When trying to release the extent states due to memory pressure we'll
set all the bits except LOCKED, NODATASUM, and DELALLOC_NEW. This
includes some of the CTL bits, which isn't really a problem but isn't
correct either. Exclude the CTL bits from this clearing.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This was used as an optimization for count_range_bits(EXTENT_DIRTY),
which was used by the failed record code. However this was removed in
this series by patch "btrfs: convert the io_failure_tree to a plain
rb_tree" which was the last user of this optimization. Remove the
->dirty_bytes as nobody cares anymore.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Since commit 78361f64ff42 ("btrfs: remove unnecessary EXTENT_UPTODATE
state in buffered I/O path") we no longer check ->track_uptodate, remove
it.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We have two variants of lock/unlock extent, one set that takes a cached
state, another that does not. This is slightly annoying, and generally
speaking there are only a few places where we don't have a cached state.
Simplify this by making lock_extent/unlock_extent the only variant and
make it take a cached state, then convert all the callers appropriately.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The only places that set extent_changeset is set_record_extent_bits,
everywhere else sets it to NULL. Drop this argument from
set_extent_bit.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This is only used for internal locking related helpers, everybody else
just passes in NULL. I've changed set_extent_bit to __set_extent_bit
and made it static, removed failed_start from set_extent_bit and have it
call __set_extent_bit with a NULL failed_start, and I've moved some code
down below the now static __set_extent_bit.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This is only used in the case that we are clearing EXTENT_LOCKED, so
infer this value from the bits passed in instead of taking it as an
argument.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This is only ever set if we have EXTENT_LOCKED set, so simply push this
into the function itself and remove the function argument.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
These prototypes have nothing to do with the extent_io_tree helpers,
move them to their appropriate header.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We use rb_next/rb_prev and then get the entry for the adjacent items in
an extent io tree. We have helpers for this, so convert merge_state to
use next_state/prev_state and simplify the code.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Instead of doing the rb_entry again once we return from this function,
simply return the actual states themselves, and then clean up the only
user of this helper to handle states instead of nodes.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We use this to search for an extent state, or return the nodes we need
to insert a new extent state. This means we have the following pattern
node = tree_search_for_insert();
if (!node) {
/* alloc and insert. */
goto again;
}
state = rb_entry(node, struct extent_state, rb_node);
we don't use the node for anything else. Making
tree_search_for_insert() return the extent_state means we can drop the
rb_node and clean this up by eliminating the rb_entry.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We have a consistent pattern of
n = tree_search();
if (!n)
goto out;
state = rb_entry(n, struct extent_state, rb_node);
while (state) {
/* do something. */
}
which is a bit redundant. If we make tree_search return the state we
can simply have
state = tree_search();
while (state) {
/* do something. */
}
which cleans up the code quite a bit.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We can simplify a lot of these functions where we have to cycle through
extent_state's by simply using next_state() instead of rb_next(). In
many spots this allows us to do things like
while (state) {
/* whatever */
state = next_state(state);
}
instead of
while (1) {
state = rb_entry(n, struct extent_state, rb_node);
n = rb_next(n);
if (!n)
break;
}
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This existed when we overloaded the tree manipulation functions for both
the extent_io_tree and the extent buffer tree. However the extent
buffers are now stored in a radix tree, so we no longer need this
abstraction. Remove struct tree_entry and use extent_state directly
instead.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Now that we've moved everything we can unexport all the temporary
exports, move the random helpers, and mark everything as static again.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We no longer need to export this as all users are in extent-io-tree.c,
remove it from the header and put it into extent-io-tree.c.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This is still huge, but unfortunately I cannot make it smaller without
renaming tree_search() and changing all the callers to use the new name,
then moving those chunks and then changing the name back. This feels
like too much churn for code movement, so I've limited this to only
things that called tree_search(). With this patch all of the
extent_io_tree code is now in extent-io-tree.c.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
These are the last few helpers that do not rely on tree_search() and
who's other helpers are exported and in extent-io-tree.c already. Move
these across now in order to make the core move smaller.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
In order to avoid moving all of the related code at once temporarily
export all of the extent state related helpers. Then move these helpers
into extent-io-tree.c. We will clean up the exports and make them
static in followup patches.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
A lot of the various internals of extent_io_tree call these two
functions for insert or searching the rb tree for entries, so
temporarily export them and then move them to extent-io-tree.c. We
can't move tree_search() without renaming it, and I don't want to
introduce a bunch of churn just to do that, so move these functions
first and then we can move a few big functions and then the remaining
users of tree_search().
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This helper is used by a lot of the core extent_io_tree helpers, so
temporarily export it and move it into extent-io-tree.c in order to make
it straightforward to migrate the helpers in batches.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This is used by the subpage code in addition to lock_extent_bits, so
export it so we can move it out of extent_io.c
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
These are just variants and wrappers around the actual work horses of
the extent state. Extract these out of extent_io.c.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We only call these functions from the qgroup code which doesn't call
with EXTENT_BIT_LOCKED. These are BUG_ON()'s that exist to keep us
developers from using these functions with EXTENT_BIT_LOCKED, so convert
them to ASSERT()'s.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Start cleaning up extent_io.c by moving the extent state code out of it.
This patch starts with the extent state allocation code and the
extent_io_tree init code.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We're going to move this code in stages, but while we're doing that we
need to export these helpers so we can more easily move the code into
the new file.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently we have the add/del functions generic so that we can use them
for both extent buffers and extent states. We want to separate this
code however, so separate these helpers into per-object helpers in
anticipation of the split.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
In order to help separate the extent buffer from the extent io tree code
we need to break up the init functions.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently we're using find_first_extent_bit_state to check if our state
contains the given failrec range, however this is more of an internal
extent_io_tree helper, and is technically unsafe to use because we're
accessing the state outside of the extent_io_tree lock.
Instead use the normal helper find_first_extent_bit which returns the
range of the extent state we find in find_first_extent_bit_state and use
that to do our sanity checking.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We still have this oddity of stashing the io_failure_record in the
extent state for the io_failure_tree, which is leftover from when we
used to stuff private pointers in extent_io_trees.
However this doesn't make a lot of sense for the io failure records, we
can simply use a normal rb_tree for this. This will allow us to further
simplify the extent_io_tree code by removing the io_failure_rec pointer
from the extent state.
Convert the io_failure_tree to an rb tree + spinlock in the inode, and
then use our rb tree simple helpers to insert and find failed records.
This greatly cleans up this code and makes it easier to separate out the
extent_io_tree code.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
These are internally used functions and are not used outside of
extent_io.c.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This is exported, so rename it to btrfs_clean_io_failure. Additionally
we are passing in the io tree's and such from the inode, so instead of
doing all that simply pass in the inode itself and get all the
components we need directly inside of btrfs_clean_io_failure.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
KCSAN reports that there's unlocked access mixed with locked access,
which is technically correct but is not a bug. To avoid false alerts at
least from KCSAN, add annotation and use a wrapper whenever ->full is
accessed for read outside of lock.
It is used as a fast check and only advisory. In the worst case the
block reserve is found !full and becomes full in the meantime, but
properly handled.
Depending on the value of ->full, btrfs_block_rsv_release decides
where to return the reservation, and block_rsv_release_bytes handles a
NULL pointer for block_rsv and if it's not NULL then it double checks
the full status under a lock.
Link: https://lore.kernel.org/linux-btrfs/CAAwBoOJDjei5Hnem155N_cJwiEkVwJYvgN-tQrwWbZQGhFU=cA@mail.gmail.com/
Link: https://lore.kernel.org/linux-btrfs/YvHU/vsXd7uz5V6j@hungrycats.org
Reported-by: Zygo Blaxell <ce3g8jdj@umail.furryterror.org>
Signed-off-by: David Sterba <dsterba@suse.com>
At space-info.c:__reserve_bytes(), we increment the 'used' variable, but
then we don't use the variable anymore, making the increment pointless.
The increment became useless with commit 2e294c6049 ("btrfs: simplify
the logic in need_preemptive_flushing"), so just remove it.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
btrfs_check_zoned_mode is really hard to follow, mostly due to the
fact that a lot of the checks use duplicate conditions after support
for zone emulation for conventional devices on file systems with the
ZONED flag was added. Fix this by factoring out the check for host
managed devices for !ZONED file systems into a separate helper and
then simplifying the rest of the code.
Reviewed-by: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David Sterba <dsterba@suse.com>
Add a missing 'r'. s/qgoup/qgroup/ . Codespell does not catch that for
some reason.
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
btrfs_bit_radix_cachep has been removed since
commit 45c06543af ("Btrfs: remove unused btrfs_bit_radix slab"),
so remove it.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Gaosheng Cui <cuigaosheng1@huawei.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Btrfs qgroup has a long history of bringing performance penalty in
btrfs_commit_transaction().
Although we tried our best to migrate such impact, there is still an
unsolved call site, btrfs_drop_snapshot().
This function will find the highest shared tree block and modify its
extent ownership to do a subvolume/snapshot dropping.
Such change will affect the whole subtree, and cause tons of qgroup
dirty extents and stall btrfs_commit_transaction().
To avoid such problem, here we introduce a new sysfs interface,
/sys/fs/btrfs/<uuid>/qgroups/drop_subptree_threshold, to determine at
whether and at which level we should skip qgroup accounting for subtree
dropping.
The default value is BTRFS_MAX_LEVEL, thus every subtree drop will go
through qgroup accounting, to ensure qgroup numbers are kept as
consistent as possible.
While for performance sensitive cases, add a way to change the values to
more reasonable values like 3, to make any subtree, which is at or higher
than level 3, to mark qgroup inconsistent and skip the accounting.
The cost is obvious, the qgroup number is no longer consistent, but at
least performance is more reasonable, and users have the control.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The new flag will make btrfs qgroup skip all its time consuming
qgroup accounting.
The lifespan is the same as BTRFS_QGROUP_RUNTIME_FLAG_CANCEL_RESCAN,
only get cleared after a new rescan.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Introduce a new runtime flag, BTRFS_QGROUP_RUNTIME_FLAG_CANCEL_RESCAN,
which will inform qgroup rescan to cancel its work asynchronously.
This is to address the window when an operation makes qgroup numbers
inconsistent (like qgroup inheriting) while a qgroup rescan is running.
In that case, qgroup inconsistent flag will be cleared when qgroup
rescan finishes.
But we changed the ownership of some extents, which means the rescan is
already meaningless, and the qgroup inconsistent flag should not be
cleared.
With the new flag, each time we set INCONSISTENT flag, we also set this
new flag to inform any running qgroup rescan to exit immediately, and
leaving the INCONSISTENT flag there.
The new runtime flag can only be cleared when a new rescan is started.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently we only have 3 qgroup flags:
- BTRFS_QGROUP_STATUS_FLAG_ON
- BTRFS_QGROUP_STATUS_FLAG_RESCAN
- BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT
These flags match the on-disk flags used in btrfs_qgroup_status.
But we're going to introduce extra runtime flags which will not reach
disks.
