vhost: add fdset-event thread name

This patch adds the name for vhost fdset thread.
It can help us to know whether the thread is running.

Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Reviewed-by: Jianfeng Tan <jianfeng.tan@intel.com>

vhost: raise error on fdset-thread creation

When first call the 'rte_vhost_driver_start', the
fdset_event_dispatch thread should be created successfully.
Because the vhost uses it to poll socket events for vhost
server or clients. Without it, for example, vhost will not
get the connection event.

This patch returns err code directly when created not successful.

Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Reviewed-by: Jianfeng Tan <jianfeng.tan@intel.com>

net/vhost: fix crash when creating vdev dynamically

When creating vdev dynamically, vhost pmd driver starts directly without
checking TX/RX queues are ready or not, and thus causes segmentation fault
when vhost library accesses queues. This patch adds a flag to check whether
queues are setup or not, and adds queues setup into dev_start function to
allow user to start them after setting up.

Fixes: aed0b12930b3 ("net/vhost: fix socket file deleted on stop")
Cc: stable@dpdk.org
Signed-off-by: Junjie Chen <junjie.j.chen@intel.com>
Tested-by: Zhiyong Yang <zhiyong.yang@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

vhost: avoid concurrency when logging dirty pages

This patch aims at fixing a migration performance regression
faced since atomic operation is used to log pages as dirty when
doing live migration.

Instead of setting a single bit by doing an atomic read-modify-write
operation to log a page as dirty, this patch write 0xFF to the
corresponding byte, and so logs 8 page as dirty.

The advantage is that it avoids concurrent atomic operations by
multiple PMD threads, the drawback is that some clean pages are
marked as dirty and so are transferred twice.

Fixes: 897f13a1f726 ("vhost: make page logging atomic")
Cc: stable@dpdk.org
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Jianfeng Tan <jianfeng.tan@intel.com>

ethdev: fix port accessing after release

rte_eth_dev_pci_release() function wrongly releases an ethdev port and
then releases internal fields of this port.
This behavior is problematic, because after the release, the port may
be reallocated again by another thread or just be invalid for any
usage.

Move the release operation to the end of the function.

Fixes: dcd5c8112bc3 ("ethdev: add PCI driver helpers")
Cc: stable@dpdk.org
Suggested-by: Ferruh Yigit <ferruh.yigit@intel.com>
Signed-off-by: Matan Azrad <matan@mellanox.com>
Acked-by: Thomas Monjalon <thomas@monjalon.net>

net/qede: fix to prevent overwriting packet type

Fixes: 3f72dd780e10 ("net/qede: check tunnel L3 header")
Cc: stable@dpdk.org
Signed-off-by: Harish Patil <harish.patil@cavium.com>

net/qede: fix device stop to remove primary MAC

This fix is to remove primary MAC filter during dev stop. Without the
fix device start attempt to add the primary MAC fails. Perform MAC set
remove under IS_PF() check.

Fixes: dd28bc8c6ef4 ("net/qede: fix VF port creation sequence")
Cc: stable@dpdk.org
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>

net/qede: add devargs to disable NPAR Tx switching

Added a run-time argument npar_tx_switching to enable/disable
Tx switching for multi-fucntion devices (NPAR).
Rename vf_txswitch option to vf_tx_switching.

Sample usage to disable Tx switching for NPAR or VF is something like...

 -w 05:00.0,npar_tx_switching=0 -w 05:00.1,npar_tx_switching=0
 -w 05:00.0,vf_tx_switching=0 -w 05:00.1,vf_tx_switching=0

Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>

net/qede: enable IPGRE offload support

Signed-off-by: Harish Patil <harish.patil@cavium.com>

net/qede: convert to new Rx/Tx offloads API

Ethdev RX/TX offloads API has changed since:
commit ce17eddefc20 ("ethdev: introduce Rx queue offloads API")
commit cba7f53b717d ("ethdev: introduce Tx queue offloads API")

This patch makes use of new offload flags as suggested and fix
the default RX/TX configuration. Also indent the code properly
around those changes.

