1IOPRIO_SET(2) Linux Programmer's Manual IOPRIO_SET(2)
2
3
4
6 ioprio_get, ioprio_set - get/set I/O scheduling class and priority
7
9 int ioprio_get(int which, int who);
10 int ioprio_set(int which, int who, int ioprio);
11
12 Note: There are no glibc wrappers for these system calls; see NOTES.
13
15 The ioprio_get() and ioprio_set() system calls get and set the I/O
16 scheduling class and priority of one or more threads.
17
18 The which and who arguments identify the thread(s) on which the system
19 calls operate. The which argument determines how who is interpreted,
20 and has one of the following values:
21
22 IOPRIO_WHO_PROCESS
23 who is a process ID or thread ID identifying a single process or
24 thread. If who is 0, then operate on the calling thread.
25
26 IOPRIO_WHO_PGRP
27 who is a process group ID identifying all the members of a
28 process group. If who is 0, then operate on the process group
29 of which the caller is a member.
30
31 IOPRIO_WHO_USER
32 who is a user ID identifying all of the processes that have a
33 matching real UID.
34
35 If which is specified as IOPRIO_WHO_PGRP or IOPRIO_WHO_USER when call‐
36 ing ioprio_get(), and more than one process matches who, then the
37 returned priority will be the highest one found among all of the match‐
38 ing processes. One priority is said to be higher than another one if
39 it belongs to a higher priority class (IOPRIO_CLASS_RT is the highest
40 priority class; IOPRIO_CLASS_IDLE is the lowest) or if it belongs to
41 the same priority class as the other process but has a higher priority
42 level (a lower priority number means a higher priority level).
43
44 The ioprio argument given to ioprio_set() is a bit mask that specifies
45 both the scheduling class and the priority to be assigned to the target
46 process(es). The following macros are used for assembling and dissect‐
47 ing ioprio values:
48
49 IOPRIO_PRIO_VALUE(class, data)
50 Given a scheduling class and priority (data), this macro com‐
51 bines the two values to produce an ioprio value, which is
52 returned as the result of the macro.
53
54 IOPRIO_PRIO_CLASS(mask)
55 Given mask (an ioprio value), this macro returns its I/O class
56 component, that is, one of the values IOPRIO_CLASS_RT,
57 IOPRIO_CLASS_BE, or IOPRIO_CLASS_IDLE.
58
59 IOPRIO_PRIO_DATA(mask)
60 Given mask (an ioprio value), this macro returns its priority
61 (data) component.
62
63 See the NOTES section for more information on scheduling classes and
64 priorities, as well as the meaning of specifying ioprio as 0.
65
66 I/O priorities are supported for reads and for synchronous (O_DIRECT,
67 O_SYNC) writes. I/O priorities are not supported for asynchronous
68 writes because they are issued outside the context of the program
69 dirtying the memory, and thus program-specific priorities do not apply.
70
72 On success, ioprio_get() returns the ioprio value of the process with
73 highest I/O priority of any of the processes that match the criteria
74 specified in which and who. On error, -1 is returned, and errno is set
75 to indicate the error.
76
77 On success, ioprio_set() returns 0. On error, -1 is returned, and
78 errno is set to indicate the error.
79
81 EINVAL Invalid value for which or ioprio. Refer to the NOTES section
82 for available scheduler classes and priority levels for ioprio.
83
84 EPERM The calling process does not have the privilege needed to assign
85 this ioprio to the specified process(es). See the NOTES section
86 for more information on required privileges for ioprio_set().
87
88 ESRCH No process(es) could be found that matched the specification in
89 which and who.
90
92 These system calls have been available on Linux since kernel 2.6.13.
93
95 These system calls are Linux-specific.
96
98 Glibc does not provide a wrapper for these system calls; call them
99 using syscall(2).
100
101 Two or more processes or threads can share an I/O context. This will
102 be the case when clone(2) was called with the CLONE_IO flag. However,
103 by default, the distinct threads of a process will not share the same
104 I/O context. This means that if you want to change the I/O priority of
105 all threads in a process, you may need to call ioprio_set() on each of
106 the threads. The thread ID that you would need for this operation is
107 the one that is returned by gettid(2) or clone(2).
108
109 These system calls have an effect only when used in conjunction with an
110 I/O scheduler that supports I/O priorities. As at kernel 2.6.17 the
111 only such scheduler is the Completely Fair Queuing (CFQ) I/O scheduler.
