[9fans] vmware 5.0

[9fans] vmware 5.0

[Tim Newsham]

I just tried loading up vmware 5 and booting a plan9 in vmware.
Ugh, the screen doesnt update properly.  I see the cursor and
some of the window borders and scroll bars.  The main parts of
the window never seem to refresh.

Has anyone played with this much?  Are there any known work-arounds?

Tim Newsham
http://www.lava.net/~newsham/

Re: [9fans] vmware 5.0

[Russ Cox]

> Has anyone played with this much?  
> Are there any known work-arounds?

there are no known work-arounds.
i asked vmware for documentation, was promised it,
didn't get it, tried to follow up once, and then
dropped it.  i don't care enough.  qemu with
the kqemu patch on linux is plenty fast for me.
even plain qemu is pretty fast.

russ

Re: [9fans] vmware 5.0

[Ronald G Minnich]

Russ Cox wrote:
> qemu with
> the kqemu patch on linux is plenty fast for me.
> even plain qemu is pretty fast.


yup. I'm done with vmware as well. When I can't use xen, I use qemu, and 
it's just fine.

Plus, you can make qemu better, and you can't make vmware better. vmware 
just does not care about Plan 9, for good reasons.

I just don't see any point in vmware any more for Plan 9 users.

ron

Re: [9fans] vmware 5.0

[Ronald G Minnich]

I should probably mention here that once the VT and Pacifica become 
widely available, you'll be able to boot unmodified plan 9 as a xen 
guest ...

ron

Re: [9fans] vmware 5.0

[Gorka guardiola]

What about windows?. Qemu is flaky on windows yet and xen doesnt exist...

On 8/23/05, Ronald G Minnich <rminnich@lanl.gov> wrote:
> Russ Cox wrote:
> I just don't see any point in vmware any more for Plan 9 users.


-- 
- curiosity sKilled the cat

Re: [9fans] vmware 5.0

[Ronald G Minnich]

Gorka guardiola wrote:
> What about windows?. Qemu is flaky on windows yet and xen doesnt exist...

lucky me! I keep forgetting about windows, except when it does something 
like take the entire LANL network down ...
http://lanl-the-real-story.blogspot.com/2005/08/major-disruption-in-internet-service.html

Still, I think you're better off putting effort into helping qemu than 
trying to make vmware work, because vmware (for sound business reasons, 
I"m sure) is not going to help plan 9.

ron

Re: [9fans] vmware 5.0

[Steve Simon]

> What about windows?. Qemu is flaky on windows yet and xen doesnt exist...

Sadly its worse than that, a port of WinXP was done to xen
but licensing restrictions mean it cannot be released.

-Steve

Re: [9fans] vmware 5.0

[Devon H. O'Dell]

On Tue, Aug 23, 2005 at 07:55:30PM +0100, Steve Simon wrote:
> > What about windows?. Qemu is flaky on windows yet and xen doesnt exist...
> 
> Sadly its worse than that, a port of WinXP was done to xen
> but licensing restrictions mean it cannot be released.
> 
> -Steve

It's a big pity since microsoft obviously has a vested interest
in Xen (google for xen site:research.microsoft.com yields
interesting results), but it's really improbable that they'd
ever release a patch for paid licenses or sell packages for it.
My guess is that they are simply interested for a VMWare ESX
competitor in VirtualPC.

--Devon

Re: [9fans] vmware 5.0

[Gorka guardiola]

I meant as a host. I believe you are talking about running windows
inside xen. I was
talking about running xen inside windows (with plan 9...). 

On 8/23/05, Steve Simon <steve@quintile.net> wrote:
> > What about windows?. Qemu is flaky on windows yet and xen doesnt exist...
> 
> Sadly its worse than that, a port of WinXP was done to xen
> but licensing restrictions mean it cannot be released.
> 
> -Steve
> 


-- 
- curiosity sKilled the cat

Re: [9fans] vmware 5.0

[Scott Stout]

"A data type extension for XML and table manipulation (formerly known
as Xen and as X#):"

-http://research.microsoft.com/Comega/

scott

On 8/23/05, Devon H. O'Dell <dodell@offmyserver.com> wrote:
> On Tue, Aug 23, 2005 at 07:55:30PM +0100, Steve Simon wrote:
> > > What about windows?. Qemu is flaky on windows yet and xen doesnt exist...
> >
> > Sadly its worse than that, a port of WinXP was done to xen
> > but licensing restrictions mean it cannot be released.
> >
> > -Steve
> 
> It's a big pity since microsoft obviously has a vested interest
> in Xen (google for xen site:research.microsoft.com yields
> interesting results), but it's really improbable that they'd
> ever release a patch for paid licenses or sell packages for it.
> My guess is that they are simply interested for a VMWare ESX
> competitor in VirtualPC.
> 
> --Devon
>

Re: [9fans] vmware 5.0

[Devon H. O'Dell]

On Tue, Aug 23, 2005 at 02:28:59PM -0500, Scott Stout wrote:
> "A data type extension for XML and table manipulation (formerly known
> as Xen and as X#):"
> 
> -http://research.microsoft.com/Comega/
> 
> scott

Did you read any other links?

research.microsoft.com/~tharris/
research.microsoft.com/research/sv/vexedd/
research.microsoft.com/~helenw/papers/vground.pdf

These are before and after the link you pasted, respectively,
in my google search results.

-Devon
 
> On 8/23/05, Devon H. O'Dell <dodell@offmyserver.com> wrote:
> > On Tue, Aug 23, 2005 at 07:55:30PM +0100, Steve Simon wrote:
> > > > What about windows?. Qemu is flaky on windows yet and xen doesnt exist...
> > >
> > > Sadly its worse than that, a port of WinXP was done to xen
> > > but licensing restrictions mean it cannot be released.
> > >
> > > -Steve
> > 
> > It's a big pity since microsoft obviously has a vested interest
> > in Xen (google for xen site:research.microsoft.com yields
> > interesting results), but it's really improbable that they'd
> > ever release a patch for paid licenses or sell packages for it.
> > My guess is that they are simply interested for a VMWare ESX
> > competitor in VirtualPC.
> > 
> > --Devon
> >

Re: [9fans] vmware 5.0

[Latchesar Ionkov]

I don't think you can run xen inside Windows. Xen is always in control of
the hardware, it can let some of the domains to access some of it directly,
but the MMU and the interrupts are always handled by Xen. 

	Lucho

On Tue, Aug 23, 2005 at 02:26:27PM -0500, Gorka guardiola said:
> I meant as a host. I believe you are talking about running windows
> inside xen. I was
> talking about running xen inside windows (with plan 9...). 
> 
> On 8/23/05, Steve Simon <steve@quintile.net> wrote:
> > > What about windows?. Qemu is flaky on windows yet and xen doesnt exist...
> > 
> > Sadly its worse than that, a port of WinXP was done to xen
> > but licensing restrictions mean it cannot be released.
> > 
> > -Steve
> > 
> 
> 
> -- 
> - curiosity sKilled the cat

[9fans] procreate vs threadcreate?

[rattan]

I need some help in understnading the difference
between proccreate() and threadcreate(). What
are the scenario where usage of one is better than
the other?

-ishwar

Re: [9fans] procreate vs threadcreate?

[Brantley Coile]

Both share address space, but proccreate creates another process.
This process can block on doing IO or some other system call that will
block the progress of the process.  Threads are co-routines in a
single process.  None of them will run if the process blocks on, say,
a read.  So, proccreate processes to do the IO and have them use
channels to send messages to worker threads.

Sape has a good paper on all this, as is Rob's paper on Newsqueak.

> I need some help in understnading the difference
> between proccreate() and threadcreate(). What
> are the scenario where usage of one is better than
> the other?
> 
> -ishwar

Re: [9fans] procreate vs threadcreate?

[andrey mirtchovski]

> I need some help in understnading the difference
> between proccreate() and threadcreate(). What
> are the scenario where usage of one is better than
> the other?
> 
> -ishwar

see section 4 of:

	http://www.cs.unibo.it/ersads/tutorials/mullender.ps

Re: [9fans] procreate vs threadcreate?

[Russ Cox]

> I need some help in understnading the difference
> between proccreate() and threadcreate(). What
> are the scenario where usage of one is better than
> the other?

see http://plan9.bell-labs.com/~rsc/thread/ and
read the introduction to alef by bob flandrena
linked there.

in terms of trite academic examples,

int x;
void inc(void*) { x=x+1; }
void
main(void)
{
    int i;
    for(i=0; i<10; i++) 
        threadcreate(inc, nil, STACK);
    while(x < 10)
        yield();
    print("everyone finished\n");
}

is correct (though tortuous) code.
using proccreate would introduce
the usual race.

russ

Re: [9fans] vmware 5.0

[Gorka guardiola]

And windows.

On 8/23/05, Ronald G Minnich <rminnich@lanl.gov> wrote:
> I should probably mention here that once the VT and Pacifica become
> widely available, you'll be able to boot unmodified plan 9 as a xen
> guest ...
> 
> ron
> 


-- 
- curiosity sKilled the cat

Re: [9fans] vmware 5.0

[Francisco Ballesteros]

seems that we should probably switch to linux VMs as well.
At least, we wont have to program with damn ms apis.

On 8/23/05, Gorka guardiola <paurea@gmail.com> wrote:
> And windows.
> 
> On 8/23/05, Ronald G Minnich <rminnich@lanl.gov> wrote:
> > I should probably mention here that once the VT and Pacifica become
> > widely available, you'll be able to boot unmodified plan 9 as a xen
> > guest ...
> >
> > ron
> >
> 
> 
> --
> - curiosity sKilled the cat
>

Re: [9fans] vmware 5.0

[LiteStar numnums]

What, you don't like Hungarian notation, multiple calling methods,
bizzare & inconsitent function names and overly bloated libraries?
Sheesh, what do you think this is {Plan9 | Unix | OpenVMS}?

On 8/24/05, Francisco Ballesteros <capt.nemo@gmail.com> wrote:
> seems that we should probably switch to linux VMs as well.
> At least, we wont have to program with damn ms apis.
> 
> On 8/23/05, Gorka guardiola <paurea@gmail.com> wrote:
> > And windows.
> >
> > On 8/23/05, Ronald G Minnich <rminnich@lanl.gov> wrote:
> > > I should probably mention here that once the VT and Pacifica become
> > > widely available, you'll be able to boot unmodified plan 9 as a xen
> > > guest ...
> > >
> > > ron
> > >
> >
> >
> > --
> > - curiosity sKilled the cat
> >
> 


-- 
The subject of this essay (the Myth of Sisyphus) is precisely
this relationship between the absurd and suicide, the exact
degree to which suicide is a solution to the absurd. The 
principle can be established that for a man who does not cheat,
what he believes to be true must determine his action.
Belief in the absurdity of existence must then dictate his
conduct. It is legitimate to wonder, clearly and without
false pathos, whether a conclusion of this importance
requires forsaking as rapidly possiblean imcompre-
hensible condition. I am speaking, of course, of men
inclined to be in harmony with themselves.
  << Albert Camus>>

Re: [9fans] vmware 5.0

[Martin C. Atkins]

On Tue, 23 Aug 2005 15:48:56 -0400 Latchesar Ionkov <lucho@gmx.net> wrote:
> I don't think you can run xen inside Windows. Xen is always in control of
> the hardware, it can let some of the domains to access some of it directly,
> but the MMU and the interrupts are always handled by Xen. 

I've been wondering if the techniques that colinux uses to run Linux
inside windows could also be used to make Xen in Windows possible.

However, I have no real details about what exactly those techniques are,
so this is pure speculation!

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

Re: [9fans] vmware 5.0

[Tim Newsham]

> I've been wondering if the techniques that colinux uses to run Linux
> inside windows could also be used to make Xen in Windows possible.

I'm sure you could take the xen dom0<->domU interface and use
it to allow domU's to be run on top of win32 (or the NT kernel, or
any other kernel, for that matter).  If you didn't want to interpret
the cpu instructions you would need a little hook in the native kernel
to catch trap instructions and handle page faults.  If you wanted to
emulate the cpu (ie. bochs/flex86/qemu) you could directly provide
a xen "machine."  Definitely not a trivial undertaking, though.

> Martin C. Atkins			martin_ml@parvat.com

Tim Newsham
http://www.lava.net/~newsham/

Re: [9fans] vmware 5.0

[Vester Thacker]

On 8/25/05, Martin C. Atkins <martin_ml@parvat.com> wrote:
> On Tue, 23 Aug 2005 15:48:56 -0400 Latchesar Ionkov <lucho@gmx.net> wrote:
> > I don't think you can run xen inside Windows. Xen is always in control of
> > the hardware, it can let some of the domains to access some of it directly,
> > but the MMU and the interrupts are always handled by Xen.
> 
> I've been wondering if the techniques that colinux uses to run Linux
> inside windows could also be used to make Xen in Windows possible.

Jun Okamoto is a colinux developer and a fellow Plan 9er, I'll be sure
to mention your comments to him at the next tip9ug meeting. Jun can be
reached at: okajima@digitalinfra.co.jp.

More about Jun at: http://www.colinux.org/?section=devteam
Btw Jun, also, owns plan9.jp domain. :-)

--vester

Re: [9fans] vmware 5.0

[Vester Thacker]

On 8/27/05, Vester Thacker <vester.thacker@gmail.com> wrote:
> 
> Jun Okamoto is a colinux developer and a fellow Plan 9er, I'll be sure
> to mention your comments to him at the next tip9ug meeting. Jun can be

Oops, that's Jun Okajima not Okamoto. *sigh*   Sorry, Jun.

--vester

Re: [9fans] vmware 5.0

[Jun OKAJIMA]

I think it is possible to run Dom0 of Xen on the top of Windows, without VT.
The biggest issue would be not technical one, but just who will do it.

So, we are welcome if you would port Xen to Windows using "Cooperative 
Virtual Machine" technology, which Dan Aloni has invented. But dont expect
we will do it,  especially under the situation that we realize that once VT
becomes major, CVM becomes obsolete. And paravirtualization, which is the key
point of Xen in  the technological aspect also would become obsolete, although
Xen has hopeful future in business aspect.


                    --- Okajima, Jun. Tokyo, Japan.


>On 8/25/05, Martin C. Atkins <martin_ml@parvat.com> wrote:
>> On Tue, 23 Aug 2005 15:48:56 -0400 Latchesar Ionkov <lucho@gmx.net> wrote:
>> > I don't think you can run xen inside Windows. Xen is always in control of
>> > the hardware, it can let some of the domains to access some of it directly,
>> > but the MMU and the interrupts are always handled by Xen.
>> 
>> I've been wondering if the techniques that colinux uses to run Linux
>> inside windows could also be used to make Xen in Windows possible.
>
>Jun Okamoto is a colinux developer and a fellow Plan 9er, I'll be sure
>to mention your comments to him at the next tip9ug meeting. Jun can be
>reached at: okajima@digitalinfra.co.jp.
>
>More about Jun at: http://www.colinux.org/?section=devteam
>Btw Jun, also, owns plan9.jp domain. :-)
>
>--vester
>

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Jun OKAJIMA]

And another way is, you make coPlan9, which
uses same CVM way on Plan9. This is a simpler way,
if you just want to run only Plan9.


                    --- Okajima.

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[kokamoto]

> if you just want to run only Plan9.

Actually, I never understand such kind of attempts.
What is the merit to use mutiple OSs on a machine?
>From a bad humor sense, I can realise it only for saving
power...

Once, I thought it'd be nice if I could use Xen for Linux and Plan 9
web server to use UTF-8 encoded our page.   However, I realised
it not so essential after that.  Now, I'm thinking it I need Windows,
lets have a machine for it, if I need Plan 9 lets have three machine
for it, etc.   Machines are not expensive these days, son that must
be only for saving power for mother earth.☺

Kenji

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Steve Simon]

Re: forign OS servers

I have never tried this but I have read that one can
run Windows terminal server and terminal client on a
WinXP machine. If you then run vnc server on it you can
have multiple users connecting with VNC, each running a
seperate Windows GUI sessions - IE a single XP machine
can become an IE6 server.