Signed-off-by: Harish Patil <harish.patil@cavium.com>
Signed-off-by: Rasesh Mody <rasesh.mody@cavium.com>

net/mrvl: rename PMD as mvpp2

The name "mrvl" for Marvell PMD driver for PPv2 Marvell PPv2
(Packet Processor v2) 1/10 Gbps adapter is too generic and causes
problem for adding new PMD drivers for other Marvell devices.
Changed to "mvpp2" for specific Marvell PPv2 PMD.

This patch doesn't introduce any change except renaming.

Signed-off-by: Natalie Samsonov <nsamsono@marvell.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/sfc: support Medford2 family adapters

Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Andy Moreton <amoreton@solarflare.com>

net/sfc: support runtime VI window size

Medford2 uses a configurable VI window size, and requires
updates to register accesses to use a runtime VI window size
rather than the *_STEP register constants used for earlier
controllers.

Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Andy Moreton <amoreton@solarflare.com>

net/sfc: support link speeds up to 100G

Add 25G, 50G and 100G.

Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Andy Moreton <amoreton@solarflare.com>

net/avf: fix link autoneg value

Intention of the logic is:
If ETH_LINK_SPEED_FIXED is set in dev_conf, set link_autoneg to
ETH_LINK_FIXED, else set it to ETH_LINK_AUTONEG

ETH_LINK_SPEED_FIXED value is 1, and ETH_LINK_FIXED is 0;

Current logic is broken, fix it according above values.

Fixes: 48de41ca11f0 ("net/avf: enable link status update")
Cc: stable@dpdk.org
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>

net/i40e: fix support DDP packages group 0xff

Group ID 0xFF indicates that packages does not change parser
graph so compatible with any other packages.

Fixes: b1ec717bfff5 ("net/i40e: fix multiple DDP packages conflict")
Cc: stable@dpdk.org
Signed-off-by: Kirill Rybalchenko <kirill.rybalchenko@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>

net/i40e: add log when setting input set

This patch adds log when setting input set since global
configuration is changed.

Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Reviewed-by: Kirill Rybalchenko <kirill.rybalchenko@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>

net/ixgbe: fix intr callback unregister by adding retry

The nic's interrupt source has some active callbacks, when
the port hotplug. Add a retry to give more port's a chance
to uninit before returning an error.

Fixes: 2866c5f1b87e ("ixgbe: support port hotplug")
Cc: stable@dpdk.org
Signed-off-by: Yunjian Wang <wangyunjian@huawei.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>

vhost: do not generate signal when sendmsg fails

More precisely, do not generate a SIGPIPE signal if the peer
has closed the connection. Otherwise, it will terminate the
process by default. As a library, we should avoid terminating
the application process when error happens and just need to
return with an error.

Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

vhost: support sending fds via slave channel

This function will be used to send fds to QEMU via slave channel.

Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/virtio: add check for cvq

Add checking for cvq to judge if virtio_ack_link_announce should be called.
The existing code doesn't cause issue, and add the checking just to look
more reasonable.

Signed-off-by: Zhiyong Yang <zhiyong.yang@intel.com>
Reviewed-by: Jianfeng Tan <jianfeng.tan@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/virtio: fix queues pointer check

It is necessary to add pointer checking because in some case the
code will cause crash. For example, the code goes here before
memory allocation of rxvq is finished.

Fixes: 7365504f77e3 ("net/virtio: support guest announce")
Cc: stable@dpdk.org
Signed-off-by: Zhiyong Yang <zhiyong.yang@intel.com>
Reviewed-by: Jianfeng Tan <jianfeng.tan@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/virtio-user: specify MAC of the tap

When using virtio-user with vhost-kernel to exchange
packet with kernel networking stack, application can
set the MAC of the tap interface via parameter.

Signed-off-by: Ning Li <muziding001@163.com>
Reviewed-by: Seán Harte <seanbh@gmail.com>
Tested-by: Seán Harte <seanbh@gmail.com>
Reviewed-by: Jianfeng Tan <jianfeng.tan@intel.com>

net/cxgbe: convert to SPDX license tags

Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
Signed-off-by: Kumar Sanghvi <kumaras@chelsio.com>

net/cxgbe: support setting MAC address

Signed-off-by: Kumar Sanghvi <kumaras@chelsio.com>
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>

net/cxgbe: add VF port statistics

Signed-off-by: Kumar Sanghvi <kumaras@chelsio.com>
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>

net/cxgbe: update Tx and Rx path for VF

On TX path, add fw_eth_tx_pkt_vm_wr to transmit packets over VF.
Use is_pf4() to correctly calculate the work request size and
offsets within the work request.  On RX path, use pktshift to adjust
data offset within the mbuf.

