platform/kernel/linux-starfive.git
8 years agonet_sched: sch_plug: use a private throttled status
Eric Dumazet [Fri, 10 Jun 2016 23:41:36 +0000 (16:41 -0700)]
net_sched: sch_plug: use a private throttled status

We want to get rid of generic qdisc throttled management,
so this qdisc has to use a private flag.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'mdio-iProc-SOC'
David S. Miller [Sat, 11 Jun 2016 06:24:55 +0000 (23:24 -0700)]
Merge branch 'mdio-iProc-SOC'

Pramod Kumar says:

====================
Add MDIO bus multiplexer support for iProc SoCs

Broadcom iProc based SoCs use a MDIO bus multiplexer where child buses
could be internal as well external to SoCs. These buses could supports
MDIO transaction compatible to C-22/C-45.

Broadcom MDIO bus multiplexer is an integrated multiplexer where child bus
selection and mdio transaction logic lies inside multiplexer itself.
To accommodate this multiplexer in existing mux framework below changes
were required-

1. Passed MDIO parent bus via mdio_mux_init to MDIO mux framework.

This patch set includes MDIO bus multiplexer driver along with above
framework change. It includes one external bus node having Ethernet PHY
attached and two internal bus node holding PCIe PHYs.

This patch series is based on v4.7-rc1 and is available from github-
repo: https://github.com/Broadcom/arm64-linux.git
branch:mdio-mux-v5

-Changes from v4:
- disabled PCIe PHYs from dtsi and enabled in dts file.

-Changes from v3:
- Unregister and free the parent MDIO bus.
- rebased on net-next/master branch.

Reason for resend:
-Rebased on v4.7-rc1

Changes from v2:
-Addressed Rob's comments in this patch regarding typo/grammers.
-Addressed David's comments regarding local variables order.
-Removed property "mdio-integrated-mux" and used mdiobus_register()
in place of of_mdiobus_regsiter().
-removed usage of IS_ERR_OR_NULL to IS_ERR in PCIe PHY driver.

Changes from v1:
- stop using "brcm,is_c45" from bus node as suggested by Andrew. MDIO
PHY driver will logically OR MII_ADDR_C45 into the address when issues
any C45 MDIO read/write transaction.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agophy: Add Northstar2 PCI Phy support
Pramod Kumar [Fri, 10 Jun 2016 05:33:51 +0000 (11:03 +0530)]
phy: Add Northstar2 PCI Phy support

Add PCI Phy support for Broadcom Northstar2 SoCs.  This driver uses the
interface from the iproc mdio mux driver to enable the devices
respective phys.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Jon Mason <jonmason@broadcom.com>
Signed-off-by: Pramod Kumar <pramod.kumar@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobinding: PHY: Binding doc for NS2 PCIe PHYs.
Pramod Kumar [Fri, 10 Jun 2016 05:33:50 +0000 (11:03 +0530)]
binding: PHY: Binding doc for NS2 PCIe PHYs.

Binding doc for NS2 PCIe PHYs.

Signed-off-by: Pramod Kumar <pramod.kumar@broadcom.com>
Signed-off-by: Jon Mason <jonmason@broadcom.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: mdio-mux: Add MDIO mux driver for iProc SoCs
Pramod Kumar [Fri, 10 Jun 2016 05:33:49 +0000 (11:03 +0530)]
net: mdio-mux: Add MDIO mux driver for iProc SoCs

iProc based SoCs supports the integrated mdio multiplexer which
has the bus selection as well as mdio transaction generation logic
inside.

This multiplexer has child buses for PCIe, SATA, USB and ETH. These
buses could be internal or external to SOC where PHYs are attached.
These buses could use C-45 or C-22 mdio transaction.

Signed-off-by: Pramod Kumar <pramod.kumar@broadcom.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agodt: mdio-mux: Add mdio multiplexer driver node
Pramod Kumar [Fri, 10 Jun 2016 05:33:48 +0000 (11:03 +0530)]
dt: mdio-mux: Add mdio multiplexer driver node

Add integrated MDIO multiplexer driver node which contains
two mux PCIe bus and one ethernet bus along with phys
lying on these bus.

Signed-off-by: Pramod Kumar <pramod.kumar@broadcom.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobinding: mdio-mux: Add DT binding doc for Broadcom MDIO bus multiplexer
Pramod Kumar [Fri, 10 Jun 2016 05:33:47 +0000 (11:03 +0530)]
binding: mdio-mux: Add DT binding doc for Broadcom MDIO bus multiplexer

Add DT binding doc for Broadcom MDIO bus multiplexer driver.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Pramod Kumar <pramod.kumar@broadcom.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobinding: Make "mdio-parent-bus" property from mandatory to optional
Pramod Kumar [Fri, 10 Jun 2016 05:33:46 +0000 (11:03 +0530)]
binding: Make "mdio-parent-bus" property from mandatory to optional

Change "mdio-parent-bus" from mandatory section to optional
as it won't be required by integrated MDIO multiplexer
which has bus selection and mdio transaction generation logic,
integrated inside.

Signed-off-by: Pramod Kumar <pramod.kumar@broadcom.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agomdio: mux: Enhanced MDIO mux framework for integrated multiplexers
Pramod Kumar [Fri, 10 Jun 2016 05:33:45 +0000 (11:03 +0530)]
mdio: mux: Enhanced MDIO mux framework for integrated multiplexers

An integrated multiplexer uses same address space for
"muxed bus selection" and "generation of mdio transaction"
hence its good to register parent bus from mux driver.

Hence added a mechanism where mux driver could register a
parent bus and pass it down to framework via mdio_mux_init api.

Signed-off-by: Pramod Kumar <pramod.kumar@broadcom.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agosctp: sctp should change socket state when shutdown is received
Xin Long [Thu, 9 Jun 2016 14:48:18 +0000 (22:48 +0800)]
sctp: sctp should change socket state when shutdown is received

Now sctp doesn't change socket state upon shutdown reception. It changes
just the assoc state, even though it's a TCP-style socket.

For some cases, if we really need to check sk->sk_state, it's necessary to
fix this issue, at least when we use ss or netstat to dump, we can get a
more exact information.

As an improvement, we will change sk->sk_state when we change asoc->state
to SHUTDOWN_RECEIVED, and also do it in sctp_shutdown to keep consistent
with sctp_close.

Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Marcelo R. Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge tag 'mac80211-next-for-davem-2016-06-09' of git://git.kernel.org/pub/scm/linux...
David S. Miller [Sat, 11 Jun 2016 06:13:32 +0000 (23:13 -0700)]
Merge tag 'mac80211-next-for-davem-2016-06-09' of git://git./linux/kernel/git/jberg/mac80211-next

Johannes Berg says:

====================
For the next cycle, we have the following:
 * the biggest change is MichaƂ's work on integrating FQ/codel
   with the mac80211 internal software queues
 * cfg80211 connect result gets clarified for the
   "no connection at all" case
 * advertisement of per-interface type capabilities, in case
   they differ (which makes a lot of sense for some capabilities)
 * most of the nl80211 & hwsim unprivileged namespace operation
   changes
 * human-readable VHT capabilities in debugfs
 * some other cleanups, like spelling
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'arm64-bpf'
David S. Miller [Sat, 11 Jun 2016 06:11:50 +0000 (23:11 -0700)]
Merge branch 'arm64-bpf'

Zi Shen Lim says:

====================
arm64 BPF JIT updates

Updates for arm64 eBPF JIT.
The main addition here is implementation of bpf_tail_call.

Changes since v2:
 - None. Resubmit per David Miller.

Changes since v1:
 - Added patch #1 to address build error due to missing header inclusion
   in linux/bpf.h. (Thanks to suggestion and ack by Daniel Borkmann)
   Ordered it ahead of bpf_tail_call patch #2 so build error is not
   triggered.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoarm64: bpf: optimize LD_ABS, LD_IND
Zi Shen Lim [Thu, 9 Jun 2016 04:18:50 +0000 (21:18 -0700)]
arm64: bpf: optimize LD_ABS, LD_IND

Remove superfluous stack frame, saving us 3 instructions for every
LD_ABS or LD_IND.

Signed-off-by: Zi Shen Lim <zlim.lnx@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoarm64: bpf: optimize JMP_CALL
Zi Shen Lim [Thu, 9 Jun 2016 04:18:49 +0000 (21:18 -0700)]
arm64: bpf: optimize JMP_CALL

Remove superfluous stack frame, saving us 3 instructions for
every JMP_CALL.

Signed-off-by: Zi Shen Lim <zlim.lnx@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoarm64: bpf: implement bpf_tail_call() helper
Zi Shen Lim [Thu, 9 Jun 2016 04:18:48 +0000 (21:18 -0700)]
arm64: bpf: implement bpf_tail_call() helper

Add support for JMP_CALL_X (tail call) introduced by commit 04fd61ab36ec
("bpf: allow bpf programs to tail-call other bpf programs").

bpf_tail_call() arguments:
  ctx   - context pointer passed to next program
  array - pointer to map which type is BPF_MAP_TYPE_PROG_ARRAY
  index - index inside array that selects specific program to run

In this implementation arm64 JIT jumps into callee program after prologue,
so callee program reuses the same stack. For tail_call_cnt, we use the
callee-saved R26 (which was already saved/restored but previously unused
by JIT).

