Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

The main problem is dealing with inode->i_size: we need to set the
inode->i_lock on all attribute updates, and so vmtruncate won't cut it.
Make an NFS-private version of vmtruncate that has the necessary locking
semantics.

The result should be that the following inode attribute updates are
protected by inode->i_lock
	nfsi->cache_validity
	nfsi->read_cache_jiffies
	nfsi->attrtimeo
	nfsi->attrtimeo_timestamp
	nfsi->change_attr
	nfsi->last_updated
	nfsi->cache_change_attribute
	nfsi->access_cache
	nfsi->access_cache_entry_lru
	nfsi->access_cache_inode_lru
	nfsi->acl_access
	nfsi->acl_default
	nfsi->nfs_page_tree
	nfsi->ncommit
	nfsi->npages
	nfsi->open_files
	nfsi->silly_list
	nfsi->acl
	nfsi->open_states
	inode->i_size
	inode->i_atime
	inode->i_mtime
	inode->i_ctime
	inode->i_nlink
	inode->i_uid
	inode->i_gid

The following is protected by dir->i_mutex
	nfsi->cookieverf
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

fcntl(F_GETLK) on an nfs client incorrectly returns
the values for the conflicting lock. fl_len value is
always 1.
If the conflicting lock is (0, 4095) the F_GETLK
request for (1024, 10) returns (0, 1), which doesn't
even cover the requested range, and is quite confusing.
The fix is trivial, set fl_end from the fl_end value
recieved from the nfs server.
Signed-off-by: NFelix Blyakher <felixb@sgi.com>
Signed-off-by: N"J. Bruce Fields" <bfields@citi.umich.edu>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Introduce a new API to register RPC services on IPv6 interfaces to allow
the NFS server and lockd to advertise on IPv6 networks.

Unlike rpcb_register(), the new rpcb_v4_register() function uses rpcbind
protocol version 4 to contact the local rpcbind daemon.  The version 4
SET/UNSET procedures allow services to register address families besides
AF_INET, register at specific network interfaces, and register transport
protocols besides UDP and TCP.  All of this functionality is exposed via
the new rpcb_v4_register() kernel API.

A user-space rpcbind daemon implementation that supports version 4 of the
rpcbind protocol is required in order to make use of this new API.

Note that rpcbind version 3 is sufficient to support the new rpcbind
facilities listed above, but most extant implementations use version 4.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

rpcbind version 4 registration will reuse part of rpcb_register, so just
split it out into a separate function now.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up: Callers that required a privileged source port now use
rpcb_create_local(), so we can remove the @privileged argument from
rpcb_create().
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Add rpcb_create_local() for use by rpcb_register() and upcoming IPv6
registration functions.

Ensure any errors encountered by rpcb_create_local() are properly
reported.

We can also use a statically allocated constant loopback socket address
instead of one allocated on the stack and initialized every time the
function is called.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

The rpcbind versions 3 and 4 SET and UNSET procedures use the same
arguments as the GETADDR procedure.

While definitely a bug, this hasn't been a problem so far since the
kernel hasn't used version 3 or 4 SET and UNSET.  But this will change
in just a moment.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NJ. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

If another task is busy in rpcb_getport_async number, it is more efficient
to have it wake us up when it has finished instead of arbitrarily sleeping
for 5 seconds.

Also ensure that rpcb_wake_rpcbind_waiters() is called regardless of
whether or not rpcb_getport_done() gets called.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

The kernel's NFS client mount option parser currently doesn't allow
unrecognized or incorrect mount options.  This prevents misspellings or
incorrectly specified mount options from possibly causing silent data
corruption.

However, NFS mount options are not standardized, so different operating
systems can use differently spelled mount options to support similar
features, or can support mount options which no other operating system
supports.

"Sloppy" mount option parsing, which allows the parser to ignore any
option it doesn't recognize, is needed to support automounters that often
use maps that are shared between heterogenous operating systems.

The legacy mount command ignores the validity of the values of mount
options entirely, except for the "sec=" and "proto=" options.  If an
incorrect value is specified, the out-of-range value is passed to the
kernel; if a value is specified that contains non-numeric characters,
it appears as though the legacy mount command sets that option to zero
(probably incorrect behavior in general).

In any case, this sets a precedent which we will partially follow for
the kernel mount option parser:

	+ if "sloppy" is not set, the parser will be strict about both
	  unrecognized options (same as legacy) and invalid option
	  values (stricter than legacy)

	+ if "sloppy" is set, the parser will ignore unrecognized
	  options and invalid option values (same as legacy)

An "invalid" option value in this case means that either the type
(integer, short, or string) or sign (for integer values) of the specified
value is incorrect.

