.. _glossary-b: ============================= Memory Management Glossary: B ============================= .. include:: alphabet.txt .. glossary:: backing store Backing :term:`store (2)` is typically part of a hard disk that is used by a :term:`paging` or :term:`swapping` system to store information not currently in :term:`main memory`. Backing store is slower and cheaper than main memory. Other :term:`storage ` may, less commonly, be used in place of a hard disk (for instance, magnetic tape, floppy disk, or historically, magnetic drum). In general, backing store may mean any locations used to store information when its preferred or natural location is otherwise being used: for example, memory used by a graphical interface to keep a copy of the contents of obscured windows. .. similar:: :term:`swap space`. barrier (1) A barrier is a block on reading from or writing to certain :term:`memory (2)` :term:`locations ` by certain threads or processes. Barriers can be implemented in either software or hardware. Software barriers involve additional instructions around :term:`load` or :term:`store (1)` operations, which would typically be added by a cooperative compiler. Hardware barriers don't require compiler support, and may be implemented on common operating systems by using :term:`memory protection`. .. relevance:: Barriers are used for :term:`incremental ` or :term:`concurrent ` :term:`garbage collection`. .. seealso:: :term:`read barrier`, :term:`write barrier`. .. bibref:: :ref:`Pirinen (1998) `, :ref:`Zorn (1990) `. barrier (2) A memory barrier is an instruction on certain processor architectures that will ensure certain guarantees about the order of accesses to memory. Some processor architectures make very few guarantees about the relative orders of :term:`load` and :term:`store (1)` operations in the instruction stream and the actual order of accesses to :term:`main memory`. These architectures will often have special instructions that make stronger guarantees. For example, the ARM has the ``DMB`` (Data Memory Barrier) instruction: It ensures that all explicit memory accesses that appear in program order before the DMB instruction are observed before any explicit memory accesses that appear in program order after the DMB instruction. These instructions are vital for certain synchronization operations. barrier hit .. see:: :term:`protection fault`. base pointer A *base pointer* is a :term:`pointer` to the base or start of an :term:`object`. This term is commonly used in opposition to :term:`derived pointer`. Note that :ref:`Boehm & Chase (1992) ` define "base pointer" to be "any pointer value directly recognizable by the :term:`collector (1)`", and this may well include :term:`interior pointers`. .. opposite:: :term:`derived pointer`. .. mps:specific:: For objects with :term:`in-band headers`, the MPS distinguishes between the base pointer, which points to the start of the header, and the :term:`client pointer`, which points to the first word after the end of the header. best fit The :term:`allocation policy` that always allocates from the smallest suitable :term:`free block`. Suitable :term:`allocation mechanisms` include :term:`sequential fit` searching for a :term:`perfect fit`, :term:`first fit` on a size-ordered :term:`free block chain`, :term:`segregated fits`, and :term:`indexed fits`. Many :term:`good fit` allocators are also described as :term:`best fit`. In theory, best fit may exhibit bad :term:`fragmentation`, but in practice this is not commonly observed. .. seealso:: :term:`allocation policy`, :term:`first fit`, :term:`sequential fit`. .. bibref:: :ref:`Wilson et al. (1995) `. BIBOP .. aka:: *big bag of pages*. BIBOP, or *BIg Bag Of Pages*, is a technique that encodes :term:`object` type in the high-order bits of their :term:`address`, by using a lookup table that maps from those bits to a type. Despite the name, the blocks involved need not be the size of a :term:`page`. BIBOP requires storing only objects of the same type in a block, but this has the same advantages as :term:`segregated fits` in general. .. historical:: This technique was invented for the PDP-10 MACLISP by JonL White and Stavros Macrakis. It was an advance on earlier techniques that divided the :term:`address space` into contiguous blocks for each type. .. bibref:: :ref:`Baker (1979) `, :ref:`Steele (1977) `. big bag of pages .. see:: :term:`BIBOP`. binary buddies The most common :term:`buddy system` :term:`allocation mechanism`, in which all block sizes are a power of two. Finding a block's buddy is then a matter of flipping