[Qemu-devel] question for QEMU users

[Qemu-devel] question for QEMU users

[Hetz Ben Hamo]

Hi people,

I wanted to ask you this question:

What plugin for QEMU would you consider to buy and for what terms
(support etc) and how much would you pay?

Here are 2 options:

* Virtual IPAQ hardware emulator (all the hardware sans the ROM [legal reasons])
* X86 plugin (to make QEMU use kenel mode and have at least VMWare
speed and performance)
* Others (if you have suggestions - go ahead)..

Comments? Opinions? share with us..

Thanks,
Hetz

Re: [Qemu-devel] question for QEMU users

[John R. Hogerhuis]

What would I buy? I don't know. I'd certainly buy the first item on my
list below, since I've needed it in the past. But I'll answer a broader
question... commercial products involving QEMU I think could work:

1. Utilties that can reliably convert an existing Windows XX image to a
virtual machine image. I know this is a hard problem, but it is THE
killer app for this market. EMC is just entering it, but they're
targeting their marketing to corporate clients. You might even be able
to build a service company around it... you take an image and adapt it
to QEMU by whatever means.

2. Figure out a legal way to create and sell canned, working, licensed
Windows and MacOSX images. You can do the same thing for Linux, BSD,
etc. you just can't charge quite as much.

3. QEMU User + Wine running rootless windows applications under the
native windowing system, for non x86 platforms. Make it smooth and
polished, don't overcharge, and the Macish folk will open their wallets
for you.

4. I think you're on to something with ARM based PDA emulators. Keep in
mind there is wide variation between devices in this market... not quite
as close knit family as Palm stuff.

5. Just make it work, box it up, print manuals, provide support, ala
Crossover Office.

6. Polish the QEMU-Win32 running Knoppix zero-install, and sell it as a
solution to book publishers of entry level Linux stuff. You might be
able to get a small royalty off the CD in the back of the book. 

Ideas are a dime-a-dozen. You got a bargain this time at $.00 for half
dozen.

-- John.

Re: [Qemu-devel] question for QEMU users

[Philipp Gühring]

[-- Attachment #1: Type: text/plain, Size: 311 bytes --]

Hi,

I wouldn´t buy it myself, since I am developing those tools myself. But I 
think that others might buy it:

* QEMU Forensic Toolkit, to do a forensic analysis of a Virtual Machine

* QEMU Task manager, manages the running tasks in a guest-machine from outside

Many greetings,
Philipp Gühring

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [Qemu-devel] question for QEMU users

[Tim]

> * QEMU Forensic Toolkit, to do a forensic analysis of a Virtual Machine

I would pay for the ability to do this.  I would love to be able to dd
off an image of a system, then boot it read-only (via cow image) under
QEMU, and analyze it's behaviour on the network and otherwise without
having to worry about making a full HD copy to a seperate physical
machine, set up the network, etc.  The problem is, almost all systems
being analyzed would be running Windows, and I am not yet convinced
running Windows under Qemu is close to being stable.  I only use a Linux
host, so that part shouldn't be problematic, but being able to drop in a
Windows image to a VM that accurately emulates a wide variety of
hardware is.

In addition, having a stable VM to run Windows under will allow me to
set up honey pots quickly and painlessly.  A variety of emulated
hardware would also allow the VM to look more like a real machine to an
attacker, and that I would also pay for.

So, I guess my position is:  Don't worry too much about supporting lots
of host OSes outside of unix, but any x86 guest should work, and have
options for different hardware (video, network, etc).

thanks,
tim

Re: [Qemu-devel] question for QEMU users

[Joe Batt]

On the professional side, I currently use VMWare under Linux to do win32
development and testing.  It is very convenient to have a portable
(across machines) image of my development workstation.  I also use
different VM images for testing under different environments (OSs and
configurations).  What I don't like about VMWare is that I must load
kernel modules and that it is licensed per (OS,user), not per user, so I
would need to buy another license to use a Windows host.  My machines
are very fluid.  I tend to run things on my laptop, workstation, wife's
machine, client's machines (anywhere there is an open seat), so the
simplicity of qemu compared to VMWare is a huge benefit.  Rebooting to
Knoppix might work, but then I don't have the native tools like Outlook
at the client site, or would require me to reboot my wife's machine
(which is against the Law).

