Age | Commit message (Collapse) | Author | Files | Lines |
|
Set the mask of calculating the next cqe index to the corresponding CQ
size instead of rxq size.
Type: fix
Signed-off-by: Jieqiang Wang <jieqiang.wang@arm.com>
Change-Id: I67494f029967af64051f51452eba1fd699984cd9
|
|
Previously we encountered the issue of failing to create completion
queues on some Arm platforms because DPDK may set MLX5_CQE_SIZE to 128
if DPDK MLX PMDs are built and DPDK plugin is loaded, which does not
satisfy the requirement of 64B size CQE by RDMA plugin.
We fixed this issue in 844a0e8b0("always use 64 byte CQEs for MLX5"),
but some of CSIT test cases failed due to this code change. It turns out
that we don't need to specify compressed CQE mode for txq CQ because
RDMA tx doesn't have the code logic to handle compressed CQEs, which
might cause unexpected behavior if it is enabled.
Type: fix
Fixes: 844a0e8b0 ("always use 64 byte CQEs for MLX5")
Signed-off-by: Jieqiang Wang <jieqiang.wang@arm.com>
Change-Id: I7909a6d44b15bcf39c15dfac9377b65520a0cbfb
|
|
When DPDK MLX PMDs are built, and the DPDK plugin is loaded, DPDK may
set the MLX5_CQE_SIZE environment variable to 128. This causes the RDMA
plugin to be unable to create completion queues. Since the RDMA plugin
expects the CQEs to be 64 bytes, set the cqe_size explicitly when
creating the CQ. This avoids any issues with different values for the
MLX5_CQE_SIZE environment variable.
Type: improvement
Signed-off-by: Nathan Brown <nathan.brown@arm.com>
Change-Id: Idfd078d3045a4dcb674325ef36f85a89df6fbebc
|
|
When n_rx_packets is less then 16(VEC256) or 8(VEC128), code execution
will fall into scalar path of processing packets. But with a wrong
initialization value for n_left set to zero, i in the for-loop will
equal to n_rx_packets. This leads to the bypass of required ip4 checksum
validation and byte count endianness conversion in scalar path.
Besides, refactor the code using while instead of for-loop to keep
consistency with VPP code style.
Type: fix
Fixes: bf93670c515d ("rdma: fix ipv4 checksum check in rdma-input node")
Signed-off-by: Lijian Zhang <lijian.zhang@arm.com>
Signed-off-by: Jieqiang Wang <jieqiang.wang@arm.com>
Change-Id: Ib4e8cb5202735f8b060c99caddf26035657551e1
|
|
CQE flags located in bits 16-31 at offset 0x1c should be defined as
actual numbers instead of indexes. Besides, L3 header type for IPv4 is
10(2 in decimal) and for IPv6 is 01(1 in decimal) according to CQE entry
fields description of page 120 in Mellanox Programmer Reference Manual.
(https://network.nvidia.com/files/doc-2020/ethernet-adapters-programming-manual.pdf)
Fixing this issue will lead to correct CQE flags printing for rdma-input
node when buffer trace is enabled.
Type: fix
Signed-off-by: Jieqiang Wang <jieqiang.wang@arm.com>
Change-Id: I9b578ca5cbd8cd93a577aa83131e31c79f60430e
|
|
- cqe_flags pointer should be incremented accordingly otherwise only the
first element in cqe_flags will be updated
- flag l3_ok should be set for match variable when verifying if packets
are IPv4 packets with flag l3_ok set
- mask/match variables should be converted to network byte order to
match the endianness of cqe_flags
- vector processing of checking cqe flags will set return value to
0xFFFF by mistake if packet numbers are not multiple of 16(VEC256) or
8(VEC128)
Type: fix
Signed-off-by: Jieqiang Wang <jieqiang.wang@arm.com>
Change-Id: I9fec09e449fdffbb0ace8e5a6ccfeb6869b5cac1
|
|
Change-Id: I2040b560b2a00f8bd176ae6ad46035678a2b249e
Type: improvement
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
|
|
flags is u64, makes sure we do not overflow when shifting.
Type: fix
Change-Id: Ieea34187c0b568dc4d24c9415b9cff36907a5a87
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
- fix branch prediction for checking rdma ERROR flag
- add the missing right angle bracket to help message
Type: improvement
Signed-off-by: Jieqiang Wang <jieqiang.wang@arm.com>
Reviewed-by: Lijian Zhang <lijian.zhang@arm.com>
Reviewed-by: Tianyu Li <tianyu.li@arm.com>
Change-Id: I2ce667631b3e3f60939069e2a16ddba0ff12a695
|
|
Type: improvement
Change-Id: I3659de6599f402c92e3855e3bf0e5e3388f2bea0
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
- per hw-interface-class handlers
- ethernet set_mtu callback
- driver can now refuse MTU change
Type: improvement
Change-Id: I3d37c9129930ebec7bb70caf4263025413873048
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Prep for supporting multiple callbacks, optional args, etc.
