Re: [PATCH v5] RO/NX protection for loadable kernel modules

[Greg KH <greg@kroah.com>]

On Sun, Jul 12, 2009 at 08:32:27AM -0700, Arjan van de Ven wrote:
> On Sun, 12 Jul 2009 19:28:27 +0930
> Rusty Russell <rusty@rustcorp.com.au> wrote:
> 
> > On Sun, 12 Jul 2009 05:15:24 pm Arjan van de Ven wrote:
> > > Rusty Russell <rusty@rustcorp.com.au> wrote:
> > > > (I like the idea of trying kmalloc and falling back, simply
> > > > because it reduces TLB pressure, but that's probably best done
> > > > after unification).
> > >
> > > or using a non-power-of-two get_free_pages() thing...
> > >
> > > some architectures will need to know that memory needs to be
> > > executable at allocation time so that it can be put in an
> > > executable address range etc...
> > 
> > Yes, maybe that's better than kmalloc.  On my laptop I have 105
> > modules loaded, with 3778464 total length: I'm wasting 206944 bytes
> > on unused tails of pages.  But that's only 0.06% of my memory.
> >
> 
> 105 is also a sign that you picked a somewhat suboptimal config...
> that's of course your choice but it's a choice that has a small price,
> if you don't want to pay that price, changing the config to not be
> entirely insane is a good answer as well ;-)

But this is the "common" case in the world of Linux where the distros
are forced to build everything as modules.  So it should be considered
as a very valid case, and not merely dismissed out of hand.

thanks,

greg k-h