fsverity builtin signatures (CONFIG_FS_VERITY_BUILTIN_SIGNATURES) aren't
the only way to do signatures with fsverity, and they have some major
limitations.  Yet, more users have tried to use them, e.g. recently by
https://github.com/ostreedev/ostree/pull/2640.  In most cases this seems
to be because users aren't sufficiently familiar with the limitations of
this feature and what the alternatives are.

Therefore, make some updates to the documentation to try to clarify the
properties of this feature and nudge users in the right direction.

Note that the Integrity Policy Enforcement (IPE) LSM, which is not yet
upstream, is planned to use the builtin signatures.  (This differs from
IMA, which uses its own signature mechanism.)  For that reason, my
earlier patch "fsverity: mark builtin signatures as deprecated"
(https://lore.kernel.org/r/20221208033548.122704-1-ebiggers@kernel.org),
which marked builtin signatures as "deprecated", was controversial.

This patch therefore stops short of marking the feature as deprecated.
I've also revised the language to focus on better explaining the feature
and what its alternatives are.

Link: https://lore.kernel.org/r/20230620041937.5809-1-ebiggers@kernel.org
Reviewed-by: Colin Walters <walters@verbum.org>
Reviewed-by: Luca Boccassi <bluca@debian.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>

Address several issues with the calling convention and documentation of
fsverity_get_digest():

- Make it provide the hash algorithm as either a FS_VERITY_HASH_ALG_*
  value or HASH_ALGO_* value, at the caller's choice, rather than only a
  HASH_ALGO_* value as it did before.  This allows callers to work with
  the fsverity native algorithm numbers if they want to.  HASH_ALGO_* is
  what IMA uses, but other users (e.g. overlayfs) should use
  FS_VERITY_HASH_ALG_* to match fsverity-utils and the fsverity UAPI.

- Make it return the digest size so that it doesn't need to be looked up
  separately.  Use the return value for this, since 0 works nicely for
  the "file doesn't have fsverity enabled" case.  This also makes it
  clear that no other errors are possible.

- Rename the 'digest' parameter to 'raw_digest' and clearly document
  that it is only useful in combination with the algorithm ID.  This
  hopefully clears up a point of confusion.

- Export it to modules, since overlayfs will need it for checking the
  fsverity digests of lowerdata files
  (https://lore.kernel.org/r/dd294a44e8f401e6b5140029d8355f88748cd8fd.1686565330.git.alexl@redhat.com).

Acked-by: Mimi Zohar <zohar@linux.ibm.com> # for the IMA piece
Link: https://lore.kernel.org/r/20230612190047.59755-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>

Clean up the error handling in verify_data_block() to (a) eliminate the
'err' variable which has caused some confusion because the function
actually returns a bool, (b) reduce the compiled code size slightly, and
(c) execute one fewer branch in the success case.

Link: https://lore.kernel.org/r/20230604022312.48532-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>

bio_first_page_all(bio)->mapping->host is not compatible with large
folios, since the first page of the bio is not necessarily the head page
of the folio, and therefore it might not have the mapping pointer set.

Therefore, move the dereference of ->mapping->host into
verify_data_blocks(), which works with a folio.

(Like the commit that this Fixes, this hasn't actually been tested with
large folios yet, since the filesystems that use fs/verity/ don't
support that yet.  But based on code review, I think this is needed.)

Fixes: 5d0f0e57ed90 ("fsverity: support verifying data from large folios")
Link: https://lore.kernel.org/r/20230604022101.48342-1-ebiggers@kernel.org
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>

Now that fsverity_hash_alg doesn't have an embedded mempool, it can be
'const' almost everywhere.  Add it.

Link: https://lore.kernel.org/r/20230604022348.48658-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>

The "ahash" API, like the other scatterlist-based crypto APIs such as
"skcipher", comes with some well-known limitations.  First, it can't
easily be used with vmalloc addresses.  Second, the request struct can't
be allocated on the stack.  This adds complexity and a possible failure
point that needs to be worked around, e.g. using a mempool.

