smith@NRL-AIC.arpa (Russ Smith) (10/21/86)
We recently installed a couple memory controllers in our VAX 780 running fairly vanilla 4.2BSD. In so doing, we put 16 Megabytes on one controller, 8 on the other (this is a valid configuration and not the cause of the following problem). On booting UNIX, the machine gives us: . . . real mem = 25161728 panic: sys pt too small and crashes back to the boot level. This panic message originates in the .../sys/machine/machdep.c file right after a check of the form: if (((int)(ecmap+1))&~0x80000000) > SYSPTSIZE*NBPG) Question: Has anyone run into and fixed this problem? I won't go into the interesting comments that are above the "define" of SYSPTSIZE in the same directory's vmparam.h... We'd really like to be able to use all 24 Megs...any help would be MUCH appreciated. Russ <Smith@nrl-aic> JAYCOR Navy Center for Applied Research in Artificial Intelligence (whew!)
smith@NRL-AIC.arpa (Russ Smith) (10/22/86)
Sigh. We are running 16 megabytes currently. The problem unfortunately is not the "8 megabyte wall" but some other value over and above that which is not set right. Sorry for not making this clear in my initial note (please note that our system SEES and "locore.s loops" through all 25 megabytes, it craps out later during other non-8-meg based stuff.) Thanks, and keep those responses coming! Russ <Smith@nrl-aic> JAYCOR Navy Center for Applied Research in Artificial Intelligence (whew!)