Relax memory requirement for event timers when internal mempool used is
octeontx2 mempool.
Add debug log to print the memory used.
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
If an application issues rte_event_enqueue_new_burst() or
rte_event_enqueue_forward_burst() call with a burst of events longer
than the configured max enqueue burst size, DSW allocates credits not
only for events actually enqueued, but for the complete burst. If this
process is repeated, enough credits will have leaked to cause the
event device to backpressure (i.e. disallow) any new enqueue
operations.
In addition, the port-level enqueue xstats will log the wrong number
of events enqueued for oversized enqueues.
This patch makes DSW gracefully handle oversized enqueue bursts.
Fixes: 1c8e3caa3bfb ("event/dsw: add event scheduling and device start/stop")
Cc: stable@dpdk.org
Signed-off-by: Mattias Rönnblom <mattias.ronnblom@ericsson.com>
Fix DSW's rte_event_enqueue_burst(), so that a call with a zero-sized
event array immediately flushes the port's output buffers. Prior to
this patch, the flush operation would be deferred to the next enqueue
or dequeue call, which is inconsistent with DSW documentation.
Fixes: 1c8e3caa3bfb ("event/dsw: add event scheduling and device start/stop")
Cc: stable@dpdk.org
Signed-off-by: Mattias Rönnblom <mattias.ronnblom@ericsson.com>
A new vDPA class was recently introduced.
IFC driver implements the vDPA operations,
hence it should be moved to the vDPA class.
Move it.
Signed-off-by: Matan Azrad <matan@mellanox.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
The vDPA (vhost data path acceleration) drivers provide support for
the vDPA operations introduced by the rte_vhost library.
Any driver which provides the vDPA operations should be moved\added to
the vdpa class under drivers/vdpa/.
Create the general files for vDPA class in drivers and in documentation.
The management tree for vDPA drivers is
git://dpdk.org/next/dpdk-next-virtio.
Signed-off-by: Matan Azrad <matan@mellanox.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
The soname for each stable ABI version should be just the ABI version major
number without the minor number. Unfortunately both major and minor were
used causing version 20.1 to be incompatible with 20.0.
This patch fixes the issue by switching from 2-part to 3-part ABI version
numbers so that we can keep 20.0 as soname and using the final digits to
identify the 20.x releases which are ABI compatible. This requires changes
to both make and meson builds to handle the three-digit version and shrink
it to 2-digit for soname.
The final fix needed in this patch is to adjust the library version number
for the ethtool example library, which needs to be upped to 2-digits, as
external libraries using the DPDK build system also use the logic in this
file.
Fixes: cba806e07d6f ("build: change ABI versioning to global")
Signed-off-by: Thomas Monjalon <thomas@monjalon.net>
Signed-off-by: Bruce Richardson <bruce.richardson@intel.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Tested-by: Ray Kinsella <mdr@ashroe.eu>
Tested-by: Ferruh Yigit <ferruh.yigit@intel.com>
Tested-by: Kevin Laatz <kevin.laatz@intel.com>
Tested-by: David Marchand <david.marchand@redhat.com>
This rewrites the MKSTR macro appending an empty string to its arguments
to resolve build failures similar to:
drivers/net/mlx4/mlx4.c:461:14: fatal error: format string is not a
string literal [-Wformat-nonliteral]
MKSTR(path, "%s/device/uevent", device->ibdev_path);
drivers/net/mlx4/mlx4_utils.h:82:30: note: expanded from macro 'MKSTR'
char name[snprintf(NULL, 0, __VA_ARGS__) + 1]; \
drivers/net/mlx5/mlx5_stats.c:144:15: fatal error: format string is not a
string literal [-Wformat-nonliteral]
MKSTR(path, "%s/ports/%d/hw_counters/%s",
drivers/net/mlx5/mlx5_utils.h:149:30: note: expanded from macro 'MKSTR'
char name[snprintf(NULL, 0, __VA_ARGS__) + 1]; \
The errors reproduce with clang version 9.0.0, and the release notes
don't mention what could have caused them.
