Call rcu_read_lock() before exiting the loop in
try_release_subpage_extent_buffer() because there is a rcu_read_unlock()
call past the loop.
This has been detected by the Clang thread-safety analyzer.
Fixes: ad580dfa38 ("btrfs: fix subpage deadlock in try_release_subpage_extent_buffer()")
CC: stable@vger.kernel.org # 6.18+
Reviewed-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We have recently observed a number of subvolumes with broken dentries.
ls-ing the parent dir looks like:
drwxrwxrwt 1 root root 16 Jan 23 16:49 .
drwxr-xr-x 1 root root 24 Jan 23 16:48 ..
d????????? ? ? ? ? ? broken_subvol
and similarly stat-ing the file fails.
In this state, deleting the subvol fails with ENOENT, but attempting to
create a new file or subvol over it errors out with EEXIST and even
aborts the fs. Which leaves us a bit stuck.
dmesg contains a single notable error message reading:
"could not do orphan cleanup -2"
2 is ENOENT and the error comes from the failure handling path of
btrfs_orphan_cleanup(), with the stack leading back up to
btrfs_lookup().
btrfs_lookup
btrfs_lookup_dentry
btrfs_orphan_cleanup // prints that message and returns -ENOENT
After some detailed inspection of the internal state, it became clear
that:
- there are no orphan items for the subvol
- the subvol is otherwise healthy looking, it is not half-deleted or
anything, there is no drop progress, etc.
- the subvol was created a while ago and does the meaningful first
btrfs_orphan_cleanup() call that sets BTRFS_ROOT_ORPHAN_CLEANUP much
later.
- after btrfs_orphan_cleanup() fails, btrfs_lookup_dentry() returns -ENOENT,
which results in a negative dentry for the subvolume via
d_splice_alias(NULL, dentry), leading to the observed behavior. The
bug can be mitigated by dropping the dentry cache, at which point we
can successfully delete the subvolume if we want.
i.e.,
btrfs_lookup()
btrfs_lookup_dentry()
if (!sb_rdonly(inode->vfs_inode)->vfs_inode)
btrfs_orphan_cleanup(sub_root)
test_and_set_bit(BTRFS_ROOT_ORPHAN_CLEANUP)
btrfs_search_slot() // finds orphan item for inode N
...
prints "could not do orphan cleanup -2"
if (inode == ERR_PTR(-ENOENT))
inode = NULL;
return d_splice_alias(NULL, dentry) // NEGATIVE DENTRY for valid subvolume
btrfs_orphan_cleanup() does test_and_set_bit(BTRFS_ROOT_ORPHAN_CLEANUP)
on the root when it runs, so it cannot run more than once on a given
root, so something else must run concurrently. However, the obvious
routes to deleting an orphan when nlinks goes to 0 should not be able to
run without first doing a lookup into the subvolume, which should run
btrfs_orphan_cleanup() and set the bit.
The final important observation is that create_subvol() calls
d_instantiate_new() but does not set BTRFS_ROOT_ORPHAN_CLEANUP, so if
the dentry cache gets dropped, the next lookup into the subvolume will
make a real call into btrfs_orphan_cleanup() for the first time. This
opens up the possibility of concurrently deleting the inode/orphan items
but most typical evict() paths will be holding a reference on the parent
dentry (child dentry holds parent->d_lockref.count via dget in
d_alloc(), released in __dentry_kill()) and prevent the parent from
being removed from the dentry cache.
The one exception is delayed iputs. Ordered extent creation calls
igrab() on the inode. If the file is unlinked and closed while those
refs are held, iput() in __dentry_kill() decrements i_count but does
not trigger eviction (i_count > 0). The child dentry is freed and the
subvol dentry's d_lockref.count drops to 0, making it evictable while
the inode is still alive.
Since there are two races (the race between writeback and unlink and
the race between lookup and delayed iputs), and there are too many moving
parts, the following three diagrams show the complete picture.
(Only the second and third are races)
Phase 1:
Create Subvol in dentry cache without BTRFS_ROOT_ORPHAN_CLEANUP set
btrfs_mksubvol()
lookup_one_len()
__lookup_slow()
d_alloc_parallel()
__d_alloc() // d_lockref.count = 1
create_subvol(dentry)
// doesn't touch the bit..
d_instantiate_new(dentry, inode) // dentry in cache with d_lockref.count == 1
Phase 2:
Create a delayed iput for a file in the subvol but leave the subvol in
state where its dentry can be evicted (d_lockref.count == 0)
T1 (task) T2 (writeback) T3 (OE workqueue)
write() // dirty pages
btrfs_writepages()
btrfs_run_delalloc_range()
cow_file_range()
btrfs_alloc_ordered_extent()
igrab() // i_count: 1 -> 2
btrfs_unlink_inode()
btrfs_orphan_add()
close()
__fput()
dput()
finish_dput()
__dentry_kill()
dentry_unlink_inode()
iput() // 2 -> 1
--parent->d_lockref.count // 1 -> 0; evictable
finish_ordered_fn()
btrfs_finish_ordered_io()
btrfs_put_ordered_extent()
btrfs_add_delayed_iput()
Phase 3:
Once the delayed iput is pending and the subvol dentry is evictable,
the shrinker can free it, causing the next lookup to go through
btrfs_lookup() and call btrfs_orphan_cleanup() for the first time.
