Make all IO sizes multiple of PAGE_SIZE, either negotiated by the
server or passed through rsize, wsize and bsize mount options, to
prevent from breaking DIO reads and writes against servers that
enforce alignment as specified in MS-FSA 2.1.5.3 and 2.1.5.4.
Cc: linux-cifs@vger.kernel.org
Reviewed-by: David Howells <dhowells@redhat.com>
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
This reverts commit e9f2517a3e.
Commit e9f2517a3e ("smb: client: fix TCP timers deadlock after
rmmod") is intended to fix a null-ptr-deref in LOCKDEP, which is
mentioned as CVE-2024-54680, but is actually did not fix anything;
The issue can be reproduced on top of it. [0]
Also, it reverted the change by commit ef7134c7fc ("smb: client:
Fix use-after-free of network namespace.") and introduced a real
issue by reviving the kernel TCP socket.
When a reconnect happens for a CIFS connection, the socket state
transitions to FIN_WAIT_1. Then, inet_csk_clear_xmit_timers_sync()
in tcp_close() stops all timers for the socket.
If an incoming FIN packet is lost, the socket will stay at FIN_WAIT_1
forever, and such sockets could be leaked up to net.ipv4.tcp_max_orphans.
Usually, FIN can be retransmitted by the peer, but if the peer aborts
the connection, the issue comes into reality.
I warned about this privately by pointing out the exact report [1],
but the bogus fix was finally merged.
So, we should not stop the timers to finally kill the connection on
our side in that case, meaning we must not use a kernel socket for
TCP whose sk->sk_net_refcnt is 0.
The kernel socket does not have a reference to its netns to make it
possible to tear down netns without cleaning up every resource in it.
For example, tunnel devices use a UDP socket internally, but we can
destroy netns without removing such devices and let it complete
during exit. Otherwise, netns would be leaked when the last application
died.
However, this is problematic for TCP sockets because TCP has timers to
close the connection gracefully even after the socket is close()d. The
lifetime of the socket and its netns is different from the lifetime of
the underlying connection.
If the socket user does not maintain the netns lifetime, the timer could
be fired after the socket is close()d and its netns is freed up, resulting
in use-after-free.
Actually, we have seen so many similar issues and converted such sockets
to have a reference to netns.
That's why I converted the CIFS client socket to have a reference to
netns (sk->sk_net_refcnt == 1), which is somehow mentioned as out-of-scope
of CIFS and technically wrong in e9f2517a3e, but **is in-scope and right
fix**.
Regarding the LOCKDEP issue, we can prevent the module unload by
bumping the module refcount when switching the LOCKDDEP key in
sock_lock_init_class_and_name(). [2]
For a while, let's revert the bogus fix.
Note that now we can use sk_net_refcnt_upgrade() for the socket
conversion, but I'll do so later separately to make backport easy.
Link: https://lore.kernel.org/all/20250402020807.28583-1-kuniyu@amazon.com/ #[0]
Link: https://lore.kernel.org/netdev/c08bd5378da647a2a4c16698125d180a@huawei.com/ #[1]
Link: https://lore.kernel.org/lkml/20250402005841.19846-1-kuniyu@amazon.com/ #[2]
Fixes: e9f2517a3e ("smb: client: fix TCP timers deadlock after rmmod")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Cc: stable@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
This reverts commit 4e7f1644f2.
The commit e9f2517a3e ("smb: client: fix TCP timers deadlock after
rmmod") is not only a bogus fix for LOCKDEP null-ptr-deref but also
introduces a real issue, TCP sockets leak, which will be explained in
detail in the next revert.
Also, CNA assigned CVE-2024-54680 to it but is rejecting it. [0]
Thus, we are reverting the commit and its follow-up commit 4e7f1644f2
("smb: client: Fix netns refcount imbalance causing leaks and
use-after-free").
Link: https://lore.kernel.org/all/2025040248-tummy-smilingly-4240@gregkh/ #[0]
Fixes: 4e7f1644f2 ("smb: client: Fix netns refcount imbalance causing leaks and use-after-free")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Cc: stable@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
When mounting the same share twice, once with the "linux" mount parameter
(or equivalently "posix") and then once without (or e.g. with "nolinux"),
we were incorrectly reusing the same tree connection for both mounts.
This meant that the first mount of the share on the client, would
cause subsequent mounts of that same share on the same client to
ignore that mount parm ("linux" vs. "nolinux") and incorrectly reuse
the same tcon.
Cc: stable@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
SMB negotiate retry functionality in cifs_negotiate() is currently broken
and does not work when doing socket reconnect. Caller of this function,
which is cifs_negotiate_protocol() requires that tcpStatus after successful
execution of negotiate callback stay in CifsInNegotiate. But if the
CIFSSMBNegotiate() called from cifs_negotiate() fails due to connection
issues then tcpStatus is changed as so repeated CIFSSMBNegotiate() call
does not help.
Fix this problem by moving retrying code from negotiate callback (which is
either cifs_negotiate() or smb2_negotiate()) to cifs_negotiate_protocol()
which is caller of those callbacks. This allows to properly handle and
implement correct transistions between tcpStatus states as function
cifs_negotiate_protocol() already handles it.
With this change, cifs_negotiate_protocol() now handles also -EAGAIN error
set by the RFC1002_NEGATIVE_SESSION_RESPONSE processing after reconnecting
with NetBIOS session.
