Kernel crash after upgrade to 17.7

[mgerlach_itss]

Hello,

After updating to 17.7 the OPNSENSE crashes directly after the "Configuring firewall" appears in the console.
The message is "Fatal trap 12 page fault while in kernel mode". (See also attached pictures).
When I select old kernel in the boot menu however everything works fine.

I am using the 64bit version.
Hardware: Intel Atom D525.

How can I fix this?

Regards,
mg

[franco]

You can boot the old kernel from the boot menu, just have to switch it.

It shouldn't crash, it's the same kernel as 17.1 basically... I can only imagine the disk write was faulty.

What is your config? Are you using LAGG by any chance?

[mgerlach_itss]

It works with the old kernel.
I am not using LAGG. But I don't think it is a faulty disk write because the problem also occurs when I boot from a USB stick and import the configuration.

Anything I could do to track down the problem?

[franco]

Hm, do you have a PPPoE connection?

Did you use the 17.7-RC or did you come from 17.1.11?


Cheers,
Franco

[franco]

Or maybe IPsec?

It could also be a reconfigure race based on slightly adapted interface / firewall configuration code.

[odites999]

Hello:

I have exactly the same problem. 64 bit version. J1900 motherboard in my case. No need to reload old configuration. With live version I set up a pppoe connection (through vlan 6, in my case) and inmediately crashes.

[franco]

Thanks, I'll try to reproduce!

[franco]

Posting here too, please try if this patch helps.

https://github.com/opnsense/core/commit/065244ed

Apply with:

opnsense-patch 065244ed

Apply again to revert.


Cheers,
Franco

[odites999]

I assume the patch command needs a working internet connection to be downloaded and applied. That connection, in my case, is PPPOE based. But every time I try to establish it, it crashes. I'm wrong?

Regards,

[jwe]

Just in case it is the really same problem,
i found a possible solution in my thread: https://forum.opnsense.org/index.php?topic=5697.0

[franco]

Hi,

Yes, a connection is needed. We can also try a modified image to see if we can find the source of the problem. It  looks like a dormant problem in the operating system code that we now trigger with our modified interface configuration code.

See the thread that jwe posted, the details are in there.


Thanks,
Franco