Signed-off-by: Kumar Sanghvi <kumaras@chelsio.com>
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>

net/cxgbe: enable RSS for VF

Query firmware and configure RSS for VF.

Signed-off-by: Kumar Sanghvi <kumaras@chelsio.com>
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>

net/cxgbe: initialize SGE and queues for VF

Query firmware and initialize SGE parameters and enable queue
allocation for VF.  Calculate pcie channel and queue congestion
management for VF.

Signed-off-by: Kumar Sanghvi <kumaras@chelsio.com>
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>

net/cxgbe: add probe to initialize VF devices

Add probe to initialize VF devices.  Separate init/de-init paths
for PF and VF.  Do firmware state initialization wrt VF and retrieve
various operational parameters by querying firmware.  Finally configure
and initialize ports.

Signed-off-by: Kumar Sanghvi <kumaras@chelsio.com>
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>

net/cxgbe: add base for enabling VF ports

Add base to enable VF ports in subsequent patches.

Signed-off-by: Kumar Sanghvi <kumaras@chelsio.com>
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>

net/cxgbe: add VF firmware mailbox

Add firmware mailbox communication support for VF.  Add is_pf4()
to check if driver is attached to PF4.  Use is_pf4() to determine
whether to use PF or VF mailbox communication.

Signed-off-by: Kumar Sanghvi <kumaras@chelsio.com>
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>

net/cxgbe: add skeleton VF driver

Add minimal VF driver.  Declare functions common to both PF and VF
functionality in separate header file and import the header file.

Signed-off-by: Kumar Sanghvi <kumaras@chelsio.com>
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>

net/avp: convert to SPDX license tags

As per the following commit the AVP files maintained by Wind River are
converted to use the SPDX license tag format.

    commit a4862c9e1a98 ("license: introduce SPDX identifiers")

Signed-off-by: Allain Legacy <allain.legacy@windriver.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/liquidio: move to new offload API

Make use of new offloads member instead of bit fields in port Rx conf.

Signed-off-by: Shijith Thotton <shijith.thotton@caviumnetworks.com>

net/mlx5: fix RSS key length query

The RSS key length returned by rte_eth_dev_info_get command was taken
from the
PMD private structure. This structure initialization was done only after
the port configuration.

Considering Mellanox device supports only 40B long RSS key, reporting
the fixed number instead.

Fixes: 29c1d8bb3e79 ("net/mlx5: handle a single RSS hash key for all protocols")
Cc: stable@dpdk.org
Signed-off-by: Shahaf Shuler <shahafs@mellanox.com>
Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

net/mlx5: enforce RSS key length limitation

RSS hash key must be 40 Bytes long.

Cc: stable@dpdk.org
Signed-off-by: Shahaf Shuler <shahafs@mellanox.com>
Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

net/mlx5: setup RSS regardless of queue count

In some environments it is desirable to have the NIC perform RSS
normally on the packet regardless of the number of queues configured.
The RSS hash result that is stored in the mbuf can then be used by
the application to make decisions about how to distribute workloads
to threads, secondary processes, or even virtual machines if the
application is a virtual switch.  This change to the mlx5 driver
aligns with how other drivers in the Intel family work.

Signed-off-by: Allain Legacy <allain.legacy@windriver.com>
Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Yongseok Koh <yskoh@mellanox.com>
Tested-by: Allain Legacy <allain.legacy@windriver.com>

net/mlx5: fix icc build

Remove the second declaration of device_attr [1] inside the loop as well as
the query_device_ex() which has already been done outside of the loop.

[1] https://dpdk.org/ml/archives/dev/2018-March/091744.html

Fixes: 9a761de8ea14 ("net/mlx5: flow counter support")
Cc: stable@dpdk.org
Reported-by: Ferruh Yigit <ferruh.yigit@intel.com>
Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Yongseok Koh <yskoh@mellanox.com>
Tested-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/mlx5: fix TSO enablement

TSO should be set if either of the TSO offload flags is requested.

