>>> On 13.02.18 at 12:27, <roger....@citrix.com> wrote:
> On Tue, Feb 13, 2018 at 04:04:17AM -0700, Jan Beulich wrote:
>> >>> On 13.02.18 at 10:59, <roger....@citrix.com> wrote:
>> > On Tue, Feb 13, 2018 at 02:29:08AM -0700, Jan Beulich wrote:
>> >> >>> On 08.02.18 at 13:25, <roger....@citrix.com> wrote:
>> >> > Signed-off-by: Roger Pau Monné <roger....@citrix.com>
>> >> 
>> >> A change like this should not come without description, providing a
>> >> reason for the change. Otherwise how will someone wanting to
>> >> understand the change in a couple of years actually be able to
>> >> make any sense of it. This is in particular because I continue to be
>> >> not fully convinced that white listing is appropriate in the Dom0
>> >> case (and for the record I'm similarly unconvinced that black listing
>> >> is the best choice, yet obviously we need to pick on of the two).
>> > 
>> > I'm sorry, I thought we agreed at the summit to convert this to
>> > whitelisting because it was likely better to simply not expose unknown
>> > ACPI tables to guests.
>> 
>> "to guests" != "to Dom0".
>> 
>> > I guess the commit message could be something like:
>> > 
>> > "The following list of whitelisted APIC tables are either known to work
>> > or don't require any resources to be mapped in either the IO or the
>> > memory space.
>> 
>> Even if the white listing vs black listing question wasn't still
>> undecided, I think we should revert the patch in favor of one
>> with a description. The one above might be fine with "ACPI" in
>> place of "APIC" as far as tables actively white listed are
>> concerned, but then it still remains open why certain tables
>> haven't been included. I'm in particular worried about various
>> APEI related tables, but invisibility of e.g. an IBFT could also
>> lead to boot problems.
> 
> Regarding APEI I think ERST, EINJ and HEST could be passed through,
> BERT however requires that the BOOT Error Region is mapped into Dom0
> p2m.
> 
> Since PVH Dom0 creation still ends up in a panic, I see no problem in
> adding those in follow up patches.
> 
> IBFT also looks safe to pass through.

But you realize I've named only the few that came to mind
immediately?

Jan

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Reply via email to