Re: [Xen-devel] [PATCH v1 10/20] acpi/hvmloader: Replace mem_alloc() and virt_to_phys() with memory ops

[Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx>]

On 07/19/2016 05:11 AM, Jan Beulich wrote:
>>>> Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx> 07/08/16 6:20 PM >>>
>> On 07/08/2016 11:35 AM, Jan Beulich wrote:
>>>>>> On 08.07.16 at 17:23, <boris.ostrovsky@xxxxxxxxxx> wrote:
>>>> Is it up to the builder to decide which tables are important and which
>>>> are not?
>>> I'm afraid that's not so easy to tell. If for example we can't fit the
>>> HPET table, the guest could be run without HPET unless a HPET
>>> was specifically requested (rather than just defaulted to).
>> But again --- how will the caller know that it was only HPET table that
>> was not built?
> Why would the caller care? I guess examples could be found where it is
> necessary for the caller to know, but for the specific example (and at least
> some others) failure is of no interest to the caller - it's only the guest 
> which
> is affected.

HPET was just an example, the same question could be asked for (almost)
any other table.

But I can see that we can defer to the guest to deal with ACPI
brokenness, although some not built tables will almost certainly lead to
guest's failure.

(We probably will not get to use this new free() op anyway since failure
to allocate memory is currently the only possible error and there is one
allocation per table)


-boris






_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel