Convert to struct mnt_idmap.
Last cycle we merged the necessary infrastructure in
256c8aed2b ("fs: introduce dedicated idmap type for mounts").
This is just the conversion to struct mnt_idmap.
Currently we still pass around the plain namespace that was attached to a
mount. This is in general pretty convenient but it makes it easy to
conflate namespaces that are relevant on the filesystem with namespaces
that are relevent on the mount level. Especially for non-vfs developers
without detailed knowledge in this area this can be a potential source for
bugs.
Once the conversion to struct mnt_idmap is done all helpers down to the
really low-level helpers will take a struct mnt_idmap argument instead of
two namespace arguments. This way it becomes impossible to conflate the two
eliminating the possibility of any bugs. All of the vfs and all filesystems
only operate on struct mnt_idmap.
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Convert to struct mnt_idmap.
Last cycle we merged the necessary infrastructure in
256c8aed2b ("fs: introduce dedicated idmap type for mounts").
This is just the conversion to struct mnt_idmap.
Currently we still pass around the plain namespace that was attached to a
mount. This is in general pretty convenient but it makes it easy to
conflate namespaces that are relevant on the filesystem with namespaces
that are relevent on the mount level. Especially for non-vfs developers
without detailed knowledge in this area this can be a potential source for
bugs.
Once the conversion to struct mnt_idmap is done all helpers down to the
really low-level helpers will take a struct mnt_idmap argument instead of
two namespace arguments. This way it becomes impossible to conflate the two
eliminating the possibility of any bugs. All of the vfs and all filesystems
only operate on struct mnt_idmap.
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Convert to struct mnt_idmap.
Last cycle we merged the necessary infrastructure in
256c8aed2b ("fs: introduce dedicated idmap type for mounts").
This is just the conversion to struct mnt_idmap.
Currently we still pass around the plain namespace that was attached to a
mount. This is in general pretty convenient but it makes it easy to
conflate namespaces that are relevant on the filesystem with namespaces
that are relevent on the mount level. Especially for non-vfs developers
without detailed knowledge in this area this can be a potential source for
bugs.
Once the conversion to struct mnt_idmap is done all helpers down to the
really low-level helpers will take a struct mnt_idmap argument instead of
two namespace arguments. This way it becomes impossible to conflate the two
eliminating the possibility of any bugs. All of the vfs and all filesystems
only operate on struct mnt_idmap.
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Convert to struct mnt_idmap.
Last cycle we merged the necessary infrastructure in
256c8aed2b ("fs: introduce dedicated idmap type for mounts").
This is just the conversion to struct mnt_idmap.
Currently we still pass around the plain namespace that was attached to a
mount. This is in general pretty convenient but it makes it easy to
conflate namespaces that are relevant on the filesystem with namespaces
that are relevent on the mount level. Especially for non-vfs developers
without detailed knowledge in this area this can be a potential source for
bugs.
Once the conversion to struct mnt_idmap is done all helpers down to the
really low-level helpers will take a struct mnt_idmap argument instead of
two namespace arguments. This way it becomes impossible to conflate the two
eliminating the possibility of any bugs. All of the vfs and all filesystems
only operate on struct mnt_idmap.
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Convert to struct mnt_idmap.
Last cycle we merged the necessary infrastructure in
256c8aed2b ("fs: introduce dedicated idmap type for mounts").
This is just the conversion to struct mnt_idmap.
Currently we still pass around the plain namespace that was attached to a
mount. This is in general pretty convenient but it makes it easy to
conflate namespaces that are relevant on the filesystem with namespaces
that are relevent on the mount level. Especially for non-vfs developers
without detailed knowledge in this area this can be a potential source for
bugs.
Once the conversion to struct mnt_idmap is done all helpers down to the
really low-level helpers will take a struct mnt_idmap argument instead of
two namespace arguments. This way it becomes impossible to conflate the two
eliminating the possibility of any bugs. All of the vfs and all filesystems
only operate on struct mnt_idmap.
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Convert to struct mnt_idmap.
Last cycle we merged the necessary infrastructure in
256c8aed2b ("fs: introduce dedicated idmap type for mounts").
This is just the conversion to struct mnt_idmap.
Currently we still pass around the plain namespace that was attached to a
mount. This is in general pretty convenient but it makes it easy to
conflate namespaces that are relevant on the filesystem with namespaces
that are relevent on the mount level. Especially for non-vfs developers
without detailed knowledge in this area this can be a potential source for
bugs.
Once the conversion to struct mnt_idmap is done all helpers down to the
really low-level helpers will take a struct mnt_idmap argument instead of
two namespace arguments. This way it becomes impossible to conflate the two
eliminating the possibility of any bugs. All of the vfs and all filesystems
only operate on struct mnt_idmap.
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Convert to struct mnt_idmap.
Last cycle we merged the necessary infrastructure in
256c8aed2b ("fs: introduce dedicated idmap type for mounts").
This is just the conversion to struct mnt_idmap.
Currently we still pass around the plain namespace that was attached to a
mount. This is in general pretty convenient but it makes it easy to
conflate namespaces that are relevant on the filesystem with namespaces
that are relevent on the mount level. Especially for non-vfs developers
without detailed knowledge in this area this can be a potential source for
bugs.
Once the conversion to struct mnt_idmap is done all helpers down to the
really low-level helpers will take a struct mnt_idmap argument instead of
two namespace arguments. This way it becomes impossible to conflate the two
eliminating the possibility of any bugs. All of the vfs and all filesystems
only operate on struct mnt_idmap.
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Convert to struct mnt_idmap.
Last cycle we merged the necessary infrastructure in
256c8aed2b ("fs: introduce dedicated idmap type for mounts").
This is just the conversion to struct mnt_idmap.
Currently we still pass around the plain namespace that was attached to a
mount. This is in general pretty convenient but it makes it easy to
conflate namespaces that are relevant on the filesystem with namespaces
that are relevent on the mount level. Especially for non-vfs developers
without detailed knowledge in this area this can be a potential source for
bugs.
Once the conversion to struct mnt_idmap is done all helpers down to the
really low-level helpers will take a struct mnt_idmap argument instead of
two namespace arguments. This way it becomes impossible to conflate the two
eliminating the possibility of any bugs. All of the vfs and all filesystems
only operate on struct mnt_idmap.
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Convert to struct mnt_idmap.
Last cycle we merged the necessary infrastructure in
256c8aed2b ("fs: introduce dedicated idmap type for mounts").
This is just the conversion to struct mnt_idmap.
Currently we still pass around the plain namespace that was attached to a
mount. This is in general pretty convenient but it makes it easy to
conflate namespaces that are relevant on the filesystem with namespaces
that are relevent on the mount level. Especially for non-vfs developers
without detailed knowledge in this area this can be a potential source for
bugs.
Once the conversion to struct mnt_idmap is done all helpers down to the
really low-level helpers will take a struct mnt_idmap argument instead of
two namespace arguments. This way it becomes impossible to conflate the two
eliminating the possibility of any bugs. All of the vfs and all filesystems
only operate on struct mnt_idmap.
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Convert to struct mnt_idmap.
Last cycle we merged the necessary infrastructure in
256c8aed2b ("fs: introduce dedicated idmap type for mounts").
This is just the conversion to struct mnt_idmap.
Currently we still pass around the plain namespace that was attached to a
mount. This is in general pretty convenient but it makes it easy to
conflate namespaces that are relevant on the filesystem with namespaces
that are relevent on the mount level. Especially for non-vfs developers
without detailed knowledge in this area this can be a potential source for
bugs.
