f2fs recovery flow is relying on dnode block link list, it means fsynced
file recovery depends on previous dnode's persistence in the list, so
during fsync() we should wait on all regular inode's dnode writebacked
before issuing flush.
By this way, we can avoid dnode block list being broken by out-of-order
IO submission due to IO scheduler or driver.
Sheng Yong helps to do the test with this patch:
Target:/data (f2fs, -)
64MB / 32768KB / 4KB / 8
1 / PERSIST / Index
Base:
SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS)
1 867.82 204.15 41440.03 41370.54 680.8 1025.94 1031.08
2 871.87 205.87 41370.3 40275.2 791.14 1065.84 1101.7
3 866.52 205.69 41795.67 40596.16 694.69 1037.16 1031.48
Avg 868.7366667 205.2366667 41535.33333 40747.3 722.21 1042.98 1054.753333
After:
SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS)
1 798.81 202.5 41143 40613.87 602.71 838.08 913.83
2 805.79 206.47 40297.2 41291.46 604.44 840.75 924.27
3 814.83 206.17 41209.57 40453.62 602.85 834.66 927.91
Avg 806.4766667 205.0466667 40883.25667 40786.31667 603.3333333 837.83 922.0033333
Patched/Original:
0.928332713 0.999074239 0.984300676 1.000957528 0.835398753 0.803303994 0.874141189
It looks like atomic write will suffer performance regression.
I suspect that the criminal is that we forcing to wait all dnode being in
storage cache before we issue PREFLUSH+FUA.
BTW, will commit ("f2fs: don't need to wait for node writes for atomic write")
cause the problem: we will lose data of last transaction after SPO, even if
atomic write return no error:
- atomic_open();
- write() P1, P2, P3;
- atomic_commit();
- writeback data: P1, P2, P3;
- writeback node: N1, N2, N3; <--- If N1, N2 is not writebacked, N3 with fsync_mark is
writebacked, In SPOR, we won't find N3 since node chain is broken, turns out that losing
last transaction.
- preflush + fua;
- power-cut
If we don't wait dnode writeback for atomic_write:
SEQ-RD(MB/s) SEQ-WR(MB/s) RND-RD(IOPS) RND-WR(IOPS) Insert(TPS) Update(TPS) Delete(TPS)
1 779.91 206.03 41621.5 40333.16 716.9 1038.21 1034.85
2 848.51 204.35 40082.44 39486.17 791.83 1119.96 1083.77
3 772.12 206.27 41335.25 41599.65 723.29 1055.07 971.92
Avg 800.18 205.55 41013.06333 40472.99333 744.0066667 1071.08 1030.18
Patched/Original:
0.92108464 1.001526693 0.987425886 0.993268102 1.030180511 1.026942031 0.976702294
SQLite's performance recovers.
Jaegeuk:
"Practically, I don't see db corruption becase of this. We can excuse to lose
the last transaction."
Finally, we decide to keep original implementation of atomic write interface
sematics that we don't wait all dnode writeback before preflush+fua submission.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Return statements in functions returning bool should use true or false
instead of an integer value.
This issue was detected with the help of Coccinelle.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
There is a subtle race condition to invoke f2fs_bug_on() in shutdown tests. I've
confirmed that the last checkpoint is preserved in consistent state, so it'd be
fine to just return error at this moment.
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
PG_checked flag will be set on data page during GC, later, we can
recognize such page by the flag and migrate page to cold segment.
But previously, we don't clear this flag when invalidating data page,
after page redirtying, we will write it into wrong log.
Let's clear PG_checked flag in set_page_dirty() to avoid this.
Signed-off-by: Weichao Guo <guoweichao@huawei.com>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Rebuild the AGI header items with some help from the rmapbt.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
mounting with the "snapshots=" mount parm allows a read-only
view of a previous version of a file system (see MS-SMB2
and "timewarp" tokens, section 2.2.13.2.6) based on the timestamp
passed in on the snapshots mount parm.
Add processing to optionally send this create context.
Example output:
/mnt1 is mounted with "snapshots=..." and will see an earlier
version of the directory, with three fewer files than /mnt2
the current version of the directory.
root@Ubuntu-17-Virtual-Machine:~/cifs-2.6# cat /proc/mounts | grep cifs
//172.22.149.186/public /mnt1 cifs
ro,relatime,vers=default,cache=strict,username=smfrench,uid=0,noforceuid,gid=0,noforcegid,addr=172.22.149.186,file_mode=0755,dir_mode=0755,soft,nounix,mapposix,rsize=1048576,wsize=1048576,echo_interval=60,snapshot=131748608570000000,actimeo=1
//172.22.149.186/public /mnt2 cifs
rw,relatime,vers=default,cache=strict,username=smfrench,uid=0,noforceuid,gid=0,noforcegid,addr=172.22.149.186,file_mode=0755,dir_mode=0755,soft,nounix,mapposix,rsize=1048576,wsize=1048576,echo_interval=60,actimeo=1
root@Ubuntu-17-Virtual-Machine:~/cifs-2.6# ls /mnt1
EmptyDir newerdir
root@Ubuntu-17-Virtual-Machine:~/cifs-2.6# ls /mnt1/newerdir
root@Ubuntu-17-Virtual-Machine:~/cifs-2.6# ls /mnt2
EmptyDir file newerdir newestdir timestamp-trace.cap
root@Ubuntu-17-Virtual-Machine:~/cifs-2.6# ls /mnt2/newerdir
new-file-not-in-snapshot
Snapshots are extremely useful for comparing previous versions of files or directories,
and recovering from data corruptions or mistakes.
Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
When enumerating snapshots, the last few bytes of the final
snapshot could be left off since we were miscalculating the
length returned (leaving off the sizeof struct SRV_SNAPSHOT_ARRAY)
See MS-SMB2 section 2.2.32.2. In addition fixup the length used
to allow smaller buffer to be passed in, in order to allow
returning the size of the whole snapshot array more easily.
Sample userspace output with a kernel patched with this
(mounted to a Windows volume with two snapshots).
Before this patch, the second snapshot would be missing a
few bytes at the end.
~/cifs-2.6# ~/enum-snapshots /mnt/file
press enter to issue the ioctl to retrieve snapshot information ...
size of snapshot array = 102
Num snapshots: 2 Num returned: 2 Array Size: 102
Snapshot 0:@GMT-2018.06.30-19.34.17
Snapshot 1:@GMT-2018.06.30-19.33.37
CC: Stable <stable@vger.kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-by: Pavel Shilovsky <pshilov@microsoft.com>
RCU pathwalk relies upon the assumption that anything that changes
->d_inode of a dentry will invalidate its ->d_seq. That's almost
true - the one exception is that the final dput() of already unhashed
dentry does *not* touch ->d_seq at all. Unhashing does, though,
so for anything we'd found by RCU dcache lookup we are fine.
Unfortunately, we can *start* with an unhashed dentry or jump into
it.
We could try and be careful in the (few) places where that could
happen. Or we could just make the final dput() invalidate the damn
thing, unhashed or not. The latter is much simpler and easier to
backport, so let's do it that way.
Reported-by: "Dae R. Jeong" <threeearcat@gmail.com>
Cc: stable@vger.kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
__legitimize_mnt() has two problems - one is that in case of success
the check of mount_lock is not ordered wrt preceding increment of
refcount, making it possible to have successful __legitimize_mnt()
on one CPU just before the otherwise final mntpu() on another,
with __legitimize_mnt() not seeing mntput() taking the lock and
mntput() not seeing the increment done by __legitimize_mnt().
Solved by a pair of barriers.
Another is that failure of __legitimize_mnt() on the second
read_seqretry() leaves us with reference that'll need to be
dropped by caller; however, if that races with final mntput()
we can end up with caller dropping rcu_read_lock() and doing
mntput() to release that reference - with the first mntput()
having freed the damn thing just as rcu_read_lock() had been
dropped. Solution: in "do mntput() yourself" failure case
grab mount_lock, check if MNT_DOOMED has been set by racing
final mntput() that has missed our increment and if it has -
undo the increment and treat that as "failure, caller doesn't
need to drop anything" case.
It's not easy to hit - the final mntput() has to come right
after the first read_seqretry() in __legitimize_mnt() *and*
manage to miss the increment done by __legitimize_mnt() before
the second read_seqretry() in there. The things that are almost
impossible to hit on bare hardware are not impossible on SMP
KVM, though...
Reported-by: Oleg Nesterov <oleg@redhat.com>
Fixes: 48a066e72d ("RCU'd vsfmounts")
Cc: stable@vger.kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
mntput_no_expire() does the calculation of total refcount under mount_lock;
unfortunately, the decrement (as well as all increments) are done outside
of it, leading to false positives in the "are we dropping the last reference"
test. Consider the following situation:
* mnt is a lazy-umounted mount, kept alive by two opened files. One
of those files gets closed. Total refcount of mnt is 2. On CPU 42
mntput(mnt) (called from __fput()) drops one reference, decrementing component
* After it has looked at component #0, the process on CPU 0 does
mntget(), incrementing component #0, gets preempted and gets to run again -
on CPU 69. There it does mntput(), which drops the reference (component #69)
and proceeds to spin on mount_lock.
