When IN_ORDER feature is negotiated, device may just write out a
single used ring entry for a batch of buffers:
"""
Some devices always use descriptors in the same order in which they
have been made available. These devices can offer the VIRTIO_F_IN_ORDER
feature. If negotiated, this knowledge allows devices to notify the
use of a batch of buffers to the driver by only writing out a single
used ring entry with the id corresponding to the head entry of the
descriptor chain describing the last buffer in the batch.
The device then skips forward in the ring according to the size of
the batch. Accordingly, it increments the used idx by the size of
the batch.
The driver needs to look up the used id and calculate the batch size
to be able to advance to where the next used ring entry will be written
by the device.
"""
Currently, the in-order Tx path in split ring can't handle this.
With this patch, driver will allocate desc_extra[] based on the
index in avail/used ring instead of the index in descriptor table.
And driver can just relay on the used->idx written by device to
reclaim the descriptors and Tx buffers.
Fixes: e5f456a98d ("net/virtio: support in-order Rx and Tx")
Cc: stable@dpdk.org
Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
We should try to cleanup at least the 'need' number of descs.
Fixes: 892dc798fa ("net/virtio: implement Tx path for packed queues")
Cc: stable@dpdk.org
Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
If the port has received less than ``pkt_per_port`` packets (for
example, the port has missed some packets), the test is in an infinite
loop.
Instead of expecting a number of packet to receive, let the port to be
drained by itself. If no more packets are received, the test can
continue.
Fixes: 002ade70e9 ("app/test: measure cycles per packet in Rx/Tx")
Cc: stable@dpdk.org
Signed-off-by: Julien Meunier <julien.meunier@nokia.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
test_alb_reply_from_client test fails due to incorrect active slave
array's index. This was due to invalid active slave count.
Count of internals->active_slave is not updated even when active slave
is deactivated.
Hence active slave count always keeps incrementing beyond the actual
active slaves.
Fix is to set the internals->active_slave to starting index 0 whenever
it exceeds the number of slaves in active slave list and also update
the active slave count during slave de-activation.
Fixes: e1110e9776 ("net/bonding: fix Rx slave fairness")
Cc: stable@dpdk.org
Signed-off-by: Hari Kumar Vemula <hari.kumarx.vemula@intel.com>
Acked-by: Radu Nicolau <radu.nicolau@intel.com>
Acked-by: Chas Williams <chas3@att.com>
For active-backup, tlb, and alb mode,
bond_ethdev_promiscuous_{enable,disable} tries to set promisc mode on
the primary port, even when there are no slaves. It is harmless, as
rte_eth_promiscuous_{enable,disable} does nothing if the port number
is invalid. But, it does print a warning message. Here is an example
from testpmd.
testpmd> create bonded device 5 0
Created new bonded device net_bonding_testpmd_0 on (port 4).
Invalid port_id=33
testpmd> set promisc 4 off
Invalid port_id=33
33 in this case is RTE_MAX_ETHPORTS + 1, the invalid primary port
number used within the bonding driver. This warning message is
harmless but can be confusing to the user. So do not try to set
promisc on a primary port when we know it does not exist (i.e. no
slaves).
Fixes: 2efb58cbab ("bond: new link bonding library")
Cc: stable@dpdk.org
Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
Acked-by: Chas Williams <chas3@att.com>
This patch fixes a regression introduced by
commit 49d3978d57 ("net/qede: fix Tx tunnel offload support mask")
in which qede_xmit_prep_pkts() breaks the loop for successful
check of Tunneling offload flags instead of continuing, resulting
in tx_pkt_prepare return a failure.
Fixes: 49d3978d57 ("net/qede: fix Tx tunnel offload support mask")
Cc: stable@dpdk.org
Signed-off-by: Shahed Shaikh <shshaikh@marvell.com>
When removing the old attach function, the racy variable for getting
the last port id became unused.
Fixes: c9cce42876 ("ethdev: remove deprecated attach/detach functions")
Cc: stable@dpdk.org
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Andrew Rybchenko <arybchenko@solarflare.com>
Whenever the app is calling rte_eth_dev_info_get(), it shouldn't use the
admin command. It was causing problems, if it was called from the
secondary process - the main process was crashing, and the secondary app
wasn't getting any result, as the admin request couldn't be processed by
the process which was requesting the data.
To fix that, the data is being written to the adapter structure during
device initialization routine.
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Copying the link properties of the first slave added may copy an
invalid link status. The speed and duplex of the slave may not
be known at this time. Delay setting the properties until the
first slave reports as link up. Note that we are still ignoring
an error from link_properties_valid. For some bonding modes,
802.3ad, we should not activate the slave if it does not have
matching link properties.
Fixes: a45b288ef2 ("bond: support link status polling")
Cc: stable@dpdk.org
Signed-off-by: Chas Williams <chas3@att.com>
The Tx burst routine always needs to check for pending LACPDUs
and send them if available. Do this first to prioritize the
control traffic. We can still early exit, before calculating
the distribution slaves, if there isn't any data packets.
