[...] > > So I am - for myself - positively sure that the 2 FS corruptions I met > were related to Arch kernel 5.2 on this machine, as it happened right > after I had upgraded the kernel, had never happened before, and doesn't > happen since I downgraded the kernel.
Could you give more detailed history, including each reboot? Like: CASE 1 # Upgrade kernel (running 5.1) # Reboot # Kernel mount failure (running 5.2) CASE 2 # Upgrade kernel (running 5.1) # Reboot # Kernel mount success (running 5.2) # Doing some operations (running 5.2) # Reboot # Kernel mount failure (running 5.2) For case 1, as already explained, the damage is done using 5.1 not 5.2. For case 2, it's indeed more likely 5.2's fault. BTW, working case makes no sense here, as that's expected. (It's a really pity that the original corrupted leaf kernel message can't be preserved, that could really help a lot to detect memory corruption or things like that) Thanks, Qu > > I have to add however that I upgraded another little machine to Manjaro > kernel 5.2 - after taking a full clone of the FS - and I don't have met > any filesystem corruption so far. > > It is worth noting that Manjaro is the same family as Arch. > > > So even though I have no better logs to provide, here is my experience : > > - Arch kernel 5.2 : BTRFS over LVM over LUKS on a SSD, and BTRFS over > LUKS on an USB HD : 2 filesystem corruptions. Both using numerous > snapshots, some were deleted (either by snapper or manually). Downgraded > to 5.1 now OK. > > > - Manjaro kernel 5.2 on a small laptop, BTRFS over LUKS on eMMC, no > compression, no snapshots, no problem so far. > > > - Manjaro kernel 5.2 on another laptop for a very short while before > reverting to 5.1, BTRFS over LVM over LUKS on SSD, a few snapshots, I > dunno if some were deleted (snapper) : Still OK. > > - Manjaro kernel 5.2 on a desktop for a very short while before > reverting to 5.1, BTRFS RAID-1 over bcache over LUKS on a 2 HD + 1 SSD > mix, a few snapshots, I dunno if some were deleted (snapper) : Still OK. > > So you see the setups can be a bit complex : Always a LUKS layer, > compression used on mechanical HDs, sometimes LVM or bcache, some BTRFS > RAID on one system... > > As far as I can tell, the issue doesn't relate to the most complex setups. > > > I am under the unproved but strong feeling that the mess has something > to do with snapshots deletion with kernel 5.2... > > Dunno if it can be of some help. > > Kind regards. > > ॐ >
