At btrfs_commit_super(), called in a few contexts such as when unmounting
a filesystem, we use btrfs_join_transaction() to catch any running
transaction and then commit it. This will however create a new and empty
transaction in case there's no running transaction or there's a running
transaction with a state >= TRANS_STATE_UNBLOCKED.
As we just want to be sure that any existing transaction is fully
committed, we can use btrfs_attach_transaction_barrier() instead of
btrfs_join_transaction(), therefore avoiding the creation and commit of
empty transactions, which only waste IO and causes rotation of the
precious backup roots.
Example where we create and commit a pointless empty transaction:
$ mkfs.btrfs -f /dev/sdj
$ btrfs inspect-internal dump-super /dev/sdj | grep -e '^generation'
generation 6
$ mount /dev/sdj /mnt/sdj
$ touch /mnt/sdj/foo
# Commit the currently open transaction. Just 'sync' or wait ~30
# seconds for the transaction kthread to commit it.
$ sync
$ btrfs inspect-internal dump-super /dev/sdj | grep -e '^generation'
generation 7
$ umount /mnt/sdj
$ btrfs inspect-internal dump-super /dev/sdj | grep -e '^generation'
generation 8
The transaction with id 8 was pointless, an empty transaction that did
not achieve anything.
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 range is specified only in two ways, we can simplify the case for
the whole filesystem range as a NULL block group parameter.
Signed-off-by: David Sterba <dsterba@suse.com>
The function btrfs_block_group_root() is declared in disk-io.c; however,
all its callers are in block-group.c. Move it to the latter file and
declare it static.
Reviewed-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
On 64 bits platforms we don't really need to have a dedicated member (the
objectid field) for the inode's number since we store in the VFS inode's
i_ino member, which is an unsigned long and this type is 64 bits wide on
64 bits platforms. We only need that field in case we are on a 32 bits
platform because the unsigned long type is 32 bits wide on such platforms
See commit 33345d0152 ("Btrfs: Always use 64bit inode number") regarding
this 64/32 bits detail.
The objectid field of struct btrfs_inode is also used to store the ID of
a root for directories that are stubs for unreferenced roots. In such
cases the inode is a directory and has the BTRFS_INODE_ROOT_STUB runtime
flag set.
So in order to reduce the size of btrfs_inode structure on 64 bits
platforms we can remove the objectid member and use the VFS inode's i_ino
member instead whenever we need to get the inode number. In case the inode
is a root stub (BTRFS_INODE_ROOT_STUB set) we can use the member
last_reflink_trans to store the ID of the unreferenced root, since such
inode is a directory and reflinks can't be done against directories.
So remove the objectid fields for 64 bits platforms and alias the
last_reflink_trans field with a name of ref_root_id in a union.
On a release kernel config, this reduces the size of struct btrfs_inode
from 1040 bytes down to 1032 bytes.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently struct btrfs_inode has a key member, named "location", that is
either:
1) The key of the inode's item. In this case the objectid is the number
of the inode;
2) A key stored in a dir entry with a type of BTRFS_ROOT_ITEM_KEY, for
the case where we have a root that is a snapshot of a subvolume that
points to other subvolumes. In this case the objectid is the ID of
a subvolume inside the snapshotted parent subvolume.
The key is only used to lookup the inode item for the first case, while
for the second it's never used since it corresponds to directory stubs
created with new_simple_dir() and which are marked as dummy, so there's
no actual inode item to ever update. In the second case we only check
the key type at btrfs_ino() for 32 bits platforms and its objectid is
only needed for unlink.
Instead of using a key we can do fine with just the objectid, since we
can generate the key whenever we need it having only the objectid, as
in all use cases the type is always BTRFS_INODE_ITEM_KEY and the offset
is always 0.
So use only an objectid instead of a full key. This reduces the size of
struct btrfs_inode from 1048 bytes down to 1040 bytes on a release kernel.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently we use the spinlock inode_lock from struct btrfs_root to
serialize access to two different data structures:
1) The delayed inodes xarray (struct btrfs_root::delayed_nodes);
2) The inodes xarray (struct btrfs_root::inodes).
Instead of using our own lock, we can use the spinlock that is part of the
xarray implementation, by using the xa_lock() and xa_unlock() APIs and
using the xarray APIs with the double underscore prefix that don't take
the xarray locks and assume the caller is using xa_lock() and xa_unlock().
So remove the spinlock inode_lock from struct btrfs_root and use the
corresponding xarray locks. This brings 2 benefits:
1) We reduce the size of struct btrfs_root, from 1336 bytes down to
1328 bytes on a 64 bits release kernel config;
2) We reduce lock contention by not using anymore the same lock for
changing two different and unrelated xarrays.
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>
Currently we use a red black tree (rb-tree) to track the currently open
inodes of a root (in struct btrfs_root::inode_tree). This however is not
very efficient when the number of inodes is large since rb-trees are
binary trees. For example for 100K open inodes, the tree has a depth of
17. Besides that, inserting into the tree requires navigating through it
and pulling useless cache lines in the process since the red black tree
nodes are embedded within the btrfs inode - on the other hand, by being
embedded, it requires no extra memory allocations.
We can improve this by using an xarray instead, which is efficient when
indices are densely clustered (such as inode numbers), is more cache
friendly and behaves like a resizable array, with a much better search
and insertion complexity than a red black tree. This only has one small
disadvantage which is that insertion will sometimes require allocating
memory for the xarray - which may fail (not that often since it uses a
kmem_cache) - but on the other hand we can reduce the btrfs inode
structure size by 24 bytes (from 1080 down to 1056 bytes) after removing
the embedded red black tree node, which after the next patches will allow
to reduce the size of the structure to 1024 bytes, meaning we will be able
to store 4 inodes per 4K page instead of 3 inodes.
This change does a straightforward change to use an xarray, and results
in a transaction abort if we can't allocate memory for the xarray when
creating an inode - but the next patch changes things so that we don't
need to abort.
Running the following fs_mark test showed some improvements:
$ cat test.sh
#!/bin/bash
DEV=/dev/nullb0
MNT=/mnt/nullb0
MOUNT_OPTIONS="-o ssd"
FILES=100000
THREADS=$(nproc --all)
echo "performance" | \
tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor
mkfs.btrfs -f $DEV
mount $MOUNT_OPTIONS $DEV $MNT
OPTS="-S 0 -L 5 -n $FILES -s 0 -t $THREADS -k"
for ((i = 1; i <= $THREADS; i++)); do
OPTS="$OPTS -d $MNT/d$i"
done
fs_mark $OPTS
umount $MNT
Before this patch:
FSUse% Count Size Files/sec App Overhead
10 1200000 0 92081.6 12505547
16 2400000 0 138222.6 13067072
23 3600000 0 148833.1 13290336
43 4800000 0 97864.7 13931248
53 6000000 0 85597.3 14384313
After this patch:
FSUse% Count Size Files/sec App Overhead
10 1200000 0 93225.1 12571078
16 2400000 0 146720.3 12805007
23 3600000 0 160626.4 13073835
46 4800000 0 116286.2 13802927
53 6000000 0 90087.9 14754892
The test was run with a release kernel config (Debian's default config).
Also capturing the insertion times into the rb tree and into the xarray,
that is measuring the duration of the old function inode_tree_add() and
the duration of the new btrfs_add_inode_to_root() function, gave the
following results (in nanoseconds):
Before this patch, inode_tree_add() execution times:
Count: 5000000
Range: 0.000 - 5536887.000; Mean: 775.674; Median: 729.000; Stddev: 4820.961
Percentiles: 90th: 1015.000; 95th: 1139.000; 99th: 1397.000
0.000 - 7.816: 40 |
7.816 - 37.858: 209 |
37.858 - 170.278: 6059 |
170.278 - 753.961: 2754890 #####################################################
753.961 - 3326.728: 2232312 ###########################################
3326.728 - 14667.018: 4366 |
14667.018 - 64652.943: 852 |
64652.943 - 284981.761: 550 |
284981.761 - 1256150.914: 221 |
1256150.914 - 5536887.000: 7 |
After this patch, btrfs_add_inode_to_root() execution times:
Count: 5000000
Range: 0.000 - 2900652.000; Mean: 272.148; Median: 241.000; Stddev: 2873.369
Percentiles: 90th: 342.000; 95th: 432.000; 99th: 572.000
0.000 - 7.264: 104 |
7.264 - 33.145: 352 |
33.145 - 140.081: 109606 #
140.081 - 581.930: 4840090 #####################################################
581.930 - 2407.590: 43532 |
2407.590 - 9950.979: 2245 |
9950.979 - 41119.278: 514 |
41119.278 - 169902.616: 155 |
169902.616 - 702018.539: 47 |
702018.539 - 2900652.000: 9 |
Average, percentiles, standard deviation, etc, are all much better.
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>
-----BEGIN PGP SIGNATURE-----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=7rUb
-----END PGP SIGNATURE-----
Merge tag 'for-6.10-rc2-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux
Pull btrfs fixes from David Sterba:
- fix handling of folio private changes.
The private value holds pointer to our extent buffer structure
representing a metadata range. Release and create of the range was
not properly synchronized when updating the private bit which ended
up in double folio_put, leading to all sorts of breakage
- fix a crash, reported as duplicate key in metadata, but caused by a
race of fsync and size extending write. Requires prealloc target
range + fsync and other conditions (log tree state, timing)
- fix leak of qgroup extent records after transaction abort
* tag 'for-6.10-rc2-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
btrfs: protect folio::private when attaching extent buffer folios
btrfs: fix leak of qgroup extent records after transaction abort
btrfs: fix crash on racing fsync and size-extending write into prealloc
Qgroup extent records are created when delayed ref heads are created and
then released after accounting extents at btrfs_qgroup_account_extents(),
called during the transaction commit path.