Once the conversion to struct mnt_idmap is done all helpers down to the
really low-level helpers will take a struct mnt_idmap argument instead of
two namespace arguments. This way it becomes impossible to conflate the two
eliminating the possibility of any bugs. All of the vfs and all filesystems
only operate on struct mnt_idmap.
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
This patch adds POSIX_FADV_NOREUSE to vma_has_recency() so that the LRU
algorithm can ignore access to mapped files marked by this flag.
The advantages of POSIX_FADV_NOREUSE are:
1. Unlike MADV_SEQUENTIAL and MADV_RANDOM, it does not alter the
default readahead behavior.
2. Unlike MADV_SEQUENTIAL and MADV_RANDOM, it does not split VMAs and
therefore does not take mmap_lock.
3. Unlike MADV_COLD, setting it has a negligible cost, regardless of
how many pages it affects.
Its limitations are:
1. Like POSIX_FADV_RANDOM and POSIX_FADV_SEQUENTIAL, it currently does
not support range. IOW, its scope is the entire file.
2. It currently does not ignore access through file descriptors.
Specifically, for the active/inactive LRU, given a file page shared
by two users and one of them having set POSIX_FADV_NOREUSE on the
file, this page will be activated upon the second user accessing
it. This corner case can be covered by checking POSIX_FADV_NOREUSE
before calling folio_mark_accessed() on the read path. But it is
considered not worth the effort.
There have been a few attempts to support POSIX_FADV_NOREUSE, e.g., [1].
This time the goal is to fill a niche: a few desktop applications, e.g.,
large file transferring and video encoding/decoding, want fast file
streaming with mmap() rather than direct IO. Among those applications, an
SVT-AV1 regression was reported when running with MGLRU [2]. The
following test can reproduce that regression.
kb=$(awk '/MemTotal/ { print $2 }' /proc/meminfo)
kb=$((kb - 8*1024*1024))
modprobe brd rd_nr=1 rd_size=$kb
dd if=/dev/zero of=/dev/ram0 bs=1M
mkfs.ext4 /dev/ram0
mount /dev/ram0 /mnt/
swapoff -a
fallocate -l 8G /mnt/swapfile
mkswap /mnt/swapfile
swapon /mnt/swapfile
wget http://ultravideo.cs.tut.fi/video/Bosphorus_3840x2160_120fps_420_8bit_YUV_Y4M.7z
7z e -o/mnt/ Bosphorus_3840x2160_120fps_420_8bit_YUV_Y4M.7z
SvtAv1EncApp --preset 12 -w 3840 -h 2160 \
-i /mnt/Bosphorus_3840x2160.y4m
For MGLRU, the following change showed a [9-11]% increase in FPS,
which makes it on par with the active/inactive LRU.
patch Source/App/EncApp/EbAppMain.c <<EOF
31a32
> #include <fcntl.h>
35d35
< #include <fcntl.h> /* _O_BINARY */
117a118
> posix_fadvise(config->mmap.fd, 0, 0, POSIX_FADV_NOREUSE);
EOF
[1] https://lore.kernel.org/r/1308923350-7932-1-git-send-email-andrea@betterlinux.com/
[2] https://openbenchmarking.org/result/2209259-PTS-MGLRU8GB57
Link: https://lkml.kernel.org/r/20221230215252.2628425-2-yuzhao@google.com
Signed-off-by: Yu Zhao <yuzhao@google.com>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Andrea Righi <andrea.righi@canonical.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michael Larabel <Michael@MichaelLarabel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Convert to struct mnt_idmap.
Last cycle we merged the necessary infrastructure in
256c8aed2b ("fs: introduce dedicated idmap type for mounts").
This is just the conversion to struct mnt_idmap.
Currently we still pass around the plain namespace that was attached to a
mount. This is in general pretty convenient but it makes it easy to
conflate namespaces that are relevant on the filesystem with namespaces
that are relevent on the mount level. Especially for non-vfs developers
without detailed knowledge in this area this can be a potential source for
bugs.
Once the conversion to struct mnt_idmap is done all helpers down to the
really low-level helpers will take a struct mnt_idmap argument instead of
two namespace arguments. This way it becomes impossible to conflate the two
eliminating the possibility of any bugs. All of the vfs and all filesystems
only operate on struct mnt_idmap.
Acked-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
The file locking definitions have lived in fs.h since the dawn of time,
but they are only used by a small subset of the source files that
include it.
Move the file locking definitions to a new header file, and add the
appropriate #include directives to the source files that need them. By
doing this we trim down fs.h a bit and limit the amount of rebuilding
that has to be done when we make changes to the file locking APIs.
Reviewed-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: David Howells <dhowells@redhat.com>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Acked-by: Chuck Lever <chuck.lever@oracle.com>
Acked-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Acked-by: Steve French <stfrench@microsoft.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
When ceph releasing the file_lock it will try to get the inode pointer
from the fl->fl_file, which the memory could already be released by
another thread in filp_close(). Because in VFS layer the fl->fl_file
doesn't increase the file's reference counter.
Will switch to use ceph dedicate lock info to track the inode.
And in ceph_fl_release_lock() we should skip all the operations if the
fl->fl_u.ceph.inode is not set, which should come from the request
file_lock. And we will set fl->fl_u.ceph.inode when inserting it to the
inode lock list, which is when copying the lock.
Link: https://tracker.ceph.com/issues/57986
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Pull idmapping updates from Christian Brauner:
"Last cycle we've already made the interaction with idmapped mounts
more robust and type safe by introducing the vfs{g,u}id_t type. This
cycle we concluded the conversion and removed the legacy helpers.
Currently we still pass around the plain namespace that was attached
to a mount. This is in general pretty convenient but it makes it easy
to conflate namespaces that are relevant on the filesystem - with
namespaces that are relevent on the mount level. Especially for
filesystem developers without detailed knowledge in this area this can
be a potential source for bugs.
Instead of passing the plain namespace we introduce a dedicated type
struct mnt_idmap and replace the pointer with a pointer to a struct
mnt_idmap. There are no semantic or size changes for the mount struct
caused by this.
We then start converting all places aware of idmapped mounts to rely
on struct mnt_idmap. Once the conversion is done all helpers down to
the really low-level make_vfs{g,u}id() and from_vfs{g,u}id() will take
a struct mnt_idmap argument instead of two namespace arguments. This
way it becomes impossible to conflate the two removing and thus
eliminating the possibility of any bugs. Fwiw, I fixed some issues in
that area a while ago in ntfs3 and ksmbd in the past. Afterwards only
low-level code can ultimately use the associated namespace for any
permission checks. Even most of the vfs can be completely obivious
about this ultimately and filesystems will never interact with it in
any form in the future.
A struct mnt_idmap currently encompasses a simple refcount and pointer
to the relevant namespace the mount is idmapped to. If a mount isn't
idmapped then it will point to a static nop_mnt_idmap and if it
doesn't that it is idmapped. As usual there are no allocations or
anything happening for non-idmapped mounts. Everthing is carefully
written to be a nop for non-idmapped mounts as has always been the
case.
If an idmapped mount is created a struct mnt_idmap is allocated and a
reference taken on the relevant namespace. Each mount that gets
idmapped or inherits the idmap simply bumps the reference count on
struct mnt_idmap. Just a reminder that we only allow a mount to change
it's idmapping a single time and only if it hasn't already been
attached to the filesystems and has no active writers.
