From 53c20f17c78d1d8a0fc2505868f201e69ff59cc5 Mon Sep 17 00:00:00 2001 From: Vittorio Giovara Date: Fri, 28 Mar 2014 01:13:54 +0100 Subject: vp8: K&R formatting cosmetics Signed-off-by: Diego Biurrun --- libavcodec/vp8dsp.c | 566 ++++++++++++++++++++++++++++------------------------ 1 file changed, 302 insertions(+), 264 deletions(-) (limited to 'libavcodec/vp8dsp.c') diff --git a/libavcodec/vp8dsp.c b/libavcodec/vp8dsp.c index ad7c60309c..431c96ea77 100644 --- a/libavcodec/vp8dsp.c +++ b/libavcodec/vp8dsp.c @@ -24,9 +24,10 @@ * VP8 compatible video decoder */ +#include "libavutil/common.h" + #include "mathops.h" #include "vp8dsp.h" -#include "libavutil/common.h" // TODO: Maybe add dequant static void vp8_luma_dc_wht_c(int16_t block[4][4][16], int16_t dc[16]) @@ -34,26 +35,26 @@ static void vp8_luma_dc_wht_c(int16_t block[4][4][16], int16_t dc[16]) int i, t0, t1, t2, t3; for (i = 0; i < 4; i++) { - t0 = dc[0*4+i] + dc[3*4+i]; - t1 = dc[1*4+i] + dc[2*4+i]; - t2 = dc[1*4+i] - dc[2*4+i]; - t3 = dc[0*4+i] - dc[3*4+i]; - - dc[0*4+i] = t0 + t1; - dc[1*4+i] = t3 + t2; - dc[2*4+i] = t0 - t1; - dc[3*4+i] = t3 - t2; + t0 = dc[0 * 4 + i] + dc[3 * 4 + i]; + t1 = dc[1 * 4 + i] + dc[2 * 4 + i]; + t2 = dc[1 * 4 + i] - dc[2 * 4 + i]; + t3 = dc[0 * 4 + i] - dc[3 * 4 + i]; + + dc[0 * 4 + i] = t0 + t1; + dc[1 * 4 + i] = t3 + t2; + dc[2 * 4 + i] = t0 - t1; + dc[3 * 4 + i] = t3 - t2; } for (i = 0; i < 4; i++) { - t0 = dc[i*4+0] + dc[i*4+3] + 3; // rounding - t1 = dc[i*4+1] + dc[i*4+2]; - t2 = dc[i*4+1] - dc[i*4+2]; - t3 = dc[i*4+0] - dc[i*4+3] + 3; // rounding - dc[i*4+0] = 0; - dc[i*4+1] = 0; - dc[i*4+2] = 0; - dc[i*4+3] = 0; + t0 = dc[i * 4 + 0] + dc[i * 4 + 3] + 3; // rounding + t1 = dc[i * 4 + 1] + dc[i * 4 + 2]; + t2 = dc[i * 4 + 1] - dc[i * 4 + 2]; + t3 = dc[i * 4 + 0] - dc[i * 4 + 3] + 3; // rounding + dc[i * 4 + 0] = 0; + dc[i * 4 + 1] = 0; + dc[i * 4 + 2] = 0; + dc[i * 4 + 3] = 0; block[i][0][0] = (t0 + t1) >> 3; block[i][1][0] = (t3 + t2) >> 3; @@ -75,8 +76,8 @@ static void vp8_luma_dc_wht_dc_c(int16_t block[4][4][16], int16_t dc[16]) } } -#define MUL_20091(a) ((((a)*20091) >> 16) + (a)) -#define MUL_35468(a) (((a)*35468) >> 16) +#define MUL_20091(a) ((((a) * 20091) >> 16) + (a)) +#define MUL_35468(a) (((a) * 35468) >> 16) static void vp8_idct_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride) { @@ -84,32 +85,32 @@ static void vp8_idct_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride) int16_t tmp[16]; for (i = 0; i < 4; i++) { - t0 = block[0*4+i] + block[2*4+i]; - t1 = block[0*4+i] - block[2*4+i]; - t2 = MUL_35468(block[1*4+i]) - MUL_20091(block[3*4+i]); - t3 = MUL_20091(block[1*4+i]) + MUL_35468(block[3*4+i]); - block[0*4+i] = 0; - block[1*4+i] = 0; - block[2*4+i] = 0; - block[3*4+i] = 0; - - tmp[i*4+0] = t0 + t3; - tmp[i*4+1] = t1 + t2; - tmp[i*4+2] = t1 - t2; - tmp[i*4+3] = t0 - t3; + t0 = block[0 * 4 + i] + block[2 * 4 + i]; + t1 = block[0 * 4 + i] - block[2 * 4 + i]; + t2 = MUL_35468(block[1 * 4 + i]) - MUL_20091(block[3 * 4 + i]); + t3 = MUL_20091(block[1 * 4 + i]) + MUL_35468(block[3 * 4 + i]); + block[0 * 4 + i] = 0; + block[1 * 4 + i] = 0; + block[2 * 4 + i] = 0; + block[3 * 4 + i] = 0; + + tmp[i * 4 + 0] = t0 + t3; + tmp[i * 4 + 1] = t1 + t2; + tmp[i * 4 + 2] = t1 - t2; + tmp[i * 4 + 3] = t0 - t3; } for (i = 0; i < 4; i++) { - t0 = tmp[0*4+i] + tmp[2*4+i]; - t1 = tmp[0*4+i] - tmp[2*4+i]; - t2 = MUL_35468(tmp[1*4+i]) - MUL_20091(tmp[3*4+i]); - t3 = MUL_20091(tmp[1*4+i]) + MUL_35468(tmp[3*4+i]); + t0 = tmp[0 * 4 + i] + tmp[2 * 4 + i]; + t1 = tmp[0 * 4 + i] - tmp[2 * 4 + i]; + t2 = MUL_35468(tmp[1 * 4 + i]) - MUL_20091(tmp[3 * 4 + i]); + t3 = MUL_20091(tmp[1 * 4 + i]) + MUL_35468(tmp[3 * 4 + i]); dst[0] = av_clip_uint8(dst[0] + ((t0 + t3 + 4) >> 3)); dst[1] = av_clip_uint8(dst[1] + ((t1 + t2 + 4) >> 3)); dst[2] = av_clip_uint8(dst[2] + ((t1 - t2 + 4) >> 3)); dst[3] = av_clip_uint8(dst[3] + ((t0 - t3 + 4) >> 3)); - dst += stride; + dst += stride; } } @@ -123,46 +124,49 @@ static void vp8_idct_dc_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride) dst[1] = av_clip_uint8(dst[1] + dc); dst[2] = av_clip_uint8(dst[2] + dc); dst[3] = av_clip_uint8(dst[3] + dc); - dst += stride; + dst += stride; } } -static void vp8_idct_dc_add4uv_c(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride) +static void vp8_idct_dc_add4uv_c(uint8_t *dst, int16_t block[4][16], + ptrdiff_t stride) { - vp8_idct_dc_add_c(dst+stride*0+0, block[0], stride); - vp8_idct_dc_add_c(dst+stride*0+4, block[1], stride); - vp8_idct_dc_add_c(dst+stride*4+0, block[2], stride); - vp8_idct_dc_add_c(dst+stride*4+4, block[3], stride); + vp8_idct_dc_add_c(dst + stride * 0 + 0, block[0], stride); + vp8_idct_dc_add_c(dst + stride * 0 + 4, block[1], stride); + vp8_idct_dc_add_c(dst + stride * 4 + 0, block[2], stride); + vp8_idct_dc_add_c(dst + stride * 4 + 4, block[3], stride); } -static void vp8_idct_dc_add4y_c(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride) +static void vp8_idct_dc_add4y_c(uint8_t *dst, int16_t block[4][16], + ptrdiff_t stride) { - vp8_idct_dc_add_c(dst+ 0, block[0], stride); - vp8_idct_dc_add_c(dst+ 4, block[1], stride); - vp8_idct_dc_add_c(dst+ 8, block[2], stride); - vp8_idct_dc_add_c(dst+12, block[3], stride); + vp8_idct_dc_add_c(dst + 0, block[0], stride); + vp8_idct_dc_add_c(dst + 4, block[1], stride); + vp8_idct_dc_add_c(dst + 8, block[2], stride); + vp8_idct_dc_add_c(dst + 12, block[3], stride); } // because I like only having two parameters to pass functions... -#define LOAD_PIXELS\ - int av_unused p3 = p[-4*stride];\ - int av_unused p2 = p[-3*stride];\ - int av_unused p1 = p[-2*stride];\ - int av_unused p0 = p[-1*stride];\ - int av_unused q0 = p[ 0*stride];\ - int av_unused q1 = p[ 1*stride];\ - int av_unused q2 = p[ 2*stride];\ - int av_unused q3 = p[ 3*stride]; - -#define clip_int8(n) (cm[n+0x80]-0x80) - -static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride, int