Commit bd2b6b33 authored by Måns Rullgård's avatar Måns Rullgård

snow: move scratch buffer from stack to malloced buffer in context

This is required to ensure 16-byte alignment.

Originally committed as revision 15934 to svn://svn.ffmpeg.org/ffmpeg/trunk
parent ab56b321
...@@ -488,6 +488,8 @@ typedef struct SnowContext{ ...@@ -488,6 +488,8 @@ typedef struct SnowContext{
slice_buffer sb; slice_buffer sb;
MpegEncContext m; // needed for motion estimation, should not be used for anything else, the idea is to eventually make the motion estimation independent of MpegEncContext, so this will be removed then (FIXME/XXX) MpegEncContext m; // needed for motion estimation, should not be used for anything else, the idea is to eventually make the motion estimation independent of MpegEncContext, so this will be removed then (FIXME/XXX)
uint8_t *scratchbuf;
}SnowContext; }SnowContext;
typedef struct { typedef struct {
...@@ -2423,7 +2425,7 @@ static av_always_inline void add_yblock(SnowContext *s, int sliced, slice_buffer ...@@ -2423,7 +2425,7 @@ static av_always_inline void add_yblock(SnowContext *s, int sliced, slice_buffer
BlockNode *rb= lb+1; BlockNode *rb= lb+1;
uint8_t *block[4]; uint8_t *block[4];
int tmp_step= src_stride >= 7*MB_SIZE ? MB_SIZE : MB_SIZE*src_stride; int tmp_step= src_stride >= 7*MB_SIZE ? MB_SIZE : MB_SIZE*src_stride;
uint8_t tmp[src_stride*7*MB_SIZE]; //FIXME align uint8_t *tmp = s->scratchbuf;
uint8_t *ptmp; uint8_t *ptmp;
int x,y; int x,y;
...@@ -2785,7 +2787,7 @@ static int get_block_rd(SnowContext *s, int mb_x, int mb_y, int plane_index, con ...@@ -2785,7 +2787,7 @@ static int get_block_rd(SnowContext *s, int mb_x, int mb_y, int plane_index, con
uint8_t *dst= s->current_picture.data[plane_index]; uint8_t *dst= s->current_picture.data[plane_index];
uint8_t *src= s-> input_picture.data[plane_index]; uint8_t *src= s-> input_picture.data[plane_index];
IDWTELEM *pred= (IDWTELEM*)s->m.obmc_scratchpad + plane_index*block_size*block_size*4; IDWTELEM *pred= (IDWTELEM*)s->m.obmc_scratchpad + plane_index*block_size*block_size*4;
uint8_t cur[ref_stride*2*MB_SIZE]; //FIXME alignment uint8_t *cur = s->scratchbuf;
uint8_t tmp[ref_stride*(2*MB_SIZE+HTAPS_MAX-1)]; uint8_t tmp[ref_stride*(2*MB_SIZE+HTAPS_MAX-1)];
const int b_stride = s->b_width << s->block_max_depth; const int b_stride = s->b_width << s->block_max_depth;
const int b_height = s->b_height<< s->block_max_depth; const int b_height = s->b_height<< s->block_max_depth;
...@@ -3703,6 +3705,7 @@ static av_cold int common_init(AVCodecContext *avctx){ ...@@ -3703,6 +3705,7 @@ static av_cold int common_init(AVCodecContext *avctx){
scale_mv_ref[i][j] = 256*(i+1)/(j+1); scale_mv_ref[i][j] = 256*(i+1)/(j+1);
s->avctx->get_buffer(s->avctx, &s->mconly_picture); s->avctx->get_buffer(s->avctx, &s->mconly_picture);
s->scratchbuf = av_malloc(s->mconly_picture.linesize[0]*7*MB_SIZE);
return 0; return 0;
} }
...@@ -4438,6 +4441,7 @@ static av_cold void common_end(SnowContext *s){ ...@@ -4438,6 +4441,7 @@ static av_cold void common_end(SnowContext *s){
av_freep(&s->m.obmc_scratchpad); av_freep(&s->m.obmc_scratchpad);
av_freep(&s->block); av_freep(&s->block);
av_freep(&s->scratchbuf);
for(i=0; i<MAX_REF_FRAMES; i++){ for(i=0; i<MAX_REF_FRAMES; i++){
av_freep(&s->ref_mvs[i]); av_freep(&s->ref_mvs[i]);
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment