[PATCH v2, part3 07/12] mm: accurately calculate zone->managed_pages for highmem zones

Commit "mm: introduce new field 'managed_pages' to struct zone" assumes that all highmem pages will be freed into the buddy system by function mem_init(). But that's not always true, some architectures may reserve some highmem pages during boot. For example PPC may allocate highmem pages for

[PATCH v2, part3 07/12] mm: accurately calculate zone-managed_pages for highmem zones

Commit mm: introduce new field 'managed_pages' to struct zone assumes that all highmem pages will be freed into the buddy system by function mem_init(). But that's not always true, some architectures may reserve some highmem pages during boot. For example PPC may allocate highmem pages for giagant