... and fix the directory locking documentation and proof of correctness.
Holding ->s_vfs_rename_mutex *almost* prevents ->d_parent changes; the
case where we really don't want it is splicing the root of disconnected
tree to somewhere.
In other words, ->s_vfs_rename_mutex is sufficient to stabilize "X is an
ancestor of Y" only if X and Y are already in the same tree. Otherwise
it can go from false to true, and one can construct a deadlock on that.
Make lock_two_directories() report an error in such case and update the
callers of lock_rename()/lock_rename_child() to handle such errors.
And yes, such conditions are not impossible to create ;-/
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Pull vfs fixes from Christian Brauner:
- Avoid calling back into LSMs from vfs_getattr_nosec() calls.
IMA used to query inode properties accessing raw inode fields without
dedicated helpers. That was finally fixed a few releases ago by
forcing IMA to use vfs_getattr_nosec() helpers.
The goal of the vfs_getattr_nosec() helper is to query for attributes
without calling into the LSM layer which would be quite problematic
because incredibly IMA is called from __fput()...
__fput()
-> ima_file_free()
What it does is to call back into the filesystem to update the file's
IMA xattr. Querying the inode without using vfs_getattr_nosec() meant
that IMA didn't handle stacking filesystems such as overlayfs
correctly. So the switch to vfs_getattr_nosec() is quite correct. But
the switch to vfs_getattr_nosec() revealed another bug when used on
stacking filesystems:
__fput()
-> ima_file_free()
-> vfs_getattr_nosec()
-> i_op->getattr::ovl_getattr()
-> vfs_getattr()
-> i_op->getattr::$WHATEVER_UNDERLYING_FS_getattr()
-> security_inode_getattr() # calls back into LSMs
Now, if that __fput() happens from task_work_run() of an exiting task
current->fs and various other pointer could already be NULL. So
anything in the LSM layer relying on that not being NULL would be
quite surprised.
Fix that by passing the information that this is a security request
through to the stacking filesystem by adding a new internal
ATT_GETATTR_NOSEC flag. Now the callchain becomes:
__fput()
-> ima_file_free()
-> vfs_getattr_nosec()
-> i_op->getattr::ovl_getattr()
-> if (AT_GETATTR_NOSEC)
vfs_getattr_nosec()
else
vfs_getattr()
-> i_op->getattr::$WHATEVER_UNDERLYING_FS_getattr()
- Fix a bug introduced with the iov_iter rework from last cycle.
This broke /proc/kcore by copying too much and without the correct
offset.
- Add a missing NULL check when allocating the root inode in
autofs_fill_super().
- Fix stable writes for multi-device filesystems (xfs, btrfs etc) and
the block device pseudo filesystem.
Stable writes used to be a superblock flag only, making it a per
filesystem property. Add an additional AS_STABLE_WRITES mapping flag
to allow for fine-grained control.
- Ensure that offset_iterate_dir() returns 0 after reaching the end of
a directory so it adheres to getdents() convention.
* tag 'vfs-6.7-rc3.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs:
libfs: getdents() should return 0 after reaching EOD
xfs: respect the stable writes flag on the RT device
xfs: clean up FS_XFLAG_REALTIME handling in xfs_ioctl_setattr_xflags
block: update the stable_writes flag in bdev_add
filemap: add a per-mapping stable writes flag
autofs: add: new_inode check in autofs_fill_super()
iov_iter: fix copy_page_to_iter_nofault()
fs: Pass AT_GETATTR_NOSEC flag to getattr interface function
In vfs code, sb_start_write() is usually called after the permission hook
in rw_verify_area(). vfs_iocb_iter_write() is an exception to this rule,
where kiocb_start_write() is called by its callers.
Move kiocb_start_write() from the callers into vfs_iocb_iter_write()
after the rw_verify_area() checks, to make them "start-write-safe".
The semantics of vfs_iocb_iter_write() is changed, so that the caller is
responsible for calling kiocb_end_write() on completion only if async
iocb was queued. The completion handlers of both callers were adapted
to this semantic change.
This is needed for fanotify "pre content" events.
Suggested-by: Jan Kara <jack@suse.cz>
Suggested-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Link: https://lore.kernel.org/r/20231122122715.2561213-14-amir73il@gmail.com
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Christian Brauner <brauner@kernel.org>
All the callers of vfs_iter_write() call file_start_write() just before
calling vfs_iter_write() except for target_core_file's fd_do_rw().
Move file_start_write() from the callers into vfs_iter_write().
fd_do_rw() calls vfs_iter_write() with a non-regular file, so
file_start_write() is a no-op.
