2 @comment node-name, next, previous, up
5 SBCL supports a fairly low-level threading interface that maps onto
6 the host operating system's concept of threads or lightweight
7 processes. This means that threads may take advantage of hardware
8 multiprocessing on machines that have more than one CPU, but it does
9 not allow Lisp control of the scheduler. This is found in the
12 This requires Linux (2.6+ or systems with NPTL backports) running on the
13 x86 or x86-64 architecture, or SunOS (Solaris) on the x86. Support for
14 threading on Darwin (Mac OS X) and FreeBSD on the x86 is experimental.
20 * Waitqueue/condition variables::
21 * Sessions/Debugging::
22 * Implementation (Linux x86)::
25 @node Threading basics
26 @comment node-name, next, previous, up
27 @section Threading basics
30 (make-thread (lambda () (write-line "Hello, world")))
33 @include struct-sb-thread-thread.texinfo
34 @include var-sb-thread-star-current-thread-star.texinfo
35 @include fun-sb-thread-make-thread.texinfo
36 @include fun-sb-thread-join-thread.texinfo
37 @include condition-sb-thread-join-thread-error.texinfo
38 @include fun-sb-thread-join-thread-error-thread.texinfo
39 @include fun-sb-thread-thread-alive-p.texinfo
40 @include fun-sb-thread-list-all-threads.texinfo
41 @include condition-sb-thread-interrupt-thread-error.texinfo
42 @include fun-sb-thread-interrupt-thread-error-thread.texinfo
43 @include fun-sb-thread-interrupt-thread.texinfo
44 @include fun-sb-thread-terminate-thread.texinfo
46 @node Special Variables
47 @comment node-name, next, previous, up
48 @section Special Variables
50 The interaction of special variables with multiple threads is mostly
51 as one would expect, with behaviour very similar to other
56 global special values are visible across all threads;
58 bindings (e.g. using LET) are local to the thread;
60 threads do not inherit dynamic bindings from the parent thread
63 The last point means that
68 (sb-thread:make-thread (lambda () (print *x*))))
71 prints @code{0} and not @code{1} as of 0.9.6.
74 @comment node-name, next, previous, up
75 @section Mutex Support
77 Mutexes are used for controlling access to a shared resource. One
78 thread is allowed to hold the mutex, others which attempt to take it
79 will be made to wait until it's free. Threads are woken in the order
80 that they go to sleep.
82 There isn't a timeout on mutex acquisition, but the usual WITH-TIMEOUT
83 macro (which throws a TIMEOUT condition after n seconds) can be used
84 if you want a bounded wait.
87 (defpackage :demo (:use "CL" "SB-THREAD" "SB-EXT"))
91 (defvar *a-mutex* (make-mutex :name "my lock"))
94 (format t "Thread ~A running ~%" *current-thread*)
95 (with-mutex (*a-mutex*)
96 (format t "Thread ~A got the lock~%" *current-thread*)
98 (format t "Thread ~A dropped lock, dying now~%" *current-thread*))
100 (make-thread #'thread-fn)
101 (make-thread #'thread-fn)
104 @include struct-sb-thread-mutex.texinfo
105 @include fun-sb-thread-make-mutex.texinfo
106 @include fun-sb-thread-mutex-name.texinfo
107 @include fun-sb-thread-mutex-value.texinfo
108 @include fun-sb-thread-get-mutex.texinfo
109 @include fun-sb-thread-release-mutex.texinfo
110 @include macro-sb-thread-with-mutex.texinfo
111 @include macro-sb-thread-with-recursive-lock.texinfo
113 @node Waitqueue/condition variables
114 @comment node-name, next, previous, up
115 @section Waitqueue/condition variables
117 These are based on the POSIX condition variable design, hence the
118 annoyingly CL-conflicting name. For use when you want to check a
119 condition and sleep until it's true. For example: you have a shared
120 queue, a writer process checking ``queue is empty'' and one or more
121 readers that need to know when ``queue is not empty''. It sounds
122 simple, but is astonishingly easy to deadlock if another process runs
123 when you weren't expecting it to.