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Now all NetBIOS session logic is handled in ip_rfc1001_connect() function,
so cleanup is_smb_response() function which contains generic handling of
incoming SMB packets. Note that function is_smb_response() is not used
directly or indirectly (e.g. over cifs_demultiplex_thread() by
ip_rfc1001_connect() function.
Except the Negative Session Response and the Session Keep Alive packet, the
cifs_demultiplex_thread() should not receive any NetBIOS session packets.
And Session Keep Alive packet may be received only when the NetBIOS session
was established by ip_rfc1001_connect() function. So treat any such packet
as error and schedule reconnect.
Negative Session Response packet is returned from Windows SMB server (from
Windows 98 and also from Windows Server 2022) if client sent over port 139
SMB negotiate request without previously establishing a NetBIOS session.
The common scenario is that Negative Session Response packet is returned
for the SMB negotiate packet, which is the first one which SMB client
sends (if it is not establishing a NetBIOS session).
Note that server port 139 may be forwarded and mapped between virtual
machines to different number. And Linux SMB client do not call function
ip_rfc1001_connect() when prot is not 139. So nowadays when using port
mapping or port forwarding between VMs, it is not so uncommon to see this
error.
Currently the logic on Negative Session Response packet changes server port
to 445 and force reconnection. But this logic does not work when using
non-standard port numbers and also does not help if the server on specified
port is requiring establishing a NetBIOS session.
Fix this Negative Session Response logic and instead of changing server
port (on which server does not have to listen), force reconnection with
establishing a NetBIOS session.
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Currently SMB client always tries to initialize NetBIOS session when the
server port is 139. This is useful for default cases, but nowadays when
using non-standard routing or testing between VMs, it is common that
servers are listening on non-standard ports.
So add a new mount option -o nbsessinit and -o nonbsessinit which either
forces initialization or disables initialization regardless of server port
number.
This allows Linux SMB client to connect to older SMB1 server listening on
non-standard port, which requires initialization of NetBIOS session, by
using additional mount options -o port= and -o nbsessinit.
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Currently, when a SMB connection is reset and renegotiated with the
server, there's no way to update all related mount points with new
negotiated sizes. This is because while superblocks (cifs_sb_info)
maintain references to tree connections (tcon) through tcon_link
structures, there is no reverse mapping from a tcon back to all the
superblocks using it.
This patch adds a bidirectional relationship between tcon and
cifs_sb_info structures by:
1. Adding a cifs_sb_list to tcon structure with appropriate locking
2. Adding tcon_sb_link to cifs_sb_info to join the list
3. Managing the list entries during mount and umount operations
The bidirectional relationship enables future functionality to locate and
update all superblocks connected to a specific tree connection, such as:
- Updating negotiated parameters after reconnection
- Efficiently notifying all affected mounts of capability changes
This is the first part of a series to improve connection resilience
by keeping all mount parameters in sync with server capabilities
after reconnection.
Signed-off-by: Wang Zhaolong <wangzhaolong1@huawei.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
echo_interval is checked at mount time, the code has become
unreachable.
Signed-off-by: Roman Smirnov <r.smirnov@omp.ru>
Signed-off-by: Steve French <stfrench@microsoft.com>
SMB1 protocol supports non-UNICODE (8-bit OEM character set) and
UNICODE (UTF-16) modes.
Linux SMB1 client implements both of them but currently does not allow to
choose non-UNICODE mode when SMB1 server announce UNICODE mode support.
This change adds a new mount option -o nounicode to disable UNICODE mode
and force usage of non-UNICODE (8-bit OEM character set) mode.
This allows to test non-UNICODE implementation of Linux SMB1 client against
any SMB1 server, including modern and recent Windows SMB1 server.
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Commit ef7134c7fc ("smb: client: Fix use-after-free of network
namespace.") attempted to fix a netns use-after-free issue by manually
adjusting reference counts via sk->sk_net_refcnt and sock_inuse_add().
However, a later commit e9f2517a3e ("smb: client: fix TCP timers deadlock
after rmmod") pointed out that the approach of manually setting
sk->sk_net_refcnt in the first commit was technically incorrect, as
sk->sk_net_refcnt should only be set for user sockets. It led to issues
like TCP timers not being cleared properly on close. The second commit
moved to a model of just holding an extra netns reference for
server->ssocket using get_net(), and dropping it when the server is torn
down.
But there remain some gaps in the get_net()/put_net() balancing added by
these commits. The incomplete reference handling in these fixes results
in two issues:
1. Netns refcount leaks[1]
The problem process is as follows:
```
mount.cifs cifsd
cifs_do_mount
cifs_mount
cifs_mount_get_session
cifs_get_tcp_session
get_net() /* First get net. */
ip_connect
generic_ip_connect /* Try port 445 */
get_net()
->connect() /* Failed */
put_net()
generic_ip_connect /* Try port 139 */
get_net() /* Missing matching put_net() for this get_net().*/
cifs_get_smb_ses
cifs_negotiate_protocol
smb2_negotiate
SMB2_negotiate
cifs_send_recv
wait_for_response
cifs_demultiplex_thread
cifs_read_from_socket
cifs_readv_from_socket
cifs_reconnect
cifs_abort_connection
sock_release();
server->ssocket = NULL;
/* Missing put_net() here. */
generic_ip_connect
get_net()
->connect() /* Failed */
put_net()
sock_release();
server->ssocket = NULL;
free_rsp_buf
...
clean_demultiplex_info
/* It's only called once here. */
put_net()
```
When cifs_reconnect() is triggered, the server->ssocket is released
without a corresponding put_net() for the reference acquired in
generic_ip_connect() before. it ends up calling generic_ip_connect()
again to retry get_net(). After that, server->ssocket is set to NULL
in the error path of generic_ip_connect(), and the net count cannot be
released in the final clean_demultiplex_info() function.
