This is workaround for 8000-series EF10 hardware TSO bug.
Innermost IP length and outer UDP datagram length must be
greater than or equal to the corresponding values derived
from the MSS; otherwise, the checksum offloads will break.
Fixes: c1ce2ba218 ("net/sfc: support tunnel TSO on EF10 native Tx datapath")
Fixes: 6bc985e411 ("net/sfc: support TSO in EF10 Tx datapath")
Fixes: fec33d5bb3 ("net/sfc: support firmware-assisted TSO")
Cc: stable@dpdk.org
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Reviewed-by: Andy Moreton <amoreton@xilinx.com>
Release-on-close has been implemented for the NFP PMD. Remove the
UNMAINTAINED flag.
Signed-off-by: Heinrich Kuhn <heinrich.kuhn@netronome.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
Here fixes the Kunpeng introduction address link with hns3.rst
Fixes: 565829db8b ("net/hns3: add build and doc infrastructure")
Cc: stable@dpdk.org
Signed-off-by: Lijun Ou <oulijun@huawei.com>
Document FEC in NIC features, add information about FEC and add
implementation related support.
Fixes: b7ccfb09da ("ethdev: introduce FEC API")
Fixes: 9bf2ea8dbc ("net/hns3: support FEC")
Fixes: 62aafe0358 ("net/cxgbe: support configuring link FEC")
Cc: stable@dpdk.org
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Signed-off-by: Lijun Ou <oulijun@huawei.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Currently, X722 firmware doesn't support to add more than
one mirror rule in one VSI.
Signed-off-by: Lingyu Liu <lingyu.liu@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>
For pattern MAC_PPPOE_IPV4/6, add ETH_RSS_ETH into input_set_mask
to fix RSS rule cannot be created when set eth as RSS type.
Fixes: 0d84f86c30 ("net/ice: fix GTPU header parsing")
Signed-off-by: Xuan Ding <xuan.ding@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
VIRTIO_OPS() macro relies on the port ID stored in the
virtio_hw struct. Issue is that it is used before being
assigned at init time. It results in all devices setting
ops on port ID 0, causing crash later when calling ops
for port IDs other than 0.
This patch ensures port ID assignment is done at early
primary process probe time, before it is being used.
Bugzilla ID: 631
Fixes: 512e27eeb7 ("net/virtio: move PCI specific dev init to PCI ethdev init")
Reported-by: Wei Ling <weix.ling@intel.com>
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: David Marchand <david.marchand@redhat.com>
Tested-by: Wei Ling <weix.ling@intel.com>
When running in client mode, the listen file descriptor
is not initialized, and so has value 0. At destroy time,
the listen FD is closed if its value is greater than or
equal to zero, which causes STDIN to be closed.
Bugzilla ID: 630
Fixes: 949735312f ("net/virtio: move vhost-user specifics to its backend")
Reported-by: Jun W Zhou <junx.w.zhou@intel.com>
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: David Marchand <david.marchand@redhat.com>
Tested-by: Jun W Zhou <junx.w.zhou@intel.com>
The filter type (struct filter_v2.type) should always be set to
FILTER_DPDK_1, when advanced filtering is enabled in firmware.
Otherwise, for some old firmware versions, the driver sets
it to FILTER_USNIC_IP, and attempts to install filters fail. This
behavior matches that of the now-removed flow director implementation
(enic_clsf.c).
Fixes: 26faa126d8 ("net/enic: flow API for NICs with advanced filters disabled")
Cc: stable@dpdk.org
Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
Reviewed-by: John Daley <johndale@cisco.com>
The ionic PMD only supports Tx queue version 1 or greater.
Version 1 introduced a new SGL format with support for more
fragments per descriptor.
Add release notes and an explanation to the docs.
Signed-off-by: Andrew Boyer <aboyer@pensando.io>
This patch support LSC (Link Status Change) event report.
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Lijun Ou <oulijun@huawei.com>
In the Testpmd Flow rules management section, correct
the TPID values in the Sample QinQ flow rules sub section.
Also replace the keyword qinq_strip with extend in the
vlan set command.
Fixes: bef3bfe7d5 ("doc: revise sample testpmd flow commands")
Cc: stable@dpdk.org
Signed-off-by: Bernard Iremonger <bernard.iremonger@intel.com>
Acked-by: Xiaoyun Li <xiaoyun.li@intel.com>
Add support to specify PF or VF as targets in "set sample_actions"
command.
Signed-off-by: Satheesh Paul <psatheesh@marvell.com>
Acked-by: Ori Kam <orika@nvidia.com>
This patch includes several updates of the shared RSS action:
(1)
The shared RSS action, introduced recently, uses existing definitions
of the regular RSS action.
