panic: sbuf_clear makes no sense on sbuf 0xfffff80004234000 with drain

root@mercat1:/tmp # cat sysctl.sh
#!/bin/sh

kldstat | grep -q zfs.ko || kldload zfs.ko
start=`date +%s`
while [ $((`date +%s` - start)) -lt 180 ]; do
        for i in `jot 3`; do
                sysctl -a > /dev/null 2>&1 &
        done
        wait
done
kldstat | grep -q zfs.ko && kldunload zfs.ko
exit 0
root@mercat1:/tmp # ./sysctl.sh
ZFS filesystem version: 5
ZFS storage pool version: features support (5000)
panic: sbuf_clear makes no sense on sbuf 0xfffff80004234000 with drain
cpuid = 0
time = 1703841857
KDB: stack backtrace:
db_trace_self_wrapper() at db_trace_self_wrapper+0x2b/frame 0xfffffe016a270960
vpanic() at vpanic+0x131/frame 0xfffffe016a270a90
panic() at panic+0x43/frame 0xfffffe016a270af0
sbuf_clear() at sbuf_clear+0xa8/frame 0xfffffe016a270b00
sbuf_cpy() at sbuf_cpy+0x56/frame 0xfffffe016a270b20
spa_taskq_write_param() at spa_taskq_write_param+0x85/frame 0xfffffe016a270bd0 (src/sys/contrib/openzfs/module/zfs/spa.c:1486)
sysctl_root_handler_locked() at sysctl_root_handler_locked+0x9c/frame 0xfffffe016a270c20
sysctl_root() at sysctl_root+0x21e/frame 0xfffffe016a270ca0
userland_sysctl() at userland_sysctl+0x184/frame 0xfffffe016a270d50
sys___sysctl() at sys___sysctl+0x60/frame 0xfffffe016a270e00
amd64_syscall() at amd64_syscall+0x153/frame 0xfffffe016a270f30
fast_syscall_common() at fast_syscall_common+0xf8/frame 0xfffffe016a270f30
--- syscall (202, FreeBSD ELF64, __sysctl), rip = 0x116eb469819a, rsp = 0x116eb273c6e8, rbp = 0x116eb273c720 ---
KDB: enter: panic
[ thread pid 3697 tid 100308 ]
Stopped at      kdb_enter+0x33: movq    $0,0xe38ff2(%rip)
db> x/s version
FreeBSD 15.0-CURRENT #0 main-n267275-4b817fc1f02c: Thu Dec 28 07:42:48 CET 2023
pho@mercat1.netperf.freebsd.org:/usr/src/sys/amd64/compile/PHO
db> 

- Peter