You are not logged in.
Hello,
when I launch a guest (win or lin) I get that message:
[ 40.402290] SUPR0GipMap: fGetGipCpu=0x3b
--- omissis ---
[ 41.312465] vboxdrv: 00000000b64afd71 VMMR0.r0
[ 41.567624] vboxdrv: 000000000ece7ba1 VBoxDDR0.r0
[ 41.603192] vboxdrv: 00000000f3bd3f3f VBoxNvmeR0.r0
[ 41.617219] vboxdrv: 00000000de53e062 VBoxEhciR0.r0
[ 41.619617] VMMR0InitVM: eflags=246 fKernelFeatures=0x0 (SUPKERNELFEATURES_SMAP=0)
The vms works, but I I had to downgrade Vbox and kernel packages because the vm Fedora was unusable.
sudo pacman -U linux-lts-5.15.43-1-x86_64.pkg.tar.zst linux-lts-headers-5.15.43-1-x86_64.pkg.tar.zst acpi_call-lts-1.2.2-45-x86_64.pkg.tar.zst virtualbox-6.1.34-3-x86_64.pkg.tar.zst virtualbox-host-dkms-6.1.34-3-x86_64.pkg.tar.zst
The problem before was resizing display, but it could work. Today, after updates, it crashed. Crashed gnome-shell, eclipse, chrome and every application for graphic driver. Updated the drivers on guest not solved.
How I can found some log on the system or trace new kernel modules?
PS: the host machines are AMD platform
Offline
Hi,
today i've update all packages, but problem persist.
Downgraded kernel and virtualbox again.
I've tried to upgrade only virtualbox to 6.1.34-4 and continue to use kernel 5.15.43-1 and it's work (virtualbox)
What's changed in the new kernels? It's breaking virtualbox on amd cpu...
Offline
intel core 4th cpu use virtualbox 6.1.34 and linux-lts-5.15.43 no problem , but update to linux-lts-5.15.44 / 45 win vm will bluescreen quickly, tested win 7 and win 10
Offline
intel core 4th cpu use virtualbox 6.1.34 and linux-lts-5.15.43 no problem , but update to linux-lts-5.15.44 / 45 win vm will bluescreen quickly, tested win 7 and win 10
try to change cpu to 1 and graphic driver to vboxsvga
Offline
i switch to kernel 5.15.46 and now 2 on 3 vms not working (only win works... uff!)
fedora 36 and ubuntu 21.04 are unusable (not unstable)
still problem with gnome-shell (in wayland or x11)
Offline
For the past few days I have had lots of stability problems with Virtualbox. I updated my PC 8 days ago, it is possible that the instability started then, not sure exactly when it started. Yesterday (Saturday) I updated again and the instability continues. Flakiness is another way to describe it. I have 15 virtual machines. Of these 1 is Windows that I have not used in a long time, 4 are BSDs and the remaining are Linux. I have flakiness in both the BSD and Linux VMs, however more in Linux. It seems that the problem is corruption of files inside the virtual machine, and as a result of this corruption some of the errors I got are:
- not being able to update due to hashes or signatures not matching the contents of packages
- segmentation faults
- not booting
- virtual machine hanging
- installation of new instance of VM failing.
To give some examples, my FreeBSD and NetBSD are working fine, but not the OpenBSD. Two instances of Debian are working fine but I got problems in just about all the other 8 Linux VMs.
Last edited by archdub (2022-06-11 23:24:31)
Offline
For a bit more context, I am using the lts kernel on a Intel machine. I don't know when the issues started. The reason is that I have been off work for the past week (Monday to Friday, my break has now ended) so over this past week I spent more time than usual using Virtualbox and that is why I came across so many issues. The problems may have started weeks ago without me noticing. I have not tried downgrading anything, though I tried upgrading my system (which I did yesterday) and reinstalling VMs.
