Initial commit

This commit is contained in:
Cabillot Julien
2018-11-26 00:49:19 +01:00
commit 368127e848
355 changed files with 67999 additions and 0 deletions
@@ -0,0 +1,313 @@
#ifndef __INC_M0_CLOCKLESS_H
#define __INC_M0_CLOCKLESS_H
struct M0ClocklessData {
uint8_t d[3];
uint8_t s[3];
uint8_t e[3];
uint8_t adj;
};
template<int HI_OFFSET, int LO_OFFSET, int T1, int T2, int T3, EOrder RGB_ORDER, int WAIT_TIME>int
showLedData(volatile uint32_t *_port, uint32_t _bitmask, const uint8_t *_leds, uint32_t num_leds, struct M0ClocklessData *pData) {
// Lo register variables
register uint32_t scratch=0;
register struct M0ClocklessData *base = pData;
register volatile uint32_t *port = _port;
register uint32_t d=0;
register uint32_t counter=num_leds;
register uint32_t bn=0;
register uint32_t b=0;
register uint32_t bitmask = _bitmask;
// high register variable
register const uint8_t *leds = _leds;
asm __volatile__ (
///////////////////////////////////////////////////////////////////////////
//
// asm macro definitions - used to assemble the clockless output
//
".ifnotdef fl_delay_def;"
#ifdef FASTLED_ARM_M0_PLUS
" .set fl_is_m0p, 1;"
" .macro m0pad;"
" nop;"
" .endm;"
#else
" .set fl_is_m0p, 0;"
" .macro m0pad;"
" .endm;"
#endif
" .set fl_delay_def, 1;"
" .set fl_delay_mod, 4;"
" .if fl_is_m0p == 1;"
" .set fl_delay_mod, 3;"
" .endif;"
" .macro fl_delay dtime, reg=r0;"
" .if (\\dtime > 0);"
" .set dcycle, (\\dtime / fl_delay_mod);"
" .set dwork, (dcycle * fl_delay_mod);"
" .set drem, (\\dtime - dwork);"
" .rept (drem);"
" nop;"
" .endr;"
" .if dcycle > 0;"
" mov \\reg, #dcycle;"
" delayloop_\\@:;"
" sub \\reg, #1;"
" bne delayloop_\\@;"
" .if fl_is_m0p == 0;"
" nop;"
" .endif;"
" .endif;"
" .endif;"
" .endm;"
" .macro mod_delay dtime,b1,b2,reg;"
" .set adj, (\\b1 + \\b2);"
" .if adj < \\dtime;"
" .set dtime2, (\\dtime - adj);"
" fl_delay dtime2, \\reg;"
" .endif;"
" .endm;"
// check the bit and drop the line low if it isn't set
" .macro qlo4 b,bitmask,port,loff ;"
" lsl \\b, #1 ;"
" bcs skip_\\@ ;"
" str \\bitmask, [\\port, \\loff] ;"
" skip_\\@: ;"
" m0pad;"
" .endm ;"
// set the pin hi or low (determined by the offset passed in )
" .macro qset2 bitmask,port,loff;"
" str \\bitmask, [\\port, \\loff];"
" m0pad;"
" .endm;"
// Load up the next led byte to work with, put it in bn
" .macro loadleds3 leds, bn, rled, scratch;"
" mov \\scratch, \\leds;"
" ldrb \\bn, [\\scratch, \\rled];"
" .endm;"
// check whether or not we should dither
" .macro loaddither7 bn,d,base,rdither;"
" ldrb \\d, [\\base, \\rdither];"
" lsl \\d, #24;" //; shift high for the qadd w/bn
" lsl \\bn, #24;" //; shift high for the qadd w/d
" bne chkskip_\\@;" //; if bn==0, clear d;"
" eor \\d, \\d;" //; clear d;"
" m0pad;"
" chkskip_\\@:;"
" .endm;"
// Do the qadd8 for dithering -- there's two versions of this. The m0 version
// takes advantage of the 3 cycle branch to do two things after the branch,
// while keeping timing constant. The m0+, however, branches in 2 cycles, so
// we have to work around that a bit more. This is one of the few times
// where the m0 will actually be _more_ efficient than the m0+
" .macro dither5 bn,d;"
" .syntax unified;"
" .if fl_is_m0p == 0;"
" adds \\bn, \\d;" // do the add
" bcc dither5_1_\\@;"
" mvns \\bn, \\bn;" // set the low 24bits ot 1's
" lsls \\bn, \\bn, #24;" // move low 8 bits to the high bits
" dither5_1_\\@:;"
" nop;" // nop to keep timing in line
" .else;"
" adds \\bn, \\d;" // do the add"
" bcc dither5_2_\\@;"
" mvns \\bn, \\bn;" // set the low 24bits ot 1's
" dither5_2_\\@:;"
" bcc dither5_3_\\@;"
" lsls \\bn, \\bn, #24;" // move low 8 bits to the high bits
" dither5_3_\\@:;"
" .endif;"
" .syntax divided;"
" .endm;"
// Do our scaling
" .macro scale4 bn, base, scale, scratch;"
" ldrb \\scratch, [\\base, \\scale];"
" lsr \\bn, \\bn, #24;" // bring bn back down to its low 8 bits
" mul \\bn, \\scratch;" // do the multiply
" .endm;"
// swap bn into b
" .macro swapbbn1 b,bn;"
" lsl \\b, \\bn, #16;" // put the 8 bits we want for output high
" .endm;"
// adjust the dithering value for the next time around (load e from memory
// to do the math)
" .macro adjdither7 base,d,rled,eoffset,scratch;"
" ldrb \\d, [\\base, \\rled];"
" ldrb \\scratch,[\\base,\\eoffset];" // load e
" .syntax unified;"
" subs \\d, \\scratch, \\d;" // d=e-d
" .syntax divided;"
" strb \\d, [\\base, \\rled];" // save d
" .endm;"
// increment the led pointer (base+9 has what we're incrementing by)
" .macro incleds3 leds, base, scratch;"
" ldrb \\scratch, [\\base, #9];" // load incremen
" add \\leds, \\leds, \\scratch;" // update leds pointer
" .endm;"
// compare and loop
" .macro cmploop5 counter,label;"
" .syntax unified;"
" subs \\counter, #1;"
" .syntax divided;"
" beq done_\\@;"
" m0pad;"
" b \\label;"
" done_\\@:;"
" .endm;"
" .endif;"
);
#define M0_ASM_ARGS : \
[leds] "+h" (leds), \
[counter] "+l" (counter), \
[scratch] "+l" (scratch), \
[d] "+l" (d), \
[bn] "+l" (bn), \
[b] "+l" (b) \
: \
[port] "l" (port), \
[base] "l" (base), \
[bitmask] "l" (bitmask), \
[hi_off] "I" (HI_OFFSET), \
[lo_off] "I" (LO_OFFSET), \
[led0] "I" (RO(0)), \
[led1] "I" (RO(1)), \
[led2] "I" (RO(2)), \
[scale0] "I" (3+RO(0)), \
[scale1] "I" (3+RO(1)), \
[scale2] "I" (3+RO(2)), \
[e0] "I" (6+RO(0)), \
[e1] "I" (6+RO(1)), \
[e2] "I" (6+RO(2)), \
[T1] "I" (T1), \
[T2] "I" (T2), \
[T3] "I" (T3) \
:
/////////////////////////////////////////////////////////////////////////
// now for some convinience macros to make building our lines a bit cleaner
#define LOOP " loop_%=:"
#define HI2 " qset2 %[bitmask], %[port], %[hi_off];"
#define D1 " mod_delay %c[T1],2,0,%[scratch];"
#define QLO4 " qlo4 %[b],%[bitmask],%[port], %[lo_off];"
#define LOADLEDS3(X) " loadleds3 %[leds], %[bn], %[led" #X "] ,%[scratch];"
#define D2(ADJ) " mod_delay %c[T2],4," #ADJ ",%[scratch];"
#define LO2 " qset2 %[bitmask], %[port], %[lo_off];"
#define D3(ADJ) " mod_delay %c[T3],2," #ADJ ",%[scratch];"
#define LOADDITHER7(X) " loaddither7 %[bn], %[d], %[base], %[led" #X "];"
#define DITHER5 " dither5 %[bn], %[d];"
#define SCALE4(X) " scale4 %[bn], %[base], %[scale" #X "], %[scratch];"
#define SWAPBBN1 " swapbbn1 %[b], %[bn];"
#define ADJDITHER7(X) " adjdither7 %[base],%[d],%[led" #X "],%[e" #X "],%[scratch];"
#define INCLEDS3 " incleds3 %[leds],%[base],%[scratch];"
#define CMPLOOP5 " cmploop5 %[counter], loop_%=;"
#define NOTHING ""
#if !(defined(SEI_CHK) && (FASTLED_ALLOW_INTERRUPTS == 1))
// We're not allowing interrupts - run the entire loop in asm to keep things
// as tight as possible. In an ideal world, we should be pushing out ws281x
// leds (or other 3-wire leds) with zero gaps between pixels.
asm __volatile__ (
// pre-load byte 0
LOADLEDS3(0) LOADDITHER7(0) DITHER5 SCALE4(0) ADJDITHER7(0) SWAPBBN1
// loop over writing out the data
LOOP
// Write out byte 0, prepping byte 1
HI2 D1 QLO4 NOTHING D2(0) LO2 D3(0)
HI2 D1 QLO4 LOADLEDS3(1) D2(3) LO2 D3(0)
HI2 D1 QLO4 LOADDITHER7(1) D2(7) LO2 D3(0)
HI2 D1 QLO4 DITHER5 D2(5) LO2 D3(0)
HI2 D1 QLO4 SCALE4(1) D2(4) LO2 D3(0)
HI2 D1 QLO4 ADJDITHER7(1) D2(7) LO2 D3(0)
HI2 D1 QLO4 NOTHING D2(0) LO2 D3(0)
HI2 D1 QLO4 SWAPBBN1 D2(1) LO2 D3(0)
// Write out byte 1, prepping byte 2
HI2 D1 QLO4 NOTHING D2(0) LO2 D3(0)
HI2 D1 QLO4 LOADLEDS3(2) D2(3) LO2 D3(0)
HI2 D1 QLO4 LOADDITHER7(2) D2(7) LO2 D3(0)
HI2 D1 QLO4 DITHER5 D2(5) LO2 D3(0)
HI2 D1 QLO4 SCALE4(2) D2(4) LO2 D3(0)
HI2 D1 QLO4 ADJDITHER7(2) D2(7) LO2 D3(0)
HI2 D1 QLO4 INCLEDS3 D2(3) LO2 D3(0)
HI2 D1 QLO4 SWAPBBN1 D2(1) LO2 D3(0)
// Write out byte 2, prepping byte 0
HI2 D1 QLO4 NOTHING D2(0) LO2 D3(0)
HI2 D1 QLO4 LOADLEDS3(0) D2(3) LO2 D3(0)
HI2 D1 QLO4 LOADDITHER7(0) D2(7) LO2 D3(0)
HI2 D1 QLO4 DITHER5 D2(5) LO2 D3(0)
HI2 D1 QLO4 SCALE4(0) D2(4) LO2 D3(0)
HI2 D1 QLO4 ADJDITHER7(0) D2(7) LO2 D3(0)
HI2 D1 QLO4 NOTHING D2(0) LO2 D3(0)
HI2 D1 QLO4 SWAPBBN1 D2(1) LO2 D3(5) CMPLOOP5
M0_ASM_ARGS
);
#else
// We're allowing interrupts - track the loop outside the asm code, to allow
// inserting the interrupt overrun checks.
asm __volatile__ (
// pre-load byte 0
LOADLEDS3(0) LOADDITHER7(0) DITHER5 SCALE4(0) ADJDITHER7(0) SWAPBBN1
M0_ASM_ARGS);
do {
asm __volatile__ (
// Write out byte 0, prepping byte 1
HI2 D1 QLO4 NOTHING D2(0) LO2 D3(0)
HI2 D1 QLO4 LOADLEDS3(1) D2(3) LO2 D3(0)
HI2 D1 QLO4 LOADDITHER7(1) D2(7) LO2 D3(0)
HI2 D1 QLO4 DITHER5 D2(5) LO2 D3(0)
HI2 D1 QLO4 SCALE4(1) D2(4) LO2 D3(0)
HI2 D1 QLO4 ADJDITHER7(1) D2(7) LO2 D3(0)
HI2 D1 QLO4 NOTHING D2(0) LO2 D3(0)
HI2 D1 QLO4 SWAPBBN1 D2(1) LO2 D3(0)
// Write out byte 1, prepping byte 2
HI2 D1 QLO4 NOTHING D2(0) LO2 D3(0)
HI2 D1 QLO4 LOADLEDS3(2) D2(3) LO2 D3(0)
HI2 D1 QLO4 LOADDITHER7(2) D2(7) LO2 D3(0)
HI2 D1 QLO4 DITHER5 D2(5) LO2 D3(0)
HI2 D1 QLO4 SCALE4(2) D2(4) LO2 D3(0)
HI2 D1 QLO4 ADJDITHER7(2) D2(7) LO2 D3(0)
HI2 D1 QLO4 INCLEDS3 D2(3) LO2 D3(0)
HI2 D1 QLO4 SWAPBBN1 D2(1) LO2 D3(0)
// Write out byte 2, prepping byte 0
HI2 D1 QLO4 NOTHING D2(0) LO2 D3(0)
HI2 D1 QLO4 LOADLEDS3(0) D2(3) LO2 D3(0)
HI2 D1 QLO4 LOADDITHER7(0) D2(7) LO2 D3(0)
HI2 D1 QLO4 DITHER5 D2(5) LO2 D3(0)
HI2 D1 QLO4 SCALE4(0) D2(4) LO2 D3(0)
HI2 D1 QLO4 ADJDITHER7(0) D2(7) LO2 D3(0)
HI2 D1 QLO4 NOTHING D2(0) LO2 D3(0)
HI2 D1 QLO4 SWAPBBN1 D2(1) LO2 D3(5)
M0_ASM_ARGS
);
SEI_CHK; INNER_SEI; --counter; CLI_CHK;
} while(counter);
#endif
return num_leds;
}
#endif
@@ -0,0 +1,89 @@
#ifndef __INC_CLOCKLESS_ARM_D21
#define __INC_CLOCKLESS_ARM_D21
#include "platforms/arm/common/m0clockless.h"
FASTLED_NAMESPACE_BEGIN
#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>
class ClocklessController : public CLEDController {
typedef typename FastPinBB<DATA_PIN>::port_ptr_t data_ptr_t;
typedef typename FastPinBB<DATA_PIN>::port_t data_t;
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 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> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
cli();
showRGBInternal(pixels);
sei();
mWait.mark();
}
virtual void show(const struct CRGB *rgbdata, int nLeds, CRGB scale) {
PixelController<RGB_ORDER> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
cli();
showRGBInternal(pixels);
sei();
mWait.mark();
}
#ifdef SUPPORT_ARGB
virtual void show(const struct CARGB *rgbdata, int nLeds, CRGB scale) {
PixelController<RGB_ORDER> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
cli();
showRGBInternal(pixels);
sei();
mWait.mark();
}
#endif
// 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();
showLedData<8,4,T1,T2,T3,RGB_ORDER, WAIT_TIME>(portBase, FastPin<DATA_PIN>::mask(), pixels.mData, pixels.mLen, &data);
return 0; // 0x00FFFFFF - _VAL;
}
};
FASTLED_NAMESPACE_END
#endif // __INC_CLOCKLESS_ARM_D21
@@ -0,0 +1,8 @@
#ifndef __INC_FASTLED_ARM_D21_H
#define __INC_FASTLED_ARM_D21_H
#include "fastled_delay.h"
#include "fastpin_arm_d21.h"
#include "clockless_arm_d21.h"
#endif
@@ -0,0 +1,95 @@
#ifndef __INC_FASTPIN_ARM_SAM_H
#define __INC_FASTPIN_ARM_SAM_H
FASTLED_NAMESPACE_BEGIN
#if defined(FASTLED_FORCE_SOFTWARE_PINS)
#warning "Software pin support forced, pin access will be sloightly slower."
#define NO_HARDWARE_PIN_SUPPORT
#undef HAS_HARDWARE_PIN_SUPPORT
#else
/// Template definition for STM32 style ARM pins, providing direct access to the various GPIO registers. Note that this
/// uses the full port GPIO registers. In theory, in some way, bit-band register access -should- be faster, however I have found
/// that something about the way gcc does register allocation results in the bit-band code being slower. It will need more fine tuning.
/// The registers are data output, set output, clear output, toggle output, input, and direction
template<uint8_t PIN, uint8_t _BIT, uint32_t _MASK, int _GRP> class _ARMPIN {
public:
typedef volatile uint32_t * port_ptr_t;
typedef uint32_t port_t;
#if 0
inline static void setOutput() {
if(_BIT<8) {
_CRL::r() = (_CRL::r() & (0xF << (_BIT*4)) | (0x1 << (_BIT*4));
} else {
_CRH::r() = (_CRH::r() & (0xF << ((_BIT-8)*4))) | (0x1 << ((_BIT-8)*4));
}
}
inline static void setInput() { /* TODO */ } // TODO: preform MUX config { _PDDR::r() &= ~_MASK; }
#endif
inline static void setOutput() { pinMode(PIN, OUTPUT); } // TODO: perform MUX config { _PDDR::r() |= _MASK; }
inline static void setInput() { pinMode(PIN, INPUT); } // TODO: preform MUX config { _PDDR::r() &= ~_MASK; }
inline static void hi() __attribute__ ((always_inline)) { PORT->Group[_GRP].OUTSET.reg = _MASK; }
inline static void lo() __attribute__ ((always_inline)) { PORT->Group[_GRP].OUTCLR.reg = _MASK; }
// inline static void lo() __attribute__ ((always_inline)) { PORT->Group[_GRP].BSRR = (_MASK<<16); }
inline static void set(register port_t val) __attribute__ ((always_inline)) { PORT->Group[_GRP].OUT.reg = val; }
inline static void strobe() __attribute__ ((always_inline)) { toggle(); toggle(); }
inline static void toggle() __attribute__ ((always_inline)) { PORT->Group[_GRP].OUTTGL.reg = _MASK; }
inline static void hi(register port_ptr_t port) __attribute__ ((always_inline)) { hi(); }
inline static void lo(register port_ptr_t port) __attribute__ ((always_inline)) { lo(); }
inline static void fastset(register port_ptr_t port, register port_t val) __attribute__ ((always_inline)) { *port = val; }
inline static port_t hival() __attribute__ ((always_inline)) { return PORT->Group[_GRP].OUT.reg | _MASK; }
inline static port_t loval() __attribute__ ((always_inline)) { return PORT->Group[_GRP].OUT.reg & ~_MASK; }
inline static port_ptr_t port() __attribute__ ((always_inline)) { return &PORT->Group[_GRP].OUT.reg; }
inline static port_ptr_t sport() __attribute__ ((always_inline)) { return &PORT->Group[_GRP].OUTSET.reg; }
inline static port_ptr_t cport() __attribute__ ((always_inline)) { return &PORT->Group[_GRP].OUTCLR.reg; }
inline static port_t mask() __attribute__ ((always_inline)) { return _MASK; }
};
#define _R(T) struct __gen_struct_ ## T
#define _RD32(T) struct __gen_struct_ ## T { static __attribute__((always_inline)) inline volatile PortGroup * r() { return T; } };
#define _IO32(L) _RD32(GPIO ## L)
#define _DEFPIN_ARM(PIN, L, BIT) template<> class FastPin<PIN> : public _ARMPIN<PIN, BIT, 1 << BIT, L> {};
// Actual pin definitions
#if defined(ARDUINO_SAMD_ZERO)
#define MAX_PIN 42
_DEFPIN_ARM( 0,0,10); _DEFPIN_ARM( 1,0,11); _DEFPIN_ARM( 2,0, 8); _DEFPIN_ARM( 3,0, 9);
_DEFPIN_ARM( 4,0,14); _DEFPIN_ARM( 5,0,15); _DEFPIN_ARM( 6,0,20); _DEFPIN_ARM( 7,0,21);
_DEFPIN_ARM( 8,0, 6); _DEFPIN_ARM( 9,0, 7); _DEFPIN_ARM(10,0,18); _DEFPIN_ARM(11,0,16);
_DEFPIN_ARM(12,0,19); _DEFPIN_ARM(13,0,17); _DEFPIN_ARM(14,0, 2); _DEFPIN_ARM(15,1, 8);
_DEFPIN_ARM(16,1, 9); _DEFPIN_ARM(17,0, 4); _DEFPIN_ARM(18,0, 5); _DEFPIN_ARM(19,1, 2);
_DEFPIN_ARM(20,0,22); _DEFPIN_ARM(21,0,23); _DEFPIN_ARM(22,0,12); _DEFPIN_ARM(23,1,11);
_DEFPIN_ARM(24,1,10); _DEFPIN_ARM(25,1, 3); _DEFPIN_ARM(26,0,27); _DEFPIN_ARM(27,0,28);
_DEFPIN_ARM(28,0,24); _DEFPIN_ARM(29,0,25); _DEFPIN_ARM(30,1,22); _DEFPIN_ARM(31,1,23);
_DEFPIN_ARM(32,0,22); _DEFPIN_ARM(33,0,23); _DEFPIN_ARM(34,0,19); _DEFPIN_ARM(35,0,16);
_DEFPIN_ARM(36,0,18); _DEFPIN_ARM(37,0,17); _DEFPIN_ARM(38,0,13); _DEFPIN_ARM(39,0,21);
_DEFPIN_ARM(40,0, 6); _DEFPIN_ARM(41,0, 7); _DEFPIN_ARM(42,0, 3);
#define SPI_DATA 24
#define SPI_CLOCK 23
#define HAS_HARDWARE_PIN_SUPPORT
#endif
#endif // FASTLED_FORCE_SOFTWARE_PINS
FASTLED_NAMESPACE_END
#endif // __INC_FASTPIN_ARM_SAM_H
@@ -0,0 +1,26 @@
#ifndef __INC_LED_SYSDEFS_ARM_D21_H
#define __INC_LED_SYSDEFS_ARM_D21_H
#define FASTLED_ARM
#define FASTLED_ARM_M0_PLUS
#ifndef INTERRUPT_THRESHOLD
#define INTERRUPT_THRESHOLD 1
#endif
// Default to allowing interrupts
#ifndef FASTLED_ALLOW_INTERRUPTS
#define FASTLED_ALLOW_INTERRUPTS 0
#endif
#if FASTLED_ALLOW_INTERRUPTS == 1
#define FASTLED_ACCURATE_CLOCK
#endif
// reuseing/abusing cli/sei defs for due
#define cli() __disable_irq();
#define sei() __enable_irq();
#endif
@@ -0,0 +1,145 @@
#ifndef __INC_CLOCKLESS_ARM_K20_H
#define __INC_CLOCKLESS_ARM_K20_H
FASTLED_NAMESPACE_BEGIN
// Definition for a single channel clockless controller for the k20 family of chips, like that used in the teensy 3.0/3.1
// See clockless.h for detailed info on how the template parameters are used.
#if defined(FASTLED_TEENSY3)
#define FASTLED_HAS_CLOCKLESS 1
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 CLEDController {
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;
public:
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 clearLeds(int nLeds) {
showColor(CRGB(0, 0, 0), nLeds, 0);
}
protected:
// 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> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
showRGBInternal(pixels);
mWait.mark();
}
virtual void show(const struct CRGB *rgbdata, int nLeds, CRGB scale) {
PixelController<RGB_ORDER> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
showRGBInternal(pixels);
mWait.mark();
}
#ifdef SUPPORT_ARGB
virtual void show(const struct CARGB *rgbdata, int nLeds, CRGB scale) {
PixelController<RGB_ORDER> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
showRGBInternal(pixels);
mWait.mark();
}
#endif
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 ARM_DWT_CYCCNT; }
}
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;
}
};
#endif
FASTLED_NAMESPACE_END
#endif
@@ -0,0 +1,396 @@
#ifndef __INC_BLOCK_CLOCKLESS_ARM_K20_H
#define __INC_BLOCK_CLOCKLESS_ARM_K20_H
// Definition for a single channel clockless controller for the k20 family of chips, like that used in the teensy 3.0/3.1
// See clockless.h for detailed info on how the template parameters are used.
