Re: October 2026 stabilization week

From: Gleb Smirnoff <glebius_at_freebsd.org>
Date: Fri, 30 Oct 2026 21:14:47 UTC 
On Mon, Oct 26, 2026 at 01:00:17AM -0700, Gleb Smirnoff wrote:
T> This is an automated email to inform you that the October 2026 stabilization week
T> started with FreeBSD/main at main-n290336-175595ba6c52, which was tagged as
T> main-stabweek-2026-Oct.

The A/B testing at Netflix didn't discover new performance or stability
regressions.

Experience with amd64 laptop (Intel) and desktop (NVIDIA) is also positive.

fluffy@ reported assertion failure in libtht on aarch64.  After a week of
bisecting he came to a conclusion that there was some ABI breakage, and if all
binaries are rebuilt the problem is gone.  He may update with more details in
separate email.

Testing runs experience increased rate of panics induced by
tests/net/if_clone_test:lo_ipv6_up_stress. It seems to me not much has changed
in the relevant code in the last month.  There are known bugs there, but it is
unclear what made these races more probably to hit.

Since it is already Friday, the stabweek declared as closed, albeit we have
two not resolved problems :(

-- 
Gleb Smirnoff