-
Bug
-
Resolution: Done-Errata
-
Normal
-
rhel-9.5
-
kernel-5.14.0-486.el9
-
Yes
-
Critical
-
CustomerScenariosInitiative
-
1
-
rhel-sst-virtualization-hwe
-
ssg_virtualization
-
20
-
26
-
200
-
False
-
-
None
-
zKVM CY24Q3
-
-
s390x
-
None
What were you trying to do that didn't work?
Connect to the guest console via virtio-gpu
Please provide the package NVR for which bug is seen:
guest kernel 5.14.0-467
host kernel 5.14.0-460
How reproducible:
100%
Steps to reproduce
- Set up a VM with virtio-gpu (but without an actual desktop environment), e.g.
# virt-install --import --disk <installed-qcow2> --os-variant rhel9-unknown --name latest_rhel9 --memory 2048 --vcpus 2 --graphics vnc,listen=0.0.0.0 --input keyboard,bus=virtio --input tablet,bus=virtio --name rhel9 - Use VNC or virt-viewer to connect to the guest console
Expected results
The text console is displayed
Actual results
The session doesn't receive any graphics
Additional notes
thuth@redhat.com reported this. He already narrowed down the affected kernel version: up to 5.14.0-464 the console is still shown, since then not.
Note that this only affects the text console, if the guest has a working desktop environment the graphics still work as expected. In that case the issue can be reproduced by stopping the gdm daemon.
This was hit on s390x. I'll ask for help to reproduce this on other archs.
- links to
-
RHSA-2024:128795
kernel bug fix and enhancement update
- mentioned in
-
Page
No Confluence page found with the given URL.
[RHEL-45158] text console via virtio-gpu is not working on s390x anymore
Since the problem described in this issue should be resolved in a recent advisory, it has been closed.
For information on the advisory (Moderate: kernel security update), and where to find the updated files, follow the link below.
If the solution does not work for you, open a new bug report.
https://access.redhat.com/errata/RHSA-2024:9315
FWIW: I replied in https://issues.redhat.com/browse/RHEL-52444 with regards to the --extra-args "inst.text" problem.
Hi smitterl@redhat.com , https://issues.redhat.com/browse/RHEL-52444 filed against the issue I met.
Thank you lcong@redhat.com! It seems that it's another issue. I can reproduce what you describe but after the installation finishes and reboots I do see the console. 
So I'll set this to release pending.
However, we should understand why we can't see the text console over virtio-gpu. Please can you open a ticket against the anaconda component? Btw I didn't use the --cdrom parameter, only the -l for the installation.
Hi thuth@redhat.com , I found a similar issue when installing guest in text mode over vnc, could help to clarify whether it is the same issue?
Host package versions:
qemu-kvm-9.0.0-7.el9.s390x
libvirt-10.5.0-4.el9.s390x
kernel-5.14.0-486.el9.s390x
Guest package versions:
kernel-5.14.0-487.el9.s390x
Test steps:
1. Download an boot iso from compse: RHEL-9.5.0-20240731.4, and save to /var/lib/libvirt/images/boot.iso
2. install guest with virt-install cmd like below:
# virt-install --memory 2048 --vcpus 2 --os-variant rhel9-unknown --graphics vnc,listen=0.0.0.0 --input keyboard,bus=virtio --input tablet,bus=virtio --disk size=20 --cdrom /var/lib/libvirt/images/boot.iso -l http://<repourl>/rhel-9/composes/RHEL-9/RHEL-9.5.0-20240731.4/compose/BaseOS/s390x/os/ --extra-args "inst.text"
3. Connect to the guest with vnc, but get black screen like screenshot of "black_screen_from_vnc.png" in attachment.
4. Use virsh console rhel9 to connect to guest serial console, then I could see the text installation info, which is like screenshot of "info_from_virsh_console.png" in attachment.
bfu@redhat.com : I think the status will change once this ticket gets attached to an erratum. Let's wait some more days, but if it then still did not change, we need to have a closer look.
Leo Fu
added a comment - thuth@redhat.com I think the status of this is not changed manually before, and it keep in "in progress" for a while could you please help check?
Leo Fu
added a comment - thuth@redhat.com I think the status of this is not changed manually before, and it keep in "in progress" for a while could you please help check? Leo Fu
added a comment - as we discussed, we would take the risk of not triggering critical regression to verify this issue, and since it's graphical issue, please check the attachment of the test result, the preliminary test is PASS for this one
Leo Fu
added a comment - as we discussed, we would take the risk of not triggering critical regression to verify this issue, and since it's graphical issue, please check the attachment of the test result, the preliminary test is PASS for this one smitterl@redhat.com , I also think that the regression risk is very, very small. It's only changing the drm_fbdev code, so it can hardly influence any other parts.
Since the problem described in this issue should be resolved in a recent advisory, it has been closed.
For information on the advisory (Moderate: kernel security update), and where to find the updated files, follow the link below.
If the solution does not work for you, open a new bug report.
https://access.redhat.com/errata/RHSA-2024:9315