If a transaction is aborted we free the qgroup records by calling
btrfs_qgroup_destroy_extent_records() at btrfs_destroy_delayed_refs(),
unless we don't have delayed references. We are incorrectly assuming
that no delayed references means we don't have qgroup extents records.
We can currently have no delayed references because we ran them all
during a transaction commit and the transaction was aborted after that
due to some error in the commit path.
So fix this by ensuring we btrfs_qgroup_destroy_extent_records() at
btrfs_destroy_delayed_refs() even if we don't have any delayed references.
Reported-by: syzbot+0fecc032fa134afd49df@syzkaller.appspotmail.com
Link: https://lore.kernel.org/linux-btrfs/0000000000004e7f980619f91835@google.com/
Fixes: 81f7eb00ff ("btrfs: destroy qgroup extent records on transaction abort")
CC: stable@vger.kernel.org # 6.1+
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>
Replacement of bdev->bd_inode with sane(r) set of primitives.
-----BEGIN PGP SIGNATURE-----
iHUEABYIAB0WIQQqUNBr3gm4hGXdBJlZ7Krx/gZQ6wUCZkwjlgAKCRBZ7Krx/gZQ
66OmAP9nhZLASn/iM2+979I6O0GW+vid+uLh48uW3d+LbsmVIgD9GYpR+cuLQ/xj
mJESWfYKOVSpFFSrqlzKg9PQlU/GFgs=
=6LRp
-----END PGP SIGNATURE-----
Merge tag 'pull-bd_inode-1' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Pull bdev bd_inode updates from Al Viro:
"Replacement of bdev->bd_inode with sane(r) set of primitives by me and
Yu Kuai"
* tag 'pull-bd_inode-1' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
RIP ->bd_inode
dasd_format(): killing the last remaining user of ->bd_inode
nilfs_attach_log_writer(): use ->bd_mapping->host instead of ->bd_inode
block/bdev.c: use the knowledge of inode/bdev coallocation
gfs2: more obvious initializations of mapping->host
fs/buffer.c: massage the remaining users of ->bd_inode to ->bd_mapping
blk_ioctl_{discard,zeroout}(): we only want ->bd_inode->i_mapping here...
grow_dev_folio(): we only want ->bd_inode->i_mapping there
use ->bd_mapping instead of ->bd_inode->i_mapping
block_device: add a pointer to struct address_space (page cache of bdev)
missing helpers: bdev_unhash(), bdev_drop()
block: move two helpers into bdev.c
block2mtd: prevent direct access of bd_inode
dm-vdo: use bdev_nr_bytes(bdev) instead of i_size_read(bdev->bd_inode)
blkdev_write_iter(): saner way to get inode and bdev
bcachefs: remove dead function bdev_sectors()
ext4: remove block_device_ejected()
erofs_buf: store address_space instead of inode
erofs: switch erofs_bread() to passing offset instead of block number
Currently the error status of super block write is tracked in page/folio
status bit Error. For that we need to keep the reference for the whole
duration of write and wait.
Count the number of superblock writeback errors in the btrfs_device.
That means we don't need the folio to stay around until it's waited for,
and can avoid the extra call to folio_get/put.
Also remove a mention of PageError in a comment as it's the last mention
of the page Error state.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Iterate over folios instead of bvecs. Switch the order of unlock and put
to be the usual order; we know this folio can't be put until it's been
waited for, but that's fragile. Remove the calls to ClearPageUptodate /
SetPageUptodate -- if PAGE_SIZE is larger than BTRFS_SUPER_INFO_SIZE,
we'd be marking the entire folio uptodate without having actually
initialised all the bytes in the page.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This is a direct conversion from pages to folios, assuming single page
folio. Also removes some calls to obsolete APIs and some hidden calls to
compound_head().
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This is a direct conversion from pages to folios, assuming single page
folio. Also removes a few calls to compound_head() and calls to
obsolete APIs.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
There are open coded tests of BTRFS_FS_STATE_DUMMY_FS_INFO and we have a
wrapper for that that's a compile-time constant when self-tests are not
built in. As this is only for development we can save some bytes and
conditions on release configs by using the helper in the remaining
cases.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
There's no need to initialize the delayed inodes xarray with a GFP_ATOMIC
flag because that actually does nothing on the xarray operations. That was
needed for radix trees, but for xarrays the allocation flags are passed as
the last argument to xa_store() (which we are using correctly).
So initialize the delayed inodes xarray with a simple xa_init().
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>
Add a per cpu counter that tracks the total number of extent maps that are
in extent trees of inodes that belong to fs trees. This is going to be
used in an upcoming change that adds a shrinker for extent maps. Only
extent maps for fs trees are considered, because for special trees such as
the data relocation tree we don't want to evict their extent maps which
are critical for the relocation to work, and since those are limited, it's
not a concern to have them in memory during the relocation of a block
group. Another case are extent maps for free space cache inodes, which
must always remain in memory, but those are limited (there's only one per
free space cache inode, which means one per block group).
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>
A comment from Filipe on one of my previous cleanups brought my
attention to a new helper we have for getting the root id of a root,
which makes it easier to read in the code.
The changes where made with the following Coccinelle semantic patch:
// <smpl>
@@
expression E,E1;
@@
(
E->root_key.objectid = E1
|
- E->root_key.objectid
+ btrfs_root_id(E)
)
// </smpl>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ minor style fixups ]
Signed-off-by: David Sterba <dsterba@suse.com>
At warn_about_uncommitted_trans(), there's no need to check if the list
is empty and return, because list_for_each_entry_safe() is safe to call
for an empty list, it simply does nothing. So remove the check.
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>
Some of the operations after the free might convert more PERTRANS
metadata. Do the freeing as late as possible to eliminate a source of
leaked PERTRANS metadata.
This helps with the pass rate of generic/269 and generic/475.
Reviewed-by: Qu Wenruo <qwu@suse.com>
Signed-off-by: Boris Burkov <boris@bur.io>
Signed-off-by: David Sterba <dsterba@suse.com>
The helpers btrfs_del_delalloc_inode() and __btrfs_del_delalloc_inode()
don't follow the pattern when the "__" helper does a special case and
are in fact reversed regarding the naming. We can merge them into one as
there's only one place that needs to be open coded.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
There's no need to pass a root argument to __btrfs_del_delalloc_inode()
and btrfs_del_delalloc_inode(), we can just pass the inode since the root
is always the root associated to that inode. Some remove the root argument
from these functions.
Reviewed-by: Boris Burkov <boris@bur.io>
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>
Add a convenience helper to get a fs_info from a VFS inode pointer
instead of open coding the chain or using btrfs_sb() that in some cases
does one more pointer hop. This is implemented as a macro (still with
type checking) so we don't need full definitions of struct btrfs_inode,
btrfs_root or btrfs_fs_info.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Add convenience helpers to get a fs_info from a page or folio pointer
instead of open coding the chain or using btrfs_sb() that in some cases
does one more pointer hop. This is implemented as a macro (still with
type checking) so we don't need full definitions of struct page, folio,
btrfs_root and btrfs_fs_info. The latter can't be static inlines as this
would create loop between ctree.h <-> fs.h, or the headers would have to
be restructured.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Add convenience helpers to get a struct btrfs_inode from a page or folio
pointer instead of open coding the chain or intermediate BTRFS_I. This
is implemented as a macro (still with type checking) so we don't need
full definitions of struct page or address_space.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
There's one caller of btrfs_read_roots() and that already uses the
tree_root pointer, it's pointless to BUG_ON on it. As it's an assumption
of the initialization helpers make it an assert instead.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The btrfs_init_root_free_objectid() looks up a root by a key, allowing
to do an inexact search when key->offset is -1. It's never expected to
find such item, as it would break the allowed range of a root id.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
With help of neovim, LSP and clangd we can identify header files that
are not actually needed to be included in the .c files. This is focused
only on removal (with minor fixups), further cleanups are possible but
will require doing the header files properly with forward declarations,
minimized includes and include-what-you-use care.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The block size stored in the super block is used by subsystems outside
of btrfs and it's a copy of fs_info::sectorsize. Unify that to always
use our sectorsize, with the exception of mount where we first need to
use fixed values (4K) until we read the super block and can set the
sectorsize.
Replace all uses, in most cases it's fewer pointer indirections.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
fstests looks for WARN_ON's in dmesg. Add WARN_ON_ONCE() to our leak
detection code (enabled only in debug builds) so that fstests will fail
if these things trip at all. This will allow us to easily catch
problems with our reference counting that may otherwise go unnoticed.
Reviewed-by: Neal Gompa <neal@gompa.dev>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
After the conversion to folio interfaces (but without the patch to
enable larger folio allocation), there is an LTP report about observable
performance drop on metadata heavy operations.
https://lore.kernel.org/linux-btrfs/202312221750.571925bd-oliver.sang@intel.com/
This drop is caused by the extra code of calculating the
folio_size()/folio_shift(), instead of the old hard coded
PAGE_SIZE/PAGE_SHIFT.
To slightly reduce the overhead, just cache both folio_size and
folio_shift in extent_buffer.
The two new members (u32 folio_size and u8 folio_shift) are stored
inside the holes of extent_buffer. folio_size is shared with len, which
is reduced to u32. The size of eb does not change.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When creating a snapshot we may do a double free of an anonymous device
in case there's an error committing the transaction. The second free may
result in freeing an anonymous device number that was allocated by some
other subsystem in the kernel or another btrfs filesystem.