#if defined(FASTLED_TEENSY3)
#define FASTLED_HAS_BLOCKLESS 1
#define PORTC_FIRST_PIN 15
#define PORTD_FIRST_PIN 2
#define HAS_PORTDC 1
#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))
#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 {
typedef typename FastPin<FIRST_PIN>::port_ptr_t data_ptr_t;
typedef typename FastPin<FIRST_PIN>::port_t data_t;
data_t mPinMask;
data_ptr_t mPort;
CMinWait<WAIT_TIME> mWait;
public:
virtual void init() {
if(FIRST_PIN == PORTC_FIRST_PIN) { // PORTC
switch(LANES) {
case 12: FastPin<30>::setOutput();
case 11: FastPin<29>::setOutput();
case 10: FastPin<27>::setOutput();
case 9: FastPin<28>::setOutput();
case 8: FastPin<12>::setOutput();
case 7: FastPin<11>::setOutput();
case 6: FastPin<13>::setOutput();
case 5: FastPin<10>::setOutput();
case 4: FastPin<9>::setOutput();
case 3: FastPin<23>::setOutput();
case 2: FastPin<22>::setOutput();
case 1: FastPin<15>::setOutput();
}
} else if(FIRST_PIN == PORTD_FIRST_PIN) { // PORTD
switch(LANES) {
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();
}
}
mPinMask = FastPin<FIRST_PIN>::mask();
mPort = FastPin<FIRST_PIN>::port();
}
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) {
MultiPixelController<LANES,PORT_MASK,RGB_ORDER> 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) {
MultiPixelController<LANES,PORT_MASK,RGB_ORDER> 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();
}
#ifdef SUPPORT_ARGB
virtual void show(const struct CARGB *rgbdata, int nLeds, CRGB scale) {
MultiPixelController<LANES,PORT_MASK,RGB_ORDER> 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();
}
#endif
typedef union {
uint8_t bytes[12];
uint16_t shorts[6];
uint32_t raw[3];
} Lines;
template<int BITS,int PX> __attribute__ ((always_inline)) inline static void writeBits(register uint32_t & next_mark, register Lines & b, MultiPixelController<LANES, PORT_MASK, RGB_ORDER> &pixels) { // , register uint32_t & b2) {
register Lines b2;
if(LANES>8) {
transpose8<1,2>(b.bytes,b2.bytes);
transpose8<1,2>(b.bytes+8,b2.bytes+1);
} else {
transpose8x1(b.bytes,b2.bytes);
}
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++) {
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) {
*FastPin<FIRST_PIN>::cport() = ((~b2.shorts[i]) & PORT_MASK);
} else {
*FastPin<FIRST_PIN>::cport() = ((~b2.bytes[7-i]) & PORT_MASK);
}
while((next_mark - ARM_DWT_CYCCNT) > (T3));
*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);
}
// 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);
}
for(register uint32_t i = 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) {
*FastPin<FIRST_PIN>::cport() = ((~b2.shorts[i]) & PORT_MASK);
} else {
// b2.bytes[0] = 0;
*FastPin<FIRST_PIN>::cport() = ((~b2.bytes[7-i]) & PORT_MASK);
}
while((next_mark - ARM_DWT_CYCCNT) > (T3));
*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(MultiPixelController<LANES, PORT_MASK, RGB_ORDER> &allpixels, int nLeds) {
// Get access to the clock
ARM_DEMCR |= ARM_DEMCR_TRCENA;
ARM_DWT_CTRL |= ARM_DWT_CTRL_CYCCNTENA;
ARM_DWT_CYCCNT = 0;
// Setup the pixel controller and load/scale the first byte
allpixels.preStepFirstByteDithering();
register Lines b0;
allpixels.preStepFirstByteDithering();
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--) {
#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-5)*CLKS_PER_US)) { sei(); return ARM_DWT_CYCCNT; }
}
#endif
allpixels.stepDithering();
// Write first byte, read next byte
writeBits<8+XTRA0,1>(next_mark, b0, allpixels);
// Write second byte, read 3rd byte
writeBits<8+XTRA0,2>(next_mark, b0, allpixels);
allpixels.advanceData();
// Write third byte
writeBits<8+XTRA0,0>(next_mark, b0, allpixels);
#if (FASTLED_ALLOW_INTERRUPTS == 1)
sei();
#endif
};
return ARM_DWT_CYCCNT;
}
};
#define DLANES (MIN(__LANES,16))
#define PMASK ((1<<(DLANES))-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 {
typedef typename FastPin<PORTC_FIRST_PIN>::port_ptr_t data_ptr_t;
typedef typename FastPin<PORTC_FIRST_PIN>::port_t data_t;
data_t mPinMask;
data_ptr_t mPort;
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);
}
// set all the leds on the controller to a given color
virtual void showColor(const struct CRGB & rgbdata, int nLeds, CRGB scale) {
MultiPixelController<DLANES,PMASK,RGB_ORDER> pixels(rgbdata,nLeds, scale, getDither() );
mWait.wait();
showRGBInternal(pixels,nLeds);
mWait.mark();
}
virtual void show(const struct CRGB *rgbdata, int nLeds, CRGB scale) {
MultiPixelController<DLANES,PMASK,RGB_ORDER> pixels(rgbdata,nLeds, scale, getDither() );
mWait.wait();
showRGBInternal(pixels,nLeds);
mWait.mark();
}
#ifdef SUPPORT_ARGB
virtual void show(const struct CARGB *rgbdata, int nLeds, CRGB scale) {
MultiPixelController<DLANES,PMASK,RGB_ORDER> pixels(rgbdata,nLeds, scale, getDither() );
mWait.wait();
showRGBInternal(pixels,nLeds);
mWait.mark();
}
#endif
typedef union {
uint8_t bytes[16];
uint16_t shorts[8];
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, MultiPixelController<DLANES, PMASK, RGB_ORDER> &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++) {
while(ARM_DWT_CYCCNT < next_mark);
next_mark = ARM_DWT_CYCCNT + (T1+T2+T3)-3;
*FastPin<PORTD_FIRST_PIN>::sport() = PMASK_LO;
*FastPin<PORTC_FIRST_PIN>::sport() = PMASK_HI;
while((next_mark - ARM_DWT_CYCCNT) > (T2+T3+6));
*FastPin<PORTD_FIRST_PIN>::cport() = ((~b2.bytes[7-i]) & PMASK_LO);
*FastPin<PORTC_FIRST_PIN>::cport() = ((~b2.bytes[15-i]) & PMASK_HI);
while((next_mark - ARM_DWT_CYCCNT) > (T3));
*FastPin<PORTD_FIRST_PIN>::cport() = PMASK_LO;
*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))) {
b.bytes[i+8] = pixels.template loadAndScale<PX>(pixels,i+8,d,scale);
}
}
}
// 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(MultiPixelController<DLANES, PMASK, RGB_ORDER> &allpixels, int nLeds) {
// Get access to the clock
ARM_DEMCR |= ARM_DEMCR_TRCENA;
ARM_DWT_CTRL |= ARM_DWT_CTRL_CYCCNTENA;
ARM_DWT_CYCCNT = 0;
// Setup the pixel controller and load/scale the first byte
allpixels.preStepFirstByteDithering();
register Lines b0;
allpixels.preStepFirstByteDithering();
for(int i = 0; i < DLANES; i++) {
b0.bytes[i] = allpixels.loadAndScale0(i);
}
cli();
uint32_t next_mark = ARM_DWT_CYCCNT + (T1+T2+T3);
while(nLeds--) {
allpixels.stepDithering();
#if 0 && (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 ARM_DWT_CYCCNT; }
}
#endif
// Write first byte, read next byte
writeBits<8+XTRA0,1>(next_mark, b0, allpixels);
// Write second byte, read 3rd byte
writeBits<8+XTRA0,2>(next_mark, b0, allpixels);
allpixels.advanceData();
// Write third byte
writeBits<8+XTRA0,0>(next_mark, b0, allpixels);
#if 0 && (FASTLED_ALLOW_INTERRUPTS == 1)
sei();
#endif
};
sei();
return ARM_DWT_CYCCNT;
}
};
FASTLED_NAMESPACE_END
#endif
#endif
@@ -0,0 +1,14 @@
#ifndef __INC_FASTLED_ARM_K20_H
#define __INC_FASTLED_ARM_K20_H
// Include the k20 headers
#include "bitswap.h"
#include "fastled_delay.h"
#include "fastpin_arm_k20.h"
#include "fastspi_arm_k20.h"
#include "octows2811_controller.h"
#include "smartmatrix_t3.h"
#include "clockless_arm_k20.h"
#include "clockless_block_arm_k20.h"
#endif
@@ -0,0 +1,120 @@
#ifndef __FASTPIN_ARM_K20_H
#define __FASTPIN_ARM_K20_H
FASTLED_NAMESPACE_BEGIN
#if defined(FASTLED_FORCE_SOFTWARE_PINS)
#warning "Software pin support forced, pin access will be sloightly slower."
#define NO_HARDWARE_PIN_SUPPORT
#undef HAS_HARDWARE_PIN_SUPPORT
#else
/// Template definition for teensy 3.0 style ARM pins, providing direct access to the various GPIO registers. Note that this
/// uses the full port GPIO registers. In theory, in some way, bit-band register access -should- be faster, however I have found
/// that something about the way gcc does register allocation results in the bit-band code being slower. It will need more fine tuning.
/// The registers are data output, set output, clear output, toggle output, input, and direction
template<uint8_t PIN, uint32_t _MASK, typename _PDOR, typename _PSOR, typename _PCOR, typename _PTOR, typename _PDIR, typename _PDDR> class _ARMPIN {
public:
typedef volatile uint32_t * port_ptr_t;
typedef uint32_t port_t;
inline static void setOutput() { pinMode(PIN, OUTPUT); } // TODO: perform MUX config { _PDDR::r() |= _MASK; }
inline static void setInput() { pinMode(PIN, INPUT); } // TODO: preform MUX config { _PDDR::r() &= ~_MASK; }
inline static void hi() __attribute__ ((always_inline)) { _PSOR::r() = _MASK; }
inline static void lo() __attribute__ ((always_inline)) { _PCOR::r() = _MASK; }
inline static void set(register port_t val) __attribute__ ((always_inline)) { _PDOR::r() = val; }
inline static void strobe() __attribute__ ((always_inline)) { toggle(); toggle(); }
inline static void toggle() __attribute__ ((always_inline)) { _PTOR::r() = _MASK; }
inline static void hi(register port_ptr_t port) __attribute__ ((always_inline)) { hi(); }
inline static void lo(register port_ptr_t port) __attribute__ ((always_inline)) { lo(); }
inline static void fastset(register port_ptr_t port, register port_t val) __attribute__ ((always_inline)) { *port = val; }
inline static port_t hival() __attribute__ ((always_inline)) { return _PDOR::r() | _MASK; }
inline static port_t loval() __attribute__ ((always_inline)) { return _PDOR::r() & ~_MASK; }
inline static port_ptr_t port() __attribute__ ((always_inline)) { return &_PDOR::r(); }
inline static port_ptr_t sport() __attribute__ ((always_inline)) { return &_PSOR::r(); }
inline static port_ptr_t cport() __attribute__ ((always_inline)) { return &_PCOR::r(); }
inline static port_t mask() __attribute__ ((always_inline)) { return _MASK; }
};
/// Template definition for teensy 3.0 style ARM pins using bit banding, providing direct access to the various GPIO registers. GCC
/// does a poor job of optimizing around these accesses so they are not being used just yet.
template<uint8_t PIN, int _BIT, typename _PDOR, typename _PSOR, typename _PCOR, typename _PTOR, typename _PDIR, typename _PDDR> class _ARMPIN_BITBAND {
public:
typedef volatile uint32_t * port_ptr_t;
typedef uint32_t port_t;
inline static void setOutput() { pinMode(PIN, OUTPUT); } // TODO: perform MUX config { _PDDR::r() |= _MASK; }
inline static void setInput() { pinMode(PIN, INPUT); } // TODO: preform MUX config { _PDDR::r() &= ~_MASK; }
inline static void hi() __attribute__ ((always_inline)) { *_PDOR::template rx<_BIT>() = 1; }
inline static void lo() __attribute__ ((always_inline)) { *_PDOR::template rx<_BIT>() = 0; }
inline static void set(register port_t val) __attribute__ ((always_inline)) { *_PDOR::template rx<_BIT>() = val; }
inline static void strobe() __attribute__ ((always_inline)) { toggle(); toggle(); }
inline static void toggle() __attribute__ ((always_inline)) { *_PTOR::template rx<_BIT>() = 1; }
inline static void hi(register port_ptr_t port) __attribute__ ((always_inline)) { hi(); }
inline static void lo(register port_ptr_t port) __attribute__ ((always_inline)) { lo(); }
inline static void fastset(register port_ptr_t port, register port_t val) __attribute__ ((always_inline)) { *_PDOR::template rx<_BIT>() = val; }
inline static port_t hival() __attribute__ ((always_inline)) { return 1; }
inline static port_t loval() __attribute__ ((always_inline)) { return 0; }
inline static port_ptr_t port() __attribute__ ((always_inline)) { return _PDOR::template rx<_BIT>(); }
inline static port_t mask() __attribute__ ((always_inline)) { return 1; }
};
// Macros for k20 pin access/definition
#define GPIO_BITBAND_ADDR(reg, bit) (((uint32_t)&(reg) - 0x40000000) * 32 + (bit) * 4 + 0x42000000)
#define GPIO_BITBAND_PTR(reg, bit) ((uint32_t *)GPIO_BITBAND_ADDR((reg), (bit)))
#define _R(T) struct __gen_struct_ ## T
#define _RD32(T) struct __gen_struct_ ## T { static __attribute__((always_inline)) inline reg32_t r() { return T; } \
template<int BIT> static __attribute__((always_inline)) inline ptr_reg32_t rx() { return GPIO_BITBAND_PTR(T, BIT); } };
#define _IO32(L) _RD32(GPIO ## L ## _PDOR); _RD32(GPIO ## L ## _PSOR); _RD32(GPIO ## L ## _PCOR); _RD32(GPIO ## L ## _PTOR); _RD32(GPIO ## L ## _PDIR); _RD32(GPIO ## L ## _PDDR);
#define _DEFPIN_ARM(PIN, BIT, L) template<> class FastPin<PIN> : public _ARMPIN<PIN, 1 << BIT, _R(GPIO ## L ## _PDOR), _R(GPIO ## L ## _PSOR), _R(GPIO ## L ## _PCOR), \
_R(GPIO ## L ## _PTOR), _R(GPIO ## L ## _PDIR), _R(GPIO ## L ## _PDDR)> {}; \
template<> class FastPinBB<PIN> : public _ARMPIN_BITBAND<PIN, BIT, _R(GPIO ## L ## _PDOR), _R(GPIO ## L ## _PSOR), _R(GPIO ## L ## _PCOR), \
_R(GPIO ## L ## _PTOR), _R(GPIO ## L ## _PDIR), _R(GPIO ## L ## _PDDR)> {};
// Actual pin definitions
#if defined(FASTLED_TEENSY3) && defined(CORE_TEENSY)
_IO32(A); _IO32(B); _IO32(C); _IO32(D); _IO32(E);
#define MAX_PIN 33
_DEFPIN_ARM(0, 16, B); _DEFPIN_ARM(1, 17, B); _DEFPIN_ARM(2, 0, D); _DEFPIN_ARM(3, 12, A);
_DEFPIN_ARM(4, 13, A); _DEFPIN_ARM(5, 7, D); _DEFPIN_ARM(6, 4, D); _DEFPIN_ARM(7, 2, D);
_DEFPIN_ARM(8, 3, D); _DEFPIN_ARM(9, 3, C); _DEFPIN_ARM(10, 4, C); _DEFPIN_ARM(11, 6, C);
_DEFPIN_ARM(12, 7, C); _DEFPIN_ARM(13, 5, C); _DEFPIN_ARM(14, 1, D); _DEFPIN_ARM(15, 0, C);
_DEFPIN_ARM(16, 0, B); _DEFPIN_ARM(17, 1, B); _DEFPIN_ARM(18, 3, B); _DEFPIN_ARM(19, 2, B);
_DEFPIN_ARM(20, 5, D); _DEFPIN_ARM(21, 6, D); _DEFPIN_ARM(22, 1, C); _DEFPIN_ARM(23, 2, C);
_DEFPIN_ARM(24, 5, A); _DEFPIN_ARM(25, 19, B); _DEFPIN_ARM(26, 1, E); _DEFPIN_ARM(27, 9, C);
_DEFPIN_ARM(28, 8, C); _DEFPIN_ARM(29, 10, C); _DEFPIN_ARM(30, 11, C); _DEFPIN_ARM(31, 0, E);
_DEFPIN_ARM(32, 18, B); _DEFPIN_ARM(33, 4, A);
#define SPI_DATA 11
#define SPI_CLOCK 13
#define SPI1 (*(SPI_t *)0x4002D000)
#define SPI2_DATA 7
#define SPI2_CLOCK 14
#define FASTLED_TEENSY3
#define ARM_HARDWARE_SPI
#define HAS_HARDWARE_PIN_SUPPORT
#endif
#endif // FASTLED_FORCE_SOFTWARE_PINS
FASTLED_NAMESPACE_END
#endif // __INC_FASTPIN_ARM_K20
@@ -0,0 +1,446 @@
#ifndef __INC_FASTSPI_ARM_H
#define __INC_FASTSPI_ARM_H
FASTLED_NAMESPACE_BEGIN
#if defined(FASTLED_TEENSY3) && defined(CORE_TEENSY)
// Version 1.20 renamed SPI_t to KINETISK_SPI_t
#if TEENSYDUINO >= 120
#define SPI_t KINETISK_SPI_t
#endif
#ifndef KINETISK_SPI0
#define KINETISK_SPI0 SPI0
#endif
#ifndef SPI_PUSHR_CONT
#define SPI_PUSHR_CONT SPIX.PUSHR_CONT
#define SPI_PUSHR_CTAS(X) SPIX.PUSHR_CTAS(X)
#define SPI_PUSHR_EOQ SPIX.PUSHR_EOQ
#define SPI_PUSHR_CTCNT SPIX.PUSHR_CTCNT
#define SPI_PUSHR_PCS(X) SPIX.PUSHR_PCS(X)
#endif
// Template function that, on compilation, expands to a constant representing the highest bit set in a byte. Right now,
// if no bits are set (value is 0), it returns 0, which is also the value returned if the lowest bit is the only bit
// set (the zero-th bit). Unclear if I will want this to change at some point.
template<int VAL, int BIT> class BitWork {
public:
static int highestBit() __attribute__((always_inline)) { return (VAL & 1 << BIT) ? BIT : BitWork<VAL, BIT-1>::highestBit(); }
};
template<int VAL> class BitWork<VAL, 0> {
public:
static int highestBit() __attribute__((always_inline)) { return 0; }
};
#define MAX(A, B) (( (A) > (B) ) ? (A) : (B))
#define USE_CONT 0
// intra-frame backup data
struct SPIState {
uint32_t _ctar0,_ctar1;
uint32_t pins[4];
};
// extern SPIState gState;
// Templated function to translate a clock divider value into the prescalar, scalar, and clock doubling setting for the world.
template <int VAL> void getScalars(uint32_t & preScalar, uint32_t & scalar, uint32_t & dbl) {
switch(VAL) {
// Handle the dbl clock cases
case 0: case 1:
case 2: preScalar = 0; scalar = 0; dbl = 1; break;
case 3: preScalar = 1; scalar = 0; dbl = 1; break;
case 5: preScalar = 2; scalar = 0; dbl = 1; break;
case 7: preScalar = 3; scalar = 0; dbl = 1; break;
// Handle the scalar value 6 cases (since it's not a power of two, it won't get caught
// below)
case 9: preScalar = 1; scalar = 2; dbl = 1; break;
case 18: case 19: preScalar = 1; scalar = 2; dbl = 0; break;
case 15: preScalar = 2; scalar = 2; dbl = 1; break;
case 30: case 31: preScalar = 2; scalar = 2; dbl = 0; break;
case 21: case 22: case 23: preScalar = 3; scalar = 2; dbl = 1; break;
case 42: case 43: case 44: case 45: case 46: case 47: preScalar = 3; scalar = 2; dbl = 0; break;
default: {
int p2 = BitWork<VAL/2, 15>::highestBit();
int p3 = BitWork<VAL/3, 15>::highestBit();
int p5 = BitWork<VAL/5, 15>::highestBit();
int p7 = BitWork<VAL/7, 15>::highestBit();
int w2 = 2 * (1 << p2);
int w3 = (VAL/3) > 0 ? 3 * (1 << p3) : 0;
int w5 = (VAL/5) > 0 ? 5 * (1 << p5) : 0;
int w7 = (VAL/7) > 0 ? 7 * (1 << p7) : 0;
int maxval = MAX(MAX(w2, w3), MAX(w5, w7));
if(w2 == maxval) { preScalar = 0; scalar = p2; }
else if(w3 == maxval) { preScalar = 1; scalar = p3; }
else if(w5 == maxval) { preScalar = 2; scalar = p5; }
else if(w7 == maxval) { preScalar = 3; scalar = p7; }
dbl = 0;
if(scalar == 0) { dbl = 1; }
else if(scalar < 3) { scalar--; }
}
}
return;
}
#define SPIX (*(SPI_t*)pSPIX)
template <uint8_t _DATA_PIN, uint8_t _CLOCK_PIN, uint8_t _SPI_CLOCK_DIVIDER, uint32_t pSPIX>
class ARMHardwareSPIOutput {
Selectable *m_pSelect;
SPIState gState;
// Borrowed from the teensy3 SPSR emulation code -- note, enabling pin 7 disables pin 11 (and vice versa),
// and likewise enabling pin 14 disables pin 13 (and vice versa)
inline void enable_pins(void) __attribute__((always_inline)) {
//serial_print("enable_pins\n");
switch(_DATA_PIN) {
case 7:
CORE_PIN7_CONFIG = PORT_PCR_DSE | PORT_PCR_MUX(2);
CORE_PIN11_CONFIG = PORT_PCR_SRE | PORT_PCR_DSE | PORT_PCR_MUX(1);
break;
case 11:
CORE_PIN11_CONFIG = PORT_PCR_DSE | PORT_PCR_MUX(2);
CORE_PIN7_CONFIG = PORT_PCR_SRE | PORT_PCR_DSE | PORT_PCR_MUX(1);
break;
}
switch(_CLOCK_PIN) {
case 13:
CORE_PIN13_CONFIG = PORT_PCR_DSE | PORT_PCR_MUX(2);
CORE_PIN14_CONFIG = PORT_PCR_SRE | PORT_PCR_DSE | PORT_PCR_MUX(1);
break;
case 14:
CORE_PIN14_CONFIG = PORT_PCR_DSE | PORT_PCR_MUX(2);
CORE_PIN13_CONFIG = PORT_PCR_SRE | PORT_PCR_DSE | PORT_PCR_MUX(1);
break;
}
}
// Borrowed from the teensy3 SPSR emulation code. We disable the pins that we're using, and restore the state on the pins that we aren't using
inline void disable_pins(void) __attribute__((always_inline)) {
switch(_DATA_PIN) {
case 7: CORE_PIN7_CONFIG = PORT_PCR_SRE | PORT_PCR_DSE | PORT_PCR_MUX(1); CORE_PIN11_CONFIG = gState.pins[1]; break;
case 11: CORE_PIN11_CONFIG = PORT_PCR_SRE | PORT_PCR_DSE | PORT_PCR_MUX(1); CORE_PIN7_CONFIG = gState.pins[0]; break;
}
switch(_CLOCK_PIN) {
case 13: CORE_PIN13_CONFIG = PORT_PCR_SRE | PORT_PCR_DSE | PORT_PCR_MUX(1); CORE_PIN14_CONFIG = gState.pins[3]; break;
case 14: CORE_PIN14_CONFIG = PORT_PCR_SRE | PORT_PCR_DSE | PORT_PCR_MUX(1); CORE_PIN13_CONFIG = gState.pins[2]; break;
}
}
static inline void update_ctars(uint32_t ctar0, uint32_t ctar1) __attribute__((always_inline)) {
if(SPIX.CTAR0 == ctar0 && SPIX.CTAR1 == ctar1) return;
uint32_t mcr = SPIX.MCR;
if(mcr & SPI_MCR_MDIS) {
SPIX.CTAR0 = ctar0;
SPIX.CTAR1 = ctar1;
} else {
SPIX.MCR = mcr | SPI_MCR_MDIS | SPI_MCR_HALT;
SPIX.CTAR0 = ctar0;
SPIX.CTAR1 = ctar1;
SPIX.MCR = mcr;
}
}
static inline void update_ctar0(uint32_t ctar) __attribute__((always_inline)) {
if (SPIX.CTAR0 == ctar) return;
uint32_t mcr = SPIX.MCR;
if (mcr & SPI_MCR_MDIS) {
SPIX.CTAR0 = ctar;
} else {
SPIX.MCR = mcr | SPI_MCR_MDIS | SPI_MCR_HALT;
SPIX.CTAR0 = ctar;
SPIX.MCR = mcr;
}
}
static inline void update_ctar1(uint32_t ctar) __attribute__((always_inline)) {
if (SPIX.CTAR1 == ctar) return;
uint32_t mcr = SPIX.MCR;
if (mcr & SPI_MCR_MDIS) {
SPIX.CTAR1 = ctar;
} else {
SPIX.MCR = mcr | SPI_MCR_MDIS | SPI_MCR_HALT;
SPIX.CTAR1 = ctar;
SPIX.MCR = mcr;
}
}
void setSPIRate() {
// Configure CTAR0, defaulting to 8 bits and CTAR1, defaulting to 16 bits
uint32_t _PBR = 0;
uint32_t _BR = 0;
uint32_t _CSSCK = 0;
uint32_t _DBR = 0;
// if(_SPI_CLOCK_DIVIDER >= 256) { _PBR = 0; _BR = _CSSCK = 7; _DBR = 0; } // osc/256
// else if(_SPI_CLOCK_DIVIDER >= 128) { _PBR = 0; _BR = _CSSCK = 6; _DBR = 0; } // osc/128
// else if(_SPI_CLOCK_DIVIDER >= 64) { _PBR = 0; _BR = _CSSCK = 5; _DBR = 0; } // osc/64
// else if(_SPI_CLOCK_DIVIDER >= 32) { _PBR = 0; _BR = _CSSCK = 4; _DBR = 0; } // osc/32
// else if(_SPI_CLOCK_DIVIDER >= 16) { _PBR = 0; _BR = _CSSCK = 3; _DBR = 0; } // osc/16
// else if(_SPI_CLOCK_DIVIDER >= 8) { _PBR = 0; _BR = _CSSCK = 1; _DBR = 0; } // osc/8
// else if(_SPI_CLOCK_DIVIDER >= 7) { _PBR = 3; _BR = _CSSCK = 0; _DBR = 1; } // osc/7
// else if(_SPI_CLOCK_DIVIDER >= 5) { _PBR = 2; _BR = _CSSCK = 0; _DBR = 1; } // osc/5
// else if(_SPI_CLOCK_DIVIDER >= 4) { _PBR = 0; _BR = _CSSCK = 0; _DBR = 0; } // osc/4
// else if(_SPI_CLOCK_DIVIDER >= 3) { _PBR = 1; _BR = _CSSCK = 0; _DBR = 1; } // osc/3
// else { _PBR = 0; _BR = _CSSCK = 0; _DBR = 1; } // osc/2
getScalars<_SPI_CLOCK_DIVIDER>(_PBR, _BR, _DBR);
_CSSCK = _BR;
uint32_t ctar0 = SPI_CTAR_FMSZ(7) | SPI_CTAR_PBR(_PBR) | SPI_CTAR_BR(_BR) | SPI_CTAR_CSSCK(_CSSCK);
uint32_t ctar1 = SPI_CTAR_FMSZ(15) | SPI_CTAR_PBR(_PBR) | SPI_CTAR_BR(_BR) | SPI_CTAR_CSSCK(_CSSCK);
#if USE_CONT == 1
ctar0 |= SPI_CTAR_CPHA | SPI_CTAR_CPOL;
ctar1 |= SPI_CTAR_CPHA | SPI_CTAR_CPOL;
#endif
if(_DBR) {
ctar0 |= SPI_CTAR_DBR;
ctar1 |= SPI_CTAR_DBR;
}
update_ctars(ctar0,ctar1);
}
void inline save_spi_state() __attribute__ ((always_inline)) {
// save ctar data
gState._ctar0 = SPIX.CTAR0;
gState._ctar1 = SPIX.CTAR1;
// save data for the not-us pins
gState.pins[0] = CORE_PIN7_CONFIG;
gState.pins[1] = CORE_PIN11_CONFIG;
gState.pins[2] = CORE_PIN13_CONFIG;
gState.pins[3] = CORE_PIN14_CONFIG;
}
void inline restore_spi_state() __attribute__ ((always_inline)) {
// restore ctar data
update_ctars(gState._ctar0,gState._ctar1);
// restore data for the not-us pins (not necessary because disable_pins will do this)
// CORE_PIN7_CONFIG = gState.pins[0];
// CORE_PIN11_CONFIG = gState.pins[1];
// CORE_PIN13_CONFIG = gState.pins[2];
// CORE_PIN14_CONFIG = gState.pins[3];
}
public:
ARMHardwareSPIOutput() { m_pSelect = NULL; }
ARMHardwareSPIOutput(Selectable *pSelect) { m_pSelect = pSelect; }
void setSelect(Selectable *pSelect) { m_pSelect = pSelect; }
void init() {
// set the pins to output
FastPin<_DATA_PIN>::setOutput();
FastPin<_CLOCK_PIN>::setOutput();
// Enable SPI0 clock
uint32_t sim6 = SIM_SCGC6;
if((SPI_t*)pSPIX == &KINETISK_SPI0) {
if (!(sim6 & SIM_SCGC6_SPI0)) {
//serial_print("init1\n");
SIM_SCGC6 = sim6 | SIM_SCGC6_SPI0;
SPIX.CTAR0 = SPI_CTAR_FMSZ(7) | SPI_CTAR_PBR(1) | SPI_CTAR_BR(1);
}
} else if((SPI_t*)pSPIX == &SPI1) {
if (!(sim6 & SIM_SCGC6_SPI1)) {
//serial_print("init1\n");
SIM_SCGC6 = sim6 | SIM_SCGC6_SPI1;
SPIX.CTAR0 = SPI_CTAR_FMSZ(7) | SPI_CTAR_PBR(1) | SPI_CTAR_BR(1);
}
}
// Configure SPI as the master and enable
SPIX.MCR |= SPI_MCR_MSTR; // | SPI_MCR_CONT_SCKE);
SPIX.MCR &= ~(SPI_MCR_MDIS | SPI_MCR_HALT);
// pin/spi configuration happens on select
}
static void waitFully() __attribute__((always_inline)) {
while( (SPIX.SR & 0xF000) > 0);
while (!(SPIX.SR & SPI_SR_TCF));
SPIX.SR |= (SPI_SR_TCF | SPI_SR_EOQF);
}
static bool needwait() __attribute__((always_inline)) { return (SPIX.SR & 0x4000); }
static void wait() __attribute__((always_inline)) { while( (SPIX.SR & 0x4000) ); }
static void wait1() __attribute__((always_inline)) { while( (SPIX.SR & 0xF000) >= 0x2000); }
enum ECont { CONT, NOCONT };
enum EWait { PRE, POST, NONE };
enum ELast { NOTLAST, LAST };
#if USE_CONT == 1
#define CM CONT
#else
#define CM NOCONT
#endif
#define WM PRE
template<ECont CONT_STATE, EWait WAIT_STATE, ELast LAST_STATE> class Write {
public:
static void writeWord(uint16_t w) __attribute__((always_inline)) {
if(WAIT_STATE == PRE) { wait(); }
SPIX.PUSHR = ((LAST_STATE == LAST) ? SPI_PUSHR_EOQ : 0) |
((CONT_STATE == CONT) ? SPI_PUSHR_CONT : 0) |
SPI_PUSHR_CTAS(1) | (w & 0xFFFF);
if(WAIT_STATE == POST) { wait(); }
}
static void writeByte(uint8_t b) __attribute__((always_inline)) {
if(WAIT_STATE == PRE) { wait(); }
SPIX.PUSHR = ((LAST_STATE == LAST) ? SPI_PUSHR_EOQ : 0) |
((CONT_STATE == CONT) ? SPI_PUSHR_CONT : 0) |
SPI_PUSHR_CTAS(0) | (b & 0xFF);
if(WAIT_STATE == POST) { wait(); }
}
};
static void writeWord(uint16_t w) __attribute__((always_inline)) { wait(); SPIX.PUSHR = SPI_PUSHR_CTAS(1) | (w & 0xFFFF); }
static void writeWordNoWait(uint16_t w) __attribute__((always_inline)) { SPIX.PUSHR = SPI_PUSHR_CTAS(1) | (w & 0xFFFF); }
static void writeByte(uint8_t b) __attribute__((always_inline)) { wait(); SPIX.PUSHR = SPI_PUSHR_CTAS(0) | (b & 0xFF); }
static void writeBytePostWait(uint8_t b) __attribute__((always_inline)) { SPIX.PUSHR = SPI_PUSHR_CTAS(0) | (b & 0xFF); wait(); }
static void writeByteNoWait(uint8_t b) __attribute__((always_inline)) { SPIX.PUSHR = SPI_PUSHR_CTAS(0) | (b & 0xFF); }
static void writeWordCont(uint16_t w) __attribute__((always_inline)) { wait(); SPIX.PUSHR = SPI_PUSHR_CONT | SPI_PUSHR_CTAS(1) | (w & 0xFFFF); }
static void writeWordContNoWait(uint16_t w) __attribute__((always_inline)) { SPIX.PUSHR = SPI_PUSHR_CONT | SPI_PUSHR_CTAS(1) | (w & 0xFFFF); }
static void writeByteCont(uint8_t b) __attribute__((always_inline)) { wait(); SPIX.PUSHR = SPI_PUSHR_CONT | SPI_PUSHR_CTAS(0) | (b & 0xFF); }
static void writeByteContPostWait(uint8_t b) __attribute__((always_inline)) { SPIX.PUSHR = SPI_PUSHR_CONT | SPI_PUSHR_CTAS(0) | (b & 0xFF); wait(); }
static void writeByteContNoWait(uint8_t b) __attribute__((always_inline)) { SPIX.PUSHR = SPI_PUSHR_CONT | SPI_PUSHR_CTAS(0) | (b & 0xFF); }
// not the most efficient mechanism in the world - but should be enough for sm16716 and friends
template <uint8_t BIT> inline static void writeBit(uint8_t b) {
uint32_t ctar1_save = SPIX.CTAR1;
// Clear out the FMSZ bits, reset them for 1 bit transferd for the start bit
uint32_t ctar1 = (ctar1_save & (~SPI_CTAR_FMSZ(15))) | SPI_CTAR_FMSZ(0);
update_ctar1(ctar1);
writeWord( (b & (1 << BIT)) != 0);
update_ctar1(ctar1_save);
}
void inline select() __attribute__((always_inline)) {
save_spi_state();
if(m_pSelect != NULL) { m_pSelect->select(); }
setSPIRate();
enable_pins();
}
void inline release() __attribute__((always_inline)) {
disable_pins();
if(m_pSelect != NULL) { m_pSelect->release(); }
restore_spi_state();
}
static void writeBytesValueRaw(uint8_t value, int len) {
while(len--) { Write<CM, WM, NOTLAST>::writeByte(value); }
}
void writeBytesValue(uint8_t value, int len) {
select();
while(len--) {
writeByte(value);
}
waitFully();
release();
}
// Write a block of n uint8_ts out
template <class D> void writeBytes(register uint8_t *data, int len) {
uint8_t *end = data + len;
select();
// could be optimized to write 16bit words out instead of 8bit bytes
while(data != end) {
writeByte(D::adjust(*data++));
}
D::postBlock(len);
waitFully();
release();
}
void writeBytes(register uint8_t *data, int len) { writeBytes<DATA_NOP>(data, len); }
// write a block of uint8_ts out in groups of three. len is the total number of uint8_ts to write out. The template
// parameters indicate how many uint8_ts to skip at the beginning and/or end of each grouping
template <uint8_t FLAGS, class D, EOrder RGB_ORDER> void writePixels(PixelController<RGB_ORDER> pixels) {
select();
int len = pixels.mLen;
// Setup the pixel controller
if((FLAGS & FLAG_START_BIT) == 0) {
//If no start bit stupiditiy, write out as many 16-bit blocks as we can
while(pixels.has(2)) {
// Load and write out the first two bytes
if(WM == NONE) { wait1(); }
Write<CM, WM, NOTLAST>::writeWord(D::adjust(pixels.loadAndScale0()) << 8 | D::adjust(pixels.loadAndScale1()));
// Load and write out the next two bytes (step dithering, advance data in between since we
// cross pixels here)
Write<CM, WM, NOTLAST>::writeWord(D::adjust(pixels.loadAndScale2()) << 8 | D::adjust(pixels.stepAdvanceAndLoadAndScale0()));
// Load and write out the next two bytes
Write<CM, WM, NOTLAST>::writeWord(D::adjust(pixels.loadAndScale1()) << 8 | D::adjust(pixels.loadAndScale2()));
pixels.stepDithering();
pixels.advanceData();
}
if(pixels.has(1)) {
if(WM == NONE) { wait1(); }
// write out the rest as alternating 16/8-bit blocks (likely to be just one)
Write<CM, WM, NOTLAST>::writeWord(D::adjust(pixels.loadAndScale0()) << 8 | D::adjust(pixels.loadAndScale1()));
Write<CM, WM, NOTLAST>::writeByte(D::adjust(pixels.loadAndScale2()));
}
D::postBlock(len);
waitFully();
} else if(FLAGS & FLAG_START_BIT) {
uint32_t ctar1_save = SPIX.CTAR1;
// Clear out the FMSZ bits, reset them for 9 bits transferd for the start bit
uint32_t ctar1 = (ctar1_save & (~SPI_CTAR_FMSZ(15))) | SPI_CTAR_FMSZ(8);
update_ctar1(ctar1);
while(pixels.has(1)) {
writeWord( 0x100 | D::adjust(pixels.loadAndScale0()));
writeByte(D::adjust(pixels.loadAndScale1()));
writeByte(D::adjust(pixels.loadAndScale2()));
pixels.advanceData();
pixels.stepDithering();
}
D::postBlock(len);
waitFully();
// restore ctar1
update_ctar1(ctar1_save);
}
release();
}
};
#endif
FASTLED_NAMESPACE_END
#endif
@@ -0,0 +1,46 @@
#ifndef __INC_LED_SYSDEFS_ARM_K20_H
#define __INC_LED_SYSDEFS_ARM_K20_H
#define FASTLED_TEENSY3
#define FASTLED_ARM
#ifndef INTERRUPT_THRESHOLD
#define INTERRUPT_THRESHOLD 1
#endif
// Default to allowing interrupts
#ifndef FASTLED_ALLOW_INTERRUPTS
#define FASTLED_ALLOW_INTERRUPTS 1
#endif
#if FASTLED_ALLOW_INTERRUPTS == 1
#define FASTLED_ACCURATE_CLOCK
#endif
#if (F_CPU == 96000000)
#define CLK_DBL 1
#endif
// Get some system include files
#include <avr/io.h>
#include <avr/interrupt.h> // for cli/se definitions
// Define the register types
#if defined(ARDUINO) // && ARDUINO < 150
typedef volatile uint8_t RoReg; /**< Read only 8-bit register (volatile const unsigned int) */
typedef volatile uint8_t RwReg; /**< Read-Write 8-bit register (volatile unsigned int) */
#endif
extern volatile uint32_t systick_millis_count;
# define MS_COUNTER systick_millis_count
// Default to using PROGMEM, since TEENSY3 provides it
// even though all it does is ignore it. Just being
// conservative here in case TEENSY3 changes.
#ifndef FASTLED_USE_PROGMEM
#define FASTLED_USE_PROGMEM 1
#endif
#endif
@@ -0,0 +1,96 @@
#ifndef __INC_OCTOWS2811_CONTROLLER_H
#define __INC_OCTOWS2811_CONTROLLER_H
#ifdef USE_OCTOWS2811
// #include "OctoWS2811.h"
FASTLED_NAMESPACE_BEGIN
template<EOrder RGB_ORDER = GRB, boolean SLOW=false>
class COctoWS2811Controller : public CLEDController {
OctoWS2811 *pocto;
uint8_t *drawbuffer,*framebuffer;
void _init(int nLeds) {
if(pocto == NULL) {
drawbuffer = (uint8_t*)malloc(nLeds * 8 * 3);
framebuffer = (uint8_t*)malloc(nLeds * 8 * 3);
// byte ordering is handled in show by the pixel controller
int config = WS2811_RGB;
if(SLOW) {
config |= WS2811_400kHz;
}
pocto = new OctoWS2811(nLeds, framebuffer, drawbuffer, config);
pocto->begin();
}
}
public:
COctoWS2811Controller() { pocto = NULL; }
virtual void init() { /* do nothing yet */ }
virtual void clearLeds(int nLeds) {
_init(nLeds);
showColor(CRGB(0,0,0),nLeds,CRGB(0,0,0));
}
virtual void showColor(const struct CRGB & data, int nLeds, CRGB scale) {
_init(nLeds);
// Get our pixel values
PixelController<RGB_ORDER> pixels(data, nLeds, scale, getDither());
uint8_t ball[3][8];
memset(ball[0],pixels.loadAndScale0(),8);
memset(ball[1],pixels.loadAndScale1(),8);
memset(ball[2],pixels.loadAndScale2(),8);
uint8_t bout[24];
transpose8x1_MSB(ball[0],bout);
transpose8x1_MSB(ball[1],bout+8);
transpose8x1_MSB(ball[2],bout+16);
uint8_t *pdata = drawbuffer;
while(nLeds--) {
memcpy(pdata,bout,24);
pdata += 24;
}
pocto->show();
}
typedef union {
uint8_t bytes[8];
uint32_t raw[2];
} Lines;
virtual void show(const struct CRGB *rgbdata, int nLeds, CRGB scale) {
_init(nLeds);
MultiPixelController<8,0xFF,RGB_ORDER> pixels(rgbdata,nLeds, scale, getDither() );
uint8_t *pData = drawbuffer;
while(nLeds--) {
Lines b;
for(int i = 0; i < 8; i++) { b.bytes[i] = pixels.loadAndScale0(i); }
transpose8x1_MSB(b.bytes,pData); pData += 8;
for(int i = 0; i < 8; i++) { b.bytes[i] = pixels.loadAndScale1(i); }
transpose8x1_MSB(b.bytes,pData); pData += 8;
for(int i = 0; i < 8; i++) { b.bytes[i] = pixels.loadAndScale2(i); }
transpose8x1_MSB(b.bytes,pData); pData += 8;
pixels.stepDithering();
pixels.advanceData();
}
pocto->show();
}
};
FASTLED_NAMESPACE_END
#endif
#endif
@@ -0,0 +1,83 @@
#ifndef __INC_SMARTMATRIX_T3_H
#define __INC_SMARTMATRIX_T3_H
#ifdef SmartMatrix_h
#include<SmartMatrix.h>
FASTLED_NAMESPACE_BEGIN
extern SmartMatrix *pSmartMatrix;
// note - dmx simple must be included before FastSPI for this code to be enabled
class CSmartMatrixController : public CLEDController {
SmartMatrix matrix;
public:
// initialize the LED controller
virtual void init() {
// Initialize 32x32 LED Matrix
matrix.begin();
matrix.setBrightness(255);
matrix.setColorCorrection(ccNone);
// Clear screen
clearLeds(0);
matrix.swapBuffers();
pSmartMatrix = &matrix;
}
// clear out/zero out the given number of leds.
virtual void clearLeds(int nLeds) {
const rgb24 black = {0,0,0};
matrix.fillScreen(black);
matrix.swapBuffers();
}
// set all the leds on the controller to a given color
virtual void showColor(const struct CRGB & data, int nLeds,CRGB scale) {
PixelController<RGB> pixels(data, nLeds, scale, getDither());
rgb24 *md = matrix.backBuffer();
while(nLeds--) {
md->red = pixels.loadAndScale0();
md->green = pixels.loadAndScale1();
md->blue = pixels.loadAndScale2();
md++;
pixels.stepDithering();
}
matrix.swapBuffers();
}
// note that the uint8_ts will be in the order that you want them sent out to the device.
// nLeds is the number of RGB leds being written to
virtual void show(const struct CRGB *data, int nLeds, CRGB scale) {
PixelController<RGB> pixels(data, nLeds, scale, getDither());
#ifdef SMART_MATRIX_CAN_TRIPLE_BUFFER
rgb24 *md = matrix.getRealBackBuffer();
#else
rgb24 *md = matrix.backBuffer();
#endif
while(nLeds--) {
md->red = pixels.loadAndScale0();
md->green = pixels.loadAndScale1();
md->blue = pixels.loadAndScale2();
md++;
pixels.advanceData();
pixels.stepDithering();
}
matrix.swapBuffers();
#ifdef SMART_MATRIX_CAN_TRIPLE_BUFFER
matrix.setBackBuffer((rgb24*)data);
#endif
}
#ifdef SUPPORT_ARGB
// as above, but every 4th uint8_t is assumed to be alpha channel data, and will be skipped
virtual void show(const struct CARGB *data, int nLeds, CRGB scale) = 0;
#endif
};
FASTLED_NAMESPACE_END
#endif
#endif
@@ -0,0 +1,89 @@
#ifndef __INC_CLOCKLESS_ARM_KL26
#define __INC_CLOCKLESS_ARM_KL26
#include "platforms/arm/common/m0clockless.h"
FASTLED_NAMESPACE_BEGIN
#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>
class ClocklessController : public CLEDController {
typedef typename FastPinBB<DATA_PIN>::port_ptr_t data_ptr_t;
typedef typename FastPinBB<DATA_PIN>::port_t data_t;
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 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> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
cli();
showRGBInternal(pixels);
sei();
mWait.mark();
}
virtual void show(const struct CRGB *rgbdata, int nLeds, CRGB scale) {
PixelController<RGB_ORDER> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
cli();
showRGBInternal(pixels);
sei();
mWait.mark();
}
#ifdef SUPPORT_ARGB
virtual void show(const struct CARGB *rgbdata, int nLeds, CRGB scale) {
PixelController<RGB_ORDER> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
cli();
showRGBInternal(pixels);
sei();
mWait.mark();
}
#endif
// 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();
showLedData<4,8,T1,T2,T3,RGB_ORDER, WAIT_TIME>(portBase, FastPin<DATA_PIN>::mask(), pixels.mData, pixels.mLen, &data);
return 0; // 0x00FFFFFF - _VAL;
}
};
FASTLED_NAMESPACE_END
#endif // __INC_CLOCKLESS_ARM_D21
@@ -0,0 +1,10 @@
#ifndef __INC_FASTLED_ARM_KL26_H
#define __INC_FASTLED_ARM_KL26_H
// Include the k20 headers
#include "fastled_delay.h"
#include "fastpin_arm_kl26.h"
#include "fastspi_arm_kl26.h"
#include "clockless_arm_kl26.h"
#endif
@@ -0,0 +1,88 @@
#ifndef __FASTPIN_ARM_KL26_H
#define __FASTPIN_ARM_KL26_H
FASTLED_NAMESPACE_BEGIN
#if defined(FASTLED_FORCE_SOFTWARE_PINS)
#warning "Software pin support forced, pin access will be sloightly slower."
#define NO_HARDWARE_PIN_SUPPORT
#undef HAS_HARDWARE_PIN_SUPPORT
#else
/// Template definition for teensy LC style ARM pins, providing direct access to the various GPIO registers. Note that this
/// uses the full port GPIO registers. In theory, in some way, bit-band register access -should- be faster, however I have found
/// that something about the way gcc does register allocation results in the bit-band code being slower. It will need more fine tuning.
/// The registers are data output, set output, clear output, toggle output, input, and direction
template<uint8_t PIN, uint32_t _MASK, typename _PDOR, typename _PSOR, typename _PCOR, typename _PTOR, typename _PDIR, typename _PDDR> class _ARMPIN {
public:
typedef volatile uint32_t * port_ptr_t;
typedef uint32_t port_t;
inline static void setOutput() { pinMode(PIN, OUTPUT); } // TODO: perform MUX config { _PDDR::r() |= _MASK; }
inline static void setInput() { pinMode(PIN, INPUT); } // TODO: preform MUX config { _PDDR::r() &= ~_MASK; }
inline static void hi() __attribute__ ((always_inline)) { _PSOR::r() = _MASK; }
inline static void lo() __attribute__ ((always_inline)) { _PCOR::r() = _MASK; }
inline static void set(register port_t val) __attribute__ ((always_inline)) { _PDOR::r() = val; }
inline static void strobe() __attribute__ ((always_inline)) { toggle(); toggle(); }
inline static void toggle() __attribute__ ((always_inline)) { _PTOR::r() = _MASK; }
inline static void hi(register port_ptr_t port) __attribute__ ((always_inline)) { hi(); }
inline static void lo(register port_ptr_t port) __attribute__ ((always_inline)) { lo(); }
inline static void fastset(register port_ptr_t port, register port_t val) __attribute__ ((always_inline)) { *port = val; }
inline static port_t hival() __attribute__ ((always_inline)) { return _PDOR::r() | _MASK; }
inline static port_t loval() __attribute__ ((always_inline)) { return _PDOR::r() & ~_MASK; }
inline static port_ptr_t port() __attribute__ ((always_inline)) { return &_PDOR::r(); }
inline static port_ptr_t sport() __attribute__ ((always_inline)) { return &_PSOR::r(); }
inline static port_ptr_t cport() __attribute__ ((always_inline)) { return &_PCOR::r(); }
inline static port_t mask() __attribute__ ((always_inline)) { return _MASK; }
};
// Macros for kl26 pin access/definition
#define GPIO_BITBAND_ADDR(reg, bit) (((uint32_t)&(reg) - 0x40000000) * 32 + (bit) * 4 + 0x42000000)
#define GPIO_BITBAND_PTR(reg, bit) ((uint32_t *)GPIO_BITBAND_ADDR((reg), (bit)))
#define _R(T) struct __gen_struct_ ## T
#define _RD32(T) struct __gen_struct_ ## T { static __attribute__((always_inline)) inline reg32_t r() { return T; } \
template<int BIT> static __attribute__((always_inline)) inline ptr_reg32_t rx() { return GPIO_BITBAND_PTR(T, BIT); } };
#define _IO32(L) _RD32(FGPIO ## L ## _PDOR); _RD32(FGPIO ## L ## _PSOR); _RD32(FGPIO ## L ## _PCOR); _RD32(GPIO ## L ## _PTOR); _RD32(FGPIO ## L ## _PDIR); _RD32(FGPIO ## L ## _PDDR);
#define _DEFPIN_ARM(PIN, BIT, L) template<> class FastPin<PIN> : public _ARMPIN<PIN, 1 << BIT, _R(FGPIO ## L ## _PDOR), _R(FGPIO ## L ## _PSOR), _R(FGPIO ## L ## _PCOR), \
_R(GPIO ## L ## _PTOR), _R(FGPIO ## L ## _PDIR), _R(FGPIO ## L ## _PDDR)> {}; \
/* template<> class FastPinBB<PIN> : public _ARMPIN_BITBAND<PIN, BIT, _R(GPIO ## L ## _PDOR), _R(GPIO ## L ## _PSOR), _R(GPIO ## L ## _PCOR), \
_R(GPIO ## L ## _PTOR), _R(GPIO ## L ## _PDIR), _R(GPIO ## L ## _PDDR)> {}; */
// Actual pin definitions
#if defined(FASTLED_TEENSYLC) && defined(CORE_TEENSY)
_IO32(A); _IO32(B); _IO32(C); _IO32(D); _IO32(E);
#define MAX_PIN 26
_DEFPIN_ARM(0, 16, B); _DEFPIN_ARM(1, 17, B); _DEFPIN_ARM(2, 0, D); _DEFPIN_ARM(3, 1, A);
_DEFPIN_ARM(4, 2, A); _DEFPIN_ARM(5, 7, D); _DEFPIN_ARM(6, 4, D); _DEFPIN_ARM(7, 2, D);
_DEFPIN_ARM(8, 3, D); _DEFPIN_ARM(9, 3, C); _DEFPIN_ARM(10, 4, C); _DEFPIN_ARM(11, 6, C);
_DEFPIN_ARM(12, 7, C); _DEFPIN_ARM(13, 5, C); _DEFPIN_ARM(14, 1, D); _DEFPIN_ARM(15, 0, C);
_DEFPIN_ARM(16, 0, B); _DEFPIN_ARM(17, 1, B); _DEFPIN_ARM(18, 3, B); _DEFPIN_ARM(19, 2, B);
_DEFPIN_ARM(20, 5, D); _DEFPIN_ARM(21, 6, D); _DEFPIN_ARM(22, 1, C); _DEFPIN_ARM(23, 2, C);
_DEFPIN_ARM(24, 20, E); _DEFPIN_ARM(25, 21, E); _DEFPIN_ARM(26, 30, E);
#define SPI_DATA 11
#define SPI_CLOCK 13
// #define SPI1 (*(SPI_t *)0x4002D000)
#define SPI2_DATA 0
#define SPI2_CLOCK 20
#define HAS_HARDWARE_PIN_SUPPORT
#endif
#endif // FASTLED_FORCE_SOFTWARE_PINS
FASTLED_NAMESPACE_END
#endif // __INC_FASTPIN_ARM_K20
@@ -0,0 +1,252 @@
#ifndef __INC_FASTSPI_ARM_KL26_H
#define __INC_FASTSPI_ARM_KL26_h
FASTLED_NAMESPACE_BEGIN
template <int VAL> void getScalars(uint8_t & sppr, uint8_t & spr) {
if(VAL > 4096) { sppr=7; spr=8; }
else if(VAL > 3584) { sppr=6; spr=8; }
else if(VAL > 3072) { sppr=5; spr=8; }
else if(VAL > 2560) { sppr=4; spr=8; }
else if(VAL > 2048) { sppr=7; spr=7; }
else if(VAL > 2048) { sppr=3; spr=8; }
else if(VAL > 1792) { sppr=6; spr=7; }
else if(VAL > 1536) { sppr=5; spr=7; }
else if(VAL > 1536) { sppr=2; spr=8; }
else if(VAL > 1280) { sppr=4; spr=7; }
else if(VAL > 1024) { sppr=7; spr=6; }
else if(VAL > 1024) { sppr=3; spr=7; }
else if(VAL > 1024) { sppr=1; spr=8; }
else if(VAL > 896) { sppr=6; spr=6; }
else if(VAL > 768) { sppr=5; spr=6; }
else if(VAL > 768) { sppr=2; spr=7; }
else if(VAL > 640) { sppr=4; spr=6; }
else if(VAL > 512) { sppr=7; spr=5; }
else if(VAL > 512) { sppr=3; spr=6; }
else if(VAL > 512) { sppr=1; spr=7; }
else if(VAL > 512) { sppr=0; spr=8; }
else if(VAL > 448) { sppr=6; spr=5; }
else if(VAL > 384) { sppr=5; spr=5; }
else if(VAL > 384) { sppr=2; spr=6; }
else if(VAL > 320) { sppr=4; spr=5; }
else if(VAL > 256) { sppr=7; spr=4; }
else if(VAL > 256) { sppr=3; spr=5; }
else if(VAL > 256) { sppr=1; spr=6; }
else if(VAL > 256) { sppr=0; spr=7; }
else if(VAL > 224) { sppr=6; spr=4; }
else if(VAL > 192) { sppr=5; spr=4; }
else if(VAL > 192) { sppr=2; spr=5; }
else if(VAL > 160) { sppr=4; spr=4; }
else if(VAL > 128) { sppr=7; spr=3; }
else if(VAL > 128) { sppr=3; spr=4; }
else if(VAL > 128) { sppr=1; spr=5; }
else if(VAL > 128) { sppr=0; spr=6; }
else if(VAL > 112) { sppr=6; spr=3; }
else if(VAL > 96) { sppr=5; spr=3; }
else if(VAL > 96) { sppr=2; spr=4; }
else if(VAL > 80) { sppr=4; spr=3; }
else if(VAL > 64) { sppr=7; spr=2; }
else if(VAL > 64) { sppr=3; spr=3; }
else if(VAL > 64) { sppr=1; spr=4; }
else if(VAL > 64) { sppr=0; spr=5; }
else if(VAL > 56) { sppr=6; spr=2; }
else if(VAL > 48) { sppr=5; spr=2; }
else if(VAL > 48) { sppr=2; spr=3; }
else if(VAL > 40) { sppr=4; spr=2; }
else if(VAL > 32) { sppr=7; spr=1; }
else if(VAL > 32) { sppr=3; spr=2; }
else if(VAL > 32) { sppr=1; spr=3; }
else if(VAL > 32) { sppr=0; spr=4; }
else if(VAL > 28) { sppr=6; spr=1; }
