Problem with kernel 6.15.3-1-siduction-amd64

[Pip]

Thanks for that. It came up with an error

Code Select Expand

/var/lib/dkms/nvidia-tesla-470/470.256.02/build/nvidia-modeset/nvidia-modeset-linux.c: In function 'nvkms_kthread_q_callback':
/var/lib/dkms/nvidia-tesla-470/470.256.02/build/nvidia-modeset/nvidia-modeset-linux.c:721:9: error: implicit declaration of function 'timer_delete_sync' [-Werror=implicit-function-declaration]
  721 |         timer_delete_sync(&timer->kernel_timer);
I tried changing the nvidia-modeset-linux.c source so that it had

Code Select Expand

#include <linux/timer.h> in it just in case, but the same error turned up. After poking around for a bit, it looks like it is somewhat out of my league. Or would be if I'd every reached a league.

[Teriarch]

Well, someone is obviously working on it in the meantime
(maybe towo, and hopefully incorporates the patch):

$ apt-get download nvidia-tesla-470-kernel-dkms (now, 23:03h):
nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.3_amd64.deb

$ apt-get download nvidia-tesla-470-kernel-dkms (then, 17:00h):
nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.1_amd64.deb

But it's not working yet:
  Applying patch nvidia-470xx-fix-linux-6.14.patch...(bad exit status: 1)
  Failed command:
  patch -p1 < ./patches/nvidia-470xx-fix-linux-6.14.patch

But I'm sure, he will figure it out. Maybe I'll fix it in the meantime,
make it a race...

[Teriarch]

...and here we go:
<https://wormhole.app/vbNaAk#XPEJPPIdE2Qb3mN2eayW1Q>

md5sum:                                                   filename:
592eca374baacce37bd2e9ac4e3a82b9 nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.3b_amd64.deb

@Pip:
I had no problems installing the 3b version neither with kernel 6.15 nor with 6.14 (and all the other tesla-470 packages).
If it still fails for you, report back with some details...

and just in case (I'm getting ahead of myself...):

https://wormhole.app/5KWzrP#0ojWtHwfQ0MgJbCkzuCU-g
320ef59ed0825db1978c7c221a5b607b  nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4b_amd64.deb

https://wormhole.app/nA6apD#Vx3We1pMBLJgDrCFg53CKw
d1f58d1e79f53b9935958bc90ecd3c5d  nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.5b_amd64.deb

[Teriarch]

@towo (or so I guess)

the final patch has still problems (4 vs.4b:

  Applying patch use-kbuild-flags.patch...(bad exit status: 1)
  Failed command:
  patch -p1 < ./patches/use-kbuild-flags.patch

But you are closing in  (hint: EXTRA_CCFLAGS in wrong context after 6.15 patch, I guess)
I still have a working 5b version to offer...

[towo]

with your 5b version, i get:

Code Select Expand

towo:Defiant> patch -p1 < ./patches/use-kbuild-flags.patch
patching file Kbuild
Hunk #1 FAILED at 108.
1 out of 1 hunk FAILED -- saving rejects to file Kbuild.rej
patching file nvidia/nvidia.Kbuild
Hunk #1 succeeded at 94 (offset 2 lines).
patching file Makefile
patching file nvidia-modeset/nvidia-modeset.Kbuild
Hunk #1 succeeded at 79 (offset 2 lines).

if i apply

cc_version_check-gcc5.patch
bashisms.patch
0001-some-power-management-features-were-not-yet-in-Linux.patch
0033-refuse-to-load-legacy-module-if-IBT-is-enabled.patch
0034-fix-typos.patch
0037-import-pfn_valid-w-o-GPL-rcu_read_lock-unlock-from-v.patch
0042-Log-an-error-message-when-nv_mem_client_init-fails-d.patch
0043-backport-follow_pfn-changes-from-550.90.07.patch
0045-let-the-virt_addr_valid-macro-use-nv_pfn_valid-on-pp.patch
0046-virt_addr_valid-wants-a-void-argument.patch
kernel-6.12.patch
kernel-6.13.patch
nvidia-470xx-fix-linux-6.14.patch
fragile-ARCH.patch
conftest-verbose.patch
use-kbuild-compiler.patch

before.

[Teriarch]

That's strange. Did you check the 4b version, because last time I checked,
it worked, but let's try again... hang on...

