======================================================
WARNING: possible circular locking dependency detected
syzkaller #0 Tainted: G             L     
------------------------------------------------------
kworker/u10:16/3918 is trying to acquire lock:
ffff8881bcdc00c8 (&ei->xattr_sem){++++}-{4:4}, at: ext4_destroy_inline_data+0x28/0xe0

but task is already holding lock:
ffff8881a4b88b98 (&sbi->s_writepages_rwsem){++++}-{0:0}, at: ext4_writepages+0x205/0x3b0

which lock already depends on the new lock.


the existing dependency chain (in reverse order) is:

-> #1 (&sbi->s_writepages_rwsem){++++}-{0:0}:
       percpu_down_read_internal+0x48/0x1c0
       ext4_writepages+0x205/0x3b0
       do_writepages+0x32e/0x550
       __writeback_single_inode+0x133/0x1230
       writeback_single_inode+0x493/0xc70
       write_inode_now+0x1c2/0x290
       iput+0xa73/0x1020
       ext4_xattr_block_set+0x1fd4/0x2ad0
       ext4_expand_extra_isize_ea+0x12cf/0x1ea0
       __ext4_expand_extra_isize+0x30d/0x400
       __ext4_mark_inode_dirty+0x45c/0x6e0
       ext4_evict_inode+0x7a1/0xe60
       evict+0x61e/0xb10
       ext4_orphan_cleanup+0xc38/0x1470
       ext4_fill_super+0x5932/0x61f0
       get_tree_bdev_flags+0x431/0x4f0
       vfs_get_tree+0x92/0x2a0
       do_new_mount+0x329/0xa50
       __se_sys_mount+0x31d/0x420
       do_syscall_64+0xe2/0xf80
       entry_SYSCALL_64_after_hwframe+0x77/0x7f

-> #0 (&ei->xattr_sem){++++}-{4:4}:
       __lock_acquire+0x15a5/0x2cf0
       lock_acquire+0x106/0x330
       down_write+0x96/0x200
       ext4_destroy_inline_data+0x28/0xe0
       ext4_do_writepages+0x51e/0x4600
       ext4_writepages+0x241/0x3b0
       do_writepages+0x32e/0x550
       __writeback_single_inode+0x133/0x1230
       writeback_sb_inodes+0x92c/0x1820
       wb_writeback+0x445/0xad0
       wb_workfn+0x3f8/0xef0
       process_scheduled_works+0xaec/0x17a0
       worker_thread+0x89f/0xd90
       kthread+0x726/0x8b0
       ret_from_fork+0x51b/0xa40
       ret_from_fork_asm+0x1a/0x30

other info that might help us debug this:

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  rlock(&sbi->s_writepages_rwsem);
                               lock(&ei->xattr_sem);
                               lock(&sbi->s_writepages_rwsem);
  lock(&ei->xattr_sem);

 *** DEADLOCK ***

3 locks held by kworker/u10:16/3918:
 #0: ffff888161aef948 ((wq_completion)writeback){+.+.}-{0:0}, at: process_scheduled_works+0x9d4/0x17a0
 #1: ffffc90023cd7bc0 ((work_completion)(&(&wb->dwork)->work)){+.+.}-{0:0}, at: process_scheduled_works+0xa0f/0x17a0
 #2: ffff8881a4b88b98 (&sbi->s_writepages_rwsem){++++}-{0:0}, at: ext4_writepages+0x205/0x3b0

stack backtrace:
CPU: 1 UID: 0 PID: 3918 Comm: kworker/u10:16 Tainted: G             L      syzkaller #0 PREEMPT(full) 
Tainted: [L]=SOFTLOCKUP
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.2-debian-1.16.2-1 04/01/2014
Workqueue: writeback wb_workfn (flush-7:4)
Call Trace:
 <TASK>
 dump_stack_lvl+0xe8/0x150
 print_circular_bug+0x2e1/0x300
 check_noncircular+0x12e/0x150
 __lock_acquire+0x15a5/0x2cf0
 lock_acquire+0x106/0x330
 down_write+0x96/0x200
 ext4_destroy_inline_data+0x28/0xe0
 ext4_do_writepages+0x51e/0x4600
 ext4_writepages+0x241/0x3b0
 do_writepages+0x32e/0x550
 __writeback_single_inode+0x133/0x1230
 writeback_sb_inodes+0x92c/0x1820
 wb_writeback+0x445/0xad0
 wb_workfn+0x3f8/0xef0
 process_scheduled_works+0xaec/0x17a0
 worker_thread+0x89f/0xd90
 kthread+0x726/0x8b0
 ret_from_fork+0x51b/0xa40
 ret_from_fork_asm+0x1a/0x30
 </TASK>
EXT4-fs error (device loop4): ext4_mb_generate_buddy:1306: group 0, block bitmap and bg descriptor inconsistent: 25 vs 281 free clusters
EXT4-fs (loop4): Delayed block allocation failed for inode 15 at logical offset 0 with max blocks 1 with error 28
EXT4-fs (loop4): This should not happen!! Data will be lost

EXT4-fs (loop4): Total free blocks count 0
EXT4-fs (loop4): Free/Dirty block details
EXT4-fs (loop4): free_blocks=4096
EXT4-fs (loop4): dirty_blocks=512
EXT4-fs (loop4): Block reservation details
EXT4-fs (loop4): i_reserved_data_blocks=32
