dangernoodle-io / breadboard / 29462080289

96%

push

github

feat: add a per-cause reboot rate-limit budget to bb_system (#891)

wifi_reconn's future safeguard reboot needs the same cooldown+daily-cap
rate-limiting bb_net_health's egress tier-3 already has, so a later PR
(B1-878, PR4 of the WiFi FSM lane) can route BOTH reboot paths through
one shared budget instead of duplicating the rate-limit logic. This PR
adds the budget facility only; it has zero production callers.

- bb_system: new pure decision core (bb_system_reboot_budget_should_allow,
  state_record/encode/decode) plus a portable, bb_storage-backed
  allows_at/record_at pair (backend "nvs" on device, "ram" on host, no
  ESP_PLATFORM gate beyond that pick). The epoch-resolving
  allows/record wrapper is implemented per-platform (espidf: real
  bb_ntp_is_synced()/time(); host: straight-line unsynced) so the synced
  decision logic stays fully host-tested via explicit args instead of an
  untestable host branch.
- bb_system: new Kconfig knobs BB_SYSTEM_REBOOT_BUDGET_MIN_INTERVAL_S
  (1800) / _DAILY_CAP (4), shared by every cause; gains bb_storage in
  PRIV_REQUIRES.
- bb_net_health is untouched -- its tier-3 keeps its own
  should_reboot/reboot_state_record/_encode/_decode machinery and NVS
  state for now; re-pointing it at the new budget (and retiring the old
  machinery) is PR4's job (B1-878), not this one.


Claude-Session: https://claude.ai/code/session_01V5JXJNaYWPwjrcS8nk183m

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

7164 of 7798 branches covered (91.87%)

Branch coverage included in aggregate %.

121 of 121 new or added lines in 2 files covered. (100.0%)

12045 of 12297 relevant lines covered (97.95%)

2382.9 hits per line