github actions: fix CI by requesting python verson 3.9

this is the same fix used in Adafruit_Learning_System_Guides

.github/ISSUE_TEMPLATE.md

@@ -0,0 +1,46 @@
+Thank you for opening an issue on an Adafruit Arduino library repository.  To
+improve the speed of resolution please review the following guidelines and
+common troubleshooting steps below before creating the issue:
+
+- **Do not use GitHub issues for troubleshooting projects and issues.**  Instead use
+  the forums at http://forums.adafruit.com to ask questions and troubleshoot why
+  something isn't working as expected.  In many cases the problem is a common issue
+  that you will more quickly receive help from the forum community.  GitHub issues
+  are meant for known defects in the code.  If you don't know if there is a defect
+  in the code then start with troubleshooting on the forum first.
+
+- **If following a tutorial or guide be sure you didn't miss a step.** Carefully
+  check all of the steps and commands to run have been followed.  Consult the
+  forum if you're unsure or have questions about steps in a guide/tutorial.
+
+- **For Arduino projects check these very common issues to ensure they don't apply**:
+
+  - For uploading sketches or communicating with the board make sure you're using
+    a **USB data cable** and **not** a **USB charge-only cable**.  It is sometimes
+    very hard to tell the difference between a data and charge cable!  Try using the
+    cable with other devices or swapping to another cable to confirm it is not
+    the problem.
+
+  - **Be sure you are supplying adequate power to the board.**  Check the specs of
+    your board and plug in an external power supply.  In many cases just
+    plugging a board into your computer is not enough to power it and other
+    peripherals.
+
+  - **Double check all soldering joints and connections.**  Flakey connections
+    cause many mysterious problems.  See the [guide to excellent soldering](https://learn.adafruit.com/adafruit-guide-excellent-soldering/tools) for examples of good solder joints.
+
+  - **Ensure you are using an official Arduino or Adafruit board.** We can't
+    guarantee a clone board will have the same functionality and work as expected
+    with this code and don't support them.
+
+If you're sure this issue is a defect in the code and checked the steps above
+please fill in the following fields to provide enough troubleshooting information.
+You may delete the guideline and text above to just leave the following details:
+
+- Arduino board:  **INSERT ARDUINO BOARD NAME/TYPE HERE**
+
+- Arduino IDE version (found in Arduino -> About Arduino menu):  **INSERT ARDUINO
+  VERSION HERE**
+
+- List the steps to reproduce the problem below (if possible attach a sketch or
+  copy the sketch code in too): **LIST REPRO STEPS BELOW**

.github/PULL_REQUEST_TEMPLATE.md

@@ -0,0 +1,26 @@
+Thank you for creating a pull request to contribute to Adafruit's GitHub code!
+Before you open the request please review the following guidelines and tips to
+help it be more easily integrated:
+
+- **Describe the scope of your change--i.e. what the change does and what parts
+  of the code were modified.**  This will help us understand any risks of integrating
+  the code.
+
+- **Describe any known limitations with your change.**  For example if the change
+  doesn't apply to a supported platform of the library please mention it.
+
+- **Please run any tests or examples that can exercise your modified code.**  We
+  strive to not break users of the code and running tests/examples helps with this
+  process.
+
+Thank you again for contributing!  We will try to test and integrate the change
+as soon as we can, but be aware we have many GitHub repositories to manage and
+can't immediately respond to every request.  There is no need to bump or check in
+on a pull request (it will clutter the discussion of the request).
+
+Also don't be worried if the request is closed or not integrated--sometimes the
+priorities of Adafruit's GitHub code (education, ease of use) might not match the
+priorities of the pull request.  Don't fret, the open source community thrives on
+forks and GitHub makes it easy to keep your changes in a forked repo.
+
+After reviewing the guidelines above you can delete this text from the pull request.

.github/workflows/githubci.yml

@@ -0,0 +1,37 @@
+name: Arduino Library CI
+
+on: [pull_request, push, repository_dispatch]
+
+jobs:
+  build:
+    strategy:
+      fail-fast: false
+      matrix:
+        arduino-platform: ["metro_m0", "metro_m4", "nrf52840", "esp32"]
+
+    runs-on: ubuntu-latest
+    
+    steps:
+    - uses: actions/setup-python@v1
+      with:
+        python-version: '3.8'
+    - uses: actions/checkout@v2
+    - uses: actions/checkout@v2
+      with:
+         repository: adafruit/ci-arduino
+         path: ci
+
+    - name: pre-install
+      run: bash ci/actions_install.sh
+
+    - name: test platforms
+      run: python3 ci/build_platform.py ${{ matrix.arduino-platform }}
+
+    - name: clang
+      run: python3 ci/run-clang-format.py -e "ci/*" -e "bin/*" -r . 
+
+    - name: doxygen
+      env:
+        GH_REPO_TOKEN: ${{ secrets.GH_REPO_TOKEN }}
+        PRETTYNAME : "Adafruit Protomatter"
+      run: bash ci/doxy_gen_and_deploy.sh

.gitignore

@@ -0,0 +1,4 @@
+# Our handy .gitignore for automation ease
+Doxyfile*
+doxygen_sqlite3.db
+html

Adafruit_Protomatter.cpp

@@ -1,93 +0,0 @@
-// Arduino-specific wrapper for the Protomatter C library (provides
-// constructor and so forth, builds on Adafruit_GFX). There should
-// not be any device-specific #ifdefs here. See notes in core.c and
-// arch.h regarding portability.
-
-#include "Adafruit_Protomatter.h" // Also includes core.h & Adafruit_GFX.h
-
-extern Protomatter_core *_PM_protoPtr; // In core.c (via arch.h)
-
-// Overall matrix refresh rate (frames/second) is a function of matrix width
-// and chain length, number of address lines, number of bit planes, CPU speed
-// and whether or not a GPIO toggle register is available. There is no "this
-// will run at X-frames-per-second" constant figure. You typically just have
-// to try it out and perhaps trade off some bit planes for refresh rate until
-// the image looks good and stable. Anything over 100 Hz is usually passable,
-// around 250 Hz is where things firm up. And while this could proceed higher
-// in some situations, the tradeoff is that faster rates use progressively
-// more CPU time (because it's timer interrupt based and not using DMA or
-// special peripherals). So a throttle is set here, an approximate maximum
-// frame rate which the software will attempt to avoid exceeding (but may
-// refresh slower than this, and in many cases will...just need to set an
-// upper limit to avoid excessive CPU load). An incredibly long comment block
-// for a single constant, thank you for coming to my TED talk!
-#define _PM_MAX_REFRESH_HZ 250
-
-// Time (in milliseconds) to pause following any change in address lines
-// (individually or collectively). Some matrices respond slowly there...
-// must pause on change for matrix to catch up. Defined here (rather than
-// arch.h) because it's not architecture-specific.
-#define _PM_ROW_DELAY 8
-
-
-Adafruit_Protomatter::Adafruit_Protomatter(
-  uint16_t bitWidth, uint8_t bitDepth,
-  uint8_t rgbCount, uint8_t *rgbList,
-  uint8_t addrCount, uint8_t *addrList,
-  uint8_t clockPin, uint8_t latchPin, uint8_t oePin,
-  bool doubleBuffer, void *timer) :
-  GFXcanvas16(bitWidth, (2 << min(addrCount, 5)) * min(rgbCount, 5)) {
-    if(bitDepth > 6)  bitDepth = 6;   // GFXcanvas16 color limit (565)
-
-    // Arguments are passed through to the C _PM_init() function which does
-    // some input validation and minor allocation. Return value is ignored
-    // because we can't really do anything about it in a C++ constructor.
-    // The class begin() function checks rgbPins for NULL to determine
-    // whether to proceed or indicate an error.
-    (void)_PM_init(&core, bitWidth, bitDepth, rgbCount, rgbList,
-      addrCount, addrList, clockPin, latchPin, oePin, doubleBuffer, timer);
-}
-
-Adafruit_Protomatter::~Adafruit_Protomatter(void) {
-    _PM_free(&core);
-    _PM_protoPtr = NULL;
-}
-
-ProtomatterStatus Adafruit_Protomatter::begin(void) {
-    _PM_protoPtr = &core;
-    _PM_begin(&core);
-    return PROTOMATTER_OK;
-}
-
-// Transfer data from GFXcanvas16 to the matrix framebuffer's weird
-// internal format. The actual conversion functions referenced below
-// are in core.c, reasoning is explained there.
-void Adafruit_Protomatter::show(void) {
-
-    // Destination address is computed in convert function
-    // (based on active buffer value, if double-buffering),
-    // just need to pass in the canvas buffer address and
-    // width in pixels.
-    if(core.bytesPerElement == 1) {
-        _PM_convert_565_byte(&core, getBuffer(), WIDTH);
-    } else if(core.bytesPerElement == 2) {
-        _PM_convert_565_word(&core, getBuffer(), WIDTH);
-    } else {
-        _PM_convert_565_long(&core, getBuffer(), WIDTH);
-    }
-
-    if(core.doubleBuffer) {
-        core.swapBuffers = 1;
-        // To avoid overwriting data on the matrix, don't return
-        // until the timer ISR has performed the swap at the right time.
-        while(core.swapBuffers);
-    }
-}
-
-// Returns current value of frame counter and resets its value to zero.
-// Two calls to this, timed one second apart (or use math with other
-// intervals), can be used to get a rough frames-per-second value for
-// the matrix (since this is difficult to estimate beforehand).
-uint32_t Adafruit_Protomatter::getFrameCount(void) {
-    return _PM_getFrameCount(_PM_protoPtr);
-}

Adafruit_Protomatter.h

@@ -1,28 +0,0 @@
-// Arduino-specific header, accompanies Adafruit_Protomatter.cpp.
-// There should not be any device-specific #ifdefs here.
-
-#ifndef _ADAFRUIT_PROTOMATTER_H_
-#define _ADAFRUIT_PROTOMATTER_H_
-
-#include "core.h"
-#include <Adafruit_GFX.h>
-
-class Adafruit_Protomatter : public GFXcanvas16 {
-  public:
-    Adafruit_Protomatter(uint16_t bitWidth, uint8_t bitDepth,
-      uint8_t rgbCount, uint8_t *rgbList,
-      uint8_t addrCount, uint8_t *addrList,
-      uint8_t clockPin, uint8_t latchPin, uint8_t oePin,
-      bool doubleBuffer, void *timer=NULL);
-    ~Adafruit_Protomatter(void);
-    ProtomatterStatus begin(void);
-    void              show(void);
-    uint32_t          getFrameCount(void);
-  private:
-    Protomatter_core  core;                         // Underlying C struct
-    void              convert_byte(uint8_t  *dest); // GFXcanvas16-to-matrix
-    void              convert_word(uint16_t *dest); // conversion functions
-    void              convert_long(uint32_t *dest); // for 8/16/32 bit bufs
-};
-
-#endif // _ADAFRUIT_PROTOMATTER_H_

README.md

@@ -1,4 +1,4 @@
-# Adafruit_Protomatter
+# Adafruit_Protomatter [![Build Status](https://github.com/adafruit/Adafruit_Protomatter/workflows/Arduino%20Library%20CI/badge.svg)](https://github.com/adafruit/Adafruit_Protomatter/actions)
 
 "I used protomatter in the Genesis matrix." - David Marcus, Star Trek III
 
@@ -6,8 +6,6 @@ Code for driving HUB75-style RGB LED matrices, targeted at 32-bit MCUs
 using brute-force GPIO (that is, not relying on DMA or other specialized
 peripherals beyond a timer interrupt, goal being portability).
 
-Name might change as it's nondescriptive and tedious to type in code.
-
 # Matrix Concepts and Jargon
 
 HUB75 RGB LED matrices are basically a set of six concurrent shift register
@@ -41,14 +39,16 @@ help keep up-to-date with any future changes here!
 
 The common ground for architectures to support this library:
 
-* 32-bit device (e.g. ARM core, but potentially ESP32 and others in future)
+* 32-bit device (e.g. ARM core, ESP32 and others)
 * One or more 32-bit GPIO PORTs with atomic (single-cycle, not
   read-modify-write) bitmask SET and CLEAR registers. A bitmask TOGGLE
   register, if present, may improve performance but is NOT required.
-* Tolerate 8-bit or word-aligned 16-bit accesses within the 32-bit PORT
-  registers (e.g. writing just one of four bytes, rather than the whole
-  32 bits). The library does not use any unaligned accesses (i.e. the
-  "middle word" of a 32-bit register), even if a device tolerates such.
+* There may be performance or storage benefits if the architecture tolerates
+  8-bit or word-aligned 16-bit accesses within the 32-bit PORT registers
+  (e.g. writing just one of four bytes, rather than the whole 32 bits), but
+  this is NOT a hardware requirement. Also, the library does not use any
+  unaligned accesses (i.e. "middle word" of a 32-bit register), even if a
+  device tolerates such.
 
 # Software Components
 
@@ -58,14 +58,15 @@ This repository currently consists of:
   Adafruit_Protomatter.h, plus the "examples" directory). The Arduino code
   is dependent on the Adafruit_GFX library.
 
-* An underlying C library (files core.c, core.h and arch.h) that might be
-  adaptable to other runtime environments (e.g. CircuitPython).
+* An underlying C library (files core.c, core.h and headers in the arch
+  directory) that might be adaptable to other runtime environments (e.g.
+  CircuitPython).
 
 # Arduino Library
 
-This *might* supersede the RGBmatrixPanel library on non-AVR devices, as the
-older library has painted itself into a few corners. The newer library uses
-a single constructor for all matrix setups, potentially handling parallel
+This will likely supersede the RGBmatrixPanel library on non-AVR devices, as
+the older library has painted itself into a few corners. The newer library
+uses a single constructor for all matrix setups, potentially handling parallel
 chains (not yet fully implemented), various matrix sizes and chain lengths,
 and variable bit depths from 1 to 6 (refresh rate is a function of all of
 these). Note however that it is NOT A DROP-IN REPLACEMENT for RGBmatrixPanel.
@@ -93,20 +94,22 @@ drawing functions.
 
 The C code has the same limitations as the Arduino library: all RGB data
 pins and the clock pin MUST be on the same PORT register, and it's most
-memory efficient (though still a bit gluttonous) if those pins are all
+memory efficient (though still slightly gluttonous) if those pins are all
 within the same 8-bit byte within the PORT (they do not need to be
 contiguous or sequential within that byte). Other pins (matrix address lines,
 latch and output enable) can reside on any PORT or bit.
 
-When adapting this code to new devices (e.g. nRF52, ESP32) or new runtime
-environments (e.g. CircuitPython), goal is to put all the device- or
-platform-specific code into the arch.h file (or completely separate source
-files, as in the Arduino library .cpp and .h). core.c contains only the
-device-neutral bitbang code and should not have any "#ifdef DEVICE"- or
-"#ifdef ENVIRONMENT"-like lines. Macros for things like getting a PORT
-register address from a pin, or setting up a timer peripheral, all occur
-in arch.h, which is ONLY #included by core.c (to prevent problems like
-multiple instances of ISR functions, which must be singularly declared at
+When adapting this code to new devices (e.g. iMX) or new runtime environments
+(e.g. CircuitPython), goal is to put all the device- or platform-specific
+code into a new header file in the arch directory (or completely separate
+source files, as in the Arduino library .cpp and .h). core.c contains only
+the device-neutral bitbang code and should not have any "#ifdef DEVICE"- or
+"#ifdef ENVIRONMENT"-like lines (exception for the 565 color conversion
+functions, since the internal representation is common to both Arduino and
+CircuitPython). Macros for things like getting a PORT register address from
+a pin, or setting up a timer peripheral, all occur in the arch header files,
+which are ONLY #included by core.c (to prevent problems like multiple
+instances of ISR functions, which must be singularly declared at
 compile-time).
 
