From 4ca6d206d1b5beea42c4290d2ee801aaf5cd31f0 Mon Sep 17 00:00:00 2001 From: Mans Rullgard Date: Sun, 1 Jul 2012 13:36:30 +0100 Subject: alsdec: remove dead assignments Signed-off-by: Mans Rullgard --- libavcodec/alsdec.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'libavcodec') diff --git a/libavcodec/alsdec.c b/libavcodec/alsdec.c index 3d4f2bdd70..c411d5739a 100644 --- a/libavcodec/alsdec.c +++ b/libavcodec/alsdec.c @@ -756,7 +756,7 @@ static int read_var_block_data(ALSDecContext *ctx, ALSBlockData *bd) int delta[8]; unsigned int k [8]; unsigned int b = av_clip((av_ceil_log2(bd->block_length) - 3) >> 1, 0, 5); - unsigned int i = start; + unsigned int i; // read most significant bits unsigned int high; @@ -767,7 +767,7 @@ static int read_var_block_data(ALSDecContext *ctx, ALSBlockData *bd) current_res = bd->raw_samples + start; - for (sb = 0; sb < sub_blocks; sb++, i = 0) { + for (sb = 0; sb < sub_blocks; sb++) { k [sb] = s[sb] > b ? s[sb] - b : 0; delta[sb] = 5 - s[sb] + k[sb]; -- cgit v1.2.3 From 1c2c64edac4bd2b65b6d2f8d1489eed40f8a5458 Mon Sep 17 00:00:00 2001 From: Mans Rullgard Date: Sun, 1 Jul 2012 13:38:12 +0100 Subject: proresenc: make a variable local to the loop where it is used This moves the mbs_per_slice declaration inside the only loop where it is used. Fixes a dead assignment. Signed-off-by: Mans Rullgard --- libavcodec/proresenc.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'libavcodec') diff --git a/libavcodec/proresenc.c b/libavcodec/proresenc.c index bee49ee60a..f82153309b 100644 --- a/libavcodec/proresenc.c +++ b/libavcodec/proresenc.c @@ -724,7 +724,6 @@ static int encode_frame(AVCodecContext *avctx, AVPacket *pkt, int sizes[4] = { 0 }; int slice_hdr_size = 2 + 2 * (ctx->num_planes - 1); int frame_size, picture_size, slice_size; - int mbs_per_slice = ctx->mbs_per_slice; int pkt_size, ret; *avctx->coded_frame = *pic; @@ -792,7 +791,7 @@ static int encode_frame(AVCodecContext *avctx, AVPacket *pkt, } for (y = 0; y < ctx->mb_height; y++) { - mbs_per_slice = ctx->mbs_per_slice; + int mbs_per_slice = ctx->mbs_per_slice; for (x = mb = 0; x < ctx->mb_width; x += mbs_per_slice, mb++) { q = ctx->force_quant ? ctx->force_quant : ctx->slice_q[mb + y * ctx->slices_width]; -- cgit v1.2.3 From 800ab1bafa0698bdac32061c1bc72d5e04821e96 Mon Sep 17 00:00:00 2001 From: Mans Rullgard Date: Sun, 1 Jul 2012 13:43:44 +0100 Subject: rl2: remove dead assignment Signed-off-by: Mans Rullgard --- libavcodec/rl2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'libavcodec') diff --git a/libavcodec/rl2.c b/libavcodec/rl2.c index 9b21bb6c68..4d1937a4b4 100644 --- a/libavcodec/rl2.c +++ b/libavcodec/rl2.c @@ -64,7 +64,7 @@ static void rl2_rle_decode(Rl2Context *s,const unsigned char* in,int size, const unsigned char* back_frame = s->back_frame; const unsigned char* in_end = in + size; const unsigned char* out_end = out + stride * s->avctx->height; - unsigned char* line_end = out + s->avctx->width; + unsigned char* line_end; /** copy start of the background frame */ for(i=0;i<=base_y;i++){ -- cgit v1.2.3 From 3c928477d2f42c456f4c560db032a4380a3b6953 Mon Sep 17 00:00:00 2001 From: Mans Rullgard Date: Sun, 1 Jul 2012 13:50:09 +0100 Subject: aacdec: remove dead assignment Signed-off-by: Mans Rullgard --- libavcodec/aacdec.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'libavcodec') diff --git a/libavcodec/aacdec.c b/libavcodec/aacdec.c index f16b831ccd..4be525555a 100644 --- a/libavcodec/aacdec.c +++ b/libavcodec/aacdec.c @@ -1271,7 +1271,7 @@ static inline float *VMUL4S(float *dst, const float *v, unsigned idx, t.i = s.i ^ (sign & 1U<<31); *dst++ = v[idx>>4 & 3] * t.f; - sign <<= nz & 1; nz >>= 1; + sign <<= nz & 1; t.i = s.i ^ (sign & 1U<<31); *dst++ = v[idx>>6 & 3] * t.f; -- cgit v1.2.3 From d55e5e6bae4ba1ddc18683f181d27c4e84923281 Mon Sep 17 00:00:00 2001 From: Mans Rullgard Date: Sun, 1 Jul 2012 14:28:50 +0100 Subject: dsicin: remove dead assignment Signed-off-by: Mans Rullgard --- libavcodec/dsicinav.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'libavcodec') diff --git a/libavcodec/dsicinav.c b/libavcodec/dsicinav.c index d3b2cbded6..7e64b97b25 100644 --- a/libavcodec/dsicinav.c +++ b/libavcodec/dsicinav.c @@ -121,7 +121,7 @@ static int cin_decode_huffman(const unsigned char *src, int src_size, unsigned c unsigned char *dst_end = dst + dst_size; const unsigned char *src_end = src + src_size; - memcpy(huff_code_table, src, 15); src += 15; src_size -= 15; + memcpy(huff_code_table, src, 15); src += 15; while (src < src_end) { huff_code = *src++; -- cgit v1.2.3 From 779f8bc24ea94ec6f9d03ea52bb10b29c621c44b Mon Sep 17 00:00:00 2001 From: Mans Rullgard Date: Sun, 1 Jul 2012 14:56:16 +0100 Subject: smacker: remove some unused code This removes some code apparently left over from vlc reader debugging. Signed-off-by: Mans Rullgard --- libavcodec/smacker.