media/libvpx/vp9/common/x86/vp9_copy_sse2.asm@ac0c01689b40 (annotated)
media/libvpx/vp9/common/x86/vp9_copy_sse2.asm
Thu, 15 Jan 2015 15:59:08 +0100
- author
- Michael Schloh von Bennewitz <michael@schloh.com>
- date
- Thu, 15 Jan 2015 15:59:08 +0100
- branch
- TOR_BUG_9701
- changeset 10
- ac0c01689b40
- permissions
- -rw-r--r--
Implement a real Private Browsing Mode condition by changing the API/ABI;
This solves Tor bug #9701, complying with disk avoidance documented in
https://www.torproject.org/projects/torbrowser/design/#disk-avoidance.
| michael@0 | 1 | ; |
| michael@0 | 2 | ; Copyright (c) 2010 The WebM project authors. All Rights Reserved. |
| michael@0 | 3 | ; |
| michael@0 | 4 | ; Use of this source code is governed by a BSD-style license |
| michael@0 | 5 | ; that can be found in the LICENSE file in the root of the source |
| michael@0 | 6 | ; tree. An additional intellectual property rights grant can be found |
| michael@0 | 7 | ; in the file PATENTS. All contributing project authors may |
| michael@0 | 8 | ; be found in the AUTHORS file in the root of the source tree. |
| michael@0 | 9 | ; |
| michael@0 | 10 | |
| michael@0 | 11 | %include "third_party/x86inc/x86inc.asm" |
| michael@0 | 12 | |
| michael@0 | 13 | SECTION .text |
| michael@0 | 14 | |
| michael@0 | 15 | %macro convolve_fn 1 |
| michael@0 | 16 | INIT_XMM sse2 |
| michael@0 | 17 | cglobal convolve_%1, 4, 7, 4, src, src_stride, dst, dst_stride, \ |
| michael@0 | 18 | fx, fxs, fy, fys, w, h |
| michael@0 | 19 | mov r4d, dword wm |
| michael@0 | 20 | cmp r4d, 4 |
| michael@0 | 21 | je .w4 |
| michael@0 | 22 | cmp r4d, 8 |
| michael@0 | 23 | je .w8 |
| michael@0 | 24 | cmp r4d, 16 |
| michael@0 | 25 | je .w16 |
| michael@0 | 26 | cmp r4d, 32 |
| michael@0 | 27 | je .w32 |
| michael@0 | 28 | |
| michael@0 | 29 | mov r4d, dword hm |
| michael@0 | 30 | .loop64: |
| michael@0 | 31 | movu m0, [srcq] |
| michael@0 | 32 | movu m1, [srcq+16] |
| michael@0 | 33 | movu m2, [srcq+32] |
| michael@0 | 34 | movu m3, [srcq+48] |
| michael@0 | 35 | add srcq, src_strideq |
| michael@0 | 36 | %ifidn %1, avg |
| michael@0 | 37 | pavgb m0, [dstq] |
| michael@0 | 38 | pavgb m1, [dstq+16] |
| michael@0 | 39 | pavgb m2, [dstq+32] |
| michael@0 | 40 | pavgb m3, [dstq+48] |
| michael@0 | 41 | %endif |
| michael@0 | 42 | mova [dstq ], m0 |
| michael@0 | 43 | mova [dstq+16], m1 |
| michael@0 | 44 | mova [dstq+32], m2 |
| michael@0 | 45 | mova [dstq+48], m3 |
| michael@0 | 46 | add dstq, dst_strideq |
| michael@0 | 47 | dec r4d |
| michael@0 | 48 | jnz .loop64 |
| michael@0 | 49 | RET |
| michael@0 | 50 | |
| michael@0 | 51 | .w32: |
| michael@0 | 52 | mov r4d, dword hm |
| michael@0 | 53 | .loop32: |
| michael@0 | 54 | movu m0, [srcq] |
| michael@0 | 55 | movu m1, [srcq+16] |
| michael@0 | 56 | movu m2, [srcq+src_strideq] |
| michael@0 | 57 | movu m3, [srcq+src_strideq+16] |
| michael@0 | 58 | lea srcq, [srcq+src_strideq*2] |
| michael@0 | 59 | %ifidn %1, avg |
| michael@0 | 60 | pavgb m0, [dstq] |
| michael@0 | 61 | pavgb m1, [dstq +16] |
| michael@0 | 62 | pavgb m2, [dstq+dst_strideq] |
| michael@0 | 63 | pavgb m3, [dstq+dst_strideq+16] |
| michael@0 | 64 | %endif |
| michael@0 | 65 | mova [dstq ], m0 |
| michael@0 | 66 | mova [dstq +16], m1 |
| michael@0 | 67 | mova [dstq+dst_strideq ], m2 |
| michael@0 | 68 | mova [dstq+dst_strideq+16], m3 |