 Most macros and functions begin with the prefix **\_PM\_** in order to

arch.h

@@ -1,836 +0,0 @@
-// Establishes some very low-level things specific to each supported device.
-// This should ONLY be included by core.c, nowhere else. Ever.
-
-#if !defined(_PROTOMATTER_ARCH_H_)
-#define _PROTOMATTER_ARCH_H_
-
-/*
-Common ground for architectures to support this library:
-
-- 32-bit device (e.g. ARM core, but potentially ESP32 and others in future)
-- One or more 32-bit GPIO PORTs with atomic bitmask SET and CLEAR registers.
-  A TOGGLE register, if present, may improve performance but is NOT required.
-- Tolerate 8-bit or word-aligned 16-bit accesses within the 32-bit PORT
-  registers (e.g. writing just one of four bytes, rather than the whole
-  32 bits). The library does not use any unaligned accesses (i.e. the
-  "middle word" of a 32-bit register), even if a device tolerates such.
-
-"Pin" as used in this code is always a uint8_t value, but the semantics
-of what it means may vary between Arduino and non-Arduino situations.
-In Arduino, it's the pin index one would pass to functions such as
-digitalWrite(), and doesn't necessarily correspond to physical hardware
-pins or any other arrangement. Some may have names like 'A0' that really
-just map to higher indices.
-In non-Arduino settings (CircuitPython, other languages, etc.), how a
-pin index relates to hardware is entirely implementation dependent, and
-how to get from one to the other is what must be implemented in this file.
-Quite often an environment will follow the Arduino pin designations
-(since the numbers are on a board's silkscreen) and will have an internal
-table mapping those indices to registers and bitmasks...but probably not
-an identically-named and -structured table to the Arduino code, hence the
-reason for many "else" situations in this code.
-
-Each architecture defines the following macros and/or functions (the _PM_
-prefix on each is to reduce likelihood of naming collisions...especially
-on ESP32, which has some similarly-named timer functions...though note
-that this library is NOT CURRENTLY PORTED to ESP32):
-
-GPIO-related macros/functions:
-
-_PM_portOutRegister(pin):    Get address of PORT out register. Code calling
-                             this can cast it to whatever type's needed.
-_PM_portSetRegister(pin):    Get address of PORT set-bits register.
-_PM_portClearRegister(pin):  Get address of PORT clear-bits register.
-_PM_portToggleRegister(pin): Get address of PORT toggle-bits register.
-                             Not all devices support this, in which case
-                             it must be left undefined.
-_PM_portBitMask(pin):        Get bit mask within PORT register corresponding
-                             to a pin number. When compiling for Arduino,
-                             this just maps to digitalPinToBitMask(), other
-                             environments will need an equivalent.
-_PM_byteOffset(pin):         Get index of byte (0 to 3) within 32-bit PORT
-                             corresponding to a pin number.
-_PM_wordOffset(pin):         Get index of word (0 or 1) within 32-bit PORT
-                             corresponding to a pin number.
-_PM_pinOutput(pin):          Set a pin to output mode. In Arduino this maps
-                             to pinMode(pin, OUTPUT). Other environments
-                             will need an equivalent.
-_PM_pinInput(pin):           Set a pin to input mode, no pullup. In Arduino
-                             this maps to pinMode(pin, INPUT).
-_PM_pinHigh(pin):            Set an output pin to a high or 1 state. In
-                             Arduino this maps to digitalWrite(pin, HIGH).
-_PM_pinLow(pin):             Set an output pin to a low or 0 state. In
-                             Arduino this maps to digitalWrite(pin, LOW).
-
-Timer-related macros/functions:
-
-_PM_timerFreq:               A numerical constant - the source clock rate
-                             (in Hz) that's fed to the timer peripheral.
-_PM_timerInit(void*):        Initialize (but do not start) timer.
-_PM_timerStart(void*,count): (Re)start timer for a given timer-tick interval.
-_PM_timerStop(void*):        Stop timer, return current timer counter value.
-_PM_timerGetCount(void*):    Get current timer counter value (whether timer
-                             is running or stopped).
-A timer interrupt service routine is also required, syntax for which varies
-between architectures.
-The void* argument passed to the timer functions is some indeterminate type
-used to uniquely identify a timer peripheral within a given environment. For
-example, in the Arduino wrapper for this library, compiling for SAMD chips,
-it's just a pointer directly to a timer/counter peripheral base address. If
-an implementation needs more data associated alongside a peripheral, this
-could instead be a pointer to a struct, or an integer index.
-
-Other macros/functions:
-
-_PM_chunkSize:               Matrix bitmap width (both in RAM and as issued
-                             to the device) is rounded up (if necessary) to
-                             a multiple of this value as a way of explicitly
-                             unrolling the innermost data-stuffing loops.
-                             So far all HUB75 displays I've encountered are
-                             a multiple of 32 pixels wide, but in case
-                             something new comes along, or if a larger
-                             unroll actually decreases performance due to
-                             cache size, this can be set to whatever works
-                             best (any additional data is simply shifted
-                             out the other end of the matrix). Default if
-                             unspecified is 8 (e.g. four loop passes on a
-                             32-pixel matrix, eight if 64-pixel). Only
-                             certain chunkSizes are actually implemented,
-                             see .cpp code (avoiding GCC-specific tricks
-                             that would handle arbitrary chunk sizes).
-_PM_delayMicroseconds(us):   Function or macro to delay some number of
-                             microseconds. For Arduino, this just maps to
-                             delayMicroseconds(). Other environments will
-                             need to provide their own or map to an
-                             an equivalent function.
-_PM_clockHoldHigh:           Additional code (typically some number of NOPs)
-                             needed to delay the clock fall after RGB data is
-                             written to PORT. Only required on fast devices.
-                             If left undefined, no delay happens.
-_PM_clockHoldLow:            Additional code (e.g. NOPs) needed to delay
-                             clock rise after writing RGB data to PORT.
-                             No delay if left undefined.
-_PM_minMinPeriod:            Mininum value for the "minPeriod" class member,
-                             so bit-angle-modulation time always doubles with
-                             each bitplane (else lower bits may be the same).
-*/
-
-#if defined(ARDUINO)   // If compiling in Arduino IDE...
-  #include <Arduino.h> // pull in all that stuff.
-
-  #define _PM_delayMicroseconds(us) delayMicroseconds(us)
-  #define _PM_pinOutput(pin)        pinMode(pin, OUTPUT)
-  #define _PM_pinInput(pin)         pinMode(pin, INPUT)
-  #define _PM_pinHigh(pin)          digitalWrite(pin, HIGH)
-  #define _PM_pinLow(pin)           digitalWrite(pin, LOW)
-  #define _PM_portBitMask(pin)      digitalPinToBitMask(pin)
-
-// No #else here. In non-Arduino case, declare things in the arch-specific
-// sections below...unless other environments provide device-neutral
-// functions as above, in which case those could go here (w/#elif).
-
-#endif // end defined(ARDUINO)
-
-
-// CODE COMMON TO BOTH SAMD51 AND SAMD21 -----------------------------------
-
-#if defined(__SAMD51__) || defined(_SAMD21_)
-  #if defined(ARDUINO)
-
-    // g_APinDescription[] table and pin indices are Arduino specific:
-    #if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
-      #define _PM_byteOffset(pin) (g_APinDescription[pin].ulPin / 8)
-    #else
-      #define _PM_byteOffset(pin) (3 - (g_APinDescription[pin].ulPin / 8))
-    #endif
-
-    #if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
-      #define _PM_wordOffset(pin) (g_APinDescription[pin].ulPin / 16)
-    #else
-      #define _PM_wordOffset(pin) (1 - (g_APinDescription[pin].ulPin / 16))
-    #endif
-
-    // Arduino implementation is tied to a specific timer/counter & freq:
-    #define _PM_TIMER_DEFAULT TC4
-    #define _PM_IRQ_HANDLER   TC4_Handler
-    #define _PM_timerFreq     48000000
-    // Partly because IRQs must be declared at compile-time, and partly
-    // because we know Arduino's already set up one of the GCLK sources
-    // for 48 MHz.
-
-    // Because it's tied to a specific timer right now, there can be only
-    // one instance of the Protomatter_core struct. The Arduino library
-    // sets up this pointer when calling begin().
-    void *_PM_protoPtr = NULL;
-
-    // Timer interrupt service routine
-    void _PM_IRQ_HANDLER(void) {
-        // Clear overflow flag:
-        _PM_TIMER_DEFAULT->COUNT16.INTFLAG.reg = TC_INTFLAG_OVF;
-        _PM_row_handler(_PM_protoPtr); // In core.c
-    }
-
-  #else
-
-    // Non-arduino byte offset macros, timer and ISR work go here.
-
-  #endif
-
-  // Code below diverges for SAMD51 vs SAMD21, but is still very similar...
-  // If making a change or bug fix in one, check to see if an equivalent
-  // change should be made in the other!
-
-#endif // __SAMD51__ || _SAMD21_
-
-
-// SAMD51-SPECIFIC CODE ----------------------------------------------------
-
-#if defined(__SAMD51__)
-
-  #if defined(ARDUINO)
-
-    // g_APinDescription[] table and pin indices are Arduino specific:
-    #define _PM_portOutRegister(pin) \
-      &PORT->Group[g_APinDescription[pin].ulPort].OUT.reg
-
-    #define _PM_portSetRegister(pin) \
-      &PORT->Group[g_APinDescription[pin].ulPort].OUTSET.reg
-
-    #define _PM_portClearRegister(pin) \
-      &PORT->Group[g_APinDescription[pin].ulPort].OUTCLR.reg
-
-    #define _PM_portToggleRegister(pin) \
-      &PORT->Group[g_APinDescription[pin].ulPort].OUTTGL.reg
-
-  #else
-
-    // Non-Arduino port register lookups go here
-
-  #endif
-
-  // Initialize, but do not start, timer
-  void _PM_timerInit(void *tptr) {
-      static const struct {
-          Tc       *tc;      // -> Timer/counter peripheral base address
-          IRQn_Type IRQn;    // Interrupt number
-          uint8_t   GCLK_ID; // Peripheral channel # for clock source
-      } timer[] = {
-          TC0, TC0_IRQn, TC0_GCLK_ID,
-          TC1, TC1_IRQn, TC1_GCLK_ID,
-          TC2, TC2_IRQn, TC2_GCLK_ID,
-          TC3, TC3_IRQn, TC3_GCLK_ID,
-          TC4, TC4_IRQn, TC4_GCLK_ID,
-          TC5, TC5_IRQn, TC5_GCLK_ID,
-  #if defined(TC6)
-          TC6, TC6_IRQn, TC6_GCLK_ID,
-  #endif
-  #if defined(TC7)
-          TC7, TC7_IRQn, TC7_GCLK_ID,
-  #endif
-  #if defined(TC8)
-          TC8, TC8_IRQn, TC8_GCLK_ID,
-  #endif
-  #if defined(TC9)
-          TC9, TC9_IRQn, TC9_GCLK_ID,
-  #endif
-  #if defined(TC10)
-          TC10, TC10_IRQn, TC10_GCLK_ID,
-  #endif
-  #if defined(TC11)
-          TC11, TC11_IRQn, TC11_GCLK_ID,
-  #endif
-  #if defined(TC12)
-          TC12, TC12_IRQn, TC12_GCLK_ID,
-  #endif
-      };
-      #define NUM_TIMERS (sizeof timer / sizeof timer[0])
-
-      Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
-
-      uint8_t timerNum = 0;
-      while((timerNum < NUM_TIMERS) && (timer[timerNum].tc != tc)) {
-          timerNum++;
-      }
-      if(timerNum >= NUM_TIMERS) return;
-
-      // Feed timer/counter off GCLK1 (already set 48 MHz by Arduino core).
-      // Sure, SAMD51 can run timers up to F_CPU (e.g. 120 MHz or up to
-      // 200 MHz with overclocking), but on higher bitplanes (which have
-      // progressively longer timer periods) I could see this possibly
-      // exceeding a 16-bit timer, and would have to switch prescalers.
-      // We don't actually need atomic precision on the timer -- point is
-      // simply that the period doubles with each bitplane, and this can
-      // work fine at 48 MHz.
-      GCLK->PCHCTRL[timer[timerNum].GCLK_ID].bit.CHEN = 0;    // Disable
-      while(GCLK->PCHCTRL[timer[timerNum].GCLK_ID].bit.CHEN); // Wait for it
-      GCLK_PCHCTRL_Type pchctrl; // Read-modify-store
-      pchctrl.reg      = GCLK->PCHCTRL[timer[timerNum].GCLK_ID].reg;
-      pchctrl.bit.GEN  = GCLK_PCHCTRL_GEN_GCLK1_Val;
-      pchctrl.bit.CHEN = 1;
-      GCLK->PCHCTRL[timer[timerNum].GCLK_ID].reg = pchctrl.reg;
-      while(!GCLK->PCHCTRL[timer[timerNum].GCLK_ID].bit.CHEN);
-
-      // Disable timer before configuring it
-      tc->COUNT16.CTRLA.bit.ENABLE = 0;
-      while(tc->COUNT16.SYNCBUSY.bit.ENABLE);
-
-      // 16-bit counter mode, 1:1 prescale
-      tc->COUNT16.CTRLA.bit.MODE      = TC_CTRLA_MODE_COUNT16;
-      tc->COUNT16.CTRLA.bit.PRESCALER = TC_CTRLA_PRESCALER_DIV1_Val;
-
-      tc->COUNT16.WAVE.bit.WAVEGEN =
-        TC_WAVE_WAVEGEN_MFRQ_Val; // Match frequency generation mode (MFRQ)
-
-      tc->COUNT16.CTRLBCLR.reg = TC_CTRLBCLR_DIR; // Count up
-      while(tc->COUNT16.SYNCBUSY.bit.CTRLB);
-
-      // Overflow interrupt
-      tc->COUNT16.INTENSET.reg = TC_INTENSET_OVF;
-
-      NVIC_DisableIRQ(timer[timerNum].IRQn);
-      NVIC_ClearPendingIRQ(timer[timerNum].IRQn);
-      NVIC_SetPriority(timer[timerNum].IRQn, 0); // Top priority
-      NVIC_EnableIRQ(timer[timerNum].IRQn);
-
-      // Timer is configured but NOT enabled by default
-  }
-
-  // Set timer period, initialize count value to zero, enable timer.
-  // Timer must be initialized to 16-bit mode using the init function
-  // above, but must be inactive before calling this.
-  inline void _PM_timerStart(void *tptr, uint32_t period) {
-      Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
-      tc->COUNT16.COUNT.reg = 0;
-      while(tc->COUNT16.SYNCBUSY.bit.COUNT);
-      tc->COUNT16.CC[0].reg = period;
-      while(tc->COUNT16.SYNCBUSY.bit.CC0);
-      tc->COUNT16.CTRLA.bit.ENABLE = 1;
-      while(tc->COUNT16.SYNCBUSY.bit.STATUS);
-  }
-
-  // Return current count value (timer enabled or not).
-  // Timer must be previously initialized.
-  inline uint32_t _PM_timerGetCount(void *tptr) {
-      Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
-      tc->COUNT16.CTRLBSET.bit.CMD = 0x4;  // Sync COUNT
-      while(tc->COUNT16.CTRLBSET.bit.CMD); // Wait for command
-      return tc->COUNT16.COUNT.reg;
-  }
-
-  // Disable timer and return current count value.
-  // Timer must be previously initialized.
-  uint32_t _PM_timerStop(void *tptr) {
-      Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
-      uint32_t count = _PM_timerGetCount(tptr);
-      tc->COUNT16.CTRLA.bit.ENABLE = 0;
-      while(tc->COUNT16.SYNCBUSY.bit.STATUS);
-      return count;
-  }
-
-  // See notes in core.c before the "blast" functions
-  #if F_CPU >= 200000000
-    #define _PM_clockHoldHigh asm("nop; nop; nop; nop; nop");
-    #define _PM_clockHoldLow  asm("nop; nop");
-  #elif F_CPU >= 180000000
-    #define _PM_clockHoldHigh asm("nop; nop; nop; nop");
-    #define _PM_clockHoldLow  asm("nop");
-  #elif F_CPU >= 150000000
-    #define _PM_clockHoldHigh asm("nop; nop; nop");
-    #define _PM_clockHoldLow  asm("nop");
-  #else
-    #define _PM_clockHoldHigh asm("nop; nop; nop");
-    #define _PM_clockHoldLow  asm("nop");
-  #endif
-
-  #define _PM_minMinPeriod 160
-
-#endif // end __SAMD51__
-
-
-// SAMD21-SPECIFIC CODE ----------------------------------------------------
-
-#if defined(_SAMD21_)
-
-  #if defined(ARDUINO)
-
-    // g_APinDescription[] table and pin indices are Arduino specific:
-    #define _PM_portOutRegister(pin) \
-      &PORT_IOBUS->Group[g_APinDescription[pin].ulPort].OUT.reg
-
-    #define _PM_portSetRegister(pin) \
-      &PORT_IOBUS->Group[g_APinDescription[pin].ulPort].OUTSET.reg
-
-    #define _PM_portClearRegister(pin) \
-      &PORT_IOBUS->Group[g_APinDescription[pin].ulPort].OUTCLR.reg
-
-    #define _PM_portToggleRegister(pin) \
-      &PORT_IOBUS->Group[g_APinDescription[pin].ulPort].OUTTGL.reg
-
-  #else
-
-    // Non-Arduino port register lookups go here
-
-  #endif
-
-
-    // Initialize, but do not start, timer
-    void _PM_timerInit(void *tptr) {
-        static const struct {
-            Tc       *tc;     // -> Timer/counter peripheral base address
-            IRQn_Type IRQn;   // Interrupt number
-            uint8_t   GCM_ID; // GCLK selection ID
-        } timer[] = {
-            TC0, TC0_IRQn, GCM_TCC0_TCC1,
-            TC1, TC1_IRQn, GCM_TCC0_TCC1,
-#if defined(TC2)
-            TC2, TC2_IRQn, GCM_TCC2_TC3,
-#endif
-#if defined(TC3)
-            TC3, TC3_IRQn, GCM_TCC2_TC3,
-#endif
-#if defined(TC4)
-            TC4, TC4_IRQn, GCM_TC4_TC5,
-#endif
-        };
-        #define NUM_TIMERS (sizeof timer / sizeof timer[0])
-
-        Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
-
-        uint8_t timerNum = 0;
-        while((timerNum < NUM_TIMERS) && (timer[timerNum].tc != tc)) {
-            timerNum++;
-        }
-        if(timerNum >= NUM_TIMERS) return;
-
-        // Enable GCLK for timer/counter
-        GCLK->CLKCTRL.reg = (uint16_t)(GCLK_CLKCTRL_CLKEN |
-          GCLK_CLKCTRL_GEN_GCLK0 | GCLK_CLKCTRL_ID(timer[timerNum].GCM_ID));
-        while(GCLK->STATUS.bit.SYNCBUSY == 1);
-
-        // Counter must first be disabled to configure it
-        tc->COUNT16.CTRLA.bit.ENABLE = 0;
-        while(tc->COUNT16.STATUS.bit.SYNCBUSY);
-
-        tc->COUNT16.CTRLA.reg =     // Configure timer counter
-          TC_CTRLA_PRESCALER_DIV1 | // 1:1 Prescale
-          TC_CTRLA_WAVEGEN_MFRQ   | // Match frequency generation mode (MFRQ)
-          TC_CTRLA_MODE_COUNT16;    // 16-bit counter mode
-        while(tc->COUNT16.STATUS.bit.SYNCBUSY);
-
-        tc->COUNT16.CTRLBCLR.reg = TCC_CTRLBCLR_DIR; // Count up
-        while(tc->COUNT16.STATUS.bit.SYNCBUSY);
-
-        // Overflow interrupt
-        tc->COUNT16.INTENSET.reg = TC_INTENSET_OVF;
-
-        NVIC_DisableIRQ(timer[timerNum].IRQn);
-        NVIC_ClearPendingIRQ(timer[timerNum].IRQn);
-        NVIC_SetPriority(timer[timerNum].IRQn, 0); // Top priority
-        NVIC_EnableIRQ(timer[timerNum].IRQn);
-
-        // Timer is configured but NOT enabled by default
-    }
-
-    // Set timer period, initialize count value to zero, enable timer.
-    // Timer must be initialized to 16-bit mode using the init function
-    // above, but must be inactive before calling this.
-    inline void _PM_timerStart(void *tptr, uint32_t period) {
-        Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
-        tc->COUNT16.COUNT.reg = 0;
-        while(tc->COUNT16.STATUS.bit.SYNCBUSY);
-        tc->COUNT16.CC[0].reg = period;
-        while(tc->COUNT16.STATUS.bit.SYNCBUSY);
-        tc->COUNT16.CTRLA.bit.ENABLE = 1;
-        while(tc->COUNT16.STATUS.bit.SYNCBUSY);
-    }
-
-    // Return current count value (timer enabled or not).
-    // Timer must be previously initialized.
-    inline uint32_t _PM_timerGetCount(void *tptr) {
-        Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
-        tc->COUNT16.READREQ.reg = TC_READREQ_RCONT | TC_READREQ_ADDR(0x10);
-        while(tc->COUNT16.STATUS.bit.SYNCBUSY);
-        return tc->COUNT16.COUNT.reg;
-    }
-
-    // Disable timer and return current count value.
-    // Timer must be previously initialized.
-    inline uint32_t _PM_timerStop(void *tptr) {
-        Tc      *tc    = (Tc *)tptr; // Cast peripheral address passed in
-        uint32_t count = _PM_timerGetCount(tptr);
-        tc->COUNT16.CTRLA.bit.ENABLE = 0;
-        while(tc->COUNT16.STATUS.bit.SYNCBUSY);
-        return count;
-    }
-
-#endif // _SAMD21_
-
-
-// NRF52-SPECIFIC CODE -----------------------------------------------------
-
-#if defined(NRF52_SERIES)
-#endif // NRF52_SERIES
-
-
-// ESP32-SPECIFIC CODE -----------------------------------------------------
-
-#if defined(ARDUINO_ARCH_ESP32)
-#endif // ARDUINO_ARCH_ESP32
-
-
-// DEFAULTS IF NOT DEFINED ABOVE -------------------------------------------
-
-#if !defined(_PM_chunkSize)
-  #define _PM_chunkSize 8
-#endif
-
-#if !defined(_PM_clockHoldHigh)
-  #define _PM_clockHoldHigh
-#endif
-
-#if !defined(_PM_clockHoldLow)
-  #define _PM_clockHoldLow
-#endif
-
-#if !defined(_PM_minMinPeriod)
-  #define _PM_minMinPeriod 100
-#endif
-
-// ARDUINO SPECIFIC CODE ---------------------------------------------------
-
-#if defined(ARDUINO)
-
-// 16-bit (565) color conversion functions go here (rather than in the
-// Arduino lib .cpp) because knowledge is required of chunksize and the
-// toggle register (or lack thereof), which are only known to this file,
-// not the .cpp or anywhere else
-// However...this file knows nothing of the GFXcanvas16 type (from
-// Adafruit_GFX...another C++ lib), so the .cpp just passes down some
-// pointers and minimal info about the canvas buffer.
-// It's probably not ideal but this is my life now, oh well.
-
-// Different runtime environments (which might not use the 565 canvas
-// format) will need their own conversion functions.
-
-// There are THREE COPIES of the following function -- one each for byte,
-// word and long. If changes are made in any one of them, the others MUST
-// be updated to match! Note that they are not simple duplicates of each
-// other. The byte case, for example, doesn't need to handle parallel
-// matrix chains (matrix data can only be byte-sized if one chain).
-
-// width argument comes from GFX canvas width, which may be less than
-// core's bitWidth (due to padding). height isn't needed, it can be
-// inferred from core->numRowPairs.
-void _PM_convert_565_byte(Protomatter_core *core, uint16_t *source,
-  uint16_t width) {
-    uint16_t *upperSrc = source;                             // Canvas top half
-    uint16_t *lowerSrc = source + width * core->numRowPairs; // " bottom half
-    uint8_t  *pinMask  = (uint8_t *)core->rgbMask;           // Pin bitmasks
-    uint8_t  *dest     = (uint8_t *)core->screenData;
-    if(core->doubleBuffer) {
-        dest += core->bufferSize * (1 - core->activeBuffer);
-    }
-
-    // No need to clear matrix buffer, loops below do a full overwrite
-    // (except for any scanline pad, which was already initialized in the
-    // begin() function and won't be touched here).
-
-    // Determine matrix bytes per bitplane & row (row pair really):
-
-    uint32_t bitplaneSize = _PM_chunkSize *
-      ((width + (_PM_chunkSize - 1)) / _PM_chunkSize); // 1 plane of row pair
-    uint8_t  pad          = bitplaneSize - width;      // Start-of-plane pad
-
-    // Skip initial scanline padding if present (HUB75 matrices shift data
-    // in from right-to-left, so if we need scanline padding it occurs at
-    // the start of a line, rather than the usual end). Destination pointer
-    // passed in already handles double-buffer math, so we don't need to
-    // handle that here, just the pad...
-    dest += pad;
-
-    uint32_t initialRedBit, initialGreenBit, initialBlueBit;
-    if(core->numPlanes == 6) {
-        // If numPlanes is 6, red and blue are expanded from 5 to 6 bits.
-        // This involves duplicating the MSB of the 5-bit value to the LSB
-        // of its corresponding 6-bit value...or in this case, bitmasks for
-        // red and blue are initially assigned to canvas MSBs, while green
-        // starts at LSB (because it's already 6-bit). Inner loop below then
-        // wraps red & blue after the first bitplane.
-        initialRedBit   = 0b1000000000000000; // MSB red
-        initialGreenBit = 0b0000000000100000; // LSB green
-        initialBlueBit  = 0b0000000000010000; // MSB blue
-    } else {
-        // If numPlanes is 1 to 5, no expansion is needed, and one or all
-        // three color components might be decimated by some number of bits.
-        // The initial bitmasks are set to the components' numPlanesth bit
-        // (e.g. for 5 planes, start at red & blue bit #0, green bit #1,
-        // for 4 planes, everything starts at the next bit up, etc.).
-        uint8_t shiftLeft = 5 - core->numPlanes;
-        initialRedBit   = 0b0000100000000000 << shiftLeft;
-        initialGreenBit = 0b0000000001000000 << shiftLeft;
-        initialBlueBit  = 0b0000000000000001 << shiftLeft;
-    }
-
-    // This works sequentially-ish through the destination buffer,
-    // reading from the canvas source pixels in repeated passes,
-    // beginning from the least bit.
-    for(uint8_t row=0; row<core->numRowPairs; row++) {
-        uint32_t redBit   = initialRedBit;
-        uint32_t greenBit = initialGreenBit;
-        uint32_t blueBit  = initialBlueBit;
-        for(uint8_t plane=0; plane<core->numPlanes; plane++) {
-#if defined(_PM_portToggleRegister)
-            uint8_t prior = core->clockMask; // Set clock bit on 1st out
-#endif
-            for(uint16_t x=0; x<width; x++) {
-                uint16_t upperRGB = upperSrc[x]; // Pixel in upper half
-                uint16_t lowerRGB = lowerSrc[x]; // Pixel in lower half
-                uint8_t  result   = 0;
-                if(upperRGB & redBit)   result |= pinMask[0];
-                if(upperRGB & greenBit) result |= pinMask[1];
-                if(upperRGB & blueBit)  result |= pinMask[2];
-                if(lowerRGB & redBit)   result |= pinMask[3];
-                if(lowerRGB & greenBit) result |= pinMask[4];
-                if(lowerRGB & blueBit)  result |= pinMask[5];
-#if defined(_PM_portToggleRegister)
-                dest[x] = result ^ prior;
-                prior   = result | core->clockMask; // Set clock bit on next out
-#else
-                dest[x] = result;
-#endif
-            } // end x
-            greenBit <<= 1;
-            if(plane || (core->numPlanes < 6)) {
-                // In most cases red & blue bit scoot 1 left...
-                redBit  <<= 1;
-                blueBit <<= 1;
-            } else {
-                // Exception being after bit 0 with 6-plane display,
-                // in which case they're reset to red & blue LSBs
-                // (so 5-bit colors are expanded to 6 bits).
-                redBit  = 0b0000100000000000;
-                blueBit = 0b0000000000000001;
-            }
-#if defined(_PM_portToggleRegister)
-            // If using bit-toggle register, erase the toggle bit on the
-            // first element of each bitplane & row pair. The matrix-driving
-            // interrupt functions correspondingly set the clock low before
-            // finishing. This is all done for legibility on oscilloscope --
-            // so idle clock appears LOW -- but really the matrix samples on
-            // a rising edge and we could leave it high, but at this stage
-            // in development just want the scope "readable."
-            dest[-pad] &= ~core->clockMask; // Negative index is legal & intentional
-#endif
-            dest += bitplaneSize; // Advance one scanline in dest buffer
-        } // end plane
-        upperSrc += width; // Advance one scanline in source buffer
-        lowerSrc += width;
-    } // end row
-}
-
-// Corresponding function for word output -- either 12 RGB bits (2 parallel
-// matrix chains), or 1 chain with RGB bits not in the same byte (but in the
-// same 16-bit word). Some of the comments have been stripped out since it's
-// largely the same operation, but changes are noted.
-void _PM_convert_565_word(Protomatter_core *core, uint16_t *source,
-  uint16_t width) {
-    uint16_t *upperSrc = source;                             // Matrix top half
-    uint16_t *lowerSrc = source + width * core->numRowPairs; // " bottom half
-    uint16_t *pinMask  = (uint16_t *)core->rgbMask;          // Pin bitmasks
-    uint16_t *dest     = (uint16_t *)core->screenData;
-    if(core->doubleBuffer) {
-        dest += core->bufferSize / core->bytesPerElement *
-          (1 - core->activeBuffer);
-    }
-
-    uint32_t bitplaneSize = _PM_chunkSize *
-      ((width + (_PM_chunkSize - 1)) / _PM_chunkSize); // 1 plane of row pair
-    uint8_t  pad          = bitplaneSize - width;      // Start-of-plane pad
-
-    uint32_t initialRedBit, initialGreenBit, initialBlueBit;
-    if(core->numPlanes == 6) {
-        initialRedBit   = 0b1000000000000000; // MSB red
-        initialGreenBit = 0b0000000000100000; // LSB green
-        initialBlueBit  = 0b0000000000010000; // MSB blue
-    } else {
-        uint8_t shiftLeft = 5 - core->numPlanes;
-        initialRedBit   = 0b0000100000000000 << shiftLeft;
-        initialGreenBit = 0b0000000001000000 << shiftLeft;
-        initialBlueBit  = 0b0000000000000001 << shiftLeft;
-    }
-
-    // Unlike the 565 byte converter, the word converter DOES clear out the
-    // matrix buffer (because each chain is OR'd into place). If a toggle
-    // register exists, "clear" really means the clock mask is set in all
-    // but the first element on a scanline (per bitplane). If no toggle
-    // register, can just zero everything out.
-#if defined(_PM_portToggleRegister)
-    // No per-chain loop is required; one clock bit handles all chains
-    uint32_t offset = 0; // Current position in the 'dest' buffer
-    for(uint8_t row=0; row<core->numRowPairs; row++) {
-        for(uint8_t plane=0; plane<core->numPlanes; plane++) {
-            dest[offset++] = 0; // First element of each plane
-            for(uint16_t x=1; x<bitplaneSize; x++) { // All subsequent items
-                dest[offset++] = core->clockMask;
-            }
-        }
-    }
-#else
-    memset(dest, 0, core->bufferSize);
-#endif
-
-    dest += pad; // Pad value is in 'elements,' not bytes, so this is OK
-
-    // After a set of rows+bitplanes are processed, upperSrc and lowerSrc
-    // have advanced halfway down one matrix. This offset is used after
-    // each chain to advance them to the start/middle of the next matrix.
-    uint32_t halfMatrixOffset = width * core->numPlanes * core->numRowPairs;
-
-    for(uint8_t chain=0; chain<core->parallel; chain++) {
-        for(uint8_t row=0; row<core->numRowPairs; row++) {
-            uint32_t redBit   = initialRedBit;
-            uint32_t greenBit = initialGreenBit;
-            uint32_t blueBit  = initialBlueBit;
-            for(uint8_t plane=0; plane<core->numPlanes; plane++) {
-#if defined(_PM_portToggleRegister)
-                // Since we're ORing in bits over an existing clock bit,
-                // prior is 0 rather than clockMask as in the byte case.
-                uint16_t prior = 0;
-#endif
-                for(uint16_t x=0; x<width; x++) {
-                    uint16_t upperRGB = upperSrc[x]; // Pixel in upper half
-                    uint16_t lowerRGB = lowerSrc[x]; // Pixel in lower half
-                    uint16_t result   = 0;
-                    if(upperRGB & redBit)   result |= pinMask[0];
-                    if(upperRGB & greenBit) result |= pinMask[1];
-                    if(upperRGB & blueBit)  result |= pinMask[2];
-                    if(lowerRGB & redBit)   result |= pinMask[3];
-                    if(lowerRGB & greenBit) result |= pinMask[4];
-                    if(lowerRGB & blueBit)  result |= pinMask[5];
-                    // Main difference here vs byte converter is each chain
-                    // ORs new bits into place (vs single-pass overwrite).
-#if defined(_PM_portToggleRegister)
-                    dest[x] |= result ^ prior; // Bitwise OR
-                    prior    = result;
-#else
-                    dest[x] |= result;         // Bitwise OR
-#endif
-                } // end x
-                greenBit <<= 1;
-                if(plane || (core->numPlanes < 6)) {
-                    redBit  <<= 1;
-                    blueBit <<= 1;
-                } else {
-                    redBit  = 0b0000100000000000;
-                    blueBit = 0b0000000000000001;
-                }
-                dest += bitplaneSize; // Advance one scanline in dest buffer
-            } // end plane
-            upperSrc += width; // Advance one scanline in source buffer
-            lowerSrc += width;
-        } // end row
-        pinMask  += 6;                // Next chain's RGB pin masks
-        upperSrc += halfMatrixOffset; // Advance to next matrix start pos
-        lowerSrc += halfMatrixOffset;
-    }
-}
-
-// Corresponding function for long output -- either several parallel chains
-// (up to 5), or 1 chain with RGB bits scattered widely about the PORT.
-// Same deal, comments are pared back, see above functions for explanations.
-void _PM_convert_565_long(Protomatter_core *core, uint16_t *source,
-  uint16_t width) {
-    uint16_t *upperSrc = source;                             // Matrix top half
-    uint16_t *lowerSrc = source + width * core->numRowPairs; // " bottom half
-    uint32_t *pinMask  = (uint32_t *)core->rgbMask;          // Pin bitmasks
-    uint32_t *dest     = (uint32_t *)core->screenData;
-    if(core->doubleBuffer) {
-        dest += core->bufferSize / core->bytesPerElement *
-          (1 - core->activeBuffer);
-    }
-
-    uint32_t bitplaneSize = _PM_chunkSize *
-      ((width + (_PM_chunkSize - 1)) / _PM_chunkSize); // 1 plane of row pair
-    uint8_t  pad          = bitplaneSize - width;      // Start-of-plane pad
-
-    uint32_t initialRedBit, initialGreenBit, initialBlueBit;
-    if(core->numPlanes == 6) {
-        initialRedBit   = 0b1000000000000000; // MSB red
-        initialGreenBit = 0b0000000000100000; // LSB green
-        initialBlueBit  = 0b0000000000010000; // MSB blue
-    } else {
-        uint8_t shiftLeft = 5 - core->numPlanes;
-        initialRedBit   = 0b0000100000000000 << shiftLeft;
-        initialGreenBit = 0b0000000001000000 << shiftLeft;
-        initialBlueBit  = 0b0000000000000001 << shiftLeft;
-    }
-
-#if defined(_PM_portToggleRegister)
-    // No per-chain loop is required; one clock bit handles all chains
-    uint32_t offset = 0; // Current position in the 'dest' buffer
-    for(uint8_t row=0; row<core->numRowPairs; row++) {
-        for(uint8_t plane=0; plane<core->numPlanes; plane++) {
-            dest[offset++] = 0; // First element of each plane
-            for(uint16_t x=1; x<bitplaneSize; x++) { // All subsequent items
-                dest[offset++] = core->clockMask;
-            }
-        }
-    }
-#else
-    memset(dest, 0, core->bufferSize);
-#endif
-
-    dest += pad; // Pad value is in 'elements,' not bytes, so this is OK
-
-    uint32_t halfMatrixOffset = width * core->numPlanes * core->numRowPairs;
-
-    for(uint8_t chain=0; chain<core->parallel; chain++) {
-        for(uint8_t row=0; row<core->numRowPairs; row++) {
-            uint32_t redBit   = initialRedBit;
-            uint32_t greenBit = initialGreenBit;
-            uint32_t blueBit  = initialBlueBit;
-            for(uint8_t plane=0; plane<core->numPlanes; plane++) {
-#if defined(_PM_portToggleRegister)
-                uint32_t prior = 0;
-#endif
-                for(uint16_t x=0; x<width; x++) {
-                    uint16_t upperRGB = upperSrc[x]; // Pixel in upper half
-                    uint16_t lowerRGB = lowerSrc[x]; // Pixel in lower half
-                    uint32_t result   = 0;
-                    if(upperRGB & redBit)   result |= pinMask[0];
-                    if(upperRGB & greenBit) result |= pinMask[1];
-                    if(upperRGB & blueBit)  result |= pinMask[2];
-                    if(lowerRGB & redBit)   result |= pinMask[3];
-                    if(lowerRGB & greenBit) result |= pinMask[4];
-                    if(lowerRGB & blueBit)  result |= pinMask[5];
-                    // Main difference here vs byte converter is each chain
-                    // ORs new bits into place (vs single-pass overwrite).
-#if defined(_PM_portToggleRegister)
-                    dest[x] |= result ^ prior; // Bitwise OR
-                    prior    = result;
-#else
-                    dest[x] |= result;         // Bitwise OR
-#endif
-                } // end x
-                greenBit <<= 1;
-                if(plane || (core->numPlanes < 6)) {
-                    redBit  <<= 1;
-                    blueBit <<= 1;
-                } else {
-                    redBit  = 0b0000100000000000;
-                    blueBit = 0b0000000000000001;
-                }
-                dest += bitplaneSize; // Advance one scanline in dest buffer
-            } // end plane
-            upperSrc += width; // Advance one scanline in source buffer
-            lowerSrc += width;
-        } // end row
-        pinMask  += 6;                // Next chain's RGB pin masks
-        upperSrc += halfMatrixOffset; // Advance to next matrix start pos
-        lowerSrc += halfMatrixOffset;
-    }
-}
-
-#endif // ARDUINO
-
-#endif // _PROTOMATTER_ARCH_H_

