At ctree.c:key_search(), the assertion that verifies the first key on a
child extent buffer corresponds to the key at a specific slot in the
parent has a disadvantage: we effectively hit a BUG_ON() which requires
rebooting the machine later. It also does not tell any information about
which extent buffer is affected, from which root, the expected and found
keys, etc.
However as of commit 581c176041 ("btrfs: Validate child tree block's
level and first key"), that assertion is not needed since at the time we
read an extent buffer from disk we validate that its first key matches the
key, at the respective slot, in the parent extent buffer. Therefore just
remove the assertion at key_search().
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The function map_private_extent_buffer() can return an -EINVAL error, and
it is called by generic_bin_search() which will return back the error. The
btrfs_bin_search() function in turn calls generic_bin_search() and the
key_search() function calls btrfs_bin_search(), so both can return the
-EINVAL error coming from the map_private_extent_buffer() function. Some
callers of these functions were ignoring that these functions can return
an error, so fix them to deal with error return values.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The last caller that does not have a fixed value of lock is
btrfs_set_path_blocking, that actually does the same conditional swtich
by the lock type so we can merge the branches together and remove the
helper.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: David Sterba <dsterba@suse.com>
btrfs_set_lock_blocking is now only a simple wrapper around
btrfs_set_lock_blocking_write. The name does not bring any semantic
value that could not be inferred from the new function so there's no
point keeping it.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: David Sterba <dsterba@suse.com>
We can use the right helper where the lock type is a fixed parameter.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: David Sterba <dsterba@suse.com>
Before this patch, qgroup code traces the whole subtree of subvolume and
reloc trees unconditionally.
This makes qgroup numbers consistent, but it could cause tons of
unnecessary extent tracing, which causes a lot of overhead.
However for subtree swap of balance, just swap both subtrees because
they contain the same contents and tree structure, so qgroup numbers
won't change.
It's the race window between subtree swap and transaction commit could
cause qgroup number change.
This patch will delay the qgroup subtree scan until COW happens for the
subtree root.
So if there is no other operations for the fs, balance won't cause extra
qgroup overhead. (best case scenario)
Depending on the workload, most of the subtree scan can still be
avoided.
Only for worst case scenario, it will fall back to old subtree swap
overhead. (scan all swapped subtrees)
[[Benchmark]]
Hardware:
VM 4G vRAM, 8 vCPUs,
disk is using 'unsafe' cache mode,
backing device is SAMSUNG 850 evo SSD.
Host has 16G ram.
Mkfs parameter:
--nodesize 4K (To bump up tree size)
Initial subvolume contents:
4G data copied from /usr and /lib.
(With enough regular small files)
Snapshots:
16 snapshots of the original subvolume.
each snapshot has 3 random files modified.
balance parameter:
-m
So the content should be pretty similar to a real world root fs layout.
And after file system population, there is no other activity, so it
should be the best case scenario.
| v4.20-rc1 | w/ patchset | diff
-----------------------------------------------------------------------
relocated extents | 22615 | 22457 | -0.1%
qgroup dirty extents | 163457 | 121606 | -25.6%
time (sys) | 22.884s | 18.842s | -17.6%
time (real) | 27.724s | 22.884s | -17.5%
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When splitting a leaf or node from one of the trees that are modified when
flushing pending block groups (extent, chunk, device and free space trees),
we need to allocate a new tree block, which in turn can result in the need
to allocate a new block group. After allocating the new block group we may
need to flush new block groups that were previously allocated during the
course of the current transaction, which is what may cause a deadlock due
to attempts to write lock twice the same leaf or node, as when splitting
a leaf or node we are holding a write lock on it and its parent node.
The same type of deadlock can also happen when increasing the tree's
height, since we are holding a lock on the existing root while allocating
the tree block to use as the new root node.
An example trace when the deadlock happens during the leaf split path is:
[27175.293054] CPU: 0 PID: 3005 Comm: kworker/u17:6 Tainted: G W 4.19.16 #1
[27175.293942] Hardware name: Penguin Computing Relion 1900/MD90-FS0-ZB-XX, BIOS R15 06/25/2018
[27175.294846] Workqueue: btrfs-extent-refs btrfs_extent_refs_helper [btrfs]
(...)
