hns3: ignore lower-level new coming reset

It is unnecessary to deal with the new coming reset if
it is lower than the ongoing one.

Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns3: deactive the reset timer when reset successfully
Huazhong Tan [Sat, 6 Apr 2019 07:43:29 +0000 (15:43 +0800)]
net: hns3: deactive the reset timer when reset successfully

If the reset has been done successfully, the ongoing reset timer
is unnecessary.

Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns3: adjust the timing of hns3_client_stop when unloading
Huazhong Tan [Sat, 6 Apr 2019 07:43:28 +0000 (15:43 +0800)]
net: hns3: adjust the timing of hns3_client_stop when unloading

hns3_client_stop() should be called after unregister_netdev(),
otherwise the ongoing reset task may start the client just after it.

Fixes: a6d818e31d08 ("net: hns3: Add vport alive state checking support")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns3: not reset vport who not alive when PF reset
Huazhong Tan [Sat, 6 Apr 2019 07:43:27 +0000 (15:43 +0800)]
net: hns3: not reset vport who not alive when PF reset

If a vport is not alive, it is unnecessary to notify it to reset
before PF asserting a reset. So before inform vport to reset,
we need to check its alive state firstly.

Fixes: aa5c4f175be6 ("net: hns3: add reset handling for VF when doing PF reset")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns3: set up the vport alive state while reinitializing
Huazhong Tan [Sat, 6 Apr 2019 07:43:26 +0000 (15:43 +0800)]
net: hns3: set up the vport alive state while reinitializing

When reinitializing, the vport alive state needs to be set up.

Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: hns3: set vport alive state to default while resetting
Huazhong Tan [Sat, 6 Apr 2019 07:43:25 +0000 (15:43 +0800)]
net: hns3: set vport alive state to default while resetting

When resetting, the vport alive state should be set to default,
otherwise the alive state of the vport whose driver not running
is wrong before the timer to check it out.

Fixes: a6d818e31d08 ("net: hns3: Add vport alive state checking support")
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: Peng Li <lipeng321@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'ipv4-Enable-support-for-IPv6-gateway-with-IPv4-routes'
David S. Miller [Mon, 8 Apr 2019 22:22:41 +0000 (15:22 -0700)]
Merge branch 'ipv4-Enable-support-for-IPv6-gateway-with-IPv4-routes'

David Ahern says:

====================
ipv4: Enable support for IPv6 gateway with IPv4 routes

Last set of three with the end goal of enabling IPv6 gateways with IPv4
routes.

This set adds fib6_nh_init and release to the IPv6 stubs, and adds neighbor
helpers that IPv4 code invokes to resolve an IPv6 address. When using
an IPv6 neighbor entry the hh_cache is bypassed as it contains the wrong
ethernet header for an IPv4 packet.

The nh_common nhc_has_gw was a temporary field used to convert existing
code from fib{6}_nh to fib_nh_common. That field is now converted to
nhc_gw_family to differentiate the address family of the gateway entry
as opposed to the address family of the container of fib_nh_common.

Existing code for rtable and fib_config is refactored to prepare
for a v6 address and then support is added. From there various
miscellaneous functions are updated to handle a v6 gateway - from
validating the v6 address to lookups in bpf code to verifying the
nexthop state.

Offload drivers - mlxsw and rocker - are modified to detect the v6
gateway and reject the route as 'unsupported'. e.g.,

    $ ip ro add 172.16.101.0/24 via inet6 fe80::202:ff:fe00:b dev swp1s0
    Error: mlxsw_spectrum: IPv6 gateway with IPv4 route is not supported.

This can be removed in time once support is added to each.

With the infrastructure changes in place, patch 17 enables it by adding
support for RTA_VIA to IPv4. RTA_VIA can be used for IPv4 addresses as
well. Only one of RTA_VIA and RTA_GATEWAY can be passed in a request.

Patch 18 adds a few test cases to fib_tests.sh.

v2
- comments from Ido - fixed typos as noted and updated messages
- add commit message to patch 1
- In patch 9, ipv4: Add fib_check_nh_v6_gw, moved the call to
  fib6_nh_release under the 'if (!err)' check as the intention is
  that release should not be called if init fails.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoselftests: fib_tests: Add tests for ipv6 gateway with ipv4 route
David Ahern [Fri, 5 Apr 2019 23:30:41 +0000 (16:30 -0700)]
selftests: fib_tests: Add tests for ipv6 gateway with ipv4 route

Add tests for ipv6 gateway with ipv4 route. Tests include basic
single path with ping to verify connectivity and multipath.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Allow ipv6 gateway with ipv4 routes
David Ahern [Fri, 5 Apr 2019 23:30:40 +0000 (16:30 -0700)]
ipv4: Allow ipv6 gateway with ipv4 routes

Add support for RTA_VIA and allow an IPv6 nexthop for v4 routes:
   $ ip ro add 172.16.1.0/24 via inet6 2001:db8::1 dev eth0
   $ ip ro ls
   ...
   172.16.1.0/24 via inet6 2001:db8::1 dev eth0

For convenience and simplicity, userspace can use RTA_VIA to specify
AF_INET or AF_INET6 gateway.

The common fib_nexthop_info dump function compares the gateway address
family to the nh_common family to know if the gateway should be encoded
as RTA_VIA or RTA_GATEWAY.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Flag fib_info with a fib_nh using IPv6 gateway
David Ahern [Fri, 5 Apr 2019 23:30:39 +0000 (16:30 -0700)]
ipv4: Flag fib_info with a fib_nh using IPv6 gateway

Until support is added to the offload drivers, they need to be able to
reject routes with an IPv6 gateway. To that end add a flag to fib_info
that indicates if any fib_nh has a v6 gateway. The flag allows the drivers
to efficiently know the use of a v6 gateway without walking all fib_nh
tied to a fib_info each time a route is added.

Update mlxsw and rocker to reject the routes with extack message as to why.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Handle ipv6 gateway in fib_good_nh
David Ahern [Fri, 5 Apr 2019 23:30:38 +0000 (16:30 -0700)]
ipv4: Handle ipv6 gateway in fib_good_nh

Update fib_good_nh to handle an ipv6 gateway.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Handle ipv6 gateway in fib_detect_death
David Ahern [Fri, 5 Apr 2019 23:30:37 +0000 (16:30 -0700)]
ipv4: Handle ipv6 gateway in fib_detect_death

Update fib_detect_death to handle an ipv6 gateway.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Handle ipv6 gateway in ipv4_confirm_neigh
David Ahern [Fri, 5 Apr 2019 23:30:36 +0000 (16:30 -0700)]
ipv4: Handle ipv6 gateway in ipv4_confirm_neigh

Update ipv4_confirm_neigh to handle an ipv6 gateway.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobpf: Handle ipv6 gateway in bpf_ipv4_fib_lookup
David Ahern [Fri, 5 Apr 2019 23:30:35 +0000 (16:30 -0700)]
bpf: Handle ipv6 gateway in bpf_ipv4_fib_lookup

Update bpf_ipv4_fib_lookup to handle an ipv6 gateway.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Add helpers for neigh lookup for nexthop
David Ahern [Fri, 5 Apr 2019 23:30:34 +0000 (16:30 -0700)]
ipv4: Add helpers for neigh lookup for nexthop

A common theme in the output path is looking up a neigh entry for a
nexthop, either the gateway in an rtable or a fallback to the daddr
in the skb:

        nexthop = (__force u32)rt_nexthop(rt, ip_hdr(skb)->daddr);
        neigh = __ipv4_neigh_lookup_noref(dev, nexthop);
        if (unlikely(!neigh))
                neigh = __neigh_create(&arp_tbl, &nexthop, dev, false);

To allow the nexthop to be an IPv6 address we need to consider the
family of the nexthop and then call __ipv{4,6}_neigh_lookup_noref based
on it.

