This creates the reference to a new snapshot in the same commit as the
snapshot itself.  This avoids the need for a second commit in order for a
snapshot to be persistent, and also avoids the problem of "leaking" a
new snapshot tree root if the host crashes before the second commit takes
place.

It is not at all clear to me why it wasn't always done this way.  If there
is still a reason for the two-stage {create,finish}_pending_snapshots()
approach I'm missing something!  :)

I've been running this for a couple weeks under pretty heavy usage (a few
snapshots per minute) without obvious problems.
Signed-off-by: NSage Weil <sage@newdream.net>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Everytime we start a new flushing thread, we init the waitqueue if there isn't a
flushing thread running.  The problem with this is we check
space_info->flushing, which we clear right before doing a wake_up on the
flushing waitqueue, which causes problems if we init the waitqueue in the middle
of clearing the flushing flagh and calling wake_up.  This is hard to hit, but
the code is wrong anyway, so init the flushing/allocating waitqueue when
creating the space info and let it be.  I haven't seen the panic since I've been
using this patch.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Pagecache pages should be allocated with __page_cache_alloc, so they
obey pagecache memory policies.

add_to_page_cache_lru is exported, so it should be used. Benefits over
using a private pagevec: neater code, 128 bytes fewer stack used, percpu
lru ordering is preserved, and finally don't need to flush pagevec
before returning so batching may be shared with other LRU insertions.

Signed-off-by: Nick Piggin <npiggin@suse.de>:
Signed-off-by: NChris Mason <chris.mason@oracle.com>

If the amount of free space left in a device is less than what we think should
be the minimum size, just ignore the minimum size and use the amount we have.  I
ran into this running tests on a 600mb volume, the chunk allocator wouldn't let
me allocate the last 52mb of the disk for data because we want to have at least
64mb chunks for data.  This patch fixes that problem.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

As Yan pointed out, theres not much reason for all this complicated math to
account for file extents being split up into max_extent chunks, since they are
likely to all end up in the same leaf anyway.  Since there isn't much reason to
use max_extent, just remove the option altogether so we have one less thing we
need to test.
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

We don't actually check the return value of btrfs_read_block_groups, so we can
possibly succeed to mount, but then fail to say read the superblock xattr for
selinux which will cause the vfs code to deactivate the super.

This is a problem because in find_free_extent we just assume that we
will find the right space_info for the allocation we want.  But if we
failed to read the block groups, we won't have setup any space_info's,
and we'll hit a NULL pointer deref in find_free_extent.

This patch fixes that problem by checking the return value of
btrfs_read_block_groups, and failing out properly.  I've also added a
check in find_free_extent so if for some reason we don't find an
appropriate space_info, we just return -ENOSPC.
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

btrfs_get_extent() never returns NULL, only a valid pointer or ERR_PTR()
Signed-off-by: NDan Carpenter <error27@gmail.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Return -ENOMEM if kmalloc() fails.
Signed-off-by: NDan Carpenter <error27@gmail.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The original code dereferenced range on the next line.
Signed-off-by: NDan Carpenter <error27@gmail.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

We can use this simple method to make source more readable.
Signed-off-by: NZhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

We need to check return value of btrfs_search_slot() in
btrfs_read_chunk_tree() and do corresponding error handing.
Signed-off-by: NZhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

We only need to call finish_wait() after wait loop.

By the way, this patch makes code of waiting loop similar to
example in wait.h(no functional change)
Signed-off-by: NZhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

btrfs_find_create_tree_block() may return NULL, so we must check the returned
value, or we will access a NULL pointer.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

fs/btrfs/ioctl.c: ctree.h is included more than once.
Signed-off-by: NAndrea Gelmini <andrea.gelmini@gelma.net>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

This is used by the inode lookup ioctl to follow all the backrefs up
to the subvol root.  But the search being done would sometimes land one
past the last item in the leaf instead of finding the backref.

This changes the search to look for the highest possible backref and hop
back one item.  It also fixes a leaked path on failure to find the root.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

When a root id of 0 is sent to the inode lookup ioctl, it will
use the root of the file we're ioctling and pass the root id
back to userland along with the results.

