openwrt/staging/blogic.git
7 years agoqed: Correct endian order of MAC passed to MFW
Mintz, Yuval [Thu, 23 Mar 2017 13:50:17 +0000 (15:50 +0200)]
qed: Correct endian order of MAC passed to MFW

The management firmware is running on a Big Endian processor,
and when running on LE platform HW is configured to swap access
to memory shared between management firmware and driver on
32-bit granulariy.

As a result, for matters of simplicity most of the APIs between
driver and management firmware are based on 32-bit variables.
MAC settings are one exception, as driver needs to fill a byte
array when indicating to management firmware that primary MAC
has changed.
Due to the swap, driver must make sure that the mac that was
provided in byte-order would be translated into native order,
otherwise after the swap the management firmware would read
it swapped.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Pass src/dst sizes when interacting with MFW
Tomer Tayar [Thu, 23 Mar 2017 13:50:16 +0000 (15:50 +0200)]
qed: Pass src/dst sizes when interacting with MFW

The driver interaction with management firmware involves a union
of all the data-members relating to the commands the driver prepares.

Current interface assumes the caller always passes such a union -
but thats cumbersome as well as risky [chancing a stack corruption
in case caller accidentally passes a smaller member instead of union].

Change implementation so that caller could pass a pointer to any
of the members instead of the union.

Signed-off-by: Tomer Tayar <Tomer.Tayar@cavium.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Revise MFW command locking
Tomer Tayar [Thu, 23 Mar 2017 13:50:15 +0000 (15:50 +0200)]
qed: Revise MFW command locking

Interaction of driver -> management firmware is based
on a one-pending mailbox [per interface], and various
mailbox commands need to be synchronized.

Current scheme is messy, and there's a difficulty extending
it as it deals differently with various commands as well as
making assumption on the required behavior for load/unload
requests.

Drop the current scheme into a completion-list-based approach;
Each flow would try sending the command when possible,
allowing one flow to complete another flow's completion and
relieve the mailbox before sending its own command.

Signed-off-by: Tomer Tayar <Tomer.Tayar@cavium.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'netvsc-bug-fixes-and-cleanups'
David S. Miller [Thu, 23 Mar 2017 02:38:57 +0000 (19:38 -0700)]
Merge branch 'netvsc-bug-fixes-and-cleanups'

Stephen Hemminger says:

====================
netvsc: bug fixes and cleanups

These fix NAPI issues and bugs found during testing of shutdown
testing.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: fix and cleanup rndis_filter_set_packet_filter
stephen hemminger [Wed, 22 Mar 2017 21:51:05 +0000 (14:51 -0700)]
netvsc: fix and cleanup rndis_filter_set_packet_filter

Fix warning from unused set_complete variable. And rearrange code
to eliminate unnecessary goto's.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: eliminate unnecessary skb == NULL checks
stephen hemminger [Wed, 22 Mar 2017 21:51:04 +0000 (14:51 -0700)]
netvsc: eliminate unnecessary skb == NULL checks

Since there already is a special case goto for control messages (skb == NULL)
in netvsc_send, there is no need for later checks in same code path.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: remove unnecessary lock on shutdown
stephen hemminger [Wed, 22 Mar 2017 21:51:03 +0000 (14:51 -0700)]
netvsc: remove unnecessary lock on shutdown

The channel inbound lock was not being used at all by the netvsc
device, but the spin_lock was helpful by providing necessary
barrier before waiting.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: use refcount_t for keeping track of sub channels
stephen hemminger [Wed, 22 Mar 2017 21:51:02 +0000 (14:51 -0700)]
netvsc: use refcount_t for keeping track of sub channels

Rather than a lock and variable, use a refcount_t to keep track
of the number of sub channels.  Don't need to wait for subchannels
on device removal since wait was already done in device_add.

Also fix the error handling; don't wait forever in case of
an error on request to create sub channels.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: uses RCU instead of removal flag
stephen hemminger [Wed, 22 Mar 2017 21:51:01 +0000 (14:51 -0700)]
netvsc: uses RCU instead of removal flag

It is cleaner to use RCU protected pointer (nvdev_ctx->nvdev)
to indicate device is in removed state, rather than having a separate
boolean flag. By using the pointer the context can be checked
by static checkers and dynamic lockdep.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: use RCU to protect inner device structure
stephen hemminger [Wed, 22 Mar 2017 21:51:00 +0000 (14:51 -0700)]
netvsc: use RCU to protect inner device structure

The netvsc driver has an internal structure (netvsc_device) which
is created when device is opened and released when device is closed.
And also opened/released when MTU or number of channels change.

Since this is referenced in the receive and transmit path, it is
safer to use RCU to protect/prevent use after free problems.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: change max channel calculation
stephen hemminger [Wed, 22 Mar 2017 21:50:59 +0000 (14:50 -0700)]
netvsc: change max channel calculation

The default number of maximum channels should be limited to the
number of cpus available on the numa node of the primary channel.
This also makes sure maximum channels <= num_online_cpus

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: handle offline mtu and channel change
stephen hemminger [Wed, 22 Mar 2017 21:50:58 +0000 (14:50 -0700)]
netvsc: handle offline mtu and channel change

If device is not up, then changing MTU (or number of channels)
should not re-enable the device.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: fix NAPI performance regression
stephen hemminger [Wed, 22 Mar 2017 21:50:57 +0000 (14:50 -0700)]
netvsc: fix NAPI performance regression

When using NAPI, the single stream performance declined signifcantly
because the poll routine was updating host after every burst
of packets. This excess signalling caused host throttling.

This fix restores the old behavior. Host is only signalled
after the ring has been emptied.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoliquidio: fix tx completions in napi poll
VSR Burru [Wed, 22 Mar 2017 18:54:50 +0000 (11:54 -0700)]
liquidio: fix tx completions in napi poll

If there are no egress packets pending, then don't look for tx completions
in napi poll.  Also, fix broken tx queue wakeup logic.

Signed-off-by: VSR Burru <veerasenareddy.burru@cavium.com>
Signed-off-by: Felix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: Satanand Burla <satananda.burla@cavium.com>
Signed-off-by: Derek Chickles <derek.chickles@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoliquidio: allocate RX buffers in OOM conditions in PF and VF
Satanand Burla [Wed, 22 Mar 2017 18:31:13 +0000 (11:31 -0700)]
liquidio: allocate RX buffers in OOM conditions in PF and VF

Add workqueue that is periodically run to try to allocate RX buffers in OOM
conditions in PF and VF.

