xref: /openbmc/linux/Documentation/core-api/memory-allocation.rst (revision 1ac731c529cd4d6adbce134754b51ff7d822b145)

.. _memory_allocation:

=======================
Memory Allocation Guide
=======================

Linux provides a variety of APIs for memory allocation. You can
allocate small chunks using `kmalloc` or `kmem_cache_alloc` families,
large virtually contiguous areas using `vmalloc` and its derivatives,
or you can directly request pages from the page allocator with
`alloc_pages`. It is also possible to use more specialized allocators,
for instance `cma_alloc` or `zs_malloc`.

Most of the memory allocation APIs use GFP flags to express how that
memory should be allocated. The GFP acronym stands for "get free
pages", the underlying memory allocation function.

Diversity of the allocation APIs combined with the numerous GFP flags
makes the question "How should I allocate memory?" not that easy to
answer, although very likely you should use

::

  kzalloc(<size>, GFP_KERNEL);

Of course there are cases when other allocation APIs and different GFP
flags must be used.

Get Free Page flags
===================

The GFP flags control the allocators behavior. They tell what memory
zones can be used, how hard the allocator should try to find free
memory, whether the memory can be accessed by the userspace etc. The
:ref:`Documentation/core-api/mm-api.rst <mm-api-gfp-flags>` provides
reference documentation for the GFP flags and their combinations and
here we briefly outline their recommended usage:

  * Most of the time ``GFP_KERNEL`` is what you need. Memory for the
    kernel data structures, DMAable memory, inode cache, all these and
    many other allocations types can use ``GFP_KERNEL``. Note, that
    using ``GFP_KERNEL`` implies ``GFP_RECLAIM``, which means that
    direct reclaim may be triggered under memory pressure; the calling
    context must be allowed to sleep.
  * If the allocation is performed from an atomic context, e.g interrupt
    handler, use ``GFP_NOWAIT``. This flag prevents direct reclaim and
    IO or filesystem operations. Consequently, under memory pressure
    ``GFP_NOWAIT`` allocation is likely to fail. Allocations which
    have a reasonable fallback should be using ``GFP_NOWARN``.
  * If you think that accessing memory reserves is justified and the kernel
    will be stressed unless allocation succeeds, you may use ``GFP_ATOMIC``.
  * Untrusted allocations triggered from userspace should be a subject
    of kmem accounting and must have ``__GFP_ACCOUNT`` bit set. There
    is the handy ``GFP_KERNEL_ACCOUNT`` shortcut for ``GFP_KERNEL``
    allocations that should be accounted.
  * Userspace allocations should use either of the ``GFP_USER``,
    ``GFP_HIGHUSER`` or ``GFP_HIGHUSER_MOVABLE`` flags. The longer
    the flag name the less restrictive it is.

    ``GFP_HIGHUSER_MOVABLE`` does not require that allocated memory
    will be directly accessible by the kernel and implies that the
    data is movable.

    ``GFP_HIGHUSER`` means that the allocated memory is not movable,
    but it is not required to be directly accessible by the kernel. An
    example may be a hardware allocation that maps data directly into
    userspace but has no addressing limitations.

    ``GFP_USER`` means that the allocated memory is not movable and it
    must be directly accessible by the kernel.

You may notice that quite a few allocations in the existing code
specify ``GFP_NOIO`` or ``GFP_NOFS``. Historically, they were used to
prevent recursion deadlocks caused by direct memory reclaim calling
back into the FS or IO paths and blocking on already held
resources. Since 4.12 the preferred way to address this issue is to
use new scope APIs described in
:ref:`Documentation/core-api/gfp_mask-from-fs-io.rst <gfp_mask_from_fs_io>`.

Other legacy GFP flags are ``GFP_DMA`` and ``GFP_DMA32``. They are
used to ensure that the allocated memory is accessible by hardware
with limited addressing capabilities. So unless you are writing a
driver for a device with such restrictions, avoid using these flags.
And even with hardware with restrictions it is preferable to use
`dma_alloc*` APIs.

GFP flags and reclaim behavior
------------------------------
Memory allocations may trigger direct or background reclaim and it is
useful to understand how hard the page allocator will try to satisfy that
or another request.

  * ``GFP_KERNEL & ~__GFP_RECLAIM`` - optimistic allocation without _any_
    attempt to free memory at all. The most light weight mode which even
    doesn't kick the background reclaim. Should be used carefully because it
    might deplete the memory and the next user might hit the more aggressive
    reclaim.