2. Potential use-after-free
The current refcounting scheme can lead to a potential use-after-free issue
in the following scenario:
```
cifs_do_mount
cifs_mount
cifs_mount_get_session
cifs_get_tcp_session
get_net() /* First get net */
ip_connect
generic_ip_connect
get_net()
bind_socket
kernel_bind /* failed */
put_net()
/* after out_err_crypto_release label */
put_net()
/* after out_err label */
put_net()
```
In the exception handling process where binding the socket fails, the
get_net() and put_net() calls are unbalanced, which may cause the
server->net reference count to drop to zero and be prematurely released.
To address both issues, this patch ties the netns reference counting to
the server->ssocket and server lifecycles. The extra reference is now
acquired when the server or socket is created, and released when the
socket is destroyed or the server is torn down.
[1]: https://bugzilla.kernel.org/show_bug.cgi?id=219792
Fixes: ef7134c7fc ("smb: client: Fix use-after-free of network namespace.")
Fixes: e9f2517a3e ("smb: client: fix TCP timers deadlock after rmmod")
Signed-off-by: Wang Zhaolong <wangzhaolong1@huawei.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
In a multichannel setup, it was observed that a few fields were not being
copied over to the secondary channels, which impacted performance in cases
where these options were relevant but not properly synchronized. To address
this, this patch introduces copying the following parameters from the
primary channel to the secondary channels:
- min_offload
- compression.requested
- dfs_conn
- ignore_signature
- leaf_fullpath
- noblockcnt
- retrans
- sign
By copying these parameters, we ensure consistency across channels and
prevent performance degradation due to missing or outdated settings.
Cc: stable@vger.kernel.org
Signed-off-by: Aman <aman1@microsoft.com>
Reviewed-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Function ip_rfc1001_connect() send NetBIOS session request but currently
does not read response. It even does not wait for the response. Instead it
just calls usleep_range(1000, 2000) and explain in comment that some
servers require short break before sending SMB negotiate packet. Response
is later handled in generic is_smb_response() function called from
cifs_demultiplex_thread().
That comment probably refers to the old DOS SMB server which cannot process
incoming SMB negotiate packet if it has not sent NetBIOS session response
packet. Note that current sleep timeout is too small when trying to
establish connection to DOS SMB server running in qemu virtual machine
connected over qemu user networking with guestfwd netcat options. So that
usleep_range() call is not useful at all.
NetBIOS session response packet contains useful error information, like
the server name specified NetBIOS session request packet is incorrect.
Old Windows SMB servers and even the latest SMB server on the latest
Windows Server 2022 version requires that the name is the correct server
name, otherwise they return error RFC1002_NOT_PRESENT. This applies for all
SMB dialects (old SMB1, and also modern SMB2 and SMB3).
Therefore read the reply of NetBIOS session request and implement parsing
of the reply. Log received error to dmesg to help debugging reason why
connection was refused. Also convert NetBIOS error to useful errno.
Note that ip_rfc1001_connect() function is used only when doing connection
over port 139. So the common SMB scenario over port 445 is not affected by
this change at all.
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Function ip_rfc1001_connect() which establish NetBIOS session for SMB
connections, currently uses smb_send() function for sending NetBIOS Session
Request packet. This function expects that the passed buffer is SMB packet
and for SMB2+ connections it mangles packet header, which breaks prepared
NetBIOS Session Request packet. Result is that this function send garbage
packet for SMB2+ connection, which SMB2+ server cannot parse. That function
is not mangling packets for SMB1 connections, so it somehow works for SMB1.
Fix this problem and instead of smb_send(), use smb_send_kvec() function
which does not mangle prepared packet, this function send them as is. Just
API of this function takes struct msghdr (kvec) instead of packet buffer.
[MS-SMB2] specification allows SMB2 protocol to use NetBIOS as a transport
protocol. NetBIOS can be used over TCP via port 139. So this is a valid
configuration, just not so common. And even recent Windows versions (e.g.
Windows Server 2022) still supports this configuration: SMB over TCP port
139, including for modern SMB2 and SMB3 dialects.
This change fixes SMB2 and SMB3 connections over TCP port 139 which
requires establishing of NetBIOS session. Tested that this change fixes
establishing of SMB2 and SMB3 connections with Windows Server 2022.
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Native Windows sockets created by WinSock on Windows 10 April 2018 Update
(version 1803) or Windows Server 2019 (version 1809) or later versions is
reparse point with IO_REPARSE_TAG_AF_UNIX tag, with empty reparse point
data buffer and without any EAs.
Create AF_UNIX sockets in this native format if -o nonativesocket was not
specified.
This change makes AF_UNIX sockets created by Linux CIFS client compatible
with AF_UNIX sockets created by Windows applications on NTFS volumes.
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Currently Linux CIFS client creates a new symlink of the first flavor which
is allowed by mount options, parsed in this order: -o (no)mfsymlinks,
-o (no)sfu, -o (no)unix (+ its aliases) and -o reparse=[type].
Introduce a new mount option -o symlink= for explicitly choosing a symlink
flavor. Possible options are:
-o symlink=default - The default behavior, like before this change.