The only benefit of ahash over "shash" is that ahash is needed to access
traditional memory-to-memory crypto accelerators, i.e. drivers/crypto/.
However, this style of crypto acceleration has largely fallen out of
favor and been superseded by CPU-based acceleration or inline crypto
engines.  Also, ahash needs to be used asynchronously to take full
advantage of such hardware, but fs/verity/ has never done this.

On all systems that aren't actually using one of these ahash-only crypto
accelerators, ahash just adds unnecessary overhead as it sits between
the user and the underlying shash algorithms.

Also, XFS is planned to cache fsverity Merkle tree blocks in the
existing XFS buffer cache.  As a result, it will be possible for a
single Merkle tree block to be split across discontiguous pages
(https://lore.kernel.org/r/20230405233753.GU3223426@dread.disaster.area).
This data will need to be hashed.  It is easiest to work with a vmapped
address in this case.  However, ahash is incompatible with this.

Therefore, let's convert fs/verity/ from ahash to shash.  This
simplifies the code, and it should also slightly improve performance for
everyone who wasn't actually using one of these ahash-only crypto
accelerators, i.e. almost everyone (or maybe even everyone)!

Link: https://lore.kernel.org/r/20230516052306.99600-1-ebiggers@kernel.org
Reviewed-by: Christoph Hellwig <hch@lst.de>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>

Commit 56124d6c87fd ("fsverity: support enabling with tree block size <
PAGE_SIZE") changed FS_IOC_ENABLE_VERITY to use __kernel_read() to read
the file's data, instead of direct pagecache accesses.

An unintended consequence of this is that the
'WARN_ON_ONCE(!(file->f_mode & FMODE_READ))' in __kernel_read() became
reachable by fuzz tests.  This happens if FS_IOC_ENABLE_VERITY is called
on a fd opened with access mode 3, which means "ioctl access only".

Arguably, FS_IOC_ENABLE_VERITY should work on ioctl-only fds.  But
ioctl-only fds are a weird Linux extension that is rarely used and that
few people even know about.  (The documentation for FS_IOC_ENABLE_VERITY
even specifically says it requires O_RDONLY.)  It's probably not
worthwhile to make the ioctl internally open a new fd just to handle
this case.  Thus, just reject the ioctl on such fds for now.

Fixes: 56124d6c87fd ("fsverity: support enabling with tree block size < PAGE_SIZE")
Reported-by: syzbot+51177e4144d764827c45@syzkaller.appspotmail.com
Link: https://syzkaller.appspot.com/bug?id=2281afcbbfa8fdb92f9887479cc0e4180f1c6b28
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230406215106.235829-1-ebiggers@kernel.org
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>

The new Merkle tree construction algorithm is a bit fragile in that it
may overflow the 'root_hash' array if the tree actually generated does
not match the calculated tree parameters.

This should never happen unless there is a filesystem bug that allows
the file size to change despite deny_write_access(), or a bug in the
Merkle tree logic itself.  Regardless, it's fairly easy to check for
buffer overflow here, so let's do so.

This is a robustness improvement only; this case is not currently known
to be reachable.  I've added a Fixes tag anyway, since I recommend that
this be included in kernels that have the mentioned commit.

Fixes: 56124d6c87fd ("fsverity: support enabling with tree block size < PAGE_SIZE")
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230328041505.110162-1-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>

As per Linus's suggestion
(https://lore.kernel.org/r/CAHk-=whefxRGyNGzCzG6BVeM=5vnvgb-XhSeFJVxJyAxAF8XRA@mail.gmail.com),
use WARN_ON_ONCE instead of WARN_ON.  This barely adds any extra
overhead, and it makes it so that if any of these ever becomes reachable
(they shouldn't, but that's the point), the logs can't be flooded.

Link: https://lore.kernel.org/r/20230406181542.38894-1-ebiggers@kernel.org
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Reviewed-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>

register_sysctl_paths() is only needed if your child (directories) have
entries but this does not so just use register_sysctl() so to do away
with the path specification.

Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>
Link: https://lore.kernel.org/r/20230302202826.776286-10-mcgrof@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>