else if(VAL > 24) { sppr=5; spr=1; }
else if(VAL > 24) { sppr=2; spr=2; }
else if(VAL > 20) { sppr=4; spr=1; }
else if(VAL > 16) { sppr=7; spr=0; }
else if(VAL > 16) { sppr=3; spr=1; }
else if(VAL > 16) { sppr=1; spr=2; }
else if(VAL > 16) { sppr=0; spr=3; }
else if(VAL > 14) { sppr=6; spr=0; }
else if(VAL > 12) { sppr=5; spr=0; }
else if(VAL > 12) { sppr=2; spr=1; }
else if(VAL > 10) { sppr=4; spr=0; }
else if(VAL > 8) { sppr=3; spr=0; }
else if(VAL > 8) { sppr=1; spr=1; }
else if(VAL > 8) { sppr=0; spr=2; }
else if(VAL > 6) { sppr=2; spr=0; }
else if(VAL > 4) { sppr=1; spr=0; }
else if(VAL > 4) { sppr=0; spr=1; }
else /* if(VAL > 2) */ { sppr=0; spr=0; }
}
#define SPIX (*(KINETISL_SPI_t*)pSPIX)
#define ARM_HARDWARE_SPI
template <uint8_t _DATA_PIN, uint8_t _CLOCK_PIN, uint8_t _SPI_CLOCK_DIVIDER, uint32_t pSPIX>
class ARMHardwareSPIOutput {
Selectable *m_pSelect;
static inline void enable_pins(void) __attribute__((always_inline)) {
switch(_DATA_PIN) {
case 0: CORE_PIN0_CONFIG = PORT_PCR_MUX(2); break;
case 1: CORE_PIN1_CONFIG = PORT_PCR_MUX(5); break;
case 7: CORE_PIN7_CONFIG = PORT_PCR_MUX(2); break;
case 8: CORE_PIN8_CONFIG = PORT_PCR_MUX(5); break;
case 11: CORE_PIN11_CONFIG = PORT_PCR_MUX(2); break;
case 12: CORE_PIN12_CONFIG = PORT_PCR_MUX(5); break;
case 21: CORE_PIN21_CONFIG = PORT_PCR_MUX(2); break;
}
switch(_CLOCK_PIN) {
case 13: CORE_PIN13_CONFIG = PORT_PCR_MUX(2); break;
case 14: CORE_PIN14_CONFIG = PORT_PCR_MUX(2); break;
case 20: CORE_PIN20_CONFIG = PORT_PCR_MUX(2); break;
}
}
static inline void disable_pins(void) __attribute((always_inline)) {
switch(_DATA_PIN) {
case 0: CORE_PIN0_CONFIG = PORT_PCR_SRE | PORT_PCR_MUX(1); break;
case 1: CORE_PIN1_CONFIG = PORT_PCR_SRE | PORT_PCR_MUX(1); break;
case 7: CORE_PIN7_CONFIG = PORT_PCR_SRE | PORT_PCR_MUX(1); break;
case 8: CORE_PIN8_CONFIG = PORT_PCR_SRE | PORT_PCR_MUX(1); break;
case 11: CORE_PIN11_CONFIG = PORT_PCR_SRE | PORT_PCR_MUX(1); break;
case 12: CORE_PIN12_CONFIG = PORT_PCR_SRE | PORT_PCR_MUX(1); break;
case 21: CORE_PIN21_CONFIG = PORT_PCR_SRE | PORT_PCR_MUX(1); break;
}
switch(_CLOCK_PIN) {
case 13: CORE_PIN13_CONFIG = PORT_PCR_SRE | PORT_PCR_MUX(1); break;
case 14: CORE_PIN14_CONFIG = PORT_PCR_SRE | PORT_PCR_MUX(1); break;
case 20: CORE_PIN20_CONFIG = PORT_PCR_SRE | PORT_PCR_MUX(1); break;
}
}
void setSPIRate() {
uint8_t sppr, spr;
getScalars<_SPI_CLOCK_DIVIDER>(sppr, spr);
// Set the speed
SPIX.BR = SPI_BR_SPPR(sppr) | SPI_BR_SPR(spr);
// Also, force 8 bit transfers (don't want to juggle 8/16 since that flushes the world)
SPIX.C2 = 0;
SPIX.C1 |= SPI_C1_SPE;
}
public:
ARMHardwareSPIOutput() { m_pSelect = NULL; }
ARMHardwareSPIOutput(Selectable *pSelect) { m_pSelect = pSelect; }
// set the object representing the selectable
void setSelect(Selectable *pSelect) { m_pSelect = pSelect; }
// initialize the SPI subssytem
void init() {
FastPin<_DATA_PIN>::setOutput();
FastPin<_CLOCK_PIN>::setOutput();
// Enable the SPI clocks
uint32_t sim4 = SIM_SCGC4;
if ((pSPIX == 0x40076000) && !(sim4 & SIM_SCGC4_SPI0)) {
SIM_SCGC4 = sim4 | SIM_SCGC4_SPI0;
}
if ( (pSPIX == 0x40077000) && !(sim4 & SIM_SCGC4_SPI1)) {
SIM_SCGC4 = sim4 | SIM_SCGC4_SPI1;
}
SPIX.C1 = SPI_C1_MSTR | SPI_C1_SPE;
SPIX.C2 = 0;
SPIX.BR = SPI_BR_SPPR(1) | SPI_BR_SPR(0);
}
// latch the CS select
void inline select() __attribute__((always_inline)) {
if(m_pSelect != NULL) { m_pSelect->select(); }
setSPIRate();
enable_pins();
}
// release the CS select
void inline release() __attribute__((always_inline)) {
disable_pins();
if(m_pSelect != NULL) { m_pSelect->release(); }
}
// Wait for the world to be clear
static void wait() __attribute__((always_inline)) { while(!(SPIX.S & SPI_S_SPTEF)); }
// wait until all queued up data has been written
void waitFully() { wait(); }
// not the most efficient mechanism in the world - but should be enough for sm16716 and friends
template <uint8_t BIT> inline static void writeBit(uint8_t b) { /* TODO */ }
// write a byte out via SPI (returns immediately on writing register)
static void writeByte(uint8_t b) __attribute__((always_inline)) { wait(); SPIX.DL = b; }
// write a word out via SPI (returns immediately on writing register)
static void writeWord(uint16_t w) __attribute__((always_inline)) { writeByte(w>>8); writeByte(w & 0xFF); }
// A raw set of writing byte values, assumes setup/init/waiting done elsewhere (static for use by adjustment classes)
static void writeBytesValueRaw(uint8_t value, int len) {
while(len--) { writeByte(value); }
}
// A full cycle of writing a value for len bytes, including select, release, and waiting
void writeBytesValue(uint8_t value, int len) {
setSPIRate();
select();
while(len--) {
writeByte(value);
}
waitFully();
release();
}
// A full cycle of writing a raw block of data out, including select, release, and waiting
template <class D> void writeBytes(register uint8_t *data, int len) {
setSPIRate();
uint8_t *end = data + len;
select();
// could be optimized to write 16bit words out instead of 8bit bytes
while(data != end) {
writeByte(D::adjust(*data++));
}
D::postBlock(len);
waitFully();
release();
}
void writeBytes(register uint8_t *data, int len) { writeBytes<DATA_NOP>(data, len); }
template <uint8_t FLAGS, class D, EOrder RGB_ORDER> void writePixels(PixelController<RGB_ORDER> pixels) {
int len = pixels.mLen;
select();
while(pixels.has(1)) {
if(FLAGS & FLAG_START_BIT) {
writeBit<0>(1);
writeByte(D::adjust(pixels.loadAndScale0()));
writeByte(D::adjust(pixels.loadAndScale1()));
writeByte(D::adjust(pixels.loadAndScale2()));
} else {
writeByte(D::adjust(pixels.loadAndScale0()));
writeByte(D::adjust(pixels.loadAndScale1()));
writeByte(D::adjust(pixels.loadAndScale2()));
}
pixels.advanceData();
pixels.stepDithering();
}
D::postBlock(len);
release();
}
};
FASTLED_NAMESPACE_END
#endif
@@ -0,0 +1,45 @@
#ifndef __INC_LED_SYSDEFS_ARM_KL26_H
#define __INC_LED_SYSDEFS_ARM_KL26_H
#define FASTLED_TEENSYLC
#define FASTLED_ARM
#define FASTLED_ARM_M0_PLUS
#ifndef INTERRUPT_THRESHOLD
#define INTERRUPT_THRESHOLD 1
#endif
// Default to allowing interrupts
#ifndef FASTLED_ALLOW_INTERRUPTS
#define FASTLED_ALLOW_INTERRUPTS 1
#endif
#if FASTLED_ALLOW_INTERRUPTS == 1
#define FASTLED_ACCURATE_CLOCK
#endif
#if (F_CPU == 96000000)
#define CLK_DBL 1
#endif
// Get some system include files
#include <avr/io.h>
#include <avr/interrupt.h> // for cli/se definitions
// Define the register types
#if defined(ARDUINO) // && ARDUINO < 150
typedef volatile uint8_t RoReg; /**< Read only 8-bit register (volatile const unsigned int) */
typedef volatile uint8_t RwReg; /**< Read-Write 8-bit register (volatile unsigned int) */
#endif
extern volatile uint32_t systick_millis_count;
# define MS_COUNTER systick_millis_count
// Default to using PROGMEM since TEENSYLC provides it
// even though all it does is ignore it. Just being
// conservative here in case TEENSYLC changes.
#ifndef FASTLED_USE_PROGMEM
#define FASTLED_USE_PROGMEM 1
#endif
#endif
@@ -0,0 +1,117 @@
#ifndef __INC_CLOCKLESS_ARM_NRF51
#define __INC_CLOCKLESS_ARM_NRF51
#if defined(NRF51)
#include "nrf51_bitfields.h"
#define FASTLED_HAS_CLOCKLESS 1
#if (FASTLED_ALLOW_INTERRUPTS==1)
#define SEI_CHK LED_TIMER->CC[0] = (WAIT_TIME * (F_CPU/1000000)); LED_TIMER->TASKS_CLEAR; LED_TIMER->EVENTS_COMPARE[0] = 0;
#define CLI_CHK cli(); if(LED_TIMER->EVENTS_COMPARE[0]) { LED_TIMER->TASKS_STOP = 1; return 0; }
#define INNER_SEI sei();
#else
#define SEI_CHK
#define CLI_CHK
#define INNER_SEI delaycycles<1>();
#endif
#include "platforms/arm/common/m0clockless.h"
template <uint8_t DATA_PIN, int T1, int T2, int T3, EOrder RGB_ORDER = RGB, int XTRA0 = 0, bool FLIP = false, int WAIT_TIME = 75>
class ClocklessController : public CLEDController {
typedef typename FastPinBB<DATA_PIN>::port_ptr_t data_ptr_t;
typedef typename FastPinBB<DATA_PIN>::port_t data_t;
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 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> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
cli();
// attempt to re-show a frame if we exit early because of interrupts.
if(!showRGBInternal(pixels)) {
sei(); delayMicroseconds(WAIT_TIME); cli();
showRGBInternal(pixels);
}
sei();
mWait.mark();
}
virtual void show(const struct CRGB *rgbdata, int nLeds, CRGB scale) {
PixelController<RGB_ORDER> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
cli();
if(!showRGBInternal(pixels)) {
sei(); delayMicroseconds(WAIT_TIME); cli();
showRGBInternal(pixels);
}
sei();
mWait.mark();
}
#ifdef SUPPORT_ARGB
virtual void show(const struct CARGB *rgbdata, int nLeds, CRGB scale) {
PixelController<RGB_ORDER> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
cli();
showRGBInternal(pixels);
sei();
mWait.mark();
}
#endif
// 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();
// timer mode w/prescaler of 0
LED_TIMER->MODE = TIMER_MODE_MODE_Timer;
LED_TIMER->PRESCALER = 0;
LED_TIMER->EVENTS_COMPARE[0] = 0;
LED_TIMER->BITMODE = TIMER_BITMODE_BITMODE_16Bit;
LED_TIMER->SHORTS = TIMER_SHORTS_COMPARE0_CLEAR_Msk;
LED_TIMER->TASKS_START = 1;
int ret = showLedData<4,8,T1,T2,T3,RGB_ORDER,WAIT_TIME>(portBase, FastPin<DATA_PIN>::mask(), pixels.mData, pixels.mLen, &data);
LED_TIMER->TASKS_STOP = 1;
return ret; // 0x00FFFFFF - _VAL;
}
};
#endif // NRF51
#endif // __INC_CLOCKLESS_ARM_NRF51
@@ -0,0 +1,11 @@
#ifndef __INC_FASTLED_ARM_NRF51_H
#define __INC_FASTLED_ARM_NRF51_H
// Include the k20 headers
#include "bitswap.h"
#include "fastled_delay.h"
#include "fastpin_arm_nrf51.h"
#include "fastspi_arm_nrf51.h"
#include "clockless_arm_nrf51.h"
#endif
@@ -0,0 +1,119 @@
#ifndef __FASTPIN_ARM_NRF51_H
#define __FASTPIN_ARM_NRF51_H
#if defined(NRF51)
/// Template definition for teensy 3.0 style ARM pins, providing direct access to the various GPIO registers. Note that this
/// uses the full port GPIO registers. In theory, in some way, bit-band register access -should- be faster, however I have found
/// that something about the way gcc does register allocation results in the bit-band code being slower. It will need more fine tuning.
/// The registers are data output, set output, clear output, toggle output, input, and direction
#if 0
template<uint8_t PIN, uint32_t _MASK, typename _DIRSET, typename _DIRCLR, typename _OUTSET, typename _OUTCLR, typename _OUT> class _ARMPIN {
public:
typedef volatile uint32_t * port_ptr_t;
typedef uint32_t port_t;
inline static void setOutput() { _DIRSET::r() = _MASK; }
inline static void setInput() { _DIRCLR::r() = _MASK; }
inline static void hi() __attribute__ ((always_inline)) { _OUTSET::r() = _MASK; }
inline static void lo() __attribute__ ((always_inline)) { _OUTCLR::r() = _MASK; }
inline static void set(register port_t val) __attribute__ ((always_inline)) { _OUT::r() = val; }
inline static void strobe() __attribute__ ((always_inline)) { toggle(); toggle(); }
inline static void toggle() __attribute__ ((always_inline)) { _OUT::r() ^= _MASK; }
inline static void hi(register port_ptr_t port) __attribute__ ((always_inline)) { hi(); }
inline static void lo(register port_ptr_t port) __attribute__ ((always_inline)) { lo(); }
inline static void fastset(register port_ptr_t port, register port_t val) __attribute__ ((always_inline)) { *port = val; }
inline static port_t hival() __attribute__ ((always_inline)) { return _OUT::r() | _MASK; }
inline static port_t loval() __attribute__ ((always_inline)) { return _OUT::r() & ~_MASK; }
inline static port_ptr_t port() __attribute__ ((always_inline)) { return &_OUT::r(); }
inline static port_t mask() __attribute__ ((always_inline)) { return _MASK; }
};
#define ADDR(X) *(volatile uint32_t*)X
#define NR_GPIO_ADDR(base,offset) (*(volatile uint32_t *))((uint32_t)(base + offset))
#define NR_DIRSET ADDR(0x50000518UL) // NR_GPIO_ADDR(NRF_GPIO_BASE, 0x518)
#define NR_DIRCLR ADDR(0x5000051CUL) // NR_GPIO_ADDR(NRF_GPIO_BASE, 0x51C)
#define NR_OUTSET ADDR(0x50000508UL) // NR_GPIO_ADDR(NRF_GPIO_BASE, 0x508)
#define NR_OUTCLR ADDR(0x5000050CUL) // NR_GPIO_ADDR(NRF_GPIO_BASE, 0x50C)
#define NR_OUT ADDR(0x50000504UL) // NR_GPIO_ADDR(NRF_GPIO_BASE, 0x504)
#define _RD32_NRF(T) struct __gen_struct_ ## T { static __attribute__((always_inline)) inline reg32_t r() { return T; }};
_RD32_NRF(NR_DIRSET);
_RD32_NRF(NR_DIRCLR);
_RD32_NRF(NR_OUTSET);
_RD32_NRF(NR_OUTCLR);
_RD32_NRF(NR_OUT);
#define _DEFPIN_ARM(PIN) template<> class FastPin<PIN> : public _ARMPIN<PIN, 1 << PIN, \
_R(NR_DIRSET), _R(NR_DIRCLR), _R(NR_OUTSET), _R(NR_OUTCLR), _R(NR_OUT)> {};
#else
typedef struct { /*!< GPIO Structure */
// __I uint32_t RESERVED0[321];
__IO uint32_t OUT; /*!< Write GPIO port. */
__IO uint32_t OUTSET; /*!< Set individual bits in GPIO port. */
__IO uint32_t OUTCLR; /*!< Clear individual bits in GPIO port. */
__I uint32_t IN; /*!< Read GPIO port. */
__IO uint32_t DIR; /*!< Direction of GPIO pins. */
__IO uint32_t DIRSET; /*!< DIR set register. */
__IO uint32_t DIRCLR; /*!< DIR clear register. */
__I uint32_t RESERVED1[120];
__IO uint32_t PIN_CNF[32]; /*!< Configuration of GPIO pins. */
} FL_NRF_GPIO_Type;
#define FL_NRF_GPIO_BASE 0x50000504UL
#define FL_NRF_GPIO ((FL_NRF_GPIO_Type *) FL_NRF_GPIO_BASE)
template<uint8_t PIN, uint32_t _MASK> class _ARMPIN {
public:
typedef volatile uint32_t * port_ptr_t;
typedef uint32_t port_t;
inline static void setOutput() { FL_NRF_GPIO->DIRSET = _MASK; }
inline static void setInput() { FL_NRF_GPIO->DIRCLR = _MASK; }
inline static void hi() __attribute__ ((always_inline)) { FL_NRF_GPIO->OUTSET = _MASK; }
inline static void lo() __attribute__ ((always_inline)) { FL_NRF_GPIO->OUTCLR= _MASK; }
inline static void set(register port_t val) __attribute__ ((always_inline)) { FL_NRF_GPIO->OUT = val; }
inline static void strobe() __attribute__ ((always_inline)) { toggle(); toggle(); }
inline static void toggle() __attribute__ ((always_inline)) { FL_NRF_GPIO->OUT ^= _MASK; }
inline static void hi(register port_ptr_t port) __attribute__ ((always_inline)) { hi(); }
inline static void lo(register port_ptr_t port) __attribute__ ((always_inline)) { lo(); }
inline static void fastset(register port_ptr_t port, register port_t val) __attribute__ ((always_inline)) { *port = val; }
inline static port_t hival() __attribute__ ((always_inline)) { return FL_NRF_GPIO->OUT | _MASK; }
inline static port_t loval() __attribute__ ((always_inline)) { return FL_NRF_GPIO->OUT & ~_MASK; }
inline static port_ptr_t port() __attribute__ ((always_inline)) { return &FL_NRF_GPIO->OUT; }
inline static port_t mask() __attribute__ ((always_inline)) { return _MASK; }
inline static bool isset() __attribute__ ((always_inline)) { return (FL_NRF_GPIO->IN & _MASK) != 0; }
};
#define _DEFPIN_ARM(PIN) template<> class FastPin<PIN> : public _ARMPIN<PIN, 1 << PIN> {};
#endif
// Actual pin definitions
#define MAX_PIN 31
_DEFPIN_ARM(0); _DEFPIN_ARM(1); _DEFPIN_ARM(2); _DEFPIN_ARM(3);
_DEFPIN_ARM(4); _DEFPIN_ARM(5); _DEFPIN_ARM(6); _DEFPIN_ARM(7);
_DEFPIN_ARM(8); _DEFPIN_ARM(9); _DEFPIN_ARM(10); _DEFPIN_ARM(11);
_DEFPIN_ARM(12); _DEFPIN_ARM(13); _DEFPIN_ARM(14); _DEFPIN_ARM(15);
_DEFPIN_ARM(16); _DEFPIN_ARM(17); _DEFPIN_ARM(18); _DEFPIN_ARM(19);
_DEFPIN_ARM(20); _DEFPIN_ARM(21); _DEFPIN_ARM(22); _DEFPIN_ARM(23);
_DEFPIN_ARM(24); _DEFPIN_ARM(25); _DEFPIN_ARM(26); _DEFPIN_ARM(27);
_DEFPIN_ARM(28); _DEFPIN_ARM(29); _DEFPIN_ARM(30); _DEFPIN_ARM(31);
#define HAS_HARDWARE_PIN_SUPPORT
#endif
#endif
@@ -0,0 +1,146 @@
#ifndef __INC_FASTSPI_NRF_H
#define __INC_FASTSPI_NRF_H
#ifdef NRF51
// A nop/stub class, mostly to show the SPI methods that are needed/used by the various SPI chipset implementations. Should
// be used as a definition for the set of methods that the spi implementation classes should use (since C++ doesn't support the
// idea of interfaces - it's possible this could be done with virtual classes, need to decide if i want that overhead)
template <uint8_t _DATA_PIN, uint8_t _CLOCK_PIN, uint8_t _SPI_CLOCK_DIVIDER>
class NRF51SPIOutput {
struct saveData {
uint32_t sck;
uint32_t mosi;
uint32_t miso;
uint32_t freq;
uint32_t enable;
} mSavedData;
void saveSPIData() {
mSavedData.sck = NRF_SPI0->PSELSCK;
mSavedData.mosi = NRF_SPI0->PSELMOSI;
mSavedData.miso = NRF_SPI0->PSELMISO;
mSavedData.freq = NRF_SPI0->FREQUENCY;
mSavedData.enable = NRF_SPI0->ENABLE;
}
void restoreSPIData() {
NRF_SPI0->PSELSCK = mSavedData.sck;
NRF_SPI0->PSELMOSI = mSavedData.mosi;
NRF_SPI0->PSELMISO = mSavedData.miso;
NRF_SPI0->FREQUENCY = mSavedData.freq;
mSavedData.enable = NRF_SPI0->ENABLE;
}
public:
NRF51SPIOutput() { FastPin<_DATA_PIN>::setOutput(); FastPin<_CLOCK_PIN>::setOutput(); }
NRF51SPIOutput(Selectable *pSelect) { FastPin<_DATA_PIN>::setOutput(); FastPin<_CLOCK_PIN>::setOutput(); }
// set the object representing the selectable
void setSelect(Selectable *pSelect) { /* TODO */ }
// initialize the SPI subssytem
void init() {
FastPin<_DATA_PIN>::setOutput();
FastPin<_CLOCK_PIN>::setOutput();
NRF_SPI0->PSELSCK = _CLOCK_PIN;
NRF_SPI0->PSELMOSI = _DATA_PIN;
NRF_SPI0->PSELMISO = 0xFFFFFFFF;
NRF_SPI0->FREQUENCY = 0x80000000;
NRF_SPI0->ENABLE = 1;
NRF_SPI0->EVENTS_READY = 0;
}
// latch the CS select
void select() { saveSPIData(); init(); }
// release the CS select
void release() { restoreSPIData(); }
static bool shouldWait(bool wait = false) __attribute__((always_inline)) __attribute__((always_inline)) {
static bool sWait=false;
return false; // if(sWait) { sWait = wait; return true; } else { sWait = wait; return false; }
}
// wait until all queued up data has been written
void waitFully() __attribute__((always_inline)){ if(shouldWait()) { while(NRF_SPI0->EVENTS_READY==0); } NRF_SPI0->EVENTS_READY=0; uint8_t b = NRF_SPI0->RXD; }
void wait() __attribute__((always_inline)){ if(shouldWait()) { while(NRF_SPI0->EVENTS_READY==0); } NRF_SPI0->EVENTS_READY=0; uint8_t b = NRF_SPI0->RXD; }
// void waitFully() { while(NRF_SPI0->EVENTS_READY==0); NRF_SPI0->EVENTS_READY=0; uint8_t b = NRF_SPI0->RXD; }
// void wait() { while(NRF_SPI0->EVENTS_READY==0); NRF_SPI0->EVENTS_READY=0; uint8_t b = NRF_SPI0->RXD; }
// write a byte out via SPI (returns immediately on writing register)
// void writeByte(uint8_t b) { wait(); NRF_SPI0->TXD = b; shouldWait(true); }
// void writeByte(uint8_t b) __attribute__((always_inline)){ wait(); NRF_SPI0->TXD = b; shouldWait(true); }
void writeByte(uint8_t b) __attribute__((always_inline)) { NRF_SPI0->EVENTS_READY=0; /*uint8_t x = NRF_SPI0->RXD;*/ NRF_SPI0->TXD = b; }
// write a word out via SPI (returns immediately on writing register)
void writeWord(uint16_t w) __attribute__((always_inline)){ writeByte(w>>8); writeByte(w & 0xFF); }
// A raw set of writing byte values, assumes setup/init/waiting done elsewhere (static for use by adjustment classes)
static void writeBytesValueRaw(uint8_t value, int len) { while(len--) { writeByte(value); } }
// A full cycle of writing a value for len bytes, including select, release, and waiting
void writeBytesValue(uint8_t value, int len) {
select();
while(len--) {
writeByte(value);
}
waitFully();
release();
}
// A full cycle of writing a raw block of data out, including select, release, and waiting
template<class D> void writeBytes(uint8_t *data, int len) {
uint8_t *end = data + len;
select();
while(data != end) {
writeByte(D::adjust(*data++));
}
D::postBlock(len);
waitFully();
release();
}
void writeBytes(uint8_t *data, int len) {
writeBytes<DATA_NOP>(data, len);
}
// write a single bit out, which bit from the passed in byte is determined by template parameter
template <uint8_t BIT> inline static void writeBit(uint8_t b) {
waitFully();
NRF_SPI0->ENABLE = 0;
if(b & 1<<BIT) {
FastPin<_DATA_PIN>::hi();
} else {
FastPin<_DATA_PIN>::lo();
}
FastPin<_CLOCK_PIN>::toggle();
FastPin<_CLOCK_PIN>::toggle();
NRF_SPI0->ENABLE = 1;
}
template <uint8_t FLAGS, class D, EOrder RGB_ORDER> void writePixels(PixelController<RGB_ORDER> pixels) {
select();
int len = pixels.mLen;
while(pixels.has(1)) {
if(FLAGS & FLAG_START_BIT) {
writeBit<0>(1);
}
writeByte(D::adjust(pixels.loadAndScale0()));
writeByte(D::adjust(pixels.loadAndScale1()));
writeByte(D::adjust(pixels.loadAndScale2()));
pixels.advanceData();
pixels.stepDithering();
}
D::postBlock(len);
waitFully();
release();
}
};
#endif
#endif
@@ -0,0 +1,44 @@
#ifndef __LED_SYSDEFS_ARM_NRF51
#define __LED_SYSDEFS_ARM_NRF51
#ifndef NRF51
#define NRF51
#endif
#define LED_TIMER NRF_TIMER1
#define FASTLED_NO_PINMAP
#define FASTLED_HAS_CLOCKLESS
#define FASTLED_ARM
#define FASTLED_ARM_M0
#ifndef F_CPU
#define F_CPU 16000000
#endif
#include <stdint.h>
#include "nrf51.h"
#include "core_cm0.h"
typedef volatile uint32_t RoReg;
typedef volatile uint32_t RwReg;
typedef uint32_t prog_uint32_t;
typedef uint8_t boolean;
#define PROGMEM
#define NO_PROGMEM
#define NEED_CXX_BITS
// Default to NOT using PROGMEM here
#ifndef FASTLED_USE_PROGMEM
#define FASTLED_USE_PROGMEM 0
#endif
#ifndef FASTLED_ALLOW_INTERRUPTS
#define FASTLED_ALLOW_INTERRUPTS 1
#endif
#define cli() __disable_irq();
#define sei() __enable_irq();
#endif
@@ -0,0 +1,145 @@
#ifndef __INC_CLOCKLESS_ARM_SAM_H
#define __INC_CLOCKLESS_ARM_SAM_H
FASTLED_NAMESPACE_BEGIN
// Definition for a single channel clockless controller for the sam family of arm chips, like that used in the due and rfduino
// See clockless.h for detailed info on how the template parameters are used.
