Adding initial version of rte_flow support for cnxk family device.
Supported rte_flow ops are flow_validate, flow_create, flow_destroy,
flow_flush, flow_query, flow_isolate.
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Adding support to configure NPC on device initialization. This involves
reading the MKEX and initializing the necessary data.
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Once mbufs are transmitted, mbufs are freed by H/W. No mbufs are
accumalated as a pending mbuf.
Hence operation is NOP for cnxk platform.
Signed-off-by: Sunil Kumar Kori <skori@marvell.com>
CN9K and CN10K support platform specific mempool ops.
This patch implements API to validate whether given mempool
ops is supported or not.
Signed-off-by: Sunil Kumar Kori <skori@marvell.com>
Application may choose to enable/disable interrupts on Rx queues
so that application can select its processing if no packets are
available on queues for a longer period.
Signed-off-by: Sunil Kumar Kori <skori@marvell.com>
Add device start and stop operation callbacks for
CN9K and CN10K. Device stop is common for both platforms
while device start as some platform dependent portion where
the platform specific offload flags are recomputed and
the right Rx/Tx burst function is chosen.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Add support for packet type lookup on Rx to translate HW
specific types to RTE_PTYPE_* defines
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
aDD tx queue setup and release for CN9K and CN10K.
Release is common while setup is platform dependent due
to differences in fast path Tx queue structures.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Add Rx queue setup and release op for CN9K and CN10K
SoC. Release is completely common while setup is platform
dependent due to fast path Rx queue structure variation.
Fastpath is platform dependent partly due to core cacheline
size difference.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Add device configuration op for CN9K and CN10K. Most of the
device configuration is common between two platforms except for
some supported offloads.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Add various devargs parsing command line arguments
parsing functions supported by CN9K and CN10K.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Add platform specific probe and remove callbacks for CN9K
and CN10K which use common probe and remove functions.
Register ethdev driver for CN9K and CN10K.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Add build infrastructure and common probe and remove for cnxk driver
which is used by both CN10K and CN9K SoC.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Add common devargs key definition for "bus", "class" and "driver".
Signed-off-by: Xueming Li <xuemingl@nvidia.com>
Acked-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Metadata were stored in the CPU order (little-endian format on x86),
while all the packet header fields are stored in the network order.
That caused wrong results whenever we tried to use metadata value
in the modify_field action: bytes were swapped as a result.
Convert the metadata value into big-endian format before storing it
in the Mellanox NIC to achieve consistent behaviour.
Fixes: 641dbe4fb0 ("net/mlx5: support modify field flow action")
Cc: stable@dpdk.org
Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
MAC addresses are split into 2 parts inside Mellanox NIC:
bits 0-15 are separate from bits 16-47. That makes a copy
from another packet field tricky because any other field
is aligned to 32 bits, not 16. This causes unexpected
results when using the MODIFY_FIELD action with MAC addresses.
Track crossing MAC addresses boundary and arrange a proper
order for the MODIFY_FIELD action involving MAC addresses.
Fixes: 641dbe4fb0 ("net/mlx5: support modify field flow action")
Cc: stable@dpdk.org
Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
A flow rule must not include multiple tunnel layers.
An attempt to create such a rule, for example:
testpmd> flow create .../ vxlan / eth / ipv4 proto is 4 / end <actions>
results in an unclear error.
In the current implementation there is a check for
multiple IPIP tunnels, but not for combination of IPIP
and a different kind of tunnel, such as VXLAN. The fix
is to enhance the above check to use MLX5_FLOW_LAYER_TUNNEL
that consists of all the tunnel masks. The error message
will be "multiple tunnel not supported".
Fixes: 5e33bebdd8 ("net/mlx5: support IP-in-IP tunnel")
Cc: stable@dpdk.org
Signed-off-by: Lior Margalit <lmargalit@nvidia.com>
Acked-by: Ori Kam <orika@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
The timestamp format was not configured correctly for the
receiving queues created via DevX calls. It caused non-UTC
timestamps in CQEs for real time configurations.
Fixes: d61381ad46 ("net/mlx5: support timestamp format")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
The new kernels might add the switch_id attribute to the
Netlink replies and this caused the wrong recognition
of the E-Switch presence. The single uplink device was
erroneously recognized as master and it caused the
extending match for source vport index on all installed
flows, including the default ones, and adding extra hops
in the steering engine, that affected the maximal
throughput packet rate.
The extra check for the new device name format (it supposes
the new kernel) and the device is only one is added. If this
check succeeds the E-Switch presence is considered as wrongly
detected and overridden.
Fixes: 30a86157f6 ("net/mlx5: support PF representor")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
When a counter is destroyed and used for aging action, the driver should
remove the counter object from the age-out list if it is there.
The counter memory of the list entry and of the counter shared
information is shared because, currently, shared counter cannot be used
for aging.
When the support for counter action in action handle API was added, the
counter shared information was reused and moved to be used also for
non-shared case. Wrongly, it is used for aging case too.
Remove the usage of shared information in case of aging.
Fixes: f3191849f2 ("net/mlx5: support flow count action handle")
Cc: stable@dpdk.org
Signed-off-by: Matan Azrad <matan@nvidia.com>
Acked-by: Bing Zhao <bingz@nvidia.com>
When an error appears in the policy creation,
the IDs mapping between the user policy ID to
the driver policy ID is skipped.
Wrongly, the driver tried to clean the mapping in
this case what caused an error.
Skip the clearance in this case.
Fixes: afb4aa4f12 ("net/mlx5: support meter policy operations")
Cc: stable@dpdk.org
Signed-off-by: Li Zhang <lizh@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
The modify field implementation in mlx5 driver has a check to
prevent a copy from a field to the same field. But the level
is not taken into account which prevents a copy from different
tags. Check the level and allow a copy from one tag to another.
Fixes: 641dbe4fb0 ("net/mlx5: support modify field flow action")
Cc: stable@dpdk.org
Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Flow rule pattern may be implicitly expanded by the PMD if the rule
has RSS flow action. The expansion adds network headers to the
original pattern. The new pattern lists all network levels that
participate in the rule RSS action.
The patch fixes expanded pattern for cases when original pattern
included meta items like MARK, TAG, META.
Fixes: c7870bfe09 ("ethdev: move RSS expansion code to mlx5 driver")
Cc: stable@dpdk.org
Signed-off-by: Gregory Etelson <getelson@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
'dev_gen' is a variable to trigger all cores to flush their local caches
once the global MR cache has been rebuilt.
This is due to MR cache's R/W lock can maintain synchronization between
threads:
1. dev_gen and global cache updating ordering inside the lock protected
section does not matter. Because other threads cannot take the lock
until global cache has been updated. Thus, in out of order platform,
even if other agents firstly observe updated dev_gen but global does
not update, they also have to wait the lock. As a result, it is
unnecessary to add a wmb between global cache rebuilding and updating
the dev_gen to keep the memory store order.
2. Store-Release of unlock provides the implicit wmb at the level
visible by software. This makes 'rebuilding global cache' and 'updating
dev_gen' be observed before local_cache starts to be updated by other
agents. Thus, wmb after 'updating dev_gen' can be removed.
Suggested-by: Ruifeng Wang <ruifeng.wang@arm.com>
Signed-off-by: Feifei Wang <feifei.wang2@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
'dev_gen' is a variable to trigger all cores to flush their local caches
once the global MR cache has been rebuilt.
This is due to MR cache's R/W lock can maintain synchronization between
threads:
1. dev_gen and global cache updating ordering inside the lock protected
section does not matter. Because other threads cannot take the lock
until global cache has been updated. Thus, in out of order platform,
even if other agents firstly observe updated dev_gen but global does
not update, they still have to wait the lock. As a result, it is
unnecessary to add a wmb between global cache rebuilding and updating
the dev_gen to keep the memory store order.
2. Store-Release of unlock provides the implicit wmb at the level
visible by software. This makes 'rebuilding global cache' and 'updating
dev_gen' be observed before local_cache starts to be updated by other
agents. Thus, wmb after 'updating dev_gen' can be removed.
Suggested-by: Ruifeng Wang <ruifeng.wang@arm.com>
Signed-off-by: Feifei Wang <feifei.wang2@arm.com>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
This patch uses the new device config ops to get and set
the MAC address if supported.
If a valid MAC address is passed as devarg of the
Virtio-user PMD, the driver will try to store it in the
device config space. Otherwise the one provided in
the device config space will be used, if available.
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Chenbo Xia <chenbo.xia@intel.com>
This patch introduces two virtio-user callbacks to get
and set device's config, and implements it for vDPA
backends.
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Chenbo Xia <chenbo.xia@intel.com>
This patch is preliminary rework to add support for getting
and setting device's config space.
In order to get or set a device config such as its MAC address,
we need to know whether the device itself support the feature,
or if it is emulated by the frontend.
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: Chenbo Xia <chenbo.xia@intel.com>
Restore the original code, where VHOST_SET_FEATURES is applied to
all vhostfds of the device.
Fixes: cc0151b34d ("net/virtio: add virtio-user features ops")
Cc: stable@dpdk.org
Signed-off-by: Thierry Herbelot <thierry.herbelot@6wind.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
IPv4 and IPv6 fragment ptypes are supposed to be separated from IP
other ptypes. New bitmaps for IP fragment ptypes were created, but the
IP fragment ptypes were not deleted from the previous non-frag bitmaps,
which will cause conflicts. This patch removes IP fragment ptypes from
the non-frag bitmaps.
Fixes: 8434528175 ("net/ice/base: support IP fragment RSS and FDIR")
Cc: stable@dpdk.org
Signed-off-by: Ting Xu <ting.xu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
The new allocated mbuf should be updated to the SW
ring.
Fixes: a2b29a7733 ("net/avf: enable basic Rx Tx")
Fixes: b8b4c54ef9 ("net/iavf: support flexible Rx descriptor in normal path")
Cc: stable@dpdk.org
Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Jingjing Wu <jingjing.wu@intel.com>
The original code use a heap pointer after it is freed.
Fixes: 460d167958 ("drivers/net: delete HW rings while freeing queues")
Cc: stable@dpdk.org
Signed-off-by: Dapeng Yu <dapengx.yu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
When creating FDIR rule and parsing the pattern, if IPv4 fragment type is
detected, the flow type is not changed to ICE_FLTR_PTYPE_FRAG_IPV4 from
ICE_FLTR_PTYPE_NONF_IPV4_OTHER. It will cause profile confilict with
other FDIR rules for IPv4 other type.
