Use a new per-sb alloc_mutex instead of abusing i_mutex of the alloc_file
to protect block allocations. This gets rid of lockdep nesting warnings
and prepares for extending the scope of alloc_mutex.
Signed-off-by: Christoph Hellwig <hch@tuxera.com>
Use i_mutex for protecting against concurrent setflags ioctls like in
other filesystems and get rid of the BKL in hfsplus_ioctl.
Signed-off-by: Christoph Hellwig <hch@tuxera.com>
Currenly the HFSPLUS_IOC_EXT2_GETFLAGS case never unlocks the BKL, which
can lead to easily reproduced lockups when doing multiple GETFLAGS ioctls.
Fix this by only taking the BKL for the HFSPLUS_IOC_EXT2_SETFLAGS case
as neither HFSPLUS_IOC_EXT2_GETFLAGS not the default error case needs it.
Signed-off-by: Christoph Hellwig <hch@tuxera.com>
This patch fixes a GFS2 problem whereby the first rename after a
mount can result in a file system consistency error being flagged
improperly and cause the file system to withdraw. The problem is
that the rename code tries to run the rgrp list with function
gfs2_blk2rgrpd before the rgrp list is guaranteed to be read in
from disk. The patch makes the rename function hold the rindex
glock (as the gfs2_unlink code does today) which reads in the rgrp
list if need be. There were a total of three places in the rename
code that improperly referenced the rgrp list without the rindex
glock and this patch fixes all three.
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
ocfs2 fast symlinks are NUL terminated strings stored inline in the
inode data area. However, disk corruption or a local attacker could, in
theory, remove that NUL. Because we're using strlen() (my fault,
introduced in a731d1 when removing vfs_follow_link()), we could walk off
the end of that string.
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Cc: stable@kernel.org
Testing on very recent kernel (2.6.36-rc6) made this warning pop:
WARNING: at fs/fs-writeback.c:87 inode_to_bdi+0x65/0x70()
Hardware name:
Dirtiable inode bdi default != sb bdi cifs
...the following patch fixes it and seems to be the obviously correct
thing to do for cifs.
Cc: stable@kernel.org
Acked-by: Dave Kleikamp <shaggy@linux.vnet.ibm.com>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
Recently a feature was added to GFS2 to allow journal id allocation
via sysfs. This patch builds upon that so that a negative journal id
will be treated as an error code to be passed back as the return code
from mount. This allows termination of the mount process if there is
a failure.
Also, the process has been updated so that the kernel will wait
for a journal id, even in the "spectator" case. This is required
in order to avoid mounting a filesystem in case there is an error
while joining the cluster. In the spectator case, 0 is written into
the file to indicate that all is well, and that mount should continue.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
XFS supports the "norecovery" mount option which is basically the
same as the GFS2 spectator mode. This adds support for "norecovery"
as a synonym for spectator mode, which is hopefully a more obvious
description of what it actually does.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
The tests further down the recovery function relating to
unlocking the journal need to be updated to match the
intial test. Also, a test in the umount code which was
surplus to requirements has been removed. Umounting
spectator mounts now works correctly, as expected.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
I have been seeing occasional pauses in transaction throughput up to
30s long under heavy parallel workloads. The only notable thing was
that the xfsaild was trying to be active during the pauses, but
making no progress. It was running exactly 20 times a second (on the
50ms no-progress backoff), and the number of pushbuf events was
constant across this time as well. IOWs, the xfsaild appeared to be
stuck on buffers that it could not push out.
Further investigation indicated that it was trying to push out inode
buffers that were pinned and/or locked. The xfsbufd was also getting
woken at the same frequency (by the xfsaild, no doubt) to push out
delayed write buffers. The xfsbufd was not making any progress
because all the buffers in the delwri queue were pinned. This scan-
and-make-no-progress dance went one in the trace for some seconds,
before the xfssyncd came along an issued a log force, and then
things started going again.
However, I noticed something strange about the log force - there
were way too many IO's issued. 516 log buffers were written, to be
exact. That added up to 129MB of log IO, which got me very
interested because it's almost exactly 25% of the size of the log.
He delayed logging code is suppose to aggregate the minimum of 25%
of the log or 8MB worth of changes before flushing. That's what
really puzzled me - why did a log force write 129MB instead of only
8MB?
Essentially what has happened is that no CIL pushes had occurred
since the previous tail push which cleared out 25% of the log space.
That caused all the new transactions to block because there wasn't
log space for them, but they kick the xfsaild to push the tail.
However, the xfsaild was not making progress because there were
buffers it could not lock and flush, and the xfsbufd could not flush
them because they were pinned. As a result, both the xfsaild and the
xfsbufd could not move the tail of the log forward without the CIL
first committing.
