Merge remote-tracking branch 'qatar/master'

* qatar/master:
  ffmpeg: get rid of the -vglobal option.
  dct32: Add AVX implementation of 32-point DCT
  dct32: Change pass 6 permutation to allow for AVX implementation
  dct32: port SSE 32-point DCT to YASM
  multiple inclusion guard cleanup
  avio: document buffer must created with av_malloc() and friends
  avio: check AVIOContext malloc failure
  swscale: point out an alternative to sws_getContext
  svq3: Do initialization after parsing the extradata
  add changelog entries for 0.7_beta2
  mp3lame: add #include required for AV_RB32 macro.

Conflicts:
	Changelog
	libavcodec/svq3.c
	libavcodec/x86/dct32_sse.c
	libavfilter/vsrc_buffer.h

Merged-by: Michael Niedermayer <michaelni@gmx.at>

ffmpeg.c

@@ -218,7 +218,6 @@ static float audio_drift_threshold= 0.1;
 static int copy_ts= 0;
 static int copy_tb= 0;
 static int opt_shortest = 0;
-static int video_global_header = 0;
 static char *vstats_filename;
 static FILE *vstats_file;
 static int opt_programid = 0;
@@ -3555,15 +3554,10 @@ static void new_video_stream(AVFormatContext *oc, int file_idx)
     if(video_codec_tag)
         video_enc->codec_tag= video_codec_tag;
 
-    if(   (video_global_header&1)
-       || (video_global_header==0 && (oc->oformat->flags & AVFMT_GLOBALHEADER))){
+    if(oc->oformat->flags & AVFMT_GLOBALHEADER) {
         video_enc->flags |= CODEC_FLAG_GLOBAL_HEADER;
         avcodec_opts[AVMEDIA_TYPE_VIDEO]->flags|= CODEC_FLAG_GLOBAL_HEADER;
     }
-    if(video_global_header&2){
-        video_enc->flags2 |= CODEC_FLAG2_LOCAL_HEADER;
-        avcodec_opts[AVMEDIA_TYPE_VIDEO]->flags2|= CODEC_FLAG2_LOCAL_HEADER;
-    }
 
     if (video_stream_copy) {
         st->stream_copy = 1;
@@ -4408,7 +4402,6 @@ static const OptionDef options[] = {
     { "vsync", HAS_ARG | OPT_INT | OPT_EXPERT, {(void*)&video_sync_method}, "video sync method", "" },
     { "async", HAS_ARG | OPT_INT | OPT_EXPERT, {(void*)&audio_sync_method}, "audio sync method", "" },
     { "adrift_threshold", HAS_ARG | OPT_FLOAT | OPT_EXPERT, {(void*)&audio_drift_threshold}, "audio drift threshold", "threshold" },
-    { "vglobal", HAS_ARG | OPT_INT | OPT_EXPERT, {(void*)&video_global_header}, "video global header storage type", "" },
     { "copyts", OPT_BOOL | OPT_EXPERT, {(void*)&copy_ts}, "copy timestamps" },
     { "copytb", OPT_BOOL | OPT_EXPERT, {(void*)&copy_tb}, "copy input stream time base when stream copying" },
     { "shortest", OPT_BOOL | OPT_EXPERT, {(void*)&opt_shortest}, "finish encoding within shortest input" }, //

libavcodec/aac_tablegen.h

@@ -20,8 +20,8 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef AAC_TABLEGEN_H
-#define AAC_TABLEGEN_H
+#ifndef AVCODEC_AAC_TABLEGEN_H
+#define AVCODEC_AAC_TABLEGEN_H
 
 #include "aac_tablegen_decl.h"
 
@@ -40,4 +40,4 @@ void ff_aac_tableinit(void)
 }
 #endif /* CONFIG_HARDCODED_TABLES */
 
-#endif /* AAC_TABLEGEN_H */
+#endif /* AVCODEC_AAC_TABLEGEN_H */

libavcodec/aac_tablegen_decl.h

@@ -20,8 +20,8 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef AAC_TABLEGEN_DECL_H
-#define AAC_TABLEGEN_DECL_H
+#ifndef AVCODEC_AAC_TABLEGEN_DECL_H
+#define AVCODEC_AAC_TABLEGEN_DECL_H
 
 #if CONFIG_HARDCODED_TABLES
 #define ff_aac_tableinit()
@@ -31,4 +31,4 @@ void ff_aac_tableinit(void);
 extern       float ff_aac_pow2sf_tab[428];
 #endif /* CONFIG_HARDCODED_TABLES */
 
-#endif /* AAC_TABLEGEN_DECL_H */
+#endif /* AVCODEC_AAC_TABLEGEN_DECL_H */

libavcodec/amrwbdata.h

@@ -1887,4 +1887,4 @@ static const uint16_t cf_sizes_wb[] = {
     40 /// SID/comfort noise frame
 };
 
-#endif
+#endif /* AVCODEC_AMRWBDATA_H */

libavcodec/arm/asm-offsets.h

@@ -36,4 +36,4 @@
 #define H263_AIC                 0xf0
 #define INTER_SCANTAB_RASTER_END 0x138
 
-#endif
+#endif /* AVCODEC_ARM_ASM_OFFSETS_H */

libavcodec/arm/dsputil_arm.h

@@ -30,4 +30,4 @@ void ff_dsputil_init_vfp(DSPContext* c, AVCodecContext *avctx);
 void ff_dsputil_init_neon(DSPContext *c, AVCodecContext *avctx);
 void ff_dsputil_init_iwmmxt(DSPContext* c, AVCodecContext *avctx);
 
-#endif
+#endif /* AVCODEC_ARM_DSPUTIL_H */

libavcodec/arm/mpegvideo_arm.h

@@ -24,4 +24,4 @@
 void MPV_common_init_iwmmxt(MpegEncContext *s);
 void MPV_common_init_armv5te(MpegEncContext *s);
 
-#endif
+#endif /* AVCODEC_ARM_MPEGVIDEO_H */

libavcodec/arm/vp56_arith.h

@@ -89,4 +89,4 @@ static inline int vp56_rac_get_prob_branchy_armv6(VP56RangeCoder *c, int pr)
 
 #endif
 
-#endif
+#endif /* AVCODEC_ARM_VP56_ARITH_H */

libavcodec/arm/vp8.h

@@ -26,4 +26,4 @@ int ff_decode_block_coeffs_armv6(VP56RangeCoder *rc, DCTELEM block[16],
                                  int i, uint8_t *token_prob, int16_t qmul[2]);
 #endif
 
-#endif
+#endif /* AVCODEC_ARM_VP8_H */

libavcodec/cavsdsp.h

@@ -38,4 +38,4 @@ typedef struct CAVSDSPContext {
 void ff_cavsdsp_init(CAVSDSPContext* c, AVCodecContext *avctx);
 void ff_cavsdsp_init_mmx(CAVSDSPContext* c, AVCodecContext *avctx);
 
-#endif
+#endif /* AVCODEC_CAVSDSP_H */

libavcodec/cbrt_tablegen.h

@@ -20,8 +20,8 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef CBRT_TABLEGEN_H
-#define CBRT_TABLEGEN_H
+#ifndef AVCODEC_CBRT_TABLEGEN_H
+#define AVCODEC_CBRT_TABLEGEN_H
 
