| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Pull scheduler updates from Ingo Molnar:
"Debuggability:
- Change most occurances of BUG_ON() to WARN_ON_ONCE()
- Reorganize & fix TASK_ state comparisons, turn it into a bitmap
- Update/fix misc scheduler debugging facilities
Load-balancing & regular scheduling:
- Improve the behavior of the scheduler in presence of lot of
SCHED_IDLE tasks - in particular they should not impact other
scheduling classes.
- Optimize task load tracking, cleanups & fixes
- Clean up & simplify misc load-balancing code
Freezer:
- Rewrite the core freezer to behave better wrt thawing and be
simpler in general, by replacing PF_FROZEN with TASK_FROZEN &
fixing/adjusting all the fallout.
Deadline scheduler:
- Fix the DL capacity-aware code
- Factor out dl_task_is_earliest_deadline() &
replenish_dl_new_period()
- Relax/optimize locking in task_non_contending()
Cleanups:
- Factor out the update_current_exec_runtime() helper
- Various cleanups, simplifications"
* tag 'sched-core-2022-10-07' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (41 commits)
sched: Fix more TASK_state comparisons
sched: Fix TASK_state comparisons
sched/fair: Move call to list_last_entry() in detach_tasks
sched/fair: Cleanup loop_max and loop_break
sched/fair: Make sure to try to detach at least one movable task
sched: Show PF_flag holes
freezer,sched: Rewrite core freezer logic
sched: Widen TAKS_state literals
sched/wait: Add wait_event_state()
sched/completion: Add wait_for_completion_state()
sched: Add TASK_ANY for wait_task_inactive()
sched: Change wait_task_inactive()s match_state
freezer,umh: Clean up freezer/initrd interaction
freezer: Have {,un}lock_system_sleep() save/restore flags
sched: Rename task_running() to task_on_cpu()
sched/fair: Cleanup for SIS_PROP
sched/fair: Default to false in test_idle_cores()
sched/fair: Remove useless check in select_idle_core()
sched/fair: Avoid double search on same cpu
sched/fair: Remove redundant check in select_idle_smt()
...
|
|
So far we were just lucky because the uninitialized members
of struct msghdr are not used by default on a SOCK_STREAM tcp
socket.
But as new things like msg_ubuf and sg_from_iter where added
recently, we should play on the safe side and avoid potention
problems in future.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Cc: stable@vger.kernel.org
Reviewed-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
This is ignored anyway by the tcp layer.
Signed-off-by: Stefan Metzmacher <metze@samba.org>
Cc: stable@vger.kernel.org
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Rewrite the core freezer to behave better wrt thawing and be simpler
in general.
By replacing PF_FROZEN with TASK_FROZEN, a special block state, it is
ensured frozen tasks stay frozen until thawed and don't randomly wake
up early, as is currently possible.
As such, it does away with PF_FROZEN and PF_FREEZER_SKIP, freeing up
two PF_flags (yay!).
Specifically; the current scheme works a little like:
freezer_do_not_count();
schedule();
freezer_count();
And either the task is blocked, or it lands in try_to_freezer()
through freezer_count(). Now, when it is blocked, the freezer
considers it frozen and continues.
However, on thawing, once pm_freezing is cleared, freezer_count()
stops working, and any random/spurious wakeup will let a task run
before its time.
That is, thawing tries to thaw things in explicit order; kernel
threads and workqueues before doing bringing SMP back before userspace
etc.. However due to the above mentioned races it is entirely possible
for userspace tasks to thaw (by accident) before SMP is back.
This can be a fatal problem in asymmetric ISA architectures (eg ARMv9)
where the userspace task requires a special CPU to run.
As said; replace this with a special task state TASK_FROZEN and add
the following state transitions:
TASK_FREEZABLE -> TASK_FROZEN
__TASK_STOPPED -> TASK_FROZEN
__TASK_TRACED -> TASK_FROZEN
The new TASK_FREEZABLE can be set on any state part of TASK_NORMAL
(IOW. TASK_INTERRUPTIBLE and TASK_UNINTERRUPTIBLE) -- any such state
is already required to deal with spurious wakeups and the freezer
causes one such when thawing the task (since the original state is
lost).
The special __TASK_{STOPPED,TRACED} states *can* be restored since
their canonical state is in ->jobctl.
With this, frozen tasks need an explicit TASK_FROZEN wakeup and are
free of undue (early / spurious) wakeups.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://lore.kernel.org/r/20220822114649.055452969@infradead.org
|
|
SMB1 server's header_preamble_size is not 0, add use is_smb1 function
to simplify the code, no actual functional changes.
