Problem with kernel 6.15.3-1-siduction-amd64

[Geier0815]

Maybe a stupid question but for what reason do you need such an old kernel (6.1.0-1 released end of 2022)?

[michaa7]

I have no time at the moment to fix that package, maybe until next week i can't do that.

Hi towo,

thanks for letting us know!

As far as I am concerned, I am happy when the fix is available in the foreseeable future, in weeks. Thanks for even caring about proprietary legacy drivers.

[Pip]

Maybe a stupid question but for what reason do you need such an old kernel (6.1.0-1 released end of 2022)?

A fine question! I hadn't even noticed that. apt auto-remove and the kernel remover don't think they're there, but the header files certainly are. I'll trim that. Thanks.

[Teriarch]

@Pip
Thanks to Geier0815 it all makes sense now. I can almost guarantee that after the old kernel
removal everything works fine. Unfortunately the link already expired, so I will renew it for
your convenience. And please report back about success or failure (you are so close now)...

https://wormhole.app/D1x576#VaPpC8kIpQjuecFyK6rblw
md5sum                                                    filename
d93464b119ea55f44006614a40848a4c  nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4c_amd64.deb

@towo
Thanks for clarifying!

[Pip]

I downloaded it before I reported back last time, that was the result with it. I'll give it another try tomorrow without the old headers.

[michaa7]

I don't know the state of your installation, but "without something old" does not seem to be the appropriate prerequisite to overcome modul building problems with the newest sources. An up-to-date system without any cruft could be.

[Pip]

I don't know the state of your installation, but "without something old" does not seem to be the appropriate prerequisite to overcome modul building problems with the newest sources. An up-to-date system without any cruft could be.

Quite so, quite so. And I didn't really think it would solve anything since the module was being built against the latest kernel with the latest headers.
However...
The module got built, the system is stable again!
Thanks to Teriarch, Geier0815 and towo for putting in the time, effort and "stupid" questions.

[Teriarch]

Glad to hear that you finally solved your problem (with a little help from your friends...).
And there are no stupid questions, just stupid answers!

[n4ai9i522]

Not sure if this thread is pertinent to the problem I am having.
Last working kernel is 6.14.10-1, as both 6.15.3-1 and now even 6.15.4-1 will not allow building nvidia-current module for them, the process will fail.

The following is just an excerpt of what I mean:

Code Select Expand

Autoinstall of module nvidia-current/550.163.01 for kernel 6.15.4-1-siduction-amd64 (x86_64)
Building module(s)..............(bad exit status: 2)                                                                 
Failed command:                                                                                                       
env NV_VERBOSE=1 make -j12 modules KERNEL_UNAME=6.15.4-1-siduction-amd64
                                                                                                                     
Error! Bad return status for module build on kernel: 6.15.4-1-siduction-amd64 (x86_64)
Consult /var/lib/dkms/nvidia-current/550.163.01/build/make.log for more information.                               
                                                           
Autoinstall on 6.15.4-1-siduction-amd64 failed for module(s) nvidia-current(10).

Error! One or more modules failed to install during autoinstall.
Refer to previous errors for more information.
run-parts: /etc/kernel/header_postinst.d/dkms exited with return code 1
Failed to process /etc/kernel/header_postinst.d at /var/lib/dpkg/info/linux-headers-6.15.4-1-siduction-amd64.postinst
line 11.
dpkg: error processing package linux-headers-6.15.4-1-siduction-amd64 (--configure):
installed linux-headers-6.15.4-1-siduction-amd64 package post-installation script subprocess returned error exit stat
us 1
Setting up linux-image-6.15.4-1-siduction-amd64 (6.15-4) ...
I: /initrd.img is now a symlink to boot/initrd.img-6.15.4-1-siduction-amd64
/etc/kernel/postinst.d/dkms:
The kernel is built without module signing facility, modules won't be signed

Autoinstall of module nvidia-current/550.163.01 for kernel 6.15.4-1-siduction-amd64 (x86_64)
Building module(s)..............(bad exit status: 2)
Failed command:
env NV_VERBOSE=1 make -j12 modules KERNEL_UNAME=6.15.4-1-siduction-amd64


[ro_sid]

For me it got built (for both kernels, the last one today). So there must be something peculiar in your environment.
Problem is: what?


Edit: Corrected the date (today, not yesterday)

[n4ai9i522]

I had the GPT guide me, and after sharing the nvidia make.log with it, here's part of its response:

❌ Root Cause:
The build fails due to missing or incompatible kernel symbols in Linux 6.15.x:

/var/lib/dkms/nvidia-current/550.163.01/build/nvidia/nv-dma.c: In function 'nv_dma_map_pages':
... warning: passing argument from incompatible pointer type
...
/var/lib/dkms/nvidia-current/550.163.01/build/nvidia/nv-pci.c: error: implicit declaration of function 'pci_enable_msi'; did you mean 'pci_enable_msix'?

🧨 Most critical error:

error: implicit declaration of function 'pci_enable_msi'

This means that the NVIDIA driver is using kernel APIs that have changed or been removed in kernel 6.15. These changes break compatibility with NVIDIA 550.163.01 — which was last known to support kernels up to ~6.8 or 6.9.
✅ Confirmed: NVIDIA 550.163.01 does not yet support Linux 6.15

As of now:

    NVIDIA hasn't updated their proprietary drivers to support Linux 6.15 kernel headers.

    The issue is due to removed/changed kernel functions, not your system setup.

[Geier0815]

Do you have the repo "deb https://packages.siduction.org/fixes unstable main contrib non-free non-free-firmware" activated? If not, do so

[n4ai9i522]

...I just discovered that all .list files under /etc/apt/sources.list.d/ have been renamed to *.list.bak so there's that

EDIT: they were just updated into .sources files. I added `contrib non-free non-free-firmware` near to `main` in fixes.sources and it found more packages to update, churning now.

EDIT 2: should I also add `contrib non-free non-free-firmware` to the other repos, for good measure?

EDIT 3: Thanks! I could boot into 6.15.4. Now virtualbox complains about a missing driver 

EDIT 4: virtualbox 7.1.10 fixed the issue

[ro_sid]

@n4ai9i522:
To "EDIT 2": Well, I did so - add the "rest" - and I have not experienced any negative effects from it.

[michaa7]

I have no time at the moment to fix that package, maybe until next week i can't do that.

Hi towo,

would be nice if you find some time to make a fixed nvidia-470 package.

OTOH, if, as you wrote, the  tp_smapi thing is the package to fix, is there already a DEBIAN bug report and will it get fixed there eventually? As Debian/experimental still provides only an 6.12xx kernel, I think it will last a year until Debian arrives at 6.15 Kernel source and gets aware of this problem.

So, I am in no hurry for a fixed nvidia-driver, but I definitivley would like to not being cut out forever ;-)


BTW: Isn't "tp-smapi" a thinkpad thing? Does this thing influence the nvidia driver for all other computers?