GitHub/moto-9609/android_kernel_motorola_exynos9610.git
8 years agousbnet: smsc95xx: silence an uninitialized variable warning
Dan Carpenter [Wed, 4 May 2016 06:22:01 +0000 (09:22 +0300)]
usbnet: smsc95xx: silence an uninitialized variable warning

If the call to fn() fails then "buf" is uninitialized.  Just return the
error code in that case.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agousbnet/smsc75xx: silence uninitialized variable warning
Dan Carpenter [Wed, 4 May 2016 06:21:02 +0000 (09:21 +0300)]
usbnet/smsc75xx: silence uninitialized variable warning

If the fn() calls fail then "buf" is uninitialized.  Just return early
in that situation.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotcp: must block bh in __inet_twsk_hashdance()
Eric Dumazet [Wed, 4 May 2016 00:10:50 +0000 (17:10 -0700)]
tcp: must block bh in __inet_twsk_hashdance()

__inet_twsk_hashdance() might be called from process context,
better block BH before acquiring bind hash and established locks

Fixes: c10d9310edf5 ("tcp: do not assume TCP code is non preemptible")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotcp: fix lockdep splat in tcp_snd_una_update()
Eric Dumazet [Tue, 3 May 2016 23:56:03 +0000 (16:56 -0700)]
tcp: fix lockdep splat in tcp_snd_una_update()

tcp_snd_una_update() and tcp_rcv_nxt_update() call
u64_stats_update_begin() either from process context or BH handler.

This triggers a lockdep splat on 32bit & SMP builds.

We could add u64_stats_update_begin_bh() variant but this would
slow down 32bit builds with useless local_disable_bh() and
local_enable_bh() pairs, since we own the socket lock at this point.

I add sock_owned_by_me() helper to have proper lockdep support
even on 64bit builds, and new u64_stats_update_begin_raw()
and u64_stats_update_end_raw methods.

Fixes: c10d9310edf5 ("tcp: do not assume TCP code is non preemptible")
Reported-by: Fabio Estevam <festevam@gmail.com>
Diagnosed-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Tested-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge tag 'batman-adv-for-davem' of git://git.open-mesh.org/linux-merge
David S. Miller [Wed, 4 May 2016 20:21:08 +0000 (16:21 -0400)]
Merge tag 'batman-adv-for-davem' of git://git.open-mesh.org/linux-merge

Antonio Quartulli says:

====================
pull request: batman-adv 20160504

In this pull request you have:
- two changes to the MAINTAINERS file where one marks our mailing list
  as moderated and the other adds a missing documentation file
- kernel-doc fixes
- code refactoring and various cleanups
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agomdio_bus: don't return NULL from mdiobus_scan()
Sergei Shtylyov [Tue, 3 May 2016 20:14:41 +0000 (23:14 +0300)]
mdio_bus: don't return NULL from mdiobus_scan()

I've finally noticed that mdiobus_scan() also returns either NULL or error
value on failure.  Return ERR_PTR(-ENODEV) instead of NULL since this is
the  error value  already filtered out by the callers that want to ignore
the  MDIO address scan failure...

Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'kill_trans_start'
David S. Miller [Wed, 4 May 2016 18:16:51 +0000 (14:16 -0400)]
Merge branch 'kill_trans_start'

Florian Westphal says:

====================
net: remove trans_start from struct net_device

We currently have two instances for trans_start, once in
net_device and once in netdev_queue.

This series removes trans_start from net_device.
Updates to dev->trans_start are replaced with updates to netdev queue 0.

This series is compile-tested only.
Replacement is done in 3 steps:

1. Replace read-accesses:
  x = dev->trans_start

  gets replaced by
  x = dev_trans_start(dev)

2. Replace write accesses:
  dev->trans_start = jiffies;

  gets replaced with new helper:
  netif_trans_update(dev);

3. This helper is then changed to set
   netdev_get_tx_queue(dev, 0)->trans_start
   instead of dev->trans_start.

After this dev->trans_start can be removed.

It should be noted that after this series several instances
of netif_trans_update() are useless (if they occur in
.ndo_start_xmit and driver doesn't set LLTX flag -- stack already
did an update).

Comments welcome.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: remove dev->trans_start
Florian Westphal [Tue, 3 May 2016 14:33:14 +0000 (16:33 +0200)]
net: remove dev->trans_start

previous patches removed all direct accesses to dev->trans_start,
so change the netif_trans_update helper to update trans_start of
netdev queue 0 instead and then remove trans_start from struct net_device.

AFAICS a lot of the netif_trans_update() invocations are now useless
because they occur in ndo_start_xmit and driver doesn't set LLTX
(i.e. stack already took care of the update).

As I can't test any of them it seems better to just leave them alone.

Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotreewide: replace dev->trans_start update with helper
Florian Westphal [Tue, 3 May 2016 14:33:13 +0000 (16:33 +0200)]
treewide: replace dev->trans_start update with helper

Replace all trans_start updates with netif_trans_update helper.
change was done via spatch:

struct net_device *d;
@@
- d->trans_start = jiffies
+ netif_trans_update(d)

Compile tested only.

Cc: user-mode-linux-devel@lists.sourceforge.net
Cc: linux-xtensa@linux-xtensa.org
Cc: linux1394-devel@lists.sourceforge.net
Cc: linux-rdma@vger.kernel.org
Cc: netdev@vger.kernel.org
Cc: MPT-FusionLinux.pdl@broadcom.com
Cc: linux-scsi@vger.kernel.org
Cc: linux-can@vger.kernel.org
Cc: linux-parisc@vger.kernel.org
Cc: linux-omap@vger.kernel.org
Cc: linux-hams@vger.kernel.org
Cc: linux-usb@vger.kernel.org
Cc: linux-wireless@vger.kernel.org
Cc: linux-s390@vger.kernel.org
Cc: devel@driverdev.osuosl.org
Cc: b.a.t.m.a.n@lists.open-mesh.org
Cc: linux-bluetooth@vger.kernel.org
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Acked-by: Mugunthan V N <mugunthanvnm@ti.com>
Acked-by: Antonio Quartulli <a@unstable.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonetdevice: add helper to update trans_start
Florian Westphal [Tue, 3 May 2016 14:31:00 +0000 (16:31 +0200)]
netdevice: add helper to update trans_start

trans_start exists twice:
- as member of net_device (legacy)
- as member of netdev_queue

In order to get rid of the legacy case, add a helper for the
dev->trans_update (this patch), then convert spots that do

dev->trans_start = jiffies

to use this helper (next patch).

This would then allow us to change the helper so that it updates the
trans_stamp of netdev queue 0 instead.

Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agodrivers: replace dev->trans_start accesses with dev_trans_start
Florian Westphal [Tue, 3 May 2016 14:30:59 +0000 (16:30 +0200)]
drivers: replace dev->trans_start accesses with dev_trans_start

a trans_start struct member exists twice:
- in struct net_device (legacy)
- in struct netdev_queue

Instead of open-coding dev->trans_start usage to obtain the current
trans_start value, use dev_trans_start() instead.

This is not exactly the same, as dev_trans_start also considers
the trans_start values of the netdev queues owned by the device
and provides the most recent one.

For legacy devices this doesn't matter as dev_trans_start can cope
with netdev trans_start values of 0 (they are ignored).

This is a prerequisite to eventual removal of dev->trans_start.

Cc: linux-rdma@vger.kernel.org
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agodmfe: kill DEVICE define
Florian Westphal [Tue, 3 May 2016 16:53:15 +0000 (18:53 +0200)]
dmfe: kill DEVICE define

use net_device directly. Compile tested, objdiff shows no changes.

Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agogre6: add Kconfig dependency for NET_IPGRE_DEMUX
Arnd Bergmann [Tue, 3 May 2016 15:19:57 +0000 (17:19 +0200)]
gre6: add Kconfig dependency for NET_IPGRE_DEMUX

The ipv6 gre implementation was cleaned up to share more code
with the ipv4 version, but it can be enabled even when NET_IPGRE_DEMUX
is disabled, resulting in a link error:

net/built-in.o: In function `gre_rcv':
:(.text+0x17f5d0): undefined reference to `gre_parse_header'
ERROR: "gre_parse_header" [net/ipv6/ip6_gre.ko] undefined!

