# Conflicts:
#	fs/btrfs/file.c

# Conflicts:
#	fs/btrfs/file.c

Xfstests btrfs/011 complains about a deadlock warning,

[ 1226.649039] =========================================================
[ 1226.649039] [ INFO: possible irq lock inversion dependency detected ]
[ 1226.649039] 4.1.0+ #270 Not tainted
[ 1226.649039] ---------------------------------------------------------
[ 1226.652955] kswapd0/46 just changed the state of lock:
[ 1226.652955]  (&delayed_node->mutex){+.+.-.}, at: [<ffffffff81458735>] __btrfs_release_delayed_node+0x45/0x1d0
[ 1226.652955] but this lock took another, RECLAIM_FS-unsafe lock in the past:
[ 1226.652955]  (&fs_info->dev_replace.lock){+.+.+.}

and interrupts could create inverse lock ordering between them.

[ 1226.652955]
other info that might help us debug this:
[ 1226.652955] Chain exists of:
  &delayed_node->mutex --> &found->groups_sem --> &fs_info->dev_replace.lock

[ 1226.652955]  Possible interrupt unsafe locking scenario:

[ 1226.652955]        CPU0                    CPU1
[ 1226.652955]        ----                    ----
[ 1226.652955]   lock(&fs_info->dev_replace.lock);
[ 1226.652955]                                local_irq_disable();
[ 1226.652955]                                lock(&delayed_node->mutex);
[ 1226.652955]                                lock(&found->groups_sem);
[ 1226.652955]   <Interrupt>
[ 1226.652955]     lock(&delayed_node->mutex);
[ 1226.652955]
 *** DEADLOCK ***

Commit 084b6e7c ("btrfs: Fix a lockdep warning when running xfstest.") tried
to fix a similar one that has the exactly same warning, but with that, we still
run to this.

The above lock chain comes from
btrfs_commit_transaction
  ->btrfs_run_delayed_items
    ...
    ->__btrfs_update_delayed_inode
      ...
      ->__btrfs_cow_block
         ...
         ->find_free_extent
            ->cache_block_group
              ->load_free_space_cache
                ->btrfs_readpages
                  ->submit_one_bio
                    ...
                    ->__btrfs_map_block
                      ->btrfs_dev_replace_lock

However, with high memory pressure, tasks which hold dev_replace.lock can
be interrupted by kswapd and then kswapd is intended to release memory occupied
by superblock, inodes and dentries, where we may call evict_inode, and it comes
to

[ 1226.652955]  [<ffffffff81458735>] __btrfs_release_delayed_node+0x45/0x1d0
[ 1226.652955]  [<ffffffff81459e74>] btrfs_remove_delayed_node+0x24/0x30
[ 1226.652955]  [<ffffffff8140c5fe>] btrfs_evict_inode+0x34e/0x700

delayed_node->mutex may be acquired in __btrfs_release_delayed_node(), and it leads
to a ABBA deadlock.

To fix this, we can use "blocking rwlock" used in the case of extent_buffer, but
things are simpler here since we only needs read's spinlock to blocking lock.

With this, btrfs/011 no more produces warnings in dmesg.
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>

Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>

The control device is accessible when no filesystem is mounted and we
may want to query features supported by the module. This is already
possible using the sysfs files, this ioctl is for parity and
convenience.
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>

The current practical default is ~4k on x86_64 (the logic is more complex,
simplified for brevity), the inlined files land in the metadata group and
thus consume space that could be needed for the real metadata.

The inlining brings some usability surprises:

1) total space consumption measured on various filesystems and btrfs
   with DUP metadata was quite visible because of the duplicated data
   within metadata

2) inlined data may exhaust the metadata, which are more precious in case
   the entire device space is allocated to chunks (ie. balance cannot
   make the space more compact)

3) performance suffers a bit as the inlined blocks are duplicate and
   stored far away on the device.

Proposed fix: set the default to 2048

This fixes namely 1), the total filesysystem space consumption will be on
par with other filesystems.

