Add support for SFF-8024 Rev 4.0 of pluggable I/O configuration
and some common utilities for SFF-8436/8636 and SFF-8472/8079.
Signed-off-by: Robin Zhang <robinx.zhang@intel.com>
Signed-off-by: Kevin Liu <kevinx.liu@intel.com>
Acked-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Add a new telemetry command /ethdev/module_eeprom to dump the module
EEPROM of each port. The format of module EEPROM information follows
the SFF(Small Form Factor) Committee specifications.
Signed-off-by: Robin Zhang <robinx.zhang@intel.com>
Signed-off-by: Kevin Liu <kevinx.liu@intel.com>
Acked-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
The packet type is set even if it is a fragmented packet.
Tested, using testpmd, for different hardware versions on ESXi 7.0
Update 2.
Signed-off-by: Pankaj Gupta <pagupta@vmware.com>
Reviewed-by: Jochen Behrens <jbehrens@vmware.com>
Currently the driver assumes that the size of the RETA table is
four times the number of Rx queues at multiple places. However,
it sets it to four times the maximum number of queues (4 * 32 = 128)
when it first initializes the device. Change the other uses to use
the stored value, not the assumed default.
Tested, using testpmd, for different hardware versions on ESXi 7.0
Update 2.
Signed-off-by: Pankaj Gupta <pagupta@vmware.com>
Reviewed-by: Jochen Behrens <jbehrens@vmware.com>
VMXNET3 version 6 supports some new features, including but not limited to:
- Increased maximum MTU up to 9190
- Increased maximum number of Rx and Tx queues
- Removes power-of-two limitations on queue sizes
- Extended interrupt structures (required for additional queues)
Tested, using testpmd, for different hardware versions on ESXi 7.0
Update 2.
Signed-off-by: Pankaj Gupta <pagupta@vmware.com>
Reviewed-by: Jochen Behrens <jbehrens@vmware.com>
Support rte_eth_dev_fw_version_get() API.
Tested, using testpmd, for different hardware versions on ESXi 7.0
Update 2.
Signed-off-by: Pankaj Gupta <pagupta@vmware.com>
Reviewed-by: Jochen Behrens <jbehrens@vmware.com>
Signed-off-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Driver callbacks initialization should follow callbacks order in the
structure definition.
Do not reorder callbacks implementation to avoid extra noise in
the code.
Signed-off-by: Pankaj Gupta <pagupta@vmware.com>
Signed-off-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Count the number of entries in the Rx queue for debugging.
Tested, using testpmd, for different hardware versions on ESXi 7.0
Update 2.
Signed-off-by: Pankaj Gupta <pagupta@vmware.com>
Reviewed-by: Jochen Behrens <jbehrens@vmware.com>
Add RETA query and RETA update support for VMXNET3.
Tested, using testpmd, for different hardware versions on ESXi 7.0
Update 2.
Signed-off-by: Pankaj Gupta <pagupta@vmware.com>
Reviewed-by: Jochen Behrens <jbehrens@vmware.com>
The help message of create meter command is incomplete,
lack of policy_id param, update help string.
Fixes: f29fa2c59b ("app/testpmd: support policy actions per color")
Cc: stable@dpdk.org
Signed-off-by: Jin Liu <jin.liu@corigine.com>
Signed-off-by: Chaoyong He <chaoyong.he@corigine.com>
Signed-off-by: Niklas Söderlund <niklas.soderlund@corigine.com>
Acked-by: Ferruh Yigit <ferruh.yigit@xilinx.com>
"vlan set strip off 0" can't work, due to the incorrect usage
of the mask, it just represent that the status is changed or not,
not represent offloaded or not, so that the driver send the
error control message to the nic.
Now, by first inspect the mask of things that changed, and then
change the requested state if VLAN stripping according the
requested offload status. So this change can fix this bug.
Fixes: d4a27a3b09 ("nfp: add basic features")
Cc: stable@dpdk.org
Signed-off-by: Peng Zhang <peng.zhang@corigine.com>
Signed-off-by: Yong Xu <yong.xu@corigine.com>
Signed-off-by: Chaoyong He <chaoyong.he@corigine.com>
Signed-off-by: Niklas Söderlund <niklas.soderlund@corigine.com>
Hardware restrictions require a maximum of 4 queues for every pool.
Fixes: a6712cd029 ("net/txgbe: add PF module init and uninit for SRIOV")
Cc: stable@dpdk.org
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Add to read efuse values from flash, and disable EEE to improve signal
quality. Remove PHY semaphore to access PHY registers faster. And remove
unnecessary page selection where quick access is required.
