Qiming Chen [Mon, 30 Aug 2021 03:31:57 +0000 (11:31 +0800)]
net/ixgbe: fix Rx multicast statistics after reset
In the implementation of the VF driver ixgbevf_update_stats to obtain
statistics, the multicast count hw_stats->vfmprc has been obtained,
but it is not cleared in the corresponding ixgbevf_dev_stats_reset
interface.
Fixes: abf7275bbaa2 ("ixgbe: move to drivers/net/") Cc: stable@dpdk.org Signed-off-by: Qiming Chen <chenqiming_huawei@163.com> Acked-by: Haiyue Wang <haiyue.wang@intel.com>
Tudor Cornea [Thu, 5 Aug 2021 11:35:23 +0000 (14:35 +0300)]
net/iavf: fix overflow in maximum packet length config
The len variable, used in the computation of max_pkt_len could
overflow, if used to store the result of the following computation:
rxq->rx_buf_len * IAVF_MAX_CHAINED_RX_BUFFERS
Since, we could define the mbuf size to have a large value (i.e 13312),
and IAVF_MAX_CHAINED_RX_BUFFERS is defined as 5, the computation
mentioned above could potentially result in a value which might be
bigger than MAX_USHORT.
The result will be that Jumbo Frames will not work properly
Wenjun Wu [Tue, 17 Aug 2021 08:45:02 +0000 (16:45 +0800)]
net/ice: fix default RSS hash for IP fragments
Previously, hash value is calculated by src IP address, dst IP address
and IP ID. However, default RSS field only needs src and dst IP address.
This patch removes IP ID from default RSS field for IP fragment packets
to improve default RSS configuration.
Fixes: 4027fffe86f4 ("net/ice: support default RSS for IP fragment packet") Cc: stable@dpdk.org Signed-off-by: Wenjun Wu <wenjun1.wu@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Wenjun Wu [Tue, 17 Aug 2021 06:49:55 +0000 (14:49 +0800)]
net/iavf: fix flow director L3 field for IPv4 fragments
Originally, the value of field_selector for IPV4_FRAG header hdr1 is
the same as the previous header hdr2. For IPv4 packets, field_selector
for hdr2 can be any value between 0 and 4, depending on the selected
field. Actually, this value for IPV4_FRAG should be constant 0,
which denotes the field packet ID.
This patch adds an assignment to hdr1->field_selector to make sure that
it is always 0.
Fixes: 3334513ef484 ("net/iavf: support flow director for IP fragment") Cc: stable@dpdk.org Signed-off-by: Wenjun Wu <wenjun1.wu@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
Heinrich Kuhn [Mon, 23 Aug 2021 07:59:21 +0000 (09:59 +0200)]
net/nfp: fix minimum descriptor sizes
The NFP4000/6000 supports a minimum of 256 Tx/Rx descriptors and not 64.
Before this patch when a DPDK application configured < 256 Tx
descriptors the hardware read/write pointers would be unmasked and not
wrapped at the expected size of the ring. The PMD logic to determine the
amount of free space in the Tx ring is incompatible with the unmasked
hwqueue pointers and this will result in the PMD not updating the
read pointer at all.The knock-on effect is that under high load the
PMD will potentially re-use a Tx descriptor before the hw has had a
chance to process that particular descriptor. This issue can manifest as
a DMA error when the hardware tries to perform a DMA with info from a
partially populated descriptor.
Fixes: defb9a5dd156 ("nfp: introduce driver initialization") Cc: stable@dpdk.org Signed-off-by: Heinrich Kuhn <heinrich.kuhn@corigine.com> Signed-off-by: Simon Horman <simon.horman@corigine.com>
This patch add support to handle PDCP short MAC-I domain
along with standard control and data domains as it has to
be treaty as special case with PDCP protocol offload support.
ShortMAC-I is the 16 least significant bits of calculated MAC-I. Usually
when a RRC message is exchanged between UE and eNodeB it is integrity &
ciphered protected.
MAC-I = f(key, varShortMAC-I, count, bearer, direction).
Here varShortMAC-I is prepared by using (current cellId, pci of source cell
and C-RNTI of old cell). Other parameters like count, bearer and
direction set to all 1.
crypto-perf app is updated to take short MAC as input mode.
Srujana Challa [Wed, 18 Aug 2021 11:56:35 +0000 (17:26 +0530)]
common/cnxk: fix attaching NPA LF to CPT VF
ATTACH_RESOURCES mailbox for CPT LFs is being called without
setting modify bit. Because of this previously attached NPA LF
to the CPT VF is getting removed, when only CPT VF is whitelisted.
