Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fscrypt prevents ssh public key authentication even if home directory has been unlocked #116

Open
RomanValov opened this issue Dec 26, 2018 · 7 comments
Labels

Comments

@RomanValov
Copy link

@RomanValov RomanValov commented Dec 26, 2018

I've set up fscrypt home directory encryption on Ubuntu 18.04 following guide by @tlbdk.

Then I've tried to access my Ubuntu system using ssh public key authentication. Logically ssh server refuses to authorize using public key and forces ssh client to ask user for a password. However in contrast to pam_ecryptfs/pam_mount solutions password is still required for following simultaneous login attempts when home directory has been unlocked.

Having to say if ssh server is restarted after home directory has been unlocked -- public key authentication works.

@obadz

This comment has been minimized.

Copy link

@obadz obadz commented Jan 19, 2019

That's annoying.

Also it'd be nice if there was a way to have the "clear version" of the directory (which would just contain your ~/.ssh/authorized_keys as well as the encrypted version which contains everything so that you can still log into your account even if home not mounted yet (ecryptfs does do that).

fengguang pushed a commit to 0day-ci/linux that referenced this issue Mar 6, 2019
Add a new fscrypt ioctl, FS_IOC_ADD_ENCRYPTION_KEY.  This ioctl adds an
encryption key to the filesystem's fscrypt keyring ->s_master_keys,
making any files encrypted with that key appear "unlocked".

Why we need this
~~~~~~~~~~~~~~~~

The main problem is that the "locked/unlocked" (ciphertext/plaintext)
status of encrypted files is global, but the fscrypt keys are not.
fscrypt only looks for keys in the keyring(s) the process accessing the
filesystem is subscribed to: the thread keyring, process keyring, and
session keyring, where the session keyring may contain the user keyring.

Therefore, userspace has to put fscrypt keys in the keyrings for
individual users or sessions.  But this means that when a process with a
different keyring tries to access encrypted files, whether they appear
"unlocked" or not is nondeterministic.  This is because it depends on
whether the files are currently present in the inode cache.

Fixing this by consistently providing each process its own view of the
filesystem depending on whether it has the key or not isn't feasible due
to how the VFS caches work.  Furthermore, while sometimes users expect
this behavior, it is misguided for two reasons.  First, it would be an
OS-level access control mechanism largely redundant with existing access
control mechanisms such as UNIX file permissions, ACLs, LSMs, etc.
Encryption is actually for protecting the data at rest.

Second, almost all users of fscrypt actually do need the keys to be
global.  The largest users of fscrypt, Android and Chromium OS, achieve
this by having PID 1 create a "session keyring" that is inherited by
every process.  This works, but it isn't scalable because it prevents
session keyrings from being used for any other purpose.

On general-purpose Linux distros, the 'fscrypt' userspace tool [1] can't
similarly abuse the session keyring, so to make 'sudo' work on all
systems it has to link all the user keyrings into root's user keyring
[2].  This is ugly and raises security concerns.  Moreover it can't make
the keys available to system services, such as sshd trying to access the
user's '~/.ssh' directory (see [3], [4]) or NetworkManager trying to
read certificates from the user's home directory (see [5]).

By having an API to add a key to the *filesystem* we'll be able to fix
the above bugs, remove userspace workarounds, and clearly express the
intended semantics: the locked/unlocked status of an encrypted directory
is global, and encryption is orthogonal to OS-level access control.

Why not use the add_key() syscall
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

We use an ioctl for this API rather than the existing add_key() system
call because the ioctl gives us the flexibility needed to implement
fscrypt-specific semantics that will be introduced in later patches:

- Supporting key removal with the semantics such that the secret is
  removed immediately and any unused inodes using the key are evicted;
  also, the eviction of any in-use inodes can be retried.

- Calculating a key-dependent cryptographic identifier and returning it
  to userspace.

- Allowing keys to be added and removed by non-root users, but only keys
  for v2 encryption policies; and to prevent denial-of-service attacks,
  users can only remove keys they themselves have added, and a key is
  only really removed after all users who added it have removed it.

Trying to shoehorn these semantics into the keyrings syscalls would be
very difficult, whereas the ioctls make things much easier.

However, to reuse code the implementation still uses the keyrings
service internally.  Thus we get lockless RCU-mode key lookups without
having to re-implement it, and the keys automatically show up in
/proc/keys for debugging purposes.

References:

    [1] https://github.com/google/fscrypt
    [2] https://goo.gl/55cCrI#heading=h.vf09isp98isb
    [3] google/fscrypt#111 (comment)
    [4] google/fscrypt#116
    [5] https://bugs.launchpad.net/ubuntu/+source/fscrypt/+bug/1770715

Signed-off-by: Eric Biggers <ebiggers@google.com>
ColinIanKing pushed a commit to ColinIanKing/linux-next-mirror that referenced this issue Aug 16, 2019
Add a new fscrypt ioctl, FS_IOC_ADD_ENCRYPTION_KEY.  This ioctl adds an
encryption key to the filesystem's fscrypt keyring ->s_master_keys,
making any files encrypted with that key appear "unlocked".

Why we need this
~~~~~~~~~~~~~~~~

