dangernoodle-io / breadboard / 29221207311

95%

push

github

feat: add bb_bqueue blocking queue (#841)

- new component: mailbox (capacity==1) / bounded-MPSC (capacity>1) on one
  opaque type, mirroring FreeRTOS's own xQueueCreate(1,...)+xQueueOverwrite
  mailbox idiom (B1-821)
- mode restriction enforced at runtime, not just documented:
  overwrite()/reset() reject capacity>1, send()/dropped() reject capacity==1
- peek() is non-consuming and N-peeker-safe in both modes -- the host
  backend wakes waiters via bb_lock_cond_broadcast() (never signal()),
  protecting TaipanMiner's never-drained, two-concurrent-peeker work_queue
- zero heap: Kconfig-sized static instance pool (BB_BQUEUE_MAX_INSTANCES,
  default 2), per-instance storage bounded by BB_BQUEUE_MAX_CAPACITY /
  BB_BQUEUE_MAX_ITEM_BYTES
- host backend built on bb_lock_t/bb_lock_cond_t using the mandated
  absolute-deadline/remaining-time wait idiom (pure, host-testable helpers);
  espidf backend is a thin xQueue wrap; arduino ships an UNSUPPORTED stub
- promotes bb_lock_cond_ms_to_ticks() to bb_core's public bb_lock.h so
  non-bb_core consumers (this component's espidf backend) can reuse the
  overflow-safe ms->ticks conversion instead of re-deriving it
- 100% line/branch/function host coverage; two-concurrent-peekers test's
  non-vacuity proven by temporarily swapping broadcast() for signal() and
  observing a clean BB_ERR_TIMEOUT failure on the starved peeker


Claude-Session: https://claude.ai/code/session_01WcT1TPAV2UE4VeGuDL7K8p

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

4075 of 4075 branches covered (100.0%)

Branch coverage included in aggregate %.

155 of 155 new or added lines in 2 files covered. (100.0%)

6611 of 6611 relevant lines covered (100.0%)

8051.54 hits per line