There is a nasty bug in the SCSI SG_IO ioctl that in some circumstances
leads to one process writing data into the address space of some other
random unrelated process if the ioctl is interrupted by a signal.
What happens is the following:
- A process issues an SG_IO ioctl with direction DXFER_FROM_DEV (ie the
underlying SCSI command will transfer data from the SCSI device to
the buffer provided in the ioctl)
- Before the command finishes, a signal is sent to the process waiting
in the ioctl. This will end up waking up the sg_ioctl() code:
result = wait_event_interruptible(sfp->read_wait,
(srp_done(sfp, srp) || sdp->detached));
but neither srp_done() nor sdp->detached is true, so we end up just
setting srp->orphan and returning to userspace:
srp->orphan = 1;
write_unlock_irq(&sfp->rq_list_lock);
return result; /* -ERESTARTSYS because signal hit process */
At this point the original process is done with the ioctl and
blithely goes ahead handling the signal, reissuing the ioctl, etc.
- Eventually, the SCSI command issued by the first ioctl finishes and
ends up in sg_rq_end_io(). At the end of that function, we run through:
write_lock_irqsave(&sfp->rq_list_lock, iflags);
if (unlikely(srp->orphan)) {
if (sfp->keep_orphan)
srp->sg_io_owned = 0;
else
done = 0;
}
srp->done = done;
write_unlock_irqrestore(&sfp->rq_list_lock, iflags);
if (likely(done)) {
/* Now wake up any sg_read() that is waiting for this
* packet.
*/
wake_up_interruptible(&sfp->read_wait);
kill_fasync(&sfp->async_qp, SIGPOLL, POLL_IN);
kref_put(&sfp->f_ref, sg_remove_sfp);
} else {
INIT_WORK(&srp->ew.work, sg_rq_end_io_usercontext);
schedule_work(&srp->ew.work);
}
Since srp->orphan *is* set, we set done to 0 (assuming the
userspace app has not set keep_orphan via an SG_SET_KEEP_ORPHAN
ioctl), and therefore we end up scheduling sg_rq_end_io_usercontext()
to run in a workqueue.
- In workqueue context we go through sg_rq_end_io_usercontext() ->
sg_finish_rem_req() -> blk_rq_unmap_user() -> ... ->
bio_uncopy_user() -> __bio_copy_iov() -> copy_to_user().
The key point here is that we are doing copy_to_user() on a
workqueue -- that is, we're on a kernel thread with current->mm
equal to whatever random previous user process was scheduled before
this kernel thread. So we end up copying whatever data the SCSI
command returned to the virtual address of the buffer passed into
the original ioctl, but it's quite likely we do this copying into a
different address space!
As suggested by James Bottomley <James.Bottomley@hansenpartnership.com>,
add a check for current->mm (which is NULL if we're on a kernel thread
without a real userspace address space) in bio_uncopy_user(), and skip
the copy if we're on a kernel thread.
There's no reason that I can think of for any caller of bio_uncopy_user()
to want to do copying on a kernel thread with a random active userspace
address space.
Huge thanks to Costa Sapuntzakis <costa@purestorage.com> for the
original pointer to this bug in the sg code.