Partially fixes 2), more data are pushed to the data block groups.

The characteristics of 3) are based on actual small file size
distribution.

The change is independent of the metadata blockgroup type (though it's
most visible with DUP) or system page size as these parameters are not
trival to find out, compared to file size.
Signed-off-by: David Sterba <dsterba@suse.com>

Let's remove the error message that appears when the tree_id is not
present. This can happen with the quota tree and has been observed in
practice. The applications are supposed to handle -ENOENT and we don't
need to report that in the system log as it's not a fatal error.
Reported-by: Vlastimil Babka <vbabka@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>

With CONFIG_SMP and CONFIG_PREEMPT both disabled, gcc decides
to partially inline the get_state_failrec() function but cannot
figure out that means the failrec pointer is always valid
if the function returns success, which causes a harmless
warning:

fs/btrfs/extent_io.c: In function 'clean_io_failure':
fs/btrfs/extent_io.c:2131:4: error: 'failrec' may be used uninitialized in this function [-Werror=maybe-uninitialized]

This marks get_state_failrec() and set_state_failrec() both
as 'noinline', which avoids the warning in all cases for me,
and seems less ugly than adding a fake initialization.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: 47dc196a ("btrfs: use proper type for failrec in extent_state")
Signed-off-by: David Sterba <dsterba@suse.com>

When starting up linux with btrfs filesystem, I got many memory leak
messages by kmemleak as,

unreferenced object 0xffff880066882000 (size 4096):
  comm "modprobe", pid 730, jiffies 4294690024 (age 196.599s)
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  backtrace:
    [<ffffffff8174d52e>] kmemleak_alloc+0x4e/0xb0
    [<ffffffff811d09aa>] kmem_cache_alloc_trace+0xea/0x1e0
    [<ffffffffa03620fb>] btrfs_alloc_dummy_fs_info+0x6b/0x2a0 [btrfs]
    [<ffffffffa03624fc>] btrfs_alloc_dummy_block_group+0x5c/0x120 [btrfs]
    [<ffffffffa0360aa9>] btrfs_test_free_space_cache+0x39/0xed0 [btrfs]
    [<ffffffffa03b5a74>] trace_raw_output_xfs_attr_class+0x54/0xe0 [xfs]
    [<ffffffff81002122>] do_one_initcall+0xb2/0x1f0
    [<ffffffff811765aa>] do_init_module+0x5e/0x1e9
    [<ffffffff810fec09>] load_module+0x20a9/0x2690
    [<ffffffff810ff439>] SyS_finit_module+0xb9/0xf0
    [<ffffffff81757daf>] entry_SYSCALL_64_fastpath+0x12/0x76
    [<ffffffffffffffff>] 0xffffffffffffffff
unreferenced object 0xffff8800573f8000 (size 10256):
  comm "modprobe", pid 730, jiffies 4294690185 (age 196.460s)
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  backtrace:
    [<ffffffff8174d52e>] kmemleak_alloc+0x4e/0xb0
    [<ffffffff8119ca6e>] kmalloc_order+0x5e/0x70
    [<ffffffff8119caa4>] kmalloc_order_trace+0x24/0x90
    [<ffffffffa03620b3>] btrfs_alloc_dummy_fs_info+0x23/0x2a0 [btrfs]
    [<ffffffffa03624fc>] btrfs_alloc_dummy_block_group+0x5c/0x120 [btrfs]
    [<ffffffffa036603d>] run_test+0xfd/0x320 [btrfs]
    [<ffffffffa0366f34>] btrfs_test_free_space_tree+0x94/0xee [btrfs]
    [<ffffffffa03b5aab>] trace_raw_output_xfs_attr_class+0x8b/0xe0 [xfs]
    [<ffffffff81002122>] do_one_initcall+0xb2/0x1f0
    [<ffffffff811765aa>] do_init_module+0x5e/0x1e9
    [<ffffffff810fec09>] load_module+0x20a9/0x2690
    [<ffffffff810ff439>] SyS_finit_module+0xb9/0xf0
    [<ffffffff81757daf>] entry_SYSCALL_64_fastpath+0x12/0x76
    [<ffffffffffffffff>] 0xffffffffffffffff

This patch lets btrfs using fs_info stored in btrfs_root for
block group cache directly without allocating a new one.

