Re: [PATCH net-next] net: dropreason: add SKB_DROP_REASON_IP_TTL_EXCEEDED

[Junjie Cao]

On Tue, Aug 25, 2026 at 11:36 AM Fernando Fernandez Mancera
<fmancera@xxxxxxx> wrote:
> The patch is sound but isn't this drop reason a bit redundant? I mean, a
> simple check on traffic should have shown to the user that TTL is been
> exceeded.

Checking traffic for TTL presumes TTL is already the suspect. The reason
string is what points the investigation: in the linked bug, IP_INHDR read
as a broken header and the capture time went to checksums -- RX offload,
tc csum actions, both libvirt firewall backends. Nobody filters on TTL
while the kernel is saying the header is bad.

dropwatch and perf report the reason without a capture; that only helps
if the string is precise. IPV6_NDISC_HOP_LIMIT and TCP_MINTTL are as
visible in a capture as this one, and both have their own reason.

The exthdrs.c hunks also cover three drops that report no reason at all.

> In any case, net-next is currently closed [1]. Please send this after it
> re-opens.

Will repost when it reopens.