Have just spent three hours on this trying to get kernel-linus to come to terms with nvidia, trying all the things done in the past and getting nowhere.  

Removed all the kernel-linus packages and started installing again:
# urpmi kernel-linus-5.4.12-1.mga7
    http://ftp.fi.muni.cz/pub/linux/mageia/distrib/7/x86_64/media/core/updates_testing/kernel-linus-5.4.12-1.mga7-1-1.mga7.x86_64.rpm
installing kernel-linus-5.4.12-1.mga7-1-1.mga7.x86_64.rpm from /var/cache/urpmi/rpms
Preparing...                     #############################################
      1/1: kernel-linus-5.4.12-1.mga7
                                 #############################################
cannot get info for device (6:0:0:1) at /usr/lib/libDrakX/detect_devices.pm line 272.
Creating: target|kernel|dracut args|basicmodules 
remove-boot-splash: Format of /boot/initrd-5.4.12-1.mga7.img not recognized
<why does this come up?>
nvidia-current (430.64-2.mga7.nonfree): Already installed on this kernel.
vboxadditions (6.0.16-1.mga7): Already installed on this kernel.
virtualbox (6.0.16-1.mga7): Already installed on this kernel.
xtables-addons (3.7-1.mga7): Already installed on this kernel.
You should restart your computer for kernel-linus-5.4.12-1.mga7

# dkms status
This issues many warnings about differences between built and installed modules,
e.g.
nvidia-current, 430.64-2.mga7.nonfree, 5.4.12-desktop-1.mga7, x86_64: installed 
nvidia-current, 430.64-2.mga7.nonfree, 5.4.12-1.mga7, x86_64: installed  (WARNING! Diff between built and installed module!) (WARNING! Diff between built and installed module!) (WARNING! Diff between built and installed module!) (WARNING! Diff between built and installed module!)

On the previous pass the kmods are rebuilt during the boot sequence so those differences should have been taken care of - no?

vboxadditions, 6.0.16-1.mga7, 5.4.12-desktop-1.mga7, x86_64: installed 
vboxadditions, 6.0.16-1.mga7, 5.4.12-1.mga7, x86_64: installed  (WARNING! Diff between built and installed module!) (WARNING! Diff between built and installed module!) (WARNING! Diff between built and installed module!)

CC: (none) => tarazed25

Given the part about nvidia, this is on real hardware? If so, the virtualbox
guest additions should not be installed. The guest additions are only for use
in a virtualbox guest, not on the host.

Is kernel-linus-devel-latest installed?

CC: (none) => davidwhodgins

Yes, this is real hardware - always test kernels on real hardware.  Leave virtualbox for others.  I take your point about vbox-guest-additions - I just blindly check all the boxes when installing and have never been sure about what belongs where, not being a habitual vbox user.

Tried this kernel on an nvidia390 machine.  The nvidia390 kmod was built and installed on reboot but the infamous "There has been a display driver change - you need to reboot for it to take effect" came up but rebooting does not help.  Installed nvidia390 from a virtual console, ran "dracut -f" and rebooted again.  Back into the display driver change loop.

No more time to spare on this, except it might be worth removing the vbox-guest-additions files and seeing what difference that makes.  Otherwise I am finished.
Note that vbox-guest-additions has never interfered, as far as I can see, in the ossibly hundreds of kernel tests I have performed.  Anyway, here goes.

Nope.  It still ends up with nouveau.  Every time nvidia is installed via drakx11  it fails to "take" and the system has to be rebooted into nouveau.  Finished with this update for the time being.  It would be interesting to see how non-nvidia machines behave and other testers' nvidia systems.

@Dave: yes, always install the devel packages, all the packages in fact.  source, etc can always be removed later.

I guess "nokmsboot" has gone missing from kernel command line, preventing the nvidia driver to load properly

OK, shall check that and add it if necessary.

Yes, nokmsboot comes and goes.  Ran drakx11 and specified nvidia.  dracut -f.  Reboot.  nokmsboot is there.  Display driver change detected - reboot.  nokmsboot has disappeared.  Added it and continued the boot.  Logged in to the desktop and nouveau was back.  ??

I shall try the house-cleaning approach tomorrow.  No more tonight.

Try removing /etc/X11/xorg.conf

On one laptop I now have which uses the intel i915 module, I found that removing
the file (even though the file called for the i915 module), that fixed it's
problems. No idea why that works.

Yes, that does sound strange.  Shall try it, maybe, after looking at a few other things.  

Meanwhile all the usual tests run fine; stress c/i/m/d, glmark2, common desktop applications, TV viewing via vlc, sound and video.