Fixes: b7e8781de7 ("net/ice: support flow director for IP fragment packet")
Cc: stable@dpdk.org
Signed-off-by: Ting Xu <ting.xu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
The rte_eth_devices array is not in share memory, it should not be
referenced by ice_adapter which is shared by primary and secondary.
Any process set ice_adapter->eth_dev will corrupt another process'
context.
The patch removed the field "eth_dev" from ice_adapter.
Now, when the data paths try to access the rte_eth_dev_data instance,
they should replace adapter->eth_dev->data with adapter->pf.dev_data.
Fixes: f9cf4f8641 ("net/ice: support device initialization")
Cc: stable@dpdk.org
Reported-by: Yixue Wang <yixue.wang@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Tested-by: Yixue Wang <yixue.wang@intel.com>
The flag use_avx2 and use_avx512 are defined as local variables, they
will not be aware by the secondary process, then wrong data path is
selected. Fix the issue by moving them into struct ice_adapter.
Fixes: ae60d3c9b2 ("net/ice: support Rx AVX2 vector")
Fixes: 2d5f6953d5 ("net/ice: support vector AVX2 in Tx")
Fixes: 7f85d5ebcf ("net/ice: add AVX512 vector path")
Cc: stable@dpdk.org
Reported-by: Yixue Wang <yixue.wang@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Tested-by: Yixue Wang <yixue.wang@intel.com>
remove the VSI info from previous aggregator after moving the VSI to a
new aggregator.
Signed-off-by: Victor Raj <victor.raj@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
ice_aq_set_pfc_mode is used to configure DSCP.
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Performance improvement: use a write combining store
instead of a regular mmio write to update queue tail
registers.
Signed-off-by: Gordon Noonan <gordon.noonan@intel.com>
Signed-off-by: Radu Nicolau <radu.nicolau@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
When user configured the flow rule with raw packet via command
"flow_director_filter", it would reset all previous fdir input set
flags with "i40e_flow_set_fdir_inset()".
Ignore to configure the flow input set with raw packet rule used.
Fixes: ff04964ea6 ("net/i40e: fix flow director for common pctypes")
Cc: stable@dpdk.org
Signed-off-by: Steve Yang <stevex.yang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>
In original implementation, device reconfiguration will generate
a new default RSS key if there is no one from user, it is unexpected
when updating a completely unrelated configuration.
This patch makes default RSS key unchanged, during the lifetime of the
DPDK application even if there are multiple reconfigurations.
Fixes: 50370662b7 ("net/ice: support device and queue ops")
Cc: stable@dpdk.org
Signed-off-by: Dapeng Yu <dapengx.yu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
The array rss_key has size 'vf->vf_res->rss_key_size', the array index
should be less than that.
Cc: stable@dpdk.org
Fixes: 69dd4c3d08 ("net/avf: enable queue and device")
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>
The existing log message is missing a space. Modified it to
a more meaningful log as part of this change.
Before this patch:
bnxt_dev_init(): bnxtfound at mem D67E0000, node addr 0x2101112000M
With this patch:
bnxt_dev_init(): Found bnxt device at mem D67E0000, node addr 0x2101112000M
Fixes: 1bf01f5135 ("net/bnxt: prevent device access when device is in reset")
Cc: stable@dpdk.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
On Thor, driver must use HWRM to access the timestamp information.
Driver should not advertise PTP support to application
if PTP information is not available via HWRM commands.
Fixes: 6cbd89f9f3 ("net/bnxt: support PTP for Thor")
Cc: stable@dpdk.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Use the function bnxt_vnic_destroy() to destroy VNIC resources
and thereby eliminate few duplicate code.
Fixes: 8d0a244b40 ("net/bnxt: cleanup VNIC after flow validate")
Fixes: 49d0709b25 ("net/bnxt: delete and flush L2 filters cleanly")
Cc: stable@dpdk.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
During flow destroy, when bnxt_hwrm_tunnel_redirect_free() fails,
driver is not setting flow error using "rte_flow_error_set".
Fixes: 11e5e19695 ("net/bnxt: support redirecting tunnel packets to VF")
Cc: stable@dpdk.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Resources should be freed on error conditions. i.e, VNIC and
VNIC context created in HW and memory allocated in
bnxt_vnic_grp_alloc() should be freed.
Added a new function bnxt_vnic_destroy() to do the cleanup.
This lightweight function can be used in flow destroy/flush
path to avoid duplicate code as well.
Fixes: d24610f7bf ("net/bnxt: allow flow creation when RSS is enabled")
Cc: stable@dpdk.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Also removed a log message which does not convey any
useful information.
Fixes: d24610f7bf ("net/bnxt: allow flow creation when RSS is enabled")
Cc: stable@dpdk.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
In bnxt_flow_validate(), when bnxt_get_unused_filter() fails due to
no filter resources available, driver is not setting flow error using
"rte_flow_error_set".
Also, fixed the error code.
Fixes: 5ef3b79fdf ("net/bnxt: support flow filter ops")
Cc: stable@dpdk.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
The bnxt_vnic_prep() can fail due to multiple reasons.
But when bnxt_vnic_prep() fails, PMD is not returning
the actual error/string to the application.
Fix it by moving the "rte_flow_error_set" to bnxt_vnic_prep()
to set the actual error code.
Fixes: d24610f7bf ("net/bnxt: allow flow creation when RSS is enabled")
Cc: stable@dpdk.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Reviewed-by: Venkat Duvvuru <venkatkumar.duvvuru@broadcom.com>
There is a HW bug that can result in certain stats being reported as
zero.
Workaround this by ignoring stats with a value of zero based on the
previously stored snapshot of the same stat.
This bug mainly manifests in the output of func_qstats as FW aggregrates
each ring's stat value to give the per function stat and if one of
them is zero, the per function stat value ends up being lower than the
previous snapshot which shows up as a zero PPS value in testpmd.
Eliminate invocation of func_qstats and aggregate the per-ring stat
values in the driver itself to derive the func_qstats output post
accounting for the spurious zero stat value.
Bugzilla ID: 641
Fixes: f8168ca0e6 ("net/bnxt: support thor controller")
Cc: stable@dpdk.org
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Lance Richardson <lance.richardson@broadcom.com>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Acked-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
There is a rare hardware bug that can cause a bad opaque value in the RX
or TPA start completion. When this happens, the hardware may have used the
same buffer twice for 2 Rx packets. In addition, the driver might also
crash later using the bad opaque as an index into the ring.
The Rx opaque value is predictable and is always monotonically increasing.
The workaround is to keep track of the expected next opaque value and
compare it with the one returned by hardware during RX and TPA start
completions. If they miscompare, log it, discard the completion,
schedule a ring reset and move on to the next one.
Fixes: 0958d8b643 ("net/bnxt: support LRO")
Cc: stable@dpdk.org
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
The burst receive function should return all packets currently
present in the receive ring up to the requested burst size,
update vector mode receive functions accordingly.
Fixes: 3983583414 ("net/bnxt: support NEON")
Fixes: bc4a000f2f ("net/bnxt: implement SSE vector mode")
Cc: stable@dpdk.org
Signed-off-by: Lance Richardson <lance.richardson@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Make the definition of the table used to map hardware packet type
information to DPDK packet type more generic.
Add macro definitions for constants used in creating table
indices, use these to eliminate raw constants in code.
Add build-time assertions to validate ptype mapping constants.
Signed-off-by: Lance Richardson <lance.richardson@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Check that pointers are valid before using them.
Fixes: 7bc8e9a227 ("net/bnxt: support async link notification")
Cc: stable@dpdk.org
Signed-off-by: Thierry Herbelot <thierry.herbelot@6wind.com>
Acked-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
The VF reset can be triggered by the PF reset event, then the PCI bus
master will be cleared, the VF will be not allowed to issue any Memory
or I/O Requests.
So after the reset event is detected, always enable the PCI bus master.
And if failed, the device or system may be in an invalid state, so keep
the VF reset state to mark it as I/O error.
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>
The VF reset can be triggered by the PF reset event, then the PCI bus
master will be cleared, the VF will be not allowed to issue any Memory
or I/O Requests.
So after the reset event is detected, always enable the PCI bus master.
And if failed, the device or system may be in an invalid state, so keep
the VF reset state to mark it as I/O error.
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>
The modify_field flow API assumes that the META item is 32 bits wide.
But the C register that is used for META item can be 16 or 32 bits
wide depending on kernel and firmware configurations.
Take this into consideration and use the appropriate META width.
Fixes: 641dbe4fb0 ("net/mlx5: support modify field flow action")
Cc: stable@dpdk.org
Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
In the past, all the queues and other hardware objects were created
through Verbs interface. Currently, most of the objects creation are
migrated to Devx interface by default, including queues. Only when
the DV is disabled by device arg or eswitch is enabled, all or some
of the objects are created through Verbs interface.
When using Devx interface to create queues, the kernel driver
behavior is different from the case using Verbs. The Tx loopback
cannot work properly even if the Tx and Rx queues are configured
with loopback attribute. To fix the support self loopback for Tx, a
Verbs dummy queue pair needs to be created to trigger the kernel to
enable the global loopback capability.
This is only required when TIR is created for Rx and loopback is
needed. Only CQ and QP are needed for this case, no WQ(RQ) needs to
be created.
Bugzilla ID: 645
Fixes: 6deb19e1b2 ("net/mlx5: separate Rx queue object creations")
Cc: stable@dpdk.org
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
When the port is link down state, it is meaningless to display the
port link speed. It should be an undefined state.
Fixes: 59fad0f321 ("net/hns3: support link update operation")
Cc: stable@dpdk.org
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Whether the enable bit of the pfc ("pfc_en") is changed or not is one of
the conditions for reconfiguring the DCB. Currently, pfc_en is not
rolled back when DCB configuration fails. This patch fixes it.
Fixes: 62e3ccc2b9 ("net/hns3: support flow control")
Cc: stable@dpdk.org
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Currently, the DCB configuration takes effect in the dev_start stage, and
the mapping between TCs and queues are also updated in this stage.
However, the DCB configuration is delivered in the dev_configure stage.
If the configuration fails, it should be intercepted in this stage. If
the configuration succeeds, the user should be able to obtain the
corresponding updated information, such as the mapping between TCs and
queues. So this patch moves DCB configuration to dev_configure.
