Switch Teensy parallel output to use CPixelLEDController objects

2 files changed, 71 insertions, 114 deletions

diff --git a/FastLED.h b/FastLED.h
index 496411a1..5cf6d3fa 100644
--- a/FastLED.h
+++ b/FastLED.h
@@ -388,8 +388,8 @@ public:
 				case WS2811_400_PORTD: return addLeds(new InlineBlockClocklessController<NUM_LANES, PORTD_FIRST_PIN, NS(800), NS(800), NS(900), RGB_ORDER>(), data, nLedsOrOffset, nLedsIfOffset);
 		#endif
 		#ifdef HAS_PORTDC
-				case WS2811_PORTDC: return addLeds(new SixteenWayInlineBlockClocklessController<16,NS(320), NS(320), NS(640), RGB_ORDER>(), data, nLedsOrOffset, nLedsIfOffset);
-				case WS2811_400_PORTDC: return addLeds(new SixteenWayInlineBlockClocklessController<16,NS(800), NS(800), NS(900), RGB_ORDER>(), data, nLedsOrOffset, nLedsIfOffset);
+				case WS2811_PORTDC: return addLeds(new SixteenWayInlineBlockClocklessController<NUM_LANES,NS(320), NS(320), NS(640), RGB_ORDER>(), data, nLedsOrOffset, nLedsIfOffset);
+				case WS2811_400_PORTDC: return addLeds(new SixteenWayInlineBlockClocklessController<NUM_LANES,NS(800), NS(800), NS(900), RGB_ORDER>(), data, nLedsOrOffset, nLedsIfOffset);
 		#endif
 		}
 	}
diff --git a/platforms/arm/k20/clockless_block_arm_k20.h b/platforms/arm/k20/clockless_block_arm_k20.h
index b5e4007a..f2b76468 100644
--- a/platforms/arm/k20/clockless_block_arm_k20.h
+++ b/platforms/arm/k20/clockless_block_arm_k20.h
@@ -13,14 +13,14 @@
 #define PORT_MASK (((1<<LANES)-1) & ((FIRST_PIN==2) ? 0xFF : 0xFFF))
 
 #define MIN(X,Y) (((X)<(Y)) ? (X):(Y))
-#define LANES ((FIRST_PIN==2) ? MIN(__LANES,8) : MIN(__LANES,12))
+#define USED_LANES ((FIRST_PIN==2) ? MIN(LANES,8) : MIN(LANES,12))
 
 #include "kinetis.h"
 
 FASTLED_NAMESPACE_BEGIN
 
-template <uint8_t __LANES, int FIRST_PIN, int T1, int T2, int T3, EOrder RGB_ORDER = GRB, int XTRA0 = 0, bool FLIP = false, int WAIT_TIME = 40>
-class InlineBlockClocklessController : public CLEDController {
+template <uint8_t LANES, int FIRST_PIN, int T1, int T2, int T3, EOrder RGB_ORDER = GRB, int XTRA0 = 0, bool FLIP = false, int WAIT_TIME = 40>
+class InlineBlockClocklessController : public CPixelLEDController<RGB_ORDER, LANES, PORT_MASK> {
 	typedef typename FastPin<FIRST_PIN>::port_ptr_t data_ptr_t;
 	typedef typename FastPin<FIRST_PIN>::port_t data_t;
 
@@ -28,11 +28,23 @@ class InlineBlockClocklessController : public CLEDController {
 	data_ptr_t mPort;
 	CMinWait<WAIT_TIME> mWait;
 public:
-	virtual int size() { return m_nLeds * __LANES; }
+	virtual int size() { return CLEDController::size() * LANES; }
+
+	virtual void showPixels(PixelController<RGB_ORDER, LANES, PORT_MASK> & pixels) { 
+		mWait.wait();
+		uint32_t clocks = showRGBInternal(pixels);
+		#if FASTLED_ALLOW_INTTERUPTS == 0
+		// Adjust the timer
+		long microsTaken = CLKS_TO_MICROS(clocks);
+		MS_COUNTER += (1 + (microsTaken / 1000));
+		#endif
+
+		mWait.mark();
+	}
 
