tail head cat sleep
QR code linking to this page

Manual Pages  — JEMALLOC

NAME

jemalloc - general purpose memory allocation functions

CONTENTS

LIBRARY

This manual describes jemalloc 5amp;.1amp;.0-0-g61efbda7098de6fe64c362d309824864308c36d4amp;. More information can be found at the jemalloc websiteamp;[1]amp;.

The following configuration options are enabled in libc's built-in jemalloc: --enable-fill, --enable-lazy-lock, --enable-stats, --enable-utrace, --enable-xmalloc, and --with-malloc-conf=abort_conf:falseamp;. Additionally, --enable-debug is enabled in development versions of FreeBSD (controlled by the MK_MALLOC_PRODUCTION make variable)amp;.

SYNOPSIS

#include <stdlibamp;.h>
#include <malloc_npamp;.h>

Standard API

void *malloc(size_t size);
void *calloc(size_t number, size_t size);
int posix_memalign(void **ptr, size_t alignment, size_t size);
void *aligned_alloc(size_t alignment, size_t size);
void *realloc(void *ptr, size_t size);
void free(void *ptr);

Non-standard API


void *mallocx(size_t size, int flags);
void *rallocx(void *ptr, size_t size, int flags);
size_t xallocx(void *ptr, size_t size, size_t extra, int flags);
size_t sallocx(void *ptr, int flags);
void dallocx(void *ptr, int flags);
void sdallocx(void *ptr, size_t size, int flags);
size_t nallocx(size_t size, int flags);
int mallctl(const char *name, void *oldp, size_t *oldlenp, void *newp, size_t newlen);
int mallctlnametomib(const char *name, size_t *mibp, size_t *miblenp);
int mallctlbymib(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp, void *newp, size_t newlen);
void malloc_stats_print(void (*write_cb) (void *, const char *), void *cbopaque, const char *opts);
size_t malloc_usable_size(const void *ptr);
void (*malloc_message)(void *cbopaque, const char *s);

const char *malloc_conf;

DESCRIPTION

Standard API

The malloc() function allocates size bytes of uninitialized memoryamp;. The allocated space is suitably aligned (after possible pointer coercion) for storage of any type of objectamp;.

The calloc() function allocates space for number objects, each size bytes in lengthamp;. The result is identical to calling malloc() with an argument of number * size, with the exception that the allocated memory is explicitly initialized to zero bytesamp;.

The posix_memalign() function allocates size bytes of memory such that the allocation's base address is a multiple of alignment, and returns the allocation in the value pointed to by ptramp;. The requested alignment must be a power of 2 at least as large as sizeof(void *)amp;.

The aligned_alloc() function allocates size bytes of memory such that the allocation's base address is a multiple of alignmentamp;. The requested alignment must be a power of 2amp;. Behavior is undefined if size is not an integral multiple of alignmentamp;.

The realloc() function changes the size of the previously allocated memory referenced by ptr to size bytesamp;. The contents of the memory are unchanged up to the lesser of the new and old sizesamp;. If the new size is larger, the contents of the newly allocated portion of the memory are undefinedamp;. Upon success, the memory referenced by ptr is freed and a pointer to the newly allocated memory is returnedamp;. Note that realloc() may move the memory allocation, resulting in a different return value than ptramp;. If ptr is NULL, the realloc() function behaves identically to malloc() for the specified sizeamp;.

The free() function causes the allocated memory referenced by ptr to be made available for future allocationsamp;. If ptr is NULL, no action occursamp;.

Non-standard API

The mallocx(), rallocx(), xallocx(), sallocx(), dallocx(), sdallocx(), and nallocx() functions all have a flags argument that can be used to specify optionsamp;. The functions only check the options that are contextually relevantamp;. Use bitwise or (|) operations to specify one or more of the following:

MALLOCX_LG_ALIGN(la)

Align the memory allocation to start at an address that is a multiple of (1 << la)amp;. This macro does not validate that la is within the valid rangeamp;.

MALLOCX_ALIGN(a)

Align the memory allocation to start at an address that is a multiple of a, where a is a power of twoamp;. This macro does not validate that a is a power of 2amp;.

MALLOCX_ZERO

Initialize newly allocated memory to contain zero bytesamp;. In the growing reallocation case, the real size prior to reallocation defines the boundary between untouched bytes and those that are initialized to contain zero bytesamp;. If this macro is absent, newly allocated memory is uninitializedamp;.

MALLOCX_TCACHE(tc)

Use the thread-specific cache (tcache) specified by the identifier tc, which must have been acquired via the tcacheamp;.create mallctlamp;. This macro does not validate that tc specifies a valid identifieramp;.

MALLOCX_TCACHE_NONE

Do not use a thread-specific cache (tcache)amp;. Unless MALLOCX_TCACHE(tc) or MALLOCX_TCACHE_NONE is specified, an automatically managed tcache will be used under many circumstancesamp;. This macro cannot be used in the same flags argument as MALLOCX_TCACHE(tc)amp;.

MALLOCX_ARENA(a)

Use the arena specified by the index aamp;. This macro has no effect for regions that were allocated via an arena other than the one specifiedamp;. This macro does not validate that a specifies an arena index in the valid rangeamp;.

The mallocx() function allocates at least size bytes of memory, and returns a pointer to the base address of the allocationamp;. Behavior is undefined if size is 0amp;.

The rallocx() function resizes the allocation at ptr to be at least size bytes, and returns a pointer to the base address of the resulting allocation, which may or may not have moved from its original locationamp;. Behavior is undefined if size is 0amp;.

The xallocx() function resizes the allocation at ptr in place to be at least size bytes, and returns the real size of the allocationamp;. If extra is non-zero, an attempt is made to resize the allocation to be at least (size + extra) bytes, though inability to allocate the extra byte(s) will not by itself result in failure to resizeamp;. Behavior is undefined if size is 0, or if (size + extra > SIZE_T_MAX)amp;.

The sallocx() function returns the real size of the allocation at ptramp;.

The dallocx() function causes the memory referenced by ptr to be made available for future allocationsamp;.

The sdallocx() function is an extension of dallocx() with a size parameter to allow the caller to pass in the allocation size as an optimizationamp;. The minimum valid input size is the original requested size of the allocation, and the maximum valid input size is the corresponding value returned by nallocx() or sallocx()amp;.

The nallocx() function allocates no memory, but it performs the same size computation as the mallocx() function, and returns the real size of the allocation that would result from the equivalent mallocx() function call, or 0 if the inputs exceed the maximum supported size class and/or alignmentamp;. Behavior is undefined if size is 0amp;.

The mallctl() function provides a general interface for introspecting the memory allocator, as well as setting modifiable parameters and triggering actionsamp;. The period-separated name argument specifies a location in a tree-structured namespace; see the MALLCTL NAMESPACE section for documentation on the tree contentsamp;. To read a value, pass a pointer via oldp to adequate space to contain the value, and a pointer to its length via oldlenp; otherwise pass NULL and NULLamp;. Similarly, to write a value, pass a pointer to the value via newp, and its length via newlen; otherwise pass NULL and 0amp;.

The mallctlnametomib() function provides a way to avoid repeated name lookups for applications that repeatedly query the same portion of the namespace, by translating a name to a &#147;Management Information Base&#148; (MIB) that can be passed repeatedly to mallctlbymib()amp;. Upon successful return from mallctlnametomib(), mibp contains an array of *miblenp integers, where *miblenp is the lesser of the number of components in name and the input value of *miblenpamp;. Thus it is possible to pass a *miblenp that is smaller than the number of period-separated name components, which results in a partial MIB that can be used as the basis for constructing a complete MIBamp;. For name components that are integers (eamp;.gamp;. the 2 in arenasamp;.binamp;.2amp;.size), the corresponding MIB component will always be that integeramp;. Therefore, it is legitimate to construct code like the following:

unsigned nbins, i;
size_t mib[4];
size_t len, miblen;

len = sizeof(nbins); mallctl("arenasamp;.nbins", &nbins, &len, NULL, 0);

miblen = 4; mallctlnametomib("arenasamp;.binamp;.0amp;.size", mib, &miblen); for (i = 0; i < nbins; i++) {         size_t bin_size;

        mib[2] = i;         len = sizeof(bin_size);         mallctlbymib(mib, miblen, (void *)&bin_size, &len, NULL, 0);         /* Do something with bin_sizeamp;.amp;.amp;. */ }

The malloc_stats_print() function writes summary statistics via the write_cb callback function pointer and cbopaque data passed to write_cb, or malloc_message() if write_cb is NULLamp;. The statistics are presented in human-readable form unless &#147;J&#148; is specified as a character within the opts string, in which case the statistics are presented in JSON formatamp;[2]amp;. This function can be called repeatedlyamp;. General information that never changes during execution can be omitted by specifying &#147;g&#148; as a character within the opts stringamp;. Note that malloc_message() uses the mallctl*() functions internally, so inconsistent statistics can be reported if multiple threads use these functions simultaneouslyamp;. If --enable-stats is specified during configuration, &#147;m&#148;, &#147;d&#148;, and &#147;a&#148; can be specified to omit merged arena, destroyed merged arena, and per arena statistics, respectively; &#147;b&#148; and &#147;l&#148; can be specified to omit per size class statistics for bins and large objects, respectively; &#147;x&#148; can be specified to omit all mutex statisticsamp;. Unrecognized characters are silently ignoredamp;. Note that thread caching may prevent some statistics from being completely up to date, since extra locking would be required to merge counters that track thread cache operationsamp;.

