Since many servers (Windows clients, and non-clustered servers) do not
support persistent handles but do support resilient handles, allow
the user to specify a mount option "resilienthandles" in order
to get more reliable connections and less chance of data loss
(at least when SMB2.1 or later).  Default resilient handle
timeout (120 seconds to recent Windows server) is used.
Reviewed-by: NPavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: NSteve French <steve.french@primarydata.com>

Version 2 of the patch. Thanks to Dan Carpenter and the smatch
tool for finding a problem in the first version of this patch.

CC: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: NPavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: NSteve French <steve.french@primarydata.com>

Validate "persistenthandles" and "nopersistenthandles" mount options against
the support the server claims in negotiate and tree connect SMB3 responses.
Signed-off-by: NSteve French <steve.french@primarydata.com>
Reviewed-by: NPavel Shilovsky <pshilovsky@samba.org>

"nopersistenthandles" and "persistenthandles" mount options added.
The former will not request persistent handles on open even when
SMB3 negotiated and Continuous Availability share.  The latter
will request persistent handles (as long as server notes the
capability in protocol negotiation) even if share is not Continuous
Availability share.
Signed-off-by: NSteve French <steve.french@primarydata.com>
Reviewed-by: NPavel Shilovsky <pshilovsky@samba.org>

set integrity increases reliability of files stored on SMB3 servers.
Add ioctl to allow setting this on files on SMB3 and later mounts.
Signed-off-by: NSteve French <steve.french@primarydata.com>

 Getting fantastic copy performance with cp --reflink over SMB3.11
 using the new FSCTL_DUPLICATE_EXTENTS.

 This FSCTL was added in the SMB3.11 dialect (testing was
 against REFS file system) so have put it as a 3.11 protocol
 specific operation ("vers=3.1.1" on the mount).  Tested at
 the SMB3 plugfest in Redmond.

 It depends on the new FS Attribute (BLOCK_REFCOUNTING) which
 is used to advertise support for the ability to do this ioctl
 (if you can support multiple files pointing to the same block
 than this refcounting ability or equivalent is needed to
 support the new reflink-like duplicate extent SMB3 ioctl.
Signed-off-by: NSteve French <steve.french@primarydata.com>

Most people think of SMB 3.1.1 as SMB version 3.11 so add synonym
for "vers=3.1.1" of "vers=3.11" on mount.

Also make sure that unlike SMB3.0 and 3.02 we don't send
validate negotiate on mount (it is handled by negotiate contexts) -
add list of SMB3.11 specific functions (distinct from 3.0 dialect).

Signed-off-by: Steve French <steve.french@primarydata.com>w

Parses and recognizes "vers=3.1.1" on cifs mount and allows sending
0x0311 as a new CIFS/SMB3 dialect. Subsequent patches will add
the new negotiate contexts and updated session setup
Reviewed-by: NJeff Layton <jlayton@primarydata.com>
Signed-off-by: NSteve French <steve.french@primarydata.com>

We have encountered failures when When testing smb2 mounts on ppc64
machines when using both Samba as well as Windows 2012.

On poking around, the problem was determined to be caused by the
high endian MessageID passed in the header for smb2. On checking the
corresponding MID for smb1 is converted to LE before being sent on the
wire.

We have tested this patch successfully on a ppc64 machine.
Signed-off-by: NSachin Prabhu <sprabhu@redhat.com>

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

This is a bigger patch, but its size is mostly due to
a single change for how we check for remapping illegal characters
in file names - a lot of repeated, small changes to
the way callers request converting file names.

The final patch in the series does the following:

1) changes default behavior for cifs to be more intuitive.
Currently we do not map by default to seven reserved characters,
ie those valid in POSIX but not in NTFS/CIFS/SMB3/Windows,
unless a mount option (mapchars) is specified.  Change this
to by default always map and map using the SFM maping
(like the Mac uses) unless the server negotiates the CIFS Unix
Extensions (like Samba does when mounting with the cifs protocol)
when the remapping of the characters is unnecessary.  This should
help SMB3 mounts in particular since Samba will likely be
able to implement this mapping with its new "vfs_fruit" module
as it will be doing for the Mac.
2) if the user specifies the existing "mapchars" mount option then
use the "SFU" (Microsoft Services for Unix, SUA) style mapping of
the seven characters instead.
3) if the user specifies "nomapposix" then disable SFM/MAC style mapping
(so no character remapping would be used unless the user specifies
"mapchars" on mount as well, as above).
4) change all the places in the code that check for the superblock
flag on the mount which is set by mapchars and passed in on all
path based operation and change it to use a small function call
instead to set the mapping type properly (and check for the
mapping type in the cifs unicode functions)
Signed-off-by: NSteve French <smfrench@gmail.com>

The "sfu" mount option did not work on SMB2/SMB3 mounts.
With these changes when the "sfu" mount option is passed in
on an smb2/smb2.1/smb3 mount the client can emulate (and
recognize) fifo and device (character and device files).