To make this simpler, add a ip_neigh_gw4 helper similar to ip_neigh_gw6
added in an earlier patch which handles:

        neigh = __ipv4_neigh_lookup_noref(dev, nexthop);
        if (unlikely(!neigh))
                neigh = __neigh_create(&arp_tbl, &nexthop, dev, false);

And then add a second one, ip_neigh_for_gw, that calls either
ip_neigh_gw4 or ip_neigh_gw6 based on the address family of the gateway.

Update the output paths in the VRF driver and core v4 code to use
ip_neigh_for_gw simplifying the family based lookup and making both
ready for a v6 nexthop.

ipv4_neigh_lookup has a different need - the potential to resolve a
passed in address in addition to any gateway in the rtable or skb. Since
this is a one-off, add ip_neigh_gw4 and ip_neigh_gw6 diectly. The
difference between __neigh_create used by the helpers and neigh_create
called by ipv4_neigh_lookup is taking a refcount, so add rcu_read_lock_bh
and bump the refcnt on the neigh entry.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoneighbor: Add skip_cache argument to neigh_output
David Ahern [Fri, 5 Apr 2019 23:30:33 +0000 (16:30 -0700)]
neighbor: Add skip_cache argument to neigh_output

A later patch allows an IPv6 gateway with an IPv4 route. The neighbor
entry will exist in the v6 ndisc table and the cached header will contain
the ipv6 protocol which is wrong for an IPv4 packet. For an IPv4 packet to
use the v6 neighbor entry, neigh_output needs to skip the cached header
and just use the output callback for the neigh entry.

A future patchset can look at expanding the hh_cache to handle 2
protocols. For now, IPv6 gateways with an IPv4 route will take the
extra overhead of generating the header.

Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Add fib_check_nh_v6_gw
David Ahern [Fri, 5 Apr 2019 23:30:32 +0000 (16:30 -0700)]
ipv4: Add fib_check_nh_v6_gw

Add helper to use fib6_nh_init to validate a nexthop spec with an IPv6
gateway.

Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Refactor fib_check_nh
David Ahern [Fri, 5 Apr 2019 23:30:31 +0000 (16:30 -0700)]
ipv4: Refactor fib_check_nh

fib_check_nh is currently huge covering multiple uses cases - device only,
device + gateway, and device + gateway with ONLINK. The next patch adds
validation checks for IPv6 which only further complicates it. So, break
fib_check_nh into 2 helpers - one for gateway validation and one for device
only.

Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Add support to fib_config for IPv6 gateway
David Ahern [Fri, 5 Apr 2019 23:30:30 +0000 (16:30 -0700)]
ipv4: Add support to fib_config for IPv6 gateway

Add support for an IPv6 gateway to fib_config. Since a gateway is either
IPv4 or IPv6, make it a union with fc_gw4 where fc_gw_family decides
which address is in use. Update current checks on family and gw4 to
handle ipv6 as well.

Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Add support to rtable for ipv6 gateway
David Ahern [Fri, 5 Apr 2019 23:30:29 +0000 (16:30 -0700)]
ipv4: Add support to rtable for ipv6 gateway

Add support for an IPv6 gateway to rtable. Since a gateway is either
IPv4 or IPv6, make it a union with rt_gw4 where rt_gw_family decides
which address is in use.

When dumping the route data, encode an ipv6 nexthop using RTA_VIA.

Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Prepare fib_config for IPv6 gateway
David Ahern [Fri, 5 Apr 2019 23:30:28 +0000 (16:30 -0700)]
ipv4: Prepare fib_config for IPv6 gateway

Similar to rtable, fib_config needs to allow the gateway to be either an
IPv4 or an IPv6 address. To that end, rename fc_gw to fc_gw4 to mean an
IPv4 address and add fc_gw_family. Checks on 'is a gateway set' are changed
to see if fc_gw_family is set. In the process prepare the code for a
fc_gw_family == AF_INET6.

Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv4: Prepare rtable for IPv6 gateway
David Ahern [Fri, 5 Apr 2019 23:30:27 +0000 (16:30 -0700)]
ipv4: Prepare rtable for IPv6 gateway

To allow the gateway to be either an IPv4 or IPv6 address, remove
rt_uses_gateway from rtable and replace with rt_gw_family. If
rt_gw_family is set it implies rt_uses_gateway. Rename rt_gateway
to rt_gw4 to represent the IPv4 version.

Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: Replace nhc_has_gw with nhc_gw_family
David Ahern [Fri, 5 Apr 2019 23:30:26 +0000 (16:30 -0700)]
net: Replace nhc_has_gw with nhc_gw_family

Allow the gateway in a fib_nh_common to be from a different address
family than the outer fib{6}_nh. To that end, replace nhc_has_gw with
nhc_gw_family and update users of nhc_has_gw to check nhc_gw_family.
Now nhc_family is used to know if the nh_common is part of a fib_nh
or fib6_nh (used for container_of to get to route family specific data),
and nhc_gw_family represents the address family for the gateway.

Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Add neighbor helpers that use the ipv6 stub
David Ahern [Fri, 5 Apr 2019 23:30:25 +0000 (16:30 -0700)]
ipv6: Add neighbor helpers that use the ipv6 stub

Add ipv6 helpers to handle ndisc references via the stub. Update
bpf_ipv6_fib_lookup to use __ipv6_neigh_lookup_noref_stub instead of
the open code ___neigh_lookup_noref with the stub.

Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Add fib6_nh_init and release to stubs
David Ahern [Fri, 5 Apr 2019 23:30:24 +0000 (16:30 -0700)]
ipv6: Add fib6_nh_init and release to stubs

Add fib6_nh_init and fib6_nh_release to ipv6_stubs. If fib6_nh_init fails,
callers should not invoke fib6_nh_release, so there is no reason to have
a dummy stub for the IPv6 is not enabled case.

Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agor8169: disable ASPM again
Heiner Kallweit [Fri, 5 Apr 2019 18:46:46 +0000 (20:46 +0200)]
r8169: disable ASPM again

There's a significant number of reports that re-enabling ASPM causes
different issues, ranging from decreased performance to system not
booting at all. This affects only a minority of users, but the number
of affected users is big enough that we better switch off ASPM again.

This will hurt notebook users who are not affected by the issues, they
may see decreased battery runtime w/o ASPM. With the PCI core folks is
being discussed to add generic sysfs attributes to control ASPM.
Once this is in place brave enough users can re-enable ASPM on their
system.

Fixes: a99790bf5c7f ("r8169: Reinstate ASPM Support")
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: phy: improve link partner capability detection
Heiner Kallweit [Fri, 5 Apr 2019 19:23:13 +0000 (21:23 +0200)]
net: phy: improve link partner capability detection

genphy_read_status() so far checks phydev->supported, not the actual
PHY capabilities. This can make a difference if the supported speeds
have been limited by of_set_phy_supported() or phy_set_max_speed().

It seems that this issue only affects the link partner advertisements
as displayed by ethtool. Also this patch wouldn't apply to older
kernels because linkmode bitmaps have been introduced recently.
Therefore net-next.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge tag 'mlx5-updates-2019-04-02' of git://git.kernel.org/pub/scm/linux/kernel...
David S. Miller [Mon, 8 Apr 2019 21:31:25 +0000 (14:31 -0700)]
Merge tag 'mlx5-updates-2019-04-02' of git://git./linux/kernel/git/saeed/linux

Saeed Mamameed says:

====================
mlx5-updates-2019-04-02

This series provides misc updates to mlx5 driver

1) Aya Levin (1): Handle event of power detection in the PCIE slot

2) Eli Britstein (6):
  Some TC VLAN related updates and fixes to the previous VLAN modify action
  support patchset.
  Offload TC e-switch rules with egress/ingress VLAN devices

3) Max Gurtovoy (1): Fix double mutex initialization in esiwtch.c

4) Tariq Toukan (3): Misc small updates
  A write memory barrier is sufficient in EQ ci update
  Obsolete param field holding a constant value
  Unify logic of MTU boundaries