core.c

@@ -1,668 +0,0 @@
-// Device- and environment-neutral core matrix-driving functionality.
-// See notes near top of arch.h regarding assumptions of hardware
-// "common ground." If you find yourself doing an "#ifdef ARDUINO" or
-// "#ifdef _SAMD21_" in this file, STOP. Idea is that the code in this
-// file is neutral and portable (within aforementioned assumptions).
-// Nonportable elements should appear in arch.h. If arch.h functionality
-// is lacking, extend it there, do not go making device- or environment-
-// specific cases within this file.
-
-// Function names are intentionally a little obtuse, idea is that one writes
-// a more sensible wrapper around this for specific environments (e.g. the
-// Arduino stuff in Adafruit_Protomatter.cpp). The "_PM_" prefix on most
-// things hopefully makes function and variable name collisions much less
-// likely with one's own code.
-
-#include "core.h" // enums and structs
-#include "arch.h" // Do NOT include this in any other source files
-
-// Overall matrix refresh rate (frames/second) is a function of matrix width
-// and chain length, number of address lines, number of bit planes, CPU speed
-// and whether or not a GPIO toggle register is available. There is no "this
-// will run at X-frames-per-second" constant figure. You typically just have
-// to try it out and perhaps trade off some bit planes for refresh rate until
-// the image looks good and stable. Anything over 100 Hz is usually passable,
-// around 250 Hz is where things firm up. And while this could proceed higher
-// in some situations, the tradeoff is that faster rates use progressively
-// more CPU time (because it's timer interrupt based and not using DMA or
-// special peripherals). So a throttle is set here, an approximate maximum
-// frame rate which the software will attempt to avoid exceeding (but may
-// refresh slower than this, and in many cases will...just need to set an
-// upper limit to avoid excessive CPU load). An incredibly long comment block
-// for a single constant, thank you for coming to my TED talk!
-#define _PM_MAX_REFRESH_HZ 250
-
-// Time (in microseconds) to pause following any change in address lines
-// (individually or collectively). Some matrices respond slowly there...
-// must pause on change for matrix to catch up. Defined here (rather than
-// arch.h) because it's not architecture-specific.
-#define _PM_ROW_DELAY 8
-
-// These are the lowest-level functions for issing data to matrices.
-// There are three versions because it depends on how the six RGB data bits
-// (and clock bit) are arranged within a 32-bit PORT register. If all six
-// (seven) fit within one byte or word of the PORT, the library's memory
-// use (and corresponding data-issuing function) change. This will also have
-// an impact on parallel chains in the future, where the number of concurrent
-// RGB data bits isn't always six, but some multiple thereof (i.e. up to five
-// parallel outputs -- 30 RGB bits + clock -- on a 32-bit PORT, though that's
-// largely hypothetical as the chance of finding a PORT with that many bits
-// exposed and NOT interfering with other peripherals on a board is highly
-// improbable. But I could see four happening, maybe on a Grand Central or
-// other kitchen-sink board.
-static void blast_byte(Protomatter_core *core, uint8_t *data);
-static void blast_word(Protomatter_core *core, uint16_t *data);
-static void blast_long(Protomatter_core *core, uint32_t *data);
-
-// Validate and populate vital elements of core structure.
-// Does NOT allocate core struct -- calling function must provide that.
-// (In the Arduino C++ library, it’s part of the Protomatter class.)
-ProtomatterStatus _PM_init(Protomatter_core *core,
-  uint16_t bitWidth, uint8_t bitDepth,
-  uint8_t rgbCount, uint8_t *rgbList,
-  uint8_t addrCount, uint8_t *addrList,
-  uint8_t clockPin, uint8_t latchPin, uint8_t oePin,
-  bool doubleBuffer, void *timer) {
-    if(!core) return PROTOMATTER_ERR_ARG;
-
-    if(rgbCount  > 5) rgbCount  = 5; // Max 5 in parallel (32-bit PORT)
-    if(addrCount > 5) addrCount = 5; // Max 5 address lines (A-E)
-    // bitDepth is NOT constrained here, handle in calling function
-    // (varies with implementation, e.g. GFX lib is max 6 bitplanes,
-    // but might be more or less elsewhere)
-
-    // If NULL timer was passed in (the default case for the constructor),
-    // use default value from arch.h. For example, in the Arduino case it's
-    // tied to TC4 specifically.
-    if(timer == NULL) timer = _PM_TIMER_DEFAULT;
-
-    core->timer           = timer;
-    core->width           = bitWidth; // Total matrix chain length in bits
-    core->numPlanes       = bitDepth;
-    core->parallel        = rgbCount;
-    core->numAddressLines = addrCount;
-    core->clockPin        = clockPin;
-    core->latch.pin       = latchPin;
-    core->oe.pin          = oePin;
-    core->doubleBuffer    = doubleBuffer;
-    core->addr            = NULL;
-    core->screenData      = NULL;
-
-    // Make a copy of the rgbList and addrList tables in case they're
-    // passed from local vars on the stack or some other non-persistent
-    // source. screenData is NOT allocated here because data size (byte,
-    // word, long) is not known until the begin function evaluates all
-    // the pin bitmasks.
-
-    rgbCount *= 6; // Convert parallel count to pin count
-    if((core->rgbPins = (uint8_t *)malloc(rgbCount * sizeof(uint8_t)))) {
-        if((core->addr = (_PM_pin *)malloc(addrCount * sizeof(_PM_pin)))) {
-            memcpy(core->rgbPins, rgbList, rgbCount * sizeof(uint8_t));
-            for(uint8_t i=0; i<addrCount; i++) {
-                core->addr[i].pin = addrList[i];
-            }
-            return PROTOMATTER_OK;
-        }
-        free(core->rgbPins);
-        core->rgbPins = NULL;
-    }
-    return PROTOMATTER_ERR_MALLOC;
-}
-
-// Allocate display buffers and populate additional elements.
-ProtomatterStatus _PM_begin(Protomatter_core *core) {
-    if(!core) return PROTOMATTER_ERR_ARG;
-
-    if(!core->rgbPins) { // NULL if copy failed to allocate
-        return PROTOMATTER_ERR_MALLOC;
-    }
-
-    // Verify that rgbPins and clockPin are all on the same PORT. If not,
-    // return an error. Pin list is not freed; please call dealloc function.
-    // Also get bitmask of which bits within 32-bit PORT register are
-    // referenced.
-    uint8_t *port = (uint8_t *)_PM_portOutRegister(core->clockPin);
-#if defined(_PM_portToggleRegister)
-    // If a bit-toggle register is present, the clock pin is included
-    // in determining which bytes of the PORT register are used (and thus
-    // the data storage efficiency).
-    uint32_t bitMask = _PM_portBitMask(core->clockPin);
-#else
-    // If no bit-toggle register, clock pin can be on any bit, doesn't
-    // affect storage efficiency.
-    uint32_t bitMask = 0;
-#endif
-
-    for(uint8_t i=0; i<core->parallel * 6; i++) {
-        uint8_t *p2 = (uint8_t *)_PM_portOutRegister(core->rgbPins[i]);
-        if(p2 != port) {
-            return PROTOMATTER_ERR_PINS;
-        }
-        bitMask |= _PM_portBitMask(core->rgbPins[i]);
-    }
-
-    // RGB + clock are on same port, we can proceed...
-
-    // Determine data type for internal representation. If all the data
-    // bitmasks (and possibly clock bitmask, depending whether toggle-bits
-    // register is present) are in the same byte, this can be stored more
-    // compact than if they're spread across a word or long.
-    uint8_t byteMask = 0;
-    if(bitMask & 0xFF000000) byteMask |= 0b1000;
-    if(bitMask & 0x00FF0000) byteMask |= 0b0100;
-    if(bitMask & 0x0000FF00) byteMask |= 0b0010;
-    if(bitMask & 0x000000FF) byteMask |= 0b0001;
-    switch(byteMask) {
-      case 0b0001:                 // If all PORT bits are in the same byte...
-      case 0b0010:
-      case 0b0100:
-      case 0b1000:
-        core->bytesPerElement = 1; // Use 8-bit PORT accesses.
-        break;
-      case 0b0011:                 // If all PORT bits in upper/lower word...
-      case 0b1100:
-        core->bytesPerElement = 2; // Use 16-bit PORT accesses.
-        // Although some devices might tolerate unaligned 16-bit accesses
-        // ('middle' word of 32-bit PORT), that is NOT handled here.
-        // It's a portability liability.
-        break;
-      default:                     // Any other situation...
-        core->bytesPerElement = 4; // Use 32-bit PORT accesses.
-        break;
-    }
-
-    // Planning for screen data allocation...
-    core->numRowPairs    = 1 << core->numAddressLines;
-    uint8_t  chunks      = (core->width + (_PM_chunkSize - 1)) / _PM_chunkSize;
-    uint16_t columns     = chunks * _PM_chunkSize; // Padded matrix width
-    uint32_t screenBytes = columns * core->numRowPairs * core->numPlanes *
-      core->bytesPerElement;
-
-    core->bufferSize = screenBytes;    // Bytes per matrix buffer (1 or 2)
-    if(core->doubleBuffer) screenBytes *= 2; // Total for matrix buffer(s)
-    uint32_t rgbMaskBytes = core->parallel * 6 * core->bytesPerElement;
-
-    // Allocate matrix buffer(s). Don't worry about the return type...
-    // though we might be using words or longs for certain pin configs,
-    // malloc() by definition always aligns to the longest type.
-    if(!(core->screenData = (uint8_t *)malloc(screenBytes + rgbMaskBytes))) {
-        return PROTOMATTER_ERR_MALLOC;
-    }
-
-    // rgbMask data follows the matrix buffer(s)
-    core->rgbMask = core->screenData + screenBytes;
-
-#if !defined(_PM_portToggleRegister)
-    // Clear entire screenData buffer so there's no cruft in any pad bytes
-    // (if using toggle register, each is set to clockMask below instead).
-    memset(core->screenData, 0, screenBytes);
-#endif
-
-    // Figure out clockMask and rgbAndClockMask, clear matrix buffers
-    if(core->bytesPerElement == 1) {
-        core->portOffset = _PM_byteOffset(core->rgbPins[0]);
-#if defined(_PM_portToggleRegister)
-        // Clock and rgbAndClockMask are 8-bit values
-        core->clockMask = _PM_portBitMask(core->clockPin) >>
-          (core->portOffset * 8);
-        core->rgbAndClockMask = (bitMask >> (core->portOffset * 8)) |
-          core->clockMask;
-        memset(core->screenData, core->clockMask, screenBytes);
-#else
-        // Clock and rgbAndClockMask are 32-bit values
-        core->clockMask       = _PM_portBitMask(core->clockPin);
-        core->rgbAndClockMask = bitMask | core->clockMask;
-#endif
-        for(uint8_t i=0; i<core->parallel * 6; i++) {
-            ((uint8_t *)core->rgbMask)[i] = // Pin bitmasks are 8-bit
-              _PM_portBitMask(core->rgbPins[i]) >> (core->portOffset * 8);
-        }
-    } else if(core->bytesPerElement == 2) {
-        core->portOffset = _PM_wordOffset(core->rgbPins[0]);
-#if defined(_PM_portToggleRegister)
-        // Clock and rgbAndClockMask are 16-bit values
-        core->clockMask = _PM_portBitMask(core->clockPin) >>
-          (core->portOffset * 16);
-        core->rgbAndClockMask = (bitMask >> (core->portOffset * 16)) |
-          core->clockMask;
-        uint32_t elements = screenBytes / 2;
-        for(uint32_t i=0; i<elements; i++) {
-            ((uint16_t *)core->screenData)[i] = core->clockMask;
-        }
-#else
-        // Clock and rgbAndClockMask are 32-bit values
-        core->clockMask       = _PM_portBitMask(core->clockPin);
-        core->rgbAndClockMask = bitMask | core->clockMask;
-#endif
-        for(uint8_t i=0; i<core->parallel * 6; i++) {
-            ((uint16_t *)core->rgbMask)[i] = // Pin bitmasks are 16-bit
-              _PM_portBitMask(core->rgbPins[i]) >> (core->portOffset * 16);
-        }
-    } else {
-        core->portOffset      = 0;
-        core->clockMask       = _PM_portBitMask(core->clockPin);
-        core->rgbAndClockMask = bitMask | core->clockMask;
-#if defined(_PM_portToggleRegister)
-        uint32_t elements = screenBytes / 4;
-        for(uint32_t i=0; i<elements; i++) {
-            ((uint32_t *)core->screenData)[i] = core->clockMask;
-        }
-#endif
-        for(uint8_t i=0; i<core->parallel * 6; i++) {
-            ((uint32_t *)core->rgbMask)[i] = // Pin bitmasks are 32-bit
-              _PM_portBitMask(core->rgbPins[i]);
-        }
-    }
-
-    // Estimate minimum bitplane #0 period for _PM_MAX_REFRESH_HZ rate.
-    uint32_t minPeriodPerFrame = _PM_timerFreq / _PM_MAX_REFRESH_HZ;
-    uint32_t minPeriodPerLine  = minPeriodPerFrame / core->numRowPairs;
-    core->minPeriod = minPeriodPerLine / ((1 << core->numPlanes) - 1);
-    if(core->minPeriod < _PM_minMinPeriod) {
-        core->minPeriod = _PM_minMinPeriod;
-    }
-    // Actual frame rate may be lower than this...it's only an estimate
-    // and does not factor in things like address line selection delays
-    // or interrupt overhead. That's OK, just don't want to exceed this
-    // rate, as it'll eat all the CPU cycles.
-    // Make a wild guess for the initial bit-zero interval. It's okay
-    // that this is off, code adapts to actual timer results pretty quick.
-
-    core->bitZeroPeriod = core->width * 5; // Initial guesstimate
-
-    core->activeBuffer  = 0;
-
-    // Configure pins as outputs and initialize their states.
-
-    core->latch.setReg   = _PM_portSetRegister(core->latch.pin);
-    core->latch.clearReg = _PM_portClearRegister(core->latch.pin);
-    core->latch.bit      = _PM_portBitMask(core->latch.pin);
-    core->oe.setReg      = _PM_portSetRegister(core->oe.pin);
-    core->oe.clearReg    = _PM_portClearRegister(core->oe.pin);
-    core->oe.bit         = _PM_portBitMask(core->oe.pin);
-
-    _PM_pinOutput(core->clockPin);
-    _PM_pinLow(core->clockPin);  // Init clock LOW
-    _PM_pinOutput(core->latch.pin);
-    _PM_pinLow(core->latch.pin); // Init latch LOW
-    _PM_pinOutput(core->oe.pin);
-    _PM_pinHigh(core->oe.pin);   // Init OE HIGH (disable output)
-
-    for(uint8_t i=0; i<core->parallel * 6; i++) {
-        _PM_pinOutput(core->rgbPins[i]);
-        _PM_pinLow(core->rgbPins[i]);
-    }
-#if defined(_PM_portToggleRegister)
-    core->addrPortToggle = _PM_portToggleRegister(core->addr[0].pin);
-    core->singleAddrPort = 1;
-#endif
-    for(uint8_t line=0,bit=1; line<core->numAddressLines; line++, bit<<=1) {
-        core->addr[line].setReg =
-          _PM_portSetRegister(core->addr[line].pin);
-        core->addr[line].clearReg =
-          _PM_portClearRegister(core->addr[line].pin);
-        core->addr[line].bit =
-          _PM_portBitMask(core->addr[line].pin);
-        _PM_pinOutput(core->addr[line].pin);
-        if(core->prevRow & bit) {
-            _PM_pinHigh(core->addr[line].pin);
-        } else {
-            _PM_pinLow(core->addr[line].pin);
-        }
-#if defined(_PM_portToggleRegister)
-        // If address pin on different port than addr 0, no singleAddrPort.
-        if(_PM_portToggleRegister(core->addr[line].pin) !=
-          core->addrPortToggle) {
-            core->singleAddrPort = 0;
-        }
-#endif
-    }
-
-    // Get pointers to bit set and clear registers (and toggle, if present)
-    core->setReg    = (uint8_t *)_PM_portSetRegister(core->clockPin);
-    core->clearReg  = (uint8_t *)_PM_portClearRegister(core->clockPin);
-#if defined(_PM_portToggleRegister)
-    core->toggleReg = (uint8_t *)_PM_portToggleRegister(core->clockPin);
-#endif
-
-    // Reset plane/row counters, config and start timer
-    _PM_resume(core);
-
-    return PROTOMATTER_OK;
-}
-
-// Disable (but do not deallocate) a Protomatter matrix. Disables matrix by
-// setting OE pin HIGH and writing all-zero data to matrix shift registers,
-// so it won't halt with lit LEDs.
-void _PM_stop(Protomatter_core *core) {
-    if((core)) {
-        while(core->swapBuffers);        // Wait for any pending buffer swap
-        _PM_timerStop(core->timer);      // Halt timer
-        *core->oe.setReg = core->oe.bit; // Set OE HIGH (disable output)
-        // So, in PRINCIPLE, setting OE high would be sufficient...
-        // but in case that pin is shared with another function such
-        // as the onloard LED (which pulses during bootloading) let's
-        // also clear out the matrix shift registers for good measure.
-        // Set all RGB pins LOW...
-        for(uint8_t i=0; i<core->parallel * 6; i++) {
-            _PM_pinLow(core->rgbPins[i]);
-        }
-        // Clock out bits (just need to toggle clock with RGBs held low)
-        for(uint32_t i=0; i<core->width; i++) {
-            _PM_pinHigh(core->clockPin);
-            _PM_clockHoldHigh;
-            _PM_pinLow(core->clockPin);
-            _PM_clockHoldLow;
-        }
-        // Latch data
-        *core->latch.setReg   = core->latch.bit;
-        *core->latch.clearReg = core->latch.bit;
-    }
-}
-
-void _PM_resume(Protomatter_core *core) {
-    if((core)) {
-        // Init plane & row to max values so they roll over on 1st interrupt
-        core->plane       = core->numPlanes   - 1;
-        core->row         = core->numRowPairs - 1;
-        core->prevRow     = (core->numRowPairs > 1) ? (core->row - 1) : 1;
-        core->swapBuffers = 0;
-        core->frameCount  = 0;
-
-        _PM_timerInit(core->timer);        // Configure timer
-        _PM_timerStart(core->timer, 1000); // Start timer
-    }
-}
-
-// Free memory associated with core structure. Does NOT dealloc struct.
-void _PM_free(Protomatter_core *core) {
-    if((core)) {
-        _PM_stop(core);
-        // TO DO: Set all pins back to inputs here?
-        if(core->screenData) free(core->screenData);
-        if(core->addr)       free(core->addr);
-        if(core->rgbPins) {
-            free(core->rgbPins);
-            core->rgbPins = NULL;
-        }
-    }
-}
-
-
-// ISR function (in arch.h) calls this function which it extern'd.
-void _PM_row_handler(Protomatter_core *core) {
-
-    *core->oe.setReg = core->oe.bit; // Disable LED output
-
-    *core->latch.setReg   = core->latch.bit; // Latch data from PRIOR pass
-    // Stop timer, save count value at stop
-    uint32_t elapsed = _PM_timerStop(core->timer);
-    uint8_t prevPlane = core->plane; // Save that plane # for later timing
-    *core->latch.clearReg = core->latch.bit; // (split to add a few cycles)
-
-    // If plane 0 just finished being displayed (plane 1 was loaded on prior
-    // pass, or there's only one plane...I know, it's confusing), take note
-    // of the elapsed timer value, for subsequent bitplane timing (each
-    // plane period is double the previous). Value is filtered slightly to
-    // avoid jitter.
-    if((prevPlane == 1) || (core->numPlanes == 1)) {
-        core->bitZeroPeriod = ((core->bitZeroPeriod * 7) + elapsed) / 8;
-        if(core->bitZeroPeriod < core->minPeriod) {
-            core->bitZeroPeriod = core->minPeriod;
-        }
-    }
-
-    if(prevPlane == 0) { // Plane 0 just finished loading
-#if defined(_PM_portToggleRegister)
-        // If all address lines are on a single PORT (and bit toggle is
-        // available), do address line change all at once. Even doing all
-        // this math takes MUCH less time than the delays required when
-        // doing line-by-line changes.
-        if(core->singleAddrPort) {
-            // Make bitmasks of prior and new row bits
-            uint32_t priorBits = 0, newBits = 0;
-            for(uint8_t line=0,bit=1; line<core->numAddressLines;
-              line++, bit<<=1) {
-                if(core->row & bit) {
-                    newBits |= core->addr[line].bit;
-                }
-                if(core->prevRow & bit) {
-                    priorBits |= core->addr[line].bit;
-                }
-            }
-            *core->addrPortToggle = newBits ^ priorBits;
-            _PM_delayMicroseconds(_PM_ROW_DELAY);
-        } else {
-#endif
-            // Configure row address lines individually, making changes
-            // (with delays) only where necessary.
-            for(uint8_t line=0,bit=1; line<core->numAddressLines;
-              line++, bit<<=1) {
-                if((core->row & bit) != (core->prevRow & bit)) {
-                    if(core->row & bit) { // Set addr line high
-                         *core->addr[line].setReg = core->addr[line].bit;
-                    } else { // Set addr line low
-                         *core->addr[line].clearReg = core->addr[line].bit;
-                    }
-                    _PM_delayMicroseconds(_PM_ROW_DELAY);
-                }
-            }
-#if defined(_PM_portToggleRegister)
-        }
-#endif
-        core->prevRow = core->row;
-    }
-
-    // Advance bitplane index and/or row as necessary
-    if(++core->plane >= core->numPlanes) {     // Next data bitplane, or
-        core->plane = 0;                       // roll over bitplane to start
-        if(++core->row >= core->numRowPairs) { // Next row, or
-            core->row = 0;                     // roll over row to start
-            // Switch matrix buffers if due (only if double-buffered)
-            if(core->swapBuffers) {
-                core->activeBuffer = 1 - core->activeBuffer;
-                core->swapBuffers  = 0;        // Swapped!
-            }
-            core->frameCount++;
-        }
-    }
-
-    // 'plane' now is index of data to issue, NOT data to display.
-    // 'prevPlane' is the previously-loaded data, which gets displayed
-    // now while the next plane data is loaded.
-
-    // Set timer and enable LED output for data loaded on PRIOR pass:
-    _PM_timerStart(core->timer, core->bitZeroPeriod << prevPlane);
-    *core->oe.clearReg = core->oe.bit; // Enable LED output
-
-    uint32_t elementsPerLine = _PM_chunkSize *
-        ((core->width + (_PM_chunkSize - 1)) / _PM_chunkSize);
-    uint32_t srcOffset = elementsPerLine *
-      (core->numPlanes * core->row + core->plane) * core->bytesPerElement;
-    if(core->doubleBuffer) {
-        srcOffset += core->bufferSize * core->activeBuffer;
-    }
-
-    if(core->bytesPerElement == 1) {
-        blast_byte(core, (uint8_t *)(core->screenData + srcOffset));
-    } else if(core->bytesPerElement == 2) {
-        blast_word(core, (uint16_t *)(core->screenData + srcOffset));
-    } else {
-        blast_long(core, (uint32_t *)(core->screenData + srcOffset));
-    }
-
-    // 'plane' data is now loaded, will be shown on NEXT pass
-}
-
-// Innermost data-stuffing loop functions
-
-// The presence of a bit-toggle register can make the data-stuffing loop a
-// fair bit faster (2 PORT accesses per column vs 3). But ironically, some
-// devices (e.g. SAMD51) can outpace the matrix max CLK speed, so we slow
-// them down with a few NOPs. These are defined in arch.h as needed.
-// _PM_clockHoldLow is whatever code necessary to delay the clock rise
-// after data is placed on the PORT. _PM_clockHoldHigh is code for delay
-// before setting the clock back low. If undefined, nothing goes there.
-
-#if defined(_PM_portToggleRegister)
-  #define PEW \
-    *toggle  = *data++; /* Toggle in new data + toggle clock low */ \
-    _PM_clockHoldLow; \
-    *toggle  =  clock;  /* Toggle clock high */ \
-    _PM_clockHoldHigh;
-#else
-  #define PEW \
-    *set     = *data++;   /* Set RGB data high */ \
-    _PM_clockHoldLow; \
-    *set32   =  clock;    /* Set clock high */ \
-    _PM_clockHoldHigh; \
-    *clear32 =  rgbclock; /* Clear RGB data + clock */
-#endif
-
-#if _PM_chunkSize == 1
-  #define PEW_UNROLL PEW
-#elif _PM_chunkSize == 8
-  #define PEW_UNROLL PEW PEW PEW PEW PEW PEW PEW PEW
-#elif _PM_chunkSize == 16
-  #define PEW_UNROLL \
-    PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW
-#elif _PM_chunkSize == 32
-  #define PEW_UNROLL \
-    PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW \
-    PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW
-#elif _PM_chunkSize == 64
-  #define PEW_UNROLL \
-    PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW \
-    PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW \
-    PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW \
-    PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW PEW
-#else
-  #error "Unimplemented _PM_chunkSize value"
-#endif
-
-// There are THREE COPIES of the following function -- one each for byte,
-// word and long. If changes are made in any one of them, the others MUST
-// be updated to match! (Decided against using macro tricks for the
-// function, too often ends in disaster...but must be vigilant in the
-// three-function maintenance then.)
-
-static void blast_byte(Protomatter_core *core, uint8_t *data) {
-#if defined(_PM_portToggleRegister)
-    // If here, it was established in begin() that the RGB data bits and
-    // clock are all within the same byte of a PORT register, else we'd be
-    // in the word- or long-blasting functions now. So we just need an
-    // 8-bit pointer to the PORT.
-    volatile uint8_t *toggle = (volatile uint8_t *)core->toggleReg +
-        core->portOffset;
-#else
-    // No-toggle version is a little different. If here, RGB data is all
-    // in one byte of PORT register, clock can be any bit in 32-bit PORT.
-    volatile uint8_t  *set;     // For RGB data set
-    volatile uint32_t *set32;   // For clock set
-    volatile uint32_t *clear32; // For RGB data + clock clear
-    set     = (volatile uint8_t *)core->setReg + portOffset;
-    set32   = (volatile uint32_t *)core->setReg;
-    clear32 = (volatile uint32_t *)core->clearReg;
-    uint32_t rgbclock = core->rgbAndClockMask; // RGB + clock bit
-#endif
-    uint32_t clock  = core->clockMask; // Clock bit
-    uint8_t  chunks = (core->width + (_PM_chunkSize - 1)) / _PM_chunkSize;
-
-    // PORT has already been initialized with RGB data + clock bits
-    // all LOW, so we don't need to initialize that state here.
-
-    while(chunks--) {
-        PEW_UNROLL // _PM_chunkSize RGB+clock writes
-    }
-
-#if defined(_PM_portToggleRegister)
-    // Want the PORT left with RGB data and clock LOW on function exit
-    // (so it's easier to see on 'scope, and to prime it for the next call).
-    // This is implicit in the no-toggle case (due to how the PEW macro
-    // works), but toggle case requires explicitly clearing those bits.
-    // rgbAndClockMask is an 8-bit value when toggling, hence offset here.
-    *((volatile uint8_t *)core->clearReg + core->portOffset) =
-      core->rgbAndClockMask;
-#endif
-}
-
-static void blast_word(Protomatter_core *core, uint16_t *data) {
-#if defined(_PM_portToggleRegister)
-    // See notes above -- except now 16-bit word in PORT.
-    volatile uint16_t *toggle = (volatile uint16_t *)core->toggleReg +
-        core->portOffset;
-#else
-    volatile uint16_t *set;     // For RGB data set
-    volatile uint32_t *set32;   // For clock set
-    volatile uint32_t *clear32; // For RGB data + clock clear
-    set     = (volatile uint16_t *)core->setReg + core->portOffset;
-    set32   = (volatile uint32_t *)core->setReg;
-    clear32 = (volatile uint32_t *)core->clearReg;
-    uint32_t rgbclock = core->rgbAndClockMask; // RGB + clock bit
-#endif
-    uint32_t clock  = core->clockMask; // Clock bit
-    uint8_t  chunks = (core->width + (_PM_chunkSize - 1)) / _PM_chunkSize;
-    while(chunks--) {
-        PEW_UNROLL // _PM_chunkSize RGB+clock writes
-    }
-#if defined(_PM_portToggleRegister)
-    // rgbAndClockMask is a 16-bit value when toggling, hence offset here.
-    *((volatile uint16_t *)core->clearReg + core->portOffset) =
-        core->rgbAndClockMask;
-#endif
-}
-
-static void blast_long(Protomatter_core *core, uint32_t *data) {
-#if defined(_PM_portToggleRegister)
-    // See notes above -- except now full 32-bit PORT.
-    volatile uint32_t *toggle = (volatile uint32_t *)core->toggleReg;
-#else
-    // Note in this case two copies exist of the PORT set register.
-    // The optimizer will most likely simplify this; leaving as-is, not
-    // wanting a special case of the PEW macro due to divergence risk.
-    volatile uint32_t *set;     // For RGB data set
-    volatile uint32_t *set32;   // For clock set
-    volatile uint32_t *clear32; // For RGB data + clock clear
-    set     = (volatile uint32_t *)core->setReg;
-    set32   = (volatile uint32_t *)core->setReg;
-    clear32 = (volatile uint32_t *)core->clearReg;
-    uint32_t rgbclock = core->rgbAndClockMask; // RGB + clock bit
-#endif
-    uint32_t clock  = core->clockMask; // Clock bit
-    uint8_t  chunks = (core->width + (_PM_chunkSize - 1)) / _PM_chunkSize;
-    while(chunks--) {
-        PEW_UNROLL // _PM_chunkSize RGB+clock writes
-    }
-#if defined(_PM_portToggleRegister)
-    *(volatile uint32_t *)core->clearReg = core->rgbAndClockMask;
-#endif
-}
-
-// Returns current value of frame counter and resets its value to zero.
-// Two calls to this, timed one second apart (or use math with other
-// intervals), can be used to get a rough frames-per-second value for
-// the matrix (since this is difficult to estimate beforehand).
-uint32_t _PM_getFrameCount(Protomatter_core *core) {
-    uint32_t count = 0;
-    if((core)) {
-        count = core->frameCount;
-        core->frameCount = 0;
-    }
-    return count;
-}
-
-// Note to future self: I've gone back and forth between implementing all
-// this either as it currently is (with byte, word and long cases for various
-// steps), or using a uint32_t[64] table for expanding RGB bit combos to PORT
-// bit combos. The latter would certainly simplify the code a ton, and the
-// additional table lookup step wouldn't significantly impact performance,
-// especially going forward with faster processors (the SAMD51 code already
-// requires a few NOPs in the innermost loop to avoid outpacing the matrix).
-// BUT, the reason this is NOT currently done is that it only allows for a
-// single matrix chain (doing parallel chains would require either an
-// impractically large lookup table, or adding together multiple tables'
-// worth of bitmasks, which would slow things down in the vital inner loop).
-// Although parallel matrix chains aren't yet 100% implemented in this code
-// right now, I wanted to leave that possibility for the future, as a way to
-// handle larger matrix combos, because long chains will slow down the
-// refresh rate.