I tried to get Linux and Xvnc to cooperate with plan9
but never got the cut and paste buffers to synchronise
properly, perhaps things have changed.

-Steve

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[kokamoto]

Last night I must have been too sleepy.   I can't read my own
posting this morning.☺  Then, I'll try once more.

> if you just want to run only Plan9.

Actually, I never understand such kind of attempts.
What is the merit to use mutiple OSs on a machine?
>From a bad humor sense, I can realize it only for saving
power...

Once, I thought it'd be nice if I could use Xen for Linux and Plan 9
web server to use UTF-8 encoded our page.   However, I realized
it not so essential after that.  Now, I'm thinking like this: if I need 
Windows, let's have a machine for it. If I need Plan 9 let's have 
three machine for it, etc.   Machines are cheeper these days, 
so it must be only for saving power for mother earth.☺

Better?

I'm not offending the one machine model for Plan 9 from the
view point of more convenience to more people.   However, I think 
Geoff's effort should be payed more attention by more Plan 9ers.

Kenji

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Brantley Coile]

[-- Attachment #1: Type: text/plain, Size: 902 bytes --]

i too am both curious as to the motivations for VM and completely open
minded with no preconceived notions about VM. except my aversion to
hype.  but hype is independent from the quality of an idea.

i was asking Friday here at work, what are the modivations behind VM?
the only answers that were offered were variations on the ability to
rent someone a machine that has root access without having as many
machines are renters.  the earliest VM i know of is VM/CMS, from IBM,
which is still used today.  its purpose was to provide early
timesharing, and was also used to debug MVS.  so those are two
motivation, although Xen can't be used for debugging OSes since it's a
paravirtual machine.  i don't think VMware would be too good either
because it rewrites parts of your code.  maybe that's not a problem in
practice.

maybe Ron can give us insight into the motivations for using VM.

[-- Attachment #2: Type: message/rfc822, Size: 3484 bytes --]

From: kokamoto@hera.eonet.ne.jp
To: 9fans@cse.psu.edu
Subject: Re: [9fans] Xen for Windows(Was:vmware 5.0)
Date: Sun, 28 Aug 2005 10:22:55 +0900
Message-ID: <6a65a8b751540b784b8cbff84466ad36@hera.eonet.ne.jp>

Last night I must have been too sleepy.   I can't read my own
posting this morning.☺  Then, I'll try once more.

> if you just want to run only Plan9.

Actually, I never understand such kind of attempts.
What is the merit to use mutiple OSs on a machine?
>From a bad humor sense, I can realize it only for saving
power...

Once, I thought it'd be nice if I could use Xen for Linux and Plan 9
web server to use UTF-8 encoded our page.   However, I realized
it not so essential after that.  Now, I'm thinking like this: if I need 
Windows, let's have a machine for it. If I need Plan 9 let's have 
three machine for it, etc.   Machines are cheeper these days, 
so it must be only for saving power for mother earth.☺

Better?

I'm not offending the one machine model for Plan 9 from the
view point of more convenience to more people.   However, I think 
Geoff's effort should be payed more attention by more Plan 9ers.

Ken ji

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[lucio]

> i too am both curious as to the motivations for VM and completely open
> minded with no preconceived notions about VM. except my aversion to
> hype.  but hype is independent from the quality of an idea.

I understood that IBM's motive for VM was to allow different OSes to
co-exist on a single hardware platform.  That the same OS also managed
to co-exist with multiple instances of itself was an added bonus and
greatly simplified the multitasking process.

Basically, there were radical incompatibilies between successive OS
releases from IBM (you may recall those days as being a quick
succession of software discoveries/inventions, unmatched by recent
developments) and IBM could not compel users to rewrite their
applications, no matter how exciting the new platform.  In fact, I
suspect IBM themselves made good use of the backwards compatibility
they provided with VM.

By providing VM capabilities _at_the_hardware_level_ IBM could entice
customers to upgrade and thus appreciate the improvements in the newer
equipment.  The enormous investment of man power as well as
intellectual effort involved in producing custom applications made
this critical.

We live in a different world, today, with disposable equipment on
every desk.  But there is a price tag and it's not just higher
electricity bills.  For example, disk warranties.  Ten years ago you
could buy a 2 Gig drive with a lifetime guarantee (rough guess, I
can't recall very accurately that far back), now you _expect_ you 320
Gig drive to pack up on you within two years.  What happens to the 320
Gig of data you entrusted to the drive, then?

I would much rather have a single, reliable computer with the
essential instruction set (I'm not exclusively sold on RISC, but I do
think it makes a lot more sense) and replaceable peripherals than
invest in the latest, greatest Wintel box with the fanciest, most
irrelevant multimedia instruction set in the dual-core CPU and a
Winmodem for my Internet connectivity :-(

But ranting isn't going to help.

++L

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Martin Harriss]

See the following paper (by a colleague) for more than you wanted to 
know about the development of VM.  It was, in the beginning, a skunk 
works project.

http://pucc.Princeton.EDU/~melinda/25paper.pdf

Martin


Brantley Coile wrote:
> i too am both curious as to the motivations for VM and completely open
> minded with no preconceived notions about VM. except my aversion to
> hype.  but hype is independent from the quality of an idea.
> 
> i was asking Friday here at work, what are the modivations behind VM?
> the only answers that were offered were variations on the ability to
> rent someone a machine that has root access without having as many
> machines are renters.  the earliest VM i know of is VM/CMS, from IBM,
> which is still used today.  its purpose was to provide early
> timesharing, and was also used to debug MVS.  so those are two
> motivation, although Xen can't be used for debugging OSes since it's a
> paravirtual machine.  i don't think VMware would be too good either
> because it rewrites parts of your code.  maybe that's not a problem in
> practice.
> 
> maybe Ron can give us insight into the motivations for using VM.

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Ronald G Minnich]

kokamoto@hera.eonet.ne.jp wrote:

> Actually, I never understand such kind of attempts.
> What is the merit to use mutiple OSs on a machine?

I do all me development on my laptop, and sometimes need to test 
linux/plan9 working together.

ron

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Ronald G Minnich]

Brantley Coile wrote:

> i was asking Friday here at work, what are the modivations behind VM?

I forgot the other one. Drivers, drivers, drivers. I can run Plan 9 on 
machines that I don't have time/brains to write drivers for. This is 
particularly interesting with a 1024-node cluster here called Pink; I 
can run Plan under Linux on 1024 nodes and do some kind of scaling 
tests. I can also run 10,240 instances of Plan9 on that machine (VM, 
right?) and at least see what kinds of things break when you have a 
10240 node Plan 9 cluster. Of course, it's not going to run at speed, 
but you can still learn a thing or two.

The only interconnect on Pink is Myrinet. Maybe I should write Myrinet 
drivers for Plan 9, but even if I did ...

There's infiniband. I don't want to write IB drivers for Plan 9, at 
least not at present. They're very complex and unsettled.

So, the basic idea is that VM lets you run OSes on machines that 
otherwise it would be very hard to get to, and use many more processors 
than you have in reality.

Also, booting Plan 9 in a second is nice when you're in developer mode. 
Skipping 9load is a good thing, all the way around. Kernel crashes are 
painless.

True story: I went to IBM Palo Alto in 1990 to talk about various 
supercomputing things and the issue of AIX/370 came up. AIX had always 
run under VM to that point. There was really some question about whether 
any living person knew enough about the IBM I/O channels to make AIX 
native. VM knew the tricks; did any human know the tricks? Nobody knew. 
IBM had kind of screwed themselves on this score, as VM went 
closed-source in 1982, and the entire external VM community no longer 
knew enough to help.

