| === Kangarooo_ is now known as Kangarooo | ||
| === xenial is now known as Guest5281 | ||
| bluesceada | luis220413, thanks ... I thought that these "hwe" kernels were meant to be somewhat up-to-date ... | 11:28 |
|---|---|---|
| bluesceada | I think this fix is really necessary for a couple of AMD graphics users, so maybe I should open a ubuntu bug separately that this fix should be backported to the ubuntu kernel? | 11:29 |
| bluesceada | or try to get mainline kernel maintainers to get it in the next 5.15 maintenance release | 11:30 |
| luis220413 | bluesceada: You should open both a bug in the Ubuntu kernel you are using (with the command ubuntu-bug linux) and a bug in the upstream kernel and link the Ubuntu bug to the upstream bug. | 12:05 |
| luis220413 | To file bugs in the upstream kernel: https://bugzilla.kernel.org | 12:05 |
| bluesceada | I think the bug is fixed, following that freedesktop bug | 12:05 |
| bluesceada | Need to just find the upstream kernel bug, I guess | 12:05 |
| luis220413 | Please run the command uname -srvmpio and paste the output here | 12:06 |
| luis220413 | There is no need for paste services, the output is just 1 line long | 12:06 |
| bluesceada | Linux 5.15.0-47-generic #51-Ubuntu SMP Thu Aug 11 07:51:15 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux | 12:07 |
| bluesceada | According to the freedesktop bug, just running a 5.17 kernel might fix it ... maybe I should just do that for testing purpose. | 12:08 |
| bluesceada | imho, the AMD maintainer's job would be to make sure the respective patches get into the 5.15 vanilla kernel | 12:09 |
| luis220413 | bluesceada: Is the bug harming your system right now? | 12:11 |
| bluesceada | Yes, it's not stable | 12:12 |
| bluesceada | However not even sure if it is exactly that bug, there was another one, but it seems they all get fixed in recent kernels, and ubuntu just has that old kernel | 12:14 |
| luis220413 | I am looking at the kernels in Ubuntu 22.04 and Kinetic (the current development release, also mentioned as 22.10). | 12:15 |
| luis220413 | Kinetic has some 5.19 kernels, but they be in -proposed and Kinetic is unstable. | 12:15 |
| luis220413 | And 5.18 and 5.17 kernels as well | 12:15 |
| bluesceada | that might be another issue somewhat related, and also fixed by a more recent kernel: https://gitlab.freedesktop.org/drm/amd/-/issues/1819 | 12:16 |
| ubottu | Issue 1819 in drm/amd "RX 6600 XT fails to resume from suspend in 5.15.y series" [Opened] | 12:16 |
| bluesceada | i also have a 6600 XT and the bug is always triggered when I set the screen to go to suspend after a while | 12:17 |
| bluesceada | currently I set it to "Never" | 12:17 |
| bluesceada | Just to clarify: the screen is sufficient, not the whole system going to suspend | 12:17 |
| luis220413 | bluesceada: I also had to do that in my Ubuntu 21.10 (now EOL) and 22.04 VMs, because they failed to resume! | 12:17 |
| bluesceada | I set "Screen Blank: Never" | 12:18 |
| luis220413 | I set "Automatically lock: Never" | 12:18 |
| bluesceada | Funnily, the bug can sometimes even be triggered when shutting down the system, or when not immediately logging in after it booted | 12:18 |
| bluesceada | so the system might be stuck on shutdown, once that kept the machine running for the night x.x | 12:19 |
| luis220413 | bluesceada: You can try the 5.17 OEM kernel. | 12:19 |
| luis220413 | sudo apt install linux-oem-5.17 | 12:19 |
| bluesceada | ok | 12:19 |
| bluesceada | it only has 5.17-something | 12:19 |
| bluesceada | headers or tools | 12:20 |
| bluesceada | do i need some extra repo? or go with linux-image-5.17.0-1015-oem ? | 12:20 |
| luis220413 | Sorry, wrong command. | 12:20 |
| luis220413 | Do you believe that a 5.17 kernel fixes your issue? | 12:21 |
| luis220413 | sudo apt install linux-image-oem-22.04 | 12:21 |
| bluesceada | <luis220413> Do you believe that a 5.17 kernel fixes your issue? ---> yes it might | 12:22 |
| bluesceada | however one of those bugs mention a 5.18 kernel, let's try it | 12:22 |
| luis220413 | This command will install an Ubuntu kernel based on Linux v5.17.15. | 12:23 |
| luis220413 | bluesceada: Only if you upgrade to the development release or find a backport from a PPA. | 12:23 |
| bluesceada | ok i need also the headers for virtualbox dkms, but thanks, let's see | 12:23 |
| luis220413 | You can also monitor the 22.04 repositories with these two commands: | 12:23 |
| luis220413 | apt list linux\*5.18\* | 12:24 |
| bluesceada | This OEM sounds strange, that's why I didn't test it yet | 12:24 |
