[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH 1/2] pvh: clearly specify used parameters in vcpu_guest_context

On 20/11/13 19:19, George Dunlap wrote:
> On 20/11/13 10:29, Roger Pau Monnà wrote:
>> On 20/11/13 10:54, Jan Beulich wrote:
>>>>>> On 20.11.13 at 10:37, Roger Pau MonnÃ<roger.pau@xxxxxxxxxx> wrote:
>>>> On 20/11/13 10:28, Jan Beulich wrote:
>>>>>>>> On 20.11.13 at 10:18, Roger Pau MonnÃ<roger.pau@xxxxxxxxxx> wrote:
>>>>>> On 19/11/13 17:53, Jan Beulich wrote:
>>>>>>>>>> On 19.11.13 at 17:42, Roger Pau MonnÃ<roger.pau@xxxxxxxxxx>
>>>>>>>>>> wrote:
>>>>>>>> Trying to make all those fields functional on PVH (or HVM) is quite
>>>>>>>> useless IMHO, it's going to add more code that I doubt anyone is
>>>>>>>> going
>>>>>>>> to use when you can instead use the bare metal functions to set all
>>>>>>>> those things (and from an OS point of view it's also more
>>>>>>>> comfortable
>>>>>>>> because you need less Xen specific stuff).
>>>>>>> That last part I certainly agree to, but that would apply to CR0
>>>>>>> and CR4 just as much.
>>>>>> I've removed the usage of anything that's not strictly necessary in
>>>>>> order to do AP bringup, so I've removed the setting of debugregs:
>>>>> You can't - this code is also used for HVM guests.
>>>> Yes, my fault, I erroneously thought this was introduced by 35b1e076,
>>>> but it has been there longer than that. Would you agree to a patch
>>>> similar to the one posted, but without touching the setting of
>>>> debugregs?
>>> Yes, if Mukesh and George confirm that this is not going to break
>>> things.
> Well it does change the interface, by not setting gs_base_kernel; but
> that was part of the point. :-)
> The rest of it looks OK to me -- Roger, have you tested Linux?

No, I expect Mukesh to test it, since this requires Linux to not try to
set gs_base_kernel.

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.