Replayed rename operations (after an mds failure/recovery) were broken
because the request paths were regenerated from the dentry names, which
get mangled when d_move() is called.

Instead, resend the previous request message when replaying completed
operations.  Just make sure the REPLAY flag is set and the target ino is
filled in.

This fixes problems with workloads doing renames when the MDS restarts,
where the rename operation appears to succeed, but on mds restart then
fails (leading to client confusion, app breakage, etc.).
Signed-off-by: Sage Weil <sage@newdream.net>

Use the address family from the peer address instead of assuming IPv4.
Signed-off-by: Sage Weil <sage@newdream.net>

Check for brackets around the ipv6 address to avoid ambiguity with the port
number.
Signed-off-by: Sage Weil <sage@newdream.net>

The buffer was too small.  Make it bigger, use snprintf(), put brackets
around the ipv6 address to avoid mixing it up with the :port, and use the
ever-so-handy %pI[46] formats.
Signed-off-by: Sage Weil <sage@newdream.net>

We leak a "pi" on this error path.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Sage Weil <sage@newdream.net>

Fix leak of a struct ceph_buffer on umount.
Signed-off-by: Sage Weil <sage@newdream.net>

A message can be on a queue (pending or sent), or out_msg (sending), or
both.  We were assuming that if it's not on a queue it couldn't be out_msg,
but that was false in the case of lossy connections like the OSD.  Fix
ceph_con_revoke() to treat these cases independently.  Also, fix the
out_kvec_is_message check to only trigger if we are currently sending
_this_ message.

This fixes a GPF in tcp_sendpage, triggered by OSD restarts.
Signed-off-by: Sage Weil <sage@newdream.net>

Fix a typo that made any OSD weighted between 0.1 and 1.0 effectively
weighted as 1.0 (fully in).
Signed-off-by: Sage Weil <sage@newdream.net>

We need to increase the total and used counters when allocating a new cap
in the non-reserved (cap import) case.
Signed-off-by: Sage Weil <sage@newdream.net>

We can drop caps with an mds request.  Ensure we only drop unused AND
clean caps, since the MDS doesn't support cap writeback in that context,
nor do we track it.  If caps are dirty, and the MDS needs them back, we
it will revoke and we will flush in the normal fashion.

This fixes a possibly loss of metadata.
Signed-off-by: Sage Weil <sage@newdream.net>

We may not recurse for CHOOSE_LEAF if we start with a leaf node.  When
that happens, the out2 vector needs to be filled in with the result.
Signed-off-by: Sage Weil <sage@newdream.net>

There was a longstanding problem with recursion through intervening
bucket types on complex hierarchies.
Signed-off-by: Sage Weil <sage@newdream.net>

The ceph client structure was not set correctly.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>

This fixes a race between handle_reply finishing an mds request, signalling
completion, and then dropping the request structing and its dentry+inode
refs, and pre_umount function waiting for requests to finish before
letting the vfs tear down the dcache.  If umount was delayed waiting for
mds requests, we could race and BUG in shrink_dcache_for_umount_subtree
because of a slow dput.

This delays umount until the msgr queue flushes, which means handle_reply
will exit and will have dropped the ceph_mds_request struct.  I'm assuming
the VFS has already ensured that its calls have all completed and those
request refs have thus been dropped as well (I haven't seen that race, at
least).
Signed-off-by: Sage Weil <sage@newdream.net>

Handle a splice_dentry failure (due to a d_materialize_unique error)
without crashing.  (Also, report the error code.)
Signed-off-by: Sage Weil <sage@newdream.net>

If the incremental osdmap has a new crush map, advance the position after
decoding so that we can parse the rest of the osdmap properly.
Signed-off-by: Sage Weil <sage@newdream.net>

We need to properly initialize skip, as not all alloc_msg op instances
set it.

Also, BUG if someone says skip but also allocates a message.
Signed-off-by: Sage Weil <sage@newdream.net>

Don't leak message if we receive an unexpected message type.
Signed-off-by: Sage Weil <sage@newdream.net>

Fix some problems that came up with sparse.
Signed-off-by: Yehuda Sadeh <yehuda@hq.newdream.net>
Signed-off-by: Sage Weil <sage@newdream.net>

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
  wimax/i2400m: fix missing endian correction read in fw loader
  net8139: fix a race at the end of NAPI
  pktgen: Fix accuracy of inter-packet delay.
  pkt_sched: gen_estimator: add a new lock
  net: deliver skbs on inactive slaves to exact matches
  ipv6: fix ICMP6_MIB_OUTERRORS
  r8169: fix mdio_read and update mdio_write according to hw specs
  gianfar: Revive the driver for eTSEC devices (disable timestamping)
  caif: fix a couple range checks
  phylib: Add support for the LXT973 phy.
  net: Print num_rx_queues imbalance warning only when there are allocated queues

* 'pm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/suspend-2.6:
  PM / x86: Save/restore MISC_ENABLE register

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable:
  Btrfs: The file argument for fsync() is never null
  Btrfs: handle ERR_PTR from posix_acl_from_xattr()
  Btrfs: avoid BUG when dropping root and reference in same transaction
  Btrfs: prohibit a operation of changing acl's mask when noacl mount option used
  Btrfs: should add a permission check for setfacl
  Btrfs: btrfs_lookup_dir_item() can return ERR_PTR
  Btrfs: btrfs_read_fs_root_no_name() returns ERR_PTRs
  Btrfs: unwind after btrfs_start_transaction() errors
  Btrfs: btrfs_iget() returns ERR_PTR
  Btrfs: handle kzalloc() failure in open_ctree()
  Btrfs: handle error returns from btrfs_lookup_dir_item()
  Btrfs: Fix BUG_ON for fs converted from extN
  Btrfs: Fix null dereference in relocation.c
  Btrfs: fix remap_file_pages error
  Btrfs: uninitialized data is check_path_shared()
  Btrfs: fix fallocate regression
  Btrfs: fix loop device on top of btrfs