Reviewed-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Signed-off-by: Zhang Xiaoxu <zhangxiaoxu5@huawei.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
It's better to use HEADER_PREAMBLE_SIZE because the unfolded expression
too long. No actual functional changes, minor readability improvement.
Signed-off-by: Zhang Xiaoxu <zhangxiaoxu5@huawei.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
It is only used in transport.c.
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
DeleteMidQEntry() was just a proxy for cifs_mid_q_entry_release().
- remove DeleteMidQEntry()
- rename cifs_mid_q_entry_release() to release_mid()
- rename kref_put() callback _cifs_mid_q_entry_release to __release_mid
- rename AllocMidQEntry() to alloc_mid()
- rename cifs_delete_mid() to delete_mid()
Update callers to use new names.
Signed-off-by: Enzo Matsumiya <ematsumiya@suse.de>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Currently much of the smb1 code is built even when
CONFIG_CIFS_ALLOW_INSECURE_LEGACY is disabled.
Move cifssmb.c to only be compiled when insecure legacy is disabled,
and move various SMB1/CIFS helper functions to that ifdef. Some
functions that were not SMB1/CIFS specific needed to be moved out of
cifssmb.c
This shrinks cifs.ko by more than 10% which is good - but also will
help with the eventual movement of the legacy code to a distinct
module. Follow on patches can shrink the number of ifdefs by
code restructuring where smb1 code is wedged in functions that
should be calling dialect specific helper functions instead,
and also by moving some functions from file.c/dir.c/inode.c into
smb1 specific c files.
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
During analysis of multichannel perf, it was seen that
the global locks cifs_tcp_ses_lock and GlobalMid_Lock, which
were shared between various data structures were causing a
lot of contention points.
With this change, we're breaking down the use of these locks
by introducing new locks at more granular levels. i.e.
server->srv_lock, ses->ses_lock and tcon->tc_lock to protect
the unprotected fields of server, session and tcon structs;
and server->mid_lock to protect mid related lists and entries
at server level.
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Remove warnings for five global variables. For example:
fs/cifs/cifsglob.h:1984:24: warning: symbol 'midCount' was not declared. Should it be static?
Also change them from camelCase (e.g. "midCount" to "mid_count")
Reviewed-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
The srv_mutex is used during writeback so cifs should ensure that
allocations done when that mutex is held are done with GFP_NOFS, to
avoid having direct reclaim ending up waiting for the same mutex and
causing a deadlock. This is detected by lockdep with the splat below:
======================================================
WARNING: possible circular locking dependency detected
5.18.0 #70 Not tainted
------------------------------------------------------
kswapd0/49 is trying to acquire lock:
ffff8880195782e0 (&tcp_ses->srv_mutex){+.+.}-{3:3}, at: compound_send_recv
but task is already holding lock:
ffffffffa98e66c0 (fs_reclaim){+.+.}-{0:0}, at: balance_pgdat
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (fs_reclaim){+.+.}-{0:0}:
fs_reclaim_acquire
kmem_cache_alloc_trace
__request_module
crypto_alg_mod_lookup
crypto_alloc_tfm_node
crypto_alloc_shash
cifs_alloc_hash
smb311_crypto_shash_allocate
smb311_update_preauth_hash
compound_send_recv
cifs_send_recv
SMB2_negotiate
smb2_negotiate
cifs_negotiate_protocol
cifs_get_smb_ses
cifs_mount
cifs_smb3_do_mount
smb3_get_tree
vfs_get_tree
path_mount
__x64_sys_mount
do_syscall_64
entry_SYSCALL_64_after_hwframe
-> #0 (&tcp_ses->srv_mutex){+.+.}-{3:3}:
__lock_acquire
lock_acquire
__mutex_lock
mutex_lock_nested
compound_send_recv
cifs_send_recv
SMB2_write
smb2_sync_write
cifs_write
cifs_writepage_locked
cifs_writepage
shrink_page_list
shrink_lruvec
shrink_node
balance_pgdat
kswapd
kthread
ret_from_fork
other info that might help us debug this:
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(fs_reclaim);
lock(&tcp_ses->srv_mutex);
lock(fs_reclaim);
lock(&tcp_ses->srv_mutex);
*** DEADLOCK ***
1 lock held by kswapd0/49:
#0: ffffffffa98e66c0 (fs_reclaim){+.+.}-{0:0}, at: balance_pgdat
stack backtrace:
CPU: 2 PID: 49 Comm: kswapd0 Not tainted 5.18.0 #70
Call Trace:
<TASK>
dump_stack_lvl
dump_stack
print_circular_bug.cold
check_noncircular
__lock_acquire
lock_acquire
__mutex_lock
mutex_lock_nested
compound_send_recv
cifs_send_recv
SMB2_write
smb2_sync_write
cifs_write
cifs_writepage_locked
cifs_writepage
shrink_page_list
shrink_lruvec
shrink_node
balance_pgdat
kswapd
kthread
ret_from_fork
</TASK>
Fix this by using the memalloc_nofs_save/restore APIs around the places
where the srv_mutex is held. Do this in a wrapper function for the
lock/unlock of the srv_mutex, and rename the srv_mutex to avoid missing
call sites in the conversion.
