gaming/godot
gaming/godot
1
0
Fork 0
mirror of https://github.com/godotengine/godot synced 2024-09-21 10:56:31 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge remote-tracking branch 'upstream/master'

Browse source
This commit is contained in:
Kostadin Damyanov 2015-10-13 05:01:54 +03:00
parent c5f574b914 79e5ced7e6
commit 29caf2bb5b
634 changed files with 116816 additions and 91751 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
.gitattributes vendored Normal file
View file

@ -0,0 +1,4 @@
*.cpp eol=lf
*.h eol=lf
*.py eol=lf
*.hpp eol=lf

10
.gitignore vendored
View file

@ -50,6 +50,14 @@ platform/android/libs/play_licensing/gen/*
*.d
*.so
*.os
*.Plo
*.lo
*.Po
# Libs generated files
.deps/*
.dirstamp
# QT project files
*.config
@ -282,4 +290,4 @@ cscope.in.out
cscope.po.out
godot.creator.*
projects/
projects/

3
SConstruct
View file

@ -102,6 +102,7 @@ opts.Add('p','Platform (same as platform=).',"")
opts.Add('tools','Build Tools (Including Editor): (yes/no)','yes')
opts.Add('gdscript','Build GDSCript support: (yes/no)','yes')
opts.Add('vorbis','Build Ogg Vorbis Support: (yes/no)','yes')
opts.Add('opus','Build Opus Audio Format Support: (yes/no)','yes')
opts.Add('minizip','Build Minizip Archive Support: (yes/no)','yes')
opts.Add('squish','Squish BC Texture Compression in editor (yes/no)','yes')
opts.Add('theora','Theora Video (yes/no)','yes')
@ -299,6 +300,8 @@ if selected_platform in platform_list:
if (env['vorbis']=='yes'):
env.Append(CPPFLAGS=['-DVORBIS_ENABLED']);
if (env['opus']=='yes'):
env.Append(CPPFLAGS=['-DOPUS_ENABLED']);
if (env['theora']=='yes'):
env.Append(CPPFLAGS=['-DTHEORA_ENABLED']);

6
core/bind/core_bind.cpp
View file

@ -732,6 +732,11 @@ int _OS::find_scancode_from_string(const String& p_code) const {
return find_keycode(p_code);
}
void _OS::alert(const String& p_alert,const String& p_title) {
OS::get_singleton()->alert(p_alert,p_title);
}
_OS *_OS::singleton=NULL;
void _OS::_bind_methods() {
@ -859,6 +864,7 @@ void _OS::_bind_methods() {
ObjectTypeDB::bind_method(_MD("set_use_file_access_save_and_swap","enabled"),&_OS::set_use_file_access_save_and_swap);
ObjectTypeDB::bind_method(_MD("alert","text","title"),&_OS::alert,DEFVAL("Alert!"));
BIND_CONSTANT( DAY_SUNDAY );

3
core/bind/core_bind.h
View file

@ -256,6 +256,9 @@ public:
String get_data_dir() const;
void alert(const String& p_alert,const String& p_title="ALERT!");
void set_screen_orientation(ScreenOrientation p_orientation);
ScreenOrientation get_screen_orientation() const;

107
core/image.cpp
View file

@ -400,6 +400,102 @@ Image::Format Image::get_format() const{
return format;
}
static double _bicubic_interp_kernel( double x ) {
x = ABS(x);
double bc = 0;
if ( x <= 1 )
bc = ( 1.5 * x - 2.5 ) * x * x + 1;
else if ( x < 2 )
bc = ( ( -0.5 * x + 2.5 ) * x - 4 ) * x + 2;
return bc;
}
template<int CC>
static void _scale_cubic(const uint8_t* p_src, uint8_t* p_dst, uint32_t p_src_width, uint32_t p_src_height, uint32_t p_dst_width, uint32_t p_dst_height) {
// get source image size
int width = p_src_width;
int height = p_src_height;
double xfac = (double) width / p_dst_width;
double yfac = (double) height / p_dst_height;
// coordinates of source points and cooefficiens
double ox, oy, dx, dy, k1, k2;
int ox1, oy1, ox2, oy2;
// destination pixel values
// width and height decreased by 1
int ymax = height - 1;
int xmax = width - 1;
// temporary pointer
for ( int y = 0; y < p_dst_height; y++ ) {
// Y coordinates
oy = (double) y * yfac - 0.5f;
oy1 = (int) oy;
dy = oy - (double) oy1;
for ( int x = 0; x < p_dst_width; x++ ) {
// X coordinates
ox = (double) x * xfac - 0.5f;
ox1 = (int) ox;
dx = ox - (double) ox1;
// initial pixel value
uint8_t *dst=p_dst + (y*p_dst_width+x)*CC;
double color[CC];
for(int i=0;i<CC;i++) {
color[i]=0;
}
for ( int n = -1; n < 3; n++ ) {
// get Y cooefficient
k1 = _bicubic_interp_kernel( dy - (double) n );
oy2 = oy1 + n;
if ( oy2 < 0 )
oy2 = 0;
if ( oy2 > ymax )
oy2 = ymax;
for ( int m = -1; m < 3; m++ ) {
// get X cooefficient
k2 = k1 * _bicubic_interp_kernel( (double) m - dx );
ox2 = ox1 + m;
if ( ox2 < 0 )
ox2 = 0;
if ( ox2 > xmax )
ox2 = xmax;
// get pixel of original image
const uint8_t *p = p_src + (oy2 * p_src_width + ox2)*CC;
for(int i=0;i<CC;i++) {
color[i]+=p[i]*k2;
}
}
}
for(int i=0;i<CC;i++) {
dst[i]=CLAMP(Math::fast_ftoi(color[i]),0,255);
}
}
}
}
template<int CC>
static void _scale_bilinear(const uint8_t* p_src, uint8_t* p_dst, uint32_t p_src_width, uint32_t p_src_height, uint32_t p_dst_width, uint32_t p_dst_height) {
@ -559,6 +655,17 @@ void Image::resize( int p_width, int p_height, Interpolation p_interpolation ) {
}
} break;
case INTERPOLATE_CUBIC: {
switch(get_format_pixel_size(format)) {
case 1: _scale_cubic<1>(r_ptr,w_ptr,width,height,p_width,p_height); break;
case 2: _scale_cubic<2>(r_ptr,w_ptr,width,height,p_width,p_height); break;
case 3: _scale_cubic<3>(r_ptr,w_ptr,width,height,p_width,p_height); break;
case 4: _scale_cubic<4>(r_ptr,w_ptr,width,height,p_width,p_height); break;
}
} break;
}

1
core/image.h
View file

@ -91,6 +91,7 @@ public:
INTERPOLATE_NEAREST,
INTERPOLATE_BILINEAR,
INTERPOLATE_CUBIC,
/* INTERPOLATE GAUSS */
};

718
core/io/aes256.cpp
View file

@ -1,359 +1,359 @@
/*
* Byte-oriented AES-256 implementation.
* All lookup tables replaced with 'on the fly' calculations.
*
* Copyright (c) 2007-2009 Ilya O. Levin, http://www.literatecode.com
* Other contributors: Hal Finney
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
#include "aes256.h"
#define F(x) (((x)<<1) ^ ((((x)>>7) & 1) * 0x1b))
#define FD(x) (((x) >> 1) ^ (((x) & 1) ? 0x8d : 0))
// #define BACK_TO_TABLES
#ifdef BACK_TO_TABLES
const uint8_t sbox[256] = {
0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5,
0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76,
0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0,
0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0,
0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc,
0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15,
0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a,
0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75,
0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0,
0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84,
0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b,
0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf,
0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85,
0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8,
0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5,
0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2,
0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17,
0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73,
0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88,
0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb,
0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c,
0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79,
0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9,
0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08,
0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6,
0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a,
0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e,
0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e,
0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94,
0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf,
0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68,
0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16
};
const uint8_t sboxinv[256] = {
0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38,
0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb,
0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87,
0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb,
0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d,
0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e,
0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2,
0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25,
0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16,
0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92,
0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda,
0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84,
0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a,
0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06,
0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02,
0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b,
0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea,
0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73,
0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85,
0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e,
0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89,
0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b,
0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20,
0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4,
0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31,
0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f,
0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d,
0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef,
0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0,
0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61,
0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26,
0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d
};
#define rj_sbox(x) sbox[(x)]
#define rj_sbox_inv(x) sboxinv[(x)]
#else /* tableless subroutines */
/* -------------------------------------------------------------------------- */
uint8_t gf_alog(uint8_t x) // calculate anti-logarithm gen 3
{
uint8_t atb = 1, z;
while (x--) {z = atb; atb <<= 1; if (z & 0x80) atb^= 0x1b; atb ^= z;}
return atb;
} /* gf_alog */
/* -------------------------------------------------------------------------- */
uint8_t gf_log(uint8_t x) // calculate logarithm gen 3
{
uint8_t atb = 1, i = 0, z;
do {
if (atb == x) break;
z = atb; atb <<= 1; if (z & 0x80) atb^= 0x1b; atb ^= z;
} while (++i > 0);
return i;
} /* gf_log */
/* -------------------------------------------------------------------------- */
uint8_t gf_mulinv(uint8_t x) // calculate multiplicative inverse
{
return (x) ? gf_alog(255 - gf_log(x)) : 0;
} /* gf_mulinv */
/* -------------------------------------------------------------------------- */
uint8_t rj_sbox(uint8_t x)
{
uint8_t y, sb;
sb = y = gf_mulinv(x);
y = (y<<1)|(y>>7); sb ^= y; y = (y<<1)|(y>>7); sb ^= y;
y = (y<<1)|(y>>7); sb ^= y; y = (y<<1)|(y>>7); sb ^= y;
return (sb ^ 0x63);
} /* rj_sbox */
/* -------------------------------------------------------------------------- */
uint8_t rj_sbox_inv(uint8_t x)
{
uint8_t y, sb;
y = x ^ 0x63;
sb = y = (y<<1)|(y>>7);
y = (y<<2)|(y>>6); sb ^= y; y = (y<<3)|(y>>5); sb ^= y;
return gf_mulinv(sb);
} /* rj_sbox_inv */
#endif
/* -------------------------------------------------------------------------- */
uint8_t rj_xtime(uint8_t x)
{
return (x & 0x80) ? ((x << 1) ^ 0x1b) : (x << 1);
} /* rj_xtime */
/* -------------------------------------------------------------------------- */
void aes_subBytes(uint8_t *buf)
{
register uint8_t i = 16;
while (i--) buf[i] = rj_sbox(buf[i]);
} /* aes_subBytes */
/* -------------------------------------------------------------------------- */
void aes_subBytes_inv(uint8_t *buf)
{
register uint8_t i = 16;
while (i--) buf[i] = rj_sbox_inv(buf[i]);
} /* aes_subBytes_inv */
/* -------------------------------------------------------------------------- */
void aes_addRoundKey(uint8_t *buf, uint8_t *key)
{
register uint8_t i = 16;
while (i--) buf[i] ^= key[i];
} /* aes_addRoundKey */
/* -------------------------------------------------------------------------- */
void aes_addRoundKey_cpy(uint8_t *buf, uint8_t *key, uint8_t *cpk)
{
register uint8_t i = 16;
while (i--) buf[i] ^= (cpk[i] = key[i]), cpk[16+i] = key[16 + i];
} /* aes_addRoundKey_cpy */
/* -------------------------------------------------------------------------- */
void aes_shiftRows(uint8_t *buf)
{
register uint8_t i, j; /* to make it potentially parallelable :) */
i = buf[1]; buf[1] = buf[5]; buf[5] = buf[9]; buf[9] = buf[13]; buf[13] = i;
i = buf[10]; buf[10] = buf[2]; buf[2] = i;
j = buf[3]; buf[3] = buf[15]; buf[15] = buf[11]; buf[11] = buf[7]; buf[7] = j;
j = buf[14]; buf[14] = buf[6]; buf[6] = j;
} /* aes_shiftRows */
/* -------------------------------------------------------------------------- */
void aes_shiftRows_inv(uint8_t *buf)
{
register uint8_t i, j; /* same as above :) */
i = buf[1]; buf[1] = buf[13]; buf[13] = buf[9]; buf[9] = buf[5]; buf[5] = i;
i = buf[2]; buf[2] = buf[10]; buf[10] = i;
j = buf[3]; buf[3] = buf[7]; buf[7] = buf[11]; buf[11] = buf[15]; buf[15] = j;
j = buf[6]; buf[6] = buf[14]; buf[14] = j;
} /* aes_shiftRows_inv */
/* -------------------------------------------------------------------------- */
void aes_mixColumns(uint8_t *buf)
{
register uint8_t i, a, b, c, d, e;
for (i = 0; i < 16; i += 4)
{
a = buf[i]; b = buf[i + 1]; c = buf[i + 2]; d = buf[i + 3];
e = a ^ b ^ c ^ d;
buf[i] ^= e ^ rj_xtime(a^b); buf[i+1] ^= e ^ rj_xtime(b^c);
buf[i+2] ^= e ^ rj_xtime(c^d); buf[i+3] ^= e ^ rj_xtime(d^a);
}
} /* aes_mixColumns */
/* -------------------------------------------------------------------------- */
void aes_mixColumns_inv(uint8_t *buf)
{
register uint8_t i, a, b, c, d, e, x, y, z;
for (i = 0; i < 16; i += 4)
{
a = buf[i]; b = buf[i + 1]; c = buf[i + 2]; d = buf[i + 3];
e = a ^ b ^ c ^ d;
z = rj_xtime(e);
x = e ^ rj_xtime(rj_xtime(z^a^c)); y = e ^ rj_xtime(rj_xtime(z^b^d));
buf[i] ^= x ^ rj_xtime(a^b); buf[i+1] ^= y ^ rj_xtime(b^c);
buf[i+2] ^= x ^ rj_xtime(c^d); buf[i+3] ^= y ^ rj_xtime(d^a);
}
} /* aes_mixColumns_inv */
/* -------------------------------------------------------------------------- */
void aes_expandEncKey(uint8_t *k, uint8_t *rc)
{
register uint8_t i;
k[0] ^= rj_sbox(k[29]) ^ (*rc);
k[1] ^= rj_sbox(k[30]);
k[2] ^= rj_sbox(k[31]);
k[3] ^= rj_sbox(k[28]);
*rc = F( *rc);
for(i = 4; i < 16; i += 4) k[i] ^= k[i-4], k[i+1] ^= k[i-3],
k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
k[16] ^= rj_sbox(k[12]);
k[17] ^= rj_sbox(k[13]);
k[18] ^= rj_sbox(k[14]);
k[19] ^= rj_sbox(k[15]);
for(i = 20; i < 32; i += 4) k[i] ^= k[i-4], k[i+1] ^= k[i-3],
k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
} /* aes_expandEncKey */
/* -------------------------------------------------------------------------- */
void aes_expandDecKey(uint8_t *k, uint8_t *rc)
{
uint8_t i;
for(i = 28; i > 16; i -= 4) k[i+0] ^= k[i-4], k[i+1] ^= k[i-3],
k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
k[16] ^= rj_sbox(k[12]);
k[17] ^= rj_sbox(k[13]);
k[18] ^= rj_sbox(k[14]);
k[19] ^= rj_sbox(k[15]);
for(i = 12; i > 0; i -= 4) k[i+0] ^= k[i-4], k[i+1] ^= k[i-3],
k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
*rc = FD(*rc);
k[0] ^= rj_sbox(k[29]) ^ (*rc);
k[1] ^= rj_sbox(k[30]);
k[2] ^= rj_sbox(k[31]);
k[3] ^= rj_sbox(k[28]);
} /* aes_expandDecKey */
/* -------------------------------------------------------------------------- */
void aes256_init(aes256_context *ctx, uint8_t *k)
{
uint8_t rcon = 1;
register uint8_t i;
for (i = 0; i < sizeof(ctx->key); i++) ctx->enckey[i] = ctx->deckey[i] = k[i];
for (i = 8;--i;) aes_expandEncKey(ctx->deckey, &rcon);
} /* aes256_init */
/* -------------------------------------------------------------------------- */
void aes256_done(aes256_context *ctx)
{
register uint8_t i;
for (i = 0; i < sizeof(ctx->key); i++)
ctx->key[i] = ctx->enckey[i] = ctx->deckey[i] = 0;
} /* aes256_done */
/* -------------------------------------------------------------------------- */
void aes256_encrypt_ecb(aes256_context *ctx, uint8_t *buf)
{
uint8_t i, rcon;
aes_addRoundKey_cpy(buf, ctx->enckey, ctx->key);
for(i = 1, rcon = 1; i < 14; ++i)
{
aes_subBytes(buf);
aes_shiftRows(buf);
aes_mixColumns(buf);
if( i & 1 ) aes_addRoundKey( buf, &ctx->key[16]);
else aes_expandEncKey(ctx->key, &rcon), aes_addRoundKey(buf, ctx->key);
}
aes_subBytes(buf);
aes_shiftRows(buf);
aes_expandEncKey(ctx->key, &rcon);
aes_addRoundKey(buf, ctx->key);
} /* aes256_encrypt */
/* -------------------------------------------------------------------------- */
void aes256_decrypt_ecb(aes256_context *ctx, uint8_t *buf)
{
uint8_t i, rcon;
aes_addRoundKey_cpy(buf, ctx->deckey, ctx->key);
aes_shiftRows_inv(buf);
aes_subBytes_inv(buf);
for (i = 14, rcon = 0x80; --i;)
{
if( ( i & 1 ) )
{
aes_expandDecKey(ctx->key, &rcon);
aes_addRoundKey(buf, &ctx->key[16]);
}
else aes_addRoundKey(buf, ctx->key);
aes_mixColumns_inv(buf);
aes_shiftRows_inv(buf);
aes_subBytes_inv(buf);
}
aes_addRoundKey( buf, ctx->key);
} /* aes256_decrypt */
/*
* Byte-oriented AES-256 implementation.
* All lookup tables replaced with 'on the fly' calculations.
*
* Copyright (c) 2007-2009 Ilya O. Levin, http://www.literatecode.com
* Other contributors: Hal Finney
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
#include "aes256.h"
#define F(x) (((x)<<1) ^ ((((x)>>7) & 1) * 0x1b))
#define FD(x) (((x) >> 1) ^ (((x) & 1) ? 0x8d : 0))
// #define BACK_TO_TABLES
#ifdef BACK_TO_TABLES
const uint8_t sbox[256] = {
0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5,
0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76,
0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0,
0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0,
0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc,
0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15,
0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a,
0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75,
0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0,
0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84,
0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b,
0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf,
0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85,
0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8,
0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5,
0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2,
0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17,
0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73,
0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88,
0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb,
0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c,
0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79,
0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9,
0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08,
0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6,
0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a,
0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e,
0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e,
0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94,
0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf,
0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68,
0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16
};
const uint8_t sboxinv[256] = {
0x52, 0x09, 0x6a, 0xd5, 0x30, 0x36, 0xa5, 0x38,
0xbf, 0x40, 0xa3, 0x9e, 0x81, 0xf3, 0xd7, 0xfb,
0x7c, 0xe3, 0x39, 0x82, 0x9b, 0x2f, 0xff, 0x87,
0x34, 0x8e, 0x43, 0x44, 0xc4, 0xde, 0xe9, 0xcb,
0x54, 0x7b, 0x94, 0x32, 0xa6, 0xc2, 0x23, 0x3d,
0xee, 0x4c, 0x95, 0x0b, 0x42, 0xfa, 0xc3, 0x4e,
0x08, 0x2e, 0xa1, 0x66, 0x28, 0xd9, 0x24, 0xb2,
0x76, 0x5b, 0xa2, 0x49, 0x6d, 0x8b, 0xd1, 0x25,
0x72, 0xf8, 0xf6, 0x64, 0x86, 0x68, 0x98, 0x16,
0xd4, 0xa4, 0x5c, 0xcc, 0x5d, 0x65, 0xb6, 0x92,
0x6c, 0x70, 0x48, 0x50, 0xfd, 0xed, 0xb9, 0xda,
0x5e, 0x15, 0x46, 0x57, 0xa7, 0x8d, 0x9d, 0x84,
0x90, 0xd8, 0xab, 0x00, 0x8c, 0xbc, 0xd3, 0x0a,
0xf7, 0xe4, 0x58, 0x05, 0xb8, 0xb3, 0x45, 0x06,
0xd0, 0x2c, 0x1e, 0x8f, 0xca, 0x3f, 0x0f, 0x02,
0xc1, 0xaf, 0xbd, 0x03, 0x01, 0x13, 0x8a, 0x6b,
0x3a, 0x91, 0x11, 0x41, 0x4f, 0x67, 0xdc, 0xea,
0x97, 0xf2, 0xcf, 0xce, 0xf0, 0xb4, 0xe6, 0x73,
0x96, 0xac, 0x74, 0x22, 0xe7, 0xad, 0x35, 0x85,
0xe2, 0xf9, 0x37, 0xe8, 0x1c, 0x75, 0xdf, 0x6e,
0x47, 0xf1, 0x1a, 0x71, 0x1d, 0x29, 0xc5, 0x89,
0x6f, 0xb7, 0x62, 0x0e, 0xaa, 0x18, 0xbe, 0x1b,
0xfc, 0x56, 0x3e, 0x4b, 0xc6, 0xd2, 0x79, 0x20,
0x9a, 0xdb, 0xc0, 0xfe, 0x78, 0xcd, 0x5a, 0xf4,
0x1f, 0xdd, 0xa8, 0x33, 0x88, 0x07, 0xc7, 0x31,
0xb1, 0x12, 0x10, 0x59, 0x27, 0x80, 0xec, 0x5f,
0x60, 0x51, 0x7f, 0xa9, 0x19, 0xb5, 0x4a, 0x0d,
0x2d, 0xe5, 0x7a, 0x9f, 0x93, 0xc9, 0x9c, 0xef,
0xa0, 0xe0, 0x3b, 0x4d, 0xae, 0x2a, 0xf5, 0xb0,
0xc8, 0xeb, 0xbb, 0x3c, 0x83, 0x53, 0x99, 0x61,
0x17, 0x2b, 0x04, 0x7e, 0xba, 0x77, 0xd6, 0x26,
0xe1, 0x69, 0x14, 0x63, 0x55, 0x21, 0x0c, 0x7d
};
#define rj_sbox(x) sbox[(x)]
#define rj_sbox_inv(x) sboxinv[(x)]
#else /* tableless subroutines */
/* -------------------------------------------------------------------------- */
uint8_t gf_alog(uint8_t x) // calculate anti-logarithm gen 3
{
uint8_t atb = 1, z;
while (x--) {z = atb; atb <<= 1; if (z & 0x80) atb^= 0x1b; atb ^= z;}
return atb;
} /* gf_alog */
/* -------------------------------------------------------------------------- */
uint8_t gf_log(uint8_t x) // calculate logarithm gen 3
{
uint8_t atb = 1, i = 0, z;
do {
if (atb == x) break;
z = atb; atb <<= 1; if (z & 0x80) atb^= 0x1b; atb ^= z;
} while (++i > 0);
return i;
} /* gf_log */
/* -------------------------------------------------------------------------- */
uint8_t gf_mulinv(uint8_t x) // calculate multiplicative inverse
{
return (x) ? gf_alog(255 - gf_log(x)) : 0;
} /* gf_mulinv */
/* -------------------------------------------------------------------------- */
uint8_t rj_sbox(uint8_t x)
{
uint8_t y, sb;
sb = y = gf_mulinv(x);
y = (y<<1)|(y>>7); sb ^= y; y = (y<<1)|(y>>7); sb ^= y;
y = (y<<1)|(y>>7); sb ^= y; y = (y<<1)|(y>>7); sb ^= y;
return (sb ^ 0x63);
} /* rj_sbox */
/* -------------------------------------------------------------------------- */
uint8_t rj_sbox_inv(uint8_t x)
{
uint8_t y, sb;
y = x ^ 0x63;
sb = y = (y<<1)|(y>>7);
y = (y<<2)|(y>>6); sb ^= y; y = (y<<3)|(y>>5); sb ^= y;
return gf_mulinv(sb);
} /* rj_sbox_inv */
#endif
/* -------------------------------------------------------------------------- */
uint8_t rj_xtime(uint8_t x)
{
return (x & 0x80) ? ((x << 1) ^ 0x1b) : (x << 1);
} /* rj_xtime */
/* -------------------------------------------------------------------------- */
void aes_subBytes(uint8_t *buf)
{
register uint8_t i = 16;
while (i--) buf[i] = rj_sbox(buf[i]);
} /* aes_subBytes */
/* -------------------------------------------------------------------------- */
void aes_subBytes_inv(uint8_t *buf)
{
register uint8_t i = 16;
while (i--) buf[i] = rj_sbox_inv(buf[i]);
} /* aes_subBytes_inv */
/* -------------------------------------------------------------------------- */
void aes_addRoundKey(uint8_t *buf, uint8_t *key)
{
register uint8_t i = 16;
while (i--) buf[i] ^= key[i];
} /* aes_addRoundKey */
/* -------------------------------------------------------------------------- */
void aes_addRoundKey_cpy(uint8_t *buf, uint8_t *key, uint8_t *cpk)
{
register uint8_t i = 16;
while (i--) buf[i] ^= (cpk[i] = key[i]), cpk[16+i] = key[16 + i];
} /* aes_addRoundKey_cpy */
/* -------------------------------------------------------------------------- */
void aes_shiftRows(uint8_t *buf)
{
register uint8_t i, j; /* to make it potentially parallelable :) */
i = buf[1]; buf[1] = buf[5]; buf[5] = buf[9]; buf[9] = buf[13]; buf[13] = i;
i = buf[10]; buf[10] = buf[2]; buf[2] = i;
j = buf[3]; buf[3] = buf[15]; buf[15] = buf[11]; buf[11] = buf[7]; buf[7] = j;
j = buf[14]; buf[14] = buf[6]; buf[6] = j;
} /* aes_shiftRows */
/* -------------------------------------------------------------------------- */
void aes_shiftRows_inv(uint8_t *buf)
{
register uint8_t i, j; /* same as above :) */
i = buf[1]; buf[1] = buf[13]; buf[13] = buf[9]; buf[9] = buf[5]; buf[5] = i;
i = buf[2]; buf[2] = buf[10]; buf[10] = i;
j = buf[3]; buf[3] = buf[7]; buf[7] = buf[11]; buf[11] = buf[15]; buf[15] = j;
j = buf[6]; buf[6] = buf[14]; buf[14] = j;
} /* aes_shiftRows_inv */
/* -------------------------------------------------------------------------- */
void aes_mixColumns(uint8_t *buf)
{
register uint8_t i, a, b, c, d, e;
for (i = 0; i < 16; i += 4)
{
a = buf[i]; b = buf[i + 1]; c = buf[i + 2]; d = buf[i + 3];
e = a ^ b ^ c ^ d;
buf[i] ^= e ^ rj_xtime(a^b); buf[i+1] ^= e ^ rj_xtime(b^c);
buf[i+2] ^= e ^ rj_xtime(c^d); buf[i+3] ^= e ^ rj_xtime(d^a);
}
} /* aes_mixColumns */
/* -------------------------------------------------------------------------- */
void aes_mixColumns_inv(uint8_t *buf)
{
register uint8_t i, a, b, c, d, e, x, y, z;
for (i = 0; i < 16; i += 4)
{
a = buf[i]; b = buf[i + 1]; c = buf[i + 2]; d = buf[i + 3];
e = a ^ b ^ c ^ d;
z = rj_xtime(e);
x = e ^ rj_xtime(rj_xtime(z^a^c)); y = e ^ rj_xtime(rj_xtime(z^b^d));
buf[i] ^= x ^ rj_xtime(a^b); buf[i+1] ^= y ^ rj_xtime(b^c);
buf[i+2] ^= x ^ rj_xtime(c^d); buf[i+3] ^= y ^ rj_xtime(d^a);
}
} /* aes_mixColumns_inv */
/* -------------------------------------------------------------------------- */
void aes_expandEncKey(uint8_t *k, uint8_t *rc)
{
register uint8_t i;
k[0] ^= rj_sbox(k[29]) ^ (*rc);
k[1] ^= rj_sbox(k[30]);
k[2] ^= rj_sbox(k[31]);
k[3] ^= rj_sbox(k[28]);
*rc = F( *rc);
for(i = 4; i < 16; i += 4) k[i] ^= k[i-4], k[i+1] ^= k[i-3],
k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
k[16] ^= rj_sbox(k[12]);
k[17] ^= rj_sbox(k[13]);
k[18] ^= rj_sbox(k[14]);
k[19] ^= rj_sbox(k[15]);
for(i = 20; i < 32; i += 4) k[i] ^= k[i-4], k[i+1] ^= k[i-3],
k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
} /* aes_expandEncKey */
/* -------------------------------------------------------------------------- */
void aes_expandDecKey(uint8_t *k, uint8_t *rc)
{
uint8_t i;
for(i = 28; i > 16; i -= 4) k[i+0] ^= k[i-4], k[i+1] ^= k[i-3],
k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
k[16] ^= rj_sbox(k[12]);
k[17] ^= rj_sbox(k[13]);
k[18] ^= rj_sbox(k[14]);
k[19] ^= rj_sbox(k[15]);
for(i = 12; i > 0; i -= 4) k[i+0] ^= k[i-4], k[i+1] ^= k[i-3],
k[i+2] ^= k[i-2], k[i+3] ^= k[i-1];
*rc = FD(*rc);
k[0] ^= rj_sbox(k[29]) ^ (*rc);
k[1] ^= rj_sbox(k[30]);
k[2] ^= rj_sbox(k[31]);
k[3] ^= rj_sbox(k[28]);
} /* aes_expandDecKey */
/* -------------------------------------------------------------------------- */
void aes256_init(aes256_context *ctx, uint8_t *k)
{
uint8_t rcon = 1;
register uint8_t i;
for (i = 0; i < sizeof(ctx->key); i++) ctx->enckey[i] = ctx->deckey[i] = k[i];
for (i = 8;--i;) aes_expandEncKey(ctx->deckey, &rcon);
} /* aes256_init */
/* -------------------------------------------------------------------------- */
void aes256_done(aes256_context *ctx)
{
register uint8_t i;
for (i = 0; i < sizeof(ctx->key); i++)
ctx->key[i] = ctx->enckey[i] = ctx->deckey[i] = 0;
} /* aes256_done */
/* -------------------------------------------------------------------------- */
void aes256_encrypt_ecb(aes256_context *ctx, uint8_t *buf)
{
uint8_t i, rcon;
aes_addRoundKey_cpy(buf, ctx->enckey, ctx->key);
for(i = 1, rcon = 1; i < 14; ++i)
{
aes_subBytes(buf);
aes_shiftRows(buf);
aes_mixColumns(buf);
if( i & 1 ) aes_addRoundKey( buf, &ctx->key[16]);
else aes_expandEncKey(ctx->key, &rcon), aes_addRoundKey(buf, ctx->key);
}
aes_subBytes(buf);
aes_shiftRows(buf);
aes_expandEncKey(ctx->key, &rcon);
aes_addRoundKey(buf, ctx->key);
} /* aes256_encrypt */
/* -------------------------------------------------------------------------- */
void aes256_decrypt_ecb(aes256_context *ctx, uint8_t *buf)
{
uint8_t i, rcon;
aes_addRoundKey_cpy(buf, ctx->deckey, ctx->key);
aes_shiftRows_inv(buf);
aes_subBytes_inv(buf);
for (i = 14, rcon = 0x80; --i;)
{
if( ( i & 1 ) )
{
aes_expandDecKey(ctx->key, &rcon);
aes_addRoundKey(buf, &ctx->key[16]);
}
else aes_addRoundKey(buf, ctx->key);
aes_mixColumns_inv(buf);
aes_shiftRows_inv(buf);
aes_subBytes_inv(buf);
}
aes_addRoundKey( buf, ctx->key);
} /* aes256_decrypt */

92
core/io/aes256.h
View file

@ -1,46 +1,46 @@
/*
* Byte-oriented AES-256 implementation.
* All lookup tables replaced with 'on the fly' calculations.
*
* Copyright (c) 2007-2009 Ilya O. Levin, http://www.literatecode.com
* Other contributors: Hal Finney
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
#ifndef AES_256_H
#define AES_256_H
#include "typedefs.h"
#ifdef __cplusplus
extern "C" {
#endif
typedef struct {
uint8_t key[32];
uint8_t enckey[32];
uint8_t deckey[32];
} aes256_context;
void aes256_init(aes256_context *, uint8_t * /* key */);
void aes256_done(aes256_context *);
void aes256_encrypt_ecb(aes256_context *, uint8_t * /* plaintext */);
void aes256_decrypt_ecb(aes256_context *, uint8_t * /* cipertext */);
#ifdef __cplusplus
}
#endif
#endif
/*
* Byte-oriented AES-256 implementation.
* All lookup tables replaced with 'on the fly' calculations.
*
* Copyright (c) 2007-2009 Ilya O. Levin, http://www.literatecode.com
* Other contributors: Hal Finney
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
#ifndef AES_256_H
#define AES_256_H
#include "typedefs.h"
#ifdef __cplusplus
extern "C" {
#endif
typedef struct {
uint8_t key[32];
uint8_t enckey[32];
uint8_t deckey[32];
} aes256_context;
void aes256_init(aes256_context *, uint8_t * /* key */);
void aes256_done(aes256_context *);
void aes256_encrypt_ecb(aes256_context *, uint8_t * /* plaintext */);
void aes256_decrypt_ecb(aes256_context *, uint8_t * /* cipertext */);
#ifdef __cplusplus
}
#endif
#endif

4
core/math/math_funcs.h
View file

@ -79,9 +79,9 @@ public:
return Math::log( p_linear ) * 8.6858896380650365530225783783321;
}
static inline double db2linear(double p_linear) {
static inline double db2linear(double p_db) {
return Math::exp( p_linear * 0.11512925464970228420089957273422 );
return Math::exp( p_db * 0.11512925464970228420089957273422 );
}
static bool is_nan(double p_val);

307
core/os/input.cpp
View file

@ -64,6 +64,7 @@ void Input::_bind_methods() {
ObjectTypeDB::bind_method(_MD("warp_mouse_pos","to"),&Input::warp_mouse_pos);
ObjectTypeDB::bind_method(_MD("action_press"),&Input::action_press);
ObjectTypeDB::bind_method(_MD("action_release"),&Input::action_release);
ObjectTypeDB::bind_method(_MD("set_custom_mouse_cursor","image:Texture","hotspot"),&Input::set_custom_mouse_cursor,DEFVAL(Vector2()));
BIND_CONSTANT( MOUSE_MODE_VISIBLE );
BIND_CONSTANT( MOUSE_MODE_HIDDEN );
@ -104,309 +105,3 @@ Input::Input() {
//////////////////////////////////////////////////////////
void InputDefault::SpeedTrack::update(const Vector2& p_delta_p) {
uint64_t tick = OS::get_singleton()->get_ticks_usec();
uint32_t tdiff = tick-last_tick;
float delta_t = tdiff / 1000000.0;
last_tick=tick;
accum+=p_delta_p;
accum_t+=delta_t;
if (accum_t>max_ref_frame*10)
accum_t=max_ref_frame*10;
while( accum_t>=min_ref_frame ) {
float slice_t = min_ref_frame / accum_t;
Vector2 slice = accum*slice_t;
accum=accum-slice;
accum_t-=min_ref_frame;
speed=(slice/min_ref_frame).linear_interpolate(speed,min_ref_frame/max_ref_frame);
}
}
void InputDefault::SpeedTrack::reset() {
last_tick = OS::get_singleton()->get_ticks_usec();
speed=Vector2();
accum_t=0;
}
InputDefault::SpeedTrack::SpeedTrack() {
min_ref_frame=0.1;
max_ref_frame=0.3;
reset();
}
bool InputDefault::is_key_pressed(int p_scancode) {
_THREAD_SAFE_METHOD_
return keys_pressed.has(p_scancode);
}
bool InputDefault::is_mouse_button_pressed(int p_button) {
_THREAD_SAFE_METHOD_
return (mouse_button_mask&(1<<p_button))!=0;
}
static int _combine_device(int p_value,int p_device) {
return p_value|(p_device<<20);
}
bool InputDefault::is_joy_button_pressed(int p_device, int p_button) {
_THREAD_SAFE_METHOD_
return joy_buttons_pressed.has(_combine_device(p_button,p_device));
}
bool InputDefault::is_action_pressed(const StringName& p_action) {
if (custom_action_press.has(p_action))
return true; //simpler
const List<InputEvent> *alist = InputMap::get_singleton()->get_action_list(p_action);
if (!alist)
return NULL;
for (const List<InputEvent>::Element *E=alist->front();E;E=E->next()) {
int device=E->get().device;
switch(E->get().type) {
case InputEvent::KEY: {
const InputEventKey &iek=E->get().key;
if ((keys_pressed.has(iek.scancode)))
return true;
} break;
case InputEvent::MOUSE_BUTTON: {
const InputEventMouseButton &iemb=E->get().mouse_button;
if(mouse_button_mask&(1<<iemb.button_index))
return true;
} break;
case InputEvent::JOYSTICK_BUTTON: {
const InputEventJoystickButton &iejb=E->get().joy_button;
int c = _combine_device(iejb.button_index,device);
if (joy_buttons_pressed.has(c))
return true;
} break;
}
}
return false;
}
float InputDefault::get_joy_axis(int p_device,int p_axis) {
_THREAD_SAFE_METHOD_
int c = _combine_device(p_axis,p_device);
if (joy_axis.has(c)) {
return joy_axis[c];
} else {
return 0;
}
}
String InputDefault::get_joy_name(int p_idx) {
_THREAD_SAFE_METHOD_
return joy_names[p_idx];
};
void InputDefault::joy_connection_changed(int p_idx, bool p_connected, String p_name) {
_THREAD_SAFE_METHOD_
joy_names[p_idx] = p_connected ? p_name : "";
emit_signal("joy_connection_changed", p_idx, p_connected);
};
Vector3 InputDefault::get_accelerometer() {
_THREAD_SAFE_METHOD_
return accelerometer;
}
void InputDefault::parse_input_event(const InputEvent& p_event) {
_THREAD_SAFE_METHOD_
switch(p_event.type) {
case InputEvent::KEY: {
if (p_event.key.echo)
break;
if (p_event.key.scancode==0)
break;
// print_line(p_event);
if (p_event.key.pressed)
keys_pressed.insert(p_event.key.scancode);
else
keys_pressed.erase(p_event.key.scancode);
} break;
case InputEvent::MOUSE_BUTTON: {
if (p_event.mouse_button.doubleclick)
break;
if (p_event.mouse_button.pressed)
mouse_button_mask|=(1<<p_event.mouse_button.button_index);
else
mouse_button_mask&=~(1<<p_event.mouse_button.button_index);
if (main_loop && emulate_touch && p_event.mouse_button.button_index==1) {
InputEventScreenTouch touch_event;
touch_event.index=0;
touch_event.pressed=p_event.mouse_button.pressed;
touch_event.x=p_event.mouse_button.x;
touch_event.y=p_event.mouse_button.y;
InputEvent ev;
ev.type=InputEvent::SCREEN_TOUCH;
ev.screen_touch=touch_event;
main_loop->input_event(ev);
}
} break;
case InputEvent::MOUSE_MOTION: {
if (main_loop && emulate_touch && p_event.mouse_motion.button_mask&1) {
InputEventScreenDrag drag_event;
drag_event.index=0;
drag_event.x=p_event.mouse_motion.x;
drag_event.y=p_event.mouse_motion.y;
drag_event.relative_x=p_event.mouse_motion.relative_x;
drag_event.relative_y=p_event.mouse_motion.relative_y;
drag_event.speed_x=p_event.mouse_motion.speed_x;
drag_event.speed_y=p_event.mouse_motion.speed_y;
InputEvent ev;
ev.type=InputEvent::SCREEN_DRAG;
ev.screen_drag=drag_event;
main_loop->input_event(ev);
}
} break;
case InputEvent::JOYSTICK_BUTTON: {
int c = _combine_device(p_event.joy_button.button_index,p_event.device);
if (p_event.joy_button.pressed)
joy_buttons_pressed.insert(c);
else
joy_buttons_pressed.erase(c);
} break;
case InputEvent::JOYSTICK_MOTION: {
set_joy_axis(p_event.device, p_event.joy_motion.axis, p_event.joy_motion.axis_value);
} break;
}
if (main_loop)
main_loop->input_event(p_event);
}
void InputDefault::set_joy_axis(int p_device,int p_axis,float p_value) {
_THREAD_SAFE_METHOD_
int c = _combine_device(p_axis,p_device);
joy_axis[c]=p_value;
}
void InputDefault::set_accelerometer(const Vector3& p_accel) {
_THREAD_SAFE_METHOD_
accelerometer=p_accel;
}
void InputDefault::set_main_loop(MainLoop *p_main_loop) {
main_loop=p_main_loop;
}
void InputDefault::set_mouse_pos(const Point2& p_posf) {
mouse_speed_track.update(p_posf-mouse_pos);
mouse_pos=p_posf;
}
Point2 InputDefault::get_mouse_pos() const {
return mouse_pos;
}
Point2 InputDefault::get_mouse_speed() const {
return mouse_speed_track.speed;
}
int InputDefault::get_mouse_button_mask() const {
return OS::get_singleton()->get_mouse_button_state();
}
void InputDefault::warp_mouse_pos(const Vector2& p_to) {
OS::get_singleton()->warp_mouse_pos(p_to);
}
void InputDefault::iteration(float p_step) {
}
void InputDefault::action_press(const StringName& p_action) {
if (custom_action_press.has(p_action)) {
custom_action_press[p_action]++;
} else {
custom_action_press[p_action]=1;
}
}
void InputDefault::action_release(const StringName& p_action){
ERR_FAIL_COND(!custom_action_press.has(p_action));
custom_action_press[p_action]--;
if (custom_action_press[p_action]==0) {
custom_action_press.erase(p_action);
}
}
void InputDefault::set_emulate_touch(bool p_emulate) {
emulate_touch=p_emulate;
}
bool InputDefault::is_emulating_touchscreen() const {
return emulate_touch;
}
InputDefault::InputDefault() {
mouse_button_mask=0;
emulate_touch=false;
main_loop=NULL;
}

73
core/os/input.h
View file

@ -80,82 +80,13 @@ public:
virtual bool is_emulating_touchscreen() const=0;
virtual void set_custom_mouse_cursor(const RES& p_cursor,const Vector2& p_hotspot=Vector2())=0;
virtual void set_mouse_in_window(bool p_in_window)=0;
Input();
};
VARIANT_ENUM_CAST(Input::MouseMode);
class InputDefault : public Input {
OBJ_TYPE( InputDefault, Input );
_THREAD_SAFE_CLASS_
int mouse_button_mask;
Set<int> keys_pressed;
Set<int> joy_buttons_pressed;
Map<int,float> joy_axis;
Map<StringName,int> custom_action_press;
Map<int, String> joy_names;
Vector3 accelerometer;
Vector2 mouse_pos;
MainLoop *main_loop;
bool emulate_touch;
struct SpeedTrack {
uint64_t last_tick;
Vector2 speed;
Vector2 accum;
float accum_t;
float min_ref_frame;
float max_ref_frame;
void update(const Vector2& p_delta_p);
void reset();
SpeedTrack();
};
SpeedTrack mouse_speed_track;
public:
virtual bool is_key_pressed(int p_scancode);
virtual bool is_mouse_button_pressed(int p_button);
virtual bool is_joy_button_pressed(int p_device, int p_button);
virtual bool is_action_pressed(const StringName& p_action);
virtual float get_joy_axis(int p_device,int p_axis);
String get_joy_name(int p_idx);
void joy_connection_changed(int p_idx, bool p_connected, String p_name);
virtual Vector3 get_accelerometer();
virtual Point2 get_mouse_pos() const;
virtual Point2 get_mouse_speed() const;
virtual int get_mouse_button_mask() const;
virtual void warp_mouse_pos(const Vector2& p_to);
void parse_input_event(const InputEvent& p_event);
void set_accelerometer(const Vector3& p_accel);
void set_joy_axis(int p_device,int p_axis,float p_value);
void set_main_loop(MainLoop *main_loop);
void set_mouse_pos(const Point2& p_posf);
void action_press(const StringName& p_action);
void action_release(const StringName& p_action);
void iteration(float p_step);
void set_emulate_touch(bool p_emulate);
virtual bool is_emulating_touchscreen() const;
InputDefault();
};
#endif // INPUT_H

3
core/os/main_loop.cpp
View file

@ -45,7 +45,8 @@ void MainLoop::_bind_methods() {
BIND_VMETHOD( MethodInfo("_idle",PropertyInfo(Variant::REAL,"delta")) );
BIND_VMETHOD( MethodInfo("_finalize") );
BIND_CONSTANT(NOTIFICATION_WM_MOUSE_ENTER);
BIND_CONSTANT(NOTIFICATION_WM_MOUSE_EXIT);
BIND_CONSTANT(NOTIFICATION_WM_FOCUS_IN);
BIND_CONSTANT(NOTIFICATION_WM_FOCUS_OUT);
BIND_CONSTANT(NOTIFICATION_WM_QUIT_REQUEST);

2
core/os/main_loop.h
View file

@ -47,6 +47,8 @@ protected:
public:
enum {
NOTIFICATION_WM_MOUSE_ENTER = 3,
NOTIFICATION_WM_MOUSE_EXIT = 4,
NOTIFICATION_WM_FOCUS_IN = 5,
NOTIFICATION_WM_FOCUS_OUT = 6,
NOTIFICATION_WM_QUIT_REQUEST = 7,

1
core/os/os.cpp
View file

@ -516,6 +516,7 @@ OS::OS() {
_target_fps=0;
_render_thread_mode=RENDER_THREAD_SAFE;
_time_scale=1.0;
_pixel_snap=false;
Math::seed(1234567);
}

3
core/os/os.h
View file

@ -58,6 +58,7 @@ class OS {
float _fps;
int _target_fps;
float _time_scale;
bool _pixel_snap;
char *last_error;
@ -393,7 +394,7 @@ public:
void set_time_scale(float p_scale);
float get_time_scale() const;
_FORCE_INLINE_ bool get_use_pixel_snap() const { return _pixel_snap; }
OS();
virtual ~OS();

8
core/ustring.cpp
View file

@ -3119,8 +3119,8 @@ String String::xml_escape(bool p_escape_quotes) const {
String str=*this;
str=str.replace("&","&amp;");
str=str.replace("<","&gt;");
str=str.replace(">","&lt;");
str=str.replace("<","&lt;");
str=str.replace(">","&gt;");
if (p_escape_quotes) {
str=str.replace("'","&apos;");
str=str.replace("\"","&quot;");
@ -3172,12 +3172,12 @@ static _FORCE_INLINE_ int _xml_unescape(const CharType *p_src,int p_src_len,Char
} else if (p_src_len>=4 && p_src[1]=='g' && p_src[2]=='t' && p_src[3]==';') {
if (p_dst)
*p_dst='<';
*p_dst='>';
eat=4;
} else if (p_src_len>=4 && p_src[1]=='l' && p_src[2]=='t' && p_src[3]==';') {
if (p_dst)
*p_dst='>';
*p_dst='<';
eat=4;
} else if (p_src_len>=5 && p_src[1]=='a' && p_src[2]=='m' && p_src[3]=='p' && p_src[4]==';') {

21
demos/2d/dynamic_collision_shapes/ball.gd Normal file
View file

@ -0,0 +1,21 @@
extends RigidBody2D
# member variables here, example:
# var a=2
# var b="textvar"
var timeout=5
func _process(delta):
timeout-=delta
if (timeout<1):
set_opacity(timeout)
if (timeout<0):
queue_free()
func _ready():
set_process(true)
# Initialization here
pass

BIN
demos/2d/dynamic_collision_shapes/ball.png Normal file
View file

Binary file not shown.
Side by side

After

Width:  |  Height:  |  Size: 321 B

BIN
demos/2d/dynamic_collision_shapes/ball.scn Normal file
View file

Binary file not shown.

BIN
demos/2d/dynamic_collision_shapes/box.png Normal file
View file

Binary file not shown.
Side by side

After

Width:  |  Height:  |  Size: 253 B

BIN
demos/2d/dynamic_collision_shapes/circle.png Normal file
View file

Binary file not shown.
Side by side

After

Width:  |  Height:  |  Size: 889 B

23
demos/2d/dynamic_collision_shapes/dynamic_colobjs.gd Normal file
View file

@ -0,0 +1,23 @@
extends Node2D
# member variables here, example:
# var a=2
# var b="textvar"
const EMIT_INTERVAL=0.1
var timeout=EMIT_INTERVAL
func _process(delta):
timeout-=delta
if (timeout<0):
timeout=EMIT_INTERVAL
var ball = preload("res://ball.scn").instance()
ball.set_pos( Vector2(randf() * get_viewport_rect().size.x, 0) )
add_child(ball)
func _ready():
set_process(true)
# Initialization here
pass

BIN
demos/2d/dynamic_collision_shapes/dynamic_colobjs.scn Normal file
View file

Binary file not shown.

4
demos/2d/dynamic_collision_shapes/engine.cfg Normal file
View file

@ -0,0 +1,4 @@
[application]
name="Run-Time CollisionShape"
main_scene="res://dynamic_colobjs.scn"

BIN
demos/2d/dynamic_collision_shapes/poly.png Normal file
View file

Binary file not shown.
Side by side

After

Width:  |  Height:  |  Size: 1.6 KiB

BIN
demos/2d/isometric/dungeon.scn
View file

Binary file not shown.

BIN
demos/2d/lights_shadows/light_shadows.scn
View file

Binary file not shown.