#if defined(__SAM3X8E__)
#define TADJUST 0
#define TOTAL ( (T1+TADJUST) + (T2+TADJUST) + (T3+TADJUST) )
#define T1_MARK (TOTAL - (T1+TADJUST))
#define T2_MARK (T1_MARK - (T2+TADJUST))
#define SCALE(S,V) scale8_video(S,V)
// #define SCALE(S,V) scale8(S,V)
#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>
class ClocklessController : public CLEDController {
typedef typename FastPinBB<DATA_PIN>::port_ptr_t data_ptr_t;
typedef typename FastPinBB<DATA_PIN>::port_t data_t;
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 uint16_t getMaxRefreshRate() const { return 400; }
virtual void clearLeds(int nLeds) {
showColor(CRGB(0, 0, 0), nLeds, 0);
}
protected:
// 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> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
showRGBInternal(pixels);
mWait.mark();
}
virtual void show(const struct CRGB *rgbdata, int nLeds, CRGB scale) {
PixelController<RGB_ORDER> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
showRGBInternal(pixels);
mWait.mark();
}
#ifdef SUPPORT_ARGB
virtual void show(const struct CARGB *rgbdata, int nLeds, CRGB scale) {
PixelController<RGB_ORDER> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
showRGBInternal(pixels);
sei();
mWait.mark();
}
#endif
template<int BITS> __attribute__ ((always_inline)) inline static void writeBits(register uint32_t & next_mark, register data_ptr_t port, register uint8_t & b) {
// Make sure we don't slot into a wrapping spot, this will delay up to 12.5µs for WS2812
// bool bShift=0;
// while(VAL < (TOTAL*10)) { bShift=true; }
// if(bShift) { next_mark = (VAL-TOTAL); };
for(register uint32_t i = BITS; i > 0; i--) {
// wait to start the bit, then set the pin high
while(DUE_TIMER_VAL < next_mark);
next_mark = (DUE_TIMER_VAL+TOTAL);
*port = 1;
// how long we want to wait next depends on whether or not our bit is set to 1 or 0
if(b&0x80) {
// we're a 1, wait until there's less than T3 clocks left
while((next_mark - DUE_TIMER_VAL) > (T3));
} else {
// we're a 0, wait until there's less than (T2+T3+slop) clocks left in this bit
while((next_mark - DUE_TIMER_VAL) > (T2+T3+6+TADJUST+TADJUST));
}
*port=0;
b <<= 1;
}
}
#define FORCE_REFERENCE(var) asm volatile( "" : : "r" (var) )
// 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) {
// Setup and start the clock
TC_Configure(DUE_TIMER,DUE_TIMER_CHANNEL,TC_CMR_TCCLKS_TIMER_CLOCK1);
pmc_enable_periph_clk(DUE_TIMER_ID);
TC_Start(DUE_TIMER,DUE_TIMER_CHANNEL);
register data_ptr_t port asm("r7") = FastPinBB<DATA_PIN>::port(); FORCE_REFERENCE(port);
*port = 0;
// Setup the pixel controller and load/scale the first byte
pixels.preStepFirstByteDithering();
register uint8_t b = pixels.loadAndScale0();
uint32_t next_mark = (DUE_TIMER_VAL + (TOTAL));
while(pixels.has(1)) {
pixels.stepDithering();
#if (FASTLED_ALLOW_INTERRUPTS == 1)
cli();
if(DUE_TIMER_VAL > next_mark) {
if((DUE_TIMER_VAL - next_mark) > ((WAIT_TIME-INTERRUPT_THRESHOLD)*CLKS_PER_US)) { sei(); TC_Stop(DUE_TIMER,DUE_TIMER_CHANNEL); return DUE_TIMER_VAL; }
}
#endif
writeBits<8+XTRA0>(next_mark, port, b);
b = pixels.loadAndScale1();
writeBits<8+XTRA0>(next_mark, port,b);
b = pixels.loadAndScale2();
writeBits<8+XTRA0>(next_mark, port,b);
b = pixels.advanceAndLoadAndScale0();
#if (FASTLED_ALLOW_INTERRUPTS == 1)
sei();
#endif
};
TC_Stop(DUE_TIMER,DUE_TIMER_CHANNEL);
return DUE_TIMER_VAL;
}
};
#endif
FASTLED_NAMESPACE_END
#endif
@@ -0,0 +1,206 @@
#ifndef __INC_BLOCK_CLOCKLESS_H
#define __INC_BLOCK_CLOCKLESS_H
FASTLED_NAMESPACE_BEGIN
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
//
// Base template for clockless controllers. These controllers have 3 control points in their cycle for each bit. The first point
// is where the line is raised hi. The second pointsnt is where the line is dropped low for a zero. The third point is where the
// line is dropped low for a one. T1, T2, and T3 correspond to the timings for those three in clock cycles.
//
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
#if defined(__SAM3X8E__)
#define PORT_MASK (((1<<LANES)-1) & ((FIRST_PIN==2) ? 0xFF : 0xFF))
#define FASTLED_HAS_BLOCKLESS 1
#define PORTD_FIRST_PIN 25
#define PORTA_FIRST_PIN 69
#define PORTB_FIRST_PIN 90
typedef union {
uint8_t bytes[8];
uint32_t raw[2];
} Lines;
#define TADJUST 0
#define TOTAL ( (T1+TADJUST) + (T2+TADJUST) + (T3+TADJUST) )
#define T1_MARK (TOTAL - (T1+TADJUST))
#define T2_MARK (T1_MARK - (T2+TADJUST))
template <uint8_t LANES, int FIRST_PIN, int T1, int T2, int T3, EOrder RGB_ORDER = RGB, int XTRA0 = 0, bool FLIP = false, int WAIT_TIME = 50>
class InlineBlockClocklessController : public CLEDController {
typedef typename FastPin<FIRST_PIN>::port_ptr_t data_ptr_t;
typedef typename FastPin<FIRST_PIN>::port_t data_t;
data_t mPinMask;
data_ptr_t mPort;
CMinWait<WAIT_TIME> mWait;
public:
virtual void init() {
if(FIRST_PIN == PORTA_FIRST_PIN) {
switch(LANES) {
case 8: FastPin<31>::setOutput();
case 7: FastPin<58>::setOutput();
case 6: FastPin<100>::setOutput();
case 5: FastPin<59>::setOutput();
case 4: FastPin<60>::setOutput();
case 3: FastPin<61>::setOutput();
case 2: FastPin<68>::setOutput();
case 1: FastPin<69>::setOutput();
}
} else if(FIRST_PIN == PORTD_FIRST_PIN) {
switch(LANES) {
case 8: FastPin<11>::setOutput();
case 7: FastPin<29>::setOutput();
case 6: FastPin<15>::setOutput();
case 5: FastPin<14>::setOutput();
case 4: FastPin<28>::setOutput();
case 3: FastPin<27>::setOutput();
case 2: FastPin<26>::setOutput();
case 1: FastPin<25>::setOutput();
}
} else if(FIRST_PIN == PORTB_FIRST_PIN) {
switch(LANES) {
case 8: FastPin<97>::setOutput();
case 7: FastPin<96>::setOutput();
case 6: FastPin<95>::setOutput();
case 5: FastPin<94>::setOutput();
case 4: FastPin<93>::setOutput();
case 3: FastPin<92>::setOutput();
case 2: FastPin<91>::setOutput();
case 1: FastPin<90>::setOutput();
}
}
mPinMask = FastPin<FIRST_PIN>::mask();
mPort = FastPin<FIRST_PIN>::port();
}
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) {
MultiPixelController<LANES,PORT_MASK,RGB_ORDER> pixels(rgbdata,nLeds, scale, getDither() );
mWait.wait();
showRGBInternal(pixels, nLeds);
sei();
mWait.mark();
}
// #define ADV_RGB
#define ADV_RGB if(maskbit & PORT_MASK) { rgbdata += nLeds; } maskbit <<= 1;
virtual void show(const struct CRGB *rgbdata, int nLeds, CRGB scale) {
MultiPixelController<LANES,PORT_MASK,RGB_ORDER> pixels(rgbdata,nLeds, scale, getDither() );
mWait.wait();
showRGBInternal(pixels, nLeds);
mWait.mark();
}
#ifdef SUPPORT_ARGB
virtual void show(const struct CARGB *rgbdata, int nLeds, CRGB scale) {
mWait.wait();
showRGBInternal(PixelController<RGB_ORDER>(rgbdata, nLeds, scale, getDither()));
mWait.mark();
}
#endif
template<int BITS,int PX> __attribute__ ((always_inline)) inline static void writeBits(register uint32_t & next_mark, register Lines & b, Lines & b3, MultiPixelController<LANES, PORT_MASK, RGB_ORDER> &pixels) { // , register uint32_t & b2) {
register Lines b2;
transpose8x1(b.bytes,b2.bytes);
register uint8_t d = pixels.template getd<PX>(pixels);
register uint8_t scale = pixels.template getscale<PX>(pixels);
for(uint32_t i = 0; (i < LANES) && (i<8); i++) {
while(DUE_TIMER_VAL < next_mark);
next_mark = (DUE_TIMER_VAL+TOTAL);
*FastPin<FIRST_PIN>::sport() = PORT_MASK;
while((next_mark - DUE_TIMER_VAL) > (T2+T3+6));
*FastPin<FIRST_PIN>::cport() = (~b2.bytes[7-i]) & PORT_MASK;
while((next_mark - (DUE_TIMER_VAL)) > T3);
*FastPin<FIRST_PIN>::cport() = PORT_MASK;
b3.bytes[i] = pixels.template loadAndScale<PX>(pixels,i,d,scale);
}
for(uint32_t i = LANES; i < 8; i++) {
while(DUE_TIMER_VAL > next_mark);
next_mark = DUE_TIMER_VAL - (TOTAL-3);
*FastPin<FIRST_PIN>::sport() = PORT_MASK;
while((next_mark - DUE_TIMER_VAL) > (T2+T3+6));
*FastPin<FIRST_PIN>::cport() = (~b2.bytes[7-i]) & PORT_MASK;
while((next_mark - DUE_TIMER_VAL) > 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(MultiPixelController<LANES, PORT_MASK, RGB_ORDER> &allpixels, int nLeds) {
// Serial.println("Entering show");
// Setup the pixel controller and load/scale the first byte
Lines b0,b1,b2;
allpixels.preStepFirstByteDithering();
for(int i = 0; i < LANES; i++) {
b0.bytes[i] = allpixels.loadAndScale0(i);
}
// Setup and start the clock
TC_Configure(DUE_TIMER,DUE_TIMER_CHANNEL,TC_CMR_TCCLKS_TIMER_CLOCK1);
pmc_enable_periph_clk(DUE_TIMER_ID);
TC_Start(DUE_TIMER,DUE_TIMER_CHANNEL);
#if (FASTLED_ALLOW_INTERRUPTS == 1)
cli();
#endif
uint32_t next_mark = (DUE_TIMER_VAL + (TOTAL));
while(nLeds--) {
allpixels.stepDithering();
#if (FASTLED_ALLOW_INTERRUPTS == 1)
cli();
if(DUE_TIMER_VAL > next_mark) {
if((DUE_TIMER_VAL - next_mark) > ((WAIT_TIME-INTERRUPT_THRESHOLD)*CLKS_PER_US)) {
sei(); TC_Stop(DUE_TIMER,DUE_TIMER_CHANNEL); return DUE_TIMER_VAL;
}
}
#endif
// Write first byte, read next byte
writeBits<8+XTRA0,1>(next_mark, b0, b1, allpixels);
// Write second byte, read 3rd byte
writeBits<8+XTRA0,2>(next_mark, b1, b2, allpixels);
allpixels.advanceData();
// Write third byte
writeBits<8+XTRA0,0>(next_mark, b2, b0, allpixels);
#if (FASTLED_ALLOW_INTERRUPTS == 1)
sei();
#endif
}
return DUE_TIMER_VAL;
}
};
#endif
FASTLED_NAMESPACE_END
#endif
@@ -0,0 +1,11 @@
#ifndef __INC_FASTLED_ARM_SAM_H
#define __INC_FASTLED_ARM_SAM_H
// Include the sam headers
#include "fastled_delay.h"
#include "fastpin_arm_sam.h"
#include "fastspi_arm_sam.h"
#include "clockless_arm_sam.h"
#include "clockless_block_arm_sam.h"
#endif
@@ -0,0 +1,137 @@
#ifndef __INC_FASTPIN_ARM_SAM_H
#define __INC_FASTPIN_ARM_SAM_H
FASTLED_NAMESPACE_BEGIN
#if defined(FASTLED_FORCE_SOFTWARE_PINS)
#warning "Software pin support forced, pin access will be sloightly slower."
#define NO_HARDWARE_PIN_SUPPORT
#undef HAS_HARDWARE_PIN_SUPPORT
#else
/// Template definition for arduino due style ARM pins, providing direct access to the various GPIO registers. Note that this
/// uses the full port GPIO registers. In theory, in some way, bit-band register access -should- be faster, however I have found
/// that something about the way gcc does register allocation results in the bit-band code being slower. It will need more fine tuning.
/// The registers are data register, set output register, clear output register, set data direction register
template<uint8_t PIN, uint32_t _MASK, typename _PDOR, typename _PSOR, typename _PCOR, typename _PDDR> class _DUEPIN {
public:
typedef volatile uint32_t * port_ptr_t;
typedef uint32_t port_t;
inline static void setOutput() { pinMode(PIN, OUTPUT); } // TODO: perform MUX config { _PDDR::r() |= _MASK; }
inline static void setInput() { pinMode(PIN, INPUT); } // TODO: preform MUX config { _PDDR::r() &= ~_MASK; }
inline static void hi() __attribute__ ((always_inline)) { _PSOR::r() = _MASK; }
inline static void lo() __attribute__ ((always_inline)) { _PCOR::r() = _MASK; }
inline static void set(register port_t val) __attribute__ ((always_inline)) { _PDOR::r() = val; }
inline static void strobe() __attribute__ ((always_inline)) { toggle(); toggle(); }
inline static void toggle() __attribute__ ((always_inline)) { _PDOR::r() ^= _MASK; }
inline static void hi(register port_ptr_t port) __attribute__ ((always_inline)) { hi(); }
inline static void lo(register port_ptr_t port) __attribute__ ((always_inline)) { lo(); }
inline static void fastset(register port_ptr_t port, register port_t val) __attribute__ ((always_inline)) { *port = val; }
inline static port_t hival() __attribute__ ((always_inline)) { return _PDOR::r() | _MASK; }
inline static port_t loval() __attribute__ ((always_inline)) { return _PDOR::r() & ~_MASK; }
inline static port_ptr_t port() __attribute__ ((always_inline)) { return &_PDOR::r(); }
inline static port_ptr_t sport() __attribute__ ((always_inline)) { return &_PSOR::r(); }
inline static port_ptr_t cport() __attribute__ ((always_inline)) { return &_PCOR::r(); }
inline static port_t mask() __attribute__ ((always_inline)) { return _MASK; }
};
/// Template definition for DUE style ARM pins using bit banding, providing direct access to the various GPIO registers. GCC
/// does a poor job of optimizing around these accesses so they are not being used just yet.
template<uint8_t PIN, uint32_t _BIT, typename _PDOR, typename _PSOR, typename _PCOR, typename _PDDR> class _DUEPIN_BITBAND {
public:
typedef volatile uint32_t * port_ptr_t;
typedef uint32_t port_t;
inline static void setOutput() { pinMode(PIN, OUTPUT); } // TODO: perform MUX config { _PDDR::r() |= _MASK; }
inline static void setInput() { pinMode(PIN, INPUT); } // TODO: preform MUX config { _PDDR::r() &= ~_MASK; }
inline static void hi() __attribute__ ((always_inline)) { *_PDOR::template rx<_BIT>() = 1; }
inline static void lo() __attribute__ ((always_inline)) { *_PDOR::template rx<_BIT>() = 0; }
inline static void set(register port_t val) __attribute__ ((always_inline)) { *_PDOR::template rx<_BIT>() = val; }
inline static void strobe() __attribute__ ((always_inline)) { toggle(); toggle(); }
inline static void toggle() __attribute__ ((always_inline)) { *_PDOR::template rx<_BIT>() ^= 1; }
inline static void hi(register port_ptr_t port) __attribute__ ((always_inline)) { hi(); }
inline static void lo(register port_ptr_t port) __attribute__ ((always_inline)) { lo(); }
inline static void fastset(register port_ptr_t port, register port_t val) __attribute__ ((always_inline)) { *port = val; }
inline static port_t hival() __attribute__ ((always_inline)) { return 1; }
inline static port_t loval() __attribute__ ((always_inline)) { return 0; }
inline static port_ptr_t port() __attribute__ ((always_inline)) { return _PDOR::template rx<_BIT>(); }
inline static port_t mask() __attribute__ ((always_inline)) { return 1; }
};
#define GPIO_BITBAND_ADDR(reg, bit) (((uint32_t)&(reg) - 0x40000000) * 32 + (bit) * 4 + 0x42000000)
#define GPIO_BITBAND_PTR(reg, bit) ((uint32_t *)GPIO_BITBAND_ADDR((reg), (bit)))
#define _R(T) struct __gen_struct_ ## T
#define _RD32(T) struct __gen_struct_ ## T { static __attribute__((always_inline)) inline reg32_t r() { return T; } \
template<int BIT> static __attribute__((always_inline)) inline ptr_reg32_t rx() { return GPIO_BITBAND_PTR(T, BIT); } };
#define DUE_IO32(L) _RD32(REG_PIO ## L ## _ODSR); _RD32(REG_PIO ## L ## _SODR); _RD32(REG_PIO ## L ## _CODR); _RD32(REG_PIO ## L ## _OER);
#define _DEFPIN_DUE(PIN, BIT, L) template<> class FastPin<PIN> : public _DUEPIN<PIN, 1 << BIT, _R(REG_PIO ## L ## _ODSR), _R(REG_PIO ## L ## _SODR), _R(REG_PIO ## L ## _CODR), \
_R(GPIO ## L ## _OER)> {}; \
template<> class FastPinBB<PIN> : public _DUEPIN_BITBAND<PIN, BIT, _R(REG_PIO ## L ## _ODSR), _R(REG_PIO ## L ## _SODR), _R(REG_PIO ## L ## _CODR), \
_R(GPIO ## L ## _OER)> {};
#if defined(__SAM3X8E__)
DUE_IO32(A);
DUE_IO32(B);
DUE_IO32(C);
DUE_IO32(D);
#define MAX_PIN 78
_DEFPIN_DUE(0, 8, A); _DEFPIN_DUE(1, 9, A); _DEFPIN_DUE(2, 25, B); _DEFPIN_DUE(3, 28, C);
_DEFPIN_DUE(4, 26, C); _DEFPIN_DUE(5, 25, C); _DEFPIN_DUE(6, 24, C); _DEFPIN_DUE(7, 23, C);
_DEFPIN_DUE(8, 22, C); _DEFPIN_DUE(9, 21, C); _DEFPIN_DUE(10, 29, C); _DEFPIN_DUE(11, 7, D);
_DEFPIN_DUE(12, 8, D); _DEFPIN_DUE(13, 27, B); _DEFPIN_DUE(14, 4, D); _DEFPIN_DUE(15, 5, D);
_DEFPIN_DUE(16, 13, A); _DEFPIN_DUE(17, 12, A); _DEFPIN_DUE(18, 11, A); _DEFPIN_DUE(19, 10, A);
_DEFPIN_DUE(20, 12, B); _DEFPIN_DUE(21, 13, B); _DEFPIN_DUE(22, 26, B); _DEFPIN_DUE(23, 14, A);
_DEFPIN_DUE(24, 15, A); _DEFPIN_DUE(25, 0, D); _DEFPIN_DUE(26, 1, D); _DEFPIN_DUE(27, 2, D);
_DEFPIN_DUE(28, 3, D); _DEFPIN_DUE(29, 6, D); _DEFPIN_DUE(30, 9, D); _DEFPIN_DUE(31, 7, A);
_DEFPIN_DUE(32, 10, D); _DEFPIN_DUE(33, 1, C); _DEFPIN_DUE(34, 2, C); _DEFPIN_DUE(35, 3, C);
_DEFPIN_DUE(36, 4, C); _DEFPIN_DUE(37, 5, C); _DEFPIN_DUE(38, 6, C); _DEFPIN_DUE(39, 7, C);
_DEFPIN_DUE(40, 8, C); _DEFPIN_DUE(41, 9, C); _DEFPIN_DUE(42, 19, A); _DEFPIN_DUE(43, 20, A);
_DEFPIN_DUE(44, 19, C); _DEFPIN_DUE(45, 18, C); _DEFPIN_DUE(46, 17, C); _DEFPIN_DUE(47, 16, C);
_DEFPIN_DUE(48, 15, C); _DEFPIN_DUE(49, 14, C); _DEFPIN_DUE(50, 13, C); _DEFPIN_DUE(51, 12, C);
_DEFPIN_DUE(52, 21, B); _DEFPIN_DUE(53, 14, B); _DEFPIN_DUE(54, 16, A); _DEFPIN_DUE(55, 24, A);
_DEFPIN_DUE(56, 23, A); _DEFPIN_DUE(57, 22, A); _DEFPIN_DUE(58, 6, A); _DEFPIN_DUE(59, 4, A);
_DEFPIN_DUE(60, 3, A); _DEFPIN_DUE(61, 2, A); _DEFPIN_DUE(62, 17, B); _DEFPIN_DUE(63, 18, B);
_DEFPIN_DUE(64, 19, B); _DEFPIN_DUE(65, 20, B); _DEFPIN_DUE(66, 15, B); _DEFPIN_DUE(67, 16, B);
_DEFPIN_DUE(68, 1, A); _DEFPIN_DUE(69, 0, A); _DEFPIN_DUE(70, 17, A); _DEFPIN_DUE(71, 18, A);
_DEFPIN_DUE(72, 30, C); _DEFPIN_DUE(73, 21, A); _DEFPIN_DUE(74, 25, A); _DEFPIN_DUE(75, 26, A);
_DEFPIN_DUE(76, 27, A); _DEFPIN_DUE(77, 28, A); _DEFPIN_DUE(78, 23, B);
// digix pins
_DEFPIN_DUE(90, 0, B); _DEFPIN_DUE(91, 1, B); _DEFPIN_DUE(92, 2, B); _DEFPIN_DUE(93, 3, B);
_DEFPIN_DUE(94, 4, B); _DEFPIN_DUE(95, 5, B); _DEFPIN_DUE(96, 6, B); _DEFPIN_DUE(97, 7, B);
_DEFPIN_DUE(98, 8, B); _DEFPIN_DUE(99, 9, B); _DEFPIN_DUE(100, 5, A); _DEFPIN_DUE(101, 22, B);
_DEFPIN_DUE(102, 23, B); _DEFPIN_DUE(103, 24, B); _DEFPIN_DUE(104, 27, C); _DEFPIN_DUE(105, 20, C);
_DEFPIN_DUE(106, 11, C); _DEFPIN_DUE(107, 10, C); _DEFPIN_DUE(108, 21, A); _DEFPIN_DUE(109, 30, C);
_DEFPIN_DUE(110, 29, B); _DEFPIN_DUE(111, 30, B); _DEFPIN_DUE(112, 31, B); _DEFPIN_DUE(113, 28, B);
#define SPI_DATA 75
#define SPI_CLOCK 76
#define ARM_HARDWARE_SPI
#define HAS_HARDWARE_PIN_SUPPORT
#endif
#endif // FASTLED_FORCE_SOFTWARE_PINS
FASTLED_NAMESPACE_END
#endif // __INC_FASTPIN_ARM_SAM_H
@@ -0,0 +1,163 @@
#ifndef __INC_FASTSPI_ARM_SAM_H
#define __INC_FASTSPI_ARM_SAM_H
FASTLED_NAMESPACE_BEGIN
#if defined(__SAM3X8E__)
#define m_SPI ((Spi*)SPI0)
template <uint8_t _DATA_PIN, uint8_t _CLOCK_PIN, uint8_t _SPI_CLOCK_DIVIDER>
class SAMHardwareSPIOutput {
Selectable *m_pSelect;
static inline void waitForEmpty() { while ((m_SPI->SPI_SR & SPI_SR_TDRE) == 0); }
void enableConfig() { m_SPI->SPI_WPMR &= ~SPI_WPMR_WPEN; }
void disableConfig() { m_SPI->SPI_WPMR |= SPI_WPMR_WPEN; }
void enableSPI() { m_SPI->SPI_CR = SPI_CR_SPIEN; }
void disableSPI() { m_SPI->SPI_CR = SPI_CR_SPIDIS; }
void resetSPI() { m_SPI->SPI_CR = SPI_CR_SWRST; }
static inline void readyTransferBits(register uint32_t bits) {
bits -= 8;
// don't change the number of transfer bits while data is still being transferred from TDR to the shift register
waitForEmpty();
m_SPI->SPI_CSR[0] = SPI_CSR_NCPHA | SPI_CSR_CSAAT | (bits << SPI_CSR_BITS_Pos) | SPI_CSR_DLYBCT(1) | SPI_CSR_SCBR(_SPI_CLOCK_DIVIDER);
}
template<int BITS> static inline void writeBits(uint16_t w) {
waitForEmpty();
m_SPI->SPI_TDR = (uint32_t)w | SPI_PCS(0);
}
public:
SAMHardwareSPIOutput() { m_pSelect = NULL; }
SAMHardwareSPIOutput(Selectable *pSelect) { m_pSelect = pSelect; }
// set the object representing the selectable
void setSelect(Selectable *pSelect) { /* TODO */ }
// initialize the SPI subssytem
void init() {
// m_SPI = SPI0;
// set the output pins master out, master in, clock. Note doing this here because I still don't
// know how I want to expose this type of functionality in FastPin.
