All VM fails when changing dual monitor setup

[gyust]

Hello, here is what I get when I change the setup several times :

virtualbox --debug
Gtk-Message: Failed to load module "canberra-gtk-module"
Qt WARNING: QObject::connect: No such signal QApplication::screenRemoved(QScreen *)
Qt WARNING: QObject::connect: No such signal QScreen::availableGeometryChanged(const QRect &)
Qt WARNING: QObject::connect: No such signal QScreen::availableGeometryChanged(const QRect &)
Qt WARNING: QPainter::begin: Paint device returned engine == 0, type: 2
Qt WARNING: QPainter::end: Painter not active, aborted
Segmentation fault

All virtualbox VM crashes.

I'm running debian 8.7 with 4.9 linux kernel. 01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Venus XTX [Radeon HD 8890M / R9 M275X/M375X] (rev 83) I'm using the radeon opensource driver. The dual monitor is working perfectly except for this issue.

[gyust]

This also happens every 3 suspend wake up cycle, also if you change the display setting with a dual monitor several times.

[gyust]

I'm still having the same issue with the version 5.1.16. My VM are crashing, maybe related to gnome-shell but this is very annoying and I can't trust Virtualboxany more for my professional usage.

[grazvi]

Same here with 5.1.18. Vbox crashes every time after windows host turn offs the monitors. I use 2 Dells connected via Display Port on Nvidia Quadro k2200. I use Debian with mate on guest. Non of this happens with 5.0.32 Vbox. Tahanks.

[Frank Mehnert]

I would like to get a core dump together with the information about which package (exact name please) is used on which Linux distribution. If you can provide a core dump on some URL where I can fetch it, please contact me via private email at frank _dot_ mehnert _at_ oracle _dot_ com.

[Frank Mehnert]

gyust, thank you for your core dump. It turns out that the crash you observe is due to a bug in Qt. The version of Debian 8 is quite old (5.3.2) and the multi-monitor stuff was fixed in Qt 5.5 or 5.6, I don't know the exact version. For Debian 8 we decided to not ship a self-compiled version of Qt because in particular Debian users prefer to use the system-provided Qt version. Qt 5.3.2 works well in most cases but as I said, it has a couple of multi-monitor bugs. We already attempted to show users with Qt < 5.5 a warning about possible bugs....

You have two options: If you are using the universal install (the .run package) then this one comes with Qt 5.6.2 and your problem is most likely fixed. As an alternative you might want to upgrade your host to Debian 9. I know, it's not released yet but the freeze is in progress for a few weeks and I don't expect major changes until Debian 9 is released. We provide "Stretch" packages which requires Qt 5.6 or later (Stretch currently has 5.7.1).

grazvi, which Qt version is used in your case? Have a look at the VBox.log file...

[Frank Mehnert]

grazvi, from your logs (please attach them to this ticket!) I saw that your host is Window so your problem is different. Of course your Qt version is 5.6.2 because on Windows the shipped version of Qt is used.

I'm confused about your problem. You say that with VBox 5.1.18 crashes as soon as one of the multiple monitors goest to sleep. And you wrote that the VM state is marked as "aborted". Later you write that only the guest crashes. What is the actual behavior? Does the VM process crash (VM window vanishes, the VirtualBox process which runs your guest doesn't exist anymore) or does the guest crash (VirtualBox process continues to exist, only the Debian 8 guest crashes? Please clarify.

[grazvi]

Fully functional VM 5.0.36

[grazvi]

additional 5.1.18 log of crashing VM

[grazvi]

crashing VM 5.1.18 log

[grazvi]

Windows error msg after the VM process crash

[grazvi]

the VM process crash when the one of the monitors (Display Port) go to sleep. After weaking the windwos gives the error message complaining about some addresses of RAM (see Attachment Vbox Error2.png). The Vbox GUI still their but no VM process with the guest OS. GUI states the VM is aborted. Though non of this issues on the older build of VM 5.0.36. Thanks

[Frank Mehnert]

In that case we would appreciate a minidump, see here (scroll down to the Windows part).

[grazvi]

Thanks for the quick response. I would like to generate the minidump but i followed the instruction and no dump was generated. May be because the VM gui still runs? please see my registry. What else could be dune to generate the dump?

[grazvi]

regisry config - still no mini dump

[Frank Mehnert]

It's still not clear what happens. What do you mean by "VM GUI is still there" but the "VM crashed"? Are you talking about the VM selector?

And are you starting the VM in "separate" mode?

1. If you are starting your VM in separate mode then starting your VM will start actually 2 processes: One process is the "frontend" process (VirtualBox.exe) which is used to interact with the user and another process (VBoxHeadless.exe) which runs the guest process. It's possible that the frontend process (VirtualBox.exe) crashes and the backend process (VBoxHeadless.exe) continues to run but it's unlikely that the other way around happens.
2. If you just start the VM normally then there is only one process (VirtualBox.exe). If that process crashes then your guest stopped working -- this is like a real computer would be powered off.

I assume that you are working in the 2nd (normal) mode but I want to be 100% sure.

The registry settings look correct. You also need to make sure that there is enough free space on your drive C:.

[grazvi]

Hello,

yes, I just start the VirtualBox normally (by running the VirtualBox.exe) no other option or so. By the GUI I mean even if the guest crashes the VM selector still running the whole guest is gone then.

using older version 5.0.x I have no crashes though the the screens after the sleep gets mixed up. So windows always knows what is the monitor 1 what is monitor 2 this is not a case for the VB. In the screen menu of the VBox it is very often same number representing different physical screen after sleep. By sleep I mean when Windows turns of the monitors (I can reproduce the behavior by locking the PC (win+L) and turning on of the monitors off). No hibernation or so. So what i think that new version of VBox has some problem that one of the screen is gone. As the windows gives mamory error (attached above). So on of the pointers is pointing on the screen but it isn't existing for the VBox and so it crashes.