The Android Open Source Project | dd7bc33 | 2009-03-03 19:32:55 -0800 | [diff] [blame^] | 1 | /* libs/pixelflinger/t32cb16blend.S |
| 2 | ** |
| 3 | ** Copyright 2006, The Android Open Source Project |
| 4 | ** |
| 5 | ** Licensed under the Apache License, Version 2.0 (the "License"); |
| 6 | ** you may not use this file except in compliance with the License. |
| 7 | ** You may obtain a copy of the License at |
| 8 | ** |
| 9 | ** http://www.apache.org/licenses/LICENSE-2.0 |
| 10 | ** |
| 11 | ** Unless required by applicable law or agreed to in writing, software |
| 12 | ** distributed under the License is distributed on an "AS IS" BASIS, |
| 13 | ** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 14 | ** See the License for the specific language governing permissions and |
| 15 | ** limitations under the License. |
| 16 | */ |
| 17 | |
| 18 | |
| 19 | .text |
| 20 | .align |
| 21 | |
| 22 | .global scanline_t32cb16blend_arm |
| 23 | |
| 24 | // uses r6, r7, lr |
| 25 | |
| 26 | .macro pixel, DREG, SRC, FB, OFFSET |
| 27 | |
| 28 | // SRC = AARRGGBB |
| 29 | mov r7, \SRC, lsr #24 // sA |
| 30 | add r7, r7, r7, lsr #7 // sA + (sA >> 7) |
| 31 | rsb r7, r7, #0x100 // sA = 0x100 - (sA+(sA>>7)) |
| 32 | |
| 33 | 1: |
| 34 | |
| 35 | .if \OFFSET |
| 36 | |
| 37 | // red |
| 38 | mov lr, \DREG, lsr #(\OFFSET + 6 + 5) |
| 39 | smulbb lr, r7, lr |
| 40 | mov r6, \SRC, lsr #3 |
| 41 | and r6, r6, #0x1F |
| 42 | add lr, r6, lr, lsr #8 |
| 43 | orr \FB, lr, lsl #(\OFFSET + 11) |
| 44 | |
| 45 | // green |
| 46 | and r6, \DREG, #(0x3F<<(\OFFSET + 5)) |
| 47 | smulbt r6, r7, r6 |
| 48 | mov lr, \SRC, lsr #(8+2) |
| 49 | and lr, lr, #0x3F |
| 50 | add r6, lr, r6, lsr #(5+8) |
| 51 | orr \FB, \FB, r6, lsl #(\OFFSET + 5) |
| 52 | |
| 53 | // blue |
| 54 | and lr, \DREG, #(0x1F << \OFFSET) |
| 55 | smulbt lr, r7, lr |
| 56 | mov r6, \SRC, lsr #(8+8+3) |
| 57 | and r6, r6, #0x1F |
| 58 | add lr, r6, lr, lsr #8 |
| 59 | orr \FB, \FB, lr, lsl #\OFFSET |
| 60 | |
| 61 | .else |
| 62 | |
| 63 | // red |
| 64 | mov lr, \DREG, lsr #(6+5) |
| 65 | and lr, lr, #0x1F |
| 66 | smulbb lr, r7, lr |
| 67 | mov r6, \SRC, lsr #3 |
| 68 | and r6, r6, #0x1F |
| 69 | add lr, r6, lr, lsr #8 |
| 70 | mov \FB, lr, lsl #11 |
| 71 | |
| 72 | // green |
| 73 | and r6, \DREG, #(0x3F<<5) |
| 74 | smulbb r6, r7, r6 |
| 75 | mov lr, \SRC, lsr #(8+2) |
| 76 | and lr, lr, #0x3F |
| 77 | add r6, lr, r6, lsr #(5+8) |
| 78 | orr \FB, \FB, r6, lsl #5 |
| 79 | |
| 80 | // blue |
| 81 | and lr, \DREG, #0x1F |
| 82 | smulbb lr, r7, lr |
| 83 | mov r6, \SRC, lsr #(8+8+3) |
| 84 | and r6, r6, #0x1F |
| 85 | add lr, r6, lr, lsr #8 |
| 86 | orr \FB, \FB, lr |
| 87 | |
| 88 | .endif |
| 89 | |
| 90 | .endm |
| 91 | |
| 92 | |
| 93 | // r0: dst ptr |
| 94 | // r1: src ptr |
| 95 | // r2: count |
| 96 | // r3: d |
| 97 | // r4: s0 |
| 98 | // r5: s1 |
| 99 | // r6: pixel |
| 100 | // r7: pixel |
| 101 | // r8: free |
| 102 | // r9: free |
| 103 | // r10: free |
| 104 | // r11: free |
| 105 | // r12: scratch |
| 106 | // r14: pixel |
| 107 | |
| 108 | scanline_t32cb16blend_arm: |
| 109 | stmfd sp!, {r4-r7, lr} |
| 110 | |
| 111 | pld [r0] |
| 112 | pld [r1] |
| 113 | |
| 114 | // align DST to 32 bits |
| 115 | tst r0, #0x3 |
| 116 | beq aligned |
| 117 | subs r2, r2, #1 |
| 118 | ldmlofd sp!, {r4-r7, lr} // return |
| 119 | bxlo lr |
| 120 | |
| 121 | last: |
| 122 | ldr r4, [r1], #4 |
| 123 | ldrh r3, [r0] |
| 124 | pixel r3, r4, r12, 0 |
| 125 | strh r12, [r0], #2 |
| 126 | |
| 127 | aligned: |
| 128 | subs r2, r2, #2 |
| 129 | blo 9f |
| 130 | |
| 131 | // The main loop is unrolled twice and process 4 pixels |
| 132 | 8: ldmia r1!, {r4, r5} |
| 133 | // stream the source |
| 134 | pld [r1, #32] |
| 135 | add r0, r0, #4 |
| 136 | // it's all zero, skip this pixel |
| 137 | orrs r3, r4, r5 |
| 138 | beq 7f |
| 139 | |
| 140 | // load the destination |
| 141 | ldr r3, [r0, #-4] |
| 142 | // stream the destination |
| 143 | pld [r0, #32] |
| 144 | pixel r3, r4, r12, 0 |
| 145 | pixel r3, r5, r12, 16 |
| 146 | // effectively, we're getting write-combining by virtue of the |
| 147 | // cpu's write-back cache. |
| 148 | str r12, [r0, #-4] |
| 149 | |
| 150 | // 2nd iterration of the loop, don't stream anything |
| 151 | subs r2, r2, #2 |
| 152 | movlt r4, r5 |
| 153 | blt 9f |
| 154 | ldmia r1!, {r4, r5} |
| 155 | add r0, r0, #4 |
| 156 | orrs r3, r4, r5 |
| 157 | beq 7f |
| 158 | ldr r3, [r0, #-4] |
| 159 | pixel r3, r4, r12, 0 |
| 160 | pixel r3, r5, r12, 16 |
| 161 | str r12, [r0, #-4] |
| 162 | |
| 163 | |
| 164 | 7: subs r2, r2, #2 |
| 165 | bhs 8b |
| 166 | mov r4, r5 |
| 167 | |
| 168 | 9: adds r2, r2, #1 |
| 169 | ldmlofd sp!, {r4-r7, lr} // return |
| 170 | bxlo lr |
| 171 | b last |