Overlayfs's volatile option allows the user to bypass all forced sync calls
to the upperdir filesystem. This comes at the cost of safety. We can never
ensure that the user's data is intact, but we can make a best effort to
expose whether or not the data is likely to be in a bad state.
The best way to handle this in the time being is that if an overlayfs's
upperdir experiences an error after a volatile mount occurs, that error
will be returned on fsync, fdatasync, sync, and syncfs. This is
contradictory to the traditional behaviour of VFS which fails the call
once, and only raises an error if a subsequent fsync error has occurred,
and been raised by the filesystem.
One awkward aspect of the patch is that we have to manually set the
superblock's errseq_t after the sync_fs callback as opposed to just
returning an error from syncfs. This is because the call chain looks
something like this:
sys_syncfs ->
sync_filesystem ->
__sync_filesystem ->
/* The return value is ignored here
sb->s_op->sync_fs(sb)
_sync_blockdev
/* Where the VFS fetches the error to raise to userspace */
errseq_check_and_advance
Because of this we call errseq_set every time the sync_fs callback occurs.
Due to the nature of this seen / unseen dichotomy, if the upperdir is an
inconsistent state at the initial mount time, overlayfs will refuse to
mount, as overlayfs cannot get a snapshot of the upperdir's errseq that
will increment on error until the user calls syncfs.
Signed-off-by: Sargun Dhillon <sargun@sargun.me>
Suggested-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Fixes: c86243b090 ("ovl: provide a mount option "volatile"")
Cc: stable@vger.kernel.org
Reviewed-by: Vivek Goyal <vgoyal@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
The function ovl_dir_real_file() currently uses the inode lock to serialize
writes to the od->upperfile field.
However, this function will get called by ovl_ioctl_set_flags(), which
utilizes the inode lock too. In this case ovl_dir_real_file() will try to
claim a lock that is owned by a function in its call stack, which won't get
released before ovl_dir_real_file() returns.
Fix by replacing the open coded compare and exchange by an explicit atomic
op.
Fixes: 61536bed21 ("ovl: support [S|G]ETFLAGS and FS[S|G]ETXATTR ioctls for directories")
Cc: stable@vger.kernel.org # v5.10
Reported-by: Icenowy Zheng <icenowy@aosc.io>
Tested-by: Icenowy Zheng <icenowy@aosc.io>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
[S|G]ETFLAGS and FS[S|G]ETXATTR ioctls are applicable to both files and
directories, so add ioctl operations to dir as well.
We teach ovl_real_fdget() to get the realfile of directories which use
a different type of file->private_data.
Ifdef away compat ioctl implementation to conform to standard practice.
With this change, xfstest generic/079 which tests these ioctls on files
and directories passes.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Xiao Yang <yangx.jy@cn.fujitsu.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Container folks are complaining that dnf/yum issues too many sync while
installing packages and this slows down the image build. Build requirement
is such that they don't care if a node goes down while build was still
going on. In that case, they will simply throw away unfinished layer and
start new build. So they don't care about syncing intermediate state to the
disk and hence don't want to pay the price associated with sync.
So they are asking for mount options where they can disable sync on overlay
mount point.
They primarily seem to have two use cases.
- For building images, they will mount overlay with nosync and then sync
upper layer after unmounting overlay and reuse upper as lower for next
layer.
- For running containers, they don't seem to care about syncing upper layer
because if node goes down, they will simply throw away upper layer and
create a fresh one.
So this patch provides a mount option "volatile" which disables all forms
of sync. Now it is caller's responsibility to throw away upper if system
crashes or shuts down and start fresh.
With "volatile", I am seeing roughly 20% speed up in my VM where I am just
installing emacs in an image. Installation time drops from 31 seconds to 25
seconds when nosync option is used. This is for the case of building on top
of an image where all packages are already cached. That way I take out the
network operations latency out of the measurement.
Giuseppe is also looking to cut down on number of iops done on the disk. He
is complaining that often in cloud their VMs are throttled if they cross
the limit. This option can help them where they reduce number of iops (by
cutting down on frequent sync and writebacks).
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
An incompatible feature is marked by a non-empty directory nested
2 levels deep under "work" dir, e.g.:
workdir/work/incompat/volatile.
This commit checks for marked incompat features, warns about them
and fails to mount the overlay, for example:
overlayfs: overlay with incompat feature 'volatile' cannot be mounted
Very old kernels (i.e. v3.18) will fail to remove a non-empty "work"
dir and fail the mount. Newer kernels will fail to remove a "work"
dir with entries nested 3 levels and fall back to read-only mount.
