sdrangel/plugins/samplemimo/bladerf2mimo/bladerf2mithread.cpp

344 lines
10 KiB
C++

///////////////////////////////////////////////////////////////////////////////////
// Copyright (C) 2019-2020 Edouard Griffiths, F4EXB <f4exb06@gmail.com> //
// //
// 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 //
// (at your option) any later version. //
// //
// 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 <algorithm>
#include "bladerf2/devicebladerf2.h"
#include "dsp/samplemififo.h"
#include "bladerf2mithread.h"
BladeRF2MIThread::BladeRF2MIThread(struct bladerf* dev, QObject* parent) :
QThread(parent),
m_running(false),
m_dev(dev),
m_sampleFifo(nullptr),
m_iqOrder(true)
{
qDebug("BladeRF2MIThread::BladeRF2MIThread");
m_buf = new qint16[2*DeviceBladeRF2::blockSize*2];
for (unsigned int i = 0; i < 2; i++) {
m_convertBuffer[i].resize(DeviceBladeRF2::blockSize, Sample{0,0});
}
}
BladeRF2MIThread::~BladeRF2MIThread()
{
qDebug("BladeRF2MIThread::~BladeRF2MIThread");
if (m_running) {
stopWork();
}
delete[] m_buf;
}
void BladeRF2MIThread::startWork()
{
m_startWaitMutex.lock();
start();
while(!m_running) {
m_startWaiter.wait(&m_startWaitMutex, 100);
}
m_startWaitMutex.unlock();
}
void BladeRF2MIThread::stopWork()
{
m_running = false;
wait();
}
void BladeRF2MIThread::setLog2Decimation(unsigned int log2_decim)
{
m_log2Decim = log2_decim;
}
unsigned int BladeRF2MIThread::getLog2Decimation() const
{
return m_log2Decim;
}
void BladeRF2MIThread::setFcPos(int fcPos)
{
m_fcPos = fcPos;
}
int BladeRF2MIThread::getFcPos() const
{
return m_fcPos;
}
void BladeRF2MIThread::run()
{
int res;
m_running = true;
m_startWaiter.wakeAll();
int status = bladerf_sync_config(m_dev, BLADERF_RX_X2, BLADERF_FORMAT_SC16_Q11, 64, DeviceBladeRF2::blockSize, 32, 1500);
if (status < 0)
{
qCritical("BladeRF2MIThread::run: cannot configure streams: %s", bladerf_strerror(status));
}
else
{
qDebug("BladeRF2MIThread::run: start running loop");
while (m_running)
{
res = bladerf_sync_rx(m_dev, m_buf, DeviceBladeRF2::blockSize*2, nullptr, 1500);
if (res < 0)
{
qCritical("BladeRF2MIThread::run sync Rx error: %s", bladerf_strerror(res));
break;
}
callback(m_buf, DeviceBladeRF2::blockSize);
}
qDebug("BladeRF2MIThread::run: stop running loop");
m_running = false;
}
}
void BladeRF2MIThread::callback(const qint16* buf, qint32 samplesPerChannel)
{
int status = bladerf_deinterleave_stream_buffer(BLADERF_RX_X2, BLADERF_FORMAT_SC16_Q11 , samplesPerChannel*2, (void *) buf);
if (status < 0)
{
qCritical("BladeRF2MIThread::callback: cannot de-interleave buffer: %s", bladerf_strerror(status));
return;
}
std::vector<SampleVector::const_iterator> vbegin;
int lengths[2];
for (unsigned int channel = 0; channel < 2; channel++)
{
if (m_iqOrder) {
lengths[channel] = channelCallbackIQ(&buf[2*samplesPerChannel*channel], 2*samplesPerChannel, channel);
} else {
lengths[channel] = channelCallbackQI(&buf[2*samplesPerChannel*channel], 2*samplesPerChannel, channel);
}
vbegin.push_back(m_convertBuffer[channel].begin());
}
if (lengths[0] == lengths[1])
{
m_sampleFifo->writeSync(vbegin, lengths[0]);
}
else
{
qWarning("BladeRF2MIThread::callback: unequal channel lengths: [0]=%d [1]=%d", lengths[0], lengths[1]);
m_sampleFifo->writeSync(vbegin, (std::min)(lengths[0], lengths[1]));
}
}
int BladeRF2MIThread::channelCallbackIQ(const qint16* buf, qint32 len, int channel)
{
SampleVector::iterator it = m_convertBuffer[channel].begin();