In addition the "sfu" mount option should not conflict
with "mfsymlinks" (symlink emulation) as we will never
create "sfu" style symlinks, but using "sfu" mount option
will allow us to recognize existing symlinks, created with
Microsoft "Services for Unix" (SFU and SUA).

To enable the "sfu" mount option for SMB2/SMB3 the calling
syntax of the generic cifs/smb2/smb3 sync_read and sync_write
protocol dependent function needed to be changed (we
don't have a file struct in all cases), but this actually
ended up simplifying the code a little.
Signed-off-by: NSteve French <smfrench@gmail.com>

The existing code uses the old MAX_NAME constant. This causes
XFS test generic/013 to fail. Fix it by replacing MAX_NAME with
PATH_MAX that SMB1 uses. Also remove an unused MAX_NAME constant
definition.

Cc: <stable@vger.kernel.org> # v3.7+
Signed-off-by: NPavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: NSteve French <smfrench@gmail.com>

Implement FALLOC_FL_PUNCH_HOLE (which does not change the file size
fortunately so this matches the behavior of the equivalent SMB3
fsctl call) for SMB3 mounts.  This allows "fallocate -p" to work.
It requires that the server support setting files as sparse
(which Windows allows).
Signed-off-by: NSteve French <smfrench@gmail.com>

SMB2 servers indicates the end of a directory search with
STATUS_NO_MORE_FILE error code that is not processed now.
This causes generic/257 xfstest to fail. Fix this by triggering
the end of search by this error code in SMB2_query_directory.

Also when negotiating CIFS protocol we tell the server to close
the search automatically at the end and there is no need to do
it itself. In the case of SMB2 protocol, we need to close it
explicitly - separate close directory checks for different
protocols.

Cc: <stable@vger.kernel.org>
Signed-off-by: NPavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: NSteve French <smfrench@gmail.com>

Many Linux filesystes make a file "sparse" when extending
a file with ftruncate. This does work for CIFS to Samba
(only) but not for SMB2/SMB3 (to Samba or Windows) since
there is a "set sparse" fsctl which is supposed to be
sent to mark a file as sparse.

This patch marks a file as sparse by sending this simple
set sparse fsctl if it is extended more than 2 pages.
It has been tested to Windows 8.1, Samba and various
SMB2/SMB3 servers which do support setting sparse (and
MacOS which does not appear to support the fsctl yet).
If a server share does not support setting a file
as sparse, then we do not retry setting sparse on that
share.

The disk space savings for sparse files can be quite
large (even more significant on Windows servers than Samba).
Signed-off-by: NSteve French <smfrench@gmail.com>
Reviewed-by: NShirish Pargaonkar <spargaonkar@suse.com>

and don't mix it with the number of bytes that was requested.
Signed-off-by: NPavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: NSteve French <smfrench@gmail.com>

If we negotiate SMB 2.1 and higher version of the protocol and
a server supports large read buffer size, we need to consume 1
credit per 65536 bytes. So, we need to know how many credits
we have and obtain the required number of them before constructing
a readdata structure in readpages and user read.
Reviewed-by: NShirish Pargaonkar <spargaonkar@suse.com>
Signed-off-by: NPavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: NSteve French <smfrench@gmail.com>

If we negotiate SMB 2.1 and higher version of the protocol and
a server supports large write buffer size, we need to consume 1
credit per 65536 bytes. So, we need to know how many credits
we have and obtain the required number of them before constructing
a writedata structure in writepages and iovec write.
Reviewed-by: NShirish Pargaonkar <spargaonkar@suse.com>
Signed-off-by: NPavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: NSteve French <smfrench@gmail.com>

If wsize changes on reconnect we need to use new writedata structure
that for retrying.
Reviewed-by: NShirish Pargaonkar <spargaonkar@suse.com>
Signed-off-by: NPavel Shilovsky <pshilovsky@samba.org>
Signed-off-by: NSteve French <smfrench@gmail.com>

When mounting from a Windows 2012R2 server, we hit the following
problem:
1) Mount with any of the following versions - 2.0, 2.1 or 3.0
2) unmount
3) Attempt a mount again using a different SMB version >= 2.0.

