rockbox/apps/codecs/libffmpegFLAC/shndec.c
Dave Chapman 8e46ab85a9 Patch #1426489 - Shorten codec optimisations from Mark Arigo
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@8615 a1c6a512-1295-4272-9138-f99709370657
2006-02-07 22:16:35 +00:00

493 lines
14 KiB
C

/*
* Shorten decoder
* Copyright (c) 2005 Jeff Muizelaar
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
/**
* @file shorten.c
* Shorten decoder
* @author Jeff Muizelaar
*
*/
#include "bitstream.h"
#include "golomb.h"
#include "shndec.h"
#define ULONGSIZE 2
#define WAVE_FORMAT_PCM 0x0001
#define TYPESIZE 4
#define CHANSIZE 0
#define LPCQSIZE 2
#define ENERGYSIZE 3
#define BITSHIFTSIZE 2
#define TYPE_S16HL 3 /* signed 16 bit shorts: high-low */
#define TYPE_S16LH 5 /* signed 16 bit shorts: low-high */
#define NWRAP 3
#define NSKIPSIZE 1
#define LPCQUANT 5
#define V2LPCQOFFSET (1 << LPCQUANT)
#define FNSIZE 2
#define VERBATIM_CKSIZE_SIZE 5
#define VERBATIM_BYTE_SIZE 8
#define CANONICAL_HEADER_SIZE 44
#define FFMAX(a,b) ((a) > (b) ? (a) : (b))
#define FFMIN(a,b) ((a) > (b) ? (b) : (a))
#define MKTAG(a,b,c,d) (a | (b << 8) | (c << 16) | (d << 24))
#define get_le16(gb) bswap_16(get_bits_long(gb, 16))
#define get_le32(gb) bswap_32(get_bits_long(gb, 32))
static uint32_t bswap_32(uint32_t x){
x= ((x<<8)&0xFF00FF00) | ((x>>8)&0x00FF00FF);
return (x>>16) | (x<<16);
}
static uint16_t bswap_16(uint16_t x){
return (x>>8) | (x<<8);
}
/* converts fourcc string to int */
static int ff_get_fourcc(const char *s){
//assert( strlen(s)==4 );
return (s[0]) + (s[1]<<8) + (s[2]<<16) + (s[3]<<24);
}
static unsigned int get_uint(ShortenContext *s, int k)
{
if (s->version != 0)
k = get_ur_golomb_shorten(&s->gb, ULONGSIZE);
return get_ur_golomb_shorten(&s->gb, k);
}
#if defined(CPU_COLDFIRE) && !defined(SIMULATOR)
static void coldfire_lshift_samples(int n, int shift, int32_t *samples) ICODE_ATTR;
static void coldfire_lshift_samples(int n, int shift, int32_t *samples)
{
/*
for (i = 0; i < n; i++)
samples[i] =<< shift;
*/
asm volatile (
"move.l %[n], %%d0 \n" /* d0 = loop counter */
"asr.l #2, %%d0 \n"
"beq l1_shift \n"
"l2_shift:" /* main loop (unroll by 4) */
"movem.l (%[x]), %%d4-%%d7 \n"
"asl.l %[s], %%d4 \n"
"asl.l %[s], %%d5 \n"
"asl.l %[s], %%d6 \n"
"asl.l %[s], %%d7 \n"
"movem.l %%d4-%%d7, (%[x]) \n"
"add.l #16, %[x] \n"
"subq.l #1, %%d0 \n"
"bne l2_shift \n"
"l1_shift:" /* any loops left? */
"and.l #3, %[n] \n"
"beq l4_shift \n"
"l3_shift:" /* remaining loops */
"move.l (%[x]), %%d4 \n"
"asl.l %[s], %%d4 \n"
"move.l %%d4, (%[x])+ \n"
"subq.l #1, %[n] \n"
"bne l3_shift \n"
"l4_shift:" /* exit */
: [n] "+d" (n), /* d1 */
[s] "+d" (shift), /* d2 */
[x] "+a" (samples) /* a0 */
:
: "%d0", "%d4", "%d5", "%d6", "%d7"
);
}
#endif
static inline void fix_bitshift(ShortenContext *s, int32_t *samples)
{
int i;
/* Wrapped samples don't get bitshifted, so we'll do them during
the next iteration. */
if (s->bitshift != 0) {
#if defined(CPU_COLDFIRE) && !defined(SIMULATOR)
coldfire_lshift_samples(s->blocksize, s->bitshift, samples - s->nwrap);
#else
for (i = -s->nwrap; i < (s->blocksize - s->nwrap); i++)
samples[i] <<= s->bitshift;
#endif
}
/* Also, when we have to remember to fix the wrapped samples when
the bitshift changes.*/
if (s->bitshift != s->last_bitshift) {
if (s->last_bitshift != 0)
