Pull STATX_DIOALIGN support from Eric Biggers:
"Make statx() support reporting direct I/O (DIO) alignment information.
This provides a generic interface for userspace programs to determine
whether a file supports DIO, and if so with what alignment
restrictions. Specifically, STATX_DIOALIGN works on block devices, and
on regular files when their containing filesystem has implemented
support.
An interface like this has been requested for years, since the
conditions for when DIO is supported in Linux have gotten increasingly
complex over time. Today, DIO support and alignment requirements can
be affected by various filesystem features such as multi-device
support, data journalling, inline data, encryption, verity,
compression, checkpoint disabling, log-structured mode, etc.
Further complicating things, Linux v6.0 relaxed the traditional rule
of DIO needing to be aligned to the block device's logical block size;
now user buffers (but not file offsets) only need to be aligned to the
DMA alignment.
The approach of uplifting the XFS specific ioctl XFS_IOC_DIOINFO was
discarded in favor of creating a clean new interface with statx().
For more information, see the individual commits and the man page
update[1]"
Link: https://lore.kernel.org/r/20220722074229.148925-1-ebiggers@kernel.org [1]
* tag 'statx-dioalign-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiggers/linux:
xfs: support STATX_DIOALIGN
f2fs: support STATX_DIOALIGN
f2fs: simplify f2fs_force_buffered_io()
f2fs: move f2fs_force_buffered_io() into file.c
ext4: support STATX_DIOALIGN
fscrypt: change fscrypt_dio_supported() to prepare for STATX_DIOALIGN
vfs: support STATX_DIOALIGN on block devices
statx: add direct I/O alignment information
request_queues are a block layer implementation detail that should not
leak into file systems. Change the fscrypt inline crypto code to
retrieve block devices instead of request_queues from the file system.
As part of that, clean up the interaction with multi-device file systems
by returning both the number of devices and the actual device array in a
single method call.
Signed-off-by: Christoph Hellwig <hch@lst.de>
[ebiggers: bug fixes and minor tweaks]
Signed-off-by: Eric Biggers <ebiggers@google.com>
Link: https://lore.kernel.org/r/20220901193208.138056-4-ebiggers@kernel.org
Now that the fscrypt_master_key lifetime has been reworked to not be
subject to the quirks of the keyrings subsystem, blk_crypto_evict_key()
no longer gets called after the filesystem has already been unmounted.
Therefore, there is no longer any need to hold extra references to the
filesystem's request_queue(s). (And these references didn't always do
their intended job anyway, as pinning a request_queue doesn't
necessarily pin the corresponding blk_crypto_profile.)
Stop taking these extra references. Instead, just pass the super_block
to fscrypt_destroy_inline_crypt_key(), and use it to get the list of
block devices the key needs to be evicted from.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Link: https://lore.kernel.org/r/20220901193208.138056-3-ebiggers@kernel.org
The approach of fs/crypto/ internally managing the fscrypt_master_key
structs as the payloads of "struct key" objects contained in a
"struct key" keyring has outlived its usefulness. The original idea was
to simplify the code by reusing code from the keyrings subsystem.
However, several issues have arisen that can't easily be resolved:
- When a master key struct is destroyed, blk_crypto_evict_key() must be
called on any per-mode keys embedded in it. (This started being the
case when inline encryption support was added.) Yet, the keyrings
subsystem can arbitrarily delay the destruction of keys, even past the
time the filesystem was unmounted. Therefore, currently there is no
easy way to call blk_crypto_evict_key() when a master key is
destroyed. Currently, this is worked around by holding an extra
reference to the filesystem's request_queue(s). But it was overlooked
that the request_queue reference is *not* guaranteed to pin the
corresponding blk_crypto_profile too; for device-mapper devices that
support inline crypto, it doesn't. This can cause a use-after-free.
- When the last inode that was using an incompletely-removed master key
is evicted, the master key removal is completed by removing the key
struct from the keyring. Currently this is done via key_invalidate().
Yet, key_invalidate() takes the key semaphore. This can deadlock when
called from the shrinker, since in fscrypt_ioctl_add_key(), memory is
allocated with GFP_KERNEL under the same semaphore.
- More generally, the fact that the keyrings subsystem can arbitrarily
delay the destruction of keys (via garbage collection delay, or via
random processes getting temporary key references) is undesirable, as
it means we can't strictly guarantee that all secrets are ever wiped.
- Doing the master key lookups via the keyrings subsystem results in the
key_permission LSM hook being called. fscrypt doesn't want this, as
all access control for encrypted files is designed to happen via the
files themselves, like any other files. The workaround which SELinux
users are using is to change their SELinux policy to grant key search
access to all domains. This works, but it is an odd extra step that
shouldn't really have to be done.