125 There are three components:
129 the condition itself (not represented in code)
132 the condition variable (a.k.a waitqueue) which proxies for it
135 a lock to hold while testing the condition
138 Important stuff to be aware of:
142 when calling condition-wait, you must hold the mutex. condition-wait
143 will drop the mutex while it waits, and obtain it again before
144 returning for whatever reason;
147 likewise, you must be holding the mutex around calls to
151 a process may return from condition-wait in several circumstances: it
152 is not guaranteed that the underlying condition has become true. You
153 must check that the resource is ready for whatever you want to do to
159 (defvar *buffer-queue* (make-waitqueue))
160 (defvar *buffer-lock* (make-mutex :name "buffer lock"))
162 (defvar *buffer* (list nil))
165 (with-mutex (*buffer-lock*)
167 (condition-wait *buffer-queue* *buffer-lock*)
169 (unless *buffer* (return))
170 (let ((head (car *buffer*)))
171 (setf *buffer* (cdr *buffer*))
172 (format t "reader ~A woke, read ~A~%"
173 *current-thread* head))))))
178 (with-mutex (*buffer-lock*)
181 (+ (char-code #\A) (random 26)))))))
182 (setf *buffer* (cons el *buffer*)))
183 (condition-notify *buffer-queue*))))
185 (make-thread #'writer)
186 (make-thread #'reader)
187 (make-thread #'reader)
190 @include struct-sb-thread-waitqueue.texinfo
191 @include fun-sb-thread-make-waitqueue.texinfo
192 @include fun-sb-thread-waitqueue-name.texinfo
193 @include fun-sb-thread-condition-wait.texinfo
194 @include fun-sb-thread-condition-notify.texinfo
195 @include fun-sb-thread-condition-broadcast.texinfo
197 @node Sessions/Debugging
198 @comment node-name, next, previous, up
199 @section Sessions/Debugging
201 If the user has multiple views onto the same Lisp image (for example,
202 using multiple terminals, or a windowing system, or network access)
203 they are typically set up as multiple @dfn{sessions} such that each
204 view has its own collection of foreground/background/stopped threads.
205 A thread which wishes to create a new session can use
206 @code{sb-thread:with-new-session} to remove itself from the current
207 session (which it shares with its parent and siblings) and create a
209 # See also @code{sb-thread:make-listener-thread}.
211 Within a single session, threads arbitrate between themselves for the
212 user's attention. A thread may be in one of three notional states:
213 foreground, background, or stopped. When a background process
214 attempts to print a repl prompt or to enter the debugger, it will stop
215 and print a message saying that it has stopped. The user at his
216 leisure may switch to that thread to find out what it needs. If a
217 background thread enters the debugger, selecting any restart will put
218 it back into the background before it resumes. Arbitration for the
219 input stream is managed by calls to @code{sb-thread:get-foreground}
220 (which may block) and @code{sb-thread:release-foreground}.
222 @code{sb-ext:quit} terminates all threads in the current session, but
223 leaves other sessions running.
225 @node Implementation (Linux x86)
226 @comment node-name, next, previous, up
227 @section Implementation (Linux x86/x86-64)
229 Threading is implemented using pthreads and some Linux specific bits
232 On x86 the per-thread local bindings for special variables is achieved
233 using the %fs segment register to point to a per-thread storage area.
234 This may cause interesting results if you link to foreign code that
235 expects threading or creates new threads, and the thread library in
236 question uses %fs in an incompatible way. On x86-64 the r12 register
239 Queues require the @code{sys_futex()} system call to be available:
240 this is the reason for the NPTL requirement. We test at runtime that
241 this system call exists.
243 Garbage collection is done with the existing Conservative Generational
244 GC. Allocation is done in small (typically 8k) regions: each thread
245 has its own region so this involves no stopping. However, when a
246 region fills, a lock must be obtained while another is allocated, and
247 when a collection is required, all processes are stopped. This is
248 achieved by sending them signals, which may make for interesting
249 behaviour if they are interrupted in system calls. The streams
250 interface is believed to handle the required system call restarting
251 correctly, but this may be a consideration when making other blocking
252 calls e.g. from foreign library code.
254 Large amounts of the SBCL library have not been inspected for
255 thread-safety. Some of the obviously unsafe areas have large locks
256 around them, so compilation and fasl loading, for example, cannot be
257 parallelized. Work is ongoing in this area.
259 A new thread by default is created in the same POSIX process group and
260 session as the thread it was created by. This has an impact on
261 keyboard interrupt handling: pressing your terminal's intr key
262 (typically @kbd{Control-C}) will interrupt all processes in the
263 foreground process group, including Lisp threads that SBCL considers
264 to be notionally `background'. This is undesirable, so background
265 threads are set to ignore the SIGINT signal.
267 @code{sb-thread:make-listener-thread} in addition to creating a new
268 Lisp session makes a new POSIX session, so that pressing
269 @kbd{Control-C} in one window will not interrupt another listener -
270 this has been found to be embarrassing.