the appropriate bit in the block's address. :term:`Internal fragmentation` is usually high, because objects are often not a good fit for power-of-two sized blocks. .. seealso:: :term:`allocation mechanism`, :term:`buddy system`. .. bibref:: :ref:`Wilson et al. (1995) `. bit array .. see:: :term:`bitmap`. bit table .. see:: :term:`bitmap`. bit vector .. see:: :term:`bitmap`. bitmap .. aka:: *bit array*, *bit table*, *bit vector*, *bitset*. A table of bits. .. relevance:: Bitmaps are sometimes used to represent the marks in a :term:`mark-sweep` collector (see :term:`bitmap marking`), or the used memory in a :term:`bitmapped fits` :term:`allocator`. bitmap marking In :term:`mark-sweep` collectors, bitmap marking is a technique for :term:`marking` objects that stores the mark bits for the objects in a contiguous range of memory in a separate :term:`bitmap`. This improves the collector's :term:`locality of reference` and cache performance, because it avoids setting the :term:`dirty bit` on the :term:`pages` containing the marked objects. .. bibref:: :ref:`Zorn (1989) `. bitmapped fit A class of :term:`allocation mechanisms` that use a :term:`bitmap` to represent the usage of the :term:`heap`. Each bit in the map corresponds to a part of the heap, typically a :term:`word`, and is set if that part is in use. Allocation is done by searching the bitmap for a run of clear bits. Bitmapped fit mechanisms have good :term:`locality of reference`, as they avoid examining :term:`in-band headers` when allocating. .. seealso:: :term:`allocation mechanism`, :term:`indexed fit`, :term:`sequential fit`. .. bibref:: :ref:`Wilson et al. (1995) `. bitmask A :term:`bitmap` used to select or exclude a set of bits in another bitmap. bitset .. see:: :term:`bitmap`. black In a :term:`tri-color marking` scheme, black :term:`objects` are objects that have been :term:`scanned `. More precisely, black objects have been noted :term:`reachable` and the :term:`collector (2)` has finished with them and need not visit them again (for the purposes of :term:`tracing `). .. opposite:: :term:`gray`, :term:`white`. blacklisting black-listing A :term:`conservative garbage collector ` can be made more effective by *blacklisting* values which resemble :term:`addresses` that may be :term:`allocated` at in the future, but are known not to be :term:`pointers` . This list is then used to avoid allocation at those addresses. For example, such values can be gathered by scanning the :term:`roots` before any :term:`objects` have been allocated. .. bibref:: :ref:`Boehm (1993) `. block Block is a vague term for an (often contiguous) area of :term:`memory (1)`. Often used to describe :term:`memory (2)` :term:`allocated` by an :term:`allocator` such as :term:`malloc`. .. mps:specific:: The term *block* is used as a general term for a unit of allocation, with *object* being reserved for :term:`formatted objects`. bounds error .. see:: :term:`overwriting error`. boxed Boxed :term:`objects` are represented by a :term:`pointer` to a :term:`block` of :term:`memory (2)` that contains the object data. Sometimes the pointer is :term:`tagged ` to distinguish it from an :term:`unboxed` object, or to represent its type. Only the pointer is duplicated when the object is passed around, so updates to the object are reflected everywhere. .. opposite:: :term:`unboxed`. .. seealso:: :term:`BIBOP`, :term:`tag`. .. bibref:: :ref:`Gudeman (1993) `. break-table A break-table is a data structure used by a :term:`mark-compact` collector to store the :term:`relocation` information. .. seealso:: :term:`mark-compact`. brk ``brk`` is a Unix system call that sets the limit of the data segment. This limit is known as the *break*. ``brk`` and its companion :term:`sbrk` are obsolete on Unix systems that support :term:`virtual memory` and the ``mmap`` system call. The :term:`C` library implementation of :term:`malloc` formerly :term:`allocated` :term:`memory (2)` for the :term:`heap` by extending the data segment using ``brk`` or ``sbrk``. The data segment resided immediately above the program code and :term:`static data ` (the "text segment") in the :term:`address space`. .. figure:: ../diagrams/brk.svg :align: center :alt: Diagram: A simplified view of the address space of a Unix process. A simplified view of the address space of a Unix process. More modern Unix systems use :term:`address space layout randomization` to place these segments at randomized locations in :term:`address space`, so that the :term:`heap` is no longer adjacent to the static data. broken heart :term:`Copying garbage