nvidia-580xx-dkms

gcb · 2026-11-01 07:02:42

trying to follow https://wiki.archlinux.org/title/NVIDIA

my card is a "pascal" that uses nvidia-580xx-dkms. It is one of the awfull optimus/bumblebee setups though.

But I am 100% lost, as those pages mention nothing besides installing the driver, and installing the driver doesn't seem to be enough.

the AUR git builds 3 packages. utils, dkms, opencl. I install the three of them fine. They place no modules anywhere i can see, they pretend to re-run mkinitcpio as if they added something to the system, but as far as i can see, there's nothing.

i see some modprobe configs but they are in /usr/lib/modprobe.d, not even sure those are picked up normally. For example, one things those files contains, they try to blacklist nouveau, but after i reboot, i still have nouveau loaded.

I can't find any modules anywhere

there's a binary in the utils package called nvidia-modprobe, but it just return error code 1 and does/output nothing.

$ lspci -k -d ::03xx
00:02.0 VGA compatible controller: Intel Corporation Kaby Lake-H GT2 [HD Graphics 630] (rev 04)
DeviceName: Onboard IGD
Subsystem: Dell Device 07be
Kernel driver in use: i915
Kernel modules: i915
01:00.0 3D controller: NVIDIA Corporation GP107M [GeForce GTX 1050 Mobile] (rev a1)
Subsystem: Dell Device 07be
Kernel driver in use: nouveau
Kernel modules: nouveau

$ lsmod | grep nvidia

$ lsmod | grep nouveau
nouveau 3780608 1
gpu_sched 77824 1 nouveau
drm_gpuvm 57344 1 nouveau
drm_exec 12288 2 drm_gpuvm,nouveau
drm_ttm_helper 20480 1 nouveau
mxm_wmi 12288 1 nouveau
i2c_algo_bit 24576 2 i915,nouveau
ttm 155648 3 drm_ttm_helper,i915,nouveau
video 81920 4 dell_wmi,dell_laptop,i915,nouveau
drm_display_helper 303104 2 i915,nouveau
wmi 40960 9 video,intel_wmi_thunderbolt,dell_wmi,wmi_bmof,dell_smm_hwmon,dell_smbios,dell_wmi_descriptor,mxm_wmi,nouveau

Nov 01 06:28:16 kernel: pci 0000:01:00.0: optimus capabilities: enabled, status dynamic power,
Nov 01 06:28:16 kernel: VGA switcheroo: detected Optimus DSM method \_SB_.PCI0.PEG0.PEGP handle
Nov 01 06:28:16 kernel: nouveau: detected PR support, will not use DSM
Nov 01 06:28:16 kernel: nouveau 0000:01:00.0: enabling device (0006 -> 0007)
Nov 01 06:28:16 kernel: nouveau 0000:01:00.0: NVIDIA GP107 (137000a1)
Nov 01 06:28:16 kernel: nouveau 0000:01:00.0: bios: version 86.07.3e.00.1c
Nov 01 06:28:16 kernel: nouveau 0000:01:00.0: pmu: firmware unavailable
Nov 01 06:28:16 kernel: nouveau 0000:01:00.0: fb: 4096 MiB GDDR5
Nov 01 06:28:16 kernel: vga_switcheroo: enabled

loqs · 2026-11-01 07:10:34

What is the output of `dkms status`. Please use code tags.

seth · 2026-11-01 09:17:24

On a limb, read https://wiki.archlinux.org/title/Dynami … stallation three times as well as https://wiki.archlinux.org/title/NVIDIA … rly%20boot.

gcb · Yesterday 04:03:49

$ dkms status
bbswitch/0.8, 7.2.11-arch1-1, x86_64: installed
nvidia/580.196.07, 7.2.11-arch1-1, x86_64: installed
v4l2loopback/0.15.4, 7.2.11-arch1-1, x86_64: installed

> read https://wiki ...

I did try my best to follow all the "Install" sections I could find from those pages.

Edit:
If I blacklist the nouveau module, card is not detected at all

$ lspci -k -d ::03xx
00:02.0 VGA compatible controller: Intel Corporation Kaby Lake-H GT2 [HD Graphics 630] (rev 04)
        DeviceName:  Onboard IGD
        Subsystem: Dell Device 07be
        Kernel driver in use: i915
        Kernel modules: i915

seth · Yesterday 07:16:56

The modules will have no impact on the presence of the pci device.
Please post your complete system journal for the boot:

sudo journalctl -b | nc termbin.com 9999

WorMzy · Yesterday 09:16:48

Mod note: moving to AUR Issues.

gcb · Yesterday 19:52:11

boot with fresh arch install, no AUR or proprietary drivers whatsoever
https://termbin.com/lg5l3

 # journalctl -b -11 | grep -i nvidia
 nouveau 0000:01:00.0: NVIDIA GP107 (137000a1)

Card shows up on lspci just fine.

boot with AUR drivers built, and no bumblebee/bswitch installed
AUR driver blocks nouveau.
https://termbin.com/mfs5

# journalctl -b -5 | grep -i nvidia
 nvidia: module license 'NVIDIA' taints kernel.
 nvidia: module license taints kernel.
 nvidia-nvlink: Nvlink Core is being initialized, major device number 510
 NVRM: No NVIDIA GPU found.

card doesn't show up on lspci

lastly, with AUR driver and bumblebeed+bbswitch
https://termbin.com/lkjt
...i think this made things worse as bumblebee disables the AUR driver at boot too?

# journalctl -b 0  | grep -i nvid
 Module 'nvidia' is deny-listed (by kmod)
 Module 'nvidia_modeset' is deny-listed (by kmod)
 Module 'nvidia_uvm' is deny-listed (by kmod)
 Module 'nvidia_drm' is deny-listed (by kmod)
 Configuration file /etc/systemd/system/nvidia-enable.service is marked world-inaccessible. This has no effect as configuration data is accessible via APIs without restrictions. Proceeding anyway.

The lot of pcie errors everywhere, i'm thinking it have to do with the thunderbolt dock.

# mkinitcpio.conf
MODULES=(nvidia nvidia_modeset bbswitch nvidia_uvm nvidia_drm)
BINARIES=()
FILES=()
HOOKS=(systemd keyboard autodetect microcode modconf kms sd-vconsole block sd-encrypt lvm2 filesystems resume fsck acpi_override)
COMPRESSION="zstd"
COMPRESSION_OPTIONS=(-v --long -9)
MODULES_DECOMPRESS="yes"

BIOS have no option to disable optimus or either of the GPUs

GPU is soldered on the same mainboard as CPU, not external/thunderbolt in any way.

Last edited by gcb (Yesterday 20:00:23)

seth · Yesterday 20:35:54

card doesn't show up on lspci

Yes, does.

pci 0000:01:00.0: [10de:1c8d] type 00 class 0x030200 PCIe Endpoint

10de  NVIDIA Corporation
	1c8d  GP107M [GeForce GTX 1050 Mobile]

But

pcieport 0000:00:1d.6: AER: Correctable error message received from 0000:06:00.0

there's tremendous noise from that bus, not sure whether the GPU is behind that

Please post the output of

lspci -tvnn

when the GPU still shows up

nvidia then complains

 nvidia-nvlink: Nvlink Core is being initialized, major device number 510
 NVRM: No NVIDIA GPU found.

but there's no issue w/ the device but

 bbswitch: Found discrete VGA device 0000:01:00.0: \_SB_.PCI0.PEG0.PEGP
 pci 0000:01:00.0: enabling device (0006 -> 0007)
 bbswitch: Succesfully loaded. Discrete card 0000:01:00.0 is on

recognizes it but then

 BUG: kernel NULL pointer dereference, address: 00000000000002a4
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 0 P4D 0 
 Oops: Oops: 0000 [#1] SMP NOPTI
 CPU: 0 UID: 0 PID: 757 Comm: sh Tainted: G           OE       7.2.11-arch1-1 #1 PREEMPT(full)  1e38049adb503fbd3a6552a982bd2d91fe8786ca
 Tainted: [O]=OOT_MODULE, [E]=UNSIGNED_MODULE
 Hardware name: Dell Inc. XPS 15 9560/0JHP5H, BIOS 1.24.0 08/10/2021
 RIP: 0010:__pm_runtime_resume+0x1a/0x90
 Code: 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 f3 0f 1e fa 0f 1f 44 00 00 41 54 55 53 48 89 fb 48 83 ec 08 40 f6 c6 01 75 09 <f6> 87 d1 01 00 00 04 74 4b 40 f6 c6 04 74 07 f0 ff 83 c8 01 00 00
 RSP: 0018:ffffd206039ffd78 EFLAGS: 00010246
 RAX: 0000000000000003 RBX: 00000000000000d3 RCX: 0000000000000000
 RDX: 0000000000000003 RSI: 0000000000000004 RDI: 00000000000000d3
 RBP: 000055d0b5094970 R08: ffffd206039ffdb8 R09: 0000000000000001
 R10: 0000000000000000 R11: 0000000000000000 R12: ffffd206039ffe88
 R13: ffff8cee87113880 R14: 0000000000000001 R15: 0000000000000000
 FS:  00007f8d350c1b80(0000) GS:ffff8cf05a32a000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 00000000000002a4 CR3: 000000014085a002 CR4: 00000000003726f0
 Call Trace:
  <TASK>
  bbswitch_proc_write+0x9a/0x170 [bbswitch 5ab81ceb5af416bab0b01347850acf6ce3f10a76]
  ? rw_verify_area+0x61/0x1a0
  proc_reg_write+0x5a/0xb0
  vfs_write+0xe6/0x560
  ? do_dup2+0xb0/0x150
  ? do_sys_openat2+0x9c/0xe0