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) (limited to 'libavcodec') diff --git a/libavcodec/smacker.c b/libavcodec/smacker.c index 43c206f25d..32ce48ff63 100644 --- a/libavcodec/smacker.c +++ b/libavcodec/smacker.c @@ -133,13 +133,9 @@ static int smacker_decode_bigtree(GetBitContext *gb, HuffContext *hc, DBCtx *ctx return -1; } if(!get_bits1(gb)){ //Leaf - int val, i1, i2, b1, b2; - b1 = get_bits_count(gb); + int val, i1, i2; i1 = ctx->v1->table ? get_vlc2(gb, ctx->v1->table, SMKTREE_BITS, 3) : 0; - b1 = get_bits_count(gb) - b1; - b2 = get_bits_count(gb); i2 = ctx->v2->table ? get_vlc2(gb, ctx->v2->table, SMKTREE_BITS, 3) : 0; - b2 = get_bits_count(gb) - b2; if (i1 < 0 || i2 < 0) return -1; val = ctx->recode1[i1] | (ctx->recode2[i2] << 8); -- cgit v1.2.3 From fc31508e3a7894efd8d084e3bc483f958dfdd726 Mon Sep 17 00:00:00 2001 From: Martin Storsjö Date: Sun, 1 Jul 2012 23:46:28 +0300 Subject: roqvideo: Remove a totally unused dspcontext MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Martin Storsjö --- libavcodec/roqvideo.h | 1 - 1 file changed, 1 deletion(-) (limited to 'libavcodec') diff --git a/libavcodec/roqvideo.h b/libavcodec/roqvideo.h index fba3d41a4a..4018e357a9 100644 --- a/libavcodec/roqvideo.h +++ b/libavcodec/roqvideo.h @@ -45,7 +45,6 @@ struct RoqTempData; typedef struct RoqContext { AVCodecContext *avctx; - DSPContext dsp; AVFrame frames[2]; AVFrame *last_frame; AVFrame *current_frame; -- cgit v1.2.3 From edf35d83429b1cba76ae8fd4717bdd5e6a9fe0eb Mon Sep 17 00:00:00 2001 From: Kostya Shishkov Date: Thu, 28 Jun 2012 21:18:23 +0200 Subject: MS ATC Screen (aka MSS3) decoder --- Changelog | 1 + doc/general.texi | 2 + libavcodec/Makefile | 1 + libavcodec/allcodecs.c | 1 + libavcodec/avcodec.h | 1 + libavcodec/mss3.c | 967 +++++++++++++++++++++++++++++++++++++++++++++++++ libavcodec/version.h | 2 +- libavformat/riff.c | 1 + 8 files changed, 975 insertions(+), 1 deletion(-) create mode 100644 libavcodec/mss3.c (limited to 'libavcodec') diff --git a/Changelog b/Changelog index 95f79bcd6e..39ad8a3fec 100644 --- a/Changelog +++ b/Changelog @@ -30,6 +30,7 @@ version : - Microsoft Screen 1 decoder - join audio filter - audio channel mapping filter +- Microsoft ATC Screen decoder version 0.8: diff --git a/doc/general.texi b/doc/general.texi index 1d00ef593e..4ea47b45d8 100644 --- a/doc/general.texi +++ b/doc/general.texi @@ -531,6 +531,8 @@ following image formats are supported: @item LCL (LossLess Codec Library) ZLIB @tab E @tab E @item LOCO @tab @tab X @item lossless MJPEG @tab X @tab X +@item Microsoft ATC Screen @tab @tab X + @tab Also known as Microsoft Screen 3. @item Microsoft RLE @tab @tab X @item Microsoft Screen 1 @tab @tab X @tab Also known as Windows Media Video V7 Screen. diff --git a/libavcodec/Makefile b/libavcodec/Makefile index 8204a1848d..efedee90fc 100644 --- a/libavcodec/Makefile +++ b/libavcodec/Makefile @@ -292,6 +292,7 @@ OBJS-$(CONFIG_MSMPEG4V3_ENCODER) += msmpeg4.o msmpeg4enc.o msmpeg4data.o \ h263dec.o h263.o ituh263dec.o \ mpeg4videodec.o OBJS-$(CONFIG_MSRLE_DECODER) += msrle.o msrledec.o +OBJS-$(CONFIG_MSA1_DECODER) += mss3.o OBJS-$(CONFIG_MSS1_DECODER) += mss1.o OBJS-$(CONFIG_MSVIDEO1_DECODER) += msvideo1.o OBJS-$(CONFIG_MSZH_DECODER) += lcldec.o diff --git a/libavcodec/allcodecs.c b/libavcodec/allcodecs.c index f6a2df8a94..cb08e3349e 100644 --- a/libavcodec/allcodecs.c +++ b/libavcodec/allcodecs.c @@ -152,6 +152,7 @@ void avcodec_register_all(void) REGISTER_DECODER (MPEG4_VDPAU, mpeg4_vdpau); REGISTER_DECODER (MPEG_VDPAU, mpeg_vdpau); REGISTER_DECODER (MPEG1_VDPAU, mpeg1_vdpau); + REGISTER_DECODER (MSA1, msa1); REGISTER_DECODER (MSMPEG4V1, msmpeg4v1); REGISTER_ENCDEC (MSMPEG4V2, msmpeg4v2); REGISTER_ENCDEC (MSMPEG4V3, msmpeg4v3); diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h index 5ba021c873..2b70b96779 100644 --- a/libavcodec/avcodec.h +++ b/libavcodec/avcodec.h @@ -256,6 +256,7 @@ enum CodecID { CODEC_ID_XBM, CODEC_ID_ZEROCODEC, CODEC_ID_MSS1, + CODEC_ID_MSA1, /* various PCM "codecs" */ CODEC_ID_FIRST_AUDIO = 0x10000, ///< A dummy id pointing at the start of audio codecs diff --git a/libavcodec/mss3.c b/libavcodec/mss3.c new file mode 100644 index 0000000000..37d8aee4ec --- /dev/null +++ b/libavcodec/mss3.c @@ -0,0 +1,967 @@ +/* + * Microsoft Screen 3 (aka Microsoft ATC Screen) decoder + * Copyright (c) 2012 Konstantin Shishkov + * + * 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 + */ + +/** + * @file + * Microsoft Screen 3 (aka Microsoft ATC Screen) decoder + */ + +#include "avcodec.h" +#include "bytestream.h" + +#define HEADER_SIZE 27 + +#define MODEL2_SCALE 13 +#define MODEL_SCALE 15 +#define MODEL256_SEC_SCALE 9 + +typedef struct Model2 { + int upd_val, till_rescale; + unsigned zero_freq, zero_weight; + unsigned total_freq, total_weight; +} Model2; + +typedef struct Model { + int weights[16], freqs[16]; + int num_syms; + int tot_weight; + int upd_val, max_upd_val, till_rescale; +} Model; + +typedef struct Model256 { + int weights[256], freqs[256]; + int tot_weight; + int secondary[68]; + int sec_size; + int upd_val, max_upd_val, till_rescale; +} Model256; + +#define RAC_BOTTOM 0x01000000 +typedef struct RangeCoder { + const uint8_t *src, *src_end; + + uint32_t range, low; + int got_error; +} RangeCoder; + +enum BlockType { + FILL_BLOCK = 0, + IMAGE_BLOCK, + DCT_BLOCK, + HAAR_BLOCK, + SKIP_BLOCK +}; + +typedef struct BlockTypeContext { + int last_type; + Model