Added in manitou48 's fixes to FastLed for ItsyBitsy M4 (SADM51 boards) (#803)

* Files fixed by manitou48 m4 feather express update

I have not been able to find user manitou48 so I forked his changes and am trying to make a pull request to get it into FastLED.

* delete led_sysdefs.h

move led_sysdef.h to root of FastLED where it belongs.

* Delete platforms.h

Move platform.h to root of FastLED where it belongs.

* move files to root of FastLED 

The led_sysdefs.h and platforms.h files belong in the root of the FastLED directory, not in platforms/arm/d51

5 files changed, 131 insertions, 40 deletions

diff --git a/led_sysdefs.h b/led_sysdefs.h
index 13fbf4c3..ea8c14f4 100644
--- a/led_sysdefs.h
+++ b/led_sysdefs.h
@@ -21,7 +21,7 @@
 #include "platforms/arm/sam/led_sysdefs_arm_sam.h"
 #elif defined(STM32F10X_MD) || defined(__STM32F1__)
 #include "platforms/arm/stm32/led_sysdefs_arm_stm32.h"
-#elif defined(__SAMD21G18A__) || defined(__SAMD21J18A__) || defined(__SAMD21E17A__) || defined(__SAMD21E18A__) || defined(__SAMD51G19A__)
+#elif defined(__SAMD21G18A__) || defined(__SAMD21J18A__) || defined(__SAMD21E17A__) || defined(__SAMD21E18A__) || defined(__SAMD51G19A__) || defined(__SAMD51J19A__)
 #include "platforms/arm/d21/led_sysdefs_arm_d21.h"
 #elif defined(ESP8266)
 #include "platforms/esp/8266/led_sysdefs_esp8266.h"
diff --git a/platforms.h b/platforms.h
index 88bf462d..625791bc 100644
--- a/platforms.h
+++ b/platforms.h
@@ -23,7 +23,7 @@
 #include "platforms/arm/stm32/fastled_arm_stm32.h"
 #elif defined(__SAMD21G18A__) || defined(__SAMD21J18A__) || defined(__SAMD21E17A__) || defined(__SAMD21E18A__)
 #include "platforms/arm/d21/fastled_arm_d21.h"
-#elif defined(__SAMD51G19A__)
+#elif defined(__SAMD51G19A__) || defined(__SAMD51J19A__)
 #include "platforms/arm/d51/fastled_arm_d51.h"
 #elif defined(ESP8266)
 #include "platforms/esp/8266/fastled_esp8266.h"
diff --git a/platforms/arm/d51/README.txt b/platforms/arm/d51/README.txt
new file mode 100644
index 00000000..b00fb670
--- /dev/null
+++ b/platforms/arm/d51/README.txt
@@ -0,0 +1,4 @@
+FastLED updates for adafruit FEATHER M4 and fixes to ITSBITSY M4 compiles
+  SAMD51
+
+only tested on FEATHER M4 with DOTSTAR and neopixel strips
diff --git a/platforms/arm/d51/clockless_arm_d51.h b/platforms/arm/d51/clockless_arm_d51.h
index a543ec18..0c3f6d4d 100644
--- a/platforms/arm/d51/clockless_arm_d51.h
+++ b/platforms/arm/d51/clockless_arm_d51.h
@@ -1,62 +1,128 @@
 #ifndef __INC_CLOCKLESS_ARM_D51
 #define __INC_CLOCKLESS_ARM_D51
 
-// D51 is an M4 chip, however the M0 clockless logic seems to work.
-#include "../common/m0clockless.h"
 FASTLED_NAMESPACE_BEGIN
+
+// Definition for a single channel clockless controller for SAMD51
+// See clockless.h for detailed info on how the template parameters are used.
+#define ARM_DEMCR               (*(volatile uint32_t *)0xE000EDFC) // Debug Exception and Monitor Control
+#define ARM_DEMCR_TRCENA                (1 << 24)        // Enable debugging & monitoring blocks
+#define ARM_DWT_CTRL            (*(volatile uint32_t *)0xE0001000) // DWT control register
+#define ARM_DWT_CTRL_CYCCNTENA          (1 << 0)                // Enable cycle count
+#define ARM_DWT_CYCCNT          (*(volatile uint32_t *)0xE0001004) // Cycle count register
+
+
 #define FASTLED_HAS_CLOCKLESS 1
 
-template <uint8_t DATA_PIN, int T1, int T2, int T3, EOrder RGB_ORDER = RGB, int XTRA0 = 0, bool FLIP = false, int WAIT_TIME = 50>
+template <int DATA_PIN, int T1, int T2, int T3, EOrder RGB_ORDER = RGB, int XTRA0 = 0, bool FLIP = false, int WAIT_TIME = 50>
 class ClocklessController : public CPixelLEDController<RGB_ORDER> {
-  typedef typename FastPinBB<DATA_PIN>::port_ptr_t data_ptr_t;
-  typedef typename FastPinBB<DATA_PIN>::port_t data_t;
+	typedef typename FastPin<DATA_PIN>::port_ptr_t data_ptr_t;
+	typedef typename FastPin<DATA_PIN>::port_t data_t;
 