The main problem is that the "locked/unlocked" (ciphertext/plaintext)
status of encrypted files is global, but the fscrypt keys are not.
fscrypt only looks for keys in the keyring(s) the process accessing the
filesystem is subscribed to: the thread keyring, process keyring, and
session keyring, where the session keyring may contain the user keyring.

Therefore, userspace has to put fscrypt keys in the keyrings for
individual users or sessions.  But this means that when a process with a
different keyring tries to access encrypted files, whether they appear
"unlocked" or not is nondeterministic.  This is because it depends on
whether the files are currently present in the inode cache.

Fixing this by consistently providing each process its own view of the
filesystem depending on whether it has the key or not isn't feasible due
to how the VFS caches work.  Furthermore, while sometimes users expect
this behavior, it is misguided for two reasons.  First, it would be an
OS-level access control mechanism largely redundant with existing access
control mechanisms such as UNIX file permissions, ACLs, LSMs, etc.
Encryption is actually for protecting the data at rest.

Second, almost all users of fscrypt actually do need the keys to be
global.  The largest users of fscrypt, Android and Chromium OS, achieve
this by having PID 1 create a "session keyring" that is inherited by
every process.  This works, but it isn't scalable because it prevents
session keyrings from being used for any other purpose.

On general-purpose Linux distros, the 'fscrypt' userspace tool [1] can't
similarly abuse the session keyring, so to make 'sudo' work on all
systems it has to link all the user keyrings into root's user keyring
[2].  This is ugly and raises security concerns.  Moreover it can't make
the keys available to system services, such as sshd trying to access the
user's '~/.ssh' directory (see [3], [4]) or NetworkManager trying to
read certificates from the user's home directory (see [5]); or to Docker
containers (see [6], [7]).

By having an API to add a key to the *filesystem* we'll be able to fix
the above bugs, remove userspace workarounds, and clearly express the
intended semantics: the locked/unlocked status of an encrypted directory
is global, and encryption is orthogonal to OS-level access control.

Why not use the add_key() syscall
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

We use an ioctl for this API rather than the existing add_key() system
call because the ioctl gives us the flexibility needed to implement
fscrypt-specific semantics that will be introduced in later patches:

- Supporting key removal with the semantics such that the secret is
  removed immediately and any unused inodes using the key are evicted;
  also, the eviction of any in-use inodes can be retried.

- Calculating a key-dependent cryptographic identifier and returning it
  to userspace.

- Allowing keys to be added and removed by non-root users, but only keys
  for v2 encryption policies; and to prevent denial-of-service attacks,
  users can only remove keys they themselves have added, and a key is
  only really removed after all users who added it have removed it.

Trying to shoehorn these semantics into the keyrings syscalls would be
very difficult, whereas the ioctls make things much easier.

However, to reuse code the implementation still uses the keyrings
service internally.  Thus we get lockless RCU-mode key lookups without
having to re-implement it, and the keys automatically show up in
/proc/keys for debugging purposes.