Note that there is another lockdep warning involving internal crypto
locks, which was masked by this problem and is visible after this fix,
see the discussion in this thread:
https://lore.kernel.org/all/20220523123755.GA13668@axis.com/
Link: https://lore.kernel.org/r/CANT5p=rqcYfYMVHirqvdnnca4Mo+JQSw5Qu12v=kPfpk5yhhmg@mail.gmail.com/
Reported-by: Shyam Prasad N <nspmangalore@gmail.com>
Suggested-by: Lars Persson <larper@axis.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Reviewed-by: Enzo Matsumiya <ematsumiya@suse.de>
Signed-off-by: Vincent Whitchurch <vincent.whitchurch@axis.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
ses->status today shares statusEnum with server->tcpStatus.
This has been confusing, and tcon->status has deviated to use
a new enum. Follow suit and use new enum for ses_status as well.
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Use kzalloc rather than duplicating its implementation, which
makes code simple and easy to understand.
Signed-off-by: Haowen Bai <baihaowen@meizu.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Split the smb3_add_credits tracepoint to make it more obvious when looking
at the logs which line corresponds to what credit change. Also add a
tracepoint for credit overflow when it's being added back.
Note that it might be better to add another field to the tracepoint for
the information rather than splitting it. It would also be useful to store
the MID potentially, though that isn't available when the credits are first
obtained.
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Shyam Prasad N <nspmangalore@gmail.com>
cc: Rohith Surabattula <rohiths.msft@gmail.com>
cc: linux-cifs@vger.kernel.org
Acked-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Reviewed-by: Enzo Matsumiya <ematsumiya@suse.de>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
The cifs_demultiplexer_thread should only call cifs_reconnect.
If any other thread wants to trigger a reconnect, they can do
so by updating the server tcpStatus to CifsNeedReconnect.
The last patch attempted to use the same helper function for
both types of threads, but that causes other issues
with lock dependencies.
This patch creates a new helper for non-cifsd threads, that
will indicate to cifsd that the server needs reconnect.
Fixes: 2a05137a0575 ("cifs: mark sessions for reconnection in helper function")
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
cifs_mark_tcp_ses_conns_for_reconnect helper function is now
meant to be used by any of the threads to mark a channel
(or all the channels) for reconnect.
Replace all such manual changes to tcpStatus to use this
helper function, which takes care that the right channels,
smb sessions and tcons are marked for reconnect.
Also includes one line minor change
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
A spin lock called chan_lock was introduced recently.
But not all accesses were protected. Doing that with
this change.
To make sure that a channel is not freed when in use,
we need to introduce a ref count. But today, we don't
ever free channels.
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Recent changes to multichannel required some adjustments in
the way connection states transitioned during/after reconnect.
Also some minor fixes:
1. A pending switch of GlobalMid_Lock to cifs_tcp_ses_lock
2. Relocations of the code that logs reconnect
3. Changed some code in allocate_mid to suit the new scheme
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
When a TCP connection gets reestablished by the sender in cifs_reconnect,
There is a chance for race condition with demultiplex thread waiting in
cifs_readv_from_socket on the old socket. It will now return -ECONNRESET.
This condition is handled by comparing socket pointer before and after
sock_recvmsg. If the socket pointer has changed, we should not call
cifs_reconnect again, but instead retry with new socket.
Also fixed another bug in my prev mchan commits.
We should always reestablish session (even if binding) on a channel
that needs reconnection.
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
While checking/updating status for tcp ses, smb ses or tcon,
we take GlobalMid_Lock. This doesn't make any sense.
Replaced it with cifs_tcp_ses_lock.
Ideally, we should take a spin lock per struct.
But since tcp ses, smb ses and tcon objects won't add up to a lot,
I think there should not be too much contention.
Also, in few other places, these are checked without locking.
Added locking for these.
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
We use the concept of "binding" when one of the secondary channel
is in the process of connecting/reconnecting to the server. Till this
binding process completes, and the channel is bound to an existing session,
we redirect traffic from other established channels on the binding channel,
effectively blocking all traffic till individual channels get reconnected.
With my last set of commits, we can get rid of this binding serialization.
We now have a bitmap of connection states for each channel. We will use
this bitmap instead for tracking channel status.
Having a bitmap also now enables us to keep the session alive, as long
as even a single channel underneath is alive.
Unfortunately, this also meant that we need to supply the tcp connection
info for the channel during all negotiate and session setup functions.
These changes have resulted in a slightly bigger code churn.
However, I expect perf and robustness improvements in the mchan scenario
after this change.
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Introducing a new spin lock to protect all the channel related
fields in a cifs_ses struct. This lock should be taken
whenever dealing with the channel fields, and should be held
only for very short intervals which will not sleep.
Currently, all channel related fields in cifs_ses structure
are protected by session_mutex. However, this mutex is held for
long periods (sometimes while waiting for a reply from server).
This makes the codepath quite tricky to change.
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
checkpatch complains about source files with filenames (e.g. in
these cases just below the SPDX header in comments at the top of
various files in fs/cifs). It also is helpful to change this now
so will be less confusing when the parent directory is renamed
e.g. from fs/cifs to fs/smb_client (or fs/smbfs)
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
There were three places where we were not taking the spinlock
around updates to server->tcpStatus when it was being modified.
To be consistent (also removes Coverity warning) and to remove
possibility of race best to lock all places where it is updated.
Two of the three were in initialization of the field and can't
race - but added lock around the other.
Addresses-Coverity: 1399512 ("Data race condition")
Reviewed-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Add SPDX license identifier and replace license boilerplate.
Corrects various checkpatch errors with the older format for
noting the LGPL license.
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
smb311_update_preauth_hash() uses the shash in server->secmech without
appropriate locking, and this can lead to sessions corrupting each
other's preauth hashes.
The following script can easily trigger the problem:
#!/bin/sh -e
NMOUNTS=10
for i in $(seq $NMOUNTS);
mkdir -p /tmp/mnt$i
umount /tmp/mnt$i 2>/dev/null || :
done
while :; do
for i in $(seq $NMOUNTS); do
mount -t cifs //192.168.0.1/test /tmp/mnt$i -o ... &
done
wait
for i in $(seq $NMOUNTS); do
umount /tmp/mnt$i
done
done
Usually within seconds this leads to one or more of the mounts failing
with the following errors, and a "Bad SMB2 signature for message" is
seen in the server logs:
CIFS: VFS: \\192.168.0.1 failed to connect to IPC (rc=-13)
CIFS: VFS: cifs_mount failed w/return code = -13
Fix it by holding the server mutex just like in the other places where
the shashes are used.
Fixes: 8bd68c6e47abff34e4 ("CIFS: implement v3.11 preauth integrity")
Signed-off-by: Vincent Whitchurch <vincent.whitchurch@axis.com>
CC: <stable@vger.kernel.org>
Reviewed-by: Aurelien Aptel <aaptel@suse.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
In case of interrupted syscalls, prevent sending CLOSE commands for
compound CREATE+CLOSE requests by introducing an
CIFS_CP_CREATE_CLOSE_OP flag to indicate lower layers that it should
not send a CLOSE command to the MIDs corresponding the compound
CREATE+CLOSE request.
A simple reproducer:
#!/bin/bash
mount //server/share /mnt -o username=foo,password=***
tc qdisc add dev eth0 root netem delay 450ms
stat -f /mnt &>/dev/null & pid=$!
sleep 0.01
kill $pid
tc qdisc del dev eth0 root
umount /mnt
Before patch:
...
6 0.256893470 192.168.122.2 → 192.168.122.15 SMB2 402 Create Request File: ;GetInfo Request FS_INFO/FileFsFullSizeInformation;Close Request
7 0.257144491 192.168.122.15 → 192.168.122.2 SMB2 498 Create Response File: ;GetInfo Response;Close Response
9 0.260798209 192.168.122.2 → 192.168.122.15 SMB2 146 Close Request File:
10 0.260841089 192.168.122.15 → 192.168.122.2 SMB2 130 Close Response, Error: STATUS_FILE_CLOSED
Signed-off-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Reviewed-by: Aurelien Aptel <aaptel@suse.com>
CC: <stable@vger.kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
A customer has reported that their dmesg were being flooded by
CIFS: VFS: \\server Cancelling wait for mid xxx cmd: a
CIFS: VFS: \\server Cancelling wait for mid yyy cmd: b
CIFS: VFS: \\server Cancelling wait for mid zzz cmd: c
because some processes that were performing statfs(2) on the share had
been interrupted due to their automount setup when certain users
logged in and out.
Change it to FYI as they should be mostly informative rather than
error messages.
