...

Processor Tools

Memory fences: Mfence

See Section 8.2 (Memory Ordering) in Intel 3a System Programming Manual. This suggests you can actually get away with a fair bit. I think it also says that the LOCK prefix and XCHG instruction act like an mfence (TODO: confirm).

Compiler Tools

Inline assembly

...

So it seems the only safe way to make sure an asm block is not not moved or deleted by the compiler is to use volatile together with a memory clobber.

Volatile keyword

...

On data

On pointers

on data or pointers

lkml emails from linus http://lwn.net/Articles/233481/:

If the memory barriers are right, then the "volatile" doesn't matter. And if the memory barriers aren't right, then "volatile" doesn't help.

Basically, a volatile on a data structure can NEVER be right. If it makes a difference, it's a sign of improper locking.

potentially make the compiler generate worse code for no reason (the "no reason" being that if there aren't any barriers in between, the compiler should merge accesses)

The only thing "volatile" can ever do is generate worse code, WITH NO UPSIDES.

actual accessor functions can use it in a cast to make one particular access follow the rules of "don't cache this one dereference". That is useful as part of a bigger set of rules about that access (i.e., it might be the internal implementation of a "readb()", for example).

On pointers

In Section 7.1.6.1 (The cv-qualifiers) of the C++ standard:

A pointer or reference to a cv-qualified type need not actually point or refer to a cv-qualified object, but it is treated as if it does.

So you can use a pointer marked volatile to signify volatile accesses, i.e., those that should be read from memory. However, this is useful for interacting with IO devices, not for concurrency where you need memory barriers anyway.

Survey of Existing Concurrency Primitives

...

where __volatile is defined as volatile in sys/cdefs.h. I couldn't find the definition for __asm, but I assume it's the same as asm.

Google perf tools

http://google-perftools.googlecode.com/svn/trunk/src/base/
They have a set of BSD-licensed atomic primitives, a spinlock, etc.

References

Extended Asm in the GCC manual describes how to use inline assembly in gcc. Unfortunately, it is not very precise or thorough. When describing the '+' constraint modifier, it says you should not use '+m', but Linus claims this is not true (see Re: [PATCH 3/8] i386: bitops: Rectify bogus "+m" constraints).

...