Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 1 | config SELECT_MEMORY_MODEL |
| 2 | def_bool y |
| 3 | depends on EXPERIMENTAL || ARCH_SELECT_MEMORY_MODEL |
| 4 | |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 5 | choice |
| 6 | prompt "Memory model" |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 7 | depends on SELECT_MEMORY_MODEL |
| 8 | default DISCONTIGMEM_MANUAL if ARCH_DISCONTIGMEM_DEFAULT |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 9 | default SPARSEMEM_MANUAL if ARCH_SPARSEMEM_DEFAULT |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 10 | default FLATMEM_MANUAL |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 11 | |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 12 | config FLATMEM_MANUAL |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 13 | bool "Flat Memory" |
Anton Blanchard | c898ec1 | 2006-01-06 00:12:07 -0800 | [diff] [blame] | 14 | depends on !(ARCH_DISCONTIGMEM_ENABLE || ARCH_SPARSEMEM_ENABLE) || ARCH_FLATMEM_ENABLE |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 15 | help |
| 16 | This option allows you to change some of the ways that |
| 17 | Linux manages its memory internally. Most users will |
| 18 | only have one option here: FLATMEM. This is normal |
| 19 | and a correct option. |
| 20 | |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 21 | Some users of more advanced features like NUMA and |
| 22 | memory hotplug may have different options here. |
| 23 | DISCONTIGMEM is an more mature, better tested system, |
| 24 | but is incompatible with memory hotplug and may suffer |
| 25 | decreased performance over SPARSEMEM. If unsure between |
| 26 | "Sparse Memory" and "Discontiguous Memory", choose |
| 27 | "Discontiguous Memory". |
| 28 | |
| 29 | If unsure, choose this option (Flat Memory) over any other. |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 30 | |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 31 | config DISCONTIGMEM_MANUAL |
Dave Hansen | f3519f9 | 2005-09-16 19:27:54 -0700 | [diff] [blame] | 32 | bool "Discontiguous Memory" |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 33 | depends on ARCH_DISCONTIGMEM_ENABLE |
| 34 | help |
Dave Hansen | 785dcd4 | 2005-06-23 00:07:50 -0700 | [diff] [blame] | 35 | This option provides enhanced support for discontiguous |
| 36 | memory systems, over FLATMEM. These systems have holes |
| 37 | in their physical address spaces, and this option provides |
| 38 | more efficient handling of these holes. However, the vast |
| 39 | majority of hardware has quite flat address spaces, and |
Philipp Marek | ad3d0a3 | 2007-10-20 02:46:58 +0200 | [diff] [blame] | 40 | can have degraded performance from the extra overhead that |
Dave Hansen | 785dcd4 | 2005-06-23 00:07:50 -0700 | [diff] [blame] | 41 | this option imposes. |
| 42 | |
| 43 | Many NUMA configurations will have this as the only option. |
| 44 | |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 45 | If unsure, choose "Flat Memory" over this option. |
| 46 | |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 47 | config SPARSEMEM_MANUAL |
| 48 | bool "Sparse Memory" |
| 49 | depends on ARCH_SPARSEMEM_ENABLE |
| 50 | help |
| 51 | This will be the only option for some systems, including |
| 52 | memory hotplug systems. This is normal. |
| 53 | |
| 54 | For many other systems, this will be an alternative to |
