NAME¶

memkind_arena - jemalloc arena allocation memkind operations.
Note: This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.

SYNOPSIS¶

int memkind_arena_create(struct memkind *kind, struct memkind_ops *ops, const char *name); int memkind_arena_create_map(struct memkind *kind, extent_hooks_t *hooks); int memkind_arena_destroy(struct memkind *kind); void *memkind_arena_malloc(struct memkind *kind, size_t size); void *memkind_arena_calloc(struct memkind *kind, size_t num, size_t size); int memkind_arena_posix_memalign(struct memkind *kind, void **memptr, size_t alignment, size_t size); void *memkind_arena_realloc(struct memkind *kind, void *ptr, size_t size); void *memkind_arena_realloc_with_kind_detect(void *ptr, size_t size); int memkind_thread_get_arena(struct memkind *kind, unsigned int *arena, size_t size); int memkind_bijective_get_arena(struct memkind *kind, unsigned int *arena, size_t size); struct memkind *get_kind_by_arena(unsigned arena_ind); struct memkind *memkind_arena_detect_kind(void *ptr); int memkind_arena_finalize(struct memkind *kind); void memkind_arena_init(struct memkind *kind); void memkind_arena_free(struct memkind *kind, void *ptr); void memkind_arena_free_with_kind_detect(void *ptr); size_t memkind_arena_malloc_usable_size(void *ptr); int memkind_arena_update_memory_usage_policy(struct memkind *kind, memkind_mem_usage_policy policy); int memkind_arena_enable_background_threads(size_t threads_limit);

DESCRIPTION¶

This header file is a collection of functions can be used to populate the memkind operations structure for memory kinds that use jemalloc.

memkind_arena_create() is an implementation of the memkind "create" operation for memory kinds that use jemalloc. This calls memkind_default_create() (see memkind_default(3)) followed by memkind_arena_create_map() described below.

memkind_arena_create_map() creates the arena_map array for the memkind structure pointed to by kind which can be indexed by the ops.get_arena() function from the kind's operations. If get_arena points memkind_thread_get_arena() then there will be four arenas created for each processor, and if get_arena points to memkind_bijective_get_arena() then just one arena is created.

memkind_arena_destroy() is an implementation of the memkind "destroy" operation for memory kinds that use jemalloc. This releases all of the resources allocated by memkind_arena_create().

memkind_arena_malloc() is an implementation of the memkind "malloc" operation for memory kinds that use jemalloc. This allocates memory using the arenas created by memkind_arena_create() through the jemalloc's mallocx() interface. It uses the memkind "get_arena" operation to select the arena.

memkind_arena_calloc() is an implementation of the memkind "calloc" operation for memory kinds that use jemalloc. This allocates memory using the arenas created by memkind_arena_create() through the jemalloc's mallocx() interface. It uses the memkind "get_arena" operation to select the arena.

memkind_arena_posix_memalign() is an implementation of the memkind "posix_memalign" operation for memory kinds that use jemalloc. This allocates memory using the arenas created by memkind_arena_create() through the jemalloc's mallocx() interface. It uses the memkind "get_arena" operation to select the arena. The POSIX standard requires that posix_memalign(3) may not set errno however the jemalloc's mallocx() routine may. In an attempt to abide by the standard errno is recorded before calling jemalloc's mallocx() and then reset after the call.

memkind_arena_realloc() is an implementation of the memkind "realloc" operation for memory kinds that use jemalloc. This allocates memory using the arenas created by memkind_arena_create() through the jemalloc's mallocx() interface. It uses the memkind "get_arena" operation to select the arena.

memkind_arena_realloc_with_kind_detect() function will look up for kind associated to the allocated memory referenced by ptr and call (depending on kind value) memkind_arena_realloc() or memkind_default_realloc().

memkind_thread_get_arena() retrieves the arena index that is bound to to the calling thread based on a hash of its thread ID. The arena index can be used with the MALLOCX_ARENA macro to set flags for jemalloc's mallocx().

memkind_bijective_arena_get_arena() retrieves the arena index to be used with the MALLOCX_ARENA macro to set flags for jemalloc's mallocx(). Use of this operation implies that only one arena is used for the kind.

memkind_arena_free() is an implementation of the memkind "free" operation for memory kinds that use jemalloc. It causes the allocated memory referenced by ptr, which must have been returned by a previous call to memkind_arena_malloc(), memkind_arena_calloc() or memkind_arena_realloc() to be made available for future allocations. It uses the memkind "get_arena" operation to select the arena.

memkind_arena_free_with_kind_detect() function will look up for kind associated to the allocated memory referenced by ptr and call memkind_arena_free().

memkind_arena_detect_kind() returns pointer to memory kind structure associated with given allocated memory referenced by ptr.

get_kind_by_arena() returns pointer to memory kind structure associated with given arena.

memkind_arena_malloc_usable_size() is an implementation of the memkind "usable_size" operation for memory kinds that use jemalloc.

memkind_arena_finalize() is an implementation of the memkind "finalize" operation for memory kinds that use jemalloc. This function releases all resources allocated by memkind_arena_create() and it's called when main() finishes or after calling exit() function.

memkind_arena_init() creates arena map with proper hooks per specified kind.

memkind_arena_update_memory_usage_policy() function changes time, which determine how fast jemalloc returns unused pages back to the operating system, in other words how fast it deallocates file space.

memkind_arena_enable_background_threads() enables internal background worker threads in jemalloc. The threads_limit specify limit of background threads which will be enabled ( 0 means no limit).

COPYRIGHT¶

Copyright (C) 2014 - 2020 Intel Corporation. All rights reserved.

SEE ALSO¶

memkind(3), memkind_default(3), memkind_hbw(3), memkind_hugetlb(3), memkind_pmem(3), jemalloc(3), mbind(2), mmap(2)