baa070cca6
This enables the ability to allocate (and free) memory dynamically without fragmentation, through compaction. This means allocations can move and fragmentation be reduced. Most changes are preparing Rockbox for this, which many times means adding a move callback which can temporarily disable movement when the corresponding code is in a critical section. For now, the audio buffer allocation has a central role, because it's the one having allocated most. This buffer is able to shrink itself, for which it needs to stop playback for a very short moment. For this, audio_buffer_available() returns the size of the audio buffer which can possibly be used by other allocations because the audio buffer can shrink. lastfm scrobbling and timestretch can now be toggled at runtime without requiring a reboot. git-svn-id: svn://svn.rockbox.org/rockbox/trunk@30381 a1c6a512-1295-4272-9138-f99709370657
450 lines
12 KiB
C
450 lines
12 KiB
C
/***************************************************************************
|
|
* __________ __ ___.
|
|
* Open \______ \ ____ ____ | | _\_ |__ _______ ___
|
|
* Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
|
|
* Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
|
|
* Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
|
|
* \/ \/ \/ \/ \/
|
|
* $Id$
|
|
*
|
|
* Copyright (C) 2006 by Nicolas Pitre <nico@cam.org>
|
|
* Copyright (C) 2006-2007 by Stéphane Doyon <s.doyon@videotron.ca>
|
|
*
|
|
* This program is free software; you can redistribute it and/or
|
|
* modify it under the terms of the GNU General Public License
|
|
* as published by the Free Software Foundation; either version 2
|
|
* of the License, or (at your option) any later version.
|
|
*
|
|
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
|
|
* KIND, either express or implied.
|
|
*
|
|
****************************************************************************/
|
|
|
|
#include <inttypes.h>
|
|
#include <stddef.h>
|
|
#include <stdio.h>
|
|
#include <string.h>
|
|
#include "sound.h"
|
|
#include "core_alloc.h"
|
|
#include "system.h"
|
|
#include "tdspeed.h"
|
|
#include "settings.h"
|
|
|
|
#define assert(cond)
|
|
|
|
#define MIN_RATE 8000
|
|
#define MAX_RATE 48000 /* double buffer for double rate */
|
|
#define MINFREQ 100
|
|
|
|
#define FIXED_BUFSIZE 3072 /* 48KHz factor 3.0 */
|
|
|
|
static int32_t** dsp_src;
|
|
static int handles[4];
|
|
static int32_t *overlap_buffer[2] = { NULL, NULL };
|
|
static int32_t *outbuf[2] = { NULL, NULL };
|
|
|
|
static int move_callback(int handle, void* current, void* new)
|
|
{
|
|
/* TODO */
|
|
(void)handle;
|
|
if (dsp_src)
|
|
{
|
|
int ch = (current == outbuf[0]) ? 0 : 1;
|
|
dsp_src[ch] = outbuf[ch] = new;
|
|
}
|
|
return BUFLIB_CB_OK;
|
|
}
|
|
|
|
static struct buflib_callbacks ops = {
|
|
.move_callback = move_callback,
|
|
.shrink_callback = NULL,
|
|
};
|
|
|
|
static int ovl_move_callback(int handle, void* current, void* new)
|
|
{
|
|
/* TODO */
|
|
(void)handle;
|
|
if (dsp_src)
|
|
{
|
|
int ch = (current == overlap_buffer[0]) ? 0 : 1;
|
|
overlap_buffer[ch] = new;
|
|
}
|
|
return BUFLIB_CB_OK;
|
|
}
|
|
|
|
static struct buflib_callbacks ovl_ops = {
|
|
.move_callback = ovl_move_callback,
|
|
.shrink_callback = NULL,
|
|
};
|
|
|
|
|
|
