A memory management unit (MMU), sometimes called paged memory management unit (PMMU), is a computer hardware component responsible for handling accesses to memory requested by the central processing unit (CPU). Its functions include translation of virtual addresses to physical addresses (i.e., virtual memory management), memory protection, cache control, bus arbitration, and, in simpler computer architectures (especially 8-bit systems), bank switching.
Modern MMUs typically divide the virtual address space (the range of addresses used by the processor) into pages, each having a size which is a power of 2, usually a few kilobytes. The bottom n bits of the address (the offset within a page) are left unchanged. The upper address bits are the (virtual) page number. The MMU normally translates virtual page numbers to physical page numbers via an associative cache called a Translation Lookaside Buffer (TLB). When the TLB lacks a translation, a slower mechanism involving hardware-specific data structures or software assistance is used. The data found in such data structures are typically called page table entries (PTEs), and the data structure itself is typically called a page table. The physical page number is combined with the page offset to give the complete physical address.
A PTE or TLB entry may also include information about whether the page has been written to (the dirty bit), when it was last used (the accessed bit, for a least recently used page replacement algorithm), what kind of processes (user mode, supervisor mode) may read and write it, and whether it should be cached.
Sometimes, a TLB entry or PTE prohibits access to a virtual page, perhaps because no physical random access memory has been allocated to that virtual page. In this case the MMU signals a page fault to the CPU. The operating system (OS) then handles the situation, perhaps by trying to find a spare frame of RAM and set up a new PTE to map it to the requested virtual address. If no RAM is free, it may be necessary to choose an existing page, using some replacement algorithm, and save it to disk (this is called "paging"). With some MMUs, there can also be a shortage of PTEs or TLB entries, in which case the OS will have to free one for the new mapping.
In some cases a "page fault" may indicate a software bug. A key benefit of an MMU is memory protection: an OS can use it to protect against errant programs, by disallowing access to memory that a particular program should not have access to. Typically, an OS assigns each program its own virtual address space.
An MMU also reduces the problem of fragmentation of memory. After blocks of memory have been allocated and freed, the free memory may become fragmented (discontinuous) so that the largest contiguous block of free memory may be much smaller than the total amount. With virtual memory, a contiguous range of virtual addresses can be mapped to several non-contiguous blocks of physical memory.
In some early microprocessor designs, memory management was performed by a separate integrated circuit such as the MC 68851 used with the Motorola 68020 CPU in the Macintosh II or the Z8015 used with the Zilog Z80 family of processors. Later microprocessors such as the Motorola 68030 and the ZILOG Z280 placed the MMU together with the CPU on the same integrated circuit, as did the Intel 80286 and later x86 microprocessors.
While this article concentrates on modern MMUs, commonly based on pages, early systems used a similar concept for base-limit addressing, that further developed into segmentation. Those are occasionally also present on modern architectures. The x86 architecture provided segmentation rather than paging in the 80286, and provides both paging and segmentation in the 80386 and later processors.
Most modern systems divide memory into pages that are 4 KiB to 64 KiB in size, often with the possibility to use huge pages from 2 MiB to 512 MiB in size. Page translations are cached in a TLB. Some systems, mainly older RISC designs, trap into the OS when a page translation is not found in the TLB. Most systems use a hardware-based tree walker. Most systems allow the MMU to be disabled; some disable the MMU when trapping into OS code.
VAX pages are 512 bytes, which is very small. An OS may treat multiple pages as if they were a single larger page, for example Linux on VAX groups 8 pages together, so that the system is viewed as having 4 KiB pages. The VAX divides memory into 4 fixed-purpose regions, each 1 GiB in size. They are:
Page tables are big linear arrays. Normally this would be very wasteful when addresses are used at both ends of the possible range, but the page table for applications is itself stored in the kernel's paged memory. Thus there is effectively a 2-level tree, allowing applications to have sparse memory layout without wasting lots of space on unused page table entries. The VAX MMU is notable for lacking an accessed bit. OSes which implement paging must find some way to emulate the accessed bit if they are to operate efficiently. Typically, the OS will periodically unmap pages so that page-not-present faults can be used to let the OS set an accessed bit.
ARM architecture based application processors implement an MMU defined by ARM's Virtual Memory System Architecture. The current architecture defines PTEs for describing 4KiB and 64KiB pages, 1MiB sections and 16MiB super-sections; legacy versions also defined a 1KiB tiny page.
TLB updates are performed automatically by page-table walking hardware.