With this patch a tail call generates the following code on arm64:

  if (index >= array->map.max_entries)
      goto out;

  34:   mov     x10, #0x10                      // #16
  38:   ldr     w10, [x1,x10]
  3c:   cmp     w2, w10
  40:   b.ge    0x0000000000000074

  if (tail_call_cnt > MAX_TAIL_CALL_CNT)
      goto out;
  tail_call_cnt++;

  44:   mov     x10, #0x20                      // #32
  48:   cmp     x26, x10
  4c:   b.gt    0x0000000000000074
  50:   add     x26, x26, #0x1

  prog = array->ptrs[index];
  if (prog == NULL)
      goto out;

  54:   mov     x10, #0x68                      // #104
  58:   ldr     x10, [x1,x10]
  5c:   ldr     x11, [x10,x2]
  60:   cbz     x11, 0x0000000000000074

  goto *(prog->bpf_func + prologue_size);

  64:   mov     x10, #0x20                      // #32
  68:   ldr     x10, [x11,x10]
  6c:   add     x10, x10, #0x20
  70:   br      x10
  74:

Signed-off-by: Zi Shen Lim <zlim.lnx@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobpf: fix missing header inclusion
Zi Shen Lim [Thu, 9 Jun 2016 04:18:47 +0000 (21:18 -0700)]
bpf: fix missing header inclusion

Commit 0fc174dea545 ("ebpf: make internal bpf API independent of
CONFIG_BPF_SYSCALL ifdefs") introduced usage of ERR_PTR() in
bpf_prog_get(), however did not include linux/err.h.

Without this patch, when compiling arm64 BPF without CONFIG_BPF_SYSCALL:
...
In file included from arch/arm64/net/bpf_jit_comp.c:21:0:
include/linux/bpf.h: In function 'bpf_prog_get':
include/linux/bpf.h:235:9: error: implicit declaration of function 'ERR_PTR' [-Werror=implicit-function-declaration]
  return ERR_PTR(-EOPNOTSUPP);
         ^
include/linux/bpf.h:235:9: warning: return makes pointer from integer without a cast [-Wint-conversion]
In file included from include/linux/rwsem.h:17:0,
                 from include/linux/mm_types.h:10,
                 from include/linux/sched.h:27,
                 from arch/arm64/include/asm/compat.h:25,
                 from arch/arm64/include/asm/stat.h:23,
                 from include/linux/stat.h:5,
                 from include/linux/compat.h:12,
                 from include/linux/filter.h:10,
                 from arch/arm64/net/bpf_jit_comp.c:22:
include/linux/err.h: At top level:
include/linux/err.h:23:35: error: conflicting types for 'ERR_PTR'
 static inline void * __must_check ERR_PTR(long error)
                                   ^
In file included from arch/arm64/net/bpf_jit_comp.c:21:0:
include/linux/bpf.h:235:9: note: previous implicit declaration of 'ERR_PTR' was here
  return ERR_PTR(-EOPNOTSUPP);
         ^
...

Fixes: 0fc174dea545 ("ebpf: make internal bpf API independent of CONFIG_BPF_SYSCALL ifdefs")
Suggested-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Zi Shen Lim <zlim.lnx@gmail.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'tcp_nv'
David S. Miller [Sat, 11 Jun 2016 06:07:50 +0000 (23:07 -0700)]
Merge branch 'tcp_nv'

Lawrence Brakmo says:

====================
tcp: add NV congestion control

Removed most of the module parameters

Tested in a rack using between 1 and 380 active TCP-NV flows.

Consists of the following patches:
[PATCH net-next v2 1/2] tcp: add in_flight to tcp_skb_cb
[PATCH net-next v2 2/2] tcp: add NV congestion control
====================

Signed-off-by: Lawrence Brakmo <brakmo@fb.com>
8 years agotcp: add NV congestion control
Lawrence Brakmo [Thu, 9 Jun 2016 04:16:45 +0000 (21:16 -0700)]
tcp: add NV congestion control

TCP-NV (New Vegas) is a major update to TCP-Vegas.
An earlier version of NV was presented at 2010's LPC.
It is a delayed based congestion avoidance for the
data center. This version has been tested within a
10G rack where the HW RTTs are 20-50us and with
1 to 400 flows.

A description of TCP-NV, including implementation
details as well as experimental results, can be found at:
http://www.brakmo.org/networking/tcp-nv/TCPNV.html

Signed-off-by: Lawrence Brakmo <brakmo@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotcp: add in_flight to tcp_skb_cb
Lawrence Brakmo [Thu, 9 Jun 2016 04:16:44 +0000 (21:16 -0700)]
tcp: add in_flight to tcp_skb_cb

Add in_flight (bytes in flight when packet was sent) field
to tx component of tcp_skb_cb and make it available to
congestion modules' pkts_acked() function through the
ack_sample function argument.

Signed-off-by: Lawrence Brakmo <brakmo@fb.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'virtio_net-GSO-helpers'
David S. Miller [Sat, 11 Jun 2016 06:03:56 +0000 (23:03 -0700)]
Merge branch 'virtio_net-GSO-helpers'

Mike Rapoport says:

====================
virtio_net: use common code for virtio_net_hdr and skb GSO conversion

This patches introduce virtio_net_hdr_{from,to}_skb functions for
conversion of GSO information between skb and virtio_net_hdr.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agopacket: use common code for virtio_net_hdr and skb GSO conversion
Mike Rapoport [Wed, 8 Jun 2016 13:09:22 +0000 (16:09 +0300)]
packet: use common code for virtio_net_hdr and skb GSO conversion

Replace open coded conversion between virtio_net_hdr to skb GSO info with
virtio_net_hdr_from_skb

Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agovirtio_net: use common code for virtio_net_hdr and skb GSO conversion
Mike Rapoport [Wed, 8 Jun 2016 13:09:21 +0000 (16:09 +0300)]
virtio_net: use common code for virtio_net_hdr and skb GSO conversion

Replace open coded conversion between virtio_net_hdr to skb GSO info with
virtio_net_hdr_{from,to}_skb

Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agotuntap: use common code for virtio_net_hdr and skb GSO conversion
Mike Rapoport [Wed, 8 Jun 2016 13:09:20 +0000 (16:09 +0300)]
tuntap: use common code for virtio_net_hdr and skb GSO conversion

Replace open coded conversion between virtio_net_hdr to skb GSO info with
virtio_net_hdr_{from,to}_skb

Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agomacvtap: use common code for virtio_net_hdr and skb GSO conversion
Mike Rapoport [Wed, 8 Jun 2016 13:09:19 +0000 (16:09 +0300)]
macvtap: use common code for virtio_net_hdr and skb GSO conversion

Replace open coded conversion between virtio_net_hdr to skb GSO info with
virtio_net_hdr_{from,to}_skb

Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agovirtio_net: introduce virtio_net_hdr_{from,to}_skb
Mike Rapoport [Wed, 8 Jun 2016 13:09:18 +0000 (16:09 +0300)]
virtio_net: introduce virtio_net_hdr_{from,to}_skb

The code for conversion between virtio_net_hdr and skb GSO info is
duplicated at several places. Let's put it to a common place to allow
reuse.

Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agovirtio_net: add _UAPI prefix to virtio_net header guards
Mike Rapoport [Wed, 8 Jun 2016 13:09:17 +0000 (16:09 +0300)]
virtio_net: add _UAPI prefix to virtio_net header guards

This gives better namespacing and prevents conflicts with no-uapi version
of virtio_net header that will be introduced in the following patch.

Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoRDS: IB: Remove deprecated create_workqueue
Bhaktipriya Shridhar [Tue, 7 Jun 2016 19:33:45 +0000 (01:03 +0530)]
RDS: IB: Remove deprecated create_workqueue

alloc_workqueue replaces deprecated create_workqueue().

Since the driver is infiniband which can be used as block device and the
workqueue seems involved in regular operation of the device, so a
dedicated workqueue has been used  with WQ_MEM_RECLAIM set to guarantee
forward progress under memory pressure.
Since there are only a fixed number of work items, explicit concurrency
limit is unnecessary here.

Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoNET: PHY: adds driver for Intel XWAY PHY
Hauke Mehrtens [Sun, 5 Jun 2016 21:41:11 +0000 (23:41 +0200)]
NET: PHY: adds driver for Intel XWAY PHY

This adds support for the Intel (former Lantiq) XWAY 11G and 22E PHYs.
These PHYs are also named PEF 7061, PEF 7071, PEF 7072.

Signed-off-by: John Crispin <john@phrozen.org>
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agorxrpc: Limit the listening backlog
David Howells [Fri, 10 Jun 2016 21:30:37 +0000 (22:30 +0100)]
rxrpc: Limit the listening backlog

Limit the socket incoming call backlog queue size so that a remote client
can't pump in sufficient new calls that the server runs out of memory.  Note
that this is partially theoretical at the moment since whilst the number of
calls is limited, the number of packets trying to set up new calls is not.
This will be addressed in a later patch.

If the caller of listen() specifies a backlog INT_MAX, then they get the
current maximum; anything else greater than max_backlog or anything
negative incurs EINVAL.

The limit on the maximum queue size can be set by:

echo N >/proc/sys/net/rxrpc/max_backlog

where 4<=N<=32.

Further, set the default backlog to 0, requiring listen() to be called
before we start actually queueing new calls.  Whilst this kind of is a
change in the UAPI, the caller can't actually *accept* new calls anyway
unless they've first called listen() to put the socket into the LISTENING
state - thus the aforementioned new calls would otherwise just sit there,
eating up kernel memory.  (Note that sockets that don't have a non-zero
service ID bound don't get incoming calls anyway.)

Given that the default backlog is now 0, make the AFS filesystem call
kernel_listen() to set the maximum backlog for itself.

Possible improvements include:

 (1) Trimming a too-large backlog to max_backlog when listen is called.

 (2) Trimming the backlog value whenever the value is used so that changes
     to max_backlog are applied to an open socket automatically.  Note that
     the AFS filesystem opens one socket and keeps it open for extended
     periods, so would miss out on changes to max_backlog.

