1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
|
#ifndef __INC_M0_CLOCKLESS_H
#define __INC_M0_CLOCKLESS_H
struct M0ClocklessData {
uint8_t d[3];
uint8_t e[3];
uint8_t adj;
uint8_t pad;
uint32_t s[3];
};
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;
#if (FASTLED_SCALE8_FIXED == 1)
pData->s[0]++;
pData->s[1]++;
pData->s[2]++;
#endif
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;"
" ldr \\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+6 has what we're incrementing by)
" .macro incleds3 leds, base, scratch;"
" ldrb \\scratch, [\\base, #6];" // 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)), \
[e0] "I" (3+RO(0)), \
[e1] "I" (3+RO(1)), \
[e2] "I" (3+RO(2)), \
[scale0] "I" (4*(2+RO(0))), \
[scale1] "I" (4*(2+RO(1))), \
[scale2] "I" (4*(2+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 NOTHING D2(0) LO2 D3(0)
HI2 D1 QLO4 SWAPBBN1 D2(1) LO2 D3(0)
// Write out byte 2, prepping byte 0
HI2 D1 QLO4 INCLEDS3 D2(3) 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
|