This is needed for fanotify "pre content" events.
Suggested-by: Jan Kara <jack@suse.cz>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Link: https://lore.kernel.org/r/20231122122715.2561213-11-amir73il@gmail.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
The main callers of do_splice_direct() also call rw_verify_area() for
the entire range that is being copied, e.g. by vfs_copy_file_range()
or do_sendfile() before calling do_splice_direct().
The only caller that does not have those checks for entire range is
ovl_copy_up_file(). In preparation for removing the checks inside
do_splice_direct(), add rw_verify_area() call in ovl_copy_up_file().
For extra safety, perform minimal sanity checks from rw_verify_area()
for non negative offsets also in the copy up do_splice_direct() loop
without calling the file permission hooks.
This is needed for fanotify "pre content" events.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Link: https://lore.kernel.org/r/20231122122715.2561213-2-amir73il@gmail.com
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Christian Brauner <brauner@kernel.org>
The ->destname member of struct ovl_copy_up_ctx is initialized inside
ovl_copy_up_one() to ->d_name of the overlayfs dentry being copied up
and then it may be overridden by index name inside ovl_do_copy_up().
ovl_inode_lock() in ovl_copy_up_start() and ovl_copy_up() in ovl_rename()
effectively stabilze ->d_name of the overlayfs dentry being copied up,
but ovl_inode_lock() is not held when ->d_name is being read.
It is not a correctness bug, because if ovl_do_copy_up() races with
ovl_rename() and ctx.destname is freed, we will not end up calling
ovl_do_copy_up() with the dead name reference.
The code becomes much easier to understand and to document if the
initialization of c->destname is always done inside ovl_do_copy_up(),
either to the index entry name, or to the overlay dentry ->d_name.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
When the index feature is disabled, ofs->indexdir is NULL.
When the index feature is enabled, ofs->indexdir has the same value as
ofs->workdir and takes an extra reference.
This makes the code harder to understand when it is not always clear
that ofs->indexdir in one function is the same dentry as ofs->workdir
in another function.
Remove this redundancy, by referencing ofs->workdir directly in index
helpers and by using the ovl_indexdir() accessor in generic code.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Commit f9c34674bc ("vfs: factor out helpers d_instantiate_anon() and
d_alloc_anon()") was introduced so overlayfs could initialize a non-dir
disconnected overlay dentry before overlay inode is attached to it.
Since commit ("0af950f57fef ovl: move ovl_entry into ovl_inode"), all
ovl_obtain_alias() can do is set DCACHE_OP_*REVALIDATE flags in ->d_flags
and OVL_E_UPPER_ALIAS flag in ->d_fsdata.
The DCACHE_OP_*REVALIDATE flags and OVL_E_UPPER_ALIAS flag are irrelevant
for a disconnected non-dir dentry, so it is better to use d_obtain_alias()
instead of open coding it.
Suggested-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
When vfs_getattr_nosec() calls a filesystem's getattr interface function
then the 'nosec' should propagate into this function so that
vfs_getattr_nosec() can again be called from the filesystem's gettattr
rather than vfs_getattr(). The latter would add unnecessary security
checks that the initial vfs_getattr_nosec() call wanted to avoid.
Therefore, introduce the getattr flag GETATTR_NOSEC and allow to pass
with the new getattr_flags parameter to the getattr interface function.
In overlayfs and ecryptfs use this flag to determine which one of the
two functions to call.
In a recent code change introduced to IMA vfs_getattr_nosec() ended up
calling vfs_getattr() in overlayfs, which in turn called
security_inode_getattr() on an exiting process that did not have
current->fs set anymore, which then caused a kernel NULL pointer
dereference. With this change the call to security_inode_getattr() can
be avoided, thus avoiding the NULL pointer dereference.
Reported-by: <syzbot+a67fc5321ffb4b311c98@syzkaller.appspotmail.com>
Fixes: db1d1e8b98 ("IMA: use vfs_getattr_nosec to get the i_version")
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: <linux-fsdevel@vger.kernel.org>
Cc: Miklos Szeredi <miklos@szeredi.hu>
Cc: Amir Goldstein <amir73il@gmail.com>
Cc: Tyler Hicks <code@tyhicks.com>
Cc: Mimi Zohar <zohar@linux.ibm.com>
Suggested-by: Christian Brauner <brauner@kernel.org>
Co-developed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Link: https://lore.kernel.org/r/20231002125733.1251467-1-stefanb@linux.vnet.ibm.com
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
On failure to parse parameters in ovl_parse_param_lowerdir(), it is
necessary to update ctx->nr with the correct nr before using
ovl_reset_lowerdirs() to release l->name.