 (3) Having a separate setting for the AFS filesystem.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agorxrpc: Trim line-terminal whitespace
David Howells [Fri, 10 Jun 2016 21:30:27 +0000 (22:30 +0100)]
rxrpc: Trim line-terminal whitespace

Trim line-terminal whitespace in net/rxrpc/

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet, cls: allow for deleting all filters for given parent
Daniel Borkmann [Fri, 10 Jun 2016 21:10:22 +0000 (23:10 +0200)]
net, cls: allow for deleting all filters for given parent

Add a possibility where the user can just specify the parent and
all filters under that parent are then being purged. Currently,
for example for scripting, one needs to specify pref/prio to have
a well-defined number for 'tc filter del' command for addressing
the previously created instance or additionally filter handle in
case of priorities being the same. Improve usage by allowing the
option for tc to specify the parent and removing the whole chain
for that given parent.

Example usage after patch, no tc changes required:

  # tc qdisc replace dev foo clsact
  # tc filter add dev foo egress bpf da obj ./bpf.o
  # tc filter add dev foo egress bpf da obj ./bpf.o
  # tc filter show dev foo egress
  filter protocol all pref 49151 bpf
  filter protocol all pref 49151 bpf handle 0x1 bpf.o:[classifier] direct-action
  filter protocol all pref 49152 bpf
  filter protocol all pref 49152 bpf handle 0x1 bpf.o:[classifier] direct-action
  # tc filter del dev foo egress
  # tc filter show dev foo egress
  #

Previously, RTM_DELTFILTER requests with invalid prio of 0 were
rejected, so only netlink requests with RTM_NEWTFILTER and NLM_F_CREATE
flag were allowed where the kernel would auto-generate a pref/prio.
We can piggyback on that and use prio of 0 as a wildcard for
requests of RTM_DELTFILTER.

For notifying tc netlink monitoring users (e.g. libnl uses this
for caching), there are two options, that is, sending individual
tfilter_notify() notifications for each tcf_proto, or sending a
single one indicating wildcard removal. I tried both and there
are pros and cons for each, eventually I decided for sending
individual tfilter_notify(), so that user space can support this
seamlessly and there won't be a mess of changing each and every
application to make sure expectations from the kernel won't break
when they don't understand single notification. Since linear chains
don't really scale, I expect only a handful of classifiers to be
attached at max for a given parent anyway.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'bpf-fixes'
David S. Miller [Sat, 11 Jun 2016 01:00:57 +0000 (18:00 -0700)]
Merge branch 'bpf-fixes'

Daniel Borkmann says:

====================
bpf: couple of fixes

These are two fixes for BPF, one to introduce xmit recursion limiter for
tc bpf programs and the other one to reject filters a bit earlier. For
more details please see individual patches. I have no strong opinion
to which tree they should go, they apply to both, but I think net-next
seems okay to me.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobpf: reject wrong sized filters earlier
Daniel Borkmann [Fri, 10 Jun 2016 19:19:07 +0000 (21:19 +0200)]
bpf: reject wrong sized filters earlier

Add a bpf_check_basics_ok() and reject filters that are of invalid
size much earlier, so we don't do any useless work such as invoking
bpf_prog_alloc(). Currently, rejection happens in bpf_check_classic()
only, but it's really unnecessarily late and they should be rejected
at earliest point. While at it, also clean up one bpf_prog_size() to
make it consistent with the remaining invocations.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobpf: enforce recursion limit on redirects
Daniel Borkmann [Fri, 10 Jun 2016 19:19:06 +0000 (21:19 +0200)]
bpf: enforce recursion limit on redirects

Respect the stack's xmit_recursion limit for calls into dev_queue_xmit().
Currently, they are not handeled by the limiter when attached to clsact's
egress parent, for example, and a buggy program redirecting it to the
same device again could run into stack overflow eventually. It would be
good if we could notify an admin to give him a chance to react. We reuse
xmit_recursion instead of having one private to eBPF, so that the stack's
current recursion depth will be taken into account as well. Follow-up to
commit 3896d655f4d4 ("bpf: introduce bpf_clone_redirect() helper") and
27b29f63058d ("bpf: add bpf_redirect() helper").

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoopenvswitch: Add packet truncation support.
William Tu [Fri, 10 Jun 2016 18:49:33 +0000 (11:49 -0700)]
openvswitch: Add packet truncation support.

The patch adds a new OVS action, OVS_ACTION_ATTR_TRUNC, in order to
truncate packets. A 'max_len' is added for setting up the maximum
packet size, and a 'cutlen' field is to record the number of bytes
to trim the packet when the packet is outputting to a port, or when
the packet is sent to userspace.

Signed-off-by: William Tu <u9012063@gmail.com>
Cc: Pravin Shelar <pshelar@nicira.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
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 [Fri, 10 Jun 2016 18:52:24 +0000 (11:52 -0700)]
Merge git://git./linux/kernel/git/davem/net

Conflicts:
net/sched/act_police.c
net/sched/sch_drr.c
net/sched/sch_hfsc.c
net/sched/sch_prio.c
net/sched/sch_red.c
net/sched/sch_tbf.c

In net-next the drop methods of the packet schedulers got removed, so
the bug fixes to them in 'net' are irrelevant.

A packet action unload crash fix conflicts with the addition of the
new firstuse timestamp.

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 [Fri, 10 Jun 2016 15:32:24 +0000 (08:32 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) nfnetlink timestamp taken from wrong skb, fix from Florian Westphal.

 2) Revert some msleep conversions in rtlwifi as these spots are in
    atomic context, from Larry Finger.

 3) Validate that NFTA_SET_TABLE attribute is actually specified when we
    call nf_tables_getset().  From Phil Turnbull.

 4) Don't do mdio_reset in stmmac driver with spinlock held as that can
    sleep, from Vincent Palatin.

 5) sk_filter() does things other than run a BPF filter, so we should
    not elide it's call just because sk->sk_filter is NULL.  Fix from
    Eric Dumazet.

 6) Fix missing backlog updates in several packet schedulers, from Cong
    Wang.

 7) bnx2x driver should allow VLAN add/remove while the interface is
    down, from Michal Schmidt.

 8) Several RDS/TCP race fixes from Sowmini Varadhan.

 9) fq_codel scheduler doesn't return correct queue length in dumps,
    from Eric Dumazet.

10) Fix TCP stats for tail loss probe and early retransmit in ipv6, from
    Yuchung Cheng.

11) Properly initialize udp_tunnel_socket_cfg in l2tp_tunnel_create(),
    from Guillaume Nault.

12) qfq scheduler leaks SKBs if a kzalloc fails, fix from Florian
    Westphal.

13) sock_fprog passed into PACKET_FANOUT_DATA needs compat handling,
    from Willem de Bruijn.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (85 commits)
  vmxnet3: segCnt can be 1 for LRO packets
  packet: compat support for sock_fprog
  stmmac: fix parameter to dwmac4_set_umac_addr()
  net/mlx5e: Fix blue flame quota logic
  net/mlx5e: Use ndo_stop explicitly at shutdown flow
  net/mlx5: E-Switch, always set mc_promisc for allmulti vports
  net/mlx5: E-Switch, Modify node guid on vf set MAC
  net/mlx5: E-Switch, Fix vport enable flow
  net/mlx5: E-Switch, Use the correct error check on returned pointers
  net/mlx5: E-Switch, Use the correct free() function
  net/mlx5: Fix E-Switch flow steering capabilities check
  net/mlx5: Fix flow steering NIC capabilities check
  net/mlx5: Fix root flow table update
  net/mlx5: Fix MLX5_CMD_OP_MAX to be defined correctly
  net/mlx5: Fix masking of reserved bits in XRCD number
  net/mlx5: Fix the size of modify QP mailbox
  mlxsw: spectrum: Don't sleep during ndo_get_phys_port_name()
  mlxsw: spectrum: Make split flow match firmware requirements
  wext: Fix 32 bit iwpriv compatibility issue with 64 bit Kernel
  cfg80211: remove get/set antenna and tx power warnings
  ...

8 years agoMerge tag 'sound-4.7-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Fri, 10 Jun 2016 15:27:30 +0000 (08:27 -0700)]
Merge tag 'sound-4.7-rc3' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "We have only few, mainly HD-audio device-specific fixes.  Realtek
  codec driver got a slightly more LOC, but they are all for the new
  codec chip, and won't affect others at all"

* tag 'sound-4.7-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: hda - Add PCI ID for Kabylake
  ALSA: hda/realtek: Add T560 docking unit fixup
  ALSA: hda - Fix headset mic detection problem for Dell machine
  ALSA: uapi: Add three missing header files to Kbuild file
  ALSA: hda/realtek - Add support for new codecs ALC700/ALC701/ALC703
  ALSA: hda/realtek - ALC256 speaker noise issue

8 years agoMerge tag 'drm-fixes-for-v4.7-rc3' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Fri, 10 Jun 2016 15:21:06 +0000 (08:21 -0700)]
Merge tag 'drm-fixes-for-v4.7-rc3' of git://people.freedesktop.org/~airlied/linux

Pull drm fixes from Dave Airlie:
 "This weeks instalment of fixes:

  amdgpu:
     Lots of memory leak and firmware leak fixes

  nouveau:
     Collection of display fixes, KASAN fixes

  vc4:
     vblank/pageflipping fixes

  fsl-dcu:
     Regmap cache fix

  omap:
     Unused variable warning fix.

