msaladna/optska

uptime fix for kano

This is a fix to the problem with GetTicks() under Windows which returns a DWORD only allocating 4 bytes of storage, so if your uptime is greater than 2^32/1000 seconds (ticks are in millisecs) then it experiences a rollover. I guess Windows developers never considered a non WinXP/2k/NT box to have any uptime greater than 49.7 days.

At any rate this returns the seconds elapsed since the last reboot through the performance counter and there are no sanity checks as to whether you have the performance counter dll present (WinXP/NT/2k) or not... it's on the to-do list, as is future improvements to give XiRCON functionality to obtain other counter information.

Call the proc under XiRCON as 'opts' and from there, if using kano, you can run it by 'since', e.g.: set uptime [since [opts]] Which will return something along the lines of: [ka] Tcl: 2mn 4d 15h 51m 1s

This fix returns a 64 bit wide int (LONGLONG) that'll guarantee accurate non-rollover uptime reports for up to 2^64 seconds, or 584942417355.07203247 years... I don't think I'll need to write another patch any time soon.

Do not call opts for any high-resolution benchmarking such as the time elapsed during an interval where the resolution needs to be high. The value returned is in seconds and definitely not milliseconds (as stated before). Also too because of this property, you don't need to divide the value by 1000 to get seconds, that'd be what, kiloseconds? Well something odd.

There's a slight start-up cost though when the query is first instantiated when calling opts for the first time, but remains resident for subsequent calls-- that's why kano will hang for a second or two on /uptime.

The opts addon does give some interesting ideas to add some more nifty functionality such as the ability to add and query performance counters, but that'll come around when I have time.

To install, just extract the opts.dll and opts.ka to the addons directory of kano then pop open kano.tcl in your favorite text editory (XeD works here) and find:

if {[osuptime] > [get_cookie winuptime 0]} {set_cookie winuptime \
        [osuptime]}

within the minute hook (approximately line 9212), change to:

if {[opts] > [get_cookie winuptime 0]} {set_cookie winuptime \
        [opts]}

that should fix the best os uptime and /reload - done.

• Matt Saladna [email protected]