This adds a Kconfig dependency to prevent that now invalid
configuration.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: 308edfdf1563 ("gre6: Cleanup GREv6 receive path, call common GRE functions")
Acked-by: Tom Herbert <tom@herbertland.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'gre-teb'
David S. Miller [Wed, 4 May 2016 18:11:32 +0000 (14:11 -0400)]
Merge branch 'gre-teb'

Jiri Benc says:

====================
gre: receive also TEB packets for lwtunnels

NOTE: this patchset needs net merged to net-next.

This allows lwtunnel users to get also packets with ETH_P_TEB protocol
specified in GRE header through an ipgre interface. There's really nothing
special about these packets in the case of lwtunnels - it's just an inner
protocol like any other. The only complications stem from keeping
compatibility with other uses of GRE.

This will be used by openvswitch to support eth_push and eth_pop actions.
I'd also like to see tc support for lwtunnels (this feature included) in the
future.

The first patch is not directly related and can be submitted standalone if
needed.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agogre: receive also TEB packets for lwtunnels
Jiri Benc [Tue, 3 May 2016 15:10:08 +0000 (17:10 +0200)]
gre: receive also TEB packets for lwtunnels

For ipgre interfaces in collect metadata mode, receive also traffic with
encapsulated Ethernet headers. The lwtunnel users are supposed to sort this
out correctly. This allows to have mixed Ethernet + L3-only traffic on the
same lwtunnel interface. This is the same way as VXLAN-GPE behaves.

To keep backwards compatibility and prevent any surprises, gretap interfaces
have priority in receiving packets with Ethernet headers.

Signed-off-by: Jiri Benc <jbenc@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agogre: move iptunnel_pull_header down to ipgre_rcv
Jiri Benc [Tue, 3 May 2016 15:10:07 +0000 (17:10 +0200)]
gre: move iptunnel_pull_header down to ipgre_rcv

This will allow to make the pull dependent on the tunnel type.

Signed-off-by: Jiri Benc <jbenc@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agogre: remove superfluous pskb_may_pull
Jiri Benc [Tue, 3 May 2016 15:10:06 +0000 (17:10 +0200)]
gre: remove superfluous pskb_may_pull

The call to gre_parse_header is either followed by iptunnel_pull_header, or
in the case of ICMP error path, the actual header is not accessed at all.

In the first case, iptunnel_pull_header will call pskb_may_pull anyway and
it's pointless to do it twice. The only difference is what call will fail
with what error code but the net effect is still the same in all call sites.

In the second case, pskb_may_pull is pointless, as skb->data is at the outer
IP header and not at the GRE header.

Signed-off-by: Jiri Benc <jbenc@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'mlx5-sriov-updates'
David S. Miller [Wed, 4 May 2016 18:04:49 +0000 (14:04 -0400)]
Merge branch 'mlx5-sriov-updates'

Saeed Mahameed says:

====================
Mellanox 100G ethernet SRIOV Upgrades

This series introduces new features and upgrades for mlx5 etherenet SRIOV,
while the first patch provides a bug fixes for a compilation issue introduced
buy the previous aRFS series for when CONFIG_RFS_ACCEL=y and CONFIG_MLX5_CORE_EN=n.

Changes from V0:
    - 1st patch: Don't add a new Kconfig flag.  Instead, compile out en_arfs.c \
contents when CONFIG_RFS_ACCEL=n

SRIOV upgrades:
    - Use synchronize_irq instead of the vport events spin_lock
    - Fix memory leak in error flow
    - Added full VST support
    - Spoofcheck support
    - Trusted VF promiscuous and allmulti support

VST and Spoofcheck in details:
    - Adding Low level firmware commands support for creating ACLs
     (Access Control Lists) Flow tables.  ACLs are regular flow tables with
     the only exception that they are bound to a specific e-Switch vport (VF)
     and they can be one of two types
        > egress ACL: filters traffic going from e-Switch to VF.
        > ingress ACL: filters traffic going from VF to e-Switch.
    - Ingress/Egress ACLs (per vport) for VF VST mode filtering.
    - Ingress/Egress ACLs (per vport) for VF spoofcheck filtering.
    - Ingress/Egress ACLs (per vport) configuration:
        > Created only when at least one of (VST, spoofcheck) is configured.
> if (!spoofchk && !vst) allow all traffic.  i.e. no ACLs.
        > if (spoofchk && vst) allow only untagged traffic with smac=original mac \
                sent from the VF.
        > if (spoofchk && !vst) allow only traffic with smac=original mac sent from \
the VF.  > if (!spoofchk && vst) allow only untagged traffic.

Trusted VF promiscuous and allmulti support in details:
    - Added two flow groups for allmulti and promisc VFs to the e-Switch FDB table
        > Allmulti group: One rule that forwards any mcast traffic coming from
                          either uplink or VFs/PF vports.
        > Promisc group: One rule that forwards all unmatched traffic coming from \
                uplink.
    - Add vport context change event handling for promisc and allmulti
      If VF is trusted respect the request and:
        > if allmulti request: add the vport to the allmulti group.
          and to all other L2 mcast address in the FDB table.
        > if promisc request: add the vport to the promisc group.
        > Note: A promisc VF can only see traffic that was not explicitly matched to
                or requested by any other VF.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: E-Switch, Implement trust vf ndo
Mohamad Haj Yahia [Tue, 3 May 2016 14:14:04 +0000 (17:14 +0300)]
net/mlx5: E-Switch, Implement trust vf ndo

- Add support to configure trusted vf attribute through trust_vf_ndo.

- Upon VF trust setting change we update vport context to refresh
 allmulti/promisc or any trusted vf attributes that we didn't trust the
 VF for before.

- Lock the eswitch state lock on vport event in order to synchronise the
 vport context updates , this will prevent contention with vport trust
 setting change which will trigger vport mac list update.

Signed-off-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: E-Switch, Implement promiscuous rx modes vf request handling
Mohamad Haj Yahia [Tue, 3 May 2016 14:14:03 +0000 (17:14 +0300)]
net/mlx5: E-Switch, Implement promiscuous rx modes vf request handling

Add promisc_change as a trigger to vport context change event.
Add set vport promisc/allmulti functions to add vport to promiscuous
flowtable rules.
Upon promisc/allmulti rx mode vf request add the vport to
the relevant promiscuous group (Allmulti/Promisc group) so the relevant
traffic will be forwarded to it.
Upon allmulti vf request add the vport to each existing multicast fdb
rule.
Upon adding/removing mcast address from a vport, update all other
allmulti vports.

Signed-off-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: E-Switch, Add promiscuous and allmulti FDB flowtable groups
Mohamad Haj Yahia [Tue, 3 May 2016 14:14:02 +0000 (17:14 +0300)]
net/mlx5: E-Switch, Add promiscuous and allmulti FDB flowtable groups

Add promiscuous and allmulti steering groups in FDB table.
Besides the full match L2 steering rules group, we added
two more groups to catch the "miss" rules traffic:
* Allmulti group: One rule that forwards any mcast traffic coming from
either uplink or VFs/PF vports
* Promisc group: One rule that forwards all unmatched traffic coming
from uplink.

Needed for downstream privileged VF promisc and allmulti support.

Signed-off-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: E-Switch, Use vport event handler for vport cleanup
Mohamad Haj Yahia [Tue, 3 May 2016 14:14:01 +0000 (17:14 +0300)]
net/mlx5: E-Switch, Use vport event handler for vport cleanup

Remove the usage of explicit cleanup function and use existing vport
change handler. Calling vport change handler while vport
is disabled will cleanup the vport resources.

Signed-off-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: E-Switch, Enable/disable ACL tables on demand
Mohamad Haj Yahia [Tue, 3 May 2016 14:14:00 +0000 (17:14 +0300)]
net/mlx5: E-Switch, Enable/disable ACL tables on demand

Enable ingress/egress ACL tables only when we need to configure ACL
rules.
Disable ingress/egress ACL tables once all ACL rules are removed.

