rockbox/apps/codec_thread.c

752 lines
20 KiB
C
Raw Normal View History

/***************************************************************************
* __________ __ ___.
* Open \______ \ ____ ____ | | _\_ |__ _______ ___
* Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
* Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
* Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
* \/ \/ \/ \/ \/
* $Id$
*
* Copyright (C) 2005-2007 Miika Pekkarinen
* Copyright (C) 2007-2008 Nicolas Pennequin
* Copyright (C) 2011 Michael Sevakis
*
* 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 "config.h"
#include "system.h"
#include "kernel.h"
#include "codecs.h"
#include "codec_thread.h"
#include "pcmbuf.h"
#include "audio_thread.h"
#include "playback.h"
#include "buffering.h"
#include "dsp_core.h"
#include "metadata.h"
#include "settings.h"
/* Define LOGF_ENABLE to enable logf output in this file */
/*#define LOGF_ENABLE*/
#include "logf.h"
/* macros to enable logf for queues
logging on SYS_TIMEOUT can be disabled */
#ifdef SIMULATOR
/* Define this for logf output of all queuing except SYS_TIMEOUT */
#define PLAYBACK_LOGQUEUES
/* Define this to logf SYS_TIMEOUT messages */
/*#define PLAYBACK_LOGQUEUES_SYS_TIMEOUT*/
#endif
#ifdef PLAYBACK_LOGQUEUES
#define LOGFQUEUE logf
#else
#define LOGFQUEUE(...)
#endif
#ifdef PLAYBACK_LOGQUEUES_SYS_TIMEOUT
#define LOGFQUEUE_SYS_TIMEOUT logf
#else
#define LOGFQUEUE_SYS_TIMEOUT(...)
#endif
/* Variables are commented with the threads that use them:
* A=audio, C=codec
* - = reads only
*
* Unless otherwise noted, the extern variables are located
* in playback.c.
*/
/* Q_LOAD_CODEC parameter data */
struct codec_load_info
{
int hid; /* audio handle id (specify < 0 to use afmt) */
int afmt; /* codec specification (AFMT_*) */
};
/** --- Main state control --- **/
static int codec_type = AFMT_UNKNOWN; /* Codec type (C,A-) */
/* Private interfaces to main playback control */
extern void audio_codec_update_elapsed(unsigned long elapsed);
extern void audio_codec_update_offset(size_t offset);
extern void audio_codec_complete(int status);
extern void audio_codec_seek_complete(void);
extern struct codec_api ci; /* from codecs.c */
/* Codec thread */
static unsigned int codec_thread_id; /* For modifying thread priority later */
static struct event_queue codec_queue SHAREDBSS_ATTR;
static struct queue_sender_list codec_queue_sender_list SHAREDBSS_ATTR;
/* Workaround stack overflow in opus codec on highmem devices (see FS#13060). */
#if !defined(CPU_COLDFIRE) && (MEMORYSIZE >= 8) && defined(IRAMSIZE) && IRAMSIZE > (32 * 1024)
#define WORKAROUND_FS13060 0x800
#else
#define WORKAROUND_FS13060 0
#endif
static long codec_stack[(DEFAULT_STACK_SIZE + 0x2000 + WORKAROUND_FS13060)/sizeof(long)] IBSS_ATTR;
static const char codec_thread_name[] = "codec";
static void unload_codec(void);
/* Messages are only ever sent one at a time to the codec from the audio
thread. This is important for correct operation unless playback is
stopped. */
/* static routines */
static void codec_queue_ack(intptr_t ackme)
{
queue_reply(&codec_queue, ackme);
}
static intptr_t codec_queue_send(long id, intptr_t data)
{
return queue_send(&codec_queue, id, data);
}
/* Poll the state of the codec queue. Returns < 0 if the message is urgent
and any state should exit, > 0 if it's a run message (and it was
scrubbed), 0 if message was ignored. */
static int codec_check_queue__have_msg(void)
{
struct queue_event ev;
queue_peek(&codec_queue, &ev);
/* Seek, pause or stop? Just peek and return if so. Codec
must handle the command after returing. Inserts will not
be allowed until it complies. */
switch (ev.id)
{
case Q_CODEC_SEEK:
LOGFQUEUE("codec - Q_CODEC_SEEK %ld", ev.id);
return -1;
case Q_CODEC_PAUSE:
LOGFQUEUE("codec - Q_CODEC_PAUSE %ld", ev.id);
return -1;
case Q_CODEC_STOP:
LOGFQUEUE("codec - Q_CODEC_STOP %ld", ev.id);
return -1;
}
/* This is in error in this context unless it's "go, go, go!" */
queue_wait(&codec_queue, &ev);
if (ev.id == Q_CODEC_RUN)
{
logf("codec < Q_CODEC_RUN: already running!");
codec_queue_ack(Q_CODEC_RUN);
return 1;
}
/* Ignore it */
logf("codec < bad req %ld (%s)", ev.id, __func__);
codec_queue_ack(Q_NULL);
return 0;
}
/* Does the audio format type equal CODEC_TYPE_ENCODER? */
static inline bool type_is_encoder(int afmt)
{
#ifdef AUDIO_HAVE_RECORDING
return (afmt & CODEC_TYPE_MASK) == CODEC_TYPE_ENCODER;
#else
return false;
(void)afmt;
#endif
}
/**************************************/
/** --- Miscellaneous external functions --- **/
const char * get_codec_filename(int cod_spec)
{
const char *fname;
#ifdef HAVE_RECORDING
/* Can choose decoder or encoder if one available */
int type = cod_spec & CODEC_TYPE_MASK;
int afmt = cod_spec & CODEC_AFMT_MASK;
if ((unsigned)afmt >= AFMT_NUM_CODECS)
type = AFMT_UNKNOWN | (type & CODEC_TYPE_MASK);
fname = (type == CODEC_TYPE_ENCODER) ?
audio_formats[afmt].codec_enc_root_fn :
audio_formats[afmt].codec_root_fn;
logf("%s: %d - %s",
(type == CODEC_TYPE_ENCODER) ? "Encoder" : "Decoder",
afmt, fname ? fname : "<unknown>");
#else /* !HAVE_RECORDING */
/* Always decoder */
if ((unsigned)cod_spec >= AFMT_NUM_CODECS)
cod_spec = AFMT_UNKNOWN;
fname = audio_formats[cod_spec].codec_root_fn;
logf("Codec: %d - %s", cod_spec, fname ? fname : "<unknown>");
#endif /* HAVE_RECORDING */
return fname;
}
/* Borrow the codec thread and return the ID */
void codec_thread_do_callback(void (*fn)(void), unsigned int *id)
{
/* Set id before telling thread to call something; it may be
* needed before this function returns. */
if (id != NULL)
*id = codec_thread_id;
/* Codec thread will signal just before entering callback */
LOGFQUEUE("codec >| Q_CODEC_DO_CALLBACK");
codec_queue_send(Q_CODEC_DO_CALLBACK, (intptr_t)fn);
}
/** --- codec API callbacks --- **/
static void codec_pcmbuf_insert_callback(
const void *ch1, const void *ch2, int count)
{
struct dsp_buffer src;
src.remcount = count;
src.pin[0] = ch1;
src.pin[1] = ch2;
src.proc_mask = 0;
while (LIKELY(queue_empty(&codec_queue)) ||
codec_check_queue__have_msg() >= 0)
{
struct dsp_buffer dst;
dst.remcount = 0;
dst.bufcount = MAX(src.remcount, 1024); /* Arbitrary min request */
if ((dst.p16out = pcmbuf_request_buffer(&dst.bufcount)) == NULL)
{
cancel_cpu_boost();
/* It may be awhile before space is available but we want
"instant" response to any message */
queue_wait_w_tmo(&codec_queue, NULL, HZ/20);
}
else
{
dsp_process(ci.dsp, &src, &dst);
if (dst.remcount > 0)
{
pcmbuf_write_complete(dst.remcount, ci.id3->elapsed,
ci.id3->offset);
}
else if (src.remcount <= 0)
{
return; /* No input remains and DSP purged */
}
}
Update software recording engine to latest codec interface. Basically, just give it a good rewrite. Software codec recording can be implemented in a more straightforward and simple manner and made more robust through the better codec control now available. Encoded audio buffer uses a packed format instead of fixed-size chunks and uses smaller data headers leading to more efficient usage. The greatest benefit is with a VBR format like wavpack which needs to request a maximum size but only actually ends up committing part of that request. No guard buffers are used for either PCM or encoded audio. PCM is read into the codec's provided buffer and mono conversion done at that time in the core if required. Any highly-specialized sample conversion is still done within the codec itself, such as 32-bit (wavpack) or interleaved mono (mp3). There is no longer a separate filename array. All metadata goes onto the main encoded audio buffer, eliminating any predermined file limit on the buffer as well as not wasting the space for unused path queue slots. The core and codec interface is less awkward and a bit more sensible. Some less useful interface features were removed. Threads are kept on narrow code paths ie. the audio thread never calls encoding functions and the codec thread never calls file functions as before. Codecs no longer call file functions directly. Writes are buffered in the core and data written to storage in larger chunks to speed up flushing of data. In fact, codecs are no longer aware of the stream being a file at all and have no access to the fd. SPDIF frequency detection no longer requires a restart of recording or plugging the source before entering the screen. It will poll for changes and update when stopped or prerecording (which does discard now-invalid prerecorded data). I've seen to it that writing a proper header on full disk works when the format makes it reasonably practical to do so. Other cases may have incorrect data sizes but sample info will be in tact. File left that way may play anyway. mp3_enc.codec acquires the ability to write 'Info' headers with LAME tags to make it gapless (bonus). Change-Id: I670685166d5eb32ef58ef317f50b8af766ceb653 Reviewed-on: http://gerrit.rockbox.org/493 Reviewed-by: Michael Sevakis <jethead71@rockbox.org> Tested-by: Michael Sevakis <jethead71@rockbox.org>
2013-06-22 20:41:16 +00:00
}
}
/* helper function, not a callback */
static bool codec_advance_buffer_counters(size_t amount)
{
if (bufadvance(ci.audio_hid, amount) < 0)
{
ci.curpos = ci.filesize;
return false;
}
ci.curpos += amount;
return true;
}
/* copy up-to size bytes into ptr and return the actual size copied */
static size_t codec_filebuf_callback(void *ptr, size_t size)
{