Hmm.  Checked what was installed and found the devel package WAS missing - apologies Dave - even though I was sure I had installed it manually.  Another senior moment.

Rectified that.

Removed dkms-nvidia-current and then reinstalled it.

Error messages in the dracut section - possibly because  the linus kernel is missing bells and whistles.

No nvidia after reboot - ran drakx11 to install the nvidia x11 driver - rebooted.  Reverts to nouveau after the display driver change message.  nokmsboot was part of the kernel command each time - no need to edit.

Retraced the loop a few times, removing xorg.conf one time.  No progress.

AMD Phenom II X4 910, 8GB RAM, Radeon HD 8490 graphics, atheros wifi, 64-bit Plasma system.

Installed kernel-linus and kernel-linus-latest. Rebooted to a working desktop. Tried all the usual suspects with no issues noted. Looks OK on this hardware.

Sorry Len, but when I recently upgraded the processor on this system I also switched out the old nvidia Geforce 210 video card, so I can't test for nvidia issues any more.

CC: (none) => andrewsfarm

Dell inspiron 5100, 32-bit P4, 2GB RAM, Radeon RV200 graphics, old atheros wifi, 32-bit Xfce system.

It has been a couple of weeks since I booted this system. After getting all the pending updates, I installed the 32-bit kernel-linus and rebooted.

Unfortunately, this kernel shows the same graphical symptoms that were seen with some 5.4 series versions of kernel-desktop. See Bug 25897.  

In its current state, it is unusable on this hardware.

I too see problem with Nvidia driver.
(nvidia-current - NVIDIA kernel module for GeForce 635 and later cards​)

System was running kernel-desktop-5.4.12-1.mga7
Installed kernel-linus, kernel-linus-latest and corresponding devel packages.
After reboot it fail to switch to graphical.

Changing to nouveau, i can boot and use the computer as normal.

Thunderbird, LibreOffice, FreeCAD, Ktorrent, Syncthing, Nextcloud client...
Video with sound in Firefox
VirtualBox 6.0.16 running MSW7 incl host folder sharing, USB2 flash stick, firefox video with sound.

  ( One exeption, maybe related to nvidia: BOINC simply fail - maybe it try to use nvidia as it use to (it had one GPU project running i think) but fail and also fail telling me (a bug in it IMO); issuing "boincmgr" in konsole simply exits with no message.  I let boincmgr start the client after i log in to desktop. I leave that problem out for now. )


__SYSTEM
My machine "svarten": Mainboard: Sabertooth P67, CPU: i7-3770, RAM 16G, Nvidia GTX760 (GK104) using proprietary driver GeForce 420 and later

Disk&Filesystem: SSD with EFI and ext 4 /boot, then an encrypted partition for LVM, containing swap and ext4 /home & /

__Investigating

# dkms status
virtualbox, 6.0.16-1.mga7, 5.4.12-1.mga7, x86_64: installed 
virtualbox, 6.0.16-1.mga7, 5.4.12-desktop-1.mga7, x86_64: installed 
nvidia-current, 430.64-2.mga7.nonfree, 5.4.12-1.mga7, x86_64: installed 
nvidia-current, 430.64-2.mga7.nonfree, 5.4.12-desktop-1.mga7, x86_64: installed 
nvidia-current, 430.64-2.mga7.nonfree, 5.4.10-desktop-1.mga7, x86_64: installed 
virtualbox, 6.0.16-1.mga7, 5.4.12-desktop-1.mga7, x86_64: installed-binary from 5.4.12-desktop-1.mga7
virtualbox, 6.0.14-2.mga7, 5.4.10-desktop-1.mga7, x86_64: installed-binary from 5.4.10-desktop-1.mga7


From journal:

