You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am going to try to provide as much details as I can, however I am new to debugging kernel modules, and filesystems, please let me know if any additional information is needing, willing to debug if necessary. This does seem easy to reproduce though . My goal is to mount an encrypted bcachefs rootfs volume at boot.
Env:
-arch rolling
-util-linux 2.39.3
-kernel 6.7.x and 6.8.0
-config in mkinitcpio.conf for bcachefs in MODULES, HOOKS and BINARIES
How to reproduce:
Created a new filesystem (originally with bcachefs-tools last 1.3.x, updated to latest git version 1.6.4), --compression=zstd --acl --discard --metadata_replicas=2 --encryption, enter passphrase
Configured mkinitcpio.conf per above
Attempt to boot, asks for passphrase, I put in passphrase, enter it in, I see
"fsck binary is version 1.6: btree_subvolume_children but filesystem is 1.3: rebalance_work and kernel is 1.4 member_seq, using kernel fsck"
"bcachefs (someuid string): error requesting encryption key: ENOKEY"
Now if I use stable version of bcache-tools in arch (vs one in git, without the hooks, I see this when it tries to mount:
"mounting version 1.3: rebalance_work opts=ro,metadata_replicas=2,compression=zstd,degraded,fsck,fix_errors=yes,read_only
recovering from clean shutdown, journal seq 4302"
"Doing compatible version upgrade from 1.3: rebalance_work to 1.6: btree_subvolume_children"
running recovery passes: check_subvols,check_dirents"
... Along with a series of checks that say done"
and then finally see "Fatal error: No such device or address (os error 6)"
Failed to mount PARTLABEL=arch on real root
In emergency shell, If I try to mount manually with stable 1.6.4(mount /dev/sda4 /root) I get same error with os error 6, I get same error if using bcachefs mount. If I do bcachefs unlock /dev/sda4, I can then mount it. If using 1.6.4 (git version), I can not mount, even after unlock.
The text was updated successfully, but these errors were encountered:
FYI, I looked at how it was being mounted in the initrd script, it is doing something like this:
mount -t auto PARTLABEL=arch /new_root -o ro, which does not look that too out of the ordinary.
I am going to try to provide as much details as I can, however I am new to debugging kernel modules, and filesystems, please let me know if any additional information is needing, willing to debug if necessary. This does seem easy to reproduce though . My goal is to mount an encrypted bcachefs rootfs volume at boot.
Env:
-arch rolling
-util-linux 2.39.3
-kernel 6.7.x and 6.8.0
-config in mkinitcpio.conf for bcachefs in MODULES, HOOKS and BINARIES
How to reproduce:
"fsck binary is version 1.6: btree_subvolume_children but filesystem is 1.3: rebalance_work and kernel is 1.4 member_seq, using kernel fsck"
"bcachefs (someuid string): error requesting encryption key: ENOKEY"
Now if I use stable version of bcache-tools in arch (vs one in git, without the hooks, I see this when it tries to mount:
"mounting version 1.3: rebalance_work opts=ro,metadata_replicas=2,compression=zstd,degraded,fsck,fix_errors=yes,read_only
recovering from clean shutdown, journal seq 4302"
"Doing compatible version upgrade from 1.3: rebalance_work to 1.6: btree_subvolume_children"
running recovery passes: check_subvols,check_dirents"
... Along with a series of checks that say done"
and then finally see "Fatal error: No such device or address (os error 6)"
Failed to mount PARTLABEL=arch on real root
In emergency shell, If I try to mount manually with stable 1.6.4(mount /dev/sda4 /root) I get same error with os error 6, I get same error if using bcachefs mount. If I do bcachefs unlock /dev/sda4, I can then mount it. If using 1.6.4 (git version), I can not mount, even after unlock.
The text was updated successfully, but these errors were encountered: