While testing with my buffer read fio jobs[1], I find that btrfs does not
perform well enough.

Here is a scenario in fio jobs:

We have 4 threads, "t1 t2 t3 t4", starting to buffer read a same file,
and all of them will race on add_to_page_cache_lru(), and if one thread
successfully puts its page into the page cache, it takes the responsibility
to read the page's data.

And what's more, reading a page needs a period of time to finish, in which
other threads can slide in and process rest pages:

     t1          t2          t3          t4
   add Page1
   read Page1  add Page2
     |         read Page2  add Page3
     |            |        read Page3  add Page4
     |            |           |        read Page4
-----|------------|-----------|-----------|--------
     v            v           v           v
    bio          bio         bio         bio

Now we have four bios, each of which holds only one page since we need to
maintain consecutive pages in bio.  Thus, we can end up with far more bios
than we need.

Here we're going to
a) delay the real read-page section and
b) try to put more pages into page cache.

With that said, we can make each bio hold more pages and reduce the number
of bios we need.

Here is some numbers taken from fio results:
         w/o patch                 w patch
       -------------  --------  ---------------
READ:    745MB/s        +25%       934MB/s

[1]:
[global]
group_reporting
thread
numjobs=4
bs=32k
rw=read
ioengine=sync
directory=/mnt/btrfs/

[READ]
filename=foobar
size=2000M
invalidate=1
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

For backref walking, we've introduce delayed ref's sequence.  However,
it changes our preallocation behavior.

The story is that when we preallocate an extent and then mark it written
piece by piece, the ideal case should be that we don't need to COW the
extent, which is why we use 'preallocate'.

But we may not make use of preallocation, since when we check for cross refs on
the extent, we may have two ref entries which have the same content except
the sequence value, and we recognize them as cross refs and do COW to allocate
another extent.

So we end up with several pieces of space instead of an whole extent.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

There is a small window where an eb can have no IO bits set on it, which
could potentially result in extent_buffer_under_io() returning false when we
want it to return true, which could result in not fun things happening.  So
in order to protect this case we need to hold the refs_lock when we make
this transition to make sure we get reliable results out of
extent_buffer_udner_io().  Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

This sounds sort of impossible but it is the only thing I can think of and
at the very least it is theoretically possible so here it goes.

If we are in try_release_extent_buffer we will check that the ref count on
the extent buffer is 1 and not under IO, and then go down and clear the tree
ref.  If between this check and clearing the tree ref somebody else comes in
and grabs a ref on the eb and the marks it dirty before
try_release_extent_buffer() does it's tree ref clear we can end up with a
dirty eb that will be freed while it is still dirty which will result in a
panic.  Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

I noticed while looking at an extent_buffer race that we will
unconditionally return 1 if we get down to release_extent_buffer after
clearing the tree ref.  However we can easily race in here and get a ref on
the eb and not actually free the eb.  So make release_extent_buffer return 1
if it free'd the eb and 0 if not so we can be a little kinder to the vm.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

Code is added to suppress the I/O stats printing at mount time if all
statistic values are zero.
Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>

People complained about the annoying kernel log message
"btrfs: no dev_stats entry found ... (OK on first mount after mkfs)"
everytime a filesystem is mounted for the first time after running
mkfs. Since the distribution of the btrfs-progs is not synchronized
to the kernel version, mkfs like it is now will be used also in the
future. Then this message is not useful to find errors, it is just
annoying. This commit removes the printk().
Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>

BTRFS_SETGET_FUNCS macro is used to generate btrfs_set_foo() and
btrfs_foo() functions, which read and write specific fields in the
extent buffer.

The total number of set/get functions is ~200, but in fact we only
need 8 functions: 2 for u8 field, 2 for u16, 2 for u32 and 2 for u64.

It results in redunction of ~37K bytes.

   text    data     bss     dec     hex filename
 629661   12489     216  642366   9cd3e fs/btrfs/btrfs.o.orig
 592637   12489     216  605342   93c9e fs/btrfs/btrfs.o
Signed-off-by: Li Zefan <lizefan@huawei.com>

The otime field is not zeroed, so users will see random otime in an old
filesystem with a new kernel which has otime support in the future.