is4tap) +#define LOAD_PIXELS \ + int av_unused p3 = p[-4 * stride]; \ + int av_unused p2 = p[-3 * stride]; \ + int av_unused p1 = p[-2 * stride]; \ + int av_unused p0 = p[-1 * stride]; \ + int av_unused q0 = p[ 0 * stride]; \ + int av_unused q1 = p[ 1 * stride]; \ + int av_unused q2 = p[ 2 * stride]; \ + int av_unused q3 = p[ 3 * stride]; + +#define clip_int8(n) (cm[n + 0x80] - 0x80) + +static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride, + int is4tap) { LOAD_PIXELS int a, f1, f2; const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; - a = 3*(q0 - p0); + a = 3 * (q0 - p0); if (is4tap) a += clip_int8(p1 - q1); @@ -171,45 +175,50 @@ static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride, int is4 // We deviate from the spec here with c(a+3) >> 3 // since that's what libvpx does. - f1 = FFMIN(a+4, 127) >> 3; - f2 = FFMIN(a+3, 127) >> 3; + f1 = FFMIN(a + 4, 127) >> 3; + f2 = FFMIN(a + 3, 127) >> 3; // Despite what the spec says, we do need to clamp here to // be bitexact with libvpx. - p[-1*stride] = cm[p0 + f2]; - p[ 0*stride] = cm[q0 - f1]; + p[-1 * stride] = cm[p0 + f2]; + p[ 0 * stride] = cm[q0 - f1]; // only used for _inner on blocks without high edge variance if (!is4tap) { - a = (f1+1)>>1; - p[-2*stride] = cm[p1 + a]; - p[ 1*stride] = cm[q1 - a]; + a = (f1 + 1) >> 1; + p[-2 * stride] = cm[p1 + a]; + p[ 1 * stride] = cm[q1 - a]; } } static av_always_inline int simple_limit(uint8_t *p, ptrdiff_t stride, int flim) { LOAD_PIXELS - return 2*FFABS(p0-q0) + (FFABS(p1-q1) >> 1) <= flim; + return 2 * FFABS(p0 - q0) + (FFABS(p1 - q1) >> 1) <= flim; } /** * E - limit at the macroblock edge * I - limit for interior difference */ -static av_always_inline int normal_limit(uint8_t *p, ptrdiff_t stride, int E, int I) +static av_always_inline int normal_limit(uint8_t *p, ptrdiff_t stride, + int E, int I) { LOAD_PIXELS - return simple_limit(p, stride, E) - && FFABS(p3-p2) <= I && FFABS(p2-p1) <= I && FFABS(p1-p0) <= I - && FFABS(q3-q2) <= I && FFABS(q2-q1) <= I && FFABS(q1-q0) <= I; + return simple_limit(p, stride, E) && + FFABS(p3 - p2) <= I && + FFABS(p2 - p1) <= I && + FFABS(p1 - p0) <= I && + FFABS(q3 - q2) <= I && + FFABS(q2 - q1) <= I && + FFABS(q1 - q0) <= I; } // high edge variance static av_always_inline int hev(uint8_t *p, ptrdiff_t stride, int thresh) { LOAD_PIXELS - return FFABS(p1-p0) > thresh || FFABS(q1-q0) > thresh; + return FFABS(p1 - p0) > thresh || FFABS(q1 - q0) > thresh; } static av_always_inline void filter_mbedge(uint8_t *p, ptrdiff_t stride) @@ -219,67 +228,75 @@ static av_always_inline void filter_mbedge(uint8_t *p, ptrdiff_t stride) LOAD_PIXELS - w = clip_int8(p1-q1); - w = clip_int8(w + 3*(q0-p0)); + w = clip_int8(p1 - q1); + w = clip_int8(w + 3 * (q0 - p0)); - a0 = (27*w + 63) >> 7; - a1 = (18*w + 63) >> 7; - a2 = ( 9*w + 63) >> 7; + a0 = (27 * w + 63) >> 7; + a1 = (18 * w + 63) >> 7; + a2 = (9 * w + 