We used 3 variants to track the state of the transaction, it was complex
and wasted the memory space. Besides that, it was hard to understand that
which types of the transaction handles should be blocked in each transaction
state, so the developers often made mistakes.

This patch improved the above problem. In this patch, we define 6 states
for the transaction,
  enum btrfs_trans_state {
	TRANS_STATE_RUNNING		= 0,
	TRANS_STATE_BLOCKED		= 1,
	TRANS_STATE_COMMIT_START	= 2,
	TRANS_STATE_COMMIT_DOING	= 3,
	TRANS_STATE_UNBLOCKED		= 4,
	TRANS_STATE_COMPLETED		= 5,
	TRANS_STATE_MAX			= 6,
  }
and just use 1 variant to track those state.

In order to make the blocked handle types for each state more clear,
we introduce a array:
  unsigned int btrfs_blocked_trans_types[TRANS_STATE_MAX] = {
	[TRANS_STATE_RUNNING]		= 0U,
	[TRANS_STATE_BLOCKED]		= (__TRANS_USERSPACE |
					   __TRANS_START),
	[TRANS_STATE_COMMIT_START]	= (__TRANS_USERSPACE |
					   __TRANS_START |
					   __TRANS_ATTACH),
	[TRANS_STATE_COMMIT_DOING]	= (__TRANS_USERSPACE |
					   __TRANS_START |
					   __TRANS_ATTACH |
					   __TRANS_JOIN),
	[TRANS_STATE_UNBLOCKED]		= (__TRANS_USERSPACE |
					   __TRANS_START |
					   __TRANS_ATTACH |
					   __TRANS_JOIN |
					   __TRANS_JOIN_NOLOCK),
	[TRANS_STATE_COMPLETED]		= (__TRANS_USERSPACE |
					   __TRANS_START |
					   __TRANS_ATTACH |
					   __TRANS_JOIN |
					   __TRANS_JOIN_NOLOCK),
  }
it is very intuitionistic.

Besides that, because we remove ->in_commit in transaction structure, so
the lock ->commit_lock which was used to protect it is unnecessary, remove
->commit_lock.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

We checked the commit time to avoid committing the transaction
frequently, but it is unnecessary because:
- It made the transaction commit spend more time, and delayed the
  operation of the external writers(TRANS_START/TRANS_USERSPACE).
- Except the space that we have to commit transaction, such as
  snapshot creation, btrfs doesn't commit the transaction on its
  own initiative.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

We used ->num_joined track if there were some writers which join the current
transaction when the committer was sleeping. If some writers joined the current
transaction, we has to continue the while loop to do some necessary stuff, such
as flush the ordered operations. But it is unnecessary because we will do it
after the while loop.

Besides that, tracking ->num_joined would make the committer drop into the while
loop when there are lots of internal writers(TRANS_JOIN).

So we remove ->num_joined and don't track if there are some writers which join
the current transaction when the committer is sleeping.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

It is unnecessary to flush the delalloc inodes again and again because
we don't care the dirty pages which are introduced after the flush, and
they will be flush in the transaction commit.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

btrfs_commit_transaction has the following loop before we commit the
transaction.

do {
    // attempt to do some useful stuff and/or sleep
} while (atomic_read(&cur_trans->num_writers) > 1 ||
	 (should_grow && cur_trans->num_joined != joined));

This is used to prevent from the TRANS_START to get in the way of a
committing transaction. But it does not prevent from TRANS_JOIN, that
is we would do this loop for a long time if some writers JOIN the
current transaction endlessly.

Because we need join the current transaction to do some useful stuff,
we can not block TRANS_JOIN here. So we introduce a external writer
counter, which is used to count the TRANS_USERSPACE/TRANS_START writers.
If the external writer counter is zero, we can break the above loop.