User mounting with old kernel will see a warning like these in dmesg:
overlayfs: cleanup of 'incompat/...' failed (-39)
overlayfs: cleanup of 'work/incompat' failed (-39)
overlayfs: cleanup of 'ovl-work/work' failed (-39)
overlayfs: failed to create directory /vdf/ovl-work/work (errno: 17);
mounting read-only
These warnings should give the hint to the user that:
1. mount failure is caused by backward incompatible features
2. mount failure can be resolved by manually removing the "work" directory
There is nothing preventing users on old kernels from manually removing
workdir entirely or mounting overlay with a new workdir, so this is in
no way a full proof backward compatibility enforcement, but only a best
effort.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
In preparation for more permission checking, override credentials for
directory operations on the underlying filesystems.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
The three instances of ovl_path_open() in overlayfs/readdir.c do three
different things:
- pass f_flags from overlay file
- pass O_RDONLY | O_DIRECTORY
- pass just O_RDONLY
The value of f_flags can be (other than O_RDONLY):
O_WRONLY - not possible for a directory
O_RDWR - not possible for a directory
O_CREAT - masked out by dentry_open()
O_EXCL - masked out by dentry_open()
O_NOCTTY - masked out by dentry_open()
O_TRUNC - masked out by dentry_open()
O_APPEND - no effect on directory ops
O_NDELAY - no effect on directory ops
O_NONBLOCK - no effect on directory ops
__O_SYNC - no effect on directory ops
O_DSYNC - no effect on directory ops
FASYNC - no effect on directory ops
O_DIRECT - no effect on directory ops
O_LARGEFILE - ?
O_DIRECTORY - only affects lookup
O_NOFOLLOW - only affects lookup
O_NOATIME - overlay sets this unconditionally in ovl_path_open()
O_CLOEXEC - only affects fd allocation
O_PATH - no effect on directory ops
__O_TMPFILE - not possible for a directory
Fon non-merge directories we use the underlying filesystem's iterate; in
this case honor O_LARGEFILE from the original file to make sure that open
doesn't get rejected.
For merge directories it's safe to pass O_LARGEFILE unconditionally since
userspace will only see the artificial offsets created by overlayfs.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Share inode with different whiteout files for saving inode and speeding up
delete operation.
If EMLINK is encountered when linking a shared whiteout, create a new one.
In case of any other error, disable sharing for this super block.
Note: ofs->whiteout is protected by inode lock on workdir.
Signed-off-by: Chengguang Xu <cgxu519@mykernel.net>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Teach ovl_indexdir_cleanup() to remove temp directories containing
whiteouts to prepare for using index dir instead of work dir for removing
merge directories.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
So far, with xino=auto, we only enable xino if we know that all
underlying filesystem use 32bit inode numbers.
When users configure overlay with xino=auto, they already declare that
they are ready to handle 64bit inode number from overlay.
It is a very common case, that underlying filesystem uses 64bit ino,
but rarely or never uses the high inode number bits (e.g. tmpfs, xfs).
Leaving it for the users to declare high ino bits are unused with
xino=on is not a recipe for many users to enjoy the benefits of xino.
There appears to be very little reason not to enable xino when users
declare xino=auto even if we do not know how many bits underlying
filesystem uses for inode numbers.
In the worst case of xino bits overflow by real inode number, we
already fall back to the non-xino behavior - real inode number with
unique pseudo dev or to non persistent inode number and overlay st_dev
(for directories).
The only annoyance from auto enabling xino is that xino bits overflow
emits a warning to kmsg. Suppress those warnings unless users explicitly
asked for xino=on, suggesting that they expected high ino bits to be
unused by underlying filesystem.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
When xino feature is enabled and a real directory inode number overflows
the lower xino bits, we cannot map this directory inode number to a unique
and persistent inode number and we fall back to the real inode st_ino and
overlay st_dev.
The real inode st_ino with high bits may collide with a lower inode number
on overlay st_dev that was mapped using xino.
To avoid possible collision with legitimate xino values, map a non
persistent inode number to a dedicated range in the xino address space.
The dedicated range is created by adding one more bit to the number of
reserved high xino bits. We could have added just one more fsid, but that
would have had the undesired effect of changing persistent overlay inode
numbers on kernel or require more complex xino mapping code.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
No code uses the sb returned from this helper, so make it retrun a boolean
and rename it to ovl_same_fs().
