Add missing arm vector build support to meson.build.
Signed-off-by: Reshma Pattan <reshma.pattan@intel.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
Add stubs for ixgbe_xmit_fixed_burst_vec,
ixgbe_rx_queue_release_mbufs_vec and
ixgbe_txq_vec_setup
Signed-off-by: Reshma Pattan <reshma.pattan@intel.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
Remove weak symbols from ixgbe_rxtx.c file as
it is done in i40e driver in commit "02ad704708"
(net/i40e: eliminate weak symbols in data path)
Signed-off-by: Reshma Pattan <reshma.pattan@intel.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
CONFIG_RTE_IXGBE_INC_VECTOR is enabled by default, so remove
it and use architecture specific flags.
Signed-off-by: Reshma Pattan <reshma.pattan@intel.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
Remove the unused definitions, rewrite the IO data read/write helpers,
and put the common definitions related to RTE defines under the macro
__INTEL_NET_BASE_OSDEP__, so it works like OS(RTE) dependency.
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
Replaces the redefined TRUE and FALSE values with standard ones to
match the 'bool' type definition.
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
Replaces the redefined TRUE and FALSE values with standard ones to
match the 'bool' type definition.
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
Hardware limits that max buffer size per Tx descriptor should be
(16K-1)B. So when TSO enabled, the mbuf data size may exceed the
limit and cause malicious behavior to the NIC. This patch fixes
this issue by using more Tx descs for this kind of large buffer.
Fixes: 17c7d0f9d6 ("net/ice: support basic Rx/Tx")
Cc: stable@dpdk.org
Signed-off-by: Xiaoyun Li <xiaoyun.li@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Build error:
In function ‘ixgbe_recv_pkts_lro_bulk_alloc’:
../drivers/net/ixgbe/ixgbe_rxtx.c:2209:24:
error: ‘next_sc_entry’ may be used uninitialized in this function
[-Werror=maybe-uninitialized]
next_sc_entry->fbuf = first_seg;
^
http://mails.dpdk.org/archives/test-report/2020-January/113891.html
This is a compiler false positive and error not seen by newer compilers,
or clang but to fix the warning initializing the complained variable.
According git bisect, no idea how:
Fixes: ad43b7bce9 ("net/ixgbe: avoid multiple definitions of bool")
Reported-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
Tested-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
The doorbell register is mapped using mmap() and offset
must have off_t instead of unsigned int. Bug is not critical
due to only least significant bits of offset are currently
tested to determine mapping mode.
Fixes: 8409a28573 ("net/mlx5: control transmit doorbell register mapping")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
This patch fixes the situation where data path does not work properly
when vhost reconnects to virtio in server mode with packed ring.
Currently, virtio and vhost share memory of vring. For split ring, vhost
can read the status of descriptors directly from the available ring and
the used ring during reconnection. Therefore, the data path can
continue.
But for packed ring, when reconnecting to virtio, vhost cannot get the
status of descriptors via the descriptor ring. By resetting Tx
and Rx queues, the data path can restart from the beginning.
Fixes: 4c3f5822eb ("net/virtio: add packed virtqueue defines")
Cc: stable@dpdk.org
Signed-off-by: Xuan Ding <xuan.ding@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
The function vhost_kernel_tap_set_offload() could return errors,
the return value need to be checked. And there is no need to fail
when error is -ENOTSUP.
Fixes: 1db4d2330b ("net/virtio-user: check negotiated features before set")
Cc: stable@dpdk.org
Signed-off-by: Yunjian Wang <wangyunjian@huawei.com>
Reviewed-by: Tiwei Bie <tiwei.bie@intel.com>
This patch supports the feature that the driver passes extra data
(besides identifying the virtqueue) in its device notifications,
expanding the notifications to include the avail index and avail
wrap counter.
Signed-off-by: Cheng Jiang <cheng1.jiang@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Do not close the tap fds when disabling queue pairs, instead,
we just need to unbind the backend. Otherwise, tap port can be
destroyed unexpectedly.