Type: improvement
Change-Id: I96244c098712e8213374678623f12527b0e7f387
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Make it shorter to type, easier to debug, make adding callbacks in
future simpler.
Type: improvement
Change-Id: I6cdd6375e36da23bd452a7c7273ff42789e94433
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
REPLY_MSG_ID_BASE is the standard way to define reply message id base,
so this refactor makes all the files use that. This is a preparation
patch for future safety add-ons which rely on REPLY_MACRO* parameters to
be preprocessor tokens identifying the message instead,
Type: refactor
Signed-off-by: Klement Sekera <ksekera@cisco.com>
Change-Id: Ibe3e056a3d9326d08af45bbcb25588b11e870141
|
|
Type: improvement
Change-Id: I7e821cce1feae229e1be4baeed249b9cca658135
Signed-off-by: Nathan Skrzypczak <nathan.skrzypczak@gmail.com>
|
|
Remove aggressive inlining outside of the main loop to improve build
time (from 146s to 22s).
Type: refactor
Change-Id: I3824516a85b5e8d02894e66f19d891569c1a68fb
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: refactor
Change-Id: Id10cbf52e8f2dd809080a228d8fa282308be84ac
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
When switching to the direct verb chain buffer tx path, we must account
for all remaining packets, including the packets that would wrapped
around.
Previously we were using the 'n' counter but ignoring the 'n_wrap'
counter: if some packets would have wrapped around in the default path,
it would be ignored by the chained buffer tx path.
Compute the correct number of remaining packets based on the old and
current txq tail instead.
Also simplify the chained tx function parameters.
Type: fix
Change-Id: If12b41a8f143fda80290342e2904792f7501c559
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: feature
Change-Id: I911ee94b624ef5edf46b34666597eff93df1d1ed
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: improvement
Change-Id: I7b78a8470aa3fd271120587d08e82afb23cb195b
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: fix
Change-Id: Ifc4b30675365ebb5d69c018a34fd17a5090b30e1
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: make
Change-Id: I780c1c81a50cb92bc89c05856efd8ef88479c0ab
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
The memory region is already registered right above, looks like a
copy/paste error.
Type: fix
Change-Id: I97aed821e719e1a34ac38c86d0473a8fdd671d4e
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Current rdma input L3 validating behavior for scalar path is:
if any packet L3_OK flag matches, then unset skip_ip4_cksum.
The correct behavior should be if any packet L3_OK NOT match,
then unset skip_ip4_cksum. The logic is also different from
the vector path. This patch fixes the wrong behavior for scalar path.
Type: fix
Signed-off-by: Tianyu Li <tianyu.li@arm.com>
Change-Id: I5ca5ed3aa0c07d441f3c87b33f03ea8f7a3c9826
|
|
Change-Id: I17228bd35cc7aa0d207a16502bf1376c75675302
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
Type: fix
|
|
Type: improvement
This patch adds flags to represent the modern NICs capabilities.
Change-Id: I96d38d9ab7eac55974d72795cd100d8337168e1e
Signed-off-by: Mohsin Kazmi <sykazmi@cisco.com>
|
|
Also apply style edits as proprosed by checkstyle.
Ticket: VPP-1971
Type: fix
Change-Id: I4332a4e32220f3076b4a373da01cc0022cde32f5
Signed-off-by: Vratko Polak <vrpolak@cisco.com>
|
|
Change-Id: Id539d36635f0ab9625dc2fc73630be39bead09af
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
Type: improvement
|
|
Change-Id: Id81b4d27845c4e91cef90a4b8649662942d3cba1
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
Type: fix
|
|
Change-Id: I623617ad3c80610805dd3cf2a5f371e6677f4844
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
Type: improvement
|
|
When using classifier to filter traces, not all packets will be traced.
In that case, we should only count traced packets.
