t206
commented
4 years ago I'm also experiencing this failure to boot at board power up on the Raspberry Pi 4B. It seems like it started a couple of weeks ago and doesn't affect rebooting an already powered-on Pi - it only fails at power cycling. On average, I have to power cycle 3-4 times but once yesterday I power cycled 7 times before it came up. Thinking that this might be an issue with 18.04 (kernel 4.19.105 v8), I upgraded the Pi to Ubuntu 19.10 which I'm running quite successfully now. Unfortunately, this didn't resolve the power-cycle lockups. They're no worse under 19.10 but no better. Finally, I also tried upgrading 19.10 to 20.04 using the 4.19.105 v8 kernel but was not successful. Numerous packages failed to install so I finally gave up on this path.
I then started over with the newest Unbuntu 20.04 server distro and added the Mate desktop. That did indeed come up with the 5.4.x kernel but overall, the Pi4B is unstable. It'll run fine for an hour or so and then both LEDs go out and the system becomes unresponsive and requires a power cycle to start up again. What did get fixed however is the power-cycle lock ups - those are completely gone using Ubuntu server 20.04/5.4.x kernel. The boot/firmware directory under 20.04 is different than what we have under 18.04 and 19.10.
I don't see anything on my monitor when the kernel locks up in 18.04/19.10 but I'd be happy to troubleshoot the power-cycle issue if someone knows how to get the monitor to display startup messages.
TheRemote
Stuck/kernel paniced boots at random with likely v3d conflicts, i.e. "IRQ request failed on -22" [v28]
Hi James!
The Ubuntu Server 64bit has been working excellently for months up until now - good times. However, apparently some of the recent cutting-edge driver updates for the RPi4B's internal chips have caused things to go haywire very often during the boot-up process.
Strangely enough, and on somewhat of a bright side, the issue seems to resolve itself by unplugging the power and rebooting the Pi - usually after a 1-3 reboot attempts, I'm back into the OS and it's running again without any problems at all and everything is working as if nothing had happened.
The kernel panics/stuck boots occur during the boot sequence when Ubuntu is loading up and the Xubuntu desktop hasn't started up yet. I've tried to replicate this problem on two up-to-date RPi4B units and it persists in both of them.
The issue has started to occur quite recently. During both cold boots and soft reboots when the actual Ubuntu's side is loading up, I get stuck with (usually) either a kernel paging request error or an IRQ request failing at -22.
I was wondering whether that could be related to the task prioritization in this particular Ubuntu Server fork that's been done back in the day to get the audio working without crackling?
The problem seems to revolve around the v3d modules and drivers for the onboard chips. It's gotten worse with recent driver updates, so... Here's a snippet of the
dmesgerror messages I've been getting (note also the nags about the tainted kernel -- could it be one of the culprits that since there's already 5.4.y kernel tree being pushed out there, and from what I've read, for Ubuntu 18.04.4 LTS as well, that there's probably "a bit of a mismatch" now..?)As a side question (which is likely related to this issue), any idea on when you'll be able to crank things up to the latest 5.4.y kernel branch or if you're even planning to do it? If you're not interested, can you give any tips on how to try the latest official 5.4.y kernel headers on the 64bit Ubuntu Server? Thanks so much for your hard work, hope you are doing alright & all the best.