[SOLVED] Kernel panic after removing installation USB

apollinaris_mons · 2026-08-05 15:08:56

Good day. I'm installing arch on an unused 2018 HP envy (13-ah0003nk) laptop and I'm having some issues which are still beyond me after a week of troubleshooting and a few fresh installations. I hope the answer isn't too obvious and forgive me for the formatting.

Arch installs (manually; from the wiki) without a problem and seems to run perfectly until I remove my installation USB. Whether the laptop's turned off or on when I remove it, my system keeps accepting commands for a few seconds (or from off, fully boots, gives me time to log in and quickly run a command), and then the cursor disappears, the image freezes, and after a few seconds my caps lock light starts blinking (more than 50 blinks, so not one of HP's own signals for hardware failure, not that this rules out hardware failure). Sometimes this leads to it booting again, and sometimes it stays shuts down. If I plug in the USB from off and power on, the system boots back to normal. Only once so far have I had the same thing happen when booting to the installation USB. The boot manager shows OS boot Manager (UEFI) — GRUB as well as the disk name, and seems to boot to the right place.

Please let me know what information I should supply. I'm not dual-booting; the arch image is verified and the latest. Below is what I've noticed by myself.

journalctl contains the following:
x86/cpu: SGX disabled or unsupported by BIOS; remains after I enable SGX in BIOS, currently disabled
MDS CPU bug present and STM on, data leak possible. See https..
MMIO Stale Data CPU bug present and STM on, data leak possible. See https..
VMSCAPE: SMT on, STIBP is required for full protection. See https..
A number of ACPI errors.
nouveau 000:01:00.0: [drm] No compatible format found; I haven't installed NVIDIA drivers assuming/really, hoping they may not be related to this.
File /var/log/journal/df724743e98b42809d04415b7f68a4ee/system.journal corrupted or uncleanly shut down, renaming and replacing.
FAT-fs (nvme0n1p1): Volume was not properly unmounted. Some data may be corrupted. Please run fsck. This is the target disk's boot partition. I checked lsblk after umount -R /mnt and before rebooting at the end of the installation process and none of the partitions remained mounted, and only the swap partition was on ([SWAP]). I'm guessing this isn't what that means.
HP wmi fan errors
hp_bioscfg: Returned error 0x3, "Invalid command/value/Feature not supplied"
And finally Activation request for 'org.freedesktop.resolve1' failed: The systemd unit 'dbus-org.freedesktop.resolve1.service' could not be found.

Running fsck /dev/nvme0n1 before partitioning, after partitioning, after formatting and after completing the installation returns the following:
ext2fs_open2: Bad magic number in super-block
fsck.ext2: Superblock invalid, trying backup blocks...
fsck.ext2: Bad magic number in super-block while trying to open /dev/nvme0n1

The superblock could not be read or does not describe a valid ext2/ext3/ext4 filesystem. If the device is valid and it really contains an ext2/ext3/ext4 filesystem (and not swap or ufs or something else), then the superblock is corrupt, and you might try running e2fsck with an alternate superblock...
Running fsck /dev/nvme0n1p1 (on the boot partition) says that there are differences between boot sector and its backup, and that Dirty bit is set. Fs was not properly unmounted and some data may be corrupt. It finds no problems with the root and home partitions, and ignores swap.
I'm only running fsck from the installation USB and while unmounted.

Formatting root and home (I'm partitioning them separately) returns /dev/nvme0n1pX contains 'ISO-8859 text, with very long lines (1028), with no line terminators' data; Proceed anyway? I haven't seen this message on others' installations online so far.

My fstab file reads fmask=0022,dmask=0022, at odds with the wiki's examples (https://wiki.archlinux.org/title/Fstab#Usage).

When working through the localisation commands in chroot, I run vim /etc/vconsole.conf and enter my KEYMAP variable; write and quit; and cat /etc/vconsole.conf returns my input. Running mkinitcpio -P shows ==> WARNING: sd-vconsole: "/etc/vconsole.conf" not found, will use default values. After this, cat /etc/vconsole.conf isn't found.