[27175.298384] RSP: 0018:ffffab2087107758 EFLAGS: 00010246
[27175.299269] RAX: 0000000000000bbd RBX: ffff9fadc7141c48 RCX: 0000000000000001
[27175.300155] RDX: 0000000000000001 RSI: 0000000000000002 RDI: ffff9fadc7141c48
[27175.301023] RBP: 0000000000000001 R08: ffff9faeb6ac1040 R09: ffff9fa9c0000000
[27175.301887] R10: 0000000000000000 R11: 0000000000000040 R12: ffff9fb21aac8000
[27175.302743] R13: ffff9fb1a64d6a20 R14: 0000000000000001 R15: ffff9fb1a64d6a18
[27175.303601] FS: 0000000000000000(0000) GS:ffff9fb21fa00000(0000) knlGS:0000000000000000
[27175.304468] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[27175.305339] CR2: 00007fdc8743ead8 CR3: 0000000763e0a006 CR4: 00000000003606f0
[27175.306220] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[27175.307087] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[27175.307940] Call Trace:
[27175.308802] btrfs_search_slot+0x779/0x9a0 [btrfs]
[27175.309669] ? update_space_info+0xba/0xe0 [btrfs]
[27175.310534] btrfs_insert_empty_items+0x67/0xc0 [btrfs]
[27175.311397] btrfs_insert_item+0x60/0xd0 [btrfs]
[27175.312253] btrfs_create_pending_block_groups+0xee/0x210 [btrfs]
[27175.313116] do_chunk_alloc+0x25f/0x300 [btrfs]
[27175.313984] find_free_extent+0x706/0x10d0 [btrfs]
[27175.314855] btrfs_reserve_extent+0x9b/0x1d0 [btrfs]
[27175.315707] btrfs_alloc_tree_block+0x100/0x5b0 [btrfs]
[27175.316548] split_leaf+0x130/0x610 [btrfs]
[27175.317390] btrfs_search_slot+0x94d/0x9a0 [btrfs]
[27175.318235] btrfs_insert_empty_items+0x67/0xc0 [btrfs]
[27175.319087] alloc_reserved_file_extent+0x84/0x2c0 [btrfs]
[27175.319938] __btrfs_run_delayed_refs+0x596/0x1150 [btrfs]
[27175.320792] btrfs_run_delayed_refs+0xed/0x1b0 [btrfs]
[27175.321643] delayed_ref_async_start+0x81/0x90 [btrfs]
[27175.322491] normal_work_helper+0xd0/0x320 [btrfs]
[27175.323328] ? move_linked_works+0x6e/0xa0
[27175.324160] process_one_work+0x191/0x370
[27175.324976] worker_thread+0x4f/0x3b0
[27175.325763] kthread+0xf8/0x130
[27175.326531] ? rescuer_thread+0x320/0x320
[27175.327284] ? kthread_create_worker_on_cpu+0x50/0x50
[27175.328027] ret_from_fork+0x35/0x40
[27175.328741] ---[ end trace 300a1b9f0ac30e26 ]---
Fix this by preventing the flushing of new blocks groups when splitting a
leaf/node and when inserting a new root node for one of the trees modified
by the flushing operation, similar to what is done when COWing a node/leaf
from on of these trees.
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=202383
Reported-by: Eli V <eliventer@gmail.com>
CC: stable@vger.kernel.org # 4.4+
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When modifying the free space tree we can end up COWing one of its extent
buffers which in turn might result in allocating a new chunk, which in
turn can result in flushing (finish creation) of pending block groups. If
that happens we can deadlock because creating a pending block group needs
to update the free space tree, and if any of the updates tries to modify
the same extent buffer that we are COWing, we end up in a deadlock since
we try to write lock twice the same extent buffer.
So fix this by skipping pending block group creation if we are COWing an
extent buffer from the free space tree. This is a case missed by commit
5ce555578e ("Btrfs: fix deadlock when writing out free space caches").
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=202173
Fixes: 5ce555578e ("Btrfs: fix deadlock when writing out free space caches")
CC: stable@vger.kernel.org # 4.18+
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The typos accumulate over time so once in a while time they get fixed in
a large patch.
Signed-off-by: Andrea Gelmini <andrea.gelmini@gelma.net>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
If we create a snapshot of a snapshot currently being used by a send
operation, we can end up with send failing unexpectedly (returning
-ENOENT error to user space for example). The following diagram shows
how this happens.