Fixes: 62e3ccc2b9 ("net/hns3: support flow control")
Cc: stable@dpdk.org
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Packet buffer allocation and hardware pause configuration fail normally
when a reset occurs. If the execution fails, rollback of the packet
buffer still fails. So this rollback is meaningless.
Fixes: 62e3ccc2b9 ("net/hns3: support flow control")
Cc: stable@dpdk.org
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Currently, the "requested_fc_mode" lacks rollback when enabling link
FC or PFC fails.
For example, this may result an incorrect FC mode after a reset.
Fixes: d4fdb71a0e ("net/hns3: fix flow control mode")
Fixes: 62e3ccc2b9 ("net/hns3: support flow control")
Cc: stable@dpdk.org
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
The Rx/Tx queue numbers should be greater than TC number, this patch adds
this check for PF before updating the mapping between TC and queue.
Fixes: a951c1ed3a ("net/hns3: support different numbers of Rx and Tx queues")
Fixes: 76d794566d ("net/hns3: maximize queue number")
Cc: stable@dpdk.org
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
The net/vhost PMD does not comply with the ethdev offload API as it does
not report Rx/Tx offload capabilities wrt TSO and checksum offloading.
On the other hand, the net/vhost PMD lets guest negotiates TSO and
checksum offloading.
Changing the behavior for Rx/Tx offload flags handling won't
improve/fix this situation and will break applications that might have
been relying on implicit support of TSO in this driver.
Revert this behavior change until we have a complete fix.
Fixes: ca7036b4af ("vhost: fix offload flags in Rx path")
Cc: stable@dpdk.org
Signed-off-by: David Marchand <david.marchand@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
A meter may handle Rx queue reference in his sub-policies.
In stop operation, all the Rx queues are released.
Wrongly, the meter reference was not released before
destroying the Rx queues what cause an error in stop.
Release the Rx queues meter references in stop operation.
Fixes: fc6ce56bba ("net/mlx5: prepare sub-policy for flow with meter")
Signed-off-by: Li Zhang <lizh@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Currently, the counter offset support is discovered by creating the
rule with invalid offset counter and drop action in root table. If
the rule creation fails with EINVAL errno, that mean counter offset
is not supported in root table.
However, drop action may not be supported in some rdma-core version
in root table. In this case, the discover code will not work properly.
This commits changes flow attribute to egress. That removes all the
extra fate actions in the flow to avoid any unsupported fate actions
make the discover code fail time to time.
Fixes: 994829e695 ("net/mlx5: remove single counter container")
Cc: stable@dpdk.org
Signed-off-by: Suanming Mou <suanmingm@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Currently, configuring a mlx device, it will allocate its
own process private in mlx5_proc_priv_init() and only frees
it when closing the device. This will lead to a memory leak,
when a device is configured repeatedly.
For example:
for(...)
do
rte_eth_dev_configure
rte_eth_rx_queue_setup
rte_eth_tx_queue_setup
rte_eth_dev_start
rte_eth_dev_stop
done
Fixes: 120dc4a7dc ("net/mlx5: remove device register remap")
Cc: stable@dpdk.org
Signed-off-by: Yunjian Wang <wangyunjian@huawei.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Currently, configuring a mlx device, it will allocate its
own process private in mlx5_proc_priv_init() and only frees
it when closing the device. This will lead to a memory leak,
when a device is configured repeatedly.
For example:
for(...)
do
rte_eth_dev_configure
rte_eth_rx_queue_setup
rte_eth_tx_queue_setup
rte_eth_dev_start
rte_eth_dev_stop
done
Fixes: 97d37d2c1f ("net/mlx4: remove device register remap")
Cc: stable@dpdk.org
Signed-off-by: Yunjian Wang <wangyunjian@huawei.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Depending on the breakout mode of the physical ports the internal NFP
port number might differ from the actual physical port number. Prior to
this patch the physical port number was used when making configuration
changes to the physical ports (enable, admin up etc). After this change
the internal port number is now correctly used for configuration
changes.
Fixes: 5e15e799d6 ("net/nfp: create separate entity for PF device")
Cc: stable@dpdk.org
Signed-off-by: Heinrich Kuhn <heinrich.kuhn@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
When getting meter flow_id bits, there's an issue that not handling
correctly if flow_id is 0.
This fix this issue that when flow_id is 0, treat it as 1 bit.
Fixes: 83306d6c46 ("net/mlx5: fix meter statistics")
Cc: stable@dpdk.org
Signed-off-by: Shun Hao <shunh@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
One of the user parameters for the flow AGE action is the
action context. This context should be provided back to the
user when the action is aged-out.
While this context is NULL, a default value should be provided
by the PMD: the rte_flow pointer in case of rte_flow_create API
and the action pointer in case of the rte_flow_action_handle API.
The default for rte_flow_action_handle was set correctly,
while in case of rte_flow_create it wrongly remained NULL.
This patch set the default value for rte_flow_create case to be
the rte_flow pointer.
Fixes: f9bc5274a6 ("net/mlx5: allow age modes combination")
Cc: stable@dpdk.org
Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Current the ASO age action was supported in the non-root table,
and the counter based age action was be used in the root table.
The FDB table skips group 0 on MLX5 PMD by adding implicit rule
that jump to non-root table, but PMD code use the original group
value for checking.
This patch adds the transfer checking for ASO age action.
Fixes: f9bc5274a6 ("net/mlx5: allow age modes combination")
Cc: stable@dpdk.org
Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Currently RSS expansion only supports GRE and GRE KEY.
This patch adds RSS expansion for NVGRE item so PMD can expand flow item
correctly.
Fixes: ea81c1b816 ("net/mlx5: fix NVGRE matching")
Cc: stable@dpdk.org
Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
Acked-by: Xiaoyu Min <jackmin@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
While there's mirror action prior to the meter action in the E-Switch
flow, means that the packets should be duplicated into port firstly,
and then do meter and send to the original destination.
MLX5 PMD will split the above E-Switch flow into two sub flows,
similar as mirror with modify action before.
Fixes: 07627fbf15 ("net/mlx5: support E-Switch mirroring with modify action")
Cc: stable@dpdk.org
Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
In case of bonding, orchestrator wants to use same devargs for LAG and
non-LAG scenario to probe representor on PF1 using PF1 PCI address
like "<DBDF_PF1>,representor=pf1vf[0-3]".
This patch changes PCI address check policy to allow PF1 PCI address for
representors on PF1.
Note: detaching PF0 device can't remove representors on PF1. It's
recommended to use primary(PF0) PCI address to probe representors on
both PFs.
Fixes: f926cce3fa ("net/mlx5: refactor bonding representor probing")
Cc: stable@dpdk.org
Signed-off-by: Xueming Li <xuemingl@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The memory barrier is used to ensure that the response is returned
only after the Tx/Rx function is set, it should place after the Rx/Tx
function is set.
Fixes: 2aac5b5d11 ("net/mlx5: sync stop/start with secondary process")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The memory barrier is used to ensure that the response is returned
only after the Tx/Rx function is set, it should place after the Rx/Tx
function is set.
Fixes: 0203d33a10 ("net/mlx4: support secondary process")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Reproduced with '--buildtype=debugoptimized' config,
compiler version: gcc (GCC) 12.0.0 20210509 (experimental)
There are multiple build errors, like:
In file included from ../drivers/net/tap/tap_flow.c:13:
In function ‘rte_jhash_2hashes’,
inlined from ‘rte_jhash’ at ../lib/hash/rte_jhash.h:284:2,
inlined from ‘tap_flow_set_handle’ at
../drivers/net/tap/tap_flow.c:1306:12,
inlined from ‘rss_enable’ at ../drivers/net/tap/tap_flow.c:1909:3,
inlined from ‘priv_flow_process’ at
../drivers/net/tap/tap_flow.c:1228:11:
../lib/hash/rte_jhash.h:238:9:
warning: ‘flow’ may be used uninitialized [-Wmaybe-uninitialized]
238 | __rte_jhash_2hashes(key, length, pc, pb, 1);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../drivers/net/tap/tap_flow.c: In function ‘priv_flow_process’:
../lib/hash/rte_jhash.h:81:1: note: by argument 1 of type ‘const void *’
to ‘__rte_jhash_2hashes.constprop’ declared here
81 | __rte_jhash_2hashes(const void *key, uint32_t length, uint32_t *pc,
| ^~~~~~~~~~~~~~~~~~~
../drivers/net/tap/tap_flow.c:1028:1: note: ‘flow’ declared here
1028 | priv_flow_process(struct pmd_internals *pmd,
| ^~~~~~~~~~~~~~~~~
Fix strict aliasing rule by using union.
Bugzilla ID: 690
Fixes: de96fe68ae ("net/tap: add basic flow API patterns and actions")
Cc: stable@dpdk.org
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Kevin Traynor <ktraynor@redhat.com>
Reproduced with '--buildtype=debugoptimized' config,
compiler version: gcc (GCC) 12.0.0 20210509 (experimental)
There are multiple build errors, like:
../drivers/net/ice/base/ice_switch.c: In function ‘ice_add_marker_act’:
../drivers/net/ice/base/ice_switch.c:3727:15:
warning: array subscript ‘struct ice_aqc_sw_rules_elem[0]’
is partly outside array bounds of ‘unsigned char[52]’
[-Warray-bounds]
3727 | lg_act->type = CPU_TO_LE16(ICE_AQC_SW_RULES_T_LG_ACT);
| ^~
In file included from ../drivers/net/ice/base/ice_type.h:52,
from ../drivers/net/ice/base/ice_common.h:8,
from ../drivers/net/ice/base/ice_switch.h:8,
from ../drivers/net/ice/base/ice_switch.c:5:
../drivers/net/ice/base/ice_osdep.h:209:29:
note: referencing an object of size 52 allocated by ‘rte_zmalloc’
209 | #define ice_malloc(h, s) rte_zmalloc(NULL, s, 0)
| ^~~~~~~~~~~~~~~~~~~~~~~
../drivers/net/ice/base/ice_switch.c:3720:50:
note: in expansion of macro ‘ice_malloc’
lg_act = (struct ice_aqc_sw_rules_elem *)ice_malloc(hw, rules_size);
These errors are mainly because allocated memory is cast to
"struct ice_aqc_sw_rules_elem *" but allocated size is less than the size
of "struct ice_aqc_sw_rules_elem".
"struct ice_aqc_sw_rules_elem" has multiple other structs has unions,
based on which one is used allocated memory being less than the size of
"struct ice_aqc_sw_rules_elem" is logically correct but compiler is
complaining about it.