-----------------------------------------------------
Regular 'sudo apt-get sudo apt-get install nvidia-tesla-470-driver' install with
nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4_amd64.deb:

Loading new nvidia-tesla-470/470.256.02 DKMS files...
Building for 6.15.3-1-siduction-amd64

Building initial module nvidia-tesla-470/470.256.02 for 6.15.3-1-siduction-amd64
The kernel is built without module signing facility, modules won't be signed

Applying patch cc_version_check-gcc5.patch... done.
Applying patch bashisms.patch... done.
Applying patch 0001-some-power-management-features-were-not-yet-in-Linux.patch... done.
Applying patch 0033-refuse-to-load-legacy-module-if-IBT-is-enabled.patch... done.
Applying patch 0034-fix-typos.patch... done.
Applying patch 0037-import-pfn_valid-w-o-GPL-rcu_read_lock-unlock-from-v.patch... done.
Applying patch 0042-Log-an-error-message-when-nv_mem_client_init-fails-d.patch... done.
Applying patch 0043-backport-follow_pfn-changes-from-550.90.07.patch... done.
Applying patch 0045-let-the-virt_addr_valid-macro-use-nv_pfn_valid-on-pp.patch... done.
Applying patch 0046-virt_addr_valid-wants-a-void-argument.patch... done.
Applying patch kernel-6.12.patch... done.
Applying patch kernel-6.13.patch... done.
Applying patch nvidia-470xx-fix-linux-6.14.patch... done.
Applying patch nvidia-470xx-fix-linux-6.15.patch... done.
Applying patch fragile-ARCH.patch... done.
Applying patch conftest-verbose.patch... done.
Applying patch use-kbuild-compiler.patch... done.
Applying patch use-kbuild-flags.patch...(bad exit status: 1)
Failed command:
patch -p1 < ./patches/use-kbuild-flags.patch

Error! Application of patch use-kbuild-flags.patch failed.
-----------------------------------------------------
After the error I did a
sudo dpkg -i nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4b_amd64.deb
with the following result:

(Lese Datenbank ... 312816 Dateien und Verzeichnisse sind derzeit installiert.)
Vorbereitung zum Entpacken von nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4b_amd64.deb ...
Deleting module nvidia-tesla-470/470.256.02 completely from the DKMS tree.
Entpacken von nvidia-tesla-470-kernel-dkms (470.256.02-7.siduction.4b) über (470.256.02-7.siduction.4) ...
nvidia-tesla-470-kernel-dkms (470.256.02-7.siduction.4b) wird eingerichtet ...
Loading new nvidia-tesla-470/470.256.02 DKMS files...
Building for 6.15.3-1-siduction-amd64

Building initial module nvidia-tesla-470/470.256.02 for 6.15.3-1-siduction-amd64
The kernel is built without module signing facility, modules won't be signed

Applying patch cc_version_check-gcc5.patch... done.
Applying patch bashisms.patch... done.
Applying patch 0001-some-power-management-features-were-not-yet-in-Linux.patch... done.
Applying patch 0033-refuse-to-load-legacy-module-if-IBT-is-enabled.patch... done.
Applying patch 0034-fix-typos.patch... done.
Applying patch 0037-import-pfn_valid-w-o-GPL-rcu_read_lock-unlock-from-v.patch... done.
Applying patch 0042-Log-an-error-message-when-nv_mem_client_init-fails-d.patch... done.
Applying patch 0043-backport-follow_pfn-changes-from-550.90.07.patch... done.
Applying patch 0045-let-the-virt_addr_valid-macro-use-nv_pfn_valid-on-pp.patch... done.
Applying patch 0046-virt_addr_valid-wants-a-void-argument.patch... done.
Applying patch kernel-6.12.patch... done.
Applying patch kernel-6.13.patch... done.
Applying patch nvidia-470xx-fix-linux-6.14.patch... done.
Applying patch nvidia-470xx-fix-linux-6.15.patch... done.
Applying patch fragile-ARCH.patch... done.
Applying patch conftest-verbose.patch... done.
Applying patch use-kbuild-compiler.patch... done.
Applying patch conftest-prefer-arch-headers.patch... done.
Applying patch use-kbuild-flags.patch... done.
Building module(s)........................ done.
Installing /lib/modules/6.15.3-1-siduction-amd64/updates/dkms/nvidia-tesla-470.ko
Installing /lib/modules/6.15.3-1-siduction-amd64/updates/dkms/nvidia-tesla-470-modeset.ko
Installing /lib/modules/6.15.3-1-siduction-amd64/updates/dkms/nvidia-tesla-470-drm.ko
Installing /lib/modules/6.15.3-1-siduction-amd64/updates/dkms/nvidia-tesla-470-uvm.ko
Installing /lib/modules/6.15.3-1-siduction-amd64/updates/dkms/nvidia-tesla-470-peermem.ko
Running depmod... done.
------------------------------------------------------------------------------------------
The 5b file is identical to 4b except for the version information in the debian control file.