i'm thinking it have to do with the thunderbolt dock.

Is it an eGPU?
Remove the dock, see whether the PCI errors go away.

Sidebar: feel free to obfuscate your hostname or usernames or whatever all you want, but please preserve the file structure (columns) for processing and most importantly don't just sed away timestamps.

Last edited by seth (Yesterday 20:36:39)

gcb · Today 05:12:56

nope, not an egpu. it is on the same main board as the cpu. But maybe there's some innerworkings I don't understand. The HDMI port on the laptop only outputs low refresh rate, while the hdmi port via the dock outputs full refresh rate for higher resolutions.

not using the thunderbolt dock reduce those warnings (as much as I could figure out, they are incompatible powermanagement messages from the bios acpi and linux). For those i'm playing with acpi_osi values, but nothing seems to work very well.

But, I think you found the smoking gun. bbswitch finding the board and disabling it at boot. I saw some comments on the driver (can't find it now) that it might only work during boot.

I found that bbswitch have a modprobe.d file that disables the board at start, i removed that and now bbswitch doesn't disable it right away:

Nov 03 04:17:53 archlinux kernel: bbswitch: version 0.8
Nov 03 04:17:53 archlinux kernel: bbswitch: Found integrated VGA device 0000:00:02.0: \_SB_.PCI0.GFX0
Nov 03 04:17:53 archlinux kernel: bbswitch: Found discrete VGA device 0000:01:00.0: \_SB_.PCI0.PEG0.PEGP
Nov 03 04:17:53 archlinux kernel: bbswitch: detected an Optimus _DSM function
Nov 03 04:17:53 archlinux kernel: pci 0000:01:00.0: enabling device (0006 -> 0007)
Nov 03 04:17:53 archlinux kernel: bbswitch: Succesfully loaded. Discrete card 0000:01:00.0 is on

Now it stays on... but doesn't. Still hidden from lscpi on that boot. But now there's nothing saying it was turned off

anything that touches bbswitch or that gpu, crashes. I will try the nouveau back

Oops: Oops: 0000 [#2] SMP NOPTI
RIP: 0010:pci_bus_read_config_dword+0x2e/0x90
bbswitch_proc_write+0x110/0x170 [bbswitch 5ab81ceb5af416bab0b01347850acf6ce3f10a76]

i tried fully disabling pcie_port_pm=off (read somewhere it might fight for pm control of the card with bbswitch) but that did nothing to improve.

seth · Today 08:05:10

not using the thunderbolt dock reduce those warnings

"reduced" or "removed"?

bbswitch finding the board and disabling it at boot

 nvidia-nvlink: Nvlink Core is being initialized, major device number 510
 NVRM: No NVIDIA GPU found.

happens w/o bbswitch - they're both symptoms of the same problem.

seth wrote:

Please post the output of
lspci -tvnn
when the GPU still shows up

Did you by any chance disable the GPU in the UEFI and it just randomly shows up n the beginning still?
Can you disable the GPU there? (Turn it off, reboot, turn it on …)

Do you have another external power supply than the dock?

Arch Linux

#1 2026-11-01 07:02:42

nvidia-580xx-dkms

#2 2026-11-01 07:10:34

Re: nvidia-580xx-dkms

#3 2026-11-01 09:17:24

Re: nvidia-580xx-dkms

#4 Yesterday 04:03:49

Re: nvidia-580xx-dkms

#5 Yesterday 07:16:56

Re: nvidia-580xx-dkms

#6 Yesterday 09:16:48

Re: nvidia-580xx-dkms

#7 Yesterday 19:52:11

Re: nvidia-580xx-dkms

#8 Yesterday 20:35:54

Re: nvidia-580xx-dkms

#9 Today 05:12:56

Re: nvidia-580xx-dkms

#10 Today 08:05:10

Re: nvidia-580xx-dkms

Board footer