The actual changes are fairly straightforward but this will have huge
benefits for maintenance and security in the long run even if it
causes some churn.
Note that this also makes it possible to extend struct mount_idmap in
the future. For example, it would be possible to place the namespace
pointer in an anonymous union together with an idmapping struct. This
would allow us to expose an api to userspace that would let it specify
idmappings directly instead of having to go through the detour of
setting up namespaces at all"
* tag 'fs.idmapped.mnt_idmap.v6.2' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping:
acl: conver higher-level helpers to rely on mnt_idmap
fs: introduce dedicated idmap type for mounts
Pull vfsuid updates from Christian Brauner:
"Last cycle we introduced the vfs{g,u}id_t types and associated helpers
to gain type safety when dealing with idmapped mounts. That initial
work already converted a lot of places over but there were still some
left,
This converts all remaining places that still make use of non-type
safe idmapping helpers to rely on the new type safe vfs{g,u}id based
helpers.
Afterwards it removes all the old non-type safe helpers"
* tag 'fs.vfsuid.conversion.v6.2' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping:
fs: remove unused idmapping helpers
ovl: port to vfs{g,u}id_t and associated helpers
fuse: port to vfs{g,u}id_t and associated helpers
ima: use type safe idmapping helpers
apparmor: use type safe idmapping helpers
caps: use type safe idmapping helpers
fs: use type safe idmapping helpers
mnt_idmapping: add missing helpers
Pull setgid inheritance updates from Christian Brauner:
"This contains the work to make setgid inheritance consistent between
modifying a file and when changing ownership or mode as this has been
a repeated source of very subtle bugs. The gist is that we perform the
same permission checks in the write path as we do in the ownership and
mode changing paths after this series where we're currently doing
different things.
We've already made setgid inheritance a lot more consistent and
reliable in the last releases by moving setgid stripping from the
individual filesystems up into the vfs. This aims to make the logic
even more consistent and easier to understand and also to fix
long-standing overlayfs setgid inheritance bugs. Miklos was nice
enough to just let me carry the trivial overlayfs patches from Amir
too.
Below is a more detailed explanation how the current difference in
setgid handling lead to very subtle bugs exemplified via overlayfs
which is a victim of the current rules. I hope this explains why I
think taking the regression risk here is worth it.
A long while ago I found a few setgid inheritance bugs in overlayfs in
the write path in certain conditions. Amir recently picked this back
up in [1] and I jumped on board to fix this more generally.
On the surface all that overlayfs would need to fix setgid inheritance
would be to call file_remove_privs() or file_modified() but actually
that isn't enough because the setgid inheritance api is wildly
inconsistent in that area.
Before this pr setgid stripping in file_remove_privs()'s old
should_remove_suid() helper was inconsistent with other parts of the
vfs. Specifically, it only raises ATTR_KILL_SGID if the inode is
S_ISGID and S_IXGRP but not if the inode isn't in the caller's groups
and the caller isn't privileged over the inode although we require
this already in setattr_prepare() and setattr_copy() and so all
filesystem implement this requirement implicitly because they have to
use setattr_{prepare,copy}() anyway.
But the inconsistency shows up in setgid stripping bugs for overlayfs
in xfstests (e.g., generic/673, generic/683, generic/685, generic/686,
generic/687). For example, we test whether suid and setgid stripping
works correctly when performing various write-like operations as an
unprivileged user (fallocate, reflink, write, etc.):
echo "Test 1 - qa_user, non-exec file $verb"
setup_testfile
chmod a+rws $junk_file
commit_and_check "$qa_user" "$verb" 64k 64k
The test basically creates a file with 6666 permissions. While the
file has the S_ISUID and S_ISGID bits set it does not have the S_IXGRP
set.
On a regular filesystem like xfs what will happen is:
sys_fallocate()
-> vfs_fallocate()
-> xfs_file_fallocate()
-> file_modified()
-> __file_remove_privs()
-> dentry_needs_remove_privs()
-> should_remove_suid()
-> __remove_privs()
newattrs.ia_valid = ATTR_FORCE | kill;
-> notify_change()
-> setattr_copy()
In should_remove_suid() we can see that ATTR_KILL_SUID is raised
unconditionally because the file in the test has S_ISUID set.
But we also see that ATTR_KILL_SGID won't be set because while the
file is S_ISGID it is not S_IXGRP (see above) which is a condition for
ATTR_KILL_SGID being raised.
So by the time we call notify_change() we have attr->ia_valid set to
ATTR_KILL_SUID | ATTR_FORCE.
Now notify_change() sees that ATTR_KILL_SUID is set and does:
ia_valid = attr->ia_valid |= ATTR_MODE
attr->ia_mode = (inode->i_mode & ~S_ISUID);
which means that when we call setattr_copy() later we will definitely
update inode->i_mode. Note that attr->ia_mode still contains S_ISGID.
Now we call into the filesystem's ->setattr() inode operation which
will end up calling setattr_copy(). Since ATTR_MODE is set we will
hit:
if (ia_valid & ATTR_MODE) {
umode_t mode = attr->ia_mode;
vfsgid_t vfsgid = i_gid_into_vfsgid(mnt_userns, inode);
if (!vfsgid_in_group_p(vfsgid) &&
!capable_wrt_inode_uidgid(mnt_userns, inode, CAP_FSETID))
mode &= ~S_ISGID;
inode->i_mode = mode;
}
and since the caller in the test is neither capable nor in the group
of the inode the S_ISGID bit is stripped.
But assume the file isn't suid then ATTR_KILL_SUID won't be raised
which has the consequence that neither the setgid nor the suid bits
are stripped even though it should be stripped because the inode isn't
in the caller's groups and the caller isn't privileged over the inode.
If overlayfs is in the mix things become a bit more complicated and
the bug shows up more clearly.
When e.g., ovl_setattr() is hit from ovl_fallocate()'s call to
file_remove_privs() then ATTR_KILL_SUID and ATTR_KILL_SGID might be
raised but because the check in notify_change() is questioning the
ATTR_KILL_SGID flag again by requiring S_IXGRP for it to be stripped
the S_ISGID bit isn't removed even though it should be stripped:
sys_fallocate()
-> vfs_fallocate()
-> ovl_fallocate()
-> file_remove_privs()
-> dentry_needs_remove_privs()
-> should_remove_suid()
-> __remove_privs()
newattrs.ia_valid = ATTR_FORCE | kill;
-> notify_change()
-> ovl_setattr()
/* TAKE ON MOUNTER'S CREDS */
-> ovl_do_notify_change()
-> notify_change()
/* GIVE UP MOUNTER'S CREDS */
/* TAKE ON MOUNTER'S CREDS */
-> vfs_fallocate()
-> xfs_file_fallocate()
-> file_modified()
-> __file_remove_privs()
-> dentry_needs_remove_privs()
-> should_remove_suid()
-> __remove_privs()
newattrs.ia_valid = attr_force | kill;
-> notify_change()
The fix for all of this is to make file_remove_privs()'s
should_remove_suid() helper perform the same checks as we already
require in setattr_prepare() and setattr_copy() and have
notify_change() not pointlessly requiring S_IXGRP again. It doesn't
make any sense in the first place because the caller must calculate
the flags via should_remove_suid() anyway which would raise
ATTR_KILL_SGID
Note that some xfstests will now fail as these patches will cause the
setgid bit to be lost in certain conditions for unprivileged users
modifying a setgid file when they would've been kept otherwise. I
think this risk is worth taking and I explained and mentioned this
multiple times on the list [2].