345
demos/2d/platformer/enemy.xml
View file

@ -1,10 +1,10 @@
<?xml version="1.0" encoding="UTF-8" ?>
<resource_file type="PackedScene" subresource_count="11" version="1.0" version_name="Godot Engine v1.0.3917-beta1">
<ext_resource path="res://sound_explode.*" type="Sample"></ext_resource>
<ext_resource path="res://enemy.*" type="Texture"></ext_resource>
<ext_resource path="res://enemy.*" type="Script"></ext_resource>
<ext_resource path="res://sound_hit.*" type="Sample"></ext_resource>
<ext_resource path="res://bullet.*" type="Texture"></ext_resource>
<resource_file type="PackedScene" subresource_count="12" version="2.0" version_name="Godot Engine v2.0.alpha.custom_build">
<ext_resource path="res://bullet.png" type="Texture" index="2"></ext_resource>
<ext_resource path="res://enemy.gd" type="Script" index="0"></ext_resource>
<ext_resource path="res://enemy.png" type="Texture" index="1"></ext_resource>
<ext_resource path="res://sound_hit.wav" type="Sample" index="4"></ext_resource>
<ext_resource path="res://sound_explode.wav" type="Sample" index="3"></ext_resource>
<resource type="CircleShape2D" path="local://1">
<real name="custom_solver_bias"> 0 </real>
<real name="radius"> 14 </real>
@ -21,6 +21,8 @@
<dictionary name="tracks/0/keys" shared="false">
<string> "cont" </string>
<bool> False </bool>
<string> "times" </string>
<real_array len="10"> 0, 0.75, 1.5, 2.25, 3, 3.75, 4.5, 5.25, 6, 6.75 </real_array>
<string> "transitions" </string>
<real_array len="10"> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 </real_array>
<string> "values" </string>
@ -36,8 +38,33 @@
<int> 7 </int>
<int> 5 </int>
</array>
</dictionary>
</resource>
<resource type="Animation" path="local://4">
<string name="resource/name"> "walk" </string>
<real name="length"> 1.25 </real>
<bool name="loop"> True </bool>
<real name="step"> 0.25 </real>
<string name="tracks/0/type"> "value" </string>
<node_path name="tracks/0/path"> "sprite:frame" </node_path>
<int name="tracks/0/interp"> 1 </int>
<dictionary name="tracks/0/keys" shared="false">
<string> "cont" </string>
<bool> False </bool>
<string> "times" </string>
<real_array len="10"> 0, 0.75, 1.5, 2.25, 3, 3.75, 4.5, 5.25, 6, 6.75 </real_array>
<real_array len="6"> 0, 0.25, 0.5, 0.75, 1, 1.25 </real_array>
<string> "transitions" </string>
<real_array len="6"> 1, 1, 1, 1, 1, 1 </real_array>
<string> "values" </string>
<array len="6" shared="false">
<int> 0 </int>
<int> 1 </int>
<int> 2 </int>
<int> 3 </int>
<int> 4 </int>
<int> 0 </int>
</array>
</dictionary>
</resource>
@ -52,6 +79,8 @@
<dictionary name="tracks/0/keys" shared="false">
<string> "cont" </string>
<bool> True </bool>
<string> "times" </string>
<real_array len="2"> 3.58422, 4.33851 </real_array>
<string> "transitions" </string>
<real_array len="2"> 1, 1 </real_array>
<string> "values" </string>
@ -59,8 +88,6 @@
<real> 1 </real>
<real> 0 </real>
</array>
<string> "times" </string>
<real_array len="2"> 3.58422, 4.33851 </real_array>
</dictionary>
<string name="tracks/1/type"> "value" </string>
<node_path name="tracks/1/path"> "sprite:frame" </node_path>
@ -68,14 +95,14 @@
<dictionary name="tracks/1/keys" shared="false">
<string> "cont" </string>
<bool> True </bool>
<string> "times" </string>
<real_array len="1"> 0 </real_array>
<string> "transitions" </string>
<real_array len="1"> 1 </real_array>
<string> "values" </string>
<array len="1" shared="false">
<int> 4 </int>
</array>
<string> "times" </string>
<real_array len="1"> 0 </real_array>
</dictionary>
<string name="tracks/2/type"> "value" </string>
<node_path name="tracks/2/path"> "Particles2D:config/emitting" </node_path>
@ -83,19 +110,21 @@
<dictionary name="tracks/2/keys" shared="false">
<string> "cont" </string>
<bool> False </bool>
<string> "times" </string>
<real_array len="1"> 3.47394 </real_array>
<string> "transitions" </string>
<real_array len="1"> 1 </real_array>
<string> "values" </string>
<array len="1" shared="false">
<bool> True </bool>
</array>
<string> "times" </string>
<real_array len="1"> 3.47394 </real_array>
</dictionary>
<string name="tracks/3/type"> "method" </string>
<node_path name="tracks/3/path"> "." </node_path>
<int name="tracks/3/interp"> 1 </int>
<dictionary name="tracks/3/keys" shared="false">
<string> "times" </string>
<real_array len="2"> 3.20357, 5.07305 </real_array>
<string> "transitions" </string>
<real_array len="2"> 1, 1 </real_array>
<string> "values" </string>
@ -115,36 +144,12 @@
<string> "_die" </string>
</dictionary>
</array>
<string> "times" </string>
<real_array len="2"> 3.20357, 5.07305 </real_array>
</dictionary>
</resource>
<resource type="Animation" path="local://4">
<string name="resource/name"> "walk" </string>
<real name="length"> 1.25 </real>
<bool name="loop"> True </bool>
<real name="step"> 0.25 </real>
<string name="tracks/0/type"> "value" </string>
<node_path name="tracks/0/path"> "sprite:frame" </node_path>
<int name="tracks/0/interp"> 1 </int>
<dictionary name="tracks/0/keys" shared="false">
<string> "cont" </string>
<bool> False </bool>
<string> "transitions" </string>
<real_array len="6"> 1, 1, 1, 1, 1, 1 </real_array>
<string> "values" </string>
<array len="6" shared="false">
<int> 0 </int>
<int> 1 </int>
<int> 2 </int>
<int> 3 </int>
<int> 4 </int>
<int> 0 </int>
</array>
<string> "times" </string>
<real_array len="6"> 0, 0.25, 0.5, 0.75, 1, 1.25 </real_array>
</dictionary>
<resource type="ColorRamp" path="local://6">
<real_array name="offsets" len="2"> 0, 1 </real_array>
<color_array name="colors" len="2"> 1, 0.884956, 0.823009, 1, 0.768627, 0.389381, 0, 0 </color_array>
</resource>
<resource type="SampleLibrary" path="local://5">
@ -154,7 +159,7 @@
<string> "pitch" </string>
<real> 1 </real>
<string> "sample" </string>
<resource resource_type="Sample" path="res://sound_explode.*"> </resource>
<resource external="3"> </resource>
</dictionary>
<dictionary name="samples/hit" shared="false">
<string> "db" </string>
@ -162,24 +167,21 @@
<string> "pitch" </string>
<real> 1 </real>
<string> "sample" </string>
<resource resource_type="Sample" path="res://sound_hit.*"> </resource>
<resource external="4"> </resource>
</dictionary>
</resource>
<main_resource>
<dictionary name="_bundled" shared="false">
<string> "conn_count" </string>
<int> 0 </int>
<string> "conns" </string>
<int_array len="0"> </int_array>
<string> "names" </string>
<string_array len="132">
<string_array len="107">
<string> "enemy" </string>
<string> "RigidBody2D" </string>
<string> "visibility/visible" </string>
<string> "visibility/opacity" </string>
<string> "visibility/self_opacity" </string>
<string> "visibility/on_top" </string>
<string> "transform/pos" </string>
<string> "transform/rot" </string>
<string> "transform/scale" </string>
<string> "shape_count" </string>
<string> "input/pickable" </string>
<string> "shapes/0/shape" </string>
<string> "shapes/0/transform" </string>
<string> "shapes/0/trigger" </string>
@ -189,71 +191,71 @@
<string> "shapes/2/shape" </string>
<string> "shapes/2/transform" </string>
<string> "shapes/2/trigger" </string>
<string> "collision/layers" </string>
<string> "collision/mask" </string>
<string> "mode" </string>
<string> "mass" </string>
<string> "friction" </string>
<string> "bounce" </string>
<string> "gravity_scale" </string>
<string> "custom_integrator" </string>
<string> "continuous_cd" </string>
<string> "contacts_reported" </string>
<string> "contact_monitor" </string>
<string> "active" </string>
<string> "sleeping" </string>
<string> "can_sleep" </string>
<string> "velocity/linear" </string>
<string> "velocity/angular" </string>
<string> "damp_override/linear" </string>
<string> "damp_override/angular" </string>
<string> "script/script" </string>
<string> "__meta__" </string>
<string> "enabler" </string>
<string> "VisibilityEnabler2D" </string>
<string> "transform/pos" </string>
<string> "transform/scale" </string>
<string> "rect" </string>
<string> "enabler/pause_animations" </string>
<string> "enabler/freeze_bodies" </string>
<string> "enabler/pause_particles" </string>
<string> "enabler/process_parent" </string>
<string> "enabler/fixed_process_parent" </string>
<string> "anim" </string>
<string> "AnimationPlayer" </string>
<string> "playback/process_mode" </string>
<string> "playback/default_blend_time" </string>
<string> "root/root" </string>
<string> "anims/idle" </string>
<string> "anims/explode" </string>
<string> "anims/walk" </string>
<string> "anims/explode" </string>
<string> "playback/active" </string>
<string> "playback/speed" </string>
<string> "blend_times" </string>
<string> "autoplay" </string>
<string> "CollisionShape2D" </string>
<string> "shape" </string>
<string> "trigger" </string>
<string> "CollisionShape2D 2" </string>
<string> "CollisionShape2D 3" </string>
<string> "sprite" </string>
<string> "Sprite" </string>
<string> "texture" </string>
<string> "centered" </string>
<string> "offset" </string>
<string> "flip_h" </string>
<string> "flip_v" </string>
<string> "vframes" </string>
<string> "hframes" </string>
<string> "frame" </string>
<string> "modulate" </string>
<string> "region" </string>
<string> "region_rect" </string>
<string> "CollisionShape2D" </string>
<string> "shape" </string>
<string> "trigger" </string>
<string> "_update_shape_index" </string>
<string> "CollisionShape2D 2" </string>
<string> "CollisionShape2D 3" </string>
<string> "raycast_left" </string>
<string> "RayCast2D" </string>
<string> "enabled" </string>
<string> "cast_to" </string>
<string> "layer_mask" </string>
<string> "raycast_right" </string>
<string> "Particles2D" </string>
<string> "visibility/self_opacity" </string>
<string> "visibility/blend_mode" </string>
<string> "config/amount" </string>
<string> "config/lifetime" </string>
<string> "config/time_scale" </string>
<string> "config/preprocess" </string>
<string> "config/emit_timeout" </string>
<string> "config/emitting" </string>
<string> "config/offset" </string>
<string> "config/half_extents" </string>
<string> "config/local_space" </string>
<string> "config/explosiveness" </string>
<string> "config/texture" </string>
<string> "params/direction" </string>
@ -266,32 +268,14 @@
<string> "params/radial_accel" </string>
<string> "params/tangential_accel" </string>
<string> "params/damping" </string>
<string> "params/initial_angle" </string>
<string> "params/initial_size" </string>
<string> "params/final_size" </string>
<string> "params/hue_variation" </string>
<string> "randomness/direction" </string>
<string> "randomness/spread" </string>
<string> "randomness/linear_velocity" </string>
<string> "params/anim_speed_scale" </string>
<string> "params/anim_initial_pos" </string>
<string> "randomness/spin_velocity" </string>
<string> "randomness/orbit_velocity" </string>
<string> "randomness/gravity_direction" </string>
<string> "randomness/gravity_strength" </string>
<string> "randomness/radial_accel" </string>
<string> "randomness/tangential_accel" </string>
<string> "randomness/damping" </string>
<string> "randomness/initial_size" </string>
<string> "randomness/final_size" </string>
<string> "randomness/hue_variation" </string>
<string> "color_phases/count" </string>
<string> "phase_0/pos" </string>
<string> "phase_0/color" </string>
<string> "phase_1/pos" </string>
<string> "phase_1/color" </string>
<string> "phase_2/pos" </string>
<string> "phase_2/color" </string>
<string> "phase_3/pos" </string>
<string> "phase_3/color" </string>
<string> "emission_points" </string>
<string> "color/color_ramp" </string>
<string> "sound" </string>
<string> "SamplePlayer2D" </string>
<string> "params/volume_db" </string>
@ -303,125 +287,154 @@
<string> "config/samples" </string>
<string> "config/pitch_random" </string>
</string_array>
<string> "version" </string>
<int> 1 </int>
<string> "conn_count" </string>
<int> 0 </int>
<string> "node_count" </string>
<int> 11 </int>
<string> "nodes" </string>
<int_array len="285"> -1, -1, 1, 0, -1, 29, 2, 0, 3, 1, 4, 2, 5, 0, 6, 1, 7, 3, 8, 0, 9, 1, 10, 4, 11, 0, 12, 5, 13, 5, 14, 6, 15, 7, 16, 8, 17, 8, 18, 7, 19, 0, 20, 9, 21, 10, 22, 0, 23, 0, 24, 11, 25, 12, 26, 8, 27, 13, 28, 13, 29, 14, 30, 15, 0, 0, 0, 32, 31, -1, 8, 33, 16, 34, 17, 35, 18, 36, 11, 37, 11, 38, 11, 39, 0, 40, 0, 0, 0, 0, 42, 41, -1, 10, 43, 5, 44, 8, 45, 19, 46, 20, 47, 21, 48, 22, 49, 11, 50, 23, 51, 24, 52, 25, 0, 0, 0, 54, 53, -1, 3, 55, 26, 56, 27, 57, 10, 0, 0, 0, 58, 58, -1, 4, 33, 28, 59, 1, 60, 0, 61, 29, 0, 0, 0, 58, 62, -1, 4, 33, 30, 59, 1, 60, 0, 61, 29, 0, 0, 0, 58, 63, -1, 4, 33, 31, 59, 1, 60, 0, 61, 29, 0, 0, 0, 65, 64, -1, 4, 33, 32, 66, 11, 67, 33, 68, 5, 0, 0, 0, 65, 69, -1, 4, 33, 34, 66, 11, 67, 33, 68, 5, 0, 0, 0, 70, 70, -1, 26, 71, 35, 72, 5, 73, 36, 74, 37, 75, 37, 76, 0, 77, 38, 78, 39, 79, 8, 80, 40, 81, 41, 82, 42, 83, 8, 84, 8, 85, 43, 86, 8, 87, 8, 88, 8, 89, 8, 90, 42, 91, 23, 92, 8, 93, 7, 94, 8, 95, 7, 96, 44, 0, 0, 0, 98, 97, -1, 8, 99, 8, 100, 7, 101, 7, 102, 45, 103, 7, 104, 46, 105, 47, 106, 8, 0 </int_array>
<string> "variants" </string>
<array len="51" shared="false">
<bool> True </bool>
<real> 1 </real>
<vector2> 0, 0 </vector2>
<real> 0 </real>
<vector2> 1, 1 </vector2>
<int> 3 </int>
<array len="48" shared="false">
<bool> False </bool>
<resource resource_type="Shape2D" path="local://1"> </resource>
<matrix32> 1, -0, 0, 1, -1.08072, -2.16144 </matrix32>
<bool> False </bool>
<matrix32> 1, -0, 0, 1, 6.48431, 3.24216 </matrix32>
<matrix32> 1, -0, 0, 1, -12.495, 3.53415 </matrix32>
<int> 1 </int>
<int> 2 </int>
<real> 1 </real>
<real> 0 </real>
<int> 0 </int>
<int> 4 </int>
<resource resource_type="Script" path="res://enemy.*"> </resource>
<bool> True </bool>
<vector2> 0, 0 </vector2>
<real> -1 </real>
<resource external="0"> </resource>
<dictionary shared="false">
<string> "__editor_plugin_screen__" </string>
<string> "2D" </string>
<string> "__editor_plugin_states__" </string>
<dictionary shared="false">
<string> "Script" </string>
<dictionary shared="false">
<string> "current" </string>
<int> 0 </int>
<string> "sources" </string>
<array len="1" shared="false">
<string> "res://enemy.gd" </string>
</array>
</dictionary>
<string> "2D" </string>
<dictionary shared="false">
<string> "pixel_snap" </string>
<bool> False </bool>
<string> "zoom" </string>
<real> 1.108033 </real>
<string> "ofs" </string>
<vector2> -227.625, -197.9 </vector2>
<string> "snap_grid" </string>
<bool> False </bool>
<string> "snap_offset" </string>
<vector2> 0, 0 </vector2>
<string> "snap_pixel" </string>
<bool> False </bool>
<string> "snap_relative" </string>
<bool> False </bool>
<string> "snap_rotation" </string>
<bool> False </bool>
<string> "snap_rotation_offset" </string>
<real> 0 </real>
<string> "snap_rotation_step" </string>
<real> 0.261799 </real>
<string> "snap_show_grid" </string>
<bool> False </bool>
<string> "snap_step" </string>
<vector2> 10, 10 </vector2>
<string> "zoom" </string>
<real> 1.108033 </real>
</dictionary>
<string> "3D" </string>
<dictionary shared="false">
<string> "zfar" </string>
<real> 500 </real>
<string> "ambient_light_color" </string>
<color> 0.15, 0.15, 0.15, 1 </color>
<string> "default_light" </string>
<bool> True </bool>
<string> "default_srgb" </string>
<bool> False </bool>
<string> "deflight_rot_x" </string>
<real> 0.942478 </real>
<string> "deflight_rot_y" </string>
<real> 0.628319 </real>
<string> "fov" </string>
<real> 45 </real>
<string> "show_grid" </string>
<bool> True </bool>
<string> "show_origin" </string>
<bool> True </bool>
<string> "viewport_mode" </string>
<int> 1 </int>
<string> "viewports" </string>
<array len="4" shared="false">
<dictionary shared="false">
<string> "distance" </string>
<real> 4 </real>
<string> "listener" </string>
<bool> True </bool>
<string> "pos" </string>
<vector3> 0, 0, 0 </vector3>
<string> "use_environment" </string>
<bool> False </bool>
<string> "use_orthogonal" </string>
<bool> False </bool>
<string> "x_rot" </string>
<real> 0 </real>
<string> "y_rot" </string>
<real> 0 </real>
<string> "use_orthogonal" </string>
<bool> False </bool>
<string> "use_environment" </string>
<bool> False </bool>
<string> "pos" </string>
<vector3> 0, 0, 0 </vector3>
</dictionary>
<dictionary shared="false">
<string> "distance" </string>
<real> 4 </real>
<string> "listener" </string>
<bool> False </bool>
<string> "pos" </string>
<vector3> 0, 0, 0 </vector3>
<string> "use_environment" </string>
<bool> False </bool>
<string> "use_orthogonal" </string>
<bool> False </bool>
<string> "x_rot" </string>
<real> 0 </real>
<string> "y_rot" </string>
<real> 0 </real>
<string> "use_orthogonal" </string>
<bool> False </bool>
<string> "use_environment" </string>
<bool> False </bool>
<string> "pos" </string>
<vector3> 0, 0, 0 </vector3>
</dictionary>
<dictionary shared="false">
<string> "distance" </string>
<real> 4 </real>
<string> "listener" </string>
<bool> False </bool>
<string> "pos" </string>
<vector3> 0, 0, 0 </vector3>
<string> "use_environment" </string>
<bool> False </bool>
<string> "use_orthogonal" </string>
<bool> False </bool>
<string> "x_rot" </string>
<real> 0 </real>
<string> "y_rot" </string>
<real> 0 </real>
<string> "use_orthogonal" </string>
<bool> False </bool>
<string> "use_environment" </string>
<bool> False </bool>
<string> "pos" </string>
<vector3> 0, 0, 0 </vector3>
</dictionary>
<dictionary shared="false">
<string> "distance" </string>
<real> 4 </real>
<string> "listener" </string>
<bool> False </bool>
<string> "pos" </string>
<vector3> 0, 0, 0 </vector3>
<string> "use_environment" </string>
<bool> False </bool>
<string> "use_orthogonal" </string>
<bool> False </bool>
<string> "x_rot" </string>
<real> 0 </real>
<string> "y_rot" </string>
<real> 0 </real>
<string> "use_orthogonal" </string>
<bool> False </bool>
<string> "use_environment" </string>
<bool> False </bool>
<string> "pos" </string>
<vector3> 0, 0, 0 </vector3>
</dictionary>
</array>
<string> "viewport_mode" </string>
<int> 1 </int>
<string> "default_light" </string>
<bool> True </bool>
<string> "show_grid" </string>
<bool> True </bool>
<string> "show_origin" </string>
<bool> True </bool>
<string> "zfar" </string>
<real> 500 </real>
<string> "znear" </string>
<real> 0.1 </real>
</dictionary>
<string> "Anim" </string>
<dictionary shared="false">
<string> "visible" </string>
<bool> False </bool>
</dictionary>
</dictionary>
<string> "__editor_run_settings__" </string>
<dictionary shared="false">
@ -430,28 +443,24 @@
<string> "run_mode" </string>
<int> 0 </int>
</dictionary>
<string> "__editor_plugin_screen__" </string>
<string> "2D" </string>
</dictionary>
<vector2> 16.2569, 11.0034 </vector2>
<vector2> 23.5056, 10.8629 </vector2>
<rect2> -10, -10, 20, 20 </rect2>
<int> 1 </int>
<node_path> ".." </node_path>
<resource resource_type="Animation" path="local://2"> </resource>
<resource resource_type="Animation" path="local://3"> </resource>
<resource resource_type="Animation" path="local://4"> </resource>
<resource resource_type="Animation" path="local://3"> </resource>
<real> 3 </real>
<array len="0" shared="false">
</array>
<string> "" </string>
<resource external="1"> </resource>
<int> 8 </int>
<vector2> -1.08072, -2.16144 </vector2>
<int> -1 </int>
<vector2> 6.48431, 3.24216 </vector2>
<vector2> -12.495, 3.53415 </vector2>
<resource resource_type="Texture" path="res://enemy.*"> </resource>
<int> 8 </int>
<color> 1, 1, 1, 1 </color>
<rect2> 0, 0, 0, 0 </rect2>
<vector2> -33.2868, -9.34363 </vector2>
<vector2> 0, 45 </vector2>
<vector2> 29.1987, -9.34363 </vector2>
@ -459,22 +468,18 @@
<int> 32 </int>
<real> 0.5 </real>
<real> 0.1 </real>
<resource resource_type="Texture" path="res://bullet.*"> </resource>
<resource external="2"> </resource>
<real> 180 </real>
<real> 90 </real>
<real> 2 </real>
<real> 9.8 </real>
<color> 1, 0.884956, 0.823009, 1 </color>
<color> 0.768627, 0.389381, 0, 0 </color>
<color> 0, 0, 0, 1 </color>
<vector2_array len="0"> </vector2_array>
<resource resource_type="ColorRamp" path="local://6"> </resource>
<real> 2048 </real>
<int> 3 </int>
<resource resource_type="SampleLibrary" path="local://5"> </resource>
</array>
<string> "nodes" </string>
<int_array len="445"> -1, -1, 1, 0, -1, 31, 2, 0, 3, 1, 4, 1, 5, 0, 6, 2, 7, 3, 8, 4, 9, 5, 10, 6, 11, 7, 12, 8, 13, 6, 14, 9, 15, 8, 16, 6, 17, 10, 18, 8, 19, 11, 20, 1, 21, 3, 22, 3, 23, 8, 24, 8, 25, 12, 26, 8, 27, 0, 28, 0, 29, 2, 30, 3, 31, 13, 32, 14, 0, 0, 0, 34, 33, -1, 10, 2, 0, 3, 1, 4, 1, 5, 0, 6, 15, 7, 3, 8, 16, 35, 17, 36, 0, 37, 0, 0, 0, 0, 39, 38, -1, 10, 40, 18, 41, 3, 42, 19, 43, 20, 44, 21, 45, 22, 46, 0, 47, 23, 48, 24, 49, 25, 0, 0, 0, 50, 50, -1, 9, 2, 0, 3, 1, 4, 1, 5, 0, 6, 26, 7, 3, 8, 4, 51, 6, 52, 8, 0, 0, 0, 50, 53, -1, 9, 2, 0, 3, 1, 4, 1, 5, 0, 6, 27, 7, 3, 8, 4, 51, 6, 52, 8, 0, 0, 0, 50, 54, -1, 9, 2, 0, 3, 1, 4, 1, 5, 0, 6, 28, 7, 3, 8, 4, 51, 6, 52, 8, 0, 0, 0, 56, 55, -1, 18, 2, 0, 3, 1, 4, 1, 5, 0, 6, 2, 7, 3, 8, 4, 57, 29, 58, 0, 59, 2, 60, 8, 61, 8, 62, 18, 63, 30, 64, 12, 65, 31, 66, 8, 67, 32, 0, 0, 0, 69, 68, -1, 9, 2, 0, 3, 1, 4, 1, 5, 0, 6, 33, 7, 3, 8, 4, 70, 0, 71, 34, 0, 0, 0, 69, 72, -1, 9, 2, 0, 3, 1, 4, 1, 5, 0, 6, 35, 7, 3, 8, 4, 70, 0, 71, 34, 0, 0, 0, 73, 73, -1, 55, 2, 0, 3, 1, 4, 36, 5, 0, 74, 18, 6, 2, 7, 3, 8, 4, 75, 37, 76, 38, 77, 1, 78, 3, 79, 38, 80, 8, 81, 2, 82, 2, 83, 0, 84, 39, 85, 40, 86, 3, 87, 41, 88, 42, 89, 43, 90, 3, 91, 3, 92, 44, 93, 3, 94, 3, 95, 3, 96, 43, 97, 23, 98, 3, 99, 3, 100, 3, 101, 3, 102, 1, 103, 3, 104, 3, 105, 3, 106, 3, 107, 3, 108, 3, 109, 3, 110, 3, 111, 3, 112, 11, 113, 3, 114, 45, 115, 1, 116, 46, 117, 1, 118, 47, 119, 1, 120, 47, 121, 48, 0, 0, 0, 123, 122, -1, 15, 2, 0, 3, 1, 4, 1, 5, 0, 6, 2, 7, 3, 8, 4, 124, 3, 125, 1, 126, 1, 127, 49, 128, 1, 129, 5, 130, 50, 131, 3, 0 </int_array>
<string> "conns" </string>
<int_array len="0"> </int_array>
<string> "version" </string>
<int> 1 </int>
</dictionary>
</main_resource>

3
demos/2d/platformer/engine.cfg
View file

@ -10,9 +10,6 @@ target_fps="60"
width=800
height=480
#stretch_2d=false
#stretch_mode="viewport"
#stretch_aspect="keep"
stretch_mode="2d"
stretch_aspect="keep_height"

56
demos/2d/platformer/stage.xml
View file

File diff suppressed because one or more lines are too long

6
demos/3d/platformer/player.gd
View file

@ -69,9 +69,9 @@ func _integrate_forces( state ):
var lv = state.get_linear_velocity() # linear velocity
var g = state.get_total_gravity()
var delta = state.get_step()
var d = 1.0 - delta*state.get_total_density()
if (d<0):
d=0
# var d = 1.0 - delta*state.get_total_density()
# if (d<0):
# d=0
lv += g * delta #apply gravity
var anim = ANIM_FLOOR

6328
doc/base/classes.xml
View file

File diff suppressed because it is too large Load diff

8
drivers/SCsub
View file

@ -31,15 +31,17 @@ SConscript("rtaudio/SCsub");
SConscript("nedmalloc/SCsub");
SConscript("nrex/SCsub");
SConscript("chibi/SCsub");
if (env["vorbis"]=="yes" or env["speex"]=="yes" or env["theora"]=="yes"):
if (env["vorbis"]=="yes" or env["speex"]=="yes" or env["theora"]=="yes" or env["opus"]=="yes"):
SConscript("ogg/SCsub");
if (env["vorbis"]=="yes"):
SConscript("vorbis/SCsub");
if (env["opus"]=="yes"):
SConscript('opus/SCsub');
if (env["tools"]=="yes"):
SConscript("convex_decomp/SCsub");
if env["theora"]=="yes":
SConscript("theoraplayer/SCsub")
#if env["theora"]=="yes":
# SConscript("theoraplayer/SCsub")
if (env["theora"]=="yes"):
SConscript("theora/SCsub");
if (env['speex']=='yes'):

3
drivers/convex_decomp/b2Glue.h
View file

@ -20,7 +20,8 @@
#define B2GLUE_H
#include "math_2d.h"
#include <limits>
#include <limits.h>
namespace b2ConvexDecomp {
typedef real_t float32;

4
drivers/convex_decomp/b2Polygon.cpp
View file

@ -21,8 +21,8 @@
#include "b2Triangle.h"
#include "b2Polygon.h"
#include <cmath>
#include <climits>
#include <math.h>
#include <limits.h>
#include <assert.h>
#define b2Assert assert

2
drivers/convex_decomp/b2Polygon.h
View file

@ -22,7 +22,7 @@
#include "b2Triangle.h"
#include "stdio.h"
#include <string.h>
#include <limits>
#include <limits.h>
namespace b2ConvexDecomp {
static bool B2_POLYGON_REPORT_ERRORS = false;

4908
drivers/etc1/rg_etc1.cpp
View file

File diff suppressed because it is too large Load diff

152
drivers/etc1/rg_etc1.h
View file

@ -1,76 +1,76 @@
// File: rg_etc1.h - Fast, high quality ETC1 block packer/unpacker - Rich Geldreich <richgel99@gmail.com>
// Please see ZLIB license at the end of this file.
#pragma once
namespace rg_etc1
{
// Unpacks an 8-byte ETC1 compressed block to a block of 4x4 32bpp RGBA pixels.
// Returns false if the block is invalid. Invalid blocks will still be unpacked with clamping.
// This function is thread safe, and does not dynamically allocate any memory.
// If preserve_alpha is true, the alpha channel of the destination pixels will not be overwritten. Otherwise, alpha will be set to 255.
bool unpack_etc1_block(const void *pETC1_block, unsigned int* pDst_pixels_rgba, bool preserve_alpha = false);
// Quality setting = the higher the quality, the slower.
// To pack large textures, it is highly recommended to call pack_etc1_block() in parallel, on different blocks, from multiple threads (particularly when using cHighQuality).
enum etc1_quality
{
cLowQuality,
cMediumQuality,
cHighQuality,
};
struct etc1_pack_params
{
etc1_quality m_quality;
bool m_dithering;
inline etc1_pack_params()
{
clear();
}
void clear()
{
m_quality = cHighQuality;
m_dithering = false;
}
};
// Important: pack_etc1_block_init() must be called before calling pack_etc1_block().
void pack_etc1_block_init();
// Packs a 4x4 block of 32bpp RGBA pixels to an 8-byte ETC1 block.
// 32-bit RGBA pixels must always be arranged as (R,G,B,A) (R first, A last) in memory, independent of platform endianness. A should always be 255.
// Returns squared error of result.
// This function is thread safe, and does not dynamically allocate any memory.
// pack_etc1_block() does not currently support "perceptual" colorspace metrics - it primarily optimizes for RGB RMSE.
unsigned int pack_etc1_block(void* pETC1_block, const unsigned int* pSrc_pixels_rgba, etc1_pack_params& pack_params);
} // namespace rg_etc1
//------------------------------------------------------------------------------
//
// rg_etc1 uses the ZLIB license:
// http://opensource.org/licenses/Zlib
//
// Copyright (c) 2012 Rich Geldreich
//
// This software is provided 'as-is', without any express or implied
// warranty. In no event will the authors be held liable for any damages
// arising from the use of this software.
//
// Permission is granted to anyone to use this software for any purpose,
// including commercial applications, and to alter it and redistribute it
// freely, subject to the following restrictions:
//
// 1. The origin of this software must not be misrepresented; you must not
// claim that you wrote the original software. If you use this software
// in a product, an acknowledgment in the product documentation would be
// appreciated but is not required.
//
// 2. Altered source versions must be plainly marked as such, and must not be
// misrepresented as being the original software.
//
// 3. This notice may not be removed or altered from any source distribution.
//
//------------------------------------------------------------------------------
// File: rg_etc1.h - Fast, high quality ETC1 block packer/unpacker - Rich Geldreich <richgel99@gmail.com>
// Please see ZLIB license at the end of this file.
#pragma once
namespace rg_etc1
{
// Unpacks an 8-byte ETC1 compressed block to a block of 4x4 32bpp RGBA pixels.
// Returns false if the block is invalid. Invalid blocks will still be unpacked with clamping.
// This function is thread safe, and does not dynamically allocate any memory.
// If preserve_alpha is true, the alpha channel of the destination pixels will not be overwritten. Otherwise, alpha will be set to 255.
bool unpack_etc1_block(const void *pETC1_block, unsigned int* pDst_pixels_rgba, bool preserve_alpha = false);
// Quality setting = the higher the quality, the slower.
// To pack large textures, it is highly recommended to call pack_etc1_block() in parallel, on different blocks, from multiple threads (particularly when using cHighQuality).
enum etc1_quality
{
cLowQuality,
cMediumQuality,
cHighQuality,
};
struct etc1_pack_params
{
etc1_quality m_quality;
bool m_dithering;
inline etc1_pack_params()
{
clear();
}
void clear()
{
m_quality = cHighQuality;
m_dithering = false;
}
};
// Important: pack_etc1_block_init() must be called before calling pack_etc1_block().
void pack_etc1_block_init();
// Packs a 4x4 block of 32bpp RGBA pixels to an 8-byte ETC1 block.
// 32-bit RGBA pixels must always be arranged as (R,G,B,A) (R first, A last) in memory, independent of platform endianness. A should always be 255.
// Returns squared error of result.
// This function is thread safe, and does not dynamically allocate any memory.
// pack_etc1_block() does not currently support "perceptual" colorspace metrics - it primarily optimizes for RGB RMSE.
unsigned int pack_etc1_block(void* pETC1_block, const unsigned int* pSrc_pixels_rgba, etc1_pack_params& pack_params);
} // namespace rg_etc1
//------------------------------------------------------------------------------
//
// rg_etc1 uses the ZLIB license:
// http://opensource.org/licenses/Zlib
//
// Copyright (c) 2012 Rich Geldreich
//
// This software is provided 'as-is', without any express or implied
// warranty. In no event will the authors be held liable for any damages
// arising from the use of this software.
//
// Permission is granted to anyone to use this software for any purpose,
// including commercial applications, and to alter it and redistribute it
// freely, subject to the following restrictions:
//
// 1. The origin of this software must not be misrepresented; you must not
// claim that you wrote the original software. If you use this software
// in a product, an acknowledgment in the product documentation would be
// appreciated but is not required.
//
// 2. Altered source versions must be plainly marked as such, and must not be
// misrepresented as being the original software.
//
// 3. This notice may not be removed or altered from any source distribution.
//
//------------------------------------------------------------------------------

15
drivers/gles2/rasterizer_gles2.cpp
View file

@ -4145,7 +4145,7 @@ void RasterizerGLES2::begin_frame() {
//fragment_lighting=Globals::get_singleton()->get("rasterizer/use_fragment_lighting");
#ifdef TOOLS_ENABLED
canvas_shader.set_conditional(CanvasShaderGLES2::USE_PIXEL_SNAP,GLOBAL_DEF("rasterizer/use_pixel_snap",false));
canvas_shader.set_conditional(CanvasShaderGLES2::USE_PIXEL_SNAP,GLOBAL_DEF("display/use_2d_pixel_snap",false));
shadow_filter=ShadowFilterTechnique(int(Globals::get_singleton()->get("rasterizer/shadow_filter")));
#endif
@ -4160,7 +4160,6 @@ void RasterizerGLES2::begin_frame() {
time_delta=time-last_time;
last_time=time;
frame++;
clear_viewport(Color(1,0,0.5));
_rinfo.vertex_count=0;
_rinfo.object_count=0;
@ -5970,6 +5969,10 @@ void RasterizerGLES2::_render(const Geometry *p_geometry,const Material *p_mater
if (element_count==0)
return;
if (mm->visible>=0) {
element_count=MIN(element_count,mm->visible);
}
const MultiMesh::Element *elements=&mm->elements[0];
_rinfo.vertex_count+=s->array_len*element_count;
@ -10804,7 +10807,7 @@ void RasterizerGLES2::init() {
copy_shader.set_conditional(CopyShaderGLES2::USE_8BIT_HDR,!use_fp16_fb);
canvas_shader.set_conditional(CanvasShaderGLES2::USE_DEPTH_SHADOWS,read_depth_supported);
canvas_shader.set_conditional(CanvasShaderGLES2::USE_PIXEL_SNAP,GLOBAL_DEF("rasterizer/use_pixel_snap",false));
canvas_shader.set_conditional(CanvasShaderGLES2::USE_PIXEL_SNAP,GLOBAL_DEF("display/use_2d_pixel_snap",false));
npo2_textures_available=true;
//fragment_lighting=false;
@ -11188,6 +11191,12 @@ RasterizerGLES2::RasterizerGLES2(bool p_compress_arrays,bool p_keep_ram_copy,boo
tc0_idx=0;
};
void RasterizerGLES2::restore_framebuffer() {
glBindFramebuffer(GL_FRAMEBUFFER, base_framebuffer);
}
RasterizerGLES2::~RasterizerGLES2() {
memdelete_arr(skinned_buffer);

2
drivers/gles2/rasterizer_gles2.h
View file

@ -1695,6 +1695,8 @@ public:
void reload_vram();
virtual bool has_feature(VS::Features p_feature) const;
virtual void restore_framebuffer();
static RasterizerGLES2* get_singleton();

128
drivers/mpc/audio_stream_mpc.cpp
View file

@ -1,7 +1,7 @@
#include "audio_stream_mpc.h"
Error AudioStreamMPC::_open_file() {
Error AudioStreamPlaybackMPC::_open_file() {
if (f) {
memdelete(f);
@ -41,7 +41,7 @@ Error AudioStreamMPC::_open_file() {
return OK;
}
void AudioStreamMPC::_close_file() {
void AudioStreamPlaybackMPC::_close_file() {
if (f) {
memdelete(f);
@ -52,7 +52,7 @@ void AudioStreamMPC::_close_file() {
data_ofs=0;
}
int AudioStreamMPC::_read_file(void *p_dst,int p_bytes) {
int AudioStreamPlaybackMPC::_read_file(void *p_dst,int p_bytes) {
if (f)
return f->get_buffer((uint8_t*)p_dst,p_bytes);
@ -68,7 +68,7 @@ int AudioStreamMPC::_read_file(void *p_dst,int p_bytes) {
return p_bytes;
}
bool AudioStreamMPC::_seek_file(int p_pos){
bool AudioStreamPlaybackMPC::_seek_file(int p_pos){
if (p_pos<0 || p_pos>streamlen)
return false;
@ -83,7 +83,7 @@ bool AudioStreamMPC::_seek_file(int p_pos){
return true;
}
int AudioStreamMPC::_tell_file() const{
int AudioStreamPlaybackMPC::_tell_file() const{
if (f)
return f->get_pos();
@ -93,13 +93,13 @@ int AudioStreamMPC::_tell_file() const{
}
int AudioStreamMPC::_sizeof_file() const{
int AudioStreamPlaybackMPC::_sizeof_file() const{
//print_line("sizeof file, get: "+itos(streamlen));
return streamlen;
}
bool AudioStreamMPC::_canseek_file() const{
bool AudioStreamPlaybackMPC::_canseek_file() const{
//print_line("canseek file, get true");
return true;
@ -107,51 +107,46 @@ bool AudioStreamMPC::_canseek_file() const{
/////////////////////
mpc_int32_t AudioStreamMPC::_mpc_read(mpc_reader *p_reader,void *p_dst, mpc_int32_t p_bytes) {
mpc_int32_t AudioStreamPlaybackMPC::_mpc_read(mpc_reader *p_reader,void *p_dst, mpc_int32_t p_bytes) {
AudioStreamMPC *smpc=(AudioStreamMPC *)p_reader->data;
AudioStreamPlaybackMPC *smpc=(AudioStreamPlaybackMPC *)p_reader->data;
return smpc->_read_file(p_dst,p_bytes);
}
mpc_bool_t AudioStreamMPC::_mpc_seek(mpc_reader *p_reader,mpc_int32_t p_offset) {
mpc_bool_t AudioStreamPlaybackMPC::_mpc_seek(mpc_reader *p_reader,mpc_int32_t p_offset) {
AudioStreamMPC *smpc=(AudioStreamMPC *)p_reader->data;
AudioStreamPlaybackMPC *smpc=(AudioStreamPlaybackMPC *)p_reader->data;
return smpc->_seek_file(p_offset);
}
mpc_int32_t AudioStreamMPC::_mpc_tell(mpc_reader *p_reader) {
mpc_int32_t AudioStreamPlaybackMPC::_mpc_tell(mpc_reader *p_reader) {
AudioStreamMPC *smpc=(AudioStreamMPC *)p_reader->data;
AudioStreamPlaybackMPC *smpc=(AudioStreamPlaybackMPC *)p_reader->data;
return smpc->_tell_file();
}
mpc_int32_t AudioStreamMPC::_mpc_get_size(mpc_reader *p_reader) {
mpc_int32_t AudioStreamPlaybackMPC::_mpc_get_size(mpc_reader *p_reader) {
AudioStreamMPC *smpc=(AudioStreamMPC *)p_reader->data;
AudioStreamPlaybackMPC *smpc=(AudioStreamPlaybackMPC *)p_reader->data;
return smpc->_sizeof_file();
}
mpc_bool_t AudioStreamMPC::_mpc_canseek(mpc_reader *p_reader) {
mpc_bool_t AudioStreamPlaybackMPC::_mpc_canseek(mpc_reader *p_reader) {
AudioStreamMPC *smpc=(AudioStreamMPC *)p_reader->data;
AudioStreamPlaybackMPC *smpc=(AudioStreamPlaybackMPC *)p_reader->data;
return smpc->_canseek_file();
}
bool AudioStreamMPC::_can_mix() const {
return /*active &&*/ !paused;
}
void AudioStreamMPC::update() {
int AudioStreamPlaybackMPC::mix(int16_t* p_bufer,int p_frames) {
if (!active || paused)
return;
return 0;
int todo=get_todo();
int todo=p_frames;
while(todo>MPC_DECODER_BUFFER_LENGTH/si.channels) {
@ -162,7 +157,7 @@ void AudioStreamMPC::update() {
mpc_status err = mpc_demux_decode(demux, &frame);
if (frame.bits!=-1) {
int16_t *dst_buff = get_write_buffer();
int16_t *dst_buff = p_bufer;
#ifdef MPC_FIXED_POINT
@ -185,21 +180,21 @@ void AudioStreamMPC::update() {
#endif
int frames = frame.samples;
write(frames);
p_bufer+=si.channels*frames;
todo-=frames;
} else {
if (err != MPC_STATUS_OK) {
stop();
ERR_EXPLAIN("Error decoding MPC");
ERR_FAIL();
ERR_PRINT("Error decoding MPC");
break;
} else {
//finished
if (!loop) {
stop();
return;
break;
} else {
@ -213,9 +208,11 @@ void AudioStreamMPC::update() {
}
}
}
return p_frames-todo;
}
Error AudioStreamMPC::_reload() {
Error AudioStreamPlaybackMPC::_reload() {
ERR_FAIL_COND_V(demux!=NULL, ERR_FILE_ALREADY_IN_USE);
@ -224,31 +221,40 @@ Error AudioStreamMPC::_reload() {
demux = mpc_demux_init(&reader);
ERR_FAIL_COND_V(!demux,ERR_CANT_CREATE);
mpc_demux_get_info(demux, &si);
_setup(si.channels,si.sample_freq,MPC_DECODER_BUFFER_LENGTH*2/si.channels);
return OK;
}
void AudioStreamMPC::set_file(const String& p_file) {
void AudioStreamPlaybackMPC::set_file(const String& p_file) {
file=p_file;
Error err = _open_file();
ERR_FAIL_COND(err!=OK);
demux = mpc_demux_init(&reader);
ERR_FAIL_COND(!demux);
mpc_demux_get_info(demux, &si);
stream_min_size=MPC_DECODER_BUFFER_LENGTH*2/si.channels;
stream_rate=si.sample_freq;
stream_channels=si.channels;
mpc_demux_exit(demux);
demux=NULL;
_close_file();
}
String AudioStreamMPC::get_file() const {
String AudioStreamPlaybackMPC::get_file() const {
return file;
}
void AudioStreamMPC::play() {
void AudioStreamPlaybackMPC::play(float p_offset) {
_THREAD_SAFE_METHOD_
if (active)
stop();
active=false;
@ -262,9 +268,9 @@ void AudioStreamMPC::play() {
}
void AudioStreamMPC::stop() {
void AudioStreamPlaybackMPC::stop() {
_THREAD_SAFE_METHOD_
if (!active)
return;
if (demux) {
@ -275,70 +281,58 @@ void AudioStreamMPC::stop() {
active=false;
}
bool AudioStreamMPC::is_playing() const {
bool AudioStreamPlaybackMPC::is_playing() const {
return active || (get_total() - get_todo() -1 > 0);
return active;
}
void AudioStreamMPC::set_paused(bool p_paused) {
paused=p_paused;
}
bool AudioStreamMPC::is_paused(bool p_paused) const {
return paused;
}
void AudioStreamMPC::set_loop(bool p_enable) {
void AudioStreamPlaybackMPC::set_loop(bool p_enable) {
loop=p_enable;
}
bool AudioStreamMPC::has_loop() const {
bool AudioStreamPlaybackMPC::has_loop() const {
return loop;
}
float AudioStreamMPC::get_length() const {
float AudioStreamPlaybackMPC::get_length() const {
return 0;
}
String AudioStreamMPC::get_stream_name() const {
String AudioStreamPlaybackMPC::get_stream_name() const {
return "";
}
int AudioStreamMPC::get_loop_count() const {
int AudioStreamPlaybackMPC::get_loop_count() const {
return 0;
}
float AudioStreamMPC::get_pos() const {
float AudioStreamPlaybackMPC::get_pos() const {
return 0;
}
void AudioStreamMPC::seek_pos(float p_time) {
void AudioStreamPlaybackMPC::seek_pos(float p_time) {
}
AudioStream::UpdateMode AudioStreamMPC::get_update_mode() const {
return UPDATE_THREAD;
}
void AudioStreamPlaybackMPC::_bind_methods() {
void AudioStreamMPC::_bind_methods() {
ObjectTypeDB::bind_method(_MD("set_file","name"),&AudioStreamMPC::set_file);
ObjectTypeDB::bind_method(_MD("get_file"),&AudioStreamMPC::get_file);
ObjectTypeDB::bind_method(_MD("set_file","name"),&AudioStreamPlaybackMPC::set_file);
ObjectTypeDB::bind_method(_MD("get_file"),&AudioStreamPlaybackMPC::get_file);
ADD_PROPERTYNZ( PropertyInfo(Variant::STRING,"file",PROPERTY_HINT_FILE,"mpc"), _SCS("set_file"), _SCS("get_file"));
}
AudioStreamMPC::AudioStreamMPC() {
AudioStreamPlaybackMPC::AudioStreamPlaybackMPC() {
preload=true;
preload=false;
f=NULL;
streamlen=0;
data_ofs=0;
@ -356,7 +350,7 @@ AudioStreamMPC::AudioStreamMPC() {
}
AudioStreamMPC::~AudioStreamMPC() {
AudioStreamPlaybackMPC::~AudioStreamPlaybackMPC() {
stop();

45
drivers/mpc/audio_stream_mpc.h
View file

@ -1,18 +1,17 @@
#ifndef AUDIO_STREAM_MPC_H
#define AUDIO_STREAM_MPC_H
#include "scene/resources/audio_stream_resampled.h"
#include "scene/resources/audio_stream.h"
#include "os/file_access.h"
#include "mpc/mpcdec.h"
#include "os/thread_safe.h"
#include "io/resource_loader.h"
//#include "../libmpcdec/decoder.h"
//#include "../libmpcdec/internal.h"
class AudioStreamMPC : public AudioStreamResampled {
OBJ_TYPE( AudioStreamMPC, AudioStreamResampled );
class AudioStreamPlaybackMPC : public AudioStreamPlayback {
_THREAD_SAFE_CLASS_
OBJ_TYPE( AudioStreamPlaybackMPC, AudioStreamPlayback );
bool preload;
FileAccess *f;
@ -39,7 +38,9 @@ class AudioStreamMPC : public AudioStreamResampled {
static mpc_int32_t _mpc_get_size(mpc_reader *p_reader);
static mpc_bool_t _mpc_canseek(mpc_reader *p_reader);
virtual bool _can_mix() const ;
int stream_min_size;
int stream_rate;
int stream_channels;
protected:
Error _open_file();
@ -59,12 +60,10 @@ public:
void set_file(const String& p_file);
String get_file() const;
virtual void play();
virtual void play(float p_offset=0);
virtual void stop();
virtual bool is_playing() const;
virtual void set_paused(bool p_paused);
virtual bool is_paused(bool p_paused) const;
virtual void set_loop(bool p_enable);
virtual bool has_loop() const;
@ -78,13 +77,35 @@ public:
virtual float get_pos() const;
virtual void seek_pos(float p_time);
virtual UpdateMode get_update_mode() const;
virtual void update();
virtual int get_channels() const { return stream_channels; }
virtual int get_mix_rate() const { return stream_rate; }
AudioStreamMPC();
~AudioStreamMPC();
virtual int get_minimum_buffer_size() const { return stream_min_size; }
virtual int mix(int16_t* p_bufer,int p_frames);
virtual void set_loop_restart_time(float p_time) { }
AudioStreamPlaybackMPC();
~AudioStreamPlaybackMPC();
};
class AudioStreamMPC : public AudioStream {
OBJ_TYPE( AudioStreamMPC, AudioStream );
String file;
public:
Ref<AudioStreamPlayback> instance_playback() {
Ref<AudioStreamPlaybackMPC> pb = memnew( AudioStreamPlaybackMPC );
pb->set_file(file);
return pb;
}
void set_file(const String& p_file) { file=p_file; }
};
class ResourceFormatLoaderAudioStreamMPC : public ResourceFormatLoader {
public:

11628
drivers/nedmalloc/malloc.c.h
View file

File diff suppressed because it is too large Load diff

2934
drivers/nedmalloc/nedmalloc.cpp
View file

File diff suppressed because it is too large Load diff

604
drivers/nedmalloc/nedmalloc.h
View file

@ -1,302 +1,302 @@
#ifdef NEDMALLOC_ENABLED
/* nedalloc, an alternative malloc implementation for multiple threads without
lock contention based on dlmalloc v2.8.3. (C) 2005-2009 Niall Douglas
Boost Software License - Version 1.0 - August 17th, 2003
Permission is hereby granted, free of charge, to any person or organization
obtaining a copy of the software and accompanying documentation covered by
this license (the "Software") to use, reproduce, display, distribute,
execute, and transmit the Software, and to prepare derivative works of the
Software, and to permit third-parties to whom the Software is furnished to
do so, all subject to the following:
The copyright notices in the Software and this entire statement, including
the above license grant, this restriction and the following disclaimer,
must be included in all copies of the Software, in whole or in part, and
all derivative works of the Software, unless such copies or derivative
works are solely in the form of machine-executable object code generated by
a source language processor.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT
SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE
FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE,
ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
DEALINGS IN THE SOFTWARE.
*/
#ifndef NEDMALLOC_H
#define NEDMALLOC_H
#include "typedefs.h"
#define MALLOC_ALIGNMENT DEFAULT_ALIGNMENT
#ifdef PSP_ENABLED
#define USE_LOCKS 0
#define HAVE_MMAP 0
#endif
/* See malloc.c.h for what each function does.
REPLACE_SYSTEM_ALLOCATOR on POSIX causes nedalloc's functions to be called
malloc, free etc. instead of nedmalloc, nedfree etc. You may or may not want
this. On Windows it causes nedmalloc to patch all loaded DLLs and binaries
to replace usage of the system allocator.
NO_NED_NAMESPACE prevents the functions from being defined in the nedalloc
namespace when in C++ (uses the global namespace instead).
NEDMALLOCEXTSPEC can be defined to be __declspec(dllexport) or
__attribute__ ((visibility("default"))) or whatever you like. It defaults
to extern unless NEDMALLOC_DLL_EXPORTS is set as it would be when building
nedmalloc.dll.
USE_LOCKS can be 2 if you want to define your own MLOCK_T, INITIAL_LOCK,
ACQUIRE_LOCK, RELEASE_LOCK, TRY_LOCK, IS_LOCKED and NULL_LOCK_INITIALIZER.
NEDMALLOC_DEBUG can be defined to cause DEBUG to be set differently for nedmalloc
than for the rest of the build. Remember to set NDEBUG to disable all assertion
checking too.
USE_MAGIC_HEADERS causes nedalloc to allocate an extra three sizeof(size_t)
to each block. nedpfree() and nedprealloc() can then automagically know when
to free a system allocated block. Enabling this typically adds 20-50% to
application memory usage.
ENABLE_TOLERANT_NEDMALLOC is automatically turned on if REPLACE_SYSTEM_ALLOCATOR
is set or the Windows DLL is being built. This causes nedmalloc to detect when a
system allocator block is passed to it and to handle it appropriately. Note that
without USE_MAGIC_HEADERS there is a very tiny chance that nedmalloc will segfault
on non-Windows builds (it uses Win32 SEH to trap segfaults on Windows and there
is no comparable system on POSIX).
USE_ALLOCATOR can be one of these settings (it defaults to 1):
0: System allocator (nedmalloc now simply acts as a threadcache).
WARNING: Intended for DEBUG USE ONLY - not all functions work correctly.
1: dlmalloc
ENABLE_LARGE_PAGES enables support for requesting memory from the system in large
(typically >=2Mb) pages if the host OS supports this. These occupy just a single
TLB entry and can significantly improve performance in large working set applications.
ENABLE_FAST_HEAP_DETECTION enables special logic to detect blocks allocated
by the system heap. This avoids 1.5%-2% overhead when checking for non-nedmalloc
blocks, but it assumes that the NT and glibc heaps function in a very specific
fashion which may not hold true across OS upgrades.
*/
#include <stddef.h> /* for size_t */
#ifndef NEDMALLOCEXTSPEC
#ifdef NEDMALLOC_DLL_EXPORTS
#ifdef WIN32
#define NEDMALLOCEXTSPEC extern __declspec(dllexport)
#elif defined(__GNUC__)
#define NEDMALLOCEXTSPEC extern __attribute__ ((visibility("default")))
#endif
#ifndef ENABLE_TOLERANT_NEDMALLOC
#define ENABLE_TOLERANT_NEDMALLOC 1
#endif
#else
#define NEDMALLOCEXTSPEC extern
#endif
#endif
#if __STDC_VERSION__ >= 199901L /* C99 or better */
#define RESTRICT restrict
#else
#if defined(_MSC_VER) && _MSC_VER>=1400
#define RESTRICT __restrict
#endif
#ifdef __GNUC__
#define RESTRICT __restrict
#endif
#endif
#ifndef RESTRICT
#define RESTRICT
#endif
#if defined(_MSC_VER) && _MSC_VER>=1400
#define NEDMALLOCPTRATTR __declspec(restrict)
#define NEDMALLOCNOALIASATTR __declspec(noalias)
#endif
#ifdef __GNUC__
#define NEDMALLOCPTRATTR __attribute__ ((malloc))
#endif
#ifndef NEDMALLOCPTRATTR
#define NEDMALLOCPTRATTR
#endif
#ifndef NEDMALLOCNOALIASATTR
#define NEDMALLOCNOALIASATTR
#endif
#ifndef USE_MAGIC_HEADERS
#define USE_MAGIC_HEADERS 0
#endif
#ifndef USE_ALLOCATOR
#define USE_ALLOCATOR 1 /* dlmalloc */
#endif
#if !USE_ALLOCATOR && !USE_MAGIC_HEADERS
#error If you are using the system allocator then you MUST use magic headers
#endif
#ifdef REPLACE_SYSTEM_ALLOCATOR
#if USE_ALLOCATOR==0
#error Cannot combine using the system allocator with replacing the system allocator
#endif
#ifndef ENABLE_TOLERANT_NEDMALLOC
#define ENABLE_TOLERANT_NEDMALLOC 1
#endif
#ifndef WIN32 /* We have a dedicated patcher for Windows */
#define nedmalloc malloc
#define nedcalloc calloc
#define nedrealloc realloc
#define nedfree free
#define nedmemalign memalign
#define nedmallinfo mallinfo
#define nedmallopt mallopt
#define nedmalloc_trim malloc_trim
#define nedmalloc_stats malloc_stats
#define nedmalloc_footprint malloc_footprint
#define nedindependent_calloc independent_calloc
#define nedindependent_comalloc independent_comalloc
#ifdef _MSC_VER
#define nedblksize _msize
#endif
#endif
#endif
#if defined(__cplusplus)
extern "C" {
#endif
struct nedmallinfo {
size_t arena; /* non-mmapped space allocated from system */
size_t ordblks; /* number of free chunks */
size_t smblks; /* always 0 */
size_t hblks; /* always 0 */
size_t hblkhd; /* space in mmapped regions */
size_t usmblks; /* maximum total allocated space */
size_t fsmblks; /* always 0 */
size_t uordblks; /* total allocated space */
size_t fordblks; /* total free space */
size_t keepcost; /* releasable (via malloc_trim) space */
};
#if defined(__cplusplus)
}
#endif
#if defined(__cplusplus)
#if !defined(NO_NED_NAMESPACE)
namespace nedalloc {
#else
extern "C" {
#endif
#define THROWSPEC throw()
#else
#define THROWSPEC
#endif
/* These are the global functions */
/* Gets the usable size of an allocated block. Note this will always be bigger than what was
asked for due to rounding etc. Optionally returns 1 in isforeign if the block came from the
system allocator - note that there is a small (>0.01%) but real chance of segfault on non-Windows
systems when passing non-nedmalloc blocks if you don't use USE_MAGIC_HEADERS.
*/
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR size_t nedblksize(int *RESTRICT isforeign, void *RESTRICT mem) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR void nedsetvalue(void *v) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedmalloc(size_t size) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedcalloc(size_t no, size_t size) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedrealloc(void *mem, size_t size) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR void nedfree(void *mem) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedmemalign(size_t alignment, size_t bytes) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR struct nedmallinfo nedmallinfo(void) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR int nedmallopt(int parno, int value) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR void* nedmalloc_internals(size_t *granularity, size_t *magic) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR int nedmalloc_trim(size_t pad) THROWSPEC;
NEDMALLOCEXTSPEC void nedmalloc_stats(void) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR size_t nedmalloc_footprint(void) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void **nedindependent_calloc(size_t elemsno, size_t elemsize, void **chunks) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void **nedindependent_comalloc(size_t elems, size_t *sizes, void **chunks) THROWSPEC;
/* Destroys the system memory pool used by the functions above.
Useful for when you have nedmalloc in a DLL you're about to unload.
If you call ANY nedmalloc functions after calling this you will
get a fatal exception!
*/
NEDMALLOCEXTSPEC void neddestroysyspool() THROWSPEC;
/* These are the pool functions */
struct nedpool_t;
typedef struct nedpool_t nedpool;
/* Creates a memory pool for use with the nedp* functions below.
Capacity is how much to allocate immediately (if you know you'll be allocating a lot
of memory very soon) which you can leave at zero. Threads specifies how many threads
will *normally* be accessing the pool concurrently. Setting this to zero means it
extends on demand, but be careful of this as it can rapidly consume system resources
where bursts of concurrent threads use a pool at once.
*/
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR nedpool *nedcreatepool(size_t capacity, int threads) THROWSPEC;
/* Destroys a memory pool previously created by nedcreatepool().
*/
NEDMALLOCEXTSPEC void neddestroypool(nedpool *p) THROWSPEC;
/* Returns a zero terminated snapshot of threadpools existing at the time of call. Call
nedfree() on the returned list when you are done. Returns zero if there is only the
system pool in existence.
*/
NEDMALLOCEXTSPEC nedpool **nedpoollist() THROWSPEC;
/* Sets a value to be associated with a pool. You can retrieve this value by passing
any memory block allocated from that pool.
*/
NEDMALLOCEXTSPEC void nedpsetvalue(nedpool *p, void *v) THROWSPEC;
/* Gets a previously set value using nedpsetvalue() or zero if memory is unknown.
Optionally can also retrieve pool. You can detect an unknown block by the return
being zero and *p being unmodifed.
*/
NEDMALLOCEXTSPEC void *nedgetvalue(nedpool **p, void *mem) THROWSPEC;
/* Trims the thread cache for the calling thread, returning any existing cache
data to the central pool. Remember to ALWAYS call with zero if you used the
system pool. Setting disable to non-zero replicates neddisablethreadcache().
*/
NEDMALLOCEXTSPEC void nedtrimthreadcache(nedpool *p, int disable) THROWSPEC;
/* Disables the thread cache for the calling thread, returning any existing cache
data to the central pool. Remember to ALWAYS call with zero if you used the
system pool.
*/
NEDMALLOCEXTSPEC void neddisablethreadcache(nedpool *p) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedpmalloc(nedpool *p, size_t size) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedpcalloc(nedpool *p, size_t no, size_t size) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedprealloc(nedpool *p, void *mem, size_t size) THROWSPEC;
NEDMALLOCEXTSPEC void nedpfree(nedpool *p, void *mem) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedpmemalign(nedpool *p, size_t alignment, size_t bytes) THROWSPEC;
NEDMALLOCEXTSPEC struct nedmallinfo nedpmallinfo(nedpool *p) THROWSPEC;
NEDMALLOCEXTSPEC int nedpmallopt(nedpool *p, int parno, int value) THROWSPEC;
NEDMALLOCEXTSPEC int nedpmalloc_trim(nedpool *p, size_t pad) THROWSPEC;
NEDMALLOCEXTSPEC void nedpmalloc_stats(nedpool *p) THROWSPEC;
NEDMALLOCEXTSPEC size_t nedpmalloc_footprint(nedpool *p) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void **nedpindependent_calloc(nedpool *p, size_t elemsno, size_t elemsize, void **chunks) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void **nedpindependent_comalloc(nedpool *p, size_t elems, size_t *sizes, void **chunks) THROWSPEC;
#if defined(__cplusplus)
}
#endif
#endif
#endif
#ifdef NEDMALLOC_ENABLED
/* nedalloc, an alternative malloc implementation for multiple threads without
lock contention based on dlmalloc v2.8.3. (C) 2005-2009 Niall Douglas
Boost Software License - Version 1.0 - August 17th, 2003
Permission is hereby granted, free of charge, to any person or organization
obtaining a copy of the software and accompanying documentation covered by
this license (the "Software") to use, reproduce, display, distribute,
execute, and transmit the Software, and to prepare derivative works of the
Software, and to permit third-parties to whom the Software is furnished to
do so, all subject to the following:
The copyright notices in the Software and this entire statement, including
the above license grant, this restriction and the following disclaimer,
must be included in all copies of the Software, in whole or in part, and
all derivative works of the Software, unless such copies or derivative
works are solely in the form of machine-executable object code generated by
a source language processor.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE, TITLE AND NON-INFRINGEMENT. IN NO EVENT
SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE
FOR ANY DAMAGES OR OTHER LIABILITY, WHETHER IN CONTRACT, TORT OR OTHERWISE,
ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
DEALINGS IN THE SOFTWARE.
*/
#ifndef NEDMALLOC_H
#define NEDMALLOC_H
#include "typedefs.h"
#define MALLOC_ALIGNMENT DEFAULT_ALIGNMENT
#ifdef PSP_ENABLED
#define USE_LOCKS 0
#define HAVE_MMAP 0
#endif
/* See malloc.c.h for what each function does.
REPLACE_SYSTEM_ALLOCATOR on POSIX causes nedalloc's functions to be called
malloc, free etc. instead of nedmalloc, nedfree etc. You may or may not want
this. On Windows it causes nedmalloc to patch all loaded DLLs and binaries
to replace usage of the system allocator.
NO_NED_NAMESPACE prevents the functions from being defined in the nedalloc
namespace when in C++ (uses the global namespace instead).
NEDMALLOCEXTSPEC can be defined to be __declspec(dllexport) or
__attribute__ ((visibility("default"))) or whatever you like. It defaults
to extern unless NEDMALLOC_DLL_EXPORTS is set as it would be when building
nedmalloc.dll.
USE_LOCKS can be 2 if you want to define your own MLOCK_T, INITIAL_LOCK,
ACQUIRE_LOCK, RELEASE_LOCK, TRY_LOCK, IS_LOCKED and NULL_LOCK_INITIALIZER.
NEDMALLOC_DEBUG can be defined to cause DEBUG to be set differently for nedmalloc
than for the rest of the build. Remember to set NDEBUG to disable all assertion
checking too.
USE_MAGIC_HEADERS causes nedalloc to allocate an extra three sizeof(size_t)
to each block. nedpfree() and nedprealloc() can then automagically know when
to free a system allocated block. Enabling this typically adds 20-50% to
application memory usage.
ENABLE_TOLERANT_NEDMALLOC is automatically turned on if REPLACE_SYSTEM_ALLOCATOR
is set or the Windows DLL is being built. This causes nedmalloc to detect when a
system allocator block is passed to it and to handle it appropriately. Note that
without USE_MAGIC_HEADERS there is a very tiny chance that nedmalloc will segfault
on non-Windows builds (it uses Win32 SEH to trap segfaults on Windows and there
is no comparable system on POSIX).
USE_ALLOCATOR can be one of these settings (it defaults to 1):
0: System allocator (nedmalloc now simply acts as a threadcache).
WARNING: Intended for DEBUG USE ONLY - not all functions work correctly.
1: dlmalloc
ENABLE_LARGE_PAGES enables support for requesting memory from the system in large
(typically >=2Mb) pages if the host OS supports this. These occupy just a single
TLB entry and can significantly improve performance in large working set applications.
ENABLE_FAST_HEAP_DETECTION enables special logic to detect blocks allocated
by the system heap. This avoids 1.5%-2% overhead when checking for non-nedmalloc
blocks, but it assumes that the NT and glibc heaps function in a very specific
fashion which may not hold true across OS upgrades.
*/
#include <stddef.h> /* for size_t */
#ifndef NEDMALLOCEXTSPEC
#ifdef NEDMALLOC_DLL_EXPORTS
#ifdef WIN32
#define NEDMALLOCEXTSPEC extern __declspec(dllexport)
#elif defined(__GNUC__)
#define NEDMALLOCEXTSPEC extern __attribute__ ((visibility("default")))
#endif
#ifndef ENABLE_TOLERANT_NEDMALLOC
#define ENABLE_TOLERANT_NEDMALLOC 1
#endif
#else
#define NEDMALLOCEXTSPEC extern
#endif
#endif
#if __STDC_VERSION__ >= 199901L /* C99 or better */
#define RESTRICT restrict
#else
#if defined(_MSC_VER) && _MSC_VER>=1400
#define RESTRICT __restrict
#endif
#ifdef __GNUC__
#define RESTRICT __restrict
#endif
#endif
#ifndef RESTRICT
#define RESTRICT
#endif
#if defined(_MSC_VER) && _MSC_VER>=1400
#define NEDMALLOCPTRATTR __declspec(restrict)
#define NEDMALLOCNOALIASATTR __declspec(noalias)
#endif
#ifdef __GNUC__
#define NEDMALLOCPTRATTR __attribute__ ((malloc))
#endif
#ifndef NEDMALLOCPTRATTR
#define NEDMALLOCPTRATTR
#endif
#ifndef NEDMALLOCNOALIASATTR
#define NEDMALLOCNOALIASATTR
#endif
#ifndef USE_MAGIC_HEADERS
#define USE_MAGIC_HEADERS 0
#endif
#ifndef USE_ALLOCATOR
#define USE_ALLOCATOR 1 /* dlmalloc */
#endif
#if !USE_ALLOCATOR && !USE_MAGIC_HEADERS
#error If you are using the system allocator then you MUST use magic headers
#endif
#ifdef REPLACE_SYSTEM_ALLOCATOR
#if USE_ALLOCATOR==0
#error Cannot combine using the system allocator with replacing the system allocator
#endif
#ifndef ENABLE_TOLERANT_NEDMALLOC
#define ENABLE_TOLERANT_NEDMALLOC 1
#endif
#ifndef WIN32 /* We have a dedicated patcher for Windows */
#define nedmalloc malloc
#define nedcalloc calloc
#define nedrealloc realloc
#define nedfree free
#define nedmemalign memalign
#define nedmallinfo mallinfo
#define nedmallopt mallopt
#define nedmalloc_trim malloc_trim
#define nedmalloc_stats malloc_stats
#define nedmalloc_footprint malloc_footprint
#define nedindependent_calloc independent_calloc
#define nedindependent_comalloc independent_comalloc
#ifdef _MSC_VER
#define nedblksize _msize
#endif
#endif
#endif
#if defined(__cplusplus)
extern "C" {
#endif
struct nedmallinfo {
size_t arena; /* non-mmapped space allocated from system */
size_t ordblks; /* number of free chunks */
size_t smblks; /* always 0 */
size_t hblks; /* always 0 */
size_t hblkhd; /* space in mmapped regions */
size_t usmblks; /* maximum total allocated space */
size_t fsmblks; /* always 0 */
size_t uordblks; /* total allocated space */
size_t fordblks; /* total free space */
size_t keepcost; /* releasable (via malloc_trim) space */
};
#if defined(__cplusplus)
}
#endif
#if defined(__cplusplus)
#if !defined(NO_NED_NAMESPACE)
namespace nedalloc {
#else
extern "C" {
#endif
#define THROWSPEC throw()
#else
#define THROWSPEC
#endif
/* These are the global functions */
/* Gets the usable size of an allocated block. Note this will always be bigger than what was
asked for due to rounding etc. Optionally returns 1 in isforeign if the block came from the
system allocator - note that there is a small (>0.01%) but real chance of segfault on non-Windows
systems when passing non-nedmalloc blocks if you don't use USE_MAGIC_HEADERS.
*/
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR size_t nedblksize(int *RESTRICT isforeign, void *RESTRICT mem) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR void nedsetvalue(void *v) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedmalloc(size_t size) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedcalloc(size_t no, size_t size) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedrealloc(void *mem, size_t size) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR void nedfree(void *mem) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void * nedmemalign(size_t alignment, size_t bytes) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR struct nedmallinfo nedmallinfo(void) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR int nedmallopt(int parno, int value) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR void* nedmalloc_internals(size_t *granularity, size_t *magic) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR int nedmalloc_trim(size_t pad) THROWSPEC;
NEDMALLOCEXTSPEC void nedmalloc_stats(void) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR size_t nedmalloc_footprint(void) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void **nedindependent_calloc(size_t elemsno, size_t elemsize, void **chunks) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCNOALIASATTR NEDMALLOCPTRATTR void **nedindependent_comalloc(size_t elems, size_t *sizes, void **chunks) THROWSPEC;
/* Destroys the system memory pool used by the functions above.
Useful for when you have nedmalloc in a DLL you're about to unload.
If you call ANY nedmalloc functions after calling this you will
get a fatal exception!
*/
NEDMALLOCEXTSPEC void neddestroysyspool() THROWSPEC;
/* These are the pool functions */
struct nedpool_t;
typedef struct nedpool_t nedpool;
/* Creates a memory pool for use with the nedp* functions below.
Capacity is how much to allocate immediately (if you know you'll be allocating a lot
of memory very soon) which you can leave at zero. Threads specifies how many threads
will *normally* be accessing the pool concurrently. Setting this to zero means it
extends on demand, but be careful of this as it can rapidly consume system resources
where bursts of concurrent threads use a pool at once.
*/
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR nedpool *nedcreatepool(size_t capacity, int threads) THROWSPEC;
/* Destroys a memory pool previously created by nedcreatepool().
*/
NEDMALLOCEXTSPEC void neddestroypool(nedpool *p) THROWSPEC;
/* Returns a zero terminated snapshot of threadpools existing at the time of call. Call
nedfree() on the returned list when you are done. Returns zero if there is only the
system pool in existence.
*/
NEDMALLOCEXTSPEC nedpool **nedpoollist() THROWSPEC;
/* Sets a value to be associated with a pool. You can retrieve this value by passing
any memory block allocated from that pool.
*/
NEDMALLOCEXTSPEC void nedpsetvalue(nedpool *p, void *v) THROWSPEC;
/* Gets a previously set value using nedpsetvalue() or zero if memory is unknown.
Optionally can also retrieve pool. You can detect an unknown block by the return
being zero and *p being unmodifed.
*/
NEDMALLOCEXTSPEC void *nedgetvalue(nedpool **p, void *mem) THROWSPEC;
/* Trims the thread cache for the calling thread, returning any existing cache
data to the central pool. Remember to ALWAYS call with zero if you used the
system pool. Setting disable to non-zero replicates neddisablethreadcache().
*/
NEDMALLOCEXTSPEC void nedtrimthreadcache(nedpool *p, int disable) THROWSPEC;
/* Disables the thread cache for the calling thread, returning any existing cache
data to the central pool. Remember to ALWAYS call with zero if you used the
system pool.
*/
NEDMALLOCEXTSPEC void neddisablethreadcache(nedpool *p) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedpmalloc(nedpool *p, size_t size) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedpcalloc(nedpool *p, size_t no, size_t size) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedprealloc(nedpool *p, void *mem, size_t size) THROWSPEC;
NEDMALLOCEXTSPEC void nedpfree(nedpool *p, void *mem) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void * nedpmemalign(nedpool *p, size_t alignment, size_t bytes) THROWSPEC;
NEDMALLOCEXTSPEC struct nedmallinfo nedpmallinfo(nedpool *p) THROWSPEC;
NEDMALLOCEXTSPEC int nedpmallopt(nedpool *p, int parno, int value) THROWSPEC;
NEDMALLOCEXTSPEC int nedpmalloc_trim(nedpool *p, size_t pad) THROWSPEC;
NEDMALLOCEXTSPEC void nedpmalloc_stats(nedpool *p) THROWSPEC;
NEDMALLOCEXTSPEC size_t nedpmalloc_footprint(nedpool *p) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void **nedpindependent_calloc(nedpool *p, size_t elemsno, size_t elemsize, void **chunks) THROWSPEC;
NEDMALLOCEXTSPEC NEDMALLOCPTRATTR void **nedpindependent_comalloc(nedpool *p, size_t elems, size_t *sizes, void **chunks) THROWSPEC;
#if defined(__cplusplus)
}
#endif
#endif
#endif

