Welcome, Guest. Please login or register.

Author Topic: Open Source 3.1 compatible OS?  (Read 3834 times)

Description:

0 Members and 1 Guest are viewing this topic.

Offline Piru

  • \' union select name,pwd--
  • Hero Member
  • *****
  • Join Date: Aug 2002
  • Posts: 6946
    • Show only replies by Piru
    • http://www.iki.fi/sintonen/
Re: Open Source 3.1 compatible OS?
« Reply #14 from previous page: September 23, 2006, 12:47:52 PM »
@Tomas

No you really can't have any effective memory protection if you intend to use AmigaOS. It's not just applications, it's the OS itself. You'd need to totally rewrite the OS and the apps. Then it wouldn't be AmigaOS anymore, though.
 

Offline Fats

  • Hero Member
  • *****
  • Join Date: Mar 2002
  • Posts: 672
    • Show only replies by Fats
Re: Open Source 3.1 compatible OS?
« Reply #15 on: September 23, 2006, 02:35:27 PM »
Quote

Piru wrote:
@Tomas

No you really can't have any effective memory protection if you intend to use AmigaOS. It's not just applications, it's the OS itself.


I don't fully agree with this statement. Sure you can't have UNIX style MP where every process lives in it's own adress space. But I think it should be possible to have one adress space but make some of your memory read-only to other tasks. At least this should make it possible to make the core OS uncrashable. Of course programs not using this feature to protect their own memory can still be crashed by other programs that have a wild pointer.
I agree though this will be a major undertaking especially to get it right and performant.
Trust me...                                              I know what I\'m doing
 

Offline Piru

  • \' union select name,pwd--
  • Hero Member
  • *****
  • Join Date: Aug 2002
  • Posts: 6946
    • Show only replies by Piru
    • http://www.iki.fi/sintonen/
Re: Open Source 3.1 compatible OS?
« Reply #16 on: September 23, 2006, 02:48:50 PM »
@Fats
Quote
But I think it should be possible to have one adress space but make some of your memory read-only to other tasks.

This is possible. This is not the problem.

Quote
At least this should make it possible to make the core OS uncrashable.

No it doesn't. The core OS depends on being able to poke things freely.

Also, all apps pass their memory to the OS, and the OS assumes it can poke it. Apps pass memory to each other and assume they can access it freely. It just doesn't work, sorry.

The best semi-protection can be achieved from mmu.library MuLink, but even that only protects the given code segments against alteration. MuProctectModules can protect RAM loaded modules, but again, this is only protecting the code from alteration.