Re: SCO: "thread creation is about a thousand times faster than

From: Theodore Y. Ts'o (tytso@MIT.EDU)
Date: Mon Aug 28 2000 - 12:03:58 EST


   Date: Sun, 27 Aug 2000 22:36:46 -0500 (CDT)
   From: Dave McCracken <dcm@mccr.org>

   I'll also admit signals were always a mess. They never fit very well
   with pthreads, and we pretty much punted in the early drafts. I
   added sigwait() to the original draft as a way an application could
   direct all its signal handling to a single synchronous 'signal
   handler' thread, but it never addressed the larger question of how to
   make signals and threads play together.

I'll note that signals is one of the priamry places where Linuxthreads
aren't compatible with Posix threads, and most of the various attempts
to fix this involved kernel hacks that would have penelized the
performance of the kernel in general, not just when dealing with Posix
brain damage. (In one case, I believe the estimates were that it would
slow signal delivery time by a factor of six.)

So that's probably one of the reasons why some Linux kernel folks have
simply taken to referring to Posix threads as being incurably broken.
(And we haven't even gotten to pthread_cancel yet....)

                                                - Ted

P.S. I still very firmly believe that given how few systems programmers
can really grok asynchronous programming in the kernel, with all of its
pitfalls and locking requirements, that giving threads to Applications
Programmers is roughly the equivalent of giving a loaded .45 to a
chimpanzee. :-)

So I've never been particular fan of threads in user space to begin
with.... and so I have a theory that the combination of folks who (a)
have good taste, (b) understand Posix threads, and (c) are psyched about
trying to bring Posix threads to Linux has been a (lamentably) small
number.

There are those who believe that (b) removes any possibility of (a),
sort of like Dykstra's comment about exposure to goto causing permanent
brain damage making it impossible for that person to write good programs
ever again. While I really don't like threads, I don't *think* that's
true.....

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Thu Aug 31 2000 - 21:00:21 EST