Re: [BUG] Kernel Panic in iptfs_reassem_cont when handling large packets

[Fernando Fernandez Mancera <fmancera@suse.de>]

On 3/1/26 10:49 AM, Hao Long wrote:
> Hello,
> 
> Recently I set up a strongSwan tunnel in AGGFRAG mode[1] in order to see
> how it fragments large packets.
> 
> Later I found out the receiver node will kernel panic when handling
> large packets, I tested in different distro and both panic.
> 
> Tested environment:
> - Arch Linux 6.18.13-arch1-1 strongswan-6.0.4-2
> - Arch Linux 7.0.0-rc1-1-mainline strongswan-6.0.4-2
> - NixOS 6.18.13 strongswan-6.0.4
> 
> Step to reproduce:
> 1. install strongSwan and create tunnel interface in vm1, see the
> attachment init_env.sh
> 2. do step1 in vm2, but remember to switch local_addrs and remote_addrs,
> also the ip assignment
> 3. run `ping -s 3333 10.0.1.2` in vm1, 10.0.1.2 is the ip from vm2
> 4. kernel panic in vm2
> 

I have reproduced this. I will send a patch a shortly so you can test it 
in your environment too.

Thanks,
Fernando.

> I'm not familiar in C programming and kernel developing, so sorry I can't
> provide a useful root case analyze and a fix.
> 
> Regards,
> Hao Long
> 
> [1] https://docs.strongswan.org/docs/latest/features/iptfs.html
>