Reported-and-tested-by: syzbot+26eedf3631650972f17c@syzkaller.appspotmail.com
Fixes: c835110b58 ("ovl: remove unused code in lowerdir param parsing")
Co-authored-by: Edward Adam Davis <eadavis@qq.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Pull vfs fanotify fsid updates from Christian Brauner:
"This work is part of the plan to enable fanotify to serve as a drop-in
replacement for inotify. While inotify is availabe on all filesystems,
fanotify currently isn't.
In order to support fanotify on all filesystems two things are needed:
(1) all filesystems need to support AT_HANDLE_FID
(2) all filesystems need to report a non-zero f_fsid
This contains (1) and allows filesystems to encode non-decodable file
handlers for fanotify without implementing any exportfs operations by
encoding a file id of type FILEID_INO64_GEN from i_ino and
i_generation.
Filesystems that want to opt out of encoding non-decodable file ids
for fanotify that don't support NFS export can do so by providing an
empty export_operations struct.
This also partially addresses (2) by generating f_fsid for simple
filesystems as well as freevxfs. Remaining filesystems will be dealt
with by separate patches.
Finally, this contains the patch from the current exportfs maintainers
which moves exportfs under vfs with Chuck, Jeff, and Amir as
maintainers and vfs.git as tree"
* tag 'vfs-6.7.fsid' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs:
MAINTAINERS: create an entry for exportfs
fs: fix build error with CONFIG_EXPORTFS=m or not defined
freevxfs: derive f_fsid from bdev->bd_dev
fs: report f_fsid from s_dev for "simple" filesystems
exportfs: support encoding non-decodeable file handles by default
exportfs: define FILEID_INO64_GEN* file handle types
exportfs: make ->encode_fh() a mandatory method for NFS export
exportfs: add helpers to check if filesystem can encode/decode file handles
Pull overlayfs updates from Amir Goldstein:
- Overlayfs aio cleanups and fixes
Cleanups and minor fixes in preparation for factoring out of
read/write passthrough code.
- Overlayfs lock ordering changes
Hold mnt_writers only throughout copy up instead of a long lived
elevated refcount.
- Add support for nesting overlayfs private xattrs
There are cases where you want to use an overlayfs mount as a
lowerdir for another overlayfs mount. For example, if the system
rootfs is on overlayfs due to composefs, or to make it volatile (via
tmpfs), then you cannot currently store a lowerdir on the rootfs,
because the inner overlayfs will eat all the whiteouts and overlay
xattrs. This means you can't e.g. store on the rootfs a prepared
container image for use with overlayfs.
This adds support for nesting of overlayfs mounts by escaping the
problematic features and unescaping them when exposing to the
overlayfs user.
- Add new mount options for appending lowerdirs
* tag 'ovl-update-6.7' of git://git.kernel.org/pub/scm/linux/kernel/git/overlayfs/vfs:
ovl: add support for appending lowerdirs one by one
ovl: refactor layer parsing helpers
ovl: store and show the user provided lowerdir mount option
ovl: remove unused code in lowerdir param parsing
ovl: Add documentation on nesting of overlayfs mounts
ovl: Add an alternative type of whiteout
ovl: Support escaped overlay.* xattrs
ovl: Add OVL_XATTR_TRUSTED/USER_PREFIX_LEN macros
ovl: Move xattr support to new xattrs.c file
ovl: do not encode lower fh with upper sb_writers held
ovl: do not open/llseek lower file with upper sb_writers held
ovl: reorder ovl_want_write() after ovl_inode_lock()
ovl: split ovl_want_write() into two helpers
ovl: add helper ovl_file_modified()
ovl: protect copying of realinode attributes to ovl inode
ovl: punt write aio completion to workqueue
ovl: propagate IOCB_APPEND flag on writes to realfile
ovl: use simpler function to convert iocb to rw flags
Pull integrity updates from Mimi Zohar:
"Four integrity changes: two IMA-overlay updates, an integrity Kconfig
cleanup, and a secondary keyring update"
* tag 'integrity-v6.7' of git://git.kernel.org/pub/scm/linux/kernel/git/zohar/linux-integrity:
ima: detect changes to the backing overlay file
certs: Only allow certs signed by keys on the builtin keyring
integrity: fix indentation of config attributes
ima: annotate iint mutex to avoid lockdep false positive warnings
Commit 18b44bc5a6 ("ovl: Always reevaluate the file signature for
IMA") forced signature re-evaulation on every file access.
Instead of always re-evaluating the file's integrity, detect a change
to the backing file, by comparing the cached file metadata with the
backing file's metadata. Verifying just the i_version has not changed
is insufficient. In addition save and compare the i_ino and s_dev
as well.