Dave Hansen | f3519f9 | 2005-09-16 19:27:54 -0700 | [diff] [blame] | 55 | "Discontiguous Memory". This option provides some potential |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 56 | performance benefits, along with decreased code complexity, |
| 57 | but it is newer, and more experimental. |
| 58 | |
| 59 | If unsure, choose "Discontiguous Memory" or "Flat Memory" |
| 60 | over this option. |
| 61 | |
Dave Hansen | 3a9da76 | 2005-06-23 00:07:42 -0700 | [diff] [blame] | 62 | endchoice |
| 63 | |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 64 | config DISCONTIGMEM |
| 65 | def_bool y |
| 66 | depends on (!SELECT_MEMORY_MODEL && ARCH_DISCONTIGMEM_ENABLE) || DISCONTIGMEM_MANUAL |
| 67 | |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 68 | config SPARSEMEM |
| 69 | def_bool y |
Russell King | 1a83e17 | 2009-10-26 16:50:12 -0700 | [diff] [blame] | 70 | depends on (!SELECT_MEMORY_MODEL && ARCH_SPARSEMEM_ENABLE) || SPARSEMEM_MANUAL |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 71 | |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 72 | config FLATMEM |
| 73 | def_bool y |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 74 | depends on (!DISCONTIGMEM && !SPARSEMEM) || FLATMEM_MANUAL |
| 75 | |
| 76 | config FLAT_NODE_MEM_MAP |
| 77 | def_bool y |
| 78 | depends on !SPARSEMEM |
Dave Hansen | e1785e8 | 2005-06-23 00:07:49 -0700 | [diff] [blame] | 79 | |
Dave Hansen | 93b7504 | 2005-06-23 00:07:47 -0700 | [diff] [blame] | 80 | # |
| 81 | # Both the NUMA code and DISCONTIGMEM use arrays of pg_data_t's |
| 82 | # to represent different areas of memory. This variable allows |
| 83 | # those dependencies to exist individually. |
| 84 | # |
| 85 | config NEED_MULTIPLE_NODES |
| 86 | def_bool y |
| 87 | depends on DISCONTIGMEM || NUMA |
Andy Whitcroft | af70536 | 2005-06-23 00:07:53 -0700 | [diff] [blame] | 88 | |
| 89 | config HAVE_MEMORY_PRESENT |
| 90 | def_bool y |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 91 | depends on ARCH_HAVE_MEMORY_PRESENT || SPARSEMEM |
Bob Picco | 802f192 | 2005-09-03 15:54:26 -0700 | [diff] [blame] | 92 | |
| 93 | # |
Bob Picco | 3e34726 | 2005-09-03 15:54:28 -0700 | [diff] [blame] | 94 | # SPARSEMEM_EXTREME (which is the default) does some bootmem |
Matt LaPlante | 84eb8d0 | 2006-10-03 22:53:09 +0200 | [diff] [blame] | 95 | # allocations when memory_present() is called. If this cannot |
Bob Picco | 3e34726 | 2005-09-03 15:54:28 -0700 | [diff] [blame] | 96 | # be done on your architecture, select this option. However, |
| 97 | # statically allocating the mem_section[] array can potentially |
| 98 | # consume vast quantities of .bss, so be careful. |
| 99 | # |
| 100 | # This option will also potentially produce smaller runtime code |
| 101 | # with gcc 3.4 and later. |
| 102 | # |
| 103 | config SPARSEMEM_STATIC |
Jan Beulich | 9ba1608 | 2008-10-15 22:01:38 -0700 | [diff] [blame] | 104 | bool |
Bob Picco | 3e34726 | 2005-09-03 15:54:28 -0700 | [diff] [blame] | 105 | |
| 106 | # |
Matt LaPlante | 44c0920 | 2006-10-03 22:34:14 +0200 | [diff] [blame] | 107 | # Architecture platforms which require a two level mem_section in SPARSEMEM |
Bob Picco | 802f192 | 2005-09-03 15:54:26 -0700 | [diff] [blame] | 108 | # must select this option. This is usually for architecture platforms with |
| 109 | # an extremely sparse physical address space. |
| 110 | # |
Bob Picco | 3e34726 | 2005-09-03 15:54:28 -0700 | [diff] [blame] | 111 | config SPARSEMEM_EXTREME |
