After VF reset, FDIR rule still takes effect. To solve the issue,
this patch adds to flush all flows before flow uninit. VIRTCHNL
sends message to PF by Admin Queue, so flow flush should be implemented
before Admin Queue shut down.
Type: fix
Signed-off-by: Chenmin Sun <chenmin.sun@intel.com>
Change-Id: I8ba0db7cd7646eaabd5745f74952016b0b968bbb
0001 ~ 0014 patches are for virtual channel and PMD
0015 is the iavf fdir framework
0016 ~ 0017 are for the iavf fidr driver
Type: feature
Signed-off-by: Chenmin Sun <chenmin.sun@intel.com>
Change-Id: I38e69ca0065a71cc6ba0b44ef7c7db51193a0899
This issue is observed with X520-2 NICs on FD.io lab Taishan server.
After VPP booting up and bringing up the interfaces with command "set
interface state <interface> up", it still shows link down status from
the command "show hardware-interfaces". However, the hardware link
status is actually up. dpdk_process() cannot get the hardware link
status correctly via rte_eth_link_get_nowait().
In ixgbe_dev_link_update_share(), if the media type is fiber and the
link is down, a flag (IXGBE_FLAG_NEED_LINK_CONFIG) is set. A callback to
ixgbe_dev_setup_link_alarm_handler() is scheduled trying to set up the
link and clear the flag afterwards.
If the device is started or stopped before the flag is cleared, the
scheduled callback is canceled. This causes the flag to remain set and
subsequent calls to ixgbe_dev_link_update_share() return without trying
to retrieve the link state because the flag is set.
When the callback is canceled by either interface start or stop
operation, in ixgbe_dev_cancel_link_thread(), after cancelling the
callback/thread, unset the flag on the device to avoid this condition.
Type: fix
Signed-off-by: Lijian Zhang <Lijian.Zhang@arm.com>
Reviewed-by: Honnappa Nagarahalli <Honnappa.Nagarahalli@arm.com>
Reviewed-by: Jieqiang Wang <Jieqiang.Wang@arm.com>
Change-Id: I04de377dc048307a78a5b7109ebdfaf376d5e029
Type: fix
Some fiber ports that are managed by the ixgbe PMD have the
possibility to get into a state where link can never be brought up.
This patch should fix it and will be submitted to upstream DPDK.
Change-Id: Ia4d0df2e70d098b2151e513b96e8bd742151e8ce
Signed-off-by: Matthew Smith <mgsmith@netgate.com>
This patch bumps the engine's intel-ipsec-mb version from 0.52 to
0.53, to avail performance improvement brought by the library.
Type: feature
Signed-off-by: Fan Zhang <roy.fan.zhang@intel.com>
Change-Id: I939803015dece91ca59a7fe2120e2131d29e1c55
- Replace the solution to the quicly time skew assert
with h2o/quicly PR#222 which was merged upstream
after verification that it resolved the vpp issue.
Type: refactor
Change-Id: I26df08e4108b054a28f50c964ddff1c80a592339
Signed-off-by: Dave Wallace <dwallacelf@gmail.com>
This adds support to compile the bnxt PMD that supports Broadcom's
10/25/40/50/100/200 Gbps NICs. Tested with modified DPDK driver on
x86_64 and aarch64 targets that is pending upstream acceptance, but
should compile just fine today.
Type: feature
Signed-off-by: Andy Gospodarek <gospo@broadcom.com>
Change-Id: I735a991c4cae4fa77e5605094facea54b781a1db
- This patch to quicly fixes an intermittent failure
in the rtt calculation when there is a time skew
Type: fix
Change-Id: If89c47401ee75c88f0eb65ae888b6914695b2aec
Signed-off-by: Dave Wallace <dwallacelf@gmail.com>
Type: fix
The scenario is touch of a dpdk source file in places like
<TOP>/build-root/build-vpp-native/external/dpdk-19.08/. The subsequent
build will trigger an rm -rf $(PACKAGE_INSTALL_DIR) by
build-root/Makefile at around line 709. Because the package in
question, "external", consists of various "sub" packages, all of which
are controlled with files like .quicly.install.ok located in
build-root/build-vpp[_debug]-native/external, we end up not handling the
.*.install.ok files to reflect the removal of that PACKAGE_INSTALL_DIR.
