W dniu 23.09.2025 o 22:23, Gleb Smirnoff pisze:
On Mon, Sep 22, 2025 at 01:00:30AM -0700, Gleb Smirnoff wrote:
T> This is an automated email to inform you that the September 2025
stabilization week
T> started with FreeBSD/main at main-n280533-f72908c94df9, which was tagged as
T> main-stabweek-2025-Sep.
The A/B testing at Netflix did not discover any performance neither stability
regressions compared to August stabweek.
My personal desktop with nVidia upgrade went almost smoothly, except that
COMPAT_FREEBSD14 seems to be needed for X to work. Note that X doesn't exit
due to bad syscall, just doesn't see the nvidia device. I wasn't able to find
out what is the real requirement for COMPAT_FREEBSD14.
Laptop upgrade has a regression - USB ports do not work after resume. Poul-
Henning also reports the same for a different Thinkpad version. This definitely
needs to be bisected. The problem is that both me and Poul-Henning have
skipped the August stabweek on our laptops (this is why it is important to have
the packages in the repo, btw!), so the regression could come from August.
Right now at EuroBSDCon using laptop as main terminal, I can't bisect this
until next week.
There is a known regression with nd_ifinfo ABI, and a fix 0ec13430c583 has been
committed.
The stabweek is declared as done and advisory freeze on src is thawed.
After updating to FreeBSD 16.0-CURRENT #119 main-n280607-942916a548ab
(based on commit 56e01d0d7e0f7c2129c05467ca99d5f5538f52bc in main
branch, post-stabweek tag), the USB regression is gone and the desktop
is working properly again. This might also be related to having upgraded
packages to the native ABI, since 16.0-CURRENT packages have only
recently become available.
Cheers
--
Marek Zarychta
