Threads

Threads — Functions for using GDK in multi-threaded programs

Synopsis


#include <gdk/gdk.h>

#define             GDK_THREADS_ENTER                   ()
#define             GDK_THREADS_LEAVE                   ()
void                gdk_threads_init                    (void);
void                gdk_threads_enter                   (void);
void                gdk_threads_leave                   (void);
extern              GMutex *gdk_threads_mutex;
void                gdk_threads_set_lock_functions      (GCallback enter_fn,
                                                         GCallback leave_fn);
guint               gdk_threads_add_idle                (GSourceFunc function,
                                                         gpointer data);
guint               gdk_threads_add_idle_full           (gint priority,
                                                         GSourceFunc function,
                                                         gpointer data,
                                                         GDestroyNotify notify);
guint               gdk_threads_add_timeout             (guint interval,
                                                         GSourceFunc function,
                                                         gpointer data);
guint               gdk_threads_add_timeout_full        (gint priority,
                                                         guint interval,
                                                         GSourceFunc function,
                                                         gpointer data,
                                                         GDestroyNotify notify);

Description

For thread safety, GDK relies on the thread primitives in GLib, and on the thread-safe GLib main loop.

GLib is completely thread safe (all global data is automatically locked), but individual data structure instances are not automatically locked for performance reasons. So e.g. you must coordinate accesses to the same GHashTable from multiple threads.

GTK+ is "thread aware" but not thread safe — it provides a global lock controlled by gdk_threads_enter()/gdk_threads_leave() which protects all use of GTK+. That is, only one thread can use GTK+ at any given time.

Unfortunately the above holds with the X11 backend only. With the Win32 backend, GDK calls should not be attempted from multiple threads at all.

You must call g_thread_init() and gdk_threads_init() before executing any other GTK+ or GDK functions in a threaded GTK+ program.

Idles, timeouts, and input functions from GLib, such as g_idle_add(), are executed outside of the main GTK+ lock. So, if you need to call GTK+ inside of such a callback, you must surround the callback with a gdk_threads_enter()/gdk_threads_leave() pair or use gdk_threads_add_idle_full() which does this for you. However, event dispatching from the mainloop is still executed within the main GTK+ lock, so callback functions connected to event signals like GtkWidget::button-press-event, do not need thread protection.

In particular, this means, if you are writing widgets that might be used in threaded programs, you must surround timeouts and idle functions in this matter.

As always, you must also surround any calls to GTK+ not made within a signal handler with a gdk_threads_enter()/gdk_threads_leave() pair.

Before calling gdk_threads_leave() from a thread other than your main thread, you probably want to call gdk_flush() to send all pending commands to the windowing system. (The reason you don't need to do this from the main thread is that GDK always automatically flushes pending commands when it runs out of incoming events to process and has to sleep while waiting for more events.)

A minimal main program for a threaded GTK+ application looks like:

int
main (int argc, char *argv[])
{
  GtkWidget *window;

  g_thread_init (NULL);
  gdk_threads_init ();
  gdk_threads_enter ();

  gtk_init (&argc, &argv);

  window = create_window ();
  gtk_widget_show (window);

  gtk_main ();
  gdk_threads_leave ();

  return 0;
}

Callbacks require a bit of attention. Callbacks from GTK+ signals are made within the GTK+ lock. However callbacks from GLib (timeouts, IO callbacks, and idle functions) are made outside of the GTK+ lock. So, within a signal handler you do not need to call gdk_threads_enter(), but within the other types of callbacks, you do.

Erik Mouw contributed the following code example to illustrate how to use threads within GTK+ programs.

/*-------------------------------------------------------------------------
 * Filename:      gtk-thread.c
 * Version:       0.99.1
 * Copyright:     Copyright (C) 1999, Erik Mouw
 * Author:        Erik Mouw <J.A.K.Mouw@its.tudelft.nl>
 * Description:   GTK threads example. 
 * Created at:    Sun Oct 17 21:27:09 1999
 * Modified by:   Erik Mouw <J.A.K.Mouw@its.tudelft.nl>
 * Modified at:   Sun Oct 24 17:21:41 1999
 *-----------------------------------------------------------------------*/
/*
 * Compile with:
 *
 * cc -o gtk-thread gtk-thread.c `gtk-config --cflags --libs gthread`
 *
 * Thanks to Sebastian Wilhelmi and Owen Taylor for pointing out some
 * bugs.
 *
 */

#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <time.h>
#include <gtk/gtk.h>
#include <glib.h>
#include <pthread.h>

#define YES_IT_IS    (1)
#define NO_IT_IS_NOT (0)

typedef struct 
{
  GtkWidget *label;
  int what;
} yes_or_no_args;