Fixes: dbccb4cddcd2 ("net/mlx5: convert to new Tx offloads API")
Cc: stable@dpdk.org
Signed-off-by: Shahaf Shuler <shahafs@mellanox.com>
Acked-by: Yongseok Koh <yskoh@mellanox.com>

net/igb: fix flow RSS queue index

Igb has a bug about queue index for rss rule flow configuration
when set redirection table.

Fixes: 424ae915baf0 ("net/e1000: move RSS to flow API")
Cc: stable@dpdk.org
Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Tested-by: Yuan Peng <yuan.peng@intel.com>
Acked-by: Wenzhuo Lu <wenzhuo.lu@intel.com>

net/i40evf: add multicast MAC address filtering

Add support the set_mc_addr_list device operation in the i40evf PMD.

The configured addresses are stored in the device private area, so
they can be flushed before adding new ones.

Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>

net/i40e: fix flow RSS configuration error

I40e needs a protection for rss rule flow configuration
after a rule is set for a port.

Fixes: ecad87d22383 ("net/i40e: move RSS to flow API")
Cc: stable@dpdk.org
Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Tested-by: Yuan Peng <yuan.peng@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>

net/i40e: fix intr callback unregister by adding retry

The nic's interrupt source has some active callbacks, when
the port hotplug. Add a retry to give more port's a chance
to uninit before returning an error.

Fixes: d42aaf30008b ("i40e: support port hotplug")
Cc: stable@dpdk.org
Signed-off-by: Yunjian Wang <wangyunjian@huawei.com>
Reviewed-by: Kirill Rybalchenko <kirill.rybalchenko@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>

net/sfc: add device parameter to choose FW variant

Add support of choice the preferred firmware variant to use in device
parameters.

Signed-off-by: Roman Zhukov <roman.zhukov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>

net/sfc/base: add values for RxDPCPU firmware id recognition

Signed-off-by: Roman Zhukov <roman.zhukov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>

net/sfc/base: support choosing firmware variant

Signed-off-by: Gautam Dawar <gdawar@solarflare.com>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>

net/bnxt: fix flow director with same cmd different queue

When user reissues same flow director cmd with a different queue
update the existing filter to redirect flow to the new desired
queue as destination just like the other filters like 5 tuple and
generic flow.

Fixes: 2d64da097aa0 ("net/bnxt: support FDIR")
Cc: stable@dpdk.org
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>

net/sfc: fix type of opaque pointer in perf profile handler

The 'opaque' pointer in handler function is the last argument
of sfc_kvargs_process() function and it is pointer to the adapter
'evq_flags' that has a uint32_t type. So 'value' must be pointer
to uint32_t.

Fixes: c22d3c508e0c ("net/sfc: support parameter to choose performance profile")
Cc: stable@dpdk.org
Signed-off-by: Roman Zhukov <roman.zhukov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>

ethdev: fix queue start

Device must be started before start any queue.

Fixes: 0748be2cf9a2 ("ethdev: queue start and stop")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>

ethdev: return named opaque type instead of void pointer

"struct rte_eth_rxtx_callback" is defined as internal data structure and
used as named opaque type.

So the functions that are adding callbacks can return objects in this
type instead of void pointer.

Also const qualifier added to "struct rte_eth_rxtx_callback *" to
protect it better from application modification.

Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>

net/avp: conform to new ethdev offload API

This change updates the AVP driver to conform with the new ethdev
offload API.  As per these commits:

    commit cba7f53b717d ("ethdev: introduce Tx queue offloads API")'
    commit ce17eddefc20 ("ethdev: introduce Rx queue offloads API")'

Signed-off-by: Allain Legacy <allain.legacy@windriver.com>

net/sfc: fix mbuf data alignment calculation

Unlike ffs() rte_bsf32() counts bit position from 0.

Fixes: 0c7a0c35f24c ("net/sfc: calculate Rx buffer size which may be used")
Cc: stable@dpdk.org
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>

ethdev: remove unused struct forward declaration

Fixes: 331c447ad913 ("ethdev: separate internal structures into own header")

Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Shreyansh Jain <shreyansh.jain@nxp.com>

ethdev: support dynamic logging

Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Reviewed-by: Shreyansh Jain <shreyansh.jain@nxp.com>

net/dpaa: prefer defines for link speed values

Use existing defines instead of hardcoded vales.

Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Shreyansh Jain <shreyansh.jain@nxp.com>

net/bnxt: fix LRO disable

When the vnic_tpa_cfg HWRM command is sent to the FW,
we are not passing the VNIC ID in case of disable.
This can cause the FW to return an error.
Correct VNIC ID needs to be passed for both enable and disable.

Fixes: 0958d8b6435d ("net/bnxt: support LRO")
Cc: stable@dpdk.org
Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com>

ethdev: fix port id storage

port_id is now 16bits, update function parameter according.

Fixes: 4c270218aa26 ("ethdev: support security APIs")
Cc: stable@dpdk.org
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Hemant Agrawal <hemant.agrawal@nxp.com>
Acked-by: Thomas Monjalon <thomas@monjalon.net>

net/sfc: add dynamic log level for MCDI messages

Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Andy Moreton <amoreton@solarflare.com>

net/sfc: remove dedicated init log parameter

The previous patches in the set convert per-port
logging to use NOTICE level and make this level default.
This provides the possibility to remove the dedicated
toggle for init-related messages and merge init logging
with the main log type. In order to keep these logs silent
by default, INFO level should be used.

Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Andy Moreton <amoreton@solarflare.com>

net/sfc: prepare to merge init logs with main log type

Conversion to dynamic logging done in the previous patches
makes it possible to simplify internal controls for init
logging. This patch allows to prepare for such a change.
It makes init-unrelated messages use NOTICE level so that
the following patch will be able to convert init logging
to use INFO level and remain silent by default.

Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Andy Moreton <amoreton@solarflare.com>

net/sfc: support per-port dynamic logging

Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Andy Moreton <amoreton@solarflare.com>

net/sfc: support driver-wide dynamic logging

Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Andy Moreton <amoreton@solarflare.com>

eal: register log type and pick level from args

Dynamic log types are registered on RTE_INIT() step.
This allows one to set log levels by EAL options on
application launch. However, this does not allow to
manage log types if they are created during runtime.

EAL does not store log levels and types passed from
the command line. Thus, they cannot be picked later.
This is an obvious flaw since it would be better to
be able to pick levels for dynamic types registered
for runtime-determined facilities such as NIC ports.

This patch provides a mechanism to store log levels
passed from EAL options and adds an API to register
log types and pick levels from the internal storage.

Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Andy Moreton <amoreton@solarflare.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/mlx5: refuse empty VLAN flow specification

Verbs specification doesn't help to distinguish between packets having an
VLAN and those which do not have, this ends by having flow rule which does
not react as the user expects e.g.

 flow create 0 ingress pattern eth / vlan / end action queue index 0 / end
 flow create 0 ingress pattern eth / end action queue index 1 / end

are colliding in Verbs definition as in both rule are matching packets with
or without VLAN.
For this reason, the VLAN specification must not be empty, otherwise the
PMD has to refuse it.

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

net/mlx5: improve flow error explanation

Fill the error context in conversion function to provide a better reason on
why it cannot be done to the user.

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

net/mlx5: change tunnel flow priority

Packet matching inner and outer flow rules are caught by the first one
added in the device as both flows are configured with the same priority.
To avoid such situation, the inner flow can have an higher priority than
the outer ones as their pattern matching will otherwise collide.

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

net/mlx5: fix link status to use wait to complete

Wait to complete is present to let the application get a correct status
when it requires it, it should not be ignored.

Fixes: e313ef4c2fe8 ("net/mlx5: fix link state on device start")
Fixes: cb8faed7dde8 ("mlx5: support link status update")
Cc: stable@dpdk.org
Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: fix link status behavior

This behavior is mixed between what should be handled by the application
and what is under PMD responsibility.

According to DPDK API:
- link_update() should only query the link status [1]
- link_set_{up,down}() should only set the link to the according status [1]
- dev_{start,stop}() should enable/disable traffic reception/emission [2]

On this PMD, the link status is retrieved from the net device associated
owned by the Linux Kernel, it does not means that even when this interface
is down, the PMD cannot send/receive traffic from the NIC those two
information are unrelated, until the physical port is active and has a
link, the PMD can receive/send traffic on the wire.

