Kernel Panic Please Help!!!

[ErkDog]

About once a week my proxmox hypervisor will lock up.

I finally had a KVM on it when it did and got the error messages:

https://photos.app.goo.gl/g1hzgeCo1gSMnCyc7

kernel BUG at drivers/immu/iova.c:814!
invalid opcode: 0000 [#1] SMB NOPTI
CPU: 17 PID: 0 Comm: swapper/17 Tainted: P W O 5.3.13-3-pve #1


System report attached.

 

[t.lamprecht]

Hi, the 5.3.13 kernel is a bit older, could you please first try to upgrade to the more recent 5.3.18-3 kernel, or if that doesn't help trying the upcoming 5.4 Kernel would be also a good idea, see: https://forum.proxmox.com/threads/linux-kernel-5-4-for-proxmox-ve.66854/#post-300133

While it's a preview, we heard mostly good things about it and it will move as new default kernel in PVE pretty soon.

IOVA is the IO Virtual address, and in your case it seems that your realtek card tries to "free" such an address which never belonged to it, so the IOVA code cannot do much but shouting "bug" and probably tearing down network with that.

Can you tell us a bit more about the system, from the screenshots I see that it's using and AMD Threadripper CPU and the realtek network interface is probably the onboard from the mainboard?

 

[ErkDog]

Thank you for helping to decipher that error message. It's worth noting that the r8125 does not appear to have native support in the PVE Kernel.

I had to download the driver and add it to the kernel directly from Realtek's website.

I'm currently using this version: r8125-9.003.02

Which I got from here: https://www.realtek.com/en/componen...0-1000m-gigabit-ethernet-pci-express-software

If the @proxmox team could perhaps add a driver directly to the Kernel it may help absolve these issues as well.

Yes, It's a 3960X on an ASRock TRX 40 Creator motherboard.

The aquantia driver for the 10G is built into the kernel, but the r8125 I had to source as noted above.

Thanks,
Matt

 

[ErkDog]

My (cheap) Realtek RTL8125 based 2.5 Gbps network card now finally works! quoted from other thread so looks like 5.4 will actually solve that problem indeed!!!!

Question though - when will 5.4 be rolled into 6.x naturally? or do you have to opt in with the commands listed either way?

Thanks,
Matt

 

[t.lamprecht]

Question though - when will 5.4 be rolled into 6.x naturally? or do you have to opt in with the commands listed either way?

Soon, but opting in earlier works too

 

[Sarisa149]

My (cheap) Realtek RTL8125 based 2.5 Gbps network card now finally works! quoted from other thread so looks like 5.4 will actually solve that problem indeed!!!!

Question though - when will 5.4 be rolled into 6.x naturally? or do you have to opt in with the commands listed either way?

Thanks,
Matt

Hi,
Strange, i'm getting the error regarding the network controller and cant install.
I'm install PVE6.2 and i think the kernel is 5.4 !!
Any help how to solve it!?

Thanks.

 

[Bupple]

Hi,
Strange, i'm getting the error regarding the network controller and cant install.
I'm install PVE6.2 and i think the kernel is 5.4 !!
Any help how to solve it!?

Thanks.

I am also having the same issue even thought it seems that kernel 5.4 should have support for this nic