man2/set_mempolicy.2

   1 .\" Copyright 2003,2004 Andi Kleen, SuSE Labs.
   2 .\" and Copyright 2007 Lee Schermerhorn, Hewlett Packard
   3 .\"
   4 .\" %%%LICENSE_START(VERBATIM_PROF)
   5 .\" Permission is granted to make and distribute verbatim copies of this
   6 .\" manual provided the copyright notice and this permission notice are
   7 .\" preserved on all copies.
   8 .\"
   9 .\" Permission is granted to copy and distribute modified versions of this
  10 .\" manual under the conditions for verbatim copying, provided that the
  11 .\" entire resulting derived work is distributed under the terms of a
  12 .\" permission notice identical to this one.
  13 .\"
  14 .\" Since the Linux kernel and libraries are constantly changing, this
  15 .\" manual page may be incorrect or out-of-date.  The author(s) assume no
  16 .\" responsibility for errors or omissions, or for damages resulting from
  17 .\" the use of the information contained herein.
  18 .\"
  19 .\" Formatted or processed versions of this manual, if unaccompanied by
  20 .\" the source, must acknowledge the copyright and authors of this work.
  21 .\" %%%LICENSE_END
  22 .\"
  23 .\" 2006-02-03, mtk, substantial wording changes and other improvements
  24 .\" 2007-08-27, Lee Schermerhorn <Lee.Schermerhorn@hp.com>
  25 .\"     more precise specification of behavior.
  26 .\"
  27 .TH SET_MEMPOLICY 2 2015-05-07 Linux "Linux Programmer's Manual"
  28 .SH NAME
  29 set_mempolicy \- set default NUMA memory policy for a thread and its children
  30 .SH SYNOPSIS
  31 .nf
  32 .B "#include <numaif.h>"
  33 .sp
  34 .BI "long set_mempolicy(int " mode ", const unsigned long *" nodemask ,
  35 .BI "                   unsigned long " maxnode );
  36 .sp
  37 Link with \fI\-lnuma\fP.
  38 .fi
  39 .SH DESCRIPTION
  40 .BR set_mempolicy ()
  41 sets the NUMA memory policy of the calling thread,
  42 which consists of a policy mode and zero or more nodes,
  43 to the values specified by the
  44 .IR mode ,
  45 .I nodemask
  46 and
  47 .I maxnode
  48 arguments.
  49
  50 A NUMA machine has different
  51 memory controllers with different distances to specific CPUs.
  52 The memory policy defines from which node memory is allocated for
  53 the thread.
  54
  55 This system call defines the default policy for the thread.
  56 The thread policy governs allocation of pages in the process's
  57 address space outside of memory ranges
  58 controlled by a more specific policy set by
  59 .BR mbind (2).
  60 The thread default policy also controls allocation of any pages for
  61 memory-mapped files mapped using the
  62 .BR mmap (2)
  63 call with the
  64 .B MAP_PRIVATE
  65 flag and that are only read [loaded] from by the thread
  66 and of memory-mapped files mapped using the
  67 .BR mmap (2)
  68 call with the
  69 .B MAP_SHARED
  70 flag, regardless of the access type.
  71 The policy is applied only when a new page is allocated
  72 for the thread.
  73 For anonymous memory this is when the page is first
  74 touched by the thread.
  75
  76 The
  77 .I mode
  78 argument must specify one of
  79 .BR MPOL_DEFAULT ,
  80 .BR MPOL_BIND ,
  81 .BR MPOL_INTERLEAVE ,
  82 .BR MPOL_PREFERRED ,
  83 or
  84 .BR MPOL_LOCAL .
  85 All modes except
  86 .B MPOL_DEFAULT
  87 require the caller to specify via the
  88 .I nodemask
  89 argument one or more nodes.
  90
  91 The
  92 .I mode
  93 argument may also include an optional
  94 .IR "mode flag" .
  95 The supported
  96 .I "mode flags"
  97 are:
  98 .TP
  99 .BR MPOL_F_STATIC_NODES " (since Linux 2.6.26)"
 100 A nonempty
 101 .I nodemask
 102 specifies physical node ids.
 103 Linux will not remap the
 104 .I nodemask
 105 when the process moves to a different cpuset context,
 106 nor when the set of nodes allowed by the process's
 107 current cpuset context changes.
 108 .TP
 109 .BR MPOL_F_RELATIVE_NODES " (since Linux 2.6.26)"
 110 A nonempty
 111 .I nodemask
 112 specifies node ids that are relative to the set of
 113 node ids allowed by the process's current cpuset.
 114 .PP
 115 .I nodemask
 116 points to a bit mask of node IDs that contains up to
 117 .I maxnode
 118 bits.
 119 The bit mask size is rounded to the next multiple of
 120 .IR "sizeof(unsigned long)" ,
 121 but the kernel will use bits only up to
 122 .IR maxnode .
 123 A NULL value of
 124 .I nodemask
 125 or a
 126 .I maxnode
 127 value of zero specifies the empty set of nodes.
 128 If the value of
 129 .I maxnode
 130 is zero,
 131 the
 132 .I nodemask
 133 argument is ignored.
 134
 135 Where a
 136 .I nodemask
 137 is required, it must contain at least one node that is on-line,
 138 allowed by the process's current cpuset context,
 139 [unless the
 140 .B MPOL_F_STATIC_NODES
 141 mode flag is specified],
 142 and contains memory.
 143 If the
 144 .B MPOL_F_STATIC_NODES
 145 is set in
 146 .I mode
 147 and a required
 148 .I nodemask
 149 contains no nodes that are allowed by the process's current cpuset context,
 150 the memory policy reverts to
 151 .IR "local allocation" .