Sorry for supplying a lot of information which is likely unrelated. I don't know how I could go about saving a copy of journalctl -b after unplugging the USB and before the kernel panicking, so if this would be helpful to have, please let me know how to.

Thanks!

Last edited by apollinaris_mons (2026-08-06 13:53:29)

seth · 2026-08-05 20:31:20

journalctl contains the following:

Please use [code][/code] tags and don't post isolated line fragments out of context, https://bbs.archlinux.org/viewtopic.php?id=57855

and after a few seconds my caps lock light starts blinking

That's a kernel panic, the corrupted filesystem will be a consequence of the hard reboot.

Formatting root and home (I'm partitioning them separately) returns /dev/nvme0n1pX contains 'ISO-8859 text

Is there currently supposed to be a filesystem on those partitions?

Can you boot the system after the installation w/o the install iso usb key being plugged in at all?

I don't know how I could go about saving a copy of journalctl -b after unplugging the USB and before the kernel panicking

What would likely not help anyway.
Boot the multi-user.target (2nd link below), login on the console (no gui!) run "dmesg -W" and yank the usb key.
In doubt make a photo of the screen to record the error messages - if they fly by too fast to read: your phone has a camera that can read faster than you. Use a tripod, in doubt makeshift one out of two books: -\_
Handheld video will probably just be a blurry mess.

apollinaris_mons · 2026-08-05 21:43:47

I'm running without a GUI and only a handful of packages (vim networkmanager linux linux-headers intel-ucode man-db man-pages), in case that helps.

Is there currently supposed to be a filesystem on those partitions?

They're newly made partitions each time I install arch. Formatting the boot partition doesn't give me the same message, but I figure that might be to do with formatting it as fat32, where home and root are ext4. I don't know if that answers your question.

Can you boot the system after the installation w/o the install iso usb key being plugged in at all?

I can, and I have enough time to sign in and run a few commands as it takes about 20 seconds for the screen to freeze. Might be worth mentioning that I can also boot without the iso USB plugged in, then plug it in before panic, and avoid panic. So it's as if something were happening about 20 seconds into a boot which depends on the USB. I can sit in BIOS as long as I like.

Boot the multi-user.target (2nd link below), login on the console (no gui!) run "dmesg -W" and yank the usb key.

I'm afraid the screen freezes before it outputs anything.

That's a kernel panic, the corrupted filesystem will be a consequence of the hard reboot.

Does every hard reboot result in a corrupted filesystem?

Sorry about the formatting and isolating logs.

seth · 2026-08-05 22:06:31

each time I install arch

Why do you repeatedly install the system?
Does it no longer boot after showing issues for the first time?
mkfs thinks you're asking it to format a textfile - some random disk sectors could contain some text, but not if the partition was already formated before…

Does every hard reboot result in a corrupted filesystem?

It will leave the dirty bit set for sure - actual integrity loss only occurs when (writing) changes were not synced before the reboot.

Does it have to be the install iso or do random (empty) usb keys or some usb periphery (keyboard/mouse/…) do as well?
After booting the system w/ the iso key plugged, what's the output of

lsblk -f

Maybe also just post a system journal from that condition
Please post your complete system journal for the boot:

sudo journalctl -b | curl -s -H "Accept: application/json, */*" --upload-file - 'https://paste.c-net.org/'

The key has to be used in some capacity unless this is a freak-electric hardware damage…

dimich · 2026-08-05 22:42:45

apollinaris_mons wrote:

Running fsck /dev/nvme0n1 before partitioning, after partitioning, after formatting and after completing the installation returns the following:
ext2fs_open2: Bad magic number in super-block

/dev/nvme0n1 is NVMe namespace. Partitions should be nvme0n1p1, nvme0n1p2 etc

apollinaris_mons · 2026-08-05 23:02:33

Why do you repeatedly install the system?
Does it no longer boot after showing issues for the first time?

