zdb: SIGSEGV in dsl_deadlist_close while verifying pool checksum

[zzy9001]

System information

Type	Version/Name
Distribution Name	CachyOS
Distribution Version	rolling
Kernel Version	6.6.69-1-cachyos-lts
Architecture	x86_64
OpenZFS Version	zfs-2.3.0-1;zfs-kmod-2.2.7-1

Describe the problem you're observing

After experiencing several unexpected system shutdowns, the ZFS pool appeared to be functioning normally upon system restart. As a precautionary measure, I decided to run zdb -cc to verify data integrity. But the command crashes with a SIGSEGV signal while traversing blocks for checksum verification. The crash occurs during the deadlist processing (dsl_deadlist_close). Before the crash, multiple "error 52" reading errors were encountered on various blocks.

Describe how to reproduce the problem

1. Run the following command on the affected pool:

sudo zdb -cc zpcachyos

2. The command initially proceeds with block traversal and verification
3. Multiple error 52 readings appear
4. The process eventually crashes with SIGSEGV while processing deadlist

Include any warning/errors/backtraces from the system logs

Traversing all blocks to verify checksums and verify nothing leaked ...
loading concrete vdev 0, metaslab 231 of 232 ...
6.44G completed ( 103MB/s) estimated time remaining: 3hr 10min 51sec
zdb_blkptr_cb: Got error 52 reading <529, 18472, 0, 15>  -- skipping
zdb_blkptr_cb: Got error 52 reading <529, 979002, 0, 2>  -- skipping
zdb_blkptr_cb: Got error 52 reading <529, 980200, 0, 0>  -- skipping
zdb_blkptr_cb: Got error 52 reading <529, 2973302, 0, 6>  -- skipping
zdb_blkptr_cb: Got error 52 reading <529, 6645040, 0, 0>  -- skipping
zdb_blkptr_cb: Got error 52 reading <529, 7932296, 0, 7>  -- skipping
zdb_blkptr_cb: Got error 52 reading <529, 7932296, 0, 4>  -- skipping
zdb_blkptr_cb: Got error 52 reading <529, 7932296, 0, 0>  -- skipping
zdb_blkptr_cb: Got error 52 reading <529, 7946710, 0, 0>  -- skipping
zdb_blkptr_cb: Got error 52 reading <529, 7946684, 0, 0>  -- skipping
769G completed ( 125MB/s) estimated time remaining: 0hr 53min 09sec
zdb_blkptr_cb: Got error 52 reading <771, 0, -1, 0>  -- skipping
774G completed ( 125MB/s) estimated time remaining: 0hr 52min 35sec

Call trace:
/usr/lib/libzpool.so.6(libspl_backtrace+0x38) [0x77be8c409db8]
zdb(+0xc50c) [0x555e20d8450c]
/usr/lib/libc.so.6(+0x42150) [0x77be8b8f4150]
/usr/lib/libc.so.6(pthread_mutex_lock+0x4) [0x77be8b954c74]
/usr/lib/libzpool.so.6(dsl_deadlist_close+0x229) [0x77be8c1ac6f9]
/usr/lib/libzpool.so.6(dsl_dir_hold_obj+0x86c) [0x77be8c1bda4c]
/usr/lib/libzpool.so.6(dsl_dataset_hold_obj+0x3bb) [0x77be8c19971b]
/usr/lib/libzpool.so.6(traverse_pool+0x200) [0x77be8c1731e0]
zdb(+0x240c3) [0x555e20d9c0c3]
zdb(+0xa812) [0x555e20d82812]
/usr/lib/libc.so.6(+0x2618e) [0x77be8b8d818e]
/usr/lib/libc.so.6(__libc_start_main+0x8a) [0x77be8b8d824a]
zdb(+0xb635) [0x555e20d83635]

fish: Job 1, 'sudo zdb -cc zpcachyos' terminated by signal SIGSEGV (Address boundary error)

$zpool status -v
  pool: zpcachyos
 state: ONLINE
  scan: scrub repaired 0B in 00:10:56 with 0 errors on Wed Jan 15 09:31:10 2025
config:

        NAME                                    STATE     READ WRITE CKSUM
        zpcachyos                               ONLINE       0     0     0
          a5d06edf-0256-4f19-9065-00613a102ba5  ONLINE       0     0     0

errors: No known data errors