Add tested Intel platforms with Intel NICs to v21.11 release note.
Signed-off-by: Yan Xia <yanx.xia@intel.com>
Acked-by: Xueqin Lin <xueqin.lin@intel.com>
The size of delay request message sent out by the DPDK
ptpclient application was observed to have extra length
than expected. Due to this, bad messages were observed
on the master side and delay response was not received.
This patch fixes this bug.
Fixes: ab129e9065 ("examples/ptpclient: add minimal PTP client")
Cc: stable@dpdk.org
Signed-off-by: Vanshika Shukla <vanshika.shukla@nxp.com>
Acked-by: Nipun Gupta <nipun.gupta@nxp.com>
Device iterator RTE_DEV_FOREACH() failed to return devices from
classifier like "class=vdpa", because matching name from empty kvargs
returns no result. If device name not specified in kvargs, the function
should iterate all devices.
This patch allows empty devargs or devargs without name specified.
Fixes: 6aebb94290 ("kvargs: add function to get from key and value")
Signed-off-by: Xueming Li <xuemingl@nvidia.com>
Signed-off-by: Olivier Matz <olivier.matz@6wind.com>
Acked-by: Xueming Li <xuemingl@nvidia.com>
examples.dox is built inside builddir/doc/api and so doxygen generates
some dir_HASH.html that includes the builddir name and this may prevent
DPDK documentation to be correctly generated in some distributions, for
example CentOS Stream 9 and RHEL9, since the builddir includes the
architecture.
This commit adds builddir/doc/api (the path where examples.dox is
generated) to STRIP_FROM_PATH, so the generated documentation doesn't
change if builddir changes.
Fixes: a6090630f4 ("doc: automate examples file list for API")
Cc: stable@dpdk.org
Signed-off-by: Timothy Redaelli <tredaelli@redhat.com>
Acked-by: Luca Boccassi <bluca@debian.org>
Tested-by: David Marchand <david.marchand@redhat.com>
The dump of dynamic fields and flags fails if the shm is already
allocated. Add a check to fix the issue.
Fixes: d4902ed31c ("mbuf: check shared memory before dumping dynamic space")
Cc: stable@dpdk.org
Signed-off-by: Alexander Bechikov <asb.tyum@gmail.com>
Acked-by: Olivier Matz <olivier.matz@6wind.com>
To enable bifurcated device support, rtnl_lock is released before calling
userspace callbacks and asynchronous requests are enabled.
But these changes caused more issues, like bug #809, #816. To reduce the
scope of the problems, the bifurcated device support related changes are
only enabled when it is requested explicitly with new 'enable_bifurcated'
module parameter.
And bifurcated device support is disabled by default.
So the bifurcated device related problems are isolated and they can be
fixed without impacting all use cases.
Bugzilla ID: 816
Fixes: 631217c761 ("kni: fix kernel deadlock with bifurcated device")
Cc: stable@dpdk.org
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Igor Ryzhov <iryzhov@nfware.com>
Wakeup happens quite often (several hundred times a second) in
l3fwd-power example app in PMD power management mode, so this
message is appearing too often to be useful.
This patch reverts that info message addition.
Fixes: 931e3a9945 ("examples/l3fwd-power: add wakeup log")
Signed-off-by: David Hunt <david.hunt@intel.com>
Tested-by: Reshma Pattan <reshma.pattan@intel.com>
Acked-by: Reshma Pattan <reshma.pattan@intel.com>
Application calculates segment size based on buffer size plus
digest size only, But if the operation mode is IPsec then
packet length can be increased by some more bytes depending on
the algorithm.
In this patch, increasing segment size with RTE_PKTMBUF_HEADROOM
when there is no user given segment size.
Fixes: 28dde5da50 ("app/crypto-perf: support lookaside IPsec")
Signed-off-by: Gagandeep Singh <g.singh@nxp.com>
Acked-by: Akhil Goyal <gakhil@marvell.com>
Updated AESNI MB and AESNI GCM, KASUMI, ZUC and SNOW3G PMD documentation
guides with information about the latest Intel IPSec Multi-buffer
library supported.