When rte_eth_link_get_nowait() is called frequently with LSC disabled by
self-developed applications, eventually the PHY status register will be
accessed frequently. It will cause internal PHY init failure, if they
are done simultaneously. So there is a protection added for internal PHY
init.
Fixes: 3518df5774 ("net/ngbe: support device start/stop")
Fixes: 91bc12c522 ("net/ngbe: optimize PHY initialization process")
Cc: stable@dpdk.org
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
When using mailbox to request firmware to enable or disable PCIe bus
master, there is a small probability that mailbox cannot respond.
Change to use rte_pci_read_config() and rte_pci_write_config(), to
avoid this problem.
Fixes: ac6c5e9af5 ("net/ngbe: fix Tx hang on queue disable")
Cc: stable@dpdk.org
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
For M88E1512 PHY mixed mode, PXE driver overrides PHY mode at load
time. To workaround this problem, change to read PHY mode from flash
instead of register.
Fixes: 1c44384fce ("net/ngbe: support custom PHY interfaces")
Cc: stable@dpdk.org
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
External PHY cannot power down after LAN reset, so need to manually
power down when device stopped.
Fixes: 3d0af70667 ("net/ngbe: setup PHY link")
Fixes: 1c44384fce ("net/ngbe: support custom PHY interfaces")
Cc: stable@dpdk.org
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Change to check low ID register to determine the valid PHY address,
for yt8521s PHY with high ID register value 0. And fix polling
register when expect value is 0, to complete MDIO read.
Fixes: 44e97550ca ("net/ngbe: identify and reset PHY")
Cc: stable@dpdk.org
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Since the bit of ETH_LINK_SPEED_FIXED was set for the force link
speed, it conflicts with '~allowed_speeds'.
Fixes: 3518df5774 ("net/ngbe: support device start/stop")
Cc: stable@dpdk.org
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Currently, 'dev_started' is always set to be 0 when dev stop, whether
it succeeded or failed. This is unreasonable and this patch fixed it.
Fixes: 62024eb827 ("ethdev: change stop operation callback to return int")
Cc: stable@dpdk.org
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Acked-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Ferruh Yigit <ferruh.yigit@xilinx.com>
The 'dst_off' was reset in multi segment case.
This caused memif buffer segment to write to
beginning of mbuf, overwriting previous data.
Fix it with this patch.
Fixes: 09c7e63a71 ("net/memif: introduce memory interface PMD")
Cc: stable@dpdk.org
Reported-by: Ferruh Yigit <ferruh.yigit@xilinx.com>
Signed-off-by: Joyce Kong <joyce.kong@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Usage of 'RTE_ETH_TX_OFFLOAD_MBUF_FAST_FREE' offload has two
constraints: per-queue all mbufs comes from the same mempool and
has refcnt = 1.
Bonding mode Broadcast, Tx mbuf has more than one refcnt.
Bonding mode 8023AD, It contains two mempools separately for LACP
packets and other packets. In Tx or Rx, Fast mbuf free will operate
mbuf from different mempool.
This patch will prevent 'RTE_ETH_TX_OFFLOAD_MBUF_FAST_FREE' offload
when in bonding mode Broadcast and mode 8023AD.
Fixes: 78aecefed9 ("bond: move param parsing in configure step")
Cc: stable@dpdk.org
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Calling Rx/Tx functions on a stopped queue is not supported.
Do not run packet forwarding for streams that use stopped queues.
Each stream has a read-only "disabled" field,
so that lcore function can skip such streams.
Forwarding engines can set this field
using a new "stream_init" callback function
by checking relevant queue states,
which are stored along with queue configurations
(not all PMDs implement rte_eth_rx/tx_queue_info_get()
to query the state from there).
Fixes: 5f4ec54f1d ("testpmd: queue start and stop")
Cc: stable@dpdk.org
Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
When a port was closed, indirect actions could remain
with their handles no longer valid.
If a newly attached device was assigned the same ID as the closed port,
those indirect actions became accessible again.
Any attempt to use them resulted in an undefined behavior.
Automatically flush indirect actions when a port is closed.
Fixes: 4b61b8774b ("ethdev: introduce indirect flow action")
Cc: stable@dpdk.org
Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Acked-by: Aman Singh <aman.deep.singh@intel.com>
Whether it is allowed to call Rx/Tx functions for a stopped queue
was undocumented. Some PMDs make this behavior a no-op
either by explicitly checking the queue state
or by the way how their routines are implemented or HW works.
No-op behavior may be convenient for application developers.
But it also means that pollers of stopped queues
would go all the way down to PMD Rx/Tx routines, wasting cycles.
Some PMDs would do a check for the queue state on data path,
even though it may never be needed for a particular application.
