The multi-process has the following problems:
1) After a port in primary process is closed, the mp action of the
process is unregistered. Which will cause that other device in the
primary process cannot respond to requests from secondary processes.
2) Because variable "hns3_inited" is set to true without returning an
initial value, the mp action cannot be registered again after it is
unregistered.
3) The mp action of primary and secondary process need to be registered
only once regardless of port numbers in the process. That's what
variable "hns3_inited" does. But the variable is difficult to
understand.
This patch adds a hns3_process_local_data structure to resolve above
problems.
Fixes: 9570b1fdbd ("net/hns3: check multi-process action register result")
Fixes: 23d4b61fee ("net/hns3: support multiple process")
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
The "secondary_cnt" will be increased when a secondary process
initialized. But the value of this variable is not decreased when the
secondary process exits, which causes the primary process senses that
the secondary process still exists. As a result, the primary process
fails to send messages to the secondary process after the secondary
process exits.
Fixes: 23d4b61fee ("net/hns3: support multiple process")
Cc: stable@dpdk.org
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
It's possible that a switch rule can't be redirect successfully due
to kernel driver is busy to handle an ongoing VF reset, so the
redirect action need to be deferred into next redirect request which
is promised by kernel driver after VF reset done.
This patch uses the saved flow rule's data to replay switch rule
remove/add during next flow redirect.
Fixes: 397b4b3c50 ("net/ice: enable flow redirect on switch")
Cc: stable@dpdk.org
Signed-off-by: Dapeng Yu <dapengx.yu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
The VSI number, lookup elements and rule information for creating switch
filter are abandoned when switch filter is created in original
implementation.
This patch saved the abandoned data in RTE flow, it is for future
use on replay when handling exception at flow redirect.
Cc: stable@dpdk.org
Signed-off-by: Dapeng Yu <dapengx.yu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
The order of flow filter parser list was not definite and
linked to the register order of parsers. It caused ACL filter
covered by switch filter in some cases.
This patch fixed order of parser list to guarantee the usage
of each filter. Below lists the order.
ACL filter > Switch filter > FDIR > Hash filter.
Fixes: e4a0a7599d ("net/ice: fix flow priority support in non-pipeline mode")
Cc: stable@dpdk.org
Signed-off-by: Yuying Zhang <yuying.zhang@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
The poll call can return POLLERR which is ignored, or it can return
POLLOUT, even if there are no free frames in the mmap-ed area.
We can account for both of these cases by re-checking if the next
frame is empty before writing into it.
We have attempted to reproduce this issue with pktgen-dpdk, using the
following configuration.
pktgen -l 1-4 -n 4 --proc-type=primary --no-pci --no-telemetry \
--no-huge -m 512 \
--vdev=net_af_packet0,iface=eth1,blocksz=16384,framesz=8192, \
framecnt=2048,qpairs=1,qdisc_bypass=0 \
-- \
-P \
-T \
-m "3.0" \
-f themes/black-yellow.theme
We configure a low tx rate (~ 335 packets / second) and a small
packet size, of about 300 Bytes from the pktgen CLI.
set 0 size 300
set 0 rate 0.008
set 0 burst 1
start 0
After bringing the interface down, and up again, we seem to arrive
in a state in which the tx rate is inconsistent, and does not recover.
ifconfig eth1 down; sleep 7; ifconfig eth1 up
[1] http://code.dpdk.org/pktgen-dpdk/pktgen-20.11.2/source/INSTALL.md
Fixes: 364e08f2bb ("af_packet: add PMD for AF_PACKET-based virtual devices")
Cc: stable@dpdk.org
Signed-off-by: Mihai Pogonaru <pogonarumihai@gmail.com>
Signed-off-by: Tudor Cornea <tudor.cornea@gmail.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Add the device and vendor numbers to the PCI ID map so
that a VF can be probed.
Signed-off-by: Igor Romanov <igor.romanov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
The GTP, GTP-U, GTP-C header fields can be matched, however NIC does not
support GTP tunneling so no items after the GTP header can be specified.
If a GTP-U or GTP-C item is specified without a preceding UDP item, the
UDP destination port is implicitly matched. For GTP, the destination UDP
port must be specified but its value is not enforced.
Signed-off-by: John Daley <johndale@cisco.com>
Reviewed-by: Hyong Youb Kim <hyonkim@cisco.com>
Enable protocol agnostic flow offloading to support raw pattern input
for RSS hash flow rule creation. It is based on Parser Library feature.