References:

    [1] https://github.com/google/fscrypt
    [2] https://goo.gl/55cCrI#heading=h.vf09isp98isb
    [3] google/fscrypt#111 (comment)
    [4] google/fscrypt#116
    [5] https://bugs.launchpad.net/ubuntu/+source/fscrypt/+bug/1770715
    [6] google/fscrypt#128
    [7] https://askubuntu.com/questions/1130306/cannot-run-docker-on-an-encrypted-filesystem

Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Eric Biggers <ebiggers@google.com>
@Redsandro

This comment has been minimized.

Copy link

@Redsandro Redsandro commented Aug 29, 2019

@RomanValov commented:

Having to say if ssh server is restarted after home directory has been unlocked -- public key authentication works.

This is actually not true; it will only accidentally read the files if they are still in the file system cache. I have experienced the same issue, including experiencing the ssh connection works as long as the user is already logged in. However, 24 hours later, I can no longer login. Running sshd in debug mode reveals why:

Aug 29 18:50:04 red sshd[20743]: debug1: trying public key file /home/redsandro/.ssh/authorized_keys
Aug 29 18:50:04 red sshd[20743]: debug1: Could not open authorized keys '/home/redsandro/.ssh/authorized_keys': Required key not available

Could not open authorized keys '/home/sander/.ssh/authorized_keys': Required key not available

This is the same issue as #128 and I assume the kernel patches by @ebiggers will address this. But perhaps @ebiggers can comment on that.

Old ecryptfs logged in encrypted home ssh works because the keys are available permanently, not per user per read request (or something). The trick was just to stay logged in (e.g. a screen session).

For fscrypt until upstream patches are accepted, we need a workaround.

@ebiggers

This comment has been minimized.

Copy link
Collaborator

@ebiggers ebiggers commented Sep 2, 2019

Yes, this (along with other issues where system services and containers can't access encrypted files) will be fixed by putting the keys in a filesystem-level keyring instead of a user keyring, the kernel side of which is planned to be merged for Linux v5.4. After that, there will still be some significant changes to the fscrypt tool needed, which I'm working on. For existing encrypted directories I'm planning to make it so that you can set an option in /etc/fscrypt.conf to use the new keyring, if you don't mind that fscrypt unlock will start requiring root. For new encrypted directories, there will be a new encryption policy version that will always use the new keyring and won't require root.

@Redsandro

This comment has been minimized.

Copy link

@Redsandro Redsandro commented Sep 3, 2019

@ebiggers thank you for the update. Do I understand correctly that (after these updates) we can simply update our encrypted home, even if it's on the same device (that we previously used fscrypt setup on)?

E.g.:

mv /home/user /home/user-bup
mkdir /home/user
fscrypt encrypt /home/user
rsync -vaH /home/user-bup/ /home/user/
rm -rf /home/user-bup

Or will the renaming cause problems? (This might be a stupid question but please bear in mind that I don't understand what's going on inside /home/.fscrypt/.)

PS - I hope this will find it's way back into Ubuntu's Ubiquity installer. Looking to volunteers to choose "this affects me too" before someone closes the issue without addressing it based on a technicality.

@ebiggers

This comment has been minimized.

Copy link
Collaborator

@ebiggers ebiggers commented Sep 3, 2019

Yes, assuming all the kernel and userspace patches go in as planned, the best way to fix the problem will be to set "policy_version": "2" in the options section of /etc/fscrypt.conf, then re-create your encrypted home directory as you showed. (But please don't actually do this yet unless you're really adventurous; I'm still actively working on the userspace patches!)

Eventually we may make fscrypt encrypt create a v2 policy automatically if the kernel supports it, without needing to opt-in. But I think we'll make it opt-in initially.

I'm also planning to make it possible to opt-in to the filesystem keyring for v1 encrypted directories by setting "use_fs_keyring_for_v1_policies": true in /etc/fscrypt.conf. But that will have some disadvantages, like making fscrypt unlock start requiring root on those directories.

@zabbal

This comment has been minimized.

Copy link

@zabbal zabbal commented Sep 3, 2019

Could you clarify what "requiring root" means in this context? Like an actual logged-in root user or just a "sudo fscrypt..." or particular set of capabilities?

