Be more careful when tearing down the subrequests of an O_DIRECT write
as part of a retransmission.

Reported-by: Chris Mason <clm@fb.com>
Fixes: ed5d588f

 ("NFS: Try to join page groups before an O_DIRECT retransmission")
Cc: stable@vger.kernel.org
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Commmit f5ea1613

 ("NFSv4: Retry LOCK on OLD_STATEID during delegation
return") attempted to solve this problem by using nfs4's generic async error
handling, but introduced a regression where v4.0 lock recovery would hang.
The additional complexity introduced by overloading that error handling is
not necessary for this case.  This patch expects that commit to be
reverted.

The problem as originally explained in the above commit is:

    There's a small window where a LOCK sent during a delegation return can
    race with another OPEN on client, but the open stateid has not yet been
    updated.  In this case, the client doesn't handle the OLD_STATEID error
    from the server and will lose this lock, emitting:
    "NFS: nfs4_handle_delegation_recall_error: unhandled error -10024".

Fix this by using the old_stateid refresh helpers if the server replies
with OLD_STATEID.

Suggested-by: Trond Myklebust <trondmy@hammerspace.com>
Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Olga Kornievskaia reports that this patch breaks NFSv4.0 state recovery.
It also introduces additional complexity in the error paths for cases not
related to the original problem.  Let's revert it for now, and address the
original problem in another manner.

This reverts commit f5ea1613.

Fixes: f5ea1613

 ("NFSv4: Retry LOCK on OLD_STATEID during delegation return")
Reported-by: Kornievskaia, Olga <Olga.Kornievskaia@netapp.com>
Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Its possible to end up in nfs_free_server() before the server's superblock
sysfs entry has been initialized, in which case calling kobject_put() will
emit a WARNING.  Check if the kobject has been initialized before cleaning
it up.

Fixes: 1c725118

 ("NFS: add superblock sysfs entries")
Reported-by: Nathan Chancellor <nathan@kernel.org>
Tested-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

NFS doesn't properly support reporting the btime in getattr (yet), but
61a968b4 mistakenly added it to the request_mask. This causes statx
for STATX_BTIME to report a zeroed out btime instead of properly
clearing the flag.

Cc: stable@vger.kernel.org # v6.3+
Fixes: 61a968b4

 ("nfs: report the inode version in getattr if requested")
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Link: https://bugzilla.redhat.com/show_bug.cgi?id=2214134


Reported-by: Boyang Xue <bxue@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

When the client received NFS4ERR_BADSESSION, it schedules recovery
and start the state manager thread which in turn freezes the
session table and does not allow for any new requests to use the
no-longer valid session. However, it is possible that before
the state manager thread runs, a new operation would use the
released slot that received BADSESSION and was therefore not
updated its sequence number. Such re-use of the slot can lead
the application errors.

Fixes: 5c441544

 ("NFSv4.x: Handle bad/dead sessions correctly in nfs41_sequence_process()")
Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Currently, the list_lru::shrinker_id corresponding to the nfs4_xattr
shrinkers is wrong:

>>> prog["nfs4_xattr_cache_lru"].shrinker_id
(int)0
>>> prog["nfs4_xattr_entry_lru"].shrinker_id
(int)0
>>> prog["nfs4_xattr_large_entry_lru"].shrinker_id
(int)0
>>> prog["nfs4_xattr_cache_shrinker"].id
(int)18
>>> prog["nfs4_xattr_entry_shrinker"].id
(int)19
>>> prog["nfs4_xattr_large_entry_shrinker"].id
(int)20

This is not what we expect, which will cause these shrinkers
not to be found in shrink_slab_memcg().

We should assign shrinker::id before calling list_lru_init_memcg(),
so that the corresponding list_lru::shrinker_id will be assigned
the correct value like below:

>>> prog["nfs4_xattr_cache_lru"].shrinker_id
(int)16
>>> prog["nfs4_xattr_entry_lru"].shrinker_id
(int)17
>>> prog["nfs4_xattr_large_entry_lru"].shrinker_id
(int)18
>>> prog["nfs4_xattr_cache_shrinker"].id
(int)16
>>> prog["nfs4_xattr_entry_shrinker"].id
(int)17
>>> prog["nfs4_xattr_large_entry_shrinker"].id
(int)18

So just do it.

