[PATCH 24/24] document volatile atomic_read() behavior

[Chris Snook <csnook@redhat.com>]

From: Chris Snook <csnook@redhat.com>

Update atomic_ops.txt to reflect the newly consistent behavior of
atomic_read(), and to note that volatile (in declarations) is now
considered harmful.

Signed-off-by: Chris Snook <csnook@redhat.com>

--- linux-2.6.23-rc2-orig/Documentation/atomic_ops.txt	2007-07-08 19:32:17.000000000 -0400
+++ linux-2.6.23-rc2/Documentation/atomic_ops.txt	2007-08-09 08:24:32.000000000 -0400
@@ -12,7 +12,7 @@
 C integer type will fail.  Something like the following should
 suffice:
 
-	typedef struct { volatile int counter; } atomic_t;
+	typedef struct { int counter; } atomic_t;
 
 	The first operations to implement for atomic_t's are the
 initializers and plain reads.
@@ -38,9 +38,17 @@
 
 Next, we have:
 
-	#define atomic_read(v)	((v)->counter)
+	#define atomic_read(v)	(*(volatile int *)&(v)->counter)
 
-which simply reads the current value of the counter.
+which reads the counter as though it were volatile.  This prevents the
+compiler from optimizing away repeated atomic_read() invocations without
+requiring a more expensive barrier().  Historically this has been
+accomplished by declaring the counter itself to be volatile, but the
+ambiguity of the C standard on the semantics of volatile make this practice
+vulnerable to overly creative interpretation by compilers.  Explicit
+casting in atomic_read() ensures consistent behavior across architectures
+and compilers.  Even with this convenience in atomic_read(), busy-waiters
+should call cpu_relax().
 
 Now, we move onto the actual atomic operation interfaces.