1SET_MEMPOLICY(2)           Linux Programmer's Manual          SET_MEMPOLICY(2)


6       set_mempolicy  -  set  default  NUMA memory policy for a thread and its
7       children


10       #include <numaif.h>
12       long set_mempolicy(int mode, const unsigned long *nodemask,
13                          unsigned long maxnode);
15       Link with -lnuma.


18       set_mempolicy() sets the NUMA memory  policy  of  the  calling  thread,
19       which  consists  of a policy mode and zero or more nodes, to the values
20       specified by the mode, nodemask and maxnode arguments.
22       A NUMA machine has different memory  controllers  with  different  dis‐
23       tances  to  specific  CPUs.   The memory policy defines from which node
24       memory is allocated for the thread.
26       This system call defines the default policy for the thread.  The thread
27       policy  governs allocation of pages in the process's address space out‐
28       side of memory ranges controlled by  a  more  specific  policy  set  by
29       mbind(2).   The  thread  default policy also controls allocation of any
30       pages for memory-mapped files mapped using the mmap(2)  call  with  the
31       MAP_PRIVATE flag and that are only read (loaded) from by the thread and
32       of  memory-mapped  files  mapped  using  the  mmap(2)  call  with   the
33       MAP_SHARED  flag, regardless of the access type.  The policy is applied
34       only when a new page is allocated for the thread.  For anonymous memory
35       this is when the page is first touched by the thread.
37       The   mode  argument  must  specify  one  of  MPOL_DEFAULT,  MPOL_BIND,
38       MPOL_INTERLEAVE, MPOL_PREFERRED, or MPOL_LOCAL (which are described  in
39       detail  below).   All  modes  except MPOL_DEFAULT require the caller to
40       specify the node or nodes to which the mode applies, via  the  nodemask
41       argument.
43       The  mode  argument  may  also include an optional mode flag.  The sup‐
44       ported mode flags are:
46       MPOL_F_STATIC_NODES (since Linux 2.6.26)
47              A nonempty nodemask specifies physical node IDs.  Linux will not
48              remap  the nodemask when the process moves to a different cpuset
49              context, nor when the set of nodes allowed by the process's cur‐
50              rent cpuset context changes.
52       MPOL_F_RELATIVE_NODES (since Linux 2.6.26)
53              A  nonempty nodemask specifies node IDs that are relative to the
54              set of node IDs allowed by the process's current cpuset.
56       nodemask points to a bit mask of node IDs that contains up  to  maxnode
57       bits.    The  bit  mask  size  is  rounded  to  the  next  multiple  of
58       sizeof(unsigned long), but the kernel will use bits only up to maxnode.
59       A NULL value of nodemask or a maxnode value of zero specifies the empty
60       set of nodes.  If the value of maxnode is zero, the  nodemask  argument
61       is ignored.
63       Where a nodemask is required, it must contain at least one node that is
64       on-line, allowed by the process's current cpuset context,  (unless  the
65       MPOL_F_STATIC_NODES  mode  flag is specified), and contains memory.  If
66       the MPOL_F_STATIC_NODES is set in mode and a required nodemask contains
67       no  nodes that are allowed by the process's current cpuset context, the
68       memory policy reverts to local allocation.  This effectively  overrides
69       the specified policy until the process's cpuset context includes one or
70       more of the nodes specified by nodemask.
72       The mode argument must include one of the following values:
75              This mode specifies that any nondefault thread memory policy  be
76              removed,  so  that  the memory policy "falls back" to the system
77              default policy.  The system default  policy  is  "local  alloca‐
78              tion"—that is, allocate memory on the node of the CPU that trig‐
79              gered the allocation.  nodemask must be specified as  NULL.   If
80              the  "local  node"  contains  no  free  memory,  the system will
81              attempt to allocate memory from a "near by" node.
83       MPOL_BIND
84              This mode defines a strict policy that restricts memory  alloca‐
85              tion  to the nodes specified in nodemask.  If nodemask specifies
86              more than one node, page allocations will  come  from  the  node
87              with  the lowest numeric node ID first, until that node contains
88              no free memory.  Allocations will then come from the  node  with
89              the  next  highest  node  ID specified in nodemask and so forth,
90              until none of the specified nodes contain  free  memory.   Pages
91              will  not  be allocated from any node not specified in the node‐
92              mask.
95              This mode interleaves page allocations across the  nodes  speci‐
96              fied  in  nodemask in numeric node ID order.  This optimizes for
97              bandwidth instead of latency by spreading out pages  and  memory
98              accesses   to  those  pages  across  multiple  nodes.   However,
99              accesses to a single page will still be limited  to  the  memory
100              bandwidth of a single node.
103              This  mode  sets  the preferred node for allocation.  The kernel
104              will try to allocate pages from this node first and fall back to
105              "near by" nodes if the preferred node is low on free memory.  If
106              nodemask specifies more than one node ID, the first node in  the
107              mask  will  be  selected as the preferred node.  If the nodemask
108              and maxnode arguments specify the empty  set,  then  the  policy
109              specifies  "local  allocation"  (like  the system default policy
110              discussed above).
112       MPOL_LOCAL (since Linux 3.8)
113              This mode specifies "local allocation"; the memory is  allocated
114              on the node of the CPU that triggered the allocation (the "local
115              node").  The nodemask and maxnode  arguments  must  specify  the
116              empty  set.  If the "local node" is low on free memory, the ker‐
117              nel will try to allocate memory from other  nodes.   The  kernel
118              will  allocate  memory from the "local node" whenever memory for
119              this node is available.  If the "local node" is not  allowed  by
120              the  process's  current  cpuset  context, the kernel will try to
121              allocate memory from other nodes.  The kernel will allocate mem‐
122              ory  from  the  "local  node" whenever it becomes allowed by the
123              process's current cpuset context.
125       The thread memory policy is  preserved  across  an  execve(2),  and  is
126       inherited by child threads created using fork(2) or clone(2).


129       On  success,  set_mempolicy()  returns  0; on error, -1 is returned and
130       errno is set to indicate the error.


133       EFAULT Part of all of the memory range specified by nodemask and  maxn‐
134              ode points outside your accessible address space.
136       EINVAL mode  is  invalid.   Or,  mode  is  MPOL_DEFAULT and nodemask is
137              nonempty, or mode is MPOL_BIND or MPOL_INTERLEAVE  and  nodemask
138              is empty.  Or, maxnode specifies more than a page worth of bits.
139              Or, nodemask specifies one or more node  IDs  that  are  greater
140              than  the  maximum  supported node ID.  Or, none of the node IDs
141              specified by nodemask are on-line and allowed by  the  process's
142              current  cpuset  context, or none of the specified nodes contain
143              memory.     Or,    the    mode    argument    specified     both
146       ENOMEM Insufficient kernel memory was available.


149       The  set_mempolicy()  system call was added to the Linux kernel in ver‐
150       sion 2.6.7.


153       This system call is Linux-specific.


156       Memory policy is not remembered if the page is swapped out.  When  such
157       a page is paged back in, it will use the policy of the thread or memory
158       range that is in effect at the time the page is allocated.
160       For information on library support, see numa(7).


163       get_mempolicy(2), getcpu(2),  mbind(2),  mmap(2),  numa(3),  cpuset(7),
164       numa(7), numactl(8)


167       This  page  is  part of release 4.15 of the Linux man-pages project.  A
168       description of the project, information about reporting bugs,  and  the
169       latest     version     of     this    page,    can    be    found    at
170       https://www.kernel.org/doc/man-pages/.
174Linux                             2017-09-15                  SET_MEMPOLICY(2)