Robert Elz writes:

 > How certain are we that this is really a SRM bug?    

It's all scuttlebutt and rumour at this point :)

I will be trying the "downgrade to 4.9, then upgrade to 5.5" SRM
routine one evening this week.  I will attempt to confirm what others
have reported: that this makes the problem go away.

 > One workaround that has been mentioned on the list is to explicitly set the
 > env vars just before a power off, that seems to cause them to be saved.

That is
 a) no use if you don't get back to SRM after "halt"ing netbsd
 b) no more or less work than setting them after power-on
 c) nonetheless, an interesting datapoint, if not a workaround

--
Dan.