summaryrefslogtreecommitdiff
path: root/gpsd.c
Commit message (Collapse)AuthorAgeFilesLines
* If the profiling code isn't compiled in, timing policy bit must be forced off.Eric S. Raymond2011-09-231-0/+3
|
* Re-enable the & and ! control-socket commands.Eric S. Raymond2011-09-211-2/+0
|
* Remove the old latency-profiling machinery.Eric S. Raymond2011-09-201-18/+1
| | | | | | It hasn't actually worked since we switched protocols to JSON. This diff puts the framework in place for new machinery using timing attributes in the TPV sentence.
* -G option now works under IPV6.Eric S. Raymond2011-09-191-9/+13
| | | | | | | | See http://stackoverflow.com/questions/7474066/how-to-listen-on-all-ipv6-addresses-using-c-sockets-api for discussion; thanks to Dietrich Epps for a correct answer.
* Revert "Try once again to make -G work with IPV6."Eric S. Raymond2011-09-181-3/+10
| | | | | That change led to a spurious message about the listen address already being in use.
* Add the force_global configure option (always operate in -G mode).Eric S. Raymond2011-09-161-8/+18
|
* Try once again to make -G work with IPV6.Eric S. Raymond2011-09-161-10/+3
| | | | | | There was a comment that claimed assigning sat.sa_in6.sin6_addr = in6addr_an wouldn't work, but that's exactly what's done in examples such as this one: http://docsrv.sco.com/SDK_netapi/sockC.TheIPv6sockaddrstructure.html
* Define default gpsd_user name in SConstruct.Michael Tatarinov2011-09-151-5/+0
| | | | Signed-off-by: Eric S. Raymond <esr@thyrsus.com>
* Prevent use of prototype device when GPSD_GROUP is defined.Eric S. Raymond2011-09-141-6/+7
|
* No need to use setlocale(NUMERIC) any more, so lose the code overhead.Eric S. Raymond2011-08-271-2/+0
| | | | All regression tests pass.
* Fix the DBUS build.Eric S. Raymond2011-08-251-4/+0
|
* Fix problems with reconfigure=no and netfeed=no builds.Eric S. Raymond2011-08-241-1/+1
|
* The -N option no longer drops privileges.Eric S. Raymond2011-08-191-15/+29
| | | | | | | | | | The price is that only devices specified on the gpsd commmand line will do PPS, because the calls to set line discipline stop being available after we drop privileges. In practice, this shouldn't be a problem, as PPS devices ar RS232 ports that can't be hotplugged and thus *must* be specified on the command line. All regression tests pass.
* Typo fix.Eric S. Raymond2011-08-111-1/+1
|
* Improved explanation of the priv-dropping problem. All regression tests pass.Eric S. Raymond2011-08-111-4/+4
|
* Revert to old -N behavior of dripping privileges, and explain why.Eric S. Raymond2011-08-111-2/+10
|
* Drop privileges in foreground as well as background.Eric S. Raymond2011-08-111-1/+2
| | | | All regression tests pass.
* Consider a device subscribed inly if the subscriber has requested WATCH on it.Gleb Smirnoff2011-07-041-1/+1
| | | | | | | | This fixes a bug introduced by glebius's earlier patch. Actually this bug might have been causing subtle problems since 2.95. All regression tests pass. Signed-off-by: Eric S. Raymond <esr@thyrsus.com>
* Compiler warning suppression.Eric S. Raymond2011-07-011-3/+0
|
* Comment out some code that is unused after the gpsmon featurectomy.Eric S. Raymond2011-06-271-4/+7
|
* Comment typo fixes.Eric S. Raymond2011-06-261-2/+2
|
* Correct some documentation.Eric S. Raymond2011-06-261-3/+4
|
* Document some obscurities. All regression tests pass.Eric S. Raymond2011-06-251-0/+11
|
* Minor cleanup on Gleb Smirnoff's patch. All regression tests pass.Eric S. Raymond2011-06-231-3/+2
|
* Fix a problem when a device has disconnected and then reappeared. Gleb Smirnoff2011-06-231-36/+49
| | | | | | | | | | | | | | | | Before this patch gpsd would ignore a re-connected device until a new client connects and subscribes for it. To do that, utilize already exisiting cycle in the main daemon loop, that goes through all devices and through all subscribers to determine whether device is needed. Before this change the cycle did some treatment on connected devices that aren't needed, now it also polls disconnected devices that _are_ needed. I've utilized device->opentime to keep polling interval. This field isn't used by libgpsd_core while device is disconnected. Signed-off-by: Eric S. Raymond <esr@thyrsus.com>
* Tiny formatting nit.Gleb Smirnoff2011-06-221-1/+1
| | | | Signed-off-by: Eric S. Raymond <esr@thyrsus.com>
* Use macro instead of hand-rolled version.Gleb Smirnoff2011-06-221-3/+1
| | | | Signed-off-by: Eric S. Raymond <esr@thyrsus.com>
* Path-rewrite for JSON packets works. All regression tests pass.Eric S. Raymond2011-06-211-3/+17
|
* Elimination of gpsd_hexdump_level.Eric S. Raymond2011-06-171-2/+0
| | | | The few gpsd_hexdump() calls left have explicit guards.