The new defined value MLX5_RSS_HASH_IPV4_TCP uses existing definition
IBV_RX_HASH_SRC_PORT_TCP twice, instead of using
IBV_RX_HASH_SRC_PORT_TCP and IBV_RX_HASH_DST_PORT_TCP.
--- ---
The same is true for IPv4-UDP, IPv6-TCP, IPv6-UDP.
As result, a shared RSS action with L4 type is specified as src-only.
Flow rule using such shared action, while specifying L4 item in flow
pattern, will fail to create.
This patch updates the new definitions, to use the existing values
correctly.
(2)
On shared RSS action destroy, in function __flow_dv_action_rss_release,
the indirection table shared_rss->ind_tbl was released before
shared_rss->refcnt was checked.
This order is incorrect, since the indirection table should be
released only when the shared RSS action is destroyed.
This patch puts release function calls in correct order.
(3)
Variables declared of type "struct mlx5_shared_action_rss" are named
"shared_rss", "action", and "shared_action".
To improve code readability, this patch renames all to "shared_rss".
Fixes: d7cfcddded ("net/mlx5: translate shared action for RSS action")
Fixes: d2046c09aa ("net/mlx5: support shared action for RSS")
Cc: stable@dpdk.org
Signed-off-by: Dekel Peled <dekelp@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Existing names of the flags denoting flow meter capability are unclear
and may be misleading.
This patch updates the names to align with the updated documentation.
Comments were edited, describing the names clearly.
Signed-off-by: Dekel Peled <dekelp@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
The size of the receive completion ring should be recalculated
when MTU is increased to a size that requires scattered receive
or when LRO is enabled. Move logic for this calculation from
the ring configuration path to the device start path.
- Made size calculation dependent only on scattered_rx
status.
- Moved calculation of scattered_rx up in the initialization
sequence.
- Made LRO offload status part of scattered_rx calculation.
When the completion ring size is too small, completion overflows
can occur causing the ring to be disabled in hardware.
Fixes: 04067844a3 ("net/bnxt: reduce CQ queue size without aggregation ring")
Cc: stable@dpdk.org
Signed-off-by: Lance Richardson <lance.richardson@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Add advanced RSS offloads check due to some legacy driver (kernel/DPDK
PF) does not support virtual channel command VIRTCHNL_OP_RSS_HENA with
hena = 0 and VIRTCHNL_OP_ADD_RSS_CFG.
Fixes: 95f2f0e9fc ("net/iavf: improve default RSS")
Signed-off-by: Xuan Ding <xuan.ding@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Since we don't check ABI on the x86-default target anymore, installation
of the target must always happen for examples external compilation check
to work.
Fixes: 6a426d733e ("devtools: reduce ABI checks and static binaries")
Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Thomas Monjalon <thomas@monjalon.net>
There are compiler issues when building with -mcpu=native with popular
compilers, such as GCC-8.4:
In file included from ../lib/librte_eal/arm/include/rte_vect.h:11,
from ../lib/librte_net/net_crc_neon.c:10:
../lib/librte_net/net_crc_neon.c: In function ‘crcr32_folding_round’:
/usr/lib/gcc/aarch64-linux-gnu/8/include/arm_neon.h:26094:1: error:
inlining failed in call to always_inline ‘vmull_p64’:
target specific option mismatch
vmull_p64 (poly64_t a, poly64_t b)
../lib/librte_net/net_crc_neon.c:50:20: note: called from here
uint64x2_t tmp1 = vreinterpretq_u64_p128(vmull_p64(
vgetq_lane_p64(vreinterpretq_p64_u64(fold), 0),
vgetq_lane_p64(vreinterpretq_p64_u64(precomp), 1)));
and clang:
gcc -E -dM -mcpu="native" - < /dev/null | grep __ARM_FEATURE_ATOMICS
clang-9 -E -dM -mcpu="native" - < /dev/null | grep __ARM_FEATURE_ATOMICS
<no output> # no clang support
Fix this by always specifying the proper machine args and never using
the native flags.
Fixes: 78ac8eac7e ("config/arm: use native machine build arguments")
Signed-off-by: Juraj Linkeš <juraj.linkes@pantheon.tech>
A config variable rename seems missed to update all config files and new
config file get with old variable names.
Reflect config variable rename to all config files, changed
'implementor_id' to 'implementer_id',
'implementor_pn' to 'part_number'.
Fixes: 3d01d65ba0 ("config: add aarch64 clang for Ubuntu 18.04")
Fixes: 7870ae8994 ("config/arm: rename variables")
Reported-by: Thierry Herbelot <thierry.herbelot@6wind.com>
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Juraj Linkeš <juraj.linkes@pantheon.tech>
Ubuntu 18.04 GHA virtual machine images point at an invalid APT
repository.
