Re: [PATCH v6 11/14] x86: add KHO support

[Mike Rapoport <rppt@kernel.org>]

On Mon, Apr 28, 2025 at 03:05:55PM -0700, Dave Hansen wrote:
> On 4/10/25 22:37, Changyuan Lyu wrote:
> > From: Alexander Graf <graf@amazon.com>
> > 
> > +/*
> > + * If KHO is active, only process its scratch areas to ensure we are not
> > + * stepping onto preserved memory.
> > + */
> 
> Same thing on the imperative voice here.
> 
> I'm also not fully understanding the comment. Do these "scratch" regions
> basically represent all the memory that's not being handed over? It's
> not obvious.

Scratch memory represents areas created at the boot of the first kernel and
it's known that scratch areas won't contain any memory that's being handed
over.
 
> > diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c
> > index 57120f0749cc3..c314212a5ecd5 100644
> > --- a/arch/x86/kernel/e820.c
> > +++ b/arch/x86/kernel/e820.c
> > @@ -1300,6 +1300,24 @@ void __init e820__memblock_setup(void)
> >  		memblock_add(entry->addr, entry->size);
> >  	}
> >  
> > +	/*
> > +	 * At this point with KHO we only allocate from scratch memory.
> > +	 * At the same time, we configure memblock to only allow
> > +	 * allocations from memory below ISA_END_ADDRESS which is not
> > +	 * a natural scratch region, because Linux ignores memory below
> > +	 * ISA_END_ADDRESS at runtime. Beside very few (if any) early
> > +	 * allocations, we must allocate real-mode trapoline below
> 
> 						trampoline ^
> 
> > +	 * ISA_END_ADDRESS.
> > +	 *
> > +	 * To make sure that we can actually perform allocations during
> > +	 * this phase, let's mark memory below ISA_END_ADDRESS as scratch
> > +	 * so we can allocate from there in a scratch-only world.
> > +	 *
> > +	 * After real mode trampoline is allocated, we clear scratch
> > +	 * marking from the memory below ISA_END_ADDRESS
> > +	 */
> > +	memblock_mark_kho_scratch(0, ISA_END_ADDRESS);
> 
> This isn't making a whole ton of sense to me.
> 
> Is this *only* to facilitate possible users that need <ISA_END_ADDRESS
> allocations? If so, please say that.
> 
> I _think_ this is trying to say that KHO kernels are special and are
> trying to only allocate from scratch areas. But <ISA_END_ADDRESS
> allocations are both necessary and not marked by KHO _as_ a scratch area
> which causes a problem.

Yes :)

-- 
Sincerely yours,
Mike.