Re: [PATCH] [1/5] CPA: Split static_protections into required_static_prot and advised_static_prot

[Andi Kleen <andi@firstfloor.org>]

On Sat, Feb 09, 2008 at 06:09:23PM +0100, Thomas Gleixner wrote:
> On Sat, 9 Feb 2008, Andi Kleen wrote:
> > > > Note the behaviour for pageattr and thus DEBUG_RODATA / debugging
> > > > sitations where you don't care about your TLB this
> > > > does not change, this makes only a difference for the initial init_32
> > > > direct mapping setup.
> > > 
> > > Your patches do change the behaviour. The range checking breaks the
> > > enforcement of some restrictions for the sake of keeping the large
> > > page intact.
> > 
> > You mean in try_preserve_large_page()?
> > 
> > No actually they were not completely enforced previously at all, because
> > it did only check the restrictions of the first page.
> 
> Right, you poked my nose to it. I did not think about it when I coded
> it. It is wrong and needs to be fixed, but not by the range check you
> introduced.

Well I need the range check for a different piece of code (init_memory_mapping())
For that a range check is definitely needed and the existing code there
also does an (although quite fishy) range check. The DEBUG_RODATA case
 is also handled correctly there because DEBUG_RODATA is applied explicitely 
using pageattr later. 

You have not commented on that at all so I assume it's ok for you.

> > On the end of my patch series the enforcement is actually stricter
> > than it was before, although not 100%.
> 
> As far as I can tell it is more relaxed, as it will make overlapping
> regions of rodata and rwdata completely rw instead of splitting it up.

In try_preserve_large_page()? No because it only checks the first page.

In all other cases (in the existing code; my patchkit adds a new case 
in mm/init_32.c) it always only checks single 4K pages so the only
overlap case would be sub 4K. For that there can be no split up.

-Andi