Re: [Xen-devel] [PATCH 2 of 2] x86/mm: New mem access type to log access

["Andres Lagar-Cavilla" <andres@xxxxxxxxxxxxxxxx>]

> At 16:58 -0500 on 29 Nov (1322585905), Andres Lagar-Cavilla wrote:
>> @@ -1162,10 +1167,13 @@ int p2m_mem_access_check(unsigned long g
>>          }
>>          else
>>          {
>> -            /* A listener is not required, so clear the access
>> restrictions */
>> -            p2m_lock(p2m);
>> -            p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K, p2mt,
>> p2m_access_rwx);
>> -            p2m_unlock(p2m);
>> +            if ( p2ma != p2m_access_n2rwx )
>> +            {
>> +                /* A listener is not required, so clear the access
>> restrictions */
>> +                p2m_lock(p2m);
>> +                p2m->set_entry(p2m, gfn, mfn, PAGE_ORDER_4K, p2mt,
>> p2m_access_rwx);
>> +                p2m_unlock(p2m);
>> +            }
>>              return 1;
>>          }
>
> This logic is getting a bit convoluted, and I'm not sure it's correct.
> If a page is marked n2rwx and there's no listener, it looks like this
> will cause it to spin forever re-taking the fault rather than pausing it
> waiting for the listener to attach.

The entry is set to p2m_access_rwx, so no additional faults will be
generated. In the n2rwx case, the entry was promoted to rwx previously, in
 a p2m_lock protected section. We're avoiding a repeat here.

Andres

>
> Cheers,
>
> Tim.
>



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