zfs change-key <fs> causes kernel panic on one (of many) encrypted filesystems

[jstenback]

System information

Type	Version/Name
Distribution Name	Arch
Distribution Version	current, lts kernel
Linux Kernel	4.19.92-1-lts
Architecture	x86_64
ZFS Version	0.8.2-1
SPL Version	0.8.2-1

Describe the problem you're observing

zfs change-key causes kernel panic on one of my encrypted filesystems (I have many, only one exhibits this problem). See journal log info below.

Describe how to reproduce the problem

Unclear, except in my case on one of my systems this reproduces every time with this one particular filesystem.

The one thing that sets this filesystem apart from the others I have where I don't see this problem on this host is that it contains zfs volumes that are not encrypted.

Include any warning/errors/backtraces from the system logs

kernel:  ret_from_fork+0x35/0x40
kernel:  ? kthread_park+0x80/0x80
kernel:  ? __thread_exit+0x20/0x20 [spl]
kernel:  kthread+0xfb/0x130
kernel:  thread_generic_wrapper+0x6f/0x80 [spl]
kernel:  ? txg_thread_exit.isra.0+0x60/0x60 [zfs]
kernel:  ? __switch_to+0xc5/0x490
kernel:  txg_sync_thread+0x2d2/0x4b0 [zfs]
kernel:  ? spa_txg_history_init_io+0x123/0x130 [zfs]
kernel:  spa_sync+0x5a9/0xf90 [zfs]
kernel:  dsl_pool_sync+0x3e1/0x4d0 [zfs]
kernel:  dsl_sync_task_sync+0xa6/0xf0 [zfs]
kernel:  spa_keystore_change_key_sync+0x161/0x3c0 [zfs]
kernel:  spa_keystore_change_key_sync_impl+0x1e5/0x320 [zfs]
kernel:  spa_keystore_change_key_sync_impl+0x1e5/0x320 [zfs]
kernel:  spa_keystore_change_key_sync_impl+0x253/0x320 [zfs]
kernel:  ? dsl_dir_hold_obj+0xad/0x3f0 [zfs]
kernel:  ? dmu_buf_replace_user+0x4d/0x70 [zfs]
kernel:  spl_panic+0xd4/0xfc [spl]
kernel:  dump_stack+0x66/0x90
kernel: Call Trace:
kernel: Hardware name: Supermicro Super Server/X10DRL-i, BIOS 2.0 12/18/2015
kernel: CPU: 23 PID: 3002 Comm: txg_sync Tainted: P         C OE     4.19.92-1-lts #1
kernel: Showing stack for process 3002
kernel: PANIC at dsl_crypt.c:1448:spa_keystore_change_key_sync_impl()
kernel: VERIFY3(0 == dsl_dir_get_encryption_root_ddobj(dd, &curr_rddobj)) failed (0 == 2)

[behlendorf]

This is a duplicate of #9524 which was resolved by commit bae11ba. This fix is queued up for the pending 0.8.3 tag.

[jstenback]

Ah, indeed! Apologies for not finding that one before filing this issue. Thanks!