The VirtualBox Kernel Driver Is Tainted Crap

[Zajec]

I got huge problem with VirtualBox some time ago when I was testing it. Every time I started my virtual machine, Opera was crashing. I guess kernel module had to corrupt some memory used by Opera. Don't see other explanation. Not sure how it works now, with recent versions of VirtualBox.

Strange. I am using VirtualBox moderatly often for 3 years now, and I'm always working on bleding edge kernel versions. I had never, ever had crash because of VirtualBox. Hover I had reported lots of problems with kernel itself (about 30 genuine new bugs), and it was always problem in kernel, not vboxdrv. So, now what, if I have loaded vboxdrv, developers will reject my bug reports automatically (even If I do nothing with virtualbox or guest system, beside loading module itself)? This is some bullshit.

It's not do-able to track all memory access done by tained module. That would be required to be sure vboxdrv didn't corrupt anything.

[kebabbert]

One difficulty that VirtualBox has, is it supports many more platforms and guest OSes. It is like Windows; it can run on many different machines, unsupported and supported. This makes Windows fragile. Mac OS X runs only on a set hardware and there are less factors that can go wrong.

It is easier to support only a few OSes and few guest OS, than support many. I have now an support issue with VMware workstation, it crashed and they ask me for even more log files now. Duh

[locovaca]

I got huge problem with VirtualBox some time ago when I was testing it. Every time I started my virtual machine, Opera was crashing. I guess kernel module had to corrupt some memory used by Opera. Don't see other explanation. Not sure how it works now, with recent versions of VirtualBox.

1. I start VirtualBox
2. Opera crashes
3. Therefore, VirtualBox is corrupting memory

That's one hell of a jump to that conclusion.

[Hohlraum]

1. I start VirtualBox
2. Opera crashes
3. Therefore, VirtualBox is corrupting memory

That's one hell of a jump to that conclusion.

Sounds like he's got a chair to keyboard interface error. I kid! I've really never had any of these issues with virtualbox on linux but I only ever use it on ubuntu and always use the official builds from the virtualbox repo.

[locovaca]

Sounds like he's got a chair to keyboard interface error. I kid! I've really never had any of these issues with virtualbox on linux but I only ever use it on ubuntu and always use the official builds from the virtualbox repo.

Ditto, and I *do* run Opera as well.

[devius]

The problem isn't virtual box, which runs phenominally well on windows platforms, the problem is that linux is a huge undocumented mess of spaghetti code.

OR some software vendors just don't invest as much on all platforms they support.

[Zajec]

1. I start VirtualBox
2. Opera crashes
3. Therefore, VirtualBox is corrupting memory

That's one hell of a jump to that conclusion.

Sure it isn't any proof, just the easiest explanation for something I've observed to be very consistent. My Opera could run for days without crashing. And every time I've started virtual machine - bang - it crashed.
I'll test it again when I find some time, maybe will find out sth more and grab real proofs.

[DeepDayze]

Sure it isn't any proof, just the easiest explanation for something I've observed to be very consistent. My Opera could run for days without crashing. And every time I've started virtual machine - bang - it crashed.
I'll test it again when I find some time, maybe will find out sth more and grab real proofs.

You should get backtraces and gather log info so that opera team can determine it it is an Opera issue or issue pointing to vbox. Do any other running browsers (for example firefox or chrome) or other programs crash when you start vbox? If so then it could be a memory corruption issue that the kernel team can investigate as well.

[Hohlraum]

Vbox and my web browsers are really the only things I have running that us an inordinate amount of memory as well. How much memory do you have?