| 1 | #ifndef _OP_LIGHT_ARM_H_ |
| 2 | #define _OP_LIGHT_ARM_H_ |
| 3 | |
| 4 | //////////////////////////////////////////////////////////////////////////////// |
| 5 | // Extract bgr555 color from Gouraud u32 fixed-pt 8.3:8.3:8.2 rgb triplet |
| 6 | // |
| 7 | // INPUT: |
| 8 | // 'gCol' input: rrrrrrrrXXXggggggggXXXbbbbbbbbXX |
| 9 | // ^ bit 31 |
| 10 | // RETURNS: |
| 11 | // u16 output: 0bbbbbgggggrrrrr |
| 12 | // ^ bit 16 |
| 13 | // Where 'r,g,b' are integer bits of colors, 'X' fixed-pt, and '0' zero |
| 14 | //////////////////////////////////////////////////////////////////////////////// |
| 15 | GPU_INLINE uint_fast16_t gpuLightingRGBARM(u32 gCol) |
| 16 | { |
| 17 | uint_fast16_t out = 0x03E0; // don't need the mask after starting to write output |
| 18 | u32 tmp; |
| 19 | |
| 20 | asm ("and %[tmp], %[gCol], %[out]\n\t" // tmp holds 0x000000bbbbb00000 |
| 21 | "and %[out], %[out], %[gCol], lsr #0x0B\n\t" // out holds 0x000000ggggg00000 |
| 22 | "orr %[tmp], %[out], %[tmp], lsl #0x05\n\t" // tmp holds 0x0bbbbbggggg00000 |
| 23 | "orr %[out], %[tmp], %[gCol], lsr #0x1B\n\t" // out holds 0x0bbbbbgggggrrrrr |
| 24 | : [out] "+&r" (out), [tmp] "=&r" (tmp) |
| 25 | : [gCol] "r" (gCol) |
| 26 | ); |
| 27 | |
| 28 | return out; |
| 29 | } |
| 30 | |
| 31 | //////////////////////////////////////////////////////////////////////////////// |
| 32 | // Apply fast (low-precision) 5-bit lighting to bgr555 texture color: |
| 33 | // |
| 34 | // INPUT: |
| 35 | // 'r5','g5','b5' are unsigned 5-bit color values, value of 15 |
| 36 | // is midpoint that doesn't modify that component of texture |
| 37 | // 'uSrc' input: mbbbbbgggggrrrrr |
| 38 | // ^ bit 16 |
| 39 | // RETURNS: |
| 40 | // u16 output: mbbbbbgggggrrrrr |
| 41 | // Where 'X' are fixed-pt bits. |
| 42 | //////////////////////////////////////////////////////////////////////////////// |
| 43 | GPU_INLINE uint_fast16_t gpuLightingTXTARM(uint_fast16_t uSrc, u8 r5, u8 g5, u8 b5) |
| 44 | { |
| 45 | uint_fast16_t out = 0x03E0; |
| 46 | u32 db, dg; |
| 47 | |
| 48 | // Using `g` for src, `G` for dest |
| 49 | asm ("and %[dg], %[out], %[src] \n\t" // dg holds 0x000000ggggg00000 |
| 50 | "orr %[dg], %[dg], %[g5] \n\t" // dg holds 0x000000gggggGGGGG |
| 51 | "and %[db], %[out], %[src], lsr #0x05 \n\t" // db holds 0x000000bbbbb00000 |
| 52 | "ldrb %[dg], [%[lut], %[dg]] \n\t" // dg holds result 0x00000000000ggggg |
| 53 | "and %[out], %[out], %[src], lsl #0x05 \n\t" // out holds 0x000000rrrrr00000 |
| 54 | "orr %[out], %[out], %[r5] \n\t" // out holds 0x000000rrrrrRRRRR |
| 55 | "orr %[db], %[db], %[b5] \n\t" // db holds 0x000000bbbbbBBBBB |
| 56 | "ldrb %[out], [%[lut], %[out]] \n\t" // out holds result 0x00000000000rrrrr |
| 57 | "ldrb %[db], [%[lut], %[db]] \n\t" // db holds result 0x00000000000bbbbb |
| 58 | "tst %[src], #0x8000\n\t" // check whether msb was set on uSrc |
| 59 | "orr %[out], %[out], %[dg], lsl #0x05 \n\t" // out holds 0x000000gggggrrrrr |