[[Another argument for open source: your company might forget how your 
own software works, but the larger community might remember. This type 
of forgetfulness happens way more often than you might think. ]]

To this day, at least Linux is not native, or so I understand; Linux on 
the zSeries always runs under VM. Again, feature: IBM has shown cases 
where 7,000 or more instances of Linux can be running on a zSeries 
machine small enough to fit in your kitchen -- air-cooled at that. No 
need to buy rackfuls of machines in that case!

ron

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Ronald G Minnich]

another true story. IBM really knows emulation. At E.I. Dupont, where as 
a teenager I was a computer operator, we had (as of 1975) the last 
vacuum tube computer in the US. It was a 705 (NOT 704). It ran payroll.

Oops, vacuum tubes were getting hard to get. So Dupont got a 7080 (NOT 
7090). It emulated a 705.

Oops, 7080 was expensive to maintain. So Dupont got a 360, emulates 
7080, emulating 705.

Oh, heck, 360 went off support. So, on our 370/158, we emulated 360, 
emulating 7080, emulating 705. This was really emulation! virtual card 
punches, virtual card readers!

Then they rewrote the payroll system to run native. "I'm sure there are 
some bugs in there, but I can't find them," the programmer told me, on 
saturday at about 2 am. "We're going live next week".

2 weeks later, all us weekend dudes got a 4x-larger paycheck than we 
were supposed to.

Geez, I hope nobody from Dupont comes after me for telling this story!

ron

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Ronald G Minnich]

Martin Harriss wrote:
> See the following paper (by a colleague) for more than you wanted to 
> know about the development of VM.  It was, in the beginning, a skunk 
> works project.
> 
> http://pucc.Princeton.EDU/~melinda/25paper.pdf


Great paper, also details the "go closed source" decision in 1982, which 
I doubt IBM would do today.

ron

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[kokamoto]

[-- Attachment #1: Type: text/plain, Size: 440 bytes --]

I'm not an expert of this.  However, I think I know the motivation
of IBM's VM attempt.   Because they decided their main job as
"ON DEMAND" business.   To do this, they need such a frexible
mechanism to offer various on demand, on line business.

However, this is Plan 9 community.   Is their anyone trying to
make on demand business? ☺

By the way, anyone Plan 9ers here is invlolved in the development 
of DeepMail?

Kenji

[-- Attachment #2: Type: message/rfc822, Size: 3527 bytes --]

From: Ronald G Minnich <rminnich@lanl.gov>
To: Fans of the OS Plan 9 from Bell Labs <9fans@cse.psu.edu>
Subject: Re: [9fans] Xen for Windows(Was:vmware 5.0)
Date: Sun, 28 Aug 2005 08:25:25 -0600
Message-ID: <4311C955.8040509@lanl.gov>

Martin Harriss wrote:
> See the following paper (by a colleague) for more than you wanted to 
> know about the development of VM.  It was, in the beginning, a skunk 
> works project.
> 
> http://pucc.Princeton.EDU/~melinda/25paper.pdf


Great paper, also details the "go closed source" decision in 1982, which 
I doubt IBM would do today.

ron

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Ronald G Minnich]

kokamoto@hera.eonet.ne.jp wrote:

> However, this is Plan 9 community.   Is their anyone trying to
> make on demand business? ☺


I'd just like somebody to continue to pay for Plan 9 work :-)

ron

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Latchesar Ionkov]

Another good thing about VMs is that you can migrate them to another server
if you want to service the one that they are running on...

Thanks,
	Lucho

On Sun, Aug 28, 2005 at 08:04:34AM -0400, Brantley Coile said:
> i too am both curious as to the motivations for VM and completely open
> minded with no preconceived notions about VM. except my aversion to
> hype.  but hype is independent from the quality of an idea.
> 
> i was asking Friday here at work, what are the modivations behind VM?
> the only answers that were offered were variations on the ability to
> rent someone a machine that has root access without having as many
> machines are renters.  the earliest VM i know of is VM/CMS, from IBM,
> which is still used today.  its purpose was to provide early
> timesharing, and was also used to debug MVS.  so those are two
> motivation, although Xen can't be used for debugging OSes since it's a
> paravirtual machine.  i don't think VMware would be too good either
> because it rewrites parts of your code.  maybe that's not a problem in
> practice.
> 
> maybe Ron can give us insight into the motivations for using VM.

> To: 9fans@cse.psu.edu
> Subject: Re: [9fans] Xen for Windows(Was:vmware 5.0)
> Date: Sun, 28 Aug 2005 10:22:55 +0900
> From: kokamoto@hera.eonet.ne.jp
> 
> Last night I must have been too sleepy.   I can't read my own
> posting this morning.☺  Then, I'll try once more.
> 
> > if you just want to run only Plan9.
> 
> Actually, I never understand such kind of attempts.
> What is the merit to use mutiple OSs on a machine?
> >From a bad humor sense, I can realize it only for saving
> power...
> 
> Once, I thought it'd be nice if I could use Xen for Linux and Plan 9
> web server to use UTF-8 encoded our page.   However, I realized
> it not so essential after that.  Now, I'm thinking like this: if I need 
> Windows, let's have a machine for it. If I need Plan 9 let's have 
> three machine for it, etc.   Machines are cheeper these days, 
> so it must be only for saving power for mother earth.☺
> 
> Better?
> 
> I'm not offending the one machine model for Plan 9 from the
> view point of more convenience to more people.   However, I think 
> Geoff's effort should be payed more attention by more Plan 9ers.
> 
> Ken ji

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Brantley Coile]

i guess this would be another rent-a-computer app.  if i want
to bring up another machine with other services i just do it.

On Aug 28, 2005, at 2:18 PM, Latchesar Ionkov wrote:

> Another good thing about VMs is that you can migrate them to another 
> server
> if you want to service the one that they are running on...
>
> Thanks,
> 	Lucho

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Gorka guardiola]

On 8/28/05, Brantley Coile <brantley@coraid.com> wrote:
> i too am both curious as to the motivations for VM and completely open
> minded with no preconceived notions about VM. except my aversion to
> hype.  but hype is independent from the quality of an idea.
> 
> i was asking Friday here at work, what are the modivations behind VM?

Mine's are

a) not writing drivers or writing easier ones.
b) developing/debugging a kernel while reading my email on my laptop
(I dont carry
various laptops around)
c) running different OS for compatibility (e.j. Word) on my laptop
without having
to reboot my machine.

-- 
- curiosity sKilled the cat

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[McLone]

On 8/28/05, Brantley Coile <brantley@coraid.com> wrote:
> i was asking Friday here at work, what are the modivations behind VM?
i need to run multiple OSes, more than 4 (primary PC partitions limit),
on a single box (which i _finally_ bought, being a poor student).

But not only this - i am unable to write drivers for some hardware
because of closed specifications (nVidia, various WiFi, Adaptecs etc).

Besides, i dislike toying with kernels in silence, and iPod is pricey.
-- 
wbr,                        |\      _,,,---,,_           dog bless ya!
`                       Zzz /,`.-'`'    -.  ;-;;,_
McLone at GMail dotcom     |,4-  ) )-,_. ,\ (  `'-'
  net- and *BSD admin     '---''(_/--'  `-'\_)

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Martin C. Atkins]

On Sun, 28 Aug 2005 08:04:34 -0400 Brantley Coile <brantley@coraid.com> wrote:
> i too am both curious as to the motivations for VM and completely open

My reasons:
1) I hate mucking with partitions - too much chance of zapping something important
by mistake, and they *always* end up being the wrong size, regardless of how much
planning I do... With a VM, the filesystems can be in normal files, which doesn't
completely avoid the problem, but removes the day-to-day need to re-partition.

2) Machines are cheap, but (desk/office/etc) space is not. Virtual machines
don't take up real space!

3) 2 in another guise - I don't want to carry n>1 laptops!

4) I can move virtual machines between physical machines just by copying
the filesystems.

5) Compatibility with OSs I don't want to run all the time (without rebooting)

All the other things people have mentioned...

Martin

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

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Brantley Coile]

[-- Attachment #1: Type: text/plain, Size: 651 bytes --]

these are good reasons and bring to mind a question.  why don't we
define a generic microkernel that all these systems can use and not
have to go to all the tricks of VM?

i now that the usual definition for a microkernel is the part that
supports all the processes that really make up the OS. but if we had a
generic microkernel then people could write device drivers for that an
all the other OSs could use that.

i'm not holding my breath on the above, so next question.  is the structure
of windows device driver environments well known enough to build a
small `container' to use off-the-shelf windows drivers on plan 9?

