| 1 |
From 065925f7363a38ebeaa25b492e4e6c35b0d52393 Mon Sep 17 00:00:00 2001 |
| 2 |
From: elupus <elupus@xbmc.org> |
| 3 |
Date: Thu, 29 Mar 2012 21:11:18 +0200 |
| 4 |
Subject: [PATCH 103/145] Drop neon patch that doesn't apply anymore (cherry |
| 5 |
picked from commit |
| 6 |
2d8334be4f7112d543630af00c53f8052fcfbc12) |
| 7 |
|
| 8 |
--- |
| 9 |
.../patches/0057-Fixes-neon-usage-under-iOS5.patch | 307 -------------------- |
| 10 |
1 file changed, 307 deletions(-) |
| 11 |
delete mode 100644 lib/ffmpeg/patches/0057-Fixes-neon-usage-under-iOS5.patch |
| 12 |
|
| 13 |
diff --git a/lib/ffmpeg/patches/0057-Fixes-neon-usage-under-iOS5.patch b/lib/ffmpeg/patches/0057-Fixes-neon-usage-under-iOS5.patch |
| 14 |
deleted file mode 100644 |
| 15 |
index 7b55691..0000000 |
| 16 |
--- a/lib/ffmpeg/patches/0057-Fixes-neon-usage-under-iOS5.patch |
| 17 |
+++ /dev/null |
| 18 |
@@ -1,307 +0,0 @@ |
| 19 |
-diff --git a/lib/ffmpeg/libavcodec/arm/fft_neon.S b/lib/ffmpeg/libavcodec/arm/fft_neon.S |
| 20 |
-index 1db7abd..6390065 100644 |
| 21 |
---- a/lib/ffmpeg/libavcodec/arm/fft_neon.S |
| 22 |
-+++ b/lib/ffmpeg/libavcodec/arm/fft_neon.S |
| 23 |
-@@ -101,8 +101,12 @@ function fft8_neon |
| 24 |
- bx lr |
| 25 |
- endfunc |
| 26 |
- |
| 27 |
-+ .align 4 |
| 28 |
-+pmmp: .float +1.0, -1.0, -1.0, +1.0 |
| 29 |
-+mppm: .float -M_SQRT1_2, M_SQRT1_2, M_SQRT1_2, -M_SQRT1_2 |
| 30 |
-+ |
| 31 |
- function fft16_neon |
| 32 |
-- movrel r1, mppm |
| 33 |
-+ adr r1, mppm |
| 34 |
- vld1.32 {d16-d19}, [r0,:128]! @ q8{r0,i0,r1,i1} q9{r2,i2,r3,i3} |
| 35 |
- pld [r0, #32] |
| 36 |
- vld1.32 {d2-d3}, [r1,:128] |
| 37 |
-@@ -144,12 +148,16 @@ function fft16_neon |
| 38 |
- vswp d29, d30 @ q14{r12,i12,i14,r15} q15{r13,i13,i15,r14} |
| 39 |
- vadd.f32 q0, q12, q13 @ {t1,t2,t5,t6} |
| 40 |
- vadd.f32 q1, q14, q15 @ {t1a,t2a,t5a,t6a} |
| 41 |
-- movrel r2, X(ff_cos_16) |
| 42 |
-+ ldr r2, _neon_label |
| 43 |
-+ ldr r3, L$diff1 |
| 44 |
-+ add r2, r3 |
| 45 |
-+local_label1: |
| 46 |
-+ ldr r2, [pc, r2] |
| 47 |
- vsub.f32 q13, q12, q13 @ {t3,t4,t7,t8} |
| 48 |
- vrev64.32 d1, d1 |
| 49 |
- vsub.f32 q15, q14, q15 @ {t3a,t4a,t7a,t8a} |
| 50 |
- vrev64.32 d3, d3 |
| 51 |
-- movrel r3, pmmp |
| 52 |
-+ adr r3, pmmp |
| 53 |
- vswp d1, d26 @ q0{t1,t2,t3,t4} q13{t6,t5,t7,t8} |
| 54 |
- vswp d3, d30 @ q1{t1a,t2a,t3a,t4a} q15{t6a,t5a,t7a,t8a} |
| 55 |
- vadd.f32 q12, q0, q13 @ {r8,i8,r9,i9} |
| 56 |
-@@ -214,7 +222,7 @@ function fft_pass_neon |
| 57 |
- add r2, r2, r0 @ &z[o2] |
| 58 |
- add r3, r3, r0 @ &z[o3] |
| 59 |
- vld1.32 {d20-d21},[r2,:128] @ {z[o2],z[o2+1]} |
| 60 |
-- movrel r12, pmmp |
| 61 |
-+ adr r12, pmmp |
| 62 |
- vld1.32 {d22-d23},[r3,:128] @ {z[o3],z[o3+1]} |
| 63 |
- add r5, r5, r1 @ wim |
| 64 |
- vld1.32 {d6-d7}, [r12,:128] @ pmmp |
| 65 |
-@@ -279,6 +287,11 @@ function fft_pass_neon |
| 66 |
- pop {r4-r6,pc} |
| 67 |
- endfunc |
| 68 |
- |
| 69 |
-+.set L$offs16, 0 |
| 70 |
-+.macro setTabOffs n, n2 |
| 71 |
-+.set L$offs\n, L$offs\n2 + 4 |
| 72 |
-+.endm |
| 73 |
-+ |
| 74 |
- .macro def_fft n, n2, n4 |
| 75 |
- .align 6 |
| 76 |
- function fft\n\()_neon |
| 77 |
-@@ -291,10 +304,15 @@ function fft\n\()_neon |
| 78 |
- bl fft\n4\()_neon |
| 79 |
- mov r0, r4 |
| 80 |
- pop {r4, lr} |
| 81 |
-- movrel r1, X(ff_cos_\n) |
| 82 |
-+ ldr r1, _neon_label |
| 83 |
-+ add r1, #L$diff\n |
| 84 |
-+local_label\n: |
| 85 |
-+ ldr r1, [pc, r1] |
| 86 |
- mov r2, #\n4/2 |
| 87 |
- b fft_pass_neon |
| 88 |
- endfunc |
| 89 |
-+setTabOffs \n, \n2 |
| 90 |
-+.set L$diff\n, _neon_label - local_label\n + L$offs\n - 8 |
| 91 |
- .endm |
| 92 |
- |
| 93 |
- def_fft 32, 16, 8 |
| 94 |
-@@ -310,10 +328,14 @@ endfunc |
| 95 |
- def_fft 32768, 16384, 8192 |
| 96 |
- def_fft 65536, 32768, 16384 |
| 97 |
- |
| 98 |
-+.set L$diffTab, fft_tab_neon_offs - local_label_tab - 8 |
| 99 |
- function ff_fft_calc_neon, export=1 |
| 100 |
- ldr r2, [r0] |
| 101 |
- sub r2, r2, #2 |
| 102 |
-- movrel r3, fft_tab_neon |
| 103 |
-+ ldr r3, fft_tab_neon_offs |
| 104 |
-+ add r3, #L$diffTab |
| 105 |
-+local_label_tab: |
| 106 |
-+ add r3, pc |
| 107 |
- ldr r3, [r3, r2, lsl #2] |
| 108 |
- mov r0, r1 |
| 109 |
- bx r3 |
| 110 |
-@@ -349,9 +371,22 @@ function ff_fft_permute_neon, export=1 |
| 111 |
- pop {r4,pc} |
| 112 |
- endfunc |
| 113 |
- |
| 114 |
-- .section .rodata |
| 115 |
-+ |
| 116 |
-+.global _neon_label |
| 117 |
-+_neon_label: |
| 118 |
-+.word _neon_cos_tab - . |
| 119 |
-+ |
| 120 |
-+L$diff1: |
| 121 |
-+.word _neon_label - local_label1 - 8 |
| 122 |
-+ |
| 123 |
-+fft_tab_neon_offs: |
| 124 |
-+.word _fft_tab_neon - . |
| 125 |
-+ |
| 126 |
-+ |
| 127 |
-+.section .rodata |
| 128 |
-+ |
| 129 |
- .align 4 |
| 130 |
--fft_tab_neon: |
| 131 |