| luis220413 | apt list linux\*5.19\* | 12:24 |
| bluesceada | thanks | 12:24 |
| luis220413_ | There is a caveat of using the OEM kernel: "However, if it is run on hardware that is not certified by Canonical, we cannot guarantee it works without issues." | 12:28 |
| luis220413_ | https://wiki.ubuntu.com/Kernel/OEMKernel#Should_I_run_it_on_my_machine.3F | 12:28 |
| luis220413_ | These two commands monitor the repositories for 5.18 and 5.19 kernels. | 12:29 |
| === luis220413_ is now known as luis220413 | ||
| luis220413 | Ubuntu 22.04.2 (that should be released in February 2023) will ship with the 5.19 kernel: https://www.omgubuntu.co.uk/2022/08/ubuntu-22-04-point-release-available | 12:30 |
| bluesceada | yes I understand, but it's still maintained for security patches I hope | 12:30 |
| luis220413 | bluesceada: It is supported according to the "Security fixes" section in the first link I posted now | 12:30 |
| bluesceada | is there any list of those supported OEM hardware? if those are purely Intel systems for example, I need to be more careful I guess | 12:30 |
| bluesceada | luis220413, thanks. OK, let's just see, I just rebooted this kernel | 12:31 |
| bluesceada | and set to blank after 10min again | 12:31 |
| luis220413 | https://ubuntu.com/certified | 12:31 |
| luis220413 | This is a list of hardware certified by Canonical. | 12:32 |
| bluesceada | at least some AMD notebooks, ok | 12:32 |
| bluesceada | thanks | 12:32 |
| luis220413 | Check that they were certified for your Ubuntu version. | 12:32 |
| luis220413 | bluesceada: Is your specific desktop/laptop certified? | 12:33 |
| luis220413 | If you encounter any issues with this kernel, I recommend that you file a bug with the command ubuntu-bug linux | 12:34 |
| bluesceada | ok | 12:45 |
| bluesceada | <luis220413> bluesceada: Is your specific desktop/laptop certified? ---> I don't think so. It is a custom built desktop | 12:45 |
| luis220413 | bluesceada: Then you may encounter issues, as with any kernel on any system. | 12:46 |
| bluesceada | Of course | 12:46 |
| luis220413 | Have you filed a bug for the 5.15 kernel in Ubuntu? | 12:46 |
| bluesceada | No | 12:46 |
| bluesceada | To be it looks like it is handled upstream and it's the AMD maintainers job to get it into the 5.15 vanilla LTS kernel (kernel.org kernel) | 12:47 |
| luis220413 | I recommend that you file a bug by rebooting into the 5.15 kernel (through "Advanced options for Ubuntu") and using ubuntu-bug linux | 12:47 |
| bluesceada | but maybe he doesn't care enough | 12:47 |
| luis220413 | This is a severe issue and I strongly recommend that you file a bug both in Ubuntu and upstream, *if those bugs do not exist*. | 12:48 |
| bluesceada | they do exist | 12:48 |
| luis220413 | Could you provide links here? | 12:48 |
| bluesceada | several critical amdgpu bugs exist, I am pretty annoyed by AMD's graphics division recently | 12:48 |
| bluesceada | I did, they are freedesktop bugs, so not sure if there are respective kernel bugs ....Wait, here in summary: | 12:49 |
| bluesceada | [1] https://gitlab.freedesktop.org/drm/amd/-/issues/1819 [2] https://gitlab.freedesktop.org/drm/amd/-/issues/1871 [3] https://gitlab.freedesktop.org/drm/amd/-/issues/1887 | 12:49 |
| ubottu | Issue 1819 in drm/amd "RX 6600 XT fails to resume from suspend in 5.15.y series" [Opened] | 12:49 |
| ubottu | Issue 1871 in drm/amd "Random system reboot on 5.15+ and no video out until a full shutdown and start up" [Opened] | 12:50 |
| ubottu | Issue 1887 in drm/amd "[drm:dc_dmub_srv_wait_idle [amdgpu]] *ERROR* Error waiting for DMUB idle: status=3 log spam on 5.16.3" [Opened] | 12:50 |
| luis220413 | Bugs in freedesktop.org GitLab are not bugs in the kernel. You should find them in the Kernel Bugzilla (https://bugzilla.kernel.org) and in the bug pages for the 5.15 kernels: | 12:50 |
| bluesceada | I understand, but I might not care enough if the solution is simply to go with a more recent kernel | 12:51 |
| luis220413 | https://bugs.launchpad.net/ubuntu/+source/linux | 12:51 |
| luis220413 | https://bugs.launchpad.net/ubuntu/+source/linux-hwe-5.15 | 12:51 |
| bluesceada | ubuntu should just go as they did in the past and use more recent kernels for HWE - imho | 12:53 |
| bluesceada | after all, it is relatively recent hardware, so that's the main "issue" | 12:54 |
| bluesceada | But maybe manpower to maintain that is missing in Ubuntu, I understand | 12:54 |
| bluesceada | let me see a while how the oem kernel peforms and then I can report a bug on https://bugs.launchpad.net/ubuntu/+source/linux-hwe-5.15 | 12:58 |