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Acked-by: Fan Zhang <roy.fan.zhang@intel.com>
dpaa, dpaa2 and caam_jr drivers do not support
SA expiry. This result in failure of test cases in
test app. This patch returns ENOTSUP to skip the
SA lifetime test cases.
Signed-off-by: Gagandeep Singh <g.singh@nxp.com>
Acked-by: Akhil Goyal <gakhil@marvell.com>
dpaa, dpaa2 and caam_jr drivers decrement the inner IP header
TTL for all packets and ignoring the dec_ttl option of SA.
In this patch, using the dec_ttl to decide to decrement the
packets inner IP header TTL or not.
Fixes: 0a23d4b6f4 ("crypto/dpaa2_sec: support protocol offload IPsec")
Fixes: 3e33486f80 ("crypto/caam_jr: add security offload")
Fixes: 1f14d500bc ("crypto/dpaa_sec: support IPsec protocol offload")
Cc: stable@dpdk.org
Signed-off-by: Gagandeep Singh <g.singh@nxp.com>
Acked-by: Akhil Goyal <gakhil@marvell.com>
An indirect mkey is created for each descriptor in a QP, number of
descriptors per QP is configured by the user on QP setup callback.
In mlx cryptodev autotest, the max number of QPs (provided by the driver)
is created, and due to mkey resource limits, QPs creation fail which leads
to the test failing.
Since there is no capability of max number of descriptors provided to
the user, we can't give an exact number of max QPs available.
Reduce the max number of QPs to 128, which supports standard descriptors
numbers, including the 4K number provided in the test.
Fixes: 6152534e21 ("crypto/mlx5: support queue pairs operations")
Cc: stable@dpdk.org
Signed-off-by: Raja Zidane <rzidane@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
When authenticating with SNOW3G, KASUMI and ZUC,
the pointers for encryption/decryption keys is not set.
If a cipher algorithm such as AES-CBC is also used,
the application would seg fault.
Hence, these pointers should be set to some value by default.
Command line to replicate the issue:
./build/app/dpdk-test-crypto-perf -l 4,5 -n 6 --vdev="crypto_aesni_mb" -- \
--devtype="crypto_aesni_mb" --optype=cipher-then-auth --auth-algo \
snow3g-uia2 --auth-key-sz 16 --auth-iv-sz 16 --digest-sz 4 --silent \
--total-ops 1000000 --auth-op generate --burst-sz 32 \
--cipher-algo aes-ctr --cipher-key-sz 16 --cipher-iv-sz 16
Fixes: ae8e085c60 ("crypto/aesni_mb: support KASUMI F8/F9")
Fixes: 6c42e0cf4d ("crypto/aesni_mb: support SNOW3G-UEA2/UIA2")
Fixes: fd8df85487 ("crypto/aesni_mb: support ZUC-EEA3/EIA3")
Cc: stable@dpdk.org
Signed-off-by: Pablo de Lara <pablo.de.lara.guarch@intel.com>
Acked-by: Ciara Power <ciara.power@intel.com>
Acked-by: Fan Zhang <roy.fan.zhang@intel.com>
Set correct rte_errno variable in CUDA driver
and return -rte_errno in case of error.
rte_errno values are compliant with the gpudev library documentation.
Fixes: 1306a73b19 ("gpu/cuda: introduce CUDA driver")
Signed-off-by: Elena Agostini <eagostini@nvidia.com>
The gpudev functions free, register and unregister
return gracefully if input pointer is NULL or size 0,
as API doc was indicating no-op accepted values.
CUDA driver checks are removed because redundant
with the checks added in gpudev library.
Fixes: e818c4e2bf ("gpudev: add memory API")
Signed-off-by: Elena Agostini <eagostini@nvidia.com>
Multiple drivers are defining macros for VLAN header length, to remove
the redundancy defining macro in the ether header.
And updated drivers to use the new macro.