You end up with the following failure:
Status code returned 0xc0000203 STATUS_USER_SESSION_DELETED
CIFS VFS: Send error in SessSetup = -5
CIFS VFS: cifs_mount failed w/return code = -5

I cannot reproduce this issue using a Windows 2008 R2 server.

This appears to be caused because we use the same client guid for the
connection on first mount which we then disconnect and attempt to mount
again using a different protocol version. By generating a new guid each
time a new connection is Negotiated, we avoid hitting this problem.
Signed-off-by: NSachin Prabhu <sprabhu@redhat.com>
Signed-off-by: NSteve French <smfrench@gmail.com>

The handling of the CIFS_INO_INVALID_MAPPING flag is racy. It's possible
for two tasks to attempt to revalidate the mapping at the same time. The
first sees that CIFS_INO_INVALID_MAPPING is set. It clears the flag and
then calls invalidate_inode_pages2 to start shooting down the pagecache.

While that's going on, another task checks the flag and sees that it's
clear. It then ends up trusting the pagecache to satisfy a read when it
shouldn't.

Fix this by adding a bitlock to ensure that the clearing of the flag is
atomic with respect to the actual cache invalidation. Also, move the
other existing users of cifs_invalidate_mapping to use a new
cifs_zap_mapping() function that just sets the INVALID_MAPPING bit and
then uses the standard codepath to handle the invalidation.
Signed-off-by: NJeff Layton <jlayton@poochiereds.net>
Signed-off-by: NSteve French <smfrench@gmail.com>

In later patches, we'll need to have a bitlock, so go ahead and convert
these bools to use atomic bitops instead.

Also, clean up the initialization of the flags field. There's no need
to unset each bit individually just after it was zeroed on allocation.
Signed-off-by: NJeff Layton <jlayton@poochiereds.net>
Signed-off-by: NSteve French <smfrench@gmail.com>

Problem reported in Red Hat bz 1040329 for strict writes where we cache
only when we hold oplock and write direct to the server when we don't.

When we receive an oplock break, we first change the oplock value for
the inode in cifsInodeInfo->oplock to indicate that we no longer hold
the oplock before we enqueue a task to flush changes to the backing
device. Once we have completed flushing the changes, we return the
oplock to the server.

There are 2 ways here where we can have data corruption
1) While we flush changes to the backing device as part of the oplock
break, we can have processes write to the file. These writes check for
the oplock, find none and attempt to write directly to the server.
These direct writes made while we are flushing from cache could be
overwritten by data being flushed from the cache causing data
corruption.
2) While a thread runs in cifs_strict_writev, the machine could receive
and process an oplock break after the thread has checked the oplock and
found that it allows us to cache and before we have made changes to the
cache. In that case, we end up with a dirty page in cache when we
shouldn't have any. This will be flushed later and will overwrite all
subsequent writes to the part of the file represented by this page.

Before making any writes to the server, we need to confirm that we are
not in the process of flushing data to the server and if we are, we
should wait until the process is complete before we attempt the write.
We should also wait for existing writes to complete before we process
an oplock break request which changes oplock values.

We add a version specific  downgrade_oplock() operation to allow for
differences in the oplock values set for the different smb versions.

Cc: stable@vger.kernel.org
Signed-off-by: NSachin Prabhu <sprabhu@redhat.com>
Reviewed-by: NJeff Layton <jlayton@redhat.com>
Reviewed-by: NPavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: NSteve French <smfrench@gmail.com>

The rfc1002 length actually includes a type byte, which we aren't
masking off. In most cases, it's not a problem since the
RFC1002_SESSION_MESSAGE type is 0, but when doing a RFC1002 session
establishment, the type is non-zero and that throws off the returned
length.
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Tested-by: NSachin Prabhu <sprabhu@redhat.com>
Signed-off-by: NSteve French <smfrench@gmail.com>

When mounting with smb2/smb3 (e.g. vers=2.1) and cifsacl mount option,
it was trying to get the mode by querying the acl over the cifs
rather than smb2 protocol.  This patch makes that protocol
independent and makes cifsacl smb2 mounts return a more intuitive
operation not supported error (until we add a worker function
for smb2_get_acl).

