[FFmpeg-devel] [PATCH 4/8] avcodec/hapdec: use the common texturedsp decode function
Marton Balint
cus at passwd.hu
Wed Mar 30 23:32:01 EEST 2022
Signed-off-by: Marton Balint <cus at passwd.hu>
---
libavcodec/hap.h | 11 ++---
libavcodec/hapdec.c | 110 ++++++++++----------------------------------
2 files changed, 28 insertions(+), 93 deletions(-)
diff --git a/libavcodec/hap.h b/libavcodec/hap.h
index 00c3dbb32d..7e065e4838 100644
--- a/libavcodec/hap.h
+++ b/libavcodec/hap.h
@@ -72,23 +72,18 @@ typedef struct HapContext {
HapChunk *chunks;
int *chunk_results; /* Results from threaded operations */
- int tex_rat; /* Compression ratio */
- int tex_rat2; /* Compression ratio of the second texture */
- const uint8_t *tex_data; /* Compressed texture */
uint8_t *tex_buf; /* Buffer for compressed texture */
size_t tex_size; /* Size of the compressed texture */
size_t max_snappy; /* Maximum compressed size for snappy buffer */
- int slice_count; /* Number of slices for threaded operations */
-
int texture_count; /* 2 for HAQA, 1 for other version */
int texture_section_size; /* size of the part of the texture section (for HAPQA) */
- int uncompress_pix_size; /* nb of byte / pixel for the target picture */
- /* Pointer to the selected compress or decompress function */
+ /* Pointer to the selected compress function (encoder only) */
int (*tex_fun)(uint8_t *dst, ptrdiff_t stride, const uint8_t *block);
- int (*tex_fun2)(uint8_t *dst, ptrdiff_t stride, const uint8_t *block);
+
+ TextureDSPThreadContext dec[2];
} HapContext;
/*
diff --git a/libavcodec/hapdec.c b/libavcodec/hapdec.c
index 4a7ac15a8e..685d65d8b8 100644
--- a/libavcodec/hapdec.c
+++ b/libavcodec/hapdec.c
@@ -247,60 +247,6 @@ static int decompress_chunks_thread(AVCodecContext *avctx, void *arg,
return 0;
}
-static int decompress_texture_thread_internal(AVCodecContext *avctx, void *arg,
- int slice, int thread_nb, int texture_num)
-{
- HapContext *ctx = avctx->priv_data;
- AVFrame *frame = arg;
- const uint8_t *d = ctx->tex_data;
- int w_block = avctx->coded_width / TEXTURE_BLOCK_W;
- int h_block = avctx->coded_height / TEXTURE_BLOCK_H;
- int x, y;
- int start_slice, end_slice;
- int base_blocks_per_slice = h_block / ctx->slice_count;
- int remainder_blocks = h_block % ctx->slice_count;
-
- /* When the frame height (in blocks) doesn't divide evenly between the
- * number of slices, spread the remaining blocks evenly between the first
- * operations */
- start_slice = slice * base_blocks_per_slice;
- /* Add any extra blocks (one per slice) that have been added before this slice */
- start_slice += FFMIN(slice, remainder_blocks);
-
- end_slice = start_slice + base_blocks_per_slice;
- /* Add an extra block if there are still remainder blocks to be accounted for */
- if (slice < remainder_blocks)
- end_slice++;
-
- for (y = start_slice; y < end_slice; y++) {
- uint8_t *p = frame->data[0] + y * frame->linesize[0] * TEXTURE_BLOCK_H;
- int off = y * w_block;
- for (x = 0; x < w_block; x++) {
- if (texture_num == 0) {
- ctx->tex_fun(p + x * 4 * ctx->uncompress_pix_size, frame->linesize[0],
- d + (off + x) * ctx->tex_rat);
- } else {
- ctx->tex_fun2(p + x * 4 * ctx->uncompress_pix_size, frame->linesize[0],
- d + (off + x) * ctx->tex_rat2);
- }
- }
- }
-
- return 0;
-}
-
-static int decompress_texture_thread(AVCodecContext *avctx, void *arg,
- int slice, int thread_nb)
-{
- return decompress_texture_thread_internal(avctx, arg, slice, thread_nb, 0);
-}
-
-static int decompress_texture2_thread(AVCodecContext *avctx, void *arg,
- int slice, int thread_nb)
-{
- return decompress_texture_thread_internal(avctx, arg, slice, thread_nb, 1);
-}
-
static int hap_decode(AVCodecContext *avctx, void *data,
int *got_frame, AVPacket *avpkt)
{
@@ -310,12 +256,9 @@ static int hap_decode(AVCodecContext *avctx, void *data,
int section_size;
enum HapSectionType section_type;
int start_texture_section = 0;
- int tex_rat[2] = {0, 0};
bytestream2_init(&ctx->gbc, avpkt->data, avpkt->size);
- tex_rat[0] = ctx->tex_rat;
-
/* check for multi texture header */
