* Improved heap and PSRAM handling
- Segment `allocateData()` uses more elaborate DRAM checking to reduce fragmentation and allow for larger setups to run on low heap
- Segment data allocation fails if minimum contiguous block size runs low to keep the UI working
- Increased `MAX_SEGMENT_DATA` to account for better segment data handling
- Memory allocation functions try to keep enough DRAM for segment data
- Added constant `PSRAM_THRESHOLD` to improve PSARM usage
- Increase MIN_HEAP_SIZE to reduce risk of breaking UI due to low memory for JSON response
- ESP32 makes use of IRAM (no 8bit access) for pixeluffers, freeing up to 50kB of RAM
- Fix to properly get available heap on all platforms: added function `getFreeHeapSize()`
- Bugfix for effects that divide by SEGLEN: don't run FX in service() if segment is not active
-Syntax fix in AR: calloc() uses (numelements, size) as arguments
* Added new functions for allocation and heap checking
- added `allocate_buffer()` function that can be used to allocate large buffers: takes parameters to set preferred ram location, including 32bit accessible RAM on ESP32. Returns null if heap runs low or switches to PSRAM
- getFreeHeapSize() and getContiguousFreeHeap() helper functions for all platforms to correctly report free useable heap
- updated some constants
- updated segment data allocation to free the data if it is large
- replaced "psramsafe" variable with it's #ifdef: BOARD_HAS_PSRAM and made accomodating changes
- added some compile-time checks to handle invalid env. definitions
- updated all allocation functions and some of the logic behind them
- added use of fast RTC-Memory where available
- increased MIN_HEAP_SIZE for all systems (improved stability in tests)
- updated memory calculation in web-UI to account for required segment buffer
- added UI alerts if buffer allocation fails
- made getUsedSegmentData() non-private (used in buffer alloc function)
- changed MAX_SEGMENT_DATA
- added more detailed memory log to DEBUG output
- added debug output to buffer alloc function
Any repeating crash that prevents a human from logging in and fixing
the config should be treated as a boot loop. Increase the detection
timeout, so anything that's fast enough to preclude a user fix will
trigger the recovery behaviour.
Don't treat consecutive but infrequent crashes as bootloops. The
bootloop recovery actions only make sense when there is no opportunity
for a user to reconfigure their system.
Suggested by @coderabbitai
ESP8266 RTC RAM requires 32-bit accesses, but there's no need to jump
through a bunch of functions for it. Use references to simplify access
and harmonize the implementation with ESP32.
* Fix bootloop if config missing/reset
Can't reset the config if there's nothing to reset!
* ESP8266: Commit ACTIONT_TRACKER
* Use consistent naming for backups and reset cfgs
Use 'rst.cfg.json' instead of 'cfg.json.rst.json' for configs that were
reset.
* Add a little more PSTR to bootloop handling
* added boot loop detection and config backup
* automatic OTA rollback if loading backup does not fix it
* added new file handling functions
* adding verification of json files, added config restore at bootup if broken
* added function to compare contents of two files for future use (currently not used)
- Bugfixes in FX data allocation: realloc was not handled properly.
- Added *intermediate* fix for waitForIt(), see https://github.com/wled/WLED/issues/4779
- Bugfix in 1D->2D expansions: corner-expansion MUST be boundary checked as it blindly writes the max dimension
- removed some realloc(), improving fragmentation on large setups
- increased min heap constant
- ESP32 C3 has no PSRAM, it now uses default alloc functions
- also added missing UI info for "Error 7"
- Photoshop-style segment/layer blending
- return of strip ABL
- remove global LED buffer in favour of segment-local buffer
- new effect blending modes/transitions
- custom palettes moved out of WS2812FX class
- increased limits (matrix size, LED RAM)
- added "rainbow"-mode colorwheel
- replaced palettes with gamma unmodified ones
- move gamma adjustment to last step before sending to LEDs
- Segment & WS2812FX class reorganisation (mutable members, reordered members, protected members)
Both ESP8266 and ESP32 have a hardware random register. This update makes use of that. It is slightly faster than the fastled variants but mostly it is truly random, even when the timing limitations stated in the datasheet are disregarded. Also saves a bit on code size.
- Replaced all random8() and random16() calls with new hw_random() versions
- Not replaced in FX where PRNG is required
- `sin16_t() / cos16_t()` are faster and more accurate than fastled versions
- `sin_approx() / cos_approx()` are float wrappers for `sin16_t() / cos16_t()` and are accurate enough to replace `sinf()/cosf()`
- `atan2()` is used only in octopus to calculate center offset, new approximated version saves flash
- `tan(), atan(), asin(), acos(), floor(), fmod()` are used only for sunrise/sunset calculation, using wled_math version saves flash
- `beatsinx()` replacements are to make use of new `sin16_t()/sin8_t()` functions to reduce flash size
- Extensively tested surnise/sunset calculation: deviation is 1min. max
- Tested some of the relevant FX and found no visual difference: Julia, 2D Drift, Drift Rose, Ghost rider, Rotozoomer, Palette, Arc 1D expansion
- total flash savings: 7.4k
Remove the large stack buffer as we're just going to copy it in to a
heap buffer anyways. Later we can refine the length estimation or use a
rope-style dynamic data structure like DynamicBufferList.
On ESP8266, it isn't permissible to call delay() in system context;
ensure this is legal before waiting.
On ESP32, use an operating system mutex to ensure consistent variable
state in a multicore environment, and manage the wait without needing
to loop.