Fixes: e3b434818b ("net/virtio-user: support kernel vhost")
Cc: stable@dpdk.org
Reported-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Tiwei Bie <tiwei.bie@intel.com>
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Add support to the igb vf for the API rte_eth_tx_done_cleanup to force
free consumed buffers on Tx ring.
Signed-off-by: Chenxu Di <chenxux.di@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Add support to the ixgbe driver for the API rte_eth_tx_done_cleanup
to force free consumed buffers on Tx ring.
Signed-off-by: Chenxu Di <chenxux.di@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
Add support to the ice driver for the API rte_eth_tx_done_cleanup
to force free consumed buffers on Tx ring.
Signed-off-by: Chenxu Di <chenxux.di@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
Add support to the i40e driver for the API rte_eth_tx_done_cleanup
to force free consumed buffers on Tx ring.
Signed-off-by: Chenxu Di <chenxux.di@intel.com>
Acked-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
octeontx2 PMD does not support both PTP and HIGIG2 together.
Added a check to enforce this and updated the Rx offload capabilities when
Higig2 mode enabled.
Fixes: 602009ee2d ("net/octeontx2: support HIGIG2")
Cc: stable@dpdk.org
Signed-off-by: Kiran Kumar K <kirankumark@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Multi-segs Tx is already supported by CXGBE PMD. So, add the missing
DEV_TX_OFFLOAD_MULTI_SEGS flag to the list of supported Tx offload
features.
Fixes: 436125e641 ("net/cxgbe: update to Rx/Tx offload API")
Cc: stable@dpdk.org
Reported-by: Chas Williams <3chas3@gmail.com>
Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
Move a macro from a widely included header file to a header file
used only by the one caller of the macro.
Signed-off-by: John Daley <johndale@cisco.com>
Reviewed-by: Hyong Youb Kim <hyonkim@cisco.com>
There were defines which originally allowed sharing of some code with
the enic kernel driver. The code has long since diverged and now the
abstraction just makes the code harder to read. Mostly mechanical
replacement of defines and reformatting.
Signed-off-by: John Daley <johndale@cisco.com>
Reviewed-by: Hyong Youb Kim <hyonkim@cisco.com>
Use the memzone namesize, Ethernet address length defines from the RTE
header files instead of locally defined versions.
Use the RTE byte swap functions instead of the x86 specific locally
defined versions.
Signed-off-by: John Daley <johndale@cisco.com>
Reviewed-by: Hyong Youb Kim <hyonkim@cisco.com>
The control messages do not need NUMA specific allocation.
Numa node is not set anyway in most kernels anyway.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
The secondary process would crash trying to get the offload
capability of the device. Since the device capabilities were
already set by the primary process this was unnecessary.
Fixes: 4e9c73e96e ("net/netvsc: add Hyper-V network device")
Cc: stable@dpdk.org
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Avoid using PCI subsystem device id for SoC revision
identification and just use PCI revision id to support C0 silicon.
This patch also reduces SQB threshold to 70% to have
sufficient buffer before we overflow SQ.
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
At BGX, DMAC filter table can be enabled to allow defined hosts.
Patch implements 'mac_addr_add' and 'mac_addr_remove' ops to
add/remove MAC filters which uses BGX DMAC filter table.
Signed-off-by: Sunil Kumar Kori <skori@marvell.com>
Acked-by: Harman Kalra <hkalra@marvell.com>
Use scalar loads instead of vector loads for fields
that don't need any vector operations.
Signed-off-by: Pavan Nikhilesh <pbhagavatula@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Use lightweight transmit handler which handles non-offloaded Tx data
path. We get CPU utilization improvement of ~8%.
Signed-off-by: Shahed Shaikh <shshaikh@marvell.com>
Use light weight receive packet handler for non-LRO and
non-scatter packets to improve CPU utilization in receive data path.