Signed-off-by: Roland Dreier <roland@purestorage.com>
Tested-by: David Milburn <dmilburn@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: <stable@vger.kernel.org>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>
Pull CIFS fixes from Steve French:
"A set of small cifs fixes, including 3 relating to symlink handling"
* 'for-next' of git://git.samba.org/sfrench/cifs-2.6:
cifs: don't instantiate new dentries in readdir for inodes that need to be revalidated immediately
cifs: set sb->s_d_op before calling d_make_root()
cifs: fix bad error handling in crypto code
cifs: file: initialize oparms.reconnect before using it
Do not attempt to do cifs operations reading symlinks with SMB2
cifs: extend the buffer length enought for sprintf() using
so that if ext4 is built as a module, to allow it to be unloaded.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)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=NuDG
-----END PGP SIGNATURE-----
Merge tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4
Pull more ext4 bugfixes from Ted Ts'o:
"A number of miscellaneous ext4 bugs fixes for v3.11, including a fix
so that if ext4 is built as a module, to allow it to be unloaded"
* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
ext4: flush the extent status cache during EXT4_IOC_SWAP_BOOT
ext4: fix mount/remount error messages for incompatible mount options
ext4: allow the mount options nodelalloc and data=journal
Previously we weren't swapping only some of the extent_status LRU
fields during the processing of the EXT4_IOC_SWAP_BOOT ioctl. The
much safer thing to do is to just completely flush the extent status
tree when doing the swap.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Cc: Zheng Liu <gnehzuil.liu@gmail.com>
Cc: stable@vger.kernel.org
Pull btrfs fixes from Chris Mason:
"These are assorted fixes, mostly from Josef nailing down xfstests
runs. Zach also has a long standing fix for problems with readdir
wrapping f_pos (or ctx->pos)
These patches were spread out over different bases, so I rebased
things on top of rc4 and retested overnight"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
btrfs: don't loop on large offsets in readdir
Btrfs: check to see if root_list is empty before adding it to dead roots
Btrfs: release both paths before logging dir/changed extents
Btrfs: allow splitting of hole em's when dropping extent cache
Btrfs: make sure the backref walker catches all refs to our extent
Btrfs: fix backref walking when we hit a compressed extent
Btrfs: do not offset physical if we're compressed
Btrfs: fix extent buffer leak after backref walking
Btrfs: fix a bug of snapshot-aware defrag to make it work on partial extents
btrfs: fix file truncation if FALLOC_FL_KEEP_SIZE is specified
- Stable patch for lockd to fix Oopses due to inappropriate calls to
utsname()->nodename
- Stable patches for sunrpc to fix Oopses on shutdown when using
AF_LOCAL sockets with rpcbind
- Fix memory leak and error checking issues in nfs4_proc_lookup_mountpoint
- Fix a regression with the sync mount option failing to work for nfs4 mounts
- Fix a writeback performance issue when doing cache invalidation
- Remove an incorrect call to nfs_setsecurity in nfs_fhget
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.13 (GNU/Linux)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=W12x
-----END PGP SIGNATURE-----
Merge tag 'nfs-for-3.11-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Pull NFS client bugfixes from Trond Myklebust:
- Stable patch for lockd to fix Oopses due to inappropriate calls to
utsname()->nodename
- Stable patches for sunrpc to fix Oopses on shutdown when using
AF_LOCAL sockets with rpcbind
- Fix memory leak and error checking issues in nfs4_proc_lookup_mountpoint
- Fix a regression with the sync mount option failing to work for nfs4
mounts
- Fix a writeback performance issue when doing cache invalidation
- Remove an incorrect call to nfs_setsecurity in nfs_fhget
* tag 'nfs-for-3.11-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
NFSv4: Fix up nfs4_proc_lookup_mountpoint
NFS: Remove unnecessary call to nfs_setsecurity in nfs_fhget()
NFSv4: Fix the sync mount option for nfs4 mounts
NFS: Fix writeback performance issue on cache invalidation
SUNRPC: If the rpcbind channel is disconnected, fail the call to unregister
SUNRPC: Don't auto-disconnect from the local rpcbind socket
LOCKD: Don't call utsname()->nodename from nlmclnt_setlockargs
Pull nfsd fixes from Bruce Fields:
"Some fixes for a 4.1 feature that in retrospect probably should have
waited for 3.12.... But it appears to be working now"
* 'for-3.11' of git://linux-nfs.org/~bfields/linux:
nfsd: Fix SP4_MACH_CRED negotiation in EXCHANGE_ID
nfsd4: Fix MACH_CRED NULL dereference
When btrfs readdir() hits the last entry it sets the readdir offset to a
huge value to stop buggy apps from breaking when the same name is
returned by readdir() with concurrent rename()s.
But unconditionally setting the offset to INT_MAX causes readdir() to
loop returning any entries with offsets past INT_MAX. It only takes a
few hours of constant file creation and removal to create entries past
INT_MAX.
So let's set the huge offset to LLONG_MAX if the last entry has already
overflowed 32bit loff_t. Without large offsets behaviour is identical.