The xino mode is irrelevant when all layers are on same fs, so instead of
describing samefs with mode OVL_XINO_OFF, use a new xino_mode state, which
is 0 in the case of samefs, -1 in the case of xino=off and > 0 with xino
enabled.
Create a new helper ovl_same_dev(), to use instead of the common check for
(ovl_same_fs() || xinobits).
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
The WARN_ON() that child entry is always on overlay st_dev became wrong
when we allowed this function to update d_ino in non-samefs setup with xino
enabled.
It is not true in case of xino bits overflow on a non-dir inode. Leave the
WARN_ON() only for directories, where assertion is still true.
Fixes: adbf4f7ea8 ("ovl: consistent d_ino for non-samefs with xino")
Cc: <stable@vger.kernel.org> # v4.17+
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Based on 2 normalized pattern(s):
this program is free software you can redistribute it and or modify
it under the terms of the gnu general public license version 2 as
published by the free software foundation
this program is free software you can redistribute it and or modify
it under the terms of the gnu general public license version 2 as
published by the free software foundation #
extracted by the scancode license scanner the SPDX license identifier
GPL-2.0-only
has been chosen to replace the boilerplate/reference in 4122 file(s).
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Enrico Weigelt <info@metux.net>
Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org>
Reviewed-by: Allison Randal <allison@lohutok.net>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190604081206.933168790@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Only upper dir can be impure, but if we are in the middle of
iterating a lower real dir, dir could be copied up and marked
impure. We only want the impure cache if we started iterating
a real upper dir to begin with.
Aditya Kali reported that the following reproducer hits the
WARN_ON(!cache->refcount) in ovl_get_cache():
docker run --rm drupal:8.5.4-fpm-alpine \
sh -c 'cd /var/www/html/vendor/symfony && \
chown -R www-data:www-data . && ls -l .'
Reported-by: Aditya Kali <adityakali@google.com>
Tested-by: Aditya Kali <adityakali@google.com>
Fixes: 4edb83bb10 ('ovl: constant d_ino for non-merge dirs')
Cc: <stable@vger.kernel.org> # v4.14
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
When overlay layers are not all on the same fs, but all inode numbers
of underlying fs do not use the high 'xino' bits, overlay st_ino values
are constant and persistent.
In that case, relax non-samefs constraint for consistent d_ino and always
iterate non-merge dir using ovl_fill_real() actor so we can remap lower
inode numbers to unique lower fs range.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Orphan index entries are non-dir index entries whose union nlink count
dropped to zero. With index=on, orphan index entries are removed on
mount. With NFS export feature enabled, orphan index entries are replaced
with white out index entries to block future open by handle from opening
the lower file.
When dir index has a stale 'upper' xattr, we assume that the upper dir
was removed and we treat the dir index as orphan entry that needs to be
whited out or removed.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Pass the fs instance with lower_layers array instead of the dentry
lowerstack array to ovl_check_origin_fh(), because the dentry members
of lowerstack play no role in this helper.
This change simplifies the argument list of ovl_check_origin(),
ovl_cleanup_index() and ovl_verify_index().
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
The optimization in ovl_cache_get_impure() that tries to remove an
unneeded "impure" xattr needs to take mnt_want_write() on upper fs.
Fixes: 4edb83bb10 ("ovl: constant d_ino for non-merge dirs")
Cc: <stable@vger.kernel.org> #v4.14
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
The functions ovl_lower_positive() and ovl_check_empty_dir() both take
inode mutex on the real lower dir under ovl_want_write() which takes
the upper_mnt sb_writers lock.
While this is not a clear locking order or layering violation, it creates
an undesired lock dependency between two unrelated layers for no good
reason.
This lock dependency materializes to a false(?) positive lockdep warning
when calling rmdir() on a nested overlayfs, where both nested and
underlying overlayfs both use the same fs type as upper layer.
rmdir() on the nested overlayfs creates the lock chain:
sb_writers of upper_mnt (e.g. tmpfs) in ovl_do_remove()
ovl_i_mutex_dir_key[] of lower overlay dir in ovl_lower_positive()
rmdir() on the underlying overlayfs creates the lock chain in
reverse order:
ovl_i_mutex_dir_key[] of lower overlay dir in vfs_rmdir()
sb_writers of nested upper_mnt (e.g. tmpfs) in ovl_do_remove()
To rid of the unneeded locking dependency, move both ovl_lower_positive()
and ovl_check_empty_dir() to before ovl_want_write() in rmdir() and
rename() implementation.