Just out of curiosity though, I do realize that it's on-going development and things are still in flux.

@ebiggers

This comment has been minimized.

Copy link
Collaborator

@ebiggers ebiggers commented Sep 3, 2019

I mean you'd need sudo fscrypt unlock to unlock the directories --- or have the PAM module unlock them using your login passphrase. Regular users will still be able to use the directories as long as they were unlocked by sudo (or by something else running as root) first.

That's only for the existing "v1" encrypted directories. Directories using the new v2 encryption policies will be unlockable as regular users, and they always use the filesystem-level keyring so they always avoid the key visibility issue with sshd/docker/NetworkManager/etc.

jaegeuk pushed a commit to jaegeuk/f2fs-stable that referenced this issue Sep 23, 2019
Add a new fscrypt ioctl, FS_IOC_ADD_ENCRYPTION_KEY.  This ioctl adds an
encryption key to the filesystem's fscrypt keyring ->s_master_keys,
making any files encrypted with that key appear "unlocked".

Why we need this
~~~~~~~~~~~~~~~~

The main problem is that the "locked/unlocked" (ciphertext/plaintext)
status of encrypted files is global, but the fscrypt keys are not.
fscrypt only looks for keys in the keyring(s) the process accessing the
filesystem is subscribed to: the thread keyring, process keyring, and
session keyring, where the session keyring may contain the user keyring.

Therefore, userspace has to put fscrypt keys in the keyrings for
individual users or sessions.  But this means that when a process with a
different keyring tries to access encrypted files, whether they appear
"unlocked" or not is nondeterministic.  This is because it depends on
whether the files are currently present in the inode cache.

Fixing this by consistently providing each process its own view of the
filesystem depending on whether it has the key or not isn't feasible due
to how the VFS caches work.  Furthermore, while sometimes users expect
this behavior, it is misguided for two reasons.  First, it would be an
OS-level access control mechanism largely redundant with existing access
control mechanisms such as UNIX file permissions, ACLs, LSMs, etc.
Encryption is actually for protecting the data at rest.

Second, almost all users of fscrypt actually do need the keys to be
global.  The largest users of fscrypt, Android and Chromium OS, achieve
this by having PID 1 create a "session keyring" that is inherited by
every process.  This works, but it isn't scalable because it prevents
session keyrings from being used for any other purpose.

On general-purpose Linux distros, the 'fscrypt' userspace tool [1] can't
similarly abuse the session keyring, so to make 'sudo' work on all
systems it has to link all the user keyrings into root's user keyring
[2].  This is ugly and raises security concerns.  Moreover it can't make
the keys available to system services, such as sshd trying to access the
user's '~/.ssh' directory (see [3], [4]) or NetworkManager trying to
read certificates from the user's home directory (see [5]); or to Docker
containers (see [6], [7]).

By having an API to add a key to the *filesystem* we'll be able to fix
the above bugs, remove userspace workarounds, and clearly express the
intended semantics: the locked/unlocked status of an encrypted directory
is global, and encryption is orthogonal to OS-level access control.

Why not use the add_key() syscall
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

We use an ioctl for this API rather than the existing add_key() system
call because the ioctl gives us the flexibility needed to implement
fscrypt-specific semantics that will be introduced in later patches:

- Supporting key removal with the semantics such that the secret is
  removed immediately and any unused inodes using the key are evicted;
  also, the eviction of any in-use inodes can be retried.

- Calculating a key-dependent cryptographic identifier and returning it
  to userspace.

- Allowing keys to be added and removed by non-root users, but only keys
  for v2 encryption policies; and to prevent denial-of-service attacks,
  users can only remove keys they themselves have added, and a key is
  only really removed after all users who added it have removed it.

Trying to shoehorn these semantics into the keyrings syscalls would be
very difficult, whereas the ioctls make things much easier.

However, to reuse code the implementation still uses the keyrings
service internally.  Thus we get lockless RCU-mode key lookups without
having to re-implement it, and the keys automatically show up in
/proc/keys for debugging purposes.

References:

    [1] https://github.com/google/fscrypt
    [2] https://goo.gl/55cCrI#heading=h.vf09isp98isb
    [3] google/fscrypt#111 (comment)
    [4] google/fscrypt#116
    [5] https://bugs.launchpad.net/ubuntu/+source/fscrypt/+bug/1770715
    [6] google/fscrypt#128
    [7] https://askubuntu.com/questions/1130306/cannot-run-docker-on-an-encrypted-filesystem

Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Eric Biggers <ebiggers@google.com>
jaegeuk pushed a commit to jaegeuk/f2fs-stable that referenced this issue Sep 23, 2019
Add a new fscrypt ioctl, FS_IOC_ADD_ENCRYPTION_KEY.  This ioctl adds an
encryption key to the filesystem's fscrypt keyring ->s_master_keys,
making any files encrypted with that key appear "unlocked".

Why we need this
~~~~~~~~~~~~~~~~