 	virtual void init() {
 		if(FIRST_PIN == PORTC_FIRST_PIN) { // PORTC
-			switch(LANES) {
+			switch(USED_LANES) {
 				case 12: FastPin<30>::setOutput();
 				case 11: FastPin<29>::setOutput();
 				case 10: FastPin<27>::setOutput();
@@ -47,7 +59,7 @@ public:
 				case 1: FastPin<15>::setOutput();
 			}
 		} else if(FIRST_PIN == PORTD_FIRST_PIN) { // PORTD
-			switch(LANES) {
+			switch(USED_LANES) {
 				case 8: FastPin<5>::setOutput();
 				case 7: FastPin<21>::setOutput();
 				case 6: FastPin<20>::setOutput();
@@ -64,39 +76,6 @@ public:
 
 	virtual uint16_t getMaxRefreshRate() const { return 400; }
 
-	virtual void clearLeds(int nLeds) {
-		showColor(CRGB(0, 0, 0), nLeds, 0);
-	}
-
-	// set all the leds on the controller to a given color
-	virtual void showColor(const struct CRGB & rgbdata, int nLeds, CRGB scale) {
-		PixelController<RGB_ORDER,LANES,PORT_MASK> pixels(rgbdata,nLeds, scale, getDither() );
-		mWait.wait();
-		uint32_t clocks = showRGBInternal(pixels,nLeds);
-		#if FASTLED_ALLOW_INTTERUPTS == 0
-		// Adjust the timer
-		long microsTaken = CLKS_TO_MICROS(clocks);
-		MS_COUNTER += (1 + (microsTaken / 1000));
-		#endif
-
-		mWait.mark();
-	}
-
-	virtual void show(const struct CRGB *rgbdata, int nLeds, CRGB scale) {
-		PixelController<RGB_ORDER,LANES,PORT_MASK> pixels(rgbdata,nLeds, scale, getDither() );
-		mWait.wait();
-		uint32_t clocks = showRGBInternal(pixels,nLeds);
-		#if FASTLED_ALLOW_INTTERUPTS == 0
-		// Adjust the timer
-		long microsTaken = CLKS_TO_MICROS(clocks);
-		MS_COUNTER += (1 + (microsTaken / 1000));
-		#endif
-
-		mWait.mark();
-	}
-
-
-
 	typedef union {
 		uint8_t bytes[12];
 		uint16_t shorts[6];
@@ -105,7 +84,7 @@ public:
 
 	template<int BITS,int PX> __attribute__ ((always_inline)) inline static void writeBits(register uint32_t & next_mark, register Lines & b, PixelController<RGB_ORDER, LANES, PORT_MASK> &pixels) { // , register uint32_t & b2)  {
 		register Lines b2;
-		if(LANES>8) {
+		if(USED_LANES>8) {
 			transpose8<1,2>(b.bytes,b2.bytes);
 			transpose8<1,2>(b.bytes+8,b2.bytes+1);
 		} else {
@@ -114,13 +93,13 @@ public:
 		register uint8_t d = pixels.template getd<PX>(pixels);
 		register uint8_t scale = pixels.template getscale<PX>(pixels);
 
-		for(register uint32_t i = 0; i < (LANES/2); i++) {
+		for(register uint32_t i = 0; i < (USED_LANES/2); i++) {
 			while(ARM_DWT_CYCCNT < next_mark);
 			next_mark = ARM_DWT_CYCCNT + (T1+T2+T3)-3;
 			*FastPin<FIRST_PIN>::sport() = PORT_MASK;
 
