Every boring problem found in eBPF (2022)
161 points by udev4096 6 days ago | 11 comments
  • magicalhippo a day ago |
    • tptacek a day ago |
      This is so weird to me. It's not an interoperable standard. It isn't even interoperable on Linux, the one OS where it's popular.
      • kimixa a day ago |
        If you wanted it to become an interoperable standard, that's the obvious step, right?
      • maxmcd a day ago |
        • tptacek a day ago |
          Right, for offload, but XDP programs also depend on helper definitions, which themselves have not been consistent between versions of the Linux kernel.
          • stefan_ a day ago |
            And some (most?) of the helpers end up being "read memory at X", which has some obvious problems with offloading.
            • tptacek a day ago |
              I mean, there really is working XDP offload (Netronome, right?) so it can be made to work, but this spec doesn't define the hardest part of interoperability.
              • _flux 21 hours ago |
                It probably still helps to have a standardised baseline. Then the vendors can just document deviations from it, and the deviations are probably quite small, not necessating new compilers for example (just guessing here ;)).
  • Retr0id a day ago |
    By the way, this article is published as part of the tmp.0ut zine, and the CFP for the next issue is currently open: https://tmpout.sh/blog/vol4-cfp.html
  • humanfromearth9 a day ago |
    Wow that's a tl;dr for me. Not only because of the length, also because of format. Not very readable on a smartphone...
  • remram 21 hours ago |
    I'm usually against separate "mobile versions" of websites but wow, this needs one.