If the cleaner kthread processes the delayed iput concurrently, the
two race:
T1 (shrinker) T2 (cleaner kthread) T3 (lookup)
super_cache_scan()
prune_dcache_sb()
__dentry_kill()
// subvol dentry freed
btrfs_run_delayed_iputs()
iput() // i_count -> 0
evict() // sets I_FREEING
btrfs_evict_inode()
// truncation loop
btrfs_lookup()
btrfs_lookup_dentry()
btrfs_orphan_cleanup()
// first call (bit never set)
btrfs_iget()
// blocks on I_FREEING
btrfs_orphan_del()
// inode freed
// returns -ENOENT
btrfs_del_orphan_item()
// -ENOENT
// "could not do orphan cleanup -2"
d_splice_alias(NULL, dentry)
// negative dentry for valid subvol
The most straightforward fix is to ensure the invariant that a dentry
for a subvolume can exist if and only if that subvolume has
BTRFS_ROOT_ORPHAN_CLEANUP set on its root (and is known to have no
orphans or ran btrfs_orphan_cleanup()).
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Boris Burkov <boris@bur.io>
Signed-off-by: David Sterba <dsterba@suse.com>
btrfs_zoned_reserve_data_reloc_bg() is called on each mount of a file
system and allocates a new block-group, to assign it to be the dedicated
relocation target, if no pre-existing usable block-group for this task is
found.
If for some reason the transaction is aborted, btrfs_end_transaction()
will wake up the transaction kthread. But the transaction kthread is not
yet initialized at the time btrfs_zoned_reserve_data_reloc_bg() is
called, leading to the following NULL-pointer dereference:
RSP: 0018:ffffc9000c617c98 EFLAGS: 00010046
RAX: 0000000000000000 RBX: 000000000000073c RCX: 0000000000000002
RDX: 0000000000000001 RSI: 0000000000000003 RDI: 0000000000000001
RBP: 0000000000000207 R08: ffffffff8223c71d R09: 0000000000000635
R10: ffff888108588000 R11: 0000000000000003 R12: 0000000000000003
R13: 000000000000073c R14: 0000000000000000 R15: ffff888114dd6000
FS: 00007f2993745840(0000) GS:ffff8882b508d000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 000000000000073c CR3: 0000000121a82006 CR4: 0000000000770eb0
PKRU: 55555554
Call Trace:
<TASK>
try_to_wake_up (./include/linux/spinlock.h:557 kernel/sched/core.c:4106)
__btrfs_end_transaction (fs/btrfs/transaction.c:1115 (discriminator 2))
btrfs_zoned_reserve_data_reloc_bg (fs/btrfs/zoned.c:2840)
open_ctree (fs/btrfs/disk-io.c:3588)
btrfs_get_tree.cold (fs/btrfs/super.c:982 fs/btrfs/super.c:1944 fs/btrfs/super.c:2087 fs/btrfs/super.c:2121)
vfs_get_tree (fs/super.c:1752)
__do_sys_fsconfig (fs/fsopen.c:231 fs/fsopen.c:295 fs/fsopen.c:473)
do_syscall_64 (arch/x86/entry/syscall_64.c:63 (discriminator 1) arch/x86/entry/syscall_64.c:94 (discriminator 1))
entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:131)
RIP: 0033:0x7f299392740e
Move the call to btrfs_zoned_reserve_data_reloc_bg() after the
transaction_kthread has been initialized to fix this problem.
Fixes: 694ce5e143 ("btrfs: zoned: reserve data_reloc block group on mount")
Reviewed-by: Filipe Manana <fdmanana@suse.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>
btrfs_set_periodic_reclaim_ready() requires space_info->lock to be held,
as enforced by lockdep_assert_held(). However, btrfs_reclaim_sweep() was
calling it after do_reclaim_sweep() returns, at which point
space_info->lock is no longer held.
Fix this by explicitly acquiring space_info->lock before clearing the
periodic reclaim ready flag in btrfs_reclaim_sweep().
Reported-by: Chris Mason <clm@meta.com>
Link: https://lore.kernel.org/linux-btrfs/20260208182556.891815-1-clm@meta.com/
Fixes: 19eff93dc7 ("btrfs: fix periodic reclaim condition")
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: Sun YangKai <sunk67188@gmail.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Add the definitions for the remap tree to print-tree.c, so that we get
more useful information if a tree is dumped to dmesg.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Mark Harmstone <mark@harmstone.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Fix a chunk map leak in btrfs_map_block(): if we return early with -EINVAL,
we're not freeing the chunk map that we've just looked up.
Fixes: 0ae653fbec ("btrfs: reduce chunk_map lookups in btrfs_map_block()")
CC: stable@vger.kernel.org # 6.12+
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Mark Harmstone <mark@harmstone.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Commit d7f67ac9a9 ("btrfs: relax block-group-tree feature dependency
checks") introduced a regression when it comes to handling unsupported
incompat or compat_ro flags. Beforehand we only printed the flags that
we didn't recognize, afterwards we printed them all, which is less
useful. Fix the error handling so it behaves like it used to.