This allows userland to do searches based on that root later on.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The search ioctl was skipping large items entirely (ones that are too
big for the results buffer).  This changes things to at least copy
the item header so that we can send information about the item back to
userland.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The search ioctl was working well for finding tree roots, but using it for
generic searches requires a few changes to how the keys are advanced.
This treats the search control min fields for objectid, type and offset
more like a key, where we drop the offset to zero once we bump the type,
etc.

The downside of this is that we are changing the min_type and min_offset
fields during the search, and so the ioctl caller needs extra checks to make sure
the keys in the result are the ones it wanted.

This also changes key_in_sk to use btrfs_comp_cpu_keys, just to make
things more readable.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The space_info ioctl was using copy_to_user inside rcu_read_lock.  This
commit changes things to copy into a buffer first and then dump the
result down to userland.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Signed-off-by: NSage Weil <sage@newdream.net>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

key->type is u8, not u64.

fs/btrfs/ioctl.c: In function 'copy_to_sk':
fs/btrfs/ioctl.c:1024: warning: comparison is always true due to limited range of data type
Signed-off-by: NSage Weil <sage@newdream.net>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

GFP_FS must be masked out, NOFS can't be or'd in.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

After callling submit_bio, the bio can be freed at any time.  The
btrfs submission thread helper was checking the bio flags too late,
which might not give the correct answer.

When CONFIG_DEBUG_PAGE_ALLOC is turned on, it can lead to oopsen.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

We can use btrfs_stack_device_id() to get dev_item->devid
Signed-off-by: NXiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Use memparse() instead of its own private implementation.
Signed-off-by: NAkinobu Mita <akinobu.mita@gmail.com>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: linux-btrfs@vger.kernel.org
Signed-off-by: NChris Mason <chris.mason@oracle.com>

df is a very loaded question in btrfs.  This gives us a way to get the per-space
usage information so we can tell exactly what is in use where.  This will help
us figure out ENOSPC problems, and help users better understand where their disk
space is going.
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

This patch just goes through and fixes everybody that does

lock_extent()
blah
unlock_extent()

to use

lock_extent_bits()
blah
unlock_extent_cached()

and pass around a extent_state so we only have to do the searches once per
function.  This gives me about a 3 mb/s boots on my random write test.  I have
not converted some things, like the relocation and ioctl's, since they aren't
heavily used and the relocation stuff is in the middle of being re-written.  I
also changed the clear_extent_bit() to only unset the cached state if we are
clearing EXTENT_LOCKED and related stuff, so we can do things like this

lock_extent_bits()
clear delalloc bits
unlock_extent_cached()

without losing our cached state.  I tested this thoroughly and turned on
LEAK_DEBUG to make sure we weren't leaking extent states, everything worked out
fine.
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

When finishing io we run btrfs_dec_test_ordered_pending, and then immediately
run btrfs_lookup_ordered_extent, but btrfs_dec_test_ordered_pending does that
already, so we're searching twice when we don't have to.  This patch lets us
pass a btrfs_ordered_extent in to btrfs_dec_test_ordered_pending so if we do
complete io on that ordered extent we can just use the one we found then instead
of having to do another btrfs_lookup_ordered_extent.  This made my fio job with
the other patch go from 24 mb/s to 29 mb/s.
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

This patch makes us cache the extent state we find in find_delalloc_range since
we'll have to lock the extent later on in the function.  This will keep us from
re-searching for the rang when we try to lock the extent.
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The ordered tree used to need a mutex, but currently all we use it for is to
protect the rb_tree, and a spin_lock is just fine for that.  Using a spin_lock
instead makes dbench run a little faster, 58 mb/s instead of 51 mb/s, and have
less latency, 3445.138 ms instead of 3820.633 ms.
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The endio is done at reverse order of bio vectors.

That means for a sequential read, the page first submitted will finish
last in a bio. Considering we will do checksum (making cache hot) for
every page, this does introduce delay (and chance to squeeze cache used
soon) for pages submitted at the begining.