 			while((next_mark - ARM_DWT_CYCCNT) > (T2+T3+(2*(F_CPU/24000000))));
-			if(LANES>8) {
+			if(USED_LANES>8) {
 				*FastPin<FIRST_PIN>::cport() = ((~b2.shorts[i]) & PORT_MASK);
 			} else {
 				*FastPin<FIRST_PIN>::cport() = ((~b2.bytes[7-i]) & PORT_MASK);
@@ -130,20 +109,20 @@ public:
 			*FastPin<FIRST_PIN>::cport() = PORT_MASK;
 
 			b.bytes[i] = pixels.template loadAndScale<PX>(pixels,i,d,scale);
-			b.bytes[i+(LANES/2)] = pixels.template loadAndScale<PX>(pixels,i+(LANES/2),d,scale);
+			b.bytes[i+(USED_LANES/2)] = pixels.template loadAndScale<PX>(pixels,i+(USED_LANES/2),d,scale);
 		}
 
 		// if folks use an odd numnber of lanes, get the last byte's value here
-		if(LANES & 0x01) {
-			b.bytes[LANES-1] = pixels.template loadAndScale<PX>(pixels,LANES-1,d,scale);
+		if(USED_LANES & 0x01) {
+			b.bytes[USED_LANES-1] = pixels.template loadAndScale<PX>(pixels,USED_LANES-1,d,scale);
 		}
 
-		for(register uint32_t i = LANES/2; i < 8; i++) {
+		for(register uint32_t i = USED_LANES/2; i < 8; i++) {
 			while(ARM_DWT_CYCCNT < next_mark);
 			next_mark = ARM_DWT_CYCCNT + (T1+T2+T3)-3;
 			*FastPin<FIRST_PIN>::sport() = PORT_MASK;
 			while((next_mark - ARM_DWT_CYCCNT) > (T2+T3+(2*(F_CPU/24000000))));
-			if(LANES>8) {
+			if(USED_LANES>8) {
 				*FastPin<FIRST_PIN>::cport() = ((~b2.shorts[i]) & PORT_MASK);
 			} else {
 				// b2.bytes[0] = 0;
@@ -154,28 +133,13 @@ public:
 			*FastPin<FIRST_PIN>::cport() = PORT_MASK;
 
 		}
-
-
-		// while(ARM_DWT_CYCCNT < next_mark);
-		// next_mark = ARM_DWT_CYCCNT + (T1+T2+T3)-3;
-		// *FastPin<FIRST_PIN>::sport() = PORT_MASK;
-		//
-		// while((next_mark - ARM_DWT_CYCCNT) > (T2+T3+6));
-		// if(LANES>8) {
-		// 	*FastPin<FIRST_PIN>::cport() = ((~b2.shorts[7]) & PORT_MASK);
-		// } else {
-		// 	*FastPin<FIRST_PIN>::cport() = PORT_MASK; // ((~b2.bytes[7-i]) & PORT_MASK);
-		// }
-		//
-		// while((next_mark - ARM_DWT_CYCCNT) > (T3));
-		// *FastPin<FIRST_PIN>::cport() = PORT_MASK;
 	}
 
 
 
 	// 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, LANES, PORT_MASK> &allpixels, int nLeds) {
+		static uint32_t showRGBInternal(PixelController<RGB_ORDER, LANES, PORT_MASK> &allpixels) {
 		// Get access to the clock
 		ARM_DEMCR    |= ARM_DEMCR_TRCENA;
 		ARM_DWT_CTRL |= ARM_DWT_CTRL_CYCCNTENA;
@@ -186,14 +150,14 @@ public:
 		register Lines b0;
 
 		allpixels.preStepFirstByteDithering();
-		for(int i = 0; i < LANES; i++) {
+		for(int i = 0; i < USED_LANES; i++) {
 			b0.bytes[i] = allpixels.loadAndScale0(i);
 		}
 
 		cli();
 		uint32_t next_mark = ARM_DWT_CYCCNT + (T1+T2+T3);
 