| 112 | def_bool y |
| 113 | depends on SPARSEMEM && !SPARSEMEM_STATIC |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 114 | |
Andy Whitcroft | 29c7111 | 2007-10-16 01:24:14 -0700 | [diff] [blame] | 115 | config SPARSEMEM_VMEMMAP_ENABLE |
Jan Beulich | 9ba1608 | 2008-10-15 22:01:38 -0700 | [diff] [blame] | 116 | bool |
Andy Whitcroft | 29c7111 | 2007-10-16 01:24:14 -0700 | [diff] [blame] | 117 | |
Yinghai Lu | 9bdac91 | 2010-02-10 01:20:22 -0800 | [diff] [blame] | 118 | config SPARSEMEM_ALLOC_MEM_MAP_TOGETHER |
| 119 | def_bool y |
| 120 | depends on SPARSEMEM && X86_64 |
| 121 | |
Andy Whitcroft | 29c7111 | 2007-10-16 01:24:14 -0700 | [diff] [blame] | 122 | config SPARSEMEM_VMEMMAP |
Geoff Levand | a5ee6da | 2007-12-17 16:19:53 -0800 | [diff] [blame] | 123 | bool "Sparse Memory virtual memmap" |
| 124 | depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE |
| 125 | default y |
| 126 | help |
| 127 | SPARSEMEM_VMEMMAP uses a virtually mapped memmap to optimise |
| 128 | pfn_to_page and page_to_pfn operations. This is the most |
| 129 | efficient option when sufficient kernel resources are available. |
Andy Whitcroft | 29c7111 | 2007-10-16 01:24:14 -0700 | [diff] [blame] | 130 | |
Yinghai Lu | 95f72d1 | 2010-07-12 14:36:09 +1000 | [diff] [blame] | 131 | config HAVE_MEMBLOCK |
| 132 | boolean |
| 133 | |
Dave Hansen | 3947be1 | 2005-10-29 18:16:54 -0700 | [diff] [blame] | 134 | # eventually, we can have this option just 'select SPARSEMEM' |
| 135 | config MEMORY_HOTPLUG |
| 136 | bool "Allow for memory hot-add" |
Keith Mannthey | ec69acb | 2006-09-30 23:27:05 -0700 | [diff] [blame] | 137 | depends on SPARSEMEM || X86_64_ACPI_NUMA |
Andi Kleen | 6ad696d | 2009-11-17 14:06:22 -0800 | [diff] [blame] | 138 | depends on HOTPLUG && ARCH_ENABLE_MEMORY_HOTPLUG |
Kumar Gala | ed84a07 | 2009-10-16 07:21:36 +0000 | [diff] [blame] | 139 | depends on (IA64 || X86 || PPC_BOOK3S_64 || SUPERH || S390) |
Dave Hansen | 3947be1 | 2005-10-29 18:16:54 -0700 | [diff] [blame] | 140 | |
Keith Mannthey | ec69acb | 2006-09-30 23:27:05 -0700 | [diff] [blame] | 141 | config MEMORY_HOTPLUG_SPARSE |
| 142 | def_bool y |
| 143 | depends on SPARSEMEM && MEMORY_HOTPLUG |
| 144 | |
KAMEZAWA Hiroyuki | 0c0e619 | 2007-10-16 01:26:12 -0700 | [diff] [blame] | 145 | config MEMORY_HOTREMOVE |
| 146 | bool "Allow for memory hot remove" |
| 147 | depends on MEMORY_HOTPLUG && ARCH_ENABLE_MEMORY_HOTREMOVE |
| 148 | depends on MIGRATION |
| 149 | |
Christoph Lameter | e20b8cc | 2008-04-28 02:12:55 -0700 | [diff] [blame] | 150 | # |
| 151 | # If we have space for more page flags then we can enable additional |
| 152 | # optimizations and functionality. |
| 153 | # |
| 154 | # Regular Sparsemem takes page flag bits for the sectionid if it does not |
| 155 | # use a virtual memmap. Disable extended page flags for 32 bit platforms |
| 156 | # that require the use of a sectionid in the page flags. |
| 157 | # |
| 158 | config PAGEFLAGS_EXTENDED |
| 159 | def_bool y |
H. Peter Anvin | a269cca | 2009-08-31 11:17:44 -0700 | [diff] [blame] | 160 | depends on 64BIT || SPARSEMEM_VMEMMAP || !SPARSEMEM |
Christoph Lameter | e20b8cc | 2008-04-28 02:12:55 -0700 | [diff] [blame] | 161 | |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 162 | # Heavily threaded applications may benefit from splitting the mm-wide |
| 163 | # page_table_lock, so that faults on different parts of the user address |