If I remember correctly I ran into the same error as you did, so I not only corrected
the order of the patches in dkms.conf, but also the content of "use-kbuild-flags.patch" to
account for the context change of the 470xx 6.15 patch (which also was adapted) and make
the hunk succeed.

All the installation was done on a virtualized latest siduction iso image with the latest
changes from debian and siduction. I suggest you unpack the 4b file and have a closer look at
the 6.15 and kbuild patch. They could be different from yours.

[Teriarch]

@towo:
Got it!
Simply replace line 19 in "use-kbuild-flags.patch"
(of your current nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4_amd64.deb, 15:00h):
       NV_CONFTEST_CFLAGS = $(NV_CFLAGS_FROM_CONFTEST) $(EXTRA_CFLAGS) -fno-pie
by
       NV_CONFTEST_CFLAGS = $(NV_CFLAGS_FROM_CONFTEST) $(ccflags-y) -fno-pie
and it should work fine!

[Pip]

Sorry for the delay in replying, I'd lost my GUI. Very careless of me.
I did try the latest update and put in place the suggested line 19 change, but it still came back with the error I mentioned in my last post. It's now taken me a while to strip out nvidia and set up/etc/X11/xorg.conf.d/20-nouveau.conf but now I have my GUI back.
Thank you for all the attempts. If nouveau handles things for me, maybe I won't need to worry about this any more.

[Teriarch]

Hi Pip,

Don't give up so easily, and besides: The nouveau driver offers only restricted functionality.

Proceed a follows:

$ sudo apt-get update
$ sudo apt-get install nvidia-tesla-470-driver

The last command fails on nvidia-tesla-470-kernel-dkms (unless towo applied the patch in the meantime,
hence check for version nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4).

Get the patched version from

https://wormhole.app/08x65v#teV4IvWkirPB-eXLrijzPA
md5sum                                                 filename
6acbeb508691f14d481cc44798e748ff  nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4c_amd64.deb
(Don't worry about the 4c, it simply contains the latest patch and will reinstall the 4 version, btw: the link is
only valid for the next 24 hours)

Install it:
$ sudo dpkg -i nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4c_amd64.deb

If it fails, please attach the output of the last command in your next message.
Good luck!

PS.: I'd like to see an output like the following with all the gory details:
-----------------------------------------------------
After the error I did a
sudo dpkg -i nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4c_amd64.deb
with the following result:

(Lese Datenbank ... 312816 Dateien und Verzeichnisse sind derzeit installiert.)
Vorbereitung zum Entpacken von nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4c_amd64.deb ...
Deleting module nvidia-tesla-470/470.256.02 completely from the DKMS tree.
Entpacken von nvidia-tesla-470-kernel-dkms (470.256.02-7.siduction.4b) über (470.256.02-7.siduction.4) ...
nvidia-tesla-470-kernel-dkms (470.256.02-7.siduction.4b) wird eingerichtet ...
Loading new nvidia-tesla-470/470.256.02 DKMS files...
Building for 6.15.3-1-siduction-amd64

Building initial module nvidia-tesla-470/470.256.02 for 6.15.3-1-siduction-amd64
The kernel is built without module signing facility, modules won't be signed

Applying patch cc_version_check-gcc5.patch... done.
Applying patch bashisms.patch... done.
Applying patch 0001-some-power-management-features-were-not-yet-in-Linux.patch... done.
Applying patch 0033-refuse-to-load-legacy-module-if-IBT-is-enabled.patch... done.
Applying patch 0034-fix-typos.patch... done.
Applying patch 0037-import-pfn_valid-w-o-GPL-rcu_read_lock-unlock-from-v.patch... done.
Applying patch 0042-Log-an-error-message-when-nv_mem_client_init-fails-d.patch... done.
Applying patch 0043-backport-follow_pfn-changes-from-550.90.07.patch... done.
Applying patch 0045-let-the-virt_addr_valid-macro-use-nv_pfn_valid-on-pp.patch... done.
Applying patch 0046-virt_addr_valid-wants-a-void-argument.patch... done.
Applying patch kernel-6.12.patch... done.
Applying patch kernel-6.13.patch... done.
Applying patch nvidia-470xx-fix-linux-6.14.patch... done.
Applying patch nvidia-470xx-fix-linux-6.15.patch... done.
Applying patch fragile-ARCH.patch... done.
Applying patch conftest-verbose.patch... done.
Applying patch use-kbuild-compiler.patch... done.
Applying patch conftest-prefer-arch-headers.patch... done.
Applying patch use-kbuild-flags.patch... done.
Building module(s)........................ done.
Installing /lib/modules/6.15.3-1-siduction-amd64/updates/dkms/nvidia-tesla-470.ko
Installing /lib/modules/6.15.3-1-siduction-amd64/updates/dkms/nvidia-tesla-470-modeset.ko
Installing /lib/modules/6.15.3-1-siduction-amd64/updates/dkms/nvidia-tesla-470-drm.ko
Installing /lib/modules/6.15.3-1-siduction-amd64/updates/dkms/nvidia-tesla-470-uvm.ko
Installing /lib/modules/6.15.3-1-siduction-amd64/updates/dkms/nvidia-tesla-470-peermem.ko
Running depmod... done.
------------------------------------------------------------------------------------------

If the patches succeed, it has nothing to do with the deb file, but rather with a somehow broken build
system. We have to know, otherwise we are digging forever...