I was convinced I'd done something wrong during the installation, so I worked my way through the wiki page more thoroughly each time. I'm coming from windows and a few raspberry pis, so between my inexperience and arch's towering reputation I figured I'd better make sure it wasn't obvious. I'm learning a lot.
The system boots again fine after showing issues and doesn't need reinstalling. Only once did the kernel panic seem to happen when I was booted into the iso usb, which as far as I know doesn't make sense, but it made both the image and the system keep panicking despite the usb being inserted, and required flashing and reinstalling. The usb has successfully installed three different images so far, so I'm not yet desperate enough to start doubting it.

Does it have to be the install iso or do random (empty) usb keys or some usb periphery (keyboard/mouse/…) do as well?

I hadn't thought of that. After checking, it seems to have to be the iso, and no matter the slot, either.

The output of lsblk -f and journalctl -b below.

https://paste.c-net.org/MundaneSpeaking
https://paste.c-net.org/RookieGerbil

apollinaris_mons · 2026-08-05 23:37:35

Below is the journalctl -b output for booting the system without the iso usb plugged in, then inserting it when the login screen appeared. I've diff'd the files after removing the dates and times but don't know what I'm looking for...

https://paste.c-net.org/JealousCabin

dimich · 2026-08-06 06:02:55

apollinaris_mons wrote:

I'm installing arch on an unused 2018 HP envy (13-ah0003nk) laptop

Search by this particular model gives several vague repors about instability with Linux, likely caused by some non-standard ACPI and power management issues. If so, the behavior may vary from one kernel version to another.
Can you try these kernel options, one by one?

acpi=noirq
acpi=off
pcie_aspm=off
pcie_ports=compat
pcie_port_pm=off
nvme_core.default_ps_max_latency_us=0

Also you can try to install linux-lts with Arch ISO and then boot it.

apollinaris_mons · 2026-08-06 08:01:02

The same thing happens when booting from linux-lts.

Can you try these kernel options, one by one?

Other than nvme_core (but not nvme_core.default_ps_max_latency_us), these don't appear in the output for lsmod, modprobe can't find them in /lib/modules/7.1.5-arch1-2, and does insmod says it can't find such files or directories. Please let me know if there's a different way of trying these.

Thank you for the help so far. I'll see what I can find about the laptop model. Is there any way of confirming or ruling out the hardware? How could the iso usb specifically be keeping the system alive?

dimich · 2026-08-06 08:31:37

apollinaris_mons wrote:

these don't appear in the output for lsmod, modprobe can't find them in /lib/modules/7.1.5-arch1-2, and does insmod says it can't find such files or directories.

These are not modules but kernel command line options. While in grub menu, press 'e' and append an option to existing command line. This is the most convenient method for one-time try.
Alternatively you can edit /boot/grub/grub.cfg or /etc/default/grub (and re-generate grub.cfg). However, this will be permanent change, so not convenient for test. If one of the options solves the issue, you'll probably want to make it permanent. In this case /etc/default/grub will be the proper place.

apollinaris_mons wrote:

Is there any way of confirming or ruling out the hardware?

I guess only by trial and error. It would be helpful to see full kernel panic message.

apollinaris_mons wrote:

How could the iso usb specifically be keeping the system alive?

Speculating, something keeps USB-related stuff open (partition block devices etc), USB hub or parent PCIe device remains powered, no power management transitions.

apollinaris_mons · 2026-08-06 08:46:47

I've tried appending your kernel options to the end of the grub commands-before-booting screen, on the line starting "linux". I left the iso usb in until after login, and if I didn't make it past login I removed it anyway after confirming I couldn't. Here's what I found.

acpi=noirq loaded the login screen with blinking cursor but took no keyboard inputs and forced my caps lock light on; unplugging iso usb made the cursor disappear and kernel panic.

acpi=off took over a minute on

    Booting a command list
Loading Linux linux-lts ...
Loading initial ramdisk ...

where it usually takes five or so seconds, then gave me a few lines of red error messages and loaded me into the login screen. Let me know if it would be useful for me to record these to catch them. After logging in, I unplugged the iso usb and it didn't panic! However, the when the arrow keys are used and there is no function they provide (pressing left and right when there are no characters on either side), the laptop makes the kind of beeping sound that it's made when booting to the BIOS. journalctl -b here https://paste.c-net.org/BradfordHyena

pcie_aspm=off, pcie_ports=compat, pcie_port_pm=off all load to successful login and usual kernel panic.

nvme_core.default_ps_max_latency_us=0 boots without lingering on the above Booting a command list, and doesn't panic when removing the iso USB. So, apparently perfect?

