Permanently resident pages:

Every virtual memory systems have memory areas that are "pinned down", for example cannot be swapped out to secondary storage, for instance:

• Interrupt mechanisms usually rely on an array of pointers to the handlers for different types of interrupt (I/O completion, program error, timer event, page fault, etc.). If the pages containing these pointers or the code that they invoke were page able, interrupt-handling would become even more complicated and time-consuming; and it would be especially hard in the case of page fault interrupts.
• The page tables are typically not page able.
• Data buffers that are accessed outside of the CPU, for instance by peripheral devices that use direct memory access (DMA) or by I/O channels. Typically such devices and the buses (connection paths) to which they are connected use physical memory addresses rather than virtual memory addresses. Even on buses  with  an  IOMMU,  that  is  a  special  memory  management  unit  that  may  translate  virtual addresses used on an I/O bus to physical addresses that transfer can't be stopped if a page fault take place and then restarted when the page fault has been processed. So pages that contain locations to which or from which a peripheral device is transferring data are either enduringly pinned down or pinned down whilst the transfer is in progress.
• Timing-dependent kernel/application cannot stand the varying response time caused by paging.