System works with linux514 but not with linux515

Thanks for the tip! Glad to see there are other in our camp. I noticed that, on my machine, these hangs would typically come from network-related drivers. Things such as ‘igb’, ‘rtl8169’ and ‘cfg80211’. Though, not sure why it would effect programs like ‘sudo’, except that I know it tries to do host resolution.

For me, 5.15.0 works OK, but 5.15.1 is broken. There is a small handful of changes that were made, but nothing effecting networking :thinking:

I don’t think it’s Manjaro specific either. I had Ubuntu mate installed with a self compiled 5.15.x when the problem started. Things got so confusing that I opted to reinstall with Manjaro, all good until i tried 5.15 again. As pointed out before, 5.15.0 was ok for me and I think it happened when i updated it to 5.15.4 from there. So maybe the issue appeared between .0 and .4.

PS. I was going to give .8 a shot too so thanks for saving me the time chuck_r.

In my case , lspci is hanging, ifconfig is hanging so i thought my Intel NIC was bad, I replaced it with a new one ( still with the igb driver ), but the problem persisted and that’s when I started to look at a software issue.

If anything , I’ll see if i can recompile 5.15.0 and move through .1 and .2 to see when the problem appears.

So far the problem seems to have appeared in .3. I reverted to .2 for a while. Unfortunately the changelog fro, .2 to .3 is quite extensive

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/diff/?id=v5.15.2&id2=v5.15.3&dt=2

Update: attempting a bisect now to try to isolate the issue. I just saw they released v5.15.10 as well but at this point i’m too curious :slight_smile:

So I thought of posting an update. In my bisecting i found out the issue goes back to 5.15.0, so I ran a bisect up until 5.14.21 which works - In fact , the way to reproduce in my case was to simply reboot. Any 5.15.x kernel up to .10 would cause NetworkManager to hang on stop.

I filed https://bugzilla.kernel.org/show_bug.cgi?id=215359 with the details, but bisect pointed to

commit f32a213765739f2a1db319346799f130a3d08820 Author: Heiner Kallweit <hkallweit1@gmail.com> Date: Sun Aug 1 12:36:48 2021 +0200 ethtool: runtime-resume netdev parent before ethtool ioctl ops

I reverted it to test, and 5.15.10 now reboots correctly - will wait the outcome of that bug now

1 Like