We BUG if we fail to commit the transaction when creating a snapshot, which
is just obnoxious.  Remove the BUG_ON().  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

After cloning root's node, we forgot to dec the src's ref
which can lead to a memory leak.
Signed-off-by: NLiu Bo <bo.li.liu@oracle.com>
Signed-off-by: NChris Mason <chris.mason@fusionio.com>

On a really full file system I was getting ENOSPC back from
btrfs_update_inode when trying to update the parent inode when creating a
snapshot.  Just use the fallback method so we can update the inode and not
have to worry about having a delayed ref.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

This patch also requires a change in the user-space part of "receive".
We need to use "lchown" instead of "chown". We will do this in the
following patch.
Signed-off-by: NAlex Lyakas <alex.btrfs@zadarastorage.com>

 	if (S_ISREG(sctx->cur_inode_mode)) {

Steps to reproduce:
 # mkfs.btrfs -m raid1 <disk1> <disk2>
 # btrfstune -S 1 <disk1>
 # mount <disk1> <mnt>
 # btrfs device add <disk3> <disk4> <mnt>
 # mount -o remount,rw <mnt>
 # dd if=/dev/zero of=<mnt>/tmpfile bs=1M count=1
 Deadlock happened.

It is because of the nested chunk allocation. When we wrote the data
into the filesystem, we would allocate the data chunk because there was
no data chunk in the filesystem. At the end of the data chunk allocation,
we should insert the metadata of the data chunk into the extent tree, but
there was no raid1 chunk, so we tried to lock the chunk allocation mutex to
allocate the new chunk, but we had held the mutex, the deadlock happened.

By rights, we would allocate the raid1 chunk when we added the second device
because the profile of the seed filesystem is raid1 and we had two devices.
But we didn't do that in fact. It is because the last step of the first device
insertion didn't commit the transaction. So when we added the second device,
we didn't cow the tree, and just inserted the relative metadata into the leaves
which were generated by the first device insertion, and its profile was dup.

So, I fix this problem by commiting the transaction at the end of the first
device insertion.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>

Currently if len argument in btrfs_ioctl_fitrim() is smaller than
one FSB we will continue and finally return 0 bytes discarded.
However if the length to discard is smaller then file system block
we should really return EINVAL.
Signed-off-by: NLukas Czerner <lczerner@redhat.com>

We should free quota_root before returning from the error
handling code.
Signed-off-by: NTsutomu Itoh <t-itoh@jp.fujitsu.com>

When sending a device file, the stream was missing the mode. Also the
rdev was encoded wrongly.
Signed-off-by: NArne Jansen <sensille@gmx.net>

This adds support for the new extended inode refs to btrfs send.
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>

gcc says "warning: comparison of unsigned expression >= 0 is always
true" because i is an unsigned long. And gcc is right this time.
Signed-off-by: NStefan Behrens <sbehrens@giantdisaster.de>

To see the problem, create many hardlinks to the same file (120 should do it),
then look up paths by inode with:

  ls -i
  btrfs inspect inode-resolve -v $ino /mnt/btrfs

I noticed the memory layout of the fspath->val data had some irregularities
(some unnecessary gaps that stop appearing about halfway),
so I'm not sure there aren't any bugs left in it.

Emphasis the way tree_mod_log_insert_move avoids adding
MOD_LOG_KEY_REMOVE_WHILE_MOVING operations, depending on the direction of
the move operation.
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>

In get_old_root we grab a lock on the extent buffer before we obtain a
reference on that buffer. That order is changed now.
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>

In btrfs_find_all_roots' termination condition, we compare the level of the
old buffer we got from btrfs_search_old_slot to the level of the current
root node. We'd better compare it to the level of the rewinded root node.
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>

Tree mod log treated old root buffers as always empty buffers when starting
the rewind operations. However, the old root may still be part of the
current tree at a lower level, with still some valid entries.
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>

Avoid the implicit free by tree_mod_log_set_root_pointer, which is wrong in
two places. Where needed, we call tree_mod_log_free_eb explicitly now.
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>

Independant of the check (push_items < src_items) tree_mod_log_eb_copy did
log the removal of the old data entries from the source buffer. Therefore,
we must not call tree_mod_log_eb_move if the check evaluates to true, as
that would log the removal twice, finally resulting in (rewinded) buffers
with wrong values for header_nritems.
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>

Signed-off-by: NChris Mason <chris.mason@fusionio.com>

In csum_dirty_buffer, we first get eb from page->private.
Then we check if the page is the first page of eb. Later
we check it again. Remove the repeated check here.
Signed-off-by: NWang Sheng-Hui <shhuiw@gmail.com>

Alloc_dummy_extent_buffer will not free the first page in the eb array if we
fail to allocate a page, fix this.  Thanks,
Reported-by: NDavid Sterba <dave@jikos.cz>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

It's just annoying and the user will have gotten a nice OOM killer message
so they are already fully aware they are screwed :).  Thanks,
Reported-by: NJérôme Poulin <jeromepoulin@gmail.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

Get rid of the BUG_ON(ret == -ENOMEM) in __extent_read_full_page.  Thanks,
Reported-by: NJérôme Poulin <jeromepoulin@gmail.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

We were freeing non-existent pages which was causing a panic for a user who
was suffering from ENOMEM.  This patch fixes the problem.  Thanks,
Reported-by: NJérôme Poulin <jeromepoulin@gmail.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

So far the return code of barrier_all_devices() is ignored, which
means that errors are ignored. The result can be a corrupt
filesystem which is not consistent.
This commit adds code to evaluate the return code of
barrier_all_devices(). The normal btrfs_error() mechanism is used to
switch the filesystem into read-only mode when errors are detected.