Current rte_flow raw API is utilized.
Signed-off-by: Ting Xu <ting.xu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Based on the parser library, we can directly set HW profile and
associate VSI for RSS raw flows. Add symmetric hash configuration
for raw flow.
Signed-off-by: Ting Xu <ting.xu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Probing the availability of Flow Manager API may print the following
error log.
PMD: rte_enic_pmd: Devcmd 88 failed with error code -1
The error indicates a flow manager operation failed and happens when
advanced filtering is disabled on vNIC. It is harmless but confusing
to the user. Since advanced filtering is a prerequisite, check first
if it is available and avoid the error message altogether.
Fixes: ea7768b5bb ("net/enic: add flow implementation based on Flow Manager API")
Cc: stable@dpdk.org
Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
Reviewed-by: John Daley <johndale@cisco.com>
Changing MTU after the device start causes a segfault in the Rx
handler. The MTU handler (enic_set_mtu) performs the following steps.
1. Stop NIC Rx
2. Change Rx handler '(struct rte_eth_dev)->rx_pkt_burst' to
the dummy handler and sleep a while to quiesce
3. Re-allocate/initialize Rx structures
4. Change Rx handler back to the real handler
(e.g. enic_noscatter_recv_pkts)
enic_set_mtu does not update the recently introduced fast-path pointer
'(struct rte_eth_fp_ops)->rx_pkt_burst'. Since rte_eth_rx_burst only
uses the fast-path pointer, it keeps invoking the real Rx handler, not
the dummy one set by (2). And, (3) causes a segfault in the real Rx
handler (e.g. dereferencing freed structures).
Fix the segfault by updating the fast-path pointer as well.
Fixes: c87d435a4d ("ethdev: copy fast-path API into separate structure")
Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
Reviewed-by: John Daley <johndale@cisco.com>
Generally it is good practice to include all headers that provide APIs
which are being used. This is especially true in situations where 3rd
party apps include our public headers and assume that all should work
out of the box.
Including all headers explicitly helps to achieve that.
Signed-off-by: Tomasz Duszynski <tduszynski@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Device naming might be misleading which is especially true if one takes
it from lspci output. In order to keep naming consistent keep leading
zero in front of pci bus number.
Signed-off-by: Tomasz Duszynski <tduszynski@marvell.com>
Reviewed-by: Jakub Palider <jpalider@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Bandwidth profile count is updated in meter capabilities during device
initialization using mbox interface.
Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
This patch resolves problem with internal header
inclusion. In addition prevents C++ name mangling.
Signed-off-by: Jakub Palider <jpalider@marvell.com>
Reviewed-by: Tomasz Duszynski <tduszynski@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Handling the case where number of MSIX interrupts are greater
than default value i.e. PLT_MAX_RXTX_INTR_VEC_ID. On PCI probe
device is queried for supported MSIX interrupts, and respective
interrupt resources are reallocated with this value. Same MSIX
count should be used while registering new interrupt vectors.
Fixes: 8cb5d08db9 ("interrupts: extend event list")
Signed-off-by: Harman Kalra <hkalra@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Fix a few typos.
Fixes: fa8f86a14e ("common/cnxk: add build infrastructre and HW definition")
Fixes: f6d567b03d ("common/cnxk: support NIX IRQ")
Fixes: 5e076b609f ("common/cnxk: add SE set key for crypto")
Cc: stable@dpdk.org
Signed-off-by: Tomasz Duszynski <tduszynski@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
To support lower pps in packet mode we are changing adjust value,
same needs to be consider for burst size calculations.
When both peak and committed rates requested, then peak rate should
be larger than committed rate.
Signed-off-by: Satha Rao <skoteshwar@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Ingress meter rate is calculated based on hardcoded
policer time unit. Patch adds mbox interface to
retrieve configured policer time unit.
Signed-off-by: Rakesh Kudurumalla <rkudurumalla@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Disable drop_re i.e dropping packets with receive errors on
vector enable for few cn10k revisions due to HW errata.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Allow flow control on LBK VF's and enable TM to listen on
backpressure when Rx pause is enabled.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Adds support to write CPT CTX through microcode op(SET_CTX/WRITE_SA)
for cn10k inline mode.
Signed-off-by: Srujana Challa <schalla@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Added API to toggle drop_re flag after nix_lf_alloc() so that it
can be used to toggle it runtime.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Enable TM topology to listen on backpressure received when
Rx pause frame is enabled. Only one TM node in Tl3/TL2 per
channel can listen on backpressure on that channel.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Support flow control enable/disable on LBK VF's as HW
supports backpressure on LBK links.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
As previously announced, this patch renames struct
vhost_device_ops to struct rte_vhost_device_ops.
