# DF-3067 run (guest, stock INVARIANTS kernel #0): run3067.sh crafts an
# all-PAD 256KB undo ring (vol_version=6, first=next=UNDO|0x1000,
# alloc=UNDO|0x40000) and mounts it in the background with
# vfs.hammer.debug_general=0x80.  The mount never returned; the ssh session
# running the script hung (background mount never exits, no cancellation
# points in the kernel loop) and was killed locally after 150s.
#
# Evidence that the walk is live and cyclic (boot.log = serial console):
#   - 47,039 'hammer_recover_scan_rev: rev scan_offset ...' lines and growing
#   - first traced offset = 3000000000001000 == crafted first_offset
#   - offset wraps UP 3000000000000000 -> 300000000003fff8 at iterations 513
#     and 33281 (hammer_recover.c:789-790 ring wrap at alloc_offset)
#   - 32,769 distinct offsets = the entire ring revisited (lap >= 2 at
#     capture time), still descending when captured
#   - guest otherwise alive: fresh ssh session answered 'date; echo PROBE_OK'
#   - the mount process cannot be reaped: kill -9 pends forever (kernel
#     spin); vn0 stays configured
# See bootlog_walk_trace.log for the raw trace (first 2.7MB of console).