Re: [PATCH 05/10] VMX: add help functions to support PML

[Kai Huang <kai.huang@linux.intel.com>]

On 04/10/2015 05:03 PM, Tim Deegan wrote:
> At 15:05 +0800 on 10 Apr (1428678346), Kai Huang wrote:
>>
>> On 04/09/2015 08:00 PM, Tim Deegan wrote:
>>> Hi,
>>>
>>> That's also buggy, because there's no locking here to make sure
>>> gfn->mfn->gfn ends up in the right place. :(
>> Yes you are right. Thanks for pointing out.
>>
>> Just curious, looks if I use get_gfn_type_access and put_gfn here, the
>> gfn->mfn->gfn can be guaranteed, and it's safe to use them here, right?
> Yes, that would be the right way to do it.
>
> In this particular case, there's another race too: the gfn might have
> been entirely deleted after it's logged by PML but before the entry is
> processed here.  I don't think that's a problem here, because if the
> gfn's not mapped any more it's not relevant that it's dirty.  Just
> mentioning it out of interest.
Agreed. I think the gone gfn should be irrelevant. Thanks.

Thanks,
-Kai
>
> Cheers,
>
> Tim.