Fixes: d7f67ac9a9 ("btrfs: relax block-group-tree feature dependency checks")
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Mark Harmstone <mark@harmstone.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Fix the error message in btrfs_delete_subvolume() if we can't delete a
subvolume because it has an active swapfile: we were printing the number
of the parent rather than the target.
Fixes: 60021bd754 ("btrfs: prevent subvol with swapfile from being deleted")
Reviewed-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Mark Harmstone <mark@harmstone.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Commit b471965fdb ("btrfs: fix replace/scrub failure with
metadata_uuid") fixed the comparison in scrub_verify_one_metadata() to
use metadata_uuid rather than fsid, but left the warning as it was. Fix
it so it matches what we're doing.
Fixes: b471965fdb ("btrfs: fix replace/scrub failure with metadata_uuid")
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Mark Harmstone <mark@harmstone.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Fix a copy-paste error in check_extent_data_ref(): we're printing root
as in the message above, we should be printing objectid.
Fixes: f333a3c7e8 ("btrfs: tree-checker: validate dref root and objectid")
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Mark Harmstone <mark@harmstone.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Fix the error message in check_dev_extent_item(), when an overlapping
stripe is encountered. For dev extents, objectid is the disk number and
offset the physical address, so prev_key->objectid should actually be
prev_key->offset.
(I can't take any credit for this one - this was discovered by Chris and
his friend Claude.)
Reported-by: Chris Mason <clm@fb.com>
Fixes: 008e2512dc ("btrfs: tree-checker: add dev extent item checks")
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Mark Harmstone <mark@harmstone.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Fix the error message in btrfs_delete_delayed_dir_index() if
__btrfs_add_delayed_item() fails: the message says root, inode, index,
error, but we're actually passing index, root, inode, error.
Fixes: adc1ef55dc ("btrfs: add details to error messages at btrfs_delete_delayed_dir_index()")
Signed-off-by: Mark Harmstone <mark@harmstone.com>
Reviewed-by: Filipe Manana <fdmanana@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>
When unmounting a filesystem we will try, among many other things, to
commit the super block. On a filesystem that was shutdown, though, this
will always fail with -EROFS as writes are forbidden on this context;
and an error will be reported.
Don't commit the super block on this situation, which should be fine as
the filesystem is frozen before shutdown and, therefore, it should be at
a consistent state.
Signed-off-by: Miquel Sabaté Solà <mssola@mssola.com>
Reviewed-by: Filipe Manana <fdmanana@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>
In this function the 'pages' object is never freed in the hopes that it is
picked up by btrfs_uring_read_finished() whenever that executes in the
future. But that's just the happy path. Along the way previous
allocations might have gone wrong, or we might not get -EIOCBQUEUED from
btrfs_encoded_read_regular_fill_pages(). In all these cases, we go to a
cleanup section that frees all memory allocated by this function without
assuming any deferred execution, and this also needs to happen for the
'pages' allocation.
Fixes: 34310c442e ("btrfs: add io_uring command for encoded reads (ENCODED_READ ioctl)")
Signed-off-by: Miquel Sabaté Solà <mssola@mssola.com>
Reviewed-by: Filipe Manana <fdmanana@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>
This WARN_ON(ret) is never executed since the previous if statement makes
us jump into the 'out_put' label when ret is not zero. The existing
transaction abort inside the if statement also gives us a stack trace,
so we don't need to move the WARN_ON(ret) into the if statement either.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We are logging messages as warnings but they should really have an error
level instead, as if the respective conditions are met the mount will
fail. So convert them to error level and also log the error code returned
by read_tree_block().
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
There is no need to call btrfs_handle_fs_error() (which we are trying to
deprecate) if we fail to recover log trees:
1) Such a failure results in failing the mount immediately;
2) If the recovery started a transaction before failing, it has already
aborted the transaction down in the call chain.
So remove the btrfs_handle_fs_error() call, replace it with an error
message and assert that the FS is in error state (so that no partial
updates are committed due to a transaction that was not aborted).
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
If we fail to start the UUID rescan kthread, btrfs_check_uuid_tree() logs
an error message and returns the error to the single caller, open_ctree().
This however is redundant since the caller already logs an error message,
which is also more informative since it logs the error code. Some remove
the warning message from btrfs_check_uuid_tree() as it doesn't add any
value.
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>
Failure to read the fs root results in a mount error, but we log a warning
message. Same goes for checking the UUID tree, an error results in a mount
failure but we log a warning message. Change the level of the logged
messages from warning to error.
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>
[BUG]
Commit c28214bde6 ("btrfs: refactor the main loop of
cow_file_range()") refactored the handling of COWing one range.
However it changed the error handling of the reserved extent.
The old cleanup looks like this:
out_drop_extent_cache:
btrfs_drop_extent_map_range(inode, start, start + cur_alloc_size - 1, false);
out_reserve:
btrfs_dec_block_group_reservations(fs_info, ins.objectid);
btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, true);
[...]
clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |
EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV;
page_ops = PAGE_UNLOCK | PAGE_START_WRITEBACK | PAGE_END_WRITEBACK;
/*
* For the range (2). If we reserved an extent for our delalloc range
* (or a subrange) and failed to create the respective ordered extent,
* then it means that when we reserved the extent we decremented the
* extent's size from the data space_info's bytes_may_use counter and
* incremented the space_info's bytes_reserved counter by the same
* amount. We must make sure extent_clear_unlock_delalloc() does not try
* to decrement again the data space_info's bytes_may_use counter,
* therefore we do not pass it the flag EXTENT_CLEAR_DATA_RESV.