PIO_Configure(PIOA, PIO_PERIPH_A, FastPin<_DATA_PIN>::mask(), PIO_DEFAULT);
PIO_Configure(PIOA, PIO_PERIPH_A, FastPin<_DATA_PIN-1>::mask(), PIO_DEFAULT);
PIO_Configure(PIOA, PIO_PERIPH_A, FastPin<_CLOCK_PIN>::mask(), PIO_DEFAULT);
release();
// Configure the SPI clock, divider between 1-255
// SCBR = _SPI_CLOCK_DIVIDER
pmc_enable_periph_clk(ID_SPI0);
disableSPI();
// reset twice (what the sam code does, not sure why?)
resetSPI();
resetSPI();
// Configure SPI as master, enable
// Bits we want in MR: master, disable mode fault detection, variable peripheral select
m_SPI->SPI_MR = SPI_MR_MSTR | SPI_MR_MODFDIS | SPI_MR_PS;
enableSPI();
// Send everything out in 8 bit chunks, other sizes appear to work, poorly...
readyTransferBits(8);
}
// latch the CS select
void inline select() __attribute__((always_inline)) { if(m_pSelect != NULL) { m_pSelect->select(); } }
// release the CS select
void inline release() __attribute__((always_inline)) { if(m_pSelect != NULL) { m_pSelect->release(); } }
// wait until all queued up data has been written
void waitFully() { while((m_SPI->SPI_SR & SPI_SR_TXEMPTY) == 0); }
// write a byte out via SPI (returns immediately on writing register)
static void writeByte(uint8_t b) {
writeBits<8>(b);
}
// write a word out via SPI (returns immediately on writing register)
static void writeWord(uint16_t w) {
writeBits<16>(w);
}
// A raw set of writing byte values, assumes setup/init/waiting done elsewhere
static void writeBytesValueRaw(uint8_t value, int len) {
while(len--) { writeByte(value); }
}
// A full cycle of writing a value for len bytes, including select, release, and waiting
void writeBytesValue(uint8_t value, int len) {
select(); writeBytesValueRaw(value, len); release();
}
template <class D> void writeBytes(register uint8_t *data, int len) {
uint8_t *end = data + len;
select();
// could be optimized to write 16bit words out instead of 8bit bytes
while(data != end) {
writeByte(D::adjust(*data++));
}
D::postBlock(len);
waitFully();
release();
}
void writeBytes(register uint8_t *data, int len) { writeBytes<DATA_NOP>(data, len); }
// write a single bit out, which bit from the passed in byte is determined by template parameter
// not the most efficient mechanism in the world - but should be enough for sm16716 and friends
template <uint8_t BIT> inline void writeBit(uint8_t b) {
// need to wait for all exisiting data to go out the door, first
waitFully();
disableSPI();
if(b & (1 << BIT)) {
FastPin<_DATA_PIN>::hi();
} else {
FastPin<_DATA_PIN>::lo();
}
FastPin<_CLOCK_PIN>::hi();
FastPin<_CLOCK_PIN>::lo();
enableSPI();
}
// write a block of uint8_ts out in groups of three. len is the total number of uint8_ts to write out. The template
// parameters indicate how many uint8_ts to skip at the beginning and/or end of each grouping
template <uint8_t FLAGS, class D, EOrder RGB_ORDER> void writePixels(PixelController<RGB_ORDER> pixels) {
select();
int len = pixels.mLen;
if(FLAGS & FLAG_START_BIT) {
while(pixels.has(1)) {
writeBits<9>((1<<8) | D::adjust(pixels.loadAndScale0()));
writeByte(D::adjust(pixels.loadAndScale1()));
writeByte(D::adjust(pixels.loadAndScale2()));
pixels.advanceData();
pixels.stepDithering();
}
} else {
while(pixels.has(1)) {
writeByte(D::adjust(pixels.loadAndScale0()));
writeByte(D::adjust(pixels.loadAndScale1()));
writeByte(D::adjust(pixels.loadAndScale2()));
pixels.advanceData();
pixels.stepDithering();
}
}
D::postBlock(len);
release();
}
};
#endif
FASTLED_NAMESPACE_END
#endif
@@ -0,0 +1,39 @@
#ifndef __INC_LED_SYSDEFS_ARM_SAM_H
#define __INC_LED_SYSDEFS_ARM_SAM_H
#define FASTLED_ARM
// Setup DUE timer defines/channels/etc...
#ifndef DUE_TIMER_CHANNEL
#define DUE_TIMER_GROUP 0
#endif
#ifndef DUE_TIMER_CHANNEL
#define DUE_TIMER_CHANNEL 0
#endif
#define DUE_TIMER ((DUE_TIMER_GROUP==0) ? TC0 : ((DUE_TIMER_GROUP==1) ? TC1 : TC2))
#define DUE_TIMER_ID (ID_TC0 + (DUE_TIMER_GROUP*3) + DUE_TIMER_CHANNEL)
#define DUE_TIMER_VAL (DUE_TIMER->TC_CHANNEL[DUE_TIMER_CHANNEL].TC_CV << 1)
#define DUE_TIMER_RUNNING ((DUE_TIMER->TC_CHANNEL[DUE_TIMER_CHANNEL].TC_SR & TC_SR_CLKSTA) != 0)
#ifndef INTERRUPT_THRESHOLD
#define INTERRUPT_THRESHOLD 1
#endif
// Default to allowing interrupts
#ifndef FASTLED_ALLOW_INTERRUPTS
#define FASTLED_ALLOW_INTERRUPTS 1
#endif
#if FASTLED_ALLOW_INTERRUPTS == 1
#define FASTLED_ACCURATE_CLOCK
#endif
// reusing/abusing cli/sei defs for due
#define cli() __disable_irq(); __disable_fault_irq();
#define sei() __enable_irq(); __enable_fault_irq();
#endif
@@ -0,0 +1,147 @@
#ifndef __INC_CLOCKLESS_ARM_STM32_H
#define __INC_CLOCKLESS_ARM_STM32_H
FASTLED_NAMESPACE_BEGIN
// Definition for a single channel clockless controller for the stm32 family of chips, like that used in the spark core
// See clockless.h for detailed info on how the template parameters are used.
#define FASTLED_HAS_CLOCKLESS 1
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 CLEDController {
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;
public:
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 clearLeds(int nLeds) {
showColor(CRGB(0, 0, 0), nLeds, 0);
}
protected:
// 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> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
showRGBInternal(pixels);
mWait.mark();
}
virtual void show(const struct CRGB *rgbdata, int nLeds, CRGB scale) {
PixelController<RGB_ORDER> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
showRGBInternal(pixels);
mWait.mark();
}
#ifdef SUPPORT_ARGB
virtual void show(const struct CARGB *rgbdata, int nLeds, CRGB scale) {
PixelController<RGB_ORDER> pixels(rgbdata, nLeds, scale, getDither());
mWait.wait();
showRGBInternal(pixels);
mWait.mark();
}
#endif
#define _CYCCNT (*(volatile uint32_t*)(0xE0001004UL))
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(_CYCCNT < (T1+T2+T3-20));
FastPin<DATA_PIN>::fastset(port, hi);
_CYCCNT = 4;
if(b&0x80) {
while(_CYCCNT < (T1+T2-20));
FastPin<DATA_PIN>::fastset(port, lo);
} else {
while(_CYCCNT < (T1-10));
FastPin<DATA_PIN>::fastset(port, lo);
}
b <<= 1;
}
while(_CYCCNT < (T1+T2+T3-20));
FastPin<DATA_PIN>::fastset(port, hi);
_CYCCNT = 4;
if(b&0x80) {
while(_CYCCNT < (T1+T2-20));
FastPin<DATA_PIN>::fastset(port, lo);
} else {
while(_CYCCNT < (T1-10));
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
CoreDebug->DEMCR |= CoreDebug_DEMCR_TRCENA_Msk;
DWT->CTRL |= DWT_CTRL_CYCCNTENA_Msk;
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 = (T1+T2+T3);
DWT->CYCCNT = 0;
while(pixels.has(1)) {
pixels.stepDithering();
#if (FASTLED_ALLOW_INTERRUPTS == 1)
cli();
// if interrupts took longer than 45µs, punt on the current frame
if(DWT->CYCCNT > next_mark) {
if((DWT->CYCCNT-next_mark) > ((WAIT_TIME-INTERRUPT_THRESHOLD)*CLKS_PER_US)) { sei(); return DWT->CYCCNT; }
}
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 DWT->CYCCNT;
}
};
FASTLED_NAMESPACE_END
#endif
@@ -0,0 +1,10 @@
#ifndef __INC_FASTLED_ARM_SAM_H
#define __INC_FASTLED_ARM_SAM_H
// Include the sam headers
#include "fastled_delay.h"
#include "fastpin_arm_stm32.h"
// #include "fastspi_arm_stm32.h"
#include "clockless_arm_stm32.h"
#endif
@@ -0,0 +1,105 @@
#ifndef __FASTPIN_ARM_STM32_H
#define __FASTPIN_ARM_STM32_H
FASTLED_NAMESPACE_BEGIN
#if defined(FASTLED_FORCE_SOFTWARE_PINS)
#warning "Software pin support forced, pin access will be sloightly slower."
#define NO_HARDWARE_PIN_SUPPORT
#undef HAS_HARDWARE_PIN_SUPPORT
#else
/// Template definition for STM32 style ARM pins, providing direct access to the various GPIO registers. Note that this
/// uses the full port GPIO registers. In theory, in some way, bit-band register access -should- be faster, however I have found
/// that something about the way gcc does register allocation results in the bit-band code being slower. It will need more fine tuning.
/// The registers are data output, set output, clear output, toggle output, input, and direction
template<uint8_t PIN, uint8_t _BIT, uint32_t _MASK, typename _GPIO> class _ARMPIN {
public:
typedef volatile uint32_t * port_ptr_t;
typedef uint32_t port_t;
#if 0
inline static void setOutput() {
if(_BIT<8) {
_CRL::r() = (_CRL::r() & (0xF << (_BIT*4)) | (0x1 << (_BIT*4));
} else {
_CRH::r() = (_CRH::r() & (0xF << ((_BIT-8)*4))) | (0x1 << ((_BIT-8)*4));
}
}
inline static void setInput() { /* TODO */ } // TODO: preform MUX config { _PDDR::r() &= ~_MASK; }
#endif
inline static void setOutput() { pinMode(PIN, OUTPUT); } // TODO: perform MUX config { _PDDR::r() |= _MASK; }
inline static void setInput() { pinMode(PIN, INPUT); } // TODO: preform MUX config { _PDDR::r() &= ~_MASK; }
inline static void hi() __attribute__ ((always_inline)) { _GPIO::r()->BSRR = _MASK; }
inline static void lo() __attribute__ ((always_inline)) { _GPIO::r()->BRR = _MASK; }
// inline static void lo() __attribute__ ((always_inline)) { _GPIO::r()->BSRR = (_MASK<<16); }
inline static void set(register port_t val) __attribute__ ((always_inline)) { _GPIO::r()->ODR = val; }
inline static void strobe() __attribute__ ((always_inline)) { toggle(); toggle(); }
inline static void toggle() __attribute__ ((always_inline)) { if(_GPIO::r()->ODR & _MASK) { lo(); } else { hi(); } }
inline static void hi(register port_ptr_t port) __attribute__ ((always_inline)) { hi(); }
inline static void lo(register port_ptr_t port) __attribute__ ((always_inline)) { lo(); }
inline static void fastset(register port_ptr_t port, register port_t val) __attribute__ ((always_inline)) { *port = val; }
inline static port_t hival() __attribute__ ((always_inline)) { return _GPIO::r()->ODR | _MASK; }
inline static port_t loval() __attribute__ ((always_inline)) { return _GPIO::r()->ODR & ~_MASK; }
inline static port_ptr_t port() __attribute__ ((always_inline)) { return &_GPIO::r()->ODR; }
inline static port_ptr_t sport() __attribute__ ((always_inline)) { return &_GPIO::r()->BSRR; }
inline static port_ptr_t cport() __attribute__ ((always_inline)) { return &_GPIO::r()->BRR; }
inline static port_t mask() __attribute__ ((always_inline)) { return _MASK; }
};
#define _R(T) struct __gen_struct_ ## T
#define _RD32(T) struct __gen_struct_ ## T { static __attribute__((always_inline)) inline volatile GPIO_TypeDef * r() { return T; } };
#define _IO32(L) _RD32(GPIO ## L)
#define _DEFPIN_ARM(PIN, BIT, L) template<> class FastPin<PIN> : public _ARMPIN<PIN, BIT, 1 << BIT, _R(GPIO ## L)> {};
// Actual pin definitions
#if defined(SPARK)
_IO32(A); _IO32(B); _IO32(C); _IO32(D); _IO32(E); _IO32(F); _IO32(G);
#define MAX_PIN 19
_DEFPIN_ARM(0, 7, B);
_DEFPIN_ARM(1, 6, B);
_DEFPIN_ARM(2, 5, B);
_DEFPIN_ARM(3, 4, B);
_DEFPIN_ARM(4, 3, B);
_DEFPIN_ARM(5, 15, A);
_DEFPIN_ARM(6, 14, A);
_DEFPIN_ARM(7, 13, A);
_DEFPIN_ARM(8, 8, A);
_DEFPIN_ARM(9, 9, A);
_DEFPIN_ARM(10, 0, A);
_DEFPIN_ARM(11, 1, A);
_DEFPIN_ARM(12, 4, A);
_DEFPIN_ARM(13, 5, A);
_DEFPIN_ARM(14, 6, A);
_DEFPIN_ARM(15, 7, A);
_DEFPIN_ARM(16, 0, B);
_DEFPIN_ARM(17, 1, B);
_DEFPIN_ARM(18, 3, A);
_DEFPIN_ARM(19, 2, A);
#define SPI_DATA 15
#define SPI_CLOCK 13
#define HAS_HARDWARE_PIN_SUPPORT
#endif
#endif // FASTLED_FORCE_SOFTWARE_PINS
FASTLED_NAMESPACE_END
#endif // __INC_FASTPIN_ARM_STM32
@@ -0,0 +1,47 @@
#ifndef __INC_LED_SYSDEFS_ARM_SAM_H
#define __INC_LED_SYSDEFS_ARM_SAM_H
#include "application.h"
#define FASTLED_NAMESPACE_BEGIN namespace NSFastLED {
#define FASTLED_NAMESPACE_END }
#define FASTLED_USING_NAMESPACE using namespace NSFastLED;
#define FASTLED_ARM
#ifndef INTERRUPT_THRESHOLD
#define INTERRUPT_THRESHOLD 1
#endif
// Default to allowing interrupts
#ifndef FASTLED_ALLOW_INTERRUPTS
#define FASTLED_ALLOW_INTERRUPTS 0
#endif
#if FASTLED_ALLOW_INTERRUPTS == 1
#define FASTLED_ACCURATE_CLOCK
#endif
// reusing/abusing cli/sei defs for due
#define cli() __disable_irq(); __disable_fault_irq();
#define sei() __enable_irq(); __enable_fault_irq();
// pgmspace definitions
#define PROGMEM
#define pgm_read_dword(addr) (*(const unsigned long *)(addr))
#define pgm_read_dword_near(addr) pgm_read_dword(addr)
// Default to NOT using PROGMEM here
#ifndef FASTLED_USE_PROGMEM
#define FASTLED_USE_PROGMEM 0
#endif
// data type defs
typedef volatile uint8_t RoReg; /**< Read only 8-bit register (volatile const unsigned int) */
typedef volatile uint8_t RwReg; /**< Read-Write 8-bit register (volatile unsigned int) */
#define FASTLED_NO_PINMAP
#define F_CPU 72000000
#endif
@@ -0,0 +1,514 @@
#ifndef __INC_CLOCKLESS_TRINKET_H
#define __INC_CLOCKLESS_TRINKET_H
#include "controller.h"
#include "lib8tion.h"
#include <avr/interrupt.h> // for cli/se definitions
FASTLED_NAMESPACE_BEGIN
#if defined(FASTLED_AVR)
// Scaling macro choice
#ifndef TRINKET_SCALE
#define TRINKET_SCALE 1
// whether or not to use dithering
#define DITHER 1
#endif
#if (F_CPU==8000000)
#define FASTLED_SLOW_CLOCK_ADJUST asm __volatile__ ("mov r0,r0\n\t");
#else
#define FASTLED_SLOW_CLOCK_ADJUST
#endif
#define US_PER_TICK (64 / (F_CPU/1000000))
// Variations on the functions in delay.h - w/a loop var passed in to preserve registers across calls by the optimizer/compiler
template<int CYCLES> inline void _dc(register uint8_t & loopvar);
template<int _LOOP, int PAD> __attribute__((always_inline)) inline void _dc_AVR(register uint8_t & loopvar) {
_dc<PAD>(loopvar);
// The convolution in here is to ensure that the state of the carry flag coming into the delay loop is preserved
asm __volatile__ ( "BRCS L_PC%=\n\t"
" LDI %[loopvar], %[_LOOP]\n\tL_%=: DEC %[loopvar]\n\t BRNE L_%=\n\tBREQ L_DONE%=\n\t"
"L_PC%=: LDI %[loopvar], %[_LOOP]\n\tLL_%=: DEC %[loopvar]\n\t BRNE LL_%=\n\tBSET 0\n\t"
"L_DONE%=:\n\t"
:
[loopvar] "+a" (loopvar) : [_LOOP] "M" (_LOOP) : );
}
template<int CYCLES> __attribute__((always_inline)) inline void _dc(register uint8_t & loopvar) {
_dc_AVR<CYCLES/6,CYCLES%6>(loopvar);
}
template<> __attribute__((always_inline)) inline void _dc<-6>(register uint8_t & loopvar) {}
template<> __attribute__((always_inline)) inline void _dc<-5>(register uint8_t & loopvar) {}
template<> __attribute__((always_inline)) inline void _dc<-4>(register uint8_t & loopvar) {}
template<> __attribute__((always_inline)) inline void _dc<-3>(register uint8_t & loopvar) {}
template<> __attribute__((always_inline)) inline void _dc<-2>(register uint8_t & loopvar) {}
template<> __attribute__((always_inline)) inline void _dc<-1>(register uint8_t & loopvar) {}
template<> __attribute__((always_inline)) inline void _dc< 0>(register uint8_t & loopvar) {}
template<> __attribute__((always_inline)) inline void _dc< 1>(register uint8_t & loopvar) {asm __volatile__("mov r0,r0":::);}
template<> __attribute__((always_inline)) inline void _dc< 2>(register uint8_t & loopvar) {asm __volatile__("rjmp .+0":::);}
template<> __attribute__((always_inline)) inline void _dc< 3>(register uint8_t & loopvar) { _dc<2>(loopvar); _dc<1>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc< 4>(register uint8_t & loopvar) { _dc<2>(loopvar); _dc<2>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc< 5>(register uint8_t & loopvar) { _dc<2>(loopvar); _dc<3>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc< 6>(register uint8_t & loopvar) { _dc<2>(loopvar); _dc<2>(loopvar); _dc<2>(loopvar);}
template<> __attribute__((always_inline)) inline void _dc< 7>(register uint8_t & loopvar) { _dc<4>(loopvar); _dc<3>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc< 8>(register uint8_t & loopvar) { _dc<4>(loopvar); _dc<4>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc< 9>(register uint8_t & loopvar) { _dc<5>(loopvar); _dc<4>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc<10>(register uint8_t & loopvar) { _dc<6>(loopvar); _dc<4>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc<11>(register uint8_t & loopvar) { _dc<10>(loopvar); _dc<1>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc<12>(register uint8_t & loopvar) { _dc<10>(loopvar); _dc<2>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc<13>(register uint8_t & loopvar) { _dc<10>(loopvar); _dc<3>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc<14>(register uint8_t & loopvar) { _dc<10>(loopvar); _dc<4>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc<15>(register uint8_t & loopvar) { _dc<10>(loopvar); _dc<5>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc<16>(register uint8_t & loopvar) { _dc<10>(loopvar); _dc<6>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc<17>(register uint8_t & loopvar) { _dc<10>(loopvar); _dc<7>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc<18>(register uint8_t & loopvar) { _dc<10>(loopvar); _dc<8>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc<19>(register uint8_t & loopvar) { _dc<10>(loopvar); _dc<9>(loopvar); }
template<> __attribute__((always_inline)) inline void _dc<20>(register uint8_t & loopvar) { _dc<10>(loopvar); _dc<10>(loopvar); }
#define DINTPIN(T,ADJ,PINADJ) (T-(PINADJ+ADJ)>0) ? _dc<T-(PINADJ+ADJ)>(loopvar) : _dc<0>(loopvar);
#define DINT(T,ADJ) if(AVR_PIN_CYCLES(DATA_PIN)==1) { DINTPIN(T,ADJ,1) } else { DINTPIN(T,ADJ,2); }
#define D1(ADJ) DINT(T1,ADJ)
#define D2(ADJ) DINT(T2,ADJ)
#define D3(ADJ) DINT(T3,ADJ)
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
//
// Base template for clockless controllers. These controllers have 3 control points in their cycle for each bit. The first point
// is where the line is raised hi. The second point is where the line is dropped low for a zero. The third point is where the
// line is dropped low for a one. T1, T2, and T3 correspond to the timings for those three in clock cycles.
//
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
#if (!defined(NO_CORRECTION) || (NO_CORRECTION == 0)) && (FASTLED_ALLOW_INTERRUPTS == 0)
static uint8_t gTimeErrorAccum256ths;
#endif
#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 = 10>
class ClocklessController : public CLEDController {
typedef typename FastPin<DATA_PIN>::port_ptr_t data_ptr_t;
typedef typename FastPin<DATA_PIN>::port_t data_t;
CMinWait<WAIT_TIME> mWait;
public:
virtual void init() {
FastPin<DATA_PIN>::setOutput();
}
virtual uint16_t getMaxRefreshRate() const { return 400; }
virtual void clearLeds(int nLeds) {
CRGB zeros(0,0,0);
showAdjTime((uint8_t*)&zeros, nLeds, zeros, false, 0);
}
protected:
// set all the leds on the controller to a given color
virtual void showColor(const struct CRGB & rgbdata, int nLeds, CRGB scale) {
showAdjTime((uint8_t*)&rgbdata, nLeds, scale, false, 0);
}
virtual void show(const struct CRGB *rgbdata, int nLeds, CRGB scale) {
showAdjTime((uint8_t*)rgbdata, nLeds, scale, true, 0);
}
#ifdef SUPPORT_ARGB
virtual void show(const struct CARGB *rgbdata, int nLeds, CRGB scale) {
showAdjTime((uint8_t*)rgbdata, nLeds, scale, true, 1);
}
#endif
void showAdjTime(const uint8_t *data, int nLeds, CRGB & scale, bool advance, int skip) {
PixelController<RGB_ORDER> pixels(data, nLeds, scale, getDither(), advance, skip);
mWait.wait();
cli();
showRGBInternal(pixels);
// Adjust the timer
#if (!defined(NO_CORRECTION) || (NO_CORRECTION == 0)) && (FASTLED_ALLOW_INTERRUPTS == 0)
uint32_t microsTaken = (uint32_t)nLeds * (uint32_t)CLKS_TO_MICROS(24 * (T1 + T2 + T3));
// adust for approximate observed actal runtime (as of January 2015)
// roughly 9.6 cycles per pixel, which is 0.6us/pixel at 16MHz
// microsTaken += nLeds * 0.6 * CLKS_TO_MICROS(16);
microsTaken += scale16by8(nLeds,(0.6 * 256) + 1) * CLKS_TO_MICROS(16);
// if less than 1000us, there is NO timer impact,
// this is because the ONE interrupt that might come in while interrupts
// are disabled is queued up, and it will be serviced as soon as
// interrupts are re-enabled.
// This actually should technically also account for the runtime of the
// interrupt handler itself, but we're just not going to worry about that.
if( microsTaken > 1000) {
// Since up to one timer tick will be queued, we don't need
// to adjust the MS_COUNTER for that one.
microsTaken -= 1000;
// Now convert microseconds to 256ths of a second, approximately like this:
// 250ths = (us/4)
// 256ths = 250ths * (263/256);
uint16_t x256ths = microsTaken >> 2;
x256ths += scale16by8(x256ths,7);
x256ths += gTimeErrorAccum256ths;
MS_COUNTER += (x256ths >> 8);
gTimeErrorAccum256ths = x256ths & 0xFF;
}
#if 0
// For pixel counts of 30 and under at 16Mhz, no correction is necessary.
// For pixel counts of 15 and under at 8Mhz, no correction is necessary.
//
// This code, below, is smaller, and quicker clock correction, which drifts much
// more significantly, but is a few bytes smaller. Presented here for consideration
// as an alternate on the ATtiny, which can't have more than about 150 pixels MAX
// anyway, meaning that microsTaken will never be more than about 4,500, which fits in
// a 16-bit variable. The difference between /1000 and /1024 only starts showing
// up in the range of about 100 pixels, so many ATtiny projects won't even
// see a clock difference due to the approximation there.
uint16_t microsTaken = (uint32_t)nLeds * (uint32_t)CLKS_TO_MICROS((24) * (T1 + T2 + T3));
MS_COUNTER += (microsTaken >> 10);
#endif
#endif
sei();
mWait.mark();
}
#define USE_ASM_MACROS
// The variables that our various asm statemetns use. The same block of variables needs to be declared for
// all the asm blocks because GCC is pretty stupid and it would clobber variables happily or optimize code away too aggressively
#define ASM_VARS : /* write variables */ \
[count] "+x" (count), \
[data] "+z" (data), \
[b1] "+a" (b1), \
[d0] "+r" (d0), \
[d1] "+r" (d1), \
[d2] "+r" (d2), \
[loopvar] "+a" (loopvar), \
[scale_base] "+a" (scale_base) \
: /* use variables */ \
[ADV] "r" (advanceBy), \
[b0] "a" (b0), \
[hi] "r" (hi), \
[lo] "r" (lo), \
[s0] "r" (s0), \
[s1] "r" (s1), \
[s2] "r" (s2), \
[e0] "r" (e0), \
[e1] "r" (e1), \
[e2] "r" (e2), \
[PORT] "M" (FastPin<DATA_PIN>::port()-0x20), \
[O0] "M" (RGB_BYTE0(RGB_ORDER)), \
[O1] "M" (RGB_BYTE1(RGB_ORDER)), \
[O2] "M" (RGB_BYTE2(RGB_ORDER)) \
: "cc" /* clobber registers */
// Note: the code in the else in HI1/LO1 will be turned into an sts (2 cycle, 2 word) opcode
// 1 cycle, write hi to the port
#define HI1 FASTLED_SLOW_CLOCK_ADJUST if((int)(FastPin<DATA_PIN>::port())-0x20 < 64) { asm __volatile__("out %[PORT], %[hi]" ASM_VARS ); } else { *FastPin<DATA_PIN>::port()=hi; }
// 1 cycle, write lo to the port
#define LO1 if((int)(FastPin<DATA_PIN>::port())-0x20 < 64) { asm __volatile__("out %[PORT], %[lo]" ASM_VARS ); } else { *FastPin<DATA_PIN>::port()=lo; }
// 2 cycles, sbrs on flipping the line to lo if we're pushing out a 0
#define QLO2(B, N) asm __volatile__("sbrs %[" #B "], " #N ASM_VARS ); LO1;
// load a byte from ram into the given var with the given offset
#define LD2(B,O) asm __volatile__("ldd %[" #B "], Z + %[" #O "]\n\t" ASM_VARS );
// 4 cycles - load a byte from ram into the scaling scratch space with the given offset, clear the target var, clear carry
#define LDSCL4(B,O) asm __volatile__("ldd %[scale_base], Z + %[" #O "]\n\tclr %[" #B "]\n\tclc\n\t" ASM_VARS );
#if (DITHER==1)
// apply dithering value before we do anything with scale_base
#define PRESCALE4(D) asm __volatile__("cpse %[scale_base], __zero_reg__\n\t add %[scale_base],%[" #D "]\n\tbrcc L_%=\n\tldi %[scale_base], 0xFF\n\tL_%=:\n\t" ASM_VARS);
// Do the add for the prescale
#define PRESCALEA2(D) asm __volatile__("cpse %[scale_base], __zero_reg__\n\t add %[scale_base],%[" #D "]\n\t" ASM_VARS);
// Do the clamp for the prescale, clear carry when we're done - NOTE: Must ensure carry flag state is preserved!
