net/i40e/base: fix compile issue for GCC 6.3

When using GCC 6.3.0 and EXTRA_CFLAGS=-Og, the compiler is not smart
enough to see that when you have an integrated VF that just one of the
if statemets has to be true so that ntu is always set. Fix the way that
the code is for integrated VFs so that the compiler won't complain.

Fixes: 9783eb15eb98 ("i40e/base: wrap the register definitions for PF and VF")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>

net/i40e/base: add AQ critical error type

The FW has the ability to return a critical error on every AQ command.
So add the new return type as critical error to sync with firmware.

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

net/i40e/base: add rearrange process AQ command

Add new AQ command that help to inform software how advanced
rearrangement process is.

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

net/i40e/base: enhance loopback AQ command

Enhanced loopback AQ command to allow variety of loopback modes.

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

net/i40e/base: add defines for flat NVM

New control bits were defined for flat NVM structure.
One for pointing which NVM bank is currently used and
one used as indicator of structure type.

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

net/i40e/base: fix link LED blink

This fix solves an issue occurring while calling i40e_led_set function
from the driver with "blink" parameter set as TRUE. This call resulted
in Activity LED blinking instead of Link LED, which may lead to errors
in physically identyfying the port, since Activity LED may be blinking
for different reasons aswell.

Fixes: 51b27bc5d898 ("i40e/base: update LED blinking")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>

net/i40e/base: enable AQ event get in NVM update

This patch adds new I40E_NVMUPD_GET_AQ_EVENT state
to allow retrieval of AdminQ events as a result.

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

net/i40e/base: add NVM update preservation flags

Add preservation flags support on X722 devices for NVM update AdminQ
function wrapper. Add new parameter and handling to nvm update admin
queue function intended to allow nvmupdate tool to configure the
preservation flags in the AdminQ command.

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

net/i40e/base: clean up

1. Cleanup logical in __i40e_read_nvm_word and reuse it in
i40e_read_nvm_word.
2. Change couple functions to static so we can remove uncessary function
declarations
3. Cleanup logical in i40e_validate_nvm_checksum.
4. Add more debug information in PF reset
5. correct comments
6. Minor code style fix.

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

net/i40e/base: fix NVM lock

NVM lock is only required at below situation:
1. its X722 device or
2. API version >= 1.5
or may have issue at firmware downgrade.

Fixes: c5846a125b26 ("net/i40e/base: fix NVM access interfering")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>

net/i40e/base: add AQ command for DCB parameters

Add new Adminq command i40e_aqc_opc_set_dcb_parameters that
helps to get DCB parameters in data structure i40e_aqc_set_dcb_parameters.

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

net/i40e/base: add link speed convert function

Introduce a helper function which can convert the AdminQ link speed
definition into a virtchnl definition. This can then be used by host
driver implementations in order to safely convert to the interface
recognized by the virtual functions.

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

net/i40e/base: code refactoring for LED blink

Code refactory, wrap repeated code in function.

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

net/i40e/base: add macro for 25G device

Add macro to idenfity 25G device by device ID

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

net/i40e/base: add byte swaps in PHY register access

Add byte swap commandwhen transferring data between memory and firmware
while using PHY register access functions. Endianness also need to be
handled correctly during these swaps.

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

net/i40e/base: retry AQC to overcome IRCRead hangs

When the I2C is busy, the PHY reads are delayed. The firmware will
return EGAIN in these cases with an expectation that the SW will trigger
the reads again. This patch retries the operation for a maximum period
of 500ms.

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

net/i40e/base: release spinlock before function returns

Fix a bug that adminq spin lock is not be released before function
i40e_nvmupd_command returns.

Fixes: 5a2e46a3e405 ("net/i40e/base: refactor NVM update command processing")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>

net/i40e/base: add (Q)SFP module memory access definitions

Add definations used for accessing and interrupting contents
of (Q)SFP module internal EEPROM memory.

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

net/i40e/base: add capability macros

Add capability macros for 25G AOC/ACC PHYS.

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

net/i40e/base: add new PHY type

Add new PHY type I40E_PHY_TYPE_NOT_SUPPORTED_HIGH_TEMP, it will be set
by firmware if device module does not meet the thermal requirements.

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

doc: add DDP in i40e guide

Update i40e.rst to declare DDP support and input set configuration,
as implemented in DPDK 17.11.

