Bug 2220888
Summary: | updating to kernel 6.3.11 breaks amd gpu drivers | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | oli <olze> | ||||||
Component: | linux-firmware | Assignee: | Kernel Maintainer List <kernel-maint> | ||||||
Status: | CLOSED ERRATA | QA Contact: | Fedora Extras Quality Assurance <extras-qa> | ||||||
Severity: | urgent | Docs Contact: | |||||||
Priority: | unspecified | ||||||||
Version: | 38 | CC: | acaringi, adscvr, airlied, alciregi, bskeggs, dwmw2, gwk2112, hdegoede, hpa, jarodwilson, jforbes, josef, joshua, jwboyer, kernel-maint, laura, lgoncalv, linville, masami256, mchehab, pbrobinson, ptalbert, steved, stijn, vbraun.name, vitezslav.zivota | ||||||
Target Milestone: | --- | Keywords: | Desktop | ||||||
Target Release: | --- | ||||||||
Hardware: | Unspecified | ||||||||
OS: | Linux | ||||||||
Whiteboard: | |||||||||
Fixed In Version: | linux-firmware-20230804-152.fc38 linux-firmware-20230804-153.fc37 | Doc Type: | If docs needed, set a value | ||||||
Doc Text: | Story Points: | --- | |||||||
Clone Of: | Environment: | ||||||||
Last Closed: | 2023-08-11 00:41:47 UTC | Type: | --- | ||||||
Regression: | --- | Mount Type: | --- | ||||||
Documentation: | --- | CRM: | |||||||
Verified Versions: | Category: | --- | |||||||
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||||
Cloudforms Team: | --- | Target Upstream Version: | |||||||
Embargoed: | |||||||||
Attachments: |
|
Description
oli
2023-07-06 12:34:23 UTC
the current "workaround" is to delete the file /lib/firmware/amdgpu/dcn_3_1_5_dmcub.bin.xz, then regenerate the initrd with dracut --regenerate-all --force. the igpu is dead but at least the system "works". i tried the workaround from the freedesktop.org url above and after updating that dcn 315 file my system was fully broken. the terminal got spammed by these dmub messages, no login to lightdm possible, system was slow as f... even in terminal only. its a 7950x3d with a 7900xtx On my system f38, Ryzen 7900X, iGPU with dual display: after kernel update to 6.3.11 and reboot both monitors were black. After hard reset, on primary display connected to DP appeared some graphic artifacts during boot and then showed GDM. I was able to login and system seemed to work, but secondary display connected to hdmi remained black without signal. I switched back to 6.3.8. when you edit the boot params and remove rhgb quiet, you probably will also notice the dead screen of DMUB ;) 6.3.8 was working for me too, more or less (some stuters, small freezes), but with newer kernels its totally broken. unfortunately, i dont know how to go back. i rm -rf'd /lib/firmware/amdgpu/dcn* and did a reinstall of amd-gpu-firmware, then dracut regenerate all with force, but no help, it still shows me the newest version that is fully broken > /lib/firmware/amdgpu/dcn* and did a reinstall of amd-gpu-firmware, then
> dracut regenerate all with force, but no help, it still shows me the newest
> version that is fully broken
A reinstall just reinstalls the latest version so that would be expected if the latest version is broken. You need to do "dnf downgrade amd-gpu-firmware" which will take you back to the GA release in Fedora (20230310-148.fc38) which contains the last rev that was in Fedora.
It looks like upstream is updating the firmware (or reverting it or something) but it hasn't landed upstream as yet. (In reply to Peter Robinson from comment #5) > It looks like upstream is updating the firmware (or reverting it or > something) but it hasn't landed upstream as yet. Actually it has (but a dnf downgrade will have the same effect in the short term if you're affected): commit d3f66064cf43bd7338a79174bd0ff60c4ecbdf6d (HEAD -> main, origin/main) Author: Hamza Mahfooz <hamza.mahfooz> Date: Wed Jul 5 16:56:35 2023 -0400 Partially revert "amdgpu: DMCUB updates for DCN 3.1.4 and 3.1.5" This partially reverts commit ade163aaaeae0c1ad20cb3dd8ce878bf61c91b3a. The DCN315 DMCUB firmware update provided by the aforementioned commit wasn't thoroughly tested before being sent for public consumption and as such there are a number of issues with it. So, revert to the previous version until it can be fixed properly. Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2666 Signed-off-by: Hamza Mahfooz <hamza.mahfooz> Signed-off-by: Josh Boyer <jwboyer> downgrading did break something else it seems: [oli@DESKTOP-SJIB21T ~]$ dmesg | grep DMUB [ 5.480892] [drm] Loading DMUB firmware via PSP: version=0x07001900 [ 5.919524] [drm] DMUB hardware initialized: version=0x07001900 [ 6.280890] [drm:dm_early_init [amdgpu]] *ERROR* DMUB firmware loading failed: -19 2nd screen is still dead, have this now installed: amd-gpu-firmware-20230310-148.fc38.noarch I have now amd-gpu-firmware-20230625-151.fc38 and kernel 6.3.8, both display OK, without problems. dmesg | grep DMUB [ 4.277782] [drm] Loading DMUB firmware via PSP: version=0x05000500 Latest working version is amd-gpu-firmware-20230515-150.fc38.noarch.rpm which you can download from https://koji.fedoraproject.org/koji/buildinfo?buildID=2201258, this fixes the "Error waiting for DMUB idle". Do not forget to regenerate initramfs (dracut --regenerate-all --force) after firmware downgrade. PS: This is a different issue, but FYI: if you get a white screen instead of GUI then add amdgpu.sg_display=0 kernel parameter (https://gitlab.freedesktop.org/drm/amd/-/issues/2354) Today update to kernel 6.3.12 didn't fix the issue. Then I downgraded amdgpu firmware to amd-gpu-firmware-20230515-150.fc38 from koji and it works with 6.3.12. Thanks for workaround! $ sudo rpm -Uvh --oldpackage amd-gpu-firmware-20230515-150.fc38.noarch.rpm $ sudo dracut --regenerate-all --force I too suffer from breakage of Dual Monitor capabilities on F38. I noticed it with the 6.3 kernel. Downgrading to amd-gpu-firmware-20230515-150.fc38.noarch.rpm and regenerating the initrd doesn't change anything for me. I've also tried the kernel-6.5.0-0.rc1.20230711git3f01e9fed845.12 just for kicks, no luck there either. For now I'm stuck on 6.2.15-300.fc38.x86_64 until something upstream gets fixed or is otherwise reverted (In reply to joshua from comment #11) > I too suffer from breakage of Dual Monitor capabilities on F38. I noticed > it with the 6.3 kernel. This bug is purely about a regression on a single firmware for a single class of devices as described in the first comment. If you're not suffering from that problem it's a different bug so please file a new bug against the kernel for that. Created attachment 1976815 [details]
broken latest 6.3.12 kernel
this is how it looks with the latest kernel, so its unusable
the .11 kernel seem to work but some apps are fckd... like chromium. i guess because of hw acceleration
Created attachment 1980132 [details]
the new error on kernel 6.4.4
updating to 6.4.4-200 breaks the system even with old amd gpu drivers.
its getting worse and worse.
(In reply to oli from comment #14) > Created attachment 1980132 [details] > the new error on kernel 6.4.4 > > updating to 6.4.4-200 breaks the system even with old amd gpu drivers. > its getting worse and worse. I think that is a different bug, but we should have a new linux-firmware shortly. is there some roadmap or so for the linux firmware? i checked the online git repo but it looks like its dead since 3 or 4 weeks. (In reply to oli from comment #16) > is there some roadmap or so for the linux firmware? i checked the online git > repo but it looks like its dead since 3 or 4 weeks. It releases basically monthly, you could actually see that given the versions, there's be a lot of recent commits upstream this week so I have no idea where you're looking. yeah you are right, i did not re-check it. it was dead for about 2 weeks, that was the time where i checked it. thanks FEDORA-2023-85168977a9 has been submitted as an update to Fedora 37. https://bodhi.fedoraproject.org/updates/FEDORA-2023-85168977a9 FEDORA-2023-d15f5a186a has been submitted as an update to Fedora 38. https://bodhi.fedoraproject.org/updates/FEDORA-2023-d15f5a186a FEDORA-2023-d15f5a186a has been pushed to the Fedora 38 testing repository. Soon you'll be able to install the update with the following command: `sudo dnf upgrade --enablerepo=updates-testing --refresh --advisory=FEDORA-2023-d15f5a186a` You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2023-d15f5a186a See also https://fedoraproject.org/wiki/QA:Updates_Testing for more information on how to test updates. FEDORA-2023-85168977a9 has been pushed to the Fedora 37 testing repository. Soon you'll be able to install the update with the following command: `sudo dnf upgrade --enablerepo=updates-testing --refresh --advisory=FEDORA-2023-85168977a9` You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2023-85168977a9 See also https://fedoraproject.org/wiki/QA:Updates_Testing for more information on how to test updates. Can anyone who was suffering from this problem verify that these F37 and F38 fixes resolve the issue on their system? it is way better than before. before that update it was 100% dmcub errors which lead to instable system, now i can even boot the latest 6.4 kernel but 1 of 10 times the system goes nuts again with dmcub errors, not sure what exactly causes that. i disabled hw acceleration in chromium (because it was not possible to use hw acceleration in any application because of dmcub errors), will enable that now again firmware update FEDORA-2023-d15f5a186a works on my Ryzen 7900X, iGPU. Both displays OK. I did this, I suppose that's enough: sudo dnf upgrade --enablerepo=updates-testing --refresh --advisory=FEDORA-2023-d15f5a186a sudo dracut --force Thanks FEDORA-2023-d15f5a186a has been pushed to the Fedora 38 stable repository. If problem still persists, please make note of it in this bug report. FEDORA-2023-eabbf4ca4d has been pushed to the Fedora 37 testing repository. Soon you'll be able to install the update with the following command: `sudo dnf upgrade --enablerepo=updates-testing --refresh --advisory=FEDORA-2023-eabbf4ca4d` You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2023-eabbf4ca4d See also https://fedoraproject.org/wiki/QA:Updates_Testing for more information on how to test updates. FEDORA-2023-eabbf4ca4d has been pushed to the Fedora 37 stable repository. If problem still persists, please make note of it in this bug report. |