[LispM-Hackers] Do we need a read barrier to support e3 GC?
Sat Apr 13 08:22:00 2002
If you're handling forwarding pointers and otherwise simulating VM in
software already, then no. But if you're doing that, and in my message of
a couple of weeks ago I argued that you need to, then you don't need to do
very much to the Intel MMU at all.
On Sat, 13 Apr 2002, John Morrison wrote:
> As near as I can tell, unless I muck about with non-privileged segments
> (which I am not currently doing -- I currently have only one segment),
> the standard intel architecture MMU does not support read-denied pages
> (although I can probably fake it by pretending the page is not mapped
> and doing more work). Does anybody knowledgeable about the E2 GC know
> whether the GC depends upon a read barrier?
> ==== John Morrison
> ==== MAK Technologies Inc.
> ==== 185 Alewife Brook Parkway, Cambridge, MA 02138
> ==== http://www.mak.com/
> ==== vox:617-876-8085 x115
> ==== fax:617-876-9208
> ==== firstname.lastname@example.org