The malloc_usable_size() function returns the usable size of the allocation pointed to by ptramp;. The return value may be larger than the size that was requested during allocationamp;. The malloc_usable_size() function is not a mechanism for in-place realloc(); rather it is provided solely as a tool for introspection purposesamp;. Any discrepancy between the requested allocation size and the size reported by malloc_usable_size() should not be depended on, since such behavior is entirely implementation-dependentamp;.

TUNING

Once, when the first call is made to one of the memory allocation routines, the allocator initializes its internals based in part on various options that can be specified at compile- or run-timeamp;.

The string specified via --with-malloc-conf, the string pointed to by the global variable malloc_conf, the &#147;name&#148; of the file referenced by the symbolic link named /etc/mallocamp;.conf, and the value of the environment variable MALLOC_CONF, will be interpreted, in that order, from left to right as optionsamp;. Note that malloc_conf may be read before main() is entered, so the declaration of malloc_conf should specify an initializer that contains the final value to be read by jemallocamp;. --with-malloc-conf and malloc_conf are compile-time mechanisms, whereas /etc/mallocamp;.conf and MALLOC_CONF can be safely set any time prior to program invocationamp;.

An options string is a comma-separated list of option:value pairsamp;. There is one key corresponding to each optamp;.* mallctl (see the MALLCTL NAMESPACE section for options documentation)amp;. For example, abort:true,narenas:1 sets the optamp;.abort and optamp;.narenas optionsamp;. Some options have boolean values (true/false), others have integer values (base 8, 10, or 16, depending on prefix), and yet others have raw string valuesamp;.

IMPLEMENTATION NOTES

Traditionally, allocators have used sbrk(2) to obtain memory, which is suboptimal for several reasons, including race conditions, increased fragmentation, and artificial limitations on maximum usable memoryamp;. If sbrk(2) is supported by the operating system, this allocator uses both mmap(2) and sbrk(2), in that order of preference; otherwise only mmap(2) is usedamp;.

This allocator uses multiple arenas in order to reduce lock contention for threaded programs on multi-processor systemsamp;. This works well with regard to threading scalability, but incurs some costsamp;. There is a small fixed per-arena overhead, and additionally, arenas manage memory completely independently of each other, which means a small fixed increase in overall memory fragmentationamp;. These overheads are not generally an issue, given the number of arenas normally usedamp;. Note that using substantially more arenas than the default is not likely to improve performance, mainly due to reduced cache performanceamp;. However, it may make sense to reduce the number of arenas if an application does not make much use of the allocation functionsamp;.

In addition to multiple arenas, this allocator supports thread-specific caching, in order to make it possible to completely avoid synchronization for most allocation requestsamp;. Such caching allows very fast allocation in the common case, but it increases memory usage and fragmentation, since a bounded number of objects can remain allocated in each thread cacheamp;.

Memory is conceptually broken into extentsamp;. Extents are always aligned to multiples of the page sizeamp;. This alignment makes it possible to find metadata for user objects quicklyamp;. User objects are broken into two categories according to size: small and largeamp;. Contiguous small objects comprise a slab, which resides within a single extent, whereas large objects each have their own extents backing themamp;.

Small objects are managed in groups by slabsamp;. Each slab maintains a bitmap to track which regions are in useamp;. Allocation requests that are no more than half the quantum (8 or 16, depending on architecture) are rounded up to the nearest power of two that is at least sizeof(double)amp;. All other object size classes are multiples of the quantum, spaced such that there are four size classes for each doubling in size, which limits internal fragmentation to approximately 20% for all but the smallest size classesamp;. Small size classes are smaller than four times the page size, and large size classes extend from four times the page size up to the largest size class that does not exceed PTRDIFF_MAXamp;.

Allocations are packed tightly together, which can be an issue for multi-threaded applicationsamp;. If you need to assure that allocations do not suffer from cacheline sharing, round your allocation requests up to the nearest multiple of the cacheline size, or specify cacheline alignment when allocatingamp;.

The realloc(), rallocx(), and xallocx() functions may resize allocations without moving them under limited circumstancesamp;. Unlike the *allocx() API, the standard API does not officially round up the usable size of an allocation to the nearest size class, so technically it is necessary to call realloc() to grow eamp;.gamp;. a 9-byte allocation to 16 bytes, or shrink a 16-byte allocation to 9 bytesamp;. Growth and shrinkage trivially succeeds in place as long as the pre-size and post-size both round up to the same size classamp;. No other API guarantees are made regarding in-place resizing, but the current implementation also tries to resize large allocations in place, as long as the pre-size and post-size are both largeamp;. For shrinkage to succeed, the extent allocator must support splitting (see arenaamp;.<i>amp;.extent_hooks)amp;. Growth only succeeds if the trailing memory is currently available, and the extent allocator supports mergingamp;.

Assuming 4 KiB pages and a 16-byte quantum on a 64-bit system, the size classes in each category are as shown in Table 1amp;.

Table amp;1. amp;Size classes

Category Spacing Size
l r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l
l r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l
^ r l.
Small lg [8]
 16 [16, 32, 48, 64, 80, 96, 112, 128]
 32 [160, 192, 224, 256]
 64 [320, 384, 448, 512]
 128 [640, 768, 896, 1024]
 256 [1280, 1536, 1792, 2048]
 512 [2560, 3072, 3584, 4096]
 1 KiB [5 KiB, 6 KiB, 7 KiB, 8 KiB]
 2 KiB [10 KiB, 12 KiB, 14 KiB]
Large 2 KiB [16 KiB]
 4 KiB [20 KiB, 24 KiB, 28 KiB, 32 KiB]
 8 KiB [40 KiB, 48 KiB, 54 KiB, 64 KiB]
 16 KiB [80 KiB, 96 KiB, 112 KiB, 128 KiB]
 32 KiB [160 KiB, 192 KiB, 224 KiB, 256 KiB]
 64 KiB [320 KiB, 384 KiB, 448 KiB, 512 KiB]
 128 KiB [640 KiB, 768 KiB, 896 KiB, 1 MiB]
 256 KiB [1280 KiB, 1536 KiB, 1792 KiB, 2 MiB]
 512 KiB [2560 KiB, 3 MiB, 3584 KiB, 4 MiB]
 1 MiB [5 MiB, 6 MiB, 7 MiB, 8 MiB]
 2 MiB [10 MiB, 12 MiB, 14 MiB, 16 MiB]
 4 MiB [20 MiB, 24 MiB, 28 MiB, 32 MiB]
 8 MiB [40 MiB, 48 MiB, 56 MiB, 64 MiB]
 amp;.amp;.amp;. amp;.amp;.amp;.
 512 PiB [2560 PiB, 3 EiB, 3584 PiB, 4 EiB]
 1 EiB [5 EiB, 6 EiB, 7 EiB]

MALLCTL NAMESPACE

The following names are defined in the namespace accessible via the mallctl*() functionsamp;. Value types are specified in parentheses, their readable/writable statuses are encoded as rw, r-, -w, or --, and required build configuration flags follow, if anyamp;. A name element encoded as <i> or <j> indicates an integer component, where the integer varies from 0 to some upper value that must be determined via introspectionamp;. In the case of statsamp;.arenasamp;.<i>amp;.* and arenaamp;.<i>amp;.{initialized,purge,decay,dss}, <i> equal to MALLCTL_ARENAS_ALL can be used to operate on all arenas or access the summation of statistics from all arenas; similarly <i> equal to MALLCTL_ARENAS_DESTROYED can be used to access the summation of statistics from all destroyed arenasamp;. These constants can be utilized either via mallctlnametomib() followed by mallctlbymib(), or via code such as the following:

#define STRINGIFY_HELPER(x) #x
#define STRINGIFY(x) STRINGIFY_HELPER(x)

mallctl("arenaamp;." STRINGIFY(MALLCTL_ARENAS_ALL) "amp;.decay", NULL, NULL, NULL, 0);

Take special note of the epoch mallctl, which controls refreshing of cached dynamic statisticsamp;.

version (const char *) r-

Return the jemalloc version stringamp;.

epoch (uint64_t) rw

If a value is passed in, refresh the data from which the mallctl*() functions report values, and increment the epochamp;. Return the current epochamp;. This is useful for detecting whether another thread caused a refreshamp;.

background_thread (bool) rw

Enable/disable internal background worker threadsamp;. When set to true, background threads are created on demand (the number of background threads will be no more than the number of CPUs or active arenas)amp;. Threads run periodically, and handle purging asynchronouslyamp;. When switching off, background threads are terminated synchronouslyamp;. Note that after fork(2) function, the state in the child process will be disabled regardless the state in parent processamp;. See statsamp;.background_thread for related statsamp;. optamp;.background_thread can be used to set the default optionamp;. This option is only available on selected pthread-based platformsamp;.

max_background_threads (size_t) rw

Maximum number of background worker threads that will be createdamp;. This value is capped at optamp;.max_background_threads at startupamp;.

configamp;.cache_oblivious (bool) r-

--enable-cache-oblivious was specified during build configurationamp;.

configamp;.debug (bool) r-

--enable-debug was specified during build configurationamp;.

configamp;.fill (bool) r-

--enable-fill was specified during build configurationamp;.

configamp;.lazy_lock (bool) r-

--enable-lazy-lock was specified during build configurationamp;.

configamp;.malloc_conf (const char *) r-

Embedded configure-time-specified run-time options string, empty unless --with-malloc-conf was specified during build configurationamp;.

configamp;.prof (bool) r-

--enable-prof was specified during build configurationamp;.

configamp;.prof_libgcc (bool) r-

--disable-prof-libgcc was not specified during build configurationamp;.

configamp;.prof_libunwind (bool) r-

--enable-prof-libunwind was specified during build configurationamp;.

configamp;.stats (bool) r-

--enable-stats was specified during build configurationamp;.

configamp;.utrace (bool) r-

--enable-utrace was specified during build configurationamp;.

configamp;.xmalloc (bool) r-

--enable-xmalloc was specified during build configurationamp;.

optamp;.abort (bool) r-

Abort-on-warning enabled/disabledamp;. If true, most warnings are fatalamp;. Note that runtime option warnings are not included (see optamp;.abort_conf for that)amp;. The process will call abort(3) in these casesamp;. This option is disabled by default unless --enable-debug is specified during configuration, in which case it is enabled by defaultamp;.

optamp;.abort_conf (bool) r-

Abort-on-invalid-configuration enabled/disabledamp;. If true, invalid runtime options are fatalamp;. The process will call abort(3) in these casesamp;. This option is disabled by default unless --enable-debug is specified during configuration, in which case it is enabled by defaultamp;.

optamp;.metadata_thp (const char *) r-

Controls whether to allow jemalloc to use transparent huge page (THP) for internal metadata (see statsamp;.metadata)amp;. &#147;always&#148; allows such usageamp;. &#147;auto&#148; uses no THP initially, but may begin to do so when metadata usage reaches certain levelamp;. The default is &#147;disabled&#148;amp;.

optamp;.retain (bool) r-

If true, retain unused virtual memory for later reuse rather than discarding it by calling munmap(2) or equivalent (see statsamp;.retained for related details)amp;. This option is disabled by default unless discarding virtual memory is known to trigger platform-specific performance problems, eamp;.gamp;. for [64-bit] Linux, which has a quirk in its virtual memory allocation algorithm that causes semi-permanent VM map holes under normal jemalloc operationamp;. Although munmap(2) causes issues on 32-bit Linux as well, retaining virtual memory for 32-bit Linux is disabled by default due to the practical possibility of address space exhaustionamp;.

optamp;.dss (const char *) r-

dss (sbrk(2)) allocation precedence as related to mmap(2) allocationamp;. The following settings are supported if sbrk(2) is supported by the operating system: &#147;disabled&#148;, &#147;primary&#148;, and &#147;secondary&#148;; otherwise only &#147;disabled&#148; is supportedamp;. The default is &#147;secondary&#148; if sbrk(2) is supported by the operating system; &#147;disabled&#148; otherwiseamp;.

optamp;.narenas (unsigned) r-

Maximum number of arenas to use for automatic multiplexing of threads and arenasamp;. The default is four times the number of CPUs, or one if there is a single CPUamp;.

optamp;.percpu_arena (const char *) r-

Per CPU arena modeamp;. Use the &#147;percpu&#148; setting to enable this feature, which uses number of CPUs to determine number of arenas, and bind threads to arenas dynamically based on the CPU the thread runs on currentlyamp;. &#147;phycpu&#148; setting uses one arena per physical CPU, which means the two hyper threads on the same CPU share one arenaamp;. Note that no runtime checking regarding the availability of hyper threading is done at the momentamp;. When set to &#147;disabled&#148;, narenas and thread to arena association will not be impacted by this optionamp;. The default is &#147;disabled&#148;amp;.

optamp;.background_thread (const bool) r-

Internal background worker threads enabled/disabledamp;. Because of potential circular dependencies, enabling background thread using this option may cause crash or deadlock during initializationamp;. For a reliable way to use this feature, see background_thread for dynamic control options and detailsamp;. This option is disabled by defaultamp;.

optamp;.max_background_threads (const size_t) r-

Maximum number of background threads that will be created if background_thread is setamp;. Defaults to number of cpusamp;.

optamp;.dirty_decay_ms (ssize_t) r-

Approximate time in milliseconds from the creation of a set of unused dirty pages until an equivalent set of unused dirty pages is purged (iamp;.eamp;. converted to muzzy via eamp;.gamp;. madvise(amp;.amp;.amp;.MADV_FREE) if supported by the operating system, or converted to clean otherwise) and/or reusedamp;. Dirty pages are defined as previously having been potentially written to by the application, and therefore consuming physical memory, yet having no current useamp;. The pages are incrementally purged according to a sigmoidal decay curve that starts and ends with zero purge rateamp;. A decay time of 0 causes all unused dirty pages to be purged immediately upon creationamp;. A decay time of -1 disables purgingamp;. The default decay time is 10 secondsamp;. See arenasamp;.dirty_decay_ms and arenaamp;.<i>amp;.dirty_decay_ms for related dynamic control optionsamp;. See optamp;.muzzy_decay_ms for a description of muzzy pagesamp;.

optamp;.muzzy_decay_ms (ssize_t) r-

Approximate time in milliseconds from the creation of a set of unused muzzy pages until an equivalent set of unused muzzy pages is purged (iamp;.eamp;. converted to clean) and/or reusedamp;. Muzzy pages are defined as previously having been unused dirty pages that were subsequently purged in a manner that left them subject to the reclamation whims of the operating system (eamp;.gamp;. madvise(amp;.amp;.amp;.MADV_FREE)), and therefore in an indeterminate stateamp;. The pages are incrementally purged according to a sigmoidal decay curve that starts and ends with zero purge rateamp;. A decay time of 0 causes all unused muzzy pages to be purged immediately upon creationamp;. A decay time of -1 disables purgingamp;. The default decay time is 10 secondsamp;. See arenasamp;.muzzy_decay_ms and arenaamp;.<i>amp;.muzzy_decay_ms for related dynamic control optionsamp;.

optamp;.lg_extent_max_active_fit (size_t) r-

When reusing dirty extents, this determines the (log base 2 of the) maximum ratio between the size of the active extent selected (to split off from) and the size of the requested allocationamp;. This prevents the splitting of large active extents for smaller allocations, which can reduce fragmentation over the long run (especially for non-active extents)amp;. Lower value may reduce fragmentation, at the cost of extra active extentsamp;. The default value is 6, which gives a maximum ratio of 64 (2^6)amp;.

optamp;.stats_print (bool) r-

Enable/disable statistics printing at exitamp;. If enabled, the malloc_stats_print() function is called at program exit via an atexit(3) functionamp;. optamp;.stats_print_opts can be combined to specify output optionsamp;. If --enable-stats is specified during configuration, this has the potential to cause deadlock for a multi-threaded process that exits while one or more threads are executing in the memory allocation functionsamp;. Furthermore, atexit() may allocate memory during application initialization and then deadlock internally when jemalloc in turn calls atexit(), so this option is not universally usable (though the application can register its own atexit() function with equivalent functionality)amp;. Therefore, this option should only be used with care; it is primarily intended as a performance tuning aid during application developmentamp;. This option is disabled by defaultamp;.

optamp;.stats_print_opts (const char *) r-

Options (the opts string) to pass to the malloc_stats_print() at exit (enabled through optamp;.stats_print)amp;. See available options in malloc_stats_print()amp;. Has no effect unless optamp;.stats_print is enabledamp;. The default is &#147;&#148;amp;.

optamp;.junk (const char *) r- [--enable-fill]

Junk fillingamp;. If set to &#147;alloc&#148;, each byte of uninitialized allocated memory will be initialized to 0xa5amp;. If set to &#147;free&#148;, all deallocated memory will be initialized to 0x5aamp;. If set to &#147;true&#148;, both allocated and deallocated memory will be initialized, and if set to &#147;false&#148;, junk filling be disabled entirelyamp;. This is intended for debugging and will impact performance negativelyamp;. This option is &#147;false&#148; by default unless --enable-debug is specified during configuration, in which case it is &#147;true&#148; by defaultamp;.

optamp;.zero (bool) r- [--enable-fill]

Zero filling enabled/disabledamp;. If enabled, each byte of uninitialized allocated memory will be initialized to 0amp;. Note that this initialization only happens once for each byte, so realloc() and rallocx() calls do not zero memory that was previously allocatedamp;. This is intended for debugging and will impact performance negativelyamp;. This option is disabled by defaultamp;.

optamp;.utrace (bool) r- [--enable-utrace]

Allocation tracing based on utrace(2) enabled/disabledamp;. This option is disabled by defaultamp;.

optamp;.xmalloc (bool) r- [--enable-xmalloc]

Abort-on-out-of-memory enabled/disabledamp;. If enabled, rather than returning failure for any allocation function, display a diagnostic message on STDERR_FILENO and cause the program to drop core (using abort(3))amp;. If an application is designed to depend on this behavior, set the option at compile time by including the following in the source code:

malloc_conf = "xmalloc:true";

This option is disabled by defaultamp;.

optamp;.tcache (bool) r-

Thread-specific caching (tcache) enabled/disabledamp;. When there are multiple threads, each thread uses a tcache for objects up to a certain sizeamp;. Thread-specific caching allows many allocations to be satisfied without performing any thread synchronization, at the cost of increased memory useamp;. See the optamp;.lg_tcache_max option for related tuning informationamp;. This option is enabled by defaultamp;.

optamp;.lg_tcache_max (size_t) r-

Maximum size class (log base 2) to cache in the thread-specific cache (tcache)amp;. At a minimum, all small size classes are cached, and at a maximum all large size classes are cachedamp;. The default maximum is 32 KiB (2^15)amp;.