-o symlink=none - Disallow creating a new symlinks
-o symlink=native - Create as native SMB symlink reparse point
-o symlink=unix - Create via SMB1 unix extension command
-o symlink=mfsymlinks - Create as regular file of mfsymlinks format
-o symlink=sfu - Create as regular system file of SFU format
-o symlink=nfs - Create as NFS reparse point
-o symlink=wsl - Create as WSL reparse point
So for example specifying -o sfu,mfsymlinks,symlink=native will allow to
parse symlinks also of SFU and mfsymlinks types (which are disabled by
default unless mount option is explicitly specified), but new symlinks will
be created under native SMB type (which parsing is always enabled).
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
The matching of DFS connections is already handled by @dfs_conn, so
remove @leaf_fullpath matching altogether.
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
TCP_Server_Info::leaf_fullpath is allocated in cifs_get_tcp_session()
and never changed afterwards, so there is no need to serialize its
access.
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
If TCP Server is about to be destroyed (e.g. CifsExiting was set) and
it is reconnecting, stop retrying DFS targets from cached DFS referral
as this would potentially delay server shutdown in several seconds.
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Some places pass hostnames rather than UNC paths to resolve them to ip
addresses, so provide helpers to handle both cases and then stop
converting hostnames to UNC paths by inserting path delimiters into
them. Also kill @expiry parameter as it's not used anywhere.
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
If the user specified a DNS domain name in domain= mount option, then
use it instead of parsing it in NTLMSSP CHALLENGE_MESSAGE message.
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Old Windows servers will return not fully qualified DFS targets by
default as specified in
MS-DFSC 3.2.5.5 Receiving a Root Referral Request or Link Referral
Request
| Servers SHOULD<30> return fully qualified DNS host names of
| targets in responses to root referral requests and link referral
| requests.
| ...
| <30> Section 3.2.5.5: By default, Windows Server 2003, Windows
| Server 2008, Windows Server 2008 R2, Windows Server 2012, and
| Windows Server 2012 R2 return DNS host names that are not fully
| qualified for targets.
Fix this by converting all NetBIOS host names from DFS targets to
FQDNs and try resolving them first if DNS domain name was provided in
NTLMSSP CHALLENGE_MESSAGE message from previous SMB2_SESSION_SETUP.
This also prevents the client from translating the DFS target
hostnames to another domain depending on the network domain search
order.
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Commit ef7134c7fc ("smb: client: Fix use-after-free of network namespace.")
fixed a netns UAF by manually enabled socket refcounting
(sk->sk_net_refcnt=1 and sock_inuse_add(net, 1)).
The reason the patch worked for that bug was because we now hold
references to the netns (get_net_track() gets a ref internally)
and they're properly released (internally, on __sk_destruct()),
but only because sk->sk_net_refcnt was set.
Problem:
(this happens regardless of CONFIG_NET_NS_REFCNT_TRACKER and regardless
if init_net or other)
Setting sk->sk_net_refcnt=1 *manually* and *after* socket creation is not
only out of cifs scope, but also technically wrong -- it's set conditionally
based on user (=1) vs kernel (=0) sockets. And net/ implementations
seem to base their user vs kernel space operations on it.
e.g. upon TCP socket close, the TCP timers are not cleared because
sk->sk_net_refcnt=1:
(cf. commit 151c9c724d ("tcp: properly terminate timers for kernel sockets"))
net/ipv4/tcp.c:
void tcp_close(struct sock *sk, long timeout)
{
lock_sock(sk);
__tcp_close(sk, timeout);
release_sock(sk);
if (!sk->sk_net_refcnt)
inet_csk_clear_xmit_timers_sync(sk);
sock_put(sk);
}
Which will throw a lockdep warning and then, as expected, deadlock on
tcp_write_timer().
A way to reproduce this is by running the reproducer from ef7134c7fc
and then 'rmmod cifs'. A few seconds later, the deadlock/lockdep
warning shows up.
Fix:
We shouldn't mess with socket internals ourselves, so do not set
sk_net_refcnt manually.
Also change __sock_create() to sock_create_kern() for explicitness.
As for non-init_net network namespaces, we deal with it the best way
we can -- hold an extra netns reference for server->ssocket and drop it
when it's released. This ensures that the netns still exists whenever
we need to create/destroy server->ssocket, but is not directly tied to
it.
Fixes: ef7134c7fc ("smb: client: Fix use-after-free of network namespace.")
Cc: stable@vger.kernel.org
Signed-off-by: Enzo Matsumiya <ematsumiya@suse.de>
Signed-off-by: Steve French <stfrench@microsoft.com>
dfs_cache_refresh() delayed worker could race with cifs_put_tcon(), so
make sure to call list_replace_init() on @tcon->dfs_ses_list after
kworker is cancelled or finished.
Fixes: 4f42a8b54b ("smb: client: fix DFS interlink failover")
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Fixes the case for example where the password specified on mount is a
recently expired password, but password2 is valid. Without this patch
this mount scenario would fail.
This patch introduces the following changes to support password rotation on
mount:
1. If an existing session is not found and the new session setup results in
EACCES, EKEYEXPIRED or EKEYREVOKED, swap password and password2 (if
available), and retry the mount.
2. To match the new mount with an existing session, add conditions to check
if a) password and password2 of the new mount and the existing session are
the same, or b) password of the new mount is the same as the password2 of
the existing session, and password2 of the new mount is the same as the
password of the existing session.