Note that a previous patch fixed getxattr/setxattr for the CIFSACL xattr
which would unconditionally call cifs_get_acl and cifs_set_acl (even when
mounted smb2). I made those protocol independent last week (new protocol
version operations "get_acl" and "set_acl" but did not add an
smb2_get_acl and smb2_set_acl yet so those now simply return EOPNOTSUPP
which at least is better than sending cifs requests on smb2 mount)

The previous patches did not fix the one remaining case though ie
mounting with "cifsacl" when getting mode from acl would unconditionally
end up calling "cifs_get_acl_from_fid" even for smb2 - so made that protocol
independent but to make that protocol independent had to make sure that the callers
were passing the protocol independent handle structure (cifs_fid) instead
of cifs specific _u16 network file handle (ie cifs_fid instead of cifs_fid->fid)

Now mount with smb2 and cifsacl mount options will return EOPNOTSUP (instead
of timing out) and a future patch will add smb2 operations (e.g. get_smb2_acl)
to enable this.
Signed-off-by: NSteve French <smfrench@gmail.com>

In the event that a send fails in an uncached write, or we end up
needing to reissue it (-EAGAIN case), we'll kfree the wdata but
the pages currently leak.

Fix this by adding a new kref release routine for uncached writedata
that releases the pages, and have the uncached codepaths use that.

[original patch by Jeff modified to fix minor formatting problems]
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Reviewed-by: NPavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: NSteve French <smfrench@gmail.com>

The cifs_writedata code uses a single element trailing array, which
just adds unneeded complexity. Use a flexarray instead.
Signed-off-by: NJeff Layton <jlayton@redhat.com>
Reviewed-by: NPavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: NSteve French <smfrench@gmail.com>

The get/set ACL xattr support for CIFS ACLs attempts to send old
cifs dialect protocol requests even when mounted with SMB2 or later
dialects. Sending cifs requests on an smb2 session causes problems -
the server drops the session due to the illegal request.

This patch makes CIFS ACL operations protocol specific to fix that.

Attempting to query/set CIFS ACLs for SMB2 will now return
EOPNOTSUPP (until we add worker routines for sending query
ACL requests via SMB2) instead of sending invalid (cifs)
requests.

A separate followon patch will be needed to fix cifs_acl_to_fattr
(which takes a cifs specific u16 fid so can't be abstracted
to work with SMB2 until that is changed) and will be needed
to fix mount problems when "cifsacl" is specified on mount
with e.g. vers=2.1
Signed-off-by: NSteve French <smfrench@gmail.com>
Reviewed-by: NShirish Pargaonkar <spargaonkar@suse.com>
CC: Stable <stable@kernel.org>

When mounting with smb2 (or smb2.1 or smb3) we need to check to make
sure that attempts to query or set extended attributes do not
attempt to send the request with the older cifs protocol instead
(eventually we also need to add the support in SMB2
to query/set extended attributes but this patch prevents us from
using the wrong protocol for extended attribute operations).
Signed-off-by: NSteve French <smfrench@gmail.com>

Add a new protocol ops function create_mf_symlink and have
create_mf_symlink() use it.

This patchset moves the MFSymlink operations completely to the
ops structure so that we only use the right protocol versions when
querying or creating MFSymlinks.
Signed-off-by: NSachin Prabhu <sprabhu@redhat.com>
Reviewed-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NSteve French <smfrench@gmail.com>

Rename open_query_close_cifs_symlink to cifs_query_mf_symlink() to make
the name more consistent with other protocol version specific functions.

We also pass tcon as an argument to the function. This is already
available in the calling functions and we can avoid having to make an
unnecessary lookup.
Signed-off-by: NSachin Prabhu <sprabhu@redhat.com>
Reviewed-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NSteve French <smfrench@gmail.com>

When we are running SMB3 or SMB3.02 connections which are signed
we need to validate the protocol negotiation information,
to ensure that the negotiate protocol response was not tampered with.

Add the missing FSCTL which is sent at mount time (immediately after
the SMB3 Tree Connect) to validate that the capabilities match
what we think the server sent.

"Secure dialect negotiation is introduced in SMB3 to protect against
man-in-the-middle attempt to downgrade dialect negotiation.
The idea is to prevent an eavesdropper from downgrading the initially
negotiated dialect and capabilities between the client and the server."

For more explanation see 2.2.31.4 of MS-SMB2 or
http://blogs.msdn.com/b/openspecification/archive/2012/06/28/smb3-secure-dialect-negotiation.aspxReviewed-by: NPavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: NSteve French <smfrench@gmail.com>

Patch 2 of the copy chunk series (the final patch will
use these to handle copies of files larger than the chunk size.