core.h

@@ -1,110 +0,0 @@
-#ifndef _PROTOMATTER_CORE_H_
-#define _PROTOMATTER_CORE_H_
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-#include <stdint.h>
-#include <stdbool.h>
-
-// Status type returned by some functions.
-typedef enum {
-    PROTOMATTER_OK,         // Everything is hunky-dory!
-    PROTOMATTER_ERR_PINS,   // Clock and/or data pins on different PORTs
-    PROTOMATTER_ERR_MALLOC, // Couldn't allocate memory for display
-    PROTOMATTER_ERR_ARG,    // Bad input to function
-} ProtomatterStatus;
-
-// Struct for matrix control lines NOT related to RGB data or clock, i.e.
-// latch, OE and address lines. RGB data and clock ("RGBC") are handled
-// differently as they have specific requirements (and might use a toggle
-// register if present). The data conversion functions need bitmasks for
-// RGB data but do NOT need the set or clear registers, so those items
-// are also declared as separate things in the core structure that follows.
-typedef struct {
-    volatile uint32_t *setReg;          // GPIO bit set register
-    volatile uint32_t *clearReg;        // GPIO bit clear register
-    uint32_t           bit;             // GPIO bitmask
-    uint8_t            pin;             // Some identifier, e.g. Arduino pin #
-} _PM_pin;
-
-// Struct with info about an RGB matrix chain and lots of state and buffer
-// details for the library. Toggle-related items in this structure MUST be
-// declared even if the device lacks GPIO bit-toggle registers (i.e. don't
-// do an ifdef check around these). All hardware-specific details (including
-// the presence or lack of toggle registers) are isolated to a single
-// file -- arch.h -- which should ONLY be included by core.c, and ifdef'ing
-// them would result in differing representations of this structure which
-// must be shared between the library and calling code. (An alternative is
-// to put any toggle-specific stuff at the end of the struct with an ifdef
-// check, but that's just dirty pool and asking for trouble.)
-typedef struct {
-    void              *timer;           // Arch-specific timer/counter info
-    void              *setReg;          // RGBC bit set register (cast to use)
-    void              *clearReg;        // RGBC bit clear register "
-    void              *toggleReg;       // RGBC bit toggle register "
-    uint8_t           *rgbPins;         // Array of RGB data pins (mult of 6)
-    void              *rgbMask;         // PORT bit mask for each RGB pin
-    uint32_t           clockMask;       // PORT bit mask for RGB clock
-    uint32_t           rgbAndClockMask; // PORT bit mask for RGB data + clock
-    volatile uint32_t *addrPortToggle;  // See singleAddrPort below
-    void              *screenData;      // Per-bitplane RGB data for matrix
-    _PM_pin            latch;           // RGB data latch
-    _PM_pin            oe;              // !OE (LOW out enable)
-    _PM_pin           *addr;            // Array of address pins
-    uint32_t           bufferSize;      // Bytes per matrix buffer
-    uint32_t           bitZeroPeriod;   // Bitplane 0 timer period
-    uint32_t           minPeriod;       // Plane 0 timer period for ~250Hz
-    volatile uint32_t  frameCount;      // For estimating refresh rate
-    uint16_t           width;           // Matrix chain width in bits
-    uint8_t            bytesPerElement; // Using 8, 16 or 32 bits of PORT?
-    uint8_t            clockPin;        // RGB clock pin identifier
-    uint8_t            parallel;        // Number of concurrent matrix outs
-    uint8_t            numAddressLines; // Number of address line pins
-    uint8_t            portOffset;      // Active 8- or 16-bit pos. in PORT
-    uint8_t            numPlanes;       // Display bitplanes (1 to 6)
-    uint8_t            numRowPairs;     // Addressable row pairs
-    bool               doubleBuffer;    // 2X buffers for clean switchover
-    bool               singleAddrPort;  // If 1, all addr lines on same PORT
-    volatile uint8_t   activeBuffer;    // Index of currently-displayed buf
-    volatile uint8_t   plane;           // Current bitplane (changes in ISR)
-    volatile uint8_t   row;             // Current scanline (changes in ISR)
-    volatile uint8_t   prevRow;         // Scanline from prior ISR
-    volatile bool      swapBuffers;     // If 1, awaiting double-buf switch
-} Protomatter_core;
-
-// Protomatter core function prototypes. Environment-specific code (like the
-// Adafruit_Protomatter class for Arduino) calls on these underlying things,
-// and has to provide a few extras of its own (interrupt handlers and such).
-// User code shouldn't need to invoke any of them directly.
-extern ProtomatterStatus _PM_init(Protomatter_core *core,
-  uint16_t bitWidth, uint8_t bitDepth,
-  uint8_t rgbCount, uint8_t *rgbList,
-  uint8_t addrCount, uint8_t *addrList,
-  uint8_t clockPin, uint8_t latchPin, uint8_t oePin,
-  bool doubleBuffer, void *timer);
-extern ProtomatterStatus _PM_begin(Protomatter_core *core);
-extern void              _PM_stop(Protomatter_core *core);
-extern void              _PM_resume(Protomatter_core *core);
-extern void              _PM_free(Protomatter_core *core);
-extern void              _PM_row_handler(Protomatter_core *core);
-extern uint32_t          _PM_getFrameCount(Protomatter_core *core);
-extern void              _PM_timerStart(void *tptr, uint32_t period);
-extern uint32_t          _PM_timerStop(void *tptr);
-extern uint32_t          _PM_timerGetCount(void *tptr);
-
-#if defined(ARDUINO)
-extern void _PM_convert_565_byte(Protomatter_core *core,
-  uint16_t *source, uint16_t width);
-extern void _PM_convert_565_word(Protomatter_core *core,
-  uint16_t *source, uint16_t width);
-extern void _PM_convert_565_long(Protomatter_core *core,
-  uint16_t *source, uint16_t width);
-#endif // ARDUINO
-
-#ifdef __cplusplus
-} // extern "C"
-#endif
-
-#endif // _PROTOMATTER_CORE_H_

examples/animated_gif/animated_gif.ino

@@ -0,0 +1,365 @@
+// Play GIFs from CIRCUITPY drive (USB-accessible filesystem) to LED matrix.
+// Designed for Adafruit MatrixPortal M4, but may run on some other M4 & M0
+// and nRF52 boards (relies on TinyUSB stack). As written, runs on 64x32 pixel
+// matrix, this can be changed by editing the addrPins[] array (height) and/or
+// matrix constructor call (width).
+// Adapted from examples from Larry Bank's AnimatedGIF library and
+// msc_external_flash example in Adafruit_TinyUSB_Arduino.
+// Prerequisite libraries:
+//   - Adafruit_Protomatter
+//   - Adafruit_SPIFlash
+//   - Adafruit_TinyUSB
+//   - SdFat (Adafruit fork)
+//   - AnimatedGIF
+// Set ENABLE_EXTENDED_TRANSFER_CLASS and FAT12_SUPPORT in SdFatConfig.h.
+// Select Tools->USB Stack->TinyUSB before compiling.
+
+#include <Adafruit_Protomatter.h>
+#include <Adafruit_SPIFlash.h>
+#include <Adafruit_TinyUSB.h>
+#include <AnimatedGIF.h>
+#include <SPI.h>
+#include <SdFat.h>
+
+// CONFIGURABLE SETTINGS ---------------------------------------------------
+
+char GIFpath[] = "/gifs";     // Absolute path to GIFs on CIRCUITPY drive
+uint16_t GIFminimumTime = 10; // Min. repeat time (seconds) until next GIF
+
+// FLASH FILESYSTEM STUFF --------------------------------------------------
+
+// External flash macros for QSPI or SPI are defined in board variant file.
+#if defined(EXTERNAL_FLASH_USE_QSPI)
+Adafruit_FlashTransport_QSPI flashTransport;
+#elif defined(EXTERNAL_FLASH_USE_SPI)
+Adafruit_FlashTransport_SPI flashTransport(EXTERNAL_FLASH_USE_CS,
+                                           EXTERNAL_FLASH_USE_SPI);
+#else
+#error No QSPI/SPI flash are defined in your board variant.h!
+#endif
+
+Adafruit_SPIFlash flash(&flashTransport);
+FatFileSystem filesys;     // Filesystem object from SdFat
+Adafruit_USBD_MSC usb_msc; // USB mass storage object
+
+// RGB MATRIX (PROTOMATTER) LIBRARY STUFF ----------------------------------
+
+#if defined(_VARIANT_MATRIXPORTAL_M4_)
+uint8_t rgbPins[] = {7, 8, 9, 10, 11, 12};
+uint8_t addrPins[] = {17, 18, 19, 20};
+uint8_t clockPin = 14;
+uint8_t latchPin = 15;
+uint8_t oePin = 16;
+#define BACK_BUTTON 2
+#define NEXT_BUTTON 3
+#elif defined(_VARIANT_METRO_M4_)
+uint8_t rgbPins[] = {2, 3, 4, 5, 6, 7};
+uint8_t addrPins[] = {A0, A1, A2, A3};
+uint8_t clockPin = A4;
+uint8_t latchPin = 10;
+uint8_t oePin = 9;
+#elif defined(_VARIANT_FEATHER_M4_)
+uint8_t rgbPins[] = {6, 5, 9, 11, 10, 12};
+uint8_t addrPins[] = {A5, A4, A3, A2};
+uint8_t clockPin = 13;
+uint8_t latchPin = 0;
+uint8_t oePin = 1;
+#endif
+
+// Matrix width is first arg here, height is inferred from addrPins[]
+Adafruit_Protomatter matrix(64, 6, 1, rgbPins, sizeof addrPins, addrPins,
+                            clockPin, latchPin, oePin, true);
+
+// ANIMATEDGIF LIBRARY STUFF -----------------------------------------------
+
+AnimatedGIF GIF;
+File GIFfile;
+int16_t xPos = 0, yPos = 0; // Top-left pixel coord of GIF in matrix space
+
+// FILE ACCESS FUNCTIONS REQUIRED BY ANIMATED GIF LIB ----------------------
+
+// Pass in ABSOLUTE PATH of GIF file to open
+void *GIFOpenFile(char *filename, int32_t *pSize) {
+  GIFfile = filesys.open(filename);
+  if (GIFfile) {
+    *pSize = GIFfile.size();
+    return (void *)&GIFfile;
+  }
+  return NULL;
+}
+
+void GIFCloseFile(void *pHandle) {
+  File *f = static_cast<File *>(pHandle);
+  if (f) f->close();
+}
+
+int32_t GIFReadFile(GIFFILE *pFile, uint8_t *pBuf, int32_t iLen) {
+  int32_t iBytesRead = iLen;
+  File *f = static_cast<File *>(pFile->fHandle);
+  // If a file is read all the way to last byte, seek() stops working
+  if ((pFile->iSize - pFile->iPos) < iLen)
+    iBytesRead = pFile->iSize - pFile->iPos - 1; // ugly work-around
+  if (iBytesRead <= 0) return 0;
+  iBytesRead = (int32_t)f->read(pBuf, iBytesRead);
+  pFile->iPos = f->position();
+  return iBytesRead;
+}
+
+int32_t GIFSeekFile(GIFFILE *pFile, int32_t iPosition) {
+  File *f = static_cast<File *>(pFile->fHandle);
+  f->seek(iPosition);
+  pFile->iPos = (int32_t)f->position();
+  return pFile->iPos;
+}
+
+// Draw one line of image to matrix back buffer
+void GIFDraw(GIFDRAW *pDraw) {
+  uint8_t *s;
+  uint16_t *d, *usPalette, usTemp[320];
+  int x, y;
+
+  y = pDraw->iY + pDraw->y; // current line in image
+
+  // Vertical clip
+  int16_t screenY = yPos + y; // current row on matrix
+  if ((screenY < 0) || (screenY >= matrix.height())) return;
+
+  usPalette = pDraw->pPalette;
+
+  s = pDraw->pPixels;
+  // Apply the new pixels to the main image
+  if (pDraw->ucHasTransparency) { // if transparency used
+    uint8_t *pEnd, c, ucTransparent = pDraw->ucTransparent;
+    int x, iCount;
+    pEnd = s + pDraw->iWidth;
+    x = 0;
+    iCount = 0; // count non-transparent pixels
+    while (x < pDraw->iWidth) {
+      c = ucTransparent - 1;
+      d = usTemp;
+      while (c != ucTransparent && s < pEnd) {
+        c = *s++;
+        if (c == ucTransparent) { // done, stop
+          s--;                    // back up to treat it like transparent
+        } else {                  // opaque
+          *d++ = usPalette[c];
+          iCount++;
+        }
+      }             // while looking for opaque pixels
+      if (iCount) { // any opaque pixels?
+        span(usTemp, xPos + pDraw->iX + x, screenY, iCount);
+        x += iCount;
+        iCount = 0;
+      }
+      // no, look for a run of transparent pixels
+      c = ucTransparent;
+      while (c == ucTransparent && s < pEnd) {
+        c = *s++;
+        if (c == ucTransparent)
+          iCount++;
+        else
+          s--;
+      }
+      if (iCount) {
+        x += iCount; // skip these
+        iCount = 0;
+      }
+    }
+  } else {
+    s = pDraw->pPixels;
+    // Translate 8-bit pixels through RGB565 palette (already byte reversed)
+    for (x = 0; x < pDraw->iWidth; x++)
+      usTemp[x] = usPalette[*s++];
+    span(usTemp, xPos + pDraw->iX, screenY, pDraw->iWidth);
+  }
+}
+
+// Copy a horizontal span of pixels from a source buffer to an X,Y position
+// in matrix back buffer, applying horizontal clipping. Vertical clipping is
+// handled in GIFDraw() above -- y can safely be assumed valid here.
+void span(uint16_t *src, int16_t x, int16_t y, int16_t width) {
+  if (x >= matrix.width()) return; // Span entirely off right of matrix
+  int16_t x2 = x + width - 1;      // Rightmost pixel
+  if (x2 < 0) return;              // Span entirely off left of matrix
+  if (x < 0) {                     // Span partially off left of matrix
+    width += x;                    // Decrease span width
+    src -= x;                      // Increment source pointer to new start
+    x = 0;                         // Leftmost pixel is first column
+  }
+  if (x2 >= matrix.width()) {      // Span partially off right of matrix
+    width -= (x2 - matrix.width() + 1);
+  }
+  if(matrix.getRotation() == 0) {
+    memcpy(matrix.getBuffer() + y * matrix.width() + x, src, width * 2);
+  } else {
+    while(x <= x2) {
+      matrix.drawPixel(x++, y, *src++);
+    }
+  }
+}
+
+// FUNCTIONS REQUIRED FOR USB MASS STORAGE ---------------------------------
+
+static bool msc_changed = true; // Is set true on filesystem changes
+
+// Callback on READ10 command.
+int32_t msc_read_cb(uint32_t lba, void *buffer, uint32_t bufsize) {
+  return flash.readBlocks(lba, (uint8_t *)buffer, bufsize / 512) ? bufsize : -1;
+}
+
+// Callback on WRITE10 command.
+int32_t msc_write_cb(uint32_t lba, uint8_t *buffer, uint32_t bufsize) {
+  digitalWrite(LED_BUILTIN, HIGH);
+  return flash.writeBlocks(lba, buffer, bufsize / 512) ? bufsize : -1;
+}
+
+// Callback on WRITE10 completion.
+void msc_flush_cb(void) {
+  flash.syncBlocks();   // Sync with flash
+  filesys.cacheClear(); // Clear filesystem cache to force refresh
+  digitalWrite(LED_BUILTIN, LOW);
+  msc_changed = true;
+}
+
+// Get number of files in a specified path that match extension ('filter').
+// Pass in absolute path (e.g. "/" or "/gifs") and extension WITHOUT period
+// (e.g. "gif", NOT ".gif").
+int16_t numFiles(const char *path, const char *filter) {
+  File dir = filesys.open(path);
+  if (!dir) return -1;
+  char filename[256];
+  for(int16_t num_files = 0;;) {
+    File entry = dir.openNextFile();
+    if (!entry) return num_files; // No more files
+    entry.getName(filename, sizeof(filename) - 1);
+    entry.close();
+    if (!entry.isDirectory() &&       // Skip directories
+        strncmp(filename, "._", 2)) { // and Mac junk files
+      char *extension = strrchr(filename, '.');
+      if (extension && !strcasecmp(&extension[1], filter)) num_files++;
+    }
+  }
+  return -1;
+}
+
+// Return name of file (matching extension) by index (0 to numFiles()-1)
+char *filenameByIndex(const char *path, const char *filter, int16_t index) {
+  static char filename[256]; // Must be static, we return a pointer to this!
+  File entry, dir = filesys.open(path);
+  if (!dir) return NULL;
+  while(entry = dir.openNextFile()) {
+    entry.getName(filename, sizeof(filename) - 1);
+    entry.close();
+    if(!entry.isDirectory() &&       // Skip directories
+       strncmp(filename, "._", 2)) { // and Mac junk files
+      char *extension = strrchr(filename, '.');
+      if (extension  && !strcasecmp(&extension[1], filter)) {
+        if(!index--) {
+          return filename;
+        }
+      }
+    }
+  }
+  return NULL;
+}
+
+// SETUP FUNCTION - RUNS ONCE AT STARTUP -----------------------------------
+
+void setup() {
+  pinMode(LED_BUILTIN, OUTPUT);
+#if defined(BACK_BUTTON)
+  pinMode(BACK_BUTTON, INPUT_PULLUP);
+#endif
+#if defined(NEXT_BUTTON)
+  pinMode(NEXT_BUTTON, INPUT_PULLUP);
+#endif
+
+  // USB mass storage / filesystem setup (do BEFORE Serial init)
+  flash.begin();
+  // Set disk vendor id, product id and revision
+  usb_msc.setID("Adafruit", "External Flash", "1.0");
+  // Set disk size, block size is 512 regardless of spi flash page size
+  usb_msc.setCapacity(flash.pageSize() * flash.numPages() / 512, 512);
+  usb_msc.setReadWriteCallback(msc_read_cb, msc_write_cb, msc_flush_cb);
+  usb_msc.setUnitReady(true); // MSC is ready for read/write
+  usb_msc.begin();
+  filesys.begin(&flash); // Start filesystem on the flash
+
+  Serial.begin(115200);
+  //while (!Serial);
+
+  // Protomatter (RGB matrix) setup
+  ProtomatterStatus status = matrix.begin();
+  Serial.print("Protomatter begin() status: ");
+  Serial.println((int)status);
+  matrix.fillScreen(0);
+  matrix.show();
+
+  // GIF setup
+  GIF.begin(LITTLE_ENDIAN_PIXELS);
+}
+
+// LOOP FUNCTION - RUNS REPEATEDLY UNTIL RESET / POWER OFF -----------------
+
+int16_t GIFindex = -1;     // Current file index in GIFpath
+int8_t GIFincrement = 1;   // +1 = next GIF, -1 = prev, 0 = same
+uint32_t GIFstartTime = 0; // When current GIF started playing
+bool GIFisOpen = false;    // True if GIF is currently open
+
+void loop() {
+  if (msc_changed) {     // If filesystem has changed...
+    msc_changed = false; // Clear flag
+    GIFincrement = 1;    // Set index to next file when we resume here
+    return;              // Prioritize USB, handled in calling func
+  }
+
+#if defined(BACK_BUTTON)
+  if(!digitalRead(BACK_BUTTON)) {
+    GIFincrement = -1;                // Back
+    while(!digitalRead(BACK_BUTTON)); // Wait for release
+  }
+#endif
+#if defined(NEXT_BUTTON)
+  if(!digitalRead(NEXT_BUTTON)) {
+    GIFincrement = 1;                 // Forward
+    while(!digitalRead(NEXT_BUTTON)); // Wait for release
+  }
+#endif
+
+  if (GIFincrement) { // Change file?
+    if (GIFisOpen) {  // If currently playing,
+      GIF.close();    // stop it
+      GIFisOpen = false;
+    }
+    GIFindex += GIFincrement; // Fwd or back 1 file
+    GIFincrement = 0;         // Reset increment flag
+    int num_files = numFiles(GIFpath, "GIF");
+    if(GIFindex >= num_files) GIFindex = 0;         // 'Wrap around' file index
+    else if(GIFindex < 0) GIFindex = num_files - 1; // both directions
+
+    char *filename = filenameByIndex(GIFpath, "GIF", GIFindex);
+    if (filename) {
+      char fullname[sizeof GIFpath + 256];
+      sprintf(fullname, "%s/%s", GIFpath, filename); // Absolute path to GIF
+      Serial.printf("Opening file '%s'\n", fullname);
+      if (GIF.open(fullname, GIFOpenFile, GIFCloseFile,
+                   GIFReadFile, GIFSeekFile, GIFDraw)) {
+        matrix.fillScreen(0);
+        Serial.printf("GIF dimensions: %d x %d\n",
+                      GIF.getCanvasWidth(), GIF.getCanvasHeight());
+        xPos = (matrix.width() - GIF.getCanvasWidth()) / 2; // Center on matrix
+        yPos = (matrix.height() - GIF.getCanvasHeight()) / 2;
+        GIFisOpen = true;
+        GIFstartTime = millis();
+      }
+    }
+  } else if(GIFisOpen) {
+    if (GIF.playFrame(true, NULL)) {
+      matrix.show();
+    } else if ((millis() - GIFstartTime) < (GIFminimumTime * 1000)) {
+      GIF.reset(); // Minimum time hasn't elapsed yet, repeat this GIF
+    } else {
+      GIFincrement = 1; // Minimum time has elapsed, proceed to next GIF
+    }
+  }
+}

examples/doublebuffer/doublebuffer.ino

@@ -41,6 +41,26 @@ PA05 A4    PA13       PA21 D7    PB05       PB13
 PA06       PA14       PA22 SDA   PB06       PB14
 PA07 D9    PA15 D5    PA23 SCL   PB07       PB15
 