CPU 1 CPU2 CPU3
btrfs_ioctl_send()
(...)
create_snapshot()
-> creates snapshot of a
root used by the send
task
btrfs_commit_transaction()
create_pending_snapshot()
__get_inode_info()
btrfs_search_slot()
btrfs_search_slot_get_root()
down_read commit_root_sem
get reference on eb of the
commit root
-> eb with bytenr == X
up_read commit_root_sem
btrfs_cow_block(root node)
btrfs_free_tree_block()
-> creates delayed ref to
free the extent
btrfs_run_delayed_refs()
-> runs the delayed ref,
adds extent to
fs_info->pinned_extents
btrfs_finish_extent_commit()
unpin_extent_range()
-> marks extent as free
in the free space cache
transaction commit finishes
btrfs_start_transaction()
(...)
btrfs_cow_block()
btrfs_alloc_tree_block()
btrfs_reserve_extent()
-> allocates extent at
bytenr == X
btrfs_init_new_buffer(bytenr X)
btrfs_find_create_tree_block()
alloc_extent_buffer(bytenr X)
find_extent_buffer(bytenr X)
-> returns existing eb,
which the send task got
(...)
-> modifies content of the
eb with bytenr == X
-> uses an eb that now
belongs to some other
tree and no more matches
the commit root of the
snapshot, resuts will be
unpredictable
The consequences of this race can be various, and can lead to searches in
the commit root performed by the send task failing unexpectedly (unable to
find inode items, returning -ENOENT to user space, for example) or not
failing because an inode item with the same number was added to the tree
that reused the metadata extent, in which case send can behave incorrectly
in the worst case or just fail later for some reason.
Fix this by performing a copy of the commit root's extent buffer when doing
a search in the context of a send operation.
CC: stable@vger.kernel.org # 4.4.x: 1fc28d8e2e9: Btrfs: move get root out of btrfs_search_slot to a helper
CC: stable@vger.kernel.org # 4.4.x: f9ddfd0592a: Btrfs: remove unused check of skip_locking
CC: stable@vger.kernel.org # 4.4.x
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When debugging some weird extent reference bug I suspected that we were
changing a snapshot while we were deleting it, which could explain my
bug. This was indeed what was happening, and this patch helped me
verify my theory. It is never correct to modify the snapshot once it's
being deleted, so mark the root when we are deleting it and make sure we
complain about it when it happens.
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently btrfs_fs_info structure contains a copy of the
fsid/metadata_uuid fields. Same values are also contained in the
btrfs_fs_devices structure which fs_info has a reference to. Let's
reduce duplication by removing the fields from fs_info and always refer
to the ones in fs_devices. No functional changes.
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This field is going to be used when the user wants to change the UUID
of the filesystem without having to rewrite all metadata blocks. This
field adds another level of indirection such that when the FSID is
changed what really happens is the current UUID (the one with which the
fs was created) is copied to the 'metadata_uuid' field in the superblock
as well as a new incompat flag is set METADATA_UUID. When the kernel
detects this flag is set it knows that the superblock in fact has 2
UUIDs:
1. Is the UUID which is user-visible, currently known as FSID.
2. Metadata UUID - this is the UUID which is stamped into all on-disk
datastructures belonging to this file system.
When the new incompat flag is present device scanning checks whether
both fsid/metadata_uuid of the scanned device match any of the
registered filesystems. When the flag is not set then both UUIDs are
equal and only the FSID is retained on disk, metadata_uuid is set only
in-memory during mount.
Additionally a new metadata_uuid field is also added to the fs_info
struct. It's initialised either with the FSID in case METADATA_UUID
incompat flag is not set or with the metdata_uuid of the superblock
otherwise.
This commit introduces the new fields as well as the new incompat flag
and switches all users of the fsid to the new logic.
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ minor updates in comments ]
Signed-off-by: David Sterba <dsterba@suse.com>
When the 2 comparison trees roots are initialised they are private to
the function and already have reference counts of 1 each. There is no
need to further increment the reference count since the cloned buffers
are already accessed via struct btrfs_path. Eventually the 2 paths used
for comparison are going to be released, effectively disposing of the
cloned buffers.
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When a rewound buffer is created it already has a ref count of 1 and the
dummy flag set. Then another ref is taken bumping the count to 2.
Finally when this buffer is released from btrfs_release_path the extra
reference is decremented by the special handling code in
free_extent_buffer.
However, this special code is in fact redundant sinca ref count of 1 is
still correct since the buffer is only accessed via btrfs_path struct.
This paves the way forward of removing the special handling in
free_extent_buffer.
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
get_old_root used used only by btrfs_search_old_slot to initialise the
path structure. The old root is always a cloned buffer (either via alloc
dummy or via btrfs_clone_extent_buffer) and its reference count is 2: 1
from allocation, 1 from extent_buffer_get call in get_old_root.