The cause of the problem was that the background CIL push, which
should happen when 8MB of aggregated changes have been committed, is
being held off by the concurrent transaction commit load. The
background push does a down_write_trylock() which will fail if there
is a concurrent transaction commit holding the push lock in read
mode. With 8 CPUs all doing transactions as fast as they can, there
was enough concurrent transaction commits to hold off the background
push until tail-pushing could no longer free log space, and the halt
would occur.
It should be noted that there is no reason why it would halt at 25%
of log space used by a single CIL checkpoint. This bug could
definitely violate the "no transaction should be larger than half
the log" requirement and hence result in corruption if the system
crashed under heavy load. This sort of bug is exactly the reason why
delayed logging was tagged as experimental....
The fix is to start blocking background pushes once the threshold
has been exceeded. Rework the threshold calculations to keep the
amount of log space a CIL checkpoint can use to below that of the
AIL push threshold to avoid the problem completely.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Alex Elder <aelder@sgi.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
This shouldn't really be required, but gcc can't tell that
"al" is only accessed when initialised.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Some of the functions in GFS2 were not reserving space in the transaction for
the resource group header and the resource groups bitblocks that get added
when you do allocation. GFS2 now makes sure to reserve space for the
resource group header and either all the bitblocks in the resource group, or
one for each block that it may allocate, whichever is smaller using the new
gfs2_rg_blocks() inline function.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
When checking journals for spectator mounts, we cannot rely on the
journal being locked, whatever its jid might be. This patch
ensures that we always get the journal locks when checking
journals for a spectator mount.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jlbec/ocfs2:
o2dlm: force free mles during dlm exit
ocfs2: Sync inode flags with ext2.
ocfs2: Move 'wanted' into parens of ocfs2_resmap_resv_bits.
ocfs2: Use cpu_to_le16 for e_leaf_clusters in ocfs2_bg_discontig_add_extent.
ocfs2: update ctime when changing the file's permission by setfacl
ocfs2/net: fix uninitialized ret in o2net_send_message_vec()
Ocfs2: Handle empty list in lockres_seq_start() for dlmdebug.c
Ocfs2: Re-access the journal after ocfs2_insert_extent() in dxdir codes.
ocfs2: Fix lockdep warning in reflink.
ocfs2/lockdep: Move ip_xattr_sem out of ocfs2_xattr_get_nolock.
This option has never done anything useful. Also at the same time
this cleans up the sb checks which are done at mount time. The
debug option will be accepted, but ignored in future. Since it
didn't do anything, there didn't seem much point in retaining it.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
While umounting, a block mle doesn't get freed if dlm is shutdown after
master request is received but before assert master. This results in unclean
shutdown of dlm domain.
This patch frees all mles that lie around after other nodes were notified about
exiting the dlm and marking dlm state as leaving. Only block mles are expected
to be around, so we log ERROR for other mles but still free them.
Signed-off-by: Srinivas Eeda <srinivas.eeda@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
We sync our inode flags with ext2 and define them by hex
values. But actually in commit 3669567(4 years ago), all
these values are moved to include/linux/fs.h. So we'd
better also use them as what ext2 did. So sync our inode
flags with ext2 by using FS_*.
Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
The first time I read the function ocfs2_resmap_resv_bits, I consider
about what 'wanted' will be used and consider about the comments.
Then I find it is only used if the reservation is empty. ;)
So we'd better move it to the parens so that it make the code more
readable, what's more, ocfs2_resmap_resv_bits is used so frequently
and we should save some cpus.
Acked-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
e_leaf_clusters is a le16, so use cpu_to_le16 instead
of cpu_to_le32.
What's more, we change 'clusters' to unsigned int to
signify that the size of 'clusters' isn't important here.
Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
In commit 30e2bab, ext3 fixed it. So change it accordingly in ocfs2.
Steps to reproduce:
# touch aaa
# stat -c %Z aaa
1283760364
# setfacl -m 'u::x,g::x,o::x' aaa
# stat -c %Z aaa
1283760364
Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
This option defaulted to on for lock_nolock mounts and off
otherwise. The only function was to avoid the revalidation of
dentries. In the cluster case, that is entirely pointless and
liable to cause coherency problems.