The main problem is that the "locked/unlocked" (ciphertext/plaintext)
status of encrypted files is global, but the fscrypt keys are not.
fscrypt only looks for keys in the keyring(s) the process accessing the
filesystem is subscribed to: the thread keyring, process keyring, and
session keyring, where the session keyring may contain the user keyring.

Therefore, userspace has to put fscrypt keys in the keyrings for
individual users or sessions.  But this means that when a process with a
different keyring tries to access encrypted files, whether they appear
"unlocked" or not is nondeterministic.  This is because it depends on
whether the files are currently present in the inode cache.

Fixing this by consistently providing each process its own view of the
filesystem depending on whether it has the key or not isn't feasible due
to how the VFS caches work.  Furthermore, while sometimes users expect
this behavior, it is misguided for two reasons.  First, it would be an
OS-level access control mechanism largely redundant with existing access
control mechanisms such as UNIX file permissions, ACLs, LSMs, etc.
Encryption is actually for protecting the data at rest.

Second, almost all users of fscrypt actually do need the keys to be
global.  The largest users of fscrypt, Android and Chromium OS, achieve
this by having PID 1 create a "session keyring" that is inherited by
every process.  This works, but it isn't scalable because it prevents
session keyrings from being used for any other purpose.

On general-purpose Linux distros, the 'fscrypt' userspace tool [1] can't
similarly abuse the session keyring, so to make 'sudo' work on all
systems it has to link all the user keyrings into root's user keyring
[2].  This is ugly and raises security concerns.  Moreover it can't make
the keys available to system services, such as sshd trying to access the
user's '~/.ssh' directory (see [3], [4]) or NetworkManager trying to
read certificates from the user's home directory (see [5]); or to Docker
containers (see [6], [7]).

By having an API to add a key to the *filesystem* we'll be able to fix
the above bugs, remove userspace workarounds, and clearly express the
intended semantics: the locked/unlocked status of an encrypted directory
is global, and encryption is orthogonal to OS-level access control.

Why not use the add_key() syscall
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

We use an ioctl for this API rather than the existing add_key() system
call because the ioctl gives us the flexibility needed to implement
fscrypt-specific semantics that will be introduced in later patches:

- Supporting key removal with the semantics such that the secret is
  removed immediately and any unused inodes using the key are evicted;
  also, the eviction of any in-use inodes can be retried.

- Calculating a key-dependent cryptographic identifier and returning it
  to userspace.

- Allowing keys to be added and removed by non-root users, but only keys
  for v2 encryption policies; and to prevent denial-of-service attacks,
  users can only remove keys they themselves have added, and a key is
  only really removed after all users who added it have removed it.

Trying to shoehorn these semantics into the keyrings syscalls would be
very difficult, whereas the ioctls make things much easier.

However, to reuse code the implementation still uses the keyrings
service internally.  Thus we get lockless RCU-mode key lookups without
having to re-implement it, and the keys automatically show up in
/proc/keys for debugging purposes.

