Re: OSX virtualbox 3 beta still dumps core on NetBSD 5.0 release .iso

[Michal Suchanek <hramrach%centrum.cz@localhost>]

2009/6/21 Chavdar Ivanov <ci4ic4%gmail.com@localhost>:
> 2009/6/20 Michal Suchanek <hramrach%centrum.cz@localhost>:
>> 2009/6/20 Chavdar Ivanov <ci4ic4%gmail.com@localhost>:
>>> 2009/6/20 Thor Lancelot Simon <tls%rek.tjls.com@localhost>:
>>>> On Sat, Jun 20, 2009 at 11:38:01AM +0100, Chavdar Ivanov wrote:
>>>>> 2009/6/20 Matthias Scheler <tron%zhadum.org.uk@localhost>:
>>>>> >
>>>>> > Well, I'm inclined to blame this on VirtualBox as NetBSD works
>>>>> > like a charm under VMware Fusion.
>>>>>
>>>>> This logic is flawed. Here I can list all the operating systems which
>>>>> work fine under VirtualBox.
>>>>
>>>> Uh, no. ÂNetBSD works fine on the physical hardware VirtualBox is
>>>> purportedly emulating. ÂIf it doesn't work on VirtualBox, that can't
>>>> possibly be caused by anything but bugs in VirtualBox.
>>>
>>> I accepted that above. What I added was that NetBSD 4.0 is working,
>>> NetBSD 5.0 and later is not, so something clever has happened in
>>> between, which we might be able to trace from 4.99.xx commits, in
>>> order to avoid it or submit a bug report of some kind to Sun.
>>>
>>> The remark about the flawed logic was simply result of the frustration
>>> knowing that most operating systems I daily use work fine under
>>> VirtualBox, so I can use it for testing and prototyping; not NetBSD.
>>
>> That's not exactly true. Windows work quite well with VirtualBox
>> because that's what VirtualBox was tested with but if you install some
>> rootkit (called antivirus by the vendor) they may suddenly become
>> unable to boot pretty much as any of the other systems that fail out
>> of the box.
>
> For instance?

For instance installing "Application and device control" feature of
Symantec Endpoint Protection 11 causes windows XP processes to
segfault. It seems that processses would segfault at random but so
often that you usually cannot get to the login prompt or your
login.exe segfaults shortly after login which logs you off again (and
you do not get back to the login prompt then).

>
>>
>>> This is not a problem for me per se, as I have enough physical
>>> hardware running NetBSD, as well as VMWare machines for similar
>>> purposes. I can't imagine it being *that* difficult to fix,
>>
>> It depends on the kind of flaw VirtualBox has and how fundamental to
>> its current implementation that flaw is.
>> One of the hardest part might be to actually identify the flaw, though.
>>
>>> unfortunately this is well beyond me to help, except with a trace or a
>>> dump. I also realize that this can be fairly low priority for any
>>> serious NetBSD developer, and, of course, realize that this is a
>>> volunteer project.
>>>
>>> May be someone from VirtualBox development will hear this noise and
>>> patch a thing or two; there were a few NetBSD-related patches in their
>>> recent builds...
>>>
>>
>> I guess they are quite well aware of the fact that not everything
>> works in VirtualBox (this also includes OpenBSD)
>
> Nice, but not cigar... dmesg.boot follows:
<removed dmesg>
>
> It is not very stable, the installation often segfaults, but
> continues. I have now an instance of OpenBSD-Current, running under
> VBox3.0beta. Xorg segfaults, which is not important, although of
> course adds a bit to your argument.

That's the problem here. If stuff segfaults at random it is not much
use that you can boot the system and see the dmesg.

Thanks

Michal