linux.git/fs/nilfs2, branch v6.15 Linux kernel source tree nilfs2: fix deadlock warnings caused by lock dependency in init_nilfs() 2025-05-08T06:39:42+00:00 Ryusuke Konishi konishi.ryusuke@gmail.com 2025-05-03T05:33:14+00:00 fb881cd7604536b17a1927fb0533f9a6982ffcc5 After commit c0e473a0d226 ("block: fix race between set_blocksize and read paths") was merged, set_blocksize() called by sb_set_blocksize() now locks the inode of the backing device file. As a result of this change, syzbot started reporting deadlock warnings due to a circular dependency involving the semaphore "ns_sem" of the nilfs object, the inode lock of the backing device file, and the locks that this inode lock is transitively dependent on. This is caused by a new lock dependency added by the above change, since init_nilfs() calls sb_set_blocksize() in the lock section of "ns_sem". However, these warnings are false positives because init_nilfs() is called in the early stage of the mount operation and the filesystem has not yet started. The reason why "ns_sem" is locked in init_nilfs() was to avoid a race condition in nilfs_fill_super() caused by sharing a nilfs object among multiple filesystem instances (super block structures) in the early implementation. However, nilfs objects and super block structures have long ago become one-to-one, and there is no longer any need to use the semaphore there. So, fix this issue by removing the use of the semaphore "ns_sem" in init_nilfs(). Link: https://lkml.kernel.org/r/20250503053327.12294-1-konishi.ryusuke@gmail.com Fixes: c0e473a0d226 ("block: fix race between set_blocksize and read paths") Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com> Reported-by: syzbot+00f7f5b884b117ee6773@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=00f7f5b884b117ee6773 Tested-by: syzbot+00f7f5b884b117ee6773@syzkaller.appspotmail.com Reported-by: syzbot+f30591e72bfc24d4715b@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=f30591e72bfc24d4715b Tested-by: syzbot+f30591e72bfc24d4715b@syzkaller.appspotmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> 
After commit c0e473a0d226 ("block: fix race between set_blocksize and read
paths") was merged, set_blocksize() called by sb_set_blocksize() now locks
the inode of the backing device file.  As a result of this change, syzbot
started reporting deadlock warnings due to a circular dependency involving
the semaphore "ns_sem" of the nilfs object, the inode lock of the backing
device file, and the locks that this inode lock is transitively dependent
on.

This is caused by a new lock dependency added by the above change, since
init_nilfs() calls sb_set_blocksize() in the lock section of "ns_sem". 
However, these warnings are false positives because init_nilfs() is called
in the early stage of the mount operation and the filesystem has not yet
started.

The reason why "ns_sem" is locked in init_nilfs() was to avoid a race
condition in nilfs_fill_super() caused by sharing a nilfs object among
multiple filesystem instances (super block structures) in the early
implementation.  However, nilfs objects and super block structures have
long ago become one-to-one, and there is no longer any need to use the
semaphore there.

So, fix this issue by removing the use of the semaphore "ns_sem" in
init_nilfs().

Link: https://lkml.kernel.org/r/20250503053327.12294-1-konishi.ryusuke@gmail.com
Fixes: c0e473a0d226 ("block: fix race between set_blocksize and read paths")
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Reported-by: syzbot+00f7f5b884b117ee6773@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=00f7f5b884b117ee6773
Tested-by: syzbot+00f7f5b884b117ee6773@syzkaller.appspotmail.com
Reported-by: syzbot+f30591e72bfc24d4715b@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=f30591e72bfc24d4715b
Tested-by: syzbot+f30591e72bfc24d4715b@syzkaller.appspotmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
treewide: Switch/rename to timer_delete[_sync]() 2025-04-05T08:30:12+00:00 Thomas Gleixner tglx@linutronix.de 2025-04-05T08:17:26+00:00 8fa7292fee5c5240402371ea89ab285ec856c916 timer_delete[_sync]() replaces del_timer[_sync](). Convert the whole tree over and remove the historical wrapper inlines. Conversion was done with coccinelle plus manual fixups where necessary. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org> 
timer_delete[_sync]() replaces del_timer[_sync](). Convert the whole tree
over and remove the historical wrapper inlines.