Fixes: 7fae69eeff13 ("mlx4: new poll mode driver")
Fixes: 771fa900b73a ("mlx5: introduce new driver for Mellanox ConnectX-4 adapters")
Cc: stable@dpdk.org
Signed-off-by: Ali Alnubani <alialnu@mellanox.com>
Signed-off-by: Matan Azrad <matan@mellanox.com>
When net/octeontx is connected to event/octeontx as an event Rx adapter,
PKI aka 'net/octeontx' can forward packets directly to SSO aka
'event/octeontx'.
When pumping traffic to PKI if flow control is disabled internal FIFOs
might be overrun causing partial l2 packets to be enqueued.
SSO receives <31:0> TAG tag calculated by PKI, in normal cases <31:28>
is always 0 which signifies RTE_EVENT_TYPE_ETHDEV. But in case of
partial received packets PKI sets the <31:0> TAG as 0xFFFFFFFF which
is an invalid event type.
Add a check to see if TAG is 0xFFFFFFFF and free the partial receive
packet.
Fixes: d0d654986018 ("net/octeontx: support event Rx adapter")
Cc: stable@dpdk.org
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
The legacy multi-packet write is the feature allowing to
put multiple packets into one transmitting descriptor,
this feature is supported by only NIC ConnectX-4LX.
The number of packets should be limited to provide
optimal size descriptor and better performance.
Fixes: 18a1c20044c0 ("net/mlx5: implement Tx burst template")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
The legacy multi-packet write is the feature allowing to
put multiple packets into one transmitting descriptor,
this feature is supported by only NIC ConnectX-4LX.
The number of packets should be limited to provide
optimal size descriptor and better performance.
Fixes: 18a1c20044c0 ("net/mlx5: implement Tx burst template")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
In extensive metadata mode the MARK copy table is engaged,
if the application creates the flow with zero MARK ID action:
flow create 1 ingress pattern eth / ... / end actions mark id 0 / .. end
And then destroys that, the traffic to the port stops. This happens
due to default flow for the copy table has the zero ID and is removed
with the application rule. The patch extends internal ID variable
to 64 bits and provide the UINT64_MAX ID for the copy table default
rule.
Fixes: dd3c774f6ffb ("net/mlx5: add metadata register copy table")
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Ori Kam <orika@mellanox.com>
When set the GRE item, GRE key should follow after GRE header, or the
header gre_item pointer used by the key will be invalid.
Currently in the mlx5_flow_validate_item_gre_key() function, the header
gre_item pointer is access before checking if the key is after the
header or not. Once the key item is before the header, invalid gre_item
pointer access happens.
Move the gre_item pointer access after the GRE header check to avoid the
crash issue.
Fixes: a7a0365565a4 ("net/mlx5: match GRE key and present bits")
Cc: stable@dpdk.org
Signed-off-by: Suanming Mou <suanmingm@mellanox.com>
Acked-by: Ori Kam <orika@mellanox.com>
GENEVE is available in tunnel offloads. Add it as the default support
option.
Signed-off-by: Suanming Mou <suanmingm@mellanox.com>
Acked-by: Ori Kam <orika@mellanox.com>
The header mlx5_glue.h includes rte_byteorder.h, which isn't
pulled by global_inc. This patch adds the missing include directory
and fixes the build error:
In file included from drivers/net/mlx5/mlx5_glue.c:31:
drivers/net/mlx5/mlx5_glue.h:12:10: fatal error:
rte_byteorder.h: No such file or directory
The build failure can be triggered with:
meson -Dibverbs_link=dlopen build && ninja -C build
Fixes: 400d985eb586 ("net/mlx5: add VLAN push/pop DR commands to glue")
Signed-off-by: Ali Alnubani <alialnu@mellanox.com>
Acked-by: Thomas Monjalon <thomas@monjalon.net>
ConnectX-4LX supports multiple packets within the single Tx
descriptor. This feature is named as "Legacy Multi-Packet Write"
and imposes a lot of limitations:
- no ACLs, it means no NIC Tx Flows are supported and Tx metadata
become meaningless
- the required minimal inline data must be zero
- no SR-IOV, it means no support in E-Switch configurations,
- no priority and dscp forcing
- no VLAN insertion
- no TSO
- all packets within MPW session must have the same size
This legacy MPW feature is mainly intended for test purposes.
To explicitly engage the feature on ConnectX-4LX the devargs
should be specified:
- txq_mpw_en=1
This feature was dropped in 19.08, this patch reverts it back.
