lib8tion.c
6439 bytes
1#define FASTLED_INTERNAL
2#include <stdint.h>
3
4#define RAND16_SEED 1337
5uint16_t rand16seed = RAND16_SEED;
6
7
8// memset8, memcpy8, memmove8:
9// optimized avr replacements for the standard "C" library
10// routines memset, memcpy, and memmove.
11//
12// There are two techniques that make these routines
13// faster than the standard avr-libc routines.
14// First, the loops are unrolled 2X, meaning that
15// the average loop overhead is cut in half.
16// And second, the compare-and-branch at the bottom
17// of each loop decrements the low byte of the
18// counter, and if the carry is clear, it branches
19// back up immediately. Only if the low byte math
20// causes carry do we bother to decrement the high
21// byte and check that result for carry as well.
22// Results for a 100-byte buffer are 20-40% faster
23// than standard avr-libc, at a cost of a few extra
24// bytes of code.
25
26#if defined(__AVR__)
27//__attribute__ ((noinline))
28void * memset8 ( void * ptr, uint8_t val, uint16_t num )
29{
30 asm volatile(
31 " movw r26, %[ptr] \n\t"
32 " sbrs %A[num], 0 \n\t"
33 " rjmp Lseteven_%= \n\t"
34 " rjmp Lsetodd_%= \n\t"
35 "Lsetloop_%=: \n\t"
36 " st X+, %[val] \n\t"
37 "Lsetodd_%=: \n\t"
38 " st X+, %[val] \n\t"
39 "Lseteven_%=: \n\t"
40 " subi %A[num], 2 \n\t"
41 " brcc Lsetloop_%= \n\t"
42 " sbci %B[num], 0 \n\t"
43 " brcc Lsetloop_%= \n\t"
44 : [num] "+r" (num)
45 : [ptr] "r" (ptr),
46 [val] "r" (val)
47 : "memory"
48 );
49 return ptr;
50}
51
52
53
54//__attribute__ ((noinline))
55void * memcpy8 ( void * dst, const void* src, uint16_t num )
56{
57 asm volatile(
58 " movw r30, %[src] \n\t"
59 " movw r26, %[dst] \n\t"
60 " sbrs %A[num], 0 \n\t"
61 " rjmp Lcpyeven_%= \n\t"
62 " rjmp Lcpyodd_%= \n\t"
63 "Lcpyloop_%=: \n\t"
64 " ld __tmp_reg__, Z+ \n\t"
65 " st X+, __tmp_reg__ \n\t"
66 "Lcpyodd_%=: \n\t"
67 " ld __tmp_reg__, Z+ \n\t"
68 " st X+, __tmp_reg__ \n\t"
69 "Lcpyeven_%=: \n\t"
70 " subi %A[num], 2 \n\t"
71 " brcc Lcpyloop_%= \n\t"
72 " sbci %B[num], 0 \n\t"
73 " brcc Lcpyloop_%= \n\t"
74 : [num] "+r" (num)
75 : [src] "r" (src),
76 [dst] "r" (dst)
77 : "memory"
78 );
79 return dst;
80}
81
82//__attribute__ ((noinline))
83void * memmove8 ( void * dst, const void* src, uint16_t num )
84{
85 if( src > dst) {
86 // if src > dst then we can use the forward-stepping memcpy8
87 return memcpy8( dst, src, num);
88 } else {
89 // if src < dst then we have to step backward:
90 dst = (char*)dst + num;
91 src = (char*)src + num;
92 asm volatile(
93 " movw r30, %[src] \n\t"
94 " movw r26, %[dst] \n\t"
95 " sbrs %A[num], 0 \n\t"
96 " rjmp Lmoveven_%= \n\t"
97 " rjmp Lmovodd_%= \n\t"
98 "Lmovloop_%=: \n\t"
99 " ld __tmp_reg__, -Z \n\t"
100 " st -X, __tmp_reg__ \n\t"
101 "Lmovodd_%=: \n\t"
102 " ld __tmp_reg__, -Z \n\t"