This patch does two things: it changes the NFS client's mount option
parsing loop so that it parses the whole string instead of failing at
the first unrecognized option or invalid option value.  An unrecognized
option or an invalid option value cause the option to be skipped.

Then, the patch adds a "sloppy" mount option that allows the parsing
to succeed anyway if there were any problems during parsing.  When
parsing a set of options is complete, if there are errors and "sloppy"
was specified, return success anyway.  Otherwise, only return success
if there are no errors.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Set the default security flavor when we set the other mount option
default values for NFSv4.  This cleans up the NFSv4 mount option parsing
path to look like the NFSv2/v3 one.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Set the default security flavor when we set the other mount option default
values.  After this change, only the legacy user-space mount path needs to
set the NFS_MOUNT_SECFLAVOUR flag.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up: Refactor the NFS mount option parsing function to extract the
security flavor parsing logic into a separate function.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

The remount path does not need to set the port in the server address.
Since it's not really a part of option parsing, move the nfs_set_port()
call to nfs_parse_mount_options()'s callers.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Move the UDP/TCP default timeo/retrans settings for text mounts to
nfs_init_timeout_values(), which was were they were always being
initialised (and sanity checked) for binary mounts.
Document the default timeout values using appropriate #defines.

Ensure that we initialise and sanity check the transport protocols that
may have been specified by the user.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Some server vendors support the higher versions of rpcbind only for
AF_INET6.  The kernel doesn't need to use v3 or v4 for AF_INET anyway,
so change the kernel's rpcbind client to query AF_INET servers over
rpcbind v2 only.

This has a few interesting benefits:

1. If the rpcbind request is going over TCP, and the server doesn't
   support rpcbind versions 3 or 4, the client reduces by two the number
   of ephemeral ports left in TIME_WAIT for each rpcbind request.  This
   will help during NFS mount storms.

2. The rpcbind interaction with servers that don't support rpcbind
   versions 3 or 4 will use less network traffic.  Also helpful
   during mount storms.

3. We can eliminate the kernel build option that controls whether the
   kernel's rpcbind client uses rpcbind version 3 and 4 for AF_INET
   servers.  Less complicated kernel configuration...
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Some rpcbind servers that do support rpcbind version 4 do not support
the GETVERSADDR procedure.  Use GETADDR for querying rpcbind servers
via rpcbind version 4 instead of GETVERSADDR.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up: Change the version 2 procedure name to GETPORT.  It's the same
procedure number as GETADDR, but version 2 implementations usually refer
to it as GETPORT.

This also now matches the procedure name used in the version 2 procedure
entry in the rpcb_next_version[] array, making it slightly less confusing.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up: Replace naked integers that represent rpcbind protocol versions.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up dprintk's in rpcb client's XDR decoder functions.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

If the nfsv4 callback thread is rapidly brought up and down, it's
possible that nfs_callback_svc might never get a chance to run. If
this happens, the cleanup at thread exit might never occur, throwing
the refcounting off and nfs_callback_info in an incorrect state.

Move the clean functions into nfs_callback_down. Also change the
nfs_callback_info struct to track the svc_rqst rather than svc_serv
since we need to know that to call svc_exit_thread.
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

The nfs_callback_mutex is sufficient protection.
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

gcc (4.3.0) rightfully warns about this:
/usr0/export/dev/bhalevy/git/linux-pnfs-bh-nfs41/fs/nfs/nfs4proc.c: In function nfs4_proc_setclientid_confirm:
/usr0/export/dev/bhalevy/git/linux-pnfs-bh-nfs41/fs/nfs/nfs4proc.c:2936: warning: timeout may be used uninitialized in this function

nfs4_delay that's passed a pointer to 'timeout' is looking at its value
and sets it up to some value in the range: NFS4_POLL_RETRY_MIN..NFS4_POLL_RETRY_MAX
	if (*timeout <= 0)
		*timeout = NFS4_POLL_RETRY_MIN;
	if (*timeout > NFS4_POLL_RETRY_MAX)
		*timeout = NFS4_POLL_RETRY_MAX;

Therefore it will end up set to some sane, though rather indeterministic, value.
Signed-off-by: NBenny Halevy <bhalevy@panasas.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Add support in the kernel NFS client's address parser for interface
identifiers.

IPv6 link-local addresses require an additional "interface identifier",
which is a network device name or an integer that indexes the array of
local network interfaces.  They are suffixed to the address with a '%'.
For example:

	fe80::215:c5ff:fe3b:e1b2%2

indicates an interface index of 2.  Or

	fe80::215:c5ff:fe3b:e1b2%eth0

indicates that requests should be routed through the eth0 device.
Without the interface ID, link-local addresses are not usable for NFS.