The steps that lead to this:
1) At ioctl.c:create_snapshot() we allocate an anonymous device number
and assign it to pending_snapshot->anon_dev;
2) Then we call btrfs_commit_transaction() and end up at
transaction.c:create_pending_snapshot();
3) There we call btrfs_get_new_fs_root() and pass it the anonymous device
number stored in pending_snapshot->anon_dev;
4) btrfs_get_new_fs_root() frees that anonymous device number because
btrfs_lookup_fs_root() returned a root - someone else did a lookup
of the new root already, which could some task doing backref walking;
5) After that some error happens in the transaction commit path, and at
ioctl.c:create_snapshot() we jump to the 'fail' label, and after
that we free again the same anonymous device number, which in the
meanwhile may have been reallocated somewhere else, because
pending_snapshot->anon_dev still has the same value as in step 1.
Recently syzbot ran into this and reported the following trace:
------------[ cut here ]------------
ida_free called for id=51 which is not allocated.
WARNING: CPU: 1 PID: 31038 at lib/idr.c:525 ida_free+0x370/0x420 lib/idr.c:525
Modules linked in:
CPU: 1 PID: 31038 Comm: syz-executor.2 Not tainted 6.8.0-rc4-syzkaller-00410-gc02197fc9076 #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/25/2024
RIP: 0010:ida_free+0x370/0x420 lib/idr.c:525
Code: 10 42 80 3c 28 (...)
RSP: 0018:ffffc90015a67300 EFLAGS: 00010246
RAX: be5130472f5dd000 RBX: 0000000000000033 RCX: 0000000000040000
RDX: ffffc90009a7a000 RSI: 000000000003ffff RDI: 0000000000040000
RBP: ffffc90015a673f0 R08: ffffffff81577992 R09: 1ffff92002b4cdb4
R10: dffffc0000000000 R11: fffff52002b4cdb5 R12: 0000000000000246
R13: dffffc0000000000 R14: ffffffff8e256b80 R15: 0000000000000246
FS: 00007fca3f4b46c0(0000) GS:ffff8880b9500000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f167a17b978 CR3: 000000001ed26000 CR4: 0000000000350ef0
Call Trace:
<TASK>
btrfs_get_root_ref+0xa48/0xaf0 fs/btrfs/disk-io.c:1346
create_pending_snapshot+0xff2/0x2bc0 fs/btrfs/transaction.c:1837
create_pending_snapshots+0x195/0x1d0 fs/btrfs/transaction.c:1931
btrfs_commit_transaction+0xf1c/0x3740 fs/btrfs/transaction.c:2404
create_snapshot+0x507/0x880 fs/btrfs/ioctl.c:848
btrfs_mksubvol+0x5d0/0x750 fs/btrfs/ioctl.c:998
btrfs_mksnapshot+0xb5/0xf0 fs/btrfs/ioctl.c:1044
__btrfs_ioctl_snap_create+0x387/0x4b0 fs/btrfs/ioctl.c:1306
btrfs_ioctl_snap_create_v2+0x1ca/0x400 fs/btrfs/ioctl.c:1393
btrfs_ioctl+0xa74/0xd40
vfs_ioctl fs/ioctl.c:51 [inline]
__do_sys_ioctl fs/ioctl.c:871 [inline]
__se_sys_ioctl+0xfe/0x170 fs/ioctl.c:857
do_syscall_64+0xfb/0x240
entry_SYSCALL_64_after_hwframe+0x6f/0x77
RIP: 0033:0x7fca3e67dda9
Code: 28 00 00 00 (...)
RSP: 002b:00007fca3f4b40c8 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
RAX: ffffffffffffffda RBX: 00007fca3e7abf80 RCX: 00007fca3e67dda9
RDX: 00000000200005c0 RSI: 0000000050009417 RDI: 0000000000000003
RBP: 00007fca3e6ca47a R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
R13: 000000000000000b R14: 00007fca3e7abf80 R15: 00007fff6bf95658
</TASK>
Where we get an explicit message where we attempt to free an anonymous
device number that is not currently allocated. It happens in a different
code path from the example below, at btrfs_get_root_ref(), so this change
may not fix the case triggered by syzbot.
To fix at least the code path from the example above, change
btrfs_get_root_ref() and its callers to receive a dev_t pointer argument
for the anonymous device number, so that in case it frees the number, it
also resets it to 0, so that up in the call chain we don't attempt to do
the double free.
CC: stable@vger.kernel.org # 5.10+
Link: https://lore.kernel.org/linux-btrfs/000000000000f673a1061202f630@google.com/
Fixes: e03ee2fe87 ("btrfs: do not ASSERT() if the newly created subvolume already got read")
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
[BUG]
There is a syzbot crash, triggered by the ASSERT() during subvolume
creation:
assertion failed: !anon_dev, in fs/btrfs/disk-io.c:1319
------------[ cut here ]------------
kernel BUG at fs/btrfs/disk-io.c:1319!
invalid opcode: 0000 [#1] PREEMPT SMP KASAN
RIP: 0010:btrfs_get_root_ref.part.0+0x9aa/0xa60
<TASK>
btrfs_get_new_fs_root+0xd3/0xf0
create_subvol+0xd02/0x1650
btrfs_mksubvol+0xe95/0x12b0
__btrfs_ioctl_snap_create+0x2f9/0x4f0
btrfs_ioctl_snap_create+0x16b/0x200
btrfs_ioctl+0x35f0/0x5cf0
__x64_sys_ioctl+0x19d/0x210
do_syscall_64+0x3f/0xe0
entry_SYSCALL_64_after_hwframe+0x63/0x6b
---[ end trace 0000000000000000 ]---
[CAUSE]
During create_subvol(), after inserting root item for the newly created
subvolume, we would trigger btrfs_get_new_fs_root() to get the
btrfs_root of that subvolume.
The idea here is, we have preallocated an anonymous device number for
the subvolume, thus we can assign it to the new subvolume.
But there is really nothing preventing things like backref walk to read
the new subvolume.
If that happens before we call btrfs_get_new_fs_root(), the subvolume
would be read out, with a new anonymous device number assigned already.
In that case, we would trigger ASSERT(), as we really expect no one to
read out that subvolume (which is not yet accessible from the fs).
But things like backref walk is still possible to trigger the read on
the subvolume.
Thus our assumption on the ASSERT() is not correct in the first place.
[FIX]
Fix it by removing the ASSERT(), and just free the @anon_dev, reset it
to 0, and continue.
If the subvolume tree is read out by something else, it should have
already get a new anon_dev assigned thus we only need to free the
preallocated one.
Reported-by: Chenyuan Yang <chenyuan0y@gmail.com>
Fixes: 2dfb1e43f5 ("btrfs: preallocate anon block device at first phase of snapshot creation")
CC: stable@vger.kernel.org # 5.15+
Reviewed-by: Filipe Manana <fdmanana@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>
[BUG]
Test case btrfs/124 failed if larger metadata folio is enabled, the
dying message looks like this:
BTRFS error (device dm-2): bad tree block start, mirror 2 want 31686656 have 0
BTRFS info (device dm-2): read error corrected: ino 0 off 31686656 (dev /dev/mapper/test-scratch2 sector 20928)
BUG: kernel NULL pointer dereference, address: 0000000000000020
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
CPU: 6 PID: 350881 Comm: btrfs Tainted: G OE 6.7.0-rc3-custom+ #128
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS unknown 2/2/2022
RIP: 0010:btrfs_read_extent_buffer+0x106/0x180 [btrfs]
PKRU: 55555554
Call Trace:
<TASK>
read_tree_block+0x33/0xb0 [btrfs]
read_block_for_search+0x23e/0x340 [btrfs]
btrfs_search_slot+0x2f9/0xe60 [btrfs]
btrfs_lookup_csum+0x75/0x160 [btrfs]
btrfs_lookup_bio_sums+0x21a/0x560 [btrfs]
btrfs_submit_chunk+0x152/0x680 [btrfs]
btrfs_submit_bio+0x1c/0x50 [btrfs]
submit_one_bio+0x40/0x80 [btrfs]
submit_extent_page+0x158/0x390 [btrfs]
btrfs_do_readpage+0x330/0x740 [btrfs]
extent_readahead+0x38d/0x6c0 [btrfs]
read_pages+0x94/0x2c0
page_cache_ra_unbounded+0x12d/0x190
relocate_file_extent_cluster+0x7c1/0x9d0 [btrfs]
relocate_block_group+0x2d3/0x560 [btrfs]
btrfs_relocate_block_group+0x2c7/0x4b0 [btrfs]
btrfs_relocate_chunk+0x4c/0x1a0 [btrfs]
btrfs_balance+0x925/0x13c0 [btrfs]
btrfs_ioctl+0x19f1/0x25d0 [btrfs]
__x64_sys_ioctl+0x90/0xd0
do_syscall_64+0x3f/0xf0
entry_SYSCALL_64_after_hwframe+0x6e/0x76
[CAUSE]
The dying line is at btrfs_repair_io_failure() call inside
btrfs_repair_eb_io_failure().
The function is still relying on the extent buffer using page sized
folios.
When the extent buffer is using larger folio, we go into the 2nd slot of
folios[], and triggered the NULL pointer dereference.
[FIX]
Migrate btrfs_repair_io_failure() to folio interfaces.
So that when we hit a larger folio, we just submit the whole folio in
one go.
This also affects data repair path through btrfs_end_repair_bio(),
thankfully data is still fully page based, we can just add an
ASSERT(), and use page_folio() to convert the page to folio.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Although subpage itself is conflicting with higher folio, since subpage
(sectorsize < PAGE_SIZE and nodesize < PAGE_SIZE) means we will never
need higher order folio, there is a hidden pitfall:
- btrfs_page_*() helpers
Those helpers are an abstraction to handle both subpage and non-subpage
cases, which means we're going to pass pages pointers to those helpers.
And since those helpers are shared between data and metadata paths, it's
unavoidable to let them to handle folios, including higher order
folios).