optamp;.thp (const char *) r-

Transparent hugepage (THP) modeamp;. Settings "always", "never" and "default" are available if THP is supported by the operating systemamp;. The "always" setting enables transparent hugepage for all user memory mappings with MADV_HUGEPAGE; "never" ensures no transparent hugepage with MADV_NOHUGEPAGE; the default setting "default" makes no changesamp;. Note that: this option does not affect THP for jemalloc internal metadata (see optamp;.metadata_thp); in addition, for arenas with customized extent_hooks, this option is bypassed as it is implemented as part of the default extent hooksamp;.

optamp;.prof (bool) r- [--enable-prof]

Memory profiling enabled/disabledamp;. If enabled, profile memory allocation activityamp;. See the optamp;.prof_active option for on-the-fly activation/deactivationamp;. See the optamp;.lg_prof_sample option for probabilistic sampling controlamp;. See the optamp;.prof_accum option for control of cumulative sample reportingamp;. See the optamp;.lg_prof_interval option for information on interval-triggered profile dumping, the optamp;.prof_gdump option for information on high-water-triggered profile dumping, and the optamp;.prof_final option for final profile dumpingamp;. Profile output is compatible with the jeprof command, which is based on the pprof that is developed as part of the gperftools packageamp;[3]amp;. See HEAP PROFILE FORMAT for heap profile format documentationamp;.

optamp;.prof_prefix (const char *) r- [--enable-prof]

Filename prefix for profile dumpsamp;. If the prefix is set to the empty string, no automatic dumps will occur; this is primarily useful for disabling the automatic final heap dump (which also disables leak reporting, if enabled)amp;. The default prefix is jeprofamp;.

optamp;.prof_active (bool) r- [--enable-prof]

Profiling activated/deactivatedamp;. This is a secondary control mechanism that makes it possible to start the application with profiling enabled (see the optamp;.prof option) but inactive, then toggle profiling at any time during program execution with the profamp;.active mallctlamp;. This option is enabled by defaultamp;.

optamp;.prof_thread_active_init (bool) r- [--enable-prof]

Initial setting for threadamp;.profamp;.active in newly created threadsamp;. The initial setting for newly created threads can also be changed during execution via the profamp;.thread_active_init mallctlamp;. This option is enabled by defaultamp;.

optamp;.lg_prof_sample (size_t) r- [--enable-prof]

Average interval (log base 2) between allocation samples, as measured in bytes of allocation activityamp;. Increasing the sampling interval decreases profile fidelity, but also decreases the computational overheadamp;. The default sample interval is 512 KiB (2^19 B)amp;.

optamp;.prof_accum (bool) r- [--enable-prof]

Reporting of cumulative object/byte counts in profile dumps enabled/disabledamp;. If this option is enabled, every unique backtrace must be stored for the duration of executionamp;. Depending on the application, this can impose a large memory overhead, and the cumulative counts are not always of interestamp;. This option is disabled by defaultamp;.

optamp;.lg_prof_interval (ssize_t) r- [--enable-prof]

Average interval (log base 2) between memory profile dumps, as measured in bytes of allocation activityamp;. The actual interval between dumps may be sporadic because decentralized allocation counters are used to avoid synchronization bottlenecksamp;. Profiles are dumped to files named according to the pattern <prefix>amp;.<pid>amp;.<seq>amp;.i<iseq>amp;.heap, where <prefix> is controlled by the optamp;.prof_prefix optionamp;. By default, interval-triggered profile dumping is disabled (encoded as -1)amp;.

optamp;.prof_gdump (bool) r- [--enable-prof]

Set the initial state of profamp;.gdump, which when enabled triggers a memory profile dump every time the total virtual memory exceeds the previous maximumamp;. This option is disabled by defaultamp;.

optamp;.prof_final (bool) r- [--enable-prof]

Use an atexit(3) function to dump final memory usage to a file named according to the pattern <prefix>amp;.<pid>amp;.<seq>amp;.famp;.heap, where <prefix> is controlled by the optamp;.prof_prefix optionamp;. Note that atexit() may allocate memory during application initialization and then deadlock internally when jemalloc in turn calls atexit(), so this option is not universally usable (though the application can register its own atexit() function with equivalent functionality)amp;. This option is disabled by defaultamp;.

optamp;.prof_leak (bool) r- [--enable-prof]

Leak reporting enabled/disabledamp;. If enabled, use an atexit(3) function to report memory leaks detected by allocation samplingamp;. See the optamp;.prof option for information on analyzing heap profile outputamp;. This option is disabled by defaultamp;.

threadamp;.arena (unsigned) rw

Get or set the arena associated with the calling threadamp;. If the specified arena was not initialized beforehand (see the arenaamp;.iamp;.initialized mallctl), it will be automatically initialized as a side effect of calling this interfaceamp;.

threadamp;.allocated (uint64_t) r- [--enable-stats]

Get the total number of bytes ever allocated by the calling threadamp;. This counter has the potential to wrap around; it is up to the application to appropriately interpret the counter in such casesamp;.

threadamp;.allocatedp (uint64_t *) r- [--enable-stats]

Get a pointer to the the value that is returned by the threadamp;.allocated mallctlamp;. This is useful for avoiding the overhead of repeated mallctl*() callsamp;.

threadamp;.deallocated (uint64_t) r- [--enable-stats]

Get the total number of bytes ever deallocated by the calling threadamp;. This counter has the potential to wrap around; it is up to the application to appropriately interpret the counter in such casesamp;.

threadamp;.deallocatedp (uint64_t *) r- [--enable-stats]

Get a pointer to the the value that is returned by the threadamp;.deallocated mallctlamp;. This is useful for avoiding the overhead of repeated mallctl*() callsamp;.

threadamp;.tcacheamp;.enabled (bool) rw

Enable/disable calling thread's tcacheamp;. The tcache is implicitly flushed as a side effect of becoming disabled (see threadamp;.tcacheamp;.flush)amp;.

threadamp;.tcacheamp;.flush (void) --

Flush calling thread's thread-specific cache (tcache)amp;. This interface releases all cached objects and internal data structures associated with the calling thread's tcacheamp;. Ordinarily, this interface need not be called, since automatic periodic incremental garbage collection occurs, and the thread cache is automatically discarded when a thread exitsamp;. However, garbage collection is triggered by allocation activity, so it is possible for a thread that stops allocating/deallocating to retain its cache indefinitely, in which case the developer may find manual flushing usefulamp;.

threadamp;.profamp;.name (const char *) r- or -w [--enable-prof]

Get/set the descriptive name associated with the calling thread in memory profile dumpsamp;. An internal copy of the name string is created, so the input string need not be maintained after this interface completes executionamp;. The output string of this interface should be copied for non-ephemeral uses, because multiple implementation details can cause asynchronous string deallocationamp;. Furthermore, each invocation of this interface can only read or write; simultaneous read/write is not supported due to string lifetime limitationsamp;. The name string must be nil-terminated and comprised only of characters in the sets recognized by isgraph(3) and isblank(3)amp;.

threadamp;.profamp;.active (bool) rw [--enable-prof]

Control whether sampling is currently active for the calling threadamp;. This is an activation mechanism in addition to profamp;.active; both must be active for the calling thread to sampleamp;. This flag is enabled by defaultamp;.

tcacheamp;.create (unsigned) r-

Create an explicit thread-specific cache (tcache) and return an identifier that can be passed to the MALLOCX_TCACHE(tc) macro to explicitly use the specified cache rather than the automatically managed one that is used by defaultamp;. Each explicit cache can be used by only one thread at a time; the application must assure that this constraint holdsamp;.

tcacheamp;.flush (unsigned) -w

Flush the specified thread-specific cache (tcache)amp;. The same considerations apply to this interface as to threadamp;.tcacheamp;.flush, except that the tcache will never be automatically discardedamp;.

tcacheamp;.destroy (unsigned) -w

Flush the specified thread-specific cache (tcache) and make the identifier available for use during a future tcache creationamp;.

arenaamp;.<i>amp;.initialized (bool) r-

Get whether the specified arena's statistics are initialized (iamp;.eamp;. the arena was initialized prior to the current epoch)amp;. This interface can also be nominally used to query whether the merged statistics corresponding to MALLCTL_ARENAS_ALL are initialized (always true)amp;.

arenaamp;.<i>amp;.decay (void) --

Trigger decay-based purging of unused dirty/muzzy pages for arena <i>, or for all arenas if <i> equals MALLCTL_ARENAS_ALLamp;. The proportion of unused dirty/muzzy pages to be purged depends on the current time; see optamp;.dirty_decay_ms and optamp;.muzy_decay_ms for detailsamp;.

arenaamp;.<i>amp;.purge (void) --

Purge all unused dirty pages for arena <i>, or for all arenas if <i> equals MALLCTL_ARENAS_ALLamp;.

arenaamp;.<i>amp;.reset (void) --

Discard all of the arena's extant allocationsamp;. This interface can only be used with arenas explicitly created via arenasamp;.createamp;. None of the arena's discarded/cached allocations may accessed afterwardamp;. As part of this requirement, all thread caches which were used to allocate/deallocate in conjunction with the arena must be flushed beforehandamp;.