Signed-off-by: Satanand Burla <satananda.burla@cavium.com>
Signed-off-by: Felix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: vmxnet3: use new api ethtool_{get|set}_link_ksettings
Philippe Reynes [Wed, 22 Mar 2017 07:27:57 +0000 (08:27 +0100)]
net: vmxnet3: use new api ethtool_{get|set}_link_ksettings

The ethtool api {get|set}_settings is deprecated.
We move this 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>
7 years agonet: virtio_net: use new api ethtool_{get|set}_link_ksettings
Philippe Reynes [Tue, 21 Mar 2017 22:24:24 +0000 (23:24 +0100)]
net: virtio_net: use new api ethtool_{get|set}_link_ksettings

The ethtool api {get|set}_settings is deprecated.
We move this 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>
7 years agoMerge branch 'xgene-v2-mdio-and-ethtool'
David S. Miller [Thu, 23 Mar 2017 02:17:15 +0000 (19:17 -0700)]
Merge branch 'xgene-v2-mdio-and-ethtool'

Iyappan Subramanian says:

====================
drivers: net: xgene-v2: Add MDIO and ethtool support

This patch set,

- adds phy management and ethtool support
- fixes ethernet reset
- addresses review comments from previous patch set

v2: Address review comments from v1
- removed mdio_lock, since there is a top level lock in mdio_bus.c

v1:
- Initial version
====================

Signed-off-by: Iyappan Subramanian <isubramanian@apm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodrivers: net: xgene-v2: misc fixes
Iyappan Subramanian [Wed, 22 Mar 2017 01:18:05 +0000 (18:18 -0700)]
drivers: net: xgene-v2: misc fixes

Fixed review comments from the previous patch-set.

- changed return value check of platform_get_irq() to < 0
- replaced devm_request(free)_irq() calls by request(free)_irq() since
  they are called from open() and close()
- changed sizeof(struct mystruct) to sizeof(*mystruct)
- reduced indentation on tx_timeout()

Signed-off-by: Iyappan Subramanian <isubramanian@apm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodrivers: net: xgene-v2: Fix port reset
Iyappan Subramanian [Wed, 22 Mar 2017 01:18:04 +0000 (18:18 -0700)]
drivers: net: xgene-v2: Fix port reset

Fixed port reset sequence by adding ECC init.

Signed-off-by: Iyappan Subramanian <isubramanian@apm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodrivers: net: xgene-v2: Add ethtool support
Iyappan Subramanian [Wed, 22 Mar 2017 01:18:03 +0000 (18:18 -0700)]
drivers: net: xgene-v2: Add ethtool support

Added basic ethtool support.

Signed-off-by: Iyappan Subramanian <isubramanian@apm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodrivers: net: xgene-v2: Add MDIO support
Iyappan Subramanian [Wed, 22 Mar 2017 01:18:02 +0000 (18:18 -0700)]
drivers: net: xgene-v2: Add MDIO support

Added phy management support by using phy abstraction layer APIs.

Signed-off-by: Iyappan Subramanian <isubramanian@apm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'bpf-map-in-map'
David S. Miller [Wed, 22 Mar 2017 22:45:46 +0000 (15:45 -0700)]
Merge branch 'bpf-map-in-map'

Martin KaFai Lau says:

====================
bpf: Add map-in-map support

This patchset adds map-in-map support (map->map).
One use case is the (vips -> webservers) in the L4 load balancer so
that different vips can be backed by different set of webservers.

Please refer to the individual commit log for details.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: Add tests for map-in-map
Martin KaFai Lau [Wed, 22 Mar 2017 17:00:35 +0000 (10:00 -0700)]
bpf: Add tests for map-in-map

Test cases for array of maps and hash of maps.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: Add hash of maps support
Martin KaFai Lau [Wed, 22 Mar 2017 17:00:34 +0000 (10:00 -0700)]
bpf: Add hash of maps support

This patch adds hash of maps support (hashmap->bpf_map).
BPF_MAP_TYPE_HASH_OF_MAPS is added.

A map-in-map contains a pointer to another map and lets call
this pointer 'inner_map_ptr'.

Notes on deleting inner_map_ptr from a hash map:

1. For BPF_F_NO_PREALLOC map-in-map, when deleting
   an inner_map_ptr, the htab_elem itself will go through
   a rcu grace period and the inner_map_ptr resides
   in the htab_elem.

2. For pre-allocated htab_elem (!BPF_F_NO_PREALLOC),
   when deleting an inner_map_ptr, the htab_elem may
   get reused immediately.  This situation is similar
   to the existing prealloc-ated use cases.

   However, the bpf_map_fd_put_ptr() calls bpf_map_put() which calls
   inner_map->ops->map_free(inner_map) which will go
   through a rcu grace period (i.e. all bpf_map's map_free
   currently goes through a rcu grace period).  Hence,
   the inner_map_ptr is still safe for the rcu reader side.

This patch also includes BPF_MAP_TYPE_HASH_OF_MAPS to the
check_map_prealloc() in the verifier.  preallocation is a
must for BPF_PROG_TYPE_PERF_EVENT.  Hence, even we don't expect
heavy updates to map-in-map, enforcing BPF_F_NO_PREALLOC for map-in-map
is impossible without disallowing BPF_PROG_TYPE_PERF_EVENT from using
map-in-map first.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: Add array of maps support
Martin KaFai Lau [Wed, 22 Mar 2017 17:00:33 +0000 (10:00 -0700)]
bpf: Add array of maps support

This patch adds a few helper funcs to enable map-in-map
support (i.e. outer_map->inner_map).  The first outer_map type
BPF_MAP_TYPE_ARRAY_OF_MAPS is also added in this patch.
The next patch will introduce a hash of maps type.

Any bpf map type can be acted as an inner_map.  The exception
is BPF_MAP_TYPE_PROG_ARRAY because the extra level of
indirection makes it harder to verify the owner_prog_type
and owner_jited.

Multi-level map-in-map is not supported (i.e. map->map is ok
but not map->map->map).

When adding an inner_map to an outer_map, it currently checks the
map_type, key_size, value_size, map_flags, max_entries and ops.
The verifier also uses those map's properties to do static analysis.
map_flags is needed because we need to ensure BPF_PROG_TYPE_PERF_EVENT
is using a preallocated hashtab for the inner_hash also.  ops and
max_entries are needed to generate inlined map-lookup instructions.
For simplicity reason, a simple '==' test is used for both map_flags
and max_entries.  The equality of ops is implied by the equality of
map_type.

During outer_map creation time, an inner_map_fd is needed to create an
outer_map.  However, the inner_map_fd's life time does not depend on the
outer_map.  The inner_map_fd is merely used to initialize
the inner_map_meta of the outer_map.

Also, for the outer_map:

* It allows element update and delete from syscall
* It allows element lookup from bpf_prog

The above is similar to the current fd_array pattern.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: Fix and simplifications on inline map lookup
Martin KaFai Lau [Wed, 22 Mar 2017 17:00:32 +0000 (10:00 -0700)]
bpf: Fix and simplifications on inline map lookup

Fix in verifier:
For the same bpf_map_lookup_elem() instruction (i.e. "call 1"),
a broken case is "a different type of map could be used for the
same lookup instruction". For example, an array in one case and a
hashmap in another.  We have to resort to the old dynamic call behavior
in this case.  The fix is to check for collision on insn_aux->map_ptr.
If there is collision, don't inline the map lookup.

