Description of problem: Users asking for kernel 6.12 I have been running them Giuseppe already provide in backport testing for a while. Seems to work perfectly minus it need a workaround for running VirtualBox - the same that is needed in Cauldron: Bug 34408 - VirtualBox borked with newer kernels - VMX Root Mode Currently I use kernel-stable-desktop-6.12.41-1.stable.mga9-1-1.mga9 on my workstation, and also ASUS G75V laptop. Machine details in https://bugs.mageia.org/show_bug.cgi?id=34529#c21 and c23. VirtualBox host on both, nvidia470 on the laptop, suspend-resume, hibernate-restore - no glitches noted. I think we can QA this! I think we can release it as backport despite Bug 34408, if we note it somewhere, and I can add it to errata9, poining at an entry I will make for backport kernel in https://wiki.mageia.org/en/Kernel_flavours. How to install it, and the current quirks regarding VirtualBox and the following: A couple tools need be adjusted, it seems: § rok (remove-old-kernels) need be fixed to detect and handle those kernels (probably also for cauldron/mga10!), but it can be handled in errata9 and that /Kernel_flavours entry until fixed. § drakrpm go bananas if trying to install kernel-stable-userspace-headers; wanting to uninstall gcc, glibc etc.... What we need to install seems to be only two packages: kernel-stable-desktop-6.12... kernel-stable-desktop-devel-6.12... And optionally the corresponding -latest virtual packages, automatically selected. The latest non-backport lib64bpf1, cpupower, kernel-userspace-headers should do - and that is also in order to not break using 6.6 kernels. Remember to that before starting a VirtualBox guest, you need to $ sudo rmmod kvm_intel $ sudo rmmod kvm per https://bugs.mageia.org/show_bug.cgi?id=34408#c3 --- One interesting trick to test is to install 6.12 kernel on a mga9 live with persistence. Then that stick may be booted on hardware that need kernel 6.12, and can also be used to install that mga9 system to such hardware.
AFAIK the backport is not to be made in i586
CC: (none) => kernel, qa-bugs
CC tools maintainers for the problems when attempting to install kernel-stable-userspace-headers CC Barry for the rok script
CC: (none) => mageiatools, zen25000
CC: (none) => Joselp
This looks too complicated to pass to QA yet, especially: "A couple tools need be adjusted, it seems: § rok (remove-old-kernels) need be fixed to detect and handle those kernels (probably also for cauldron/mga10!), but it can be handled in errata9 and that /Kernel_flavours entry until fixed. § drakrpm go bananas if trying to install kernel-stable-userspace-headers; wanting to uninstall gcc, glibc etc.... [I do not understand this; can you clarify?]" Can we push this with these constraints?
See Also: (none) => https://bugs.mageia.org/show_bug.cgi?id=34408CC: (none) => lewyssmith
CC: (none) => mageia
(In reply to Lewis Smith from comment #3) > This looks too complicated to pass to QA yet, especially: > "A couple tools need be adjusted, it seems: ... > Can we push this with these constraints? I am thinking that if a user is skilled and interested enough to enable bakckport and install this kernel, the user can handle this. Especially if as now i have a thread on forum pointing here, add ti errata9, and that wiki page on kernels. For rok, we lived many years without it. I do not think it misbehaves, except it do not see, count, and handle the 6.12 kernels. So like before, users need to uninstall old/unwanted kernels to keep the number down (especially if they have a separate /boot and it is is small.) IF the backport kernel and maybe an update to it consumes too much space. kernel-stable-userspace-headers is not needed to install. And it is not auto-selected.
Created attachment 15074 [details] Telling drakrpm to install kernel-stable-userspace-headers (In reply to Lewis Smith from comment #3) > § drakrpm go bananas if trying to install kernel-stable-userspace-headers; > wanting to uninstall gcc, glibc etc.... > [I do not understand this; can you clarify?]" Attaching the message here And then also what urpmi say - wrong in other ways
Created attachment 15075 [details] Telling urpmi to install kernel-stable-userspace-headers
(In reply to Morgan Leijström from comment #1) > AFAIK the backport is not to be made in i586 yes, -desktop flavour (no server) only and x86_64 only.
CC: (none) => ghibomgx
In reply to Morgan Leijström from comment #4) > For rok, we lived many years without it. I do not think it misbehaves, > except it do not see, count, and handle the 6.12 kernels. So like before, > users need to uninstall old/unwanted kernels to keep the number down Fair enough, so long as it is documented (you probably have). > kernel-stable-userspace-headers is not needed to install. > And it is not auto-selected. So are comments 5 & 6 to advise users *not* to try installing 'kernel-stable-userspace-headers'? Again to be documented - but you have probably done that. > § drakrpm go bananas if trying to install kernel-stable-userspace-headers; > wanting to uninstall gcc, glibc etc.... I see now! I leave it to you to assign to QA, but it will need the package list - and why kernel 6.12 is necessary at all. Ideally the install would have a 'to read' section with at least: - need to install only two packages (from Backports): kernel-stable-desktop-6.12... kernel-stable-desktop-devel-6.12... - The latest non-backport lib64bpf1, cpupower, kernel-userspace-headers should do - Bug 34408 - VirtualBox borked with newer kernels , but easy avoidance: $ sudo rmmod kvm_intel $ sudo rmmod kvm - Do not try to install 'kernel-stable-userspace-headers' (re this bug). - desktop flavour (no server) only and x86_64 only
(In reply to Morgan Leijström from comment #4) > (In reply to Lewis Smith from comment #3) > > This looks too complicated to pass to QA yet, especially: > > "A couple tools need be adjusted, it seems: > ... > > Can we push this with these constraints? > > I am thinking that if a user is skilled and interested enough to enable > bakckport and install this kernel, the user can handle this. Especially if > as now i have a thread on forum pointing here, add ti errata9, and that wiki > page on kernels. > > For rok, we lived many years without it. I do not think it misbehaves, > except it do not see, count, and handle the 6.12 kernels. So like before, > users need to uninstall old/unwanted kernels to keep the number down > (especially if they have a separate /boot and it is is small.) IF the > backport kernel and maybe an update to it consumes too much space. > > kernel-stable-userspace-headers is not needed to install. > And it is not auto-selected. Anyway as for kernel-stable-userspace-headers, at the moment, i.e. for now, it was removed the production of kernel-stable-userspace-headers for version 6.12.41. I'm working to see if possible to fix that problem (but in a round after 6.6.101 validation). What you see now in the mirrors are files from previous releases, from 6.12.33 to 6.12.39, that should be cleaned by sysadmins). The lib64bpf1, cpupower is also not provided. Note that the kernel-stable in backports is in old naming scheme (as in mga8, nothing new from this point of view), where the version is included in the name. To install from command line without specifying the versions, it should be referred with the "-latest" suffix. E.g. "urpmi kernel-stable-desktop-latest kernel-stable-desktop-devel-latest" is the minimal command to install the kernel and devel stuff for building external dkms modules, if any. As for the rest, the installation would be sort of "transparent" and won't interfere with main kernel (6.6.101 at the moment).
In what case do a user need kernel-stable-userspace-headers? And can (when install problems are solved) it be installed alongside the normal 6.6 kernel kernel-userspace-headers ? - If not important for normal use, I think we can start QA this. ... Unless there is a reason to wait for another kernel version.
Summary: Backport kernel 6.12 => Backport kernel 6.12 - desktop x86_64 only
Morgan, I am leaving this with you to pass to QA when you see fit.
CC: lewyssmith => (none)
Lets see if Giuseppe solve the kernel-stable-userspace-headers install issue after 6.6.101 is shipped. And maybe another update to kernel, who knows ;-) Apart from that and wiki entry it seems ready to QA - I am using 6.12.41-desktop-1 on my production machine.
(In reply to Morgan Leijström from comment #10) > In what case do a user need kernel-stable-userspace-headers? Mostly for building applications. But anyway there is the standard headers package. > > And can (when install problems are solved) it be installed alongside the > normal 6.6 kernel kernel-userspace-headers ? No, while kernel-desktop and kernel-stable-desktop can be installed alongside each other, only one of kernel-stable-userspace-headers or kernel-userspace-headers is expected to be installed at a time.
Wiki entries: https://wiki.mageia.org/en/Kernel_flavours#Backport_kernels And also https://wiki.mageia.org/en/Mageia_9_Errata#Backport_kernel pointing to the one above. I think this kernel is usable as is for inclined users. Giuseppe, I think that if there is no update soon to kernel version or installable kernel-stable-userspace-headers around the corner, we can send this as is to QA now?
At https://wiki.mageia.org/en/Kernel_flavours#Backport_kernels corrected per Comment 13, and added instruction to fix rok per per https://bugs.mageia.org/show_bug.cgi?id=34560#c8 + Kicking the ball rolling For QA testing, read that wiki section!
Assignee: bugsquad => qa-bugsURL: (none) => https://wiki.mageia.org/en/Kernel_flavours#Backport_kernels
My dear friends, Next time you assign such thing to QA, plse remember that QA-ers are not developers, so a little guidance would have been in place. Anyway, here's what I did to test this backport. Installed in a separate partition a completely fresh M9 clasical installer from aug. 2023. Once there, used MCC to define the wifi internet access, no problem. Used MCC to get the repos active, disabled all Update ones to be sure I didn't pull in anything I didn't want at that stage, activate Core Backports Testing. And here my problems start: there are something like 6 different versions in Backports, which one have you been talking about here??? Went for the latest on (41) and tried to select the packages in MCC (yes, I want to seen what I'm doing, not being surprised by som quickly passing event in the CLI). I could select kernel-stable-desktop and its "latest" dummy, kernel-stable desktop-development and its latest, kernel-stable-docs. There was no 41 for kernel-stable-userspace-headers, only a 39. Trying to select that one would throw out a load as indicated in the attachment above, so I did not select that one. On submitting I then saw that the developement packs where not included in the list of packages to be installed, despite having been selected before. Continued installation without problems. closed MCC and rebooted. Checked in grub that the 41 kernel was being selected, booting went on till after 66 seconds (thats quite normal time lapse on this old laptop) I got the last messages: Startingplymouth-quit.service netfiler PSD loaded - (c) astero AG IFWlog: register target and then nothing... Had the idea of plugging in the Ethernet connection and got at second 227 r8169 000:04:00.0 enp4s0: Link is up etc... net-fw dropIN=enps4s0 OUT etc.... And that last line was repeated with increasing intervals untill; the 7th at second 317, and then finally nothing. Up to you now to point where I went wrong...
CC: (none) => herman.viaene
Sorry, forgot to say it is the latest, 6.12.41-1, we are testing. For guidance I did tell to read the wiki when handing this to QA in comment 15. (https://wiki.mageia.org/en/Kernel_flavours#Backport_kernels) That is: Only install kernel-stable-desktop-6.12... kernel-stable-desktop-devel-6.12... They will also pull in the -latest packages. I believe the install went OK, so setting feedback for the problem. Please describe the machine i.e output from inxi -SMCGN @Giuseppe - any idea? BTW please remove the old testing kernels from repo.
Summary: Backport kernel 6.12 - desktop x86_64 only => Backport kernel 6.12.41 - desktop x86_64 onlyKeywords: (none) => feedback
We forgot the good old files list. Also the 6.12.39 stuff should be cleaned from repo, and shouldn't be installed even if it provides packages not included in 6.12.41. Anyway a 6.12.42 shouldn't be too far.
I suggest we pause QA until 6.12.42 then. Please when it is ready, provide files list and set to QA.
Keywords: feedback => (none)Assignee: qa-bugs => kernel
From a working M9: # inxi -SMCGN System: Host: myname Kernel: 6.6.101-server-1.mga9 arch: x86_64 bits: 64 Console: pty pts/0 Distro: Mageia 9 Machine: Type: Laptop System: Hewlett-Packard product: Compaq 15 Notebook PC v: 0975100000405F00001620180 serial: CND41217KS Mobo: Hewlett-Packard model: 21F7 v: 52.16 serial: PEBRUG2WV6H4AI UEFI: Insyde v: F.05 date: 01/03/2014 CPU: Info: dual core model: AMD E1-2100 APU with Radeon HD Graphics bits: 64 type: MCP cache: L2: 1024 KiB Speed (MHz): avg: 800 min/max: 800/1000 cores: 1: 800 2: 800 Graphics: Device-1: Advanced Micro Devices [AMD/ATI] Kabini [Radeon HD 8210] driver: amdgpu v: kernel Device-2: Chicony HP Truevision HD camera driver: uvcvideo type: USB Display: server: X.Org v: 22.1.9 with: Xwayland v: 22.1.9 driver: X: loaded: amdgpu,v4l dri: radeonsi gpu: amdgpu resolution: 1366x768~60Hz API: OpenGL v: 4.6 vendor: amd mesa v: 25.0.7 renderer: AMD Radeon HD 8210 (radeonsi kabini ACO DRM 3.54 6.6.101-server-1.mga9) API: EGL Message: EGL data requires eglinfo. Check --recommends. Network: Device-1: Qualcomm Atheros QCA9565 / AR9565 Wireless Network Adapter driver: ath9k Device-2: Realtek RTL810xE PCI Express Fast Ethernet driver: r8169
Now using Backport kernel desktop 6.12.43-1 on my workstation Works well using drakrpm; I could select it and the -devel as an update (had 6.12.41 installed), and also I selected kernel-stable-userspace-headers, and it responded it need to uninstall kernel-userspace-headers, and it all worked. Note: for ...userspace-headers logic to work I had first updated from updates_testing: - mageia-repos-9-4.mga9.x86_64 - mageia-repos-keys-9-4.mga9.noarch - mageia-repos-pkgprefs-9-4.mga9.noarch - meta-task-9-4.mga9.noarch [morgan@svarten ~]$ rpm -qa | grep ^kernel-sta kernel-stable-desktop-devel-6.12.41-1.stable.mga9-1-1.mga9 kernel-stable-desktop-6.12.41-1.stable.mga9-1-1.mga9 kernel-stable-desktop-devel-6.12.43-1.stable.mga9-1-1.mga9 kernel-stable-desktop-devel-latest-6.12.43-1.stable.mga9 kernel-stable-desktop-latest-6.12.43-1.stable.mga9 kernel-stable-desktop-6.12.43-1.stable.mga9-1-1.mga9 kernel-stable-userspace-headers-6.12.43-1.stable.mga9 I presume the fix is in so drakboot etc works again https://bugs.mageia.org/show_bug.cgi?id=34563#c25. As before as described in wiki user need some trivial manual fixing for VirtualBox and rok. Giuseppe, is it ready for QA? Then make a files list and set to QA :-)
Summary: Backport kernel 6.12.41 - desktop x86_64 only => Backport kernel 6.12.43 - desktop x86_64 only
Updated https://wiki.mageia.org/en/Kernel_flavours#Backport_kernels
(In reply to Morgan Leijström from comment #22) > Updated https://wiki.mageia.org/en/Kernel_flavours#Backport_kernels A few notes for that page, kernel-stablenew is substatially obsoleted by kernel-stable-testing, and will be removed/obsoleted to leave kernel-stable-testing; as for version now mainline is at 6.17.0-rc3, stable-testing at 6.16.2|3, stable at 6.15.11, kernel at 6.12.43.
