From: Diego Biurrun Date: Wed, 18 Dec 2013 22:27:35 +0000 (+0100) Subject: dsputil: Move RV40-specific bits into rv40dsp X-Git-Tag: v11_alpha1~809 X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=da5be235250a61d6994408b054e3e3acf2e0f90f;p=platform%2Fupstream%2Flibav.git dsputil: Move RV40-specific bits into rv40dsp --- diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c index 4a2011a..843c244 100644 --- a/libavcodec/dsputil.c +++ b/libavcodec/dsputil.c @@ -1250,28 +1250,6 @@ static void wmv2_mspel8_h_lowpass(uint8_t *dst, uint8_t *src, } } -#if CONFIG_RV40_DECODER -void ff_put_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) -{ - put_pixels16_xy2_8_c(dst, src, stride, 16); -} - -void ff_avg_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) -{ - avg_pixels16_xy2_8_c(dst, src, stride, 16); -} - -void ff_put_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) -{ - put_pixels8_xy2_8_c(dst, src, stride, 8); -} - -void ff_avg_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) -{ - avg_pixels8_xy2_8_c(dst, src, stride, 8); -} -#endif /* CONFIG_RV40_DECODER */ - static void wmv2_mspel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int w) { diff --git a/libavcodec/dsputil.h b/libavcodec/dsputil.h index 052ac50..7f467d1 100644 --- a/libavcodec/dsputil.h +++ b/libavcodec/dsputil.h @@ -44,12 +44,6 @@ void ff_avg_pixels8x8_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); void ff_put_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); void ff_avg_pixels16x16_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); -/* RV40 functions */ -void ff_put_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); -void ff_avg_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); -void ff_put_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); -void ff_avg_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride); - void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy, int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height); diff --git a/libavcodec/dsputil_template.c b/libavcodec/dsputil_template.c index 7d8ae15..8085078 100644 --- a/libavcodec/dsputil_template.c +++ b/libavcodec/dsputil_template.c @@ -253,58 +253,6 @@ static inline void OPNAME ## _no_rnd_pixels16_l4_8(uint8_t *dst, \ src_stride2, src_stride3, \ src_stride4, h); \ } \ - \ -static inline void OPNAME ## _pixels8_xy2_8_c(uint8_t *block, \ - const uint8_t *pixels, \ - ptrdiff_t line_size, \ - int h) \ -{ \ - /* FIXME HIGH BIT DEPTH */ \ - int j; \ - \ - for (j = 0; j < 2; j++) { \ - int i; \ - const uint32_t a = AV_RN32(pixels); \ - const uint32_t b = AV_RN32(pixels + 1); \ - uint32_t l0 = (a & 0x03030303UL) + \ - (b & 0x03030303UL) + \ - 0x02020202UL; \ - uint32_t h0 = ((a & 0xFCFCFCFCUL) >> 2) + \ - ((b & 0xFCFCFCFCUL) >> 2); \ - uint32_t l1, h1; \ - \ - pixels += line_size; \ - for (i = 0; i < h; i += 2) { \ - uint32_t a = AV_RN32(pixels); \ - uint32_t b = AV_RN32(pixels + 1); \ - l1 = (a & 0x03030303UL) + \ - (b & 0x03030303UL); \ - h1 = ((a & 0xFCFCFCFCUL) >> 2) + \ - ((b & 0xFCFCFCFCUL) >> 2); \ - OP(*((uint32_t *) block), \ - h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL)); \ - pixels += line_size; \ - block += line_size; \ - a = AV_RN32(pixels); \ - b = AV_RN32(pixels + 1); \ - l0 = (a & 0x03030303UL) + \ - (b & 0x03030303UL) + \ - 0x02020202UL; \ - h0 = ((a & 0xFCFCFCFCUL) >> 2) + \ - ((b & 0xFCFCFCFCUL) >> 2); \ - OP(*((uint32_t *) block), \ - h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL)); \ - pixels += line_size; \ - block += line_size; \ - } \ - pixels += 4 - line_size * (h + 1); \ - block += 4 - line_size * h; \ - } \ -} \ - \ -CALL_2X_PIXELS(OPNAME ## _pixels16_xy2_8_c, \ - OPNAME ## _pixels8_xy2_8_c, \ - 8) \ #define op_avg(a, b) a = rnd_avg32(a, b) #define op_put(a, b) a = b diff --git a/libavcodec/rv40dsp.c b/libavcodec/rv40dsp.c index 122d14d..cca0294 100644 --- a/libavcodec/rv40dsp.c +++ b/libavcodec/rv40dsp.c @@ -24,11 +24,14 @@ * RV40 decoder motion compensation functions */ +#include "libavutil/common.h" +#include "libavutil/intreadwrite.h" #include "avcodec.h" #include "h264qpel.h" #include "mathops.h" +#include "pixels.h" +#include "rnd_avg.h" #include "rv34dsp.h" -#include "libavutil/common.h" #define RV40_LOWPASS(OPNAME, OP) \ static