[dpdk-dev] [PATCH v7 0/5] Support VxLAN & NVGRE checksum off-load on X550
Liu, Yong
yong.liu at intel.com
Fri Mar 4 06:45:09 CET 2016
Tested-by: Yong Liu <yong.liu at intel.com>
- Tested Branch: dpdk-next-net
- Tested Commit: 5fa83b5398e26af7537b09605432fcb3d0cc1d41
- OS: Fedora21 3.17.4-301.fc21.x86_64
- GCC: gcc version 4.9.2 20150212
- CPU: Intel(R) Xeon(R) CPU D-1540 @ 2.00GHz
- NIC: Intel Corporation Device Sageville [8086:15ad]
- Default x86_64-native-linuxapp-gcc configuration
- Total 4 cases, 4 passed, 0 failed
- Prerequisites command / instruction:
Bound PF device to igb_uio
./tools/dpdk_nic_bind.py --bind=igb_uio 03:00.0
Create 2VF devices from PF device.
echo 2 > /sys/bus/pci/devices/0000\:03\:00.0/max_vfs
Detach VFs from the host, bind them to pci-stub driver
virsh # nodedev-dettach pci_0000_03_10_0
virsh # nodedev-dettach pci_0000_03_10_2
Passthrough VF 03:10.0 & 03:10.2 to vm0 and start vm0
/usr/bin/qemu-system-x86_64 -name vm0 -enable-kvm \
-cpu host -smp 4 -m 2048 -drive file=/home/image/fedora20.img -vnc :1 \
-device pci-assign,host=03:10.0 \
-device pci-assign,host=03:10.2
Login vm0 and them bind VF devices to igb_uio driver.
./tools/dpdk_nic_bind.py --bind=igb_uio 00:04.0 00:05.0
Bind PF to vfio-pci and start testpmd in host
testpmd -c f -n 3 -- -i
Config to rxonly mode and enable verbose output
testpmd> set fwd rxonly
testpmd> set verbose 1
testpmd> start
Start testpmd in guest, configured to mac forward mode
testpmd -c 0x3 -n 1 -- -i --txqflags=0x0
testpmd> set fwd mac
testpmd> start
- Case: E-tag tunnel filter
Description: check that E-tag tunnel filter can work as expected
Command / instruction:
Enable E-tag l2 tunnel support means enabling ability of parsing E-tag
packet. This ability should be enabled before we enable filtering,
forwarding, offloading for this specific type of tunnel
host testpmd> port config 0 l2-tunnel E-tag enable
Send 802.1BR packet to PF and VFs, check packet normally received
- Case: E-tag filter and forwarding
Description: check that E-tag forwarding work as expected
Command / instruction:
Enable E-tag packet forwarding and enable E-tag 1000 packet forward to VF0
testpmd> E-tag set forwarding on port 0
testpmd> E-tag set filter add e-tag-id 1000 dst-pool 0 port 0
Send E-tag packet with broadcast mac and check packet only received on VF0
Set E-tag with ID 1000 forwarding to VF1
testpmd> E-tag set filter add e-tag-id 1000 dst-pool 1 port 0
Send 802.1BR packet with broadcast mac and check packet only received on VF1
Set E-tag with ID 1000 PF0
testpmd> E-tag set filter add e-tag-id 1000 dst-pool 2 port 0
Send 802.1BR packet with broadcast mac and check packet only received on PF
Remove E-tag
testpmd> E-tag set filter del e-tag-id 1000 port 0
Send 802.1BR packet with broadcast mac and check packet not received
- Case: E-tag insertion
Description: check that E-tag insertion feature can work as expected
Command / instruction:
Enable E-tag insertion in VF0
testpmd> E-tag set insertion on port-tag-id 1000 port 0 vf 0
Send normal packet to VF1 and check forwarded packet contain E-tag
Remove E-tag insertion in VF0
testpmd> E-tag set insertion off port 0 vf 0
Send normal packet to VF1 and check forwarded packet not include E-tag
- Case: E-tag strip
Description: check that E-tag strip feature can work as expected
Enable E-tag strip on PF
testpmd> E-tag set filter add e-tag-id 1000 dst-pool 0 port 0
testpmd> E-tag set stripping on port 0
Send 802.1BR packet to VF and check forwarded packet without E-tag
Disable E-tag strip on PF
testpmd> E-tag set stripping off port 0
Send 802.1BR packet and check forwarded packet with E-tag.
> -----Original Message-----
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Wenzhuo Lu
> Sent: Friday, March 04, 2016 10:35 AM
> To: dev at dpdk.org
> Subject: [dpdk-dev] [PATCH v7 0/5] Support VxLAN & NVGRE checksum off-
> load on X550
>
> This patch set add the VxLAN & NVGRE checksum off-load support.
> Both RX and TX checksum off-load can be used for VxLAN & NVGRE.
> And the VxLAN port can be set, it's implemented in this patch
> set either.
>
> v2:
> - Update release note.
>
> v3:
> - Update RX/TX offload capability.
> - Reuse PKT_RX_EIP_CKSUM_BAD but not add a new one.
> - Correct the tunnel len for TX, and remove the useless out_l2_len.
> - Don't set the tunnel type for TX, and remove the unused ol_flag_nvgre.
>
> v4:
> - Fix the issue that not setting the MAC length correctly.
>
> v5:
> - Change the behavior of VxLAN port add/del to make it align with i40e.
>
> v6:
> - Fix x86_64-native-linuxapp-gcc-shared compile error.
>
> v7:
> - Change the return value from hardcode to macro.
>
> Wenzhuo Lu (5):
> lib/librte_ether: change function name of tunnel port config
> i40e: rename the tunnel port config functions
> ixgbe: support UDP tunnel port config
> ixgbe: support VxLAN & NVGRE RX checksum off-load
> ixgbe: support VxLAN & NVGRE TX checksum off-load
>
> app/test-pmd/cmdline.c | 6 +-
> doc/guides/rel_notes/release_16_04.rst | 9 +++
> drivers/net/i40e/i40e_ethdev.c | 22 +++---
> drivers/net/ixgbe/ixgbe_ethdev.c | 131
> +++++++++++++++++++++++++++++++++
> drivers/net/ixgbe/ixgbe_rxtx.c | 67 ++++++++++++++---
> drivers/net/ixgbe/ixgbe_rxtx.h | 6 +-
> examples/tep_termination/vxlan_setup.c | 2 +-
> lib/librte_ether/rte_ethdev.c | 45 +++++++++++
> lib/librte_ether/rte_ethdev.h | 19 +++++
> lib/librte_ether/rte_ether_version.map | 2 +
> lib/librte_mbuf/rte_mbuf.c | 2 +-
> lib/librte_mbuf/rte_mbuf.h | 2 +-
> 12 files changed, 285 insertions(+), 28 deletions(-)
>
> Acked-by: Konstantin Ananyev <konstantin.ananyev at intel.com>
>
> --
> 1.9.3
More information about the dev
mailing list