| 164 | # space can be handled with less contention: split it at this NR_CPUS. |
| 165 | # Default to 4 for wider testing, though 8 might be more appropriate. |
| 166 | # ARM's adjust_pte (unused if VIPT) depends on mm-wide page_table_lock. |
Hugh Dickins | 7b6ac9d | 2005-11-23 13:37:37 -0800 | [diff] [blame] | 167 | # PA-RISC 7xxx's spinlock_t would enlarge struct page from 32 to 44 bytes. |
Hugh Dickins | a70caa8 | 2009-12-14 17:59:02 -0800 | [diff] [blame] | 168 | # DEBUG_SPINLOCK and DEBUG_LOCK_ALLOC spinlock_t also enlarge struct page. |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 169 | # |
| 170 | config SPLIT_PTLOCK_CPUS |
| 171 | int |
Hugh Dickins | a70caa8 | 2009-12-14 17:59:02 -0800 | [diff] [blame] | 172 | default "999999" if ARM && !CPU_CACHE_VIPT |
| 173 | default "999999" if PARISC && !PA20 |
| 174 | default "999999" if DEBUG_SPINLOCK || DEBUG_LOCK_ALLOC |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 175 | default "4" |
Christoph Lameter | 7cbe34c | 2006-01-08 01:00:49 -0800 | [diff] [blame] | 176 | |
| 177 | # |
Mel Gorman | e9e96b3 | 2010-05-24 14:32:21 -0700 | [diff] [blame] | 178 | # support for memory compaction |
| 179 | config COMPACTION |
| 180 | bool "Allow for memory compaction" |
| 181 | select MIGRATION |
| 182 | depends on EXPERIMENTAL && HUGETLB_PAGE && MMU |
| 183 | help |
| 184 | Allows the compaction of memory for the allocation of huge pages. |
| 185 | |
| 186 | # |
Christoph Lameter | 7cbe34c | 2006-01-08 01:00:49 -0800 | [diff] [blame] | 187 | # support for page migration |
| 188 | # |
| 189 | config MIGRATION |
Christoph Lameter | b20a350 | 2006-03-22 00:09:12 -0800 | [diff] [blame] | 190 | bool "Page migration" |
Christoph Lameter | 6c5240a | 2006-06-23 02:03:37 -0700 | [diff] [blame] | 191 | def_bool y |
Andrea Arcangeli | 152e065 | 2010-09-09 16:37:53 -0700 | [diff] [blame] | 192 | depends on NUMA || ARCH_ENABLE_MEMORY_HOTREMOVE || COMPACTION |
Christoph Lameter | b20a350 | 2006-03-22 00:09:12 -0800 | [diff] [blame] | 193 | help |
| 194 | Allows the migration of the physical location of pages of processes |
Mel Gorman | e9e96b3 | 2010-05-24 14:32:21 -0700 | [diff] [blame] | 195 | while the virtual addresses are not changed. This is useful in |
| 196 | two situations. The first is on NUMA systems to put pages nearer |
| 197 | to the processors accessing. The second is when allocating huge |
| 198 | pages as migration can relocate pages to satisfy a huge page |
| 199 | allocation instead of reclaiming. |
Greg Kroah-Hartman | 6550e07 | 2006-06-12 17:11:31 -0700 | [diff] [blame] | 200 | |
Jeremy Fitzhardinge | 600715d | 2008-09-11 01:31:45 -0700 | [diff] [blame] | 201 | config PHYS_ADDR_T_64BIT |
| 202 | def_bool 64BIT || ARCH_PHYS_ADDR_T_64BIT |
| 203 | |
Christoph Lameter | 4b51d66 | 2007-02-10 01:43:10 -0800 | [diff] [blame] | 204 | config ZONE_DMA_FLAG |
| 205 | int |
| 206 | default "0" if !ZONE_DMA |
| 207 | default "1" |
| 208 | |
Christoph Lameter | 2a7326b | 2007-07-17 04:03:37 -0700 | [diff] [blame] | 209 | config BOUNCE |
| 210 | def_bool y |
| 211 | depends on BLOCK && MMU && (ZONE_DMA || HIGHMEM) |
| 212 | |
Christoph Lameter | 6225e93 | 2007-05-06 14:49:50 -0700 | [diff] [blame] | 213 | config NR_QUICK |
| 214 | int |
| 215 | depends on QUICKLIST |
Paul Mundt | 0176bd3 | 2010-01-05 12:35:00 +0900 | [diff] [blame] | 216 | default "2" if AVR32 |
Christoph Lameter | 6225e93 | 2007-05-06 14:49:50 -0700 | [diff] [blame] | 217 | default "1" |
