OSDN Git Service

Merge commit 'dfe224f377be3e45758c69d881ca7874b82d647a'
authorMichael Niedermayer <michaelni@gmx.at>
Sun, 9 Mar 2014 12:26:50 +0000 (13:26 +0100)
committerMichael Niedermayer <michaelni@gmx.at>
Sun, 9 Mar 2014 12:27:04 +0000 (13:27 +0100)
* commit 'dfe224f377be3e45758c69d881ca7874b82d647a':
  aarch64: get_cabac inline asm

Merged-by: Michael Niedermayer <michaelni@gmx.at>
1  2 
libavcodec/aarch64/cabac.h
libavcodec/cabac.h
libavcodec/cabac_functions.h

index 0000000,e12953e..6b9b77e
mode 000000,100644..100644
--- /dev/null
@@@ -1,0 -1,104 +1,104 @@@
 - * This file is part of Libav.
+ /*
 - * Libav is free software; you can redistribute it and/or
++ * This file is part of FFmpeg.
+  *
 - * Libav is distributed in the hope that it will be useful,
++ * FFmpeg is free software; you can redistribute it and/or
+  * modify it under the terms of the GNU Lesser General Public
+  * License as published by the Free Software Foundation; either
+  * version 2.1 of the License, or (at your option) any later version.
+  *
 - * License along with Libav; if not, write to the Free Software
++ * FFmpeg is distributed in the hope that it will be useful,
+  * but WITHOUT ANY WARRANTY; without even the implied warranty of
+  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+  * Lesser General Public License for more details.
+  *
+  * You should have received a copy of the GNU Lesser General Public
++ * License along with FFmpeg; if not, write to the Free Software
+  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+  */
+ #ifndef AVCODEC_AARCH64_CABAC_H
+ #define AVCODEC_AARCH64_CABAC_H
+ #include "config.h"
+ #if HAVE_INLINE_ASM
+ #include "libavutil/attributes.h"
+ #include "libavutil/internal.h"
+ #include "libavcodec/cabac.h"
+ #define get_cabac_inline get_cabac_inline_aarch64
+ static av_always_inline int get_cabac_inline_aarch64(CABACContext *c,
+                                                      uint8_t *const state)
+ {
+     int bit;
+     void *reg_a, *reg_b, *reg_c, *tmp;
+     __asm__ volatile(
+         "ldrb       %w[bit]       , [%[state]]                  \n\t"
+         "add        %[r_b]        , %[tables]   , %[lps_off]    \n\t"
+         "mov        %w[tmp]       , %w[range]                   \n\t"
+         "and        %w[range]     , %w[range]   , #0xC0         \n\t"
+         "lsl        %w[r_c]       , %w[range]   , #1            \n\t"
+         "add        %[r_b]        , %[r_b]      , %w[bit], UXTW \n\t"
+         "ldrb       %w[range]     , [%[r_b], %w[r_c], SXTW]     \n\t"
+         "sub        %w[r_c]       , %w[tmp]     , %w[range]     \n\t"
+         "lsl        %w[tmp]       , %w[r_c]     , #17           \n\t"
+         "cmp        %w[tmp]       , %w[low]                     \n\t"
+         "csel       %w[tmp]       , %w[tmp]     , wzr      , cc \n\t"
+         "csel       %w[range]     , %w[r_c]     , %w[range], gt \n\t"
+         "cinv       %w[bit]       , %w[bit]     , cc            \n\t"
+         "sub        %w[low]       , %w[low]     , %w[tmp]       \n\t"
+         "add        %[r_b]        , %[tables]   , %[norm_off]   \n\t"
+         "add        %[r_a]        , %[tables]   , %[mlps_off]   \n\t"
+         "ldrb       %w[tmp]       , [%[r_b], %w[range], SXTW]   \n\t"
+         "ldrb       %w[r_a]       , [%[r_a], %w[bit], SXTW]     \n\t"
+         "lsl        %w[low]       , %w[low]     , %w[tmp]       \n\t"
+         "lsl        %w[range]     , %w[range]   , %w[tmp]       \n\t"
+         "uxth       %w[r_c]       , %w[low]                     \n\t"
+         "strb       %w[r_a]       , [%[state]]                  \n\t"
+         "cbnz       %w[r_c]       , 2f                          \n\t"
+         "ldr        %[r_c]        , [%[c], %[byte]]             \n\t"
+         "ldr        %[r_a]        , [%[c], %[end]]              \n\t"
+         "ldrh       %w[tmp]       , [%[r_c]]                    \n\t"
+         "cmp        %[r_c]        , %[r_a]                      \n\t"
+         "b.ge       1f                                          \n\t"
+         "add        %[r_a]        , %[r_c]      , #2            \n\t"
+         "str        %[r_a]        , [%[c], %[byte]]             \n\t"
+         "1:                                                     \n\t"
+         "sub        %w[r_c]       , %w[low]     , #1            \n\t"
+         "eor        %w[r_c]       , %w[r_c]     , %w[low]       \n\t"
+         "rev        %w[tmp]       , %w[tmp]                     \n\t"
+         "lsr        %w[r_c]       , %w[r_c]     , #15           \n\t"
+         "lsr        %w[tmp]       , %w[tmp]     , #15           \n\t"
+         "ldrb       %w[r_c]       , [%[r_b], %w[r_c], SXTW]     \n\t"
+         "mov        %w[r_b]       , #0xFFFF                     \n\t"
+         "mov        %w[r_a]       , #7                          \n\t"
+         "sub        %w[tmp]       , %w[tmp]     , %w[r_b]       \n\t"
+         "sub        %w[r_c]       , %w[r_a]     , %w[r_c]       \n\t"
+         "lsl        %w[tmp]       , %w[tmp]     , %w[r_c]       \n\t"
+         "add        %w[low]       , %w[low]     , %w[tmp]       \n\t"
+         "2:                                                     \n\t"
+         :    [bit]"=&r"(bit),
+              [low]"+&r"(c->low),
+            [range]"+&r"(c->range),
+              [r_a]"=&r"(reg_a),
+              [r_b]"=&r"(reg_b),
+              [r_c]"=&r"(reg_c),
+              [tmp]"=&r"(tmp)
+         :        [c]"r"(c),
+              [state]"r"(state),
+             [tables]"r"(ff_h264_cabac_tables),
+               [byte]"i"(offsetof(CABACContext, bytestream)),
+                [end]"i"(offsetof(CABACContext, bytestream_end)),
+           [norm_off]"I"(H264_NORM_SHIFT_OFFSET),
+            [lps_off]"I"(H264_LPS_RANGE_OFFSET),
+           [mlps_off]"I"(H264_MLPS_STATE_OFFSET + 128)
+         : "memory", "cc"
+         );
+     return bit & 1;
+ }
+ #endif /* HAVE_INLINE_ASM */
+ #endif /* AVCODEC_AARCH64_CABAC_H */
Simple merge
Simple merge