Personally, I would like to use qemu to do distributed processing with
an openmosix guest across a variety of hosts.  I've been excited about
single image clusters since the early 90s and I think qemu could be a
very handy tool in management of them, especially when you consider
running the desktops as peers in the cluster, instead of clients of the
cluster.

I don't currently need GL/DirectX/high end graphics support, but I can
easily imaging the day that it will be required (for apps that use
complex data displays like 3d charts).  Audio is just a nicety for me
(to test failure beeps in the apps I write).

Speed is very important.  Last time I tried, Knoppix was not what I
would consider usable inside qemu on my dual 2.4 Ghz Xeon with 1 GB RAM.

-- 
Joe Batt <Joe@soliddesign.net>

Re: [Qemu-devel] question for QEMU users

[Carlos Valiente]

On Wed, 17 Nov 2004 16:15:12 -0500, Joe Batt <joe@soliddesign.net> wrote:

> Personally, I would like to use qemu to do distributed processing with
> an openmosix guest across a variety of hosts. 

You can do that already, Joe: I have patched together a couple of scripts to set
up the right networking environment
(http://virutass.net/software/qemu/index.html) and I'm happily (and
slowly) running a four node virtual openMosix cluster .

C

Re: [Qemu-devel] question for QEMU users

[Pavel Janík]

   From: Hetz Ben Hamo <hetzbh@gmail.com>
   Date: Wed, 17 Nov 2004 03:07:36 +0200

Hi,

   > * Others (if you have suggestions - go ahead)..

- use vmnet like approach for networking where you simply get IP from your
  local network, no bridges, etc.
-- 
Pavel Janík

I'm simply not a good maintainer, because I'm too impatient and get too
bored with it.
                  -- Linus Torvalds in LKML

Re: [Qemu-devel] question for QEMU users

[Gwenole Beauchesne]

On Wed, 17 Nov 2004, Pavel [iso-8859-2] Janík wrote:

>    > * Others (if you have suggestions - go ahead)..
> 
> - use vmnet like approach for networking where you simply get IP from your
>   local network, no bridges, etc.

I believe Samuel has fixed sheep_net for kernel 2.6 in the MoL tree. That 
could help too.

Re: [Qemu-devel] question for QEMU users

[Jim C. Brown]

On Wed, Nov 17, 2004 at 09:23:23AM +0100, Pavel Jan?k wrote:
>    From: Hetz Ben Hamo <hetzbh@gmail.com>
>    Date: Wed, 17 Nov 2004 03:07:36 +0200
> 
> Hi,
> 
>    > * Others (if you have suggestions - go ahead)..
> 
> - use vmnet like approach for networking where you simply get IP from your
>   local network, no bridges, etc.
> -- 
> Pavel Jan?k

	I do not know how vmnet works. VDE does this (you get your IP from
the local network(1) and you don't have to deal with bridges). The problem
is that you do have firewall/router headaches sometimes.

	It would be nice if we could just run a script or program that works
on all linux systems and fixes up firewall trouble ... this is actually not
difficult to do (assuming the script knows what tapN device (basicly a kind
of ethN device) VDE is using and what the ip addresses are) it is just no one
has bothered to write a generic script to handle this yet. This script could
easily handle automaticly setting up routing as well (but for some people that
includes a kernel recompile).

	It would be nicer if the program could be used on non-nix OSes as well
(modifying a script so it works on various *nixes is not a trivial task but it
is far from unreasonably difficult). But I only use *nix OSes for qemu hosts
so I don't have to think about this part at all.

(1) This "local" network is actually a virtual network. So qemu would get its
IP for the virtual network but it would not be visible on your real LAN without
some router work. Hence the need for that magic script.

