Tor Browser crashing in Whonix VirtualBox since upgrade to Host Linux Kernel version 5.10.0-15

Found these posts as I had same reported Tor browser crashing (and noise/artifacts prior to crashes) the other day in Virtualbox on updated to current Pop!_OS 22.04 LTS; this morning I set up qemu using Whonix for KVM on same device and said OS; all issues gone, Whonix and Tor browser acting as they should (tested through using for browsing over four hours, no problems). Hope this is useful.

1 Like

OK, Debian had a new kernel update from 5.10.0-15-amd64 to 5.10.0-16-amd64.

With the new kernel the issue still exists: Tor browser crashing and noise/artifacts prior to crashes.

Still no issues with kernel 5.10.0-14-amd64. Issue only exists with 5.10.0-15 and 5.10.0-16.

Still don’t know if the issue is caused by TOR browser, Whonix, Debian kernel or VirtualBox! I really have not got the machinery or time to test all combinations. I am starting to think that I may switch host distribution (but that is a major undertaking and I would like a guarantee that it works under the new distro before doing so).

The error is as follows:

Fontconfig warning: “/home/user/.tb/tor-browser/Browser/TorBrowser/Data/fontconfig/fonts.conf”, line 85: unknown element “blank”

###!!! [Parent][MessageChannel] Error: (msgtype=0x390142,name=PContent::Msg_CommitBrowsingContextTransaction) Channel error: cannot send/recv

Fontconfig warning: “/home/user/.tb/tor-browser/Browser/TorBrowser/Data/fontconfig/fonts.conf”, line 85: unknown element “blank”
Exiting due to channel error.
Exiting due to channel error.
Exiting due to channel error.
Exiting due to channel error.
Exiting due to channel error.
Exiting due to channel error.
Exiting due to channel error.
Exiting due to channel error.
Exiting due to channel error.
/home/user/.tb/tor-browser/Browser/start-tor-browser: line 362: 3883 Segmentation fault (core dumped) TOR_CONTROL_PASSWD=${TOR_CONTROL_PASSWD} ./firefox --class “Tor Browser” -profile TorBrowser/Data/Browser/profile.default “${@}” < /dev/null
[ERROR] [torbrowser] Tor Browser ended with non-zero (error) exit code!

Tor Browser was started with:
/home/user/.tb/tor-browser/Browser/start-tor-browser --verbose --allow-remote .

Tor Browser exited with code: 139

Hope this helps someone! :slight_smile:

1 Like

If a host kernel upgrade breaks something inside VirtualBox VMs such as:

  • Debian
  • Tor Browser in Debian

then these are Linux host kernel and/or VirtualBox host software issues.

Debian stable (buster at time of writing) or Kicksecure is/are the distribution(s) used by developers.

I doubt anyone will give a guarantee. See legal / licenses.

These happened with earlier versions and are therefore likely unrelated. Also most likely happening outside of Whonix. Therefore please check if this was reported upstream and report a bug if not so yet.

This is the actual issue.

If a host kernel upgrade breaks something inside VirtualBox VMs such as:

  • Debian
  • Tor Browser in Debian

then these are Linux host kernel and/or VirtualBox host software issues.

Debian stable (buster at time of writing) or Kicksecure is/are the distribution(s) used by developers.

I doubt anyone will give a guarantee. See legal / licenses.

These happened with earlier versions and are therefore likely unrelated. Also most likely happening outside of Whonix. Therefore please check if this was reported upstream and report a bug if not so yet.

This is the actual issue.

Did you read this forum thread?
The questions and answer will be the same.

This forum thread was about the host kernel issue.
Not VM kernel issue.

Are you using a Linux based host?

Primary answer is still this link:

Potentially it’s a different issue and resolved by above link.

Hopefully not.

Possibly using https://snapshot.debian.org/

https://www.kicksecure.com/wiki/Free_Support_Principle

“Sources.list.d” file is empty

It’s a a folder. Not a file.

I recommend separate questions. Could you please try to learn these skills as per Potential Solutions Beyond Whonix! /
https://www.kicksecure.com/wiki/Free_Support_Principle / in separate forum threads.

It’s a VirtualBox issue:

https://forums.virtualbox.org/viewtopic.php?f=7&t=106272

Kicksecure / Whonix inside VirtualBox is affected but not the root cause

A workaround will be documented probably within the hour and I’ll post again once available.