This change spreads the pieces of ovl_check_empty_and_clear() directly
inside the rmdir()/rename() implementations so the helper is no longer
needed and removed.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
As a writable mount, it is not expected for overlayfs to return
EINVAL/EROFS for fsync, even if dir/file is not changed.
This commit fixes the case of fsync of directory, which is easier to
address, because overlayfs already implements fsync file operation for
directories.
The problem reported by Raphael is that new PostgreSQL 10.0 with a
database in overlayfs where lower layer in squashfs fails to start.
The failure is due to fsync error, when PostgreSQL does fsync on all
existing db directories on startup and a specific directory exists
lower layer with no changes.
Reported-by: Raphael Hertzog <raphael@ouaza.com>
Cc: <stable@vger.kernel.org> # v3.18
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Tested-by: Raphaël Hertzog <hertzog@debian.org>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Conform two stray warning messages to the standard overlayfs: prefix.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
This fixes a regression with readdir of impure dir in overlayfs
that is shared to VM via 9p fs.
Reported-by: Miguel Bernal Marin <miguel.bernal.marin@linux.intel.com>
Fixes: 4edb83bb10 ("ovl: constant d_ino for non-merge dirs")
Cc: <stable@vger.kernel.org> #4.14
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Tested-by: Miguel Bernal Marin <miguel.bernal.marin@linux.intel.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Pull overlayfs updates from Miklos Szeredi:
- Report constant st_ino values across copy-up even if underlying
layers are on different filesystems, but using different st_dev
values for each layer.
Ideally we'd report the same st_dev across the overlay, and it's
possible to do for filesystems that use only 32bits for st_ino by
unifying the inum space. It would be nice if it wasn't a choice of 32
or 64, rather filesystems could report their current maximum (that
could change on resize, so it wouldn't be set in stone).
- miscellaneus fixes and a cleanup of ovl_fill_super(), that was long
overdue.
- created a path_put_init() helper that clears out the pointers after
putting the ref.
I think this could be useful elsewhere, so added it to <linux/path.h>
* 'overlayfs-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs: (30 commits)
ovl: remove unneeded arg from ovl_verify_origin()
ovl: Put upperdentry if ovl_check_origin() fails
ovl: rename ufs to ofs
ovl: clean up getting lower layers
ovl: clean up workdir creation
ovl: clean up getting upper layer
ovl: move ovl_get_workdir() and ovl_get_lower_layers()
ovl: reduce the number of arguments for ovl_workdir_create()
ovl: change order of setup in ovl_fill_super()
ovl: factor out ovl_free_fs() helper
ovl: grab reference to workbasedir early
ovl: split out ovl_get_indexdir() from ovl_fill_super()
ovl: split out ovl_get_lower_layers() from ovl_fill_super()
ovl: split out ovl_get_workdir() from ovl_fill_super()
ovl: split out ovl_get_upper() from ovl_fill_super()
ovl: split out ovl_get_lowerstack() from ovl_fill_super()
ovl: split out ovl_get_workpath() from ovl_fill_super()
ovl: split out ovl_get_upperpath() from ovl_fill_super()
ovl: use path_put_init() in error paths for ovl_fill_super()
vfs: add path_put_init()
...
Define new structures to represent overlay instance lower layers and
overlay merge dir lower layers to make room for storing more per layer
information in-memory.
Instead of keeping the fs instance lower layers in an array of struct
vfsmount, keep them in an array of new struct ovl_layer, that has a
pointer to struct vfsmount.
Instead of keeping the dentry lower layers in an array of struct path,
keep them in an array of new struct ovl_path, that has a pointer to
struct dentry and to struct ovl_layer.
Add a small helper to find the fs layer id that correspopnds to a lower
struct ovl_path and use it in ovl_lookup().
[amir: split re-structure from anonymous bdev patch]
Signed-off-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Filter out non-whiteout non-upper entries from list of merge dir entries
while checking if merge dir is empty in ovl_check_empty_dir().
The remaining work for ovl_clear_empty() is to clear all entries on the
list.
[amir: split patch from rmdir bug fix]
Signed-off-by: zhangyi (F) <yi.zhang@huawei.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
If a non-merge dir in an overlay mount has an overlay.origin xattr, it
means it was once an upper merge dir, which may contain whiteouts and
then the lower dir was removed under it.
Do not iterate real dir directly in this case to avoid exposing whiteouts.
[SzM] Set OVL_WHITEOUT for all merge directories as well.