(In reply to Morgan Leijström from comment #21) > Now using Backport kernel desktop 6.12.43-1 on my workstation > > Works well using drakrpm; I could select it and the -devel as an update (had > 6.12.41 installed), and also I selected kernel-stable-userspace-headers, and > it responded it need to uninstall kernel-userspace-headers, and it all > worked. > > Note: for ...userspace-headers logic to work I had first updated from > updates_testing: > - mageia-repos-9-4.mga9.x86_64 > - mageia-repos-keys-9-4.mga9.noarch > - mageia-repos-pkgprefs-9-4.mga9.noarch > - meta-task-9-4.mga9.noarch exactly. Now the logic is: urpmi kernel-stable-userspace-headers will uninstall kernel-userspace-headers and install it; then from there you can go back to kernel-userspace-headers just installing it again and will remove kernel-stable-userspace-headers. The default for "auto" is kernel-userspace-headers, so packages requiring it, or the buildsystem, will use the "standard" version. There is still a further little fix in the same subject which will be included in 6.6.103.
(In reply to Morgan Leijström from comment #21) > I presume the fix is in so drakboot etc works again > https://bugs.mageia.org/show_bug.cgi?id=34563#c25. > > As before as described in wiki user need some trivial manual fixing for > VirtualBox and rok. Are you refering to the kvm|kvm_intel|kvm_amd kernel modules? > > Giuseppe, is it ready for QA? It will be the 6.12.44 for that.
(In reply to Giuseppe Ghibò from comment #23) > (In reply to Morgan Leijström from comment #22) > > A few notes for that page, kernel-stablenew is substatially obsoleted by > kernel-stable-testing, and will be removed/obsoleted to leave > kernel-stable-testing; as for version now mainline is at 6.17.0-rc3, > stable-testing at 6.16.2|3, stable at 6.15.11, kernel at 6.12.43. So like this? https://wiki.mageia.org/en/Kernel_flavours#Kernel_series_in_Cauldron I let -stable-testing inherit the description of -stablenew.
(In reply to Giuseppe Ghibò from comment #25) > (In reply to Morgan Leijström from comment #21) ... > > As before as described in wiki user need some trivial manual fixing for > > VirtualBox and rok. > > Are you refering to the kvm|kvm_intel|kvm_amd kernel modules? Yes. Ah, so on a AMD machine you need to rmmod kvm_amd instead of for intel rmmod kvm_intel ? > > Giuseppe, is it ready for QA? > > It will be the 6.12.44 for that. OK :-)
(In reply to Giuseppe Ghibò from comment #24) > Now the logic is: > > urpmi kernel-stable-userspace-headers will uninstall > kernel-userspace-headers and install it; Works with drakrpm > then from there you can go back to > kernel-userspace-headers just installing it again and will remove > kernel-stable-userspace-headers. Unfortunately, that does not work: drakrpm do not recognise the conflict when selecting kernel-userspace-headers, but complain and refuse when trying to execute. And trying to deselect kernel-stable-userspace-headers, it wants to uninstall gcc, glibc, kernels, dkms, ... Same goes for sudo urpme --test kernel-stable-userspace-headers > The default for "auto" is > kernel-userspace-headers, so packages requiring it, or the buildsystem, will > use the "standard" version. Sounds good, I did not test. > There is still a further little fix in the same subject which will be > included in 6.6.103. For the above problem?
> Ah, so on a AMD machine you need to > rmmod kvm_amd > instead of for intel > rmmod kvm_intel > ? > > Exactly. But on vb 7.1.12 (actually in cauldron) they are removed automatically at vbox startup if detected (it will be prompted for the root pwd). Instead with virtualbox-kvm they can leave modprobed, and would work side by side with qemu (with kvm acceleration). Maybe worthwhile to add to vbox-kvm to backports. Only limit, as you already seen, is a CPU with the XSAVE (IIRC) instruction set supported, so SandyBridge (2011) and beyond for Intel, Bulldozer and beyond (2011) for AMD, or also any x86_64-v3.
(In reply to Morgan Leijström from comment #28) > (In reply to Giuseppe Ghibò from comment #24) > > > Now the logic is: > > > > urpmi kernel-stable-userspace-headers will uninstall > > kernel-userspace-headers and install it; > > Works with drakrpm > > > then from there you can go back to > > kernel-userspace-headers just installing it again and will remove > > kernel-stable-userspace-headers. > > Unfortunately, that does not work: > > drakrpm do not recognise the conflict when selecting > kernel-userspace-headers, but complain and refuse when trying to execute. > > And trying to deselect kernel-stable-userspace-headers, it wants to > uninstall gcc, glibc, kernels, dkms, ... I was referring on installing or passing from one to another. Of course uninstalling, if you have it installed, it would should act exactly as you were uninstalling plain kernel-userspace-headers, so I think it's normal it would also remove compilers and development kits that depends on it. > > Same goes for > sudo urpme --test kernel-stable-userspace-headers > > > > The default for "auto" is > > kernel-userspace-headers, so packages requiring it, or the buildsystem, will > > use the "standard" version. > > Sounds good, I did not test. > > > > There is still a further little fix in the same subject which will be > > included in 6.6.103. > > For the above problem? Yes.
(In reply to Morgan Leijström from comment #26) > (In reply to Giuseppe Ghibò from comment #23) > > (In reply to Morgan Leijström from comment #22) > > > > A few notes for that page, kernel-stablenew is substatially obsoleted by > > kernel-stable-testing, and will be removed/obsoleted to leave > > kernel-stable-testing; as for version now mainline is at 6.17.0-rc3, > > stable-testing at 6.16.2|3, stable at 6.15.11, kernel at 6.12.43. > > So like this? > https://wiki.mageia.org/en/Kernel_flavours#Kernel_series_in_Cauldron > yes.
(In reply to Giuseppe Ghibò from comment #30) > (In reply to Morgan Leijström from comment #28) ... > I was referring on installing or passing from one to another. Of course > uninstalling, if you have it installed, it would should act exactly as you > were uninstalling plain kernel-userspace-headers, so I think it's normal it > would also remove compilers and development kits that depends on it. Yes, but consider: User have kernel-stable-userspace-headers installed, In drakrpm selects also kernel-serspace-headers: -> no warning message, both re now selected. User deselect kernel-stable-userspace-headers, (kernel-userspace-headers still selected) -> drakrpm wants to uninstall gcc, kernels, etc... > > Same goes for > > sudo urpme --test kernel-stable-userspace-headers Better example: [morgan@svarten ~]$ LC_ALL=C sudo urpmi --test kernel-userspace-headers Package kernel-stable-userspace-headers-6.12.43-1.stable.mga9.x86_64 is already installed installing kernel-userspace-headers-6.6.101-1.mga9.x86_64.rpm from /var/cache/urpmi/rpms Installation failed: kernel-userspace-headers conflicts with (installed) kernel-stable-userspace-headers-6.12.43-1.stable.mga9.x86_64 linux-userspace-headers < 6.12.43-1.stable.mga9 conflicts with (installed) kernel-stable-userspace-headers-6.12.43-1.stable.mga9.x86_64 How is user supposed to switch from kernel-stable-userspace-headers to kernel-userspace-headers ?
(In reply to Morgan Leijström from comment #32) > (In reply to Giuseppe Ghibò from comment #30) > > (In reply to Morgan Leijström from comment #28) > ... > > I was referring on installing or passing from one to another. Of course > > uninstalling, if you have it installed, it would should act exactly as you > > were uninstalling plain kernel-userspace-headers, so I think it's normal it > > would also remove compilers and development kits that depends on it. > > Yes, but consider: > > User have kernel-stable-userspace-headers installed, > In drakrpm selects also kernel-serspace-headers: > -> no warning message, both re now selected. > User deselect kernel-stable-userspace-headers, > (kernel-userspace-headers still selected) > -> drakrpm wants to uninstall gcc, kernels, etc... > > > > > Same goes for > > > sudo urpme --test kernel-stable-userspace-headers > > Better example: > > [morgan@svarten ~]$ LC_ALL=C sudo urpmi --test kernel-userspace-headers > Package kernel-stable-userspace-headers-6.12.43-1.stable.mga9.x86_64 is > already installed > > > installing kernel-userspace-headers-6.6.101-1.mga9.x86_64.rpm from > /var/cache/urpmi/rpms > Installation failed: kernel-userspace-headers conflicts with (installed) > kernel-stable-userspace-headers-6.12.43-1.stable.mga9.x86_64 > linux-userspace-headers < 6.12.43-1.stable.mga9 conflicts with > (installed) kernel-stable-userspace-headers-6.12.43-1.stable.mga9.x86_64 > > How is user supposed to switch from kernel-stable-userspace-headers to > kernel-userspace-headers ? You're right, seems still wrong
I see 6.12.44-1 is in backport testing now. And 6.6.105 in updates testing, you are busy :) I note a difference in how drakrpm handle the updates of 6.6 and 6.12 kernels, regarding needing kernel* -latest packages. In drakrpm, the 6.6 kernel and kernel-devel updates shows up as updates regardless of weather any -latest package is installed or not. And they also is handled via update applet without -latest packages installed. The backport packages behaves differently: Neither kernel-stable-desktop nor kernel-stable-desktop-devel shows up when in drakrpm dropdown selecting all updates. kernel-stable-desktop-devel-latest and kernel-stable-desktop-latest shows up, and when selecting them, drakrpm pops up dialogue to accept kernel-stable-desktop and kernel-stable-desktop-devel too.
(In reply to Morgan Leijström from comment #34) > I see 6.12.44-1 is in backport testing now. > And 6.6.105 in updates testing, you are busy :) yep, before a pause... :) 6.6.103 should be completed now, it will be opened a BR for 6.6.103. There is also virtualbox 7.1.12-1.mga9 in updates_testing, but actually the kmods were built for 7.1.10, so to avoid problems with newer kernel, let's push 6.6.103 before and vb later (actually 7.1.12 can be used but the vbox modules should be built with dkms-virtualbox). with the bundle 6.6.103/6.12.44 the kernel-userspace-headers|kernel-stable-userspace-headers should be working now, as a toggle. > > I note a difference in how drakrpm handle the updates of 6.6 and 6.12 > kernels, regarding needing kernel* -latest packages. > > In drakrpm, the 6.6 kernel and kernel-devel updates shows up as updates > regardless of weather any -latest package is installed or not. > And they also is handled via update applet without -latest packages > installed. > > The backport packages behaves differently: > > Neither kernel-stable-desktop nor kernel-stable-desktop-devel shows up when > in drakrpm dropdown selecting all updates. > kernel-stable-desktop-devel-latest and kernel-stable-desktop-latest shows > up, and when selecting them, drakrpm pops up dialogue to accept > kernel-stable-desktop and kernel-stable-desktop-devel too. Good observation. However, if you were referring to something new from 6.6.101|6.6.41 towards 6.6.103|6.12.44 there was no change in this subject. If you are referring to the general case, yes, the naming scheme differs. kernel-stable in backport is in old naming versioned scheme (like actually in cauldron and mga8), so the version is "embedded" in the package name, and here kernel-<series>-desktop-latest is required to point to the latest version, while in kernel (standard) package the version is not "embedded" in the name (at least not yet...), and the version is like any other regular package.
Great. Yeah no news really about drakrpm. A user interaction quirk by design (I suppose), is that when you have say kernel-stable-userspace-headers installed and select kernel-userspace-headers it correctly pops up message and user confirm to uninstall kernel-stable-userspace-headers, still drakrpm show a checkmark to keep it installed. And if user tries (logically correctly) to uncheck it (keeping the checkmark on kernel-userspace-headers, drakrpm pops up a message it then need to install that great bunch of gcc etc... Confusing. User should just sit still in the boat leaving both selected, drakrpm does what it said in the popup, not as it shows in the selection... Anyhow, I now updated https://wiki.mageia.org/en/Kernel_flavours#Backport_kernels , saying userspace-headers flavours are now handled correctly (neglecting the user interaction ambiguity I reported above), and a note on -latest packages. I now use 6.12.44-1 for work. May have time to test 6.6.103 in the weekend. For both: If ready for QA, provide a files list and set to QA :-)
(In reply to Morgan Leijström from comment #36) > Great. > > Yeah no news really about drakrpm. > > A user interaction quirk by design (I suppose), is that when you have say > kernel-stable-userspace-headers installed and select > kernel-userspace-headers it correctly pops up message and user confirm to > uninstall kernel-stable-userspace-headers, still drakrpm show a checkmark to > keep it installed. > And if user tries (logically correctly) to uncheck it (keeping the > checkmark on kernel-userspace-headers, drakrpm pops up a message it then > need to install that great bunch of gcc etc... Confusing. > User should just sit still in the boat leaving both selected, drakrpm does > what it said in the popup, not as it shows in the selection... > > So, if I start from this position: a) kernel-userspace-headers installed with what it requires or might have required, e.g. gcc-devel, dkms, etc. b) then type "userspace-headers" in the rpmdrake "find" string form. Both kernel-userspace-headers and kernel-stable-userspace-headers appears, with the checkmark [v] on kernel-userspace-headers. c) then I click on kernel-stable-userspace-headers. It popup a requester saying "The following package has to be removed for others to be upgraded: kernel-userspace-headers-6.6.103-1.mga9.x86_64 (due to conflicts with linux-userspace-headers[> 6.6.103-1.mga9])" If I click "No", nothing happens. If I click "Yes", both are remained selected. Then I click "Apply", then kernel-userspace-headers is uninstalled and kernel-stable-userspace-headers installed (pretty quickly, no other packages are uninstalled). If was undecided and I unselect "kernel-stable-userspace-headers", then nothing happens, it just remove the checkmark from it. If I when both are selected, unselect "kernel-userspace-headers", and click "Apply", then it will ask to remove gcc and its toolchain, this because kernel-stable-userspace-headers is not (yet) installed, and gcc virtually depends still on the installed package which in this case is kernel-userspace-headers. Isn't the same "checkmark behaviour" as when you toggle the packages pulseaudio<->pipewire-pulseaudio (generating other side effects on sound, but that's another story, as it requires proper usage of draksound)? btw, speaking of rpmdrake I just noticed in cauldron that when I run rpmdrake as plain user, it prompts for a root password then it regularly starts, but if I start directly as root it ends up with error/crash. Do you get the same? In mga9 instead this won't happen.