Type: fix
Change-Id: I87d1e217b580ebff8c6ade7860eb43950420ae78
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: fix
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
Change-Id: I428c87e581db335362fef30e274db8947a896416
|
|
/vpp/src/plugins/rdma/rdma.h:203:17: error: field 'buffer_template' with variable sized type 'vlib_buffer_t' not at the end of a struct or class is a GNU extension [-Werror,-Wgnu-variable-sized-type-not-at-end]
vlib_buffer_t buffer_template;
Type: fix
Change-Id: I4661839f262e01fe274a2ee7b3cb70f9bc6b7c62
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
This change leverages the striding RQ feature of
ConnectX-5 adapters to support chained buffers on
the RX path. In Striding RQ mode, WQE are SG lists
of data segments, each mapped to a vlib_buffer.
When a packet is received, it can consume one or
multiple data segments belonging to the WQE,
without wasting the whole WQE.
Change-Id: I74eba5b2c2c66538e75e046335058ba011cb27fd
Type: improvement
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
|
|
Type: feature
Change-Id: I8b0d918e6f13325954b29bf34e4ef224c1315c51
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
allocation, the remainder modulo 8 must be returned to the pool so that the number of kept buffers is a multiple of 8.
Type: fix
Change-Id: Ifd97b03ea220300e7e6fe81a8ff4a25060fea6c1
Signed-off-by: Mohammed Hawari <mohammed@hawari.fr>
|
|
Fix and optimize DMAC check in ethernet-input node to utilize NIC or
driver which support L3 DMAC-filtering mode so that DMAC check can be
bypassed safely for interfaces/sub-interfaces in L3 mode.
Checking of interface in L3-DMAC-filtering state to avoid DMAC check
require the following:
a) Fix interface driver init sequence for devices which supports L3
DMAC-filtering to indicate its capability and initialize interface
to L3 DMAC-filtering state.
b) Fix ethernet_set_flags() function and its associated callback
flags_change() functions registered by various drivers in interface
infra to provide proper L3 DMAC filtering status.
Maintain interface/sub-interface L3 config count so DMAC checks can be
bypassed if L3 forwarding is not setup on any main/sub-interfaces.
Type: fix
Ticket: VPP-1868
Signed-off-by: John Lo <loj@cisco.com>
Change-Id: I204d90459c13e9e486cfcba4e64e3d479bc9f2ae
|
|
Type: fix
Change-Id: I0ff7a6f3354066c2252c1ebd0d43ac59db278bf0
Signed-off-by: Damjan Marion <damarion@cisco.com>
Signed-off-by: Elias Rudberg <elias.rudberg@bahnhof.net>
|
|
Type: fix
Change-Id: I9b613f0af484f601dd20a851e2f59ee5e06b5c37
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: improvement
Change-Id: Ic2d9b17cf5e524f3ad2a3c5343fe1230aa360e73
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: improvement
Change-Id: I275bbca17c5a0263b3e017b48aa6ccd8f59bc7c3
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: I09b017d60bc38ff8a4a7a73aff42b27611bd8955
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: improvement
Change-Id: I7f28a3f03ab1ea8461c52743c61dc23a57965237
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Direct Verb allows for direct access to NIC HW rx/tx rings. This patch
introduce TX direct verb support for Mellanox ConnectX-4/5 adapters.
'dv' mode must be explicitely selected at interface creation to benefit
from this.
Type: feature
Change-Id: If830ba9f33db73299acdbddc68b5c09eaf6add98
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|
|
Type: feature
Change-Id: I3f287ab536a482c366ad7df47e1c04e640992ebc
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: If81847bc0c92d167ce03e1e94a2f8e18f8154af2
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: Ib9921718ec88e6dc56a38e40a140f16f86ed2e44
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Type: improvement
Change-Id: Ie6f2c79e0a769f70eab079c75a500b9308dd51e6
Signed-off-by: Damjan Marion <damarion@cisco.com>
|
|
Fix a bug that caused some input packets to be dropped due to errors of
the type 'ip4 length > l2 length'. The change is related to the second
call to the rdma_device_input_bufs() function that happens when the end
of the ring buffer is reached.
Type: fix
Change-Id: I332d69ab22242b3443a0baca6e5dd86349a54765
Signed-off-by: Elias Rudberg <elias.rudberg@bahnhof.net>
|
|
Prepare rdma interface creation API for direct verb support:
- add new optional 'mode' parameters to select between ibverb or direct
verb backend (optional, default to 'auto')
- set default value for rxq_num (1), rxq_size (1024) and txq_size
(1024) so they are now optional
- bump default create value for rxq_size and txq_size to 1024 if
unset (0) so they are coherent with default values above
Type: feature
Change-Id: Id9eae2b8eb0baaf34a0fcd55da6ad09515f57a93
Signed-off-by: Benoît Ganne <bganne@cisco.com>
|