Also, use cases for stopping queues or starting them deferred
do not logically require polling stopped queues.
Use case 1: a secondary that was polling the queue has crashed,
the primary is doing a recovery to free all mbufs.
By definition the queue to be restarted is not polled.
Use case 2: deferred queue start or queue reconfiguration.
The polling thread must be synchronized anyway,
because queue start and stop are non-atomic.
Prohibit calling Rx/Tx functions on stopped queues.
Fixes: 0748be2cf9 ("ethdev: queue start and stop")
Cc: stable@dpdk.org
Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
Acked-by: Tyler Retzlaff <roretzla@linux.microsoft.com>
Add macro MIN_TOTAL_NUM_MBUFS (1024) to indicate
what the value of total-num-mbufs should bigger than.
Fixes: c87988187f ("app/testpmd: add --total-num-mbufs option")
Cc: stable@dpdk.org
Signed-off-by: Mingxia Liu <mingxia.liu@intel.com>
Acked-by: Yuying Zhang <yuying.zhang@intel.com>
Acked-by: Ferruh Yigit <ferruh.yigit@xilinx.com>
The return value of rte_tel_data_alloc() may be null pointer.
Add missing check vs null.
Fixes: 5ea354a1f2 ("net/cnxk: support telemetry")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Acked-by: Morten Brørup <mb@smartsharesystems.com>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
The rte_tel_data_alloc() may return NULL, so the caller should add
judgement for it.
Fixes: 083b0b310b ("ethdev: add common stats for telemetry")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Acked-by: Morten Brørup <mb@smartsharesystems.com>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Many user (e.g. telemetry) invokes rte_eth_xstats_get(port_id, NULL, 0)
to retrieve the required number of elements, but currently axgbe PMD
returns zero when xstats is null.
Remove the logic of "return zero when xstats is NULL", and add the logic
of "return the required number of entries when n is lower than the
required number of entries".
Fixes: 9d1ef6b2e7 ("net/axgbe: add xstats")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Acked-by: Morten Brørup <mb@smartsharesystems.com>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Many user (e.g. telemetry) invokes rte_eth_xstats_get(port_id, NULL, 0)
to retrieve the required number of elements, but currently mvpp2 PMD
returns zero when xstats is null.
Remove the logic of "return zero when xstats is NULL", and add the logic
of "return the required number of entries when n is lower than the
required number of entries".
Fixes: a77b5378cd ("net/mrvl: add extended statistics")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Acked-by: Morten Brørup <mb@smartsharesystems.com>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Many user (e.g. telemetry) invokes rte_eth_xstats_get(port_id, NULL, 0)
to retrieve the required number of elements, but currently ipn3ke PMD
returns zero when xstats is null.
Dedicated check for xstats vs null is not required, since ethdev layer
guarantees that it may be null only if number of entries n is 0 (which
is definitely smaller than total xstats count).
Fixes: 5a6d883878 ("net/ipn3ke: implement statistics")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Acked-by: Morten Brørup <mb@smartsharesystems.com>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Many user (e.g. telemetry) invokes rte_eth_xstats_get(port_id, NULL, 0)
to retrieve the required number of elements, but currently hns3 PMD
returns zero when xstats is null.
Dedicated check for xstats vs null is not required, since ethdev layer
guarantees that it may be null only if number of entries n is 0 (which
is definitely smaller than total xstats count).
Fixes: 8839c5e202 ("net/hns3: support device stats")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Acked-by: Morten Brørup <mb@smartsharesystems.com>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
When xstats location is null in rte_eth_xstats_get() the return value
is not clearly specified. Some PMDs (eg. hns3/ipn3ke/mvpp2/axgbe) return
zero while others return the required number of elements.
In this patch, special parameter combinations are restricted:
1. highlight that xstats location may be null if and only if n is 0.
2. amend n parameter description to specify that if n is lower than
the required number of elements, the function returns the required
number of elements.
3. specify that if n is zero, the xstats must be NULL, the function
returns the required number of elements (a duplicate which should
help to not very attentive readers).
Add sanity check for null xstats and non-zero n case on API level to
make it unnecessary to care about it in drivers.
Fixes: ce757f5c9a ("ethdev: new method to retrieve extended statistics")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Acked-by: Morten Brørup <mb@smartsharesystems.com>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Parameters to this static helper can't be NULL.
str has already been dereferenced in caller.
dst and size point to variable in stack.
