[HECnet] How long has your 20 been up?

[Thomas DeBellis]

Check it out:

    @_uptiME.EXE.2_
    TOMMYT up for 1 Year, 2 Days, 16 Hours, 36 Minutes, 26 Seconds and
    582 Milliseconds
    @_timET2.EXE.4 _
    Tops-20 was booted on Saturday, January 16, 2021 8:46:38PM-EST.
    It will crash with an UP2LNG BUGHLT on Friday, February 18, 2022
    1:08:57PM-EST.
    @

UPTIME and TIMET2 are two programs that I wrote several years; ago, 
UPTIME in 2017 and TIMET2 in 2006.  I wrote them because I found the 
hours format given by SYSTAT SYSTEM and INFORMATION MONITOR-STATISTICS 
to be a little opaque and they didn't have the information I wanted, viz:

    @_sysTAT sySTEM_
      Tue 18-Jan-2022 13:28:45  Up *8800:42:07!!!!!!!!*
      4+7 Jobs   Load av   0.01   0.02   0.02

      No operator in attendance

    @_iNFORMATION (ABOUT) monITOR-STATISTICS_
      Up *8800:42:10!!!!!!!!*
      Idle 99%  Waiting 0%  Sched ovh 0%  Pager traps 0%
      Swap reads 28049 Writes 385176  File reads 4643699 Writes 3677925
      8092 Pages of user memory
      4259771001 Term wakeups  1255 Term interrupts
      NBAL av   0.32  NRUN av   0.02
      Runtime of jobs on sched queues 0-6 (sec)
         143964    16409    14494    1012    1756 4846    4215
    @

I wrote TIMET2 because I needed to know when to set a shutdown. If you 
don't do that and you hit the uptime limit, then the machine simple 
crashes with an UP2LNG BUGHLT. This happens because the uptime 
millisecond counter is a signed 36 bit integer, which will roll over and 
become negative after 1 Year, 4 Weeks, 5 Days, 16 Hours, 22 Minutes, 18 
Seconds and 367 Milliseconds.  You would think the code would self-set a 
'graceful' shutdown, but it doesn't...  Maybe I'll do that one of these 
days; it's easy enough.

Since the PDP-10 does not have unsigned compares, fixing this would 
involve changing a rather large number of comparisons. Assuming you kept 
the storage at a single word, going unsigned would get you out to 2 
Years, 9 Weeks, 4 Days, 8 Hours, 44 Minutes, 36 Seconds and 735 
Milliseconds.  That's probably still not that great given today's hardware.

Going to an unsigned double word format would obviously get you a far 
larger uptime.  However, the current arithmetic that I use can 'only'  
handle a maximum uptime of 34,359,738,367,999 of milliseconds.  This 
works out to be 1092 Years, 27 Weeks, 5 Days, 3 Hours, 46 Minutes, 7 
Seconds and 999 Milliseconds.  Since the time of day counter expires in 
7-Aug-2576 7:59:59 PM EDT, I used a faster instruction...  Otherwise, 
going over a millennium between reboots seemed reasonable unless you 
find yourself in a significantly sub-luminal spacecraft going someplace 
inter-galactic...

I know that XKL fixed at least part of the uptime problem, but I don't 
remember what that limit is.  What are the limits for other systems?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.sonic.net/mailman/private/hecnet-list/attachments/20220118/6eab747d/attachment.htm>