Header files for Linux 5.19.0-1-amd64
Common header files for Linux 5.19.0-1

They may not be backports, I have them installed.

Tried this yesterday/today on Laptop with FX 770m graphics,

but it seems Ceres not offering the desired solution (yet).

My Question: it has been alluded that it could be solved by using a backported kernel from Beowulf, but how to achieve this?

Installing kernel 4.19 + headers for it and just booting from it?
(EDIT: it seems there are no kernel headers for 4.19 on Beowulf Backports available).

Thank you.

Thank you.  8^)

At the moment my box runs beowulf on a backported kernel ...

~$ uname -a
Linux devuan 5.10.0-0.bpo.12-amd64 #1 SMP Debian 5.10.103-1~bpo10+1 (2022-03-08) x86_64 GNU/Linux
~$ 

... and xorg-xserver-1.20:

~$ cat /var/log/Xorg.0.log
[    33.934] 
X.Org X Server 1.20.4
X Protocol Version 11, Revision 0
[    33.935] Build Operating System: Linux 5.10.0-10-amd64 x86_64 Debian
[    33.935] Current Operating System: Linux devuan 5.10.0-0.bpo.12-amd64 #1 SMP Debian 5.10.103-1~bpo10+1 (2022-03-08) x86_64
--- snip ---

Maybe the issue is just with 5.4 kernels.
ie: 5.4 included.

I don't think I'm up to using anything other than the official Devuan kernels.

Yes.

The XFCE compositor is a PITA, screws up many things.
Has been a problem for quite a while but the devs/maintainers have taken no steps to fix it.
The old let's wait for the new release song and dance while putting the blame on something else.

So I'm slowly but steadily heading towards Openbox as the XFCE 4.16 and it's path to 5.x does not look too bright.

See:
https://forum.xfce.org/viewtopic.php?pid=56141#p56141
https://forum.xfce.org/viewtopic.php?pid=56143#p56143
https://forum.xfce.org/viewtopic.php?pid=56144#p56144

Thank you very much for taking the time to write this up.
I'll just stay with the backported beowulf kernel for the time being.

*************
Edit:

Given that the proposed solution involves downgrading the kernel (ie: to a pre 5.4 kernel) it would seem that it is not a suitable one.
At least, not a suitable answer to the OP's question: install nvidia-legacy-340xx drivers on Devuan chimaera.

This thread has had 2K views, so it's evidently of some interest to many.
ie: not a unique/one off setting.

From where I am seeing it, it should not be marked as solved.
But it's the admins say that counts.

*************

Maybe the Ubuntu people will get the chaps at Debian to do something about this Nvidia problem.
Maybe not, we'll see.

But I'm not giving up my video cards again.

Best,

A.

The link below is to a post by a Ubuntu specialist, enigma9o7 on how to compile an
nvidia-legacy-304xx driver straight from Nvidia for Ubuntu 18.04.
Before you examine it in detail, please review the following preamble:

-The nvidia-legacy-304xx driver has two major restrictions:
   i)  It will not work on Linux kernels newer than 4.13 without patches.
  (enigma9o7 has provided a patch for kernels 4.14 and 4.15).
  ii)  A lesser know fact is that it will not work for xorg-xserver newer than
   1.19.n
-I have successfully installed the driver on Debian 11, Debian 10, antiX-19 and
MX-19 by downgrading xorg-xserver to 1.19.2 and downgrading the kernel,
except for antiX-19, to 4.9.0-n, and then following enigma9o7's procedure
to the letter.
***********************************************************************
Turning then to the nvidia-legacy-340xx driver:

The Nvidia website description for version 340.107 states that the driver does
work for xorg-xserver-1.20.    Chimaera and Beowulf are both the same in this
regard, so no change should be required.

It is widely documented that the driver does not work on kernels newer than 5.4.
Therefore, a kernel downgrade is required.

I recommend the antiX-21 kernel, 4.19.0-222-antix.1.
It is a relatively fast procedure to download and install the antiX-21 base .iso,
and then run sudo apt-get download linux-headers-4.19.0-222.antix.1
               and sudo ap-get download linux-image-4.19.0-222.antix.1
to obtain the 2 packages necessary for the downgrade.