38
drivers/openssl/register_openssl.cpp
View file

@ -1,19 +1,19 @@
#include "register_openssl.h"
#include "stream_peer_openssl.h"
#ifdef OPENSSL_ENABLED
void register_openssl() {
ObjectTypeDB::register_type<StreamPeerOpenSSL>();
StreamPeerOpenSSL::initialize_ssl();
}
void unregister_openssl() {
StreamPeerOpenSSL::finalize_ssl();
}
#endif
#include "register_openssl.h"
#include "stream_peer_openssl.h"
#ifdef OPENSSL_ENABLED
void register_openssl() {
ObjectTypeDB::register_type<StreamPeerOpenSSL>();
StreamPeerOpenSSL::initialize_ssl();
}
void unregister_openssl() {
StreamPeerOpenSSL::finalize_ssl();
}
#endif

22
drivers/openssl/register_openssl.h
View file

@ -1,11 +1,11 @@
#ifndef REGISTER_OPENSSL_H
#define REGISTER_OPENSSL_H
#ifdef OPENSSL_ENABLED
void register_openssl();
void unregister_openssl();
#endif
#endif // REGISTER_OPENSSL_H
#ifndef REGISTER_OPENSSL_H
#define REGISTER_OPENSSL_H
#ifdef OPENSSL_ENABLED
void register_openssl();
void unregister_openssl();
#endif
#endif // REGISTER_OPENSSL_H

200
drivers/opus/SCsub Normal file
View file

@ -0,0 +1,200 @@
Import('env')
opus_sources = [
"opus/audio_stream_opus.cpp",
]
opus_sources_silk=[]
opus_sources_lib = [
"opus/celt/bands.c",
"opus/celt/celt_lpc.c",
"opus/celt/entenc.c",
"opus/celt/mdct.c",
"opus/celt/quant_bands.c",
"opus/celt/celt.c",
"opus/celt/cwrs.c",
"opus/celt/kiss_fft.c",
"opus/celt/modes.c",
"opus/celt/rate.c",
"opus/celt/celt_decoder.c",
"opus/celt/entcode.c",
"opus/celt/laplace.c",
#opus/celt/opus_custom_demo.c",
"opus/celt/vq.c",
"opus/celt/celt_encoder.c",
"opus/celt/entdec.c",
"opus/celt/mathops.c",
"opus/celt/pitch.c",
"opus/silk/A2NLSF.c",
"opus/silk/decoder_set_fs.c",
"opus/silk/NLSF_stabilize.c",
"opus/silk/sigm_Q15.c",
"opus/silk/ana_filt_bank_1.c",
"opus/silk/enc_API.c",
"opus/silk/NLSF_unpack.c",
"opus/silk/sort.c",
"opus/silk/biquad_alt.c",
"opus/silk/encode_indices.c",
"opus/silk/NLSF_VQ.c",
"opus/silk/stereo_decode_pred.c",
"opus/silk/bwexpander_32.c",
"opus/silk/encode_pulses.c",
"opus/silk/NLSF_VQ_weights_laroia.c",
"opus/silk/stereo_encode_pred.c",
"opus/silk/bwexpander.c",
"opus/silk/gain_quant.c",
"opus/silk/NSQ.c",
"opus/silk/stereo_find_predictor.c",
"opus/silk/check_control_input.c",
"opus/silk/HP_variable_cutoff.c",
"opus/silk/NSQ_del_dec.c",
"opus/silk/stereo_LR_to_MS.c",
"opus/silk/CNG.c",
"opus/silk/init_decoder.c",
"opus/silk/pitch_est_tables.c",
"opus/silk/stereo_MS_to_LR.c",
"opus/silk/code_signs.c",
"opus/silk/init_encoder.c",
"opus/silk/PLC.c",
"opus/silk/stereo_quant_pred.c",
"opus/silk/control_audio_bandwidth.c",
"opus/silk/inner_prod_aligned.c",
"opus/silk/process_NLSFs.c",
"opus/silk/sum_sqr_shift.c",
"opus/silk/control_codec.c",
"opus/silk/interpolate.c",
"opus/silk/quant_LTP_gains.c",
"opus/silk/table_LSF_cos.c",
"opus/silk/control_SNR.c",
"opus/silk/lin2log.c",
"opus/silk/resampler.c",
"opus/silk/tables_gain.c",
"opus/silk/debug.c",
"opus/silk/log2lin.c",
"opus/silk/resampler_down2_3.c",
"opus/silk/tables_LTP.c",
"opus/silk/dec_API.c",
"opus/silk/LPC_analysis_filter.c",
"opus/silk/resampler_down2.c",
"opus/silk/tables_NLSF_CB_NB_MB.c",
"opus/silk/decode_core.c",
"opus/silk/LPC_inv_pred_gain.c",
"opus/silk/resampler_private_AR2.c",
"opus/silk/tables_NLSF_CB_WB.c",
"opus/silk/decode_frame.c",
"opus/silk/LP_variable_cutoff.c",
"opus/silk/resampler_private_down_FIR.c",
"opus/silk/tables_other.c",
"opus/silk/decode_indices.c",
"opus/silk/NLSF2A.c",
"opus/silk/resampler_private_IIR_FIR.c",
"opus/silk/tables_pitch_lag.c",
"opus/silk/decode_parameters.c",
"opus/silk/NLSF_decode.c",
"opus/silk/resampler_private_up2_HQ.c",
"opus/silk/tables_pulses_per_block.c",
"opus/silk/decode_pitch.c",
"opus/silk/NLSF_del_dec_quant.c",
"opus/silk/resampler_rom.c",
"opus/silk/VAD.c",
"opus/silk/decode_pulses.c",
"opus/silk/NLSF_encode.c",
"opus/silk/shell_coder.c",
"opus/silk/VQ_WMat_EC.c",
"opus/analysis.c",
"opus/internal.c",
"opus/opus.c",
#"opus/opus_demo.c",
"opus/opus_multistream.c",
"opus/repacketizer.c",
"opus/wincerts.c",
"opus/http.c",
"opus/mlp.c",
#"opus/opus_compare.c",
"opus/opus_encoder.c",
"opus/opus_multistream_decoder.c",
#"opus/repacketizer_demo.c",
"opus/info.c",
"opus/mlp_data.c",
"opus/opus_decoder.c",
"opus/opusfile.c",
"opus/opus_multistream_encoder.c",
"opus/stream.c"
]
if("opus_fixed_point" in env and env.opus_fixed_point=="yes"):
env.Append(CPPPATH=["#drivers/opus/silk/fixed"], CFLAGS=["-DOPUS_FIXED_POINT"])
opus_sources_silk = [
"opus/silk/fixed/apply_sine_window_FIX.c",
"opus/silk/fixed/k2a_FIX.c",
"opus/silk/fixed/residual_energy16_FIX.c",
"opus/silk/fixed/autocorr_FIX.c",
"opus/silk/fixed/k2a_Q16_FIX.c",
"opus/silk/fixed/residual_energy_FIX.c",
"opus/silk/fixed/burg_modified_FIX.c",
"opus/silk/fixed/LTP_analysis_filter_FIX.c",
"opus/silk/fixed/schur64_FIX.c",
"opus/silk/fixed/corrMatrix_FIX.c",
"opus/silk/fixed/LTP_scale_ctrl_FIX.c",
"opus/silk/fixed/schur_FIX.c",
"opus/silk/fixed/encode_frame_FIX.c",
"opus/silk/fixed/noise_shape_analysis_FIX.c",
"opus/silk/fixed/solve_LS_FIX.c",
"opus/silk/fixed/find_LPC_FIX.c",
"opus/silk/fixed/pitch_analysis_core_FIX.c",
"opus/silk/fixed/vector_ops_FIX.c",
"opus/silk/fixed/find_LTP_FIX.c",
"opus/silk/fixed/prefilter_FIX.c",
"opus/silk/fixed/warped_autocorrelation_FIX.c",
"opus/silk/fixed/find_pitch_lags_FIX.c",
"opus/silk/fixed/process_gains_FIX.c",
"opus/silk/fixed/find_pred_coefs_FIX.c",
"opus/silk/fixed/regularize_correlations_FIX.c"
]
else:
env.Append(CPPPATH=["#drivers/opus/silk/float"])
opus_sources_silk = [
"opus/silk/float/apply_sine_window_FLP.c",
"opus/silk/float/inner_product_FLP.c",
"opus/silk/float/regularize_correlations_FLP.c",
"opus/silk/float/autocorrelation_FLP.c",
"opus/silk/float/k2a_FLP.c",
"opus/silk/float/residual_energy_FLP.c",
"opus/silk/float/burg_modified_FLP.c",
"opus/silk/float/levinsondurbin_FLP.c",
"opus/silk/float/scale_copy_vector_FLP.c",
"opus/silk/float/bwexpander_FLP.c",
"opus/silk/float/LPC_analysis_filter_FLP.c",
"opus/silk/float/scale_vector_FLP.c",
"opus/silk/float/corrMatrix_FLP.c",
"opus/silk/float/LPC_inv_pred_gain_FLP.c",
"opus/silk/float/schur_FLP.c",
"opus/silk/float/encode_frame_FLP.c",
"opus/silk/float/LTP_analysis_filter_FLP.c",
"opus/silk/float/solve_LS_FLP.c",
"opus/silk/float/energy_FLP.c",
"opus/silk/float/LTP_scale_ctrl_FLP.c",
"opus/silk/float/sort_FLP.c",
"opus/silk/float/find_LPC_FLP.c",
"opus/silk/float/noise_shape_analysis_FLP.c",
"opus/silk/float/warped_autocorrelation_FLP.c",
"opus/silk/float/find_LTP_FLP.c",
"opus/silk/float/pitch_analysis_core_FLP.c",
"opus/silk/float/wrappers_FLP.c",
"opus/silk/float/find_pitch_lags_FLP.c",
"opus/silk/float/prefilter_FLP.c",
"opus/silk/float/find_pred_coefs_FLP.c",
"opus/silk/float/process_gains_FLP.c"
]
opus_sources_lib+=opus_sources_silk
env.drivers_sources+=opus_sources_lib
env.drivers_sources+=opus_sources
env.Append(CPPPATH=["#drivers/opus"])
env.Append(CPPPATH=["#drivers/opus/celt","#drivers/opus/silk","#drivers/opus/silk/float"])
env.Append(CFLAGS=["-DOPUS_HAVE_CONFIG_H"])
Export('env')

645
drivers/opus/analysis.c Normal file
View file

@ -0,0 +1,645 @@
/* Copyright (c) 2011 Xiph.Org Foundation
Written by Jean-Marc Valin */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR
CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#include "kiss_fft.h"
#include "celt.h"
#include "opus_modes.h"
#include "arch.h"
#include "quant_bands.h"
#include <stdio.h>
#include "analysis.h"
#include "mlp.h"
#include "stack_alloc.h"
extern const MLP net;
#ifndef M_PI
#define M_PI 3.141592653
#endif
static const float dct_table[128] = {
0.250000f, 0.250000f, 0.250000f, 0.250000f, 0.250000f, 0.250000f, 0.250000f, 0.250000f,
0.250000f, 0.250000f, 0.250000f, 0.250000f, 0.250000f, 0.250000f, 0.250000f, 0.250000f,
0.351851f, 0.338330f, 0.311806f, 0.273300f, 0.224292f, 0.166664f, 0.102631f, 0.034654f,
-0.034654f,-0.102631f,-0.166664f,-0.224292f,-0.273300f,-0.311806f,-0.338330f,-0.351851f,
0.346760f, 0.293969f, 0.196424f, 0.068975f,-0.068975f,-0.196424f,-0.293969f,-0.346760f,
-0.346760f,-0.293969f,-0.196424f,-0.068975f, 0.068975f, 0.196424f, 0.293969f, 0.346760f,
0.338330f, 0.224292f, 0.034654f,-0.166664f,-0.311806f,-0.351851f,-0.273300f,-0.102631f,
0.102631f, 0.273300f, 0.351851f, 0.311806f, 0.166664f,-0.034654f,-0.224292f,-0.338330f,
0.326641f, 0.135299f,-0.135299f,-0.326641f,-0.326641f,-0.135299f, 0.135299f, 0.326641f,
0.326641f, 0.135299f,-0.135299f,-0.326641f,-0.326641f,-0.135299f, 0.135299f, 0.326641f,
0.311806f, 0.034654f,-0.273300f,-0.338330f,-0.102631f, 0.224292f, 0.351851f, 0.166664f,
-0.166664f,-0.351851f,-0.224292f, 0.102631f, 0.338330f, 0.273300f,-0.034654f,-0.311806f,
0.293969f,-0.068975f,-0.346760f,-0.196424f, 0.196424f, 0.346760f, 0.068975f,-0.293969f,
-0.293969f, 0.068975f, 0.346760f, 0.196424f,-0.196424f,-0.346760f,-0.068975f, 0.293969f,
0.273300f,-0.166664f,-0.338330f, 0.034654f, 0.351851f, 0.102631f,-0.311806f,-0.224292f,
0.224292f, 0.311806f,-0.102631f,-0.351851f,-0.034654f, 0.338330f, 0.166664f,-0.273300f,
};
static const float analysis_window[240] = {
0.000043f, 0.000171f, 0.000385f, 0.000685f, 0.001071f, 0.001541f, 0.002098f, 0.002739f,
0.003466f, 0.004278f, 0.005174f, 0.006156f, 0.007222f, 0.008373f, 0.009607f, 0.010926f,
0.012329f, 0.013815f, 0.015385f, 0.017037f, 0.018772f, 0.020590f, 0.022490f, 0.024472f,
0.026535f, 0.028679f, 0.030904f, 0.033210f, 0.035595f, 0.038060f, 0.040604f, 0.043227f,
0.045928f, 0.048707f, 0.051564f, 0.054497f, 0.057506f, 0.060591f, 0.063752f, 0.066987f,
0.070297f, 0.073680f, 0.077136f, 0.080665f, 0.084265f, 0.087937f, 0.091679f, 0.095492f,
0.099373f, 0.103323f, 0.107342f, 0.111427f, 0.115579f, 0.119797f, 0.124080f, 0.128428f,
0.132839f, 0.137313f, 0.141849f, 0.146447f, 0.151105f, 0.155823f, 0.160600f, 0.165435f,
0.170327f, 0.175276f, 0.180280f, 0.185340f, 0.190453f, 0.195619f, 0.200838f, 0.206107f,
0.211427f, 0.216797f, 0.222215f, 0.227680f, 0.233193f, 0.238751f, 0.244353f, 0.250000f,
0.255689f, 0.261421f, 0.267193f, 0.273005f, 0.278856f, 0.284744f, 0.290670f, 0.296632f,
0.302628f, 0.308658f, 0.314721f, 0.320816f, 0.326941f, 0.333097f, 0.339280f, 0.345492f,
0.351729f, 0.357992f, 0.364280f, 0.370590f, 0.376923f, 0.383277f, 0.389651f, 0.396044f,
0.402455f, 0.408882f, 0.415325f, 0.421783f, 0.428254f, 0.434737f, 0.441231f, 0.447736f,
0.454249f, 0.460770f, 0.467298f, 0.473832f, 0.480370f, 0.486912f, 0.493455f, 0.500000f,
0.506545f, 0.513088f, 0.519630f, 0.526168f, 0.532702f, 0.539230f, 0.545751f, 0.552264f,
0.558769f, 0.565263f, 0.571746f, 0.578217f, 0.584675f, 0.591118f, 0.597545f, 0.603956f,
0.610349f, 0.616723f, 0.623077f, 0.629410f, 0.635720f, 0.642008f, 0.648271f, 0.654508f,
0.660720f, 0.666903f, 0.673059f, 0.679184f, 0.685279f, 0.691342f, 0.697372f, 0.703368f,
0.709330f, 0.715256f, 0.721144f, 0.726995f, 0.732807f, 0.738579f, 0.744311f, 0.750000f,
0.755647f, 0.761249f, 0.766807f, 0.772320f, 0.777785f, 0.783203f, 0.788573f, 0.793893f,
0.799162f, 0.804381f, 0.809547f, 0.814660f, 0.819720f, 0.824724f, 0.829673f, 0.834565f,
0.839400f, 0.844177f, 0.848895f, 0.853553f, 0.858151f, 0.862687f, 0.867161f, 0.871572f,
0.875920f, 0.880203f, 0.884421f, 0.888573f, 0.892658f, 0.896677f, 0.900627f, 0.904508f,
0.908321f, 0.912063f, 0.915735f, 0.919335f, 0.922864f, 0.926320f, 0.929703f, 0.933013f,
0.936248f, 0.939409f, 0.942494f, 0.945503f, 0.948436f, 0.951293f, 0.954072f, 0.956773f,
0.959396f, 0.961940f, 0.964405f, 0.966790f, 0.969096f, 0.971321f, 0.973465f, 0.975528f,
0.977510f, 0.979410f, 0.981228f, 0.982963f, 0.984615f, 0.986185f, 0.987671f, 0.989074f,
0.990393f, 0.991627f, 0.992778f, 0.993844f, 0.994826f, 0.995722f, 0.996534f, 0.997261f,
0.997902f, 0.998459f, 0.998929f, 0.999315f, 0.999615f, 0.999829f, 0.999957f, 1.000000f,
};
static const int tbands[NB_TBANDS+1] = {
2, 4, 6, 8, 10, 12, 14, 16, 20, 24, 28, 32, 40, 48, 56, 68, 80, 96, 120
};
static const int extra_bands[NB_TOT_BANDS+1] = {
1, 2, 4, 6, 8, 10, 12, 14, 16, 20, 24, 28, 32, 40, 48, 56, 68, 80, 96, 120, 160, 200
};
/*static const float tweight[NB_TBANDS+1] = {
.3, .4, .5, .6, .7, .8, .9, 1., 1., 1., 1., 1., 1., 1., .8, .7, .6, .5
};*/
#define NB_TONAL_SKIP_BANDS 9
#define cA 0.43157974f
#define cB 0.67848403f
#define cC 0.08595542f
#define cE ((float)M_PI/2)
static OPUS_INLINE float fast_atan2f(float y, float x) {
float x2, y2;
/* Should avoid underflow on the values we'll get */
if (ABS16(x)+ABS16(y)<1e-9f)
{
x*=1e12f;
y*=1e12f;
}
x2 = x*x;
y2 = y*y;
if(x2<y2){
float den = (y2 + cB*x2) * (y2 + cC*x2);
if (den!=0)
return -x*y*(y2 + cA*x2) / den + (y<0 ? -cE : cE);
else
return (y<0 ? -cE : cE);
}else{
float den = (x2 + cB*y2) * (x2 + cC*y2);
if (den!=0)
return x*y*(x2 + cA*y2) / den + (y<0 ? -cE : cE) - (x*y<0 ? -cE : cE);
else
return (y<0 ? -cE : cE) - (x*y<0 ? -cE : cE);
}
}
void tonality_get_info(TonalityAnalysisState *tonal, AnalysisInfo *info_out, int len)
{
int pos;
int curr_lookahead;
float psum;
int i;
pos = tonal->read_pos;
curr_lookahead = tonal->write_pos-tonal->read_pos;
if (curr_lookahead<0)
curr_lookahead += DETECT_SIZE;
if (len > 480 && pos != tonal->write_pos)
{
pos++;
if (pos==DETECT_SIZE)
pos=0;
}
if (pos == tonal->write_pos)
pos--;
if (pos<0)
pos = DETECT_SIZE-1;
OPUS_COPY(info_out, &tonal->info[pos], 1);
tonal->read_subframe += len/120;
while (tonal->read_subframe>=4)
{
tonal->read_subframe -= 4;
tonal->read_pos++;
}
if (tonal->read_pos>=DETECT_SIZE)
tonal->read_pos-=DETECT_SIZE;
/* Compensate for the delay in the features themselves.
FIXME: Need a better estimate the 10 I just made up */
curr_lookahead = IMAX(curr_lookahead-10, 0);
psum=0;
/* Summing the probability of transition patterns that involve music at
time (DETECT_SIZE-curr_lookahead-1) */
for (i=0;i<DETECT_SIZE-curr_lookahead;i++)
psum += tonal->pmusic[i];
for (;i<DETECT_SIZE;i++)
psum += tonal->pspeech[i];
psum = psum*tonal->music_confidence + (1-psum)*tonal->speech_confidence;
/*printf("%f %f %f\n", psum, info_out->music_prob, info_out->tonality);*/
info_out->music_prob = psum;
}
void tonality_analysis(TonalityAnalysisState *tonal, AnalysisInfo *info_out, const CELTMode *celt_mode, const void *x, int len, int offset, int c1, int c2, int C, int lsb_depth, downmix_func downmix)
{
int i, b;
const kiss_fft_state *kfft;
VARDECL(kiss_fft_cpx, in);
VARDECL(kiss_fft_cpx, out);
int N = 480, N2=240;
float * OPUS_RESTRICT A = tonal->angle;
float * OPUS_RESTRICT dA = tonal->d_angle;
float * OPUS_RESTRICT d2A = tonal->d2_angle;
VARDECL(float, tonality);
VARDECL(float, noisiness);
float band_tonality[NB_TBANDS];
float logE[NB_TBANDS];
float BFCC[8];
float features[25];
float frame_tonality;
float max_frame_tonality;
/*float tw_sum=0;*/
float frame_noisiness;
const float pi4 = (float)(M_PI*M_PI*M_PI*M_PI);
float slope=0;
float frame_stationarity;
float relativeE;
float frame_probs[2];
float alpha, alphaE, alphaE2;
float frame_loudness;
float bandwidth_mask;
int bandwidth=0;
float maxE = 0;
float noise_floor;
int remaining;
AnalysisInfo *info;
SAVE_STACK;
tonal->last_transition++;
alpha = 1.f/IMIN(20, 1+tonal->count);
alphaE = 1.f/IMIN(50, 1+tonal->count);
alphaE2 = 1.f/IMIN(1000, 1+tonal->count);
if (tonal->count<4)
tonal->music_prob = .5;
kfft = celt_mode->mdct.kfft[0];
if (tonal->count==0)
tonal->mem_fill = 240;
downmix(x, &tonal->inmem[tonal->mem_fill], IMIN(len, ANALYSIS_BUF_SIZE-tonal->mem_fill), offset, c1, c2, C);
if (tonal->mem_fill+len < ANALYSIS_BUF_SIZE)
{
tonal->mem_fill += len;
/* Don't have enough to update the analysis */
RESTORE_STACK;
return;
}
info = &tonal->info[tonal->write_pos++];
if (tonal->write_pos>=DETECT_SIZE)
tonal->write_pos-=DETECT_SIZE;
ALLOC(in, 480, kiss_fft_cpx);
ALLOC(out, 480, kiss_fft_cpx);
ALLOC(tonality, 240, float);
ALLOC(noisiness, 240, float);
for (i=0;i<N2;i++)
{
float w = analysis_window[i];
in[i].r = (kiss_fft_scalar)(w*tonal->inmem[i]);
in[i].i = (kiss_fft_scalar)(w*tonal->inmem[N2+i]);
in[N-i-1].r = (kiss_fft_scalar)(w*tonal->inmem[N-i-1]);
in[N-i-1].i = (kiss_fft_scalar)(w*tonal->inmem[N+N2-i-1]);
}
OPUS_MOVE(tonal->inmem, tonal->inmem+ANALYSIS_BUF_SIZE-240, 240);
remaining = len - (ANALYSIS_BUF_SIZE-tonal->mem_fill);
downmix(x, &tonal->inmem[240], remaining, offset+ANALYSIS_BUF_SIZE-tonal->mem_fill, c1, c2, C);
tonal->mem_fill = 240 + remaining;
opus_fft(kfft, in, out);
for (i=1;i<N2;i++)
{
float X1r, X2r, X1i, X2i;
float angle, d_angle, d2_angle;
float angle2, d_angle2, d2_angle2;
float mod1, mod2, avg_mod;
X1r = (float)out[i].r+out[N-i].r;
X1i = (float)out[i].i-out[N-i].i;
X2r = (float)out[i].i+out[N-i].i;
X2i = (float)out[N-i].r-out[i].r;
angle = (float)(.5f/M_PI)*fast_atan2f(X1i, X1r);
d_angle = angle - A[i];
d2_angle = d_angle - dA[i];
angle2 = (float)(.5f/M_PI)*fast_atan2f(X2i, X2r);
d_angle2 = angle2 - angle;
d2_angle2 = d_angle2 - d_angle;
mod1 = d2_angle - (float)floor(.5+d2_angle);
noisiness[i] = ABS16(mod1);
mod1 *= mod1;
mod1 *= mod1;
mod2 = d2_angle2 - (float)floor(.5+d2_angle2);
noisiness[i] += ABS16(mod2);
mod2 *= mod2;
mod2 *= mod2;
avg_mod = .25f*(d2A[i]+2.f*mod1+mod2);
tonality[i] = 1.f/(1.f+40.f*16.f*pi4*avg_mod)-.015f;
A[i] = angle2;
dA[i] = d_angle2;
d2A[i] = mod2;
}
frame_tonality = 0;
max_frame_tonality = 0;
/*tw_sum = 0;*/
info->activity = 0;
frame_noisiness = 0;
frame_stationarity = 0;
if (!tonal->count)
{
for (b=0;b<NB_TBANDS;b++)
{
tonal->lowE[b] = 1e10;
tonal->highE[b] = -1e10;
}
}
relativeE = 0;
frame_loudness = 0;
for (b=0;b<NB_TBANDS;b++)
{
float E=0, tE=0, nE=0;
float L1, L2;
float stationarity;
for (i=tbands[b];i<tbands[b+1];i++)
{
float binE = out[i].r*(float)out[i].r + out[N-i].r*(float)out[N-i].r
+ out[i].i*(float)out[i].i + out[N-i].i*(float)out[N-i].i;
#ifdef OPUS_FIXED_POINT
/* FIXME: It's probably best to change the BFCC filter initial state instead */
binE *= 5.55e-17f;
#endif
E += binE;
tE += binE*tonality[i];
nE += binE*2.f*(.5f-noisiness[i]);
}
tonal->E[tonal->E_count][b] = E;
frame_noisiness += nE/(1e-15f+E);
frame_loudness += (float)sqrt(E+1e-10f);
logE[b] = (float)log(E+1e-10f);
tonal->lowE[b] = MIN32(logE[b], tonal->lowE[b]+.01f);
tonal->highE[b] = MAX32(logE[b], tonal->highE[b]-.1f);
if (tonal->highE[b] < tonal->lowE[b]+1.f)
{
tonal->highE[b]+=.5f;
tonal->lowE[b]-=.5f;
}
relativeE += (logE[b]-tonal->lowE[b])/(1e-15f+tonal->highE[b]-tonal->lowE[b]);
L1=L2=0;
for (i=0;i<NB_FRAMES;i++)
{
L1 += (float)sqrt(tonal->E[i][b]);
L2 += tonal->E[i][b];
}
stationarity = MIN16(0.99f,L1/(float)sqrt(1e-15+NB_FRAMES*L2));
stationarity *= stationarity;
stationarity *= stationarity;
frame_stationarity += stationarity;
/*band_tonality[b] = tE/(1e-15+E)*/;
band_tonality[b] = MAX16(tE/(1e-15f+E), stationarity*tonal->prev_band_tonality[b]);
#if 0
if (b>=NB_TONAL_SKIP_BANDS)
{
frame_tonality += tweight[b]*band_tonality[b];
tw_sum += tweight[b];
}
#else
frame_tonality += band_tonality[b];
if (b>=NB_TBANDS-NB_TONAL_SKIP_BANDS)
frame_tonality -= band_tonality[b-NB_TBANDS+NB_TONAL_SKIP_BANDS];
#endif
max_frame_tonality = MAX16(max_frame_tonality, (1.f+.03f*(b-NB_TBANDS))*frame_tonality);
slope += band_tonality[b]*(b-8);
/*printf("%f %f ", band_tonality[b], stationarity);*/
tonal->prev_band_tonality[b] = band_tonality[b];
}
bandwidth_mask = 0;
bandwidth = 0;
maxE = 0;
noise_floor = 5.7e-4f/(1<<(IMAX(0,lsb_depth-8)));
#ifdef OPUS_FIXED_POINT
noise_floor *= 1<<(15+SIG_SHIFT);
#endif
noise_floor *= noise_floor;
for (b=0;b<NB_TOT_BANDS;b++)
{
float E=0;
int band_start, band_end;
/* Keep a margin of 300 Hz for aliasing */
band_start = extra_bands[b];
band_end = extra_bands[b+1];
for (i=band_start;i<band_end;i++)
{
float binE = out[i].r*(float)out[i].r + out[N-i].r*(float)out[N-i].r
+ out[i].i*(float)out[i].i + out[N-i].i*(float)out[N-i].i;
E += binE;
}
maxE = MAX32(maxE, E);
tonal->meanE[b] = MAX32((1-alphaE2)*tonal->meanE[b], E);
E = MAX32(E, tonal->meanE[b]);
/* Use a simple follower with 13 dB/Bark slope for spreading function */
bandwidth_mask = MAX32(.05f*bandwidth_mask, E);
/* Consider the band "active" only if all these conditions are met:
1) less than 10 dB below the simple follower
2) less than 90 dB below the peak band (maximal masking possible considering
both the ATH and the loudness-dependent slope of the spreading function)
3) above the PCM quantization noise floor
*/
if (E>.1*bandwidth_mask && E*1e9f > maxE && E > noise_floor*(band_end-band_start))
bandwidth = b;
}
if (tonal->count<=2)
bandwidth = 20;
frame_loudness = 20*(float)log10(frame_loudness);
tonal->Etracker = MAX32(tonal->Etracker-.03f, frame_loudness);
tonal->lowECount *= (1-alphaE);
if (frame_loudness < tonal->Etracker-30)
tonal->lowECount += alphaE;
for (i=0;i<8;i++)
{
float sum=0;
for (b=0;b<16;b++)
sum += dct_table[i*16+b]*logE[b];
BFCC[i] = sum;
}
frame_stationarity /= NB_TBANDS;
relativeE /= NB_TBANDS;
if (tonal->count<10)
relativeE = .5;
frame_noisiness /= NB_TBANDS;
#if 1
info->activity = frame_noisiness + (1-frame_noisiness)*relativeE;
#else
info->activity = .5*(1+frame_noisiness-frame_stationarity);
#endif
frame_tonality = (max_frame_tonality/(NB_TBANDS-NB_TONAL_SKIP_BANDS));
frame_tonality = MAX16(frame_tonality, tonal->prev_tonality*.8f);
tonal->prev_tonality = frame_tonality;
slope /= 8*8;
info->tonality_slope = slope;
tonal->E_count = (tonal->E_count+1)%NB_FRAMES;
tonal->count++;
info->tonality = frame_tonality;
for (i=0;i<4;i++)
features[i] = -0.12299f*(BFCC[i]+tonal->mem[i+24]) + 0.49195f*(tonal->mem[i]+tonal->mem[i+16]) + 0.69693f*tonal->mem[i+8] - 1.4349f*tonal->cmean[i];
for (i=0;i<4;i++)
tonal->cmean[i] = (1-alpha)*tonal->cmean[i] + alpha*BFCC[i];
for (i=0;i<4;i++)
features[4+i] = 0.63246f*(BFCC[i]-tonal->mem[i+24]) + 0.31623f*(tonal->mem[i]-tonal->mem[i+16]);
for (i=0;i<3;i++)
features[8+i] = 0.53452f*(BFCC[i]+tonal->mem[i+24]) - 0.26726f*(tonal->mem[i]+tonal->mem[i+16]) -0.53452f*tonal->mem[i+8];
if (tonal->count > 5)
{
for (i=0;i<9;i++)
tonal->std[i] = (1-alpha)*tonal->std[i] + alpha*features[i]*features[i];
}
for (i=0;i<8;i++)
{
tonal->mem[i+24] = tonal->mem[i+16];
tonal->mem[i+16] = tonal->mem[i+8];
tonal->mem[i+8] = tonal->mem[i];
tonal->mem[i] = BFCC[i];
}
for (i=0;i<9;i++)
features[11+i] = (float)sqrt(tonal->std[i]);
features[20] = info->tonality;
features[21] = info->activity;
features[22] = frame_stationarity;
features[23] = info->tonality_slope;
features[24] = tonal->lowECount;
#ifndef DISABLE_FLOAT_API
mlp_process(&net, features, frame_probs);
frame_probs[0] = .5f*(frame_probs[0]+1);
/* Curve fitting between the MLP probability and the actual probability */
frame_probs[0] = .01f + 1.21f*frame_probs[0]*frame_probs[0] - .23f*(float)pow(frame_probs[0], 10);
/* Probability of active audio (as opposed to silence) */
frame_probs[1] = .5f*frame_probs[1]+.5f;
/* Consider that silence has a 50-50 probability. */
frame_probs[0] = frame_probs[1]*frame_probs[0] + (1-frame_probs[1])*.5f;
/*printf("%f %f ", frame_probs[0], frame_probs[1]);*/
{
/* Probability of state transition */
float tau;
/* Represents independence of the MLP probabilities, where
beta=1 means fully independent. */
float beta;
/* Denormalized probability of speech (p0) and music (p1) after update */
float p0, p1;
/* Probabilities for "all speech" and "all music" */
float s0, m0;
/* Probability sum for renormalisation */
float psum;
/* Instantaneous probability of speech and music, with beta pre-applied. */
float speech0;
float music0;
/* One transition every 3 minutes of active audio */
tau = .00005f*frame_probs[1];
beta = .05f;
if (1) {
/* Adapt beta based on how "unexpected" the new prob is */
float p, q;
p = MAX16(.05f,MIN16(.95f,frame_probs[0]));
q = MAX16(.05f,MIN16(.95f,tonal->music_prob));
beta = .01f+.05f*ABS16(p-q)/(p*(1-q)+q*(1-p));
}
/* p0 and p1 are the probabilities of speech and music at this frame
using only information from previous frame and applying the
state transition model */
p0 = (1-tonal->music_prob)*(1-tau) + tonal->music_prob *tau;
p1 = tonal->music_prob *(1-tau) + (1-tonal->music_prob)*tau;
/* We apply the current probability with exponent beta to work around
the fact that the probability estimates aren't independent. */
p0 *= (float)pow(1-frame_probs[0], beta);
p1 *= (float)pow(frame_probs[0], beta);
/* Normalise the probabilities to get the Marokv probability of music. */
tonal->music_prob = p1/(p0+p1);
info->music_prob = tonal->music_prob;
/* This chunk of code deals with delayed decision. */
psum=1e-20f;
/* Instantaneous probability of speech and music, with beta pre-applied. */
speech0 = (float)pow(1-frame_probs[0], beta);
music0 = (float)pow(frame_probs[0], beta);
if (tonal->count==1)
{
tonal->pspeech[0]=.5;
tonal->pmusic [0]=.5;
}
/* Updated probability of having only speech (s0) or only music (m0),
before considering the new observation. */
s0 = tonal->pspeech[0] + tonal->pspeech[1];
m0 = tonal->pmusic [0] + tonal->pmusic [1];
/* Updates s0 and m0 with instantaneous probability. */
tonal->pspeech[0] = s0*(1-tau)*speech0;
tonal->pmusic [0] = m0*(1-tau)*music0;
/* Propagate the transition probabilities */
for (i=1;i<DETECT_SIZE-1;i++)
{
tonal->pspeech[i] = tonal->pspeech[i+1]*speech0;
tonal->pmusic [i] = tonal->pmusic [i+1]*music0;
}
/* Probability that the latest frame is speech, when all the previous ones were music. */
tonal->pspeech[DETECT_SIZE-1] = m0*tau*speech0;
/* Probability that the latest frame is music, when all the previous ones were speech. */
tonal->pmusic [DETECT_SIZE-1] = s0*tau*music0;
/* Renormalise probabilities to 1 */
for (i=0;i<DETECT_SIZE;i++)
psum += tonal->pspeech[i] + tonal->pmusic[i];
psum = 1.f/psum;
for (i=0;i<DETECT_SIZE;i++)
{
tonal->pspeech[i] *= psum;
tonal->pmusic [i] *= psum;
}
psum = tonal->pmusic[0];
for (i=1;i<DETECT_SIZE;i++)
psum += tonal->pspeech[i];
/* Estimate our confidence in the speech/music decisions */
if (frame_probs[1]>.75)
{
if (tonal->music_prob>.9)
{
float adapt;
adapt = 1.f/(++tonal->music_confidence_count);
tonal->music_confidence_count = IMIN(tonal->music_confidence_count, 500);
tonal->music_confidence += adapt*MAX16(-.2f,frame_probs[0]-tonal->music_confidence);
}
if (tonal->music_prob<.1)
{
float adapt;
adapt = 1.f/(++tonal->speech_confidence_count);
tonal->speech_confidence_count = IMIN(tonal->speech_confidence_count, 500);
tonal->speech_confidence += adapt*MIN16(.2f,frame_probs[0]-tonal->speech_confidence);
}
} else {
if (tonal->music_confidence_count==0)
tonal->music_confidence = .9f;
if (tonal->speech_confidence_count==0)
tonal->speech_confidence = .1f;
}
}
if (tonal->last_music != (tonal->music_prob>.5f))
tonal->last_transition=0;
tonal->last_music = tonal->music_prob>.5f;
#else
info->music_prob = 0;
#endif
/*for (i=0;i<25;i++)
printf("%f ", features[i]);
printf("\n");*/
info->bandwidth = bandwidth;
/*printf("%d %d\n", info->bandwidth, info->opus_bandwidth);*/
info->noisiness = frame_noisiness;
info->valid = 1;
if (info_out!=NULL)
OPUS_COPY(info_out, info, 1);
RESTORE_STACK;
}
void run_analysis(TonalityAnalysisState *analysis, const CELTMode *celt_mode, const void *analysis_pcm,
int analysis_frame_size, int frame_size, int c1, int c2, int C, opus_int32 Fs,
int lsb_depth, downmix_func downmix, AnalysisInfo *analysis_info)
{
int offset;
int pcm_len;
if (analysis_pcm != NULL)
{
/* Avoid overflow/wrap-around of the analysis buffer */
analysis_frame_size = IMIN((DETECT_SIZE-5)*Fs/100, analysis_frame_size);
pcm_len = analysis_frame_size - analysis->analysis_offset;
offset = analysis->analysis_offset;
do {
tonality_analysis(analysis, NULL, celt_mode, analysis_pcm, IMIN(480, pcm_len), offset, c1, c2, C, lsb_depth, downmix);
offset += 480;
pcm_len -= 480;
} while (pcm_len>0);
analysis->analysis_offset = analysis_frame_size;
analysis->analysis_offset -= frame_size;
}
analysis_info->valid = 0;
tonality_get_info(analysis, analysis_info, frame_size);
}

90
drivers/opus/analysis.h Normal file
View file

@ -0,0 +1,90 @@
/* Copyright (c) 2011 Xiph.Org Foundation
Written by Jean-Marc Valin */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR
CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef ANALYSIS_H
#define ANALYSIS_H
#include "celt.h"
#include "opus_private.h"
#define NB_FRAMES 8
#define NB_TBANDS 18
#define NB_TOT_BANDS 21
#define ANALYSIS_BUF_SIZE 720 /* 15 ms at 48 kHz */
#define DETECT_SIZE 200
typedef struct {
float angle[240];
float d_angle[240];
float d2_angle[240];
opus_val32 inmem[ANALYSIS_BUF_SIZE];
int mem_fill; /* number of usable samples in the buffer */
float prev_band_tonality[NB_TBANDS];
float prev_tonality;
float E[NB_FRAMES][NB_TBANDS];
float lowE[NB_TBANDS];
float highE[NB_TBANDS];
float meanE[NB_TOT_BANDS];
float mem[32];
float cmean[8];
float std[9];
float music_prob;
float Etracker;
float lowECount;
int E_count;
int last_music;
int last_transition;
int count;
float subframe_mem[3];
int analysis_offset;
/** Probability of having speech for time i to DETECT_SIZE-1 (and music before).
pspeech[0] is the probability that all frames in the window are speech. */
float pspeech[DETECT_SIZE];
/** Probability of having music for time i to DETECT_SIZE-1 (and speech before).
pmusic[0] is the probability that all frames in the window are music. */
float pmusic[DETECT_SIZE];
float speech_confidence;
float music_confidence;
int speech_confidence_count;
int music_confidence_count;
int write_pos;
int read_pos;
int read_subframe;
AnalysisInfo info[DETECT_SIZE];
} TonalityAnalysisState;
void tonality_analysis(TonalityAnalysisState *tonal, AnalysisInfo *info,
const CELTMode *celt_mode, const void *x, int len, int offset, int c1, int c2, int C, int lsb_depth, downmix_func downmix);
void tonality_get_info(TonalityAnalysisState *tonal, AnalysisInfo *info_out, int len);
void run_analysis(TonalityAnalysisState *analysis, const CELTMode *celt_mode, const void *analysis_pcm,
int analysis_frame_size, int frame_size, int c1, int c2, int C, opus_int32 Fs,
int lsb_depth, downmix_func downmix, AnalysisInfo *analysis_info);
#endif

376
drivers/opus/audio_stream_opus.cpp Normal file
View file

@ -0,0 +1,376 @@
/*************************************************************************/
/* audio_stream_opus.cpp */
/*************************************************************************/
/* This file is part of: */
/* GODOT ENGINE */
/* http://www.godotengine.org */
/*************************************************************************/
/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur. */
/* */
/* Author: George Marques <george@gmarqu.es> */
/* */
/* Permission is hereby granted, free of charge, to any person obtaining */
/* a copy of this software and associated documentation files (the */
/* "Software"), to deal in the Software without restriction, including */
/* without limitation the rights to use, copy, modify, merge, publish, */
/* distribute, sublicense, and/or sell copies of the Software, and to */
/* permit persons to whom the Software is furnished to do so, subject to */
/* the following conditions: */
/* */
/* The above copyright notice and this permission notice shall be */
/* included in all copies or substantial portions of the Software. */
/* */
/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */
/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */
/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */
/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */
/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */
/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */
/*************************************************************************/
#include "audio_stream_opus.h"
const float AudioStreamPlaybackOpus::osrate=48000.0f;
int AudioStreamPlaybackOpus::_op_read_func(void *_stream, unsigned char *_ptr, int _nbytes) {
FileAccess *fa=(FileAccess*)_stream;
if(fa->eof_reached())
return 0;
uint8_t *dst = (uint8_t*)_ptr;
int read = fa->get_buffer(dst, _nbytes);
return read;
}
int AudioStreamPlaybackOpus::_op_seek_func(void *_stream, opus_int64 _offset, int _whence){
#ifdef SEEK_SET
FileAccess *fa=(FileAccess*)_stream;
switch (_whence) {
case SEEK_SET: {
fa->seek(_offset);
} break;
case SEEK_CUR: {
fa->seek(fa->get_pos()+_offset);
} break;
case SEEK_END: {
fa->seek_end(_offset);
} break;
default: {
ERR_PRINT("BUG, wtf was whence set to?\n");
}
}
int ret=fa->eof_reached()?-1:0;
return ret;
#else
return -1; // no seeking
#endif
}
int AudioStreamPlaybackOpus::_op_close_func(void *_stream) {
if (!_stream)
return 0;
FileAccess *fa=(FileAccess*)_stream;
if (fa->is_open())
fa->close();
return 0;
}
opus_int64 AudioStreamPlaybackOpus::_op_tell_func(void *_stream) {
FileAccess *_fa = (FileAccess*)_stream;
return (opus_int64)_fa->get_pos();
}
void AudioStreamPlaybackOpus::_clear_stream() {
if(!stream_loaded)
return;
op_free(opus_file);
_close_file();
stream_loaded=false;
stream_channels=1;
playing=false;
}
void AudioStreamPlaybackOpus::_close_file() {
if (f) {
memdelete(f);
f=NULL;
}
}
Error AudioStreamPlaybackOpus::_load_stream() {
ERR_FAIL_COND_V(!stream_valid,ERR_UNCONFIGURED);
_clear_stream();
if (file=="")
return ERR_INVALID_DATA;
Error err;
f=FileAccess::open(file,FileAccess::READ,&err);
if (err) {
ERR_FAIL_COND_V( err, err );
}
int _err = 0;
opus_file = op_open_callbacks(f,&_op_callbacks,NULL,0,&_err);
switch (_err) {
case OP_EREAD: { // - Can't read the file.
memdelete(f); f=NULL;
ERR_FAIL_V( ERR_FILE_CANT_READ );
} break;
case OP_EVERSION: // - Unrecognized version number.
case OP_ENOTFORMAT: // - Stream is not Opus data.
case OP_EIMPL : { // - Stream used non-implemented feature.
memdelete(f); f=NULL;
ERR_FAIL_V( ERR_FILE_UNRECOGNIZED );
} break;
case OP_EBADLINK: // - Failed to find old data after seeking.
case OP_EBADTIMESTAMP: // - Timestamp failed the validity checks.
case OP_EBADHEADER: { // - Invalid or mising Opus bitstream header.
memdelete(f); f=NULL;
ERR_FAIL_V( ERR_FILE_CORRUPT );
} break;
case OP_EFAULT: { // - Internal logic fault; indicates a bug or heap/stack corruption.
memdelete(f); f=NULL;
ERR_FAIL_V( ERR_BUG );
} break;
}
repeats=0;
stream_loaded=true;
return OK;
}
AudioStreamPlaybackOpus::AudioStreamPlaybackOpus() {
loops=false;
playing=false;
f = NULL;
stream_loaded=false;
stream_valid=false;
repeats=0;
paused=true;
stream_channels=0;
current_section=0;
length=0;
loop_restart_time=0;
pre_skip=0;
_op_callbacks.read = _op_read_func;
_op_callbacks.seek = _op_seek_func;
_op_callbacks.tell = _op_tell_func;
_op_callbacks.close = _op_close_func;
}
Error AudioStreamPlaybackOpus::set_file(const String &p_file) {
file=p_file;
stream_valid=false;
Error err;
f=FileAccess::open(file,FileAccess::READ,&err);
if (err) {
ERR_FAIL_COND_V( err, err );
}
int _err;
opus_file = op_open_callbacks(f,&_op_callbacks,NULL,0,&_err);
switch (_err) {
case OP_EREAD: { // - Can't read the file.
memdelete(f); f=NULL;
ERR_FAIL_V( ERR_FILE_CANT_READ );
} break;
case OP_EVERSION: // - Unrecognized version number.
case OP_ENOTFORMAT: // - Stream is not Opus data.
case OP_EIMPL : { // - Stream used non-implemented feature.
memdelete(f); f=NULL;
ERR_FAIL_V( ERR_FILE_UNRECOGNIZED );
} break;
case OP_EBADLINK: // - Failed to find old data after seeking.
case OP_EBADTIMESTAMP: // - Timestamp failed the validity checks.
case OP_EBADHEADER: { // - Invalid or mising Opus bitstream header.
memdelete(f); f=NULL;
ERR_FAIL_V( ERR_FILE_CORRUPT );
} break;
case OP_EFAULT: { // - Internal logic fault; indicates a bug or heap/stack corruption.
memdelete(f); f=NULL;
ERR_FAIL_V( ERR_BUG );
} break;
}
const OpusHead *oinfo = op_head(opus_file,-1);
stream_channels=oinfo->channel_count;
pre_skip=oinfo->pre_skip;
frames_mixed=pre_skip;
ogg_int64_t len = op_pcm_total(opus_file,-1);
if(len < 0) {
length = 0;
} else {
length=(len/osrate);
}
op_free(opus_file);
memdelete(f);
f=NULL;
stream_valid=true;
return OK;
}
void AudioStreamPlaybackOpus::play(float p_from) {
if (playing)
stop();
if (_load_stream()!=OK)
return;
frames_mixed=pre_skip;
playing=true;
if (p_from>0) {
seek_pos(p_from);
}
}
void AudioStreamPlaybackOpus::stop() {
_clear_stream();
playing=false;
}
void AudioStreamPlaybackOpus::seek_pos(float p_time) {
if(!playing) return;
ogg_int64_t pcm_offset = (ogg_int64_t)(p_time * osrate);
bool ok = op_pcm_seek(opus_file,pcm_offset)==0;
if(!ok) {
ERR_PRINT("Seek time over stream size.");
return;
}
frames_mixed=osrate*p_time;
}
int AudioStreamPlaybackOpus::mix(int16_t* p_bufer,int p_frames) {
if (!playing)
return 0;
int total=p_frames;
while (true) {
int todo = p_frames;
if (todo==0 || todo<MIN_MIX) {
break;
}
int ret=op_read(opus_file,(opus_int16*)p_bufer,todo*stream_channels,&current_section);
if (ret<0) {
playing = false;
ERR_EXPLAIN("Error reading Opus File: "+file);
ERR_BREAK(ret<0);
} else if (ret==0) { // end of song, reload?
op_free(opus_file);
_close_file();
f=FileAccess::open(file,FileAccess::READ);
int errv = 0;
opus_file = op_open_callbacks(f,&_op_callbacks,NULL,0,&errv);
if (errv!=0) {
playing=false;
break; // :(
}
if (!has_loop()) {
playing=false;
repeats=1;
break;
}
if (loop_restart_time) {
bool ok = op_pcm_seek(opus_file, (loop_restart_time*osrate)+pre_skip)==0;
if (!ok) {
playing=false;
ERR_PRINT("loop restart time rejected")
}
frames_mixed=(loop_restart_time*osrate)+pre_skip;
} else {
frames_mixed=pre_skip;
}
repeats++;
continue;
}
stream_channels=op_head(opus_file,current_section)->channel_count;
frames_mixed+=ret;
p_bufer+=ret*stream_channels;
p_frames-=ret;
}
return total-p_frames;
}
float AudioStreamPlaybackOpus::get_length() const {
if(!stream_loaded) {
if(const_cast<AudioStreamPlaybackOpus*>(this)->_load_stream() != OK)
return 0;
}
return length;
}
float AudioStreamPlaybackOpus::get_pos() const {
int32_t frames = int32_t(frames_mixed);
if (frames < 0)
frames=0;
return double(frames) / osrate;
}
int AudioStreamPlaybackOpus::get_minimum_buffer_size() const {
return MIN_MIX;
}
AudioStreamPlaybackOpus::~AudioStreamPlaybackOpus() {
_clear_stream();
}
RES ResourceFormatLoaderAudioStreamOpus::load(const String &p_path, const String& p_original_path, Error *r_error) {
if (r_error)
*r_error=OK;
AudioStreamOpus *opus_stream = memnew(AudioStreamOpus);
opus_stream->set_file(p_path);
return Ref<AudioStreamOpus>(opus_stream);
}
void ResourceFormatLoaderAudioStreamOpus::get_recognized_extensions(List<String> *p_extensions) const {
p_extensions->push_back("opus");
}
String ResourceFormatLoaderAudioStreamOpus::get_resource_type(const String &p_path) const {
if (p_path.extension().to_lower()=="opus")
return "AudioStreamOpus";
return "";
}
bool ResourceFormatLoaderAudioStreamOpus::handles_type(const String& p_type) const {
return (p_type=="AudioStream" || p_type=="AudioStreamOpus");
}

