Hi,
On Sat, May 02, 2026 at 08:12:42AM +0200, Salvatore Bonaccorso wrote:
> Hi,
>
> Thanks for your report.
>
> On Fri, May 01, 2026 at 09:24:43PM -0400, podorski wrote:
> > Source: linux
> > Version: Regression introduced in 6.1.170-1 / 6.1.0-45-amd64
> > Severity: important
> >
> > Dear Maintainer,
> >
> > *** Reporter, please consider answering these questions, where appropriate
> > ***
> >
> > * What led up to the situation?
> > updating kernel via apt on 20260501
> >
> > any ping from my host to a "down" host on my lan causes these messages to
> > pile up in the ring buffer
>
> This looks to be a consequence of/related to the backport of
> 81b84de32bb2 ("xfrm: fix ip_rt_bug race in icmp_route_lookup reverse
> path") which landed in 6.1.165 upstream and thus included as well in
> the recent bookworm-security update.
>
> The change is net_warn_ratelimited() but it is not immediately clear
> to me yet why this triggers on a bookworm kernel, whereas it does not
> with on a trixie host with 6.12.85-1.
>
> We will have a look.
So indeed, given this is the one emmiting the message reverting the
commit in 6.1.y "fixes" the problem. I'm going to contact upstream /
stable maintainers to see how to proceed, there might be a missing
requisite in 6.1.y causing the problem.
Regards,
Salvatore