All VF outgoing/incoming traffic need to go through the ingress/egress ACL
tables.
Adding/Removing these tables on demand will save unnecessary hops in the
flow steering when the ACL tables are empty.

Signed-off-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: E-Switch, Vport ingress/egress ACLs rules for spoofchk
Mohamad Haj Yahia [Tue, 3 May 2016 14:13:59 +0000 (17:13 +0300)]
net/mlx5: E-Switch, Vport ingress/egress ACLs rules for spoofchk

Configure ingress and egress vport ACL rules according to spoofchk
admin parameters.

Ingress ACL flow table rules:
if (!spoofchk && !vst) allow all traffic.
else :
1) one of the following rules :
* if (spoofchk && vst) allow only untagged traffic with smac=original
mac sent from the VF.
* if (spoofchk && !vst) allow only traffic with smac=original mac sent
from the VF.
* if (!spoofchk && vst) allow only untagged traffic.
2) drop all traffic that didn't hit #1.

Add support for set vf spoofchk ndo.

Add non zero mac validation in case of spoofchk to set mac ndo:
when setting new mac we need to validate that the new mac is
not zero while the spoofchk is on because it is illegal
combination.

Signed-off-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: E-Switch, Vport ingress/egress ACLs rules for VST mode
Mohamad Haj Yahia [Tue, 3 May 2016 14:13:58 +0000 (17:13 +0300)]
net/mlx5: E-Switch, Vport ingress/egress ACLs rules for VST mode

Configure ingress and egress vport ACL rules according to
vlan and qos admin parameters.

Ingress ACL flow table rules:
1) drop any tagged packet sent from the VF
2) allow other traffic (default behavior)

Egress ACL flow table rules:
1) allow only tagged traffic with vlan_tag=vst_vid.
2) drop other traffic.

Signed-off-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: E-Switch, Introduce VST vport ingress/egress ACLs
Mohamad Haj Yahia [Tue, 3 May 2016 14:13:57 +0000 (17:13 +0300)]
net/mlx5: E-Switch, Introduce VST vport ingress/egress ACLs

Create egress/ingress ACLs per VF vport at vport enable.

Ingress ACL:
- one flow group to drop all tagged traffic in VST mode.

Egress ACL:
- one flow group that allows only untagged traffic with
          smac that is equals to the original mac (anti-spoofing).
        - one flow group that allows only untagged traffic.
        - one flow group that allows only  smac that is equals
          to the original mac (anti-spoofing).
        (note: only one of the above group has active rule)
- star rule will be used to drop all other traffic.

By default no rules are generated, unless VST is explicitly requested.

Signed-off-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: E-Switch, Fix error flow memory leak
Mohamad Haj Yahia [Tue, 3 May 2016 14:13:56 +0000 (17:13 +0300)]
net/mlx5: E-Switch, Fix error flow memory leak

Fix memory leak in case query nic vport command failed.

Fixes: 81848731ff40 ('net/mlx5: E-Switch, Add SR-IOV (FDB) support')
Signed-off-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: E-Switch, Replace vport spin lock with synchronize_irq()
Mohamad Haj Yahia [Tue, 3 May 2016 14:13:55 +0000 (17:13 +0300)]
net/mlx5: E-Switch, Replace vport spin lock with synchronize_irq()

Vport spin lock can be replaced with synchronize_irq() in the right
place, this will remove the need of locking inside irq context.
Locking in esw_enable_vport is not required since vport events are yet
to be enabled, and at esw_disable_vport it is sufficient to
synchronize_irq() to guarantee no further vport events handlers will be
scheduled.

Signed-off-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: Flow steering, Add vport ACL support
Mohamad Haj Yahia [Tue, 3 May 2016 14:13:54 +0000 (17:13 +0300)]
net/mlx5: Flow steering, Add vport ACL support

Update the relevant flow steering device structs and commands to
support vport.
Update the flow steering core API to receive vport number.
Add ingress and egress ACL flow table name spaces.
Add ACL flow table support:
* ACL (Access Control List) flow table is a table that contains
only allow/drop steering rules.

* We have two types of ACL flow tables - ingress and egress.

* ACLs handle traffic sent from/to E-Switch FDB table, Ingress refers to
traffic sent from Vport to E-Switch and Egress refers to traffic sent
from E-Switch to vport.

* Ingress ACL flow table allow/drop rules is checked against traffic
sent from VF.

* Egress ACL flow table allow/drop rules is checked against traffic sent
to VF.

Signed-off-by: Mohamad Haj Yahia <mohamad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5e: Fix aRFS compilation dependency
Maor Gottlieb [Tue, 3 May 2016 14:13:53 +0000 (17:13 +0300)]
net/mlx5e: Fix aRFS compilation dependency

en_arfs.o should be compiled only if both CONFIG_MLX5_CORE_EN
and CONFIG_RFS_ACCEL are enabled. en_arfs calls to rps_may_expire_flow
which is compiled only if CONFIG_RFS_ACCEL is defined.

Move en_arfs.o compilation dependency to be under CONFIG_MLX5_CORE_EN
and wrap the en_arfs.c content with ifdef of CONFIG_RFS_ACCEL.

Fixes: 1cabe6b0965e ('net/mlx5e: Create aRFS flow tables')
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Reported-by: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'cxgb4-mbox'
David S. Miller [Wed, 4 May 2016 17:59:28 +0000 (13:59 -0400)]
Merge branch 'cxgb4-mbox'

Hariprasad Shenai says:

====================
cxgb4: mbox enhancements for cxgb4

This patch series checks for firmware errors when we are waiting for
mbox response in a loop and breaks out. When negative timeout is passed
to mailbox code, don't sleep. Negative timeout is passed only from
interrupt context.

This patch series has been created against net-next tree and includes
patches on cxgb4 driver.

We have included all the maintainers of respective drivers. Kindly review
the change and let us know in case of any review comments.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agocxgb4: Check for firmware errors in the mailbox command loop
Hariprasad Shenai [Tue, 3 May 2016 13:28:02 +0000 (18:58 +0530)]
cxgb4: Check for firmware errors in the mailbox command loop

Check for firmware errors in the mailbox command loop and report
them differently rather than simply timing out when the firmware goes
belly up.

Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agocxgb4: Don't sleep when mbox cmd is issued from interrupt context
Hariprasad Shenai [Tue, 3 May 2016 13:28:01 +0000 (18:58 +0530)]
cxgb4: Don't sleep when mbox cmd is issued from interrupt context

When link goes down, from the interrupt handler DCB priority for the
Tx queues needs to be unset. We issue mbox command to unset the Tx queue
priority with negative timeout. In t4_wr_mbox_meat_timeout() do not sleep
when negative timeout is passed, since it is called from interrupt context.

Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agodrivers: net: emac: add Atheros AR8035 phy initialization code
Christian Lamparter [Tue, 3 May 2016 12:08:30 +0000 (14:08 +0200)]
drivers: net: emac: add Atheros AR8035 phy initialization code

This patch adds the phy initialization code for Qualcomm
Atheros AR8035 phy. This configuration is found in the
Cisco Meraki MR24.

Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'tunnel-features-and-gso-partial'
David S. Miller [Wed, 4 May 2016 17:32:29 +0000 (13:32 -0400)]
Merge branch 'tunnel-features-and-gso-partial'

Alexander Duyck says:

====================
Fix Tunnel features and enable GSO partial for several drivers

This patch series is meant to allow us to get the best performance possible
for Mellanox ConnectX-3/4 and Broadcom NetXtreme-C/E adapters in terms of
VXLAN and GRE tunnels.

The first 3 patches address issues I found in regards to GSO_PARTIAL and
TSO_MANGLEID.

The next 4 patches go through and enable GSO_PARTIAL for VXLAN tunnels that
have an outer checksum enabled, and then enable IPv6 support where I can.
One outstanding issue is that I wasn't able to get offloads working with
outer IPv6 headers on mlx4.  However that wasn't a feature that was enabled
before so it isn't technically a regression, however I believe Engineers
from Mellanox said they would look into it since they thought it should be
supported.