Fixes: 95ad37f9

 ("NFSv4.2: add client side xattr caching.")
Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

If a SEQUENCE call receives -EIO for a shutdown client, it will retry the
RPC call.  Instead of doing that for a shutdown client, just bail out.

Likewise, if the state manager decides to perform recovery for a shutdown
client, it will continuously retry.  As above, just bail out.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Walk existing RPC tasks and cancel them with -EIO when the client is
shutdown.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Within each nfs_server sysfs tree, add an entry named "shutdown".  Writing
1 to this file will set the cl_shutdown bit on the rpc_clnt structs
associated with that mount.  If cl_shutdown is set, the task scheduler
immediately returns -EIO for new tasks.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

After lockd is started, add a symlink for lockd's rpc_client under
NFS' superblock sysfs.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

For the general and state management nfs_client under each mount, create
symlinks to their respective rpc_client sysfs entries.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Create a sysfs directory for each mount that corresponds to the mount's
nfs_server struct.  As the mount is being constructed, use the name
"server-n", but rename it to the "MAJOR:MINOR" of the mount after assigning
a device_id. The rename approach allows us to populate the mount's directory
with links to the various rpc_client objects during the mount's
construction.  The naming convention (MAJOR:MINOR) can be used to reference
a particular NFS mount's sysfs tree.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Expand the NFS network-namespaced sysfs from /sys/fs/nfs/net down one level
into /sys/fs/nfs by moving the "net" kobject onto struct
nfs_netns_client and setting it up during network namespace init.

This prepares the way for superblock kobjects within /sys/fs/nfs that will
only be visible to matching network namespaces.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

In preparation to make objects below /sys/fs/nfs namespace aware, we need
to define our own kobj_type for the nfs kset so that we can add the
.child_ns_type member in a following patch.  No functional change here, only
the unrolling of kset_create_and_add().

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Match the variable names to the sysfs structure.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Be brief and match the subsystem name.  There's no need to distinguish this
kset variable from the server.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

After some discussion, we decided that controlling transport layer
security policy should be separate from the setting for the user
authentication flavor. To accomplish this, add a new NFS mount
option to select a transport layer security policy for RPC
operations associated with the mount point.

  xprtsec=none     - Transport layer security is forced off.

  xprtsec=tls      - Establish an encryption-only TLS session. If
                     the initial handshake fails, the mount fails.
                     If TLS is not available on a reconnect, drop
                     the connection and try again.

  xprtsec=mtls     - Both sides authenticate and an encrypted
                     session is created. If the initial handshake
                     fails, the mount fails. If TLS is not available
                     on a reconnect, drop the connection and try
                     again.

To support client peer authentication (mtls), the handshake daemon
will have configurable default authentication material (certificate
or pre-shared key). In the future, mount options can be added that
can provide this material on a per-mount basis.

Updates to mount.nfs (to support xprtsec=auto) and nfs(5) will be
sent under separate cover.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

The new field is used to match struct nfs_clients that have the same
TLS policy setting.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Use the new TLS handshake API to enable the SunRPC client code
to request a TLS handshake. This implements support for RFC 9289,
only on TCP sockets.

Upper layers such as NFS use RPC-with-TLS to protect in-transit
traffic.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

kTLS sockets use CMSG to report decryption errors and the need
for session re-keying.

For RPC-with-TLS, an "application data" message contains a ULP
payload, and that is passed along to the RPC client. An "alert"
message triggers connection reset. Everything else is discarded.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

The RPC header parser doesn't recognize TLS handshake traffic, so it
will close the connection prematurely with an error. To avoid that,
shunt the transport's data_ready callback when there is a TLS
handshake in progress.

The XPRT_SOCK_IGNORE_RECV flag will be toggled by code added in a
subsequent patch.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

The new authentication flavor is used only to discover peer support
for RPC-over-TLS.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Pass the upper layer's rpc_create_args to the rpc_clnt_new()
tracepoint so additional parts of the upper layer's request can be
recorded.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Add an initial set of policies along with fields for upper layers to
pass the requested policy down to the transport layer.

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Add some missing observability to the fs_context tracepoints
added by commit 33ce83ef

 ("NFS: Replace fs_context-related
dprintk() call sites with tracepoints").

Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

NFS is primarily name-spaced using network namespaces.  However it
contacts rpcbind (and gss_proxy) using AF_UNIX sockets which are
name-spaced using the mount namespaces.  This requires a container using
NFSv3 (the form that requires rpcbind) to manage both network and mount
namespaces, which can seem an unnecessary burden.

As NFS is primarily a network service it makes sense to use network
namespaces as much as possible, and to prefer to communicate with an
rpcbind running in the same network namespace.  This can be done, while
preserving the benefits of AF_UNIX sockets, by using an abstract socket
address.

An abstract address has a nul at the start of sun_path, and a length
that is exactly the complete size of the sockaddr_un up to the end of
the name, NOT including any trailing nul (which is not part of the
address).
Abstract addresses are local to a network namespace - regular AF_UNIX
path names a resolved in the mount namespace ignoring the network
namespace.

This patch causes rpcb to first try an abstract address before
continuing with regular AF_UNIX and then IP addresses.  This ensures
backwards compatibility.

Choosing the name needs some care as the same address will be configured
for rpcbind, and needs to be built in to libtirpc for this enhancement
to be fully successful.  There is no formal standard for choosing
abstract addresses.  The defacto standard appears to be to use a path
name similar to what would be used for a filesystem AF_UNIX address -
but with a leading nul.

In that case
   "\0/var/run/rpcbind.sock"
seems like the best choice.  However at this time /var/run is deprecated
in favour of /run, so
   "\0/run/rpcbind.sock"
might be better.
Though as we are deliberately moving away from using the filesystem it
might seem more sensible to explicitly break the connection and just
have
   "\0rpcbind.socket"
using the same name as the systemd unit file..

This patch chooses the second option, which seems least likely to raise
objections.

Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

An "abtract" address for an AF_UNIX socket start with a nul and can
contain any bytes for the given length, but traditionally doesn't
contain other nuls.  When reported, the leading nul is replaced by '@'.

sunrpc currently rejects connections to these addresses and reports them
as an empty string.  To provide support for future use of these
addresses, allow them for outgoing connections and report them more
usefully.

Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Otherwise, `stat` will report a stale value to users.

Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Fold them into the other NFS v4.2 operations in the right spots and
adjust spacing to keep the same style.

Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

I add commends above each function to match the style of the other
nfs4_xdr_dec_*() functions. I also remove the unnecessary #ifdef
CONFIG_NFS_V4_2 that was added around this code, since we are already in
a v4.2-only file.

Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

They should be in the nfs4_xdr_enc_*() section, and not at the bottom of
the file.

Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Move them out of the encode_*() section and into the decode_*() section
where it belongs.

Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Move the function to be with the other encode_*() functions, instead of
in the middle of the nfs4_xdr_enc_*() section.

Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>

Pull SCSI fixes from James Bottomley:
 "Four fixes, all in drivers: three fairly obvious small ones and a
  large one in aacraid to add block queue completion mapping and fix a
  CPU offline hang"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: lpfc: Fix incorrect big endian type assignment in bsg loopback path
  scsi: target: core: Fix error path in target_setup_session()
  scsi: storvsc: Always set no_report_opcodes
  scsi: aacraid: Reply queue mapping to CPUs based on IRQ affinity

Pull ata fix from Damien Le Moal:

 - Avoid deadlocks on resume from sleep by delaying scsi rescan until
   the scsi device is also fully resumed.

* tag 'ata-6.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal/libata:
  ata: libata-scsi: Avoid deadlock on rescan after device resume

Pull parisc fix from Helge Deller:

 - Drop redundant register definitions to fix build with latest binutils

* tag 'parisc-for-6.4-4' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Delete redundant register definitions in <asm/assembly.h>

The error unrolling was leaving the VMAs detached in many cases and
leaving the locked_vm statistic altered, and skipping the unrolling
entirely in the case of the vma tree write failing.

Fix the error path by re-attaching the detached VMAs and adding the
necessary goto for the failed vma tree write, and fix the locked_vm
statistic by only updating after the vma tree write succeeds.

Fixes: 763ecb03

 ("mm: remove the vma linked list")
Reported-by: Vegard Nossum <vegard.nossum@oracle.com>
Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>