So here we introduce a new mask, BTRFS_QGROUP_STATUS_FLAGS_MASK, to
make sure only those flags can reach disks.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Although we already have info kobject for each qgroup, we don't have
global qgroup info attributes to show things like enabled or
inconsistent status flags.
Add this qgroups attribute groups, and the first member is qgroup_flags,
which is a read-only attribute to show human readable qgroup flags.
The path is:
/sys/fs/btrfs/<uuid>/qgroups/enabled
/sys/fs/btrfs/<uuid>/qgroups/inconsistent
The output is simple, just 1 or 0.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The current fiemap implementation does not scale very well with the number
of extents a file has. This is both because the main algorithm to find out
the extents has a high algorithmic complexity and because for each extent
we have to check if it's shared. This second part, checking if an extent
is shared, is significantly improved by the two previous patches in this
patchset, while the first part is improved by this specific patch. Every
now and then we get reports from users mentioning fiemap is too slow or
even unusable for files with a very large number of extents, such as the
two recent reports referred to by the Link tags at the bottom of this
change log.
To understand why the part of finding which extents a file has is very
inefficient, consider the example of doing a full ranged fiemap against
a file that has over 100K extents (normal for example for a file with
more than 10G of data and using compression, which limits the extent size
to 128K). When we enter fiemap at extent_fiemap(), the following happens:
1) Before entering the main loop, we call get_extent_skip_holes() to get
the first extent map. This leads us to btrfs_get_extent_fiemap(), which
in turn calls btrfs_get_extent(), to find the first extent map that
covers the file range [0, LLONG_MAX).
btrfs_get_extent() will first search the inode's extent map tree, to
see if we have an extent map there that covers the range. If it does
not find one, then it will search the inode's subvolume b+tree for a
fitting file extent item. After finding the file extent item, it will
allocate an extent map, fill it in with information extracted from the
file extent item, and add it to the inode's extent map tree (which
requires a search for insertion in the tree).
2) Then we enter the main loop at extent_fiemap(), emit the details of
the extent, and call again get_extent_skip_holes(), with a start
offset matching the end of the extent map we previously processed.
We end up at btrfs_get_extent() again, will search the extent map tree
and then search the subvolume b+tree for a file extent item if we could
not find an extent map in the extent tree. We allocate an extent map,
fill it in with the details in the file extent item, and then insert
it into the extent map tree (yet another search in this tree).
3) The second step is repeated over and over, until we have processed the
whole file range. Each iteration ends at btrfs_get_extent(), which
does a red black tree search on the extent map tree, then searches the
subvolume b+tree, allocates an extent map and then does another search
in the extent map tree in order to insert the extent map.
In the best scenario we have all the extent maps already in the extent
tree, and so for each extent we do a single search on a red black tree,
so we have a complexity of O(n log n).
In the worst scenario we don't have any extent map already loaded in
the extent map tree, or have very few already there. In this case the
complexity is much higher since we do:
- A red black tree search on the extent map tree, which has O(log n)
complexity, initially very fast since the tree is empty or very
small, but as we end up allocating extent maps and adding them to
the tree when we don't find them there, each subsequent search on
the tree gets slower, since it's getting bigger and bigger after
each iteration.
- A search on the subvolume b+tree, also O(log n) complexity, but it
has items for all inodes in the subvolume, not just items for our
inode. Plus on a filesystem with concurrent operations on other
inodes, we can block doing the search due to lock contention on
b+tree nodes/leaves.
- Allocate an extent map - this can block, and can also fail if we
are under serious memory pressure.
- Do another search on the extent maps red black tree, with the goal
of inserting the extent map we just allocated. Again, after every
iteration this tree is getting bigger by 1 element, so after many
iterations the searches are slower and slower.
- We will not need the allocated extent map anymore, so it's pointless
to add it to the extent map tree. It's just wasting time and memory.
In short we end up searching the extent map tree multiple times, on a
tree that is growing bigger and bigger after each iteration. And
besides that we visit the same leaf of the subvolume b+tree many times,
since a leaf with the default size of 16K can easily have more than 200
file extent items.
This is very inefficient overall. This patch changes the algorithm to
instead iterate over the subvolume b+tree, visiting each leaf only once,
and only searching in the extent map tree for file ranges that have holes
or prealloc extents, in order to figure out if we have delalloc there.
It will never allocate an extent map and add it to the extent map tree.
This is very similar to what was previously done for the lseek's hole and
data seeking features.
Also, the current implementation relying on extent maps for figuring out
which extents we have is not correct. This is because extent maps can be
merged even if they represent different extents - we do this to minimize
memory utilization and keep extent map trees smaller. For example if we
have two extents that are contiguous on disk, once we load the two extent
maps, they get merged into a single one - however if only one of the
extents is shared, we end up reporting both as shared or both as not
shared, which is incorrect.
This reproducer triggers that bug:
$ cat fiemap-bug.sh
#!/bin/bash
DEV=/dev/sdj
MNT=/mnt/sdj
mkfs.btrfs -f $DEV
mount $DEV $MNT
# Create a file with two 256K extents.
# Since there is no other write activity, they will be contiguous,
# and their extent maps merged, despite having two distinct extents.
xfs_io -f -c "pwrite -S 0xab 0 256K" \
-c "fsync" \
-c "pwrite -S 0xcd 256K 256K" \
-c "fsync" \
$MNT/foo
# Now clone only the second extent into another file.
xfs_io -f -c "reflink $MNT/foo 256K 0 256K" $MNT/bar
# Filefrag will report a single 512K extent, and say it's not shared.
echo
filefrag -v $MNT/foo
umount $MNT
Running the reproducer:
$ ./fiemap-bug.sh
wrote 262144/262144 bytes at offset 0
256 KiB, 64 ops; 0.0038 sec (65.479 MiB/sec and 16762.7030 ops/sec)
wrote 262144/262144 bytes at offset 262144
256 KiB, 64 ops; 0.0040 sec (61.125 MiB/sec and 15647.9218 ops/sec)
linked 262144/262144 bytes at offset 0
256 KiB, 1 ops; 0.0002 sec (1.034 GiB/sec and 4237.2881 ops/sec)
Filesystem type is: 9123683e
File size of /mnt/sdj/foo is 524288 (128 blocks of 4096 bytes)
ext: logical_offset: physical_offset: length: expected: flags:
0: 0.. 127: 3328.. 3455: 128: last,eof
/mnt/sdj/foo: 1 extent found
We end up reporting that we have a single 512K that is not shared, however
we have two 256K extents, and the second one is shared. Changing the
reproducer to clone instead the first extent into file 'bar', makes us
report a single 512K extent that is shared, which is algo incorrect since
we have two 256K extents and only the first one is shared.
This patch is part of a larger patchset that is comprised of the following
patches:
btrfs: allow hole and data seeking to be interruptible
btrfs: make hole and data seeking a lot more efficient
btrfs: remove check for impossible block start for an extent map at fiemap
btrfs: remove zero length check when entering fiemap
btrfs: properly flush delalloc when entering fiemap
btrfs: allow fiemap to be interruptible
btrfs: rename btrfs_check_shared() to a more descriptive name
btrfs: speedup checking for extent sharedness during fiemap
btrfs: skip unnecessary extent buffer sharedness checks during fiemap
btrfs: make fiemap more efficient and accurate reporting extent sharedness
The patchset was tested on a machine running a non-debug kernel (Debian's
default config) and compared the tests below on a branch without the
patchset versus the same branch with the whole patchset applied.
The following test for a large compressed file without holes:
$ cat fiemap-perf-test.sh
#!/bin/bash
DEV=/dev/sdi
MNT=/mnt/sdi
mkfs.btrfs -f $DEV
mount -o compress=lzo $DEV $MNT
# 40G gives 327680 128K file extents (due to compression).
xfs_io -f -c "pwrite -S 0xab -b 1M 0 20G" $MNT/foobar
umount $MNT
mount -o compress=lzo $DEV $MNT
start=$(date +%s%N)
filefrag $MNT/foobar
end=$(date +%s%N)
dur=$(( (end - start) / 1000000 ))
echo "fiemap took $dur milliseconds (metadata not cached)"
start=$(date +%s%N)
filefrag $MNT/foobar
end=$(date +%s%N)
dur=$(( (end - start) / 1000000 ))
echo "fiemap took $dur milliseconds (metadata cached)"
umount $MNT
Before patchset:
$ ./fiemap-perf-test.sh
(...)
/mnt/sdi/foobar: 327680 extents found
fiemap took 3597 milliseconds (metadata not cached)
/mnt/sdi/foobar: 327680 extents found
fiemap took 2107 milliseconds (metadata cached)
After patchset:
$ ./fiemap-perf-test.sh
(...)
/mnt/sdi/foobar: 327680 extents found
fiemap took 1214 milliseconds (metadata not cached)
/mnt/sdi/foobar: 327680 extents found
fiemap took 684 milliseconds (metadata cached)
That's a speedup of about 3x for both cases (no metadata cached and all
metadata cached).
The test provided by Pavel (first Link tag at the bottom), which uses
files with a large number of holes, was also used to measure the gains,
and it consists on a small C program and a shell script to invoke it.