arenaamp;.<i>amp;.destroy (void) --

Destroy the arenaamp;. Discard all of the arena's extant allocations using the same mechanism as for arenaamp;.<i>amp;.reset (with all the same constraints and side effects), merge the arena stats into those accessible at arena index MALLCTL_ARENAS_DESTROYED, and then completely discard all metadata associated with the arenaamp;. Future calls to arenasamp;.create may recycle the arena indexamp;. Destruction will fail if any threads are currently associated with the arena as a result of calls to threadamp;.arenaamp;.

arenaamp;.<i>amp;.dss (const char *) rw

Set the precedence of dss allocation as related to mmap allocation for arena <i>, or for all arenas if <i> equals MALLCTL_ARENAS_ALLamp;. See optamp;.dss for supported settingsamp;.

arenaamp;.<i>amp;.dirty_decay_ms (ssize_t) rw

Current per-arena approximate time in milliseconds from the creation of a set of unused dirty pages until an equivalent set of unused dirty pages is purged and/or reusedamp;. Each time this interface is set, all currently unused dirty pages are considered to have fully decayed, which causes immediate purging of all unused dirty pages unless the decay time is set to -1 (iamp;.eamp;. purging disabled)amp;. See optamp;.dirty_decay_ms for additional informationamp;.

arenaamp;.<i>amp;.muzzy_decay_ms (ssize_t) rw

Current per-arena approximate time in milliseconds from the creation of a set of unused muzzy pages until an equivalent set of unused muzzy pages is purged and/or reusedamp;. Each time this interface is set, all currently unused muzzy pages are considered to have fully decayed, which causes immediate purging of all unused muzzy pages unless the decay time is set to -1 (iamp;.eamp;. purging disabled)amp;. See optamp;.muzzy_decay_ms for additional informationamp;.

arenaamp;.<i>amp;.retain_grow_limit (size_t) rw

Maximum size to grow retained region (only relevant when optamp;.retain is enabled)amp;. This controls the maximum increment to expand virtual memory, or allocation through arenaamp;.<i>extent_hooksamp;. In particular, if customized extent hooks reserve physical memory (eamp;.gamp;. 1G huge pages), this is useful to control the allocation hook's input sizeamp;. The default is no limitamp;.

arenaamp;.<i>amp;.extent_hooks (extent_hooks_t *) rw

Get or set the extent management hook functions for arena <i>amp;. The functions must be capable of operating on all extant extents associated with arena <i>, usually by passing unknown extents to the replaced functionsamp;. In practice, it is feasible to control allocation for arenas explicitly created via arenasamp;.create such that all extents originate from an application-supplied extent allocator (by specifying the custom extent hook functions during arena creation), but the automatically created arenas will have already created extents prior to the application having an opportunity to take over extent allocationamp;.

typedef extent_hooks_s extent_hooks_t;
struct extent_hooks_s {
        extent_alloc_t          *alloc;
        extent_dalloc_t         *dalloc;
        extent_destroy_t        *destroy;
        extent_commit_t         *commit;
        extent_decommit_t       *decommit;
        extent_purge_t          *purge_lazy;
        extent_purge_t          *purge_forced;
        extent_split_t          *split;
        extent_merge_t          *merge;
};

The extent_hooks_t structure comprises function pointers which are described individually belowamp;. jemalloc uses these functions to manage extent lifetime, which starts off with allocation of mapped committed memory, in the simplest case followed by deallocationamp;. However, there are performance and platform reasons to retain extents for later reuseamp;. Cleanup attempts cascade from deallocation to decommit to forced purging to lazy purging, which gives the extent management functions opportunities to reject the most permanent cleanup operations in favor of less permanent (and often less costly) operationsamp;. All operations except allocation can be universally opted out of by setting the hook pointers to NULL, or selectively opted out of by returning failureamp;. Note that once the extent hook is set, the structure is accessed directly by the associated arenas, so it must remain valid for the entire lifetime of the arenasamp;.
typedef void *(extent_alloc_t)(extent_hooks_t *extent_hooks, void *new_addr, size_t size, size_t alignment, bool *zero, bool *commit, unsigned arena_ind);

 

An extent allocation function conforms to the extent_alloc_t type and upon success returns a pointer to size bytes of mapped memory on behalf of arena arena_ind such that the extent's base address is a multiple of alignment, as well as setting *zero to indicate whether the extent is zeroed and *commit to indicate whether the extent is committedamp;. Upon error the function returns NULL and leaves *zero and *commit unmodifiedamp;. The size parameter is always a multiple of the page sizeamp;. The alignment parameter is always a power of two at least as large as the page sizeamp;. Zeroing is mandatory if *zero is true upon function entryamp;. Committing is mandatory if *commit is true upon function entryamp;. If new_addr is not NULL, the returned pointer must be new_addr on success or NULL on erroramp;. Committed memory may be committed in absolute terms as on a system that does not overcommit, or in implicit terms as on a system that overcommits and satisfies physical memory needs on demand via soft page faultsamp;. Note that replacing the default extent allocation function makes the arena's arenaamp;.<i>amp;.dss setting irrelevantamp;.
typedef bool (extent_dalloc_t)(extent_hooks_t *extent_hooks, void *addr, size_t size, bool committed, unsigned arena_ind);

 

An extent deallocation function conforms to the extent_dalloc_t type and deallocates an extent at given addr and size with committed/decommited memory as indicated, on behalf of arena arena_ind, returning false upon successamp;. If the function returns true, this indicates opt-out from deallocation; the virtual memory mapping associated with the extent remains mapped, in the same commit state, and available for future use, in which case it will be automatically retained for later reuseamp;.
typedef void (extent_destroy_t)(extent_hooks_t *extent_hooks, void *addr, size_t size, bool committed, unsigned arena_ind);

 

An extent destruction function conforms to the extent_destroy_t type and unconditionally destroys an extent at given addr and size with committed/decommited memory as indicated, on behalf of arena arena_indamp;. This function may be called to destroy retained extents during arena destruction (see arenaamp;.<i>amp;.destroy)amp;.
typedef bool (extent_commit_t)(extent_hooks_t *extent_hooks, void *addr, size_t size, size_t offset, size_t length, unsigned arena_ind);

 

An extent commit function conforms to the extent_commit_t type and commits zeroed physical memory to back pages within an extent at given addr and size at offset bytes, extending for length on behalf of arena arena_ind, returning false upon successamp;. Committed memory may be committed in absolute terms as on a system that does not overcommit, or in implicit terms as on a system that overcommits and satisfies physical memory needs on demand via soft page faultsamp;. If the function returns true, this indicates insufficient physical memory to satisfy the requestamp;.
typedef bool (extent_decommit_t)(extent_hooks_t *extent_hooks, void *addr, size_t size, size_t offset, size_t length, unsigned arena_ind);

 

An extent decommit function conforms to the extent_decommit_t type and decommits any physical memory that is backing pages within an extent at given addr and size at offset bytes, extending for length on behalf of arena arena_ind, returning false upon success, in which case the pages will be committed via the extent commit function before being reusedamp;. If the function returns true, this indicates opt-out from decommit; the memory remains committed and available for future use, in which case it will be automatically retained for later reuseamp;.
typedef bool (extent_purge_t)(extent_hooks_t *extent_hooks, void *addr, size_t size, size_t offset, size_t length, unsigned arena_ind);

 

An extent purge function conforms to the extent_purge_t type and discards physical pages within the virtual memory mapping associated with an extent at given addr and size at offset bytes, extending for length on behalf of arena arena_indamp;. A lazy extent purge function (eamp;.gamp;. implemented via madvise(amp;.amp;.amp;.MADV_FREE)) can delay purging indefinitely and leave the pages within the purged virtual memory range in an indeterminite state, whereas a forced extent purge function immediately purges, and the pages within the virtual memory range will be zero-filled the next time they are accessedamp;. If the function returns true, this indicates failure to purgeamp;.
typedef bool (extent_split_t)(extent_hooks_t *extent_hooks, void *addr, size_t size, size_t size_a, size_t size_b, bool committed, unsigned arena_ind);

 

An extent split function conforms to the extent_split_t type and optionally splits an extent at given addr and size into two adjacent extents, the first of size_a bytes, and the second of size_b bytes, operating on committed/decommitted memory as indicated, on behalf of arena arena_ind, returning false upon successamp;. If the function returns true, this indicates that the extent remains unsplit and therefore should continue to be operated on as a wholeamp;.
typedef bool (extent_merge_t)(extent_hooks_t *extent_hooks, void *addr_a, size_t size_a, void *addr_b, size_t size_b, bool committed, unsigned arena_ind);

 

An extent merge function conforms to the extent_merge_t type and optionally merges adjacent extents, at given addr_a and size_a with given addr_b and size_b into one contiguous extent, operating on committed/decommitted memory as indicated, on behalf of arena arena_ind, returning false upon successamp;. If the function returns true, this indicates that the extents remain distinct mappings and therefore should continue to be operated on independentlyamp;.

arenasamp;.narenas (unsigned) r-

Current limit on number of arenasamp;.

arenasamp;.dirty_decay_ms (ssize_t) rw

Current default per-arena approximate time in milliseconds from the creation of a set of unused dirty pages until an equivalent set of unused dirty pages is purged and/or reused, used to initialize arenaamp;.<i>amp;.dirty_decay_ms during arena creationamp;. See optamp;.dirty_decay_ms for additional informationamp;.

