dump1090-fa/dump1090.c

952 lines
39 KiB
C
Raw Normal View History

2015-01-20 16:49:01 +00:00
// Part of dump1090, a Mode S message decoder for RTLSDR devices.
//
2015-01-20 16:49:01 +00:00
// dump1090.c: main program & miscellany
//
// Copyright (c) 2014-2016 Oliver Jowett <oliver@mutability.co.uk>
//
// This file is free software: you may copy, redistribute and/or modify it
2015-01-20 16:49:01 +00:00
// 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 file 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
2015-01-20 16:49:01 +00:00
// General Public License for more details.
//
// You should have received a copy of the GNU General Public License
2015-01-20 16:49:01 +00:00
// along with this program. If not, see <http://www.gnu.org/licenses/>.
// This file incorporates work covered by the following copyright and
2015-01-20 16:49:01 +00:00
// permission notice:
//
// Copyright (C) 2012 by Salvatore Sanfilippo <antirez@gmail.com>
//
// All rights reserved.
//
2015-01-20 16:49:01 +00:00
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
2015-01-20 16:49:01 +00:00
// * 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
// HOLDER 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 "dump1090.h"
Move all converters to starch-based implementations (#97) * Switch all conversion routines to use starch. main user-visible changes: * ensure you check out submodules ('git clone --recurse-submodules") * --version shows the CPU features and DSP implementations in use * --wisdom allows overriding of the built-in architecture wisdom * --dcfilter no longer supported * "starch-benchmark" binary will benchmark all options on the current machine and can produce a wisdom file to feed to the --wisdom option If you have a usecase for --dcfilter, please get in touch and let me know - it's an edge case and for now there's no starch/DSP support for it, but support can be written if needed. In almost all cases the new conversion routines are slightly or substantially faster than the old conversion routines. The only case that is slower is SC16/SC16Q11 on a Pi 0, which is around 10% slower due to changing from heavily approximated lookup tables to higher quality results (but SC16 is probably already out of reach of a Pi 0) * No need to build with SC16Q11_TABLE_BITS any more * Add oneoff/uc8_capture_stats (reads a UC8 capture; measures min/max/mean I and Q) * Switch UC8 conversion to 127.4 center, 128 range. Looking at actual UC8 captures from a RTL2832, the mean I and Q are actually at 127.4, so use that as the zero point. This means that the resulting I/Q maximum values could be as large as 127.6. Switch to 128 for simplicity. * Switch to the new UC8 zero offset in benchmarks, fix some bugs * Fix some bugs in SC16/SC16Q11 validation, tighten the max error requirements * Ditch UC8 approximation path, add a NEON VRQSQRTE path. * Tweak the SC16 exact path, add a new impl that uses a mix of u32 & floats. * SC16Q11 impl tweaks: * add a u32->float exact path * ditch the approximation path * add a NEON VRSQRTE path * add a 12-bit table path (using the full signed I/Q value, not absolute value) * Ditch SC16 approximation path, add NEON vrsqrte path * Add oneoff/dsp_error_measurement This runs sample input through the DSP functions that are allowed to be inexact and dumps the results as a TSV suitable for feeding to gnuplot to look at the actual errors. * Update make clean, make wisdom targets * Update wisdom based on benchmarking * Preserve the raw wisdom benchmark data * Update to latest starch * Update .gitignore for new wisdom files * Update starch generated code * Build starch-benchmark as part of the 'all' target * Use wisdom from /etc/dump1090-fa/wisdom.local if present * Package starch-benchmark and a helper script to generate local wisdom data * Remove submodules in preparation for importing them directly * Import cpu_features v0.6.0 from https://github.com/google/cpu_features/releases/tag/v0.6.0 * Import starch at commit a725c8491dc33a321565d451b385131e589d8490 from https://github.com/flightaware/starch
2021-01-21 11:45:00 +00:00
#include "cpu.h"
#include <stdarg.h>
struct _Modes Modes;
//
// ============================= Utility functions ==========================
//
static void log_with_timestamp(const char *format, ...) __attribute__((format (printf, 1, 2) ));
static void log_with_timestamp(const char *format, ...)
{
char timebuf[128];
char msg[1024];
time_t now;
struct tm local;
va_list ap;
now = time(NULL);
localtime_r(&now, &local);
strftime(timebuf, 128, "%c %Z", &local);
timebuf[127] = 0;
va_start(ap, format);
vsnprintf(msg, 1024, format, ap);
va_end(ap);
msg[1023] = 0;
fprintf(stderr, "%s %s\n", timebuf, msg);
}
static void sigintHandler(int dummy) {
MODES_NOTUSED(dummy);
signal(SIGINT, SIG_DFL); // reset signal handler - bit extra safety
Modes.exit = 1; // Signal to threads that we are done
log_with_timestamp("Caught SIGINT, shutting down..\n");
}
static void sigtermHandler(int dummy) {
MODES_NOTUSED(dummy);
signal(SIGTERM, SIG_DFL); // reset signal handler - bit extra safety
Modes.exit = 1; // Signal to threads that we are done
log_with_timestamp("Caught SIGTERM, shutting down..\n");
}
void receiverPositionChanged(float lat, float lon, float alt)
{
log_with_timestamp("Autodetected receiver location: %.5f, %.5f at %.0fm AMSL", lat, lon, alt);
writeJsonToFile("receiver.json", generateReceiverJson); // location changed
}
//
// =============================== Initialization ===========================
//
static void modesInitConfig(void) {
// Default everything to zero/NULL
memset(&Modes, 0, sizeof(Modes));
// Now initialise things that should not be 0/NULL to their defaults
Modes.gain = MODES_DEFAULT_GAIN;
Modes.freq = MODES_DEFAULT_FREQ;
Modes.check_crc = 1;
Modes.fix_df = 1;
Modes.interactive_display_ttl = MODES_INTERACTIVE_DISPLAY_TTL;
Modes.json_interval = 1000;
Modes.json_stats_interval = 60000;
Modes.json_location_accuracy = 1;
Modes.maxRange = 1852 * 300; // 300NM default max range
Modes.mode_ac_auto = 1;
Modes.net_heartbeat_interval = MODES_NET_HEARTBEAT_INTERVAL;
Modes.net_output_flush_size = 1300;
Modes.net_output_flush_interval = 500;
// adaptive
Modes.adaptive_min_gain_db = 0;
Modes.adaptive_max_gain_db = 99999;
Modes.adaptive_duty_cycle = 0.5;
Modes.adaptive_burst_control = false;
Modes.adaptive_burst_alpha = 2.0 / (5 + 1);
Modes.adaptive_burst_change_delay = 5;
Modes.adaptive_burst_loud_runlength = 10;
Modes.adaptive_burst_loud_rate = 5.0;
Modes.adaptive_burst_quiet_runlength = 10;
Modes.adaptive_burst_quiet_rate = 5.0;
Modes.adaptive_range_control = false;
Modes.adaptive_range_alpha = 2.0 / (5 + 1);
Modes.adaptive_range_percentile = 40;
Modes.adaptive_range_change_delay = 10;
Modes.adaptive_range_scan_delay = 300;
Modes.adaptive_range_rescan_delay = 3600;
sdrInitConfig();
}
//
//=========================================================================
//
static void modesInit(void) {
int i;
Modes.sample_rate = 2400000.0;
// Allocate the various buffers used by Modes
Modes.trailing_samples = (MODES_PREAMBLE_US + MODES_LONG_MSG_BITS + 16) * 1e-6 * Modes.sample_rate;
if ( ((Modes.log10lut = (uint16_t *) malloc(sizeof(uint16_t) * 256 * 256) ) == NULL) )
{
fprintf(stderr, "Out of memory allocating data buffer.\n");
exit(1);
}
if (!fifo_create(MODES_MAG_BUFFERS, MODES_MAG_BUF_SAMPLES + Modes.trailing_samples, Modes.trailing_samples)) {
fprintf(stderr, "Out of memory allocating FIFO\n");
exit(1);
}
// Validate the users Lat/Lon home location inputs
if ( (Modes.fUserLat > 90.0) // Latitude must be -90 to +90
|| (Modes.fUserLat < -90.0) // and
|| (Modes.fUserLon > 360.0) // Longitude must be -180 to +360
|| (Modes.fUserLon < -180.0) ) {
Modes.fUserLat = Modes.fUserLon = 0.0;
} else if (Modes.fUserLon > 180.0) { // If Longitude is +180 to +360, make it -180 to 0
Modes.fUserLon -= 360.0;
}
// If both Lat and Lon are 0.0 then the users location is either invalid/not-set, or (s)he's in the
// Atlantic ocean off the west coast of Africa. This is unlikely to be correct.
// Set the user LatLon valid flag only if either Lat or Lon are non zero. Note the Greenwich meridian
// is at 0.0 Lon,so we must check for either fLat or fLon being non zero not both.