The patch changes the revalidation to depend upon whether the
fs is a local or cluster fs (i.e. it follows the existing default
behaviour). I very much doubt anybody ever used this option as
there is no reason to. Even so we will continue to accept it
on the mount command line, but ignore it.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
This is been a no-op for a very long time now. I'm pretty sure
nobody uses it, but just in case we'll still accept it on the
command line, but ignore it.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
OCFS2 can return ERESTARTSYS from its write function when the process is
signalled while waiting for a cluster lock (and the filesystem is mounted
with intr mount option). Generally, it seems reasonable to allow
filesystems to return this error code from its IO functions. As we must
not leak ERESTARTSYS (and similar error codes) to userspace as a result of
an AIO operation, we have to properly convert it to EINTR inside AIO code
(restarting the syscall isn't really an option because other AIO could
have been already submitted by the same io_submit syscall).
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Zach Brown <zach.brown@oracle.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 73296bc611 ("procfs: Use generic_file_llseek in /proc/vmcore")
broke seeking on /proc/vmcore. This changes it back to use default_llseek
in order to restore the original behaviour.
The problem with generic_file_llseek is that it only allows seeks up to
inode->i_sb->s_maxbytes, which is zero on procfs and some other virtual
file systems. We should merge generic_file_llseek and default_llseek some
day and clean this up in a proper way, but for 2.6.35/36, reverting vmcore
is the safer solution.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Reported-by: CAI Qian <caiqian@redhat.com>
Tested-by: CAI Qian <caiqian@redhat.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In 32-bit compatibility mode, the error handling for
compat_do_readv_writev() may free an uninitialized pointer, potentially
leading to all sorts of ugly memory corruption. This is reliably
triggerable by unprivileged users by invoking the readv()/writev()
syscalls with an invalid iovec pointer. The below patch fixes this to
emulate the non-compat version.
Introduced by commit b83733639a ("compat: factor out
compat_rw_copy_check_uvector from compat_do_readv_writev")
Signed-off-by: Dan Rosenberg <dan.j.rosenberg@gmail.com>
Cc: stable@kernel.org (2.6.35)
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'for-linus' of git://git.kernel.dk/linux-2.6-block:
bdi: Fix warnings in __mark_inode_dirty for /dev/zero and friends
char: Mark /dev/zero and /dev/kmem as not capable of writeback
bdi: Initialize noop_backing_dev_info properly
cfq-iosched: fix a kernel OOPs when usb key is inserted
block: fix blk_rq_map_kern bio direction flag
cciss: freeing uninitialized data on error path
Inodes of devices such as /dev/zero can get dirty for example via
utime(2) syscall or due to atime update. Backing device of such inodes
(zero_bdi, etc.) is however unable to handle dirty inodes and thus
__mark_inode_dirty complains. In fact, inode should be rather dirtied
against backing device of the filesystem holding it. This is generally a
good rule except for filesystems such as 'bdev' or 'mtd_inodefs'. Inodes
in these pseudofilesystems are referenced from ordinary filesystem
inodes and carry mapping with real data of the device. Thus for these
inodes we have to use inode->i_mapping->backing_dev_info as we did so
far. We distinguish these filesystems by checking whether sb->s_bdi
points to a non-trivial backing device or not.
Example: Assume we have an ext3 filesystem on /dev/sda1 mounted on /.
There's a device inode A described by a path "/dev/sdb" on this
filesystem. This inode will be dirtied against backing device "8:0"
after this patch. bdev filesystem contains block device inode B coupled
with our inode A. When someone modifies a page of /dev/sdb, it's B that
gets dirtied and the dirtying happens against the backing device "8:16".
Thus both inodes get filed to a correct bdi list.
Cc: stable@kernel.org
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
These devices don't do any writeback but their device inodes still can get
dirty so mark bdi appropriately so that bdi code does the right thing and files
inodes to lists of bdi carrying the device inodes.
Cc: stable@kernel.org
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
ceph: select CRYPTO
ceph: check mapping to determine if FILE_CACHE cap is used
ceph: only send one flushsnap per cap_snap per mds session
ceph: fix cap_snap and realm split
ceph: stop sending FLUSHSNAPs when we hit a dirty capsnap
ceph: correctly set 'follows' in flushsnap messages
ceph: fix dn offset during readdir_prepopulate
ceph: fix file offset wrapping at 4GB on 32-bit archs
ceph: fix reconnect encoding for old servers
ceph: fix pagelist kunmap tail
ceph: fix null pointer deref on anon root dentry release
Rather than calculating the qstrs for . and .. each time
we need them, its better to keep a constant version of
these and just refer to them when required.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Reviewed-by: Christoph Hellwig <hch@infradead.org>
The recovery workqueue can be freezable since
we want it to finish what it is doing if the system is to
be frozen (although why you'd want to freeze a cluster node
is beyond me since it will result in it being ejected from
the cluster). It does still make sense for single node
GFS2 filesystems though.