Enforcing the rules consistently across write operations and
chmod/chown will lead to losing the setgid bit in cases were it
might've been retained before.
While I've mentioned this a few times but it's worth repeating just to
make sure that this is understood. For the sake of maintainability,
consistency, and security this is a risk worth taking.
If we really see regressions for workloads the fix is to have special
setgid handling in the write path again with different semantics from
chmod/chown and possibly additional duct tape for overlayfs. I'll
update the relevant xfstests with if you should decide to merge this
second setgid cleanup.
Before that people should be aware that there might be failures for
fstests where unprivileged users modify a setgid file"
Link: https://lore.kernel.org/linux-fsdevel/20221003123040.900827-1-amir73il@gmail.com [1]
Link: https://lore.kernel.org/linux-fsdevel/20221122142010.zchf2jz2oymx55qi@wittgenstein [2]
* tag 'fs.ovl.setgid.v6.2' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping:
fs: use consistent setgid checks in is_sxid()
ovl: remove privs in ovl_fallocate()
ovl: remove privs in ovl_copyfile()
attr: use consistent sgid stripping checks
attr: add setattr_should_drop_sgid()
fs: move should_remove_suid()
attr: add in_group_or_capable()
Pull VFS acl updates from Christian Brauner:
"This contains the work that builds a dedicated vfs posix acl api.
The origins of this work trace back to v5.19 but it took quite a while
to understand the various filesystem specific implementations in
sufficient detail and also come up with an acceptable solution.
As we discussed and seen multiple times the current state of how posix
acls are handled isn't nice and comes with a lot of problems: The
current way of handling posix acls via the generic xattr api is error
prone, hard to maintain, and type unsafe for the vfs until we call
into the filesystem's dedicated get and set inode operations.
It is already the case that posix acls are special-cased to death all
the way through the vfs. There are an uncounted number of hacks that
operate on the uapi posix acl struct instead of the dedicated vfs
struct posix_acl. And the vfs must be involved in order to interpret
and fixup posix acls before storing them to the backing store, caching
them, reporting them to userspace, or for permission checking.
Currently a range of hacks and duct tape exist to make this work. As
with most things this is really no ones fault it's just something that
happened over time. But the code is hard to understand and difficult
to maintain and one is constantly at risk of introducing bugs and
regressions when having to touch it.
Instead of continuing to hack posix acls through the xattr handlers
this series builds a dedicated posix acl api solely around the get and
set inode operations.
Going forward, the vfs_get_acl(), vfs_remove_acl(), and vfs_set_acl()
helpers must be used in order to interact with posix acls. They
operate directly on the vfs internal struct posix_acl instead of
abusing the uapi posix acl struct as we currently do. In the end this
removes all of the hackiness, makes the codepaths easier to maintain,
and gets us type safety.
This series passes the LTP and xfstests suites without any
regressions. For xfstests the following combinations were tested:
- xfs
- ext4
- btrfs
- overlayfs
- overlayfs on top of idmapped mounts
- orangefs
- (limited) cifs
There's more simplifications for posix acls that we can make in the
future if the basic api has made it.
A few implementation details:
- The series makes sure to retain exactly the same security and
integrity module permission checks. Especially for the integrity
modules this api is a win because right now they convert the uapi
posix acl struct passed to them via a void pointer into the vfs
struct posix_acl format to perform permission checking on the mode.
There's a new dedicated security hook for setting posix acls which
passes the vfs struct posix_acl not a void pointer. Basing checking
on the posix acl stored in the uapi format is really unreliable.
The vfs currently hacks around directly in the uapi struct storing
values that frankly the security and integrity modules can't
correctly interpret as evidenced by bugs we reported and fixed in
this area. It's not necessarily even their fault it's just that the
format we provide to them is sub optimal.
- Some filesystems like 9p and cifs need access to the dentry in
order to get and set posix acls which is why they either only
partially or not even at all implement get and set inode
operations. For example, cifs allows setxattr() and getxattr()
operations but doesn't allow permission checking based on posix
acls because it can't implement a get acl inode operation.
Thus, this patch series updates the set acl inode operation to take
a dentry instead of an inode argument. However, for the get acl
inode operation we can't do this as the old get acl method is
called in e.g., generic_permission() and inode_permission(). These
helpers in turn are called in various filesystem's permission inode
operation. So passing a dentry argument to the old get acl inode
operation would amount to passing a dentry to the permission inode
operation which we shouldn't and probably can't do.
So instead of extending the existing inode operation Christoph
suggested to add a new one. He also requested to ensure that the
get and set acl inode operation taking a dentry are consistently
named. So for this version the old get acl operation is renamed to
->get_inode_acl() and a new ->get_acl() inode operation taking a
dentry is added. With this we can give both 9p and cifs get and set
acl inode operations and in turn remove their complex custom posix
xattr handlers.
In the future I hope to get rid of the inode method duplication but
it isn't like we have never had this situation. Readdir is just one
example. And frankly, the overall gain in type safety and the more
pleasant api wise are simply too big of a benefit to not accept
this duplication for a while.
- We've done a full audit of every codepaths using variant of the
current generic xattr api to get and set posix acls and
surprisingly it isn't that many places. There's of course always a
chance that we might have missed some and if so I'm sure we'll find
them soon enough.
The crucial codepaths to be converted are obviously stacking
filesystems such as ecryptfs and overlayfs.
For a list of all callers currently using generic xattr api helpers
see [2] including comments whether they support posix acls or not.
- The old vfs generic posix acl infrastructure doesn't obey the
create and replace semantics promised on the setxattr(2) manpage.
This patch series doesn't address this. It really is something we
should revisit later though.
The patches are roughly organized as follows:
(1) Change existing set acl inode operation to take a dentry
argument (Intended to be a non-functional change)
(2) Rename existing get acl method (Intended to be a non-functional
change)
(3) Implement get and set acl inode operations for filesystems that
couldn't implement one before because of the missing dentry.
That's mostly 9p and cifs (Intended to be a non-functional
change)
(4) Build posix acl api, i.e., add vfs_get_acl(), vfs_remove_acl(),
and vfs_set_acl() including security and integrity hooks
(Intended to be a non-functional change)
(5) Implement get and set acl inode operations for stacking
filesystems (Intended to be a non-functional change)
(6) Switch posix acl handling in stacking filesystems to new posix
acl api now that all filesystems it can stack upon support it.
(7) Switch vfs to new posix acl api (semantical change)
(8) Remove all now unused helpers
(9) Additional regression fixes reported after we merged this into
linux-next
Thanks to Seth for a lot of good discussion around this and
encouragement and input from Christoph"
* tag 'fs.acl.rework.v6.2' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/idmapping: (36 commits)
posix_acl: Fix the type of sentinel in get_acl
orangefs: fix mode handling
ovl: call posix_acl_release() after error checking
evm: remove dead code in evm_inode_set_acl()
cifs: check whether acl is valid early
acl: make vfs_posix_acl_to_xattr() static
acl: remove a slew of now unused helpers
9p: use stub posix acl handlers
cifs: use stub posix acl handlers
ovl: use stub posix acl handlers
ecryptfs: use stub posix acl handlers
evm: remove evm_xattr_acl_change()
xattr: use posix acl api
ovl: use posix acl api
ovl: implement set acl method
ovl: implement get acl method
ecryptfs: implement set acl method
ecryptfs: implement get acl method
ksmbd: use vfs_remove_acl()
acl: add vfs_remove_acl()
...