3. If an existing session is found, but needs reconnect, retry the session
setup after swapping password and password2 (if available), in case the
previous attempt results in EACCES, EKEYEXPIRED or EKEYREVOKED.
Cc: stable@vger.kernel.org
Signed-off-by: Meetakshi Setiya <msetiya@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
We can access local_nls directly from @tcon->ses, so there is no need
to pass it as parameter in cifs_tree_connect().
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Setting dir_cache_timeout to zero should disable the caching of
directory contents. Currently, even when dir_cache_timeout is zero,
some caching related functions are still invoked, which is unintended
behavior.
Fix the issue by setting tcon->nohandlecache to true when
dir_cache_timeout is zero, ensuring that directory handle caching
is properly disabled.
Fixes: 238b351d09 ("smb3: allow controlling length of time directory entries are cached with dir leases")
Reviewed-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Reviewed-by: Enzo Matsumiya <ematsumiya@suse.de>
Signed-off-by: Henrique Carvalho <henrique.carvalho@suse.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
In the current implementation, the SMB filesystem on a mount point can
trigger upcalls from the kernel to the userspace to enable certain
functionalities like spnego, dns_resolution, amongst others. These upcalls
usually either happen in the context of the mount or in the context of an
application/user. The upcall handler for cifs, cifs.upcall already has
existing code which switches the namespaces to the caller's namespace
before handling the upcall. This behaviour is expected for scenarios like
multiuser mounts, but might not cover all single user scenario with
services such as Kubernetes, where the mount can happen from different
locations such as on the host, from an app container, or a driver pod
which does the mount on behalf of a different pod.
This patch introduces a new mount option called upcall_target, to
customise the upcall behaviour. upcall_target can take 'mount' and 'app'
as possible values. This aids use cases like Kubernetes where the mount
happens on behalf of the application in another container altogether.
Having this new mount option allows the mount command to specify where the
upcall should happen: 'mount' for resolving the upcall to the host
namespace, and 'app' for resolving the upcall to the ns of the calling
thread. This will enable both the scenarios where the Kerberos credentials
can be found on the application namespace or the host namespace to which
just the mount operation is "delegated".
Reviewed-by: Shyam Prasad <shyam.prasad@microsoft.com>
Reviewed-by: Bharath S M <bharathsm@microsoft.com>
Reviewed-by: Ronnie Sahlberg <ronniesahlberg@gmail.com>
Signed-off-by: Ritvik Budhiraja <rbudhiraja@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Recently, we got a customer report that CIFS triggers oops while
reconnecting to a server. [0]
The workload runs on Kubernetes, and some pods mount CIFS servers
in non-root network namespaces. The problem rarely happened, but
it was always while the pod was dying.
The root cause is wrong reference counting for network namespace.
CIFS uses kernel sockets, which do not hold refcnt of the netns that
the socket belongs to. That means CIFS must ensure the socket is
always freed before its netns; otherwise, use-after-free happens.
The repro steps are roughly:
1. mount CIFS in a non-root netns
2. drop packets from the netns
3. destroy the netns
4. unmount CIFS
We can reproduce the issue quickly with the script [1] below and see
the splat [2] if CONFIG_NET_NS_REFCNT_TRACKER is enabled.
When the socket is TCP, it is hard to guarantee the netns lifetime
without holding refcnt due to async timers.
Let's hold netns refcnt for each socket as done for SMC in commit
9744d2bf19 ("smc: Fix use-after-free in tcp_write_timer_handler().").
Note that we need to move put_net() from cifs_put_tcp_session() to
clean_demultiplex_info(); otherwise, __sock_create() still could touch a
freed netns while cifsd tries to reconnect from cifs_demultiplex_thread().
Also, maybe_get_net() cannot be put just before __sock_create() because
the code is not under RCU and there is a small chance that the same
address happened to be reallocated to another netns.
[0]:
CIFS: VFS: \\XXXXXXXXXXX has not responded in 15 seconds. Reconnecting...