In order to make the code more clear, we don't use enum variant
to define the type of the transaction handle, use bitmask instead.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

If the transaction is removed from the transaction list, it means the
transaction has been committed successfully. So it is impossible to
call cleanup_transaction(), otherwise there is something wrong with
the code logic. Thus, we use BUG_ON() instead of the original handle.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

When we umount a fs with serious errors, we will invoke btrfs_cleanup_transactions()
to clean up the residual transaction. At this time, It is impossible to start a new
transaction, so we needn't assign trans_no_join to 1, and also needn't clear running
transaction every time we destroy a residual transaction.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

Before applying this patch, we need flush all the delalloc inodes in
the fs when we want to create a snapshot, it wastes time, and make
the transaction commit be blocked for a long time. It means some other
user operation would also be blocked for a long time.

This patch improves this problem, we just flush the delalloc inodes that
in the source trees before snapshot creation, so the transaction commit
will complete quickly.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

The reason we introduce per-subvolume ordered extent list is the same
as the per-subvolume delalloc inode list.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

When we create a snapshot, we need flush all delalloc inodes in the
fs, just flushing the inodes in the source tree is OK. So we introduce
per-subvolume delalloc inode list.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

The grab/put funtions will be used in the next patch, which need grab
the root object and ensure it is not freed. We use reference counter
instead of the srcu lock is to aovid blocking the memory reclaim task,
which invokes synchronize_srcu().
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

There are several functions whose code is similar, such as
  btrfs_find_last_root()
  btrfs_read_fs_root_no_radix()

Besides that, some functions are invoked twice, it is unnecessary,
for example, we are sure that all roots which is found in
  btrfs_find_orphan_roots()
have their orphan items, so it is unnecessary to check the orphan
item again.

So cleanup it.
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

The snapshot/subvolume deletion might spend lots of time, it would make
the remount task wait for a long time. This patch improve this problem,
we will break the deletion if the fs is remounted to be R/O. It will make
the users happy.

Cc: David Sterba <dsterba@suse.cz>
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

If the fs is remounted to be R/O, it is unnecessary to call
btrfs_clean_one_deleted_snapshot(), so move the R/O check out of
this function. And besides that, it can make the check logic in the
caller more clear.

Cc: David Sterba <dsterba@suse.cz>
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

Cc: David Sterba <dsterba@suse.cz>
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

In order to avoid the R/O remount, we acquired ->s_umount lock during
we deleted the dead snapshots and subvolumes. But it is unnecessary,
because we have cleaner_mutex.

We use cleaner_mutex to protect the process of the dead snapshots/subvolumes
deletion. And when we remount the fs to be R/O, we also acquire this mutex to
do cleanup after we change the status of the fs. That is this lock can serialize
the above operations, the cleaner can be aware of the status of the fs, and if
the cleaner is deleting the dead snapshots/subvolumes, the remount task will
wait for it. So it is safe to remove ->s_umount in cleaner_kthread().

Cc: David Sterba <dsterba@suse.cz>
Signed-off-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

btrfs_read_fs_root_no_name() already checks if btrfs_root_refs()
is zero and returns ENOENT in this case. There is no need to do
it again in six places.
Signed-off-by: NStefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

No need to check for NULL in send.c and disk-io.c.
Signed-off-by: NStefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

Signed-off-by: NStefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

We don't need to copy it back to user side as it remains unchanged.
Signed-off-by: NLiu Bo <bo.li.liu@oracle.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

Clean up the format of the definitions of BTRFS_BLOCK_GROUP_RAID5 and
BTRFS_BLOCK_GROUP_RAID6.
Signed-off-by: NAndreas Philipp <philipp.andreas@gmail.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

sctx is removed from the argument of the function that
doesn't use sctx.
Signed-off-by: NTsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

The size parameter to btrfs_extend_item() is the number of bytes
to add to the item, not the size of the item after the operation
(like it is for btrfs_truncate_item(), there the size parameter
is not the number of bytes to take away, but the total size of
the item after truncation).
Fix it in the comment.
Signed-off-by: NStefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

btrfs_qgroup_wait_for_completion waits until the currently running qgroup
operation completes. It returns immediately when no rescan process is in
progress. This is useful to automate things around the rescan process (e.g.
testing).
Signed-off-by: NJan Schmidt <list.btrfs@jan-o-sch.net>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

When doing qgroup accounting, we call ulist_alloc()/ulist_free() every time
when we want to walk qgroup tree.

