In the replay case, the

	renew_client(session->se_client);

happens after we've droppped the sessionid_lock, and without holding a
reference on the session; so there's nothing preventing the session
being freed before we get here.

Thanks to Benny Halevy for catching a bug in an earlier version of this
patch.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>
Acked-by: NBenny Halevy <bhalevy@panasas.com>

We "goto finish" from several places where "exp" is an ERR_PTR.  Also I
changed the check for "fsid_key" so that it was consistent with the check
I added.
Signed-off-by: NDan Carpenter <error27@gmail.com>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Enforce the rules about compound op ordering.

Motivated by implementing RECLAIM_COMPLETE, for which the client is
implicit in the current session, so it is important to ensure a
succesful SEQUENCE proceeds the RECLAIM_COMPLETE.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

The rfc allows a client to change the callback parameters, but we didn't
previously implement it.

Teach the callbacks to rerun themselves (by placing themselves on a
workqueue) when they recognize that their rpc task has been killed and
that the callback connection has changed.

Then we can change the callback connection by setting up a new rpc
client, modifying the nfs4 client to point at it, waiting for any work
in progress to complete, and then shutting down the old client.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Mainly I just want to separate the arguments used for setting up the tcp
client from the rest.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Now that the shutdown sequence guarantees callbacks are shut down before
the client is destroyed, we no longer have a use for cl_count.

We'll probably reinstate a reference count on the client some day, but
it will be held by users other than callbacks.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

The NFSv4 server's fl_break callback can sleep (dropping the BKL), in
order to allocate a new rpc task to send a recall to the client.

As far as I can tell this doesn't cause any races in the current code,
but the analysis is difficult.  Also, the sleep here may complicate the
move away from the BKL.

So, just schedule some work to do the job for us instead.  The work will
later also prove useful for restarting a call after the callback
information is changed.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Looks like a put-and-paste mistake.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

We should clear these flags on any new create_session, not just on the
first one.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Any null probe rpc will be synchronously destroyed by the
rpc_shutdown_client() in expire_client(), so the rpc task cannot outlast
the nfs4 client.  Therefore there's no need for that task to hold a
reference on the client.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

I haven't found this useful.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Once we've expired the client, there's no further purpose to the
callbacks; go ahead and shut down the callback client rather than
waiting for the last reference to go.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Instead of allocating this small structure, just include it in the
delegation.

The nfsd4_callback structure isn't really necessary yet, but we plan to
add to it all the information necessary to perform a callback.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

This is the second attempt to fix the problem whereby a COMMIT call
causes a lease break and triggers a possible deadlock.

The problem is that nfsd attempts to break a lease on a COMMIT call.
This triggers a delegation recall if the lease is held for a delegation.
If the client is the one holding the delegation and it's the same one on
which it's issuing the COMMIT, then it can't return that delegation
until the COMMIT is complete. But, nfsd won't complete the COMMIT until
the delegation is returned. The client and server are essentially
deadlocked until the state is marked bad (due to the client not
responding on the callback channel).

The first patch attempted to deal with this by eliminating the open of
the file altogether and simply had nfsd_commit pass a NULL file pointer
to the vfs_fsync_range. That would conflict with some work in progress
by Christoph Hellwig to clean up the fsync interface, so this patch
takes a different approach.

This declares a new NFSD_MAY_NOT_BREAK_LEASE access flag that indicates
to nfsd_open that it should not break any leases when opening the file,
and has nfsd_commit set that flag on the nfsd_open call.

For now, this patch leaves nfsd_commit opening the file with write
access since I'm not clear on what sort of access would be more
appropriate.
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Cc: stable@kernel.org
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Both the _lookup and the _update functions for these two caches
independently calculate the hash of the key.
So factor out that code for improved reuse.
Signed-off-by: NNeilBrown <neilb@suse.de>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

The current documentation here is out of date, and not quite right.

(Future work: some user documentation would be useful.)
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Allow explicit configuration of the grace period time as well as the
lease period time.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

We'll soon allow setting the grace period, so we'll want to share this
code.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

This is another layer of indirection that doesn't really buy us
anything.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

The original code here assumed we'd allow the user to change the lease
any time, but only allow the change to take effect on restart.  Since
then we modified the code to allow setting the lease on when the server
is down.  Update the rest of the code to reflect that fact, clarify
variable names, and add document.

Also, the code insisted that the grace period always be the longer of
the old and new lease periods, but that's overly conservative--as long
as it lasts at least the old lease period, old clients should still know
to recover in time.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Instead of accessing the lease time directly, some users call
nfs4_lease_time(), and some a macro, NFSD_LEASE_TIME, defined as
nfs4_lease_time().  Neither layer of indirection serves any purpose.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

We'll introduce FMODE_RANDOM which will be runtime modified.  So protect
all runtime modification to f_mode with f_lock to avoid races.
Signed-off-by: NWu Fengguang <fengguang.wu@intel.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: <stable@kernel.org>			[2.6.33.x]
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Currently various places in the VFS call vfs_dq_init directly.  This means
we tie the quota code into the VFS.  Get rid of that and make the
filesystem responsible for the initialization.   For most metadata operations
this is a straight forward move into the methods, but for truncate and
open it's a bit more complicated.

For truncate we currently only call vfs_dq_init for the sys_truncate case
because open already takes care of it for ftruncate and open(O_TRUNC) - the
new code causes an additional vfs_dq_init for those which is harmless.