 #include <stdint.h>
 #include <math.h>
@@ -48,4 +48,4 @@ static void cbrt_tableinit(void)
 }
 #endif /* CONFIG_HARDCODED_TABLES */
 
-#endif /* CBRT_TABLEGEN_H */
+#endif /* AVCODEC_CBRT_TABLEGEN_H */

libavcodec/cga_data.h

@@ -45,4 +45,4 @@ extern const uint32_t ff_ega_palette[64];
  */
 void ff_draw_pc_font(uint8_t *dst, int linesize, const uint8_t *font, int font_height, int ch, int fg, int bg);
 
-#endif
+#endif /* AVCODEC_CGA_DATA_H */

libavcodec/dct.h

@@ -49,4 +49,4 @@ void ff_dct_end (DCTContext *s);
 
 void ff_dct_init_mmx(DCTContext *s);
 
-#endif
+#endif /* AVCODEC_DCT_H */

libavcodec/dctref.h

@@ -28,4 +28,4 @@ void ff_ref_fdct(DCTELEM *block);
 void ff_ref_idct(DCTELEM *block);
 void ff_ref_dct_init(void);
 
-#endif
+#endif /* AVCODEC_DCTREF_H */

libavcodec/dv_tablegen.h

@@ -20,8 +20,8 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef DV_TABLEGEN_H
-#define DV_TABLEGEN_H
+#ifndef AVCODEC_DV_TABLEGEN_H
+#define AVCODEC_DV_TABLEGEN_H
 
 #include <stdint.h>
 #include "dv_vlc_data.h"
@@ -93,4 +93,4 @@ static void dv_vlc_map_tableinit(void)
 }
 #endif /* CONFIG_HARDCODED_TABLES */
 
-#endif /* DV_TABLEGEN_H */
+#endif /* AVCODEC_DV_TABLEGEN_H */

libavcodec/flv.h

@@ -30,5 +30,4 @@ void ff_flv2_encode_ac_esc(PutBitContext *pb, int slevel, int level, int run, in
 int ff_flv_decode_picture_header(MpegEncContext *s);
 void ff_flv2_decode_ac_esc(GetBitContext *gb, int *level, int *run, int *last);
 
-#endif
-
+#endif /* AVCODEC_FLV_H */

libavcodec/gsmdec_data.h

@@ -19,8 +19,8 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef GSMDEC_DATA
-#define GSMDEC_DATA
+#ifndef AVCODEC_GSMDEC_DATA
+#define AVCODEC_GSMDEC_DATA
 
 #include <stdint.h>
 
@@ -44,4 +44,4 @@ typedef struct {
 extern const uint16_t ff_gsm_long_term_gain_tab[4];
 extern const int16_t ff_gsm_dequant_tab[64][8];
 
-#endif
+#endif /* AVCODEC_GSMDEC_DATA */

libavcodec/h263.h

@@ -248,4 +248,5 @@ static inline void memsetw(short *tab, int val, int n)
     for(i=0;i<n;i++)
         tab[i] = val;
 }
-#endif
+
+#endif /* AVCODEC_H263_H */

libavcodec/kbdwin.h

@@ -32,4 +32,4 @@
  */
 void ff_kbd_window_init(float *window, float alpha, int n);
 
-#endif
+#endif /* AVCODEC_KBDWIN_H */

libavcodec/motionpixels_tablegen.h

@@ -20,8 +20,8 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef MOTIONPIXELS_TABLEGEN_H
-#define MOTIONPIXELS_TABLEGEN_H
+#ifndef AVCODEC_MOTIONPIXELS_TABLEGEN_H
+#define AVCODEC_MOTIONPIXELS_TABLEGEN_H
 
 #include <stdint.h>
 
@@ -88,4 +88,4 @@ static void motionpixels_tableinit(void)
 }
 #endif /* CONFIG_HARDCODED_TABLES */
 
-#endif /* MOTIONPIXELS_TABLEGEN_H */
+#endif /* AVCODEC_MOTIONPIXELS_TABLEGEN_H */

libavcodec/mpeg4video.h

@@ -196,4 +196,4 @@ static inline int ff_mpeg4_pred_dc(MpegEncContext * s, int n, int level, int *di
 
     return ret;
 }
-#endif
+#endif /* AVCODEC_MPEG4VIDEO_H */

libavcodec/mpegaudio_tablegen.h

@@ -20,8 +20,8 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef MPEGAUDIO_TABLEGEN_H
-#define MPEGAUDIO_TABLEGEN_H
+#ifndef AVCODEC_MPEGAUDIO_TABLEGEN_H
+#define AVCODEC_MPEGAUDIO_TABLEGEN_H
 
 #include <stdint.h>
 #include <math.h>
@@ -68,4 +68,4 @@ static void mpegaudio_tableinit(void)
 }
 #endif /* CONFIG_HARDCODED_TABLES */
 
-#endif /* MPEGAUDIO_TABLEGEN_H */
+#endif /* AVCODEC_MPEGAUDIO_TABLEGEN_H */

libavcodec/mpegaudiodec.c

@@ -69,9 +69,9 @@ typedef struct MPADecodeContext {
     uint32_t free_format_next_header;
     GetBitContext gb;
     GetBitContext in_gb;
-    DECLARE_ALIGNED(16, MPA_INT, synth_buf)[MPA_MAX_CHANNELS][512 * 2];
+    DECLARE_ALIGNED(32, MPA_INT, synth_buf)[MPA_MAX_CHANNELS][512 * 2];
     int synth_buf_offset[MPA_MAX_CHANNELS];
-    DECLARE_ALIGNED(16, INTFLOAT, sb_samples)[MPA_MAX_CHANNELS][36][SBLIMIT];
+    DECLARE_ALIGNED(32, INTFLOAT, sb_samples)[MPA_MAX_CHANNELS][36][SBLIMIT];
     INTFLOAT mdct_buf[MPA_MAX_CHANNELS][SBLIMIT * 18]; /* previous samples, for layer 3 MDCT */
     GranuleDef granules[2][2]; /* Used in Layer 3 */
 #ifdef DEBUG

libavcodec/mpegaudiodsp.h

@@ -60,4 +60,4 @@ void ff_mpadsp_apply_window_fixed(int32_t *synth_buf, int32_t *window,
                                   int *dither_state, int16_t *samples,
                                   int incr);
 
-#endif
+#endif /* AVCODEC_MPEGAUDIODSP_H */

libavcodec/msgsmdec.h

@@ -19,12 +19,12 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef MSGSMDEC_H
-#define MSGSMDEC_H
+#ifndef AVCODEC_MSGSMDEC_H
+#define AVCODEC_MSGSMDEC_H
 
 #include "avcodec.h"
 
 int ff_msgsm_decode_block(AVCodecContext *avctx, int16_t *samples,
                           const uint8_t *buf);
 
-#endif
+#endif /* AVCODEC_MSGSMDEC_H */

libavcodec/nellymoser.h

@@ -54,4 +54,4 @@ extern const int16_t  ff_nelly_delta_table[32];
 
 void ff_nelly_get_sample_bits(const float *buf, int *bits);
 