Since the allocation is done explicitly and both producer and consumer
are internal, safe to ignore the warnings. Also to prevent any side
affect disabling the compiler warning for now, until proper fix done.
Reducing the warning disable to gcc >= 11 version.
Bugzilla ID: 678
Fixes: c7dd159311 ("net/ice/base: add virtual switch code")
Fixes: 02acdce2f5 ("net/ice/base: add MAC filter with marker and counter")
Fixes: f89aa3affa ("net/ice/base: support removing advanced rule")
Cc: stable@dpdk.org
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Acked-by: Haiyue Wang <haiyue.wang@intel.com>
Fixed speed mode is not supported currently, this patch
removes configurations for this mode and adds fault handling
for ETH_LINK_SPEED_FIXED.
Fixes: 4f09bc55ac ("net/igc: implement device base operations")
Cc: stable@dpdk.org
Signed-off-by: Alvin Zhang <alvinx.zhang@intel.com>
Acked-by: Junfeng Guo <junfeng.guo@intel.com>
The kernel driver supports VF RSS configuration message
"VIRTCHNL_OP_GET_RSS_HENA_CAPS and VIRTCHNL_OP_SET_RSS_HENA",
this patch adds PMD support for these messages.
Fixes: b81295c474 ("net/i40e: add user callback for VF to PF message")
Cc: stable@dpdk.org
Signed-off-by: Alvin Zhang <alvinx.zhang@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>
In the loop, when the index of array "vsi->rss_key" is equal
to "vsi->rss_key_size", the array will be accessed out of bounds.
Fixes: 50370662b7 ("net/ice: support device and queue ops")
Cc: stable@dpdk.org
Signed-off-by: Jie Wang <jie1x.wang@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
The QW0 of Tx context descriptor should be reset to 0, otherwise the
previous hardware writeback value may pollute the next context descriptor
write.
Fixes: a2b29a7733 ("net/avf: enable basic Rx Tx")
Cc: stable@dpdk.org
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Signed-off-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
For dev_ops not supported by the secondary process, either return -EPERM
or return without doing anything. In both cases log a warning.
It's still application's responsibility to avoid calls like that and
those changes are for debugging/informational purposes.
Signed-off-by: Stanislaw Kardach <kda@semihalf.com>
Reviewed-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Shay Agroskin <shayagr@amazon.com>
rte_pci_device and rte_eth_dev are process-local structures. Therefore
ena_adapter::pdev and ena_adapter::rte_dev cannot be used universally.
Both ena_timer_wd_callback and ena_interrupt_handler_rte needs access to
the rte_eth_dev, but as they are being setup and executed in the primary
process, it is safe to pass there the same pointer, which is used for
the device configuration.
In all other cases, except the eth_ena_dev_init(), the rte_eth_dev_data
is used instead.
Signed-off-by: Stanislaw Kardach <kda@semihalf.com>
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Shay Agroskin <shayagr@amazon.com>
To make it possible to the app to determine if the hash was calculated
for the packet or not, the PKT_RX_RSS_HASH should be set in the mbuf's
ol_flags.
As the PMD wasn't setting that, the application couldn't check if there
is a hash in a proper way.
The hash is valid only if it's UDP or TCP and the IP packet wasn't
fragmented.
Fixes: e5df9f33db ("net/ena: fix passing RSS hash to mbuf")
Cc: stable@dpdk.org
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Amit Bernstein <amitbern@amazon.com>
Thread argument changed to wrong value during thread name addition,
fixing that bug.
Fixes: fdefe038eb ("net/ark: set generator delay thread name")
Signed-off-by: Ed Czeck <ed.czeck@atomicrules.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
Tunnel offload API requires application to query PMD for specific flow
items and actions. Application uses these PMD specific elements to
build flow rules according to the tunnel offload model.
The model does not restrict private elements location in a flow rule,
but the current MLX5 PMD implementation expects that tunnel offload
rule will begin with PMD specific elements.
The patch removes that placement limitation.
Fixes: 4ec6360de3 ("net/mlx5: implement tunnel offload")
Cc: stable@dpdk.org
Signed-off-by: Gregory Etelson <getelson@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The admin-configured vNIC settings (i.e. via CIMC or UCSM) now include
Geneve offload. Use that setting to decide whether to enable or
disable Geneve offload and remove the devarg 'geneve-opt'.
Also, the firmware now allows the driver to change the Geneve port
number. So extend udp_tunnel_port_{add,del} to accept Geneve port, in
addition to VXLAN.
Fixes: 93fb21fdbe ("net/enic: enable overlay offload for VXLAN and GENEVE")
Cc: stable@dpdk.org
Signed-off-by: John Daley <johndale@cisco.com>
Reviewed-by: Hyong Youb Kim <hyonkim@cisco.com>
A terminated pthread should be joined or detached so that its associated
resources are released.
The "ice-reset-<vf_id>" threads are used to service some reset task in
the background, but they are never joined by the thread that created
them.
The easiest solution is to detach new threads.
The Windows EAL did not provide a pthread_detach wrapper but there is no
resource to release for Windows threads, so add an empty wrapper.
Fixes: 3b3757bda3 ("net/ice: get VF hardware index in DCF")
Cc: stable@dpdk.org
Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Haiyue Wang <haiyue.wang@intel.com>
Acked-by: Dmitry Kozlyuk <dmitry.kozliuk@gmail.com>
A terminated pthread should be joined or detached so that its associated
resources are released.
The "ark-delay-pg" thread is just used to delay some task but it is never
joined by the thread that created it.
The easiest solution is to detach the new thread.
Fixes: 727b3fe292 ("net/ark: integrate PMD")
Cc: stable@dpdk.org
Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Ed Czeck <ed.czeck@atomicrules.com>
If the FEC mode was not supported, it should return error code.
This patch also adds a space when log error info.
Fixes: 9bf2ea8dbc ("net/hns3: support FEC")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
The memory barrier is used to ensure that the response is returned
only after the Tx/Rx function is set, it should place after the Rx/Tx
function is set.
Fixes: 23d4b61fee ("net/hns3: support multiple process")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
This secondary process should not send request to start/stop Rx/Tx,
this patch fixes it.
Fixes: 23d4b61fee ("net/hns3: support multiple process")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
The mailbox message id is uint8_t, but the unsupported mailbox message
id was logged by uint16.
Fixes: 463e748964 ("net/hns3: support mailbox")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
The TM QCN error event should report by RAS other than MSIX.
Also this patch adds fifo int enable configuration before the TM QCN
error event is enabled.
Fixes: f53a793bb7 ("net/hns3: add more hardware error types")
Fixes: 3903c05382 ("net/hns3: remove read when enabling TM QCN error event")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Support to enable and disable QINQ hardware strip, when configure VLAN
offload with QINQ strip mask. If there are packets have QINQ tag to RSS,
users should enable QINQ strip before configuring the RSS.
Fixes: 220b0e49bc ("net/txgbe: support VLAN")
Cc: stable@dpdk.org
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
The hardware doesn't support counting the number of bytes that through
the fdir rule. Therefore, the corresponding out parameters (e.g.
bytes_set/bytes) is set to zero.
Fixes: fcba820d9b ("net/hns3: support flow director")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Currently in the VF reset scenario, the VF performs the set
alive operation before restoring the configuration completed,
which may cause the hardware to work in an abnormal state.
This patch fix this problem by set VF alive after restoring
the configuration is completed.
Fixes: a5475d61fa ("net/hns3: support VF")
Cc: stable@dpdk.org
Signed-off-by: Hongbo Zheng <zhenghongbo3@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
The fdir hash map hold the pointers of fdir rule elements, it needs to
be set to NULL when clear all fdir rules.
Fixes: fcba820d9b ("net/hns3: support flow director")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
If clear FDIR rules fail, the error code was logged, but the error code
was useless because it was the sum of all fail code.
This patch fixes it by log the success cnt and fail cnt.
Fixes: fcba820d9b ("net/hns3: support flow director")
Fixes: 8eed8acc81 ("net/hns3: add error code to some logs")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Prior to this change, two implementations of rx_syscall_handler
existed although only one was needed (for the zero copy path which
is only available from kernel 5.4 and onwards). Remove the second
definition from compat.h and move the first definition back to where
it is called in the Rx function. Doing this removes a build warning
on kernels before 5.4 which complained about the second function
being defined but not used.
Fixes: 2aa51cdd55 ("net/af_xdp: fix trigger for syscall on Tx")
Signed-off-by: Ciara Loftus <ciara.loftus@intel.com>
Before this change the dev_infos callback always reported RSS
capabilities regardless of whether the capability is supported by the
device or not. First check the capabilities field in the BAR of the
device and advertise RSS functionality accordingly.
Fixes: 8b945a7f7d ("drivers/net: update Rx RSS hash offload capabilities")
Cc: stable@dpdk.org
Signed-off-by: Heinrich Kuhn <heinrich.kuhn@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
This version update contains:
* memcpy mapping to the dpdk-optimized version.
* ena_com (HAL) update to the latest version (from 18.09.2020).
* Bug fixes for the large LLQ headers and devargs parsing.
* Bug fix for the default ring size.
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Remove invalid ring size alignment logic and add default Rx and Tx port
ring sizes to the device info spec.
The logic in lines 1297 and 1371 is invalid. The
RTE_ETH_DEV_FALLBACK_RX_RINGSIZE (and the TX counterpart) is a value
that rte_eth_rx_queue_setup() will set if
dev_info.default_rxportconf.ring_size is 0 and user provided 0 in
nb_rx_desc argument. However the current code treats it as a hint for
the PMD to change the ring size to internal defaults.
Additionally since the ENA_DEFAULT_RING_SIZE is defined, report it in
the device capabilities so that both rte_ethdev code and the user can
utilize it for device configuration.
Fixes: ea93d37eb4 ("net/ena: add HW queues depth setup")
Cc: stable@dpdk.org
Signed-off-by: Stanislaw Kardach <kda@semihalf.com>
Reviewed-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Shay Agroskin <shayagr@amazon.com>
pthread_cond_timedwait() may spuriously wakeup according to POSIX.
Therefore it is required to check whether predicate is actually true
before finishing the waiting loop.
Signed-off-by: Stanislaw Kardach <kda@semihalf.com>
Reviewed-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Shay Agroskin <shayagr@amazon.com>
swap*_*_le() functions are not used anywhere and besides there are rte
alternatives already present.
