Re: [Xen-ia64-devel] Modify to introduce delayed p2m table destruction

[Doi.Tsunehisa@xxxxxxxxxxxxxx]

You (yamahata) said:
>>   We have studied it. So, we think that we can't realize to defer
>> p2m table destruction with the convension. Thus, we'll take anothor
>> approach to avoid the hypervisor hang. Basicaly, we'll introduce a
>> reference count of p2m table like page struct.
> 
> Why do you think so?
> I believe that it is possible without such a reference count.

  At first, we think that we can't realize to defer p2m table destruction
because the convention requires that p2m table exists while page struct
exists.

  About p2m table reference count, we tried to implement to check
as if d->arch.mm.pgd is NULL, at first. But, we can't protect to
access the table during p2m table destruction, it might start to
travese p2m table directory structure.

  Next, we considered to check domain dying before p2m table access.
But the solution requires that domain_kill has to wait for complesion
of p2m table access at test_and_set_bit().

  Therefore, we think to introduce the reference count.

Thanks,
- Tsunehisa Doi


  

_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel