[fix] patched kernel:
DragonFly 6.5-DEVELOPMENT #1: Sat Aug  8 21:54:58 UTC 2026
[fix] NFS server up
MOUNT_OK
-rw-r--r--  1 maxx  wheel  10 Aug  8 22:01 /mnt/nfs/f
[fix] steady state:
[trigger] iter 15: fchown rc=0 errno=0 (ok)
[trigger] iter 16: fchown rc=0 errno=0 (ok)
[fix] kill NFS server
nfsd_dead
[fix] wait 90s; on patched kernel the guest must STAY UP
===== trigger_fix.out (tail) =====
[trigger] iter 8: fchown rc=0 errno=0 (ok)
[trigger] iter 9: fchown rc=0 errno=0 (ok)
[trigger] iter 10: fchown rc=0 errno=0 (ok)
[trigger] iter 11: fchown rc=0 errno=0 (ok)
[trigger] iter 12: fchown rc=0 errno=0 (ok)
[trigger] iter 13: fchown rc=0 errno=0 (ok)
[trigger] iter 14: fchown rc=0 errno=0 (ok)
[trigger] iter 15: fchown rc=0 errno=0 (ok)
[trigger] iter 16: fchown rc=0 errno=0 (ok)
[trigger] iter 17: fchown rc=0 errno=0 (ok)
===== guest status (expect up) =====
up
===== serial panic check (expect none) =====
[fix] RESULT: guest still UP after NFS failure -> NO PANIC -> FIX VALIDATED

=== POST-WAIT (patched kernel) ===
RESULT: iter 18 = rc=-1 EINTR (failing VOP_GETATTR, same as baseline); iter 19+
did NOT panic -- the fix's vput(vp) released the exclusive lock on the error
path. Guest stayed UP. Baseline #0 panicked at the same point. FIX VALIDATED.