-		while(nLeds--) {
+		while(allpixels.has(1)) {
 			#if (FASTLED_ALLOW_INTERRUPTS == 1)
 			cli();
 			// if interrupts took longer than 45µs, punt on the current frame
@@ -221,13 +185,12 @@ public:
 	}
 };
 
-#define DLANES (MIN(__LANES,16))
-#define PMASK ((1<<(DLANES))-1)
+#define PMASK ((1<<(LANES))-1)
 #define PMASK_HI (PMASK>>8 & 0xFF)
 #define PMASK_LO (PMASK & 0xFF)
 
-template <uint8_t __LANES, int T1, int T2, int T3, EOrder RGB_ORDER = GRB, int XTRA0 = 0, bool FLIP = false, int WAIT_TIME = 50>
-class SixteenWayInlineBlockClocklessController : public CLEDController {
+template <uint8_t LANES, int T1, int T2, int T3, EOrder RGB_ORDER = GRB, int XTRA0 = 0, bool FLIP = false, int WAIT_TIME = 50>
+class SixteenWayInlineBlockClocklessController : public CPixelLEDController<RGB_ORDER, LANES, PMASK> {
 	typedef typename FastPin<PORTC_FIRST_PIN>::port_ptr_t data_ptr_t;
 	typedef typename FastPin<PORTC_FIRST_PIN>::port_t data_t;
 
@@ -236,47 +199,41 @@ class SixteenWayInlineBlockClocklessController : public CLEDController {
 	CMinWait<WAIT_TIME> mWait;
 public:
 	virtual void init() {
-				// FastPin<30>::setOutput();
-				// FastPin<29>::setOutput();
-				// FastPin<27>::setOutput();
-				// FastPin<28>::setOutput();
-				switch(DLANES) {
-					case 16: FastPin<12>::setOutput();
-					case 15: FastPin<11>::setOutput();
-					case 14: FastPin<13>::setOutput();
-					case 13: FastPin<10>::setOutput();
-					case 12: FastPin<9>::setOutput();
-					case 11: FastPin<23>::setOutput();
-					case 10: FastPin<22>::setOutput();
-					case 9:  FastPin<15>::setOutput();
-
-					case 8:  FastPin<5>::setOutput();
-					case 7:  FastPin<21>::setOutput();
-					case 6:  FastPin<20>::setOutput();
-					case 5:  FastPin<6>::setOutput();
-					case 4:  FastPin<8>::setOutput();
-					case 3:  FastPin<7>::setOutput();
-					case 2:  FastPin<14>::setOutput();
-					case 1:  FastPin<2>::setOutput();
-				}
-	}
-
-	virtual void clearLeds(int nLeds) {
-		showColor(CRGB(0, 0, 0), nLeds, 0);
+		static_assert(LANES <= 16, "Maximum of 16 lanes for Teensy parallel controllers!");
+		// FastPin<30>::setOutput();
+		// FastPin<29>::setOutput();
+		// FastPin<27>::setOutput();
+		// FastPin<28>::setOutput();
+		switch(LANES) {
+			case 16: FastPin<12>::setOutput();
+			case 15: FastPin<11>::setOutput();
+			case 14: FastPin<13>::setOutput();
+			case 13: FastPin<10>::setOutput();
+			case 12: FastPin<9>::setOutput();
+			case 11: FastPin<23>::setOutput();
+			case 10: FastPin<22>::setOutput();
+			case 9:  FastPin<15>::setOutput();
+
+			case 8:  FastPin<5>::setOutput();
+			case 7:  FastPin<21>::setOutput();
+			case 6:  FastPin<20>::setOutput();
+			case 5:  FastPin<6>::setOutput();
+			case 4:  FastPin<8>::setOutput();
+			case 3:  FastPin<7>::setOutput();
+			case 2:  FastPin<14>::setOutput();
+			case 1:  FastPin<2>::setOutput();
+		}
 	}
 