References:

    [1] https://github.com/google/fscrypt
    [2] https://goo.gl/55cCrI#heading=h.vf09isp98isb
    [3] google/fscrypt#111 (comment)
    [4] google/fscrypt#116
    [5] https://bugs.launchpad.net/ubuntu/+source/fscrypt/+bug/1770715
    [6] google/fscrypt#128
    [7] https://askubuntu.com/questions/1130306/cannot-run-docker-on-an-encrypted-filesystem

Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Eric Biggers <ebiggers@google.com>
jaegeuk pushed a commit to jaegeuk/f2fs-stable that referenced this issue Sep 23, 2019
Add a new fscrypt ioctl, FS_IOC_ADD_ENCRYPTION_KEY.  This ioctl adds an
encryption key to the filesystem's fscrypt keyring ->s_master_keys,
making any files encrypted with that key appear "unlocked".

Why we need this
~~~~~~~~~~~~~~~~

The main problem is that the "locked/unlocked" (ciphertext/plaintext)
status of encrypted files is global, but the fscrypt keys are not.
fscrypt only looks for keys in the keyring(s) the process accessing the
filesystem is subscribed to: the thread keyring, process keyring, and
session keyring, where the session keyring may contain the user keyring.

Therefore, userspace has to put fscrypt keys in the keyrings for
individual users or sessions.  But this means that when a process with a
different keyring tries to access encrypted files, whether they appear
"unlocked" or not is nondeterministic.  This is because it depends on
whether the files are currently present in the inode cache.

Fixing this by consistently providing each process its own view of the
filesystem depending on whether it has the key or not isn't feasible due
to how the VFS caches work.  Furthermore, while sometimes users expect
this behavior, it is misguided for two reasons.  First, it would be an
OS-level access control mechanism largely redundant with existing access
control mechanisms such as UNIX file permissions, ACLs, LSMs, etc.
Encryption is actually for protecting the data at rest.

Second, almost all users of fscrypt actually do need the keys to be
global.  The largest users of fscrypt, Android and Chromium OS, achieve
this by having PID 1 create a "session keyring" that is inherited by
every process.  This works, but it isn't scalable because it prevents
session keyrings from being used for any other purpose.

On general-purpose Linux distros, the 'fscrypt' userspace tool [1] can't
similarly abuse the session keyring, so to make 'sudo' work on all
systems it has to link all the user keyrings into root's user keyring
[2].  This is ugly and raises security concerns.  Moreover it can't make
the keys available to system services, such as sshd trying to access the
user's '~/.ssh' directory (see [3], [4]) or NetworkManager trying to
read certificates from the user's home directory (see [5]); or to Docker
containers (see [6], [7]).

By having an API to add a key to the *filesystem* we'll be able to fix
the above bugs, remove userspace workarounds, and clearly express the
intended semantics: the locked/unlocked status of an encrypted directory
is global, and encryption is orthogonal to OS-level access control.

Why not use the add_key() syscall
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

We use an ioctl for this API rather than the existing add_key() system
call because the ioctl gives us the flexibility needed to implement
fscrypt-specific semantics that will be introduced in later patches:

- Supporting key removal with the semantics such that the secret is
  removed immediately and any unused inodes using the key are evicted;
  also, the eviction of any in-use inodes can be retried.

- Calculating a key-dependent cryptographic identifier and returning it
  to userspace.

- Allowing keys to be added and removed by non-root users, but only keys
  for v2 encryption policies; and to prevent denial-of-service attacks,
  users can only remove keys they themselves have added, and a key is
  only really removed after all users who added it have removed it.

Trying to shoehorn these semantics into the keyrings syscalls would be
very difficult, whereas the ioctls make things much easier.

However, to reuse code the implementation still uses the keyrings
service internally.  Thus we get lockless RCU-mode key lookups without
having to re-implement it, and the keys automatically show up in
/proc/keys for debugging purposes.

References:

    [1] https://github.com/google/fscrypt
    [2] https://goo.gl/55cCrI#heading=h.vf09isp98isb
    [3] google/fscrypt#111 (comment)
    [4] google/fscrypt#116
    [5] https://bugs.launchpad.net/ubuntu/+source/fscrypt/+bug/1770715
    [6] google/fscrypt#128
    [7] https://askubuntu.com/questions/1130306/cannot-run-docker-on-an-encrypted-filesystem

Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Eric Biggers <ebiggers@google.com>
jaegeuk pushed a commit to jaegeuk/f2fs-stable that referenced this issue Sep 24, 2019
Add a new fscrypt ioctl, FS_IOC_ADD_ENCRYPTION_KEY.  This ioctl adds an
encryption key to the filesystem's fscrypt keyring ->s_master_keys,
making any files encrypted with that key appear "unlocked".

Why we need this
~~~~~~~~~~~~~~~~