With large offsets 64bit apps will work and 32bit apps will be no more
broken than they currently are if they see large offsets.
Signed-off-by: Zach Brown <zab@redhat.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
A user reported a panic when running with autodefrag and deleting snapshots.
This is because we could end up trying to add the root to the dead roots list
twice. To fix this check to see if we are empty before adding ourselves to the
dead roots list. Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
The ceph guys tripped over this bug where we were still holding onto the
original path that we used to copy the inode with when logging. This is based
on Chris's fix which was reported to fix the problem. We need to drop the paths
in two cases anyway so just move the drop up so that we don't have duplicate
code. Thanks,
Cc: stable@vger.kernel.org
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
I noticed while running multi-threaded fsync tests that sometimes fsck would
complain about an improper gap. This happens because we fail to add a hole
extent to the file, which was happening when we'd split a hole EM because
btrfs_drop_extent_cache was just discarding the whole em instead of splitting
it. So this patch fixes this by allowing us to split a hole em properly, which
means that added holes actually get logged properly and we no longer see this
fsck error. Thankfully we're tolerant of these sort of problems so a user would
not see any adverse effects of this bug, other than fsck complaining. Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
Because we don't mess with the offset into the extent for compressed we will
properly find both extents for this case
[extent a][extent b][rest of extent a]
but because we already added a ref for the front half we won't add the inode
information for the second half. This causes us to leak that memory and not
print out the other offset when we do logical-resolve. So fix this by calling
ulist_add_merge and then add our eie to the existing entry if there is one.
With this patch we get both offsets out of logical-resolve. With this and the
other 2 patches I've sent we now pass btrfs/276 on my vm with compress-force=lzo
set. Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
If you do btrfs inspect-internal logical-resolve on a compressed extent that has
been partly overwritten it won't find anything. This is because we try and
match the extent offset we've searched for based on the extent offset in the
data extent entry. However this doesn't work for compressed extents because the
offsets are for the uncompressed size, not the compressed size. So instead only
do this check if we are not compressed, that way we can get an actual entry for
the physical offset rather than nothing for compressed. Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
xfstest btrfs/276 was freaking out on slower boxes partly because fiemap was
offsetting the physical based on the extent offset. This is perfectly fine with
uncompressed extents, however the extent offset is into the uncompressed area,
not the compressed. So we can return a physical value that isn't at all within
the area we have allocated on disk. Fix this by returning the start of the
extent if it is compressed no matter what the offset. Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
commit 47fb091fb787420cd195e66f162737401cce023f(Btrfs: fix unlock after free on rewinded tree blocks)
takes an extra increment on the reference of allocated dummy extent buffer, so now we
cannot free this dummy one, and end up with extent buffer leak.
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Reviewed-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
For partial extents, snapshot-aware defrag does not work as expected,
since
a) we use the wrong logical offset to search for parents, which should be
disk_bytenr + extent_offset, not just disk_bytenr,
b) 'offset' returned by the backref walking just refers to key.offset, not
the 'offset' stored in btrfs_extent_data_ref which is
(key.offset - extent_offset).
The reproducer:
$ mkfs.btrfs sda
$ mount sda /mnt
$ btrfs sub create /mnt/sub
$ for i in `seq 5 -1 1`; do dd if=/dev/zero of=/mnt/sub/foo bs=5k count=1 seek=$i conv=notrunc oflag=sync; done
$ btrfs sub snap /mnt/sub /mnt/snap1
$ btrfs sub snap /mnt/sub /mnt/snap2
$ sync; btrfs filesystem defrag /mnt/sub/foo;
$ umount /mnt
$ btrfs-debug-tree sda (Here we can check whether the defrag operation is snapshot-awared.
This addresses the above two problems.
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
Create a small file and fallocate it to a big size with
FALLOC_FL_KEEP_SIZE option, then truncate it back to the
small size again, the disk free space is not changed back
in this case. i.e,
total 4
-rw-r--r-- 1 root root 512 Jun 28 11:35 test
Filesystem Size Used Avail Use% Mounted on
....