Pull misc vfs updates from Al Viro:
"misc pile"
* tag 'pull-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
fs: sysv: Fix sysv_nblocks() returns wrong value
get rid of INT_LIMIT, use type_max() instead
btrfs: replace INT_LIMIT(loff_t) with OFFSET_MAX
fs: simplify vfs_get_super
fs: drop useless condition from inode_needs_update_time
Pull non-MM updates from Andrew Morton:
- A ptrace API cleanup series from Sergey Shtylyov
- Fixes and cleanups for kexec from ye xingchen
- nilfs2 updates from Ryusuke Konishi
- squashfs feature work from Xiaoming Ni: permit configuration of the
filesystem's compression concurrency from the mount command line
- A series from Akinobu Mita which addresses bound checking errors when
writing to debugfs files
- A series from Yang Yingliang to address rapidio memory leaks
- A series from Zheng Yejian to address possible overflow errors in
encode_comp_t()
- And a whole shower of singleton patches all over the place
* tag 'mm-nonmm-stable-2022-12-12' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm: (79 commits)
ipc: fix memory leak in init_mqueue_fs()
hfsplus: fix bug causing custom uid and gid being unable to be assigned with mount
rapidio: devices: fix missing put_device in mport_cdev_open
kcov: fix spelling typos in comments
hfs: Fix OOB Write in hfs_asc2mac
hfs: fix OOB Read in __hfs_brec_find
relay: fix type mismatch when allocating memory in relay_create_buf()
ocfs2: always read both high and low parts of dinode link count
io-mapping: move some code within the include guarded section
kernel: kcsan: kcsan_test: build without structleak plugin
mailmap: update email for Iskren Chernev
eventfd: change int to __u64 in eventfd_signal() ifndef CONFIG_EVENTFD
rapidio: fix possible UAF when kfifo_alloc() fails
relay: use strscpy() is more robust and safer
cpumask: limit visibility of FORCE_NR_CPUS
acct: fix potential integer overflow in encode_comp_t()
acct: fix accuracy loss for input value of encode_comp_t()
linux/init.h: include <linux/build_bug.h> and <linux/stringify.h>
rapidio: rio: fix possible name leak in rio_register_mport()
rapidio: fix possible name leaks when rio_add_device() fails
...
Pull file locking updates from Jeff Layton:
"The main change here is to add the new locks_inode_context helper, and
convert all of the places that dereference inode->i_flctx directly to
use that instead.
There is a new helper to indicate whether any locks are held on an
inode. This is mostly for Ceph but may be usable elsewhere too.
Andi Kleen requested that we print the PID when the LOCK_MAND warning
fires, to help track down applications trying to use it.
Finally, we added some new warnings to some of the file locking
functions that fire when the ->fl_file and filp arguments differ. This
helped us find some long-standing bugs in lockd. Patches for those are
in Chuck Lever's tree and should be in his v6.2 PR. After that patch,
people using NFSv2/v3 locking may see some warnings fire until those
go in.
Happy Holidays!"
* tag 'locks-v6.2' of git://git.kernel.org/pub/scm/linux/kernel/git/jlayton/linux:
Add process name and pid to locks warning
nfsd: use locks_inode_context helper
nfs: use locks_inode_context helper
lockd: use locks_inode_context helper
ksmbd: use locks_inode_context helper
cifs: use locks_inode_context helper
ceph: use locks_inode_context helper
filelock: add a new locks_inode_context accessor function
filelock: new helper: vfs_inode_has_locks
filelock: WARN_ON_ONCE when ->fl_file and filp don't match
There are a number of places in the kernel that are accessing the
inode->i_flctx field without smp_load_acquire. This is required to
ensure that the caller doesn't see a partially-initialized structure.
Add a new accessor function for it to make this clear and convert all of
the relevant accesses in locks.c to use it. Also, convert
locks_free_lock_context to use the helper as well instead of just doing
a "bare" assignment.
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Ceph has a need to know whether a particular inode has any locks set on
it. It's currently tracking that by a num_locks field in its
filp->private_data, but that's problematic as it tries to decrement this
field when releasing locks and that can race with the file being torn
down.
Add a new vfs_inode_has_locks helper that just returns whether any locks
are currently held on the inode.
Reviewed-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Christoph Hellwig <hch@infradead.org>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
INT_LIMIT() tries to do what type_max() does, except that type_max()
doesn't rely upon undefined behaviour[*], might as well use type_max()
instead.
[*] if T is an N-bit signed integer type, the maximal value in T is
pow(2, N - 1) - 1, all right, but naive expression for that value
ends up with a couple of wraparounds and as usual for wraparounds
in signed types, that's an undefined behaviour. type_max() takes
care to avoid those...
Caught-by: UBSAN
Suggested-by: Eric Biggers <ebiggers@kernel.org>
Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Commit 868f9f2f8e ("vfs: fix copy_file_range() regression in cross-fs
copies") removed fallback to generic_copy_file_range() for cross-fs
cases inside vfs_copy_file_range().
To preserve behavior of nfsd and ksmbd server-side-copy, the fallback to
generic_copy_file_range() was added in nfsd and ksmbd code, but that
call is missing sb_start_write(), fsnotify hooks and more.
Ideally, nfsd and ksmbd would pass a flag to vfs_copy_file_range() that
will take care of the fallback, but that code would be subtle and we got
vfs_copy_file_range() logic wrong too many times already.
Instead, add a flag to explicitly request vfs_copy_file_range() to
perform only generic_copy_file_range() and let nfsd and ksmbd use this
flag only in the fallback path.
This choise keeps the logic changes to minimum in the non-nfsd/ksmbd code
paths to reduce the risk of further regressions.
Fixes: 868f9f2f8e ("vfs: fix copy_file_range() regression in cross-fs copies")
Tested-by: Namjae Jeon <linkinjeon@kernel.org>
Tested-by: Luis Henriques <lhenriques@suse.de>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Now that we made the VFS setgid checking consistent an inode can't be
marked security irrelevant even if the setgid bit is still set. Make
this function consistent with all other helpers.
Note that enforcing consistent setgid stripping checks for file
modification and mode- and ownership changes will cause the setgid bit
to be lost in more cases than useed to be the case. If an unprivileged
user wrote to a non-executable setgid file that they don't have
privilege over the setgid bit will be dropped. This will lead to
temporary failures in some xfstests until they have been updated.
Reported-by: Miklos Szeredi <miklos@szeredi.hu>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Last cycle we've already made the interaction with idmapped mounts more
robust and type safe by introducing the vfs{g,u}id_t type. This cycle we
concluded the conversion and removed the legacy helpers.
Currently we still pass around the plain namespace that was attached to
a mount. This is in general pretty convenient but it makes it easy to
conflate filesystem and mount namespaces and what different roles they
have to play. Especially for filesystem developers without much
experience in this area this is an easy source for bugs.
Instead of passing the plain namespace we introduce a dedicated type
struct mnt_idmap and replace the pointer with a pointer to a struct
mnt_idmap. There are no semantic or size changes for the mount struct
caused by this.