-#endif
+#endif /* AVCODEC_NELLYMOSER_H */

libavcodec/opt.h

@@ -13,4 +13,4 @@
 #include "libavutil/opt.h"
 #endif
 
-#endif
+#endif /* AVCODEC_OPT_H */

libavcodec/pcm_tablegen.h

@@ -20,8 +20,8 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef PCM_TABLEGEN_H
-#define PCM_TABLEGEN_H
+#ifndef AVCODEC_PCM_TABLEGEN_H
+#define AVCODEC_PCM_TABLEGEN_H
 
 #include <stdint.h>
 #include "libavutil/attributes.h"
@@ -116,4 +116,4 @@ static void pcm_ulaw_tableinit(void)
 }
 #endif /* CONFIG_HARDCODED_TABLES */
 
-#endif /* PCM_TABLEGEN_H */
+#endif /* AVCODEC_PCM_TABLEGEN_H */

libavcodec/qdm2_tablegen.h

@@ -20,8 +20,8 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef QDM2_TABLEGEN_H
-#define QDM2_TABLEGEN_H
+#ifndef AVCODEC_QDM2_TABLEGEN_H
+#define AVCODEC_QDM2_TABLEGEN_H
 
 #include <stdint.h>
 #include <math.h>
@@ -99,4 +99,4 @@ static av_cold void init_noise_samples(void) {
 }
 #endif /* CONFIG_HARDCODED_TABLES */
 
-#endif /* QDM2_TABLEGEN_H */
+#endif /* AVCODEC_QDM2_TABLEGEN_H */

libavcodec/rdft.h

@@ -71,4 +71,4 @@ void ff_rdft_end(RDFTContext *s);
 void ff_rdft_init_arm(RDFTContext *s);
 
 
-#endif
+#endif /* AVCODEC_RDFT_H */

libavcodec/sh4/dsputil_sh4.h

@@ -25,4 +25,4 @@
 void idct_sh4(DCTELEM *block);
 void dsputil_init_align(DSPContext* c, AVCodecContext *avctx);
 
-#endif
+#endif /* AVCODEC_SH4_DSPUTIL_SH4_H */

libavcodec/sinewin.h

@@ -56,4 +56,4 @@ extern SINETABLE(4096);
 
 extern SINETABLE_CONST float * const ff_sine_windows[13];
 
-#endif
+#endif /* AVCODEC_SINEWIN_H */

libavcodec/sinewin_tablegen.h

@@ -20,6 +20,9 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
+#ifndef AVCODEC_SINEWIN_TABLEGEN_H
+#define AVCODEC_SINEWIN_TABLEGEN_H
+
 #include <assert.h>
 // do not use libavutil/libm.h since this is compiled both
 // for the host and the target and config.h is only valid for the target
@@ -58,3 +61,5 @@ av_cold void ff_init_ff_sine_windows(int index) {
     ff_sine_window_init(ff_sine_windows[index], 1 << index);
 #endif
 }
+
+#endif /* AVCODEC_SINEWIN_TABLEGEN_H */

libavcodec/sparc/dsputil_vis.h

@@ -26,4 +26,4 @@ void ff_simple_idct_put_vis(uint8_t *dest, int line_size, DCTELEM *data);
 void ff_simple_idct_add_vis(uint8_t *dest, int line_size, DCTELEM *data);
 void ff_simple_idct_vis(DCTELEM *data);
 
-#endif
+#endif /* AVCODEC_SPARC_DSPUTIL_VIS_H */

libavcodec/svq3.c

@@ -819,21 +819,12 @@ static av_cold int svq3_decode_init(AVCodecContext *avctx)
     avctx->pix_fmt = avctx->codec->pix_fmts[0];
 
     if (!s->context_initialized) {
-        s->width  = avctx->width;
-        s->height = avctx->height;
         h->chroma_qp[0] = h->chroma_qp[1] = 4;
 
         svq3->halfpel_flag = 1;
         svq3->thirdpel_flag = 1;
         svq3->unknown_flag = 0;
 
-        if (MPV_common_init(s) < 0)
-            return -1;
-
-        h->b_stride = 4*s->mb_width;
-
-        ff_h264_alloc_tables(h);
-
         /* prowl for the "SEQH" marker in the extradata */
         extradata = (unsigned char *)avctx->extradata;
         for (m = 0; m < avctx->extradata_size; m++) {
@@ -920,6 +911,16 @@ static av_cold int svq3_decode_init(AVCodecContext *avctx)
 #endif
             }
         }
+
+        s->width  = avctx->width;
+        s->height = avctx->height;
+
+        if (MPV_common_init(s) < 0)
+            return -1;
+
+        h->b_stride = 4*s->mb_width;
+
+        ff_h264_alloc_tables(h);
     }
 
     return 0;

libavcodec/targa.h

@@ -16,6 +16,9 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
+#ifndef AVCODEC_TARGA_H
+#define AVCODEC_TARGA_H
+
 /**
  * @file
  * targa file common definitions
@@ -34,3 +37,5 @@ enum TargaCompr {
     TGA_BW     = 3, // black & white or grayscale
     TGA_RLE    = 8, // flag pointing that data is RLE-coded
 };
+
+#endif /* AVCODEC_TARGA_H */

libavcodec/vp8.h

@@ -239,4 +239,4 @@ typedef struct {
     AVFrame frames[5];
 } VP8Context;
 
-#endif
+#endif /* AVCODEC_VP8_H */

libavcodec/vp8data.h

@@ -685,4 +685,4 @@ static const uint8_t vp8_mv_default_prob[2][19] = {
       128, 130, 130,  74, 148, 180, 203, 236, 254, 254 }
 };
 
-#endif
+#endif /* AVCODEC_VP8DATA_H */

libavcodec/x86/Makefile

@@ -1,6 +1,8 @@
 OBJS-$(CONFIG_MLP_DECODER)             += x86/mlpdsp.o
 OBJS-$(CONFIG_TRUEHD_DECODER)          += x86/mlpdsp.o
 
+YASM-OBJS-$(CONFIG_DCT)                += x86/dct32_sse.o
+
 YASM-OBJS-FFT-$(HAVE_AMD3DNOW)         += x86/fft_3dn.o
 YASM-OBJS-FFT-$(HAVE_AMD3DNOWEXT)      += x86/fft_3dn2.o
 YASM-OBJS-FFT-$(HAVE_SSE)              += x86/fft_sse.o
@@ -55,4 +57,3 @@ OBJS-$(HAVE_MMX)                       += x86/dnxhd_mmx.o               \
                                           x86/mpegvideo_mmx.o           \
                                           x86/simple_idct_mmx.o         \
 