By introducing 'qgroup_ulist', we only need to call ulist_alloc()/ulist_free()
once. This reduce some sys time to allocate memory, see the measurements below

fsstress -p 4 -n 10000 -d $dir

With this patch:

real    0m50.153s
user    0m0.081s
sys     0m6.294s

real    0m51.113s
user    0m0.092s
sys     0m6.220s

real    0m52.610s
user    0m0.096s
sys     0m6.125s	avg 6.213
-----------------------------------------------------
Without the patch:

real    0m54.825s
user    0m0.061s
sys     0m10.665s

real    1m6.401s
user    0m0.089s
sys     0m11.218s

real    1m13.768s
user    0m0.087s
sys     0m10.665s       avg 10.849

we can see the sys time reduce ~43%.
Signed-off-by: NWang Shilong <wangsl-fnst@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

We want to know if there are debugging features compiled in, this may
affect performance. The message is printed before the sanity checks.
Also kill version.h file that serves no purpose, we don't use any
version tag for kernel module.
Signed-off-by: NDavid Sterba <dsterba@suse.cz>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

Signed-off-by: NDavid Sterba <dsterba@suse.cz>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

And change the message level to KERN_INFO.
Signed-off-by: NDavid Sterba <dsterba@suse.cz>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

The 'end' value must exactly cover the end of the interval, which means
one byte less than the expected block alignment, or in case of a file
smaller than one block, one byte less than the inode size.
Signed-off-by: NDavid Sterba <dsterba@suse.cz>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

Code checked for raid 5 flag in two else-if branches, so code would never be reached. Probably a copy-paste bug.
Signed-off-by: NHenrik Nordvik <henrikno@gmail.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

Dave reported a panic because the extent_root->commit_root was NULL in the
caching kthread.  That is because we just unset it in free_root_pointers, which
is not the correct thing to do, we have to either wait for the caching kthread
to complete or hold the extent_commit_sem lock so we know the thread has exited.
This patch makes the kthreads all stop first and then we do our cleanup.  This
should fix the race.  Thanks,
Reported-by: NDavid Sterba <dsterba@suse.cz>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

Commit be283b2e
(    Btrfs: use helper to cleanup tree roots) introduced the following bug,

 BUG: unable to handle kernel NULL pointer dereference at 0000000000000034
 IP: [<ffffffffa039368c>] extent_buffer_get+0x4/0xa [btrfs]
[...]
 Pid: 2463, comm: btrfs-cache-1 Tainted: G           O 3.9.0+ #4 innotek GmbH VirtualBox/VirtualBox
 RIP: 0010:[<ffffffffa039368c>]  [<ffffffffa039368c>] extent_buffer_get+0x4/0xa [btrfs]
 Process btrfs-cache-1 (pid: 2463, threadinfo ffff880112d60000, task ffff880117679730)
[...]
 Call Trace:
  [<ffffffffa0398a99>] btrfs_search_slot+0x104/0x64d [btrfs]
  [<ffffffffa039aea4>] btrfs_next_old_leaf+0xa7/0x334 [btrfs]
  [<ffffffffa039b141>] btrfs_next_leaf+0x10/0x12 [btrfs]
  [<ffffffffa039ea13>] caching_thread+0x1a3/0x2e0 [btrfs]
  [<ffffffffa03d8811>] worker_loop+0x14b/0x48e [btrfs]
  [<ffffffffa03d86c6>] ? btrfs_queue_worker+0x25c/0x25c [btrfs]
  [<ffffffff81068d3d>] kthread+0x8d/0x95
  [<ffffffff81068cb0>] ? kthread_freezable_should_stop+0x43/0x43
  [<ffffffff8151e5ac>] ret_from_fork+0x7c/0xb0
  [<ffffffff81068cb0>] ? kthread_freezable_should_stop+0x43/0x43
RIP  [<ffffffffa039368c>] extent_buffer_get+0x4/0xa [btrfs]