Signed-off-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Reviewed-by: Aurelien Aptel <aaptel@suse.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Introduced a new field conn_id in TCP_Server_Info structure.
This is a non-persistent unique identifier maintained by the client
for a connection to a file server. For this, a global counter named
tcpSesNextId is maintained. On allocating a new TCP_Server_Info,
this counter is incremented and assigned.
Changed the dynamic tracepoints related to reconnects and
crediting to be more informative (with conn_id printed).
Debugging a crediting issue helped me understand the
important things to print here.
Always call dynamic tracepoints outside the scope of spinlocks.
To do this, copy out the credits and in_flight fields of the
server struct before dropping the lock.
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Pavel Shilovsky <pshilov@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
For failure by timeout waiting for credits, changed the error
returned to the app with EBUSY, instead of ENOTSUPP. This is done
because this situation is possible even in non-buggy cases. i.e.
overloaded server can return 0 credits until done with outstanding
requests. And this feels like a better error to return to the app.
For cases of zero credits found even when there are no requests
in flight, replaced ENOTSUPP with EDEADLK, since we're avoiding
deadlock here by returning error.
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Pavel Shilovsky <pshilov@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
We used to share the CIFS_NEG_OP flag between negotiate and
session authentication. There was an assumption in the code that
CIFS_NEG_OP is used by negotiate only. So introcuded CIFS_SESS_OP
and used it for session setup optypes.
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Pavel Shilovsky <pshilov@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Currently we try to guess if a compound request is going to
succeed waiting for credits or not based on the number of
requests in flight. This approach doesn't work correctly
all the time because there may be only one request in
flight which is going to bring multiple credits satisfying
the compound request.
Change the behavior to fail a request only if there are no requests
in flight at all and proceed waiting for credits otherwise.
Cc: <stable@vger.kernel.org> # 5.1+
Signed-off-by: Pavel Shilovsky <pshilov@microsoft.com>
Reviewed-by: Tom Talpey <tom@talpey.com>
Reviewed-by: Shyam Prasad N <nspmangalore@gmail.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
RHBZ 1848178
The original intent of returning an error in this function
in the patch:
"CIFS: Mask off signals when sending SMB packets"
was to avoid interrupting packet send in the middle of
sending the data (and thus breaking an SMB connection),
but we also don't want to fail the request for non-fatal
signals even before we have had a chance to try to
send it (the reported problem could be reproduced e.g.
by exiting a child process when the parent process was in
the midst of calling futimens to update a file's timestamps).
In addition, since the signal may remain pending when we enter the
sending loop, we may end up not sending the whole packet before
TCP buffers become full. In this case the code returns -EINTR
but what we need here is to return -ERESTARTSYS instead to
allow system calls to be restarted.
Fixes: b30c74c73c78 ("CIFS: Mask off signals when sending SMB packets")
Cc: stable@vger.kernel.org # v5.1+
Signed-off-by: Ronnie Sahlberg <lsahlber@redhat.com>
Reviewed-by: Pavel Shilovsky <pshilov@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
There is at least one suspected bug in crediting changes in cifs.ko
which has come up a few times in the discussions and in a customer
case.
This change adds tracepoints to the code which modifies the server
credit values in any way. The goal is to be able to track the changes
to the credit values of the session to be able to catch when there is
a crediting bug.
Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
A customer has reported that several files in their multi-threaded app
were left with size of 0 because most of the read(2) calls returned
-EINTR and they assumed no bytes were read. Obviously, they could
have fixed it by simply retrying on -EINTR.
We noticed that most of the -EINTR on read(2) were due to real-time
signals sent by glibc to process wide credential changes (SIGRT_1),
and its signal handler had been established with SA_RESTART, in which
case those calls could have been automatically restarted by the
kernel.
Let the kernel decide to whether or not restart the syscalls when
there is a signal pending in __smb_send_rqst() by returning
-ERESTARTSYS. If it can't, it will return -EINTR anyway.
Signed-off-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
CC: Stable <stable@vger.kernel.org>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Reviewed-by: Pavel Shilovsky <pshilov@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
SMB3 crediting is used for flow control, and it can be useful to
trace for problem determination how many credits were acquired
and for which operation.
Here is an example ("trace-cmd record -e *add_credits"):
cifsd-9522 [010] .... 5995.202712: smb3_add_credits:
server=localhost current_mid=0x12 credits=373 credits_to_add=10
cifsd-9522 [010] .... 5995.204040: smb3_add_credits:
server=localhost current_mid=0x15 credits=400 credits_to_add=30
Reviewed-by: Aurelien Aptel <aaptel@suse.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
If server returns ERRBaduid but does not reset transport connection,
we'll keep sending command with a non-valid UID for the server as long
as transport is healthy, without actually recovering. This have been
observed on the field.
