On 13/06/16 18:08, Diego Biurrun wrote: > On Fri, Jun 10, 2016 at 10:25:38AM +0200, Anton Khirnov wrote: >> Quoting Anton Khirnov (2016-06-10 10:16:00) >>> Quoting Diego Biurrun (2016-06-09 17:11:46) >>>> From: Alexandra Hájková <[email protected]> >>>> --- a/libavcodec/eamad.c >>>> +++ b/libavcodec/eamad.c >>>> @@ -51,7 +52,7 @@ typedef struct MadContext { >>>> BswapDSPContext bbdsp; >>>> IDCTDSPContext idsp; >>>> AVFrame *last_frame; >>>> - GetBitContext gb; >>>> + BitstreamContext bc; >>>> void *bitstream_buf; >>>> unsigned int bitstream_buf_size; >>>> DECLARE_ALIGNED(16, int16_t, block)[64]; >>>> @@ -129,17 +130,15 @@ static inline void decode_block_intra(MadContext *s, >>>> int16_t * block) >>>> const uint8_t *scantable = s->scantable.permutated; >>>> int16_t *quant_matrix = s->quant_matrix; >>>> >>>> - block[0] = (128 + get_sbits(&s->gb, 8)) * quant_matrix[0]; >>>> + block[0] = (128 + bitstream_read_signed(&s->bc, 8)) * quant_matrix[0]; >>>> >>>> /* The RL decoder is derived from mpeg1_decode_block_intra; >>>> Escaped level and run values a decoded differently */ >>>> i = 0; >>>> { >>>> - OPEN_READER(re, &s->gb); >>>> /* now quantify & encode AC coefficients */ >>>> for (;;) { >>>> - UPDATE_CACHE(re, &s->gb); >>>> - GET_RL_VLC(level, run, re, &s->gb, rl->rl_vlc[0], >>>> TEX_VLC_BITS, 2, 0); >>>> + BITSTREAM_RL_VLC(level, run, &s->bc, rl->rl_vlc[0], >>>> TEX_VLC_BITS, 2); >>>> >>>> if (level == 127) { >>>> break; >>>> @@ -153,15 +152,13 @@ static inline void decode_block_intra(MadContext *s, >>>> int16_t * block) >>>> j = scantable[i]; >>>> level = (level*quant_matrix[j]) >> 4; >>>> level = (level-1)|1; >>>> - level = (level ^ SHOW_SBITS(re, &s->gb, 1)) - >>>> SHOW_SBITS(re, &s->gb, 1); >>>> - LAST_SKIP_BITS(re, &s->gb, 1); >>>> + level = (level ^ bitstream_peek_signed(&s->bc, 1)) - >>>> bitstream_peek_signed(&s->bc, 1); >>>> + bitstream_skip(&s->bc, 1); >>> >>> This looks rather awkward. Isn't it identical to: >>> int sign = bitstream_read(&s->bc, 1); >> ^^^^ >> read_signed of course > > You're probably right, but your suggestion seems to go beyond the scope > of this patch. >
I'd spun it off but apply it since it is much easier to read. lu _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