The last patch enables GSO_PARTIAL for VXLAN and GRE tunnels on the bnxt
driver.  One piece of feedback I received on the patch was that the
hardware has globally set IPv6 UDP tunnels to always have the checksum
field computed.  I plan to work with Broadcom to get that addressed so that
we only populate the checksum field if it was requested by the network
stack.

v2: Rebased patches off of latest changes to the mlx4/mlx5 drivers.
    Added bnxt driver patch as I received feedback on the RFC.
v3: Moved 2 patches into series for net as they were generic fixes.
    Added patch to disable GSO partial if frame is less than 2x size of MSS

    There are outstanding issues as called out above that need to be
    addressed, however they were present before these patches so it isn't
    as if they introduce a regression.  In addition gains can be easily
    seen so there should be no issue with applying the driver patches while
    the IPv6 mlx4_en and bnxt issues are being researched.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobnxt: Add support for segmentation of tunnels with outer checksums
Alexander Duyck [Mon, 2 May 2016 16:38:55 +0000 (09:38 -0700)]
bnxt: Add support for segmentation of tunnels with outer checksums

This patch assumes that the bnxt hardware will ignore existing IPv4/v6
header fields for length and checksum as well as the length and checksum
fields for outer UDP and GRE headers.

I have been told by Michael Chan that this is working.  Though this might
be somewhat redundant for IPv6 as they are forcing the checksum to be
computed for all IPv6 frames that are offloaded.  A follow-up patch may be
necessary in order to fix this as it is essentially mangling the outer IPv6
headers to add a checksum where none was requested.

Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5e: Fix IPv6 tunnel checksum offload
Alexander Duyck [Mon, 2 May 2016 16:38:49 +0000 (09:38 -0700)]
net/mlx5e: Fix IPv6 tunnel checksum offload

The mlx5 driver exposes support for TSO6 but not IPv6 csum for hardware
encapsulated tunnels.  This leads to issues as it triggers warnings in
skb_checksum_help as it ends up being called as we report supporting the
segmentation but not the checksumming for IPv6 frames.

This patch corrects that and drops 2 features that don't actually need to
be supported in hw_enc_features since they are Rx features and don't
actually impact anything by being present in hw_enc_features.

Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5e: Add support for UDP tunnel segmentation with outer checksum offload
Alexander Duyck [Mon, 2 May 2016 16:38:43 +0000 (09:38 -0700)]
net/mlx5e: Add support for UDP tunnel segmentation with outer checksum offload

This patch assumes that the mlx5 hardware will ignore existing IPv4/v6
header fields for length and checksum as well as the length and checksum
fields for outer UDP headers.

Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx4_en: Add support for inner IPv6 checksum offloads and TSO
Alexander Duyck [Mon, 2 May 2016 16:38:37 +0000 (09:38 -0700)]
net/mlx4_en: Add support for inner IPv6 checksum offloads and TSO

>From what I can tell the ConnectX-3 will support an inner IPv6 checksum and
segmentation offload, however it cannot support outer IPv6 headers.  This
assumption is based on the fact that I could see the checksum being
offloaded for inner header on IPv4 tunnels, but not on IPv6 tunnels.

For this reason I am adding the feature to the hw_enc_features and adding
an extra check to the features_check call that will disable GSO and
checksum offload in the case that the encapsulated frame has an outer IP
version of that is not 4.  The check in mlx4_en_features_check could be
removed if at some point in the future a fix is found that allows the
hardware to offload segmentation/checksum on tunnels with an outer IPv6
header.

Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx4_en: Add support for UDP tunnel segmentation with outer checksum offload
Alexander Duyck [Mon, 2 May 2016 16:38:30 +0000 (09:38 -0700)]
net/mlx4_en: Add support for UDP tunnel segmentation with outer checksum offload

This patch assumes that the mlx4 hardware will ignore existing IPv4/v6
header fields for length and checksum as well as the length and checksum
fields for outer UDP headers.

Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: Fix netdev_fix_features so that TSO_MANGLEID is only available with TSO
Alexander Duyck [Mon, 2 May 2016 16:38:24 +0000 (09:38 -0700)]
net: Fix netdev_fix_features so that TSO_MANGLEID is only available with TSO

This change makes it so that we will strip the TSO_MANGLEID bit if TSO is
not present.  This way we will also handle ECN correctly of TSO is not
present.

Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agogso: Only allow GSO_PARTIAL if we can checksum the inner protocol
Alexander Duyck [Mon, 2 May 2016 16:38:18 +0000 (09:38 -0700)]
gso: Only allow GSO_PARTIAL if we can checksum the inner protocol

This patch addresses a possible issue that can occur if we get into any odd
corner cases where we support TSO for a given protocol but not the checksum
or scatter-gather offload.  There are few drivers floating around that
setup their tunnels this way and by enforcing the checksum piece we can
avoid mangling any frames.

Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agogso: Do not perform partial GSO if number of partial segments is 1 or less
Alexander Duyck [Mon, 2 May 2016 16:38:12 +0000 (09:38 -0700)]
gso: Do not perform partial GSO if number of partial segments is 1 or less

In the event that the number of partial segments is equal to 1 we don't
really need to perform partial segmentation offload.  As such we should
skip multiplying the MSS and instead just clear the partial_segs value
since it will not provide any gain to advertise the frame as being GSO when
it is a single frame.

Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agogre: change gre_parse_header to return the header length
Jiri Benc [Tue, 3 May 2016 13:00:21 +0000 (15:00 +0200)]
gre: change gre_parse_header to return the header length

It's easier for gre_parse_header to return the header length instead of
filing it into a parameter. That way, the callers that don't care about the
header length can just check whether the returned value is lower than zero.

In gre_err, the tunnel header must not be pulled. See commit b7f8fe251e46
("gre: do not pull header in ICMP error processing") for details.

This patch reduces the conflict between the mentioned commit and commit
95f5c64c3c13 ("gre: Move utility functions to common headers").

Signed-off-by: Jiri Benc <jbenc@redhat.com>
Acked-by: Tom Herbert <tom@herbertland.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotcp: guarantee forward progress in tcp_sendmsg()
Eric Dumazet [Tue, 3 May 2016 04:49:25 +0000 (21:49 -0700)]
tcp: guarantee forward progress in tcp_sendmsg()

Under high rx pressure, it is possible tcp_sendmsg() never has a
chance to allocate an skb and loop forever as sk_flush_backlog()
would always return true.

Fix this by calling sk_flush_backlog() only if one skb had been
allocated and filled before last backlog check.

Fixes: d41a69f1d390 ("tcp: make tcp_sendmsg() aware of socket backlog")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.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 [Wed, 4 May 2016 04:52:29 +0000 (00:52 -0400)]
Merge git://git./linux/kernel/git/davem/net

Conflicts:
net/ipv4/ip_gre.c

Minor conflicts between tunnel bug fixes in net and
ipv6 tunnel cleanups in net-next.

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Tue, 3 May 2016 22:07:50 +0000 (15:07 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:
 "Some straggler bug fixes:

   1) Batman-adv DAT must consider VLAN IDs when choosing candidate
      nodes, from Antonio Quartulli.

   2) Fix botched reference counting of vlan objects and neigh nodes in
      batman-adv, from Sven Eckelmann.

   3) netem can crash when it sees GSO packets, the fix is to segment
      then upon ->enqueue.  Fix from Neil Horman with help from Eric
      Dumazet.

   4) Fix VXLAN dependencies in mlx5 driver Kconfig, from Matthew
      Finlay.

   5) Handle VXLAN ops outside of rcu lock, via a workqueue, in mlx5,
      since it can sleep.  Fix also from Matthew Finlay.

   6) Check mdiobus_scan() return values properly in pxa168_eth and macb
      drivers.  From Sergei Shtylyov.

   7) If the netdevice doesn't support checksumming, disable
      segmentation.  From Alexandery Duyck.

   8) Fix races between RDS tcp accept and sending, from Sowmini
      Varadhan.

   9) In macb driver, probe MDIO bus before we register the netdev,
      otherwise we can try to open the device before it is really ready
      for that.  Fix from Florian Fainelli.

