1 .\" This man page is Copyright (C) 2006 Andi Kleen <ak@muc.de>.
2 .\" Permission is granted to distribute possibly modified copies
3 .\" of this page provided the header is included verbatim,
4 .\" and in case of nontrivial modification author and date
5 .\" of the modification is added to the header.
6 .\" 2008, mtk, various edits
7 .TH GETCPU 2 2012-05-09 "Linux" "Linux Programmer's Manual"
9 getcpu \- determine CPU and NUMA node on which the calling thread is running
12 .B #include <linux/getcpu.h>
14 .BI "int getcpu(unsigned *" cpu ", unsigned *" node \
15 ", struct getcpu_cache *" tcache );
20 system call identifies the processor and node on which the calling
21 thread or process is currently running and writes them into the
22 integers pointed to by the
27 The processor is a unique small integer identifying a CPU.
28 The node is a unique small identifier identifying a NUMA node.
33 is NULL nothing is written to the respective pointer.
35 The third argument to this system call is nowadays unused (see NOTES).
37 The information placed in
39 is only guaranteed to be current at the time of the call:
40 unless the CPU affinity has been fixed using
41 .BR sched_setaffinity (2),
42 the kernel might change the CPU at any time.
43 (Normally this does not happen
44 because the scheduler tries to minimize movements between CPUs to
45 keep caches hot, but it is possible.)
46 The caller must allow for the possibility that the information returned in
50 is no longer current by the time the call returns.
52 On success, 0 is returned.
53 On error, \-1 is returned, and
59 Arguments point outside the calling process's address space.
62 was added in kernel 2.6.19 for x86_64 and i386.
67 Linux makes a best effort to make this call as fast possible.
70 is to allow programs to make optimizations with per-CPU data
71 or for NUMA optimization.
73 Glibc does not provide a wrapper for this system call; call it using
81 argument is unused since Linux 2.6.24.
82 .\" commit 4307d1e5ada595c87f9a4d16db16ba5edb70dcb1
83 .\" Author: Ingo Molnar <mingo@elte.hu>
84 .\" Date: Wed Nov 7 18:37:48 2007 +0100
85 .\" x86: ignore the sys_getcpu() tcache parameter
87 if this argument was non-NULL,
88 then it specified a pointer to a caller-allocated buffer in thread-local
89 storage that was used to provide a caching mechanism for
91 Use of the cache could speed
93 calls, at the cost that there was a very small chance that
94 the returned information would be out of date.
95 The caching mechanism was considered to cause problems when
96 migrating threads between CPUs, and so the argument is now ignored.
98 .\" ===== Before kernel 2.6.24: =====
100 .\" is a pointer to a
101 .\" .IR "struct getcpu_cache"
102 .\" that is used as a cache by
104 .\" The caller should put the cache into a thread-local variable
105 .\" if the process is multithreaded,
106 .\" because the cache cannot be shared between different threads.
109 .\" If it is not NULL
111 .\" will use it to speed up operation.
112 .\" The information inside the cache is private to the system call
113 .\" and should not be accessed by the user program.
114 .\" The information placed in the cache can change between kernel releases.
116 .\" When no cache is specified
119 .\" but always retrieve the current CPU and node information.
123 .\" However, the cached information is only updated once per jiffy (see
125 .\" This means that the information could theoretically be out of date,
126 .\" although in practice the scheduler's attempt to maintain
127 .\" soft CPU affinity means that the information is unlikely to change
128 .\" over the course of the caching interval.
131 .BR sched_setaffinity (2),
132 .BR set_mempolicy (2),
133 .BR sched_getcpu (3),