Cc: stable@dpdk.org
Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: John McNamara <john.mcnamara@intel.com>

net/ixgbe: fix the failure of number of Tx queue check

Tx_q check fails when the SRIOV is active and tx_q > rx_q. It has
to use the maximum HW queue number for calculation, to get the
right number of queue can be used.

Fixes: 27b609cbd1c6 ("ethdev: move the multi-queue mode check to specific drivers")
Cc: stable@dpdk.org
Signed-off-by: Yanglong Wu <yanglong.wu@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>

net/i40e: exclude LLDP packet count

When use port stats register to calculate the packet count, LLDP packets
are counted in statistics which is not expected, the patch exclude this
number from total number.

Fixes: 763de290cbd1 ("net/i40e: fix packet count for PF")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>

net/i40e: turn off flexible payload on driver init

Function i40e_GLQF_reg_init() overwrites global register for
flexible payload, forcing extraction of first 16 bytes of
L2/L3/L4 payload to the field vector even if flexible payload
is not used by an application. Such unconditional turn on of
flexible payload effectively disables ability to use outer
IP Destination address for RSS/FDIR for tunnelled packets,
as flexible payload overwrites outer IP destination address
on the field vector.
Now flexible payload turned on only when flow director is
enabled and configured.

Signed-off-by: Kirill Rybalchenko <kirill.rybalchenko@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>

net/i40e: determine number of queues per VF at run time

Without this patch, the number of queues per i40e VF is set to 4
by CONFIG_RTE_LIBRTE_I40E_QUEUE_NUM_PER_VF=4 in config/common_base.
It is a fixed value determined at compile time and can't be changed
at run time.
With this patch, the number of queues per i40e VF can be determined
at run time. For example, if the PCI address of an i40e PF is
aaaa:bb.cc, with the EAL parameter -w aaaa:bb.cc,queue-num-per-vf=8,
the number of queues per VF created from this PF is set to 8.
If there is no "queue-num-per-vf" setting in EAL parameters, it uses
the default value of 4. And if the value after the "queue-num-per-vf"
is invalid, it will also use the default value. The valid values can
be 1, 2, 4, 8, or 16.

Signed-off-by: Wei Dai <wei.dai@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>

net/i40e: change VF default ITR interval

Current default ITR interval I40E_QUEUE_ITR_INTERVAL_DEFAULT is
used by both DPDK PF and DPDK VF, but this ITR interval value
will cause lots of interrupts when using VF. This patch modifies
VF default ITR interval to reduce interrupts.
ITR interval can be reconfigured by RTE_LIBRTE_I40E_ITR_INTERVAL
if expecting low latency.

Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>

net/i40e: fix FDIR rule confiliction issue

Failed to create two FDIR rules with different vlan id for
the same PCTYPE. Root cause is that wrong hash key length
is used.

Fixes: 4149825bbdb9 ("net/i40e: finish integration FDIR with generic flow API")
Cc: stable@dpdk.org
Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>

net/ixgbe: move RSS to flow API

Rte_flow was defined to include RSS, this patch moves ixgbe
existing RSS to rte_flow. The old RSS configuration is kept
as it was, and can be deprecated in the future.

Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Wei Dai <wei.dai@intel.com>

net/e1000: move RSS to flow API

Rte_flow was defined to include RSS, this patch moves igb
existing RSS functionalities into rte_flow. The old RSS
configuration method is kept as it was, and can be deprecated in
the future.

Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Wei Dai <wei.dai@intel.com>

net/ixgbe: fix parsing FDIR NVGRE issue

This patch fixes the issue of mask check in NVGRE parser
for flow API.

Fixes: 11777435c727 ("net/ixgbe: parse flow director filter")
Cc: stable@dpdk.org
Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Wenzhuo Lu <wenzhuo.lu@intel.com>

doc: add sfc tunnels support to release features

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

net/sfc: support inner checksum offload on transmit

There is no dedicated controls for inner checksum offload on
device/queue level. So, enable together with outer offloads.

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

net/sfc: correct Rx checksum offloads for tunnel packets

In the case of tunnel packet, PKT_RX_{IP,L4}_CSUM_* flags correspond
to inner packet checksums. There is only one flag to indicate bad
external IPv4 header checksum.

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

net/sfc: support VXLAN and NVGRE packet types classification

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

net/sfc: fix incorrect bitwise ORing of L3/L4 packet types

Not a bug since value is set only once, but it is still incorrect.

Fixes: 638bddc99faa ("net/sfc: implement EF10 native Rx datapath")
Cc: stable@dpdk.org
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>

net/sfc: support UDP tunnel ports configuration

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

net/sfc/base: add API to control UDP tunnel ports

HW needs to know which UDP packets should be treated as tunnel
encapsulation to do inner packet recognition, classification and
offloads.

