so this time, 1.5C crashes with a stack trace like this:

panic
amap_wipeout(df3756bc,df478144,df478144,0,df3b1ec8) at amap_wipeout+0x30
amap_unref(df478144,0) at amap_unref+0x1a
uvm_unmap_detach(df4782c4,0,df203228,df1fee10,df4782c40 at uvm_unmap_detach+0x21
uvm_unmap(df203228,0,bfbfe000,df203228,df3b1f28) at uvm_unmap+0xa3
uvm_deallocate(df203228,0,bfbfe000) at uvm_deallocate+0x38
exit1(df1fee10,100,df3b1fa8,c022bf0f,df1fee10) at exit1+0x13f
sys_exit(df1fee10,df3b1f88,df3b1f80,480faefc,ffffffff) at sys_exit+0x14
syscall() at syscall+0x1e7
--- syscall (number 1) ---

could this be a "bad RAM" thing ?  What makes me wonder is that the last
thing I see while doing a kernel build (which was happening at the time)
was:

cc: Internal compiler error: program cc1 got fatal signal 6
*** Error code 1

and I've had a lot of trouble with gcc coring of late.

Darren