Hi, This week I got a prototype patch running, along the ideas discussed last week. Essentially, we keep the BRIDGE_LOCK for any add/delete of interfaces or rt entries, but use CK_LIST and epoch in the data path. This is a relatively straightforward change, and currently passes the regression tests (WITNESS/INVARIANTS enabled). I’ve also run traffic through it without issue. My current test setup fails to generate sufficient packets to truly stress the code. I’m hoping to get access to a more suitable setup next week so I can usefully measure the improvement. The prototype patch is also running on my home gateway right now. So far so good! Assuming no major issues come up in the next week or two I hope to post the patch for initial review in the near future. Best regards, KristofReceived on Sat Mar 28 2020 - 15:35:54 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:41:23 UTC