Please see the "do_reg_lookup()" in test_map_in_map_kern.c in the later
patch for how-to trigger the above case.

Simplifications on array_map_gen_lookup():
1. Calculate elem_size from map->value_size.  It removes the
   need for 'struct bpf_array' which makes the later map-in-map
   implementation easier.
2. Remove the 'elem_size == 1' test

Fixes: 81ed18ab3098 ("bpf: add helper inlining infra and optimize map_array lookup")
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: stmmac: fix dma operation mode config for older versions
Joao Pinto [Wed, 22 Mar 2017 11:56:05 +0000 (11:56 +0000)]
net: stmmac: fix dma operation mode config for older versions

The dma operation mode configuration routine was wrongly moved to a
function (stmmac_mtl_configuration) that is only executed if the
core version is >= 4.00.

Fixes: 6deee2221e11 ("net: stmmac: prepare dma op mode config for multiple queues")
Reported-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Reviewed-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Joao Pinto <jpinto@synopsys.com>
Tested-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: ipv6: Add sysctl for minimum prefix len acceptable in RIOs.
Joel Scherpelz [Wed, 22 Mar 2017 09:19:04 +0000 (18:19 +0900)]
net: ipv6: Add sysctl for minimum prefix len acceptable in RIOs.

This commit adds a new sysctl accept_ra_rt_info_min_plen that
defines the minimum acceptable prefix length of Route Information
Options. The new sysctl is intended to be used together with
accept_ra_rt_info_max_plen to configure a range of acceptable
prefix lengths. It is useful to prevent misconfigurations from
unintentionally blackholing too much of the IPv6 address space
(e.g., home routers announcing RIOs for fc00::/7, which is
incorrect).

Signed-off-by: Joel Scherpelz <jscherpelz@google.com>
Acked-by: Lorenzo Colitti <lorenzo@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'nfp-concurrency'
David S. Miller [Wed, 22 Mar 2017 19:59:10 +0000 (12:59 -0700)]
Merge branch 'nfp-concurrency'

Jakub Kicinski says:

====================
nfp: allow concurrency in core and minor fixes

The first 10 patches of this series prepare nfpcore for concurrent
accesses.  This will be needed by upcoming hwmon and devlink patches.
Most locking is already in place, the patches in this series iron out
a few bugs.

Last 5 patches are fixes and cleanups to the netdev code, including
removal of doorbell pointers used only on old versions of the chip,
removal of unnecessarily defensive code and flushing xmit_more more
carefully on error paths.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: disable FW on reconfiguration errors
Jakub Kicinski [Wed, 22 Mar 2017 00:59:21 +0000 (17:59 -0700)]
nfp: disable FW on reconfiguration errors

Since we no longer need to keep the FW enabled for .ndo_close()
to work we can always stop FW after reconfiguration failure.
This seems to make most FWs more resilient to faults (at least
in error injection scenarios).

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: remove defensive checks around ndo_open()/ndo_close()
Jakub Kicinski [Wed, 22 Mar 2017 00:59:20 +0000 (17:59 -0700)]
nfp: remove defensive checks around ndo_open()/ndo_close()

Device open and close handlers check if the device is already
in the desired state.  Thanks to our reconfig infrastructure
this should not be necessary, there doesn't seem to be any
code in the driver which depends on it.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: flush xmit_more on error paths
Jakub Kicinski [Wed, 22 Mar 2017 00:59:19 +0000 (17:59 -0700)]
nfp: flush xmit_more on error paths

In case of ring full or DMA mapping error remember to flush xmit_more
delayed kicks.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: remove RX queue pointers
Jakub Kicinski [Wed, 22 Mar 2017 00:59:18 +0000 (17:59 -0700)]
nfp: remove RX queue pointers

NFP6000 doesn't use queue pointers/doorbells for RX, it uses
'done' bit in descriptors.  Remove the pointers from data structures.
Since we are saving space in rx_ring structure make fields we
previously compressed to 16bits word size again.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: don't use netdev_warn() before netdev is registered
Jakub Kicinski [Wed, 22 Mar 2017 00:59:17 +0000 (17:59 -0700)]
nfp: don't use netdev_warn() before netdev is registered

Fix warning which was using netdev_warn() instead of dev_warn()
to early.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: fix nfp_cpp_read()/nfp_cpp_write() error paths
Jakub Kicinski [Wed, 22 Mar 2017 00:59:16 +0000 (17:59 -0700)]
nfp: fix nfp_cpp_read()/nfp_cpp_write() error paths

When acquiring an area fails we can't call function doing both
release and free.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: fix invalid area detection
Jakub Kicinski [Wed, 22 Mar 2017 00:59:15 +0000 (17:59 -0700)]
nfp: fix invalid area detection

Core should detect when someone is trying to request an access
window which is too large for a given type of access.  Otherwise
the requester will be put on a wait queue for ever without any
error message.

Add const qualifiers to clarify that we are only looking at read-
-only members in relevant functions.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: don't ignore return value of wait_event_interruptible
Jakub Kicinski [Wed, 22 Mar 2017 00:59:14 +0000 (17:59 -0700)]
nfp: don't ignore return value of wait_event_interruptible

When signal interrupts waiting for an area to become available
we assume success.  Pay attention to the return code.  Unpack
the code a little bit to make it more readable.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: correct return codes when msleep gets interrupted
Jakub Kicinski [Wed, 22 Mar 2017 00:59:13 +0000 (17:59 -0700)]
nfp: correct return codes when msleep gets interrupted

msleep_interruptible() returns time left to wait, not error
code.  Return ERESTARTSYS when interrupted.

While at it correct a comment and make the polling a bit
more aggressive.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: lock area cache earlier
Jakub Kicinski [Wed, 22 Mar 2017 00:59:12 +0000 (17:59 -0700)]
nfp: lock area cache earlier

We shouldn't access area_cache_list without its lock even
to check if it's empty.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: document expected locking in the core
Jakub Kicinski [Wed, 22 Mar 2017 00:59:11 +0000 (17:59 -0700)]
nfp: document expected locking in the core

Document which fields of nfp_cpp are protected by which locks.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: move mutex code out of nfp_cppcore.c
Jakub Kicinski [Wed, 22 Mar 2017 00:59:10 +0000 (17:59 -0700)]
nfp: move mutex code out of nfp_cppcore.c

After mutex cache removal we can put the mutex code in a separate
source file.  This makes it clear it doesn't play with internals
of struct nfp_cpp any more.

No functional changes.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: remove cpp mutex cache
Jakub Kicinski [Wed, 22 Mar 2017 00:59:09 +0000 (17:59 -0700)]
nfp: remove cpp mutex cache

CPP mutex cache was introduced to work around the fact that the
same host could successfully acquire a lock multiple times.  It
used to collapse multiple users to the same struct nfp_cpp_mutex
and track use count.  Unfortunately it's racy.  Since we now force
all nfp_mutex_lock() callers within the host to actually succeed
at acquiring the lock we no longer need the cache, let's remove it.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: fail graciously when someone tries to grab global lock
Jakub Kicinski [Wed, 22 Mar 2017 00:59:08 +0000 (17:59 -0700)]
nfp: fail graciously when someone tries to grab global lock

The global device lock is acquired to search the resource table.
The lock is actually itself part of the table (entry 0).
Therefore if someone asks for resource 0 we would deadlock since
double locking is no longer allowed.

