Can't this be corrected somewere (is a patch posible)? Or is PHYS_SIZE hardcoded everywhere. That would make patching harder, but not imposible. I never made an VM, but I can imagine we could change the value of PHYS_SIZE everywhere. I'm not behind my beos computer, so I can't test this, but if the kernel is 1MB, the chance of a unrelated int32 with exacly the same value is about 25% By simply using sed, we sould make a kernel with 25% chance of still functioning as it should, right? I'm probably very wrong, but it's a nice try..... Leon Timmermans ----- Original Message ----- From: "François Revol" <revol@xxxxxxx> To: <openbeos@xxxxxxxxxxxxx> Sent: Monday, June 02, 2003 12:55 PM Subject: [openbeos] Re: kernel memory limit > En réponse à Bruno van Dooren <bruno_van_dooren@xxxxxxxxxxx>: > > > > > > No, NewOS doesn't map all available memory - that's a BeOS only > > > feature, and is responsible for this hard limit; neither NewOS nor > > > OpenBeOS have it. > I stand corrected. > > > > > what do you mean by this? > > i thought memory was allocated on a by-need basis, and that for example > > all > > PCI memory is mapped in an unused region > xxx GB. > > > > i think i do not understand what you mean by mapping. > > > > kind regards, > > At boot, the BeOS kernel does the equivalent of: > > map_physical_memory("physical_ram", 0, PHYS_SIZE, ...); > > So it "wastes" its 2G virtual space by creating an area that is as big > as the physical RAM... > > Just check yourself: > > listarea 1 | head -25 > > > François. > > >