+FEATHER nRF52840:
+P0.00      P0.08 D12       P0.24 RXD  P1.08 D5
+P0.01      P0.09           P0.25 TXD  P1.09 D13
+P0.02 A4   P0.10 D2 (NFC)  P0.26 D9   P1.10
+P0.03 A5   P0.11 SCL       P0.27 D10  P1.11
+P0.04 A0   P0.12 SDA       P0.28 A3   P1.12
+P0.05 A1   P0.13 MOSI      P0.29      P1.13
+P0.06 D11  P0.14 SCK       P0.30 A2   P1.14
+P0.07 D6   P0.15 MISO      P0.31      P1.15
+
+FEATHER ESP32:
+P0.00          P0.08          P0.16 16/RX    P0.24          P1.00 32/A7
+P0.01          P0.09          P0.17 17/TX    P0.25 25/A1    P1.01 33/A9/SS
+P0.02          P0.10          P0.18 18/MOSI  P0.26 26/A0    P1.02 34/A2 (in)
+P0.03          P0.11          P0.19 19/MISO  P0.27 27/A10   P1.03
+P0.04 4/A5     P0.12 12/A11   P0.20          P0.28          P1.04 36/A4 (in)
+P0.05 5/SCK    P0.13 13/A12   P0.21 21       P0.29          P1.05
+P0.06          P0.14 14/A6    P0.22 22/SCL   P0.30          P1.06
+P0.07          P0.15 15/A8    P0.23 23/SDA   P0.31          P1.07 39/A3 (in)
+
 RGB Matrix FeatherWing:
 R1  D6    A   A5
 G1  D5    B   A4
@@ -55,6 +75,7 @@ the code could run there (with some work to be done in the convert_*
 functions), but would be super RAM-inefficient. Should be fine on other
 M0 devices like a Metro, if wiring manually so one can pick a contiguous
 byte of PORT bits.
+RGB+clock are on different PORTs on nRF52840.
 */
 
 #if defined(__SAMD51__)
@@ -64,12 +85,41 @@ byte of PORT bits.
   uint8_t clockPin   = 13;
   uint8_t latchPin   = 0;
   uint8_t oePin      = 1;
-#else // SAMD21
+#elif defined(_SAMD21_)
   uint8_t rgbPins[]  = {6, 7, 10, 11, 12, 13};
   uint8_t addrPins[] = {0, 1, 2, 3};
   uint8_t clockPin   = SDA;
   uint8_t latchPin   = 4;
   uint8_t oePin      = 5;
+#elif defined(NRF52_SERIES)
+  // Special nRF52840 FeatherWing pinout
+  uint8_t rgbPins[]  = {6, A5, A1, A0, A4, 11};
+  uint8_t addrPins[] = {10, 5, 13, 9};
+  uint8_t clockPin   = 12;
+  uint8_t latchPin   = PIN_SERIAL1_RX;
+  uint8_t oePin      = PIN_SERIAL1_TX;
+#elif defined(ESP32)
+  // 'Safe' pins (not overlapping any peripherals):
+  // GPIO.out: 4, 12, 13, 14, 15, 21, 27, GPIO.out1: 32, 33
+  // Peripheral-overlapping pins, sorted from 'most expendible':
+  // 16, 17 (RX, TX), 25, 26 (A0, A1), 18, 5, 9 (MOSI, SCK, MISO), 22, 23 (SCL, SDA)
+  uint8_t rgbPins[]  = {4, 12, 13, 14, 15, 21};
+  uint8_t addrPins[] = {16, 17, 25, 26};
+  uint8_t clockPin   = 27; // Must be on same port as rgbPins
+  uint8_t latchPin   = 32;
+  uint8_t oePin      = 33;
+#elif defined(ARDUINO_TEENSY40)
+  uint8_t rgbPins[]  = {15, 16, 17, 20, 21, 22}; // A1-A3, A6-A8, skips SDA,SCL
+  uint8_t addrPins[] = {2, 3, 4, 5};
+  uint8_t clockPin   = 23; // A9
+  uint8_t latchPin   = 6;
+  uint8_t oePin      = 9;
+#elif defined(ARDUINO_TEENSY41)
+  uint8_t rgbPins[]  = {26, 27, 38, 20, 21, 22}; // A12-14, A6-A8 (yes that's a 38, NOT 28!)
+  uint8_t addrPins[] = {2, 3, 4, 5};
+  uint8_t clockPin   = 23; // A9
+  uint8_t latchPin   = 6;
+  uint8_t oePin      = 9;
 #endif
 
 // Last arg here enables double-buffering

examples/pixeldust/pixeldust.ino

@@ -0,0 +1,131 @@
+#include <Wire.h>               // For I2C communication
+#include <Adafruit_LIS3DH.h>
+#include <Adafruit_PixelDust.h> // For simulation
+#include "Adafruit_Protomatter.h"
+
+uint8_t rgbPins[]  = {7, 8, 9, 10, 11, 12};
+uint8_t addrPins[] = {17, 18, 19, 20};
+uint8_t clockPin   = 14;
+uint8_t latchPin   = 15;
+uint8_t oePin      = 16;
+
+Adafruit_LIS3DH accel = Adafruit_LIS3DH();
+
+Adafruit_Protomatter matrix(
+  64, 4, 1, rgbPins, 4, addrPins, clockPin, latchPin, oePin, true);
+
+
+#define WIDTH        64 // Display width in pixels
+#define HEIGHT       32 // Display height in pixels
+#define MAX_FPS      45 // Maximum redraw rate, frames/second
+
+#define N_COLORS 8
+#define BOX_HEIGHT 8
+#define N_GRAINS (BOX_HEIGHT*N_COLORS*8)
+uint16_t colors[N_COLORS];
+
+// Sand object, last 2 args are accelerometer scaling and grain elasticity
+Adafruit_PixelDust sand(WIDTH, HEIGHT, N_GRAINS, 1, 128, false);
+
+
+uint32_t        prevTime   = 0;      // Used for frames-per-second throttle
+
+// SETUP - RUNS ONCE AT PROGRAM START --------------------------------------
+
+void err(int x) {
+  uint8_t i;
+  pinMode(LED_BUILTIN, OUTPUT);       // Using onboard LED
+  for(i=1;;i++) {                     // Loop forever...
+    digitalWrite(LED_BUILTIN, i & 1); // LED on/off blink to alert user
+    delay(x);
+  }
+}
+
+void setup(void) {
+  uint8_t i, j, bytes;
+  Serial.begin(115200);
+  //while (!Serial) delay(10);
+
+  ProtomatterStatus status = matrix.begin();
+  Serial.printf("Protomatter begin() status: %d\n", status);
+
+  if  (!sand.begin()) {
+    Serial.println("Couldn't start sand");
+    err(1000); // Slow blink = malloc error
+  }
+
+  if  (!accel.begin(0x19)) {
+    Serial.println("Couldn't find accelerometer");
+    err(250);  // Fast bink = I2C error
+  }
+  accel.setRange(LIS3DH_RANGE_4_G);   // 2, 4, 8 or 16 G!
+  
+  //sand.randomize(); // Initialize random sand positions
+
+  // Set up initial sand coordinates, in 8x8 blocks
+  int n = 0;
+  for(int i=0; i<N_COLORS; i++) {
+    int xx = i * WIDTH / N_COLORS;
+    int yy =  HEIGHT - BOX_HEIGHT;
+    for(int y=0; y<BOX_HEIGHT; y++) {
+      for(int x=0; x < WIDTH / N_COLORS; x++) {
+        //Serial.printf("#%d -> (%d, %d)\n", n,  xx + x, yy + y);
+        sand.setPosition(n++, xx + x, yy + y);
+      }
+    }
+  }
+  Serial.printf("%d total pixels\n", n);
+
+  colors[0] = color565(64, 64, 64);   // Dark Gray
+  colors[1] = color565(120, 79, 23);   // Brown
+  colors[2] = color565(228,  3,  3);   // Red
+  colors[3] = color565(255,140,  0);   // Orange
+  colors[4] = color565(255,237,  0);   // Yellow
+  colors[5] = color565(  0,128, 38);   // Green
+  colors[6] = color565(  0, 77,255);   // Blue
+  colors[7] = color565(117,  7,135); // Purple  
+}
+
+uint16_t color565(uint8_t red, uint8_t green, uint8_t blue) {
+  return ((red & 0xF8) << 8) | ((green & 0xFC) << 3) | (blue >> 3);
+}
+// MAIN LOOP - RUNS ONCE PER FRAME OF ANIMATION ----------------------------
+
+void loop() {
+  // Limit the animation frame rate to MAX_FPS.  Because the subsequent sand
+  // calculations are non-deterministic (don't always take the same amount
+  // of time, depending on their current states), this helps ensure that
+  // things like gravity appear constant in the simulation.
+  uint32_t t;
+  while(((t = micros()) - prevTime) < (1000000L / MAX_FPS));
+  prevTime = t;
+  
+  // Read accelerometer...
+  sensors_event_t event;
+  accel.getEvent(&event);
+  //Serial.printf("(%0.1f, %0.1f, %0.1f)\n", event.acceleration.x, event.acceleration.y, event.acceleration.z);
+
+  double xx, yy, zz;
+  xx = event.acceleration.x * 1000;
+  yy = event.acceleration.y * 1000;
+  zz = event.acceleration.z * 1000;
+
+  // Run one frame of the simulation
+  sand.iterate(xx, yy, zz);
+  
+  //sand.iterate(-accel.y, accel.x, accel.z);
+
+  // Update pixel data in LED driver
+  dimension_t x, y;
+  matrix.fillScreen(0x0);
+  for(int i=0; i<N_GRAINS ; i++) {
+    sand.getPosition(i, &x, &y);
+    int n = i / ((WIDTH / N_COLORS) * BOX_HEIGHT); // Color index
+    uint16_t flakeColor = colors[n];
+    matrix.drawPixel(x, y, flakeColor);
+    //Serial.printf("(%d, %d)\n", x, y);
+  }
+  matrix.show(); // Copy data to matrix buffers
+  
+
+}

examples/pixeldust_64x64/pixeldust_64x64.ino

@@ -0,0 +1,131 @@
+#include <Wire.h>               // For I2C communication
+#include <Adafruit_LIS3DH.h>
+#include <Adafruit_PixelDust.h> // For simulation
+#include "Adafruit_Protomatter.h"
+
+uint8_t rgbPins[]  = {7, 8, 9, 10, 11, 12};
+uint8_t addrPins[] = {17, 18, 19, 20, 21};
+uint8_t clockPin   = 14;
+uint8_t latchPin   = 15;
+uint8_t oePin      = 16;
+
+Adafruit_LIS3DH accel = Adafruit_LIS3DH();
+
+Adafruit_Protomatter matrix(
+  64, 4, 1, rgbPins, 5, addrPins, clockPin, latchPin, oePin, true);
+
+
+#define WIDTH        64 // Display width in pixels
+#define HEIGHT       64 // Display height in pixels
+#define MAX_FPS      45 // Maximum redraw rate, frames/second
+
+#define N_COLORS 8
+#define BOX_HEIGHT 8
+#define N_GRAINS (BOX_HEIGHT*N_COLORS*8)
+uint16_t colors[N_COLORS];
+
+// Sand object, last 2 args are accelerometer scaling and grain elasticity
+Adafruit_PixelDust sand(WIDTH, HEIGHT, N_GRAINS, 1, 128, false);
+
+
+uint32_t        prevTime   = 0;      // Used for frames-per-second throttle
+
+// SETUP - RUNS ONCE AT PROGRAM START --------------------------------------
+
+void err(int x) {
+  uint8_t i;
+  pinMode(LED_BUILTIN, OUTPUT);       // Using onboard LED
+  for(i=1;;i++) {                     // Loop forever...
+    digitalWrite(LED_BUILTIN, i & 1); // LED on/off blink to alert user
+    delay(x);
+  }
+}
+
+void setup(void) {
+  uint8_t i, j, bytes;
+  Serial.begin(115200);
+  //while (!Serial) delay(10);
+
+  ProtomatterStatus status = matrix.begin();
+  Serial.printf("Protomatter begin() status: %d\n", status);
+
+  if  (!sand.begin()) {
+    Serial.println("Couldn't start sand");
+    err(1000); // Slow blink = malloc error
+  }
+
+  if  (!accel.begin(0x19)) {
+    Serial.println("Couldn't find accelerometer");
+    err(250);  // Fast bink = I2C error
+  }
+  accel.setRange(LIS3DH_RANGE_4_G);   // 2, 4, 8 or 16 G!
+  
+  //sand.randomize(); // Initialize random sand positions
+
+  // Set up initial sand coordinates, in 8x8 blocks
+  int n = 0;
+  for(int i=0; i<N_COLORS; i++) {
+    int xx = i * WIDTH / N_COLORS;
+    int yy =  HEIGHT - BOX_HEIGHT;
+    for(int y=0; y<BOX_HEIGHT; y++) {
+      for(int x=0; x < WIDTH / N_COLORS; x++) {
+        //Serial.printf("#%d -> (%d, %d)\n", n,  xx + x, yy + y);
+        sand.setPosition(n++, xx + x, yy + y);
+      }
+    }
+  }
+  Serial.printf("%d total pixels\n", n);
+
+  colors[0] = color565(64, 64, 64);   // Dark Gray
+  colors[1] = color565(120, 79, 23);   // Brown
+  colors[2] = color565(228,  3,  3);   // Red
+  colors[3] = color565(255,140,  0);   // Orange
+  colors[4] = color565(255,237,  0);   // Yellow
+  colors[5] = color565(  0,128, 38);   // Green
+  colors[6] = color565(  0, 77,255);   // Blue
+  colors[7] = color565(117,  7,135); // Purple  
+}
+
+uint16_t color565(uint8_t red, uint8_t green, uint8_t blue) {
+  return ((red & 0xF8) << 8) | ((green & 0xFC) << 3) | (blue >> 3);
+}
+// MAIN LOOP - RUNS ONCE PER FRAME OF ANIMATION ----------------------------
+
+void loop() {
+  // Limit the animation frame rate to MAX_FPS.  Because the subsequent sand
+  // calculations are non-deterministic (don't always take the same amount
+  // of time, depending on their current states), this helps ensure that
+  // things like gravity appear constant in the simulation.
+  uint32_t t;
+  while(((t = micros()) - prevTime) < (1000000L / MAX_FPS));
+  prevTime = t;
+  
+  // Read accelerometer...
+  sensors_event_t event;
+  accel.getEvent(&event);
+  //Serial.printf("(%0.1f, %0.1f, %0.1f)\n", event.acceleration.x, event.acceleration.y, event.acceleration.z);
+
+  double xx, yy, zz;
+  xx = event.acceleration.x * 1000;
+  yy = event.acceleration.y * 1000;
+  zz = event.acceleration.z * 1000;
+
+  // Run one frame of the simulation
+  sand.iterate(xx, yy, zz);
+  
+  //sand.iterate(-accel.y, accel.x, accel.z);
+
+  // Update pixel data in LED driver
+  dimension_t x, y;
+  matrix.fillScreen(0x0);
+  for(int i=0; i<N_GRAINS ; i++) {
+    sand.getPosition(i, &x, &y);
+    int n = i / ((WIDTH / N_COLORS) * BOX_HEIGHT); // Color index
+    uint16_t flakeColor = colors[n];
+    matrix.drawPixel(x, y, flakeColor);
+    //Serial.printf("(%d, %d)\n", x, y);
+  }
+  matrix.show(); // Copy data to matrix buffers
+  
+
+}

examples/protomatter/protomatter.ino

@@ -41,6 +41,26 @@ PA05 A4    PA13       PA21 D7    PB05       PB13
 PA06       PA14       PA22 SDA   PB06       PB14
 PA07 D9    PA15 D5    PA23 SCL   PB07       PB15
 
+FEATHER nRF52840:
+P0.00      P0.08 D12       P0.24 RXD  P1.08 D5
+P0.01      P0.09           P0.25 TXD  P1.09 D13
+P0.02 A4   P0.10 D2 (NFC)  P0.26 D9   P1.10
+P0.03 A5   P0.11 SCL       P0.27 D10  P1.11
+P0.04 A0   P0.12 SDA       P0.28 A3   P1.12
+P0.05 A1   P0.13 MOSI      P0.29      P1.13
+P0.06 D11  P0.14 SCK       P0.30 A2   P1.14
+P0.07 D6   P0.15 MISO      P0.31      P1.15
+
+FEATHER ESP32:
+P0.00          P0.08          P0.16 16/RX    P0.24          P1.00 32/A7
+P0.01          P0.09          P0.17 17/TX    P0.25 25/A1    P1.01 33/A9/SS
+P0.02          P0.10          P0.18 18/MOSI  P0.26 26/A0    P1.02 34/A2 (in)
+P0.03          P0.11          P0.19 19/MISO  P0.27 27/A10   P1.03
+P0.04 4/A5     P0.12 12/A11   P0.20          P0.28          P1.04 36/A4 (in)
+P0.05 5/SCK    P0.13 13/A12   P0.21 21       P0.29          P1.05
+P0.06          P0.14 14/A6    P0.22 22/SCL   P0.30          P1.06
+P0.07          P0.15 15/A8    P0.23 23/SDA   P0.31          P1.07 39/A3 (in)
+
 RGB Matrix FeatherWing:
 R1  D6    A   A5
 G1  D5    B   A4
@@ -55,6 +75,7 @@ the code could run there (with some work to be done in the convert_*
 functions), but would be super RAM-inefficient. Should be fine on other
 M0 devices like a Metro, if wiring manually so one can pick a contiguous
 byte of PORT bits.
+RGB+clock are on different PORTs on nRF52840.
 */
 
 #if defined(__SAMD51__)
@@ -64,12 +85,41 @@ byte of PORT bits.
   uint8_t clockPin   = 13;
   uint8_t latchPin   = 0;
   uint8_t oePin      = 1;
-#else // SAMD21
+#elif defined(_SAMD21_)
   uint8_t rgbPins[]  = {6, 7, 10, 11, 12, 13};
   uint8_t addrPins[] = {0, 1, 2, 3};
   uint8_t clockPin   = SDA;
   uint8_t latchPin   = 4;
   uint8_t oePin      = 5;
+#elif defined(NRF52_SERIES)
+  // Special nRF52840 FeatherWing pinout
+  uint8_t rgbPins[]  = {6, A5, A1, A0, A4, 11};
+  uint8_t addrPins[] = {10, 5, 13, 9};
+  uint8_t clockPin   = 12;
+  uint8_t latchPin   = PIN_SERIAL1_RX;
+  uint8_t oePin      = PIN_SERIAL1_TX;
+#elif defined(ESP32)
+  // 'Safe' pins (not overlapping any peripherals):
+  // GPIO.out: 4, 12, 13, 14, 15, 21, 27, GPIO.out1: 32, 33
+  // Peripheral-overlapping pins, sorted from 'most expendible':
+  // 16, 17 (RX, TX), 25, 26 (A0, A1), 18, 5, 9 (MOSI, SCK, MISO), 22, 23 (SCL, SDA)
+  uint8_t rgbPins[]  = {4, 12, 13, 14, 15, 21};
+  uint8_t addrPins[] = {16, 17, 25, 26};
+  uint8_t clockPin   = 27; // Must be on same port as rgbPins
+  uint8_t latchPin   = 32;
+  uint8_t oePin      = 33;
+#elif defined(ARDUINO_TEENSY40)
+  uint8_t rgbPins[]  = {15, 16, 17, 20, 21, 22}; // A1-A3, A6-A8, skips SDA,SCL
+  uint8_t addrPins[] = {2, 3, 4, 5};
+  uint8_t clockPin   = 23; // A9
+  uint8_t latchPin   = 6;
+  uint8_t oePin      = 9;
+#elif defined(ARDUINO_TEENSY41)
+  uint8_t rgbPins[]  = {26, 27, 38, 20, 21, 22}; // A12-14, A6-A8 (yes that's a 38, NOT 28!)
+  uint8_t addrPins[] = {2, 3, 4, 5};
+  uint8_t clockPin   = 23; // A9
+  uint8_t latchPin   = 6;
+  uint8_t oePin      = 9;
 #endif
 