141
drivers/opus/audio_stream_opus.h Normal file
View file

@ -0,0 +1,141 @@
/*************************************************************************/
/* audio_stream_opus.h */
/*************************************************************************/
/* This file is part of: */
/* GODOT ENGINE */
/* http://www.godotengine.org */
/*************************************************************************/
/* Copyright (c) 2007-2015 Juan Linietsky, Ariel Manzur. */
/* */
/* Author: George Marques <george@gmarqu.es> */
/* */
/* Permission is hereby granted, free of charge, to any person obtaining */
/* a copy of this software and associated documentation files (the */
/* "Software"), to deal in the Software without restriction, including */
/* without limitation the rights to use, copy, modify, merge, publish, */
/* distribute, sublicense, and/or sell copies of the Software, and to */
/* permit persons to whom the Software is furnished to do so, subject to */
/* the following conditions: */
/* */
/* The above copyright notice and this permission notice shall be */
/* included in all copies or substantial portions of the Software. */
/* */
/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */
/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */
/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */
/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */
/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */
/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */
/*************************************************************************/
#ifndef AUDIO_STREAM_OPUS_H
#define AUDIO_STREAM_OPUS_H
#include "scene/resources/audio_stream.h"
#include "opus/opusfile.h"
#include "opus/internal.h"
#include "os/file_access.h"
#include "io/resource_loader.h"
class AudioStreamPlaybackOpus : public AudioStreamPlayback {
OBJ_TYPE(AudioStreamPlaybackOpus,AudioStreamPlayback)
enum {
MIN_MIX=1024
};
FileAccess *f;
OpusFileCallbacks _op_callbacks;
float length;
static int _op_read_func(void *_stream, unsigned char *_ptr, int _nbytes);
static int _op_seek_func(void *_stream, opus_int64 _offset, int _whence);
static int _op_close_func(void *_stream);
static opus_int64 _op_tell_func(void *_stream);
static const float osrate;
String file;
int64_t frames_mixed;
bool stream_loaded;
volatile bool playing;
OggOpusFile *opus_file;
int stream_channels;
int current_section;
int pre_skip;
bool paused;
bool loops;
int repeats;
Error _load_stream();
void _clear_stream();
void _close_file();
bool stream_valid;
float loop_restart_time;
public:
Error set_file(const String& p_file);
virtual void play(float p_from=0);
virtual void stop();
virtual bool is_playing() const { return playing; }
virtual void set_loop_restart_time(float p_time) { loop_restart_time=p_time; }
virtual void set_paused(bool p_paused) { paused=p_paused; }
virtual bool is_paused() const { return paused; }
virtual void set_loop(bool p_enable) { loops=p_enable; }
virtual bool has_loop() const {return loops; }
virtual float get_length() const;
virtual String get_stream_name() const { return ""; }
virtual int get_loop_count() const { return repeats; }
virtual float get_pos() const;
virtual void seek_pos(float p_time);
virtual int get_channels() const { return stream_channels; }
virtual int get_mix_rate() const { return osrate; }
virtual int get_minimum_buffer_size() const;
virtual int mix(int16_t* p_bufer,int p_frames);
AudioStreamPlaybackOpus();
~AudioStreamPlaybackOpus();
};
class AudioStreamOpus: public AudioStream {
OBJ_TYPE(AudioStreamOpus,AudioStream)
String file;
public:
Ref<AudioStreamPlayback> instance_playback() {
Ref<AudioStreamPlaybackOpus> pb = memnew( AudioStreamPlaybackOpus );
pb->set_file(file);
return pb;
}
void set_file(const String& p_file) { file=p_file; }
};
class ResourceFormatLoaderAudioStreamOpus: public ResourceFormatLoader {
public:
virtual RES load(const String &p_path,const String& p_original_path="",Error *r_error=NULL);
virtual void get_recognized_extensions(List<String> *p_extensions) const;
virtual bool handles_type(const String& p_type) const;
virtual String get_resource_type(const String &p_path) const;
};
#endif // AUDIO_STREAM_OPUS_H

183
drivers/opus/celt/_kiss_fft_guts.h Normal file
View file

@ -0,0 +1,183 @@
/*Copyright (c) 2003-2004, Mark Borgerding
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice,
this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.*/
#ifndef KISS_FFT_GUTS_H
#define KISS_FFT_GUTS_H
#define MIN(a,b) ((a)<(b) ? (a):(b))
#define MAX(a,b) ((a)>(b) ? (a):(b))
/* kiss_fft.h
defines kiss_fft_scalar as either short or a float type
and defines
typedef struct { kiss_fft_scalar r; kiss_fft_scalar i; }kiss_fft_cpx; */
#include "kiss_fft.h"
/*
Explanation of macros dealing with complex math:
C_MUL(m,a,b) : m = a*b
C_FIXDIV( c , div ) : if a fixed point impl., c /= div. noop otherwise
C_SUB( res, a,b) : res = a - b
C_SUBFROM( res , a) : res -= a
C_ADDTO( res , a) : res += a
* */
#ifdef OPUS_FIXED_POINT
#include "arch.h"
#define SAMP_MAX 2147483647
#define TWID_MAX 32767
#define TRIG_UPSCALE 1
#define SAMP_MIN -SAMP_MAX
# define S_MUL(a,b) MULT16_32_Q15(b, a)
# define C_MUL(m,a,b) \
do{ (m).r = SUB32(S_MUL((a).r,(b).r) , S_MUL((a).i,(b).i)); \
(m).i = ADD32(S_MUL((a).r,(b).i) , S_MUL((a).i,(b).r)); }while(0)
# define C_MULC(m,a,b) \
do{ (m).r = ADD32(S_MUL((a).r,(b).r) , S_MUL((a).i,(b).i)); \
(m).i = SUB32(S_MUL((a).i,(b).r) , S_MUL((a).r,(b).i)); }while(0)
# define C_MUL4(m,a,b) \
do{ (m).r = SHR32(SUB32(S_MUL((a).r,(b).r) , S_MUL((a).i,(b).i)),2); \
(m).i = SHR32(ADD32(S_MUL((a).r,(b).i) , S_MUL((a).i,(b).r)),2); }while(0)
# define C_MULBYSCALAR( c, s ) \
do{ (c).r = S_MUL( (c).r , s ) ;\
(c).i = S_MUL( (c).i , s ) ; }while(0)
# define DIVSCALAR(x,k) \
(x) = S_MUL( x, (TWID_MAX-((k)>>1))/(k)+1 )
# define C_FIXDIV(c,div) \
do { DIVSCALAR( (c).r , div); \
DIVSCALAR( (c).i , div); }while (0)
#define C_ADD( res, a,b)\
do {(res).r=ADD32((a).r,(b).r); (res).i=ADD32((a).i,(b).i); \
}while(0)
#define C_SUB( res, a,b)\
do {(res).r=SUB32((a).r,(b).r); (res).i=SUB32((a).i,(b).i); \
}while(0)
#define C_ADDTO( res , a)\
do {(res).r = ADD32((res).r, (a).r); (res).i = ADD32((res).i,(a).i);\
}while(0)
#define C_SUBFROM( res , a)\
do {(res).r = ADD32((res).r,(a).r); (res).i = SUB32((res).i,(a).i); \
}while(0)
#if defined(OPUS_ARM_INLINE_ASM)
#include "arm/kiss_fft_armv4.h"
#endif
#if defined(OPUS_ARM_INLINE_EDSP)
#include "arm/kiss_fft_armv5e.h"
#endif
#else /* not OPUS_FIXED_POINT*/
# define S_MUL(a,b) ( (a)*(b) )
#define C_MUL(m,a,b) \
do{ (m).r = (a).r*(b).r - (a).i*(b).i;\
(m).i = (a).r*(b).i + (a).i*(b).r; }while(0)
#define C_MULC(m,a,b) \
do{ (m).r = (a).r*(b).r + (a).i*(b).i;\
(m).i = (a).i*(b).r - (a).r*(b).i; }while(0)
#define C_MUL4(m,a,b) C_MUL(m,a,b)
# define C_FIXDIV(c,div) /* NOOP */
# define C_MULBYSCALAR( c, s ) \
do{ (c).r *= (s);\
(c).i *= (s); }while(0)
#endif
#ifndef CHECK_OVERFLOW_OP
# define CHECK_OVERFLOW_OP(a,op,b) /* noop */
#endif
#ifndef C_ADD
#define C_ADD( res, a,b)\
do { \
CHECK_OVERFLOW_OP((a).r,+,(b).r)\
CHECK_OVERFLOW_OP((a).i,+,(b).i)\
(res).r=(a).r+(b).r; (res).i=(a).i+(b).i; \
}while(0)
#define C_SUB( res, a,b)\
do { \
CHECK_OVERFLOW_OP((a).r,-,(b).r)\
CHECK_OVERFLOW_OP((a).i,-,(b).i)\
(res).r=(a).r-(b).r; (res).i=(a).i-(b).i; \
}while(0)
#define C_ADDTO( res , a)\
do { \
CHECK_OVERFLOW_OP((res).r,+,(a).r)\
CHECK_OVERFLOW_OP((res).i,+,(a).i)\
(res).r += (a).r; (res).i += (a).i;\
}while(0)
#define C_SUBFROM( res , a)\
do {\
CHECK_OVERFLOW_OP((res).r,-,(a).r)\
CHECK_OVERFLOW_OP((res).i,-,(a).i)\
(res).r -= (a).r; (res).i -= (a).i; \
}while(0)
#endif /* C_ADD defined */
#ifdef OPUS_FIXED_POINT
# define KISS_FFT_COS(phase) TRIG_UPSCALE*floor(MIN(32767,MAX(-32767,.5+32768 * cos (phase))))
# define KISS_FFT_SIN(phase) TRIG_UPSCALE*floor(MIN(32767,MAX(-32767,.5+32768 * sin (phase))))
# define KISS_FFT_COS(phase) floor(.5+TWID_MAX*cos (phase))
# define KISS_FFT_SIN(phase) floor(.5+TWID_MAX*sin (phase))
# define HALF_OF(x) ((x)>>1)
#elif defined(USE_SIMD)
# define KISS_FFT_COS(phase) _mm_set1_ps( cos(phase) )
# define KISS_FFT_SIN(phase) _mm_set1_ps( sin(phase) )
# define HALF_OF(x) ((x)*_mm_set1_ps(.5f))
#else
# define KISS_FFT_COS(phase) (kiss_fft_scalar) cos(phase)
# define KISS_FFT_SIN(phase) (kiss_fft_scalar) sin(phase)
# define HALF_OF(x) ((x)*.5f)
#endif
#define kf_cexp(x,phase) \
do{ \
(x)->r = KISS_FFT_COS(phase);\
(x)->i = KISS_FFT_SIN(phase);\
}while(0)
#define kf_cexp2(x,phase) \
do{ \
(x)->r = TRIG_UPSCALE*celt_cos_norm((phase));\
(x)->i = TRIG_UPSCALE*celt_cos_norm((phase)-32768);\
}while(0)
#endif /* KISS_FFT_GUTS_H */

214
drivers/opus/celt/arch.h Normal file
View file

@ -0,0 +1,214 @@
/* Copyright (c) 2003-2008 Jean-Marc Valin
Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2009 Xiph.Org Foundation
Written by Jean-Marc Valin */
/**
@file arch.h
@brief Various architecture definitions for CELT
*/
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef ARCH_H
#define ARCH_H
#include "opus_types.h"
#include "opus_defines.h"
# if !defined(__GNUC_PREREQ)
# if defined(__GNUC__)&&defined(__GNUC_MINOR__)
# define __GNUC_PREREQ(_maj,_min) \
((__GNUC__<<16)+__GNUC_MINOR__>=((_maj)<<16)+(_min))
# else
# define __GNUC_PREREQ(_maj,_min) 0
# endif
# endif
#define CELT_SIG_SCALE 32768.f
#define celt_fatal(str) _celt_fatal(str, __FILE__, __LINE__);
#ifdef ENABLE_ASSERTIONS
#include <stdio.h>
#include <stdlib.h>
#ifdef __GNUC__
__attribute__((noreturn))
#endif
static OPUS_INLINE void _celt_fatal(const char *str, const char *file, int line)
{
fprintf (stderr, "Fatal (internal) error in %s, line %d: %s\n", file, line, str);
abort();
}
#define celt_assert(cond) {if (!(cond)) {celt_fatal("assertion failed: " #cond);}}
#define celt_assert2(cond, message) {if (!(cond)) {celt_fatal("assertion failed: " #cond "\n" message);}}
#else
#define celt_assert(cond)
#define celt_assert2(cond, message)
#endif
#define IMUL32(a,b) ((a)*(b))
#define ABS(x) ((x) < 0 ? (-(x)) : (x)) /**< Absolute integer value. */
#define ABS16(x) ((x) < 0 ? (-(x)) : (x)) /**< Absolute 16-bit value. */
#define MIN16(a,b) ((a) < (b) ? (a) : (b)) /**< Minimum 16-bit value. */
#define MAX16(a,b) ((a) > (b) ? (a) : (b)) /**< Maximum 16-bit value. */
#define ABS32(x) ((x) < 0 ? (-(x)) : (x)) /**< Absolute 32-bit value. */
#define MIN32(a,b) ((a) < (b) ? (a) : (b)) /**< Minimum 32-bit value. */
#define MAX32(a,b) ((a) > (b) ? (a) : (b)) /**< Maximum 32-bit value. */
#define IMIN(a,b) ((a) < (b) ? (a) : (b)) /**< Minimum int value. */
#define IMAX(a,b) ((a) > (b) ? (a) : (b)) /**< Maximum int value. */
#define UADD32(a,b) ((a)+(b))
#define USUB32(a,b) ((a)-(b))
#define PRINT_MIPS(file)
#ifdef OPUS_FIXED_POINT
typedef opus_int16 opus_val16;
typedef opus_int32 opus_val32;
typedef opus_val32 celt_sig;
typedef opus_val16 celt_norm;
typedef opus_val32 celt_ener;
#define Q15ONE 32767
#define SIG_SHIFT 12
#define NORM_SCALING 16384
#define DB_SHIFT 10
#define EPSILON 1
#define VERY_SMALL 0
#define VERY_LARGE16 ((opus_val16)32767)
#define Q15_ONE ((opus_val16)32767)
#define SCALEIN(a) (a)
#define SCALEOUT(a) (a)
#ifdef FIXED_DEBUG
#include "fixed_debug.h"
#else
#include "fixed_generic.h"
#ifdef OPUS_ARM_INLINE_EDSP
#include "arm/fixed_armv5e.h"
#elif defined (OPUS_ARM_INLINE_ASM)
#include "arm/fixed_armv4.h"
#elif defined (BFIN_ASM)
#include "fixed_bfin.h"
#elif defined (TI_C5X_ASM)
#include "fixed_c5x.h"
#elif defined (TI_C6X_ASM)
#include "fixed_c6x.h"
#endif
#endif
#else /* OPUS_FIXED_POINT */
typedef float opus_val16;
typedef float opus_val32;
typedef float celt_sig;
typedef float celt_norm;
typedef float celt_ener;
#define Q15ONE 1.0f
#define NORM_SCALING 1.f
#define EPSILON 1e-15f
#define VERY_SMALL 1e-30f
#define VERY_LARGE16 1e15f
#define Q15_ONE ((opus_val16)1.f)
#define QCONST16(x,bits) (x)
#define QCONST32(x,bits) (x)
#define NEG16(x) (-(x))
#define NEG32(x) (-(x))
#define EXTRACT16(x) (x)
#define EXTEND32(x) (x)
#define SHR16(a,shift) (a)
#define SHL16(a,shift) (a)
#define SHR32(a,shift) (a)
#define SHL32(a,shift) (a)
#define PSHR32(a,shift) (a)
#define VSHR32(a,shift) (a)
#define PSHR(a,shift) (a)
#define SHR(a,shift) (a)
#define SHL(a,shift) (a)
#define SATURATE(x,a) (x)
#define SATURATE16(x) (x)
#define ROUND16(a,shift) (a)
#define HALF16(x) (.5f*(x))
#define HALF32(x) (.5f*(x))
#define ADD16(a,b) ((a)+(b))
#define SUB16(a,b) ((a)-(b))
#define ADD32(a,b) ((a)+(b))
#define SUB32(a,b) ((a)-(b))
#define MULT16_16_16(a,b) ((a)*(b))
#define MULT16_16(a,b) ((opus_val32)(a)*(opus_val32)(b))
#define MAC16_16(c,a,b) ((c)+(opus_val32)(a)*(opus_val32)(b))
#define MULT16_32_Q15(a,b) ((a)*(b))
#define MULT16_32_Q16(a,b) ((a)*(b))
#define MULT32_32_Q31(a,b) ((a)*(b))
#define MAC16_32_Q15(c,a,b) ((c)+(a)*(b))
#define MULT16_16_Q11_32(a,b) ((a)*(b))
#define MULT16_16_Q11(a,b) ((a)*(b))
#define MULT16_16_Q13(a,b) ((a)*(b))
#define MULT16_16_Q14(a,b) ((a)*(b))
#define MULT16_16_Q15(a,b) ((a)*(b))
#define MULT16_16_P15(a,b) ((a)*(b))
#define MULT16_16_P13(a,b) ((a)*(b))
#define MULT16_16_P14(a,b) ((a)*(b))
#define MULT16_32_P16(a,b) ((a)*(b))
#define DIV32_16(a,b) (((opus_val32)(a))/(opus_val16)(b))
#define DIV32(a,b) (((opus_val32)(a))/(opus_val32)(b))
#define SCALEIN(a) ((a)*CELT_SIG_SCALE)
#define SCALEOUT(a) ((a)*(1/CELT_SIG_SCALE))
#endif /* !OPUS_FIXED_POINT */
#ifndef GLOBAL_STACK_SIZE
#ifdef OPUS_FIXED_POINT
#define GLOBAL_STACK_SIZE 100000
#else
#define GLOBAL_STACK_SIZE 100000
#endif
#endif
#endif /* ARCH_H */

316
drivers/opus/celt/arm/arm2gnu.pl Executable file
View file

@ -0,0 +1,316 @@
#!/usr/bin/perl
my $bigend; # little/big endian
my $nxstack;
$nxstack = 0;
eval 'exec /usr/local/bin/perl -S $0 ${1+"$@"}'
if $running_under_some_shell;
while ($ARGV[0] =~ /^-/) {
$_ = shift;
last if /^--/;
if (/^-n/) {
$nflag++;
next;
}
die "I don't recognize this switch: $_\\n";
}
$printit++ unless $nflag;
$\ = "\n"; # automatically add newline on print
$n=0;
$thumb = 0; # ARM mode by default, not Thumb.
@proc_stack = ();
LINE:
while (<>) {
# For ADRLs we need to add a new line after the substituted one.
$addPadding = 0;
# First, we do not dare to touch *anything* inside double quotes, do we?
# Second, if you want a dollar character in the string,
# insert two of them -- that's how ARM C and assembler treat strings.
s/^([A-Za-z_]\w*)[ \t]+DCB[ \t]*\"/$1: .ascii \"/ && do { s/\$\$/\$/g; next };
s/\bDCB\b[ \t]*\"/.ascii \"/ && do { s/\$\$/\$/g; next };
s/^(\S+)\s+RN\s+(\S+)/$1 .req r$2/ && do { s/\$\$/\$/g; next };
# If there's nothing on a line but a comment, don't try to apply any further
# substitutions (this is a cheap hack to avoid mucking up the license header)
s/^([ \t]*);/$1@/ && do { s/\$\$/\$/g; next };
# If substituted -- leave immediately !
s/@/,:/;
s/;/@/;
while ( /@.*'/ ) {
s/(@.*)'/$1/g;
}
s/\{FALSE\}/0/g;
s/\{TRUE\}/1/g;
s/\{(\w\w\w\w+)\}/$1/g;
s/\bINCLUDE[ \t]*([^ \t\n]+)/.include \"$1\"/;
s/\bGET[ \t]*([^ \t\n]+)/.include \"${ my $x=$1; $x =~ s|\.s|-gnu.S|; \$x }\"/;
s/\bIMPORT\b/.extern/;
s/\bEXPORT\b/.global/;
s/^(\s+)\[/$1IF/;
s/^(\s+)\|/$1ELSE/;
s/^(\s+)\]/$1ENDIF/;
s/IF *:DEF:/ .ifdef/;
s/IF *:LNOT: *:DEF:/ .ifndef/;
s/ELSE/ .else/;
s/ENDIF/ .endif/;
if( /\bIF\b/ ) {
s/\bIF\b/ .if/;
s/=/==/;
}
if ( $n == 2) {
s/\$/\\/g;
}
if ($n == 1) {
s/\$//g;
s/label//g;
$n = 2;
}
if ( /MACRO/ ) {
s/MACRO *\n/.macro/;
$n=1;
}
if ( /\bMEND\b/ ) {
s/\bMEND\b/.endm/;
$n=0;
}
# ".rdata" doesn't work in 'as' version 2.13.2, as it is ".rodata" there.
#
if ( /\bAREA\b/ ) {
my $align;
$align = "2";
if ( /ALIGN=(\d+)/ ) {
$align = $1;
}
if ( /CODE/ ) {
$nxstack = 1;
}
s/^(.+)CODE(.+)READONLY(.*)/ .text/;
s/^(.+)DATA(.+)READONLY(.*)/ .section .rdata/;
s/^(.+)\|\|\.data\|\|(.+)/ .data/;
s/^(.+)\|\|\.bss\|\|(.+)/ .bss/;
s/$/; .p2align $align/;
# Enable NEON instructions but don't produce a binary that requires
# ARMv7. RVCT does not have equivalent directives, so we just do this
# for all CODE areas.
if ( /.text/ ) {
# Separating .arch, .fpu, etc., by semicolons does not work (gas
# thinks the semicolon is part of the arch name, even when there's
# whitespace separating them). Sadly this means our line numbers
# won't match the original source file (we could use the .line
# directive, which is documented to be obsolete, but then gdb will
# show the wrong line in the translated source file).
s/$/; .arch armv7-a\n .fpu neon\n .object_arch armv4t/;
}
}
s/\|\|\.constdata\$(\d+)\|\|/.L_CONST$1/; # ||.constdata$3||
s/\|\|\.bss\$(\d+)\|\|/.L_BSS$1/; # ||.bss$2||
s/\|\|\.data\$(\d+)\|\|/.L_DATA$1/; # ||.data$2||
s/\|\|([a-zA-Z0-9_]+)\@([a-zA-Z0-9_]+)\|\|/@ $&/;
s/^(\s+)\%(\s)/ .space $1/;
s/\|(.+)\.(\d+)\|/\.$1_$2/; # |L80.123| -> .L80_123
s/\bCODE32\b/.code 32/ && do {$thumb = 0};
s/\bCODE16\b/.code 16/ && do {$thumb = 1};
if (/\bPROC\b/)
{
my $prefix;
my $proc;
/^([A-Za-z_\.]\w+)\b/;
$proc = $1;
$prefix = "";
if ($proc)
{
$prefix = $prefix.sprintf("\t.type\t%s, %%function; ",$proc);
push(@proc_stack, $proc);
s/^[A-Za-z_\.]\w+/$&:/;
}
$prefix = $prefix."\t.thumb_func; " if ($thumb);
s/\bPROC\b/@ $&/;
$_ = $prefix.$_;
}
s/^(\s*)(S|Q|SH|U|UQ|UH)ASX\b/$1$2ADDSUBX/;
s/^(\s*)(S|Q|SH|U|UQ|UH)SAX\b/$1$2SUBADDX/;
if (/\bENDP\b/)
{
my $proc;
s/\bENDP\b/@ $&/;
$proc = pop(@proc_stack);
$_ = "\t.size $proc, .-$proc".$_ if ($proc);
}
s/\bSUBT\b/@ $&/;
s/\bDATA\b/@ $&/; # DATA directive is deprecated -- Asm guide, p.7-25
s/\bKEEP\b/@ $&/;
s/\bEXPORTAS\b/@ $&/;
s/\|\|(.)+\bEQU\b/@ $&/;
s/\|\|([\w\$]+)\|\|/$1/;
s/\bENTRY\b/@ $&/;
s/\bASSERT\b/@ $&/;
s/\bGBLL\b/@ $&/;
s/\bGBLA\b/@ $&/;
s/^\W+OPT\b/@ $&/;
s/:OR:/|/g;
s/:SHL:/<</g;
s/:SHR:/>>/g;
s/:AND:/&/g;
s/:LAND:/&&/g;
s/CPSR/cpsr/;
s/SPSR/spsr/;
s/ALIGN$/.balign 4/;
s/ALIGN\s+([0-9x]+)$/.balign $1/;
s/psr_cxsf/psr_all/;
s/LTORG/.ltorg/;
s/^([A-Za-z_]\w*)[ \t]+EQU/ .set $1,/;
s/^([A-Za-z_]\w*)[ \t]+SETL/ .set $1,/;
s/^([A-Za-z_]\w*)[ \t]+SETA/ .set $1,/;
s/^([A-Za-z_]\w*)[ \t]+\*/ .set $1,/;
# {PC} + 0xdeadfeed --> . + 0xdeadfeed
s/\{PC\} \+/ \. +/;
# Single hex constant on the line !
#
# >>> NOTE <<<
# Double-precision floats in gcc are always mixed-endian, which means
# bytes in two words are little-endian, but words are big-endian.
# So, 0x0000deadfeed0000 would be stored as 0x0000dead at low address
# and 0xfeed0000 at high address.
#
s/\bDCFD\b[ \t]+0x([a-fA-F0-9]{8})([a-fA-F0-9]{8})/.long 0x$1, 0x$2/;
# Only decimal constants on the line, no hex !
s/\bDCFD\b[ \t]+([0-9\.\-]+)/.double $1/;
# Single hex constant on the line !
# s/\bDCFS\b[ \t]+0x([a-f0-9]{8})([a-f0-9]{8})/.long 0x$1, 0x$2/;
# Only decimal constants on the line, no hex !
# s/\bDCFS\b[ \t]+([0-9\.\-]+)/.double $1/;
s/\bDCFS[ \t]+0x/.word 0x/;
s/\bDCFS\b/.float/;
s/^([A-Za-z_]\w*)[ \t]+DCD/$1 .word/;
s/\bDCD\b/.word/;
s/^([A-Za-z_]\w*)[ \t]+DCW/$1 .short/;
s/\bDCW\b/.short/;
s/^([A-Za-z_]\w*)[ \t]+DCB/$1 .byte/;
s/\bDCB\b/.byte/;
s/^([A-Za-z_]\w*)[ \t]+\%/.comm $1,/;
s/^[A-Za-z_\.]\w+/$&:/;
s/^(\d+)/$1:/;
s/\%(\d+)/$1b_or_f/;
s/\%[Bb](\d+)/$1b/;
s/\%[Ff](\d+)/$1f/;
s/\%[Ff][Tt](\d+)/$1f/;
s/&([\dA-Fa-f]+)/0x$1/;
if ( /\b2_[01]+\b/ ) {
s/\b2_([01]+)\b/conv$1&&&&/g;
while ( /[01][01][01][01]&&&&/ ) {
s/0000&&&&/&&&&0/g;
s/0001&&&&/&&&&1/g;
s/0010&&&&/&&&&2/g;
s/0011&&&&/&&&&3/g;
s/0100&&&&/&&&&4/g;
s/0101&&&&/&&&&5/g;
s/0110&&&&/&&&&6/g;
s/0111&&&&/&&&&7/g;
s/1000&&&&/&&&&8/g;
s/1001&&&&/&&&&9/g;
s/1010&&&&/&&&&A/g;
s/1011&&&&/&&&&B/g;
s/1100&&&&/&&&&C/g;
s/1101&&&&/&&&&D/g;
s/1110&&&&/&&&&E/g;
s/1111&&&&/&&&&F/g;
}
s/000&&&&/&&&&0/g;
s/001&&&&/&&&&1/g;
s/010&&&&/&&&&2/g;
s/011&&&&/&&&&3/g;
s/100&&&&/&&&&4/g;
s/101&&&&/&&&&5/g;
s/110&&&&/&&&&6/g;
s/111&&&&/&&&&7/g;
s/00&&&&/&&&&0/g;
s/01&&&&/&&&&1/g;
s/10&&&&/&&&&2/g;
s/11&&&&/&&&&3/g;
s/0&&&&/&&&&0/g;
s/1&&&&/&&&&1/g;
s/conv&&&&/0x/g;
}
if ( /commandline/)
{
if( /-bigend/)
{
$bigend=1;
}
}
if ( /\bDCDU\b/ )
{
my $cmd=$_;
my $value;
my $prefix;
my $w1;
my $w2;
my $w3;
my $w4;
s/\s+DCDU\b/@ $&/;
$cmd =~ /\bDCDU\b\s+0x(\d+)/;
$value = $1;
$value =~ /(\w\w)(\w\w)(\w\w)(\w\w)/;
$w1 = $1;
$w2 = $2;
$w3 = $3;
$w4 = $4;
if( $bigend ne "")
{
# big endian
$prefix = "\t.byte\t0x".$w1.";".
"\t.byte\t0x".$w2.";".
"\t.byte\t0x".$w3.";".
"\t.byte\t0x".$w4."; ";
}
else
{
# little endian
$prefix = "\t.byte\t0x".$w4.";".
"\t.byte\t0x".$w3.";".
"\t.byte\t0x".$w2.";".
"\t.byte\t0x".$w1."; ";
}
$_=$prefix.$_;
}
if ( /\badrl\b/i )
{
s/\badrl\s+(\w+)\s*,\s*(\w+)/ldr $1,=$2/i;
$addPadding = 1;
}
s/\bEND\b/@ END/;
} continue {
printf ("%s", $_) if $printit;
if ($addPadding != 0)
{
printf (" mov r0,r0\n");
$addPadding = 0;
}
}
#If we had a code section, mark that this object doesn't need an executable
# stack.
if ($nxstack) {
printf (" .section\t.note.GNU-stack,\"\",\%\%progbits\n");
}

49
drivers/opus/celt/arm/arm_celt_map.c Normal file
View file

@ -0,0 +1,49 @@
/* Copyright (c) 2010 Xiph.Org Foundation
* Copyright (c) 2013 Parrot */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#include "pitch.h"
#if defined(OPUS_HAVE_RTCD)
# if defined(OPUS_FIXED_POINT)
opus_val32 (*const CELT_PITCH_XCORR_IMPL[OPUS_ARCHMASK+1])(const opus_val16 *,
const opus_val16 *, opus_val32 *, int , int) = {
celt_pitch_xcorr_c, /* ARMv4 */
MAY_HAVE_EDSP(celt_pitch_xcorr), /* EDSP */
MAY_HAVE_MEDIA(celt_pitch_xcorr), /* Media */
MAY_HAVE_NEON(celt_pitch_xcorr) /* NEON */
};
# else
# error "Floating-point implementation is not supported by ARM asm yet." \
"Reconfigure with --disable-rtcd or send patches."
# endif
#endif

174
drivers/opus/celt/arm/armcpu.c Normal file
View file

@ -0,0 +1,174 @@
/* Copyright (c) 2010 Xiph.Org Foundation
* Copyright (c) 2013 Parrot */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/* Original code from libtheora modified to suit to Opus */
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#ifdef OPUS_HAVE_RTCD
#include "armcpu.h"
#include "cpu_support.h"
#include "os_support.h"
#include "opus_types.h"
#define OPUS_CPU_ARM_V4 (1)
#define OPUS_CPU_ARM_EDSP (1<<1)
#define OPUS_CPU_ARM_MEDIA (1<<2)
#define OPUS_CPU_ARM_NEON (1<<3)
#if defined(_MSC_VER)
/*For GetExceptionCode() and EXCEPTION_ILLEGAL_INSTRUCTION.*/
# define WIN32_LEAN_AND_MEAN
# define WIN32_EXTRA_LEAN
# include <windows.h>
static OPUS_INLINE opus_uint32 opus_cpu_capabilities(void){
opus_uint32 flags;
flags=0;
/* MSVC has no OPUS_INLINE __asm support for ARM, but it does let you __emit
* instructions via their assembled hex code.
* All of these instructions should be essentially nops. */
# if defined(OPUS_ARM_MAY_HAVE_EDSP)
__try{
/*PLD [r13]*/
__emit(0xF5DDF000);
flags|=OPUS_CPU_ARM_EDSP;
}
__except(GetExceptionCode()==EXCEPTION_ILLEGAL_INSTRUCTION){
/*Ignore exception.*/
}
# if defined(OPUS_ARM_MAY_HAVE_MEDIA)
__try{
/*SHADD8 r3,r3,r3*/
__emit(0xE6333F93);
flags|=OPUS_CPU_ARM_MEDIA;
}
__except(GetExceptionCode()==EXCEPTION_ILLEGAL_INSTRUCTION){
/*Ignore exception.*/
}
# if defined(OPUS_ARM_MAY_HAVE_NEON)
__try{
/*VORR q0,q0,q0*/
__emit(0xF2200150);
flags|=OPUS_CPU_ARM_NEON;
}
__except(GetExceptionCode()==EXCEPTION_ILLEGAL_INSTRUCTION){
/*Ignore exception.*/
}
# endif
# endif
# endif
return flags;
}
#elif defined(__linux__)
/* Linux based */
opus_uint32 opus_cpu_capabilities(void)
{
opus_uint32 flags = 0;
FILE *cpuinfo;
/* Reading /proc/self/auxv would be easier, but that doesn't work reliably on
* Android */
cpuinfo = fopen("/proc/cpuinfo", "r");
if(cpuinfo != NULL)
{
/* 512 should be enough for anybody (it's even enough for all the flags that
* x86 has accumulated... so far). */
char buf[512];
while(fgets(buf, 512, cpuinfo) != NULL)
{
# if defined(OPUS_ARM_MAY_HAVE_EDSP) || defined(OPUS_ARM_MAY_HAVE_NEON)
/* Search for edsp and neon flag */
if(memcmp(buf, "Features", 8) == 0)
{
char *p;
# if defined(OPUS_ARM_MAY_HAVE_EDSP)
p = strstr(buf, " edsp");
if(p != NULL && (p[5] == ' ' || p[5] == '\n'))
flags |= OPUS_CPU_ARM_EDSP;
# endif
# if defined(OPUS_ARM_MAY_HAVE_NEON)
p = strstr(buf, " neon");
if(p != NULL && (p[5] == ' ' || p[5] == '\n'))
flags |= OPUS_CPU_ARM_NEON;
# endif
}
# endif
# if defined(OPUS_ARM_MAY_HAVE_MEDIA)
/* Search for media capabilities (>= ARMv6) */
if(memcmp(buf, "CPU architecture:", 17) == 0)
{
int version;
version = atoi(buf+17);
if(version >= 6)
flags |= OPUS_CPU_ARM_MEDIA;
}
# endif
}
fclose(cpuinfo);
}
return flags;
}
#else
/* The feature registers which can tell us what the processor supports are
* accessible in priveleged modes only, so we can't have a general user-space
* detection method like on x86.*/
# error "Configured to use ARM asm but no CPU detection method available for " \
"your platform. Reconfigure with --disable-rtcd (or send patches)."
#endif
int opus_select_arch(void)
{
opus_uint32 flags = opus_cpu_capabilities();
int arch = 0;
if(!(flags & OPUS_CPU_ARM_EDSP))
return arch;
arch++;
if(!(flags & OPUS_CPU_ARM_MEDIA))
return arch;
arch++;
if(!(flags & OPUS_CPU_ARM_NEON))
return arch;
arch++;
return arch;
}
#endif

71
drivers/opus/celt/arm/armcpu.h Normal file
View file

@ -0,0 +1,71 @@
/* Copyright (c) 2010 Xiph.Org Foundation
* Copyright (c) 2013 Parrot */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#if !defined(ARMCPU_H)
# define ARMCPU_H
# if defined(OPUS_ARM_MAY_HAVE_EDSP)
# define MAY_HAVE_EDSP(name) name ## _edsp
# else
# define MAY_HAVE_EDSP(name) name ## _c
# endif
# if defined(OPUS_ARM_MAY_HAVE_MEDIA)
# define MAY_HAVE_MEDIA(name) name ## _media
# else
# define MAY_HAVE_MEDIA(name) MAY_HAVE_EDSP(name)
# endif
# if defined(OPUS_ARM_MAY_HAVE_NEON)
# define MAY_HAVE_NEON(name) name ## _neon
# else
# define MAY_HAVE_NEON(name) MAY_HAVE_MEDIA(name)
# endif
# if defined(OPUS_ARM_PRESUME_EDSP)
# define PRESUME_EDSP(name) name ## _edsp
# else
# define PRESUME_EDSP(name) name ## _c
# endif
# if defined(OPUS_ARM_PRESUME_MEDIA)
# define PRESUME_MEDIA(name) name ## _media
# else
# define PRESUME_MEDIA(name) PRESUME_EDSP(name)
# endif
# if defined(OPUS_ARM_PRESUME_NEON)
# define PRESUME_NEON(name) name ## _neon
# else
# define PRESUME_NEON(name) PRESUME_MEDIA(name)
# endif
# if defined(OPUS_HAVE_RTCD)
int opus_select_arch(void);
# endif
#endif

37
drivers/opus/celt/arm/armopts.s Normal file
View file

@ -0,0 +1,37 @@
/* Copyright (C) 2013 Mozilla Corporation */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
; Set the following to 1 if we have EDSP instructions
; (LDRD/STRD, etc., ARMv5E and later).
OPUS_ARM_MAY_HAVE_EDSP *
; Set the following to 1 if we have ARMv6 media instructions.
OPUS_ARM_MAY_HAVE_MEDIA *
; Set the following to 1 if we have NEON (some ARMv7)
OPUS_ARM_MAY_HAVE_NEON *
END

37
drivers/opus/celt/arm/armopts.s.in Normal file
View file

@ -0,0 +1,37 @@
/* Copyright (C) 2013 Mozilla Corporation */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
; Set the following to 1 if we have EDSP instructions
; (LDRD/STRD, etc., ARMv5E and later).
OPUS_ARM_MAY_HAVE_EDSP * @OPUS_ARM_MAY_HAVE_EDSP@
; Set the following to 1 if we have ARMv6 media instructions.
OPUS_ARM_MAY_HAVE_MEDIA * @OPUS_ARM_MAY_HAVE_MEDIA@
; Set the following to 1 if we have NEON (some ARMv7)
OPUS_ARM_MAY_HAVE_NEON * @OPUS_ARM_MAY_HAVE_NEON@
END

545
drivers/opus/celt/arm/celt_pitch_xcorr_arm.s Normal file
View file

@ -0,0 +1,545 @@
; Copyright (c) 2007-2008 CSIRO
; Copyright (c) 2007-2009 Xiph.Org Foundation
; Copyright (c) 2013 Parrot
; Written by Aurélien Zanelli
;
; Redistribution and use in source and binary forms, with or without
; modification, are permitted provided that the following conditions
; are met:
;
; - Redistributions of source code must retain the above copyright
; notice, this list of conditions and the following disclaimer.
;
; - Redistributions in binary form must reproduce the above copyright
; notice, this list of conditions and the following disclaimer in the
; documentation and/or other materials provided with the distribution.
;
; THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
; ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
; LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
; A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
; OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
; EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
; PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
; PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
; LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
; NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
; SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
AREA |.text|, CODE, READONLY
GET celt/arm/armopts.s
IF OPUS_ARM_MAY_HAVE_EDSP
EXPORT celt_pitch_xcorr_edsp
ENDIF
IF OPUS_ARM_MAY_HAVE_NEON
EXPORT celt_pitch_xcorr_neon
ENDIF
IF OPUS_ARM_MAY_HAVE_NEON
; Compute sum[k]=sum(x[j]*y[j+k],j=0...len-1), k=0...3
xcorr_kernel_neon PROC
; input:
; r3 = int len
; r4 = opus_val16 *x
; r5 = opus_val16 *y
; q0 = opus_val32 sum[4]
; output:
; q0 = opus_val32 sum[4]
; preserved: r0-r3, r6-r11, d2, q4-q7, q9-q15
; internal usage:
; r12 = int j
; d3 = y_3|y_2|y_1|y_0
; q2 = y_B|y_A|y_9|y_8|y_7|y_6|y_5|y_4
; q3 = x_7|x_6|x_5|x_4|x_3|x_2|x_1|x_0
; q8 = scratch
;
; Load y[0...3]
; This requires len>0 to always be valid (which we assert in the C code).
VLD1.16 {d5}, [r5]!
SUBS r12, r3, #8
BLE xcorr_kernel_neon_process4
; Process 8 samples at a time.
; This loop loads one y value more than we actually need. Therefore we have to
; stop as soon as there are 8 or fewer samples left (instead of 7), to avoid
; reading past the end of the array.
xcorr_kernel_neon_process8
; This loop has 19 total instructions (10 cycles to issue, minimum), with
; - 2 cycles of ARM insrtuctions,
; - 10 cycles of load/store/byte permute instructions, and
; - 9 cycles of data processing instructions.
; On a Cortex A8, we dual-issue the maximum amount (9 cycles) between the
; latter two categories, meaning the whole loop should run in 10 cycles per
; iteration, barring cache misses.
;
; Load x[0...7]
VLD1.16 {d6, d7}, [r4]!
; Unlike VMOV, VAND is a data processsing instruction (and doesn't get
; assembled to VMOV, like VORR would), so it dual-issues with the prior VLD1.
VAND d3, d5, d5
SUBS r12, r12, #8
; Load y[4...11]
VLD1.16 {d4, d5}, [r5]!
VMLAL.S16 q0, d3, d6[0]
VEXT.16 d16, d3, d4, #1
VMLAL.S16 q0, d4, d7[0]
VEXT.16 d17, d4, d5, #1
VMLAL.S16 q0, d16, d6[1]
VEXT.16 d16, d3, d4, #2
VMLAL.S16 q0, d17, d7[1]
VEXT.16 d17, d4, d5, #2
VMLAL.S16 q0, d16, d6[2]
VEXT.16 d16, d3, d4, #3
VMLAL.S16 q0, d17, d7[2]
VEXT.16 d17, d4, d5, #3
VMLAL.S16 q0, d16, d6[3]
VMLAL.S16 q0, d17, d7[3]
BGT xcorr_kernel_neon_process8
; Process 4 samples here if we have > 4 left (still reading one extra y value).
xcorr_kernel_neon_process4
ADDS r12, r12, #4
BLE xcorr_kernel_neon_process2
; Load x[0...3]
VLD1.16 d6, [r4]!
; Use VAND since it's a data processing instruction again.
VAND d4, d5, d5
SUB r12, r12, #4
; Load y[4...7]
VLD1.16 d5, [r5]!
VMLAL.S16 q0, d4, d6[0]
VEXT.16 d16, d4, d5, #1
VMLAL.S16 q0, d16, d6[1]
VEXT.16 d16, d4, d5, #2
VMLAL.S16 q0, d16, d6[2]
VEXT.16 d16, d4, d5, #3
VMLAL.S16 q0, d16, d6[3]
; Process 2 samples here if we have > 2 left (still reading one extra y value).
xcorr_kernel_neon_process2
ADDS r12, r12, #2
BLE xcorr_kernel_neon_process1
; Load x[0...1]
VLD2.16 {d6[],d7[]}, [r4]!
; Use VAND since it's a data processing instruction again.
VAND d4, d5, d5
SUB r12, r12, #2
; Load y[4...5]
VLD1.32 {d5[]}, [r5]!
VMLAL.S16 q0, d4, d6
VEXT.16 d16, d4, d5, #1
; Replace bottom copy of {y5,y4} in d5 with {y3,y2} from d4, using VSRI
; instead of VEXT, since it's a data-processing instruction.
VSRI.64 d5, d4, #32
VMLAL.S16 q0, d16, d7
; Process 1 sample using the extra y value we loaded above.
xcorr_kernel_neon_process1
; Load next *x
VLD1.16 {d6[]}, [r4]!
ADDS r12, r12, #1
; y[0...3] are left in d5 from prior iteration(s) (if any)
VMLAL.S16 q0, d5, d6
MOVLE pc, lr
; Now process 1 last sample, not reading ahead.
; Load last *y
VLD1.16 {d4[]}, [r5]!
VSRI.64 d4, d5, #16
; Load last *x
VLD1.16 {d6[]}, [r4]!
VMLAL.S16 q0, d4, d6
MOV pc, lr
ENDP
; opus_val32 celt_pitch_xcorr_neon(opus_val16 *_x, opus_val16 *_y,
; opus_val32 *xcorr, int len, int max_pitch)
celt_pitch_xcorr_neon PROC
; input:
; r0 = opus_val16 *_x
; r1 = opus_val16 *_y
; r2 = opus_val32 *xcorr
; r3 = int len
; output:
; r0 = int maxcorr
; internal usage:
; r4 = opus_val16 *x (for xcorr_kernel_neon())
; r5 = opus_val16 *y (for xcorr_kernel_neon())
; r6 = int max_pitch
; r12 = int j
; q15 = int maxcorr[4] (q15 is not used by xcorr_kernel_neon())
STMFD sp!, {r4-r6, lr}
LDR r6, [sp, #16]
VMOV.S32 q15, #1
; if (max_pitch < 4) goto celt_pitch_xcorr_neon_process4_done
SUBS r6, r6, #4
BLT celt_pitch_xcorr_neon_process4_done
celt_pitch_xcorr_neon_process4
; xcorr_kernel_neon parameters:
; r3 = len, r4 = _x, r5 = _y, q0 = {0, 0, 0, 0}
MOV r4, r0
MOV r5, r1
VEOR q0, q0, q0
; xcorr_kernel_neon only modifies r4, r5, r12, and q0...q3.
; So we don't save/restore any other registers.
BL xcorr_kernel_neon
SUBS r6, r6, #4
VST1.32 {q0}, [r2]!
; _y += 4
ADD r1, r1, #8
VMAX.S32 q15, q15, q0
; if (max_pitch < 4) goto celt_pitch_xcorr_neon_process4_done
BGE celt_pitch_xcorr_neon_process4
; We have less than 4 sums left to compute.
celt_pitch_xcorr_neon_process4_done
ADDS r6, r6, #4
; Reduce maxcorr to a single value
VMAX.S32 d30, d30, d31
VPMAX.S32 d30, d30, d30
; if (max_pitch <= 0) goto celt_pitch_xcorr_neon_done
BLE celt_pitch_xcorr_neon_done
; Now compute each remaining sum one at a time.
celt_pitch_xcorr_neon_process_remaining
MOV r4, r0
MOV r5, r1
VMOV.I32 q0, #0
SUBS r12, r3, #8
BLT celt_pitch_xcorr_neon_process_remaining4
; Sum terms 8 at a time.
celt_pitch_xcorr_neon_process_remaining_loop8
; Load x[0...7]
VLD1.16 {q1}, [r4]!
; Load y[0...7]
VLD1.16 {q2}, [r5]!
SUBS r12, r12, #8
VMLAL.S16 q0, d4, d2
VMLAL.S16 q0, d5, d3
BGE celt_pitch_xcorr_neon_process_remaining_loop8
; Sum terms 4 at a time.
celt_pitch_xcorr_neon_process_remaining4
ADDS r12, r12, #4
BLT celt_pitch_xcorr_neon_process_remaining4_done
; Load x[0...3]
VLD1.16 {d2}, [r4]!
; Load y[0...3]
VLD1.16 {d3}, [r5]!
SUB r12, r12, #4
VMLAL.S16 q0, d3, d2
celt_pitch_xcorr_neon_process_remaining4_done
; Reduce the sum to a single value.
VADD.S32 d0, d0, d1
VPADDL.S32 d0, d0
ADDS r12, r12, #4
BLE celt_pitch_xcorr_neon_process_remaining_loop_done
; Sum terms 1 at a time.
celt_pitch_xcorr_neon_process_remaining_loop1
VLD1.16 {d2[]}, [r4]!
VLD1.16 {d3[]}, [r5]!
SUBS r12, r12, #1
VMLAL.S16 q0, d2, d3
BGT celt_pitch_xcorr_neon_process_remaining_loop1
celt_pitch_xcorr_neon_process_remaining_loop_done
VST1.32 {d0[0]}, [r2]!
VMAX.S32 d30, d30, d0
SUBS r6, r6, #1
; _y++
ADD r1, r1, #2
; if (--max_pitch > 0) goto celt_pitch_xcorr_neon_process_remaining
BGT celt_pitch_xcorr_neon_process_remaining
celt_pitch_xcorr_neon_done
VMOV.32 r0, d30[0]
LDMFD sp!, {r4-r6, pc}
ENDP
ENDIF
IF OPUS_ARM_MAY_HAVE_EDSP
; This will get used on ARMv7 devices without NEON, so it has been optimized
; to take advantage of dual-issuing where possible.
xcorr_kernel_edsp PROC
; input:
; r3 = int len
; r4 = opus_val16 *_x (must be 32-bit aligned)
; r5 = opus_val16 *_y (must be 32-bit aligned)
; r6...r9 = opus_val32 sum[4]
; output:
; r6...r9 = opus_val32 sum[4]
; preserved: r0-r5
; internal usage
; r2 = int j
; r12,r14 = opus_val16 x[4]
; r10,r11 = opus_val16 y[4]
STMFD sp!, {r2,r4,r5,lr}
LDR r10, [r5], #4 ; Load y[0...1]
SUBS r2, r3, #4 ; j = len-4
LDR r11, [r5], #4 ; Load y[2...3]
BLE xcorr_kernel_edsp_process4_done
LDR r12, [r4], #4 ; Load x[0...1]
; Stall
xcorr_kernel_edsp_process4
; The multiplies must issue from pipeline 0, and can't dual-issue with each
; other. Every other instruction here dual-issues with a multiply, and is
; thus "free". There should be no stalls in the body of the loop.
SMLABB r6, r12, r10, r6 ; sum[0] = MAC16_16(sum[0],x_0,y_0)
LDR r14, [r4], #4 ; Load x[2...3]
SMLABT r7, r12, r10, r7 ; sum[1] = MAC16_16(sum[1],x_0,y_1)
SUBS r2, r2, #4 ; j-=4
SMLABB r8, r12, r11, r8 ; sum[2] = MAC16_16(sum[2],x_0,y_2)
SMLABT r9, r12, r11, r9 ; sum[3] = MAC16_16(sum[3],x_0,y_3)
SMLATT r6, r12, r10, r6 ; sum[0] = MAC16_16(sum[0],x_1,y_1)
LDR r10, [r5], #4 ; Load y[4...5]
SMLATB r7, r12, r11, r7 ; sum[1] = MAC16_16(sum[1],x_1,y_2)
SMLATT r8, r12, r11, r8 ; sum[2] = MAC16_16(sum[2],x_1,y_3)
SMLATB r9, r12, r10, r9 ; sum[3] = MAC16_16(sum[3],x_1,y_4)
LDRGT r12, [r4], #4 ; Load x[0...1]
SMLABB r6, r14, r11, r6 ; sum[0] = MAC16_16(sum[0],x_2,y_2)
SMLABT r7, r14, r11, r7 ; sum[1] = MAC16_16(sum[1],x_2,y_3)
SMLABB r8, r14, r10, r8 ; sum[2] = MAC16_16(sum[2],x_2,y_4)
SMLABT r9, r14, r10, r9 ; sum[3] = MAC16_16(sum[3],x_2,y_5)
SMLATT r6, r14, r11, r6 ; sum[0] = MAC16_16(sum[0],x_3,y_3)
LDR r11, [r5], #4 ; Load y[6...7]
SMLATB r7, r14, r10, r7 ; sum[1] = MAC16_16(sum[1],x_3,y_4)
SMLATT r8, r14, r10, r8 ; sum[2] = MAC16_16(sum[2],x_3,y_5)
SMLATB r9, r14, r11, r9 ; sum[3] = MAC16_16(sum[3],x_3,y_6)
BGT xcorr_kernel_edsp_process4
xcorr_kernel_edsp_process4_done
ADDS r2, r2, #4
BLE xcorr_kernel_edsp_done
LDRH r12, [r4], #2 ; r12 = *x++
SUBS r2, r2, #1 ; j--
; Stall
SMLABB r6, r12, r10, r6 ; sum[0] = MAC16_16(sum[0],x,y_0)
LDRGTH r14, [r4], #2 ; r14 = *x++
SMLABT r7, r12, r10, r7 ; sum[1] = MAC16_16(sum[1],x,y_1)
SMLABB r8, r12, r11, r8 ; sum[2] = MAC16_16(sum[2],x,y_2)
SMLABT r9, r12, r11, r9 ; sum[3] = MAC16_16(sum[3],x,y_3)
BLE xcorr_kernel_edsp_done
SMLABT r6, r14, r10, r6 ; sum[0] = MAC16_16(sum[0],x,y_1)
SUBS r2, r2, #1 ; j--
SMLABB r7, r14, r11, r7 ; sum[1] = MAC16_16(sum[1],x,y_2)
LDRH r10, [r5], #2 ; r10 = y_4 = *y++
SMLABT r8, r14, r11, r8 ; sum[2] = MAC16_16(sum[2],x,y_3)
LDRGTH r12, [r4], #2 ; r12 = *x++
SMLABB r9, r14, r10, r9 ; sum[3] = MAC16_16(sum[3],x,y_4)
BLE xcorr_kernel_edsp_done
SMLABB r6, r12, r11, r6 ; sum[0] = MAC16_16(sum[0],tmp,y_2)
CMP r2, #1 ; j--
SMLABT r7, r12, r11, r7 ; sum[1] = MAC16_16(sum[1],tmp,y_3)
LDRH r2, [r5], #2 ; r2 = y_5 = *y++
SMLABB r8, r12, r10, r8 ; sum[2] = MAC16_16(sum[2],tmp,y_4)
LDRGTH r14, [r4] ; r14 = *x
SMLABB r9, r12, r2, r9 ; sum[3] = MAC16_16(sum[3],tmp,y_5)
BLE xcorr_kernel_edsp_done
SMLABT r6, r14, r11, r6 ; sum[0] = MAC16_16(sum[0],tmp,y_3)
LDRH r11, [r5] ; r11 = y_6 = *y
SMLABB r7, r14, r10, r7 ; sum[1] = MAC16_16(sum[1],tmp,y_4)
SMLABB r8, r14, r2, r8 ; sum[2] = MAC16_16(sum[2],tmp,y_5)
SMLABB r9, r14, r11, r9 ; sum[3] = MAC16_16(sum[3],tmp,y_6)
xcorr_kernel_edsp_done
LDMFD sp!, {r2,r4,r5,pc}
ENDP
celt_pitch_xcorr_edsp PROC
; input:
; r0 = opus_val16 *_x (must be 32-bit aligned)
; r1 = opus_val16 *_y (only needs to be 16-bit aligned)
; r2 = opus_val32 *xcorr
; r3 = int len
; output:
; r0 = maxcorr
; internal usage
; r4 = opus_val16 *x
; r5 = opus_val16 *y
; r6 = opus_val32 sum0
; r7 = opus_val32 sum1
; r8 = opus_val32 sum2
; r9 = opus_val32 sum3
; r1 = int max_pitch
; r12 = int j
STMFD sp!, {r4-r11, lr}
MOV r5, r1
LDR r1, [sp, #36]
MOV r4, r0
TST r5, #3
; maxcorr = 1
MOV r0, #1
BEQ celt_pitch_xcorr_edsp_process1u_done
; Compute one sum at the start to make y 32-bit aligned.
SUBS r12, r3, #4
; r14 = sum = 0
MOV r14, #0
LDRH r8, [r5], #2
BLE celt_pitch_xcorr_edsp_process1u_loop4_done
LDR r6, [r4], #4
MOV r8, r8, LSL #16
celt_pitch_xcorr_edsp_process1u_loop4
LDR r9, [r5], #4
SMLABT r14, r6, r8, r14 ; sum = MAC16_16(sum, x_0, y_0)
LDR r7, [r4], #4
SMLATB r14, r6, r9, r14 ; sum = MAC16_16(sum, x_1, y_1)
LDR r8, [r5], #4
SMLABT r14, r7, r9, r14 ; sum = MAC16_16(sum, x_2, y_2)
SUBS r12, r12, #4 ; j-=4
SMLATB r14, r7, r8, r14 ; sum = MAC16_16(sum, x_3, y_3)
LDRGT r6, [r4], #4
BGT celt_pitch_xcorr_edsp_process1u_loop4
MOV r8, r8, LSR #16
celt_pitch_xcorr_edsp_process1u_loop4_done
ADDS r12, r12, #4
celt_pitch_xcorr_edsp_process1u_loop1
LDRGEH r6, [r4], #2
; Stall
SMLABBGE r14, r6, r8, r14 ; sum = MAC16_16(sum, *x, *y)
SUBGES r12, r12, #1
LDRGTH r8, [r5], #2
BGT celt_pitch_xcorr_edsp_process1u_loop1
; Restore _x
SUB r4, r4, r3, LSL #1
; Restore and advance _y
SUB r5, r5, r3, LSL #1
; maxcorr = max(maxcorr, sum)
CMP r0, r14
ADD r5, r5, #2
MOVLT r0, r14
SUBS r1, r1, #1
; xcorr[i] = sum
STR r14, [r2], #4
BLE celt_pitch_xcorr_edsp_done
celt_pitch_xcorr_edsp_process1u_done
; if (max_pitch < 4) goto celt_pitch_xcorr_edsp_process2
SUBS r1, r1, #4
BLT celt_pitch_xcorr_edsp_process2
celt_pitch_xcorr_edsp_process4
; xcorr_kernel_edsp parameters:
; r3 = len, r4 = _x, r5 = _y, r6...r9 = sum[4] = {0, 0, 0, 0}
MOV r6, #0
MOV r7, #0
MOV r8, #0
MOV r9, #0
BL xcorr_kernel_edsp ; xcorr_kernel_edsp(_x, _y+i, xcorr+i, len)
; maxcorr = max(maxcorr, sum0, sum1, sum2, sum3)
CMP r0, r6
; _y+=4
ADD r5, r5, #8
MOVLT r0, r6
CMP r0, r7
MOVLT r0, r7
CMP r0, r8
MOVLT r0, r8
CMP r0, r9
MOVLT r0, r9
STMIA r2!, {r6-r9}
SUBS r1, r1, #4
BGE celt_pitch_xcorr_edsp_process4
celt_pitch_xcorr_edsp_process2
ADDS r1, r1, #2
BLT celt_pitch_xcorr_edsp_process1a
SUBS r12, r3, #4
; {r10, r11} = {sum0, sum1} = {0, 0}
MOV r10, #0
MOV r11, #0
LDR r8, [r5], #4
BLE celt_pitch_xcorr_edsp_process2_loop_done
LDR r6, [r4], #4
LDR r9, [r5], #4
celt_pitch_xcorr_edsp_process2_loop4
SMLABB r10, r6, r8, r10 ; sum0 = MAC16_16(sum0, x_0, y_0)
LDR r7, [r4], #4
SMLABT r11, r6, r8, r11 ; sum1 = MAC16_16(sum1, x_0, y_1)
SUBS r12, r12, #4 ; j-=4
SMLATT r10, r6, r8, r10 ; sum0 = MAC16_16(sum0, x_1, y_1)
LDR r8, [r5], #4
SMLATB r11, r6, r9, r11 ; sum1 = MAC16_16(sum1, x_1, y_2)
LDRGT r6, [r4], #4
SMLABB r10, r7, r9, r10 ; sum0 = MAC16_16(sum0, x_2, y_2)
SMLABT r11, r7, r9, r11 ; sum1 = MAC16_16(sum1, x_2, y_3)
SMLATT r10, r7, r9, r10 ; sum0 = MAC16_16(sum0, x_3, y_3)
LDRGT r9, [r5], #4
SMLATB r11, r7, r8, r11 ; sum1 = MAC16_16(sum1, x_3, y_4)
BGT celt_pitch_xcorr_edsp_process2_loop4
celt_pitch_xcorr_edsp_process2_loop_done
ADDS r12, r12, #2
BLE celt_pitch_xcorr_edsp_process2_1
LDR r6, [r4], #4
; Stall
SMLABB r10, r6, r8, r10 ; sum0 = MAC16_16(sum0, x_0, y_0)
LDR r9, [r5], #4
SMLABT r11, r6, r8, r11 ; sum1 = MAC16_16(sum1, x_0, y_1)
SUB r12, r12, #2
SMLATT r10, r6, r8, r10 ; sum0 = MAC16_16(sum0, x_1, y_1)
MOV r8, r9
SMLATB r11, r6, r9, r11 ; sum1 = MAC16_16(sum1, x_1, y_2)
celt_pitch_xcorr_edsp_process2_1
LDRH r6, [r4], #2
ADDS r12, r12, #1
; Stall
SMLABB r10, r6, r8, r10 ; sum0 = MAC16_16(sum0, x_0, y_0)
LDRGTH r7, [r4], #2
SMLABT r11, r6, r8, r11 ; sum1 = MAC16_16(sum1, x_0, y_1)
BLE celt_pitch_xcorr_edsp_process2_done
LDRH r9, [r5], #2
SMLABT r10, r7, r8, r10 ; sum0 = MAC16_16(sum0, x_0, y_1)
SMLABB r11, r7, r9, r11 ; sum1 = MAC16_16(sum1, x_0, y_2)
celt_pitch_xcorr_edsp_process2_done
; Restore _x
SUB r4, r4, r3, LSL #1
; Restore and advance _y
SUB r5, r5, r3, LSL #1
; maxcorr = max(maxcorr, sum0)
CMP r0, r10
ADD r5, r5, #2
MOVLT r0, r10
SUB r1, r1, #2
; maxcorr = max(maxcorr, sum1)
CMP r0, r11
; xcorr[i] = sum
STR r10, [r2], #4
MOVLT r0, r11
STR r11, [r2], #4
celt_pitch_xcorr_edsp_process1a
ADDS r1, r1, #1
BLT celt_pitch_xcorr_edsp_done
SUBS r12, r3, #4
; r14 = sum = 0
MOV r14, #0
BLT celt_pitch_xcorr_edsp_process1a_loop_done
LDR r6, [r4], #4
LDR r8, [r5], #4
LDR r7, [r4], #4
LDR r9, [r5], #4
celt_pitch_xcorr_edsp_process1a_loop4
SMLABB r14, r6, r8, r14 ; sum = MAC16_16(sum, x_0, y_0)
SUBS r12, r12, #4 ; j-=4
SMLATT r14, r6, r8, r14 ; sum = MAC16_16(sum, x_1, y_1)
LDRGE r6, [r4], #4
SMLABB r14, r7, r9, r14 ; sum = MAC16_16(sum, x_2, y_2)
LDRGE r8, [r5], #4
SMLATT r14, r7, r9, r14 ; sum = MAC16_16(sum, x_3, y_3)
LDRGE r7, [r4], #4
LDRGE r9, [r5], #4
BGE celt_pitch_xcorr_edsp_process1a_loop4
celt_pitch_xcorr_edsp_process1a_loop_done
ADDS r12, r12, #2
LDRGE r6, [r4], #4
LDRGE r8, [r5], #4
; Stall
SMLABBGE r14, r6, r8, r14 ; sum = MAC16_16(sum, x_0, y_0)
SUBGE r12, r12, #2
SMLATTGE r14, r6, r8, r14 ; sum = MAC16_16(sum, x_1, y_1)
ADDS r12, r12, #1
LDRGEH r6, [r4], #2
LDRGEH r8, [r5], #2
; Stall
SMLABBGE r14, r6, r8, r14 ; sum = MAC16_16(sum, *x, *y)
; maxcorr = max(maxcorr, sum)
CMP r0, r14
; xcorr[i] = sum
STR r14, [r2], #4
MOVLT r0, r14
celt_pitch_xcorr_edsp_done
LDMFD sp!, {r4-r11, pc}
ENDP
ENDIF
END