// Testing the flag at runtime will be much quicker than ((fLon != 0.0) || (fLat != 0.0))
Modes.bUserFlags &= ~MODES_USER_LATLON_VALID;
if ((Modes.fUserLat != 0.0) || (Modes.fUserLon != 0.0)) {
Modes.bUserFlags |= MODES_USER_LATLON_VALID;
}
// Limit the maximum requested raw output size to less than one Ethernet Block
if (Modes.net_output_flush_size > (MODES_OUT_FLUSH_SIZE))
{Modes.net_output_flush_size = MODES_OUT_FLUSH_SIZE;}
if (Modes.net_output_flush_interval > (MODES_OUT_FLUSH_INTERVAL))
{Modes.net_output_flush_interval = MODES_OUT_FLUSH_INTERVAL;}
if (Modes.net_sndbuf_size > (MODES_NET_SNDBUF_MAX))
{Modes.net_sndbuf_size = MODES_NET_SNDBUF_MAX;}
// Prepare the log10 lookup table: 100log10(x)
Modes.log10lut[0] = 0; // poorly defined..
for (i = 1; i <= 65535; i++) {
Modes.log10lut[i] = (uint16_t) round(100.0 * log10(i));
}
// Prepare error correction tables
modesChecksumInit(Modes.nfix_crc);
icaoFilterInit();
modeACInit();
if (Modes.show_only)
icaoFilterAdd(Modes.show_only);
}
//
//=========================================================================
//
// We use a thread reading data in background, while the main thread
// handles decoding and visualization of data to the user.
//
// The reading thread calls the RTLSDR API to read data asynchronously, and
// uses a callback to populate the data buffer.
//
// A Mutex is used to avoid races with the decoding thread.
//
//
//=========================================================================
//
// We read data using a thread, so the main thread only handles decoding
// without caring about data acquisition
//
static void *readerThreadEntryPoint(void *arg)
{
MODES_NOTUSED(arg);
sdrRun();
if (!Modes.exit)
Modes.exit = 2; // unexpected exit
fifo_halt(); // wakes the main thread, if it's still waiting
return NULL;
}
//
// ============================== Snip mode =================================
//
// Get raw IQ samples and filter everything is < than the specified level
// for more than 256 samples in order to reduce example file size
//
static void snipMode(int level) {
int i, q;
uint64_t c = 0;
while ((i = getchar()) != EOF && (q = getchar()) != EOF) {
if (abs(i-127) < level && abs(q-127) < level) {
c++;
if (c > MODES_PREAMBLE_SIZE) continue;
} else {
c = 0;
}
putchar(i);
putchar(q);
}
}
//
// ================================ Main ====================================
//
static void showVersion()
{
printf("-----------------------------------------------------------------------------\n");
printf("| dump1090 ModeS Receiver %45s |\n", MODES_DUMP1090_VARIANT " " MODES_DUMP1090_VERSION);
printf("| build options: %-58s |\n",
""
#ifdef ENABLE_RTLSDR
"ENABLE_RTLSDR "
#endif
#ifdef ENABLE_BLADERF
"ENABLE_BLADERF "
#endif
2019-12-09 21:55:40 +00:00
#ifdef ENABLE_HACKRF
"ENABLE_HACKRF "
#endif
#ifdef ENABLE_LIMESDR
"ENABLE_LIMESDR "
#endif
);
printf("-----------------------------------------------------------------------------\n");
Move all converters to starch-based implementations (#97) * Switch all conversion routines to use starch. main user-visible changes: * ensure you check out submodules ('git clone --recurse-submodules") * --version shows the CPU features and DSP implementations in use * --wisdom allows overriding of the built-in architecture wisdom * --dcfilter no longer supported * "starch-benchmark" binary will benchmark all options on the current machine and can produce a wisdom file to feed to the --wisdom option If you have a usecase for --dcfilter, please get in touch and let me know - it's an edge case and for now there's no starch/DSP support for it, but support can be written if needed. In almost all cases the new conversion routines are slightly or substantially faster than the old conversion routines. The only case that is slower is SC16/SC16Q11 on a Pi 0, which is around 10% slower due to changing from heavily approximated lookup tables to higher quality results (but SC16 is probably already out of reach of a Pi 0) * No need to build with SC16Q11_TABLE_BITS any more * Add oneoff/uc8_capture_stats (reads a UC8 capture; measures min/max/mean I and Q) * Switch UC8 conversion to 127.4 center, 128 range. Looking at actual UC8 captures from a RTL2832, the mean I and Q are actually at 127.4, so use that as the zero point. This means that the resulting I/Q maximum values could be as large as 127.6. Switch to 128 for simplicity. * Switch to the new UC8 zero offset in benchmarks, fix some bugs * Fix some bugs in SC16/SC16Q11 validation, tighten the max error requirements * Ditch UC8 approximation path, add a NEON VRQSQRTE path. * Tweak the SC16 exact path, add a new impl that uses a mix of u32 & floats. * SC16Q11 impl tweaks: * add a u32->float exact path * ditch the approximation path * add a NEON VRSQRTE path * add a 12-bit table path (using the full signed I/Q value, not absolute value) * Ditch SC16 approximation path, add NEON vrsqrte path * Add oneoff/dsp_error_measurement This runs sample input through the DSP functions that are allowed to be inexact and dumps the results as a TSV suitable for feeding to gnuplot to look at the actual errors. * Update make clean, make wisdom targets * Update wisdom based on benchmarking * Preserve the raw wisdom benchmark data * Update to latest starch * Update .gitignore for new wisdom files * Update starch generated code * Build starch-benchmark as part of the 'all' target * Use wisdom from /etc/dump1090-fa/wisdom.local if present * Package starch-benchmark and a helper script to generate local wisdom data * Remove submodules in preparation for importing them directly * Import cpu_features v0.6.0 from https://github.com/google/cpu_features/releases/tag/v0.6.0 * Import starch at commit a725c8491dc33a321565d451b385131e589d8490 from https://github.com/flightaware/starch
2021-01-21 11:45:00 +00:00
}
static void showDSP()
{
printf(" detected runtime CPU features: ");
if (cpu_supports_avx())
printf("AVX ");
if (cpu_supports_avx2())
printf("AVX2 ");
if (cpu_supports_armv7_neon_vfpv4())
printf("ARMv7+NEON+VFPv4 ");
printf("\n");
printf(" selected DSP implementations: \n");
#define SHOW(x) do { \
printf(" %-40s %s\n", #x , starch_ ## x ## _select()->name); \
printf(" %-40s %s\n", #x "_aligned", starch_ ## x ## _aligned_select()->name); \
} while(0)
SHOW(magnitude_uc8);
SHOW(magnitude_power_uc8);
SHOW(magnitude_sc16);
SHOW(magnitude_sc16q11);
SHOW(mean_power_u16);
SHOW(count_above_u16);
Move all converters to starch-based implementations (#97) * Switch all conversion routines to use starch. main user-visible changes: * ensure you check out submodules ('git clone --recurse-submodules") * --version shows the CPU features and DSP implementations in use * --wisdom allows overriding of the built-in architecture wisdom * --dcfilter no longer supported * "starch-benchmark" binary will benchmark all options on the current machine and can produce a wisdom file to feed to the --wisdom option If you have a usecase for --dcfilter, please get in touch and let me know - it's an edge case and for now there's no starch/DSP support for it, but support can be written if needed. In almost all cases the new conversion routines are slightly or substantially faster than the old conversion routines. The only case that is slower is SC16/SC16Q11 on a Pi 0, which is around 10% slower due to changing from heavily approximated lookup tables to higher quality results (but SC16 is probably already out of reach of a Pi 0) * No need to build with SC16Q11_TABLE_BITS any more * Add oneoff/uc8_capture_stats (reads a UC8 capture; measures min/max/mean I and Q) * Switch UC8 conversion to 127.4 center, 128 range. Looking at actual UC8 captures from a RTL2832, the mean I and Q are actually at 127.4, so use that as the zero point. This means that the resulting I/Q maximum values could be as large as 127.6. Switch to 128 for simplicity. * Switch to the new UC8 zero offset in benchmarks, fix some bugs * Fix some bugs in SC16/SC16Q11 validation, tighten the max error requirements * Ditch UC8 approximation path, add a NEON VRQSQRTE path. * Tweak the SC16 exact path, add a new impl that uses a mix of u32 & floats. * SC16Q11 impl tweaks: * add a u32->float exact path * ditch the approximation path * add a NEON VRSQRTE path * add a 12-bit table path (using the full signed I/Q value, not absolute value) * Ditch SC16 approximation path, add NEON vrsqrte path * Add oneoff/dsp_error_measurement This runs sample input through the DSP functions that are allowed to be inexact and dumps the results as a TSV suitable for feeding to gnuplot to look at the actual errors. * Update make clean, make wisdom targets * Update wisdom based on benchmarking * Preserve the raw wisdom benchmark data * Update to latest starch * Update .gitignore for new wisdom files * Update starch generated code * Build starch-benchmark as part of the 'all' target * Use wisdom from /etc/dump1090-fa/wisdom.local if present * Package starch-benchmark and a helper script to generate local wisdom data * Remove submodules in preparation for importing them directly * Import cpu_features v0.6.0 from https://github.com/google/cpu_features/releases/tag/v0.6.0 * Import starch at commit a725c8491dc33a321565d451b385131e589d8490 from https://github.com/flightaware/starch