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Tested-by: Eric Snowberg <eric.snowberg@oracle.com>
Tested-by: Raul E Rangel <rrangel@chromium.org>
Cc: stable@vger.kernel.org
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Add new mount options lowerdir+ and datadir+ that can be used to add
layers to lower layers stack one by one.
Unlike the legacy lowerdir mount option, special characters (i.e. colons
and cammas) are not unescaped with these new mount options.
The new mount options can be repeated to compose a large stack of lower
layers, but they may not be mixed with the lagacy lowerdir mount option,
because for displaying lower layers in mountinfo, we do not want to mix
escaped with unescaped lower layers path syntax.
Similar to data-only layer rules with the lowerdir mount option, the
datadir+ option must follow at least one lowerdir+ option and the
lowerdir+ option must not follow the datadir+ option.
If the legacy lowerdir mount option follows lowerdir+ and datadir+
mount options, it overrides them. Sepcifically, calling:
fsconfig(FSCONFIG_SET_STRING, "lowerdir", "", 0);
can be used to reset previously setup lower layers.
Suggested-by: Miklos Szeredi <miklos@szeredi.hu>
Link: https://lore.kernel.org/r/CAJfpegt7VC94KkRtb1dfHG8+4OzwPBLYqhtc8=QFUxpFJE+=RQ@mail.gmail.com/
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
We are about to add new mount options for adding lowerdir one by one,
but those mount options will not support escaping.
For the existing case, where lowerdir mount option is provided as a colon
separated list, store the user provided (possibly escaped) string and
display it as is when showing the lowerdir mount option.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Commit beae836e9c ("ovl: temporarily disable appending lowedirs")
removed the ability to append lowerdirs with syntax lowerdir=":<path>".
Remove leftover code and comments that are irrelevant with lowerdir
append mode disabled.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
An xattr whiteout (called "xwhiteout" in the code) is a reguar file of
zero size with the "overlay.whiteout" xattr set. A file like this in a
directory with the "overlay.whiteouts" xattrs set will be treated the
same way as a regular whiteout.
The "overlay.whiteouts" directory xattr is used in order to
efficiently handle overlay checks in readdir(), as we only need to
checks xattrs in affected directories.
The advantage of this kind of whiteout is that they can be escaped
using the standard overlay xattr escaping mechanism. So, a file with a
"overlay.overlay.whiteout" xattr would be unescaped to
"overlay.whiteout", which could then be consumed by another overlayfs
as a whiteout.
Overlayfs itself doesn't create whiteouts like this, but a userspace
mechanism could use this alternative mechanism to convert images that
may contain whiteouts to be used with overlayfs.
To work as a whiteout for both regular overlayfs mounts as well as
userxattr mounts both the "user.overlay.whiteout*" and the
"trusted.overlay.whiteout*" xattrs will need to be created.
Signed-off-by: Alexander Larsson <alexl@redhat.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
There are cases where you want to use an overlayfs mount as a lowerdir
for another overlayfs mount. For example, if the system rootfs is on
overlayfs due to composefs, or to make it volatile (via tmps), then
you cannot currently store a lowerdir on the rootfs. This means you
can't e.g. store on the rootfs a prepared container image for use
using overlayfs.
To work around this, we introduce an escapment mechanism for overlayfs
xattrs. Whenever the lower/upper dir has a xattr named
"overlay.overlay.XYZ", we list it as "overlay.XYZ" in listxattrs, and
when the user calls getxattr or setxattr on "overlay.XYZ", we apply to
"overlay.overlay.XYZ" in the backing directories.
This allows storing any kind of overlay xattrs in a overlayfs mount
that can be used as a lowerdir in another mount. It is possible to
stack this mechanism multiple times, such that
"overlay.overlay.overlay.XYZ" will survive two levels of overlay mounts,
however this is not all that useful in practice because of stack depth
limitations of overlayfs mounts.
Note: These escaped xattrs are copied to upper during copy-up.
Signed-off-by: Alexander Larsson <alexl@redhat.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
This moves the code from super.c and inode.c, and makes ovl_xattr_get/set()
static.
This is in preparation for doing more work on xattrs support.
Signed-off-by: Alexander Larsson <alexl@redhat.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
When lower fs is a nested overlayfs, calling encode_fh() on a lower
directory dentry may trigger copy up and take sb_writers on the upper fs
of the lower nested overlayfs.
The lower nested overlayfs may have the same upper fs as this overlayfs,
so nested sb_writers lock is illegal.