We then start converting all places aware of idmapped mounts to rely on
struct mnt_idmap. Once the conversion is done all helpers down to the
really low-level make_vfs{g,u}id() and from_vfs{g,u}id() will take a
struct mnt_idmap argument instead of two namespace arguments. This way
it becomes impossible to conflate the two, removing and thus eliminating
the possibility of any bugs. Fwiw, I fixed some issues in that area a
while ago in ntfs3 and ksmbd in the past. Afterwards, only low-level
code can ultimately use the associated namespace for any permission
checks. Even most of the vfs can be ultimately completely oblivious
about this and filesystems will never interact with it directly in any
form in the future.
A struct mnt_idmap currently encompasses a simple refcount and a pointer
to the relevant namespace the mount is idmapped to. If a mount isn't
idmapped then it will point to a static nop_mnt_idmap. If it is an
idmapped mount it will point to a new struct mnt_idmap. As usual there
are no allocations or anything happening for non-idmapped mounts.
Everthing is carefully written to be a nop for non-idmapped mounts as
has always been the case.
If an idmapped mount or mount tree is created a new struct mnt_idmap is
allocated and a reference taken on the relevant namespace. For each
mount in a mount tree that gets idmapped or a mount that inherits the
idmap when it is cloned the reference count on the associated struct
mnt_idmap is bumped. Just a reminder that we only allow a mount to
change it's idmapping a single time and only if it hasn't already been
attached to the filesystems and has no active writers.
The actual changes are fairly straightforward. This will have huge
benefits for maintenance and security in the long run even if it causes
some churn. I'm aware that there's some cost for all of you. And I'll
commit to doing this work and make this as painless as I can.
Note that this also makes it possible to extend struct mount_idmap in
the future. For example, it would be possible to place the namespace
pointer in an anonymous union together with an idmapping struct. This
would allow us to expose an api to userspace that would let it specify
idmappings directly instead of having to go through the detour of
setting up namespaces at all.
This just adds the infrastructure and doesn't do any conversions.
Reviewed-by: Seth Forshee (DigitalOcean) <sforshee@kernel.org>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Now that all places can deal with the new type safe helpers remove all
of the old helpers.
Reviewed-by: Seth Forshee (DigitalOcean) <sforshee@kernel.org>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
The current way of setting and getting posix acls through the generic
xattr interface is error prone and type unsafe. The vfs needs to
interpret and fixup posix acls before storing or reporting it to
userspace. Various hacks exist to make this work. The code is hard to
understand and difficult to maintain in it's current form. Instead of
making this work by hacking posix acls through xattr handlers we are
building a dedicated posix acl api around the get and set inode
operations. This removes a lot of hackiness and makes the codepaths
easier to maintain. A lot of background can be found in [1].
Since some filesystem rely on the dentry being available to them when
setting posix acls (e.g., 9p and cifs) they cannot rely on the old get
acl inode operation to retrieve posix acl and need to implement their
own custom handlers because of that.
In a previous patch we renamed the old get acl inode operation to
->get_inode_acl(). We decided to rename it and implement a new one since
->get_inode_acl() is called generic_permission() and inode_permission()
both of which can be called during an filesystem's ->permission()
handler. So simply passing a dentry argument to ->get_acl() would have
amounted to also having to pass a dentry argument to ->permission(). We
avoided that change.
This adds a new ->get_acl() inode operations which takes a dentry
argument which filesystems such as 9p, cifs, and overlayfs can implement
to get posix acls.
Link: https://lore.kernel.org/all/20220801145520.1532837-1-brauner@kernel.org [1]
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
The current way of setting and getting posix acls through the generic
xattr interface is error prone and type unsafe. The vfs needs to
interpret and fixup posix acls before storing or reporting it to
userspace. Various hacks exist to make this work. The code is hard to
understand and difficult to maintain in it's current form. Instead of
making this work by hacking posix acls through xattr handlers we are
building a dedicated posix acl api around the get and set inode
operations. This removes a lot of hackiness and makes the codepaths
easier to maintain. A lot of background can be found in [1].
The current inode operation for getting posix acls takes an inode
argument but various filesystems (e.g., 9p, cifs, overlayfs) need access
to the dentry. In contrast to the ->set_acl() inode operation we cannot
simply extend ->get_acl() to take a dentry argument. The ->get_acl()
inode operation is called from:
acl_permission_check()
-> check_acl()
-> get_acl()
which is part of generic_permission() which in turn is part of
inode_permission(). Both generic_permission() and inode_permission() are
called in the ->permission() handler of various filesystems (e.g.,
overlayfs). So simply passing a dentry argument to ->get_acl() would
amount to also having to pass a dentry argument to ->permission(). We
should avoid this unnecessary change.
So instead of extending the existing inode operation rename it from
->get_acl() to ->get_inode_acl() and add a ->get_acl() method later that
passes a dentry argument and which filesystems that need access to the
dentry can implement instead of ->get_inode_acl(). Filesystems like cifs
which allow setting and getting posix acls but not using them for
permission checking during lookup can simply not implement
->get_inode_acl().
This is intended to be a non-functional change.
Link: https://lore.kernel.org/all/20220801145520.1532837-1-brauner@kernel.org [1]
Suggested-by/Inspired-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
The current way of setting and getting posix acls through the generic
xattr interface is error prone and type unsafe. The vfs needs to
interpret and fixup posix acls before storing or reporting it to
userspace. Various hacks exist to make this work. The code is hard to
understand and difficult to maintain in it's current form. Instead of
making this work by hacking posix acls through xattr handlers we are
building a dedicated posix acl api around the get and set inode
operations. This removes a lot of hackiness and makes the codepaths
easier to maintain. A lot of background can be found in [1].
Since some filesystem rely on the dentry being available to them when
setting posix acls (e.g., 9p and cifs) they cannot rely on set acl inode
operation. But since ->set_acl() is required in order to use the generic
posix acl xattr handlers filesystems that do not implement this inode
operation cannot use the handler and need to implement their own
dedicated posix acl handlers.
Update the ->set_acl() inode method to take a dentry argument. This
allows all filesystems to rely on ->set_acl().
As far as I can tell all codepaths can be switched to rely on the dentry
instead of just the inode. Note that the original motivation for passing
the dentry separate from the inode instead of just the dentry in the
xattr handlers was because of security modules that call
security_d_instantiate(). This hook is called during
d_instantiate_new(), d_add(), __d_instantiate_anon(), and
d_splice_alias() to initialize the inode's security context and possibly
to set security.* xattrs. Since this only affects security.* xattrs this
is completely irrelevant for posix acls.
Link: https://lore.kernel.org/all/20220801145520.1532837-1-brauner@kernel.org [1]
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Currently setgid stripping in file_remove_privs()'s should_remove_suid()
helper is inconsistent with other parts of the vfs. Specifically, it only
raises ATTR_KILL_SGID if the inode is S_ISGID and S_IXGRP but not if the
inode isn't in the caller's groups and the caller isn't privileged over the
inode although we require this already in setattr_prepare() and
setattr_copy() and so all filesystem implement this requirement implicitly
because they have to use setattr_{prepare,copy}() anyway.