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

net/sfc/base: allow to request inner classes for Rx packets

If HW/FW supports tunnel encapsulations, Rx event may contain
either inner or outer packet classes. By default outer classes
are requested. Make it possible to request inner classes to
have more information about packet type and allow to interpret
inner frame checksum validation results correctly.

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

net/sfc/base: add function to create packed stream RxQ

Encoding packed stream buffer size in RxQ type is not a future-proof
idea taking into account a new RxQ types with extra parameters.
To be consistent make packet stream buffer size a separate parameter.
In order to avoid blowing of the default RxQ create function prototype
add a dedicated function to create packed stream RxQ without not
applicable paramters.

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

net/sfc/base: control RxQ scatter using flag instead of type

Rx scatter may be applicable to different Rx queue types.

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

net/sfc: retry port start to handle MC reboot in the middle

MC reboot may be provoked by the other function which is either
starting in parallel or, for example, reconfiguring UDP tunnel
ports.

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

net/sfc: handle MC reboot event

Implement handling of the MC reboot event received on management
event queue or detected by MCDI processing.

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

net/sfc: do not hold management event queue lock while MCDI

MCDI execution may require MCDI proxy handling which involves
management event queue polling. So, it is a bad idea to hold
managment event queue lock when MCDI is executed.

Event queue creation and destruction are MCDI operations.

Fixes: 4650ed44c120 ("net/sfc: support MCDI proxy")
Cc: stable@dpdk.org
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andy Moreton <amoreton@solarflare.com>

net/sfc: fix label name to be consistent

Management event queue is the right name of event queue 0.

Fixes: 3b809c27b1fe ("net/sfc: support link status change interrupt")
Cc: stable@dpdk.org
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Reviewed-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andy Moreton <amoreton@solarflare.com>

event/octeontx: fix Rx adapter port id mapping

When octeontx event dev receives a packet for the event Rx adapter, the
mbuf port id should contain the appropriate ethdev id instead of
internal channel info.

Fixes: 45a914c5bd71 ("event/octeontx: support event Rx adapter")

Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
Acked-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>

net/octeontx: add channel to port id mapping

The channel to port id map is used by event octeontx to map the received
wqe to the respective ethdev port.

Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
Acked-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>

doc: remove unused config options

Remove unused debug config options for thunder nicvf and octeontx as they
are now controlled by dynamic logging.

Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>

event/octeontx: implement dynamic logging

Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
Acked-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>

mempool/octeontx: implement dynamic logging

Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
Acked-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>

net/octeontx: implement dynamic logging

Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
Acked-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>

net/thunderx: implement dynamic logging

Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
Acked-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>

net/liquidio: implement dynamic logging

This driver mostly uses the common pattern. Convert this to the
dynamic logging.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>

net/qede: implement dynamic logging

This driver is mostly like others with slightly different logging
macros. The semantics were retained, with some minor reformatting.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>

net/ena: implement dynamic logging

Use dynamic rather static configuration for logging.
Also include ena_assert_msg.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>

net/fm10k: implement dynamic logging

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>

net/nfp: implement dynamic logging

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
v2: remove unsed macro RTE_LIBRTE_NFP_NET_DEBUG_INIT

net/virtio: implement dynamic logging

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Tested-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>

net/e1000: implement dynamic logging

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>

net/ixgbe: implement dynamic logging

Convert all drivers possible to dynamic logging.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>

net/vmxnet3: implement dynamic logging

Replace compile time configuration with runtime.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>

net/bnx2x: implement dynamic logging

Replace compile time option for init and driver log with
dynamic value.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
v2: remove remaining RTE_LIBRTE_BNX2X_DEBUG usage

net/avp: implement dynamic logging

All PMD should be using dynamic log levels.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>

net/tap: renamed netlink functions

Functions like nl_recev and nl_send name clash functions in the
libnl library (https://www.infradead.org/~tgr/libnl/).
All functions declared in tap_netlink.h were decorated with tap_
for consistency.

Signed-off-by: Radu Nicolau <radu.nicolau@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>

net/octeontx: advertise supported mbuf pool ops

Advertise mempool/octeontx as the only supported mempool ops when the
application checks using `rte_eth_dev_pool_ops_supported`.