[amir] A directory that was just copied up does not have the OVL_WHITEOUTS
flag. We need to set it to fix merge dir iteration.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
With index=on, ovl_indexdir_cleanup() tries to cleanup invalid index
entries (e.g. bad index name). This behavior could result in cleaning of
entries created by newer kernels and is therefore undesirable.
Instead, abort mount if such entries are encountered. We still cleanup
'stale' entries and 'orphan' entries, both those cases can be a result
of offline changes to lower and upper dirs.
When encoutering an index entry of type directory or whiteout, kernel
was supposed to fallback to read-only mount, but the fill_super()
operation returns EROFS in this case instead of returning success with
read-only mount flag, so mount fails when encoutering directory or
whiteout index entries. Bless this behavior by returning -EINVAL on
directory and whiteout index entries as we do for all unsupported index
entries.
Fixes: 61b674710c ("ovl: do not cleanup directory and whiteout index..")
Cc: <stable@vger.kernel.org> # v4.13
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
READ_ONCE() now has an implicit smp_read_barrier_depends() call, so it
can be used instead of lockless_dereference() without any change in
semantics.
Signed-off-by: Will Deacon <will.deacon@arm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1508840570-22169-4-git-send-email-will.deacon@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
index dentry was not released when breaking out of the loop
due to index verification error.
Fixes: 415543d5c6 ("ovl: cleanup bad and stale index entries on mount")
Cc: <stable@vger.kernel.org> # v4.13
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Pull overlayfs updates from Miklos Szeredi:
"This fixes d_ino correctness in readdir, which brings overlayfs on par
with normal filesystems regarding inode number semantics, as long as
all layers are on the same filesystem.
There are also some bug fixes, one in particular (random ioctl's
shouldn't be able to modify lower layers) that touches some vfs code,
but of course no-op for non-overlay fs"
* 'overlayfs-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs:
ovl: fix false positive ESTALE on lookup
ovl: don't allow writing ioctl on lower layer
ovl: fix relatime for directories
vfs: add flags to d_real()
ovl: cleanup d_real for negative
ovl: constant d_ino for non-merge dirs
ovl: constant d_ino across copy up
ovl: fix readdir error value
ovl: check snprintf return
While we could replace the smp_mb__before_spinlock() with the new
smp_mb__after_spinlock(), the normal pattern is to use
smp_store_release() to publish an object that is used for
lockless_dereference() -- and mirrors the regular rcu_assign_pointer()
/ rcu_dereference() patterns.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Impure directories are ones which contain objects with origins (i.e. those
that have been copied up). These are relevant to readdir operation only
because of the d_ino field, no other transformation is necessary. Also a
directory can become impure between two getdents(2) calls.
This patch creates a cache for impure directories. Unlike the cache for
merged directories, this one only contains entries with origin and is not
refcounted but has a its lifetime tied to that of the dentry.
Similarly to the merged cache, the impure cache is invalidated based on a
version number. This version number is incremented when an entry with
origin is added or removed from the directory.
If the cache is empty, then the impure xattr is removed from the directory.
This patch also fixes up handling of d_ino for the ".." entry if the parent
directory is merged.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
When all layers are on the same fs, and iterating a directory which may
contain copy up entries, call vfs_getattr() on the overlay entries to make
sure that d_ino will be consistent with st_ino from stat(2).
There is an overhead of lookup per upper entry in readdir.
The overhead is minimal if the iterated entries are already in dcache. It
is also quite useful for the common case of 'ls -l' that readdir() pre
populates the dcache with the listed entries, making the following stat()
calls faster.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
actor's return value is taken as a bool (filled/not filled) so we need to
return the error in the context.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Directory index entries are going to be used for looking up
redirected upper dirs by lower dir fh when decoding an overlay
file handle of a merge dir.
Whiteout index entries are going to be used as an indication that
an exported overlay file handle should be treated as stale (i.e.
after unlink of the overlay inode).
We don't know the verification rules for directory and whiteout
index entries, because they have not been implemented yet, so fail
to mount overlay rw if those entries are found to avoid corrupting
an index that was created by a newer kernel.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Bad index entries are entries whose name does not match the
origin file handle stored in trusted.overlay.origin xattr.
Bad index entries could be a result of a system power off in
the middle of copy up.
Stale index entries are entries whose origin file handle is
stale. Stale index entries could be a result of copying layers
or removing lower entries while the overlay is not mounted.
The case of copying layers should be detected earlier by the
verification of upper root dir origin and index dir origin.