arenasamp;.muzzy_decay_ms (ssize_t) rw

Current default per-arena approximate time in milliseconds from the creation of a set of unused muzzy pages until an equivalent set of unused muzzy pages is purged and/or reused, used to initialize arenaamp;.<i>amp;.muzzy_decay_ms during arena creationamp;. See optamp;.muzzy_decay_ms for additional informationamp;.

arenasamp;.quantum (size_t) r-

Quantum sizeamp;.

arenasamp;.page (size_t) r-

Page sizeamp;.

arenasamp;.tcache_max (size_t) r-

Maximum thread-cached size classamp;.

arenasamp;.nbins (unsigned) r-

Number of bin size classesamp;.

arenasamp;.nhbins (unsigned) r-

Total number of thread cache bin size classesamp;.

arenasamp;.binamp;.<i>amp;.size (size_t) r-

Maximum size supported by size classamp;.

arenasamp;.binamp;.<i>amp;.nregs (uint32_t) r-

Number of regions per slabamp;.

arenasamp;.binamp;.<i>amp;.slab_size (size_t) r-

Number of bytes per slabamp;.

arenasamp;.nlextents (unsigned) r-

Total number of large size classesamp;.

arenasamp;.lextentamp;.<i>amp;.size (size_t) r-

Maximum size supported by this large size classamp;.

arenasamp;.create (unsigned, extent_hooks_t *) rw

Explicitly create a new arena outside the range of automatically managed arenas, with optionally specified extent hooks, and return the new arena indexamp;.

arenasamp;.lookup (unsigned, void*) rw

Index of the arena to which an allocation belongs toamp;.

profamp;.thread_active_init (bool) rw [--enable-prof]

Control the initial setting for threadamp;.profamp;.active in newly created threadsamp;. See the optamp;.prof_thread_active_init option for additional informationamp;.

profamp;.active (bool) rw [--enable-prof]

Control whether sampling is currently activeamp;. See the optamp;.prof_active option for additional information, as well as the interrelated threadamp;.profamp;.active mallctlamp;.

profamp;.dump (const char *) -w [--enable-prof]

Dump a memory profile to the specified file, or if NULL is specified, to a file according to the pattern <prefix>amp;.<pid>amp;.<seq>amp;.m<mseq>amp;.heap, where <prefix> is controlled by the optamp;.prof_prefix optionamp;.

profamp;.gdump (bool) rw [--enable-prof]

When enabled, trigger a memory profile dump every time the total virtual memory exceeds the previous maximumamp;. Profiles are dumped to files named according to the pattern <prefix>amp;.<pid>amp;.<seq>amp;.u<useq>amp;.heap, where <prefix> is controlled by the optamp;.prof_prefix optionamp;.

profamp;.reset (size_t) -w [--enable-prof]

Reset all memory profile statistics, and optionally update the sample rate (see optamp;.lg_prof_sample and profamp;.lg_sample)amp;.

profamp;.lg_sample (size_t) r- [--enable-prof]

Get the current sample rate (see optamp;.lg_prof_sample)amp;.

profamp;.interval (uint64_t) r- [--enable-prof]

Average number of bytes allocated between interval-based profile dumpsamp;. See the optamp;.lg_prof_interval option for additional informationamp;.

statsamp;.allocated (size_t) r- [--enable-stats]

Total number of bytes allocated by the applicationamp;.

statsamp;.active (size_t) r- [--enable-stats]

Total number of bytes in active pages allocated by the applicationamp;. This is a multiple of the page size, and greater than or equal to statsamp;.allocatedamp;. This does not include statsamp;.arenasamp;.<i>amp;.pdirty, statsamp;.arenasamp;.<i>amp;.pmuzzy, nor pages entirely devoted to allocator metadataamp;.

statsamp;.metadata (size_t) r- [--enable-stats]

Total number of bytes dedicated to metadata, which comprise base allocations used for bootstrap-sensitive allocator metadata structures (see statsamp;.arenasamp;.<i>amp;.base) and internal allocations (see statsamp;.arenasamp;.<i>amp;.internal)amp;. Transparent huge page (enabled with optamp;.metadata_thp) usage is not consideredamp;.

statsamp;.metadata_thp (size_t) r- [--enable-stats]

Number of transparent huge pages (THP) used for metadataamp;. See statsamp;.metadata and optamp;.metadata_thp) for detailsamp;.

statsamp;.resident (size_t) r- [--enable-stats]

Maximum number of bytes in physically resident data pages mapped by the allocator, comprising all pages dedicated to allocator metadata, pages backing active allocations, and unused dirty pagesamp;. This is a maximum rather than precise because pages may not actually be physically resident if they correspond to demand-zeroed virtual memory that has not yet been touchedamp;. This is a multiple of the page size, and is larger than statsamp;.activeamp;.

statsamp;.mapped (size_t) r- [--enable-stats]

Total number of bytes in active extents mapped by the allocatoramp;. This is larger than statsamp;.activeamp;. This does not include inactive extents, even those that contain unused dirty pages, which means that there is no strict ordering between this and statsamp;.residentamp;.

statsamp;.retained (size_t) r- [--enable-stats]

Total number of bytes in virtual memory mappings that were retained rather than being returned to the operating system via eamp;.gamp;. munmap(2) or similaramp;. Retained virtual memory is typically untouched, decommitted, or purged, so it has no strongly associated physical memory (see extent hooks for details)amp;. Retained memory is excluded from mapped memory statistics, eamp;.gamp;. statsamp;.mappedamp;.

statsamp;.background_threadamp;.num_threads (size_t) r- [--enable-stats]

Number of background threads running currentlyamp;.

statsamp;.background_threadamp;.num_runs (uint64_t) r- [--enable-stats]

Total number of runs from all background threadsamp;.

statsamp;.background_threadamp;.run_interval (uint64_t) r- [--enable-stats]

Average run interval in nanoseconds of background threadsamp;.

statsamp;.mutexesamp;.ctlamp;.{counter}; (counter specific type) r- [--enable-stats]

Statistics on ctl mutex (global scope; mallctl related)amp;. {counter} is one of the counters below:

num_ops (uint64_t): Total number of lock acquisition operations on this mutexamp;.

num_spin_acq (uint64_t): Number of times the mutex was spin-acquiredamp;. When the mutex is currently locked and cannot be acquired immediately, a short period of spin-retry within jemalloc will be performedamp;. Acquired through spin generally means the contention was lightweight and not causing context switchesamp;.

num_wait (uint64_t): Number of times the mutex was wait-acquired, which means the mutex contention was not solved by spin-retry, and blocking operation was likely involved in order to acquire the mutexamp;. This event generally implies higher cost / longer delay, and should be investigated if it happens oftenamp;.

max_wait_time (uint64_t): Maximum length of time in nanoseconds spent on a single wait-acquired lock operationamp;. Note that to avoid profiling overhead on the common path, this does not consider spin-acquired casesamp;.

total_wait_time (uint64_t): Cumulative time in nanoseconds spent on wait-acquired lock operationsamp;. Similarly, spin-acquired cases are not consideredamp;.

max_num_thds (uint32_t): Maximum number of threads waiting on this mutex simultaneouslyamp;. Similarly, spin-acquired cases are not consideredamp;.

num_owner_switch (uint64_t): Number of times the current mutex owner is different from the previous oneamp;. This event does not generally imply an issue; rather it is an indicator of how often the protected data are accessed by different threadsamp;.

statsamp;.mutexesamp;.background_threadamp;.{counter} (counter specific type) r- [--enable-stats]

Statistics on background_thread mutex (global scope; background_thread related)amp;. {counter} is one of the counters in mutex profiling countersamp;.

statsamp;.mutexesamp;.profamp;.{counter} (counter specific type) r- [--enable-stats]

Statistics on prof mutex (global scope; profiling related)amp;. {counter} is one of the counters in mutex profiling countersamp;.

statsamp;.mutexesamp;.reset (void) -- [--enable-stats]

Reset all mutex profile statistics, including global mutexes, arena mutexes and bin mutexesamp;.

statsamp;.arenasamp;.<i>amp;.dss (const char *) r-

dss (sbrk(2)) allocation precedence as related to mmap(2) allocationamp;. See optamp;.dss for detailsamp;.

statsamp;.arenasamp;.<i>amp;.dirty_decay_ms (ssize_t) r-

Approximate time in milliseconds from the creation of a set of unused dirty pages until an equivalent set of unused dirty pages is purged and/or reusedamp;. See optamp;.dirty_decay_ms for detailsamp;.

statsamp;.arenasamp;.<i>amp;.muzzy_decay_ms (ssize_t) r-

Approximate time in milliseconds from the creation of a set of unused muzzy pages until an equivalent set of unused muzzy pages is purged and/or reusedamp;. See optamp;.muzzy_decay_ms for detailsamp;.

statsamp;.arenasamp;.<i>amp;.nthreads (unsigned) r-

Number of threads currently assigned to arenaamp;.

statsamp;.arenasamp;.<i>amp;.uptime (uint64_t) r-

Time elapsed (in nanoseconds) since the arena was createdamp;. If <i> equals 0 or MALLCTL_ARENAS_ALL, this is the uptime since malloc initializationamp;.

statsamp;.arenasamp;.<i>amp;.pactive (size_t) r-

Number of pages in active extentsamp;.

statsamp;.arenasamp;.<i>amp;.pdirty (size_t) r-

Number of pages within unused extents that are potentially dirty, and for which madvise() or similar has not been calledamp;. See optamp;.dirty_decay_ms for a description of dirty pagesamp;.

