Remove duplicate #includes, avcodec.h #includes common.h.

[ffmpeg] / libavcodec / i386 / h264dsp_mmx.c

diff --git a/libavcodec/i386/h264dsp_mmx.c b/libavcodec/i386/h264dsp_mmx.c
index 904d8ea5fb1374efcd3dd194dc492248ae123c21..e5cf57970217e3f7b882cc5c14489708759f506a 100644 (file)
--- a/libavcodec/i386/h264dsp_mmx.c
+++ b/libavcodec/i386/h264dsp_mmx.c
@@ -48,12 +48,6 @@
     SUMSUBD2_AB( s13, d13, t )\
     SUMSUB_BADC( d13, s02, s13, d02 )
 
-#define TRANSPOSE4(a,b,c,d,t)\
-    SBUTTERFLY(a,b,t,wd) /* a=aebf t=cgdh */\
-    SBUTTERFLY(c,d,b,wd) /* c=imjn b=kolp */\
-    SBUTTERFLY(a,c,d,dq) /* a=aeim d=bfjn */\
-    SBUTTERFLY(t,b,c,dq) /* t=cgko c=dhlp */
-
 #define STORE_DIFF_4P( p, t, z ) \
     "psraw      $6,     "#p" \n\t"\
     "movd       (%0),   "#t" \n\t"\
@@ -176,7 +170,7 @@ static void ff_h264_idct8_add_mmx(uint8_t *dst, int16_t *block, int stride)
     block[0] += 32;
 
     for(i=0; i<2; i++){
-        uint64_t tmp;
+        DECLARE_ALIGNED_8(uint64_t, tmp);
 
         h264_idct8_1d(block+4*i);
 
@@ -370,7 +364,7 @@ static void ff_h264_idct8_dc_add_mmx2(uint8_t *dst, int16_t *block, int stride)
         "paddusb %%mm6              , %%mm2 \n\t"
 
 // in: mm0=p1 mm1=p0 mm2=q0 mm3=q1 mm7=(tc&mask) %8=mm_bone
-// out: (q1addr) = clip( (q2+((p0+q0+1)>>1))>>1, q1-tc0, q1+tc0 )
+// out: (q1addr) = av_clip( (q2+((p0+q0+1)>>1))>>1, q1-tc0, q1+tc0 )
 // clobbers: q2, tmp, tc0
 #define H264_DEBLOCK_Q1(p1, q2, q2addr, q1addr, tc0, tmp)\
         "movq     %%mm1,  "#tmp"   \n\t"\
@@ -388,7 +382,7 @@ static void ff_h264_idct8_dc_add_mmx2(uint8_t *dst, int16_t *block, int stride)
 
 static inline void h264_loop_filter_luma_mmx2(uint8_t *pix, int stride, int alpha1, int beta1, int8_t *tc0)
 {
-    uint64_t tmp0[2];
+    DECLARE_ALIGNED_8(uint64_t, tmp0[2]);
 
     asm volatile(
         "movq    (%1,%3), %%mm0    \n\t" //p1
@@ -450,7 +444,7 @@ static void h264_h_loop_filter_luma_mmx2(uint8_t *pix, int stride, int alpha, in
 {
     //FIXME: could cut some load/stores by merging transpose with filter
     // also, it only needs to transpose 6x8
-    uint8_t trans[8*8];
+    DECLARE_ALIGNED_8(uint8_t, trans[8*8]);
     int i;
     for(i=0; i<2; i++, pix+=8*stride, tc0+=2) {
         if((tc0[0] & tc0[1]) < 0)
@@ -494,7 +488,7 @@ static void h264_v_loop_filter_chroma_mmx2(uint8_t *pix, int stride, int alpha,
 static void h264_h_loop_filter_chroma_mmx2(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
 {
     //FIXME: could cut some load/stores by merging transpose with filter
-    uint8_t trans[8*4];
+    DECLARE_ALIGNED_8(uint8_t, trans[8*4]);
     transpose4x4(trans, pix-2, 8, stride);
     transpose4x4(trans+4, pix-2+4*stride, 8, stride);
     h264_loop_filter_chroma_mmx2(trans+2*8, 8, alpha-1, beta-1, tc0);
@@ -544,7 +538,7 @@ static void h264_v_loop_filter_chroma_intra_mmx2(uint8_t *pix, int stride, int a
 static void h264_h_loop_filter_chroma_intra_mmx2(uint8_t *pix, int stride, int alpha, int beta)
 {
     //FIXME: could cut some load/stores by merging transpose with filter
-    uint8_t trans[8*4];
+    DECLARE_ALIGNED_8(uint8_t, trans[8*4]);
     transpose4x4(trans, pix-2, 8, stride);
     transpose4x4(trans+4, pix-2+4*stride, 8, stride);
     h264_loop_filter_chroma_intra_mmx2(trans+2*8, 8, alpha-1, beta-1);
@@ -567,7 +561,7 @@ static void h264_loop_filter_strength_mmx2( int16_t bS[2][4][4], uint8_t nnz[40]
     for( dir=1; dir>=0; dir-- ) {
         const int d_idx = dir ? -8 : -1;
         const int mask_mv = dir ? mask_mv1 : mask_mv0;
-        const uint64_t mask_dir = dir ? 0 : 0xffffffffffffffffULL;
+        DECLARE_ALIGNED_8(const uint64_t, mask_dir) = dir ? 0 : 0xffffffffffffffffULL;
         int b_idx, edge, l;
         for( b_idx=12, edge=0; edge<edges; edge+=step, b_idx+=8*step ) {
             asm volatile(
@@ -684,7 +678,7 @@ static void h264_loop_filter_strength_mmx2( int16_t bS[2][4][4], uint8_t nnz[40]
         "movq %%mm6, "#OF"(%1)      \n\t"
 