Meanwhile for true subpage case, we should only have a single page
backed folios anyway, thus add a new ASSERT() for btrfs_subpage_assert()
to ensure that.
Also since those helpers are shared between both data and metadata, add
some extra ASSERT()s for data path to make sure we only get single page
backed folio for now.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
These two functions are still using the old page based code, which is
not going to handle larger folios at all.
The migration itself is going to involve the following changes:
- PAGE_SIZE -> folio_size()
- PAGE_SHIFT -> folio_shift()
- get_eb_page_index() -> get_eb_folio_index()
- get_eb_offset_in_page() -> get_eb_offset_in_folio()
And since we're going to support larger folios, although above straight
conversion is good enough, this patch would add extra comments in the
involved functions to explain why the same single line code can now
cover 3 cases:
- folio_size == PAGE_SIZE, sectorsize == PAGE_SIZE, nodesize >= PAGE_SIZE
The common, non-subpage case with per-page folio.
- folio_size > PAGE_SIZE, sectorsize == PAGE_SIZE, nodesize >= PAGE_SIZE
The incoming larger folio, non-subpage case.
- folio_size == PAGE_SIZE, sectorsize < PAGE_SIZE, nodesize < PAGE_SIZE
The existing subpage case, we won't larger folio anyway.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Although we have migrated extent_buffer::pages[] to folios[], we're
still mostly using the folio_page() help to grab the page.
This patch would do the following cleanups for metadata:
- Introduce num_extent_folios() helper
This is to replace most num_extent_pages() callers.
- Use num_extent_folios() to iterate future large folios
This allows us to use things like
bio_add_folio()/bio_add_folio_nofail(), and only set the needed flags
for the folio (aka the leading/tailing page), which reduces the loop
iteration to 1 for large folios.
- Change metadata related functions to use folio pointers
Including their function name, involving:
* attach_extent_buffer_page()
* detach_extent_buffer_page()
* page_range_has_eb()
* btrfs_release_extent_buffer_pages()
* btree_clear_page_dirty()
* btrfs_page_inc_eb_refs()
* btrfs_page_dec_eb_refs()
- Change btrfs_is_subpage() to accept an address_space pointer
This is to allow both page->mapping and folio->mapping to be utilized.
As data is still using the old per-page code, and may keep so for a
while.
- Special corner case place holder for future order mismatches between
extent buffer and inode filemap
For now it's just a block of comments and a dead ASSERT(), no real
handling yet.
The subpage code would still go page, just because subpage and large
folio are conflicting conditions, thus we don't need to bother subpage
with higher order folios at all. Just folio_page(folio, 0) would be
enough.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ minor styling tweaks ]
Signed-off-by: David Sterba <dsterba@suse.com>
For now extent_buffer::pages[] are still only accepting single page
pointer, thus we can migrate to folios pretty easily.
As for single page, page and folio are 1:1 mapped, including their page
flags.
This patch would just do the conversion from struct page to struct
folio, providing the first step to higher order folio in the future.
This conversion is pretty simple:
- extent_buffer::pages[] -> extent_buffer::folios[]
- page_address(eb->pages[i]) -> folio_address(eb->pages[i])
- eb->pages[i] -> folio_page(eb->folios[i], 0)
There would be more specific cleanups preparing for the incoming higher
order folio support.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The radix-tree has been superseded by the xarray
(https://lwn.net/Articles/745073), this patch converts the
btrfs_root::delayed_nodes, the APIs are used in a simple way.
First idea is to do xa_insert() but this would require GFP_ATOMIC
allocation which we want to avoid if possible. The preload mechanism of
radix-tree can be emulated within the xarray API.
- xa_reserve() with GFP_NOFS outside of the lock, the reserved entry
is inserted atomically at most once
- xa_store() under a lock, in case something races in we can detect that
and xa_load() returns a valid pointer
All uses of xa_load() must check for a valid pointer in case they manage
to get between the xa_reserve() and xa_store(), this is handled in
btrfs_get_delayed_node().
Otherwise the functionality is equivalent, xarray implements the
radix-tree and there should be no performance difference.
The patch continues the efforts started in 253bf57555 ("btrfs: turn
delayed_nodes_tree into an XArray") and fixes the problems with locking
and GFP flags 088aea3b97 ("Revert "btrfs: turn delayed_nodes_tree
into an XArray"").
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We're currently setting this when we try to load the roots and we see
that usebackuproot is set. Instead set this at mount option parsing
time.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Acked-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
There's no reason this has to happen in open_ctree, and in fact in the
old mount API we had to call this from remount. Move this to super.c,
unexport it, and call it from both mount and reconfigure.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Acked-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We add these mount options based on the fs_devices settings, which can
be set once we've opened the fs_devices. Move these into their own
helper and call it from get_tree_super.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Acked-by: Christian Brauner <brauner@kernel.org>
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 have all of the parts in place to use the new mount API,
switch our fs_type to use the new callbacks.
There are a few things that have to be done at the same time because of
the order of operations changes that come along with the new mount API.
These must be done in the same patch otherwise things will go wrong.
1. Export and use btrfs_check_options in open_ctree(). This is because
the options are done ahead of time, and we need to check them once we
have the feature flags loaded.
2. Update the free space cache settings. Since we're coming in with the
options already set we need to make sure we don't undo what the user
has asked for.
3. Set our sb_flags at init_fs_context time, the fs_context stuff is
trying to manage the sb_flagss itself, so move that into
init_fs_context and out of the fill super part.
Additionally I've marked the unused functions with __maybe_unused and
will remove them in a future patch.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Acked-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
With the old mount API we'd pre-populate the mount options with the
space cache settings of the file system, and then the user toggled them
on or off with the mount options. When we switch to the new mount API
the mount options will be set before we get into opening the file
system, so we need a flag to indicate that the user explicitly asked for
-o nospace_cache so we can make the appropriate changes after the fact.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Acked-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We currently don't allow these options to be set if we're extent tree v2
via the mount option parsing. However when we switch to the new mount
API we'll no longer have the super block loaded, so won't be able to
make this distinction at mount option parsing time. Address this by
checking for extent tree v2 at the point where we make the decision to
rebuild the free space tree.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Acked-by: Christian Brauner <brauner@kernel.org>
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 pre-load the space cache settings in btrfs_parse_options,
however when we switch to the new mount API the mount option parsing
will happen before we have the super block loaded. Add a helper to set
the appropriate options based on the fs settings, this will allow us to
have consistent free space cache settings.
This also folds in the space cache related decisions we make for subpage
sectorsize support, so all of this is done in one place.
Since this was being called by parse options it looks like we're
changing the behavior of remount, but in fact we aren't. The
pre-loading of the free space cache settings is done because we want to
handle the case of users not using any space_cache options, we'll derive
the appropriate mount option based on the on disk state. On remount
this wouldn't reset anything as we'll have cleared the v1 cache
generation if we mounted -o nospace_cache. Similarly it's impossible to
turn off the free space tree without specifically saying -o
nospace_cache,clear_cache, which will delete the free space tree and
clear the compat_ro option. Again in this case calling this code in
remount wouldn't result in any change.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Acked-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
With the new mount API we'll be setting our compression well before we
call open_ctree. We don't want to overwrite our settings, so set the
default in btrfs_init_fs_info instead of open_ctree.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Acked-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently btrfs extent buffer helpers are doing all the cross-page
handling, as there is no guarantee that all those eb pages are
contiguous.
However on systems with enough memory, there is a very high chance the
page cache for btree_inode are allocated with physically contiguous
pages.
In that case, we can skip all the complex cross-page handling, thus
speeding up the code.
This patch adds a new member, extent_buffer::addr, which is only set to
non-NULL if all the extent buffer pages are physically contiguous.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
One a zoned filesystem, never clear the dirty flag of an extent buffer,
but instead mark it as zeroout.
On writeout, when encountering a marked extent_buffer, zero it out.
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
EXTENT_BUFFER_ZONED_ZEROOUT better describes the state of the extent buffer,
namely it is written as all zeros. This is needed in zoned mode, to
preserve I/O ordering.
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently we abuse the extent_map structure for two purposes:
1) To actually represent extents for inodes;
2) To represent chunk mappings.
This is odd and has several disadvantages:
1) To create a chunk map, we need to do two memory allocations: one for
an extent_map structure and another one for a map_lookup structure, so
more potential for an allocation failure and more complicated code to
manage and link two structures;
2) For a chunk map we actually only use 3 fields (24 bytes) of the
respective extent map structure: the 'start' field to have the logical
start address of the chunk, the 'len' field to have the chunk's size,
and the 'orig_block_len' field to contain the chunk's stripe size.
Besides wasting a memory, it's also odd and not intuitive at all to
have the stripe size in a field named 'orig_block_len'.
We are also using 'block_len' of the extent_map structure to contain
the chunk size, so we have 2 fields for the same value, 'len' and
'block_len', which is pointless;
3) When an extent map is associated to a chunk mapping, we set the bit
EXTENT_FLAG_FS_MAPPING on its flags and then make its member named
'map_lookup' point to the associated map_lookup structure. This means
that for an extent map associated to an inode extent, we are not using
this 'map_lookup' pointer, so wasting 8 bytes (on a 64 bits platform);
4) Extent maps associated to a chunk mapping are never merged or split so
it's pointless to use the existing extent map infrastructure.
So add a dedicated data structure named 'btrfs_chunk_map' to represent
chunk mappings, this is basically the existing map_lookup structure with
some extra fields:
1) 'start' to contain the chunk logical address;
2) 'chunk_len' to contain the chunk's length;
3) 'stripe_size' for the stripe size;
4) 'rb_node' for insertion into a rb tree;
5) 'refs' for reference counting.
This way we do a single memory allocation for chunk mappings and we don't
waste memory for them with unused/unnecessary fields from an extent_map.