Fixes: 1173fca25a ("ena: add polling-mode driver")
Cc: stable@dpdk.org
Signed-off-by: Stanislaw Kardach <kda@semihalf.com>
Reviewed-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Shay Agroskin <shayagr@amazon.com>
As the documentation of rte_kvargs_parse() states, the valid_keys
argument must be NULL terminated. Lack of this feature may cause
segmentation fault if the passed devarg will be different then the
supported value.
Fixes: 8a7a73f26c ("net/ena: support large LLQ headers")
Cc: stable@dpdk.org
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Shay Agroskin <shayagr@amazon.com>
Acked-by: Ferruh Yigit <ferruh.yigit@intel.com>
The code incorrectly checked the return value of comparison when parsing
the argument key name. The return value of strcmp should be compared
to 0 to identify a match.
Fixes: 8a7a73f26c ("net/ena: support large LLQ headers")
Cc: stable@dpdk.org
Signed-off-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Shay Agroskin <shayagr@amazon.com>
Reviewed-by: Michal Krawczyk <mk@semihalf.com>
1. As memzone uses unique names, changed alloc coherent macro to use
64 bit size atomic variable to increase the memzone name space
2. "handle" param name change to be consistent with other macros
3. Variable definition displacement
4. Backslash alignment to column 80
Signed-off-by: Amit Bernstein <amitbern@amazon.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Shay Agroskin <shayagr@amazon.com>
Reviewed-by: Michal Krawczyk <mk@semihalf.com>
As in the v20.11 rte_prefetch0_write API was added, it should be used
in the platform file for the definition of the macro prefetchw, instead
of using simply prefetch0.
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Amit Bernstein <amitbern@amazon.com>
This feature wasn't used by any of the drivers. Because of that, it was
removed.
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Shay Agroskin <shayagr@amazon.com>
The RSS indirection table shouldn't be included as a part of the
device features context.
If the driver has to acquire it, it can be done using the separate API
whenever the RSS configuration is needed.
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Amit Bernstein <amitbern@amazon.com>
Although the ENA DPDK PMD doesn't have to perform any actions for
destroying the wait event, some other platforms have to.
The macro "ENA_WAIT_EVENT_DESTROY" was renamed to
"ENA_WAIT_EVENTS_DESTROY" and also whole implementation responsible for
that was moved to a separate function for better readability.
Fixes: 3adcba9a89 ("net/ena: update HAL to the newer version")
Cc: stable@dpdk.org
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Guy Tzalik <gtzalik@amazon.com>
To silence error messages from the static code analysis, make the type
conversions explicit where they're intended.
Also fix the type for the DMA width value.
Fixes: 99ecfbf845 ("ena: import communication layer")
Cc: stable@dpdk.org
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Guy Tzalik <gtzalik@amazon.com>
List of changes:
* Comment style was adjusted for the functions
* The keys_num at "struct ena_admin_feature_rss_flow_hash_control" was
renamed to the key_parts to better describe it's meaning
* The RSS indirection table was called "REDIRECTION" -> changed to
INDIRECTION
* Change AENQ field "syndrom" -> "syndrome"
* Calculate number of the RSS key parts or whole key by using the
common way: sizeof of the first element of the RSS key
* Add description of the "enum ena_admin_aq_feature_id"
* Rename "map_rx_buf_bidirectional" field as "rx_buf_mirroring"
* Other minor style fixes (remove extra spaces, add missing line break,
improve indentation)
* Remove unused macros ENA_ADMIN_EXTRA_PROPERTIES_*
* Restructure the "if {} else if {} else" conditional statement for
setting up the meta descriptor
Fixes: 99ecfbf845 ("ena: import communication layer")
Fixes: b68309be44 ("net/ena/base: update communication layer for the ENAv2")
Fixes: b2b02edeb0 ("net/ena/base: upgrade HAL for new HW features")
Cc: stable@dpdk.org
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Guy Tzalik <gtzalik@amazon.com>
Some platforms may make use of the device argument to make the logs more
verbose and specific for the appropriate device.
As it's not used by the ENA DPDK PMD for the logging, the type is just
defined, but never used.
It may be reconsidered to change this in the future by adding port ID
to the message logs, but as for now the logging behavior won't change.
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Guy Tzalik <gtzalik@amazon.com>
Instead of using 'queue' for struct ena_com_admin_queue and 'dev' for
struct ena_com_dev variables, use more descriptive 'admin_queue' and
'ena_dev'.
This also unifies the names of variables of the type struct ena_com_dev
in the driver.
Signed-off-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Guy Tzalik <gtzalik@amazon.com>
memcpy is now mapped to rte_memcpy macro on x86 architectures.
Fixes: 9ba7981ec9 ("ena: add communication layer for DPDK")
Cc: stable@dpdk.org
Signed-off-by: Igor Chauskin <igorch@amazon.com>
Reviewed-by: Michal Krawczyk <mk@semihalf.com>
Reviewed-by: Artur Rojek <ar@semihalf.com>
This patch fixes a bug where firmware version was not
copied from ibv_device_attr structure into mlx5_dev_attr
structure, resulting in inability to read firmware
version.
Fixes: e85f623e13 ("net/mlx5: remove attributes dependency on Verbs")
Cc: stable@dpdk.org
Signed-off-by: Kamil Vojanec <xvojan00@stud.fit.vutbr.cz>
Acked-by: Matan Azrad <matan@nvidia.com>
There was a typo - the device context was wrongly provided
instead of counter's one for the DevX query operation.
Fixes: e6988afdc7 ("net/mlx5: fix imissed statistics")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
If the Send Queue (backing one for PMD Tx queue) the was
created with DevX API the doorbell record offset for the
producer index field was incorrect. If hardware missed the
doorbell register write event the wrong content of doorbell
record might cause queue malfunction. For the Send Queues
created with Verbs API the doorbell record offset was
configured correctly.
Fixes: 86d259cec8 ("net/mlx5: separate Tx queue object creations")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The support of RSS expansion for the flows with IPv6 GRE item was added
to mlx5 PMD. And the GRE KEY item support in expansion was missed
and the flows with GRE and GRE KEY items were expanded in the wrong
way causing the flow creation failure.
This patch adds the RSS expansion support for GRE KEY and mlx5 PMD
performs RSS expansion correctly.
Fixes: 048f0d45e3 ("net/mlx5: support RSS expansion for IPv6 GRE")
Cc: stable@dpdk.org
Signed-off-by: Jiawei Wang <jiaweiw@nvidia.com>
Acked-by: Xiaoyu Min <jackmin@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The crypto driver added new fields to the mkey attributes struct:
crypto_en and set_remote_rw.
The entire mkey struct was not initialized, only specific fields in it,
which caused the new added fields not to be initialized resulting in a
mkey creation error.
This is fixed by initializing the entire mkey attributes struct to 0
which will prevent this issue from reoccurring if any fields are added
to the mkey struct in the future.
Fixes: 0111a74e13 ("common/mlx5: adjust DevX mkey fields for crypto")
Signed-off-by: Shiri Kuzin <shirik@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Originally, the default RSS for GTPU is inner fields. Now, we hope outer
RSS for GTPU to be the default.
Since RSS for IPv4, RSS for IPv6, RSS for UDP and RSS for TCP can cover
the cases of outer RSS for GTPU, this patch deletes redundant default
RSS configurations for GTPU.
Signed-off-by: Wenjun Wu <wenjun1.wu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
To enable IP fragment RSS hash, ICE_FLOW_SEG_HDR_IPV_FRAG is added to the
IP RSS configuration template, together with ICE_FLOW_SEG_HDR_IPV_OTHER.
It will cause error when associating flow profile. And packet id field
for RSS is not correctly added when IP fragment is enabled. To fix this
issue, this patch only selects one of the above two segment header types
based on RSS types.
Fixes: f1ea76eb63 ("net/ice: support RSS hash for IP fragment")
Cc: stable@dpdk.org
Signed-off-by: Ting Xu <ting.xu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
If vector mode is not allowed for Tx, no need to perform vector
related setup for Tx queue.
The patch deferred vector setup for Tx queue to the place that
vector mode is confirmed to be allowed.
Fixes: 28f9002ab6 ("net/ice: add Tx AVX512 offload path")
Signed-off-by: Alvin Zhang <alvinx.zhang@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
This is reported by our internal covscan:
1. dpdk-20.11/drivers/net/ice/base/ice_switch.c:4214: sign_extension:
Suspicious implicit sign extension: "s_rule_size" with type "u16" (16
bits, unsigned) is promoted in "num_unicast * s_rule_size" to type "int"
(32 bits, signed), then sign-extended to type "unsigned long" (64 bits,
unsigned).
If "num_unicast * s_rule_size" is greater than 0x7FFFFFFF, the upper bits
of the result will all be 1.
# 4212| s_rule_size = ICE_SW_RULE_RX_TX_ETH_HDR_SIZE;
# 4213| s_rule = (struct ice_aqc_sw_rules_elem *)
# 4214|-> ice_calloc(hw, num_unicast, s_rule_size);
# 4215| if (!s_rule) {
# 4216| status = ICE_ERR_NO_MEMORY;
Even if this condition is not likely to happen, in any case, it is more
straightforward to rely on the existing rte_calloc.
Fixes: 5f0978e962 ("net/ice/base: add OS specific implementation")
Cc: stable@dpdk.org
Signed-off-by: David Marchand <david.marchand@redhat.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Let's try to enforce the convention where most drivers use a pmd. logtype
with their class reflected in it, and libraries use a lib. logtype.
Introduce two new macros:
- RTE_LOG_REGISTER_DEFAULT can be used when a single logtype is
used in a component. It is associated to the default name provided
by the build system,
- RTE_LOG_REGISTER_SUFFIX can be used when multiple logtypes are used,
and then the passed name is appended to the default name,
RTE_LOG_REGISTER is left untouched for existing external users
and for components that do not comply with the convention.
There is a new Meson variable log_prefix to adapt the default name
for baseband (pmd.bb.), bus (no pmd.) and mempool (no pmd.) classes.
