f2fs: fix to avoid potential deadlock

As Jiaming Zhang and syzbot reported, there is potential deadlock in
f2fs as below:

Chain exists of:
  &sbi->cp_rwsem --> fs_reclaim --> sb_internal#2

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  rlock(sb_internal#2);
                               lock(fs_reclaim);
                               lock(sb_internal#2);
  rlock(&sbi->cp_rwsem);

 *** DEADLOCK ***

3 locks held by kswapd0/73:
 #0: ffffffff8e247a40 (fs_reclaim){+.+.}-{0:0}, at: balance_pgdat mm/vmscan.c:7015 [inline]
 #0: ffffffff8e247a40 (fs_reclaim){+.+.}-{0:0}, at: kswapd+0x951/0x2800 mm/vmscan.c:7389
 #1: ffff8880118400e0 (&type->s_umount_key#50){.+.+}-{4:4}, at: super_trylock_shared fs/super.c:562 [inline]
 #1: ffff8880118400e0 (&type->s_umount_key#50){.+.+}-{4:4}, at: super_cache_scan+0x91/0x4b0 fs/super.c:197
 #2: ffff888011840610 (sb_internal#2){.+.+}-{0:0}, at: f2fs_evict_inode+0x8d9/0x1b60 fs/f2fs/inode.c:890

stack backtrace:
CPU: 0 UID: 0 PID: 73 Comm: kswapd0 Not tainted syzkaller #0 PREEMPT(full)
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2~bpo12+1 04/01/2014
Call Trace:
 <TASK>
 dump_stack_lvl+0x189/0x250 lib/dump_stack.c:120
 print_circular_bug+0x2ee/0x310 kernel/locking/lockdep.c:2043
 check_noncircular+0x134/0x160 kernel/locking/lockdep.c:2175
 check_prev_add kernel/locking/lockdep.c:3165 [inline]
 check_prevs_add kernel/locking/lockdep.c:3284 [inline]
 validate_chain+0xb9b/0x2140 kernel/locking/lockdep.c:3908
 __lock_acquire+0xab9/0xd20 kernel/locking/lockdep.c:5237
 lock_acquire+0x120/0x360 kernel/locking/lockdep.c:5868
 down_read+0x46/0x2e0 kernel/locking/rwsem.c:1537
 f2fs_down_read fs/f2fs/f2fs.h:2278 [inline]
 f2fs_lock_op fs/f2fs/f2fs.h:2357 [inline]
 f2fs_do_truncate_blocks+0x21c/0x10c0 fs/f2fs/file.c:791
 f2fs_truncate_blocks+0x10a/0x300 fs/f2fs/file.c:867
 f2fs_truncate+0x489/0x7c0 fs/f2fs/file.c:925
 f2fs_evict_inode+0x9f2/0x1b60 fs/f2fs/inode.c:897
 evict+0x504/0x9c0 fs/inode.c:810
 f2fs_evict_inode+0x1dc/0x1b60 fs/f2fs/inode.c:853
 evict+0x504/0x9c0 fs/inode.c:810
 dispose_list fs/inode.c:852 [inline]
 prune_icache_sb+0x21b/0x2c0 fs/inode.c:1000
 super_cache_scan+0x39b/0x4b0 fs/super.c:224
 do_shrink_slab+0x6ef/0x1110 mm/shrinker.c:437
 shrink_slab_memcg mm/shrinker.c:550 [inline]
 shrink_slab+0x7ef/0x10d0 mm/shrinker.c:628
 shrink_one+0x28a/0x7c0 mm/vmscan.c:4955
 shrink_many mm/vmscan.c:5016 [inline]
 lru_gen_shrink_node mm/vmscan.c:5094 [inline]
 shrink_node+0x315d/0x3780 mm/vmscan.c:6081
 kswapd_shrink_node mm/vmscan.c:6941 [inline]
 balance_pgdat mm/vmscan.c:7124 [inline]
 kswapd+0x147c/0x2800 mm/vmscan.c:7389
 kthread+0x70e/0x8a0 kernel/kthread.c:463
 ret_from_fork+0x4bc/0x870 arch/x86/kernel/process.c:158
 ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:245
 </TASK>

The root cause is deadlock among four locks as below:

kswapd
- fs_reclaim				--- Lock A
 - shrink_one
  - evict
   - f2fs_evict_inode
    - sb_start_intwrite			--- Lock B

- iput
 - evict
  - f2fs_evict_inode
   - sb_start_intwrite			--- Lock B
   - f2fs_truncate
    - f2fs_truncate_blocks
     - f2fs_do_truncate_blocks
      - f2fs_lock_op			--- Lock C

ioctl
- f2fs_ioc_commit_atomic_write
 - f2fs_lock_op				--- Lock C
  - __f2fs_commit_atomic_write
   - __replace_atomic_write_block
    - f2fs_get_dnode_of_data
     - __get_node_folio
      - f2fs_check_nid_range
       - f2fs_handle_error
        - f2fs_record_errors
         - f2fs_down_write		--- Lock D

open
- do_open
 - do_truncate
  - security_inode_need_killpriv
   - f2fs_getxattr
    - lookup_all_xattrs
     - f2fs_handle_error
      - f2fs_record_errors
       - f2fs_down_write		--- Lock D
        - f2fs_commit_super
         - read_mapping_folio
          - filemap_alloc_folio_noprof
           - prepare_alloc_pages
            - fs_reclaim_acquire	--- Lock A

In order to avoid such deadlock, we need to avoid grabbing sb_lock in
f2fs_handle_error(), so, let's use asynchronous method instead:
- remove f2fs_handle_error() implementation
- rename f2fs_handle_error_async() to f2fs_handle_error()
- spread f2fs_handle_error()

Fixes: 95fa90c9e5a7 ("f2fs: support recording errors into superblock")
Cc: stable@kernel.org
Reported-by: syzbot+14b90e1156b9f6fc1266@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/linux-f2fs-devel/68eae49b.050a0220.ac43.0001.GAE@google.com
Reported-by: Jiaming Zhang <r772577952@gmail.com>
Closes: https://lore.kernel.org/lkml/CANypQFa-Gy9sD-N35o3PC+FystOWkNuN8pv6S75HLT0ga-Tzgw@mail.gmail.com
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

0 files changed, 0 insertions, 0 deletions