(thread-yield)
(return-from get-spinlock t))))
(if (and (not *interrupts-enabled*) *allow-with-interrupts*)
- ;; If interrupts are enabled, but we are allowed to enabled them,
- ;; check for pending interrupts every once in a while.
+ ;; If interrupts are disabled, but we are allowed to
+ ;; enabled them, check for pending interrupts every once
+ ;; in a while.
(loop
(loop repeat 128 do (cas)) ; 128 is arbitrary here
(sb!unix::%check-interrupts))
(defun release-spinlock (spinlock)
(declare (optimize (speed 3) (safety 0)))
- ;; Simply setting SPINLOCK-VALUE to NIL is not enough as it does not
- ;; propagate to other processors, plus without a memory barrier the
- ;; CPU might reorder instructions allowing code from the critical
- ;; section to leak out. Use COMPARE-AND-SWAP for the memory barrier
- ;; effect and do some sanity checking while we are at it.
- (unless (eq *current-thread*
- (sb!ext:compare-and-swap (spinlock-value spinlock)
- *current-thread* nil))
- (error "Only the owner can release the spinlock ~S." spinlock)))
+ ;; On x86 and x86-64 we can get away with no memory barriers, (see
+ ;; Linux kernel mailing list "spin_unlock optimization(i386)"
+ ;; thread, summary at
+ ;; http://kt.iserv.nl/kernel-traffic/kt19991220_47.html#1.
+ ;;
+ ;; If the compiler may reorder this with other instructions, insert
+ ;; compiler barrier here.
+ ;;
+ ;; FIXME: this does not work on SMP Pentium Pro and OOSTORE systems,
+ ;; neither on most non-x86 architectures (but we don't have threads
+ ;; on those).
+ (setf (spinlock-value spinlock) nil))
\f
;;;; Mutexes
(defconstant +lock-taken+ 1)
(defconstant +lock-contested+ 2))
+(defun mutex-owner (mutex)
+ "Current owner of the mutex, NIL if the mutex is free. Naturally,
+this is racy by design (another thread may acquire the mutex after
+this function returns), it is intended for informative purposes. For
+testing whether the current thread is holding a mutex see
+HOLDING-MUTEX-P."
+ ;; Make sure to get the current value.
+ (sb!ext:compare-and-swap (mutex-%owner mutex) nil nil))
+
(defun get-mutex (mutex &optional (new-owner *current-thread*) (waitp t))
#!+sb-doc
"Acquire MUTEX for NEW-OWNER, which must be a thread or NIL. If
(when (eq new-owner old)
(error "Recursive lock attempt ~S." mutex))
#!-sb-thread
- (if old
- (error "Strange deadlock on ~S in an unithreaded build?" mutex)
- (setf (mutex-%owner mutex) new-owner)))
+ (when old
+ (error "Strange deadlock on ~S in an unithreaded build?" mutex)))
+ #!-sb-thread
+ (setf (mutex-%owner mutex) new-owner)
#!+sb-thread
(progn
;; FIXME: Lutexes do not currently support deadlines, as at least
(setf (mutex-%owner mutex) new-owner)
t)
#!-sb-lutex
+ ;; This is a direct tranlation of the Mutex 2 algorithm from
+ ;; "Futexes are Tricky" by Ulrich Drepper.
(let ((old (sb!ext:compare-and-swap (mutex-state mutex)
+lock-free+
+lock-taken+)))
RELEASE-MUTEX is not interrupt safe: interrupts should be disabled
around calls to it.
-Signals a WARNING is current thread is not the current owner of the
+Signals a WARNING if current thread is not the current owner of the
mutex."
(declare (type mutex mutex))
;; Order matters: set owner to NIL before releasing state.
(with-lutex-address (lutex (mutex-lutex mutex))
(%lutex-unlock lutex))
#!-sb-lutex
+ ;; FIXME: once ATOMIC-INCF supports struct slots with word sized
+ ;; unsigned-byte type this can be used:
+ ;;
+ ;; (let ((old (sb!ext:atomic-incf (mutex-state mutex) -1)))
+ ;; (unless (eql old +lock-free+)
+ ;; (setf (mutex-state mutex) +lock-free+)
+ ;; (with-pinned-objects (mutex)
+ ;; (futex-wake (mutex-state-address mutex) 1))))
(let ((old (sb!ext:compare-and-swap (mutex-state mutex)
+lock-taken+ +lock-free+)))
(when (eql old +lock-contested+)
#!+sb-thread
(defun handle-thread-exit (thread)
(/show0 "HANDLING THREAD EXIT")
- ;; We're going down, can't handle interrupts sanely anymore. GC
- ;; remains enabled.