  10) Netlink attribute size for ILA "tunnels" not calculated properly,
      fix from Nicolas Dichtel"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
  ipv6/ila: fix nlsize calculation for lwtunnel
  net: macb: Probe MDIO bus before registering netdev
  RDS: TCP: Synchronize accept() and connect() paths on t_conn_lock.
  RDS:TCP: Synchronize rds_tcp_accept_one with rds_send_xmit when resetting t_sock
  vxlan: Add checksum check to the features check function
  net: Disable segmentation if checksumming is not supported
  net: mvneta: Remove superfluous SMP function call
  macb: fix mdiobus_scan() error check
  pxa168_eth: fix mdiobus_scan() error check
  net/mlx5e: Use workqueue for vxlan ops
  net/mlx5e: Implement a mlx5e workqueue
  net/mlx5: Kconfig: Fix MLX5_EN/VXLAN build issue
  net/mlx5: Unmap only the relevant IO memory mapping
  netem: Segment GSO packets on enqueue
  batman-adv: Fix reference counting of hardif_neigh_node object for neigh_node
  batman-adv: Fix reference counting of vlan object for tt_local_entry
  batman-adv: B.A.T.M.A.N V - make sure iface is reactivated upon NETDEV_UP event
  batman-adv: fix DAT candidate selection (must use vid)

8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi...
Linus Torvalds [Tue, 3 May 2016 21:23:58 +0000 (14:23 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/mszeredi/fuse

Pull fuse fixes from Miklos Szeredi:
 "Fix a regression and update the MAINTAINERS entry for fuse"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse:
  fuse: update mailing list in MAINTAINERS
  fuse: Fix return value from fuse_get_user_pages()

8 years agoipv6/ila: fix nlsize calculation for lwtunnel
Nicolas Dichtel [Tue, 3 May 2016 07:58:27 +0000 (09:58 +0200)]
ipv6/ila: fix nlsize calculation for lwtunnel

The handler 'ila_fill_encap_info' adds one attribute: ILA_ATTR_LOCATOR.

Fixes: 65d7ab8de582 ("net: Identifier Locator Addressing module")
CC: Tom Herbert <tom@herbertland.com>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoipv6: add new struct ipcm6_cookie
Wei Wang [Tue, 3 May 2016 04:40:07 +0000 (21:40 -0700)]
ipv6: add new struct ipcm6_cookie

In the sendmsg function of UDP, raw, ICMP and l2tp sockets, we use local
variables like hlimits, tclass, opt and dontfrag and pass them to corresponding
functions like ip6_make_skb, ip6_append_data and xxx_push_pending_frames.
This is not a good practice and makes it hard to add new parameters.
This fix introduces a new struct ipcm6_cookie similar to ipcm_cookie in
ipv4 and include the above mentioned variables. And we only pass the
pointer to this structure to corresponding functions. This makes it easier
to add new parameters in the future and makes the function cleaner.

Signed-off-by: Wei Wang <weiwan@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: macb: Probe MDIO bus before registering netdev
Florian Fainelli [Tue, 3 May 2016 01:38:45 +0000 (18:38 -0700)]
net: macb: Probe MDIO bus before registering netdev

The current sequence makes us register for a network device prior to
registering and probing the MDIO bus which could lead to some unwanted
consequences, like a thread of execution calling into ndo_open before
register_netdev() returns, while the MDIO bus is not ready yet.

Rework the sequence to register for the MDIO bus, and therefore attach
to a PHY prior to calling register_netdev(), which implies reworking the
error path a bit.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'rds-fixes'
David S. Miller [Tue, 3 May 2016 20:03:45 +0000 (16:03 -0400)]
Merge branch 'rds-fixes'

Sowmini Varadhan says:

====================
RDS: TCP: sychronization during connection startup

This patch series ensures that the passive (accept) side of the
TCP connection used for RDS-TCP is correctly synchronized with
any concurrent active (connect) attempts for a given pair of peers.

Patch 1 in the series makes sure that the t_sock in struct
rds_tcp_connection is only reset after any threads in rds_tcp_xmit
have completed (otherwise a null-ptr deref may be encountered).
Patch 2 synchronizes rds_tcp_accept_one() with the rds_tcp*connect()
path.

v2: review comments from Santosh Shilimkar, other spelling corrections
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoRDS: TCP: Synchronize accept() and connect() paths on t_conn_lock.
Sowmini Varadhan [Mon, 2 May 2016 18:24:52 +0000 (11:24 -0700)]
RDS: TCP: Synchronize accept() and connect() paths on t_conn_lock.

An arbitration scheme for duelling SYNs is implemented as part of
commit 241b271952eb ("RDS-TCP: Reset tcp callbacks if re-using an
outgoing socket in rds_tcp_accept_one()") which ensures that both nodes
involved will arrive at the same arbitration decision. However, this
needs to be synchronized with an outgoing SYN to be generated by
rds_tcp_conn_connect(). This commit achieves the synchronization
through the t_conn_lock mutex in struct rds_tcp_connection.

The rds_conn_state is checked in rds_tcp_conn_connect() after acquiring
the t_conn_lock mutex.  A SYN is sent out only if the RDS connection is
not already UP (an UP would indicate that rds_tcp_accept_one() has
completed 3WH, so no SYN needs to be generated).

Similarly, the rds_conn_state is checked in rds_tcp_accept_one() after
acquiring the t_conn_lock mutex. The only acceptable states (to
allow continuation of the arbitration logic) are UP (i.e., outgoing SYN
was SYN-ACKed by peer after it sent us the SYN) or CONNECTING (we sent
outgoing SYN before we saw incoming SYN).

Signed-off-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoRDS:TCP: Synchronize rds_tcp_accept_one with rds_send_xmit when resetting t_sock
Sowmini Varadhan [Mon, 2 May 2016 18:24:51 +0000 (11:24 -0700)]
RDS:TCP: Synchronize rds_tcp_accept_one with rds_send_xmit when resetting t_sock

There is a race condition between rds_send_xmit -> rds_tcp_xmit
and the code that deals with resolution of duelling syns added
by commit 241b271952eb ("RDS-TCP: Reset tcp callbacks if re-using an
outgoing socket in rds_tcp_accept_one()").

Specifically, we may end up derefencing a null pointer in rds_send_xmit
if we have the interleaving sequence:
           rds_tcp_accept_one                  rds_send_xmit

                                             conn is RDS_CONN_UP, so
      invoke rds_tcp_xmit

                                             tc = conn->c_transport_data
        rds_tcp_restore_callbacks
            /* reset t_sock */
      null ptr deref from tc->t_sock

The race condition can be avoided without adding the overhead of
additional locking in the xmit path: have rds_tcp_accept_one wait
for rds_tcp_xmit threads to complete before resetting callbacks.
The synchronization can be done in the same manner as rds_conn_shutdown().
First set the rds_conn_state to something other than RDS_CONN_UP
(so that new threads cannot get into rds_tcp_xmit()), then wait for
RDS_IN_XMIT to be cleared in the conn->c_flags indicating that any
threads in rds_tcp_xmit are done.

Fixes: 241b271952eb ("RDS-TCP: Reset tcp callbacks if re-using an
outgoing socket in rds_tcp_accept_one()")
Signed-off-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: add __sock_wfree() helper
Eric Dumazet [Mon, 2 May 2016 17:56:27 +0000 (10:56 -0700)]
net: add __sock_wfree() helper

Hosts sending lot of ACK packets exhibit high sock_wfree() cost
because of cache line miss to test SOCK_USE_WRITE_QUEUE

We could move this flag close to sk_wmem_alloc but it is better
to perform the atomic_sub_and_test() on a clean cache line,
as it avoid one extra bus transaction.

skb_orphan_partial() can also have a fast track for packets that either
are TCP acks, or already went through another skb_orphan_partial()

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'tunnel-csum-and-sg-offloads'
David S. Miller [Tue, 3 May 2016 20:00:55 +0000 (16:00 -0400)]
Merge branch 'tunnel-csum-and-sg-offloads'

Alexander Duyck says:

====================
Fixes for tunnel checksum and segmentation offloads

This patch series is a subset of patches I had submitted for net-next.  I
plan to drop these two patches from the v3 of "Fix Tunnel features and
enable GSO partial for several drivers" and I am instead submitting them
for net since these are truly fixes and likely will need to be backported
to stable branches.