This latter explicit ref count acquire operation is in fact unnecessary
since the semantic is such that the newly allocated buffer is handed
over to the btrfs_path for lifetime management. Considering this just
remove the extra extent_buffer_get in get_old_root.
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When writing out a block group free space cache we can end deadlocking
with ourselves on an extent buffer lock resulting in a warning like the
following:
[245043.379979] WARNING: CPU: 4 PID: 2608 at fs/btrfs/locking.c:251 btrfs_tree_lock+0x1be/0x1d0 [btrfs]
[245043.392792] CPU: 4 PID: 2608 Comm: btrfs-transacti Tainted: G
W I 4.16.8 #1
[245043.395489] RIP: 0010:btrfs_tree_lock+0x1be/0x1d0 [btrfs]
[245043.396791] RSP: 0018:ffffc9000424b840 EFLAGS: 00010246
[245043.398093] RAX: 0000000000000a30 RBX: ffff8807e20a3d20 RCX: 0000000000000001
[245043.399414] RDX: 0000000000000001 RSI: 0000000000000002 RDI: ffff8807e20a3d20
[245043.400732] RBP: 0000000000000001 R08: ffff88041f39a700 R09: ffff880000000000
[245043.402021] R10: 0000000000000040 R11: ffff8807e20a3d20 R12: ffff8807cb220630
[245043.403296] R13: 0000000000000001 R14: ffff8807cb220628 R15: ffff88041fbdf000
[245043.404780] FS: 0000000000000000(0000) GS:ffff88082fc80000(0000) knlGS:0000000000000000
[245043.406050] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[245043.407321] CR2: 00007fffdbdb9f10 CR3: 0000000001c09005 CR4: 00000000000206e0
[245043.408670] Call Trace:
[245043.409977] btrfs_search_slot+0x761/0xa60 [btrfs]
[245043.411278] btrfs_insert_empty_items+0x62/0xb0 [btrfs]
[245043.412572] btrfs_insert_item+0x5b/0xc0 [btrfs]
[245043.413922] btrfs_create_pending_block_groups+0xfb/0x1e0 [btrfs]
[245043.415216] do_chunk_alloc+0x1e5/0x2a0 [btrfs]
[245043.416487] find_free_extent+0xcd0/0xf60 [btrfs]
[245043.417813] btrfs_reserve_extent+0x96/0x1e0 [btrfs]
[245043.419105] btrfs_alloc_tree_block+0xfb/0x4a0 [btrfs]
[245043.420378] __btrfs_cow_block+0x127/0x550 [btrfs]
[245043.421652] btrfs_cow_block+0xee/0x190 [btrfs]
[245043.422979] btrfs_search_slot+0x227/0xa60 [btrfs]
[245043.424279] ? btrfs_update_inode_item+0x59/0x100 [btrfs]
[245043.425538] ? iput+0x72/0x1e0
[245043.426798] write_one_cache_group.isra.49+0x20/0x90 [btrfs]
[245043.428131] btrfs_start_dirty_block_groups+0x102/0x420 [btrfs]
[245043.429419] btrfs_commit_transaction+0x11b/0x880 [btrfs]
[245043.430712] ? start_transaction+0x8e/0x410 [btrfs]
[245043.432006] transaction_kthread+0x184/0x1a0 [btrfs]
[245043.433341] kthread+0xf0/0x130
[245043.434628] ? btrfs_cleanup_transaction+0x4e0/0x4e0 [btrfs]
[245043.435928] ? kthread_create_worker_on_cpu+0x40/0x40
[245043.437236] ret_from_fork+0x1f/0x30
[245043.441054] ---[ end trace 15abaa2aaf36827f ]---
This is because at write_one_cache_group() when we are COWing a leaf from
the extent tree we end up allocating a new block group (chunk) and,
because we have hit a threshold on the number of bytes reserved for system
chunks, we attempt to finalize the creation of new block groups from the
current transaction, by calling btrfs_create_pending_block_groups().
However here we also need to modify the extent tree in order to insert
a block group item, and if the location for this new block group item
happens to be in the same leaf that we were COWing earlier, we deadlock
since btrfs_search_slot() tries to write lock the extent buffer that we
locked before at write_one_cache_group().
We have already hit similar cases in the past and commit d9a0540a79
("Btrfs: fix deadlock when finalizing block group creation") fixed some
of those cases by delaying the creation of pending block groups at the
known specific spots that could lead to a deadlock. This change reworks
that commit to be more generic so that we don't have to add similar logic
to every possible path that can lead to a deadlock. This is done by
making __btrfs_cow_block() disallowing the creation of new block groups
(setting the transaction's can_flush_pending_bgs to false) before it
attempts to allocate a new extent buffer for either the extent, chunk or
device trees, since those are the trees that pending block creation
modifies. Once the new extent buffer is allocated, it allows creation of
pending block groups to happen again.
This change depends on a recent patch from Josef which is not yet in
Linus' tree, named "btrfs: make sure we create all new block groups" in
order to avoid occasional warnings at btrfs_trans_release_chunk_metadata().
Fixes: d9a0540a79 ("Btrfs: fix deadlock when finalizing block group creation")
CC: stable@vger.kernel.org # 4.4+
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=199753
Link: https://lore.kernel.org/linux-btrfs/CAJtFHUTHna09ST-_EEiyWmDH6gAqS6wa=zMNMBsifj8ABu99cw@mail.gmail.com/
Reported-by: E V <eliventer@gmail.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Btrfs's btree locking has two modes, spinning mode and blocking mode,
while searching btree, locking is always acquired in spinning mode and
then converted to blocking mode if necessary, and in some hot paths we may
switch the locking back to spinning mode by btrfs_clear_path_blocking().
When acquiring locks, both of reader and writer need to wait for blocking
readers and writers to complete before doing read_lock()/write_lock().
The problem is that btrfs_clear_path_blocking() needs to switch nodes
in the path to blocking mode at first (by btrfs_set_path_blocking) to
make lockdep happy before doing its actual clearing blocking job.
When switching to blocking mode from spinning mode, it consists of
step 1) bumping up blocking readers counter and
step 2) read_unlock()/write_unlock(),
this has caused serious ping-pong effect if there're a great amount of
concurrent readers/writers, as waiters will be woken up and go to
sleep immediately.
1) Killing this kind of ping-pong results in a big improvement in my 1600k
files creation script,
MNT=/mnt/btrfs
mkfs.btrfs -f /dev/sdf
mount /dev/def $MNT
time fsmark -D 10000 -S0 -n 100000 -s 0 -L 1 -l /tmp/fs_log.txt \
-d $MNT/0 -d $MNT/1 \
-d $MNT/2 -d $MNT/3 \
-d $MNT/4 -d $MNT/5 \
-d $MNT/6 -d $MNT/7 \
-d $MNT/8 -d $MNT/9 \
-d $MNT/10 -d $MNT/11 \
-d $MNT/12 -d $MNT/13 \
-d $MNT/14 -d $MNT/15
w/o patch:
real 2m27.307s
user 0m12.839s
sys 13m42.831s
w/ patch:
real 1m2.273s
user 0m15.802s
sys 8m16.495s
1.1) latency histogram from funclatency[1]
Overall with the patch, there're ~50% less write lock acquisition and
the 95% max latency that write lock takes also reduces to ~100ms from
>500ms.
--------------------------------------------
w/o patch:
--------------------------------------------
Function = btrfs_tree_lock
msecs : count distribution
0 -> 1 : 2385222 |****************************************|
2 -> 3 : 37147 | |
4 -> 7 : 20452 | |
8 -> 15 : 13131 | |
16 -> 31 : 3877 | |
32 -> 63 : 3900 | |
64 -> 127 : 2612 | |
128 -> 255 : 974 | |
256 -> 511 : 165 | |
512 -> 1023 : 13 | |
Function = btrfs_tree_read_lock
msecs : count distribution
0 -> 1 : 6743860 |****************************************|
2 -> 3 : 2146 | |
4 -> 7 : 190 | |
8 -> 15 : 38 | |
16 -> 31 : 4 | |
--------------------------------------------
w/ patch:
--------------------------------------------
Function = btrfs_tree_lock
msecs : count distribution
0 -> 1 : 1318454 |****************************************|
2 -> 3 : 6800 | |
4 -> 7 : 3664 | |
8 -> 15 : 2145 | |
16 -> 31 : 809 | |
32 -> 63 : 219 | |
64 -> 127 : 10 | |
Function = btrfs_tree_read_lock
msecs : count distribution
0 -> 1 : 6854317 |****************************************|
2 -> 3 : 2383 | |
4 -> 7 : 601 | |
8 -> 15 : 92 | |
2) dbench also proves the improvement,
dbench -t 120 -D /mnt/btrfs 16
w/o patch:
Throughput 158.363 MB/sec
w/ patch:
Throughput 449.52 MB/sec
3) xfstests didn't show any additional failures.