/dev/sdb1 8.0G 56K 7.2G 1% /mnt
-rw-r--r-- 1 root root 512 Jun 28 11:35 /mnt/test
Filesystem Size Used Avail Use% Mounted on
....
/dev/sdb1 8.0G 5.1G 2.2G 70% /mnt
Filesystem Size Used Avail Use% Mounted on
....
/dev/sdb1 8.0G 5.1G 2.2G 70% /mnt
With this fix, the truncated up space is back as:
Filesystem Size Used Avail Use% Mounted on
....
/dev/sdb1 8.0G 56K 7.2G 1% /mnt
Signed-off-by: Jie Liu <jeff.liu@oracle.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
Signed-off-by: Chris Mason <chris.mason@fusionio.com>
device_close()->recalc_sigpending() is not needed, sigprocmask() takes
care of TIF_SIGPENDING correctly.
And without ->siglock it is racy and wrong, it can wrongly clear
TIF_SIGPENDING and miss a signal.
But even with this patch device_close() is still buggy:
1. sigprocmask() should not be used, we have set_task_blocked(),
but this is minor.
2. We should never block SIGKILL or SIGSTOP, and this is what
the code tries to do.
3. This can't protect against SIGKILL or SIGSTOP anyway. Another
thread can do signal_wake_up(), say, do_signal_stop() or
complete_signal() or debugger.
4. sigprocmask(SIG_BLOCK, allsigs) doesn't necessarily clears
TIF_SIGPENDING, say, freezing() or ->jobctl.
5. device_write() looks equally wrong by the same reason.
Looks like, this tries to protect some wait_event_interruptible() logic
from signals, it should be turned into uninterruptible wait. Or we need
to implement something like signals_stop/start for such a use-case.
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 5688978 ("ext4: improve handling of conflicting mount options")
introduced incorrect messages shown while choosing wrong mount options.
First of all, both cases of incorrect mount options,
"data=journal,delalloc" and "data=journal,dioread_nolock" result in
the same error message.
Secondly, the problem above isn't solved for remount option: the
mismatched parameter is simply ignored. Moreover, ext4_msg states
that remount with options "data=journal,delalloc" succeeded, which is
not true.
To fix it up, I added a simple check after parse_options() call to
ensure that data=journal and delalloc/dioread_nolock parameters are
not present at the same time.
Signed-off-by: Piotr Sarna <p.sarna@partner.samsung.com>
Acked-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Cc: stable@vger.kernel.org
Commit 26092bf ("ext4: use a table-driven handler for mount options")
wrongly disallows the specifying the mount options nodelalloc and
data=journal simultaneously. This is incorrect; it should have only
disallowed the combination of delalloc and data=journal
simultaneously.
Reported-by: Piotr Sarna <p.sarna@partner.samsung.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Cc: stable@vger.kernel.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)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=xqB3
-----END PGP SIGNATURE-----
Merge tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4
Pull ext4 bugfixes from Ted Ts'o.
Misc ext4 fixes, delayed by Ted moving mail servers and email getting
marked as spam due to bad spf records.
* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
ext4: add WARN_ON to check the length of allocated blocks
ext4: fix retry handling in ext4_ext_truncate()
ext4: destroy ext4_es_cachep on module unload
ext4: make sure group number is bumped after a inode allocation race
Currently, we do not check the return value of client = rpc_clone_client(),
nor do we shut down the resulting cloned rpc_clnt in the case where a
NFS4ERR_WRONGSEC has caused nfs4_proc_lookup_common() to replace the
original value of 'client' (causing a memory leak).
Fix both issues and simplify the code by moving the call to
rpc_clone_client() until after nfs4_proc_lookup_common() has
done its business.
Reported-by: Andy Adamson <andros@netapp.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
We only need to call it on the creation of the inode.
Reported-by: Julia Lawall <Julia.Lawall@lip6.fr>
Cc: Steve Dickson <SteveD@redhat.com>
Cc: Dave Quigley <dpquigl@davequigley.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The sync mount option stopped working for NFSv4 mounts after commit
c02d7adf8c (NFSv4: Replace nfs4_path_walk() with
FS path lookup in a private namespace). If MS_SYNCHRONOUS is set in the
super_block that we're cloning from, then it should be set in the new
super_block as well.
