xref: /freebsd/share/man/man9/kern_yield.9 (revision 2008043f386721d58158e37e0d7e50df8095942d)
1.\"
2.\" Copyright (c) 2023 The FreeBSD Foundation
3.\"
4.\" This documentation was written by Mitchell Horne <mhorne@FreeBSD.org> under
5.\" sponsorship from the FreeBSD Foundation.
6.\"
7.\" Redistribution and use in source and binary forms, with or without
8.\" modification, are permitted provided that the following conditions
9.\" are met:
10.\" 1. Redistributions of source code must retain the above copyright
11.\"    notice, this list of conditions and the following disclaimer.
12.\" 2. Redistributions in binary form must reproduce the above copyright
13.\"    notice, this list of conditions and the following disclaimer in the
14.\"    documentation and/or other materials provided with the distribution.
15.\"
16.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26.\" SUCH DAMAGE.
27.\"
28.Dd January 30, 2023
29.Dt KERN_YIELD 9
30.Os
31.Sh NAME
32.Nm kern_yield ,
33.Nm maybe_yield ,
34.Nm should_yield
35.Nd "functions for yielding execution of the current thread"
36.Sh SYNOPSIS
37.Ft void
38.Fn kern_yield "int prio"
39.Ft void
40.Fn maybe_yield
41.Ft bool
42.Fn should_yield
43.Sh DESCRIPTION
44The
45.Fn kern_yield
46function causes the currently running thread to voluntarily, but
47unconditionally, surrender its execution to the scheduler.
48The
49.Va prio
50argument specifies the scheduling priority to be assigned before the context
51switch, which has an influence on when execution will resume.
52Note that the requested priority will take effect until the thread returns to
53usermode, after which its base user priority will be restored.
54Valid values for
55.Va prio
56are any of the
57.Dv PRI_*
58values defined in
59.In sys/priority.h ,
60as well as the following special values:
61.Bl -tag -width "PRI_UNCHANGED"
62.It Dv PRI_USER
63Schedule the thread with its base user priority; the value corresponding to
64.Xr setpriority 2 /
65.Xr nice 3 .
66.It Dv PRI_UNCHANGED
67Yield the thread without changing its priority.
68.El
69.Pp
70The
71.Fn should_yield
72function checks if sufficient time has passed since the thread's last voluntary
73context switch that yielding would be a useful service to other threads.
74It returns
75.Va true
76when this is the case.
77See
78.Sx USAGE NOTES
79for an elaboration of what this means.
80.Pp
81The
82.Fn maybe_yield
83function is a helper function for the common task of optionally yielding the
84processor.
85Internally,
86.Fn kern_yield "PRI_USER"
87will be called if
88.Fn should_yield
89returns
90.Va true .
91.Sh USAGE NOTES
92Although the kernel supports preemption, this is usually reserved for
93high-priority realtime or interrupt threads.
94Kernel worker threads and timesharing threads are not guaranteed to preempt
95each another.
96Thus, threads executing in the kernel are expected to behave cooperatively
97with respect to other threads in the system.
98The yield functions are mostly intended to be used by threads which perform a
99lot of work inside the kernel.
100For example:
101.Fn maybe_yield
102is called by the
103.Dv vlnru
104process each time it reclaims a vnode.
105.Pp
106The scheduler aims to identify threads which monopolize the CPU, and will
107schedule them with decreased priority.
108Threads which regularly yield the processor will be given the chance to run
109more often.
110The possibly surprising effect of this is that, depending on the disposition of
111other threads on the CPU's runqueue, a call to
112.Fn kern_yield
113does not guarantee that the yielding thread will be taken off the CPU.
114.Pp
115With the above considerations in mind, it is advised that code written using
116.Fn kern_yield
117be measured to confirm that its use has a positive effect on relevant
118performance or responsiveness metrics.
119Switching thread contexts has a non-zero cost, and thus yielding the processor
120too eagerly could have a negative impact on performance.
121.Pp
122Additionally, since yielding is a cooperative action, it is advised that the
123yielding thread release any locks that it may be holding, when possible.
124Otherwise, threads which have been given the chance to run could end up waiting
125on the yielding thread to release the lock, largely defeating the purpose of
126the yield.
127.Sh SEE ALSO
128.Xr setpriority 2 ,
129.Xr nice 3 ,
130.Xr mi_switch 9
131.Sh AUTHORS
132.An -nosplit
133This manual page was written by
134.An Mitchell Horne Aq Mt mhorne@FreeBSD.org .
135