This is a changeset adding encryption to btrfs.
Last October, Omar Sandoval sent out a design document for having fscrypt
integration with btrfs [1]. In summary, it proposes btrfs storing its
own encryption IVs on a per-file-extent basis. fscrypt usually encrypts
files using an IV derived from per-inode information; this would prevent
snapshotting or reflinking or data relocation for btrfs. We have
refined this into a fscrypt extent context object, opaque to the
filesystem, which fscrypt uses to generate an IV associated with each
block in an extent. Thus, all the inodes sharing a particular
key and file extent may decrypt the extent.
This series implements this integration for the simple case,
non-compressed data extents, and for verity items. Followup changes will
allow encryption of compressed extents, inline extents, and will add
tests around subvolume encryption. This series should provide encryption
for the simplest cases, but this series should not be used in
production, as there are likely bugs.
This set hopefully reflects all of the feedback given on v3.
I apologize if I missed any past feedback.
[1]
https://lore.kernel.org/linux-btrfs/YXGyq+buM79A1S0L@relinquished.localdomain/
Changelog:
v4:
- Dropped the partial directory encryption trio of patches, as it's an
easy and orthogonal followon.
- fscrypt: Changed extent-based encryption to require a direct key policy.
- fscrypt: Changed to allow direct key policies with mixed
filename/contents encryption modes, to enable usage of existing AES
modes with btrfs.
- fscrypt: Changed terminology used for extent context contents to 'nonce'
instead of 'IV', to match the nonces used in direct-key policies
elsewhere.
- fscrypt: Changed IV generation for extent-based encryption to generate
16-byte IVs (if needed) from a 16-byte nonce plus extent offset, hopefully
as discussed.
- fscrypt: Updated documentation change to remove partial directory
encryption.
- fscrypt: Fixed another place to refer to 'inode-based' encryption
instead of a more generic term.
- btrfs: Factored btrfs_fscrypt_set_extent_context() as per Josef's
feedback.
- btrfs: Fixed a couple of style nits.
- https://lore.kernel.org/linux-btrfs/cover.1666651724.git.sweettea-kernel@xxxxxxxxxx
v3:
- fscrypt: changed to generate extent contexts the same way as IVs are
generated for inode-based encryption, allowing use of any existing
policy and making the difference in encryption more minimal.
- Changed to use qstr's, then fscrypt_strs, then fscrypt_names only
where absolutely necessary, rather than fscrypt_names everywhere.
- Reordered changes to put partially-encrypted directories, and no-key
name handling, in their own sections at the end of the patchset.
- Expanded on descriptions of how no-key name handling works.
- Added encryption of verity items (but see notes in change about
outstanding questions there)
- Stylistic fixes
- Renamed flags from FSCRYPT to ENCRYPT in multiple instances.
- Made the incompat encrypt superblock flag be set the first time a
directory is set to be encrypted, so there isn't a need for a mkfs
option.
- Hopefully addressed all other minor feedback points.
- https://lore.kernel.org/linux-btrfs/cover.1666281276.git.sweettea-kernel@xxxxxxxxxx
v2:
- Amended the fscrypt side to generically add extent contexts,
hopefully as per Eric Biggers' past comments. IVs are now entirely
abstracted within an extent context, and there is no longer a new
encryption policy, as DIRECT_KEY sufficiently encapsulates the
needs of extent-based encryption. Documented its usage in btrfs
briefly in the documentation.
- Adjusted the btrfs side to deal in opaque extent contexts. Improved
optimization to skip storing inode contexts if they are the same as
the inode's root item's inode context.
- Combined 'add fscrypt operation table to superblock' into 'start
using fscrypt hooks'.
- https://lore.kernel.org/linux-btrfs/cover.1662420176.git.sweettea-kernel@xxxxxxxxxx
- progs: https://lore.kernel.org/linux-btrfs/cover.1662417859.git.sweettea-kernel@xxxxxxxxxx
- tests: https://lore.kernel.org/linux-btrfs/cover.1662417905.git.sweettea-kernel@xxxxxxxxxx
v1:
- Recombined the fscrypt changes back into this patchset.
- Fixed several races and incorrectly ordered operations.
- Improved IV retrieval to correctly distinguish between
filename/symlink encryption and encryption of block 0 of a file.