| luis220413 | That link is for reporting bugs in the 5.15 kernel in Ubuntu *20.04* LTS. | 13:05 |
| luis220413 | You should report the bug on the previous link. | 13:05 |
| luis220413 | bluesceada: The 5.19 kernel should be available in the Ubuntu 22.04 repositories in October, by looking at the publishing history of the first HWE kernel for 20.04: https://launchpad.net/ubuntu/+source/linux-hwe-5.8/+publishinghistory | 13:09 |
| luis220413 | Manpower is not missing in Ubuntu to maintain the kernels. Ubuntu has a kernel release schedule: https://ubuntu.com/kernel/lifecycle | 13:11 |
| bluesceada | OK I guess I was just lucky before | 13:12 |
| bluesceada | luis220413, did 20.04 also start with a 5.8 kernel while the hwe kernel was also 5.8 ? | 13:14 |
| luis220413 | A release never starts with an HWE kernel. | 13:15 |
| luis220413 | The initial kernel for 20.04 was 5.4, according to that page. | 13:15 |
| bluesceada | OK, anyway, thanks for all the information, I will just see what's going on with the OEM kernel, report a bug to get the information out, worst-case get a more recent kernel from somewhere else until october | 13:18 |
| luis220413 | You can upgrade to the development release with do-release-upgrade -d. However, it is unstable and new features may be decided until Thursday. | 13:22 |
| luis220413 | Once in the development release, you need to manually install the 5.18 or 5.19 (recommended) kernel. | 13:23 |
| luis220413 | The schedule has a few errors. I reported them in https://github.com/canonical/ubuntu.com/issues/11964 | 13:36 |
| ubottu | Issue 11964 in canonical/ubuntu.com "Wrong and missing versions in https://ubuntu.com/kernel/lifecycle" [Open] | 13:36 |
| bluesceada | Sorry no interest in an entire development release | 13:55 |
| bluesceada | I might use the dev repo to get the kernel out of that however | 13:55 |
| bluesceada | OK, screen blank/suspend seems to work with the 5.17 oem kernel | 15:13 |
| bluesceada | luis220413, https://bugs.launchpad.net/ubuntu/+source/linux-hwe-5.15/+bug/1987413 | 20:31 |
| ubottu | Launchpad bug 1987413 in linux-hwe-5.15 (Ubuntu) "[amdgpu] Several stability bugs in pre-5.17 kernels, e.g. on AMD RX 6600 XT" [Undecided, New] | 20:31 |
| bluesceada | just FYI | 20:31 |
| luis220413 | bluesceada: You filed the bug against the wrong kernel (the 5.15 kernel in Ubuntu *20.04* LTS). | 20:32 |
| bluesceada | i thought I filed it for "linux" only, somehow it went into linux-hwe-5.15 | 20:32 |
| luis220413 | Paste here the output of lsb_release -a | 20:32 |
| bluesceada | it is also true for that one anyway | 20:32 |
| luis220413 | bluesceada: I will do a debdiff between linux-hwe-5.15 and linux. | 20:33 |
| bluesceada | i have the bug since 5.15 hwe in 20.04 anyway, and it just still persists | 20:33 |
| luis220413 | Thanks for the information! | 20:34 |
| luis220413 | I added a bugtask for linux (Ubuntu) because this affects the linux source package in Ubuntu 22.04 (that is, the 5.15 kernel). | 20:35 |
| bluesceada | thanks | 20:43 |
| luis220413 | I corrected two major errors in the description. | 20:46 |
| luis220413 | I have subscribed to the bug report as well. | 20:47 |
| bluesceada | major errors? which were? | 20:55 |
| bluesceada | ah the HWE kernel, ok hx | 20:55 |
| bluesceada | thx | 20:55 |
| bluesceada | but you know there is a package "linux-generic-hwe-22.04" | 20:57 |
| luis220413 | That package is a metapackage that installs the GA kernel when an HWE kernel is not available, and the HWE kernel when it is available. | 20:57 |
| bluesceada | so not sure why you say there is no HWE ... okay it might be identical with 20.04 - but the package named after 22.04 exists | 20:57 |
| luis220413 | s/available/released/ | 20:57 |
| luis220413 | bluesceada: See https://ubuntu.com/kernel/lifecycle | 20:58 |
| bluesceada | I get it :D ... but from looking at the packages only, one gets the impression there is a hwe kernel | 20:59 |
| bluesceada | i added a kern.log of one of the crashes that I managed to get out of using sysrq | 21:05 |
| luis220413 | You should reboot into the buggy kernel in recovery mode and run apport-collect 1987413 | 21:13 |
| luis220413 | Through the option "Advanced options for Ubuntu" in the boot menu | 21:13 |
| bluesceada | yes when I have time for that... | 21:31 |
| luis220413 | You have 60 days until the bug is automatically expired (closed), but you should this as soon as possible because this is a critical issue. | 21:36 |
| luis220413 | *should do this | 21:36 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!