Next: , Previous: , Up: Multitasker   [Contents][Index]


6.26.1 Ptheads

Tasks can be created with newtask or newtask4 with a given amount of stack space (either all the same or each stack’s size specified); these tasks neet to be activated or send an xt through initiate. Tasks can stop themselves when they are done or wait for new instructions.

newtask ( stacksize – task  ) gforth-experimental “newtask”

creates a task, uses stacksize for stack, rstack, fpstack, locals

task ( stacksize "name" –  ) gforth-experimental “task”

create a named task with stacksize stacksize

execute-task ( xt – task  ) gforth-experimental “execute-task”

create a new task task and initiate it with xt

stacksize ( – u  ) gforth-experimental “stacksize”

u is the data stack size of the main task.

newtask4 ( dsize rsize fsize lsize – task  ) gforth-experimental “newtask4”

creates a task, each stack individually sized

stacksize4 ( – u-data u-return u-fp u-locals  ) gforth-experimental “stacksize4”

Pushes the data, return, FP, and locals stack sizes of the main task.

activate ( task –  ) gforth-experimental “activate”

activates a task. The remaining part of the word calling activate will be executed in the context of the task.

pass ( x1 .. xn n task –  ) gforth-experimental “pass”

activates task, and passes n parameters from the data stack

initiate ( xt task –  ) gforth-experimental “initiate”

pass an xt to a task (VFX compatible)

pause ( ) gforth-experimental “pause”

voluntarily switch to the next waiting task (pause is the traditional cooperative task switcher; in the pthread multitasker, you don’t need pause for cooperation, but you still can use it e.g. when you have to resort to polling for some reason). This also checks for events in the queue.

restart ( task –  ) gforth-experimental “restart”

Wake a task

halt ( task –  ) gforth-experimental “halt”

Stop a task

stop ( ) gforth-experimental “stop”

stops the current task, and waits for events (which may restart it)

stop-ns ( timeout –  ) gforth-experimental “stop-ns”

Stop with timeout (in nanoseconds), better replacement for ms

A cooperative multitasker can ensure that there is no other task interacting between two invocations of pause. Pthreads however are really concurrent tasks (at least on a multi-core CPU), and therefore, several techniques to avoid conflicts when accessing the same resources.

6.26.1.1 Special User Variables

Aside from the user variables that are already defined in the kernel, tasks may want to have user values and user defers, optain the offset of a user variable, or the address of those related to another task to initialize that task’s user area.

UValue ( "name" –  ) gforth-1.0 “UValue”
UDefer ( "name" –  ) gforth-experimental “UDefer”

Define a per-thread deferred word

user' ( ’user’ – n  ) gforth-experimental “user”’

USER’ computes the task offset of a user variable

's ( user task – user’  ) gforth-experimental “’s”

get the tasks’s address of our user variable

6.26.1.2 Semaphores

Semaphores can only be aquired by one thread, all other threads have to wait until the semapohre is released.

semaphore ( "name" –  ) gforth-experimental “semaphore”

create a named semaphore "name" \\ "name"-execution: ( – semaphore )

lock ( semaphore –  ) gforth-experimental “lock”

lock the semaphore

unlock ( semaphore –  ) gforth-experimental “unlock”

unlock the semaphore

The other approach to prevent concurrent access is the critical section. Here, we implement a critical section with a semaphore, so you have to specify the semaphore which is used for the critical section. Only those critical sections which use the same semaphore are mutually exclusive.

critical-section ( xt semaphore –  ) gforth-experimental “critical-section”

implement a critical section that will unlock the semaphore even in case there’s an exception within.

6.26.1.3 Atomic operations

Atomic operations can be used to synchronize tasks without using slow OS primitives.

!@ ( u1 a-addr – u2 ) gforth-experimental “store-fetch”

load u2 from a_addr, and store u1 there, as atomic operation

+!@ ( u1 a-addr – u2 ) gforth-experimental “add-store-fetch”

load u2 from a_addr, and increment this location by u1, as atomic operation

?!@ ( unew uold a-addr – uprev ) gforth-experimental “question-store-fetch”

load uprev from a_addr, compare it to uold, and if equal, store unew there, as atomic operation

barrier ( ) gforth-experimental “barrier”

Insert a full memory barrier

6.26.1.4 Message Queues

Gforth implements executable message queues for event driven programs: you send instructions to other tasks, enclosed in <event and event>; the entire event sequence is executed atomically. You can pass integers, floats, and strings (only the addresses, so treat the string as read-only after you have send it to another task). The messages you send are defined with event: name, which, when invoked, will add the code for its execution to the message queue, and when recieved, will execute the code following. The message queue is queried when you stop a task, or when you check for events with ?events. You can define a maximum of 256 different events.

<event ( ) gforth-experimental “<event”

starts a sequence of events.

event> ( task –  ) gforth-experimental “event>”

ends a sequence and sends it to the mentioned task

event: ( "name" –  ) gforth-experimental “event:”

defines an event and the reaction to it as Forth code. If name is invoked, the event gets assembled to the event buffer. If the event name is received, the Forth definition that follows the event declaration is executed.

?events ( ) gforth-experimental “?events”

checks for events and executes them

event-loop ( ) gforth-experimental “event-loop”

Tasks that are controlled by sending events to them should go into an event-loop

elit, ( x –  ) gforth-experimental “elit,”

sends a literal

e$, ( addr u –  ) gforth-experimental “e$,”

sends a string (actually only the address and the count, because it’s shared memory

eflit, ( x –  ) gforth-experimental “eflit,”

sends a float

The naming conventions for events is :>name.

6.26.1.5 Conditions

The pthreads library also provides conditional variables, which allow to wait for a condition. Using the message queue is generally preferred.

cond ( "name" –  ) gforth-experimental “cond”

create a named condition

pthread_cond_signal ( cond – r  ) gforth-experimental “pthread_cond_signal”
pthread_cond_broadcast ( cond – r  ) gforth-experimental “pthread_cond_broadcast”
pthread_cond_wait ( cond mutex – r  ) gforth-experimental “pthread_cond_wait”
pthread_cond_timedwait ( cond mutex abstime – r  ) gforth-experimental “pthread_cond_timedwait”

Next: Cilk, Previous: Multitasker, Up: Multitasker   [Contents][Index]