- https://lore.kernel.org/linux-btrfs/cover.1660744500.git.sweettea-kernel@xxxxxxxxxx
- progs: https://lore.kernel.org/linux-btrfs/cover.1660729916.git.sweettea-kernel@xxxxxxxxxx
- tests: https://lore.kernel.org/linux-btrfs/cover.1660729861.git.sweettea-kernel@xxxxxxxxxx
RFC v2:
- Fixed all warnings and known incorrectnesses.
- Split fscrypt changes into their own patchset:
https://lore.kernel.org/linux-fscrypt/cover.1658623235.git.sweettea-kernel@xxxxxxxxxx
- Combined and reordered changes so that enabling fscrypt is the last change.
- Removed unnecessary factoring.
- Split a cleanup change off.
- https://lore.kernel.org/linux-btrfs/cover.1658623319.git.sweettea-kernel@xxxxxxxxxx
RFC v1:
- https://lore.kernel.org/linux-btrfs/cover.1657707686.git.sweettea-kernel@xxxxxxxxxx
Omar Sandoval (11):
fscrypt: expose fscrypt_nokey_name
fscrypt: add fscrypt_have_same_policy() to check inode compatibility
btrfs: store directory encryption state
btrfs: disable various operations on encrypted inodes
btrfs: start using fscrypt hooks
btrfs: add fscrypt_context items
btrfs: translate btrfs encryption flags and encrypted inode flag
btrfs: encrypt normal file extent data if appropriate
btrfs: Add new FEATURE_INCOMPAT_ENCRYPT feature flag.
btrfs: implement fscrypt ioctls
btrfs: permit searching for nokey names for removal
Sweet Tea Dorminy (10):
fscrypt: allow fscrypt_generate_iv() to distinguish filenames
fscrypt: add extent-based encryption
fscrypt: direct key policies for extent-based encryption
fscrypt: document btrfs' fscrypt quirks.
btrfs: use struct qstr instead of name and namelen
btrfs: setup qstrings from dentrys using fscrypt helper
btrfs: use struct fscrypt_str instead of struct qstr
btrfs: store a fscrypt extent context per normal file extent
btrfs: use correct name hash for nokey names
btrfs: encrypt verity items
Documentation/filesystems/fscrypt.rst | 31 +-
fs/btrfs/Makefile | 1 +
fs/btrfs/accessors.h | 44 +-
fs/btrfs/btrfs_inode.h | 3 +
fs/btrfs/ctree.h | 46 ++-
fs/btrfs/delayed-inode.c | 36 +-
fs/btrfs/delayed-inode.h | 6 +-
fs/btrfs/dir-item.c | 170 ++++++--
fs/btrfs/extent_io.c | 94 ++++-
fs/btrfs/extent_io.h | 3 +
fs/btrfs/extent_map.c | 7 +
fs/btrfs/extent_map.h | 4 +
fs/btrfs/file-item.c | 25 +-
fs/btrfs/file.c | 11 +-
fs/btrfs/fs.h | 5 +-
fs/btrfs/fscrypt.c | 287 +++++++++++++
fs/btrfs/fscrypt.h | 63 +++
fs/btrfs/inode-item.c | 73 ++--
fs/btrfs/inode-item.h | 20 +-
fs/btrfs/inode.c | 560 +++++++++++++++++++-------
fs/btrfs/ioctl.c | 55 ++-
fs/btrfs/ordered-data.c | 11 +-
fs/btrfs/ordered-data.h | 4 +-
fs/btrfs/print-tree.c | 4 +-
fs/btrfs/reflink.c | 8 +
fs/btrfs/root-tree.c | 21 +-
fs/btrfs/send.c | 13 +-
fs/btrfs/super.c | 10 +-
fs/btrfs/transaction.c | 40 +-
fs/btrfs/tree-checker.c | 51 ++-
fs/btrfs/tree-log.c | 304 +++++++-------
fs/btrfs/tree-log.h | 4 +-
fs/btrfs/verity.c | 114 +++++-
fs/crypto/crypto.c | 40 +-
fs/crypto/fname.c | 43 +-
fs/crypto/fscrypt_private.h | 23 +-
fs/crypto/inline_crypt.c | 28 +-
fs/crypto/policy.c | 96 +++++
include/linux/fscrypt.h | 91 +++++
include/uapi/linux/btrfs.h | 1 +
include/uapi/linux/btrfs_tree.h | 28 ++
41 files changed, 1935 insertions(+), 543 deletions(-)
create mode 100644 fs/btrfs/fscrypt.c
create mode 100644 fs/btrfs/fscrypt.h
base-commit: cc159678aee59042e24fd1b039405b4fdf7b0538
--
2.35.1
|