Re: [PATCH] x86/efistub: Omit physical KASLR when memory reservations exist

[Ard Biesheuvel]

On Thu, 16 May 2024 at 20:45, Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>
> On Thu, May 16, 2024 at 05:29:11PM +0000, Chaney, Ben wrote:
> > > +static efi_status_t parse_options(const char *cmdline)
> > > +{
> > > + static const char opts[][14] = {
> > > + "mem=", "memmap=", "efi_fake_mem=", "hugepages="
> > > + };
> > > +
> >
> > I think we probably want to include both crashkernel and pstore as arguments that can disable this randomization.
>
> The carve-outs that pstore uses should already appear in the physical
> memory mapping that EFI has. (i.e. those things get listed in e820 as
> non-RAM, etc)
>
> I don't know anything about crashkernel, but if we really do have a lot
> of these, we likely need to find a way to express them to EFI...
>

Perhaps. But the fact that the current KASLR code ignores it entirely
suggests that this has not been a problem up to this point.