Hi, this must be a host kernel issue.
Had the same problem, google search lead me here. I’m running tails in kvm, tor browser was crashing with the latest kernels 5.18 and 5.17 on host (even older versions of tails 4.x), after downgrading to kernel version 5.15.55 (probably LTS version) on kvm host, the browser is stable again. My system is gentoo, except the kernel all the relevant versions are up-to-date (at the time of posting).

I am pretty certain this is a Linux host kernel version vs VirtualBox version issue.

This might be related:

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1012627

(A Linux kernel security patch was backported to the stablized Debian bullseye Linux kernel, which might be the cause for these issues.)

Workaround documented here just now:
Linux Host Kernel versus Tor Browser and other Crashes

Workaround doesn’t solve the issue for me. Both Gateway and Workstation running on 5.10.0-14.

VirtualBox DKMS on Arch.

Only whonix or so as other operating systems inside vbox?

If vbox is hopeless case for you then please use Whonix in KVM version.

Since when does this happen?

In that case it could be a different issue and Tor Browser Crash Errors applies.

Tor Browser Crash Errors was a thing long before this issue kernel issue happened.

I’ve been using Whonix since its early days, updating regularly, and it was flawless until these Tor crashes at the begining of June. As I use it through VB, I thought the solution in this thread would stop them but it’s still unstable.

In my opinion, this issue is rooted in virtualization, the same browser crash is exhibiting on tails running in kvm (see my previous post here), but when I run same tails from usb directly on hardware, tor browser is not crashing. Also, I can consistently reproduce the problem on kvm with host kernels 5.16-5.18, but on 5.15.latest its working fine. My cpu is AMD btw. Still thinking what all these here described details means and what is the common culprit. Something must have been changed in tor browser, which does not like the virtualization, maybe some compile flag, or some specific cpu version, or some unexpected combination of something.

solution partially worked for me: limit the number of virtual CPUs to 1 in Virtualbox settings for Workstation. With this crashes less often.

Also there is an another bug here when Firefox crashes and crash the xorg xserver too. It occurs on some sites that use webgl with the latest extesnions like shaders or so. Seems the problem is in video adapter driver which doesn’t implement these extensions. The last bug occurs both on old real hardware adapters too

1 Like

please upload VirtualBox 6.1.38 to Debian fasttrack


Whonix VirtualBox Integration and Upgrades

My setup: Linux OS (jammy jellyfish) - Oracle VM - Whonix (where i use tor browser) Linux version: No LSB modules are available. Distributor ID: Ubuntu Description: Ubuntu 22.04.1 LTS Release: 22.04 Codename: jammy I have all the lastest software version for the software that I’m using. (I’ve checked)

The problem - My tor browser started crashing displaying a exit node zero error:

“ERROR: Tor Browser ended with non-zero (error) exit code! Tor Browser was started with: /home/user/.tb/tor-browser/Browser/start-tor-browser --verbose --allow-remote . Tor Browser exited with code: 139”

I’ve searched for a while and the only answer i’ve found is to degrade my kernel version but since security is my focus this wouldn’t make any sense:

“ This issue happens after a Linux host kernel upgrade. Common symptoms of this issue are: • APT showing hashsum mismatch. • other kernel issues • Tor Browser crashes

It is happening because VirtualBox does not support yet that kernel version. [5] This wiki chapter was written on 19 July 2022 and might be improved later perhaps with better options as they become available. The only known workaround for now is to downgrade to Debian Linux kernel version 5.10.0-14. Obviously this is a bad solution. “

Can you help me find any other alternative? Or at least let me know if I should wait for an upgrade? Thanks in advance.

Which VirtualBox version are you using?

If…

6.1.32

…too old.


Even…

https://packages.ubuntu.com/jammy-updates/virtualbox

6.1.34

…is too old.


kinetic…

https://packages.ubuntu.com/kinetic/virtualbox

6.1.38

That’s good but I understand Ubuntu Kinetic hasn’t been released yet.

This might be fixed in VirtualBox 6.1.36. And even more likely so in Virtualbox 6.1.38 which is already available from upstream VirtualBox.

The problem is…

…Debian fasttrack at time of writing is shipping only version 6.1.34 and I have no idea when or if Debian will update VirtualBox.

Installation of VirtualBox from VirtualBox.org Repository (6.1.38) might be possible but it’s cumbersome and quote:

Note: after upgrading the VirtualBox host version, this can cause issues with the guest VMs such as broken VM size adjustment (full screen) [2] unless the VirtualBox guest additions are also upgraded inside your virtual machine (which is not covered in these instructions).

related:
Previous VirtualBox Integration