bt_model[5]; +} BlockTypeContext; + +typedef struct FillBlockCoder { + int fill_val; + Model coef_model; +} FillBlockCoder; + +typedef struct ImageBlockCoder { + Model256 esc_model, vec_entry_model; + Model vec_size_model; + Model vq_model[125]; +} ImageBlockCoder; + +typedef struct DCTBlockCoder { + int *prev_dc; + int prev_dc_stride; + int prev_dc_height; + int quality; + uint16_t qmat[64]; + Model dc_model; + Model2 sign_model; + Model256 ac_model; +} DCTBlockCoder; + +typedef struct HaarBlockCoder { + int quality, scale; + Model256 coef_model; + Model coef_hi_model; +} HaarBlockCoder; + +typedef struct MSS3Context { + AVCodecContext *avctx; + AVFrame pic; + + int got_error; + RangeCoder coder; + BlockTypeContext btype[3]; + FillBlockCoder fill_coder[3]; + ImageBlockCoder image_coder[3]; + DCTBlockCoder dct_coder[3]; + HaarBlockCoder haar_coder[3]; + + int dctblock[64]; + int hblock[16 * 16]; +} MSS3Context; + +static const uint8_t mss3_luma_quant[64] = { + 16, 11, 10, 16, 24, 40, 51, 61, + 12, 12, 14, 19, 26, 58, 60, 55, + 14, 13, 16, 24, 40, 57, 69, 56, + 14, 17, 22, 29, 51, 87, 80, 62, + 18, 22, 37, 56, 68, 109, 103, 77, + 24, 35, 55, 64, 81, 104, 113, 92, + 49, 64, 78, 87, 103, 121, 120, 101, + 72, 92, 95, 98, 112, 100, 103, 99 +}; + +static const uint8_t mss3_chroma_quant[64] = { + 17, 18, 24, 47, 99, 99, 99, 99, + 18, 21, 26, 66, 99, 99, 99, 99, + 24, 26, 56, 99, 99, 99, 99, 99, + 47, 66, 99, 99, 99, 99, 99, 99, + 99, 99, 99, 99, 99, 99, 99, 99, + 99, 99, 99, 99, 99, 99, 99, 99, + 99, 99, 99, 99, 99, 99, 99, 99, + 99, 99, 99, 99, 99, 99, 99, 99 +}; + +const uint8_t zigzag_scan[64] = { + 0, 1, 8, 16, 9, 2, 3, 10, + 17, 24, 32, 25, 18, 11, 4, 5, + 12, 19, 26, 33, 40, 48, 41, 34, + 27, 20, 13, 6, 7, 14, 21, 28, + 35, 42, 49, 56, 57, 50, 43, 36, + 29, 22, 15, 23, 30, 37, 44, 51, + 58, 59, 52, 45, 38, 31, 39, 46, + 53, 60, 61, 54, 47, 55, 62, 63 +}; + + +static void model2_reset(Model2 *m) +{ + m->zero_weight = 1; + m->total_weight = 2; + m->zero_freq = 0x1000; + m->total_freq = 0x2000; + m->upd_val = 4; + m->till_rescale = 4; +} + +static void model2_update(Model2 *m, int bit) +{ + unsigned scale; + + if (!bit) + m->zero_weight++; + m->till_rescale--; + if (m->till_rescale) + return; + + m->total_weight += m->upd_val; + if (m->total_weight > 0x2000) { + m->total_weight = (m->total_weight + 1) >> 1; + m->zero_weight = (m->zero_weight + 1) >> 1; + if (m->total_weight == m->zero_weight) + m->total_weight = m->zero_weight + 1; + } + m->upd_val = m->upd_val * 5 >> 2; + if (m->upd_val > 64) + m->upd_val = 64; + scale = 0x80000000u / m->total_weight; + m->zero_freq = m->zero_weight * scale >> 18; + m->total_freq = m->total_weight * scale >> 18; + m->till_rescale = m->upd_val; +} + +static void model_update(Model *m, int val) +{ + int i, sum = 0; + unsigned scale; + + m->weights[val]++; + m->till_rescale--; + if (m->till_rescale) + return; + m->tot_weight += m->upd_val; + + if (m->tot_weight > 0x8000) { + m->tot_weight = 0; + for (i = 0; i < m->num_syms; i++) { + m->weights[i] = (m->weights[i] + 1) >> 1; + m->tot_weight += m->weights[i]; + } + } + scale = 0x80000000u / m->tot_weight; + for (i = 0; i < m->num_syms; i++) { + m->freqs[i] = sum * scale >> 16; + sum += m->weights[i]; + } + + m->upd_val = m->upd_val * 5 >> 2; + if (m->upd_val > m->max_upd_val) + m->upd_val = m->max_upd_val; + m->till_rescale = m->upd_val; +} + +static void model_reset(Model *m) +{ + int i; + + m->tot_weight = 0; + for (i = 0; i < m->num_syms - 1; i++) + m->weights[i] = 1; + m->weights[m->num_syms - 1] = 0; + + m->upd_val = m->num_syms; + m->till_rescale = 1; + model_update(m, m->num_syms - 1); + m->till_rescale = + m->upd_val = (m->num_syms + 6) >> 1; +} + +static av_cold void model_init(Model *m, int num_syms) +{ + m->num_syms = num_syms; + m->max_upd_val = 8 * num_syms + 48; + + model_reset(m); +} + +static void model256_update(Model256 *m, int val) +{ + int i, sum = 0; + unsigned scale; + int send, sidx = 1; + + m->weights[val]++; + m->till_rescale--; + if (m->till_rescale) + return; + m->tot_weight += m->upd_val; + + if (m->tot_weight > 0x8000) { + m->tot_weight = 0; + for (i = 0; i < 256; i++) { + m->weights[i] = (m->weights[i] + 1) >> 1; + m->tot_weight += m->weights[i]; + } + } + scale = 0x80000000u / m->tot_weight; + m->secondary[0] = 0; + for (i = 0; i < 256; i++) { + m->freqs[i] = sum * scale >> 16; + sum += m->weights[i]; + send = m->freqs[i] >> MODEL256_SEC_SCALE; + while (sidx <= send) + m->secondary[sidx++] = i - 1; + } + while (sidx < m->sec_size) + m->secondary[sidx++] = 255; + + m->upd_val = m->upd_val * 5 >> 2; + if (m->upd_val > m->max_upd_val) + m->upd_val = m->max_upd_val; + m->till_rescale = m->upd_val; +} + +static void model256_reset(Model256 *m) +{ + int i; + + for (i = 0; i < 255; i++) + m->weights[i] = 1; + m->weights[255] = 0; + + m->tot_weight = 0; + m->upd_val = 256; + m->till_rescale = 1; + model256_update(m, 255); + m->till_rescale = + m->upd_val = (256 + 6) >> 1; +} + +static av_cold void model256_init(Model256 *m) +{ + m->max_upd_val = 8 * 256 + 48; + m->sec_size = (1 << 6) + 2; + + model256_reset(m); +} + +static void rac_init(RangeCoder *c, const uint8_t *src, int size) +{ + int i; + + c->src = src; + c->src_end = src + size; + c->low = 0; + for (i = 0; i < FFMIN(size, 4); i++) + c->low = (c->low << 8) | *c->src++; + c->range = 0xFFFFFFFF; + c->got_error = 0; +} + +static void rac_normalise(RangeCoder *c) +{ + for (;;) { + c->range <<= 8; + c->low <<= 8; + if (c->src < c->src_end) { + c->low |= *c->src++; + } else if (!c->low) { + c->got_error = 1; + return; + } + if (c->range >= RAC_BOTTOM) + return; + } +} + +static int rac_get_bit(RangeCoder *c) +{ + int bit; + + c->range >>= 1; + + bit = (c->range <= c->low); + if (bit) + c->low -= c->range; + + if (c->range < RAC_BOTTOM) + rac_normalise(c); + + return bit; +} + +static int rac_get_bits(RangeCoder *c, int nbits) +{ + int val; + + c->range >>= nbits; + val = c->low / c->range; + c->low -= c->range * val; + + if (c->range < RAC_BOTTOM) + rac_normalise(c); + + return val; +} + +static int rac_get_model2_sym(RangeCoder *c, Model2 *m) +{ + int bit, helper; + + helper = m->zero_freq * (c->range >> MODEL2_SCALE); + bit = (c->low >= helper); + if (bit) { + c->low -= helper; + c->range -= helper; + } else { + c->range = helper; + } + + if (c->range < RAC_BOTTOM) + rac_normalise(c); + + model2_update(m, bit); + + return bit; +} + +static int rac_get_model_sym(RangeCoder *c, Model *m) +{ + int prob, prob2, helper, val; + int end, end2; + + prob = 0; + prob2 = c->range; + c->range >>= MODEL_SCALE; + val = 0; + end = m->num_syms >> 1; + end2 = m->num_syms; + do { + helper = m->freqs[end] * c->range; + if (helper <= c->low) { + val = end; + prob = helper; + } else { + end2 = end; + prob2 = helper; + } + end = (end2 + val) >> 1; + } while (end != val); + c->low -= prob; + c->range = prob2 - prob; + if (c->range < RAC_BOTTOM) + rac_normalise(c); + + model_update(m, val); + + return val; +} + +static int rac_get_model256_sym(RangeCoder *c, Model256 *m) +{ + int prob, prob2, helper, val; + int start, end; + int ssym; + + prob2 = c->range; + c->range >>= MODEL_SCALE; + + helper = c->low / c->range; + ssym = helper >> MODEL256_SEC_SCALE; + val = m->secondary[ssym]; + + end = start = m->secondary[ssym + 1] + 1; + while (end > val + 1) { + ssym = (end + val) >> 1; + if (m->freqs[ssym] <= helper) { + end = start; + val = ssym; + } else { + end = (end + val) >> 1; + start = ssym; + } + } + prob = m->freqs[val] * c->range; + if (val != 255) + prob2 = m->freqs[val + 1] * c->range; + + c->low -= prob; + c->range = prob2 - prob; + if (c->range < RAC_BOTTOM) + rac_normalise(c); + + model256_update(m, val); + + return val; +} + +static int decode_block_type(RangeCoder *c, BlockTypeContext *bt) +{ + bt->last_type = rac_get_model_sym(c, &bt->bt_model[bt->last_type]); + + return bt->last_type; +} + +static int decode_coeff(RangeCoder *c, Model *m) +{ + int val, sign; + + val = rac_get_model_sym(c, m); + if (val) { + sign = rac_get_bit(c); + if (val > 1) { + val--; + val = (1 << val) + rac_get_bits(c, val); + } + if (!sign) + val = -val; + } + + return val; +} + +static void decode_fill_block(RangeCoder *c, FillBlockCoder *fc, + uint8_t *dst, int stride, int block_size) +{ + int i; + + fc->fill_val += decode_coeff(c, &fc->coef_model); + + for (i = 0; i < block_size; i++, dst += stride) + memset(dst, fc->fill_val, block_size); +} + +static void decode_image_block(RangeCoder *c, ImageBlockCoder *ic, + uint8_t *dst, int stride, int block_size) +{ + int i, j; + int vec_size; + int vec[4]; + int prev_line[16]; + int A, B, C; + + vec_size = rac_get_model_sym(c, &ic->vec_size_model) + 2; + for (i = 0; i < vec_size; i++) + vec[i] = rac_get_model256_sym(c, &ic->vec_entry_model); + for (; i < 4; i++) + vec[i] = 0; + memset(prev_line, 0, sizeof(prev_line)); + + for (j = 0; j < block_size; j++) { + A = 0; + B = 0; + for (i = 0; i < block_size; i++) { + C = B; + B = prev_line[i]; + A = rac_get_model_sym(c, &ic->vq_model[A + B * 5 + C * 25]); + + prev_line[i] = A; + if (A < 4) + dst[i] = vec[A]; + else + dst[i] = rac_get_model256_sym(c, &ic->esc_model); + } + dst += stride; + } +} + +static int decode_dct(RangeCoder *c, DCTBlockCoder *bc, int *block, + int bx, int by) +{ + int skip, val, sign, pos = 1, zz_pos, dc; + int blk_pos = bx + by * bc->prev_dc_stride; + + memset(block, 0, sizeof(*block) * 64); + + dc = decode_coeff(c, &bc->dc_model); + if (by) { + if (bx) { + int l, tl, t; + + l = bc->prev_dc[blk_pos - 1]; + tl = bc->prev_dc[blk_pos - 1 - bc->prev_dc_stride]; + t = bc->prev_dc[blk_pos - bc->prev_dc_stride]; + + if (FFABS(t - tl) <= FFABS(l - tl)) + dc += l; + else + dc += t; + } else { + dc += bc->prev_dc[blk_pos - bc->prev_dc_stride]; + } + } else if (bx) { + dc += bc->prev_dc[bx - 1]; + } + bc->prev_dc[blk_pos] = dc; + block[0] = dc * bc->qmat[0]; + + while (pos < 64) { + val = rac_get_model256_sym(c, &bc->ac_model); + if (!val) + return 0; + if (val == 0xF0) { + pos += 16; + continue; + } + skip = val >> 4; + val = val & 0xF; + if (!val) + return -1; + pos += skip; + if (pos >= 64) + return -1; + + sign = rac_get_model2_sym(c, &bc->sign_model); + if (val > 1) { + val--; + val = (1 << val) + rac_get_bits(c, val); + } + if (!sign) + val = -val; + + zz_pos = zigzag_scan[pos]; + block[zz_pos] = val * bc->qmat[zz_pos]; + pos++; + } + + return pos == 64 ? 