You are not logged in.
Hi,
Have been trying to figure out what the hick-up is with my Lenovo G50-45 (AMD A8-6410 APU with AMD Radeon R5 Graphics) but realize I need help. Upgraded from 5.18.14 to 5.18.15 yesterday and after that the boot get stuck around initializing usb devices and I have to do a hard reset. I have been trying to extract logs etc. but I guess the boot gets stuck before systemd is started so there are no logs generated. If I revert back (using usb installation image) to linux + linux-headers to 5.18.14 everything works again.
At the failed boot I get to this:
[ 1.879597] usb 2-1.3: Manufacturer: Broadcom Corp
[ 1.879599] usb 2-1.3: SerialNumber: 38B1DBD9DBA6
STUCK
For 5.18.14 boot the next line is:
[ 1.879929] random: fast init done
So I have thinking that it has something to do with random, but don't know how to proceed.
Last thing is that it is more like a stuck, not a freeze or crash because if I attach a usb stick a new line is added with "... New USB device found..."
Help me Obi-Wan Kenobi, you are my only hope
EDIT:
I can confirm that adding "spectre_v2=off" to kernel boot string fixes the issue. Thanks everyone! I have no idea what the "spectre_v2=off" is about but no I have something to read up on
Cheers / Henrik
Last edited by k96hkh (2022-08-29 11:03:01)
Offline
You didn't post full log output. Start system, wait to stuck, restart from arch live usb, arch-chroot to your installed system and post full output of command "journalctl -b" or "journalctl -b-1".
That Broadcom usb device is WiFi dongle?
Maybe you didn't have mounted /boot when you last updated?
Offline
Does it look like any of the screenshots in https://bbs.archlinux.org/viewtopic.php?id=278529 or the other thread referenced there?
Online
Hi,
I would like to notice that I have the same problem or at least it seems. I upgraded to linux-5.18.15 about 40 minutes ago and rebooted. Then my kernel just stops at the stage of detecting USB devices (listing them in particular). Then nothing happens. If I plug in USB stick at the moment, kernel logs detection of a new usb device as usual. Expected behavior here is to continue loading. Usually it writes something like `starting version 250.2-arch`, it happens relatively fast so I'm not sure what and then prompts for my luks key for root partition. Since there was no root partiton mounted I would have no logs for the time. Let me see if I can have a repro with kernel logs sent via UDP.
As for similarity to screenshots, well, it happens after switching clocksource. But I don't have any stacktrace on screen.
Last edited by akshimassar (2022-07-31 21:39:21)
Offline
I got almost full logs. Not best quality, but at least something.
1 - https://ibb.co/614ms4m
2 - https://ibb.co/YXv3Qn4
Runs fine with LTS.
Last edited by lukp0t (2022-07-31 21:58:26)
Offline
Does it look like any of the screenshots in https://bbs.archlinux.org/viewtopic.php?id=278529 or the other thread referenced there?
Also, seems like in all cases it has to do something with kworker. But I'm not sure.
Last edited by lukp0t (2022-07-31 22:07:16)
Offline
Same for me here. Hangs with kworker, RIP: 0010:efi_call_rts
No UDP logs unfortunately, just screen capture via phone.
And with boot_delay=50 it stops working at all, like no messages after "loading initrd"
Last edited by akshimassar (2022-07-31 22:20:03)
Offline
Lower the number, make sure to remove the quiet parameter - it delays every (line of) kernel message w/ 50ms so 100 lines take 5 seconds to print (whether you can see them or not)
This looks like the same, ensure to enable early microcode patches.
efi_call_rts is probably already part of the bail from the actual problem.
Online
All of you have AMD CPU?
Offline
Yes, forgot to mention. I have AMD APU with Radeon Graphics too.
Offline
Confirming, updating an old system with an AMD Athlon 5350 to 5.18.15 and boot stuck with RIP: efi_call_rts at the top of the screen and then USB shenanigans continue to work as the devices enter and exit low power modes.
Seems related to retbleed fixes? https://lore.kernel.org/lkml/2022072812 … cal.com/T/
For my disgrace this just happened after I exchanged SATA cables... not funny looking at the issue wrongly for half an hour!
Offline
Yup, that was it, add "spectre_v2=off" to the boot command line, and the issue is gone. As per the kernel.org thread above, there is already a fix, I guess it will be available soon in the packaged kernel here.
Last edited by uka (2022-08-01 14:06:51)
Offline
Yup, that was it, add "spectre_v2=off" to the boot command line, and the issue it's gone. As per the kernel.org above, the issue seems to be fixed already, I guess it will be available soon in the packaged kernel here.
this flag fixed this issue on my machine too. thanks uka!
Offline
Someone should report this issue to the Arch bug tracker.
Offline
I'm not sure if we should doing this now as this affects not only Arch kernel. We need to wait until new version will be available.
Offline
Offline
Yup, that was it, add "spectre_v2=off" to the boot command line, and the issue is gone. As per the kernel.org thread above, there is already a fix, I guess it will be available soon in the packaged kernel here.
Hello!
I also have an AMD Radeon and had this same problem with this kernel version (I use Zen kernel). Luckily I also have the LTS kernel installed, and through that I downgraded to version 5.18.14.
How do I add this flag "spectre_v2=off" at boot? I don't use GRUB, I use systemd-bootctl.
Last edited by kevinlucasilva (2022-08-01 16:24:51)
Offline
Nothing is too wonderful to be true, if it be consistent with the laws of nature -- Michael Faraday
Sometimes it is the people no one can imagine anything of who do the things no one can imagine. -- Alan Turing
---
How to Ask Questions the Smart Way
Online
See https://wiki.archlinux.org/title/Kernel … figuration and follow the instructions for your boot loader.
Offline
See https://wiki.archlinux.org/title/Kernel … figuration and follow the instructions for your boot loader.
Thx buddy, works!
Offline
But it it still interesting: retbleed mitigation was introduced with 5.18.14 kernel version, but your problems started with version 5.18.15. There was something more to change... but somewhat related.
Does "retbleed=off" kernel parameter also work?
UPDATE:
That is not really solved! That is just a workaround.
Last edited by xerxes_ (2022-08-01 20:35:17)
Offline
The 'spectre_v2=off' worked for me as well. What is the best way to track when I should remove the flag from my bootloader entry?
Last edited by branthebuilder (2022-08-03 13:33:17)
Offline
Online
FWIW installed the updated kernel in testing and booted without the spectre_v2=off flag and all is well.
Offline
New kernel patch is out of testing. I can confirm that I was able to boot on AMD E2-6110 without spectre_v2=off parameter.
Now the issue is completely solved I guess.
Last edited by lukp0t (2022-08-04 09:20:27)
Offline