I don't observe obvious performance difference with below patch at my
simple test, but seems more natural to finish read in the order they are
submitted.
Signed-off-by: NShaohua Li <shaohua.li@intel.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

btrfs_mkdir() must jump to the place of ending transaction after
btrfs_find_free_objectid() failed. Or this transaction can't end.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Flush any delalloc extents when we create a snapshot, so that recently
written file data is always included in the snapshot.

A later commit will add the ability to snapshot without the flush, but
most people expect flushing.
Signed-off-by: NSage Weil <sage@newdream.net>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The way we report df usage is way confusing for everybody, including some other
utilities (bacula for one).  So this patch makes df a little bit more
understandable.  First we make used actually count the total amount of used
space in all space info's.  This will give us a real view of how much disk space
is in use.  Second, for blocks available, only count data space.  This makes
things like bacula work because it says 0 when you can no longer write anymore
data to the disk.  I think this is a nice compromise, since you will end up with
something like the following

[root@alpha ~]# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/VolGroup-lv_root
                      148G   30G  111G  21% /
/dev/sda1             194M  116M   68M  64% /boot
tmpfs                 985M   12K  985M   1% /dev/shm
/dev/mapper/VolGroup-LogVol02
                      145G  140G     0 100% /mnt/btrfs-test

Compare this with btrfsctl -i output

[root@alpha btrfs-progs-unstable]# ./btrfsctl -i /mnt/btrfs-test/
Metadata, DUP: total=4.62GB, used=2.46GB
System, DUP: total=8.00MB, used=24.00KB
Data: total=134.80GB, used=134.80GB
Metadata: total=8.00MB, used=0.00
System: total=4.00MB, used=0.00
operation complete

This way we show that there is no more data space to be used, but we have
another 5GB of space left for metadata.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

When we scan devices in a multi-device filesystem, we memorize the original
name.  If the device gets a new name, later scans don't update the
in-kernel structures related to it, and we're not able to mount the
filesystem.

This patch updates device name during scaning.
Signed-off-by: NTARUISI Hiroaki <taruishi.hiroak@jp.fujitsu.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The btrfs defrag ioctl was limited to doing the entire file.  This
commit adds a new interface that can defrag a specific range inside
the file.

It can also force compression on the file, allowing you to selectively
compress individual files after they were created, even when mount -o
compress isn't turned on.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The btrfs defrag ioctl had some bugs around delalloc accounting, and it
wasn't properly skipping pages that were not in the mapping.

It wasn't properly clearing the page checked flag, which could make the
writeback code ignore the page forever while pinning it as dirty.

This commit fixes those problems and makes defrag a little smarter.  It
skips holes and it doesn't waste time defragging large extents.  If a
tiny extent comes before a very large extent, it will defrag both of
them to make sure the tiny extent ends up next to something big.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

The submit_bio helper thread can decide to loop back around to
service more bios.  This commit forces it to unplug first, which helps
reduce the latency seen by submitters.
Signed-off-by: NChris Mason <chris.mason@oracle.com>

Since theres not a good way to make sure the user sees the original default root
tree id, and not to mention it's 5 so is way different than any other volume,
just make subvol=0 mount the original default root.  This makes it a bit easier
for users to handle in the long run.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>

This patch needs to go along with my previous patch.  This lets us set the
default dir item's location to whatever root we want to use as our default
mounting subvol.  With this we don't have to use mount -o subvol=<tree id>
anymore to mount a different subvol, we can just set the new one and it will
just magically work.  I've done some moderate testing with this, mostly just
switching the default mount around, mounting subvols and the default mount at
the same time and such, everything seems to work.  Thanks,

Older kernels would generally be able to still mount the filesystem with the
default subvolume set, but it would result in a different volume being mounted,
which could be an even more unpleasant suprise for users.  So if you set your
default subvolume, you can't go back to older kernels.  Thanks,
Signed-off-by: NJosef Bacik <josef@redhat.com>
Signed-off-by: NChris Mason <chris.mason@oracle.com>