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6:
  PCI: clear bridge resource range if BIOS assigned bad one
  PCI: hotplug/cpqphp, fix NULL dereference
  Revert "PCI: create function symlinks in /sys/bus/pci/slots/N/"
  PCI: change resource collision messages from KERN_ERR to KERN_INFO

Yannick found that video does not work with 2.6.34.  The cause of this
bug was that the BIOS had assigned the wrong range to the PCI bridge
above the video device.  Before 2.6.34 the kernel would have shrunk
the size of the bridge window, but since
  d65245c3 PCI: don't shrink bridge resources
the kernel will avoid shrinking BIOS ranges.

So zero out the old range if we fail to claim it at boot time; this will
cause us to allocate a new range at startup, restoring the 2.6.34
behavior.

Fixes regression https://bugzilla.kernel.org/show_bug.cgi?id=16009.
Reported-by: Yannick <yannick.roehlly@free.fr>
Acked-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>

There are devices out there which are PCI Hot-plug controllers with
compaq PCI IDs, but are not bridges, hence have pdev->subordinate
NULL. But cpqphp expects the pointer to be non-NULL.

Add a check to the probe function to avoid oopses like:
BUG: unable to handle kernel NULL pointer dereference at 00000050
IP: [<f82e3c41>] cpqhpc_probe+0x951/0x1120 [cpqphp]
*pdpt = 0000000033779001 *pde = 0000000000000000
...

The device here was:
00:0b.0 PCI Hot-plug controller [0804]: Compaq Computer Corporation PCI Hotplug Controller [0e11:a0f7] (rev 11)
	Subsystem: Compaq Computer Corporation Device [0e11:a2f8]
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>

This reverts commit 75568f80.

Since they're just a convenience anyway, remove these symlinks since
they're causing duplicate filename errors in the wild.
Acked-by: Alex Chiang <achiang@canonical.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>

We can often deal with PCI resource issues by moving devices around.  In
that case, there's no point in alarming the user with messages like these.
There are many bug reports where the message itself is the only problem,
e.g., https://bugs.launchpad.net/ubuntu/+source/linux/+bug/413419 .
Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>

The "file" argument for fsync is never null so we can remove this check.

What drew my attention here is that 7ea80859: "drop unused dentry
argument to ->fsync" introduced an unconditional dereference at the
start of the function and that generated a smatch warning.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

posix_acl_from_xattr() returns both ERR_PTRs and null, but it's OK to
pass null values to set_cached_acl()
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

If btrfs_ioctl_snap_destroy() deletes a snapshot but finishes
with end_transaction(), the cleaner kthread may come in and
drop the root in the same transaction.  If that's the case, the
root's refs still == 1 in the tree when btrfs_del_root() deletes
the item, because commit_fs_roots() hasn't updated it yet (that
happens during the commit).

This wasn't a problem before only because
btrfs_ioctl_snap_destroy() would commit the transaction before dropping
the dentry reference, so the dead root wouldn't get queued up until
after the fs root item was updated in the btree.

Since it is not an error to drop the root reference and the root in the
same transaction, just drop the BUG_ON() in btrfs_del_root().
Signed-off-by: Sage Weil <sage@newdream.net>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

when used Posix File System Test Suite(pjd-fstest) to test btrfs,
some cases about setfacl failed when noacl mount option used.
I simplified used commands in pjd-fstest, and the following steps
can reproduce it.
------------------------
# cd btrfs-part/
# mkdir aaa
# setfacl -m m::rw aaa    <- successed, but not expected by pjd-fstest.
------------------------
I checked ext3, a warning message occured, like as:
  setfacl: aaa/: Operation not supported
Certainly, it's expected by pjd-fstest.

So, i compared acl.c of btrfs and ext3. Based on that, a patch created.
Fortunately, it works.
Signed-off-by: Shi Weihua <shiwh@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

On btrfs, do the following
------------------
# su user1
# cd btrfs-part/
# touch aaa
# getfacl aaa
  # file: aaa
  # owner: user1
  # group: user1
  user::rw-
  group::rw-
  other::r--
# su user2
# cd btrfs-part/
# setfacl -m u::rwx aaa
# getfacl aaa
  # file: aaa
  # owner: user1
  # group: user1
  user::rwx           <- successed to setfacl
  group::rw-
  other::r--
------------------
but we should prohibit it that user2 changing user1's acl.
In fact, on ext3 and other fs, a message occurs:
  setfacl: aaa: Operation not permitted

This patch fixed it.
Signed-off-by: Shi Weihua <shiwh@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

btrfs_lookup_dir_item() can return either ERR_PTRs or null.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

btrfs_read_fs_root_no_name() returns ERR_PTRs on error so I added a
check for that.  It's not clear to me if it can also return NULL
pointers or not so I left the original NULL pointer check as is.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

This was added by a22285a6: "Btrfs: Integrate metadata reservation
with start_transaction".  If we goto out here then we skip all the
unwinding and there are locks still held etc.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

btrfs_iget() returns an ERR_PTR() on failure and not null.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

Unwind and return -ENOMEM if the allocation fails here.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

If btrfs_lookup_dir_item() fails, we should can just let the mount fail
with an error.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>

Tree blocks can live in data block groups in FS converted from extN.
So it's easy to trigger the BUG_ON.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>