if (m_log2Decim == 0)
{
m_decimatorsIQ[channel].decimate1(&it, buf, len);
}
else
{
if (m_fcPos == 0) // Infra
{
switch (m_log2Decim)
{
case 1:
m_decimatorsIQ[channel].decimate2_inf(&it, buf, len);
break;
case 2:
m_decimatorsIQ[channel].decimate4_inf(&it, buf, len);
break;
case 3:
m_decimatorsIQ[channel].decimate8_inf(&it, buf, len);
break;
case 4:
m_decimatorsIQ[channel].decimate16_inf(&it, buf, len);
break;
case 5:
m_decimatorsIQ[channel].decimate32_inf(&it, buf, len);
break;
case 6:
m_decimatorsIQ[channel].decimate64_inf(&it, buf, len);
break;
default:
break;
}
}
else if (m_fcPos == 1) // Supra
{
switch (m_log2Decim)
{
case 1:
m_decimatorsIQ[channel].decimate2_sup(&it, buf, len);
break;
case 2:
m_decimatorsIQ[channel].decimate4_sup(&it, buf, len);
break;
case 3:
m_decimatorsIQ[channel].decimate8_sup(&it, buf, len);
break;
case 4:
m_decimatorsIQ[channel].decimate16_sup(&it, buf, len);
break;
case 5:
m_decimatorsIQ[channel].decimate32_sup(&it, buf, len);
break;
case 6:
m_decimatorsIQ[channel].decimate64_sup(&it, buf, len);
break;
default:
break;
}
}
else if (m_fcPos == 2) // Center
{
switch (m_log2Decim)
{
case 1:
m_decimatorsIQ[channel].decimate2_cen(&it, buf, len);
break;
case 2:
m_decimatorsIQ[channel].decimate4_cen(&it, buf, len);
break;
case 3:
m_decimatorsIQ[channel].decimate8_cen(&it, buf, len);
break;
case 4:
m_decimatorsIQ[channel].decimate16_cen(&it, buf, len);
break;
case 5:
m_decimatorsIQ[channel].decimate32_cen(&it, buf, len);
break;
case 6:
m_decimatorsIQ[channel].decimate64_cen(&it, buf, len);
break;
default:
break;
}
}
}
return it - m_convertBuffer[channel].begin();
}
int BladeRF2MIThread::channelCallbackQI(const qint16* buf, qint32 len, int channel)
{
SampleVector::iterator it = m_convertBuffer[channel].begin();
if (m_log2Decim == 0)
{
m_decimatorsQI[channel].decimate1(&it, buf, len);
}
else
{
if (m_fcPos == 0) // Infra
{
switch (m_log2Decim)
{
case 1:
m_decimatorsQI[channel].decimate2_inf(&it, buf, len);
break;
case 2:
m_decimatorsQI[channel].decimate4_inf(&it, buf, len);
break;
case 3:
m_decimatorsQI[channel].decimate8_inf(&it, buf, len);
break;
case 4:
m_decimatorsQI[channel].decimate16_inf(&it, buf, len);
break;
case 5:
m_decimatorsQI[channel].decimate32_inf(&it, buf, len);
break;
case 6:
m_decimatorsQI[channel].decimate64_inf(&it, buf, len);
break;
default:
break;
}
}
else if (m_fcPos == 1) // Supra
{
switch (m_log2Decim)
{
case 1:
m_decimatorsQI[channel].decimate2_sup(&it, buf, len);
break;
case 2:
m_decimatorsQI[channel].decimate4_sup(&it, buf, len);
break;
case 3:
m_decimatorsQI[channel].decimate8_sup(&it, buf, len);
break;
case 4:
m_decimatorsQI[channel].decimate16_sup(&it, buf, len);
break;
case 5:
m_decimatorsQI[channel].decimate32_sup(&it, buf, len);
break;
case 6:
m_decimatorsQI[channel].decimate64_sup(&it, buf, len);
break;
default:
break;
}
}
else if (m_fcPos == 2) // Center
{
switch (m_log2Decim)
{
case 1:
m_decimatorsQI[channel].decimate2_cen(&it, buf, len);
break;
case 2:
m_decimatorsQI[channel].decimate4_cen(&it, buf, len);
break;
case 3:
m_decimatorsQI[channel].decimate8_cen(&it, buf, len);
break;
case 4:
m_decimatorsQI[channel].decimate16_cen(&it, buf, len);
break;
case 5:
m_decimatorsQI[channel].decimate32_cen(&it, buf, len);
break;
case 6:
m_decimatorsQI[channel].decimate64_cen(&it, buf, len);
break;
default:
break;
}
}
}
return it - m_convertBuffer[channel].begin();
}