RWLock#

Added in version 2.32.

class RWLock(*args, **kwargs)#

The GRWLock struct is an opaque data structure to represent a reader-writer lock. It is similar to a Mutex in that it allows multiple threads to coordinate access to a shared resource.

The difference to a mutex is that a reader-writer lock discriminates between read-only (‘reader’) and full (‘writer’) access. While only one thread at a time is allowed write access (by holding the ‘writer’ lock via writer_lock()), multiple threads can gain simultaneous read-only access (by holding the ‘reader’ lock via reader_lock()).

It is unspecified whether readers or writers have priority in acquiring the lock when a reader already holds the lock and a writer is queued to acquire it.

Here is an example for an array with access functions:

GRWLock lock;
GPtrArray *array;

gpointer
my_array_get (guint index)
{
  gpointer retval = NULL;

  if (!array)
    return NULL;

  g_rw_lock_reader_lock (&lock);
  if (index < array->len)
    retval = g_ptr_array_index (array, index);
  g_rw_lock_reader_unlock (&lock);

  return retval;
}

void
my_array_set (guint index, gpointer data)
{
  g_rw_lock_writer_lock (&lock);

  if (!array)
    array = g_ptr_array_new ();

  if (index >= array->len)
    g_ptr_array_set_size (array, index+1);
  g_ptr_array_index (array, index) = data;

  g_rw_lock_writer_unlock (&lock);
}

This example shows an array which can be accessed by many readers (the my_array_get() function) simultaneously, whereas the writers (the my_array_set() function) will only be allowed one at a time and only if no readers currently access the array. This is because of the potentially dangerous resizing of the array. Using these functions is fully multi-thread safe now.

If a RWLock is allocated in static storage then it can be used without initialisation. Otherwise, you should call init() on it and clear() when done.

A GRWLock should only be accessed with the g_rw_lock_ functions.

Methods#

class RWLock
clear() None#

Frees the resources allocated to a lock with init().

This function should not be used with a RWLock that has been statically allocated.

Calling clear() when any thread holds the lock leads to undefined behaviour.

Added in version 2.32.

init() None#

Initializes a RWLock so that it can be used.

This function is useful to initialize a lock that has been allocated on the stack, or as part of a larger structure. It is not necessary to initialise a reader-writer lock that has been statically allocated.

  typedef struct {
    GRWLock l;
    ...
  } Blob;

Blob *b;

b = g_new (Blob, 1);
g_rw_lock_init (&b->l);

To undo the effect of init() when a lock is no longer needed, use clear().

Calling init() on an already initialized RWLock leads to undefined behaviour.

Added in version 2.32.

reader_lock() None#

Obtain a read lock on rw_lock. If another thread currently holds the write lock on rw_lock, the current thread will block until the write lock was (held and) released. If another thread does not hold the write lock, but is waiting for it, it is implementation defined whether the reader or writer will block. Read locks can be taken recursively.

Calling reader_lock() while the current thread already owns a write lock leads to undefined behaviour. Read locks however can be taken recursively, in which case you need to make sure to call reader_unlock() the same amount of times.

It is implementation-defined how many read locks are allowed to be held on the same lock simultaneously. If the limit is hit, or if a deadlock is detected, a critical warning will be emitted.

Added in version 2.32.

reader_trylock() bool#

Tries to obtain a read lock on rw_lock and returns True if the read lock was successfully obtained. Otherwise it returns False.

Added in version 2.32.

reader_unlock() None#

Release a read lock on rw_lock.

Calling reader_unlock() on a lock that is not held by the current thread leads to undefined behaviour.

Added in version 2.32.

writer_lock() None#

Obtain a write lock on rw_lock. If another thread currently holds a read or write lock on rw_lock, the current thread will block until all other threads have dropped their locks on rw_lock.

Calling writer_lock() while the current thread already owns a read or write lock on rw_lock leads to undefined behaviour.

Added in version 2.32.

writer_trylock() bool#

Tries to obtain a write lock on rw_lock. If another thread currently holds a read or write lock on rw_lock, it immediately returns False. Otherwise it locks rw_lock and returns True.

Added in version 2.32.

writer_unlock() None#

Release a write lock on rw_lock.

Calling writer_unlock() on a lock that is not held by the current thread leads to undefined behaviour.

Added in version 2.32.

Fields#

class RWLock
i#
p#