This patch fixes the same.
Akhil Goyal [Tue, 10 Aug 2021 19:50:19 +0000 (01:20 +0530)]
examples/fips_validation: remove illegal usage of API
Some of the cryptodev APIs are not allowed to be used
by application directly. Hence removing the usage of
1. queue_pair_release: it is not required, as configure
of queue pair release the previous queue pairs and the
dev is not directly exposed to application, hence cannot
use its ops from app.
2. rte_cryptodev_stop: it can be used directly without
checking if the device is started or not.
3. rte_cryptodev_pmd_destroy: application should use
rte_cryptodev_close instead.
Akhil Goyal [Tue, 10 Aug 2021 19:50:18 +0000 (01:20 +0530)]
cryptodev: rename function to check device validity
The API rte_cryptodev_pmd_is_valid_dev, can be used
by the application as well as PMD to check whether
the device is valid or not. Hence, _pmd is removed
from the API.
The applications and drivers which use this API are
also updated.
Akhil Goyal [Tue, 10 Aug 2021 19:50:17 +0000 (01:20 +0530)]
test/crypto: remove illegal PMD header include
rte_cryptodev_pmd.h is an interface between
driver and library and it is mentioned in the
file that application cannot use it directly.
Hence, removing the include.
When given a source length 0 for CCM, the encryption and decryption
functions did not call the EVP_ENCRYPTUPDATE/EVP_DECRYPTUPDATE functions
with a src and dst, causing some FIPS validation failures for testcases
with PLen=0:
process_openssl_auth_encryption_ccm() line 1131:
Process openssl auth encryption ccm failed
Fixes: 1a4998dc4d94 ("crypto/openssl: support AES-CCM") Cc: stable@dpdk.org Signed-off-by: Ciara Power <ciara.power@intel.com> Acked-by: Fan Zhang <roy.fan.zhang@intel.com>
Previously, the scheduler unit test only created and attached 1 or 2
AESNI-MB cryptodev PMDs as workers if less than 2 AESNI-MB PMDs had
already been initialized.
This commit changes this to always create and attach 2 new AESNI-MB
cryptodev PMDs, regardless of previously initialized AESNI-MB PMDs.
Signed-off-by: Rebecca Troy <rebecca.troy@intel.com> Acked-by: Fan Zhang <roy.fan.zhang@intel.com>
Suppress the gcc warning "note: the layout of aggregates containing
vectors with 4-byte alignment has changed in GCC 5" on POWER systems
by setting "-Wno-psabi". Warning was originally added to gcc in
commit https://gcc.gnu.org/git/gitweb.cgi?p=gcc.git;h=9832651 to warn
of the vector alignment changes introduced in GCC 5. Older gcc
versions forced vector alignment to 16 bytes due to requirements for
POWER 6 and earlier CPUs, but these restrictions don't apply to CPUs
supported by DPDK.
Bugzilla ID: 739 Cc: stable@dpdk.org Signed-off-by: David Christensen <drc@linux.vnet.ibm.com>
Suppress gcc warning "warning: writing 16 bytes into a region of
size 0" for users of the POWER rte_memcpy() function. Existing
rte_memcpy() code takes different code paths based on the actual
size of the move so the warning is already addressed. See also
commit b5b3ea803e47 ("eal/x86: ignore gcc 10 stringop-overflow warnings")
Cc: stable@dpdk.org Signed-off-by: David Christensen <drc@linux.vnet.ibm.com>
Michael Baum [Sun, 5 Sep 2021 08:04:15 +0000 (11:04 +0300)]
crypto/mlx5: fix timestamp format configuration
This patch adds support for the timestamp format settings for
the receive and send queues. If the firmware version x.30.1000
or above is installed and the NIC timestamps are configured
with the real-time format, the default zero values for newly
added fields cause the queue creation to fail.
The patch queries the timestamp formats supported by the hardware
and sets the configuration values in queue context accordingly.
Fixes: 6152534e211e ("crypto/mlx5: support queue pairs operations") Cc: stable@dpdk.org Signed-off-by: Michael Baum <michaelba@nvidia.com> Acked-by: Matan Azrad <matan@nvidia.com>
CPT can be told to submit events to SSO upon completion.
Crypto adapter uses this feature and the new flag can be
used to optimize receive path in those cases.
Michael Baum [Tue, 31 Aug 2021 20:39:41 +0000 (23:39 +0300)]
compress/mlx5: fix leak on QP setup failure
The QP setup function allocates buffer for its
opaque MR and register it into MR structure.