According to DPDK API, calling the rte_eth_dev_start() even when the Linux
interface link is down is then possible and allowed, as the traffic will
flow between the DPDK application and the Physical port.

This also means that a synchronization between the Linux interface and the
DPDK application remains under the DPDK application responsibility.

To handle such synchronization the application should behave as the
following scheme, to start:

 rte_eth_get_link(port_id, &link);
 if (link.link_status == ETH_DOWN)
rte_eth_dev_set_link_up(port_id);
 rte_eth_dev_start(port_id);

Taking in account the possible returned values for each function.

and to stop:

 rte_eth_dev_stop(port_id);
 rte_eth_dev_set_link_down(port_id);

The application should also set the LSC interrupt callbacks to catch and
behave accordingly when the administrator set the Linux device down/up.
The same callbacks are called when the link on the medium falls/raise.

[1] https://dpdk.org/browse/dpdk/tree/lib/librte_ether/rte_ethdev_core.h
[2] https://dpdk.org/browse/dpdk/tree/lib/librte_ether/rte_ethdev.h#n1677

Fixes: c7bf62255edf ("net/mlx5: fix handling link status event")
Fixes: e313ef4c2fe8 ("net/mlx5: fix link state on device start")
Cc: stable@dpdk.org
Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Acked-by: Yongseok Koh <yskoh@mellanox.com>

net/mlx5: remove kernel version check

Kernel version check was introduced in
commit 3a49ffe38a95 ("net/mlx5: fix link status query")
due to a bug fixed by
commit ef09a7fc7620 ("net/mlx5: fix inconsistent link status query")

This patch restore the previous behavior as described in Linux API.

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: fix ARM build

rdma-core v16 has a bug. The following compilation error occurs on ARM
hosts.

In file included
from drivers/net/mlx5/mlx5_glue.h:16:0,
from drivers/net/mlx5/mlx5_glue.c:11:
/usr/include/infiniband/mlx5dv.h:144:2: error: unknown type name 'off_t'
off_t   uar_mmap_offset;
^

As a temporary fix, sys/types.h is included in PMD. This has been fixed in
rdma-core v17. This can be removed when all the Linux distros are shipped
with rdma-core v17 or back-ported fix. As of now, RedHat 7.5 is known to
have rdma-core v16.

Cc: stable@dpdk.org
Signed-off-by: Yongseok Koh <yskoh@mellanox.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: fix existing file removal

There is no guarantee that the file won't be removed by external
user/application between the stat() and remove() syscalls, remove() will
fail if the file no longer exists.

Fixes: f8b9a3bad467 ("net/mlx5: install a socket to exchange a file descriptor")
Cc: stable@dpdk.org
Signed-off-by: Xueming Li <xuemingl@mellanox.com>
Acked-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>

net/mlx5: use dynamic logging

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: use port id in PMD log

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: standardize on negative errno values

Set rte_errno systematically as well.

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: change non failing function return values

These functions return int although they are not supposed to fail,
resulting in unnecessary checks in their callers.
Some are returning error where is should be a boolean.

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: prefix all functions with mlx5

This change removes the need to distinguish unlocked priv_*() functions
which are therefore renamed using a mlx5_*() prefix for consistency.

At the same time, all functions from mlx5 uses a pointer to the ETH device
instead of the one to the PMD private data.

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: remove control path locks

In priv struct only the memory region needs to be protected against
concurrent access between the control plane and the data plane.

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: remove useless empty lines

Some empty lines have been added in the middle of the code without any
reason.  This commit removes them.

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: add missing function documentation

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: normalize function prototypes

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: mark parameters with unused attribute

Replaces all (void)foo; by __rte_unused macro except when variables are
under #if statements.

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: name parameters in function prototypes

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mlx5: fix sriov flag

priv_get_num_vfs() was used to help the PMD in prefetching the mbuf in
datapath when the PMD was behaving in VF mode.
This knowledge is no more used.

