dm-thin-pool kworker与drop cache并发触发死锁

```
======================================================
WARNING: possible circular locking dependency detected
5.10.0-00468-g646bcb923d89 #28 Not tainted
------------------------------------------------------
kworker/u8:0/7 is trying to acquire lock:
ffffffff867fe2f0 (shrinker_rwsem){++++}-{3:3}, at: register_shrinker_prepared+0x23/0x120 mm/vmscan.c:610

but task is already holding lock:
ffff888111072148 (&t->lock){++++}-{3:3}, at: throttle_work_update drivers/md/dm-thin.c:165 [inline]
ffff888111072148 (&t->lock){++++}-{3:3}, at: throttle_work_update drivers/md/dm-thin.c:162 [inline]
ffff888111072148 (&t->lock){++++}-{3:3}, at: process_thin_deferred_bios+0x59d/0x7b0 drivers/md/dm-thin.c:2217

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #1 (&t->lock){++++}-{3:3}:
       validate_chain+0x9b7/0x1650 kernel/locking/lockdep.c:3728
       __lock_acquire+0x7ac/0x1070 kernel/locking/lockdep.c:4954
       lock_acquire+0x197/0x4b0 kernel/locking/lockdep.c:5565
       down_read+0x9b/0x450 kernel/locking/rwsem.c:1507
       throttle_lock drivers/md/dm-thin.c:180 [inline]
       thin_defer_bio_with_throttle drivers/md/dm-thin.c:2674 [inline]
       thin_bio_map.isra.0+0x474/0xb50 drivers/md/dm-thin.c:2730
       __map_bio.isra.0+0x15d/0x9a0 drivers/md/dm.c:1310
       __clone_and_map_data_bio drivers/md/dm.c:1504 [inline]
       __split_and_process_non_flush+0x522/0x850 drivers/md/dm.c:1595
       __split_and_process_bio+0x1c5/0x860 drivers/md/dm.c:1639
       dm_submit_bio+0x1ad/0x5b0 drivers/md/dm.c:1711
       __submit_bio block/blk-core.c:957 [inline]
       __submit_bio_noacct+0x216/0x7e0 block/blk-core.c:1005
       submit_bio_noacct+0xf3/0x270 block/blk-core.c:1084
       submit_bio+0x10f/0x450 block/blk-core.c:1144
       submit_bh_wbc+0x68a/0x930 fs/buffer.c:3053
       ext4_commit_super+0x37e/0x6b0 fs/ext4/super.c:5723
       ext4_handle_error+0x4b2/0x5d0 fs/ext4/super.c:709
       __ext4_error+0x23c/0x4c0 fs/ext4/super.c:818
       ext4_wait_block_bitmap+0x1bf/0x220 fs/ext4/balloc.c:556
       ext4_mb_init_cache+0x672/0x2030 fs/ext4/mballoc.c:937
       ext4_mb_load_buddy_gfp+0x1c2b/0x26a0 fs/ext4/mballoc.c:1220
       ext4_discard_preallocations+0x988/0x1020 fs/ext4/mballoc.c:4466
       ext4_clear_inode+0x40/0x1e0 fs/ext4/super.c:1462
       ext4_evict_inode+0x256/0x1d10 fs/ext4/inode.c:349
       evict+0x2f2/0x780 fs/inode.c:584
       dispose_list+0x11c/0x1e0 fs/inode.c:619
       prune_icache_sb+0xec/0x150 fs/inode.c:814
       super_cache_scan+0x388/0x580 fs/super.c:107
       do_shrink_slab+0x432/0xb50 mm/vmscan.c:715
       shrink_slab+0x172/0x280 mm/vmscan.c:875
       drop_slab_node+0xa1/0x170 mm/vmscan.c:910
       drop_slab+0x55/0xa0 mm/vmscan.c:920
       drop_caches_sysctl_handler+0xff/0x120 fs/drop_caches.c:65
       proc_sys_call_handler+0x34d/0x630 fs/proc/proc_sysctl.c:592
       call_write_iter include/linux/fs.h:1961 [inline]
       new_sync_write+0x42e/0x670 fs/read_write.c:513
       vfs_write+0x73c/0xac0 fs/read_write.c:600
       ksys_write+0x132/0x260 fs/read_write.c:653
       do_syscall_64+0x30/0x40 arch/x86/entry/common.c:46
       entry_SYSCALL_64_after_hwframe+0x61/0xc6

-> #0 (shrinker_rwsem){++++}-{3:3}:
       check_prev_add+0x160/0x2540 kernel/locking/lockdep.c:2988
       check_prevs_add+0x25f/0x660 kernel/locking/lockdep.c:3113
       validate_chain+0x9b7/0x1650 kernel/locking/lockdep.c:3728
       __lock_acquire+0x7ac/0x1070 kernel/locking/lockdep.c:4954
       lock_acquire+0x197/0x4b0 kernel/locking/lockdep.c:5565
       down_write+0x92/0x160 kernel/locking/rwsem.c:1560
       register_shrinker_prepared+0x23/0x120 mm/vmscan.c:610
       register_shrinker+0x36/0x50 mm/vmscan.c:622
       dm_bufio_client_create+0xb65/0xf10 drivers/md/dm-bufio.c:1810
       dm_block_manager_create+0x82/0x140 drivers/md/persistent-data/dm-block-manager.c:391
       dm_pool_abort_metadata+0x9f/0x3a0 drivers/md/dm-thin-metadata.c:1899
       abort_transaction+0x8d/0x110 drivers/md/dm-thin.c:2630
       metadata_operation_failed+0x48/0xb0 drivers/md/dm-thin.c:2646
       commit+0xf6/0x330 drivers/md/dm-thin.c:1501
       process_deferred_bios+0x50d/0x6d0 drivers/md/dm-thin.c:2376
       do_worker+0x417/0x6e0 drivers/md/dm-thin.c:2413
       process_one_work+0x83e/0x1420 kernel/workqueue.c:2280
       worker_thread+0xe1/0xd80 kernel/workqueue.c:2426
       kthread+0x353/0x470 kernel/kthread.c:313
       ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:299

other info that might help us debug this:

Possible unsafe locking scenario:

CPU0                    CPU1
       ----                    ----
  lock(&t->lock);
                               lock(shrinker_rwsem);
                               lock(&t->lock);
  lock(shrinker_rwsem);

*** DEADLOCK ***
```

GVP openEuler/kernel

内容风险标识

评论 (1)

GVPopenEuler/kernel

内容风险标识