Move all the callers that encode lower fh to before ovl_want_write().
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
overlayfs file open (ovl_maybe_lookup_lowerdata) and overlay file llseek
take the ovl_inode_lock, without holding upper sb_writers.
In case of nested lower overlay that uses same upper fs as this overlay,
lockdep will warn about (possibly false positive) circular lock
dependency when doing open/llseek of lower ovl file during copy up with
our upper sb_writers held, because the locking ordering seems reverse to
the locking order in ovl_copy_up_start():
- lower ovl_inode_lock
- upper sb_writers
Let the copy up "transaction" keeps an elevated mnt write count on upper
mnt, but leaves taking upper sb_writers to lower level helpers only when
they actually need it. This allows to avoid holding upper sb_writers
during lower file open/llseek and prevents the lockdep warning.
Minimizing the scope of upper sb_writers during copy up is also needed
for fixing another possible deadlocks by a following patch.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Make the locking order of ovl_inode_lock() strictly between the two
vfs stacked layers, i.e.:
- ovl vfs locks: sb_writers, inode_lock, ...
- ovl_inode_lock
- upper vfs locks: sb_writers, inode_lock, ...
To that effect, move ovl_want_write() into the helpers ovl_nlink_start()
and ovl_copy_up_start which currently take the ovl_inode_lock() after
ovl_want_write().
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
ovl_get_write_access() gets write access to upper mnt without taking
freeze protection on upper sb and ovl_start_write() only takes freeze
protection on upper sb.
These helpers will be used to breakup the large ovl_want_write() scope
during copy up into finer grained freeze protection scopes.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
ovl_copyattr() may be called concurrently from aio completion context
without any lock and that could lead to overlay inode attributes getting
permanently out of sync with real inode attributes.
Use ovl inode spinlock to protect ovl_copyattr().
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
We want to protect concurrent updates of ovl inode size and mtime
(i.e. ovl_copyattr()) from aio completion context.
Punt write aio completion to a workqueue so that we can protect
ovl_copyattr() with a spinlock.
Export sb_init_dio_done_wq(), so that overlayfs can use its own
dio workqueue to punt aio completions.
Suggested-by: Jens Axboe <axboe@kernel.dk>
Link: https://lore.kernel.org/r/8620dfd3-372d-4ae0-aa3f-2fe97dda1bca@kernel.dk/
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
If ovl file is opened O_APPEND, the underlying realfile is also
opened O_APPEND, so it makes sense to propagate the IOCB_APPEND flags
on sync writes to realfile, just as we do with aio writes.
Effectively, because sync ovl writes are protected by inode lock,
this change only makes a difference if the realfile is written to (size
extending writes) from underneath overlayfs. The behavior in this case
is undefined, so it is ok if we change the behavior (to fail the ovl
IOCB_APPEND write).
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Overlayfs implements its own function to translate iocb flags into rw
flags, so that they can be passed into another vfs call.
With commit ce71bfea20 ("fs: align IOCB_* flags with RWF_* flags")
Jens created a 1:1 matching between the iocb flags and rw flags,
simplifying the conversion.
Signed-off-by: Alessio Balsini <balsini@android.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Pull vfs inode time accessor updates from Christian Brauner:
"This finishes the conversion of all inode time fields to accessor
functions as discussed on list. Changing timestamps manually as we
used to do before is error prone. Using accessors function makes this
robust.
It does not contain the switch of the time fields to discrete 64 bit
integers to replace struct timespec and free up space in struct inode.
But after this, the switch can be trivially made and the patch should
only affect the vfs if we decide to do it"
* tag 'vfs-6.7.ctime' of gitolite.kernel.org:pub/scm/linux/kernel/git/vfs/vfs: (86 commits)
fs: rename inode i_atime and i_mtime fields
security: convert to new timestamp accessors
selinux: convert to new timestamp accessors
apparmor: convert to new timestamp accessors
sunrpc: convert to new timestamp accessors
mm: convert to new timestamp accessors
bpf: convert to new timestamp accessors
ipc: convert to new timestamp accessors
linux: convert to new timestamp accessors
zonefs: convert to new timestamp accessors
xfs: convert to new timestamp accessors
vboxsf: convert to new timestamp accessors
ufs: convert to new timestamp accessors
udf: convert to new timestamp accessors
ubifs: convert to new timestamp accessors
tracefs: convert to new timestamp accessors
sysv: convert to new timestamp accessors
squashfs: convert to new timestamp accessors
server: convert to new timestamp accessors
client: convert to new timestamp accessors
...