static struct tdspeed_state_s
|
|
{
|
|
bool stereo;
|
|
int32_t shift_max; /* maximum displacement on a frame */
|
|
int32_t src_step; /* source window pace */
|
|
int32_t dst_step; /* destination window pace */
|
|
int32_t dst_order; /* power of two for dst_step */
|
|
int32_t ovl_shift; /* overlap buffer frame shift */
|
|
int32_t ovl_size; /* overlap buffer used size */
|
|
int32_t ovl_space; /* overlap buffer size */
|
|
int32_t *ovl_buff[2]; /* overlap buffer */
|
|
} tdspeed_state;
|
|
|
|
void tdspeed_init(void)
|
|
{
|
|
if (!global_settings.timestretch_enabled)
|
|
return;
|
|
|
|
/* Allocate buffers */
|
|
if (overlap_buffer[0] == NULL)
|
|
{
|
|
handles[0] = core_alloc_ex("tdspeed ovl left", FIXED_BUFSIZE * sizeof(int32_t), &ovl_ops);
|
|
overlap_buffer[0] = core_get_data(handles[0]);
|
|
}
|
|
if (overlap_buffer[1] == NULL)
|
|
{
|
|
handles[1] = core_alloc_ex("tdspeed ovl right", FIXED_BUFSIZE * sizeof(int32_t), &ovl_ops);
|
|
overlap_buffer[1] = core_get_data(handles[1]);
|
|
}
|
|
if (outbuf[0] == NULL)
|
|
{
|
|
handles[2] = core_alloc_ex("tdspeed left", TDSPEED_OUTBUFSIZE * sizeof(int32_t), &ops);
|
|
outbuf[0] = core_get_data(handles[2]);
|
|
}
|
|
if (outbuf[1] == NULL)
|
|
{
|
|
handles[3] = core_alloc_ex("tdspeed right", TDSPEED_OUTBUFSIZE * sizeof(int32_t), &ops);
|
|
outbuf[1] = core_get_data(handles[3]);
|
|
}
|
|
}
|
|
|
|
void tdspeed_finish(void)
|
|
{
|
|
for(unsigned i = 0; i < ARRAYLEN(handles); i++)
|
|
{
|
|
if (handles[i] > 0)
|
|
{
|
|
core_free(handles[i]);
|
|
handles[i] = 0;
|
|
}
|
|
}
|
|
overlap_buffer[0] = overlap_buffer[1] = NULL;
|
|
outbuf[0] = outbuf[1] = NULL;
|
|
}
|
|
|
|
bool tdspeed_config(int samplerate, bool stereo, int32_t factor)
|
|
{
|
|
struct tdspeed_state_s *st = &tdspeed_state;
|
|
int src_frame_sz;
|
|
|
|
/* Check buffers were allocated ok */
|
|
if (overlap_buffer[0] == NULL || overlap_buffer[1] == NULL)
|
|
return false;
|
|
|
|
if (outbuf[0] == NULL || outbuf[1] == NULL)
|
|
return false;
|
|
|
|
/* Check parameters */
|
|
if (factor == PITCH_SPEED_100)
|
|
return false;
|
|
|
|
if (samplerate < MIN_RATE || samplerate > MAX_RATE)
|
|
return false;
|
|
|
|
if (factor < STRETCH_MIN || factor > STRETCH_MAX)
|
|
return false;
|
|
|
|
st->stereo = stereo;
|
|
st->dst_step = samplerate / MINFREQ;
|
|
|
|
if (factor > PITCH_SPEED_100)
|
|
st->dst_step = st->dst_step * PITCH_SPEED_100 / factor;
|
|
|
|
st->dst_order = 1;
|
|
|
|
while (st->dst_step >>= 1)
|
|
st->dst_order++;
|
|
|
|
st->dst_step = (1 << st->dst_order);
|
|
st->src_step = st->dst_step * factor / PITCH_SPEED_100;
|
|
st->shift_max = (st->dst_step > st->src_step) ? st->dst_step : st->src_step;
|
|
|
|
src_frame_sz = st->shift_max + st->dst_step;
|
|
|
|
if (st->dst_step > st->src_step)
|
|
src_frame_sz += st->dst_step - st->src_step;
|
|
|
|
st->ovl_space = ((src_frame_sz - 2) / st->src_step) * st->src_step
|
|
+ src_frame_sz;
|
|
|
|
if (st->src_step > st->dst_step)
|
|
st->ovl_space += 2*st->src_step - st->dst_step;
|
|
|
|
if (st->ovl_space > FIXED_BUFSIZE)
|
|
st->ovl_space = FIXED_BUFSIZE;
|
|
|
|
st->ovl_size = 0;
|
|
st->ovl_shift = 0;
|
|
|
|
st->ovl_buff[0] = overlap_buffer[0];
|
|
|
|
if (stereo)
|
|
st->ovl_buff[1] = overlap_buffer[1];
|
|
else
|
|
st->ovl_buff[1] = st->ovl_buff[0];