-MMX-OBJS-$(CONFIG_DCT)                 += x86/dct32_sse.o

libavcodec/x86/dct32_sse.asm

@@ -0,0 +1,391 @@
+;******************************************************************************
+;* 32 point SSE-optimized DCT transform
+;* Copyright (c) 2010 Vitor Sessak
+;*
+;* This file is part of Libav.
+;*
+;* Libav 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.
+;*
+;* Libav 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 Libav; if not, write to the Free Software
+;* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+;******************************************************************************
+
+%include "x86inc.asm"
+%include "config.asm"
+
+SECTION_RODATA 32
+
+align 32
+ps_cos_vec: dd   0.500603,  0.505471,  0.515447,  0.531043
+            dd   0.553104,  0.582935,  0.622504,  0.674808
+            dd -10.190008, -3.407609, -2.057781, -1.484165
+            dd  -1.169440, -0.972568, -0.839350, -0.744536
+            dd   0.502419,  0.522499,  0.566944,  0.646822
+            dd   0.788155,  1.060678,  1.722447,  5.101149
+            dd   0.509796,  0.601345,  0.899976,  2.562916
+            dd   0.509796,  0.601345,  0.899976,  2.562916
+            dd   1.000000,  1.000000,  1.306563,  0.541196
+            dd   1.000000,  1.000000,  1.306563,  0.541196
+            dd   1.000000,  0.707107,  1.000000, -0.707107
+            dd   1.000000,  0.707107,  1.000000, -0.707107
+
+
+ps_p1p1m1m1: dd 0, 0, 0x80000000, 0x80000000, 0, 0, 0x80000000, 0x80000000
+
+%macro BUTTERFLY_SSE 4
+    movaps %4, %1
+    subps  %1, %2
+    addps  %2, %4
+    mulps  %1, %3
+%endmacro
+
+%macro BUTTERFLY_AVX 4
+    vsubps  %4, %1, %2
+    vaddps  %2, %2, %1
+    vmulps  %1, %4, %3
+%endmacro
+
+%macro BUTTERFLY0_SSE 5
+    movaps %4, %1
+    shufps %1, %1, %5
+    xorps  %4, %2
+    addps  %1, %4
+    mulps  %1, %3
+%endmacro
+
+%macro BUTTERFLY0_AVX 5
+    vshufps %4, %1, %1, %5
+    vxorps  %1, %1, %2
+    vaddps  %4, %4, %1
+    vmulps  %1, %4, %3
+%endmacro
+
+%macro BUTTERFLY2 4
+    BUTTERFLY0 %1, %2, %3, %4, 0x1b
+%endmacro
+
+%macro BUTTERFLY3 4
+    BUTTERFLY0 %1, %2, %3, %4, 0xb1
+%endmacro
+
+%macro PASS6_AND_PERMUTE 0
+    mov         tmpd, [outq+4]
+    movss         m7, [outq+72]
+    addss         m7, [outq+76]
+    movss         m3, [outq+56]
+    addss         m3, [outq+60]
+    addss         m4, m3
+    movss         m2, [outq+52]
+    addss         m2, m3
+    movss         m3, [outq+104]
+    addss         m3, [outq+108]
+    addss         m1, m3
+    addss         m5, m4
+    movss [outq+ 16], m1
+    movss         m1, [outq+100]
+    addss         m1, m3
+    movss         m3, [outq+40]
+    movss [outq+ 48], m1
+    addss         m3, [outq+44]
+    movss         m1, [outq+100]
+    addss         m4, m3
+    addss         m3, m2
+    addss         m1, [outq+108]
+    movss [outq+ 40], m3
+    addss         m2, [outq+36]
+    movss         m3, [outq+8]
+    movss [outq+ 56], m2
+    addss         m3, [outq+12]
+    movss [outq+ 32], m3
+    movss         m3, [outq+80]
+    movss [outq+  8], m5
+    movss [outq+ 80], m1
+    movss         m2, [outq+52]
+    movss         m5, [outq+120]
+    addss         m5, [outq+124]
+    movss         m1, [outq+64]
+    addss         m2, [outq+60]
+    addss         m0, m5
+    addss         m5, [outq+116]
+    mov    [outq+64], tmpd
+    addss         m6, m0
+    addss         m1, m6
+    mov         tmpd, [outq+12]
+    mov   [outq+ 96], tmpd
+    movss [outq+  4], m1
+    movss         m1, [outq+24]
+    movss [outq+ 24], m4
+    movss         m4, [outq+88]
+    addss         m4, [outq+92]
+    addss         m3, m4
+    addss         m4, [outq+84]
+    mov         tmpd, [outq+108]
+    addss         m1, [outq+28]
+    addss         m0, m1
+    addss         m1, m5
+    addss         m6, m3
+    addss         m3, m0
+    addss         m0, m7
+    addss         m5, [outq+20]
+    addss         m7, m1
+    movss [outq+ 12], m6
+    mov   [outq+112], tmpd
+    movss         m6, [outq+28]
+    movss [outq+ 28], m0
+    movss         m0, [outq+36]
+    movss [outq+ 36], m7
+    addss         m1, m4
+    movss         m7, [outq+116]
+    addss         m0, m2
+    addss         m7, [outq+124]
+    movss [outq+ 72], m0
+    movss         m0, [outq+44]
+    addss         m2, m0
+    movss [outq+ 44], m1
+    movss [outq+ 88], m2
+    addss         m0, [outq+60]
+    mov         tmpd, [outq+60]
+    mov   [outq+120], tmpd
+    movss [outq+104], m0
+    addss         m4, m5
+    addss         m5, [outq+68]
+    movss  [outq+52], m4
+    movss  [outq+60], m5
+    movss         m4, [outq+68]
+    movss         m5, [outq+20]
+    movss [outq+ 20], m3
+    addss         m5, m7
+    addss         m7, m6
+    addss         m4, m5
+    movss         m2, [outq+84]
+    addss         m2, [outq+92]
+    addss         m5, m2
+    movss [outq+ 68], m4
+    addss         m2, m7
+    movss         m4, [outq+76]
+    movss [outq+ 84], m2
+    movss [outq+ 76], m5
