dangernoodle-io / breadboard / 27175034216

100%

push

github

fix(bb_ota_push): abort a slow/stalled upload via a transfer deadline (B1-254) (#374)

The receive loop only aborted on N consecutive recv timeouts, but timeout_count
resets on every successful recv -- so a slow trickle (e.g. 8 KB/s) never tripped
it and held the device with mining paused + the WDT extended (300s) for the whole
crawl, looking hung until a manual reboot. Board-agnostic (the push pause hook is
wired on every board).

Add a content-length-derived transfer deadline from a minimum-throughput floor,
checked each loop iteration: if exceeded, esp_ota_abort + 408 + resume_and_exit
(restores the WDT, resumes mining, emits BB_OTA_PHASE_FAIL) -- well before the WDT.

- pure host-testable bb_ota_push_deadline_ms (clamp content*1000/min_bps to [floor,ceil])
- Kconfig BB_OTA_PUSH_MIN_BPS (10 KB/s) + BB_OTA_PUSH_MAX_DURATION_S (240s, < WDT 300s)
- 7 host tests

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

1618 of 1618 branches covered (100.0%)

Branch coverage included in aggregate %.

2709 of 2709 relevant lines covered (100.0%)

976.09 hits per line