Currently the driver doesn't try to lock that resource so let's
simply make sure we fail graciously and not add special handling
of this case until really need.  Hide the relevant defines in
the source file.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: disallow sharing mutexes on the same machine
Jakub Kicinski [Wed, 22 Mar 2017 00:59:07 +0000 (17:59 -0700)]
nfp: disallow sharing mutexes on the same machine

NFP can be connected to multiple machines via PCI or other buses.
Access to hardware resources is arbitrated using locks residing
in device memory.  Currently nfpcore only respects the mutexes
when it comes to inter-host locking, but if we try to acquire
the same lock again, on one host - it will simply return success
because owner of the lock is already set to that host.

This makes the locks useless for arbitration within one host
and unfair because whichever host grabbed the lock will have
a chance to reacquire it without others getting a shot.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dwc-xlgmac: fix an error code in xlgmac_alloc_pages()
Dan Carpenter [Tue, 21 Mar 2017 20:42:27 +0000 (23:42 +0300)]
net: dwc-xlgmac: fix an error code in xlgmac_alloc_pages()

The dma_mapping_error() returns true if there is an error but we want
to return -ENOMEM and not 1.

Fixes: 65e0ace2c5cd ("net: dwc-xlgmac: Initial driver for DesignWare Enterprise Ethernet")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Jie Deng <jiedeng@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agortnetlink: Add dump all for netconf
David Ahern [Tue, 21 Mar 2017 19:22:26 +0000 (12:22 -0700)]
rtnetlink: Add dump all for netconf

Use the rtnl_dump_all to dump all netconf handlers that have been
registered. Allows userspace to send a dump request for PF_UNSPEC
and get all families.

Cc: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'phy-mmd-cleanup'
David S. Miller [Wed, 22 Mar 2017 19:43:01 +0000 (12:43 -0700)]
Merge branch 'phy-mmd-cleanup'

Russell King says:

====================
Clean up PHY MMD accessors

This series cleans up phylib's MMD accessors, so that we have a common
way of accessing the Clause 45 register set.

The current situation is far from ideal - we have phy_(read|write)_mmd()
which accesses Clause 45 registers over Clause 45 accesses, and we have
phy_(read|write)_mmd_indirect(), which accesses Clause 45 registers via
Clause 22 register 13/14.

Generic code uses the indirect methods to access standard Clause 45
features, and when we come to add Clause 45 PHY support to phylib, we
would need to make these conditional upon the PHY type, or duplicate
these functions.

An alternative solution is to merge these accessors together, and select
the appropriate access method depending upon the 802.3 clause that the
PHY conforms with.  The result is that we have a single set of
phy_(read|write)_mmd() accessors.

For cases which require special handling, we still allow PHY drivers to
override all MMD accesses - except rather than just overriding the
indirect accesses.  This keeps existing overrides working.

Combining the two also has another beneficial side effect - we get rid
of similar functions that take arguments in different orders.  The
old direct accessors took the phy structure, devad and register number,
whereas the indirect accessors took the phy structure, register number
and devad in that order.  Care must be taken when updating future
drivers that the argument order is correct, and the function name is
not merely replaced.

This patch set is against net-next.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: phy: clean up mmd_phy_indirect()
Russell King [Tue, 21 Mar 2017 16:37:08 +0000 (16:37 +0000)]
net: phy: clean up mmd_phy_indirect()

Make mmd_phy_indirect() use the same terminology as the rest of the
code, making clear what each address is - phy address, devad, and
register number.

While here, remove the "inline" from this static function, leaving
it to the compiler to decide whether to inline this function, and
get rid of unnecessary parens.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: phy: remove the indirect MMD read/write methods
Russell King [Tue, 21 Mar 2017 16:37:03 +0000 (16:37 +0000)]
net: phy: remove the indirect MMD read/write methods

Remove the indirect MMD read/write methods which are now no longer
necessary.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: phy: convert micrel to new read_mmd/write_mmd driver methods
Russell King [Tue, 21 Mar 2017 16:36:58 +0000 (16:36 +0000)]
net: phy: convert micrel to new read_mmd/write_mmd driver methods

Convert micrel to the new read_mmd/write_mmd driver methods.  This
Clause 22 PHY does not support any MMD access method.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: phy: switch remaining users to phy_(read|write)_mmd()
Russell King [Tue, 21 Mar 2017 16:36:53 +0000 (16:36 +0000)]
net: phy: switch remaining users to phy_(read|write)_mmd()

Switch everyone over to using phy_read_mmd() and phy_write_mmd() now
that they are able to handle both Clause 22 indirect addressing and
Clause 45 direct addressing methods to the MMD registers.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: lan78xx: update for phy_(read|write)_mmd_indirect() removal
Russell King [Tue, 21 Mar 2017 16:36:48 +0000 (16:36 +0000)]
net: lan78xx: update for phy_(read|write)_mmd_indirect() removal

lan78xx appears to use phylib in a rather weird way, accessing the PHY
partly through phylib, and partly by making direct accesses to it,
including to the Clause 45 registers.  As the indirect MMD accessors are
going away, update this driver to use the plain phy_(read|write)_mmd()
accessors instead.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Acked-by: Woojung Huh <Woojung.Huh@microchip.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: phy: make phy_(read|write)_mmd() generic MMD accessors
Russell King [Tue, 21 Mar 2017 16:36:43 +0000 (16:36 +0000)]
net: phy: make phy_(read|write)_mmd() generic MMD accessors

Make phy_(read|write)_mmd() generic 802.3 clause 45 register accessors
for both Clause 22 and Clause 45 PHYs, using either the direct register
reading for Clause 45, or the indirect method for Clause 22 PHYs.
Allow this behaviour to be overriden by PHY drivers where necessary.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: phy: move phy MMD accessors to phy-core.c
Russell King [Tue, 21 Mar 2017 16:36:37 +0000 (16:36 +0000)]
net: phy: move phy MMD accessors to phy-core.c

Move the phy_(read|write)__mmd() helpers out of line, they will become
our main MMD accessor functions, and so will be a little more complex.
This complexity doesn't belong in an inline function.  Also move the
_indirect variants as well to keep like functionality together.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: stmmac: Use AVB mode by default
Thierry Reding [Tue, 21 Mar 2017 15:12:11 +0000 (16:12 +0100)]
net: stmmac: Use AVB mode by default

Prior to the recent multi-queue changes the driver would configure the
queues to use the AVB mode, but the mode then got switched to DCB. The
hardware still works fine in DCB mode, but my testing capabilities are
limited, so it's safer to revert to the prior setting anyway.

