Building firmware on NetBSD

Building firmware on NetBSD

[Pierrick Brossin]

Hey guys,

I am trying to get firmware (hvmloader) to work on NetBSD/XEN 3.0.

I have succesfully built it but it segfaults on _start when I run it.

Is anyone skilled enough to help a bit with the asm code ?
Or maybe someone can give me information.

Thank you for your time


-Pierrick Brossin

RE: Building firmware on NetBSD

[Petersson, Mats]

 

> -----Original Message-----
> From: xen-devel-bounces@lists.xensource.com 
> [mailto:xen-devel-bounces@lists.xensource.com] On Behalf Of 
> Pierrick Brossin
> Sent: 03 October 2006 14:17
> To: xen-devel@lists.xensource.com
> Subject: [Xen-devel] Building firmware on NetBSD
> 
> Hey guys,
> 
> I am trying to get firmware (hvmloader) to work on NetBSD/XEN 3.0.
> 
> I have succesfully built it but it segfaults on _start when I run it.

What do you mean "when I run it"? Hvmloader is not a regular
application, it's a piece of code intended to be loaded by Xen into the
virtual machine and executed there - is that what you're doing, or are
you typing "hvmloader" at the command prompt on the machine - the latter
is almost guaranteed to fail immediately (_start sounds like the first
thing that happens). 

As far as I can think, there shouldn't be any difference to this on a
NetBSD or Linux platform, as both use gcc and related tools to build the
binary, and unless something has gone wrong with the link-scripts or
there are bugs in the gcc & friends for NetBSD that generates different
code or the linker locates the code to a different location, I don't see
how the OS that you build hvmloader on would of any matter... 

--
Mats
> 
> Is anyone skilled enough to help a bit with the asm code ?
> Or maybe someone can give me information.
> 
> Thank you for your time
> 
> 
> -Pierrick Brossin
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
> 
> 
> 

Re: Building firmware on NetBSD

[Pierrick Brossin]

On Tue, Oct 03, 2006 at 03:53:53PM +0200, Petersson, Mats wrote:
> What do you mean "when I run it"? Hvmloader is not a regular
> application, it's a piece of code intended to be loaded by Xen into the
> virtual machine and executed there - is that what you're doing, or are
> you typing "hvmloader" at the command prompt on the machine - the latter
> is almost guaranteed to fail immediately (_start sounds like the first
> thing that happens). 

Yes I am running in on CLI. I am happy to hear it. I have successfully
been able to build hvmloader and will therefore try to run something
using it tonite when I'll be home.

Is it the only thing we will need to be able to run unmodified guest os?

Thanks you for your time

-Pierrick Brossin

RE: Building firmware on NetBSD

[Petersson, Mats]

 

> -----Original Message-----
> From: xen-devel-bounces@lists.xensource.com 
> [mailto:xen-devel-bounces@lists.xensource.com] On Behalf Of 
> Pierrick Brossin
> Sent: 03 October 2006 15:05
> To: Petersson, Mats
> Cc: xen-devel@lists.xensource.com
> Subject: Re: [Xen-devel] Building firmware on NetBSD
> 
> On Tue, Oct 03, 2006 at 03:53:53PM +0200, Petersson, Mats wrote:
> > What do you mean "when I run it"? Hvmloader is not a regular
> > application, it's a piece of code intended to be loaded by 
> Xen into the
> > virtual machine and executed there - is that what you're 
> doing, or are
> > you typing "hvmloader" at the command prompt on the machine 
> - the latter
> > is almost guaranteed to fail immediately (_start sounds 
> like the first
> > thing that happens). 
> 
> Yes I am running in on CLI. I am happy to hear it. I have successfully
> been able to build hvmloader and will therefore try to run something
> using it tonite when I'll be home.
> 
> Is it the only thing we will need to be able to run 
> unmodified guest os?