  Brantley

[-- Attachment #2: Type: message/rfc822, Size: 3719 bytes --]

From: "Martin C. Atkins" <martin_ml@parvat.com>
To: Fans of the OS Plan 9 from Bell Labs <9fans@cse.psu.edu>
Subject: Re: [9fans] Xen for Windows(Was:vmware 5.0)
Date: Wed, 31 Aug 2005 18:52:10 +0100
Message-ID: <20050831185210.1218a746@localhost.localdomain>

On Sun, 28 Aug 2005 08:04:34 -0400 Brantley Coile <brantley@coraid.com> wrote:
> i too am both curious as to the motivations for VM and completely open

My reasons:
1) I hate mucking with partitions - too much chance of zapping something important
by mistake, and they *always* end up being the wrong size, regardless of how much
planning I do... With a VM, the filesystems can be in normal files, which doesn't
completely avoid the problem, but removes the day-to-day need to re-partition.

2) Machines are cheap, but (desk/office/etc) space is not. Virtual machines
don't take up real space!

3) 2 in another guise - I don't want to carry n>1 laptops!

4) I can move virtual machines between physical machines just by copying
the filesystems.

5) Compatibility with OSs I don't want to run all the time (without rebooting)

All the other things people have mentioned...

Martin

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

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Tim Newsham]

> these are good reasons and bring to mind a question.  why don't we
> define a generic microkernel that all these systems can use and not
> have to go to all the tricks of VM?

Mach?  OSKit?  Xen?

>  Brantley

Tim Newsham
http://www.lava.net/~newsham/

[9fans] ndis-ulator for plan 9

[Ronald G Minnich]

McLone wrote:

> But not only this - i am unable to write drivers for some hardware
> because of closed specifications (nVidia, various WiFi, Adaptecs etc).

speaking of this:

if anybody wanted to port the bsd ndis-ulator to plan 9, or use it as a 
model for a Plan 9 driver, it would be a HUGE service, and would get us 
tons of drivers.

ron

Re: [9fans] ndis-ulator for plan 9

[Dave Lukes]

Ron,
I'm going to regret this, but I'm gonna trust that you're right and go 
take a look at it
(unless someone else steps up to the plate/wicket).

I'm not going to even announce it, let alone discuss it,
I'm going to take a look at how much work is involved and either start 
doing it or not.

If I even mention I'm doing it, there will be a flood of "windoze 
sukz", "is it GPLed?"
and "driver X won't work so it ain't worth doing" messages to distract 
me.

Do you happen to know anything about the beast to get me started?

Cheers,
     Dave.


On 1 Sep 2005, at 04:59, Ronald G Minnich wrote:

> McLone wrote:
>
>> But not only this - i am unable to write drivers for some hardware
>> because of closed specifications (nVidia, various WiFi, Adaptecs etc).
>
> speaking of this:
>
> if anybody wanted to port the bsd ndis-ulator to plan 9, or use it as 
> a model for a Plan 9 driver, it would be a HUGE service, and would get 
> us tons of drivers.
>
> ron

Re: [9fans] ndis-ulator for plan 9

[Devon H. O'Dell]

On Thu, Sep 01, 2005 at 08:31:49AM +0100, Dave Lukes wrote:
> Ron,
> I'm going to regret this, but I'm gonna trust that you're right and go 
> take a look at it
> (unless someone else steps up to the plate/wicket).
> 
> I'm not going to even announce it, let alone discuss it,
> I'm going to take a look at how much work is involved and either start 
> doing it or not.
> 
> If I even mention I'm doing it, there will be a flood of "windoze 
> sukz", "is it GPLed?"
> and "driver X won't work so it ain't worth doing" messages to distract 
> me.
> 
> Do you happen to know anything about the beast to get me started?
> 
> Cheers,
>     Dave.

Yes. Both Linux and Solaris have used Bill Paul's FreeBSD
work to get `Project Evil' running in these operating systems.
Take a look around at:

/sys/dev/if_ndis/*
/sys/compat/ndis/*
/usr/src/usr.sbin/ndiscvt/*

in the FreeBSD source repository for more info :)

--Devon
 
> 
> On 1 Sep 2005, at 04:59, Ronald G Minnich wrote:
> 
> >McLone wrote:
> >
> >>But not only this - i am unable to write drivers for some hardware
> >>because of closed specifications (nVidia, various WiFi, Adaptecs etc).
> >
> >speaking of this:
> >
> >if anybody wanted to port the bsd ndis-ulator to plan 9, or use it as 
> >a model for a Plan 9 driver, it would be a HUGE service, and would get 
> >us tons of drivers.
> >
> >ron

Re: [9fans] ndis-ulator for plan 9

[balaji]

there was a thread sometime ago regarding "caged drivers"
and Dave Presotto had some thoughts on it.

On 9/1/05, Devon H. O'Dell <dodell@offmyserver.com> wrote:
> On Thu, Sep 01, 2005 at 08:31:49AM +0100, Dave Lukes wrote:
> > Ron,
> > I'm going to regret this, but I'm gonna trust that you're right and go
> > take a look at it
> > (unless someone else steps up to the plate/wicket).
> >
> > I'm not going to even announce it, let alone discuss it,
> > I'm going to take a look at how much work is involved and either start
> > doing it or not.
> >
> > If I even mention I'm doing it, there will be a flood of "windoze
> > sukz", "is it GPLed?"
> > and "driver X won't work so it ain't worth doing" messages to distract
> > me.
> >
> > Do you happen to know anything about the beast to get me started?
> >
> > Cheers,
> >     Dave.
> 
> Yes. Both Linux and Solaris have used Bill Paul's FreeBSD
> work to get `Project Evil' running in these operating systems.
> Take a look around at:
> 
> /sys/dev/if_ndis/*
> /sys/compat/ndis/*
> /usr/src/usr.sbin/ndiscvt/*
> 
> in the FreeBSD source repository for more info :)
> 
> --Devon
> 
> >
> > On 1 Sep 2005, at 04:59, Ronald G Minnich wrote:
> >
> > >McLone wrote:
> > >
> > >>But not only this - i am unable to write drivers for some hardware
> > >>because of closed specifications (nVidia, various WiFi, Adaptecs etc).
> > >
> > >speaking of this:
> > >
> > >if anybody wanted to port the bsd ndis-ulator to plan 9, or use it as
> > >a model for a Plan 9 driver, it would be a HUGE service, and would get
> > >us tons of drivers.
> > >
> > >ron
>

Re: [9fans] ndis-ulator for plan 9

[Ronald G Minnich]

Dave Lukes wrote:

> Do you happen to know anything about the beast to get me started?
> 

no, I only know that both linux and freebsd, at least, support windoze 
drivers in the kernel. This has been useful for wireless cards.

ron

Re: [9fans] ndis-ulator for plan 9

[McLone]

On 9/1/05, Ronald G Minnich <rminnich@lanl.gov> wrote:
> This has been useful for wireless cards.
aren't Theo's campaign for open specs is more wise move?

AFAIK, NDISulator is tied very closely with FreeBSD 5 kernel.
Porting it = alot of work, that's for sure. But it is useable.
-- 
wbr,                        |\      _,,,---,,_           dog bless ya!
`                       Zzz /,`.-'`'    -.  ;-;;,_
McLone at GMail dot com    |,4-  ) )-,_. ,\ (  `'-'
  net- and *BSD admin     '---''(_/--'  `-'\_)   ...sorry for translit

Re: [9fans] ndis-ulator for plan 9

[Ronald G Minnich]

McLone wrote:
> On 9/1/05, Ronald G Minnich <rminnich@lanl.gov> wrote:
> 
>>This has been useful for wireless cards.
> 
> aren't Theo's campaign for open specs is more wise move?

david lukes, it's starting already.

open specs are fine. we have a driver problem with plan 9, however, and 
an ndis-ulator for plan 9 could help. people are not dropping out of the 
  sky with plan 9 drivers in their hands, even for hardware we have 
specs for.

ron

Re: [9fans] ndis-ulator for plan 9

[Dave Lukes]

 > no, I only know that both linux and freebsd, at least, support 
windoze drivers in the kernel. This has been useful for wireless cards.