We also save 8 bytes from the extent_map structure by removing the
'map_lookup' pointer, so the size of struct extent_map is reduced from
144 bytes down to 136 bytes, and we can now have 30 extents map per 4K
page instead of 28.
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>
The logged_list[2] and log_extents_lock[2] members of struct btrfs_root
are no longer used, their last use was removed in commit 5636cf7d6d
("btrfs: remove the logged extents infrastructure"). So remove these
fields. This reduces the size of struct btrfs_root, on a release kernel,
from 1392 bytes down to 1352 bytes.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
If we abort a transaction, we never run the code that frees the pertrans
qgroup reservation. This results in warnings on unmount as that
reservation has been leaked. The leak isn't a huge issue since the fs is
read-only, but it's better to clean it up when we know we can/should. Do
it during the cleanup_transaction step of aborting.
CC: stable@vger.kernel.org # 5.15+
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Boris Burkov <boris@bur.io>
Signed-off-by: David Sterba <dsterba@suse.com>
There is a feature request to add dmesg output when unmounting a btrfs.
There are several alternative methods to do the same thing, but with
their own problems:
- Use eBPF to watch btrfs_put_super()/open_ctree()
Not end user friendly, they have to dip their head into the source
code.
- Watch for directory /sys/fs/<uuid>/
This is way more simple, but still requires some simple device -> uuid
lookups. And a script needs to use inotify to watch /sys/fs/.
Compared to all these, directly outputting the information into dmesg
would be the most simple one, with both device and UUID included.
And since we're here, also add the output when mounting a filesystem for
the first time for parity. A more fine grained monitoring of subvolume
mounts should be done by another layer, like audit.
Now mounting a btrfs with all default mkfs options would look like this:
[81.906566] BTRFS info (device dm-8): first mount of filesystem 633b5c16-afe3-4b79-b195-138fe145e4f2
[81.907494] BTRFS info (device dm-8): using crc32c (crc32c-intel) checksum algorithm
[81.908258] BTRFS info (device dm-8): using free space tree
[81.912644] BTRFS info (device dm-8): auto enabling async discard
[81.913277] BTRFS info (device dm-8): checking UUID tree
[91.668256] BTRFS info (device dm-8): last unmount of filesystem 633b5c16-afe3-4b79-b195-138fe145e4f2
CC: stable@vger.kernel.org # 5.4+
Link: https://github.com/kdave/btrfs-progs/issues/689
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ update changelog ]
Signed-off-by: David Sterba <dsterba@suse.com>
Currently the last_trans_committed field of struct btrfs_fs_info is
modified and read without any locking or other protection. For example
early in the fsync path, skip_inode_logging() is called which reads
fs_info->last_trans_committed, but at the same time we can have a
transaction commit completing and updating that field.
In the case of an fsync this is harmless and any data race should be
rare and at most cause an unnecessary logging of an inode.
To avoid data race warnings from tools like KCSAN and other issues such
as load and store tearing (amongst others, see [1]), create helpers to
access the last_trans_committed field of struct btrfs_fs_info using
READ_ONCE() and WRITE_ONCE(), and use these helpers everywhere.
[1] https://lwn.net/Articles/793253/
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently the log_transid field of a root is always modified while holding
the root's log_mutex locked. Most readers of a root's log_transid are also
holding the root's log_mutex locked, however there is one exception which
is btrfs_set_inode_last_trans() where we don't take the lock to avoid
blocking several operations if log syncing is happening in parallel.
Any races here should be harmless, and in the worst case they may cause a
fsync to log an inode when it's not really needed, so nothing bad from a
functional perspective.
To avoid data race warnings from tools like KCSAN and other issues such
as load and store tearing (amongst others, see [1]), create helpers to
access the log_transid field of a root using READ_ONCE() and WRITE_ONCE(),
and use these helpers where needed.
[1] https://lwn.net/Articles/793253/
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently, the last_log_commit of a root can be accessed concurrently
without any lock protection. Readers can be calling btrfs_inode_in_log()
early in a fsync call, which reads a root's last_log_commit, while a
writer can change the last_log_commit while a log tree if being synced,
at btrfs_sync_log(). Any races here should be harmless, and in the worst
case they may cause a fsync to log an inode when it's not really needed,
so nothing bad from a functional perspective.
To avoid data race warnings from tools like KCSAN and other issues such
as load and store tearing (amongst others, see [1]), create helpers to
access the last_log_commit field of a root using READ_ONCE() and
WRITE_ONCE(), and use these helpers everywhere.
[1] https://lwn.net/Articles/793253/
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Guilherme's previous work [1] aimed at the mounting of cloned devices
using a superblock flag SINGLE_DEV during mkfs.
[1] https://lore.kernel.org/linux-btrfs/20230831001544.3379273-1-gpiccoli@igalia.com/
Building upon this work, here is in memory only approach. As it mounts
we determine if the same fsid is already mounted if then we generate a
random temp fsid which shall be used the mount, in memory only not
written to the disk. We distinguish devices by devt.
Example:
$ fallocate -l 300m ./disk1.img
$ mkfs.btrfs -f ./disk1.img
$ cp ./disk1.img ./disk2.img
$ cp ./disk1.img ./disk3.img
$ mount -o loop ./disk1.img /btrfs
$ mount -o ./disk2.img /btrfs1
$ mount -o ./disk3.img /btrfs2
$ btrfs fi show -m
Label: none uuid: 4a212b48-1bec-46a5-938a-783c8c1f0b02
Total devices 1 FS bytes used 144.00KiB
devid 1 size 300.00MiB used 88.00MiB path /dev/loop0
Label: none uuid: adabf2fe-5515-4ad0-95b4-7b1609218c16
Total devices 1 FS bytes used 144.00KiB
devid 1 size 300.00MiB used 88.00MiB path /dev/loop1
Label: none uuid: 1d77d0df-7d92-439e-adbd-20b9b86fdedb
Total devices 1 FS bytes used 144.00KiB
devid 1 size 300.00MiB used 88.00MiB path /dev/loop2
Co-developed-by: Guilherme G. Piccoli <gpiccoli@igalia.com>
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Space for block group item updates, necessary after allocating or
deallocating an extent from a block group, is reserved in the delayed
refs block reserve. Currently we do this by incrementing the transaction
handle's delayed_ref_updates counter and then calling
btrfs_update_delayed_refs_rsv(), which will increase the size of the
delayed refs block reserve by an amount that corresponds to the same
amount we use for delayed refs, given by btrfs_calc_delayed_ref_bytes().
That is an excessive amount because it corresponds to the amount of space
needed to insert one item in a btree (btrfs_calc_insert_metadata_size())
times 2 when the free space tree feature is enabled. All we need is an
amount as given by btrfs_calc_metadata_size(), since we only need to
update an existing block group item in the extent tree (or block group
tree if this feature is enabled). By using btrfs_calc_metadata_size() we
will need to reserve 4 times less space when using the free space tree
and 2 times less space when not using it, putting less pressure on space
reservation.
So use helpers to reserve and release space for block group item updates
that use btrfs_calc_metadata_size() for calculation of the space.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The BTRFS_SUPER_FLAG_CHANGING_FSID_V2 flag indicates a transient state
where the device in the userspace btrfstune -m|-M operation failed to
complete changing the fsid.
This flag makes the kernel to automatically determine the other
partner devices to which a given device can be associated, based on the
fsid, metadata_uuid and generation values.
btrfstune -m|M feature is especially useful in virtual cloud setups, where
compute instances (disk images) are quickly copied, fsid changed, and
launched. Given numerous disk images with the same metadata_uuid but
different fsid, there's no clear way a device can be correctly assembled
with the proper partners when the CHANGING_FSID_V2 flag is set. So, the
disk could be assembled incorrectly, as in the example below:
Before this patch:
Consider the following two filesystems:
/dev/loop[2-3] are raw copies of /dev/loop[0-1] and the btrsftune -m
operation fails.
In this scenario, as the /dev/loop0's fsid change is interrupted, and the
CHANGING_FSID_V2 flag is set as shown below.
$ p="device|devid|^metadata_uuid|^fsid|^incom|^generation|^flags"
$ btrfs inspect dump-super /dev/loop0 | egrep '$p'
superblock: bytenr=65536, device=/dev/loop0
flags 0x1000000001
fsid 7d4b4b93-2b27-4432-b4e4-4be1fbccbd45
metadata_uuid bb040a9f-233a-4de2-ad84-49aa5a28059b
generation 9
num_devices 2
incompat_flags 0x741
dev_item.devid 1
$ btrfs inspect dump-super /dev/loop1 | egrep '$p'
superblock: bytenr=65536, device=/dev/loop1
flags 0x1
fsid 11d2af4d-1b71-45a9-83f6-f2100766939d
metadata_uuid bb040a9f-233a-4de2-ad84-49aa5a28059b
generation 10
num_devices 2
incompat_flags 0x741
dev_item.devid 2
$ btrfs inspect dump-super /dev/loop2 | egrep '$p'
superblock: bytenr=65536, device=/dev/loop2
flags 0x1
fsid 7d4b4b93-2b27-4432-b4e4-4be1fbccbd45
metadata_uuid bb040a9f-233a-4de2-ad84-49aa5a28059b
generation 8
num_devices 2
incompat_flags 0x741
dev_item.devid 1
$ btrfs inspect dump-super /dev/loop3 | egrep '$p'
superblock: bytenr=65536, device=/dev/loop3
flags 0x1
fsid 7d4b4b93-2b27-4432-b4e4-4be1fbccbd45
metadata_uuid bb040a9f-233a-4de2-ad84-49aa5a28059b
generation 8
num_devices 2
incompat_flags 0x741
dev_item.devid 2
It is normal that some devices aren't instantly discovered during
system boot or iSCSI discovery. The controlled scan below demonstrates
this.