*/
if (cur_alloc_size) {
extent_clear_unlock_delalloc(inode, start,
start + cur_alloc_size - 1,
locked_folio, &cached, clear_bits,
page_ops);
btrfs_qgroup_free_data(inode, NULL, start, cur_alloc_size, NULL);
}
Which only calls EXTENT_CLEAR_META_RESV.
As the reserved extent is properly handled by
btrfs_free_reserved_extent().
However the new cleanup is:
extent_clear_unlock_delalloc(inode, file_offset, cur_end, locked_folio, cached,
EXTENT_LOCKED | EXTENT_DELALLOC |
EXTENT_DELALLOC_NEW |
EXTENT_DEFRAG | EXTENT_DO_ACCOUNTING,
PAGE_UNLOCK | PAGE_START_WRITEBACK |
PAGE_END_WRITEBACK);
btrfs_qgroup_free_data(inode, NULL, file_offset, cur_len, NULL);
btrfs_dec_block_group_reservations(fs_info, ins->objectid);
btrfs_free_reserved_extent(fs_info, ins->objectid, ins->offset, true);
The flag EXTENT_DO_ACCOUNTING implies both EXTENT_CLEAR_META_RESV and
EXTENT_CLEAR_DATA_RESV, which will release the bytes_may_use, which
later btrfs_free_reserved_extent() will do again, causing incorrect
double release (and may underflow bytes_may_use).
[FIX]
Use EXTENT_CLEAR_META_RESV to replace EXTENT_DO_ACCOUNTING, and add back
the comments on why we only use EXTENT_CLEAR_META_RESV.
Fixes: c28214bde6 ("btrfs: refactor the main loop of cow_file_range()")
Reported-by: Chris Mason <clm@meta.com>
Link: https://lore.kernel.org/linux-btrfs/20260208184920.1102719-1-clm@meta.com/
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Coverity (ID: 1226842) reported that the return value of
btrfs_discard_extent() is assigned to ret but is immediately
overwritten by unpin_extent_range() without being checked.
Use the same error handling that is done later in the same function.
Signed-off-by: Jingkai Tan <contact@jingk.ai>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
If btrfs_search_slot_for_read() returns 1, it means we did not find any
key greater than or equals to the key we asked for, meaning we have
reached the end of the tree and therefore the path is not valid. If
this happens we need to break out of the loop and stop, instead of
continuing and accessing an invalid path.
Fixes: 5223cc60b4 ("btrfs: drop the path before adding qgroup items when enabling qgroups")
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
If the root nodes for the chunk root, tree root or log root are not sector
size aligned, we are logging a warning message but these are in fact
errors that makes the super block validation fail. So change the level of
the messages from warning to error.
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>
When initializing the delayed refs block reserve for a transaction handle
we are passing a type of BTRFS_BLOCK_RSV_DELOPS, which is meant for
delayed items and not for delayed refs. The correct type for delayed refs
is BTRFS_BLOCK_RSV_DELREFS.
On release of any excess space reserved in a local delayed refs reserve,
we also should transfer that excess space to the global block reserve
(it it's full, we return to the space info for general availability).
By initializing a transaction's local delayed refs block reserve with a
type of BTRFS_BLOCK_RSV_DELOPS, we were also causing any excess space
released from the delayed block reserve (fs_info->delayed_block_rsv, used
for delayed inodes and items) to be transferred to the global block
reserve instead of the global delayed refs block reserve. This was an
unintentional change in commit 28270e25c6 ("btrfs: always reserve space
for delayed refs when starting transaction"), but it's not particularly
serious as things tend to cancel out each other most of the time and it's
relatively rare to be anywhere near exhaustion of the global reserve.
Fix this by initializing a transaction's local delayed refs reserve with
a type of BTRFS_BLOCK_RSV_DELREFS and making btrfs_block_rsv_release()
attempt to transfer unused space from such a reserve into the global block
reserve, just as we did before that commit for when the block reserve is
a delayed refs rsv.
Reported-by: Alex Lyakas <alex.lyakas@zadara.com>
Link: https://lore.kernel.org/linux-btrfs/CAOcd+r0FHG5LWzTSu=LknwSoqxfw+C00gFAW7fuX71+Z5AfEew@mail.gmail.com/
Fixes: 28270e25c6 ("btrfs: always reserve space for delayed refs when starting transaction")
Reviewed-by: Alex Lyakas <alex.lyakas@zadara.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
[BUG]
There is a bug report that when btrfs hits ENOSPC error in a critical
path, btrfs flips RO (this part is expected, although the ENOSPC bug
still needs to be addressed).
The problem is after the RO flip, if there is a read repair pending, we
can hit the ASSERT() inside btrfs_repair_io_failure() like the following:
BTRFS info (device vdc): relocating block group 30408704 flags metadata|raid1
------------[ cut here ]------------
BTRFS: Transaction aborted (error -28)
WARNING: fs/btrfs/extent-tree.c:3235 at __btrfs_free_extent.isra.0+0x453/0xfd0, CPU#1: btrfs/383844
Modules linked in: kvm_intel kvm irqbypass
[...]
