Re: Kernel without SUN support
David Brownlee <[email protected]> Sun, 12 Nov 2023 12:03:30 +0000
| Newsgroups | gmane.os.netbsd.ports.amiga |
|---|---|
| Message-ID | <CAGN_6pYSanjOYNVkYvX9RcR_b59v2GvoNezBUH9NO6_6SeSVyQ@mail.gmail.com> |
On Sun, 12 Nov 2023 at 09:16, Michael van Elst <[email protected]> wrote: > > On Sat, Nov 11, 2023 at 02:09:19PM +0100, Harald Meinzer wrote: > > Hi out there, > > > > is there somebody out, who could compile a Kernel without this stupid 384 MB > > Ram per Task limit? This is completely obsolete. > > The code supports 512MB address space and leaves 480MB for a user process, > a small test program can malloc 438MB RAM. Just checking - is that with the default limits and COMPAT_SUNOS enabled? > You need to rewrite the pmap code (and other things) to exceed that, 68k > supports a 4GB user address space, so that's not impossible. > > On the other hand, there is a rising number of programs that require more > than 4GB RAM to build, the success might not last long. Nothing lasts forever, but that doesn't mean removing a software limitation to allow using more of the hardware's available address space isn't worth doing :) Maybe we can ping the other NetBSD/m68k mail lists and put together a little bounty for someone interested in reworking the pmap? I'll put up $50 to start a supply of $beverageOfChoice towards that end :) David