We've free'ed commit_root before actually getting to free block groups where
caching thread needs valid extent_root->commit_root.
Signed-off-by: NLiu Bo <bo.li.liu@oracle.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>
Signed-off-by: NChris Mason <chris.mason@fusionio.com>

Dave reported a NULL pointer deref.  This is caused because he thought he'd be
smart and add sanity checks to the extent_io bit operations, but he didn't
expect a tree to have a NULL mapping.  To fix this we just need to init the
relocation's processed_blocks with the btree_inode->i_mapping.  Thanks,
Reported-by: NDavid Sterba <dsterba@suse.cz>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>
Signed-off-by: NChris Mason <chris.mason@fusionio.com>

There is a path where btrfs_drop_inode() is called with its inode's root
is NULL: In btrfs_new_inode(), when btrfs_set_inode_index() fails,
iput() is called. We should handle this case before taking look at the
root->root_item.
Signed-off-by: NNaohiro Aota <naota@elisp.net>
Reviewed-by: NMiao Xie <miaox@cn.fujitsu.com>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>
Signed-off-by: NChris Mason <chris.mason@fusionio.com>

We get a use after free if we had a transaction to cleanup since there could be
delayed inodes which refer to their respective fs_root.  Thanks
Reported-by: NDavid Sterba <dsterba@suse.cz>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>
Signed-off-by: NChris Mason <chris.mason@fusionio.com>

Btrfs has been pointer tagging bi_private and using bi_bdev
to store the stripe index and mirror number of failed IOs.

As bios bubble back up through the call chain, we use these
to decide if and how to retry our IOs.  They are also used
to count IO failures on a per device basis.

Recently a bio tracepoint was added lead to crashes because
we were abusing bi_bdev.

This commit adds a btrfs bioset, and creates explicit fields
for the mirror number and stripe index.  The plan is to
extend this structure for all of the fields currently in
struct btrfs_bio, which will mean one less kmalloc in
our IO path.
Signed-off-by: NChris Mason <chris.mason@fusionio.com>
Reported-by: NTejun Heo <tj@kernel.org>

If we fail to load the chunk tree we'll call free_root_pointers, except we may
not have assigned the roots for the dev_root/extent_root/csum_root yet, so we
could NULL pointer deref at this point.  Just add checks to make sure these
roots are set to keep us from panicing.  Thanks,
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

The quota_tree was set up to use the empty_block_rsv before
which would be problematic when the filesystem is filled up
and ENOSPC happens during internal operations while the quota
tree is updated and COWed (when the btrfs_qgroup_info_item
items) are written. In fact, use_block_rsv() which is used
in btrfs_cow_block() falls back to the global_block_rsv in
this case. But just in order to make it more clear what is
happening, change it to explicitly use the global_block_rsv.
Signed-off-by: NStefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>

end_bio_extent_readpage computes whole_page based on bv_offset and
bv_len, without taking into account that blk_update_request may modify
them when some of the blocks to be read into a page produce a read
error.  This would cause the read to unlock only part of the file
range associated with the page, which would in turn leave the entire
page locked, which would not only keep the process blocked instead of
returning -EIO to it, but also prevent any further access to the file.

It turns out that btrfs always issues whole-page reads and writes.
The special handling of non-whole_page appears to be a mistake or a
left-over from a time when this wasn't the case.  Indeed,
end_bio_extent_writepage distinguished between whole_page and
non-whole_page writes but behaved identically in both cases!

I've replaced the whole_page computations with warnings, just to be
sure that we're not issuing partial page reads or writes.  The
warnings should probably just go away some time.
Signed-off-by: NAlexandre Oliva <oliva@gnu.org>
Signed-off-by: NJosef Bacik <jbacik@fusionio.com>