#define PRESCALEB3(D) asm __volatile__("brcc L_%=\n\tldi %[scale_base], 0xFF\n\tL_%=:\n\tCLC" ASM_VARS);
#else
#define PRESCALE4(D) _dc<4>(loopvar);
#define PRESCALEA2(D) _dc<2>(loopvar);
#define PRESCALEB3(D) _dc<3>(loopvar);
#endif
// 2 cycles - perform one step of the scaling (if a given bit is set in scale, add scale-base to the scratch space)
#define _SCALE02(B, N) "sbrc %[s0], " #N "\n\tadd %[" #B "], %[scale_base]\n\t"
#define _SCALE12(B, N) "sbrc %[s1], " #N "\n\tadd %[" #B "], %[scale_base]\n\t"
#define _SCALE22(B, N) "sbrc %[s2], " #N "\n\tadd %[" #B "], %[scale_base]\n\t"
#define SCALE02(B,N) asm __volatile__( _SCALE02(B,N) ASM_VARS );
#define SCALE12(B,N) asm __volatile__( _SCALE12(B,N) ASM_VARS );
#define SCALE22(B,N) asm __volatile__( _SCALE22(B,N) ASM_VARS );
// 1 cycle - rotate right, pulling in from carry
#define _ROR1(B) "ror %[" #B "]\n\t"
#define ROR1(B) asm __volatile__( _ROR1(B) ASM_VARS);
// 1 cycle, clear the carry bit
#define _CLC1 "clc\n\t"
#define CLC1 asm __volatile__( _CLC1 ASM_VARS );
// 2 cycles, rortate right, pulling in from carry then clear the carry bit
#define RORCLC2(B) asm __volatile__( _ROR1(B) _CLC1 ASM_VARS );
// 4 cycles, rotate, clear carry, scale next bit
#define RORSC04(B, N) asm __volatile__( _ROR1(B) _CLC1 _SCALE02(B, N) ASM_VARS );
#define RORSC14(B, N) asm __volatile__( _ROR1(B) _CLC1 _SCALE12(B, N) ASM_VARS );
#define RORSC24(B, N) asm __volatile__( _ROR1(B) _CLC1 _SCALE22(B, N) ASM_VARS );
// 4 cycles, scale bit, rotate, clear carry
#define SCROR04(B, N) asm __volatile__( _SCALE02(B,N) _ROR1(B) _CLC1 ASM_VARS );
#define SCROR14(B, N) asm __volatile__( _SCALE12(B,N) _ROR1(B) _CLC1 ASM_VARS );
#define SCROR24(B, N) asm __volatile__( _SCALE22(B,N) _ROR1(B) _CLC1 ASM_VARS );
/////////////////////////////////////////////////////////////////////////////////////
// Loop life cycle
// dither adjustment macro - should be kept in sync w/what's in stepDithering
// #define ADJDITHER2(D, E) D = E - D;
#define ADJDITHER2(D, E) asm __volatile__ ("neg %[" #D "]\n\tadd %[" #D "],%[" #E "]\n\t" ASM_VARS);
// #define xstr(a) str(a)
// #define str(a) #a
// #define ADJDITHER2(D,E) asm __volatile__("subi %[" #D "], " xstr(DUSE) "\n\tand %[" #D "], %[" #E "]\n\t" ASM_VARS);
// define the beginning of the loop
#define LOOP asm __volatile__("1:" ASM_VARS );
// define the end of the loop
#define DONE asm __volatile__("2:" ASM_VARS );
// 2 cycles - increment the data pointer
#define IDATA2 asm __volatile__("add %A[data], %[ADV]\n\tadc %B[data], __zero_reg__\n\t" ASM_VARS );
#define IDATACLC3 asm __volatile__("add %A[data], %[ADV]\n\tadc %B[data], __zero_reg__\n\t" _CLC1 ASM_VARS );
// 1 cycle mov
#define MOV1(B1, B2) asm __volatile__("mov %[" #B1 "], %[" #B2 "]" ASM_VARS );
// 2 cycles - decrement the counter
#define DCOUNT2 asm __volatile__("sbiw %[count], 1" ASM_VARS );
// 2 cycles - jump to the beginning of the loop
#define JMPLOOP2 asm __volatile__("rjmp 1b" ASM_VARS );
// 2 cycles - jump out of the loop
#define BRLOOP1 asm __volatile__("brne 3\n\trjmp 2f\n\t3:" ASM_VARS );
// 5 cycles 2 sbiw, 3 for the breq/rjmp
#define ENDLOOP5 asm __volatile__("sbiw %[count], 1\n\tbreq L_%=\n\trjmp 1b\n\tL_%=:\n\t" ASM_VARS);
// NOP using the variables, forcing a move
#define DNOP asm __volatile__("mov r0,r0" ASM_VARS);
#define DADVANCE 3
#define DUSE (0xFF - (DADVANCE-1))
// 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 void /*__attribute__((optimize("O0")))*/ /*__attribute__ ((always_inline))*/ showRGBInternal(PixelController<RGB_ORDER> & pixels) {
uint8_t *data = (uint8_t*)pixels.mData;
data_ptr_t port = FastPin<DATA_PIN>::port();
data_t mask = FastPin<DATA_PIN>::mask();
uint8_t scale_base = 0;
// register uint8_t *end = data + nLeds;
data_t hi = *port | mask;
data_t lo = *port & ~mask;
*port = lo;
// the byte currently being written out
uint8_t b0 = 0;
// the byte currently being worked on to write the next out
uint8_t b1 = 0;
// Setup the pixel controller
pixels.preStepFirstByteDithering();
// pull the dithering/adjustment values out of the pixels object for direct asm access
uint8_t advanceBy = pixels.advanceBy();
uint16_t count = pixels.mLen;
uint8_t s0 = pixels.mScale.raw[RO(0)];
uint8_t s1 = pixels.mScale.raw[RO(1)];
uint8_t s2 = pixels.mScale.raw[RO(2)];
uint8_t d0 = pixels.d[RO(0)];
uint8_t d1 = pixels.d[RO(1)];
uint8_t d2 = pixels.d[RO(2)];
uint8_t e0 = pixels.e[RO(0)];
uint8_t e1 = pixels.e[RO(1)];
uint8_t e2 = pixels.e[RO(2)];
uint8_t loopvar=0;
// load/scale the first byte
#if !defined(LIB8_ATTINY)
// we have a hardware multiply, can use loadAndScale0
b0 = pixels.loadAndScale0();
#else
// no hardware multiply, we have to do our own mul by hand here, lest we incur a
// function call which will kill all of our register usage/allocations below
b0 = data[RO(0)];
{
LDSCL4(b0,O0) PRESCALEA2(d0)
PRESCALEB3(d0) SCALE02(b0,0)
RORSC04(b0,1) ROR1(b0) CLC1
SCROR04(b0,2) SCALE02(b0,3)
RORSC04(b0,4) ROR1(b0) CLC1
SCROR04(b0,5) SCALE02(b0,6)
RORSC04(b0,7) ROR1(b0) CLC1
}
#endif
// #if (FASTLED_ALLOW_INTERRUPTS == 1)
// TCCR0A |= 0x30;
// OCR0B = (uint8_t)(TCNT0 + ((WAIT_TIME-INTERRUPT_THRESHOLD)/US_PER_TICK));
// TIFR0 = 0x04;
// #endif
{
// while(--count)
{
// Loop beginning, does some stuff that's outside of the pixel write cycle, namely incrementing d0-2 and masking off
// by the E values (see the definition )
DNOP;
LOOP;
// ADJDITHER2(d0,e0);
// ADJDITHER2(d1,e1);
// ADJDITHER2(d2,e2);
// NOP;
// #if (FASTLED_ALLOW_INTERRUPTS == 1)
// cli();
// if(TIFR0 & 0x04) {
// sei();
// TCCR0A &= ~0x30;
// return;
// }
// hi = *port | mask;
// lo = *port & ~mask;
// #endif
// Sum of the clock counts across each row should be 10 for 8Mhz, WS2811
// The values in the D1/D2/D3 indicate how many cycles the previous column takes
// to allow things to line back up.
//
// While writing out byte 0, we're loading up byte 1, applying the dithering adjustment,
// then scaling it using 8 cycles of shift/add interleaved in between writing the bits
// out. When doing byte 1, we're doing the above for byte 2. When we're doing byte 2,
// we're cycling back around and doing the above for byte 0.
#if TRINKET_SCALE
// Inline scaling - RGB ordering
// DNOP
HI1 D1(1) QLO2(b0, 7) LDSCL4(b1,O1) D2(4) LO1 PRESCALEA2(d1) D3(2)
HI1 D1(1) QLO2(b0, 6) PRESCALEB3(d1) D2(3) LO1 SCALE12(b1,0) D3(2)
HI1 D1(1) QLO2(b0, 5) RORSC14(b1,1) D2(4) LO1 RORCLC2(b1) D3(2)
HI1 D1(1) QLO2(b0, 4) SCROR14(b1,2) D2(4) LO1 SCALE12(b1,3) D3(2)
HI1 D1(1) QLO2(b0, 3) RORSC14(b1,4) D2(4) LO1 RORCLC2(b1) D3(2)
HI1 D1(1) QLO2(b0, 2) SCROR14(b1,5) D2(4) LO1 SCALE12(b1,6) D3(2)
HI1 D1(1) QLO2(b0, 1) RORSC14(b1,7) D2(4) LO1 RORCLC2(b1) D3(2)
HI1 D1(1) QLO2(b0, 0)
switch(XTRA0) {
case 4: D2(0) LO1 D3(0) HI1 D1(1) QLO2(b0,0)
case 3: D2(0) LO1 D3(0) HI1 D1(1) QLO2(b0,0)
case 2: D2(0) LO1 D3(0) HI1 D1(1) QLO2(b0,0)
case 1: D2(0) LO1 D3(0) HI1 D1(1) QLO2(b0,0)
}
ADJDITHER2(d1,e1) D2(2) LO1 MOV1(b0,b1) D3(1)
HI1 D1(1) QLO2(b0, 7) LDSCL4(b1,O2) D2(4) LO1 PRESCALEA2(d2) D3(2)
HI1 D1(1) QLO2(b0, 6) PRESCALEB3(d2) D2(3) LO1 SCALE22(b1,0) D3(2)
HI1 D1(1) QLO2(b0, 5) RORSC24(b1,1) D2(4) LO1 RORCLC2(b1) D3(2)
HI1 D1(1) QLO2(b0, 4) SCROR24(b1,2) D2(4) LO1 SCALE22(b1,3) D3(2)
HI1 D1(1) QLO2(b0, 3) RORSC24(b1,4) D2(4) LO1 RORCLC2(b1) D3(2)
HI1 D1(1) QLO2(b0, 2) SCROR24(b1,5) D2(4) LO1 SCALE22(b1,6) D3(2)
HI1 D1(1) QLO2(b0, 1) RORSC24(b1,7) D2(4) LO1 RORCLC2(b1) D3(2)
HI1 D1(1) QLO2(b0, 0)
switch(XTRA0) {
case 4: D2(0) LO1 D3(0) HI1 D1(1) QLO2(b0,0)
case 3: D2(0) LO1 D3(0) HI1 D1(1) QLO2(b0,0)
case 2: D2(0) LO1 D3(0) HI1 D1(1) QLO2(b0,0)
case 1: D2(0) LO1 D3(0) HI1 D1(1) QLO2(b0,0)
}
IDATACLC3 MOV1(b0,b1) D2(4) LO1 ADJDITHER2(d2,e2) D3(2)
HI1 D1(1) QLO2(b0, 7) LDSCL4(b1,O0) D2(4) LO1 PRESCALEA2(d0) D3(2)
HI1 D1(1) QLO2(b0, 6) PRESCALEB3(d0) D2(3) LO1 SCALE02(b1,0) D3(2)
HI1 D1(1) QLO2(b0, 5) RORSC04(b1,1) D2(4) LO1 RORCLC2(b1) D3(2)
HI1 D1(1) QLO2(b0, 4) SCROR04(b1,2) D2(4) LO1 SCALE02(b1,3) D3(2)
HI1 D1(1) QLO2(b0, 3) RORSC04(b1,4) D2(4) LO1 RORCLC2(b1) D3(2)
HI1 D1(1) QLO2(b0, 2) SCROR04(b1,5) D2(4) LO1 SCALE02(b1,6) D3(2)
HI1 D1(1) QLO2(b0, 1) RORSC04(b1,7) D2(4) LO1 RORCLC2(b1) D3(2)
HI1 D1(1) QLO2(b0, 0)
switch(XTRA0) {
case 4: D2(0) LO1 D3(0) HI1 D1(1) QLO2(b0,0)
case 3: D2(0) LO1 D3(0) HI1 D1(1) QLO2(b0,0)
case 2: D2(0) LO1 D3(0) HI1 D1(1) QLO2(b0,0)
case 1: D2(0) LO1 D3(0) HI1 D1(1) QLO2(b0,0)
}
ADJDITHER2(d0,e0) MOV1(b0,b1) D2(3) LO1 D3(6)
ENDLOOP5
#else
// no inline scaling - non-straight RGB ordering -- no longer in line with the actual asm macros above, left for
// reference only
HI1 D1(1) QLO2(b0, 7) LD2(b1,O1) D2(2) LO1 D3(0)
HI1 D1(1) QLO2(b0, 6) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b0, 5) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b0, 4) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b0, 3) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b0, 2) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b0, 1) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b0, 0) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 7) LD2(b1,O2) D2(2) LO1 D3(0)
HI1 D1(1) QLO2(b1, 6) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 5) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 4) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 3) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 2) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 1) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 0) IDATA2 D2(2) LO1 D3(0)
HI1 D1(1) QLO2(b1, 7) LD2(b0,O0) D2(2) LO1 D3(0)
HI1 D1(1) QLO2(b1, 6) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 5) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 4) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 3) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 2) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 1) D2(0) LO1 D3(0)
HI1 D1(1) QLO2(b1, 0) D2(0) LO1 D3(0)
#endif
// #if (FASTLED_ALLOW_INTERRUPTS == 1)
// // set the counter mark
// OCR0B = (uint8_t)(TCNT0 + ((WAIT_TIME-INTERRUPT_THRESHOLD)/US_PER_TICK));
// TIFR0 = 0x04;
// sei();
// #endif
}
DONE;
}
#if (FASTLED_ALLOW_INTERRUPTS == 1)
// stop using the clock juggler
TCCR0A &= ~0x30;
#endif
}
#ifdef SUPPORT_ARGB
virtual void showARGB(struct CARGB *data, int nLeds) {
// TODO: IMPLEMENTME
}
#endif
};
#endif
FASTLED_NAMESPACE_END
#endif
+14
View File
@@ -0,0 +1,14 @@
#ifndef __INC_FASTLED_AVR_H
#define __INC_FASTLED_AVR_H
#include "fastled_delay.h"
#include "fastpin_avr.h"
#include "fastspi_avr.h"
#include "clockless_trinket.h"
// Default to using PROGMEM
#ifndef FASTLED_USE_PROGMEM
#define FASTLED_USE_PROGMEM 1
#endif
#endif
+280
View File
@@ -0,0 +1,280 @@
#ifndef __INC_FASTPIN_AVR_H
#define __INC_FASTPIN_AVR_H
FASTLED_NAMESPACE_BEGIN
#if defined(FASTLED_FORCE_SOFTWARE_PINS)
#warning "Software pin support forced, pin access will be sloightly slower."
#define NO_HARDWARE_PIN_SUPPORT
#undef HAS_HARDWARE_PIN_SUPPORT
#else
#define AVR_PIN_CYCLES(_PIN) ((((int)FastPin<_PIN>::port())-0x20 < 64) ? 1 : 2)
/// Class definition for a Pin where we know the port registers at compile time for said pin. This allows us to make
/// a lot of optimizations, as the inlined hi/lo methods will devolve to a single io register write/bitset.
template<uint8_t PIN, uint8_t _MASK, typename _PORT, typename _DDR, typename _PIN> class _AVRPIN {
public:
typedef volatile uint8_t * port_ptr_t;
typedef uint8_t port_t;
inline static void setOutput() { _DDR::r() |= _MASK; }
inline static void setInput() { _DDR::r() &= ~_MASK; }
inline static void hi() __attribute__ ((always_inline)) { _PORT::r() |= _MASK; }
inline static void lo() __attribute__ ((always_inline)) { _PORT::r() &= ~_MASK; }
inline static void set(register uint8_t val) __attribute__ ((always_inline)) { _PORT::r() = val; }
inline static void strobe() __attribute__ ((always_inline)) { toggle(); toggle(); }
inline static void toggle() __attribute__ ((always_inline)) { _PIN::r() = _MASK; }
inline static void hi(register port_ptr_t port) __attribute__ ((always_inline)) { hi(); }
inline static void lo(register port_ptr_t port) __attribute__ ((always_inline)) { lo(); }
inline static void fastset(register port_ptr_t port, register uint8_t val) __attribute__ ((always_inline)) { set(val); }
inline static port_t hival() __attribute__ ((always_inline)) { return _PORT::r() | _MASK; }
inline static port_t loval() __attribute__ ((always_inline)) { return _PORT::r() & ~_MASK; }
inline static port_ptr_t port() __attribute__ ((always_inline)) { return &_PORT::r(); }
inline static port_t mask() __attribute__ ((always_inline)) { return _MASK; }
};
/// AVR definitions for pins. Getting around the fact that I can't pass GPIO register addresses in as template arguments by instead creating
/// a custom type for each GPIO register with a single, static, aggressively inlined function that returns that specific GPIO register. A similar
/// trick is used a bit further below for the ARM GPIO registers (of which there are far more than on AVR!)
typedef volatile uint8_t & reg8_t;
#define _R(T) struct __gen_struct_ ## T
#define _RD8(T) struct __gen_struct_ ## T { static inline reg8_t r() { return T; }};
#define _IO(L) _RD8(DDR ## L); _RD8(PORT ## L); _RD8(PIN ## L);
#define _DEFPIN_AVR(_PIN, MASK, L) template<> class FastPin<_PIN> : public _AVRPIN<_PIN, MASK, _R(PORT ## L), _R(DDR ## L), _R(PIN ## L)> {};
#if defined(__AVR_ATtiny85__) || defined(__AVR_ATtiny45__)
_IO(B);
#define MAX_PIN 5
_DEFPIN_AVR(0, 0x01, B); _DEFPIN_AVR(1, 0x02, B); _DEFPIN_AVR(2, 0x04, B); _DEFPIN_AVR(3, 0x08, B);
_DEFPIN_AVR(4, 0x10, B); _DEFPIN_AVR(5, 0x20, B);
#define HAS_HARDWARE_PIN_SUPPORT 1
#elif defined(ARDUINO_AVR_DIGISPARK) // digispark pin layout
#define MAX_PIN 5
#define HAS_HARDWARE_PIN_SUPPORT 1
_IO(A); _IO(B);
_DEFPIN_AVR(0, 0x01, B); _DEFPIN_AVR(1, 0x02, B); _DEFPIN_AVR(2, 0x04, B);
_DEFPIN_AVR(3, 0x80, A); _DEFPIN_AVR(4, 0x40, A); _DEFPIN_AVR(5, 0x20, A);
#elif defined(__AVR_ATtiny24__) || defined(__AVR_ATtiny44__) || defined(__AVR_ATtiny84__) || defined(__AVR_ATtiny25__)
_IO(A); _IO(B);
#define MAX_PIN 10
_DEFPIN_AVR(0, 0x01, A); _DEFPIN_AVR(1, 0x02, A); _DEFPIN_AVR(2, 0x04, A); _DEFPIN_AVR(3, 0x08, A);
_DEFPIN_AVR(4, 0x10, A); _DEFPIN_AVR(5, 0x20, A); _DEFPIN_AVR(6, 0x40, A); _DEFPIN_AVR(7, 0x80, A);
_DEFPIN_AVR(8, 0x04, B); _DEFPIN_AVR(9, 0x02, B); _DEFPIN_AVR(10, 0x01, B);
#define HAS_HARDWARE_PIN_SUPPORT 1
#elif defined(ARDUINO_AVR_DIGISPARKPRO)
_IO(A); _IO(B);
#define MAX_PIN 12
_DEFPIN_AVR(0, 0x01, B); _DEFPIN_AVR(1, 0x02, B); _DEFPIN_AVR(2, 0x04, B); _DEFPIN_AVR(3, 0x20, B);
_DEFPIN_AVR(4, 0x08, B); _DEFPIN_AVR(5, 0x80, A); _DEFPIN_AVR(6, 0x01, A); _DEFPIN_AVR(7, 0x02, A);
_DEFPIN_AVR(8, 0x04, A); _DEFPIN_AVR(9, 0x08, A); _DEFPIN_AVR(10, 0x10, A); _DEFPIN_AVR(11, 0x20, A);
_DEFPIN_AVR(12, 0x40, A);
#elif defined(__AVR_ATtiny167__) || defined(__AVR_ATtiny87__)
_IO(A); _IO(B);
#define MAX_PIN 15
_DEFPIN_AVR(0, 0x01, A); _DEFPIN_AVR(1, 0x02, A); _DEFPIN_AVR(2, 0x04, A); _DEFPIN_AVR(3, 0x08, A);
_DEFPIN_AVR(4, 0x10, A); _DEFPIN_AVR(5, 0x20, A); _DEFPIN_AVR(6, 0x40, A); _DEFPIN_AVR(7, 0x80, A);
_DEFPIN_AVR(8, 0x01, B); _DEFPIN_AVR(9, 0x02, B); _DEFPIN_AVR(10, 0x04, B); _DEFPIN_AVR(11, 0x08, B);
_DEFPIN_AVR(12, 0x10, B); _DEFPIN_AVR(13, 0x20, B); _DEFPIN_AVR(14, 0x40, B); _DEFPIN_AVR(15, 0x80, B);
#define SPI_DATA 4
#define SPI_CLOCK 5
#define AVR_HARDWARE_SPI 1
#define HAS_HARDWARE_PIN_SUPPORT 1
#elif defined(ARDUINO_HOODLOADER2) && (defined(__AVR_ATmega32U2__) || defined(__AVR_ATmega16U2__) || defined(__AVR_ATmega8U2__)) || defined(__AVR_AT90USB82__) || defined(__AVR_AT90USB162__)
_IO(D); _IO(B); _IO(C);
#define MAX_PIN 20
_DEFPIN_AVR( 0, 0x01, B); _DEFPIN_AVR( 1, 0x02, B); _DEFPIN_AVR( 2, 0x04, B); _DEFPIN_AVR( 3, 0x08, B);
_DEFPIN_AVR( 4, 0x10, B); _DEFPIN_AVR( 5, 0x20, B); _DEFPIN_AVR( 6, 0x40, B); _DEFPIN_AVR( 7, 0x80, B);
_DEFPIN_AVR( 8, 0x80, C); _DEFPIN_AVR( 9, 0x40, C); _DEFPIN_AVR( 10, 0x20,C); _DEFPIN_AVR( 11, 0x10, C);
_DEFPIN_AVR( 12, 0x04, C); _DEFPIN_AVR( 13, 0x01, D); _DEFPIN_AVR( 14, 0x02, D); _DEFPIN_AVR(15, 0x04, D);
_DEFPIN_AVR( 16, 0x08, D); _DEFPIN_AVR( 17, 0x10, D); _DEFPIN_AVR( 18, 0x20, D); _DEFPIN_AVR( 19, 0x40, D);
_DEFPIN_AVR( 20, 0x80, D);
#define HAS_HARDWARE_PIN_SUPPORT 1
// #define SPI_DATA 2
// #define SPI_CLOCK 1
// #define AVR_HARDWARE_SPI 1
#elif defined(__AVR_ATmega328P__) || defined(__AVR_ATmega168__) || defined(__AVR_ATmega168P__) || defined(__AVR_ATmega8__)
// Accelerated port definitions for arduino avrs
_IO(D); _IO(B); _IO(C);
#define MAX_PIN 19
_DEFPIN_AVR( 0, 0x01, D); _DEFPIN_AVR( 1, 0x02, D); _DEFPIN_AVR( 2, 0x04, D); _DEFPIN_AVR( 3, 0x08, D);
_DEFPIN_AVR( 4, 0x10, D); _DEFPIN_AVR( 5, 0x20, D); _DEFPIN_AVR( 6, 0x40, D); _DEFPIN_AVR( 7, 0x80, D);
_DEFPIN_AVR( 8, 0x01, B); _DEFPIN_AVR( 9, 0x02, B); _DEFPIN_AVR(10, 0x04, B); _DEFPIN_AVR(11, 0x08, B);
_DEFPIN_AVR(12, 0x10, B); _DEFPIN_AVR(13, 0x20, B); _DEFPIN_AVR(14, 0x01, C); _DEFPIN_AVR(15, 0x02, C);
_DEFPIN_AVR(16, 0x04, C); _DEFPIN_AVR(17, 0x08, C); _DEFPIN_AVR(18, 0x10, C); _DEFPIN_AVR(19, 0x20, C);
#define SPI_DATA 11
#define SPI_CLOCK 13
#define SPI_SELECT 10
#define AVR_HARDWARE_SPI 1
#define HAS_HARDWARE_PIN_SUPPORT 1
#define SPI_UART0_DATA 1
#define SPI_UART0_CLOCK 4
#elif defined(__AVR_ATmega1284P__)
_IO(A); _IO(B); _IO(C); _IO(D);
_DEFPIN_AVR(0, 1<<0, D); _DEFPIN_AVR(1, 1<<1, D); _DEFPIN_AVR(2, 1<<2, B); _DEFPIN_AVR(3, 1<<3, B);
_DEFPIN_AVR(4, 1<<0, B); _DEFPIN_AVR(5, 1<<1, B); _DEFPIN_AVR(6, 1<<2, D); _DEFPIN_AVR(7, 1<<3, D);
_DEFPIN_AVR(8, 1<<5, D); _DEFPIN_AVR(9, 1<<6, D); _DEFPIN_AVR(10, 1<<4, B); _DEFPIN_AVR(11, 1<<5, B);
_DEFPIN_AVR(12, 1<<6, B); _DEFPIN_AVR(13, 1<<7, B); _DEFPIN_AVR(14, 1<<7, A); _DEFPIN_AVR(15, 1<<6, A);
_DEFPIN_AVR(16, 1<<5, A); _DEFPIN_AVR(17, 1<<4, A); _DEFPIN_AVR(18, 1<<3, A); _DEFPIN_AVR(19, 1<<2, A);
_DEFPIN_AVR(20, 1<<1, A); _DEFPIN_AVR(21, 1<<0, A); _DEFPIN_AVR(22, 1<<0, C); _DEFPIN_AVR(23, 1<<1, C);
_DEFPIN_AVR(24, 1<<2, C); _DEFPIN_AVR(25, 1<<3, C); _DEFPIN_AVR(26, 1<<4, C); _DEFPIN_AVR(27, 1<<5, C);
_DEFPIN_AVR(28, 1<<6, C); _DEFPIN_AVR(29, 1<<7, C); _DEFPIN_AVR(30, 1<<4, D); _DEFPIN_AVR(31, 1<<7, D);
#define SPI_DATA 11
#define SPI_CLOCK 13
#define SPI_SELECT 10
#define AVR_HARDWARE_SPI 1
#define HAS_HARDWARE_PIN_SUPPORT 1
#elif defined(__AVR_ATmega1280__) || defined(__AVR_ATmega2560__)
// megas
_IO(A); _IO(B); _IO(C); _IO(D); _IO(E); _IO(F); _IO(G); _IO(H); _IO(J); _IO(K); _IO(L);
#define MAX_PIN 69
_DEFPIN_AVR(0, 1, E); _DEFPIN_AVR(1, 2, E); _DEFPIN_AVR(2, 16, E); _DEFPIN_AVR(3, 32, E);
_DEFPIN_AVR(4, 32, G); _DEFPIN_AVR(5, 8, E); _DEFPIN_AVR(6, 8, H); _DEFPIN_AVR(7, 16, H);
_DEFPIN_AVR(8, 32, H); _DEFPIN_AVR(9, 64, H); _DEFPIN_AVR(10, 16, B); _DEFPIN_AVR(11, 32, B);
_DEFPIN_AVR(12, 64, B); _DEFPIN_AVR(13, 128, B); _DEFPIN_AVR(14, 2, J); _DEFPIN_AVR(15, 1, J);
_DEFPIN_AVR(16, 2, H); _DEFPIN_AVR(17, 1, H); _DEFPIN_AVR(18, 8, D); _DEFPIN_AVR(19, 4, D);
