Re: [9fans] [OT] linux origins, why not?

Re: [9fans] [OT] linux origins, why not?

[erik quanstrom]

sorry for the stupid question.  i didn't mean to step in that.

- erik

On Wed Mar 29 22:28:01 CST 2006, jmk@plan9.bell-labs.com wrote:
> On Wed Mar 29 20:52:09 EST 2006, quanstro@quanstro.net wrote:
> > do you have some pointers to papers from these guys?
> >
> > from my uneducated position, it seems to me that plan9 has a large
> > percentage of what microkernels claim.  one thing one can't
> > do is write a hardware driver that lives in userspace.  one advantage
> > of this could be the ability to load drivers depending on configuration.
> >
> > has anybody invested some brain cells in this?
> >
> > - erik
>
> yes, brain cells have been burned on that and it has been done.
> but we're waiting for uriel's replacement for 9load before the
> details are finalised. after all, we don't want to waste our time
> doing something that's already been done but just not released.
>
> --jim

Re: [9fans] [OT] linux origins, why not?

[Dave Eckhardt]

> i don't see how mach is an improvement over linux.  especially
> early linux.  mach kernels were three times the size of linux
> kernels of the day and didn't do anything useful by themselves.

What the AT&T lawsuit blocked distribution of was Mach 2.5, which
was BSD Unix with parts of the kernel replaced.  One unified source
tree ran on several architectures (VAX, 68k, MIPS, x86; later 88k,
Sparc).  Commercial multiprocessor Unix machines were sold running
that code base (the Encore Multimax, and I *think* the Sequent
Symmetry).

As for utility, Mach on VAX and 68k Sun and DEC PMAX and PC hardware
was the computing environment for hundreds of people at a time at CMU
for a good chunk of a decade.  Multiprocessor desktop workstations
(4-way 88k "Luna" boxes made by Omron) were not uncommon in 1992.
By no means was it perfect, and it was never really good at the
distributed-computing thing, but I think it would have provided *very*
stiff competition for Linux if it hadn't been embargoed.

Dave Eckhardt

Re: [9fans] [OT] linux origins, why not?

[David Leimbach]

On 3/29/06, quanstro@quanstro.net <quanstro@quanstro.net> wrote:
> do you have some pointers to papers from these guys?
>
> from my uneducated position, it seems to me that plan9 has a large
> percentage of what microkernels claim.  one thing one can't
> do is write a hardware driver that lives in userspace.  one advantage
> of this could be the ability to load drivers depending on configuration.
>
> has anybody invested some brain cells in this?

Userspace drivers?  You can load L4 Linux into userspace and use it's
drivers over L4 IPC calls and get reasonable performance.

Heck, you can even download a live ISO and just boot it that uses some
of this technology.  The FreeBSD IDE driver is pushed into userspace
and accessed by the rest of the L4 system.

Now if I could only find the link....

Dave

>
> - erik
>
> On Wed Mar 29 19:37:48 CST 2006, rminnich@lanl.gov wrote:
> > Lots of good research came out of mach ... not what you think. sandia
> > national labs has done lots of great OS work for 10 years, or so,
> > spurred on by the unusable Mach-derived OSF-1/MK-AD that came on their
> > paragon, and the need to toss it and start clean. SNL did some very nice
> > work, all due to the need to get rid of the "micro kernel".
> >
> > ron
>

Re: [9fans] [OT] linux origins, why not?

[David Leimbach]

On 3/29/06, Ronald G Minnich <rminnich@lanl.gov> wrote:
> quanstro@quanstro.net wrote:
>
> > mach was developed at cmu and freely available, wasn't it? the documentation
> > was (tree killers).
>
>
> best Mach phrase: "micro kernel doesn't mean it is small, just that it
> does not do much".
>
> from a flame war that erupted when the leviathan mach 3.0 came out.
>
> Well, it may have been big, but at least it was slow.
>
> Lots of good research came out of mach ... not what you think. sandia
> national labs has done lots of great OS work for 10 years, or so,
> spurred on by the unusable Mach-derived OSF-1/MK-AD that came on their
> paragon, and the need to toss it and start clean. SNL did some very nice
> work, all due to the need to get rid of the "micro kernel".
>

In case anyone was interested.  The madmen at UNSW are porting Darwin
(the mac os x unix portion that used to be freely available until the
intel macs came out) to L4

http://www.ertos.nicta.com.au/software/darbat/

I know Qualcomm also uses L4 in real production hardware now for
embedded systems.