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Chenbo Xia <chenbo.xia@intel.com>
This patch marks the vDPA driver APIs as internal and
rename the corresponding header file to vdpa_driver.h.
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Thomas Monjalon <thomas@monjalon.net>
Reviewed-by: Chenbo Xia <chenbo.xia@intel.com>
Protocol agnostic flow offloading in Flow Director is enabled by this
patch based on the Parser Library, using existing rte_flow raw API.
Note that the raw flow requires:
1. byte string of raw target packet bits.
2. byte string of mask of target packet.
Here is an example:
FDIR matching ipv4 dst addr with 1.2.3.4 and redirect to queue 3:
flow create 0 ingress pattern raw \
pattern spec \
00000000000000000000000008004500001400004000401000000000000001020304 \
pattern mask \
000000000000000000000000000000000000000000000000000000000000ffffffff \
/ end actions queue index 3 / mark id 3 / end
Note that mask of some key bits (e.g., 0x0800 to indicate ipv4 proto)
is optional in our cases. To avoid redundancy, we just omit the mask
of 0x0800 (with 0xFFFF) in the mask byte string example. The prefix
'0x' for the spec and mask byte (hex) strings are also omitted here.
Also update the ice feature list with rte_flow item raw.
Signed-off-by: Junfeng Guo <junfeng.guo@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Based on the parser library, we can directly set HW profile and
associate the main/ctrl vsi.
This patch set also updated the base code BSD release version.
Signed-off-by: Junfeng Guo <junfeng.guo@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
In this patch, we introduced a new parameter to enable/disable the
FDIR SWAP option by setting the swap and inset register set with
certain values.
Signed-off-by: Junfeng Guo <junfeng.guo@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
UBSan testing revealed undefined shift here.
The firmware returns the version in bytes; and shifting a 8 bit
quantity here can lead to undefined behaviour or truncation.
The fix is to promote the bytes to 32 bit before shifting.
Bugzilla ID: 838
Fixes: 9a891c1764 ("net/bnxt: update HWRM to version 1.9.2")
Cc: stable@dpdk.org
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Somnath Kotur <somnath.kotur@broadcom.com>
Acked-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
[1] has deprecated the use of direction attributes in "transfer"
flows. Ignore them during the transition period.
[1]
commit 9d2a349b38 ("ethdev: deprecate direction attributes in transfer flows")
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Add support for actions PORT_REPRESENTOR and REPRESENTED_PORT.
The former should be used instead of ambiguous PORT_ID.
The latter sends traffic to the entity represented by
the given ethdev (network port or VF).
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Add support for item REPRESENTED_PORT to match on traffic entering
the embedded switch from the entity represented by the given
ethdev (network port or VF).
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
In accordance with patches [1-4], MAE admin ethdev represents a
network port and not the PF which it sits on. Rework the way
how "ethdev" and "entity" m-ports are assigned in SW switch
port entries of independent ethdevs. Explain in comments.
[1] commit 081e42dab1 ("ethdev: add port representor item to flow API")
[2] commit 49863ae2bf ("ethdev: add represented port item to flow API")
[3] commit 8edb6bc026 ("ethdev: add port representor action to flow API")
[4] commit 88caad251c ("ethdev: add represented port action to flow API")
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
The function in question has an unfortunate name that reads
like finding a SW switch port entry. In fact just one of
the two m-ports is retrieved from that entry.
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
VF representors do not own dedicated m-ports and thus cannot
be referred to as traffic endpoints in flow items or actions.
Fixes: a62ec90522 ("net/sfc: add port representors infrastructure")
Fixes: f55b61cec9 ("net/sfc: support port representor flow item")
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
MLX5 PMD uses reference counting to manage RX queue resources.
After port stop shared RSS actions kept references to RX queues,
preventing resource release. As a result, internal PMD mempool
for such queues had been exhausted after a number of port restarts.
Diagnostic message from rte_eth_dev_start():
Rx queue allocation failed: Cannot allocate memory
Dereference RX queues used by indirect actions on port stop (detach)
and restore references on port start (attach) in order to allow RX queue
resource release, but keep indirect RSS across the port restart.
Replace queue IDs in HW by drop queue ID on detach and restore actual
queue IDs on attach.
