Maybe this question belongs on another list, but I haven't convinced
myself I know enough to follow tech-kern, so here goes....

My /var/log/messages is filled with lines like:

  Jun  6 01:49:03 oak /netbsd: ex0: uplistptr was 0

Is this considered good form?  

As I understand it, the message is harmless, a little note from the driver
writer to the driver writer that the uplistptr turned out to be zero. 
Nothing to worry my pretty little head about.  Except, except, I don't
like noise in my message logs. :-(

I know: "If thine line offends thee, cut it out."  I could expunge the
noise emitter, recompile my kernel, and be done.  

Supposing I were to do more than merely redact, though.  Supposing I were
to try to improve the code and its s/n ratio, what would be the best aim? 
Is there a general theory for what to do with driver statistics?  Some
framework to tie into?  Is there nothing better for it than splatf() to
the log?  

As far as I can tell, the uplistptr message indicates not a logic error
but a runtime condition, some resource exhaustion, perhaps resulting in
suboptimal network behavior.  If so, it would be better to count the silly
things and pony them up for an snmp query or something.  I'm hoping
someone is willing to explain what that "something" is.  

Because I expect my delete-this-line pr wouldn't be accepted.... 

Thank you.

--jkl