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Haiyue Wang <haiyue.wang@intel.com>
Acked-by: Rosen Xu <rosen.xu@intel.com>
Acked-by: Jiawen Wu <jiawenwu@trustnetic.com>
Currently, ethtool directly ends the process after 'quit' cmd. In this
case, software resources are not released and hardware resources of the
device are not uninstalled.
This patch adds closing port operation to release resources.
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
In DPDK, 'rte_socket_id' means the running socket while
'rte_eth_dev_socket_id' is the device socket.
For better performance, memory which queue setup used and device
should be in the same socket.
This patch make sure it calls rte_eth_dev_socket_id API to get device
socket_id when setting ringparam.
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
This patch adds a comment for RTE_HASH_BUCKET_ENTRIES
explaining why a particular value was chosen.
Signed-off-by: Vladimir Medvedkin <vladimir.medvedkin@intel.com>
Reviewed-by: Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>
This library can be made optional.
drivers/gpu and app/test-gpudev depend on this library,
so they are automatically disabled if the lib is disabled.
Signed-off-by: Elena Agostini <eagostini@nvidia.com>
If the modify field action requests the field copy/set transaction
from other field, the destination field hardware bit offset was
assigned incorrectly with non-zero byte offset, causing wrong
translations for the fields with sizes larger than 32 bits.
Fixes: 40c8fb1fd3 ("net/mlx5: update modify field action")
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The "mlx5_devx_cmd_mkey_create" DevX cmd allocates DevX object using
mlx5_malloc and then creates MKey using glue function.
Compatibly, "mlx5_devx_cmd_destroy" cmd releases first the MKey using
glue function, and then free the DevX object using mlx5_free.
On Windows OS, the reg_mr function creates Mkey using
"mlx5_devx_cmd_mkey_create" cmd, but dereg_mr function using directly
glue function without freeing the object.
This behavior causes memory leak at each MR release.
In addition, the dereg_mr function makes sure that the MR pointer is
valid before destroying its fields, but always calls the memset function
that makes a difference to it.
This patch moves the dereg_mr function to use "mlx5_devx_cmd_destroy"
instead of glue function, and extends the validity test to the whole
function.
Fixes: ba42071982 ("common/mlx5: add reg/dereg MR on Windows")
Cc: stable@dpdk.org
Signed-off-by: Michael Baum <michaelba@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
To detect the timestamp mode configured on the NIC the mlx5 PMD uses the
firmware command ACCESS_REGISTER_USER.
The HCA capability command has an attribute flag checking whether
firmware supports the command.
However, the HCA capability query command read the flag from wrong place
in PRM structure.
This patch move the flag to correct place.
Fixes: 972a1bf812 ("common/mlx5: fix user mode register access command")
Cc: stable@dpdk.org
Signed-off-by: Michael Baum <michaelba@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
mlx5 PMD incorrectly overwrote outer layer fields in MPLS tunnel
rte_flow patterns using defaults for MPLS tunnels. This included
overwriting UDP destination port in MPLSoUDP and GRE protocol field in
MPLSoGRE.
This patch fixes this behavior. If application provides the values in
flow pattern items preceding the MPLS flow item the provided values will
be used, otherwise the defaults will be applied.
Fixes: d1abe664dd ("net/mlx5: add MPLS to Direct Verbs flow engine")
Cc: stable@dpdk.org
Signed-off-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Assuming a user tried to send multi-segment packets, with
RTE_PMD_MLX5_FINE_GRANULARITY_INLINE flag set, using a device with
minimum inlining requirements (such as ConnectX-4 Lx or when user
specified them explicitly), sending such packets caused segfault.
Segfault was caused by failed invariants in
mlx5_tx_packet_multi_inline function.
This patch introduces a logic for multi-segment packets, with
RTE_PMD_MLX5_FINE_GRANULARITY_INLINE flag set, to omit mbuf scanning for
filling inline buffer and inline only minimal amount of data required.
Fixes: ec837ad0fc ("net/mlx5: fix multi-segment inline for the first segments")
Cc: stable@dpdk.org
Signed-off-by: Dariusz Sosnowski <dsosnowski@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Users are confused with a feature with "P", added necessary
explanation for this.
