]> git.baikalelectronics.ru Git - kernel.git/commit
btrfs: move btrfs_rm_dev_replace_free_srcdev outside of all locks
authorJosef Bacik <josef@toxicpanda.com>
Thu, 20 Aug 2020 15:18:27 +0000 (11:18 -0400)
committerDavid Sterba <dsterba@suse.com>
Wed, 30 Sep 2020 17:34:24 +0000 (19:34 +0200)
commit5fb115703f6b156a5d7cbd14023f5f99f95b0fb8
tree4e9222bfc3dc83be7f267f22ad29ebaaf1ca3dd2
parentc4763e4965db5671578020fc745020f144acbd8a
btrfs: move btrfs_rm_dev_replace_free_srcdev outside of all locks

When closing and freeing the source device we could end up doing our
final blkdev_put() on the bdev, which will grab the bd_mutex.  As such
we want to be holding as few locks as possible, so move this call
outside of the dev_replace->lock_finishing_cancel_unmount lock.  Since
we're modifying the fs_devices we need to make sure we're holding the
uuid_mutex here, so take that as well.

There's a report from syzbot probably hitting one of the cases where
the bd_mutex and device_list_mutex are taken in the wrong order, however
it's not with device replace, like this patch fixes. As there's no
reproducer available so far, we can't verify the fix.

https://lore.kernel.org/lkml/000000000000fc04d105afcf86d7@google.com/
link: https://syzkaller.appspot.com/bug?extid=84a0634dc5d21d488419
  WARNING: possible circular locking dependency detected
  5.9.0-rc5-syzkaller #0 Not tainted
  ------------------------------------------------------
  syz-executor.0/6878 is trying to acquire lock:
  ffff88804c17d780 (&bdev->bd_mutex){+.+.}-{3:3}, at: blkdev_put+0x30/0x520 fs/block_dev.c:1804

  but task is already holding lock:
  ffff8880908cfce0 (&fs_devs->device_list_mutex){+.+.}-{3:3}, at: close_fs_devices.part.0+0x2e/0x800 fs/btrfs/volumes.c:1159

  which lock already depends on the new lock.

  the existing dependency chain (in reverse order) is:

  -> #4 (&fs_devs->device_list_mutex){+.+.}-{3:3}:
 __mutex_lock_common kernel/locking/mutex.c:956 [inline]
 __mutex_lock+0x134/0x10e0 kernel/locking/mutex.c:1103
 btrfs_finish_chunk_alloc+0x281/0xf90 fs/btrfs/volumes.c:5255
 btrfs_create_pending_block_groups+0x2f3/0x700 fs/btrfs/block-group.c:2109
 __btrfs_end_transaction+0xf5/0x690 fs/btrfs/transaction.c:916
 find_free_extent_update_loop fs/btrfs/extent-tree.c:3807 [inline]
 find_free_extent+0x23b7/0x2e60 fs/btrfs/extent-tree.c:4127
 btrfs_reserve_extent+0x166/0x460 fs/btrfs/extent-tree.c:4206
 cow_file_range+0x3de/0x9b0 fs/btrfs/inode.c:1063
 btrfs_run_delalloc_range+0x2cf/0x1410 fs/btrfs/inode.c:1838
 writepage_delalloc+0x150/0x460 fs/btrfs/extent_io.c:3439
 __extent_writepage+0x441/0xd00 fs/btrfs/extent_io.c:3653
 extent_write_cache_pages.constprop.0+0x69d/0x1040 fs/btrfs/extent_io.c:4249
 extent_writepages+0xcd/0x2b0 fs/btrfs/extent_io.c:4370
 do_writepages+0xec/0x290 mm/page-writeback.c:2352
 __writeback_single_inode+0x125/0x1400 fs/fs-writeback.c:1461
 writeback_sb_inodes+0x53d/0xf40 fs/fs-writeback.c:1721
 wb_writeback+0x2ad/0xd40 fs/fs-writeback.c:1894
 wb_do_writeback fs/fs-writeback.c:2039 [inline]
 wb_workfn+0x2dc/0x13e0 fs/fs-writeback.c:2080
 process_one_work+0x94c/0x1670 kernel/workqueue.c:2269
 worker_thread+0x64c/0x1120 kernel/workqueue.c:2415
 kthread+0x3b5/0x4a0 kernel/kthread.c:292
 ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:294

