Hi,
On Fri, Feb 13, 2026 at 05:12:07PM +0200, Hostinger NOC wrote:
> Package: linux-image-6.1.0-43-amd64
> Version: 6.1.162-1
> Severity: important
> Tags: upstream
>
> Hello,
>
> The ip6_tunnel kernel module is broken in linux-image-6.1.0-43-amd64
> (upstream 6.1.162). IPv6 tunnels (mode "any") fail to encapsulate or
> transmit any packets.
>
> WORKING CONFIGURATION:
> - Kernel: 6.1.0-37-amd64 (upstream 6.1.140)
> - Tunnel works correctly with identical configuration
>
> BROKEN CONFIGURATION:
> - Kernel: 6.1.0-43-amd64 (upstream 6.1.162)
> - Same tunnel configuration fails completely
>
> REPRODUCTION STEPS:
> 1. Create IPv6 tunnel with mode "any":
> ip link add tun0 type ip6tnl mode any \
> remote 2001:db8::2 \
> local 2001:db8::1 \
> encaplimit 4
> ip addr add 192.0.2.1/30 dev tun0
> ip addr add 2001:db8:1::1/126 dev tun0
> ip link set tun0 up mtu 1500
>
> 2. Send traffic through tunnel:
> ping -c 3 -I tun0 192.0.2.2
>
> 3. Monitor physical interface:
> tcpdump -i eth0 -nn 'ip6 and host 2001:db8::2'
>
> EXPECTED BEHAVIOR (kernel 6.1.0-37):
> - tcpdump shows encapsulated packets with protocol 60 (DSTOPT) or 4 (IPIP)
> - Tunnel traffic flows successfully
>
> ACTUAL BEHAVIOR (kernel 6.1.0-43):
> - tcpdump captures ZERO packets
> - TX counter increments on tunnel interface but packets disappear in
> kernel
> - RX packets show errors
>
> EVIDENCE - Side-by-side comparison:
>
> Kernel 6.1.0-37 (WORKING):
> Protocol 60 (DSTOPT) packets visible on wire
> Encapsulation functioning correctly
>
> Kernel 6.1.0-43 (BROKEN):
> 0 packets captured
> No packets transmitted to physical interface
>
> SYSTEM INFORMATION:
> - Debian version: 12 (bookworm)
> - Architecture: amd64
> - Broken kernel: 6.1.0-43-amd64 (6.1.162-1)
> - Working kernel: 6.1.0-37-amd64 (6.1.140-1)
> - Boot parameters tested: default and mitigations=off (both fail)
>
> IMPACT:
> - All IPv6 tunnels using ip6tnl module are affected
> - Workaround: downgrade to 6.1.0-37
>
> ADDITIONAL NOTES:
> - Issue persists regardless of tunnel parameters (MTU, encaplimit,
> offloads)
> - Module parameter changes have no effect
> - Issue affects both encapsulation (TX) and decapsulation (RX)
> - No relevant errors in dmesg or kernel logs
>
> Could you please investigate this regression? Is there any additional
> debugging information or testing I can provide to help identify the root
> cause?
> Would additional debugging output be helpful? I can provide kernel
> logs, strace output, or test patches if needed.
See #1127597.
Can you confirm that applying 9990ddf47d41 ("net: tunnel: make
skb_vlan_inet_prepare() return drop reasons") on top as well of
6.1.162-1 (or 6.1.163 upstream) fixes the issue and report back
please?
Regards,
Salvatore