statsamp;.arenasamp;.<i>amp;.pmuzzy (size_t) r-

Number of pages within unused extents that are muzzyamp;. See optamp;.muzzy_decay_ms for a description of muzzy pagesamp;.

statsamp;.arenasamp;.<i>amp;.mapped (size_t) r- [--enable-stats]

Number of mapped bytesamp;.

statsamp;.arenasamp;.<i>amp;.retained (size_t) r- [--enable-stats]

Number of retained bytesamp;. See statsamp;.retained for detailsamp;.

statsamp;.arenasamp;.<i>amp;.base (size_t) r- [--enable-stats]

Number of bytes dedicated to bootstrap-sensitive allocator metadata structuresamp;.

statsamp;.arenasamp;.<i>amp;.internal (size_t) r- [--enable-stats]

Number of bytes dedicated to internal allocationsamp;. Internal allocations differ from application-originated allocations in that they are for internal use, and that they are omitted from heap profilesamp;.

statsamp;.arenasamp;.<i>amp;.metadata_thp (size_t) r- [--enable-stats]

Number of transparent huge pages (THP) used for metadataamp;. See optamp;.metadata_thp for detailsamp;.

statsamp;.arenasamp;.<i>amp;.resident (size_t) r- [--enable-stats]

Maximum number of bytes in physically resident data pages mapped by the arena, comprising all pages dedicated to allocator metadata, pages backing active allocations, and unused dirty pagesamp;. This is a maximum rather than precise because pages may not actually be physically resident if they correspond to demand-zeroed virtual memory that has not yet been touchedamp;. This is a multiple of the page sizeamp;.

statsamp;.arenasamp;.<i>amp;.dirty_npurge (uint64_t) r- [--enable-stats]

Number of dirty page purge sweeps performedamp;.

statsamp;.arenasamp;.<i>amp;.dirty_nmadvise (uint64_t) r- [--enable-stats]

Number of madvise() or similar calls made to purge dirty pagesamp;.

statsamp;.arenasamp;.<i>amp;.dirty_purged (uint64_t) r- [--enable-stats]

Number of dirty pages purgedamp;.

statsamp;.arenasamp;.<i>amp;.muzzy_npurge (uint64_t) r- [--enable-stats]

Number of muzzy page purge sweeps performedamp;.

statsamp;.arenasamp;.<i>amp;.muzzy_nmadvise (uint64_t) r- [--enable-stats]

Number of madvise() or similar calls made to purge muzzy pagesamp;.

statsamp;.arenasamp;.<i>amp;.muzzy_purged (uint64_t) r- [--enable-stats]

Number of muzzy pages purgedamp;.

statsamp;.arenasamp;.<i>amp;.smallamp;.allocated (size_t) r- [--enable-stats]

Number of bytes currently allocated by small objectsamp;.

statsamp;.arenasamp;.<i>amp;.smallamp;.nmalloc (uint64_t) r- [--enable-stats]

Cumulative number of times a small allocation was requested from the arena's bins, whether to fill the relevant tcache if optamp;.tcache is enabled, or to directly satisfy an allocation request otherwiseamp;.

statsamp;.arenasamp;.<i>amp;.smallamp;.ndalloc (uint64_t) r- [--enable-stats]

Cumulative number of times a small allocation was returned to the arena's bins, whether to flush the relevant tcache if optamp;.tcache is enabled, or to directly deallocate an allocation otherwiseamp;.

statsamp;.arenasamp;.<i>amp;.smallamp;.nrequests (uint64_t) r- [--enable-stats]

Cumulative number of allocation requests satisfied by all bin size classesamp;.

statsamp;.arenasamp;.<i>amp;.largeamp;.allocated (size_t) r- [--enable-stats]

Number of bytes currently allocated by large objectsamp;.

statsamp;.arenasamp;.<i>amp;.largeamp;.nmalloc (uint64_t) r- [--enable-stats]

Cumulative number of times a large extent was allocated from the arena, whether to fill the relevant tcache if optamp;.tcache is enabled and the size class is within the range being cached, or to directly satisfy an allocation request otherwiseamp;.

statsamp;.arenasamp;.<i>amp;.largeamp;.ndalloc (uint64_t) r- [--enable-stats]

Cumulative number of times a large extent was returned to the arena, whether to flush the relevant tcache if optamp;.tcache is enabled and the size class is within the range being cached, or to directly deallocate an allocation otherwiseamp;.

statsamp;.arenasamp;.<i>amp;.largeamp;.nrequests (uint64_t) r- [--enable-stats]

Cumulative number of allocation requests satisfied by all large size classesamp;.

statsamp;.arenasamp;.<i>amp;.binsamp;.<j>amp;.nmalloc (uint64_t) r- [--enable-stats]

Cumulative number of times a bin region of the corresponding size class was allocated from the arena, whether to fill the relevant tcache if optamp;.tcache is enabled, or to directly satisfy an allocation request otherwiseamp;.

statsamp;.arenasamp;.<i>amp;.binsamp;.<j>amp;.ndalloc (uint64_t) r- [--enable-stats]

Cumulative number of times a bin region of the corresponding size class was returned to the arena, whether to flush the relevant tcache if optamp;.tcache is enabled, or to directly deallocate an allocation otherwiseamp;.

statsamp;.arenasamp;.<i>amp;.binsamp;.<j>amp;.nrequests (uint64_t) r- [--enable-stats]

Cumulative number of allocation requests satisfied by bin regions of the corresponding size classamp;.

statsamp;.arenasamp;.<i>amp;.binsamp;.<j>amp;.curregs (size_t) r- [--enable-stats]

Current number of regions for this size classamp;.

statsamp;.arenasamp;.<i>amp;.binsamp;.<j>amp;.nfills (uint64_t) r-

Cumulative number of tcache fillsamp;.

statsamp;.arenasamp;.<i>amp;.binsamp;.<j>amp;.nflushes (uint64_t) r-

Cumulative number of tcache flushesamp;.

statsamp;.arenasamp;.<i>amp;.binsamp;.<j>amp;.nslabs (uint64_t) r- [--enable-stats]

Cumulative number of slabs createdamp;.

statsamp;.arenasamp;.<i>amp;.binsamp;.<j>amp;.nreslabs (uint64_t) r- [--enable-stats]

Cumulative number of times the current slab from which to allocate changedamp;.

statsamp;.arenasamp;.<i>amp;.binsamp;.<j>amp;.curslabs (size_t) r- [--enable-stats]

Current number of slabsamp;.

statsamp;.arenasamp;.<i>amp;.binsamp;.<j>amp;.mutexamp;.{counter} (counter specific type) r- [--enable-stats]

Statistics on arenaamp;.<i>amp;.binsamp;.<j> mutex (arena bin scope; bin operation related)amp;. {counter} is one of the counters in mutex profiling countersamp;.

statsamp;.arenasamp;.<i>amp;.lextentsamp;.<j>amp;.nmalloc (uint64_t) r- [--enable-stats]

Cumulative number of times a large extent of the corresponding size class was allocated from the arena, whether to fill the relevant tcache if optamp;.tcache is enabled and the size class is within the range being cached, or to directly satisfy an allocation request otherwiseamp;.

statsamp;.arenasamp;.<i>amp;.lextentsamp;.<j>amp;.ndalloc (uint64_t) r- [--enable-stats]

Cumulative number of times a large extent of the corresponding size class was returned to the arena, whether to flush the relevant tcache if optamp;.tcache is enabled and the size class is within the range being cached, or to directly deallocate an allocation otherwiseamp;.

statsamp;.arenasamp;.<i>amp;.lextentsamp;.<j>amp;.nrequests (uint64_t) r- [--enable-stats]

Cumulative number of allocation requests satisfied by large extents of the corresponding size classamp;.

statsamp;.arenasamp;.<i>amp;.lextentsamp;.<j>amp;.curlextents (size_t) r- [--enable-stats]

Current number of large allocations for this size classamp;.

statsamp;.arenasamp;.<i>amp;.mutexesamp;.largeamp;.{counter} (counter specific type) r- [--enable-stats]

Statistics on arenaamp;.<i>amp;.large mutex (arena scope; large allocation related)amp;. {counter} is one of the counters in mutex profiling countersamp;.

statsamp;.arenasamp;.<i>amp;.mutexesamp;.extent_availamp;.{counter} (counter specific type) r- [--enable-stats]

Statistics on arenaamp;.<i>amp;.extent_avail mutex (arena scope; extent avail related)amp;. {counter} is one of the counters in mutex profiling countersamp;.

statsamp;.arenasamp;.<i>amp;.mutexesamp;.extents_dirtyamp;.{counter} (counter specific type) r- [--enable-stats]

Statistics on arenaamp;.<i>amp;.extents_dirty mutex (arena scope; dirty extents related)amp;. {counter} is one of the counters in mutex profiling countersamp;.

statsamp;.arenasamp;.<i>amp;.mutexesamp;.extents_muzzyamp;.{counter} (counter specific type) r- [--enable-stats]

Statistics on arenaamp;.<i>amp;.extents_muzzy mutex (arena scope; muzzy extents related)amp;. {counter} is one of the counters in mutex profiling countersamp;.

statsamp;.arenasamp;.<i>amp;.mutexesamp;.extents_retainedamp;.{counter} (counter specific type) r- [--enable-stats]

Statistics on arenaamp;.<i>amp;.extents_retained mutex (arena scope; retained extents related)amp;. {counter} is one of the counters in mutex profiling countersamp;.

statsamp;.arenasamp;.<i>amp;.mutexesamp;.decay_dirtyamp;.{counter} (counter specific type) r- [--enable-stats]

Statistics on arenaamp;.<i>amp;.decay_dirty mutex (arena scope; decay for dirty pages related)amp;. {counter} is one of the counters in mutex profiling countersamp;.

statsamp;.arenasamp;.<i>amp;.mutexesamp;.decay_muzzyamp;.{counter} (counter specific type) r- [--enable-stats]

Statistics on arenaamp;.<i>amp;.decay_muzzy mutex (arena scope; decay for muzzy pages related)amp;. {counter} is one of the counters in mutex profiling countersamp;.

statsamp;.arenasamp;.<i>amp;.mutexesamp;.baseamp;.{counter} (counter specific type) r- [--enable-stats]

Statistics on arenaamp;.<i>amp;.base mutex (arena scope; base allocator related)amp;. {counter} is one of the counters in mutex profiling countersamp;.

statsamp;.arenasamp;.<i>amp;.mutexesamp;.tcache_listamp;.{counter} (counter specific type) r- [--enable-stats]

Statistics on arenaamp;.<i>amp;.tcache_list mutex (arena scope; tcache to arena association related)amp;. This mutex is expected to be accessed less oftenamp;. {counter} is one of the counters in mutex profiling countersamp;.