Fixes: 09150784a7 ("net/bonding: burst mode hash calculation")
Cc: stable@dpdk.org
Reported-by: Hui Zhao <zhaohui8@huawei.com>
Signed-off-by: Chas Williams <chas3@att.com>
VLAN filter is required to be configured during dev_start according to
dev_conf.rxmod.offloads setting, so ice_vlan_offload_set is called to
handle this.
Fixes: 690175ee51 ("net/ice: support getting device information")
Cc: stable@dpdk.org
Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Enable CLEARPBA bit is required by ice NIC of A0/A1 version to
enable admin queue interrupt.
Also enable CLEARPBA bit does no impact on AVF behaviour when be
hosted by other devices, so we can make it as default.
Cc: stable@dpdk.org
Signed-off-by: Wei Zhao <wei.zhao1@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Set RTE_PCI_DRV_IOVA_AS_VA in drv_flags: this allows initializing bnxt
PMD as non-root also on Linux v4.x, where /proc/self/pagemap can't be
accessed without CAP_SYS_ADMIN privileges.
Cc: stable@dpdk.org
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Acked-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
We can transmit if there is at least one distributing slave.
Fixes: 09150784a7 ("net/bonding: burst mode hash calculation")
Cc: stable@dpdk.org
Signed-off-by: Chas Williams <chas3@att.com>
Fix segment fault when detach a device due to some redundant function
call in ice_dev_uninit.
Fixes: f9cf4f8641 ("net/ice: support device initialization")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Wenzhuo Lu <wenzhuo.lu@intel.com>
The API function rte_eth_dev_fw_version_get() is querying drivers
via the operation callback fw_version_get().
The implementation of this operation is added for mlx4 and mlx5.
Both functions are copying the same ibverbs field fw_ver
which is retrieved when calling ibv_query_device[_ex]()
during the port probing.
It is tested with command "drvinfo" of examples/ethtool/.
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Shahaf Shuler <shahafs@mellanox.com>
Original patch implemented the use of match_metadata offload in the
different burst functions.
The concurrent use of match_metadata and multi_segs offloads was
not handled.
This patch updates function txq_scatter_v(), to pass metadata value
from mbuf to wqe, when indicated by offload flags.
Fixes: 6bd7fbd03c ("net/mlx5: support metadata as flow rule criteria")
Cc: stable@dpdk.org
Signed-off-by: Dekel Peled <dekelp@mellanox.com>
Acked-by: Yongseok Koh <yskoh@mellanox.com>
This patch fixes a broken build; when running
make test
we get the following error:
...
Traceback (most recent call last):
File "test/test/autotest.py", line 49, in <module>
num_fails = runner.run_all_tests()
File "test/test/autotest_runner.py", line 345, in run_all_tests
self.parallel_tests)
File "test/test/autotest_runner.py", line 264, in __filter_test
test_cmd = test["Command"]
KeyError: 'Command'
...
The reason is that the pdump autotest entry uses "Comamnd"
instead of "Command"; this patch fixes it.
Bugzilla ID: 205
Fixes: 086eb64db3 ("test/pdump: add unit test for pdump library")
Cc: stable@dpdk.org
Signed-off-by: Rami Rosen <ramirose@gmail.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Add the strlcat function to DPDK to exist alongside the strlcpy one.
While strncat is generally safe for use for concatenation, the API for the
strlcat function is perhaps a little nicer to use, and supports truncation
detection.
See commit 5364de644a ("eal: support strlcpy function") for more
details on the function selection logic, since we only should be using the
DPDK-provided version when no system-provided version is present.
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
External message callbacks are used e.g. by vhost crypto
to parse crypto-specific vhost-user messages.
We are now publishing the API to register those callbacks,
so that other backends outside of DPDK can use them as well.
Signed-off-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
This minor cleanup patch removes an unnecessary forward
declaration of virtio_intr_enable() in net/virtio PMD.
Fixes: fe19d49cb5 ("net/virtio: fix Rx interrupt with VFIO")
Cc: stable@dpdk.org
Signed-off-by: Rami Rosen <ramirose@gmail.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
We don't need to relay available ring and check the desc, vdpa device
can access the available ring in the guest directly. With this patch,
we can achieve better throughput and lower CPU usage.
Signed-off-by: Xiao Wang <xiao.w.wang@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Port configuration fails because offload flags don't match the expected
value when max-pkt-len is set to a value that should enable receive port
offloading but doesn't.
The .dev_infos_get callback can be called before the configure callback.
At that time we don't know the maximum packet size yet because it is
only set up when ports are started. So in virtio_dev_info_get() just
always set the jumbo packet offload flag.
Check the maximum packet length at device configure time, because then we
have access to the max-pkt-len value provided by the user. If the
max-pkt-len exceeds the maximum MTU supported by the device we remove
the VIRTIO_NET_F_MTU flag from requested features.
Fixes: a4996bd89c ("ethdev: new Rx/Tx offloads API")
Cc: stable@dpdk.org
Signed-off-by: Jens Freimann <jfreimann@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Read barrier is required between reading the flags (desc_is_used)
and the content of descriptor to ensure the ordering.