 152 This effectively overrides the specified policy until the process's
 153 cpuset context includes one or more of the nodes specified by
 154 .IR nodemask .
 155
 156 The
 157 .B MPOL_DEFAULT
 158 mode specifies that any nondefault thread memory policy be removed,
 159 so that the memory policy "falls back" to the system default policy.
 160 The system default policy is "local allocation"\(emthat is,
 161 allocate memory on the node of the CPU that triggered the allocation.
 162 .I nodemask
 163 must be specified as NULL.
 164 If the "local node" contains no free memory, the system will
 165 attempt to allocate memory from a "near by" node.
 166
 167 The
 168 .B MPOL_BIND
 169 mode defines a strict policy that restricts memory allocation to the
 170 nodes specified in
 171 .IR nodemask .
 172 If
 173 .I nodemask
 174 specifies more than one node, page allocations will come from
 175 the node with the lowest numeric node ID first, until that node
 176 contains no free memory.
 177 Allocations will then come from the node with the next highest
 178 node ID specified in
 179 .I nodemask
 180 and so forth, until none of the specified nodes contain free memory.
 181 Pages will not be allocated from any node not specified in the
 182 .IR nodemask .
 183
 184 .B MPOL_INTERLEAVE
 185 interleaves page allocations across the nodes specified in
 186 .I nodemask
 187 in numeric node ID order.
 188 This optimizes for bandwidth instead of latency
 189 by spreading out pages and memory accesses to those pages across
 190 multiple nodes.
 191 However, accesses to a single page will still be limited to
 192 the memory bandwidth of a single node.
 193 .\" NOTE:  the following sentence doesn't make sense in the context
 194 .\" of set_mempolicy() -- no memory area specified.
 195 .\" To be effective the memory area should be fairly large,
 196 .\" at least 1MB or bigger.
 197
 198 .B MPOL_PREFERRED
 199 sets the preferred node for allocation.
 200 The kernel will try to allocate pages from this node first
 201 and fall back to "near by" nodes if the preferred node is low on free
 202 memory.
 203 If
 204 .I nodemask
 205 specifies more than one node ID, the first node in the
 206 mask will be selected as the preferred node.
 207 If the
 208 .I nodemask
 209 and
 210 .I maxnode
 211 arguments specify the empty set, then the policy
 212 specifies "local allocation"
 213 (like the system default policy discussed above).
 214
 215 .B MPOL_LOCAL
 216 specifies the "local allocation", the memory is allocated on
 217 the node of the CPU that triggered the allocation, "local node".
 218 The
 219 .I nodemask
 220 and
 221 .I maxnode
 222 arguments must specify the empty set. If the "local node" is low
 223 on free memory the kernel will try to allocate memory from other
 224 nodes. The kernel will allocate memory from the "local node"
 225 whenever memory for this node is available. If the "local node"
 226 is not allowed by the process's current cpuset context the kernel
 227 will try to allocate memory from other nodes. The kernel will
 228 allocate memory from the "local node" whenever it becomes allowed
 229 by the process's current cpuset context.
 230
 231 The thread memory policy is preserved across an
 232 .BR execve (2),
 233 and is inherited by child threads created using
 234 .BR fork (2)
 235 or
 236 .BR clone (2).
 237 .SH RETURN VALUE
 238 On success,
 239 .BR set_mempolicy ()
 240 returns 0;
 241 on error, \-1 is returned and
 242 .I errno
 243 is set to indicate the error.
 244 .SH ERRORS
 245 .TP
 246 .B EFAULT
 247 Part of all of the memory range specified by
 248 .I nodemask
 249 and
 250 .I maxnode
 251 points outside your accessible address space.
 252 .TP
 253 .B EINVAL
 254 .I mode
 255 is invalid.
 256 Or,
 257 .I mode
 258 is
 259 .B MPOL_DEFAULT
 260 and
 261 .I nodemask
 262 is nonempty,
 263 or
 264 .I mode
 265 is
 266 .B MPOL_BIND
 267 or
 268 .B MPOL_INTERLEAVE
 269 and
 270 .I nodemask
 271 is empty.
 272 Or,
 273 .I maxnode
 274 specifies more than a page worth of bits.
 275 Or,
 276 .I nodemask
 277 specifies one or more node IDs that are
 278 greater than the maximum supported node ID.
 279 Or, none of the node IDs specified by
 280 .I nodemask
 281 are on-line and allowed by the process's current cpuset context,
 282 or none of the specified nodes contain memory.
 283 Or, the
 284 .I mode
 285 argument specified both
 286 .B MPOL_F_STATIC_NODES
 287 and
 288 .BR MPOL_F_RELATIVE_NODES .
 289 .TP
 290 .B ENOMEM
 291 Insufficient kernel memory was available.
 292 .SH VERSIONS
 293 The
 294 .BR set_mempolicy (),
 295 system call was added to the Linux kernel in version 2.6.7.
 296 .SH CONFORMING TO
 297 This system call is Linux-specific.
 298 .SH NOTES
 299 Memory policy is not remembered if the page is swapped out.
 300 When such a page is paged back in, it will use the policy of
 301 the thread or memory range that is in effect at the time the
 302 page is allocated.
 303
 304 For information on library support, see
 305 .BR numa (7).
 306 .SH SEE ALSO
 307 .BR get_mempolicy (2),
 308 .BR getcpu (2),
 309 .BR mbind (2),
 310 .BR mmap (2),
 311 .BR numa (3),
 312 .BR cpuset (7),
 313 .BR numa (7),
 314 .BR numactl (8)