 Adafruit_Protomatter matrix(

library.properties

@@ -1,10 +1,10 @@
 name=Adafruit Protomatter
-version=0.0.0
+version=1.0.5
 author=Adafruit
 maintainer=Adafruit <info@adafruit.com>
-sentence=This is a library for the Adafruit RGB LED matrix.
+sentence=A library for Adafruit RGB LED matrices.
 paragraph=RGB LED matrix.
 category=Display
 url=https://github.com/adafruit/Adafruit_protomatter
-architectures=*
-depends=Adafruit GFX Library
+architectures=samd,nrf52,stm32,esp32
+depends=Adafruit GFX Library, Adafruit LIS3DH, Adafruit PixelDust, AnimatedGIF, Adafruit SPIFlash, Adafruit TinyUSB Library

src/Adafruit_Protomatter.cpp

@@ -0,0 +1,89 @@
+/*!
+ * @file Adafruit_Protomatter.cpp
+ *
+ * @mainpage Adafruit Protomatter RGB LED matrix library.
+ *
+ * @section intro_sec Introduction
+ *
+ * This is documentation for Adafruit's protomatter library for HUB75-style
+ * RGB LED matrices. It is designed to work with various matrices sold by
+ * Adafruit ("HUB75" is a vague term and other similar matrices are not
+ * guaranteed to work). This file is the Arduino-specific calls; the
+ * underlying C code is more platform-neutral.
+ *
+ * Adafruit invests time and resources providing this open source code,
+ * please support Adafruit and open-source hardware by purchasing products
+ * from Adafruit!
+ *
+ * @section dependencies Dependencies
+ *
+ * This library depends on
+ * <a href="https://github.com/adafruit/Adafruit-GFX-Library">Adafruit_GFX</a>
+ * being present on your system. Please make sure you have installed the
+ * latest version before using this library.
+ *
+ * @section author Author
+ *
+ * Written by Phil "Paint Your Dragon" Burgess and Jeff Epler for
+ * Adafruit Industries, with contributions from the open source community.
+ *
+ * @section license License
+ *
+ * BSD license, all text here must be included in any redistribution.
+ *
+ */
+
+// Arduino-specific wrapper for the Protomatter C library (provides
+// constructor and so forth, builds on Adafruit_GFX). There should
+// not be any device-specific #ifdefs here. See notes in core.c and
+// arch/arch.h regarding portability.
+
+#include "Adafruit_Protomatter.h" // Also includes core.h & Adafruit_GFX.h
+
+extern Protomatter_core *_PM_protoPtr; ///< In core.c (via arch.h)
+
+Adafruit_Protomatter::Adafruit_Protomatter(uint16_t bitWidth, uint8_t bitDepth,
+                                           uint8_t rgbCount, uint8_t *rgbList,
+                                           uint8_t addrCount, uint8_t *addrList,
+                                           uint8_t clockPin, uint8_t latchPin,
+                                           uint8_t oePin, bool doubleBuffer,
+                                           void *timer)
+    : GFXcanvas16(bitWidth,
+                  (2 << min((int)addrCount, 5)) * min((int)rgbCount, 5)) {
+  if (bitDepth > 6)
+    bitDepth = 6; // GFXcanvas16 color limit (565)
+
+  // Arguments are passed through to the C _PM_init() function which does
+  // some input validation and minor allocation. Return value is ignored
+  // because we can't really do anything about it in a C++ constructor.
+  // The class begin() function checks rgbPins for NULL to determine
+  // whether to proceed or indicate an error.
+  (void)_PM_init(&core, bitWidth, bitDepth, rgbCount, rgbList, addrCount,
+                 addrList, clockPin, latchPin, oePin, doubleBuffer, timer);
+}
+
+Adafruit_Protomatter::~Adafruit_Protomatter(void) {
+  _PM_deallocate(&core);
+  _PM_protoPtr = NULL;
+}
+
+ProtomatterStatus Adafruit_Protomatter::begin(void) {
+  _PM_protoPtr = &core;
+  return _PM_begin(&core);
+}
+
+// Transfer data from GFXcanvas16 to the matrix framebuffer's weird
+// internal format. The actual conversion functions referenced below
+// are in core.c, reasoning is explained there.
+void Adafruit_Protomatter::show(void) {
+  _PM_convert_565(&core, getBuffer(), WIDTH);
+  _PM_swapbuffer_maybe(&core);
+}
+
+// Returns current value of frame counter and resets its value to zero.
+// Two calls to this, timed one second apart (or use math with other
+// intervals), can be used to get a rough frames-per-second value for
+// the matrix (since this is difficult to estimate beforehand).
+uint32_t Adafruit_Protomatter::getFrameCount(void) {
+  return _PM_getFrameCount(_PM_protoPtr);
+}

src/Adafruit_Protomatter.h

@@ -0,0 +1,102 @@
+// Arduino-specific header, accompanies Adafruit_Protomatter.cpp.
+// There should not be any device-specific #ifdefs here.
+
+#pragma once
+
+#include "core.h"
+#include <Adafruit_GFX.h>
+
+/*!
+    @brief  Class representing the Arduino-facing side of the Protomatter
+            library. Subclass of Adafruit_GFX's GFXcanvas16 to allow all
+            the drawing operations.
+*/
+class Adafruit_Protomatter : public GFXcanvas16 {
+public:
+  /*!
+    @brief  Adafruit_Protomatter constructor.
+    @param  bitWidth      Total width of RGB matrix chain, in pixels.
+                          Usu. some multiple of 32, but maybe exceptions.
+    @param  bitDepth      Color "depth" in bitplanes, determines range of
+                          shades of red, green and blue. e.g. passing 4
+                          bits = 16 shades ea. R,G,B = 16x16x16 = 4096
+                          colors. Max is 6, since the GFX library works
+                          with "565" RGB colors (6 bits green, 5 red/blue).
+    @param  rgbCount      Number of "sets" of RGB data pins, each set
+                          containing 6 pins (2 ea. R,G,B). Typically 1,
+                          indicating a single matrix (or matrix chain).
+                          In theory (but not yet extensively tested),
+                          multiple sets of pins can be driven in parallel,
+                          up to 5 on some devices (if the hardware design
+                          provides all those bits on one PORT).
+    @param  rgbList       A uint8_t array of pins (Arduino pin numbering),
+                          6X the prior rgbCount value, corresponding to
+                          the 6 output color bits for a matrix (or chain).
+                          Order is upper-half red, green, blue, lower-half
+                          red, green blue (repeat for each add'l chain).
+                          All the RGB pins (plus the clock pin below on
+                          some architectures) MUST be on the same PORT
+                          register. It's recommended (but not required)
+                          that all RGB pins (and clock depending on arch)
+                          be within the same byte of a PORT (but do not
+                          need to be sequential or contiguous within that
+                          byte) for more efficient RAM utilization. For
+                          two concurrent chains, same principle but 16-bit
+                          word instead of byte.
+    @param  addrCount     Number of row address lines required of matrix.
+                          Total pixel height is then 2 x 2^addrCount, e.g.
+                          32-pixel-tall matrices have 4 row address lines.
+    @param  addrList      A uint8_t array of pins (Arduino pin numbering),
+                          one per row address line.
+    @param  clockPin      RGB clock pin (Arduino pin #).
+    @param  latchPin      RGB data latch pin (Arduino pin #).
+    @param  oePin         Output enable pin (Arduino pin #), active low.
+    @param  doubleBuffer  If true, two matrix buffers are allocated,
+                          so changing display contents doesn't introduce
+                          artifacts mid-conversion. Requires ~2X RAM.
+    @param  timer         Pointer to timer peripheral or timer-related
+                          struct (architecture-dependent), or NULL to
+                          use a default timer ID (also arch-dependent).
+  */
+  Adafruit_Protomatter(uint16_t bitWidth, uint8_t bitDepth, uint8_t rgbCount,
+                       uint8_t *rgbList, uint8_t addrCount, uint8_t *addrList,
+                       uint8_t clockPin, uint8_t latchPin, uint8_t oePin,
+                       bool doubleBuffer, void *timer = NULL);
+  ~Adafruit_Protomatter(void);
+
+  /*!
+    @brief  Start a Protomatter matrix display running -- initialize
+            pins, timer and interrupt into existence.
+    @return A ProtomatterStatus status, one of:
+            PROTOMATTER_OK if everything is good.
+            PROTOMATTER_ERR_PINS if data and/or clock pins are split
+            across different PORTs.
+            PROTOMATTER_ERR_MALLOC if insufficient RAM to allocate
+            display memory.
+            PROTOMATTER_ERR_ARG if a bad value was passed to the
+            constructor.
+  */
+  ProtomatterStatus begin(void);
+
+  /*!
+    @brief Process data from GFXcanvas16 to the matrix framebuffer's
+           internal format for display.
+  */
+  void show(void);
+
+  /*!
+    @brief  Returns current value of frame counter and resets its value
+            to zero. Two calls to this, timed one second apart (or use
+            math with other intervals), can be used to get a rough
+            frames-per-second value for the matrix (since this is
+            difficult to estimate beforehand).
+    @return Frame count since previous call to function, as a uint32_t.
+  */
+  uint32_t getFrameCount(void);
+
+private:
+  Protomatter_core core;             // Underlying C struct
+  void convert_byte(uint8_t *dest);  // GFXcanvas16-to-matrix
+  void convert_word(uint16_t *dest); // conversion functions
+  void convert_long(uint32_t *dest); // for 8/16/32 bit bufs
+};

src/arch/arch.h

@@ -0,0 +1,208 @@
+/*!
+ * @file arch.h
+ *
+ * Part of Adafruit's Protomatter library for HUB75-style RGB LED matrices.
+ * This file establishes some very low-level things and includes headers
+ * specific to each supported device. This should ONLY be included by
+ * core.c, nowhere else. Ever.
+ *
+ * Adafruit invests time and resources providing this open source code,
+ * please support Adafruit and open-source hardware by purchasing
+ * products from Adafruit!
+ *
+ * Written by Phil "Paint Your Dragon" Burgess and Jeff Epler for
+ * Adafruit Industries, with contributions from the open source community.
+ *
+ * BSD license, all text here must be included in any redistribution.
+ *
+ */
+
+#pragma once
+
+#include <string.h>
+
+/*
+Common ground for architectures to support this library:
+
+- 32-bit device (e.g. ARM core, ESP32, potentially others in the future)
+- One or more 32-bit GPIO PORTs with atomic bitmask SET and CLEAR registers.
+  A TOGGLE register, if present, may improve performance but is NOT required.
+- Tolerate 8-bit or word-aligned 16-bit accesses within the 32-bit PORT
+  registers (e.g. writing just one of four bytes, rather than the whole
+  32 bits). The library does not use any unaligned accesses (i.e. the
+  "middle word" of a 32-bit register), even if a device tolerates such.
+
+"Pin" as used in this code is always a uint8_t value, but the semantics
+of what it means may vary between Arduino and non-Arduino situations.
+In Arduino, it's the pin index one would pass to functions such as
+digitalWrite(), and doesn't necessarily correspond to physical hardware
+pins or any other arrangement. Some may have names like 'A0' that really
+just map to higher indices.
+In non-Arduino settings (CircuitPython, other languages, etc.), how a
+pin index relates to hardware is entirely implementation dependent, and
+how to get from one to the other is what must be implemented in this file.
+Quite often an environment will follow the Arduino pin designations
+(since the numbers are on a board's silkscreen) and will have an internal
+table mapping those indices to registers and bitmasks...but probably not
+an identically-named and -structured table to the Arduino code, hence the
+reason for many "else" situations in this code.
+
+Each architecture defines the following macros and/or functions (the _PM_
+prefix on each is to reduce likelihood of naming collisions...especially
+on ESP32, which has some similarly-named timer functions:
+
+GPIO-related macros/functions:
+
+_PM_portOutRegister(pin):    Get address of PORT out register. Code calling
+                             this can cast it to whatever type's needed.
+_PM_portSetRegister(pin):    Get address of PORT set-bits register.
+_PM_portClearRegister(pin):  Get address of PORT clear-bits register.
+_PM_portToggleRegister(pin): Get address of PORT toggle-bits register.
+                             Not all devices support this, in which case
+                             it must be left undefined.
+_PM_portBitMask(pin):        Get bit mask within PORT register corresponding
+                             to a pin number. When compiling for Arduino,
+                             this just maps to digitalPinToBitMask(), other
+                             environments will need an equivalent.
+_PM_byteOffset(pin):         Get index of byte (0 to 3) within 32-bit PORT
+                             corresponding to a pin number.
+_PM_wordOffset(pin):         Get index of word (0 or 1) within 32-bit PORT
+                             corresponding to a pin number.
+_PM_pinOutput(pin):          Set a pin to output mode. In Arduino this maps
+                             to pinMode(pin, OUTPUT). Other environments
+                             will need an equivalent.
+_PM_pinInput(pin):           Set a pin to input mode, no pullup. In Arduino
+                             this maps to pinMode(pin, INPUT).
+_PM_pinHigh(pin):            Set an output pin to a high or 1 state. In
+                             Arduino this maps to digitalWrite(pin, HIGH).
+_PM_pinLow(pin):             Set an output pin to a low or 0 state. In
+                             Arduino this maps to digitalWrite(pin, LOW).
+
+Timer-related macros/functions:
+
+_PM_timerFreq:               A numerical constant - the source clock rate
+                             (in Hz) that's fed to the timer peripheral.
+_PM_timerInit(void*):        Initialize (but do not start) timer.
+_PM_timerStart(void*,count): (Re)start timer for a given timer-tick interval.
+_PM_timerStop(void*):        Stop timer, return current timer counter value.
+_PM_timerGetCount(void*):    Get current timer counter value (whether timer
+                             is running or stopped).
+A timer interrupt service routine is also required, syntax for which varies
+between architectures.
+The void* argument passed to the timer functions is some indeterminate type
+used to uniquely identify a timer peripheral within a given environment. For
+example, in the Arduino wrapper for this library, compiling for SAMD chips,
+it's just a pointer directly to a timer/counter peripheral base address. If
+an implementation needs more data associated alongside a peripheral, this
+could instead be a pointer to a struct, or an integer index.
+
+Other macros/functions:
+
+_PM_chunkSize:               Matrix bitmap width (both in RAM and as issued
+                             to the device) is rounded up (if necessary) to
+                             a multiple of this value as a way of explicitly
+                             unrolling the innermost data-stuffing loops.
+                             So far all HUB75 displays I've encountered are
+                             a multiple of 32 pixels wide, but in case
+                             something new comes along, or if a larger
+                             unroll actually decreases performance due to
+                             cache size, this can be set to whatever works
+                             best (any additional data is simply shifted
+                             out the other end of the matrix). Default if
+                             unspecified is 8 (e.g. four loop passes on a
+                             32-pixel matrix, eight if 64-pixel). Only
+                             certain chunkSizes are actually implemented,
+                             see .cpp code (avoiding GCC-specific tricks
+                             that would handle arbitrary chunk sizes).
+_PM_delayMicroseconds(us):   Function or macro to delay some number of
+                             microseconds. For Arduino, this just maps to
+                             delayMicroseconds(). Other environments will
+                             need to provide their own or map to an
+                             an equivalent function.
+_PM_clockHoldHigh:           Additional code (typically some number of NOPs)
+                             needed to delay the clock fall after RGB data is
+                             written to PORT. Only required on fast devices.
+                             If left undefined, no delay happens.
+_PM_clockHoldLow:            Additional code (e.g. NOPs) needed to delay
+                             clock rise after writing RGB data to PORT.
+                             No delay if left undefined.
+_PM_minMinPeriod:            Mininum value for the "minPeriod" class member,
+                             so bit-angle-modulation time always doubles with
+                             each bitplane (else lower bits may be the same).
+_PM_allocate:                Memory allocation function, should return a
+                             pointer to a buffer of requested size, aligned
+                             to the architecture's largest native type.
+                             If not defined, malloc() is used.
+_PM_free:                    Corresponding deallocator for _PM_allocate().
+                             If not defined, free() is used.
+*/
+
+// ENVIRONMENT-SPECIFIC DECLARATIONS ---------------------------------------
+
+#if defined(ARDUINO) // COMPILING FOR ARDUINO ------------------------------
+
+#include <Arduino.h> // Pull in all that stuff.
+
+#define _PM_delayMicroseconds(us) delayMicroseconds(us)
+#define _PM_pinOutput(pin) pinMode(pin, OUTPUT)
+#define _PM_pinInput(pin) pinMode(pin, INPUT)
+#define _PM_pinHigh(pin) digitalWrite(pin, HIGH)
+#define _PM_pinLow(pin) digitalWrite(pin, LOW)
+#define _PM_portBitMask(pin) digitalPinToBitMask(pin)
+
+#elif defined(CIRCUITPY) // COMPILING FOR CIRCUITPYTHON --------------------
+
+#include "py/mphal.h"
+#include "shared-bindings/microcontroller/Pin.h"
+
+#define _PM_delayMicroseconds(us) mp_hal_delay_us(us)
+
+// No #else here. In non-Arduino case, declare things in the arch-specific
+// files below...unless other environments provide device-neutral functions
+// as above, in which case those could go here (w/#elif).
+
+#endif // END CIRCUITPYTHON ------------------------------------------------
+
+// ARCHITECTURE-SPECIFIC HEADERS -------------------------------------------
+
+#include "esp32.h"
+#include "nrf52.h"
+#include "samd-common.h"
+#include "samd21.h"
+#include "samd51.h"
+#include "stm32.h"
+#include "teensy4.h"
+
+// DEFAULTS IF NOT DEFINED ABOVE -------------------------------------------
+
+#if !defined(_PM_chunkSize)
+#define _PM_chunkSize 8 ///< Unroll data-stuffing loop to this size
+#endif
+
+#if !defined(_PM_clockHoldHigh)
+#define _PM_clockHoldHigh ///< Extra cycles (if any) on clock HIGH signal
+#endif
+
+#if !defined(_PM_clockHoldLow)
+#define _PM_clockHoldLow ///< Extra cycles (if any) on clock LOW signal
+#endif
+
+#if !defined(_PM_minMinPeriod)
+#define _PM_minMinPeriod 100 ///< Minimum timer interval for least bit
+#endif
+
+#if !defined(_PM_allocate)
+#define _PM_allocate(x) (malloc((x))) ///< Memory alloc call
+#endif
+
+#if !defined(_PM_free)
+#define _PM_free(x) (free((x))) ///< Corresponding memory free call
+#endif
+
+#if !defined(IRAM_ATTR)
+#define IRAM_ATTR ///< Neutralize ESP32-specific attribute in core.c
+#endif
+
+#if !defined(_PM_PORT_TYPE)
+#define _PM_PORT_TYPE uint32_t ///< PORT register size/type
+#endif

src/arch/esp32.h

@@ -0,0 +1,123 @@
+/*!
+ * @file esp32.h
+ *
+ * Part of Adafruit's Protomatter library for HUB75-style RGB LED matrices.
+ * This file contains ESP32-SPECIFIC CODE.
+ *
+ * Adafruit invests time and resources providing this open source code,
+ * please support Adafruit and open-source hardware by purchasing
+ * products from Adafruit!
+ *
+ * Written by Phil "Paint Your Dragon" Burgess and Jeff Epler for
+ * Adafruit Industries, with contributions from the open source community.
+ *
+ * BSD license, all text here must be included in any redistribution.
+ *
+ */
+
+#pragma once
+
+#if defined(ESP32)
+
+#if defined(ARDUINO) // COMPILING FOR ARDUINO ------------------------------
+
+#include "driver/timer.h"
+
+#define _PM_portOutRegister(pin)                                               \
+  (volatile uint32_t *)((pin < 32) ? &GPIO.out : &GPIO.out1.val)
+
+#define _PM_portSetRegister(pin)                                               \
+  (volatile uint32_t *)((pin < 32) ? &GPIO.out_w1ts : &GPIO.out1_w1ts.val)
+
+#define _PM_portClearRegister(pin)                                             \
+  (volatile uint32_t *)((pin < 32) ? &GPIO.out_w1tc : &GPIO.out1_w1tc.val)
+
+#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+#define _PM_byteOffset(pin) ((pin & 31) / 8)
+#define _PM_wordOffset(pin) ((pin & 31) / 16)
+#else
+#define _PM_byteOffset(pin) (3 - ((pin & 31) / 8))
+#define _PM_wordOffset(pin) (1 - ((pin & 31) / 16))
+#endif
+
+// ESP32 requires a custom PEW declaration (issues one set of RGB color bits
+// followed by clock pulse). Turns out the bit set/clear registers are not
+// actually atomic. If two writes are made in quick succession, the second
+// has no effect. One option is NOPs, other is to write a 0 (no effect) to
+// the opposing register (set vs clear) to synchronize the next write.
+#define PEW                                                                    \
+  *set = *data++;         /* Set RGB data high */                              \
+  *clear_full = 0;        /* ESP32 MUST sync before 2nd 'set' */               \
+  *set_full = clock;      /* Set clock high */                                 \
+  *clear_full = rgbclock; /* Clear RGB data + clock */                         \
+  ///< Bitbang one set of RGB data bits to matrix
+
+// As written, because it's tied to a specific timer right now, the
+// Arduino lib only permits one instance of the Protomatter_core struct,
+// which it sets up when calling begin().
+void *_PM_protoPtr = NULL;
+
+#define _PM_timerFreq 40000000 // 40 MHz (1:2 prescale)
+#define _PM_timerNum 0         // Timer #0 (can be 0-3)
+
+// This is the default aforementioned singular timer. IN THEORY, other
+// timers could be used, IF an Arduino sketch passes the address of its
+// own hw_timer_t* to the Protomatter constructor and initializes that
+// timer using ESP32's timerBegin(). All of the timer-related functions
+// below pass around a handle rather than accessing _PM_esp32timer
+// directly, in case that's ever actually used in the future.
+static hw_timer_t *_PM_esp32timer = NULL;
+#define _PM_TIMER_DEFAULT &_PM_esp32timer
+
+extern IRAM_ATTR void _PM_row_handler(Protomatter_core *core);
+
+// Timer interrupt handler. This, _PM_row_handler() and any functions
+// called by _PM_row_handler() should all have the IRAM_ATTR attribute
+// (RAM-resident functions). This isn't really the ISR itself, but a
+// callback invoked by the real ISR (in arduino-esp32's esp32-hal-timer.c)
+// which takes care of interrupt status bits & such.
+IRAM_ATTR static void _PM_esp32timerCallback(void) {
+  _PM_row_handler(_PM_protoPtr); // In core.c
+}
+
+// Initialize, but do not start, timer.
+void _PM_timerInit(void *tptr) {
+  hw_timer_t **timer = (hw_timer_t **)tptr; // pointer-to-pointer
+  if (timer == _PM_TIMER_DEFAULT) {
+    *timer = timerBegin(_PM_timerNum, 2, true); // 1:2 prescale, count up
+  }
+  timerAttachInterrupt(*timer, &_PM_esp32timerCallback, true);
+}
+
+// Set timer period, initialize count value to zero, enable timer.
+IRAM_ATTR inline void _PM_timerStart(void *tptr, uint32_t period) {
+  hw_timer_t *timer = *(hw_timer_t **)tptr;
+  timerAlarmWrite(timer, period, true);
+  timerAlarmEnable(timer);
+  timerStart(timer);
+}
+
+// Return current count value (timer enabled or not).
+// Timer must be previously initialized.
+IRAM_ATTR inline uint32_t _PM_timerGetCount(void *tptr) {
+  hw_timer_t *timer = *(hw_timer_t **)tptr;
+  return (uint32_t)timerRead(timer);
+}
+
+// Disable timer and return current count value.
+// Timer must be previously initialized.
+IRAM_ATTR uint32_t _PM_timerStop(void *tptr) {
+  hw_timer_t *timer = *(hw_timer_t **)tptr;
+  timerStop(timer);
+  return _PM_timerGetCount(tptr);
+}
+
+#elif defined(CIRCUITPY) // COMPILING FOR CIRCUITPYTHON --------------------
+
+// ESP32 CircuitPython magic goes here. If any of the above Arduino-specific
+// defines, structs or functions are useful as-is, don't copy them, just
+// move them above the ARDUINO check so fixes/changes carry over, thx.
+
+#endif // END CIRCUITPYTHON ------------------------------------------------
+
+#endif // END ESP32

src/arch/nrf52.h

@@ -0,0 +1,216 @@
+/*!
+ * @file nrf52.h
+ *
+ * Part of Adafruit's Protomatter library for HUB75-style RGB LED matrices.
+ * This file contains NRF52-SPECIFIC CODE.
+ *
+ * Adafruit invests time and resources providing this open source code,
+ * please support Adafruit and open-source hardware by purchasing
+ * products from Adafruit!
+ *
+ * Written by Phil "Paint Your Dragon" Burgess and Jeff Epler for
+ * Adafruit Industries, with contributions from the open source community.
+ *
+ * BSD license, all text here must be included in any redistribution.
+ *
+ */
+
+#pragma once
+
+#if defined(NRF52_SERIES)
+
+#if defined(ARDUINO) // COMPILING FOR ARDUINO ------------------------------
+
+// digitalPinToPort, g_ADigitalPinMap[] are Arduino specific:
+
+void *_PM_portOutRegister(uint32_t pin) {
+  NRF_GPIO_Type *port = digitalPinToPort(pin);
+  return &port->OUT;
+}
+
+void *_PM_portSetRegister(uint32_t pin) {
+  NRF_GPIO_Type *port = digitalPinToPort(pin);
+  return &port->OUTSET;
+}
+
+void *_PM_portClearRegister(uint32_t pin) {
+  NRF_GPIO_Type *port = digitalPinToPort(pin);
+  return &port->OUTCLR;
+}
+
+// Leave _PM_portToggleRegister(pin) undefined on nRF!
+
+#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+#define _PM_byteOffset(pin) ((g_ADigitalPinMap[pin] & 0x1F) / 8)
+#define _PM_wordOffset(pin) ((g_ADigitalPinMap[pin] & 0x1F) / 16)
+#else
+#define _PM_byteOffset(pin) (3 - ((g_ADigitalPinMap[pin] & 0x1F) / 8))
+#define _PM_wordOffset(pin) (1 - ((g_ADigitalPinMap[pin] & 0x1F) / 16))
+#endif
+
+// Because it's tied to a specific timer right now, there can be only
+// one instance of the Protomatter_core struct. The Arduino library
+// sets up this pointer when calling begin().
+void *_PM_protoPtr = NULL;
+
+// Arduino implementation is tied to a specific timer/counter,
+// Partly because IRQs must be declared at compile-time.
+#define _PM_IRQ_HANDLER TIMER4_IRQHandler
+#define _PM_timerFreq 16000000
+#define _PM_TIMER_DEFAULT NRF_TIMER4
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+// Timer interrupt service routine
+void _PM_IRQ_HANDLER(void) {
+  if (_PM_TIMER_DEFAULT->EVENTS_COMPARE[0]) {
+    _PM_TIMER_DEFAULT->EVENTS_COMPARE[0] = 0;
+  }
+  _PM_row_handler(_PM_protoPtr); // In core.c
+}
+
+#ifdef __cplusplus
+}
+#endif
+
+#elif defined(CIRCUITPY) // COMPILING FOR CIRCUITPYTHON --------------------
+
+#include "nrf_gpio.h"
+
+volatile uint32_t *_PM_portOutRegister(uint32_t pin) {
+  NRF_GPIO_Type *port = nrf_gpio_pin_port_decode(&pin);
+  return &port->OUT;
+}
+
+volatile uint32_t *_PM_portSetRegister(uint32_t pin) {
+  NRF_GPIO_Type *port = nrf_gpio_pin_port_decode(&pin);
+  return &port->OUTSET;
+}
+
+volatile uint32_t *_PM_portClearRegister(uint32_t pin) {
+  NRF_GPIO_Type *port = nrf_gpio_pin_port_decode(&pin);
+  return &port->OUTCLR;
+}
+#define _PM_pinOutput(pin)                                                     \
+  nrf_gpio_cfg(pin, NRF_GPIO_PIN_DIR_OUTPUT, NRF_GPIO_PIN_INPUT_DISCONNECT,    \
+               NRF_GPIO_PIN_NOPULL, NRF_GPIO_PIN_H0H1, NRF_GPIO_PIN_NOSENSE)
+#define _PM_pinInput(pin) nrf_gpio_cfg_input(pin)
+#define _PM_pinHigh(pin) nrf_gpio_pin_set(pin)
+#define _PM_pinLow(pin) nrf_gpio_pin_clear(pin)
+#define _PM_portBitMask(pin) (1u << ((pin)&31))
+
+#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+#define _PM_byteOffset(pin) ((pin & 31) / 8)
+#define _PM_wordOffset(pin) ((pin & 31) / 16)
+#else
+#define _PM_byteOffset(pin) (3 - ((pin & 31) / 8))
+#define _PM_wordOffset(pin) (1 - ((pin & 31) / 16))
+#endif
+
+// CircuitPython implementation is tied to a specific freq (but the counter
+// is dynamically allocated):
+#define _PM_timerFreq 16000000
+
+// Because it's tied to a specific timer right now, there can be only
+// one instance of the Protomatter_core struct. The Arduino library
+// sets up this pointer when calling begin().
+void *_PM_protoPtr = NULL;
+
+// Timer interrupt service routine
+void _PM_IRQ_HANDLER(void) {
+  NRF_TIMER_Type *timer = (((Protomatter_core *)_PM_protoPtr)->timer);
+  if (timer->EVENTS_COMPARE[0]) {
+    timer->EVENTS_COMPARE[0] = 0;
+  }
+
+  _PM_row_handler(_PM_protoPtr); // In core.c
+}
+
+#else // END CIRCUITPYTHON -------------------------------------------------
+
+// Byte offset macros, timer and ISR work for other environments go here.
+
+#endif
+
+// CODE COMMON TO ALL ENVIRONMENTS -----------------------------------------
+
+void _PM_timerInit(void *tptr) {
+  static const struct {
+    NRF_TIMER_Type *tc; // -> Timer peripheral base address
+    IRQn_Type IRQn;     // Interrupt number
+  } timer[] = {
+#if defined(NRF_TIMER0)
+    {NRF_TIMER0, TIMER0_IRQn},
+#endif
+#if defined(NRF_TIMER1)
+    {NRF_TIMER1, TIMER1_IRQn},
+#endif
+#if defined(NRF_TIMER2)
+    {NRF_TIMER2, TIMER2_IRQn},
+#endif
+#if defined(NRF_TIMER3)
+    {NRF_TIMER3, TIMER3_IRQn},
+#endif
+#if defined(NRF_TIMER4)
+    {NRF_TIMER4, TIMER4_IRQn},
+#endif
+  };
+#define NUM_TIMERS (sizeof timer / sizeof timer[0])
+
+  // Determine IRQn from timer address
+  uint8_t timerNum = 0;
+  while ((timerNum < NUM_TIMERS) && (timer[timerNum].tc != tptr)) {
+    timerNum++;
+  }
+  if (timerNum >= NUM_TIMERS)
+    return;
+
+  NRF_TIMER_Type *tc = timer[timerNum].tc;
+
+  tc->TASKS_STOP = 1;               // Stop timer
+  tc->MODE = TIMER_MODE_MODE_Timer; // Timer (not counter) mode
+  tc->TASKS_CLEAR = 1;
+  tc->BITMODE = TIMER_BITMODE_BITMODE_16Bit
+                << TIMER_BITMODE_BITMODE_Pos; // 16-bit timer res
+  tc->PRESCALER = 0;                          // 1:1 prescale (16 MHz)
+  tc->INTENSET = TIMER_INTENSET_COMPARE0_Enabled
+                 << TIMER_INTENSET_COMPARE0_Pos; // Event 0 interrupt
+  // NVIC_DisableIRQ(timer[timerNum].IRQn);
+  // NVIC_ClearPendingIRQ(timer[timerNum].IRQn);
+  // NVIC_SetPriority(timer[timerNum].IRQn, 0); // Top priority
+  NVIC_EnableIRQ(timer[timerNum].IRQn);
+}
+
+inline void _PM_timerStart(void *tptr, uint32_t period) {
+  volatile NRF_TIMER_Type *tc = (volatile NRF_TIMER_Type *)tptr;
+  tc->TASKS_STOP = 1;  // Stop timer
+  tc->TASKS_CLEAR = 1; // Reset to 0
+  tc->CC[0] = period;
+  tc->TASKS_START = 1; // Start timer
+}
+
+inline uint32_t _PM_timerGetCount(void *tptr) {
+  volatile NRF_TIMER_Type *tc = (volatile NRF_TIMER_Type *)tptr;
+  tc->TASKS_CAPTURE[0] = 1; // Capture timer to CC[n] register
+  return tc->CC[0];
+}
+
+uint32_t _PM_timerStop(void *tptr) {
+  volatile NRF_TIMER_Type *tc = (volatile NRF_TIMER_Type *)tptr;
+  tc->TASKS_STOP = 1; // Stop timer
+  __attribute__((unused)) uint32_t count = _PM_timerGetCount(tptr);
+  // NOTE TO FUTURE SELF: I don't know why the GetCount code isn't
+  // working. It does the expected thing in a small test program but
+  // not here. I need to get on with testing on an actual matrix, so
+  // this is just a nonsense fudge value for now:
+  return 100;
+  // return count;
+}
+
+#define _PM_clockHoldHigh asm("nop; nop");
+
+#define _PM_minMinPeriod 100
+
+#endif // END NRF52_SERIES