This patch adds ERRBaduid handling so that we set CifsNeedReconnect.
map_and_check_smb_error() can be modified to extend use cases.
Signed-off-by: Roberto Bergantinos Corpas <rbergant@redhat.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
|
|
The wait_event_... defines evaluate to long so we should not assign it an int as this may truncate
the value.
Reported-by: Marshall Midden <marshallmidden@gmail.com>
Signed-off-by: Ronnie Sahlberg <lsahlber@redhat.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Pull cifs updates from Steve French:
"22 changesets, 2 for stable.
Includes big performance improvement for large i/o when using
multichannel, also includes DFS fixes"
* tag '5.8-rc-smb3-fixes-part-1' of git://git.samba.org/sfrench/cifs-2.6: (22 commits)
cifs: update internal module version number
cifs: multichannel: try to rebind when reconnecting a channel
cifs: multichannel: use pointer for binding channel
smb3: remove static checker warning
cifs: multichannel: move channel selection above transport layer
cifs: multichannel: always zero struct cifs_io_parms
cifs: dump Security Type info in DebugData
smb3: fix incorrect number of credits when ioctl MaxOutputResponse > 64K
smb3: default to minimum of two channels when multichannel specified
cifs: multichannel: move channel selection in function
cifs: fix minor typos in comments and log messages
smb3: minor update to compression header definitions
cifs: minor fix to two debug messages
cifs: Standardize logging output
smb3: Add new parm "nodelete"
cifs: move some variables off the stack in smb2_ioctl_query_info
cifs: reduce stack use in smb2_compound_op
cifs: get rid of unused parameter in reconn_setup_dfs_targets()
cifs: handle hostnames that resolve to same ip in failover
cifs: set up next DFS target before generic_ip_connect()
...
|
|
Move the channel (TCP_Server_Info*) selection from the tranport
layer to higher in the call stack so that:
- credit handling is done with the server that will actually be used
to send.
* ->wait_mtu_credit
* ->set_credits / set_credits
* ->add_credits / add_credits
* add_credits_and_wake_if
- potential reconnection (smb2_reconnect) done when initializing a
request is checked and done with the server that will actually be
used to send.
To do this:
- remove the cifs_pick_channel() call out of compound_send_recv()
- select channel and pass it down by adding a cifs_pick_channel(ses)
call in:
- smb311_posix_mkdir
- SMB2_open
- SMB2_ioctl
- __SMB2_close
- query_info
- SMB2_change_notify
- SMB2_flush
- smb2_async_readv (if none provided in context param)
- SMB2_read (if none provided in context param)
- smb2_async_writev (if none provided in context param)
- SMB2_write (if none provided in context param)
- SMB2_query_directory
- send_set_info
- SMB2_oplock_break
- SMB311_posix_qfs_info
- SMB2_QFS_info
- SMB2_QFS_attr
- smb2_lockv
- SMB2_lease_break
- smb2_compound_op
- smb2_set_ea
- smb2_ioctl_query_info
- smb2_query_dir_first
- smb2_query_info_comound
- smb2_query_symlink
- cifs_writepages
- cifs_write_from_iter
- cifs_send_async_read
- cifs_read
- cifs_readpages
- add TCP_Server_Info *server param argument to:
- cifs_send_recv
- compound_send_recv
- SMB2_open_init
- SMB2_query_info_init
- SMB2_set_info_init
- SMB2_close_init
- SMB2_ioctl_init
- smb2_iotcl_req_init
- SMB2_query_directory_init
- SMB2_notify_init
- SMB2_flush_init
- build_qfs_info_req
- smb2_hdr_assemble
- smb2_reconnect
- fill_small_buf
- smb2_plain_req_init
- __smb2_plain_req_init
The read/write codepath is different than the rest as it is using
pages, io iterators and async calls. To deal with those we add a
server pointer in the cifs_writedata/cifs_readdata/cifs_io_parms
context struct and set it in:
- cifs_writepages (wdata)
- cifs_write_from_iter (wdata)
- cifs_readpages (rdata)
- cifs_send_async_read (rdata)
The [rw]data->server pointer is eventually copied to
cifs_io_parms->server to pass it down to SMB2_read/SMB2_write.
If SMB2_read/SMB2_write is called from a different place that doesn't
set the server field it will pick a channel.
Some places do not pick a channel and just use ses->server or
cifs_ses_server(ses). All cifs_ses_server(ses) calls are in codepaths
involving negprot/sess.setup.