76
drivers/opus/celt/arm/fixed_armv4.h Normal file
View file

@ -0,0 +1,76 @@
/* Copyright (C) 2013 Xiph.Org Foundation and contributors */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef FIXED_ARMv4_H
#define FIXED_ARMv4_H
/** 16x32 multiplication, followed by a 16-bit shift right. Results fits in 32 bits */
#undef MULT16_32_Q16
static OPUS_INLINE opus_val32 MULT16_32_Q16_armv4(opus_val16 a, opus_val32 b)
{
unsigned rd_lo;
int rd_hi;
__asm__(
"#MULT16_32_Q16\n\t"
"smull %0, %1, %2, %3\n\t"
: "=&r"(rd_lo), "=&r"(rd_hi)
: "%r"(b),"r"(a<<16)
);
return rd_hi;
}
#define MULT16_32_Q16(a, b) (MULT16_32_Q16_armv4(a, b))
/** 16x32 multiplication, followed by a 15-bit shift right. Results fits in 32 bits */
#undef MULT16_32_Q15
static OPUS_INLINE opus_val32 MULT16_32_Q15_armv4(opus_val16 a, opus_val32 b)
{
unsigned rd_lo;
int rd_hi;
__asm__(
"#MULT16_32_Q15\n\t"
"smull %0, %1, %2, %3\n\t"
: "=&r"(rd_lo), "=&r"(rd_hi)
: "%r"(b), "r"(a<<16)
);
/*We intentionally don't OR in the high bit of rd_lo for speed.*/
return rd_hi<<1;
}
#define MULT16_32_Q15(a, b) (MULT16_32_Q15_armv4(a, b))
/** 16x32 multiply, followed by a 15-bit shift right and 32-bit add.
b must fit in 31 bits.
Result fits in 32 bits. */
#undef MAC16_32_Q15
#define MAC16_32_Q15(c, a, b) ADD32(c, MULT16_32_Q15(a, b))
/** 32x32 multiplication, followed by a 31-bit shift right. Results fits in 32 bits */
#undef MULT32_32_Q31
#define MULT32_32_Q31(a,b) (opus_val32)((((opus_int64)(a)) * ((opus_int64)(b)))>>31)
#endif

116
drivers/opus/celt/arm/fixed_armv5e.h Normal file
View file

@ -0,0 +1,116 @@
/* Copyright (C) 2007-2009 Xiph.Org Foundation
Copyright (C) 2003-2008 Jean-Marc Valin
Copyright (C) 2007-2008 CSIRO
Copyright (C) 2013 Parrot */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef FIXED_ARMv5E_H
#define FIXED_ARMv5E_H
#include "fixed_armv4.h"
/** 16x32 multiplication, followed by a 16-bit shift right. Results fits in 32 bits */
#undef MULT16_32_Q16
static OPUS_INLINE opus_val32 MULT16_32_Q16_armv5e(opus_val16 a, opus_val32 b)
{
int res;
__asm__(
"#MULT16_32_Q16\n\t"
"smulwb %0, %1, %2\n\t"
: "=r"(res)
: "r"(b),"r"(a)
);
return res;
}
#define MULT16_32_Q16(a, b) (MULT16_32_Q16_armv5e(a, b))
/** 16x32 multiplication, followed by a 15-bit shift right. Results fits in 32 bits */
#undef MULT16_32_Q15
static OPUS_INLINE opus_val32 MULT16_32_Q15_armv5e(opus_val16 a, opus_val32 b)
{
int res;
__asm__(
"#MULT16_32_Q15\n\t"
"smulwb %0, %1, %2\n\t"
: "=r"(res)
: "r"(b), "r"(a)
);
return res<<1;
}
#define MULT16_32_Q15(a, b) (MULT16_32_Q15_armv5e(a, b))
/** 16x32 multiply, followed by a 15-bit shift right and 32-bit add.
b must fit in 31 bits.
Result fits in 32 bits. */
#undef MAC16_32_Q15
static OPUS_INLINE opus_val32 MAC16_32_Q15_armv5e(opus_val32 c, opus_val16 a,
opus_val32 b)
{
int res;
__asm__(
"#MAC16_32_Q15\n\t"
"smlawb %0, %1, %2, %3;\n"
: "=r"(res)
: "r"(b<<1), "r"(a), "r"(c)
);
return res;
}
#define MAC16_32_Q15(c, a, b) (MAC16_32_Q15_armv5e(c, a, b))
/** 16x16 multiply-add where the result fits in 32 bits */
#undef MAC16_16
static OPUS_INLINE opus_val32 MAC16_16_armv5e(opus_val32 c, opus_val16 a,
opus_val16 b)
{
int res;
__asm__(
"#MAC16_16\n\t"
"smlabb %0, %1, %2, %3;\n"
: "=r"(res)
: "r"(a), "r"(b), "r"(c)
);
return res;
}
#define MAC16_16(c, a, b) (MAC16_16_armv5e(c, a, b))
/** 16x16 multiplication where the result fits in 32 bits */
#undef MULT16_16
static OPUS_INLINE opus_val32 MULT16_16_armv5e(opus_val16 a, opus_val16 b)
{
int res;
__asm__(
"#MULT16_16\n\t"
"smulbb %0, %1, %2;\n"
: "=r"(res)
: "r"(a), "r"(b)
);
return res;
}
#define MULT16_16(a, b) (MULT16_16_armv5e(a, b))
#endif

121
drivers/opus/celt/arm/kiss_fft_armv4.h Normal file
View file

@ -0,0 +1,121 @@
/*Copyright (c) 2013, Xiph.Org Foundation and contributors.
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice,
this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.*/
#ifndef KISS_FFT_ARMv4_H
#define KISS_FFT_ARMv4_H
#if !defined(KISS_FFT_GUTS_H)
#error "This file should only be included from _kiss_fft_guts.h"
#endif
#ifdef OPUS_FIXED_POINT
#undef C_MUL
#define C_MUL(m,a,b) \
do{ \
int br__; \
int bi__; \
int tt__; \
__asm__ __volatile__( \
"#C_MUL\n\t" \
"ldrsh %[br], [%[bp], #0]\n\t" \
"ldm %[ap], {r0,r1}\n\t" \
"ldrsh %[bi], [%[bp], #2]\n\t" \
"smull %[tt], %[mi], r1, %[br]\n\t" \
"smlal %[tt], %[mi], r0, %[bi]\n\t" \
"rsb %[bi], %[bi], #0\n\t" \
"smull %[br], %[mr], r0, %[br]\n\t" \
"mov %[tt], %[tt], lsr #15\n\t" \
"smlal %[br], %[mr], r1, %[bi]\n\t" \
"orr %[mi], %[tt], %[mi], lsl #17\n\t" \
"mov %[br], %[br], lsr #15\n\t" \
"orr %[mr], %[br], %[mr], lsl #17\n\t" \
: [mr]"=r"((m).r), [mi]"=r"((m).i), \
[br]"=&r"(br__), [bi]"=r"(bi__), [tt]"=r"(tt__) \
: [ap]"r"(&(a)), [bp]"r"(&(b)) \
: "r0", "r1" \
); \
} \
while(0)
#undef C_MUL4
#define C_MUL4(m,a,b) \
do{ \
int br__; \
int bi__; \
int tt__; \
__asm__ __volatile__( \
"#C_MUL4\n\t" \
"ldrsh %[br], [%[bp], #0]\n\t" \
"ldm %[ap], {r0,r1}\n\t" \
"ldrsh %[bi], [%[bp], #2]\n\t" \
"smull %[tt], %[mi], r1, %[br]\n\t" \
"smlal %[tt], %[mi], r0, %[bi]\n\t" \
"rsb %[bi], %[bi], #0\n\t" \
"smull %[br], %[mr], r0, %[br]\n\t" \
"mov %[tt], %[tt], lsr #17\n\t" \
"smlal %[br], %[mr], r1, %[bi]\n\t" \
"orr %[mi], %[tt], %[mi], lsl #15\n\t" \
"mov %[br], %[br], lsr #17\n\t" \
"orr %[mr], %[br], %[mr], lsl #15\n\t" \
: [mr]"=r"((m).r), [mi]"=r"((m).i), \
[br]"=&r"(br__), [bi]"=r"(bi__), [tt]"=r"(tt__) \
: [ap]"r"(&(a)), [bp]"r"(&(b)) \
: "r0", "r1" \
); \
} \
while(0)
#undef C_MULC
#define C_MULC(m,a,b) \
do{ \
int br__; \
int bi__; \
int tt__; \
__asm__ __volatile__( \
"#C_MULC\n\t" \
"ldrsh %[br], [%[bp], #0]\n\t" \
"ldm %[ap], {r0,r1}\n\t" \
"ldrsh %[bi], [%[bp], #2]\n\t" \
"smull %[tt], %[mr], r0, %[br]\n\t" \
"smlal %[tt], %[mr], r1, %[bi]\n\t" \
"rsb %[bi], %[bi], #0\n\t" \
"smull %[br], %[mi], r1, %[br]\n\t" \
"mov %[tt], %[tt], lsr #15\n\t" \
"smlal %[br], %[mi], r0, %[bi]\n\t" \
"orr %[mr], %[tt], %[mr], lsl #17\n\t" \
"mov %[br], %[br], lsr #15\n\t" \
"orr %[mi], %[br], %[mi], lsl #17\n\t" \
: [mr]"=r"((m).r), [mi]"=r"((m).i), \
[br]"=&r"(br__), [bi]"=r"(bi__), [tt]"=r"(tt__) \
: [ap]"r"(&(a)), [bp]"r"(&(b)) \
: "r0", "r1" \
); \
} \
while(0)
#endif /* OPUS_FIXED_POINT */
#endif /* KISS_FFT_ARMv4_H */

118
drivers/opus/celt/arm/kiss_fft_armv5e.h Normal file
View file

@ -0,0 +1,118 @@
/*Copyright (c) 2013, Xiph.Org Foundation and contributors.
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice,
this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.*/
#ifndef KISS_FFT_ARMv5E_H
#define KISS_FFT_ARMv5E_H
#if !defined(KISS_FFT_GUTS_H)
#error "This file should only be included from _kiss_fft_guts.h"
#endif
#ifdef OPUS_FIXED_POINT
#if defined(__thumb__)||defined(__thumb2__)
#define LDRD_CONS "Q"
#else
#define LDRD_CONS "Uq"
#endif
#undef C_MUL
#define C_MUL(m,a,b) \
do{ \
int mr1__; \
int mr2__; \
int mi__; \
long long aval__; \
int bval__; \
__asm__( \
"#C_MUL\n\t" \
"ldrd %[aval], %H[aval], %[ap]\n\t" \
"ldr %[bval], %[bp]\n\t" \
"smulwb %[mi], %H[aval], %[bval]\n\t" \
"smulwb %[mr1], %[aval], %[bval]\n\t" \
"smulwt %[mr2], %H[aval], %[bval]\n\t" \
"smlawt %[mi], %[aval], %[bval], %[mi]\n\t" \
: [mr1]"=r"(mr1__), [mr2]"=r"(mr2__), [mi]"=r"(mi__), \
[aval]"=&r"(aval__), [bval]"=r"(bval__) \
: [ap]LDRD_CONS(a), [bp]"m"(b) \
); \
(m).r = SHL32(SUB32(mr1__, mr2__), 1); \
(m).i = SHL32(mi__, 1); \
} \
while(0)
#undef C_MUL4
#define C_MUL4(m,a,b) \
do{ \
int mr1__; \
int mr2__; \
int mi__; \
long long aval__; \
int bval__; \
__asm__( \
"#C_MUL4\n\t" \
"ldrd %[aval], %H[aval], %[ap]\n\t" \
"ldr %[bval], %[bp]\n\t" \
"smulwb %[mi], %H[aval], %[bval]\n\t" \
"smulwb %[mr1], %[aval], %[bval]\n\t" \
"smulwt %[mr2], %H[aval], %[bval]\n\t" \
"smlawt %[mi], %[aval], %[bval], %[mi]\n\t" \
: [mr1]"=r"(mr1__), [mr2]"=r"(mr2__), [mi]"=r"(mi__), \
[aval]"=&r"(aval__), [bval]"=r"(bval__) \
: [ap]LDRD_CONS(a), [bp]"m"(b) \
); \
(m).r = SHR32(SUB32(mr1__, mr2__), 1); \
(m).i = SHR32(mi__, 1); \
} \
while(0)
#undef C_MULC
#define C_MULC(m,a,b) \
do{ \
int mr__; \
int mi1__; \
int mi2__; \
long long aval__; \
int bval__; \
__asm__( \
"#C_MULC\n\t" \
"ldrd %[aval], %H[aval], %[ap]\n\t" \
"ldr %[bval], %[bp]\n\t" \
"smulwb %[mr], %[aval], %[bval]\n\t" \
"smulwb %[mi1], %H[aval], %[bval]\n\t" \
"smulwt %[mi2], %[aval], %[bval]\n\t" \
"smlawt %[mr], %H[aval], %[bval], %[mr]\n\t" \
: [mr]"=r"(mr__), [mi1]"=r"(mi1__), [mi2]"=r"(mi2__), \
[aval]"=&r"(aval__), [bval]"=r"(bval__) \
: [ap]LDRD_CONS(a), [bp]"m"(b) \
); \
(m).r = SHL32(mr__, 1); \
(m).i = SHL32(SUB32(mi1__, mi2__), 1); \
} \
while(0)
#endif /* OPUS_FIXED_POINT */
#endif /* KISS_FFT_GUTS_H */

57
drivers/opus/celt/arm/pitch_arm.h Normal file
View file

@ -0,0 +1,57 @@
/* Copyright (c) 2010 Xiph.Org Foundation
* Copyright (c) 2013 Parrot */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#if !defined(PITCH_ARM_H)
# define PITCH_ARM_H
# include "armcpu.h"
# if defined(OPUS_FIXED_POINT)
# if defined(OPUS_ARM_MAY_HAVE_NEON)
opus_val32 celt_pitch_xcorr_neon(const opus_val16 *_x, const opus_val16 *_y,
opus_val32 *xcorr, int len, int max_pitch);
# endif
# if defined(OPUS_ARM_MAY_HAVE_MEDIA)
# define celt_pitch_xcorr_media MAY_HAVE_EDSP(celt_pitch_xcorr)
# endif
# if defined(OPUS_ARM_MAY_HAVE_EDSP)
opus_val32 celt_pitch_xcorr_edsp(const opus_val16 *_x, const opus_val16 *_y,
opus_val32 *xcorr, int len, int max_pitch);
# endif
# if !defined(OPUS_HAVE_RTCD)
# define OVERRIDE_PITCH_XCORR (1)
# define celt_pitch_xcorr(_x, _y, xcorr, len, max_pitch, arch) \
((void)(arch),PRESUME_NEON(celt_pitch_xcorr)(_x, _y, xcorr, len, max_pitch))
# endif
# endif
#endif

1518
drivers/opus/celt/bands.c Normal file
View file

File diff suppressed because it is too large Load diff

114
drivers/opus/celt/bands.h Normal file
View file

@ -0,0 +1,114 @@
/* Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2009 Xiph.Org Foundation
Copyright (c) 2008-2009 Gregory Maxwell
Written by Jean-Marc Valin and Gregory Maxwell */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef BANDS_H
#define BANDS_H
#include "arch.h"
#include "opus_modes.h"
#include "entenc.h"
#include "entdec.h"
#include "rate.h"
/** Compute the amplitude (sqrt energy) in each of the bands
* @param m Mode data
* @param X Spectrum
* @param bandE Square root of the energy for each band (returned)
*/
void compute_band_energies(const CELTMode *m, const celt_sig *X, celt_ener *bandE, int end, int C, int M);
/*void compute_noise_energies(const CELTMode *m, const celt_sig *X, const opus_val16 *tonality, celt_ener *bandE);*/
/** Normalise each band of X such that the energy in each band is
equal to 1
* @param m Mode data
* @param X Spectrum (returned normalised)
* @param bandE Square root of the energy for each band
*/
void normalise_bands(const CELTMode *m, const celt_sig * OPUS_RESTRICT freq, celt_norm * OPUS_RESTRICT X, const celt_ener *bandE, int end, int C, int M);
/** Denormalise each band of X to restore full amplitude
* @param m Mode data
* @param X Spectrum (returned de-normalised)
* @param bandE Square root of the energy for each band
*/
void denormalise_bands(const CELTMode *m, const celt_norm * OPUS_RESTRICT X,
celt_sig * OPUS_RESTRICT freq, const opus_val16 *bandE, int start, int end, int C, int M);
#define SPREAD_NONE (0)
#define SPREAD_LIGHT (1)
#define SPREAD_NORMAL (2)
#define SPREAD_AGGRESSIVE (3)
int spreading_decision(const CELTMode *m, celt_norm *X, int *average,
int last_decision, int *hf_average, int *tapset_decision, int update_hf,
int end, int C, int M);
#ifdef MEASURE_NORM_MSE
void measure_norm_mse(const CELTMode *m, float *X, float *X0, float *bandE, float *bandE0, int M, int N, int C);
#endif
void haar1(celt_norm *X, int N0, int stride);
/** Quantisation/encoding of the residual spectrum
* @param encode flag that indicates whether we're encoding (1) or decoding (0)
* @param m Mode data
* @param start First band to process
* @param end Last band to process + 1
* @param X Residual (normalised)
* @param Y Residual (normalised) for second channel (or NULL for mono)
* @param collapse_masks Anti-collapse tracking mask
* @param bandE Square root of the energy for each band
* @param pulses Bit allocation (per band) for PVQ
* @param shortBlocks Zero for long blocks, non-zero for short blocks
* @param spread Amount of spreading to use
* @param dual_stereo Zero for MS stereo, non-zero for dual stereo
* @param intensity First band to use intensity stereo
* @param tf_res Time-frequency resolution change
* @param total_bits Total number of bits that can be used for the frame (including the ones already spent)
* @param balance Number of unallocated bits
* @param en Entropy coder state
* @param LM log2() of the number of 2.5 subframes in the frame
* @param codedBands Last band to receive bits + 1
* @param seed Random generator seed
*/
void quant_all_bands(int encode, const CELTMode *m, int start, int end,
celt_norm * X, celt_norm * Y, unsigned char *collapse_masks, const celt_ener *bandE, int *pulses,
int shortBlocks, int spread, int dual_stereo, int intensity, int *tf_res,
opus_int32 total_bits, opus_int32 balance, ec_ctx *ec, int M, int codedBands, opus_uint32 *seed);
void anti_collapse(const CELTMode *m, celt_norm *X_, unsigned char *collapse_masks, int LM, int C, int size,
int start, int end, opus_val16 *logE, opus_val16 *prev1logE,
opus_val16 *prev2logE, int *pulses, opus_uint32 seed);
opus_uint32 celt_lcg_rand(opus_uint32 seed);
int hysteresis_decision(opus_val16 val, const opus_val16 *thresholds, const opus_val16 *hysteresis, int N, int prev);
#endif /* BANDS_H */

223
drivers/opus/celt/celt.c Normal file
View file

@ -0,0 +1,223 @@
/* Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2010 Xiph.Org Foundation
Copyright (c) 2008 Gregory Maxwell
Written by Jean-Marc Valin and Gregory Maxwell */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#define CELT_C
#include "os_support.h"
#include "mdct.h"
#include <math.h>
#include "celt.h"
#include "pitch.h"
#include "bands.h"
#include "opus_modes.h"
#include "entcode.h"
#include "quant_bands.h"
#include "rate.h"
#include "stack_alloc.h"
#include "mathops.h"
#include "float_cast.h"
#include <stdarg.h>
#include "celt_lpc.h"
#include "vq.h"
#ifndef PACKAGE_VERSION
#define PACKAGE_VERSION "unknown"
#endif
int resampling_factor(opus_int32 rate)
{
int ret;
switch (rate)
{
case 48000:
ret = 1;
break;
case 24000:
ret = 2;
break;
case 16000:
ret = 3;
break;
case 12000:
ret = 4;
break;
case 8000:
ret = 6;
break;
default:
#ifndef CUSTOM_MODES
celt_assert(0);
#endif
ret = 0;
break;
}
return ret;
}
#ifndef OVERRIDE_COMB_FILTER_CONST
static void comb_filter_const(opus_val32 *y, opus_val32 *x, int T, int N,
opus_val16 g10, opus_val16 g11, opus_val16 g12)
{
opus_val32 x0, x1, x2, x3, x4;
int i;
x4 = x[-T-2];
x3 = x[-T-1];
x2 = x[-T];
x1 = x[-T+1];
for (i=0;i<N;i++)
{
x0=x[i-T+2];
y[i] = x[i]
+ MULT16_32_Q15(g10,x2)
+ MULT16_32_Q15(g11,ADD32(x1,x3))
+ MULT16_32_Q15(g12,ADD32(x0,x4));
x4=x3;
x3=x2;
x2=x1;
x1=x0;
}
}
#endif
void comb_filter(opus_val32 *y, opus_val32 *x, int T0, int T1, int N,
opus_val16 g0, opus_val16 g1, int tapset0, int tapset1,
const opus_val16 *window, int overlap)
{
int i;
/* printf ("%d %d %f %f\n", T0, T1, g0, g1); */
opus_val16 g00, g01, g02, g10, g11, g12;
opus_val32 x0, x1, x2, x3, x4;
static const opus_val16 gains[3][3] = {
{QCONST16(0.3066406250f, 15), QCONST16(0.2170410156f, 15), QCONST16(0.1296386719f, 15)},
{QCONST16(0.4638671875f, 15), QCONST16(0.2680664062f, 15), QCONST16(0.f, 15)},
{QCONST16(0.7998046875f, 15), QCONST16(0.1000976562f, 15), QCONST16(0.f, 15)}};
if (g0==0 && g1==0)
{
/* OPT: Happens to work without the OPUS_MOVE(), but only because the current encoder already copies x to y */
if (x!=y)
OPUS_MOVE(y, x, N);
return;
}
g00 = MULT16_16_Q15(g0, gains[tapset0][0]);
g01 = MULT16_16_Q15(g0, gains[tapset0][1]);
g02 = MULT16_16_Q15(g0, gains[tapset0][2]);
g10 = MULT16_16_Q15(g1, gains[tapset1][0]);
g11 = MULT16_16_Q15(g1, gains[tapset1][1]);
g12 = MULT16_16_Q15(g1, gains[tapset1][2]);
x1 = x[-T1+1];
x2 = x[-T1 ];
x3 = x[-T1-1];
x4 = x[-T1-2];
for (i=0;i<overlap;i++)
{
opus_val16 f;
x0=x[i-T1+2];
f = MULT16_16_Q15(window[i],window[i]);
y[i] = x[i]
+ MULT16_32_Q15(MULT16_16_Q15((Q15ONE-f),g00),x[i-T0])
+ MULT16_32_Q15(MULT16_16_Q15((Q15ONE-f),g01),ADD32(x[i-T0+1],x[i-T0-1]))
+ MULT16_32_Q15(MULT16_16_Q15((Q15ONE-f),g02),ADD32(x[i-T0+2],x[i-T0-2]))
+ MULT16_32_Q15(MULT16_16_Q15(f,g10),x2)
+ MULT16_32_Q15(MULT16_16_Q15(f,g11),ADD32(x1,x3))
+ MULT16_32_Q15(MULT16_16_Q15(f,g12),ADD32(x0,x4));
x4=x3;
x3=x2;
x2=x1;
x1=x0;
}
if (g1==0)
{
/* OPT: Happens to work without the OPUS_MOVE(), but only because the current encoder already copies x to y */
if (x!=y)
OPUS_MOVE(y+overlap, x+overlap, N-overlap);
return;
}
/* Compute the part with the constant filter. */
comb_filter_const(y+i, x+i, T1, N-i, g10, g11, g12);
}
const signed char tf_select_table[4][8] = {
{0, -1, 0, -1, 0,-1, 0,-1},
{0, -1, 0, -2, 1, 0, 1,-1},
{0, -2, 0, -3, 2, 0, 1,-1},
{0, -2, 0, -3, 3, 0, 1,-1},
};
void init_caps(const CELTMode *m,int *cap,int LM,int C)
{
int i;
for (i=0;i<m->nbEBands;i++)
{
int N;
N=(m->eBands[i+1]-m->eBands[i])<<LM;
cap[i] = (m->cache.caps[m->nbEBands*(2*LM+C-1)+i]+64)*C*N>>2;
}
}
const char *opus_strerror(int error)
{
static const char * const error_strings[8] = {
"success",
"invalid argument",
"buffer too small",
"internal error",
"corrupted stream",
"request not implemented",
"invalid state",
"memory allocation failed"
};
if (error > 0 || error < -7)
return "unknown error";
else
return error_strings[-error];
}
const char *opus_get_version_string(void)
{
return "libopus " PACKAGE_VERSION
#ifdef OPUS_FIXED_POINT
"-fixed"
#endif
#ifdef FUZZING
"-fuzzing"
#endif
;
}

218
drivers/opus/celt/celt.h Normal file
View file

@ -0,0 +1,218 @@
/* Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2009 Xiph.Org Foundation
Copyright (c) 2008 Gregory Maxwell
Written by Jean-Marc Valin and Gregory Maxwell */
/**
@file celt.h
@brief Contains all the functions for encoding and decoding audio
*/
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef CELT_H
#define CELT_H
#include "opus_types.h"
#include "opus_defines.h"
#include "opus_custom.h"
#include "entenc.h"
#include "entdec.h"
#include "arch.h"
#ifdef __cplusplus
extern "C" {
#endif
#define CELTEncoder OpusCustomEncoder
#define CELTDecoder OpusCustomDecoder
#define CELTMode OpusCustomMode
typedef struct {
int valid;
float tonality;
float tonality_slope;
float noisiness;
float activity;
float music_prob;
int bandwidth;
}AnalysisInfo;
#define __celt_check_mode_ptr_ptr(ptr) ((ptr) + ((ptr) - (const CELTMode**)(ptr)))
#define __celt_check_analysis_ptr(ptr) ((ptr) + ((ptr) - (const AnalysisInfo*)(ptr)))
/* Encoder/decoder Requests */
/* Expose this option again when variable framesize actually works */
#define OPUS_FRAMESIZE_VARIABLE 5010 /**< Optimize the frame size dynamically */
#define CELT_SET_PREDICTION_REQUEST 10002
/** Controls the use of interframe prediction.
0=Independent frames
1=Short term interframe prediction allowed
2=Long term prediction allowed
*/
#define CELT_SET_PREDICTION(x) CELT_SET_PREDICTION_REQUEST, __opus_check_int(x)
#define CELT_SET_INPUT_CLIPPING_REQUEST 10004
#define CELT_SET_INPUT_CLIPPING(x) CELT_SET_INPUT_CLIPPING_REQUEST, __opus_check_int(x)
#define CELT_GET_AND_CLEAR_ERROR_REQUEST 10007
#define CELT_GET_AND_CLEAR_ERROR(x) CELT_GET_AND_CLEAR_ERROR_REQUEST, __opus_check_int_ptr(x)
#define CELT_SET_CHANNELS_REQUEST 10008
#define CELT_SET_CHANNELS(x) CELT_SET_CHANNELS_REQUEST, __opus_check_int(x)
/* Internal */
#define CELT_SET_START_BAND_REQUEST 10010
#define CELT_SET_START_BAND(x) CELT_SET_START_BAND_REQUEST, __opus_check_int(x)
#define CELT_SET_END_BAND_REQUEST 10012
#define CELT_SET_END_BAND(x) CELT_SET_END_BAND_REQUEST, __opus_check_int(x)
#define CELT_GET_MODE_REQUEST 10015
/** Get the CELTMode used by an encoder or decoder */
#define CELT_GET_MODE(x) CELT_GET_MODE_REQUEST, __celt_check_mode_ptr_ptr(x)
#define CELT_SET_SIGNALLING_REQUEST 10016
#define CELT_SET_SIGNALLING(x) CELT_SET_SIGNALLING_REQUEST, __opus_check_int(x)
#define CELT_SET_TONALITY_REQUEST 10018
#define CELT_SET_TONALITY(x) CELT_SET_TONALITY_REQUEST, __opus_check_int(x)
#define CELT_SET_TONALITY_SLOPE_REQUEST 10020
#define CELT_SET_TONALITY_SLOPE(x) CELT_SET_TONALITY_SLOPE_REQUEST, __opus_check_int(x)
#define CELT_SET_ANALYSIS_REQUEST 10022
#define CELT_SET_ANALYSIS(x) CELT_SET_ANALYSIS_REQUEST, __celt_check_analysis_ptr(x)
#define OPUS_SET_LFE_REQUEST 10024
#define OPUS_SET_LFE(x) OPUS_SET_LFE_REQUEST, __opus_check_int(x)
#define OPUS_SET_ENERGY_MASK_REQUEST 10026
#define OPUS_SET_ENERGY_MASK(x) OPUS_SET_ENERGY_MASK_REQUEST, __opus_check_val16_ptr(x)
/* Encoder stuff */
int celt_encoder_get_size(int channels);
int celt_encode_with_ec(OpusCustomEncoder * OPUS_RESTRICT st, const opus_val16 * pcm, int frame_size, unsigned char *compressed, int nbCompressedBytes, ec_enc *enc);
int celt_encoder_init(CELTEncoder *st, opus_int32 sampling_rate, int channels,
int arch);
/* Decoder stuff */
int celt_decoder_get_size(int channels);
int celt_decoder_init(CELTDecoder *st, opus_int32 sampling_rate, int channels);
int celt_decode_with_ec(OpusCustomDecoder * OPUS_RESTRICT st, const unsigned char *data, int len, opus_val16 * OPUS_RESTRICT pcm, int frame_size, ec_dec *dec);
#define celt_encoder_ctl opus_custom_encoder_ctl
#define celt_decoder_ctl opus_custom_decoder_ctl
#ifdef CUSTOM_MODES
#define OPUS_CUSTOM_NOSTATIC
#else
#define OPUS_CUSTOM_NOSTATIC static OPUS_INLINE
#endif
static const unsigned char trim_icdf[11] = {126, 124, 119, 109, 87, 41, 19, 9, 4, 2, 0};
/* Probs: NONE: 21.875%, LIGHT: 6.25%, NORMAL: 65.625%, AGGRESSIVE: 6.25% */
static const unsigned char spread_icdf[4] = {25, 23, 2, 0};
static const unsigned char tapset_icdf[3]={2,1,0};
#ifdef CUSTOM_MODES
static const unsigned char toOpusTable[20] = {
0xE0, 0xE8, 0xF0, 0xF8,
0xC0, 0xC8, 0xD0, 0xD8,
0xA0, 0xA8, 0xB0, 0xB8,
0x00, 0x00, 0x00, 0x00,
0x80, 0x88, 0x90, 0x98,
};
static const unsigned char fromOpusTable[16] = {
0x80, 0x88, 0x90, 0x98,
0x40, 0x48, 0x50, 0x58,
0x20, 0x28, 0x30, 0x38,
0x00, 0x08, 0x10, 0x18
};
static OPUS_INLINE int toOpus(unsigned char c)
{
int ret=0;
if (c<0xA0)
ret = toOpusTable[c>>3];
if (ret == 0)
return -1;
else
return ret|(c&0x7);
}
static OPUS_INLINE int fromOpus(unsigned char c)
{
if (c<0x80)
return -1;
else
return fromOpusTable[(c>>3)-16] | (c&0x7);
}
#endif /* CUSTOM_MODES */
#define COMBFILTER_MAXPERIOD 1024
#define COMBFILTER_MINPERIOD 15
extern const signed char tf_select_table[4][8];
int resampling_factor(opus_int32 rate);
void celt_preemphasis(const opus_val16 * OPUS_RESTRICT pcmp, celt_sig * OPUS_RESTRICT inp,
int N, int CC, int upsample, const opus_val16 *coef, celt_sig *mem, int clip);
void comb_filter(opus_val32 *y, opus_val32 *x, int T0, int T1, int N,
opus_val16 g0, opus_val16 g1, int tapset0, int tapset1,
const opus_val16 *window, int overlap);
void init_caps(const CELTMode *m,int *cap,int LM,int C);
#ifdef RESYNTH
void deemphasis(celt_sig *in[], opus_val16 *pcm, int N, int C, int downsample, const opus_val16 *coef, celt_sig *mem, celt_sig * OPUS_RESTRICT scratch);
void compute_inv_mdcts(const CELTMode *mode, int shortBlocks, celt_sig *X,
celt_sig * OPUS_RESTRICT out_mem[], int C, int LM);
#endif
#ifdef __cplusplus
}
#endif
#endif /* CELT_H */

1195
drivers/opus/celt/celt_decoder.c Normal file
View file

File diff suppressed because it is too large Load diff

2353
drivers/opus/celt/celt_encoder.c Normal file
View file

File diff suppressed because it is too large Load diff

309
drivers/opus/celt/celt_lpc.c Normal file
View file

@ -0,0 +1,309 @@
/* Copyright (c) 2009-2010 Xiph.Org Foundation
Written by Jean-Marc Valin */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#include "celt_lpc.h"
#include "stack_alloc.h"
#include "mathops.h"
#include "pitch.h"
void _celt_lpc(
opus_val16 *_lpc, /* out: [0...p-1] LPC coefficients */
const opus_val32 *ac, /* in: [0...p] autocorrelation values */
int p
)
{
int i, j;
opus_val32 r;
opus_val32 error = ac[0];
#ifdef OPUS_FIXED_POINT
opus_val32 lpc[LPC_ORDER];
#else
float *lpc = _lpc;
#endif
for (i = 0; i < p; i++)
lpc[i] = 0;
if (ac[0] != 0)
{
for (i = 0; i < p; i++) {
/* Sum up this iteration's reflection coefficient */
opus_val32 rr = 0;
for (j = 0; j < i; j++)
rr += MULT32_32_Q31(lpc[j],ac[i - j]);
rr += SHR32(ac[i + 1],3);
r = -frac_div32(SHL32(rr,3), error);
/* Update LPC coefficients and total error */
lpc[i] = SHR32(r,3);
for (j = 0; j < (i+1)>>1; j++)
{
opus_val32 tmp1, tmp2;
tmp1 = lpc[j];
tmp2 = lpc[i-1-j];
lpc[j] = tmp1 + MULT32_32_Q31(r,tmp2);
lpc[i-1-j] = tmp2 + MULT32_32_Q31(r,tmp1);
}
error = error - MULT32_32_Q31(MULT32_32_Q31(r,r),error);
/* Bail out once we get 30 dB gain */
#ifdef OPUS_FIXED_POINT
if (error<SHR32(ac[0],10))
break;
#else
if (error<.001f*ac[0])
break;
#endif
}
}
#ifdef OPUS_FIXED_POINT
for (i=0;i<p;i++)
_lpc[i] = ROUND16(lpc[i],16);
#endif
}
void celt_fir(const opus_val16 *_x,
const opus_val16 *num,
opus_val16 *_y,
int N,
int ord,
opus_val16 *mem)
{
int i,j;
VARDECL(opus_val16, rnum);
VARDECL(opus_val16, x);
SAVE_STACK;
ALLOC(rnum, ord, opus_val16);
ALLOC(x, N+ord, opus_val16);
for(i=0;i<ord;i++)
rnum[i] = num[ord-i-1];
for(i=0;i<ord;i++)
x[i] = mem[ord-i-1];
for (i=0;i<N;i++)
x[i+ord]=_x[i];
for(i=0;i<ord;i++)
mem[i] = _x[N-i-1];
#ifdef SMALL_FOOTPRINT
for (i=0;i<N;i++)
{
opus_val32 sum = SHL32(EXTEND32(_x[i]), SIG_SHIFT);
for (j=0;j<ord;j++)
{
sum = MAC16_16(sum,rnum[j],x[i+j]);
}
_y[i] = SATURATE16(PSHR32(sum, SIG_SHIFT));
}
#else
for (i=0;i<N-3;i+=4)
{
opus_val32 sum[4]={0,0,0,0};
xcorr_kernel(rnum, x+i, sum, ord);
_y[i ] = SATURATE16(ADD32(EXTEND32(_x[i ]), PSHR32(sum[0], SIG_SHIFT)));
_y[i+1] = SATURATE16(ADD32(EXTEND32(_x[i+1]), PSHR32(sum[1], SIG_SHIFT)));
_y[i+2] = SATURATE16(ADD32(EXTEND32(_x[i+2]), PSHR32(sum[2], SIG_SHIFT)));
_y[i+3] = SATURATE16(ADD32(EXTEND32(_x[i+3]), PSHR32(sum[3], SIG_SHIFT)));
}
for (;i<N;i++)
{
opus_val32 sum = 0;
for (j=0;j<ord;j++)
sum = MAC16_16(sum,rnum[j],x[i+j]);
_y[i] = SATURATE16(ADD32(EXTEND32(_x[i]), PSHR32(sum, SIG_SHIFT)));
}
#endif
RESTORE_STACK;
}
void celt_iir(const opus_val32 *_x,
const opus_val16 *den,
opus_val32 *_y,
int N,
int ord,
opus_val16 *mem)
{
#ifdef SMALL_FOOTPRINT
int i,j;
for (i=0;i<N;i++)
{
opus_val32 sum = _x[i];
for (j=0;j<ord;j++)
{
sum -= MULT16_16(den[j],mem[j]);
}
for (j=ord-1;j>=1;j--)
{
mem[j]=mem[j-1];
}
mem[0] = ROUND16(sum,SIG_SHIFT);
_y[i] = sum;
}
#else
int i,j;
VARDECL(opus_val16, rden);
VARDECL(opus_val16, y);
SAVE_STACK;
celt_assert((ord&3)==0);
ALLOC(rden, ord, opus_val16);
ALLOC(y, N+ord, opus_val16);
for(i=0;i<ord;i++)
rden[i] = den[ord-i-1];
for(i=0;i<ord;i++)
y[i] = -mem[ord-i-1];
for(;i<N+ord;i++)
y[i]=0;
for (i=0;i<N-3;i+=4)
{
/* Unroll by 4 as if it were an FIR filter */
opus_val32 sum[4];
sum[0]=_x[i];
sum[1]=_x[i+1];
sum[2]=_x[i+2];
sum[3]=_x[i+3];
xcorr_kernel(rden, y+i, sum, ord);
/* Patch up the result to compensate for the fact that this is an IIR */
y[i+ord ] = -ROUND16(sum[0],SIG_SHIFT);
_y[i ] = sum[0];
sum[1] = MAC16_16(sum[1], y[i+ord ], den[0]);
y[i+ord+1] = -ROUND16(sum[1],SIG_SHIFT);
_y[i+1] = sum[1];
sum[2] = MAC16_16(sum[2], y[i+ord+1], den[0]);
sum[2] = MAC16_16(sum[2], y[i+ord ], den[1]);
y[i+ord+2] = -ROUND16(sum[2],SIG_SHIFT);
_y[i+2] = sum[2];
sum[3] = MAC16_16(sum[3], y[i+ord+2], den[0]);
sum[3] = MAC16_16(sum[3], y[i+ord+1], den[1]);
sum[3] = MAC16_16(sum[3], y[i+ord ], den[2]);
y[i+ord+3] = -ROUND16(sum[3],SIG_SHIFT);
_y[i+3] = sum[3];
}
for (;i<N;i++)
{
opus_val32 sum = _x[i];
for (j=0;j<ord;j++)
sum -= MULT16_16(rden[j],y[i+j]);
y[i+ord] = ROUND16(sum,SIG_SHIFT);
_y[i] = sum;
}
for(i=0;i<ord;i++)
mem[i] = _y[N-i-1];
RESTORE_STACK;
#endif
}
int _celt_autocorr(
const opus_val16 *x, /* in: [0...n-1] samples x */
opus_val32 *ac, /* out: [0...lag-1] ac values */
const opus_val16 *window,
int overlap,
int lag,
int n,
int arch
)
{
opus_val32 d;
int i, k;
int fastN=n-lag;
int shift;
const opus_val16 *xptr;
VARDECL(opus_val16, xx);
SAVE_STACK;
ALLOC(xx, n, opus_val16);
celt_assert(n>0);
celt_assert(overlap>=0);
if (overlap == 0)
{
xptr = x;
} else {
for (i=0;i<n;i++)
xx[i] = x[i];
for (i=0;i<overlap;i++)
{
xx[i] = MULT16_16_Q15(x[i],window[i]);
xx[n-i-1] = MULT16_16_Q15(x[n-i-1],window[i]);
}
xptr = xx;
}
shift=0;
#ifdef OPUS_FIXED_POINT
{
opus_val32 ac0;
ac0 = 1+(n<<7);
if (n&1) ac0 += SHR32(MULT16_16(xptr[0],xptr[0]),9);
for(i=(n&1);i<n;i+=2)
{
ac0 += SHR32(MULT16_16(xptr[i],xptr[i]),9);
ac0 += SHR32(MULT16_16(xptr[i+1],xptr[i+1]),9);
}
shift = celt_ilog2(ac0)-30+10;
shift = (shift)/2;
if (shift>0)
{
for(i=0;i<n;i++)
xx[i] = PSHR32(xptr[i], shift);
xptr = xx;
} else
shift = 0;
}
#endif
celt_pitch_xcorr(xptr, xptr, ac, fastN, lag+1, arch);
for (k=0;k<=lag;k++)
{
for (i = k+fastN, d = 0; i < n; i++)
d = MAC16_16(d, xptr[i], xptr[i-k]);
ac[k] += d;
}
#ifdef OPUS_FIXED_POINT
shift = 2*shift;
if (shift<=0)
ac[0] += SHL32((opus_int32)1, -shift);
if (ac[0] < 268435456)
{
int shift2 = 29 - EC_ILOG(ac[0]);
for (i=0;i<=lag;i++)
ac[i] = SHL32(ac[i], shift2);
shift -= shift2;
} else if (ac[0] >= 536870912)
{
int shift2=1;
if (ac[0] >= 1073741824)
shift2++;
for (i=0;i<=lag;i++)
ac[i] = SHR32(ac[i], shift2);
shift += shift2;
}
#endif
RESTORE_STACK;
return shift;
}

54
drivers/opus/celt/celt_lpc.h Normal file
View file

@ -0,0 +1,54 @@
/* Copyright (c) 2009-2010 Xiph.Org Foundation
Written by Jean-Marc Valin */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef PLC_H
#define PLC_H
#include "arch.h"
#define LPC_ORDER 24
void _celt_lpc(opus_val16 *_lpc, const opus_val32 *ac, int p);
void celt_fir(const opus_val16 *x,
const opus_val16 *num,
opus_val16 *y,
int N,
int ord,
opus_val16 *mem);
void celt_iir(const opus_val32 *x,
const opus_val16 *den,
opus_val32 *y,
int N,
int ord,
opus_val16 *mem);
int _celt_autocorr(const opus_val16 *x, opus_val32 *ac,
const opus_val16 *window, int overlap, int lag, int n, int arch);
#endif /* PLC_H */

54
drivers/opus/celt/cpu_support.h Normal file
View file

@ -0,0 +1,54 @@
/* Copyright (c) 2010 Xiph.Org Foundation
* Copyright (c) 2013 Parrot */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef CPU_SUPPORT_H
#define CPU_SUPPORT_H
#include "opus_types.h"
#include "opus_defines.h"
#if defined(OPUS_HAVE_RTCD) && defined(OPUS_ARM_ASM)
#include "arm/armcpu.h"
/* We currently support 4 ARM variants:
* arch[0] -> ARMv4
* arch[1] -> ARMv5E
* arch[2] -> ARMv6
* arch[3] -> NEON
*/
#define OPUS_ARCHMASK 3
#else
#define OPUS_ARCHMASK 0
static OPUS_INLINE int opus_select_arch(void)
{
return 0;
}
#endif
#endif