Signed-off-by: Thierry Reding <treding@nvidia.com>
Acked-By: Joao Pinto <jpinto@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: stmmac: Restore DT backwards-compatibility
Thierry Reding [Tue, 21 Mar 2017 15:12:10 +0000 (16:12 +0100)]
net: stmmac: Restore DT backwards-compatibility

Recent changes to support multiple queues in the device tree bindings
resulted in the number of RX and TX queues to be initialized to zero for
device trees not adhering to the new bindings.

Restore backwards-compatibility with those device trees by falling back
to a single RX and TX queues each.

Signed-off-by: Thierry Reding <treding@nvidia.com>
Acked-By: Joao Pinto <jpinto@synopsys.com>
Tested-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: stmmac: Always enable MAC RX queues
Thierry Reding [Tue, 21 Mar 2017 15:12:09 +0000 (16:12 +0100)]
net: stmmac: Always enable MAC RX queues

The MAC RX queues always need to be enabled in order to receive network
packets. Remove the condition that this only needs to be done for multi-
queue configurations.

Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: convert sk_filter.refcnt from atomic_t to refcount_t
Reshetova, Elena [Tue, 21 Mar 2017 11:59:19 +0000 (13:59 +0200)]
net: convert sk_filter.refcnt from atomic_t to refcount_t

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: greth: Utilize of_get_mac_address()
Tobias Klauser [Tue, 21 Mar 2017 10:24:38 +0000 (11:24 +0100)]
net: greth: Utilize of_get_mac_address()

Do not open code getting the MAC address exclusively from the
"local-mac-address" property, but instead use of_get_mac_address() which
looks up the MAC address using the 3 typical property names.

Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoliquidio: fix Coverity scan errors
Felix Manlunas [Tue, 21 Mar 2017 02:04:48 +0000 (19:04 -0700)]
liquidio: fix Coverity scan errors

Fix Coverity scan errors by not dereferencing lio->glists_dma_base pointer
if it's NULL.

See http://marc.info/?l=linux-netdev&m=149002294305614&w=2

Reported-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Felix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: VSR Burru <veerasenareddy.burru@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: tcp: Permit user set TCP_MAXSEG to default value
Gao Feng [Tue, 21 Mar 2017 01:28:03 +0000 (09:28 +0800)]
net: tcp: Permit user set TCP_MAXSEG to default value

When user_mss is zero, it means use the default value. But the current
codes don't permit user set TCP_MAXSEG to the default value.
It would return the -EINVAL when val is zero.

Signed-off-by: Gao Feng <fgao@ikuai8.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'ovs-sample-action-optimization'
David S. Miller [Wed, 22 Mar 2017 18:28:35 +0000 (11:28 -0700)]
Merge branch 'ovs-sample-action-optimization'

Andy Zhou says:

====================
net-next sample action optimization v4

The sample action can be used for translating Openflow 'clone' action.
However its implementation has not been sufficiently optimized for this
use case. This series attempts to close the gap.

Patch 3 commit message has more details on the specific optimizations
implemented.

---
v3->v4: Enhance patch 4.
        Fix two bugs pointed out by Pravin,
        Remove 'is_sample' variable.

v2->v3: Enhance patch 4, Rafctor to move more common logic to clone_execute().

v1->v2: Address Pravin's comment, Refactor recirc and sample
        to share more common code
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoOpenvswitch: Refactor sample and recirc actions implementation
andy zhou [Mon, 20 Mar 2017 23:32:30 +0000 (16:32 -0700)]
Openvswitch: Refactor sample and recirc actions implementation

Added clone_execute() that both the sample and the recirc
action implementation can use.

Signed-off-by: Andy Zhou <azhou@ovn.org>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoopenvswitch: Optimize sample action for the clone use cases
andy zhou [Mon, 20 Mar 2017 23:32:29 +0000 (16:32 -0700)]
openvswitch: Optimize sample action for the clone use cases

With the introduction of open flow 'clone' action, the OVS user space
can now translate the 'clone' action into kernel datapath 'sample'
action, with 100% probability, to ensure that the clone semantics,
which is that the packet seen by the clone action is the same as the
packet seen by the action after clone, is faithfully carried out
in the datapath.

While the sample action in the datpath has the matching semantics,
its implementation is only optimized for its original use.
Specifically, there are two limitation: First, there is a 3 level of
nesting restriction, enforced at the flow downloading time. This
limit turns out to be too restrictive for the 'clone' use case.
Second, the implementation avoid recursive call only if the sample
action list has a single userspace action.

The main optimization implemented in this series removes the static
nesting limit check, instead, implement the run time recursion limit
check, and recursion avoidance similar to that of the 'recirc' action.
This optimization solve both #1 and #2 issues above.

One related optimization attempts to avoid copying flow key as
long as the actions enclosed does not change the flow key. The
detection is performed only once at the flow downloading time.

Another related optimization is to rewrite the action list
at flow downloading time in order to save the fast path from parsing
the sample action list in its original form repeatedly.

Signed-off-by: Andy Zhou <azhou@ovn.org>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoopenvswitch: Refactor recirc key allocation.
andy zhou [Mon, 20 Mar 2017 23:32:28 +0000 (16:32 -0700)]
openvswitch: Refactor recirc key allocation.

The logic of allocating and copy key for each 'exec_actions_level'
was specific to execute_recirc(). However, future patches will reuse
as well.  Refactor the logic into its own function clone_key().

Signed-off-by: Andy Zhou <azhou@ovn.org>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoopenvswitch: Deferred fifo API change.
andy zhou [Mon, 20 Mar 2017 23:32:27 +0000 (16:32 -0700)]
openvswitch: Deferred fifo API change.

add_deferred_actions() API currently requires actions to be passed in
as a fully encoded netlink message. So far both 'sample' and 'recirc'
actions happens to carry actions as fully encoded netlink messages.
However, this requirement is more restrictive than necessary, future
patch will need to pass in action lists that are not fully encoded
by themselves.

Signed-off-by: Andy Zhou <azhou@ovn.org>
Acked-by: Joe Stringer <joe@ovn.org>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'vrf-perf'
David S. Miller [Wed, 22 Mar 2017 18:19:48 +0000 (11:19 -0700)]
Merge branch 'vrf-perf'

David Ahern says:

====================
net: vrf: performance improvements

Device based features for VRF such as qdisc, netfilter and packet
captures are implemented by switching the dst on skbuffs to its per-VRF
dst. This has the effect of controlling the output function which points
a function in the VRF driver. [1] The skb proceeds down the stack with
dst->dev pointing to the VRF device. Netfilter, qdisc and tc rules and
network taps are evaluated based on this device. Finally, the skb makes
it to the vrf_xmit function which resets the dst based on a FIB lookup.

