Commit Graph
21873 Commits
Author SHA1 Message Date
J. Nick Koston 7516e418f2 Merge branch 'ms8607_remove_set_retry' into integration 2026-02-07 17:23:25 -06:00
J. Nick Koston 3864f06a15 Merge branch 'esp32_hosted_remove_set_retry' into integration 2026-02-07 17:23:21 -06:00
J. Nick Koston 98dcea6e7d Merge branch 'speaker_media_player_remove_set_retry' into integration 2026-02-07 17:23:15 -06:00
J. Nick Koston 9ee51b06fa Merge branch 'deprecate_set_retry' into integration 2026-02-07 17:23:12 -06:00
J. Nick Koston 4efca40457 Merge branch 'lps22_remove_set_retry' into integration 2026-02-07 17:23:06 -06:00
J. Nick Koston f64f71b9ac Merge remote-tracking branch 'upstream/dev' into integration 2026-02-07 15:13:45 -06:00
J. Nick Koston 60298f67b8 [ms8607] Replace set_retry with set_timeout chain to avoid heap allocation
set_retry internally does a std::make_shared<RetryArgs>() heap allocation
on every invocation. Replace with a try_reset_() method that chains
set_timeout calls with manual backoff, preserving the same timing
(immediate, +5ms, +25ms).
2026-02-07 14:53:18 -06:00
J. Nick Koston 4d2354da2e [esp32_hosted] Replace set_retry with set_interval to avoid heap allocation
set_retry internally does a std::make_shared<RetryArgs>() heap allocation
on every invocation. Replace with set_interval + countdown counter which
avoids this entirely. The original code used fixed-interval polling
(no backoff), making set_interval a direct fit.
2026-02-07 14:51:45 -06:00
J. Nick Koston 6a3da67a1e [speaker] Replace set_retry with set_interval to avoid heap allocation
set_retry internally does a std::make_shared<RetryArgs>() heap allocation
on every invocation. Replace with set_interval + countdown counter which
avoids this entirely. All 3 call sites used fixed-interval polling
(no backoff), making set_interval a direct fit.
2026-02-07 14:48:34 -06:00
J. Nick Koston 6ebafa8a9e [core] Deprecate set_retry, cancel_retry, and RetryResult
set_retry does a std::make_shared<RetryArgs>() heap allocation on every
invocation. No core component needs this pattern - all callers have been
migrated to set_timeout or set_interval in prior PRs. The feature wastes
flash and RAM on every firmware for a pattern that set_interval covers
better, and the hidden heap allocation is a footgun for component authors.

Deprecated in 2026.2.0, removal in 2026.8.0.

Depends on:
- #13841 [lps22] Replace set_retry with set_interval
- #13842 [ms8607] Replace set_retry with set_timeout chain
- #13843 [speaker] Replace set_retry with set_interval
- #13844 [esp32_hosted] Replace set_retry with set_interval
2026-02-07 14:44:09 -06:00
J. Nick Koston 3ba7e48615 [lps22] Replace set_retry with set_interval to avoid heap allocation
set_retry internally does a std::make_shared<RetryArgs>() heap allocation
on every invocation. Replace with set_interval + countdown counter which
avoids this entirely.
2026-02-07 14:27:10 -06:00
schrob 9de91539e6 [epaper_spi] Add Waveshare 1.54-G (#13758) 2026-02-08 06:24:57 +11:00
J. Nick Koston 51b0661d9d Merge branch 'scheduler-inplace-cleanup' into integration 2026-02-07 19:56:42 +01:00
J. Nick Koston 3c85ff4744 try to avoid bloat 2026-02-07 19:56:20 +01:00
J. Nick Koston 6a383a62b8 Merge branch 'scheduler-inplace-cleanup' into integration 2026-02-07 19:52:42 +01:00
J. Nick Koston 0fa7050b1c remove temp test 2026-02-07 10:01:57 +01:00
J. Nick Koston fa1554cac0 [scheduler] Eliminate heap allocation in full_cleanup_removed_items_
Replace the temporary std::vector copy with in-place compaction using a
read/write pointer pattern. This avoids a heap allocation+deallocation
cycle during scheduler cleanup, reducing heap fragmentation on
long-running ESP devices.

