LIBHUGETLBFS(7) Miscellaneous Information Manual LIBHUGETLBFS(7)
libhugetlbfs - preload library to back text, data, malloc() or shared
memory with hugepages
export [environment options]
libhugetlbfs is a library that can back application text, data, mal‐
loc() and shared memory with hugepages. This is of benefit to applica‐
tions that use large amounts of address space and suffer a performance
hit due to TLB misses. Wall-clock time or oprofile can be used to
determine if there is a performance benefit from using libhugetlbfs or
not. In all cases but shared memory, a hugetlbfs mount must exist and
a hugepage pool defined for hugepages to be used.
Some limited functionality is available for unmodified dynamically
linked applications. By preloading the library, the library can back
malloc() and shared memory, and text and data segments can be partially
backed if they are large enough.
For the effective backing of text and data with huge pages, the appli‐
cation must be linked to the library and the ELF segments correctly
aligned using the ld helpers. Once linked, malloc or shared memory can
still be backed but no pre-loading is required. See /usr/share/doc/lib‐
hugetlbfs/HOWTO and ld.hugetlbfs(1) for detailed instructions on
For applications that are hugepage-aware and linked to the library
get_huge_pages() can be used for the direct allocation of hugepage-
Unless otherwise specified, libhugetlbfs will use the default hugepage
size to back memory regions. The default size is the value of Hugepage‐
size displayed in /proc/meminfo. The size can be specified in bytes or
in kilobytes, megabytes, or gigabytes by appending K, M, or G respec‐
tively. It is an error to specify a invalid, unsupported, or otherwise
unconfigured huge page size. Kernel 2.6.27 or later is required to
specify any pagesize other than the default.
See /usr/share/docs/libhugetlbfs/HOWTO for detailed instructions on how
the library should be used, particularly when relinking the applica‐
tion. This manual page provides a brief synopsis of the environment
variables as a quick reference.
The following variables affect what memory regions are backed by
hugepages. In all cases, the environment being unset implies the fea‐
ture should remain disabled.
This sets the default hugepage size to be used by libhugetlbfs.
If not set, libhugetlbfs will use the kernel's default hugepage
This enables the hugepage malloc() feature, instructing lib‐
hugetlbfs to override glibc's normal morecore() function with a
hugepage version and use it for malloc(). All application mal‐
loc() memory should come from hugepage memory until it runs out,
it will then fallback to base pages. Note that applications
that use custom allocators may not be able to back their heaps
using hugepages and this environment variable. It may be neces‐
sary to modify the custom allocator to use get_huge_pages().
When this environment variable is set, the SHM_HUGETLB flag is
added to the shmget() call and the size parameter is aligned to
back the shared memory segment with hugepages. In the event
hugepages cannot be used, base pages will be used instead and a
warning will be printed to explain the failure. The pagesize
cannot be specified with this parameter. To change the kernels
default hugepage size, use the pagesize= kernel boot parameter
(2.6.26 or later required).
If the application has been relinked (see the HOWTO for instruc‐
tions), this environment variable determines whether read-only,
read-write, both or no segments are backed by hugepages and what
pagesize should be used. If the recommended relinking method has
been used, then hugeedit can be used to automatically back the
text or data by default.
Force the use of hugepages for text and data segments even if
the application has not been relinked to align the ELF segments
on a hugepage boundary. Partial segment remapping is not guar‐
anteed to work and the segments must be large enough to contain
at least one hugepage for the remapping to occur.
The following options affect how libhugetlbfs behaves.
By default, libhugetlbfs will act on any program that it is
loaded with, either via LD_PRELOAD or by explicitly linking with
There are situations in which it is desirable to restrict lib‐
hugetlbfs' actions to specific programs. For example, some ISV
applications are wrapped in a series of scripts that invoke
bash, python, and/or perl. It is more convenient to set the
environment variables related to libhugetlbfs before invoking
the wrapper scripts, yet this has the unintended and undesirable
consequence of causing the script interpreters to use and con‐
sume hugepages. There is no obvious benefit to causing the
script interpreters to use hugepages, and there is a clear dis‐
advantage: fewer hugepages are available to the actual applica‐
To address this scenario, set HUGETLB_RESTRICT_EXE to a colon-
separated list of programs to which the other libhugetlbfs envi‐
ronment variables should apply. (If not set, libhugetlbfs will
attempt to apply the requested actions to all programs.) For
will restrict libhugetlbfs' actions to programs named
/home/fred/hpcc and /bench/long_hpcc but not /bin/hpcc_no.
By default, the hugepage heap does not shrink. Shrinking is
enabled by setting this environment variable. It is disabled by
default as glibc occasionally exhibits strange behaviour if it
mistakes the heap returned by libhugetlbfs as a foreign brk().
By default libhugetlbfs will prefault regions it creates to
ensure they can be referenced without receiving a SIGKILL. On
kernels older than 2.6.27, this was necessary as the system did
not guarantee that future faults would succeed on regions mapped
MAP_PRIVATE. Prefaulting impacts the performance of malloc()
and can result in poor placement on NUMA systems. If it is known
the hugepage pool is large enough to run the application or the
kernel is 2.6.27 or later, this environment variable should be
By default, the kernel will reserve huge pages at mmap() time to
ensure that future faults will succeed. This avoids unexpected
application failure at fault time but some applications depend
on memory overcommit to create large sparse mappings. For this
type of application, setting this environment variable will cre‐
ate huge page backed mappings without a reservation. Use this
option with extreme care as in the event huge pages are not
available when the mapping is used, the application will be
killed. On older kernels, the use of this feature can trigger
the OOM killer. Hence, even with this variable set, reservations
may still be used for safety.
libhugetlbfs normally picks an address to use as the base of the
heap for malloc() automatically. This environment variable fixes
which address is used.
The path to the hugetlbfs mount is automatically determined at
run-time. In the event there are multiple mounts and the wrong
one is being selected, use this option to select the correct
one. This may be the case if an application-specific mount with
a fixed quota has been created for example.
By default, libhugetlbfs uses unlinked hugetlbfs files to store
remapped program segment data. If the same program is started
multiple times using hugepage segments, multiple hugepages will
be used to store the same program data. The reduce this wastage,
setting this environment variable will share read-only segments
between multiple invocations of a program at the cost of the
memory being used whether the applications are running or not.
It is also possible that a malicious application inferfere with
other applications executable code. See the HOWTO for more
detailed information on this topic.
The following options control the verbosity of libhugetlbfs.
The default value for this is 1 and the range of the value is
from 0 to 99. The higher the value, the more verbose the output
is. 0 is quiet and 3 will output much debugging information.
Once set, this will give very detailed output on what is happen‐
ing in the library and run extra diagnostics.
oprofile(1), ld.hugetlbfs(1), hugectl(8), hugeedit(8), gethugepage‐
size(3), gethugepagesizes(3), getpagesizes(3), hugetlbfs_test_path(3),
libhugetlbfs was written by various people on the libhugetlbfs-devel
September 27, 2008 LIBHUGETLBFS(7)