Fixes: 18a1c20044c0 ("net/mlx5: implement Tx burst template")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
'close()' calls 'stop()' and 'stop()' cancels pthread without any check.
Calling 'stop()' & 'close()' sequentially tries to cancel pthread twice
which will cause a crash.
Adding a state check in 'stop()' before canceling the pthread to prevent
multiple stop.
Fixes: 696fbc7bb4fc ("net/kni: remove resources when port is closed")
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Tested-by: Yinan Wang <yinan.wang@intel.com>
It's possible that we fail to get the IOMMU group of ifcvf device, this
patch adds a check on the return value.
Coverity issue: 349894
Fixes: a3f8150eac6d ("net/ifcvf: add ifcvf vDPA driver")
Cc: stable@dpdk.org
Signed-off-by: Xiao Wang <xiao.w.wang@intel.com>
Reviewed-by: Xiaolong Ye <xiaolong.ye@intel.com>
Commit in fixes line sets the DV (Direct Verbs) flow engine as default.
Newer versions of DV flow engine use the DR (Direct Rules) features.
DR is supported from RDMA Core library version rdma-core-24.0.
This cause failure to start port when using older rdma-core version,
without DR support.
This patch selects DV flow engine if rdma-core version is v24.0 or
higher. Verbs flow engine is selected otherwise.
Fixes: cd4569d2bf3c ("net/mlx5: change default flow engine to DV")
Signed-off-by: Dekel Peled <dekelp@mellanox.com>
Acked-by: Ori Kam <orika@mellanox.com>
loopback mode is not supported for VFs so returning error if
VF is being configured with loopback mode.
Fixes: 920717e4d8ba ("net/octeontx2: add device start operation")
Cc: stable@dpdk.org
Signed-off-by: Sunil Kumar Kori <skori@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
The link status got from link status register was not correct,
because register has been reset when ports reset.
After port reset, set the link status down.
Fixes: c431ec66c54c ("net/igb: support setting link up or down")
Cc: stable@dpdk.org
Signed-off-by: Lunyuan Cui <lunyuanx.cui@intel.com>
Acked-by: Wenzhuo Lu <wenzhuo.lu@intel.com>
The link status for 82599eb got from link status register was not
correct. Check the enable/disable flag of tx laser, set the link
status down if tx laser disabled. Then, we can get correct status.
But after port reset, tx laser register will be reset enable.
Link status will always be up. So set tx laser disable when port resets.
When hw->mac.autotry_restart is true, whether tx laser is disable or
enable, it will be set enable in ixgbe_flap_tx_laser_multispeed_fiber().
hw->mac.autotry_restart can be set true in both port init and port start.
Because we don't need this treatment before port starts, set
hw->mac.autotry_restart false when port init.
Fixes: 0408f47ba4d6 ("net/ixgbe: fix busy polling while fiber link update")
Cc: stable@dpdk.org
Signed-off-by: Lunyuan Cui <lunyuanx.cui@intel.com>
Acked-by: Wenzhuo Lu <wenzhuo.lu@intel.com>
Meaningless to judge the link state according to the memset'ed
link variable, this patch fixes this logical issue.
In addition, this patch changes the variable from link_check to link_up
according to its real meaning.
Fixes: 80ba61115e77 ("net/e1000: use link status helper functions")
Cc: stable@dpdk.org
Signed-off-by: Lunyuan Cui <lunyuanx.cui@intel.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
Acked-by: Wenzhuo Lu <wenzhuo.lu@intel.com>
In .tx_pkt_burst ops implementation function of hns3 PMD driver,
there is one check whether there are enough BDs in the TX queue.
If not, driver will stop sending the packets.
Currently in the 'for' process loop, the next_to_use member of
TX queue is not updated in time after processing BDs of one packet,
which results in the invalid action of checking whether there are
enough BDs and failure in sending packets.
This patch fixes it by moving the assignment statment of the
next_to_use member of TX queue to the place after porcessing TX BDs
in the 'for' loop.
Fixes: bba636698316 ("net/hns3: support Rx/Tx and related operations")
Cc: stable@dpdk.org
Signed-off-by: Hongbo Zheng <zhenghongbo3@huawei.com>
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Enable packet type parsing by default, only disable when
`rte_eth_dev_set_ptypes()` is called with ptype_mask as 0.