|
|
|
|
return true;
|
|
}
|
|
|
|
static int tdspeed_apply(int32_t *buf_out[2], int32_t *buf_in[2],
|
|
int data_len, int last, int out_size)
|
|
/* data_len in samples */
|
|
{
|
|
struct tdspeed_state_s *st = &tdspeed_state;
|
|
int32_t *curr, *prev, *dest[2], *d;
|
|
int32_t i, j, next_frame, prev_frame, shift, src_frame_sz;
|
|
bool stereo = buf_in[0] != buf_in[1];
|
|
|
|
assert(stereo == st->stereo);
|
|
|
|
src_frame_sz = st->shift_max + st->dst_step;
|
|
|
|
if (st->dst_step > st->src_step)
|
|
src_frame_sz += st->dst_step - st->src_step;
|
|
|
|
/* deal with overlap data first, if any */
|
|
if (st->ovl_size)
|
|
{
|
|
int32_t have, copy, steps;
|
|
have = st->ovl_size;
|
|
|
|
if (st->ovl_shift > 0)
|
|
have -= st->ovl_shift;
|
|
|
|
/* append just enough data to have all of the overlap buffer consumed */
|
|
steps = (have - 1) / st->src_step;
|
|
copy = steps * st->src_step + src_frame_sz - have;
|
|
|
|
if (copy < src_frame_sz - st->dst_step)
|
|
copy += st->src_step; /* one more step to allow for pregap data */
|
|
|
|
if (copy > data_len)
|
|
copy = data_len;
|
|
|
|
assert(st->ovl_size + copy <= FIXED_BUFSIZE);
|
|
memcpy(st->ovl_buff[0] + st->ovl_size, buf_in[0],
|
|
copy * sizeof(int32_t));
|
|
|
|
if (stereo)
|
|
memcpy(st->ovl_buff[1] + st->ovl_size, buf_in[1],
|
|
copy * sizeof(int32_t));
|
|
|
|
if (!last && have + copy < src_frame_sz)
|
|
{
|
|
/* still not enough to process at least one frame */
|
|
st->ovl_size += copy;
|
|
return 0;
|
|
}
|
|
|
|
/* recursively call ourselves to process the overlap buffer */
|
|
have = st->ovl_size;
|
|
st->ovl_size = 0;
|
|
|
|
if (copy == data_len)
|
|
{
|
|
assert(have + copy <= FIXED_BUFSIZE);
|
|
return tdspeed_apply(buf_out, st->ovl_buff, have+copy, last,
|
|
out_size);
|
|
}
|
|
|
|
assert(have + copy <= FIXED_BUFSIZE);
|
|
i = tdspeed_apply(buf_out, st->ovl_buff, have+copy, -1, out_size);
|
|
|
|
dest[0] = buf_out[0] + i;
|
|
dest[1] = buf_out[1] + i;
|
|
|
|
/* readjust pointers to account for data already consumed */
|
|
next_frame = copy - src_frame_sz + st->src_step;
|
|
prev_frame = next_frame - st->ovl_shift;
|
|
}
|
|
else
|
|
{
|
|
dest[0] = buf_out[0];
|
|
dest[1] = buf_out[1];
|
|
|
|
next_frame = prev_frame = 0;
|
|
|
|
if (st->ovl_shift > 0)
|
|
next_frame += st->ovl_shift;
|
|
else
|
|
prev_frame += -st->ovl_shift;
|
|
}
|
|
|
|
st->ovl_shift = 0;
|
|
|
|
/* process all complete frames */
|
|
while (data_len - next_frame >= src_frame_sz)
|
|
{
|
|
/* find frame overlap by autocorelation */
|
|
int32_t const INC1 = 8;
|
|
int32_t const INC2 = 32;
|
|
|
|
int64_t min_delta = ~(1ll << 63); /* most positive */
|
|
shift = 0;
|
|
|
|
/* Power of 2 of a 28bit number requires 56bits, can accumulate
|
|
256times in a 64bit variable. */
|
|
assert(st->dst_step / INC2 <= 256);
|
|
assert(next_frame + st->shift_max - 1 + st->dst_step - 1 < data_len);
|
|
assert(prev_frame + st->dst_step - 1 < data_len);
|
|
|
|
for (i = 0; i < st->shift_max; i += INC1)
|
|
{
|
|
int64_t delta = 0;
|
|
|
|
curr = buf_in[0] + next_frame + i;
|
|
prev = buf_in[0] + prev_frame;
|
|
|
|
for (j = 0; j < st->dst_step; j += INC2, curr += INC2, prev += INC2)
|
|
{
|
|
int32_t diff = *curr - *prev;
|
|
delta += (int64_t)diff * diff;
|
|
|
|
if (delta >= min_delta)
