diff options
author | Anton Khirnov <anton@khirnov.net> | 2013-12-17 08:56:13 +0100 |
---|---|---|
committer | Anton Khirnov <anton@khirnov.net> | 2014-01-06 08:14:14 +0100 |
commit | fe80fa53bc4977ef6e8e749a9df8ca5590a09307 (patch) | |
tree | 49155b0270191f2e1a87eef6bf3ee0ccbc7ad18d | |
parent | fffca3d278c2a2422c2f61f21c5a9d5f690d328e (diff) | |
download | ffmpeg-fe80fa53bc4977ef6e8e749a9df8ca5590a09307.tar.gz |
4xm: replace forcing EMU_EDGE by a copy
The decoder currently sets CODEC_FLAG_EMU_EDGE and relies on
get_buffer2() to always provide buffers with linesize == 2 * width.
This is wrong, since we place no such restriction on get_buffer2()
implementations.
Fix this by decoding into internal buffers and copying them to output
frames. Since this is a very obscure decoder, the performance hit should
not be an issue.
-rw-r--r-- | libavcodec/4xm.c | 121 |
1 files changed, 59 insertions, 62 deletions
diff --git a/libavcodec/4xm.c b/libavcodec/4xm.c index 3ae19182ea..687fb895d5 100644 --- a/libavcodec/4xm.c +++ b/libavcodec/4xm.c @@ -25,6 +25,7 @@ */ #include "libavutil/frame.h" +#include "libavutil/imgutils.h" #include "libavutil/intreadwrite.h" #include "avcodec.h" #include "bytestream.h" @@ -129,7 +130,8 @@ typedef struct CFrameBuffer { typedef struct FourXContext { AVCodecContext *avctx; DSPContext dsp; - AVFrame *last_picture; + uint16_t *frame_buffer; + uint16_t *last_frame_buffer; GetBitContext pre_gb; ///< ac/dc prefix GetBitContext gb; GetByteContext g; @@ -339,7 +341,7 @@ static int decode_p_block(FourXContext *f, uint16_t *dst, uint16_t *src, int code = get_vlc2(&f->gb, block_type_vlc[1 - (f->version > 1)][index].table, BLOCK_TYPE_VLC_BITS, 1); - uint16_t *start = (uint16_t *)f->last_picture->data[0]; + uint16_t *start = f->last_frame_buffer; uint16_t *end = start + stride * (f->avctx->height - h + 1) - (1 << log2w); int ret; int scale = 1; @@ -395,30 +397,18 @@ static int decode_p_block(FourXContext *f, uint16_t *dst, uint16_t *src, return 0; } -static int decode_p_frame(FourXContext *f, AVFrame *frame, - const uint8_t *buf, int length) +static int decode_p_frame(FourXContext *f, const uint8_t *buf, int length) { int x, y; const int width = f->avctx->width; const int height = f->avctx->height; - uint16_t *dst = (uint16_t *)frame->data[0]; - const int stride = frame->linesize[0] >> 1; + uint16_t *dst = f->frame_buffer; uint16_t *src; unsigned int bitstream_size, bytestream_size, wordstream_size, extra, bytestream_offset, wordstream_offset; int ret; - if (!f->last_picture->data[0]) { - if ((ret = ff_get_buffer(f->avctx, f->last_picture, - AV_GET_BUFFER_FLAG_REF)) < 0) { - av_log(f->avctx, AV_LOG_ERROR, "get_buffer() failed\n"); - return ret; - } - memset(f->last_picture->data[0], 0, - f->avctx->height * FFABS(f->last_picture->linesize[0])); - } - - src = (uint16_t *)f->last_picture->data[0]; + src = f->last_frame_buffer; if (f->version > 1) { if (length < 20) @@ -461,14 +451,14 @@ static int decode_p_frame(FourXContext *f, AVFrame *frame, bytestream2_init(&f->g, buf + bytestream_offset, length - bytestream_offset); - init_mv(f, frame->linesize[0]); + init_mv(f, width * 2); for (y = 0; y < height; y += 8) { for (x = 0; x < width; x += 8) - if ((ret = decode_p_block(f, dst + x, src + x, 3, 3, stride)) < 0) + if ((ret = decode_p_block(f, dst + x, src + x, 3, 3, width)) < 0) return ret; - src += 8 * stride; - dst += 8 * stride; + src += 8 * width; + dst += 8 * width; } return 0; @@ -521,12 +511,12 @@ static int decode_i_block(FourXContext *f, int16_t *block) return 0; } -static inline void idct_put(FourXContext *f, AVFrame *frame, int x, int y) +static inline void idct_put(FourXContext *f, int x, int y) { int16_t (*block)[64] = f->block; - int stride = frame->linesize[0] >> 1; + int stride = f->avctx->width; int i; - uint16_t *dst = ((uint16_t*)frame->data[0]) + y * stride + x; + uint16_t *dst = f->frame_buffer + y * stride + x; for (i = 0; i < 4; i++) { block[i][0] += 0x80 * 8 * 8; @@ -683,14 +673,13 @@ static int mix(int c0, int c1) return red / 3 * 1024 + green / 3 * 32 + blue / 3; } -static int decode_i2_frame(FourXContext *f, AVFrame *frame, const uint8_t *buf, int length) +static int decode_i2_frame(FourXContext *f, const uint8_t *buf, int length) { int x, y, x2, y2; const int width = f->avctx->width; const int height = f->avctx->height; const int mbs = (FFALIGN(width, 16) >> 4) * (FFALIGN(height, 16) >> 4); - uint16_t *dst = (uint16_t*)frame->data[0]; - const int stride = frame->linesize[0]>>1; + uint16_t *dst = f->frame_buffer; GetByteContext g3; if (length < mbs * 8) { @@ -718,18 +707,18 @@ static int decode_i2_frame(FourXContext *f, AVFrame *frame, const uint8_t *buf, for (y2 = 0; y2 < 16; y2++) { for (x2 = 0; x2 < 16; x2++) { int index = 2 * (x2 >> 2) + 8 * (y2 >> 2); - dst[y2 * stride + x2] = color[(bits >> index) & 3]; + dst[y2 * width + x2] = color[(bits >> index) & 3]; } } dst += 16; } - dst += 16 * stride - x; + dst += 16 * width - x; } return 0; } -static int decode_i_frame(FourXContext *f, AVFrame *frame, const uint8_t *buf, int length) +static int decode_i_frame(FourXContext *f, const uint8_t *buf, int length) { int x, y, ret; const int width = f->avctx->width; @@ -785,7 +774,7 @@ static int decode_i_frame(FourXContext *f, AVFrame *frame, const uint8_t *buf, i if ((ret = decode_i_mb(f)) < 0) return ret; - idct_put(f, frame, x, y); + idct_put(f, x, y); } } @@ -882,25 +871,23 @@ static int decode_frame(AVCodecContext *avctx, void *data, frame_size = buf_size - 12; } - // alternatively we would have to use our own buffer management - avctx->flags |= CODEC_FLAG_EMU_EDGE; - if ((ret = ff_get_buffer(avctx, picture, AV_GET_BUFFER_FLAG_REF)) < 0) { + if ((ret = ff_get_buffer(avctx, picture, 0)) < 0) { av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n"); return ret; } if (frame_4cc == AV_RL32("ifr2")) { picture->pict_type = AV_PICTURE_TYPE_I; - if ((ret = decode_i2_frame(f, picture, buf - 4, frame_size + 4)) < 0) + if ((ret = decode_i2_frame(f, buf - 4, frame_size + 4)) < 0) return ret; } else if (frame_4cc == AV_RL32("ifrm")) { picture->pict_type = AV_PICTURE_TYPE_I; - if ((ret = decode_i_frame(f, picture, buf, frame_size)) < 0) + if ((ret = decode_i_frame(f, buf, frame_size)) < 0) return ret; } else if (frame_4cc == AV_RL32("pfrm") || frame_4cc == AV_RL32("pfr2")) { picture->pict_type = AV_PICTURE_TYPE_P; - if ((ret = decode_p_frame(f, picture, buf, frame_size)) < 0) + if ((ret = decode_p_frame(f, buf, frame_size)) < 0) return ret; } else if (frame_4cc == AV_RL32("snd_")) { av_log(avctx, AV_LOG_ERROR, "ignoring snd_ chunk length:%d\n", @@ -912,9 +899,11 @@ static int decode_frame(AVCodecContext *avctx, void *data, picture->key_frame = picture->pict_type == AV_PICTURE_TYPE_I; - av_frame_unref(f->last_picture); - if ((ret = av_frame_ref(f->last_picture, picture)) < 0) - return ret; + av_image_copy_plane(picture->data[0], picture->linesize[0], + (const uint8_t*)f->frame_buffer, avctx->width * 2, + avctx->width * 2, avctx->height); + FFSWAP(uint16_t *, f->frame_buffer, f->last_frame_buffer); + *got_frame = 1; emms_c(); @@ -922,15 +911,45 @@ static int decode_frame(AVCodecContext *avctx, void *data, return buf_size; } +static av_cold int decode_end(AVCodecContext *avctx) +{ + FourXContext * const f = avctx->priv_data; + int i; + + av_freep(&f->frame_buffer); + av_freep(&f->last_frame_buffer); + av_freep(&f->bitstream_buffer); + f->bitstream_buffer_size = 0; + for (i = 0; i < CFRAME_BUFFER_COUNT; i++) { + av_freep(&f->cfrm[i].data); + f->cfrm[i].allocated_size = 0; + } + ff_free_vlc(&f->pre_vlc); + + return 0; +} + static av_cold int decode_init(AVCodecContext *avctx) { FourXContext * const f = avctx->priv_data; + int ret; if (avctx->extradata_size != 4 || !avctx->extradata) { av_log(avctx, AV_LOG_ERROR, "extradata wrong or missing\n"); return AVERROR_INVALIDDATA; } + ret = av_image_check_size(avctx->width, avctx->height, 0, avctx); + if (ret < 0) + return ret; + + f->frame_buffer = av_mallocz(avctx->width * avctx->height * 2); + f->last_frame_buffer = av_mallocz(avctx->width * avctx->height * 2); + if (!f->frame_buffer || !f->last_frame_buffer) { + decode_end(avctx); + return AVERROR(ENOMEM); + } + f->version = AV_RL32(avctx->extradata) >> 16; ff_dsputil_init(&f->dsp, avctx); f->avctx = avctx; @@ -941,28 +960,6 @@ static av_cold int decode_init(AVCodecContext *avctx) else avctx->pix_fmt = AV_PIX_FMT_BGR555; - f->last_picture = av_frame_alloc(); - if (!f->last_picture) - return AVERROR(ENOMEM); - - return 0; -} - - -static av_cold int decode_end(AVCodecContext *avctx) -{ - FourXContext * const f = avctx->priv_data; - int i; - - av_freep(&f->bitstream_buffer); - f->bitstream_buffer_size = 0; - for (i = 0; i < CFRAME_BUFFER_COUNT; i++) { - av_freep(&f->cfrm[i].data); - f->cfrm[i].allocated_size = 0; - } - ff_free_vlc(&f->pre_vlc); - av_frame_free(&f->last_picture); - return 0; } |