Mercurial > core / lisp/std/task.lisp

 ;;; task.lisp --- Standard Task API
 
 ;; 
 
 ;;; Code:
 (in-package :std/task)
 
 ;;; Vars
 (defvar *task-pool*)
 (defvar *tasks* (make-queue :name "tasks"))
 (defvar *jobs*)
 (defvar *stages*)
 (sb-ext:defglobal *worker-threads* nil)
 (sb-ext:defglobal *oracles* nil)
 (sb-ext:defglobal *oracle-threads* nil)
 (sb-ext:defglobal *supervisor-threads* nil)
 
 (eval-when (:compile-toplevel)
   (defvar *task*)
   (defvar *task-result* nil))
 
 (define-condition task-error (thread-error) ()
   (:report (lambda (condition stream)
              (format stream "Unhandled task error in thread ~A" 
                      (thread-error-thread condition)))))
 
 (defun task-error (thread)
   (error 'task-error :thread thread))
 
 ;;; Kernel
 (defmacro gen-task-kernel (name args lock queue mailbox timeout &body body)
   `(compile ,name 
             (lambda ,args 
               (wait-on-semaphore ,lock ,@(when timeout `((:timeout ,timeout))))
               (let ((*task* (dequeue ,queue)))
                 (unwind-protect 
                      (handler-case (setf *task-result* (progn ,@body))
                        (error () (task-error *current-thread*)))
                   (send-message ,mailbox *task-result*)
                   (release-foreground))))))
 
 (defmacro define-task-kernel (name (&key lock timeout mailbox queue) args &body body)
   "Define a task kernel.
 
 (define-task-kernel NAME (&key ARGS ACCESSORS)
 
 The kernel should process all options and return a function - the
 'kernel function'.
 
 The kernel function is installed in worker threads by passing it to
 SB-THREAD:MAKE-THREAD. It may accept a varying number of arguments
 specified by ARGS.
 
 Within the BODY the variable *task* is bound to the result of (DEQUEUE QUEUE)
 and *task-result* is bound to the return value of BODY.
 
 This interface is experimental and subject to change."
   `(gen-task-kernel ,name ,args ,lock ,queue ,mailbox ,timeout
      ,@body))
 
 (defun make-ephemeral-thread (name)
     (sb-thread::%make-thread name t (make-semaphore :name name)))
 
 ;;; Proto
 (defgeneric designate-oracle (host guest))
 (defgeneric assign-supervisor (worker supervisor))
 
 (defgeneric make-workers (count &rest initargs &key &allow-other-keys)
   (:method ((count number) &key thread kernel input)
     (let ((ret))
       (dotimes (i count ret)
         (push (make-worker :thread thread :kernel kernel :input input) ret)))))
 
 (defgeneric tasks (self))
 (defgeneric results (self))
 (defgeneric run (self object &rest initargs &key &allow-other-keys))
 (defgeneric run-object (self))
 (defgeneric work (self &key &allow-other-keys))
 (defgeneric workers (self))
 
 ;;; Supervisor
 (defclass supervisor ()
   ((thread :initform (make-ephemeral-thread (symbol-name (gensym "supervisor"))) :accessor supervisor-thread)
    (domain)
    (scope))
   (:documentation "Supervisors are threads which are responsible for a set of worker threads
 within their DOMAIN and SCOPE."))
 
 (defmethod initialize-instance :after ((self supervisor) &key &allow-other-keys)
   (push (supervisor-thread self) *supervisor-threads*))
 
 ;;; Worker
 ;; unix-getrusage  
 ;; 0,-1,-2
 ;; (multiple-value-list (sb-unix:unix-getrusage 0))
 ;; (setf sb-unix::*on-dangerous-wait* :error)
 
 ;; TODO 2024-10-03: with-cas-lock?
 (defclass worker ()
   ((thread :initform (make-ephemeral-thread (symbol-name (gensym "worker")))
            :accessor worker-thread
            :initarg :thread)
    (kernel :type function :accessor worker-kernel :initarg :kernel)
    (tasks :initform nil :accessor tasks :initarg :input)))
 
 (defmethod initialize-instance :after ((self worker) &key &allow-other-keys)
   (push (worker-thread self) *worker-threads*))
 
 (defun make-worker (&key thread kernel input)
   (apply #'make-instance 'worker
          `(,@(when thread `(:thread ,thread))
            ,@(when kernel `(:kernel ,kernel))
            ,@(when input `(:input ,input)))))
 
 ;; TODO 2024-10-03: pause/resume
 
 (declaim (inline kill-worker join-worker start-worker run-worker))
 (defun start-worker (worker) 
   (sb-thread::start-thread (worker-thread worker) (worker-kernel worker) (tasks worker)))
 
 (defun run-worker (worker &key input wait)
   (when input
     (setf (tasks worker) input))
   (start-worker worker)
   (if wait (join-worker worker)
       worker))
 
 (defmethod run-object ((self worker))
   (run-worker self))
 
 (defmethod run ((self worker) (object t) &key wait &allow-other-keys)
   (run-worker self :input object :wait wait))
 
 (defun kill-worker (worker) 
   (declare (worker worker))
   (let ((th (worker-thread worker)))
     (unwind-protect (kill-thread th)
       (deletef *worker-threads* th))))
 
 (defun join-worker (worker)
   (declare (worker worker))
   (let ((th (worker-thread worker)))
     (unwind-protect (join-thread th)
       (deletef *worker-threads* th))))
 
 ;;; Oracle           
 (defstruct (oracle (:constructor %make-oracle (id thread)))
   "Oracles provide a tagged view into some threaded scope of work."
   (id 0 :type (unsigned-byte 32) :read-only t)
   (thread *current-thread* :read-only t))
 