The IBM System/370 have had an MMU since early 1970s, it was initially known as DAT box. It has the unusual feature of storing accessed and dirty bits outside of the page table. They refer to physical memory rather than virtual memory. They are accessed by special-purpose instructions. This reduces overhead for the OS, which would otherwise need to propagate accessed and dirty bits from the page tables to a more physically-oriented data structure. This makes OS-level virtualization easier. These features have been inherited by succeeding mainframe architectures, up to the current z/Architecture.
The DEC Alpha processor divides memory into 8192-byte pages. After a TLB miss, low-level firmware machine code (here called PALcode) walks a 3-level tree-structured page table. Addresses are broken down as follows: 21 bits unused, 10 bits to index the root level of the tree, 10 bits to index the middle level of the tree, 10 bits to index the leaf level of the tree, and 13 bits that pass through to the physical address without modification. Full read/write/execute permission bits are supported.
The original Sun 1 was a single-board computer built around the Motorola 68000 microprocessor and introduced in 1982. It included the original Sun 1 Memory Management Unit, that provided address translation, memory protection, memory sharing and memory allocation for multiple processes running on the CPU. All access of the CPU to private on-board RAM, external Multibus memory, on-board I/O and the Multibus I/O ran through the MMU where they were translated and protected in uniform fashion. The MMU was implemented in hardware on the CPU board.
The MMU consisted of a context register, a segment map and a page map. Virtual addresses from the CPU were translated into intermediate addresses by the segment map, which in turn were translated into physical addresses by the page map. The page size was 2 KiB and the segment size was 32 KiB which gave 16 pages per segment. Up to 16 contexts could be mapped concurrently. The maximum logical address space for a context was 1024 pages or 2 MiB. The maximum physical address that could be mapped simultaneously was also 2 MiB.
The context register was important in a multitasking operating system because it allowed the CPU to switch between processes without reloading all the translation state information. The 4-bit context register could switch between 16 sections of the segment map under supervisor control which allowed 16 contexts to be mapped concurrently. Each context had its own virtual address space. Sharing of virtual address space and inter-context communications could be provided by writing the same values in to the segment or page maps of different contexts. Additional contexts could be handled by treating the segment map as a context cache and replacing out-of-date contexts on a least-recently-used basis.
The context register made no distinction between user and supervisor states; interrupts and traps did not switch contexts which required that all valid interrupt vectors always be mapped in page 0 of context, as well as the valid Supervisor Stack.
In PowerPC G1, G2, G3, and G4, pages are normally 4 KiB. After a TLB miss, the standard PowerPC MMU begins two simultaneous lookups. One lookup attempts to match the address with one of 4 or 8 Data Block Address Translation (DBAT) registers, or 4 or 8 Instruction Block Address Translation registers (IBAT) as appropriate. The BAT registers can map linear chunks of memory as large as 256 MiB, and are normally used by an OS to map large portions of the address space for the OS kernel's own use. If the BAT lookup succeeds, the other lookup is halted and ignored.
The other lookup, not directly supported by all processors in this family, is via a so-called "inverted page table" which acts as a hashed off-chip extension of the TLB. First, the top 4 bits of the address are used to select one of 16 segment registers. 24 bits from the segment register replace those 4 bits, producing a 52-bit address. The use of segment registers allows multiple processes to share the same hash table. The 52-bit address is hashed, then used as an index into the off-chip table. There, a group of 8 page table entries is scanned for one that matches. If none match due to excessive hash collisions, the processor tries again with a slightly different hash function. If this too fails, the CPU traps into the OS (with MMU disabled) so that the problem may be resolved. The OS needs to discard an entry from the hash table to make space for a new entry. The OS may generate the new entry from a more-normal tree-like page table or from per-mapping data structures which are likely to be slower and more space-efficient. Support for no-execute control is in the segment registers, leading to 256-MiB granularity.
A major problem with this design is poor cache locality caused by the hash function. Tree-based designs avoid this by placing the page table entries for adjacent pages in adjacent locations. An operating system running on the PowerPC may minimize the size of the hash table to reduce this problem.
It is also somewhat slow to remove the page table entries of a process; the OS may avoid reusing segment values to delay facing this or it may elect to suffer the waste of memory associated with per-process hash tables. G1 chips do not search for page table entries, but they do generate the hash with the expectation that an OS will search the standard hash table via software. The OS can write to the TLB. G2, G3, and early G4 chips use hardware to search the hash table. The latest chips allow the OS to choose either method. On chips that make this optional or do not support it at all, the OS may choose to use a tree-based page table exclusively.
The x86 architecture has evolved over a long time while maintaining full software compatibility even for OS code. Thus the MMU is extremely complex, with many different possible operating modes. Normal operation of the traditional 80386 CPU and its successors is described here.