_DEFPIN_AVR(20, 2, D); _DEFPIN_AVR(21, 1, D); _DEFPIN_AVR(22, 1, A); _DEFPIN_AVR(23, 2, A);
_DEFPIN_AVR(24, 4, A); _DEFPIN_AVR(25, 8, A); _DEFPIN_AVR(26, 16, A); _DEFPIN_AVR(27, 32, A);
_DEFPIN_AVR(28, 64, A); _DEFPIN_AVR(29, 128, A); _DEFPIN_AVR(30, 128, C); _DEFPIN_AVR(31, 64, C);
_DEFPIN_AVR(32, 32, C); _DEFPIN_AVR(33, 16, C); _DEFPIN_AVR(34, 8, C); _DEFPIN_AVR(35, 4, C);
_DEFPIN_AVR(36, 2, C); _DEFPIN_AVR(37, 1, C); _DEFPIN_AVR(38, 128, D); _DEFPIN_AVR(39, 4, G);
_DEFPIN_AVR(40, 2, G); _DEFPIN_AVR(41, 1, G); _DEFPIN_AVR(42, 128, L); _DEFPIN_AVR(43, 64, L);
_DEFPIN_AVR(44, 32, L); _DEFPIN_AVR(45, 16, L); _DEFPIN_AVR(46, 8, L); _DEFPIN_AVR(47, 4, L);
_DEFPIN_AVR(48, 2, L); _DEFPIN_AVR(49, 1, L); _DEFPIN_AVR(50, 8, B); _DEFPIN_AVR(51, 4, B);
_DEFPIN_AVR(52, 2, B); _DEFPIN_AVR(53, 1, B); _DEFPIN_AVR(54, 1, F); _DEFPIN_AVR(55, 2, F);
_DEFPIN_AVR(56, 4, F); _DEFPIN_AVR(57, 8, F); _DEFPIN_AVR(58, 16, F); _DEFPIN_AVR(59, 32, F);
_DEFPIN_AVR(60, 64, F); _DEFPIN_AVR(61, 128, F); _DEFPIN_AVR(62, 1, K); _DEFPIN_AVR(63, 2, K);
_DEFPIN_AVR(64, 4, K); _DEFPIN_AVR(65, 8, K); _DEFPIN_AVR(66, 16, K); _DEFPIN_AVR(67, 32, K);
_DEFPIN_AVR(68, 64, K); _DEFPIN_AVR(69, 128, K);
#define SPI_DATA 51
#define SPI_CLOCK 52
#define SPI_SELECT 53
#define AVR_HARDWARE_SPI 1
#define HAS_HARDWARE_PIN_SUPPORT 1
// Leonardo, teensy, blinkm
#elif defined(__AVR_ATmega32U4__) && defined(CORE_TEENSY)
// teensy defs
_IO(B); _IO(C); _IO(D); _IO(E); _IO(F);
#define MAX_PIN 23
_DEFPIN_AVR(0, 1, B); _DEFPIN_AVR(1, 2, B); _DEFPIN_AVR(2, 4, B); _DEFPIN_AVR(3, 8, B);
_DEFPIN_AVR(4, 128, B); _DEFPIN_AVR(5, 1, D); _DEFPIN_AVR(6, 2, D); _DEFPIN_AVR(7, 4, D);
_DEFPIN_AVR(8, 8, D); _DEFPIN_AVR(9, 64, C); _DEFPIN_AVR(10, 128, C); _DEFPIN_AVR(11, 64, D);
_DEFPIN_AVR(12, 128, D); _DEFPIN_AVR(13, 16, B); _DEFPIN_AVR(14, 32, B); _DEFPIN_AVR(15, 64, B);
_DEFPIN_AVR(16, 128, F); _DEFPIN_AVR(17, 64, F); _DEFPIN_AVR(18, 32, F); _DEFPIN_AVR(19, 16, F);
_DEFPIN_AVR(20, 2, F); _DEFPIN_AVR(21, 1, F); _DEFPIN_AVR(22, 16, D); _DEFPIN_AVR(23, 32, D);
#define SPI_DATA 2
#define SPI_CLOCK 1
#define SPI_SELECT 0
#define AVR_HARDWARE_SPI 1
#define HAS_HARDWARE_PIN_SUPPORT 1
// PD3/PD5
#define SPI_UART1_DATA 8
#define SPI_UART1_CLOCK 23
#elif defined(__AVR_AT90USB646__) || defined(__AVR_AT90USB1286__)
// teensy++ 2 defs
_IO(A); _IO(B); _IO(C); _IO(D); _IO(E); _IO(F);
#define MAX_PIN 45
_DEFPIN_AVR(0, 1, D); _DEFPIN_AVR(1, 2, D); _DEFPIN_AVR(2, 4, D); _DEFPIN_AVR(3, 8, D);
_DEFPIN_AVR(4, 16, D); _DEFPIN_AVR(5, 32, D); _DEFPIN_AVR(6, 64, D); _DEFPIN_AVR(7, 128, D);
_DEFPIN_AVR(8, 1, E); _DEFPIN_AVR(9, 2, E); _DEFPIN_AVR(10, 1, C); _DEFPIN_AVR(11, 2, C);
_DEFPIN_AVR(12, 4, C); _DEFPIN_AVR(13, 8, C); _DEFPIN_AVR(14, 16, C); _DEFPIN_AVR(15, 32, C);
_DEFPIN_AVR(16, 64, C); _DEFPIN_AVR(17, 128, C); _DEFPIN_AVR(18, 64, E); _DEFPIN_AVR(19, 128, E);
_DEFPIN_AVR(20, 1, B); _DEFPIN_AVR(21, 2, B); _DEFPIN_AVR(22, 4, B); _DEFPIN_AVR(23, 8, B);
_DEFPIN_AVR(24, 16, B); _DEFPIN_AVR(25, 32, B); _DEFPIN_AVR(26, 64, B); _DEFPIN_AVR(27, 128, B);
_DEFPIN_AVR(28, 1, A); _DEFPIN_AVR(29, 2, A); _DEFPIN_AVR(30, 4, A); _DEFPIN_AVR(31, 8, A);
_DEFPIN_AVR(32, 16, A); _DEFPIN_AVR(33, 32, A); _DEFPIN_AVR(34, 64, A); _DEFPIN_AVR(35, 128, A);
_DEFPIN_AVR(36, 16, E); _DEFPIN_AVR(37, 32, E); _DEFPIN_AVR(38, 1, F); _DEFPIN_AVR(39, 2, F);
_DEFPIN_AVR(40, 4, F); _DEFPIN_AVR(41, 8, F); _DEFPIN_AVR(42, 16, F); _DEFPIN_AVR(43, 32, F);
_DEFPIN_AVR(44, 64, F); _DEFPIN_AVR(45, 128, F);
#define SPI_DATA 22
#define SPI_CLOCK 21
#define SPI_SELECT 20
#define AVR_HARDWARE_SPI 1
#define HAS_HARDWARE_PIN_SUPPORT 1
// PD3/PD5
#define SPI_UART1_DATA 3
#define SPI_UART1_CLOCK 5
#elif defined(__AVR_ATmega32U4__)
// leonard defs
_IO(B); _IO(C); _IO(D); _IO(E); _IO(F);
#define MAX_PIN 23
_DEFPIN_AVR(0, 4, D); _DEFPIN_AVR(1, 8, D); _DEFPIN_AVR(2, 2, D); _DEFPIN_AVR(3, 1, D);
_DEFPIN_AVR(4, 16, D); _DEFPIN_AVR(5, 64, C); _DEFPIN_AVR(6, 128, D); _DEFPIN_AVR(7, 64, E);
_DEFPIN_AVR(8, 16, B); _DEFPIN_AVR(9, 32, B); _DEFPIN_AVR(10, 64, B); _DEFPIN_AVR(11, 128, B);
_DEFPIN_AVR(12, 64, D); _DEFPIN_AVR(13, 128, C); _DEFPIN_AVR(14, 8, B); _DEFPIN_AVR(15, 2, B);
_DEFPIN_AVR(16, 4, B); _DEFPIN_AVR(17, 1, B); _DEFPIN_AVR(18, 128, F); _DEFPIN_AVR(19, 64, F);
_DEFPIN_AVR(20, 32, F); _DEFPIN_AVR(21, 16, F); _DEFPIN_AVR(22, 2, F); _DEFPIN_AVR(23, 1, F);
#define SPI_DATA 16
#define SPI_CLOCK 15
#define AVR_HARDWARE_SPI 1
#define HAS_HARDWARE_PIN_SUPPORT 1
// PD3/PD5
// #define SPI_UART1_DATA 1
// #define SPI_UART1_CLOCK 4
#endif
#endif // FASTLED_FORCE_SOFTWARE_PINS
FASTLED_NAMESPACE_END
#endif // __INC_FASTPIN_AVR_H
+505
View File
@@ -0,0 +1,505 @@
#ifndef __INC_FASTSPI_AVR_H
#define __INC_FASTSPI_AVR_H
FASTLED_NAMESPACE_BEGIN
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
//
// Hardware SPI support using USART registers and friends
//
// TODO: Complete/test implementation - right now this doesn't work
//
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
// uno/mini/duemilanove
#if defined(AVR_HARDWARE_SPI)
#if defined(UBRR1)
#ifndef UCPHA1
#define UCPHA1 1
#endif
template <uint8_t _DATA_PIN, uint8_t _CLOCK_PIN, uint8_t _SPI_CLOCK_DIVIDER>
class AVRUSART1SPIOutput {
Selectable *m_pSelect;
public:
AVRUSART1SPIOutput() { m_pSelect = NULL; }
AVRUSART1SPIOutput(Selectable *pSelect) { m_pSelect = pSelect; }
void setSelect(Selectable *pSelect) { m_pSelect = pSelect; }
void init() {
UBRR1 = 0;
/* Set MSPI mode of operation and SPI data mode 0. */
UCSR1C = (1<<UMSEL11)|(1<<UMSEL10)|(0<<UCPHA1)|(0<<UCPOL1);
/* Enable receiver and transmitter. */
UCSR1B = (1<<RXEN1)|(1<<TXEN1);
FastPin<_CLOCK_PIN>::setOutput();
FastPin<_DATA_PIN>::setOutput();
// must be done last, see page 206
setSPIRate();
}
void setSPIRate() {
if(_SPI_CLOCK_DIVIDER > 2) {
UBRR1 = (_SPI_CLOCK_DIVIDER/2)-1;
} else {
UBRR1 = 0;
}
}
static void stop() {
// TODO: stop the uart spi output
}
static bool shouldWait(bool wait = false) __attribute__((always_inline)) {
static bool sWait=false;
if(sWait) {
sWait = wait; return true;
} else {
sWait = wait; return false;
}
// return true;
}
static void wait() __attribute__((always_inline)) {
if(shouldWait()) {
while(!(UCSR1A & (1<<UDRE1)));
}
}
static void waitFully() __attribute__((always_inline)) { wait(); }
static void writeWord(uint16_t w) __attribute__((always_inline)) { writeByte(w>>8); writeByte(w&0xFF); }
static void writeByte(uint8_t b) __attribute__((always_inline)) { wait(); UDR1=b; shouldWait(true); }
static void writeBytePostWait(uint8_t b) __attribute__((always_inline)) { UDR1=b; shouldWait(true); wait(); }
static void writeByteNoWait(uint8_t b) __attribute__((always_inline)) { UDR1=b; shouldWait(true); }
template <uint8_t BIT> inline static void writeBit(uint8_t b) {
if(b && (1 << BIT)) {
FastPin<_DATA_PIN>::hi();
} else {
FastPin<_DATA_PIN>::lo();
}
FastPin<_CLOCK_PIN>::hi();
FastPin<_CLOCK_PIN>::lo();
}
void enable_pins() { }
void disable_pins() { }
void select() {
if(m_pSelect != NULL) {
m_pSelect->select();
}
enable_pins();
setSPIRate();
}
void release() {
if(m_pSelect != NULL) {
m_pSelect->release();
}
disable_pins();
}
static void writeBytesValueRaw(uint8_t value, int len) {
while(len--) {
writeByte(value);
}
}
void writeBytesValue(uint8_t value, int len) {
//setSPIRate();
select();
while(len--) {
writeByte(value);
}
release();
}
// Write a block of n uint8_ts out
template <class D> void writeBytes(register uint8_t *data, int len) {
//setSPIRate();
uint8_t *end = data + len;
select();
while(data != end) {
// a slight touch of delay here helps optimize the timing of the status register check loop (not used on ARM)
writeByte(D::adjust(*data++)); delaycycles<3>();
}
release();
}
void writeBytes(register uint8_t *data, int len) { writeBytes<DATA_NOP>(data, len); }
// write a block of uint8_ts out in groups of three. len is the total number of uint8_ts to write out. The template
// parameters indicate how many uint8_ts to skip at the beginning and/or end of each grouping
template <uint8_t FLAGS, class D, EOrder RGB_ORDER> void writePixels(PixelController<RGB_ORDER> pixels) {
//setSPIRate();
int len = pixels.mLen;
select();
while(pixels.has(1)) {
if(FLAGS & FLAG_START_BIT) {
writeBit<0>(1);
writeBytePostWait(D::adjust(pixels.loadAndScale0()));
writeBytePostWait(D::adjust(pixels.loadAndScale1()));
writeBytePostWait(D::adjust(pixels.loadAndScale2()));
} else {
writeByte(D::adjust(pixels.loadAndScale0()));
writeByte(D::adjust(pixels.loadAndScale1()));
writeByte(D::adjust(pixels.loadAndScale2()));
}
pixels.advanceData();
pixels.stepDithering();
}
D::postBlock(len);
release();
}
};
#endif
#if defined(UBRR0)
template <uint8_t _DATA_PIN, uint8_t _CLOCK_PIN, uint8_t _SPI_CLOCK_DIVIDER>
class AVRUSART0SPIOutput {
Selectable *m_pSelect;
public:
AVRUSART0SPIOutput() { m_pSelect = NULL; }
AVRUSART0SPIOutput(Selectable *pSelect) { m_pSelect = pSelect; }
void setSelect(Selectable *pSelect) { m_pSelect = pSelect; }
void init() {
UBRR0 = 0;
/* Set MSPI mode of operation and SPI data mode 0. */
UCSR0C = (1<<UMSEL01)|(1<<UMSEL00)/*|(0<<UCPHA0)*/|(0<<UCPOL0);
/* Enable receiver and transmitter. */
UCSR0B = (1<<RXEN0)|(1<<TXEN0);
FastPin<_CLOCK_PIN>::setOutput();
FastPin<_DATA_PIN>::setOutput();
// must be done last, see page 206
setSPIRate();
}
void setSPIRate() {
if(_SPI_CLOCK_DIVIDER > 2) {
UBRR0 = (_SPI_CLOCK_DIVIDER/2)-1;
} else {
UBRR0 = 0;
}
}
static void stop() {
// TODO: stop the uart spi output
}
static bool shouldWait(bool wait = false) __attribute__((always_inline)) {
static bool sWait=false;
if(sWait) {
sWait = wait; return true;
} else {
sWait = wait; return false;
}
// return true;
}
static void wait() __attribute__((always_inline)) {
if(shouldWait()) {
while(!(UCSR0A & (1<<UDRE0)));
}
}
static void waitFully() __attribute__((always_inline)) { wait(); }
static void writeWord(uint16_t w) __attribute__((always_inline)) { writeByte(w>>8); writeByte(w&0xFF); }
static void writeByte(uint8_t b) __attribute__((always_inline)) { wait(); UDR0=b; shouldWait(true); }
static void writeBytePostWait(uint8_t b) __attribute__((always_inline)) { UDR0=b; shouldWait(true); wait(); }
static void writeByteNoWait(uint8_t b) __attribute__((always_inline)) { UDR0=b; shouldWait(true); }
template <uint8_t BIT> inline static void writeBit(uint8_t b) {
if(b && (1 << BIT)) {
FastPin<_DATA_PIN>::hi();
} else {
FastPin<_DATA_PIN>::lo();
}
FastPin<_CLOCK_PIN>::hi();
FastPin<_CLOCK_PIN>::lo();
}
void enable_pins() { }
void disable_pins() { }
void select() {
if(m_pSelect != NULL) {
m_pSelect->select();
}
enable_pins();
setSPIRate();
}
void release() {
if(m_pSelect != NULL) {
m_pSelect->release();
}
disable_pins();
}
static void writeBytesValueRaw(uint8_t value, int len) {
while(len--) {
writeByte(value);
}
}
void writeBytesValue(uint8_t value, int len) {
//setSPIRate();
select();
while(len--) {
writeByte(value);
}
release();
}
// Write a block of n uint8_ts out
template <class D> void writeBytes(register uint8_t *data, int len) {
//setSPIRate();
uint8_t *end = data + len;
select();
while(data != end) {
// a slight touch of delay here helps optimize the timing of the status register check loop (not used on ARM)
writeByte(D::adjust(*data++)); delaycycles<3>();
}
release();
}
void writeBytes(register uint8_t *data, int len) { writeBytes<DATA_NOP>(data, len); }
// write a block of uint8_ts out in groups of three. len is the total number of uint8_ts to write out. The template
// parameters indicate how many uint8_ts to skip at the beginning and/or end of each grouping
template <uint8_t FLAGS, class D, EOrder RGB_ORDER> void writePixels(PixelController<RGB_ORDER> pixels) {
//setSPIRate();
int len = pixels.mLen;
select();
while(pixels.has(1)) {
if(FLAGS & FLAG_START_BIT) {
writeBit<0>(1);
writeBytePostWait(D::adjust(pixels.loadAndScale0()));
writeBytePostWait(D::adjust(pixels.loadAndScale1()));
writeBytePostWait(D::adjust(pixels.loadAndScale2()));
} else {
writeByte(D::adjust(pixels.loadAndScale0()));
writeByte(D::adjust(pixels.loadAndScale1()));
writeByte(D::adjust(pixels.loadAndScale2()));
}
pixels.advanceData();
pixels.stepDithering();
}
D::postBlock(len);
waitFully();
release();
}
};
#endif
#if defined(SPSR)
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
//
// Hardware SPI support using SPDR registers and friends
//
// Technically speaking, this uses the AVR SPI registers. This will work on the Teensy 3.0 because Paul made a set of compatability
// classes that map the AVR SPI registers to ARM's, however this caps the performance of output.
//
// TODO: implement ARMHardwareSPIOutput
//
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
template <uint8_t _DATA_PIN, uint8_t _CLOCK_PIN, uint8_t _SPI_CLOCK_DIVIDER>
class AVRHardwareSPIOutput {
Selectable *m_pSelect;
bool mWait;
public:
AVRHardwareSPIOutput() { m_pSelect = NULL; mWait = false;}
AVRHardwareSPIOutput(Selectable *pSelect) { m_pSelect = pSelect; }
void setSelect(Selectable *pSelect) { m_pSelect = pSelect; }
void setSPIRate() {
SPCR &= ~ ( (1<<SPR1) | (1<<SPR0) ); // clear out the prescalar bits
bool b2x = false;
if(_SPI_CLOCK_DIVIDER >= 128) { SPCR |= (1<<SPR1); SPCR |= (1<<SPR0); }
else if(_SPI_CLOCK_DIVIDER >= 64) { SPCR |= (1<<SPR1);}
else if(_SPI_CLOCK_DIVIDER >= 32) { SPCR |= (1<<SPR1); b2x = true; }
else if(_SPI_CLOCK_DIVIDER >= 16) { SPCR |= (1<<SPR0); }
else if(_SPI_CLOCK_DIVIDER >= 8) { SPCR |= (1<<SPR0); b2x = true; }
else if(_SPI_CLOCK_DIVIDER >= 4) { /* do nothing - default rate */ }
else { b2x = true; }
if(b2x) { SPSR |= (1<<SPI2X); }
else { SPSR &= ~ (1<<SPI2X); }
}
void init() {
volatile uint8_t clr;
// set the pins to output
FastPin<_DATA_PIN>::setOutput();
FastPin<_CLOCK_PIN>::setOutput();
#ifdef SPI_SELECT
// Make sure the slave select line is set to output, or arduino will block us
FastPin<SPI_SELECT>::setOutput();
FastPin<SPI_SELECT>::lo();
#endif
SPCR |= ((1<<SPE) | (1<<MSTR) ); // enable SPI as master
SPCR &= ~ ( (1<<SPR1) | (1<<SPR0) ); // clear out the prescalar bits
clr = SPSR; // clear SPI status register
clr = SPDR; // clear SPI data register
clr;
bool b2x = false;
if(_SPI_CLOCK_DIVIDER >= 128) { SPCR |= (1<<SPR1); SPCR |= (1<<SPR0); }
else if(_SPI_CLOCK_DIVIDER >= 64) { SPCR |= (1<<SPR1);}
else if(_SPI_CLOCK_DIVIDER >= 32) { SPCR |= (1<<SPR1); b2x = true; }
else if(_SPI_CLOCK_DIVIDER >= 16) { SPCR |= (1<<SPR0); }
else if(_SPI_CLOCK_DIVIDER >= 8) { SPCR |= (1<<SPR0); b2x = true; }
else if(_SPI_CLOCK_DIVIDER >= 4) { /* do nothing - default rate */ }
else { b2x = true; }
if(b2x) { SPSR |= (1<<SPI2X); }
else { SPSR &= ~ (1<<SPI2X); }
SPDR=0;
shouldWait(false);
release();
}
static bool shouldWait(bool wait = false) __attribute__((always_inline)) {
static bool sWait=false;
if(sWait) { sWait = wait; return true; } else { sWait = wait; return false; }
// return true;
}
static void wait() __attribute__((always_inline)) { if(shouldWait()) { while(!(SPSR & (1<<SPIF))); } }
static void waitFully() __attribute__((always_inline)) { wait(); }
static void writeWord(uint16_t w) __attribute__((always_inline)) { writeByte(w>>8); writeByte(w&0xFF); }
static void writeByte(uint8_t b) __attribute__((always_inline)) { wait(); SPDR=b; shouldWait(true); }
static void writeBytePostWait(uint8_t b) __attribute__((always_inline)) { SPDR=b; shouldWait(true); wait(); }
static void writeByteNoWait(uint8_t b) __attribute__((always_inline)) { SPDR=b; shouldWait(true); }
template <uint8_t BIT> inline static void writeBit(uint8_t b) {
SPCR &= ~(1 << SPE);
if(b & (1 << BIT)) {
FastPin<_DATA_PIN>::hi();
} else {
FastPin<_DATA_PIN>::lo();
}
FastPin<_CLOCK_PIN>::hi();
FastPin<_CLOCK_PIN>::lo();
SPCR |= 1 << SPE;
shouldWait(false);
}
void enable_pins() {
SPCR |= ((1<<SPE) | (1<<MSTR) ); // enable SPI as master
}
void disable_pins() {
SPCR &= ~(((1<<SPE) | (1<<MSTR) )); // disable SPI
}
void select() {
if(m_pSelect != NULL) { m_pSelect->select(); }
enable_pins();
setSPIRate();
}
void release() {
if(m_pSelect != NULL) { m_pSelect->release(); }
disable_pins();
}
static void writeBytesValueRaw(uint8_t value, int len) {
while(len--) { writeByte(value); }
}
void writeBytesValue(uint8_t value, int len) {
//setSPIRate();
select();
while(len--) {
writeByte(value);
}
release();
}
// Write a block of n uint8_ts out
template <class D> void writeBytes(register uint8_t *data, int len) {
//setSPIRate();
uint8_t *end = data + len;
select();
while(data != end) {
// a slight touch of delay here helps optimize the timing of the status register check loop (not used on ARM)
writeByte(D::adjust(*data++)); delaycycles<3>();
}
release();
}
void writeBytes(register uint8_t *data, int len) { writeBytes<DATA_NOP>(data, len); }
// write a block of uint8_ts out in groups of three. len is the total number of uint8_ts to write out. The template
// parameters indicate how many uint8_ts to skip at the beginning and/or end of each grouping
template <uint8_t FLAGS, class D, EOrder RGB_ORDER> void writePixels(PixelController<RGB_ORDER> pixels) {
//setSPIRate();
int len = pixels.mLen;
select();
while(pixels.has(1)) {
if(FLAGS & FLAG_START_BIT) {
writeBit<0>(1);
writeBytePostWait(D::adjust(pixels.loadAndScale0()));
writeBytePostWait(D::adjust(pixels.loadAndScale1()));
writeBytePostWait(D::adjust(pixels.loadAndScale2()));
} else {
writeByte(D::adjust(pixels.loadAndScale0()));
writeByte(D::adjust(pixels.loadAndScale1()));
writeByte(D::adjust(pixels.loadAndScale2()));
}
pixels.advanceData();
pixels.stepDithering();
}
D::postBlock(len);
waitFully();
release();
}
};
#endif
#else
// #define FASTLED_FORCE_SOFTWARE_SPI
#endif
FASTLED_NAMESPACE_END;
#endif
@@ -0,0 +1,52 @@
#ifndef __INC_LED_SYSDEFS_AVR_H
#define __INC_LED_SYSDEFS_AVR_H
#define FASTLED_AVR
#ifndef INTERRUPT_THRESHOLD
#define INTERRUPT_THRESHOLD 2
#endif
#include <avr/io.h>
#include <avr/interrupt.h> // for cli/se definitions
// Define the register types
#if defined(ARDUINO) // && ARDUINO < 150
typedef volatile uint8_t RoReg; /**< Read only 8-bit register (volatile const unsigned int) */
typedef volatile uint8_t RwReg; /**< Read-Write 8-bit register (volatile unsigned int) */
#endif
// Default to disallowing interrupts (may want to gate this on teensy2 vs. other arm platforms, since the
// teensy2 has a good, fast millis interrupt implementation)
#ifndef FASTLED_ALLOW_INTERRUPTS
#define FASTLED_ALLOW_INTERRUPTS 0
#endif
#if FASTLED_ALLOW_INTERRUPTS == 1
#define FASTLED_ACCURATE_CLOCK
#endif
// Default to using PROGMEM here
#ifndef FASTLED_USE_PROGMEM
#define FASTLED_USE_PROGMEM 1
#endif
#if defined(ARDUINO_AVR_DIGISPARK) || defined(ARDUINO_AVR_DIGISPARKPRO)
#ifndef NO_CORRECTION
#define NO_CORRECTION 1
#endif
#endif
extern "C" {
# if defined(CORE_TEENSY) || defined(TEENSYDUINO)
extern volatile unsigned long timer0_millis_count;
# define MS_COUNTER timer0_millis_count
# else
extern volatile unsigned long timer0_millis;
# define MS_COUNTER timer0_millis
# endif
};
#endif