This series addresses 2 specific issues.  The first is that we could
request TSO on a v4 inner header while not supporting checksum offload of
the outer IPv6 header.  The second is that we could request an IPv6 inner
checksum offload without validating that we could actually support an inner
IPv6 checksum offload.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agovxlan: Add checksum check to the features check function
Alexander Duyck [Mon, 2 May 2016 16:25:16 +0000 (09:25 -0700)]
vxlan: Add checksum check to the features check function

We need to perform an additional check on the inner headers to determine if
we can offload the checksum for them.  Previously this check didn't occur
so we would generate an invalid frame in the case of an IPv6 header
encapsulated inside of an IPv4 tunnel.  To fix this I added a secondary
check to vxlan_features_check so that we can verify that we can offload the
inner checksum.

Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: Disable segmentation if checksumming is not supported
Alexander Duyck [Mon, 2 May 2016 16:25:10 +0000 (09:25 -0700)]
net: Disable segmentation if checksumming is not supported

In the case of the mlx4 and mlx5 driver they do not support IPv6 checksum
offload for tunnels.  With this being the case we should disable GSO in
addition to the checksum offload features when we find that a device cannot
perform a checksum on a given packet type.

Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'tipc-next'
David S. Miller [Tue, 3 May 2016 19:51:17 +0000 (15:51 -0400)]
Merge branch 'tipc-next'

Jon Maloy says:

====================
tipc: redesign socket-level flow control

The socket-level flow control in TIPC has long been due for a major
overhaul. This series fixes this.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotipc: redesign connection-level flow control
Jon Paul Maloy [Mon, 2 May 2016 15:58:47 +0000 (11:58 -0400)]
tipc: redesign connection-level flow control

There are two flow control mechanisms in TIPC; one at link level that
handles network congestion, burst control, and retransmission, and one
at connection level which' only remaining task is to prevent overflow
in the receiving socket buffer. In TIPC, the latter task has to be
solved end-to-end because messages can not be thrown away once they
have been accepted and delivered upwards from the link layer, i.e, we
can never permit the receive buffer to overflow.

Currently, this algorithm is message based. A counter in the receiving
socket keeps track of number of consumed messages, and sends a dedicated
acknowledge message back to the sender for each 256 consumed message.
A counter at the sending end keeps track of the sent, not yet
acknowledged messages, and blocks the sender if this number ever reaches
512 unacknowledged messages. When the missing acknowledge arrives, the
socket is then woken up for renewed transmission. This works well for
keeping the message flow running, as it almost never happens that a
sender socket is blocked this way.

A problem with the current mechanism is that it potentially is very
memory consuming. Since we don't distinguish between small and large
messages, we have to dimension the socket receive buffer according
to a worst-case of both. I.e., the window size must be chosen large
enough to sustain a reasonable throughput even for the smallest
messages, while we must still consider a scenario where all messages
are of maximum size. Hence, the current fix window size of 512 messages
and a maximum message size of 66k results in a receive buffer of 66 MB
when truesize(66k) = 131k is taken into account. It is possible to do
much better.

This commit introduces an algorithm where we instead use 1024-byte
blocks as base unit. This unit, always rounded upwards from the
actual message size, is used when we advertise windows as well as when
we count and acknowledge transmitted data. The advertised window is
based on the configured receive buffer size in such a way that even
the worst-case truesize/msgsize ratio always is covered. Since the
smallest possible message size (from a flow control viewpoint) now is
1024 bytes, we can safely assume this ratio to be less than four, which
is the value we are now using.

This way, we have been able to reduce the default receive buffer size
from 66 MB to 2 MB with maintained performance.

In order to keep this solution backwards compatible, we introduce a
new capability bit in the discovery protocol, and use this throughout
the message sending/reception path to always select the right unit.

Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotipc: propagate peer node capabilities to socket layer
Jon Paul Maloy [Mon, 2 May 2016 15:58:46 +0000 (11:58 -0400)]
tipc: propagate peer node capabilities to socket layer

During neighbor discovery, nodes advertise their capabilities as a bit
map in a dedicated 16-bit field in the discovery message header. This
bit map has so far only be stored in the node structure on the peer
nodes, but we now see the need to keep a copy even in the socket
structure.

This commit adds this functionality.

Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotipc: re-enable compensation for socket receive buffer double counting
Jon Paul Maloy [Mon, 2 May 2016 15:58:45 +0000 (11:58 -0400)]
tipc: re-enable compensation for socket receive buffer double counting

In the refactoring commit d570d86497ee ("tipc: enqueue arrived buffers
in socket in separate function") we did by accident replace the test

if (sk->sk_backlog.len == 0)
     atomic_set(&tsk->dupl_rcvcnt, 0);

with

if (sk->sk_backlog.len)
     atomic_set(&tsk->dupl_rcvcnt, 0);

This effectively disables the compensation we have for the double
receive buffer accounting that occurs temporarily when buffers are
moved from the backlog to the socket receive queue. Until now, this
has gone unnoticed because of the large receive buffer limits we are
applying, but becomes indispensable when we reduce this buffer limit
later in this series.

We now fix this by inverting the mentioned condition.

Acked-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agortl8152: correct speed testing
Oliver Neukum [Mon, 2 May 2016 11:06:14 +0000 (13:06 +0200)]
rtl8152: correct speed testing

Allow for SS+ USB

Signed-off-by: Oliver Neukum <ONeukum@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agousbnet: correct speed testing
Oliver Neukum [Mon, 2 May 2016 11:06:13 +0000 (13:06 +0200)]
usbnet: correct speed testing

Allow for SS+ USB

Signed-off-by: Oliver Neukum <ONeukum@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobrcm80211: correct speed testing
Oliver Neukum [Mon, 2 May 2016 11:06:12 +0000 (13:06 +0200)]
brcm80211: correct speed testing

Allow for SS+ USB

Signed-off-by: Oliver Neukum <ONeukum@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoqed: Apply tunnel configurations after PF start
Manish Chopra [Mon, 2 May 2016 10:16:04 +0000 (06:16 -0400)]
qed: Apply tunnel configurations after PF start

Configure and enable various tunnels on the
adapter after PF start.

This change was missed as a part of
'commit 464f664501816ef5fbbc00b8de96f4ae5a1c9325
("qed: Add infrastructure support for tunneling")'

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: Yuval Mintz <yuval.mintz@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: mvneta: Remove superfluous SMP function call
Anna-Maria Gleixner [Mon, 2 May 2016 09:02:51 +0000 (11:02 +0200)]
net: mvneta: Remove superfluous SMP function call

Since commit 3b9d6da67e11 ("cpu/hotplug: Fix rollback during error-out
in __cpu_disable()") it is ensured that callbacks of CPU_ONLINE and
CPU_DOWN_PREPARE are processed on the hotplugged CPU. Due to this SMP
function calls are no longer required.

Replace smp_call_function_single() with a direct call to
mvneta_percpu_enable() or mvneta_percpu_disable(). The functions do
not require to be called with interrupts disabled, therefore the
smp_call_function_single() calling convention is not preserved.

Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: netdev@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'stmmac-dwmac-socfpga-cleanup'
David S. Miller [Tue, 3 May 2016 19:22:21 +0000 (15:22 -0400)]
Merge branch 'stmmac-dwmac-socfpga-cleanup'

Joachim Eastwood says:

====================
stmmac: dwmac-socfpga refactor+cleanup

This patch aims to remove the init/exit callbacks from the dwmac-
socfpga driver and instead use standard PM callbacks. Doing this
will also allow us to cleanup the driver.

Eventually the init/exit callbacks will be deprecated and removed
from all drivers dwmac-* except for dwmac-generic. Drivers will be
refactored to use standard PM and remove callbacks.

