>>> @@ -528,7 +532,7 @@ ENTRY(_stext)
>>> /*
>>> * BSS section
>>> */
>>> -.section ".bss.page_aligned","w"
>>> +.section ".bss.page_aligned"
>>>
>>
>> Why?
>>
>
>I got complaints about section attribute mismatches without it.
Then perhaps ... "aw" is meant?
>>> +fastcall unsigned long long xen_pgd_val(pgd_t pgd)
>>> +{
>>> + unsigned long long ret = pgd.pgd;
>>> + if (ret)
>>> + ret = machine_to_phys(XMADDR(ret)).paddr | 1;
>>>
>>
>> Why can they be 0 here anyways?
>>
>> Normally they are all considered undefined when not present
>>
>
>Not sure.
This should probably get sync-ed up with the page table handling changes
submitted to xen-devel yesterday. Using zero/non-zero tests in contexts
like this was always broken; should check for _PAGE_PRESENT.
Jan