Pull vfs xattr updates from Christian Brauner:
"The 's_xattr' field of 'struct super_block' currently requires a
mutable table of 'struct xattr_handler' entries (although each handler
itself is const). However, no code in vfs actually modifies the
tables.
This changes the type of 's_xattr' to allow const tables, and modifies
existing file systems to move their tables to .rodata. This is
desirable because these tables contain entries with function pointers
in them; moving them to .rodata makes it considerably less likely to
be modified accidentally or maliciously at runtime"
* tag 'vfs-6.7.xattr' of gitolite.kernel.org:pub/scm/linux/kernel/git/vfs/vfs: (30 commits)
const_structs.checkpatch: add xattr_handler
net: move sockfs_xattr_handlers to .rodata
shmem: move shmem_xattr_handlers to .rodata
overlayfs: move xattr tables to .rodata
xfs: move xfs_xattr_handlers to .rodata
ubifs: move ubifs_xattr_handlers to .rodata
squashfs: move squashfs_xattr_handlers to .rodata
smb: move cifs_xattr_handlers to .rodata
reiserfs: move reiserfs_xattr_handlers to .rodata
orangefs: move orangefs_xattr_handlers to .rodata
ocfs2: move ocfs2_xattr_handlers and ocfs2_xattr_handler_map to .rodata
ntfs3: move ntfs_xattr_handlers to .rodata
nfs: move nfs4_xattr_handlers to .rodata
kernfs: move kernfs_xattr_handlers to .rodata
jfs: move jfs_xattr_handlers to .rodata
jffs2: move jffs2_xattr_handlers to .rodata
hfsplus: move hfsplus_xattr_handlers to .rodata
hfs: move hfs_xattr_handlers to .rodata
gfs2: move gfs2_xattr_handlers_max to .rodata
fuse: move fuse_xattr_handlers to .rodata
...
Pull misc vfs updates from Christian Brauner:
"This contains the usual miscellaneous features, cleanups, and fixes
for vfs and individual fses.
Features:
- Rename and export helpers that get write access to a mount. They
are used in overlayfs to get write access to the upper mount.
- Print the pretty name of the root device on boot failure. This
helps in scenarios where we would usually only print
"unknown-block(1,2)".
- Add an internal SB_I_NOUMASK flag. This is another part in the
endless POSIX ACL saga in a way.
When POSIX ACLs are enabled via SB_POSIXACL the vfs cannot strip
the umask because if the relevant inode has POSIX ACLs set it might
take the umask from there. But if the inode doesn't have any POSIX
ACLs set then we apply the umask in the filesytem itself. So we end
up with:
(1) no SB_POSIXACL -> strip umask in vfs
(2) SB_POSIXACL -> strip umask in filesystem
The umask semantics associated with SB_POSIXACL allowed filesystems
that don't even support POSIX ACLs at all to raise SB_POSIXACL
purely to avoid umask stripping. That specifically means NFS v4 and
Overlayfs. NFS v4 does it because it delegates this to the server
and Overlayfs because it needs to delegate umask stripping to the
upper filesystem, i.e., the filesystem used as the writable layer.
This went so far that SB_POSIXACL is raised eve on kernels that
don't even have POSIX ACL support at all.
Stop this blatant abuse and add SB_I_NOUMASK which is an internal
superblock flag that filesystems can raise to opt out of umask
handling. That should really only be the two mentioned above. It's
not that we want any filesystems to do this. Ideally we have all
umask handling always in the vfs.
- Make overlayfs use SB_I_NOUMASK too.
- Now that we have SB_I_NOUMASK, stop checking for SB_POSIXACL in
IS_POSIXACL() if the kernel doesn't have support for it. This is a
very old patch but it's only possible to do this now with the wider
cleanup that was done.
- Follow-up work on fake path handling from last cycle. Citing mostly
from Amir:
When overlayfs was first merged, overlayfs files of regular files
and directories, the ones that are installed in file table, had a
"fake" path, namely, f_path is the overlayfs path and f_inode is
the "real" inode on the underlying filesystem.
In v6.5, we took another small step by introducing of the
backing_file container and the file_real_path() helper. This change
allowed vfs and filesystem code to get the "real" path of an
overlayfs backing file. With this change, we were able to make
fsnotify work correctly and report events on the "real" filesystem
objects that were accessed via overlayfs.
This method works fine, but it still leaves the vfs vulnerable to
new code that is not aware of files with fake path. A recent
example is commit db1d1e8b98 ("IMA: use vfs_getattr_nosec to get
the i_version"). This commit uses direct referencing to f_path in
IMA code that otherwise uses file_inode() and file_dentry() to
reference the filesystem objects that it is measuring.
