Hi, Robert,
There is another minor problem. in dt_init_action:
+                if (_32pae_l3 || c_curr_accessed)
+                    action |= a_present;
+             /* else make present when accessed */
+                if (c_curr_rw)
+                    action |= a_rw;

According to your comments, the following snippet may be more correct:
+                if (_32pae_l3 || c_curr_accessed)
+                    action |= a_present;
+             /* else make present when accessed */
+                if (c_curr_rw && !_32_pae_l3)
+                    action |= a_rw;
The doc said we should retain R/W be zero for pae_l3_pte in legacy mode. 
  If returned a_rw, resync_spte will set _PAGE_RW for the corresponding 
spte.
_______________________________________________________
Best Regards,
hanzhu


Robert Phillips дµÀ:
> Yes, that's true.  If the caller supplied a bogus PFN (one larger than ram
> size), we'd crash.
> BTW, we are avoiding use of grant table transfers because they fragment PSE
> pages.
> -- rsp
> 
> 
> On 6/30/06, zhu <vanbas.han@gmail.com> wrote:
>>
>> I'm afraid it could trigger some certain problems only when you open l2
>> PSE flag and use grant table to transfer pages between domains. I'm not
>> very sure about it. :-)
>>
>> _______________________________________________________
>> Best Regards,
>> hanzhu
>>
>>
>> Robert Phillips $B<LF;(B:
>> > Hi Han,
>> > You are right again.  I have not seen this cause a problem but we'll
>> > certainly fix it.
>> > -- rsp
>> >
>> > On 6/29/06, zhu <vanbas.han@gmail.com> wrote:
>> >>
>> >> Hi, Robert,
>> >> I found out another confusing code snippet:
>> >> in void xi_invl_mfn(struct domain *d, unsigned long mfn)
>> >>      if (ext && pfn < ext->large_page_aligned_size)
>> >>
>> >> According to the code, it should be
>> >>      if (ext && (pfn>>SPT_ENTRIES_ORDER) <
>> ext->large_page_aligned_size)
>> >>
>> >> If I made any mistake, please point it out.
>> >> _______________________________________________________
>> >> Best Regards,
>> >> hanzhu
>> >>
>> >
>> >
>>
> 
> 
>