The new approach compacts valid items forward in the existing vector,
recycles removed items as they are encountered, then resizes the vector
(no reallocation since size only shrinks). Same O(n) complexity, same
behavior, zero allocations.
2026-02-07 09:54:43 +01:00
J. Nick Koston 14071086bb Merge branch 'logger_thread_name_cleanup' into integration 2026-02-07 09:02:06 +01:00
J. Nick Koston 30f9bfaf83 [logger] Resolve thread name once and pass through logging chain
Eliminate redundant xTaskGetCurrentTaskHandle() and pcTaskGetName()
calls on the hot path by resolving the thread name once in log_vprintf_
and passing it through as const char* to all downstream functions.

- Main task fast path passes nullptr (no task handle lookup needed)
- Non-main thread path resolves name once, passes to both ring buffer
  and emergency console fallback
- Unify log_vprintf_non_main_thread_ to single signature across platforms
- Change send_message_thread_safe() on all platforms from TaskHandle_t
  to const char* thread_name
- Add TaskHandle_t overload for get_thread_name_ as primary on
  ESP32/LibreTiny, with no-arg convenience wrapper
- Use std::span<char> for Host/Zephyr get_thread_name_ buffer parameter
- Document Zephyr single-task path thread safety limitation
2026-02-07 07:47:00 +01:00
J. Nick Koston daebc2cc39 Merge branch 'dashboard-ws-close-on-exit' into integration 2026-02-07 06:15:23 +01:00
J. Nick Koston 6b089a611c [dashboard] Close WebSocket after process exit to prevent zombie connections
When a subprocess exited, _proc_on_exit sent the exit event but never
closed the server-side WebSocket. This left zombie connections open
until the client eventually disconnected.
2026-02-07 06:14:44 +01:00
J. Nick Koston 8a2c5407d8 Merge branch 'ld2450_batch_read' into integration 2026-02-06 23:40:24 +01:00
J. Nick Koston 52a039585d Merge branch 'ld2410_batch_read' into integration 2026-02-06 23:40:20 +01:00
J. Nick Koston fd6bd7fb67 Merge branch 'ld2412_batch_read' into integration 2026-02-06 23:40:15 +01:00
J. Nick Koston b544cf2ffe [ld2410] Batch UART reads to reduce loop overhead 2026-02-06 23:39:31 +01:00
J. Nick Koston 6d1281301f [ld2412] Batch UART reads to reduce loop overhead
Read all available bytes in batches via read_array() instead of
byte-at-a-time read() calls. Each read() internally chains through
read_byte -> read_array(1) -> check_read_timeout_ -> available(),
resulting in 3 UART calls per byte. Batching reduces this
significantly.
2026-02-06 23:36:01 +01:00
J. Nick Koston 901192cca1 [ld2450] Batch UART reads to reduce loop overhead
Read all available bytes in batches via read_array() instead of
byte-at-a-time read() calls. Each read() internally chains through
read_byte -> read_array(1) -> check_read_timeout_ -> available(),
resulting in 3 UART calls per byte. At 256000 baud with ~235 bytes
per loop iteration, this was ~706 UART operations per loop call.
Batching reduces this to ~12.