The C program is the following:
$ cat pavels-test.c
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <fcntl.h>
#include <sys/stat.h>
#include <sys/time.h>
#include <sys/ioctl.h>
#include <linux/fs.h>
#include <linux/fiemap.h>
#define FILE_INTERVAL (1<<13) /* 8Kb */
long long interval(struct timeval t1, struct timeval t2)
{
long long val = 0;
val += (t2.tv_usec - t1.tv_usec);
val += (t2.tv_sec - t1.tv_sec) * 1000 * 1000;
return val;
}
int main(int argc, char **argv)
{
struct fiemap fiemap = {};
struct timeval t1, t2;
char data = 'a';
struct stat st;
int fd, off, file_size = FILE_INTERVAL;
if (argc != 3 && argc != 2) {
printf("usage: %s <path> [size]\n", argv[0]);
return 1;
}
if (argc == 3)
file_size = atoi(argv[2]);
if (file_size < FILE_INTERVAL)
file_size = FILE_INTERVAL;
file_size -= file_size % FILE_INTERVAL;
fd = open(argv[1], O_RDWR | O_CREAT | O_TRUNC, 0644);
if (fd < 0) {
perror("open");
return 1;
}
for (off = 0; off < file_size; off += FILE_INTERVAL) {
if (pwrite(fd, &data, 1, off) != 1) {
perror("pwrite");
close(fd);
return 1;
}
}
if (ftruncate(fd, file_size)) {
perror("ftruncate");
close(fd);
return 1;
}
if (fstat(fd, &st) < 0) {
perror("fstat");
close(fd);
return 1;
}
printf("size: %ld\n", st.st_size);
printf("actual size: %ld\n", st.st_blocks * 512);
fiemap.fm_length = FIEMAP_MAX_OFFSET;
gettimeofday(&t1, NULL);
if (ioctl(fd, FS_IOC_FIEMAP, &fiemap) < 0) {
perror("fiemap");
close(fd);
return 1;
}
gettimeofday(&t2, NULL);
printf("fiemap: fm_mapped_extents = %d\n",
fiemap.fm_mapped_extents);
printf("time = %lld us\n", interval(t1, t2));
close(fd);
return 0;
}
$ gcc -o pavels_test pavels_test.c
And the wrapper shell script:
$ cat fiemap-pavels-test.sh
#!/bin/bash
DEV=/dev/sdi
MNT=/mnt/sdi
mkfs.btrfs -f -O no-holes $DEV
mount $DEV $MNT
echo
echo "*********** 256M ***********"
echo
./pavels-test $MNT/testfile $((1 << 28))
echo
./pavels-test $MNT/testfile $((1 << 28))
echo
echo "*********** 512M ***********"
echo
./pavels-test $MNT/testfile $((1 << 29))
echo
./pavels-test $MNT/testfile $((1 << 29))
echo
echo "*********** 1G ***********"
echo
./pavels-test $MNT/testfile $((1 << 30))
echo
./pavels-test $MNT/testfile $((1 << 30))
umount $MNT
Running his reproducer before applying the patchset:
*********** 256M ***********
size: 268435456
actual size: 134217728
fiemap: fm_mapped_extents = 32768
time = 4003133 us
size: 268435456
actual size: 134217728
fiemap: fm_mapped_extents = 32768
time = 4895330 us
*********** 512M ***********
size: 536870912
actual size: 268435456
fiemap: fm_mapped_extents = 65536
time = 30123675 us
size: 536870912
actual size: 268435456
fiemap: fm_mapped_extents = 65536
time = 33450934 us
*********** 1G ***********
size: 1073741824
actual size: 536870912
fiemap: fm_mapped_extents = 131072
time = 224924074 us
size: 1073741824
actual size: 536870912
fiemap: fm_mapped_extents = 131072
time = 217239242 us
Running it after applying the patchset:
*********** 256M ***********
size: 268435456
actual size: 134217728
fiemap: fm_mapped_extents = 32768
time = 29475 us
size: 268435456
actual size: 134217728
fiemap: fm_mapped_extents = 32768
time = 29307 us
*********** 512M ***********
size: 536870912
actual size: 268435456
fiemap: fm_mapped_extents = 65536
time = 58996 us
size: 536870912
actual size: 268435456
fiemap: fm_mapped_extents = 65536
time = 59115 us
*********** 1G ***********
size: 1073741824
actual size: 536870912
fiemap: fm_mapped_extents = 116251
time = 124141 us
size: 1073741824
actual size: 536870912
fiemap: fm_mapped_extents = 131072
time = 119387 us
The speedup is massive, both on the first fiemap call and on the second
one as well, as his test creates files with many holes and small extents
(every extent follows a hole and precedes another hole).
For the 256M file we go from 4 seconds down to 29 milliseconds in the
first run, and then from 4.9 seconds down to 29 milliseconds again in the
second run, a speedup of 138x and 169x, respectively.
For the 512M file we go from 30.1 seconds down to 59 milliseconds in the
first run, and then from 33.5 seconds down to 59 milliseconds again in the
second run, a speedup of 510x and 568x, respectively.
For the 1G file, we go from 225 seconds down to 124 milliseconds in the
first run, and then from 217 seconds down to 119 milliseconds in the
second run, a speedup of 1815x and 1824x, respectively.
Reported-by: Pavel Tikhomirov <ptikhomirov@virtuozzo.com>
Link: https://lore.kernel.org/linux-btrfs/21dd32c6-f1f9-f44a-466a-e18fdc6788a7@virtuozzo.com/
Reported-by: Dominique MARTINET <dominique.martinet@atmark-techno.com>
Link: https://lore.kernel.org/linux-btrfs/Ysace25wh5BbLd5f@atmark-techno.com/
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
During fiemap, for each file extent we find, we must check if it's shared
or not. The sharedness check starts by verifying if the extent is directly
shared (its refcount in the extent tree is > 1), and if it is not directly
shared, then we will check if every node in the subvolume b+tree leading
from the root to the leaf that has the file extent item (in reverse order),
is shared (through snapshots).
However this second step is not needed if our extent was created in a
transaction more recent than the last transaction where a snapshot of the
inode's root happened, because it can't be shared indirectly (through
shared subtrees) without a snapshot created in a more recent transaction.
So grab the generation of the extent from the extent map and pass it to
btrfs_is_data_extent_shared(), which will skip this second phase when the
generation is more recent than the root's last snapshot value. Note that
we skip this optimization if the extent map is the result of merging 2
or more extent maps, because in this case its generation is the maximum
of the generations of all merged extent maps.
The fact the we use extent maps and they can be merged despite the
underlying extents being distinct (different file extent items in the
subvolume b+tree and different extent items in the extent b+tree), can
result in some bugs when reporting shared extents. But this is a problem
of the current implementation of fiemap relying on extent maps.
One example where we get incorrect results is:
$ cat fiemap-bug.sh
#!/bin/bash
DEV=/dev/sdj
MNT=/mnt/sdj
mkfs.btrfs -f $DEV
mount $DEV $MNT
# Create a file with two 256K extents.
# Since there is no other write activity, they will be contiguous,
# and their extent maps merged, despite having two distinct extents.
xfs_io -f -c "pwrite -S 0xab 0 256K" \
-c "fsync" \
-c "pwrite -S 0xcd 256K 256K" \
-c "fsync" \
$MNT/foo
# Now clone only the second extent into another file.
xfs_io -f -c "reflink $MNT/foo 256K 0 256K" $MNT/bar
# Filefrag will report a single 512K extent, and say it's not shared.
echo
filefrag -v $MNT/foo
umount $MNT
Running the reproducer:
$ ./fiemap-bug.sh
wrote 262144/262144 bytes at offset 0
256 KiB, 64 ops; 0.0038 sec (65.479 MiB/sec and 16762.7030 ops/sec)
wrote 262144/262144 bytes at offset 262144
256 KiB, 64 ops; 0.0040 sec (61.125 MiB/sec and 15647.9218 ops/sec)
linked 262144/262144 bytes at offset 0
256 KiB, 1 ops; 0.0002 sec (1.034 GiB/sec and 4237.2881 ops/sec)
Filesystem type is: 9123683e
File size of /mnt/sdj/foo is 524288 (128 blocks of 4096 bytes)
ext: logical_offset: physical_offset: length: expected: flags:
0: 0.. 127: 3328.. 3455: 128: last,eof
/mnt/sdj/foo: 1 extent found
We end up reporting that we have a single 512K that is not shared, however
we have two 256K extents, and the second one is shared. Changing the
reproducer to clone instead the first extent into file 'bar', makes us
report a single 512K extent that is shared, which is algo incorrect since
we have two 256K extents and only the first one is shared.
This is z problem that existed before this change, and remains after this
change, as it can't be easily fixed. The next patch in the series reworks
fiemap to primarily use file extent items instead of extent maps (except
for checking for delalloc ranges), with the goal of improving its
scalability and performance, but it also ends up fixing this particular
bug caused by extent map merging.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
One of the most expensive tasks performed during fiemap is to check if
an extent is shared. This task has two major steps:
1) Check if the data extent is shared. This implies checking the extent
item in the extent tree, checking delayed references, etc. If we
find the data extent is directly shared, we terminate immediately;
2) If the data extent is not directly shared (its extent item has a
refcount of 1), then it may be shared if we have snapshots that share
subtrees of the inode's subvolume b+tree. So we check if the leaf
containing the file extent item is shared, then its parent node, then
the parent node of the parent node, etc, until we reach the root node
or we find one of them is shared - in which case we stop immediately.
During fiemap we process the extents of a file from left to right, from
file offset 0 to EOF. This means that we iterate b+tree leaves from left
to right, and has the implication that we keep repeating that second step
above several times for the same b+tree path of the inode's subvolume
b+tree.
For example, if we have two file extent items in leaf X, and the path to
leaf X is A -> B -> C -> X, then when we try to determine if the data
extent referenced by the first extent item is shared, we check if the data
extent is shared - if it's not, then we check if leaf X is shared, if not,
then we check if node C is shared, if not, then check if node B is shared,
if not than check if node A is shared. When we move to the next file
extent item, after determining the data extent is not shared, we repeat
the checks for X, C, B and A - doing all the expensive searches in the
extent tree, delayed refs, etc. If we have thousands of tile extents, then
we keep repeating the sharedness checks for the same paths over and over.
On a file that has no shared extents or only a small portion, it's easy
to see that this scales terribly with the number of extents in the file
and the sizes of the extent and subvolume b+trees.
This change eliminates the repeated sharedness check on extent buffers
by caching the results of the last path used. The results can be used as
long as no snapshots were created since they were cached (for not shared
extent buffers) or no roots were dropped since they were cached (for
shared extent buffers). This greatly reduces the time spent by fiemap for
files with thousands of extents and/or large extent and subvolume b+trees.
Example performance test:
$ cat fiemap-perf-test.sh
#!/bin/bash
DEV=/dev/sdi
MNT=/mnt/sdi
mkfs.btrfs -f $DEV
mount -o compress=lzo $DEV $MNT
# 40G gives 327680 128K file extents (due to compression).
xfs_io -f -c "pwrite -S 0xab -b 1M 0 40G" $MNT/foobar
umount $MNT
mount -o compress=lzo $DEV $MNT
start=$(date +%s%N)
filefrag $MNT/foobar
end=$(date +%s%N)
dur=$(( (end - start) / 1000000 ))
echo "fiemap took $dur milliseconds (metadata not cached)"
start=$(date +%s%N)
filefrag $MNT/foobar
end=$(date +%s%N)
dur=$(( (end - start) / 1000000 ))
echo "fiemap took $dur milliseconds (metadata cached)"
umount $MNT
Before this patch:
$ ./fiemap-perf-test.sh
(...)
/mnt/sdi/foobar: 327680 extents found
fiemap took 3597 milliseconds (metadata not cached)
/mnt/sdi/foobar: 327680 extents found
fiemap took 2107 milliseconds (metadata cached)
After this patch:
$ ./fiemap-perf-test.sh
(...)
/mnt/sdi/foobar: 327680 extents found
fiemap took 1646 milliseconds (metadata not cached)
/mnt/sdi/foobar: 327680 extents found
fiemap took 698 milliseconds (metadata cached)
That's about 2.2x faster when no metadata is cached, and about 3x faster
when all metadata is cached. On a real filesystem with many other files,
data, directories, etc, the b+trees will be 2 or 3 levels higher,
therefore this optimization will have a higher impact.
Several reports of a slow fiemap show up often, the two Link tags below
refer to two recent reports of such slowness. This patch, together with
the next ones in the series, is meant to address that.
Link: https://lore.kernel.org/linux-btrfs/21dd32c6-f1f9-f44a-466a-e18fdc6788a7@virtuozzo.com/
Link: https://lore.kernel.org/linux-btrfs/Ysace25wh5BbLd5f@atmark-techno.com/
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The function btrfs_check_shared() is supposed to be used to check if a
data extent is shared, but its name is too generic, may easily cause
confusion in the sense that it may be used for metadata extents.