-  data_t mPinMask;
-  data_ptr_t mPort;
-  CMinWait<WAIT_TIME> mWait;
+	data_t mPinMask;
+	data_ptr_t mPort;
+	CMinWait<WAIT_TIME> mWait;
 public:
-  virtual void init() {
-    FastPinBB<DATA_PIN>::setOutput();
-    mPinMask = FastPinBB<DATA_PIN>::mask();
-    mPort = FastPinBB<DATA_PIN>::port();
-  }
+	virtual void init() {
+		FastPin<DATA_PIN>::setOutput();
+		mPinMask = FastPin<DATA_PIN>::mask();
+		mPort = FastPin<DATA_PIN>::port();
+	}
 
 	virtual uint16_t getMaxRefreshRate() const { return 400; }
 
-  virtual void showPixels(PixelController<RGB_ORDER> & pixels) {
+protected:
+
+	virtual void showPixels(PixelController<RGB_ORDER> & pixels) {
     mWait.wait();
-    cli();
-    if(!showRGBInternal(pixels)) {
+		if(!showRGBInternal(pixels)) {
       sei(); delayMicroseconds(WAIT_TIME); cli();
       showRGBInternal(pixels);
     }
-    sei();
     mWait.mark();
   }
 
-  // This method is made static to force making register Y available to use for data on AVR - if the method is non-static, then
-  // gcc will use register Y for the this pointer.
-  static uint32_t showRGBInternal(PixelController<RGB_ORDER> pixels) {
-    struct M0ClocklessData data;
-    data.d[0] = pixels.d[0];
-    data.d[1] = pixels.d[1];
-    data.d[2] = pixels.d[2];
-    data.s[0] = pixels.mScale[0];
-    data.s[1] = pixels.mScale[1];
-    data.s[2] = pixels.mScale[2];
-    data.e[0] = pixels.e[0];
-    data.e[1] = pixels.e[1];
-    data.e[2] = pixels.e[2];
-    data.adj = pixels.mAdvance;
-
-    typename FastPin<DATA_PIN>::port_ptr_t portBase = FastPin<DATA_PIN>::port();
-    return showLedData<8,4,T1,T2,T3,RGB_ORDER, WAIT_TIME>(portBase, FastPin<DATA_PIN>::mask(), pixels.mData, pixels.mLen, &data);
-  }
+	template<int BITS> __attribute__ ((always_inline)) inline static void writeBits(register uint32_t & next_mark, register data_ptr_t port, register data_t hi, register data_t lo, register uint8_t & b)  {
+		for(register uint32_t i = BITS-1; i > 0; i--) {
+			while(ARM_DWT_CYCCNT < next_mark);
+			next_mark = ARM_DWT_CYCCNT + (T1+T2+T3);
+			FastPin<DATA_PIN>::fastset(port, hi);
+			if(b&0x80) {
+				while((next_mark - ARM_DWT_CYCCNT) > (T3+(2*(F_CPU/24000000))));
+				FastPin<DATA_PIN>::fastset(port, lo);
+			} else {
+				while((next_mark - ARM_DWT_CYCCNT) > (T2+T3+(2*(F_CPU/24000000))));
+				FastPin<DATA_PIN>::fastset(port, lo);
+			}
+			b <<= 1;
+		}
+
+		while(ARM_DWT_CYCCNT < next_mark);
+		next_mark = ARM_DWT_CYCCNT + (T1+T2+T3);
+		FastPin<DATA_PIN>::fastset(port, hi);
+
+		if(b&0x80) {
+			while((next_mark - ARM_DWT_CYCCNT) > (T3+(2*(F_CPU/24000000))));
+			FastPin<DATA_PIN>::fastset(port, lo);
+		} else {
+			while((next_mark - ARM_DWT_CYCCNT) > (T2+T3+(2*(F_CPU/24000000))));
+			FastPin<DATA_PIN>::fastset(port, lo);
+		}
+	}
 
+	// This method is made static to force making register Y available to use for data on AVR - if the method is non-static, then
+	// gcc will use register Y for the this pointer.
+	static uint32_t showRGBInternal(PixelController<RGB_ORDER> pixels) {
+	    // Get access to the clock
+		ARM_DEMCR    |= ARM_DEMCR_TRCENA;
+		ARM_DWT_CTRL |= ARM_DWT_CTRL_CYCCNTENA;
+		ARM_DWT_CYCCNT = 0;
 