Fixes: 528a9fbec6de ("net/mlx5: support ConnectX-5 devices")
Cc: stable@dpdk.org
Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>

net/mrvl: fix Rx descriptors number

Since filling hardware buffer pool (bpool) is Rx related
constant describing maximum number of rx descriptors
instead of maximum number of Tx descriptors should be used.

Fixes: 0ddc9b815b11 ("net/mrvl: add net PMD skeleton")
Cc: stable@dpdk.org
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: Tomasz Duszynski <tdu@semihalf.com>

net/mrvl: fix typo in log message

Show appropriate log message in case Tx offloads are either
not supported or missing.

Fixes: 7d8f6c20cc7c ("net/mrvl: switch to the new Tx offload API")
Cc: stable@dpdk.org
Signed-off-by: Tomasz Duszynski <tdu@semihalf.com>

net/tap: allow user MAC to be passed as args

Allow TAP PMD to pass user desired MAC address as argument.
The argument value is processed as string delimited by  ':',
is parsed and converted to HEX MAC address after validation.

Signed-off-by: Vipin Varghese <vipin.varghese@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Pascal Mazon <pascal.mazon@6wind.com>

net/null: support bulk allocation

Bulk allocation of multiple mbufs increased more than ~2%  and less
than 8% throughput on single core (1.8 GHz), based on usage for example
1: Testpmd case: Two null devices with copy 8% improvement.
    testpmd -c 0x3 -n 4 --socket-mem 1024,1024
--vdev 'eth_null0,size=64,copy=1' --vdev 'eth_null1,size=64,copy=1'
-- -i -a --coremask=0x2 --txrst=64 --txfreet=64 --txd=256
--rxd=256 --rxfreet=64 --burst=64 --txpt=64 --txq=1 --rxq=1 --numa
2. Ovs switch case: 2% improvement.
$VSCTL add-port ovs-br dpdk1 -- set Interface dpdk1 type=dpdk \
        options:dpdk-devargs=eth_null0,size=64,copy=1
$VSCTL add-port ovs-br dpdk2 -- set Interface dpdk2 type=dpdk \
        options:dpdk-devargs=eth_null1,size=64,copy=1

Signed-off-by: Mallesh Koujalagi <malleshx.koujalagi@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/null: set different MAC address for each device

Previously all null devices were getting MAC address "00:00:00:00:00:00"

After attaching two Null device to OVS, seeing "00.00.00.00.00.00" mac
address for both null devices.

Fix this issue, by setting different mac address.

Signed-off-by: Mallesh Koujalagi <malleshx.koujalagi@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/bonding: set MTU on slave configure

When a slave is configured make sure that the slave's MTU
matches the bonding's idea of the MTU.

Signed-off-by: Chas Williams <chas3@att.com>
Acked-by: Radu Nicolau <radu.nicolau@intel.com>

net/enic: use link status helper functions

Use new rte_eth_linkstatus_get/set helper functions to handle link
status update.

This driver was not doing atomic update of link status information.
And the return value was different than others.
The hardware also does not do autonegotiation (at least on Linux).

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/octeontx: use link status helper functions

Use new rte_eth_linkstatus_get/set helper functions to handle link
status update.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/szedata2: use link status helper functions

Yet another driver which was not returing correct value on
link change.

Use new rte_eth_linkstatus_get/set helper functions to handle link
status update.

Since this driver can't be built on x86 could not even
do a compile test.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/thunderx: use link status helper functions

Use new rte_eth_linkstatus_get/set helper functions to handle link
status update.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/liquidio: use link status helper functions

Use new rte_eth_linkstatus_get/set helper functions to handle link
status update.
And cleanup the logic in the the link update routine.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Tested-by: Shijith Thotton <shijith.thotton@caviumnetworks.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/i40e: use link status helper functions

Use new rte_eth_linkstatus_get/set helper functions to handle link
status update.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/sfc: use link status helper functions

Use new rte_eth_linkstatus_get/set helper functions to handle link
status update.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/ixgbe: use link status helper functions

Use new rte_eth_linkstatus_get/set helper functions to handle link
status update.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/e1000: use link status helper functions

Use new rte_eth_linkstatus_get/set helper functions to handle link
status update.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/nfp: use link status helper functions

Use new rte_eth_linkstatus_get/set helper functions to handle link
status update.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>