Bug 2072556 - Rawhide does not boot in Virtualbox with Kernel 5.18-rc0 or -rc1
Summary: Rawhide does not boot in Virtualbox with Kernel 5.18-rc0 or -rc1
Keywords:
Status: CLOSED RAWHIDE
Alias: None
Product: Fedora
Classification: Fedora
Component: virtualbox-guest-additions
Version: rawhide
Hardware: x86_64
OS: Linux
unspecified
unspecified
Target Milestone: ---
Assignee: Hans de Goede
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2022-04-06 14:11 UTC by Luna Jernberg
Modified: 2022-05-22 01:33 UTC (History)
4 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2022-05-22 01:33:05 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)

Description Luna Jernberg 2022-04-06 14:11:41 UTC
Hey!

Since i ran a dnf update earlier today my VMs with Rawhide and 36 Beta Frozen, does not boot with Kernel 5.18-rc0 (not sure what the problem is but they boot with Kernel 5.17 just fyi)

Comment 1 Sergio Basto 2022-04-06 16:59:49 UTC
5.18-rc0  is kernel of rawhide , not F36 Beta , on F36 you should get kernel 5.17.1

Comment 2 Luna Jernberg 2022-04-07 13:59:27 UTC
Have the same problem with 5.18-rc1 in rawhide

Comment 3 Luna Jernberg 2022-04-08 10:46:20 UTC
Still broken with todays compose

Comment 4 Luna Jernberg 2022-04-10 12:20:31 UTC
Still broken with todays 10th April composehttps://cdimage.ubuntu.com/daily-live/20220409/jammy-desktop-arm64.manifest

Comment 5 Sergio Basto 2022-04-10 19:53:53 UTC
Linux vmdevf36.local 5.18.0-0.rc1.20220408git1831fed559732b1.20.fc37.x86_64 #1 SMP PREEMPT_DYNAMIC Fri Apr 8 16:16:01 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux
boots for me on F36 [2] but without logging and with a gap of 36 seconds or more [1] seems it is "detected fb_set_par" ?

[1]
[   16.709503] daemon-init (1056) used greatest stack depth: 11760 bytes left
[   52.761659] [drm:vmw_msg_ioctl [vmwgfx]] *ERROR* Failed to open channel.
(...)
[   65.118481] detected fb_set_par error, error code: -22

[2]
dnf --disablerepo='*' --enablerepo=rawhide update "kernel*"

Comment 6 Hans de Goede 2022-04-11 07:00:40 UTC
(In reply to Sergio Basto from comment #5)
> Linux vmdevf36.local 5.18.0-0.rc1.20220408git1831fed559732b1.20.fc37.x86_64
> #1 SMP PREEMPT_DYNAMIC Fri Apr 8 16:16:01 UTC 2022 x86_64 x86_64 x86_64
> GNU/Linux
> boots for me on F36 [2] but without logging and with a gap of 36 seconds or
> more [1] seems it is "detected fb_set_par" ?
> 
> [1]
> [   16.709503] daemon-init (1056) used greatest stack depth: 11760 bytes left
> [   52.761659] [drm:vmw_msg_ioctl [vmwgfx]] *ERROR* Failed to open channel.
> (...)
> [   65.118481] detected fb_set_par error, error code: -22
> 
> [2]
> dnf --disablerepo='*' --enablerepo=rawhide update "kernel*"

Thanks that is helpful. This seems to be a regression in the vmwgfx driver.

I assume that if you add "nomodeset" to the kernel commandline, or if you change the virtual GPU to VboxSVGA that the VM will boot with 5.18-rc# properly then?

Comment 7 Luna Jernberg 2022-04-11 07:03:07 UTC
@Hans de Goede: yeah switching it to VboxSVGA makes it work

Comment 8 Luna Jernberg 2022-04-11 07:14:06 UTC
seems gdm, does not like it for GNOME however (but worked on the VMs without X/Wayland)

Comment 9 Hans de Goede 2022-05-09 11:44:49 UTC
I've taken some time today to work on fixing this and I've posted a patch fixing the vmwgfx issue upstream:

https://lore.kernel.org/dri-devel/20220509110425.165537-1-hdegoede@redhat.com/

hopefully upstream will pick this up before 5.18.0 is released, if not then we can add it as a downstream patch in Fedora before we push 5.18.y kernel updates to F35 and F36.

Comment 10 Sergio Basto 2022-05-22 01:33:05 UTC
with  kernel-5.18.0-0.rc7  is fixed , closing , thank you


Note You need to log in before you can comment on or make changes to this bug.