Merge remote-tracking branch 'qatar/master'

* qatar/master:
  Revert "swscale: use 15-bit intermediates for 9/10-bit scaling."
  swscale: use 15-bit intermediates for 9/10-bit scaling.
  dct32: Add SSE2 ASM optimizations
  Correct chroma vector calculation for RealVideo 3.
  lavf: Add an option to discard corrupted frames
  mpegts: Mark wrongly-sized packets as corrupted
  mpegts: Move scan test to handle_packets
  mpegts: Mark corrupted packets
  mpegts: Reset continuity counter on seek
  mpegts: Fix for continuity counter
  mpegts: Silence "can't seek" warning on unseekable
  apichange: add an entry for AV_PKT_FLAG_CORRUPT
  avpacket: signal possibly corrupted packets
  mpeg4videodec: remove dead code that would have detected erroneous encoding
  aac: Remove some suspicious illegal memcpy()s from LTP.
  bink: Eliminate unnecessary shadow declaration.

Conflicts:
	doc/APIchanges
	libavcodec/version.h
	libavformat/avformat.h
	libavformat/options.c
	libavformat/utils.c

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

8 files changed, 41 insertions, 16 deletions

diff --git a/libavcodec/aacdec.c b/libavcodec/aacdec.c
index 7fce228193..927d4314fa 100644
--- a/libavcodec/aacdec.c
+++ b/libavcodec/aacdec.c
@@ -1791,12 +1791,10 @@ static void windowing_and_mdct_ltp(AACContext *ac, float *out,
     } else {
         memset(in, 0, 448 * sizeof(float));
         ac->dsp.vector_fmul(in + 448, in + 448, swindow_prev, 128);
-        memcpy(in + 576, in + 576, 448 * sizeof(float));
     }
     if (ics->window_sequence[0] != LONG_START_SEQUENCE) {
         ac->dsp.vector_fmul_reverse(in + 1024, in + 1024, lwindow, 1024);
     } else {
-        memcpy(in + 1024, in + 1024, 448 * sizeof(float));
         ac->dsp.vector_fmul_reverse(in + 1024 + 448, in + 1024 + 448, swindow, 128);
         memset(in + 1024 + 576, 0, 448 * sizeof(float));
     }
diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h
index 0bb86690bc..2a438ccc13 100644
--- a/libavcodec/avcodec.h
+++ b/libavcodec/avcodec.h
@@ -800,6 +800,9 @@ typedef struct AVPacket {
     uint8_t *data;
     int   size;
     int   stream_index;
+    /**
+     * A combination of AV_PKT_FLAG values
+     */
     int   flags;
     /**
      * Additional packet data that can be provided by the container.
@@ -840,7 +843,8 @@ typedef struct AVPacket {
      */
     int64_t convergence_duration;
 } AVPacket;
-#define AV_PKT_FLAG_KEY   0x0001
+#define AV_PKT_FLAG_KEY     0x0001 ///< The packet contains a keyframe
+#define AV_PKT_FLAG_CORRUPT 0x0002 ///< The packet content is corrupted
 
 /**
  * Audio Video Frame.
diff --git a/libavcodec/bink.c b/libavcodec/bink.c
index 6b58f7716b..e137312693 100644
--- a/libavcodec/bink.c
+++ b/libavcodec/bink.c
@@ -624,7 +624,6 @@ static int read_dct_coeffs(GetBitContext *gb, int32_t block[64], const uint8_t *
                         coef_list[--list_start] = ccoef;
                         mode_list[  list_start] = 3;
                     } else {
-                        int t;
                         if (!bits) {
                             t = 1 - (get_bits1(gb) << 1);
                         } else {
diff --git a/libavcodec/rv34.c b/libavcodec/rv34.c
index 2be01041fe..ee9dd7db6b 100644
--- a/libavcodec/rv34.c
+++ b/libavcodec/rv34.c
@@ -737,8 +737,8 @@ static inline void rv34_mc(RV34DecContext *r, const int block_type,
         my = (s->current_picture_ptr->f.motion_val[dir][mv_pos][1] + (3 << 24)) / 3 - (1 << 24);
         lx = (s->current_picture_ptr->f.motion_val[dir][mv_pos][0] + (3 << 24)) % 3;
         ly = (s->current_picture_ptr->f.motion_val[dir][mv_pos][1] + (3 << 24)) % 3;
-        chroma_mx = (s->current_picture_ptr->f.motion_val[dir][mv_pos][0] + 1) >> 1;
-        chroma_my = (s->current_picture_ptr->f.motion_val[dir][mv_pos][1] + 1) >> 1;
+        chroma_mx = s->current_picture_ptr->f.motion_val[dir][mv_pos][0] / 2;
+        chroma_my = s->current_picture_ptr->f.motion_val[dir][mv_pos][1] / 2;
         umx = (chroma_mx + (3 << 24)) / 3 - (1 << 24);
         umy = (chroma_my + (3 << 24)) / 3 - (1 << 24);
         uvmx = chroma_coeffs[(chroma_mx + (3 << 24)) % 3];
diff --git a/libavcodec/version.h b/libavcodec/version.h
index 69040a5195..b19e8946d8 100644
--- a/libavcodec/version.h
+++ b/libavcodec/version.h
@@ -22,7 +22,7 @@
 