112
113 If no I/O scheduler has been set for a thread, then by default the I/O
114 priority will follow the CPU nice value (setpriority(2)). In Linux
115 kernels before version 2.6.24, once an I/O priority had been set using
116 ioprio_set(), there was no way to reset the I/O scheduling behavior to
117 the default. Since Linux 2.6.24, specifying ioprio as 0 can be used to
118 reset to the default I/O scheduling behavior.
119
120 Selecting an I/O scheduler
121 I/O schedulers are selected on a per-device basis via the special file
122 /sys/block/<device>/queue/scheduler.
123
124 One can view the current I/O scheduler via the /sys filesystem. For
125 example, the following command displays a list of all schedulers cur‐
126 rently loaded in the kernel:
127
128 $ cat /sys/block/sda/queue/scheduler
129 noop anticipatory deadline [cfq]
130
131 The scheduler surrounded by brackets is the one actually in use for the
132 device (sda in the example). Setting another scheduler is done by
133 writing the name of the new scheduler to this file. For example, the
134 following command will set the scheduler for the sda device to cfq:
135
136 $ su
137 Password:
138 # echo cfq > /sys/block/sda/queue/scheduler
139
140 The Completely Fair Queuing (CFQ) I/O scheduler
141 Since version 3 (also known as CFQ Time Sliced), CFQ implements I/O
142 nice levels similar to those of CPU scheduling. These nice levels are
143 grouped into three scheduling classes, each one containing one or more
144 priority levels:
145
146 IOPRIO_CLASS_RT (1)
147 This is the real-time I/O class. This scheduling class is given
148 higher priority than any other class: processes from this class
149 are given first access to the disk every time. Thus, this I/O
150 class needs to be used with some care: one I/O real-time process
151 can starve the entire system. Within the real-time class, there
152 are 8 levels of class data (priority) that determine exactly how
153 much time this process needs the disk for on each service. The
154 highest real-time priority level is 0; the lowest is 7. In the
155 future, this might change to be more directly mappable to per‐
156 formance, by passing in a desired data rate instead.
157
158 IOPRIO_CLASS_BE (2)
159 This is the best-effort scheduling class, which is the default
160 for any process that hasn't set a specific I/O priority. The
161 class data (priority) determines how much I/O bandwidth the
162 process will get. Best-effort priority levels are analogous to
163 CPU nice values (see getpriority(2)). The priority level deter‐
164 mines a priority relative to other processes in the best-effort
165 scheduling class. Priority levels range from 0 (highest) to 7
166 (lowest).
167
168 IOPRIO_CLASS_IDLE (3)
169 This is the idle scheduling class. Processes running at this
170 level get I/O time only when no one else needs the disk. The
171 idle class has no class data. Attention is required when
172 assigning this priority class to a process, since it may become
173 starved if higher priority processes are constantly accessing
174 the disk.
175
176 Refer to the kernel source file Documentation/block/ioprio.txt for more
177 information on the CFQ I/O Scheduler and an example program.
178
179 Required permissions to set I/O priorities
180 Permission to change a process's priority is granted or denied based on
181 two criteria:
182
183 Process ownership
184 An unprivileged process may set the I/O priority only for a
185 process whose real UID matches the real or effective UID of the
186 calling process. A process which has the CAP_SYS_NICE capabil‐
187 ity can change the priority of any process.
188
189 What is the desired priority
190 Attempts to set very high priorities (IOPRIO_CLASS_RT) require
191 the CAP_SYS_ADMIN capability. Kernel versions up to 2.6.24 also
192 required CAP_SYS_ADMIN to set a very low priority
193 (IOPRIO_CLASS_IDLE), but since Linux 2.6.25, this is no longer
194 required.
195
196 A call to ioprio_set() must follow both rules, or the call will fail
197 with the error EPERM.
198
200 Glibc does not yet provide a suitable header file defining the function
201 prototypes and macros described on this page. Suitable definitions can
202 be found in linux/ioprio.h.
203
205 ionice(1), getpriority(2), open(2), capabilities(7), cgroups(7)
206
207 Documentation/block/ioprio.txt in the Linux kernel source tree
208
210 This page is part of release 5.07 of the Linux man-pages project. A
211 description of the project, information about reporting bugs, and the
212 latest version of this page, can be found at
213 https://www.kernel.org/doc/man-pages/.
214
215
216
217Linux 2019-03-06 IOPRIO_SET(2)