  Nothing too surprising so far"

* tag 'drm-fixes-for-v4.7-rc3' of git://people.freedesktop.org/~airlied/linux: (46 commits)
  drm/amdgpu: fix warning with powerplay disabled.
  drm/amd/powerplay: delete useless code as pptable changed in vbios.
  drm/amd/powerplay: fix bug visit array out of bounds
  drm/amdgpu: fix smu ucode memleak (v2)
  drm/amdgpu: add release firmware for cgs
  drm/amdgpu: fix tonga smu_fini mem leak
  drm/amdgpu: fix fiji smu fini mem leak
  drm/amdgpu: fix cik sdma ucode memleak
  drm/amdgpu: fix sdma24 ucode mem leak
  drm/amdgpu: fix sdma3 ucode mem leak
  drm/amdgpu: fix uvd fini mem leak
  drm/amdgpu: fix gfx 7 ucode mem leak
  drm/amdgpu: fix gfx8 ucode mem leak
  drm/amdgpu: fix missing free wb for cond_exec
  drm/amdgpu: fix memleak in pptable_init
  drm/amdgpu: fix mem leak in atombios
  drm/amdgpu: fix mem leak in pplib/hwmgr
  drm/amdgpu: fix mem leak in smumgr
  drm/amdgpu: add pipeline sync while vmid switch in same ctx
  drm/amdgpu: vBIOS post only call when mem_size zero
  ...

8 years agoMerge tag 'acpi-4.7-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Fri, 10 Jun 2016 15:15:37 +0000 (08:15 -0700)]
Merge tag 'acpi-4.7-rc3' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI fix from Rafael Wysocki:
 "A recently introduced boot regression related to the ACPI EC
  initialization is addressed by restoring the previous behavior (Lv
  Zheng)"

* tag 'acpi-4.7-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  ACPI / EC: Fix a boot EC regresion by restoring boot EC support for the DSDT EC

8 years agoMerge tag 'pm-4.7-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Linus Torvalds [Fri, 10 Jun 2016 15:09:12 +0000 (08:09 -0700)]
Merge tag 'pm-4.7-rc3' of git://git./linux/kernel/git/rafael/linux-pm

Pull power management fixes from Rafael Wysocki:
 "Stable-candidate fixes for the intel_pstate driver and the cpuidle
  core.

  Specifics:

   - Fix two intel_pstate initialization issues, one of which was
     introduced during the 4.4 cycle (Srinivas Pandruvada)

   - Fix kernel build with CONFIG_UBSAN set and CONFIG_CPU_IDLE unset
     (Catalin Marinas)"

* tag 'pm-4.7-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpufreq: intel_pstate: Fix ->set_policy() interface for no_turbo
  cpufreq: intel_pstate: Fix code ordering in intel_pstate_set_policy()
  cpuidle: Do not access cpuidle_devices when !CONFIG_CPU_IDLE

8 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Fri, 10 Jun 2016 15:00:47 +0000 (08:00 -0700)]
Merge branch 'akpm' (patches from Andrew)

Merge misc fixes from Andrew Morton:
 "7 fixes"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  mm/fadvise.c: do not discard partial pages with POSIX_FADV_DONTNEED
  mm: introduce dedicated WQ_MEM_RECLAIM workqueue to do lru_add_drain_all
  kernel/relay.c: fix potential memory leak
  mm: thp: broken page count after commit aa88b68c3b1d
  revert "mm: memcontrol: fix possible css ref leak on oom"
  kasan: change memory hot-add error messages to info messages
  mm/hugetlb: fix huge page reserve accounting for private mappings

8 years agovmxnet3: segCnt can be 1 for LRO packets
Shrikrishna Khare [Wed, 8 Jun 2016 14:40:53 +0000 (07:40 -0700)]
vmxnet3: segCnt can be 1 for LRO packets

The device emulation may send segCnt of 1 for LRO packets.

Signed-off-by: Shrikrishna Khare <skhare@vmware.com>
Signed-off-by: Jin Heo <heoj@vmware.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agomlxsw: core: Remove deprecated create_workqueue
Bhaktipriya Shridhar [Tue, 7 Jun 2016 19:59:46 +0000 (01:29 +0530)]
mlxsw: core: Remove deprecated create_workqueue

alloc_workqueue replaces deprecated create_workqueue().

A dedicated workqueue has been used since the workqueue
mlxsw_wq is used for FDB notif. processing with workitems that are
involved in normal device operation && because it's a network device
which can be depended upon during memory reclaim.

Workitems &trans->timeout_dw and &mlxsw_sp->fdb_notify.dw,
map to mlxsw_sp_fdb_notify_work (processes FDB notifications from the
underlying device and resolves the netdev to which the entry points to
and notifies the bridge using the switchdev notifier) and
mlxsw_emad_trans_timeout_work (provides async EMAD register access)
respectively. They require forward progress under memory pressure and
hence, WQ_MEM_RECLAIM has been set.

Since there are only a fixed number of work items, explicit concurrency
limit is unnecessary here.

Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com>
Tested-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: cavium: liquidio: Remove deprecated create_workqueue
Bhaktipriya Shridhar [Tue, 7 Jun 2016 20:17:59 +0000 (01:47 +0530)]
net: cavium: liquidio: Remove deprecated create_workqueue

alloc_workqueue replaces deprecated create_workqueue().

A dedicated workqueue has been used since the workitem viz
(&lio->txq_status_wq.wk.work which maps to octnet_poll_check_txq_status)
is involved in a brief poll routine for checking transmit queue status
and is an intergral part of normal device operation.
WQ_MEM_RECLAIM has been set to guarantee forward progress under memory
pressure, which is a requirement here.
Since there are only a fixed number of work items, explicit concurrency
limit is unnecessary.

flush_workqueue is unnecessary since destroy_workqueue() itself calls
drain_workqueue() which flushes repeatedly till the workqueue
becomes empty.

Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agopacket: compat support for sock_fprog
Willem de Bruijn [Tue, 7 Jun 2016 16:06:34 +0000 (12:06 -0400)]
packet: compat support for sock_fprog

Socket option PACKET_FANOUT_DATA takes a struct sock_fprog as argument
if PACKET_FANOUT has mode PACKET_FANOUT_CBPF. This structure contains
a pointer into user memory. If userland is 32-bit and kernel is 64-bit
the two disagree about the layout of struct sock_fprog.

Add compat setsockopt support to convert a 32-bit compat_sock_fprog to
a 64-bit sock_fprog. This is analogous to compat_sock_fprog support for
SO_REUSEPORT added in commit 1957598840f4 ("soreuseport: add compat
case for setsockopt SO_ATTACH_REUSEPORT_CBPF").

Reported-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Willem de Bruijn <willemb@google.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx4_en: fix ethtool -x
Eric Dumazet [Wed, 8 Jun 2016 04:24:18 +0000 (21:24 -0700)]
net/mlx4_en: fix ethtool -x

mlx4 RSS is limited to spread incoming packets to a power of two number
of queues.

An uniformly distibuted traffic would be split on queues 0 to N-1, N
being a power of two, each queue having a 1/N weight.

If number of RX queues is not a power of two, upper RX queues do not
receive traffic.

ethtool -x is lying, because it pretends some queues have higher weight.

Before patch:

lpaa24:~# ethtool -L eth1 rx 24
lpaa24:~# ethtool -x eth1
RX flow hash indirection table for eth1 with 24 RX ring(s):
    0:      0     1     2     3     4     5     6     7
    8:      8     9    10    11    12    13    14    15
   16:      0     1     2     3     4     5     6     7
RSS hash key:
e0:7c:3a:89:07:55:b6:58:69:cc:f4:e5:24:62:e3:25:88:6c:42:5b:d2:cb:9a:d2:e0:06:e1:dc:f9:09:a1:89:0f:a0:30:43:73:6f:0c:b6

If this information was correct, user space tools could expect queues 0
to 7 to receive twice more traffic than queues 8 to 15

After patch :

lpaa24:~# ethtool -L eth1 rx 24
lpaa24:~# ethtool -x eth1
RX flow hash indirection table for eth1 with 24 RX ring(s):
    0:      0     1     2     3     4     5     6     7
    8:      8     9    10    11    12    13    14    15
RSS hash key:
da:7b:09:60:f1:ac:67:b4:d0:72:d4:ec:a2:e5:80:0a:ad:50:22:1a:f8:f9:66:54:5f:22:45:c3:88:f4:57:82:c1:c1:90:ed:70:cb:40:ce
lpaa24:~# ethtool -X eth1 equal 8
lpaa24:~# ethtool -x eth1
RX flow hash indirection table for eth1 with 24 RX ring(s):
    0:      0     1     2     3     4     5     6     7
    8:      0     1     2     3     4     5     6     7
RSS hash key:
da:7b:09:60:f1:ac:67:b4:d0:72:d4:ec:a2:e5:80:0a:ad:50:22:1a:f8:f9:66:54:5f:22:45:c3:88:f4:57:82:c1:c1:90:ed:70:cb:40:ce

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Maciej ƻenczykowski <maze@google.com>
Cc: Eugenia Emantayev <eugenia@mellanox.com>
Cc: Wei Wang <weiwan@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agostmmac: fix parameter to dwmac4_set_umac_addr()
Ben Dooks [Wed, 8 Jun 2016 18:21:17 +0000 (19:21 +0100)]
stmmac: fix parameter to dwmac4_set_umac_addr()

The dwmac4_set_umac_addr() takes a struct mac_device_info as
the first parameter, but is being passed a ioaddr instead from
dwmac4_set_filter(). Fix the warning/bug by changing the first
parameter.

drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c:159:46: warning: incorrect type in argument 1 (different address spaces)
drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c:159:46:    expected struct mac_device_info *hw
drivers/net/ethernet/stmicro/stmmac/dwmac4_core.c:159:46:    got void [noderef] <asn:2>*ioaddr

Note, only compile tested this as do not have any
hardware with it in.

