]>
Commit | Line | Data |
---|---|---|
314093c9 | 1 | .\" Copyright 2003,2004 Andi Kleen, SuSE Labs. |
73ae0a09 | 2 | .\" and Copyright 2007 Lee Schermerhorn, Hewlett Packard |
314093c9 | 3 | .\" |
9f882130 | 4 | .\" %%%LICENSE_START(VERBATIM_PROF) |
314093c9 MK |
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. | |
c13182ef | 13 | .\" |
314093c9 MK |
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15 | .\" manual page may be incorrect or out-of-date.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 | |
c13182ef MK |
17 | .\" the use of the information contained herein. |
18 | .\" | |
314093c9 MK |
19 | .\" Formatted or processed versions of this manual, if unaccompanied by |
20 | .\" the source, must acknowledge the copyright and authors of this work. | |
9f882130 | 21 | .\" %%%LICENSE_END |
c13182ef | 22 | .\" |
314093c9 | 23 | .\" 2006-02-03, mtk, substantial wording changes and other improvements |
00045cbb MK |
24 | .\" 2007-08-27, Lee Schermerhorn <Lee.Schermerhorn@hp.com> |
25 | .\" more precise specification of behavior. | |
314093c9 | 26 | .\" |
adfbcbeb | 27 | .TH SET_MEMPOLICY 2 2008-08-15 Linux "Linux Programmer's Manual" |
314093c9 | 28 | .SH NAME |
73ae0a09 | 29 | set_mempolicy \- set default NUMA memory policy for a process and its children |
314093c9 | 30 | .SH SYNOPSIS |
521bf584 | 31 | .nf |
c13182ef | 32 | .B "#include <numaif.h>" |
314093c9 | 33 | .sp |
73ae0a09 | 34 | .BI "int set_mempolicy(int " mode ", unsigned long *" nodemask , |
521bf584 | 35 | .BI " unsigned long " maxnode ); |
73ae0a09 | 36 | .sp |
4ed3353d | 37 | Link with \fI\-lnuma\fP. |
521bf584 | 38 | .fi |
314093c9 MK |
39 | .SH DESCRIPTION |
40 | .BR set_mempolicy () | |
73ae0a09 MK |
41 | sets the NUMA memory policy of the calling process, |
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 | |
0daa9e92 | 47 | .I maxnode |
73ae0a09 | 48 | arguments. |
314093c9 MK |
49 | |
50 | A NUMA machine has different | |
51 | memory controllers with different distances to specific CPUs. | |
73ae0a09 | 52 | The memory policy defines from which node memory is allocated for |
c13182ef | 53 | the process. |
314093c9 | 54 | |
73ae0a09 | 55 | This system call defines the default policy for the process. |
ecccf7c2 | 56 | The process policy governs allocation of pages in the process's |
73ae0a09 MK |
57 | address space outside of memory ranges |
58 | controlled by a more specific policy set by | |
314093c9 | 59 | .BR mbind (2). |
73ae0a09 MK |
60 | The process 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 | |
cdba9253 | 65 | flag and that are only read [loaded] from by the process |
73ae0a09 MK |
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. | |
33a0ccb2 | 71 | The policy is applied only when a new page is allocated |
c13182ef MK |
72 | for the process. |
73 | For anonymous memory this is when the page is first | |
314093c9 MK |
74 | touched by the application. |
75 | ||
73ae0a09 MK |
76 | The |
77 | .I mode | |
78 | argument must specify one of | |
314093c9 MK |
79 | .BR MPOL_DEFAULT , |
80 | .BR MPOL_BIND , | |
73ae0a09 MK |
81 | .B MPOL_INTERLEAVE |
82 | or | |
314093c9 | 83 | .BR MPOL_PREFERRED . |
73ae0a09 | 84 | All modes except |
314093c9 | 85 | .B MPOL_DEFAULT |
73ae0a09 | 86 | require the caller to specify via the |
c13182ef | 87 | .I nodemask |
c4bb193f | 88 | argument one or more nodes. |
73ae0a09 | 89 | |
f98b728e MK |
90 | The |
91 | .I mode | |
92 | argument may also include an optional | |
adfbcbeb | 93 | .IR "mode flag" . |
f98b728e MK |
94 | The supported |
95 | .I "mode flags" | |
96 | are: | |
97 | .TP | |
98 | .BR MPOL_F_STATIC_NODES " (since Linux 2.6.26)" | |
aa796481 | 99 | A nonempty |
f98b728e MK |
100 | .I nodemask |
101 | specifies physical node ids. | |
102 | Linux does will not remap the | |
103 | .I nodemask | |
104 | when the process moves to a different cpuset context, | |
105 | nor when the set of nodes allowed by the process's | |
106 | current cpuset context changes. | |
107 | .TP | |
108 | .BR MPOL_F_RELATIVE_NODES " (since Linux 2.6.26)" | |
aa796481 | 109 | A nonempty |
f98b728e MK |
110 | .I nodemask |
111 | specifies node ids that are relative to the set of | |
4e836144 | 112 | node ids allowed by the process's current cpuset. |
f98b728e | 113 | .PP |
c13182ef | 114 | .I nodemask |
00045cbb | 115 | points to a bit mask of node IDs that contains up to |
314093c9 | 116 | .I maxnode |
c13182ef | 117 | bits. |
73ae0a09 | 118 | The bit mask size is rounded to the next multiple of |
c13182ef | 119 | .IR "sizeof(unsigned long)" , |
33a0ccb2 | 120 | but the kernel will use bits only up to |
314093c9 | 121 | .IR maxnode . |
73ae0a09 MK |
122 | A NULL value of |
123 | .I nodemask | |
124 | or a | |
125 | .I maxnode | |
126 | value of zero specifies the empty set of nodes. | |
127 | If the value of | |
128 | .I maxnode | |
129 | is zero, | |
130 | the | |
131 | .I nodemask | |
132 | argument is ignored. | |
f98b728e | 133 | |
cdba9253 MK |
134 | Where a |
135 | .I nodemask | |
136 | is required, it must contain at least one node that is on-line, | |
137 | allowed by the process's current cpuset context, | |
f98b728e MK |
138 | [unless the |
139 | .B MPOL_F_STATIC_NODES | |
140 | mode flag is specified], | |
cdba9253 | 141 | and contains memory. |
f98b728e MK |
142 | If the |
143 | .B MPOL_F_STATIC_NODES | |
144 | is set in | |
145 | .I mode | |
146 | and a required | |
147 | .I nodemask | |
148 | contains no nodes that are allowed by the process's current cpuset context, | |
149 | the memory policy reverts to | |
150 | .IR "local allocation" . | |
151 | This effectively overrides the specified policy until the process's | |
152 | cpuset context includes one or more of the nodes specified by | |
fe48639f | 153 | .IR nodemask . |
314093c9 | 154 | |
c13182ef | 155 | The |
314093c9 | 156 | .B MPOL_DEFAULT |
24b74457 | 157 | mode specifies that any nondefault process memory policy be removed, |
f98b728e | 158 | so that the memory policy "falls back" to the system default policy. |
88879aeb MK |
159 | The system default policy is "local allocation"\(emthat is, |
160 | allocate memory on the node of the CPU that triggered the allocation. | |
c13182ef | 161 | .I nodemask |
73ae0a09 MK |
162 | must be specified as NULL. |
163 | If the "local node" contains no free memory, the system will | |
164 | attempt to allocate memory from a "near by" node. | |
314093c9 MK |
165 | |
166 | The | |
167 | .B MPOL_BIND | |
73ae0a09 | 168 | mode defines a strict policy that restricts memory allocation to the |
c13182ef | 169 | nodes specified in |
314093c9 | 170 | .IR nodemask . |
73ae0a09 MK |
171 | If |
172 | .I nodemask | |
173 | specifies more than one node, page allocations will come from | |
00045cbb | 174 | the node with the lowest numeric node ID first, until that node |
73ae0a09 MK |
175 | contains no free memory. |
176 | Allocations will then come from the node with the next highest | |
00045cbb | 177 | node ID specified in |
73ae0a09 MK |
178 | .I nodemask |
179 | and so forth, until none of the specified nodes contain free memory. | |
180 | Pages will not be allocated from any node not specified in the | |
181 | .IR nodemask . | |
314093c9 MK |
182 | |
183 | .B MPOL_INTERLEAVE | |
73ae0a09 MK |
184 | interleaves page allocations across the nodes specified in |
185 | .I nodemask | |
00045cbb | 186 | in numeric node ID order. |
73ae0a09 MK |
187 | This optimizes for bandwidth instead of latency |
188 | by spreading out pages and memory accesses to those pages across | |
189 | multiple nodes. | |
190 | However, accesses to a single page will still be limited to | |
191 | the memory bandwidth of a single node. | |
192 | .\" NOTE: the following sentence doesn't make sense in the context | |