Stephen Rothwell | f057eac | 2007-07-15 23:40:05 -0700 | [diff] [blame] | 218 | |
| 219 | config VIRT_TO_BUS |
| 220 | def_bool y |
| 221 | depends on !ARCH_NO_VIRT_TO_BUS |
Andrea Arcangeli | cddb8a5 | 2008-07-28 15:46:29 -0700 | [diff] [blame] | 222 | |
| 223 | config MMU_NOTIFIER |
| 224 | bool |
David Howells | fc4d5c2 | 2009-05-06 16:03:05 -0700 | [diff] [blame] | 225 | |
Hugh Dickins | f8af4da | 2009-09-21 17:01:57 -0700 | [diff] [blame] | 226 | config KSM |
| 227 | bool "Enable KSM for page merging" |
| 228 | depends on MMU |
| 229 | help |
| 230 | Enable Kernel Samepage Merging: KSM periodically scans those areas |
| 231 | of an application's address space that an app has advised may be |
| 232 | mergeable. When it finds pages of identical content, it replaces |
Hugh Dickins | d0f209f | 2009-12-14 17:59:34 -0800 | [diff] [blame] | 233 | the many instances by a single page with that content, so |
Hugh Dickins | f8af4da | 2009-09-21 17:01:57 -0700 | [diff] [blame] | 234 | saving memory until one or another app needs to modify the content. |
| 235 | Recommended for use with KVM, or with other duplicative applications. |
Hugh Dickins | c73602a | 2009-10-07 16:32:22 -0700 | [diff] [blame] | 236 | See Documentation/vm/ksm.txt for more information: KSM is inactive |
| 237 | until a program has madvised that an area is MADV_MERGEABLE, and |
| 238 | root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set). |
Hugh Dickins | f8af4da | 2009-09-21 17:01:57 -0700 | [diff] [blame] | 239 | |
Christoph Lameter | e0a94c2 | 2009-06-03 16:04:31 -0400 | [diff] [blame] | 240 | config DEFAULT_MMAP_MIN_ADDR |
| 241 | int "Low address space to protect from user allocation" |
David Howells | 6e14154 | 2009-12-15 19:27:45 +0000 | [diff] [blame] | 242 | depends on MMU |
Christoph Lameter | e0a94c2 | 2009-06-03 16:04:31 -0400 | [diff] [blame] | 243 | default 4096 |
| 244 | help |
| 245 | This is the portion of low virtual memory which should be protected |
| 246 | from userspace allocation. Keeping a user from writing to low pages |
| 247 | can help reduce the impact of kernel NULL pointer bugs. |
| 248 | |
| 249 | For most ia64, ppc64 and x86 users with lots of address space |
| 250 | a value of 65536 is reasonable and should cause no problems. |
| 251 | On arm and other archs it should not be higher than 32768. |
Eric Paris | 788084a | 2009-07-31 12:54:11 -0400 | [diff] [blame] | 252 | Programs which use vm86 functionality or have some need to map |
| 253 | this low address space will need CAP_SYS_RAWIO or disable this |
| 254 | protection by setting the value to 0. |
Christoph Lameter | e0a94c2 | 2009-06-03 16:04:31 -0400 | [diff] [blame] | 255 | |
| 256 | This value can be changed after boot using the |
| 257 | /proc/sys/vm/mmap_min_addr tunable. |
| 258 | |
Linus Torvalds | d949f36 | 2009-09-26 09:35:07 -0700 | [diff] [blame] | 259 | config ARCH_SUPPORTS_MEMORY_FAILURE |
| 260 | bool |
Christoph Lameter | e0a94c2 | 2009-06-03 16:04:31 -0400 | [diff] [blame] | 261 | |
Andi Kleen | 6a46079 | 2009-09-16 11:50:15 +0200 | [diff] [blame] | 262 | config MEMORY_FAILURE |
| 263 | depends on MMU |
Linus Torvalds | d949f36 | 2009-09-26 09:35:07 -0700 | [diff] [blame] | 264 | depends on ARCH_SUPPORTS_MEMORY_FAILURE |
Andi Kleen | 6a46079 | 2009-09-16 11:50:15 +0200 | [diff] [blame] | 265 | bool "Enable recovery from hardware memory errors" |
| 266 | help |
