Wake (resume) regression?

Has anything changed in the past two months that might cause wake to 
become unreliable?

nvidia-driver-470, if that's relevant. Built from source.

HP ZBook 17 G2.

IIRC the unreliability became noticeable not long after the bump to 
1500023. That was around the end of August (with pkgbase).

Previously, I'd say that wake was 99% reliable. Now, wake more often 
fails than succeeds. The greatest frustration: it's not consistently 
reproducible. I have not yet identified the ingredients that precede a 
failure, and critically:

- when failure occurs, there's nothing (no flicker of the display, no 
HDD activity).

In other words, no sense of what might be wrong at the time.

<https://gist.github.com/grahamperrin/86359cb7c6d1316d75f340fb305484ed> 
lists all recent boot environments.

<https://pastebin.com/raw/xBwqeBxr> is the subset that remains after 
routine weeding of environments. This includes:

1500021-006-base          -      -          2.81G 2024-07-25 02:31
1500022-001-base-ports    -      -          17.0M 2024-07-29 17:22
1500022-002-base-ports    -      -          3.09G 2024-07-30 06:50
1500022-004-base          -      -          2.81G 2024-07-30 14:45
1500023-001-base          -      -          2.89G 2024-07-31 03:14
1500023-002-base          -      -          2.80G 2024-08-01 05:26

----

I'm aware of this report (not involving NVIDIA):

<https://github.com/freebsd/drm-kmod/issues/312>
<https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=280923>

Blind carbon copy: the reporter.