[Ffmpeg-devel] [RFC] smallcpy for h264

[Luca Barbato]

Michael Niedermayer wrote:
> 
> rejected, the only memcpy i have found which touch these are in the init code

Ok, looks like I should do my homework better, still I'm puzzled by the
effect of the patch.

decode_ref_pic_list_reordering
	memcpy(h->ref_list[list], h->default_ref_list[list],
		sizeof(Picture)*h->ref_count[list]);

fill_mbaff_ref_list
	memcpy(h->implicit_weight[16+2*j],   h->implicit_weight[j], 		
		sizeof(*h->implicit_weight));
	memcpy(h->implicit_weight[16+2*j+1], h->implicit_weight[j],
		sizeof(*h->implicit_weight));

decode_slice_header
	memcpy(h->zigzag_scan, zigzag_scan, 16*sizeof(uint8_t));
	memcpy(h-> field_scan,  field_scan, 16*sizeof(uint8_t));
	memcpy(h->zigzag_scan8x8,       zigzag_scan8x8,       			
		64*sizeof(uint8_t));
	memcpy(h->zigzag_scan8x8_cavlc, zigzag_scan8x8_cavlc,
		64*sizeof(uint8_t));
	memcpy(h->field_scan8x8,        field_scan8x8,
		64*sizeof(uint8_t));
	memcpy(h->field_scan8x8_cavlc,  field_scan8x8_cavlc,
		64*sizeof(uint8_t));

decode_scaling_list
	memcpy(factors, fallback_list, size*sizeof(uint8_t));
	memcpy(factors, jvt_list, size*sizeof(uint8_t));

decode_scaling_matrices
	memcpy(scaling_matrix4, sps->scaling_matrix4,
		6*16*sizeof(uint8_t));
	memcpy(scaling_matrix8, sps->scaling_matrix8,
		2*64*sizeof(uint8_t));

lu - that will start drinking more cola.

-- 

Luca Barbato

Gentoo/linux Gentoo/PPC
http://dev.gentoo.org/~lu_zero