5) Tonghao Zhang (4): Misc updates to en_tc.c
  Make the log friendly when decapsulation offload not supported
  Remove 'parse_attr' argument in parse_tc_fdb_actions()
  Deletes unnecessary setting of esw_attr->parse_attr
  Return -EOPNOTSUPP when attempting to offload an unsupported action
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMAINTAINERS: ieee802154: update documentation file pattern
Stefan Schmidt [Mon, 8 Apr 2019 16:08:04 +0000 (18:08 +0200)]
MAINTAINERS: ieee802154: update documentation file pattern

When moving the documentation for the ieee802154 subsystem from
plain text to rst the file pattern in the MAINTAINERS file got wrong.
Updating it here to fix scripts using this file.

Reported-by: Joe Perches <joe@perches.com>
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agocxgb4: Don't return EAGAIN when TCAM is full.
Vishal Kulkarni [Mon, 8 Apr 2019 12:33:49 +0000 (18:03 +0530)]
cxgb4: Don't return EAGAIN when TCAM is full.

During hash filter programming, driver needs to return ENOSPC error
intead of EAGAIN when TCAM is full.

Signed-off-by: Vishal Kulkarni <vishal@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: xilinx: emaclite: add minimal ndo_do_ioctl hook
Alexandru Ardelean [Mon, 8 Apr 2019 09:01:57 +0000 (12:01 +0300)]
net: xilinx: emaclite: add minimal ndo_do_ioctl hook

This hook only implements a minimal set of ioctl hooks to be able to access
MII regs by using phytool.
When using this simple MAC controller, it's pretty difficult to do
debugging of the PHY chip without checking MII regs.

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Reviewed-by: Radhey Shyam Pandey <radhey.shyam.pandey@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agonet: xilinx: emaclite: add minimal ethtool ops
Alexandru Ardelean [Mon, 8 Apr 2019 09:01:36 +0000 (12:01 +0300)]
net: xilinx: emaclite: add minimal ethtool ops

This set adds a minimal set of ethtool hooks to the driver, which provide a
decent amount of link information via ethtool.
With this change, running `ethtool ethX` in user-space provides all the
neatly-formatted information about the link (what was negotiated, what is
advertised, etc).

Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Reviewed-by: Radhey Shyam Pandey <radhey.shyam.pandey@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agodatagram: remove rendundant 'peeked' argument
Paolo Abeni [Mon, 8 Apr 2019 08:15:59 +0000 (10:15 +0200)]
datagram: remove rendundant 'peeked' argument

After commit a297569fe00a ("net/udp: do not touch skb->peeked unless
really needed") the 'peeked' argument of __skb_try_recv_datagram()
and friends is always equal to !!'flags & MSG_PEEK'.

Since such argument is really a boolean info, and the callers have
already 'flags & MSG_PEEK' handy, we can remove it and clean-up the
code a bit.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Mon, 8 Apr 2019 05:51:09 +0000 (19:51 -1000)]
Merge branch 'linus' of git://git./linux/kernel/git/herbert/crypto-2.6

Pull crypto fix from Herbert Xu:
 "This fixes a bug in the implementation of xcbc and cmac in caam"

* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: caam - fix copy of next buffer for xcbc and cmac

5 years agonet: vrf: Fix ping failed when vrf mtu is set to 0
Miaohe Lin [Mon, 8 Apr 2019 02:04:20 +0000 (10:04 +0800)]
net: vrf: Fix ping failed when vrf mtu is set to 0

When the mtu of a vrf device is set to 0, it would cause ping
failed. So I think we should limit vrf mtu in a reasonable range
to solve this problem. I set dev->min_mtu to IPV6_MIN_MTU, so it
will works for both ipv4 and ipv6. And if dev->max_mtu still be 0
can be confusing, so I set dev->max_mtu to ETH_MAX_MTU.