This patch set should not change the behavior of the driver itself,
it only moves code around. The only exception to this is patch
number 4 which restores the resume callback behavior which was
changed in the "net: stmmac: socfpga: Remove re-registration of
reset controller" patch. I belive calling phy_resume() only
from the resume callback and not probe is the right thing to do.

Changes from v1:
 - Rebase on net-next

One heads-up here:
The first patch changes the prototype of a couple of
functions used in Alexandre's "add Ethernet glue logic for
stm32 chip" patch [1] and will cause build failures for
dwmac-stm32.c if not fixed up!
If Alexandre's patch set is applied first I will gladly
rebase my patch set to account for his driver as well.

[1] https://patchwork.ozlabs.org/patch/614405/
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agostmmac: dwmac-socfpga: kill init() and rename setup() to set_phy_mode()
Joachim Eastwood [Sun, 1 May 2016 20:58:23 +0000 (22:58 +0200)]
stmmac: dwmac-socfpga: kill init() and rename setup() to set_phy_mode()

Remove old init callback which now contains only a call to
socfpga_dwmac_setup(). Also rename socfpga_dwmac_setup() to indicate
what the function really does.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
Tested-by: Marek Vasut <marex@denx.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agostmmac: dwmac-socfpga: call phy_resume() only in resume callback
Joachim Eastwood [Sun, 1 May 2016 20:58:22 +0000 (22:58 +0200)]
stmmac: dwmac-socfpga: call phy_resume() only in resume callback

Calling phy_resume() should only be need during driver resume to
workaround a hardware errata.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
Tested-by: Marek Vasut <marex@denx.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agostmmac: dwmac-socfpga: keep a copy of stmmac_rst in driver priv data
Joachim Eastwood [Sun, 1 May 2016 20:58:21 +0000 (22:58 +0200)]
stmmac: dwmac-socfpga: keep a copy of stmmac_rst in driver priv data

The dwmac-socfpga driver needs to control the reset usually managed
by the core driver to set the PHY mode. Take a copy of the reset
handle from core priv data so it can be used by the driver later.

This also allow us to move reset handling into socfpga_dwmac_setup()
where the code that needs it is located.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
Tested-by: Marek Vasut <marex@denx.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agostmmac: dwmac-socfpga: add PM ops and resume function
Joachim Eastwood [Sun, 1 May 2016 20:58:20 +0000 (22:58 +0200)]
stmmac: dwmac-socfpga: add PM ops and resume function

Implement the needed PM callbacks in the driver instead of
relying on the init/exit hooks in stmmac_platform. This gives
the driver more flexibility in how the code is organized.

Eventually the init/exit callbacks will be deprecated in favor
of the standard PM callbacks and driver remove function.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
Tested-by: Marek Vasut <marex@denx.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agostmmac: let remove/resume/suspend functions take device pointer
Joachim Eastwood [Sun, 1 May 2016 20:58:19 +0000 (22:58 +0200)]
stmmac: let remove/resume/suspend functions take device pointer

Change stmmac_remove/resume/suspend to take a device pointer so
they can be used directly by drivers that doesn't need to perform
anything device specific.

This lets us remove the PCI pm functions and later simplifiy the
platform drivers.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
Tested-by: Marek Vasut <marex@denx.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agomacb: fix mdiobus_scan() error check
Sergei Shtylyov [Sat, 30 Apr 2016 22:47:36 +0000 (01:47 +0300)]
macb: fix mdiobus_scan() error check

Now mdiobus_scan() returns ERR_PTR(-ENODEV) instead of NULL if the PHY
device ID was read as all ones. As this was not  an error before, this
value  should be filtered out now in this driver.

Fixes: b74766a0a0fe ("phylib: don't return NULL from get_phy_device()")
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agopxa168_eth: fix mdiobus_scan() error check
Sergei Shtylyov [Sat, 30 Apr 2016 20:35:11 +0000 (23:35 +0300)]
pxa168_eth: fix mdiobus_scan() error check

Since mdiobus_scan() returns either an error code or NULL on error, the
driver should check  for both,  not only for NULL, otherwise a crash is
imminent...

Reported-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobatman-adv: Split batadv_iv_ogm_orig_del_if function
Sven Eckelmann [Mon, 22 Feb 2016 21:56:34 +0000 (22:56 +0100)]
batman-adv: Split batadv_iv_ogm_orig_del_if function

batadv_iv_ogm_orig_del_if handles two different buffers bcast_own and
bcast_own_sum which should be resized. The error handling two for
allocating these buffers causes the complexity of this function. This can
be avoided completely when the function is split into a main function
handling the locking, freeing and call of the subfunctions.

The subfunction can then independently handle the resize of the buffers.
This also allows to easily reuse the old buffer (which always is larger) in
case a smaller buffer could not be allocated without increasing the code
complexity.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agobatman-adv: Merge batadv_v_ogm_orig_update into batadv_v_ogm_route_update
Simon Wunderlich [Mon, 1 Feb 2016 14:21:38 +0000 (15:21 +0100)]
batman-adv: Merge batadv_v_ogm_orig_update into batadv_v_ogm_route_update

Since batadv_v_ogm_orig_update() was only called from one place and the
calling function became very short, merge these two functions together.

This should also reflect the protocol description of B.A.T.M.A.N. V
better.

Signed-off-by: Simon Wunderlich <simon@open-mesh.com>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agobatman-adv: move and restructure batadv_v_ogm_forward
Simon Wunderlich [Mon, 1 Feb 2016 14:21:37 +0000 (15:21 +0100)]
batman-adv: move and restructure batadv_v_ogm_forward

To match our code better to the protocol description of B.A.T.M.A.N. V,
move batadv_v_ogm_forward() out into batadv_v_ogm_process_per_outif()
and move all checks directly deciding whether the OGM should be
forwarded into batadv_v_ogm_forward().

Signed-off-by: Simon Wunderlich <simon@open-mesh.com>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agobatman-adv: fix debuginfo macro style issue
Simon Wunderlich [Fri, 11 Mar 2016 13:01:11 +0000 (14:01 +0100)]
batman-adv: fix debuginfo macro style issue

Structure initialization within the macros should follow the general
coding style used in the kernel: put the initialization of the first
variable and the closing brace on a separate line.

Reported-by: Antonio Quartulli <a@unstable.cc>
Signed-off-by: Simon Wunderlich <simon.wunderlich@open-mesh.com>
[sven@narfation.org: fix conflicts with current version]
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agobatman-adv: Fix function names on new line starting with '*'
Sven Eckelmann [Sat, 5 Mar 2016 14:56:01 +0000 (15:56 +0100)]
batman-adv: Fix function names on new line starting with '*'

Some really long function names in batman-adv require a newline between
return type and the function name. This has lead to some lines starting
with *batadv_...

This * belongs to the return type and thus should be on the same line as
the return type.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agobatman-adv: Add kernel-doc for batadv_interface_rx
Sven Eckelmann [Sun, 28 Feb 2016 10:38:50 +0000 (11:38 +0100)]
batman-adv: Add kernel-doc for batadv_interface_rx

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agobatman-adv: Fix kerneldoc for batadv_compare_claim
Sven Eckelmann [Mon, 22 Feb 2016 17:55:32 +0000 (18:55 +0100)]
batman-adv: Fix kerneldoc for batadv_compare_claim

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agobatman-adv: Fix checkpatch warning about 'unsigned' type
Sven Eckelmann [Wed, 9 Mar 2016 21:22:51 +0000 (22:22 +0100)]
batman-adv: Fix checkpatch warning about 'unsigned' type

checkpatch.pl warns about the use of 'unsigned' as a short form for
'unsigned int'.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agobatman-adv: fix wrong names in kerneldoc
Antonio Quartulli [Fri, 11 Mar 2016 15:36:19 +0000 (16:36 +0100)]
batman-adv: fix wrong names in kerneldoc

Signed-off-by: Antonio Quartulli <a@unstable.cc>
[sven@narfation.org: Fix additional names]
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
8 years agobatman-adv: use to_delayed_work
Geliang Tang [Mon, 28 Dec 2015 15:43:37 +0000 (23:43 +0800)]
batman-adv: use to_delayed_work

Use to_delayed_work() instead of open-coding it.