Note: achieved with below commands + reverted change on net/bonding +
edits on crypto/virtio, compress/mlx5, regex/mlx5
$ git grep -l RTE_LOG_REGISTER drivers/ |
while read file; do
pattern=${file##drivers/};
class=${pattern%%/*};
pattern=${pattern#$class/};
drv=${pattern%%/*};
case "$class" in
baseband) pattern=pmd.bb.$drv;;
bus) pattern=bus.$drv;;
mempool) pattern=mempool.$drv;;
*) pattern=pmd.$class.$drv;;
esac
sed -i -e 's/RTE_LOG_REGISTER(\(.*\), '$pattern',/RTE_LOG_REGISTER_DEFAULT(\1,/' $file;
sed -i -e 's/RTE_LOG_REGISTER(\(.*\), '$pattern'\.\(.*\),/RTE_LOG_REGISTER_SUFFIX(\1, \2,/' $file;
done
$ git grep -l RTE_LOG_REGISTER lib/ |
while read file; do
pattern=${file##lib/};
pattern=lib.${pattern%%/*};
sed -i -e 's/RTE_LOG_REGISTER(\(.*\), '$pattern',/RTE_LOG_REGISTER_DEFAULT(\1,/' $file;
sed -i -e 's/RTE_LOG_REGISTER(\(.*\), '$pattern'\.\(.*\),/RTE_LOG_REGISTER_SUFFIX(\1, \2,/' $file;
done
Signed-off-by: David Marchand <david.marchand@redhat.com>
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: Bruce Richardson <bruce.richardson@intel.com>
The variable "dev" is not used in hns3_get_tx_prep_needed()
in the case of RTE_LIBRTE_ETHDEV_DEBUG:
drivers/net/hns3/hns3_rxtx.c:4213:45: error: unused parameter ‘dev’
Fixes: d7ec2c0765 ("net/hns3: select Tx prepare based on Tx offload")
Cc: stable@dpdk.org
Reported-by: David Marchand <david.marchand@redhat.com>
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Acked-by: David Marchand <david.marchand@redhat.com>
After creating a connection tracking context, it can be used between
two ports. For each port, the flow for one direction traffic will
be created.
The context can only be shared between the owner port and the peer
port that was specified when being created. Only the owner port
could update the context or query it in current implementation.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The item of ASO connection tracking will be translated into the
register value when matching. The validation of this item has no
dependency on other layers, since the flow including this item
should be jumped from another group. All the layers checking was
already done in the previous groups. Only the state bits conflict
should be checked.
It is assumed that the flow with CT item will always work on the
TCP traffic.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The validation of a CT action contains two parts. The first is the
CT action configurations parameter. When creating a CT action
context, some members need to be verified.
The second is that when creating a flow, the DR action of CT should
be validated with other actions and items as well. Currently, only
the TCP protocol support connection tracking.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
When updating a connection tracking context, two separate parts
could be updated.
First, the direction. This will only update the traffic direction
recorded in the software for flow creation.
Second, the TCP parameters. The hardware context will be updated
via the WQE. This update will be blocked until the hardware status
is updated and ready for the next flow creation.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The return register of the DR action will be used for matching.
After the ASO CT checking of a TCP packet, the syndrome is filled in
the register. Only the 8 LSB should be used. A converting from
RTE_FLOW_CONNTRACK_FLAG* to the syndrome should be done after
checking the spec and mask fields.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
When creating a flow with this action context for CT, it needs to be
translated in 2 levels.
First, retrieve from action context to rte_flow action.
Second, translate it to the corresponding DR action with traffic
direction that was specified when creating or updating via
rte_flow_action_handle* API.
Before using the DR action in a flow, the CT context should be
available to use in the hardware. A synchronization is done before
inserting the flow rule with CT action to check the HW availability
of this CT context.
In order to release the DR actions and reuse the context of a CT,
the reference count should also be handled in the flow rule
destroying.
The CT index will be recorded in the rte_flow by reusing the ASO age
index to save memory, since only one ASO action is supported in one
flow rule currently. The action context type should also be saved
for CT. When destroying a flow rule, if the context type is CT and
the index is valid (non-zero), the release process should be
handled. By default, the handling will fall back to try to release
the ASO age if any.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
When trying to destroy an ASO connection tracking context, the DR
action created on this context should also be destroyed. Before
inserting the related software object into the management free list,
the reference count should be checked.
Right now, the context object will not be freed to the system and
will be reused directly from the free list.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
After the connection tracking context is created and being used by
the flows, the context will be updated by the HW automatically after
a packet passed the CT validation. E.g., the ACK, SEQ, window and
state of CT can be updated with both direction traffic.
In order to query the updated contents of this context, a WQE should
be posted to the SQ with a return buffer. The data will be filled
into the buffer. And the profile will be filled with specific value.
During the execution of query command, the context may be updated.
The result of the query command may not be the latest one.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
When freeing the IB shared context during stopping a device, the
ASO connection tracking management structure should also be cleaned
up.
All the DR actions created should be destroyed. The structures need
to be freed and ASO CT QP should be released. In the meanwhile, the
allocated and registered memory region for query should also be
deregistered and then freed.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Allocating a CT from the management pools and creating the DR actions
for both directions by default.
If there is no available connection tracking action, a new pool will
be created with a fixed size bulk allocation. Right now, all the
resources are controlled by the linked list.
The ASO connection tracking context associated with these actions
need to be updated via WQE before using for steering.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
After the connection tracking object bulk is allocated, all the
objects' contents are filled with zero by default. Every
new-allocated object must be modified via WQE operation before it is
used.
In order to reduce the latency for the flow creation, an asynchronous
way is used instead of busy waiting for the CQE to be generated.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The definitions of ASO connection tracking objects management
structures are added.
Considering performance, the bulk allocation of ASO CT objects
should be used. The maximal value per bulk and the granularity could
be fetched from HCA capabilities 2. Right now, a fixed number of 64
is used for each bulk for a better management purpose.
The ASO QP for CT is initialized, the SQ will be used for both
modify and query command.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Based on the capacity, 3 registers could be used. Due to the register
allocation, only the one REG_C_3 for meter color could be reused
right now.
Then in the same flow, no more than one ASO action can be supported.
Signed-off-by: Bing Zhao <bingz@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
From RFC the size of the TCP flags is 9, while the defined
current size is 6.
Fixes: 641dbe4fb0 ("net/mlx5: support modify field flow action")
Cc: stable@dpdk.org
Signed-off-by: Wisam Jaddo <wisamm@nvidia.com>
Reviewed-by: Alexander Kozyrev <akozyrev@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Support the PMD power management API in MLX5 driver.
The monitor policy of this API puts a CPU core to sleep until
a data in some monitored memory address is changed by the NIC.
Implement the get_monitor_addr function to return an address
of a CQE owner bit to monitor the arrival of a new packet.
Signed-off-by: Alexander Kozyrev <akozyrev@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
Due to kernel issue in direct MKEY creation using the DevX API for
physical memory, this patch replaces the ASO MR creation to use Verbs
API.
Fixes: f935ed4b64 ("net/mlx5: support flow hit action for aging")
Cc: stable@dpdk.org
Signed-off-by: Michael Baum <michaelba@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Driver should prevent any DMA with the device when it
detects an error. When firmware is in fatal state,
stop tx/rx by assigning them to dummy functions.
Fixes: be14720def ("net/bnxt: support FW reset")
Fixes: 9d0cbaecc9 ("net/bnxt: support periodic FW health monitoring")
Cc: stable@dpdk.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Andy Gospodarek <gospo@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Fix ring count calculation for Thor. VNIC count does not have a
direct bearing on the number of rings that can be used.
Fixes: fe8dd26f86 ("net/bnxt: cap max Rx rings for Thor")
Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Fix comparison between uint16_t and uint32_t types.
Fixes: 6dc83230b4 ("net/bnxt: support port representor data path")
Cc: stable@dpdk.org
Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Return value where return value of rte_pci_read_config was not checked.
Fix it.
Coverity issue: 349919
Fixes: 9d0cbaecc9 ("net/bnxt: support periodic FW health monitoring")
Cc: stable@dpdk.org
Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
dev_info.max_mac_addrs is of type uint32_t. But the counter i is
of type uint16_t. This mismatch may cause the loop condition may
always be true. Change the loop counter variable to uint32_t.
Fixes: b02f1573cd ("net/bnxt: restore MAC filters during reset recovery")
Cc: stable@dpdk.org
Signed-off-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
The check BNXT_SINGLE_PF(bp) returns false for a VF. So there is no
extra check needed for VF along with BNXT_SINGLE_PF(bp).
Also make error messages more explicit.
Fixes: ff947c6ce1 ("net/bnxt: add check for multi host PF per port")
Fixes: f86febfb46 ("net/bnxt: support VF")
Fixes: 3e12fdb78e ("net/bnxt: support VLAN pvid")
Cc: stable@dpdk.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Ensure that the current count of in-use VNICs is decremented
when a VNIC is freed. Don't attempt VNIC allocation when the
maximum supported number of VNICs is currently allocated.
Fixes: 49d0709b25 ("net/bnxt: delete and flush L2 filters cleanly")
Fixes: d24610f7bf ("net/bnxt: allow flow creation when RSS is enabled")
Cc: stable@dpdk.org
Signed-off-by: Lance Richardson <lance.richardson@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Reported-by: Stephen Hemminger <sthemmin@microsoft.com>
Fix to clear the Rx FIFO while reading the timestamp.
If the Rx FIFO has pending bit set, keep reading to clear it
and return the last valid timestamp instead of unconditionally
returning an error.
Fixes: b11cceb83a ("net/bnxt: support timesync")
Cc: stable@dpdk.org
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Eliminate separate codepath/handling for single queue
as the multiqueue code path takes care of it as well.
The only difference being the end_grp_id being 1
now instead of 0 for single queue, but that does not matter
for single queue and does not alter any functionality.
Fixes: 6133f20797 ("net/bnxt: add Rx queue create/destroy")
Cc: stable@dpdk.org
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
The new VIRTCHNL_VF_OFFLOAD_VLAN_V2 capability added support that allows
the PF to set the location of the RX VLAN tag for stripping offloads.
So the VF needs to extract the VLAN tag according to the location flags.
This patch is the fix for AVX512 path, as AVX2 is already fixed.
Fixes: 9c9aa00403 ("net/iavf: add offload path for Rx AVX512 flex descriptor")
Signed-off-by: Leyi Rong <leyi.rong@intel.com>
Tested-by: Qin Sun <qinx.sun@intel.com>
New FDIR parsing are added to handle the fragmented IPv4/IPv6 packet.
Signed-off-by: Jeff Guo <jia.guo@intel.com>
Signed-off-by: Ting Xu <ting.xu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
New pattern and RSS hash flow parsing are added to handle fragmented
IPv4/IPv6 packet.
Signed-off-by: Jeff Guo <jia.guo@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Add support for IP fragment RSS hash and FDIR function. Separate IP
fragment and IP other packet types.