63) >> 7; - p[-3*stride] = cm[p2 + a2]; - p[-2*stride] = cm[p1 + a1]; - p[-1*stride] = cm[p0 + a0]; - p[ 0*stride] = cm[q0 - a0]; - p[ 1*stride] = cm[q1 - a1]; - p[ 2*stride] = cm[q2 - a2]; + p[-3 * stride] = cm[p2 + a2]; + p[-2 * stride] = cm[p1 + a1]; + p[-1 * stride] = cm[p0 + a0]; + p[ 0 * stride] = cm[q0 - a0]; + p[ 1 * stride] = cm[q1 - a1]; + p[ 2 * stride] = cm[q2 - a2]; } -#define LOOP_FILTER(dir, size, stridea, strideb, maybe_inline) \ -static maybe_inline void vp8_ ## dir ## _loop_filter ## size ## _c(uint8_t *dst, ptrdiff_t stride,\ - int flim_E, int flim_I, int hev_thresh)\ -{\ - int i;\ -\ - for (i = 0; i < size; i++)\ - if (normal_limit(dst+i*stridea, strideb, flim_E, flim_I)) {\ - if (hev(dst+i*stridea, strideb, hev_thresh))\ - filter_common(dst+i*stridea, strideb, 1);\ - else\ - filter_mbedge(dst+i*stridea, strideb);\ - }\ -}\ -\ -static maybe_inline void vp8_ ## dir ## _loop_filter ## size ## _inner_c(uint8_t *dst, ptrdiff_t stride,\ - int flim_E, int flim_I, int hev_thresh)\ -{\ - int i;\ -\ - for (i = 0; i < size; i++)\ - if (normal_limit(dst+i*stridea, strideb, flim_E, flim_I)) {\ - int hv = hev(dst+i*stridea, strideb, hev_thresh);\ - if (hv) \ - filter_common(dst+i*stridea, strideb, 1);\ - else \ - filter_common(dst+i*stridea, strideb, 0);\ - }\ +#define LOOP_FILTER(dir, size, stridea, strideb, maybe_inline) \ +static maybe_inline \ +void vp8_ ## dir ## _loop_filter ## size ## _c(uint8_t *dst, \ + ptrdiff_t stride, \ + int flim_E, int flim_I, \ + int hev_thresh) \ +{ \ + int i; \ + for (i = 0; i < size; i++) \ + if (normal_limit(dst + i * stridea, strideb, flim_E, flim_I)) { \ + if (hev(dst + i * stridea, strideb, hev_thresh)) \ + filter_common(dst + i * stridea, strideb, 1); \ + else \ + filter_mbedge(dst + i * stridea, strideb); \ + } \ +} \ + \ +static maybe_inline \ +void vp8_ ## dir ## _loop_filter ## size ## _inner_c(uint8_t *dst, \ + ptrdiff_t stride, \ + int flim_E, int flim_I, \ + int hev_thresh) \ +{ \ + int i; \ + for (i = 0; i < size; i++) \ + if (normal_limit(dst + i * stridea, strideb, flim_E, flim_I)) { \ + int hv = hev(dst + i * stridea, strideb, hev_thresh); \ + if (hv) \ + filter_common(dst + i * stridea, strideb, 1); \ + else \ + filter_common(dst + i * stridea, strideb, 0); \ + } \ } -LOOP_FILTER(v, 16, 1, stride,) -LOOP_FILTER(h, 16, stride, 1,) - -#define UV_LOOP_FILTER(dir, stridea, strideb) \ -LOOP_FILTER(dir, 8, stridea, strideb, av_always_inline) \ -static void vp8_ ## dir ## _loop_filter8uv_c(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,\ - int fE, int fI, int hev_thresh)\ -{\ - vp8_ ## dir ## _loop_filter8_c(dstU, stride, fE, fI, hev_thresh);\ - vp8_ ## dir ## _loop_filter8_c(dstV, stride, fE, fI, hev_thresh);\ -}\ -static void vp8_ ## dir ## _loop_filter8uv_inner_c(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,\ - int fE, int fI, int hev_thresh)\ -{\ - vp8_ ## dir ## _loop_filter8_inner_c(dstU, stride, fE, fI, hev_thresh);\ - vp8_ ## dir ## _loop_filter8_inner_c(dstV, stride, fE, fI, hev_thresh);\ +LOOP_FILTER(v, 16, 1, stride, ) +LOOP_FILTER(h, 16, stride, 1, ) + +#define UV_LOOP_FILTER(dir, stridea, strideb) \ +LOOP_FILTER(dir, 8, stridea, strideb, av_always_inline) \ +static void vp8_ ## dir ## _loop_filter8uv_c(uint8_t *dstU, uint8_t *dstV, \ + ptrdiff_t stride, int fE, \ + int fI, int hev_thresh) \ +{ \ + vp8_ ## dir ## _loop_filter8_c(dstU, stride, fE, fI, hev_thresh); \ + vp8_ ## dir ## _loop_filter8_c(dstV, stride, fE, fI, hev_thresh); \ +} \ + \ +static void vp8_ ## dir ## _loop_filter8uv_inner_c(uint8_t *dstU, \ + uint8_t *dstV, \ + ptrdiff_t stride, int fE, \ + int fI, int hev_thresh) \ +{ \ + vp8_ ## dir ## _loop_filter8_inner_c(dstU, stride, fE, fI, hev_thresh); \ + vp8_ ## dir ## _loop_filter8_inner_c(dstV, stride, fE, fI, hev_thresh); \ } UV_LOOP_FILTER(v, 1, stride) @@ -290,8 +307,8 @@ static void vp8_v_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, int flim) int i; for (i = 0; i < 16; i++) - if (simple_limit(dst+i, stride, flim)) - filter_common(dst+i, stride, 1); + if (simple_limit(dst + i, stride, flim)) + filter_common(dst + i, stride, 1); } static void vp8_h_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, int flim) @@ -299,94 +316,110 @@ static void vp8_h_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, int flim) int i; for (i = 0; i < 16; i++) - if (simple_limit(dst+i*stride, 1, flim)) - filter_common(dst+i*stride, 1, 1); + if (simple_limit(dst + i * stride, 1, flim)) + filter_common(dst + i * stride, 1, 1); } static const uint8_t subpel_filters[7][6] = { - { 0, 6, 123, 12, 1, 0 }, - { 2, 11, 108, 36, 8, 1 }, - { 0, 9, 93, 50, 6, 0 }, - { 3, 16, 77, 77, 16, 3 }, - { 0, 6, 50, 93, 9, 0 }, - { 1, 8, 36, 108, 11, 2 }, - { 0, 1, 12, 123, 6, 0 }, + { 0, 6, 123, 12, 1, 0 }, + { 2, 11, 108, 36, 8, 1 }, + { 0, 9, 93, 50, 6, 0 }, + { 3, 16, 77, 77, 16, 3 }, + { 0, 6, 50, 93, 9, 0 }, + { 1, 8, 36, 108, 11, 2 }, + { 0, 1, 12, 123, 6, 0 }, }; -#define PUT_PIXELS(WIDTH) \ -static void put_vp8_pixels ## WIDTH ##_c(uint8_t *dst, ptrdiff_t dststride, uint8_t *src, ptrdiff_t srcstride, int h, int x, int y) { \ - int i; \ - for (i = 0; i < h; i++, dst+= dststride, src+= srcstride) { \ - memcpy(dst, src, WIDTH); \ - } \ +#define PUT_PIXELS(WIDTH) \ +static void put_vp8_pixels ## WIDTH ## _c(uint8_t *dst, ptrdiff_t dststride, \ + uint8_t *src, ptrdiff_t srcstride, \ + int h, int x, int y) \ +{ \ + int i; \ + for (i = 0; i < h; i++, dst += dststride, src += srcstride) \ + memcpy(dst, src, WIDTH); \ } PUT_PIXELS(16) PUT_PIXELS(8) PUT_PIXELS(4) -#define FILTER_6TAP(src, F, stride) \ - cm[(F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + F[0]*src[x-2*stride] + \ - F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + F[5]*src[x+3*stride] + 64) >> 7] - -#define FILTER_4TAP(src, F, stride) \ - cm[(F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + \ - F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + 64) >> 7] - -#define VP8_EPEL_H(SIZE, TAPS) \ -static void put_vp8_epel ## SIZE ## _h ## TAPS ## _c(uint8_t *dst, ptrdiff_t dststride, uint8_t *src, ptrdiff_t srcstride, int h, int mx, int my) \ -{ \ - const uint8_t *filter = subpel_filters[mx-1]; \ - const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ - int x, y; \ -\ - for (y = 0; y < h; y++) { \ - for (x = 0; x < SIZE; x++) \ - dst[x] = FILTER_ ## TAPS ## TAP(src, filter, 1); \ - dst += dststride; \ - src += srcstride; \ - } \ +#define FILTER_6TAP(src, F, stride) \ + cm[(F[2] * src[x + 0 * stride] - F[1] * src[x - 1 * stride] + \ + F[0] * src[x - 2 * stride] + F[3] * src[x + 1 * stride] - \ + F[4] * src[x + 2 * stride] + F[5] * src[x + 3 * stride] + 64) >> 7] + +#define FILTER_4TAP(src, F, stride) \ + cm[(F[2] * src[x + 0 * stride] - F[1] * src[x - 1 * stride] + \ + F[3] * src[x + 1 * stride] - F[4] * src[x + 2 * stride] + 64) >> 7] + +#define VP8_EPEL_H(SIZE, TAPS) \ +static void put_vp8_epel ## SIZE ## _h ## TAPS ## _c(uint8_t *dst, \ + ptrdiff_t dststride, \ + uint8_t *src, \ + ptrdiff_t srcstride, \ + int h, int mx, int my) \ +{ \ + const uint8_t *filter = subpel_filters[mx - 1]; \ + const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ + int x, y; \ + for (y = 0; y < h; y++) { \ + for (x = 0; x < SIZE; x++) \ + dst[x] = FILTER_ ## TAPS ## TAP(src, filter, 1); \ + dst += dststride; \ + src += srcstride; \ + } \ } -#define VP8_EPEL_V(SIZE, TAPS) \ -static void put_vp8_epel ## SIZE ## _v ## TAPS ## _c(uint8_t *dst, ptrdiff_t dststride, uint8_t *src, ptrdiff_t srcstride, int h, int mx, int my) \ -{ \ - const uint8_t *filter = subpel_filters[my-1]; \ - const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ - int x, y; \ -\ - for (y = 0; y < h; y++) { \ - for (x = 0; x < SIZE; x++) \ - dst[x] = FILTER_ ## TAPS ## TAP(src, filter, srcstride); \ - dst += dststride; \ - src += srcstride; \ - } \ + +#define VP8_EPEL_V(SIZE, TAPS) \ +static void put_vp8_epel ## SIZE ## _v ## TAPS ## _c(uint8_t *dst, \ + ptrdiff_t dststride, \ + uint8_t *src, \ + ptrdiff_t srcstride, \ + int h, int mx, int my) \ +{ \ + const uint8_t *filter = subpel_filters[my - 1]; \ + const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ + int x, y; \ + for (y = 0; y < h; y++) { \ + for (x = 0; x < SIZE; x++) \ + dst[x] = FILTER_ ## TAPS ## TAP(src, filter, srcstride); \ + dst += dststride; \ + src += srcstride; \ + } \ } -#define VP8_EPEL_HV(SIZE, HTAPS, VTAPS) \ -static void put_vp8_epel ## SIZE ## _h ## HTAPS ## v ## VTAPS ## _c(uint8_t *dst, ptrdiff_t dststride, uint8_t *src, ptrdiff_t srcstride, int h, int mx, int my) \ -{ \ - const uint8_t *filter = subpel_filters[mx-1]; \ - const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ - int x, y; \ - uint8_t tmp_array[(2*SIZE+VTAPS-1)*SIZE]; \ - uint8_t *tmp = tmp_array; \ - src -= (2-(VTAPS==4))*srcstride; \ -\ - for (y = 0; y < h+VTAPS-1; y++) { \ - for (x = 0; x < SIZE; x++) \ - tmp[x] = FILTER_ ## HTAPS ## TAP(src, filter, 1); \ - tmp += SIZE; \ - src += srcstride; \ - } \ -\ - tmp = tmp_array + (2-(VTAPS==4))*SIZE; \ - filter = subpel_filters[my-1]; \ -\ - for (y = 0; y < h; y++) { \ - for (x = 0; x < SIZE; x++) \ - dst[x] = FILTER_ ## VTAPS ## TAP(tmp, filter, SIZE); \ - dst += dststride; \ - tmp += SIZE; \ - } \ + +#define VP8_EPEL_HV(SIZE, HTAPS, VTAPS) \ +static void \ +put_vp8_epel ## SIZE ## _h ## HTAPS ## v ## VTAPS ## _c(uint8_t *dst, \ + ptrdiff_t dststride, \ + uint8_t *src, \ + ptrdiff_t srcstride, \ + int h, int mx, \ + int my) \ +{ \ + const uint8_t *filter = subpel_filters[mx - 1]; \ + const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP; \ + int x, y; \ + uint8_t tmp_array[(2 * SIZE + VTAPS - 1) * SIZE]; \ + uint8_t *tmp = tmp_array; \ + src -= (2 - (VTAPS == 4)) * srcstride; \ + \ + for (y = 0; y < h + VTAPS - 1; y++) { \ + for (x = 0; x < SIZE; x++) \ + tmp[x] = FILTER_ ## HTAPS ## TAP(src, filter, 1); \ + tmp += SIZE; \ + src += srcstride; \ + } \ + tmp = tmp_array + (2 - (VTAPS == 4)) * SIZE; \ + filter = subpel_filters[my - 1]; \ + \ + for (y = 0; y < h; y++) { \ + for (x = 0; x < SIZE; x++) \ + dst[x] = FILTER_ ## VTAPS ## TAP(tmp, filter, SIZE); \ + dst += dststride; \ + tmp += SIZE; \ + } \ } VP8_EPEL_H(16, 4) @@ -401,6 +434,7 @@ VP8_EPEL_V(4, 4) VP8_EPEL_V(16, 6) VP8_EPEL_V(8, 6) VP8_EPEL_V(4, 6) + VP8_EPEL_HV(16, 4, 4) VP8_EPEL_HV(8, 4, 4) VP8_EPEL_HV(4, 4, 4) @@ -414,73 +448,77 @@ VP8_EPEL_HV(16, 6, 6) VP8_EPEL_HV(8, 6, 6) VP8_EPEL_HV(4, 6, 6) -#define VP8_BILINEAR(SIZE) \ -static void put_vp8_bilinear ## SIZE ## _h_c(uint8_t *dst, ptrdiff_t dstride, uint8_t *src, ptrdiff_t sstride, int h, int mx, int my) \ -{ \ - int a = 8-mx, b = mx; \ - int x, y; \ -\ - for (y = 0; y < h; y++) { \ - for (x = 0; x < SIZE; x++) \ - dst[x] = (a*src[x] + b*src[x+1] + 4) >> 3; \ - dst += dstride; \ - src += sstride; \ - } \ -} \ -static void put_vp8_bilinear ## SIZE ## _v_c(uint8_t *dst, ptrdiff_t dstride, uint8_t *src, ptrdiff_t sstride, int h, int mx, int my) \ -{ \ - int c = 8-my, d = my; \ - int x, y; \ -\ - for (y = 0; y < h; y++) { \ - for (x = 0; x < SIZE; x++) \ - dst[x] = (c*src[x] + d*src[x+sstride] + 4) >> 3; \ - dst += dstride; \ - src += sstride; \ - } \ -} \ -\ -static void put_vp8_bilinear ## SIZE ## _hv_c(uint8_t *dst, ptrdiff_t dstride, uint8_t *src, ptrdiff_t sstride, int h, int mx, int my) \ -{ \ - int a = 8-mx, b = mx; \ - int c = 8-my, d = my; \ - int x, y; \ - uint8_t tmp_array[(2*SIZE+1)*SIZE]; \ - uint8_t *tmp = tmp_array; \ -\ - for (y = 0; y < h+1; y++) { \ - for (x = 0; x < SIZE; x++) \ - tmp[x] = (a*src[x] + b*src[x+1] + 4) >> 3; \ - tmp += SIZE; \ - src += sstride; \ - } \ -\ - tmp = tmp_array; \ -\ - for (y = 0; y < h; y++) { \ - for (x = 0; x < SIZE; x++) \ - dst[x] = (c*tmp[x] + d*tmp[x+SIZE] + 4) >> 3; \ - dst += dstride; \ - tmp += SIZE; \ - } \ +#define