-- 
Infinite complexity begets infinite beauty.
Infinite precision begets infinite perfection.

Re: [Qemu-devel] question for QEMU users

[Johannes Schindelin]

Hi,

On Wed, 17 Nov 2004, Hetz Ben Hamo wrote:

> What plugin for QEMU would you consider to buy and for what terms
> (support etc) and how much would you pay?

As I deem myself a decent programmer, I wouldn't ;-)

However, "will hack for money" applies to me, too, and I have a few ideas
myself:

- The Sharp people should be interested in a cheap development platform
for their Zaurus.

- The good people at HP, IBM and Sun should be interested in a cheap and
reliable Terminal Server technology.

and I agree with the other poster who said that just wrapping up and
polishing the GUI a la CodeWeavers is a fantastic idea.

Ciao,
Dscho

Re: [Qemu-devel] question for QEMU users

[malc]

On Wed, 17 Nov 2004, Hetz Ben Hamo wrote:

> Hi people,
>
> I wanted to ask you this question:
>
> What plugin for QEMU would you consider to buy and for what terms
> (support etc) and how much would you pay?
>
> Here are 2 options:
>
> * Virtual IPAQ hardware emulator (all the hardware sans the ROM [legal reasons])
> * X86 plugin (to make QEMU use kenel mode and have at least VMWare
> speed and performance)
> * Others (if you have suggestions - go ahead)..
>
> Comments? Opinions? share with us..

Strange world we are living in, i myself wanted to post a list of things
that would be nice to have in QEMU and ask other readers of this ML about
their feelings. Despite being a big believer in 'Most of software should
be in public comain', i have to say that i'd buy many things, should they
be accessible in my country of origin, anyway here's my list (disordered):

a. UAE (http://www.freiburg.linux.de/~uae/) for QEMU
b. MPU401 support
c. DOSEmu like MFS support
d. Howto how to get (pardon tautology) networking to work under guest DOS
e. Timing in pckbd (but i know that it's Fabrice's itch too)
f. Better v/h-retrace (ditto)
g. Speaker support
h. ....

--
mailto:malc@pulsesoft.com

Re: [Qemu-devel] question for QEMU users

[Derek Fawcus]

On Wed, Nov 17, 2004 at 05:30:46PM +0300, malc wrote:
> 
> Strange world we are living in, i myself wanted to post a list of things
> that would be nice to have in QEMU and ask other readers of this ML about
> their feelings. Despite being a big believer in 'Most of software should
> be in public comain', i have to say that i'd buy many things, should they
> be accessible in my country of origin, anyway here's my list (disordered):
> 
> c. DOSEmu like MFS support

Well I've been working on the above.  I've not touched in in the last couple
of months (and probably won't get back 'til Jan/Feb).  The joy's of the DOS
redir 'API'.

DF

Re: [Qemu-devel] question for QEMU users

[jeebs]

> What plugin for QEMU would you consider to buy and for what terms
> (support etc) and how much would you pay?

Not a lot, since I already have vmware....  And it'd have to be fully open source.  Otherwise there's not much point is changing from a closed source vmware.

 
> * X86 plugin (to make QEMU use kenel mode and have at least VMWare
> speed and performance)

The only way I'd be willing to consider paying for qemu is if it was good enough for me to give up vmware.

1) Which means it could run under Windows fast enough to actually use.

2) And also, if at all possible, have video performance good enough it would actually be possible to play some games, watch dvd's, etc.

3) And, of course, by that time you'd have fine tuned the rest of the program to better emulate hardware, etc. etc. etc.

In other words, actually good enough you could really use it as a real second computer, rather than as an emulated computer.

Number 1 is going to be an issue because it looks like Fabrice's future high speed plans will only work under Linux.

Re: [Qemu-devel] question for QEMU users

[Lennert Buytenhek]

On Wed, Nov 17, 2004 at 03:07:36AM +0200, Hetz Ben Hamo wrote:

> What plugin for QEMU would you consider to buy and for what terms
> (support etc) and how much would you pay?