Signed-off-by: Ben Dooks <ben.dooks@codethink.co.uk>
Acked-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agor8152: replace netdev_alloc_skb_ip_align with napi_alloc_skb
hayeswang [Wed, 8 Jun 2016 06:52:33 +0000 (14:52 +0800)]
r8152: replace netdev_alloc_skb_ip_align with napi_alloc_skb

Replace netdev_alloc_skb_ip_align() with napi_alloc_skb() which can save
several CPU cycles by avoiding having to disable and re-enable IRQs.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'bgmac-stats'
David S. Miller [Fri, 10 Jun 2016 06:37:24 +0000 (23:37 -0700)]
Merge branch 'bgmac-stats'

Florian Fainelli says:

====================
net: bgmac: Misc improvements

This patch series add minor changes to the bgmac driver:

- properly bind net_device with its backing device structure such that
  we can locate the device using common helper functions

- add support for ethtool statistics reading the HW MIB counters which
  is useful for debugging

- add netdev statistics throughout the TX/RX path to know what is going on
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobgmac: Maintain some netdev statistics
Florian Fainelli [Tue, 7 Jun 2016 22:06:15 +0000 (15:06 -0700)]
bgmac: Maintain some netdev statistics

Add a few netdev statistics to report transmitted and received bytes and
packets and a few obvious errors.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobgmac: Add support for ethtool statistics
Florian Fainelli [Tue, 7 Jun 2016 22:06:14 +0000 (15:06 -0700)]
bgmac: Add support for ethtool statistics

Read the statistics from the BGMAC's builtin MAC and return them to
user-space using the standard ethtool helpers.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agobgmac: Bind net_device with backing device structure
Florian Fainelli [Tue, 7 Jun 2016 22:06:13 +0000 (15:06 -0700)]
bgmac: Bind net_device with backing device structure

In preparation for allowing different helpers to be utilized against
network devices created by the bgmac driver, make sure that we bind the
net_device with core->dev.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agovirtio_net: Update the feature bit to comply with spec
Aaron Conole [Thu, 9 Jun 2016 17:41:15 +0000 (13:41 -0400)]
virtio_net: Update the feature bit to comply with spec

A draft version of the MTU Advice feature bit was specified as 25.  This
bit is not within the allowed range for network device feature bits, and
should be changed to be feature bit 3 to fully comply with the spec.

Fixes 14de9d114a82 ('virtio-net: Add initial MTU advice feature')
Signed-off-by: Aaron Conole <aconole@redhat.com>
Suggested-by: "Michael S. Tsirkin" <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: vrf: Fix crash when IPv6 is disabled at boot time
David Ahern [Thu, 9 Jun 2016 17:21:00 +0000 (10:21 -0700)]
net: vrf: Fix crash when IPv6 is disabled at boot time

Frank Kellermann reported a kernel crash with 4.5.0 when IPv6 is
disabled at boot using the kernel option ipv6.disable=1. Using
current net-next with the boot option:

$ ip link add red type vrf table 1001

Generates:
[12210.919584] BUG: unable to handle kernel NULL pointer dereference at 0000000000000748
[12210.921341] IP: [<ffffffff814b30e3>] fib6_get_table+0x2c/0x5a
[12210.922537] PGD b79e3067 PUD bb32b067 PMD 0
[12210.923479] Oops: 0000 [#1] SMP
[12210.924001] Modules linked in: ipvlan 8021q garp mrp stp llc
[12210.925130] CPU: 3 PID: 1177 Comm: ip Not tainted 4.7.0-rc1+ #235
[12210.926168] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.7.5-20140531_083030-gandalf 04/01/2014
[12210.928065] task: ffff8800b9ac4640 ti: ffff8800bacac000 task.ti: ffff8800bacac000
[12210.929328] RIP: 0010:[<ffffffff814b30e3>]  [<ffffffff814b30e3>] fib6_get_table+0x2c/0x5a
[12210.930697] RSP: 0018:ffff8800bacaf888  EFLAGS: 00010202
[12210.931563] RAX: 0000000000000748 RBX: ffffffff81a9e280 RCX: ffff8800b9ac4e28
[12210.932688] RDX: 00000000000000e9 RSI: 0000000000000002 RDI: 0000000000000286
[12210.933820] RBP: ffff8800bacaf898 R08: ffff8800b9ac4df0 R09: 000000000052001b
[12210.934941] R10: 00000000657c0000 R11: 000000000000c649 R12: 00000000000003e9
[12210.936032] R13: 00000000000003e9 R14: ffff8800bace7800 R15: ffff8800bb3ec000
[12210.937103] FS:  00007faa1766c700(0000) GS:ffff88013ac00000(0000) knlGS:0000000000000000
[12210.938321] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[12210.939166] CR2: 0000000000000748 CR3: 00000000b79d6000 CR4: 00000000000406e0
[12210.940278] Stack:
[12210.940603]  ffff8800bb3ec000 ffffffff81a9e280 ffff8800bacaf8c8 ffffffff814b3135
[12210.941818]  ffff8800bb3ec000 ffffffff81a9e280 ffffffff81a9e280 ffff8800bace7800
[12210.943040]  ffff8800bacaf8f0 ffffffff81397c88 ffff8800bb3ec000 ffffffff81a9e280
[12210.944288] Call Trace:
[12210.944688]  [<ffffffff814b3135>] fib6_new_table+0x24/0x8a
[12210.945516]  [<ffffffff81397c88>] vrf_dev_init+0xd4/0x162
[12210.946328]  [<ffffffff814091e1>] register_netdevice+0x100/0x396
[12210.947209]  [<ffffffff8139823d>] vrf_newlink+0x40/0xb3
[12210.948001]  [<ffffffff814187f0>] rtnl_newlink+0x5d3/0x6d5
...

The problem above is due to the fact that the fib hash table is not
allocated when IPv6 is disabled at boot.

As for the VRF driver it should not do any IPv6 initializations if IPv6
is disabled, so it needs to know if IPv6 is disabled at boot. The disable
parameter is private to the IPv6 module, so provide an accessor for
modules to determine if IPv6 was disabled at boot time.

Fixes: 35402e3136634 ("net: Add IPv6 support to VRF device")
Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agorxrpc: Simplify connect() implementation and simplify sendmsg() op
David Howells [Thu, 9 Jun 2016 22:02:51 +0000 (23:02 +0100)]
rxrpc: Simplify connect() implementation and simplify sendmsg() op

Simplify the RxRPC connect() implementation.  It will just note the
destination address it is given, and if a sendmsg() comes along with no
address, this will be assigned as the address.  No transport struct will be
held internally, which will allow us to remove this later.

Simplify sendmsg() also.  Whilst a call is active, userspace refers to it
by a private unique user ID specified in a control message.  When sendmsg()
sees a user ID that doesn't map to an extant call, it creates a new call
for that user ID and attempts to add it.  If, when we try to add it, the
user ID is now registered, we now reject the message with -EEXIST.  We
should never see this situation unless two threads are racing, trying to
create a call with the same ID - which would be an error.

It also isn't required to provide sendmsg() with an address - provided the
control message data holds a user ID that maps to a currently active call.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/netlink/af_netlink.h: Remove unused structure.
Fabien Siron [Tue, 7 Jun 2016 13:02:04 +0000 (13:02 +0000)]
net/netlink/af_netlink.h: Remove unused structure.

Signed-off-by: Fabien Siron <fabien.siron@epita.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx4_en: mlx4_en_netpoll() should schedule TX, not RX
Eric Dumazet [Fri, 3 Jun 2016 18:52:49 +0000 (11:52 -0700)]
net/mlx4_en: mlx4_en_netpoll() should schedule TX, not RX

I am not sure mlx4_en_netpoll() is doing anything useful right now.

mlx4 has different NAPI structures for RX and TX, and netpoll only wants
to drain TX queues.

Lets schedule NAPI polls on TX, not RX.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Maciej ƻenczykowski <maze@google.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Acked-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'BCM53xx-driver'
David S. Miller [Fri, 10 Jun 2016 05:21:30 +0000 (22:21 -0700)]
Merge branch 'BCM53xx-driver'

Florian Fainelli says:

====================
net: dsa: Broadcom BCM53xx switches support

This patch series adds support for the Broadcom BCM53xx series aka RoboSwitches.

This driver is largely based on Jonas Gorski's b53 driver for OpenWrt which can
be found here:

https://dev.openwrt.org/browser/trunk/target/linux/generic/files/drivers/net/phy/b53

a few bug fixes and DSA-ifycation later, here is what we got.

This has been successfully tested in the following configurations:

- Broadcom BCM53011 using the SRAB bus layer with 4 ports LAN, 1 port WAN

- A Broadcom BCM7445 device with an internal Starfighter 2 switch (bcm_sf2.c)
  and a Broadcom BCM53125 hanging off one of its ports connected via MDIO, creating
  two trees hanging off each other, and this works!

- A Broadcom BCM53125 MDIO connected to a Lamobo/Bananapi R1 board using the STMMAC
  MDIO driver

For now, we do not enable Broadcom tags, because there are different
generations of switches being supported which have different tag formats, but
the plan is to enable them later on.

Support for different HW features will be added later: EEE, Compact Field
Processor (TCAM) once this initial cut gets accepted.

Testing and bug reports welcome!
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: b53: Plug in VLAN support
Florian Fainelli [Fri, 10 Jun 2016 01:23:57 +0000 (18:23 -0700)]
net: dsa: b53: Plug in VLAN support

Add support for configuration VLANs on B53 devices by implementing the
port VLAN add/del/dump functions. We currently default to a behavior
which is equivalent to having VLAN filtering turned on, where all VLANs
not programmed into the VLAN port-based vector will be discarded on
ingress.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: b53: Add bridge support
Florian Fainelli [Fri, 10 Jun 2016 01:23:56 +0000 (18:23 -0700)]
net: dsa: b53: Add bridge support

Add support for HW bridging by tying the ports together in the same port
VLAN mask when they belong to the same bridge, and isolating them to be
alone with the CPU port when they are not.