for (i = -s->nwrap; i < 0; i++)
samples[i] <<= s->last_bitshift;
s->last_bitshift = s->bitshift;
}
}
static inline void decode_subframe_lpc(ShortenContext *s, int32_t *decoded,
int residual_size, int pred_order)
{
int sum, i, j;
int coeffs[MAX_PRED_ORDER];
for (i=0; i<pred_order; i++) {
coeffs[i] = get_sr_golomb_shorten(&s->gb, LPCQUANT);
}
for (i=0; i < s->blocksize; i++) {
sum = s->lpcqoffset;
for (j=0; j<pred_order; j++)
sum += coeffs[j] * decoded[i-j-1];
decoded[i] =
get_sr_golomb_shorten(&s->gb, residual_size) + (sum >> LPCQUANT);
}
}
static inline int shorten_decode_frame(ShortenContext *s, int32_t *decoded,
int32_t *offset)
{
int i;
int32_t sum;
int cmd = get_ur_golomb_shorten(&s->gb, FNSIZE);
switch (cmd) {
case FN_ZERO:
case FN_DIFF0:
case FN_DIFF1:
case FN_DIFF2:
case FN_DIFF3:
case FN_QLPC:
{
int residual_size = 0;
int32_t coffset;
if (cmd != FN_ZERO) {
residual_size = get_ur_golomb_shorten(&s->gb, ENERGYSIZE);
/* this is a hack as version 0 differed in defintion of
get_sr_golomb_shorten */
if (s->version == 0)
residual_size--;
}
if (s->nmean == 0) {
coffset = offset[0];
} else {
sum = (s->version < 2) ? 0 : s->nmean / 2;
for (i=0; i<s->nmean; i++)
sum += offset[i];
coffset = sum / s->nmean;
if (s->version >= 2)
coffset >>= FFMIN(1, s->bitshift);
}
switch (cmd) {
case FN_ZERO:
for (i=0; i<s->blocksize; i++)
decoded[i] = 0;
break;
case FN_DIFF0:
for (i=0; i<s->blocksize; i++)
decoded[i] =
get_sr_golomb_shorten(&s->gb, residual_size) +
coffset;
break;
case FN_DIFF1:
for (i=0; i<s->blocksize; i++)
decoded[i] =
get_sr_golomb_shorten(&s->gb, residual_size) +
decoded[i - 1];
break;
case FN_DIFF2:
for (i=0; i<s->blocksize; i++)
decoded[i] =
get_sr_golomb_shorten(&s->gb, residual_size) +
2*decoded[i-1] - decoded[i-2];
break;
case FN_DIFF3:
for (i=0; i<s->blocksize; i++)
decoded[i] =
get_sr_golomb_shorten(&s->gb, residual_size) +
3*decoded[i-1] - 3*decoded[i-2] + decoded[i-3];
break;
case FN_QLPC:
{
int pred_order = get_ur_golomb_shorten(&s->gb, LPCQSIZE);
for (i=0; i<pred_order; i++)
decoded[i - pred_order] -= coffset;
decode_subframe_lpc(s, decoded, residual_size, pred_order);
if (coffset != 0) {
for (i=0; i < s->blocksize; i++)
decoded[i] += coffset;
}
}
}
if (s->nmean > 0) {
sum = (s->version < 2) ? 0 : s->blocksize / 2;
for (i=0; i<s->blocksize; i++)
sum += decoded[i];
for (i=1; i<s->nmean; i++)
offset[i-1] = offset[i];
if (s->version < 2) {
offset[s->nmean - 1] = sum / s->blocksize;
} else {
offset[s->nmean - 1] =
(sum / s->blocksize) << s->bitshift;
}
}
fix_bitshift(s, decoded);
break;
}
case FN_VERBATIM:
i = get_ur_golomb_shorten(&s->gb, VERBATIM_CKSIZE_SIZE);
while (i--)
get_ur_golomb_shorten(&s->gb, VERBATIM_BYTE_SIZE);
break;
case FN_BITSHIFT:
s->bitshift = get_ur_golomb_shorten(&s->gb, BITSHIFTSIZE);
break;
case FN_BLOCKSIZE:
s->blocksize = get_uint(s, av_log2(s->blocksize));
break;
case FN_QUIT:
break;
default:
return FN_ERROR;
break;
}
return cmd;
}
int shorten_decode_frames(ShortenContext *s, int *nsamples,
int32_t *decoded0, int32_t *decoded1,
int32_t *offset0, int32_t *offset1,
uint8_t *buf, int buf_size,
void (*yield)(void))
{
int32_t *decoded, *offset;
int cmd;
*nsamples = 0;
init_get_bits(&s->gb, buf, buf_size*8);
get_bits(&s->gb, s->bitindex);
int n = 0;
while (n < NUM_DEC_LOOPS) {
int chan = n%2;
if (chan == 0) {