We achieved ~18% CPU cycles improvement using this handler.
Signed-off-by: Shahed Shaikh <shshaikh@marvell.com>
Fix the way in which the number of ports per engine of an adapter is
determined by reading port mode register. Set default value to 1.
Fixes: 3b307c55f2 ("net/qede/base: update FW to 8.40.25.0")
Cc: stable@dpdk.org
Signed-off-by: Rasesh Mody <rmody@marvell.com>
Current vector mode implementation floor-aligns pkt count
with NIX_DESCS_PER_LOOP and process that many packets.
Patch addresses the case where pkt count modulo NIX_DESCS_PER_LOOP
is non-zero, after the vector mode processing, scalar routine is
used to process if there are any leftover packets. Scalar routine
is also used when descriptor head is about to wrap and turn out to
be unaligned.
Signed-off-by: Vamsi Attunuru <vattunuru@marvell.com>
Signed-off-by: Nithin Dabilpuram <ndabilpuram@marvell.com>
Acked-by: Jerin Jacob <jerinj@marvell.com>
Currently, reset related operations can only be performed in the primary
process and are not allowed in the slave process in hns3 PMD driver.
In the internal function interface named hns3_cmd_send used for
communication between driver and firmware, if the wrong head value is
detected in the static subfunction hns3_cmd_csq_clean, driver will
trigger a function level reset to make the hardware work normally again.
This patch adds check condition to prevent triggering reset procedure in
the slave process to avoid failure.
Fixes: 2790c64647 ("net/hns3: support device reset")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Currently, the hns3 PMD driver issues command to the firmware and gets
link status information.
When the driver fails to call internal interface function named
hns3_cmd_send to query the status from firmware for some reason, the
link status queried by the driver should be down.
Fixes: 59fad0f321 ("net/hns3: support link update operation")
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>
Currently, when the API interface named rte_eth_dev_get_reg_info is
called by upper applications based on VF device, it returns error.
We can read registers directly to get ring and interrupt related
information in hns3 PF/VF PMD driver. But for some other internal table
entries and common configuration information, we can get them only
through the command interface between driver and firmware in PF driver,
and VF driver has not the related access permission.
This patch fixes it by preventing getting these information through the
command interface based on VF device in 'get_reg' ops implementation
function.
Fixes: 936eda25e8 ("net/hns3: support dump register")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
The format of the ring vector related mailbox commands between driver
and firmware is different from those of other mailbox commands in hns3
network engine.
This patch fixes the error mailbox command format about the vector of
the rings, the related command opcode as below:
HNS3_MBX_MAP_RING_TO_VECTOR
HNS3_MBX_UNMAP_RING_TO_VECTOR
HNS3_MBX_GET_RING_VECTOR_MAP
Fixes: 463e748964 ("net/hns3: support mailbox")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Currently there is a certain probability of segment error in concurrent
reset when the port is closing.
The calltrace info:
This GDB was configured as "aarch64-redhat-linux-gnu".
Reading symbols from /usr/app/testpmd...(no debugging symbols found)...
done.
[New LWP 98204]
[New LWP 98203]
[New LWP 98206]
[New LWP 98205]
[New LWP 98207]
[New LWP 98208]
Missing separate debuginfo for /root/lib/libnuma.so.1
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
Core was generated by `/usr/app/testpmd --log-level=6 --socket-mem 16'.
Program terminated with signal 11, Segmentation fault.
Missing separate debuginfos, use:
debuginfo-install glibc-2.17-260.el7.aarch64
(gdb) bt
in hns3vf_service_handler ()
1 0x00000000006988b8 in eal_alarm_callback ()
2 0x00000000006969b4 in eal_intr_thread_main ()
3 0x0000ffffb08d6c48 in start_thread () from /lib64/libpthread.so.0
4 0x0000ffffb0828600 in thread_start () from /lib64/libc.so.6
(gdb)
Reset process may turn on the cancelled link state timer whether the
current port status is on or off, in order to solve this problem, this
patch add judge the current network port state before starting the
timer, only the port in the running state can start the link state
timer, so as to solve the problem that the link state timer accesses the
null pointer and causes the segment error.