src/arch/samd-common.h

@@ -0,0 +1,98 @@
+/*!
+ * @file samd-common.h
+ *
+ * Part of Adafruit's Protomatter library for HUB75-style RGB LED matrices.
+ * This file contains SAMD-SPECIFIC CODE (SAMD51 & SAMD21).
+ *
+ * Adafruit invests time and resources providing this open source code,
+ * please support Adafruit and open-source hardware by purchasing
+ * products from Adafruit!
+ *
+ * Written by Phil "Paint Your Dragon" Burgess and Jeff Epler for
+ * Adafruit Industries, with contributions from the open source community.
+ *
+ * BSD license, all text here must be included in any redistribution.
+ *
+ */
+
+#pragma once
+
+#if defined(__SAMD51__) || defined(SAMD51) || defined(_SAMD21_) ||             \
+    defined(SAMD21)
+
+#if defined(ARDUINO) // COMPILING FOR ARDUINO ------------------------------
+
+// g_APinDescription[] table and pin indices are Arduino specific:
+#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+#define _PM_byteOffset(pin) (g_APinDescription[pin].ulPin / 8)
+#define _PM_wordOffset(pin) (g_APinDescription[pin].ulPin / 16)
+#else
+#define _PM_byteOffset(pin) (3 - (g_APinDescription[pin].ulPin / 8))
+#define _PM_wordOffset(pin) (1 - (g_APinDescription[pin].ulPin / 16))
+#endif
+
+// Arduino implementation is tied to a specific timer/counter & freq:
+#if defined(TC4)
+#define _PM_TIMER_DEFAULT TC4
+#define _PM_IRQ_HANDLER TC4_Handler
+#else // No TC4 on some M4's
+#define _PM_TIMER_DEFAULT TC3
+#define _PM_IRQ_HANDLER TC3_Handler
+#endif
+#define _PM_timerFreq 48000000
+// Partly because IRQs must be declared at compile-time, and partly
+// because we know Arduino's already set up one of the GCLK sources
+// for 48 MHz.
+
+// Because it's tied to a specific timer right now, there can be only
+// one instance of the Protomatter_core struct. The Arduino library
+// sets up this pointer when calling begin().
+void *_PM_protoPtr = NULL;
+
+// Timer interrupt service routine
+void _PM_IRQ_HANDLER(void) {
+  // Clear overflow flag:
+  _PM_TIMER_DEFAULT->COUNT16.INTFLAG.reg = TC_INTFLAG_OVF;
+  _PM_row_handler(_PM_protoPtr); // In core.c
+}
+
+#elif defined(CIRCUITPY) // COMPILING FOR CIRCUITPYTHON --------------------
+
+#include "hal_gpio.h"
+
+#define _PM_pinOutput(pin) gpio_set_pin_direction(pin, GPIO_DIRECTION_OUT)
+#define _PM_pinInput(pin) gpio_set_pin_direction(pin, GPIO_DIRECTION_IN)
+#define _PM_pinHigh(pin) gpio_set_pin_level(pin, 1)
+#define _PM_pinLow(pin) gpio_set_pin_level(pin, 0)
+#define _PM_portBitMask(pin) (1u << ((pin)&31))
+
+#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+#define _PM_byteOffset(pin) ((pin & 31) / 8)
+#define _PM_wordOffset(pin) ((pin & 31) / 16)
+#else
+#define _PM_byteOffset(pin) (3 - ((pin & 31) / 8))
+#define _PM_wordOffset(pin) (1 - ((pin & 31) / 16))
+#endif
+
+// CircuitPython implementation is tied to a specific freq (but the counter
+// is dynamically allocated):
+#define _PM_timerFreq 48000000
+
+// As currently implemented, there can be only one instance of the
+// Protomatter_core struct. This pointer is set up when starting the matrix.
+void *_PM_protoPtr = NULL;
+
+// Timer interrupt service routine
+void _PM_IRQ_HANDLER(void) {
+  ((Tc *)(((Protomatter_core *)_PM_protoPtr)->timer))->COUNT16.INTFLAG.reg =
+      TC_INTFLAG_OVF;
+  _PM_row_handler(_PM_protoPtr); // In core.c
+}
+
+#else // END CIRCUITPYTHON -------------------------------------------------
+
+// Byte offset macros, timer and ISR work for other environments go here.
+
+#endif
+
+#endif // END SAMD51/SAMD21

src/arch/samd21.h

@@ -0,0 +1,150 @@
+/*!
+ * @file samd21.h
+ *
+ * Part of Adafruit's Protomatter library for HUB75-style RGB LED matrices.
+ * This file contains SAMD21-SPECIFIC CODE.
+ *
+ * Adafruit invests time and resources providing this open source code,
+ * please support Adafruit and open-source hardware by purchasing
+ * products from Adafruit!
+ *
+ * Written by Phil "Paint Your Dragon" Burgess and Jeff Epler for
+ * Adafruit Industries, with contributions from the open source community.
+ *
+ * BSD license, all text here must be included in any redistribution.
+ *
+ */
+
+#pragma once
+
+#if defined(_SAMD21_) || defined(SAMD21) // Arduino, Circuitpy SAMD21 defs
+
+#if defined(ARDUINO) // COMPILING FOR ARDUINO ------------------------------
+
+// g_APinDescription[] table and pin indices are Arduino specific:
+#define _PM_portOutRegister(pin)                                               \
+  &PORT_IOBUS->Group[g_APinDescription[pin].ulPort].OUT.reg
+
+#define _PM_portSetRegister(pin)                                               \
+  &PORT_IOBUS->Group[g_APinDescription[pin].ulPort].OUTSET.reg
+
+#define _PM_portClearRegister(pin)                                             \
+  &PORT_IOBUS->Group[g_APinDescription[pin].ulPort].OUTCLR.reg
+
+#define _PM_portToggleRegister(pin)                                            \
+  &PORT_IOBUS->Group[g_APinDescription[pin].ulPort].OUTTGL.reg
+
+#else // END ARDUINO -------------------------------------------------------
+
+// Non-Arduino port register lookups go here, if not already declared
+// in samd-common.h.
+
+#endif
+
+// CODE COMMON TO ALL ENVIRONMENTS -----------------------------------------
+
+// Initialize, but do not start, timer
+void _PM_timerInit(void *tptr) {
+  static const struct {
+    Tc *tc;         // -> Timer/counter peripheral base address
+    IRQn_Type IRQn; // Interrupt number
+    uint8_t GCM_ID; // GCLK selection ID
+  } timer[] = {
+#if defined(TC0)
+    {TC0, TC0_IRQn, GCM_TCC0_TCC1},
+#endif
+#if defined(TC1)
+    {TC1, TC1_IRQn, GCM_TCC0_TCC1},
+#endif
+#if defined(TC2)
+    {TC2, TC2_IRQn, GCM_TCC2_TC3},
+#endif
+#if defined(TC3)
+    {TC3, TC3_IRQn, GCM_TCC2_TC3},
+#endif
+#if defined(TC4)
+    {TC4, TC4_IRQn, GCM_TC4_TC5},
+#endif
+  };
+#define NUM_TIMERS (sizeof timer / sizeof timer[0])
+
+  Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
+
+  uint8_t timerNum = 0;
+  while ((timerNum < NUM_TIMERS) && (timer[timerNum].tc != tc)) {
+    timerNum++;
+  }
+  if (timerNum >= NUM_TIMERS)
+    return;
+
+  // Enable GCLK for timer/counter
+  GCLK->CLKCTRL.reg = (uint16_t)(GCLK_CLKCTRL_CLKEN | GCLK_CLKCTRL_GEN_GCLK0 |
+                                 GCLK_CLKCTRL_ID(timer[timerNum].GCM_ID));
+  while (GCLK->STATUS.bit.SYNCBUSY == 1)
+    ;
+
+  // Counter must first be disabled to configure it
+  tc->COUNT16.CTRLA.bit.ENABLE = 0;
+  while (tc->COUNT16.STATUS.bit.SYNCBUSY)
+    ;
+
+  tc->COUNT16.CTRLA.reg =       // Configure timer counter
+      TC_CTRLA_PRESCALER_DIV1 | // 1:1 Prescale
+      TC_CTRLA_WAVEGEN_MFRQ |   // Match frequency generation mode (MFRQ)
+      TC_CTRLA_MODE_COUNT16;    // 16-bit counter mode
+  while (tc->COUNT16.STATUS.bit.SYNCBUSY)
+    ;
+
+  tc->COUNT16.CTRLBCLR.reg = TCC_CTRLBCLR_DIR; // Count up
+  while (tc->COUNT16.STATUS.bit.SYNCBUSY)
+    ;
+
+  // Overflow interrupt
+  tc->COUNT16.INTENSET.reg = TC_INTENSET_OVF;
+
+  NVIC_DisableIRQ(timer[timerNum].IRQn);
+  NVIC_ClearPendingIRQ(timer[timerNum].IRQn);
+  NVIC_SetPriority(timer[timerNum].IRQn, 0); // Top priority
+  NVIC_EnableIRQ(timer[timerNum].IRQn);
+
+  // Timer is configured but NOT enabled by default
+}
+
+// Set timer period, initialize count value to zero, enable timer.
+// Timer must be initialized to 16-bit mode using the init function
+// above, but must be inactive before calling this.
+inline void _PM_timerStart(void *tptr, uint32_t period) {
+  Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
+  tc->COUNT16.COUNT.reg = 0;
+  while (tc->COUNT16.STATUS.bit.SYNCBUSY)
+    ;
+  tc->COUNT16.CC[0].reg = period;
+  while (tc->COUNT16.STATUS.bit.SYNCBUSY)
+    ;
+  tc->COUNT16.CTRLA.bit.ENABLE = 1;
+  while (tc->COUNT16.STATUS.bit.SYNCBUSY)
+    ;
+}
+
+// Return current count value (timer enabled or not).
+// Timer must be previously initialized.
+inline uint32_t _PM_timerGetCount(void *tptr) {
+  Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
+  tc->COUNT16.READREQ.reg = TC_READREQ_RCONT | TC_READREQ_ADDR(0x10);
+  while (tc->COUNT16.STATUS.bit.SYNCBUSY)
+    ;
+  return tc->COUNT16.COUNT.reg;
+}
+
+// Disable timer and return current count value.
+// Timer must be previously initialized.
+inline uint32_t _PM_timerStop(void *tptr) {
+  Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
+  uint32_t count = _PM_timerGetCount(tptr);
+  tc->COUNT16.CTRLA.bit.ENABLE = 0;
+  while (tc->COUNT16.STATUS.bit.SYNCBUSY)
+    ;
+  return count;
+}
+
+#endif // END _SAMD21_ || SAMD21

src/arch/samd51.h

@@ -0,0 +1,215 @@
+/*!
+ * @file samd51.h
+ *
+ * Part of Adafruit's Protomatter library for HUB75-style RGB LED matrices.
+ * This file contains SAMD51-SPECIFIC CODE.
+ *
+ * Adafruit invests time and resources providing this open source code,
+ * please support Adafruit and open-source hardware by purchasing
+ * products from Adafruit!
+ *
+ * Written by Phil "Paint Your Dragon" Burgess and Jeff Epler for
+ * Adafruit Industries, with contributions from the open source community.
+ *
+ * BSD license, all text here must be included in any redistribution.
+ *
+ */
+
+#pragma once
+
+#if defined(__SAMD51__) || defined(SAMD51) // Arduino, Circuitpy SAMD51 defs
+
+#if defined(ARDUINO) // COMPILING FOR ARDUINO ------------------------------
+
+// g_APinDescription[] table and pin indices are Arduino specific:
+#define _PM_portOutRegister(pin)                                               \
+  &PORT->Group[g_APinDescription[pin].ulPort].OUT.reg
+
+#define _PM_portSetRegister(pin)                                               \
+  &PORT->Group[g_APinDescription[pin].ulPort].OUTSET.reg
+
+#define _PM_portClearRegister(pin)                                             \
+  &PORT->Group[g_APinDescription[pin].ulPort].OUTCLR.reg
+
+#define _PM_portToggleRegister(pin)                                            \
+  &PORT->Group[g_APinDescription[pin].ulPort].OUTTGL.reg
+
+#elif defined(CIRCUITPY) // COMPILING FOR CIRCUITPYTHON --------------------
+
+#define _PM_portOutRegister(pin) (&PORT->Group[(pin / 32)].OUT.reg)
+
+#define _PM_portSetRegister(pin) (&PORT->Group[(pin / 32)].OUTSET.reg)
+
+#define _PM_portClearRegister(pin) (&PORT->Group[(pin / 32)].OUTCLR.reg)
+
+#define _PM_portToggleRegister(pin) (&PORT->Group[(pin / 32)].OUTTGL.reg)
+
+#define F_CPU (120000000)
+
+#else
+
+// Other port register lookups go here
+
+#endif
+
+// CODE COMMON TO ALL ENVIRONMENTS -----------------------------------------
+
+// Initialize, but do not start, timer
+void _PM_timerInit(void *tptr) {
+  static const struct {
+    Tc *tc;          // -> Timer/counter peripheral base address
+    IRQn_Type IRQn;  // Interrupt number
+    uint8_t GCLK_ID; // Peripheral channel # for clock source
+  } timer[] = {
+#if defined(TC0)
+    {TC0, TC0_IRQn, TC0_GCLK_ID},
+#endif
+#if defined(TC1)
+    {TC1, TC1_IRQn, TC1_GCLK_ID},
+#endif
+#if defined(TC2)
+    {TC2, TC2_IRQn, TC2_GCLK_ID},
+#endif
+#if defined(TC3)
+    {TC3, TC3_IRQn, TC3_GCLK_ID},
+#endif
+#if defined(TC4)
+    {TC4, TC4_IRQn, TC4_GCLK_ID},
+#endif
+#if defined(TC5)
+    {TC5, TC5_IRQn, TC5_GCLK_ID},
+#endif
+#if defined(TC6)
+    {TC6, TC6_IRQn, TC6_GCLK_ID},
+#endif
+#if defined(TC7)
+    {TC7, TC7_IRQn, TC7_GCLK_ID},
+#endif
+#if defined(TC8)
+    {TC8, TC8_IRQn, TC8_GCLK_ID},
+#endif
+#if defined(TC9)
+    {TC9, TC9_IRQn, TC9_GCLK_ID},
+#endif
+#if defined(TC10)
+    {TC10, TC10_IRQn, TC10_GCLK_ID},
+#endif
+#if defined(TC11)
+    {TC11, TC11_IRQn, TC11_GCLK_ID},
+#endif
+#if defined(TC12)
+    {TC12, TC12_IRQn, TC12_GCLK_ID},
+#endif
+  };
+#define NUM_TIMERS (sizeof timer / sizeof timer[0])
+
+  Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
+
+  uint8_t timerNum = 0;
+  while ((timerNum < NUM_TIMERS) && (timer[timerNum].tc != tc)) {
+    timerNum++;
+  }
+  if (timerNum >= NUM_TIMERS)
+    return;
+
+  // Feed timer/counter off GCLK1 (already set 48 MHz by Arduino core).
+  // Sure, SAMD51 can run timers up to F_CPU (e.g. 120 MHz or up to
+  // 200 MHz with overclocking), but on higher bitplanes (which have
+  // progressively longer timer periods) I could see this possibly
+  // exceeding a 16-bit timer, and would have to switch prescalers.
+  // We don't actually need atomic precision on the timer -- point is
+  // simply that the period doubles with each bitplane, and this can
+  // work fine at 48 MHz.
+  GCLK->PCHCTRL[timer[timerNum].GCLK_ID].bit.CHEN = 0; // Disable
+  while (GCLK->PCHCTRL[timer[timerNum].GCLK_ID].bit.CHEN)
+    ;                        // Wait for it
+  GCLK_PCHCTRL_Type pchctrl; // Read-modify-store
+  pchctrl.reg = GCLK->PCHCTRL[timer[timerNum].GCLK_ID].reg;
+  pchctrl.bit.GEN = GCLK_PCHCTRL_GEN_GCLK1_Val;
+  pchctrl.bit.CHEN = 1;
+  GCLK->PCHCTRL[timer[timerNum].GCLK_ID].reg = pchctrl.reg;
+  while (!GCLK->PCHCTRL[timer[timerNum].GCLK_ID].bit.CHEN)
+    ;
+
+  // Disable timer before configuring it
+  tc->COUNT16.CTRLA.bit.ENABLE = 0;
+  while (tc->COUNT16.SYNCBUSY.bit.ENABLE)
+    ;
+
+  // 16-bit counter mode, 1:1 prescale
+  tc->COUNT16.CTRLA.bit.MODE = TC_CTRLA_MODE_COUNT16;
+  tc->COUNT16.CTRLA.bit.PRESCALER = TC_CTRLA_PRESCALER_DIV1_Val;
+
+  tc->COUNT16.WAVE.bit.WAVEGEN =
+      TC_WAVE_WAVEGEN_MFRQ_Val; // Match frequency generation mode (MFRQ)
+
+  tc->COUNT16.CTRLBCLR.reg = TC_CTRLBCLR_DIR; // Count up
+  while (tc->COUNT16.SYNCBUSY.bit.CTRLB)
+    ;
+
+  // Overflow interrupt
+  tc->COUNT16.INTENSET.reg = TC_INTENSET_OVF;
+
+  NVIC_DisableIRQ(timer[timerNum].IRQn);
+  NVIC_ClearPendingIRQ(timer[timerNum].IRQn);
+  NVIC_SetPriority(timer[timerNum].IRQn, 0); // Top priority
+  NVIC_EnableIRQ(timer[timerNum].IRQn);
+
+  // Timer is configured but NOT enabled by default
+}
+
+// Set timer period, initialize count value to zero, enable timer.
+// Timer must be initialized to 16-bit mode using the init function
+// above, but must be inactive before calling this.
+inline void _PM_timerStart(void *tptr, uint32_t period) {
+  Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
+  tc->COUNT16.COUNT.reg = 0;
+  while (tc->COUNT16.SYNCBUSY.bit.COUNT)
+    ;
+  tc->COUNT16.CC[0].reg = period;
+  while (tc->COUNT16.SYNCBUSY.bit.CC0)
+    ;
+  tc->COUNT16.CTRLA.bit.ENABLE = 1;
+  while (tc->COUNT16.SYNCBUSY.bit.STATUS)
+    ;
+}
+
+// Return current count value (timer enabled or not).
+// Timer must be previously initialized.
+inline uint32_t _PM_timerGetCount(void *tptr) {
+  Tc *tc = (Tc *)tptr;                // Cast peripheral address passed in
+  tc->COUNT16.CTRLBSET.bit.CMD = 0x4; // Sync COUNT
+  while (tc->COUNT16.CTRLBSET.bit.CMD)
+    ; // Wait for command
+  return tc->COUNT16.COUNT.reg;
+}
+
+// Disable timer and return current count value.
+// Timer must be previously initialized.
+uint32_t _PM_timerStop(void *tptr) {
+  Tc *tc = (Tc *)tptr; // Cast peripheral address passed in
+  uint32_t count = _PM_timerGetCount(tptr);
+  tc->COUNT16.CTRLA.bit.ENABLE = 0;
+  while (tc->COUNT16.SYNCBUSY.bit.STATUS)
+    ;
+  return count;
+}
+
+// See notes in core.c before the "blast" functions
+#if F_CPU >= 200000000
+#define _PM_clockHoldHigh asm("nop; nop; nop; nop; nop");
+#define _PM_clockHoldLow asm("nop; nop");
+#elif F_CPU >= 180000000
+#define _PM_clockHoldHigh asm("nop; nop; nop; nop");
+#define _PM_clockHoldLow asm("nop");
+#elif F_CPU >= 150000000
+#define _PM_clockHoldHigh asm("nop; nop; nop");
+#define _PM_clockHoldLow asm("nop");
+#else
+#define _PM_clockHoldHigh asm("nop; nop; nop");
+#define _PM_clockHoldLow asm("nop");
+#endif
+
+#define _PM_minMinPeriod 160
+
+#endif // END __SAMD51__ || SAMD51