- SMB2_negotiate (binding channel)
- SMB2_sess_alloc_buffer (binding channel)
- SMB2_echo (uses provided one)
- SMB2_logoff (uses master)
- SMB2_tdis (uses master)
(list not exhaustive)
Signed-off-by: Aurelien Aptel <aaptel@suse.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
This commit moves channel picking code in separate function.
Signed-off-by: Aurelien Aptel <aaptel@suse.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Use pr_fmt to standardize all logging for fs/cifs.
Some logging output had no CIFS: specific prefix.
Now all output has one of three prefixes:
o CIFS:
o CIFS: VFS:
o Root-CIFS:
Miscellanea:
o Convert printks to pr_<level>
o Neaten macro definitions
o Remove embedded CIFS: prefixes from formats
o Convert "illegal" to "invalid"
o Coalesce formats
o Add missing '\n' format terminations
o Consolidate multiple cifs_dbg continuations into single calls
o More consistent use of upper case first word output logging
o Multiline statement argument alignment and wrapping
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
Add a helper to directly set the TCP_CORK sockopt from kernel space
without going through a fake uaccess. Cleanup the callers to avoid
pointless wrappers now that this is a simple function call.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
|
|
When encryption is used, smb2_transform_hdr is defined on the stack and is
passed to the transport. This doesn't work with RDMA as the buffer needs to
be DMA'ed.
Fix it by using kmalloc.
Signed-off-by: Long Li <longli@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
The task which created the MID may be gone by the time cifsd attempts to
call the callbacks on MIDs from cifs_reconnect().
This leads to a use-after-free of the task struct in cifs_wake_up_task:
==================================================================
BUG: KASAN: use-after-free in __lock_acquire+0x31a0/0x3270
Read of size 8 at addr ffff8880103e3a68 by task cifsd/630
CPU: 0 PID: 630 Comm: cifsd Not tainted 5.5.0-rc6+ #119
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014
Call Trace:
dump_stack+0x8e/0xcb
print_address_description.constprop.5+0x1d3/0x3c0
? __lock_acquire+0x31a0/0x3270
__kasan_report+0x152/0x1aa
? __lock_acquire+0x31a0/0x3270
? __lock_acquire+0x31a0/0x3270
kasan_report+0xe/0x20
__lock_acquire+0x31a0/0x3270
? __wake_up_common+0x1dc/0x630
? _raw_spin_unlock_irqrestore+0x4c/0x60
? mark_held_locks+0xf0/0xf0
? _raw_spin_unlock_irqrestore+0x39/0x60
? __wake_up_common_lock+0xd5/0x130
? __wake_up_common+0x630/0x630
lock_acquire+0x13f/0x330
? try_to_wake_up+0xa3/0x19e0
_raw_spin_lock_irqsave+0x38/0x50
? try_to_wake_up+0xa3/0x19e0
try_to_wake_up+0xa3/0x19e0
? cifs_compound_callback+0x178/0x210
? set_cpus_allowed_ptr+0x10/0x10
cifs_reconnect+0xa1c/0x15d0
? generic_ip_connect+0x1860/0x1860
? rwlock_bug.part.0+0x90/0x90
cifs_readv_from_socket+0x479/0x690
cifs_read_from_socket+0x9d/0xe0
? cifs_readv_from_socket+0x690/0x690
? mempool_resize+0x690/0x690
? rwlock_bug.part.0+0x90/0x90
? memset+0x1f/0x40
? allocate_buffers+0xff/0x340
cifs_demultiplex_thread+0x388/0x2a50
? cifs_handle_standard+0x610/0x610
? rcu_read_lock_held_common+0x120/0x120
? mark_lock+0x11b/0xc00
? __lock_acquire+0x14ed/0x3270
? __kthread_parkme+0x78/0x100
? lockdep_hardirqs_on+0x3e8/0x560
? lock_downgrade+0x6a0/0x6a0
? lockdep_hardirqs_on+0x3e8/0x560
? _raw_spin_unlock_irqrestore+0x39/0x60
? cifs_handle_standard+0x610/0x610
kthread+0x2bb/0x3a0
? kthread_create_worker_on_cpu+0xc0/0xc0
ret_from_fork+0x3a/0x50
Allocated by task 649:
save_stack+0x19/0x70
__kasan_kmalloc.constprop.5+0xa6/0xf0
kmem_cache_alloc+0x107/0x320
copy_process+0x17bc/0x5370
_do_fork+0x103/0xbf0
__x64_sys_clone+0x168/0x1e0
do_syscall_64+0x9b/0xec0
entry_SYSCALL_64_after_hwframe+0x49/0xbe
Freed by task 0:
save_stack+0x19/0x70
__kasan_slab_free+0x11d/0x160
kmem_cache_free+0xb5/0x3d0
rcu_core+0x52f/0x1230
__do_softirq+0x24d/0x962
The buggy address belongs to the object at ffff8880103e32c0
which belongs to the cache task_struct of size 6016
The buggy address is located 1960 bytes inside of
6016-byte region [ffff8880103e32c0, ffff8880103e4a40)
The buggy address belongs to the page:
page:ffffea000040f800 refcount:1 mapcount:0 mapping:ffff8880108da5c0
index:0xffff8880103e4c00 compound_mapcount: 0
raw: 4000000000010200 ffffea00001f2208 ffffea00001e3408 ffff8880108da5c0
raw: ffff8880103e4c00 0000000000050003 00000001ffffffff 0000000000000000