As I said above, corruption of files inside the virtual machine could be the common factor among the problems I had. Some examples: I have not had any problem with my NetBSD machine, I just started and stopped it without doing any operation that would change lots of files and that may be why I did not notice any problem. On the other hand, on OpenBSD boot process libraries are reordered as a mitigation. This step sometimes succeeds sometimes not. Also during the boot process another mitigation is starting relinking the kernel. This latter operation takes longer and now it fails every time.
There is this other thread where someone reports a segmentation fault. This could be caused also by corruption of files.
https://bbs.archlinux.org/viewtopic.php?id=277134
~/ pacman -Ss virtualbox | grep inst
core/linux 5.18.3.arch1-1 [installed: 5.17.7.arch1-1]
core/linux-lts 5.15.46-1 [installed]
community/virtualbox 6.1.34-4 [installed]
community/virtualbox-guest-iso 6.1.34-1 [installed]
community/virtualbox-host-dkms 6.1.34-4 [installed]
Last edited by archdub (2022-06-12 09:27:29)
Offline
I edited this post to remove its contents because they were not directly related to the issue at hand (and there is no way to delete the post).
I did further tests. I reinstalled virtual box and its dkms driver and downgraded to version 6.1.34-2, but my tests were not conclusive. The problem with (presumed) file corruption is that it can manifest in different ways and it would take a lot of time to prove it happens. What I have is circumstantial evidence, not concrete proof! As things stand I have several virtual machines that don't boot and reinstalling all of them to do further tests would take a lot of time. I have only reinstalled OpenBSD, it seems the problem continues.
Last edited by archdub (2022-06-12 23:43:37)
Offline
rollback again... now two guest works.
sudo pacman -U linux-lts-5.15.43-1-x86_64.pkg.tar.zst linux-lts-headers-5.15.43-1-x86_64.pkg.tar.zst acpi_call-lts-1.2.2-45-x86_64.pkg.tar.zst virtualbox-6.1.34-3-x86_64.pkg.tar.zst virtualbox-host-dkms-6.1.34-3-x86_64.pkg.tar.zst
Offline
I wasted so much time due to Virtualbox last week that I decided to look for alternatives - regardless of what caused the problems I had. It is a bit sad after using VB for many years.
virt-manager+libvirt+qemu+kvm is what I will try first as it appears to be the setup that is most similar to VB as far as user interfaces go. That said I am open to suggestions.
Apparently it is possible to create create a qemu (qcow2) image out of a VB image (vdi) so in principle I would be able to keep two or three images that I care more about.
Last edited by archdub (2022-06-14 21:00:36)
Offline
I have to downgrade kernel to 5.15.41-1 in order to run certain guest VMs. With newer LTS kernels like 5.15.47-1 (latest), 5.15.46-1, 5.15.43-1 my Windows 10 guests (x64) keeps crashing with BSOD. For me, it's a kernel issue.
Offline
This is from my pacman log:
[2022-05-16T07:24:05+0100] [ALPM] upgraded linux-lts (5.15.37-1 -> 5.15.39-1)
[2022-06-03T03:11:21+0100] [ALPM] upgraded linux-lts (5.15.39-1 -> 5.15.44-1)
[2022-06-11T18:57:03+0100] [ALPM] upgraded linux-lts (5.15.44-1 -> 5.15.46-1)
I know I had problems with 5.15.44 and 5.15.46. I don't remember having issues with 5.15.39, although I cannot guarantee that there were none. So the cause of the problem could be in the kernel - though I have no plans to downgrade the kernel to confirm.
Offline
I've also been experiencing VM instability the last several days. BSOD on Windows 10 guest, various X crashes on Ubuntu 22.04 guest. My CPU is Intel(R) Core(TM) i5-3570K
Curiously, no issues on Windows XP or 32-bit Windows 7 guests, at least none that I noticed. 32-bit Ubuntu did not seem to work any better, but I have not looked very deep into that.
Testing against downgrade:
pacman -U linux-lts-5.15.43-1-x86_64.pkg.tar.zst linux-lts-headers-5.15.43-1-x86_64.pkg.tar.zst virtualbox-6.1.34-3-x86_64.pkg.tar.zst virtualbox-host-dkms-6.1.34-3-x86_64.pkg.tar.zst
Things seem to be working again.