The main problem is that the "locked/unlocked" (ciphertext/plaintext)
status of encrypted files is global, but the fscrypt keys are not.
fscrypt only looks for keys in the keyring(s) the process accessing the
filesystem is subscribed to: the thread keyring, process keyring, and
session keyring, where the session keyring may contain the user keyring.

Therefore, userspace has to put fscrypt keys in the keyrings for
individual users or sessions.  But this means that when a process with a
different keyring tries to access encrypted files, whether they appear
"unlocked" or not is nondeterministic.  This is because it depends on
whether the files are currently present in the inode cache.

Fixing this by consistently providing each process its own view of the
filesystem depending on whether it has the key or not isn't feasible due
to how the VFS caches work.  Furthermore, while sometimes users expect
this behavior, it is misguided for two reasons.  First, it would be an
OS-level access control mechanism largely redundant with existing access
control mechanisms such as UNIX file permissions, ACLs, LSMs, etc.
Encryption is actually for protecting the data at rest.

Second, almost all users of fscrypt actually do need the keys to be
global.  The largest users of fscrypt, Android and Chromium OS, achieve
this by having PID 1 create a "session keyring" that is inherited by
every process.  This works, but it isn't scalable because it prevents
session keyrings from being used for any other purpose.

On general-purpose Linux distros, the 'fscrypt' userspace tool [1] can't
similarly abuse the session keyring, so to make 'sudo' work on all
systems it has to link all the user keyrings into root's user keyring
[2].  This is ugly and raises security concerns.  Moreover it can't make
the keys available to system services, such as sshd trying to access the
user's '~/.ssh' directory (see [3], [4]) or NetworkManager trying to
read certificates from the user's home directory (see [5]); or to Docker
containers (see [6], [7]).

By having an API to add a key to the *filesystem* we'll be able to fix
the above bugs, remove userspace workarounds, and clearly express the
intended semantics: the locked/unlocked status of an encrypted directory
is global, and encryption is orthogonal to OS-level access control.

Why not use the add_key() syscall
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

We use an ioctl for this API rather than the existing add_key() system
call because the ioctl gives us the flexibility needed to implement
fscrypt-specific semantics that will be introduced in later patches:

- Supporting key removal with the semantics such that the secret is
  removed immediately and any unused inodes using the key are evicted;
  also, the eviction of any in-use inodes can be retried.

- Calculating a key-dependent cryptographic identifier and returning it
  to userspace.

- Allowing keys to be added and removed by non-root users, but only keys
  for v2 encryption policies; and to prevent denial-of-service attacks,
  users can only remove keys they themselves have added, and a key is
  only really removed after all users who added it have removed it.

Trying to shoehorn these semantics into the keyrings syscalls would be
very difficult, whereas the ioctls make things much easier.

However, to reuse code the implementation still uses the keyrings
service internally.  Thus we get lockless RCU-mode key lookups without
having to re-implement it, and the keys automatically show up in
/proc/keys for debugging purposes.

References:

    [1] https://github.com/google/fscrypt
    [2] https://goo.gl/55cCrI#heading=h.vf09isp98isb
    [3] google/fscrypt#111 (comment)
    [4] google/fscrypt#116
    [5] https://bugs.launchpad.net/ubuntu/+source/fscrypt/+bug/1770715
    [6] google/fscrypt#128
    [7] https://askubuntu.com/questions/1130306/cannot-run-docker-on-an-encrypted-filesystem

Reviewed-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Eric Biggers <ebiggers@google.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
5 participants
You can’t perform that action at this time.