This contains work to switch things around: instead of having
filesystem code opt-in to get the "real" path, have generic code
opt-in for the "fake" path in the few places that it is needed.
Is it far more likely that new filesystems code that does not use
the file_dentry() and file_real_path() helpers will end up causing
crashes or averting LSM/audit rules if we keep the "fake" path
exposed by default.
This change already makes file_dentry() moot, but for now we did
not change this helper just added a WARN_ON() in ovl_d_real() to
catch if we have made any wrong assumptions.
After the dust settles on this change, we can make file_dentry() a
plain accessor and we can drop the inode argument to ->d_real().
- Switch struct file to SLAB_TYPESAFE_BY_RCU. This looks like a small
change but it really isn't and I would like to see everyone on
their tippie toes for any possible bugs from this work.
Essentially we've been doing most of what SLAB_TYPESAFE_BY_RCU for
files since a very long time because of the nasty interactions
between the SCM_RIGHTS file descriptor garbage collection. So
extending it makes a lot of sense but it is a subtle change. There
are almost no places that fiddle with file rcu semantics directly
and the ones that did mess around with struct file internal under
rcu have been made to stop doing that because it really was always
dodgy.
I forgot to put in the link tag for this change and the discussion
in the commit so adding it into the merge message:
https://lore.kernel.org/r/20230926162228.68666-1-mjguzik@gmail.com
Cleanups:
- Various smaller pipe cleanups including the removal of a spin lock
that was only used to protect against writes without pipe_lock()
from O_NOTIFICATION_PIPE aka watch queues. As that was never
implemented remove the additional locking from pipe_write().
- Annotate struct watch_filter with the new __counted_by attribute.
- Clarify do_unlinkat() cleanup so that it doesn't look like an extra
iput() is done that would cause issues.
- Simplify file cleanup when the file has never been opened.
- Use module helper instead of open-coding it.
- Predict error unlikely for stale retry.
- Use WRITE_ONCE() for mount expiry field instead of just commenting
that one hopes the compiler doesn't get smart.
Fixes:
- Fix readahead on block devices.
- Fix writeback when layztime is enabled and inodes whose timestamp
is the only thing that changed reside on wb->b_dirty_time. This
caused excessively large zombie memory cgroup when lazytime was
enabled as such inodes weren't handled fast enough.
- Convert BUG_ON() to WARN_ON_ONCE() in open_last_lookups()"
* tag 'vfs-6.7.misc' of gitolite.kernel.org:pub/scm/linux/kernel/git/vfs/vfs: (26 commits)
file, i915: fix file reference for mmap_singleton()
vfs: Convert BUG_ON to WARN_ON_ONCE in open_last_lookups
writeback, cgroup: switch inodes with dirty timestamps to release dying cgwbs
chardev: Simplify usage of try_module_get()
ovl: rely on SB_I_NOUMASK
fs: fix umask on NFS with CONFIG_FS_POSIX_ACL=n
fs: store real path instead of fake path in backing file f_path
fs: create helper file_user_path() for user displayed mapped file path
fs: get mnt_writers count for an open backing file's real path
vfs: stop counting on gcc not messing with mnt_expiry_mark if not asked
vfs: predict the error in retry_estale as unlikely
backing file: free directly
vfs: fix readahead(2) on block devices
io_uring: use files_lookup_fd_locked()
file: convert to SLAB_TYPESAFE_BY_RCU
vfs: shave work on failed file open
fs: simplify misleading code to remove ambiguity regarding ihold()/iput()
watch_queue: Annotate struct watch_filter with __counted_by
fs/pipe: use spinlock in pipe_read() only if there is a watch_queue
fs/pipe: remove unnecessary spinlock from pipe_write()
...
In commit f61b9bb3f838 ("fs: add a new SB_I_NOUMASK flag") we added a
new SB_I_NOUMASK flag that is used by filesystems like NFS to indicate
that umask stripping is never supposed to be done in the vfs independent
of whether or not POSIX ACLs are supported.
Overlayfs falls into the same category as it raises SB_POSIXACL
unconditionally to defer umask application to the upper filesystem.
Now that we have SB_I_NOUMASK use that and make SB_POSIXACL properly
conditional on whether or not the kernel does have support for it. This
will enable use to turn IS_POSIXACL() into nop on kernels that don't
have POSIX ACL support avoding bugs from missed umask stripping.
Link: https://lore.kernel.org/r/20231012-einband-uferpromenade-80541a047a1f@brauner
Acked-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
A backing file struct stores two path's, one "real" path that is referring
to f_inode and one "fake" path, which should be displayed to users in
/proc/<pid>/maps.