$ btrfs device scan --forget
$ btrfs device scan /dev/loop0
Scanning for btrfs filesystems on '/dev/loop0'
$ mount /dev/loop3 /btrfs
$ btrfs filesystem show -m
Label: none uuid: 7d4b4b93-2b27-4432-b4e4-4be1fbccbd45
Total devices 2 FS bytes used 144.00KiB
devid 1 size 300.00MiB used 48.00MiB path /dev/loop0
devid 2 size 300.00MiB used 40.00MiB path /dev/loop3
/dev/loop0 and /dev/loop3 are incorrectly partnered.
This kernel patch removes functions and code connected to the
CHANGING_FSID_V2 flag.
With this patch, now devices with the CHANGING_FSID_V2 flag are rejected.
And its partner will fail to mount with the extra -o degraded option.
The check is removed from open_ctree(), devices are rejected during
scanning which in turn fails the mount.
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
In open_ctree, we set BTRFS_FS_QUOTA_ENABLED as soon as we see a
quota_root, as opposed to after we are done setting up the qgroup
structures. In the quota_enable path, we wait until after the structures
are set up. Likewise, in disable, we clear the bit before tearing down
the structures. I feel that this organization is less surprising for the
open_ctree path.
I don't believe this fixes any actual bug, but avoids potential
confusion when using btrfs_qgroup_mode in an intermediate state where we
are enabled but haven't yet setup the qgroup status flags. It also
avoids any risk of calling a qgroup function and attempting to use the
qgroup rbtrees before they exist/are setup.
This all occurs before we do rw setup, so I believe it should be mostly
a no-op.
Signed-off-by: Boris Burkov <boris@bur.io>
Signed-off-by: David Sterba <dsterba@suse.com>
Relocation COWs metadata blocks in two cases for the reloc root:
- copying the subvolume root item when creating the reloc root
- copying a btree node when there is a COW during relocation
In both cases, the resulting btree node hits an abnormal code path with
respect to the owner field in its btrfs_header. It first creates the
root item for the new objectid, which populates the reloc root id, and
it at this point that delayed refs are created.
Later, it fully copies the old node into the new node (including the
original owner field) which overwrites it. This results in a simple
quotas mismatch where we run the delayed ref for the reloc root which
has no simple quota effect (reloc root is not an fstree) but when we
ultimately delete the node, the owner is the real original fstree and we
do free the space.
To work around this without tampering with the behavior of relocation,
add a parameter to btrfs_add_tree_block that lets the relocation code
path specify a different owning root than the "operating" root (in this
case, owning root is the real root and the operating root is the reloc
root). These can naturally be plumbed into delayed refs that have the
same concept.
Note that this is a double count in some sense, but a relatively natural
one, as there are really two extents, and the old one will be deleted
soon. This is consistent with how data relocation extents are accounted
by simple quotas.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Boris Burkov <boris@bur.io>
Signed-off-by: David Sterba <dsterba@suse.com>
If we find the raid-stripe-tree on mount, read it from disk. This is
a backward incompatible feature. The rescue=ignorebadroots mount option
will skip this tree.
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
At btrfs_mark_buffer_dirty(), having a transaction id mismatch is never
expected to happen and it usually means there's a bug or some memory
corruption due to a bitflip for example. So mark the condition as unlikely
to optimize code generation as well as to make it obvious for human
readers that it is a very unexpected condition.
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 need to use WARN() at btrfs_mark_buffer_dirty() to print an
error message, as we have the fs_info pointer we can use btrfs_crit()
which prints device information and makes the message have a more uniform
format. As we are already aborting the transaction we already have a stack
trace printed as well. So replace the use of WARN() with btrfs_crit().
Also slightly reword the message to use 'logical' instead of 'block' as
it's what is used in other error/warning messages.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When marking an extent buffer as dirty, at btrfs_mark_buffer_dirty(),
we check if its generation matches the running transaction and if not we
just print a warning. Such mismatch is an indicator that something really
went wrong and only printing a warning message (and stack trace) is not
enough to prevent a corruption. Allowing a transaction to commit with such
an extent buffer will trigger an error if we ever try to read it from disk
due to a generation mismatch with its parent generation.
So abort the current transaction with -EUCLEAN if we notice a generation
mismatch. For this we need to pass a transaction handle to
btrfs_mark_buffer_dirty() which is always available except in test code,
in which case we can pass NULL since it operates on dummy extent buffers
and all test roots have a single node/leaf (root node at level 0).
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently when reserving space for deleting the csum items for a data
extent, when adding or updating a delayed ref head, we determine how
many leaves of csum items we can have and then pass that number to the
helper btrfs_calc_delayed_ref_bytes(). This helper is used for calculating
space for all tree modifications we need when running delayed references,
however the amount of space it computes is excessive for deleting csum
items because:
1) It uses btrfs_calc_insert_metadata_size() which is excessive because
we only need to delete csum items from the csum tree, we don't need
to insert any items, so btrfs_calc_metadata_size() is all we need (as
it computes space needed to delete an item);
2) If the free space tree is enabled, it doubles the amount of space,
which is pointless for csum deletion since we don't need to touch the
free space tree or any other tree other than the csum tree.
So improve on this by tracking how many csum deletions we have and using
a new helper to calculate space for csum deletions (just a wrapper around
btrfs_calc_metadata_size() with a comment). This reduces the amount of
space we need to reserve for csum deletions by a factor of 4, and it helps
reduce the number of times we have to block space reservations and have
the reclaim task enter the space flushing algorithm (flush delayed items,
flush delayed refs, etc) in order to satisfy tickets.
For example this results in a total time decrease when unlinking (or
truncating) files with many extents, as we end up having to block on space
metadata reservations less often. Example test:
$ cat test.sh
#!/bin/bash
DEV=/dev/nullb0
MNT=/mnt/test
umount $DEV &> /dev/null
mkfs.btrfs -f $DEV
# Use compression to quickly create files with a lot of extents
# (each with a size of 128K).
mount -o compress=lzo $DEV $MNT
# 100G gives at least 983040 extents with a size of 128K.
xfs_io -f -c "pwrite -S 0xab -b 1M 0 120G" $MNT/foobar
# Flush all delalloc and clear all metadata from memory.
umount $MNT
mount -o compress=lzo $DEV $MNT
start=$(date +%s%N)
rm -f $MNT/foobar
end=$(date +%s%N)
dur=$(( (end - start) / 1000000 ))
echo "rm took $dur milliseconds"
umount $MNT
Before this change rm took: 7504 milliseconds
After this change rm took: 6574 milliseconds (-12.4%)
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently when reserving space for delayed refs we do it on a per ref head
basis. This is generally enough because most back refs for an extent end
up being inlined in the extent item - with the default leaf size of 16K we
can have at most 33 inline back refs (this is calculated by the macro
BTRFS_MAX_EXTENT_ITEM_SIZE()). The amount of bytes reserved for each ref
head is given by btrfs_calc_delayed_ref_bytes(), which basically
corresponds to a single path for insertion into the extent tree plus
another path for insertion into the free space tree if it's enabled.
However if we have reached the limit of inline refs or we have a mix of
inline and non-inline refs, then we will need to insert a non-inline ref
and update the existing extent item to update the total number of
references for the extent. This implies we need reserved space for two
insertion paths in the extent tree, but we only reserved for one path.
The extent item and the non-inline ref item may be located in different
leaves, or even if they are located in the same leaf, after updating the
extent item and before inserting the non-inline ref item, the extent
buffers in the btree path may have been written (due to memory pressure
for e.g.), in which case we need to COW the entire path again. In this
case since we have not reserved enough space for the delayed refs block
reserve, we will use the global block reserve.
If we are in a situation where the fs has no more unallocated space enough
to allocate a new metadata block group and available space in the existing
metadata block groups is close to the maximum size of the global block
reserve (512M), we may end up consuming too much of the free metadata
space to the point where we can't commit any future transaction because it
will fail, with -ENOSPC, during its commit when trying to allocate an
extent for some COW operation (running delayed refs generated by running
delayed refs or COWing the root tree's root node at commit_cowonly_roots()
for example). Such dramatic scenario can happen if we have many delayed
refs that require the insertion of non-inline ref items, due to too many
reflinks or snapshots. We also have situations where we use the global
block reserve because we could not in advance know that we will need
space to update some trees (block group creation for example), so this
all adds up to increase the chances of exhausting the global block reserve
and making any future transaction commit to fail with -ENOSPC and turn
the fs into RO mode, or fail the mount operation in case the mount needs
to start and commit a transaction, such as when we have orphans to cleanup
for example - such case was reported and hit by someone running a SLE
(SUSE Linux Enterprise) distribution for example - where the fs had no
more unallocated space that could be used to allocate a new metadata block
group, and the available metadata space was about 1.5M, not enough to
commit a transaction to cleanup an orphan inode (or do relocation of data
block groups that were far from being full).
So reserve space for delayed refs by individual refs and not by ref heads,
as we may need to COW multiple extent tree paths due to non-inline ref
items.
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Since all check-integrity entry points have been removed, let's also
remove the config and all related code relying on that.
And since we have removed the mount option for check-integrity, we also
need to re-number all the BTRFS_MOUNT_* enums.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The function btrfsic_mount() is part of the deprecated check-integrity
functionality.
Now let's remove the main entry point of check-integrity, and thankfully
most of the check-integrity code is self-contained inside
check-integrity.c, we can safely remove the function without huge
changes to btrfs code base.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The function btrfsic_mount() is part of the deprecated check-integrity
functionality.