You can't lump all microkernels together.  Mach was/is a really poor
microkernel compared to others of today's standards.  QNX has a much
better one as well.

Dave

> ron
>

Re: [9fans] [OT] linux origins, why not?

[Brantley Coile]

>> Before that I ran my port of V7 with streams
>> and my version of TCP running on Motorola 68K VME boards.
>
> You mean you kept a customised version of the kernel source
> locally?  Completely unacceptable.

Consensus has never been my strong point. :)

Re: [9fans] [OT] linux origins, why not?

[Richard Miller]

> Before that I ran my port of V7 with streams
> and my version of TCP running on Motorola 68K VME boards.

You mean you kept a customised version of the kernel source
locally?  Completely unacceptable.

Re: [9fans] [OT] linux origins, why not?

[Brantley Coile]

> Sure you would have--at worst you'd have been delayed a couple years.

I remember the 386 version of BSD about the same time as Linux was
almost, but not quite, usable.  I ran BSDI stuff from 1992 til Plan 9
was released in 1995.  Before that I ran my port of V7 with streams
and my version of TCP running on Motorola 68K VME boards.  I don't
think there was any waiting going on.

Re: [9fans] [OT] linux origins, why not?

[lucio]

> But it was the patch-hell of maintaining it that made Linux so
> attractive in those days.

I don't know, I went from an interest in Minix directly to 386BSD and
I have never had cause to look back.  Except that Plan 9 changes the
landscape somewhat.

++L

Re: [9fans] [OT] linux origins, why not?

[jmk]

On Wed Mar 29 20:52:09 EST 2006, quanstro@quanstro.net wrote:
> do you have some pointers to papers from these guys?
>
> from my uneducated position, it seems to me that plan9 has a large
> percentage of what microkernels claim.  one thing one can't
> do is write a hardware driver that lives in userspace.  one advantage
> of this could be the ability to load drivers depending on configuration.
>
> has anybody invested some brain cells in this?
>
> - erik

yes, brain cells have been burned on that and it has been done.
but we're waiting for uriel's replacement for 9load before the
details are finalised. after all, we don't want to waste our time
doing something that's already been done but just not released.

--jim

Re: [9fans] [OT] linux origins, why not?

[quanstro]

do you have some pointers to papers from these guys?

from my uneducated position, it seems to me that plan9 has a large
percentage of what microkernels claim.  one thing one can't
do is write a hardware driver that lives in userspace.  one advantage
of this could be the ability to load drivers depending on configuration.

has anybody invested some brain cells in this?

- erik

On Wed Mar 29 19:37:48 CST 2006, rminnich@lanl.gov wrote:
> Lots of good research came out of mach ... not what you think. sandia
> national labs has done lots of great OS work for 10 years, or so,
> spurred on by the unusable Mach-derived OSF-1/MK-AD that came on their
> paragon, and the need to toss it and start clean. SNL did some very nice
> work, all due to the need to get rid of the "micro kernel".
>
> ron

Re: [9fans] [OT] linux origins, why not?

[Roman Shaposhnick]

On Wed, Mar 29, 2006 at 08:40:55PM -0600, quanstro@quanstro.net wrote:
> do you have some pointers to papers from these guys?
>
> from my uneducated position, it seems to me that plan9 has a large
> percentage of what microkernels claim.

  Personally, one virtue of the founding fathers I admire the most
  is their ability to come up with really cool ideas and then always
  know where to stop. The sort of difference cherry on top makes,
  if you know what I mean:

     without a cherry on top: C,       Plan9,  Acme
     with a cherry on top:  C++, MACH/Spring, Emacs

  both lists go on.


Thanks,
Roman.

Re: [9fans] [OT] linux origins, why not?

[Martin C. Atkins]

Indeed - I used it alot. It was the best thing available.

But it was the patch-hell of maintaining it that made Linux so
attractive in those days.

(That, and the slow-downs caused by Minix's micro-kernel architecture)

Martin

On Thu, 30 Mar 2006 01:16:03 +0200 lucio@proxima.alt.za wrote:
> > " wanting to try out the protected mode features
> > of his spiffy new '386, which I assume that minix didn't support"
>
> Bruce Evans had a minix-386 running before I heard of Linus Torvalds.
>
> ++L
>


--
Martin C. Atkins			martin_ml@parvat.com
Parvat Infotech Private Limited		http://www.parvat.com{/,/martin}

Re: [9fans] [OT] linux origins, why not?