The reserved bytes are also not zeroed, and we'll have compatibility
issue if we make use of those bytes.
Signed-off-by: Li Zefan <lizefan@huawei.com>

Inodes always allocate free space with BTRFS_BLOCK_GROUP_DATA type,
which means every inode has the same BTRFS_I(inode)->free_space pointer.

This shrinks struct btrfs_inode by 4 bytes (or 8 bytes on 64 bits).
Signed-off-by: Li Zefan <lizefan@huawei.com>

Changing printk_in_rcu to printk_ratelimited_in_rcu will suffice
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

When calling btrfs_next_old_leaf, we were leaking an extent buffer in the
rare case of using the deadlock avoidance code needed for the tree mod log.
Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

If a block group is ro, do not count its entries in when we dump space info.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

Block group has ro attributes, make dump_space_info show it.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

Here is the whole story:
1)
A free space cache consists of two parts:
o  free space cache inode, which is special becase it's stored in root tree.
o  free space info, which is stored as the above inode's file data.

But we only build up another new inode and does not flush its free space info
onto disk when we _clear and setup_ free space cache, and this ends up with
that the block group cache's cache_state remains DC_SETUP instead of DC_WRITTEN.

And holding DC_SETUP means that we will not truncate this free space cache inode,
which means the disk offset of its file extent will remain _unchanged_ at least
until next transaction finishes committing itself.

2)
We can set a block group readonly when we relocate the block group.

However,
if the readonly block group covers the disk offset where our free space cache
inode is going to write, it will force the free space cache inode into
cow_file_range() and it'll end up hitting a BUG_ON.

3)
Due to the above analysis, we fix this bug by adding the missing dirty flag.

4)
However, it's not over, there is still another case, nospace_cache.

With nospace_cache, we do not want to set dirty flag, instead we just truncate
free space cache inode and bail out with setting cache state DC_WRITTEN.

We can benifit from it since it saves us another 'pre-allocation' part which
usually costs a lot.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

During disk balance, we prealloc new file extent for file data relocation,
but we may fail in 'no available space' case, and it leads to flipping btrfs
into readonly.

It is not necessary to bail out and abort transaction since we do have several
ways to rescue ourselves from ENOSPC case.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

Since root can be fetched via BTRFS_I macro directly, we can save an args
for btrfs_is_free_space_inode().
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

For btree inode, its root is also 'tree root', so btree inode can be
misunderstood as a free space inode.

We should add one more check for btree inode.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

From btree_read_extent_buffer_pages(), currently repair_io_failure()
can be called with mirror_num being zero when submit_one_bio() returned
an error before. This used to cause a BUG_ON(!mirror_num) in
repair_io_failure() and indeed this is not a case that needs the I/O
repair code to rewrite disk blocks.
This commit prevents calling repair_io_failure() in this case and thus
avoids the BUG_ON() and malfunction.
Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

So shrink_delalloc has grown all sorts of cruft over the years thanks to
many reworkings of how we track enospc.  What happens now as we fill up the
disk is we will loop for freaking ever hoping to reclaim a arbitrary amount
of space of metadata, this was from when everybody flushed at the same time.
Now we only have people flushing one at a time.  So instead of trying to
reclaim a huge amount of space, just try to flush a decent chunk of space,
and stop looping as soon as we have enough free space to satisfy our
reservation.  This makes xfstests 224 go much faster.  Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

$ mkfs.btrfs /dev/sdb7
$ btrfstune -S1 /dev/sdb7
$ mount /dev/sdb7 /mnt/btrfs
mount: block device /dev/sdb7 is write-protected, mounting read-only
$ btrfs dev add /dev/sdb8 /mnt/btrfs/

Now we get a btrfs in which mnt flags has readonly but sb flags does
not.  So for those ioctls that only check sb flags with MS_RDONLY, it
is going to be a problem.
Setting subvolume flags is such an ioctl, we should use mnt_want_write_file()
to check RO flags.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>

mnt_want_write_file is faster when file has been opened for write.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>

mnt_want_write() and mnt_want_write_file() will check sb->s_flags with
MS_RDONLY, and we don't need to do it ourselves.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>

Move check of write access to mount into upper functions so that we can
use mnt_want_write_file instead, which is faster than mnt_want_write.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>

It should be 10 * 1024 * 1024.
Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>

There is weird logic I had to put in place to make sure that when we were
adding csums that we'd used the delalloc block rsv instead of the global
block rsv.  Part of this meant that we had to free up our transaction
reservation before we ran the delayed refs since csum deletion happens
during the delayed ref work.  The problem with this is that when we release
a reservation we will add it to the global reserve if it is not full in
order to keep us going along longer before we have to force a transaction
commit.  By releasing our reservation before we run delayed refs we don't
get the opportunity to drain down the global reserve for the work we did, so
we won't refill it as often.  This isn't a problem per-se, it just results
in us possibly committing transactions more and more often, and in rare
cases could cause those WARN_ON()'s to pop in use_block_rsv because we ran
out of space in our block rsv.