Now let's remove the main entry point of check-integrity, and thankfully
most of the check-integrity code is self-contained inside
check-integrity.c, we can safely remove the function without huge
changes to btrfs code base.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The function btrfsic_check_bio() is part of the deprecated
check-integrity functionality.
Now let's remove the main entry point of check-integrity, and thankfully
most of the check-integrity code is self-contained inside
check-integrity.c, we can safely remove the function without huge
changes to btrfs code base.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Function name in the comment does not bring much value to code not
exposed as API and we don't stick to the kdoc format anymore. Update
formatting of parameter descriptions.
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Among all the callers, only the device_list_add() function uses the
second argument of alloc_fs_devices(). It passes metadata_uuid when
available, otherwise, it passes NULL. And in turn, alloc_fs_devices()
is designed to copy either metadata_uuid or fsid into
fs_devices::metadata_uuid.
So remove the second argument in alloc_fs_devices(), and always copy the
fsid. In the caller device_list_add() function, we will overwrite it
with metadata_uuid when it is available.
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
[BUG]
After commit 72a69cd030 ("btrfs: subpage: pack all subpage bitmaps
into a larger bitmap"), the DEBUG section of btree_dirty_folio() would
no longer compile.
[CAUSE]
If DEBUG is defined, we would do extra checks for btree_dirty_folio(),
mostly to make sure the range we marked dirty has an extent buffer and
that extent buffer is dirty.
For subpage, we need to iterate through all the extent buffers covered
by that page range, and make sure they all matches the criteria.
However commit 72a69cd030 ("btrfs: subpage: pack all subpage bitmaps
into a larger bitmap") changes how we store the bitmap, we pack all the
16 bits bitmaps into a larger bitmap, which would save some space.
This means we no longer have btrfs_subpage::dirty_bitmap, instead the
dirty bitmap is starting at btrfs_subpage_info::dirty_offset, and has a
length of btrfs_subpage_info::bitmap_nr_bits.
[FIX]
Although I'm not sure if it still makes sense to maintain such code, at
least let it compile.
This patch would let us test the bits one by one through the bitmaps.
CC: stable@vger.kernel.org # 6.1+
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Internally I got a report of very long stalls on normal operations like
creating a new file when auto relocation was running. The reporter used
the 'bpf offcputime' tracer to show that we would get stuck in
start_transaction for 5 to 30 seconds, and were always being woken up by
the transaction commit.
Using my timing-everything script, which times how long a function takes
and what percentage of that total time is taken up by its children, I
saw several traces like this
1083 took 32812902424 ns
29929002926 ns 91.2110% wait_for_commit_duration
25568 ns 7.7920e-05% commit_fs_roots_duration
1007751 ns 0.00307% commit_cowonly_roots_duration
446855602 ns 1.36182% btrfs_run_delayed_refs_duration
271980 ns 0.00082% btrfs_run_delayed_items_duration
2008 ns 6.1195e-06% btrfs_apply_pending_changes_duration
9656 ns 2.9427e-05% switch_commit_roots_duration
1598 ns 4.8700e-06% btrfs_commit_device_sizes_duration
4314 ns 1.3147e-05% btrfs_free_log_root_tree_duration
Here I was only tracing functions that happen where we are between
START_COMMIT and UNBLOCKED in order to see what would be keeping us
blocked for so long. The wait_for_commit() we do is where we wait for a
previous transaction that hasn't completed it's commit. This can
include all of the unpin work and other cleanups, which tends to be the
longest part of our transaction commit.
There is no reason we should be blocking new things from entering the
transaction at this point, it just adds to random latency spikes for no
reason.
Fix this by adding a PREP stage. This allows us to properly deal with
multiple committers coming in at the same time, we retain the behavior
that the winner waits on the previous transaction and the losers all
wait for this transaction commit to occur. Nothing else is blocked
during the PREP stage, and then once the wait is complete we switch to
COMMIT_START and all of the same behavior as before is maintained.
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
fs_devices::metadata_uuid value is already updated based on the
super_block::METADATA_UUID flag for either fsid or metadata_uuid as
appropriate. So, fs_devices::metadata_uuid can be used directly.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Tested-by: Guilherme G. Piccoli <gpiccoli@igalia.com>
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The function btrfs_validate_super() should verify the metadata_uuid in
the provided superblock argument. Because, all its callers expect it to
do that.
Such as in the following stacks:
write_all_supers()
sb = fs_info->super_for_commit;
btrfs_validate_write_super(.., sb)
btrfs_validate_super(.., sb, ..)
scrub_one_super()
btrfs_validate_super(.., sb, ..)
And
check_dev_super()
btrfs_validate_super(.., sb, ..)
However, it currently verifies the fs_info::super_copy::metadata_uuid
instead. Fix this using the correct metadata_uuid in the superblock
argument.
CC: stable@vger.kernel.org # 5.4+
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Tested-by: Guilherme G. Piccoli <gpiccoli@igalia.com>
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The function btrfs_validate_super() should verify the fsid in the provided
superblock argument. Because, all its callers expect it to do that.
Such as in the following stack:
write_all_supers()
sb = fs_info->super_for_commit;
btrfs_validate_write_super(.., sb)
btrfs_validate_super(.., sb, ..)
scrub_one_super()
btrfs_validate_super(.., sb, ..)
And
check_dev_super()
btrfs_validate_super(.., sb, ..)
However, it currently verifies the fs_info::super_copy::fsid instead,
which is not correct. Fix this using the correct fsid in the superblock
argument.
CC: stable@vger.kernel.org # 5.4+
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Tested-by: Guilherme G. Piccoli <gpiccoli@igalia.com>
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Use LIST_HEAD() to initialize the list_head instead of open-coding it.
Signed-off-by: Ruan Jinjie <ruanjinjie@huawei.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Ensure a metadata and system block group can be activated on write time, by
leaving a certain number of active zones when trying to activate a data
block group.
Zones for two metadata block groups (normal and tree-log) and one system
block group are reserved, according to the profile type: two zones per
block group on the DUP profile and one zone per block group otherwise.
The reservation must be freed once a non-data block group is allocated. If
not, we over-reserve the active zones and data block group activation will
suffer. For the dynamic reservation count, we need to manage the
reservation count per device.
The reservation count variable is protected by
fs_info->zone_active_bgs_lock.
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>
btrfs_cleanup_fs_roots() is not used outside disk-io.c, so make it static,
remove its prototype from disk-io.h and move its definition above the
where it's used in disk-io.c
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently when we turn the fs into an error state, typically after a
transaction abort, we don't store the error anywhere, we just set a bit
(BTRFS_FS_STATE_ERROR) at struct btrfs_fs_info::fs_state to signal the
error state.
There are cases where it would be useful to have access to the specific
error in order to provide a more meaningful error to users/applications.
This change adds a member to struct btrfs_fs_info to store the error and
removes the BTRFS_FS_STATE_ERROR bit. When there's no error, the new
member (fs_error) has a value of 0, otherwise its value is a negative
errno value.
Followup changes will make use of this new member.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently find_first_extent_bit() returns a 0 if it found a range in the
given io tree and 1 if it didn't find any. There's no need to return any
errors, so make the return value a boolean and invert the logic to make
more sense: return true if it found a range and false if it didn't find
any range.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently btrfs_destroy_pinned_extent() is always returning 0 no matter
what and its caller ignores its return value (as well everything up in
the call chain). This is because this is called in the transaction abort
path, where we can't even deal with any errors since we are in a critical
situation already and cleanup of resources is done in a best effort
fashion.
So make btrfs_destroy_pinned_extent() return void.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently btrfs_destroy_marked_extents() is returning the value of the
last call to find_first_extent_bit(), which returns a value of 1 meaning
no more ranges found the dirty pages io tree. This value is useless to the
single caller of btrfs_destroy_marked_extents(), which ignores any return
value from btrfs_destroy_marked_extents(). This is because it's only used
in the transaction abort path, where we can't even deal with any errors
since we are in a critical situation already and cleanup of resources is
done in a best effort fashion.
So make btrfs_destroy_marked_extents() return void.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
[BUG]
Syzbot reported a crash that an ASSERT() got triggered inside
prepare_to_merge().
That ASSERT() makes sure the reloc tree is properly pointed back by its
subvolume tree.
[CAUSE]
After more debugging output, it turns out we had an invalid reloc tree:
BTRFS error (device loop1): reloc tree mismatch, root 8 has no reloc root, expect reloc root key (-8, 132, 8) gen 17
Note the above root key is (TREE_RELOC_OBJECTID, ROOT_ITEM,
QUOTA_TREE_OBJECTID), meaning it's a reloc tree for quota tree.
But reloc trees can only exist for subvolumes, as for non-subvolume
trees, we just COW the involved tree block, no need to create a reloc
tree since those tree blocks won't be shared with other trees.
Only subvolumes tree can share tree blocks with other trees (thus they
have BTRFS_ROOT_SHAREABLE flag).
Thus this new debug output proves my previous assumption that corrupted
on-disk data can trigger that ASSERT().
[FIX]
Besides the dedicated fix and the graceful exit, also let tree-checker to
check such root keys, to make sure reloc trees can only exist for subvolumes.
CC: stable@vger.kernel.org # 5.15+
Reported-by: syzbot+ae97a827ae1c3336bbb4@syzkaller.appspotmail.com
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
[BUG]
Syzbot reported a weird ASSERT() triggered inside prepare_to_merge().
assertion failed: root->reloc_root == reloc_root, in fs/btrfs/relocation.c:1919
------------[ cut here ]------------
kernel BUG at fs/btrfs/relocation.c:1919!
invalid opcode: 0000 [#1] PREEMPT SMP KASAN
CPU: 0 PID: 9904 Comm: syz-executor.3 Not tainted
6.4.0-syzkaller-08881-g533925cb7604 #0
Hardware name: Google Google Compute Engine/Google Compute Engine,
BIOS Google 05/27/2023
RIP: 0010:prepare_to_merge+0xbb2/0xc40 fs/btrfs/relocation.c:1919
Code: fe e9 f5 (...)