| 267 | Enables code to recover from some memory failures on systems |
| 268 | with MCA recovery. This allows a system to continue running |
| 269 | even when some of its memory has uncorrected errors. This requires |
| 270 | special hardware support and typically ECC memory. |
| 271 | |
Andi Kleen | cae681f | 2009-09-16 11:50:17 +0200 | [diff] [blame] | 272 | config HWPOISON_INJECT |
Andi Kleen | 413f9ef | 2009-12-16 12:20:00 +0100 | [diff] [blame] | 273 | tristate "HWPoison pages injector" |
Andi Kleen | 27df506 | 2009-12-21 19:56:42 +0100 | [diff] [blame] | 274 | depends on MEMORY_FAILURE && DEBUG_KERNEL && PROC_FS |
Wu Fengguang | 478c5ff | 2009-12-16 12:19:59 +0100 | [diff] [blame] | 275 | select PROC_PAGE_MONITOR |
Andi Kleen | cae681f | 2009-09-16 11:50:17 +0200 | [diff] [blame] | 276 | |
David Howells | fc4d5c2 | 2009-05-06 16:03:05 -0700 | [diff] [blame] | 277 | config NOMMU_INITIAL_TRIM_EXCESS |
| 278 | int "Turn on mmap() excess space trimming before booting" |
| 279 | depends on !MMU |
| 280 | default 1 |
| 281 | help |
| 282 | The NOMMU mmap() frequently needs to allocate large contiguous chunks |
| 283 | of memory on which to store mappings, but it can only ask the system |
| 284 | allocator for chunks in 2^N*PAGE_SIZE amounts - which is frequently |
| 285 | more than it requires. To deal with this, mmap() is able to trim off |
| 286 | the excess and return it to the allocator. |
| 287 | |
| 288 | If trimming is enabled, the excess is trimmed off and returned to the |
| 289 | system allocator, which can cause extra fragmentation, particularly |
| 290 | if there are a lot of transient processes. |
| 291 | |
| 292 | If trimming is disabled, the excess is kept, but not used, which for |
| 293 | long-term mappings means that the space is wasted. |
| 294 | |
| 295 | Trimming can be dynamically controlled through a sysctl option |
| 296 | (/proc/sys/vm/nr_trim_pages) which specifies the minimum number of |
| 297 | excess pages there must be before trimming should occur, or zero if |
| 298 | no trimming is to occur. |
| 299 | |
| 300 | This option specifies the initial value of this option. The default |
| 301 | of 1 says that all excess pages should be trimmed. |
| 302 | |
| 303 | See Documentation/nommu-mmap.txt for more information. |
Tejun Heo | bbddff0 | 2010-09-03 18:22:48 +0200 | [diff] [blame] | 304 | |
Andrea Arcangeli | 4c76d9d | 2011-01-13 15:46:39 -0800 | [diff] [blame] | 305 | config TRANSPARENT_HUGEPAGE |
| 306 | bool "Transparent Hugepage Support" if EMBEDDED |
Johannes Weiner | f2d6bfe | 2011-01-13 15:47:01 -0800 | [diff] [blame] | 307 | depends on X86 && MMU |
Andrea Arcangeli | 4c76d9d | 2011-01-13 15:46:39 -0800 | [diff] [blame] | 308 | default y |
| 309 | help |
| 310 | Transparent Hugepages allows the kernel to use huge pages and |
| 311 | huge tlb transparently to the applications whenever possible. |
| 312 | This feature can improve computing performance to certain |
| 313 | applications by speeding up page faults during memory |
| 314 | allocation, by reducing the number of tlb misses and by speeding |
| 315 | up the pagetable walking. |
| 316 | |
| 317 | If memory constrained on embedded, you may want to say N. |
| 318 | |
Tejun Heo | bbddff0 | 2010-09-03 18:22:48 +0200 | [diff] [blame] | 319 | # |
| 320 | # UP and nommu archs use km based percpu allocator |
| 321 | # |
| 322 | config NEED_PER_CPU_KM |
| 323 | depends on !SMP |
| 324 | bool |
| 325 | default y |