VISO: changing causes VM crash

[Technologov]

Host: VBox 6.0.0-BETA2 + Win7 x64.
Guest1: Win7 64

After changing VISO frequently multiple times on a running VM, VM finally crashed with memory exception. VM Status = Aborted.
Moreover re-starting this VM from "Aborted" state is not possible, because it will crash again. (until I remove ad-hoc.viso Image manually via VM->Settings)
Basically when generating a VISO image there's 20% chance that it will hit a bug and cause a VM crash; Once VM crashed with buggy VISO image, it's 100% reproducible crash with the same VISO image, if I reboot VM.

UPDATE: here is a bugged VISO image, that causes VM crashes.
https://www.dropbox.com/s/fe53hcaykd868 ... .viso?dl=0

[Whiskeyjack]

could please describe the exact steps to reproduce this? and how did you come up with the 20%? Did you try 5 times and get crash once. That would be a very simple sampling.

[JEBjames]

On a related note...

With 5.2.x I have two vms. One vm builds an iso. And second VM mounts and installs from that ISO.

Yes, sure, it's 100% my fault if the iso is being re-built in one VM at the same time as I'm trying to reinstall from the same iso. Yes, files are being mangled real time. My bad. I expect the guest install in the vm to go wonky and not complete or complain about files being missing or bad checksums or what not. But I don't expect virtualbox to crash.

Almost all the time it's fine.

But if I unload/mount/unload/reboot the vm or reset the vm it can crash that virtualbox vm. Or freeze the vm to the point where none of the menu actions work. Windows indicates the virtualbox guest task doing the install is non responding and eventually I have to kill it via task manager.

[socratis]

On a related note...

No, it is not a related note JEBjames.
• You're talking 5.2.x stuff in a 6.0.0 BETA discussion area.
• You're talking ISO in a VISO thread.
Please don't hijack or taint threads. Next time not related messages will be removed.

[Technologov]

socratis is right. VISO is not related to ISO, and VISO is a new feature introduced in v6.0-only.

>could please describe the exact steps to reproduce this? and how did you come up with the 20%? Did you try 5 times and get crash once. That would be a very simple sampling.

It's difficult to reproduce, but I had the bug hit me 3 times during dozens of VISO tests.
Can Oracle please add VISO debugger into Vbox.log so that all my VISO operations are logged ? This could help a lot.
It should log which files are added to VISO, full path, and file size. (as I'm testing anything between 0 bytes and 100 Gigabyte-files).

In general the idea to reproduce is to try multiple times on large folders of different files.

[bird]

I've tried to reproduce the problem, but so far without any luck. From what I could figure out from the logs, we end up in rtFsIsoMakerOutFile_GenerateDirRecPartial() with a negative off parameter (a little over 3KB). This should be impossible, so I would be very interested in a way to reproduce this.
-bird

[Technologov]

UPDATE: here is a bugged VISO image, that causes VM crashes.
https://www.dropbox.com/s/fe53hcaykd868 ... .viso?dl=0

[bird]

The filename alone does not trigger any trouble here. What's the size of the file?