Re: [MiNT] memory protection on Afterburner04

[joska@nuts.edu (Jo Even Skarstein)]

On Sun, 2 May 1999 22:37:17 +0200, Thomas Binder wrote:

> But this could only protect parts of MiNT's TEXT segmen without a
> deicated loader, because it may very likely start at - for example -
> 0x2001, which means the first 0x1fff byte couldn't be protected because
> the segment doesn't begin on a page boundary (or otherwise, memory we
> don't know of would be write-protected).

You could always pad the first 8k with NOPs or something... OK, that's a
ugly hack, but the kernel really should be protected. A "MiLO" would perhaps
be the most elegant solution.

> Have you looked where  resvector points to (type of memory protection
> there, can be determined by looking at the MMU dump created by either
> Ctrl-Shift-Alt-F9 (-DDEBUG INFO-kernel) or mmudump) and where 0x4 points
> to? Is the VBR != 0?

I'll have to compile a debug-kernel first then...
(10 mins. later...) I can dump the mmu-tree alright, but TosWin2 only reads
the first 10 lines or so I can't get a log :-(

> What exactly happens when you try to reset per software, for example via
> 
> Supexec(Setexc(1, -1L));
> 
> or
> 
> Shutdown(1);

Apart from a total freeze? Nothing ;-)


/*
** Jo Even Skarstein    http://www.stud.ntnu.no/~josk/
**
**    beer - maria mckee - atari falcon - babylon 5
*/