Re: [PATCH net-next v2 0/3] ipv4/ipv6: Cache pmtu on all paths of a multipath route

["Vladimir Vdovin" <deliran@verdict.gg>]

On Tue, Nov 10, 2026 at 10:06:44AM +0000, netdev-bot+sinfo wrote:
>  - How the issue was discovered, e.g. hit in production, hit during
>    development, syzbot report, manual code inspection, LLM or static
>    analysis tool scan.

Patch 2 (ipv6) is the counterpart of the IPv4 issue fixed in commit
7d3f3b4367f3 ("net: ipv4: Cache pmtu for all packet paths if multipath
enabled") [1]: a host with two ECMP uplinks and a smaller MTU further
down the path had flows blackholing after a Packet Too Big. The IPv6
side was deferred back then at David's request [2]. The new pmtu.sh
tests in patch 3 reproduce it and fail without patch 2.

Patch 1 (ipv4) was found by code inspection while reworking v1 after
the Sashiko (LLM) review of it [3]: the same "replace the exception"
step that review flagged in the IPv6 code is in the IPv4 fan-out from
7d3f3b4367f3 as well.

[1] https://lore.kernel.org/netdev/20241108093427.317942-1-deliran@verdict.gg/
[2] https://lore.kernel.org/netdev/ad29493c-9e48-4a94-b327-e47af3a9d137@kernel.org/
[3] https://lore.kernel.org/netdev/179236119085.3135.2701233313088371263@kernel.org/