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
These edits affect the outermost header in the current processing state
of the packet, which might have been decapsulated by prior action DECAP.
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Reviewed-by: Andy Moreton <amoreton@xilinx.com>
In a tunnel packet, these actions affect the inner header if
action DECAP is set; otherwise, they affect the outer header.
Adding these actions is done in two steps: add the action to
the action mask and indicate the MAC address entry ID to use.
This allows the user to check the order of actions first and
allocate resources when time comes to enable the action rule.
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Reviewed-by: Andy Moreton <amoreton@xilinx.com>
Add a context structure to simplify handling of action sets.
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Reviewed-by: Andy Moreton <amoreton@xilinx.com>
The number of counters being non-zero can be detected before
the action set registry traversal, so move the check outside.
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Reviewed-by: Andy Moreton <amoreton@xilinx.com>
Tunnel offload API allows application to restore packet to
its original form if chain of flows missed after DECAP action.
The main idea of the tunnel offload API was to query port PMD
to provide flow elements - actions or items.
Flow elements supplied by PMD are merged with original flow rule
elements provided by testpmd operator to create a new flow rule,
optimal for PMD, to implement the tunnel offload API.
That flow rule transformation is hidden form testpmd operator and uses
internal testpmd resources.
Current testpmd did not release tunnel offload resources if flow rule
validation failed.
The patch always releases tunnel offload resources after flow rule
validation returns.
Fixes: 1b9f274623 ("app/testpmd: add commands for tunnel offload")
Cc: stable@dpdk.org
Signed-off-by: Gregory Etelson <getelson@nvidia.com>
Reviewed-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Acked-by: Aman Singh <aman.deep.singh@intel.com>
Remove the vdev args check for secondary process which prevents the
secondary from attaching to the device created by the primary process
via the hotplug framework. This check was removed for other vdevs but
was missed for failsafe.
Fixes: 4852aa8f6e ("drivers/net: enable hotplug on secondary process")
Cc: stable@dpdk.org
Signed-off-by: Kumara Parameshwaran <kumaraparamesh92@gmail.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
DMA on SN1022 SoC requires extra mapping of the memory via MCDI.
Signed-off-by: Viacheslav Galaktionov <viacheslav.galaktionov@oktetlabs.ru>
Signed-off-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Reviewed-by: Ivan Malov <ivan.malov@oktetlabs.ru>
NIC DMA memory regions API allows to establish mapping of DMA addresses
used by NIC to host IOVA understood by the host when IOMMU is absent
and NIC cannot address entire host IOVA space because of too small
DMA mask.
The API does not allow to address entire host IOVA space, but allows
arbitrary regions of the space really used for the NIC DMA.
A DMA region needs to be mapped in order to perform MCDI initialization.
Since the NIC has not been probed at that point, its configuration cannot
be accessed and there an UNKNOWN mapping type is assumed.
Signed-off-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Reviewed-by: Andy Moreton <amoreton@xilinx.com>
Fixes: 60e53c078d ("net/sfc: support decrement IP TTL actions in transfer flows")
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Reviewed-by: Andy Moreton <amoreton@xilinx.com>
Fixes: 96fd2bd69b ("net/sfc: support flow action count in transfer rules")
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>
The driver internal variable to track the next consumer index on
the Rx ring was not being set if there was an mbuf allocation
failure. In that scenario, eventually it would fall out of sync
with the actual consumer index and raise a false alarm on Thor
needlessly causing a segmentation fault with testpmd
Fixes: 03c8f2fe11 ("net/bnxt: detect bad opaque in Rx completion")
Cc: stable@dpdk.org
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
The ULP context list was not updated when high availability
feature was deinitialized. This caused the ULP context list
to acquire the lock when it is not supposed to causing a
deadlock. The fix is to correctly clear the list.