The feature comes at cost - between 5 and 10% depending on test (TCP vs
UDP, stream vs RR and IPv4 vs IPv6). The main cost is requiring a FIB
lookup in the VRF driver for each packet sent through it. The FIB lookup
is required because the real dst gets dropped so that the skb can
traverse the stack with dst->dev set to the VRF device.

All of that is really driven by the qdisc and not replicating the
processing of __dev_queue_xmit if a qdisc is set up on the device. But,
VRF devices by default do not have a qdisc and really have no need for
multiple Tx queues. This means the performance overhead is inflicted upon
all users for the potential use case of a qdisc being configured.

The overhead can be avoided by checking if the default configuration
applies to a specific VRF device before switching the dst. If a device
does not have a qdisc, the pass through netfilter hooks and packet taps
can be done inline without dropping the dst and thus avoiding the
performance penalty. With this change performance overhead of VRF drops
to neglible (difference with run-over-run variance) to 3% depending on
test type.

netperf performance comparison for 3 cases:
1. L3_MASTER_DEVICE compiled out
2. VRF with this patch set
3. current VRF code

IPv4
----
           no-l3mdev     new-vrf     old-vrf
TCP_RR       28778        28938*       27169
TCP_CRR      10706        10490         9770
UDP_RR       30750        29813        29256

* Although higher in the final run used for submitting this patch set, I
  think what this really represents is a neglible performance overhead for
  VRF with this change (i.e, within the +-1% variance of runs). Most
  notably the FIB lookups in the Tx path are avoided for TCP_RR.

IPv6
----
           no-l3mdev     new-vrf     old-vrf
TCP_RR       29495        29432       27794
TCP_CRR      10520        10338        9870
UDP_RR       26137        27019*      26511

* UDP is consistently better with VRF for two reasons:
  1. Source address selection with L3 domains is considering fewer
     addresses since only addresses on interfaces in the domain are
     considered for the selection. Specifically, perf-top shows
     shows ipv6_get_saddr_eval, ipv6_dev_get_saddr and __ipv6_dev_get_saddr
     running much lower with vrf than without.

  2. The VRF table contains all routes (i.e, there are no separate local
     and main tables per VRF). That means ip6_pol_route_output only has 1
     lookup for VRF where it does 2 without it (1 in the local table and 1
     in the main table).

[1] http://netdevconf.org/1.2/papers/ahern-what-is-l3mdev-paper.pdf
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: vrf: performance improvements for IPv6
David Ahern [Mon, 20 Mar 2017 18:19:45 +0000 (11:19 -0700)]
net: vrf: performance improvements for IPv6

The VRF driver allows users to implement device based features for an
entire domain. For example, a qdisc or netfilter rules can be attached
to a VRF device or tcpdump can be used to view packets for all devices
in the L3 domain.

The device-based features come with a performance penalty, most
notably in the Tx path. The VRF driver uses the l3mdev_l3_out hook
to switch the dst on an skb to its private dst. This allows the skb
to traverse the xmit stack with the device set to the VRF device
which in turn enables the netfilter and qdisc features. The VRF
driver then performs the FIB lookup again and reinserts the packet.

This patch avoids the redirect for IPv6 packets if a qdisc has not
been attached to a VRF device which is the default config. In this
case the netfilter hooks and network taps are directly traversed in
the l3mdev_l3_out handler. If a qdisc is attached to a VRF device,
then the redirect using the vrf dst is done.

Additional overhead is removed by only checking packet taps if a
socket is open on the device (vrf_dev->ptype_all list is not empty).
Packet sockets bound to any device will still get a copy of the
packet via the real ingress or egress interface.

The end result of this change is a decrease in the overhead of VRF
for the default, baseline case (ie., no netfilter rules, no packet
sockets, no qdisc) from a +3% improvement for UDP which has a lookup
per packet (VRF being better than no l3mdev) to ~2% loss for TCP_CRR
which connects a socket for each request-response.

Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: vrf: performance improvements for IPv4
David Ahern [Mon, 20 Mar 2017 18:19:44 +0000 (11:19 -0700)]
net: vrf: performance improvements for IPv4

The VRF driver allows users to implement device based features for an
entire domain. For example, a qdisc or netfilter rules can be attached
to a VRF device or tcpdump can be used to view packets for all devices
in the L3 domain.

The device-based features come with a performance penalty, most
notably in the Tx path. The VRF driver uses the l3mdev_l3_out hook
to switch the dst on an skb to its private dst. This allows the skb
to traverse the xmit stack with the device set to the VRF device
which in turn enables the netfilter and qdisc features. The VRF
driver then performs the FIB lookup again and reinserts the packet.

This patch avoids the redirect for IPv4 packets if a qdisc has not
been attached to a VRF device which is the default config. In this
case the netfilter hooks and network taps are directly traversed in
the l3mdev_l3_out handler. If a qdisc is attached to a VRF device,
then the redirect using the vrf dst is done.

Additional overhead is removed by only checking packet taps if a
socket is open on the device (vrf_dev->ptype_all list is not empty).
Packet sockets bound to any device will still get a copy of the
packet via the real ingress or egress interface.

The end result of this change is a decrease in the overhead of VRF
for the default, baseline case (ie., no netfilter rules, no packet
sockets, no qdisc) to ~3% for UDP which has a lookup per packet and
< 1% overhead for connected sockets that leverage early demux and
avoid FIB lookups.

Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agosock: introduce SO_MEMINFO getsockopt
Josh Hunt [Mon, 20 Mar 2017 19:22:03 +0000 (15:22 -0400)]
sock: introduce SO_MEMINFO getsockopt

Allows reading of SK_MEMINFO_VARS via socket option. This way an
application can get all meminfo related information in single socket
option call instead of multiple calls.

Adds helper function, sk_get_meminfo(), and uses that for both
getsockopt and sock_diag_put_meminfo().

Suggested by Eric Dumazet.

Signed-off-by: Josh Hunt <johunt@akamai.com>
Reviewed-by: Jason Baron <jbaron@akamai.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum: fix swapped order of arguments packets and bytes
Colin Ian King [Mon, 20 Mar 2017 11:37:22 +0000 (11:37 +0000)]
mlxsw: spectrum: fix swapped order of arguments packets and bytes

The arguments packets and bytes to call mlxsw_sp_acl_rule_get_stats are
in the wrong order. Fix this by swapping them.

Detected by CoverityScan, CID#1419705 ("Arguments in wrong order")

Fixes: 7c1b8eb175b69add8ea ("mlxsw: spectrum: Add support for TC flower offload statistics")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agocxgb4: Update IngPad and IngPack values
Arjun Vynipadath [Mon, 20 Mar 2017 08:52:38 +0000 (14:22 +0530)]
cxgb4: Update IngPad and IngPack values

We are using the smallest padding boundary (8 bytes), which isn't
smaller than the Memory Controller Read/Write Size

We get best performance in 100G when the Packing Boundary is a multiple
of the Maximum Payload Size. Its related to inefficient chopping of DMA
packets by PCIe, that causes more overhead on bus. So driver is helping
by making the starting address alignment to be MPS size.