How widespread is NDIS:
e.g. would it buy us the occasional non-wireless driver (e.g. for 
mobo-integrated ethernet)?

Cheers,
    Dave.

Re: [9fans] Xen for Windows(Was:vmware 5.0)

[Harri Haataja]

On Wed, Aug 31, 2005 at 08:21:20PM -0400, Brantley Coile wrote:
> these are good reasons and bring to mind a question.  why don't we
> define a generic microkernel that all these systems can use and not
> have to go to all the tricks of VM?

Isn't that what some of the L4 groups have been talking about?

http://l4ka.org/publications/

-- 
Nothing is free.
To acquire anything requires an expenditure of time, energy, or money.
		-- Harry Browne

Re: [9fans] ndis-ulator for plan 9

[Ronald G Minnich]

Dave Lukes wrote:

> How widespread is NDIS:
> e.g. would it buy us the occasional non-wireless driver (e.g. for 
> mobo-integrated ethernet)?

If memory serves (jdegood can correct me) just about every network 
chip/card has an NDIS driver.

ron

Re: [9fans] ndis-ulator for plan 9

[Devon O'Dell]

2005/9/1, Ronald G Minnich <rminnich@lanl.gov>:
> Dave Lukes wrote:
> 
> > How widespread is NDIS:
> > e.g. would it buy us the occasional non-wireless driver (e.g. for
> > mobo-integrated ethernet)?
> 
> If memory serves (jdegood can correct me) just about every network
> chip/card has an NDIS driver.
> 
> ron

I tried to post to the list about this from my address here; had to
change the address since it wasn't on the list. What I said was:

====
Yes. It is already used in FreeBSD for some such chipsets (e.g. nvnet
before there was a driver for this).
====

Theoretically everything with an NDIS driver should work. This isn't
always the case; I know there are a couple wireless chips in Centrino
laptops that don't always work or some functions don't work properly.
NDIS continues to be a WIP, I think, but I'd venture to say that 96% of
the cards with NDIS drivers should work.

--Devon

Re: [9fans] ndis-ulator for plan 9

[Dave Lukes]

 > david lukes, it's starting already.

The Allies have declared War on the Axis of Pragmatism!
The Openhardwareistas are rallying their allies!!
Hark! I hear the massed mumblings of angry hordes of Jeepeyells!
Head for the hills before they lobotomise us for booting a tainted OS!!!

 > open specs are fine.

... for people who like writing drivers (like I did 20 years ago).

My faint memories of driver hacking suggest that specs, open or otherwise,
were seldom as problematical as the hardware.

Basically you treat the manual as a kind of fuzzy photo:
it tells you what's where and vaguely what it looks like.
You then fumble around blindly until it actually works well enough for 
your purposes
and move on.

What you really need is the dude(s) who carved the ASIC sitting next to you.

 > people are not dropping out of the  sky with plan 9 drivers in their 
hands, even for hardware we have specs for.

Yuk. What a gruesome thought.
And even if they did, we'd have to clean the bits of splattered corpse 
off before we could use them.

:-),
    Dave.

Re: [9fans] ndis-ulator for plan 9

[Charles Forsyth]

>>people are not dropping out of the sky with plan 9 drivers in their hands,
>>even for hardware we have  specs for.

is there a page somewhere that lists `hardware with specs but without drivers'?
if not, one might be useful, especially during those long nights or rainy afternoons
with nothing to do.  associated offers of cards might also be useful.

Re: [9fans] ndis-ulator for plan 9

[lucio]

> is there a page somewhere that lists `hardware with specs but without drivers'?
> if not, one might be useful, especially during those long nights or rainy afternoons
> with nothing to do.  associated offers of cards might also be useful.

If you're offering, OHCI is still waiting, EHCI would be absolutely
fabulous.  If I could, I'd volunteer to apprentice, but it's a bit
impractical :-)

As for adapters, they are cheap even in South Africa, do I need to
send you a couple?  Devices are more of an issue.

++L

Re: [9fans] ndis-ulator for plan 9

[Ronald G Minnich]

Charles Forsyth wrote:

> is there a page somewhere that lists `hardware with specs but without drivers'?
> if not, one might be useful, especially during those long nights or rainy afternoons
> with nothing to do.  associated offers of cards might also be useful.
> 


I don't know, but: WARNING: I have learned over the last 5 years that 
vendors consider releasing source code without specs a WAY better way to 
go than releasing specs without source code. In fact, many vendors do 
not differentiate binary from source; they know about disassemblers. 
Once the binary goes out, the feeling is that it could just as easily be 
source.

What a world.

ron

Re: [9fans] ndis-ulator for plan 9

[Russ Cox]

> > is there a page somewhere that lists `hardware with specs but without drivers'?
> > if not, one might be useful, especially during those long nights or rainy afternoons
> > with nothing to do.  associated offers of cards might also be useful.
> >
> 
> I don't know, but: WARNING: I have learned over the last 5 years that
> vendors consider releasing source code without specs a WAY better way to
> go than releasing specs without source code. In fact, many vendors do
> not differentiate binary from source; they know about disassemblers.
> Once the binary goes out, the feeling is that it could just as easily be
> source.

I think Charles was talking about Plan 9 drivers,
so I'm sure there are actually more that release specs
but not Plan 9 drivers than release Plan 9 drivers
but not specs.

Russ

Re: [9fans] ndis-ulator for plan 9

[Jack Johnson]

On 8/31/05, Ronald G Minnich <rminnich@lanl.gov> wrote:
> if anybody wanted to port the bsd ndis-ulator to plan 9, or use it as a
> model for a Plan 9 driver, it would be a HUGE service, and would get us
> tons of drivers.

Along this tack, would there be any value in resurrecting UDI (or
something like it)?

http://www.projectudi.org/

It seems like it never really caught on.  I think the UDI interface
was built into SCO, Linux 2.2 & 2.4 and maybe nBSD, but I've never
actually seen drivers.

It came to mind when someone mentioned porting to microkernel X,
because all the microkernels -- heck, all the alternative OSes on the
periphery -- are in the same boat when it comes to drivers.  It would
be nice if something done for Plan 9 could also benefit L4, etc.

I think the ndisulator would be fantastic, especially for all those
random laptop NICs, but for RAID controllers, video cards, etc.,
someone's going to have to put in the time and it sure would be nice
to share -- and occasionally reap -- the wealth.

-Jack

Re: [9fans] ndis-ulator for plan 9

[Charles Forsyth]

>>I think Charles was talking about Plan 9 drivers,

i did mean a list of hardware that people have got or would like to use
for which specifications exist (.pdf or .url reference would be useful)
but no plan 9 driver.

Re: [9fans] ndis-ulator for plan 9

[Ronald G Minnich]

Russ Cox wrote:

> I think Charles was talking about Plan 9 drivers,
> so I'm sure there are actually more that release specs
> but not Plan 9 drivers than release Plan 9 drivers
> but not specs.


true, but from what I've seen, it looks something like this:
- release Plan 9 drivers: 0%
- release specs: less than 100%
- release Linux drivers: more than release specs

The problem I've seen is that linux drivers are somewhat easier to come 
by than specs, nowadays.

ron

Re: [9fans] ndis-ulator for plan 9

[David Leimbach]

On 9/4/05, Ronald G Minnich <rminnich@lanl.gov> wrote:
> Russ Cox wrote:
> 
> > I think Charles was talking about Plan 9 drivers,
> > so I'm sure there are actually more that release specs
> > but not Plan 9 drivers than release Plan 9 drivers
> > but not specs.
> 
> 
> true, but from what I've seen, it looks something like this:
> - release Plan 9 drivers: 0%
> - release specs: less than 100%
> - release Linux drivers: more than release specs
> 
> The problem I've seen is that linux drivers are somewhat easier to come
> by than specs, nowadays.
> 

And if you don't GPL them you've tainted the kernel and can't use GPLd
data structure and other stuff.  Oh what fun it is to ride.....