Propagate STP states from the bridge layer to the switch's HW mapping
when requested.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: b53: Implement ARL add/del/dump operations
Florian Fainelli [Fri, 10 Jun 2016 01:23:55 +0000 (18:23 -0700)]
net: dsa: b53: Implement ARL add/del/dump operations

Adds support for FDB add/delete/dump using the ARL read/write logic and
the ARL search logic for faster dumps. The code is made flexible enough
it could support devices with a different register layout like BCM5325
and BCM5365 which have fewer number of entries or pack values into a
single 64 bits register.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: b53: Add BCM7445 quirk
Florian Fainelli [Fri, 10 Jun 2016 01:23:54 +0000 (18:23 -0700)]
net: dsa: b53: Add BCM7445 quirk

The Broadcom BCM7445 STB chip has an issued in its revision D0 which was
previously worked around in drivers/net/dsa/bcm_sf2.c where we may
end-up double programming the integrated BCM7445 switch (bcm_sf2) and an
external Broadcom switch such as BCM53125, since these are mostly
register compatible.

Add a small quirk which just defers probing until we are sitting on the
slave DSA MDIO bus, which will allow us to intercept reads/writes and
funnel them through the SF2 internal MDIO master (which happens to
disconnect its pseudo PHY).

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: b53: Add support for Broadcom RoboSwitch
Florian Fainelli [Fri, 10 Jun 2016 01:23:53 +0000 (18:23 -0700)]
net: dsa: b53: Add support for Broadcom RoboSwitch

This patch adds support for Broadcom's BCM53xx switch family, also known
as RoboSwitch. Some of these switches are ubiquituous, found in home
routers, Wi-Fi routers, DSL and cable modem gateways and other
networking related products.

This drivers adds the library driver (b53_common.c) as well as a few bus
glue drivers for MDIO, SPI, Switch Register Access Block (SRAB) and
memory-mapped I/O into a SoC's address space (Broadcom BCM63xx/33xx).

Basic operations are supported to bring the Layer 1/2 up and running,
but not much more at this point, subsequent patches add the remaining
features.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'bcm_sf2-vlan'
David S. Miller [Fri, 10 Jun 2016 05:12:49 +0000 (22:12 -0700)]
Merge branch 'bcm_sf2-vlan'

Florian Fainelli says:

====================
net: dsa: bcm_sf2: add VLAN support

This is long overdue, finally add support for VLANs in the Broadcom Starfigther
2 switch driver.

There are a few things that make us differ from e.g; mv88e6xxx.c:

- we keep a software cache of which VLANs are enabled and which are not to
  dramatically speed up the VLAN dump operation, we do not have any HW operation
  which would only return the list of valid VLAN entries, they would have to be
  all queried one by one, with 4K vlans, this takes a while

- the default behavior is equivalent to setting VLAN filtering to 1, still working
  on implementing a proper port_vlan_filtering callback, but I figured the most
  conservative behavior is probably okay anyway

- without enabling VLANs, the default behavior is to receive any 802.1q frames
  (per the DSA documentation), however, once we start enabling VLAN support, if
  an interface leaves the bridge, we still want it to receive all 802.1q frames
  so we utiliez the "Join all VLAN" feature of the switch to perform that
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: bcm_sf2: Add VLAN support
Florian Fainelli [Fri, 10 Jun 2016 00:42:08 +0000 (17:42 -0700)]
net: dsa: bcm_sf2: Add VLAN support

Add support for configuring VLANs on the Broadcom Starfigther2 switch.
This is all done through the bridge vlan facility just like other DSA
drivers.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: bcm_sf2: Add VLAN registers definitions
Florian Fainelli [Fri, 10 Jun 2016 00:42:07 +0000 (17:42 -0700)]
net: dsa: bcm_sf2: Add VLAN registers definitions

Add the definitions for the VLAN registers that we are going to
manipulate in subsequent patches.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: bcm_sf2: Move setup function at the far end
Florian Fainelli [Fri, 10 Jun 2016 00:42:06 +0000 (17:42 -0700)]
net: dsa: bcm_sf2: Move setup function at the far end

Re-order the bcm_sf2_sw_setup() function so that it is at the far end of
the driver to avoid any kind of forward declarations.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: dsa: bcm_sf2: Split fast age into a helper function
Florian Fainelli [Fri, 10 Jun 2016 00:42:05 +0000 (17:42 -0700)]
net: dsa: bcm_sf2: Split fast age into a helper function

Add a helper function to fast age something that is controlled by the
caller: port, VLAN. We will use this to implement a VLAN fast age
operation.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'mlx5-fixes'
David S. Miller [Fri, 10 Jun 2016 05:06:27 +0000 (22:06 -0700)]
Merge branch 'mlx5-fixes'

Saeed Mahameed says:

====================
Mellanox 100G mlx5 fixes for 4.7-rc

The following series provides some small fixes for mlx5 driver.

Two small fixes for the mlx5e netdev, the 1st is for the blue flame
quota accounting and the 2nd is a small refactoring in shutdown flow.

Five trivial fixes for mlx5 E-Switch.
- Allmulti mc_promisc flag was not set in a specific flow.
- Modify VF node guid when admin mac is changed.
- Race in vport enable flow.
- Misc code fixes (kvfree when needed and error pointers checking).

Three in mlx5 steering area.  Correct capabilities checking and root flow table update.

Three misc fixes in mlx5 commands enum and layouts.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5e: Fix blue flame quota logic
Eli Cohen [Thu, 9 Jun 2016 21:07:40 +0000 (00:07 +0300)]
net/mlx5e: Fix blue flame quota logic

Blue flame is a latency enhancement feature that allows the driver to
write the packet data directly to the NIC's registers thus making the
read of the packet data from host memory redundant.

We maintain a quota for the blue flame which is reloaded whenever we
identify that the hardware is processing send requests and processes
them fast enough so by the time we post the next send request it was
able to process all the pending ones. This indicates that the hardware
is capable of processing more blue flame requests efficiently. The blue
flame quota is decremented whenever we send using blue flame.

The current code erroneously clears the budget if we did not use blue
flame for the current post send operation and we fix it here.

Fixes: 88a85f99e51f ('net/mlx5e: TX latency optimization to save DMA reads')
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5e: Use ndo_stop explicitly at shutdown flow
Eran Ben Elisha [Thu, 9 Jun 2016 21:07:39 +0000 (00:07 +0300)]
net/mlx5e: Use ndo_stop explicitly at shutdown flow

The current implementation copies the flow of ndo_stop instead of
calling it explicitly, Fixed it.

Fixes: 5fc7197d3a25 ("net/mlx5: Add pci shutdown callback")
Signed-off-by: Eran Ben Elisha <eranbe@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, always set mc_promisc for allmulti vports
Mohamad Haj Yahia [Thu, 9 Jun 2016 21:07:38 +0000 (00:07 +0300)]
net/mlx5: E-Switch, always set mc_promisc for allmulti vports

Set the mc_promisc flag also in the case of adding new mc address to
existing allmulti vport.

Fixes: a35f71f27a61 ('net/mlx5: E-Switch, Implement promiscuous rx modes vf request handling')
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, Modify node guid on vf set MAC
Noa Osherovich [Thu, 9 Jun 2016 21:07:37 +0000 (00:07 +0300)]
net/mlx5: E-Switch, Modify node guid on vf set MAC

In RoCE, the RDMA-CM needs the node guid to establish connection
between nodes.
Today, the node guid exposed to mlx5 Ethernet VFs is zero, therefore
RDMA-CM on the VF is broken.

Whenever the administrator sets a MAC for a VF, derive the node guid
from it and set it as well in the following way:
MAC: e4:1d:2d:b3:f4:01 -> node_guid: e4:1d:2d:ff:fe:b3:f4:01

Fixes: 77256579c6b43 ('net/mlx5: E-Switch, Introduce Vport...')
Signed-off-by: Noa Osherovich <noaos@mellanox.com>
Signed-off-by: Majd Dibbiny <majd@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 vport enable flow
Mohamad Haj Yahia [Thu, 9 Jun 2016 21:07:36 +0000 (00:07 +0300)]
net/mlx5: E-Switch, Fix vport enable flow

Reorder vport enable flow to mark the vport as enabled before calling
the vport change handler which was modified to handle the case for
when vport is not enabled.

This fixes the case for when the PF netdev is open before sriov is
enabled, once sriov is enabled at esw_enable_vport,
esw_vport_change_handle_locked didn't read the PF context since it
thought the PF vport was not enabled.

When we enable the vport, arming for events is not required anymore,
since it's done on the vport change handle

Fixes: 586cfa7f1d58 ('net/mlx5: E-Switch, Use vport event handler for vport cleanup')
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 the correct error check on returned pointers
Or Gerlitz [Thu, 9 Jun 2016 21:07:35 +0000 (00:07 +0300)]
net/mlx5: E-Switch, Use the correct error check on returned pointers

The mlx5 flow-steering API (mlx5_create_flow_table/group/rule) never
returns null pointer on error. Even if it was doing that, checking
for IS_ERR_OR_NULL(p) and then returning PTR_ERR(p) would have cause
bugs, since PTR_ERR(NULL) --> success, crash.

To make things more robust and protect against related future bugs,
convert all IS_ERR_OR_NULL checks on returned values to IS_ERR.

Fixes: 5742df0f7dbe ('net/mlx5: E-Switch, Introduce VST vport ingress/egress ACLs')
Fixes: 86d722ad2c3b ('net/mlx5: Use flow steering infrastructure for mlx5_en')
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reported-by: Ilya Lesokhin <ilyal@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 the correct free() function
Or Gerlitz [Thu, 9 Jun 2016 21:07:34 +0000 (00:07 +0300)]
net/mlx5: E-Switch, Use the correct free() function

We must use kvfree() for something that could have been allocated with vzalloc(),
do that.