---[ end trace 0000000000000000 ]---
BTRFS info (device vdc state EA): 2 enospc errors during balance
BTRFS info (device vdc state EA): balance: ended with status: -30
BTRFS error (device vdc state EA): parent transid verify failed on logical 30556160 mirror 2 wanted 8 found 6
BTRFS error (device vdc state EA): bdev /dev/nvme0n1 errs: wr 0, rd 0, flush 0, corrupt 10, gen 0
[...]
assertion failed: !(fs_info->sb->s_flags & SB_RDONLY) :: 0, in fs/btrfs/bio.c:938
------------[ cut here ]------------
assertion failed: !(fs_info->sb->s_flags & SB_RDONLY) :: 0, in fs/btrfs/bio.c:938
kernel BUG at fs/btrfs/bio.c:938!
Oops: invalid opcode: 0000 [#1] SMP NOPTI
CPU: 0 UID: 0 PID: 868 Comm: kworker/u8:13 Tainted: G W N 6.19.0-rc6+ #4788 PREEMPT(full)
Tainted: [W]=WARN, [N]=TEST
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.17.0-0-gb52ca86e094d-prebuilt.qemu.org 04/01/2014
Workqueue: btrfs-endio simple_end_io_work
RIP: 0010:btrfs_repair_io_failure.cold+0xb2/0x120
RSP: 0000:ffffc90001d2bcf0 EFLAGS: 00010246
RAX: 0000000000000051 RBX: 0000000000001000 RCX: 0000000000000000
RDX: 0000000000000000 RSI: ffffffff8305cf42 RDI: 00000000ffffffff
RBP: 0000000000000002 R08: 00000000fffeffff R09: ffffffff837fa988
R10: ffffffff8327a9e0 R11: 6f69747265737361 R12: ffff88813018d310
R13: ffff888168b8a000 R14: ffffc90001d2bd90 R15: ffff88810a169000
FS: 0000000000000000(0000) GS:ffff8885e752c000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
------------[ cut here ]------------
[CAUSE]
The cause of -ENOSPC error during the test case btrfs/124 is still
unknown, although it's known that we still have cases where metadata can
be over-committed but can not be fulfilled correctly, thus if we hit
such ENOSPC error inside a critical path, we have no choice but abort
the current transaction.
This will mark the fs read-only.
The problem is inside the btrfs_repair_io_failure() path that we require
the fs not to be mount read-only. This is normally fine, but if we are
doing a read-repair meanwhile the fs flips RO due to a critical error,
we can enter btrfs_repair_io_failure() with super block set to
read-only, thus triggering the above crash.
[FIX]
Just replace the ASSERT() with a proper return if the fs is already
read-only.
Reported-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/linux-btrfs/20260126045555.GB31641@lst.de/
Tested-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Block group size classes are managed consistently everywhere.
Currently, btrfs_use_block_group_size_class() sets a block group's size
class to specialize it for a specific allocation size. However, this
size class remains "stale" even if the block group becomes completely
empty (both used and reserved bytes reach zero).
This happens in two scenarios:
1. When space reservations are freed (e.g., due to errors or transaction
aborts) via btrfs_free_reserved_bytes().
2. When the last extent in a block group is freed via
btrfs_update_block_group().
While size classes are advisory, a stale size class can cause
find_free_extent to unnecessarily skip candidate block groups during
initial search loops. This undermines the purpose of size classes to
reduce fragmentation by keeping block groups restricted to a specific
size class when they could be reused for any size.
Fix this by resetting the size class to BTRFS_BG_SZ_NONE whenever a
block group's used and reserved counts both reach zero. This ensures
that empty block groups are fully available for any allocation size in
the next cycle.
Fixes: 52bb7a2166 ("btrfs: introduce size class to block group allocator")
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: Jiasheng Jiang <jiashengjiangcool@gmail.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We search with offset (u64)-1 which should never match exactly.
Previously this was handled with BUG(). Now logs an error
and return -EUCLEAN.
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Adarsh Das <adarshdas950@gmail.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We search with offset (u64)-1 which should never match exactly.
Previously the code silently returned success without setting the index
count. Now logs an error and return -EUCLEAN instead.
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Adarsh Das <adarshdas950@gmail.com>
Reviewed-by: David Sterba <dsterba@suse.com>,
Signed-off-by: David Sterba <dsterba@suse.com>
Now there is no one utilizing that member, we can safely remove it along
with compressed_bio::nr_folios member. The size is reduced from 352 to
336 bytes on x86_64.
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently only encoded writes utilized btrfs_submit_compressed_write(),
which utilized compressed_bio::compressed_folios[] array.
Change the only call site to call the new helper,
btrfs_alloc_compressed_write(), to allocate a compressed bio, then queue
needed folios into that bio, and finally call
btrfs_submit_compressed_write() to submit the compressed bio.
This change has one hidden benefit, previously we used
btrfs_alloc_folio_array() for the folios of
btrfs_submit_compressed_read(), which doesn't utilize the compression
page pool for bs == ps cases.
