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

Re: [Xen-devel] [PATCH v6 05/11] arch, x86: check if mapping exists before memory_mapping removes it



>>> On 22.04.14 at 10:53, <julien.grall@xxxxxxxxxx> wrote:
> On 22/04/14 09:34, Jan Beulich wrote:
>>>>> On 21.04.14 at 15:44, <avanzini.arianna@xxxxxxxxx> wrote:
>>> Currently, when a memory mapping is removed with the memory_mapping
>>> DOMCTL, no check is performed on the existence of such a mapping.
>>> This commit attempts to add such a consistency check to the code
>>> performing the unmap.
>>
>> I think this goes too far: Did you check that all existing tool stacks
>> actually pass a valid MFN for the unmap? It would seem quite natural
>> to me if some didn't, since tool stacks can be expected to know what
>> they're doing.
> 
> If the toolstack doesn't give a valid MFN that would mean that the 
> toolstack can mess up the range permission.

Oh, right. So there's at least some hope that the tool stacks get this
right (but no guarantee).

> On a previous mail, I suggested to skip the MFN parameter when the 
> toolstack is unmapping the range. Xen will take care to translate the 
> GFN into an MFN.

But that won't be needed anymore when access revocation and
unmapping get done through distinct hypercalls.

Jan


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


 


Rackspace

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