|
|
goto skip;
|
|
}
|
|
|
|
if (stereo)
|
|
{
|
|
curr = buf_in[1] + next_frame + i;
|
|
prev = buf_in[1] + prev_frame;
|
|
|
|
for (j = 0; j < st->dst_step; j += INC2, curr += INC2, prev += INC2)
|
|
{
|
|
int32_t diff = *curr - *prev;
|
|
delta += (int64_t)diff * diff;
|
|
|
|
if (delta >= min_delta)
|
|
goto skip;
|
|
}
|
|
}
|
|
|
|
min_delta = delta;
|
|
shift = i;
|
|
skip:;
|
|
}
|
|
|
|
/* overlap fading-out previous frame with fading-in current frame */
|
|
curr = buf_in[0] + next_frame + shift;
|
|
prev = buf_in[0] + prev_frame;
|
|
|
|
d = dest[0];
|
|
|
|
assert(next_frame + shift + st->dst_step - 1 < data_len);
|
|
assert(prev_frame + st->dst_step - 1 < data_len);
|
|
assert(dest[0] - buf_out[0] + st->dst_step - 1 < out_size);
|
|
|
|
for (i = 0, j = st->dst_step; j; i++, j--)
|
|
{
|
|
*d++ = (*curr++ * (int64_t)i +
|
|
*prev++ * (int64_t)j) >> st->dst_order;
|
|
}
|
|
|
|
dest[0] = d;
|
|
|
|
if (stereo)
|
|
{
|
|
curr = buf_in[1] + next_frame + shift;
|
|
prev = buf_in[1] + prev_frame;
|
|
|
|
d = dest[1];
|
|
|
|
for (i = 0, j = st->dst_step; j; i++, j--)
|
|
{
|
|
assert(d < buf_out[1] + out_size);
|
|
|
|
*d++ = (*curr++ * (int64_t)i +
|
|
*prev++ * (int64_t)j) >> st->dst_order;
|
|
}
|
|
|
|
dest[1] = d;
|
|
}
|
|
|
|
/* adjust pointers for next frame */
|
|
prev_frame = next_frame + shift + st->dst_step;
|
|
next_frame += st->src_step;
|
|
|
|
/* here next_frame - prev_frame = src_step - dst_step - shift */
|
|
assert(next_frame - prev_frame == st->src_step - st->dst_step - shift);
|
|
}
|
|
|
|
/* now deal with remaining partial frames */
|
|
if (last == -1)
|
|
{
|
|
/* special overlap buffer processing: remember frame shift only */
|
|
st->ovl_shift = next_frame - prev_frame;
|
|
}
|
|
else if (last != 0)
|
|
{
|
|
/* last call: purge all remaining data to output buffer */
|
|
i = data_len - prev_frame;
|
|
|
|
assert(dest[0] + i <= buf_out[0] + out_size);
|
|
memcpy(dest[0], buf_in[0] + prev_frame, i * sizeof(int32_t));
|
|
|
|
dest[0] += i;
|
|
|
|
if (stereo)
|
|
{
|
|
assert(dest[1] + i <= buf_out[1] + out_size);
|
|
memcpy(dest[1], buf_in[1] + prev_frame, i * sizeof(int32_t));
|
|
dest[1] += i;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
/* preserve remaining data + needed overlap data for next call */
|
|
st->ovl_shift = next_frame - prev_frame;
|
|
i = (st->ovl_shift < 0) ? next_frame : prev_frame;
|
|
st->ovl_size = data_len - i;
|
|
|
|
assert(st->ovl_size <= FIXED_BUFSIZE);
|
|
memcpy(st->ovl_buff[0], buf_in[0] + i, st->ovl_size * sizeof(int32_t));
|
|
|
|
if (stereo)
|
|
memcpy(st->ovl_buff[1], buf_in[1] + i, st->ovl_size * sizeof(int32_t));
|
|
}
|
|
|
|
return dest[0] - buf_out[0];
|
|
}
|
|
|
|
long tdspeed_est_output_size()
|
|
{
|
|
return TDSPEED_OUTBUFSIZE;
|
|
}
|
|
|
|
long tdspeed_est_input_size(long size)
|
|
{
|
|
struct tdspeed_state_s *st = &tdspeed_state;
|
|
|
|
size = (size - st->ovl_size) * st->src_step / st->dst_step;
|
|
|
|
if (size < 0)
|
|
size = 0;
|
|
|
|
return size;
|
|
}
|
|
|
|
int tdspeed_doit(int32_t *src[], int count)
|
|
{
|
|
dsp_src = src;
|
|
count = tdspeed_apply( (int32_t *[2]) { outbuf[0], outbuf[1] },
|
|
src, count, 0, TDSPEED_OUTBUFSIZE);
|
|
|
|
src[0] = outbuf[0];
|
|
src[1] = outbuf[1];
|
|
|
|
return count;
|
|
}
|
|
|