Here is the reproduce step:

1.Config vrf interface and set mtu to 0:
3: enp4s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel
master vrf1 state UP mode DEFAULT group default qlen 1000
    link/ether 52:54:00:9e:dd:c1 brd ff:ff:ff:ff:ff:ff

2.Ping peer:
3: enp4s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel
master vrf1 state UP group default qlen 1000
    link/ether 52:54:00:9e:dd:c1 brd ff:ff:ff:ff:ff:ff
    inet 10.0.0.1/16 scope global enp4s0
       valid_lft forever preferred_lft forever
connect: Network is unreachable

3.Set mtu to default value, ping works:
PING 10.0.0.2 (10.0.0.2) 56(84) bytes of data.
64 bytes from 10.0.0.2: icmp_seq=1 ttl=64 time=1.88 ms

Fixes: ad49bc6361ca2 ("net: vrf: remove MTU limits for vrf device")
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoslab: fix a crash by reading /proc/slab_allocators
Qian Cai [Sat, 6 Apr 2019 22:59:01 +0000 (18:59 -0400)]
slab: fix a crash by reading /proc/slab_allocators

The commit 510ded33e075 ("slab: implement slab_root_caches list")
changes the name of the list node within "struct kmem_cache" from "list"
to "root_caches_node", but leaks_show() still use the "list" which
causes a crash when reading /proc/slab_allocators.

You need to have CONFIG_SLAB=y and CONFIG_MEMCG=y to see the problem,
because without MEMCG all slab caches are root caches, and the "list"
node happens to be the right one.

Fixes: 510ded33e075 ("slab: implement slab_root_caches list")
Signed-off-by: Qian Cai <cai@lca.pw>
Reviewed-by: Tobin C. Harding <tobin@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agonet: sched: flower: insert filter to ht before offloading it to hw
Vlad Buslov [Fri, 5 Apr 2019 17:56:26 +0000 (20:56 +0300)]
net: sched: flower: insert filter to ht before offloading it to hw

John reports:

Recent refactoring of fl_change aims to use the classifier spinlock to
avoid the need for rtnl lock. In doing so, the fl_hw_replace_filer()
function was moved to before the lock is taken. This can create problems
for drivers if duplicate filters are created (commmon in ovs tc offload
due to filters being triggered by user-space matches).

Drivers registered for such filters will now receive multiple copies of
the same rule, each with a different cookie value. This means that the
drivers would need to do a full match field lookup to determine
duplicates, repeating work that will happen in flower __fl_lookup().
Currently, drivers do not expect to receive duplicate filters.

To fix this, verify that filter with same key is not present in flower
classifier hash table and insert the new filter to the flower hash table
before offloading it to hardware. Implement helper function
fl_ht_insert_unique() to atomically verify/insert a filter.

This change makes filter visible to fast path at the beginning of
fl_change() function, which means it can no longer be freed directly in
case of error. Refactor fl_change() error handling code to deallocate the
filter with rcu timeout.

Fixes: 620da4860827 ("net: sched: flower: refactor fl_change")
Reported-by: John Hurley <john.hurley@netronome.com>
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoselftests: add a tc matchall test case
Nicolas Dichtel [Fri, 5 Apr 2019 12:20:24 +0000 (14:20 +0200)]
selftests: add a tc matchall test case

This is a follow up of the commit 0db6f8befc32 ("net/sched: fix ->get
helper of the matchall cls").

To test it:
$ cd tools/testing/selftests/tc-testing
$ ln -s ../plugin-lib/nsPlugin.py plugins/20-nsPlugin.py
$ ./tdc.py -n -e 2638

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'rhashtable-bitlocks'
David S. Miller [Mon, 8 Apr 2019 02:12:12 +0000 (19:12 -0700)]
Merge branch 'rhashtable-bitlocks'

NeilBrown says:

====================
Convert rhashtable to use bitlocks

This series converts rhashtable to use a per-bucket bitlock
rather than a separate array of spinlocks.
This:
  reduces memory usage
  results in slightly fewer memory accesses
  slightly improves parallelism
  makes a configuration option unnecessary

The main change from previous version is to use a distinct type for
the pointer in the bucket which has a bit-lock in it.  This
helped find two places where rht_ptr() was missed, one
in  rhashtable_free_and_destroy() in print_ht in the test code.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agorhashtable: add lockdep tracking to bucket bit-spin-locks.
NeilBrown [Mon, 1 Apr 2019 23:07:45 +0000 (10:07 +1100)]
rhashtable: add lockdep tracking to bucket bit-spin-locks.