This would enable applications that are dependent on packet type parsing
like l3fwd.
Also this patch preserves configuration set by `rte_eth_dev_set_ptypes`
across multiple calls of `rte_eth_dev_configure()`
Fixes: d2706e15e6fb ("net/octeontx2: support reduced set of packet types")
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
The metadata register used for META item/action within E-Switch
steering domain should be altered depending on dv_xmeta_en
devarg value.
Fixes: 3e8edd0ef848 ("net/mlx5: update metadata register ID query")
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
For tunnel mode, there may be two L3/L4 layer match pattern items, one
for inner and one for outer layers. The L3 TTL and L4 port header
modify actions should handle the outermost layer items.
Currently flow_dv_attr_init() doesn't distinguish between outer and
inner layers, since inner layer comes later than the outer layer, this
may cause inner layer values also set to the flow attributes and may
lead actions to use inner L3/L4 pattern items.
Adding a check in flow_dv_attr_init() to prevent inner layer to set the
flow attribute if the previous L3/L4 outer pattern exist.
Fixes: 4bb14c83df95 ("net/mlx5: support modify header using Direct Verbs")
Cc: stable@dpdk.org
Signed-off-by: Suanming Mou <suanmingm@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
When setting vlan, hns3 driver will add vlan entry to vlan linked list
each time, and this is unreasonable.
This patch adds a check whether the VLAN to be added already exists
in the linked list and prevents adding duplicated vlan.
Fixes: 411d23b9eafb ("net/hns3: support VLAN")
Cc: stable@dpdk.org
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Currently, port link status is "up" in VF driver after user calling the
rte_eth_dev_stop API. This is unreasonable.
Therefore, this patch adjusts the strategy of getting link status from
PF driver for VF. VF driver should stop getting link status from PF by
canceling the alarm that VF driver send mailbox message to PF driver,
when the rte_eth_dev_stop API is called. And VF driver should restore
the alarm when the rte_eth_dev_start API is called.
Fixes: a5475d61fa34 ("net/hns3: support VF")
Cc: stable@dpdk.org
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Remove the macro 'HNS3_REG_MSG_DATA_OFFSET' which is used to prevent
access violation for response array by limiting the response data length
to 4.
But the limit value is too short to get some longer information such as
6 byte MAC address.
This patch modifies the length of response data from mailbox to allow
the response data length to be 8. So that the VF driver could get more
data from PF driver by mailbox.
Fixes: 463e748964f5 ("net/hns3: support mailbox")
Cc: stable@dpdk.org
Signed-off-by: Chengchang Tang <tangchengchang@hisilicon.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Ethernet minimum packet length is 64 bytes. If upper application
sends packets with less than 60 bytes in length(no CRC), driver
adds padding processing to avoid failure.
Fixes: bba636698316 ("net/hns3: support Rx/Tx and related operations")
Cc: stable@dpdk.org
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
When the application using VF device exits abnormally, for example,
when it is killed by 'kill -9', kernel PF netdev driver also stores
the corresponding configuration table entries of VF device.
This patch fixes it by adding message of deleting VF configuration
table entry corresponds to the revision of kernel hns3 netdev
driver, the new message is added to notify the kernel PF netdev
driver to clean up the VF configuration initialization during VF
initialization.
This revision is compatible with the old version of kernel hns3
netdev driver. The old version of kernel pf netdev driver will
ignore this message.
Fixes: a5475d61fa34 ("net/hns3: support VF")
Cc: stable@dpdk.org
Signed-off-by: Hongbo Zheng <zhenghongbo3@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
This patch fixes the bug that hardware configuration called
tc_size doesn't restore to the initial value when starting
the app, configuring PFC and then restarting the app,
because of the tc_mode didn't initial when rss is disabled.
Fixes: c37ca66f2b27 ("net/hns3: support RSS")
Cc: stable@dpdk.org
Signed-off-by: Hao Chen <chenhao164@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Now that vnics are created only as part of the flow creation cmds
and not during init, we cannot rely on iterating only through
'nr_vnics'. We need to sweep all the vnics by using 'max_vnics'
otherwise a vnic with a stale 'rx_queue_cnt' might be left lingering
post a port stop/start operation. This could lead to a segfault.