jan 29 07:56:46 svarten.tribun drakrpm[16852]: transaction on / (remove=0, install=0, upgrade=4)
jan 29 07:56:47 svarten.tribun [RPM][16852]: Transaction ID 5e312cae startedd
jan 29 07:56:52 svarten.tribun bootloader-config[14135]: running: /sbin/modprobe dm_zero
...
jan 29 07:56:54 svarten.tribun bootloader-config[14135]: running: /sbin/modprobe loop
jan 29 07:56:54 svarten.tribun kernel: loop: module loaded
jan 29 07:56:54 svarten.tribun bootloader-config[14135]: running: mkinitrd -f /boot/initrd-5.4.12-1.mga7.img 5.4.12-1.mga7
jan 29 07:57:14 svarten.tribun bootloader-config[14135]: current labels: Mageia Mageia (5.4.12-desktop-1.mga7) 7 Mageia (5.4.12-desktop-1.mga7) 7 (recovery mode) Mageia (5.4.>
jan 29 07:57:14 svarten.tribun bootloader-config[14135]: running: update-grub2
...
jan 29 07:57:18 svarten.tribun bootloader-config[14135]: update-grub2 logs: Generera konfigurationsfil för grub …
...
jan 29 07:57:18 svarten.tribun bootloader-config[14135]: running: grub2-set-default gnulinux-simple-5a7166ca-67b8-4792-816e-b58dba1e5f27
jan 29 07:57:18 svarten.tribun [RPM][16852]: install kernel-linus-5.4.12-1.mga7-1-1.mga7.x86_64: success
jan 29 07:57:21 svarten.tribun [RPM][16852]: install kernel-linus-devel-5.4.12-1.mga7-1-1.mga7.x86_64: success
jan 29 07:57:21 svarten.tribun [RPM][16852]: install kernel-linus-devel-latest-5.4.12-1.mga7.x86_64: success
jan 29 07:57:21 svarten.tribun [RPM][16852]: install kernel-linus-latest-5.4.12-1.mga7.x86_64: success

...
jan 29 07:59:22 svarten.tribun [RPM][16852]: install kernel-linus-5.4.12-1.mga7-1-1.mga7.x86_64: success
jan 29 07:59:23 svarten.tribun [RPM][16852]: install kernel-linus-5.4.12-1.mga7-1-1.mga7.x86_64: success
jan 29 07:59:23 svarten.tribun [RPM][16852]: install kernel-linus-devel-5.4.12-1.mga7-1-1.mga7.x86_64: success
jan 29 07:59:23 svarten.tribun [RPM][16852]: install kernel-linus-devel-latest-5.4.12-1.mga7.x86_64: success
jan 29 07:59:23 svarten.tribun [RPM][16852]: install kernel-linus-latest-5.4.12-1.mga7.x86_64: success

CC: (none) => fri

Regarding BOINC in previous comment: it do work using nouveau driver (except GPU calculations) with the desktop kernel, so why it failed with linus and nouveau is strange.

As this one is not validated yet, and more important security fixes has emerged, I've rolled this up to the upstream 5.4.16


SRPMS:
kernel-linus-5.4.16-1.mga7.src.rpm


i586:
kernel-linus-5.4.16-1.mga7-1-1.mga7.i586.rpm
kernel-linus-devel-5.4.16-1.mga7-1-1.mga7.i586.rpm
kernel-linus-devel-latest-5.4.16-1.mga7.i586.rpm
kernel-linus-doc-5.4.16-1.mga7.noarch.rpm
kernel-linus-latest-5.4.16-1.mga7.i586.rpm
kernel-linus-source-5.4.16-1.mga7-1-1.mga7.noarch.rpm
kernel-linus-source-latest-5.4.16-1.mga7.noarch.rpm


x86_64:
kernel-linus-5.4.16-1.mga7-1-1.mga7.x86_64.rpm
kernel-linus-devel-5.4.16-1.mga7-1-1.mga7.x86_64.rpm
kernel-linus-devel-latest-5.4.16-1.mga7.x86_64.rpm
kernel-linus-doc-5.4.16-1.mga7.noarch.rpm
kernel-linus-latest-5.4.16-1.mga7.x86_64.rpm
kernel-linus-source-5.4.16-1.mga7-1-1.mga7.noarch.rpm
kernel-linus-source-latest-5.4.16-1.mga7.noarch.rpm

Summary: Update request: kernel-linus-5.4.12-1.mga7 => Update request: kernel-linus-5.4.16-1.mga7

Host: difda Kernel: 5.4.16-1.mga7 x86_64
Mobo: MSI model: Z97-G43 (MS-7816) v: 3.0
Quad Core: Intel Core i7-4790 type: MT MCP
Device-1: NVIDIA GM204 [GeForce GTX 970] driver: nouveau

After half a dozen passes gave up on nvidia.  xorg.conf specified nvidia initially but at boot time the nvidia module is not loaded.  Tried various things like xdriver=nvidia kernel parameter (don't know if that is correct name) but always produced the "There has been a display driver change message - reboot" message.  Tried drakx11 : `dracut -f` : reboot at cli, to no avail.

Apart from that the desktop is running fine and passes the stress tests.
Leaving it up for a while.

Same hardware as Comment 12. 