Fixes: d0bd4560 ("Btrfs: add fragment=* debug mount option")
Signed-off-by: Kinglong Mee <kinglongmee@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>

btrfs failed in xfstests btrfs/080 with -o nodatacow.

Can be reproduced by following script:
  DEV=/dev/vdg
  MNT=/mnt/tmp

  umount $DEV &>/dev/null
  mkfs.btrfs -f $DEV
  mount -o nodatacow $DEV $MNT

  dd if=/dev/zero of=$MNT/test bs=1 count=2048 &
  btrfs subvolume snapshot -r $MNT $MNT/test_snap &
  wait
  --
  We can see dd failed on NO_SPACE.

Reason:
  __btrfs_buffered_write should run cow write when no_cow impossible,
  and current code is designed with above logic.
  But check_can_nocow() have 2 type of return value(0 and <0) on
  can_not_no_cow, and current code only continue write on first case,
  the second case happened in doing subvolume.

Fix:
  Continue write when check_can_nocow() return 0 and <0.
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>

Cleanup.

kmem_cache_destroy has support NULL argument checking,
so drop the double null testing before calling it.
Signed-off-by: Kinglong Mee <kinglongmee@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>

We were getting build warning about:
fs/btrfs/extent-tree.c:7021:34: warning: ‘used_bg’ may be used
	uninitialized in this function

It is not a valid warning as used_bg is never used uninitilized since
locked is initially false so we can never be in the section where
'used_bg' is used. But gcc is not able to understand that and we can
initialize it while declaring to silence the warning.
Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
Signed-off-by: David Sterba <dsterba@suse.com>

We use the private member of extent_state to store the failrec and play
pointless pointer games.
Signed-off-by: David Sterba <dsterba@suse.com>

CURRENT_TIME macro is not appropriate for filesystems as it
doesn't use the right granularity for filesystem timestamps.
Use current_fs_time() instead.
Signed-off-by: Deepa Dinamani <deepa.kernel@gmail.com>
Cc: Chris Mason <clm@fb.com>
Cc: Josef Bacik <jbacik@fb.com>
Cc: linux-btrfs@vger.kernel.org
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>

The kernel provides a swap() that does the same thing as this code.
Signed-off-by: Dave Jones <dsj@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>

While running btrfs_mksubvol(), d_really_is_positive() is called twice.
First in btrfs_mksubvol() and second inside btrfs_may_create().  So I
remove the first one.
Signed-off-by: Byongho Lee <bhlee.kernel@gmail.com>
Signed-off-by: David Sterba <dsterba@suse.com>

Simplify expression in btrfs_calc_trans_metadata_size().
Signed-off-by: Byongho Lee <bhlee.kernel@gmail.com>
Reviewed-by: Stefan Behrens <sbehrens@giantdisaster.de>
Signed-off-by: David Sterba <dsterba@suse.com>

We will sometimes start background flushing the various enospc related things
(delayed nodes, delalloc, etc) if we are getting close to reserving all of our
available space.  We don't want to do this however when we are actually using
this space as it causes unneeded thrashing.  We currently try to do this by
checking bytes_used >= thresh, but bytes_used is only part of the equation, we
need to use bytes_reserved as well as this represents space that is very likely
to become bytes_used in the future.

