From: Darren Tucker Subject: Re: em(4) TX interrupt mitigation To: dtucker@dtucker.net Cc: tech@openbsd.org Date: Wed, 21 May 2025 12:29:01 +1000 On Mon, 19 May 2025 at 18:52, Darren Tucker wrote: > TL;DR: if you use em(4), particularly on a low-power device such as a > pcengines APU2, please try this diff. > After adding some instrumentation to my "baseline" kernel and rebuilding it, I saw the difference between baseline and my diff more or less vanish. I repeated the test with the original baseline kernel, which was from snapshot from a few days ago (OpenBSD 7.7-current (GENERIC) #657: Sun May 18 01:28:04 MDT 2025) and saw the difference again. It would seem that slower speed and higher CPU usage in the baseline was actually something that was broken in that snapshot and fixed the following day. Apologies for the noise. If you're running with my diff it should work fine, but you probably won't see the gains I claimed :-( -- Darren Tucker (dtucker at dtucker.net) GPG key 11EAA6FA / A86E 3E07 5B19 5880 E860 37F4 9357 ECEF 11EA A6FA Good judgement comes with experience. Unfortunately, the experience usually comes from bad judgement.