+    addss         m7, m4
+    addss         m6, [outq+124]
+    addss         m4, m6
+    addss         m6, [outq+92]
+    movss [outq+100], m4
+    movss [outq+108], m6
+    movss         m6, [outq+92]
+    movss  [outq+92], m7
+    addss         m6, [outq+124]
+    movss [outq+116], m6
+%endmacro
+
+%define BUTTERFLY  BUTTERFLY_AVX
+%define BUTTERFLY0 BUTTERFLY0_AVX
+
+INIT_YMM
+section .text align=16
+%ifdef HAVE_AVX
+; void ff_dct32_float_avx(FFTSample *out, const FFTSample *in)
+cglobal dct32_float_avx, 2,3,8, out, in, tmp
+    ; pass 1
+    vmovaps     m4, [inq+0]
+    vinsertf128 m5, m5, [inq+96], 1
+    vinsertf128 m5, m5, [inq+112], 0
+    vshufps     m5, m5, m5, 0x1b
+    BUTTERFLY   m4, m5, [ps_cos_vec], m6
+
+    vmovaps     m2, [inq+64]
+    vinsertf128 m6, m6, [inq+32], 1
+    vinsertf128 m6, m6, [inq+48], 0
+    vshufps     m6, m6, m6, 0x1b
+    BUTTERFLY   m2, m6, [ps_cos_vec+32], m0
+
+    ; pass 2
+
+    BUTTERFLY  m5, m6, [ps_cos_vec+64], m0
+    BUTTERFLY  m4, m2, [ps_cos_vec+64], m7
+
+
+    ; pass 3
+    vperm2f128  m3, m6, m4, 0x31
+    vperm2f128  m1, m6, m4, 0x20
+    vshufps     m3, m3, m3, 0x1b
+
+    BUTTERFLY   m1, m3, [ps_cos_vec+96], m6
+
+
+    vperm2f128  m4, m5, m2, 0x20
+    vperm2f128  m5, m5, m2, 0x31
+    vshufps     m5, m5, m5, 0x1b
+
+    BUTTERFLY   m4, m5, [ps_cos_vec+96], m6
+
+    ; pass 4
+    vmovaps m6, [ps_p1p1m1m1+0]
+    vmovaps m2, [ps_cos_vec+128]
+
+    BUTTERFLY2  m5, m6, m2, m7
+    BUTTERFLY2  m4, m6, m2, m7
+    BUTTERFLY2  m1, m6, m2, m7
+    BUTTERFLY2  m3, m6, m2, m7
+
+
+    ; pass 5
+    vshufps m6, m6, m6, 0xcc
+    vmovaps m2, [ps_cos_vec+160]
+
+    BUTTERFLY3  m5, m6, m2, m7
+    BUTTERFLY3  m4, m6, m2, m7
+    BUTTERFLY3  m1, m6, m2, m7
+    BUTTERFLY3  m3, m6, m2, m7
+
+    vperm2f128  m6, m3, m3, 0x31
+    vmovaps [outq], m3
+
+    vextractf128  [outq+64], m5, 1
+    vextractf128  [outq+32], m5, 0
+
+    vextractf128  [outq+80], m4, 1
+    vextractf128  [outq+48], m4, 0
+
+    vperm2f128  m0, m1, m1, 0x31
+    vmovaps [outq+96], m1
+
+    vzeroupper
+
+    ;    pass 6, no SIMD...
+INIT_XMM
+    PASS6_AND_PERMUTE
+    RET
+%endif
+
+%define BUTTERFLY  BUTTERFLY_SSE
+%define BUTTERFLY0 BUTTERFLY0_SSE
+
+INIT_XMM
+; void ff_dct32_float_sse(FFTSample *out, const FFTSample *in)
+cglobal dct32_float_sse, 2,3,8, out, in, tmp
+    ; pass 1
+
+    movaps      m0, [inq+0]
+    movaps      m1, [inq+112]
+    shufps      m1, m1, 0x1b
+    BUTTERFLY   m0, m1, [ps_cos_vec], m3
+
+    movaps      m7, [inq+64]
+    movaps      m4, [inq+48]
+    shufps      m4, m4, 0x1b
+    BUTTERFLY   m7, m4, [ps_cos_vec+32], m3
+
+    ; pass 2
+    movaps      m2, [ps_cos_vec+64]
+    BUTTERFLY   m1, m4, m2, m3
+    movaps      [outq+48], m1
+    movaps      [outq+ 0], m4
+
+    ; pass 1
+    movaps      m1, [inq+16]
+    movaps      m6, [inq+96]
+    shufps      m6, m6, 0x1b
+    BUTTERFLY   m1, m6, [ps_cos_vec+16], m3
+
+    movaps      m4, [inq+80]
+    movaps      m5, [inq+32]
+    shufps      m5, m5, 0x1b
+    BUTTERFLY   m4, m5, [ps_cos_vec+48], m3
+
+    ; pass 2
+    BUTTERFLY   m0, m7, m2, m3
+
+    movaps      m2, [ps_cos_vec+80]
+    BUTTERFLY   m6, m5, m2, m3
+
+    BUTTERFLY   m1, m4, m2, m3
+
+    ; pass 3
+    movaps      m2, [ps_cos_vec+96]
+    shufps      m1, m1, 0x1b
+    BUTTERFLY   m0, m1, m2, m3
+    movaps      [outq+112], m0
+    movaps      [outq+ 96], m1
+
+    movaps      m0, [outq+0]
+    shufps      m5, m5, 0x1b
+    BUTTERFLY   m0, m5, m2, m3
+
+    movaps      m1, [outq+48]
+    shufps      m6, m6, 0x1b
+    BUTTERFLY   m1, m6, m2, m3
+    movaps      [outq+48], m1
+
+    shufps      m4, m4, 0x1b
+    BUTTERFLY   m7, m4, m2, m3
+
+    ; pass 4
+    movaps      m3, [ps_p1p1m1m1+0]
+    movaps      m2, [ps_cos_vec+128]
+
+    BUTTERFLY2  m5, m3, m2, m1
+
+    BUTTERFLY2  m0, m3, m2, m1
+    movaps      [outq+16], m0
+
+    BUTTERFLY2  m6, m3, m2, m1
+    movaps      [outq+32], m6
+
+    movaps      m0, [outq+48]
+    BUTTERFLY2  m0, m3, m2, m1
+    movaps      [outq+48], m0
+
+    BUTTERFLY2  m4, m3, m2, m1
+
+    BUTTERFLY2  m7, m3, m2, m1
+
+    movaps      m6, [outq+96]
+    BUTTERFLY2  m6, m3, m2, m1
+
+    movaps      m0, [outq+112]
+    BUTTERFLY2  m0, m3, m2, m1
+
+    ; pass 5
+    movaps      m2, [ps_cos_vec+160]
+    shufps      m3, m3, 0xcc
+
+    BUTTERFLY3  m5, m3, m2, m1
+    movaps      [outq+0], m5
+
+    movaps      m1, [outq+16]
+    BUTTERFLY3  m1, m3, m2, m5
+    movaps      [outq+96], m1
+
+    BUTTERFLY3  m4, m3, m2, m5
+    movaps      [outq+64], m4
+
+    BUTTERFLY3  m7, m3, m2, m5
+    movaps      [outq+80], m7
+
+    movaps      m5, [outq+32]
+    BUTTERFLY3  m5, m3, m2, m7
+    movaps      [outq+32], m5
+
+    movaps      m4, [outq+48]
+    BUTTERFLY3  m4, m3, m2, m7
+    movaps      [outq+48], m4
+
+    BUTTERFLY3  m6, m3, m2, m7
+    movaps      [outq+16], m6
+
+    BUTTERFLY3  m0, m3, m2, m7
+    movaps      [outq+112], m0
+
+
+    ;    pass 6, no SIMD...
+    PASS6_AND_PERMUTE
+    RET

