J. Bruce Fields [Tue, 20 Sep 2011 20:59:58 +0000 (16:59 -0400)]
locks: break delegations on rename
Cc: David Howells <dhowells@redhat.com> Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Tue, 20 Sep 2011 13:14:34 +0000 (09:14 -0400)]
locks: break delegations on unlink
We need to break delegations on any operation that changes the set of
links pointing to an inode. Start with unlink.
Such operations also hold the i_mutex on a parent directory. Breaking a
delegation may require waiting for a timeout (by default 90 seconds) in
the case of a unresponsive NFS client. To avoid blocking all directory
operations, we therefore drop locks before waiting for the delegation.
The logic then looks like:
acquire locks
...
test for delegation; if found:
take reference on inode
release locks
wait for delegation break
drop reference on inode
retry
It is possible this could never terminate. (Even if we take precautions
to prevent another delegation being acquired on the same inode, we could
get a different inode on each retry.) But this seems very unlikely.
The initial test for a delegation happens after the lock on the target
inode is acquired, but the directory inode may have been acquired
further up the call stack. We therefore add a "struct inode **"
argument to any intervening functions, which we use to pass the inode
back up to the caller in the case it needs a delegation synchronously
broken.
Cc: David Howells <dhowells@redhat.com> Cc: Tyler Hicks <tyhicks@canonical.com> Cc: Dustin Kirkland <dustin.kirkland@gazzang.com> Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Mon, 5 Mar 2012 18:18:59 +0000 (13:18 -0500)]
locks: implement delegations
Implement NFSv4 delegations at the vfs level using the new FL_DELEG lock
type.
Note nfsd is the only delegation user and is only using read
delegations. Warn on any attempt to set a write delegation for now.
We'll come back to that case later.
Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Fri, 1 Jul 2011 19:18:34 +0000 (15:18 -0400)]
locks: introduce new FL_DELEG lock flag
For now FL_DELEG is just a synonym for FL_LEASE. So this patch doesn't
change behavior.
Next we'll modify break_lease to treat FL_DELEG leases differently, to
account for the fact that NFSv4 delegations should be broken in more
situations than Windows oplocks.
Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Mon, 5 Mar 2012 16:40:41 +0000 (11:40 -0500)]
vfs: take i_mutex on renamed file
A read delegation is used by NFSv4 as a guarantee that a client can
perform local read opens without informing the server.
The open operation takes the last component of the pathname as an
argument, thus is also a lookup operation, and giving the client the
above guarantee means informing the client before we allow anything that
would change the set of names pointing to the inode.
Therefore, we need to break delegations on rename, link, and unlink.
We also need to prevent new delegations from being acquired while one of
these operations is in progress.
We could add some completely new locking for that purpose, but it's
simpler to use the i_mutex, since that's already taken by all the
operations we care about.
The single exception is rename. So, modify rename to take the i_mutex
on the file that is being renamed.
Also fix up lockdep and Documentation/filesystems/directory-locking to
reflect the change.
Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Wed, 18 Apr 2012 19:21:34 +0000 (15:21 -0400)]
vfs: rename I_MUTEX_QUOTA now that it's not used for quotas
I_MUTEX_QUOTA is now just being used whenever we want to lock two
non-directories. So the name isn't right. I_MUTEX_NONDIR2 isn't
especially elegant but it's the best I could think of.
Also fix some outdated documentation.
Acked-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Wed, 18 Apr 2012 19:16:33 +0000 (15:16 -0400)]
vfs: pull ext4's double-i_mutex-locking into common code
We want to do this elsewhere as well.
Also catch any attempts to use it for directories (where this ordering
would conflict with ancestor-first directory ordering in lock_rename).
Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Dave Chinner <david@fromorbit.com> Acked-by: Jeff Layton <jlayton@redhat.com> Acked-by: "Theodore Ts'o" <tytso@mit.edu> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Fri, 18 Oct 2013 01:34:21 +0000 (21:34 -0400)]
exportfs: fix quadratic behavior in filehandle lookup
Suppose we're given the filehandle for a directory whose closest
ancestor in the dcache is its Nth ancestor.
The main loop in reconnect_path searches for an IS_ROOT ancestor of
target_dir, reconnects that ancestor to its parent, then recommences the
search for an IS_ROOT ancestor from target_dir.
This behavior is quadratic in N. And there's really no need to restart
the search from target_dir each time: once a directory has been looked
up, it won't become IS_ROOT again. So instead of starting from
target_dir each time, we can continue where we left off.
This simplifies the code and improves performance on very deep directory
heirachies. (I can't think of any reason anyone should need heirarchies
a hundred or more deep, but the performance improvement may be valuable
if only to limit damage in case of abuse.)
Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Thu, 17 Oct 2013 15:13:00 +0000 (11:13 -0400)]
exportfs: move most of reconnect_path to helper function
Also replace 3 easily-confused three-letter acronyms by more helpful
variable names.