-+_fft_tab_neon: |
| 132 |
- .word fft4_neon |
| 133 |
- .word fft8_neon |
| 134 |
- .word fft16_neon |
| 135 |
-@@ -367,8 +402,20 @@ fft_tab_neon: |
| 136 |
- .word fft16384_neon |
| 137 |
- .word fft32768_neon |
| 138 |
- .word fft65536_neon |
| 139 |
--ELF .size fft_tab_neon, . - fft_tab_neon |
| 140 |
-- |
| 141 |
-- .align 4 |
| 142 |
--pmmp: .float +1.0, -1.0, -1.0, +1.0 |
| 143 |
--mppm: .float -M_SQRT1_2, M_SQRT1_2, M_SQRT1_2, -M_SQRT1_2 |
| 144 |
-+ELF .size _fft_tab_neon, . - _fft_tab_neon |
| 145 |
-+ |
| 146 |
-+ .align 4 |
| 147 |
-+_neon_cos_tab: |
| 148 |
-+ .word X(ff_cos_16) |
| 149 |
-+ .word X(ff_cos_32) |
| 150 |
-+ .word X(ff_cos_64) |
| 151 |
-+ .word X(ff_cos_128) |
| 152 |
-+ .word X(ff_cos_256) |
| 153 |
-+ .word X(ff_cos_512) |
| 154 |
-+ .word X(ff_cos_1024) |
| 155 |
-+ .word X(ff_cos_2048) |
| 156 |
-+ .word X(ff_cos_4096) |
| 157 |
-+ .word X(ff_cos_8192) |
| 158 |
-+ .word X(ff_cos_16384) |
| 159 |
-+ .word X(ff_cos_32768) |
| 160 |
-+ .word X(ff_cos_65536) |
| 161 |
-diff --git a/lib/ffmpeg/libavcodec/arm/h264idct_neon.S b/lib/ffmpeg/libavcodec/arm/h264idct_neon.S |
| 162 |
-index 6b6a669..8111975 100644 |
| 163 |
---- a/lib/ffmpeg/libavcodec/arm/h264idct_neon.S |
| 164 |
-+++ b/lib/ffmpeg/libavcodec/arm/h264idct_neon.S |
| 165 |
-@@ -97,7 +97,7 @@ function ff_h264_idct_add16_neon, export=1 |
| 166 |
- mov r1, r2 |
| 167 |
- mov r2, r3 |
| 168 |
- ldr r6, [sp, #24] |
| 169 |
-- movrel r7, scan8 |
| 170 |
-+ adr r7, scan8 |
| 171 |
- mov ip, #16 |
| 172 |
- 1: ldrb r8, [r7], #1 |
| 173 |
- ldr r0, [r5], #4 |
| 174 |
-@@ -117,6 +117,16 @@ function ff_h264_idct_add16_neon, export=1 |
| 175 |
- pop {r4-r8,pc} |
| 176 |
- endfunc |
| 177 |
- |
| 178 |
-+ .align |
| 179 |
-+scan8: .byte 4+1*8, 5+1*8, 4+2*8, 5+2*8 |
| 180 |
-+ .byte 6+1*8, 7+1*8, 6+2*8, 7+2*8 |
| 181 |
-+ .byte 4+3*8, 5+3*8, 4+4*8, 5+4*8 |
| 182 |
-+ .byte 6+3*8, 7+3*8, 6+4*8, 7+4*8 |
| 183 |
-+ .byte 1+1*8, 2+1*8 |
| 184 |
-+ .byte 1+2*8, 2+2*8 |
| 185 |
-+ .byte 1+4*8, 2+4*8 |
| 186 |
-+ .byte 1+5*8, 2+5*8 |
| 187 |
-+ |
| 188 |
- function ff_h264_idct_add16intra_neon, export=1 |
| 189 |
- push {r4-r8,lr} |
| 190 |
- mov r4, r0 |
| 191 |
-@@ -124,7 +134,7 @@ function ff_h264_idct_add16intra_neon, export=1 |
| 192 |
- mov r1, r2 |
| 193 |
- mov r2, r3 |
| 194 |
- ldr r6, [sp, #24] |
| 195 |
-- movrel r7, scan8 |