The fix for all these issues is to change the implementation to what I
should have done originally: don't use the keyrings subsystem to keep
track of the filesystem's fscrypt_master_key structs. Instead, just
store them in a regular kernel data structure, and rework the reference
counting, locking, and lifetime accordingly. Retain support for
RCU-mode key lookups by using a hash table. Replace fscrypt_sb_free()
with fscrypt_sb_delete(), which releases the keys synchronously and runs
a bit earlier during unmount, so that block devices are still available.
A side effect of this patch is that neither the master keys themselves
nor the filesystem keyrings will be listed in /proc/keys anymore.
("Master key users" and the master key users keyrings will still be
listed.) However, this was mostly an implementation detail, and it was
intended just for debugging purposes. I don't know of anyone using it.
This patch does *not* change how "master key users" (->mk_users) works;
that still uses the keyrings subsystem. That is still needed for key
quotas, and changing that isn't necessary to solve the issues listed
above. If we decide to change that too, it would be a separate patch.
I've marked this as fixing the original commit that added the fscrypt
keyring, but as noted above the most important issue that this patch
fixes wasn't introduced until the addition of inline encryption support.
Fixes: 22d94f493b ("fscrypt: add FS_IOC_ADD_ENCRYPTION_KEY ioctl")
Signed-off-by: Eric Biggers <ebiggers@google.com>
Link: https://lore.kernel.org/r/20220901193208.138056-2-ebiggers@kernel.org
To prepare for STATX_DIOALIGN support, make two changes to
fscrypt_dio_supported().
First, remove the filesystem-block-alignment check and make the
filesystems handle it instead. It previously made sense to have it in
fs/crypto/; however, to support STATX_DIOALIGN the alignment restriction
would have to be returned to filesystems. It ends up being simpler if
filesystems handle this part themselves, especially for f2fs which only
allows fs-block-aligned DIO in the first place.
Second, make fscrypt_dio_supported() work on inodes whose encryption key
hasn't been set up yet, by making it set up the key if needed. This is
required for statx(), since statx() doesn't require a file descriptor.
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Eric Biggers <ebiggers@google.com>
Link: https://lore.kernel.org/r/20220827065851.135710-4-ebiggers@kernel.org
As a step towards freeing the PG_error flag for other uses, change ext4
and f2fs to stop using PG_error to track decryption errors. Instead, if
a decryption error occurs, just mark the whole bio as failed. The
coarser granularity isn't really a problem since it isn't any worse than
what the block layer provides, and errors from a multi-page readahead
aren't reported to applications unless a single-page read fails too.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Chao Yu <chao@kernel.org> # for f2fs part
Link: https://lore.kernel.org/r/20220815235052.86545-2-ebiggers@kernel.org
Pull ceph updates from Ilya Dryomov:
"We have a good pile of various fixes and cleanups from Xiubo, Jeff,
Luis and others, almost exclusively in the filesystem.
Several patches touch files outside of our normal purview to set the
stage for bringing in Jeff's long awaited ceph+fscrypt series in the
near future. All of them have appropriate acks and sat in linux-next
for a while"
* tag 'ceph-for-5.20-rc1' of https://github.com/ceph/ceph-client: (27 commits)
libceph: clean up ceph_osdc_start_request prototype
libceph: fix ceph_pagelist_reserve() comment typo
ceph: remove useless check for the folio
ceph: don't truncate file in atomic_open
ceph: make f_bsize always equal to f_frsize
ceph: flush the dirty caps immediatelly when quota is approaching
libceph: print fsid and epoch with osd id
libceph: check pointer before assigned to "c->rules[]"
ceph: don't get the inline data for new creating files
ceph: update the auth cap when the async create req is forwarded
ceph: make change_auth_cap_ses a global symbol
ceph: fix incorrect old_size length in ceph_mds_request_args
ceph: switch back to testing for NULL folio->private in ceph_dirty_folio
ceph: call netfs_subreq_terminated with was_async == false
ceph: convert to generic_file_llseek
ceph: fix the incorrect comment for the ceph_mds_caps struct
ceph: don't leak snap_rwsem in handle_cap_grant
ceph: prevent a client from exceeding the MDS maximum xattr size
ceph: choose auth MDS for getxattr with the Xs caps
ceph: add session already open notify support
...
Most filesystems just call fscrypt_set_context on new inodes, which
usually causes a setxattr. That's a bit late for ceph, which can send
along a full set of attributes with the create request.