0 : -1; +} + +#define DCT_TEMPLATE(blk, step, SOP, shift) \ + const int t0 = -39409 * blk[7 * step] - 58980 * blk[1 * step]; \ + const int t1 = 39410 * blk[1 * step] - 58980 * blk[7 * step]; \ + const int t2 = -33410 * blk[5 * step] - 167963 * blk[3 * step]; \ + const int t3 = 33410 * blk[3 * step] - 167963 * blk[5 * step]; \ + const int t4 = blk[3 * step] + blk[7 * step]; \ + const int t5 = blk[1 * step] + blk[5 * step]; \ + const int t6 = 77062 * t4 + 51491 * t5; \ + const int t7 = 77062 * t5 - 51491 * t4; \ + const int t8 = 35470 * blk[2 * step] - 85623 * blk[6 * step]; \ + const int t9 = 35470 * blk[6 * step] + 85623 * blk[2 * step]; \ + const int tA = SOP(blk[0 * step] - blk[4 * step]); \ + const int tB = SOP(blk[0 * step] + blk[4 * step]); \ + \ + blk[0 * step] = ( t1 + t6 + t9 + tB) >> shift; \ + blk[1 * step] = ( t3 + t7 + t8 + tA) >> shift; \ + blk[2 * step] = ( t2 + t6 - t8 + tA) >> shift; \ + blk[3 * step] = ( t0 + t7 - t9 + tB) >> shift; \ + blk[4 * step] = (-(t0 + t7) - t9 + tB) >> shift; \ + blk[5 * step] = (-(t2 + t6) - t8 + tA) >> shift; \ + blk[6 * step] = (-(t3 + t7) + t8 + tA) >> shift; \ + blk[7 * step] = (-(t1 + t6) + t9 + tB) >> shift; \ + +#define SOP_ROW(a) ((a) << 16) + 0x2000 +#define SOP_COL(a) ((a + 32) << 16) + +static void dct_put(uint8_t *dst, int stride, int *block) +{ + int i, j; + int *ptr; + + ptr = block; + for (i = 0; i < 8; i++) { + DCT_TEMPLATE(ptr, 1, SOP_ROW, 13); + ptr += 8; + } + + ptr = block; + for (i = 0; i < 8; i++) { + DCT_TEMPLATE(ptr, 8, SOP_COL, 22); + ptr++; + } + + ptr = block; + for (j = 0; j < 8; j++) { + for (i = 0; i < 8; i++) + dst[i] = av_clip_uint8(ptr[i] + 128); + dst += stride; + ptr += 8; + } +} + +static void decode_dct_block(RangeCoder *c, DCTBlockCoder *bc, + uint8_t *dst, int stride, int block_size, + int *block, int mb_x, int mb_y) +{ + int i, j; + int bx, by; + int nblocks = block_size >> 3; + + bx = mb_x * nblocks; + by = mb_y * nblocks; + + for (j = 0; j < nblocks; j++) { + for (i = 0; i < nblocks; i++) { + if (decode_dct(c, bc, block, bx + i, by + j)) { + c->got_error = 1; + return; + } + dct_put(dst + i * 8, stride, block); + } + dst += 8 * stride; + } +} + +static void decode_haar_block(RangeCoder *c, HaarBlockCoder *hc, + uint8_t *dst, int stride, int block_size, + int *block) +{ + const int hsize = block_size >> 1; + int A, B, C, D, t1, t2, t3, t4; + int i, j; + + for (j = 0; j < block_size; j++) { + for (i = 0; i < block_size; i++) { + if (i < hsize && j < hsize) + block[i] = rac_get_model256_sym(c, &hc->coef_model); + else + block[i] = decode_coeff(c, &hc->coef_hi_model); + block[i] *= hc->scale; + } + block += block_size; + } + block -= block_size * block_size; + + for (j = 0; j < hsize; j++) { + for (i = 0; i < hsize; i++) { + A = block[i]; + B = block[i + hsize]; + C = block[i + hsize * block_size]; + D = block[i + hsize * block_size + hsize]; + + t1 = A - B; + t2 = C - D; + t3 = A + B; + t4 = C + D; + dst[i * 2] = av_clip_uint8(t1 - t2); + dst[i * 2 + stride] = av_clip_uint8(t1 + t2); + dst[i * 2 + 1] = av_clip_uint8(t3 - t4); + dst[i * 2 + 1 + stride] = av_clip_uint8(t3 + t4); + } + block += block_size; + dst += stride * 2; + } +} + +static void gen_quant_mat(uint16_t *qmat, const uint8_t *ref, float scale) +{ + int i; + + for (i = 0; i < 64; i++) + qmat[i] = (uint16_t)(ref[i] * scale + 50.0) / 100; +} + +static void reset_coders(MSS3Context *ctx, int quality) +{ + int i, j; + + for (i = 0; i < 3; i++) { + ctx->btype[i].last_type = SKIP_BLOCK; + for (j = 0; j < 5; j++) + model_reset(&ctx->btype[i].bt_model[j]); + ctx->fill_coder[i].fill_val = 0; + model_reset(&ctx->fill_coder[i].coef_model); + model256_reset(&ctx->image_coder[i].esc_model); + model256_reset(&ctx->image_coder[i].vec_entry_model); + model_reset(&ctx->image_coder[i].vec_size_model); + for (j = 0; j < 125; j++) + model_reset(&ctx->image_coder[i].vq_model[j]); + if (ctx->dct_coder[i].quality != quality) { + float scale; + ctx->dct_coder[i].quality = quality; + if (quality > 50) + scale = 200.0f - 2 * quality; + else + scale = 5000.0f / quality; + gen_quant_mat(ctx->dct_coder[i].qmat, + i ? mss3_chroma_quant : mss3_luma_quant, + scale); + } + memset(ctx->dct_coder[i].prev_dc, 0, + sizeof(*ctx->dct_coder[i].prev_dc) * + ctx->dct_coder[i].prev_dc_stride * + ctx->dct_coder[i].prev_dc_height); + model_reset(&ctx->dct_coder[i].dc_model); + model2_reset(&ctx->dct_coder[i].sign_model); + model256_reset(&ctx->dct_coder[i].ac_model); + if (ctx->haar_coder[i].quality != quality) { + ctx->haar_coder[i].quality = quality; + ctx->haar_coder[i].scale = 17 - 7 * quality / 50; + } + model_reset(&ctx->haar_coder[i].coef_hi_model); + model256_reset(&ctx->haar_coder[i].coef_model); + } +} + +static av_cold void init_coders(MSS3Context *ctx) +{ + int i, j; + + for (i = 0; i < 3; i++) { + for (j = 0; j < 5; j++) + model_init(&ctx->btype[i].bt_model[j], 5); + model_init(&ctx->fill_coder[i].coef_model, 12); + model256_init(&ctx->image_coder[i].esc_model); + model256_init(&ctx->image_coder[i].vec_entry_model); + model_init(&ctx->image_coder[i].vec_size_model, 