Now we call btrfs_alloc_compr_folio() which will benefit from the page pool.
The other obvious benefit is that we no longer need to allocate an array
to hold all those folios, thus one less error path.
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently btrfs_submit_compressed_read() still uses
compressed_bio::compressed_folios[] array.
Change it to allocate each folio and queue them into the compressed bio
so that we do not need to allocate that array.
Considering how small each compressed read bio is (less than 128KiB), we
do not benefit that much from btrfs_alloc_folio_array() anyway,
while we may benefit more from btrfs_alloc_compr_folio() by using
the global folio pool.
So changing from btrfs_alloc_folio_array() to btrfs_alloc_compr_folio()
in a loop should still be fine.
This removes one error path, and paves the way to completely remove
compressed_folios[] array.
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Since it's been replaced by btrfs_compress_bio(), remove all involved
functions.
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This switch has the following benefits:
- A single structure to handle all compression
No more extra members like compressed_folios[] nor compress_type, all
those members.
This means the structure of async_extent is much smaller.
- Simpler error handling
A single cleanup_compressed_bio() will handle everything, no extra
compressed_folios[] array to bother.
Some extra notes:
- Compressed folios releasing
Now we go bio_for_each_folio_all() loop to release the folios of the
bio. This will work for both the old compressed_folios[] array and the
new pure bio method.
For old compressed_folios[], all folios of that array is queued into
the bio, thus releasing the folios from the bio is the same as
releasing each folio of that array. We just need to be sure no double
releasing from the array and bio.
For the new pure bio method, that array is NULL, just usual folio
releasing of the bio.
The only extra note is for end_bbio_compressed_read(), as the folios
are allocated using btrfs_alloc_folio_array(), thus the folios should
only be released by regular folio_put(), not btrfs_free_compr_folio().
- Rounding up the bio to block size
We cannot simply increase bi_size, as that will not increase the
length of the last bvec.
Thus we have to properly add the last part into the bio.
This will be done by the helper, round_up_last_block().
The reason we do not round those bios up at compression time is to get
the unaligned compressed size, so that they can be utilized for
inline extents.
If we round the bios up at *_compress_bio(), then every compressed bio
will be larger than or equal to one fs block, resulting no inline
compressed extent.
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The helper will allocate a new compressed_bio, do the compression, and
return it to the caller.
This greatly simplifies the compression path, as we no longer need to
allocate a folio array thus no extra error path, furthermore the
compressed bio structure can be utilized for submission with very minor
modifications (like rounding up the bi_size and populate the bi_sector).
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The new helper has the following enhancements against the existing
zlib_compress_folios()
- Much smaller parameter list
No more shared IN/OUT members, no need to pre-allocate a
compressed_folios[] array.
Just a workspace and compressed_bio pointer, everything we need can be
extracted from that @cb pointer.
- Ready-to-be-submitted compressed bio
Although the caller still needs to do some common works like
rounding up and zeroing the tailing part of the last fs block.
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The new helper has the following enhancements against the existing
zstd_compress_folios()
- Much smaller parameter list
No more shared IN/OUT members, no need to pre-allocate a
compressed_folios[] array.
Just a workspace and compressed_bio pointer, everything we need can be
extracted from that @cb pointer.
- Ready-to-be-submitted compressed bio
Although the caller still needs to do some common works like
rounding up and zeroing the tailing part of the last fs block.
Overall the workflow is the same as zstd_compress_folios(), but with
some minor changes:
- @start/@len is now constant
For the current input file offset, use @start + @tot_in instead.
The original change of @start and @len makes it pretty hard to know
what value we're really comparing to.
- No more @cur_len
It's only utilized when switching input buffer.
Directly use btrfs_calc_input_length() instead.
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The new helper has the following enhancements against the existing
lzo_compress_folios()
- Much smaller parameter list
No more shared IN/OUT members, no need to pre-allocate a
compressed_folios[] array.
Just a workspace list header and a compressed_bio pointer.
Everything else can be fetched from that @cb pointer.
- Read-to-be-submitted compressed bio
Although the caller still needs to do some common works like
rounding up and zeroing the tailing part of the last fs block.
Some workloads are specific to lZO that is not needed with other
multi-run compression interfaces:
- Need to write a LZO header or segment header
Use the new write_and_queue_folio() helper to do the bio_add_folio()
call and folio switching.
- Need to update the LZO header after compression is done
Use bio_first_folio_all() to grab the first folio and update the header.
- Extra corner case of error handling
This can happen when we have queued part of a folio and hit an error.
In that case those folios will be released by the bio.
Thus we can only release the folio that has no queued part.
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Separate btrfs_load_block_group_* calling path into a function, so that it
can be an entry point of unit test.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Add auto-cleanup helper functions for btrfs_free_dummy_fs_info and
btrfs_free_dummy_block_group.
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
I ran into another sort of trivial bug in v1 of the patch and concluded
that these functions really ought to be unit tested.
These two functions form the core of searching the chunk allocation
pending extent bitmap and have relatively easily definable semantics, so
unit testing them can help ensure the correctness of chunk allocation.
I also made a minor unrelated fix in volumes.h to properly forward
declare btrfs_space_info. Because of the order of the includes in the
new test, this was actually hitting a latent build warning.
Note:
This is an early example for me of a commit authored in part by an AI
agent, so I wanted to more clear about what I did. I defined a
trivial test and explained the set of tests I wanted to the agent and it
produced the large set of test cases seen here. I then checked each test
case to make sure it matched the description and simplified the
constants and numbers until they looked reasonable to me. I then checked
the looping logic to make sure it made sense to the original spirit of
the trivial test. Finally, carefully combed over all the lines it wrote
to loop over the tests it generated to make sure they followed our code
style guide.
Assisted-by: Claude:claude-opus-4-5
Signed-off-by: Boris Burkov <boris@bur.io>
Signed-off-by: David Sterba <dsterba@suse.com>
I have been observing a number of systems aborting at
insert_dev_extents() in btrfs_create_pending_block_groups(). The
following is a sample stack trace of such an abort coming from forced
chunk allocation (typically behind CONFIG_BTRFS_EXPERIMENTAL) but this
can theoretically happen to any DUP chunk allocation.
[81.801] ------------[ cut here ]------------
[81.801] BTRFS: Transaction aborted (error -17)
[81.801] WARNING: fs/btrfs/block-group.c:2876 at btrfs_create_pending_block_groups+0x721/0x770 [btrfs], CPU#1: bash/319
[81.802] Modules linked in: virtio_net btrfs xor zstd_compress raid6_pq null_blk
[81.803] CPU: 1 UID: 0 PID: 319 Comm: bash Kdump: loaded Not tainted 6.19.0-rc6+ #319 NONE
[81.803] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Arch Linux 1.17.0-2-2 04/01/2014
[81.804] RIP: 0010:btrfs_create_pending_block_groups+0x723/0x770 [btrfs]
[81.806] RSP: 0018:ffffa36241a6bce8 EFLAGS: 00010282
[81.806] RAX: 000000000000000d RBX: ffff8e699921e400 RCX: 0000000000000000
[81.807] RDX: 0000000002040001 RSI: 00000000ffffffef RDI: ffffffffc0608bf0
[81.807] RBP: 00000000ffffffef R08: ffff8e69830f6000 R09: 0000000000000007
[81.808] R10: ffff8e699921e5e8 R11: 0000000000000000 R12: ffff8e6999228000
[81.808] R13: ffff8e6984d82000 R14: ffff8e69966a69c0 R15: ffff8e69aa47b000
[81.809] FS: 00007fec6bdd9740(0000) GS:ffff8e6b1b379000(0000) knlGS:0000000000000000
[81.809] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[81.810] CR2: 00005604833670f0 CR3: 0000000116679000 CR4: 00000000000006f0
[81.810] Call Trace:
[81.810] <TASK>
[81.810] __btrfs_end_transaction+0x3e/0x2b0 [btrfs]
[81.811] btrfs_force_chunk_alloc_store+0xcd/0x140 [btrfs]
[81.811] kernfs_fop_write_iter+0x15f/0x240
[81.812] vfs_write+0x264/0x500
[81.812] ksys_write+0x6c/0xe0
[81.812] do_syscall_64+0x66/0x770
[81.812] entry_SYSCALL_64_after_hwframe+0x76/0x7e
[81.813] RIP: 0033:0x7fec6be66197
[81.814] RSP: 002b:00007fffb159dd30 EFLAGS: 00000202 ORIG_RAX: 0000000000000001
[81.815] RAX: ffffffffffffffda RBX: 00007fec6bdd9740 RCX: 00007fec6be66197
[81.815] RDX: 0000000000000002 RSI: 0000560483374f80 RDI: 0000000000000001
[81.816] RBP: 0000560483374f80 R08: 0000000000000000 R09: 0000000000000000
[81.816] R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000000002
[81.817] R13: 00007fec6bfb85c0 R14: 00007fec6bfb5ee0 R15: 00005604833729c0
[81.817] </TASK>
[81.817] irq event stamp: 20039
[81.818] hardirqs last enabled at (20047): [<ffffffff99a68302>] __up_console_sem+0x52/0x60
[81.818] hardirqs last disabled at (20056): [<ffffffff99a682e7>] __up_console_sem+0x37/0x60
[81.819] softirqs last enabled at (19470): [<ffffffff999d2b46>] __irq_exit_rcu+0x96/0xc0
[81.819] softirqs last disabled at (19463): [<ffffffff999d2b46>] __irq_exit_rcu+0x96/0xc0
[81.820] ---[ end trace 0000000000000000 ]---
[81.820] BTRFS: error (device dm-7 state A) in btrfs_create_pending_block_groups:2876: errno=-17 Object already exists
Inspecting these aborts with drgn, I observed a pattern of overlapping
chunk_maps. Note how stripe 1 of the first chunk overlaps in physical
address with stripe 0 of the second chunk.