Both the kernel NFS client mount option parser and the mount.nfs command
can take either form.  The mount.nfs command always passes the address
through getnameinfo(3), which usually re-writes interface indices as
device names.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

To make nfs_parse_server_address() more generally useful, allow it to
accept input strings that are not terminated with '\0'.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Traditionally the mount command has looked for a ":" to separate the
server's hostname from the export path in the mounted on device name,
like this:

	mount server:/export /mounted/on/dir

The server's hostname is "server" and the export path is "/export".

You can also substitute a specific IPv4 network address for the server
hostname, like this:

	mount 192.168.0.55:/export /mounted/on/dir

Raw IPv6 addresses present a problem, however, because they look
something like this:

	fe80::200:5aff:fe00:30b

Note the use of colons.

To get around the presence of colons, copy the Solaris convention used for
mounting IPv6 servers by address: wrap a raw IPv6 address with square
brackets.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

To support passing a raw IPv6 address as a server hostname, we need to
expand the logic that handles splitting the passed-in device name into
a server hostname and export path

Start by pulling device name parsing out of the mount option validation
functions and into separate helper functions.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Fix the 'nfs4_fs_type' undeclared error in nfs_remount when compiling sans
NFSv4...
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>
Cc: Jeff Layton <jlayton@redhat.com>

Currently, if an unstable write completes, we cannot redirty the page in
order to reflect a new change in the page data until after we've sent a
COMMIT request.

This patch allows a page rewrite to proceed without the unnecessary COMMIT
step, putting it immediately back onto the dirty page list, undoing the
VM unstable write accounting, and removing the NFS_PAGE_TAG_COMMIT tag from
the NFS radix tree.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Simplify the loop in nfs_update_request by moving into a separate function
the code that attempts to update an existing cached NFS write.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

Clean up:  the "intr" and "nointr" mount options were recently retired.
Document this in the NFS mount option parser.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

The kernel NFS mount option parser should ignore the retry= mount option
since it is meaningful only in user space.  Today it expects a number
rather than arbitrary text, so it ignores the option if the value is
numeric, but chokes if there are other characters in the value.

Change it to allow any text (except ",") as its value.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

...and ensure that we obey the NFS_INO_INVALID_ACL flag when retrieving the
acls.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

We're not modifying the nfs_server when we call nfs_inc_server_stats and
friends, so allow the compiler to pass 'const' pointers too.
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

The fs/nfs/iostat.h header has definitions that were designed to be exposed
to user space.  Move these definitions under include/linux so user space can
use the definitions in applications that read /proc/self/mountstats.

Also address a handful of coding style issues called out by checkpatch.pl in
fs/nfs/iostat.h.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

All instances are set to nfs_open(), so we should just remove the redundant
indirection. Ditto for the file_release op
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>

The RPC client uses the rq_xtime field in each RPC request to determine the
round-trip time of the request.  Currently, the rq_xtime field is
initialized by each transport just before it starts enqueing a request to
be sent.  However, transports do not handle initializing this value
consistently; sometimes they don't initialize it at all.

To make the measurement of request round-trip time consistent for all
RPC client transport capabilities, pull rq_xtime initialization into the
RPC client's generic transport logic.  Now all transports will get a
standardized RTT measure automatically, from:

  xprt_transmit()

to

  xprt_complete_rqst()

This makes round-trip time calculation more accurate for the TCP transport.
The socket ->sendmsg() method can return "-EAGAIN" if the socket's output
buffer is full, so the TCP transport's ->send_request() method may call
the ->sendmsg() method repeatedly until it gets all of the request's bytes
queued in the socket's buffer.

Currently, the TCP transport sets the rq_xtime field every time through
that loop so the final value is the timestamp just before the *last* call
to the underlying socket's ->sendmsg() method.  After this patch, the
rq_xtime field contains a timestamp that reflects the time just before the
*first* call to ->sendmsg().

This is consequential under heavy workloads because large requests often
take multiple ->sendmsg() calls to get all the bytes of a request queued.
The TCP transport causes the request to sleep until the remote end of the
socket has received enough bytes to clear space in the socket's local
output buffer.  This delay can be quite significant.

The method introduced by this patch is a more accurate measure of RTT
for stream transports, since the server can cause enough back pressure
to delay (ie increase the latency of) requests from the client.

Additionally, this patch corrects the behavior of the RDMA transport, which
entirely neglected to initialize the rq_xtime field.  RPC performance
metrics for RDMA transports now display correct RPC request round trip
times.
Signed-off-by: NChuck Lever <chuck.lever@oracle.com>
Acked-by: NTom Talpey <thomas.talpey@netapp.com>
Signed-off-by: NTrond Myklebust <Trond.Myklebust@netapp.com>