dangernoodle-io / breadboard / 27351305459

100%

push

github

feat(bb_health): extract from bb_info + surface device-health faults on /api/events (#387)

closes B1-249 (P0). a CYD main_task stack overflow once aborted silently
deep in init — bb_log is device-only/ephemeral, so a crashing task can't
report itself. surface health faults as retained bb_event topics so web
UI, taipan-cli, and knot see them uniformly.

- extract bb_health: move /api/health + the health-extender registry out
  of bb_info (mis-homed — bb_info is /api/info static identity, health is
  liveness). bb_info slims to /api/info (reserve 2->1); bb_temp re-points
  REQUIRES bb_info -> bb_health; bb_health owns its own extender-fn typedef.
- health.stack: stack high-water monitor (bb_health) -> retained event
  {task,free_bytes,low}; uxTaskGetSystemState gated on trace-facility,
  debounced; FREERTOS_CHECK_STACKOVERFLOW=2 added to smoke sdkconfigs so a
  real overflow is attributable, not a silent abort.
- health.display: retained event {present,panel,reason} at display init
  (bb_display_info).
- diag.boot: retained event {reset_reason,abnormal_reset_count,
  panic_available,rolled_back} at bb_diag init, republishing this boot's
  persisted state; rolled_back = !bb_ota_is_validated().

all retained + auto-attached to /api/events (bb_update_check pattern,
CONFIG_BB_*_AUTO_ATTACH); pure host-tested JSON builders. consumer support
(taipan-cli SSE, web UI fault indicators) is separate, non-blocking.

verified: 1709 host tests pass; coverage 100%; smoke green on esp32,
esp32c3, esp32-p4, tdongle-s3.

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

1665 of 1665 branches covered (100.0%)

Branch coverage included in aggregate %.

17 of 17 new or added lines in 3 files covered. (100.0%)

2772 of 2772 relevant lines covered (100.0%)

1001.08 hits per line