Jump to content

Express Data Path

From Wikipedia, the free encyclopedia
This is an old revision of this page, as edited by Teknoraver (talk | contribs) at 23:08, 17 May 2019 (add graph with flow paths). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.
XDP
Original author(s)Brenden Blanco, Tom Herbert
Initial release2016
Written inC
Operating systemLinux
TypePacket filtering
LicenseGPL
Packet flow paths in the Linux kernel. XDP bypasses the networking stack and memory allocation for packet metadata.

The program is allowed to edit the packet data and, after the eBPF program returns, an action code determines what to do with the packet:

XDP (eXpress Data Path) is an eBPF based high performance data path merged in the Linux kernel since version 4.8[1].

The idea behind XDP is to add an early hook in the RX path of the kernel, and let an user supplied eBPF program decide the fate of the packet. The hook is placed in the NIC driver just after the interrupt processing, and before any memory allocation needed by the network stack itself, because memory allocation can be an expensive operation. Due to this design, XDP can drop 26 millions of packets per second with commodity hardware[2].

The eBPF program must pass a preverifier test[3] before being loaded, to avoid executing malicious code in kernel space. The preverifier checks that the program contains no out-of-bounds accesses, loops or global variables.

  • let the packet continue through the network stack
  • silently drop the packet
  • bounce the packet back from the same NIC it arrived on
  • send the packet to an user space socket via the AF_XDP address family

XDP requires support in the NIC driver but, as not all drivers supports it, it can fallback to a generic implementation, which performs the eBPF processing in the network stack, though with slower performances[4].

XDP has infrastructure to offload the eBPF program to a NIC card which supports it, reducing the CPU load. At the time only Netronome cards supports it[5], with Intel and Mellanox working on it[6].

AF_XDP

Along with XDP, a new address family entered in the Linux kernel starting 4.18[7]. AF_XDP (formerly known as AF_PACKETv4) is a raw socket optimized for high performance packet processing and allows zero-copy between kernel and applications. As the socket can be used for both receiving and transmitting, it allows to write high performance network applications purely in user space[8].

References

  1. ^ "[GIT] Networking - David Miller". lore.kernel.org. Retrieved 2019-05-14.
  2. ^ Høiland-Jørgensen, Toke (2019-05-03), Source text and experimental data for our paper describing XDP: tohojo/xdp-paper, retrieved 2019-05-14
  3. ^ "A thorough introduction to eBPF [LWN.net]". lwn.net. Retrieved 2019-05-14.
  4. ^ "net: Generic XDP". www.mail-archive.com. Retrieved 2019-05-14.
  5. ^ "BPF, eBPF, XDP and Bpfilter… What are these things and what do they mean for the enterprise? - Netronome". www.netronome.com. Retrieved 2019-05-14.
  6. ^ "XDP acceleration using NIC metadata" (PDF). {{cite web}}: Cite has empty unknown parameter: |dead-url= (help)
  7. ^ "kernel/git/torvalds/linux.git - Linux kernel source tree". git.kernel.org. Retrieved 2019-05-16.
  8. ^ "Accelerating networking with AF_XDP [LWN.net]". lwn.net. Retrieved 2019-05-16.
  • [3] XDP walkthrough at FOSDEM 2017 by Daniel Borkmann, Cilium
  • [6] Open-sourcing Katran, a scalable network load balancer, Facebook
Retrieved from "https://en.wikipedia.org/w/index.php?title=Express_Data_Path&oldid=897572199"