>> True, but if we decide to go on that direction, a change like that
>> should then be done on all EDAC drivers, and that's an API change.
>>
>> We also need to be sure that userspace will handle this change
>> properly.
>
> We don't have a choice:
>
> [ 0.000000] e820: BIOS-provided physical RAM map:
> [ 0.000000] BIOS-e820: [mem 0x0000000000000000-0x000000000009e7ff] usable
Since we have separate trace points for EDCA and extlog - we can use different
conventions for "invalid". If you think that changing the ABI in EDAC would be a
problem - then you can keep running the almost infinitesimal chance that you have
a real error at 0x00000000000000 that you treat as unknown address.
Does X86 actually allocate that page for use? - I know on Itanium it got thrown out
when rounding usable blocks of memory to 16MB boundaries for cache coherency
reasons.
-Tony
��.n��������+%������w��{.n�����{�����ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f
