[ Resending with the full Cc list, which I dropped by mistake in my previous reply. ]
On Mon, Sep 28, 2026 at 04:19:47PM +0000, [email protected] wrote: > This is an automated message. This series looks like a fix, but its > commit messages seem to be missing some information: > > - Whether the issue was actually triggered, or is only theoretical > (e.g. found by code inspection). If it was triggered please include > the symptoms, like the stack trace or error messages. The issue was triggered, not only found by code inspection. I reproduced it under virt-ng with a local test on a VLAN-aware bridge with br_netfilter and IPv6 conntrack defrag enabled. The test sends a fragmented IPv4 packet on one VLAN, then a fragmented IPv6 packet on another VLAN that the bridge floods to several ports. The test is available on request. Symptoms: with the patch not applied, the refragmented IPv6 packets leave the bridge with the wrong VLAN tag. This is the tag of the earlier IPv4 flow when both refragmentations run on the same CPU. Otherwise the packets go out untagged. There is no crash or warning; the only effect is the wrong or missing tag on the wire. With the patch applied, all the IPv6 fragments keep the correct VLAN tag. Thanks, Andrea