Signed-off-by: Pavan Nikhilesh <pbhagavatula@caviumnetworks.com>
Acked-by: Santosh Shukla <santosh.shukla@caviumnetworks.com>

ethdev: optimize xstats by ids APIs

Introduced a check to detect if the stats IDs being
requested are all basic stats IDs. In that case,
ensured that only the basic stats would be retrieved.

Previously, both basic stats and xstats were being
retrieved even if all the IDs were basic stats IDs.

Signed-off-by: Elza Mathew <elza.mathew@intel.com>
Reviewed-by: Lee Daly <lee.daly@intel.com>

ethdev: refactor basic stats code

Moved the code to get the basic stats names and values
into static functions.

Signed-off-by: Elza Mathew <elza.mathew@intel.com>
Reviewed-by: Lee Daly <lee.daly@intel.com>

vhost: destroy unused virtqueues when multiqueue not negotiated

QEMU sends VHOST_USER_SET_VRING_CALL requests for all queues
declared in QEMU command line before the guest is started.
It has the effect in DPDK vhost-user backend to allocate vrings
for all queues declared by QEMU.

If the first driver being used does not support multiqueue,
the device never changes to VIRTIO_DEV_RUNNING state as only
the first queue pair is initialized. One driver impacted by
this bug is virtio-net's iPXE driver which does not support
VIRTIO_NET_F_MQ feature.

It is safe to destroy unused virtqueues in SET_FEATURES request
handler, as it is ensured the device is not in running state
at this stage, so virtqueues aren't being processed.

Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Laszlo Ersek <lersek@redhat.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

vhost: extract virtqueue cleaning and freeing functions

This patch extracts needed code for vhost_user.c to be able
to clean and free virtqueues unitary.

Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Laszlo Ersek <lersek@redhat.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

vhost: propagate set features handling error

Not propagating VHOST_USER_SET_FEATURES request handling
error may result in unpredictable behavior, as host and
guests features may no more be synchronized.

This patch fixes this by reporting the error to the upper
layer, which would result in the device being destroyed
and the connection with the master to be closed.

Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Laszlo Ersek <lersek@redhat.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

vhost: prevent features to be changed while device is running

As section 2.2 of the Virtio spec states about features
negotiation:
"During device initialization, the driver reads this and tells
the device the subset that it accepts. The only way to
renegotiate is to reset the device."

This patch implements a check to prevent illegal features change
while the device is running.

One exception is the VHOST_F_LOG_ALL feature bit, which is enabled
when live-migration is initiated. But this feature is not negotiated
with the Virtio driver, but directly with the Vhost master.

Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Laszlo Ersek <lersek@redhat.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

vhost: support UDP Fragmentation Offload

In virtio, UDP Fragmentation Offload (UFO) includes two parts: host UFO
and guest UFO. Guest UFO means the frontend can receive large UDP
packets, and host UFO means the backend can receive large UDP packets.
This patch supports host UFO and guest UFO for vhost-user.

Signed-off-by: Jiayu Hu <jiayu.hu@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Tested-by: Lei Yao <lei.a.yao@intel.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

examples/vhost: fix sending ARP packet to self

ARP packets are not dropped when dest vdev is itself, which breaks
RX ring inconspicuously.

Fixes: 9c5ef51207c6 ("examples/vhost: handle broadcast packet")
Cc: stable@dpdk.org
Signed-off-by: Junjie Chen <junjie.j.chen@intel.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

vhost: introduce vring call API

Users of librte_vhost currently implement the vring call operation
themselves.  Each caller performs the operation slightly differently.

This patch introduces a new librte_vhost API called
rte_vhost_vring_call() that performs the operation so that vhost-user
applications don't have to duplicate it.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

vhost: add vring call helper

Extract the callfd eventfd signal operation so virtio_net.c does not
have to repeat it multiple times.

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

net/virtio: remove redundant macro definitions for vector Rx

RTE_VIRTIO_VPMD_RX_BURST and RTE_VIRTIO_VPMD_RX_REARM_THRESH
have been defined and used in virtio_rxtx_simple.h, but are
defined again in virtio_rxtx_simple_*.c. It just happens to
work. So remove the redundant definitions from the *.c files.

Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

net/virtio: remove a redundant macro definition for ctrl vq

VIRTIO_NET_CTRL_MAC_ADDR_SET is defined two times in
virtqueue.h, the second one is obviously not wanted.

Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

net/virtio: fix typo in LRO support

Fixes: 86d59b21468a ("net/virtio: support LRO")
Fixes: ec9f3d122a58 ("net/virtio: revert not claiming LRO support")
Cc: stable@dpdk.org
Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

net/virtio: fix vector Rx flushing

The vector Rx will be broken if backend has consumed all
the descs in the avail ring before the device is started.
Because in current implementation, vector Rx will return
immediately without refilling the avail ring if the used
ring is empty. So we have to refill the avail ring after
flushing the elements in the used ring for vector Rx.

Besides, vector Rx has a different ring layout assumption
and mbuf management. So we need to handle it differently.

Fixes: d8227497ec5c ("net/virtio: flush Rx queues on start")
Cc: stable@dpdk.org
Reported-by: Antonio Fischetti <antonio.fischetti@intel.com>
Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Tested-by: Antonio Fischetti <antonio.fischetti@intel.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

net/virtio: fix incorrect cast

The rx_queues and tx_queues fields of the data structure points to a
struct virtnet_rx or virtnet_tx. Casting it to a virtqueue is an error.

It does not trigger any bug because pointer is not dereferenced inside
the function, but it can become a bug if this code is copy/pasted and
vq is dereferenced.

Fixes: 01ad44fd374f ("net/virtio: split Rx/Tx queue")
Cc: stable@dpdk.org
Signed-off-by: Didier Pallard <didier.pallard@6wind.com>
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

vhost: support Generic Segmentation Offload

In virtio, Generic Segmentation Offload (GSO) is the feature for the
backend, which means the backend can receive packets with any GSO
type.

Virtio-net enables the GSO feature by default, and vhost-net supports it.
To make live migration from vhost-net to vhost-user possible, this patch
enables GSO for vhost-user.

Signed-off-by: Jiayu Hu <jiayu.hu@intel.com>
Tested-by: Lei Yao <lei.a.yao@intel.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

net/virtio: remove unnecessary macro definitions

DPDK has already the definition of Ethernet numeric link speeds in Mbps
in the file Rte_ethdev.h, it is unnecessary to rededine virtio specific
link speeds macros again.

Signed-off-by: Zhiyong Yang <zhiyong.yang@intel.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

vhost: fix dequeue zero copy with virtio1

This fix dequeue zero copy can not work with Qemu
version >= 2.7. Since from Qemu 2.7 virtio device
use virtio-1 protocol, the zero copy code path
forget to add offset to buffer address.

Fixes: b0a985d1f340 ("vhost: add dequeue zero copy")
Cc: stable@dpdk.org
Signed-off-by: Junjie Chen <junjie.j.chen@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

vhost: fix crash

In a running VM, operations (like device attach/detach) will
trigger the QEMU to resend set_mem_table to vhost-user backend.

DPDK vhost-user handles this message rudely by unmap all existing
regions and map new ones. This might lead to segfault if there
is pmd thread just trying to touch those unmapped memory regions.

But for most cases, except VM memory hotplug, QEMU still sends the
set_mem_table message even the memory regions are not changed as
QEMU vhost-user filters out those not backed by file (fd > 0).

To fix this case, we add a check in the handler to see if the
memory regions are really changed; if not, we just keep old memory
regions.

Fixes: 8f972312b8f4 ("vhost: support vhost-user")
CC: stable@dpdk.org
Reported-by: Yang Zhang <zy107165@alibaba-inc.com>
Reported-by: Xin Long <longxin.xl@alibaba-inc.com>
Signed-off-by: Yi Yang <yi.y.yang@intel.com>
Signed-off-by: Jianfeng Tan <jianfeng.tan@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Yuanhan Liu <yliu@fridaylinux.org>

net/i40e: fix setting of MAC address on i40evf

When setting the MAC address, the ethdev layer copies the new mac
address in dev->data->mac_addrs[0] before calling the dev_ops.

Therefore, "is_same_ether_addr(mac_addr, dev->data->mac_addrs)" was
always true, and the MAC was never set. Remove this test to fix the
issue.

Fixes: 943c2d899a0c ("net/i40e: set VF MAC from VF")
Cc: stable@dpdk.org
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>

examples/flow_filtering: add Tx queues setup process

This example does not have the process to set up the Tx queues,
which is required by Intel NICs. So this patch adds that Tx setup
to the application.

Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Ori Kam <orika@mellanox.com>

net/i40e: add FDIR NVGRE parameter check

Add mask parameter check to nvgre parser for flow API.

Fixes: 30965ca34127 ("net/i40e: add NVGRE flow parsing")
Cc: stable@dpdk.org
Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>

net/ixgbe: fix tunnel filter fail problem

Add a mode type check for tunnel mode, if FDIR is in this mode,
it does not need to do sanity check for x550.