decoded = decoded0 + s->nwrap + *nsamples;
offset = offset0;
} else {
decoded = decoded1 + s->nwrap + *nsamples;
offset = offset1;
}
yield();
cmd = shorten_decode_frame(s, decoded, offset);
if (cmd == FN_VERBATIM || cmd == FN_BITSHIFT || cmd == FN_BLOCKSIZE) {
continue;
} else if (cmd == FN_QUIT || cmd == FN_ERROR) {
break;
}
*nsamples += chan * s->blocksize;
n++;
}
if (*nsamples) {
/* Wrap the samples for the next loop */
int i;
for (i = 0; i < s->nwrap; i++) {
decoded0[i] = decoded0[*nsamples + i];
decoded1[i] = decoded1[*nsamples + i];
}
/* Scale the samples for the pcmbuf */
int scale = SHN_OUTPUT_DEPTH - s->bits_per_sample;
#if defined(CPU_COLDFIRE) && !defined(SIMULATOR)
coldfire_lshift_samples(*nsamples, scale, decoded0 + s->nwrap);
coldfire_lshift_samples(*nsamples, scale, decoded1 + s->nwrap);
#else
for (i = 0; i < *nsamples; i++) {
decoded0[i + s->nwrap] <<= scale;
decoded1[i + s->nwrap] <<= scale;
}
#endif
}
return cmd;
}
static int decode_wave_header(ShortenContext *s,
uint8_t *header,
int header_size)
{
GetBitContext hb;
int len;
init_get_bits(&hb, header, header_size*8);
if (get_le32(&hb) != MKTAG('R','I','F','F')) {
return -8;
}
int chunk_size = get_le32(&hb);
if (get_le32(&hb) != MKTAG('W','A','V','E')) {
return -9;
}
while (get_le32(&hb) != MKTAG('f','m','t',' ')) {
len = get_le32(&hb);
skip_bits(&hb, 8*len);
}
len = get_le32(&hb);
if (len < 16) {
return -10;
}
if (get_le16(&hb) != WAVE_FORMAT_PCM ) {
return -11;
}
s->channels = get_le16(&hb);
if (s->channels > MAX_CHANNELS) {
return -3;
}
s->sample_rate = get_le32(&hb);
skip_bits(&hb, 32);
//s->bit_rate = 8*get_le32(&hb);
int block_align = get_le16(&hb);
s->totalsamples = (chunk_size - 36) / block_align;
s->bits_per_sample = get_le16(&hb);
if (s->bits_per_sample != 16) {
return -12;
}
len -= 16;
if (len > 0) {
return len;
}
return 0;
}
int shorten_init(ShortenContext* s, uint8_t *buf, int buf_size)
{
int i;
s->blocksize = DEFAULT_BLOCK_SIZE;
s->channels = 1;
s->nmean = -1;
init_get_bits(&s->gb, buf, buf_size*8);
get_bits(&s->gb, s->bitindex);
/* shorten signature */
if (get_bits_long(&s->gb, 32) != bswap_32(ff_get_fourcc("ajkg"))) {
return -1;
}
s->version = get_bits(&s->gb, 8);
int internal_ftype = get_uint(s, TYPESIZE);
if ((internal_ftype != TYPE_S16HL) && (internal_ftype != TYPE_S16LH)) {
return -2;
}
s->channels = get_uint(s, CHANSIZE);
if (s->channels > MAX_CHANNELS) {
return -3;
}
/* get blocksize if version > 0 */
int maxnlpc = 0;
if (s->version > 0) {
s->blocksize = get_uint(s, av_log2(DEFAULT_BLOCK_SIZE));
maxnlpc = get_uint(s, LPCQSIZE);
s->nmean = get_uint(s, 0);
int skip_bytes = get_uint(s, NSKIPSIZE);
for (i=0; i<skip_bytes; i++) {
skip_bits(&s->gb, 8);
}
}
if (s->nmean > MAX_NMEAN) {
return -4;
}
s->nwrap = FFMAX(NWRAP, maxnlpc);
if (s->nwrap > MAX_NWRAP) {
return -5;
}
if (s->version > 1)
s->lpcqoffset = V2LPCQOFFSET;
if (get_ur_golomb_shorten(&s->gb, FNSIZE) != FN_VERBATIM) {
return -6;
}
uint8_t header[MAX_HEADER_SIZE];
int header_size = get_ur_golomb_shorten(&s->gb, VERBATIM_CKSIZE_SIZE);
if (header_size >= MAX_HEADER_SIZE || header_size < CANONICAL_HEADER_SIZE) {
return -7;
}
for (i=0; i<header_size; i++)
header[i] = (char)get_ur_golomb_shorten(&s->gb, VERBATIM_BYTE_SIZE);
s->header_bits = s->gb.index;
return decode_wave_header(s, header, header_size);
}