After buffer allocation and before MR registration,
it tries allocate MR Btree.
When the MR Btree allocation fails, the buffer was
not freed what caused a memory leak.
Allocate the MR Btree before buffer alloction.
Fixes: 0165bccdb45f ("compress/mlx5: add memory region management") Cc: stable@dpdk.org Signed-off-by: Michael Baum <michaelba@nvidia.com> Acked-by: Matan Azrad <matan@nvidia.com>
Xueming Li [Tue, 13 Apr 2021 03:14:11 +0000 (03:14 +0000)]
bus: add device arguments name parsing
For device probe and iterator, devargs name was key information,
parsed by rte_devargs_parse. In legacy parser, devargs name was
extracted after bus name:
bus:name,kv_arguments,,,
Example:
pci:83:00.0,arguments,...
vdev:pcap0,...
To be compatible with legacy parser, this patch introduces new
bus driver API devargs_parse to parse devargs and update devargs name.
If devargs_parse not implemented by bus driver, the new syntax parser
rte_devargs_layers_parse default will resolve devargs name from bus's
"name" argument.
Different bus driver might choose different keys from arguments with
unified format. The PCI bus implementation fills the devargs name with
the "addr" argument, example:
-a bus=pci,addr=83:00.0/class=eth/driver=mlx5,...
name: 0000:03:00.0
-a bus=vdev,name=pcap0/class=eth/driver=pcap,...
name:pcap0
Signed-off-by: Xueming Li <xuemingl@nvidia.com> Reviewed-by: Gaetan Rivet <grive@u256.net>
Donot use outer metadata when neither outer ip checksum nor
outer udp checksum is enabled. PMD's will ignore the
outer_l2_len and outer_l3_len in cases where none of
the outer checksum is enabled and hence only l2_len and
l3_len will be used to calculate the offsets for L2 or L3
header.
Ajit Khaparde [Tue, 24 Aug 2021 01:58:20 +0000 (18:58 -0700)]
net/bnxt: fix VNIC config error in Rx queue start
During port stop/start sequence the Thor FW is returning an error.
This is because we are deriving incorrect active Rx ring and using
that wrong information in the bnxt_vnic_rss_cfg HWRM command.
Somnath Kotur [Mon, 23 Aug 2021 15:44:53 +0000 (21:14 +0530)]
net/bnxt: fix crash after port stop/start
On chips like Thor, port stop/start sequence could result in a crash
in the application. This is because of false detection of a bad
opaque in the Rx completion and the subsequent kicking-in of the ring
reset code to recover from the condition.
The root cause being that the port stop/start would result in the HW
starting with fresh values, while the driver internal tracker variable
`rx_next_cons` is still pointing to a stale value.
Fix this by resetting rx_next_cons to 0 in bnxt_init_one_rx_ring()
Fixes: 03c8f2fe111c ("net/bnxt: detect bad opaque in Rx completion") Cc: stable@dpdk.org Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com> Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Ajit Khaparde [Mon, 2 Aug 2021 04:44:30 +0000 (21:44 -0700)]
net/bnxt: update ring group after ring stop start
A Rx ring stop start sequence may result in the FW returning
a different set of Rx ring and AGG ring IDs. If the ring group
is not updated with the new IDs, the HW sees the host driver using
incorrect BD types for the Rx ring and AGG ring. This can cause
the chip to go into a bad state or encounter RE_flush issue
or leak mbufs in the HW.
Fix this by issuing a bnxt_hwrm_ring_grp_free() and an
bnxt_hwrm_ring_grp_alloc() to refresh the ring group information.
Dmitry Kozlyuk [Mon, 9 Aug 2021 14:26:46 +0000 (17:26 +0300)]
net/mlx5: fix eCPRI matching
When an ETH or VLAN flow item directly preceding ECPRI (i. e. a pattern
for eCPRI over Ethernet) did not specify the eCPRI protocol, matches
were not restricted to eCPRI traffic. For example, "eth / ecpri / end"
pattern behaved as "eth / end". Implicitly add Ethernet type condition,
so that "eth / ecpri / end" behaves as "eth type is 0xAEFE / end".
This crash is caused by an attempt to copy previously uncompressed CQEs
into non-allocated mbufs. There is a check to make sure we only use
allocated mbufs in the rxq_burst_mprq_v() function, but it is done only
before the main processing loop. Leftovers of compressed CQEs session are
handled before that loop and may lead to the mbufs overflow as seen.