The patch also update the release date in README.
Signed-off-by: Ting Xu <ting.xu@intel.com>
Signed-off-by: Jeff Guo <jia.guo@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
External topology devices (e.g. PHYs) connected to 100G or to SoC that
includes 100G IP might have a firmware engine within the device and
the firmware is usually loaded from NVM connected to the topology
device.
The topology device NVM images can be updated using SW tools but
such solution poses a security risk if there is no validation of
the integrity of an image before programming it to the device NVM.
In order to prevent security risk, the topology device NVM image might
be included as part of 100G NVM image. When the topology device
NVM image is present in the 100G NVM image, it is authenticated
and might be loaded to the topology device at startup or on command
of SW using dedicated AQ.
This patch provides support for this functionality.
Signed-off-by: Stefan Wegrzyn <stefan.wegrzyn@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
The base code support to build configuration TLVs
in DSCP mode has not been implemented before, so
the functions to do so and the flow control to determine
if we are in VLAN or DSCP mode need to be added.
The current value for maximum number of DCB APPs
(ICE_DCBX_MAX_APPS) is not sufficient when supporting
DSCP mode. Each DSCP->TC mapping will come in as a
single APP value. So, there can be up to 64 APPs for
DSCP mapping.
Need to keep track of the current DSCP to TC mapping
so that TLVs can be built up to send to the FW. Add
an u8 array to hold this info.
A u64 is also needed to keep track of the DSCP values
that have had an APP submitted to map its value to a
TC. Since it would be unwise to allow an APP to be
overwritten by subsequent APPs, reject mappings for a
DSCP value that already has a user mapped value. This
will allow us to easily track which DSCP values have
been mapped, and when the last one has been deleted.
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Currently if the driver supports QinQ there is no message/information
if the DDP and/or FW don't support QinQ. Add functionality that prints
if the DDP and/or FW don't support QinQ if the driver attempts to
configured DVM. This will make it more obvious to users in the field
that they need to update their DDP and/or FW.
This required a small refactor so some of the existing code could be
shared and used by this new print functionality.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
Currently it's not clear that only the first PF downloads the package
and configures the VLAN mode. When this is happening all other PFs are
blocked on the global configuration lock. Once the package is
successfully downloaded and the global configuration lock has been
released then all PFs resume initialization. This includes some post
package download VLAN mode configuration. To make this more obvious add
the new function ice_post_pkg_dwnld_vlan_mode_cfg() so any/all post
download VLAN mode configuration code can be put in here.
This also makes it more clear that all PFs will call this new function.
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
Add the IPv6 fragment flags and the IPv4 fragment field shift.
Signed-off-by: Ting Xu <ting.xu@intel.com>
Signed-off-by: Jeff Guo <jia.guo@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
The vhost library currently configures Tx offloading (PKT_TX_*) on any
packet received from a guest virtio device which asks for some offloading.
This is problematic, as Tx offloading is something that the application
must ask for: the application needs to configure devices
to support every used offloads (ip, tcp checksumming, tso..), and the
various l2/l3/l4 lengths must be set following any processing that
happened in the application itself.
On the other hand, the received packets are not marked wrt current
packet l3/l4 checksumming info.
Copy virtio rx processing to fix those offload flags with some
differences:
- accept VIRTIO_NET_HDR_GSO_ECN and VIRTIO_NET_HDR_GSO_UDP,
- ignore anything but the VIRTIO_NET_HDR_F_NEEDS_CSUM flag (to comply with
the virtio spec),
Some applications might rely on the current behavior, so it is left
untouched by default.
A new RTE_VHOST_USER_NET_COMPLIANT_OL_FLAGS flag is added to enable the
new behavior.
The vhost example has been updated for the new behavior: TSO is applied to
any packet marked LRO.
Fixes: 859b480d5a ("vhost: add guest offload setting")
Cc: stable@dpdk.org
Signed-off-by: David Marchand <david.marchand@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Purely cosmetic but it is rather odd to have an "offload" helper that
checks if it actually must do something.
We already have the same checks in most callers, so move this branch
in them.
Signed-off-by: David Marchand <david.marchand@redhat.com>
Reviewed-by: Flavio Leitner <fbl@sysclose.org>
Reviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Tx offload flags are of the application responsibility.
Leave the mbuf alone and use a local storage for implicit tcp checksum
offloading in case of TSO.
Signed-off-by: David Marchand <david.marchand@redhat.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Flavio Leitner <fbl@sysclose.org>
When Rx queue worked in vectorized mode and rxd <= 512, under traffic of
high PPS rate, testpmd often start and receive packets of rxd without
further growth.
Testpmd started with rxq flush which tried to rx MAX_PKT_BURST(512)
packets and drop. When Rx burst size >= Rx queue size, all descriptors
in used queue consumed without rearm, device can't receive more packets.
The next Rx burst returned at once since no used descriptors found,
rearm logic was skipped, rx vq kept in starving state.
To avoid rx vq starving, this patch always check the available queue,
rearm if needed even no used descriptor reported by device.
Fixes: fc3d66212f ("virtio: add vector Rx")
Fixes: 2d7c37194e ("net/virtio: add NEON based Rx handler")
Fixes: 52b5a707e6 ("net/virtio: add Altivec Rx")
Cc: stable@dpdk.org
Signed-off-by: Xueming Li <xuemingl@nvidia.com>
Reviewed-by: David Christensen <drc@linux.vnet.ibm.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Some logs format u64 variables, mostly using hexadecimal which was not
readable.
This patch formats most u64 variables in decimal, and add '0x' prefix
to the ones that are not adjusted.
Fixes: c37ca66f2b ("net/hns3: support RSS")
Fixes: 2790c64647 ("net/hns3: support device reset")
Fixes: 8839c5e202 ("net/hns3: support device stats")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
VMDq is not supported yet, so remove the unused code.
Fixes: d51867db65 ("net/hns3: add initialization")
Fixes: 1265b5372d ("net/hns3: add some definitions for data structure and macro")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
According to the HW manual, the read operation is unnecessary when
enabling TM QCN error event, so remove it.
Fixes: f53a793bb7 ("net/hns3: add more hardware error types")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Currently, driver uses the macro HNS3_DEFAULT_RX_BURST whose value is
32 to limit the vector Rx burst size, as a result, the burst size
can't exceed 32.
This patch fixes this problem by support big burst size.
Also adjust HNS3_DEFAULT_RX_BURST to 64 as it performs better than 32.
Fixes: a3d4f4d291 ("net/hns3: support NEON Rx")
Fixes: 952ebacce4 ("net/hns3: support SVE Rx")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
The rte flow interface does not support the API of the capability
set. Therefore, fdir configuration logs are added to facilitate
debugging.
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
This patch improves data cache usage by:
1. Rearrange the rxq frequency accessed fields in the IO path to the
first 128B.
2. Rearrange the txq frequency accessed fields in the IO path to the
first 64B.
3. Make sure ptype table align cacheline size which is 128B instead of
min cacheline size which is 64B because the L1/L2 is 64B and L3 is
128B on Kunpeng ARM platform.
The performance gains are 1.5% in 64B packet macfwd scenarios.
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
This patch uses RTE_DIM() instead of ARRAY_SIZE().
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
This patch modifies hns3_write_reg_opt() API implementation because
the rte_write32() already uses rte_io_wmb().
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
This patch deletes some unused capabilities, include:
1. Delete some unused firmware capabilities definition, which are:
UDP_GSO, ATR, INT_QL, SIMPLE_BD, TX_PUSH, FEC and PAUSE.
2. Delete some unused driver capabilities definition, which are:
UDP_GSO, TX_PUSH.
3. Also redefine HNS3_DEV_SUPPORT_* as enum type, and change some of
the values. Note: the HNS3_DEV_SUPPORT_* values is used only inside
the driver, so it's safe to change the values.
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
MLX5 PMD supports the following integrity filters for outer and
inner network headers:
- l3_ok
- l4_ok
- ipv4_csum_ok
- l4_csum_ok
`level` values 0 and 1 reference outer headers.
`level` > 1 reference inner headers.
Flow rule items supplied by application must explicitly specify
network headers referred by integrity item. For example:
flow create 0 ingress
pattern
integrity level is 0 value mask l3_ok value spec l3_ok /
eth / ipv6 / end …
or
flow create 0 ingress
pattern
integrity level is 0 value mask l4_ok value spec 0 /
eth / ipv4 proto is udp / end …
Signed-off-by: Gregory Etelson <getelson@nvidia.com>
Acked-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
When jumbo frame is enabled, Rx function will choose 'Scalar Scattered'
function which has no PTP handling.
This patch fixes it by adding PTP handling in 'Scalar Scattered'
function.
Fixes: 38b539d96e ("net/hns3: support IEEE 1588 PTP")
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
If driver fails to enable MAC, it does not need to rollback the MAC
configuration. This patch fixes it.
Fixes: bdaf190f82 ("net/hns3: support link speed autoneg for PF")
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Remove "__rte_unused" instances that are wrongly marked.
Fixes: 6dc83230b4 ("net/bnxt: support port representor data path")
Fixes: 1bf01f5135 ("net/bnxt: prevent device access when device is in reset")
Cc: stable@dpdk.org
Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Decouple user mark from user flag. Usage of mark does not require to
use flag as well. Flag is not actually supported yet.
Fixes: 1aacc3d388 ("net/sfc: support user mark and flag Rx for EF100")
Cc: stable@dpdk.org
Signed-off-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
Reviewed-by: Andy Moreton <amoreton@xilinx.com>
Reviewed-by: Ivan Malov <ivan.malov@oktetlabs.ru>
When starting VF, VF will issue reset command to PF, wait a fixed
amount of time, and assume VF reset is done on PF side. However,
compared with kernel PF, DPDK PF needs more time to setup. If we
run DPDK PF to support DPDK VF, the original delay will not be
enough.
When we first start VF after PF is launched, the execution
time of the statement info.msg_buf = rte_zmalloc("msg_buffer",
info.buf_len, 0); in the function i40e_dev_handle_aq_msg is more
than 200ms. It may cause VF start error.
Since iavf can hardly trigger this issue and i40evf will be replaced
by iavf in future DPDK versions, this patch provide a workaround.
We extend VF reset waiting time from 200ms to 500ms so that
VF can start normally when using DPDK PF and DPDK VF in most cases.
Signed-off-by: Wenjun Wu <wenjun1.wu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
When start port, all MAC addresses will be set. We should set the MAC
type of default MAC address as VIRTCHNL_ETHER_ADDR_PRIMARY.
