> It's interesting how they found the unused code.
From the article: the code was broken.
The breaking bug was discovered in 2023 by syzbot, a fuzzer, and found out to have been introduced in 2016. This means that probably nobody has been using UDP-Lite (at least on a recent kernel, even LTS) for quite some time now.
It is now 2026, it has been proposed and discussed to remove UDP-Lite entirely, the patch set has gone through several iterations on the netdev mailing list. Apparently nobody complained that, actually, they do need that and it has been merged to the netdev tree, likely ending up in the next release.
I must admit just from reading the description, it doesn't sound that the correct inference is that it's never been used.
"In 2023, syzbot found a null-ptr-deref bug triggered when UDP-Lite attempted to charge an skb after the total memory usage for UDP-Lite _and_ UDP exceeded a system-wide threshold, net.ipv4.udp_mem." to me reads that if the total memory usage never exceeded that threshold then the bug wouldn't trigger. So, wouldn't this bug only affect people who changed that threshold down below the current usage? Because otherwise, usage wouldn't go above the threshold anyway?
And just because the kernel is logging a deprecation notice, there's no guarantee that anyone would ever see that, depending how often it was logged.
But that said, I'd never even heard of this feature, and wouldn't be at all surprised if many routers hadn't just silently dropped these packets anyway because they didn't recognise the protocol version.