When the port is stopped, create indirect RSS in the detached state.
As a result, MLX5 PMD is able to keep all its indirect actions
across port restart. Advertise this capability.
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>
Drop queue creation and destruction were not implemented for DevX
flow engine and Verbs engine methods were used as a workaround.
Implement these methods for DevX so that there is a valid queue ID
that can be used regardless of queue configuration via API.
Cc: stable@dpdk.org
Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Maximum available flow priority was discovered using Verbs API
regardless of the selected flow engine. This required some Verbs
objects to be initialized in order to use DevX engine. Make priority
discovery an engine method and implement it for DevX using its API.
Cc: stable@dpdk.org
Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
When RTE_ETH_DEV_CAPA_FLOW_RULE_KEEP capability bit is zero,
the specified behavior is the same as it had been before
this bit was introduced. Explicitly reset it in all PMDs
supporting rte_flow API in order to attract the attention
of maintainers, who should eventually choose to advertise
the new capability or not. It is already known that
mlx4 and mlx5 will not support this capability.
For RTE_ETH_DEV_CAPA_FLOW_SHARED_OBJECT_KEEP
similar action is not performed,
because no PMD except mlx5 supports indirect actions.
Any PMD that starts doing so will anyway have to consider
all relevant API, including this capability.
Suggested-by: Ferruh Yigit <ferruh.yigit@intel.com>
Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
Acked-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Acked-by: Somnath Kotur <somnath.kotur@broadcom.com>
Acked-by: Hyong Youb Kim <hyonkim@cisco.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Since v0.4.0, if the underlying kernel supports it, libbpf uses 'bpf
link' to manage the programs on the interfaces of the xsks. This has two
repercussions for the PMD.
1. In the case where the PMD asks libbpf to load the default XDP
program, the PMD no longer needs to remove it on teardown. This is
because bpf link handles the unloading under the hood.
2. In the case where the PMD loads a custom program, libbpf expects this
program to be linked via bpf link prior to creating the socket.
This patch introduces probes for the libbpf version and kernel support
for bpf link and orchestrates the loading and unloading of
programs according to the capabilities of the kernel and libbpf. The
libbpf version is checked with meson and pkg-config. The probe for
kernel support mirrors how it is implemented in libbpf. A bpf_link is
created and looked up on loopback device. If successful, bpf_link will
be used for the AF_XDP netdev.
Signed-off-by: Ciara Loftus <ciara.loftus@intel.com>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
The RSS expansion algorithm is using a graph to find the possible
expansion paths. A graph node with the 'explicit' flag will be skipped,
if it is not found in the flow pattern.
The current implementation misses a check for the explicit flag when
expanding the pattern according to ETH item with EtherType.
For example:
testpmd> flow create 0 ingress pattern eth / ipv6 / udp / vxlan / eth
type is 2048 / end actions rss level 2 types udp end / end
The "eth type is 2048" item in the pattern may be expanded to "ETH IPv4".
The ETH node in the expansion graph is followed by VLAN node marked as
explicit. The fix is to skip the VLAN node and continue the expansion
with its next nodes, IPv4 and IPv6.
The expansion paths for the above example will be:
ETH IPV6 UDP VXLAN ETH END
ETH IPV6 UDP VXLAN ETH IPV4 UDP END
Fixes: 69d268b4ff ("net/mlx5: fix RSS expansion for explicit graph node")
Cc: stable@dpdk.org
Signed-off-by: Lior Margalit <lmargalit@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
The ASO meter action with flows creation could be supported on
multiple threads. The meter pools were created to manage the meter
object resources, if there is no room in the current meter pool then
resize the meter pool to the new pool size and free the old one.
There's a race condition while one thread resizes the meter pool and
the old pool resource be freed, and another thread query the meter
object by index on the old pool, the return value is invalid.
This patch adds a read-write lock to protect the pool resource while
resizing and query.
Fixes: a5835d530f ("net/mlx5: optimize Rx queue match")
Cc: stable@dpdk.org
Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
The age action with flows creation could be supported on the multiple
threads. The age pools were created to manage the age resources, if
there is no room in the current pool then resize the age pool to the new
pool size and free the old one.
There's a race condition while one thread resizes the age pool and the
old pool resource be freed, and another thread query the age action
value of the old pool so the queried value is invalid.
This patch uses the read-write lock to protect the pool resource while
resizing and query.
Fixes: a5835d530f ("net/mlx5: optimize Rx queue match")
Cc: stable@dpdk.org
Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