| 60 | "orrne %[out], %[out], #0x8000\n\t" // add msb to out if set on uSrc |
| 61 | "orr %[out], %[out], %[db], lsl #0x0A \n\t" // out holds 0xmbbbbbgggggrrrrr |
| 62 | : [out] "=&r" (out), [db] "=&r" (db), [dg] "=&r" (dg) |
| 63 | : [r5] "r" (r5), [g5] "r" (g5), [b5] "r" (b5), |
| 64 | [lut] "r" (gpu_senquack.LightLUT), [src] "r" (uSrc), "0" (out) |
| 65 | : "cc"); |
| 66 | return out; |
| 67 | } |
| 68 | |
| 69 | //////////////////////////////////////////////////////////////////////////////// |
| 70 | // Apply fast (low-precision) 5-bit Gouraud lighting to bgr555 texture color: |
| 71 | // |
| 72 | // INPUT: |
| 73 | // 'gCol' is a packed Gouraud u32 fixed-pt 8.3:8.3:8.2 rgb triplet, value of |
| 74 | // 15.0 is midpoint that does not modify color of texture |
| 75 | // gCol input : rrrrrXXXXXXgggggXXXXXXbbbbbXXXXX |
| 76 | // ^ bit 31 |
| 77 | // 'uSrc' input: mbbbbbgggggrrrrr |
| 78 | // ^ bit 16 |
| 79 | // RETURNS: |
| 80 | // u16 output: mbbbbbgggggrrrrr |
| 81 | // Where 'X' are fixed-pt bits, '0' is zero-padding, and '-' is don't care |
| 82 | //////////////////////////////////////////////////////////////////////////////// |
| 83 | GPU_INLINE uint_fast16_t gpuLightingTXTGouraudARM(uint_fast16_t uSrc, u32 gCol) |
| 84 | { |
| 85 | uint_fast16_t out = 0x03E0; // don't need the mask after starting to write output |
| 86 | u32 db,dg,gtmp; |
| 87 | |
| 88 | // Using `g` for src, `G` for dest |
| 89 | asm ("and %[dg], %[out], %[src] \n\t" // dg holds 0x000000ggggg00000 |
| 90 | "and %[gtmp],%[out], %[gCol], lsr #0x0B \n\t" // gtmp holds 0x000000GGGGG00000 |
| 91 | "and %[db], %[out], %[src], lsr #0x05 \n\t" // db holds 0x000000bbbbb00000 |
| 92 | "orr %[dg], %[dg], %[gtmp], lsr #0x05 \n\t" // dg holds 0x000000gggggGGGGG |
| 93 | "and %[gtmp],%[out], %[gCol] \n\t" // gtmp holds 0x000000BBBBB00000 |
| 94 | "ldrb %[dg], [%[lut], %[dg]] \n\t" // dg holds result 0x00000000000ggggg |
| 95 | "and %[out], %[out], %[src], lsl #0x05 \n\t" // out holds 0x000000rrrrr00000 |
| 96 | "orr %[out], %[out], %[gCol], lsr #0x1B \n\t" // out holds 0x000000rrrrrRRRRR |
| 97 | "orr %[db], %[db], %[gtmp], lsr #0x05 \n\t" // db holds 0x000000bbbbbBBBBB |
| 98 | "ldrb %[out], [%[lut], %[out]] \n\t" // out holds result 0x00000000000rrrrr |
| 99 | "ldrb %[db], [%[lut], %[db]] \n\t" // db holds result 0x00000000000bbbbb |
| 100 | "tst %[src], #0x8000\n\t" // check whether msb was set on uSrc |
| 101 | "orr %[out], %[out], %[dg], lsl #0x05 \n\t" // out holds 0x000000gggggrrrrr |
| 102 | "orrne %[out], %[out], #0x8000\n\t" // add msb to out if set on uSrc |
| 103 | "orr %[out], %[out], %[db], lsl #0x0A \n\t" // out holds 0xmbbbbbgggggrrrrr |
| 104 | : [out] "=&r" (out), [db] "=&r" (db), [dg] "=&r" (dg), |
| 105 | [gtmp] "=&r" (gtmp) \ |
| 106 | : [gCol] "r" (gCol), [lut] "r" (gpu_senquack.LightLUT), "0" (out), [src] "r" (uSrc) |
| 107 | : "cc"); |
| 108 | |
| 109 | return out; |
| 110 | } |
| 111 | |
| 112 | #endif //_OP_LIGHT_ARM_H_ |