I had replaced the hard drive, and did a clean network install before beginning tests. The system works OK with kernel-desktop 5.4.12. I updated to kernel-desktop 5.4.16 first, which also works nicely, before installing kernel-linus 5.4.16.

Kernel-linus 5.4.16 doesn't like the Radeon RV200 graphics any more than the version from Comment 12 did. I will be removing it before continuing.

Like previous version Nvidia and boincmgr fail, rest seem OK, running it currently on my workstation, plasma.
Did no other testing or trying why.
I can check logs and try something if told what.

Correction: boinc do start if i give it a few minutes. I guess it tries to use nvidia and give up after a long timeout.

Sorry,

Putting this on hold, there is a regression in 5.4.16 found upstream that needs fixing

Keywords: (none) => feedback

OK. A bit feedback anyway:  After some hours withput me, chewing boinc for 100% CPU, it had locked up;
Now i got home after a party, screen did not turn on when i touched the keyboard nor mouse.  No reaction on ctrl-alt del och backspace repeated. No response on power key pressed 2 seconds.  Had to press >4 s for power off.

After reboot, i found from prevoous run in journal:

feb 01 21:27:28 svarten.tribun kernel: nouveau 0000:01:00.0: fifo: PBDMA0: 04000000 [ACQUIRE] ch 2 [007f88d000 Xorg[5585]] subc 0 mthd 001c data 00001004
feb 01 21:27:28 svarten.tribun kernel: nouveau 0000:01:00.0: fifo: PBDMA0: 04000000 [ACQUIRE] ch 2 [007f88d000 Xorg[5585]] subc 0 mthd 001c data 00001004
feb 01 21:27:28 svarten.tribun kernel: nouveau 0000:01:00.0: fifo: PBDMA0: 04000000 [ACQUIRE] ch 2 [007f88d000 Xorg[5585]] subc 0 mthd 001c data 00001004
feb 01 21:27:56 svarten.tribun kernel: ata10.00: exception Emask 0x10 SAct 0x30000 SErr 0x400000 action 0x6 frozen
feb 01 21:27:56 svarten.tribun kernel: ata10.00: irq_stat 0x08000000, interface fatal error
feb 01 21:27:56 svarten.tribun kernel: ata10: SError: { Handshk }
feb 01 21:27:56 svarten.tribun kernel: ata10.00: failed command: WRITE FPDMA QUEUED
feb 01 21:27:56 svarten.tribun kernel: ata10.00: cmd 61/00:80:90:71:19/0a:00:4e:00:00/40 tag 16 ncq dma 1310720 ou
                                                res 40/00:8c:90:7b:19/00:00:4e:00:00/40 Emask 0x10 (ATA bus error)
feb 01 21:27:56 svarten.tribun kernel: ata10.00: status: { DRDY }
feb 01 21:27:56 svarten.tribun kernel: ata10.00: failed command: WRITE FPDMA QUEUED
feb 01 21:27:56 svarten.tribun kernel: ata10.00: cmd 61/00:88:90:7b:19/06:00:4e:00:00/40 tag 17 ncq dma 786432 out
                                                res 40/00:8c:90:7b:19/00:00:4e:00:00/40 Emask 0x10 (ATA bus error)
feb 01 21:27:56 svarten.tribun kernel: ata10.00: status: { DRDY }
feb 01 21:27:56 svarten.tribun kernel: ata10: hard resetting link
feb 01 21:27:56 svarten.tribun kernel: ata10: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
feb 01 21:27:56 svarten.tribun kernel: ata10.00: configured for UDMA/133
feb 01 21:27:56 svarten.tribun kernel: ata10: EH complete

All lines containing "noveau" or "ata10.00:" was marked red.
The UDMA/133 line is really weird; IIRC the mainboard does not even have a PATA connector...

Probably better info, en earlier line:

feb 01 15:31:10 svarten.tribun kernel: nouveau 0000:01:00.0: DRM: core notifier timeout

This one is now at 5.4.20 so it contains all security fixes I have backported in the core kernel


