unable to run hashcat in vgpu

[kajweb]

There was a problem when starting up, and the console reported an error `cuCtxCreate(): invalid argument`

Below is the log file

Code:

$ cat ./example0.sh 
./hashcat -m 0 -t 32 -a 7 example0.hash ?a?a?a?a example.dict

$ sh ./example0.sh
hashcat (v6.2.6) starting

nvmlDeviceGetFanSpeed(): Not Supported

CUDA API (CUDA 11.7)
====================
* Device #1: Tesla T4, 7583/7680 MB, 40MCU

OpenCL API (OpenCL 3.0 CUDA 11.7.101) - Platform #1 [NVIDIA Corporation]
========================================================================
* Device #2: Tesla T4, skipped

Minimum password length supported by kernel: 0
Maximum password length supported by kernel: 256

Dictionary cache hit:
* Filename..: example.dict
* Passwords.: 128416
* Bytes.....: 1069601
* Keyspace..: 128416

Hashes: 6494 digests; 6494 unique digests, 1 unique salts
Bitmaps: 16 bits, 65536 entries, 0x0000ffff mask, 262144 bytes, 5/13 rotates

Optimizers applied:
* Zero-Byte
* Early-Skip
* Not-Salted
* Not-Iterated
* Single-Salt
* Raw-Hash

[color=#ffdc00]ATTENTION! Pure (unoptimized) backend kernels selected.[/color]
[color=#ffdc00]Pure kernels can crack longer passwords, but drastically reduce performance.[/color]
[color=#ffdc00]If you want to switch to optimized kernels, append -O to your commandline.[/color]
[color=#ffdc00]See the above message to find out about the exact limits.[/color]

Watchdog: Temperature abort trigger set to 90c

[color=#e82a1f][b]cuCtxCreate(): invalid argument[/b][/color]

Started: Fri Feb  7 21:28:49 2025
Stopped: Fri Feb  7 21:28:51 2025

Below is the information of nvidia-smi

Code:

$ nvidia-smi
Fri Feb  7 21:31:22 2025     
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 515.65.01    Driver Version: 515.65.01    CUDA Version: 11.7    |
|-------------------------------+----------------------+----------------------+
| 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  Tesla T4            On  | 00000000:00:0A.0 Off |                    0 |
| N/A  67C    P0    41W /  70W |      0MiB /  7680MiB |      0%      Default |
|                              |                      |                  N/A |
+-------------------------------+----------------------+----------------------+
                                                                             
+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU  GI  CI        PID  Type  Process name                  GPU Memory |
|        ID  ID                                                  Usage      |
|=============================================================================|
|  No running processes found                                                |
+-----------------------------------------------------------------------------+

I am curious why the first time I execute `hc_cuCtxCreate` to check device_available_mem, no exception is triggered
https://github.com/hashcat/hashcat/blob/...nd.c#L5365

But the second time when create context for each device, the program outputs invalid argument
https://github.com/hashcat/hashcat/blob/...nd.c#L9638

Unfortunately, there are so many statements in between that I can't debug whether the exception is caused by cuda_context or cuda_device.

So can anyone help me figure out what's going wrong here?