| 1 | // Copyright 2014 Google Inc. All Rights Reserved. |
| 2 | // |
| 3 | // Use of this source code is governed by a BSD-style license |
| 4 | // that can be found in the COPYING file in the root of the source |
| 5 | // tree. An additional intellectual property rights grant can be found |
| 6 | // in the file PATENTS. All contributing project authors may |
| 7 | // be found in the AUTHORS file in the root of the source tree. |
| 8 | // ----------------------------------------------------------------------------- |
| 9 | // |
| 10 | // MIPS DSPr2 version of YUV to RGB upsampling functions. |
| 11 | // |
| 12 | // Author(s): Branimir Vasic (branimir.vasic@imgtec.com) |
| 13 | // Djordje Pesut (djordje.pesut@imgtec.com) |
| 14 | |
| 15 | #include "src/dsp/dsp.h" |
| 16 | |
| 17 | #if defined(WEBP_USE_MIPS_DSP_R2) |
| 18 | |
| 19 | #include "src/dsp/yuv.h" |
| 20 | |
| 21 | //------------------------------------------------------------------------------ |
| 22 | // simple point-sampling |
| 23 | |
| 24 | #define ROW_FUNC_PART_1() \ |
| 25 | "lbu %[temp3], 0(%[v]) \n\t" \ |
| 26 | "lbu %[temp4], 0(%[u]) \n\t" \ |
| 27 | "lbu %[temp0], 0(%[y]) \n\t" \ |
| 28 | "mul %[temp1], %[t_con_1], %[temp3] \n\t" \ |
| 29 | "mul %[temp3], %[t_con_2], %[temp3] \n\t" \ |
| 30 | "mul %[temp2], %[t_con_3], %[temp4] \n\t" \ |
| 31 | "mul %[temp4], %[t_con_4], %[temp4] \n\t" \ |
| 32 | "mul %[temp0], %[t_con_5], %[temp0] \n\t" \ |
| 33 | "subu %[temp1], %[temp1], %[t_con_6] \n\t" \ |
| 34 | "subu %[temp3], %[temp3], %[t_con_7] \n\t" \ |
| 35 | "addu %[temp2], %[temp2], %[temp3] \n\t" \ |
| 36 | "subu %[temp4], %[temp4], %[t_con_8] \n\t" \ |
| 37 | |
| 38 | #define ROW_FUNC_PART_2(R, G, B, K) \ |
| 39 | "addu %[temp5], %[temp0], %[temp1] \n\t" \ |
| 40 | "subu %[temp6], %[temp0], %[temp2] \n\t" \ |
| 41 | "addu %[temp7], %[temp0], %[temp4] \n\t" \ |
| 42 | ".if " #K " \n\t" \ |
| 43 | "lbu %[temp0], 1(%[y]) \n\t" \ |
| 44 | ".endif \n\t" \ |
| 45 | "shll_s.w %[temp5], %[temp5], 17 \n\t" \ |
| 46 | "shll_s.w %[temp6], %[temp6], 17 \n\t" \ |
| 47 | ".if " #K " \n\t" \ |
| 48 | "mul %[temp0], %[t_con_5], %[temp0] \n\t" \ |
| 49 | ".endif \n\t" \ |
| 50 | "shll_s.w %[temp7], %[temp7], 17 \n\t" \ |
| 51 | "precrqu_s.qb.ph %[temp5], %[temp5], $zero \n\t" \ |
| 52 | "precrqu_s.qb.ph %[temp6], %[temp6], $zero \n\t" \ |
| 53 | "precrqu_s.qb.ph %[temp7], %[temp7], $zero \n\t" \ |
| 54 | "srl %[temp5], %[temp5], 24 \n\t" \ |
| 55 | "srl %[temp6], %[temp6], 24 \n\t" \ |
| 56 | "srl %[temp7], %[temp7], 24 \n\t" \ |
| 57 | "sb %[temp5], " #R "(%[dst]) \n\t" \ |
| 58 | "sb %[temp6], " #G "(%[dst]) \n\t" \ |
| 59 | "sb %[temp7], " #B "(%[dst]) \n\t" \ |
| 60 | |
| 61 | #define ASM_CLOBBER_LIST() \ |
| 62 | : [temp0]"=&r"(temp0), [temp1]"=&r"(temp1), [temp2]"=&r"(temp2), \ |