+		register data_ptr_t port = FastPin<DATA_PIN>::port();
+		register data_t hi = *port | FastPin<DATA_PIN>::mask();;
+		register data_t lo = *port & ~FastPin<DATA_PIN>::mask();;
+		*port = lo;
+
+		// Setup the pixel controller and load/scale the first byte
+		pixels.preStepFirstByteDithering();
+		register uint8_t b = pixels.loadAndScale0();
+
+		cli();
+		uint32_t next_mark = ARM_DWT_CYCCNT + (T1+T2+T3);
+
+		while(pixels.has(1)) {
+			pixels.stepDithering();
+			#if (FASTLED_ALLOW_INTERRUPTS == 1)
+			cli();
+			// if interrupts took longer than 45µs, punt on the current frame
+			if(ARM_DWT_CYCCNT > next_mark) {
+				if((ARM_DWT_CYCCNT-next_mark) > ((WAIT_TIME-INTERRUPT_THRESHOLD)*CLKS_PER_US)) { sei(); return 0; }
+			}
+
+			hi = *port | FastPin<DATA_PIN>::mask();
+			lo = *port & ~FastPin<DATA_PIN>::mask();
+			#endif
+			// Write first byte, read next byte
+			writeBits<8+XTRA0>(next_mark, port, hi, lo, b);
+			b = pixels.loadAndScale1();
+
+			// Write second byte, read 3rd byte
+			writeBits<8+XTRA0>(next_mark, port, hi, lo, b);
+			b = pixels.loadAndScale2();
+
+			// Write third byte, read 1st byte of next pixel
+			writeBits<8+XTRA0>(next_mark, port, hi, lo, b);
+			b = pixels.advanceAndLoadAndScale0();
+			#if (FASTLED_ALLOW_INTERRUPTS == 1)
+			sei();
+			#endif
+		};
+
+		sei();
+		return ARM_DWT_CYCCNT;
+	}
 };
 
 FASTLED_NAMESPACE_END
 
-
-#endif // __INC_CLOCKLESS_ARM_D51
+#endif
diff --git a/platforms/arm/d51/fastpin_arm_d51.h b/platforms/arm/d51/fastpin_arm_d51.h
index 5562d846..6d14c633 100644
--- a/platforms/arm/d51/fastpin_arm_d51.h
+++ b/platforms/arm/d51/fastpin_arm_d51.h
@@ -75,7 +75,8 @@ _DEFPIN_ARM(14, 0,  2); _DEFPIN_ARM(15, 0,  5); _DEFPIN_ARM(16, 1,  8); _DEFPIN_
 _DEFPIN_ARM(18, 0,  4); _DEFPIN_ARM(19, 0,  6); /* A6 is present in variant.h but couldn't find it on the schematic */
 // SDA/SCL
 _DEFPIN_ARM(21, 0, 12); _DEFPIN_ARM(22, 0, 13);
-// MISO/SCK/MOSI
+
+// 23..25  MISO/SCK/MOSI
 _DEFPIN_ARM(23, 1, 23); _DEFPIN_ARM(24, 0,  1); _DEFPIN_ARM(25, 0,  0);
 
 #define SPI_DATA 25
@@ -83,6 +84,26 @@ _DEFPIN_ARM(23, 1, 23); _DEFPIN_ARM(24, 0,  1); _DEFPIN_ARM(25, 0,  0);
 
 #define HAS_HARDWARE_PIN_SUPPORT 1
 
+#elif defined(ADAFRUIT_FEATHER_M4_EXPRESS)
+
+#define MAX_PIN 19
+// D0-D13, including D8 (neopixel)  no pins 2 3
+_DEFPIN_ARM( 0, 1, 17); _DEFPIN_ARM( 1, 1, 16);
+_DEFPIN_ARM( 4, 0, 14); _DEFPIN_ARM( 5, 0, 16); _DEFPIN_ARM( 6, 0,  18);
+_DEFPIN_ARM( 8, 1,  3); _DEFPIN_ARM( 9, 0, 19); _DEFPIN_ARM(10, 0, 20); _DEFPIN_ARM(11, 0, 21);
+_DEFPIN_ARM(12, 0, 22); _DEFPIN_ARM(13, 0, 23);
+// A0-A5
+_DEFPIN_ARM(14, 0,  2); _DEFPIN_ARM(15, 0,  5); _DEFPIN_ARM(16, 1,  8); _DEFPIN_ARM(17, 1,  9);
+_DEFPIN_ARM(18, 0,  4); _DEFPIN_ARM(19, 0,  6); /* A6 is present in variant.h but couldn't find it on the schematic */
+// SDA/SCL
+_DEFPIN_ARM(21, 0, 12); _DEFPIN_ARM(22, 0, 13);
+// 23..25  MISO/MOSI/SCK
+_DEFPIN_ARM(23, 1, 22); _DEFPIN_ARM(24, 1,  23); _DEFPIN_ARM(25, 0,  17);
+
+#define SPI_DATA 24
+#define SPI_CLOCK 25
+
+#define HAS_HARDWARE_PIN_SUPPORT 1
 #endif