 #define LIBAVCODEC_VERSION_MAJOR 53
 #define LIBAVCODEC_VERSION_MINOR  9
-#define LIBAVCODEC_VERSION_MICRO  0
+#define LIBAVCODEC_VERSION_MICRO  1
 
 #define LIBAVCODEC_VERSION_INT  AV_VERSION_INT(LIBAVCODEC_VERSION_MAJOR, \
                                                LIBAVCODEC_VERSION_MINOR, \
diff --git a/libavcodec/x86/dct32_sse.asm b/libavcodec/x86/dct32_sse.asm
index 7f5f815591..9a2a6ea88c 100644
--- a/libavcodec/x86/dct32_sse.asm
+++ b/libavcodec/x86/dct32_sse.asm
@@ -63,6 +63,13 @@ ps_p1p1m1m1: dd 0, 0, 0x80000000, 0x80000000, 0, 0, 0x80000000, 0x80000000
     mulps  %1, %3
 %endmacro
 
+%macro BUTTERFLY0_SSE2 5
+    pshufd %4, %1, %5
+    xorps  %1, %2
+    addps  %1, %4
+    mulps  %1, %3
+%endmacro
+
 %macro BUTTERFLY0_AVX 5
     vshufps %4, %1, %1, %5
     vxorps  %1, %1, %2
@@ -405,18 +412,17 @@ INIT_XMM
 
 
 INIT_XMM
+%macro DCT32_FUNC 1
 ; void ff_dct32_float_sse(FFTSample *out, const FFTSample *in)
-cglobal dct32_float_sse, 2,3,16, out, in, tmp
+cglobal dct32_float_%1, 2,3,16, out, in, tmp
     ; pass 1
 
     movaps      m0, [inq+0]
-    movaps      m1, [inq+112]
-    shufps      m1, m1, 0x1b
+    LOAD_INV    m1, [inq+112]
     BUTTERFLY   m0, m1, [ps_cos_vec], m3
 
     movaps      m7, [inq+64]
-    movaps      m4, [inq+48]
-    shufps      m4, m4, 0x1b
+    LOAD_INV    m4, [inq+48]
     BUTTERFLY   m7, m4, [ps_cos_vec+32], m3
 
     ; pass 2
@@ -427,13 +433,11 @@ cglobal dct32_float_sse, 2,3,16, out, in, tmp
 
     ; pass 1
     movaps      m1, [inq+16]
-    movaps      m6, [inq+96]
-    shufps      m6, m6, 0x1b
+    LOAD_INV    m6, [inq+96]
     BUTTERFLY   m1, m6, [ps_cos_vec+16], m3
 
     movaps      m4, [inq+80]
-    movaps      m5, [inq+32]
-    shufps      m5, m5, 0x1b
+    LOAD_INV    m5, [inq+32]
     BUTTERFLY   m4, m5, [ps_cos_vec+48], m3
 
     ; pass 2
@@ -492,3 +496,20 @@ cglobal dct32_float_sse, 2,3,16, out, in, tmp
     PASS5
     PASS6
     RET
+%endmacro
+
+%macro LOAD_INV_SSE 2
+    movaps      %1, %2
+    shufps      %1, %1, 0x1b
+%endmacro
+
+%define LOAD_INV LOAD_INV_SSE
+DCT32_FUNC sse
+
+%macro LOAD_INV_SSE2 2
+    pshufd      %1, %2, 0x1b
+%endmacro
+
+%define LOAD_INV LOAD_INV_SSE2
+%define BUTTERFLY0 BUTTERFLY0_SSE2
+DCT32_FUNC sse2
diff --git a/libavcodec/x86/fft.c b/libavcodec/x86/fft.c
index 18964d88ca..d2d157c2d3 100644
--- a/libavcodec/x86/fft.c
+++ b/libavcodec/x86/fft.c
@@ -60,6 +60,8 @@ av_cold void ff_dct_init_mmx(DCTContext *s)
     int has_vectors = av_get_cpu_flags();
     if (has_vectors & AV_CPU_FLAG_AVX && HAVE_AVX)
         s->dct32 = ff_dct32_float_avx;
+    else if (has_vectors & AV_CPU_FLAG_SSE2 && HAVE_SSE)
+        s->dct32 = ff_dct32_float_sse2;
     else if (has_vectors & AV_CPU_FLAG_SSE && HAVE_SSE)
         s->dct32 = ff_dct32_float_sse;
 #endif
diff --git a/libavcodec/x86/fft.h b/libavcodec/x86/fft.h
index 79064c27cd..7fdc858a50 100644
--- a/libavcodec/x86/fft.h
+++ b/libavcodec/x86/fft.h
@@ -35,6 +35,7 @@ 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_sse2(FFTSample *out, const FFTSample *in);
 void ff_dct32_float_avx(FFTSample *out, const FFTSample *in);
 
 #endif /* AVCODEC_X86_FFT_H */