1 .\" This man page is Copyright (C) 2006 Andi Kleen <ak@muc.de>.
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.
10 .\" 2008, mtk, various edits
12 .TH GETCPU 2 2013-04-03 "Linux" "Linux Programmer's Manual"
14 getcpu \- determine CPU and NUMA node on which the calling thread is running
17 .B #include <linux/getcpu.h>
19 .BI "int getcpu(unsigned *" cpu ", unsigned *" node \
20 ", struct getcpu_cache *" tcache );
24 There is no glibc wrapper for this system call; see NOTES.
28 system call identifies the processor and node on which the calling
29 thread or process is currently running and writes them into the
30 integers pointed to by the
35 The processor is a unique small integer identifying a CPU.
36 The node is a unique small identifier identifying a NUMA node.
41 is NULL nothing is written to the respective pointer.
43 The third argument to this system call is nowadays unused,
44 and should be specified as NULL
45 unless portability to Linux 2.6.23 or earlier is required (see NOTES).
47 The information placed in
49 is guaranteed to be current only at the time of the call:
50 unless the CPU affinity has been fixed using
51 .BR sched_setaffinity (2),
52 the kernel might change the CPU at any time.
53 (Normally this does not happen
54 because the scheduler tries to minimize movements between CPUs to
55 keep caches hot, but it is possible.)
56 The caller must allow for the possibility that the information returned in
60 is no longer current by the time the call returns.
62 On success, 0 is returned.
63 On error, \-1 is returned, and
69 Arguments point outside the calling process's address space.
72 was added in kernel 2.6.19 for x86_64 and i386.
77 Linux makes a best effort to make this call as fast possible.
80 is to allow programs to make optimizations with per-CPU data
81 or for NUMA optimization.
83 Glibc does not provide a wrapper for this system call; call it using
91 argument is unused since Linux 2.6.24.
92 .\" commit 4307d1e5ada595c87f9a4d16db16ba5edb70dcb1
93 .\" Author: Ingo Molnar <mingo@elte.hu>
94 .\" Date: Wed Nov 7 18:37:48 2007 +0100
95 .\" x86: ignore the sys_getcpu() tcache parameter
97 if this argument was non-NULL,
98 then it specified a pointer to a caller-allocated buffer in thread-local
99 storage that was used to provide a caching mechanism for
101 Use of the cache could speed
103 calls, at the cost that there was a very small chance that
104 the returned information would be out of date.
105 The caching mechanism was considered to cause problems when
106 migrating threads between CPUs, and so the argument is now ignored.
108 .\" ===== Before kernel 2.6.24: =====
110 .\" is a pointer to a
111 .\" .IR "struct getcpu_cache"
112 .\" that is used as a cache by
114 .\" The caller should put the cache into a thread-local variable
115 .\" if the process is multithreaded,
116 .\" because the cache cannot be shared between different threads.
119 .\" If it is not NULL
121 .\" will use it to speed up operation.
122 .\" The information inside the cache is private to the system call
123 .\" and should not be accessed by the user program.
124 .\" The information placed in the cache can change between kernel releases.
126 .\" When no cache is specified
129 .\" but always retrieve the current CPU and node information.
133 .\" However, the cached information is updated only once per jiffy (see
135 .\" This means that the information could theoretically be out of date,
136 .\" although in practice the scheduler's attempt to maintain
137 .\" soft CPU affinity means that the information is unlikely to change
138 .\" over the course of the caching interval.
141 .BR sched_setaffinity (2),
142 .BR set_mempolicy (2),
143 .BR sched_getcpu (3),