VP8_BILINEAR(SIZE) \ +static void put_vp8_bilinear ## SIZE ## _h_c(uint8_t *dst, ptrdiff_t dstride, \ + uint8_t *src, ptrdiff_t sstride, \ + int h, int mx, int my) \ +{ \ + int a = 8 - mx, b = mx; \ + int x, y; \ + for (y = 0; y < h; y++) { \ + for (x = 0; x < SIZE; x++) \ + dst[x] = (a * src[x] + b * src[x + 1] + 4) >> 3; \ + dst += dstride; \ + src += sstride; \ + } \ +} \ + \ +static void put_vp8_bilinear ## SIZE ## _v_c(uint8_t *dst, ptrdiff_t dstride, \ + uint8_t *src, ptrdiff_t sstride, \ + int h, int mx, int my) \ +{ \ + int c = 8 - my, d = my; \ + int x, y; \ + for (y = 0; y < h; y++) { \ + for (x = 0; x < SIZE; x++) \ + dst[x] = (c * src[x] + d * src[x + sstride] + 4) >> 3; \ + dst += dstride; \ + src += sstride; \ + } \ +} \ + \ +static void put_vp8_bilinear ## SIZE ## _hv_c(uint8_t *dst, \ + ptrdiff_t dstride, \ + uint8_t *src, \ + ptrdiff_t sstride, \ + int h, int mx, int my) \ +{ \ + int a = 8 - mx, b = mx; \ + int c = 8 - my, d = my; \ + int x, y; \ + uint8_t tmp_array[(2 * SIZE + 1) * SIZE]; \ + uint8_t *tmp = tmp_array; \ + for (y = 0; y < h + 1; y++) { \ + for (x = 0; x < SIZE; x++) \ + tmp[x] = (a * src[x] + b * src[x + 1] + 4) >> 3; \ + tmp += SIZE; \ + src += sstride; \ + } \ + tmp = tmp_array; \ + for (y = 0; y < h; y++) { \ + for (x = 0; x < SIZE; x++) \ + dst[x] = (c * tmp[x] + d * tmp[x + SIZE] + 4) >> 3; \ + dst += dstride; \ + tmp += SIZE; \ + } \ } VP8_BILINEAR(16) VP8_BILINEAR(8) VP8_BILINEAR(4) -#define VP8_MC_FUNC(IDX, SIZE) \ - dsp->put_vp8_epel_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \ - dsp->put_vp8_epel_pixels_tab[IDX][0][1] = put_vp8_epel ## SIZE ## _h4_c; \ - dsp->put_vp8_epel_pixels_tab[IDX][0][2] = put_vp8_epel ## SIZE ## _h6_c; \ - dsp->put_vp8_epel_pixels_tab[IDX][1][0] = put_vp8_epel ## SIZE ## _v4_c; \ +#define VP8_MC_FUNC(IDX, SIZE) \ + dsp->put_vp8_epel_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \ + dsp->put_vp8_epel_pixels_tab[IDX][0][1] = put_vp8_epel ## SIZE ## _h4_c; \ + dsp->put_vp8_epel_pixels_tab[IDX][0][2] = put_vp8_epel ## SIZE ## _h6_c; \ + dsp->put_vp8_epel_pixels_tab[IDX][1][0] = put_vp8_epel ## SIZE ## _v4_c; \ dsp->put_vp8_epel_pixels_tab[IDX][1][1] = put_vp8_epel ## SIZE ## _h4v4_c; \ dsp->put_vp8_epel_pixels_tab[IDX][1][2] = put_vp8_epel ## SIZE ## _h6v4_c; \ - dsp->put_vp8_epel_pixels_tab[IDX][2][0] = put_vp8_epel ## SIZE ## _v6_c; \ + dsp->put_vp8_epel_pixels_tab[IDX][2][0] = put_vp8_epel ## SIZE ## _v6_c; \ dsp->put_vp8_epel_pixels_tab[IDX][2][1] = put_vp8_epel ## SIZE ## _h4v6_c; \ dsp->put_vp8_epel_pixels_tab[IDX][2][2] = put_vp8_epel ## SIZE ## _h6v6_c -#define VP8_BILINEAR_MC_FUNC(IDX, SIZE) \ +#define VP8_BILINEAR_MC_FUNC(IDX, SIZE) \ dsp->put_vp8_bilinear_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \ dsp->put_vp8_bilinear_pixels_tab[IDX][0][1] = put_vp8_bilinear ## SIZE ## _h_c; \ dsp->put_vp8_bilinear_pixels_tab[IDX][0][2] = put_vp8_bilinear ## SIZE ## _h_c; \ -- cgit v1.2.3