Multiple CUDA/RTX/Vulkan application crashing with Xid (13,109) errors

Graphics / Linux Linux
418 
Fri Sep  6 21:54:52 2024       
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 560.35.03              Driver Version: 560.35.03      CUDA Version: 12.6     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce GTX 1650        Off |   00000000:01:00.0 Off |                  N/A |
| N/A   48C    P8              6W /   50W |       2MiB /   4096MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+
                                                                                         
+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI        PID   Type   Process name                              GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|  No running processes found                                                             |
+-----------------------------------------------------------------------------------------+

nvidia-bug-report.log.gz (1.0 MB)

Issue still happens, although this time the XID didnā€™t show up in dmesg (it has happened before once or twice- but the crash and all the same behavior still occurs).

419

I tried The Last of Us with 560.35.03 (open source version), and Xid is still there.

@shelter Are you absolutely certain 550.40.71 fixes it?

edit: so I tried 550.40.71, and indeed the Xid 109 that was blocking my progression is not happening anymore!

1 Like
420

@shelter might be something I did wrong, but 550.40.71 doesnā€™t seem to work with Ghost of Tsushima and Cyberpunk (both seem to crash during loading)

Sep 12 21:25:25 tumbleweed kernel: NVRM: GPU at PCI:0000:08:00: GPU-6cd0a0c1-3c79-696f-6f7f-fde0e426d057
Sep 12 21:25:25 tumbleweed kernel: NVRM: Xid (PCI:0000:08:00): 109, pid='<unknown>', name=<unknown>, Ch 0000019c, errorString CTX SWITCH TIMEOUT, Info 0x3c0bf

nvidia-bug-report.log.gz (1.3 MB)

421

GoT works fine here, never had any XID issues with that game, RTX 4070. I donā€™t have CP installed so I canā€™t test it.

I see this in your log, not sure if itā€™s related somehow.


Sep 12 20:58:15 tumbleweed steamwebhelper[5827]: x86_64-linux-gnu-capsule-capture-libs: warning: Dependencies of libnvidia-pkcs11.so.550.40.71 not found, ignoring: Missing dependencies: Could not find "libcrypto.so.1.1" in LD_LIBRARY_PATH "/home/aditya/.var/app/com.valvesoftware.Steam/.local/share/Steam/ubuntu12_32:/home/aditya/.var/app/com.valvesoftware.Steam/.local/share/Steam/ubuntu12_32/panorama:/app/lib/i386-linux-gnu/GL/default/lib:/app/lib/i386-linux-gnu/GL/nvidia-550-40-71/lib:/app/lib/ffmpeg/.:/app/lib32:/app/lib/i386-linux-gnu:/lib64:/app/lib:/usr/lib/x86_64-linux-gnu/GL/default/lib:/usr/lib/x86_64-linux-gnu/GL/nvidia-550-40-71/lib:/usr/lib/x86_64-linux-gnu/openh264/extra:/usr/lib/x86_64-linux-gnu", ld.so.cache, DT_RUNPATH or fallback /lib:/usr/lib
422

Greatā€¦ now I get XID 109 while playing AC Valhalla instead
Vulkan dev 550.40.71

423

Hi @shelter
Could you please confirm repro frequency and nvidia bug report from repro state.

424

Hi @hemaster
Could you please test both games with latest released driver and share bug report from repro state if issue persists.
Also, it would be good to know repro frequency and how long do I need to play to trigger issue.

425

Hmmā€¦ If you can you better check Vulkan dev 550.40.75 that was released today, it totally broke AC Valhalla, vertex explosions. Other games seemed fine thoā€™, so it couldā€™ve been something weird here.

image
image1920Ɨ1080 371 KB

But I went back to Vulkan dev 550.40.71 and the vertex explosions were gone.

Iā€™ll fiddle some with 555.40.71 and see if it crashes with XID 109.

Update: Went back to 555.40.75 again, explosions galore.
Also no crashes (this session) on 555.40.71 during the time I played, seems to be a bit random.

426

Hi @amrits
I tried with the latest available 550.107.02 on Tumbleweed, and the issue still seems to occur after 15 minutes or so on Cyberpunk. Didnā€™t yet try Helldivers or GoT. But I believe about 30 minutes should be enough to repro.
Iā€™ve attached the logs for Cyberpunk. Let me know if you need anything else.
nvidia-bug-report.log.gz (1.1 MB)

427

Is The Finals ever going to get fixed?

428

I have never experienced any issues with any of the recent drivers (between 535 maybe older - 560) with Cyberpunk. At least not with Xorg. I have occasionally tested with Wayland too, and it has also worked. Whatā€™s your GPU? Considering it takes 15 min before it crashes, I wonder if the issue could be related to vRAM or something. There are options in BIOS which can affect how the vRAM is utilized/accessed, maybe you can try playing with these (Turning them off/on) and see if it helps.