(In reply to Giuseppe Ghibò from comment #37) > (In reply to Morgan Leijström from comment #36) ... > If I when both are selected, unselect "kernel-userspace-headers", and click > "Apply", then it will ask to remove gcc and its toolchain, this because > kernel-stable-userspace-headers is not (yet) installed, and gcc virtually > depends still on the installed package which in this case is > kernel-userspace-headers. Yes. The thing is that is shows it intend to install kernel-stable-userspace-headers, thus the dependencies should become satisfied. > Isn't the same "checkmark behaviour" as when you toggle the packages > pulseaudio<->pipewire-pulseaudio (generating other side effects on sound, > but that's another story, as it requires proper usage of draksound)? Possibly... I am used to drakrpm being a bit qirky changing packages. Reminds me, is it time to update about Bug 32626 in errata? And Bug 32582 Have you also seen that sometimes (seldom) after install, packages are erroneously getting listed in left pane of drakrpm (ther shul only be cathegories hierarcially), and that often coincided that a pop-up message tha tit do not find packages in repo, - i think it is the same packages it just did install. I just restart it and it is OK... > btw, speaking of rpmdrake I just noticed in cauldron that when I run > rpmdrake as plain user, it prompts for a root password then it regularly > starts, but if I start directly as root it ends up with error/crash. Do you > get the same? In mga9 instead this won't happen. Works for me too in mga9, I do not use Cauldron yet. Ask in dev list.
(In reply to Morgan Leijström from comment #38) > (In reply to Giuseppe Ghibò from comment #37) > > (In reply to Morgan Leijström from comment #36) > ... > > If I when both are selected, unselect "kernel-userspace-headers", and click > > "Apply", then it will ask to remove gcc and its toolchain, this because > > kernel-stable-userspace-headers is not (yet) installed, and gcc virtually > > depends still on the installed package which in this case is > > kernel-userspace-headers. > > Yes. The thing is that is shows it intend to install > kernel-stable-userspace-headers, thus the dependencies should become > satisfied. > > > > Isn't the same "checkmark behaviour" as when you toggle the packages > > pulseaudio<->pipewire-pulseaudio (generating other side effects on sound, > > but that's another story, as it requires proper usage of draksound)? > > Possibly... I am used to drakrpm being a bit qirky changing packages. BTW, the behaviour you are expecting with the checkmark "deferred" and mutually exclusive in both packages, you can achieve if you use dnfdragora.
(In reply to Morgan Leijström from comment #38) > (In reply to Giuseppe Ghibò from comment #37) > > (In reply to Morgan Leijström from comment #36) > ... > > If I when both are selected, unselect "kernel-userspace-headers", and click > > "Apply", then it will ask to remove gcc and its toolchain, this because > > kernel-stable-userspace-headers is not (yet) installed, and gcc virtually > > depends still on the installed package which in this case is > > kernel-userspace-headers. > > Yes. The thing is that is shows it intend to install > kernel-stable-userspace-headers, thus the dependencies should become > satisfied. > > > > Isn't the same "checkmark behaviour" as when you toggle the packages > > pulseaudio<->pipewire-pulseaudio (generating other side effects on sound, > > but that's another story, as it requires proper usage of draksound)? > > Possibly... I am used to drakrpm being a bit qirky changing packages. > > Reminds me, is it time to update about Bug 32626 in errata? > And Bug 32582 The scheme for main kernel (6.6.103) has not changed, so I think there were no changes for those two bugs. > > Have you also seen that sometimes (seldom) after install, packages are > erroneously getting listed in left pane of drakrpm (ther shul only be > cathegories hierarcially), and that often coincided that a pop-up message > tha tit do not find packages in repo, - i think it is the same packages it > just did install. I just restart it and it is OK... I think have not seen that, but you can take a screenshot. Do you think is that due to rpmdrake itself or some misbehaviour of the toolkit? The problem is that most of rpmdrake development of new features is freezed long ago, and we can also fix small things.
Summary: Backport kernel 6.12.43 - desktop x86_64 only => Backport kernel 6.12.44 - desktop x86_64 only
Created attachment 15089 [details] files list
I've updated the files list. For the original problem about kvm modules to be removed, you might try with vbox 7.1.12 (in updates testing too) where it prompts for root password for removing them on demand.
OK. Ready for QA?
Assignee: kernel => qa-bugs
Per Comment 24, I set this to depend on kernel 6.6.103 for *-userspace-headers switching. Likewise, I think we should open a bug on meta-task and mageia-repos, and set this bug to depend on that too.
Depends on: (none) => 34594
(In reply to Morgan Leijström from comment #44) > Per Comment 24, I set this to depend on kernel 6.6.103 for > *-userspace-headers switching. > > Likewise, I think we should open a bug on meta-task and mageia-repos, and > set this bug to depend on that too. You're right, however I added the two packages to the #34594, since it's also required there.
Now that is assigned to qa I remove some CCs, if the users want they can be added latter
CC: (none) => stephengermany
Took my courage in both hands from beneath my boots and installed the packages listed in the file list above from Core Backports Testing. That went apparently well. but rebooting is the same disaster as I described above in Comments 16 and 20.
For @Giuseppe: Comment 47, 16, 20
Keywords: (none) => feedback
Is it possible to make initrd less clumsy? 6.12.44 initrd is almost three times as big as 6.6.103 initrd -rw------- 1 root root 70218134 sep 1 23:26 initrd-6.12.44-desktop-1.stable.mga9.img -rw------- 1 root root 25225600 aug 31 09:47 initrd-6.6.103-desktop-1.mga9.img d! Some of us have a small separate /boot partition, rest of disk encrypted.
(In reply to Morgan Leijström from comment #49) > Is it possible to make initrd less clumsy? > 6.12.44 initrd is almost three times as big as 6.6.103 initrd > > > -rw------- 1 root root 70218134 sep 1 23:26 > initrd-6.12.44-desktop-1.stable.mga9.img > -rw------- 1 root root 25225600 aug 31 09:47 > initrd-6.6.103-desktop-1.mga9.img > d! > > Some of us have a small separate /boot partition, rest of disk encrypted. can't do much at the moment for size.
(In reply to Morgan Leijström from comment #48) > For @Giuseppe: Comment 47, 16, 20 Sorry, but I've not understand if the problem is about installing or running. For installing, let's not consider the header-userspace at the moment. The multiple kernels, some of them in backports from newer release should be cleaned, anyway there could be always multiple version once they will be approved. For installing with GUI, once already enabled the backports_testing repository is to search and select in the Find of rpmdrake: kernel-stable-desktop-latest (and if dkms are required) also kernel-stable-desktop-devel-latest, and it will retrieve also the latest kernel. Ditto for urpmi.
@Giuseppe, In my case it is the installation that hangs (left it there and went for lunch and a little siesta) and nothing can be done except power down.
Like with 6.6.103, gkrellm do not show network interfaces. See links at Bug 34594#c32 Bug 34594 Comment 32
(In reply to Herman Viaene from comment #52) > @Giuseppe, > In my case it is the installation that hangs (left it there and went for > lunch and a little siesta) and nothing can be done except power down. We need to figure out what/why is happening, as this is even before it boots with the new kernel. E.g. what was the current kernel from which you install that package? What was the situation of all the installed kernels and devels? E.g. rpm -qa | grep ^kernel |sort How much space you had on /boot partition?
I had the kernel 6.6.101 (latest at that time) kernel-desktop-6.4.9-4.mga9 kernel-desktop-6.6-101-1.mga9 kernel-desktop-latest-6.6-101-1.mga9 kernel-firmware-20250509-1.mga9 kernel-firmware-nonfree-20250509-1.mga9.nonfree /boot 128.5Mb used on 310Mb
Size of /boot is after installation of this update and restore the system from fsarchive
I must say that now revisiting 6.6 kernels for QA, i must say that my desktop workstation "svarten" really feel much snappier with 6.12 :) @Giuseppe, is still target this 6.12.44, or is there a 6.12.45 coming i.e with the fix for gkrellm?
(In reply to Herman Viaene from comment #52) > it is the installation that hangs (left it there and went for > lunch and a little siesta) and nothing can be done except power down. What tool did you then use for installataion? Does it work if you open a terminal and only use urpmi, to only install kernel-stable-desktop-6.12.44-1 ?
(In reply to Morgan Leijström from comment #57) > I must say that now revisiting 6.6 kernels for QA, i must say that my > desktop workstation "svarten" really feel much snappier with 6.12 :) > > @Giuseppe, is still target this 6.12.44, or is there a 6.12.45 coming i.e > with the fix for gkrellm? Probably worthwhile a 6.12.45 (similar to current cauldron version), but actually I've not time to work with it in the immediate.
On Comment 58: I used MCC to install the kernel then. Now: # urpmi kernel-stable-desktop-6.12.44-1.stable.mga9 To satisfy dependencies, the following packages are going to be installed: Package Version Release Arch (medium "Core Backports Testing (distrib9)") kernel-stable-desktop-6.12.44> 1 1.mga9 x86_64 kernel-stable-desktop-latest 6.12.44 1.stable.mga9 x86_64 (recommended) 126MB of additional disk space will be used. 115MB of packages will be retrieved. Proceed with the installation of the 2 packages? (Y/n) y $MIRRORLIST: media/core/backports_testing/kernel-stable-desktop-6.12.44-1.stable.mga9-1-1.mga9.x86_64.rpm $MIRRORLIST: media/core/backports_testing/kernel-stable-desktop-latest-6.12.44-1.stable.mga9.x86_64.rpm installing kernel-stable-desktop-latest-6.12.44-1.stable.mga9.x86_64.rpm kernel-stable-desktop-6.12.44-1.stable.mga9-1-1.mga9.x86_64.rpm from /var/cache/urpmi/rpms Preparing... ########################################################################### 1/2: kernel-stable-desktop-6.12.44-1.stable.mga9 ########################################################################### 2/2: kernel-stable-desktop-latest ########################################################################### remove-boot-splash: Format of /boot/initrd-6.12.44-desktop-1.stable.mga9.img not recognized You should restart your computer for kernel-stable-desktop-6.12.44-1.stable.mga9 As the one but last message indicates, rebooting has little sense, I guess. Doing it anyaway. So come back for the result.
Nope. Reboot hangs after plymouth-quit.service netfilter PSD-loaded - (c) astaro AG IFWLOG: register target. Hangs now for more than 1 hour.
I see the packages for i586 too, we keep this as i586 only?
RH x86_64 I create a repository with the packages LC_ALL=C urpmi kernel-stable-desktop kernel-stable-desktop-devel kernel-stable-userspace-headers The following package has to be removed for others to be upgraded: kernel-userspace-headers-6.6.105-1.mga9.x86_64 (due to conflicts with linux-userspace-headers[> 6.6.105-1.mga9]) (y/N) y To satisfy dependencies, the following packages are going to be installed: Package Version Release Arch (medium "QA") kernel-stable-desktop-6.12.44> 1 1.mga9 x86_64 kernel-stable-desktop-devel-6> 1 1.mga9 x86_64 kernel-stable-desktop-devel-l> 6.12.44 1.stable.mga9 x86_64 (recommended) kernel-stable-desktop-latest 6.12.44 1.stable.mga9 x86_64 (recommended) kernel-stable-userspace-heade> 6.12.44 1.stable.mga9 x86_64 221MB of additional disk space will be used. 136MB of packages will be retrieved. Proceed with the installation of the 5 packages? (Y/n) y installing kernel-stable-desktop-devel-6.12.44-1.stable.mga9-1-1.mga9.x86_64.rpm kernel-stable-desktop-latest-6.12.44-1.stable.mga9.x86_64.rpm kernel-stable-userspace-headers-6.12.44-1.stable.mga9.x86_64.rpm kernel-stable-desktop-6.12.44-1.stable.mga9-1-1.mga9.x86_64.rpm kernel-stable-desktop-devel-latest-6.12.44-1.stable.mga9.x86_64.rpm from //home/katnatek/qa-testing/x86_64 Preparing... ################################################################################################### 1/5: kernel-stable-desktop-6.12.44-1.stable.mga9 ################################################################################################### 2/5: kernel-stable-desktop-latest ################################################################################################### 3/5: kernel-stable-desktop-devel-latest ################################################################################################### 4/5: kernel-stable-desktop-devel-6.12.44-1.stable.mga9 ################################################################################################### 5/5: kernel-stable-userspace-headers ################################################################################################### removing package kernel-userspace-headers-6.6.105-1.mga9.x86_64 1/1: removing kernel-userspace-headers-6.6.105-1.mga9.x86_64 ################################################################################################### vhba (20240917-1bdk_mga9): Installing module. ................. ........ remove-boot-splash: Format of /boot/initrd-6.12.44-desktop-1.stable.mga9.img not recognized You should restart your computer for kernel-stable-desktop-6.12.44-1.stable.mga9 I don't know if with the state of my system I should report But I got I full freeze after resume a video from telegram desktop I reboot to 105 version to make the report, not find information in logs that could help
(In reply to katnatek from comment #62) > I see the packages for i586 too, we keep this as i586 only? i586 shouldn't be taken in consideration, only x86_64.
(In reply to katnatek from comment #63) > RH x86_64 > I create a repository with the packages > > > LC_ALL=C urpmi kernel-stable-desktop kernel-stable-desktop-devel > kernel-stable-userspace-headers > The following package has to be removed for others to be upgraded: > kernel-userspace-headers-6.6.105-1.mga9.x86_64 > (due to conflicts with linux-userspace-headers[> 6.6.105-1.mga9]) (y/N) y > To satisfy dependencies, the following packages are going to be installed: > Package Version Release Arch > (medium "QA") > kernel-stable-desktop-6.12.44> 1 1.mga9 x86_64 > kernel-stable-desktop-devel-6> 1 1.mga9 x86_64 > kernel-stable-desktop-devel-l> 6.12.44 1.stable.mga9 x86_64 > (recommended) > kernel-stable-desktop-latest 6.12.44 1.stable.mga9 x86_64 > (recommended) > kernel-stable-userspace-heade> 6.12.44 1.stable.mga9 x86_64 > 221MB of additional disk space will be used. > 136MB of packages will be retrieved. > Proceed with the installation of the 5 packages? (Y/n) y > The "exchange" of the headers packages is expected as there could be only one header package installed on the system, and not multiple ones. You might try also with --no-recommends. > > installing > kernel-stable-desktop-devel-6.12.44-1.stable.mga9-1-1.mga9.x86_64.rpm > kernel-stable-desktop-latest-6.12.44-1.stable.mga9.x86_64.rpm > kernel-stable-userspace-headers-6.12.44-1.stable.mga9.x86_64.rpm > kernel-stable-desktop-6.12.44-1.stable.mga9-1-1.mga9.x86_64.rpm > kernel-stable-desktop-devel-latest-6.12.44-1.stable.mga9.x86_64.rpm from > //home/katnatek/qa-testing/x86_64 > Preparing... > ############################################################################# > ###################### > 1/5: kernel-stable-desktop-6.12.44-1.stable.mga9 > > ############################################################################# > ###################### > 2/5: kernel-stable-desktop-latest > > ############################################################################# > ###################### > 3/5: kernel-stable-desktop-devel-latest > > ############################################################################# > ###################### > 4/5: kernel-stable-desktop-devel-6.12.44-1.stable.mga9 > > ############################################################################# > ###################### > 5/5: kernel-stable-userspace-headers > > ############################################################################# > ###################### > removing package kernel-userspace-headers-6.6.105-1.mga9.x86_64 > 1/1: removing kernel-userspace-headers-6.6.105-1.mga9.x86_64 > > ############################################################################# > ###################### > > vhba (20240917-1bdk_mga9): Installing module. > ................. > ........ > remove-boot-splash: Format of /boot/initrd-6.12.44-desktop-1.stable.mga9.img > not recognized > You should restart your computer for > kernel-stable-desktop-6.12.44-1.stable.mga9 > > I don't know if with the state of my system I should report > But I got I full freeze after resume a video from telegram desktop > > I reboot to 105 version to make the report, not find information in logs > that could help 6.12.44 could be affected also on some problems we have seen on .103 and .104 for the amdgpu could have fixed in 6.12.46.
