Do I assume correctly nvidia-powerd should be disabled for A100?

Graphics / Linux Linux
1

I assume I should. Periodically I have this errors on startup, having “nvidia-powerd” in them

[   99.555162] BUG: kernel NULL pointer dereference, address: 0000000000000490
[   99.558203] #PF: supervisor read access in kernel mode
[   99.559878] #PF: error_code(0x0000) - not-present page
[   99.561278] PGD ec3781f067 P4D ec3781f067 PUD ec3d2b3067 PMD 0 
[  100.177369] Oops: 0000 [#1] SMP NOPTI
[  100.177372] CPU: 2 PID: 2688 Comm: nvidia-powerd Tainted: P           OE     5.4.0-124-generic #140-Ubuntu
[  100.177375] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
[  100.179219] RIP: 0010:_nv021748rm+0x19/0x50 [nvidia]
[  100.179240] Code: 90 31 c0 c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 00 80 bf 80 0b 00 00 00 75 1f 80 bf 81 0b 00 00 00 74 2e 48 8b 87 b8 1c 00 00 <48> 8b 80 90 04 00 00 c3 0f 1f 80 00 00 00 00 8b 87 34 0a 00 00 48
[  100.179241] RSP: 0018:ffffa4dc453cf7e0 EFLAGS: 00010202
[  100.179242] RAX: 0000000000000000 RBX: ffff8f6de8ae0008 RCX: 0000000000000003
[  100.179242] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff8f6e281cc008
[  100.179243] RBP: ffff8f6dfcb65bd0 R08: ffffffffc4b4b190 R09: ffff8f6e64c07640
[  100.179243] R10: ffff8f6e281cc008 R11: 0000000000000001 R12: ffff8f6e281cc008
[  100.179244] R13: ffff8f6e51338030 R14: 0000000000000000 R15: ffff8f6e5133e410
[  100.179248] FS:  00007f380fea7740(0000) GS:ffff8f6e7d080000(0000) knlGS:0000000000000000
[  100.179248] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  100.179249] CR2: 0000000000000490 CR3: 000000ec9245e006 CR4: 0000000000360ee0

All the topics here say I should mask it. It would be nice to have it in official postinstall documentation — CUDA Installation Guide for Linux

2

OK, I think nvidia-powerd was a red herring

I still have issues with this machine even without nvidia-powerd inside the VM. I probably need to update host kernel…