[vbox-dev] New OS bring up with Vbox ?

[Huihong Luo]

As related, this article on doing new os using KVM might be useful:
 
http://www.linuxjournal.com/article/10251
 
As for vbox, I think the first thing to try is to layout your OS in physical memory (real address model) using vbox memory api. You will need to place your OS loader at the x86 BIOS entry location.

--- On Mon, 6/7/10, phi gcc <phi.gcc at gmail.com> wrote:


From: phi gcc <phi.gcc at gmail.com>
Subject: [vbox-dev] New OS bring up with Vbox ?
To: vbox-dev at virtualbox.org
Date: Monday, June 7, 2010, 11:45 PM


Hi Vbox Developpers.

I'd like to ask here a couple of naive questions, I hope I subscribed for the correct mailing list to do so. If the purpose of this mailing list is not appopriate for this, tell me so and I will try to find a better place.

I am a Vbox user (not developer) so I know enough of Vbox to manage a guest.
I am also a OS designer for the purpose of learning, i.e doing mostly micro kernel.
I am not a Vbox designer and as such have no Vbox internal knowledge.

I do have a tiny OS demonstator, with no kernel loader at the moment.

Now my questions.
- Is it possible to to create an FS system (for instance FAT32) in a raw file (say under linux), then encapsulate this raw file to create a Vbox virtual disc (.vdi or other). A kind of import disc data after a virtual disc creation. My guess is that it should be easy.

- In the same idea, is it possible to provide a inital phys mem state, a kind of pre-build snapshot, or handcrafted snap shot, i.e the ability to place in a file (a core file may be) the 'new os' text/data/bss, provide an initial processor(s) context in Guest description, and start the guest with this inital context. This would do about what a kernel loader would do later.

- I red a bit about Vbox development, saw things about the debugger, but along this line I'd like to know if the host could do at some point what could be called a 'guess crashdump' i.e instead of doing a snap shot of a guest, a guest (cooperative guest I think) could ask the host to stop it and dump it, the host would produce a crashdump with the guest phys mem, either as a core file or as a sparse raw file, i.e with whole in absent phys memory or iomem. Note that the host would simply dump the guest memory, it would not provide any a.out (as a coredump do) it is assumed that the kernel developer knows what he stuffed (what kernel) into the phys mem when starting/lauchin the guest

So does this exist already?
If yes, where could I read doccos to educate myself.
If no, is it doable? or usefull? if so I could start to work at trying to implement this, and then what would be your advise to start.
I admit that install the whole SDK and try to peek/poke in there is a bit overwhelming at first.

The all idea of this is to bring up an OS that don't have x86 loader yet or install process, and integrate from the early OS design a good level of cooperation with the host, to ensure that the new OS is well supported by the Vbox host. The idea is that Vbox is a good platform for student os development, and tiny OS demonstator can be designed simpler that using another host os, like doing all this under linux and rebooting either the dev linux kernel and the toy kernel alternativly.

Thanx in advance for your advises
Cheers
Phi


-----Inline Attachment Follows-----


_______________________________________________
vbox-dev mailing list
vbox-dev at virtualbox.org
http://vbox.innotek.de/mailman/listinfo/vbox-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.virtualbox.org/pipermail/vbox-dev/attachments/20100608/4fec92a9/attachment.html>