OK, expecting 6.12.46 to show up soon then... One thing i have been noticing occasionally with 6.6.105 and 6.12.44 is that after resuming from suspend it is extremely sluggish for a minute. CPU usage is low, no core above 20% or so, but apps are nearly unresponsive. Mouse pointer moves OK. Just wait a minute and all is OK. There was some disk (SSD) activity but only a fraction of possible bandwidth. Like kernel just need to wake up slowly before it even want to *try* to do things seriously. Like myself. Maybe you could import the coffee lib? ;-)
(In reply to Giuseppe Ghibò from comment #65) > (In reply to katnatek from comment #63) > > > RH x86_64 > > I create a repository with the packages > > > > > > LC_ALL=C urpmi kernel-stable-desktop kernel-stable-desktop-devel > > kernel-stable-userspace-headers > > The following package has to be removed for others to be upgraded: > > kernel-userspace-headers-6.6.105-1.mga9.x86_64 > > (due to conflicts with linux-userspace-headers[> 6.6.105-1.mga9]) (y/N) y > > To satisfy dependencies, the following packages are going to be installed: > > Package Version Release Arch > > (medium "QA") > > kernel-stable-desktop-6.12.44> 1 1.mga9 x86_64 > > kernel-stable-desktop-devel-6> 1 1.mga9 x86_64 > > kernel-stable-desktop-devel-l> 6.12.44 1.stable.mga9 x86_64 > > (recommended) > > kernel-stable-desktop-latest 6.12.44 1.stable.mga9 x86_64 > > (recommended) > > kernel-stable-userspace-heade> 6.12.44 1.stable.mga9 x86_64 > > 221MB of additional disk space will be used. > > 136MB of packages will be retrieved. > > Proceed with the installation of the 5 packages? (Y/n) y > > > > The "exchange" of the headers packages is expected as there could be only > one header package installed on the system, and not multiple ones. You might > try also with --no-recommends. > > > > > > > installing > > kernel-stable-desktop-devel-6.12.44-1.stable.mga9-1-1.mga9.x86_64.rpm > > kernel-stable-desktop-latest-6.12.44-1.stable.mga9.x86_64.rpm > > kernel-stable-userspace-headers-6.12.44-1.stable.mga9.x86_64.rpm > > kernel-stable-desktop-6.12.44-1.stable.mga9-1-1.mga9.x86_64.rpm > > kernel-stable-desktop-devel-latest-6.12.44-1.stable.mga9.x86_64.rpm from > > //home/katnatek/qa-testing/x86_64 > > Preparing... > > ############################################################################# > > ###################### > > 1/5: kernel-stable-desktop-6.12.44-1.stable.mga9 > > > > ############################################################################# > > ###################### > > 2/5: kernel-stable-desktop-latest > > > > ############################################################################# > > ###################### > > 3/5: kernel-stable-desktop-devel-latest > > > > ############################################################################# > > ###################### > > 4/5: kernel-stable-desktop-devel-6.12.44-1.stable.mga9 > > > > ############################################################################# > > ###################### > > 5/5: kernel-stable-userspace-headers > > > > ############################################################################# > > ###################### > > removing package kernel-userspace-headers-6.6.105-1.mga9.x86_64 > > 1/1: removing kernel-userspace-headers-6.6.105-1.mga9.x86_64 > > > > ############################################################################# > > ###################### > > > > vhba (20240917-1bdk_mga9): Installing module. > > ................. > > ........ > > remove-boot-splash: Format of /boot/initrd-6.12.44-desktop-1.stable.mga9.img > > not recognized > > You should restart your computer for > > kernel-stable-desktop-6.12.44-1.stable.mga9 > > > > I don't know if with the state of my system I should report > > But I got I full freeze after resume a video from telegram desktop > > > > I reboot to 105 version to make the report, not find information in logs > > that could help > > 6.12.44 could be affected also on some problems we have seen on .103 and > .104 for the amdgpu could have fixed in 6.12.46. The freeze was with the internal card inxi -G Graphics: Device-1: Intel Core Processor Integrated Graphics driver: i915 v: kernel Display: x11 server: X.org v: 1.21.1.8 with: Xwayland v: 22.1.9 driver: X: loaded: intel unloaded: fbdev,modesetting,vesa dri: i965 gpu: i915 resolution: 1366x768~60Hz API: OpenGL v: 2.1 vendor: intel mesa v: 25.0.7 renderer: Mesa Intel HD Graphics (ILK) API: EGL Message: EGL data requires eglinfo. Check --recommends. But as my hdd is dying, I can't discard hdd as the cause :S
@Giuseppe: New version coming soon, or we ship this? @katnatek: You have a new rig to test this now?
(In reply to Morgan Leijström from comment #68) > @Giuseppe: New version coming soon, or we ship this? > > @katnatek: You have a new rig to test this now? Not new but I have almost working my system, just have some issues with my webpages but quite sure some configuration is missing, ig Guiseppe not send new version I'll check the current one to discard my hdd issue, but have a lot of work with icu bugs
I think 12.44 is now obsolete, so maybe worthwhile for waiting a newer one. Dunno yet when, as there will be a 6.6.x (113 or 114) before.
OK, changing to depend on kernel 6.6.116 And VirtualBox need to be updated for 6.12 kernels
See Also: https://bugs.mageia.org/show_bug.cgi?id=34408 => (none)Depends on: 34594 => 34713, 34408
I see kernel-stable-6.12.59-1.stable.mga9 got built. Ready for QA testing?
Summary: Backport kernel 6.12.44 - desktop x86_64 only => Backport kernel 6.12 - desktop x86_64 only
(In reply to Morgan Leijström from comment #72) > I see kernel-stable-6.12.59-1.stable.mga9 got built. > Ready for QA testing? There is 6.12.59-2.stable.mga9 for this.
Great :-) OK on ASUS Aspire A717, Plasma All official updates plus X11 from testing I let drakrpm select 6.12.59-2 kernel as an update, had 6.12.44 installed. Filesystems: separate ext4 /boot, then a LUKS encrypted pv for LVM, which provides /, swap /home suspend-resume hibernate-restore Surfing a bit incl internet video using wifi All OK, no problem noted. This laptop have dual GPU but we only use the integrated one, nvidia not configured, office usage. [kajsa@aspire ~]$ inxi -SMCGN System: Host: aspire Kernel: 6.12.59-desktop-2.stable.mga9 arch: x86_64 bits: 64 Desktop: KDE Plasma v: 5.27.10 Distro: Mageia 9 Machine: Type: Laptop System: Acer product: Aspire A717-71G v: V1.13 serial: <superuser required> Mobo: KBL model: Charizard_KLS v: V1.13 serial: <superuser required> UEFI: Insyde v: 1.13 date: 12/26/2017 CPU: Info: quad core model: Intel Core i5-7300HQ bits: 64 type: MCP cache: L2: 1024 KiB Speed (MHz): avg: 800 min/max: 800/3500 cores: 1: 800 2: 800 3: 800 4: 800 Graphics: Device-1: Intel HD Graphics 630 driver: i915 v: kernel Device-2: NVIDIA GP107M [GeForce GTX 1050 Mobile] driver: nouveau v: kernel Device-3: Chicony Integrated HD WebCam driver: uvcvideo type: USB Display: x11 server: X.org v: 1.21.1.21 with: Xwayland v: 22.1.9 driver: X: loaded: intel,v4l dri: i965 gpu: i915 resolution: 1920x1080~60Hz API: EGL v: 1.5 drivers: iris,nouveau,swrast platforms: gbm,x11,surfaceless,device API: OpenGL v: 4.6 compat-v: 4.3 vendor: intel mesa v: 25.0.7 renderer: Mesa Intel HD Graphics 630 (KBL GT2) API: Vulkan v: 1.3.231 drivers: intel,llvmpipe surfaces: xcb,xlib Network: Device-1: Qualcomm Atheros QCA6174 802.11ac Wireless Network Adapter driver: ath10k_pci Device-2: Realtek RTL8111/8168/8411 PCI Express Gigabit Ethernet driver: r8169
Keywords: feedback => (none)
Source RPM: kernel-stable-desktop-6.12 => kernel-stable-desktop
Installed 6.12.44 which draws in 6.12.59. Neveer seen such behavior before. Rebooted after installation and choose 6.12.59 and booting process never ends. Last lin reads "Starting plymouth-quit.service...." and then nothing. Left it for more than 1 hour....
The same thing as in comment 75 happens on my Intel machine -> 6.12.59-desktop-2... Have not tried rebooting.
CC: (none) => tarazed25
@Herman & @Len, which tool- drakrpm, urpmi, dnf, or...? @Herman, can you attach a system log from failed boot?
OK on my workstation, Plasma Like in comment 74: I let drakrpm choose updates; system already had 6.12.44 Also installed backport testing x11-server-xorg-21.1.21-1 Similar filesystems, plus an extra spinning drive as one large ext4 All tests OK: Some surfing, LibreOffice, Okular, Thunderbird, flatpak update, flatpak Signal, Appimage Nextcloud client, binary Syncthing, ... vt switching (ctrl-alt-F3, etc, and back) suspend-resume hibernate-restore - Unlike latest kernel 6.6.x kernels, it is not sluggish after resume. Additionally i tried hibernating with lots of RAM used, and it refused: returned to desktop after a while, journal saying it aborted hibernating as it did not have space enough (in swap). (would have been nice with a popup message to user, but this is how it is designed) VirtualBox 7.1.14 host with Guest Windows 7, using locally dkms built kmod: Tested dynamic window resizing, full size window, USB 2 flash disk, host folder sharing write protected and not, bidirectional clipboard, drag file from Dolphin to Explorer, Internet video in Firefox, Windows update(antivirus) [morgan@svarten ~]$ inxi -SMCGN System: Host: svarten.tribun Kernel: 6.12.59-desktop-2.stable.mga9 arch: x86_64 bits: 64 Desktop: KDE Plasma v: 5.27.10 Distro: Mageia 9 Machine: Type: Desktop Mobo: ASRock model: P55 Pro serial: <superuser required> BIOS: American Megatrends v: P2.60 date: 08/20/2010 CPU: Info: quad core model: Intel Core i7 870 bits: 64 type: MT MCP cache: L2: 1024 KiB Speed (MHz): avg: 2276 min/max: 1200/2934 cores: 1: 2276 2: 2276 3: 2276 4: 2276 5: 2276 6: 2276 7: 2276 8: 2276 Graphics: Device-1: Advanced Micro Devices [AMD/ATI] Navi 24 [Radeon RX 6400/6500 XT/6500M] driver: amdgpu v: kernel Display: x11 server: X.org v: 1.21.1.21 with: Xwayland v: 22.1.9 driver: X: loaded: amdgpu,v4l dri: radeonsi gpu: amdgpu resolution: 3840x2160~60Hz API: EGL v: 1.5 drivers: kms_swrast,radeonsi,swrast platforms: gbm,x11,surfaceless,device API: OpenGL v: 4.6 compat-v: 4.5 vendor: amd mesa v: 25.0.7 renderer: AMD Radeon RX 6400 (radeonsi navi24 LLVM 15.0.6 DRM 3.61 6.12.59-desktop-2.stable.mga9) Network: Device-1: Realtek RTL8111/8168/8411 PCI Express Gigabit Ethernet driver: r8169
The packager not provide update package list That could be part of the problem and if it requires install backported version of xorg then need to be indicated to make easy the work for us
In reply to Morgan in comment comment 77: Used urpmi in the end, after a few mishaps. Had tried qarepo, Reboot was normal. Mate desktop appeared with 6.12.59-desktop-2.stable.mga9 running. Have lost the transcript for the session so I need to switch to another partition to try for 6.12.44. As far as I can tell the cz.muni server which I normally use had synced OK with coffee. It is the middle of the night here, so later....
@ Comment 77 I used drakrpm to install the kernel. By system log, do you understand something like journalctl -b 1 > logfile.txt ??
Created attachment 15184 [details] Logfile of failing boot
Result attached of # journalctl --system -b -1 > logfile.txt
Created attachment 15185 [details] files list
BTW, please also check if the "ossp" package is NOT installed. I see, it's an older package often installed by default for the older OSS sound system compatibility, but sometimes (probably due to some bug) cause some race condition with systemd, especially during system shutdown. In my opinion, it shouldn't be installed by default anymore, but I still see it included in some new installations. Regarding the kernel, note that this is NOT a replacement for the standard kernel (either the -desktop or -server flavour), and the kmod precompiled modules packages aren't provided if you're using VirtualBox. It's provided as a standalone series to work alongside the standard kernels, and from a kernel packaging perspective, it's intended to be almost "invisible", so it won't update or replace any of the stock kernel package. The minimal installation to get it working is to install the package "kernel-stable-desktop-6.12.59-2.stable.mga9-1-1.mga9.x86_64.rpm", at least if you don't have any other packages installed that require external kernel modules, such as NVIDIA proprietary drivers or VirtualBox. In these cases, you'll also need to install the "kernel-stable-desktop-devel-6.12.59-2.stable.mga9-1-1.mga9.x86_64.rpm" package, to get the autobuild of the missing modules. Furthermore "kernel-stable-desktop-latest" and "kernel-stable-desktop-devel-latest" are intended as "glue" packages to retrieve the latest version of the kernel-stable-desktop (and -devel) series. Regarding xorg-server-21.1.21-1.mga9, it's not mandatory to get the kernel-stable working. It's another backport.
