man2/getcpu.2

   1 .\" This man page is Copyright (C) 2006 Andi Kleen <ak@muc.de>.
   2 .\"
   3 .\" %%%LICENSE_START(VERBATIM_ONE_PARA)
   4 .\" Permission is granted to distribute possibly modified copies
   5 .\" of this page provided the header is included verbatim,
   6 .\" and in case of nontrivial modification author and date
   7 .\" of the modification is added to the header.
   8 .\" %%%END_LICENSE
   9 .\" 2008, mtk, various edits
  10 .TH GETCPU 2 2012-07-13 "Linux" "Linux Programmer's Manual"
  11 .SH NAME
  12 getcpu \- determine CPU and NUMA node on which the calling thread is running
  13 .SH SYNOPSIS
  14 .nf
  15 .B #include <linux/getcpu.h>
  16 .sp
  17 .BI "int getcpu(unsigned *" cpu ", unsigned *" node \
  18 ", struct getcpu_cache *" tcache );
  19 .fi
  20
  21 .IR Note :
  22 There is no glibc wrapper for this system call; see NOTES.
  23 .SH DESCRIPTION
  24 The
  25 .BR getcpu ()
  26 system call identifies the processor and node on which the calling
  27 thread or process is currently running and writes them into the
  28 integers pointed to by the
  29 .I cpu
  30 and
  31 .I node
  32 arguments.
  33 The processor is a unique small integer identifying a CPU.
  34 The node is a unique small identifier identifying a NUMA node.
  35 When either
  36 .I cpu
  37 or
  38 .I node
  39 is NULL nothing is written to the respective pointer.
  40
  41 The third argument to this system call is nowadays unused (see NOTES).
  42
  43 The information placed in
  44 .I cpu
  45 is only guaranteed to be current at the time of the call:
  46 unless the CPU affinity has been fixed using
  47 .BR sched_setaffinity (2),
  48 the kernel might change the CPU at any time.
  49 (Normally this does not happen
  50 because the scheduler tries to minimize movements between CPUs to
  51 keep caches hot, but it is possible.)
  52 The caller must allow for the possibility that the information returned in
  53 .I cpu
  54 and
  55 .I node
  56 is no longer current by the time the call returns.
  57 .SH RETURN VALUE
  58 On success, 0 is returned.
  59 On error, \-1 is returned, and
  60 .I errno
  61 is set appropriately.
  62 .SH ERRORS
  63 .TP
  64 .B EFAULT
  65 Arguments point outside the calling process's address space.
  66 .SH VERSIONS
  67 .BR getcpu ()
  68 was added in kernel 2.6.19 for x86_64 and i386.
  69 .SH CONFORMING TO
  70 .BR getcpu ()
  71 is Linux specific.
  72 .SH NOTES
  73 Linux makes a best effort to make this call as fast possible.
  74 The intention of
  75 .BR getcpu ()
  76 is to allow programs to make optimizations with per-CPU data
  77 or for NUMA optimization.
  78
  79 Glibc does not provide a wrapper for this system call; call it using
  80 .BR syscall (2);
  81 or use
  82 .BR sched_getcpu (3)
  83 instead.
  84
  85 The
  86 .I tcache
  87 argument is unused since Linux 2.6.24.
  88 .\" commit 4307d1e5ada595c87f9a4d16db16ba5edb70dcb1
  89 .\" Author: Ingo Molnar <mingo@elte.hu>
  90 .\" Date:   Wed Nov 7 18:37:48 2007 +0100
  91 .\" x86: ignore the sys_getcpu() tcache parameter
  92 In earlier kernels,
  93 if this argument was non-NULL,
  94 then it specified a pointer to a caller-allocated buffer in thread-local
  95 storage that was used to provide a caching mechanism for
  96 .BR getcpu ().
  97 Use of the cache could speed
  98 .BR getcpu ()
  99 calls, at the cost that there was a very small chance that
 100 the returned information would be out of date.
 101 The caching mechanism was considered to cause problems when
 102 migrating threads between CPUs, and so the argument is now ignored.
 103 .\"
 104 .\" ===== Before kernel 2.6.24: =====
 105 .\" .I tcache
 106 .\" is a pointer to a
 107 .\" .IR "struct getcpu_cache"
 108 .\" that is used as a cache by
 109 .\" .BR getcpu ().
 110 .\" The caller should put the cache into a thread-local variable
 111 .\" if the process is multithreaded,
 112 .\" because the cache cannot be shared between different threads.
 113 .\" .I tcache
 114 .\" can be NULL.
 115 .\" If it is not NULL
 116 .\" .BR getcpu ()
 117 .\" will use it to speed up operation.
 118 .\" The information inside the cache is private to the system call
 119 .\" and should not be accessed by the user program.
 120 .\" The information placed in the cache can change between kernel releases.
 121 .\"
 122 .\" When no cache is specified
 123 .\" .BR getcpu ()
 124 .\" will be slower,
 125 .\" but always retrieve the current CPU and node information.
 126 .\" With a cache
 127 .\" .BR getcpu ()
 128 .\" is faster.
 129 .\" However, the cached information is only updated once per jiffy (see
 130 .\" .BR time (7)).
 131 .\" This means that the information could theoretically be out of date,
 132 .\" although in practice the scheduler's attempt to maintain
 133 .\" soft CPU affinity means that the information is unlikely to change
 134 .\" over the course of the caching interval.
 135 .SH SEE ALSO
 136 .BR mbind (2),
 137 .BR sched_setaffinity (2),
 138 .BR set_mempolicy (2),
 139 .BR sched_getcpu (3),
 140 .BR cpuset (7)