Fixes: 5742df0f7dbe ('net/mlx5: E-Switch, Introduce VST vport ingress/egress ACLs')
Fixes: 86d722ad2c3b ('net/mlx5: Use flow steering infrastructure for mlx5_en')
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reported-by: Ilya Lesokhin <ilyal@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: Fix E-Switch flow steering capabilities check
Maor Gottlieb [Thu, 9 Jun 2016 21:07:33 +0000 (00:07 +0300)]
net/mlx5: Fix E-Switch flow steering capabilities check

Add missing capabilities check for E-Switch FDB and ACLs flow
tables before creating their namespace in flow steering.

Fixes: efdc810ba39d ('net/mlx5: Flow steering, Add vport ACL support')
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: Fix flow steering NIC capabilities check
Maor Gottlieb [Thu, 9 Jun 2016 21:07:32 +0000 (00:07 +0300)]
net/mlx5: Fix flow steering NIC capabilities check

Flow steering infrastructure is currently used only on link layer
ethernet, therefore the driver should initialize the flow steering
when the device link layer is ethernet.

In addition, add missing capability check before initializing the
namespace of NIC RX flow tables.

Fixes: 2530236303d9 ('net/mlx5_core: Flow steering tree initialization')
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: Fix root flow table update
Maor Gottlieb [Thu, 9 Jun 2016 21:07:31 +0000 (00:07 +0300)]
net/mlx5: Fix root flow table update

When we destroy the last flow table we need to update
the root_ft to NULL.

It fixes an issue for when the last flow table is destroyed
and recreated again, root_ft pointer will not be updated,
as a result traffic will be dropped.

Fixes: 2cc43b494a6c ('net/mlx5_core: Managing root flow table')
Signed-off-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: Fix MLX5_CMD_OP_MAX to be defined correctly
Shahar Klein [Thu, 9 Jun 2016 21:07:30 +0000 (00:07 +0300)]
net/mlx5: Fix MLX5_CMD_OP_MAX to be defined correctly

Having MLX5_CMD_OP_MAX on another file causes us to repeatedly miss
accounting new commands added to the driver and hence there're no entries
for them in debugfs. To solve that, we integrate it into the commands enum
as the last entry.

Fixes: 34a40e689393 ('net/mlx5_core: Introduce modify flow table command')
Signed-off-by: Shahar Klein <shahark@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: Fix masking of reserved bits in XRCD number
Majd Dibbiny [Thu, 9 Jun 2016 21:07:29 +0000 (00:07 +0300)]
net/mlx5: Fix masking of reserved bits in XRCD number

Mask the reserved bits when reading the number of newly
created XRCD.

Fixes: e126ba97dba9 ('mlx5: Add driver for Mellanox Connect-IB adapters')
Signed-off-by: Majd Dibbiny <majd@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet/mlx5: Fix the size of modify QP mailbox
Majd Dibbiny [Thu, 9 Jun 2016 21:07:28 +0000 (00:07 +0300)]
net/mlx5: Fix the size of modify QP mailbox

Add 16 reserved bytes at the end of mlx5_modify_qp_mbox_in to
match the hardware spec definition.

Fixes: e126ba97dba9 ('mlx5: Add driver for Mellanox Connect-IB adapters')
Signed-off-by: Majd Dibbiny <majd@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agoMerge branch 'fixes-for-v4.7-rc3' of http://git.agner.ch/git/linux-drm-fsl-dcu into...
Dave Airlie [Fri, 10 Jun 2016 02:17:46 +0000 (12:17 +1000)]
Merge branch 'fixes-for-v4.7-rc3' of git.agner.ch/git/linux-drm-fsl-dcu into drm-fixes

* 'fixes-for-v4.7-rc3' of http://git.agner.ch/git/linux-drm-fsl-dcu:
  drm/fsl-dcu: use flat regmap cache

8 years agodrm/amdgpu: fix warning with powerplay disabled.
Dave Airlie [Fri, 10 Jun 2016 01:40:49 +0000 (11:40 +1000)]
drm/amdgpu: fix warning with powerplay disabled.

This just fixes a warning when you disable powerplay.

Signed-off-by: Dave Airlie <airlied@redhat.com>
8 years agoMerge branch 'drm-fixes-4.7' of git://people.freedesktop.org/~agd5f/linux into drm...
Dave Airlie [Thu, 9 Jun 2016 23:46:59 +0000 (09:46 +1000)]
Merge branch 'drm-fixes-4.7' of git://people.freedesktop.org/~agd5f/linux into drm-fixes

Mostly memory leak and firmware leak fixes for amdgpu.  A bit bigger than
usual since this is several weeks worth of fixes.

* 'drm-fixes-4.7' of git://people.freedesktop.org/~agd5f/linux: (28 commits)
  drm/amd/powerplay: delete useless code as pptable changed in vbios.
  drm/amd/powerplay: fix bug visit array out of bounds
  drm/amdgpu: fix smu ucode memleak (v2)
  drm/amdgpu: add release firmware for cgs
  drm/amdgpu: fix tonga smu_fini mem leak
  drm/amdgpu: fix fiji smu fini mem leak
  drm/amdgpu: fix cik sdma ucode memleak
  drm/amdgpu: fix sdma24 ucode mem leak
  drm/amdgpu: fix sdma3 ucode mem leak
  drm/amdgpu: fix uvd fini mem leak
  drm/amdgpu: fix gfx 7 ucode mem leak
  drm/amdgpu: fix gfx8 ucode mem leak
  drm/amdgpu: fix missing free wb for cond_exec
  drm/amdgpu: fix memleak in pptable_init
  drm/amdgpu: fix mem leak in atombios
  drm/amdgpu: fix mem leak in pplib/hwmgr
  drm/amdgpu: fix mem leak in smumgr
  drm/amdgpu: add pipeline sync while vmid switch in same ctx
  drm/amdgpu: vBIOS post only call when mem_size zero
  drm/amdgpu: modify sdma start sequence
  ...

8 years agoMerge branch 'msm-fixes-4.7-rc3' of git://people.freedesktop.org/~robclark/linux...
Dave Airlie [Thu, 9 Jun 2016 23:45:42 +0000 (09:45 +1000)]
Merge branch 'msm-fixes-4.7-rc3' of git://people.freedesktop.org/~robclark/linux into drm-fixes

* 'msm-fixes-4.7-rc3' of git://people.freedesktop.org/~robclark/linux:
  drm/msm: fix potential submit error path issue
  drm/msm: fix some crashes in submit fail path
  drm/msm: deal with exhausted vmap space better

8 years agoMerge branches 'pm-cpufreq-fixes' and 'pm-cpuidle'
Rafael J. Wysocki [Thu, 9 Jun 2016 21:49:16 +0000 (23:49 +0200)]
Merge branches 'pm-cpufreq-fixes' and 'pm-cpuidle'

* pm-cpufreq-fixes:
  cpufreq: intel_pstate: Fix ->set_policy() interface for no_turbo
  cpufreq: intel_pstate: Fix code ordering in intel_pstate_set_policy()

* pm-cpuidle:
  cpuidle: Do not access cpuidle_devices when !CONFIG_CPU_IDLE

8 years agoMerge branch 'acpi-ec'
Rafael J. Wysocki [Thu, 9 Jun 2016 21:48:54 +0000 (23:48 +0200)]
Merge branch 'acpi-ec'

* acpi-ec:
  ACPI / EC: Fix a boot EC regresion by restoring boot EC support for the DSDT EC

8 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma
Linus Torvalds [Thu, 9 Jun 2016 21:36:12 +0000 (14:36 -0700)]
Merge tag 'for-linus' of git://git./linux/kernel/git/dledford/rdma

Pull rdma fixes from Doug Ledford:
 "This is the first -rc pull for the RDMA subsystem.  The patch count is
  high, but they are all smallish patches fixing simple things for the
  most part, and the overall line count of changes here is smaller than
  the patch count would lead a person to believe.

  Code is up and running in my labs, including direct testing of cxgb4,
  mlx4, mlx5, ocrdma, and qib.

  Summary:

   - Multiple minor fixes to the rdma core
   - Multiple minor fixes to hfi1
   - Multiple minor fixes to mlx5
   - A very few other minor fixes (SRP, IPoIB, usNIC, mlx4)"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma: (35 commits)
  IB/IPoIB: Don't update neigh validity for unresolved entries
  IB/mlx5: Fix alternate path code
  IB/mlx5: Fix pkey_index length in the QP path record
  IB/mlx5: Fix entries check in mlx5_ib_resize_cq
  IB/mlx5: Fix entries checks in mlx5_ib_create_cq
  IB/mlx5: Check BlueFlame HCA support
  IB/mlx5: Fix returned values of query QP
  IB/mlx5: Limit query HCA clock
  IB/mlx5: Fix FW version diaplay in sysfs
  IB/mlx5: Return PORT_ERR in Active to Initializing tranisition
  IB/mlx5: Set flow steering capability bit
  IB/core: Make all casts in ib_device_cap_flags enum consistent
  IB/core: Fix bit curruption in ib_device_cap_flags structure
  IB/core: Initialize sysfs attributes before sysfs create group
  IB/IPoIB: Disable bottom half when dealing with device address
  IB/core: Fix removal of default GID cache entry
  IB/IPoIB: Fix race between ipoib_remove_one to sysfs functions
  IB/core: Fix query port failure in RoCE
  IB/core: fix error unwind in sysfs hw counters code
  IB/core: Fix array length allocation
  ...