This also helps us by holding onto space while the delayed refs run so we
don't end up with as many people trying to do things at the same time, which
again will help us not force commits or hit the use_block_rsv warnings.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

We didn't check error of btrfs_update_inode(), but that error looks
easy to bubble back up.
Reviewed-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

We dereferenced "node" in the error message after freeing it.  Also
btrfs_panic() can return so we should return an error code instead of
continuing.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>

There used to be a BUG_ON(ret) there before EH patch (79787eaa) went in.
Bail out with EINVAL.

Cc: David Sterba <dsterba@suse.cz>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

When bailing from open_ctree() err is returned, not ret.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

This will be used in conjunction with btrfs device ready <dev>.  This is
needed for initrd's to have a nice and lightweight way to tell if all of the
devices needed for a file system are in the cache currently.  This keeps
them from having to do mount+sleep loops waiting for devices to show up.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

Those crazy gentoo guys have been complaining about ENOSPC errors on their
portage volumes.  This is because doing things like untar tends to create
lots of new files which will soak up all the reservation space in the
delayed inodes.  Usually this gets papered over by the fact that we will try
and commit the transaction, however if this happens in the wrong spot or we
choose not to commit the transaction you will be screwed.  So add the
ability to expclitly flush delayed inodes to free up space.  Please test
this out guys to make sure it works since as usual I cannot reproduce.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fusionio.com>

Commit c11d2c23 (Btrfs: add ioctl to get and reset the device
stats) introduced two ioctls doing almost the same thing distinguished
by just the ioctl number which encodes "do reset after read". I have
suggested

http://www.mail-archive.com/linux-btrfs@vger.kernel.org/msg16604.html

to implement it via the ioctl args. This hasn't happen, and I think we
should use a more clean way to pass flags and should not waste ioctl
numbers.

CC: Stefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: David Sterba <dsterba@suse.cz>

Rebased on btrfs-next and retested.

Inform should_defrag_range if BTRFS_DEFRAG_RANGE_COMPRESS is set. If so, skip
checks for adjacent extents and extent size when deciding whether to defrag,
as these can prevent an uncompressed and unfragmented file from being
compressed as requested.
Signed-off-by: Andrew Mahone <andrew.mahone@gmail.com>

"root->fs_info" and "fs_info" are the same, but "fs_info" is prefered
because it is shorter and that's what is used in the rest of the
function.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>

Before the update_time inode operation was indroduced, it was
not possible to prevent updates of atime on RO subvolumes. VFS
was only able to check for RO on the mount, but did not know
anything about btrfs subvolumes.

btrfs_update_time does now check if the root is RO and skip
updating of times.
Signed-off-by: Alexander Block <ablock84@googlemail.com>

Btrfs allows to turn on compression on a mounted and used filesystem
by issuing mount -o remount,compress=lzo.
This patch allows to turn compression off again
while the filesystem is mounted. As suggested by David Sterba
if the compress-force option was set, it is implicitly cleared
if compression is turned off.
Tested-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Arnd Hannemann <arnd@arndnet.de>

We do all of our inode updating when we change it, and now that we do
->update_time we don't need ->dirty_inode for atime updates anymore, so just
remove it.  Thanks,
Signed-off-by: Josef Bacik <josef@redhat.com>

The btrfs locks were unconditionally calling wake_up as the
locks were released.  This lead to extra thrashing on the waitqueue,
especially for locks that were dominated by readers.
Signed-off-by: Chris Mason <chris.mason@fusionio.com>

Waiting on spindles improves performance, but ssds want all the
IO as quickly as we can push it down.
Signed-off-by: Chris Mason <chris.mason@fusionio.com>