Bug ID | 1206651 |
---|---|
Summary | Leap 15.4 guests running VirtualBox 7.0.4 crash when run on hosts with certain Intel graphics cards |
Classification | openSUSE |
Product | openSUSE Distribution |
Version | Leap 15.4 |
Hardware | x86-64 |
OS | openSUSE Leap 15.4 |
Status | NEW |
Severity | Major |
Priority | P5 - None |
Component | Virtualization:Other |
Assignee | virt-bugs@suse.de |
Reporter | Larry.Finger@gmail.com |
QA Contact | qa-bugs@suse.de |
Found By | --- |
Blocker | --- |
Long-term tests of Leap 15.4 guests running VirtualBox 7.0.4 find crashes of the VM. If 'journalctl -f' is running on the VM, the output shows repeated instances of the following: Dec 18 07:36:28 suse15r2 kernel: [drm:drm_crtc_commit_wait [drm]] *ERROR* flip_done timed out Dec 18 07:36:28 suse15r2 kernel: [drm:drm_atomic_helper_wait_for_dependencies [drm_kms_helper]] *ERROR* [CRTC:38:crtc-0] commit wait timed out Testing showed that VMs running the Oracle version of the code did not have the problem, which led to the solution. When the kernel module vboxvideo.ko became part of the kernel, generation of that module was disabled in the openSUSE package. Although this decision had not previously caused problems, something in the 7.0.4 code is interacting badly with that kernel module from 5.14.X when used with the i915 driver for the Intel Corporation CoffeeLake-S GT2 [UHD Graphics 630] hardware. The "bug" is causing the VM to error. There may be other types of Intel graphics devices that error - this was the only one that was available for testing. The problem happened on every machine with this graphics device, and is not an obscure hardware error. On one machine with nVidia hardware using the nouveau driver, no problems were noted. We also did not find the problem with Tumbleweed guests. It seems that whatever causes the bug, it was fixed between kernel 5.14.X used in Leap 15.4 and kernel 6.0.X used in Tumbleweed. Once the module vboxvideo.ko was generated as part of the virtualbox-kmp-default package, and the new code was installed on the VM, no further problems of this type were noted. Outside the scope of this defect, but worth noting, all Windows 11 guests run with the 7.0.4 guest extensions crashed. Furthermore, these crashes corrupted the virtual disk of the guest. Oracle knows of this problem and strongly recommends that Windows 11 guests be run with the VB 6.1.40 guest extensions. We have not found any problems with this configuration. Whether Windows 10 is also affected is not clear.