The CPU primarily divides memory into 4 KiB pages. Segment registers, fundamental to the older 8088 and 80286 MMU designs, are avoided as much as possible by modern OSes. There is one major exception to this: access to thread-specific data for applications or CPU-specific data for OS kernels, which is done with explicit use of the FS and GS segment registers. All memory access involves a segment register, chosen according to the code being executed. The segment register acts as an index into a table, which provides an offset to be added to the virtual address. Except when using FS or GS as described above, the OS ensures that the offset will be zero. After the offset is added, the address is masked to be no larger than 32 bits. The result may be looked up via a tree-structured page table, with the bits of the address being split as follows: 10 bits for the root of the tree, 10 bits for the leaves of the tree, and the 12 lowest bits being directly copied to the result.
Minor revisions of the MMU introduced with the Pentium have allowed very large 2 MiB or 4 MiB pages by skipping the bottom level of the tree. Minor revisions of the MMU introduced with the Pentium Pro have allowed 36-bit physical addresses with the Physical Address Extension (PAE) feature and have allowed specification of cacheability by looking up a few high bits in a small on-CPU table.
No-execute support was originally only provided on a per-segment basis, making it very awkward to use. More recent x86 chips provide a per-page no-execute bit in the PAE mode. PaX is one way to emulate per-page non-execute support via the segments, with a performance loss and halving the available address space.
x86-64 is a 64-bit extension of x86, that uses the long mode. In long mode, all segment offsets are ignored, except FS and GS. The page table tree has four levels, instead of three. The virtual addresses are divided up as follows: 16 bits unused, 9 bits each for 4 tree levels (total: 36 bits), and the 12 lowest bits unmodified. The 16 highest bits are required to be equal to 48th bit, or in the other words, the low 48 bits are sign extended to the higher bits. This is done to allow a future expansion of the addressable range, without compromising backwards compatibility.
In the page table, the highest bit is a per-page no-execute bit.
Tanenbaum et al., recently stated that the B5000 (and descendant systems) have no MMU. To understand the functionality provided by an MMU, it is instructive to study a counter example of a system that achieves this functionality by other means.
The B5000 was the first commercial system to support virtual memory after the Atlas. It provides the two functions of an MMU in different ways. Firstly, the mapping of virtual memory addresses. Instead of needing an MMU, the MCP systems are descriptor based. Each allocated memory block is given a master descriptor with the properties of the block, ie., the size, address, and whether present in memory. When a request is made to access the block for reading or writing, the hardware checks its presence via the presence bit (pbit) in the descriptor.
A pbit of 1 indicates the presence of the block. In this case the block can be accessed via the physical address in the descriptor. If the pbit is zero, an interrupt is generated for the MCP (operating system) to make the block present. If the address field is zero, this is the first access to this block and it is allocated (an init pbit). If the address field is non-zero, it is a disk address of the block, which has previously been rolled out, so the block is fetched from disk and the pbit is set to 1 and the physical memory address updated to point to the block in memory (an other pbit). This makes descriptors equivalent to a page-table entry in an MMU system. System performance can be monitored through the number of pbits. Init pbits indicate initial allocations, but a high level of other pbits indicate that the system may be thrashing.
Note that all memory allocation is therefore completely automatic (one of the features of modern systems ) and there is no way to allocate blocks other than this mechanism. There are no such calls as malloc or dealloc, since memory blocks are also automatically discarded. The scheme is also lazy, since a block will not be allocated until it is actually referenced. When memory is near full, the MCP examines the working set, trying compaction (since the system is segmented, not paged), deallocating read-only segments (such as code-segments which can be restored from their original copy), and as a last resort, rolling dirty data segments out to disk.
Secondly, protection. Since all accesses are via the descriptor the hardware can check all accesses are within bounds, and in the case of a write that the process has write permission. The MCP system is inherently secure and thus has no need of an MMU to provide this level of memory protection. Descriptors are read only to user processes and may only be updated by the system (hardware or MCP). (Descriptors have a tag of 5 and odd-tagged words are read only – code words have a tag of 3.)
Blocks can be shared between processes via copy descriptors in the process stack – thus some processes may have write permission, whereas others not. A code segment is read only, thus reentrant and shared between processes. Copy descriptors contain a 20-bit address field giving index of the master descriptor in the master descriptor array. This also implements a very efficient and secure IPC mechanism. Blocks can easily be relocated since only the master descriptor needs update when a block's status changes.
The only other aspect is performance – do MMU- or non-MMU-based systems provide better performance? MCP systems may be implemented on top of standard hardware that does have an MMU (eg., a standard PC). Even if the system implementation uses the MMU in some way, this will not be at all visible at the MCP level.