NVIDIA driver 510.39.01 crash with Ubuntu 21.10 and NVIDIA GeForce RTX 3060 Ti

Graphics / Linux Linux
,
1

NVIDIA driver 510.39.01 crash with Ubuntu 21.10 and NVIDIA GeForce RTX 3060 Ti radomly crashes with this in the syslog:

Jan 20 21:58:43 theBeast kernel: [ 4722.854311] pcieport 0000:00:01.0: AER: Uncorrected (Non-Fatal) error received: 0000:00:01.0
Jan 20 21:58:43 theBeast kernel: [ 4722.854318] pcieport 0000:00:01.0: PCIe Bus Error: severity=Uncorrected (Non-Fatal), type=Transaction Layer, (Requester ID)
Jan 20 21:58:43 theBeast kernel: [ 4722.854321] pcieport 0000:00:01.0:   device [8086:460d] error status/mask=00100000/00010000
Jan 20 21:58:43 theBeast kernel: [ 4722.854322] pcieport 0000:00:01.0:    [20] UnsupReq               (First)
Jan 20 21:58:43 theBeast kernel: [ 4722.854324] pcieport 0000:00:01.0: AER:   TLP Header: 34000000 01000010 00000000 00000000
Jan 20 21:58:43 theBeast kernel: [ 4722.854327] nvidia 0000:01:00.0: AER: can't recover (no error_detected callback)
Jan 20 21:58:43 theBeast kernel: [ 4722.854329] snd_hda_intel 0000:01:00.1: AER: can't recover (no error_detected callback)
Jan 20 21:58:43 theBeast kernel: [ 4722.854332] NVRM: GPU at PCI:0000:01:00: GPU-b69c7f51-dae2-883e-5a7b-e3629342967a
Jan 20 21:58:43 theBeast kernel: [ 4722.854336] NVRM: Xid (PCI:0000:01:00): 79, pid=0, GPU has fallen off the bus.
Jan 20 21:58:43 theBeast kernel: [ 4722.854338] NVRM: GPU 0000:01:00.0: GPU has fallen off the bus.
Jan 20 21:58:43 theBeast kernel: [ 4722.854348] pcieport 0000:00:01.0: AER: device recovery failed
Jan 20 21:58:43 theBeast kernel: [ 4722.854483] NVRM: GPU 0000:01:00.0: GPU serial number is <FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF>
<FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF><FF>
<FF><FF><FF><FF><FF><FF>.
Jan 20 21:58:43 theBeast kernel: [ 4722.854492] NVRM: A GPU crash dump has been created. If possible, please run
Jan 20 21:58:43 theBeast kernel: [ 4722.854492] NVRM: nvidia-bug-report.sh as root to collect this data before
Jan 20 21:58:43 theBeast kernel: [ 4722.854492] NVRM: the NVIDIA kernel module is unloaded.

The only way to reboot is then the reset button.
Processing: nvidia-bug-report.log.gz…

2

Seems the log is stuck in the virus scanner. Please try reattaching.
From the log snippets, there’s a bus error, already tried reseating the crad in its slot, checked for a bios update, tried lowering bus speed?

3

Yes I have been trying to upload the nvidia-bug-report.log.gz several times but with no success so far? Shall I untar it?

4

Yes, please try unzipping and uploading the plain text file.

5

nvidia-bug-report.log (2.6 MB)
Here it is. Thanks

6

I guess you should rather contact Dell since this looks like a hardware issue. Is the device still under warranty?

7

Hello,
Yes thanks for your reply. The hardware is still under warranty yes.
I have upgraded the bios hoping this would help.
What makes you think it is a hardware issue?

8

The pcie bus errors and that the gpu is always falling off the bus with a XID79. Can have many causes but since you didn’t put it together yourself but bought it as a complete system, there’s no sense in finding out yourself. Just let Dell handle that.