Doing so allows it to avoid race windows that where the new inode could
be seen by other clients without the crypto context attached. It also
avoids the separate round trip to the server.
Refactor the fscrypt code a bit to allow us to create a new crypto
context, attach it to the inode, and write it to the buffer, but without
calling set_context on it. ceph can later use this to marshal the
context into the attributes we send along with the create request.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Xiubo Li <xiubli@redhat.com>
Acked-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
For ceph, we want to use our own scheme for handling filenames that are
are longer than NAME_MAX after encryption and Base64 encoding. This
allows us to have a consistent view of the encrypted filenames for
clients that don't support fscrypt and clients that do but that don't
have the key.
Currently, fs/crypto only supports encrypting filenames using
fscrypt_setup_filename, but that also handles encoding nokey names. Ceph
can't use that because it handles nokey names in a different way.
Export fscrypt_fname_encrypt. Rename fscrypt_fname_encrypted_size to
__fscrypt_fname_encrypted_size and add a new wrapper called
fscrypt_fname_encrypted_size that takes an inode argument rather than a
pointer to a fscrypt_policy union.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Xiubo Li <xiubli@redhat.com>
Acked-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
HCTR2 is a tweakable, length-preserving encryption mode that is intended
for use on CPUs with dedicated crypto instructions. HCTR2 has the
property that a bitflip in the plaintext changes the entire ciphertext.
This property fixes a known weakness with filename encryption: when two
filenames in the same directory share a prefix of >= 16 bytes, with
AES-CTS-CBC their encrypted filenames share a common substring, leaking
information. HCTR2 does not have this problem.
More information on HCTR2 can be found here: "Length-preserving
encryption with HCTR2": https://eprint.iacr.org/2021/1441.pdf
Signed-off-by: Nathan Huckleberry <nhuck@google.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Acked-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Unfortunately the design of fscrypt_set_test_dummy_encryption() doesn't
work properly for the new mount API, as it combines too many steps into
one function:
- Parse the argument to test_dummy_encryption
- Check the setting against the filesystem instance
- Apply the setting to the filesystem instance
The new mount API has split these into separate steps. ext4 partially
worked around this by duplicating some of the logic, but it still had
some bugs. To address this, add some new helper functions that split up
the steps of fscrypt_set_test_dummy_encryption():
- fscrypt_parse_test_dummy_encryption()
- fscrypt_dummy_policies_equal()
- fscrypt_add_test_dummy_key()
While we're add it, also add a function fscrypt_is_dummy_policy_set()
which will be useful to avoid some #ifdef's.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Link: https://lore.kernel.org/r/20220501050857.538984-5-ebiggers@kernel.org
When inline encryption is used, the usual message "fscrypt: AES-256-XTS
using implementation <impl>" doesn't appear in the kernel log. Add a
similar message for the blk-crypto case that indicates that inline
encryption was used, and whether blk-crypto-fallback was used or not.
This can be useful for debugging performance problems.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Link: https://lore.kernel.org/r/20220414053415.158986-1-ebiggers@kernel.org
FS_CRYPTO_BLOCK_SIZE is neither the filesystem block size nor the
granularity of encryption. Rather, it defines two logically separate
constraints that both arise from the block size of the AES cipher:
- The alignment required for the lengths of file contents blocks
- The minimum input/output length for the filenames encryption modes
Since there are way too many things called the "block size", and the
connection with the AES block size is not easily understood, split
FS_CRYPTO_BLOCK_SIZE into two constants FSCRYPT_CONTENTS_ALIGNMENT and
FSCRYPT_FNAME_MIN_MSG_LEN that more clearly describe what they are.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Link: https://lore.kernel.org/r/20220405010914.18519-1-ebiggers@kernel.org
All filesystems have now been converted to use ->readahead, so
remove the ->readpages operation and fix all the comments that
used to refer to it.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Pull fscrypt updates from Eric Biggers:
"Add support for direct I/O on encrypted files when blk-crypto (inline
encryption) is being used for file contents encryption"
* tag 'fscrypt-for-linus' of git://git.kernel.org/pub/scm/fs/fscrypt/fscrypt:
fscrypt: update documentation for direct I/O support
f2fs: support direct I/O with fscrypt using blk-crypto
ext4: support direct I/O with fscrypt using blk-crypto
iomap: support direct I/O with fscrypt using blk-crypto
fscrypt: add functions for direct I/O support
Encrypted files traditionally haven't supported DIO, due to the need to
encrypt/decrypt the data. However, when the encryption is implemented
using inline encryption (blk-crypto) instead of the traditional
filesystem-layer encryption, it is straightforward to support DIO.