3); + for (j = 0; j < 125; j++) + model_init(&ctx->image_coder[i].vq_model[j], 5); + model_init(&ctx->dct_coder[i].dc_model, 12); + model256_init(&ctx->dct_coder[i].ac_model); + model_init(&ctx->haar_coder[i].coef_hi_model, 12); + model256_init(&ctx->haar_coder[i].coef_model); + } +} + +static int mss3_decode_frame(AVCodecContext *avctx, void *data, int *data_size, + AVPacket *avpkt) +{ + const uint8_t *buf = avpkt->data; + int buf_size = avpkt->size; + MSS3Context *c = avctx->priv_data; + RangeCoder *acoder = &c->coder; + GetByteContext gb; + uint8_t *dst[3]; + int dec_width, dec_height, dec_x, dec_y, quality, keyframe; + int x, y, i, mb_width, mb_height, blk_size, btype; + int ret; + + if (buf_size < HEADER_SIZE) { + av_log(avctx, AV_LOG_ERROR, + "Frame should have at least %d bytes, got %d instead\n", + HEADER_SIZE, buf_size); + return AVERROR_INVALIDDATA; + } + + bytestream2_init(&gb, buf, buf_size); + keyframe = bytestream2_get_be32(&gb); + if (keyframe & ~0x301) { + av_log(avctx, AV_LOG_ERROR, "Invalid frame type %X\n", keyframe); + return AVERROR_INVALIDDATA; + } + keyframe = !(keyframe & 1); + bytestream2_skip(&gb, 6); + dec_x = bytestream2_get_be16(&gb); + dec_y = bytestream2_get_be16(&gb); + dec_width = bytestream2_get_be16(&gb); + dec_height = bytestream2_get_be16(&gb); + + if (dec_x + dec_width > avctx->width || + dec_y + dec_height > avctx->height || + (dec_width | dec_height) & 0xF) { + av_log(avctx, AV_LOG_ERROR, "Invalid frame dimensions %dx%d +%d,%d\n", + dec_width, dec_height, dec_x, dec_y); + return AVERROR_INVALIDDATA; + } + bytestream2_skip(&gb, 4); + quality = bytestream2_get_byte(&gb); + if (quality < 1 || quality > 100) { + av_log(avctx, AV_LOG_ERROR, "Invalid quality setting %d\n", quality); + return AVERROR_INVALIDDATA; + } + bytestream2_skip(&gb, 4); + + if (keyframe && !bytestream2_get_bytes_left(&gb)) { + av_log(avctx, AV_LOG_ERROR, "Keyframe without data found\n"); + return AVERROR_INVALIDDATA; + } + if (!keyframe && c->got_error) + return buf_size; + c->got_error = 0; + + c->pic.reference = 3; + c->pic.buffer_hints = FF_BUFFER_HINTS_VALID | FF_BUFFER_HINTS_PRESERVE | + FF_BUFFER_HINTS_REUSABLE; + if ((ret = avctx->reget_buffer(avctx, &c->pic)) < 0) { + av_log(avctx, AV_LOG_ERROR, "reget_buffer() failed\n"); + return ret; + } + c->pic.key_frame = keyframe; + c->pic.pict_type = keyframe ? AV_PICTURE_TYPE_I : AV_PICTURE_TYPE_P; + if (!bytestream2_get_bytes_left(&gb)) { + *data_size = sizeof(AVFrame); + *(AVFrame*)data = c->pic; + + return buf_size; + } + + reset_coders(c, quality); + + rac_init(acoder, buf + HEADER_SIZE, buf_size - HEADER_SIZE); + + mb_width = dec_width >> 4; + mb_height = dec_height >> 4; + dst[0] = c->pic.data[0] + dec_x + dec_y * c->pic.linesize[0]; + dst[1] = c->pic.data[1] + dec_x / 2 + (dec_y / 2) * c->pic.linesize[1]; + dst[2] = c->pic.data[2] + dec_x / 2 + (dec_y / 2) * c->pic.linesize[2]; + for (y = 0; y < mb_height; y++) { + for (x = 0; x < mb_width; x++) { + for (i = 0; i < 3; i++) { + blk_size = 8 << !i; + + btype = decode_block_type(acoder, c->btype + i); + switch (btype) { + case FILL_BLOCK: + decode_fill_block(acoder, c->fill_coder + i, + dst[i] + x * blk_size, + c->pic.linesize[i], blk_size); + break; + case IMAGE_BLOCK: + decode_image_block(acoder, c->image_coder + i, + dst[i] + x * blk_size, + c->pic.linesize[i], blk_size); + break; + case DCT_BLOCK: + decode_dct_block(acoder, c->dct_coder + i, + dst[i] + x * blk_size, + c->pic.linesize[i], blk_size, + c->dctblock, x, y); + break; + case HAAR_BLOCK: + decode_haar_block(acoder, c->haar_coder + i, + dst[i] + x * blk_size, + c->pic.linesize[i], blk_size, + c->hblock); + break; + } + if (c->got_error || acoder->got_error) { + av_log(avctx, AV_LOG_ERROR, "Error decoding block %d,%d\n", + x, y); + c->got_error = 1; + return AVERROR_INVALIDDATA; + } + } + } + dst[0] += c->pic.linesize[0] * 16; + dst[1] += c->pic.linesize[1] * 8; + dst[2] += c->pic.linesize[2] * 8; + } + + *data_size = sizeof(AVFrame); + *(AVFrame*)data = c->pic; + + return buf_size; +} + +static av_cold int mss3_decode_init(AVCodecContext *avctx) +{ + MSS3Context * const c = avctx->priv_data; + int i; + + c->avctx = avctx; + + if ((avctx->width & 0xF) || (avctx->height & 0xF)) { + av_log(avctx, AV_LOG_ERROR, + "Image dimensions should be a multiple of 16.\n"); + return AVERROR_INVALIDDATA; + } + + c->got_error = 0; + for (i = 0; i < 3; i++) { + int b_width = avctx->width >> (2 + !!i); + int b_height = avctx->height >> (2 + !!i); + c->dct_coder[i].prev_dc_stride = b_width; + c->dct_coder[i].prev_dc_height = b_height; + c->dct_coder[i].prev_dc = av_malloc(sizeof(*c->dct_coder[i].prev_dc) * + b_width * b_height); + if (!c->dct_coder[i].prev_dc) { + av_log(avctx, AV_LOG_ERROR, "Cannot allocate buffer\n"); + while (i >= 0) { + av_freep(&c->dct_coder[i].prev_dc); + i--; + } + return AVERROR(ENOMEM); + } + } + + avctx->pix_fmt = PIX_FMT_YUV420P; + avctx->coded_frame = &c->pic; + + init_coders(c); + + return 0; +} + +static av_cold int mss3_decode_end(AVCodecContext *avctx) +{ + MSS3Context * const c = avctx->priv_data; + int i; + + if (c->pic.data[0]) + avctx->release_buffer(avctx, &c->pic); + for (i = 0; i < 3; i++) + av_freep(&c->dct_coder[i].prev_dc); + + return 0; +} + +AVCodec ff_msa1_decoder = { + .name = "msa1", + .type = AVMEDIA_TYPE_VIDEO, + .id = CODEC_ID_MSA1, + .priv_data_size = sizeof(MSS3Context), + .init = mss3_decode_init, + .close = mss3_decode_end, + .decode = mss3_decode_frame, + .capabilities = CODEC_CAP_DR1, + .long_name = NULL_IF_CONFIG_SMALL("MS ATC Screen"), +}; diff --git a/libavcodec/version.h b/libavcodec/version.h index e049bee06a..f42aebe2bb 100644 --- a/libavcodec/version.h +++ b/libavcodec/version.h @@ -27,7 +27,7 @@ */ #define LIBAVCODEC_VERSION_MAJOR 54 -#define LIBAVCODEC_VERSION_MINOR 16 +#define LIBAVCODEC_VERSION_MINOR 17 #define LIBAVCODEC_VERSION_MICRO 0 #define LIBAVCODEC_VERSION_INT AV_VERSION_INT(LIBAVCODEC_VERSION_MAJOR, \ diff --git a/libavformat/riff.c b/libavformat/riff.c index efab7048ba..0aef2d7b71 100644 --- a/libavformat/riff.c +++ b/libavformat/riff.c @@ -285,6 +285,7 @@ const AVCodecTag ff_codec_bmp_tags[] = { { CODEC_ID_DXTORY, MKTAG('x', 't', 'o', 'r') }, { CODEC_ID_ZEROCODEC, MKTAG('Z', 'E', 'C', 'O') }, { CODEC_ID_MSS1, MKTAG('M', 'S', 'S', '1') }, + { CODEC_ID_MSA1, MKTAG('M', 'S', 'A', '1') }, { CODEC_ID_NONE, 0 } }; -- cgit v1.2.3 From 33bb63cb3e1de6d78c475cf14384089ef3f1867d Mon Sep 17 00:00:00 2001 From: "Ronald S. Bultje" Date: Fri, 15 Jun 2012 09:59:57 -0700 Subject: snow: remove a VLA. MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Martin Storsjö --- libavcodec/snowenc.c | 41 +++++++++++++++++++++-------------------- 1 file changed, 21 insertions(+), 20 deletions(-) (limited to 'libavcodec') diff --git a/libavcodec/snowenc.c b/libavcodec/snowenc.c index 7e3b68fb8a..627a406af9 100644 --- a/libavcodec/snowenc.c +++ b/libavcodec/snowenc.c @@ -662,7 +662,7 @@ static inline int get_block_bits(SnowContext *s, int x, int y, int w){ } } -static int get_block_rd(SnowContext *s, int mb_x, int mb_y, int plane_index, const uint8_t *obmc_edged){ +static int get_block_rd(SnowContext *s, int mb_x, int mb_y, int plane_index, uint8_t (*obmc_edged)[MB_SIZE * 2]){ Plane *p= &s->plane[plane_index]; const int block_size = MB_SIZE >> s->block_max_depth; const int block_w = plane_index ? block_size/2 : block_size; @@ -691,7 +691,7 @@ static int get_block_rd(SnowContext *s, int mb_x, int mb_y, int plane_index, con ff_snow_pred_block(s, cur, tmp, ref_stride, sx, sy, block_w*2, block_w*2, &s->block[mb_x + mb_y*b_stride], plane_index, w, h); for(y=y0; yb_width << s->block_max_depth; BlockNode *block= &s->block[mb_x + mb_y * b_stride]; BlockNode backup= *block; @@ -994,7 +994,7 @@ static av_always_inline int check_block(SnowContext *s, int mb_x, int mb_y, int /* special case for int[2] args we discard afterwards, * fixes compilation problem with gcc 2.95 */ -static av_always_inline int check_block_inter(SnowContext *s, int mb_x, int mb_y, int p0, int p1, const uint8_t *obmc_edged, int *best_rd){ +static av_always_inline int check_block_inter(SnowContext *s, int mb_x, int mb_y, int p0, int p1, uint8_t (*obmc_edged)[MB_SIZE * 2], int *best_rd){ int p[2] = {p0, p1}; return check_block(s, mb_x, mb_y, p, 0, obmc_edged, best_rd); } @@ -1074,7 +1074,7 @@ static void iterative_me(SnowContext *s){ BlockNode *blb= mb_x && mb_y+1block[index+b_stride-1] : NULL; BlockNode *brb= mb_x+1block[index+b_stride+1] : NULL; const int b_w= (MB_SIZE >> s->block_max_depth); - uint8_t obmc_edged[b_w*2][b_w*2]; + uint8_t obmc_edged[MB_SIZE * 2][MB_SIZE * 2]; if(pass && (block->type & BLOCK_OPT)) continue; @@ -1089,7 +1089,8 @@ static void iterative_me(SnowContext *s){ //FIXME precalculate { int x, y; - memcpy(obmc_edged, ff_obmc_tab[s->block_max_depth], b_w*b_w*4); + for (y = 0; y < b_w * 2; y++) + memcpy(obmc_edged[y], ff_obmc_tab[s->block_max_depth] + y * b_w * 2, b_w * 2); if(mb_x==0) for(y=0; y 0 && (block->type&BLOCK_INTRA)){ int color0[3]= {block->color[0], block->color[1], block->color[2]}; - check_block(s, mb_x, mb_y, color0, 1, *obmc_edged, &best_rd); + check_block(s, mb_x, mb_y, color0, 1, obmc_edged, &best_rd); }else - check_block_inter(s, mb_x, mb_y, block->mx, block->my, *obmc_edged, &best_rd); + check_block_inter(s, mb_x, mb_y, block->mx, block->my, obmc_edged, &best_rd); ref_b= *block; ref_rd= best_rd; @@ -1156,16 +1157,16 @@ static void iterative_me(SnowContext *s){ block->ref= ref; best_rd= INT_MAX; - check_block_inter(s, mb_x, mb_y, mvr[0][0], mvr[0][1], *obmc_edged, &best_rd); - check_block_inter(s, mb_x, mb_y, 0, 0, *obmc_edged, &best_rd); + check_block_inter(s, mb_x, mb_y, mvr[0][0], mvr[0][1], obmc_edged, &best_rd); + check_block_inter(s, mb_x, mb_y, 0, 0, obmc_edged, &best_rd); if(tb) - check_block_inter(s, mb_x, mb_y, mvr[-b_stride][0], mvr[-b_stride][1], *obmc_edged, &best_rd); + check_block_inter(s, mb_x, mb_y, mvr[-b_stride][0], mvr[-b_stride][1], obmc_edged, &best_rd); if(lb) - check_block_inter(s, mb_x, mb_y, mvr[-1][0], mvr[-1][1], *obmc_edged, &best_rd); + check_block_inter(s, mb_x, mb_y, mvr[-1][0], mvr[-1][1], obmc_edged, &best_rd); if(rb) - check_block_inter(s, mb_x, mb_y, mvr[1][0], mvr[1][1], *obmc_edged, &best_rd); + check_block_inter(s, mb_x, mb_y, mvr[1][0], mvr[1][1], obmc_edged, &best_rd); if(bb) - check_block_inter(s, mb_x, mb_y, mvr[b_stride][0], mvr[b_stride][1], *obmc_edged, &best_rd); + check_block_inter(s, mb_x, mb_y, mvr[b_stride][0], mvr[b_stride][1], obmc_edged, &best_rd); /* fullpel