Ken,
I think SCH_CTLFLAGS was used a long time ago to control a feature called "proactive memory reclaimation". Somewhere in the V5.4 timeframe (nearly 20 years ago). The idea was to try to identify processes which weren't doing much and steal memory back from them to give to active processes.
The theory was that for a particular workload, there would inevitably be some number of pagefaults, but you could load them up on processes that didn't matter, thereby speeding up ones that did. For example, ERRFMT would wake up every so often and format a record for the error log file. You didn't care if it took several thousand pagefaults (and several seconds) to do it, especially if it means your interactive users didn't suffer those pagefaults.
It made a significant difference to the perceived performance of a VAX with 4MB of memory.
Back then 1MB of RAM cost several thousand dollars, so it made sense to do everything you could to preserve as much as possible.
I suspect the code paths are still there, but cause more trouble than they're worth in today's memory rich environments. For example, one trick used was to pre-emptively swap out a new process. If it was active it would quickly swap itself back in. If it wasn't active it would stay swapped out.
The parameter is dynamic, so you can play with it if you like. I'd suggest leaving it where ever the upgrade set it.
(installation and release documentation is too big already, if they really noted every single piece of trivia that changed, even fewer people would bother to read them)
On the other hand, well done in spotting it! Not many system managers are that vigilant ;-)
A crucible of informative mistakes