Native bit_spin_locks are not tracked by lockdep.

The bit_spin_locks used for rhashtable buckets are local
to the rhashtable implementation, so there is little opportunity
for the sort of misuse that lockdep might detect.
However locks are held while a hash function or compare
function is called, and if one of these took a lock,
a misbehaviour is possible.

As it is quite easy to add lockdep support this unlikely
possibility seems to be enough justification.

So create a lockdep class for bucket bit_spin_lock and attach
through a lockdep_map in each bucket_table.

Without the 'nested' annotation in rhashtable_rehash_one(), lockdep
correctly reports a possible problem as this lock is taken
while another bucket lock (in another table) is held.  This
confirms that the added support works.
With the correct nested annotation in place, lockdep reports
no problems.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agorhashtable: use bit_spin_locks to protect hash bucket.
NeilBrown [Mon, 1 Apr 2019 23:07:45 +0000 (10:07 +1100)]
rhashtable: use bit_spin_locks to protect hash bucket.

This patch changes rhashtables to use a bit_spin_lock on BIT(1) of the
bucket pointer to lock the hash chain for that bucket.

The benefits of a bit spin_lock are:
 - no need to allocate a separate array of locks.
 - no need to have a configuration option to guide the
   choice of the size of this array
 - locking cost is often a single test-and-set in a cache line
   that will have to be loaded anyway.  When inserting at, or removing
   from, the head of the chain, the unlock is free - writing the new
   address in the bucket head implicitly clears the lock bit.
   For __rhashtable_insert_fast() we ensure this always happens
   when adding a new key.
 - even when lockings costs 2 updates (lock and unlock), they are
   in a cacheline that needs to be read anyway.

The cost of using a bit spin_lock is a little bit of code complexity,
which I think is quite manageable.

Bit spin_locks are sometimes inappropriate because they are not fair -
if multiple CPUs repeatedly contend of the same lock, one CPU can
easily be starved.  This is not a credible situation with rhashtable.
Multiple CPUs may want to repeatedly add or remove objects, but they
will typically do so at different buckets, so they will attempt to
acquire different locks.

As we have more bit-locks than we previously had spinlocks (by at
least a factor of two) we can expect slightly less contention to
go with the slightly better cache behavior and reduced memory
consumption.

To enhance type checking, a new struct is introduced to represent the
  pointer plus lock-bit
that is stored in the bucket-table.  This is "struct rhash_lock_head"
and is empty.  A pointer to this needs to be cast to either an
unsigned lock, or a "struct rhash_head *" to be useful.
Variables of this type are most often called "bkt".

Previously "pprev" would sometimes point to a bucket, and sometimes a
->next pointer in an rhash_head.  As these are now different types,
pprev is NULL when it would have pointed to the bucket. In that case,
'blk' is used, together with correct locking protocol.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agorhashtable: allow rht_bucket_var to return NULL.
NeilBrown [Mon, 1 Apr 2019 23:07:45 +0000 (10:07 +1100)]
rhashtable: allow rht_bucket_var to return NULL.

Rather than returning a pointer to a static nulls, rht_bucket_var()
now returns NULL if the bucket doesn't exist.
This will make the next patch, which stores a bitlock in the
bucket pointer, somewhat cleaner.

This change involves introducing __rht_bucket_nested() which is
like rht_bucket_nested(), but doesn't provide the static nulls,
and changing rht_bucket_nested() to call this and possible
provide a static nulls - as is still needed for the non-var case.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>