| michael@0 | 69 | lea dstq, [dstq+dst_strideq*2] |
| michael@0 | 70 | sub r4d, 2 |
| michael@0 | 71 | jnz .loop32 |
| michael@0 | 72 | RET |
| michael@0 | 73 | |
| michael@0 | 74 | .w16: |
| michael@0 | 75 | mov r4d, dword hm |
| michael@0 | 76 | lea r5q, [src_strideq*3] |
| michael@0 | 77 | lea r6q, [dst_strideq*3] |
| michael@0 | 78 | .loop16: |
| michael@0 | 79 | movu m0, [srcq] |
| michael@0 | 80 | movu m1, [srcq+src_strideq] |
| michael@0 | 81 | movu m2, [srcq+src_strideq*2] |
| michael@0 | 82 | movu m3, [srcq+r5q] |
| michael@0 | 83 | lea srcq, [srcq+src_strideq*4] |
| michael@0 | 84 | %ifidn %1, avg |
| michael@0 | 85 | pavgb m0, [dstq] |
| michael@0 | 86 | pavgb m1, [dstq+dst_strideq] |
| michael@0 | 87 | pavgb m2, [dstq+dst_strideq*2] |
| michael@0 | 88 | pavgb m3, [dstq+r6q] |
| michael@0 | 89 | %endif |
| michael@0 | 90 | mova [dstq ], m0 |
| michael@0 | 91 | mova [dstq+dst_strideq ], m1 |
| michael@0 | 92 | mova [dstq+dst_strideq*2], m2 |
| michael@0 | 93 | mova [dstq+r6q ], m3 |
| michael@0 | 94 | lea dstq, [dstq+dst_strideq*4] |
| michael@0 | 95 | sub r4d, 4 |
| michael@0 | 96 | jnz .loop16 |
| michael@0 | 97 | RET |
| michael@0 | 98 | |
| michael@0 | 99 | INIT_MMX sse |
| michael@0 | 100 | .w8: |
| michael@0 | 101 | mov r4d, dword hm |
| michael@0 | 102 | lea r5q, [src_strideq*3] |
| michael@0 | 103 | lea r6q, [dst_strideq*3] |
| michael@0 | 104 | .loop8: |
| michael@0 | 105 | movu m0, [srcq] |
| michael@0 | 106 | movu m1, [srcq+src_strideq] |
| michael@0 | 107 | movu m2, [srcq+src_strideq*2] |
| michael@0 | 108 | movu m3, [srcq+r5q] |
| michael@0 | 109 | lea srcq, [srcq+src_strideq*4] |
| michael@0 | 110 | %ifidn %1, avg |
| michael@0 | 111 | pavgb m0, [dstq] |
| michael@0 | 112 | pavgb m1, [dstq+dst_strideq] |
| michael@0 | 113 | pavgb m2, [dstq+dst_strideq*2] |
| michael@0 | 114 | pavgb m3, [dstq+r6q] |
| michael@0 | 115 | %endif |
| michael@0 | 116 | mova [dstq ], m0 |
| michael@0 | 117 | mova [dstq+dst_strideq ], m1 |
| michael@0 | 118 | mova [dstq+dst_strideq*2], m2 |
| michael@0 | 119 | mova [dstq+r6q ], m3 |
| michael@0 | 120 | lea dstq, [dstq+dst_strideq*4] |
| michael@0 | 121 | sub r4d, 4 |
| michael@0 | 122 | jnz .loop8 |
| michael@0 | 123 | RET |
| michael@0 | 124 | |
| michael@0 | 125 | .w4: |
| michael@0 | 126 | mov r4d, dword hm |
| michael@0 | 127 | lea r5q, [src_strideq*3] |
| michael@0 | 128 | lea r6q, [dst_strideq*3] |
| michael@0 | 129 | .loop4: |
| michael@0 | 130 | movh m0, [srcq] |
| michael@0 | 131 | movh m1, [srcq+src_strideq] |
| michael@0 | 132 | movh m2, [srcq+src_strideq*2] |
| michael@0 | 133 | movh m3, [srcq+r5q] |
| michael@0 | 134 | lea srcq, [srcq+src_strideq*4] |
| michael@0 | 135 | %ifidn %1, avg |
| michael@0 | 136 | pavgb m0, [dstq] |
| michael@0 | 137 | pavgb m1, [dstq+dst_strideq] |
| michael@0 | 138 | pavgb m2, [dstq+dst_strideq*2] |
| michael@0 | 139 | pavgb m3, [dstq+r6q] |
| michael@0 | 140 | %endif |
| michael@0 | 141 | movh [dstq ], m0 |
| michael@0 | 142 | movh [dstq+dst_strideq ], m1 |
| michael@0 | 143 | movh [dstq+dst_strideq*2], m2 |
| michael@0 | 144 | movh [dstq+r6q ], m3 |
| michael@0 | 145 | lea dstq, [dstq+dst_strideq*4] |
| michael@0 | 146 | sub r4d, 4 |
| michael@0 | 147 | jnz .loop4 |
| michael@0 | 148 | RET |
| michael@0 | 149 | %endmacro |
| michael@0 | 150 | |
| michael@0 | 151 | convolve_fn copy |
| michael@0 | 152 | convolve_fn avg |