One thing to note is that callers may set path->leave_spinning to have
all nodes in the path stay in spinning mode, which means callers are
ready to not sleep before releasing the path, but it won't cause
problems if they don't want to sleep in blocking mode.
[1]: https://github.com/iovisor/bcc/blob/master/tools/funclatency.py
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Signed-off-by: David Sterba <dsterba@suse.com>
In btrfs_search_old_slot get_old_root is always used with the assumption
it cannot fail. However, this is not true in rare circumstance it can
fail and return null. This will lead to null point dereference when the
header is read. Fix this by checking the return value and properly
handling NULL by setting ret to -EIO and returning gracefully.
Coverity-id: 1087503
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
In the callchain:
btrfs_search_slot()
if (level != 0)
setup_nodes_for_search()
balance_level()
It is just impossible to have level=0 in balance_level, we can drop the
check.
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
As we're going to return right after the call, it's not necessary to get
update the new write_lock_level from unlock_up.
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
There are two members in struct btrfs_root which indicate root's
objectid: objectid and root_key.objectid.
They are both set to the same value in __setup_root():
static void __setup_root(struct btrfs_root *root,
struct btrfs_fs_info *fs_info,
u64 objectid)
{
...
root->objectid = objectid;
...
root->root_key.objectid = objecitd;
...
}
and not changed to other value after initialization.
grep in btrfs directory shows both are used in many places:
$ grep -rI "root->root_key.objectid" | wc -l
133
$ grep -rI "root->objectid" | wc -l
55
(4.17, inc. some noise)
It is confusing to have two similar variable names and it seems
that there is no rule about which should be used in a certain case.
Since ->root_key itself is needed for tree reloc tree, let's remove
'objecitd' member and unify code to use ->root_key.objectid in all places.
Signed-off-by: Misono Tomohiro <misono.tomohiro@jp.fujitsu.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The v0 compat code was introduced in commit 5d4f98a28c
("Btrfs: Mixed back reference (FORWARD ROLLING FORMAT CHANGE)") 9
years ago, which was merged in 2.6.31. This means that the code is
there to support filesystems which are _VERY_ old and if you are using
btrfs on such an old kernel, you have much bigger problems. This coupled
with the fact that no one is likely testing/maintining this code likely
means it has bugs lurking. All things considered I think 43 kernel
releases later it's high time this remnant of the past got removed.
This patch removes all code wrapped in #ifdefs but leaves the BUG_ONs in case
we have a v0 with no support intact as a sort of safety-net.
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When a new extent buffer is allocated there are a few mandatory fields
which need to be set in order for the buffer to be sane: level,
generation, bytenr, backref_rev, owner and FSID/UUID. Currently this
is open coded in the callers of btrfs_alloc_tree_block, meaning it's
fairly high in the abstraction hierarchy of operations. This patch
solves this by simply moving this init code in btrfs_init_new_buffer,
since this is the function which initializes a newly allocated
extent buffer. No functional changes.
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This argument is unused. No functional changes.
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The check is superfluous since all callers who set search_for_commit
also have skip_locking set.
ASSERT() is put in place to ensure skip_locking is set by new callers.
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Signed-off-by: David Sterba <dsterba@suse.com>
As unlock_up() is written as
for () {
if (!path->locks[i])
break;
...
if (... && path->locks[i]) {
}
}
Apparently, @path->locks[i] is always true at this 'if'.
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Typically, when acquiring root node's lock, btrfs tries its best to get
read lock and trade for write lock if @write_lock_level implies to do so.
In case of (cow && (p->keep_locks || p->lowest_level)), write_lock_level
is set to BTRFS_MAX_LEVEL, which means we need to acquire root node's
write lock directly.
In this particular case, the dance of acquiring read lock and then trading
for write lock can be saved.
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
It's good to have a helper instead of having all get-root details
open-coded. The new helper locks (if necessary) and sets root node of
the path.
Also invert the checks to make the code flow easier to read. There is
no functional change in this commit.
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
If parent_transid "0" is passed to btrfs_buffer_uptodate(),
btrfs_buffer_uptodate() is equivalent to extent_buffer_uptodate(), but
extent_buffer_uptodate() is preferred since we don't have to look into
verify_parent_transid().
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
read_block_for_search() can be simplified as:
tmp = find_extent_buffer();
if (tmp)
return;
...
free_extent_buffer();
read_tree_block();
Apparently, @tmp must be NULL at this point, free_extent_buffer() is not
needed.
