Error while unlocking encrypted BCacheFS: Required key not available
Hi!
Kernel 6.7.0-rc8 with BCacheFS new year fixes. Compliled with Debian gcc 13.2.0-9.
BCacheFS tools 1.3.3 – according to bcachefs version – from Debian package bcachefs-tools 24+really1.3.4-2.
Linux kernel keyutils from Debian package keyutils 1.6.3-2+b2. (Not sure whether really required.)
Created BCacheFS on external 4 TB SSD:
% mkfs.bcachefs -L […] --data_checksum xxhash --metadata_checksum xxhash --compression=lz4 --encrypted /dev/sda1
(also tried without xxhash, no difference)
Unlock attempt with incorrect passphrase:
% bcachefs unlock /dev/sda1
Enter passphrase:
incorrect passphrase
Unlock attempt with correct passphrase does not yield error message "incorrect passphrase". Key seems to be available:
% grep bcachefs /proc/keys1b9e7153 I--Q--- 1 perm 3f010000 0 0 user bcachefs:[… UUID …]: 32
UUID matches filesystem.
Still I get:
% LANG=en mount /dev/sda1 /mnt/zeit
mount: /mnt/zeit: mount(2) system call failed: Required key not available.
dmesg(1) may have more information after failed mount system call.
% dmesg | tail -1
[105441.695035] bcachefs ([…]): error requesting encryption key: ENOKEY
Why? And how to fix it?
I found
error requesting encryption key #93
https://github.com/koverstreet/bcachefs/issues/93
But I am not sure whether it applies to my situation.
I use Devuan with elogind. Do I need that pam related configuration change from comment
https://github.com/koverstreet/bcachefs/issues/93#issuecomment-609430340
?
I do not like to do it in case it is not required.
Best,
Signature · 1 line
Martin
Martin Steigerwald - 07.01.24, 12:22:53 CET:
Quoted 25 lines
Hi!
Kernel 6.7.0-rc8 with BCacheFS new year fixes. Compliled with Debian gcc 13.2.0-9.
BCacheFS tools 1.3.3 – according to bcachefs version – from Debian package bcachefs-tools 24+really1.3.4-2.
Linux kernel keyutils from Debian package keyutils 1.6.3-2+b2. (Not sure whether really required.)
Created BCacheFS on external 4 TB SSD:
% mkfs.bcachefs -L […] --data_checksum xxhash --metadata_checksum xxhash --compression=lz4 --encrypted
/dev/sda1(also tried without xxhash, no difference)
Unlock attempt with incorrect passphrase:
% bcachefs unlock
/dev/sda1
Enter passphrase:
incorrect passphrase
Quoted 5 lines
Unlock attempt with correct passphrase does not yield error message "incorrect passphrase". Key seems to be available:
% grep bcachefs
/proc/keys1b9e7153I--Q--- 1 perm3f0100000 0 user bcachefs:[… UUID …]: 32
Also keyctl sees the key in root user keyring:
% keyctl list @u
1 key in keyring:
463368531: --alswrv 0 0 user: bcachefs:[… UUID …]
In case this is an issue with Debian packaging of bcachefs-tools I can report there.
Quoted 34 lines
UUID matches filesystem.
Still I get:
% LANG=en mount
/dev/sda1/mnt/zeit
mount:/mnt/zeit: mount(2) system call failed: Required key not available. dmesg(1) may have more information after failed mount system call.% dmesg | tail -1
[105441.695035] bcachefs ([…]): error requesting encryption key: ENOKEYWhy? And how to fix it?
I found
error requesting encryption key #93
https://github.com/koverstreet/bcachefs/issues/93
But I am not sure whether it applies to my situation.
I use Devuan with elogind. Do I need that pam related configuration change from comment
https://github.com/koverstreet/bcachefs/issues/93#issuecomment-609430340
?
I do not like to do it in case it is not required.
Best,
Signature · 1 line
Martin
Quoted 2 lines · Martin Steigerwald
In case this is an issue with Debian packaging of bcachefs-tools I can report there.
What version of bcachefs-tools are you using? The default one in bookworm is v0.24 and is >1 year old and that may be the issue.
I'm starting to switch from btrfs to bcachefs and I backported v1.3.4 from sid.
AP
AP - 10.01.24, 03:13:01 CET:
Quoted 6 lines · Martin Steigerwald
Quoted 2 lines
In case this is an issue with Debian packaging of bcachefs-tools I can report there.
What version of bcachefs-tools are you using? The default one in bookworm is v0.24 and is >1 year old and that may be the issue.
As I wrote:
"BCacheFS tools 1.3.3 – according to bcachefs version – from Debian package bcachefs-tools 24+really1.3.4-2.".
Quoted 2 lines
I'm starting to switch from btrfs to bcachefs and I backported v1.3.4 from sid.
That is where above package is from.