Signed-off-by: Scott Mayhew <smayhew@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
If a cache invalidation is triggered, and we happen to have a lot of
writebacks cached at the time, then the call to invalidate_inode_pages2()
will end up calling ->launder_page() on each and every dirty page in order
to sync its contents to disk, thus defeating write coalescing.
The following patch ensures that we try to sync the inode to disk before
calling invalidate_inode_pages2() so that we do the writeback as efficiently
as possible.
Reported-by: William Dauchy <william@gandi.net>
Reported-by: Pascal Bouchareine <pascal@gandi.net>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Tested-by: William Dauchy <william@gandi.net>
Reviewed-by: Jeff Layton <jlayton@redhat.com>
lead to race conditions between deleting an event and accessing an event
debugfs file. This included a fix to the debugfs system (acked by
Greg Kroah-Hartman). We think that all the holes have been patched and
hopefully we don't find more. I haven't marked all of them for stable
because I need to examine them more to figure out how far back some of
the changes need to go.
Along the way, some other fixes have been made. Alexander Z Lam fixed
some logic where the wrong buffer was being modifed.
Andrew Vagin found a possible corruption for machines that actually
allocate cpumask, as a reference to one was being zeroed out by mistake.
Dhaval Giani found a bad prototype when tracing is not configured.
And I not only had some changes to help Oleg, but also finally fixed
a long standing bug that Dave Jones and others have been hitting, where
a module unload and reload can cause the function tracing accounting
to get screwed up.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.14 (GNU/Linux)
iQEcBAABAgAGBQJR/7FvAAoJEOdOSU1xswtMCAgH/RpxS5mQcQnhrGfu1qnSk+3v
CyL1X9IkvgP5f+N59tbU3ZuE8Q3YQvTII35na38fgbHbwWrhYjLSvlf3Pwtre8zr
Rjm9b8zA6iSobHu3DyuuupzMsLE+SpBakVSmG6mi8izZyuCi0YHMZMnHTGNnW9Vv
YFqkfGgQQWMqiUHLcueetOqWAjR2JiJaLfr47rsRLNflF6dB2MbG/7YbYJ0TBk7E
hemVmR7JH7606eJZ6nR7bh/hYv0sL2SSqVVaOHO5nWFLFbI8CybpNVGcoD+nihZY
M7LrZT1C1mn3nKrfDhyXy4dwwx6wBON0fY23eJTHMVNnc0tvY1xqH52vTdPILWg=
=UfsE
-----END PGP SIGNATURE-----
Merge tag 'trace-fixes-3.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace
Pull tracing fixes from Steven Rostedt:
"Oleg Nesterov has been working hard in closing all the holes that can
lead to race conditions between deleting an event and accessing an
event debugfs file. This included a fix to the debugfs system (acked
by Greg Kroah-Hartman). We think that all the holes have been patched
and hopefully we don't find more. I haven't marked all of them for
stable because I need to examine them more to figure out how far back
some of the changes need to go.
Along the way, some other fixes have been made. Alexander Z Lam fixed
some logic where the wrong buffer was being modifed.
Andrew Vagin found a possible corruption for machines that actually
allocate cpumask, as a reference to one was being zeroed out by
mistake.
Dhaval Giani found a bad prototype when tracing is not configured.