if (ctx->texture_count == 2) {
ret = ff_hap_parse_section_header(&ctx->gbc, §ion_size, §ion_type);
@@ -326,7 +269,6 @@ static int hap_decode(AVCodecContext *avctx, void *data,
return AVERROR_INVALIDDATA;
}
start_texture_section = 4;
- tex_rat[1] = ctx->tex_rat2;
}
/* Get the output frame ready to receive data */
@@ -344,7 +286,7 @@ static int hap_decode(AVCodecContext *avctx, void *data,
if (ctx->tex_size != (avctx->coded_width / TEXTURE_BLOCK_W)
*(avctx->coded_height / TEXTURE_BLOCK_H)
- *tex_rat[t]) {
+ *ctx->dec[t].tex_ratio) {
av_log(avctx, AV_LOG_ERROR, "uncompressed size mismatches\n");
return AVERROR_INVALIDDATA;
}
@@ -355,11 +297,11 @@ static int hap_decode(AVCodecContext *avctx, void *data,
if (hap_can_use_tex_in_place(ctx)) {
int tex_size;
/* Only DXTC texture compression in a contiguous block */
- ctx->tex_data = ctx->gbc.buffer;
+ ctx->dec[t].tex_data.in = ctx->gbc.buffer;
tex_size = FFMIN(ctx->texture_section_size, bytestream2_get_bytes_left(&ctx->gbc));
if (tex_size < (avctx->coded_width / TEXTURE_BLOCK_W)
*(avctx->coded_height / TEXTURE_BLOCK_H)
- *tex_rat[t]) {
+ *ctx->dec[t].tex_ratio) {
av_log(avctx, AV_LOG_ERROR, "Insufficient data\n");
return AVERROR_INVALIDDATA;
}
@@ -377,15 +319,12 @@ static int hap_decode(AVCodecContext *avctx, void *data,
return ctx->chunk_results[i];
}
- ctx->tex_data = ctx->tex_buf;
+ ctx->dec[t].tex_data.in = ctx->tex_buf;
}
- /* Use the decompress function on the texture, one block per thread */
- if (t == 0){
- avctx->execute2(avctx, decompress_texture_thread, frame, NULL, ctx->slice_count);
- } else{
- avctx->execute2(avctx, decompress_texture2_thread, frame, NULL, ctx->slice_count);
- }
+ ctx->dec[t].frame_data.out = frame->data[0];
+ ctx->dec[t].stride = frame->linesize[0];
+ avctx->execute2(avctx, ff_texturedsp_decompress_thread, &ctx->dec[t], NULL, ctx->dec[t].slice_count);
}
/* Frame is ready to be output */
@@ -415,40 +354,44 @@ static av_cold int hap_init(AVCodecContext *avctx)
ff_texturedsp_init(&ctx->dxtc);
ctx->texture_count = 1;
- ctx->uncompress_pix_size = 4;
+ ctx->dec[0].raw_ratio = 16;
+ ctx->dec[0].slice_count = av_clip(avctx->thread_count, 1,
+ avctx->coded_height / TEXTURE_BLOCK_H);
switch (avctx->codec_tag) {
case MKTAG('H','a','p','1'):
texture_name = "DXT1";
- ctx->tex_rat = 8;
- ctx->tex_fun = ctx->dxtc.dxt1_block;
+ ctx->dec[0].tex_ratio = 8;
+ ctx->dec[0].tex_funct = ctx->dxtc.dxt1_block;
avctx->pix_fmt = AV_PIX_FMT_RGB0;
break;
case MKTAG('H','a','p','5'):
texture_name = "DXT5";
- ctx->tex_rat = 16;
- ctx->tex_fun = ctx->dxtc.dxt5_block;
+ ctx->dec[0].tex_ratio = 16;
+ ctx->dec[0].tex_funct = ctx->dxtc.dxt5_block;
avctx->pix_fmt = AV_PIX_FMT_RGBA;
break;
case MKTAG('H','a','p','Y'):
texture_name = "DXT5-YCoCg-scaled";
- ctx->tex_rat = 16;
- ctx->tex_fun = ctx->dxtc.dxt5ys_block;
+ ctx->dec[0].tex_ratio = 16;
+ ctx->dec[0].tex_funct = ctx->dxtc.dxt5ys_block;
avctx->pix_fmt = AV_PIX_FMT_RGB0;
break;
case MKTAG('H','a','p','A'):
texture_name = "RGTC1";
- ctx->tex_rat = 8;
- ctx->tex_fun = ctx->dxtc.rgtc1u_gray_block;
+ ctx->dec[0].tex_ratio = 8;
+ ctx->dec[0].tex_funct = ctx->dxtc.rgtc1u_gray_block;
+ ctx->dec[0].raw_ratio = 4;
avctx->pix_fmt = AV_PIX_FMT_GRAY8;
- ctx->uncompress_pix_size = 1;
break;
case MKTAG('H','a','p','M'):
texture_name = "DXT5-YCoCg-scaled / RGTC1";
- ctx->tex_rat = 16;
- ctx->tex_rat2 = 8;
- ctx->tex_fun = ctx->dxtc.dxt5ys_block;
- ctx->tex_fun2 = ctx->dxtc.rgtc1u_alpha_block;
+ ctx->dec[0].tex_ratio = 16;
+ ctx->dec[1].tex_ratio = 8;
+ ctx->dec[0].tex_funct = ctx->dxtc.dxt5ys_block;
+ ctx->dec[1].tex_funct = ctx->dxtc.rgtc1u_alpha_block;
+ ctx->dec[1].raw_ratio = 16;
+ ctx->dec[1].slice_count = ctx->dec[0].slice_count;
avctx->pix_fmt = AV_PIX_FMT_RGBA;
ctx->texture_count = 2;
break;
@@ -458,9 +401,6 @@ static av_cold int hap_init(AVCodecContext *avctx)
av_log(avctx, AV_LOG_DEBUG, "%s texture\n", texture_name);
- ctx->slice_count = av_clip(avctx->thread_count, 1,
- avctx->coded_height / TEXTURE_BLOCK_H);
-
return 0;
}
--
2.31.1
More information about the ffmpeg-devel
mailing list