We have no control over this, simply ignore the failure.
This was caught by Ilya for OVS and the robot just hit the same issue
for DPDK:
"""
Get:46 http://security.ubuntu.com/ubuntu bionic-security/restricted
Translation-en [29.9 kB]
Get:47 http://security.ubuntu.com/ubuntu bionic-security/universe amd64
Packages [1104 kB]
Get:48 http://security.ubuntu.com/ubuntu bionic-security/universe
Translation-en [247 kB]
Reading package lists...
E: The repository 'https://apt.postgresql.org/pub/repos/apt bionic-pgdg
Release' no longer has a Release file.
Error: Process completed with exit code 100.
"""
Fixes: 9d620630ea ("ci: fix package installation in GitHub Actions")
Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Ilya Maximets <i.maximets@ovn.org>
Acked-by: Aaron Conole <aconole@redhat.com>
Since the examples are designed to be built by end-users using Make, we
can detect and warn about broken pkg-config on the user's system as part
of the build process.
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Previous fix used `rte_malloc_heap_socket_is_external()` to check if the
heap was an external heap. However, that API is thread-safe, and when
we're inside the allocation process, we're already write-locked, so
calling `rte_malloc_heap_socket_is_external()` will result in a
deadlock followed by a timeout.
Fix it by replacing the API call with a check against maximum number of
NUMA nodes, because external heaps always have higher socket ID's.
Fixes: 7ac31e82bc ("mem: improve parameter checking on memory hotplug")
Reported-by: Jim Harris <james.r.harris@intel.com>
Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
There is the "set rxpkts" command in the testpmd interactive mode,
it configures the segment sizes to split the packet on receiving.
The mentioned segment sizes are provided on the Rx queue setup
as part of queue configuration. Hence, to take the rxpkts command
into effect the Rx queues must be explicitly reconfigured.
The explained above is related to the "set rxoffs" as well.
The patch sets the queue reconfiguration request flag for
all devices once Rx split settings are updated, to take
the changes into effect the port(s) should be restarted.
Fixes: 0f2096d7ab ("app/testpmd: add rxpkts commands and parameters")
Fixes: 91c78e090e ("app/testpmd: add rxoffs commands and parameters")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Commit "899f06130724" broke the update of mbuf flags for PTP packets.
"mbuf->ol_flags" is overwritten in bnxt_set_ol_flags() function.
Fixes: 899f061307 ("net/bnxt: add Rx logic for 58818 chips")
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
1. Packet transmit function for both otx and otx2 are added.
2. Flushing transmit(command) queue when pending commands are more than
maximum allowed value (currently 16).
3. Scatter gather support if the packet spans multiple buffers.
Signed-off-by: Nalla Pradeep <pnalla@marvell.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Function to deliver packets from DROQ to application is added. It also
fills DROQ with receive buffers timely such that device can fill them
with incoming packets.
Signed-off-by: Nalla Pradeep <pnalla@marvell.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Dev start and stop operations are added. To accomplish this internal
functions to enable or disable IO queues are incorporated.
Signed-off-by: Nalla Pradeep <pnalla@marvell.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Configuring hardware registers with command queue (IQ) and driver output
queue (OQ) parameters.
List of parameters configured for IQ after making sure it is idle
1. Base address
2. Instruction size
3. Disabling interrupts for fastpath
List of parameters configured for OQ after making sure it is idle
1. Base address
2. Output buffer size
3. Clear output queue doorbell
4. Disable interrupts for fastpath
Signed-off-by: Nalla Pradeep <pnalla@marvell.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Receive queue setup involves allocating memory for the queue,
initializing data structure representing the queue and filling queue
with receive buffers of rx descriptor count. Receive queues are referred
as droq. Hardware fills the receive buffers in queue with the packet.
In receive queue release, receive buffers are freed along with the
receive queue.
Signed-off-by: Nalla Pradeep <pnalla@marvell.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Add basic init and uninit function which includes
initializing fields of ethdev private structure.
Signed-off-by: Nalla Pradeep <pnalla@marvell.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Adding bare minimum PMD library and doc build infrastructure
and claim the maintainership for octeontx end point PMD.
Signed-off-by: Nalla Pradeep <pnalla@marvell.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Device id to be probed by octeontx_ep raw device pmd is changed from
B203 to B204. B203 is for octeontx_ep net device pmd to probe.
Signed-off-by: Nalla Pradeep <pnalla@marvell.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Add per-tx-ring flag for packets that were pushed to HW but await
doorbell. That is to prevent a situation when a doorbell is sent due to
reaching Tx burst threshold and next send fails (e.g., due to queue
full). In such case we shouldn't send another doorbell because there are
no actual packets waiting for transmission.