CIFS: Serverclose failed 4 times, giving up
Unable to handle kernel paging request at virtual address 14de99e461f84a07
Mem abort info:
ESR = 0x0000000096000004
EC = 0x25: DABT (current EL), IL = 32 bits
SET = 0, FnV = 0
EA = 0, S1PTW = 0
FSC = 0x04: level 0 translation fault
Data abort info:
ISV = 0, ISS = 0x00000004
CM = 0, WnR = 0
[14de99e461f84a07] address between user and kernel address ranges
Internal error: Oops: 0000000096000004 [#1] SMP
Modules linked in: cls_bpf sch_ingress nls_utf8 cifs cifs_arc4 cifs_md4 dns_resolver tcp_diag inet_diag veth xt_state xt_connmark nf_conntrack_netlink xt_nat xt_statistic xt_MASQUERADE xt_mark xt_addrtype ipt_REJECT nf_reject_ipv4 nft_chain_nat nf_nat xt_conntrack nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 xt_comment nft_compat nf_tables nfnetlink overlay nls_ascii nls_cp437 sunrpc vfat fat aes_ce_blk aes_ce_cipher ghash_ce sm4_ce_cipher sm4 sm3_ce sm3 sha3_ce sha512_ce sha512_arm64 sha1_ce ena button sch_fq_codel loop fuse configfs dmi_sysfs sha2_ce sha256_arm64 dm_mirror dm_region_hash dm_log dm_mod dax efivarfs
CPU: 5 PID: 2690970 Comm: cifsd Not tainted 6.1.103-109.184.amzn2023.aarch64 #1
Hardware name: Amazon EC2 r7g.4xlarge/, BIOS 1.0 11/1/2018
pstate: 00400005 (nzcv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
pc : fib_rules_lookup+0x44/0x238
lr : __fib_lookup+0x64/0xbc
sp : ffff8000265db790
x29: ffff8000265db790 x28: 0000000000000000 x27: 000000000000bd01
x26: 0000000000000000 x25: ffff000b4baf8000 x24: ffff00047b5e4580
x23: ffff8000265db7e0 x22: 0000000000000000 x21: ffff00047b5e4500
x20: ffff0010e3f694f8 x19: 14de99e461f849f7 x18: 0000000000000000
x17: 0000000000000000 x16: 0000000000000000 x15: 0000000000000000
x14: 0000000000000000 x13: 0000000000000000 x12: 3f92800abd010002
x11: 0000000000000001 x10: ffff0010e3f69420 x9 : ffff800008a6f294
x8 : 0000000000000000 x7 : 0000000000000006 x6 : 0000000000000000
x5 : 0000000000000001 x4 : ffff001924354280 x3 : ffff8000265db7e0
x2 : 0000000000000000 x1 : ffff0010e3f694f8 x0 : ffff00047b5e4500
Call trace:
fib_rules_lookup+0x44/0x238
__fib_lookup+0x64/0xbc
ip_route_output_key_hash_rcu+0x2c4/0x398
ip_route_output_key_hash+0x60/0x8c
tcp_v4_connect+0x290/0x488
__inet_stream_connect+0x108/0x3d0
inet_stream_connect+0x50/0x78
kernel_connect+0x6c/0xac
generic_ip_connect+0x10c/0x6c8 [cifs]
__reconnect_target_unlocked+0xa0/0x214 [cifs]
reconnect_dfs_server+0x144/0x460 [cifs]
cifs_reconnect+0x88/0x148 [cifs]
cifs_readv_from_socket+0x230/0x430 [cifs]
cifs_read_from_socket+0x74/0xa8 [cifs]
cifs_demultiplex_thread+0xf8/0x704 [cifs]
kthread+0xd0/0xd4
Code: aa0003f8 f8480f13 eb18027f 540006c0 (b9401264)
[1]:
CIFS_CRED="/root/cred.cifs"
CIFS_USER="Administrator"
CIFS_PASS="Password"
CIFS_IP="X.X.X.X"
CIFS_PATH="//${CIFS_IP}/Users/Administrator/Desktop/CIFS_TEST"
CIFS_MNT="/mnt/smb"
DEV="enp0s3"
cat <<EOF > ${CIFS_CRED}
username=${CIFS_USER}
password=${CIFS_PASS}
domain=EXAMPLE.COM
EOF
unshare -n bash -c "
mkdir -p ${CIFS_MNT}
ip netns attach root 1
ip link add eth0 type veth peer veth0 netns root
ip link set eth0 up
ip -n root link set veth0 up
ip addr add 192.168.0.2/24 dev eth0
ip -n root addr add 192.168.0.1/24 dev veth0
ip route add default via 192.168.0.1 dev eth0
ip netns exec root sysctl net.ipv4.ip_forward=1
ip netns exec root iptables -t nat -A POSTROUTING -s 192.168.0.2 -o ${DEV} -j MASQUERADE
mount -t cifs ${CIFS_PATH} ${CIFS_MNT} -o vers=3.0,sec=ntlmssp,credentials=${CIFS_CRED},rsize=65536,wsize=65536,cache=none,echo_interval=1
touch ${CIFS_MNT}/a.txt
ip netns exec root iptables -t nat -D POSTROUTING -s 192.168.0.2 -o ${DEV} -j MASQUERADE
"
umount ${CIFS_MNT}
[2]:
ref_tracker: net notrefcnt@000000004bbc008d has 1/1 users at
sk_alloc (./include/net/net_namespace.h:339 net/core/sock.c:2227)
inet_create (net/ipv4/af_inet.c:326 net/ipv4/af_inet.c:252)
__sock_create (net/socket.c:1576)
generic_ip_connect (fs/smb/client/connect.c:3075)
cifs_get_tcp_session.part.0 (fs/smb/client/connect.c:3160 fs/smb/client/connect.c:1798)
cifs_mount_get_session (fs/smb/client/trace.h:959 fs/smb/client/connect.c:3366)
dfs_mount_share (fs/smb/client/dfs.c:63 fs/smb/client/dfs.c:285)
cifs_mount (fs/smb/client/connect.c:3622)
cifs_smb3_do_mount (fs/smb/client/cifsfs.c:949)
smb3_get_tree (fs/smb/client/fs_context.c:784 fs/smb/client/fs_context.c:802 fs/smb/client/fs_context.c:794)
vfs_get_tree (fs/super.c:1800)
path_mount (fs/namespace.c:3508 fs/namespace.c:3834)
__x64_sys_mount (fs/namespace.c:3848 fs/namespace.c:4057 fs/namespace.c:4034 fs/namespace.c:4034)
do_syscall_64 (arch/x86/entry/common.c:52 arch/x86/entry/common.c:83)
entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:130)
Fixes: 26abe14379 ("net: Modify sk_alloc to not reference count the netns of kernel sockets.")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Acked-by: Tom Talpey <tom@talpey.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
cifs_ses_find_chan() has been unused since commit
f486ef8e20 ("cifs: use the chans_need_reconnect bitmap for reconnect status")
cifs_read_page_from_socket() has been unused since commit
d08089f649 ("cifs: Change the I/O paths to use an iterator rather than a page list")
cifs_chan_in_reconnect() has been unused since commit
bc962159e8 ("cifs: avoid race conditions with parallel reconnects")
Remove them.
Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Do not flood dmesg with failed session logoffs as kerberos tickets
getting expired or passwords being rotated is a very common scenario.
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
cifs_read_iter_from_socket() copies the iterator that's passed in for the
socket to modify as and if it will, and then advances the original iterator
by the amount sent. However, both callers revert the advancement (although
receive_encrypted_read() zeros beyond the iterator first). The problem is,
though, that cifs_readv_receive() reverts by the original length, not the
amount transmitted which can cause an oops in iov_iter_revert().
Fix this by:
(1) Remove the iov_iter_advance() from cifs_read_iter_from_socket().
(2) Remove the iov_iter_revert() from both callers. This fixes the bug in
cifs_readv_receive().
(3) In receive_encrypted_read(), if we didn't get back as much data as the
buffer will hold, copy the iterator, advance the copy and use the copy
to drive iov_iter_zero().
As a bonus, this gets rid of some unnecessary work.
This was triggered by generic/074 with the "-o sign" mount option.
Fixes: 3ee1a1fc39 ("cifs: Cut over to using netfslib")
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Steve French <sfrench@samba.org>
cc: Paulo Alcantara <pc@manguebit.com>
cc: Shyam Prasad N <nspmangalore@gmail.com>
cc: Rohith Surabattula <rohiths.msft@gmail.com>
cc: Jeff Layton <jlayton@kernel.org>
cc: linux-cifs@vger.kernel.org
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
Propagate error from cifs_construct_tcon() in cifs_sb_tlink() instead of
always returning -EACCES.
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
The DFS interlinks point to different DFS namespaces so make sure to
use the correct DFS root server to chase any DFS links under it by
storing the SMB session in dfs_ref_walk structure and then using it on
every referral walk.
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
list_head can be initialized automatically with LIST_HEAD()
instead of calling INIT_LIST_HEAD(). No functional impact.
Signed-off-by: Hongbo Li <lihongbo22@huawei.com>
Reviewed-by: Dan Carpenter <dan.carpenter@linaro.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Use ERR_CAST() as it is designed for casting an error pointer to
another type.
This macro uses the __force and __must_check modifiers, which are used
to tell the compiler to check for errors where this macro is used.
Signed-off-by: Yuesong Li <liyuesong@vivo.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Call cifs_reconnect() to wake up processes waiting on negotiate
protocol to handle the case where server abruptly shut down and had no
chance to properly close the socket.
Simple reproducer:
ssh 192.168.2.100 pkill -STOP smbd
mount.cifs //192.168.2.100/test /mnt -o ... [never returns]
Cc: Rickard Andersson <rickaran@axis.com>
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
The shutdown ioctl can leak a refcount on the tlink which can
prevent rmmod (unloading the cifs.ko) module from working.
Found while debugging xfstest generic/043
Fixes: 69ca1f5755 ("smb3: add dynamic tracepoints for shutdown ioctl")
Reviewed-by: Meetakshi Setiya <msetiya@microsoft.com>
Reviewed-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Although by default we negotiate CIFS Unix Extensions for SMB1 mounts to
Samba (and they work if the user does not specify "unix" or "posix" or
"linux" on mount), and we do properly handle when a user turns them off
with "nounix" mount parm. But with the changes to the mount API we
broke cases where the user explicitly specifies the "unix" option (or
equivalently "linux" or "posix") on mount with vers=1.0 to Samba or other
servers which support the CIFS Unix Extensions.
"mount error(95): Operation not supported"
and logged:
"CIFS: VFS: Check vers= mount option. SMB3.11 disabled but required for POSIX extensions"
even though CIFS Unix Extensions are supported for vers=1.0 This patch fixes
the case where the user specifies both "unix" (or equivalently "posix" or
"linux") and "vers=1.0" on mount to a server which supports the
CIFS Unix Extensions.
Cc: stable@vger.kernel.org
Reviewed-by: David Howells <dhowell@redhat.com>
Reviewed-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
When mounting with the SMB1 Unix Extensions (e.g. mounts
to Samba with vers=1.0), reconnects no longer reset the
Unix Extensions (SetFSInfo SET_FILE_UNIX_BASIC) after tcon so most
operations (e.g. stat, ls, open, statfs) will fail continuously
with:
"Operation not supported"
if the connection ever resets (e.g. due to brief network disconnect)
Cc: stable@vger.kernel.org
Reviewed-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Add tracing for the refcounting/lifecycle of the cifs_tcon struct, marking
different events with different labels and giving each tcon its own debug
ID so that the tracelines corresponding to individual tcons can be
distinguished. This can be enabled with:
echo 1 >/sys/kernel/debug/tracing/events/cifs/smb3_tcon_ref/enable
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
cc: Shyam Prasad N <nspmangalore@gmail.com>
cc: linux-cifs@vger.kernel.org
cc: linux-fsdevel@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
There are various use cases that are becoming more common in which password
changes are scheduled on a server(s) periodically but the clients connected
to this server need to stay connected (even in the face of brief network
reconnects) due to mounts which can not be easily unmounted and mounted at
will, and servers that do password rotation do not always have the ability
to tell the clients exactly when to the new password will be effective,
so add support for an alt password ("password2=") on mount (and also
remount) so that we can anticipate the upcoming change to the server
without risking breaking existing mounts.