[Geier0815]

Hi Teriarch,

your package works for me. NVIDIA GK107GL [Quadro 410], kernel 6.15.3-1-siduction-amd64

[Teriarch]

@Geier0815:

good to hear, but be aware that any dist-upgrade reinstalls towos version 4 package
because of the siduction's site superior priority. I could circumvent the reinstall
by upgrading the version slightly and declaring an additional "Provides:..." entry in
the Debian control file, but I am counting on towo to apply the patch and upgrading
all tesla packages to version 5 in the meantime. So this is a waste of time.

Now that the drivers are loading doesn't imply they are running smoothly. So feel
free to test them to your heart's content and report back any problems.

[Teriarch]

@towo

Just out of curiosity and because I want this case closed. The siduction repo still
contains nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4_amd64.deb
from Jun, 22nd  23:44 which fails on kernel 6.15. However, we have a working
solution confirmed by positive feedback from Geier0815. The remaining changes
to the current version are rather minimal compared to the work already done. I
also understand that the interest in the tesla version is rather low and that you
have to keep mainstream changes up to date.

So, are you planning on a recent update to the current version?
And if not, what considerations are holding you back?

Thx in adv

[towo]

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

[Pip]

For me, it was

Code Select Expand

(Reading database ... 490916 files and directories currently installed.)
Preparing to unpack .../nvidia-tesla-470-kernel-dkms_470.256.02-7.siduction.4c_amd64.deb ...
Deleting module nvidia-tesla-470/470.256.02 completely from the DKMS tree.
Unpacking nvidia-tesla-470-kernel-dkms (470.256.02-7.siduction.4) over (470.256.02-7.siduction.4) ...
Setting up nvidia-tesla-470-kernel-dkms (470.256.02-7.siduction.4) ...
Loading new nvidia-tesla-470/470.256.02 DKMS files...
Building for 6.1.0-1-amd64, 6.1.0-2-amd64, 6.14.9-1-siduction-amd64, 6.14.10-1-siduction-amd64 and 6.15.3-1-siduction-amd64

Building initial module nvidia-tesla-470/470.256.02 for 6.1.0-1-amd64
Sign command: /usr/lib/linux-kbuild-6.1/scripts/sign-file
Signing key: /var/lib/dkms/mok.key
Public certificate (MOK): /var/lib/dkms/mok.pub

Applying patch cc_version_check-gcc5.patch... done.
Applying patch bashisms.patch... done.
Applying patch 0001-some-power-management-features-were-not-yet-in-Linux.patch... done.
Applying patch 0033-refuse-to-load-legacy-module-if-IBT-is-enabled.patch... done.
Applying patch 0034-fix-typos.patch... done.
Applying patch 0037-import-pfn_valid-w-o-GPL-rcu_read_lock-unlock-from-v.patch... done.
Applying patch 0042-Log-an-error-message-when-nv_mem_client_init-fails-d.patch... done.
Applying patch 0043-backport-follow_pfn-changes-from-550.90.07.patch... done.
Applying patch 0045-let-the-virt_addr_valid-macro-use-nv_pfn_valid-on-pp.patch... done.
Applying patch 0046-virt_addr_valid-wants-a-void-argument.patch... done.
Applying patch kernel-6.12.patch... done.
Applying patch kernel-6.13.patch... done.
Applying patch nvidia-470xx-fix-linux-6.14.patch... done.
Applying patch nvidia-470xx-fix-linux-6.15.patch... done.
Applying patch fragile-ARCH.patch... done.
Applying patch conftest-verbose.patch... done.
Applying patch use-kbuild-compiler.patch... done.
Applying patch use-kbuild-flags.patch... done.
Applying patch conftest-prefer-arch-headers.patch... done.
Building module(s).........(bad exit status: 2)
Failed command:
env NV_VERBOSE=1 make -j16 modules KERNEL_UNAME=6.1.0-1-amd64

with the error being the one I mentioned before

[Pip]

Thanks to both of you for trying. I'll see about slipping back into nouveau if my GUI fails to start again.