697
drivers/opus/celt/cwrs.c Normal file
View file

@ -0,0 +1,697 @@
/* Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2009 Xiph.Org Foundation
Copyright (c) 2007-2009 Timothy B. Terriberry
Written by Timothy B. Terriberry and Jean-Marc Valin */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#include "os_support.h"
#include "cwrs.h"
#include "mathops.h"
#include "arch.h"
#ifdef CUSTOM_MODES
/*Guaranteed to return a conservatively large estimate of the binary logarithm
with frac bits of fractional precision.
Tested for all possible 32-bit inputs with frac=4, where the maximum
overestimation is 0.06254243 bits.*/
int log2_frac(opus_uint32 val, int frac)
{
int l;
l=EC_ILOG(val);
if(val&(val-1)){
/*This is (val>>l-16), but guaranteed to round up, even if adding a bias
before the shift would cause overflow (e.g., for 0xFFFFxxxx).
Doesn't work for val=0, but that case fails the test above.*/
if(l>16)val=((val-1)>>(l-16))+1;
else val<<=16-l;
l=(l-1)<<frac;
/*Note that we always need one iteration, since the rounding up above means
that we might need to adjust the integer part of the logarithm.*/
do{
int b;
b=(int)(val>>16);
l+=b<<frac;
val=(val+b)>>b;
val=(val*val+0x7FFF)>>15;
}
while(frac-->0);
/*If val is not exactly 0x8000, then we have to round up the remainder.*/
return l+(val>0x8000);
}
/*Exact powers of two require no rounding.*/
else return (l-1)<<frac;
}
#endif
/*Although derived separately, the pulse vector coding scheme is equivalent to
a Pyramid Vector Quantizer \cite{Fis86}.
Some additional notes about an early version appear at
http://people.xiph.org/~tterribe/notes/cwrs.html, but the codebook ordering
and the definitions of some terms have evolved since that was written.
The conversion from a pulse vector to an integer index (encoding) and back
(decoding) is governed by two related functions, V(N,K) and U(N,K).
V(N,K) = the number of combinations, with replacement, of N items, taken K
at a time, when a sign bit is added to each item taken at least once (i.e.,
the number of N-dimensional unit pulse vectors with K pulses).
One way to compute this is via
V(N,K) = K>0 ? sum(k=1...K,2**k*choose(N,k)*choose(K-1,k-1)) : 1,
where choose() is the binomial function.
A table of values for N<10 and K<10 looks like:
V[10][10] = {
{1, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{1, 2, 2, 2, 2, 2, 2, 2, 2, 2},
{1, 4, 8, 12, 16, 20, 24, 28, 32, 36},
{1, 6, 18, 38, 66, 102, 146, 198, 258, 326},
{1, 8, 32, 88, 192, 360, 608, 952, 1408, 1992},
{1, 10, 50, 170, 450, 1002, 1970, 3530, 5890, 9290},
{1, 12, 72, 292, 912, 2364, 5336, 10836, 20256, 35436},
{1, 14, 98, 462, 1666, 4942, 12642, 28814, 59906, 115598},
{1, 16, 128, 688, 2816, 9424, 27008, 68464, 157184, 332688},
{1, 18, 162, 978, 4482, 16722, 53154, 148626, 374274, 864146}
};
U(N,K) = the number of such combinations wherein N-1 objects are taken at
most K-1 at a time.
This is given by
U(N,K) = sum(k=0...K-1,V(N-1,k))
= K>0 ? (V(N-1,K-1) + V(N,K-1))/2 : 0.
The latter expression also makes clear that U(N,K) is half the number of such
combinations wherein the first object is taken at least once.
Although it may not be clear from either of these definitions, U(N,K) is the
natural function to work with when enumerating the pulse vector codebooks,
not V(N,K).
U(N,K) is not well-defined for N=0, but with the extension
U(0,K) = K>0 ? 0 : 1,
the function becomes symmetric: U(N,K) = U(K,N), with a similar table:
U[10][10] = {
{1, 0, 0, 0, 0, 0, 0, 0, 0, 0},
{0, 1, 1, 1, 1, 1, 1, 1, 1, 1},
{0, 1, 3, 5, 7, 9, 11, 13, 15, 17},
{0, 1, 5, 13, 25, 41, 61, 85, 113, 145},
{0, 1, 7, 25, 63, 129, 231, 377, 575, 833},
{0, 1, 9, 41, 129, 321, 681, 1289, 2241, 3649},
{0, 1, 11, 61, 231, 681, 1683, 3653, 7183, 13073},
{0, 1, 13, 85, 377, 1289, 3653, 8989, 19825, 40081},
{0, 1, 15, 113, 575, 2241, 7183, 19825, 48639, 108545},
{0, 1, 17, 145, 833, 3649, 13073, 40081, 108545, 265729}
};
With this extension, V(N,K) may be written in terms of U(N,K):
V(N,K) = U(N,K) + U(N,K+1)
for all N>=0, K>=0.
Thus U(N,K+1) represents the number of combinations where the first element
is positive or zero, and U(N,K) represents the number of combinations where
it is negative.
With a large enough table of U(N,K) values, we could write O(N) encoding
and O(min(N*log(K),N+K)) decoding routines, but such a table would be
prohibitively large for small embedded devices (K may be as large as 32767
for small N, and N may be as large as 200).
Both functions obey the same recurrence relation:
V(N,K) = V(N-1,K) + V(N,K-1) + V(N-1,K-1),
U(N,K) = U(N-1,K) + U(N,K-1) + U(N-1,K-1),
for all N>0, K>0, with different initial conditions at N=0 or K=0.
This allows us to construct a row of one of the tables above given the
previous row or the next row.
Thus we can derive O(NK) encoding and decoding routines with O(K) memory
using only addition and subtraction.
When encoding, we build up from the U(2,K) row and work our way forwards.
When decoding, we need to start at the U(N,K) row and work our way backwards,
which requires a means of computing U(N,K).
U(N,K) may be computed from two previous values with the same N:
U(N,K) = ((2*N-1)*U(N,K-1) - U(N,K-2))/(K-1) + U(N,K-2)
for all N>1, and since U(N,K) is symmetric, a similar relation holds for two
previous values with the same K:
U(N,K>1) = ((2*K-1)*U(N-1,K) - U(N-2,K))/(N-1) + U(N-2,K)
for all K>1.
This allows us to construct an arbitrary row of the U(N,K) table by starting
with the first two values, which are constants.
This saves roughly 2/3 the work in our O(NK) decoding routine, but costs O(K)
multiplications.
Similar relations can be derived for V(N,K), but are not used here.
For N>0 and K>0, U(N,K) and V(N,K) take on the form of an (N-1)-degree
polynomial for fixed N.
The first few are
U(1,K) = 1,
U(2,K) = 2*K-1,
U(3,K) = (2*K-2)*K+1,
U(4,K) = (((4*K-6)*K+8)*K-3)/3,
U(5,K) = ((((2*K-4)*K+10)*K-8)*K+3)/3,
and
V(1,K) = 2,
V(2,K) = 4*K,
V(3,K) = 4*K*K+2,
V(4,K) = 8*(K*K+2)*K/3,
V(5,K) = ((4*K*K+20)*K*K+6)/3,
for all K>0.
This allows us to derive O(N) encoding and O(N*log(K)) decoding routines for
small N (and indeed decoding is also O(N) for N<3).
@ARTICLE{Fis86,
author="Thomas R. Fischer",
title="A Pyramid Vector Quantizer",
journal="IEEE Transactions on Information Theory",
volume="IT-32",
number=4,
pages="568--583",
month=Jul,
year=1986
}*/
#if !defined(SMALL_FOOTPRINT)
/*U(N,K) = U(K,N) := N>0?K>0?U(N-1,K)+U(N,K-1)+U(N-1,K-1):0:K>0?1:0*/
# define CELT_PVQ_U(_n,_k) (CELT_PVQ_U_ROW[IMIN(_n,_k)][IMAX(_n,_k)])
/*V(N,K) := U(N,K)+U(N,K+1) = the number of PVQ codewords for a band of size N
with K pulses allocated to it.*/
# define CELT_PVQ_V(_n,_k) (CELT_PVQ_U(_n,_k)+CELT_PVQ_U(_n,(_k)+1))
/*For each V(N,K) supported, we will access element U(min(N,K+1),max(N,K+1)).
Thus, the number of entries in row I is the larger of the maximum number of
pulses we will ever allocate for a given N=I (K=128, or however many fit in
32 bits, whichever is smaller), plus one, and the maximum N for which
K=I-1 pulses fit in 32 bits.
The largest band size in an Opus Custom mode is 208.
Otherwise, we can limit things to the set of N which can be achieved by
splitting a band from a standard Opus mode: 176, 144, 96, 88, 72, 64, 48,
44, 36, 32, 24, 22, 18, 16, 8, 4, 2).*/
#if defined(CUSTOM_MODES)
static const opus_uint32 CELT_PVQ_U_DATA[1488]={
#else
static const opus_uint32 CELT_PVQ_U_DATA[1272]={
#endif
/*N=0, K=0...176:*/
1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
#if defined(CUSTOM_MODES)
/*...208:*/
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0,
#endif
/*N=1, K=1...176:*/
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
#if defined(CUSTOM_MODES)
/*...208:*/
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1,
#endif
/*N=2, K=2...176:*/
3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29, 31, 33, 35, 37, 39, 41,
43, 45, 47, 49, 51, 53, 55, 57, 59, 61, 63, 65, 67, 69, 71, 73, 75, 77, 79,
81, 83, 85, 87, 89, 91, 93, 95, 97, 99, 101, 103, 105, 107, 109, 111, 113,
115, 117, 119, 121, 123, 125, 127, 129, 131, 133, 135, 137, 139, 141, 143,
145, 147, 149, 151, 153, 155, 157, 159, 161, 163, 165, 167, 169, 171, 173,
175, 177, 179, 181, 183, 185, 187, 189, 191, 193, 195, 197, 199, 201, 203,
205, 207, 209, 211, 213, 215, 217, 219, 221, 223, 225, 227, 229, 231, 233,
235, 237, 239, 241, 243, 245, 247, 249, 251, 253, 255, 257, 259, 261, 263,
265, 267, 269, 271, 273, 275, 277, 279, 281, 283, 285, 287, 289, 291, 293,
295, 297, 299, 301, 303, 305, 307, 309, 311, 313, 315, 317, 319, 321, 323,
325, 327, 329, 331, 333, 335, 337, 339, 341, 343, 345, 347, 349, 351,
#if defined(CUSTOM_MODES)
/*...208:*/
353, 355, 357, 359, 361, 363, 365, 367, 369, 371, 373, 375, 377, 379, 381,
383, 385, 387, 389, 391, 393, 395, 397, 399, 401, 403, 405, 407, 409, 411,
413, 415,
#endif
/*N=3, K=3...176:*/
13, 25, 41, 61, 85, 113, 145, 181, 221, 265, 313, 365, 421, 481, 545, 613,
685, 761, 841, 925, 1013, 1105, 1201, 1301, 1405, 1513, 1625, 1741, 1861,
1985, 2113, 2245, 2381, 2521, 2665, 2813, 2965, 3121, 3281, 3445, 3613, 3785,
3961, 4141, 4325, 4513, 4705, 4901, 5101, 5305, 5513, 5725, 5941, 6161, 6385,
6613, 6845, 7081, 7321, 7565, 7813, 8065, 8321, 8581, 8845, 9113, 9385, 9661,
9941, 10225, 10513, 10805, 11101, 11401, 11705, 12013, 12325, 12641, 12961,
13285, 13613, 13945, 14281, 14621, 14965, 15313, 15665, 16021, 16381, 16745,
17113, 17485, 17861, 18241, 18625, 19013, 19405, 19801, 20201, 20605, 21013,
21425, 21841, 22261, 22685, 23113, 23545, 23981, 24421, 24865, 25313, 25765,
26221, 26681, 27145, 27613, 28085, 28561, 29041, 29525, 30013, 30505, 31001,
31501, 32005, 32513, 33025, 33541, 34061, 34585, 35113, 35645, 36181, 36721,
37265, 37813, 38365, 38921, 39481, 40045, 40613, 41185, 41761, 42341, 42925,
43513, 44105, 44701, 45301, 45905, 46513, 47125, 47741, 48361, 48985, 49613,
50245, 50881, 51521, 52165, 52813, 53465, 54121, 54781, 55445, 56113, 56785,
57461, 58141, 58825, 59513, 60205, 60901, 61601,
#if defined(CUSTOM_MODES)
/*...208:*/
62305, 63013, 63725, 64441, 65161, 65885, 66613, 67345, 68081, 68821, 69565,
70313, 71065, 71821, 72581, 73345, 74113, 74885, 75661, 76441, 77225, 78013,
78805, 79601, 80401, 81205, 82013, 82825, 83641, 84461, 85285, 86113,
#endif
/*N=4, K=4...176:*/
63, 129, 231, 377, 575, 833, 1159, 1561, 2047, 2625, 3303, 4089, 4991, 6017,
7175, 8473, 9919, 11521, 13287, 15225, 17343, 19649, 22151, 24857, 27775,
30913, 34279, 37881, 41727, 45825, 50183, 54809, 59711, 64897, 70375, 76153,
82239, 88641, 95367, 102425, 109823, 117569, 125671, 134137, 142975, 152193,
161799, 171801, 182207, 193025, 204263, 215929, 228031, 240577, 253575,
267033, 280959, 295361, 310247, 325625, 341503, 357889, 374791, 392217,
410175, 428673, 447719, 467321, 487487, 508225, 529543, 551449, 573951,
597057, 620775, 645113, 670079, 695681, 721927, 748825, 776383, 804609,
833511, 863097, 893375, 924353, 956039, 988441, 1021567, 1055425, 1090023,
1125369, 1161471, 1198337, 1235975, 1274393, 1313599, 1353601, 1394407,
1436025, 1478463, 1521729, 1565831, 1610777, 1656575, 1703233, 1750759,
1799161, 1848447, 1898625, 1949703, 2001689, 2054591, 2108417, 2163175,
2218873, 2275519, 2333121, 2391687, 2451225, 2511743, 2573249, 2635751,
2699257, 2763775, 2829313, 2895879, 2963481, 3032127, 3101825, 3172583,
3244409, 3317311, 3391297, 3466375, 3542553, 3619839, 3698241, 3777767,
3858425, 3940223, 4023169, 4107271, 4192537, 4278975, 4366593, 4455399,
4545401, 4636607, 4729025, 4822663, 4917529, 5013631, 5110977, 5209575,
5309433, 5410559, 5512961, 5616647, 5721625, 5827903, 5935489, 6044391,
6154617, 6266175, 6379073, 6493319, 6608921, 6725887, 6844225, 6963943,
7085049, 7207551,
#if defined(CUSTOM_MODES)
/*...208:*/
7331457, 7456775, 7583513, 7711679, 7841281, 7972327, 8104825, 8238783,
8374209, 8511111, 8649497, 8789375, 8930753, 9073639, 9218041, 9363967,
9511425, 9660423, 9810969, 9963071, 10116737, 10271975, 10428793, 10587199,
10747201, 10908807, 11072025, 11236863, 11403329, 11571431, 11741177,
11912575,
#endif
/*N=5, K=5...176:*/
321, 681, 1289, 2241, 3649, 5641, 8361, 11969, 16641, 22569, 29961, 39041,
50049, 63241, 78889, 97281, 118721, 143529, 172041, 204609, 241601, 283401,
330409, 383041, 441729, 506921, 579081, 658689, 746241, 842249, 947241,
1061761, 1186369, 1321641, 1468169, 1626561, 1797441, 1981449, 2179241,
2391489, 2618881, 2862121, 3121929, 3399041, 3694209, 4008201, 4341801,
4695809, 5071041, 5468329, 5888521, 6332481, 6801089, 7295241, 7815849,
8363841, 8940161, 9545769, 10181641, 10848769, 11548161, 12280841, 13047849,
13850241, 14689089, 15565481, 16480521, 17435329, 18431041, 19468809,
20549801, 21675201, 22846209, 24064041, 25329929, 26645121, 28010881,
29428489, 30899241, 32424449, 34005441, 35643561, 37340169, 39096641,
40914369, 42794761, 44739241, 46749249, 48826241, 50971689, 53187081,
55473921, 57833729, 60268041, 62778409, 65366401, 68033601, 70781609,
73612041, 76526529, 79526721, 82614281, 85790889, 89058241, 92418049,
95872041, 99421961, 103069569, 106816641, 110664969, 114616361, 118672641,
122835649, 127107241, 131489289, 135983681, 140592321, 145317129, 150160041,
155123009, 160208001, 165417001, 170752009, 176215041, 181808129, 187533321,
193392681, 199388289, 205522241, 211796649, 218213641, 224775361, 231483969,
238341641, 245350569, 252512961, 259831041, 267307049, 274943241, 282741889,
290705281, 298835721, 307135529, 315607041, 324252609, 333074601, 342075401,
351257409, 360623041, 370174729, 379914921, 389846081, 399970689, 410291241,
420810249, 431530241, 442453761, 453583369, 464921641, 476471169, 488234561,
500214441, 512413449, 524834241, 537479489, 550351881, 563454121, 576788929,
590359041, 604167209, 618216201, 632508801,
#if defined(CUSTOM_MODES)
/*...208:*/
647047809, 661836041, 676876329, 692171521, 707724481, 723538089, 739615241,
755958849, 772571841, 789457161, 806617769, 824056641, 841776769, 859781161,
878072841, 896654849, 915530241, 934702089, 954173481, 973947521, 994027329,
1014416041, 1035116809, 1056132801, 1077467201, 1099123209, 1121104041,
1143412929, 1166053121, 1189027881, 1212340489, 1235994241,
#endif
/*N=6, K=6...96:*/
1683, 3653, 7183, 13073, 22363, 36365, 56695, 85305, 124515, 177045, 246047,
335137, 448427, 590557, 766727, 982729, 1244979, 1560549, 1937199, 2383409,
2908411, 3522221, 4235671, 5060441, 6009091, 7095093, 8332863, 9737793,
11326283, 13115773, 15124775, 17372905, 19880915, 22670725, 25765455,
29189457, 32968347, 37129037, 41699767, 46710137, 52191139, 58175189,
64696159, 71789409, 79491819, 87841821, 96879431, 106646281, 117185651,
128542501, 140763503, 153897073, 167993403, 183104493, 199284183, 216588185,
235074115, 254801525, 275831935, 298228865, 322057867, 347386557, 374284647,
402823977, 433078547, 465124549, 499040399, 534906769, 572806619, 612825229,
655050231, 699571641, 746481891, 795875861, 847850911, 902506913, 959946283,
1020274013, 1083597703, 1150027593, 1219676595, 1292660325, 1369097135,
1449108145, 1532817275, 1620351277, 1711839767, 1807415257, 1907213187,
2011371957, 2120032959,
#if defined(CUSTOM_MODES)
/*...109:*/
2233340609U, 2351442379U, 2474488829U, 2602633639U, 2736033641U, 2874848851U,
3019242501U, 3169381071U, 3325434321U, 3487575323U, 3655980493U, 3830829623U,
4012305913U,
#endif
/*N=7, K=7...54*/
8989, 19825, 40081, 75517, 134245, 227305, 369305, 579125, 880685, 1303777,
1884961, 2668525, 3707509, 5064793, 6814249, 9041957, 11847485, 15345233,
19665841, 24957661, 31388293, 39146185, 48442297, 59511829, 72616013,
88043969, 106114625, 127178701, 151620757, 179861305, 212358985, 249612805,
292164445, 340600625, 395555537, 457713341, 527810725, 606639529, 695049433,
793950709, 904317037, 1027188385, 1163673953, 1314955181, 1482288821,
1667010073, 1870535785, 2094367717,
#if defined(CUSTOM_MODES)
/*...60:*/
2340095869U, 2609401873U, 2904062449U, 3225952925U, 3577050821U, 3959439497U,
#endif
/*N=8, K=8...37*/
48639, 108545, 224143, 433905, 795455, 1392065, 2340495, 3800305, 5984767,
9173505, 13726991, 20103025, 28875327, 40754369, 56610575, 77500017,
104692735, 139703809, 184327311, 240673265, 311207743, 398796225, 506750351,
638878193, 799538175, 993696769, 1226990095, 1505789553, 1837271615,
2229491905U,
#if defined(CUSTOM_MODES)
/*...40:*/
2691463695U, 3233240945U, 3866006015U,
#endif
/*N=9, K=9...28:*/
265729, 598417, 1256465, 2485825, 4673345, 8405905, 14546705, 24331777,
39490049, 62390545, 96220561, 145198913, 214828609, 312193553, 446304145,
628496897, 872893441, 1196924561, 1621925137, 2173806145U,
#if defined(CUSTOM_MODES)
/*...29:*/
2883810113U,
#endif
/*N=10, K=10...24:*/
1462563, 3317445, 7059735, 14218905, 27298155, 50250765, 89129247, 152951073,
254831667, 413442773, 654862247, 1014889769, 1541911931, 2300409629U,
3375210671U,
/*N=11, K=11...19:*/
8097453, 18474633, 39753273, 81270333, 158819253, 298199265, 540279585,
948062325, 1616336765,
#if defined(CUSTOM_MODES)
/*...20:*/
2684641785U,
#endif
/*N=12, K=12...18:*/
45046719, 103274625, 224298231, 464387817, 921406335, 1759885185,
3248227095U,
/*N=13, K=13...16:*/
251595969, 579168825, 1267854873, 2653649025U,
/*N=14, K=14:*/
1409933619
};
#if defined(CUSTOM_MODES)
static const opus_uint32 *const CELT_PVQ_U_ROW[15]={
CELT_PVQ_U_DATA+ 0,CELT_PVQ_U_DATA+ 208,CELT_PVQ_U_DATA+ 415,
CELT_PVQ_U_DATA+ 621,CELT_PVQ_U_DATA+ 826,CELT_PVQ_U_DATA+1030,
CELT_PVQ_U_DATA+1233,CELT_PVQ_U_DATA+1336,CELT_PVQ_U_DATA+1389,
CELT_PVQ_U_DATA+1421,CELT_PVQ_U_DATA+1441,CELT_PVQ_U_DATA+1455,
CELT_PVQ_U_DATA+1464,CELT_PVQ_U_DATA+1470,CELT_PVQ_U_DATA+1473
};
#else
static const opus_uint32 *const CELT_PVQ_U_ROW[15]={
CELT_PVQ_U_DATA+ 0,CELT_PVQ_U_DATA+ 176,CELT_PVQ_U_DATA+ 351,
CELT_PVQ_U_DATA+ 525,CELT_PVQ_U_DATA+ 698,CELT_PVQ_U_DATA+ 870,
CELT_PVQ_U_DATA+1041,CELT_PVQ_U_DATA+1131,CELT_PVQ_U_DATA+1178,
CELT_PVQ_U_DATA+1207,CELT_PVQ_U_DATA+1226,CELT_PVQ_U_DATA+1240,
CELT_PVQ_U_DATA+1248,CELT_PVQ_U_DATA+1254,CELT_PVQ_U_DATA+1257
};
#endif
#if defined(CUSTOM_MODES)
void get_required_bits(opus_int16 *_bits,int _n,int _maxk,int _frac){
int k;
/*_maxk==0 => there's nothing to do.*/
celt_assert(_maxk>0);
_bits[0]=0;
for(k=1;k<=_maxk;k++)_bits[k]=log2_frac(CELT_PVQ_V(_n,k),_frac);
}
#endif
static opus_uint32 icwrs(int _n,const int *_y){
opus_uint32 i;
int j;
int k;
celt_assert(_n>=2);
j=_n-1;
i=_y[j]<0;
k=abs(_y[j]);
do{
j--;
i+=CELT_PVQ_U(_n-j,k);
k+=abs(_y[j]);
if(_y[j]<0)i+=CELT_PVQ_U(_n-j,k+1);
}
while(j>0);
return i;
}
void encode_pulses(const int *_y,int _n,int _k,ec_enc *_enc){
celt_assert(_k>0);
ec_enc_uint(_enc,icwrs(_n,_y),CELT_PVQ_V(_n,_k));
}
static void cwrsi(int _n,int _k,opus_uint32 _i,int *_y){
opus_uint32 p;
int s;
int k0;
celt_assert(_k>0);
celt_assert(_n>1);
while(_n>2){
opus_uint32 q;
/*Lots of pulses case:*/
if(_k>=_n){
const opus_uint32 *row;
row=CELT_PVQ_U_ROW[_n];
/*Are the pulses in this dimension negative?*/
p=row[_k+1];
s=-(_i>=p);
_i-=p&s;
/*Count how many pulses were placed in this dimension.*/
k0=_k;
q=row[_n];
if(q>_i){
celt_assert(p>q);
_k=_n;
do p=CELT_PVQ_U_ROW[--_k][_n];
while(p>_i);
}
else for(p=row[_k];p>_i;p=row[_k])_k--;
_i-=p;
*_y++=(k0-_k+s)^s;
}
/*Lots of dimensions case:*/
else{
/*Are there any pulses in this dimension at all?*/
p=CELT_PVQ_U_ROW[_k][_n];
q=CELT_PVQ_U_ROW[_k+1][_n];
if(p<=_i&&_i<q){
_i-=p;
*_y++=0;
}
else{
/*Are the pulses in this dimension negative?*/
s=-(_i>=q);
_i-=q&s;
/*Count how many pulses were placed in this dimension.*/
k0=_k;
do p=CELT_PVQ_U_ROW[--_k][_n];
while(p>_i);
_i-=p;
*_y++=(k0-_k+s)^s;
}
}
_n--;
}
/*_n==2*/
p=2*_k+1;
s=-(_i>=p);
_i-=p&s;
k0=_k;
_k=(_i+1)>>1;
if(_k)_i-=2*_k-1;
*_y++=(k0-_k+s)^s;
/*_n==1*/
s=-(int)_i;
*_y=(_k+s)^s;
}
void decode_pulses(int *_y,int _n,int _k,ec_dec *_dec){
cwrsi(_n,_k,ec_dec_uint(_dec,CELT_PVQ_V(_n,_k)),_y);
}
#else /* SMALL_FOOTPRINT */
/*Computes the next row/column of any recurrence that obeys the relation
u[i][j]=u[i-1][j]+u[i][j-1]+u[i-1][j-1].
_ui0 is the base case for the new row/column.*/
static OPUS_INLINE void unext(opus_uint32 *_ui,unsigned _len,opus_uint32 _ui0){
opus_uint32 ui1;
unsigned j;
/*This do-while will overrun the array if we don't have storage for at least
2 values.*/
j=1; do {
ui1=UADD32(UADD32(_ui[j],_ui[j-1]),_ui0);
_ui[j-1]=_ui0;
_ui0=ui1;
} while (++j<_len);
_ui[j-1]=_ui0;
}
/*Computes the previous row/column of any recurrence that obeys the relation
u[i-1][j]=u[i][j]-u[i][j-1]-u[i-1][j-1].
_ui0 is the base case for the new row/column.*/
static OPUS_INLINE void uprev(opus_uint32 *_ui,unsigned _n,opus_uint32 _ui0){
opus_uint32 ui1;
unsigned j;
/*This do-while will overrun the array if we don't have storage for at least
2 values.*/
j=1; do {
ui1=USUB32(USUB32(_ui[j],_ui[j-1]),_ui0);
_ui[j-1]=_ui0;
_ui0=ui1;
} while (++j<_n);
_ui[j-1]=_ui0;
}
/*Compute V(_n,_k), as well as U(_n,0..._k+1).
_u: On exit, _u[i] contains U(_n,i) for i in [0..._k+1].*/
static opus_uint32 ncwrs_urow(unsigned _n,unsigned _k,opus_uint32 *_u){
opus_uint32 um2;
unsigned len;
unsigned k;
len=_k+2;
/*We require storage at least 3 values (e.g., _k>0).*/
celt_assert(len>=3);
_u[0]=0;
_u[1]=um2=1;
/*If _n==0, _u[0] should be 1 and the rest should be 0.*/
/*If _n==1, _u[i] should be 1 for i>1.*/
celt_assert(_n>=2);
/*If _k==0, the following do-while loop will overflow the buffer.*/
celt_assert(_k>0);
k=2;
do _u[k]=(k<<1)-1;
while(++k<len);
for(k=2;k<_n;k++)unext(_u+1,_k+1,1);
return _u[_k]+_u[_k+1];
}
/*Returns the _i'th combination of _k elements chosen from a set of size _n
with associated sign bits.
_y: Returns the vector of pulses.
_u: Must contain entries [0..._k+1] of row _n of U() on input.
Its contents will be destructively modified.*/
static void cwrsi(int _n,int _k,opus_uint32 _i,int *_y,opus_uint32 *_u){
int j;
celt_assert(_n>0);
j=0;
do{
opus_uint32 p;
int s;
int yj;
p=_u[_k+1];
s=-(_i>=p);
_i-=p&s;
yj=_k;
p=_u[_k];
while(p>_i)p=_u[--_k];
_i-=p;
yj-=_k;
_y[j]=(yj+s)^s;
uprev(_u,_k+2,0);
}
while(++j<_n);
}
/*Returns the index of the given combination of K elements chosen from a set
of size 1 with associated sign bits.
_y: The vector of pulses, whose sum of absolute values is K.
_k: Returns K.*/
static OPUS_INLINE opus_uint32 icwrs1(const int *_y,int *_k){
*_k=abs(_y[0]);
return _y[0]<0;
}
/*Returns the index of the given combination of K elements chosen from a set
of size _n with associated sign bits.
_y: The vector of pulses, whose sum of absolute values must be _k.
_nc: Returns V(_n,_k).*/
static OPUS_INLINE opus_uint32 icwrs(int _n,int _k,opus_uint32 *_nc,const int *_y,
opus_uint32 *_u){
opus_uint32 i;
int j;
int k;
/*We can't unroll the first two iterations of the loop unless _n>=2.*/
celt_assert(_n>=2);
_u[0]=0;
for(k=1;k<=_k+1;k++)_u[k]=(k<<1)-1;
i=icwrs1(_y+_n-1,&k);
j=_n-2;
i+=_u[k];
k+=abs(_y[j]);
if(_y[j]<0)i+=_u[k+1];
while(j-->0){
unext(_u,_k+2,0);
i+=_u[k];
k+=abs(_y[j]);
if(_y[j]<0)i+=_u[k+1];
}
*_nc=_u[k]+_u[k+1];
return i;
}
#ifdef CUSTOM_MODES
void get_required_bits(opus_int16 *_bits,int _n,int _maxk,int _frac){
int k;
/*_maxk==0 => there's nothing to do.*/
celt_assert(_maxk>0);
_bits[0]=0;
if (_n==1)
{
for (k=1;k<=_maxk;k++)
_bits[k] = 1<<_frac;
}
else {
VARDECL(opus_uint32,u);
SAVE_STACK;
ALLOC(u,_maxk+2U,opus_uint32);
ncwrs_urow(_n,_maxk,u);
for(k=1;k<=_maxk;k++)
_bits[k]=log2_frac(u[k]+u[k+1],_frac);
RESTORE_STACK;
}
}
#endif /* CUSTOM_MODES */
void encode_pulses(const int *_y,int _n,int _k,ec_enc *_enc){
opus_uint32 i;
VARDECL(opus_uint32,u);
opus_uint32 nc;
SAVE_STACK;
celt_assert(_k>0);
ALLOC(u,_k+2U,opus_uint32);
i=icwrs(_n,_k,&nc,_y,u);
ec_enc_uint(_enc,i,nc);
RESTORE_STACK;
}
void decode_pulses(int *_y,int _n,int _k,ec_dec *_dec){
VARDECL(opus_uint32,u);
SAVE_STACK;
celt_assert(_k>0);
ALLOC(u,_k+2U,opus_uint32);
cwrsi(_n,_k,ec_dec_uint(_dec,ncwrs_urow(_n,_k,u)),_y,u);
RESTORE_STACK;
}
#endif /* SMALL_FOOTPRINT */

48
drivers/opus/celt/cwrs.h Normal file
View file

@ -0,0 +1,48 @@
/* Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2009 Xiph.Org Foundation
Copyright (c) 2007-2009 Timothy B. Terriberry
Written by Timothy B. Terriberry and Jean-Marc Valin */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef CWRS_H
#define CWRS_H
#include "arch.h"
#include "stack_alloc.h"
#include "entenc.h"
#include "entdec.h"
#ifdef CUSTOM_MODES
int log2_frac(opus_uint32 val, int frac);
#endif
void get_required_bits(opus_int16 *bits, int N, int K, int frac);
void encode_pulses(const int *_y, int N, int K, ec_enc *enc);
void decode_pulses(int *_y, int N, int K, ec_dec *dec);
#endif /* CWRS_H */

87
drivers/opus/celt/ecintrin.h Normal file
View file

@ -0,0 +1,87 @@
/* Copyright (c) 2003-2008 Timothy B. Terriberry
Copyright (c) 2008 Xiph.Org Foundation */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/*Some common macros for potential platform-specific optimization.*/
#include "opus_types.h"
#include <math.h>
#include <limits.h>
#include "arch.h"
#if !defined(_ecintrin_H)
# define _ecintrin_H (1)
/*Some specific platforms may have optimized intrinsic or OPUS_INLINE assembly
versions of these functions which can substantially improve performance.
We define macros for them to allow easy incorporation of these non-ANSI
features.*/
/*Modern gcc (4.x) can compile the naive versions of min and max with cmov if
given an appropriate architecture, but the branchless bit-twiddling versions
are just as fast, and do not require any special target architecture.
Earlier gcc versions (3.x) compiled both code to the same assembly
instructions, because of the way they represented ((_b)>(_a)) internally.*/
# define EC_MINI(_a,_b) ((_a)+(((_b)-(_a))&-((_b)<(_a))))
/*Count leading zeros.
This macro should only be used for implementing ec_ilog(), if it is defined.
All other code should use EC_ILOG() instead.*/
#if defined(_MSC_VER) && (_MSC_VER >= 1400)
# include <intrin.h>
/*In _DEBUG mode this is not an intrinsic by default.*/
# pragma intrinsic(_BitScanReverse)
static __inline int ec_bsr(unsigned long _x){
unsigned long ret;
_BitScanReverse(&ret,_x);
return (int)ret;
}
# define EC_CLZ0 (1)
# define EC_CLZ(_x) (-ec_bsr(_x))
#elif defined(ENABLE_TI_DSPLIB)
# include "dsplib.h"
# define EC_CLZ0 (31)
# define EC_CLZ(_x) (_lnorm(_x))
#elif __GNUC_PREREQ(3,4)
# if INT_MAX>=2147483647
# define EC_CLZ0 ((int)sizeof(unsigned)*CHAR_BIT)
# define EC_CLZ(_x) (__builtin_clz(_x))
# elif LONG_MAX>=2147483647L
# define EC_CLZ0 ((int)sizeof(unsigned long)*CHAR_BIT)
# define EC_CLZ(_x) (__builtin_clzl(_x))
# endif
#endif
#if defined(EC_CLZ)
/*Note that __builtin_clz is not defined when _x==0, according to the gcc
documentation (and that of the BSR instruction that implements it on x86).
The majority of the time we can never pass it zero.
When we need to, it can be special cased.*/
# define EC_ILOG(_x) (EC_CLZ0-EC_CLZ(_x))
#else
int ec_ilog(opus_uint32 _v);
# define EC_ILOG(_x) (ec_ilog(_x))
#endif
#endif

93
drivers/opus/celt/entcode.c Normal file
View file

@ -0,0 +1,93 @@
/* Copyright (c) 2001-2011 Timothy B. Terriberry
*/
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#include "entcode.h"
#include "arch.h"
#if !defined(EC_CLZ)
/*This is a fallback for systems where we don't know how to access
a BSR or CLZ instruction (see ecintrin.h).
If you are optimizing Opus on a new platform and it has a native CLZ or
BZR (e.g. cell, MIPS, x86, etc) then making it available to Opus will be
an easy performance win.*/
int ec_ilog(opus_uint32 _v){
/*On a Pentium M, this branchless version tested as the fastest on
1,000,000,000 random 32-bit integers, edging out a similar version with
branches, and a 256-entry LUT version.*/
int ret;
int m;
ret=!!_v;
m=!!(_v&0xFFFF0000)<<4;
_v>>=m;
ret|=m;
m=!!(_v&0xFF00)<<3;
_v>>=m;
ret|=m;
m=!!(_v&0xF0)<<2;
_v>>=m;
ret|=m;
m=!!(_v&0xC)<<1;
_v>>=m;
ret|=m;
ret+=!!(_v&0x2);
return ret;
}
#endif
opus_uint32 ec_tell_frac(ec_ctx *_this){
opus_uint32 nbits;
opus_uint32 r;
int l;
int i;
/*To handle the non-integral number of bits still left in the encoder/decoder
state, we compute the worst-case number of bits of val that must be
encoded to ensure that the value is inside the range for any possible
subsequent bits.
The computation here is independent of val itself (the decoder does not
even track that value), even though the real number of bits used after
ec_enc_done() may be 1 smaller if rng is a power of two and the
corresponding trailing bits of val are all zeros.
If we did try to track that special case, then coding a value with a
probability of 1/(1<<n) might sometimes appear to use more than n bits.
This may help explain the surprising result that a newly initialized
encoder or decoder claims to have used 1 bit.*/
nbits=_this->nbits_total<<BITRES;
l=EC_ILOG(_this->rng);
r=_this->rng>>(l-16);
for(i=BITRES;i-->0;){
int b;
r=r*r>>15;
b=(int)(r>>16);
l=l<<1|b;
r>>=b;
}
return nbits-l;
}

117
drivers/opus/celt/entcode.h Normal file
View file

@ -0,0 +1,117 @@
/* Copyright (c) 2001-2011 Timothy B. Terriberry
Copyright (c) 2008-2009 Xiph.Org Foundation */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include "opus_types.h"
#include "opus_defines.h"
#if !defined(_entcode_H)
# define _entcode_H (1)
# include <limits.h>
# include <stddef.h>
# include "ecintrin.h"
/*OPT: ec_window must be at least 32 bits, but if you have fast arithmetic on a
larger type, you can speed up the decoder by using it here.*/
typedef opus_uint32 ec_window;
typedef struct ec_ctx ec_ctx;
typedef struct ec_ctx ec_enc;
typedef struct ec_ctx ec_dec;
# define EC_WINDOW_SIZE ((int)sizeof(ec_window)*CHAR_BIT)
/*The number of bits to use for the range-coded part of unsigned integers.*/
# define EC_UINT_BITS (8)
/*The resolution of fractional-precision bit usage measurements, i.e.,
3 => 1/8th bits.*/
# define BITRES 3
/*The entropy encoder/decoder context.
We use the same structure for both, so that common functions like ec_tell()
can be used on either one.*/
struct ec_ctx{
/*Buffered input/output.*/
unsigned char *buf;
/*The size of the buffer.*/
opus_uint32 storage;
/*The offset at which the last byte containing raw bits was read/written.*/
opus_uint32 end_offs;
/*Bits that will be read from/written at the end.*/
ec_window end_window;
/*Number of valid bits in end_window.*/
int nend_bits;
/*The total number of whole bits read/written.
This does not include partial bits currently in the range coder.*/
int nbits_total;
/*The offset at which the next range coder byte will be read/written.*/
opus_uint32 offs;
/*The number of values in the current range.*/
opus_uint32 rng;
/*In the decoder: the difference between the top of the current range and
the input value, minus one.
In the encoder: the low end of the current range.*/
opus_uint32 val;
/*In the decoder: the saved normalization factor from ec_decode().
In the encoder: the number of oustanding carry propagating symbols.*/
opus_uint32 ext;
/*A buffered input/output symbol, awaiting carry propagation.*/
int rem;
/*Nonzero if an error occurred.*/
int error;
};
static OPUS_INLINE opus_uint32 ec_range_bytes(ec_ctx *_this){
return _this->offs;
}
static OPUS_INLINE unsigned char *ec_get_buffer(ec_ctx *_this){
return _this->buf;
}
static OPUS_INLINE int ec_get_error(ec_ctx *_this){
return _this->error;
}
/*Returns the number of bits "used" by the encoded or decoded symbols so far.
This same number can be computed in either the encoder or the decoder, and is
suitable for making coding decisions.
Return: The number of bits.
This will always be slightly larger than the exact value (e.g., all
rounding error is in the positive direction).*/
static OPUS_INLINE int ec_tell(ec_ctx *_this){
return _this->nbits_total-EC_ILOG(_this->rng);
}
/*Returns the number of bits "used" by the encoded or decoded symbols so far.
This same number can be computed in either the encoder or the decoder, and is
suitable for making coding decisions.
Return: The number of bits scaled by 2**BITRES.
This will always be slightly larger than the exact value (e.g., all
rounding error is in the positive direction).*/
opus_uint32 ec_tell_frac(ec_ctx *_this);
#endif

245
drivers/opus/celt/entdec.c Normal file
View file

@ -0,0 +1,245 @@
/* Copyright (c) 2001-2011 Timothy B. Terriberry
Copyright (c) 2008-2009 Xiph.Org Foundation */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#include <stddef.h>
#include "os_support.h"
#include "arch.h"
#include "entdec.h"
#include "mfrngcod.h"
/*A range decoder.
This is an entropy decoder based upon \cite{Mar79}, which is itself a
rediscovery of the FIFO arithmetic code introduced by \cite{Pas76}.
It is very similar to arithmetic encoding, except that encoding is done with
digits in any base, instead of with bits, and so it is faster when using
larger bases (i.e.: a byte).
The author claims an average waste of $\frac{1}{2}\log_b(2b)$ bits, where $b$
is the base, longer than the theoretical optimum, but to my knowledge there
is no published justification for this claim.
This only seems true when using near-infinite precision arithmetic so that
the process is carried out with no rounding errors.
An excellent description of implementation details is available at
http://www.arturocampos.com/ac_range.html
A recent work \cite{MNW98} which proposes several changes to arithmetic
encoding for efficiency actually re-discovers many of the principles
behind range encoding, and presents a good theoretical analysis of them.
End of stream is handled by writing out the smallest number of bits that
ensures that the stream will be correctly decoded regardless of the value of
any subsequent bits.
ec_tell() can be used to determine how many bits were needed to decode
all the symbols thus far; other data can be packed in the remaining bits of
the input buffer.
@PHDTHESIS{Pas76,
author="Richard Clark Pasco",
title="Source coding algorithms for fast data compression",
school="Dept. of Electrical Engineering, Stanford University",
address="Stanford, CA",
month=May,
year=1976
}
@INPROCEEDINGS{Mar79,
author="Martin, G.N.N.",
title="Range encoding: an algorithm for removing redundancy from a digitised
message",
booktitle="Video & Data Recording Conference",
year=1979,
address="Southampton",
month=Jul
}
@ARTICLE{MNW98,
author="Alistair Moffat and Radford Neal and Ian H. Witten",
title="Arithmetic Coding Revisited",
journal="{ACM} Transactions on Information Systems",
year=1998,
volume=16,
number=3,
pages="256--294",
month=Jul,
URL="http://www.stanford.edu/class/ee398a/handouts/papers/Moffat98ArithmCoding.pdf"
}*/
static int ec_read_byte(ec_dec *_this){
return _this->offs<_this->storage?_this->buf[_this->offs++]:0;
}
static int ec_read_byte_from_end(ec_dec *_this){
return _this->end_offs<_this->storage?
_this->buf[_this->storage-++(_this->end_offs)]:0;
}
/*Normalizes the contents of val and rng so that rng lies entirely in the
high-order symbol.*/
static void ec_dec_normalize(ec_dec *_this){
/*If the range is too small, rescale it and input some bits.*/
while(_this->rng<=EC_CODE_BOT){
int sym;
_this->nbits_total+=EC_SYM_BITS;
_this->rng<<=EC_SYM_BITS;
/*Use up the remaining bits from our last symbol.*/
sym=_this->rem;
/*Read the next value from the input.*/
_this->rem=ec_read_byte(_this);
/*Take the rest of the bits we need from this new symbol.*/
sym=(sym<<EC_SYM_BITS|_this->rem)>>(EC_SYM_BITS-EC_CODE_EXTRA);
/*And subtract them from val, capped to be less than EC_CODE_TOP.*/
_this->val=((_this->val<<EC_SYM_BITS)+(EC_SYM_MAX&~sym))&(EC_CODE_TOP-1);
}
}
void ec_dec_init(ec_dec *_this,unsigned char *_buf,opus_uint32 _storage){
_this->buf=_buf;
_this->storage=_storage;
_this->end_offs=0;
_this->end_window=0;
_this->nend_bits=0;
/*This is the offset from which ec_tell() will subtract partial bits.
The final value after the ec_dec_normalize() call will be the same as in
the encoder, but we have to compensate for the bits that are added there.*/
_this->nbits_total=EC_CODE_BITS+1
-((EC_CODE_BITS-EC_CODE_EXTRA)/EC_SYM_BITS)*EC_SYM_BITS;
_this->offs=0;
_this->rng=1U<<EC_CODE_EXTRA;
_this->rem=ec_read_byte(_this);
_this->val=_this->rng-1-(_this->rem>>(EC_SYM_BITS-EC_CODE_EXTRA));
_this->error=0;
/*Normalize the interval.*/
ec_dec_normalize(_this);
}
unsigned ec_decode(ec_dec *_this,unsigned _ft){
unsigned s;
_this->ext=_this->rng/_ft;
s=(unsigned)(_this->val/_this->ext);
return _ft-EC_MINI(s+1,_ft);
}
unsigned ec_decode_bin(ec_dec *_this,unsigned _bits){
unsigned s;
_this->ext=_this->rng>>_bits;
s=(unsigned)(_this->val/_this->ext);
return (1U<<_bits)-EC_MINI(s+1U,1U<<_bits);
}
void ec_dec_update(ec_dec *_this,unsigned _fl,unsigned _fh,unsigned _ft){
opus_uint32 s;
s=IMUL32(_this->ext,_ft-_fh);
_this->val-=s;
_this->rng=_fl>0?IMUL32(_this->ext,_fh-_fl):_this->rng-s;
ec_dec_normalize(_this);
}
/*The probability of having a "one" is 1/(1<<_logp).*/
int ec_dec_bit_logp(ec_dec *_this,unsigned _logp){
opus_uint32 r;
opus_uint32 d;
opus_uint32 s;
int ret;
r=_this->rng;
d=_this->val;
s=r>>_logp;
ret=d<s;
if(!ret)_this->val=d-s;
_this->rng=ret?s:r-s;
ec_dec_normalize(_this);
return ret;
}
int ec_dec_icdf(ec_dec *_this,const unsigned char *_icdf,unsigned _ftb){
opus_uint32 r;
opus_uint32 d;
opus_uint32 s;
opus_uint32 t;
int ret;
s=_this->rng;
d=_this->val;
r=s>>_ftb;
ret=-1;
do{
t=s;
s=IMUL32(r,_icdf[++ret]);
}
while(d<s);
_this->val=d-s;
_this->rng=t-s;
ec_dec_normalize(_this);
return ret;
}
opus_uint32 ec_dec_uint(ec_dec *_this,opus_uint32 _ft){
unsigned ft;
unsigned s;
int ftb;
/*In order to optimize EC_ILOG(), it is undefined for the value 0.*/
celt_assert(_ft>1);
_ft--;
ftb=EC_ILOG(_ft);
if(ftb>EC_UINT_BITS){
opus_uint32 t;
ftb-=EC_UINT_BITS;
ft=(unsigned)(_ft>>ftb)+1;
s=ec_decode(_this,ft);
ec_dec_update(_this,s,s+1,ft);
t=(opus_uint32)s<<ftb|ec_dec_bits(_this,ftb);
if(t<=_ft)return t;
_this->error=1;
return _ft;
}
else{
_ft++;
s=ec_decode(_this,(unsigned)_ft);
ec_dec_update(_this,s,s+1,(unsigned)_ft);
return s;
}
}
opus_uint32 ec_dec_bits(ec_dec *_this,unsigned _bits){
ec_window window;
int available;
opus_uint32 ret;
window=_this->end_window;
available=_this->nend_bits;
if((unsigned)available<_bits){
do{
window|=(ec_window)ec_read_byte_from_end(_this)<<available;
available+=EC_SYM_BITS;
}
while(available<=EC_WINDOW_SIZE-EC_SYM_BITS);
}
ret=(opus_uint32)window&(((opus_uint32)1<<_bits)-1U);
window>>=_bits;
available-=_bits;
_this->end_window=window;
_this->nend_bits=available;
_this->nbits_total+=_bits;
return ret;
}

100
drivers/opus/celt/entdec.h Normal file
View file

@ -0,0 +1,100 @@
/* Copyright (c) 2001-2011 Timothy B. Terriberry
Copyright (c) 2008-2009 Xiph.Org Foundation */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#if !defined(_entdec_H)
# define _entdec_H (1)
# include <limits.h>
# include "entcode.h"
/*Initializes the decoder.
_buf: The input buffer to use.
Return: 0 on success, or a negative value on error.*/
void ec_dec_init(ec_dec *_this,unsigned char *_buf,opus_uint32 _storage);
/*Calculates the cumulative frequency for the next symbol.
This can then be fed into the probability model to determine what that
symbol is, and the additional frequency information required to advance to
the next symbol.
This function cannot be called more than once without a corresponding call to
ec_dec_update(), or decoding will not proceed correctly.
_ft: The total frequency of the symbols in the alphabet the next symbol was
encoded with.
Return: A cumulative frequency representing the encoded symbol.
If the cumulative frequency of all the symbols before the one that
was encoded was fl, and the cumulative frequency of all the symbols
up to and including the one encoded is fh, then the returned value
will fall in the range [fl,fh).*/
unsigned ec_decode(ec_dec *_this,unsigned _ft);
/*Equivalent to ec_decode() with _ft==1<<_bits.*/
unsigned ec_decode_bin(ec_dec *_this,unsigned _bits);
/*Advance the decoder past the next symbol using the frequency information the
symbol was encoded with.
Exactly one call to ec_decode() must have been made so that all necessary
intermediate calculations are performed.
_fl: The cumulative frequency of all symbols that come before the symbol
decoded.
_fh: The cumulative frequency of all symbols up to and including the symbol
decoded.
Together with _fl, this defines the range [_fl,_fh) in which the value
returned above must fall.
_ft: The total frequency of the symbols in the alphabet the symbol decoded
was encoded in.
This must be the same as passed to the preceding call to ec_decode().*/
void ec_dec_update(ec_dec *_this,unsigned _fl,unsigned _fh,unsigned _ft);
/* Decode a bit that has a 1/(1<<_logp) probability of being a one */
int ec_dec_bit_logp(ec_dec *_this,unsigned _logp);
/*Decodes a symbol given an "inverse" CDF table.
No call to ec_dec_update() is necessary after this call.
_icdf: The "inverse" CDF, such that symbol s falls in the range
[s>0?ft-_icdf[s-1]:0,ft-_icdf[s]), where ft=1<<_ftb.
The values must be monotonically non-increasing, and the last value
must be 0.
_ftb: The number of bits of precision in the cumulative distribution.
Return: The decoded symbol s.*/
int ec_dec_icdf(ec_dec *_this,const unsigned char *_icdf,unsigned _ftb);
/*Extracts a raw unsigned integer with a non-power-of-2 range from the stream.
The bits must have been encoded with ec_enc_uint().
No call to ec_dec_update() is necessary after this call.
_ft: The number of integers that can be decoded (one more than the max).
This must be at least one, and no more than 2**32-1.
Return: The decoded bits.*/
opus_uint32 ec_dec_uint(ec_dec *_this,opus_uint32 _ft);
/*Extracts a sequence of raw bits from the stream.
The bits must have been encoded with ec_enc_bits().
No call to ec_dec_update() is necessary after this call.
_ftb: The number of bits to extract.
This must be between 0 and 25, inclusive.
Return: The decoded bits.*/
opus_uint32 ec_dec_bits(ec_dec *_this,unsigned _ftb);
#endif

294
drivers/opus/celt/entenc.c Normal file
View file

@ -0,0 +1,294 @@
/* Copyright (c) 2001-2011 Timothy B. Terriberry
Copyright (c) 2008-2009 Xiph.Org Foundation */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#if defined(OPUS_HAVE_CONFIG_H)
# include "opus_config.h"
#endif
#include "os_support.h"
#include "arch.h"
#include "entenc.h"
#include "mfrngcod.h"
/*A range encoder.
See entdec.c and the references for implementation details \cite{Mar79,MNW98}.
@INPROCEEDINGS{Mar79,
author="Martin, G.N.N.",
title="Range encoding: an algorithm for removing redundancy from a digitised
message",
booktitle="Video \& Data Recording Conference",
year=1979,
address="Southampton",
month=Jul
}
@ARTICLE{MNW98,
author="Alistair Moffat and Radford Neal and Ian H. Witten",
title="Arithmetic Coding Revisited",
journal="{ACM} Transactions on Information Systems",
year=1998,
volume=16,
number=3,
pages="256--294",
month=Jul,
URL="http://www.stanford.edu/class/ee398/handouts/papers/Moffat98ArithmCoding.pdf"
}*/
static int ec_write_byte(ec_enc *_this,unsigned _value){
if(_this->offs+_this->end_offs>=_this->storage)return -1;
_this->buf[_this->offs++]=(unsigned char)_value;
return 0;
}
static int ec_write_byte_at_end(ec_enc *_this,unsigned _value){
if(_this->offs+_this->end_offs>=_this->storage)return -1;
_this->buf[_this->storage-++(_this->end_offs)]=(unsigned char)_value;
return 0;
}
/*Outputs a symbol, with a carry bit.
If there is a potential to propagate a carry over several symbols, they are
buffered until it can be determined whether or not an actual carry will
occur.
If the counter for the buffered symbols overflows, then the stream becomes
undecodable.
This gives a theoretical limit of a few billion symbols in a single packet on
32-bit systems.
The alternative is to truncate the range in order to force a carry, but
requires similar carry tracking in the decoder, needlessly slowing it down.*/
static void ec_enc_carry_out(ec_enc *_this,int _c){
if(_c!=EC_SYM_MAX){
/*No further carry propagation possible, flush buffer.*/
int carry;
carry=_c>>EC_SYM_BITS;
/*Don't output a byte on the first write.
This compare should be taken care of by branch-prediction thereafter.*/
if(_this->rem>=0)_this->error|=ec_write_byte(_this,_this->rem+carry);
if(_this->ext>0){
unsigned sym;
sym=(EC_SYM_MAX+carry)&EC_SYM_MAX;
do _this->error|=ec_write_byte(_this,sym);
while(--(_this->ext)>0);
}
_this->rem=_c&EC_SYM_MAX;
}
else _this->ext++;
}
static void ec_enc_normalize(ec_enc *_this){
/*If the range is too small, output some bits and rescale it.*/
while(_this->rng<=EC_CODE_BOT){
ec_enc_carry_out(_this,(int)(_this->val>>EC_CODE_SHIFT));
/*Move the next-to-high-order symbol into the high-order position.*/
_this->val=(_this->val<<EC_SYM_BITS)&(EC_CODE_TOP-1);
_this->rng<<=EC_SYM_BITS;
_this->nbits_total+=EC_SYM_BITS;
}
}
void ec_enc_init(ec_enc *_this,unsigned char *_buf,opus_uint32 _size){
_this->buf=_buf;
_this->end_offs=0;
_this->end_window=0;
_this->nend_bits=0;
/*This is the offset from which ec_tell() will subtract partial bits.*/
_this->nbits_total=EC_CODE_BITS+1;
_this->offs=0;
_this->rng=EC_CODE_TOP;
_this->rem=-1;
_this->val=0;
_this->ext=0;
_this->storage=_size;
_this->error=0;
}
void ec_encode(ec_enc *_this,unsigned _fl,unsigned _fh,unsigned _ft){
opus_uint32 r;
r=_this->rng/_ft;
if(_fl>0){
_this->val+=_this->rng-IMUL32(r,(_ft-_fl));
_this->rng=IMUL32(r,(_fh-_fl));
}
else _this->rng-=IMUL32(r,(_ft-_fh));
ec_enc_normalize(_this);
}
void ec_encode_bin(ec_enc *_this,unsigned _fl,unsigned _fh,unsigned _bits){
opus_uint32 r;
r=_this->rng>>_bits;
if(_fl>0){
_this->val+=_this->rng-IMUL32(r,((1U<<_bits)-_fl));
_this->rng=IMUL32(r,(_fh-_fl));
}
else _this->rng-=IMUL32(r,((1U<<_bits)-_fh));
ec_enc_normalize(_this);
}
/*The probability of having a "one" is 1/(1<<_logp).*/
void ec_enc_bit_logp(ec_enc *_this,int _val,unsigned _logp){
opus_uint32 r;
opus_uint32 s;
opus_uint32 l;
r=_this->rng;
l=_this->val;
s=r>>_logp;
r-=s;
if(_val)_this->val=l+r;
_this->rng=_val?s:r;
ec_enc_normalize(_this);
}
void ec_enc_icdf(ec_enc *_this,int _s,const unsigned char *_icdf,unsigned _ftb){
opus_uint32 r;
r=_this->rng>>_ftb;
if(_s>0){
_this->val+=_this->rng-IMUL32(r,_icdf[_s-1]);
_this->rng=IMUL32(r,_icdf[_s-1]-_icdf[_s]);
}
else _this->rng-=IMUL32(r,_icdf[_s]);
ec_enc_normalize(_this);
}
void ec_enc_uint(ec_enc *_this,opus_uint32 _fl,opus_uint32 _ft){
unsigned ft;
unsigned fl;
int ftb;
/*In order to optimize EC_ILOG(), it is undefined for the value 0.*/
celt_assert(_ft>1);
_ft--;
ftb=EC_ILOG(_ft);
if(ftb>EC_UINT_BITS){
ftb-=EC_UINT_BITS;
ft=(_ft>>ftb)+1;
fl=(unsigned)(_fl>>ftb);
ec_encode(_this,fl,fl+1,ft);
ec_enc_bits(_this,_fl&(((opus_uint32)1<<ftb)-1U),ftb);
}
else ec_encode(_this,_fl,_fl+1,_ft+1);
}
void ec_enc_bits(ec_enc *_this,opus_uint32 _fl,unsigned _bits){
ec_window window;
int used;
window=_this->end_window;
used=_this->nend_bits;
celt_assert(_bits>0);
if(used+_bits>EC_WINDOW_SIZE){
do{
_this->error|=ec_write_byte_at_end(_this,(unsigned)window&EC_SYM_MAX);
window>>=EC_SYM_BITS;
used-=EC_SYM_BITS;
}
while(used>=EC_SYM_BITS);
}
window|=(ec_window)_fl<<used;
used+=_bits;
_this->end_window=window;
_this->nend_bits=used;
_this->nbits_total+=_bits;
}
void ec_enc_patch_initial_bits(ec_enc *_this,unsigned _val,unsigned _nbits){
int shift;
unsigned mask;
celt_assert(_nbits<=EC_SYM_BITS);
shift=EC_SYM_BITS-_nbits;
mask=((1<<_nbits)-1)<<shift;
if(_this->offs>0){
/*The first byte has been finalized.*/
_this->buf[0]=(unsigned char)((_this->buf[0]&~mask)|_val<<shift);
}
else if(_this->rem>=0){
/*The first byte is still awaiting carry propagation.*/
_this->rem=(_this->rem&~mask)|_val<<shift;
}
else if(_this->rng<=(EC_CODE_TOP>>_nbits)){
/*The renormalization loop has never been run.*/
_this->val=(_this->val&~((opus_uint32)mask<<EC_CODE_SHIFT))|
(opus_uint32)_val<<(EC_CODE_SHIFT+shift);
}
/*The encoder hasn't even encoded _nbits of data yet.*/
else _this->error=-1;
}
void ec_enc_shrink(ec_enc *_this,opus_uint32 _size){
celt_assert(_this->offs+_this->end_offs<=_size);
OPUS_MOVE(_this->buf+_size-_this->end_offs,
_this->buf+_this->storage-_this->end_offs,_this->end_offs);
_this->storage=_size;
}
void ec_enc_done(ec_enc *_this){
ec_window window;
int used;
opus_uint32 msk;
opus_uint32 end;
int l;
/*We output the minimum number of bits that ensures that the symbols encoded
thus far will be decoded correctly regardless of the bits that follow.*/
l=EC_CODE_BITS-EC_ILOG(_this->rng);
msk=(EC_CODE_TOP-1)>>l;
end=(_this->val+msk)&~msk;
if((end|msk)>=_this->val+_this->rng){
l++;
msk>>=1;
end=(_this->val+msk)&~msk;
}
while(l>0){
ec_enc_carry_out(_this,(int)(end>>EC_CODE_SHIFT));
end=(end<<EC_SYM_BITS)&(EC_CODE_TOP-1);
l-=EC_SYM_BITS;
}
/*If we have a buffered byte flush it into the output buffer.*/
if(_this->rem>=0||_this->ext>0)ec_enc_carry_out(_this,0);
/*If we have buffered extra bits, flush them as well.*/
window=_this->end_window;
used=_this->nend_bits;
while(used>=EC_SYM_BITS){
_this->error|=ec_write_byte_at_end(_this,(unsigned)window&EC_SYM_MAX);
window>>=EC_SYM_BITS;
used-=EC_SYM_BITS;
}
/*Clear any excess space and add any remaining extra bits to the last byte.*/
if(!_this->error){
OPUS_CLEAR(_this->buf+_this->offs,
_this->storage-_this->offs-_this->end_offs);
if(used>0){
/*If there's no range coder data at all, give up.*/
if(_this->end_offs>=_this->storage)_this->error=-1;
else{
l=-l;
/*If we've busted, don't add too many extra bits to the last byte; it
would corrupt the range coder data, and that's more important.*/
if(_this->offs+_this->end_offs>=_this->storage&&l<used){
window&=(1<<l)-1;
_this->error=-1;
}
_this->buf[_this->storage-_this->end_offs-1]|=(unsigned char)window;
}
}
}
}

