]>
Commit | Line | Data |
---|---|---|
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 | .\" %%%LICENSE_END | |
9 | .\" | |
10 | .\" 2008, mtk, various edits | |
11 | .\" | |
12 | .TH GETCPU 2 2019-03-06 "Linux" "Linux Programmer's Manual" | |
13 | .SH NAME | |
14 | getcpu \- determine CPU and NUMA node on which the calling thread is running | |
15 | .SH SYNOPSIS | |
16 | .nf | |
17 | .B #include <linux/getcpu.h> | |
18 | .PP | |
19 | .BI "int getcpu(unsigned *" cpu ", unsigned *" node \ | |
20 | ", struct getcpu_cache *" tcache ); | |
21 | .fi | |
22 | .SH DESCRIPTION | |
23 | The | |
24 | .BR getcpu () | |
25 | system call identifies the processor and node on which the calling | |
26 | thread or process is currently running and writes them into the | |
27 | integers pointed to by the | |
28 | .I cpu | |
29 | and | |
30 | .I node | |
31 | arguments. | |
32 | The processor is a unique small integer identifying a CPU. | |
33 | The node is a unique small identifier identifying a NUMA node. | |
34 | When either | |
35 | .I cpu | |
36 | or | |
37 | .I node | |
38 | is NULL nothing is written to the respective pointer. | |
39 | .PP | |
40 | The third argument to this system call is nowadays unused, | |
41 | and should be specified as NULL | |
42 | unless portability to Linux 2.6.23 or earlier is required (see NOTES). | |
43 | .PP | |
44 | The information placed in | |
45 | .I cpu | |
46 | is guaranteed to be current only at the time of the call: | |
47 | unless the CPU affinity has been fixed using | |
48 | .BR sched_setaffinity (2), | |
49 | the kernel might change the CPU at any time. | |
50 | (Normally this does not happen | |
51 | because the scheduler tries to minimize movements between CPUs to | |
52 | keep caches hot, but it is possible.) | |
53 | The caller must allow for the possibility that the information returned in | |
54 | .I cpu | |
55 | and | |
56 | .I node | |
57 | is no longer current by the time the call returns. | |
58 | .SH RETURN VALUE | |
59 | On success, 0 is returned. | |
60 | On error, \-1 is returned, and | |
61 | .I errno | |
62 | is set appropriately. | |
63 | .SH ERRORS | |
64 | .TP | |
65 | .B EFAULT | |
66 | Arguments point outside the calling process's address space. | |
67 | .SH VERSIONS | |
68 | .BR getcpu () | |
69 | was added in kernel 2.6.19 for x86-64 and i386. | |
70 | Library support was added in glibc 2.29 | |
71 | (Earlier glibc versions did not provide a wrapper for this system call, | |
72 | necessitating the use of | |
73 | .BR syscall (2).) | |
74 | .SH CONFORMING TO | |
75 | .BR getcpu () | |
76 | is Linux-specific. | |
77 | .SH NOTES | |
78 | Linux makes a best effort to make this call as fast as possible. | |
79 | (On some architectures, this is done via an implementation in the | |
80 | .BR vdso (7).) | |
81 | The intention of | |
82 | .BR getcpu () | |
83 | is to allow programs to make optimizations with per-CPU data | |
84 | or for NUMA optimization. | |
85 | .PP | |
86 | The | |
87 | .I tcache | |
88 | argument is unused since Linux 2.6.24. | |
89 | .\" commit 4307d1e5ada595c87f9a4d16db16ba5edb70dcb1 | |
90 | .\" Author: Ingo Molnar <mingo@elte.hu> | |
91 | .\" Date: Wed Nov 7 18:37:48 2007 +0100 | |
92 | .\" x86: ignore the sys_getcpu() tcache parameter | |
93 | In earlier kernels, | |
94 | if this argument was non-NULL, | |
95 | then it specified a pointer to a caller-allocated buffer in thread-local | |
96 | storage that was used to provide a caching mechanism for | |
97 | .BR getcpu (). | |
98 | Use of the cache could speed | |
99 | .BR getcpu () | |
100 | calls, at the cost that there was a very small chance that | |
101 | the returned information would be out of date. | |
102 | The caching mechanism was considered to cause problems when | |
103 | migrating threads between CPUs, and so the argument is now ignored. | |
104 | .\" | |
105 | .\" ===== Before kernel 2.6.24: ===== | |
106 | .\" .I tcache | |
107 | .\" is a pointer to a | |
108 | .\" .IR "struct getcpu_cache" | |
109 | .\" that is used as a cache by | |
110 | .\" .BR getcpu (). | |
111 | .\" The caller should put the cache into a thread-local variable | |
112 | .\" if the process is multithreaded, | |
113 | .\" because the cache cannot be shared between different threads. | |
114 | .\" .I tcache | |
115 | .\" can be NULL. | |
116 | .\" If it is not NULL | |
117 | .\" .BR getcpu () | |
118 | .\" will use it to speed up operation. | |
119 | .\" The information inside the cache is private to the system call | |
120 | .\" and should not be accessed by the user program. | |
121 | .\" The information placed in the cache can change between kernel releases. | |
122 | .\" | |
123 | .\" When no cache is specified | |
124 | .\" .BR getcpu () | |
125 | .\" will be slower, | |
126 | .\" but always retrieve the current CPU and node information. | |
127 | .\" With a cache | |
128 | .\" .BR getcpu () | |
129 | .\" is faster. | |
130 | .\" However, the cached information is updated only once per jiffy (see | |
131 | .\" .BR time (7)). | |
132 | .\" This means that the information could theoretically be out of date, | |
133 | .\" although in practice the scheduler's attempt to maintain | |
134 | .\" soft CPU affinity means that the information is unlikely to change | |
135 | .\" over the course of the caching interval. | |
136 | .SH SEE ALSO | |
137 | .BR mbind (2), | |
138 | .BR sched_setaffinity (2), | |
139 | .BR set_mempolicy (2), | |
140 | .BR sched_getcpu (3), | |
141 | .BR cpuset (7), | |
142 | .BR vdso (7) |