linux-5.10/mm/Kconfig

1 # SPDX-License-Identifier: GPL-2.0-only
3 menu "Memory Management options"
10 	prompt "Memory model"
16 	  This option allows you to change some of the ways that
17 	  Linux manages its memory internally. Most users will
22 	bool "Flat Memory"
25 	  This option is best suited for non-NUMA systems with
31 	  spaces and for features like NUMA and memory hotplug,
32 	  choose "Sparse Memory".
34 	  If unsure, choose this option (Flat Memory) over any other.
37 	bool "Discontiguous Memory"
41 	  memory systems, over FLATMEM.  These systems have holes
45 	  Although "Discontiguous Memory" is still used by several
47 	  "Sparse Memory".
49 	  If unsure, choose "Sparse Memory" over this option.
52 	bool "Sparse Memory"
56 	  memory hot-plug systems.  This is normal.
59 	  holes is their physical address space and allows memory
60 	  hot-plug and hot-remove.
62 	  If unsure, choose "Flat Memory" over this option.
84 # to represent different areas of memory.  This variable allows
85 # those dependencies to exist individually.
117 	bool "Sparse Memory virtual memmap"
121 	  SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise
132 # Don't discard allocated memory used to track "memory" and "reserved" memblocks
133 # after early boot, so it can still be used to test for validity of memory.
134 # Also, memblocks are updated with memory hot(un)plug.
138 # Keep arch NUMA mapping infrastructure post-init.
146 # Only be set on architectures that have completely implemented memory hotplug
154 	bool "Allow for memory hot-add"
166 	bool "Online the newly added memory blocks by default"
169 	  This option sets the default policy setting for memory hotplug
170 	  onlining policy (/sys/devices/system/memory/auto_online_blocks) which
171 	  determines what happens to newly added memory regions. Policy setting
173 	  See Documentation/admin-guide/mm/memory-hotplug.rst for more information.
175 	  Say Y here if you want all hot-plugged memory blocks to appear in
177 	  Say N here if you want the default policy to keep all hot-plugged
178 	  memory blocks in 'offline' state.
181 	bool "Allow for memory hot remove"
186 # Heavily threaded applications may benefit from splitting the mm-wide
189 # Default to 4 for wider testing, though 8 might be more appropriate.
190 # ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock.
191 # PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes.
193 # a per-page lock leads to problems when multiple tables need to be locked
209 # support for memory balloon
214 # support for memory balloon compaction
216 	bool "Allow for balloon memory compaction/migration"
220 	  Memory fragmentation introduced by ballooning might reduce
221 	  significantly the number of 2MB contiguous memory blocks that can be
224 	  by the guest workload. Allowing the compaction & migration for memory
225 	  pages enlisted as being part of memory balloon devices avoids the
226 	  scenario aforementioned and helps improving memory defragmentation.
229 # support for memory compaction
231 	bool "Allow for memory compaction"
236 	  Compaction is the only memory management component to form
237 	  high order (larger physically contiguous) memory blocks
239 	  the lack of the feature can lead to unexpected OOM killer
240 	  invocations for high order memory requests. You shouldn't
242 	  it and then we would be really interested to hear about that at
243 	  linux-mm@kvack.org.
253 	  those pages to another entity, such as a hypervisor, so that the
254 	  memory can be freed within the host for other uses.
266 	  two situations. The first is on NUMA systems to put pages nearer
267 	  to the processors accessing. The second is when allocating huge
268 	  pages as migration can relocate pages to satisfy a huge page
289 	  the full range of memory available to the CPU. Enabled
291 	  may say n to override this.
315 	  saving memory until one or another app needs to modify the content.
319 	  root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set).
322 	int "Low address space to protect from user allocation"
326 	  This is the portion of low virtual memory which should be protected
327 	  from userspace allocation.  Keeping a user from writing to low pages
333 	  Programs which use vm86 functionality or have some need to map
335 	  protection by setting the value to 0.
346 	bool "Enable recovery from hardware memory errors"
350 	  Enables code to recover from some memory failures on systems
351 	  with MCA recovery. This allows a system to continue running
352 	  even when some of its memory has uncorrected errors. This requires
353 	  special hardware support and typically ECC memory.
365 	  The NOMMU mmap() frequently needs to allocate large contiguous chunks
366 	  of memory on which to store mappings, but it can only ask the system
367 	  allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently
368 	  more than it requires.  To deal with this, mmap() is able to trim off
369 	  the excess and return it to the allocator.
371 	  If trimming is enabled, the excess is trimmed off and returned to the
376 	  long-term mappings means that the space is wasted.
381 	  no trimming is to occur.
386 	  See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
394 	  Transparent Hugepages allows the kernel to use huge pages and
395 	  huge tlb transparently to the applications whenever possible.