ssize_t copy_n = bufread(ci.audio_hid, size, ptr);
/* Nothing requested OR nothing left */
if (copy_n <= 0)
return 0;
/* Update read and other position pointers */
codec_advance_buffer_counters(copy_n);
/* Return the actual amount of data copied to the buffer */
return copy_n;
}
static void * codec_request_buffer_callback(size_t *realsize, size_t reqsize)
{
size_t copy_n = reqsize;
ssize_t ret;
void *ptr;
ret = bufgetdata(ci.audio_hid, reqsize, &ptr);
if (ret >= 0)
copy_n = MIN((size_t)ret, reqsize);
else
copy_n = 0;
if (copy_n == 0)
ptr = NULL;
*realsize = copy_n;
return ptr;
}
static void codec_advance_buffer_callback(size_t amount)
{
if (!codec_advance_buffer_counters(amount))
return;
audio_codec_update_offset(ci.curpos);
}
static bool codec_seek_buffer_callback(size_t newpos)
{
logf("codec_seek_buffer_callback");
int ret = bufseek(ci.audio_hid, newpos);
if (ret == 0)
{
ci.curpos = newpos;
return true;
}
return false;
}
static void codec_seek_complete_callback(void)
{
logf("seek_complete");
/* Clear DSP */
dsp_configure(ci.dsp, DSP_FLUSH, 0);
/* Sync position */
audio_codec_update_offset(ci.curpos);
/* Post notification to audio thread */
audio_codec_seek_complete();
/* Wait for urgent or go message */
do
{
queue_wait(&codec_queue, NULL);
}
while (codec_check_queue__have_msg() == 0);
}
static void codec_configure_callback(int setting, intptr_t value)
{
dsp_configure(ci.dsp, setting, value);
}
static long codec_get_command_callback(intptr_t *param)
{
yield();
if (LIKELY(queue_empty(&codec_queue)))
return CODEC_ACTION_NULL; /* As you were */
/* Process the message - return requested action and data (if any should
be expected) */
while (1)
{
long action = CODEC_ACTION_NULL;
struct queue_event ev;
queue_peek(&codec_queue, &ev); /* Find out what it is */
Update software recording engine to latest codec interface. Basically, just give it a good rewrite. Software codec recording can be implemented in a more straightforward and simple manner and made more robust through the better codec control now available. Encoded audio buffer uses a packed format instead of fixed-size chunks and uses smaller data headers leading to more efficient usage. The greatest benefit is with a VBR format like wavpack which needs to request a maximum size but only actually ends up committing part of that request. No guard buffers are used for either PCM or encoded audio. PCM is read into the codec's provided buffer and mono conversion done at that time in the core if required. Any highly-specialized sample conversion is still done within the codec itself, such as 32-bit (wavpack) or interleaved mono (mp3). There is no longer a separate filename array. All metadata goes onto the main encoded audio buffer, eliminating any predermined file limit on the buffer as well as not wasting the space for unused path queue slots. The core and codec interface is less awkward and a bit more sensible. Some less useful interface features were removed. Threads are kept on narrow code paths ie. the audio thread never calls encoding functions and the codec thread never calls file functions as before. Codecs no longer call file functions directly. Writes are buffered in the core and data written to storage in larger chunks to speed up flushing of data. In fact, codecs are no longer aware of the stream being a file at all and have no access to the fd. SPDIF frequency detection no longer requires a restart of recording or plugging the source before entering the screen. It will poll for changes and update when stopped or prerecording (which does discard now-invalid prerecorded data). I've seen to it that writing a proper header on full disk works when the format makes it reasonably practical to do so. Other cases may have incorrect data sizes but sample info will be in tact. File left that way may play anyway. mp3_enc.codec acquires the ability to write 'Info' headers with LAME tags to make it gapless (bonus). Change-Id: I670685166d5eb32ef58ef317f50b8af766ceb653 Reviewed-on: http://gerrit.rockbox.org/493 Reviewed-by: Michael Sevakis <jethead71@rockbox.org> Tested-by: Michael Sevakis <jethead71@rockbox.org>
2013-06-22 20:41:16 +00:00
intptr_t id = ev.id;
switch (id)
{
