[openbeos] Re: kernel memory limit

  • From: Michael Noisternig <michael.noisternig@xxxxxxxxx>
  • To: openbeos@xxxxxxxxxxxxx
  • Date: Mon, 02 Jun 2003 15:24:29 +0200

Bruno van Dooren wrote:

OK i see what the problem is, but why would they do that?

Because it is easier. You don't have to map/unmap physical pages as needed, track their mappings...


isn't this just a waste of half their address space?

Yes.



basically this means that they can address byte 0 also at byte 0 + 1G? couldn't they just physical addresses?

*All* memory addresses are translated to physical addresses (on x86, and paging turned on). The idea is that behind a (virtual) memory address there can be a completely different physical address and this mapping can be changed on the fly. So you can do all that crucial stuff like presenting continous memory to the programmer (starting at virtual address 0), paging, swapping...


So if you address virtual byte 0 + 1G, and that address' page is mapped to the physical page starting at address 0, then you access physical byte 0.
If you address virtual byte 0 then the same translation happens and you end at the physical byte that is determined by the page mapping.



kind regards, Bruno.

Leon Timmermans wrote: > Can't this be corrected somewere (is a patch posible)? > Or is PHYS_SIZE hardcoded everywhere.

Yes, it is hardcoded by the amount of RAM you have in your system.
But no, you can change that value by removing some RAM modules from your system.
;->


Michael Noisternig


Other related posts: