summaryrefslogtreecommitdiffstats
path: root/src/vnet/bonding
AgeCommit message (Collapse)AuthorFilesLines
2018-05-27bond-input performance optimizationDamjan Marion1-291/+261
Old code ~25 clocks/packet, new ~10. Change-Id: I202cd6cbafb1ab2296939634d674f7ffd28253fc Signed-off-by: Damjan Marion <damarion@cisco.com>
2018-05-25bond: performance harvestingSteven4-133/+183
- hash is great. But it is a bit too slow for the DP. Use direct array indexing to quickly retrieve the slave interface. - the algorithm used by flow hash is great. But it is a bit too slow for the DP. Use l2_hash_hash() extracted from lb_hash.h which ECMP is using. It makes use of intrinsic crc32 instruction set. - shortcut modulo arithmetic when the operand is 2**x (where x up to 4) to avoid division instruction. - special case for link count == 1 in bond_tx_fn() - use clib_mem_unaligned to access data for the packet to avoid alignment error - Fix some typos for packet tracing. Change-Id: I8eae3ad497061c5473aa675ba894ee0211120d25 Signed-off-by: Steven <sluong@cisco.com>
2018-04-24lacp: deleting the bond subinterface may cause lacp to lose the partner ↵Steven3-23/+2
[VPP-1251] Problem: When the bond subinterface is removed, it was observed that we lost the lacp partner. Show hardware shows rx counter goes up, but show interface does not for the slave interfaces. Cause: We reset the interface promiscuous mode when the bond subinterface is deleted. This causes dpdk not to accept any packet. Leave the interface in promiscuous mode fixes the problem. Other fixes: There are few places we use hw_if_index as if they are sw_if_index. But they don't necessarily have the same value. As soon as a subinterface is created, they start to diverge. The fix is to use the correct API for the hw_if_index and sw_if_index. Change-Id: I1e6b8bca0a4aae396d217a141271cbf968500c91 Signed-off-by: Steven <sluong@cisco.com> (cherry picked from commit 42c6599bf3057a7e8f4f00f5b6a9dd72af48d283)
2018-04-13bond: ping fails between l2 BD [VPP-1238]Steven2-15/+76
In dpdk based bonding, when the bond interface is configured for l2, it automatically sets the bond interface to promiscuous mode and sets rx redirect to ethernet-input. This allows traffic to be bridged to non compute node facing interface when it is received from the compute node interface. For native vpp bonding, we need to do similar things. When the bond interface is configured for l2, we set the slave interfaces to promiscuous mode and set rx redirect to ethernet-input because dpdk does not know anything about the bond interface. Likewise, when a new interface is enslaved, we also need to do the same thing if the bond interface has already been configured for l2. Change-Id: I7e168008e8a4221be74929b2a20e6db0ce8f3110 Signed-off-by: Steven <sluong@cisco.com>
2018-04-12bond: 1 packet/frame == bad performance [VPP-1236]Steven1-6/+8
While https://gerrit.fd.io/r/#/c/11316/ took care of 1 packet/frame for most of the bonding modes, it missed the broadcast mode. This patch is to fix the 1 packet/frame for the broadcast mode. Change-Id: Iac48a2977c7f702f341479cc712a6448090dbc60 Signed-off-by: Steven <sluong@cisco.com>
2018-03-30bond: show trace causes a crash if the interface is deletedSteven2-9/+30
For the debug image, if the interface is removed and the trace was collected prior to the interface delete, show trace may cause a crash. This is because vnet_get_sw_interface_name and vnet_get_sup_hw_interface are not safe if the interface is deleted. The fix is to use format_vnet_sw_if_index_name if all we need is to get the interface name in the trace to display. It would show "DELETED" which is better than a crash. Change-Id: I912402d3e71592ece9f49d36c8a6b7af97f3b69e Signed-off-by: Steven <sluong@cisco.com>
2018-03-28bond: cli renameSteven1-11/+10
rename "enslave interface <slave> to <BondEthernetx>" to "bond add <BondEthernetx> <slave> "detach interface <slave>" to "bond del <slave>" Change-Id: I1bf8f017517b1f8a823127c7efedd3766e45cd5b Signed-off-by: Steven <sluong@cisco.com>
2018-03-27bond: coverity woesSteven2-27/+32
coverity complains about statements in function A function A { x % vec_len (y) } because vec_len (y) is a macro and may return 0 if the pointer y is null. But coverity fails to realize the same statement vec_len (y) was already invoked and checked in the caller of function A and punt if vec_len (y) is 0. We can fix the coverity warning and shave off a few cpu cycles by caching the result of vec_len (y) and pass it around to avoid calling vec_len (y) again in multiple places. Change-Id: I095166373abd3af3859646f860ee97c52f12fb50 Signed-off-by: Steven <sluong@cisco.com>
2018-03-22bond: performance enhancementSteven3-155/+220
We were only puting one packet per frame to the output node. Change to buffer multiple packets per frame. Performance is now on top of dpdk-based bonding. Put a spinlock in the tx thread in case the rug is pulled under us. Change-Id: Ifda5af086a984a7301972cd6c8e428217f676a95 Signed-off-by: Steven <sluong@cisco.com>
2018-03-21bond: Add bonding driver and LACP protocolSteven6-0/+2767
Add bonding driver to support creation of bond interface which composes of multiple slave interfaces. The slave interfaces could be physical interfaces, or just any virtual interfaces. For example, memif interfaces. The syntax to create a bond interface is create bond mode <lacp | xor | acitve-backup | broadcast | round-robin> To enslave an interface to the bond interface, enslave interface TenGigabitEthernet6/0/0 to BondEthernet0 Please see src/plugins/lacp/lacp_doc.md for more examples and additional options. LACP is a control plane protocol which manages and monitors the status of the slave interfaces. The protocol is part of 802.3ad standard. This patch implements LACPv1. LACPv2 is not supported. To enable LACP on the bond interface, specify "mode lacp" when the bond interface is created. The syntax to enslave a slave interface is the same as other bonding modes. Change-Id: I06581d3b87635972f9f0e1ec50b67560fc13e26c Signed-off-by: Steven <sluong@cisco.com>