So rename it to btrfs_is_data_extent_shared(), which will also make it
less confusing after the next change that adds a backref lookup cache for
the b+tree nodes that lead to the leaf that contains the file extent item
that points to the target data extent.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Qu Wenruo <wqu@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>
If the flag FIEMAP_FLAG_SYNC is passed to fiemap, it means all delalloc
should be flushed and writeback complete. We call the generic helper
fiemap_prep() which does a filemap_write_and_wait() in case that flag is
given, however that is not enough if we have compression. Because a
single filemap_fdatawrite_range() only starts compression (in an async
thread) and therefore returns before the compression is done and writeback
is started.
So make btrfs_fiemap(), actually wait for all writeback to start and
complete if FIEMAP_FLAG_SYNC is set. We start and wait for writeback
on the whole possible file range, from 0 to LLONG_MAX, because that is
what the generic code at fiemap_prep() does.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Qu Wenruo <wqu@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>
There's no point to check for a 0 length at extent_fiemap(), as before
calling it, we called fiemap_prep() at btrfs_fiemap(), which already
checks for a zero length and returns the same -EINVAL error. So remove
the pointless check.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Qu Wenruo <wqu@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>
During fiemap we are testing if an extent map has a block start with a
value of EXTENT_MAP_LAST_BYTE, but that is never set on an extent map,
and never was according to git history. So remove that useless check.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Qu Wenruo <wqu@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 current implementation of hole and data seeking for llseek does not
scale well in regards to the number of extents and the distance between
the start offset and the next hole or extent. This is due to a very high
algorithmic complexity. Often we also get reports of btrfs' hole and data
seeking (llseek) being too slow, such as at 2017's LSFMM (see the Link
tag at the bottom).
In order to better understand it, lets consider the case where the start
offset is 0, we are seeking for a hole and the file size is 16G. Between
file offset 0 and the first hole in the file there are 100K extents - this
is common for large files, specially if we have compression enabled, since
the maximum extent size is limited to 128K. The steps take by the main
loop of the current algorithm are the following:
1) We start by calling btrfs_get_extent_fiemap(), for file offset 0, which
calls btrfs_get_extent(). This will first lookup for an extent map in
the inode's extent map tree (a red black tree). If the extent map is
not loaded in memory, then it will do a lookup for the corresponding
file extent item in the subvolume's b+tree, create an extent map based
on the contents of the file extent item and then add the extent map to
the extent map tree of the inode;
2) The second iteration calls btrfs_get_extent_fiemap() again, this time
with a start offset matching the end offset of the previous extent.
Again, btrfs_get_extent() will first search the extent map tree, and
if it doesn't find an extent map there, it will again search in the
b+tree of the subvolume for a matching file extent item, build an
extent map based on the file extent item, and add the extent map to
to the extent map tree of the inode;
3) This repeats over and over until we find the first hole (when seeking
for holes) or until we find the first extent (when seeking for data).
If there no extent maps loaded in memory for each iteration, then on
each iteration we do 1 extent map tree search, 1 b+tree search, plus
1 more extent map tree traversal to insert an extent map - plus we
allocate memory for the extent map.
On each iteration we are growing the size of the extent map tree,
making each future search slower, and also visiting the same b+tree
leaves over and over again - taking into account with the default leaf
size of 16K we can fit more than 200 file extent items in a leaf - so
we can visit the same b+tree leaf 200+ times, on each visit walking
down a path from the root to the leaf.
So it's easy to see that what we have now doesn't scale well. Also, it
loads an extent map for every file extent item into memory, which is not
efficient - we should add extents maps only when doing IO (writing or
reading file data).
This change implements a new algorithm which scales much better, and
works like this:
1) We iterate over the subvolume's b+tree, visiting each leaf that has
file extent items once and only once;
2) For any file extent items found, that don't represent holes or prealloc
extents, it will not search the extent map tree - there's no need at
all for that - an extent map is just an in-memory representation of a
file extent item;
3) When a hole is found, or a prealloc extent, it will check if there's
delalloc for its range. For this it will search for EXTENT_DELALLOC
bits in the inode's io tree and check the extent map tree - this is
for accounting for unflushed delalloc and for flushed delalloc (the
period between running delalloc and ordered extent completion),
respectively. This is similar to what the current implementation does
when it finds a hole or prealloc extent, but without creating extent
maps and adding them to the extent map tree in case they are not
loaded in memory;
4) It never allocates extent maps, or adds extent maps to the inode's
extent map tree. This not only saves memory and time (from the tree
insertions and allocations), but also eliminates the possibility of
-ENOMEM due to allocating too many extent maps.
Part of this new code will also be used later for fiemap (which also
suffers similar scalability problems).
The following test example can be used to quickly measure the efficiency
before and after this patch:
$ cat test-seek-hole.sh
#!/bin/bash
DEV=/dev/sdi
MNT=/mnt/sdi
mkfs.btrfs -f $DEV
mount -o compress=lzo $DEV $MNT
# 16G file -> 131073 compressed extents.
xfs_io -f -c "pwrite -S 0xab -b 1M 0 16G" $MNT/foobar
# Leave a 1M hole at file offset 15G.
xfs_io -c "fpunch 15G 1M" $MNT/foobar
# Unmount and mount again, so that we can test when there's no
# metadata cached in memory.
umount $MNT
mount -o compress=lzo $DEV $MNT
# Test seeking for hole from offset 0 (hole is at offset 15G).
start=$(date +%s%N)
xfs_io -c "seek -h 0" $MNT/foobar
end=$(date +%s%N)
dur=$(( (end - start) / 1000000 ))
echo "Took $dur milliseconds to seek first hole (metadata not cached)"
echo
start=$(date +%s%N)
xfs_io -c "seek -h 0" $MNT/foobar
end=$(date +%s%N)
dur=$(( (end - start) / 1000000 ))
echo "Took $dur milliseconds to seek first hole (metadata cached)"
echo
umount $MNT
Before this change:
$ ./test-seek-hole.sh
(...)
Whence Result
HOLE 16106127360
Took 176 milliseconds to seek first hole (metadata not cached)
Whence Result
HOLE 16106127360
Took 17 milliseconds to seek first hole (metadata cached)
After this change:
$ ./test-seek-hole.sh
(...)
Whence Result
HOLE 16106127360
Took 43 milliseconds to seek first hole (metadata not cached)
Whence Result
HOLE 16106127360
Took 13 milliseconds to seek first hole (metadata cached)
That's about 4x faster when no metadata is cached and about 30% faster
when all metadata is cached.
In practice the differences may often be significantly higher, either due
to a higher number of extents in a file or because the subvolume's b+tree
is much bigger than in this example, where we only have one file.
Link: https://lwn.net/Articles/718805/
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Doing hole or data seeking on a file with a very large number of extents
can take a long time, and we have reports of it being too slow (such as
at LSFMM from 2017, see the Link below). So make it interruptible.
Link: https://lwn.net/Articles/718805/
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Qu Wenruo <wqu@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 problem of long mount time caused by block group item search is
already known for some time, and the solution of block group tree has
been proposed.
There is really no need to bound this feature into extent tree v2, just
introduce compat RO flag, BLOCK_GROUP_TREE, to correctly solve the
problem.
All the code handling block group root is already in the upstream
kernel, thus this patch really only needs to introduce the new compat RO
flag.
This patch introduces one extra artificial limitation on block group
tree feature, that free space cache v2 and no-holes feature must be
enabled to use this new compat RO feature.
This artificial requirement is mostly to reduce the test combinations,
and can be a guideline for future features, to mostly rely on the latest
default features.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The extent tree v2 needs a new root for storing all block group items,
the whole feature hasn't been finished yet so we can afford to do some
changes.
My initial proposal years ago just added a new tree rootid, and load it
from tree root, just like what we did for quota/free space tree/uuid/extent
roots.
But the extent tree v2 patches introduced a completely new way to store
block group tree root into super block which is arguably wasteful.
Currently there are only 3 trees stored in super blocks, and they all
have their valid reasons:
- Chunk root
Needed for bootstrap.
- Tree root
Really the entry point for all trees.
- Log root
This is special as log root has to be updated out of existing
transaction mechanism.
There is not even any reason to put block group root into super blocks,
the block group tree is updated at the same time as the old extent tree,
no need for extra bootstrap/out-of-transaction update.
So just move block group root from super block into tree root.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently there are two corner cases not handling compat RO flags
correctly:
- Remount
We can still mount the fs RO with compat RO flags, then remount it RW.
We should not allow any write into a fs with unsupported RO flags.
- Still try to search block group items
In fact, behavior/on-disk format change to extent tree should not
need a full incompat flag.
And since we can ensure fs with unsupported RO flags never got any
writes (with above case fixed), then we can even skip block group
items search at mount time.
This patch will enhance the unsupported RO compat flags by:
- Reject read-write remount if there are unsupported RO compat flags
- Go dummy block group items directly for unsupported RO compat flags
In fact, only changes to chunk/subvolume/root/csum trees should go
incompat flags.
The latter part should allow future change to extent tree to be compat
RO flags.
Thus this patch also needs to be backported to all stable trees.
CC: stable@vger.kernel.org # 4.9+
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We have hit some transaction abort due to -ENOSPC internally.
Normally we should always reserve enough space for metadata for every
transaction, thus hitting -ENOSPC should really indicate some cases we
didn't expect.
But unfortunately current error reporting will only give a kernel
warning and stack trace, not really helpful to debug what's causing the
problem.
And mount option debug_enospc can only help when user can reproduce the
problem, but under most cases, such transaction abort by -ENOSPC is
really hard to reproduce.
So this patch will dump all space infos (data, metadata, system) when we
abort the first transaction with -ENOSPC.
This should at least provide some clue to us.
The example of a dump would look like this:
BTRFS: Transaction aborted (error -28)
WARNING: CPU: 8 PID: 3366 at fs/btrfs/transaction.c:2137 btrfs_commit_transaction+0xf81/0xfb0 [btrfs]
<call trace skipped>
---[ end trace 0000000000000000 ]---
BTRFS info (device dm-1: state A): dumping space info:
BTRFS info (device dm-1: state A): space_info DATA has 6791168 free, is not full
BTRFS info (device dm-1: state A): space_info total=8388608, used=1597440, pinned=0, reserved=0, may_use=0, readonly=0 zone_unusable=0
BTRFS info (device dm-1: state A): space_info METADATA has 257114112 free, is not full
BTRFS info (device dm-1: state A): space_info total=268435456, used=131072, pinned=180224, reserved=65536, may_use=10878976, readonly=65536 zone_unusable=0
BTRFS info (device dm-1: state A): space_info SYSTEM has 8372224 free, is not full
BTRFS info (device dm-1: state A): space_info total=8388608, used=16384, pinned=0, reserved=0, may_use=0, readonly=0 zone_unusable=0
BTRFS info (device dm-1: state A): global_block_rsv: size 3670016 reserved 3670016
BTRFS info (device dm-1: state A): trans_block_rsv: size 0 reserved 0
BTRFS info (device dm-1: state A): chunk_block_rsv: size 0 reserved 0
BTRFS info (device dm-1: state A): delayed_block_rsv: size 4063232 reserved 4063232
BTRFS info (device dm-1: state A): delayed_refs_rsv: size 3145728 reserved 3145728
BTRFS: error (device dm-1: state A) in btrfs_commit_transaction:2137: errno=-28 No space left
BTRFS info (device dm-1: state EA): forced readonly
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
For btrfs_space_info, its flags has only 4 possible values:
- BTRFS_BLOCK_GROUP_SYSTEM
- BTRFS_BLOCK_GROUP_METADATA | BTRFS_BLOCK_GROUP_DATA
- BTRFS_BLOCK_GROUP_METADATA
- BTRFS_BLOCK_GROUP_DATA
Make the output more human readable, now it looks like:
BTRFS info (device dm-1: state A): space_info METADATA has 251494400 free, is not full
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
[BACKGROUND]
There is an incident report that, one user hibernated the system, with
one btrfs on removable device still mounted.
Then by some incident, the btrfs got mounted and modified by another
system/OS, then back to the hibernated system.
After resuming from the hibernation, new write happened into the victim btrfs.
Now the fs is completely broken, since the underlying btrfs is no longer
the same one before the hibernation, and the user lost their data due to
various transid mismatch.
[REPRODUCER]
We can emulate the situation using the following small script:
truncate -s 1G $dev
mkfs.btrfs -f $dev
mount $dev $mnt
fsstress -w -d $mnt -n 500
sync
xfs_freeze -f $mnt
cp $dev $dev.backup
# There is no way to mount the same cloned fs on the same system,
# as the conflicting fsid will be rejected by btrfs.
# Thus here we have to wipe the fs using a different btrfs.
mkfs.btrfs -f $dev.backup
dd if=$dev.backup of=$dev bs=1M
xfs_freeze -u $mnt
fsstress -w -d $mnt -n 20
umount $mnt
btrfs check $dev
The final fsck will fail due to some tree blocks has incorrect fsid.
This is enough to emulate the problem hit by the unfortunate user.
[ENHANCEMENT]
Although such case should not be that common, it can still happen from
time to time.
From the view of btrfs, we can detect any unexpected super block change,
and if there is any unexpected change, we just mark the fs read-only,
and thaw the fs.
By this we can limit the damage to minimal, and I hope no one would lose
their data by this anymore.
Suggested-by: Goffredo Baroncelli <kreijack@libero.it>
Link: https://lore.kernel.org/linux-btrfs/83bf3b4b-7f4c-387a-b286-9251e3991e34@bluemole.com/
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The I/O context structure is only used to pass the btrfs_device to
the end I/O handler for I/Os that go to a single device.
Stop allocating the I/O context for these cases by passing the optional
btrfs_io_stripe argument to __btrfs_map_block to query the mapping
information and then using a fast path submission and I/O completion
handler. As the old btrfs_io_context based I/O submission path is
only used for mirrored writes, rename the functions to make that
clear and stop setting the btrfs_bio device and mirror_num field
that is only used for reads.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Tested-by: Nikolay Borisov <nborisov@suse.com>
Tested-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David Sterba <dsterba@suse.com>
There is no need for most of the btrfs_io_context when doing I/O to a
single device. To support such I/O without the extra btrfs_io_context
allocation, turn the mirror_num argument into a pointer so that it can
be used to output the selected mirror number, and add an optional
argument that points to a btrfs_io_stripe structure, which will be
filled with a single extent if provided by the caller.
In that case the btrfs_io_context allocation can be skipped as all
information for the single device I/O is provided in the mirror_num
argument and the on-stack btrfs_io_stripe. A caller that makes use of
this new argument will be added in the next commit.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Tested-by: Nikolay Borisov <nborisov@suse.com>
Tested-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David Sterba <dsterba@suse.com>
Remove the orig_bio argument as it can be derived from the bioc, and
the clone argument as it can be calculated from bioc and dev_nr.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Split out a low-level btrfs_submit_dev_bio helper that just submits
the bio without any cloning decisions or setting up the end I/O handler
for future reuse by a different caller.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently btrfs_bio end I/O handling is a bit of a mess. The bi_end_io
handler and bi_private pointer of the embedded struct bio are both used
to handle the completion of the high-level btrfs_bio and for the I/O
completion for the low-level device that the embedded bio ends up being
sent to.
To support this bi_end_io and bi_private are saved into the
btrfs_io_context structure and then restored after the bio sent to the
underlying device has completed the actual I/O.
Untangle this by adding an end I/O handler and private data to struct
btrfs_bio for the high-level btrfs_bio based completions, and leave the
actual bio bi_end_io handler and bi_private pointer entirely to the
low-level device I/O.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Tested-by: Nikolay Borisov <nborisov@suse.com>
Tested-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David Sterba <dsterba@suse.com>
The parity raid write/recover functionality is currently not very well
abstracted from the bio submission and completion handling in volumes.c:
- the raid56 code directly completes the original btrfs_bio fed into
btrfs_submit_bio instead of dispatching back to volumes.c
- the raid56 code consumes the bioc and bio_counter references taken
by volumes.c, which also leads to special casing of the calls from
the scrub code into the raid56 code
To fix this up supply a bi_end_io handler that calls back into the
volumes.c machinery, which then puts the bioc, decrements the bio_counter
and completes the original bio, and updates the scrub code to also
take ownership of the bioc and bio_counter in all cases.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Tested-by: Nikolay Borisov <nborisov@suse.com>
Tested-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David Sterba <dsterba@suse.com>
The stripes_pending in the btrfs_io_context counts number of inflight
low-level bios for an upper btrfs_bio. For reads this is generally
one as reads are never cloned, while for writes we can trivially use
the bio remaining mechanisms that is used for chained bios.
To be able to make use of that mechanism, split out a separate trivial
end_io handler for the cloned bios that does a minimal amount of error
tracking and which then calls bio_endio on the original bio to transfer
control to that, with the remaining counter making sure it is completed
last. This then allows to merge btrfs_end_bioc into the original bio
bi_end_io handler.
To make this all work all error handling needs to happen through the
bi_end_io handler, which requires a small amount of reshuffling in
submit_stripe_bio so that the bio is cloned already by the time the
suitability of the device is checked.
This reduces the size of the btrfs_io_context and prepares splitting
the btrfs_bio at the stripe boundary.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David Sterba <dsterba@suse.com>
Stop grabbing an extra bio_counter reference for each clone bio in a
mirrored write and instead just release the one original reference in
btrfs_end_bioc once all the bios for a single btrfs_bio have completed
instead of at the end of btrfs_submit_bio once all bios have been
submitted.
This means the reference is now carried by the "upper" btrfs_bio only
instead of each lower bio.
Also remove the now unused btrfs_bio_counter_inc_noblocked helper.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David Sterba <dsterba@suse.com>
Pass the operation to btrfs_bio_alloc, matching what bio_alloc_bioset
set does.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Tested-by: Nikolay Borisov <nborisov@suse.com>
Tested-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
volumes.c is the place that implements the storage layer using the
btrfs_bio structure, so move the bio_set and allocation helpers there
as well.
To make up for the new initialization boilerplate, merge the two
init/exit helpers in extent_io.c into a single one.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Tested-by: Nikolay Borisov <nborisov@suse.com>
Tested-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
btrfs never uses bio integrity data itself, so don't allocate
the integrity pools for btrfs_bioset.
This patch is a revert of the commit b208c2f7ce ("btrfs: Fix crash due
to not allocating integrity data for a set"). The integrity data pool
is not needed, the bio-integrity code now handles allocating the
integrity payload without that.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Tested-by: Nikolay Borisov <nborisov@suse.com>
Tested-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David Sterba <dsterba@suse.com>
We are calling __btrfs_remove_free_space_cache everywhere to cleanup the
block group free space, however we can just use
btrfs_remove_free_space_cache and pass in the block group in all of
these places. Then we can remove __btrfs_remove_free_space_cache and
rename __btrfs_remove_free_space_cache_locked to
__btrfs_remove_free_space_cache.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Now that lockdep is staying enabled through our entire CI runs I started
seeing the following stack in generic/475
------------[ cut here ]------------
WARNING: CPU: 1 PID: 2171864 at fs/btrfs/discard.c:604 btrfs_discard_update_discardable+0x98/0xb0
CPU: 1 PID: 2171864 Comm: kworker/u4:0 Not tainted 5.19.0-rc8+ #789
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.13.0-2.fc32 04/01/2014
Workqueue: btrfs-cache btrfs_work_helper
RIP: 0010:btrfs_discard_update_discardable+0x98/0xb0
RSP: 0018:ffffb857c2f7bad0 EFLAGS: 00010246
RAX: 0000000000000000 RBX: ffff8c85c605c200 RCX: 0000000000000001
RDX: 0000000000000000 RSI: ffffffff86807c5b RDI: ffffffff868a831e
RBP: ffff8c85c4c54000 R08: 0000000000000000 R09: 0000000000000000
R10: ffff8c85c66932f0 R11: 0000000000000001 R12: ffff8c85c3899010
R13: ffff8c85d5be4f40 R14: ffff8c85c4c54000 R15: ffff8c86114bfa80
FS: 0000000000000000(0000) GS:ffff8c863bd00000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f2e7f168160 CR3: 000000010289a004 CR4: 0000000000370ee0
Call Trace:
__btrfs_remove_free_space_cache+0x27/0x30
load_free_space_cache+0xad2/0xaf0
caching_thread+0x40b/0x650
? lock_release+0x137/0x2d0
btrfs_work_helper+0xf2/0x3e0
? lock_is_held_type+0xe2/0x140
process_one_work+0x271/0x590
? process_one_work+0x590/0x590
worker_thread+0x52/0x3b0
? process_one_work+0x590/0x590
kthread+0xf0/0x120
? kthread_complete_and_exit+0x20/0x20
ret_from_fork+0x1f/0x30
This is the code
ctl = block_group->free_space_ctl;
discard_ctl = &block_group->fs_info->discard_ctl;
lockdep_assert_held(&ctl->tree_lock);
We have a temporary free space ctl for loading the free space cache in
order to avoid having allocations happening while we're loading the
cache. When we hit an error we free it all up, however this also calls
btrfs_discard_update_discardable, which requires
block_group->free_space_ctl->tree_lock to be held. However this is our
temporary ctl so this lock isn't held. Fix this by calling
__btrfs_remove_free_space_cache_locked instead so that we only clean up
the entries and do not mess with the discardable stats.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When enabling quotas, at btrfs_quota_enable(), after committing the
transaction, we change fs_info->quota_root to point to the quota root we
created and set BTRFS_FS_QUOTA_ENABLED at fs_info->flags. Then we try
to start the qgroup rescan worker, first by initializing it with a call
to qgroup_rescan_init() - however if that fails we end up freeing the
quota root but we leave fs_info->quota_root still pointing to it, this
can later result in a use-after-free somewhere else.
We have previously set the flags BTRFS_FS_QUOTA_ENABLED and
BTRFS_QGROUP_STATUS_FLAG_ON, so we can only fail with -EINPROGRESS at
btrfs_quota_enable(), which is possible if someone already called the
quota rescan ioctl, and therefore started the rescan worker.
So fix this by ignoring an -EINPROGRESS and asserting we can't get any
other error.
Reported-by: Ye Bin <yebin10@huawei.com>
Link: https://lore.kernel.org/linux-btrfs/20220823015931.421355-1-yebin10@huawei.com/
CC: stable@vger.kernel.org # 4.19+
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
btrfs currently prints information about space cache or free space tree
being in use on every remount, regardless whether such remount actually
enabled or disabled one of these features.
This is actually unnecessary since providing remount options changing the
state of these features will explicitly print the appropriate notice.
Let's instead print such unconditional information just on an initial mount
to avoid filling the kernel log when, for example, laptop-mode-tools
remount the fs on some events.
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
At btrfs_del_root_ref() we are using two return variables, named 'ret'
and 'err'. This makes it harder to follow and easier to return the wrong
value in case an error happens - the previous patch in the series, which
has the subject "btrfs: fix silent failure when deleting root
reference", fixed a bug due to confusion created by these two variables.
So change the function to use a single variable for tracking the return
value of the function, using only 'ret', which is consistent with most
of the codebase.
Reviewed-by: Qu Wenruo <wqu@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>
struct btrfs_caching_ctl::progress and struct
btrfs_block_group::last_byte_to_unpin were previously needed to ensure
that unpin_extent_range() didn't return a range to the free space cache
before the caching thread had a chance to cache that range. However, the
commit "btrfs: fix space cache corruption and potential double
allocations" made it so that we always synchronously cache the block
group at the time that we pin the extent, so this machinery is no longer
necessary.
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
There is a bug causing send failures when processing an orphan directory
with no links. In commit 46b2f4590a ("Btrfs: fix send failure when
root has deleted files still open")', the orphan inode issue was
addressed. The send operation fails with a ENOENT error because of any
attempts to generate a path for the inode with a link count of zero.
Therefore, in that patch, sctx->ignore_cur_inode was introduced to be
set if the current inode has a link count of zero for bypassing some
unnecessary steps. And a helper function btrfs_unlink_all_paths() was
introduced and called to clean up old paths found in the parent
snapshot. However, not only regular files but also directories can be
orphan inodes. So if the send operation meets an orphan directory, it
will issue a wrong unlink command for that directory now. Soon the
receive operation fails with a EISDIR error. Besides, the send operation
also fails with a ENOENT error later when it tries to generate a path of
it.
Similar example but making an orphan dir for an incremental send:
$ btrfs subvolume create vol
$ mkdir vol/dir
$ touch vol/dir/foo
$ btrfs subvolume snapshot -r vol snap1
$ btrfs subvolume snapshot -r vol snap2
# Turn the second snapshot to RW mode and delete the whole dir while
# holding an open file descriptor on it.
$ btrfs property set snap2 ro false
$ exec 73<snap2/dir
$ rm -rf snap2/dir
# Set the second snapshot back to RO mode and do an incremental send.
$ btrfs property set snap2 ro true
$ mkdir receive_dir
$ btrfs send snap2 -p snap1 | btrfs receive receive_dir/
At subvol snap2
At snapshot snap2
ERROR: send ioctl failed with -2: No such file or directory
ERROR: unlink dir failed. Is a directory
Actually, orphan inodes are more common use cases in cascading backups.
(Please see the illustration below.) In a cascading backup, a user wants
to replicate a couple of snapshots from Machine A to Machine B and from
Machine B to Machine C. Machine B doesn't take any RO snapshots for
sending. All a receiver does is create an RW snapshot of its parent
snapshot, apply the send stream and turn it into RO mode at the end.
Even if all paths of some inodes are deleted in applying the send
stream, these inodes would not be deleted and become orphans after
changing the subvolume from RW to RO. Moreover, orphan inodes can occur
not only in send snapshots but also in parent snapshots because Machine
B may do a batch replication of a couple of snapshots.
An illustration for cascading backups:
Machine A (snapshot {1..n}) --> Machine B --> Machine C
The idea to solve the problem is to delete all the items of orphan
inodes before using these snapshots for sending. I used to think that
the reasonable timing for doing that is during the ioctl of changing the
subvolume from RW to RO because it sounds good that we will not modify
the fs tree of a RO snapshot anymore. However, attempting to do the
orphan cleanup in the ioctl would be pointless. Because if someone is
holding an open file descriptor on the inode, the reference count of the
inode will never drop to 0. Then iput() cannot trigger eviction, which
finally deletes all the items of it. So we try to extend the original
patch to handle orphans in send/parent snapshots. Here are several cases
that need to be considered:
Case 1: BTRFS_COMPARE_TREE_NEW
| send snapshot | action
--------------------------------
nlink | 0 | ignore
In case 1, when we get a BTRFS_COMPARE_TREE_NEW tree comparison result,
it means that a new inode is found in the send snapshot and it doesn't
appear in the parent snapshot. Since this inode has a link count of zero
(It's an orphan and there're no paths for it.), we can leverage
sctx->ignore_cur_inode in the original patch to prevent it from being
created.
Case 2: BTRFS_COMPARE_TREE_DELETED
| parent snapshot | action
----------------------------------
nlink | 0 | as usual
In case 2, when we get a BTRFS_COMPARE_TREE_DELETED tree comparison
result, it means that the inode only appears in the parent snapshot.
As usual, the send operation will try to delete all its paths. However,
this inode has a link count of zero, so no paths of it will be found. No
deletion operations will be issued. We don't need to change any logic.
Case 3: BTRFS_COMPARE_TREE_CHANGED
| | parent snapshot | send snapshot | action
-----------------------------------------------------------------------
subcase 1 | nlink | 0 | 0 | ignore
subcase 2 | nlink | >0 | 0 | new_gen(deletion)
subcase 3 | nlink | 0 | >0 | new_gen(creation)
In case 3, when we get a BTRFS_COMPARE_TREE_CHANGED tree comparison result,
it means that the inode appears in both snapshots. Here are 3 subcases.
First, when the inode has link counts of zero in both snapshots. Since
there are no paths for this inode in (source/destination) parent
snapshots and we don't care about whether there is also an orphan inode
in destination or not, we can set sctx->ignore_cur_inode on to prevent
it from being created.
For the second and the third subcases, if there are paths in one
snapshot and there're no paths in the other snapshot for this inode. We
can treat this inode as a new generation. We can also leverage the logic
handling a new generation of an inode with small adjustments. Then it
will delete all old paths and create a new inode with new attributes and
paths only when there's a positive link count in the send snapshot.
In subcase 2, the send operation only needs to delete all old paths as
in the parent snapshot. But it may require more operations for a
directory to remove its old paths. If a not-empty directory is going to
be deleted (because it has a link count of zero in the send snapshot)
but there are files/directories with bigger inode numbers under it, the
send operation will need to rename it to its orphan name first. After
processing and deleting the last item under this directory, the send
operation will check this directory, aka the parent directory of the
last item, again and issue a rmdir operation to remove it finally.
Therefore, we also need to treat inodes with a link count of zero as if
they didn't exist in get_cur_inode_state(), which is used in
process_recorded_refs(). By doing this, when checking a directory with
orphan names after the last item under it has been deleted, the send
operation now can properly issue a rmdir operation. Otherwise, without
doing this, the orphan directory with an orphan name would be kept here
at the end due to the existing inode with a link count of zero being
found.
In subcase 3, as in case 2, no old paths would be found, so no deletion
operations will be issued. The send operation will only create a new one
for that inode.
Note that subcase 3 is not common. That's because it's easy to reduce
the hard links of an inode, but once all valid paths are removed,
there are no valid paths for creating other hard links. The only way to
do that is trying to send an older snapshot after a newer snapshot has
been sent.
Reviewed-by: Robbie Ko <robbieko@synology.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: BingJing Chang <bingjingc@synology.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Refactor get_inode_info() to populate all wanted fields on an output
structure. Besides, also introduce a helper function called
get_inode_gen(), which is commonly used.
Reviewed-by: Robbie Ko <robbieko@synology.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: BingJing Chang <bingjingc@synology.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
After we copied data to page cache in buffered I/O, we
1. Insert a EXTENT_UPTODATE state into inode's io_tree, by
endio_readpage_release_extent(), set_extent_delalloc() or
set_extent_defrag().
2. Set page uptodate before we unlock the page.
But the only place we check io_tree's EXTENT_UPTODATE state is in
btrfs_do_readpage(). We know we enter btrfs_do_readpage() only when we
have a non-uptodate page, so it is unnecessary to set EXTENT_UPTODATE.
For example, when performing a buffered random read:
fio --rw=randread --ioengine=libaio --direct=0 --numjobs=4 \
--filesize=32G --size=4G --bs=4k --name=job \
--filename=/mnt/file --name=job
Then check how many extent_state in io_tree:
cat /proc/slabinfo | grep btrfs_extent_state | awk '{print $2}'
w/o this patch, we got 640567 btrfs_extent_state.
w/ this patch, we got 204 btrfs_extent_state.
Maintaining such a big tree brings overhead since every I/O needs to insert
EXTENT_LOCKED, insert EXTENT_UPTODATE, then remove EXTENT_LOCKED. And in
every insert or remove, we need to lock io_tree, do tree search, alloc or
dealloc extent states. By removing unnecessary EXTENT_UPTODATE, we keep
io_tree in a minimal size and reduce overhead when performing buffered I/O.
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Robbie Ko <robbieko@synology.com>
Signed-off-by: Ethan Lien <ethanlien@synology.com>
Signed-off-by: David Sterba <dsterba@suse.com>
During log replay, when adding/replacing inode references, there are two
special cases that have special code for them:
1) When we have an inode with two or more hardlinks in the same directory,
therefore two or more names encoded in the same inode reference item,
and one of the hard links gets renamed to the old name of another hard
link - that is, the index number for a name changes. This was added in
commit 0d836392ca ("Btrfs: fix mount failure after fsync due to
hard link recreation"), and is covered by test case generic/502 from
fstests;
2) When we have several inodes that got renamed to an old name of some
other inode, in a cascading style. The code to deal with this special
case was added in commit 6b5fc433a7 ("Btrfs: fix fsync after
succession of renames of different files"), and is covered by test
cases generic/526 and generic/527 from fstests.
Both cases can be deal with by making sure __add_inode_ref() is always
called by add_inode_ref() for every name encoded in the inode reference
item, and not just for the first name that has a conflict. With such
change we no longer need that special casing for the two cases mentioned
before. So do those changes.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When discard=async was introduced there were also sysfs knobs and stats
for debugging and tuning, hidden under CONFIG_BTRFS_DEBUG. The defaults
have been set and so far seem to satisfy all users on a range of
workloads. As there are not only tunables (like iops or kbps) but also
stats tracking amount of discardable bytes, that should be available
when the async discard is on (otherwise it's not).