Fixes: 169a9fed1f ("app/testpmd: fix hex string parser support for flow API")
Cc: stable@dpdk.org
Signed-off-by: David Marchand <david.marchand@redhat.com>
The GRO/GSO library doesn't re-calculate checksums for
merged/fragmented packets. If users want the packets to
have correct IP checksums, they should select HW IP
checksum calculation for the port which the packets are
transmitted to. But if the port doesn't support HW IP
checksum, users may perform a SW IP checksum.
Fixes: b7091f1dcf ("app/testpmd: enable the heavyweight mode TCP/IPv4 GRO")
Fixes: 52f38a2055 ("app/testpmd: enable TCP/IPv4 VxLAN and GRE GSO")
Cc: stable@dpdk.org
Signed-off-by: Wenwu Ma <wenwux.ma@intel.com>
Reviewed-by: Jiayu Hu <jiayu.hu@intel.com>
Tested-by: Wei Ling <weix.ling@intel.com>
Acked-by: Yuying Zhang <yuying.zhang@intel.com>
Change ambiguous terms "jump rule" and "group rule" to
clearer "tunnel rule" and "switch rule". The new terms
reflect the purpose of these rules in virtual switches.
The module name, "flow tunnel", is replaced by "FT" in
function names to avoid the use of word "tunnel" twice.
Use term "FT context" when referring to tunnel entries.
Also, add "ctx" suffix to "ft" and "ft_id" occurrences.
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Acked-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Starting or stopping a bonded port also starts or stops all active slaves
under the bonded port. If this port is a bonded device, we need to modify
the port status of all slaves.
Fixes: 0e545d3047 ("app/testpmd: check stopping port is not in bonding")
Cc: stable@dpdk.org
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Acked-by: Aman Singh <aman.deep.singh@intel.com>
Acked-by: Konstantin Ananyev <konstantin.v.ananyev@yandex.ru>
Setting a MTU larger than mbuf size is not supported by the device but
not prohibited by the driver. This change adds a restriction to the
driver to prevent setting an MTU that is too large.
While at it define the minimum MTU in the device information to describe
the complete supported MTU range.
Fixes: d4a27a3 ("nfp: add basic features")
Cc: stable@dpdk.org
Signed-off-by: Peng Zhang <peng.zhang@corigine.com>
Signed-off-by: Chaoyong He <chaoyong.he@corigine.com>
Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Niklas Söderlund <niklas.soderlund@corigine.com>
Reviewed-by: Walter Heymans <walter.heymans@corigine.com>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Currently, meter object supports only DSCP based on input color table,
The patch enhance that to support VLAN based input color table,
color table based on inner field for the tunnel use case, and
support for fallback color per meter if packet based on a different field.
All of the above features are exposed through capability and added
additional capability to specify the implementation supports
more than one input color table per ethdev port.
Suggested-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
Signed-off-by: Jerin Jacob <jerinj@marvell.com>
Acked-by: Ray Kinsella <mdr@ashroe.eu>
Acked-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
Acked-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Add support to read firmware configuration file from
/lib/firmware/cxgb4/ path in the filesystem. The firmware
config file is used to enable or disable NIC features before
firmware initialization to help retrieve better debug data to
analyze firmware init failures. The config file can also
be used to redistribute resources, like queues, TCAMs, etc.,
from disabled physical functions (PFs) to main PF, before
firmware init.
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
Transport Processor (TP) on the NIC delivers the incoming packets
from the wire to NIC's DMA engine to place the packets in Rx buffers.
TP sends signal towards the Multi-Port Switch (MPS) near the MAC when
the Rxqs run out of Rx buffers posted by driver. These MPS buffer drop
stats are already accounted for in imissed counters. However, if a
large number of Rxqs run out of Rx buffers simultaneously, then the
TP can start dropping packets by itself when there is heavy congestion
on the channel and hence could not inform to the MPS. So, track these
packets dropped by TP in imissed counters. Also add xstats for these
counters.
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
Add CLI options to l3fwd_power to utilize the new power APIs introduced in
this patchset. These CLI options allow the user to configure the
heuritstics made available through the new API via the l3fwd_power
application options.
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>
Add new get/set API to allow the user or application to set the minimum
and maximum frequencies to use when scaling.
Previously, the frequency range was determined by the HW capabilities of
the CPU. With this new API, the user or application can constrain this
if required.
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Acked-by: Ray Kinsella <mdr@ashroe.eu>
Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>
Tested-by: David Hunt <david.hunt@intel.com>
Add new get/set API for configuring 'pause_duration' which used to adjust
the pause mode callback duration.
Signed-off-by: Kevin Laatz <kevin.laatz@intel.com>
Acked-by: Ray Kinsella <mdr@ashroe.eu>
Acked-by: Anatoly Burakov <anatoly.burakov@intel.com>
Tested-by: David Hunt <david.hunt@intel.com>