In preparation for supporting this, add the following functions:
- fscrypt_dio_supported() checks whether a DIO request is supported as
far as encryption is concerned. Encrypted files will only support DIO
when inline encryption is used and the I/O request is properly
aligned; this function checks these preconditions.
- fscrypt_limit_io_blocks() limits the length of a bio to avoid crossing
a place in the file that a bio with an encryption context cannot
cross due to a DUN discontiguity. This function is needed by
filesystems that use the iomap DIO implementation (which operates
directly on logical ranges, so it won't use fscrypt_mergeable_bio())
and that support FSCRYPT_POLICY_FLAG_IV_INO_LBLK_32.
Co-developed-by: Satya Tangirala <satyat@google.com>
Signed-off-by: Satya Tangirala <satyat@google.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20220128233940.79464-2-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
Pass the block_device that we plan to use this bio for and the
operation to bio_reset to optimize the assigment. A NULL block_device
can be passed, both for the passthrough case on a raw request_queue and
to temporarily avoid refactoring some nasty code.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20220124091107.642561-20-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Pass the block_device and operation that we plan to use this bio for to
bio_alloc to optimize the assignment. NULL/0 can be passed, both for the
passthrough case on a raw request_queue and to temporarily avoid
refactoring some nasty code.
Also move the gfp_mask argument after the nr_vecs argument for a much
more logical calling convention matching what most of the kernel does.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20220124091107.642561-18-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
fscrypt currently requires a 512-bit master key when AES-256-XTS is
used, since AES-256-XTS keys are 512-bit and fscrypt requires that the
master key be at least as long any key that will be derived from it.
However, this is overly strict because AES-256-XTS doesn't actually have
a 512-bit security strength, but rather 256-bit. The fact that XTS
takes twice the expected key size is a quirk of the XTS mode. It is
sufficient to use 256 bits of entropy for AES-256-XTS, provided that it
is first properly expanded into a 512-bit key, which HKDF-SHA512 does.
Therefore, relax the check of the master key size to use the security
strength of the derived key rather than the size of the derived key
(except for v1 encryption policies, which don't use HKDF).
Besides making things more flexible for userspace, this is needed in
order for the use of a KDF which only takes a 256-bit key to be
introduced into the fscrypt key hierarchy. This will happen with
hardware-wrapped keys support, as all known hardware which supports that
feature uses an SP800-108 KDF using AES-256-CMAC, so the wrapped keys
are wrapped 256-bit AES keys. Moreover, there is interest in fscrypt
supporting the same type of AES-256-CMAC based KDF in software as an
alternative to HKDF-SHA512. There is no security problem with such
features, so fix the key length check to work properly with them.
Reviewed-by: Paul Crowley <paulcrowley@google.com>
Link: https://lore.kernel.org/r/20210921030303.5598-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
fscrypt uses a Base64 encoding to encode no-key filenames (the filenames
that are presented to userspace when a directory is listed without its
encryption key). There are many variants of Base64, but the most common
ones are specified by RFC 4648. fscrypt can't use the regular RFC 4648
"base64" variant because "base64" uses the '/' character, which isn't
allowed in filenames. However, RFC 4648 also specifies a "base64url"
variant for use in URLs and filenames. "base64url" is less common than
"base64", but it's still implemented in many programming libraries.
Unfortunately, what fscrypt actually uses is a custom Base64 variant
that differs from "base64url" in several ways:
- The binary data is divided into 6-bit chunks differently.
- Values 62 and 63 are encoded with '+' and ',' instead of '-' and '_'.
- '='-padding isn't used. This isn't a problem per se, as the padding
isn't technically necessary, and RFC 4648 doesn't strictly require it.
But it needs to be properly documented.
There have been two attempts to copy the fscrypt Base64 code into lib/
(https://lkml.kernel.org/r/20200821182813.52570-6-jlayton@kernel.org and
https://lkml.kernel.org/r/20210716110428.9727-5-hare@suse.de), and both
have been caught up by the fscrypt Base64 variant being nonstandard and
not properly documented. Also, the planned use of the fscrypt Base64
code in the CephFS storage back-end will prevent it from being changed
later (whereas currently it can still be changed), so we need to choose
an encoding that we're happy with before it's too late.
Therefore, switch the fscrypt Base64 variant to base64url, in order to
align more closely with RFC 4648 and other implementations and uses of
Base64. However, I opted not to implement '='-padding, as '='-padding
adds complexity, is unnecessary, and isn't required by the RFC.
Link: https://lore.kernel.org/r/20210718000125.59701-1-ebiggers@kernel.org
Reviewed-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Eric Biggers <ebiggers@google.com>