8 years agoMerge tag 'arc-4.7-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/vgupta/arc
Linus Torvalds [Thu, 9 Jun 2016 21:28:39 +0000 (14:28 -0700)]
Merge tag 'arc-4.7-rc3' of git://git./linux/kernel/git/vgupta/arc

Pull ARC fixes from Vineet Gupta:

 - Revert of ll-sc backoff retry workaround in atomics/spinlocks as
   hardware is now proven to work just fine

 - Typo fixes (Thanks Andrea Gelmini)

 - Removal of obsolete DT property (Alexey)

 - Other minor fixes

* tag 'arc-4.7-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/vgupta/arc:
  Revert "ARCv2: spinlock/rwlock/atomics: Delayed retry of failed SCOND with exponential backoff"
  Revert "ARCv2: spinlock/rwlock: Reset retry delay when starting a new spin-wait cycle"
  Revert "ARCv2: spinlock/rwlock/atomics: reduce 1 instruction in exponential backoff"
  ARC: don't enable DISCONTIGMEM unconditionally
  ARC: [intc-compact] simplify code for 2 priority levels
  arc: Get rid of root core-frequency property
  Fix typos

8 years agomm/fadvise.c: do not discard partial pages with POSIX_FADV_DONTNEED
Oleg Drokin [Wed, 8 Jun 2016 22:33:59 +0000 (15:33 -0700)]
mm/fadvise.c: do not discard partial pages with POSIX_FADV_DONTNEED

I noticed that the logic in the fadvise64_64 syscall is incorrect for
partial pages.  While first page of the region is correctly skipped if
it is partial, the last page of the region is mistakenly discarded.
This leads to problems for applications that read data in
non-page-aligned chunks discarding already processed data between the
reads.

A somewhat misguided application that does something like write(XX bytes
(non-page-alligned)); drop the data it just wrote; repeat gets a
significant penalty in performance as a result.

Link: http://lkml.kernel.org/r/1464917140-1506698-1-git-send-email-green@linuxhacker.ru
Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm: introduce dedicated WQ_MEM_RECLAIM workqueue to do lru_add_drain_all
Wang Sheng-Hui [Wed, 8 Jun 2016 22:33:56 +0000 (15:33 -0700)]
mm: introduce dedicated WQ_MEM_RECLAIM workqueue to do lru_add_drain_all

This patch is based on https://patchwork.ozlabs.org/patch/574623/.

Tejun submitted commit 23d11a58a9a6 ("workqueue: skip flush dependency
checks for legacy workqueues") for the legacy create*_workqueue()
interface.

But some workq created by alloc_workqueue still reports warning on
memory reclaim, e.g nvme_workq with flag WQ_MEM_RECLAIM set:

    workqueue: WQ_MEM_RECLAIM nvme:nvme_reset_work is flushing !WQ_MEM_RECLAIM events:lru_add_drain_per_cpu
    ------------[ cut here ]------------
    WARNING: CPU: 0 PID: 6 at SoC/linux/kernel/workqueue.c:2448 check_flush_dependency+0xb4/0x10c
    ...
    check_flush_dependency+0xb4/0x10c
    flush_work+0x54/0x140
    lru_add_drain_all+0x138/0x188
    migrate_prep+0xc/0x18
    alloc_contig_range+0xf4/0x350
    cma_alloc+0xec/0x1e4
    dma_alloc_from_contiguous+0x38/0x40
    __dma_alloc+0x74/0x25c
    nvme_alloc_queue+0xcc/0x36c
    nvme_reset_work+0x5c4/0xda8
    process_one_work+0x128/0x2ec
    worker_thread+0x58/0x434
    kthread+0xd4/0xe8
    ret_from_fork+0x10/0x50

That's because lru_add_drain_all() will schedule the drain work on
system_wq, whose flag is set to 0, !WQ_MEM_RECLAIM.

Introduce a dedicated WQ_MEM_RECLAIM workqueue to do
lru_add_drain_all(), aiding in getting memory freed.

Link: http://lkml.kernel.org/r/1464917521-9775-1-git-send-email-shhuiw@foxmail.com
Signed-off-by: Wang Sheng-Hui <shhuiw@foxmail.com>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: Keith Busch <keith.busch@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thierry Reding <treding@nvidia.com>
Cc: Ingo Molnar <mingo@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agokernel/relay.c: fix potential memory leak
Zhouyi Zhou [Wed, 8 Jun 2016 22:33:53 +0000 (15:33 -0700)]
kernel/relay.c: fix potential memory leak

When relay_open_buf() fails in relay_open(), code will goto free_bufs,
but chan is nowhere freed.

Link: http://lkml.kernel.org/r/1464777927-19675-1-git-send-email-yizhouzhou@ict.ac.cn
Signed-off-by: Zhouyi Zhou <zhouzhouyi@gmail.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm: thp: broken page count after commit aa88b68c3b1d
Gerald Schaefer [Wed, 8 Jun 2016 22:33:50 +0000 (15:33 -0700)]
mm: thp: broken page count after commit aa88b68c3b1d

Christian Borntraeger reported a kernel panic after corrupt page counts,
and it turned out to be a regression introduced with commit aa88b68c3b1d
("thp: keep huge zero page pinned until tlb flush"), at least on s390.

put_huge_zero_page() was moved over from zap_huge_pmd() to
release_pages(), and it was replaced by tlb_remove_page().  However,
release_pages() might not always be triggered by (the arch-specific)
tlb_remove_page().

On s390 we call free_page_and_swap_cache() from tlb_remove_page(), and
not tlb_flush_mmu() -> free_pages_and_swap_cache() like the generic
version, because we don't use the MMU-gather logic.  Although both
functions have very similar names, they are doing very unsimilar things,
in particular free_page_xxx is just doing a put_page(), while
free_pages_xxx calls release_pages().

This of course results in very harmful put_page()s on the huge zero
page, on architectures where tlb_remove_page() is implemented in this
way.  It seems to affect only s390 and sh, but sh doesn't have THP
support, so the problem (currently) probably only exists on s390.

The following quick hack fixed the issue:

Link: http://lkml.kernel.org/r/20160602172141.75c006a9@thinkpad
Signed-off-by: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Reported-by: Christian Borntraeger <borntraeger@de.ibm.com>
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: <stable@vger.kernel.org> [4.6.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agorevert "mm: memcontrol: fix possible css ref leak on oom"
Andrew Morton [Wed, 8 Jun 2016 22:33:47 +0000 (15:33 -0700)]
revert "mm: memcontrol: fix possible css ref leak on oom"

Revert commit 1383399d7be0 ("mm: memcontrol: fix possible css ref leak
on oom").  Johannes points out "There is a task_in_memcg_oom() check
before calling mem_cgroup_oom()".

Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov@virtuozzo.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agokasan: change memory hot-add error messages to info messages
Shuah Khan [Wed, 8 Jun 2016 22:33:45 +0000 (15:33 -0700)]
kasan: change memory hot-add error messages to info messages

Change the following memory hot-add error messages to info messages.
There is no need for these to be errors.

   kasan: WARNING: KASAN doesn't support memory hot-add
   kasan: Memory hot-add will be disabled

Link: http://lkml.kernel.org/r/1464794430-5486-1-git-send-email-shuahkh@osg.samsung.com
Signed-off-by: Shuah Khan <shuahkh@osg.samsung.com>
Acked-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agomm/hugetlb: fix huge page reserve accounting for private mappings
Mike Kravetz [Wed, 8 Jun 2016 22:33:42 +0000 (15:33 -0700)]
mm/hugetlb: fix huge page reserve accounting for private mappings

When creating a private mapping of a hugetlbfs file, it is possible to
unmap pages via ftruncate or fallocate hole punch.  If subsequent faults
repopulate these mappings, the reserve counts will go negative.  This is
because the code currently assumes all faults to private mappings will
consume reserves.  The problem can be recreated as follows:

 - mmap(MAP_PRIVATE) a file in hugetlbfs filesystem
 - write fault in pages in the mapping
 - fallocate(FALLOC_FL_PUNCH_HOLE) some pages in the mapping
 - write fault in pages in the hole

This will result in negative huge page reserve counts and negative
subpool usage counts for the hugetlbfs.  Note that this can also be
recreated with ftruncate, but fallocate is more straight forward.

This patch modifies the routines vma_needs_reserves and vma_has_reserves
to examine the reserve map associated with private mappings similar to
that for shared mappings.  However, the reserve map semantics for
private and shared mappings are very different.  This results in subtly
different code that is explained in the comments.

Link: http://lkml.kernel.org/r/1464720957-15698-1-git-send-email-mike.kravetz@oracle.com
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Acked-by: Hillf Danton <hillf.zj@alibaba-inc.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Kirill Shutemov <kirill.shutemov@linux.intel.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Aneesh Kumar <aneesh.kumar@linux.vnet.ibm.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
8 years agoMerge branch 'netdev_lockdep_set_classes'
David S. Miller [Thu, 9 Jun 2016 20:28:38 +0000 (13:28 -0700)]
Merge branch 'netdev_lockdep_set_classes'

Eric Dumazet says:

====================
net: better lockdep annotations

Introduction of qdisc->running seqcount added lockdep false positives.

While chasing the bug, it came to me that we had a lot of copies of the
same stuff in virtual drivers.

This patch series has the qdisc->running fix (considers that a trylock
is attempted in lockdep terminology), and adds a generic helper so
that we no longer have to patch many virtual drivers when a new per-device
or per-qdisc lock is added.

Thanks to David Ahern for reporting the issue and testing my patches :)
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
8 years agonet: ipvlan: call netdev_lockdep_set_classes()
Eric Dumazet [Thu, 9 Jun 2016 14:45:15 +0000 (07:45 -0700)]
net: ipvlan: call netdev_lockdep_set_classes()

In case a qdisc is used on a ipvlan device, we need to use different
lockdep classes to avoid false positives.

Use the new netdev_lockdep_set_classes() generic helper.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>