G_LOCK_DEFINE_STATIC (yes_or_no);
static volatile int yes_or_no = YES_IT_IS;

void destroy (GtkWidget *widget, gpointer data)
{
  gtk_main_quit ();
}

void *argument_thread (void *args)
{
  yes_or_no_args *data = (yes_or_no_args *)args;
  gboolean say_something;

  for (;;)
    {
      /* sleep a while */
      sleep(rand() / (RAND_MAX / 3) + 1);

      /* lock the yes_or_no_variable */
      G_LOCK(yes_or_no);

      /* do we have to say something? */
      say_something = (yes_or_no != data->what);

      if(say_something)
	{
	  /* set the variable */
	  yes_or_no = data->what;
	}

      /* Unlock the yes_or_no variable */
      G_UNLOCK (yes_or_no);

      if (say_something)
	{
	  /* get GTK thread lock */
	  gdk_threads_enter ();

	  /* set label text */
	  if(data->what == YES_IT_IS)
	    gtk_label_set_text (GTK_LABEL (data->label), "O yes, it is!");
	  else
	    gtk_label_set_text (GTK_LABEL (data->label), "O no, it isn't!");

	  /* release GTK thread lock */
	  gdk_threads_leave ();
	}
    }

  return NULL;
}

int main (int argc, char *argv[])
{
  GtkWidget *window;
  GtkWidget *label;
  yes_or_no_args yes_args, no_args;
  pthread_t no_tid, yes_tid;

  /* init threads */
  g_thread_init (NULL);
  gdk_threads_init ();
  gdk_threads_enter ();

  /* init gtk */
  gtk_init(&argc, &argv);

  /* init random number generator */
  srand ((unsigned int) time (NULL));

  /* create a window */
  window = gtk_window_new (GTK_WINDOW_TOPLEVEL);

  gtk_signal_connect (GTK_OBJECT (window), "destroy",
		      GTK_SIGNAL_FUNC (destroy), NULL);

  gtk_container_set_border_width (GTK_CONTAINER (window), 10);

  /* create a label */
  label = gtk_label_new ("And now for something completely different ...");
  gtk_container_add (GTK_CONTAINER (window), label);
  
  /* show everything */
  gtk_widget_show (label);
  gtk_widget_show (window);

  /* create the threads */
  yes_args.label = label;
  yes_args.what = YES_IT_IS;
  pthread_create (&yes_tid, NULL, argument_thread, &yes_args);

  no_args.label = label;
  no_args.what = NO_IT_IS_NOT;
  pthread_create (&no_tid, NULL, argument_thread, &no_args);

  /* enter the GTK main loop */
  gtk_main ();
  gdk_threads_leave ();

  return 0;
}

Details

GDK_THREADS_ENTER()

#define             GDK_THREADS_ENTER()

This macro marks the beginning of a critical section in which GDK and GTK+ functions can be called safely and without causing race conditions. Only one thread at a time can be in such a critial section. The macro expands to a no-op if G_THREADS_ENABLED has not been defined. Typically gdk_threads_enter() should be used instead of this macro.


GDK_THREADS_LEAVE()

#define             GDK_THREADS_LEAVE()

This macro marks the end of a critical section begun with GDK_THREADS_ENTER.


gdk_threads_init ()

void                gdk_threads_init                    (void);

Initializes GDK so that it can be used from multiple threads in conjunction with gdk_threads_enter() and gdk_threads_leave(). g_thread_init() must be called previous to this function.

This call must be made before any use of the main loop from GTK+; to be safe, call it before gtk_init().


gdk_threads_enter ()

void                gdk_threads_enter                   (void);

This macro marks the beginning of a critical section in which GDK and GTK+ functions can be called safely and without causing race conditions. Only one thread at a time can be in such a critial section.


gdk_threads_leave ()

void                gdk_threads_leave                   (void);

Leaves a critical region begun with gdk_threads_enter().


gdk_threads_mutex

extern GMutex *gdk_threads_mutex; /* private */

Warning

gdk_threads_mutex is deprecated and should not be used in newly-written code.

The GMutex used to implement the critical region for gdk_threads_enter()/gdk_threads_leave().


gdk_threads_set_lock_functions ()

void                gdk_threads_set_lock_functions      (GCallback enter_fn,
                                                         GCallback leave_fn);

Allows the application to replace the standard method that GDK uses to protect its data structures. Normally, GDK creates a single GMutex that is locked by gdk_threads_enter(), and released by gdk_threads_leave(); using this function an application provides, instead, a function enter_fn that is called by gdk_threads_enter() and a function leave_fn that is called by gdk_threads_leave().

The functions must provide at least same locking functionality as the default implementation, but can also do extra application specific processing.

