Hi all,

I have a desktop PC that I have used the same Arch install on since late 2019. It has worked very well during that time, but recently started to exhibit issues. These issues started to surface at the same time, making me suspect they're related, though I don't know that for sure at the moment. I'll summarize them below. I'll try and be too verbose, but apologies in advance for the wall of text; I wanted to be sure not to miss any (potentially) relevant details. Any help will be appreciated :)

The symptoms

First off, pacman can't perform a full upgrade, mentioning multiple errors of the form

error: <package-name>: signature from "John Doe <johndoe@archlinux.org>" is unknown trust
:: File /var/cache/pacman/pkg/<package-name>-<package-ver>-x86_64.pkg.tar.zst is corrupted (invalid or corrupted package (PGP signature)).

Typically I would suspect an outdated keyring or a corrupted file in the cache. Thing is, I have already repeatedly updated the keyring (pacman -S archlinux-keyring), repopulated the keyring (pacman-key --init) and cleared the complete package cache (pacman -Sc). When I do all that and again try to upgrade, the same offending packages prevent it from succeeding. And these aren't no-name packages; they include things like the kernel and Docker.

Secondly, Firefox seems to be completely unusable due to stability issues. Sometimes this means that navigating to any address completely blanks that tab, other times that firefox will indefinitely try and load a page. Also very commonly, I get the dreaded Firefox crash screen. Somewhat curiously, Private Tabs don't seem to share this instability, so I figured it must have something to do with my profile. However, I nuked the original offending profile since I had found that it often helps with these sorts of stability issues. However, shortly after doing so I started to encounter the same issues.

Then, there was a small bug where when opening my shell, it informed me I had a corrupted .zsh_history file. I didn't think too much of it, since I was able to write the contents of that file to a new history file, and that was that. It does have me wondering, though, if there isn't some underlying issue, either hardware or partition.

Debugging I've tried

This is an AMD Ryzen 5 3600 system, B450 chipset. I have 32GB of DDR4 (4x8G; two sets of CMK16GX4M2B3200C16). The root and home partition are both on an M.2 SSD, in an LVM-on-LUKS setup. Both partitions are ext4. There is also another data drive which I won't consider further.

I first ruled out a failing SSD; smartctl reported no issues on the drive. I then booted into a live USB, and tried to see if there might have been some issues with the root and/or home partition. badblocks in read-only mode reported nothing for either partition. e2fsck -f -y -C0 also failed to report anything worthwhile.

I figured bad RAM might have something to do with it, though I was suspicious of this because the errors seemed to much more cleanly relate to something in my filesystem. Also, one of the two 2x8 kits was relatively new. Nevertheless, I did a MemTest86+ test, which first time round did report a bunch of errors without an XMP profile enabled, but when I then re-ran it with XMP, it passed the test. Also, I keep a Windows dual boot on a separate drive, and I haven't been experiencing any similar issues over there, which I would expect in case of a RAM issue.

Halp

I'm at a loss what might be the cause of this, if the issues are even related, and what might be a good solution. I've been considering using some of the free space on the SSD to create a new root partition, and see if that solves things long-term. But if someone else has a better idea, I'm all ears. Any other tests I could do to rule out corrupted ext4 partitions? Thanks!