For open the initialization is moved from do_filp_open into the open method,
which means it happens slightly earlier now, and only for regular files.
The latter is fine because we don't need to initialize it for operations
on special files, and we already do it as part of the namespace operations
for directories.

Add a dquot_file_open helper that filesystems that support generic quotas
can use to fill in ->open.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJan Kara <jack@suse.cz>

There's no need to allocate this cred more than once.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

RFC says we need to follow the chain of mounts if there's more
than one stacked on that point.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

... instead of mixing FMODE_ and O_
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

The server's callback client should stop trying to connect to the
client's callback server as soon as it gets ECONNREFUSED.

The NFS server's callback client does not call rpc_ping(), but appears
to have it's own "ping" procedure, so it wasn't covered by commit
caabea8a.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

- Add commit_metadata export_operation to allow the underlying filesystem to
decide how to commit an inode most efficiently.

- Usage of nfsd_sync_dir and write_inode_now has been replaced with the
commit_metadata function that takes a svc_fh.

- The commit_metadata function calls the commit_metadata export_op if it's
there, or else falls back to sync_inode instead of fsync and write_inode_now
because only metadata need be synced here.

- nfsd4_sync_rec_dir now uses vfs_fsync so that commit_metadata can be static
Signed-off-by: NBen Myers <bpm@sgi.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

commit 1e41568d ("Take ima_path_check()
in nfsd past dentry_open() in nfsd_open()") moved this code back to its
original location but missed the "else".
Signed-off-by: NChuck Ebbert <cebbert@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Commit f39bde24 fixed the error return from PUTROOTFH in the
case where there is no pseudofilesystem.

This is really a case we shouldn't hit on a correctly configured server:
in the absence of a root filehandle, there's no point accepting version
4 NFS rpc calls at all.

But the shared responsibility between kernel and userspace here means
the kernel on its own can't eliminate the possiblity of this happening.
And we have indeed gotten this wrong in distro's, so new client-side
mount code that attempts to negotiate v4 by default first has to work
around this case.

Therefore when commit f39bde24 arrived at roughly the same
time as the new v4-default mount code, which explicitly checked only for
the previous error, the result was previously fine mounts suddenly
failing.

We'll fix both sides for now: revert the error change, and make the
client-side mount workaround more robust.
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

ima_path_check actually deals with files!  call it ima_file_check instead.
Signed-off-by: NEric Paris <eparis@redhat.com>
Acked-by: NMimi Zohar <zohar@linux.vnet.ibm.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

The "Untangling ima mess, part 2 with counters" patch messed
up the counters.  Based on conversations with Al Viro, this patch
streamlines ima_path_check() by removing the counter maintaince.
The counters are now updated independently, from measuring the file,
in __dentry_open() and alloc_file() by calling ima_counts_get().
ima_path_check() is called from nfsd and do_filp_open().
It also did not measure all files that should have been measured.
Reason: ima_path_check() got bogus value passed as mask.
[AV: mea culpa]
[AV: add missing nfsd bits]
Signed-off-by: NMimi Zohar <zohar@us.ibm.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

The NFS COMMIT operation allows the client to specify the exact byte range
that it wishes to sync to disk in order to optimise server performance.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Try to create a PF_INET6 listener for NFSD, if IPv6 is enabled in the
kernel.

Make sure nfsd_serv's reference count is decreased if
__write_ports_addxprt() failed to create a listener.  See
__write_ports_addfd().

Our current plan is to rely on rpc.nfsd to create appropriate IPv6
listeners when server-side NFS/IPv6 support is desired.  Legacy
behavior, via the write_threads or write_svc kernel APIs, will remain
the same -- only IPv4 listeners are created.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
[bfields@citi.umich.edu: Move error-handling code to end]
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

write_ports() converts svc_create_xprt()'s ENOENT error return to
EPROTONOSUPPORT so that rpc.nfsd (in user space) can report an error
message that makes sense.

It turns out that several of the other kernel APIs rpc.nfsd use can
also return ENOENT from svc_create_xprt(), by way of lockd_up().

On the client side, an NFSv2 or NFSv3 mount request can also return
the result of lockd_up().  This error may also be returned during an
NFSv4 mount request, since the NFSv4 callback service uses
svc_create_xprt() to create the callback listener.  An ENOENT error
return results in a confusing error message from the mount command.

Let's have svc_create_xprt() return EPROTONOSUPPORT instead of ENOENT.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Signed-off-by: NRicardo Labiaga <Ricardo.Labiaga@netapp.com>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Instead of opencoding the fsync calling sequence use vfs_fsync.  This also
gets rid of the useless i_mutex over the data writeout.

Consolidate the remaining special code for syncing directories and document
it's quirks.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

Since we're checking for LAST_NFS4_OP, use FIRST_NFS4_OP to be consistent.
Signed-off-by: NRicardo Labiaga <Ricardo.Labiaga@netapp.com>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>

The server incorrectly assumes that the operations in the
array start with value 0.  The first operation (OP_ACCESS)
has a value of 3, causing the check in nfsd4_decode_compound
to be off.

Instead of comparing that the operation number is less than
the number of elements in the array, the server should verify
that it is less than the maximum valid operation number
defined by LAST_NFS4_OP.
Signed-off-by: NRicardo Labiaga <Ricardo.Labiaga@netapp.com>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>