freebsd-skq/share/man/man9/mi_switch.9
John Baldwin e7573e7ad7 Allow threads to atomically release rw and sx locks while waiting for an
event.  Locking primitives that support this (mtx, rw, and sx) now each
include their own foo_sleep() routine.
- Rename msleep() to _sleep() and change it's 'struct mtx' object to a
  'struct lock_object' pointer.  _sleep() uses the recently added
  lc_unlock() and lc_lock() function pointers for the lock class of the
  specified lock to release the lock while the thread is suspended.
- Add wrappers around _sleep() for mutexes (mtx_sleep()), rw locks
  (rw_sleep()), and sx locks (sx_sleep()).  msleep() still exists and
  is now identical to mtx_sleep(), but it is deprecated.
- Rename SLEEPQ_MSLEEP to SLEEPQ_SLEEP.
- Rewrite much of sleep.9 to not be msleep(9) centric.
- Flesh out the 'RETURN VALUES' section in sleep.9 and add an 'ERRORS'
  section.
- Add __nonnull(1) to _sleep() and msleep_spin() so that the compiler will
  warn if you try to pass a NULL wait channel.  The functions already have
  a KASSERT to that effect.
2007-03-09 22:41:01 +00:00

166 lines
5.3 KiB
Groff

.\" $NetBSD: ctxsw.9,v 1.2 1996/12/02 00:11:31 tls Exp $
.\"
.\" Copyright (c) 1996 The NetBSD Foundation, Inc.
.\" All rights reserved.
.\"
.\" This code is derived from software contributed to The NetBSD Foundation
.\" by Paul Kranenburg.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\" 1. Redistributions of source code must retain the above copyright
.\" notice, this list of conditions and the following disclaimer.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\" notice, this list of conditions and the following disclaimer in the
.\" documentation and/or other materials provided with the distribution.
.\" 3. All advertising materials mentioning features or use of this software
.\" must display the following acknowledgement:
.\" This product includes software developed by the NetBSD
.\" Foundation, Inc. and its contributors.
.\" 4. Neither the name of The NetBSD Foundation nor the names of its
.\" contributors may be used to endorse or promote products derived
.\" from this software without specific prior written permission.
.\"
.\" THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
.\" ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
.\" TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
.\" PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE
.\" LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
.\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
.\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
.\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
.\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
.\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
.\" POSSIBILITY OF SUCH DAMAGE.
.\"
.\" $FreeBSD$
.\"
.Dd November 24, 1996
.Dt MI_SWITCH 9
.Os
.Sh NAME
.Nm mi_switch ,
.Nm cpu_switch ,
.Nm cpu_throw
.Nd switch to another thread context
.Sh SYNOPSIS
.In sys/param.h
.In sys/proc.h
.Ft void
.Fn mi_switch "void"
.Ft void
.Fn cpu_switch "void"
.Ft void
.Fn cpu_throw "void"
.Sh DESCRIPTION
The
.Fn mi_switch
function implements the machine independent prelude to a thread context
switch.
It is called from only a few distinguished places in the kernel
code as a result of the principle of non-preemptable kernel mode execution.
The various major uses of
.Nm
can be enumerated as follows:
.Bl -enum -offset indent
.It
From within a function such as
.Xr cv_wait 9 ,
.Xr mtx_lock ,
or
.Xr tsleep 9
when the current thread
voluntarily relinquishes the CPU to wait for some resource or lock to become
available.
.It
After handling a trap
(e.g.\& a system call, device interrupt)
when the kernel prepares a return to user-mode execution.
This case is
typically handled by machine dependent trap-handling code after detection
of a change in the signal disposition of the current process, or when a
higher priority thread might be available to run.
The latter event is
communicated by the machine independent scheduling routines by calling
the machine defined
.Fn need_resched .
.It
In the signal handling code
(see
.Xr issignal 9 )
if a signal is delivered that causes a process to stop.
.It
When a thread dies in
.Xr thread_exit 9
and control of the processor can be passed to the next runnable thread.
.It
In
.Xr thread_suspend_check 9
where a thread needs to stop execution due to the suspension state of
the process as a whole.
.El
.Pp
.Fn mi_switch
records the amount of time the current thread has been running in the
process structures and checks this value against the CPU time limits
allocated to the process
(see
.Xr getrlimit 2 ) .
Exceeding the soft limit results in a
.Dv SIGXCPU
signal to be posted to the process, while exceeding the hard limit will
cause a
.Dv SIGKILL .
.Pp
If the thread is still in the
.Dv TDS_RUNNING
state,
.Fn mi_switch
will put it back onto the run queue, assuming that
it will want to run again soon.
If it is in one of the other
states and KSE threading is enabled, the associated
.Em KSE
will be made available to any higher priority threads from the same
group, to allow them to be scheduled next.
.Pp
After these administrative tasks are done,
.Fn mi_switch
hands over control to the machine dependent routine
.Fn cpu_switch ,
which will perform the actual thread context switch.
.Pp
.Fn cpu_switch
first saves the context of the current thread.
Next, it calls
.Fn choosethread
to determine which thread to run next.
Finally, it reads in the saved context of the new thread and starts to
execute the new thread.
.Pp
.Fn cpu_throw
is similar to
.Fn cpu_switch
except that it does not save the context of the old thread.
This function is useful when the kernel does not have an old thread
context to save, such as when CPUs other than the boot CPU perform their
first task switch, or when the kernel does not care about the state of the
old thread, such as in
.Fn thread_exit
when the kernel terminates the current thread and switches into a new
thread.
.Pp
To protect the
.Xr runqueue 9 ,
all of these functions must be called with the
.Va sched_lock
mutex held.
.Sh SEE ALSO
.Xr cv_wait 9 ,
.Xr issignal 9 ,
.Xr mutex 9 ,
.Xr runqueue 9 ,
.Xr tsleep 9 ,
.Xr wakeup 9