Fixes: a2ba854ec616 ("net/ixgbe: fix MAC VLAN filter fail problem")
Cc: stable@dpdk.org
Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Wenzhuo Lu <wenzhuo.lu@intel.com>

net/ixgbe: remove ipsec keys from private data

All ipsec related settings are being held in the driver
private data to allow easy add and remove of SAS. There
is no need to keep a record of the keys, and also
storing the keys can be a security issue.

Signed-off-by: Radu Nicolau <radu.nicolau@intel.com>
Acked-by: Declan Doherty <declan.doherty@intel.com>

net/i40e: support MAC loopback

According to loopback mode, setup loopback link or not.
If loopback link is set, packets transmitted will be
looped back to received side directly.

Signed-off-by: Yanglong Wu <yanglong.wu@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>

net/i40e: enable VF Tx offload

Enable VF Tx offload including outer ip checksum,
TSO and tunnel TSO.

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

net/i40e: improve packet type parser

Add support for PPPoE and L2TP packet types parser.
Change parser as new metadata 'OIPV4' and 'OIPV6'
added in profile to distinguish outer IP and inner
IP.

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

mbuf: add PPPoE and L2TP packet types

Add support of PPPoE and L2TP packet types.

Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>

net/ixgbe: add ntuple support to flow parser

ixgbe ntuple filter in rte_flow needs to support diversion data
with less than 5 tuple parameters. So it adds this new support
in parser code. This patch also adds parsering vlan pattern code
in ntuple filter in order to handle some cases like vlan in flow
API.

Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Wenzhuo Lu <wenzhuo.lu@intel.com>

net/i40e: fix FDIR input set conflict

The first FDIR rule for some PCTYPE will configure input set and
create flow, the following flows must use the same input set,
otherwise it will cause input set conflict and fail to create flow.
If it creates the first rule after flow flush, input set should be
re-configured.

Fixes: 42044b69c67d ("net/i40e: support input set selection for FDIR")
Cc: stable@dpdk.org
Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>

net/i40e: fix VLAN offload setting issue

When using kernel PF + DPDK VF, if setting VLAN strip on
or off in VF side after setting PVID for VF with ethtool in
PF side, it will cause VF Tx error. The root cause is that
Rx VLAN offload is not permitted after setting PVID for VF
in kernel PF side.
This patch fixes the issue by check VLAN offload capability
when setting VLAN offload.

Fixes: 5f0b95d59a98 ("net/i40e: support VLAN stripping for VF")
Cc: stable@dpdk.org
Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>

net/i40e: remove forward declaration

This trivial patch removes unnecessary forward declaration
of i40e_xmit_pkts_simple() in i40e_rxtx.c.

Signed-off-by: Rami Rosen <rami.rosen@intel.com>
Acked-by: Helin Zhang <helin.zhang@intel.com>

net/fm10k: remove RSS restriction with num of queues

FM10K HW does not have such restrictions.

Enabling RSS with single queue is not used to distribute flow, but
it can be used to compute a RSS hash value. It can reduce CPU
cycles of computing a hash value with five tuples. In addition,
there is an explicit method to disable RSS instead of an obscure
way.

Signed-off-by: Yangchao Zhou <zhouyates@gmail.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>

net/igb: fix Tx queue number assignment

Internal variable containing the number of TX queues for a device,
was being incorrectly assigned the number of RX queues, instead of TX.

Fixes: 27b609cbd1c6 ("ethdev: move the multi-queue mode check to specific drivers")
Cc: stable@dpdk.org
Signed-off-by: Yangchao Zhou <zhouyates@gmail.com>
Acked-by: Wei Dai <wei.dai@intel.com>

doc: add i40e limitation of DCB function

This patch adds limitation of DCB function.

Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Marko Kovacevic <marko.kovacevic@intel.com>

net/ixgbe: fix wrong PBA setting

The datasheet says, if using MSI-X mode, the PBA support
bit of the GPIE register must be set to 1.
DPDK uses polling mode, we cannot hit this issue in the
scenario DPDK PF + DPDK VF. If we use DPDK PF + kernel VF,
as the kernel driver uses interrpt mode, VF may hit RX hang
after running hours.

Fixes: 00e30184daa0 ("ixgbe: add PF support")
Cc: stable@dpdk.org
Signed-off-by: Wenzhuo Lu <wenzhuo.lu@intel.com>
Acked-by: Wei Dai <wei.dai@intel.com>