HEAP PROFILE FORMAT

Although the heap profiling functionality was originally designed to be compatible with the pprof command that is developed as part of the gperftools packageamp;[3], the addition of per thread heap profiling functionality required a different heap profile formatamp;. The jeprof command is derived from pprof, with enhancements to support the heap profile format described hereamp;.

In the following hypothetical heap profile, [amp;.amp;.amp;.] indicates elision for the sake of compactnessamp;.

heap_v2/524288
  t*: 28106: 56637512 [0: 0]
  [amp;.amp;.amp;.]
  t3: 352: 16777344 [0: 0]
  [amp;.amp;.amp;.]
  t99: 17754: 29341640 [0: 0]
  [amp;.amp;.amp;.]
@ 0x5f86da8 0x5f5a1dc [amp;.amp;.amp;.] 0x29e4d4e 0xa200316 0xabb2988 [amp;.amp;.amp;.]
  t*: 13: 6688 [0: 0]
  t3: 12: 6496 [0: ]
  t99: 1: 192 [0: 0]
[amp;.amp;.amp;.]

MAPPED_LIBRARIES: [amp;.amp;.amp;.]

The following matches the above heap profile, but most tokens are replaced with <description> to indicate descriptions of the corresponding fieldsamp;.

<heap_profile_format_version>/<mean_sample_interval>
  <aggregate>: <curobjs>: <curbytes> [<cumobjs>: <cumbytes>]
  [amp;.amp;.amp;.]
  <thread_3_aggregate>: <curobjs>: <curbytes>[<cumobjs>: <cumbytes>]
  [amp;.amp;.amp;.]
  <thread_99_aggregate>: <curobjs>: <curbytes>[<cumobjs>: <cumbytes>]
  [amp;.amp;.amp;.]
@ <top_frame> <frame> [amp;.amp;.amp;.] <frame> <frame> <frame> [amp;.amp;.amp;.]
  <backtrace_aggregate>: <curobjs>: <curbytes> [<cumobjs>: <cumbytes>]
  <backtrace_thread_3>: <curobjs>: <curbytes> [<cumobjs>: <cumbytes>]
  <backtrace_thread_99>: <curobjs>: <curbytes> [<cumobjs>: <cumbytes>]
[amp;.amp;.amp;.]

MAPPED_LIBRARIES: </proc/<pid>/maps>

DEBUGGING MALLOC PROBLEMS

When debugging, it is a good idea to configure/build jemalloc with the --enable-debug and --enable-fill options, and recompile the program with suitable options and symbols for debugger supportamp;. When so configured, jemalloc incorporates a wide variety of run-time assertions that catch application errors such as double-free, write-after-free, etcamp;.

Programs often accidentally depend on &#147;uninitialized&#148; memory actually being filled with zero bytesamp;. Junk filling (see the optamp;.junk option) tends to expose such bugs in the form of obviously incorrect results and/or coredumpsamp;. Conversely, zero filling (see the optamp;.zero option) eliminates the symptoms of such bugsamp;. Between these two options, it is usually possible to quickly detect, diagnose, and eliminate such bugsamp;.

This implementation does not provide much detail about the problems it detects, because the performance impact for storing such information would be prohibitiveamp;.

DIAGNOSTIC MESSAGES

If any of the memory allocation/deallocation functions detect an error or warning condition, a message will be printed to file descriptor STDERR_FILENOamp;. Errors will result in the process dumping coreamp;. If the optamp;.abort option is set, most warnings are treated as errorsamp;.

The malloc_message variable allows the programmer to override the function which emits the text strings forming the errors and warnings if for some reason the STDERR_FILENO file descriptor is not suitable for thisamp;. malloc_message() takes the cbopaque pointer argument that is NULL unless overridden by the arguments in a call to malloc_stats_print(), followed by a string pointeramp;. Please note that doing anything which tries to allocate memory in this function is likely to result in a crash or deadlockamp;.

All messages are prefixed by &#147;<jemalloc>: &#148;amp;.

RETURN VALUES

Standard API

The malloc() and calloc() functions return a pointer to the allocated memory if successful; otherwise a NULL pointer is returned and errno is set to ENOMEMamp;.

The posix_memalign() function returns the value 0 if successful; otherwise it returns an error valueamp;. The posix_memalign() function will fail if:

EINVAL

The alignment parameter is not a power of 2 at least as large as sizeof(void *)amp;.

ENOMEM

Memory allocation erroramp;.

The aligned_alloc() function returns a pointer to the allocated memory if successful; otherwise a NULL pointer is returned and errno is setamp;. The aligned_alloc() function will fail if:

EINVAL

The alignment parameter is not a power of 2amp;.

ENOMEM

Memory allocation erroramp;.

The realloc() function returns a pointer, possibly identical to ptr, to the allocated memory if successful; otherwise a NULL pointer is returned, and errno is set to ENOMEM if the error was the result of an allocation failureamp;. The realloc() function always leaves the original buffer intact when an error occursamp;.

The free() function returns no valueamp;.

Non-standard API

The mallocx() and rallocx() functions return a pointer to the allocated memory if successful; otherwise a NULL pointer is returned to indicate insufficient contiguous memory was available to service the allocation requestamp;.

The xallocx() function returns the real size of the resulting resized allocation pointed to by ptr, which is a value less than size if the allocation could not be adequately grown in placeamp;.

The sallocx() function returns the real size of the allocation pointed to by ptramp;.

The nallocx() returns the real size that would result from a successful equivalent mallocx() function call, or zero if insufficient memory is available to perform the size computationamp;.

The mallctl(), mallctlnametomib(), and mallctlbymib() functions return 0 on success; otherwise they return an error valueamp;. The functions will fail if:

EINVAL

newp is not NULL, and newlen is too large or too smallamp;. Alternatively, *oldlenp is too large or too small; in this case as much data as possible are read despite the erroramp;.

ENOENT

name or mib specifies an unknown/invalid valueamp;.

EPERM

Attempt to read or write void value, or attempt to write read-only valueamp;.

EAGAIN

A memory allocation failure occurredamp;.

EFAULT

An interface with side effects failed in some way not directly related to mallctl*() read/write processingamp;.

The malloc_usable_size() function returns the usable size of the allocation pointed to by ptramp;.

ENVIRONMENT

The following environment variable affects the execution of the allocation functions:

MALLOC_CONF

If the environment variable MALLOC_CONF is set, the characters it contains will be interpreted as optionsamp;.

EXAMPLES

To dump core whenever a problem occurs:

ln -s 'abort:true' /etc/mallocamp;.conf

To specify in the source that only one arena should be automatically created:

malloc_conf = "narenas:1";

SEE ALSO

madvise(2), mmap(2), sbrk(2), utrace(2), alloca(3), atexit(3), getpagesize(3)

STANDARDS

The malloc(), calloc(), realloc(), and free() functions conform to ISO/IEC 9899:1990 (&#147;ISO C90&#148;)amp;.

The posix_memalign() function conforms to IEEE Std 1003amp;.1-2001 (&#147;POSIXamp;.1&#148;)amp;.

HISTORY

The malloc_usable_size() and posix_memalign() functions first appeared in FreeBSD 7amp;.0amp;.

The aligned_alloc(), malloc_stats_print(), and mallctl*() functions first appeared in FreeBSD 10amp;.0amp;.

The *allocx() functions first appeared in FreeBSD 11amp;.0amp;.

AUTHOR

Jason Evans

NOTES

1. jemalloc website  http://jemalloc.net/
2. JSON format  http://www.json.org/
3. gperftools package  http://code.google.com/p/gperftools/

05/08/2018 JEMALLOC (3) jemalloc 5.1.0-0-g61efbda7098d

tail head cat sleep
QR code linking to this page


Please direct any comments about this manual page service to Ben Bullock. Privacy policy.

Our grievance is not just against Unix itself, but against the cult of Unix zealots who defend and nurture it. They take the heat, disease, and pestilence as givens, and, as ancient shamans did, display their wounds, some self-inflicted, as proof of their power and wizardry. We aim, through bluntness and humor, to show them that they pray to a tin god, and that science, not religion, is the path to useful and friendly technology.
— The Unix Haters' handbook