Update software recording engine to latest codec interface. Basically, just give it a good rewrite. Software codec recording can be implemented in a more straightforward and simple manner and made more robust through the better codec control now available. Encoded audio buffer uses a packed format instead of fixed-size chunks and uses smaller data headers leading to more efficient usage. The greatest benefit is with a VBR format like wavpack which needs to request a maximum size but only actually ends up committing part of that request. No guard buffers are used for either PCM or encoded audio. PCM is read into the codec's provided buffer and mono conversion done at that time in the core if required. Any highly-specialized sample conversion is still done within the codec itself, such as 32-bit (wavpack) or interleaved mono (mp3). There is no longer a separate filename array. All metadata goes onto the main encoded audio buffer, eliminating any predermined file limit on the buffer as well as not wasting the space for unused path queue slots. The core and codec interface is less awkward and a bit more sensible. Some less useful interface features were removed. Threads are kept on narrow code paths ie. the audio thread never calls encoding functions and the codec thread never calls file functions as before. Codecs no longer call file functions directly. Writes are buffered in the core and data written to storage in larger chunks to speed up flushing of data. In fact, codecs are no longer aware of the stream being a file at all and have no access to the fd. SPDIF frequency detection no longer requires a restart of recording or plugging the source before entering the screen. It will poll for changes and update when stopped or prerecording (which does discard now-invalid prerecorded data). I've seen to it that writing a proper header on full disk works when the format makes it reasonably practical to do so. Other cases may have incorrect data sizes but sample info will be in tact. File left that way may play anyway. mp3_enc.codec acquires the ability to write 'Info' headers with LAME tags to make it gapless (bonus). Change-Id: I670685166d5eb32ef58ef317f50b8af766ceb653 Reviewed-on: http://gerrit.rockbox.org/493 Reviewed-by: Michael Sevakis <jethead71@rockbox.org> Tested-by: Michael Sevakis <jethead71@rockbox.org>
2013-06-22 20:41:16 +00:00
case Q_NULL:
LOGFQUEUE("codec < Q_NULL");
break;
case Q_CODEC_RUN: /* Already running */
LOGFQUEUE("codec < Q_CODEC_RUN");
break;
case Q_CODEC_PAUSE: /* Stay here and wait */
LOGFQUEUE("codec < Q_CODEC_PAUSE");
queue_wait(&codec_queue, &ev); /* Remove message */
codec_queue_ack(Q_CODEC_PAUSE);
queue_wait(&codec_queue, NULL); /* Wait for next (no remove) */
continue;
case Q_CODEC_SEEK: /* Audio wants codec to seek */
LOGFQUEUE("codec < Q_CODEC_SEEK %ld", ev.data);
*param = ev.data;
action = CODEC_ACTION_SEEK_TIME;
trigger_cpu_boost();
break;
case Q_CODEC_STOP: /* Must only return 0 in main loop */
Update software recording engine to latest codec interface. Basically, just give it a good rewrite. Software codec recording can be implemented in a more straightforward and simple manner and made more robust through the better codec control now available. Encoded audio buffer uses a packed format instead of fixed-size chunks and uses smaller data headers leading to more efficient usage. The greatest benefit is with a VBR format like wavpack which needs to request a maximum size but only actually ends up committing part of that request. No guard buffers are used for either PCM or encoded audio. PCM is read into the codec's provided buffer and mono conversion done at that time in the core if required. Any highly-specialized sample conversion is still done within the codec itself, such as 32-bit (wavpack) or interleaved mono (mp3). There is no longer a separate filename array. All metadata goes onto the main encoded audio buffer, eliminating any predermined file limit on the buffer as well as not wasting the space for unused path queue slots. The core and codec interface is less awkward and a bit more sensible. Some less useful interface features were removed. Threads are kept on narrow code paths ie. the audio thread never calls encoding functions and the codec thread never calls file functions as before. Codecs no longer call file functions directly. Writes are buffered in the core and data written to storage in larger chunks to speed up flushing of data. In fact, codecs are no longer aware of the stream being a file at all and have no access to the fd. SPDIF frequency detection no longer requires a restart of recording or plugging the source before entering the screen. It will poll for changes and update when stopped or prerecording (which does discard now-invalid prerecorded data). I've seen to it that writing a proper header on full disk works when the format makes it reasonably practical to do so. Other cases may have incorrect data sizes but sample info will be in tact. File left that way may play anyway. mp3_enc.codec acquires the ability to write 'Info' headers with LAME tags to make it gapless (bonus). Change-Id: I670685166d5eb32ef58ef317f50b8af766ceb653 Reviewed-on: http://gerrit.rockbox.org/493 Reviewed-by: Michael Sevakis <jethead71@rockbox.org> Tested-by: Michael Sevakis <jethead71@rockbox.org>