That, ioemu (modified qemu) and the relevant processor, yes... Hvmloader
copies the bios into the correct place in memory, so that it's able to
boot the "disk" that you specify. Of course, you do need an operating
system to run too ;-)

--
Mats
> 
> Thanks you for your time
> 
> -Pierrick Brossin
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
> 
> 
> 

Re: Building firmware on NetBSD

[Pierrick Brossin]

On Tue, Oct 03, 2006 at 04:12:05PM +0200, Petersson, Mats wrote:
> That, ioemu (modified qemu) and the relevant processor, yes... Hvmloader
> copies the bios into the correct place in memory, so that it's able to
> boot the "disk" that you specify. Of course, you do need an operating
> system to run too ;-)

Thanks for your time.

I will try to make ioemu work on NetBSD and see how things go.
What do you mean with relevant processor. One that supports
virtualization such as Core 2 Duo right ?

Is running unmodified os as fast as running a modified one ?


-Pierrick Brossin

RE: Building firmware on NetBSD

[Petersson, Mats]

 

> -----Original Message-----
> From: xen-devel-bounces@lists.xensource.com 
> [mailto:xen-devel-bounces@lists.xensource.com] On Behalf Of 
> Pierrick Brossin
> Sent: 03 October 2006 15:17
> To: Petersson, Mats
> Cc: xen-devel@lists.xensource.com
> Subject: Re: [Xen-devel] Building firmware on NetBSD
> 
> On Tue, Oct 03, 2006 at 04:12:05PM +0200, Petersson, Mats wrote:
> > That, ioemu (modified qemu) and the relevant processor, 
> yes... Hvmloader
> > copies the bios into the correct place in memory, so that 
> it's able to
> > boot the "disk" that you specify. Of course, you do need an 
> operating
> > system to run too ;-)
> 
> Thanks for your time.
> 
> I will try to make ioemu work on NetBSD and see how things go.
> What do you mean with relevant processor. One that supports
> virtualization such as Core 2 Duo right ?

Yes, although I prefer the AMD versions, for obvious reasons... 
> 
> Is running unmodified os as fast as running a modified one ?

Depends on what you're doing, generally the modified will win - but in
some cases it's hard to get the source-code to modify [or find someone
that is ABLE to modify it for that matter]. 

--
Mats
> 
> 
> -Pierrick Brossin
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
> 
> 
> 

Re: Building firmware on NetBSD

[Pierrick Brossin]

On Tue, Oct 03, 2006 at 04:20:09PM +0200, Petersson, Mats wrote:
> Depends on what you're doing, generally the modified will win - but in
> some cases it's hard to get the source-code to modify [or find someone
> that is ABLE to modify it for that matter]. 

What I mean is. In what way running an unmodified os with xen is better
than running it directly in qemu ?

Is it faster ? Is it completely different ?


Thanks again for your time


-Pierrick Brossin

RE: Building firmware on NetBSD

[Petersson, Mats]

 

> -----Original Message-----
> From: xen-devel-bounces@lists.xensource.com 
> [mailto:xen-devel-bounces@lists.xensource.com] On Behalf Of 
> Pierrick Brossin
> Sent: 03 October 2006 15:26
> To: Petersson, Mats
> Cc: xen-devel@lists.xensource.com
> Subject: Re: [Xen-devel] Building firmware on NetBSD
> 
> On Tue, Oct 03, 2006 at 04:20:09PM +0200, Petersson, Mats wrote:
> > Depends on what you're doing, generally the modified will 
> win - but in
> > some cases it's hard to get the source-code to modify [or 
> find someone
> > that is ABLE to modify it for that matter]. 
> 
> What I mean is. In what way running an unmodified os with xen 
> is better
> than running it directly in qemu ?

It's "different", and faster than qemu, for sure - if it's not, then
something is wrong with Xen... Qemu emulates instructions, Xen fully
virt guests only have certain instructions emulated in Xen. It uses qemu
to perform hardware accesses, such as hard-disk interface and network
card, but not for the instructions that are executed in the guest. 
> 
> Is it faster ? Is it completely different ?

Should be faster, as long as you don't spend 100% of the time doing disk
or network accesses [in which case the difference is probably in QEMU's
favour - but difference would be small anyways, as most of the time is
spent simulating the IO access, and little time actually performing some
"real" work]. 

--
Mats
> 
> 
> Thanks again for your time
> 
> 
> -Pierrick Brossin
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
> 
> 
>