> Is it possible to use the page_is_ram() and kamp() path in the patch to avoid the situation you mentioned?
Maybe - it could certainly check the current attributes and match them.
Not sure whether there might be a gap if the acpi mapping is long-lived,
and the other kernel mappings change ... though this is very rare to
switch attributes
-Tony
��.n��������+%������w��{.n�����{�����ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f