* Make the debug level part of the session context.Eric S. Raymond2011-06-171-10/+9
| | | | | Probably should have been this way all along. Now it will be directy accessible from the drivers.
* First xut at JSON passthrough code.Eric S. Raymond2011-06-141-0/+10
| | | | Not turned in by default yet; it interferes with RTCM2 decoding.
* Yet more code slimming. All regression tests pass, code splints clean.Eric S. Raymond2011-06-111-2/+21
|
* More code slimming.Eric S. Raymond2011-06-111-2/+18
|
* Code slimming.Eric S. Raymond2011-06-111-0/+2
|
* Mac OS X systemd activation for gpsd.Eckhart Wörner2011-04-291-1/+31
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The attached patch allows to use gpsd together with socket activation in systemd. The idea is that systemd listens to the control socket and TCP sockets and, as soon as someone connects to them, starts gpsd and passes the file descriptors over (without accept()ing itself). Sockets are in the following order: fd 3 is the control socket, followed by up to two TCP sockets (for IPv4 and IPv6). If no socket passing happens, behaviour of gpsd should be the same as before. For using this new feature, one could use the following config *skeleton*: -> File /etc/systemd/system/gpsd.socket: [Socket] ListenStream=/var/run/gpsd.sock ListenStream=127.0.0.1:2947 -> File /etc/systemd/system/gpsd.service: [Unit] Requires=gpsd.spcket [Service] type=simple ExecStart=/usr/bin/gpsd -N What has been tested: - gpsd compiles fine with both systemd=true/false - Socket activation on the TCP socket works (gpsd is started, version string is sent) What has not been tested: - Socket activation on the control socket works Signed-off-by: Eric S. Raymond <esr@thyrsus.com>
* Revert a mistake. Noticed by mlichvar, thanks.Eric S. Raymond2011-04-141-1/+1
|
* splint cleanup.Eric S. Raymond2011-04-061-1/+3
|
* With -pthread, the compiler gets pickier about having the header included.Eric S. Raymond2011-04-051-0/+1
|
* finish renaming gpsd_dbus to dbusexportJon Schlueter2011-04-041-1/+1
| | | | | | renamed gpsd_dbus.h to dbusexport.h renamed internals of doc/* to point at dbusexport instead of gpsd_dbus updated gpsd.c and dbusexport.c to use dbusexport.h
* Give up on trying to splint the DBUS code. The headers confuse splint.Eric S. Raymond2011-04-031-8/+8
| | | | | | Also, splint production for scons. All regression tests pass.
* Regularize some names.Eric S. Raymond2011-04-011-7/+7
|
* gps_maskdump() has been a CPU hog. Drastically reduce the number of calls...Eric S. Raymond2011-03-311-11/+12
| | | | | ...and guard them with debug level so they won't be done when the output would never be shipped anywhere.
* Prevent a compiler warning.Eric S. Raymond2011-03-311-1/+2
|
* Eliminate nowait global when FORCE_NOWAIT.Eric S. Raymond2011-03-291-4/+6
|
* Another step towards complete conditionalizing on FORCE_NOWAIT.Eric S. Raymond2011-03-291-5/+7
|
* Partially revert the previous FORCE_NOWAIT change, as it broke the -n option.Eric S. Raymond2011-03-291-12/+7
|
* Squeeze more code out of the minimal build.Eric S. Raymond2011-03-291-18/+29
|
* More conditioning out of code for minimal build.Eric S. Raymond2011-03-291-5/+9
|
* Add a switch to configure out control-socket switch.Eric S. Raymond2011-03-281-6/+30
|
* Scan-build and splint cleanup. All regression tests pass.Eric S. Raymond2011-03-281-4/+4
|
View Lorry Controller Status