2021-01-21 11:45:00 +00:00
#undef SHOW
printf("\n");
}
static void showHelp(void)
{
showVersion();
sdrShowHelp();
printf(
" Output modes\n"
"\n"
// ------ 80 char limit ----------------------------------------------------------|
"--raw Show only messages hex values\n"
"--modeac Enable decoding of SSR Modes 3/A & 3/C\n"
"--mlat display raw messages in Beast ascii mode\n"
"--onlyaddr Show only ICAO addresses (testing purposes)\n"
"--metric Use metric units (meters, km/h, ...)\n"
"--gnss Show altitudes as HAE/GNSS when available\n"
"--quiet Disable output to stdout. Use for daemon applications\n"
"--show-only <addr> Show only messages from the given ICAO on stdout\n"
"--snip <level> Strip IQ file removing samples < level\n"
"\n"
" Decoder settings\n"
"\n"
// ------ 80 char limit ----------------------------------------------------------|
"--gain <db> Set gain in dB (default: varies by SDR type)\n"
"--freq <hz> Set frequency (default: 1090 Mhz)\n"
"--fix Enable single-bit error correction using CRC\n"
"--fix-2bit Enable two-bit error correction using CRC\n"
" (use with caution!)\n"
"--no-fix Disable error correction using CRC\n"
"--no-fix-df Disable error correction of the DF message field\n"
" (reduces CPU requirements)\n"
"--no-crc-check Disable messages with broken CRC (discouraged)\n"
"--enable-df24 Enable decoding of DF24 Comm-D ELM messages\n"
"--lat <latitude> Reference/receiver latitude for surface positions\n"
"--lon <longitude> Reference/receiver longitude for surface positions\n"
"--max-range <distance> Absolute maximum range for position decoding (in NM)\n"
"\n"
// ------ 80 char limit ----------------------------------------------------------|
" Adaptive gain\n"
"\n"
"--adaptive-burst Adjust gain for too-loud message bursts\n"
"--adaptive-burst-change-delay <s> Set delay after changing gain before\n"
" resuming burst control (seconds)\n"
"--adaptive-burst-alpha <a> Set burst rate smoothing factor\n"
" (0..1, smaller=more smoothing)\n"
"--adaptive-burst-loud-rate <r> Set burst rate for gain decrease\n"
"--adaptive-burst-loud-runlength <l> Set burst runlength for gain decrease\n"
"--adaptive-burst-quiet-rate <r> Set burst rate for gain increase\n"
"--adaptive-burst-quiet-runlength <l> Set burst runlength for gain increase\n"
"--adaptive-range Adjust gain for target dynamic range\n"
"--adaptive-range-target <db> Set target dynamic range in dB\n"
"--adaptive-range-alpha <a> Set dynamic range noise smoothing factor\n"
" (0..1, smaller=more smoothing)\n"
"--adaptive-range-percentile <p> Set dynamic range noise percentile\n"
"--adaptive-range-change-delay <s> Set delay after changing gain before\n"
" resuming dynamic range control (seconds)\n"
"--adaptive-range-scan-delay <s> Set scan interval for dynamic range\n"
" gain scanning following a gain decrease\n"
" due to an increase in noise (seconds)\n"
"--adaptive-range-rescan-delay <s> Set periodic rescan interval for dynamic\n"
" range gain scanning (seconds)\n"
"--adaptive-min-gain <g> Set gain adjustment range lower limit (dB)\n"
"--adaptive-max-gain <g> Set gain adjustment range upper limit (dB)\n"
"--adaptive-duty-cycle <p> Set adaptive gain duty cycle %% (1..100)\n"
"\n"
// ------ 80 char limit ----------------------------------------------------------|
" Network connections\n"
"\n"
"--net Enable networking with default ports unless overridden\n"
"--no-modeac-auto Don't enable Mode A/C if requested by a net connection\n"
"--net-only Enable just networking, no RTL device or file used\n"
"--net-bind-address <ip> IP address to bind to (use 127.0.0.1 for private)\n"
"--net-ri-port <ports> TCP raw input listen ports (default: 30001)\n"
"--net-ro-port <ports> TCP raw output listen ports (default: 30002)\n"
"--net-sbs-port <ports> TCP BaseStation output listen ports (default: 30003)\n"
"--net-bi-port <ports> TCP Beast input listen ports (default: 30004,30104)\n"
"--net-bo-port <ports> TCP Beast output listen ports (default: 30005)\n"
"--net-stratux-port <ports> TCP Stratux output listen ports (default: disabled)\n"
"--net-ro-size <size> TCP output minimum size (default: 0)\n"
"--net-ro-interval <rate> TCP output memory flush rate in seconds (default: 0)\n"
"--net-heartbeat <rate> TCP heartbeat rate in seconds\n"
" (default: 60 sec; 0 to disable)\n"
"--net-buffer <n> TCP buffer size 64Kb * (2^n) (default: n=0, 64Kb)\n"
"--net-verbatim Make output connections default to verbatim mode\n"
" (forward all messages without correction)\n"
"--forward-mlat Allow forwarding of received mlat results\n"
"\n"
// ------ 80 char limit ----------------------------------------------------------|
" Stats and json output\n"
"\n"
"--stats Show stats summary at exit.\n"
"--stats-every <seconds> Show and reset stats every <seconds> seconds\n"
"--stats-range Collect/show range histogram\n"
"--write-json <dir> Periodically write json output to <dir>\n"
" (for serving by a separate webserver)\n"
"--write-json-every <t> Write json aircraft output every t seconds (default 1)\n"
"--json-stats-every <t> Write json stats output every t seconds (default 60)\n"
"--json-location-accuracy <n> Accuracy of receiver location in json metadata\n"
" (0=no location, 1=approximate, 2=exact)\n"
"\n"
" Interactive mode\n"
"\n"
"--interactive Interactive mode refreshing data on screen.\n"
" Implies --throttle\n"
"--interactive-ttl <sec> Remove from list if idle for <sec>\n"
"--interactive-show-distance Show aircraft distance and bearing\n"
" (requires --lat and --lon)\n"
"--interactive-distance-units <u> Distance units ('km', 'sm', 'nm')\n"
"--interactive-callsign-filter <r> Filter rows by callsign against regex\n"
"\n"
" Misc\n"
"\n"
"--wisdom <path> Read DSP wisdom from given path\n"
Move all converters to starch-based implementations (#97) * Switch all conversion routines to use starch. main user-visible changes: * ensure you check out submodules ('git clone --recurse-submodules") * --version shows the CPU features and DSP implementations in use * --wisdom allows overriding of the built-in architecture wisdom * --dcfilter no longer supported * "starch-benchmark" binary will benchmark all options on the current machine and can produce a wisdom file to feed to the --wisdom option If you have a usecase for --dcfilter, please get in touch and let me know - it's an edge case and for now there's no starch/DSP support for it, but support can be written if needed. In almost all cases the new conversion routines are slightly or substantially faster than the old conversion routines. The only case that is slower is SC16/SC16Q11 on a Pi 0, which is around 10% slower due to changing from heavily approximated lookup tables to higher quality results (but SC16 is probably already out of reach of a Pi 0) * No need to build with SC16Q11_TABLE_BITS any more * Add oneoff/uc8_capture_stats (reads a UC8 capture; measures min/max/mean I and Q) * Switch UC8 conversion to 127.4 center, 128 range. Looking at actual UC8 captures from a RTL2832, the mean I and Q are actually at 127.4, so use that as the zero point. This means that the resulting I/Q maximum values could be as large as 127.6. Switch to 128 for simplicity. * Switch to the new UC8 zero offset in benchmarks, fix some bugs * Fix some bugs in SC16/SC16Q11 validation, tighten the max error requirements * Ditch UC8 approximation path, add a NEON VRQSQRTE path. * Tweak the SC16 exact path, add a new impl that uses a mix of u32 & floats. * SC16Q11 impl tweaks: * add a u32->float exact path * ditch the approximation path * add a NEON VRSQRTE path * add a 12-bit table path (using the full signed I/Q value, not absolute value) * Ditch SC16 approximation path, add NEON vrsqrte path * Add oneoff/dsp_error_measurement This runs sample input through the DSP functions that are allowed to be inexact and dumps the results as a TSV suitable for feeding to gnuplot to look at the actual errors. * Update make clean, make wisdom targets * Update wisdom based on benchmarking * Preserve the raw wisdom benchmark data * Update to latest starch * Update .gitignore for new wisdom files * Update starch generated code * Build starch-benchmark as part of the 'all' target * Use wisdom from /etc/dump1090-fa/wisdom.local if present * Package starch-benchmark and a helper script to generate local wisdom data * Remove submodules in preparation for importing them directly * Import cpu_features v0.6.0 from https://github.com/google/cpu_features/releases/tag/v0.6.0 * Import starch at commit a725c8491dc33a321565d451b385131e589d8490 from https://github.com/flightaware/starch