ME */ //FIXME avoid subpel interpolation / round to nearest integer @@ -1173,10 +1174,10 @@ static void iterative_me(SnowContext *s){ dia_change=0; for(i=0; iavctx->dia_size, 1); i++){ for(j=0; jmx+4*(i-j), block->my+(4*j), *obmc_edged, &best_rd); - dia_change |= check_block_inter(s, mb_x, mb_y, block->mx-4*(i-j), block->my-(4*j), *obmc_edged, &best_rd); - dia_change |= check_block_inter(s, mb_x, mb_y, block->mx+4*(i-j), block->my-(4*j), *obmc_edged, &best_rd); - dia_change |= check_block_inter(s, mb_x, mb_y, block->mx-4*(i-j), block->my+(4*j), *obmc_edged, &best_rd); + dia_change |= check_block_inter(s, mb_x, mb_y, block->mx+4*(i-j), block->my+(4*j), obmc_edged, &best_rd); + dia_change |= check_block_inter(s, mb_x, mb_y, block->mx-4*(i-j), block->my-(4*j), obmc_edged, &best_rd); + dia_change |= check_block_inter(s, mb_x, mb_y, block->mx+4*(i-j), block->my-(4*j), obmc_edged, &best_rd); + dia_change |= check_block_inter(s, mb_x, mb_y, block->mx-4*(i-j), block->my+(4*j), obmc_edged, &best_rd); } } }while(dia_change); @@ -1185,7 +1186,7 @@ static void iterative_me(SnowContext *s){ static const int square[8][2]= {{+1, 0},{-1, 0},{ 0,+1},{ 0,-1},{+1,+1},{-1,-1},{+1,-1},{-1,+1},}; dia_change=0; for(i=0; i<8; i++) - dia_change |= check_block_inter(s, mb_x, mb_y, block->mx+square[i][0], block->my+square[i][1], *obmc_edged, &best_rd); + dia_change |= check_block_inter(s, mb_x, mb_y, block->mx+square[i][0], block->my+square[i][1], obmc_edged, &best_rd); }while(dia_change); //FIXME or try the standard 2 pass qpel or similar @@ -1198,7 +1199,7 @@ static void iterative_me(SnowContext *s){ } best_rd= ref_rd; *block= ref_b; - check_block(s, mb_x, mb_y, color, 1, *obmc_edged, &best_rd); + check_block(s, mb_x, mb_y, color, 1, obmc_edged, &best_rd); //FIXME RD style color selection if(!same_block(block, &backup)){ if(tb ) tb ->type &= ~BLOCK_OPT; -- cgit v1.2.3 From 58e57c3052f3e0e921c5f2b27f9acb545c4d826e Mon Sep 17 00:00:00 2001 From: Mans Rullgard Date: Sun, 1 Jul 2012 15:39:22 +0100 Subject: snow: remove VLA in mc_block() Signed-off-by: Mans Rullgard --- libavcodec/snow.c | 41 +++++++++++++++++++++++++---------------- 1 file changed, 25 insertions(+), 16 deletions(-) (limited to 'libavcodec') diff --git a/libavcodec/snow.c b/libavcodec/snow.c index edd7d075e4..821b81bf47 100644 --- a/libavcodec/snow.c +++ b/libavcodec/snow.c @@ -143,7 +143,7 @@ static void mc_block(Plane *p, uint8_t *dst, const uint8_t *src, int stride, int int x, y, b, r, l; int16_t tmpIt [64*(32+HTAPS_MAX)]; - uint8_t tmp2t[3][stride*(32+HTAPS_MAX)]; + uint8_t tmp2t[3][64*(32+HTAPS_MAX)]; int16_t *tmpI= tmpIt; uint8_t *tmp2= tmp2t[0]; const uint8_t *hpel[11]; @@ -181,7 +181,7 @@ static void mc_block(Plane *p, uint8_t *dst, const uint8_t *src, int stride, int tmp2[x]= am; } tmpI+= 64; - tmp2+= stride; + tmp2+= 64; src += stride; } src -= stride*y; @@ -210,7 +210,7 @@ static void mc_block(Plane *p, uint8_t *dst, const uint8_t *src, int stride, int tmp2[x]= am; } src += stride; - tmp2+= stride; + tmp2+= 64; } src -= stride*y; } @@ -237,12 +237,12 @@ static void mc_block(Plane *p, uint8_t *dst, const uint8_t *src, int stride, int tmp2[x]= am; } tmpI+= 64; - tmp2+= stride; + tmp2+= 64; } } hpel[ 0]= src; - hpel[ 1]= tmp2t[0] + stride*(HTAPS_MAX/2-1); + hpel[ 1]= tmp2t[0] + 64*(HTAPS_MAX/2-1); hpel[ 2]= src + 1; hpel[ 4]= tmp2t[1]; @@ -250,14 +250,21 @@ static void mc_block(Plane *p, uint8_t *dst, const uint8_t *src, int stride, int hpel[ 6]= tmp2t[1] + 1; hpel[ 8]= src + stride; - hpel[ 9]= hpel[1] + stride; + hpel[ 9]= hpel[1] + 64; hpel[10]= hpel[8] + 1; +#define MC_STRIDE(x) (needs[x] ? 64 : stride) + if(b==15){ - const uint8_t *src1= hpel[dx/8 + dy/8*4 ]; - const uint8_t *src2= hpel[dx/8 + dy/8*4+1]; - const uint8_t *src3= hpel[dx/8 + dy/8*4+4]; - const uint8_t *src4= hpel[dx/8 + dy/8*4+5]; + int dxy = dx / 8 + dy / 8 * 4; + const uint8_t *src1 = hpel[dxy ]; + const uint8_t *src2 = hpel[dxy + 1]; + const uint8_t *src3 = hpel[dxy + 4]; + const uint8_t *src4 = hpel[dxy + 5]; + int stride1 = MC_STRIDE(dxy); + int stride2 = MC_STRIDE(dxy + 1); + int stride3 = MC_STRIDE(dxy + 4); + int stride4 = MC_STRIDE(dxy + 5); dx&=7; dy&=7; for(y=0; y < b_h; y++){ @@ -265,23 +272,25 @@ static void mc_block(Plane *p, uint8_t *dst, const uint8_t *src, int stride, int dst[x]= ((8-dx)*(8-dy)*src1[x] + dx*(8-dy)*src2[x]+ (8-dx)* dy *src3[x] + dx* dy *src4[x]+32)>>6; } - src1+=stride; - src2+=stride; - src3+=stride; - src4+=stride; + src1+=stride1; + src2+=stride2; + src3+=stride3; + src4+=stride4; dst +=stride; } }else{ const uint8_t *src1= hpel[l]; const uint8_t *src2= hpel[r]; + int stride1 = MC_STRIDE(l); + int stride2 = MC_STRIDE(r); int a= weight[((dx&7) + (8*(dy&7)))]; int b= 8-a; for(y=0; y < b_h; y++){ for(x=0; x < b_w; x++){ dst[x]= (a*src1[x] + b*src2[x] + 4)>>3; } - src1+=stride; - src2+=stride; + src1+=stride1; + src2+=stride2; dst +=stride; } } -- cgit v1.2.3