Physical Start Physical End Length Logical Type Stripe
----------------------------------------------------------------------------------------------------
0x0000000102500000 0x0000000142500000 1.0G 0x0000000641d00000 META|DUP 0/2
0x0000000142500000 0x0000000182500000 1.0G 0x0000000641d00000 META|DUP 1/2
0x0000000142500000 0x0000000182500000 1.0G 0x0000000601d00000 META|DUP 0/2
0x0000000182500000 0x00000001c2500000 1.0G 0x0000000601d00000 META|DUP 1/2
Now how could this possibly happen? All chunk allocation is protected by
the chunk_mutex so racing allocations should see a consistent view of
the CHUNK_ALLOCATED bit in the chunk allocation extent-io-tree
(device->alloc_state as set by chunk_map_device_set_bits()) The tree
itself is protected by a spin lock, and clearing/setting the bits is
always protected by fs_info->mapping_tree_lock, so no race is apparent.
It turns out that there is a subtle bug in the logic regarding chunk
allocations that have happened in the current transaction, known as
"pending extents". The chunk allocation as defined in
find_free_dev_extent() is a loop which searches the commit root of the
dev_root and looks for gaps between DEV_EXTENT items. For those gaps, it
then checks alloc_state bitmap for any pending extents and adjusts the
hole that it finds accordingly. However, the logic in that adjustment
assumes that the first pending extent is the only one in that range.
e.g., given a layout with two non-consecutive pending extents in a hole
passed to dev_extent_hole_check() via *hole_start and *hole_size:
|----pending A----| real hole |----pending B----|
| candidate hole |
*hole_start *hole_start + *hole_size
the code incorrectly returns a "hole" from the end of pending extent A
until the passed in hole end, failing to account for pending B.
However, it is not entirely obvious that it is actually possible to
produce such a layout. I was able to reproduce it, but with some
contortions: I continued to use the force chunk allocation sysfs file
and I introduced a long delay (10 seconds) into the start of the cleaner
thread. I also prevented the unused bgs cleaning logic from ever
deleting metadata bgs. These help make it easier to deterministically
produce the condition but shouldn't really matter if you imagine the
conditions happening by race/luck. Allocations/frees can happen
concurrently with the cleaner thread preparing to process an unused
extent and both create some used chunks with an unused chunk
interleaved, all during one transaction. Then btrfs_delete_unused_bgs()
sees the unused one and clears it, leaving a range with several pending
chunk allocations and a gap in the middle.
The basic idea is that the unused_bgs cleanup work happens on a worker
so if we allocate 3 block groups in one transaction, then the cleaner
work kicked off by the previous transaction comes through and deletes
the middle one of the 3, then the commit root shows no dev extents and
we have the bad pattern in the extent-io-tree. One final consideration
is that the code happens to loop to the next hole if there are no more
extents at all, so we need one more dev extent way past the area we are
working in. Something like the following demonstrates the technique:
# push the BG frontier out to 20G
fallocate -l 20G $mnt/foo
# allocate one more that will prevent the "no more dev extents" luck
fallocate -l 1G $mnt/sticky
# sync
sync
# clear out the allocation area
rm $mnt/foo
sync
_cleaner
# let everything quiesce
sleep 20
sync
# dev tree should have one bg 20G out and the rest at the beginning..
# sort of like an empty FS but with a random sticky chunk.
# kick off the cleaner in the background, remember it will sleep 10s
# before doing interesting work
_cleaner &
sleep 3
# create 3 trivial block groups, all empty, all immediately marked as unused.
echo 1 > "$(_btrfs_sysfs_space_info $dev metadata)/force_chunk_alloc"
echo 1 > "$(_btrfs_sysfs_space_info $dev data)/force_chunk_alloc"
echo 1 > "$(_btrfs_sysfs_space_info $dev metadata)/force_chunk_alloc"
# let the cleaner thread definitely finish, it will remove the data bg
sleep 10
# this allocation sees the non-consecutive pending metadata chunks with
# data chunk gap of 1G and allocates a 2G extent in that hole. ENOSPC!
echo 1 > "$(_btrfs_sysfs_space_info $dev metadata)/force_chunk_alloc"
As for the fix, it is not that obvious. I could not see a trivial way to
do it even by adding backup loops into find_free_dev_extent(), so I
opted to change the semantics of dev_extent_hole_check() to not stop
looping until it finds a sufficiently big hole. For clarity, this also
required changing the helper function contains_pending_extent() into two
new helpers which find the first pending extent and the first suitable
hole in a range.
I attempted to clean up the documentation and range calculations to be
as consistent and clear as possible for the future.
I also looked at the zoned case and concluded that the loop there is
different and not to be unified with this one. As far as I can tell, the
zoned check will only further constrain the hole so looping back to find
more holes is acceptable. Though given that zoned really only appends, I
find it highly unlikely that it is susceptible to this bug.
Fixes: 1b98450816 ("Btrfs: fix find_free_dev_extent() malfunction in case device tree has hole")
Reported-by: Dimitrios Apostolou <jimis@gmx.net>
Closes: https://lore.kernel.org/linux-btrfs/q7760374-q1p4-029o-5149-26p28421s468@tzk.arg/
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Boris Burkov <boris@bur.io>
Signed-off-by: David Sterba <dsterba@suse.com>
Mark Harmstone
Bart Van Assche
Boris Burkov
Johannes Thumshirn
Sun YangKai
Miquel Sabaté Solà
Filipe Manana
Qu Wenruo
Jingkai Tan
Jiasheng Jiang
Adarsh Das
Naohiro Aota