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
If a btree block, aka. extent buffer, is not available in the extent
buffer cache, it'll be read out from the disk instead, i.e.
btrfs_search_slot()
read_block_for_search() # hold parent and its lock, go to read child
btrfs_release_path()
read_tree_block() # read child
Unfortunately, the parent lock got released before reading child, so
commit 5bdd3536cb ("Btrfs: Fix block generation verification race") had
used 0 as parent transid to read the child block. It forces
read_tree_block() not to check if parent transid is different with the
generation id of the child that it reads out from disk.
A simple PoC is included in btrfs/124,
0. A two-disk raid1 btrfs,
1. Right after mkfs.btrfs, block A is allocated to be device tree's root.
2. Mount this filesystem and put it in use, after a while, device tree's
root got COW but block A hasn't been allocated/overwritten yet.
3. Umount it and reload the btrfs module to remove both disks from the
global @fs_devices list.
4. mount -odegraded dev1 and write some data, so now block A is allocated
to be a leaf in checksum tree. Note that only dev1 has the latest
metadata of this filesystem.
5. Umount it and mount it again normally (with both disks), since raid1
can pick up one disk by the writer task's pid, if btrfs_search_slot()
needs to read block A, dev2 which does NOT have the latest metadata
might be read for block A, then we got a stale block A.
6. As parent transid is not checked, block A is marked as uptodate and
put into the extent buffer cache, so the future search won't bother
to read disk again, which means it'll make changes on this stale
one and make it dirty and flush it onto disk.
To avoid the problem, parent transid needs to be passed to
read_tree_block().
In order to get a valid parent transid, we need to hold the parent's
lock until finishing reading child.
This patch needs to be slightly adapted for stable kernels, the
&first_key parameter added to read_tree_block() is from 4.16+
(581c176041). The fix is to replace 0 by 'gen'.
Fixes: 5bdd3536cb ("Btrfs: Fix block generation verification race")
CC: stable@vger.kernel.org # 4.4+
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
[ update changelog ]
Signed-off-by: David Sterba <dsterba@suse.com>
[BUG]
btrfs incremental send BUG happens when creating a snapshot of snapshot
that is being used by send.
[REASON]
The problem can happen if while we are doing a send one of the snapshots
used (parent or send) is snapshotted, because snapshoting implies COWing
the root of the source subvolume/snapshot.
1. When doing an incremental send, the send process will get the commit
roots from the parent and send snapshots, and add references to them
through extent_buffer_get().
2. When a snapshot/subvolume is snapshotted, its root node is COWed
(transaction.c:create_pending_snapshot()).
3. COWing releases the space used by the node immediately, through:
__btrfs_cow_block()
--btrfs_free_tree_block()
----btrfs_add_free_space(bytenr of node)
4. Because send doesn't hold a transaction open, it's possible that
the transaction used to create the snapshot commits, switches the
commit root and the old space used by the previous root node gets
assigned to some other node allocation. Allocation of a new node will
use the existing extent buffer found in memory, which we previously
got a reference through extent_buffer_get(), and allow the extent
buffer's content (pages) to be modified:
btrfs_alloc_tree_block
--btrfs_reserve_extent
----find_free_extent (get bytenr of old node)
--btrfs_init_new_buffer (use bytenr of old node)
----btrfs_find_create_tree_block
------alloc_extent_buffer
--------find_extent_buffer (get old node)
5. So send can access invalid memory content and have unpredictable
behaviour.
[FIX]
So we fix the problem by copying the commit roots of the send and
parent snapshots and use those copies.
CallTrace looks like this:
------------[ cut here ]------------
kernel BUG at fs/btrfs/ctree.c:1861!
invalid opcode: 0000 [#1] SMP
CPU: 6 PID: 24235 Comm: btrfs Tainted: P O 3.10.105 #23721
ffff88046652d680 ti: ffff88041b720000 task.ti: ffff88041b720000
RIP: 0010:[<ffffffffa08dd0e8>] read_node_slot+0x108/0x110 [btrfs]
RSP: 0018:ffff88041b723b68 EFLAGS: 00010246
RAX: ffff88043ca6b000 RBX: ffff88041b723c50 RCX: ffff880000000000
RDX: 000000000000004c RSI: ffff880314b133f8 RDI: ffff880458b24000
RBP: 0000000000000000 R08: 0000000000000001 R09: ffff88041b723c66
R10: 0000000000000001 R11: 0000000000001000 R12: ffff8803f3e48890
R13: ffff8803f3e48880 R14: ffff880466351800 R15: 0000000000000001
FS: 00007f8c321dc8c0(0000) GS:ffff88047fcc0000(0000)
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
R2: 00007efd1006d000 CR3: 0000000213a24000 CR4: 00000000003407e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Stack:
ffff88041b723c50 ffff8803f3e48880 ffff8803f3e48890 ffff8803f3e48880
ffff880466351800 0000000000000001 ffffffffa08dd9d7 ffff88041b723c50
ffff8803f3e48880 ffff88041b723c66 ffffffffa08dde85 a9ff88042d2c4400
Call Trace:
[<ffffffffa08dd9d7>] ? tree_move_down.isra.33+0x27/0x50 [btrfs]
[<ffffffffa08dde85>] ? tree_advance+0xb5/0xc0 [btrfs]
[<ffffffffa08e83d4>] ? btrfs_compare_trees+0x2d4/0x760 [btrfs]
[<ffffffffa0982050>] ? finish_inode_if_needed+0x870/0x870 [btrfs]
[<ffffffffa09841ea>] ? btrfs_ioctl_send+0xeda/0x1050 [btrfs]
[<ffffffffa094bd3d>] ? btrfs_ioctl+0x1e3d/0x33f0 [btrfs]
[<ffffffff81111133>] ? handle_pte_fault+0x373/0x990
[<ffffffff8153a096>] ? atomic_notifier_call_chain+0x16/0x20
[<ffffffff81063256>] ? set_task_cpu+0xb6/0x1d0
[<ffffffff811122c3>] ? handle_mm_fault+0x143/0x2a0
[<ffffffff81539cc0>] ? __do_page_fault+0x1d0/0x500
[<ffffffff81062f07>] ? check_preempt_curr+0x57/0x90
[<ffffffff8115075a>] ? do_vfs_ioctl+0x4aa/0x990
[<ffffffff81034f83>] ? do_fork+0x113/0x3b0
[<ffffffff812dd7d7>] ? trace_hardirqs_off_thunk+0x3a/0x6c
[<ffffffff81150cc8>] ? SyS_ioctl+0x88/0xa0
[<ffffffff8153e422>] ? system_call_fastpath+0x16/0x1b
---[ end trace 29576629ee80b2e1 ]---
Fixes: 7069830a9e ("Btrfs: add btrfs_compare_trees function")
CC: stable@vger.kernel.org # 3.6+
Signed-off-by: Robbie Ko <robbieko@synology.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Remove GPL boilerplate text (long, short, one-line) and keep the rest,
ie. personal, company or original source copyright statements. Add the
SPDX header.
Signed-off-by: David Sterba <dsterba@suse.com>
Once there was a simple int force_cow that was used with the plain
barriers, and then converted to a bit, so we should use the appropriate
barrier helper.
Other variables in the complex if condition do not depend on a barrier,
so we should be fine in case the atomic barrier becomes a no-op.
Signed-off-by: David Sterba <dsterba@suse.com>
We have several reports about node pointer points to incorrect child
tree blocks, which could have even wrong owner and level but still with
valid generation and checksum.
Although btrfs check could handle it and print error message like:
leaf parent key incorrect 60670574592
Kernel doesn't have enough check on this type of corruption correctly.
At least add such check to read_tree_block() and btrfs_read_buffer(),
where we need two new parameters @level and @first_key to verify the
child tree block.
The new @level check is mandatory and all call sites are already
modified to extract expected level from its call chain.
While @first_key is optional, the following call sites are skipping such
check:
1) Root node/leaf
As ROOT_ITEM doesn't contain the first key, skip @first_key check.
2) Direct backref
Only parent bytenr and level is known and we need to resolve the key
all by ourselves, skip @first_key check.
Another note of this verification is, it needs extra info from nodeptr
or ROOT_ITEM, so it can't fit into current tree-checker framework, which
is limited to node/leaf boundary.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
A useless wrapper around tree_mod_log_insert_root that hides missing
error handling. Move it to the callers.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
A trivial wrapper that can be simply opencoded and makes the GFP
allocation request more visible. The error handling is now moved to the
callers.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The wrapper is effectively an alias for tree_mod_log_insert_move but
also hides the missing error handling. To make that more visible, lift
the BUG_ON to the callers.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The wrappers are trivial and do not bring any extra value on top of the
plain locking primitives.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This function always sets keep_locks to 1 and saves the old value of
keep_locks which is restored at the end. So there is no way it can be
called without keep_locks being set. Remove comment imposing redundant
requirement on callers.
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Filipe Manana