On Thu, Sep 11, 2003 at 10:12:49PM -0700, Aaron Bannert wrote:
>...
> Oh, and FWIW, I'm proposing we get rid of nested (aka recursive)
> mutexes,
> which is when the same thread can lock a mutex over and over w/o
> deadlocking. In what we call non-nested mutexes, you get the default
> posix behavior, which is the scenario you described above (whenever
> a thread tries to lock an already-acquired mutex, it will block).
Ah. Okay, just backwards of what I was describing, but no problems. I
guess I should retroactively edit my post to say, "here is an algorithm
that I'd like to ensure that I can still do." Your answer is, "sure,
dumbass. use the right thing -- condition variables."
I can live with that :-)
Cheers,
-g
--
Greg Stein, http://www.lyra.org/
|