Conversion was done with coccinelle plus manual fixups where necessary.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Change inode_operations.mkdir to return struct dentry * 2025-02-27T19:00:17+00:00 NeilBrown neilb@suse.de 2025-02-27T01:32:53+00:00 88d5baf69082e5b410296435008329676b687549 Some filesystems, such as NFS, cifs, ceph, and fuse, do not have complete control of sequencing on the actual filesystem (e.g. on a different server) and may find that the inode created for a mkdir request already exists in the icache and dcache by the time the mkdir request returns. For example, if the filesystem is mounted twice the directory could be visible on the other mount before it is on the original mount, and a pair of name_to_handle_at(), open_by_handle_at() calls could instantiate the directory inode with an IS_ROOT() dentry before the first mkdir returns. This means that the dentry passed to ->mkdir() may not be the one that is associated with the inode after the ->mkdir() completes. Some callers need to interact with the inode after the ->mkdir completes and they currently need to perform a lookup in the (rare) case that the dentry is no longer hashed. This lookup-after-mkdir requires that the directory remains locked to avoid races. Planned future patches to lock the dentry rather than the directory will mean that this lookup cannot be performed atomically with the mkdir. To remove this barrier, this patch changes ->mkdir to return the resulting dentry if it is different from the one passed in. Possible returns are: NULL - the directory was created and no other dentry was used ERR_PTR() - an error occurred non-NULL - this other dentry was spliced in This patch only changes file-systems to return "ERR_PTR(err)" instead of "err" or equivalent transformations. Subsequent patches will make further changes to some file-systems to return a correct dentry. Not all filesystems reliably result in a positive hashed dentry: - NFS, cifs, hostfs will sometimes need to perform a lookup of the name to get inode information. Races could result in this returning something different. Note that this lookup is non-atomic which is what we are trying to avoid. Placing the lookup in filesystem code means it only happens when the filesystem has no other option. - kernfs and tracefs leave the dentry negative and the ->revalidate operation ensures that lookup will be called to correctly populate the dentry. This could be fixed but I don't think it is important to any of the users of vfs_mkdir() which look at the dentry. The recommendation to use d_drop();d_splice_alias() is ugly but fits with current practice. A planned future patch will change this. Reviewed-by: Jeff Layton <jlayton@kernel.org> Reviewed-by: Jan Kara <jack@suse.cz> Signed-off-by: NeilBrown <neilb@suse.de> Link: https://lore.kernel.org/r/20250227013949.536172-2-neilb@suse.de Signed-off-by: Christian Brauner <brauner@kernel.org> 
Some filesystems, such as NFS, cifs, ceph, and fuse, do not have
complete control of sequencing on the actual filesystem (e.g.  on a
different server) and may find that the inode created for a mkdir
request already exists in the icache and dcache by the time the mkdir
request returns.  For example, if the filesystem is mounted twice the
directory could be visible on the other mount before it is on the
original mount, and a pair of name_to_handle_at(), open_by_handle_at()
calls could instantiate the directory inode with an IS_ROOT() dentry
before the first mkdir returns.

This means that the dentry passed to ->mkdir() may not be the one that
is associated with the inode after the ->mkdir() completes.  Some
callers need to interact with the inode after the ->mkdir completes and
they currently need to perform a lookup in the (rare) case that the
dentry is no longer hashed.

This lookup-after-mkdir requires that the directory remains locked to
avoid races.  Planned future patches to lock the dentry rather than the
directory will mean that this lookup cannot be performed atomically with
the mkdir.

To remove this barrier, this patch changes ->mkdir to return the
resulting dentry if it is different from the one passed in.
Possible returns are:
  NULL - the directory was created and no other dentry was used
  ERR_PTR() - an error occurred
  non-NULL - this other dentry was spliced in

This patch only changes file-systems to return "ERR_PTR(err)" instead of
"err" or equivalent transformations.  Subsequent patches will make
further changes to some file-systems to return a correct dentry.