In order to decide whether barrier_all_devices() should return
error or success, the number of disks that are allowed to fail the
barrier submission is calculated. This calculation accounts for the
worst RAID level of metadata, system and data. If single, dup or
RAID0 is in use, a single disk error is already considered to be
fatal. Otherwise a single disk error is tolerated.

The calculation of the number of disks that are tolerated to fail
the barrier operation is performed when the filesystem gets mounted,
when a balance operation is started and finished, and when devices
are added or removed.
Signed-off-by: NStefan Behrens <sbehrens@giantdisaster.de>

In check-integrity, detect when a superblock is written that points
to blocks that have not been written to disk due to I/O write errors.
Signed-off-by: NStefan Behrens <sbehrens@giantdisaster.de>

If a filesystem is mounted with compression and then remounted by adding nodatacow,
the compression is disabled but the compress flag is still visible.
Also, if a filesystem is mounted with nodatacow and then remounted with compression,
nodatacow flag is still present but it's not active.
This patch:
- removes compress flags and notifies that the compression has been disabled if the
  filesystem is mounted with nodatacow
- removes nodatacow and nodatasum flags if mounted with compress.
Signed-off-by: NAndrei Popa <andrei.popa@i-neo.ro>

Fix various messages to include newline and module prefix.
Signed-off-by: NDaniel J Blueman <daniel@quora.org>

We can just copy the in memory inode into the tree log directly, no sense in
updating the fs tree so we can copy it into the tree log tree.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

When building btrfs from kernel code, it will report:

	fs/btrfs/extent_io.h:281: warning: 'extent_buffer_page' declared inline after being called
	fs/btrfs/extent_io.h:281: warning: previous declaration of 'extent_buffer_page' was here
	fs/btrfs/extent_io.h:280: warning: 'num_extent_pages' declared inline after being called
	fs/btrfs/extent_io.h:280: warning: previous declaration of 'num_extent_pages' was here

because of the wrong declaration of inline functions.
Signed-off-by: NRobin Dong <sanbai@taobao.com>

Because the value of extent_map is only a correct value or NULL,
so IS_ERR is unnecessary.
Signed-off-by: NTsutomu Itoh <t-itoh@jp.fujitsu.com>

The function btrfs_insert_some_items() would not be called by any other functions,
so remove it.
Signed-off-by: NRobin Dong <sanbai@taobao.com>

I don't think we have the same problem that this was supposed to fix
originally since we can allocate chunks in the enospc path now.  This code
is causing us to constantly commit the transaction as we get close to using
all of our available space in our currently allocated chunks, instead of
allocating another chunk and carrying on with life, which is not nice for
performance.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

We should confirm the value of extent_map before calling
trace_btrfs_get_extent() because the value of extent_map has the
possibility of NULL.
Signed-off-by: NTsutomu Itoh <t-itoh@jp.fujitsu.com>

When we truncate existing items in the tree log we've been searching for
each individual item and removing them.  This is unnecessary churn and
searching, just keep track of the slot we are on and how many items we need
to delete and delete them all at once.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

The tree logging stuff was looking up csums to copy over for prealloc
extents which is just work we don't need to be doing.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

Everytime we write out dirty pages we search for an offset in the tree,
convert the bits in the state, and then when we wait we search for the
offset again and clear the bits.  So for every dirty range in the io tree we
are doing 4 rb searches, which is suboptimal.  With this patch we are only
doing 2 searches for every cycle (modulo weird things happening).  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

For some reason we unlock everything except the leaf we are on, set the path
blocking and then add the extent item for the extent we just finished
writing.  I can't for the life of me figure out why we would want to do
this, and the history doesn't really indicate that there was a real reason
for it, so just remove it.  This will reduce our tree lock contention on
heavy writes.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

There are a coule scenarios where farming metadata csumming off to an async
thread doesn't help.  The first is if our processor supports crc32c, in
which case the csumming will be fast and so the overhead of the async model
is not worth the cost.  The other case is for our tree log.  We will be
making that stuff dirty and writing it out and waiting for it immediately.
Even with software crc32c this gives me a ~15% increase in speed with O_SYNC
workloads.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

commit 7ca4be45 limited csum items to
PAGE_CACHE_SIZE.  It used min() with incompatible types in 32bit which
generates warnings:

fs/btrfs/file-item.c: In function ‘btrfs_csum_file_blocks’:
fs/btrfs/file-item.c:717: warning: comparison of distinct pointer types lacks a cast

This uses min_t(u32,) to fix the warnings.  u32 seemed reasonable
because btrfs_root->leafsize is u32 and PAGE_CACHE_SIZE is unsigned
long.
Signed-off-by: NZach Brown <zab@zabbo.net>