Fixes: c7519ea5eb ("net/ena: call additional doorbells if needed")
Cc: stable@dpdk.org
Signed-off-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Michal Krawczyk <mk@semihalf.com>
Before starting transmission of Tx burst, the driver checked the
available space in the sq and limited the number of packets for
transmission accordingly.
The calculation was incorrect for fragmented packets and potentially had
significantly limited the length of Tx bursts.
This patch removes the assessment and pushes packets to the sq as long
as the burst is not exhausted and space is available in the sq.
Correct evaluation of the required space isn't possible before the burst
because it depends on the number of segments of each packet.
This patch adds per-packet space evaluation for each packet before
attempting to process it. In case there is not enough queue space, the
burst will just stop without error.
Fixes: 2061fe41f2 ("net/ena: linearize Tx mbuf")
Cc: stable@dpdk.org
Signed-off-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Michal Krawczyk <mk@semihalf.com>
Instead of verifying the Rx descriptor each time it's being used in the
driver code, now the verification happens on the HAL side.
This simplifies code a lot as instead of doing 2 validations, only
single one is needed. The driver have to check the rc value returned
by the ena_com upon reading the Rx descriptor and trigger the reset
if needed. It was previously the responsibility of the
validate_rx_req_id() function.
As part of the change, the version of the driver was bumped to v2.2.1.
Fixes: 2061fe41f2 ("net/ena: linearize Tx mbuf")
Cc: stable@dpdk.org
Signed-off-by: Ido Segev <idose@amazon.com>
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Increment Tx doorbell statistics on tx_pkt_burst
after writing to doorbell and in case max burst size achieved
Fixes: c7519ea5eb ("net/ena: call additional doorbells if needed")
Cc: stable@dpdk.org
Signed-off-by: Amit Bernstein <amitbern@amazon.com>
Reviewed-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
As the refill called as part of ena_start(), we end up the refill
progress with stuck buffers at the caller core cache.
Calling to flush the cache results with invalidate this cache and free
those stuck buffers.
Fixes: 1173fca25a ("ena: add polling-mode driver")
Cc: stable@dpdk.org
Signed-off-by: Ido Segev <idose@amazon.com>
Reviewed-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
For newer smartNIC NVRAM versions the chip model should be read from the
PluDevice register as it provides the authoritative chip model/revision.
This method of reading the chip model is backwards compatible with
legacy NVRAM versions too.
Since the model number is purely used for reporting purposes, follow the
hardware team convention of subtracting 0x10 from the PluDevice register
to obtain the chip model/revision number.
Fixes: c7e9729da6 ("net/nfp: support CPP")
Cc: stable@dpdk.org
Signed-off-by: Heinrich Kuhn <heinrich.kuhn@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Reviewed-by: Louis Peens <louis.peens@netronome.com>
In interactive mode, if testpmd exit by calling rte_exit without
restore terminal attributes, terminal will not echo keyboard input.
register a function with atexit() in prompt(), when exit() in
rte_exit() is called, the registered function restores terminal
attributes.
Fixes: 5a8fb55c48 ("app/testpmd: support unidirectional configuration")
Cc: stable@dpdk.org
Signed-off-by: Dapeng Yu <dapengx.yu@intel.com>
Acked-by: Xiaoyun Li <xiaoyun.li@intel.com>
When testing high performance numbers, it is often that CPU performance
limits the max values device can reach (both in pps and in gbps)
Here instead of recreating each packet separately, we use clones counter
to resend the same mbuf to the line multiple times.
PMDs handle that transparently due to reference counting inside of mbuf.
Reaching max PPS on small packet sizes helps here:
Some data from our 2 port x 50G device. Using 2*6 tx queues, 64b packets,
PowerEdge R7525, AMD EPYC 7452:
./build/app/dpdk-testpmd -l 32-63 -- --forward-mode=flowgen \
--rxq=6 --txq=6 --disable-crc-strip --burst=512 \
--flowgen-clones=0 --txd=4096 --stats-period=1 --txpkts=64
Gives ~46MPPS TX output:
Tx-pps: 22926849 Tx-bps: 11738590176
Tx-pps: 23642629 Tx-bps: 12105024112
Setting flowgen-clones to 512 pushes TX almost to our device
physical limit (68MPPS) using same 2*6 queues(cores):
Tx-pps: 34357556 Tx-bps: 17591073696
Tx-pps: 34353211 Tx-bps: 17588802640
Doing similar measurements per core, I see one core can do
6.9MPPS (without clones) vs 11MPPS (with clones)
Verified on Marvell qede and atlantic PMDs.
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
