Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watmath!clyde!rutgers!gatech!udel!rochester!dibble
From: dibble@rochester.UUCP
Newsgroups: comp.sys.m6809
Subject: Re: True Multitasking, and some history lessons
Message-ID: <28454@rochester.ARPA>
Date: Sat, 13-Jun-87 13:13:12 EDT
Article-I.D.: rocheste.28454
Posted: Sat Jun 13 13:13:12 1987
Date-Received: Sat, 13-Jun-87 23:53:19 EDT
References: <8706040024.AA10895@cogsci.berkeley.edu> <2194@husc6.UUCP> <1738@ihwpt.ATT.COM>
Organization: U of Rochester, CS Dept., Rochester, NY
Lines: 37

In article <1738@ihwpt.ATT.COM>, knudsen@ihwpt.ATT.COM (mike knudsen) writes:
> Seems to me that however you do a shared stdio (C runtime) library,
> at least some of its code has to map into your process's space during
> the call.  

So far OS-9/68K doesn't have a version with memory management.  EVERYTHING
is mapped into the trap handler's address space.  Since the trap handlers
only get direct access to the registers, any information that is not
in a register (like a line to print), must be seen by dereferencing
a pointer (this is a guess, but I can't think of any way around it).
If (when) a version of OS-9/68K with full memory management comes out
the trap handlers will have to use the sys-call move instructions to
follow the pointers.
> 
>  Considering subroutine libraries for the 6809.

They would be a great idea, and OS-9 was designed for them.
The original idea was that vast libraries of packaged solutions
would be available in ROM.  The idea never caught on perhaps because
there weren't enought customers in the OS-9 community to justify 
the cost of developing the software.

The two packages that come with OS-9/68K (stdio and floating point)
would certainly make my life with assembly programming for the
6809 easier.  I'd be pleased with just the floating point library
especially if it included all conversions.  Come to think of
it, it probably wouldn't be that hard to write one!

Maybe we should make a project of it.  (If we succeed I think I could
get it published.)  After a numeric library I'd go for string
handling.

Problem: If the libraries are mapped into the address spaces that use
them, they will use up space there.  They could live in a separate
address space, and use an rpc-like mechanism, but it would be slower.

Peter