collectors ` :term:`move ` :term:`reachable` :term:`objects` into another :term:`semi-space`. They leave a :term:`forwarding pointer` in the old :term:`location `, pointing to the new. The object at the old location is known as a broken heart. .. similar:: :term:`forwarding pointer`. bucket In a :term:`generational garbage collector `, it is often desirable to divide :term:`generations` by the age of the :term:`object`. These divisions are known as buckets. .. seealso:: :term:`aging space`, :term:`creation space`, :term:`generational garbage collection`. buddy system Buddy systems are a subclass of :term:`strict segregated fit` :term:`allocation mechanisms` which make :term:`splitting ` and :term:`coalescing ` fast by pairing each block with a unique adjacent *buddy* block. There is an array of :term:`free lists`, one for each allowable block size. Allocation rounds up the requested size to an allowable size and allocates from the corresponding free list. If the free list is empty, a larger block is selected and split. A block may only be split into a pair of buddies. A block may only be coalesced with its buddy, and this is only possible if the buddy has not been split into smaller blocks. The advantage of buddy systems is that the buddy of a block being freed can be quickly found by a simple address computation. The disadvantage of buddy systems is that the restricted set of block sizes leads to high :term:`internal fragmentation`, as does the limited ability to coalesce. Different sorts of buddy system are distinguished by the available block sizes and the method of splitting. They include :term:`binary buddies` (the most common), :term:`Fibonacci buddies`, :term:`weighted buddies`, and :term:`double buddies`. .. seealso:: :term:`allocation mechanism`, :term:`segregated fit`, :term:`segregated free lists`, :term:`strict segregated fit`. .. bibref:: :ref:`Wilson et al. (1995) `. buffer A *buffer* is a large :term:`block` of :term:`memory (2)` from which blocks are :term:`allocated` contiguously, as a simple technique for fast :term:`allocation `. By keeping only a *high-water* mark (that is, a :term:`pointer` to the start of unused memory), the buffer technique avoids expensive :term:`in-band headers` and the searching of :term:`free block chains`. Buffers tend to, however, lead to :term:`external fragmentation`. .. bibref:: :ref:`Appel et al. (1988) `. .. mps:specific:: Buffers are implemented using :term:`allocation points` attached to :term:`pools`. bus error Strictly speaking, *a bus error* is a fault on a hardware bus, such as when an invalid :term:`address` is issued. Generally, any hardware exception caused by a :term:`memory (2)` access (for example, :term:`loading ` an :term:`unaligned` :term:`word`) is termed a *bus error*. The term is often used more loosely as a synonym for any memory access error. .. seealso:: :term:`segmentation violation`. byte (1) A unit of storage measurement, equal to 8 bits. It does not matter how the bits are arranged: a byte is just a quantity. This is the sense of byte used in the terms :term:`kilobyte`, :term:`megabyte`, :term:`gigabyte`, :term:`terabyte`, etc. The prefixes in these terms derive from the SI prefixes for powers of 1000, but since powers of two are much more common in binary computers, they are used to denote powers of 1024 (2\ :sup:`10`). .. seealso:: :term:`word`. byte (2) A data type defined by a processor architecture. For example, the smallest :term:`addressable
` :term:`memory location` on the Intel x86 family is the 8-bit byte. .. historical:: The PDP-10 had 36-bit :term:`words`, and defined "byte" to be a general sub-:term:`word` bit-field: compare :term:`byte (3)`. On this machine it was commonplace for characters to be packed four or five to a word using 9- or 7-bit bytes respectively. .. seealso:: :term:`word`. byte (3) A contiguous set of bits used to represent a range of values compactly. The number of bits in a byte is a measure of the information content of the byte. An *n*-bit byte can represent 2\ :sup:`n` distinct values. Bytes may be packed into (or otherwise stored in bit-fields of) integers, words, or other aligned values for space efficiency. byte (4) A data type or storage unit defined by a programming language. In ANSI/ISO :term:`C`, "the unit of data storage large enough to hold the basic character set of the execution environment". In this sense, it is often used synonymously with the C type ``char``. C defines ``sizeof(char)`` to be 1. Many architectures that run C programs equate this sense of byte and :term:`byte (2)`.