2013-06-22 20:41:16 +00:00
LOGFQUEUE("codec < Q_CODEC_STOP: %ld", ev.data);
#ifdef HAVE_RECORDING
if (type_is_encoder(codec_type))
{
/* Stream finish request (soft stop)? */
if (ev.data && param)
{
/* ev.data is pointer to size */
*param = ev.data;
action = CODEC_ACTION_STREAM_FINISH;
break;
}
}
else
#endif /* HAVE_RECORDING */
{
dsp_configure(ci.dsp, DSP_FLUSH, 0); /* Discontinuity */
}
return CODEC_ACTION_HALT; /* Leave in queue */
default: /* This is in error in this context. */
logf("codec bad req %ld (%s)", ev.id, __func__);
id = Q_NULL;
}
queue_wait(&codec_queue, &ev); /* Actually remove it */
codec_queue_ack(id);
return action;
}
}
static bool codec_loop_track_callback(void)
{
return global_settings.repeat_mode == REPEAT_ONE;
}
/** --- CODEC THREAD --- **/
/* Handle Q_CODEC_LOAD */
static void load_codec(const struct codec_load_info *ev_data)
{
int status = CODEC_ERROR;
/* Save a local copy so we can let the audio thread go ASAP */
struct codec_load_info data = *ev_data;
bool const encoder = type_is_encoder(data.afmt);
if (codec_type != AFMT_UNKNOWN)
{
/* Must have unloaded it first */
logf("a codec is already loaded");
if (data.hid >= 0)
bufclose(data.hid);
return;
}
trigger_cpu_boost();
if (!encoder)
{
/* Do this now because codec may set some things up at load time */
dsp_configure(ci.dsp, DSP_RESET, 0);
}
if (data.hid >= 0)
{
/* First try buffer load */
status = codec_load_buf(data.hid, &ci);
bufclose(data.hid);
}
if (status < 0)
{
/* Either not a valid handle or the buffer method failed */
const char *codec_fn = get_codec_filename(data.afmt);
if (codec_fn)
status = codec_load_file(codec_fn, &ci);
}
Update software recording engine to latest codec interface. Basically, just give it a good rewrite. Software codec recording can be implemented in a more straightforward and simple manner and made more robust through the better codec control now available. Encoded audio buffer uses a packed format instead of fixed-size chunks and uses smaller data headers leading to more efficient usage. The greatest benefit is with a VBR format like wavpack which needs to request a maximum size but only actually ends up committing part of that request. No guard buffers are used for either PCM or encoded audio. PCM is read into the codec's provided buffer and mono conversion done at that time in the core if required. Any highly-specialized sample conversion is still done within the codec itself, such as 32-bit (wavpack) or interleaved mono (mp3). There is no longer a separate filename array. All metadata goes onto the main encoded audio buffer, eliminating any predermined file limit on the buffer as well as not wasting the space for unused path queue slots. The core and codec interface is less awkward and a bit more sensible. Some less useful interface features were removed. Threads are kept on narrow code paths ie. the audio thread never calls encoding functions and the codec thread never calls file functions as before. Codecs no longer call file functions directly. Writes are buffered in the core and data written to storage in larger chunks to speed up flushing of data. In fact, codecs are no longer aware of the stream being a file at all and have no access to the fd. SPDIF frequency detection no longer requires a restart of recording or plugging the source before entering the screen. It will poll for changes and update when stopped or prerecording (which does discard now-invalid prerecorded data). I've seen to it that writing a proper header on full disk works when the format makes it reasonably practical to do so. Other cases may have incorrect data sizes but sample info will be in tact. File left that way may play anyway. mp3_enc.codec acquires the ability to write 'Info' headers with LAME tags to make it gapless (bonus). Change-Id: I670685166d5eb32ef58ef317f50b8af766ceb653 Reviewed-on: http://gerrit.rockbox.org/493 Reviewed-by: Michael Sevakis <jethead71@rockbox.org> Tested-by: Michael Sevakis <jethead71@rockbox.org>