Should I add this into the grub config and generate it and call it a day?

It would be helpful to see full kernel panic message.

I wrote the above before you sent your last post. On panic, the screen appears to jitter before freezing which makes me think I might not be seeing everything that it's doing. Is there a command that could write any subsequent console output to the disk?

Thanks again.

seth · 2026-08-06 13:14:38

nvme_core.default_ps_max_latency_us=0 boots without lingering on the above Booting a command list, and doesn't panic when removing the iso USB. So, apparently perfect?

That'd be APST, https://wiki.archlinux.org/title/Solid_ … leshooting

After checking, it seems to have to be the iso

would be really odd and maybe more coincidentally keeps the nvme from entering a lower power state.

Should I add this into the grub config and generate it and call it a day?

You can look out for firmware updates (UEFI or the nvme) but APST issues are not uncommon so it's a frequently reasonable workaround.

apollinaris_mons · 2026-08-06 13:50:12

I'll keep an eye out since I still don't understand how I got locked out of the iso usb. Should I make a new topic and link this if I have more issues, or continue to update this one?

Thank you both for the help seth and dimich

dimich · 2026-08-06 22:33:46

apollinaris_mons wrote:

nvme_core.default_ps_max_latency_us=0 boots without lingering on the above Booting a command list, and doesn't panic when removing the iso USB. So, apparently perfect?

Hopefully it is.
It's common for non-standard hardware with vendor-specific quirks that something works fine with one kernel, got broken in next version, fixed in subsequent etc.

apollinaris_mons wrote:

Should I add this into the grub config and generate it and call it a day?

Yes, you can make the option permanent by adding it to GRUB_CMDLINE_LINUX_DEFAULT in /etc/default/grub and run "grub-mkconfig -o /boot/grub/grub.cfg" as root. Current options can be listed with "xargs -n1 -a /proc/cmdline".

On panic, the screen appears to jitter before freezing which makes me think I might not be seeing everything that it's doing. Is there a command that could write any subsequent console output to the disk?

Indeed, it may be hard to get panic message when it is caused by hardware/firmware issues and affects devices where the log shoult be displayed or stored. For future, if you want to experiment or investigate, you can try to disable drm_panic (see https://wiki.archlinux.org/title/Kernel … c_message) and maybe boot with "video=efifb". For machines with hardware serial port it's also possible to redirect output to serial console. However, with laptop it doesn't seem to be the case.

Arch Linux

#1 2026-08-05 15:08:56

[SOLVED] Kernel panic after removing installation USB

#2 2026-08-05 20:31:20

Re: [SOLVED] Kernel panic after removing installation USB

#3 2026-08-05 21:43:47

Re: [SOLVED] Kernel panic after removing installation USB

#4 2026-08-05 22:06:31

Re: [SOLVED] Kernel panic after removing installation USB

#5 2026-08-05 22:42:45

Re: [SOLVED] Kernel panic after removing installation USB

#6 2026-08-05 23:02:33

Re: [SOLVED] Kernel panic after removing installation USB

#7 2026-08-05 23:37:35

Re: [SOLVED] Kernel panic after removing installation USB

#8 2026-08-06 06:02:55

Re: [SOLVED] Kernel panic after removing installation USB

#9 2026-08-06 08:01:02

Re: [SOLVED] Kernel panic after removing installation USB

#10 2026-08-06 08:31:37

Re: [SOLVED] Kernel panic after removing installation USB

#11 2026-08-06 08:46:47

Re: [SOLVED] Kernel panic after removing installation USB

#12 2026-08-06 13:14:38

Re: [SOLVED] Kernel panic after removing installation USB

#13 2026-08-06 13:50:12

Re: [SOLVED] Kernel panic after removing installation USB

#14 2026-08-06 22:33:46

Re: [SOLVED] Kernel panic after removing installation USB

Board footer