SRPMS:
kernel-linus-5.4.20-1.mga7.src.rpm


i586:
kernel-linus-5.4.20-1.mga7-1-1.mga7.i586.rpm
kernel-linus-devel-5.4.20-1.mga7-1-1.mga7.i586.rpm
kernel-linus-devel-latest-5.4.20-1.mga7.i586.rpm
kernel-linus-doc-5.4.20-1.mga7.noarch.rpm
kernel-linus-latest-5.4.20-1.mga7.i586.rpm
kernel-linus-source-5.4.20-1.mga7-1-1.mga7.noarch.rpm
kernel-linus-source-latest-5.4.20-1.mga7.noarch.rpm


x86_64:
kernel-linus-5.4.20-1.mga7-1-1.mga7.x86_64.rpm
kernel-linus-devel-5.4.20-1.mga7-1-1.mga7.x86_64.rpm
kernel-linus-devel-latest-5.4.20-1.mga7.x86_64.rpm
kernel-linus-doc-5.4.20-1.mga7.noarch.rpm
kernel-linus-latest-5.4.20-1.mga7.x86_64.rpm
kernel-linus-source-5.4.20-1.mga7-1-1.mga7.noarch.rpm
kernel-linus-source-latest-5.4.20-1.mga7.noarch.rpm



For testers, its good eough if it works with free gpu drivers ...

Summary: Update request: kernel-linus-5.4.16-1.mga7 => Update request: kernel-linus-5.4.20-1.mga7
Keywords: feedback => (none)

Installed:
- kernel-linus-5.4.20-1.mga7-1-1.mga7.x86_64
- kernel-linus-devel-5.4.20-1.mga7-1-1.mga7.x86_64
- kernel-linus-devel-latest-5.4.20-1.mga7.x86_64
- kernel-linus-doc-5.4.20-1.mga7.noarch
- kernel-linus-latest-5.4.20-1.mga7.x86_64
- kernel-linus-source-5.4.20-1.mga7-1-1.mga7.noarch
- kernel-linus-source-latest-5.4.20-1.mga7.noarch

nvidia driver kmod built during installation as well as virtualbox and xtables-addons.

As tmb anticipated, the nvidia driver did not "take" and the second reboot failed again.  Logged in to virtual console and installed nouveau graphics driver.  Rebooted straight to the Mate desktop.

Kernel: 5.4.20-1.mga7 x86_64
Intel Core i7-4790 type: MT MCP
Realtek RTL8111/8168/8411 PCI Express Gigabit Ethernet : driver: r8169
NVIDIA GM204 [GeForce GTX 970] driver: nouveau
Ran stress tests including glmark2, teapot and glxspheres.  Launched virtualbox client concurrently but it took a while.  firefox and thunderbird running at the same time.  Sound and video working.  Other desktop applications running without problems.

All good here.

An additional test nearly burnt out the cpu.  Performed a local build of blender which went on for ages with all cores firing at 100% and internal temperatures hitting 100°C.  It succeeded though, but all I can say is "don't try this at home".

@ Len @ c#25: do you mean throttling do not work - that it should have slowed down the CPU before 100C ?


64 bit OK here.
Machine "svarten": Mainboard: Sabertooth P67, CPU: i7-3770, RAM 16G, Nvidia GTX760 (GK104)

First i used MCC to change fron nvidia to nouveau.  Then:

- kernel-linus-5.4.20-1.mga7-1-1.mga7.x86_64
- kernel-linus-devel-5.4.20-1.mga7-1-1.mga7.x86_64
- kernel-linus-devel-latest-5.4.20-1.mga7.x86_64
- kernel-linus-latest-5.4.20-1.mga7.x86_64

-reboot-

__Tests:
No issues booting.
Plasma desktop with Libreoffice, FreeCAD, firefox (the one in testing) with video
VirtualBox running MSW7 client with USB and internet working, with video in firefox while BOINC chewing on using all cores on host.   My machine have a water cooler system so never get even to 60C core temp...

Forgot to say my system is using LUKS containing LVM so that works too, as usual :)

@Morgan re throttling:
Thanks for the heads-up.  Checked cpupower.service and found it dead.  So yes throttling was not working.
Have restarted the service now - better late than never.

Kernel: 5.4.20-1.mga7 x86_64

Installed linus kernel on this hardware:
Alienware X51 
Intel Core i7-2600 type: MT MCP
NVIDIA GF114 [GeForce GTX 555]
Qualcomm Atheros AR9285 Wireless Network Adapter driver: ath9k 

Reboot foundered on nvidia390 so installed nouveau from a console and rebooted to desktop.  No network.  Manually configured wifi.

Kernel: 5.4.20-1.mga7 x86_64

# dkms status
.....
nvidia390, 390.132-2.mga7.nonfree, 5.4.17-desktop-1.mga7, x86_64: installed 
nvidia390, 390.132-2.mga7.nonfree, 5.4.20-1.mga7, x86_64: installed 

Running fine for a couple of hours now.

An update for this issue has been pushed to the Mageia Updates repository.

https://advisories.mageia.org/MGASA-2020-0089.html

Resolution: (none) => FIXED
Status: NEW => RESOLVED