Unnamed POSIX shared semaphores

[Bruce Simpson]

Vlad Galu wrote:
> ...
> Thanks, Ivan. I'll take a better look at this after our first release,
> which is due in a couple of weeks. Right now the team efforts aren't
> focused on portability, so it's a low priority issue, but something
> we'd definitely like to have in the future.
>   

I stand corrected. Having read around this, I don't see how 
process-shared sems could work at all, although I haven't actually tried 
running process-shared sem code.

POSIX semaphores were however horribly broken in kernel prior to 7.2.
The fix was essentially one liner. We got a very good test case from a 
chap in a GNATS PR.

cheers
BMS