# DF-4260 decisive first run (guest, root). The session was later cut by the
# harness timeout because `timeout 15 sysctl ...` itself never returns: its
# child is stuck un-killably in the kernel (state D3, wchan "costp").
kern.poc_deadlock: 0
--- mode 2 (fixed pattern): expect clean return ---
0 -> 0
poc_deadlock: fixed pattern completed, state=2          [dmesg]
--- mode 1 (bugged vr pattern): expect HANG ---
0                                                      [old value printed by sysctl; the write pass then never returns]
poc_deadlock: mode 1, tick INPROG on cpu4              [dmesg; no completion line ever follows]
# post-hang evidence was collected from a second ssh session (see run.2.log):
#   1705  D3  costp  sysctl -n kern.poc_deadlock=1      <- kern_timeout.c:920 ssleep
#   1731  D5  oidlk  sysctl -n kern.poc_deadlock=2      <- OID lock held by stuck writer
#   per-CPU witness: cpu4 rc=124 elapsed=8s (timers dead); cpu0-3,5 rc=0 3s