There is a race condition for the accel and mag polling rates.
Whichever one gets set first, the other will be uninitialized.
Set the mag polling rate to 1ms if uninitilized.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The workqueues measure time in ticks which is typically 10ms.
Some interrupt events in Nuttx occur at about 1ms so a more
granular workqueue is needed for POSIX.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The use of a non-static function without a declaration generates a
warning for gcc 4.8.
Clang-3.4 does not work when compiling for gprof. The executable always
generates a segv.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Gyrosim would call measure continuously because the write_checked_reg
was failing. There is no need to check faked reg writes in the
simulator so that code was removed.
The delay that was added to the simulator to pace the gyrosim reads
was also removed now that the source of the problem was determined.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The existing orb_advert_t use thoughout the code sometimes tries
to treat it as a file descriptor and there are checks for < 0
and ::close calls on orb_advert_t types which is an invalid use
of an object pointer, which is what orb_advert_t really is.
Initially I had changed the -1 initializations to 0 but it was
suggested that this should be nullptr. That was a good recommendation
but the definition of orb_advert_t had to change to void * because
you cannot initialize a uintptr_t as nullptr.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The calls to orb_advertise were being mishandled throughout the code.
There were ::close() calls on memory pointers, there were checks
against < 0 when it is a pointer to a object and values larger than
0x7ffffffff are valid. Some places orb_advert_t variables were
being initialized as 0 other places as -1.
The orb_advert_t type was changed to uintptr_t so the pointer value
would not be wrapped as a negative number. This was causing a failure
on ARM.
Tests for < 0 were changed to == 0 since a null pointer is the valid
representation for error, or uninitialized.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
There were an insufficient number of devmap entries allocated and
when they ran out, new orb_advertise requests would fail.
Also added a new logging macro for the Linux build to show the
calling pthread if enabled.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The warnx and warn calls map to PX4_WARN.
Calls to errx or err genrtate a compile error.
The px4_log.h file implements a new log format:
For DEBUG and INFO:
<level> <msg>
For ERROR and WARN:
<level> <msg> (file filepath line linenum)
The verbosity can be changed by setting the macro to use
either linux_log or linux_log_verbose in px4_log.h
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The transfer function would previously return error if the receive
buffer length was 0. This appears to be a valid condition for
requesting a measurmement be taken but no data returned.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Reworking toolchain and main.cpp for QuRT to a final link can be
done and the apps.h file is autogenerated.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
GCC was more picky about prototypes for inlines being required.
The generate_listener.py script used incorrect printf formats and
was casting %f params to float, but printf casts all %f params to
double per the spec.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The initialization functions were called after the script
commands were run causing a deadlock waiting for an
uninitialized semaphore.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The HRT call processing normally happens via HW timer interrupt
handler. Since the POSIX port has no ISR handling, the HP work
queue is used.
Instead of irq_save() and irq_restore() calls to disable/enable
interrupts, a mutex is used to protect each queue.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
In STM32, the ISR handler calls hrt_call_invoke(). There is no
interrupt timer inthe POSIX port so a work item is put on the
high priority work queue to expire at the next event (in ticks)
or at the next max delay interval.
Counter wrapping is likely still not handled properly in this code.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The commented out lock and unlock were determined to be needed and added back.
The unit test for VDev was updated. It showed the race between the poll and a
write that only does a poll_notify().
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Fixed print format for __LINE__ to %d
Fixed if/else that breaks with the debug macro expansion. The if/else
needs to use braces to allow macro expansion.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Created px4_debug,h to define:
PX4_DBG
PX4_INFO
PX4_WARN
PX4_ERR
These enable OS specific mappings to be made, filtering, etc.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Added simulated tone_alarm class and enabled led class for posix build.
The simulator implements the led_init, led_on, led_off, led_toggle calls.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
To facilitate testing, the simulated devices always return OK for
self tests.
rc.S was also upated to set CAL_XXXY_ID to the devid so tests pass the
calibration check.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
PreflightCheck was failing because it was trying to read actual
devices instad of virtual devices.
ADCSIM had a LINUXTEST ifdef that was removed.
posix_run.sh was using the wrong path
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Changed "linux" target to "posix". Most of the changes are shared with
QuRT and with OSX. The Linux specific parts are in for i2c which uses
<linux/i2c.h> and <linux/i2c-dev.h>.
There is also a check for __PX4_LINUX in mavlink for a tty ioctl that is
not supported.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Mark Charlebois