Ah files list yes :-) Yes do not test 6.12.44 anymore And contrary to what I thought before, the kernel is available in all achitectures; x86_64, i586, armv7hl, aarch64
Summary: Backport kernel 6.12 - desktop x86_64 only => Backport kernel 6.12 - desktop flavour only
i586 OK on Thinkpad T43 lxde Surfing with firefox, listen to pod, writing this, using wifi suspend-resume hibernate-restore No problem noted [ettan@localhost ~]$ inxi -SMCGN System: Host: localhost Kernel: 6.12.59-desktop-2.stable.mga9 arch: i686 bits: 32 Desktop: LXDE v: 0.10.1 Distro: Mageia 9 Machine: Type: Laptop System: IBM product: 2668R1G v: ThinkPad T43 serial: <superuser required> Mobo: IBM model: 2668R1G serial: <superuser required> BIOS: IBM v: 1YET62WW (1.27 ) date: 05/18/2006 CPU: Info: single core model: Intel Pentium M bits: 32 cache: 2 MiB note: check Speed (MHz): 1866 min/max: 800/1866 core: 1: 1866 Graphics: Device-1: Advanced Micro Devices [AMD/ATI] RV370/M22 [Mobility Radeon X300] driver: radeon v: kernel Display: x11 server: X.org v: 1.21.1.8 with: Xwayland v: 22.1.9 driver: X: loaded: radeon,v4l dri: r300 gpu: radeon resolution: 1024x768~60Hz API: EGL v: 1.4,1.5 drivers: kms_swrast,r300,swrast platforms: gbm,x11,surfaceless,device API: OpenGL v: 4.5 compat-v: 2.1 vendor: mesa v: 25.0.7 renderer: llvmpipe (LLVM 15.0.6 128 bits) Network: Device-1: Broadcom NetXtreme BCM5751M Gigabit Ethernet PCI Express driver: tg3 Device-2: Intel PRO/Wireless 2200BG [Calexico2] Network driver: ipw2200
x86_64 OK on Thinkpad T510, Plasma LUKS, LVM Surfing and watching film in firefox over wifi Libreoffice, Syncthing binary, suspend-resume OK hibernate-restore works with the usual caveat on this system: user have to shut off at hibernating when power lamp starts blinking (restores OK next power on) [ettan@localhost ~]$ inxi -SMCGN System: Host: localhost Kernel: 6.12.59-desktop-2.stable.mga9 arch: x86_64 bits: 64 Desktop: KDE Plasma v: 5.27.10 Distro: Mageia 9 Machine: Type: Laptop System: LENOVO product: 4349A13 v: ThinkPad T510 serial: <superuser required> Mobo: LENOVO model: 4349A13 serial: <superuser required> BIOS: LENOVO v: 6MET92WW (1.52 ) date: 09/26/2012 CPU: Info: dual core model: Intel Core i5 M 540 bits: 64 type: MT MCP cache: L2: 512 KiB Speed (MHz): avg: 1197 min/max: 1199/2534 cores: 1: 1197 2: 1197 3: 1197 4: 1197 Graphics: Device-1: NVIDIA GT218M [NVS 3100M] driver: nouveau v: kernel Device-2: Lenovo Integrated Webcam [R5U877] driver: uvcvideo type: USB Display: x11 server: X.org v: 1.21.1.8 with: Xwayland v: 22.1.9 driver: X: loaded: modesetting,v4l dri: nouveau gpu: nouveau resolution: 1920x1080~60Hz API: EGL v: 1.5 drivers: nouveau,swrast platforms: gbm,x11,surfaceless,device API: OpenGL v: 4.5 compat-v: 3.3 vendor: mesa v: 25.0.7 renderer: NVA8 API: Vulkan v: 1.3.231 drivers: llvmpipe surfaces: xcb,xlib Network: Device-1: Intel 82577LM Gigabit Network driver: e1000e Device-2: Intel Centrino Advanced-N 6200 driver: iwlwifi
x86_64 OK on ASUS G75, Plasma LUKS, LVM Surfing and watching film in Firefox over wifi Libreoffice, Syncthing binary, suspend-resume OK hibernate-restore OK [morgan@republic ~]$ inxi -SMCGN System: Host: republic.tribun Kernel: 6.12.59-desktop-2.stable.mga9 arch: x86_64 bits: 64 Desktop: KDE Plasma v: 5.27.10 Distro: Mageia 9 Machine: Type: Laptop System: ASUSTeK product: G75VW v: 1.0 serial: <superuser required> Mobo: ASUSTeK model: G75VW v: 1.0 serial: <superuser required> UEFI: American Megatrends v: G75VW.223 date: 01/07/2013 CPU: Info: quad core model: Intel Core i7-3610QM bits: 64 type: MT MCP cache: L2: 1024 KiB Speed (MHz): avg: 1748 min/max: 1200/3300 cores: 1: 1748 2: 1748 3: 1748 4: 1748 5: 1748 6: 1748 7: 1748 8: 1748 Graphics: Device-1: NVIDIA GK107M [GeForce GTX 660M] driver: nvidia v: 470.256.02 Device-2: Sunplus Innovation ASUS Webcam driver: uvcvideo type: USB Display: x11 server: X.org v: 1.21.1.8 with: Xwayland v: 22.1.9 driver: X: loaded: nvidia,v4l gpu: nvidia resolution: 1920x1080~60Hz API: EGL v: 1.5 drivers: nvidia,swrast platforms: x11,surfaceless,device API: OpenGL v: 4.6.0 compat-v: 4.5 vendor: nvidia mesa v: 470.256.02 renderer: NVIDIA GeForce GTX 660M/PCIe/SSE2 API: Vulkan v: 1.3.231 drivers: nvidia,llvmpipe surfaces: xcb,xlib Network: Device-1: Qualcomm Atheros AR9485 Wireless Network Adapter driver: ath9k Device-2: Qualcomm Atheros AR8151 v2.0 Gigabit Ethernet driver: atl1c
LC_ALL=C urpmi /home/katnatek/qa-testing/x86_64/*.rpm The following package has to be removed for others to be upgraded: kernel-userspace-headers-6.6.116-1.mga9.x86_64 (due to conflicts with linux-userspace-headers[> 6.6.116-1.mga9]) (y/N) y installing kernel-stable-userspace-headers-6.12.59-2.stable.mga9.x86_64.rpm kernel-stable-desktop-6.12.59-2.stable.mga9-1-1.mga9.x86_64.rpm kernel-stable-desktop-devel-6.12.59-2.stable.mga9-1-1.mga9.x86_64.rpm from /home/katnatek/qa-testing/x86_64 Preparing... ################################################################################################### 1/3: kernel-stable-desktop-devel-6.12.59-2.stable.mga9 ################################################################################################### 2/3: kernel-stable-desktop-6.12.59-2.stable.mga9 ################################################################################################### 3/3: kernel-stable-userspace-headers ################################################################################################### removing package kernel-userspace-headers-6.6.116-1.mga9.x86_64 1/1: removing kernel-userspace-headers-6.6.116-1.mga9.x86_64 ################################################################################################### vhba (20240917-1bdk_mga9): Installing module. ................ ........ remove-boot-splash: Format of /boot/initrd-6.12.59-desktop-2.stable.mga9.img not recognized You should restart your computer for kernel-stable-desktop-6.12.59-2.stable.mga9 Reboot uname -r 6.12.59-desktop-2.stable.mga9 systemctl status shorewall.service shorewall6.service Looks good But in dmesg I see this line bridge: filtering via arp/ip/ip6tables is no longer available by default. Update your scripts to load br_netfilter if you need this. I not usually check this, so I don't know if it is present in the regular kernels Ethernet OK Audio/Video OK Apache starts OK Mount ISOs with dkms-vhba + cdemu-client OK To this branch not likes the way that telegram application use the video , again get a full freeze trying to see one :S And can't get information of the issue dmesg says systemd-journald[552]: File /var/log/journal/65b43e022f9a4f1cb451e7b4b9ed8375/user-1000.journal corrupted or uncleanly shut down, renaming and replacing. Again the test was with intel card and I have to reboot for 3rd time in this test because try to get the information of inxi -G in 6.12 freeze again the system inxi -G Graphics: Device-1: Intel Core Processor Integrated Graphics driver: i915 v: kernel Display: x11 server: X.org v: 1.21.1.8 with: Xwayland v: 22.1.9 driver: X: loaded: intel,v4l dri: i965 gpu: i915 resolution: 1366x768~60Hz API: EGL v: 1.5 drivers: crocus,swrast platforms: gbm,x11,surfaceless,device API: OpenGL v: 4.5 compat-v: 2.1 vendor: intel mesa v: 25.0.7 renderer: Mesa Intel HD Graphics (ILK) API: Vulkan v: 1.3.231 drivers: llvmpipe surfaces: xcb,xlib With that issues reported we can't validate this kernel And I have
mga9, x64 Finally managed to install the kernel by going to the cz.muni site and grabbing the RPM file. Reboot OK. $ sys System partition is /dev/nvme0n1p7 6.12.44-desktop-1.stable.mga9 The system has been running perfectly for a while. Mate desktop functions normally. Common applications work. Have been doing some tidying up over ssh on other machines on the home network (wired). Personal ruby notebook works fine using hyperlinks and embedded funtions for image display and launching web pages and vlc videos, etc. The only problem seems to be that bluetooth is unstable. It finds a nearby audio device but connects and disconnects. No great loss though. Ach! I see the file list has changed. Later....
$ inxi -b Host: yildun Kernel: 6.12.59-desktop-2.stable.mga9 arch: x86_64 bits: 64 Desktop: MATE v: 1.26.1 Distro: Mageia 9 Machine: Type: Mini-pc System: Entroware product: Aura ED02R5 v: 1A Mobo: Intel model: NUC12WSBi7 v: M46422-303 serial: <superuser required> UEFI: Intel v: WSADL357.0085.2022.0718.1739 date: 07/18/2022 CPU: Info: 12-core (4-mt/8-st) 12th Gen Intel Core i7-1260P [MST AMCP] speed (MHz): avg: 400 min/max: 400/4700:3400 Graphics: Device-1: Intel Alder Lake-P Integrated Graphics driver: i915 v: kernel Display: x11 server: X.org v: 1.21.1.8 with: Xwayland v: 22.1.9 driver: X: loaded: modesetting unloaded: fbdev,vesa dri: iris gpu: i915 resolution: 2560x1440~60Hz API: OpenGL v: 4.6 vendor: intel mesa v: 25.0.7 renderer: Mesa Intel Iris Xe Graphics (ADL GT2) Network: Device-1: Intel Alder Lake-P PCH CNVi WiFi driver: iwlwifi Device-2: Intel Ethernet I225-V driver: igc Drives: Local Storage: total: 2.73 TiB used: 1.33 TiB (48.9%) Info: Memory: total: 32 GiB note: est. available: 30.92 GiB used: 2.5 GiB (8.1%) Processes: 339 Uptime: 1h 24m Shell: Bash inxi: 3.3.36 Yep, running fine so far. Bluetooth audio now connects OK after pairing request but there is no sound. Installed pipewire components but no joy there either. Back to pulseaudio and rebooted. Bluetooth sound working with pulseaudio.
(In reply to Herman Viaene from comment #82) > Created attachment 15184 [details] > Logfile of failing boot sddm knows it fails: Nov 29 12:03:29 mach3.hviaene.thuis sddm[1400]: No graphical seats found, attempt to start one on the main console anyway... ... Nov 29 12:03:34 mach3.hviaene.thuis sddm[1400]: Could not start Display server on vt 2 So why no graphical seat? I see earlier in boot that it detects radeon. What is the GPU? What do "inxi -G" tell? I believe /var/log/Xorg.0.log.old is the log from previous boot. Can you attach that, grabbed the boot after a failed boot? As a test just to get to GUI, could you try a very basic driver, like xorg vesa or what is suitable? @Giuseppe: Do the radeon driver need update for this kernel?
# inxi -G Graphics: Device-1: Advanced Micro Devices [AMD/ATI] Kabini [Radeon HD 8210] driver: amdgpu v: kernel Device-2: Chicony HP Truevision HD camera driver: uvcvideo type: USB Display: server: X.Org v: 22.1.9 with: Xwayland v: 22.1.9 driver: X: loaded: amdgpu,v4l dri: radeonsi gpu: amdgpu resolution: 1366x768~60Hz API: OpenGL v: 4.6 vendor: amd mesa v: 25.0.7 renderer: AMD Radeon HD 8210 (radeonsi kabini ACO DRM 3.54 6.6.116-desktop-1.mga9) API: EGL Message: EGL data requires eglinfo. Check --recommends. Attaching /var/log/Xorg.0.log.old, trying xorg vesa thereafter
Created attachment 15186 [details] xorg log file after failed boot
Switching to xorg vesa is a complete disaster, neither 6.12.59 nor the current 6.6.116 can boot.Hanging at the same point. Switching to LDM to check if that makes a difference.
Using again the AMD drivers, continued testing different display managers. First: all fail on 6.12.59. Other problems encountered on 6.6.116: LDM hangs when selecting Plasma LDM segfaults when selecting Xfce LDM works OK with MATE. I never used XDM that I can remember, so I don't know whether this is intended behavior; I get the dialogues for user and password, enter those then I get the jingle and the Mageia background, but that's it. I can right click and start a terminal session. But as far as deskop is concerned, nothing further happens. Leaving this here as i have other things to attend to.
Thank you Herman Hm same radeon driver works on the system I now write this on, GPU = Navi 24 [Radeon RX 6400] @Giuseppe, see xorg log in comment 95
The driver referenced in xorg log, amdgpu_drv.so, is part of our package x11-driver-video-amdgpu 23.0.0, have an update sitting in testing repo since July 31, version 25.0.0. Maybe that could be something to try. I have not installed it on my system. @Giuseppe, maybe time for an even further update?
(In reply to Morgan Leijström from comment #99) > The driver referenced in xorg log, amdgpu_drv.so, is part of our package > x11-driver-video-amdgpu 23.0.0, have an update sitting in testing repo since > July 31, version 25.0.0. > > Maybe that could be something to try. > I have not installed it on my system. > > @Giuseppe, maybe time for an even further update? We could see with kernel-6.12.60 at this point. With intel is everything fine? For the amdgpu, i.e. the x11-driver-video-amdgpu, is the DDX driver which AFAIK shouldn't be affected by the kernel; the update for 25.0.0 fixes some bug and also improve performance. Over modesetting it should have some extra feature (TearFree, etc.). What about Wayland?
My HW is different: Radeon HD8210. And which testing repos are you talking about?? I don't see something like that???
the x11-driver-video-amdgpu-25.0.0 is in mga9/core/updates_testing. The 6.12.60 has not yet built.