But the inconsistency shows up in setgid stripping bugs for overlayfs in
xfstests (e.g., generic/673, generic/683, generic/685, generic/686,
generic/687). For example, we test whether suid and setgid stripping works
correctly when performing various write-like operations as an unprivileged
user (fallocate, reflink, write, etc.):
echo "Test 1 - qa_user, non-exec file $verb"
setup_testfile
chmod a+rws $junk_file
commit_and_check "$qa_user" "$verb" 64k 64k
The test basically creates a file with 6666 permissions. While the file has
the S_ISUID and S_ISGID bits set it does not have the S_IXGRP set. On a
regular filesystem like xfs what will happen is:
sys_fallocate()
-> vfs_fallocate()
-> xfs_file_fallocate()
-> file_modified()
-> __file_remove_privs()
-> dentry_needs_remove_privs()
-> should_remove_suid()
-> __remove_privs()
newattrs.ia_valid = ATTR_FORCE | kill;
-> notify_change()
-> setattr_copy()
In should_remove_suid() we can see that ATTR_KILL_SUID is raised
unconditionally because the file in the test has S_ISUID set.
But we also see that ATTR_KILL_SGID won't be set because while the file
is S_ISGID it is not S_IXGRP (see above) which is a condition for
ATTR_KILL_SGID being raised.
So by the time we call notify_change() we have attr->ia_valid set to
ATTR_KILL_SUID | ATTR_FORCE. Now notify_change() sees that
ATTR_KILL_SUID is set and does:
ia_valid = attr->ia_valid |= ATTR_MODE
attr->ia_mode = (inode->i_mode & ~S_ISUID);
which means that when we call setattr_copy() later we will definitely
update inode->i_mode. Note that attr->ia_mode still contains S_ISGID.
Now we call into the filesystem's ->setattr() inode operation which will
end up calling setattr_copy(). Since ATTR_MODE is set we will hit:
if (ia_valid & ATTR_MODE) {
umode_t mode = attr->ia_mode;
vfsgid_t vfsgid = i_gid_into_vfsgid(mnt_userns, inode);
if (!vfsgid_in_group_p(vfsgid) &&
!capable_wrt_inode_uidgid(mnt_userns, inode, CAP_FSETID))
mode &= ~S_ISGID;
inode->i_mode = mode;
}
and since the caller in the test is neither capable nor in the group of the
inode the S_ISGID bit is stripped.
But assume the file isn't suid then ATTR_KILL_SUID won't be raised which
has the consequence that neither the setgid nor the suid bits are stripped
even though it should be stripped because the inode isn't in the caller's
groups and the caller isn't privileged over the inode.
If overlayfs is in the mix things become a bit more complicated and the bug
shows up more clearly. When e.g., ovl_setattr() is hit from
ovl_fallocate()'s call to file_remove_privs() then ATTR_KILL_SUID and
ATTR_KILL_SGID might be raised but because the check in notify_change() is
questioning the ATTR_KILL_SGID flag again by requiring S_IXGRP for it to be
stripped the S_ISGID bit isn't removed even though it should be stripped:
sys_fallocate()
-> vfs_fallocate()
-> ovl_fallocate()
-> file_remove_privs()
-> dentry_needs_remove_privs()
-> should_remove_suid()
-> __remove_privs()
newattrs.ia_valid = ATTR_FORCE | kill;
-> notify_change()
-> ovl_setattr()
// TAKE ON MOUNTER'S CREDS
-> ovl_do_notify_change()
-> notify_change()
// GIVE UP MOUNTER'S CREDS
// TAKE ON MOUNTER'S CREDS
-> vfs_fallocate()
-> xfs_file_fallocate()
-> file_modified()
-> __file_remove_privs()
-> dentry_needs_remove_privs()
-> should_remove_suid()
-> __remove_privs()
newattrs.ia_valid = attr_force | kill;
-> notify_change()
The fix for all of this is to make file_remove_privs()'s
should_remove_suid() helper to perform the same checks as we already
require in setattr_prepare() and setattr_copy() and have notify_change()
not pointlessly requiring S_IXGRP again. It doesn't make any sense in the
first place because the caller must calculate the flags via
should_remove_suid() anyway which would raise ATTR_KILL_SGID.
While we're at it we move should_remove_suid() from inode.c to attr.c
where it belongs with the rest of the iattr helpers. Especially since it
returns ATTR_KILL_S{G,U}ID flags. We also rename it to
setattr_should_drop_suidgid() to better reflect that it indicates both
setuid and setgid bit removal and also that it returns attr flags.
Running xfstests with this doesn't report any regressions. We should really
try and use consistent checks.
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Pull vfs tmpfile updates from Al Viro:
"Miklos' ->tmpfile() signature change; pass an unopened struct file to
it, let it open the damn thing. Allows to add tmpfile support to FUSE"
* tag 'pull-tmpfile' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
fuse: implement ->tmpfile()
vfs: open inside ->tmpfile()
vfs: move open right after ->tmpfile()
vfs: make vfs_tmpfile() static
ovl: use vfs_tmpfile_open() helper
cachefiles: use vfs_tmpfile_open() helper
cachefiles: only pass inode to *mark_inode_inuse() helpers
cachefiles: tmpfile error handling cleanup
hugetlbfs: cleanup mknod and tmpfile
vfs: add vfs_tmpfile_open() helper
Pull io_uring updates from Jens Axboe:
- Add supported for more directly managed task_work running.
This is beneficial for real world applications that end up issuing
lots of system calls as part of handling work. Normal task_work will
always execute as we transition in and out of the kernel, even for
"unrelated" system calls. It's more efficient to defer the handling
of io_uring's deferred work until the application wants it to be run,
generally in batches.
As part of ongoing work to write an io_uring network backend for
Thrift, this has been shown to greatly improve performance. (Dylan)
- Add IOPOLL support for passthrough (Kanchan)
- Improvements and fixes to the send zero-copy support (Pavel)
- Partial IO handling fixes (Pavel)
- CQE ordering fixes around CQ ring overflow (Pavel)
- Support sendto() for non-zc as well (Pavel)
- Support sendmsg for zerocopy (Pavel)
- Networking iov_iter fix (Stefan)
- Misc fixes and cleanups (Pavel, me)
* tag 'for-6.1/io_uring-2022-10-03' of git://git.kernel.dk/linux: (56 commits)
io_uring/net: fix notif cqe reordering
io_uring/net: don't update msg_name if not provided
io_uring: don't gate task_work run on TIF_NOTIFY_SIGNAL
io_uring/rw: defer fsnotify calls to task context
io_uring/net: fix fast_iov assignment in io_setup_async_msg()
io_uring/net: fix non-zc send with address
io_uring/net: don't skip notifs for failed requests
io_uring/rw: don't lose short results on io_setup_async_rw()
io_uring/rw: fix unexpected link breakage
io_uring/net: fix cleanup double free free_iov init
io_uring: fix CQE reordering
io_uring/net: fix UAF in io_sendrecv_fail()
selftest/net: adjust io_uring sendzc notif handling
io_uring: ensure local task_work marks task as running
io_uring/net: zerocopy sendmsg
io_uring/net: combine fail handlers
io_uring/net: rename io_sendzc()
io_uring/net: support non-zerocopy sendto
io_uring/net: refactor io_setup_async_addr
io_uring/net: don't lose partial send_zc on fail
...
Pull ext4 updates from Ted Ts'o:
"The first two changes involve files outside of fs/ext4:
- submit_bh() can never return an error, so change it to return void,
and remove the unused checks from its callers
- fix I_DIRTY_TIME handling so it will be set even if the inode
already has I_DIRTY_INODE
Performance:
- Always enable i_version counter (as btrfs and xfs already do).