2013-06-22 20:41:16 +00:00
/* Types must agree */
if (status >= 0 && encoder == !!codec_get_enc_callback())
{
codec_type = data.afmt;
codec_queue_ack(Q_CODEC_LOAD);
return;
}
/* Failed - get rid of it */
unload_codec();
}
/* Handle Q_CODEC_RUN */
static void run_codec(void)
{
bool const encoder = type_is_encoder(codec_type);
int status;
if (codec_type == AFMT_UNKNOWN)
{
logf("no codec to run");
return;
}
codec_queue_ack(Q_CODEC_RUN);
trigger_cpu_boost();
dsp_configure(ci.dsp, DSP_SET_OUT_FREQUENCY, pcmbuf_get_frequency());
if (!encoder)
{
/* This will be either the initial buffered offset or where it left off
if it remained buffered and we're skipping back to it and it is best
to have ci.curpos in sync with the handle's read position - it's the
codec's responsibility to ensure it has the correct positions -
playback is sorta dumb and only has a vague idea about what to
buffer based upon what metadata has to say */
ci.curpos = bufftell(ci.audio_hid);
/* Pin the codec's audio data in place */
buf_pin_handle(ci.audio_hid, true);
}
status = codec_run_proc();
if (!encoder)
{
/* Codec is done with it - let it move */
buf_pin_handle(ci.audio_hid, false);
/* Notify audio that we're done for better or worse - advise of the
status */
audio_codec_complete(status);
}
}
/* Handle Q_CODEC_SEEK */
static void seek_codec(unsigned long time)
{
if (codec_type == AFMT_UNKNOWN)
{
logf("no codec to seek");
codec_queue_ack(Q_CODEC_SEEK);
codec_seek_complete_callback();
return;
}
/* Post it up one level */
queue_post(&codec_queue, Q_CODEC_SEEK, time);
codec_queue_ack(Q_CODEC_SEEK);
/* Have to run it again */
run_codec();
}
/* Handle Q_CODEC_UNLOAD */
static void unload_codec(void)
{
/* Tell codec to clean up */
codec_type = AFMT_UNKNOWN;
codec_close();
}
/* Handle Q_CODEC_DO_CALLBACK */
static void do_callback(void (* callback)(void))
{
codec_queue_ack(Q_CODEC_DO_CALLBACK);
if (callback)
{
commit_discard_idcache();
callback();
commit_dcache();
}
}
/* Codec thread function */
static void NORETURN_ATTR codec_thread(void)
{
struct queue_event ev;
Update software recording engine to latest codec interface. Basically, just give it a good rewrite. Software codec recording can be implemented in a more straightforward and simple manner and made more robust through the better codec control now available. Encoded audio buffer uses a packed format instead of fixed-size chunks and uses smaller data headers leading to more efficient usage. The greatest benefit is with a VBR format like wavpack which needs to request a maximum size but only actually ends up committing part of that request. No guard buffers are used for either PCM or encoded audio. PCM is read into the codec's provided buffer and mono conversion done at that time in the core if required. Any highly-specialized sample conversion is still done within the codec itself, such as 32-bit (wavpack) or interleaved mono (mp3). There is no longer a separate filename array. All metadata goes onto the main encoded audio buffer, eliminating any predermined file limit on the buffer as well as not wasting the space for unused path queue slots. The core and codec interface is less awkward and a bit more sensible. Some less useful interface features were removed. Threads are kept on narrow code paths ie. the audio thread never calls encoding functions and the codec thread never calls file functions as before. Codecs no longer call file functions directly. Writes are buffered in the core and data written to storage in larger chunks to speed up flushing of data. In fact, codecs are no longer aware of the stream being a file at all and have no access to the fd. SPDIF frequency detection no longer requires a restart of recording or plugging the source before entering the screen. It will poll for changes and update when stopped or prerecording (which does discard now-invalid prerecorded data). I've seen to it that writing a proper header on full disk works when the format makes it reasonably practical to do so. Other cases may have incorrect data sizes but sample info will be in tact. File left that way may play anyway. mp3_enc.codec acquires the ability to write 'Info' headers with LAME tags to make it gapless (bonus). Change-Id: I670685166d5eb32ef58ef317f50b8af766ceb653 Reviewed-on: http://gerrit.rockbox.org/493 Reviewed-by: Michael Sevakis <jethead71@rockbox.org> Tested-by: Michael Sevakis <jethead71@rockbox.org>