2021-01-21 11:45:00 +00:00
"--version Show version, build and DSP options\n"
"--help Show this help\n"
);
}
// Accumulate stats data from stats_current to stats_periodic, stats_alltime and stats_latest;
// reset stats_current
void flush_stats(uint64_t now);
void flush_stats(uint64_t now)
{
add_stats(&Modes.stats_current, &Modes.stats_periodic, &Modes.stats_periodic);
add_stats(&Modes.stats_current, &Modes.stats_alltime, &Modes.stats_alltime);
add_stats(&Modes.stats_current, &Modes.stats_latest, &Modes.stats_latest);
reset_stats(&Modes.stats_current);
Modes.stats_current.start = Modes.stats_current.end = now;
}
//
//=========================================================================
//
// This function is called a few times every second by main in order to
// perform tasks we need to do continuously, like accepting new clients
// from the net, refreshing the screen in interactive mode, and so forth
//
static void backgroundTasks(void) {
static uint64_t next_stats_display;
static uint64_t next_stats_update;
static uint64_t next_json_stats_update;
static uint64_t next_json, next_history;
uint64_t now = mstime();
if (Modes.sdr_type != SDR_IFILE) {
// don't run these if processing data from a file
icaoFilterExpire();
trackPeriodicUpdate();
}
if (Modes.net) {
2018-01-09 17:13:34 +00:00
modesNetPeriodicWork();
}
// Refresh screen when in interactive mode
if (Modes.interactive) {
interactiveShowData();
}
2020-08-05 04:00:50 +00:00
// copy out reader CPU time and reset it
sdrUpdateCPUTime(&Modes.stats_current.reader_cpu);
// always update end time so it is current when requests arrive
Modes.stats_current.end = mstime();
// 1-minute stats update
if (now >= next_stats_update) {
int i;
if (next_stats_update == 0) {
next_stats_update = now + 60000;
} else {
flush_stats(now); // Ensure stats_latest is up to date
// move stats_latest into 1-min ring buffer
Modes.stats_newest_1min = (Modes.stats_newest_1min + 1) % 15;
Modes.stats_1min[Modes.stats_newest_1min] = Modes.stats_latest;
reset_stats(&Modes.stats_latest);
// recalculate 5-min window
reset_stats(&Modes.stats_5min);
for (i = 0; i < 5; ++i)
add_stats(&Modes.stats_1min[(Modes.stats_newest_1min - i + 15) % 15], &Modes.stats_5min, &Modes.stats_5min);
// recalculate 15-min window
reset_stats(&Modes.stats_15min);
for (i = 0; i < 15; ++i)
add_stats(&Modes.stats_1min[i], &Modes.stats_15min, &Modes.stats_15min);
next_stats_update += 60000;
}
}
// --stats-every display
if (Modes.stats && now >= next_stats_display) {
if (next_stats_display == 0) {
next_stats_display = now + Modes.stats;
} else {
flush_stats(now); // Ensure stats_periodic is up to date
display_stats(&Modes.stats_periodic);
reset_stats(&Modes.stats_periodic);
next_stats_display += Modes.stats;
if (next_stats_display <= now) {
/* something has gone wrong, perhaps the system clock jumped */
next_stats_display = now + Modes.stats;
}
}
}
2014-11-23 16:02:11 +00:00
// json stats update
if (Modes.json_dir && now >= next_json_stats_update) {
if (next_json_stats_update == 0) {
next_json_stats_update = now + Modes.json_stats_interval;
} else {
flush_stats(now); // Ensure everything we'll write is up to date
writeJsonToFile("stats.json", generateStatsJson);
next_json_stats_update += Modes.json_stats_interval;
}
}
if (Modes.json_dir && now >= next_json) {
writeJsonToFile("aircraft.json", generateAircraftJson);
next_json = now + Modes.json_interval;
2014-11-23 16:02:11 +00:00
}
if (now >= next_history) {
int rewrite_receiver_json = (Modes.json_dir && Modes.json_aircraft_history[HISTORY_SIZE-1].content == NULL);
free(Modes.json_aircraft_history[Modes.json_aircraft_history_next].content); // might be NULL, that's OK.
Modes.json_aircraft_history[Modes.json_aircraft_history_next].content =
generateAircraftJson("/data/aircraft.json", &Modes.json_aircraft_history[Modes.json_aircraft_history_next].clen);
if (Modes.json_dir) {
char filebuf[PATH_MAX];
snprintf(filebuf, PATH_MAX, "history_%d.json", Modes.json_aircraft_history_next);
writeJsonToFile(filebuf, generateHistoryJson);
}
Modes.json_aircraft_history_next = (Modes.json_aircraft_history_next+1) % HISTORY_SIZE;
if (rewrite_receiver_json)
writeJsonToFile("receiver.json", generateReceiverJson); // number of history entries changed
next_history = now + HISTORY_INTERVAL;
}
}
//
//=========================================================================
//
2014-07-08 22:34:43 +00:00
//
//=========================================================================
//
static void applyNetDefaults()
{
if (!Modes.net_input_raw_ports)
Modes.net_input_raw_ports = strdup("30001");
if (!Modes.net_output_raw_ports)
Modes.net_output_raw_ports = strdup("30002");
if (!Modes.net_output_sbs_ports)
Modes.net_output_sbs_ports = strdup("30003");
if (!Modes.net_input_beast_ports)
Modes.net_input_beast_ports = strdup("30004,30104");
if (!Modes.net_output_beast_ports)
Modes.net_output_beast_ports = strdup("30005");
}
int main(int argc, char **argv) {
int j;
// Set sane defaults
modesInitConfig();
// signal handlers:
signal(SIGINT, sigintHandler);
signal(SIGTERM, sigtermHandler);
// Parse the command line options
for (j = 1; j < argc; j++) {
int more = j+1 < argc; // There are more arguments
if (!strcmp(argv[j],"--freq") && more) {
Modes.freq = (int) strtoll(argv[++j],NULL,10);
} else if ( (!strcmp(argv[j], "--device") || !strcmp(argv[j], "--device-index")) && more) {
Modes.dev_name = strdup(argv[++j]);
} else if (!strcmp(argv[j],"--gain") && more) {
Modes.gain = atof(argv[++j]);
} else if (!strcmp(argv[j],"--dcfilter")) {
Move all converters to starch-based implementations (#97) * Switch all conversion routines to use starch. main user-visible changes: * ensure you check out submodules ('git clone --recurse-submodules") * --version shows the CPU features and DSP implementations in use * --wisdom allows overriding of the built-in architecture wisdom * --dcfilter no longer supported * "starch-benchmark" binary will benchmark all options on the current machine and can produce a wisdom file to feed to the --wisdom option If you have a usecase for --dcfilter, please get in touch and let me know - it's an edge case and for now there's no starch/DSP support for it, but support can be written if needed. In almost all cases the new conversion routines are slightly or substantially faster than the old conversion routines. The only case that is slower is SC16/SC16Q11 on a Pi 0, which is around 10% slower due to changing from heavily approximated lookup tables to higher quality results (but SC16 is probably already out of reach of a Pi 0) * No need to build with SC16Q11_TABLE_BITS any more * Add oneoff/uc8_capture_stats (reads a UC8 capture; measures min/max/mean I and Q) * Switch UC8 conversion to 127.4 center, 128 range. Looking at actual UC8 captures from a RTL2832, the mean I and Q are actually at 127.4, so use that as the zero point. This means that the resulting I/Q maximum values could be as large as 127.6. Switch to 128 for simplicity. * Switch to the new UC8 zero offset in benchmarks, fix some bugs * Fix some bugs in SC16/SC16Q11 validation, tighten the max error requirements * Ditch UC8 approximation path, add a NEON VRQSQRTE path. * Tweak the SC16 exact path, add a new impl that uses a mix of u32 & floats. * SC16Q11 impl tweaks: * add a u32->float exact path * ditch the approximation path * add a NEON VRSQRTE path * add a 12-bit table path (using the full signed I/Q value, not absolute value) * Ditch SC16 approximation path, add NEON vrsqrte path * Add oneoff/dsp_error_measurement This runs sample input through the DSP functions that are allowed to be inexact and dumps the results as a TSV suitable for feeding to gnuplot to look at the actual errors. * Update make clean, make wisdom targets * Update wisdom based on benchmarking * Preserve the raw wisdom benchmark data * Update to latest starch * Update .gitignore for new wisdom files * Update starch generated code * Build starch-benchmark as part of the 'all' target * Use wisdom from /etc/dump1090-fa/wisdom.local if present * Package starch-benchmark and a helper script to generate local wisdom data * Remove submodules in preparation for importing them directly * Import cpu_features v0.6.0 from https://github.com/google/cpu_features/releases/tag/v0.6.0 * Import starch at commit a725c8491dc33a321565d451b385131e589d8490 from https://github.com/flightaware/starch
