cvs commit: src/sys/kern kern_proc.c

[Poul-Henning Kamp]

In message <20040609182041.GV12007 at darkness.comp.waw.pl>, Pawel Jakub Dawidek w
rites:

>BTW. Do we assume that atomic operations are atomic on MP machines?
>I read atomic(9), but I still don't know if I should use it or not for
>things like those.

Ok, let us dispell the first misconception about atomic operations:

	Atomic operations are not cheap!

In fact, there are architectures where they are significantly more
expensive than mutexes.

Next:  Atomic operations are much harder to grasp mentally, and
close to impossible to validate automatically.

We have no WITNESS which will tell us about problems with atomic
operations.

Therefore:  Do not use atomic operations, use a mutex.

If we find a place where atomics will result in a significant
performance increase, we will apply them there after careful
consideration and considerable benchmarking.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.