2013-06-22 20:41:16 +00:00
while (1)
{
cancel_cpu_boost();
queue_wait(&codec_queue, &ev);
switch (ev.id)
{
case Q_CODEC_LOAD:
LOGFQUEUE("codec < Q_CODEC_LOAD");
load_codec((const struct codec_load_info *)ev.data);
break;
case Q_CODEC_RUN:
LOGFQUEUE("codec < Q_CODEC_RUN");
run_codec();
break;
case Q_CODEC_PAUSE:
LOGFQUEUE("codec < Q_CODEC_PAUSE");
break;
case Q_CODEC_SEEK:
LOGFQUEUE("codec < Q_CODEC_SEEK: %lu", (unsigned long)ev.data);
seek_codec(ev.data);
break;
case Q_CODEC_UNLOAD:
LOGFQUEUE("codec < Q_CODEC_UNLOAD");
unload_codec();
break;
case Q_CODEC_DO_CALLBACK:
LOGFQUEUE("codec < Q_CODEC_DO_CALLBACK");
do_callback((void (*)(void))ev.data);
break;
default:
LOGFQUEUE("codec < default : %ld", ev.id);
}
}
}
/** --- Miscellaneous external interfaces -- **/
/* Initialize playback's codec interface */
void INIT_ATTR codec_thread_init(void)
{
/* Init API */
ci.dsp = dsp_get_config(CODEC_IDX_AUDIO);
ci.codec_get_buffer = codec_get_buffer_callback;
ci.pcmbuf_insert = codec_pcmbuf_insert_callback;
ci.set_elapsed = audio_codec_update_elapsed;
ci.read_filebuf = codec_filebuf_callback;
ci.request_buffer = codec_request_buffer_callback;
ci.advance_buffer = codec_advance_buffer_callback;
ci.seek_buffer = codec_seek_buffer_callback;
ci.seek_complete = codec_seek_complete_callback;
ci.set_offset = audio_codec_update_offset;
ci.configure = codec_configure_callback;
ci.get_command = codec_get_command_callback;
ci.loop_track = codec_loop_track_callback;
/* Init threading */
queue_init(&codec_queue, false);
codec_thread_id = create_thread(
codec_thread, codec_stack, sizeof(codec_stack), 0,
codec_thread_name IF_PRIO(, PRIORITY_PLAYBACK)
IF_COP(, CPU));
queue_enable_queue_send(&codec_queue, &codec_queue_sender_list,
codec_thread_id);
}
#ifdef HAVE_PRIORITY_SCHEDULING
/* Obtain codec thread's current priority */
int codec_thread_get_priority(void)
{
return thread_get_priority(codec_thread_id);
}
/* Set the codec thread's priority and return the old value */
int codec_thread_set_priority(int priority)
{
return thread_set_priority(codec_thread_id, priority);
}
#endif /* HAVE_PRIORITY_SCHEDULING */
/** --- Functions for audio thread use --- **/
/* Load a decoder or encoder and set the format type */
bool codec_load(int hid, int cod_spec)
{
struct codec_load_info parm = { hid, cod_spec };
LOGFQUEUE("audio >| codec Q_CODEC_LOAD: %d, %d", hid, cod_spec);
return codec_queue_send(Q_CODEC_LOAD, (intptr_t)&parm) != 0;
}
/* Begin decoding the current file */
void codec_go(void)
{
LOGFQUEUE("audio >| codec Q_CODEC_RUN");
codec_queue_send(Q_CODEC_RUN, 0);
}
/* Instruct the codec to seek to the specified time (should be properly
paused or stopped first to avoid possible buffering deadlock) */
void codec_seek(long time)
{
LOGFQUEUE("audio > codec Q_CODEC_SEEK: %ld", time);
codec_queue_send(Q_CODEC_SEEK, time);
}
/* Pause the codec and make it wait for further instructions inside the
command callback */
bool codec_pause(void)
{
LOGFQUEUE("audio >| codec Q_CODEC_PAUSE");
return codec_queue_send(Q_CODEC_PAUSE, 0) != Q_NULL;
}
/* Stop codec if running - codec stays resident if loaded */
void codec_stop(void)
{
/* Wait until it's in the main loop */
Update software recording engine to latest codec interface. Basically, just give it a good rewrite. Software codec recording can be implemented in a more straightforward and simple manner and made more robust through the better codec control now available. Encoded audio buffer uses a packed format instead of fixed-size chunks and uses smaller data headers leading to more efficient usage. The greatest benefit is with a VBR format like wavpack which needs to request a maximum size but only actually ends up committing part of that request. No guard buffers are used for either PCM or encoded audio. PCM is read into the codec's provided buffer and mono conversion done at that time in the core if required. Any highly-specialized sample conversion is still done within the codec itself, such as 32-bit (wavpack) or interleaved mono (mp3). There is no longer a separate filename array. All metadata goes onto the main encoded audio buffer, eliminating any predermined file limit on the buffer as well as not wasting the space for unused path queue slots. The core and codec interface is less awkward and a bit more sensible. Some less useful interface features were removed. Threads are kept on narrow code paths ie. the audio thread never calls encoding functions and the codec