[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
your lisp and my 3min
They just met. Historical landmark! Lisp on the 3min meets Mach 3.0,
PS: I am still trying to embarass it. Gabriel does not cut it, either.
PPS: I think the problem is as follows: we need the following sequence of
events to happen:
- new code is created in vm_allocated memory (newspace?)
- that memory gets paged out
- you try to execute that code, which causes
- that page is paged back in again and
- it lands on a phys page that still has entries in the cache
So you see how unlikely this is: a) newspace will be th elast thing we page out,
after just about all of the core and other dirty pages and b) it is not easy
to provoke the last event either, especially on an emulated Unix where we execute
more dynamically placed code (Unix is pageable just as well).
Ideas/dirty tricks highly appreciated, at this point.