src/arch/stm32.h

@@ -0,0 +1,141 @@
+/*!
+ * @file stm32.h
+ *
+ * Part of Adafruit's Protomatter library for HUB75-style RGB LED matrices.
+ * This file contains STM32-SPECIFIC CODE.
+ *
+ * Adafruit invests time and resources providing this open source code,
+ * please support Adafruit and open-source hardware by purchasing
+ * products from Adafruit!
+ *
+ * Written by Phil "Paint Your Dragon" Burgess and Jeff Epler for
+ * Adafruit Industries, with contributions from the open source community.
+ *
+ * BSD license, all text here must be included in any redistribution.
+ *
+ */
+
+#pragma once
+
+#if defined(STM32F4_SERIES) || defined(STM32F405xx) // Arduino, CircuitPy
+
+#if defined(ARDUINO) // COMPILING FOR ARDUINO ------------------------------
+
+// Arduino port register lookups go here, else ones in arch.h are used.
+
+#elif defined(CIRCUITPY) // COMPILING FOR CIRCUITPYTHON --------------------
+
+#include "timers.h"
+
+#undef _PM_portBitMask
+#define _PM_portBitMask(pin) (1u << ((pin)&15))
+#define _PM_byteOffset(pin) ((pin & 15) / 8)
+#define _PM_wordOffset(pin) ((pin & 15) / 16)
+
+#define _PM_pinOutput(pin_)                                                    \
+  do {                                                                         \
+    int8_t pin = (pin_);                                                       \
+    GPIO_InitTypeDef GPIO_InitStruct = {0};                                    \
+    GPIO_InitStruct.Pin = 1 << (pin & 15);                                     \
+    GPIO_InitStruct.Mode = GPIO_MODE_OUTPUT_PP;                                \
+    GPIO_InitStruct.Pull = GPIO_NOPULL;                                        \
+    GPIO_InitStruct.Speed = GPIO_SPEED_FREQ_VERY_HIGH;                         \
+    HAL_GPIO_Init(pin_port(pin / 16), &GPIO_InitStruct);                       \
+  } while (0)
+#define _PM_pinInput(pin_)                                                     \
+  do {                                                                         \
+    int8_t pin = (pin_);                                                       \
+    GPIO_InitTypeDef GPIO_InitStruct = {0};                                    \
+    GPIO_InitStruct.Pin = 1 << (pin & 15);                                     \
+    GPIO_InitStruct.Mode = GPIO_MODE_INPUT;                                    \
+    GPIO_InitStruct.Pull = GPIO_NOPULL;                                        \
+    GPIO_InitStruct.Speed = GPIO_SPEED_FREQ_VERY_HIGH;                         \
+    HAL_GPIO_Init(pin_port(pin / 16), &GPIO_InitStruct);                       \
+  } while (0)
+#define _PM_pinHigh(pin)                                                       \
+  HAL_GPIO_WritePin(pin_port(pin / 16), 1 << (pin & 15), GPIO_PIN_SET)
+#define _PM_pinLow(pin)                                                        \
+  HAL_GPIO_WritePin(pin_port(pin / 16), 1 << (pin & 15), GPIO_PIN_RESET)
+
+#define _PM_PORT_TYPE uint16_t
+
+volatile uint16_t *_PM_portOutRegister(uint32_t pin) {
+  return (uint16_t *)&pin_port(pin / 16)->ODR;
+}
+
+volatile uint16_t *_PM_portSetRegister(uint32_t pin) {
+  return (uint16_t *)&pin_port(pin / 16)->BSRR;
+}
+
+// To make things interesting, STM32F4xx places the set and clear
+// GPIO bits within a single register.  The "clear" bits are upper, so
+// offset by 1 in uint16_ts
+volatile uint16_t *_PM_portClearRegister(uint32_t pin) {
+  return 1 + (uint16_t *)&pin_port(pin / 16)->BSRR;
+}
+
+// TODO: was this somehow specific to TIM6?
+#define _PM_timerFreq 42000000
+
+// Because it's tied to a specific timer right now, there can be only
+// one instance of the Protomatter_core struct. The Arduino library
+// sets up this pointer when calling begin().
+// TODO: this is no longer true, should it change?
+void *_PM_protoPtr = NULL;
+
+STATIC TIM_HandleTypeDef tim_handle;
+
+// Timer interrupt service routine
+void _PM_IRQ_HANDLER(void) {
+  // Clear overflow flag:
+  //_PM_TIMER_DEFAULT->COUNT16.INTFLAG.reg = TC_INTFLAG_OVF;
+  _PM_row_handler(_PM_protoPtr); // In core.c
+}
+
+// Initialize, but do not start, timer
+void _PM_timerInit(void *tptr) {
+  TIM_TypeDef *tim_instance = (TIM_TypeDef *)tptr;
+  stm_peripherals_timer_reserve(tim_instance);
+  // Set IRQs at max priority and start clock
+  stm_peripherals_timer_preinit(tim_instance, 0, _PM_IRQ_HANDLER);
+
+  tim_handle.Instance = tim_instance;
+  tim_handle.Init.Period = 1000; // immediately replaced.
+  tim_handle.Init.Prescaler = 0;
+  tim_handle.Init.ClockDivision = TIM_CLOCKDIVISION_DIV1;
+  tim_handle.Init.CounterMode = TIM_COUNTERMODE_UP;
+  tim_handle.Init.AutoReloadPreload = TIM_AUTORELOAD_PRELOAD_DISABLE;
+
+  HAL_TIM_Base_Init(&tim_handle);
+
+  size_t tim_irq = stm_peripherals_timer_get_irqnum(tim_instance);
+  HAL_NVIC_DisableIRQ(tim_irq);
+  NVIC_ClearPendingIRQ(tim_irq);
+  NVIC_SetPriority(tim_irq, 0); // Top priority
+}
+
+inline void _PM_timerStart(void *tptr, uint32_t period) {
+  TIM_TypeDef *tim = tptr;
+  tim->SR = 0;
+  tim->ARR = period;
+  tim->CR1 |= TIM_CR1_CEN;
+  tim->DIER |= TIM_DIER_UIE;
+  HAL_NVIC_EnableIRQ(stm_peripherals_timer_get_irqnum(tim));
+}
+
+uint32_t _PM_timerStop(void *tptr) {
+  TIM_TypeDef *tim = tptr;
+  HAL_NVIC_DisableIRQ(stm_peripherals_timer_get_irqnum(tim));
+  tim->CR1 &= ~TIM_CR1_CEN;
+  tim->DIER &= ~TIM_DIER_UIE;
+  return tim->CNT;
+}
+// settings from M4 for >= 150MHz, we use this part at 168MHz
+#define _PM_clockHoldHigh asm("nop; nop; nop");
+#define _PM_clockHoldLow asm("nop");
+
+#define _PM_minMinPeriod 140
+
+#endif // END CIRCUITPYTHON ------------------------------------------------
+
+#endif // END STM32F4_SERIES || STM32F405xx

src/arch/teensy4.h

@@ -0,0 +1,172 @@
+/*!
+ * @file teensy4.h
+ *
+ * Part of Adafruit's Protomatter library for HUB75-style RGB LED matrices.
+ * This file contains i.MX 1062 (Teensy 4.x) SPECIFIC CODE.
+ *
+ * Adafruit invests time and resources providing this open source code,
+ * please support Adafruit and open-source hardware by purchasing
+ * products from Adafruit!
+ *
+ * Written by Phil "Paint Your Dragon" Burgess and Jeff Epler for
+ * Adafruit Industries, with contributions from the open source community.
+ *
+ * BSD license, all text here must be included in any redistribution.
+ *
+ */
+
+#pragma once
+
+#if defined(__IMXRT1062__)
+
+// i.MX only allows full 32-bit aligned writes to GPIO.
+#define _PM_STRICT_32BIT_IO ///< Change core.c behavior for long accesses only
+
+#if defined(ARDUINO) // COMPILING FOR ARDUINO ------------------------------
+
+static const struct {
+  volatile uint32_t *base; ///< GPIO base address for pin
+  uint8_t bit;             ///< GPIO bit number for pin (0-31)
+} _PM_teensyPins[] = {
+    {&CORE_PIN0_PORTREG, CORE_PIN0_BIT},
+    {&CORE_PIN1_PORTREG, CORE_PIN1_BIT},
+    {&CORE_PIN2_PORTREG, CORE_PIN2_BIT},
+    {&CORE_PIN3_PORTREG, CORE_PIN3_BIT},
+    {&CORE_PIN4_PORTREG, CORE_PIN4_BIT},
+    {&CORE_PIN5_PORTREG, CORE_PIN5_BIT},
+    {&CORE_PIN6_PORTREG, CORE_PIN6_BIT},
+    {&CORE_PIN7_PORTREG, CORE_PIN7_BIT},
+    {&CORE_PIN8_PORTREG, CORE_PIN8_BIT},
+    {&CORE_PIN9_PORTREG, CORE_PIN9_BIT},
+    {&CORE_PIN10_PORTREG, CORE_PIN10_BIT},
+    {&CORE_PIN11_PORTREG, CORE_PIN11_BIT},
+    {&CORE_PIN12_PORTREG, CORE_PIN12_BIT},
+    {&CORE_PIN13_PORTREG, CORE_PIN13_BIT},
+    {&CORE_PIN14_PORTREG, CORE_PIN14_BIT},
+    {&CORE_PIN15_PORTREG, CORE_PIN15_BIT},
+    {&CORE_PIN16_PORTREG, CORE_PIN16_BIT},
+    {&CORE_PIN17_PORTREG, CORE_PIN17_BIT},
+    {&CORE_PIN18_PORTREG, CORE_PIN18_BIT},
+    {&CORE_PIN19_PORTREG, CORE_PIN19_BIT},
+    {&CORE_PIN20_PORTREG, CORE_PIN20_BIT},
+    {&CORE_PIN21_PORTREG, CORE_PIN21_BIT},
+    {&CORE_PIN22_PORTREG, CORE_PIN22_BIT},
+    {&CORE_PIN23_PORTREG, CORE_PIN23_BIT},
+    {&CORE_PIN24_PORTREG, CORE_PIN24_BIT},
+    {&CORE_PIN25_PORTREG, CORE_PIN25_BIT},
+    {&CORE_PIN26_PORTREG, CORE_PIN26_BIT},
+    {&CORE_PIN27_PORTREG, CORE_PIN27_BIT},
+    {&CORE_PIN28_PORTREG, CORE_PIN28_BIT},
+    {&CORE_PIN29_PORTREG, CORE_PIN29_BIT},
+    {&CORE_PIN30_PORTREG, CORE_PIN30_BIT},
+    {&CORE_PIN31_PORTREG, CORE_PIN31_BIT},
+    {&CORE_PIN32_PORTREG, CORE_PIN32_BIT},
+    {&CORE_PIN33_PORTREG, CORE_PIN33_BIT},
+    {&CORE_PIN34_PORTREG, CORE_PIN34_BIT},
+    {&CORE_PIN35_PORTREG, CORE_PIN35_BIT},
+    {&CORE_PIN36_PORTREG, CORE_PIN36_BIT},
+    {&CORE_PIN37_PORTREG, CORE_PIN37_BIT},
+    {&CORE_PIN38_PORTREG, CORE_PIN38_BIT},
+    {&CORE_PIN39_PORTREG, CORE_PIN39_BIT},
+};
+
+#define _PM_SET_OFFSET 33    ///< 0x84 byte offset = 33 longs
+#define _PM_CLEAR_OFFSET 34  ///< 0x88 byte offset = 34 longs
+#define _PM_TOGGLE_OFFSET 35 ///< 0x8C byte offset = 35 longs
+
+#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+#define _PM_byteOffset(pin) (_PM_teensyPins[pin].bit / 8)
+#define _PM_wordOffset(pin) (_PM_teensyPins[pin].bit / 16)
+#else
+#define _PM_byteOffset(pin) (3 - (_PM_teensyPins[pin].bit / 8))
+#define _PM_wordOffset(pin) (1 - (_PM_teensyPins[pin].bit / 16))
+#endif
+
+#define _PM_portOutRegister(pin) (void *)_PM_teensyPins[pin].base
+
+#define _PM_portSetRegister(pin)                                               \
+  ((volatile uint32_t *)_PM_teensyPins[pin].base + _PM_SET_OFFSET)
+
+#define _PM_portClearRegister(pin)                                             \
+  ((volatile uint32_t *)_PM_teensyPins[pin].base + _PM_CLEAR_OFFSET)
+
+#define _PM_portToggleRegister(pin)                                            \
+  ((volatile uint32_t *)_PM_teensyPins[pin].base + _PM_TOGGLE_OFFSET)
+
+// As written, because it's tied to a specific timer right now, the
+// Arduino lib only permits one instance of the Protomatter_core struct,
+// which it sets up when calling begin().
+void *_PM_protoPtr = NULL;
+
+// Code as written works with the Periodic Interrupt Timer directly,
+// rather than using the Teensy IntervalTimer library, reason being we
+// need to be able to poll the current timer value in _PM_timerGetCount(),
+// but that's not available from IntervalTimer, and the timer base address
+// it keeps is a private member (possible alternative is to do dirty pool
+// and access the pointer directly, knowing it's the first element in the
+// IntervalTimer object, but this is fraught with peril).
+
+#define _PM_timerFreq 24000000 // 24 MHz
+#define _PM_timerNum 0         // PIT timer #0 (can be 0-3)
+#define _PM_TIMER_DEFAULT (IMXRT_PIT_CHANNELS + _PM_timerNum) // PIT channel *
+
+// Interrupt service routine for Periodic Interrupt Timer
+static void _PM_timerISR(void) {
+  IMXRT_PIT_CHANNEL_t *timer = _PM_TIMER_DEFAULT;
+  _PM_row_handler(_PM_protoPtr); // In core.c
+  timer->TFLG = 1;               // Clear timer interrupt
+}
+
+// Initialize, but do not start, timer.
+void _PM_timerInit(void *tptr) {
+  IMXRT_PIT_CHANNEL_t *timer = (IMXRT_PIT_CHANNEL_t *)tptr;
+  CCM_CCGR1 |= CCM_CCGR1_PIT(CCM_CCGR_ON); // Enable clock signal to PIT
+  PIT_MCR = 1;                             // Enable PIT
+  timer->TCTRL = 0;                        // Disable timer and interrupt
+  timer->LDVAL = 100000;                   // Timer initial load value
+  // Interrupt is attached but not enabled yet
+  attachInterruptVector(IRQ_PIT, &_PM_timerISR);
+  NVIC_ENABLE_IRQ(IRQ_PIT);
+}
+
+// Set timer period, initialize count value to zero, enable timer.
+inline void _PM_timerStart(void *tptr, uint32_t period) {
+  IMXRT_PIT_CHANNEL_t *timer = (IMXRT_PIT_CHANNEL_t *)tptr;
+  timer->TCTRL = 0;      // Disable timer and interrupt
+  timer->LDVAL = period; // Set load value
+  // timer->CVAL = period; // And current value (just in case?)
+  timer->TFLG = 1;  // Clear timer interrupt
+  timer->TCTRL = 3; // Enable timer and interrupt
+}
+
+// Return current count value (timer enabled or not).
+// Timer must be previously initialized.
+inline uint32_t _PM_timerGetCount(void *tptr) {
+  IMXRT_PIT_CHANNEL_t *timer = (IMXRT_PIT_CHANNEL_t *)tptr;
+  return (timer->LDVAL - timer->CVAL);
+}
+
+// Disable timer and return current count value.
+// Timer must be previously initialized.
+uint32_t _PM_timerStop(void *tptr) {
+  IMXRT_PIT_CHANNEL_t *timer = (IMXRT_PIT_CHANNEL_t *)tptr;
+  timer->TCTRL = 0; // Disable timer and interrupt
+  return _PM_timerGetCount(tptr);
+}
+
+#define _PM_clockHoldHigh                                                      \
+  asm("nop; nop; nop; nop; nop; nop; nop;");                                   \
+  asm("nop; nop; nop; nop; nop; nop; nop;");
+#define _PM_clockHoldLow                                                       \
+  asm("nop; nop; nop; nop; nop; nop; nop; nop; nop; nop;");                    \
+  asm("nop; nop; nop; nop; nop; nop; nop; nop; nop; nop;");
+
+#define _PM_chunkSize 1 ///< DON'T unroll loop, Teensy 4 is SO FAST
+
+#elif defined(CIRCUITPY) // COMPILING FOR CIRCUITPYTHON --------------------
+
+// Teensy 4 CircuitPython magic goes here.
+
+#endif // END CIRCUITPYTHON ------------------------------------------------
+
+#endif // END __IMXRT1062__ (Teensy 4)

src/core.h

@@ -0,0 +1,264 @@
+/*!
+ * @file core.h
+ *
+ * Part of Adafruit's Protomatter library for HUB75-style RGB LED matrices.
+ *
+ * Adafruit invests time and resources providing this open source code,
+ * please support Adafruit and open-source hardware by purchasing
+ * products from Adafruit!
+ *
+ * Written by Phil "Paint Your Dragon" Burgess and Jeff Epler for
+ * Adafruit Industries, with contributions from the open source community.
+ *
+ * BSD license, all text here must be included in any redistribution.
+ *
+ */
+
+#pragma once
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <stdbool.h>
+#include <stdint.h>
+
+/** Status type returned by some functions. */
+typedef enum {
+  PROTOMATTER_OK,         // Everything is hunky-dory!
+  PROTOMATTER_ERR_PINS,   // Clock and/or data pins on different PORTs
+  PROTOMATTER_ERR_MALLOC, // Couldn't allocate memory for display
+  PROTOMATTER_ERR_ARG,    // Bad input to function
+} ProtomatterStatus;
+
+/** Struct for matrix control lines NOT related to RGB data or clock, i.e.
+    latch, OE and address lines. RGB data and clock ("RGBC") are handled
+    differently as they have specific requirements (and might use a toggle
+    register if present). The data conversion functions need bitmasks for
+    RGB data but do NOT need the set or clear registers, so those items are
+    also declared as separate things in the core structure that follows. */
+typedef struct {
+  volatile void *setReg;   ///< GPIO bit set register
+  volatile void *clearReg; ///< GPIO bit clear register
+  uint32_t bit;            ///< GPIO bitmask
+  uint8_t pin;             ///< Some unique ID, e.g. Arduino pin #
+} _PM_pin;
+
+/** Struct with info about an RGB matrix chain and lots of state and buffer
+    details for the library. Toggle-related items in this structure MUST be
+    declared even if the device lacks GPIO bit-toggle registers (i.e. don't
+    do an ifdef check around these). All hardware-specific details (including
+    the presence or lack of toggle registers) are isolated to a single
+    file -- arch.h -- which should ONLY be included by core.c, and ifdef'ing
+    them would result in differing representations of this structure which
+    must be shared between the library and calling code. (An alternative is
+    to put any toggle-specific stuff at the end of the struct with an ifdef
+    check, but that's just dirty pool and asking for trouble.) */
+typedef struct {
+  void *timer;                   ///< Arch-specific timer/counter info
+  void *setReg;                  ///< RGBC bit set register (cast to use)
+  void *clearReg;                ///< RGBC bit clear register "
+  void *toggleReg;               ///< RGBC bit toggle register "
+  uint8_t *rgbPins;              ///< Array of RGB data pins (mult of 6)
+  void *rgbMask;                 ///< PORT bit mask for each RGB pin
+  uint32_t clockMask;            ///< PORT bit mask for RGB clock
+  uint32_t rgbAndClockMask;      ///< PORT bit mask for RGB data + clock
+  volatile void *addrPortToggle; ///< See singleAddrPort below
+  void *screenData;              ///< Per-bitplane RGB data for matrix
+  _PM_pin latch;                 ///< RGB data latch
+  _PM_pin oe;                    ///< !OE (LOW out enable)
+  _PM_pin *addr;                 ///< Array of address pins
+  uint32_t bufferSize;           ///< Bytes per matrix buffer
+  uint32_t bitZeroPeriod;        ///< Bitplane 0 timer period
+  uint32_t minPeriod;            ///< Plane 0 timer period for ~250Hz
+  volatile uint32_t frameCount;  ///< For estimating refresh rate
+  uint16_t width;                ///< Matrix chain width in bits
+  uint8_t bytesPerElement;       ///< Using 8, 16 or 32 bits of PORT?
+  uint8_t clockPin;              ///< RGB clock pin identifier
+  uint8_t parallel;              ///< Number of concurrent matrix outs
+  uint8_t numAddressLines;       ///< Number of address line pins
+  uint8_t portOffset;            ///< Active 8- or 16-bit pos. in PORT
+  uint8_t numPlanes;             ///< Display bitplanes (1 to 6)
+  uint8_t numRowPairs;           ///< Addressable row pairs
+  bool doubleBuffer;             ///< 2X buffers for clean switchover
+  bool singleAddrPort;           ///< If 1, all addr lines on same PORT
+  volatile uint8_t activeBuffer; ///< Index of currently-displayed buf
+  volatile uint8_t plane;        ///< Current bitplane (changes in ISR)
+  volatile uint8_t row;          ///< Current scanline (changes in ISR)
+  volatile uint8_t prevRow;      ///< Scanline from prior ISR
+  volatile bool swapBuffers;     ///< If 1, awaiting double-buf switch
+} Protomatter_core;
+
+// Protomatter core function prototypes. Environment-specific code (like the
+// Adafruit_Protomatter class for Arduino) calls on these underlying things,
+// and has to provide a few extras of its own (interrupt handlers and such).
+// User code shouldn't need to invoke any of them directly.
+
+/*!
+  @brief  Initialize values in Protomatter_core structure.
+  @param  core  Pointer to Protomatter_core structure.
+  @param  bitWidth      Total width of RGB matrix chain, in pixels.
+                        Usu. some multiple of 32, but maybe exceptions.
+  @param  bitDepth      Color "depth" in bitplanes, determines range of
+                        shades of red, green and blue. e.g. passing 4
+                        bits = 16 shades ea. R,G,B = 16x16x16 = 4096
+                        colors.
+  @param  rgbCount      Number of "sets" of RGB data pins, each set
+                        containing 6 pins (2 ea. R,G,B). Typically 1,
+                        indicating a single matrix (or matrix chain).
+                        In theory (but not yet extensively tested),
+                        multiple sets of pins can be driven in parallel,
+                        up to 5 on some devices (if the hardware design
+                        provides all those bits on one PORT).
+  @param  rgbList       A uint8_t array of pins (values are platform-
+                        dependent), 6X the prior rgbCount value,
+                        corresponding to the 6 output color bits for a
+                        matrix (or chain). Order is upper-half red, green,
+                        blue, lower-half red, green blue (repeat for each
+                        add'l chain). All the RGB pins (plus the clock pin
+                        below on some architectures) MUST be on the same
+                        PORT register. It's recommended (but not required)
+                        that all RGB pins (and clock depending on arch) be
+                        within the same byte of a PORT (but do not need to
+                        be sequential or contiguous within that byte) for
+                        more efficient RAM utilization. For two concurrent
+                        chains, same principle but 16-bit word.
+  @param  addrCount     Number of row address lines required of matrix.
+                        Total pixel height is then 2 x 2^addrCount, e.g.
+                        32-pixel-tall matrices have 4 row address lines.
+  @param  addrList      A uint8_t array of pins (platform-dependent pin
+                        numbering), one per row address line.
+  @param  clockPin      RGB clock pin (platform-dependent pin #).
+  @param  latchPin      RGB data latch pin (platform-dependent pin #).
+  @param  oePin         Output enable pin (platform-dependent pin #),
+                        active low.
+  @param  doubleBuffer  If true, two matrix buffers are allocated,
+                        so changing display contents doesn't introduce
+                        artifacts mid-conversion. Requires ~2X RAM.
+  @param  timer         Pointer to timer peripheral or timer-related
+                        struct (architecture-dependent), or NULL to
+                        use a default timer ID (also arch-dependent).
+  @return A ProtomatterStatus status, one of:
+          PROTOMATTER_OK if everything is good.
+          PROTOMATTER_ERR_PINS if data and/or clock pins are split across
+          different PORTs.
+          PROTOMATTER_ERR_MALLOC if insufficient RAM to allocate display
+          memory.
+          PROTOMATTER_ERR_ARG if a bad value (core or timer pointer) was
+          passed in.
+*/
+extern ProtomatterStatus _PM_init(Protomatter_core *core, uint16_t bitWidth,
+                                  uint8_t bitDepth, uint8_t rgbCount,
+                                  uint8_t *rgbList, uint8_t addrCount,
+                                  uint8_t *addrList, uint8_t clockPin,
+                                  uint8_t latchPin, uint8_t oePin,
+                                  bool doubleBuffer, void *timer);
+
+/*!
+  @brief  Allocate display buffers and populate additional elements of a
+          Protomatter matrix.
+  @param  core  Pointer to Protomatter_core structure.
+  @return A ProtomatterStatus status, one of:
+          PROTOMATTER_OK if everything is good.
+          PROTOMATTER_ERR_PINS if data and/or clock pins are split across
+          different PORTs.
+          PROTOMATTER_ERR_MALLOC if insufficient RAM to allocate display
+          memory.
+          PROTOMATTER_ERR_ARG if a bad value.
+*/
+extern ProtomatterStatus _PM_begin(Protomatter_core *core);
+
+/*!
+  @brief  Disable (but do not deallocate) a Protomatter matrix. Disables
+          matrix by setting OE pin HIGH and writing all-zero data to
+          matrix shift registers, so it won't halt with lit LEDs.
+  @param  core  Pointer to Protomatter_core structure.
+*/
+extern void _PM_stop(Protomatter_core *core);
+
+/*!
+  @brief  Start or restart a matrix. Initialize counters, configure and
+          start timer.
+  @param  core  Pointer to Protomatter_core structure.
+*/
+extern void _PM_resume(Protomatter_core *core);
+
+/*!
+  @brief  Deallocate memory associated with Protomatter_core structure
+          (e.g. screen data, pin lists for data and rows). Does not
+          deallocate the structure itself.
+  @param  core  Pointer to Protomatter_core structure.
+*/
+extern void _PM_deallocate(Protomatter_core *core);
+
+/*!
+  @brief  Matrix "row handler" that's called by the timer interrupt.
+          Handles row address lines and issuing data to matrix.
+  @param  core  Pointer to Protomatter_core structure.
+*/
+extern void _PM_row_handler(Protomatter_core *core);
+
+/*!
+  @brief  Returns current value of frame counter and resets its value to
+          zero. Two calls to this, timed one second apart (or use math with
+          other intervals), can be used to get a rough frames-per-second
+          value for the matrix (since this is difficult to estimate
+          beforehand).
+  @param  core  Pointer to Protomatter_core structure.
+  @return Frame count since previous call to function, as a uint32_t.
+*/
+extern uint32_t _PM_getFrameCount(Protomatter_core *core);
+
+/*!
+  @brief  Start (or restart) a timer/counter peripheral.
+  @param  tptr    Pointer to timer/counter peripheral OR a struct
+                  encapsulating information about a timer/counter
+                  periph (architecture-dependent).
+  @param  period  Timer 'top' / rollover value.
+*/
+extern void _PM_timerStart(void *tptr, uint32_t period);
+
+/*!
+  @brief  Stop timer/counter peripheral.
+  @param  tptr    Pointer to timer/counter peripheral OR a struct
+                  encapsulating information about a timer/counter
+                  periph (architecture-dependent).
+  @return Counter value when timer was stopped.
+*/
+extern uint32_t _PM_timerStop(void *tptr);
+
+/*!
+  @brief  Query a timer/counter peripheral's current count.
+  @param  tptr    Pointer to timer/counter peripheral OR a struct
+                  encapsulating information about a timer/counter
+                  periph (architecture-dependent).
+  @return Counter value.
+*/
+extern uint32_t _PM_timerGetCount(void *tptr);
+
+/*!
+  @brief  Pauses until the next vertical blank to avoid 'tearing' animation
+          (if display is double-buffered). If single-buffered, has no effect.
+  @param  core  Pointer to Protomatter_core structure.
+*/
+extern void _PM_swapbuffer_maybe(Protomatter_core *core);
+
+#if defined(ARDUINO) || defined(CIRCUITPY)
+
+/*!
+  @brief  Converts image data from GFX16 canvas to the matrices weird
+          internal format.
+  @param  core    Pointer to Protomatter_core structure.
+  @param  source  Pointer to source image data (see Adafruit_GFX 16-bit
+                  canvas type for format).
+  @param  width   Width of canvas in pixels, as this may be different than
+                  the matrix pixel width due to row padding.
+*/
+extern void _PM_convert_565(Protomatter_core *core, uint16_t *source,
+                            uint16_t width);
+
+#endif // END ARDUINO || CIRCUITPY
+
+#ifdef __cplusplus
+} // extern "C"
+#endif