  -> #3 (sb_internal#2){.+.+}-{0:0}:
 percpu_down_read include/linux/percpu-rwsem.h:51 [inline]
 __sb_start_write+0x234/0x470 fs/super.c:1672
 sb_start_intwrite include/linux/fs.h:1690 [inline]
 start_transaction+0xbe7/0x1170 fs/btrfs/transaction.c:624
 find_free_extent_update_loop fs/btrfs/extent-tree.c:3789 [inline]
 find_free_extent+0x25e1/0x2e60 fs/btrfs/extent-tree.c:4127
 btrfs_reserve_extent+0x166/0x460 fs/btrfs/extent-tree.c:4206
 cow_file_range+0x3de/0x9b0 fs/btrfs/inode.c:1063
 btrfs_run_delalloc_range+0x2cf/0x1410 fs/btrfs/inode.c:1838
 writepage_delalloc+0x150/0x460 fs/btrfs/extent_io.c:3439
 __extent_writepage+0x441/0xd00 fs/btrfs/extent_io.c:3653
 extent_write_cache_pages.constprop.0+0x69d/0x1040 fs/btrfs/extent_io.c:4249
 extent_writepages+0xcd/0x2b0 fs/btrfs/extent_io.c:4370
 do_writepages+0xec/0x290 mm/page-writeback.c:2352
 __writeback_single_inode+0x125/0x1400 fs/fs-writeback.c:1461
 writeback_sb_inodes+0x53d/0xf40 fs/fs-writeback.c:1721
 wb_writeback+0x2ad/0xd40 fs/fs-writeback.c:1894
 wb_do_writeback fs/fs-writeback.c:2039 [inline]
 wb_workfn+0x2dc/0x13e0 fs/fs-writeback.c:2080
 process_one_work+0x94c/0x1670 kernel/workqueue.c:2269
 worker_thread+0x64c/0x1120 kernel/workqueue.c:2415
 kthread+0x3b5/0x4a0 kernel/kthread.c:292
 ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:294

  -> #2 ((work_completion)(&(&wb->dwork)->work)){+.+.}-{0:0}:
 __flush_work+0x60e/0xac0 kernel/workqueue.c:3041
 wb_shutdown+0x180/0x220 mm/backing-dev.c:355
 bdi_unregister+0x174/0x590 mm/backing-dev.c:872
 del_gendisk+0x820/0xa10 block/genhd.c:933
 loop_remove drivers/block/loop.c:2192 [inline]
 loop_control_ioctl drivers/block/loop.c:2291 [inline]
 loop_control_ioctl+0x3b1/0x480 drivers/block/loop.c:2257
 vfs_ioctl fs/ioctl.c:48 [inline]
 __do_sys_ioctl fs/ioctl.c:753 [inline]
 __se_sys_ioctl fs/ioctl.c:739 [inline]
 __x64_sys_ioctl+0x193/0x200 fs/ioctl.c:739
 do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

  -> #1 (loop_ctl_mutex){+.+.}-{3:3}:
 __mutex_lock_common kernel/locking/mutex.c:956 [inline]
 __mutex_lock+0x134/0x10e0 kernel/locking/mutex.c:1103
 lo_open+0x19/0xd0 drivers/block/loop.c:1893
 __blkdev_get+0x759/0x1aa0 fs/block_dev.c:1507
 blkdev_get fs/block_dev.c:1639 [inline]
 blkdev_open+0x227/0x300 fs/block_dev.c:1753
 do_dentry_open+0x4b9/0x11b0 fs/open.c:817
 do_open fs/namei.c:3251 [inline]
 path_openat+0x1b9a/0x2730 fs/namei.c:3368
 do_filp_open+0x17e/0x3c0 fs/namei.c:3395
 do_sys_openat2+0x16d/0x420 fs/open.c:1168
 do_sys_open fs/open.c:1184 [inline]
 __do_sys_open fs/open.c:1192 [inline]
 __se_sys_open fs/open.c:1188 [inline]
 __x64_sys_open+0x119/0x1c0 fs/open.c:1188
 do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

  -> #0 (&bdev->bd_mutex){+.+.}-{3:3}:
 check_prev_add kernel/locking/lockdep.c:2496 [inline]
 check_prevs_add kernel/locking/lockdep.c:2601 [inline]
 validate_chain kernel/locking/lockdep.c:3218 [inline]
 __lock_acquire+0x2a96/0x5780 kernel/locking/lockdep.c:4426
 lock_acquire+0x1f3/0xae0 kernel/locking/lockdep.c:5006
 __mutex_lock_common kernel/locking/mutex.c:956 [inline]
 __mutex_lock+0x134/0x10e0 kernel/locking/mutex.c:1103
 blkdev_put+0x30/0x520 fs/block_dev.c:1804
 btrfs_close_bdev fs/btrfs/volumes.c:1117 [inline]
 btrfs_close_bdev fs/btrfs/volumes.c:1107 [inline]
 btrfs_close_one_device fs/btrfs/volumes.c:1133 [inline]
 close_fs_devices.part.0+0x1a4/0x800 fs/btrfs/volumes.c:1161
 close_fs_devices fs/btrfs/volumes.c:1193 [inline]
 btrfs_close_devices+0x95/0x1f0 fs/btrfs/volumes.c:1179
 close_ctree+0x688/0x6cb fs/btrfs/disk-io.c:4149
 generic_shutdown_super+0x144/0x370 fs/super.c:464
 kill_anon_super+0x36/0x60 fs/super.c:1108
 btrfs_kill_super+0x38/0x50 fs/btrfs/super.c:2265
 deactivate_locked_super+0x94/0x160 fs/super.c:335
 deactivate_super+0xad/0xd0 fs/super.c:366
 cleanup_mnt+0x3a3/0x530 fs/namespace.c:1118
 task_work_run+0xdd/0x190 kernel/task_work.c:141
 tracehook_notify_resume include/linux/tracehook.h:188 [inline]
 exit_to_user_mode_loop kernel/entry/common.c:163 [inline]
 exit_to_user_mode_prepare+0x1e1/0x200 kernel/entry/common.c:190
 syscall_exit_to_user_mode+0x7e/0x2e0 kernel/entry/common.c:265
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

  other info that might help us debug this:

