Re: [PATCH 01/10] x86/mm: no-one passes a NULL domain to init_xen_l4_slots()

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 17/04/2020 15:25, Jan Beulich wrote:
> Drop the NULL checks - they've been introduced by commit 8d7b633ada
> ("x86/mm: Consolidate all Xen L4 slot writing into
> init_xen_l4_slots()") for no apparent reason.

:) I'll take this as conformation that all my sudden pagetable work in
Xen manage ended up being rather more subtle than Linux's equivalent
work for KPTI.

https://lists.xenproject.org/archives/html/xen-devel/2018-01/msg00281.html

Specifically, this was part of trying to arrange for fully per-pcpu
private mappings, and was used to construct the pagetables for the idle
vcpu which specifically don't have a perdomain mapping.

Seeing as this is still an outstanding task in the secret-free-Xen
plans, any dropping of it now will have to be undone at some point in
the future.  Is there a specific reason for the cleanup?

> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
>
> --- a/xen/arch/x86/mm.c
> +++ b/xen/arch/x86/mm.c
> @@ -1696,7 +1696,7 @@ void init_xen_l4_slots(l4_pgentry_t *l4t

If we continue with this patch, this comment, just out of context, needs
adjusting.

~Andrew

>       * PV vcpus need a shortened directmap.  HVM and Idle vcpus get the full
>       * directmap.
>       */
> -    bool short_directmap = d && !paging_mode_external(d);
> +    bool short_directmap = !paging_mode_external(d);
>  
>      /* Slot 256: RO M2P (if applicable). */
>      l4t[l4_table_offset(RO_MPT_VIRT_START)] =
>