Recently on this list, Paul referred to an "atomic integer swap" and an "atomic pointer swap."  This was a new concept to me (and possibly others), and this e-mail shares what I've learned.

If you access a variable from multiple threads -- even a built-in variable like 'int', it is important to control access to the variable with a mutex, semaphore, or an atomic operation.

Alexander Sandler, on his blog, wrote a couple of good articles on the subject:

   "Do you need a mutex to protect an int?"
   http://www.alexonlinux.com/do-you-need-mutex-to-protect-int

   "Multithreaded simple data type access and atomic
   variables"
   http://www.alexonlinux.com/multithreaded-simple-data-type-access-and-atomic-variables

The first article contains code that calculates a wrong answer on multiprocessor machines.  I've attached a similar example that will even fail on a single-processor machine.

There is a wealth of reading material on using Mutexes and Semaphores.  However, information on atomic operations appears to be sparse and hard-to-follow.  So, here's what I've found:

   + At the moment, there is no built-in support in
     C/C++ for atomic operations.  You will need to use
     a library, compiler extension, or write your own
     in assembly code.

   + The GCC compiler has the built-in __sync_*()
     functions[1] that provide atomic operations.
     Note that the attached example is using this.

   + glib provides the g_atomic_*() functions[2].

   + Qt 4 has the q_atomic_*() functions.[3]  While
     they are accessible, they are /not/ a part of
     their stable, public API.

   + The next version of ISO C++ (code name c++0x)
     is expected to have support for atomic operations
     (E.g. the std::atomic<T> template) and memory
     barriers.  It may even require that all built-in
     types be atomic.

   + In the x86 instruction set, these are usually
     implemented using the 'LOCK' instruction prefix.[5]

When using atomic operations, perhaps the best advice I found is near the end of Sandler's second article:

   "When using atomic variables, some extra
   precautions have to be taken....  There is nothing
   that prevents you from incrementing value of the
   atomic variable with __sync_fetch_and_add() as I
   just demonstrated and later in the code doing same
   thing with regular ++ operator.

   "To address this problem, I strongly suggest
   wrapping around atomic functions and variables with
   either ADT in C or C++ class."[4]

Peace,
Gabriel

[1] http://gcc.gnu.org/onlinedocs/gcc-4.1.0/gcc/Atomic-Builtins.html
[2] http://www.gtk.org/api/2.6/glib/glib-Atomic-Operations.html
[3] http://doc.trolltech.com/4.3/atomic-operations.html
   See also the Qt header file QtCore/qatomic_i386.h, and
   its brothers.
[4] http://www.alexonlinux.com/multithreaded-simple-data-type-access-and-atomic-variables#precautions
[5] http://siyobik.info/index.php?module=x86&id=159<atomic_fail.c>_______________________________________________
Linux-audio-dev mailing list
Linux-audio-dev@lists.linuxaudio.org
http://lists.linuxaudio.org/mailman/listinfo/linux-audio-dev