Remove some uneeded i_version bumps to avoid unnecessary nfs cache
invalidations
- Wake up journal waiters in FIFO order, to avoid some journal users
from not getting a journal handle for an unfairly long time
- In ext4_write_begin() allocate any necessary buffer heads before
starting the journal handle
- Don't try to prefetch the block allocation bitmaps for a read-only
file system
Bug Fixes:
- Fix a number of fast commit bugs, including resources leaks and out
of bound references in various error handling paths and/or if the
fast commit log is corrupted
- Avoid stopping the online resize early when expanding a file system
which is less than 16TiB to a size greater than 16TiB
- Fix apparent metadata corruption caused by a race with a metadata
buffer head getting migrated while it was trying to be read
- Mark the lazy initialization thread freezable to prevent suspend
failures
- Other miscellaneous bug fixes
Cleanups:
- Break up the incredibly long ext4_full_super() function by
refactoring to move code into more understandable, smaller
functions
- Remove the deprecated (and ignored) noacl and nouser_attr mount
option
- Factor out some common code in fast commit handling
- Other miscellaneous cleanups"
* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4: (53 commits)
ext4: fix potential out of bound read in ext4_fc_replay_scan()
ext4: factor out ext4_fc_get_tl()
ext4: introduce EXT4_FC_TAG_BASE_LEN helper
ext4: factor out ext4_free_ext_path()
ext4: remove unnecessary drop path references in mext_check_coverage()
ext4: update 'state->fc_regions_size' after successful memory allocation
ext4: fix potential memory leak in ext4_fc_record_regions()
ext4: fix potential memory leak in ext4_fc_record_modified_inode()
ext4: remove redundant checking in ext4_ioctl_checkpoint
jbd2: add miss release buffer head in fc_do_one_pass()
ext4: move DIOREAD_NOLOCK setting to ext4_set_def_opts()
ext4: remove useless local variable 'blocksize'
ext4: unify the ext4 super block loading operation
ext4: factor out ext4_journal_data_mode_check()
ext4: factor out ext4_load_and_init_journal()
ext4: factor out ext4_group_desc_init() and ext4_group_desc_free()
ext4: factor out ext4_geometry_check()
ext4: factor out ext4_check_feature_compatibility()
ext4: factor out ext4_init_metadata_csum()
ext4: factor out ext4_encoding_init()
...
Pull vfs file updates from Al Viro:
"struct file-related stuff"
* tag 'pull-file' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
dma_buf_getfile(): don't bother with ->f_flags reassignments
Change calling conventions for filldir_t
locks: fix TOCTOU race when granting write lease
Currently the I_DIRTY_TIME will never get set if the inode already has
I_DIRTY_INODE with assumption that it supersedes I_DIRTY_TIME. That's
true, however ext4 will only update the on-disk inode in
->dirty_inode(), not on actual writeback. As a result if the inode
already has I_DIRTY_INODE state by the time we get to
__mark_inode_dirty() only with I_DIRTY_TIME, the time was already filled
into on-disk inode and will not get updated until the next I_DIRTY_INODE
update, which might never come if we crash or get a power failure.
The problem can be reproduced on ext4 by running xfstest generic/622
with -o iversion mount option.
Fix it by allowing I_DIRTY_TIME to be set even if the inode already has
I_DIRTY_INODE. Also make sure that the case is properly handled in
writeback_single_inode() as well. Additionally changes in
xfs_fs_dirty_inode() was made to accommodate for I_DIRTY_TIME in flag.
Thanks Jan Kara for suggestions on how to make this work properly.
Cc: Dave Chinner <david@fromorbit.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: stable@kernel.org
Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Suggested-by: Jan Kara <jack@suse.cz>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220825100657.44217-1-lczerner@redhat.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
This is in preparation for adding tmpfile support to fuse, which requires
that the tmpfile creation and opening are done as a single operation.
Replace the 'struct dentry *' argument of i_op->tmpfile with
'struct file *'.
Call finish_open_simple() as the last thing in ->tmpfile() instances (may
be omitted in the error case).
Change d_tmpfile() argument to 'struct file *' as well to make callers more
readable.
Reviewed-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Create a helper finish_open_simple() that opens the file with the original
dentry. Handle the error case here as well to simplify callers.
Call this helper right after ->tmpfile() is called.
Next patch will change the tmpfile API and move this call into tmpfile
instances.
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
No callers outside of fs/namei.c anymore.
Reviewed-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
This helper unifies tmpfile creation with opening.
Existing vfs_tmpfile() callers outside of fs/namei.c will be converted to
using this helper. There are two such callers: cachefile and overlayfs.
The cachefiles code currently uses the open_with_fake_path() helper to open
the tmpfile, presumably to disable accounting of the open file. Overlayfs
uses tmpfile for copy_up, which means these struct file instances will be
short lived, hence it doesn't really matter if they are accounted or not.
Disable accounting in this helper too, which should be okay for both
callers.
Add MAY_OPEN permission checking for consistency. Like for create(2)
read/write permissions are not checked.
Reviewed-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
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
We need the poll_flags to know how to poll for the IO, and we should
have the batch structure in preparation for supporting batched
completions with iopoll.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
filldir_t instances (directory iterators callbacks) used to return 0 for
"OK, keep going" or -E... for "stop". Note that it's *NOT* how the
error values are reported - the rules for those are callback-dependent
and ->iterate{,_shared}() instances only care about zero vs. non-zero
(look at emit_dir() and friends).
So let's just return bool ("should we keep going?") - it's less confusing
that way. The choice between "true means keep going" and "true means
stop" is bikesheddable; we have two groups of callbacks -
do something for everything in directory, until we run into problem
and
find an entry in directory and do something to it.
The former tended to use 0/-E... conventions - -E<something> on failure.
The latter tended to use 0/1, 1 being "stop, we are done".
The callers treated anything non-zero as "stop", ignoring which
non-zero value did they get.
"true means stop" would be more natural for the second group; "true
means keep going" - for the first one. I tried both variants and
the things like
if allocation failed
something = -ENOMEM;
return true;
just looked unnatural and asking for trouble.
[folded suggestion from Matthew Wilcox <willy@infradead.org>]
Acked-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Pull io_uring fixes from Jens Axboe:
- Regression fix for this merge window, fixing a wrong order of
arguments for io_req_set_res() for passthru (Dylan)
- Fix for the audit code leaking context memory (Peilin)
- Ensure that provided buffers are memcg accounted (Pavel)
- Correctly handle short zero-copy sends (Pavel)
- Sparse warning fixes for the recvmsg multishot command (Dylan)
- Error handling fix for passthru (Anuj)
- Remove randomization of struct kiocb fields, to avoid it growing in
size if re-arranged in such a fashion that it grows more holes or
padding (Keith, Linus)
- Small series improving type safety of the sqe fields (Stefan)
* tag 'io_uring-6.0-2022-08-13' of git://git.kernel.dk/linux-block:
io_uring: add missing BUILD_BUG_ON() checks for new io_uring_sqe fields
io_uring: make io_kiocb_to_cmd() typesafe
fs: don't randomize struct kiocb fields
io_uring: consistently make use of io_notif_to_data()
io_uring: fix error handling for io_uring_cmd
io_uring: fix io_recvmsg_prep_multishot sparse warnings
io_uring/net: send retry for zerocopy
io_uring: mem-account pbuf buckets
audit, io_uring, io-wq: Fix memory leak in io_sq_thread() and io_wqe_worker()
io_uring: pass correct parameters to io_req_set_res