396 	  This feature can improve computing performance to certain
397 	  applications by speeding up page faults during memory
401 	  If memory constrained on embedded, you may want to say N.
414 	  memory footprint of applications without a guaranteed
421 	  performance improvement benefit to the applications using
422 	  madvise(MADV_HUGEPAGE) but it won't risk to increase the
423 	  memory footprint of applications without a guaranteed
449 	bool "Enable cleancache driver to cache clean pages if tmem is present"
451 	  Cleancache can be thought of as a page-granularity victim cache
453 	  (PFRA) would like to keep around, but can't since there isn't enough
454 	  memory.  So when the PFRA "evicts" a page, it first attempts to use
455 	  cleancache code to put the data contained in that page into
456 	  "transcendent memory", memory that is not directly accessible or
458 	  time-varying size.  And when a cleancache-enabled
459 	  filesystem wishes to access a page in a file on disk, it first
460 	  checks cleancache to see if it already contains it; if it does,
462 	  When a transcendent memory driver is available (such as zcache or
463 	  Xen transcendent memory), a significant I/O reduction
465 	  are reduced to a single pointer-compare-against-NULL resulting
468 	  If unsure, say Y to enable cleancache
471 	bool "Enable frontswap to cache swap pages if tmem is present"
476 	  "transcendent memory", memory that is not directly accessible or
478 	  time-varying size.  When space in transcendent memory is available,
480 	  available, all frontswap calls are reduced to a single pointer-
481 	  compare-against-NULL resulting in a negligible performance hit
484 	  If unsure, say Y to enable frontswap.
487 	bool "Contiguous Memory Allocator"
492 	  This enables the Contiguous Memory Allocator which allows other
493 	  subsystems to allocate big physically-contiguous blocks of memory.
494 	  CMA reserves a region of memory and allows only movable pages to
495 	  be allocated from it. This way, the kernel can use the memory for
497 	  allocated pages are migrated away to serve the contiguous request.
522 	  CMA allows to create CMA areas for particular purpose, mainly,
529 	bool "Track memory changes"
533 	  This option enables memory changes tracking by introducing a
534 	  soft-dirty bit on pte-s. This bit it set when someone writes
538 	  See Documentation/admin-guide/mm/soft-dirty.rst for more details.
546 	  pages that are in the process of being swapped out and attempts to
547 	  compress them into a dynamically allocated RAM-based memory pool.
553 	  v3.11) that interacts heavily with memory reclaim.  While these
554 	  interactions don't cause any known issues on simple memory setups,
567 	  a particular compression algorithm please refer to the benchmarks
676 	tristate "Common API for compressed memory storage"
678 	  Compressed memory storage API.  This allows using either zbud or
682 	tristate "Low (Up to 2x) density storage for compressed pages"
685 	  It is designed to store up to two compressed pages per physical
687 	  deterministic reclaim properties that make it preferable to a higher
691 	tristate "Up to 3x density storage for compressed pages"
695 	  It is designed to store up to three compressed pages per physical
700 	tristate "Memory allocator for compressed pages"
703 	  zsmalloc is a slab-based memory allocator designed to store
704 	  compressed RAM pages.  zsmalloc uses virtual memory mapping
705 	  in order to reduce fragmentation.  However, this results in a
706 	  non-standard allocator interface where a handle, not a pointer, is
707 	  returned by an alloc().  This handle must be mapped in order to
715 	  This option enables code in the zsmalloc to collect various
717 	  information to userspace via debugfs.
724 	int "Maximum user stack size for 32-bit processes (MB)"
729 	  This is the maximum stack size in Megabytes in the VM layout of 32-bit
731 	  arch). The stack will be located at the highest memory address minus
732 	  the given value, unless the RLIMIT_STACK hard limit is changed to a
738 	bool "Defer initialisation of struct pages to kthreads"
757 	  This feature allows to estimate the amount of user pages that have
759 	  be useful to tune memory cgroup limits and/or for job placement
762 	  See Documentation/admin-guide/mm/idle_page_tracking.rst for
769 	bool "Device memory (pmem, HMM, etc...) hotplug support"
777 	  Device memory hotplug support allows for establishing pmem,
778 	  or other device driver discovered memory regions, in the
780 	  "device-physical" addresses which is needed for using a DAX
789 # Helpers to mirror range of the CPU page tables of a process into device page
797 	bool "Unaddressable device memory (GPU memory, ...)"
802 	  Allows creation of struct pages to represent unaddressable device
803 	  memory; i.e., memory that is only accessible from the device (or
804 	  group of devices). You likely also want to select HMM_MIRROR.
818 	bool "Collect percpu memory statistics"
822 	  be used to help understand percpu memory usage.
836 	bool "Read-only THP for filesystems (EXPERIMENTAL)"
840 	  Allow khugepaged to put read-only file-backed pages in THP.
851 # required to support multiple hugepage sizes. For example a4fe3ce76