> ron
>

Re: [9fans] ndis-ulator for plan 9

[Dave Lukes]

Does anyone know anything about the Linux ndiswrapper 
(http://ndiswrapper.sourceforge.net/)?

It claims to support USB and such ...

Dave.

Ronald G Minnich wrote:

> McLone wrote:
>
>> But not only this - i am unable to write drivers for some hardware
>> because of closed specifications (nVidia, various WiFi, Adaptecs etc).
>
>
> speaking of this:
>
> if anybody wanted to port the bsd ndis-ulator to plan 9, or use it as 
> a model for a Plan 9 driver, it would be a HUGE service, and would get 
> us tons of drivers.
>
> ron

Re: [9fans] ndis-ulator for plan 9

[Uriel]

On Wed, Sep 21, 2005 at 07:35:55PM +0100, Dave Lukes wrote:
> Does anyone know anything about the Linux ndiswrapper 
I know that it's broken in the latest kernels due changes to the kernel
stack size, from reading LKML it seem like it's not going to be fixed
because NDIS just makes too many assumptions about how Windows works.

I'm on the boat that says that is a good thing; OpenHCI and/or EHCI
support would be way more useful and easier, there are open
specifications for both.


> It claims to support USB and such ...
USB network cards, that is... hardly very useful when we don't support
most USB controllers.

uriel

Re: [9fans] ndis-ulator for plan 9

[Russ Cox]

> > It claims to support USB and such ...
> USB network cards, that is... hardly very useful when we don't support
> most USB controllers.

I have yet to run into a USB controller
that didn't work in USB1 mode with Plan 9.

Russ

Re: [9fans] ndis-ulator for plan 9

[Dave Lukes]

> > It claims to support USB and such ...

 > USB network cards, that is... hardly very useful when we don't support
 > most USB controllers.

Then you'd better start coding, Uriel:-).

Seriously: my point was that it *claims* to support a wider range of 
"stuff",
Quote:
    e.g., USB to serial port device, ethernet card, home phone network 
device

than Project Evil, so maybe it's a better starting point,
in case PE is too specific to certain classes of ndis drivers.

Dave.

Re: [9fans] ndis-ulator for plan 9

[Charles Forsyth]

> I have yet to run into a USB controller
> that didn't work in USB1 mode with Plan 9.

OHCI controllers won't work yet, but UHCI does seem to be more common.
there's an Inferno OHCI driver that could be adapted.
EHCI (usb 2) is more of an extension of UHCI, however, so i thought
i'd skip an evolutionary phase (actually, i suppose it's properly a
creationist phase) and do that one, now i've finally got one i can
test against.  there are the usual giggles to be found in the data
sheets. ``what were they on???!*''  as well.

Re: [9fans] ndis-ulator for plan 9

[David Leimbach]

On 9/21/05, Russ Cox <rsc@swtch.com> wrote:
> > > It claims to support USB and such ...
> > USB network cards, that is... hardly very useful when we don't support
> > most USB controllers.
>
> I have yet to run into a USB controller
> that didn't work in USB1 mode with Plan 9.
>

OHCI is not supported by plan 9.  So I guess that's one [and more
common than UHCI it seems]

The only way I can connect a USB keyboard to my plan 9 box right now
is in "legacy" mode via the BIOS.

Dave

Re: [9fans] ndis-ulator for plan 9

[Charles Forsyth]

i say
>OHCI controllers won't work yet, but UHCI does seem to be more common.

leimy2k@gmail.com observes:
>OHCI is not supported by plan 9.  So I guess that's one [and more
>common than UHCI it seems]

obviously they are equally common; vulgar indeed.

that reminds me: an astronomer, a physicist, an engineer,
and a man from DEFRA were in a railway carriage, looking out at a field of sheep...

Re: [9fans] ndis-ulator for plan 9

[David Leimbach]

On 9/21/05, Charles Forsyth <forsyth@terzarima.net> wrote:
> i say
> >OHCI controllers won't work yet, but UHCI does seem to be more common.
>
> leimy2k@gmail.com observes:
> >OHCI is not supported by plan 9.  So I guess that's one [and more
> >common than UHCI it seems]
>
> obviously they are equally common; vulgar indeed.

I was under the impression that most if not all new USB controllers
were OHCI and not UHCI.  Perhaps that's just been my exposure.

Dave

Re: [9fans] ndis-ulator for plan 9

[Charles Forsyth]

> I was under the impression that most if not all new USB controllers
> were OHCI and not UHCI.  Perhaps that's just been my exposure.

that's interesting.  i thought that once myself last year, but suddenly
all the ones i see are UHCI/EHCI.  perhaps there is a market survey somewhere.
it doesn't really matter: there are many OHCI systems and it's sensible to drive them.
(and that's easier than the ndis wrapper FOR THAT PURPOSE given the nature of that code.)

Re: [9fans] ndis-ulator for plan 9

[Kenji Okamoto]

>> I was under the impression that most if not all new USB controllers
>> were OHCI and not UHCI.  Perhaps that's just been my exposure.
> 
> that's interesting.  i thought that once myself last year, but suddenly
> all the ones i see are UHCI/EHCI.  

Probably, David said the USB controller _card_ apart from that on mother
board.

Kenji

Re: [9fans] ndis-ulator for plan 9

[David Leimbach]

On 9/21/05, Charles Forsyth <forsyth@terzarima.net> wrote:
> > I was under the impression that most if not all new USB controllers
> > were OHCI and not UHCI.  Perhaps that's just been my exposure.
>
> that's interesting.  i thought that once myself last year, but suddenly
> all the ones i see are UHCI/EHCI.  perhaps there is a market survey somewhere.
> it doesn't really matter: there are many OHCI systems and it's sensible to drive them.
> (and that's easier than the ndis wrapper FOR THAT PURPOSE given the nature of that code.)
>
>
I sometimes think they just like to flip flop to perturb the more
esoteric OS driver support.

Either way I'm still looking at that OHCI code from Inferno you passed
me, trying to figure out what I can safely strip out and still make
compile and function.  It's got it's own namespace code and stuff in
it though and I'm not sure what that will cause.

I've finally gotten the time to integrate it into a kernel config file
and see the compilation errors but the stability of my plan 9 system
is suspect.  Oftentimes the mouse pointer will go ape-shit and I'll
have to drawterm in to shutdown fossil nicely and reboot.  Now I just
drawterm in all the time.

My spare time is variable and never large these days.

Re: [9fans] ndis-ulator for plan 9

[David Leimbach]

On 9/21/05, Kenji Okamoto <okamoto@granite.cias.osakafu-u.ac.jp> wrote:
> >> I was under the impression that most if not all new USB controllers
> >> were OHCI and not UHCI.  Perhaps that's just been my exposure.
> >
> > that's interesting.  i thought that once myself last year, but suddenly
> > all the ones i see are UHCI/EHCI.
>
> Probably, David said the USB controller _card_ apart from that on mother
> board.
>

No, the on-board is clearly OHCI :).  This is an Nforce2 chipset and
it stinks to high-heaven.

I so want to throw it out the window.

Dave

> Kenji
>
>

Re: [9fans] ndis-ulator for plan 9

[Russ Cox]

> Oftentimes the mouse pointer will go ape-shit and I'll
> have to drawterm in to shutdown fossil nicely and reboot.

Usually this means the mouse protocol has gotten
off by one byte or so.  If your mouse is a PS/2 mouse,
you can wait it out -- stop touching the mouse for at least
two seconds and then try.  I just changed the serial port
readers to do this too, but you'll have to recompile your
kernel to get that.

I've only seen this behavior on notebook trackpads
after you zap them with static electricity.

Russ

RE: [9fans] ndis-ulator for plan 9

[Nigel Roles]

9fans-bounces+ngr=9fs.org@cse.psu.edu wrote:
>> I was under the impression that most if not all new USB controllers
>> were OHCI and not UHCI.  Perhaps that's just been my exposure.
> 
> that's interesting.  i thought that once myself last year, but
> suddenly all the ones i see are UHCI/EHCI.  perhaps there is a market
> survey somewhere.
> it doesn't really matter: there are many OHCI systems and it's
> sensible to drive them. (and that's easier than the ndis wrapper FOR
> THAT PURPOSE given 
> the nature of that code.)