This fix will correct that, at least for the case of touching/changing
dpdk files. A better long term solution would be to change the "sub"
packages to be full fledged build system packages, as Dave Barach points
out.
Signed-off-by: Burt Silverman <burtms@gmail.com>
Change-Id: Ia94ccaa2795cbf354b5921c3001a7141313b66be
rdma-core-25.0 move some symbol to another library, make sure we link
to it.
Type: fix
Fixes: ec757593859bd8c16839ea8a81a2ef4eb8b29539
Change-Id: I6b8a56ea5fc7787d3de0982015ed47b767557c06
Signed-off-by: Benoît Ganne <bganne@cisco.com>
Update quicly to latest version that includes our upstreamed patch.
Change-Id: I0b26c72e49bce81daf4fb069b5818defd6cf25b9
Type: feature
Signed-off-by: Aloys Augustin <aloaugus@cisco.com>
It's fine to remove the old .rpm, but don't stop the parade if there
wasn't an old .rpm to remove.
Type: fix
Change-Id: I500c740cc4ded16d3345c1e98462a1e8533f1028
Signed-off-by: Dave Barach <dave@barachs.net>
We have native implementation and we should not maintain both....
Change-Id: Ic09ebffda52cdc733b3cfeff06690e0d3cc08084
Signed-off-by: Damjan Marion <damarion@cisco.com>
The current aarch64 version of VPP package distro in cloud repository
(https://packagecloud.io/fdio/master/ubuntu), is built on a ThunderX server,
using some arch-specific options, for example, 128Byte cache line size,
T=arm64-thunderx-linuxapp-gcc, RTE_MACHINE=thunderx
The patch is trying to build package distro with aarch64 generic features,
for both binary type targets, e.g., build/build-release, and package type
targets, e.g., pkg-deb/pkg-rpm, with the generic options, e.g.,
128Byte cache line size, T=arm64-armv8a-linuxapp-gcc, RTE_MACHINE=armv8a
If end users want to build arch specific optimized image,
TARGET_PLATFORM variable could be used, as below example,
$ make build-release/pkg-deb TARGET_PLATFORM=thunderx
Change-Id: If78bca8709fe83db6a95e8c26346f206bf5ea71d
Signed-off-by: Lijian Zhang <Lijian.Zhang@arm.com>
Reviewed-by: Sirshak Das <Sirshak.Das@arm.com>
Reviewed-by: Honnappa Nagarahalli <Honnappa.Nagarahalli@arm.com>
Add option RDMA_CORE_DEBUG=[yn] build option. rdma-core package is built
in Release (-O2) mode by default, but Debug (-O0 -g) mode can be
selected by eg.
make install-ext-deps RDMA_CORE_DEBUG=y
Change-Id: Id4db5ef1e7b13c44d478755a0bfbdbe56cfc13df
Signed-off-by: Benoît Ganne <bganne@cisco.com>
VPP rdma driver relies on an internal rdma-core not compatible with DPDK
MLX driver. Force the use of external rdma-core through
RTE_IBVERBS_LINK_DLOPEN DPDK build option and make sure internal
rdma-core symbols are not leaked outside of the rdma plugin.
Change-Id: I5b2281259f517c4e109d388d172b72eadd69986f
Signed-off-by: Benoît Ganne <bganne@cisco.com>
rx: add batching for WC processing and release
tx: improve batching for WC submission and processing
rdma-core: compile in release mode to remove assert()
Change-Id: I5fb8736db36b50f8b758cd688100477b67e72d80
Signed-off-by: Benoît Ganne <bganne@cisco.com>
Recent patches add rdma-core.