libavcodec/x86/dct32_sse.c

@@ -1,296 +0,0 @@
-/*
- * 32 point SSE-optimized DCT transform
- * Copyright (c) 2010 Vitor Sessak
- *
- * This file is part of FFmpeg.
- *
- * 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.
- *
- * 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
- */
-
-#include <stdint.h>
-
-#include "libavutil/x86_cpu.h"
-#include "libavutil/mem.h"
-#include "libavcodec/dsputil.h"
-#include "fft.h"
-
-DECLARE_ALIGNED(16, static const float, b1)[] = {
-     0.500603,  0.505471,  0.515447,  0.531043,
-     0.553104,  0.582935,  0.622504,  0.674808,
-    -1.169440, -0.972568, -0.839350, -0.744536,
-   -10.190008, -3.407609, -2.057781, -1.484165,
-     0.502419,  0.522499,  0.566944,  0.646822,
-     0.788155,  1.060678,  1.722447,  5.101149,
-     0.509796,  0.601345,  0.899976,  2.562916,
-     1.000000,  1.000000,  1.306563,  0.541196,
-     1.000000,  0.707107,  1.000000, -0.707107
-};
-
-DECLARE_ALIGNED(16, static const int32_t, smask)[4] = {
-    0, 0, 0x80000000, 0x80000000
-};
-
-/* butterfly operator */
-#define BUTTERFLY(a,b,c,tmp)                            \
-    "movaps  %%" #a    ", %%" #tmp  "             \n\t" \
-    "subps   %%" #b    ", %%" #a    "             \n\t" \
-    "addps   %%" #tmp  ", %%" #b    "             \n\t" \
-    "mulps     " #c    ", %%" #a    "             \n\t"
-
-///* Same as BUTTERFLY when vectors a and b overlap */
-#define BUTTERFLY0(val, mask, cos, tmp, shuf)                            \
-    "movaps  %%" #val  ", %%" #tmp  "             \n\t"                  \
-    "shufps    " #shuf ", %%" #val  ",%%" #val "  \n\t"                  \
-    "xorps   %%" #mask ", %%" #tmp  "             \n\t" /* flip signs */ \
-    "addps   %%" #tmp  ", %%" #val  "             \n\t"                  \
-    "mulps   %%" #cos  ", %%" #val  "             \n\t"
-
-#define BUTTERFLY2(val, mask, cos, tmp) BUTTERFLY0(val, mask, cos, tmp, $0x1b)
-#define BUTTERFLY3(val, mask, cos, tmp) BUTTERFLY0(val, mask, cos, tmp, $0xb1)
-
-void ff_dct32_float_sse(FFTSample *out, const FFTSample *in)
-{
-    int32_t tmp1 = 0;
-    __asm__ volatile(
-        /* pass 1 */
-
-        "movaps    (%4), %%xmm0           \n\t"
-        "movaps 112(%4), %%xmm1           \n\t"
-        "shufps   $0x1b, %%xmm1, %%xmm1   \n\t"
-        BUTTERFLY(xmm0, xmm1, (%2), xmm3)
-
-        "movaps  64(%4), %%xmm7           \n\t"
-        "movaps  48(%4), %%xmm4           \n\t"
-        "shufps   $0x1b, %%xmm4, %%xmm4   \n\t"
-        BUTTERFLY(xmm7, xmm4, 48(%2), xmm3)
-
-
-        /* pass 2 */
-        "movaps  64(%2), %%xmm2           \n\t"
-        BUTTERFLY(xmm1, xmm4, %%xmm2, xmm3)
-        "movaps  %%xmm1, 48(%1)           \n\t"
-        "movaps  %%xmm4, (%1)             \n\t"
-
-        /* pass 1 */
-        "movaps  16(%4), %%xmm1           \n\t"
-        "movaps  96(%4), %%xmm6           \n\t"
-        "shufps   $0x1b, %%xmm6, %%xmm6   \n\t"
-        BUTTERFLY(xmm1, xmm6, 16(%2), xmm3)
-
-        "movaps  80(%4), %%xmm4           \n\t"
-        "movaps  32(%4), %%xmm5           \n\t"
-        "shufps   $0x1b, %%xmm5, %%xmm5   \n\t"
-        BUTTERFLY(xmm4, xmm5, 32(%2), xmm3)
-
-        /* pass 2 */
-        BUTTERFLY(xmm0, xmm7, %%xmm2, xmm3)
-
-        "movaps  80(%2), %%xmm2           \n\t"
-        BUTTERFLY(xmm6, xmm5, %%xmm2, xmm3)
-
-        BUTTERFLY(xmm1, xmm4, %%xmm2, xmm3)
-
-        /* pass 3 */
-        "movaps  96(%2), %%xmm2           \n\t"
-        "shufps   $0x1b, %%xmm1, %%xmm1   \n\t"
-        BUTTERFLY(xmm0, xmm1, %%xmm2, xmm3)
-        "movaps  %%xmm0, 112(%1)          \n\t"
-        "movaps  %%xmm1,  96(%1)          \n\t"
-
-        "movaps   0(%1), %%xmm0           \n\t"
-        "shufps   $0x1b, %%xmm5, %%xmm5   \n\t"
-        BUTTERFLY(xmm0, xmm5, %%xmm2, xmm3)
-
-        "movaps  48(%1), %%xmm1           \n\t"
-        "shufps   $0x1b, %%xmm6, %%xmm6   \n\t"
-        BUTTERFLY(xmm1, xmm6, %%xmm2, xmm3)
-        "movaps  %%xmm1,  48(%1)          \n\t"
-
-        "shufps   $0x1b, %%xmm4, %%xmm4   \n\t"
-        BUTTERFLY(xmm7, xmm4, %%xmm2, xmm3)
-
-        /* pass 4 */
-        "movaps    (%3), %%xmm3           \n\t"
-        "movaps 112(%2), %%xmm2           \n\t"
-
-        BUTTERFLY2(xmm5, xmm3, xmm2, xmm1)
-
-        BUTTERFLY2(xmm0, xmm3, xmm2, xmm1)
-        "movaps  %%xmm0, 16(%1)           \n\t"
-
-        BUTTERFLY2(xmm6, xmm3, xmm2, xmm1)
-        "movaps  %%xmm6, 32(%1)           \n\t"
-
-        "movaps  48(%1), %%xmm0           \n\t"
-        BUTTERFLY2(xmm0, xmm3, xmm2, xmm1)
-        "movaps  %%xmm0, 48(%1)           \n\t"
-
-        BUTTERFLY2(xmm4, xmm3, xmm2, xmm1)
-
-        BUTTERFLY2(xmm7, xmm3, xmm2, xmm1)
-
-        "movaps  96(%1), %%xmm6           \n\t"
-        BUTTERFLY2(xmm6, xmm3, xmm2, xmm1)
-
-        "movaps 112(%1), %%xmm0           \n\t"
-        BUTTERFLY2(xmm0, xmm3, xmm2, xmm1)
-
-        /* pass 5 */
-        "movaps 128(%2), %%xmm2           \n\t"
-        "shufps   $0xCC, %%xmm3,%%xmm3    \n\t"
-
-        BUTTERFLY3(xmm5, xmm3, xmm2, xmm1)
-        "movaps  %%xmm5, (%1)             \n\t"
-
-        "movaps  16(%1), %%xmm1           \n\t"
-        BUTTERFLY3(xmm1, xmm3, xmm2, xmm5)
-        "movaps  %%xmm1, 16(%1)           \n\t"
-
-        BUTTERFLY3(xmm4, xmm3, xmm2, xmm5)
-        "movaps  %%xmm4, 64(%1)           \n\t"
-
-        BUTTERFLY3(xmm7, xmm3, xmm2, xmm5)
-        "movaps  %%xmm7, 80(%1)           \n\t"
-
-        "movaps  32(%1), %%xmm5           \n\t"
-        BUTTERFLY3(xmm5, xmm3, xmm2, xmm7)
-        "movaps  %%xmm5, 32(%1)           \n\t"
-
-        "movaps  48(%1), %%xmm4           \n\t"
-        BUTTERFLY3(xmm4, xmm3, xmm2, xmm7)
-        "movaps  %%xmm4, 48(%1)           \n\t"
-
-        BUTTERFLY3(xmm6, xmm3, xmm2, xmm7)
-        "movaps  %%xmm6, 96(%1)           \n\t"