Yes different GPU but it seems to use the same xorg driver I just now installed 25.0.0 and rebooted, runs OK. Giuseppe, why do we not send that driver to QA? Should we pause QA backport 6.12 again, waiting for 6.12.60 expected really soon? What would be the steps for Herman if he wish to try Wayland? For reference a snippet of my xorg log here [ 36.773] (II) Loading /usr/lib64/xorg/modules/drivers/amdgpu_drv.so [ 36.778] (II) Module amdgpu: vendor="X.Org Foundation" [ 36.778] compiled for 1.21.1.8, module version = 25.0.0 [ 36.778] Module class: X.Org Video Driver [ 36.778] ABI class: X.Org Video Driver, version 25.2 [ 36.778] (II) v4l driver for Video4Linux overlay mode (V4L2) [ 36.778] (II) AMDGPU: Driver for AMD Radeon: All GPUs supported by the amdgpu kernel driver [ 36.804] (WW) Falling back to old probe method for v4l [ 36.805] (**) AMDGPU(0): Depth 24, (--) framebuffer bpp 32 [ 36.805] (II) AMDGPU(0): Pixel depth = 24 bits stored in 4 bytes (32 bpp pixmaps) [ 36.805] (==) AMDGPU(0): Default visual is TrueColor [ 36.805] (II) Applying OutputClass "AMDgpu" options to /dev/dri/card0 [ 36.805] (==) AMDGPU(0): RGB weight 888 [ 36.805] (II) AMDGPU(0): Using 8 bits per RGB (8 bit DAC) [ 36.805] (--) AMDGPU(0): Chipset: "AMD Radeon RX 6400" (ChipID = 0x743f) ...and after that many AMDGPU lines and many other lines
Created attachment 15187 [details] files list
Attachment 15089 is obsolete: 0 => 1 Attachment 15185 is obsolete: 0 => 1
The new version 6.12.60‑1.mga9 is ready. I extracted the file list from the following URL: https://pkgsubmit.mageia.org/uploads/done/9/core/backports_testing/20251201174611.ghibo.duvel.1534083/kernel-stable-6.12.60-1.stable.mga9/ (see packages*.log). BTW, if we could preserve those files or make them available somewhere else instead of having to guess later from mirrors, it would greatly speed up file‑list administration.
Great, lets test 6.12.60‑1. - Running it on one system, more and reports later.
Keywords: feedback => (none)Summary: Backport kernel 6.12 - desktop flavour only => Backport kernel 6.12.60 - desktop flavour only
RH x86_64 rpm -qa|grep kernel-stable kernel-stable-userspace-headers-6.12.60-1.stable.mga9 kernel-stable-desktop-devel-6.12.60-1.stable.mga9-1-1.mga9 kernel-stable-desktop-6.12.60-1.stable.mga9-1-1.mga9 Test tomorrow, I'm setting a Cauldron VM
Still freeze my system trying to see a video in telegram desktop application That is not good for a regular use
(In reply to katnatek from comment #108) > Still freeze my system trying to see a video in telegram desktop application > That is not good for a regular use Is that the similar freeze of hardware as of https://bugs.mageia.org/show_bug.cgi?id=34545#c67 with intel modesetting? Does temporarily switching xorg.conf to DDX Driver "intel" instead of Driver "modesetting" would cause the same freeze?
Kernel: 6.12.60-desktop-1.stable.mga9 arch: x86_64 8-core AMD Ryzen 7 5700U with Radeon Graphics Advanced Micro Devices [AMD/ATI] Lucienne driver: amdgpu x11 server: X.org v: 1.21.1.8 with: Xwayland v: 22.1.9 driver: X: $ rpm -qa | grep 6.12.60 kernel-stable-desktop-6.12.60-1.stable.mga9-1-1.mga9 kernel-stable-desktop-latest-6.12.60-1.stable.mga9 kernel-stable-desktop-devel-latest-6.12.60-1.stable.mga9 kernel-stable-desktop-devel-6.12.60-1.stable.mga9-1-1.mga9 kernel-stable-source-6.12.60-1.stable.mga9-1-1.mga9 kernel-stable-source-latest-6.12.60-1.stable.mga9 Clean install, revised bootloader and faultless reboot. Running comfortably with Mate.
Repeated tests OK on four laptops now with 6.12.60-1: i586: Comment 87 x86_64: Comments 74, 88, 89
(In reply to Giuseppe Ghibò from comment #109) > (In reply to katnatek from comment #108) > > > Still freeze my system trying to see a video in telegram desktop application > > That is not good for a regular use > > Is that the similar freeze of hardware as of > https://bugs.mageia.org/show_bug.cgi?id=34545#c67 with intel modesetting? > Does temporarily switching xorg.conf to DDX Driver "intel" instead of Driver > "modesetting" would cause the same freeze? The freeze is with intel driver, I'll try with modesettings, but that will be a workaround something is rotten with the intel driver for me in this kernel
The same with modesetings, so definitive not a thing that I recommend pass the qa stage
Can you see whether with the DDX package x11-driver-video-intel-2.99.917-70.mga9 (from core/updates_testing) would freeze in the same way? Also can you verify whether passing "i915.enable_psr=0 i915.enable_dc=0 i915.enable_fbc=0 i915.enable_guc=0" to the boot cmdline option affects anything?
is the freeze affecting only usage when starting the telegram application? I wonder whether it's related to GPU H264 hardware decoding.
(In reply to Giuseppe Ghibò from comment #115) > is the freeze affecting only usage when starting the telegram application? I > wonder whether it's related to GPU H264 hardware decoding. That could be a possibility as the freeze come as fast as I try to play a video in the application x11-driver-video-intel-2.99.917-70.mga9 not makes difference
The same videos plays without issue from telegram in the browser So is something related to telegram desktop application, but as it freeze the system and the journal not helps. Due Herman issues with this kernel I think we can't validate, no matter if my issue can be considered not too bad
As a backport I think it need not be perfect. Especially as the first backport, not updating earlier package. - In contrast to a normal updates, user need to really choose 6.12 kernel by enabling backport repo and pick it. And if problem, 6.6 is still available as boot selection. For some users it solves problems over 6.6 kernels. Some hardware compatibility. In my case it solves the problem later 6.6 kernels have on my workstation with extreme sluggishness for a while after hibernation. Also, I feel all systems here runs are a bit more snappy. I see no regression on any of my systems I have tested yet. I think we can release as is, with a note that some users may experience problems and a link to this bug. Also add that note to wiki. Then we continue working on a next update.
Test OK on my workstation, details see comment #78 All tests again like there OK: VirtualBox, MSW7 as client vt switching (ctrl-alt-F3, etc, and back) suspend-resume hibernate-restore Some surfing, LibreOffice, Okular, Thunderbird, flatpak update, flatpak Signal, Appimage Nextcloud client, binary Syncthing, ...
BTW, For the problematic cases, it would be interesting to having them tried in Cauldron.
Tested new set ,and initially did not get any further than previous versions. But then I got an idea: changed the run level at boot to 2. Now the boot process finishes normally and I can login. The problem I see now is X session. I can jump up and down all different settings (driver, resolution etc...), the startx command always fails at the same point,. After loading the driver, it always ends up with "No DRM device at pci:0....." I tried the AMD driver, the xorg.amdgpu; xorg.vesa, all in vain. But that still does not explain why this error does not show up in the 6.6 kernels and now in all recent 6.12 ????
(In reply to Morgan Leijström from comment #120) > BTW, For the problematic cases, it would be interesting to having them tried > in Cauldron. For me, Not a thing to do in real hardware, that will mean left of QA for the rest of mageia 9 life. I'll not give the OK to this but if the majority think is good, will be fine for me
$ inxi -b Kernel: 6.12.60-desktop-1.stable.mga9 arch: x86_64 Intel model: NUC12WSBi7 v: M46422-303 UEFI: Intel v: WSADL357.0085.2022.0718.1739 12-core (4-mt/8-st) 12th Gen Intel Core i7-1260P Intel Alder Lake-P Integrated Graphics driver: i915 x11 server: X.org v: 1.21.1.8 with: Xwayland v: 22.1.9 driver: X: loaded: modesetting unloaded: fbdev,vesa dri: iris gpu: i915 API: OpenGL v: 4.6 vendor: intel mesa v: 25.0.7 renderer: Mesa Intel Iris Xe Graphics (ADL GT2) This has been running Mate for a day without mishap.
Comment 121 is troubling considering that my AMD system booted OK. There was a Cauldron partition on this particular machine - need to check if it is still alive.
Let’s hold off updating the kernel for now. It looks like we’re dealing with two separate problems: 1) Freeze on the Ironlake CPU. We need to investigate why this is happening. Is the memory OK on your hardware? Have you tried a deep memtest recently? Does disabling guc at boot (as described in comment #119) make any difference? 2) Herman issue. Can you try whether installing libdrm 2.4.129 from the updates_testing repository has any effect?
Installed libdrm 2.4.129 from the updates_testing repository. Kernel 6.6.116 still behaves correctly, Plasma Wayland starts OK. But no improvement with 6.12.60. Boot to runlevel 3 hangs, boot to runlevel2 completes correctly, but startx or startkde or startplasma-wayland all fail. The error on drm is thne same.
(In reply to Giuseppe Ghibò from comment #114) > Can you see whether with the DDX package > x11-driver-video-intel-2.99.917-70.mga9 (from core/updates_testing) would > freeze in the same way? Also can you verify whether passing > "i915.enable_psr=0 i915.enable_dc=0 i915.enable_fbc=0 i915.enable_guc=0" to > the boot cmdline option affects anything? cat /proc/cmdline BOOT_IMAGE=/boot/vmlinuz-6.12.60-desktop-1.stable.mga9 root=UUID=b5f84c02-99ef-4f08-a5c5-2b7593fbdd0e ro splash quiet noiswmd resume=UUID=ac50cb2a-7731-479b-94f1-e90cc4f90106 audit=0 i915.enable_psr=0 i915.enable_dc=0 i915.enable_fbc=0 i915.enable_guc=0 vga=791 Not makes difference
6.12 is the default kernel in Mageia 10? It would be very interesting to see tests on the systems that here have problems with 6.12 kernel on mga9, tested with Mageia 10 alpha1 or later!
(In reply to Morgan Leijström from comment #128) > 6.12 is the default kernel in Mageia 10? At the moment it is. Plans are to switch to 6.18 lts when it will be more stable. At the moment we have 6.18 under kernel-mainline and kernel-stable-testing, but it's not complete (no flavours, no other archs), and package doesn't even build smootly, due to other problems. > > It would be very interesting to see tests on the systems that here have > problems with 6.12 kernel on mga9, tested with Mageia 10 alpha1 or later! maybe the live
(In reply to Morgan Leijström from comment #128) > 6.12 is the default kernel in Mageia 10? > > It would be very interesting to see tests on the systems that here have > problems with 6.12 kernel on mga9, tested with Mageia 10 alpha1 or later! I'll try to make a cauldron install in the real hardware
I did and logged a bug 34842.
Now that Mageia 10 reached the alpha1 milestone, does it still make sense to waste ressource to try to backport kernel 6.12 to Mageia 9 ?
In my view we should release this with no more work. It works notably better then our 6.6 for example on two of my machines; more "snappy", and on my workstation do not suffer some obscure lazyness that 6.6 have after hibernation. Most of mine and wifes machines use this 6.12, no regression noted here. Remember this is a backport - users choice to install. The problems 6.12 have seem to maybe be there also in mga10 currently and when fixed, maybe we can rather easily make a new backport. Or maybe of 6.18 later. Anyway we still support Mageia 9 for three months after mga10 is released, so maybe until summer. Some users cant switch quickly, and benefit from a backport.
The Behavior of Live Xfce 64bit alpha1 make me thing that When I get Cauldron installed in this system, I'll we have the freezes too :S
So let's postpone for a while, maybe newer 6.12.63 or so. Apparently the problems arise with old AMD APU like E1-2100. It could be that support might be even dropped upstream for such chipset on newer kernels... Anyway when it works which drivers is used? radeon or amdgpu? And does it uses DDX? what if switching with adding some boot parameter like: radeon.si_support=1 radeon.cik_support=1 amdgpu.si_support=0 amdgpu.cik_support=0 or viceversa radeon.si_support=0 radeon.cik_support=0 amdgpu.si_support=1 amdgpu.cik_support=1 ?
I can confirm that with in real installation With the intel card, try to see video in Telegram Desktop make the system freeze
(In reply to katnatek from comment #136) > I can confirm that with in real installation > With the intel card, try to see video in Telegram Desktop make the system > freeze I mean cauldron off course
So backport seem perfect in that it keep the same problems ;-)
(In reply to Morgan Leijström from comment #138) > So backport seem perfect in that it keep the same problems ;-) That just make it worst , first test with 6.18 (please make names a few more short) Look like the issue is not present, today I hope I do more test As 6.18 will be LTS perhaps should be the default kernel for mga10?
Any issue for the moment with uname -r 6.18.1-desktop-4.stabletesting.mga10 inxi -G Graphics: Device-1: Intel Core Processor Integrated Graphics driver: i915 v: kernel Display: x11 server: X.org v: 1.21.1.21 with: Xwayland v: 24.1.9 driver: X: loaded: intel,v4l dri: crocus gpu: i915 resolution: 1366x768~60Hz API: EGL v: 1.5 drivers: crocus,swrast platforms: gbm,x11,surfaceless,device API: OpenGL v: 4.5 compat-v: 2.1 vendor: intel mesa v: 25.1.9 renderer: Mesa Intel HD Graphics (ILK) Info: Tools: api: eglinfo,glxinfo de: kscreen-doctor x11: xdpyinfo, xprop, xrandr
So maybe backport 6.18? Will 6.12 still be useful when 6.18 is final - i.e will some hardware support be lacking from 6.18 that is in 6.12?
Maybe backport 6.18. Actually it's called "kernel-stable-testing" in mga10 (it's still to be completed the various flavours), and at some point, in some weeks, it will replace stock 6.12 (in mga10).