  Chain exists of:
    &bdev->bd_mutex --> sb_internal#2 --> &fs_devs->device_list_mutex

   Possible unsafe locking scenario:

 CPU0                    CPU1
 ----                    ----
    lock(&fs_devs->device_list_mutex);
 lock(sb_internal#2);
 lock(&fs_devs->device_list_mutex);
    lock(&bdev->bd_mutex);

   *** DEADLOCK ***

  3 locks held by syz-executor.0/6878:
   #0: ffff88809070c0e0 (&type->s_umount_key#70){++++}-{3:3}, at: deactivate_super+0xa5/0xd0 fs/super.c:365
   #1: ffffffff8a5b37a8 (uuid_mutex){+.+.}-{3:3}, at: btrfs_close_devices+0x23/0x1f0 fs/btrfs/volumes.c:1178
   #2: ffff8880908cfce0 (&fs_devs->device_list_mutex){+.+.}-{3:3}, at: close_fs_devices.part.0+0x2e/0x800 fs/btrfs/volumes.c:1159

  stack backtrace:
  CPU: 0 PID: 6878 Comm: syz-executor.0 Not tainted 5.9.0-rc5-syzkaller #0
  Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
  Call Trace:
   __dump_stack lib/dump_stack.c:77 [inline]
   dump_stack+0x198/0x1fd lib/dump_stack.c:118
   check_noncircular+0x324/0x3e0 kernel/locking/lockdep.c:1827
   check_prev_add kernel/locking/lockdep.c:2496 [inline]
   check_prevs_add kernel/locking/lockdep.c:2601 [inline]
   validate_chain kernel/locking/lockdep.c:3218 [inline]
   __lock_acquire+0x2a96/0x5780 kernel/locking/lockdep.c:4426
   lock_acquire+0x1f3/0xae0 kernel/locking/lockdep.c:5006
   __mutex_lock_common kernel/locking/mutex.c:956 [inline]
   __mutex_lock+0x134/0x10e0 kernel/locking/mutex.c:1103
   blkdev_put+0x30/0x520 fs/block_dev.c:1804
   btrfs_close_bdev fs/btrfs/volumes.c:1117 [inline]
   btrfs_close_bdev fs/btrfs/volumes.c:1107 [inline]
   btrfs_close_one_device fs/btrfs/volumes.c:1133 [inline]
   close_fs_devices.part.0+0x1a4/0x800 fs/btrfs/volumes.c:1161
   close_fs_devices fs/btrfs/volumes.c:1193 [inline]
   btrfs_close_devices+0x95/0x1f0 fs/btrfs/volumes.c:1179
   close_ctree+0x688/0x6cb fs/btrfs/disk-io.c:4149
   generic_shutdown_super+0x144/0x370 fs/super.c:464
   kill_anon_super+0x36/0x60 fs/super.c:1108
   btrfs_kill_super+0x38/0x50 fs/btrfs/super.c:2265
   deactivate_locked_super+0x94/0x160 fs/super.c:335
   deactivate_super+0xad/0xd0 fs/super.c:366
   cleanup_mnt+0x3a3/0x530 fs/namespace.c:1118
   task_work_run+0xdd/0x190 kernel/task_work.c:141
   tracehook_notify_resume include/linux/tracehook.h:188 [inline]
   exit_to_user_mode_loop kernel/entry/common.c:163 [inline]
   exit_to_user_mode_prepare+0x1e1/0x200 kernel/entry/common.c:190
   syscall_exit_to_user_mode+0x7e/0x2e0 kernel/entry/common.c:265
   entry_SYSCALL_64_after_hwframe+0x44/0xa9
  RIP: 0033:0x460027
  RSP: 002b:00007fff59216328 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6
  RAX: 0000000000000000 RBX: 0000000000076035 RCX: 0000000000460027
  RDX: 0000000000403188 RSI: 0000000000000002 RDI: 00007fff592163d0
  RBP: 0000000000000333 R08: 0000000000000000 R09: 000000000000000b
  R10: 0000000000000005 R11: 0000000000000246 R12: 00007fff59217460
  R13: 0000000002df2a60 R14: 0000000000000000 R15: 00007fff59217460

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
[ add syzbot reference ]
Signed-off-by: David Sterba <dsterba@suse.com>
fs/btrfs/dev-replace.c
fs/btrfs/volumes.c