103 " st -X, __tmp_reg__ \n\t"
104 "Lmoveven_%=: \n\t"
105 " subi %A[num], 2 \n\t"
106 " brcc Lmovloop_%= \n\t"
107 " sbci %B[num], 0 \n\t"
108 " brcc Lmovloop_%= \n\t"
109 : [num] "+r" (num)
110 : [src] "r" (src),
111 [dst] "r" (dst)
112 : "memory"
113 );
114 return dst;
115 }
116}
117
118#endif /* AVR */
119
120
121
122
123#if 0
124// TEST / VERIFICATION CODE ONLY BELOW THIS POINT
125#include <Arduino.h>
126#include "lib8tion.h"
127
128void test1abs( int8_t i)
129{
130 Serial.print("abs("); Serial.print(i); Serial.print(") = ");
131 int8_t j = abs8(i);
132 Serial.print(j); Serial.println(" ");
133}
134
135void testabs()
136{
137 delay(5000);
138 for( int8_t q = -128; q != 127; q++) {
139 test1abs(q);
140 }
141 for(;;){};
142}
143
144
145void testmul8()
146{
147 delay(5000);
148 byte r, c;
149
150 Serial.println("mul8:");
151 for( r = 0; r <= 20; r += 1) {
152 Serial.print(r); Serial.print(" : ");
153 for( c = 0; c <= 20; c += 1) {
154 byte t;
155 t = mul8( r, c);
156 Serial.print(t); Serial.print(' ');
157 }
158 Serial.println(' ');
159 }
160 Serial.println("done.");
161 for(;;){};
162}
163
164
165void testscale8()
166{
167 delay(5000);
168 byte r, c;
169
170 Serial.println("scale8:");
171 for( r = 0; r <= 240; r += 10) {
172 Serial.print(r); Serial.print(" : ");
173 for( c = 0; c <= 240; c += 10) {
174 byte t;
175 t = scale8( r, c);
176 Serial.print(t); Serial.print(' ');
177 }
178 Serial.println(' ');
179 }
180
181 Serial.println(' ');
182 Serial.println("scale8_video:");
183
184 for( r = 0; r <= 100; r += 4) {
185 Serial.print(r); Serial.print(" : ");
186 for( c = 0; c <= 100; c += 4) {
187 byte t;
188 t = scale8_video( r, c);
189 Serial.print(t); Serial.print(' ');
190 }
191 Serial.println(' ');
192 }
193
194 Serial.println("done.");
195 for(;;){};
196}
197
198
199
200void testqadd8()
201{
202 delay(5000);
203 byte r, c;
204 for( r = 0; r <= 240; r += 10) {
205 Serial.print(r); Serial.print(" : ");
206 for( c = 0; c <= 240; c += 10) {
207 byte t;
208 t = qadd8( r, c);
209 Serial.print(t); Serial.print(' ');
210 }
211 Serial.println(' ');
212 }
213 Serial.println("done.");
214 for(;;){};
215}
216
217void testnscale8x3()
218{
219 delay(5000);
220 byte r, g, b, sc;
221 for( byte z = 0; z < 10; z++) {
222 r = random8(); g = random8(); b = random8(); sc = random8();
223
224 Serial.print("nscale8x3_video( ");
225 Serial.print(r); Serial.print(", ");
226 Serial.print(g); Serial.print(", ");
227 Serial.print(b); Serial.print(", ");
228 Serial.print(sc); Serial.print(") = [ ");
229
230 nscale8x3_video( r, g, b, sc);
231
232 Serial.print(r); Serial.print(", ");
233 Serial.print(g); Serial.print(", ");
234 Serial.print(b); Serial.print("]");
235
236 Serial.println(' ');
237 }
238 Serial.println("done.");
239 for(;;){};
240}
241
242#endif