Best,
Signature · 1 line
Martin
Quoted 10 lines · Martin Steigerwald
AP - 10.01.24, 03:13:01 CET:
Quoted 6 lines · Martin Steigerwald
Quoted 2 lines
In case this is an issue with Debian packaging of bcachefs-tools I can report there.
What version of bcachefs-tools are you using? The default one in bookworm is v0.24 and is >1 year old and that may be the issue.
As I wrote:
Quoted 7 lines
"BCacheFS tools 1.3.3 – according to bcachefs version – from Debian package bcachefs-tools 24+really1.3.4-2.".
Quoted 2 lines
I'm starting to switch from btrfs to bcachefs and I backported v1.3.4 from sid.
That is where above package is from.
The keyring stuff has been a perpetual utter headache.
I've been debating rewriting that stuff to just pass a memfd handle as a mount option and rip out keyring usage...
alternately - now that we're pretty much always mounting via the mount helper, perhaps it would be a little bit less fragile if the mount helper was adding the key to the keyring - that might be worth checking.
Hi!
Kent Overstreet - 10.01.24, 20:13:59 CET:
Quoted 21 lines · Martin Steigerwald
Quoted 11 lines
AP - 10.01.24, 03:13:01 CET:
Quoted 7 lines · Martin Steigerwald
Quoted 3 lines
In case this is an issue with Debian packaging of bcachefs-tools I can
report there.What version of bcachefs-tools are you using? The default one in bookworm is v0.24 and is >1 year old and that may be the issue.
As I wrote:
Quoted 8 lines
"BCacheFS tools 1.3.3 – according to bcachefs version – from Debian package bcachefs-tools 24+really1.3.4-2.".
Quoted 3 lines
I'm starting to switch from btrfs to bcachefs and I backported v1.3.4
from sid.That is where above package is from.
Quoted 8 lines
The keyring stuff has been a perpetual utter headache.
I've been debating rewriting that stuff to just pass a memfd handle as a mount option and rip out keyring usage...
alternately - now that we're pretty much always mounting via the mount helper, perhaps it would be a little bit less fragile if the mount helper was adding the key to the keyring - that might be worth checking.
So no suggestion what to try to make it work except for putting BCacheFS unto LUKS? I wanted to get rid of LUKS for removable media. Often enough I have "cryptsetup luksClose" complaining about still in use while I the filesystem on top of it clearly got unmounted already. In these case BTRFS still.
Ciao,
Signature · 1 line
Martin
Quoted 33 lines · Martin Steigerwald
Hi!
Kent Overstreet - 10.01.24, 20:13:59 CET:
Quoted 21 lines · Martin Steigerwald
Quoted 11 lines
AP - 10.01.24, 03:13:01 CET:
Quoted 7 lines · Martin Steigerwald
Quoted 3 lines
In case this is an issue with Debian packaging of bcachefs-tools I can
report there.What version of bcachefs-tools are you using? The default one in bookworm is v0.24 and is >1 year old and that may be the issue.
As I wrote:
Quoted 8 lines
"BCacheFS tools 1.3.3 – according to bcachefs version – from Debian package bcachefs-tools 24+really1.3.4-2.".
Quoted 3 lines
I'm starting to switch from btrfs to bcachefs and I backported v1.3.4
from sid.That is where above package is from.
Quoted 8 lines
The keyring stuff has been a perpetual utter headache.
I've been debating rewriting that stuff to just pass a memfd handle as a mount option and rip out keyring usage...
alternately - now that we're pretty much always mounting via the mount helper, perhaps it would be a little bit less fragile if the mount helper was adding the key to the keyring - that might be worth checking.
Quoted 5 lines
So no suggestion what to try to make it work except for putting BCacheFS unto LUKS? I wanted to get rid of LUKS for removable media. Often enough I have "cryptsetup luksClose" complaining about still in use while I the filesystem on top of it clearly got unmounted already. In these case BTRFS still.
I just made a couple suggestions. I'm sorry if they weren't exactly what you were looking for?
If you or someone else wants to help out by writing some code, I just laid out what needs to happen next - but I'm not your free helpdesk here, and you're expecting an imediate fix that's not how this works :)
Kent Overstreet - 11.01.24, 17:35:10 CET:
Quoted 6 lines
I just made a couple suggestions. I'm sorry if they weren't exactly what you were looking for?
If you or someone else wants to help out by writing some code, I just laid out what needs to happen next - but I'm not your free helpdesk here, and you're expecting an imediate fix that's not how this works :)
That is a misunderstanding. I am not expecting an immediate fix or a fix at all. I did not write I do either. So please do not read something into what I wrote that simply is not there.
My programming experience in C is more than 20 years old, certainly not kernel related and not even on the Linux platform and I thought there might be something I as an user might try to at least pin-point the cause of this issue a bit more. Apparently there is not and that is okay with me.
Nothing of what I do with BCacheFS has immediate urgency at the moment.
Signature · 1 line
Martin