The stats are moved from the per-fs debug directory, so it's under
/sys/fs/btrfs/FSID/discard
- discard_bitmap_bytes - amount of discarded bytes from data tracked as
bitmaps
- discard_extent_bytes - dtto but as extents
- discard_bytes_saved -
- discardable_bytes - amount of bytes that can be discarded
- discardable_extents - number of extents to be discarded
- iops_limit - tunable limit of number of discard IOs to be issued
- kbps_limit - tunable limit of kilobytes per second issued as discard IO
- max_discard_size - tunable limit for size of one IO discard request
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When logging a directory we start by flushing all its delayed items.
That results in adding dir index items to the subvolume btree, for new
dentries, and removing dir index items from the subvolume btree for any
dentries that were deleted.
This makes it straightforward to log a directory simply by iterating over
all the modified subvolume btree leaves, especially when we used to log
both dir index keys and dir item keys (before commit 339d035424
("btrfs: only copy dir index keys when logging a directory") and when we
used to copy old dir index entries for leaves modified in the current
transaction (before commit 732d591a5d ("btrfs: stop copying old dir
items when logging a directory")).
From an efficiency point of view this has a couple of drawbacks:
1) Adds extra latency, due to copying delayed items to the subvolume btree
and deleting dir index items from the btree.
Further if there are other tasks accessing the btree, which is common
(syscalls like creat, mkdir, rename, link, unlink, truncate, reflinks,
etc, finishing an ordered extent, etc), lock contention can cause
further delays, both to the task logging a directory and to the other
tasks accessing the btree;
2) More time spent overall flushing delayed items, if after logging the
directory further changes are done to the directory in the same
transaction.
For example, if we add 10 dentries to a directory, fsync it, add more
10 dentries, fsync it again, then add more 10 dentries and fsync it
again, then we end up inserting 3 batches of 10 items to the subvolume
btree. With the changes from this patch, we flush all the delayed items
to the btree only once - a single batch of 30 items, and outside the
logging code (transaction commit or when delayed items are flushed
asynchronously).
This change simply skips the flushing of delayed items every time we log a
directory. Instead we copy the delayed insertion items directly to the log
tree and delete delayed deletion items directly from the log tree.
Therefore avoiding changing first the subvolume btree and then scanning it
for new items to copy from it to the log tree and detecting deletions
by observing gaps in consecutive dir index keys in subvolume btree leaves.
Running the following tests on a non-debug kernel (Debian's default kernel
config), on a box with a NVMe device, a 12 cores Intel CPU and 64G of ram,
produced the results below.
The results compare a branch without this patch and all the other patches
it depends on versus the same branch with the patchset applied.
The patchset is comprised of the following patches:
btrfs: don't drop dir index range items when logging a directory
btrfs: remove the root argument from log_new_dir_dentries()
btrfs: update stale comment for log_new_dir_dentries()
btrfs: free list element sooner at log_new_dir_dentries()
btrfs: avoid memory allocation at log_new_dir_dentries() for common case
btrfs: remove root argument from btrfs_delayed_item_reserve_metadata()
btrfs: store index number instead of key in struct btrfs_delayed_item
btrfs: remove unused logic when looking up delayed items
btrfs: shrink the size of struct btrfs_delayed_item
btrfs: search for last logged dir index if it's not cached in the inode
btrfs: move need_log_inode() to above log_conflicting_inodes()
btrfs: move log_new_dir_dentries() above btrfs_log_inode()
btrfs: log conflicting inodes without holding log mutex of the initial inode
btrfs: skip logging parent dir when conflicting inode is not a dir
btrfs: use delayed items when logging a directory
Custom test script for testing time spent at btrfs_log_inode():
#!/bin/bash
DEV=/dev/nvme0n1
MNT=/mnt/nvme0n1
# Total number of files to create in the test directory.
NUM_FILES=10000
# Fsync after creating or renaming N files.
FSYNC_AFTER=100
umount $DEV &> /dev/null
mkfs.btrfs -f $DEV
mount -o ssd $DEV $MNT
TEST_DIR=$MNT/testdir
mkdir $TEST_DIR
echo "Creating files..."
for ((i = 1; i <= $NUM_FILES; i++)); do
echo -n > $TEST_DIR/file_$i
if (( ($i % $FSYNC_AFTER) == 0 )); then
xfs_io -c "fsync" $TEST_DIR
fi
done
sync
echo "Renaming files..."
for ((i = 1; i <= $NUM_FILES; i++)); do
mv $TEST_DIR/file_$i $TEST_DIR/file_$i.renamed
if (( ($i % $FSYNC_AFTER) == 0 )); then
xfs_io -c "fsync" $TEST_DIR
fi
done
umount $MNT
And using the following bpftrace script to capture the total time that is
spent at btrfs_log_inode():
#!/usr/bin/bpftrace
k:btrfs_log_inode
{
@start_log_inode[tid] = nsecs;
}
kr:btrfs_log_inode
/@start_log_inode[tid]/
{
$dur = (nsecs - @start_log_inode[tid]) / 1000;
@btrfs_log_inode_total_time = sum($dur);
delete(@start_log_inode[tid]);
}
END
{
clear(@start_log_inode);
}
Result before applying patchset:
@btrfs_log_inode_total_time: 622642
Result after applying patchset:
@btrfs_log_inode_total_time: 354134 (-43.1% time spent)
The following dbench script was also used for testing:
#!/bin/bash
NUM_JOBS=$(nproc --all)
DEV=/dev/nvme0n1
MNT=/mnt/nvme0n1
MOUNT_OPTIONS="-o ssd"
MKFS_OPTIONS="-O no-holes -R free-space-tree"
echo "performance" | \
tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor
umount $DEV &> /dev/null
mkfs.btrfs -f $MKFS_OPTIONS $DEV
mount $MOUNT_OPTIONS $DEV $MNT
dbench -D $MNT --skip-cleanup -t 120 -S $NUM_JOBS
umount $MNT
Before patchset:
Operation Count AvgLat MaxLat
----------------------------------------
NTCreateX 3322265 0.034 21.032
Close 2440562 0.002 0.994
Rename 140664 1.150 269.633
Unlink 670796 1.093 269.678
Deltree 96 5.481 15.510
Mkdir 48 0.004 0.052
Qpathinfo 3010924 0.014 8.127
Qfileinfo 528055 0.001 0.518
Qfsinfo 552113 0.003 0.372
Sfileinfo 270575 0.005 0.688
Find 1164176 0.052 13.931
WriteX 1658537 0.019 5.918
ReadX 5207412 0.003 1.034
LockX 10818 0.003 0.079
UnlockX 10818 0.002 0.313
Flush 232811 1.027 269.735
Throughput 869.867 MB/sec (sync dirs) 12 clients 12 procs max_latency=269.741 ms
After patchset:
Operation Count AvgLat MaxLat
----------------------------------------
NTCreateX 4152738 0.029 20.863
Close 3050770 0.002 1.119
Rename 175829 0.871 211.741
Unlink 838447 0.845 211.724
Deltree 120 4.798 14.162
Mkdir 60 0.003 0.005
Qpathinfo 3763807 0.011 4.673
Qfileinfo 660111 0.001 0.400
Qfsinfo 690141 0.003 0.429
Sfileinfo 338260 0.005 0.725
Find 1455273 0.046 6.787
WriteX 2073307 0.017 5.690
ReadX 6509193 0.003 1.171
LockX 13522 0.003 0.077
UnlockX 13522 0.002 0.125
Flush 291044 0.811 211.631
Throughput 1089.27 MB/sec (sync dirs) 12 clients 12 procs max_latency=211.750 ms
(+25.2% throughput, -21.5% max latency)
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When we find a conflicting inode (an inode that had the same name and
parent directory as the inode we are logging now) that was deleted in the
current transaction, we always end up logging its parent directory.
This is to deal with the case where the conflicting inode corresponds to
a deleted subvolume/snapshot or a directory that had subvolumes/snapshots
(or some subdirectory inside it had subvolumes/snapshots, etc), because
we can't deal with dropping subvolumes/snapshots during log replay. So
if we log the parent directory, and if we are dealing with these special
cases, then we fallback to a transaction commit when logging the parent,
because its last_unlink_trans will match the current transaction (which
gets set and propagated when a subvolume/snapshot is deleted).
This change skips the logging of the parent directory when the conflicting
inode is not a directory (or a subvolume/snapshot). This is ok because in
this case logging the current inode is enough to trigger an unlink of the
conflicting inode during log replay.
So for a case like this:
$ mkdir /mnt/dir
$ echo -n "first foo data" > /mnt/dir/foo
$ sync
$ rm -f /mnt/dir/foo
$ echo -n "second foo data" > /mnt/dir/foo
$ xfs_io -c "fsync" /mnt/dir/foo
We avoid logging parent directory "dir" when logging the new file "foo".
In other cases it avoids falling back to a transaction commit, when the
parent directory has a last_unlink_trans value that matches the current
transaction, due to moving a file from it to some other directory.
This is a case that happens frequently with dbench for example, where a
new file that has the name/parent of another file that was deleted in the
current transaction, is fsynced.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When logging an inode, if we detect the inode has a reference that
conflicts with some other inode that got renamed, we log that other inode
while holding the log mutex of the current inode. We then find out if
there are other inodes that conflict with the first conflicting inode,
and log them while under the log mutex of the original inode. This is
fine because the recursion can only happen once.
For the upcoming work where we directly log delayed items without flushing
them first to the subvolume tree, this recursion adds a lot of complexity
and it's hard to keep lockdep happy about it.
So collect a list of conflicting inodes and then log the inodes after
unlocking the log mutex of the inode we started with.
Also limit the maximum number of conflict inodes we log to 10, to avoid
spending too much time logging (and maybe allocating too many list
elements too), as typically we don't have more than 1 or 2 conflicting
inodes - if we go over the limit, simply fallback to a transaction commit.
It is possible to have a very long list of conflicting inodes to be
intentionally created by a user if he/she creates a very long succession
of renames like this:
(...)
rename E to F
rename D to E
rename C to D
rename B to C
rename A to B
touch A (create a new file named A)
fsync A
If that happened for a sequence of hundreds or thousands of renames, it
could massively slow down the logging and cause other secondary effects
like for example blocking other fsync operations and transaction commits
for a very long time (assuming it wouldn't run into -ENOSPC or -ENOMEM
first). However such cases are very uncommon to happen in practice,
nevertheless it's better to be prepared for them and avoid chaos.
Such long sequence of conflicting inodes could be created before this
change.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The static function log_new_dir_dentries() is currently defined below
btrfs_log_inode(), but in an upcoming patch a new function is introduced
that is called by btrfs_log_inode() and this new function needs to call
log_new_dir_dentries(). So move log_new_dir_dentries() to a location
between btrfs_log_inode() and need_log_inode() (the later is called by
log_new_dir_dentries()).
