I think that there still is a problem (please help me define it!) with
memory management - I think that Jason's commit didn't have anything to
do with it - backing it out hasn't changed much for me:

kernel: page fault trap, code=0
Stopped in make at   uvm_pageactivate+0x41: movl %eax,0(%edx)

%eax=%edx=0xf01e13a9

db> t
uvm_pageactivate(f7efabb0) at uvm_pageactivate+0x41
uvm_fault(f801d170,82b7000,0,1,0) at uvm_fault+0xa86
trap() at trap+0x40b
--- trap (number 6) ---
0x480cf7d2:

current/i386 of Tue Jul 13 13:47:15 1999 BST

Have tools - need know-how..

Cheers,

Patrick