| 63 | [temp3]"=&r"(temp3), [temp4]"=&r"(temp4), [temp5]"=&r"(temp5), \ |
| 64 | [temp6]"=&r"(temp6), [temp7]"=&r"(temp7) \ |
| 65 | : [t_con_1]"r"(t_con_1), [t_con_2]"r"(t_con_2), [t_con_3]"r"(t_con_3), \ |
| 66 | [t_con_4]"r"(t_con_4), [t_con_5]"r"(t_con_5), [t_con_6]"r"(t_con_6), \ |
| 67 | [u]"r"(u), [v]"r"(v), [y]"r"(y), [dst]"r"(dst), \ |
| 68 | [t_con_7]"r"(t_con_7), [t_con_8]"r"(t_con_8) \ |
| 69 | : "memory", "hi", "lo" \ |
| 70 | |
| 71 | #define ROW_FUNC(FUNC_NAME, XSTEP, R, G, B, A) \ |
| 72 | static void FUNC_NAME(const uint8_t* y, \ |
| 73 | const uint8_t* u, const uint8_t* v, \ |
| 74 | uint8_t* dst, int len) { \ |
| 75 | int i; \ |
| 76 | uint32_t temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7; \ |
| 77 | const int t_con_1 = 26149; \ |
| 78 | const int t_con_2 = 13320; \ |
| 79 | const int t_con_3 = 6419; \ |
| 80 | const int t_con_4 = 33050; \ |
| 81 | const int t_con_5 = 19077; \ |
| 82 | const int t_con_6 = 14234; \ |
| 83 | const int t_con_7 = 8708; \ |
| 84 | const int t_con_8 = 17685; \ |
| 85 | for (i = 0; i < (len >> 1); i++) { \ |
| 86 | __asm__ volatile ( \ |
| 87 | ROW_FUNC_PART_1() \ |
| 88 | ROW_FUNC_PART_2(R, G, B, 1) \ |
| 89 | ROW_FUNC_PART_2(R + XSTEP, G + XSTEP, B + XSTEP, 0) \ |
| 90 | ASM_CLOBBER_LIST() \ |
| 91 | ); \ |
| 92 | if (A) dst[A] = dst[A + XSTEP] = 0xff; \ |
| 93 | y += 2; \ |
| 94 | ++u; \ |
| 95 | ++v; \ |
| 96 | dst += 2 * XSTEP; \ |
| 97 | } \ |
| 98 | if (len & 1) { \ |
| 99 | __asm__ volatile ( \ |
| 100 | ROW_FUNC_PART_1() \ |
| 101 | ROW_FUNC_PART_2(R, G, B, 0) \ |
| 102 | ASM_CLOBBER_LIST() \ |
| 103 | ); \ |
| 104 | if (A) dst[A] = 0xff; \ |
| 105 | } \ |
| 106 | } |
| 107 | |
| 108 | ROW_FUNC(YuvToRgbRow_MIPSdspR2, 3, 0, 1, 2, 0) |
| 109 | ROW_FUNC(YuvToRgbaRow_MIPSdspR2, 4, 0, 1, 2, 3) |
| 110 | ROW_FUNC(YuvToBgrRow_MIPSdspR2, 3, 2, 1, 0, 0) |
| 111 | ROW_FUNC(YuvToBgraRow_MIPSdspR2, 4, 2, 1, 0, 3) |
| 112 | |
| 113 | #undef ROW_FUNC |
| 114 | #undef ASM_CLOBBER_LIST |
| 115 | #undef ROW_FUNC_PART_2 |
| 116 | #undef ROW_FUNC_PART_1 |
| 117 | |
| 118 | //------------------------------------------------------------------------------ |
| 119 | // Entry point |
| 120 | |
| 121 | extern void WebPInitSamplersMIPSdspR2(void); |
| 122 | |
| 123 | WEBP_TSAN_IGNORE_FUNCTION void WebPInitSamplersMIPSdspR2(void) { |
| 124 | WebPSamplers[MODE_RGB] = YuvToRgbRow_MIPSdspR2; |
| 125 | WebPSamplers[MODE_RGBA] = YuvToRgbaRow_MIPSdspR2; |
| 126 | WebPSamplers[MODE_BGR] = YuvToBgrRow_MIPSdspR2; |
| 127 | WebPSamplers[MODE_BGRA] = YuvToBgraRow_MIPSdspR2; |
| 128 | } |
| 129 | |
| 130 | #else // !WEBP_USE_MIPS_DSP_R2 |
| 131 | |
| 132 | WEBP_DSP_INIT_STUB(WebPInitSamplersMIPSdspR2) |
| 133 | |
| 134 | #endif // WEBP_USE_MIPS_DSP_R2 |
| 135 | |