Measured 33% reduction in loop time (2348ms -> 1577ms per 60s).
2026-02-06 23:33:21 +01:00
J. Nick Koston 3478c68af7 Merge branch 'cse7766_batch_read' into integration 2026-02-06 23:12:14 +01:00
J. Nick Koston 67e7ba4812 handle unlikely 2026-02-06 23:12:00 +01:00
J. Nick Koston 981c132cf4 Merge branch 'cse7766_batch_read' into integration 2026-02-06 23:07:21 +01:00
J. Nick Koston 572376091e loop 2026-02-06 23:07:02 +01:00
J. Nick Koston 803e73fdec Merge branch 'cse7766_batch_read' into integration 2026-02-06 22:59:59 +01:00
J. Nick Koston e7c9808b87 [cse7766] Batch UART reads to reduce loop overhead 2026-02-06 22:53:31 +01:00
J. Nick Koston 82eb8e3492 Merge branch 'ssd1306-progmem-tables' into integration 2026-02-06 21:39:50 +01:00
J. Nick Koston 21a5c2891e Merge branch 'i2c-arduino-cswtch' into integration 2026-02-06 21:39:46 +01:00
J. Nick Koston 96289775f2 [i2c] Replace switch with if-else to avoid CSWTCH table in RAM
Replace the Wire status-to-ErrorCode switch with if-else to prevent
the compiler from generating a 6-byte lookup table in DRAM on ESP8266.
2026-02-06 21:38:41 +01:00
J. Nick Koston 3e4269d32a Address review: add SSD1306_MODEL_COUNT sentinel and bounds checks
- Add SSD1306_MODEL_COUNT sentinel to enum for compile-time table size validation
- Replace 14 individual static_asserts with table size checks against SSD1306_MODEL_COUNT
- Add bounds checks in get_height_internal()/get_width_internal() to preserve default return 0
2026-02-06 21:37:51 +01:00
J. Nick Koston bd6d43de52 Merge branch 'ssd1306-progmem-tables' into integration 2026-02-06 21:28:11 +01:00
J. Nick Koston 8da986d41a [ssd1306_base] Move switch tables to PROGMEM with lookup tables
Replace three compiler-generated switch tables (CSWTCH) with PROGMEM
lookup tables, saving 84 bytes of DRAM on ESP8266.

- model_str_(): 56B string pointer table → PROGMEM_STRING_TABLE
- get_height_internal(): 14B byte table → PROGMEM struct array
- get_width_internal(): 14B byte table → PROGMEM struct array

Width and height use a single ModelDimensions struct array for
maintainability. All 14 enum values verified with static_assert.
2026-02-06 21:26:10 +01:00
tronikosandJ. Nick Koston eb7aa3420f Add target_temperature to the template water heater (#13661)
Co-authored-by: J. Nick Koston <nick@koston.org>
2026-02-06 21:23:42 +01:00
J. Nick Koston 4fb851954b Merge branch 'debug_progmem' into integration 2026-02-06 21:10:55 +01:00
J. Nick Koston d1399d9145 [debug] Move ESP8266 switch tables to flash with PROGMEM_STRING_TABLE 2026-02-06 21:10:23 +01:00
J. Nick Koston ba40fbc53d Merge remote-tracking branch 'origin/mqtt-custom-topic-no-alloc' into integration 2026-02-06 21:01:53 +01:00
J. Nick Koston c8d4d870e8 [mqtt] Use .c_str() for StringRef payloads to match non-allocating publish overload 2026-02-06 21:00:53 +01:00
J. Nick Koston fcb458e449 Merge branch 'mqtt-discovery-topic-no-alloc' into integration 2026-02-06 20:57:47 +01:00
J. Nick Koston 48a8f753f5 Merge branch 'mqtt-custom-topic-no-alloc' into integration 2026-02-06 20:57:40 +01:00
J. Nick Koston 6293f3110c Merge branch 'mqtt-friendly-name-const-ref' into integration 2026-02-06 20:57:35 +01:00
J. Nick Koston ce34b37e02 Merge branch 'mqtt-on-log-no-heap' into integration 2026-02-06 20:57:29 +01:00
J. Nick Koston b454cac1dc [mqtt] Use stack buffer for discovery topic to avoid heap allocation 2026-02-06 20:48:18 +01:00
J. Nick Koston a9a943b858 [mqtt] Add zero-allocation topic getters to MQTT_COMPONENT_CUSTOM_TOPIC macro
Add _to() variants that write into a stack buffer and return StringRef,
avoiding heap allocation on every publish. Update publish_state() in
climate (11 allocations eliminated), fan (3), cover (2), and valve (1)
to use the new stack-based getters.

The allocating getters are retained for setup-time paths (subscribe,
send_discovery, dump_config) where the allocation is one-time.
2026-02-06 20:41:51 +01:00