-
-        BUTTERFLY3(xmm0, xmm3, xmm2, xmm7)
-        "movaps  %%xmm0, 112(%1)          \n\t"
-
-
-        /* pass 6, no SIMD... */
-        "movss    56(%1),  %%xmm3           \n\t"
-        "movl      4(%1),      %0           \n\t"
-        "addss    60(%1),  %%xmm3           \n\t"
-        "movss    72(%1),  %%xmm7           \n\t"
-        "addss    %%xmm3,  %%xmm4           \n\t"
-        "movss    52(%1),  %%xmm2           \n\t"
-        "addss    %%xmm3,  %%xmm2           \n\t"
-        "movss    24(%1),  %%xmm3           \n\t"
-        "addss    28(%1),  %%xmm3           \n\t"
-        "addss    76(%1),  %%xmm7           \n\t"
-        "addss    %%xmm3,  %%xmm1           \n\t"
-        "addss    %%xmm4,  %%xmm5           \n\t"
-        "movss    %%xmm1,  16(%1)           \n\t"
-        "movss    20(%1),  %%xmm1           \n\t"
-        "addss    %%xmm3,  %%xmm1           \n\t"
-        "movss    40(%1),  %%xmm3           \n\t"
-        "movss    %%xmm1,  48(%1)           \n\t"
-        "addss    44(%1),  %%xmm3           \n\t"
-        "movss    20(%1),  %%xmm1           \n\t"
-        "addss    %%xmm3,  %%xmm4           \n\t"
-        "addss    %%xmm2,  %%xmm3           \n\t"
-        "addss    28(%1),  %%xmm1           \n\t"
-        "movss    %%xmm3,  40(%1)           \n\t"
-        "addss    36(%1),  %%xmm2           \n\t"
-        "movss     8(%1),  %%xmm3           \n\t"
-        "movss    %%xmm2,  56(%1)           \n\t"
-        "addss    12(%1),  %%xmm3           \n\t"
-        "movss    %%xmm5,   8(%1)           \n\t"
-        "movss    %%xmm3,  32(%1)           \n\t"
-        "movss    52(%1),  %%xmm2           \n\t"
-        "movss    80(%1),  %%xmm3           \n\t"
-        "movss   120(%1),  %%xmm5           \n\t"
-        "movss    %%xmm1,  80(%1)           \n\t"
-        "movss    %%xmm4,  24(%1)           \n\t"
-        "addss   124(%1),  %%xmm5           \n\t"
-        "movss    64(%1),  %%xmm1           \n\t"
-        "addss    60(%1),  %%xmm2           \n\t"
-        "addss    %%xmm5,  %%xmm0           \n\t"
-        "addss   116(%1),  %%xmm5           \n\t"
-        "movl         %0,  64(%1)           \n\t"
-        "addss    %%xmm0,  %%xmm6           \n\t"
-        "addss    %%xmm6,  %%xmm1           \n\t"
-        "movl     12(%1),      %0           \n\t"
-        "movss    %%xmm1,   4(%1)           \n\t"
-        "movss    88(%1),  %%xmm1           \n\t"
-        "movl         %0,  96(%1)           \n\t"
-        "addss    92(%1),  %%xmm1           \n\t"
-        "movss   104(%1),  %%xmm4           \n\t"
-        "movl     28(%1),      %0           \n\t"
-        "addss   108(%1),  %%xmm4           \n\t"
-        "addss    %%xmm4,  %%xmm0           \n\t"
-        "addss    %%xmm1,  %%xmm3           \n\t"
-        "addss    84(%1),  %%xmm1           \n\t"
-        "addss    %%xmm5,  %%xmm4           \n\t"
-        "addss    %%xmm3,  %%xmm6           \n\t"
-        "addss    %%xmm0,  %%xmm3           \n\t"
-        "addss    %%xmm7,  %%xmm0           \n\t"
-        "addss   100(%1),  %%xmm5           \n\t"
-        "addss    %%xmm4,  %%xmm7           \n\t"
-        "movl         %0, 112(%1)           \n\t"
-        "movss    %%xmm0,  28(%1)           \n\t"
-        "movss    36(%1),  %%xmm0           \n\t"
-        "movss    %%xmm7,  36(%1)           \n\t"
-        "addss    %%xmm1,  %%xmm4           \n\t"
-        "movss   116(%1),  %%xmm7           \n\t"
-        "addss    %%xmm2,  %%xmm0           \n\t"
-        "addss   124(%1),  %%xmm7           \n\t"
-        "movss    %%xmm0,  72(%1)           \n\t"
-        "movss    44(%1),  %%xmm0           \n\t"
-        "movss    %%xmm6,  12(%1)           \n\t"
-        "movss    %%xmm3,  20(%1)           \n\t"
-        "addss    %%xmm0,  %%xmm2           \n\t"
-        "movss    %%xmm4,  44(%1)           \n\t"
-        "movss    %%xmm2,  88(%1)           \n\t"
-        "addss    60(%1),  %%xmm0           \n\t"
-        "movl     60(%1),      %0           \n\t"
-        "movl         %0, 120(%1)           \n\t"
-        "movss    %%xmm0, 104(%1)           \n\t"
-        "addss    %%xmm5,  %%xmm1           \n\t"
-        "addss    68(%1),  %%xmm5           \n\t"
-        "movss    %%xmm1,  52(%1)           \n\t"
-        "movss    %%xmm5,  60(%1)           \n\t"
-        "movss    68(%1),  %%xmm1           \n\t"
-        "movss   100(%1),  %%xmm5           \n\t"
-        "addss    %%xmm7,  %%xmm5           \n\t"
-        "addss   108(%1),  %%xmm7           \n\t"
-        "addss    %%xmm5,  %%xmm1           \n\t"
-        "movss    84(%1),  %%xmm2           \n\t"
-        "addss    92(%1),  %%xmm2           \n\t"
-        "addss    %%xmm2,  %%xmm5           \n\t"
-        "movss    %%xmm1,  68(%1)           \n\t"
-        "addss    %%xmm7,  %%xmm2           \n\t"
-        "movss    76(%1),  %%xmm1           \n\t"
-        "movss    %%xmm2,  84(%1)           \n\t"
-        "movss    %%xmm5,  76(%1)           \n\t"
-        "movss   108(%1),  %%xmm2           \n\t"
-        "addss    %%xmm1,  %%xmm7           \n\t"
-        "addss   124(%1),  %%xmm2           \n\t"
-        "addss    %%xmm2,  %%xmm1           \n\t"
-        "addss    92(%1),  %%xmm2           \n\t"
-        "movss    %%xmm1, 100(%1)           \n\t"
-        "movss    %%xmm2, 108(%1)           \n\t"
-        "movss    92(%1),  %%xmm2           \n\t"
-        "movss    %%xmm7,  92(%1)           \n\t"
-        "addss   124(%1),  %%xmm2           \n\t"
-        "movss    %%xmm2, 116(%1)           \n\t"
-        :"+&r"(tmp1)
-        :"r"(out), "r"(b1), "r"(smask), "r"(in)
-        :XMM_CLOBBERS("%xmm0", "%xmm1", "%xmm2", "%xmm3",
-                      "%xmm4", "%xmm5", "%xmm6", "%xmm7",)
-         "memory"
-        );
-}
-