Not all filesystems reliably result in a positive hashed dentry:

- NFS, cifs, hostfs will sometimes need to perform a lookup of
  the name to get inode information.  Races could result in this
  returning something different. Note that this lookup is
  non-atomic which is what we are trying to avoid.  Placing the
  lookup in filesystem code means it only happens when the filesystem
  has no other option.
- kernfs and tracefs leave the dentry negative and the ->revalidate
  operation ensures that lookup will be called to correctly populate
  the dentry.  This could be fixed but I don't think it is important
  to any of the users of vfs_mkdir() which look at the dentry.

The recommendation to use
    d_drop();d_splice_alias()
is ugly but fits with current practice.  A planned future patch will
change this.

Reviewed-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: NeilBrown <neilb@suse.de>
Link: https://lore.kernel.org/r/20250227013949.536172-2-neilb@suse.de
Signed-off-by: Christian Brauner <brauner@kernel.org>
nilfs2: fix possible int overflows in nilfs_fiemap() 2025-02-01T11:53:26+00:00 Nikita Zhandarovich n.zhandarovich@fintech.ru 2025-01-24T22:20:53+00:00 6438ef381c183444f7f9d1de18f22661cba1e946 Since nilfs_bmap_lookup_contig() in nilfs_fiemap() calculates its result by being prepared to go through potentially maxblocks == INT_MAX blocks, the value in n may experience an overflow caused by left shift of blkbits. While it is extremely unlikely to occur, play it safe and cast right hand expression to wider type to mitigate the issue. Found by Linux Verification Center (linuxtesting.org) with static analysis tool SVACE. Link: https://lkml.kernel.org/r/20250124222133.5323-1-konishi.ryusuke@gmail.com Fixes: 622daaff0a89 ("nilfs2: fiemap support") Signed-off-by: Nikita Zhandarovich <n.zhandarovich@fintech.ru> Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com> Cc: <stable@vger.kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> 
Since nilfs_bmap_lookup_contig() in nilfs_fiemap() calculates its result
by being prepared to go through potentially maxblocks == INT_MAX blocks,
the value in n may experience an overflow caused by left shift of blkbits.

While it is extremely unlikely to occur, play it safe and cast right hand
expression to wider type to mitigate the issue.

Found by Linux Verification Center (linuxtesting.org) with static analysis
tool SVACE.

Link: https://lkml.kernel.org/r/20250124222133.5323-1-konishi.ryusuke@gmail.com
Fixes: 622daaff0a89 ("nilfs2: fiemap support")
Signed-off-by: Nikita Zhandarovich <n.zhandarovich@fintech.ru>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
nilfs2: do not update mtime of renamed directory that is not moved 2025-01-25T06:47:26+00:00 Ryusuke Konishi konishi.ryusuke@gmail.com 2025-01-11T14:26:36+00:00 e30ccbb9c022660c52648f02ab6231d5dcaf4253 A minor issue with nilfs_rename, originating from an old ext2 implementation, is that the mtime is updated even if the rename target is a directory and it is renamed within the same directory, rather than moved to a different directory. In this case, the child directory being renamed does not change in any way, so changing its mtime is unnecessary according to the specification, and can unnecessarily confuse backup tools. In ext2, this issue was fixed by commit 39fe7557b4d6 ("ext2: Do not update mtime of a moved directory") and a few subsequent fixes, but it remained in nilfs2. Fix this issue by not calling nilfs_set_link(), which rewrites the inode number of the directory entry that refers to the parent directory, when the move target is a directory and the source and destination are the same directory. Here, the directory to be moved only needs to be read if the inode number of the parent directory is rewritten with nilfs_set_link, so also adjust the execution conditions of the preparation work to avoid unnecessary directory reads. Link: https://lkml.kernel.org/r/20250111143518.7901-3-konishi.ryusuke@gmail.com Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> 
A minor issue with nilfs_rename, originating from an old ext2
implementation, is that the mtime is updated even if the rename target is
a directory and it is renamed within the same directory, rather than moved
to a different directory.