An alternative would have been to use the kernel keyring for this but the
processes doing the reconnect do not have access to the keyring but do
have access to the ses structure.
Reviewed-by: Bharath SM <bharathsm@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Skip sessions that are being teared down (status == SES_EXITING) to
avoid UAF.
Cc: stable@vger.kernel.org
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Serialise cifs_construct_tcon() with cifs_mount_mutex to handle
parallel mounts that may end up reusing the session and tcon created
by it.
Cc: stable@vger.kernel.org # 6.4+
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
The tcons created by cifs_construct_tcon() on multiuser mounts must
also be able to failover and refresh DFS referrals, so set the
appropriate fields in order to get a full DFS tcon. They could be
shared among different superblocks later, too.
Cc: stable@vger.kernel.org # 6.4+
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202404021518.3Xu2VU4s-lkp@intel.com/
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Avoid potential use-after-free bugs when walking DFS referrals,
mounting and performing DFS failover by ensuring that all children
from parent @tcon->ses are also refcounted. They're all needed across
the entire DFS mount. Get rid of @tcon->dfs_ses_list while we're at
it, too.
Cc: stable@vger.kernel.org # 6.4+
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202404021527.ZlRkIxgv-lkp@intel.com/
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
The UAF bug is due to smb2_reconnect_server() accessing a session that
is already being teared down by another thread that is executing
__cifs_put_smb_ses(). This can happen when (a) the client has
connection to the server but no session or (b) another thread ends up
setting @ses->ses_status again to something different than
SES_EXITING.
To fix this, we need to make sure to unconditionally set
@ses->ses_status to SES_EXITING and prevent any other threads from
setting a new status while we're still tearing it down.
The following can be reproduced by adding some delay to right after
the ipc is freed in __cifs_put_smb_ses() - which will give
smb2_reconnect_server() worker a chance to run and then accessing
@ses->ipc:
kinit ...
mount.cifs //srv/share /mnt/1 -o sec=krb5,nohandlecache,echo_interval=10
[disconnect srv]
ls /mnt/1 &>/dev/null
sleep 30
kdestroy
[reconnect srv]
sleep 10
umount /mnt/1
...
CIFS: VFS: Verify user has a krb5 ticket and keyutils is installed
CIFS: VFS: \\srv Send error in SessSetup = -126
CIFS: VFS: Verify user has a krb5 ticket and keyutils is installed
CIFS: VFS: \\srv Send error in SessSetup = -126
general protection fault, probably for non-canonical address
0x6b6b6b6b6b6b6b6b: 0000 [#1] PREEMPT SMP NOPTI
CPU: 3 PID: 50 Comm: kworker/3:1 Not tainted 6.9.0-rc2 #1
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-1.fc39
04/01/2014
Workqueue: cifsiod smb2_reconnect_server [cifs]
RIP: 0010:__list_del_entry_valid_or_report+0x33/0xf0
Code: 4f 08 48 85 d2 74 42 48 85 c9 74 59 48 b8 00 01 00 00 00 00 ad
de 48 39 c2 74 61 48 b8 22 01 00 00 00 00 74 69 <48> 8b 01 48 39 f8 75
7b 48 8b 72 08 48 39 c6 0f 85 88 00 00 00 b8
RSP: 0018:ffffc900001bfd70 EFLAGS: 00010a83
RAX: dead000000000122 RBX: ffff88810da53838 RCX: 6b6b6b6b6b6b6b6b
RDX: 6b6b6b6b6b6b6b6b RSI: ffffffffc02f6878 RDI: ffff88810da53800
RBP: ffff88810da53800 R08: 0000000000000001 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000001 R12: ffff88810c064000
R13: 0000000000000001 R14: ffff88810c064000 R15: ffff8881039cc000
FS: 0000000000000000(0000) GS:ffff888157c00000(0000)
knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007fe3728b1000 CR3: 000000010caa4000 CR4: 0000000000750ef0
PKRU: 55555554
Call Trace:
<TASK>
? die_addr+0x36/0x90
? exc_general_protection+0x1c1/0x3f0
? asm_exc_general_protection+0x26/0x30
? __list_del_entry_valid_or_report+0x33/0xf0
__cifs_put_smb_ses+0x1ae/0x500 [cifs]
smb2_reconnect_server+0x4ed/0x710 [cifs]
process_one_work+0x205/0x6b0
worker_thread+0x191/0x360
? __pfx_worker_thread+0x10/0x10
kthread+0xe2/0x110
? __pfx_kthread+0x10/0x10
ret_from_fork+0x34/0x50
? __pfx_kthread+0x10/0x10
ret_from_fork_asm+0x1a/0x30
</TASK>
Cc: stable@vger.kernel.org
Signed-off-by: Paulo Alcantara (Red Hat) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Move the following:
extern mempool_t *cifs_sm_req_poolp;
extern mempool_t *cifs_req_poolp;
extern mempool_t *cifs_mid_poolp;
extern bool disable_legacy_dialects;
from various .c files to cifsglob.h.
Signed-off-by: David Howells <dhowells@redhat.com>
cc: linux-cifs@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
Paulo Alcantara
Kuniyuki Iwashima
Steve French
Pali Rohár
Wang Zhaolong
Roman Smirnov
Aman
Henrique Carvalho
Enzo Matsumiya
Meetakshi Setiya
Ritvik Budhiraja
Dr. David Alan Gilbert
David Howells
Hongbo Li
Yuesong Li