My understanding:

EHCI is an extension to UHCI to enable a much greater degree of buffering
to support USB 2.0 data rates. Prior to USB 2.0, OHCI began to take over
because of VIA's implementation being in so many chipsets, and it being
available in 'soft' form and as a result appearing in a lot of SoC devices.

Now that everything needs USB 2.0, and EHCI is the only register standard,
perhaps it works out more economic to licence UHCI as it gets thrown in
'free' with EHCI? Just a guess.

RE: [9fans] ndis-ulator for plan 9

[Charles Forsyth]

> EHCI is an extension to UHCI to enable a much greater degree of buffering

it's an incompatible extension since asynchronous requests are no longer put in the
frame list but go on a separate queue elsewhere, and the descriptors are different;
it retains many of the annoying features though!
there are extra descriptors to handle the micro-frame intervals
that give the extra speed, and split transactions with the hubs.
the bizarre thing is that the EHCI controller handles only the high-speed
devices, not low-speed or full-speed ones (note that `full speed' is not `high speed').
you need a `companion controller' for the low-/full-speed devices, and that
controller can be either OHCI or UHCI.  of course, that requires special hardware
to route devices to one controller or another per root port, and new software
to manage the distribution of work between them.   of course, the implemenation
can choose from two possible routing schemes.  there is an (optional) `light'
reset (distinguished from the usual `soft' reset).  such fun!

Re: [9fans] ndis-ulator for plan 9

[David Leimbach]

On 9/22/05, Charles Forsyth <forsyth@terzarima.net> wrote:
> > EHCI is an extension to UHCI to enable a much greater degree of buffering
>
> it's an incompatible extension since asynchronous requests are no longer put in the
> frame list but go on a separate queue elsewhere, and the descriptors are different;
> it retains many of the annoying features though!
> there are extra descriptors to handle the micro-frame intervals
> that give the extra speed, and split transactions with the hubs.
> the bizarre thing is that the EHCI controller handles only the high-speed
> devices, not low-speed or full-speed ones (note that `full speed' is not `high speed').
> you need a `companion controller' for the low-/full-speed devices, and that
> controller can be either OHCI or UHCI.  of course, that requires special hardware
> to route devices to one controller or another per root port, and new software
> to manage the distribution of work between them.   of course, the implemenation
> can choose from two possible routing schemes.  there is an (optional) `light'
> reset (distinguished from the usual `soft' reset).  such fun!
>
>

All of this begins to remind me why I normally don't muck around in OS
kernels.  It's so really not fun or interesting to deal with the
braindead-ness of hardware.

There are people that get off on this stuff but it's not me.  I'd
rather drink my brains out.

Re: [9fans] ndis-ulator for plan 9

[Charles Forsyth]

> I'd rather drink my brains out.

the real trick is to do that as needed AND finish the driver(s).
and have them work, of course.

Re: [9fans] ndis-ulator for plan 9

[David Leimbach]

On 9/22/05, Charles Forsyth <forsyth@terzarima.net> wrote:
> > I'd rather drink my brains out.
>
> the real trick is to do that as needed AND finish the driver(s).
> and have them work, of course.
>
>
*nod*

Re: [9fans] ndis-ulator for plan 9

[Ronald G Minnich]

David Leimbach wrote:

> All of this begins to remind me why I normally don't muck around in OS
> kernels.  It's so really not fun or interesting to deal with the
> braindead-ness of hardware.

it's not all bad. USB was a trainwreck from day one. But there's 
infiniban- oh, wait, that's kind of icky too. Well, there's always 
BlueToot -- oh, wait, I'm told that makes USB look sane.

well, still, you still shouldn't give up so easily.

Ah, then, maybe chipset work? You could work with the chipsets and -- 
oh, wait, vendors won't let you see them. Hmm. Well, they're kinda ugly 
anwyay.

Well, there's the busses! Yes, the busses! check out PCI-E, which does 
ARQ in hardware! Now there's some fun -- oh, no, it's not fun at all, 
Forgive me, I forgot. And, I am told, it has just as high latency as 
PCI-X, because of all the extra glop that got added to support ARQ in 
hardware. Hmm.

Well, if you don't want to do hardware, consider the low-level software 
stuff that vendors are pushing, like EFI and ACPI, they're really pretty 
nic -- oh, wait, they're awful.

ah, hmm. Well! this is interesting. I'm putting my life into stuff that 
sucks. What am I doing here?

I shoulda gone to work for a railroad. If you get mad at something on 
the railroad, you get to hit it with a spike hammer (I learned this from 
experience). Although at times I would like to hit a computer with a 
spike hammer. Spike hammers have a nice heft to them, and would leave a 
nice round hole in most computers -- metal cases are no problem.

Or maybe be a window washer! ".... a window washer me ..."

ron

Re: [9fans] ndis-ulator for plan 9

[David Leimbach]

On 9/22/05, Ronald G Minnich <rminnich@lanl.gov> wrote:
> David Leimbach wrote:
>
> > All of this begins to remind me why I normally don't muck around in OS
> > kernels.  It's so really not fun or interesting to deal with the
> > braindead-ness of hardware.
>
> it's not all bad. USB was a trainwreck from day one. But there's
> infiniban- oh, wait, that's kind of icky too. Well, there's always
> BlueToot -- oh, wait, I'm told that makes USB look sane.
>
> well, still, you still shouldn't give up so easily.
>
> Ah, then, maybe chipset work? You could work with the chipsets and --
> oh, wait, vendors won't let you see them. Hmm. Well, they're kinda ugly
> anwyay.
>
> Well, there's the busses! Yes, the busses! check out PCI-E, which does
> ARQ in hardware! Now there's some fun -- oh, no, it's not fun at all,
> Forgive me, I forgot. And, I am told, it has just as high latency as
> PCI-X, because of all the extra glop that got added to support ARQ in
> hardware. Hmm.
>
> Well, if you don't want to do hardware, consider the low-level software
> stuff that vendors are pushing, like EFI and ACPI, they're really pretty
> nic -- oh, wait, they're awful.
>
> ah, hmm. Well! this is interesting. I'm putting my life into stuff that
> sucks. What am I doing here?
>
> I shoulda gone to work for a railroad. If you get mad at something on
> the railroad, you get to hit it with a spike hammer (I learned this from
> experience). Although at times I would like to hit a computer with a
> spike hammer. Spike hammers have a nice heft to them, and would leave a
> nice round hole in most computers -- metal cases are no problem.
>
> Or maybe be a window washer! ".... a window washer me ..."

I've often thought I might be able to bake the ultimate bagel.  Then I
can just get fat and go "BAM! on TV or something.



>
> ron
>

Re: [9fans] ndis-ulator for plan 9

[George Michaelson]

O 
> 
> Or maybe be a window washer! ".... a window washer me ..."

A wandering minstrel I, A thing of rags, and tatters ..

-George
> 
> ron
> 

Re: [9fans] ndis-ulator for plan 9

[Lyndon Nerenberg]

On Sep 22, 2005, at 7:56 AM, Ronald G Minnich wrote:

> I shoulda gone to work for a railroad. If you get mad at something  
> on the railroad, you get to hit it with a spike hammer (I learned  
> this from experience). Although at times I would like to hit a  
> computer with a spike hammer. Spike hammers have a nice heft to  
> them, and would leave a nice round hole in most computers -- metal  
> cases are no problem.

http://orthanc.ca/bad-computer.mpg :-)

Re: [9fans] ndis-ulator for plan 9

[Nigel Roles]

George Michaelson wrote:

>O 
>  
>
>>Or maybe be a window washer! ".... a window washer me ..."
>>    
>>
>
>A wandering minstrel I, A thing of rags, and tatters ..
>
>  
>
Shreds and patches I think. Rags and tatters would have been too obvious 
when
conducting covert romance.