Editing Page replacement algorithm (section)

===Techniques for hardware with no reference bit===
Many of the techniques discussed above assume the presence of a reference bit associated with each page.  Some hardware has no such bit, so its efficient use requires techniques that operate well without one.

One notable example is [[VAX]] hardware running [[OpenVMS]]. This system knows if a page has been modified, but not necessarily if a page has been read. Its approach is known as Secondary Page Caching. Pages removed from working sets (process-private memory, generally) are placed on special-purpose lists while remaining in physical memory for some time. Removing a page from a working set is not technically a page-replacement operation, but effectively identifies that page as a candidate. A page whose backing store is still valid (whose contents are not dirty, or otherwise do not need to be preserved) is placed on the tail of the Free Page List. A page that requires writing to backing store will be placed on the Modified Page List. These actions are typically triggered when the size of the Free Page List falls below an adjustable threshold.

Pages may be selected for working set removal in an essentially random fashion, with the expectation that if a poor choice is made, a future reference may retrieve that page from the Free or Modified list before it is removed from physical memory. A page referenced this way will be removed from the Free or Modified list and placed back into a process working set. The Modified Page List additionally provides an opportunity to write pages out to backing store in groups of more than one page, increasing efficiency. These pages can then be placed on the Free Page List. The sequence of pages that works its way to the head of the Free Page List resembles the results of a LRU or NRU mechanism and the overall effect has similarities to the Second-Chance algorithm described earlier.

Another example is used by the [[Linux kernel]] on [[ARM architecture family|ARM]]. The lack of hardware functionality is made up for by providing two page tables – the processor-native page tables, with neither referenced bits nor [[dirty bit]]s, and software-maintained page tables with the required bits present. The emulated bits in the software-maintained table are set by page faults. In order to get the page faults, clearing emulated bits in the second table revokes some of the access rights to the corresponding page, which is implemented by altering the native table.