[Ronald G Minnich]

quanstro@quanstro.net wrote:
> do you have some pointers to papers from these guys?

go to snl.gov and look up PUMA. PUMA is as different from plan 9 as you
are going to get -- but it did its job well.

ron

Re: [9fans] [OT] linux origins, why not?

[Ronald G Minnich]

quanstro@quanstro.net wrote:

> mach was developed at cmu and freely available, wasn't it? the documentation
> was (tree killers).


best Mach phrase: "micro kernel doesn't mean it is small, just that it
does not do much".

from a flame war that erupted when the leviathan mach 3.0 came out.

Well, it may have been big, but at least it was slow.

Lots of good research came out of mach ... not what you think. sandia
national labs has done lots of great OS work for 10 years, or so,
spurred on by the unusable Mach-derived OSF-1/MK-AD that came on their
paragon, and the need to toss it and start clean. SNL did some very nice
work, all due to the need to get rid of the "micro kernel".

ron

Re: [9fans] [OT] linux origins, why not?

[quanstro]

i remember linux' existance giving quite a bit of pressure to the bsd projects.
it's hard to make convincing counterfactual arguments. waiting just a few
years is a really big deal when you're the one waiting.

i wore down the academic computing guys at the college i attended till
they signed the paperwork necessiary for a academic licence way back when.
the cd arrived the summer after i graduated.

i don't see how mach is an improvement over linux.  especially early linux.
mach kernels were three times the size of linux kernels of the day
and didn't do anything useful by themselves. sure they had threads, but
they also had 31 flavors of messages.

mach was developed at cmu and freely available, wasn't it? the documentation
was (tree killers).

- erik

On Wed Mar 29 17:28:38 CST 2006, davide+p9@cs.cmu.edu wrote:
> And if you *had* suffered that delay, you'd have got not only BSD
> Unix but also Mach (multiprocessor support, multi-threaded processes).
> Or maybe Plan 9 :-)
>
> Dave Eckhardt

Re: [9fans] [OT] linux origins, why not?

[Dave Eckhardt]

> Without linux I wouldn't have ever had a chance to really learn
> unix [...]

Sure you would have--at worst you'd have been delayed a couple years.

http://www.softpanorama.org/People/Torvalds/Finland_period/att_lawsuit_as_a_launcher_for_linux.shtml

And if you *had* suffered that delay, you'd have got not only BSD
Unix but also Mach (multiprocessor support, multi-threaded processes).
Or maybe Plan 9 :-)

Dave Eckhardt

Re: [9fans] [OT] linux origins, why not?

[lucio]

> " wanting to try out the protected mode features
> of his spiffy new '386, which I assume that minix didn't support"

Bruce Evans had a minix-386 running before I heard of Linus Torvalds.

++L

Re: [9fans] [OT] linux origins, why not?

[andrey mirtchovski]

On 3/29/06, Federico G. Benavento <benavento@gmail.com> wrote:
> http://liw.iki.fi/liw/texts/linux-anecdotes.html

"What is it that Linux has that no other operating system has? A
cuddly, lovable, silly-looking mascot."

Re: [9fans] [OT] linux origins, why not?

[Victor Nazarov]

Skip Tavakkolian wrote:

>what was the nature of this "terminal emulator"?  was it something
>like a vt100 terminal emulator or something like mux (dmd, 630)?
>
>
I think it was something like vt100, imho, linux's tty driver derives a
lot from Linus's work, but I don't feel competent to answer your
question, you can investigate source code to find some evidence that I'm
right or wrong.

Burton Samograd <kruhft@gmail.com> wrote

>Us kids (well, I'm 31 today, but I still feel like a kid around these
>circles) never got to grow up with unix and linux let us learn, so it
>still has a place near my heart, warts, cancerous growths and all  ;-)
>
>
Hmm, the most important part of my answer is my age. I'm 21. I'm just a
student.

>I don't want to reject it, it just sounds like a bit of a slam, since
>I generally don't hear much good about linux on lists such as this.
>
>
No slam was intended by me, just an interesting fact, isn't it?

>Without linux I wouldn't have ever had a chance to really learn unix,
>would have never had a reason to check out plan9 and even have an iota
>of understanding about why it's such a great evolution over the
>original unix.  I know linux isn't the best but it sure gets picked on
>a lot, especially around these groups.
>
I can confirm, that my path to the world of plan9 is the same. I still
like Debian, that I used a lot. I still don't like Windows that I use
and have to use a lot. I think, the feeling I experiance when see or
hear about GNU crap is something like sadness. Seems like they loose
that way. That way wich is embodied by Plan9.
--
Victor Nazarov

Re: [9fans] [OT] linux origins, why not?