2021-01-21 11:45:00 +00:00
#if 0
Modes.dc_filter = 1;
Move all converters to starch-based implementations (#97) * Switch all conversion routines to use starch. main user-visible changes: * ensure you check out submodules ('git clone --recurse-submodules") * --version shows the CPU features and DSP implementations in use * --wisdom allows overriding of the built-in architecture wisdom * --dcfilter no longer supported * "starch-benchmark" binary will benchmark all options on the current machine and can produce a wisdom file to feed to the --wisdom option If you have a usecase for --dcfilter, please get in touch and let me know - it's an edge case and for now there's no starch/DSP support for it, but support can be written if needed. In almost all cases the new conversion routines are slightly or substantially faster than the old conversion routines. The only case that is slower is SC16/SC16Q11 on a Pi 0, which is around 10% slower due to changing from heavily approximated lookup tables to higher quality results (but SC16 is probably already out of reach of a Pi 0) * No need to build with SC16Q11_TABLE_BITS any more * Add oneoff/uc8_capture_stats (reads a UC8 capture; measures min/max/mean I and Q) * Switch UC8 conversion to 127.4 center, 128 range. Looking at actual UC8 captures from a RTL2832, the mean I and Q are actually at 127.4, so use that as the zero point. This means that the resulting I/Q maximum values could be as large as 127.6. Switch to 128 for simplicity. * Switch to the new UC8 zero offset in benchmarks, fix some bugs * Fix some bugs in SC16/SC16Q11 validation, tighten the max error requirements * Ditch UC8 approximation path, add a NEON VRQSQRTE path. * Tweak the SC16 exact path, add a new impl that uses a mix of u32 & floats. * SC16Q11 impl tweaks: * add a u32->float exact path * ditch the approximation path * add a NEON VRSQRTE path * add a 12-bit table path (using the full signed I/Q value, not absolute value) * Ditch SC16 approximation path, add NEON vrsqrte path * Add oneoff/dsp_error_measurement This runs sample input through the DSP functions that are allowed to be inexact and dumps the results as a TSV suitable for feeding to gnuplot to look at the actual errors. * Update make clean, make wisdom targets * Update wisdom based on benchmarking * Preserve the raw wisdom benchmark data * Update to latest starch * Update .gitignore for new wisdom files * Update starch generated code * Build starch-benchmark as part of the 'all' target * Use wisdom from /etc/dump1090-fa/wisdom.local if present * Package starch-benchmark and a helper script to generate local wisdom data * Remove submodules in preparation for importing them directly * Import cpu_features v0.6.0 from https://github.com/google/cpu_features/releases/tag/v0.6.0 * Import starch at commit a725c8491dc33a321565d451b385131e589d8490 from https://github.com/flightaware/starch
2021-01-21 11:45:00 +00:00
#else
fprintf(stderr, "--dcfilter option ignored (please raise an issue on github if you have a usecase that needs this)\n");
#endif
} else if (!strcmp(argv[j],"--measure-noise")) {
// Ignored
} else if (!strcmp(argv[j],"--fix")) {
if (Modes.nfix_crc < 1)
Modes.nfix_crc = 1;
} else if (!strcmp(argv[j],"--fix-2bit")) {
Modes.nfix_crc = 2;
Reduce CPU further in --no-fix-df mode. Add --enable-df24 option. This reinstates the fastpath in the 2.4MHz demodulator that aborts message demodulation early if the DF bits don't match a message type that we know how to decode, or stops early if the DF bits can only correspond to a short 56-bit message. Do this in a more general form where we test against a set of valid DF values, rather than a switch/case. Then populate the sets based on the current error-correction settings (e.g. if we are allowed to repair DF damage with 1 bit error correction, then a DF17 message could appear with any of DF=17, DF=16, DF=19, DF=21, DF=25, or DF=1 and still be correctable to a valid DF17 message) In the default case this produces a small CPU improvement as short messages might be able to stop earlier and a few DF values can be ignored early. With --no-fix-df it produces a larger improvement as relatively few DF values are valid in that mode. To further reduce CPU, the default behaviour has changed to _not_ decode DF24 messages (Comm-D ELM messages). These are rarely seen in the wild and dump1090 can't do anything useful with them. Disabling them allows us to further reduce the set of valid DF values as they effectively use all DF values from 24..31, 25% of all possible values. To re-enable DF24 decoding, use the new `--enable-df24` option. Finally, avoid doing some CRC calculations twice. This was only happening once per valid decoded message, not per demodulation attempt, so it's not such a large win.
2021-03-19 08:52:42 +00:00
} else if (!strcmp(argv[j],"--enable-df24")) {
Modes.enable_df24 = 1;
} else if (!strcmp(argv[j],"--no-fix")) {
Modes.nfix_crc = 0;
} else if (!strcmp(argv[j],"--no-fix-df")) {
Modes.fix_df = 0;
} else if (!strcmp(argv[j],"--no-crc-check")) {
Modes.check_crc = 0;
} else if (!strcmp(argv[j],"--phase-enhance")) {
// Ignored, always enabled
} else if (!strcmp(argv[j],"--raw")) {
Modes.raw = 1;
} else if (!strcmp(argv[j],"--net")) {
Modes.net = 1;
applyNetDefaults();
} else if (!strcmp(argv[j],"--modeac")) {
Modes.mode_ac = 1;
Modes.mode_ac_auto = 0;
} else if (!strcmp(argv[j],"--no-modeac-auto")) {
Modes.mode_ac_auto = 0;
2013-04-07 15:22:02 +00:00
} else if (!strcmp(argv[j],"--net-beast")) {
fprintf(stderr, "--net-beast ignored, use --net-bo-port to control where Beast output is generated\n");
} else if (!strcmp(argv[j],"--net-only")) {
Modes.net = 1;
Modes.sdr_type = SDR_NONE;
applyNetDefaults();
} else if (!strcmp(argv[j],"--net-heartbeat") && more) {
Modes.net_heartbeat_interval = (uint64_t)(1000 * atof(argv[++j]));
} else if (!strcmp(argv[j],"--net-ro-size") && more) {
Modes.net_output_flush_size = atoi(argv[++j]);
} else if (!strcmp(argv[j],"--net-ro-rate") && more) {
Modes.net_output_flush_interval = 1000 * atoi(argv[++j]) / 15; // backwards compatibility
} else if (!strcmp(argv[j],"--net-ro-interval") && more) {
Modes.net_output_flush_interval = (uint64_t)(1000 * atof(argv[++j]));
} else if (!strcmp(argv[j],"--net-ro-port") && more) {
Modes.net = 1;
free(Modes.net_output_raw_ports);
Modes.net_output_raw_ports = strdup(argv[++j]);
} else if (!strcmp(argv[j],"--net-ri-port") && more) {
Modes.net = 1;
free(Modes.net_input_raw_ports);
Modes.net_input_raw_ports = strdup(argv[++j]);
} else if (!strcmp(argv[j],"--net-bo-port") && more) {
Modes.net = 1;
free(Modes.net_output_beast_ports);
Modes.net_output_beast_ports = strdup(argv[++j]);
} else if (!strcmp(argv[j],"--net-bi-port") && more) {
Modes.net = 1;
free(Modes.net_input_beast_ports);
Modes.net_input_beast_ports = strdup(argv[++j]);
} else if (!strcmp(argv[j],"--net-bind-address") && more) {
free(Modes.net_bind_address);
Modes.net_bind_address = strdup(argv[++j]);
} else if (!strcmp(argv[j],"--net-http-port") && more) {
if (strcmp(argv[++j], "0")) {
fprintf(stderr, "warning: --net-http-port not supported in this build, option ignored.\n");
}
2013-01-17 18:12:23 +00:00
} else if (!strcmp(argv[j],"--net-sbs-port") && more) {
Modes.net = 1;
free(Modes.net_output_sbs_ports);
Modes.net_output_sbs_ports = strdup(argv[++j]);
2020-03-23 22:53:24 +00:00
} else if (!strcmp(argv[j],"--net-stratux-port") && more) {
Modes.net = 1;
free(Modes.net_output_stratux_ports);
2020-03-23 22:53:24 +00:00
Modes.net_output_stratux_ports = strdup(argv[++j]);
} else if (!strcmp(argv[j],"--net-buffer") && more) {
Modes.net_sndbuf_size = atoi(argv[++j]);
} else if (!strcmp(argv[j],"--net-verbatim")) {
Modes.net_verbatim = 1;
} else if (!strcmp(argv[j],"--forward-mlat")) {
Modes.forward_mlat = 1;
} else if (!strcmp(argv[j],"--onlyaddr")) {
Modes.onlyaddr = 1;
} else if (!strcmp(argv[j],"--metric")) {
Modes.metric = 1;