Once the downgraded kernel is installed,   lock the two packages from synaptic,
and sudo apt-mark hold them from the command line.
***********************************************************************
BEFORE STARTING enigma9o7' s procedure:

-install lshw and inxi, so that the installation results can be checked,
sudo lshw -C video, and sudo inxi  -Gxx
-check which display manager is being used, as the procedure contains a
stop service command for the display manager.

WITHIN THE PROCEDURE
NB -   before invoking the Nvidia installer,  for the XFCE desktop,
           disable the compositor.   (If this isn't done, a blank screen on
           final reboot will result).

-regardless of what error or warning messages might say,  say yes/OK/proceed
to all prompts.
-at the very end, from synaptic, reinstall libvdpau1,  to replace a file that
has likely been deleted during the procedure.

-at the beginning of the procedure, there are two links for downloading the
Nvidia packages.  The digital values within the links shown are 123708
for the 64-bit version of nivida-legacy-304xx, and 123709 for the 32-bit
version.
-FOR nvidia-legacy-340xx, the values are 135161 for 64-bit and 135160
for 32-bit.

Finally, the link:

https://ubuntuforums.org/showthread.php?t=2396263

Good luck !!
Len E.

@deepforest:

Yes, same here.

Thanks to HevyDevy* I managed to set up the legacy 340xx drivers on beowulf and posted the terminal printout of the process.
It works with beowulf but I don't know if it will work with chimaera.

See: https://dev1galaxy.org/viewtopic.php?pid=24694#p24694
Did it work for you?
Did you have to do anything differently?

* it seems that there is a HevyDevy and a hevidevi ...
One and the same maybe?

@Len E.

Please do share your solution with the forum.
Not a burden, many of us are in the same leaky Nvidia boat so all data is useful

I for one am not expecting much from nouveau and won't touch wayland.
Nvidia has recently released some source code but they did not include the source files to the 340XX blob.

Those cards are not manufactured any more so I can't see the reason for not releasing the source.
But Nvidia is Nvidia ... 8^|

I refuse to ditch a pair of perfectly working and quite suitable FX-580 Quadro cards for the lack of decent Linux drivers.
I already went through that a few years ago with a pair of Matrox G450s so if needed I'll just stay on beowulf.

A step-by-step terminal printout of the installation process like the one I posted would be of great help.

Thanks in advance.

Best,

A.

From the latest entries on this post, it's not
clear whether you've found a satisfactory solution or not.

I have an idea on how to do the installation, but I won't
burden you with the details unless you're still looking for a fix.

Please advise.
Len E.

Then use dkms to build the module

So. My solution.

1. Install or upgrade to Devuan sid
2. go to recovery mode and install just

apt install linux-headers-$(uname -r) build-essential
apt install nvidia-legacy-340xx-kernel-dkms

3. after removed nvidia.conf and "update-initramfs -u"  from

etc/modules-load.d
etc/modprobe.d

All works like a charm! Enjoy of Freedom! Glory and Long live to Devuan team!

SO its no solved yet

Now i have this errors

Fri Mar 18 21:55:14 2022: modprobe: FATAL: Module nvidia not found in directory /lib/modules/5.10.0-12-amd64
Sun Mar 20 02:16:29 2022: modprobe: FATAL: Module nvidia not found in directory /lib/modules/5.16.0-5-amd64
Mon May  2 20:15:20 2022: modprobe: FATAL: Module nvidia not found in directory /lib/modules/5.17.0-1-amd64

What this files do?

/etc/modprobe.d/nvidia.conf                                                                                                                                         716/716               100%
install nvidia modprobe -i nvidia-legacy-340xx $CMDLINE_OPTS

install nvidia-uvm modprobe nvidia ; modprobe -i nvidia-legacy-340xx-uvm $CMDLINE_OPTS

remove nvidia modprobe -r -i nvidia-uvm nvidia

alias char-major-195* nvidia

# These aliases are defined in *all* nvidia modules.
# Duplicating them here sets higher precedence and ensures the selected
# module gets loaded instead of a random first match if more than one
# version is installed. See #798207.
alias   pci:v000010DEd00000E00sv*sd*bc04sc80i00*        nvidia
alias   pci:v000010DEd00000AA3sv*sd*bc0Bsc40i00*        nvidia
alias   pci:v000010DEd*sv*sd*bc03sc02i00*               nvidia
alias   pci:v000010DEd*sv*sd*bc03sc00i00*               nvidia

Hello:

deepforest wrote:

thanks for help!