My tracing tool will keep count of the number of times we kick off the async
flusher, the following are counts for the entire run of generic/027

		No Patch	Patch
avg: 		5385		5009
median:		5500		4916

We skewed lower than the average with my patch and higher than the average with
the patch, overall it cuts the flushing from anywhere from 5-10%, which in the
case of actual ENOSPC is quite helpful.  Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
Reviewed-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>

There are a few places where we add to trans->bytes_reserved but don't have the
corresponding trace point.  With these added my tool no longer sees transaction
leaks.
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>

truncate_space_check is using btrfs_csum_bytes_to_leaves() but forgetting to
multiply by nodesize so we get an actual byte count.  We need a tracepoint here
so that we have the matching reserve for the release that will come later.  Also
add a comment to make clear what the intent of truncate_space_check is.
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>

I'm writing a tool to visualize the enospc system in order to help debug enospc
bugs and I found weird data and ran it down to when we update the global block
rsv.  We add all of the remaining free space to the block rsv, do a trace event,
then remove the extra and do another trace event.  This makes my visualization
look silly and is unintuitive code as well.  Fix this stuff to only add the
amount we are missing, or free the amount we are missing.  This is less clean to
read but more explicit in what it is doing, as well as only emitting events for
values that make sense.  Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>

For a non-existent device, old code bypasses adding it in dev's reada
queue.

And to solve problem of unfinished waitting in raid5/6,
commit 5fbc7c59 ("Btrfs: fix unfinished readahead thread for
raid5/6 degraded mounting")
adding an exception for the first stripe, in short, the first
stripe will always be processed whether the device exists or not.

Actually we have a better way for the above request: just bypass
creation of the reada_extent for non-existent device, it will make
code simple and effective.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>

Reada background works is not designed to finish all jobs
completely, it will break in following case:
1: When a device reaches workload limit (MAX_IN_FLIGHT)
2: Total reads reach max limit (10000)
3: All devices don't have queued more jobs, often happened in DUP case

And if all background works exit with remaining jobs,
btrfs_reada_wait() will wait indefinetelly.

Above problem is rarely happened in old code, because:
1: Every work queues 2x new works
   So many works reduced chances of undone jobs.
2: One work will continue 10000 times loop in case of no-jobs
   It reduced no-thread window time.

But after we fixed above case, the "undone reada extents" frequently
happened.

Fix:
 Check to ensure we have at least one thread if there are undone jobs
 in btrfs_reada_wait().
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>

Reada creates 2 works for each level of tree recursively.

In case of a tree having many levels, the number of created works
is 2^level_of_tree.
Actually we don't need so many works in parallel, this patch limits
max works to BTRFS_MAX_MIRRORS * 2.

The per-fs works_counter will be also used for btrfs_reada_wait() to
check is there are background workers.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>

No need to decrease dev->reada_in_flight in __readahead_hook()'s
internal and reada_extent_put().
reada_extent_put() have no chance to decrease dev->reada_in_flight
in free operation, because reada_extent have additional refcnt when
scheduled to a dev.

We can put inc and dec operation for dev->reada_in_flight to one
place instead to make logic simple and safe, and move useless
reada_extent->scheduled_for to a bool flag instead.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>

Remove one copy of loop to fix the typo of iterate zones.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>

Current code set nritems to 0 to make for_loop useless to bypass it,
and set generation's value which is not necessary.
Jump into cleanup directly is better choise.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>

What __readahead_hook() need exactly is fs_info, no need to convert
fs_info to root in caller and convert back in __readahead_hook()
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>

reada_start_machine_dev() already have reada_extent pointer, pass
it into __readahead_hook() directly instead of search radix_tree
will make code run faster.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>

We can't release reada_extent earlier than __readahead_hook(), because
__readahead_hook() still need to use it, it is necessary to hode a refcnt
to avoid it be freed.

Actually it is not a problem after my patch named:
  Avoid many times of empty loop
It make reada_extent in above line include at least one reada_extctl,
which keeps additional one refcnt for reada_extent.

But we still need this patch to make the code in pretty logic.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>

level is not used in severial functions, remove them from arguments,
and remove relative code for get its value.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>

When failed adding all dev_zones for a reada_extent, the extent
will have no chance to be selected to run, and keep in memory
for ever.

We should bypass this extent to avoid above case.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: David Sterba <dsterba@suse.com>