They are not building in Centos because of differences
in cmake. This patch fixes that problem by defining
CMAKE for all external builds.
Change-Id: I3eea869226385207500c5a1217bd7330140ca462
Signed-off-by: Thomas F Herbert <therbert@redhat.com>
Download a single tarball with the sources of quicly and its dependencies from github instead of cloning submodules.
Change-Id: Id9955565d46f595b3a14ba3408c24045d4acd296
Signed-off-by: Aloys Augustin <aloaugus@cisco.com>
Device speed capability should be specified based on different phy types
instead of a fixed value, this patch fix the issue.
Change-Id: Ia76231aefbcb0fe8370867b6e86a0d3bb9e169a0
Signed-off-by: Chenmin Sun <chenmin.sun@intel.com>
We dont't want git pull there, there is simple way to grab tarball
from github for every single commit
Change-Id: I7a98cab7660750e48084279d8b9892cdf229bceb
Signed-off-by: Damjan Marion <damarion@cisco.com>
Currently supports on single stream exposed through standard internal APIs
Based on libquicly & picotls by h2o
Change-Id: I7bc1ec0e399d1fb02bfd1da91aa7410076d08d14
Signed-off-by: Nathan Skrzypczak <nathan.skrzypczak@gmail.com>
RDMA ibverb is a userspace API to efficiently rx/tx packets. This is an
initial, unoptimized driver targeting Mellanox cards.
Next steps should include batching, multiqueue and additional cards.
Change-Id: I0309c7a543f75f2f9317eaf63ca502ac7a093ef9
Signed-off-by: Benoît Ganne <bganne@cisco.com>
The ingress vlan rewrite mode affects the vlan header of ingress
packet. The enic driver currently uses the 'passthrough' mode, which
leaves the vlan header intact. As all packets in UCS network are
tagged, this default mode leads to tagged ingress packets in the
following cases where VPP expects untagged packets.
1. Trunk-mode vNIC on UCS standalone server.
A remote device sends an untagged packet to the server. This packet is
ultimately tagged with vlan 0 as it reaches the driver, and VPP sees
an ingress packet tagged with vlan 0.
2. Access-mode vNIC on UCS blade or standalone server.
A remote device sends a packet on the vNIC's default vlan (e.g. 200)
to the server. This tag is untouched, and VPP sees an ingress packet
tagged with the default vlan (e.g. 200).
In both cases, VPP expects to see untagged packets. To work around the
issue, VPP currently enables vlan stripping on VIC interfaces, which
breaks vlan sub-interface features.
To avoid the current workaround, use the "untag default vlan" rewrite
mode. With this mode, the VIC adapter removes the vlan header if it
matches the default vlan. In the cases described above, VPP would see
untagged packets. Packets tagged with non-default vlan (e.g. non-0 for
case 1 and non-200 for case 2) are received with their tags intact, so
VPP sees tagged packets as expected.
The driver currently has no programmatic way to change the rewrite
mode after rte_eal_init. So use this patch to change the mode for the
time being.
Change-Id: Iff6408275363ed52d6016e7516d745214d6b30d4
Signed-off-by: Hyong Youb Kim <hyonkim@cisco.com>
Upstream DPDK changed the names of the options to use dlopen()
with libibverbs and libmlx[45] from RTE_LIBRTE_MLX[45]_DLOPEN_DEPS
to RTE_IBVERBS_LINK_DLOPEN (handles both mlx4 and mlx5).
VPP's build option to enable this configuration when building DPDK
no longer worked starting when VPP moved to DPDK 19.02. Update VPP's
build options to enable the correct option name.
Change-Id: I8e34e1d3fc4ee8aac4fd6f2a7d27177f2b0dea50
Signed-off-by: Matthew Smith <mgsmith@netgate.com>