Both bad and stale index entries are detected and removed
on mount.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
When mounting overlayfs it needs a clean "work" directory under the
supplied workdir.
Previously the mount code removed this directory if it already existed and
created a new one. If the removal failed (e.g. directory was not empty)
then it fell back to a read-only mount not using the workdir.
While this has never been reported, it is possible to get a non-empty
"work" dir from a previous mount of overlayfs in case of crash in the
middle of an operation using the work directory.
In this case the left over state should be discarded and the overlay
filesystem will be consistent, guaranteed by the atomicity of operations on
moving to/from the workdir to the upper layer.
This patch implements cleaning out any files left in workdir. It is
implemented using real recursion for simplicity, but the depth is limited
to 2, because the worst case is that of a directory containing whiteouts
under "work".
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Cc: <stable@vger.kernel.org>
Pull vfs fixes from Al Viro:
"Followups to the parallel lookup work:
- update docs
- restore killability of the places that used to take ->i_mutex
killably now that we have down_write_killable() merged
- Additionally, it turns out that I missed a prerequisite for
security_d_instantiate() stuff - ->getxattr() wasn't the only thing
that could be called before dentry is attached to inode; with smack
we needed the same treatment applied to ->setxattr() as well"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
switch ->setxattr() to passing dentry and inode separately
switch xattr_handler->set() to passing dentry and inode separately
restore killability of old mutex_lock_killable(&inode->i_mutex) users
add down_write_killable_nested()
update D/f/directory-locking
Pull overlayfs update from Miklos Szeredi:
"The meat of this is a change to use the mounter's credentials for
operations that require elevated privileges (such as whiteout
creation). This fixes behavior under user namespaces as well as being
a nice cleanup"
* 'overlayfs-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs:
ovl: Do d_type check only if work dir creation was successful
ovl: update documentation
ovl: override creds with the ones from the superblock mounter
In user namespace the whiteout creation fails with -EPERM because the
current process isn't capable(CAP_SYS_ADMIN) when setting xattr.
A simple reproducer:
$ mkdir upper lower work merged lower/dir
$ sudo mount -t overlay overlay -olowerdir=lower,upperdir=upper,workdir=work merged
$ unshare -m -p -f -U -r bash
Now as root in the user namespace:
\# touch merged/dir/{1,2,3} # this will force a copy up of lower/dir
\# rm -fR merged/*
This ends up failing with -EPERM after the files in dir has been
correctly deleted:
unlinkat(4, "2", 0) = 0
unlinkat(4, "1", 0) = 0
unlinkat(4, "3", 0) = 0
close(4) = 0
unlinkat(AT_FDCWD, "merged/dir", AT_REMOVEDIR) = -1 EPERM (Operation not
permitted)
Interestingly, if you don't place files in merged/dir you can remove it,
meaning if upper/dir does not exist, creating the char device file works
properly in that same location.
This patch uses ovl_sb_creator_cred() to get the cred struct from the
superblock mounter and override the old cred with these new ones so that
the whiteout creation is possible because overlay is wrong in assuming that
the creds it will get with prepare_creds will be in the initial user
namespace. The old cap_raise game is removed in favor of just overriding
the old cred struct.
This patch also drops from ovl_copy_up_one() the following two lines:
override_cred->fsuid = stat->uid;
override_cred->fsgid = stat->gid;
This is because the correct uid and gid are taken directly with the stat
struct and correctly set with ovl_set_attr().
Signed-off-by: Antonio Murdaca <runcom@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
ta-da!
The main issue is the lack of down_write_killable(), so the places
like readdir.c switched to plain inode_lock(); once killable
variants of rwsem primitives appear, that'll be dealt with.
lockdep side also might need more work
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
The 'is_merge' is an historical naming from when only a single lower layer
could exist. With the introduction of multiple lower layers the meaning of
this flag was changed to mean only the "lowest layer" (while all lower
layers were being merged).
So now 'is_merge' is inaccurate and hence renaming to 'is_lowest'
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
In some instances xfs has been created with ftype=0 and there if a file
on lower fs is removed, overlay leaves a whiteout in upper fs but that
whiteout does not get filtered out and is visible to overlayfs users.
And reason it does not get filtered out because upper filesystem does
not report file type of whiteout as DT_CHR during iterate_dir().
So it seems to be a requirement that upper filesystem support d_type for
overlayfs to work properly. Do this check during mount and fail if d_type
is not supported.
Suggested-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Sargun Dhillon