  * ``GFP_KERNEL & ~__GFP_DIRECT_RECLAIM`` (or ``GFP_NOWAIT``)- optimistic
    allocation without any attempt to free memory from the current
    context but can wake kswapd to reclaim memory if the zone is below
    the low watermark. Can be used from either atomic contexts or when
    the request is a performance optimization and there is another
    fallback for a slow path.

  * ``(GFP_KERNEL|__GFP_HIGH) & ~__GFP_DIRECT_RECLAIM`` (aka ``GFP_ATOMIC``) -
    non sleeping allocation with an expensive fallback so it can access
    some portion of memory reserves. Usually used from interrupt/bottom-half
    context with an expensive slow path fallback.

  * ``GFP_KERNEL`` - both background and direct reclaim are allowed and the
    **default** page allocator behavior is used. That means that not costly
    allocation requests are basically no-fail but there is no guarantee of
    that behavior so failures have to be checked properly by callers
    (e.g. OOM killer victim is allowed to fail currently).

  * ``GFP_KERNEL | __GFP_NORETRY`` - overrides the default allocator behavior
    and all allocation requests fail early rather than cause disruptive
    reclaim (one round of reclaim in this implementation). The OOM killer
    is not invoked.

  * ``GFP_KERNEL | __GFP_RETRY_MAYFAIL`` - overrides the default allocator
    behavior and all allocation requests try really hard. The request
    will fail if the reclaim cannot make any progress. The OOM killer
    won't be triggered.

  * ``GFP_KERNEL | __GFP_NOFAIL`` - overrides the default allocator behavior
    and all allocation requests will loop endlessly until they succeed.
    This might be really dangerous especially for larger orders.

Selecting memory allocator
==========================

The most straightforward way to allocate memory is to use a function
from the kmalloc() family. And, to be on the safe side it's best to use
routines that set memory to zero, like kzalloc(). If you need to
allocate memory for an array, there are kmalloc_array() and kcalloc()
helpers. The helpers struct_size(), array_size() and array3_size() can
be used to safely calculate object sizes without overflowing.

The maximal size of a chunk that can be allocated with `kmalloc` is
limited. The actual limit depends on the hardware and the kernel
configuration, but it is a good practice to use `kmalloc` for objects
smaller than page size.

The address of a chunk allocated with `kmalloc` is aligned to at least
ARCH_KMALLOC_MINALIGN bytes.  For sizes which are a power of two, the
alignment is also guaranteed to be at least the respective size.

Chunks allocated with kmalloc() can be resized with krealloc(). Similarly
to kmalloc_array(): a helper for resizing arrays is provided in the form of
krealloc_array().

For large allocations you can use vmalloc() and vzalloc(), or directly
request pages from the page allocator. The memory allocated by `vmalloc`
and related functions is not physically contiguous.

If you are not sure whether the allocation size is too large for
`kmalloc`, it is possible to use kvmalloc() and its derivatives. It will
try to allocate memory with `kmalloc` and if the allocation fails it
will be retried with `vmalloc`. There are restrictions on which GFP
flags can be used with `kvmalloc`; please see kvmalloc_node() reference
documentation. Note that `kvmalloc` may return memory that is not
physically contiguous.

If you need to allocate many identical objects you can use the slab
cache allocator. The cache should be set up with kmem_cache_create() or
kmem_cache_create_usercopy() before it can be used. The second function
should be used if a part of the cache might be copied to the userspace.
After the cache is created kmem_cache_alloc() and its convenience
wrappers can allocate memory from that cache.

When the allocated memory is no longer needed it must be freed.

Objects allocated by `kmalloc` can be freed by `kfree` or `kvfree`. Objects
allocated by `kmem_cache_alloc` can be freed with `kmem_cache_free`, `kfree`
or `kvfree`, where the latter two might be more convenient thanks to not
needing the kmem_cache pointer.

The same rules apply to _bulk and _rcu flavors of freeing functions.

Memory allocated by `vmalloc` can be freed with `vfree` or `kvfree`.
Memory allocated by `kvmalloc` can be freed with `kvfree`.
Caches created by `kmem_cache_create` should be freed with
`kmem_cache_destroy` only after freeing all the allocated objects first.