[Federico G. Benavento]

[-- Attachment #1: Type: text/plain, Size: 50 bytes --]

http://liw.iki.fi/liw/texts/linux-anecdotes.html

[-- Attachment #2: Type: message/rfc822, Size: 4947 bytes --]

From: "Burton Samograd" <kruhft@gmail.com>
To: "Fans of the OS Plan 9 from Bell Labs" <9fans@cse.psu.edu>
Subject: Re: [9fans] [OT] linux origins, why not?
Date: Wed, 29 Mar 2006 12:04:44 -0600
Message-ID: <ace468790603291004y3c87009fx72d0b779df1a47df@mail.gmail.com>

> There is no offense for linux to start as a terminal emulator.
> And Linus confirms that. And this fact doesn't contradict with
> your point (" wanting to try out the protected mode features
> of his spiffy new '386, which I assume that minix didn't support").
> Why do you want to reject this terminal emulator origins?

I don't want to reject it, it just sounds like a bit of a slam, since
I generally don't hear much good about linux on lists such as this. 
Without linux I wouldn't have ever had a chance to really learn unix,
would have never had a reason to check out plan9 and even have an iota
of understanding about why it's such a great evolution over the
original unix.  I know linux isn't the best but it sure gets picked on
a lot, especially around these groups.

Us kids (well, I'm 31 today, but I still feel like a kid around these
circles) never got to grow up with unix and linux let us learn, so it
still has a place near my heart, warts, cancerous growths and all ;-)

--
burton samograd                                                kruhft@gmail.com
kruhft.blogspot.com   www.myspace.com/kruhft   metashell.blogspot.com

Re: [9fans] [OT] linux origins, why not?

[Skip Tavakkolian]

>>I've been hearing this "terminal emulator" story quite a bit lately
>>(over the past couple of months), and, althought I do dislike talking
>>part in such arguments, the story that I've always heard (circa '94)
>>was that linux was the result of wanting to try out the protected mode
>>features of his spiffy new '386, which I assume that minix didn't
>>support.
>>
>>
> There is no offense for linux to start as a terminal emulator.
> And Linus confirms that. And this fact doesn't contradict with
> your point (" wanting to try out the protected mode features
> of his spiffy new '386, which I assume that minix didn't support").
> Why do you want to reject this terminal emulator origins?

what was the nature of this "terminal emulator"?  was it something
like a vt100 terminal emulator or something like mux (dmd, 630)?

Re: [9fans] [OT] linux origins, why not?

[Burton Samograd]

> There is no offense for linux to start as a terminal emulator.
> And Linus confirms that. And this fact doesn't contradict with
> your point (" wanting to try out the protected mode features
> of his spiffy new '386, which I assume that minix didn't support").
> Why do you want to reject this terminal emulator origins?

I don't want to reject it, it just sounds like a bit of a slam, since
I generally don't hear much good about linux on lists such as this. 
Without linux I wouldn't have ever had a chance to really learn unix,
would have never had a reason to check out plan9 and even have an iota
of understanding about why it's such a great evolution over the
original unix.  I know linux isn't the best but it sure gets picked on
a lot, especially around these groups.

Us kids (well, I'm 31 today, but I still feel like a kid around these
circles) never got to grow up with unix and linux let us learn, so it
still has a place near my heart, warts, cancerous growths and all ;-)

--
burton samograd                                                kruhft@gmail.com
kruhft.blogspot.com   www.myspace.com/kruhft   metashell.blogspot.com

Re: [9fans] [OT] linux origins, why not?

[Victor Nazarov]

Burton Samograd wrote:

>I've been hearing this "terminal emulator" story quite a bit lately
>(over the past couple of months), and, althought I do dislike talking
>part in such arguments, the story that I've always heard (circa '94)
>was that linux was the result of wanting to try out the protected mode
>features of his spiffy new '386, which I assume that minix didn't
>support.
>
>
There is no offense for linux to start as a terminal emulator.
And Linus confirms that. And this fact doesn't contradict with
your point (" wanting to try out the protected mode features
of his spiffy new '386, which I assume that minix didn't support").
Why do you want to reject this terminal emulator origins?
--
Victor Nazarov