In this case, the child directory being renamed does not change in any
way, so changing its mtime is unnecessary according to the specification,
and can unnecessarily confuse backup tools.

In ext2, this issue was fixed by commit 39fe7557b4d6 ("ext2: Do not update
mtime of a moved directory") and a few subsequent fixes, but it remained
in nilfs2.

Fix this issue by not calling nilfs_set_link(), which rewrites the inode
number of the directory entry that refers to the parent directory, when
the move target is a directory and the source and destination are the same
directory.

Here, the directory to be moved only needs to be read if the inode number
of the parent directory is rewritten with nilfs_set_link, so also adjust
the execution conditions of the preparation work to avoid unnecessary
directory reads.

Link: https://lkml.kernel.org/r/20250111143518.7901-3-konishi.ryusuke@gmail.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
nilfs2: handle errors that nilfs_prepare_chunk() may return 2025-01-25T06:47:25+00:00 Ryusuke Konishi konishi.ryusuke@gmail.com 2025-01-11T14:26:35+00:00 ee70999a988b8abc3490609142f50ebaa8344432 Patch series "nilfs2: fix issues with rename operations". This series fixes BUG_ON check failures reported by syzbot around rename operations, and a minor behavioral issue where the mtime of a child directory changes when it is renamed instead of moved. This patch (of 2): The directory manipulation routines nilfs_set_link() and nilfs_delete_entry() rewrite the directory entry in the folio/page previously read by nilfs_find_entry(), so error handling is omitted on the assumption that nilfs_prepare_chunk(), which prepares the buffer for rewriting, will always succeed for these. And if an error is returned, it triggers the legacy BUG_ON() checks in each routine. This assumption is wrong, as proven by syzbot: the buffer layer called by nilfs_prepare_chunk() may call nilfs_get_block() if necessary, which may fail due to metadata corruption or other reasons. This has been there all along, but improved sanity checks and error handling may have made it more reproducible in fuzzing tests. Fix this issue by adding missing error paths in nilfs_set_link(), nilfs_delete_entry(), and their caller nilfs_rename(). Link: https://lkml.kernel.org/r/20250111143518.7901-1-konishi.ryusuke@gmail.com Link: https://lkml.kernel.org/r/20250111143518.7901-2-konishi.ryusuke@gmail.com Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com> Reported-by: syzbot+32c3706ebf5d95046ea1@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=32c3706ebf5d95046ea1 Reported-by: syzbot+1097e95f134f37d9395c@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=1097e95f134f37d9395c Fixes: 2ba466d74ed7 ("nilfs2: directory entry operations") Signed-off-by: Andrew Morton <akpm@linux-foundation.org> 
Patch series "nilfs2: fix issues with rename operations".

This series fixes BUG_ON check failures reported by syzbot around rename
operations, and a minor behavioral issue where the mtime of a child
directory changes when it is renamed instead of moved.


This patch (of 2):

The directory manipulation routines nilfs_set_link() and
nilfs_delete_entry() rewrite the directory entry in the folio/page
previously read by nilfs_find_entry(), so error handling is omitted on the
assumption that nilfs_prepare_chunk(), which prepares the buffer for
rewriting, will always succeed for these.  And if an error is returned, it
triggers the legacy BUG_ON() checks in each routine.

This assumption is wrong, as proven by syzbot: the buffer layer called by
nilfs_prepare_chunk() may call nilfs_get_block() if necessary, which may
fail due to metadata corruption or other reasons.  This has been there all
along, but improved sanity checks and error handling may have made it more
reproducible in fuzzing tests.

Fix this issue by adding missing error paths in nilfs_set_link(),
nilfs_delete_entry(), and their caller nilfs_rename().