- (block-deferrable-signals)
;; Lisp-side cleanup
(with-all-threads-lock
(setf (thread-%alive-p thread) nil)
;; of Allegro's *cl-default-special-bindings*, as that is at
;; least accessible to users to secure their own libraries.
;; --njf, 2006-07-15
+ ;;
+ ;; As it is, this lambda must not cons until we are ready
+ ;; to run GC. Be very careful.
(let* ((*current-thread* thread)
(*restart-clusters* nil)
(*handler-clusters* (sb!kernel::initial-handler-clusters))
(format nil
"~~@<Terminate this thread (~A)~~@:>"
*current-thread*))
- (unwind-protect
- (progn
- ;; now that most things have a chance to
- ;; work properly without messing up other
- ;; threads, it's time to enable signals
- (sb!unix::reset-signal-mask)
- (setf (thread-result thread)
- (cons t
- (multiple-value-list
- (funcall real-function)))))
- (handle-thread-exit thread)))))))
+ (without-interrupts
+ (unwind-protect
+ (with-local-interrupts
+ ;; Now that most things have a chance
+ ;; to work properly without messing up
+ ;; other threads, it's time to enable
+ ;; signals.
+ (sb!unix::unblock-deferrable-signals)
+ (setf (thread-result thread)
+ (cons t
+ (multiple-value-list
+ (funcall real-function))))
+ ;; Try to block deferrables. An
+ ;; interrupt may unwind it, but for a
+ ;; normal exit it prevents interrupt
+ ;; loss.
+ (block-deferrable-signals))
+ ;; We're going down, can't handle interrupts
+ ;; sanely anymore. GC remains enabled.
+ (block-deferrable-signals)
+ ;; We don't want to run interrupts in a dead
+ ;; thread when we leave WITHOUT-INTERRUPTS.
+ ;; This potentially causes important
+ ;; interupts to be lost: SIGINT comes to
+ ;; mind.
+ (setq *interrupt-pending* nil)
+ (handle-thread-exit thread))))))))
(values))))
+ ;; If the starting thread is stopped for gc before it signals the
+ ;; semaphore then we'd be stuck.
+ (assert (not *gc-inhibit*))
;; Keep INITIAL-FUNCTION pinned until the child thread is
- ;; initialized properly.
- (with-pinned-objects (initial-function)
- (let ((os-thread
- (%create-thread
- (get-lisp-obj-address initial-function))))
- (when (zerop os-thread)
- (error "Can't create a new thread"))
- (wait-on-semaphore setup-sem)
- thread))))
+ ;; initialized properly. Wrap the whole thing in
+ ;; WITHOUT-INTERRUPTS because we pass INITIAL-FUNCTION to another
+ ;; thread.
+ (without-interrupts
+ (with-pinned-objects (initial-function)
+ (let ((os-thread
+ (%create-thread
+ (get-lisp-obj-address initial-function))))
+ (when (zerop os-thread)
+ (error "Can't create a new thread"))
+ (wait-on-semaphore setup-sem)
+ thread)))))
(define-condition join-thread-error (error)
((thread :reader join-thread-error-thread :initarg :thread))
"Suspend current thread until THREAD exits. Returns the result
values of the thread function. If the thread does not exit normally,
return DEFAULT if given or else signal JOIN-THREAD-ERROR."
- (with-mutex ((thread-result-lock thread))
+ (with-system-mutex ((thread-result-lock thread) :allow-with-interrupts t)
(cond ((car (thread-result thread))
- (values-list (cdr (thread-result thread))))
+ (return-from join-thread
+ (values-list (cdr (thread-result thread)))))
(defaultp
- default)
- (t
- (error 'join-thread-error :thread thread)))))
+ (return-from join-thread default))))
+ (error 'join-thread-error :thread thread))
(defun destroy-thread (thread)
#!+sb-doc
(defun run-interruption ()
(in-interruption ()
(loop
- (let ((interruption (with-interruptions-lock (*current-thread*)
- (pop (thread-interruptions *current-thread*)))))
- (if interruption
- (with-interrupts
- (funcall interruption))
- (return))))))
+ (let ((interruption (with-interruptions-lock (*current-thread*)
+ (pop (thread-interruptions *current-thread*)))))
+ ;; Resignalling after popping one works fine, because from the
+ ;; OS's point of view we have returned from the signal handler
+ ;; (thanks to arrange_return_to_lisp_function) so at least one
+ ;; more signal will be delivered.
+ (when (thread-interruptions *current-thread*)
+ (signal-interrupt-thread (thread-os-thread *current-thread*)))
+ (if interruption
+ (with-interrupts
+ (funcall interruption))
+ (return))))))
;;; The order of interrupt execution is peculiar. If thread A
;;; interrupts thread B with I1, I2 and B for some reason receives I1
projects / sbcl.git / blobdiff