The getopt command uses global variables and is not thread safe.
Created a minimal px4_getopt version that supports options with
or without an arg, and random placement of options on the command line.
This version modifies the order of the args in argv as does the
POSIX version of getopt.
This assumes that argv[0] is the program name. Nuttx may not support
that properly in task_spawn.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The simulator satisfies the dependencies for an accelerometer
being present.
The accel code compiles but is not fully functional.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The simulated device satisfies the factory pattern used by
MS5611 to create a specific I2C or SPI device instance.
For now the functions just return true, but should/could
return simulated data.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
I have not been able to unravel why nullptr is passed as the device
path to the constructor of ms5611_i2c.
This crashes the VDev code as it expects to create a virtual driver
with the device path passed as devname. It causes VDev to do a
strncmp with null.
Using /vdev/ms5611_i2c as the name for the now.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The command shell was spewing debug infor about the command and
parameters. Removed the debug output.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
If PX4_I2C_SIMULATE is set to 1, then the actual I2C device will
not be opened and all transfers will succeed.
If PX4_I2C_SIMULATE is false and transfer() is called, then the
appropriate ioctl is make on the actual device.
if I2C::ioctl is called via px4_ioctl() then the command fails and
a warning is printed to use I2C::transfer
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The parameter is not present in the linux implementation so removed
the documentation for the parameter.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
To avoid confusion when a real device and a virtual device is
being used, changed CDev to VDev for Linux.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Added PX4_ISFINITE(x) to px4_defines.h to handle the differences on
NuttX and Linux.
This change also picked up some file renaming for virtual character devices
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
px4_errx kills the process, so if possible we want to end the thread
but not the process. Using warnx and return exits gracefully.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The build will now fail if clang is not found. To force the use
of GCC, use:
make USE_GCC=1
The toolchain makefile was modified so it no longer checks for
various versions of clang if USE_GCC=1 is passed.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
__param_start and __param end need to be 8 byte aligned on
64bit machines. Changed linker script to 8 byte align __param
section.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Disabled gcc warnings that are tripped by Eigen.
Removed signal code that is not needed in Linux port and was
causing gcc warnings.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
PX4 uses NuttX data structures throughout so those data structures
were preserved and used to implement high and low priority queues.
A unit test for the work queues was added.
The polling rate of the queues are set in px4_config.h in
CONFIG_SCHED_WORKPERIOD. The units are milliseconds.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Now run:
make PX4_TARGET_OS=nuttx
or
make PX4_TARGET_OS=linux
To test the linux build and make sure that the required directories
exist, run:
make linuxrun
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Added linker script to resolve __param_start and __param_end.
Added mc_att_control to list of supported builtins.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
The builtin commands all have _main suffix by convention so
no need to show _main. Also nsh calls the commmands without the
_main suffix.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
If the topic has not been published, orb_copy returns a
negative number which causes update() to memset the data
contents to zero.
In some instances data is a null pointer. This causes a
segment violation crash.
Added a check for data != 0
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
I2C class derives from CDev class which requires a devname
but in at least some instances, a nullptr is passed for devname.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
When printing a uint64_t type using %llu, this works on a 32bit
system, but on a 64bit machine uint64_t is an unsigned long.
The compiler complains about unmatching types.
The time times in PX4 should likely have been unsigned long long
and not uint64_t as that type changes per architecture.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
For now it uses the bus number as the id. Not sure how this should
actually be mapped.
Seems like the I2C devices come up in random order and have random
id but that a specific device can be found in the /sys/bus/i2c
interface.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
if termios.h is included before mathlib.h then BO is undefined.
Since mathlib.h is not needed it was removed but I still don't
know why this error occurs.
Also added -lrt to link flags for clock_gettime
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Not yet functional. Full implementation will provide an IOCTL interface to
do bi-directional transfer. will model the interface after Linux.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
When running the process without sufficient privilege to use
real time scheduling, warn the user and run with SCHED_OTHER.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Mark Charlebois
Lorenz Meier