-	// set all the leds on the controller to a given color
-	virtual void showColor(const struct CRGB & rgbdata, int nLeds, CRGB scale) {
-		PixelController<RGB_ORDER,DLANES,PMASK> pixels(rgbdata,nLeds, scale, getDither() );
+	virtual void showPixels(PixelController<RGB_ORDER, LANES, PMASK> & pixels) { 
 		mWait.wait();
-		showRGBInternal(pixels,nLeds);
-		mWait.mark();
-	}
+		uint32_t clocks = showRGBInternal(pixels);
+		#if FASTLED_ALLOW_INTTERUPTS == 0
+		// Adjust the timer
+		long microsTaken = CLKS_TO_MICROS(clocks);
+		MS_COUNTER += (1 + (microsTaken / 1000));
+		#endif
 
-	virtual void show(const struct CRGB *rgbdata, int nLeds, CRGB scale) {
-		PixelController<RGB_ORDER,DLANES,PMASK> pixels(rgbdata,nLeds, scale, getDither() );
-		mWait.wait();
-		showRGBInternal(pixels,nLeds);
 		mWait.mark();
 	}
 
@@ -286,14 +243,14 @@ public:
 		uint32_t raw[4];
 	} Lines;
 
-	template<int BITS,int PX> __attribute__ ((always_inline)) inline static void writeBits(register uint32_t & next_mark, register Lines & b, PixelController<RGB_ORDER,DLANES, PMASK> &pixels) { // , register uint32_t & b2)  {
+	template<int BITS,int PX> __attribute__ ((always_inline)) inline static void writeBits(register uint32_t & next_mark, register Lines & b, PixelController<RGB_ORDER,LANES, PMASK> &pixels) { // , register uint32_t & b2)  {
 		register Lines b2;
 		transpose8x1(b.bytes,b2.bytes);
 		transpose8x1(b.bytes+8,b2.bytes+8);
 		register uint8_t d = pixels.template getd<PX>(pixels);
 		register uint8_t scale = pixels.template getscale<PX>(pixels);
 
-		for(register uint32_t i = 0; (i < DLANES) && (i < 8); i++) {
+		for(register uint32_t i = 0; (i < LANES) && (i < 8); i++) {
 			while(ARM_DWT_CYCCNT < next_mark);
 			next_mark = ARM_DWT_CYCCNT + (T1+T2+T3)-3;
 			*FastPin<PORTD_FIRST_PIN>::sport() = PMASK_LO;
@@ -308,7 +265,7 @@ public:
 			*FastPin<PORTC_FIRST_PIN>::cport() = PMASK_HI;
 
 			b.bytes[i] = pixels.template loadAndScale<PX>(pixels,i,d,scale);
-			if(DLANES==16 || (DLANES>8 && ((i+8) < DLANES))) {
+			if(LANES==16 || (LANES>8 && ((i+8) < LANES))) {
 				b.bytes[i+8] = pixels.template loadAndScale<PX>(pixels,i+8,d,scale);
 			}
 		}
@@ -318,7 +275,7 @@ public:
 
 	// 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,DLANES, PMASK> &allpixels, int nLeds) {
+		static uint32_t showRGBInternal(PixelController<RGB_ORDER,LANES, PMASK> &allpixels) {
 		// Get access to the clock
 		ARM_DEMCR    |= ARM_DEMCR_TRCENA;
 		ARM_DWT_CTRL |= ARM_DWT_CTRL_CYCCNTENA;
@@ -329,14 +286,14 @@ public:
 		register Lines b0;
 
 		allpixels.preStepFirstByteDithering();
-		for(int i = 0; i < DLANES; i++) {
+		for(int i = 0; i < LANES; i++) {
 			b0.bytes[i] = allpixels.loadAndScale0(i);
 		}
 
 		cli();
 		uint32_t next_mark = ARM_DWT_CYCCNT + (T1+T2+T3);
 
-		while(nLeds--) {
+		while(allpixels.has(1)) {
 			allpixels.stepDithering();
 			#if 0 && (FASTLED_ALLOW_INTERRUPTS == 1)
 			cli();