I found 6.18.2 in mga9 backport testing, and desktop flavour is in use on two systems so far, no regresion noted. Tested Plasma desktop, firefox internet video over wifi, suspend, hibernate, LUKS, LVM, VirtualBox, flatpak chromium with video. I also see there is a server flavour. One problem: kernel-stable-testing-userspace-headers seem to have problematic dependency If i try to install kernel-stable-testing-userspace-headers it wants o uninstall much, uncluding gcc! urpmi gives better answer: [morgan@svarten ~]$ LC_ALL=C sudo urpmi --test kernel-stable-testing-userspace-headers [sudo] password for morgan: In order to satisfy the 'kernel-linus-devel|kernel-desktop-devel|kernel-server-devel|kernel-server-devel|kernel-linus-devel|kernel-desktop-devel|kernel-server-devel|kernel-desktop-devel|kernel-linus-devel|kernel-desktop-devel|kernel-linus-devel|kernel-server-devel|kernel-linus-devel|kernel-desktop-devel|kernel-server-devel|kernel-desktop-devel|kernel-server-devel|kernel-linus-devel|kernel-server-devel|kernel-desktop-devel|kernel-linus-devel|kernel-linus-devel|kernel-desktop-devel|kernel-server-devel|kernel-server-devel|kernel-desktop-devel|kernel-linus-devel|kernel-server-devel|kernel-linus-devel|kernel-desktop-devel|kernel-server-devel|kernel-desktop-devel|kernel-linus-devel|kernel-server-devel|kernel-desktop-devel|kernel-linus-devel|kernel-server-devel|kernel-desktop-devel|kernel-linus-devel|kernel-linus-devel|kernel-desktop-devel|kernel-server-devel|kernel-linus-devel|kernel-desktop-devel|kernel-server-devel|kernel-server-devel|kernel-desktop-devel|kernel-linus-devel|kernel-server-devel|kernel-desktop-devel|kernel-linus-devel|kernel-server-devel|kernel-linus-devel|kernel-desktop-devel|kernel-desktop-devel|kernel-server-devel|kernel-linus-devel|kernel-linus-devel|kernel-desktop-devel|kernel-server-devel|kernel-linus-devel|kernel-server-devel|kernel-desktop-devel|kernel-linus-devel|kernel-desktop-devel|kernel-server-devel' dependency, one of the following packages is needed: 1- kernel-desktop-devel-6.6.105-1.mga9.x86_64: The kernel-devel files for kernel-desktop-6.6.105-1.mga9 (to upgrade) 2- kernel-linus-devel-6.6.105-1.mga9.x86_64: The kernel-linus devel files for 3rdparty modules build (to upgrade) 3- kernel-server-devel-6.6.105-1.mga9.x86_64: The kernel-devel files for kernel-server-6.6.105-1.mga9 (to upgrade) [morgan@svarten ~]$ rpm -qa | grep 'devel-6.6.' kernel-linus-devel-6.6.116-1.mga9 kernel-desktop-devel-6.6.116-1.mga9 kernel-server-devel-6.6.116-1.mga9
Summary: Backport kernel 6.12.60 - desktop flavour only => Backport kernel
(In reply to Morgan Leijström from comment #143) > I found 6.18.2 in mga9 backport testing, and desktop flavour is in use on > two systems so far, no regresion noted. > > Tested Plasma desktop, firefox internet video over wifi, suspend, hibernate, > LUKS, LVM, VirtualBox, flatpak chromium with video. > > I also see there is a server flavour. > The 6.18 seems already a good series, better than expected for a .2 release, and that should be the new default kernel in cauldron, in some week, actually completing the other flavours (but build failed actually in cauldron). > One problem: kernel-stable-testing-userspace-headers seem to have > problematic dependency > Again? That wasn't fixed in kernel-stable?
(In reply to Giuseppe Ghibò from comment #144) > The 6.18 seems already a good series, better than expected for a .2 release, > and that should be the new default kernel in cauldron, in some week, > actually completing the other flavours It would be great to have 6.18 in alpha2 > > One problem: kernel-stable-testing-userspace-headers seem to have > > problematic dependency > > > > Again? That wasn't fixed in kernel-stable? There is no problem switching kernel-stable-userspace-headers <-> kernel-userspace-headers But to kernel-stable-testing-userspace-headers there is.
(In reply to Giuseppe Ghibò from comment #50) > (In reply to Morgan Leijström from comment #49) > > Is it possible to make initrd less clumsy? > > 6.12.44 initrd is almost three times as big as 6.6.103 initrd > > > > > > -rw------- 1 root root 70218134 sep 1 23:26 > > initrd-6.12.44-desktop-1.stable.mga9.img > > -rw------- 1 root root 25225600 aug 31 09:47 > > initrd-6.6.103-desktop-1.mga9.img > > d! > > > > Some of us have a small separate /boot partition, rest of disk encrypted. > > can't do much at the moment for size. I see .img is down to good size, great :-) sizes of desktop .img: 6.6.116: 41 MB 6.12.60: 48 MB 6.18.2: 50 MB
(In reply to Morgan Leijström from comment #145) > There is no problem switching > kernel-stable-userspace-headers <-> kernel-userspace-headers > > But to kernel-stable-testing-userspace-headers there is. Solved in 6.18.2-3 :-)
(In reply to Morgan Leijström from comment #147) > (In reply to Morgan Leijström from comment #145) > > There is no problem switching > > kernel-stable-userspace-headers <-> kernel-userspace-headers > > > > But to kernel-stable-testing-userspace-headers there is. > > Solved in 6.18.2-3 :-) Finally. That could help also cauldron since it's matching also the bugs. Do you get it works flawlessly with nvidia proprietary?
I will try after Christmas on the one machine with newest nvidia i have, nvdia470... I have an incoming relatively brand new machine with only a few years old nvidia (my sons discarded PC...) beginning next year. I may also try a live mga10 on both. Sorry to say there is a regression from backport desktop 6.18.2-2 to -3: On my wifes Acer Aspirer 717 it panics trying to boot, something about not mounting. My workstation boots desktop 6.18.2-3, but restoring from hibernation hangs quickly with black screen. But on same system, server 6.18.2-3 works great so far, and like 6.12 it do not show the sluggishness after restore that later 6.6 kernels have. Both systems use LVM on a LUKS pv partition. More test after christmas.
On my Thinkpad T510, both server and desktop 6.18.2-3 successfully restore after hibernation. Also here LVM on a LUKS pv partition. nouveau. I am going offline now for a couple days at least.
> > But in dmesg I see this line > > bridge: filtering via arp/ip/ip6tables is no longer available by default. > Update your scripts to load br_netfilter if you need this. > > I not usually check this, so I don't know if it is present in the regular > kernels > This is present with the regular kernels and has been since the kernel upstream changed filtering provider. I meant to write a bug about this a while back but kept forgetting to do so.
CC: (none) => rihoward1
Note Comment 151 is in response to Comment 90 katnatek 2025-11-30 19:45:43 CET
(In reply to r howard from comment #151) > > > > But in dmesg I see this line > > > > bridge: filtering via arp/ip/ip6tables is no longer available by default. > > Update your scripts to load br_netfilter if you need this. > > > > I not usually check this, so I don't know if it is present in the regular > > kernels > > > > This is present with the regular kernels and has been since the kernel > upstream changed filtering provider. I meant to write a bug about this a > while back but kept forgetting to do so. If the 6.18.x series requires changes to the netfilter related scripts anywhere (especially for Cauldron), it should be opened a dedicated bug.
Will shorewall still work ?
CC: (none) => bruno
At the moment I see shorewall working under 6.18.x, either in mga9 or mga10, at least in a basic configuration in IPv4.
(In reply to Giuseppe Ghibò from comment #153) > (In reply to r howard from comment #151) > > > > > > But in dmesg I see this line > > > > > > bridge: filtering via arp/ip/ip6tables is no longer available by default. > > > Update your scripts to load br_netfilter if you need this. > > > > > > I not usually check this, so I don't know if it is present in the regular > > > kernels > > > > > > > This is present with the regular kernels and has been since the kernel > > upstream changed filtering provider. I meant to write a bug about this a > > while back but kept forgetting to do so. > > If the 6.18.x series requires changes to the netfilter related scripts > anywhere (especially for Cauldron), it should be opened a dedicated bug. I see the problem with Mageia 9 and the 6.6.x kernels so probably should also have a bug for that series.
Created attachment 15255 [details] file list Updated file list
Attachment 15187 is obsolete: 0 => 1
installing kernel-stable-testing-desktop-latest-6.18.2-3.stabletesting.mga9.x86_64.rpm kernel-stable-testing-desktop-6.18.2-3.stabletesting.mga9-1-1.mga9.x86_64.rpm kernel-stable-testing-desktop-devel-latest-6.18.2-3.stabletesting.mga9.x86_64.rpm kernel-stable-testing-userspace-headers-6.18.2-3.stabletesting.mga9.x86_64.rpm kernel-stable-testing-desktop-devel-6.18.2-3.stabletesting.mga9-1-1.mga9.x86_64.rpm from //home/katnatek/qa-testing/x86_64 Preparing... ################################################################################################### 1/5: kernel-stable-testing-desktop-devel-6.18.2-3.stabletesting.mga9 ################################################################################################### 2/5: kernel-stable-testing-desktop-devel-latest ################################################################################################### 3/5: kernel-stable-testing-desktop-6.18.2-3.stabletesting.mga9 ################################################################################################### 4/5: kernel-stable-testing-desktop-latest ################################################################################################### 5/5: kernel-stable-testing-userspace-headers ################################################################################################### removing package kernel-userspace-headers-6.6.116-1.mga9.x86_64 1/1: removing kernel-userspace-headers-6.6.116-1.mga9.x86_64 ################################################################################################### vhba (20240917-1bdk_mga9): Installing module. ...............(bad exit status: 10) Build failed. Installation skipped. You should restart your computer for kernel-stable-testing-desktop-6.18.2-3.stabletesting.mga9 I got this fail in cauldron before update dkms-vhba to latest version So we need to backport dkms-vhba-20250329 too
uname -r 6.18.2-desktop-3.stabletesting.mga9 [katnatek@jgrey ~]$ inxi -G Graphics: Device-1: Intel Core Processor Integrated Graphics driver: i915 v: kernel Display: x11 server: X.org v: 1.21.1.8 with: Xwayland v: 22.1.9 driver: X: loaded: intel,v4l dri: i965 gpu: i915 resolution: 1366x768~60Hz API: EGL v: 1.5 drivers: crocus,swrast platforms: gbm,x11,surfaceless,device API: OpenGL v: 4.5 compat-v: 2.1 vendor: intel mesa v: 25.0.7 renderer: Mesa Intel HD Graphics (ILK) API: Vulkan v: 1.3.231 drivers: llvmpipe surfaces: xcb,xlib rpm -q x11-driver-video-intel x11-driver-video-intel-2.99.917-69.mga9 No freeze watching videos in telegram desktop for the moment
(In reply to Morgan Leijström from comment #149) > I will try after Christmas on the one machine with newest nvidia i have, > nvdia470... nvidia470: dkms fail to build it, exit status 10 - like for vhba in comment 158 Note that nvidia470 works here with 6.12.60 and earlier.
(In reply to Morgan Leijström from comment #160) > (In reply to Morgan Leijström from comment #149) > > I will try after Christmas on the one machine with newest nvidia i have, > > nvdia470... > > nvidia470: dkms fail to build it, exit status 10 - like for vhba in comment > 158 > > Note that nvidia470 works here with 6.12.60 and earlier. yep, cauldron's nvidia470 has an extra patchset for building on newer kernels, mga9's nvidia470 doesn't.
Graphics Kabini Radeon HD 8210 Installing on 6.6.116 instance results as in 6.12 in a booting system where X-server does not start. Had to add "radeon.si_support=0 radeon.cik_support=0 amdgpu.si_support=1 amdgpu.cik_support=1" to boot command, then starting is OK. $ uname -a Linux xxx.yyy.zzzz 6.18.2-desktop-3.stabletesting.mga9 #1 SMP PREEMPT_DYNAMIC Tue Dec 23 03:37:49 UTC 2025 x86_64 GNU/Linux. Wifi, internet access, access to remote NFS shares to network printer/scanner HP, shared scanner Epson from desktop PC, all works OK.
i586 6.18.2-desktop-3 OK on Thinkpad T43: LXDE Browsing internet using Firefox, over wifi, and over wired eth. (even playing video in small size) suspend-resume hibernate-restore Writing this in Firefox
I suggest we release 6.12.60 (move from backports_testing to backports) __Reasons to release 6.12.60 It works snappier on some systems (two of mine at least) Support new stuff. Good to show we provide a new kernel series to Mageia 9. Let it out to be in use and do no more work on 6.12. Yes it fail some few systems but it is easy to select previous kernel at boot. And backport is not as guaranteed to work as a regular update anyway. And it have same issues as 6.12.60 in 10alpa1 anyway. Then we continue working on 6.18.x (both for mga10 amd 9 backport) Having 6.12 in mga9 backport means it is less important to make nvidia470 build with 6.18. backport. Whatever we do, https://wiki.mageia.org/en/Kernel_flavours#Backport_kernels to be updated accordingly, with pros and cons too. --- BTW are we going to provide nvidia470 in mga10 ? I see nvidia470-470.256.02-7 in nonfree/release.
(In reply to Morgan Leijström from comment #164) > I suggest we release 6.12.60 (move from backports_testing to backports) > > __Reasons to release 6.12.60 > It works snappier on some systems (two of mine at least) > Support new stuff. > Good to show we provide a new kernel series to Mageia 9. > Let it out to be in use and do no more work on 6.12. > > Yes it fail some few systems but it is easy to select previous kernel at > boot. > And backport is not as guaranteed to work as a regular update anyway. > And it have same issues as 6.12.60 in 10alpa1 anyway. > > > Then we continue working on 6.18.x (both for mga10 amd 9 backport) > Having 6.12 in mga9 backport means it is less important to make nvidia470 > build with 6.18. backport. > > > Whatever we do, https://wiki.mageia.org/en/Kernel_flavours#Backport_kernels > to be updated accordingly, with pros and cons too. > > --- > > BTW are we going to provide nvidia470 in mga10 ? > I see nvidia470-470.256.02-7 in nonfree/release. Actually it's not in ISO live anymore, but since it builds and starts flawlessly even in 6.18.x and current xorg, I would still allow it to be installed from the net, better than nothing, also considering that for older nv cards there are not much other "stable" alternatives. Just add a note in the release notes that it's unsupported upstream and if possible (but that would require a modification to drakx11, so adding the long TODO feature request there) popup a banner when installing.
(In reply to Giuseppe Ghibò from comment #165) > (In reply to Morgan Leijström from comment #164) > > BTW are we going to provide nvidia470 in mga10 ? > > I see nvidia470-470.256.02-7 in nonfree/release. > > Actually it's not in ISO live anymore, but since it builds and starts > flawlessly even in 6.18.x and current xorg, I would still allow it to be > installed from the net, better than nothing, also considering that for older > nv cards there are not much other "stable" alternatives. Just add a note in > the release notes that it's unsupported upstream and if possible (but that > would require a modification to drakx11, so adding the long TODO feature > request there) popup a banner when installing. I edited https://wiki.mageia.org/en/Mageia_10_Release_Notes#Proprietary_NVIDIA_driver Do you mean drakx11 need update to install nvidia470 on mga10, or just would be nice with enhancement of showing warning? I think we should not touch drakx11, but instead write in package description that the driver is not supported by security updates but supplied as is and using it is up to the user.
mga10 installer is never autoselecting R470, I presume?