Fixes: 3184b1ef66 ("net/bnxt: add HA support in ULP")
Cc: stable@dpdk.org
Signed-off-by: Kishore Padmanabha <kishore.padmanabha@broadcom.com>
Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com>
Reviewed-by: Mike Baucom <michael.baucom@broadcom.com>
Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
1. removed the global flag for TruFlow global config initialization.
2. Modified the TruFlow context lock to be a global lock instead
of per context lock.
3. The ULP context list is modified to check on the ULP configuration
data so alarm handlers can operate on the correct ULP context.
These changes help in support of multiple network cards using
single DPDK application.
Fixes: d75b55121b ("net/bnxt: add context list for timers")
Cc: stable@dpdk.org
Signed-off-by: Kishore Padmanabha <kishore.padmanabha@broadcom.com>
Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com>
Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com>
Reviewed-by: Mike Baucom <michael.baucom@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
The SRAM resource free did not reset the next block to be used
when the block is not empty. This caused the flows not be created
when max flows limit is reached and you delete one flow and try to
add a new flow. The fix calls the update of the next free block
even when block is not empty.
Fixes: 37ff91c158 ("net/bnxt: add SRAM manager model")
Signed-off-by: Kishore Padmanabha <kishore.padmanabha@broadcom.com>
Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com>
Reviewed-by: Mike Baucom <michael.baucom@broadcom.com>
Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Move wc_tcam_slices_per_row and database structure of
global_cfg and if_tbl to session structure to support
multiple TruFlow sessions with different card type under single
DPDK application instance.
Signed-off-by: Jay Ding <jay.ding@broadcom.com>
Signed-off-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com>
Reviewed-by: Farah Smith <farah.smith@broadcom.com>
Reviewed-by: Randy Schacher <stuart.schacher@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Testing has shown no performance benefit from software prefetching
of receive completion descriptors in the AVX2 burst receive path,
and slightly better performance without them on some CPU families,
so this patch removes them.
Fixes: c4e4c18963 ("net/bnxt: add AVX2 RX/Tx")
Cc: stable@dpdk.org
Signed-off-by: Lance Richardson <lance.richardson@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Each call to the AVX2 vector burst receive function makes at
least one pass through the function's inner loop, loading
256 bytes of completion descriptors and copying 8 rte_mbuf
pointers regardless of whether there are any packets to be
received.
Unidirectional forwarding performance is improved by about
3-4% if we ensure that at least one packet can be received
before entering the inner loop.
Fixes: c4e4c18963 ("net/bnxt: add AVX2 RX/Tx")
Cc: stable@dpdk.org
Signed-off-by: Lance Richardson <lance.richardson@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
The current approach detects the proxy port on each port (re-)plug and
may spam the log with error messages if the PMD does not support flows.
As testpmd is a debug tool, it must not do such implicit port handling.
Instead, the new API should be called only when the user requests that.
Revoke the existing code. Implement an explicit command-line primitive
to let the user find the proxy port themselves. Provide relevant hints.
Fixes: 1179f05cc9 ("ethdev: query proxy port to manage transfer flows")
Signed-off-by: Ivan Malov <ivan.malov@oktetlabs.ru>
Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Acked-by: Ori Kam <orika@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Mempool registration code had a wrong assumption that it is always
dealing with packet mempools and always called rte_pktmbuf_priv_flags(),
which returned a random value for different types of mempools.
In particular, it could consider MPRQ mempools as having externally
pinned buffers, which is wrong.
Packet mempools cannot be reliably recognized, but it is sufficient to
check that the mempool is not a packet one, so it cannot have externally
pinned buffers.
Compare mempool private data size to that of packet mempools to check.
Fixes: 690b2a88c2 ("common/mlx5: add mempool registration facilities")
Fixes: fec28ca0e3 ("net/mlx5: support mempool registration")
Signed-off-by: Dmitry Kozlyuk <dkozlyuk@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The lock sh->txpp.mutex was not correctly released on one path
of cleanup function return, potentially causing the deadlock.
Fixes: d133f4cdb7 ("net/mlx5: create clock queue for packet pacing")
Cc: stable@dpdk.org
Signed-off-by: Chengfeng Ye <cyeaa@connect.ust.hk>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