110
drivers/opus/celt/entenc.h Normal file
View file

@ -0,0 +1,110 @@
/* Copyright (c) 2001-2011 Timothy B. Terriberry
Copyright (c) 2008-2009 Xiph.Org Foundation */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#if !defined(_entenc_H)
# define _entenc_H (1)
# include <stddef.h>
# include "entcode.h"
/*Initializes the encoder.
_buf: The buffer to store output bytes in.
_size: The size of the buffer, in chars.*/
void ec_enc_init(ec_enc *_this,unsigned char *_buf,opus_uint32 _size);
/*Encodes a symbol given its frequency information.
The frequency information must be discernable by the decoder, assuming it
has read only the previous symbols from the stream.
It is allowable to change the frequency information, or even the entire
source alphabet, so long as the decoder can tell from the context of the
previously encoded information that it is supposed to do so as well.
_fl: The cumulative frequency of all symbols that come before the one to be
encoded.
_fh: The cumulative frequency of all symbols up to and including the one to
be encoded.
Together with _fl, this defines the range [_fl,_fh) in which the
decoded value will fall.
_ft: The sum of the frequencies of all the symbols*/
void ec_encode(ec_enc *_this,unsigned _fl,unsigned _fh,unsigned _ft);
/*Equivalent to ec_encode() with _ft==1<<_bits.*/
void ec_encode_bin(ec_enc *_this,unsigned _fl,unsigned _fh,unsigned _bits);
/* Encode a bit that has a 1/(1<<_logp) probability of being a one */
void ec_enc_bit_logp(ec_enc *_this,int _val,unsigned _logp);
/*Encodes a symbol given an "inverse" CDF table.
_s: The index of the symbol to encode.
_icdf: The "inverse" CDF, such that symbol _s falls in the range
[_s>0?ft-_icdf[_s-1]:0,ft-_icdf[_s]), where ft=1<<_ftb.
The values must be monotonically non-increasing, and the last value
must be 0.
_ftb: The number of bits of precision in the cumulative distribution.*/
void ec_enc_icdf(ec_enc *_this,int _s,const unsigned char *_icdf,unsigned _ftb);
/*Encodes a raw unsigned integer in the stream.
_fl: The integer to encode.
_ft: The number of integers that can be encoded (one more than the max).
This must be at least one, and no more than 2**32-1.*/
void ec_enc_uint(ec_enc *_this,opus_uint32 _fl,opus_uint32 _ft);
/*Encodes a sequence of raw bits in the stream.
_fl: The bits to encode.
_ftb: The number of bits to encode.
This must be between 1 and 25, inclusive.*/
void ec_enc_bits(ec_enc *_this,opus_uint32 _fl,unsigned _ftb);
/*Overwrites a few bits at the very start of an existing stream, after they
have already been encoded.
This makes it possible to have a few flags up front, where it is easy for
decoders to access them without parsing the whole stream, even if their
values are not determined until late in the encoding process, without having
to buffer all the intermediate symbols in the encoder.
In order for this to work, at least _nbits bits must have already been
encoded using probabilities that are an exact power of two.
The encoder can verify the number of encoded bits is sufficient, but cannot
check this latter condition.
_val: The bits to encode (in the least _nbits significant bits).
They will be decoded in order from most-significant to least.
_nbits: The number of bits to overwrite.
This must be no more than 8.*/
void ec_enc_patch_initial_bits(ec_enc *_this,unsigned _val,unsigned _nbits);
/*Compacts the data to fit in the target size.
This moves up the raw bits at the end of the current buffer so they are at
the end of the new buffer size.
The caller must ensure that the amount of data that's already been written
will fit in the new size.
_size: The number of bytes in the new buffer.
This must be large enough to contain the bits already written, and
must be no larger than the existing size.*/
void ec_enc_shrink(ec_enc *_this,opus_uint32 _size);
/*Indicates that there are no more symbols to encode.
All reamining output bytes are flushed to the output buffer.
ec_enc_init() must be called before the encoder can be used again.*/
void ec_enc_done(ec_enc *_this);
#endif

773
drivers/opus/celt/fixed_debug.h Normal file
View file

@ -0,0 +1,773 @@
/* Copyright (C) 2003-2008 Jean-Marc Valin
Copyright (C) 2007-2012 Xiph.Org Foundation */
/**
@file fixed_debug.h
@brief Fixed-point operations with debugging
*/
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef FIXED_DEBUG_H
#define FIXED_DEBUG_H
#include <stdio.h>
#include "opus_defines.h"
#ifdef CELT_C
OPUS_EXPORT opus_int64 celt_mips=0;
#else
extern opus_int64 celt_mips;
#endif
#define MULT16_16SU(a,b) ((opus_val32)(opus_val16)(a)*(opus_val32)(opus_uint16)(b))
#define MULT32_32_Q31(a,b) ADD32(ADD32(SHL32(MULT16_16(SHR32((a),16),SHR((b),16)),1), SHR32(MULT16_16SU(SHR32((a),16),((b)&0x0000ffff)),15)), SHR32(MULT16_16SU(SHR32((b),16),((a)&0x0000ffff)),15))
/** 16x32 multiplication, followed by a 16-bit shift right. Results fits in 32 bits */
#define MULT16_32_Q16(a,b) ADD32(MULT16_16((a),SHR32((b),16)), SHR32(MULT16_16SU((a),((b)&0x0000ffff)),16))
#define MULT16_32_P16(a,b) MULT16_32_PX(a,b,16)
#define QCONST16(x,bits) ((opus_val16)(.5+(x)*(((opus_val32)1)<<(bits))))
#define QCONST32(x,bits) ((opus_val32)(.5+(x)*(((opus_val32)1)<<(bits))))
#define VERIFY_SHORT(x) ((x)<=32767&&(x)>=-32768)
#define VERIFY_INT(x) ((x)<=2147483647LL&&(x)>=-2147483648LL)
#define VERIFY_UINT(x) ((x)<=(2147483647LLU<<1))
#define SHR(a,b) SHR32(a,b)
#define PSHR(a,b) PSHR32(a,b)
static OPUS_INLINE short NEG16(int x)
{
int res;
if (!VERIFY_SHORT(x))
{
fprintf (stderr, "NEG16: input is not short: %d\n", (int)x);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = -x;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "NEG16: output is not short: %d\n", (int)res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips++;
return res;
}
static OPUS_INLINE int NEG32(opus_int64 x)
{
opus_int64 res;
if (!VERIFY_INT(x))
{
fprintf (stderr, "NEG16: input is not int: %d\n", (int)x);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = -x;
if (!VERIFY_INT(res))
{
fprintf (stderr, "NEG16: output is not int: %d\n", (int)res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=2;
return res;
}
#define EXTRACT16(x) EXTRACT16_(x, __FILE__, __LINE__)
static OPUS_INLINE short EXTRACT16_(int x, char *file, int line)
{
int res;
if (!VERIFY_SHORT(x))
{
fprintf (stderr, "EXTRACT16: input is not short: %d in %s: line %d\n", x, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = x;
celt_mips++;
return res;
}
#define EXTEND32(x) EXTEND32_(x, __FILE__, __LINE__)
static OPUS_INLINE int EXTEND32_(int x, char *file, int line)
{
int res;
if (!VERIFY_SHORT(x))
{
fprintf (stderr, "EXTEND32: input is not short: %d in %s: line %d\n", x, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = x;
celt_mips++;
return res;
}
#define SHR16(a, shift) SHR16_(a, shift, __FILE__, __LINE__)
static OPUS_INLINE short SHR16_(int a, int shift, char *file, int line)
{
int res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(shift))
{
fprintf (stderr, "SHR16: inputs are not short: %d >> %d in %s: line %d\n", a, shift, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a>>shift;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "SHR16: output is not short: %d in %s: line %d\n", res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips++;
return res;
}
#define SHL16(a, shift) SHL16_(a, shift, __FILE__, __LINE__)
static OPUS_INLINE short SHL16_(int a, int shift, char *file, int line)
{
int res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(shift))
{
fprintf (stderr, "SHL16: inputs are not short: %d %d in %s: line %d\n", a, shift, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a<<shift;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "SHL16: output is not short: %d in %s: line %d\n", res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips++;
return res;
}
static OPUS_INLINE int SHR32(opus_int64 a, int shift)
{
opus_int64 res;
if (!VERIFY_INT(a) || !VERIFY_SHORT(shift))
{
fprintf (stderr, "SHR32: inputs are not int: %d %d\n", (int)a, shift);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a>>shift;
if (!VERIFY_INT(res))
{
fprintf (stderr, "SHR32: output is not int: %d\n", (int)res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=2;
return res;
}
#define SHL32(a, shift) SHL32_(a, shift, __FILE__, __LINE__)
static OPUS_INLINE int SHL32_(opus_int64 a, int shift, char *file, int line)
{
opus_int64 res;
if (!VERIFY_INT(a) || !VERIFY_SHORT(shift))
{
fprintf (stderr, "SHL32: inputs are not int: %lld %d in %s: line %d\n", a, shift, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a<<shift;
if (!VERIFY_INT(res))
{
fprintf (stderr, "SHL32: output is not int: %lld<<%d = %lld in %s: line %d\n", a, shift, res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=2;
return res;
}
#define PSHR32(a,shift) (celt_mips--,SHR32(ADD32((a),(((opus_val32)(1)<<((shift))>>1))),shift))
#define VSHR32(a, shift) (((shift)>0) ? SHR32(a, shift) : SHL32(a, -(shift)))
#define ROUND16(x,a) (celt_mips--,EXTRACT16(PSHR32((x),(a))))
#define HALF16(x) (SHR16(x,1))
#define HALF32(x) (SHR32(x,1))
//#define SHR(a,shift) ((a) >> (shift))
//#define SHL(a,shift) ((a) << (shift))
#define ADD16(a, b) ADD16_(a, b, __FILE__, __LINE__)
static OPUS_INLINE short ADD16_(int a, int b, char *file, int line)
{
int res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(b))
{
fprintf (stderr, "ADD16: inputs are not short: %d %d in %s: line %d\n", a, b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a+b;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "ADD16: output is not short: %d+%d=%d in %s: line %d\n", a,b,res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips++;
return res;
}
#define SUB16(a, b) SUB16_(a, b, __FILE__, __LINE__)
static OPUS_INLINE short SUB16_(int a, int b, char *file, int line)
{
int res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(b))
{
fprintf (stderr, "SUB16: inputs are not short: %d %d in %s: line %d\n", a, b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a-b;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "SUB16: output is not short: %d in %s: line %d\n", res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips++;
return res;
}
#define ADD32(a, b) ADD32_(a, b, __FILE__, __LINE__)
static OPUS_INLINE int ADD32_(opus_int64 a, opus_int64 b, char *file, int line)
{
opus_int64 res;
if (!VERIFY_INT(a) || !VERIFY_INT(b))
{
fprintf (stderr, "ADD32: inputs are not int: %d %d in %s: line %d\n", (int)a, (int)b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a+b;
if (!VERIFY_INT(res))
{
fprintf (stderr, "ADD32: output is not int: %d in %s: line %d\n", (int)res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=2;
return res;
}
#define SUB32(a, b) SUB32_(a, b, __FILE__, __LINE__)
static OPUS_INLINE int SUB32_(opus_int64 a, opus_int64 b, char *file, int line)
{
opus_int64 res;
if (!VERIFY_INT(a) || !VERIFY_INT(b))
{
fprintf (stderr, "SUB32: inputs are not int: %d %d in %s: line %d\n", (int)a, (int)b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a-b;
if (!VERIFY_INT(res))
{
fprintf (stderr, "SUB32: output is not int: %d in %s: line %d\n", (int)res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=2;
return res;
}
#undef UADD32
#define UADD32(a, b) UADD32_(a, b, __FILE__, __LINE__)
static OPUS_INLINE unsigned int UADD32_(opus_uint64 a, opus_uint64 b, char *file, int line)
{
opus_uint64 res;
if (!VERIFY_UINT(a) || !VERIFY_UINT(b))
{
fprintf (stderr, "UADD32: inputs are not uint32: %llu %llu in %s: line %d\n", a, b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a+b;
if (!VERIFY_UINT(res))
{
fprintf (stderr, "UADD32: output is not uint32: %llu in %s: line %d\n", res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=2;
return res;
}
#undef USUB32
#define USUB32(a, b) USUB32_(a, b, __FILE__, __LINE__)
static OPUS_INLINE unsigned int USUB32_(opus_uint64 a, opus_uint64 b, char *file, int line)
{
opus_uint64 res;
if (!VERIFY_UINT(a) || !VERIFY_UINT(b))
{
fprintf (stderr, "USUB32: inputs are not uint32: %llu %llu in %s: line %d\n", a, b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
if (a<b)
{
fprintf (stderr, "USUB32: inputs underflow: %llu < %llu in %s: line %d\n", a, b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a-b;
if (!VERIFY_UINT(res))
{
fprintf (stderr, "USUB32: output is not uint32: %llu - %llu = %llu in %s: line %d\n", a, b, res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=2;
return res;
}
/* result fits in 16 bits */
static OPUS_INLINE short MULT16_16_16(int a, int b)
{
int res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(b))
{
fprintf (stderr, "MULT16_16_16: inputs are not short: %d %d\n", a, b);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a*b;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "MULT16_16_16: output is not short: %d\n", res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips++;
return res;
}
#define MULT16_16(a, b) MULT16_16_(a, b, __FILE__, __LINE__)
static OPUS_INLINE int MULT16_16_(int a, int b, char *file, int line)
{
opus_int64 res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(b))
{
fprintf (stderr, "MULT16_16: inputs are not short: %d %d in %s: line %d\n", a, b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = ((opus_int64)a)*b;
if (!VERIFY_INT(res))
{
fprintf (stderr, "MULT16_16: output is not int: %d in %s: line %d\n", (int)res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips++;
return res;
}
#define MAC16_16(c,a,b) (celt_mips-=2,ADD32((c),MULT16_16((a),(b))))
#define MULT16_32_QX(a, b, Q) MULT16_32_QX_(a, b, Q, __FILE__, __LINE__)
static OPUS_INLINE int MULT16_32_QX_(int a, opus_int64 b, int Q, char *file, int line)
{
opus_int64 res;
if (!VERIFY_SHORT(a) || !VERIFY_INT(b))
{
fprintf (stderr, "MULT16_32_Q%d: inputs are not short+int: %d %d in %s: line %d\n", Q, (int)a, (int)b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
if (ABS32(b)>=((opus_val32)(1)<<(15+Q)))
{
fprintf (stderr, "MULT16_32_Q%d: second operand too large: %d %d in %s: line %d\n", Q, (int)a, (int)b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = (((opus_int64)a)*(opus_int64)b) >> Q;
if (!VERIFY_INT(res))
{
fprintf (stderr, "MULT16_32_Q%d: output is not int: %d*%d=%d in %s: line %d\n", Q, (int)a, (int)b,(int)res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
if (Q==15)
celt_mips+=3;
else
celt_mips+=4;
return res;
}
#define MULT16_32_PX(a, b, Q) MULT16_32_PX_(a, b, Q, __FILE__, __LINE__)
static OPUS_INLINE int MULT16_32_PX_(int a, opus_int64 b, int Q, char *file, int line)
{
opus_int64 res;
if (!VERIFY_SHORT(a) || !VERIFY_INT(b))
{
fprintf (stderr, "MULT16_32_P%d: inputs are not short+int: %d %d in %s: line %d\n\n", Q, (int)a, (int)b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
if (ABS32(b)>=((opus_int64)(1)<<(15+Q)))
{
fprintf (stderr, "MULT16_32_Q%d: second operand too large: %d %d in %s: line %d\n\n", Q, (int)a, (int)b,file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = ((((opus_int64)a)*(opus_int64)b) + (((opus_val32)(1)<<Q)>>1))>> Q;
if (!VERIFY_INT(res))
{
fprintf (stderr, "MULT16_32_P%d: output is not int: %d*%d=%d in %s: line %d\n\n", Q, (int)a, (int)b,(int)res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
if (Q==15)
celt_mips+=4;
else
celt_mips+=5;
return res;
}
#define MULT16_32_Q15(a,b) MULT16_32_QX(a,b,15)
#define MAC16_32_Q15(c,a,b) (celt_mips-=2,ADD32((c),MULT16_32_Q15((a),(b))))
static OPUS_INLINE int SATURATE(int a, int b)
{
if (a>b)
a=b;
if (a<-b)
a = -b;
celt_mips+=3;
return a;
}
static OPUS_INLINE opus_int16 SATURATE16(opus_int32 a)
{
celt_mips+=3;
if (a>32767)
return 32767;
else if (a<-32768)
return -32768;
else return a;
}
static OPUS_INLINE int MULT16_16_Q11_32(int a, int b)
{
opus_int64 res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(b))
{
fprintf (stderr, "MULT16_16_Q11: inputs are not short: %d %d\n", a, b);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = ((opus_int64)a)*b;
res >>= 11;
if (!VERIFY_INT(res))
{
fprintf (stderr, "MULT16_16_Q11: output is not short: %d*%d=%d\n", (int)a, (int)b, (int)res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=3;
return res;
}
static OPUS_INLINE short MULT16_16_Q13(int a, int b)
{
opus_int64 res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(b))
{
fprintf (stderr, "MULT16_16_Q13: inputs are not short: %d %d\n", a, b);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = ((opus_int64)a)*b;
res >>= 13;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "MULT16_16_Q13: output is not short: %d*%d=%d\n", a, b, (int)res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=3;
return res;
}
static OPUS_INLINE short MULT16_16_Q14(int a, int b)
{
opus_int64 res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(b))
{
fprintf (stderr, "MULT16_16_Q14: inputs are not short: %d %d\n", a, b);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = ((opus_int64)a)*b;
res >>= 14;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "MULT16_16_Q14: output is not short: %d\n", (int)res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=3;
return res;
}
#define MULT16_16_Q15(a, b) MULT16_16_Q15_(a, b, __FILE__, __LINE__)
static OPUS_INLINE short MULT16_16_Q15_(int a, int b, char *file, int line)
{
opus_int64 res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(b))
{
fprintf (stderr, "MULT16_16_Q15: inputs are not short: %d %d in %s: line %d\n", a, b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = ((opus_int64)a)*b;
res >>= 15;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "MULT16_16_Q15: output is not short: %d in %s: line %d\n", (int)res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=1;
return res;
}
static OPUS_INLINE short MULT16_16_P13(int a, int b)
{
opus_int64 res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(b))
{
fprintf (stderr, "MULT16_16_P13: inputs are not short: %d %d\n", a, b);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = ((opus_int64)a)*b;
res += 4096;
if (!VERIFY_INT(res))
{
fprintf (stderr, "MULT16_16_P13: overflow: %d*%d=%d\n", a, b, (int)res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res >>= 13;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "MULT16_16_P13: output is not short: %d*%d=%d\n", a, b, (int)res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=4;
return res;
}
static OPUS_INLINE short MULT16_16_P14(int a, int b)
{
opus_int64 res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(b))
{
fprintf (stderr, "MULT16_16_P14: inputs are not short: %d %d\n", a, b);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = ((opus_int64)a)*b;
res += 8192;
if (!VERIFY_INT(res))
{
fprintf (stderr, "MULT16_16_P14: overflow: %d*%d=%d\n", a, b, (int)res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res >>= 14;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "MULT16_16_P14: output is not short: %d*%d=%d\n", a, b, (int)res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=4;
return res;
}
static OPUS_INLINE short MULT16_16_P15(int a, int b)
{
opus_int64 res;
if (!VERIFY_SHORT(a) || !VERIFY_SHORT(b))
{
fprintf (stderr, "MULT16_16_P15: inputs are not short: %d %d\n", a, b);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = ((opus_int64)a)*b;
res += 16384;
if (!VERIFY_INT(res))
{
fprintf (stderr, "MULT16_16_P15: overflow: %d*%d=%d\n", a, b, (int)res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res >>= 15;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "MULT16_16_P15: output is not short: %d*%d=%d\n", a, b, (int)res);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=2;
return res;
}
#define DIV32_16(a, b) DIV32_16_(a, b, __FILE__, __LINE__)
static OPUS_INLINE int DIV32_16_(opus_int64 a, opus_int64 b, char *file, int line)
{
opus_int64 res;
if (b==0)
{
fprintf(stderr, "DIV32_16: divide by zero: %d/%d in %s: line %d\n", (int)a, (int)b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
return 0;
}
if (!VERIFY_INT(a) || !VERIFY_SHORT(b))
{
fprintf (stderr, "DIV32_16: inputs are not int/short: %d %d in %s: line %d\n", (int)a, (int)b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a/b;
if (!VERIFY_SHORT(res))
{
fprintf (stderr, "DIV32_16: output is not short: %d / %d = %d in %s: line %d\n", (int)a,(int)b,(int)res, file, line);
if (res>32767)
res = 32767;
if (res<-32768)
res = -32768;
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=35;
return res;
}
#define DIV32(a, b) DIV32_(a, b, __FILE__, __LINE__)
static OPUS_INLINE int DIV32_(opus_int64 a, opus_int64 b, char *file, int line)
{
opus_int64 res;
if (b==0)
{
fprintf(stderr, "DIV32: divide by zero: %d/%d in %s: line %d\n", (int)a, (int)b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
return 0;
}
if (!VERIFY_INT(a) || !VERIFY_INT(b))
{
fprintf (stderr, "DIV32: inputs are not int/short: %d %d in %s: line %d\n", (int)a, (int)b, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
res = a/b;
if (!VERIFY_INT(res))
{
fprintf (stderr, "DIV32: output is not int: %d in %s: line %d\n", (int)res, file, line);
#ifdef FIXED_DEBUG_ASSERT
celt_assert(0);
#endif
}
celt_mips+=70;
return res;
}
#undef PRINT_MIPS
#define PRINT_MIPS(file) do {fprintf (file, "total complexity = %llu MIPS\n", celt_mips);} while (0);
#endif

134
drivers/opus/celt/fixed_generic.h Normal file
View file

@ -0,0 +1,134 @@
/* Copyright (C) 2007-2009 Xiph.Org Foundation
Copyright (C) 2003-2008 Jean-Marc Valin
Copyright (C) 2007-2008 CSIRO */
/**
@file fixed_generic.h
@brief Generic fixed-point operations
*/
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef FIXED_GENERIC_H
#define FIXED_GENERIC_H
/** Multiply a 16-bit signed value by a 16-bit unsigned value. The result is a 32-bit signed value */
#define MULT16_16SU(a,b) ((opus_val32)(opus_val16)(a)*(opus_val32)(opus_uint16)(b))
/** 16x32 multiplication, followed by a 16-bit shift right. Results fits in 32 bits */
#define MULT16_32_Q16(a,b) ADD32(MULT16_16((a),SHR((b),16)), SHR(MULT16_16SU((a),((b)&0x0000ffff)),16))
/** 16x32 multiplication, followed by a 16-bit shift right (round-to-nearest). Results fits in 32 bits */
#define MULT16_32_P16(a,b) ADD32(MULT16_16((a),SHR((b),16)), PSHR(MULT16_16SU((a),((b)&0x0000ffff)),16))
/** 16x32 multiplication, followed by a 15-bit shift right. Results fits in 32 bits */
#define MULT16_32_Q15(a,b) ADD32(SHL(MULT16_16((a),SHR((b),16)),1), SHR(MULT16_16SU((a),((b)&0x0000ffff)),15))
/** 32x32 multiplication, followed by a 31-bit shift right. Results fits in 32 bits */
#define MULT32_32_Q31(a,b) ADD32(ADD32(SHL(MULT16_16(SHR((a),16),SHR((b),16)),1), SHR(MULT16_16SU(SHR((a),16),((b)&0x0000ffff)),15)), SHR(MULT16_16SU(SHR((b),16),((a)&0x0000ffff)),15))
/** Compile-time conversion of float constant to 16-bit value */
#define QCONST16(x,bits) ((opus_val16)(.5+(x)*(((opus_val32)1)<<(bits))))
/** Compile-time conversion of float constant to 32-bit value */
#define QCONST32(x,bits) ((opus_val32)(.5+(x)*(((opus_val32)1)<<(bits))))
/** Negate a 16-bit value */
#define NEG16(x) (-(x))
/** Negate a 32-bit value */
#define NEG32(x) (-(x))
/** Change a 32-bit value into a 16-bit value. The value is assumed to fit in 16-bit, otherwise the result is undefined */
#define EXTRACT16(x) ((opus_val16)(x))
/** Change a 16-bit value into a 32-bit value */
#define EXTEND32(x) ((opus_val32)(x))
/** Arithmetic shift-right of a 16-bit value */
#define SHR16(a,shift) ((a) >> (shift))
/** Arithmetic shift-left of a 16-bit value */
#define SHL16(a,shift) ((opus_int16)((opus_uint16)(a)<<(shift)))
/** Arithmetic shift-right of a 32-bit value */
#define SHR32(a,shift) ((a) >> (shift))
/** Arithmetic shift-left of a 32-bit value */
#define SHL32(a,shift) ((opus_int32)((opus_uint32)(a)<<(shift)))
/** 32-bit arithmetic shift right with rounding-to-nearest instead of rounding down */
#define PSHR32(a,shift) (SHR32((a)+((EXTEND32(1)<<((shift))>>1)),shift))
/** 32-bit arithmetic shift right where the argument can be negative */
#define VSHR32(a, shift) (((shift)>0) ? SHR32(a, shift) : SHL32(a, -(shift)))
/** "RAW" macros, should not be used outside of this header file */
#define SHR(a,shift) ((a) >> (shift))
#define SHL(a,shift) SHL32(a,shift)
#define PSHR(a,shift) (SHR((a)+((EXTEND32(1)<<((shift))>>1)),shift))
#define SATURATE(x,a) (((x)>(a) ? (a) : (x)<-(a) ? -(a) : (x)))
#define SATURATE16(x) (EXTRACT16((x)>32767 ? 32767 : (x)<-32768 ? -32768 : (x)))
/** Shift by a and round-to-neareast 32-bit value. Result is a 16-bit value */
#define ROUND16(x,a) (EXTRACT16(PSHR32((x),(a))))
/** Divide by two */
#define HALF16(x) (SHR16(x,1))
#define HALF32(x) (SHR32(x,1))
/** Add two 16-bit values */
#define ADD16(a,b) ((opus_val16)((opus_val16)(a)+(opus_val16)(b)))
/** Subtract two 16-bit values */
#define SUB16(a,b) ((opus_val16)(a)-(opus_val16)(b))
/** Add two 32-bit values */
#define ADD32(a,b) ((opus_val32)(a)+(opus_val32)(b))
/** Subtract two 32-bit values */
#define SUB32(a,b) ((opus_val32)(a)-(opus_val32)(b))
/** 16x16 multiplication where the result fits in 16 bits */
#define MULT16_16_16(a,b) ((((opus_val16)(a))*((opus_val16)(b))))
/* (opus_val32)(opus_val16) gives TI compiler a hint that it's 16x16->32 multiply */
/** 16x16 multiplication where the result fits in 32 bits */
#define MULT16_16(a,b) (((opus_val32)(opus_val16)(a))*((opus_val32)(opus_val16)(b)))
/** 16x16 multiply-add where the result fits in 32 bits */
#define MAC16_16(c,a,b) (ADD32((c),MULT16_16((a),(b))))
/** 16x32 multiply, followed by a 15-bit shift right and 32-bit add.
b must fit in 31 bits.
Result fits in 32 bits. */
#define MAC16_32_Q15(c,a,b) ADD32(c,ADD32(MULT16_16((a),SHR((b),15)), SHR(MULT16_16((a),((b)&0x00007fff)),15)))
#define MULT16_16_Q11_32(a,b) (SHR(MULT16_16((a),(b)),11))
#define MULT16_16_Q11(a,b) (SHR(MULT16_16((a),(b)),11))
#define MULT16_16_Q13(a,b) (SHR(MULT16_16((a),(b)),13))
#define MULT16_16_Q14(a,b) (SHR(MULT16_16((a),(b)),14))
#define MULT16_16_Q15(a,b) (SHR(MULT16_16((a),(b)),15))
#define MULT16_16_P13(a,b) (SHR(ADD32(4096,MULT16_16((a),(b))),13))
#define MULT16_16_P14(a,b) (SHR(ADD32(8192,MULT16_16((a),(b))),14))
#define MULT16_16_P15(a,b) (SHR(ADD32(16384,MULT16_16((a),(b))),15))
/** Divide a 32-bit value by a 16-bit value. Result fits in 16 bits */
#define DIV32_16(a,b) ((opus_val16)(((opus_val32)(a))/((opus_val16)(b))))
/** Divide a 32-bit value by a 32-bit value. Result fits in 32 bits */
#define DIV32(a,b) (((opus_val32)(a))/((opus_val32)(b)))
#endif

140
drivers/opus/celt/float_cast.h Normal file
View file

@ -0,0 +1,140 @@
/* Copyright (C) 2001 Erik de Castro Lopo <erikd AT mega-nerd DOT com> */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/* Version 1.1 */
#ifndef FLOAT_CAST_H
#define FLOAT_CAST_H
#include "arch.h"
/*============================================================================
** On Intel Pentium processors (especially PIII and probably P4), converting
** from float to int is very slow. To meet the C specs, the code produced by
** most C compilers targeting Pentium needs to change the FPU rounding mode
** before the float to int conversion is performed.
**
** Changing the FPU rounding mode causes the FPU pipeline to be flushed. It
** is this flushing of the pipeline which is so slow.
**
** Fortunately the ISO C99 specifications define the functions lrint, lrintf,
** llrint and llrintf which fix this problem as a side effect.
**
** On Unix-like systems, the configure process should have detected the
** presence of these functions. If they weren't found we have to replace them
** here with a standard C cast.
*/
/*
** The C99 prototypes for lrint and lrintf are as follows:
**
** long int lrintf (float x) ;
** long int lrint (double x) ;
*/
/* The presence of the required functions are detected during the configure
** process and the values HAVE_LRINT and HAVE_LRINTF are set accordingly in
** the config.h file.
*/
#if (HAVE_LRINTF)
/* These defines enable functionality introduced with the 1999 ISO C
** standard. They must be defined before the inclusion of math.h to
** engage them. If optimisation is enabled, these functions will be
** inlined. With optimisation switched off, you have to link in the
** maths library using -lm.
*/
#define _ISOC9X_SOURCE 1
#define _ISOC99_SOURCE 1
#define __USE_ISOC9X 1
#define __USE_ISOC99 1
#include <math.h>
#define float2int(x) lrintf(x)
#elif (defined(HAVE_LRINT))
#define _ISOC9X_SOURCE 1
#define _ISOC99_SOURCE 1
#define __USE_ISOC9X 1
#define __USE_ISOC99 1
#include <math.h>
#define float2int(x) lrint(x)
#elif (defined(_MSC_VER) && _MSC_VER >= 1400) && (defined (WIN64) || defined (_WIN64))
#include <xmmintrin.h>
__inline long int float2int(float value)
{
return _mm_cvtss_si32(_mm_load_ss(&value));
}
#elif (defined(_MSC_VER) && _MSC_VER >= 1400) && (defined (WIN32) || defined (_WIN32))
#include <math.h>
/* Win32 doesn't seem to have these functions.
** Therefore implement OPUS_INLINE versions of these functions here.
*/
__inline long int
float2int (float flt)
{ int intgr;
_asm
{ fld flt
fistp intgr
} ;
return intgr ;
}
#else
#if (defined(__GNUC__) && defined(__STDC__) && __STDC__ && __STDC_VERSION__ >= 199901L)
/* supported by gcc in C99 mode, but not by all other compilers */
#warning "Don't have the functions lrint() and lrintf ()."
#warning "Replacing these functions with a standard C cast."
#endif /* __STDC_VERSION__ >= 199901L */
#include <math.h>
#define float2int(flt) ((int)(floor(.5+flt)))
#endif
#ifndef DISABLE_FLOAT_API
static OPUS_INLINE opus_int16 FLOAT2INT16(float x)
{
x = x*CELT_SIG_SCALE;
x = MAX32(x, -32768);
x = MIN32(x, 32767);
return (opus_int16)float2int(x);
}
#endif /* DISABLE_FLOAT_API */
#endif /* FLOAT_CAST_H */

719
drivers/opus/celt/kiss_fft.c Normal file
View file

@ -0,0 +1,719 @@
/*Copyright (c) 2003-2004, Mark Borgerding
Lots of modifications by Jean-Marc Valin
Copyright (c) 2005-2007, Xiph.Org Foundation
Copyright (c) 2008, Xiph.Org Foundation, CSIRO
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice,
this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.*/
/* This code is originally from Mark Borgerding's KISS-FFT but has been
heavily modified to better suit Opus */
#ifndef SKIP_CONFIG_H
# ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
# endif
#endif
#include "_kiss_fft_guts.h"
#include "arch.h"
#include "os_support.h"
#include "mathops.h"
#include "stack_alloc.h"
/* The guts header contains all the multiplication and addition macros that are defined for
complex numbers. It also delares the kf_ internal functions.
*/
static void kf_bfly2(
kiss_fft_cpx * Fout,
const size_t fstride,
const kiss_fft_state *st,
int m,
int N,
int mm
)
{
kiss_fft_cpx * Fout2;
const kiss_twiddle_cpx * tw1;
int i,j;
kiss_fft_cpx * Fout_beg = Fout;
for (i=0;i<N;i++)
{
Fout = Fout_beg + i*mm;
Fout2 = Fout + m;
tw1 = st->twiddles;
for(j=0;j<m;j++)
{
kiss_fft_cpx t;
Fout->r = SHR32(Fout->r, 1);Fout->i = SHR32(Fout->i, 1);
Fout2->r = SHR32(Fout2->r, 1);Fout2->i = SHR32(Fout2->i, 1);
C_MUL (t, *Fout2 , *tw1);
tw1 += fstride;
C_SUB( *Fout2 , *Fout , t );
C_ADDTO( *Fout , t );
++Fout2;
++Fout;
}
}
}
static void ki_bfly2(
kiss_fft_cpx * Fout,
const size_t fstride,
const kiss_fft_state *st,
int m,
int N,
int mm
)
{
kiss_fft_cpx * Fout2;
const kiss_twiddle_cpx * tw1;
kiss_fft_cpx t;
int i,j;
kiss_fft_cpx * Fout_beg = Fout;
for (i=0;i<N;i++)
{
Fout = Fout_beg + i*mm;
Fout2 = Fout + m;
tw1 = st->twiddles;
for(j=0;j<m;j++)
{
C_MULC (t, *Fout2 , *tw1);
tw1 += fstride;
C_SUB( *Fout2 , *Fout , t );
C_ADDTO( *Fout , t );
++Fout2;
++Fout;
}
}
}
static void kf_bfly4(
kiss_fft_cpx * Fout,
const size_t fstride,
const kiss_fft_state *st,
int m,
int N,
int mm
)
{
const kiss_twiddle_cpx *tw1,*tw2,*tw3;
kiss_fft_cpx scratch[6];
const size_t m2=2*m;
const size_t m3=3*m;
int i, j;
kiss_fft_cpx * Fout_beg = Fout;
for (i=0;i<N;i++)
{
Fout = Fout_beg + i*mm;
tw3 = tw2 = tw1 = st->twiddles;
for (j=0;j<m;j++)
{
C_MUL4(scratch[0],Fout[m] , *tw1 );
C_MUL4(scratch[1],Fout[m2] , *tw2 );
C_MUL4(scratch[2],Fout[m3] , *tw3 );
Fout->r = PSHR32(Fout->r, 2);
Fout->i = PSHR32(Fout->i, 2);
C_SUB( scratch[5] , *Fout, scratch[1] );
C_ADDTO(*Fout, scratch[1]);
C_ADD( scratch[3] , scratch[0] , scratch[2] );
C_SUB( scratch[4] , scratch[0] , scratch[2] );
C_SUB( Fout[m2], *Fout, scratch[3] );
tw1 += fstride;
tw2 += fstride*2;
tw3 += fstride*3;
C_ADDTO( *Fout , scratch[3] );
Fout[m].r = scratch[5].r + scratch[4].i;
Fout[m].i = scratch[5].i - scratch[4].r;
Fout[m3].r = scratch[5].r - scratch[4].i;
Fout[m3].i = scratch[5].i + scratch[4].r;
++Fout;
}
}
}
static void ki_bfly4(
kiss_fft_cpx * Fout,
const size_t fstride,
const kiss_fft_state *st,
int m,
int N,
int mm
)
{
const kiss_twiddle_cpx *tw1,*tw2,*tw3;
kiss_fft_cpx scratch[6];
const size_t m2=2*m;
const size_t m3=3*m;
int i, j;
kiss_fft_cpx * Fout_beg = Fout;
for (i=0;i<N;i++)
{
Fout = Fout_beg + i*mm;
tw3 = tw2 = tw1 = st->twiddles;
for (j=0;j<m;j++)
{
C_MULC(scratch[0],Fout[m] , *tw1 );
C_MULC(scratch[1],Fout[m2] , *tw2 );
C_MULC(scratch[2],Fout[m3] , *tw3 );
C_SUB( scratch[5] , *Fout, scratch[1] );
C_ADDTO(*Fout, scratch[1]);
C_ADD( scratch[3] , scratch[0] , scratch[2] );
C_SUB( scratch[4] , scratch[0] , scratch[2] );
C_SUB( Fout[m2], *Fout, scratch[3] );
tw1 += fstride;
tw2 += fstride*2;
tw3 += fstride*3;
C_ADDTO( *Fout , scratch[3] );
Fout[m].r = scratch[5].r - scratch[4].i;
Fout[m].i = scratch[5].i + scratch[4].r;
Fout[m3].r = scratch[5].r + scratch[4].i;
Fout[m3].i = scratch[5].i - scratch[4].r;
++Fout;
}
}
}
#ifndef RADIX_TWO_ONLY
static void kf_bfly3(
kiss_fft_cpx * Fout,
const size_t fstride,
const kiss_fft_state *st,
int m,
int N,
int mm
)
{
int i;
size_t k;
const size_t m2 = 2*m;
const kiss_twiddle_cpx *tw1,*tw2;
kiss_fft_cpx scratch[5];
kiss_twiddle_cpx epi3;
kiss_fft_cpx * Fout_beg = Fout;
epi3 = st->twiddles[fstride*m];
for (i=0;i<N;i++)
{
Fout = Fout_beg + i*mm;
tw1=tw2=st->twiddles;
k=m;
do {
C_FIXDIV(*Fout,3); C_FIXDIV(Fout[m],3); C_FIXDIV(Fout[m2],3);
C_MUL(scratch[1],Fout[m] , *tw1);
C_MUL(scratch[2],Fout[m2] , *tw2);
C_ADD(scratch[3],scratch[1],scratch[2]);
C_SUB(scratch[0],scratch[1],scratch[2]);
tw1 += fstride;
tw2 += fstride*2;
Fout[m].r = Fout->r - HALF_OF(scratch[3].r);
Fout[m].i = Fout->i - HALF_OF(scratch[3].i);
C_MULBYSCALAR( scratch[0] , epi3.i );
C_ADDTO(*Fout,scratch[3]);
Fout[m2].r = Fout[m].r + scratch[0].i;
Fout[m2].i = Fout[m].i - scratch[0].r;
Fout[m].r -= scratch[0].i;
Fout[m].i += scratch[0].r;
++Fout;
} while(--k);
}
}
static void ki_bfly3(
kiss_fft_cpx * Fout,
const size_t fstride,
const kiss_fft_state *st,
int m,
int N,
int mm
)
{
int i, k;
const size_t m2 = 2*m;
const kiss_twiddle_cpx *tw1,*tw2;
kiss_fft_cpx scratch[5];
kiss_twiddle_cpx epi3;
kiss_fft_cpx * Fout_beg = Fout;
epi3 = st->twiddles[fstride*m];
for (i=0;i<N;i++)
{
Fout = Fout_beg + i*mm;
tw1=tw2=st->twiddles;
k=m;
do{
C_MULC(scratch[1],Fout[m] , *tw1);
C_MULC(scratch[2],Fout[m2] , *tw2);
C_ADD(scratch[3],scratch[1],scratch[2]);
C_SUB(scratch[0],scratch[1],scratch[2]);
tw1 += fstride;
tw2 += fstride*2;
Fout[m].r = Fout->r - HALF_OF(scratch[3].r);
Fout[m].i = Fout->i - HALF_OF(scratch[3].i);
C_MULBYSCALAR( scratch[0] , -epi3.i );
C_ADDTO(*Fout,scratch[3]);
Fout[m2].r = Fout[m].r + scratch[0].i;
Fout[m2].i = Fout[m].i - scratch[0].r;
Fout[m].r -= scratch[0].i;
Fout[m].i += scratch[0].r;
++Fout;
}while(--k);
}
}
static void kf_bfly5(
kiss_fft_cpx * Fout,
const size_t fstride,
const kiss_fft_state *st,
int m,
int N,
int mm
)
{
kiss_fft_cpx *Fout0,*Fout1,*Fout2,*Fout3,*Fout4;
int i, u;
kiss_fft_cpx scratch[13];
const kiss_twiddle_cpx * twiddles = st->twiddles;
const kiss_twiddle_cpx *tw;
kiss_twiddle_cpx ya,yb;
kiss_fft_cpx * Fout_beg = Fout;
ya = twiddles[fstride*m];
yb = twiddles[fstride*2*m];
tw=st->twiddles;
for (i=0;i<N;i++)
{
Fout = Fout_beg + i*mm;
Fout0=Fout;
Fout1=Fout0+m;
Fout2=Fout0+2*m;
Fout3=Fout0+3*m;
Fout4=Fout0+4*m;
for ( u=0; u<m; ++u ) {
C_FIXDIV( *Fout0,5); C_FIXDIV( *Fout1,5); C_FIXDIV( *Fout2,5); C_FIXDIV( *Fout3,5); C_FIXDIV( *Fout4,5);
scratch[0] = *Fout0;
C_MUL(scratch[1] ,*Fout1, tw[u*fstride]);
C_MUL(scratch[2] ,*Fout2, tw[2*u*fstride]);
C_MUL(scratch[3] ,*Fout3, tw[3*u*fstride]);
C_MUL(scratch[4] ,*Fout4, tw[4*u*fstride]);
C_ADD( scratch[7],scratch[1],scratch[4]);
C_SUB( scratch[10],scratch[1],scratch[4]);
C_ADD( scratch[8],scratch[2],scratch[3]);
C_SUB( scratch[9],scratch[2],scratch[3]);
Fout0->r += scratch[7].r + scratch[8].r;
Fout0->i += scratch[7].i + scratch[8].i;
scratch[5].r = scratch[0].r + S_MUL(scratch[7].r,ya.r) + S_MUL(scratch[8].r,yb.r);
scratch[5].i = scratch[0].i + S_MUL(scratch[7].i,ya.r) + S_MUL(scratch[8].i,yb.r);
scratch[6].r = S_MUL(scratch[10].i,ya.i) + S_MUL(scratch[9].i,yb.i);
scratch[6].i = -S_MUL(scratch[10].r,ya.i) - S_MUL(scratch[9].r,yb.i);
C_SUB(*Fout1,scratch[5],scratch[6]);
C_ADD(*Fout4,scratch[5],scratch[6]);
scratch[11].r = scratch[0].r + S_MUL(scratch[7].r,yb.r) + S_MUL(scratch[8].r,ya.r);
scratch[11].i = scratch[0].i + S_MUL(scratch[7].i,yb.r) + S_MUL(scratch[8].i,ya.r);
scratch[12].r = - S_MUL(scratch[10].i,yb.i) + S_MUL(scratch[9].i,ya.i);
scratch[12].i = S_MUL(scratch[10].r,yb.i) - S_MUL(scratch[9].r,ya.i);
C_ADD(*Fout2,scratch[11],scratch[12]);
C_SUB(*Fout3,scratch[11],scratch[12]);
++Fout0;++Fout1;++Fout2;++Fout3;++Fout4;
}
}
}
static void ki_bfly5(
kiss_fft_cpx * Fout,
const size_t fstride,
const kiss_fft_state *st,
int m,
int N,
int mm
)
{
kiss_fft_cpx *Fout0,*Fout1,*Fout2,*Fout3,*Fout4;
int i, u;
kiss_fft_cpx scratch[13];
const kiss_twiddle_cpx * twiddles = st->twiddles;
const kiss_twiddle_cpx *tw;
kiss_twiddle_cpx ya,yb;
kiss_fft_cpx * Fout_beg = Fout;
ya = twiddles[fstride*m];
yb = twiddles[fstride*2*m];
tw=st->twiddles;
for (i=0;i<N;i++)
{
Fout = Fout_beg + i*mm;
Fout0=Fout;
Fout1=Fout0+m;
Fout2=Fout0+2*m;
Fout3=Fout0+3*m;
Fout4=Fout0+4*m;
for ( u=0; u<m; ++u ) {
scratch[0] = *Fout0;
C_MULC(scratch[1] ,*Fout1, tw[u*fstride]);
C_MULC(scratch[2] ,*Fout2, tw[2*u*fstride]);
C_MULC(scratch[3] ,*Fout3, tw[3*u*fstride]);
C_MULC(scratch[4] ,*Fout4, tw[4*u*fstride]);
C_ADD( scratch[7],scratch[1],scratch[4]);
C_SUB( scratch[10],scratch[1],scratch[4]);
C_ADD( scratch[8],scratch[2],scratch[3]);
C_SUB( scratch[9],scratch[2],scratch[3]);
Fout0->r += scratch[7].r + scratch[8].r;
Fout0->i += scratch[7].i + scratch[8].i;
scratch[5].r = scratch[0].r + S_MUL(scratch[7].r,ya.r) + S_MUL(scratch[8].r,yb.r);
scratch[5].i = scratch[0].i + S_MUL(scratch[7].i,ya.r) + S_MUL(scratch[8].i,yb.r);
scratch[6].r = -S_MUL(scratch[10].i,ya.i) - S_MUL(scratch[9].i,yb.i);
scratch[6].i = S_MUL(scratch[10].r,ya.i) + S_MUL(scratch[9].r,yb.i);
C_SUB(*Fout1,scratch[5],scratch[6]);
C_ADD(*Fout4,scratch[5],scratch[6]);
scratch[11].r = scratch[0].r + S_MUL(scratch[7].r,yb.r) + S_MUL(scratch[8].r,ya.r);
scratch[11].i = scratch[0].i + S_MUL(scratch[7].i,yb.r) + S_MUL(scratch[8].i,ya.r);
scratch[12].r = S_MUL(scratch[10].i,yb.i) - S_MUL(scratch[9].i,ya.i);
scratch[12].i = -S_MUL(scratch[10].r,yb.i) + S_MUL(scratch[9].r,ya.i);
C_ADD(*Fout2,scratch[11],scratch[12]);
C_SUB(*Fout3,scratch[11],scratch[12]);
++Fout0;++Fout1;++Fout2;++Fout3;++Fout4;
}
}
}
#endif
#ifdef CUSTOM_MODES
static
void compute_bitrev_table(
int Fout,
opus_int16 *f,
const size_t fstride,
int in_stride,
opus_int16 * factors,
const kiss_fft_state *st
)
{
const int p=*factors++; /* the radix */
const int m=*factors++; /* stage's fft length/p */
/*printf ("fft %d %d %d %d %d %d\n", p*m, m, p, s2, fstride*in_stride, N);*/
if (m==1)
{
int j;
for (j=0;j<p;j++)
{
*f = Fout+j;
f += fstride*in_stride;
}
} else {
int j;
for (j=0;j<p;j++)
{
compute_bitrev_table( Fout , f, fstride*p, in_stride, factors,st);
f += fstride*in_stride;
Fout += m;
}
}
}
/* facbuf is populated by p1,m1,p2,m2, ...
where
p[i] * m[i] = m[i-1]
m0 = n */
static
int kf_factor(int n,opus_int16 * facbuf)
{
int p=4;
/*factor out powers of 4, powers of 2, then any remaining primes */
do {
while (n % p) {
switch (p) {
case 4: p = 2; break;
case 2: p = 3; break;
default: p += 2; break;
}
if (p>32000 || (opus_int32)p*(opus_int32)p > n)
p = n; /* no more factors, skip to end */
}
n /= p;
#ifdef RADIX_TWO_ONLY
if (p!=2 && p != 4)
#else
if (p>5)
#endif
{
return 0;
}
*facbuf++ = p;
*facbuf++ = n;
} while (n > 1);
return 1;
}
static void compute_twiddles(kiss_twiddle_cpx *twiddles, int nfft)
{
int i;
#ifdef OPUS_FIXED_POINT
for (i=0;i<nfft;++i) {
opus_val32 phase = -i;
kf_cexp2(twiddles+i, DIV32(SHL32(phase,17),nfft));
}
#else
for (i=0;i<nfft;++i) {
const double pi=3.14159265358979323846264338327;
double phase = ( -2*pi /nfft ) * i;
kf_cexp(twiddles+i, phase );
}
#endif
}
/*
*
* Allocates all necessary storage space for the fft and ifft.
* The return value is a contiguous block of memory. As such,
* It can be freed with free().
* */
kiss_fft_state *opus_fft_alloc_twiddles(int nfft,void * mem,size_t * lenmem, const kiss_fft_state *base)
{
kiss_fft_state *st=NULL;
size_t memneeded = sizeof(struct kiss_fft_state); /* twiddle factors*/
if ( lenmem==NULL ) {
st = ( kiss_fft_state*)KISS_FFT_MALLOC( memneeded );
}else{
if (mem != NULL && *lenmem >= memneeded)
st = (kiss_fft_state*)mem;
*lenmem = memneeded;
}
if (st) {
opus_int16 *bitrev;
kiss_twiddle_cpx *twiddles;
st->nfft=nfft;
#ifndef OPUS_FIXED_POINT
st->scale = 1.f/nfft;
#endif
if (base != NULL)
{
st->twiddles = base->twiddles;
st->shift = 0;
while (nfft<<st->shift != base->nfft && st->shift < 32)
st->shift++;
if (st->shift>=32)
goto fail;
} else {
st->twiddles = twiddles = (kiss_twiddle_cpx*)KISS_FFT_MALLOC(sizeof(kiss_twiddle_cpx)*nfft);
compute_twiddles(twiddles, nfft);
st->shift = -1;
}
if (!kf_factor(nfft,st->factors))
{
goto fail;
}
/* bitrev */
st->bitrev = bitrev = (opus_int16*)KISS_FFT_MALLOC(sizeof(opus_int16)*nfft);
if (st->bitrev==NULL)
goto fail;
compute_bitrev_table(0, bitrev, 1,1, st->factors,st);
}
return st;
fail:
opus_fft_free(st);
return NULL;
}
kiss_fft_state *opus_fft_alloc(int nfft,void * mem,size_t * lenmem )
{
return opus_fft_alloc_twiddles(nfft, mem, lenmem, NULL);
}
void opus_fft_free(const kiss_fft_state *cfg)
{
if (cfg)
{
opus_free((opus_int16*)cfg->bitrev);
if (cfg->shift < 0)
opus_free((kiss_twiddle_cpx*)cfg->twiddles);
opus_free((kiss_fft_state*)cfg);
}
}
#endif /* CUSTOM_MODES */
void opus_fft(const kiss_fft_state *st,const kiss_fft_cpx *fin,kiss_fft_cpx *fout)
{
int m2, m;
int p;
int L;
int fstride[MAXFACTORS];
int i;
int shift;
/* st->shift can be -1 */
shift = st->shift>0 ? st->shift : 0;
celt_assert2 (fin != fout, "In-place FFT not supported");
/* Bit-reverse the input */
for (i=0;i<st->nfft;i++)
{
fout[st->bitrev[i]] = fin[i];
#ifndef OPUS_FIXED_POINT
fout[st->bitrev[i]].r *= st->scale;
fout[st->bitrev[i]].i *= st->scale;
#endif
}
fstride[0] = 1;
L=0;
do {
p = st->factors[2*L];
m = st->factors[2*L+1];
fstride[L+1] = fstride[L]*p;
L++;
} while(m!=1);
m = st->factors[2*L-1];
for (i=L-1;i>=0;i--)
{
if (i!=0)
m2 = st->factors[2*i-1];
else
m2 = 1;
switch (st->factors[2*i])
{
case 2:
kf_bfly2(fout,fstride[i]<<shift,st,m, fstride[i], m2);
break;
case 4:
kf_bfly4(fout,fstride[i]<<shift,st,m, fstride[i], m2);
break;
#ifndef RADIX_TWO_ONLY
case 3:
kf_bfly3(fout,fstride[i]<<shift,st,m, fstride[i], m2);
break;
case 5:
kf_bfly5(fout,fstride[i]<<shift,st,m, fstride[i], m2);
break;
#endif
}
m = m2;
}
}
void opus_ifft(const kiss_fft_state *st,const kiss_fft_cpx *fin,kiss_fft_cpx *fout)
{
int m2, m;
int p;
int L;
int fstride[MAXFACTORS];
int i;
int shift;
/* st->shift can be -1 */
shift = st->shift>0 ? st->shift : 0;
celt_assert2 (fin != fout, "In-place FFT not supported");
/* Bit-reverse the input */
for (i=0;i<st->nfft;i++)
fout[st->bitrev[i]] = fin[i];
fstride[0] = 1;
L=0;
do {
p = st->factors[2*L];
m = st->factors[2*L+1];
fstride[L+1] = fstride[L]*p;
L++;
} while(m!=1);
m = st->factors[2*L-1];
for (i=L-1;i>=0;i--)
{
if (i!=0)
m2 = st->factors[2*i-1];
else
m2 = 1;
switch (st->factors[2*i])
{
case 2:
ki_bfly2(fout,fstride[i]<<shift,st,m, fstride[i], m2);
break;
case 4:
ki_bfly4(fout,fstride[i]<<shift,st,m, fstride[i], m2);
break;
#ifndef RADIX_TWO_ONLY
case 3:
ki_bfly3(fout,fstride[i]<<shift,st,m, fstride[i], m2);
break;
case 5:
ki_bfly5(fout,fstride[i]<<shift,st,m, fstride[i], m2);
break;
#endif
}
m = m2;
}
}

139
drivers/opus/celt/kiss_fft.h Normal file
View file

@ -0,0 +1,139 @@
/*Copyright (c) 2003-2004, Mark Borgerding
Lots of modifications by Jean-Marc Valin
Copyright (c) 2005-2007, Xiph.Org Foundation
Copyright (c) 2008, Xiph.Org Foundation, CSIRO
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met:
* Redistributions of source code must retain the above copyright notice,
this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright notice,
this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.*/
#ifndef KISS_FFT_H
#define KISS_FFT_H
#include <stdlib.h>
#include <math.h>
#include "arch.h"
#ifdef __cplusplus
extern "C" {
#endif
#ifdef USE_SIMD
# include <xmmintrin.h>
# define kiss_fft_scalar __m128
#define KISS_FFT_MALLOC(nbytes) memalign(16,nbytes)
#else
#define KISS_FFT_MALLOC opus_alloc
#endif
#ifdef OPUS_FIXED_POINT
#include "arch.h"
# define kiss_fft_scalar opus_int32
# define kiss_twiddle_scalar opus_int16
#else
# ifndef kiss_fft_scalar
/* default is float */
# define kiss_fft_scalar float
# define kiss_twiddle_scalar float
# define KF_SUFFIX _celt_single
# endif
#endif
typedef struct {
kiss_fft_scalar r;
kiss_fft_scalar i;
}kiss_fft_cpx;
typedef struct {
kiss_twiddle_scalar r;
kiss_twiddle_scalar i;
}kiss_twiddle_cpx;
#define MAXFACTORS 8
/* e.g. an fft of length 128 has 4 factors
as far as kissfft is concerned
4*4*4*2
*/
typedef struct kiss_fft_state{
int nfft;
#ifndef OPUS_FIXED_POINT
kiss_fft_scalar scale;
#endif
int shift;
opus_int16 factors[2*MAXFACTORS];
const opus_int16 *bitrev;
const kiss_twiddle_cpx *twiddles;
} kiss_fft_state;
/*typedef struct kiss_fft_state* kiss_fft_cfg;*/
/**
* opus_fft_alloc
*
* Initialize a FFT (or IFFT) algorithm's cfg/state buffer.
*
* typical usage: kiss_fft_cfg mycfg=opus_fft_alloc(1024,0,NULL,NULL);
*
* The return value from fft_alloc is a cfg buffer used internally
* by the fft routine or NULL.
*
* If lenmem is NULL, then opus_fft_alloc will allocate a cfg buffer using malloc.
* The returned value should be free()d when done to avoid memory leaks.
*
* The state can be placed in a user supplied buffer 'mem':
* If lenmem is not NULL and mem is not NULL and *lenmem is large enough,
* then the function places the cfg in mem and the size used in *lenmem
* and returns mem.
*
* If lenmem is not NULL and ( mem is NULL or *lenmem is not large enough),
* then the function returns NULL and places the minimum cfg
* buffer size in *lenmem.
* */
kiss_fft_state *opus_fft_alloc_twiddles(int nfft,void * mem,size_t * lenmem, const kiss_fft_state *base);
kiss_fft_state *opus_fft_alloc(int nfft,void * mem,size_t * lenmem);
/**
* opus_fft(cfg,in_out_buf)
*
* Perform an FFT on a complex input buffer.
* for a forward FFT,
* fin should be f[0] , f[1] , ... ,f[nfft-1]
* fout will be F[0] , F[1] , ... ,F[nfft-1]
* Note that each element is complex and can be accessed like
f[k].r and f[k].i
* */
void opus_fft(const kiss_fft_state *cfg,const kiss_fft_cpx *fin,kiss_fft_cpx *fout);
void opus_ifft(const kiss_fft_state *cfg,const kiss_fft_cpx *fin,kiss_fft_cpx *fout);
void opus_fft_free(const kiss_fft_state *cfg);
#ifdef __cplusplus
}
#endif
#endif