Exactly it doesn't, like is not autoselecting nvidia-current or others.
So, back to talk about backport... What about releasing 6.12.60 and moving on?
(In reply to Morgan Leijström from comment #149) > Sorry to say there is a regression from backport desktop 6.18.2-2 to -3: > On my wifes Acer Aspirer 717 it panics trying to boot, something about not > mounting. > > My workstation boots desktop 6.18.2-3, but restoring from hibernation hangs > quickly with black screen. But on same system, server 6.18.2-3 works great > so far, and like 6.12 it do not show the sluggishness after restore that > later 6.6 kernels have. > > Both systems use LVM on a LUKS pv partition. > > More test after christmas. The Acer still flawlessy runs 6.18.2-3 (desktop flavour, we never hibernate it) My workstation on 6.18.2-3 have been running server flawlessly. Now again tried desktop flavour, and it seem to work perfectly too, except at restoring after hibernation it boots to black screen shortly after entering LUKS key, no monitor signal. It need to be poweered down using REISUB. Then there is a trick: Power it on again, in grub select server flavour, and then it boots and restores the session successflully, restoring also the desktop kernel and it i now write this. In journal there is no trace of the failed restore attempt, so / partition probably never got mounted/remounted write before boot stalled.
(In reply to Morgan Leijström from comment #170) > (In reply to Morgan Leijström from comment #149) > > Sorry to say there is a regression from backport desktop 6.18.2-2 to -3: > > On my wifes Acer Aspirer 717 it panics trying to boot, something about not > > mounting. > > > > My workstation boots desktop 6.18.2-3, but restoring from hibernation hangs > > quickly with black screen. But on same system, server 6.18.2-3 works great > > so far, and like 6.12 it do not show the sluggishness after restore that > > later 6.6 kernels have. > > > > Both systems use LVM on a LUKS pv partition. > > > > More test after christmas. > > The Acer still flawlessy runs 6.18.2-3 (desktop flavour, we never hibernate > it) > > My workstation on 6.18.2-3 have been running server flawlessly. > Now again tried desktop flavour, and it seem to work perfectly too, except > at restoring after hibernation it boots to black screen shortly after > entering LUKS key, no monitor signal. It need to be poweered down using > REISUB. Then there is a trick: Power it on again, in grub select server > flavour, and then it boots and restores the session successflully, restoring > also the desktop kernel and it i now write this. In journal there is no > trace of the failed restore attempt, so / partition probably never got > mounted/remounted write before boot stalled. There is 6.18.2-4.mga9 available.
To keep bug shorter you could have avoided quoting... Is there something specific to test in 6.18.2-4 ?
Tested on two machines so far: desktop 6.18.2-4 fail booting on my Thinkpad T510, just a second after grub, screen is black and caps lock LED start blinking. I have never seen that before. On same system, *server* 6.18.2-4 works perfectly, incl suspend and hibernate (with the usual problem i need to press button to power off) System use LUKS and LVM. On my laptop with nvidia470, dkms fail building like earlier 6.18, i guess that need a backport nvidia470 - if it works on mga10? Do also other nvidia need backporting? --- I repeat I think we should release 6.12 as backport, either the enough tested 6.12.60 ASAP - Or a new, 6.12.63? to be built and sent to QA ASAP?
The 6.12.60 is now already obsolete, and would require further security fixes.
(In reply to Morgan Leijström from comment #173) > On my laptop with nvidia470, dkms fail building like earlier 6.18 Thanks, now tested the smoking new 470.256.02-4 from updates testing, it builds and works on latest 6.6, 6.12 and 6.18 kernels: [morgan@republic ~]$ dkms status|grep nvid nvidia470, 470.256.02-4.mga9.nonfree, 6.6.116-desktop-1.mga9, x86_64: installed nvidia470, 470.256.02-4.mga9.nonfree, 6.12.60-desktop-1.stable.mga9, x86_64: installed nvidia470, 470.256.02-4.mga9.nonfree, 6.18.2-desktop-4.stabletesting.mga9, x86_64: installed Tested including suspend and hibernate, video in firefox. System use LVM on LUKS. Plasma on Xorg. --- Is there something I can try for you for > desktop 6.18.2-4 fail booting on my Thinkpad T510 elder nvidia, using nouveau, LVM on LUKS, *seerver* flavour works.
6.18.2-4 Test OK on my workstation, both server and desktop flavour. suspend-resume and hibernate-restore both OK (Improvement over -3 where restore failed for desktop flavour.) / I note though that this system behaves very sluggish after restore from hibernation, sddm slow to come up, long time to log in, i click the already open firefox bit it takes a half minute to switch to that desktop instance, more seconds to get a reaction from firefox. After resting a while all works OK again. All while gkrellm do not report much disk nor CPU activity. This is same behavior as 6.6 kernels last half year. But 6.12 kernels do not show this problem. And the problem only shows on this one machine. Weird. Anyway, from 6.6 kernel this is no regression /
Removed all kernel-stable, installed 6.18.2-4 . Made sure the additional statements as given in Comment 162 are in the boot command line. Rebooted and no obvious problems seen in network conncetions and LO-files and youtube.
Tested again, but this time without the radeon and amdgpu statements in the boot command line, and sure, X does not start.
@Herman: desktop flavour? Probably not same issue, but if you like to test, try also the server flavour. On my T510, server flavour works but desktop flavour stalls quickly with black screen. Old nvidia GPU, nouveau/modesetting.
(In reply to Morgan Leijström from comment #179) > On my T510, server flavour works but desktop > flavour stalls quickly with black screen. Old nvidia GPU, > nouveau/modesetting. there is a pre 6.18.3, here: https://download.copr.fedorainfracloud.org/results/ghibo/mageia9-bonus/mageia-9-x86_64/09971813-kernel-stable-testing/
Yes the 6.18.3-0 works on the T510, both desktop and server flavour. That was also true for 6.18.2-3, but desktop 6.18.2-4 failed. [ettan@localhost ~]$ uname -a Linux localhost 6.18.3-desktop-0.stabletesting.mga9 #1 SMP PREEMPT_DYNAMIC Sat Jan 3 13:46:07 UTC 2026 x86_64 GNU/Linux --- 32 bit OK desktop 6.18.2-4 on Thinkpad T43 incl suspend and hibernate
(In reply to Morgan Leijström from comment #179) > @Herman: desktop flavour? Probably not same issue, but if you like to test, > try also the server flavour. On my T510, server flavour works but desktop > flavour stalls quickly with black screen. Old nvidia GPU, > nouveau/modesetting. Installed server flavor alongside desktop, rebooted with the radeon and amdgpu statements. Works OK. $ uname -a Linux xxx.yyy.zzz 6.18.2-server-4.stabletesting.mga9 #1 SMP PREEMPT_DYNAMIC Wed Dec 31 13:09:46 UTC 2025 x86_64 GNU/Linux. Coming back after rebooting without the radeon and amdgpu statements.
(In reply to Herman Viaene from comment #182) > (In reply to Morgan Leijström from comment #179) > > @Herman: desktop flavour? Probably not same issue, but if you like to test, > > try also the server flavour. On my T510, server flavour works but desktop > > flavour stalls quickly with black screen. Old nvidia GPU, > > nouveau/modesetting. > > Installed server flavor alongside desktop, rebooted with the radeon and > amdgpu statements. Works OK. > $ uname -a > Linux xxx.yyy.zzz 6.18.2-server-4.stabletesting.mga9 #1 SMP PREEMPT_DYNAMIC > Wed Dec 31 13:09:46 UTC 2025 x86_64 GNU/Linux. > Coming back after rebooting without the radeon and amdgpu statements. Removed the radeon and amdgpu statements from the boot command line and get stuck on the same problem, X does noet start.
Testing 6.18.3-1 x86_64 from backports testing. Three machines, testing at least boot to desktop, suspend, hibernation My laptop with nvidia470: full OK, only desktop flavour tested. My workstation, AMD GPU: full OK except the regression from 6.12 that it is extremely sluggish for minutes after restore from hibernation. (since a year also 6.6 kernels shows this behaviour but interestingly not 6.12) Additionally, desktop flavour takes several more seconds between grub and when monitor returns from black to enter LUKS key. I have a feeling server is a bit snappier than desktop overall, strange. My T510: desktop flavour full OK. *server flavour fail* with kernel panic early at boot, before to enter LUKS key: /dev/root: Can´t open blockdev List of bdev filesystems: ext3 ext4 ext2 cramfs btrfs Then kernel panic message and call trace. So this is a regression from 6.18.3-0 from fedorainfracloud
These flip-flop for T510 between -server and -desktop panic lead to think there is something wrong elsewhere, maybe in the generation of the dracut initramfs images which may be corrupt. The infracloud kernel 6.18.3-0 and current 6.18.3-1 were exactly the SAME version, same code basis except two extra patches for rtw88 and r8150 cards (in the 6.18.3-1) that shouldn't affect your hardware.
6.18.3-1.mga9 is also same as current cauldron one.
(In reply to Giuseppe Ghibò from comment #185) > These flip-flop for T510 between -server and -desktop panic lead to think > there is something wrong elsewhere, maybe in the generation of the dracut > initramfs images which may be corrupt. The infracloud kernel 6.18.3-0 and > current 6.18.3-1 were exactly the SAME version, same code basis except two > extra patches for rtw88 and r8150 cards (in the 6.18.3-1) that shouldn't > affect your hardware. I suggest to Morgan run as root dracut --regenerate-all --force And test again after reboot
Hah! Thanks. Disk intermittently full and our tools do not report that all way to user... I thought it was OK because /boot afterwards was 93%, earlier when this have happened boot ended up 100% We really should get the tools to warn when installing kernel fails. Uh, and I already reported this problem recently, Bug 34546... ... dracut: *** Creating image file '/boot/initrd-6.18.3-server-1.stabletesting.mga9.img' *** dracut: dracut: using auto-determined compression method 'gzip' cp: error writing '/boot/initrd-6.18.3-server-1.stabletesting.mga9.img': No space left on device dracut: dracut: creation of /boot/initrd-6.18.3-server-1.stabletesting.mga9.img failed ... ...and it just continues with next... ... dracut: *** Creating initramfs image file '/boot/initrd-6.6.116-desktop-1.mga9.img' done ***
§ Now my T510 pass tests on both flavours :-) § Also our Aspire717 pass tests. § i586 desktop now also tested OK on my Thinkpad T43. * Are there any known problems left? * Something specific to test? * We just continue testing on more machines, then finally release?
(In reply to Morgan Leijström from comment #189) > § Now my T510 pass tests on both flavours :-) > > § Also our Aspire717 pass tests. > > § i586 desktop now also tested OK on my Thinkpad T43. > > > * Are there any known problems left? Yes, I still need to fiddle with the command line as per Comment 162 above. Agree this is an older configuration, but a normal user is lost here if the HW is what I have (and variants alongside??) > > * Something specific to test? > > * We just continue testing on more machines, then finally release?
(In reply to Herman Viaene from comment #190) > (In reply to Morgan Leijström from comment #189) > > > > * Are there any known problems left? > > Yes, I still need to fiddle with the command line as per Comment 162 above. > Agree this is an older configuration, but a normal user is lost here if the > HW is what I have (and variants alongside??) I think we can not demand the newer kernels to work on all old hardware. I am thinking no real newbies enable and use backports repo, and if they install backport kernel they should have read our wiki page about it. So I made the section there with this issue: https://wiki.mageia.org/en/Kernel_flavours#Backport_incompatibilities We can add more there as they are detected. We will probably release updated backport kernel anyway, with the issue fixed or not. If it fail, user always have the previous working kernel to boot anyway - That is the power of backport... I think we should not delay releasing backport because of this. But please keep testing on more machines a couple days first. --- @Herman, have you tested that machine with Mageia 10/Cauldron? which kernel(s) work, and do they need same workaround? Do even the installer/live boot up usable? (Possible errata10 item)
> I have a feeling server is a bit snappier than desktop overall, strange. This is a pre-release, should be a lot snappier: https://download.copr.fedorainfracloud.org/results/ghibo/mageia9-bonus/mageia-9-x86_64/09978229-kernel-stable-testing/
Hi all, It would be nice if using flags could be tested in some kernel updates, like this one. Flags can be selected and set below the Source RPM: and Status comment: fields Using flags would give every tester the opportunity to set an individual "OK" for successful tests, by setting the test_passed_mga9_64 flag to "+": test_passed_mga9_64 + (For 32bit, there is the test_passed_mga9_32 flag) But also to set an individual "NOT_OK" by setting: test_passed_mga9_64 - That would give a better overview of the test results. Of course, when the kernel to be tested is replaced by a newer one, all + and - flags should be removed (or set to "?"), to be created again when the newer one is tested. See also https://wiki.mageia.org/en/Bugzilla#Flags and specifically https://wiki.mageia.org/en/Bugzilla#test_passed_mga9_64 I understand Morgan and Herman already tested 6.18.3-1.mga9, requesting them to + or - the test_passed_mga9_64 flag. Btw, I started obsoleting all comments about older kernels than 6.18.3-1.mga9, but that takes way too long. This report should be cloned for kernel 6.18.3-1.mga9, and then this report closed.
Summary: Backport kernel => Backport kernel 6.18.3-1.mga9Flags: (none) => test_passed_mga9_64?(fri), test_passed_mga9_64?(herman.viaene)CC: (none) => marja11
>This report should be cloned for kernel 6.18.3-1.mga9, and then this report closed. Yes.. i will, another day. But not so specific, rather 6.18.x @ Giuseppe, I assume we abandon 6.12? I see 6.18.3-2 already in backport testing, should we spend time on testing that, or wait for next? - I see 6.18.4 building for mga10. Please can you open a bug for the nvidia470 already in testing - the one updated for dkms with 6.18, I tested it OK in Comment 184
> > @ Giuseppe, > > I assume we abandon 6.12? Let's hold 6.12 for a while, since it's an indipendent series from 6.18 it wouldn't hurt. Then we'll see. > > I see 6.18.3-2 already in backport testing, should we spend time on testing > that, or wait for next? - I see 6.18.4 building for mga10. Yes, 6.18.4 is in stock for mga10. I'll sync kernel-stable-testing in backports to it soon (but version is pretty close to 6.18.3-2.mga9).
OK I will keep 6.12 mentioned nvidia470 in Bug 34961 This kernel backport continues in Bug 34962 *** This bug has been marked as a duplicate of bug 34962 ***
Status: NEW => RESOLVEDResolution: (none) => DUPLICATESummary: Backport kernel 6.18.3-1.mga9 => Backport kernelFlags: test_passed_mga9_64?(fri), test_passed_mga9_64?(herman.viaene) => test_passed_mga9_64-, test_passed_mga9_64-