Note: the Rust allocator API is implemented for
jemalloc
in thetikv-jemallocator
crate.
jemalloc
is a general purpose memory allocator, its documentation
can be found here:
Current jemalloc version: 5.2.1.
See the platform support of the tikv-jemallocator
crate.
Most features correspond to jemalloc
features - the reference is jemalloc/INSTALL.md
.
This crate provides following cargo feature flags:
profiling
(configure jemalloc
with --enable-prof
): Enable heap profiling and leak detection functionality. See jemalloc's “opt.prof” option documentation for usage details. When enabled, there are several approaches to backtracing, and the configure script chooses the first one in the following list that appears to function correctly:
libunwind
(requires --enable-prof-libunwind)libgcc
(unless --disable-prof-libgcc)gcc intrinsics
(unless --disable-prof-gcc)stats
(configure jemalloc
with --enable-stats
): Enable statistics gathering functionality. See the jemalloc
's “opt.stats_print
” option documentation for usage details.
debug
(configure jemalloc
with --enable-debug
): Enable assertions and validation code. This incurs a substantial performance hit, but is very useful during application development.
background_threads_runtime_support
(enabled by default): enables background-threads run-time support when building jemalloc-sys
on some POSIX targets supported by jemalloc
. Background threads are disabled at run-time by default. This option allows dynamically enabling them at run-time.
background_threads
(disabled by default): enables background threads by default at run-time. 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). Threads run periodically, and handle purging asynchronously. When switching off, background threads are terminated synchronously. Note that after fork(2)
function, the state in the child process will be disabled regardless the state in parent process. See stats.background_thread
for related stats. opt.background_thread
can be used to set the default option. The background thread is only available on selected pthread-based platforms.
unprefixed_malloc_on_supported_platforms
: when disabled, configure jemalloc
with --with-jemalloc-prefix=_rjem_
. Enabling this causes symbols like malloc
to be emitted without a prefix, overriding the ones defined by libc. This usually causes C and C++ code linked in the same program to use jemalloc
as well. On some platforms prefixes are always used because unprefixing is known to cause segfaults due to allocator mismatches.
disable_initial_exec_tls
(disabled by default): when enabled, jemalloc is built with the --disable-initial-exec-tls
option. It disables the initial-exec TLS model for jemalloc‘s internal thread-local storage (on those platforms that support explicit settings). This can allow jemalloc to be dynamically loaded after program startup (e.g. using dlopen). If you encounter the error yourlib.so: cannot allocate memory in static TLS block
, you’ll likely want to enable this.
disable_cache_oblivious
(disabled by default): when enabled, jemalloc is built with the --disable-cache-oblivious
option. In that case, all large allocations are page-aligned as an implementation artifact. It may severely harm CPU cache utilization. However, the cache-oblivious layout has a cost of one extra page per large allocation which can be unfeasible for certain applications.
jemalloc
options taking values are passed via environment variables using the schema JEMALLOC_SYS_{KEY}=VALUE
where the KEY
names correspond to the ./configure
options of jemalloc
where the words are capitalized and the hyphens -
are replaced with underscores _
(see jemalloc/INSTALL.md
):
JEMALLOC_SYS_WITH_MALLOC_CONF=<malloc_conf>
: Embed <malloc_conf>
as a run-time options string that is processed prior to the malloc_conf
global variable, the /etc/malloc.conf
symlink, and the MALLOC_CONF
environment variable (note: this variable might be prefixed as _RJEM_MALLOC_CONF
). For example, to change the default decay time for dirty pages to 30 seconds:
JEMALLOC_SYS_WITH_MALLOC_CONF=dirty_decay_ms:30000
JEMALLOC_SYS_WITH_LG_PAGE=<lg-page>
: Specify the base 2 log of the allocator page size, which must in turn be at least as large as the system page size. By default the configure script determines the host's page size and sets the allocator page size equal to the system page size, so this option need not be specified unless the system page size may change between configuration and execution, e.g. when cross compiling.
JEMALLOC_SYS_WITH_LG_HUGEPAGE=<lg-hugepage>
: Specify the base 2 log of the system huge page size. This option is useful when cross compiling, or when overriding the default for systems that do not explicitly support huge pages.
JEMALLOC_SYS_WITH_LG_QUANTUM=<lg-quantum>
: Specify the base 2 log of the minimum allocation alignment. jemalloc needs to know the minimum alignment that meets the following C standard requirement (quoted from the April 12, 2011 draft of the C11 standard):
The pointer returned if the allocation succeeds is suitably aligned so that it may be assigned to a pointer to any type of object with a fundamental alignment requirement and then used to access such an object or an array of such objects in the space allocated [...]
This setting is architecture-specific, and although jemalloc includes known safe values for the most commonly used modern architectures, there is a wrinkle related to GNU libc (glibc) that may impact your choice of . On most modern architectures, this mandates 16-byte alignment (=4), but the glibc developers chose not to meet this requirement for performance reasons. An old discussion can be found at https://sourceware.org/bugzilla/show_bug.cgi?id=206 . Unlike glibc, jemalloc does follow the C standard by default (caveat: jemalloc technically cheats for size classes smaller than the quantum), but the fact that Linux systems already work around this allocator noncompliance means that it is generally safe in practice to let jemalloc‘s minimum alignment follow glibc’s lead. If you specify JEMALLOC_SYS_WITH_LG_QUANTUM=3
during configuration, jemalloc will provide additional size classes that are not 16-byte-aligned (24, 40, and 56).
JEMALLOC_SYS_WITH_LG_VADDR=<lg-vaddr>
: Specify the number of significant virtual address bits. By default, the configure script attempts to detect virtual address size on those platforms where it knows how, and picks a default otherwise. This option may be useful when cross-compiling.
This project is licensed under either of
at your option.
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in tikv-jemalloc-sys
by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.