} else if (!strcmp(argv[j],"--hae") || !strcmp(argv[j],"--gnss")) {
Modes.use_gnss = 1;
2013-01-15 00:41:10 +00:00
} else if (!strcmp(argv[j],"--aggressive")) {
fprintf(stderr, "warning: --aggressive not supported in this build, option ignored (consider '--fix-2bit' instead)\n");
} else if (!strcmp(argv[j],"--interactive")) {
Modes.interactive = 1;
} else if (!strcmp(argv[j],"--interactive-ttl") && more) {
Modes.interactive_display_ttl = (uint64_t)(1000 * atof(argv[++j]));
Interactive mode updates These updates were designed to assist those using interactive mode to tune antennas and SDR gain. * Add options to display distance and bearing in interactive mode Distance and bearing instead of latitude and longitude can now be displayed in interactive mode using the following options to dump1090-fa and view1090-fa. --interactive-show-distance Show aircraft distance and bearing instead of aircraft lat/lon (requires --lat and --lon) --interactive-distance-units Distance units ('km', 'sm', 'nm') (default: 'nm')" You have to specify a reference --lat and --lon for this to work of course. * A new line now shows at the top of the interactive display that has for the current sample: Total valid aircraft count Vidible aircraft count Will be less than total if the screen hasn't enough lines to show them all. Max RSSI Min RSSI Mean RSSI Max Distance Tot: 47 Vis: 47 RSSI: Max 25.4+ Mean -29.5 Min -36.9- MaxD: 197.3nm+ * Add max distance and min/max RSSI indicators A '+' after the distance in a row indicates it's the row with the maximum distance. A '+' after the RSSI in a row indicates it's the row with the highest RSSI. A '-' after the RSSI in a row indicates it's the row with the lowest RSSI. The summary line at the top of the screen always shows the values for ALL aircraft, even those not visible. The row indicators only mark visible rows though. In this example, the first aircraft is both the farthest away and has the weakest RSSI. The second aircraft has the strongest RSSI. Tot: 47 Vis: 47 RSSI: Max 25.4+ Mean -29.5 Min -36.9- MaxD: 197.3nm+ - Hex Mode Sqwk Flight Alt Spd Hdg Dist(nm) Bearing RSSI Msgs Ti ──────────────────────────────────────────────────────────────────────────────── A8D5A4 S2 34000 438 252 197.3+ 85 -36.1- 26 2 A39A13 S2ac 5740 FFT525 30750 439 256 98.8 68 -25.4+ 123 0 A70B23 S2ac 2744 LXJ553 43000 419 258 136.1 39 -33.6 174 0 * Finally, a new option '--interactive-callsign-filter' has been added to allow filtering interactive by callsign. The value can be a simple string, in which case aircraft with that string anywhere in its callsign will be displayed, or a regular expression should you want a more precise match. Examples: --interactive-callsign-filter UAL will match all aircraft with UAL anywhere in its callsign. --interactive-callsign-filter "^UAL" will match only those callsigns that start with UAL. --interactive-callsign-filter "^(UAL|AAL)" will match only those callsigns that start with UAL or AAL. --interactive-callsign-filter "^N[0-9]" or "^N[[:digit:]]" will match only those callsigns that start with "N" and a number If you need more info on regular expressions, see this link: https://en.wikipedia.org/wiki/Regular_expression
2020-09-21 22:00:44 +00:00
} else if (!strcmp(argv[j],"--interactive-show-distance")) {
Modes.interactive_show_distance = 1;
} else if (!strcmp(argv[j], "--interactive-distance-units") && more) {
char *units = argv[++j];
if (!strcmp(units, "km")) {
Modes.interactive_distance_units = UNIT_KILOMETERS;
} else if (!strcmp(units, "sm")) {
Modes.interactive_distance_units = UNIT_STATUTE_MILES;
} else {
Modes.interactive_distance_units = UNIT_NAUTICAL_MILES;
}
} else if (!strcmp(argv[j], "--interactive-callsign-filter") && more) {
Modes.interactive_callsign_filter = strdup(argv[++j]);
} else if (!strcmp(argv[j], "--lat") && more) {
Modes.fUserLat = atof(argv[++j]);
} else if (!strcmp(argv[j],"--lon") && more) {
Modes.fUserLon = atof(argv[++j]);
} else if (!strcmp(argv[j],"--max-range") && more) {
Modes.maxRange = atof(argv[++j]) * 1852.0; // convert to metres
} else if (!strcmp(argv[j],"--debug") && more) {
fprintf(stderr, "warning: --debug is obsolete and ignored\n");
++j;
} else if (!strcmp(argv[j],"--stats")) {
if (!Modes.stats)
Modes.stats = (uint64_t)1 << 60; // "never"
2015-06-19 16:29:14 +00:00
} else if (!strcmp(argv[j],"--stats-range")) {
Modes.stats_range_histo = 1;
} else if (!strcmp(argv[j],"--stats-every") && more) {
Modes.stats = (uint64_t) (1000 * atof(argv[++j]));
} else if (!strcmp(argv[j],"--json-stats-every") && more) {
Modes.json_stats_interval = (uint64_t) (1000 * atof(argv[++j]));
} else if (!strcmp(argv[j],"--snip") && more) {
snipMode(atoi(argv[++j]));
exit(0);
} else if (!strcmp(argv[j],"--help")) {
showHelp();
exit(0);
} else if (!strcmp(argv[j],"--version")) {
showVersion();
Move all converters to starch-based implementations (#97) * Switch all conversion routines to use starch. main user-visible changes: * ensure you check out submodules ('git clone --recurse-submodules") * --version shows the CPU features and DSP implementations in use * --wisdom allows overriding of the built-in architecture wisdom * --dcfilter no longer supported * "starch-benchmark" binary will benchmark all options on the current machine and can produce a wisdom file to feed to the --wisdom option If you have a usecase for --dcfilter, please get in touch and let me know - it's an edge case and for now there's no starch/DSP support for it, but support can be written if needed. In almost all cases the new conversion routines are slightly or substantially faster than the old conversion routines. The only case that is slower is SC16/SC16Q11 on a Pi 0, which is around 10% slower due to changing from heavily approximated lookup tables to higher quality results (but SC16 is probably already out of reach of a Pi 0) * No need to build with SC16Q11_TABLE_BITS any more * Add oneoff/uc8_capture_stats (reads a UC8 capture; measures min/max/mean I and Q) * Switch UC8 conversion to 127.4 center, 128 range. Looking at actual UC8 captures from a RTL2832, the mean I and Q are actually at 127.4, so use that as the zero point. This means that the resulting I/Q maximum values could be as large as 127.6. Switch to 128 for simplicity. * Switch to the new UC8 zero offset in benchmarks, fix some bugs * Fix some bugs in SC16/SC16Q11 validation, tighten the max error requirements * Ditch UC8 approximation path, add a NEON VRQSQRTE path. * Tweak the SC16 exact path, add a new impl that uses a mix of u32 & floats. * SC16Q11 impl tweaks: * add a u32->float exact path * ditch the approximation path * add a NEON VRSQRTE path * add a 12-bit table path (using the full signed I/Q value, not absolute value) * Ditch SC16 approximation path, add NEON vrsqrte path * Add oneoff/dsp_error_measurement This runs sample input through the DSP functions that are allowed to be inexact and dumps the results as a TSV suitable for feeding to gnuplot to look at the actual errors. * Update make clean, make wisdom targets * Update wisdom based on benchmarking * Preserve the raw wisdom benchmark data * Update to latest starch * Update .gitignore for new wisdom files * Update starch generated code * Build starch-benchmark as part of the 'all' target * Use wisdom from /etc/dump1090-fa/wisdom.local if present * Package starch-benchmark and a helper script to generate local wisdom data * Remove submodules in preparation for importing them directly * Import cpu_features v0.6.0 from https://github.com/google/cpu_features/releases/tag/v0.6.0 * Import starch at commit a725c8491dc33a321565d451b385131e589d8490 from https://github.com/flightaware/starch
2021-01-21 11:45:00 +00:00
showDSP();
exit(0);
} else if (!strcmp(argv[j],"--quiet")) {
Modes.quiet = 1;
} else if (!strcmp(argv[j],"--show-only") && more) {
Modes.show_only = (uint32_t) strtoul(argv[++j], NULL, 16);
} else if (!strcmp(argv[j],"--mlat")) {
2013-04-13 00:37:02 +00:00
Modes.mlat = 1;
} else if (!strcmp(argv[j],"--oversample")) {
// Ignored
2014-11-23 16:02:11 +00:00
} else if (!strcmp(argv[j], "--write-json") && more) {
Modes.json_dir = strdup(argv[++j]);
2014-11-23 16:02:11 +00:00
} else if (!strcmp(argv[j], "--write-json-every") && more) {
Modes.json_interval = (uint64_t)(1000 * atof(argv[++j]));
if (Modes.json_interval < 100) // 0.1s
Modes.json_interval = 100;
} else if (!strcmp(argv[j], "--json-location-accuracy") && more) {
Modes.json_location_accuracy = atoi(argv[++j]);