I will probably stay downgraded for the minute, but I also found that QEMU works perfectly fine with the up-to-date packages, e.g.:
qemu-system-x86_64 -m 4G --drive file='/path/to/vdi' -smp cores=2 -enable-kvm
Has this already been reported as an issue to Arch or VirtualBox?
EDIT:
VirtualBox has an open ticket which appears to be related to this issue:
https://www.virtualbox.org/ticket/20914
Last edited by eleanorhawk (2022-06-18 01:24:44)
Offline
today rollback again... what's wrong in the new kernels?
Offline
Instability issue persists when using latest LTS kernel and 6.1.34 packages.
pacman -Ss virtualbox | grep inst
core/linux-lts 5.15.50-1 [installed]
community/virtualbox 6.1.34-4 [installed]
community/virtualbox-guest-iso 6.1.34-1 [installed]
community/virtualbox-host-dkms 6.1.34-4 [installed]
The Windows 10 VM launches, detects mouse, but shows no cursor when using the keyboard. It always freezes/crashes, many times freezing the host KDE desktop and system, leaving no tty access and requiring hard shutdown.
Last edited by zpg443 (2022-06-27 16:03:39)
Offline
Same problem here on 5.15.50 LTS kernel, guests are Windows 8.1 and Windows 11.
Using 5.18.7 stable kernel fixes the issue for now, but any of the later LTS kernels makes any OS unstable.
Offline
Hi, same issues here too. Tried many changes to the Windows guest VM settings, nothing helps. Switched to latest kernel (instead of LTS) for now. Looking forward to a solution. Regards, Zach
Offline
update every things... seems to work all (kernel 5.52-1, acpi_call 1.2.2-54, vbox 6.1.34-4)
vbox vms crashes wile battery charging... full charged and descharging works...
i've a custom configuration and optimized for my thinkpad...
news:
no. do not works, after some time crashes again...
Last edited by gca (2022-07-04 08:45:48)
Offline
rollback again...
Offline
Switched to latest kernel (instead of LTS) for now.
That is my experience too. Using the standard kernel (instead of lts which I normally use), the problems seem to go away. However some virtual machines that got corrupted were unrecoverable so I deleted them.
It seems likely the problem is in the lts kernel, or perhaps something in the lts kernel triggers a bug in VB. It would not be trivial to find the root cause though.
Last edited by archdub (2022-07-04 21:25:54)
Offline
I think that something changing in support with adm cpu/gpu. I've an Ryzen 7 PRO 4750U with Renoir... this is the problem, but no have time to investigate.
I'm waiting for vb 6.36... 'cause there are many problems with new kernels 5.18.
Every sunday update and test, and all works.
Monday work, crashes all... then rollback
I can't stop CD/CI
Offline
My Windows 10 machine was corrupted, but restoring backups of Windows10.vbox and Windows10.vbox-prev at least made it able to launch again. I immediately processed shutdown (successfully) once it launched.
Offline
My Windows 10 machine was corrupted, but restoring backups of Windows10.vbox and Windows10.vbox-prev at least made it able to launch again. I immediately processed shutdown (successfully) once it launched.
do you use UEFI bios in vm?
Offline
Hi, for me things seem to be stable again (now on kernel 5.15.52-1-lts). No issues for a full day now with various W10 VM's (which normally always are rock solid running as guests with Arch/Virtualbox as host for quite some years now). Hopefully I'm not too quick on this one. Sigh... Regards, Zach
(Virtualbox 6.1.34-4)
Last edited by zachalexy (2022-07-06 16:02:32)
Offline
zpg443 wrote:My Windows 10 machine was corrupted, but restoring backups of Windows10.vbox and Windows10.vbox-prev at least made it able to launch again. I immediately processed shutdown (successfully) once it launched.
do you use UEFI bios in vm?
My physical machine uses UEFI. The VM is configured with "Enable PAE/NX," meaning the Physical Address Extension feature of the host CPU is exposed to the VM. There is not otherwise a specific UEFI bios option in VirtualBox.
Offline