| 196 |
-+ adr r7, scan8 |
| 197 |
- mov ip, #16 |
| 198 |
- 1: ldrb r8, [r7], #1 |
| 199 |
- ldr r0, [r5], #4 |
| 200 |
-@@ -149,7 +159,7 @@ function ff_h264_idct_add8_neon, export=1 |
| 201 |
- add r1, r2, #16*32 |
| 202 |
- mov r2, r3 |
| 203 |
- ldr r6, [sp, #32] |
| 204 |
-- movrel r7, scan8+16 |
| 205 |
-+ adr r7, scan8+16 |
| 206 |
- mov ip, #7 |
| 207 |
- 1: ldrb r8, [r7], #1 |
| 208 |
- ldr r0, [r5], #4 |
| 209 |
-@@ -353,7 +363,7 @@ function ff_h264_idct8_add4_neon, export=1 |
| 210 |
- mov r1, r2 |
| 211 |
- mov r2, r3 |
| 212 |
- ldr r6, [sp, #24] |
| 213 |
-- movrel r7, scan8 |
| 214 |
-+ adr r7, scan8 |
| 215 |
- mov r12, #16 |
| 216 |
- 1: ldrb r8, [r7], #4 |
| 217 |
- ldr r0, [r5], #16 |
| 218 |
-@@ -372,13 +382,3 @@ function ff_h264_idct8_add4_neon, export=1 |
| 219 |
- bne 1b |
| 220 |
- pop {r4-r8,pc} |
| 221 |
- endfunc |
| 222 |
-- |
| 223 |
-- .section .rodata |
| 224 |
--scan8: .byte 4+1*8, 5+1*8, 4+2*8, 5+2*8 |
| 225 |
-- .byte 6+1*8, 7+1*8, 6+2*8, 7+2*8 |
| 226 |
-- .byte 4+3*8, 5+3*8, 4+4*8, 5+4*8 |
| 227 |
-- .byte 6+3*8, 7+3*8, 6+4*8, 7+4*8 |
| 228 |
-- .byte 1+1*8, 2+1*8 |
| 229 |
-- .byte 1+2*8, 2+2*8 |
| 230 |
-- .byte 1+4*8, 2+4*8 |
| 231 |
-- .byte 1+5*8, 2+5*8 |
| 232 |
-diff --git a/lib/ffmpeg/libavcodec/arm/h264pred_neon.S b/lib/ffmpeg/libavcodec/arm/h264pred_neon.S |
| 233 |
-index 63c96ee..357c10a 100644 |
| 234 |
---- a/lib/ffmpeg/libavcodec/arm/h264pred_neon.S |
| 235 |
-+++ b/lib/ffmpeg/libavcodec/arm/h264pred_neon.S |
| 236 |
-@@ -123,7 +123,7 @@ function ff_pred16x16_plane_neon, export=1 |
| 237 |
- vaddl.u8 q8, d2, d3 |
| 238 |
- vsubl.u8 q2, d2, d0 |
| 239 |
- vsubl.u8 q3, d3, d1 |
| 240 |
-- movrel r3, p16weight |
| 241 |
-+ adr r3, p16weight |
| 242 |
- vld1.8 {q0}, [r3,:128] |
| 243 |
- vmul.s16 q2, q2, q0 |
| 244 |
- vmul.s16 q3, q3, q0 |
| 245 |
-@@ -166,7 +166,6 @@ function ff_pred16x16_plane_neon, export=1 |
| 246 |
- bx lr |
| 247 |
- endfunc |
| 248 |
- |
| 249 |
-- .section .rodata |
| 250 |
- .align 4 |
| 251 |
- p16weight: |
| 252 |
- .short 1,2,3,4,5,6,7,8 |
| 253 |
-@@ -207,7 +206,7 @@ function ff_pred8x8_plane_neon, export=1 |
| 254 |
- vrev32.8 d0, d0 |
| 255 |
- vtrn.32 d2, d3 |
| 256 |
- vsubl.u8 q2, d2, d0 |
| 257 |
-- movrel r3, p16weight |
| 258 |
-+ adr r3, p16weight |
| 259 |
- vld1.16 {q0}, [r3,:128] |
| 260 |
- vmul.s16 d4, d4, d0 |