Move all converters to starch-based implementations (#97) * Switch all conversion routines to use starch. main user-visible changes: * ensure you check out submodules ('git clone --recurse-submodules") * --version shows the CPU features and DSP implementations in use * --wisdom allows overriding of the built-in architecture wisdom * --dcfilter no longer supported * "starch-benchmark" binary will benchmark all options on the current machine and can produce a wisdom file to feed to the --wisdom option If you have a usecase for --dcfilter, please get in touch and let me know - it's an edge case and for now there's no starch/DSP support for it, but support can be written if needed. In almost all cases the new conversion routines are slightly or substantially faster than the old conversion routines. The only case that is slower is SC16/SC16Q11 on a Pi 0, which is around 10% slower due to changing from heavily approximated lookup tables to higher quality results (but SC16 is probably already out of reach of a Pi 0) * No need to build with SC16Q11_TABLE_BITS any more * Add oneoff/uc8_capture_stats (reads a UC8 capture; measures min/max/mean I and Q) * Switch UC8 conversion to 127.4 center, 128 range. Looking at actual UC8 captures from a RTL2832, the mean I and Q are actually at 127.4, so use that as the zero point. This means that the resulting I/Q maximum values could be as large as 127.6. Switch to 128 for simplicity. * Switch to the new UC8 zero offset in benchmarks, fix some bugs * Fix some bugs in SC16/SC16Q11 validation, tighten the max error requirements * Ditch UC8 approximation path, add a NEON VRQSQRTE path. * Tweak the SC16 exact path, add a new impl that uses a mix of u32 & floats. * SC16Q11 impl tweaks: * add a u32->float exact path * ditch the approximation path * add a NEON VRSQRTE path * add a 12-bit table path (using the full signed I/Q value, not absolute value) * Ditch SC16 approximation path, add NEON vrsqrte path * Add oneoff/dsp_error_measurement This runs sample input through the DSP functions that are allowed to be inexact and dumps the results as a TSV suitable for feeding to gnuplot to look at the actual errors. * Update make clean, make wisdom targets * Update wisdom based on benchmarking * Preserve the raw wisdom benchmark data * Update to latest starch * Update .gitignore for new wisdom files * Update starch generated code * Build starch-benchmark as part of the 'all' target * Use wisdom from /etc/dump1090-fa/wisdom.local if present * Package starch-benchmark and a helper script to generate local wisdom data * Remove submodules in preparation for importing them directly * Import cpu_features v0.6.0 from https://github.com/google/cpu_features/releases/tag/v0.6.0 * Import starch at commit a725c8491dc33a321565d451b385131e589d8490 from https://github.com/flightaware/starch
2021-01-21 11:45:00 +00:00
} else if (!strcmp(argv[j], "--wisdom") && more) {
if (starch_read_wisdom (argv[++j]) < 0) {
fprintf(stderr,
"Failed to read wisdom file %s: %s\n", argv[j], strerror(errno));
exit(1);
}
} else if (!strcmp(argv[j], "--adaptive-min-gain") && more) {
Modes.adaptive_min_gain_db = atof(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-max-gain") && more) {
Modes.adaptive_max_gain_db = atof(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-duty-cycle") && more) {
Modes.adaptive_duty_cycle = atof(argv[++j]) / 100.0;
} else if (!strcmp(argv[j], "--adaptive-burst")) {
Modes.adaptive_burst_control = true;
} else if (!strcmp(argv[j], "--adaptive-burst-alpha") && more) {
Modes.adaptive_burst_alpha = atof(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-burst-change-delay") && more) {
Modes.adaptive_burst_change_delay = atoi(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-burst-loud-rate") && more) {
Modes.adaptive_burst_loud_rate = atof(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-burst-loud-runlength") && more) {
Modes.adaptive_burst_loud_runlength = atoi(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-burst-quiet-rate") && more) {
Modes.adaptive_burst_quiet_rate = atof(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-burst-quiet-runlength") && more) {
Modes.adaptive_burst_quiet_runlength = atoi(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-range")) {
Modes.adaptive_range_control = true;
} else if (!strcmp(argv[j], "--adaptive-range-alpha") && more) {
Modes.adaptive_range_alpha = atof(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-range-percentile") && more) {
Modes.adaptive_range_percentile = atoi(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-range-target") && more) {
Modes.adaptive_range_target = atof(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-range-change-delay") && more) {
Modes.adaptive_range_change_delay = atoi(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-range-scan-delay") && more) {
Modes.adaptive_range_scan_delay = atoi(argv[++j]);
} else if (!strcmp(argv[j], "--adaptive-range-rescan-delay") && more) {
Modes.adaptive_range_rescan_delay = atoi(argv[++j]);
} else if (sdrHandleOption(argc, argv, &j)) {
/* handled */
} else {
fprintf(stderr,
"Unknown or not enough arguments for option '%s'.\nTry %s --help for full option help.\n",
argv[j],
argv[0]);
exit(1);
}
}
if (Modes.sdr_type == SDR_NONE && !Modes.net) {
fprintf(stderr,
"No SDR available and network mode not enabled; nothing to do!\n"
"Select a SDR type (--device-type or --ifile), or enable network mode (--net)\n"
"Try %s --help for full option help.\n",
argv[0]);
exit(1);
}
if (Modes.nfix_crc > MODES_MAX_BITERRORS)
Modes.nfix_crc = MODES_MAX_BITERRORS;
// Initialization
log_with_timestamp("%s %s starting up.", MODES_DUMP1090_VARIANT, MODES_DUMP1090_VERSION);
modesInit();
if (!sdrOpen()) {
exit(1);
}
if (Modes.net) {
modesInitNet();
}
// init stats:
Modes.stats_current.start = Modes.stats_current.end =
Modes.stats_alltime.start = Modes.stats_alltime.end =
Modes.stats_periodic.start = Modes.stats_periodic.end =
Modes.stats_latest.start = Modes.stats_latest.end =
Modes.stats_5min.start = Modes.stats_5min.end =
Modes.stats_15min.start = Modes.stats_15min.end = mstime();
for (j = 0; j < 15; ++j)
Modes.stats_1min[j].start = Modes.stats_1min[j].end = Modes.stats_current.start;
adaptive_init();
// write initial json files so they're not missing
writeJsonToFile("receiver.json", generateReceiverJson);
writeJsonToFile("stats.json", generateStatsJson);
writeJsonToFile("aircraft.json", generateAircraftJson);
interactiveInit();
// If the user specifies --net-only, just run in order to serve network
// clients without reading data from the RTL device
if (Modes.sdr_type == SDR_NONE) {
while (!Modes.exit) {
struct timespec start_time;
struct timespec slp = { 0, 100 * 1000 * 1000};
start_cpu_timing(&start_time);
backgroundTasks();
end_cpu_timing(&start_time, &Modes.stats_current.background_cpu);
nanosleep(&slp, NULL);
}
} else {
Don't hang on exit if rtlsdr hardware stops sending samples. Give up and exit after 30 seconds of no sample data, rather than just warning and continuing. background & discussion: https://discussions.flightaware.com/t/cpu-hikes-crash-dump1090-fa/74759 The scenario this addresses is: * Hardware wedges, USB bulk endpoint stops providing data * librtlsdr remains in rtlsdr_read_async() waiting for either USB data which never arrives, or a cancellation via _a different thread_ calling rtlsdr_cancel_async(). * main thread notices the lack of SDR data and complains * something external e.g. piaware tries a restart and sends SIGTERM * the signal handler sets Modes.exit = 1; the main thread starts waiting for receive thread termination * because we're never getting callbacks from rtlsdr_read_async(), we never call rtlsdr_cancel_async * dump1090 hangs waiting on receive thread termination To fix this, add a sdrStop() handler function where the general contract is "make the receive thread terminate". In the rtlsdr case, this calls rtlsdr_cancel_async directly, which will make rtlsdr_read_async() return even if the hardware is stuck. The main thread then calls sdrStop() before waiting for receive thread termination. Also, as discussed in the thread above, there's not really much point in continuing to run if the SDR has wedged, so bail out after 30 seconds of no sample data. Also, if pthread_timedjoin_np is available, use it in preference to pthread_join so that we do not wait indefinitely for the receive thread on shutdown. If the join times out, give up and abort() as we can't safely continue a clean shutdown while the receive thread is running.
