1
0
mirror of https://github.com/f4exb/sdrangel.git synced 2024-11-18 14:21:49 -05:00
sdrangel/plugins/samplesink/limesdroutput/limesdroutputthread.cpp
2017-05-25 20:13:34 +02:00

177 lines
5.4 KiB
C++

///////////////////////////////////////////////////////////////////////////////////
// Copyright (C) 2017 Edouard Griffiths, F4EXB //
// //
// 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 as version 3 of the License, or //
// //
// This program is distributed in the hope that it will be useful, //
// but WITHOUT ANY WARRANTY; without even the implied warranty of //
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the //
// GNU General Public License V3 for more details. //
// //
// You should have received a copy of the GNU General Public License //
// along with this program. If not, see <http://www.gnu.org/licenses/>. //
///////////////////////////////////////////////////////////////////////////////////
#include <errno.h>
#include "limesdroutputthread.h"
#include "limesdroutputsettings.h"
LimeSDROutputThread::LimeSDROutputThread(lms_stream_t* stream, SampleSourceFifo* sampleFifo, QObject* parent) :
QThread(parent),
m_running(false),
m_stream(stream),
m_sampleFifo(sampleFifo),
m_sampleRate(5000000),
m_log2Interp(0),
m_fcPos(LimeSDROutputSettings::FC_POS_CENTER)
{
m_sampleFifo->resize(16*LIMESDROUTPUT_BLOCKSIZE);
}
LimeSDROutputThread::~LimeSDROutputThread()
{
stopWork();
}
void LimeSDROutputThread::startWork()
{
if (m_running) return; // return if running already
m_startWaitMutex.lock();
start();
while(!m_running)
m_startWaiter.wait(&m_startWaitMutex, 100);
m_startWaitMutex.unlock();
}
void LimeSDROutputThread::stopWork()
{
if (!m_running) return; // return if not running
m_running = false;
wait();
}
void LimeSDROutputThread::setLog2Interpolation(unsigned int log2_interp)
{
m_log2Interp = log2_interp;
}
void LimeSDROutputThread::setFcPos(int fcPos)
{
m_fcPos = fcPos;
}
void LimeSDROutputThread::run()
{
//int count, msleep, mdelta;
int res;
//lms_stream_status_t streamStatus;
lms_stream_meta_t metadata; //Use metadata for additional control over sample receive function behaviour
metadata.flushPartialPacket = false; //Do not discard data remainder when read size differs from packet size
metadata.waitForTimestamp = false; //Do not wait for specific timestamps
m_running = true;
m_startWaiter.wakeAll();
if (LMS_StartStream(m_stream) < 0) {
qCritical("LimeSDROutputThread::run: could not start stream");
} else {
qDebug("LimeSDROutputThread::run: stream started");
}
// count = 0;
// msleep = LIMESDROUTPUT_BLOCKSIZE/(m_sampleRate/1e6f);
// mdelta = msleep/100;
// msleep = (3*msleep)/4; // to start faster
while (m_running)
{
callback(m_buf, LIMESDROUTPUT_BLOCKSIZE);
res = LMS_SendStream(m_stream, (void *) m_buf, LIMESDROUTPUT_BLOCKSIZE, &metadata, 1000000);
if (res < 0)
{
qCritical("LimeSDROutputThread::run write error: %s", strerror(errno));
break;
}
else if (res != LIMESDROUTPUT_BLOCKSIZE)
{
qDebug("LimeSDROutputThread::run written %d/%d samples", res, LIMESDROUTPUT_BLOCKSIZE);
}
// usleep(msleep);
//
// if (count < 10)
// {
// count++;
// }
// else
// {
// if (LMS_GetStreamStatus(m_stream, &streamStatus) == 0)
// {
// if (streamStatus.fifoFilledCount < (4*streamStatus.fifoSize)/5) { // FIFO at 80%
// msleep -= mdelta;
// } else {
// msleep += mdelta;
// }
// }
//
// count = 0;
// }
}
if (LMS_StopStream(m_stream) < 0) {
qCritical("LimeSDROutputThread::run: could not stop stream");
} else {
qDebug("LimeSDROutputThread::run: stream stopped");
}
m_running = false;
}
// Interpolate according to specified log2 (ex: log2=4 => decim=16)
void LimeSDROutputThread::callback(qint16* buf, qint32 len)
{
SampleVector::iterator beginRead;
m_sampleFifo->readAdvance(beginRead, len/(1<<m_log2Interp));
beginRead -= len;
if (m_log2Interp == 0)
{
m_interpolators.interpolate1(&beginRead, buf, len*2);
}
else
{
switch (m_log2Interp)
{
case 1:
m_interpolators.interpolate2_cen(&beginRead, buf, len*2);
break;
case 2:
m_interpolators.interpolate4_cen(&beginRead, buf, len*2);
break;
case 3:
m_interpolators.interpolate8_cen(&beginRead, buf, len*2);
break;
case 4:
m_interpolators.interpolate16_cen(&beginRead, buf, len*2);
break;
case 5:
m_interpolators.interpolate32_cen(&beginRead, buf, len*2);
break;
case 6:
m_interpolators.interpolate64_cen(&beginRead, buf, len*2);
break;
default:
break;
}
}
}