RSP: 0018:ffffc9000325f760 EFLAGS: 00010246
RAX: 000000000000004f RBX: ffff888075644030 RCX: 1481ccc522da5800
RDX: ffffc90005c09000 RSI: 00000000000364ca RDI: 00000000000364cb
RBP: ffffc9000325f870 R08: ffffffff816f33ac R09: 1ffff9200064bea0
R10: dffffc0000000000 R11: fffff5200064bea1 R12: ffff888075644000
R13: ffff88803b166000 R14: ffff88803b166560 R15: ffff88803b166558
FS: 00007f4e305fd700(0000) GS:ffff8880b9800000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 000056080679c000 CR3: 00000000193ad000 CR4: 00000000003506f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
<TASK>
relocate_block_group+0xa5d/0xcd0 fs/btrfs/relocation.c:3749
btrfs_relocate_block_group+0x7ab/0xd70 fs/btrfs/relocation.c:4087
btrfs_relocate_chunk+0x12c/0x3b0 fs/btrfs/volumes.c:3283
__btrfs_balance+0x1b06/0x2690 fs/btrfs/volumes.c:4018
btrfs_balance+0xbdb/0x1120 fs/btrfs/volumes.c:4402
btrfs_ioctl_balance+0x496/0x7c0 fs/btrfs/ioctl.c:3604
vfs_ioctl fs/ioctl.c:51 [inline]
__do_sys_ioctl fs/ioctl.c:870 [inline]
__se_sys_ioctl+0xf8/0x170 fs/ioctl.c:856
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x41/0xc0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x63/0xcd
RIP: 0033:0x7f4e2f88c389
[CAUSE]
With extra debugging, the offending reloc_root is for quota tree (rootid 8).
Normally we should not use the reloc tree for quota root at all, as reloc
trees are only for subvolume trees.
But there is a race between quota enabling and relocation, this happens
after commit 85724171b3 ("btrfs: fix the btrfs_get_global_root return value").
Before that commit, for quota and free space tree, we exit immediately
if we cannot grab it from fs_info.
But now we would try to read it from disk, just as if they are fs trees,
this sets ROOT_SHAREABLE flags in such race:
Thread A | Thread B
---------------------------------+------------------------------
btrfs_quota_enable() |
| | btrfs_get_root_ref()
| | |- btrfs_get_global_root()
| | | Returned NULL
| | |- btrfs_lookup_fs_root()
| | | Returned NULL
|- btrfs_create_tree() | |
| Now quota root item is | |
| inserted | |- btrfs_read_tree_root()
| | | Got the newly inserted quota root
| | |- btrfs_init_fs_root()
| | | Set ROOT_SHAREABLE flag
[FIX]
Get back to the old behavior by returning PTR_ERR(-ENOENT) if the target
objectid is not a subvolume tree or data reloc tree.
Reported-and-tested-by: syzbot+ae97a827ae1c3336bbb4@syzkaller.appspotmail.com
Fixes: 85724171b3 ("btrfs: fix the btrfs_get_global_root return value")
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The zoned mode need to reset a zone before using it. We rely on btrfs's
original discard functionality (discarding unused block group range) to do
the resetting.
While the commit 63a7cb1307 ("btrfs: auto enable discard=async when
possible") made the discard done in an async manner, a zoned reset do not
need to be async, as it is fast enough.
Even worth, delaying zone rests prevents using those zones again. So, let's
disable async discard on the zoned mode.
Fixes: 63a7cb1307 ("btrfs: auto enable discard=async when possible")
CC: stable@vger.kernel.org # 6.3+
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ update message text ]
Signed-off-by: David Sterba <dsterba@suse.com>
The implementation of XXHASH is now CPU only but still fast enough to be
considered for the synchronous checksumming, like non-generic crc32c.
A userspace benchmark comparing it to various implementations (patched
hash-speedtest from btrfs-progs):
Block size: 4096
Iterations: 1000000
Implementation: builtin
Units: CPU cycles
NULL-NOP: cycles: 73384294, cycles/i 73
NULL-MEMCPY: cycles: 228033868, cycles/i 228, 61664.320 MiB/s
CRC32C-ref: cycles: 24758559416, cycles/i 24758, 567.950 MiB/s
CRC32C-NI: cycles: 1194350470, cycles/i 1194, 11773.433 MiB/s
CRC32C-ADLERSW: cycles: 6150186216, cycles/i 6150, 2286.372 MiB/s
CRC32C-ADLERHW: cycles: 626979180, cycles/i 626, 22427.453 MiB/s
CRC32C-PCL: cycles: 466746732, cycles/i 466, 30126.699 MiB/s
XXHASH: cycles: 860656400, cycles/i 860, 16338.188 MiB/s
Comparing purely software implementation (ref), current outdated
accelerated using crc32q instruction (NI), optimized implementations by
M. Adler (https://stackoverflow.com/questions/17645167/implementing-sse-4-2s-crc32c-in-software/17646775#17646775)
and the best one that was taken from kernel using the PCLMULQDQ
instruction (PCL).
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: David Sterba <dsterba@suse.com>
btrfs_destroy_delayed_refs() always returns 0 and its single caller does
not check its return value, as it also returns void, and so does the
callers' caller and so on. This is because we are in the transaction abort
path, where we have no way to deal with errors (we are in a critical
situation) and all cleanup of resources works in a best effort fashion.
So make btrfs_destroy_delayed_refs() return void.
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We have a few static functions at disk-io.c for which we have a forward
declaration of their prototype, but it's not needed because all those
functions are defined before they are called, so remove them.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The 'in_tree' field is really not needed in struct btrfs_delayed_ref_node,
as we can check whether a reference is in the tree or not simply by
checking its red black tree node member with RB_EMPTY_NODE(), as when we
remove it from the tree we always call RB_CLEAR_NODE(). So remove that
field and use RB_EMPTY_NODE().
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
BACKGROUND
==========
When multiple work items are queued to a workqueue, their execution order
doesn't match the queueing order. They may get executed in any order and
simultaneously. When fully serialized execution - one by one in the queueing
order - is needed, an ordered workqueue should be used which can be created
with alloc_ordered_workqueue().
However, alloc_ordered_workqueue() was a later addition. Before it, an
ordered workqueue could be obtained by creating an UNBOUND workqueue with
@max_active==1. This originally was an implementation side-effect which was
broken by 4c16bd327c ("workqueue: restore WQ_UNBOUND/max_active==1 to be
ordered"). Because there were users that depended on the ordered execution,
5c0338c687 ("workqueue: restore WQ_UNBOUND/max_active==1 to be ordered")
made workqueue allocation path to implicitly promote UNBOUND workqueues w/
@max_active==1 to ordered workqueues.
While this has worked okay, overloading the UNBOUND allocation interface
this way creates other issues. It's difficult to tell whether a given
workqueue actually needs to be ordered and users that legitimately want a
min concurrency level wq unexpectedly gets an ordered one instead. With
planned UNBOUND workqueue updates to improve execution locality and more
prevalence of chiplet designs which can benefit from such improvements, this
isn't a state we wanna be in forever.
This patch series audits all call sites that create an UNBOUND workqueue w/
@max_active==1 and converts them to alloc_ordered_workqueue() as necessary.
BTRFS
=====
* fs_info->scrub_workers initialized in scrub_workers_get() was setting
@max_active to 1 when @is_dev_replace is set and it seems that the
workqueue actually needs to be ordered if @is_dev_replace. Update the code
so that alloc_ordered_workqueue() is used if @is_dev_replace.
* fs_info->discard_ctl.discard_workers initialized in
btrfs_init_workqueues() was directly using alloc_workqueue() w/
@max_active==1. Converted to alloc_ordered_workqueue().
* fs_info->fixup_workers and fs_info->qgroup_rescan_workers initialized in
btrfs_queue_work() use the btrfs's workqueue wrapper, btrfs_workqueue,
which are allocated with btrfs_alloc_workqueue().
btrfs_workqueue implements automatic @max_active adjustment which is
disabled when the specified max limit is below a certain threshold, so
calling btrfs_alloc_workqueue() with @limit_active==1 yields an ordered
workqueue whose @max_active won't be changed as the auto-tuning is
disabled.
This is rather brittle in that nothing clearly indicates that the two
workqueues should be ordered or btrfs_alloc_workqueue() must disable
auto-tuning when @limit_active==1.
This patch factors out the common btrfs_workqueue init code into
btrfs_init_workqueue() and add explicit btrfs_alloc_ordered_workqueue().
The two workqueues are converted to use the new ordered allocation
interface.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: David Sterba <dsterba@suse.com>
btrfs_grab_root already checks for a NULL root itself.
Reviewed-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>
Use a switch statement instead of an endless chain of if statements
to make the code a little cleaner.
Reviewed-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_grab_root returns either the root or NULL, and the callers of
btrfs_get_global_root expect it to return the same. But all the more
recently added roots instead return an ERR_PTR, so fix this.
Reviewed-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>
There are three ways the fsid is validated in btrfs_validate_super():
- verify that super_copy::fsid is the same as fs_devices::fsid
- if the metadata_uuid flag is set, verify if super_copy::metadata_uuid
and fs_devices::metadata_uuid are the same.
- a few lines below, often missed out, verify if dev_item::fsid is the
same as fs_devices::metadata_uuid.
The function btrfs_validate_super() contains multiple if-statements with
memcmp() to check UUIDs. This patch consolidates them into a single
location.
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Simplify the return type of check_tree_block_fsid() from int (1 or 0) to
bool. Its only user is interested in knowing the success or failure.
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