There is a lot more potential code that needs to know the "real" path, then
code that needs to know the "fake" path.
Instead of code having to request the "real" path with file_real_path(),
store the "real" path in f_path and require code that needs to know the
"fake" path request it with file_user_path().
Replace the file_real_path() helper with a simple const accessor f_path().
After this change, file_dentry() is not expected to observe any files
with overlayfs f_path and real f_inode, so the call to ->d_real() should
not be needed. Leave the ->d_real() call for now and add an assertion
in ovl_d_real() to catch if we made wrong assumptions.
Suggested-by: Miklos Szeredi <miklos@szeredi.hu>
Link: https://lore.kernel.org/r/CAJfpegtt48eXhhjDFA1ojcHPNKj3Go6joryCPtEFAKpocyBsnw@mail.gmail.com/
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Link: https://lore.kernel.org/r/20231009153712.1566422-4-amir73il@gmail.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
Kernel v6.5 converted overlayfs to new mount api.
As an added bonus, it also added a feature to allow appending lowerdirs
using lowerdir=:/lower2,lowerdir=::/data3 syntax.
This new syntax has raised some concerns regarding escaping of colons.
We decided to try and disable this syntax, which hasn't been in the wild
for so long and introduce it again in 6.7 using explicit mount options
lowerdir+=/lower2,datadir+=/data3.
Suggested-by: Miklos Szeredi <miklos@szeredi.hu>
Link: https://lore.kernel.org/r/CAJfpegsr3A4YgF2YBevWa6n3=AcP7hNndG6EPMu3ncvV-AM71A@mail.gmail.com/
Fixes: b36a5780cb ("ovl: modify layer parameter parsing")
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Before commit b36a5780cb ("ovl: modify layer parameter parsing"),
spaces and commas in lowerdir mount option value used to be escaped using
seq_show_option().
In current upstream, when lowerdir value has a space, it is not escaped
in /proc/mounts, e.g.:
none /mnt overlay rw,relatime,lowerdir=l l,upperdir=u,workdir=w 0 0
which results in broken output of the mount utility:
none on /mnt type overlay (rw,relatime,lowerdir=l)
Store the original lowerdir mount options before unescaping and show
them using the same escaping used for seq_show_option() in addition to
escaping the colon separator character.
Fixes: b36a5780cb ("ovl: modify layer parameter parsing")
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Ever since commit 91c7794713 ("ovl: allow filenames with comma"), the
following example was legit overlayfs mount options:
mount -t overlay overlay -o 'lowerdir=/tmp/a\,b/lower' /mnt
The conversion to new mount api moved to using the common helper
generic_parse_monolithic() and discarded the specialized ovl_next_opt()
option separator.
Bring back ovl_next_opt() and use vfs_parse_monolithic_sep() to fix the
regression.
Reported-by: Ryan Hendrickson <ryan.hendrickson@alum.mit.edu>
Closes: https://lore.kernel.org/r/8da307fb-9318-cf78-8a27-ba5c5a0aef6d@alum.mit.edu/
Fixes: 1784fbc2ed ("ovl: port to new mount api")
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
ovl_permission() accesses ->layers[...].mnt; we can't have ->layers
freed without an RCU delay on fs shutdown.
Fortunately, kern_unmount_array() that is used to drop those mounts
does include an RCU delay, so freeing is delayed; unfortunately, the
array passed to kern_unmount_array() is formed by mangling ->layers
contents and that happens without any delays.
The ->layers[...].name string entries are used to store the strings to
display in "lowerdir=..." by ovl_show_options(). Those entries are not
accessed in RCU walk.
Move the name strings into a separate array ofs->config.lowerdirs and
reuse the ofs->config.lowerdirs array as the temporary mount array to
pass to kern_unmount_array().
Reported-by: Al Viro <viro@zeniv.linux.org.uk>
Link: https://lore.kernel.org/r/20231002023711.GP3389589@ZenIV/
Acked-by: Miklos Szeredi <mszeredi@redhat.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
d_inode_rcu() is right - we might be in rcu pathwalk;
however, OVL_E() hides plain d_inode() on the same dentry...
Fixes: a6ff2bc0be ("ovl: use OVL_E() and OVL_E_FLAGS() accessors")
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Commit 724768a393 ("ovl: fix incorrect fdput() on aio completion")
took a refcount on real file before submitting aio, but forgot to
avoid clearing FDPUT_FPUT from real.flags stack variable.
This can result in a file reference leak.
Fixes: 724768a393 ("ovl: fix incorrect fdput() on aio completion")
Reported-by: Gil Lev <contact@levgil.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