Just cleanup, no change in functionality, with one exception: the
dentry_connected() check in the "out_reconnected" case will now only
check the ancestors of the current dentry instead of checking all the
way from target_dir. Since we've already verified connectivity up to
this dentry, that should be sufficient.
Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Thu, 17 Oct 2013 01:09:30 +0000 (21:09 -0400)]
exportfs: stop retrying once we race with rename/remove
There are two places here where we could race with a rename or remove:
- We could find the parent, but then be removed or renamed away
from that parent directory before finding our name in that
directory.
- We could find the parent, and find our name in that parent,
but then be renamed or removed before we look ourselves up by
that name in that parent.
In both cases the concurrent rename or remove will take care of
reconnecting the directory that we're currently examining. Our target
directory should then also be connected. Check this and clear
DISCONNECTED in these cases instead of looping around again.
Note: we *do* need to check that this actually happened if we want to be
robust in the face of corrupted filesystems: a corrupted filesystem
could just return a completely wrong parent, and we want to fail with an
error in that case before starting to clear DISCONNECTED on
non-DISCONNECTED filesystems.
Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Mon, 9 Sep 2013 20:15:13 +0000 (16:15 -0400)]
exportfs: clear DISCONNECTED on all parents sooner
Once we've found any connected parent, we know all our parents are
connected--that's true even if there's a concurrent rename. May as well
clear them all at once and be done with it.
Reviewed-by: Cristoph Hellwig <hch@lst.de> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Wed, 18 Jul 2012 22:27:37 +0000 (16:27 -0600)]
dcache: don't clear DCACHE_DISCONNECTED too early
DCACHE_DISCONNECTED should not be cleared until we're sure the dentry is
connected all the way up to the root of the filesystem. It *shouldn't*
be cleared as soon as the dentry is connected to a parent. That will
cause bugs at least on exportable filesystems.
Acked-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Fri, 29 Jun 2012 20:20:47 +0000 (16:20 -0400)]
dcache: Don't set DISCONNECTED on "pseudo filesystem" dentries
I can't for the life of me see any reason why anyone should care whether
a dentry that is never hooked into the dentry cache would need
DCACHE_DISCONNECTED set.
This originates from 4b936885ab04dc6e0bb0ef35e0e23c1a7364d9e5 "fs:
improve scalability of pseudo filesystems", which probably just made the
false assumption the DCACHE_DISCONNECTED was meant to be set on anything
not connected to a parent somehow.
So this is just confusing. Ideally the only uses of DCACHE_DISCONNECTED
would be in the filehandle-lookup code, which needs it to ensure
dentries are connected into the dentry tree before use.
I left d_alloc_pseudo there even though it's now equivalent to
__d_alloc(), just on the theory the name is better documentation of its
intended use outside dcache.c.
Cc: Nick Piggin <npiggin@kernel.dk> Acked-by: Christoph Hellwig <hch@infradead.org> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Thu, 28 Jun 2012 16:10:55 +0000 (12:10 -0400)]
dcache: use IS_ROOT to decide where dentry is hashed
Every hashed dentry is either hashed in the dentry_hashtable, or a
superblock's s_anon list.
__d_drop() assumes it can determine which is the case by checking
DCACHE_DISCONNECTED; this is not true.
It is true that when DCACHE_DISCONNECTED is cleared, the dentry is not
only hashed on dentry_hashtable, but is fully connected to its parents
back to the root.
But the converse is *not* true: fs/exportfs/expfs.c:reconnect_path()
attempts to connect a directory (found by filehandle lookup) back to
root by ascending to parents and performing lookups one at a time. It
does not clear DCACHE_DISCONNECTED until it's done, and that is not at
all an atomic process.
In particular, it is possible for DCACHE_DISCONNECTED to be set on a
dentry which is hashed on the dentry_hashtable.
Instead, use IS_ROOT() to check which hash chain a dentry is on. This
*does* work:
Dentries are hashed only by:
- d_obtain_alias, which adds an IS_ROOT() dentry to sb_anon.
- __d_rehash, called by _d_rehash: hashes to the dentry's
parent, and all callers of _d_rehash appear to have d_parent
set to a "real" parent.
- __d_rehash, called by __d_move: rehashes the moved dentry to
hash chain determined by target, and assigns target's d_parent
to its d_parent, before dropping the dentry's d_lock.
Therefore I believe it's safe for a holder of a dentry's d_lock to
assume that it is hashed on sb_anon if and only if IS_ROOT(dentry) is
true.
I believe the incorrect assumption about DCACHE_DISCONNECTED was
originally introduced by ceb5bdc2d246 "fs: dcache per-bucket dcache hash
locking".
Also add a comment while we're here.
Cc: Nick Piggin <npiggin@kernel.dk> Acked-by: Christoph Hellwig <hch@infradead.org> Reviewed-by: NeilBrown <neilb@suse.de> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
J. Bruce Fields [Tue, 10 Sep 2013 15:41:12 +0000 (11:41 -0400)]
exportfs: fix 32-bit nfsd handling of 64-bit inode numbers
Symptoms were spurious -ENOENTs on stat of an NFS filesystem from a
32-bit NFS server exporting a very large XFS filesystem, when the
server's cache is cold (so the inodes in question are not in cache).