$50 for working ARM (v5) system emulation for say, a PXA255-based machine
such as the gumstik.  (www.gumstix.com)  I've been wanting to look into
this myself for ages now but suffer from a general lack of time.


--L

Re: [Qemu-devel] question for QEMU users

[brian]

My interest is also in ARM system emulation.  In particular, I'd like to
run ARM linux.  The platform I am planning to use is called the armcore-gx
(http://www.compulab.co.il/armcore-gx.htm).

Also, I'm new to qemu so I'm not quite sure if this is an existing feature
or not, but it would be great to have a solid API so folks could write
custom hardware emulation plugins without recompiling qemu.

As far as paying, I'm not quite sure.  These features would have to mature
to a point where I could freely experiment with them before I decided to
pay.  After that, perhaps I'd pay for development of additional hardware
support.  Or, perhaps I'd contract to have a board properly emulated.  In
any case, I am interested in hacking on these features.

Ciao,
Brian

> On Wed, Nov 17, 2004 at 03:07:36AM +0200, Hetz Ben Hamo wrote:
>
>> What plugin for QEMU would you consider to buy and for what terms
>> (support etc) and how much would you pay?
>
> $50 for working ARM (v5) system emulation for say, a PXA255-based machine
> such as the gumstik.  (www.gumstix.com)  I've been wanting to look into
> this myself for ages now but suffer from a general lack of time.
>
>
> --L
>
>
> _______________________________________________
> Qemu-devel mailing list
> Qemu-devel@nongnu.org
> http://lists.nongnu.org/mailman/listinfo/qemu-devel
>

Re: [Qemu-devel] question for QEMU users

[vaise]

On Wednesday 17 November 2004 02:07, Hetz Ben Hamo wrote:
> What plugin for QEMU would you consider to buy and for what terms
> (support etc) and how much would you pay?
There are two things I would give money for :
-X server add on : to mix Linux and Windows software on the Linux desktop. See 
Cygwin : on the Windows desktop, you can have a Linux application in a 
Windows window. 40 euros.
-of course, I would love a optionnal patch to have nearly-native performance 
(play games...). 60 euros.

a detail : I will never install if it is not open source (I didn't say free). 

[Qemu-devel] Re: question for QEMU users

[Emmanuel Charpentier]

Hetz Ben Hamo wrote:
> Hi people,
> 
> I wanted to ask you this question:
> 
> What plugin for QEMU would you consider to buy and for what terms
> (support etc) and how much would you pay?

Not-so-frivolous idea : have decent sound input support, so one might 
(at last !) install some Winxx setup and Windows speech recognition 
software. This is a thing I learned to enjoy (due to a broken arm some 
times ago) and never been able to run under Linux.

					Emmanuel Charpentier

Re: [Qemu-devel] Re: question for QEMU users

[Richard Neill]

Emmanuel Charpentier wrote:
> Hetz Ben Hamo wrote:
> 
>> Hi people,
>>
>> I wanted to ask you this question:
>>
>> What plugin for QEMU would you consider to buy and for what terms
>> (support etc) and how much would you pay?
> 
> 
> Not-so-frivolous idea : have decent sound input support, so one might 
> (at last !) install some Winxx setup and Windows speech recognition 
> software. This is a thing I learned to enjoy (due to a broken arm some 
> times ago) and never been able to run under Linux.
> 
>                     Emmanuel Charpentier
> 

I second that. I'd like to be able to run Dragon Naturally Speaking 
under guest XP using Linux as the host. That would need sound input from 
the Mic (with Mic boost enabled). [Also, ideally, Qemu would share the 
sound card with the host OS, rather than taking an exclusive lock on it]

I'm quite prepared to contribute the £99 I'd otherwise have to spend on 
VMWare.


Other features I'd like to see are the ability to run at 1600x1200 
resolution, and support for a wheel mouse.

Regards

Richard