134
drivers/opus/celt/laplace.c Normal file
View file

@ -0,0 +1,134 @@
/* Copyright (c) 2007 CSIRO
Copyright (c) 2007-2009 Xiph.Org Foundation
Written by Jean-Marc Valin */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#include "laplace.h"
#include "mathops.h"
/* The minimum probability of an energy delta (out of 32768). */
#define LAPLACE_LOG_MINP (0)
#define LAPLACE_MINP (1<<LAPLACE_LOG_MINP)
/* The minimum number of guaranteed representable energy deltas (in one
direction). */
#define LAPLACE_NMIN (16)
/* When called, decay is positive and at most 11456. */
static unsigned ec_laplace_get_freq1(unsigned fs0, int decay)
{
unsigned ft;
ft = 32768 - LAPLACE_MINP*(2*LAPLACE_NMIN) - fs0;
return ft*(opus_int32)(16384-decay)>>15;
}
void ec_laplace_encode(ec_enc *enc, int *value, unsigned fs, int decay)
{
unsigned fl;
int val = *value;
fl = 0;
if (val)
{
int s;
int i;
s = -(val<0);
val = (val+s)^s;
fl = fs;
fs = ec_laplace_get_freq1(fs, decay);
/* Search the decaying part of the PDF.*/
for (i=1; fs > 0 && i < val; i++)
{
fs *= 2;
fl += fs+2*LAPLACE_MINP;
fs = (fs*(opus_int32)decay)>>15;
}
/* Everything beyond that has probability LAPLACE_MINP. */
if (!fs)
{
int di;
int ndi_max;
ndi_max = (32768-fl+LAPLACE_MINP-1)>>LAPLACE_LOG_MINP;
ndi_max = (ndi_max-s)>>1;
di = IMIN(val - i, ndi_max - 1);
fl += (2*di+1+s)*LAPLACE_MINP;
fs = IMIN(LAPLACE_MINP, 32768-fl);
*value = (i+di+s)^s;
}
else
{
fs += LAPLACE_MINP;
fl += fs&~s;
}
celt_assert(fl+fs<=32768);
celt_assert(fs>0);
}
ec_encode_bin(enc, fl, fl+fs, 15);
}
int ec_laplace_decode(ec_dec *dec, unsigned fs, int decay)
{
int val=0;
unsigned fl;
unsigned fm;
fm = ec_decode_bin(dec, 15);
fl = 0;
if (fm >= fs)
{
val++;
fl = fs;
fs = ec_laplace_get_freq1(fs, decay)+LAPLACE_MINP;
/* Search the decaying part of the PDF.*/
while(fs > LAPLACE_MINP && fm >= fl+2*fs)
{
fs *= 2;
fl += fs;
fs = ((fs-2*LAPLACE_MINP)*(opus_int32)decay)>>15;
fs += LAPLACE_MINP;
val++;
}
/* Everything beyond that has probability LAPLACE_MINP. */
if (fs <= LAPLACE_MINP)
{
int di;
di = (fm-fl)>>(LAPLACE_LOG_MINP+1);
val += di;
fl += 2*di*LAPLACE_MINP;
}
if (fm < fl+fs)
val = -val;
else
fl += fs;
}
celt_assert(fl<32768);
celt_assert(fs>0);
celt_assert(fl<=fm);
celt_assert(fm<IMIN(fl+fs,32768));
ec_dec_update(dec, fl, IMIN(fl+fs,32768), 32768);
return val;
}

48
drivers/opus/celt/laplace.h Normal file
View file

@ -0,0 +1,48 @@
/* Copyright (c) 2007 CSIRO
Copyright (c) 2007-2009 Xiph.Org Foundation
Written by Jean-Marc Valin */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include "entenc.h"
#include "entdec.h"
/** Encode a value that is assumed to be the realisation of a
Laplace-distributed random process
@param enc Entropy encoder state
@param value Value to encode
@param fs Probability of 0, multiplied by 32768
@param decay Probability of the value +/- 1, multiplied by 16384
*/
void ec_laplace_encode(ec_enc *enc, int *value, unsigned fs, int decay);
/** Decode a value that is assumed to be the realisation of a
Laplace-distributed random process
@param dec Entropy decoder state
@param fs Probability of 0, multiplied by 32768
@param decay Probability of the value +/- 1, multiplied by 16384
@return Value decoded
*/
int ec_laplace_decode(ec_dec *dec, unsigned fs, int decay);

208
drivers/opus/celt/mathops.c Normal file
View file

@ -0,0 +1,208 @@
/* Copyright (c) 2002-2008 Jean-Marc Valin
Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2009 Xiph.Org Foundation
Written by Jean-Marc Valin */
/**
@file mathops.h
@brief Various math functions
*/
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#include "mathops.h"
/*Compute floor(sqrt(_val)) with exact arithmetic.
This has been tested on all possible 32-bit inputs.*/
unsigned isqrt32(opus_uint32 _val){
unsigned b;
unsigned g;
int bshift;
/*Uses the second method from
http://www.azillionmonkeys.com/qed/sqroot.html
The main idea is to search for the largest binary digit b such that
(g+b)*(g+b) <= _val, and add it to the solution g.*/
g=0;
bshift=(EC_ILOG(_val)-1)>>1;
b=1U<<bshift;
do{
opus_uint32 t;
t=(((opus_uint32)g<<1)+b)<<bshift;
if(t<=_val){
g+=b;
_val-=t;
}
b>>=1;
bshift--;
}
while(bshift>=0);
return g;
}
#ifdef OPUS_FIXED_POINT
opus_val32 frac_div32(opus_val32 a, opus_val32 b)
{
opus_val16 rcp;
opus_val32 result, rem;
int shift = celt_ilog2(b)-29;
a = VSHR32(a,shift);
b = VSHR32(b,shift);
/* 16-bit reciprocal */
rcp = ROUND16(celt_rcp(ROUND16(b,16)),3);
result = MULT16_32_Q15(rcp, a);
rem = PSHR32(a,2)-MULT32_32_Q31(result, b);
result = ADD32(result, SHL32(MULT16_32_Q15(rcp, rem),2));
if (result >= 536870912) /* 2^29 */
return 2147483647; /* 2^31 - 1 */
else if (result <= -536870912) /* -2^29 */
return -2147483647; /* -2^31 */
else
return SHL32(result, 2);
}
/** Reciprocal sqrt approximation in the range [0.25,1) (Q16 in, Q14 out) */
opus_val16 celt_rsqrt_norm(opus_val32 x)
{
opus_val16 n;
opus_val16 r;
opus_val16 r2;
opus_val16 y;
/* Range of n is [-16384,32767] ([-0.5,1) in Q15). */
n = x-32768;
/* Get a rough initial guess for the root.
The optimal minimax quadratic approximation (using relative error) is
r = 1.437799046117536+n*(-0.823394375837328+n*0.4096419668459485).
Coefficients here, and the final result r, are Q14.*/
r = ADD16(23557, MULT16_16_Q15(n, ADD16(-13490, MULT16_16_Q15(n, 6713))));
/* We want y = x*r*r-1 in Q15, but x is 32-bit Q16 and r is Q14.
We can compute the result from n and r using Q15 multiplies with some
adjustment, carefully done to avoid overflow.
Range of y is [-1564,1594]. */
r2 = MULT16_16_Q15(r, r);
y = SHL16(SUB16(ADD16(MULT16_16_Q15(r2, n), r2), 16384), 1);
/* Apply a 2nd-order Householder iteration: r += r*y*(y*0.375-0.5).
This yields the Q14 reciprocal square root of the Q16 x, with a maximum
relative error of 1.04956E-4, a (relative) RMSE of 2.80979E-5, and a
peak absolute error of 2.26591/16384. */
return ADD16(r, MULT16_16_Q15(r, MULT16_16_Q15(y,
SUB16(MULT16_16_Q15(y, 12288), 16384))));
}
/** Sqrt approximation (QX input, QX/2 output) */
opus_val32 celt_sqrt(opus_val32 x)
{
int k;
opus_val16 n;
opus_val32 rt;
static const opus_val16 C[5] = {23175, 11561, -3011, 1699, -664};
if (x==0)
return 0;
else if (x>=1073741824)
return 32767;
k = (celt_ilog2(x)>>1)-7;
x = VSHR32(x, 2*k);
n = x-32768;
rt = ADD16(C[0], MULT16_16_Q15(n, ADD16(C[1], MULT16_16_Q15(n, ADD16(C[2],
MULT16_16_Q15(n, ADD16(C[3], MULT16_16_Q15(n, (C[4])))))))));
rt = VSHR32(rt,7-k);
return rt;
}
#define L1 32767
#define L2 -7651
#define L3 8277
#define L4 -626
static OPUS_INLINE opus_val16 _celt_cos_pi_2(opus_val16 x)
{
opus_val16 x2;
x2 = MULT16_16_P15(x,x);
return ADD16(1,MIN16(32766,ADD32(SUB16(L1,x2), MULT16_16_P15(x2, ADD32(L2, MULT16_16_P15(x2, ADD32(L3, MULT16_16_P15(L4, x2
))))))));
}
#undef L1
#undef L2
#undef L3
#undef L4
opus_val16 celt_cos_norm(opus_val32 x)
{
x = x&0x0001ffff;
if (x>SHL32(EXTEND32(1), 16))
x = SUB32(SHL32(EXTEND32(1), 17),x);
if (x&0x00007fff)
{
if (x<SHL32(EXTEND32(1), 15))
{
return _celt_cos_pi_2(EXTRACT16(x));
} else {
return NEG32(_celt_cos_pi_2(EXTRACT16(65536-x)));
}
} else {
if (x&0x0000ffff)
return 0;
else if (x&0x0001ffff)
return -32767;
else
return 32767;
}
}
/** Reciprocal approximation (Q15 input, Q16 output) */
opus_val32 celt_rcp(opus_val32 x)
{
int i;
opus_val16 n;
opus_val16 r;
celt_assert2(x>0, "celt_rcp() only defined for positive values");
i = celt_ilog2(x);
/* n is Q15 with range [0,1). */
n = VSHR32(x,i-15)-32768;
/* Start with a linear approximation:
r = 1.8823529411764706-0.9411764705882353*n.
The coefficients and the result are Q14 in the range [15420,30840].*/
r = ADD16(30840, MULT16_16_Q15(-15420, n));
/* Perform two Newton iterations:
r -= r*((r*n)-1.Q15)
= r*((r*n)+(r-1.Q15)). */
r = SUB16(r, MULT16_16_Q15(r,
ADD16(MULT16_16_Q15(r, n), ADD16(r, -32768))));
/* We subtract an extra 1 in the second iteration to avoid overflow; it also
neatly compensates for truncation error in the rest of the process. */
r = SUB16(r, ADD16(1, MULT16_16_Q15(r,
ADD16(MULT16_16_Q15(r, n), ADD16(r, -32768)))));
/* r is now the Q15 solution to 2/(n+1), with a maximum relative error
of 7.05346E-5, a (relative) RMSE of 2.14418E-5, and a peak absolute
error of 1.24665/32768. */
return VSHR32(EXTEND32(r),i-16);
}
#endif

258
drivers/opus/celt/mathops.h Normal file
View file

@ -0,0 +1,258 @@
/* Copyright (c) 2002-2008 Jean-Marc Valin
Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2009 Xiph.Org Foundation
Written by Jean-Marc Valin */
/**
@file mathops.h
@brief Various math functions
*/
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef MATHOPS_H
#define MATHOPS_H
#include "arch.h"
#include "entcode.h"
#include "os_support.h"
/* Multiplies two 16-bit fractional values. Bit-exactness of this macro is important */
#define FRAC_MUL16(a,b) ((16384+((opus_int32)(opus_int16)(a)*(opus_int16)(b)))>>15)
unsigned isqrt32(opus_uint32 _val);
#ifndef OVERRIDE_CELT_MAXABS16
static OPUS_INLINE opus_val32 celt_maxabs16(const opus_val16 *x, int len)
{
int i;
opus_val16 maxval = 0;
opus_val16 minval = 0;
for (i=0;i<len;i++)
{
maxval = MAX16(maxval, x[i]);
minval = MIN16(minval, x[i]);
}
return MAX32(EXTEND32(maxval),-EXTEND32(minval));
}
#endif
#ifndef OVERRIDE_CELT_MAXABS32
#ifdef OPUS_FIXED_POINT
static OPUS_INLINE opus_val32 celt_maxabs32(const opus_val32 *x, int len)
{
int i;
opus_val32 maxval = 0;
opus_val32 minval = 0;
for (i=0;i<len;i++)
{
maxval = MAX32(maxval, x[i]);
minval = MIN32(minval, x[i]);
}
return MAX32(maxval, -minval);
}
#else
#define celt_maxabs32(x,len) celt_maxabs16(x,len)
#endif
#endif
#ifndef OPUS_FIXED_POINT
#define PI 3.141592653f
#define celt_sqrt(x) ((float)sqrt(x))
#define celt_rsqrt(x) (1.f/celt_sqrt(x))
#define celt_rsqrt_norm(x) (celt_rsqrt(x))
#define celt_cos_norm(x) ((float)cos((.5f*PI)*(x)))
#define celt_rcp(x) (1.f/(x))
#define celt_div(a,b) ((a)/(b))
#define frac_div32(a,b) ((float)(a)/(b))
#ifdef FLOAT_APPROX
/* Note: This assumes radix-2 floating point with the exponent at bits 23..30 and an offset of 127
denorm, +/- inf and NaN are *not* handled */
/** Base-2 log approximation (log2(x)). */
static OPUS_INLINE float celt_log2(float x)
{
int integer;
float frac;
union {
float f;
opus_uint32 i;
} in;
in.f = x;
integer = (in.i>>23)-127;
in.i -= integer<<23;
frac = in.f - 1.5f;
frac = -0.41445418f + frac*(0.95909232f
+ frac*(-0.33951290f + frac*0.16541097f));
return 1+integer+frac;
}
/** Base-2 exponential approximation (2^x). */
static OPUS_INLINE float celt_exp2(float x)
{
int integer;
float frac;
union {
float f;
opus_uint32 i;
} res;
integer = floor(x);
if (integer < -50)
return 0;
frac = x-integer;
/* K0 = 1, K1 = log(2), K2 = 3-4*log(2), K3 = 3*log(2) - 2 */
res.f = 0.99992522f + frac * (0.69583354f
+ frac * (0.22606716f + 0.078024523f*frac));
res.i = (res.i + (integer<<23)) & 0x7fffffff;
return res.f;
}
#else
#define celt_log2(x) ((float)(1.442695040888963387*log(x)))
#define celt_exp2(x) ((float)exp(0.6931471805599453094*(x)))
#endif
#endif
#ifdef OPUS_FIXED_POINT
#include "os_support.h"
#ifndef OVERRIDE_CELT_ILOG2
/** Integer log in base2. Undefined for zero and negative numbers */
static OPUS_INLINE opus_int16 celt_ilog2(opus_int32 x)
{
celt_assert2(x>0, "celt_ilog2() only defined for strictly positive numbers");
return EC_ILOG(x)-1;
}
#endif
/** Integer log in base2. Defined for zero, but not for negative numbers */
static OPUS_INLINE opus_int16 celt_zlog2(opus_val32 x)
{
return x <= 0 ? 0 : celt_ilog2(x);
}
opus_val16 celt_rsqrt_norm(opus_val32 x);
opus_val32 celt_sqrt(opus_val32 x);
opus_val16 celt_cos_norm(opus_val32 x);
/** Base-2 logarithm approximation (log2(x)). (Q14 input, Q10 output) */
static OPUS_INLINE opus_val16 celt_log2(opus_val32 x)
{
int i;
opus_val16 n, frac;
/* -0.41509302963303146, 0.9609890551383969, -0.31836011537636605,
0.15530808010959576, -0.08556153059057618 */
static const opus_val16 C[5] = {-6801+(1<<(13-DB_SHIFT)), 15746, -5217, 2545, -1401};
if (x==0)
return -32767;
i = celt_ilog2(x);
n = VSHR32(x,i-15)-32768-16384;
frac = ADD16(C[0], MULT16_16_Q15(n, ADD16(C[1], MULT16_16_Q15(n, ADD16(C[2], MULT16_16_Q15(n, ADD16(C[3], MULT16_16_Q15(n, C[4]))))))));
return SHL16(i-13,DB_SHIFT)+SHR16(frac,14-DB_SHIFT);
}
/*
K0 = 1
K1 = log(2)
K2 = 3-4*log(2)
K3 = 3*log(2) - 2
*/
#define D0 16383
#define D1 22804
#define D2 14819
#define D3 10204
static OPUS_INLINE opus_val32 celt_exp2_frac(opus_val16 x)
{
opus_val16 frac;
frac = SHL16(x, 4);
return ADD16(D0, MULT16_16_Q15(frac, ADD16(D1, MULT16_16_Q15(frac, ADD16(D2 , MULT16_16_Q15(D3,frac))))));
}
/** Base-2 exponential approximation (2^x). (Q10 input, Q16 output) */
static OPUS_INLINE opus_val32 celt_exp2(opus_val16 x)
{
int integer;
opus_val16 frac;
integer = SHR16(x,10);
if (integer>14)
return 0x7f000000;
else if (integer < -15)
return 0;
frac = celt_exp2_frac(x-SHL16(integer,10));
return VSHR32(EXTEND32(frac), -integer-2);
}
opus_val32 celt_rcp(opus_val32 x);
#define celt_div(a,b) MULT32_32_Q31((opus_val32)(a),celt_rcp(b))
opus_val32 frac_div32(opus_val32 a, opus_val32 b);
#define M1 32767
#define M2 -21
#define M3 -11943
#define M4 4936
/* Atan approximation using a 4th order polynomial. Input is in Q15 format
and normalized by pi/4. Output is in Q15 format */
static OPUS_INLINE opus_val16 celt_atan01(opus_val16 x)
{
return MULT16_16_P15(x, ADD32(M1, MULT16_16_P15(x, ADD32(M2, MULT16_16_P15(x, ADD32(M3, MULT16_16_P15(M4, x)))))));
}
#undef M1
#undef M2
#undef M3
#undef M4
/* atan2() approximation valid for positive input values */
static OPUS_INLINE opus_val16 celt_atan2p(opus_val16 y, opus_val16 x)
{
if (y < x)
{
opus_val32 arg;
arg = celt_div(SHL32(EXTEND32(y),15),x);
if (arg >= 32767)
arg = 32767;
return SHR16(celt_atan01(EXTRACT16(arg)),1);
} else {
opus_val32 arg;
arg = celt_div(SHL32(EXTEND32(x),15),y);
if (arg >= 32767)
arg = 32767;
return 25736-SHR16(celt_atan01(EXTRACT16(arg)),1);
}
}
#endif /* OPUS_FIXED_POINT */
#endif /* MATHOPS_H */

311
drivers/opus/celt/mdct.c Normal file
View file

@ -0,0 +1,311 @@
/* Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2008 Xiph.Org Foundation
Written by Jean-Marc Valin */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/* This is a simple MDCT implementation that uses a N/4 complex FFT
to do most of the work. It should be relatively straightforward to
plug in pretty much and FFT here.
This replaces the Vorbis FFT (and uses the exact same API), which
was a bit too messy and that was ending up duplicating code
(might as well use the same FFT everywhere).
The algorithm is similar to (and inspired from) Fabrice Bellard's
MDCT implementation in FFMPEG, but has differences in signs, ordering
and scaling in many places.
*/
#ifndef SKIP_CONFIG_H
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#endif
#include "mdct.h"
#include "kiss_fft.h"
#include "_kiss_fft_guts.h"
#include <math.h>
#include "os_support.h"
#include "mathops.h"
#include "stack_alloc.h"
#ifdef CUSTOM_MODES
int clt_mdct_init(celt_mdct_lookup *l,int N, int maxshift)
{
int i;
int N4;
kiss_twiddle_scalar *trig;
#if defined(OPUS_FIXED_POINT)
int N2=N>>1;
#endif
l->n = N;
N4 = N>>2;
l->maxshift = maxshift;
for (i=0;i<=maxshift;i++)
{
if (i==0)
l->kfft[i] = opus_fft_alloc(N>>2>>i, 0, 0);
else
l->kfft[i] = opus_fft_alloc_twiddles(N>>2>>i, 0, 0, l->kfft[0]);
#ifndef ENABLE_TI_DSPLIB55
if (l->kfft[i]==NULL)
return 0;
#endif
}
l->trig = trig = (kiss_twiddle_scalar*)opus_alloc((N4+1)*sizeof(kiss_twiddle_scalar));
if (l->trig==NULL)
return 0;
/* We have enough points that sine isn't necessary */
#if defined(OPUS_FIXED_POINT)
for (i=0;i<=N4;i++)
trig[i] = TRIG_UPSCALE*celt_cos_norm(DIV32(ADD32(SHL32(EXTEND32(i),17),N2),N));
#else
for (i=0;i<=N4;i++)
trig[i] = (kiss_twiddle_scalar)cos(2*PI*i/N);
#endif
return 1;
}
void clt_mdct_clear(celt_mdct_lookup *l)
{
int i;
for (i=0;i<=l->maxshift;i++)
opus_fft_free(l->kfft[i]);
opus_free((kiss_twiddle_scalar*)l->trig);
}
#endif /* CUSTOM_MODES */
/* Forward MDCT trashes the input array */
void clt_mdct_forward(const celt_mdct_lookup *l, kiss_fft_scalar *in, kiss_fft_scalar * OPUS_RESTRICT out,
const opus_val16 *window, int overlap, int shift, int stride)
{
int i;
int N, N2, N4;
kiss_twiddle_scalar sine;
VARDECL(kiss_fft_scalar, f);
VARDECL(kiss_fft_scalar, f2);
SAVE_STACK;
N = l->n;
N >>= shift;
N2 = N>>1;
N4 = N>>2;
ALLOC(f, N2, kiss_fft_scalar);
ALLOC(f2, N2, kiss_fft_scalar);
/* sin(x) ~= x here */
#ifdef OPUS_FIXED_POINT
sine = TRIG_UPSCALE*(QCONST16(0.7853981f, 15)+N2)/N;
#else
sine = (kiss_twiddle_scalar)2*PI*(.125f)/N;
#endif
/* Consider the input to be composed of four blocks: [a, b, c, d] */
/* Window, shuffle, fold */
{
/* Temp pointers to make it really clear to the compiler what we're doing */
const kiss_fft_scalar * OPUS_RESTRICT xp1 = in+(overlap>>1);
const kiss_fft_scalar * OPUS_RESTRICT xp2 = in+N2-1+(overlap>>1);
kiss_fft_scalar * OPUS_RESTRICT yp = f;
const opus_val16 * OPUS_RESTRICT wp1 = window+(overlap>>1);
const opus_val16 * OPUS_RESTRICT wp2 = window+(overlap>>1)-1;
for(i=0;i<((overlap+3)>>2);i++)
{
/* Real part arranged as -d-cR, Imag part arranged as -b+aR*/
*yp++ = MULT16_32_Q15(*wp2, xp1[N2]) + MULT16_32_Q15(*wp1,*xp2);
*yp++ = MULT16_32_Q15(*wp1, *xp1) - MULT16_32_Q15(*wp2, xp2[-N2]);
xp1+=2;
xp2-=2;
wp1+=2;
wp2-=2;
}
wp1 = window;
wp2 = window+overlap-1;
for(;i<N4-((overlap+3)>>2);i++)
{
/* Real part arranged as a-bR, Imag part arranged as -c-dR */
*yp++ = *xp2;
*yp++ = *xp1;
xp1+=2;
xp2-=2;
}
for(;i<N4;i++)
{
/* Real part arranged as a-bR, Imag part arranged as -c-dR */
*yp++ = -MULT16_32_Q15(*wp1, xp1[-N2]) + MULT16_32_Q15(*wp2, *xp2);
*yp++ = MULT16_32_Q15(*wp2, *xp1) + MULT16_32_Q15(*wp1, xp2[N2]);
xp1+=2;
xp2-=2;
wp1+=2;
wp2-=2;
}
}
/* Pre-rotation */
{
kiss_fft_scalar * OPUS_RESTRICT yp = f;
const kiss_twiddle_scalar *t = &l->trig[0];
for(i=0;i<N4;i++)
{
kiss_fft_scalar re, im, yr, yi;
re = yp[0];
im = yp[1];
yr = -S_MUL(re,t[i<<shift]) - S_MUL(im,t[(N4-i)<<shift]);
yi = -S_MUL(im,t[i<<shift]) + S_MUL(re,t[(N4-i)<<shift]);
/* works because the cos is nearly one */
*yp++ = yr + S_MUL(yi,sine);
*yp++ = yi - S_MUL(yr,sine);
}
}
/* N/4 complex FFT, down-scales by 4/N */
opus_fft(l->kfft[shift], (kiss_fft_cpx *)f, (kiss_fft_cpx *)f2);
/* Post-rotate */
{
/* Temp pointers to make it really clear to the compiler what we're doing */
const kiss_fft_scalar * OPUS_RESTRICT fp = f2;
kiss_fft_scalar * OPUS_RESTRICT yp1 = out;
kiss_fft_scalar * OPUS_RESTRICT yp2 = out+stride*(N2-1);
const kiss_twiddle_scalar *t = &l->trig[0];
/* Temp pointers to make it really clear to the compiler what we're doing */
for(i=0;i<N4;i++)
{
kiss_fft_scalar yr, yi;
yr = S_MUL(fp[1],t[(N4-i)<<shift]) + S_MUL(fp[0],t[i<<shift]);
yi = S_MUL(fp[0],t[(N4-i)<<shift]) - S_MUL(fp[1],t[i<<shift]);
/* works because the cos is nearly one */
*yp1 = yr - S_MUL(yi,sine);
*yp2 = yi + S_MUL(yr,sine);;
fp += 2;
yp1 += 2*stride;
yp2 -= 2*stride;
}
}
RESTORE_STACK;
}
void clt_mdct_backward(const celt_mdct_lookup *l, kiss_fft_scalar *in, kiss_fft_scalar * OPUS_RESTRICT out,
const opus_val16 * OPUS_RESTRICT window, int overlap, int shift, int stride)
{
int i;
int N, N2, N4;
kiss_twiddle_scalar sine;
VARDECL(kiss_fft_scalar, f2);
SAVE_STACK;
N = l->n;
N >>= shift;
N2 = N>>1;
N4 = N>>2;
ALLOC(f2, N2, kiss_fft_scalar);
/* sin(x) ~= x here */
#ifdef OPUS_FIXED_POINT
sine = TRIG_UPSCALE*(QCONST16(0.7853981f, 15)+N2)/N;
#else
sine = (kiss_twiddle_scalar)2*PI*(.125f)/N;
#endif
/* Pre-rotate */
{
/* Temp pointers to make it really clear to the compiler what we're doing */
const kiss_fft_scalar * OPUS_RESTRICT xp1 = in;
const kiss_fft_scalar * OPUS_RESTRICT xp2 = in+stride*(N2-1);
kiss_fft_scalar * OPUS_RESTRICT yp = f2;
const kiss_twiddle_scalar *t = &l->trig[0];
for(i=0;i<N4;i++)
{
kiss_fft_scalar yr, yi;
yr = -S_MUL(*xp2, t[i<<shift]) + S_MUL(*xp1,t[(N4-i)<<shift]);
yi = -S_MUL(*xp2, t[(N4-i)<<shift]) - S_MUL(*xp1,t[i<<shift]);
/* works because the cos is nearly one */
*yp++ = yr - S_MUL(yi,sine);
*yp++ = yi + S_MUL(yr,sine);
xp1+=2*stride;
xp2-=2*stride;
}
}
/* Inverse N/4 complex FFT. This one should *not* downscale even in fixed-point */
opus_ifft(l->kfft[shift], (kiss_fft_cpx *)f2, (kiss_fft_cpx *)(out+(overlap>>1)));
/* Post-rotate and de-shuffle from both ends of the buffer at once to make
it in-place. */
{
kiss_fft_scalar * OPUS_RESTRICT yp0 = out+(overlap>>1);
kiss_fft_scalar * OPUS_RESTRICT yp1 = out+(overlap>>1)+N2-2;
const kiss_twiddle_scalar *t = &l->trig[0];
/* Loop to (N4+1)>>1 to handle odd N4. When N4 is odd, the
middle pair will be computed twice. */
for(i=0;i<(N4+1)>>1;i++)
{
kiss_fft_scalar re, im, yr, yi;
kiss_twiddle_scalar t0, t1;
re = yp0[0];
im = yp0[1];
t0 = t[i<<shift];
t1 = t[(N4-i)<<shift];
/* We'd scale up by 2 here, but instead it's done when mixing the windows */
yr = S_MUL(re,t0) - S_MUL(im,t1);
yi = S_MUL(im,t0) + S_MUL(re,t1);
re = yp1[0];
im = yp1[1];
/* works because the cos is nearly one */
yp0[0] = -(yr - S_MUL(yi,sine));
yp1[1] = yi + S_MUL(yr,sine);
t0 = t[(N4-i-1)<<shift];
t1 = t[(i+1)<<shift];
/* We'd scale up by 2 here, but instead it's done when mixing the windows */
yr = S_MUL(re,t0) - S_MUL(im,t1);
yi = S_MUL(im,t0) + S_MUL(re,t1);
/* works because the cos is nearly one */
yp1[0] = -(yr - S_MUL(yi,sine));
yp0[1] = yi + S_MUL(yr,sine);
yp0 += 2;
yp1 -= 2;
}
}
/* Mirror on both sides for TDAC */
{
kiss_fft_scalar * OPUS_RESTRICT xp1 = out+overlap-1;
kiss_fft_scalar * OPUS_RESTRICT yp1 = out;
const opus_val16 * OPUS_RESTRICT wp1 = window;
const opus_val16 * OPUS_RESTRICT wp2 = window+overlap-1;
for(i = 0; i < overlap/2; i++)
{
kiss_fft_scalar x1, x2;
x1 = *xp1;
x2 = *yp1;
*yp1++ = MULT16_32_Q15(*wp2, x2) - MULT16_32_Q15(*wp1, x1);
*xp1-- = MULT16_32_Q15(*wp1, x2) + MULT16_32_Q15(*wp2, x1);
wp1++;
wp2--;
}
}
RESTORE_STACK;
}

70
drivers/opus/celt/mdct.h Normal file
View file

@ -0,0 +1,70 @@
/* Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2008 Xiph.Org Foundation
Written by Jean-Marc Valin */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/* This is a simple MDCT implementation that uses a N/4 complex FFT
to do most of the work. It should be relatively straightforward to
plug in pretty much and FFT here.
This replaces the Vorbis FFT (and uses the exact same API), which
was a bit too messy and that was ending up duplicating code
(might as well use the same FFT everywhere).
The algorithm is similar to (and inspired from) Fabrice Bellard's
MDCT implementation in FFMPEG, but has differences in signs, ordering
and scaling in many places.
*/
#ifndef MDCT_H
#define MDCT_H
#include "opus_defines.h"
#include "kiss_fft.h"
#include "arch.h"
typedef struct {
int n;
int maxshift;
const kiss_fft_state *kfft[4];
const kiss_twiddle_scalar * OPUS_RESTRICT trig;
} celt_mdct_lookup;
int clt_mdct_init(celt_mdct_lookup *l,int N, int maxshift);
void clt_mdct_clear(celt_mdct_lookup *l);
/** Compute a forward MDCT and scale by 4/N, trashes the input array */
void clt_mdct_forward(const celt_mdct_lookup *l, kiss_fft_scalar *in,
kiss_fft_scalar * OPUS_RESTRICT out,
const opus_val16 *window, int overlap, int shift, int stride);
/** Compute a backward MDCT (no scaling) and performs weighted overlap-add
(scales implicitly by 1/2) */
void clt_mdct_backward(const celt_mdct_lookup *l, kiss_fft_scalar *in,
kiss_fft_scalar * OPUS_RESTRICT out,
const opus_val16 * OPUS_RESTRICT window, int overlap, int shift, int stride);
#endif

48
drivers/opus/celt/mfrngcod.h Normal file
View file

@ -0,0 +1,48 @@
/* Copyright (c) 2001-2008 Timothy B. Terriberry
Copyright (c) 2008-2009 Xiph.Org Foundation */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#if !defined(_mfrngcode_H)
# define _mfrngcode_H (1)
# include "entcode.h"
/*Constants used by the entropy encoder/decoder.*/
/*The number of bits to output at a time.*/
# define EC_SYM_BITS (8)
/*The total number of bits in each of the state registers.*/
# define EC_CODE_BITS (32)
/*The maximum symbol value.*/
# define EC_SYM_MAX ((1U<<EC_SYM_BITS)-1)
/*Bits to shift by to move a symbol into the high-order position.*/
# define EC_CODE_SHIFT (EC_CODE_BITS-EC_SYM_BITS-1)
/*Carry bit of the high-order range symbol.*/
# define EC_CODE_TOP (((opus_uint32)1U)<<(EC_CODE_BITS-1))
/*Low-order bit of the high-order range symbol.*/
# define EC_CODE_BOT (EC_CODE_TOP>>EC_SYM_BITS)
/*The number of bits available for the last, partial symbol in the code field.*/
# define EC_CODE_EXTRA ((EC_CODE_BITS-2)%EC_SYM_BITS+1)
#endif

438
drivers/opus/celt/modes.c Normal file
View file

@ -0,0 +1,438 @@
/* Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2009 Xiph.Org Foundation
Copyright (c) 2008 Gregory Maxwell
Written by Jean-Marc Valin and Gregory Maxwell */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#include "celt.h"
#include "opus_modes.h"
#include "rate.h"
#include "os_support.h"
#include "stack_alloc.h"
#include "quant_bands.h"
static const opus_int16 eband5ms[] = {
/*0 200 400 600 800 1k 1.2 1.4 1.6 2k 2.4 2.8 3.2 4k 4.8 5.6 6.8 8k 9.6 12k 15.6 */
0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 24, 28, 34, 40, 48, 60, 78, 100
};
/* Alternate tuning (partially derived from Vorbis) */
#define BITALLOC_SIZE 11
/* Bit allocation table in units of 1/32 bit/sample (0.1875 dB SNR) */
static const unsigned char band_allocation[] = {
/*0 200 400 600 800 1k 1.2 1.4 1.6 2k 2.4 2.8 3.2 4k 4.8 5.6 6.8 8k 9.6 12k 15.6 */
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
90, 80, 75, 69, 63, 56, 49, 40, 34, 29, 20, 18, 10, 0, 0, 0, 0, 0, 0, 0, 0,
110,100, 90, 84, 78, 71, 65, 58, 51, 45, 39, 32, 26, 20, 12, 0, 0, 0, 0, 0, 0,
118,110,103, 93, 86, 80, 75, 70, 65, 59, 53, 47, 40, 31, 23, 15, 4, 0, 0, 0, 0,
126,119,112,104, 95, 89, 83, 78, 72, 66, 60, 54, 47, 39, 32, 25, 17, 12, 1, 0, 0,
134,127,120,114,103, 97, 91, 85, 78, 72, 66, 60, 54, 47, 41, 35, 29, 23, 16, 10, 1,
144,137,130,124,113,107,101, 95, 88, 82, 76, 70, 64, 57, 51, 45, 39, 33, 26, 15, 1,
152,145,138,132,123,117,111,105, 98, 92, 86, 80, 74, 67, 61, 55, 49, 43, 36, 20, 1,
162,155,148,142,133,127,121,115,108,102, 96, 90, 84, 77, 71, 65, 59, 53, 46, 30, 1,
172,165,158,152,143,137,131,125,118,112,106,100, 94, 87, 81, 75, 69, 63, 56, 45, 20,
200,200,200,200,200,200,200,200,198,193,188,183,178,173,168,163,158,153,148,129,104,
};
#ifndef CUSTOM_MODES_ONLY
#ifdef OPUS_FIXED_POINT
#include "static_modes_fixed.h"
#else
#include "static_modes_float.h"
#endif
#endif /* CUSTOM_MODES_ONLY */
#ifndef M_PI
#define M_PI 3.141592653
#endif
#ifdef CUSTOM_MODES
/* Defining 25 critical bands for the full 0-20 kHz audio bandwidth
Taken from http://ccrma.stanford.edu/~jos/bbt/Bark_Frequency_Scale.html */
#define BARK_BANDS 25
static const opus_int16 bark_freq[BARK_BANDS+1] = {
0, 100, 200, 300, 400,
510, 630, 770, 920, 1080,
1270, 1480, 1720, 2000, 2320,
2700, 3150, 3700, 4400, 5300,
6400, 7700, 9500, 12000, 15500,
20000};
static opus_int16 *compute_ebands(opus_int32 Fs, int frame_size, int res, int *nbEBands)
{
opus_int16 *eBands;
int i, j, lin, low, high, nBark, offset=0;
/* All modes that have 2.5 ms short blocks use the same definition */
if (Fs == 400*(opus_int32)frame_size)
{
*nbEBands = sizeof(eband5ms)/sizeof(eband5ms[0])-1;
eBands = opus_alloc(sizeof(opus_int16)*(*nbEBands+1));
for (i=0;i<*nbEBands+1;i++)
eBands[i] = eband5ms[i];
return eBands;
}
/* Find the number of critical bands supported by our sampling rate */
for (nBark=1;nBark<BARK_BANDS;nBark++)
if (bark_freq[nBark+1]*2 >= Fs)
break;
/* Find where the linear part ends (i.e. where the spacing is more than min_width */
for (lin=0;lin<nBark;lin++)
if (bark_freq[lin+1]-bark_freq[lin] >= res)
break;
low = (bark_freq[lin]+res/2)/res;
high = nBark-lin;
*nbEBands = low+high;
eBands = opus_alloc(sizeof(opus_int16)*(*nbEBands+2));
if (eBands==NULL)
return NULL;
/* Linear spacing (min_width) */
for (i=0;i<low;i++)
eBands[i] = i;
if (low>0)
offset = eBands[low-1]*res - bark_freq[lin-1];
/* Spacing follows critical bands */
for (i=0;i<high;i++)
{
int target = bark_freq[lin+i];
/* Round to an even value */
eBands[i+low] = (target+offset/2+res)/(2*res)*2;
offset = eBands[i+low]*res - target;
}
/* Enforce the minimum spacing at the boundary */
for (i=0;i<*nbEBands;i++)
if (eBands[i] < i)
eBands[i] = i;
/* Round to an even value */
eBands[*nbEBands] = (bark_freq[nBark]+res)/(2*res)*2;
if (eBands[*nbEBands] > frame_size)
eBands[*nbEBands] = frame_size;
for (i=1;i<*nbEBands-1;i++)
{
if (eBands[i+1]-eBands[i] < eBands[i]-eBands[i-1])
{
eBands[i] -= (2*eBands[i]-eBands[i-1]-eBands[i+1])/2;
}
}
/* Remove any empty bands. */
for (i=j=0;i<*nbEBands;i++)
if(eBands[i+1]>eBands[j])
eBands[++j]=eBands[i+1];
*nbEBands=j;
for (i=1;i<*nbEBands;i++)
{
/* Every band must be smaller than the last band. */
celt_assert(eBands[i]-eBands[i-1]<=eBands[*nbEBands]-eBands[*nbEBands-1]);
/* Each band must be no larger than twice the size of the previous one. */
celt_assert(eBands[i+1]-eBands[i]<=2*(eBands[i]-eBands[i-1]));
}
return eBands;
}
static void compute_allocation_table(CELTMode *mode)
{
int i, j;
unsigned char *allocVectors;
int maxBands = sizeof(eband5ms)/sizeof(eband5ms[0])-1;
mode->nbAllocVectors = BITALLOC_SIZE;
allocVectors = opus_alloc(sizeof(unsigned char)*(BITALLOC_SIZE*mode->nbEBands));
if (allocVectors==NULL)
return;
/* Check for standard mode */
if (mode->Fs == 400*(opus_int32)mode->shortMdctSize)
{
for (i=0;i<BITALLOC_SIZE*mode->nbEBands;i++)
allocVectors[i] = band_allocation[i];
mode->allocVectors = allocVectors;
return;
}
/* If not the standard mode, interpolate */
/* Compute per-codec-band allocation from per-critical-band matrix */
for (i=0;i<BITALLOC_SIZE;i++)
{
for (j=0;j<mode->nbEBands;j++)
{
int k;
for (k=0;k<maxBands;k++)
{
if (400*(opus_int32)eband5ms[k] > mode->eBands[j]*(opus_int32)mode->Fs/mode->shortMdctSize)
break;
}
if (k>maxBands-1)
allocVectors[i*mode->nbEBands+j] = band_allocation[i*maxBands + maxBands-1];
else {
opus_int32 a0, a1;
a1 = mode->eBands[j]*(opus_int32)mode->Fs/mode->shortMdctSize - 400*(opus_int32)eband5ms[k-1];
a0 = 400*(opus_int32)eband5ms[k] - mode->eBands[j]*(opus_int32)mode->Fs/mode->shortMdctSize;
allocVectors[i*mode->nbEBands+j] = (a0*band_allocation[i*maxBands+k-1]
+ a1*band_allocation[i*maxBands+k])/(a0+a1);
}
}
}
/*printf ("\n");
for (i=0;i<BITALLOC_SIZE;i++)
{
for (j=0;j<mode->nbEBands;j++)
printf ("%d ", allocVectors[i*mode->nbEBands+j]);
printf ("\n");
}
exit(0);*/
mode->allocVectors = allocVectors;
}
#endif /* CUSTOM_MODES */
CELTMode *opus_custom_mode_create(opus_int32 Fs, int frame_size, int *error)
{
int i;
#ifdef CUSTOM_MODES
CELTMode *mode=NULL;
int res;
opus_val16 *window;
opus_int16 *logN;
int LM;
ALLOC_STACK;
#if !defined(VAR_ARRAYS) && !defined(USE_ALLOCA)
if (global_stack==NULL)
goto failure;
#endif
#endif
#ifndef CUSTOM_MODES_ONLY
for (i=0;i<TOTAL_MODES;i++)
{
int j;
for (j=0;j<4;j++)
{
if (Fs == static_mode_list[i]->Fs &&
(frame_size<<j) == static_mode_list[i]->shortMdctSize*static_mode_list[i]->nbShortMdcts)
{
if (error)
*error = OPUS_OK;
return (CELTMode*)static_mode_list[i];
}
}
}
#endif /* CUSTOM_MODES_ONLY */
#ifndef CUSTOM_MODES
if (error)
*error = OPUS_BAD_ARG;
return NULL;
#else
/* The good thing here is that permutation of the arguments will automatically be invalid */
if (Fs < 8000 || Fs > 96000)
{
if (error)
*error = OPUS_BAD_ARG;
return NULL;
}
if (frame_size < 40 || frame_size > 1024 || frame_size%2!=0)
{
if (error)
*error = OPUS_BAD_ARG;
return NULL;
}
/* Frames of less than 1ms are not supported. */
if ((opus_int32)frame_size*1000 < Fs)
{
if (error)
*error = OPUS_BAD_ARG;
return NULL;
}
if ((opus_int32)frame_size*75 >= Fs && (frame_size%16)==0)
{
LM = 3;
} else if ((opus_int32)frame_size*150 >= Fs && (frame_size%8)==0)
{
LM = 2;
} else if ((opus_int32)frame_size*300 >= Fs && (frame_size%4)==0)
{
LM = 1;
} else
{
LM = 0;
}
/* Shorts longer than 3.3ms are not supported. */
if ((opus_int32)(frame_size>>LM)*300 > Fs)
{
if (error)
*error = OPUS_BAD_ARG;
return NULL;
}
mode = opus_alloc(sizeof(CELTMode));
if (mode==NULL)
goto failure;
mode->Fs = Fs;
/* Pre/de-emphasis depends on sampling rate. The "standard" pre-emphasis
is defined as A(z) = 1 - 0.85*z^-1 at 48 kHz. Other rates should
approximate that. */
if(Fs < 12000) /* 8 kHz */
{
mode->preemph[0] = QCONST16(0.3500061035f, 15);
mode->preemph[1] = -QCONST16(0.1799926758f, 15);
mode->preemph[2] = QCONST16(0.2719968125f, SIG_SHIFT); /* exact 1/preemph[3] */
mode->preemph[3] = QCONST16(3.6765136719f, 13);
} else if(Fs < 24000) /* 16 kHz */
{
mode->preemph[0] = QCONST16(0.6000061035f, 15);
mode->preemph[1] = -QCONST16(0.1799926758f, 15);
mode->preemph[2] = QCONST16(0.4424998650f, SIG_SHIFT); /* exact 1/preemph[3] */
mode->preemph[3] = QCONST16(2.2598876953f, 13);
} else if(Fs < 40000) /* 32 kHz */
{
mode->preemph[0] = QCONST16(0.7799987793f, 15);
mode->preemph[1] = -QCONST16(0.1000061035f, 15);
mode->preemph[2] = QCONST16(0.7499771125f, SIG_SHIFT); /* exact 1/preemph[3] */
mode->preemph[3] = QCONST16(1.3333740234f, 13);
} else /* 48 kHz */
{
mode->preemph[0] = QCONST16(0.8500061035f, 15);
mode->preemph[1] = QCONST16(0.0f, 15);
mode->preemph[2] = QCONST16(1.f, SIG_SHIFT);
mode->preemph[3] = QCONST16(1.f, 13);
}
mode->maxLM = LM;
mode->nbShortMdcts = 1<<LM;
mode->shortMdctSize = frame_size/mode->nbShortMdcts;
res = (mode->Fs+mode->shortMdctSize)/(2*mode->shortMdctSize);
mode->eBands = compute_ebands(Fs, mode->shortMdctSize, res, &mode->nbEBands);
if (mode->eBands==NULL)
goto failure;
#if !defined(SMALL_FOOTPRINT)
/* Make sure we don't allocate a band larger than our PVQ table.
208 should be enough, but let's be paranoid. */
if ((mode->eBands[mode->nbEBands] - mode->eBands[mode->nbEBands-1])<<LM >
208) {
goto failure;
}
#endif
mode->effEBands = mode->nbEBands;
while (mode->eBands[mode->effEBands] > mode->shortMdctSize)
mode->effEBands--;
/* Overlap must be divisible by 4 */
mode->overlap = ((mode->shortMdctSize>>2)<<2);
compute_allocation_table(mode);
if (mode->allocVectors==NULL)
goto failure;
window = (opus_val16*)opus_alloc(mode->overlap*sizeof(opus_val16));
if (window==NULL)
goto failure;
#ifndef OPUS_FIXED_POINT
for (i=0;i<mode->overlap;i++)
window[i] = Q15ONE*sin(.5*M_PI* sin(.5*M_PI*(i+.5)/mode->overlap) * sin(.5*M_PI*(i+.5)/mode->overlap));
#else
for (i=0;i<mode->overlap;i++)
window[i] = MIN32(32767,floor(.5+32768.*sin(.5*M_PI* sin(.5*M_PI*(i+.5)/mode->overlap) * sin(.5*M_PI*(i+.5)/mode->overlap))));
#endif
mode->window = window;
logN = (opus_int16*)opus_alloc(mode->nbEBands*sizeof(opus_int16));
if (logN==NULL)
goto failure;
for (i=0;i<mode->nbEBands;i++)
logN[i] = log2_frac(mode->eBands[i+1]-mode->eBands[i], BITRES);
mode->logN = logN;
compute_pulse_cache(mode, mode->maxLM);
if (clt_mdct_init(&mode->mdct, 2*mode->shortMdctSize*mode->nbShortMdcts,
mode->maxLM) == 0)
goto failure;
if (error)
*error = OPUS_OK;
return mode;
failure:
if (error)
*error = OPUS_ALLOC_FAIL;
if (mode!=NULL)
opus_custom_mode_destroy(mode);
return NULL;
#endif /* !CUSTOM_MODES */
}
#ifdef CUSTOM_MODES
void opus_custom_mode_destroy(CELTMode *mode)
{
if (mode == NULL)
return;
#ifndef CUSTOM_MODES_ONLY
{
int i;
for (i=0;i<TOTAL_MODES;i++)
{
if (mode == static_mode_list[i])
{
return;
}
}
}
#endif /* CUSTOM_MODES_ONLY */
opus_free((opus_int16*)mode->eBands);
opus_free((opus_int16*)mode->allocVectors);
opus_free((opus_val16*)mode->window);
opus_free((opus_int16*)mode->logN);
opus_free((opus_int16*)mode->cache.index);
opus_free((unsigned char*)mode->cache.bits);
opus_free((unsigned char*)mode->cache.caps);
clt_mdct_clear(&mode->mdct);
opus_free((CELTMode *)mode);
}
#endif

210
drivers/opus/celt/opus_custom_demo.c Normal file
View file

@ -0,0 +1,210 @@
/* Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2009 Xiph.Org Foundation
Written by Jean-Marc Valin */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifdef OPUS_HAVE_CONFIG_H
#include "opus_config.h"
#endif
#include "opus_custom.h"
#include "arch.h"
#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#include <string.h>
#define MAX_PACKET 1275
int main(int argc, char *argv[])
{
int err;
char *inFile, *outFile;
FILE *fin, *fout;
OpusCustomMode *mode=NULL;
OpusCustomEncoder *enc;
OpusCustomDecoder *dec;
int len;
opus_int32 frame_size, channels, rate;
int bytes_per_packet;
unsigned char data[MAX_PACKET];
int complexity;
#if !(defined (OPUS_FIXED_POINT) && !defined(CUSTOM_MODES)) && defined(RESYNTH)
int i;
double rmsd = 0;
#endif
int count = 0;
opus_int32 skip;
opus_int16 *in, *out;
if (argc != 9 && argc != 8 && argc != 7)
{
fprintf (stderr, "Usage: test_opus_custom <rate> <channels> <frame size> "
" <bytes per packet> [<complexity> [packet loss rate]] "
"<input> <output>\n");
return 1;
}
rate = (opus_int32)atol(argv[1]);
channels = atoi(argv[2]);
frame_size = atoi(argv[3]);
mode = opus_custom_mode_create(rate, frame_size, NULL);
if (mode == NULL)
{
fprintf(stderr, "failed to create a mode\n");
return 1;
}
bytes_per_packet = atoi(argv[4]);
if (bytes_per_packet < 0 || bytes_per_packet > MAX_PACKET)
{
fprintf (stderr, "bytes per packet must be between 0 and %d\n",
MAX_PACKET);
return 1;
}
inFile = argv[argc-2];
fin = fopen(inFile, "rb");
if (!fin)
{
fprintf (stderr, "Could not open input file %s\n", argv[argc-2]);
return 1;
}
outFile = argv[argc-1];
fout = fopen(outFile, "wb+");
if (!fout)
{
fprintf (stderr, "Could not open output file %s\n", argv[argc-1]);
fclose(fin);
return 1;
}
enc = opus_custom_encoder_create(mode, channels, &err);
if (err != 0)
{
fprintf(stderr, "Failed to create the encoder: %s\n", opus_strerror(err));
fclose(fin);
fclose(fout);
return 1;
}
dec = opus_custom_decoder_create(mode, channels, &err);
if (err != 0)
{
fprintf(stderr, "Failed to create the decoder: %s\n", opus_strerror(err));
fclose(fin);
fclose(fout);
return 1;
}
opus_custom_decoder_ctl(dec, OPUS_GET_LOOKAHEAD(&skip));
if (argc>7)
{
complexity=atoi(argv[5]);
opus_custom_encoder_ctl(enc,OPUS_SET_COMPLEXITY(complexity));
}
in = (opus_int16*)malloc(frame_size*channels*sizeof(opus_int16));
out = (opus_int16*)malloc(frame_size*channels*sizeof(opus_int16));
while (!feof(fin))
{
int ret;
err = fread(in, sizeof(short), frame_size*channels, fin);
if (feof(fin))
break;
len = opus_custom_encode(enc, in, frame_size, data, bytes_per_packet);
if (len <= 0)
fprintf (stderr, "opus_custom_encode() failed: %s\n", opus_strerror(len));
/* This is for simulating bit errors */
#if 0
int errors = 0;
int eid = 0;
/* This simulates random bit error */
for (i=0;i<len*8;i++)
{
if (rand()%atoi(argv[8])==0)
{
if (i<64)
{
errors++;
eid = i;
}
data[i/8] ^= 1<<(7-(i%8));
}
}
if (errors == 1)
data[eid/8] ^= 1<<(7-(eid%8));
else if (errors%2 == 1)
data[rand()%8] ^= 1<<rand()%8;
#endif
#if 1 /* Set to zero to use the encoder's output instead */
/* This is to simulate packet loss */
if (argc==9 && rand()%1000<atoi(argv[argc-3]))
/*if (errors && (errors%2==0))*/
ret = opus_custom_decode(dec, NULL, len, out, frame_size);
else
ret = opus_custom_decode(dec, data, len, out, frame_size);
if (ret < 0)
fprintf(stderr, "opus_custom_decode() failed: %s\n", opus_strerror(ret));
#else
for (i=0;i<ret*channels;i++)
out[i] = in[i];
#endif
#if !(defined (OPUS_FIXED_POINT) && !defined(CUSTOM_MODES)) && defined(RESYNTH)
for (i=0;i<ret*channels;i++)
{
rmsd += (in[i]-out[i])*1.0*(in[i]-out[i]);
/*out[i] -= in[i];*/
}
#endif
count++;
fwrite(out+skip*channels, sizeof(short), (ret-skip)*channels, fout);
skip = 0;
}
PRINT_MIPS(stderr);
opus_custom_encoder_destroy(enc);
opus_custom_decoder_destroy(dec);
fclose(fin);
fclose(fout);
opus_custom_mode_destroy(mode);
free(in);
free(out);
#if !(defined (OPUS_FIXED_POINT) && !defined(CUSTOM_MODES)) && defined(RESYNTH)
if (rmsd > 0)
{
rmsd = sqrt(rmsd/(1.0*frame_size*channels*count));
fprintf (stderr, "Error: encoder doesn't match decoder\n");
fprintf (stderr, "RMS mismatch is %f\n", rmsd);
return 1;
} else {
fprintf (stderr, "Encoder matches decoder!!\n");
}
#endif
return 0;
}

83
drivers/opus/celt/opus_modes.h Normal file
View file

@ -0,0 +1,83 @@
/* Copyright (c) 2007-2008 CSIRO
Copyright (c) 2007-2009 Xiph.Org Foundation
Copyright (c) 2008 Gregory Maxwell
Written by Jean-Marc Valin and Gregory Maxwell */
/*
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
- Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
- Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef OPUS_MODES_H
#define OPUS_MODES_H
#include "opus_types.h"
#include "celt.h"
#include "arch.h"
#include "mdct.h"
#include "entenc.h"
#include "entdec.h"
#define MAX_PERIOD 1024
#ifndef OVERLAP
#define OVERLAP(mode) ((mode)->overlap)
#endif
#ifndef FRAMESIZE
#define FRAMESIZE(mode) ((mode)->mdctSize)
#endif
typedef struct {
int size;
const opus_int16 *index;
const unsigned char *bits;
const unsigned char *caps;
} PulseCache;
/** Mode definition (opaque)
@brief Mode definition
*/
struct OpusCustomMode {
opus_int32 Fs;
int overlap;
int nbEBands;
int effEBands;
opus_val16 preemph[4];
const opus_int16 *eBands; /**< Definition for each "pseudo-critical band" */
int maxLM;
int nbShortMdcts;
int shortMdctSize;
int nbAllocVectors; /**< Number of lines in the matrix below */
const unsigned char *allocVectors; /**< Number of bits in each band for several rates */
const opus_int16 *logN;
const opus_val16 *window;
celt_mdct_lookup mdct;
PulseCache cache;
};
#endif

Some files were not shown because too many files have changed in this diff Show more