| 261 |
- vmul.s16 d5, d5, d0 |
| 262 |
-diff --git a/lib/ffmpeg/libavcodec/arm/simple_idct_neon.S b/lib/ffmpeg/libavcodec/arm/simple_idct_neon.S |
| 263 |
-index 17cde58..e61414e 100644 |
| 264 |
---- a/lib/ffmpeg/libavcodec/arm/simple_idct_neon.S |
| 265 |
-+++ b/lib/ffmpeg/libavcodec/arm/simple_idct_neon.S |
| 266 |
-@@ -239,7 +239,6 @@ function idct_col4_st8_neon |
| 267 |
- bx lr |
| 268 |
- endfunc |
| 269 |
- |
| 270 |
-- .section .rodata |
| 271 |
- .align 4 |
| 272 |
- idct_coeff_neon: |
| 273 |
- .short W1, W2, W3, W4, W5, W6, W7, W4c |
| 274 |
-@@ -249,7 +248,7 @@ idct_coeff_neon: |
| 275 |
- pld [\data] |
| 276 |
- pld [\data, #64] |
| 277 |
- vpush {d8-d15} |
| 278 |
-- movrel r3, idct_coeff_neon |
| 279 |
-+ adr r3, idct_coeff_neon |
| 280 |
- vld1.64 {d0,d1}, [r3,:128] |
| 281 |
- .endm |
| 282 |
- |
| 283 |
-diff --git a/lib/ffmpeg/libavcodec/arm/vp3dsp_neon.S b/lib/ffmpeg/libavcodec/arm/vp3dsp_neon.S |
| 284 |
-index d97ed3d..74bf7ba 100644 |
| 285 |
---- a/lib/ffmpeg/libavcodec/arm/vp3dsp_neon.S |
| 286 |
-+++ b/lib/ffmpeg/libavcodec/arm/vp3dsp_neon.S |
| 287 |
-@@ -20,12 +20,9 @@ |
| 288 |
- |
| 289 |
- #include "asm.S" |
| 290 |
- |
| 291 |
--.section .rodata |
| 292 |
-+.text |
| 293 |
- .align 4 |
| 294 |
- |
| 295 |
--vp3_idct_constants: |
| 296 |
--.short 64277, 60547, 54491, 46341, 36410, 25080, 12785 |
| 297 |
-- |
| 298 |
- #define xC1S7 d0[0] |
| 299 |
- #define xC2S6 d0[1] |
| 300 |
- #define xC3S5 d0[2] |
| 301 |
-@@ -34,8 +31,6 @@ vp3_idct_constants: |
| 302 |
- #define xC6S2 d1[1] |
| 303 |
- #define xC7S1 d1[2] |
| 304 |
- |
| 305 |
--.text |
| 306 |
-- |
| 307 |
- .macro vp3_loop_filter |
| 308 |
- vsubl.u8 q3, d18, d17 |
| 309 |
- vsubl.u8 q2, d16, d19 |
| 310 |
-@@ -109,10 +104,14 @@ function ff_vp3_h_loop_filter_neon, export=1 |
| 311 |
- bx lr |
| 312 |
- endfunc |
| 313 |
- |
| 314 |
-+.align 4 |
| 315 |
-+vp3_idct_constants: |
| 316 |
-+.short 64277, 60547, 54491, 46341, 36410, 25080, 12785 |
| 317 |
-+.align 4 |
| 318 |
- |
| 319 |
- function vp3_idct_start_neon |
| 320 |
- vpush {d8-d15} |
| 321 |
-- movrel r3, vp3_idct_constants |
| 322 |
-+ adr r3, vp3_idct_constants |
| 323 |
- vld1.64 {d0-d1}, [r3,:128] |
| 324 |
- vld1.64 {d16-d19}, [r2,:128]! |
| 325 |
- vld1.64 {d20-d23}, [r2,:128]! |
| 326 |
-- |
| 327 |
1.7.10 |
| 328 |
|
Parent Directory
| 
Revision Log