== DF-3405 fix-validation runs (patched kernel #1: Sat Sep 12 06:59:23 UTC 2026, both DF-3405+DF-3406 fixes) ==

run A (race2, old harness): 150s, NO PANIC, guest up. iters=3000 create_ok=1 destroy_busy=1 open_ok=25 write_ok=25
  -> harness stalled after an EBUSY (pre-wedge), but 25 opens interleaved with create/destroy: no crash.

run B (race3, improved always-destroy harness, post-wedge): 180s, NO PANIC, guest up.
  iters=3600 create_ok=0 destroy_busy=3600 open_ok=0 (tap0 already wedged by residual devfs issue; see VERDICT)

run C (race4, improved harness, fresh boot, refcnt=0): 150s, NO PANIC, guest up.
  iters=3000 create_ok=1 destroy_ok=0 destroy_busy=3000 destroy_err=0 open_ok=11 write_ok=11

Guest uname: DragonFly dfbsd 6.5-DEVELOPMENT #1: Sat Sep 12 06:59:23 UTC 2026
Total patched-kernel exposure: ~480s of 8-thread open vs SIOCIFCREATE/SIOCIFDESTROY hammering, ZERO panics
(stock kernel crashed within 43-90s on 2/2 completed runs)

Post-fix functional sanity: tap1/tap2 create/open/write/destroy cycle: CYCLE_OK