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The static function need_log_inode() is defined below btrfs_log_inode()
and log_conflicting_inodes(), but in the next patches in the series we
will need to call need_log_inode() in a couple new functions that will be
used by btrfs_log_inode(). So move its definition to a location above
log_conflicting_inodes().
Also make its arguments 'const', since they are not supposed to be
modified.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The key offset of the last dir index item that was logged is stored in
the inode's last_dir_index_offset field. However that field is not
persisted in the inode item or elsewhere, so if the inode gets evicted
and reloaded, it gets a value of (u64)-1, so that when we are logging
dir index items we check if they were logged before, to avoid attempts
to insert duplicated keys and fallback to a transaction commit.
Improve on this by searching for the last dir index that was logged when
we start logging a directory if the inode's last_dir_index_offset is not
set (has a value of (u64)-1) and it was logged before. This avoids
checking if each dir index item we find was already logged before, and
simplifies the logging of dir index items (process_dir_items_leaf()).
This will also be needed for an incoming change where we start logging
delayed items directly, without flushing them first.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently struct btrfs_delayed_item has a base size of 96 bytes, but its
size can be decreased by doing the following 2 tweaks:
1) Change data_len from u32 to u16. Our maximum possible leaf size is 64K,
so the data_len can never be larger than that, and in fact it is always
much smaller than that. The max length for a dentry's name is ensured
at the VFS level (PATH_MAX, 4096 bytes) and in struct btrfs_inode_ref
and btrfs_dir_item we use a u16 to store the name's length;
2) Change 'ins_or_del' to a 1 bit enum, which is all we need since it
can only have 2 values. After this there's also no longer the need to
BUG_ON() before using 'ins_or_del' in several places. Also rename the
field from 'ins_or_del' to 'type', which is more clear.
These two tweaks decrease the size of struct btrfs_delayed_item from 96
bytes down to 88 bytes. A previous patch already reduced the size of this
structure by 16 bytes, but an upcoming change will increase its size by
16 bytes (adding a struct list_head element).
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
All callers pass NULL to the 'prev' and 'next' arguments of the function
__btrfs_lookup_delayed_item(), so remove these arguments. Also, remove
the unnecessary wrapper __btrfs_lookup_delayed_insertion_item(), making
btrfs_delete_delayed_insertion_item() directly call
__btrfs_lookup_delayed_item().
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
All delayed items are for dir index keys, so there's really no point of
having an embedded struct btrfs_key in struct btrfs_delayed_item, which
makes the structure use more space than necessary (and adds a hole of 7
bytes).
So replace the key field with an index number (u64), which reduces the
size of struct btrfs_delayed_item from 112 bytes down to 96 bytes.
Some upcoming work will increase the structure size by 16 bytes, so this
change compensates for that future size increase.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The root argument of btrfs_delayed_item_reserve_metadata() is used only
to get the fs_info object, but we already have a transaction handle, which
we can use to get the fs_info. So remove the root argument.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
At log_new_dir_dentries() we always start by allocating a list element
for the starting inode and then do a while loop with the condition being
a list emptiness check.
This however is not needed, we can avoid allocating this initial list
element and then just check for the list emptiness at the end of the
loop's body. So just do that to save one memory allocation from the
kmalloc-32 slab.
This allows for not doing any memory allocation when we don't have any
subdirectory to log, which is a very common case.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
At log_new_dir_dentries(), there's no need to keep the current list
element allocated while processing the leaves with directory items for
the current directory, and while logging other inodes. Plus in case we
find a subdirectory, we also end up allocating a new list element while
the current one is still allocated, temporarily using more memory than
necessary.
So free the current list element early on, before processing leaves.
Also make the removal and release of all list elements in case of an
error more simple by eliminating the label and goto, adding an explicit
loop to release all list elements in case an error happens.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The comment refers to the function log_dir_items() in order to check why
the inodes of new directory entries need to be logged, but the relevant
comments are no longer at log_dir_items(), they were moved to the function
process_dir_items_leaf() in commit eb10d85ee7 ("btrfs: factor out the
copying loop of dir items from log_dir_items()"). So update it with the
current function name.
Also remove references with i_mutex to "VFS lock", since the inode lock
is no longer a mutex since 2016 (it's now a rw semaphore).
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
There's no point in passing a root argument to log_new_dir_dentries()
because it always corresponds to the root of the given inode. So remove
it and extract the root from the given inode.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When logging a directory that was previously logged in the current
transaction, we drop all the range items (BTRFS_DIR_LOG_INDEX_KEY key
type). This is because we will process all leaves in the subvolume's tree
that were changed in the current transaction and then add range items for
covering new dir index items and deleted dir index items, which could
cover now a larger range than before.
We used to fail if we tried to insert a range item key that already
exists, so we dropped all range items to avoid failing. However nowadays,
since commit 750ee45490 ("btrfs: fix assertion failure when logging
directory key range item"), we simply update any range item that already
exists, increasing its range's last dir index if needed. Since the range
covered by a range item can never decrease, due to the fact that dir index
values come from a monotonically increasing counter and are never reused,
we can stop dropping all range items before we start logging a directory.
By not dropping the items we can avoid having occasional tree rebalance
operations.
This will also be needed for an incoming change where we start logging
delayed items directly, without flushing them first.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
[PROBLEM]
The existing scrub code for data extents always limit the block size to
sectorsize.
This causes quite some extra scrub_block being allocated:
(there is a data extent at logical bytenr 298844160, length 64KiB)
alloc_scrub_block: new block: logical=298844160 physical=298844160 mirror=1
alloc_scrub_block: new block: logical=298848256 physical=298848256 mirror=1
alloc_scrub_block: new block: logical=298852352 physical=298852352 mirror=1
alloc_scrub_block: new block: logical=298856448 physical=298856448 mirror=1
alloc_scrub_block: new block: logical=298860544 physical=298860544 mirror=1
alloc_scrub_block: new block: logical=298864640 physical=298864640 mirror=1
alloc_scrub_block: new block: logical=298868736 physical=298868736 mirror=1
alloc_scrub_block: new block: logical=298872832 physical=298872832 mirror=1
alloc_scrub_block: new block: logical=298876928 physical=298876928 mirror=1
alloc_scrub_block: new block: logical=298881024 physical=298881024 mirror=1
alloc_scrub_block: new block: logical=298885120 physical=298885120 mirror=1
alloc_scrub_block: new block: logical=298889216 physical=298889216 mirror=1
alloc_scrub_block: new block: logical=298893312 physical=298893312 mirror=1
alloc_scrub_block: new block: logical=298897408 physical=298897408 mirror=1
alloc_scrub_block: new block: logical=298901504 physical=298901504 mirror=1
alloc_scrub_block: new block: logical=298905600 physical=298905600 mirror=1
...
scrub_block_put: free block: logical=298844160 physical=298844160 len=4096 mirror=1
scrub_block_put: free block: logical=298848256 physical=298848256 len=4096 mirror=1
scrub_block_put: free block: logical=298852352 physical=298852352 len=4096 mirror=1
scrub_block_put: free block: logical=298856448 physical=298856448 len=4096 mirror=1
scrub_block_put: free block: logical=298860544 physical=298860544 len=4096 mirror=1
scrub_block_put: free block: logical=298864640 physical=298864640 len=4096 mirror=1
scrub_block_put: free block: logical=298868736 physical=298868736 len=4096 mirror=1
scrub_block_put: free block: logical=298872832 physical=298872832 len=4096 mirror=1
scrub_block_put: free block: logical=298876928 physical=298876928 len=4096 mirror=1
scrub_block_put: free block: logical=298881024 physical=298881024 len=4096 mirror=1
scrub_block_put: free block: logical=298885120 physical=298885120 len=4096 mirror=1
scrub_block_put: free block: logical=298889216 physical=298889216 len=4096 mirror=1
scrub_block_put: free block: logical=298893312 physical=298893312 len=4096 mirror=1
scrub_block_put: free block: logical=298897408 physical=298897408 len=4096 mirror=1
scrub_block_put: free block: logical=298901504 physical=298901504 len=4096 mirror=1
scrub_block_put: free block: logical=298905600 physical=298905600 len=4096 mirror=1
This behavior will waste a lot of memory, especially after we have moved
quite some members from scrub_sector to scrub_block.
[FIX]
To reduce the allocation of scrub_block, and to reduce memory usage, use
BTRFS_STRIPE_LEN instead of sectorsize as the block size to scrub data
extents.
This results only one scrub_block to be allocated for above data extent:
alloc_scrub_block: new block: logical=298844160 physical=298844160 mirror=1
scrub_block_put: free block: logical=298844160 physical=298844160 len=65536 mirror=1
This would greatly reduce the memory usage (even it's just transient)
for larger data extents scrub.
For above example, the memory usage would be:
Old: num_sectors * (sizeof(scrub_block) + sizeof(scrub_sector))
16 * (408 + 96) = 8065
New: sizeof(scrub_block) + num_sectors * sizeof(scrub_sector)
408 + 16 * 96 = 1944
A good reduction of 75.9%.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently we store the following members in scrub_sector:
- logical
- physical
- physical_for_dev_replace
- dev
- mirror_num
However the current scrub code has ensured that scrub_blocks never cross
stripe boundary.
This is caused by the entry functions (scrub_simple_mirror,
scrub_simple_stripe), thus every scrub_block will not cross stripe
boundary.
Thus this makes it possible to move those members into scrub_block other
than putting them into scrub_sector.
This should save quite some memory, as a scrub_block can be as large as 64
sectors, even for metadata it's 16 sectors byte default.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Although scrub currently works for subpage (PAGE_SIZE > sectorsize) cases,
it will allocate one page for each scrub_sector, which can cause extra
unnecessary memory usage.
Utilize scrub_block::pages[] instead of allocating page for each
scrub_sector, this allows us to integrate larger extents while using
less memory.
For example, if our page size is 64K, sectorsize is 4K, and we got an
32K sized extent.
We will only allocate one page for scrub_block, and all 8 scrub sectors
will point to that page.
To do that properly, here we introduce several small helpers:
- scrub_page_get_logical()
Get the logical bytenr of a page.
We store the logical bytenr of the page range into page::private.
But for 32bit systems, their (void *) is not large enough to contain
a u64, so in that case we will need to allocate extra memory for it.
For 64bit systems, we can use page::private directly.
- scrub_block_get_logical()
Just get the logical bytenr of the first page.
- scrub_sector_get_page()
Return the page which the scrub_sector points to.
- scrub_sector_get_page_offset()
Return the offset inside the page which the scrub_sector points to.
- scrub_sector_get_kaddr()
Return the address which the scrub_sector points to.
Just a wrapper using scrub_sector_get_page() and
scrub_sector_get_page_offset()
- bio_add_scrub_sector()
Please note that, even with this patch, we're still allocating one page
for one sector for data extents.
This is because in scrub_extent() we split the data extent using
sectorsize.
The memory usage reduction will need extra work to make scrub to work
like data read to only use the correct sector(s).
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Josef Bacik