We will try to determine PCIE MaxPayloadSize capabiltiy  and set
IngPackBoundary based on this value. If cache line size is greater than
MPS or determinig MPS fails, we will use cache line size to determine
IngPackBoundary(as before).

Signed-off-by: Arjun Vynipadath <arjun@chelsio.com>
Signed-off-by: Casey Leedom <leedom@chelsio.com>
Signed-off-by: Ganesh Goudar <ganeshgr@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dwc-xlgmac: add module license
Arnd Bergmann [Mon, 20 Mar 2017 08:51:13 +0000 (09:51 +0100)]
net: dwc-xlgmac: add module license

When building the driver as a module, we get a warning about the
lack of a license:

WARNING: modpost: missing MODULE_LICENSE() in drivers/net/ethernet/synopsys/dwc-xlgmac.o
see include/linux/module.h for more information

Curiously the text in the .c files only mentions GPLv2+, while the license
tag in the PCI driver contains both GPL and BSD. I picked the license text
as the more definite reference here and put a GPL tag in there.

Fixes: 65e0ace2c5cd ("net: dwc-xlgmac: Initial driver for DesignWare Enterprise Ethernet")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dwc-xlgmac: include dcbnl.h
Arnd Bergmann [Mon, 20 Mar 2017 08:51:12 +0000 (09:51 +0100)]
net: dwc-xlgmac: include dcbnl.h

Without this header, we can run into a build error:

drivers/net/ethernet/synopsys/dwc-xlgmac-hw.c: In function 'xlgmac_config_queue_mapping':
drivers/net/ethernet/synopsys/dwc-xlgmac-hw.c:1548:36: error: 'IEEE_8021QAZ_MAX_TCS' undeclared (first use in this function)
  prio_queues = min_t(unsigned int, IEEE_8021QAZ_MAX_TCS,

Fixes: 65e0ace2c5cd ("net: dwc-xlgmac: Initial driver for DesignWare Enterprise Ethernet")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Jie Deng <jiedeng@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoneighbour: fix nlmsg_pid in notifications
Roopa Prabhu [Mon, 20 Mar 2017 05:01:28 +0000 (22:01 -0700)]
neighbour: fix nlmsg_pid in notifications

neigh notifications today carry pid 0 for nlmsg_pid
in all cases. This patch fixes it to carry calling process
pid when available. Applications (eg. quagga) rely on
nlmsg_pid to ignore notifications generated by their own
netlink operations. This patch follows the routing subsystem
which already sets this correctly.

Reported-by: Vivek Venkatraman <vivek@cumulusnetworks.com>
Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch '1GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next...
David S. Miller [Wed, 22 Mar 2017 17:31:27 +0000 (10:31 -0700)]
Merge branch '1GbE' of git://git./linux/kernel/git/jkirsher/next-queue

Jeff Kirsher says:

====================
Intel Wired LAN Driver Updates 2017-03-21

This series contains updates to e1000, e1000e, igb, igbvf and ixgb.

This finishes up the work Philippe Reynes did to update the Intel drivers
to the new API for ethtool (get|set)_link_ksettings.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'qed-IOV-cleanups'
David S. Miller [Wed, 22 Mar 2017 02:02:38 +0000 (19:02 -0700)]
Merge branch 'qed-IOV-cleanups'

Yuval Mintz says:

====================
qed: IOV related clenaups

This patch series targets IOV functionality [on both PF and VF].

Patches #2, #3 and #5 fix flows relating to malicious VFs, either by
upgrading and aligning current safe-guards or by correcing racy flows.

Patches #1 and #8 make some malicious/dysnfunctional VFs logging appear
by default in logs.

The rest of the patches either cleanup the existing code or else correct
some possible [yet fairly insignicant] issues in VF behavior.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Always publish VF link from leading hwfn
Mintz, Yuval [Sun, 19 Mar 2017 11:08:20 +0000 (13:08 +0200)]
qed: Always publish VF link from leading hwfn

The link information exists only on the leading hwfn,
but some of its derivatives [e.g., min/max rate] need to
be configured for each hwfn.
When re-basing the VF link view, use the leading hwfn
information as basis for all existing hwfns to allow
said configurations to stick.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Raise verbosity of Malicious VF indications
Mintz, Yuval [Sun, 19 Mar 2017 11:08:19 +0000 (13:08 +0200)]
qed: Raise verbosity of Malicious VF indications

Malicious VF existance should be interesting enough for the
hyperuser. Change the PF indication that one of its child VF
became malicious to appear by default.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Make qed_iov_mark_vf_flr() return bool
Mintz, Yuval [Sun, 19 Mar 2017 11:08:18 +0000 (13:08 +0200)]
qed: Make qed_iov_mark_vf_flr() return bool

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Deprecate VF multiple queue-stop
Mintz, Yuval [Sun, 19 Mar 2017 11:08:17 +0000 (13:08 +0200)]
qed: Deprecate VF multiple queue-stop

The PF<->VF interface allows for the VF to request
multiple queues closure via a single message, but this has
never been used by any official driver.

We now deprecate this option, forcing each queue close
to arrive via a different command; This would be required
for future TLVs that are going to extend the queue TLVs with
additional information on a per-queue basis.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Uniform IOV queue validation
Mintz, Yuval [Sun, 19 Mar 2017 11:08:16 +0000 (13:08 +0200)]
qed: Uniform IOV queue validation

PF needs to validate the status of VF queues before asking firmware
to configure anything for them, but that validation is done in various
different forms - sometimes inadequate.

Add auxillary functions that can be used for testing of the queue
state and convert the various flows to use those instead of current
existing flows; Also, add missing validations where needed.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Correct default VF coalescing configuration
Mintz, Yuval [Sun, 19 Mar 2017 11:08:15 +0000 (13:08 +0200)]
qed: Correct default VF coalescing configuration

When starting the VF's vport, the PF would first configure
the status blocks of the VF and then reset them.
That would cause some of the configured information to be lost -
specifically it would mean that all the VFs queues would use
the Rx coalescing state-machine of the status block.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Set HW-channel to ready before ACKing VF
Mintz, Yuval [Sun, 19 Mar 2017 11:08:14 +0000 (13:08 +0200)]
qed: Set HW-channel to ready before ACKing VF

When PF responds to the VF requests it also cleans the HW-channel
indication in firmware to allow further VF messages to arrive,
but the order currently applied is wrong -
The PF is copying by DMAE the response the VF is polling on for
completion, and only afterwards sets the HW-channel to ready state.

This creates a race condition where the VF would be able to send
an additional message to the PF before the channel would get ready
again, causing the firmware to consider the VF as malicious.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Clean VF malicious indication when disabling IOV
Mintz, Yuval [Sun, 19 Mar 2017 11:08:13 +0000 (13:08 +0200)]
qed: Clean VF malicious indication when disabling IOV

When a VF is considered malicious, driver handling of the VF
FLR flow would clean said indication - but not if the FLR is
part of an sriov-disable flow.
That leads to further issues, as PF wouldn't re-enable the
previously malicious VF when sriov is re-enabled.