 #define QPEL_H264(OPNAME, OP, MMX)\
-static void OPNAME ## h264_qpel4_h_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+static av_noinline void OPNAME ## h264_qpel4_h_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     int h=4;\
 \
     asm volatile(\
@@ -724,7 +718,7 @@ static void OPNAME ## h264_qpel4_h_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, i
         : "memory"\
     );\
 }\
-static void OPNAME ## h264_qpel4_h_lowpass_l2_ ## MMX(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride){\
+static av_noinline void OPNAME ## h264_qpel4_h_lowpass_l2_ ## MMX(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride){\
     int h=4;\
     asm volatile(\
         "pxor %%mm7, %%mm7          \n\t"\
@@ -768,7 +762,7 @@ static void OPNAME ## h264_qpel4_h_lowpass_l2_ ## MMX(uint8_t *dst, uint8_t *src
     );\
     }while(--h);\
 }\
-static void OPNAME ## h264_qpel4_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+static av_noinline void OPNAME ## h264_qpel4_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     src -= 2*srcStride;\
     asm volatile(\
         "pxor %%mm7, %%mm7          \n\t"\
@@ -797,7 +791,7 @@ static void OPNAME ## h264_qpel4_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, i
         : "memory"\
     );\
 }\
-static void OPNAME ## h264_qpel4_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
+static av_noinline void OPNAME ## h264_qpel4_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
     int h=4;\
     int w=3;\
     src -= 2*srcStride+2;\
@@ -861,7 +855,7 @@ static void OPNAME ## h264_qpel4_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp,
     );\
 }\
 \
-static void OPNAME ## h264_qpel8_h_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+static av_noinline void OPNAME ## h264_qpel8_h_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     int h=8;\
     asm volatile(\
         "pxor %%mm7, %%mm7          \n\t"\
@@ -918,7 +912,7 @@ static void OPNAME ## h264_qpel8_h_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, i
     );\
 }\
 \
-static void OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride){\
+static av_noinline void OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride){\
     int h=8;\
     asm volatile(\
         "pxor %%mm7, %%mm7          \n\t"\
@@ -981,7 +975,7 @@ static void OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(uint8_t *dst, uint8_t *src
     }while(--h);\
 }\
 \
-static inline void OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
+static av_noinline void OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
     int w= 2;\
     src -= 2*srcStride;\
     \
@@ -1036,7 +1030,7 @@ static inline void OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(uint8_t *dst, uint
      dst += 4-h*dstStride;\
    }\
 }\
-static inline void OPNAME ## h264_qpel8or16_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride, int size){\
+static av_noinline void OPNAME ## h264_qpel8or16_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride, int size){\
     int h = size;\
     int w = (size+8)>>2;\
     src -= 2*srcStride+2;\
@@ -1141,12 +1135,12 @@ static inline void OPNAME ## h264_qpel8or16_hv_lowpass_ ## MMX(uint8_t *dst, int
 static void OPNAME ## h264_qpel8_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(dst  , src  , dstStride, srcStride, 8);\
 }\
-static void OPNAME ## h264_qpel16_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+static av_noinline void OPNAME ## h264_qpel16_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(dst  , src  , dstStride, srcStride, 16);\
     OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(dst+8, src+8, dstStride, srcStride, 16);\
 }\
 \
-static void OPNAME ## h264_qpel16_h_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+static av_noinline void OPNAME ## h264_qpel16_h_lowpass_ ## MMX(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst  , src  , dstStride, srcStride);\
     OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst+8, src+8, dstStride, srcStride);\
     src += 8*srcStride;\
@@ -1155,7 +1149,7 @@ static void OPNAME ## h264_qpel16_h_lowpass_ ## MMX(uint8_t *dst, uint8_t *src,
     OPNAME ## h264_qpel8_h_lowpass_ ## MMX(dst+8, src+8, dstStride, srcStride);\
 }\
 \
-static void OPNAME ## h264_qpel16_h_lowpass_l2_ ## MMX(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride){\
+static av_noinline void OPNAME ## h264_qpel16_h_lowpass_l2_ ## MMX(uint8_t *dst, uint8_t *src, uint8_t *src2, int dstStride, int src2Stride){\
     OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(dst  , src  , src2  , dstStride, src2Stride);\
     OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(dst+8, src+8, src2+8, dstStride, src2Stride);\
     src += 8*dstStride;\
@@ -1173,7 +1167,7 @@ static void OPNAME ## h264_qpel16_hv_lowpass_ ## MMX(uint8_t *dst, int16_t *tmp,
     OPNAME ## h264_qpel8or16_hv_lowpass_ ## MMX(dst  , tmp  , src  , dstStride, tmpStride, srcStride, 16);\
 }\
 \
-static void OPNAME ## pixels4_l2_shift5_ ## MMX(uint8_t *dst, int16_t *src16, uint8_t *src8, int dstStride, int src8Stride, int h)\
+static av_noinline void OPNAME ## pixels4_l2_shift5_ ## MMX(uint8_t *dst, int16_t *src16, uint8_t *src8, int dstStride, int src8Stride, int h)\
 {\
     asm volatile(\
         "movq       %5,  %%mm6          \n\t"\
@@ -1207,7 +1201,7 @@ static void OPNAME ## pixels4_l2_shift5_ ## MMX(uint8_t *dst, int16_t *src16, ui
         :"S"((long)src8Stride), "D"((long)dstStride), "m"(ff_pw_16)\
         :"memory");\
 }\
-static void OPNAME ## pixels8_l2_shift5_ ## MMX(uint8_t *dst, int16_t *src16, uint8_t *src8, int dstStride, int src8Stride, int h)\
+static av_noinline void OPNAME ## pixels8_l2_shift5_ ## MMX(uint8_t *dst, int16_t *src16, uint8_t *src8, int dstStride, int src8Stride, int h)\
 {\
     asm volatile(\
         "movq       %0,  %%mm6          \n\t"\
@@ -1311,6 +1305,7 @@ static void OPNAME ## h264_qpel ## SIZE ## _mc21_ ## MMX(uint8_t *dst, uint8_t *
     uint64_t temp[SIZE*(SIZE<8?12:24)/4 + SIZE*SIZE/8];\
     uint8_t * const halfHV= (uint8_t*)temp;\
     int16_t * const tmp= ((int16_t*)temp) + SIZE*SIZE/2;\
+    assert((int)temp & 7 == 0);\
     put_h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(halfHV, tmp, src, SIZE, SIZE, stride);\
     OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src, halfHV, stride, SIZE);\
 }\
@@ -1319,6 +1314,7 @@ static void OPNAME ## h264_qpel ## SIZE ## _mc23_ ## MMX(uint8_t *dst, uint8_t *
     uint64_t temp[SIZE*(SIZE<8?12:24)/4 + SIZE*SIZE/8];\
     uint8_t * const halfHV= (uint8_t*)temp;\
     int16_t * const tmp= ((int16_t*)temp) + SIZE*SIZE/2;\
+    assert((int)temp & 7 == 0);\
     put_h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(halfHV, tmp, src, SIZE, SIZE, stride);\
     OPNAME ## h264_qpel ## SIZE ## _h_lowpass_l2_ ## MMX(dst, src+stride, halfHV, stride, SIZE);\
 }\
@@ -1327,6 +1323,7 @@ static void OPNAME ## h264_qpel ## SIZE ## _mc12_ ## MMX(uint8_t *dst, uint8_t *
     uint64_t temp[SIZE*(SIZE<8?12:24)/4 + SIZE*SIZE/8];\
     int16_t * const halfV= ((int16_t*)temp) + SIZE*SIZE/2;\
     uint8_t * const halfHV= ((uint8_t*)temp);\
+    assert((int)temp & 7 == 0);\
     put_h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(halfHV, halfV, src, SIZE, SIZE, stride);\
     OPNAME ## pixels ## SIZE ## _l2_shift5_ ## MMX(dst, halfV+2, halfHV, stride, SIZE, SIZE);\
 }\
@@ -1335,6 +1332,7 @@ static void OPNAME ## h264_qpel ## SIZE ## _mc32_ ## MMX(uint8_t *dst, uint8_t *
     uint64_t temp[SIZE*(SIZE<8?12:24)/4 + SIZE*SIZE/8];\
     int16_t * const halfV= ((int16_t*)temp) + SIZE*SIZE/2;\
     uint8_t * const halfHV= ((uint8_t*)temp);\
+    assert((int)temp & 7 == 0);\
     put_h264_qpel ## SIZE ## _hv_lowpass_ ## MMX(halfHV, halfV, src, SIZE, SIZE, stride);\
     OPNAME ## pixels ## SIZE ## _l2_shift5_ ## MMX(dst, halfV+3, halfHV, stride, SIZE, SIZE);\
 }\