thread never calls file functions as before. Codecs no longer call file functions directly. Writes are buffered in the core and data written to storage in larger chunks to speed up flushing of data. In fact, codecs are no longer aware of the stream being a file at all and have no access to the fd. SPDIF frequency detection no longer requires a restart of recording or plugging the source before entering the screen. It will poll for changes and update when stopped or prerecording (which does discard now-invalid prerecorded data). I've seen to it that writing a proper header on full disk works when the format makes it reasonably practical to do so. Other cases may have incorrect data sizes but sample info will be in tact. File left that way may play anyway. mp3_enc.codec acquires the ability to write 'Info' headers with LAME tags to make it gapless (bonus). Change-Id: I670685166d5eb32ef58ef317f50b8af766ceb653 Reviewed-on: http://gerrit.rockbox.org/493 Reviewed-by: Michael Sevakis <jethead71@rockbox.org> Tested-by: Michael Sevakis <jethead71@rockbox.org>
2013-06-22 20:41:16 +00:00
LOGFQUEUE("audio >| codec Q_CODEC_STOP: 0");
while (codec_queue_send(Q_CODEC_STOP, 0) != Q_NULL);
}
Update software recording engine to latest codec interface. Basically, just give it a good rewrite. Software codec recording can be implemented in a more straightforward and simple manner and made more robust through the better codec control now available. Encoded audio buffer uses a packed format instead of fixed-size chunks and uses smaller data headers leading to more efficient usage. The greatest benefit is with a VBR format like wavpack which needs to request a maximum size but only actually ends up committing part of that request. No guard buffers are used for either PCM or encoded audio. PCM is read into the codec's provided buffer and mono conversion done at that time in the core if required. Any highly-specialized sample conversion is still done within the codec itself, such as 32-bit (wavpack) or interleaved mono (mp3). There is no longer a separate filename array. All metadata goes onto the main encoded audio buffer, eliminating any predermined file limit on the buffer as well as not wasting the space for unused path queue slots. The core and codec interface is less awkward and a bit more sensible. Some less useful interface features were removed. Threads are kept on narrow code paths ie. the audio thread never calls encoding functions and the codec thread never calls file functions as before. Codecs no longer call file functions directly. Writes are buffered in the core and data written to storage in larger chunks to speed up flushing of data. In fact, codecs are no longer aware of the stream being a file at all and have no access to the fd. SPDIF frequency detection no longer requires a restart of recording or plugging the source before entering the screen. It will poll for changes and update when stopped or prerecording (which does discard now-invalid prerecorded data). I've seen to it that writing a proper header on full disk works when the format makes it reasonably practical to do so. Other cases may have incorrect data sizes but sample info will be in tact. File left that way may play anyway. mp3_enc.codec acquires the ability to write 'Info' headers with LAME tags to make it gapless (bonus). Change-Id: I670685166d5eb32ef58ef317f50b8af766ceb653 Reviewed-on: http://gerrit.rockbox.org/493 Reviewed-by: Michael Sevakis <jethead71@rockbox.org> Tested-by: Michael Sevakis <jethead71@rockbox.org>
2013-06-22 20:41:16 +00:00
#ifdef HAVE_RECORDING
/* Tells codec to take final encoding step and then exit -
Returns minimum buffer size required or 0 if complete */
size_t codec_finish_stream(void)
{
size_t size = 0;
LOGFQUEUE("audio >| codec Q_CODEC_STOP: &size");
if (codec_queue_send(Q_CODEC_STOP, (intptr_t)&size) != Q_NULL)
{
/* Sync to keep size in scope and get response */
LOGFQUEUE("audio >| codec Q_NULL");
codec_queue_send(Q_NULL, 0);
if (size == 0)
codec_stop(); /* Replied with 0 size */
}
/* else thread running in the main loop */
return size;
}
#endif /* HAVE_RECORDING */
/* Call the codec's exit routine and close all references */
void codec_unload(void)
{
codec_stop();
LOGFQUEUE("audio >| codec Q_CODEC_UNLOAD");
codec_queue_send(Q_CODEC_UNLOAD, 0);
}
/* Return the afmt type of the loaded codec - sticks until calling
codec_unload unless initial load failed */
int codec_loaded(void)
{
return codec_type;
}