Otherwise, speculative read of desc.id could be reordered with
reading of the desc.flags.
Fixes: a76290c8f1 ("net/virtio: implement Rx path for packed queues")
Cc: stable@dpdk.org
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Reviewed-by: Jens Freimann <jfreimann@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
There should be read barrier between checking VIRTQUEUE_NUSED (reading
the used->idx) and reading these descriptors. It's done for the first
checks at the beginning of these functions but missed while checking
for extra required descriptors.
Fixes: e5f456a98d ("net/virtio: support in-order Rx and Tx")
Fixes: 13ce5e7eb9 ("virtio: mergeable buffers")
Cc: stable@dpdk.org
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Read barrier must be implied between reading descriptor flags
and descriptor id. Otherwise, in case of reordering, we could
read wrong descriptor id.
For the reference, similar barrier for split rings is the read
barrier between VIRTQUEUE_NUSED (reading the used->idx) and
the call to the virtio_xmit_cleanup().
Additionally removed double update of 'used_idx'. It's enough
to set it in the end of the loop.
Fixes: 892dc798fa ("net/virtio: implement Tx path for packed queues")
Cc: stable@dpdk.org
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Reviewed-by: Jens Freimann <jfreimann@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Move duplicate macro to increment per-event type stats to
internal header. These stats are NIC family independent and
stored in generic event queue structure.
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Include efx_mcdi.h from main internal header efx_impl.h directly
instead of indirect inclusion via family-specific header.
It avoids duplication and fixes Medford and Medford2 cases where
the header inclusion is lost.
Correctness is still guaranteed by checks in efx_check.h which
require EFSYS_OPT_MCDI for corresponding families and do not
allow to enable the option if no family requires it.
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
If event queue timers are not supported, enc_evq_timer_quantum_ns and
enc_evq_timer_max_us should be set to 0. Make sure that division by 0
does not happen in libefx, if public function efx_ev_usecs_to_ticks()
is used in this case.
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Add EFX_FAMILY_IS_EF10() which may be used as synonym of any EF10-based
NIC family (i.e. Huntigton, Medford or Medford2 at the time of addition).
Suggested-by: Andy Moreton <amoreton@solarflare.com>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Add EFX_OPTS_EF10() which may be used as synonym of any EF10-based
NIC (i.e. Huntigton, Medford or Medford2 at the time of addition).
Suggested-by: Andy Moreton <amoreton@solarflare.com>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Riverhead requires Rx buffer size to be specified in INIT_RXQ.
If the parameter is not supported (e.g. on EF10 family adapters),
zero must be used on INIT_RXQ.
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Supports the firmware update mechanism described in SF-121352-AN.
Signed-off-by: Richard Houldsworth <rhouldsworth@solarflare.com>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
The recently added NO_CONT_EV mode is recommended for when looking for
maximum throughput on 100G links as it allows the firmware to operate
more efficiently. The biggest benefit is when using scatter and jumbo
frames, but it is also necessary to achieve line rate in other cases.
Support for NO_CONT_EV when scatter is disabled is simple - the main
datapath change is to always read the packet length from the prefix,
not the event. This requires storing a flag with the event queue so
the event handler knows NO_CONT_EV mode is in use.
Supporting NO_CONT_EV with scatter would require an API change.
(Now the ee_flags field in efx_evq_t is used on the datapath, move it
next to other fields which are also read on the datapath to try to avoid
reading an additional cache line.)
Signed-off-by: Mark Spender <mspender@solarflare.com>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
The macros are deprecated and are not used anymore, so they can
be deleted.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Symbols for maximum and minimum number of Tx, Rx and event descriptors
are deprecated. They are not used anymore, so they can be deleted.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Use of the macro for calculating a Rx queue size is deprecated. Replace
it with a call to a function that uses descriptor size specified for
every NIC.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Use of the macro for calculating a Rx queue size is deprecated. Replace
it with a call to a function that uses descriptor size specified for
every NIC.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Use of the macro for calculating a Tx queue size is deprecated. Replace
it with a call to a function that uses descriptor size specified for
every NIC.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Functions efx_*q_nbufs return the number of a queue buffers by dividing
the queue size by page size. If minimum size of a queue is smaller than
one page, we still need the page and number of buffers should not be 0.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
For consistency with the size of Tx descriptors, the size of event
descriptors should be a part of NIC config, not a macro that is
common for all NIC families. Also, add a max number of EvQ buffers
define which is needed to initialize an MCDI buffer at compile time.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
For consistency with the size of Tx descriptors, the size of Rx
descriptors should be a part of NIC config, not a macro that is
common for all NIC families.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
Size of Tx descriptor is different on Riverhead. So, the size
should be a part of NIC config, not a macro that is common for
all NIC families.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
The change is needed for the functions that accept const efx_nic_t
pointer and have to get efx_nic_cfg_t.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <arybchenko@solarflare.com>