Fixes: 3f604ddf33 ("net/i40e: fix lack of MAC type when set MAC address")
Cc: stable@dpdk.org
Signed-off-by: Robin Zhang <robinx.zhang@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
When start port, all MAC addresses will be set. We should set the MAC
type of default MAC address as VIRTCHNL_ETHER_ADDR_PRIMARY.
Fixes: b335e72034 ("net/iavf: fix lack of MAC type when set MAC address")
Cc: stable@dpdk.org
Signed-off-by: Robin Zhang <robinx.zhang@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
A performance drop is caused by that the RX scalar path
is selected when AVX512 is disabled and some HW offload
is enabled.
Actually, the HW offload is supported by AVX2 and SSE.
In this scenario AVX2 path should be chosen.
This patch removes the offload related check for SSE and AVX2
as SSE and AVX2 do support the offload features.
No implementation change about the data path.
Fixes: eff56a7b9f ("net/iavf: add offload path for Rx AVX512")
Signed-off-by: Wenzhuo Lu <wenzhuo.lu@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
The driver support 2 mechanisms in order to support AGE action:
1. Aging by counter - HW counter will be configured to the flow traffic,
the driver polls the counter values efficiently to detect flow timeout.
2. Aging by ASO flow hit bit - HW ASO flow-hit bit is allocated for the
flow, the driver polls the bit efficiently to detect flow timeout.
ASO bit is only single bit resource while counter is 16 bytes, hence, it
is better to use ASO instead of counter for aging.
When a non-shared COUNT action is also configured to the flow, the
driver can use the same counter also for AGE action and no need to
create more ASO action for it.
The current code always uses ASO when it is supported in the device,
change it to reuse the non-shared counter if it exists in the flow.
Signed-off-by: Michael Baum <michaelba@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
A FLOW_AGE event should be invoked when a new aged-out flow is detected
by the PMD after the last user get-aged query calling.
The PMD manages 2 flags for this information and check them in order to
decide if an event should be invoked:
MLX5_AGE_EVENT_NEW - a new aged-out flow was detected. after the last
check.
MLX5_AGE_TRIGGER - get-aged query was called after the last aged-out
flow.
The 2 flags were unset after the event invoking.
When the user calls get-aged query from the event callback, the TRIGGER
flag was set inside the user callback and unset directly after the
callback what may stop the event invoking forever.
Unset the TRIGGER flag before the event invoking in order to allow set
it by the user callback.
Fixes: f935ed4b64 ("net/mlx5: support flow hit action for aging")
Cc: stable@dpdk.org
Reported-by: David Bouyeure <david.bouyeure@fraudbuster.mobi>
Signed-off-by: Michael Baum <michaelba@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Existing API supports counter action to count traffic of a single flow.
The user can share the count action among different flows using the
shared flag and the same counter ID in the count action configuration.
Recent patch [1] introduced the indirect action API.
Using this API, an action can be created as indirect, unattached to any
flow rule.
Multiple flows can then be created using the same indirect action.
The new API also supports query operation of an indirect action.
The new API is more efficient because the driver gets it's own handler
for the count action instead of managing a mapping between the user ID
to the driver handle.
Support create, query and destroy indirect action operations for flow
count action.
Application will use the indirect action query operation to query this
count action.
In the meantime the old sharing mechanism (with the sharing flag)
continues to be supported, and the user can choose the way he wants to
share the counter.
The new indirect action API is only supported in DevX, so sharing
counter action in Verbs can only be done through the old mechanism.
[1] https://mails.dpdk.org/archives/dev/2020-July/174110.html
Signed-off-by: Michael Baum <michaelba@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Tx prepare should be called only when necessary to reduce the impact on
performance.
For partial TX offload, users need to call rte_eth_tx_prepare() to
invoke the tx_prepare callback of PMDs. In this callback, the PMDs
adjust the packet based on the offloading used by the user. (e.g. For
some PMDs, pseudo-headers need to be calculated when the TX cksum is
offloaded.)
However, for the users, they cannot grasp all the hardware and PMDs
characteristics. As a result, users cannot decide when they need to
actually call tx_prepare. Therefore, we should assume that the user
calls rte_eth_tx_prepare() when using any Tx offloading to ensure that
related functions work properly. Whether packets need to be adjusted
should be determined by PMDs. They can make judgments in the
dev_configure or queue_setup phase. When the related function is not
used, the pointer of tx_prepare should be set to NULL to reduce the
performance loss caused by invoking rte_eth_tx_repare().
In this patch, if tx_prepare is not required for the offloading used by
the users, the tx_prepare pointer will be set to NULL.
Fixes: bba6366983 ("net/hns3: support Rx/Tx and related operations")
Cc: stable@dpdk.org
Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
The hns3_is_csq() and cmq_ring_to_dev() macro were defined in previous
version but never used.
Fixes: 737f30e1c3 ("net/hns3: support command interface with firmware")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Currently, driver uses gettimeofday() API to get the time, and
then calculate the time delta, the delta will be used mainly in
judging timeout process.
But the time which gets from gettimeofday() API isn't monotonically
increasing. The process may fail if the system time is changed.
We use the following scheme to fix it:
1. Add hns3_clock_gettime() API which will get the monotonically
increasing time.
2. Add hns3_clock_calctime_ms() API which will get the milliseconds of
the monotonically increasing time.
3. Add hns3_clock_calctime_ms() API which will calc the milliseconds of
a given time.
Fixes: 2790c64647 ("net/hns3: support device reset")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
If the reset process cost too much time, driver will log one error
message which formats the time delta, but the formatting is using
hexadecimal which was not readable.
This patch fixes it by formatting in decimal format.
Fixes: 2790c64647 ("net/hns3: support device reset")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
This patch supports get preferred burst size and queues when call
rte_eth_dev_info_get() API with VF.
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
All rights reserved by Beijing Wangxun Technology Co., Ltd.
Part of the code references Intel.
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
Support VXLAN-GPE in UDP tunnel port add and delete.
Fix to parsing packet type to pass hardware checksum.
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
When requested MTU is bigger than mbuf size and scattered Rx is not
enabled, setting MTU fails for VF.
But scattered Rx can be enabled in next port start if required, so
enabling setting MTU bigger than mbuf size if device is stopped
independent from scattered Rx configuration.
Fixes: a2beaa4a76 ("net/txgbe: support VF MTU update")
Cc: stable@dpdk.org
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Initialize meter policy list before use to avoid segment fault
Fixes: 0d73ddf25f ("net/softnic: add meter profile")
Cc: stable@dpdk.org
Signed-off-by: Dapeng Yu <dapengx.yu@intel.com>
Acked-by: Cristian Dumitrescu <cristian.dumitrescu@intel.com>
Return value of function 'eth_kni_dev_stop' passed to 'ret' is
rewritten later, and this is unreasonable.
This patch fixes it.
Fixes: 62024eb827 ("ethdev: change stop operation callback to return int")
Cc: stable@dpdk.org
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
After restoring the remote states, the return value of ioctl() is not
checked. Therefore, users cannot know whether the remote state is
restored successfully.
This patch add log for restoring failure.
Fixes: 4810d3af83 ("net/tap: restore state of remote device when closing")
Cc: stable@dpdk.org
Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Reviewed-by: Ferruh Yigit <ferruh.yigit@intel.com>
In function cons_parse_ntuple_filter, item->spec and item->mask
should be confirmed not null before use memcmp on it, current
judgement (item->spec || item->mask) just can confirm item->spec
or item->mask is not null, and cause null pointer be used in
memcmp.
This patch fix this problem.
Fixes: b7eeecb175 ("net/txgbe: parse n-tuple filter")
Cc: stable@dpdk.org
Signed-off-by: Hongbo Zheng <zhenghongbo3@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Acked-by: Jiawen Wu <jiawenwu@trustnetic.com>
When the port is link down state, it is meaningless to display the
port link speed. It should be an undefined state.
Fixes: 59fad0f321 ("net/hns3: support link update operation")
Cc: stable@dpdk.org
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
When port is stopped, link down should be reported to user. For HNS3
PF driver, link status comes from link status of hardware. If the port
supports NCSI feature, hardware MAC will not be disabled. At this case,
even if the port is stopped, the link status is still Up. So driver
should set link down when the port is stopped.
Fixes: 59fad0f321 ("net/hns3: support link update operation")
Cc: stable@dpdk.org
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Support of the checksum offloading by checking
the relevant FW capability (csum_cap) for NIC support.
RX supported offloads:
DEV_RX_OFFLOAD_IPV4_CKSUM
DEV_RX_OFFLOAD_UDP_CKSUM
DEV_RX_OFFLOAD_TCP_CKSUM
TX supported offloads:
DEV_TX_OFFLOAD_IPV4_CKSUM
DEV_TX_OFFLOAD_UDP_CKSUM
DEV_TX_OFFLOAD_TCP_CKSUM
Signed-off-by: Tal Shnaiderman <talshn@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Tested-by: Odi Assli <odia@nvidia.com>
mlx5 offloads which are unsupported on Windows
are currently disabled by checks with IBV/DV flags
which are irrelevant to Windows.
The checks are removed until they are fully available.
Fixes: 93f4ece91a ("net/mlx5: spawn ethdev ports on Windows")
Cc: stable@dpdk.org
Signed-off-by: Tal Shnaiderman <talshn@nvidia.com>
Acked-by: Matan Azrad <matan@nvidia.com>
Tested-by: Odi Assli <odia@nvidia.com>
If the device was configured as legacy bond one (without
involving E-Switch), the mlx5 PMD erroneously tried to deduce
the vport index raising the fatal error and preventing
device from being used.
The patch checks whether there is E-Switch present and we
should use vport index indeed.
Fixes: 2eb4d0107a ("net/mlx5: refactor PCI probing on Linux")
Fixes: d5c06b1b10 ("net/mlx5: query vport index match mode and parameters")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
The mlx4 PMD tracks the buffers (mbufs) for the packets being
transmitted in the dedicated array named as "elts". The tx_burst
routine frees the mbufs from this array once it needs to rearm
the hardware descriptor and store the new mbuf, so it looks
like as replacement mbuf pointer in the elts array.
On the device stop mlx4 PMD freed only the part of elts according
tail and head pointers, leaking the rest of buffers, remained in
the elts array.
Fixes: a2ce2121c0 ("net/mlx4: separate Tx configuration functions")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@nvidia.com>
