[Solved] Raspberry Pi 2.0 freezes after latest updates

[McFox]

Hi,

My RPI2 has been running Arch Linux Arm since I got it, and it has mostly worked fine. Now, I did a pacman -Syu the other day, and since then it has started to drop its' network connection for some reason. The intervals between drops have been anything between 15min and 5 hours. Disconnecting and reconnecting the ethernet cable or rebooting the device gets it back up, but rebooting my router doesn't, so now I can't access it remotely whenever it goes down.
dmesg doesn't say anything about anything going down, all I get is the normal "interface down -> interface up" messages caused by the cable being disconnected and reconnected. Could it possibly have something to do with dhcp? I'm using that on the rpi, as it was easier to set up that way (I have static dhcp for it on my router instead).

Does anyone know what could cause this? I'm gonna set a static ip for it once I get home, but if that doesn't solve it, I don't have a clue on how to proceed :/ Also, why would it stop working all of a sudden?

[johnnyvibrant]

I'm on static IP and this is happening to me . Almost exact same issue

[WarheadsSE]

Please check dmesg & journal for error messages whhhen this occurs.

[johnnyvibrant]

ok, from journal:

Oct 05 13:43:35 rpi2 systemd[432]: Listening on D-Bus User Message Bus Socket.
Oct 05 13:43:35 rpi2 systemd[432]: Reached target Sockets.
Oct 05 13:43:35 rpi2 systemd[432]: Reached target Paths.
Oct 05 13:43:35 rpi2 systemd[432]: Reached target Timers.
Oct 05 13:43:35 rpi2 systemd[432]: Reached target Basic System.
Oct 05 13:43:35 rpi2 systemd[432]: Reached target Default.
Oct 05 13:43:35 rpi2 systemd[432]: Startup finished in 175ms.
Oct 05 19:40:15 rpi2 systemd[432]: Time has been changed
Oct 05 19:41:50 rpi2 su[468]: (to root) john on pts/0

my rpi shutdown between quarter to two and seven forty (when i got home from workand manually plugged it back out and in), as far as i know nothing happened between that time period and none of the messages above shout error to me, but im no professional, any help please?

[McFox]

It's getting worse for me now, it falls out of the network pretty much straight away after booting and disconnecting and reconnecting the ethernet cable doesn't work anymore... :/ My RPI is running "headless" (connected to my TV, no keyboard, running Kodi and controlled using CEC or a Kodi remote app when CEC fails), so I can't get any logs right now, but last time I was able to access it dmesg didn't give anything helpful.

I'll connect a keyboard to it after work and see what actually happens... The RPI doesn't freeze altogether though, as Kodi still keeps running, but yesterday when I tried to go into network settings it froze. So maybe there's a bug in the latest firmware, which is causing the NIC to stop working?

[McFox]

I didn't get anything helpful from dmesg & journalctl, but now I noticed that the system actually freezes. So I disabled a bunch of services (kodi, transmission, sickbeard, couchpotato, mpd, minidlna), and after that the system seems to run ok. Then I started enabling them one by one, starting with kodi which worked ok. Then I enabled transmission, and after starting it, it took maybe 5 minutes and the system froze! So apparently that's the culprit... But I only have one transmission package in cache, so I don't think it was affected by the update... I've enabled the rest of the services now and it's still running nicely, I'll report back tomorrow...

johnnyvibrant: are you also using transmission?

[johnnyvibrant]

Yes, I got transmission and plex server running, I had sickrage and couch potato also but I disabled them. I'll disable transmission tonight and get back to you tomorrow with my findings

[McFox]

It froze again during the night :( So transmission is not to blame after all.

Yesterday when I was debugging this, what happened when it froze was that first it would become completely unresponsive, and after a while it started printing some kind of stacktrace/dump to the screen, saying something about CPU0 having stalled. Are those logged somewhere? I sure as hell am not writing those in here by hand, but if I can copypaste them from somewhere I can share the output...

I think I'll try downgrading the firmware today, and if that doesn't work, then I'll also downgrade the kernel. Another thing that was updated was systemd, so I guess I might downgrade that too if those two first options don't work. Then there was a filesystem update, but I kinda doubt that has anything to do with it... But then again, I'm no expert :D

A bit of noob question by the way: Is it possible to do some kind of rollback with pacman to a certain date? In theory it should be quite easy, but then there'd need to be a package list from before the update, right? I'm asking because I'd hate to start downgrading packages one by one, although by doing that I'd have a chance of figuring out which one causes the freezing...

[WarheadsSE]

The problem lies either in the kernel or the firmware. I would consider starting with the kernel.

[McFox]

I downgraded my kernel to 4.1.7-2, and now it seems to work like before :) It even booted much faster than with 4.1.10-1. I'll report back tomorrow, then we'll know for sure if that really solved it.