The glock workqueue will benefit from being able to run more
work items concurrently. A test running postmark shows
improved performance and multi-threaded workloads are likely
to benefit even more. It needs to be high priority because
the latency directly affects the latency of filesystem glock
operations.
The delete workqueue is similar to the recovery workqueue in
that it must not get blocked by memory allocations, and may
run for a long time.
Potentially other GFS2 threads might also be converted to
workqueues, but I'll leave that for a later patch.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Acked-by: Tejun Heo <tj@kernel.org>
GFS2's idea of which return codes it needs to handle was based
upon those listed in dlm.h. Those didn't cover all the possible
codes and listed some which never happen. This updates GFS2 to
handle all the codes which can actually be returned from the
DLM under various circumstances.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Due to the design of the VFS, it is quite usual for operations on GFS2
to consist of a lookup (requiring a shared lock) followed by an
operation requiring an exclusive lock. If a remote node has cached an
exclusive lock, then it will receive two demote events in rapid succession
firstly for a shared lock and then to unlocked. The existing min hold time
code was triggering in this case, even if the node was otherwise idle
since the state change time was being updated by the initial demote.
This patch introduces logic to skip the min hold timer in the case that
a "double demote" of this kind has occurred. The min hold timer will
still be used in all other cases.
A new glock flag is introduced which is used to keep track of whether
there have been any newly queued holders since the last glock state
change. The min hold time is only applied if the flag is set.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Tested-by: Abhijith Das <adas@redhat.com>
This patch adds support for fallocate to gfs2. Since the gfs2 does not support
uninitialized data blocks, it must write out zeros to all the blocks. However,
since it does not need to lock any pages to read from, gfs2 can write out the
zero blocks much more efficiently. On a moderately full filesystem, fallocate
works around 5 times faster on average. The fallocate call also allows gfs2 to
add blocks to the file without changing the filesize, which will make it
possible for gfs2 to preallocate space for the rindex file, so that gfs2 can
grow a completely full filesystem.
Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
This adds a check to ensure that if we reach the block allocator
that we don't try and proceed if there is no alloc structure
hanging off the inode. This should only happen if there is a bug
in GFS2. The error return code is distinctive in order that it
will be easily spotted.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
With the update of the truncate code, ip->i_disksize and
inode->i_size are merely copies of each other. This means
we can remove ip->i_disksize and use inode->i_size exclusively
reducing the size of a GFS2 inode by 8 bytes.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
This updates GFS2's truncate code to use the new truncate
sequence correctly. This is a stepping stone to being
able to remove ip->i_disksize in favour of using i_size
everywhere now that the two sizes are always identical.
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Christoph Hellwig <hch@lst.de>
Coda's REQ_* defines were renamed to avoid clashes with the block layer
(commit 4aeefdc69f: "coda: fixup clash with block layer REQ_*
defines").
However one was missed and response messages are no longer matched with
requests and waiting threads are no longer woken up. This patch fixes
this.
Signed-off-by: Jan Harkes <jaharkes@cs.cmu.edu>
[ Also fixed up whitespace while at it -Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mmotm/fs/ocfs2/cluster/tcp.c: In function ‘o2net_send_message_vec’:
mmotm/fs/ocfs2/cluster/tcp.c:980:6: warning: ‘ret’ may be used uninitialized in this function
It seems a real bug introduced by commit 9af0b38ff3 (ocfs2/net:
Use wait_event() in o2net_send_message_vec()).
cc: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
See if the i_data mapping has any pages to determine if the FILE_CACHE
capability is currently in use, instead of assuming it is any time the
rdcache_gen value is set (i.e., issued -> used).
This allows the MDS RECALL_STATE process work for inodes that have cached
pages.
Signed-off-by: Sage Weil <sage@newdream.net>
Sending multiple flushsnap messages is problematic because we ignore
the response if the tid doesn't match, and the server may only respond to
each one once. It's also a waste.
So, skip cap_snaps that are already on the flushing list, unless the caller
tells us to resend (because we are reconnecting).
Signed-off-by: Sage Weil <sage@newdream.net>
Looks like this crept in, in a recent update.
Reported-by: Krzysztof Urbaniak <urban@bash.org.pl>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>
The cap_snap creation/queueing relies on both the current i_head_snapc
_and_ the i_snap_realm pointers being correct, so that the new cap_snap
can properly reference the old context and the new i_head_snapc can be
updated to reference the new snaprealm's context. To fix this, we:
- move inodes completely to the new (split) realm so that i_snap_realm
is correct, and
- generate the new snapc's _before_ queueing the cap_snaps in
ceph_update_snap_trace().
Signed-off-by: Sage Weil <sage@newdream.net>