libavcodec/x86/fft.c

@@ -57,7 +57,9 @@ av_cold void ff_fft_init_mmx(FFTContext *s)
 av_cold void ff_dct_init_mmx(DCTContext *s)
 {
     int has_vectors = av_get_cpu_flags();
-    if (has_vectors & AV_CPU_FLAG_SSE && HAVE_SSE)
+    if (has_vectors & AV_CPU_FLAG_AVX && HAVE_AVX)
+        s->dct32 = ff_dct32_float_avx;
+    else if (has_vectors & AV_CPU_FLAG_SSE && HAVE_SSE)
         s->dct32 = ff_dct32_float_sse;
 }
 #endif

libavcodec/x86/fft.h

@@ -35,5 +35,6 @@ void ff_imdct_calc_sse(FFTContext *s, FFTSample *output, const FFTSample *input)
 void ff_imdct_half_sse(FFTContext *s, FFTSample *output, const FFTSample *input);
 void ff_imdct_half_avx(FFTContext *s, FFTSample *output, const FFTSample *input);
 void ff_dct32_float_sse(FFTSample *out, const FFTSample *in);
+void ff_dct32_float_avx(FFTSample *out, const FFTSample *in);
 
-#endif
+#endif /* AVCODEC_X86_FFT_H */

libavcodec/xvmc_internal.h

@@ -18,8 +18,8 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef AVCODEC_INTERNAL_XVMC_H
-#define AVCODEC_INTERNAL_XVMC_H
+#ifndef AVCODEC_XVMC_INTERNAL_H
+#define AVCODEC_XVMC_INTERNAL_H
 
 #include "avcodec.h"
 #include "mpegvideo.h"
@@ -30,4 +30,4 @@ int  ff_xvmc_field_start(MpegEncContext*s, AVCodecContext *avctx);
 void ff_xvmc_field_end(MpegEncContext *s);
 void ff_xvmc_decode_mb(MpegEncContext *s);
 
-#endif /* AVCODEC_INTERNAL_XVMC_H */
+#endif /* AVCODEC_XVMC_INTERNAL_H */

libavfilter/avfilter.h

@@ -861,4 +861,4 @@ static inline void avfilter_insert_outpad(AVFilterContext *f, unsigned index,
                         &f->output_pads, &f->outputs, p);
 }
 
-#endif  /* AVFILTER_AVFILTER_H */
+#endif /* AVFILTER_AVFILTER_H */

libavfilter/avfiltergraph.h

@@ -120,4 +120,4 @@ int avfilter_graph_parse(AVFilterGraph *graph, const char *filters,
                          AVFilterInOut *inputs, AVFilterInOut *outputs,
                          AVClass *log_ctx);
 
-#endif  /* AVFILTER_AVFILTERGRAPH_H */
+#endif /* AVFILTER_AVFILTERGRAPH_H */

libavfilter/internal.h

@@ -58,4 +58,4 @@ int ff_avfilter_graph_config_formats(AVFilterGraph *graphctx, AVClass *log_ctx);
 /** default handler for freeing audio/video buffer when there are no references left */
 void ff_avfilter_default_free_buffer(AVFilterBuffer *buf);
 
-#endif  /* AVFILTER_INTERNAL_H */
+#endif /* AVFILTER_INTERNAL_H */

libavformat/avio.h

@@ -370,6 +370,7 @@ void avio_set_interrupt_cb(int (*interrupt_cb)(void));
  * freed with av_free().
  *
  * @param buffer Memory block for input/output operations via AVIOContext.
+ *        The buffer must be allocated with av_malloc() and friends.
  * @param buffer_size The buffer size is very important for performance.
  *        For protocols with fixed blocksize it should be set to this blocksize.
  *        For others a typical size is a cache page, e.g. 4kb.

libavformat/avio_internal.h

@@ -98,4 +98,4 @@ int ffio_open_dyn_packet_buf(AVIOContext **s, int max_packet_size);
  */
 int ffio_fdopen(AVIOContext **s, URLContext *h);
 
-#endif // AVFORMAT_AVIO_INTERNAL_H
+#endif /* AVFORMAT_AVIO_INTERNAL_H */

libavformat/aviobuf.c

@@ -113,6 +113,8 @@ AVIOContext *avio_alloc_context(
                   int64_t (*seek)(void *opaque, int64_t offset, int whence))
 {
     AVIOContext *s = av_mallocz(sizeof(AVIOContext));
+    if (!s)
+        return NULL;
     ffio_init_context(s, buffer, buffer_size, write_flag, opaque,
                   read_packet, write_packet, seek);
     return s;

libavformat/ffmeta.h

@@ -19,11 +19,11 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef AVFORMAT_META_H
-#define AVFORMAT_META_H
+#ifndef AVFORMAT_FFMETA_H
+#define AVFORMAT_FFMETA_H
 
 #define ID_STRING  ";FFMETADATA"
 #define ID_CHAPTER "[CHAPTER]"
 #define ID_STREAM  "[STREAM]"
 
-#endif /* AVFORMAT_META_H */
+#endif /* AVFORMAT_FFMETA_H */

libavformat/mms.h

@@ -60,4 +60,5 @@ typedef struct {
 int ff_mms_asf_header_parser(MMSContext * mms);
 int ff_mms_read_data(MMSContext *mms, uint8_t *buf, const int size);
 int ff_mms_read_header(MMSContext * mms, uint8_t * buf, const int size);
-#endif
+
+#endif /* AVFORMAT_MMS_H */

libavformat/spdif.h

@@ -19,6 +19,9 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
+#ifndef AVFORMAT_SPDIF_H
+#define AVFORMAT_SPDIF_H
+
 #include <stdint.h>
 
 #define SYNCWORD1 0xF872
@@ -55,3 +58,5 @@ static const uint16_t spdif_mpeg_pkt_offset[2][3] = {
 };
 
 void ff_spdif_bswap_buf16(uint16_t *dst, const uint16_t *src, int w);
+
+#endif /* AVFORMAT_SPDIF_H */

libavformat/url.h

@@ -173,4 +173,4 @@ int ffurl_register_protocol(URLProtocol *protocol, int size);
 int ff_udp_set_remote_url(URLContext *h, const char *uri);
 int ff_udp_get_local_port(URLContext *h);
 
-#endif //AVFORMAT_URL_H
+#endif /* AVFORMAT_URL_H */

libavformat/version.h

@@ -69,4 +69,4 @@
 #define FF_API_SDP_CREATE              (LIBAVFORMAT_VERSION_MAJOR < 54)
 #endif
 
-#endif //AVFORMAT_VERSION_H
+#endif /* AVFORMAT_VERSION_H */

libavutil/avassert.h

@@ -63,4 +63,4 @@
 #define av_assert2(cond) ((void)0)
 #endif
 
-#endif
+#endif /* AVUTIL_AVASSERT_H */

libavutil/cpu.h

@@ -51,4 +51,4 @@ int ff_get_cpu_flags_arm(void);
 int ff_get_cpu_flags_ppc(void);
 int ff_get_cpu_flags_x86(void);
 
-#endif  /* AVUTIL_CPU_H */
+#endif /* AVUTIL_CPU_H */

libswscale/swscale.h

@@ -187,6 +187,7 @@ void sws_freeContext(struct SwsContext *swsContext);
  * @return a pointer to an allocated context, or NULL in case of error
  * @note this function is to be removed after a saner alternative is
  *       written
+ * @deprecated Use sws_getCachedContext() instead.
  */
 struct SwsContext *sws_getContext(int srcW, int srcH, enum PixelFormat srcFormat,
                                   int dstW, int dstH, enum PixelFormat dstFormat,