av_unused void OPNAME ## rv40_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,\ @@ -203,6 +206,83 @@ RV40_MC(put_, 16) RV40_MC(avg_, 8) RV40_MC(avg_, 16) +#define PIXOP2(OPNAME, OP) \ +static inline void OPNAME ## _pixels8_xy2_8_c(uint8_t *block, \ + const uint8_t *pixels, \ + ptrdiff_t line_size, \ + int h) \ +{ \ + /* FIXME HIGH BIT DEPTH */ \ + int j; \ + \ + for (j = 0; j < 2; j++) { \ + int i; \ + const uint32_t a = AV_RN32(pixels); \ + const uint32_t b = AV_RN32(pixels + 1); \ + uint32_t l0 = (a & 0x03030303UL) + \ + (b & 0x03030303UL) + \ + 0x02020202UL; \ + uint32_t h0 = ((a & 0xFCFCFCFCUL) >> 2) + \ + ((b & 0xFCFCFCFCUL) >> 2); \ + uint32_t l1, h1; \ + \ + pixels += line_size; \ + for (i = 0; i < h; i += 2) { \ + uint32_t a = AV_RN32(pixels); \ + uint32_t b = AV_RN32(pixels + 1); \ + l1 = (a & 0x03030303UL) + \ + (b & 0x03030303UL); \ + h1 = ((a & 0xFCFCFCFCUL) >> 2) + \ + ((b & 0xFCFCFCFCUL) >> 2); \ + OP(*((uint32_t *) block), \ + h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL)); \ + pixels += line_size; \ + block += line_size; \ + a = AV_RN32(pixels); \ + b = AV_RN32(pixels + 1); \ + l0 = (a & 0x03030303UL) + \ + (b & 0x03030303UL) + \ + 0x02020202UL; \ + h0 = ((a & 0xFCFCFCFCUL) >> 2) + \ + ((b & 0xFCFCFCFCUL) >> 2); \ + OP(*((uint32_t *) block), \ + h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL)); \ + pixels += line_size; \ + block += line_size; \ + } \ + pixels += 4 - line_size * (h + 1); \ + block += 4 - line_size * h; \ + } \ +} \ + \ +CALL_2X_PIXELS(OPNAME ## _pixels16_xy2_8_c, \ + OPNAME ## _pixels8_xy2_8_c, \ + 8) \ + +#define op_avg(a, b) a = rnd_avg32(a, b) +#define op_put(a, b) a = b +PIXOP2(avg, op_avg) +PIXOP2(put, op_put) +#undef op_avg +#undef op_put + +static void put_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) +{ + put_pixels16_xy2_8_c(dst, src, stride, 16); +} +static void avg_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) +{ + avg_pixels16_xy2_8_c(dst, src, stride, 16); +} +static void put_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) +{ + put_pixels8_xy2_8_c(dst, src, stride, 8); +} +static void avg_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride) +{ + avg_pixels8_xy2_8_c(dst, src, stride, 8); +} + static const int rv40_bias[4][4] = { { 0, 16, 32, 16 }, { 32, 28, 32, 28 }, @@ -552,7 +632,7 @@ av_cold void ff_rv40dsp_init(RV34DSPContext *c) c->put_pixels_tab[0][12] = put_rv40_qpel16_mc03_c; c->put_pixels_tab[0][13] = put_rv40_qpel16_mc13_c; c->put_pixels_tab[0][14] = put_rv40_qpel16_mc23_c; - c->put_pixels_tab[0][15] = ff_put_rv40_qpel16_mc33_c; + c->put_pixels_tab[0][15] = put_rv40_qpel16_mc33_c; c->avg_pixels_tab[0][ 0] = qpel.avg_h264_qpel_pixels_tab[0][0]; c->avg_pixels_tab[0][ 1] = avg_rv40_qpel16_mc10_c; c->avg_pixels_tab[0][ 2] = qpel.avg_h264_qpel_pixels_tab[0][2]; @@ -568,7 +648,7 @@ av_cold void ff_rv40dsp_init(RV34DSPContext *c) c->avg_pixels_tab[0][12] = avg_rv40_qpel16_mc03_c; c->avg_pixels_tab[0][13] = avg_rv40_qpel16_mc13_c; c->avg_pixels_tab[0][14] = avg_rv40_qpel16_mc23_c; - c->avg_pixels_tab[0][15] = ff_avg_rv40_qpel16_mc33_c; + c->avg_pixels_tab[0][15] = avg_rv40_qpel16_mc33_c; c->put_pixels_tab[1][ 0] = qpel.put_h264_qpel_pixels_tab[1][0]; c->put_pixels_tab[1][ 1] = put_rv40_qpel8_mc10_c; c->put_pixels_tab[1][ 2] = qpel.put_h264_qpel_pixels_tab[1][2]; @@ -584,7 +664,7 @@ av_cold void ff_rv40dsp_init(RV34DSPContext *c) c->put_pixels_tab[1][12] = put_rv40_qpel8_mc03_c; c->put_pixels_tab[1][13] = put_rv40_qpel8_mc13_c; c->put_pixels_tab[1][14] = put_rv40_qpel8_mc23_c; - c->put_pixels_tab[1][15] = ff_put_rv40_qpel8_mc33_c; + c->put_pixels_tab[1][15] = put_rv40_qpel8_mc33_c; c->avg_pixels_tab[1][ 0] = qpel.avg_h264_qpel_pixels_tab[1][0]; c->avg_pixels_tab[1][ 1] = avg_rv40_qpel8_mc10_c; c->avg_pixels_tab[1][ 2] = qpel.avg_h264_qpel_pixels_tab[1][2]; @@ -600,7 +680,7 @@ av_cold void ff_rv40dsp_init(RV34DSPContext *c) c->avg_pixels_tab[1][12] = avg_rv40_qpel8_mc03_c; c->avg_pixels_tab[1][13] = avg_rv40_qpel8_mc13_c; c->avg_pixels_tab[1][14] = avg_rv40_qpel8_mc23_c; - c->avg_pixels_tab[1][15] = ff_avg_rv40_qpel8_mc33_c; + c->avg_pixels_tab[1][15] = avg_rv40_qpel8_mc33_c; c->put_chroma_pixels_tab[0] = put_rv40_chroma_mc8_c; c->put_chroma_pixels_tab[1] = put_rv40_chroma_mc4_c;