[Asterisk-Dev] Race issue in channel.c involving uniqueint on Asterisk 1.2.1

[Kevin P. Fleming]

David Woodhouse wrote:

> Well, there's asm/atomic.h which might compile on some architectures,
> but it's likely to be broken. Do not make the mistake of using it -- if
> it does compile, you may well get non-atomic behaviour from it.

On the x86-type architectures at least, using it seems to be fine as 
long as you set CONFIG_SMP to guarantee the lock prefixes will be 
generated. I don't know about the other architectures, but we could 
certainly take it slowly and only enable the use of these operations on 
arches/kernel versions that are known good combinations.