We set the same defaults that Windows and Samba expect for
CopyChunk.
Signed-off-by: NSteve French <smfrench@gmail.com>
Reviewed-by: NDavid Disseldorp <ddiss@samba.org>

This first patch adds the ability for us to do a server side copy
(ie fast copy offloaded to the server to perform, aka refcopy)

"cp --reflink"

of one file to another located on the same server.  This
is much faster than traditional copy (which requires
reading and writing over the network and extra
memcpys).

This first version is not going to be copy
files larger than about 1MB (to Samba) until I add
support for multiple chunks and for autoconfiguring
the chunksize.

It includes:
1) processing of the ioctl
2) marshalling and sending the SMB2/SMB3 fsctl over the network
3) simple parsing of the response

It does not include yet (these will be in followon patches to come soon):
1) support for multiple chunks
2) support for autoconfiguring and remembering the chunksize
3) Support for the older style copychunk which Samba 4.1 server supports
(because this requires write permission on the target file, which
cp does not give you, apparently per-posix).  This may require
a distinct tool (other than cp) and other ioctl to implement.
Reviewed-by: NPavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: NSteve French <smfrench@gmail.com>

Now we treat any reparse point as a symbolic link and map it to a Unix
one that is not true in a common case due to many reparse point types
supported by SMB servers.

Distinguish reparse point types into two groups:
1) that can be accessed directly through a reparse point
(junctions, deduplicated files, NFS symlinks);
2) that need to be processed manually (Windows symbolic links, DFS);

and map only Windows symbolic links to Unix ones.

Cc: <stable@vger.kernel.org>
Acked-by: NJeff Layton <jlayton@redhat.com>
Reported-and-tested-by: NJoao Correia <joaomiguelcorreia@gmail.com>
Signed-off-by: NPavel Shilovsky <piastry@etersoft.ru>
Signed-off-by: NSteve French <smfrench@gmail.com>

In SMB3 it is now possible to query the file system
alignment info, and the preferred (for performance)
sector size and whether the underlying disk
has no seek penalty (like SSD).

Query this information at mount time for SMB3,
and make it visible in /proc/fs/cifs/DebugData
for debugging purposes.

This alignment information and preferred sector
size info will be helpful for the copy offload
patches to setup the right chunks in the CopyChunk
requests.   Presumably the knowledge that the
underlying disk is SSD could also help us
make better readahead and writebehind
decisions (something to look at in the future).
Signed-off-by: NSteve French <smfrench@gmail.com>

The multiplex identifier (MID) in the SMB header is only
ever used by the client, in conjunction with PID, to match responses
from the server. As such, the endianess of the MID is not important.
However, When tracing packet sequences on the wire, protocol analyzers
such as wireshark display MID as little endian. It is much more informative
for the on-the-wire MID sequences to match debug information emitted by the
CIFS driver.  Therefore, one should write and read MID in the SMB header
assuming it is always little endian.

Observed from wireshark during the protocol negotiation
and session setup:

        Multiplex ID: 256
        Multiplex ID: 256
        Multiplex ID: 512
        Multiplex ID: 512
        Multiplex ID: 768
        Multiplex ID: 768

After this patch on-the-wire MID values begin at 1 and increase monotonically.

Introduce get_next_mid64() for the internal consumers that use the full 64 bit
multiplex identifier.

Introduce the helpers get_mid() and compare_mid() to make the endian
translation clear.
Reviewed-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NTim Gardner <timg@tpi.com>
Signed-off-by: NSteve French <smfrench@gmail.com>

Allow cifs/smb2/smb3 to return whether or not a file is compressed
via lsattr, and allow SMB2/SMB3 to set the per-file compression
flag ("chattr +c filename" on an smb3 mount).

Windows users often set the compressed flag (it can be
done from the desktop and file manager).  David Disseldorp
has patches to Samba server to support this (at least on btrfs)
which are complementary to this
Signed-off-by: NSteve French <smfrench@gmail.com>

Since we don't get info about the number of links from the readdir
linfo levels, stat() will return 0 for st_nlink, and in particular,
samba re-exported shares will show directories as files (as samba is
keying off st_nlink before evaluating how to set the dos modebits)
when doing a dir or ls.

Copy nlink to the inode, unless it wasn't provided.  Provide
sane values if we don't have an existing one and none was provided.
Signed-off-by: NJim McDonough <jmcd@samba.org>
Reviewed-by: NJeff Layton <jlayton@redhat.com>
Reviewed-by: NDavid Disseldorp <ddiss@samba.org>
Signed-off-by: NSteve French <smfrench@gmail.com>