Simon Burge wrote:
> First traffic on port-sbmips I can remember in a long time!
>
> Garrett D'Amore wrote:
>
>   
>> This diff is the same, except I've converted the rtc functions to use
>> the clock_ymdhms variants of the TODR entry points, saving some more
>> redundant code and inheriting some additional "checks."
>>     
>
> The patch seems to work.  Setting the date to an incorrect time, and
> rebooting back to the same incorrect offset worked.  Setting back to
> "normal" with ntpdate and rebooting again also worked.  What other sort
> of tests should be done?
>   

Sounds good.  Can I commit it then?
>   
>> (Also, the fewer functions using the old volatile struct timval *, the
>> better, as I intend to change the signature of that function soon, to
>> use either struct bintime or struct timespec_t.)
>>     
>
> Please discuss this on tech-kern when the time comes.  I'm not sure if
> we should be using timespecs or bintime as the "base" system time.  A
> long time ago there was talk about switching to timespecs, but that
> predated FreeBSD's bintime and timecounters.
>   

Absolutely.  I have no idea what is best.  But I definitely want to get
away from volatile struct timeval * as the main argument.    Frankly,
timespec_t is easier _for me_, because its a much more trivial
conversion (cheaper math).  I'm not so sure about bintime, which looks a
bit less "obvious" to me.  But we'll chat about it on tech-kern later. 
Or I'll drop the ball and let someone else carry the torch for a change.
:-)  I've made so many tc/todr changes I have a hard time keeping them
straight in my head.

    -- Garrett