As an example, consider an application that has its own recursive lock that when held, holds the GTK+ lock as well. When GTK+ unlocks the GTK+ lock when entering a recursive main loop, the application must temporarily release its lock as well.

Most threaded GTK+ apps won't need to use this method.

This method must be called before gdk_threads_init(), and cannot be called multiple times.

enter_fn : function called to guard GDK
leave_fn : function called to release the guard

Since 2.4


gdk_threads_add_idle ()

guint               gdk_threads_add_idle                (GSourceFunc function,
                                                         gpointer data);

A wrapper for the common usage of gdk_threads_add_idle_full() assigning the default priority, G_PRIORITY_DEFAULT_IDLE.

See gdk_threads_add_idle_full().

function : function to call
data : data to pass to function
Returns : the ID (greater than 0) of the event source.

Since 2.12


gdk_threads_add_idle_full ()

guint               gdk_threads_add_idle_full           (gint priority,
                                                         GSourceFunc function,
                                                         gpointer data,
                                                         GDestroyNotify notify);

Adds a function to be called whenever there are no higher priority events pending. If the function returns FALSE it is automatically removed from the list of event sources and will not be called again.

This variant of g_idle_add_full() calls function with the GDK lock held. It can be thought of a MT-safe version for GTK+ widgets for the following use case, where you have to worry about idle_callback() running in thread A and accessing self after it has been finalized in thread B:

static gboolean
idle_callback (gpointer data)
{
   // gdk_threads_enter(); would be needed for g_idle_add()
   
   SomeWidget *self = data;
   /* do stuff with self */
   
   self->idle_id = 0;
   
   // gdk_threads_leave(); would be needed for g_idle_add()
   return FALSE;
}
 
static void
some_widget_do_stuff_later (SomeWidget *self)
{
   self->idle_id = gdk_threads_add_idle (idle_callback, self)
   // using g_idle_add() here would require thread protection in the callback
}
 
static void
some_widget_finalize (GObject *object)
{
   SomeWidget *self = SOME_WIDGET (object);
   if (self->idle_id)
     g_source_remove (self->idle_id);
   G_OBJECT_CLASS (parent_class)->finalize (object);
}

priority : the priority of the idle source. Typically this will be in the range btweeen G_PRIORITY_DEFAULT_IDLE and G_PRIORITY_HIGH_IDLE
function : function to call
data : data to pass to function
notify : function to call when the idle is removed, or NULL
Returns : the ID (greater than 0) of the event source.

Since 2.12


gdk_threads_add_timeout ()

guint               gdk_threads_add_timeout             (guint interval,
                                                         GSourceFunc function,
                                                         gpointer data);

A wrapper for the common usage of gdk_threads_add_timeout_full() assigning the default priority, G_PRIORITY_DEFAULT.

See gdk_threads_add_timeout_full().

interval : the time between calls to the function, in milliseconds (1/1000ths of a second)
function : function to call
data : data to pass to function
Returns : the ID (greater than 0) of the event source.

Since 2.12


gdk_threads_add_timeout_full ()

guint               gdk_threads_add_timeout_full        (gint priority,
                                                         guint interval,
                                                         GSourceFunc function,
                                                         gpointer data,
                                                         GDestroyNotify notify);

Sets a function to be called at regular intervals holding the GDK lock, with the given priority. The function is called repeatedly until it returns FALSE, at which point the timeout is automatically destroyed and the function will not be called again. The notify function is called when the timeout is destroyed. The first call to the function will be at the end of the first interval.

Note that timeout functions may be delayed, due to the processing of other event sources. Thus they should not be relied on for precise timing. After each call to the timeout function, the time of the next timeout is recalculated based on the current time and the given interval (it does not try to 'catch up' time lost in delays).

This variant of g_timeout_add_full() can be thought of a MT-safe version for GTK+ widgets for the following use case:

static gboolean timeout_callback (gpointer data)
{
   SomeWidget *self = data;
   
   /* do stuff with self */
   
   self->timeout_id = 0;
   
   return FALSE;
}
 
static void some_widget_do_stuff_later (SomeWidget *self)
{
   self->timeout_id = g_timeout_add (timeout_callback, self)
}
 
static void some_widget_finalize (GObject *object)
{
   SomeWidget *self = SOME_WIDGET(object);
   
   if (self->timeout_id)
     g_source_remove (self->timeout_id);
   
   G_OBJECT_CLASS (parent_class)->finalize (object);
}

priority : the priority of the timeout source. Typically this will be in the range between G_PRIORITY_DEFAULT_IDLE and G_PRIORITY_HIGH_IDLE.
interval : the time between calls to the function, in milliseconds (1/1000ths of a second)
function : function to call
data : data to pass to function
notify : function to call when the timeout is removed, or NULL
Returns : the ID (greater than 0) of the event source.

Since 2.12