[rsbac] small conflict xfs/rsbac 2.4.20

[Amon Ott]

On Sunday 29 December 2002 20:00, Tony den Haan wrote:
> i ran into this mprotect.c.rej when applying rsbac patch on 2.4.20 with
> xfs patch:
> 
> ***************
> *** 300,305 ****
>                         goto out;
>                 }
> 
>                 if (vma->vm_end > end) {
>                         error = mprotect_fixup(vma, &prev, nstart, end,
> newflags);
>                         goto out;
> --- 313,357 ----
>                         goto out;
>                 }
> 
> + #ifdef CONFIG_RSBAC
> <<<the rsbac bit to be added.....>>>
> + #endif
>         if (vma->vm_end > end) {
>                         error = mprotect_fixup(vma, &prev, nstart, end, 
> newflags);
>                         goto out;
> 
> 
> 
> this because xfs patch inserts a test just before "if (vma->vm_end >
> end) {"

The old problem with other additional patches, there are often conflicts 
somewhere.
 
> fixing seems trivial, but made me wonder: why is the rsbac related test,
> which seems more expensive, performed *before* the 
> "if (vma->vm_end > end)" test?
> 
> it looks like a less expensive sanity check, might even prevent problems?

I wanted to be sure that the RSBAC check is always called, and that no side 
effect changes through mprotect_fixup can happen before it.

It does not really matter, where the RSBAC interception happens, because a 
failed sanity check is a rare event.

Amon.
--
http://www.rsbac.org - GnuPG: 2048g/5DEAAA30 2002-10-22