No reason for that - simply clean malicious indications in
the sriov-disable flow as well.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Increase verbosity of VF -> PF errors
Mintz, Yuval [Sun, 19 Mar 2017 11:08:12 +0000 (13:08 +0200)]
qed: Increase verbosity of VF -> PF errors

VFs are currently logging errors when communicating
with their PFs in a too-low verbosity that wouldn't
be shown by default. As timeouts and failed commands
are crucial for VF operability, make them appear by
default.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agorhashtable: Add rhashtable_lookup_get_insert_fast
Andreas Gruenbacher [Fri, 17 Mar 2017 23:36:15 +0000 (00:36 +0100)]
rhashtable: Add rhashtable_lookup_get_insert_fast

Add rhashtable_lookup_get_insert_fast for fixed keys, similar to
rhashtable_lookup_get_insert_key for explicit keys.

Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoliquidio: fix for vf mac addr command sent to nic firmware
Rick Farrington [Fri, 17 Mar 2017 22:43:26 +0000 (15:43 -0700)]
liquidio: fix for vf mac addr command sent to nic firmware

Change to support host<->firmware command return value.
Fix for vf mac addr state command.
1. Added support for firmware commands to return a value:
   - previously, the returned code overlapped with host codes, thus
     commands were only returning 0 (success) or -1 (interpreted as
     timeout)
   - per 'response_manager.h', the error codes are split into two fields
     (major/minor) now, firmware commands are grouped into their own
     'major' group, separate from the host's 'major' group, which allow f/w
     commands to return any 16-bit value
2. The command to set vf mac addr was logging a success message even if
   command failed.  Now command uses a callback function to log the status
   message.
3. The command to set vf mac addr was not logging a message when set via
   the host 'ip' command.  Now, the callback function will log an
   appropriate message.

Signed-off-by: Rick Farrington <ricardo.farrington@cavium.com>
Signed-off-by: Felix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: Derek Chickles <derek.chickles@cavium.com>
Signed-off-by: Satanand Burla <satananda.burla@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: usb: pegasus: use new api ethtool_{get|set}_link_ksettings
Philippe Reynes [Fri, 17 Mar 2017 22:34:04 +0000 (23:34 +0100)]
net: usb: pegasus: use new api ethtool_{get|set}_link_ksettings

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

As I don't have the hardware, I'd be very pleased if
someone may test this patch.

Signed-off-by: Philippe Reynes <tremyfr@gmail.com>
Acked-by: Petko Manolov <petkan@nucleusys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'ibmvnic-init'
David S. Miller [Wed, 22 Mar 2017 00:46:14 +0000 (17:46 -0700)]
Merge branch 'ibmvnic-init'

John Allen says:

====================
ibmvnic: Initialization fixes and improvements

These patches resolve issues with the ibmvnic initialization process.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoibmvnic: Correct ibmvnic handling of device open/close
John Allen [Fri, 17 Mar 2017 22:13:43 +0000 (17:13 -0500)]
ibmvnic: Correct ibmvnic handling of device open/close

When closing the ibmvnic device we need to release the resources used
in communicating to the virtual I/O server. These need to be
re-negotiated with the server at open time.

This patch moves the releasing of resources a separate routine
and updates the open and close handlers to release all resources at
close and re-negotiate and allocate these resources at open.

Signed-off-by: Nathan Fontenot <nfont@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoibmvnic: Move ibmvnic adapter intialization to its own routine
John Allen [Fri, 17 Mar 2017 22:13:42 +0000 (17:13 -0500)]
ibmvnic: Move ibmvnic adapter intialization to its own routine

The intialization of the ibmvnic driver with respect to the virtual
server it connects to should be moved to its own routine. This will
alolow the driver to initiate this process from places outside of
the drivers probe routine.

Signed-off-by: Nathan Fontenot <nfont@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoibmvnic: Move login to its own routine
John Allen [Fri, 17 Mar 2017 22:13:41 +0000 (17:13 -0500)]
ibmvnic: Move login to its own routine

Move the code that handles login and renegotiation of ibmvnic
capabilities to its own routine.

Signed-off-by: Nathan Fontenot <nfont@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoibmvnic: Move login and queue negotiation into ibmvnic_open
John Allen [Fri, 17 Mar 2017 22:13:40 +0000 (17:13 -0500)]
ibmvnic: Move login and queue negotiation into ibmvnic_open

VNIC server expects LINK_STATE_UP to be sent within 30s of the login. If we
exceed the timeout, VNIC server will attempt to fail over. Since time
between probe and open of the device is indeterminate, move login and queue
negotiation into ibmvnic open so we can guarantee that login and sending
LINK_STATE_UP occur within the 30s window.

Signed-off-by: John Allen <jallen@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: sun: sungem: rix a possible null dereference
Philippe Reynes [Fri, 17 Mar 2017 21:54:04 +0000 (22:54 +0100)]
net: sun: sungem: rix a possible null dereference

The function gem_begin_auto_negotiation dereference
the pointer ep before testing if it's null. This
patch add a check on ep before dereferencing it.

Fixes: 92552fdda557 ("net: sun: sungem: use new api
ethtool_{get|set}_link_ksettings")

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Philippe Reynes <tremyfr@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoliquidio: add debug error messages to report command timeout
Rick Farrington [Fri, 17 Mar 2017 18:23:08 +0000 (11:23 -0700)]
liquidio: add debug error messages to report command timeout

Add timeout error message in lio_process_ordered_list().  Add host failure
status in existing error message in if_cfg_callback().

Signed-off-by: Rick Farrington <ricardo.farrington@cavium.com>
Signed-off-by: Felix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoliquidio: remove duplicate code
Satanand Burla [Fri, 17 Mar 2017 17:50:05 +0000 (10:50 -0700)]
liquidio: remove duplicate code

Remove code duplicated in PF and VF; define that code once only in a common
header file included by PF and VF.

Signed-off-by: Satanand Burla <satananda.burla@cavium.com>
Signed-off-by: Felix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'stmmac-mq-part3'
David S. Miller [Wed, 22 Mar 2017 00:24:02 +0000 (17:24 -0700)]
Merge branch 'stmmac-mq-part3'

Joao Pinto says:

====================
net: stmmac: adding multiple buffers and routing

As agreed with David Miller, this patch-set is the third and last to enable
multiple queues in stmmac.

This third one focuses on:

a) Enable multiple buffering to the driver and queue independent data
b) Configuration of RX and TX queues' priority
c) Configuration of RX queues' routing
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: stmmac: RX queue routing configuration
Joao Pinto [Fri, 17 Mar 2017 16:11:07 +0000 (16:11 +0000)]
net: stmmac: RX queue routing configuration

This patch adds the configuration of RX queues' routing.

Signed-off-by: Joao Pinto <jpinto@synopsys.com>
Signed-off-by: David S. Miller <davem@davemloft.net>