Reviewed-by: Christoph Hellwig <hch@lst.de> Reported-by: Trevor Cordes <trevor@tecnopolis.ca> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
David Howells [Thu, 12 Sep 2013 18:22:53 +0000 (19:22 +0100)]
VFS: Put a small type field into struct dentry::d_flags
Put a type field into struct dentry::d_flags to indicate if the dentry is one
of the following types that relate particularly to pathwalk:
Miss (negative dentry)
Directory
"Automount" directory (defective - no i_op->lookup())
Symlink
Other (regular, socket, fifo, device)
The type field is set to one of the first five types on a dentry by calls to
__d_instantiate() and d_obtain_alias() from information in the inode (if one is
given).
The type is cleared by dentry_unlink_inode() when it reconstitutes an existing
dentry as a negative dentry.
Benjamin LaHaise [Tue, 17 Sep 2013 14:18:25 +0000 (10:18 -0400)]
rework aio migrate pages to use aio fs
Don't abuse anon_inodes.c to host private files needed by aio;
we can bloody well declare a mini-fs of our own instead of
patching up what anon_inodes can create for us.
Tested-by: Benjamin LaHaise <bcrl@kvack.org> Acked-by: Benjamin LaHaise <bcrl@kvack.org> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Al Viro [Tue, 8 Oct 2013 13:11:48 +0000 (09:11 -0400)]
make dump_emit() use vfs_write() instead of banging at ->f_op->write directly
... and deal with short writes properly - the output might be to pipe, after
all; as it is, e.g. no-MMU case of elf_fdpic coredump can write a whole lot
more than a page worth of data at one call.
Al Viro [Sat, 5 Oct 2013 19:32:35 +0000 (15:32 -0400)]
new helper: dump_emit()
dump_write() analog, takes core_dump_params instead of file,
keeps track of the amount written in cprm->written and checks for
cprm->limit. Start using it in binfmt_elf.c...
Al Viro [Mon, 30 Sep 2013 02:06:07 +0000 (22:06 -0400)]
RCU'd vfsmounts
* RCU-delayed freeing of vfsmounts
* vfsmount_lock replaced with a seqlock (mount_lock)
* sequence number from mount_lock is stored in nameidata->m_seq and
used when we exit RCU mode
* new vfsmount flag - MNT_SYNC_UMOUNT. Set by umount_tree() when its
caller knows that vfsmount will have no surviving references.
* synchronize_rcu() done between unlocking namespace_sem in namespace_unlock()
and doing pending mntput().
* new helper: legitimize_mnt(mnt, seq). Checks the mount_lock sequence
number against seq, then grabs reference to mnt. Then it rechecks mount_lock
again to close the race and either returns success or drops the reference it
has acquired. The subtle point is that in case of MNT_SYNC_UMOUNT we can
simply decrement the refcount and sod off - aforementioned synchronize_rcu()
makes sure that final mntput() won't come until we leave RCU mode. We need
that, since we don't want to end up with some lazy pathwalk racing with
umount() and stealing the final mntput() from it - caller of umount() may
expect it to return only once the fs is shut down and we don't want to break
that. In other cases (i.e. with MNT_SYNC_UMOUNT absent) we have to do
full-blown mntput() in case of mount_lock sequence number mismatch happening
just as we'd grabbed the reference, but in those cases we won't be stealing
the final mntput() from anything that would care.
* mntput_no_expire() doesn't lock anything on the fast path now. Incidentally,
SMP and UP cases are handled the same way - no ifdefs there.
* normal pathname resolution does *not* do any writes to mount_lock. It does,
of course, bump the refcounts of vfsmount and dentry in the very end, but that's
it.
Miklos Szeredi [Tue, 1 Oct 2013 14:44:54 +0000 (16:44 +0200)]
vfs: introduce d_instantiate_no_diralias()
...which just returns -EBUSY if a directory alias would be created.
This is to be used by fuse mkdir to make sure that a buggy or malicious
userspace filesystem doesn't do anything nasty. Previously fuse used a
private mutex for this purpose, which can now go away.
Al Viro [Tue, 1 Oct 2013 20:11:26 +0000 (16:11 -0400)]
split __lookup_mnt() in two functions
Instead of passing the direction as argument (and checking it on every
step through the hash chain), just have separate __lookup_mnt() and
__lookup_mnt_last(). And use the standard iterators...
Al Viro [Mon, 16 Sep 2013 00:45:11 +0000 (20:45 -0400)]
ecryptfs: don't leave RCU pathwalk immediately
If the underlying dentry doesn't have ->d_revalidate(), there's no need to
force dropping out of RCU mode. All we need for that is to make freeing
ecryptfs_dentry_info RCU-delayed.
projects / ~andy / linux / log