Link: https://lkml.kernel.org/r/20250111143518.7901-1-konishi.ryusuke@gmail.com
Link: https://lkml.kernel.org/r/20250111143518.7901-2-konishi.ryusuke@gmail.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Reported-by: syzbot+32c3706ebf5d95046ea1@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=32c3706ebf5d95046ea1
Reported-by: syzbot+1097e95f134f37d9395c@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=1097e95f134f37d9395c
Fixes: 2ba466d74ed7 ("nilfs2: directory entry operations")
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
nilfs2: revise the return value description style for consistency. 2025-01-25T06:47:25+00:00 Ryusuke Konishi konishi.ryusuke@gmail.com 2025-01-10T01:01:50+00:00 2e62857f1312606bd173b81102047e8c7bce3c3f Also for comments that do not cause kernel-doc warnings (those that list multiple error codes), revise the return value description style to match Brian G.'s suggestion of "..., or one of the following negative error codes on failure:". Link: https://lkml.kernel.org/r/CAAq45aNh1qV8P6XgDhKeNstT=PvcPUaCXsAF-f9rvmzznsZL5A@mail.gmail.com Link: https://lkml.kernel.org/r/20250110010530.21872-8-konishi.ryusuke@gmail.com Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com> Cc: "Brian G ." <gissf1@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> 
Also for comments that do not cause kernel-doc warnings (those that list
multiple error codes), revise the return value description style to match
Brian G.'s suggestion of "..., or one of the following negative error
codes on failure:".

Link: https://lkml.kernel.org/r/CAAq45aNh1qV8P6XgDhKeNstT=PvcPUaCXsAF-f9rvmzznsZL5A@mail.gmail.com
Link: https://lkml.kernel.org/r/20250110010530.21872-8-konishi.ryusuke@gmail.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: "Brian G ." <gissf1@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
nilfs2: add missing return value kernel-doc descriptions 2025-01-25T06:47:25+00:00 Ryusuke Konishi konishi.ryusuke@gmail.com 2025-01-10T01:01:49+00:00 fd4e7fadfd91023c2c5868f16af3fe8bed06636b There are a number of kernel-doc comments for functions that are missing return values, which also causes a number of warnings when the kernel-doc script is run with the "-Wall" option. Fix this issue by adding proper return value descriptions, and improve code maintainability. Link: https://lkml.kernel.org/r/20250110010530.21872-7-konishi.ryusuke@gmail.com Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com> Cc: "Brian G ." <gissf1@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> 
There are a number of kernel-doc comments for functions that are missing
return values, which also causes a number of warnings when the kernel-doc
script is run with the "-Wall" option.

Fix this issue by adding proper return value descriptions, and improve
code maintainability.

Link: https://lkml.kernel.org/r/20250110010530.21872-7-konishi.ryusuke@gmail.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: "Brian G ." <gissf1@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
nilfs2: correct return value kernel-doc descriptions for the rest 2025-01-25T06:47:25+00:00 Ryusuke Konishi konishi.ryusuke@gmail.com 2025-01-10T01:01:48+00:00 343d4a33f26d048fde4b02cfc3681f766b3f1b12 Similar to the previous changes to fix return value descriptions, this fixes the format of the return value descriptions of functions for the rest. Link: https://lkml.kernel.org/r/20250110010530.21872-6-konishi.ryusuke@gmail.com Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com> Cc: "Brian G ." <gissf1@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> 
Similar to the previous changes to fix return value descriptions, this
fixes the format of the return value descriptions of functions for the
rest.

Link: https://lkml.kernel.org/r/20250110010530.21872-6-konishi.ryusuke@gmail.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: "Brian G ." <gissf1@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
nilfs2: correct return value kernel-doc descriptions for metadata files 2025-01-25T06:47:25+00:00 Ryusuke Konishi konishi.ryusuke@gmail.com 2025-01-10T01:01:47+00:00 b8268501f1e3c49e3f90526d44b92eae4baa421b Similar to the previous changes to fix return value descriptions, this fixes the format of the return value descriptions for metadata file functions other than sufile. Link: https://lkml.kernel.org/r/20250110010530.21872-5-konishi.ryusuke@gmail.com Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com> Cc: "Brian G ." <gissf1@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> 
Similar to the previous changes to fix return value descriptions, this
fixes the format of the return value descriptions for metadata file
functions other than sufile.

Link: https://lkml.kernel.org/r/20250110010530.21872-5-konishi.ryusuke@gmail.com
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@gmail.com>
Cc: "Brian G ." <gissf1@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>