Move the check for replenished mbufs up to protect uncompressed CQEs
session leftovers from accessing non-allocated mbufs after the
mlx5_rx_mprq_replenish_bulk_mbuf() function is invoked.
Lior Margalit [Tue, 3 Aug 2021 18:13:22 +0000 (21:13 +0300)]
net/mlx5: fix RSS expansion for inner tunnel VLAN
The RSS expansion algorithm is using a graph to find the possible
expansion paths. The VLAN item in the flow pattern requires special
treatment, because it should not be added implicitly by the expansion
algorithm. If the flow pattern ends with ETH item, the pattern will be
expanded with IPv4 and IPv6.
For example:
testpmd> flow create ... eth / end actions rss / end
ETH END
ETH IPV4 END
ETH IPV6 END
If a VLAN item follows the ETH item in the flow pattern, the pattern
will be expanded with IPv4 and IPv6 following the VLAN item.
For example:
testpmd> flow create ... eth / vlan / end actions rss level 1 / end
ETH VLAN END
ETH VLAN IPV4 END
ETH VLAN IPV6 END
The case of inner tunnel VLAN item was not taken care of so the flow
pattern did not expand with IPv6 and IPv4 as expected.
Example with inner VLAN:
testpmd> flow create ... / vxlan / eth / vlan / end actions rss level 2
/ end
The current result of the expansion alg:
ETH IPV6 UDP VXLAN ETH VLAN END
The expected result of the expansion alg:
ETH IPV6 UDP VXLAN ETH VLAN END
ETH IPV6 UDP VXLAN ETH VLAN IPV4 END
ETH IPV6 UDP VXLAN ETH VLAN IPV6 END
The fix is to introduce a new flag to set on a graph expansion node
to apply the 'explicit' behavior, meaning the node is not added to
the expanded pattern, if it is not found in the flow pattern, but the
expansion alg can go deeper to its next nodes.
Heinrich Kuhn [Thu, 29 Jul 2021 13:47:11 +0000 (15:47 +0200)]
net/nfp: rename files for consistency
Rename the nfp_net.c file to nfp_common as it now contains functions
common to VF and PF functionality. Rename the header file too to be
consistent. Also remove the "net" naming from the _ctrl and _logs files
for consistency across the PMD.
Signed-off-by: Heinrich Kuhn <heinrich.kuhn@netronome.com> Signed-off-by: Simon Horman <simon.horman@corigine.com>
Heinrich Kuhn [Thu, 29 Jul 2021 13:47:08 +0000 (15:47 +0200)]
net/nfp: move common function prototypes
The majority of "ethdev" type functions are used for both PF devices and
VF devices. Prototype these functions in the nfp_net_pmd header file in
preparation of splitting PF and VF specific functions.
Signed-off-by: Heinrich Kuhn <heinrich.kuhn@netronome.com> Signed-off-by: Simon Horman <simon.horman@corigine.com>
Heinrich Kuhn [Thu, 29 Jul 2021 13:47:06 +0000 (15:47 +0200)]
net/nfp: move datapath functions to their own file
Create a new rxtx file and move the Rx/Tx functions to this file. This
commit will also move the needed shared functions to the nfp_net_pmd.h
file as needed.
Signed-off-by: Heinrich Kuhn <heinrich.kuhn@netronome.com> Signed-off-by: Simon Horman <simon.horman@corigine.com>
Huisong Li [Mon, 26 Jul 2021 10:59:40 +0000 (18:59 +0800)]
net/hns3: support set link up/down for PF
This patch adds set link up/down feature. RxTx datapath and link status
will be disabled when dev_set_link_down() is called, and can be enabled by
dev_start() or dev_set_link_up().
Signed-off-by: Huisong Li <lihuisong@huawei.com> Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
Currently, hns3 PMD has supported start/stop RxTx datapath request message
between the primary and secondary processes. However, there are some cases
only to start/stop Tx datapath. This patch adds start/stop Tx datapath
request for MP.
Signed-off-by: Huisong Li <lihuisong@huawei.com> Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
The memory size of bandwidth config parameters is not set correctly in
memory copy process, which leads to the wrong values. This patch fixed
the size to the correct value.
Fixes: 3a6bfc37eaf4 ("net/ice: support QoS config VF bandwidth in DCF") Cc: stable@dpdk.org Signed-off-by: Ting Xu <ting.xu@intel.com> Acked-by: Qi Zhang <qi.z.zhang@intel.com>
git.droids-corp.org / dpdk.git / log