This change is required because of the recent fix made by commit to
"fix flow creation with non-consecutive group ids".
Fixes: fcdd7210aa1f ("net/bnxt: fix flow creation with non-consecutive group ids")
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
The current raw_encap function selects if to decap or encap
based on the egress/ingress attribute. This concepts doesn't work
in case of FDB since all flows are considered ingress.
To solve this issue we moved to check the encap size.
if the encap size is larger then eth + ipv4 it means we are
trying to encap.
Fixes: 8ba9eee4ce32 ("net/mlx5: add raw data encap/decap to Direct Verbs")
Cc: stable@dpdk.org
Signed-off-by: Ori Kam <orika@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
Currently macsec offload will be enabled every time when device starts.
It will cause QoS sample application performance drop issue. This patch
adds check in dev_start ops to make sure macsec is only enabled when
required explicitly.
Fixes: 50556c88104c ("net/ixgbe: fix MACsec setting")
Cc: stable@dpdk.org
Signed-off-by: Shougang Wang <shougangx.wang@intel.com>
Reviewed-by: Xiaolong Ye <xiaolong.ye@intel.com>
When DR is not supported and DV is supported, tag action still can be
used by the metadata feature.
Wrongly, the tag hash list was not created what caused failure in
metadata action creation.
Create the tag hash list for each DV case.
Fixes: 860897d2895a ("net/mlx5: reorganize flow tables with hash list")
Signed-off-by: Matan Azrad <matan@mellanox.com>
The "addr" param has been used since the event callbacks are enabled
for external memory. So the "__rte_unused" should be dropped.
Besides, slightly refine the coding style by consistently assuming
tabs are 8 characters.
Fixes: f32c7c9de961 ("malloc: enable event callbacks for external memory")
Cc: stable@dpdk.org
Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Add FDIR support for normal GTP-U packet, which doesn't contain GTP_PSC
extension header.
Fixes: efc16c621415 ("net/ice: support flow director GTPU tunnel")
Signed-off-by: Yahui Cao <yahui.cao@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Pattern name of GTP-U flow with extension header should be
xxx_gtpu_eh_xxx. So it will not conflict with the pattern name of a
GTP-U flow that does not contains extension header.
Fixes: 3e4eab9c2192 ("net/ice: add pattern manifest")
Signed-off-by: Yahui Cao <yahui.cao@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Flow type "IPv4 + UDP" or "IPv4 + TCP" is conflict with "IPv4 + any"
flow type. If a rule for IPv4 + any is created, we should reject any
rule for IPv4 + UDP otherwise the first rule may be impacted, same
decision should be made on a reverse order. For IPv6 and IPv4 GTPU
inner case, we have the same limitation.
Fixes: 109e8e06249e ("net/ice: configure HW flow director rule")
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Tested-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
CCP can support varied digest sizes ranging from 1 to some max value
But the current code support only fixed max values.
This patch updates the minimum digest sizes to 1
Fixes: e0d88a394e ("crypto/ccp: support run-time CPU based auth")
Cc: stable@dpdk.org
Signed-off-by: Amaranath Somalapuram <asomalap@amd.com>
CCP driver was scheduling only one CCP in a single burst(enqueue).
Effective throughput was limited to 1 CCP performance.
Scheduling multiple ccp within one burst will increase the ccp performance.
this changes will divide the enqueue packets equally among the multiple CCP
Fixes: e0d88a394e ("crypto/ccp: support run-time CPU based auth")
Cc: stable@dpdk.org
Signed-off-by: Amaranath Somalapuram <asomalap@amd.com>
when ccp_auth_opt=1 is set and if authentication error occurred,
ccp driver crash. The enqueue referance count nb_ops miss match
with dequeue nb_ops on authentication error.
Fixes: e0d88a394e ("crypto/ccp: support run-time CPU based auth")
Cc: stable@dpdk.org
Signed-off-by: Amaranath Somalapuram <asomalap@amd.com>
CCP driver crash when running l2fwd
issue happen when changing MAX_PKT_BURST 32 to 256
increasing the CCP qp to 8 and CCP_MAX_BURST to 256
Fixes: ef4b04f87f ("crypto/ccp: support device init")
Cc: stable@dpdk.org
Signed-off-by: Amaranath Somalapuram <asomalap@amd.com>