193 | .\" of set_mempolicy() -- no memory area specified. | |
194 | .\" To be effective the memory area should be fairly large, | |
195 | .\" at least 1MB or bigger. | |
314093c9 MK |
196 | |
197 | .B MPOL_PREFERRED | |
c13182ef | 198 | sets the preferred node for allocation. |
73ae0a09 MK |
199 | The kernel will try to allocate pages from this node first |
200 | and fall back to "near by" nodes if the preferred node is low on free | |
c13182ef | 201 | memory. |
73ae0a09 MK |
202 | If |
203 | .I nodemask | |
00045cbb | 204 | specifies more than one node ID, the first node in the |
73ae0a09 MK |
205 | mask will be selected as the preferred node. |
206 | If the | |
c13182ef | 207 | .I nodemask |
73ae0a09 MK |
208 | and |
209 | .I maxnode | |
1313d297 MK |
210 | arguments specify the empty set, then the policy |
211 | specifies "local allocation" | |
212 | (like the system default policy discussed above). | |
314093c9 | 213 | |
73ae0a09 | 214 | The process memory policy is preserved across an |
3bd6a9b1 MK |
215 | .BR execve (2), |
216 | and is inherited by child processes created using | |
c13182ef MK |
217 | .BR fork (2) |
218 | or | |
314093c9 | 219 | .BR clone (2). |
314093c9 MK |
220 | .SH RETURN VALUE |
221 | On success, | |
222 | .BR set_mempolicy () | |
223 | returns 0; | |
224 | on error, \-1 is returned and | |
c13182ef | 225 | .I errno |
314093c9 | 226 | is set to indicate the error. |
73ae0a09 MK |
227 | .SH ERRORS |
228 | .TP | |
b3a7b55e MK |
229 | .B EFAULT |
230 | Part of all of the memory range specified by | |
231 | .I nodemask | |
232 | and | |
233 | .I maxnode | |
234 | points outside your accessible address space. | |
235 | .TP | |
73ae0a09 | 236 | .B EINVAL |
4d2be0ee MK |
237 | .I mode |
238 | is invalid. | |
73ae0a09 MK |
239 | Or, |
240 | .I mode | |
241 | is | |
00045cbb | 242 | .B MPOL_DEFAULT |
73ae0a09 MK |
243 | and |
244 | .I nodemask | |
aa796481 | 245 | is nonempty, |
73ae0a09 MK |
246 | or |
247 | .I mode | |
248 | is | |
00045cbb | 249 | .B MPOL_BIND |
73ae0a09 | 250 | or |
00045cbb | 251 | .B MPOL_INTERLEAVE |
73ae0a09 MK |
252 | and |
253 | .I nodemask | |
254 | is empty. | |
255 | Or, | |
256 | .I maxnode | |
257 | specifies more than a page worth of bits. | |
258 | Or, | |
259 | .I nodemask | |
00045cbb | 260 | specifies one or more node IDs that are |
cdba9253 | 261 | greater than the maximum supported node ID. |
00045cbb | 262 | Or, none of the node IDs specified by |
73ae0a09 | 263 | .I nodemask |
cdba9253 MK |
264 | are on-line and allowed by the process's current cpuset context, |
265 | or none of the specified nodes contain memory. | |
f98b728e MK |
266 | Or, the |
267 | .I mode | |
268 | argument specified both | |
269 | .B MPOL_F_STATIC_NODES | |
270 | and | |
271 | .BR MPOL_F_RELATIVE_NODES . | |
73ae0a09 | 272 | .TP |
73ae0a09 MK |
273 | .B ENOMEM |
274 | Insufficient kernel memory was available. | |
adfbcbeb MK |
275 | .SH VERSIONS |
276 | The | |
277 | .BR set_mempolicy (), | |
278 | system call was added to the Linux kernel in version 2.6.7. | |
9d9dc1e8 | 279 | .SH CONFORMING TO |
8382f16d | 280 | This system call is Linux-specific. |
a1d5f77c MK |
281 | .SH NOTES |
282 | Process policy is not remembered if the page is swapped out. | |
73ae0a09 MK |
283 | When such a page is paged back in, it will use the policy of |
284 | the process or memory range that is in effect at the time the | |
285 | page is allocated. | |
adfbcbeb MK |
286 | |
287 | For information on library support, see | |
288 | .BR numa (7). | |
314093c9 | 289 | .SH SEE ALSO |
fa23e023 | 290 | .BR get_mempolicy (2), |
f0c34053 | 291 | .BR getcpu (2), |
314093c9 | 292 | .BR mbind (2), |
73ae0a09 | 293 | .BR mmap (2), |
a18e2edb MK |
294 | .BR numa (3), |
295 | .BR cpuset (7), | |
adfbcbeb | 296 | .BR numa (7), |
a18e2edb | 297 | .BR numactl (8) |