Signed-off-by: Geliang Tang <geliangtang@163.com>
Reviewed-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agobatman-adv: use list_for_each_entry_safe
Geliang Tang [Fri, 18 Dec 2015 15:33:31 +0000 (23:33 +0800)]
batman-adv: use list_for_each_entry_safe

Use list_for_each_entry_safe() instead of list_for_each_safe() to
simplify the code.

Signed-off-by: Geliang Tang <geliangtang@163.com>
Acked-by: Antonio Quartulli <a@unstable.cc>
Reviewed-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agobatman-adv: use static string for table headers
Antonio Quartulli [Sat, 12 Mar 2016 09:30:18 +0000 (10:30 +0100)]
batman-adv: use static string for table headers

Use a static string when showing table headers rather then
a nonsense parametric one with fixed arguments.

It is easier to grep and it does not need to be recomputed
at runtime each time.

Reported-by: Joe Perches <joe@perches.com>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
[sven@narfation.org: fix conflicts with current version]
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <mareklindner@neomailbox.ch>
8 years agobatman-adv: Start new development cycle
Simon Wunderlich [Fri, 15 Apr 2016 10:14:39 +0000 (12:14 +0200)]
batman-adv: Start new development cycle

Signed-off-by: Simon Wunderlich <sw@simonwunderlich.de>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agoMAINTAINERS: Add BATMAN ADVANCED documentation files
Sven Eckelmann [Fri, 1 Apr 2016 13:17:36 +0000 (15:17 +0200)]
MAINTAINERS: Add BATMAN ADVANCED documentation files

The sysfs ABI documentation files and the batman-adv.txt are maintained by
the BATMAN ADVANCED maintainers and patches for them should therefore be
sent to them.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agoMAINTAINERS: Mark BATMAN ADVANCED mailing list as moderated
Sven Eckelmann [Fri, 1 Apr 2016 13:17:35 +0000 (15:17 +0200)]
MAINTAINERS: Mark BATMAN ADVANCED mailing list as moderated

The mailing list of b.a.t.m.a.n@lists.open-mesh.org is moderated for
non-subscribers and non-whitelisted addresses. Such mails will be delayed
but the sender will not be informed about the moderation.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Antonio Quartulli <a@unstable.cc>
8 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid
Linus Torvalds [Tue, 3 May 2016 18:06:01 +0000 (11:06 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/hid

Pull HID fixes from Jiri Kosina:
 "Fixes for the HID subsystem:

   - regression fix for Wacom driver; commit introduced in 4.6-rc1
     mistakenly removed line that should be kept.  Fix by Ping Cheng

   - two device-specific quirks, by Ping Cheng and Nazar Mokrynskyi"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid:
  HID: wacom: add missed stylus_in_proximity line back
  HID: Fix boot delay for Creative SB Omni Surround 5.1 with quirk
  HID: wacom: Add support for DTK-1651

8 years agoMerge tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 3 May 2016 17:58:29 +0000 (10:58 -0700)]
Merge tag 'clk-fixes-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clk fix from Stephen Boyd:
 "One small bug fix for the imx6qp CAN clk definition that was causing
  failures and division by zeros in the kernel on those devices"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  clk: imx6q: fix typo in CAN clock definition

8 years agoMerge branch 'mlx5-fixes'
David S. Miller [Tue, 3 May 2016 17:37:27 +0000 (13:37 -0400)]
Merge branch 'mlx5-fixes'

Saeed Mahameed says:

====================
Mellanox 100G mlx5 fixes for 4.6-rc

This small series provides some bug fixes for mlx5 driver.

A small bug fix for iounmap of a null pointer, which dumps a warning on some archs.

One patch to fix the VXLAN/MLX5_EN dependency issue reported by Arnd.

Two patches to fix the scheduling while atomic issue for ndo_add/del_vxlan_port
NDOs.  The first will add an internal mlx5e workqueue and the second will
delegate vxlan ports add/del requests to that workqueue.

Note: ('net/mlx5: Kconfig: Fix MLX5_EN/VXLAN build issue') is only needed for net
and not net-next as the issue was globally fixed for all device drivers by:
b7aade15485a ('vxlan: break dependency with netdev drivers') in net-next.

Applied on top: f27337e16f2d ('ip_tunnel: fix preempt warning in ip tunnel creation/updating')
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5e: Use workqueue for vxlan ops
Matthew Finlay [Sun, 1 May 2016 19:59:57 +0000 (22:59 +0300)]
net/mlx5e: Use workqueue for vxlan ops

The vxlan add/delete port NDOs are called under rcu lock.
The current mlx5e implementation can potentially block in these
calls, which is not allowed.  Move to using the mlx5e workqueue
to handle these NDOs.

Fixes: b3f63c3d5e2c ('net/mlx5e: Add netdev support for VXLAN tunneling')
Signed-off-by: Matthew Finlay <matt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5e: Implement a mlx5e workqueue
Matthew Finlay [Sun, 1 May 2016 19:59:56 +0000 (22:59 +0300)]
net/mlx5e: Implement a mlx5e workqueue

Implement a mlx5e workqueue to handle all mlx5e specific tasks.  Move
all tasks currently using the system workqueue to the new workqueue.
This is in preparation for vxlan using the mlx5e workqueue in order to
schedule port add/remove operations.

Signed-off-by: Matthew Finlay <matt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: Kconfig: Fix MLX5_EN/VXLAN build issue
Matthew Finlay [Sun, 1 May 2016 19:59:55 +0000 (22:59 +0300)]
net/mlx5: Kconfig: Fix MLX5_EN/VXLAN build issue

When MLX5_EN=y MLX5_CORE=y and VXLAN=m there is a linker error for
vxlan_get_rx_port() due to the fact that VXLAN is a module. Change Kconfig
to select VXLAN when MLX5_CORE=y. When MLX5_CORE=m there is no dependency
on the value of VXLAN.

Fixes: b3f63c3d5e2c ('net/mlx5e: Add netdev support for VXLAN tunneling')
Signed-off-by: Matthew Finlay <matt@mellanox.com>
Reported-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: Unmap only the relevant IO memory mapping
Gal Pressman [Sun, 1 May 2016 19:59:54 +0000 (22:59 +0300)]
net/mlx5: Unmap only the relevant IO memory mapping

When freeing UAR the driver tries to unmap uar->map and uar->bf_map
which are mutually exclusive thus always unmapping a NULL pointer.
Make sure we only call iounmap() once, for the actual mapping.

Fixes: 0ba422410bbf ('net/mlx5: Fix global UAR mapping')
Signed-off-by: Gal Pressman <galp@mellanox.com>
Reported-by: Doron Tsur <doront@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: ethernet: fec_mpc52xx: move to new ethtool api {get|set}_link_ksettings
Philippe Reynes [Sun, 1 May 2016 15:08:11 +0000 (17:08 +0200)]
net: ethernet: fec_mpc52xx: move to new ethtool api {get|set}_link_ksettings

The ethtool api {get|set}_settings is deprecated.
We move the fec_mpc52xx driver to new api {get|set}_link_ksettings.

Signed-off-by: Philippe Reynes <tremyfr@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: ethernet: fs-enet: move to new ethtool api {get|set}_link_ksettings
Philippe Reynes [Sun, 1 May 2016 15:08:10 +0000 (17:08 +0200)]
net: ethernet: fs-enet: move to new ethtool api {get|set}_link_ksettings

The ethtool api {get|set}_settings is deprecated.
We move the fs-enet driver to new api {get|set}_link_ksettings.

Signed-off-by: Philippe Reynes <tremyfr@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: ethernet: ucc: move to new ethtool api {get|set}_link_ksettings
Philippe Reynes [Sun, 1 May 2016 15:08:09 +0000 (17:08 +0200)]
net: ethernet: ucc: move to new ethtool api {get|set}_link_ksettings

The ethtool api {get|set}_settings is deprecated.
We move the ucc driver to new api {get|set}_link_ksettings.

Signed-off-by: Philippe Reynes <tremyfr@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>