LGTM, please commit ASAP and I'll restart the weekly snapshot build.

In D58230#1341164, @imp wrote:

So this is OK. A better thing would be to just drop the CAM lock. That should be safe. However, it's going to be super rare we can't allocate here. Can you commit this as is, but file a bug and assign it to me. The bug should be 'investigate dropping CAM lock for AC_GETDEV_CHANGED in nda' and I'll not forget about it that way.

I see these errors under low loads, and not always as "1 [mu]secs have passed":

I'm not convinced by the rationale here -- it seems to me that the stronger argument is "processing tx completions frees up space on the tx ring, which we might want when we respond to packets which we received". But the change itself seems reasonable.

The commit message refers to critical sections, but in fact counter_{enter,exit} are no-ops everywhere except powerpc, and somehow I doubt ENA will ever run on powerpcs. So the win here isn't about critical sections; it's just about reducing the number of atomic operations.

@imp In 824530a5c4fd59427cda0eae0a4ac85212814958 you wrote nda: Assume all cases in ndaasync can sleep and changed a different M_NOWAIT to M_WAITOK, but WITNESS complained about this code path:

Commit message edited for correctness and clarity.

@ivy Can you add the tweaks from ziaee/markj and commit this? I don't think we need to add the "how to upgrade from 14.4 pkgbase" instructions; that's an unsupported path and would bloat the document too much (and risk confusing users).

LGTM.