2021-03-20 17:51:08 +00:00
int watchdogCounter = 300; // about 30 seconds
// Create the thread that will read the data from the device.
pthread_create(&Modes.reader_thread, NULL, readerThreadEntryPoint, NULL);
while (!Modes.exit) {
// get the next sample buffer off the FIFO; wait only up to 100ms
// this is fairly aggressive as all our network I/O runs out of the background work!
struct mag_buf *buf = fifo_dequeue(100 /* milliseconds */);
struct timespec start_time;
if (buf) {
// Process one buffer
start_cpu_timing(&start_time);
demodulate2400(buf);
if (Modes.mode_ac) {
demodulate2400AC(buf);
}
Modes.stats_current.samples_processed += buf->validLength - buf->overlap;
Modes.stats_current.samples_dropped += buf->dropped;
end_cpu_timing(&start_time, &Modes.stats_current.demod_cpu);
// Return the buffer to the FIFO freelist for reuse
fifo_release(buf);
// We got something so reset the watchdog
Don't hang on exit if rtlsdr hardware stops sending samples. Give up and exit after 30 seconds of no sample data, rather than just warning and continuing. background & discussion: https://discussions.flightaware.com/t/cpu-hikes-crash-dump1090-fa/74759 The scenario this addresses is: * Hardware wedges, USB bulk endpoint stops providing data * librtlsdr remains in rtlsdr_read_async() waiting for either USB data which never arrives, or a cancellation via _a different thread_ calling rtlsdr_cancel_async(). * main thread notices the lack of SDR data and complains * something external e.g. piaware tries a restart and sends SIGTERM * the signal handler sets Modes.exit = 1; the main thread starts waiting for receive thread termination * because we're never getting callbacks from rtlsdr_read_async(), we never call rtlsdr_cancel_async * dump1090 hangs waiting on receive thread termination To fix this, add a sdrStop() handler function where the general contract is "make the receive thread terminate". In the rtlsdr case, this calls rtlsdr_cancel_async directly, which will make rtlsdr_read_async() return even if the hardware is stuck. The main thread then calls sdrStop() before waiting for receive thread termination. Also, as discussed in the thread above, there's not really much point in continuing to run if the SDR has wedged, so bail out after 30 seconds of no sample data. Also, if pthread_timedjoin_np is available, use it in preference to pthread_join so that we do not wait indefinitely for the receive thread on shutdown. If the join times out, give up and abort() as we can't safely continue a clean shutdown while the receive thread is running.
2021-03-20 17:51:08 +00:00
watchdogCounter = 300;
} else {
// Nothing to process this time around.
if (--watchdogCounter <= 0) {
Don't hang on exit if rtlsdr hardware stops sending samples. Give up and exit after 30 seconds of no sample data, rather than just warning and continuing. background & discussion: https://discussions.flightaware.com/t/cpu-hikes-crash-dump1090-fa/74759 The scenario this addresses is: * Hardware wedges, USB bulk endpoint stops providing data * librtlsdr remains in rtlsdr_read_async() waiting for either USB data which never arrives, or a cancellation via _a different thread_ calling rtlsdr_cancel_async(). * main thread notices the lack of SDR data and complains * something external e.g. piaware tries a restart and sends SIGTERM * the signal handler sets Modes.exit = 1; the main thread starts waiting for receive thread termination * because we're never getting callbacks from rtlsdr_read_async(), we never call rtlsdr_cancel_async * dump1090 hangs waiting on receive thread termination To fix this, add a sdrStop() handler function where the general contract is "make the receive thread terminate". In the rtlsdr case, this calls rtlsdr_cancel_async directly, which will make rtlsdr_read_async() return even if the hardware is stuck. The main thread then calls sdrStop() before waiting for receive thread termination. Also, as discussed in the thread above, there's not really much point in continuing to run if the SDR has wedged, so bail out after 30 seconds of no sample data. Also, if pthread_timedjoin_np is available, use it in preference to pthread_join so that we do not wait indefinitely for the receive thread on shutdown. If the join times out, give up and abort() as we can't safely continue a clean shutdown while the receive thread is running.
2021-03-20 17:51:08 +00:00
log_with_timestamp("No samples received from the SDR for a long time. Maybe the hardware is wedged? Giving up.");
Modes.exit = 2; // abnormal exit
}
}
start_cpu_timing(&start_time);
backgroundTasks();
end_cpu_timing(&start_time, &Modes.stats_current.background_cpu);
}
log_with_timestamp("Waiting for receive thread termination");
Don't hang on exit if rtlsdr hardware stops sending samples. Give up and exit after 30 seconds of no sample data, rather than just warning and continuing. background & discussion: https://discussions.flightaware.com/t/cpu-hikes-crash-dump1090-fa/74759 The scenario this addresses is: * Hardware wedges, USB bulk endpoint stops providing data * librtlsdr remains in rtlsdr_read_async() waiting for either USB data which never arrives, or a cancellation via _a different thread_ calling rtlsdr_cancel_async(). * main thread notices the lack of SDR data and complains * something external e.g. piaware tries a restart and sends SIGTERM * the signal handler sets Modes.exit = 1; the main thread starts waiting for receive thread termination * because we're never getting callbacks from rtlsdr_read_async(), we never call rtlsdr_cancel_async * dump1090 hangs waiting on receive thread termination To fix this, add a sdrStop() handler function where the general contract is "make the receive thread terminate". In the rtlsdr case, this calls rtlsdr_cancel_async directly, which will make rtlsdr_read_async() return even if the hardware is stuck. The main thread then calls sdrStop() before waiting for receive thread termination. Also, as discussed in the thread above, there's not really much point in continuing to run if the SDR has wedged, so bail out after 30 seconds of no sample data. Also, if pthread_timedjoin_np is available, use it in preference to pthread_join so that we do not wait indefinitely for the receive thread on shutdown. If the join times out, give up and abort() as we can't safely continue a clean shutdown while the receive thread is running.
2021-03-20 17:51:08 +00:00
sdrStop(); // tell reader thread to wake up and exit
fifo_halt(); // Reader thread should do this anyway, but just in case..
Don't hang on exit if rtlsdr hardware stops sending samples. Give up and exit after 30 seconds of no sample data, rather than just warning and continuing. background & discussion: https://discussions.flightaware.com/t/cpu-hikes-crash-dump1090-fa/74759 The scenario this addresses is: * Hardware wedges, USB bulk endpoint stops providing data * librtlsdr remains in rtlsdr_read_async() waiting for either USB data which never arrives, or a cancellation via _a different thread_ calling rtlsdr_cancel_async(). * main thread notices the lack of SDR data and complains * something external e.g. piaware tries a restart and sends SIGTERM * the signal handler sets Modes.exit = 1; the main thread starts waiting for receive thread termination * because we're never getting callbacks from rtlsdr_read_async(), we never call rtlsdr_cancel_async * dump1090 hangs waiting on receive thread termination To fix this, add a sdrStop() handler function where the general contract is "make the receive thread terminate". In the rtlsdr case, this calls rtlsdr_cancel_async directly, which will make rtlsdr_read_async() return even if the hardware is stuck. The main thread then calls sdrStop() before waiting for receive thread termination. Also, as discussed in the thread above, there's not really much point in continuing to run if the SDR has wedged, so bail out after 30 seconds of no sample data. Also, if pthread_timedjoin_np is available, use it in preference to pthread_join so that we do not wait indefinitely for the receive thread on shutdown. If the join times out, give up and abort() as we can't safely continue a clean shutdown while the receive thread is running.
2021-03-20 17:51:08 +00:00
// Wait on reader thread exit
if (join_thread(Modes.reader_thread, NULL, 30000) == ETIMEDOUT) {
log_with_timestamp("Receive thread did not shut down cleanly in 30 seconds, aborting.");
abort(); // Can't complete cleanup while the receive thread is active; bail out.
}
}
interactiveCleanup();
2021-02-04 08:36:13 +00:00
// Write final stats
flush_stats(0);
writeJsonToFile("stats.json", generateStatsJson);
if (Modes.stats) {
display_stats(&Modes.stats_alltime);
}
sdrClose();
fifo_destroy();
if (Modes.exit == 1) {
log_with_timestamp("Normal exit.");
return 0;
} else {
log_with_timestamp("Abnormal exit.");
return 1;
}
}
//
//=========================================================================
//