Fixes: 2790c64647 ("net/hns3: support device reset")
Cc: stable@dpdk.org
Signed-off-by: Hongbo Zheng <zhenghongbo3@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Currently, there is a certain probability of a type of RAS errors when
receiving broadcast packets. This type of RAS errors are parsed as
rx_q_search_miss error by hns3 PF PMD driver, the related log as below:
0000:bd:00.0 hns3_find_highest_level(): PPP_MFP_ABNORMAL_INT_ST2
rx_q_search_miss found [error status=0x20000000]
When receiving broadcast packet, network engine select which functions
need to accept it according to the function's promisc_bcast_en bit
configuration. And then search TQP_MAP configuration to select which
hardware queues the packet should enter, if can't find the target
hardware queue, network engine will trigger rx_q_search_miss RAS error.
The root cause as below:
1. VF's promisc_bcast_en bit configuration is not cleared by FLR reset,
and the configuration has been set in the previous application.
2. There is one bug in setting TQP_MAP configuration in the
initialization of PF device: when tqp is allocated to VF, it is still
marked as PF. This issue will affect the correctness of the TQP_MAP
configuration.
This patch fixes it with the following modification.
1. Clear all VFs promisc_bcast_en bit in the initialization of PF
device.
2. Fix the issue to ensure the correct TQP_MAP configuration.
Fixes: d51867db65 ("net/hns3: add initialization")
Cc: stable@dpdk.org
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
This patch optimizes the Rx performance by adding the rx_free_thresh
related process in the '.rx_pkt_burst' ops implementation function named
hns3_recv_pkts. The related change as follows:
1. Adding the rx_free_thresh related process to reduce the number of
writing the HNS3_RING_RX_HEAD_REG register.
2. Adjusting the internal macro named DEFAULT_RX_FREE_THRESH to 32 and
adjusting HNS3_MIN_RING_DESC to 64 to make the effect of the thresh
more obvious.
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
When receiving a packet, hns3 hardware network engine firstly writes the
packet content to the memory pointed by the 'addr' field of the Rx
Buffer Descriptor, secondly fills the result of parsing the packet
include the valid field into the Rx Buffer Descriptor in one write
operation, and thirdly writes the number of the Buffer Descriptor not
processed by the driver to the HNS3_RING_RX_FBDNUM_REG register.
This patch optimizes the Rx performance by removing one rte_io_rmb call
in the '.rx_pkt_burst' ops implementation function named hns3_recv_pkts.
The change as follows:
1. Driver no longer read HNS3_RING_RX_FBDNUM_REG register, so remove one
rte_io_rmb call, and directly read the valid flag of Rx Buffer
Descriptor to check whether the BD is ready.
2. Delete the non_vld_descs field from the statistic information of the
hns3 driver because now it has become a common case that the valid
flag of Rx Buffer Descriptor read by the driver is invalid.
Signed-off-by: Chengwen Feng <fengchengwen@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
This patch reduces the number of the judgement of the free Tx ring space
in the 'tx_pkt_burst' ops implementation function to avoid performance
loss. According to hardware constraints, we need to reserve a Tx Buffer
Descriptor in the TX ring in hns3 network engine.
Signed-off-by: Yisen Zhuang <yisen.zhuang@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Currently, we only support VF device is bound to vfio_pci or igb_uio and
then driven by DPDK driver when PF is driven by kernel mode hns3 ethdev
driver, VF is not supported when PF is driven by hns3 DPDK driver.
This patch adds support setting VF MAC address by hns3 PF kernel ethdev
driver on the host by "ip link set ..." command.
1) If the hns3 PF kernel ethdev driver sets the MAC address for VF
device before the initialization of the related VF device, hns3 VF
PMD driver should get the MAC address from PF driver through mailbox
and configure hardware using this MAC address in the initialization.