 (defun oracle-of-id (id)
   (find id *oracles* :test '= :key 'oracle-id))
 
 (defun make-oracle (thread)
   (let ((id (thread-os-tid thread)))
     (if-let ((found (oracle-of-id id)))
       (values id found)
       (let ((orc (%make-oracle id thread)))
         (push orc *oracles*)
         (push (oracle-thread orc) *oracle-threads*)
         (values id orc)))))
 
 ;;; Task Pool
 (defstruct task-pool
   (kernel 'identity :type symbol)
   (tasks *tasks*)
   (lock (make-semaphore :name "online") :type semaphore)
   ;; TODO: test weak-vector here
   (workers (make-array 0 :element-type 'worker :fill-pointer 0) :type (vector worker))
   (results (make-mailbox :name "results")))
 
 (defmethod tasks ((self task-pool)) (task-pool-tasks self))
 (defmethod results ((self task-pool)) (task-pool-results self))
 (defmethod workers ((self task-pool)) (task-pool-workers self))
 
 (defmethod print-object ((self task-pool) (stream t))
   (print-unreadable-object (self stream :type t)
     (format stream "~A :workers ~A :tasks ~A/~A :results ~A"
             (task-pool-kernel self)
             (length (workers self))
             (queue-count (tasks self))
             (semaphore-count (task-pool-lock self))
             (mailbox-count (task-pool-results self)))))
 
 (defun kill-workers (pool)
   "Call FINISH-THREADS on task-pool's workers."
   (dotimes (i (length (workers pool)))
     (kill-worker (vector-pop (workers pool)))))
 
 (defun worker-count (task-pool &key online)
   (if online
       (semaphore-count (task-pool-lock task-pool))
       (length (task-pool-workers task-pool))))
 
 (defmethod designate-oracle ((self task-pool) (guest thread))
   (designate-oracle self (make-oracle guest)))
 
 (declaim (inline push-worker push-workers pop-worker))
 (defun push-worker (worker pool)
   (vector-push-extend worker (task-pool-workers pool)))
 
 (defun push-workers (threads pool)
   (with-slots (workers) pool
     (dolist (w threads)
       (vector-push-extend w workers))))
 
 (defmethod pop-worker (pool)
   (vector-pop (task-pool-workers pool)))
 
 (defun start-task-worker (pool index)
   ;; (with-recursive-lock
   (start-worker (aref (workers pool) index)))
 
 (defun start-task-workers (pool)
   "Start all workers in the given task POOL."
   (loop for w across (workers pool)
         do (start-worker w)))
 
 ;;; Task
 (defclass task ()
   ((state :initform nil :initarg :state :accessor task-state))
   (:documentation "This object represents a single unit of work to be done by some
 worker. Tasks are typically distributed from the task-pool, but workers may
 also be granted the ability to create and distribute their own tasks. Once a
 task is assigned, the 'owner', i.e. the worker that is assigned this task, may
 modify the object. When the work associated with a task is complete, the owner
 is responsible for indicating in the state slot the result of the computation."))
 
 (defmethod print-object ((self task) stream)
   (print-unreadable-object (self stream :type t)
     (format stream ":state ~A" (task-state self))))
 
 (defun run-task (worker task)
   (run-worker worker :input task))
 
 ;;; Job
 (defclass job (task)
   ((tasks :initform (make-array 0 :element-type 'task :fill-pointer 0 :adjustable t)
           :type (array task *)
           :initarg :tasks
           :accessor tasks)
    (lock :initform (make-mutex :name "job") :type mutex
          :initarg :lock))
   (:documentation "A collection of tasks forming a single unit of work."))
 
 (declaim (inline make-job))
 (defun make-job (&rest tasks)
   (make-instance 'job
     :tasks (make-array (length tasks) 
                        :element-type 'task
                        :initial-contents tasks)))
 
 (defmethod print-object ((self job) stream)
   (print-unreadable-object (self stream :type t)
     (format stream "~A tasks" (length (tasks self)))))
 
 (defun run-job (worker job)
   (run-worker worker :input job))
 
 ;;; Work Scope
 (defclass work-scope ()
   ((tasks  :initform (make-array 0 :element-type 'task :fill-pointer 0 :adjustable t)
            :initarg :tasks
            :accessor tasks
            :type (vector task))
    (lock :initform (make-mutex :name "work-scope") :initarg :lock :accessor work-scope-lock :type mutex)))
 
 (defmethod print-object ((self work-scope) stream)
   (print-unreadable-object (self stream :type t)
     (format stream "~A" (tasks self))))
 
 ;;; Macros
 (defmacro with-task-pool ((sym &key oracle (tasks 0) lock (workers 4) start kernel results) &body body)
   (unless lock (setf lock (make-semaphore :name "online" :count workers)))
   (unless results (setf results (make-mailbox :name "results")))
   `(let ((,sym (make-task-pool :lock ,lock :results ,results 
                                :tasks (make-queue 
                                        :name "tasks"
                                        :initial-contents
                                        (make-array ,tasks 
                                                    :element-type 'task 
                                                    :initial-element (make-instance 'task))))))
      ,@(if kernel `((setf (task-pool-kernel ,sym) ,kernel))
            `((setf (task-pool-kernel ,sym)
                    (gen-task-kernel (gensym "TASK-KERNEL") ()
                        (task-pool-lock ,sym) 
                        (tasks ,sym) 
                        (results ,sym)
                        nil))))
      (loop for i below ,workers
            do (push-worker (make-worker :kernel (task-pool-kernel ,sym)) ,sym))
      ,@(when oracle `((designate-oracle ,sym ,oracle)))
      ,@(when start `((start-task-workers ,sym)))
      ,@body))