And I not only had some changes to help Oleg, but also finally fixed a
long standing bug that Dave Jones and others have been hitting, where
a module unload and reload can cause the function tracing accounting
to get screwed up"
* tag 'trace-fixes-3.11-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace:
tracing: Fix reset of time stamps during trace_clock changes
tracing: Make TRACE_ITER_STOP_ON_FREE stop the correct buffer
tracing: Fix trace_dump_stack() proto when CONFIG_TRACING is not set
tracing: Fix fields of struct trace_iterator that are zeroed by mistake
tracing/uprobes: Fail to unregister if probe event files are in use
tracing/kprobes: Fail to unregister if probe event files are in use
tracing: Add comment to describe special break case in probe_remove_event_call()
tracing: trace_remove_event_call() should fail if call/file is in use
debugfs: debugfs_remove_recursive() must not rely on list_empty(d_subdirs)
ftrace: Check module functions being traced on reload
ftrace: Consolidate some duplicate code for updating ftrace ops
tracing: Change remove_event_file_dir() to clear "d_subdirs"->i_private
tracing: Introduce remove_event_file_dir()
tracing: Change f_start() to take event_mutex and verify i_private != NULL
tracing: Change event_filter_read/write to verify i_private != NULL
tracing: Change event_enable/disable_read() to verify i_private != NULL
tracing: Turn event/id->i_private into call->event.type
- don't BUG_ON() when not SP4_NONE
- calculate recv and send reserve sizes correctly
Signed-off-by: Weston Andros Adamson <dros@netapp.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
David reported that commit c2b93e06 (cifs: only set ops for inodes in
I_NEW state) caused a regression with mfsymlinks. Prior to that patch,
if a mfsymlink dentry was instantiated at readdir time, the inode would
get a new set of ops when it was revalidated. After that patch, this
did not occur.
This patch addresses this by simply skipping instantiating dentries in
the readdir codepath when we know that they will need to be immediately
revalidated. The next attempt to use that dentry will cause a new lookup
to occur (which is basically what we want to happen anyway).
Cc: <stable@vger.kernel.org>
Cc: "Stefan (metze) Metzmacher" <metze@samba.org>
Cc: Sachin Prabhu <sprabhu@redhat.com>
Reported-and-Tested-by: David McBride <dwm37@cam.ac.uk>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
Firstly, nlmclnt_setlockargs can be called from a reclaimer thread, in
which case we're in entirely the wrong namespace.
Secondly, commit 8aac62706a (move
exit_task_namespaces() outside of exit_notify()) now means that
exit_task_work() is called after exit_task_namespaces(), which
triggers an Oops when we're freeing up the locks.
Fix this by ensuring that we initialise the nlm_host's rpc_client at mount
time, so that the cl_nodename field is initialised to the value of
utsname()->nodename that the net namespace uses. Then replace the
lockd callers of utsname()->nodename.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: Toralf Förster <toralf.foerster@gmx.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Nix <nix@esperi.org.uk>
Cc: Jeff Layton <jlayton@redhat.com>
Cc: stable@vger.kernel.org # 3.10.x
As comment in include/uapi/asm-generic/fcntl.h described, when
introducing new O_* bits, we need to check its uniqueness in
fcntl_init(). But __O_TMPFILE bit is missing. So fix it.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Every now and then someone proposes a new flink syscall, and this spawns
a long discussion of whether it would be a security problem. I think
that this is missing the point: flink is *already* allowed without
privilege as long as /proc is mounted -- it's called AT_SYMLINK_FOLLOW.
Now that O_TMPFILE is here, the ability to create a file with O_TMPFILE,
write it, and link it in is very convenient. The only problem is that
it requires that /proc be mounted so that you can do:
linkat(AT_FDCWD, "/proc/self/fd/<tmpfd>", dfd, path, AT_SYMLINK_NOFOLLOW)
This sucks -- it's much nicer to do:
linkat(tmpfd, "", dfd, path, AT_EMPTY_PATH)
Let's allow it.
If this turns out to be excessively scary, it we could instead require
that the inode in question be I_LINKABLE, but this seems pointless given
the /proc situation
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
O_TMPFILE, like O_CREAT, should respect the requested mode and should
create regular files.
This fixes two bugs: O_TMPFILE required privilege (because the mode
ended up as 000) and it produced bogus inodes with no type.
Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Since remove_proc_entry() started to wait for IO in progress (i.e.
since 2007 or so), the locking in fs/reiserfs/proc.c became wrong;
if procfs read happens between the moment when umount() locks the
victim superblock and removal of /proc/fs/reiserfs/<device>/*,
we'll get a deadlock - read will wait for s_umount (in sget(),
called by r_start()), while umount will wait in remove_proc_entry()
for that read to finish, holding s_umount all along.
Fortunately, the same change allows a much simpler race avoidance -
all we need to do is remove the procfs entries in the very beginning
of reiserfs ->kill_sb(); that'll guarantee that pointer to superblock
will remain valid for the duration for procfs IO, so we don't need
sget() to keep the sucker alive. As the matter of fact, we can
get rid of the home-grown iterator completely, and use single_open()
instead.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Add the missing NULL check of the return value of find_or_create_page() in
function ocfs2_duplicate_clusters_by_page().
[akpm@linux-foundation.org: fix layout, per Joel]
Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com>
Acked-by: Joel Becker <jlbec@evilplan.org>
Cc: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently, the s_root dentry doesn't get its d_op pointer set to
anything. This breaks lookups in the root of case-insensitive mounts
since that relies on having d_hash and d_compare routines that know to
treat the filename as case-insensitive.
cifs.ko has been broken this way for a long time, but commit 1c929cfe6
("switch cifs"), added a cryptic comment which is removed in the patch
below, which makes me wonder if this was done deliberately for some
reason. It's not clear to me why we'd want the s_root not to have d_op
set properly.
It may have something to do with d_automount or d_revalidate on the
root, but my suspicion in looking over the code is that Al was just
trying to preserve the existing behavior when changing this code over to
use s_d_op.
This patch changes it so that we set s_d_op before calling d_make_root
and removes the comment. I tested mounting, accessing and unmounting
several types of shares (including DFS referrals) and everything still
seemed to work OK afterward. I could be missing something however, so
please do let me know if I am.
Reported-by: Jan-Marek Glogowski <glogow@fbihome.de>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Ian Kent <raven@themaw.net>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
debugfs_remove_recursive() is wrong,
1. it wrongly assumes that !list_empty(d_subdirs) means that this
dir should be removed.
This is not that bad by itself, but:
2. if d_subdirs does not becomes empty after __debugfs_remove()
it gives up and silently fails, it doesn't even try to remove
other entries.
However ->d_subdirs can be non-empty because it still has the
already deleted !debugfs_positive() entries.
3. simple_release_fs() is called even if __debugfs_remove() fails.
Suppose we have
dir1/
dir2/
file2
file1
and someone opens dir1/dir2/file2.
Now, debugfs_remove_recursive(dir1/dir2) succeeds, and dir1/dir2 goes
away.
But debugfs_remove_recursive(dir1) silently fails and doesn't remove
this directory. Because it tries to delete (the already deleted)
dir1/dir2/file2 again and then fails due to "Avoid infinite loop"
logic.
Test-case:
#!/bin/sh
cd /sys/kernel/debug/tracing
echo 'p:probe/sigprocmask sigprocmask' >> kprobe_events
sleep 1000 < events/probe/sigprocmask/id &
echo -n >| kprobe_events
[ -d events/probe ] && echo "ERR!! failed to rm probe"
And after that it is not possible to create another probe entry.
With this patch debugfs_remove_recursive() skips !debugfs_positive()
files although this is not strictly needed. The most important change
is that it does not try to make ->d_subdirs empty, it simply scans
the whole list(s) recursively and removes as much as possible.
Link: http://lkml.kernel.org/r/20130726151256.GC19472@redhat.com
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
In the cifs_reopen_file function, if the following statement is
asserted:
(tcon->unix_ext && cap_unix(tcon->ses) &&
(CIFS_UNIX_POSIX_PATH_OPS_CAP &
(tcon->fsUnixInfo.Capability)))
and we succeed to open with cifs_posix_open, the function jumps
to the label reopen_success and checks for oparms.reconnect
which is not initialized.
This issue has been reported by scan.coverity.com
Signed-off-by: Andi Shyti <andi@etezian.org>
Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
When use of symlinks is enabled (mounting with mfsymlinks option) to
non-Samba servers, we always tried to use cifs, even when we
were mounted with SMB2 or SMB3, which causes the server to drop the
network connection.
This patch separates out the protocol specific operations for cifs from
the code which recognizes symlinks, and fixes the problem where
with SMB2 mounts we attempt cifs operations to open and read
symlinks. The next patch will add support for SMB2 for opening
and reading symlinks. Additional followon patches will address
the similar problem creating symlinks.
Signed-off-by: Steve French <smfrench@gmail.com>
For cifs_set_cifscreds() in "fs/cifs/connect.c", 'desc' buffer length
is 'CIFSCREDS_DESC_SIZE' (56 is less than 256), and 'ses->domainName'
length may be "255 + '\0'".
The related sprintf() may cause memory overflow, so need extend related
buffer enough to hold all things.
It is also necessary to be sure of 'ses->domainName' must be less than
256, and define the related macro instead of hard code number '256'.
Signed-off-by: Chen Gang <gang.chen@asianux.com>
Reviewed-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Shirish Pargaonkar <shirishpargaonkar@gmail.com>
Reviewed-by: Scott Lovenberg <scott.lovenberg@gmail.com>
CC: <stable@vger.kernel.org>
Signed-off-by: Steve French <smfrench@gmail.com>
In commit 921f266b: ext4: add self-testing infrastructure to do a
sanity check, some sanity checks were added in map_blocks to make sure
'retval == map->m_len'.
Enable these checks by default and report any assertion failures using
ext4_warning() and WARN_ON() since they can help us to figure out some
bugs that are otherwise hard to hit.
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
When we try to allocate an inode, and there is a race between two
CPU's trying to grab the same inode, _and_ this inode is the last free
inode in the block group, make sure the group number is bumped before
we continue searching the rest of the block groups. Otherwise, we end
up searching the current block group twice, and we end up skipping
searching the last block group. So in the unlikely situation where
almost all of the inodes are allocated, it's possible that we will
return ENOSPC even though there might be free inodes in that last
block group.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Cc: stable@vger.kernel.org
- fix for regression in commit cca9f93a52, recovery causing filesystem
corruption after a crash
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)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=hrYY
-----END PGP SIGNATURE-----
Merge tag 'for-linus-v3.11-rc3' of git://oss.sgi.com/xfs/xfs
Pull xfs fix from Ben Myers:
"Fix for regression in commit cca9f93a52 ("xfs: don't do IO when
creating an new inode"), recovery causing filesystem corruption after
a crash"
* tag 'for-linus-v3.11-rc3' of git://oss.sgi.com/xfs/xfs:
xfs: di_flushiter considered harmful
Pull nfsd fix from Bruce Fields:
"One more nfsd bugfix for 3.11"
* 'for-3.11' of git://linux-nfs.org/~bfields/linux:
nfsd: nfsd_open: when dentry_open returns an error do not propagate as struct file
When we made all inode updates transactional, we no longer needed
the log recovery detection for inodes being newer on disk than the
transaction being replayed - it was redundant as replay of the log
would always result in the latest version of the inode would be on
disk. It was redundant, but left in place because it wasn't
considered to be a problem.
However, with the new "don't read inodes on create" optimisation,
flushiter has come back to bite us. Essentially, the optimisation
made always initialises flushiter to zero in the create transaction,
and so if we then crash and run recovery and the inode already on
disk has a non-zero flushiter it will skip recovery of that inode.
As a result, log recovery does the wrong thing and we end up with a
corrupt filesystem.
Because we have to support old kernel to new kernel upgrades, we
can't just get rid of the flushiter support in log recovery as we
might be upgrading from a kernel that doesn't have fully transactional
inode updates. Unfortunately, for v4 superblocks there is no way to
guarantee that log recovery knows about this fact.
We cannot add a new inode format flag to say it's a "special inode
create" because it won't be understood by older kernels and so
recovery could do the wrong thing on downgrade. We cannot specially
detect the combination of zero mode/non-zero flushiter on disk to
non-zero mode, zero flushiter in the log item during recovery
because wrapping of the flushiter can result in false detection.
Hence that makes this "don't use flushiter" optimisation limited to
a disk format that guarantees that we don't need it. And that means
the only fix here is to limit the "no read IO on create"
optimisation to version 5 superblocks....
Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
(cherry picked from commit e60896d8f2)
The calculation of the attribute length was 4 bytes off.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Tested-by: Andre Heider <a.heider@gmail.com>
Reported-and-tested-by: Henrik Rydberg <rydberg@euromail.se>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Roland Dreier