You're welcome.

deepforest wrote:

... is it worth trying to install nvidia-legacy-340xx ...

The 340XX drivers are propiertary/closed source and Debian cannot supply a 340XX package because Nvidia will not update them for newer kernels.

True to their A1-asshole mentality, they will not release the code for the hardware to be used in newer kernels.

What about old amd cards? How do they work then? Like for example, hd 6850. I still use it with 5.10 kernel. I don't think amd still release something for them. Just curious. I also got some x1950, a prehistoric card.

1. Install or upgrade to Devuan sid
2. go to recovery mode and install just

apt install linux-headers-$(uname -r) build-essential
apt install nvidia-legacy-340xx-kernel-dkms

3. after removed nvidia.conf and "update-initramfs -u"  from

etc/modules-load.d
etc/modprobe.d

All works like a charm! Enjoy of Freedom! Glory and Long live to Devuan team!

You only need nvidia-persistenced if you are running CUDA.

And why i have error during full install driver

Errors were encountered while processing:
 nvidia-persistenced
E: Sub-process /usr/bin/dpkg returned an error code (1)

And have no error if i install nvidia-persistenced standalone?

freeartist@devuan:~$ dpkg -s nvidia-persistenced 
Package: nvidia-persistenced
Status: install ok installed
Priority: optional
Section: contrib/utils
Installed-Size: 91
Maintainer: Debian NVIDIA Maintainers <pkg-nvidia-devel@lists.alioth.debian.org>
Architecture: amd64
Multi-Arch: foreign
Version: 470.103.01-2
Depends: libnvidia-cfg1 | libnvidia-tesla-510-cfg1 | libnvidia-tesla-470-cfg1 | libnvidia-tesla-450-cfg1 | libnvidia-tesla-418-cfg1 | libnvidia-legacy-390xx-cfg1 | libnvidia-legacy-340xx-cfg1 | libnvidia-cfg.so.1 | libnvidia-cfg1-any, adduser, libc6 (>= 2.33), libtirpc3 (>= 1.0.2)
Pre-Depends: init-system-helpers (>= 1.54~)
Conffiles:
 /etc/init.d/nvidia-persistenced f24407fda40433544aaf2a2b930fcdbf
Description: daemon to maintain persistent software state in the NVIDIA driver
 When persistence mode is enabled, the daemon prevents the driver from
 releasing device state when the device is not in use.
 This can improve the startup time of new clients in this scenario.
Homepage: https://github.com/NVIDIA/nvidia-persistenced
freeartist@devuan:~$

etc/modules-load.d
etc/modprobe.d-remove

error

[    1.584181] udevd[117]: Error running install command 'modprobe -i nvidia-legacy-340xx ' for module nvidia: retcode 1

gone

Errors were encountered while processing:
nvidia-persistenced
E: Sub-process /usr/bin/dpkg returned an error code (1)

I will assume that you did not install the nvidia drivers with the "--no-install-recommends" part.  If not, then it will install the nvidia-persistenced package.
As for the other error message, I don't know about that.

Yes with "--no-install-recommends" i have no error due installation, thanks!

And what this mean? After install nvidia driver i have this files

/etc/nvidia/legacy-340xx/nvidia-modprobe.conf                                                                                                                       665/665               100%

install nvidia modprobe -i nvidia-legacy-340xx $CMDLINE_OPTS

install nvidia-uvm modprobe nvidia ; modprobe -i nvidia-legacy-340xx-uvm $CMDLINE_OPTS

remove nvidia modprobe -r -i nvidia-uvm nvidia

alias char-major-195* nvidia

# These aliases are defined in *all* nvidia modules.
# Duplicating them here sets higher precedence and ensures the selected
# module gets loaded instead of a random first match if more than one
# version is installed. See #798207.
alias   pci:v000010DEd00000E00sv*sd*bc04sc80i00*        nvidia
alias   pci:v000010DEd00000AA3sv*sd*bc0Bsc40i00*        nvidia
alias   pci:v000010DEd*sv*sd*bc03sc02i00*               nvidia
alias   pci:v000010DEd*sv*sd*bc03sc00i00*               nvidia

and this error during booting

[    1.584181] udevd[117]: Error running install command 'modprobe -i nvidia-legacy-340xx ' for module nvidia: retcode 1