The hns3 VF PMD driver get the MAC address form PF driver, if
obtaining a non-zero MAC address from mailbox, VF driver will
configure hardware using it, otherwise using a random MAC address in
the initialization of VF device.
2) If the hns3 PF kernel ethdev driver sets the MAC address for VF
device after the initialization of the related VF device, the PF
driver will notify VF driver to reset VF device to make the new MAC
address effective immediately. The hns3 VF PMD driver should check
whether the MAC address has been changed by the PF kernel netdevice
driver, if changed VF driver should configure hardware using the new
MAC address in the recovering hardware configuration stage of the
reset process.
3) When user has configured a mac address for VF device by "ip link set
..." command based on the PF device, the hns3 PF kernel ethdev driver
does not allow VF driver to request reconfiguring a different default
mac address and return -EPREM to VF driver through mailbox.
Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Signed-off-by: Hongbo Zheng <zhenghongbo3@huawei.com>
Hardware does not support individually enable/disable/reset the Tx or Rx
queue in hns3 network engine, driver must enable/disable/reset Tx and Rx
queues at the same time.
Currently, hns3 PMD driver does not support the scenarios as below:
1) When calling the following function, the input parameter nb_rx_q and
nb_tx_q are not equal.
rte_eth_dev_configure(uint16_t port_id, uint16_t nb_rx_q,
uint16_t nb_tx_q,
const struct rte_eth_conf *dev_conf);
2) When calling the following functions to setup queues, the
cumulatively setup Rx queues are not the same as the setup Tx queues.
rte_eth_rx_queue_setup(uint16_t port_id, uint16_t rx_queue_id,,,);
rte_eth_tx_queue_setup(uint16_t port_id, uint16_t tx_queue_id,,,);
However, these are common usage scenarios in some applications, such as,
l3fwd, ip_ressmbly and OVS-DPDK, etc.
This patch adds support for this usage of these functions by setup fake
Tx or Rx queues to adjust numbers of Tx/Rx queues. But these fake queues
are imperceptible, and can not be used by upper applications.
Signed-off-by: Huisong Li <lihuisong@huawei.com>
Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
Adding FDIR support for L2TPv3 over IP header matching and adding
a new customized pctype for l2tpv3 over IP.
Signed-off-by: Rory Sexton <rory.sexton@intel.com>
Signed-off-by: Dariusz Jagus <dariuszx.jagus@intel.com>
Acked-by: Beilei Xing <beilei.xing@intel.com>
The ring is used only by af_xdp PMD itself, so no need to support
multi-producer and multi-consumer mode. This patch changes the ring
to single-producer and single-consumer mode, which could yield better
performance for addr enqueue and dequeue.
Signed-off-by: Xiao Wang <xiao.w.wang@intel.com>
Tested-by: Ciara Loftus <ciara.loftus@intel.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
The tx_burst routine supporting multi-segment packets with
legacy MPW and without inline was missed, and there was no
valid selection for these options, patch adds the missing
routine.
Fixes: 82e75f8323 ("net/mlx5: fix legacy multi-packet Tx descriptors")
Cc: stable@dpdk.org
Signed-off-by: Viacheslav Ovsiienko <viacheslavo@mellanox.com>
Acked-by: Matan Azrad <matan@mellanox.com>
Use the API rte_ether_addr_copy to do MAC assignment, instead of
calling rte_memcpy function directly.
Fixes: 84dc7a95a2 ("net/ice: enable flow director engine")
Cc: stable@dpdk.org
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Acked-by: Qiming Yang <qiming.yang@intel.com>
Acked-by: Xiaolong Ye <xiaolong.ye@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Change the iavf base code as driver common library, it is used by iavf
PMD now, and it can be used by other Intel SR-IOV PMDs in the future.
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Acked-by: Qi Zhang <qi.z.zhang@intel.com>