* On CPU 42 our first mntput() finishes counting. It observes the
decrement of component #69, but not the increment of component #0. As the
result, the total it gets is not 1 as it should've been - it's 0. At which
point we decide that vfsmount needs to be killed and proceed to free it and
shut the filesystem down. However, there's still another opened file
on that filesystem, with reference to (now freed) vfsmount, etc. and we are
screwed.
It's not a wide race, but it can be reproduced with artificial slowdown of
the mnt_get_count() loop, and it should be easier to hit on SMP KVM setups.
Fix consists of moving the refcount decrement under mount_lock; the tricky
part is that we want (and can) keep the fast case (i.e. mount that still
has non-NULL ->mnt_ns) entirely out of mount_lock. All places that zero
mnt->mnt_ns are dropping some reference to mnt and they call synchronize_rcu()
before that mntput(). IOW, if mntput() observes (under rcu_read_lock())
a non-NULL ->mnt_ns, it is guaranteed that there is another reference yet to
be dropped.
Reported-by: Jann Horn <jannh@google.com>
Tested-by: Jann Horn <jannh@google.com>
Fixes: 48a066e72d ("RCU'd vsfmounts")
Cc: stable@vger.kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Return statements in functions returning bool should use true or false
instead of an integer value.
This issue was detected with the help of Coccinelle.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
write_op[] is never modified, so make it 'const'.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
READ_BUF(8);
dummy = be32_to_cpup(p++);
dummy = be32_to_cpup(p++);
...
READ_BUF(4);
dummy = be32_to_cpup(p++);
Assigning value to "dummy" here, but that stored value
is overwritten before it can be used.
At the same time READ_BUF() will re-update the pointer p.
delete invalid assignment statements
Signed-off-by: nixiaoming <nixiaoming@huawei.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <trondmy@hammerspace.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
I've given up on the idea of zero-copy handling of SYMLINK on the
server side. This is because the Linux VFS symlink API requires the
symlink pathname to be in a NUL-terminated kmalloc'd buffer. The
NUL-termination is going to be problematic (watching out for
landing on a page boundary and dealing with a 4096-byte pathname).
I don't believe that SYMLINK creation is on a performance path or is
requested frequently enough that it will cause noticeable CPU cache
pollution due to data copies.
There will be two places where a transport callout will be necessary
to fill in the rqstp: one will be in the svc_fill_symlink_pathname()
helper that is used by NFSv2 and NFSv3, and the other will be in
nfsd4_decode_create().
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
fill_in_write_vector() is nearly the same logic as
svc_fill_write_vector(), but there are a few differences so that
the former can handle multiple WRITE payloads in a single COMPOUND.
svc_fill_write_vector() can be adjusted so that it can be used in
the NFSv4 WRITE code path too. Instead of assuming the pages are
coming from rq_args.pages, have the caller pass in the page list.
The immediate benefit is a reduction of code duplication. It also
prevents the NFSv4 WRITE decoder from passing an empty vector
element when the transport has provided the payload in the xdr_buf's
page array.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.
Warning level 2 was used: -Wimplicit-fallthrough=2
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
nfsd and lockd call vfs_lock_file() to lock/unlock the inode
returned by locks_inode(file).
Many places in nfsd/lockd code use the inode returned by
file_inode(file) for lock manipulation. With Overlayfs, file_inode()
(the underlying inode) is not the same object as locks_inode() (the
overlay inode). This can result in "Leaked POSIX lock" messages
and eventually to a kernel crash as reported by Eddie Horng:
https://marc.info/?l=linux-unionfs&m=153086643202072&w=2
Fix all the call sites in nfsd/lockd that should use locks_inode().
This is a correctness bug that manifested when overlayfs gained
NFS export support in v4.16.
Reported-by: Eddie Horng <eddiehorng.tw@gmail.com>
Tested-by: Eddie Horng <eddiehorng.tw@gmail.com>
Cc: Jeff Layton <jlayton@kernel.org>
Fixes: 8383f17488 ("ovl: wire up NFS export operations")
Cc: stable@vger.kernel.org
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
A COPY with unstable write data needs a simple sync commit.
Filehandle value is gotten as a part of the inner loop so in
case of a reboot retry it should get the new value.
Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
If client sent async COPY and server replied with
ERR_OFFLOAD_NO_REQS, client should retry with a synchronous copy.
Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
When COPY is killed by the user send OFFLOAD_CANCEL to server
processing the copy.
Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
It's possible that server replies back with CB_OFFLOAD call and
COPY reply at the same time such that client will process
CB_OFFLOAD before reply to COPY. For that keep a list of pending
callback stateids received and then before waiting on completion
check the pending list.
Cleanup any pending copies on the client shutdown.
Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Change xdr to always send COPY asynchronously.
Keep the list copies send in a list under a server structure.
Once copy is sent, it waits on a completion structure that will
be signalled by the callback thread that receives CB_OFFLOAD.
If CB_OFFLOAD returned an error and even if it returned partial
bytes, ignore them (as we can't commit without a verifier to
match) and return an error.
Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
If server returns async reply, it must include a callback stateid,
wr_callback_id in the write_response4.
Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
When a direct-write completes, a work_struct is schedule to handle
the completion.
When NFS is being used for swap, the direct write might be a swap-out,
so memory allocation can block until the write completes.
The work queue currently used is not WQ_MEM_RECLAIM, so tasks
can block waiting for memory - this leads to deadlock.
So use nfsiod_workqueue instead. This will always have a running
thread, and work items should never block waiting for memory.
Signed-off-by: Neil Brown <neilb@suse.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Error code is set in the error handling cases but never used. Fix it.
Fixes: 937e3133cd ("NFSv4.1: Ensure we clear the SP4_MACH_CRED flags in nfs4_sp4_select_mode()")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Return statements in functions returning bool should use true or false
instead of an integer value.
This issue was detected with the help of Coccinelle.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Layout segment validity is determined only by the NFS_LSEG_VALID flag. If
it is set, the layout segment is finable. As it is, when the flexfiles
driver sets NFS_LSEG_LAYOUTRETURN to indicate that we cannot discard
the layout segment, but that it must be returned, then this can result
in an unnecessary layoutget storm.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.
Warning level 2 was used: -Wimplicit-fallthrough=2
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
If the object being renamed from one directory to another is also
a directory, then 'nlink' will change for both directories.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Ensure that we always bump or drop the nlink count on the parent directory
when we do a mkdir or a rmdir(). This needs to be done by hand as we don't
have pre/post op attributes.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
If the server tells us that out layoutreturn raced with another layout
update, then we must ensure that the new layout segments are not in use
before we resend with an updated layout stateid.
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
The spinlock handling in this file has changed significantly since this
comment was written, and the file_lock_lock is no more. In addition,
this overall comment no longer applies. Deleting an entry now requires
both locks.
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Function update_rgrp_lvb_unlinked used to do the same thing as
be32_add_cpu. This patch removes it in favor of using be32_add_cpu
directly.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Reviewed-by: Andrew Price <anprice@redhat.com>
When CONFIG_CIFS_STATS2 is enabled keep counters for slow
commands (ie server took longer than 1 second to respond)
by SMB2/SMB3 command code. This can help in diagnosing
whether performance problems are on server (instead of
client) and which commands are causing the problem.
Sample output (the new lines contain words "slow responses ...")
$ cat /proc/fs/cifs/Stats
Resources in use
CIFS Session: 1
Share (unique mount targets): 2
SMB Request/Response Buffer: 1 Pool size: 5
SMB Small Req/Resp Buffer: 1 Pool size: 30
Total Large 10 Small 490 Allocations
Operations (MIDs): 0
0 session 0 share reconnects
Total vfs operations: 67 maximum at one time: 2
4 slow responses from localhost for command 5
1 slow responses from localhost for command 6
1 slow responses from localhost for command 14
1 slow responses from localhost for command 16
1) \\localhost\test
SMBs: 243
Bytes read: 1024000 Bytes written: 104857600
TreeConnects: 1 total 0 failed
TreeDisconnects: 0 total 0 failed
Creates: 40 total 0 failed
Closes: 39 total 0 failed
...
Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-by: Pavel Shilovsky <pshilov@microsoft.com>
Reviewed-by: Aurelien Aptel <aaptel@suse.com>
server->secmech.sdeschmacsha256 is not properly initialized before
smb2_shash_allocate(), set shash after that call.
also fix typo in error message
Fixes: 8de8c4608f ("cifs: Fix validation of signed data in smb2")
Signed-off-by: Aurelien Aptel <aaptel@suse.com>
Reviewed-by: Paulo Alcantara <palcantara@suse.com>
Reported-by: Xiaoli Feng <xifeng@redhat.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
CC: Stable <stable@vger.kernel.org>