page dumped because: kasan: bad access detected
Memory state around the buggy address:
ffff8880103e3900: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
ffff8880103e3980: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
>ffff8880103e3a00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
^
ffff8880103e3a80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
ffff8880103e3b00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================
This can be reliably reproduced by adding the below delay to
cifs_reconnect(), running find(1) on the mount, restarting the samba
server while find is running, and killing find during the delay:
spin_unlock(&GlobalMid_Lock);
mutex_unlock(&server->srv_mutex);
+ msleep(10000);
+
cifs_dbg(FYI, "%s: issuing mid callbacks\n", __func__);
list_for_each_safe(tmp, tmp2, &retry_list) {
mid_entry = list_entry(tmp, struct mid_q_entry, qhead);
Fix this by holding a reference to the task struct until the MID is
freed.
Signed-off-by: Vincent Whitchurch <vincent.whitchurch@axis.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
CC: Stable <stable@vger.kernel.org>
Reviewed-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Reviewed-by: Pavel Shilovsky <pshilov@microsoft.com>
|
|
After doing mount() successfully we call cifs_try_adding_channels()
which will open as many channels as it can.
Channels are closed when the master session is closed.
The master connection becomes the first channel.
,-------------> global cifs_tcp_ses_list <-------------------------.
| |
'- TCP_Server_Info <--> TCP_Server_Info <--> TCP_Server_Info <-'
(master con) (chan#1 con) (chan#2 con)
| ^ ^ ^
v '--------------------|--------------------'
cifs_ses |
- chan_count = 3 |
- chans[] ---------------------'
- smb3signingkey[]
(master signing key)
Note how channel connections don't have sessions. That's because
cifs_ses can only be part of one linked list (list_head are internal
to the elements).
For signing keys, each channel has its own signing key which must be
used only after the channel has been bound. While it's binding it must
use the master session signing key.
For encryption keys, since channel connections do not have sessions
attached we must now find matching session by looping over all sessions
in smb2_get_enc_key().
Each channel is opened like a regular server connection but at the
session setup request step it must set the
SMB2_SESSION_REQ_FLAG_BINDING flag and use the session id to bind to.
Finally, while sending in compound_send_recv() for requests that
aren't negprot, ses-setup or binding related, use a channel by cycling
through the available ones (round-robin).
Signed-off-by: Aurelien Aptel <aaptel@suse.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
As we get down to the transport layer, plenty of functions are passed
the session pointer and assume the transport to use is ses->server.
Instead we modify those functions to pass (ses, server) so that we
can decouple the session from the server.
Signed-off-by: Aurelien Aptel <aaptel@suse.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
During reconnecting, the transport may have already been destroyed and is in
the process being reconnected. In this case, return -EAGAIN to not fail and
to retry this I/O.
Signed-off-by: Long Li <longli@microsoft.com>
Cc: stable@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
|
|
When an OPEN command is cancelled we mark a mid as
cancelled and let the demultiplex thread process it
by closing an open handle. The problem is there is
a race between a system call thread and the demultiplex
thread and there may be a situation when the mid has
been already processed before it is set as cancelled.
Fix this by processing cancelled requests when mids
are being destroyed which means that there is only
one thread referencing a particular mid. Also set
mids as cancelled unconditionally on their state.
Cc: Stable <stable@vger.kernel.org>
Tested-by: Frank Sorenson <sorenson@redhat.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Signed-off-by: Pavel Shilovsky <pshilov@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
|
Linus Torvalds
Stefan Metzmacher
Peter Zijlstra
Zhang Xiaoxu
Steve French
Enzo Matsumiya
Shyam Prasad N
Vincent Whitchurch
Haowen Bai
David Howells
Paulo Alcantara
Pavel Shilovsky
Ronnie Sahlberg
Roberto Bergantinos Corpas
Aurelien Aptel
Joe Perches
Christoph Hellwig
Long Li