dangernoodle-io / breadboard / 27249524371

100%

push

github

feat(bb_wdt): consolidate live Task-WDT management into a dedicated component (#377)

introduces components/bb_wdt as the single owner of esp_task_wdt_* across
breadboard and migrates the scattered consumers onto it:

- bb_wdt_set_timeout / extend_begin / extend_end — WDT timeout reconfigure
  (moved out of bb_system; reset-reason classification stays in bb_system)
- bb_wdt_task_subscribe / unsubscribe / feed — per-task add/delete/reset
- bb_wdt_park_wait — portable park primitive: unsubscribes the calling
  (parked) task, waits on a caller predicate, then resubscribes and feeds.
  on a single core a parked task cannot get scheduled to feed the WDT while
  the OTA worker is CPU-bound on cache-disabled flash writes, so it must
  leave the WDT rather than try to feed it
- migrate bb_ota_pull, bb_ota_push, bb_wifi onto bb_wdt and drop the direct
  esp_task_wdt includes

also fixes a pre-existing dangling-WDT bug surfaced during on-device
validation: bb_ota_pull's ota_task_exit re-subscribed the worker task to the
WDT and then immediately vTaskDelete'd it, leaving a freed-TCB entry that the
next WDT check reads as a garbled task name that never feeds — a spurious
task_wdt panic ~60s after a failed pull. the worker now stays unsubscribed as
it exits.

host tests cover park_wait (unsubscribe/wait/resubscribe) and the no-op host
backend; smoke build links bb_ota_pull/push + bb_wifi. validated on-device:
esp32-c3-supermini completed a full in-place pull-OTA with the mining task
parked and no task_wdt.

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

1626 of 1626 branches covered (100.0%)

Branch coverage included in aggregate %.

7 of 7 new or added lines in 1 file covered. (100.0%)

2724 of 2724 relevant lines covered (100.0%)

976.43 hits per line