_export_path and _duty_path will have been already long time gone by the
time dtors kick in.
Probably better to use OwnPtr around those. But it's better to be done
in a separate PR.
Checking the time on the tcpdump capture, it matches the first fields
from the data:
$ tshark -n -c 4 -r ~/tmp/solo/rc.pcap
1 0.000000 10.1.1.1 → 10.1.1.10 UDP 68 5005 → 5005 Len=26
2 0.019976 10.1.1.1 → 10.1.1.10 UDP 68 5005 → 5005 Len=26
3 0.040046 10.1.1.1 → 10.1.1.10 UDP 68 5005 → 5005 Len=26
4 0.059961 10.1.1.1 → 10.1.1.10 UDP 68 5005 → 5005 Len=26
From the previous commit (first 2 packets):
5fa8 f441 3414 0500 73d7 dc05 dc05 dc05 db05 e803 e803 e803 f401
73f6 f441 3414 0500 74d7 dc05 dc05 dc05 db05 e803 e803 e803 f401
0x0005143441f45fa8 - 0x0005143441f4f673 = 0x4E14 = 19988 (usec)
Which seems to approximately match for the other packets as well. We are
not using the field since we rather get the time when we receive it, but
at least use a better name.
If a thread other than the main one calls Scheduler::delay() we could
end up triggering the call of delay callbacks. Those should only ever
happen on the main thread.
We are setting a termination handler for some signals which are of not
interest. Just ignore them. Ignoring SIGWINCH allows for example to
run on a screen and change the window size later without killing
ardupilot.
RPI-based boards that use RCInput_RPI need more stack space otherwise we
end up with stack corruption. This leads to crash particularly when also
using GPIO_RPI since it may change what that driver is poking on memory.
This increases stack size to 1M which is overkill for most of other
boards with a more controllable stack usage. However this exposes that
on multiple different HWs a single point for stack size decision may not
be the best. This can be improved in future.
While at it, add final and override to mark this as being the overriden
final implementation of this method.
Thanks to Phillip Khandeliants (@khandeliants) for reporting.
While at it, add final and override to mark this as being the overriden
final implementation of this method.
Thanks to Phillip Khandeliants (@khandeliants) for reporting.
Integrate the gyro values pushed by the inertial sensor backend using
bias values sent by EKF.
Use the unblocking RingBuffer to avoid locking the callers.
- avoid trying to close fd when it's -1
- Keep includes sorted
- AP_HAL::panic() doesn't need \n terminator
- %u requires unsigned type
- #pragma once is the first thing on a header
when a library called read() it would clear the new input flag, which
would cause new_input() in the main loop to return false. This could
trigger a false RC failsafe.
When PWM_Sysfs_Base constructor is called, global variable hal may not
have been initialized resulting in NULL dereferencing error.
Move hal dependent stuff from contructor to init method.
By opening with O_CLOEXEC we make sure we don't leak the file descriptor
when we are exec'ing or calling out subprograms. Right now we currently
don't do it so there's no harm, but it's good practice in Linux to have
it.
Several return values in the constructor of the scheduler were ignored
before, while they should be respected.
I found that bug while strac'ing ardupilot as it failed at some later
point.
Signed-off-by: Ralf Ramsauer <ralf.ramsauer@othr.de>
This allows to terminate the flight stack nicely, ensuring it returns 0
so init system can check by return code if it terminated nicely or if it
was due to a crash.
This allows to wakeup the thread that is sleeping on Poller::poll()
[ which in our case is an epoll_wait() call ]. This is usually achieved
by using a special signal and using the pwait() variant of the sleeping
function (or using signalfd). However integrating the signal in the
Thread class is more complex than simply use the eventfd syscall which
can serve our needs.
Up until now we rely on Thread objects and variants thereof to be allocated
on heap or embedded in another object that's zero'ed on initialization.
However sometimes it's convenient to be able to use them on stack as
will be the case when writting unit tests.
Initialize all relevant fields to allow them to be used on stack. While
at it, prefer C++11 initialization on Poller since it's only setting the
default (invalid) value.
RC_Channel: To nullptr from NULL.
AC_Fence: To nullptr from NULL.
AC_Avoidance: To nullptr from NULL.
AC_PrecLand: To nullptr from NULL.
DataFlash: To nullptr from NULL.
SITL: To nullptr from NULL.
GCS_MAVLink: To nullptr from NULL.
DataFlash: To nullptr from NULL.
AP_Compass: To nullptr from NULL.
Global: To nullptr from NULL.
Global: To nullptr from NULL.
This patch replaces the 'old style' ringbuffer by the ByteBuffer class.
An effort was made to keep the exchange as close as possible from a
drop-in replacement to minimize the risk of introducing bugs.
Although the exchange opens opportunities for improvement and
simplification of this class.
When the buffer wraps and we do it in 2 steps, we can't actually do the
second part if it fails or if we wrote less bytes than we intended,
otherwise we will corrupt the data being sent.
This patch replaces the 'old style' ringbuffer by the ByteBuffer class.
An effort was made to keep the exchange as close as possible from a
drop-in replacement to minimize the risk of introducing bugs.
Although the exchange opens opportunities for improvement and
simplification of this class.
The constant passed to cflag is BOTHER, meaning the actual baud is set
in the other specific members. Don't define B* constants as they are
misleading here and this is why it doesn't work with e.g.
cfset[io]speed()... that function expect a B* constant which in Linux
is not the speed, but an index to an array with speeds.
Accidentally pushed in commit 298f7bffb9
The order of the motors shouldn't have been changed on version 5 because
it is specific to older versions of the ESC controller firmware
We currently check examples are buildable with waf which doesn't need
the libraries to be specified in a make.inc file. Having the makefiles
there is misleading since people try to build and realize the build is
broken.
this is a RCInput module for multiple R/C uart protocols running at
115200 baud 8-bit. We can decode multiple protocols in parallel with
this module, relying on frame timing and CRCs to get the right
protocol
Contributions from:
- Gustavo Jose de Sousa <gustavo.sousa@intel.com>
- José Roberto de Souza <jose.souza@intel.com>
- Lucas De Marchi <lucas.demarchi@intel.com>
- Patrick J.P <patrick.pereira@intel.com>
25c7e8b changed the logic of transfer(). Align
I2CDevice::read_registers_multiple() in the same way.
Signed-off-by: Ralf Ramsauer <ralf.ramsauer@othr.de>
If I2CDevice::transfer() has to do nothing it returns false. This can be
misleading, as this might feel contradictory.
Let's spend a comment on that.
Signed-off-by: Ralf Ramsauer <ralf.ramsauer@othr.de>
According to man 3 ioctl, ioctl returns other values than -1 on success.
So loop while ioctl returns -1.
Furthermore, there is no necessity to initialise r with -EINVAL,
Signed-off-by: Ralf Ramsauer <ralf.ramsauer@othr.de>
We aren't going to use all the poller infra for now and we need it's
behavior a little bit different for what we are going to use:
- Do not use any "fair" time for each ready fd since we don't want
to set a timeout
- Allow to set the fd on Pollable after constructing it since we are
likely to embed Pollable inside other structs and just later be
able to open an fd.
- Let caller use the epoll flags directly - this is not in AP_HAL,
so there's no need to abstract them
Add system's polling infrastructure to be notified whenever a
file descriptor is ready to be read from or written to.
Adds a few classes:
* Poller, as an interface to epoll()
* Pollable, as an interface to a file descriptor
../../libraries/AP_HAL_Linux/SPIDevice.cpp: In member function ‘virtual AP_HAL::OwnPtr<AP_HAL::SPIDevice> Linux::SPIDeviceManager::get_device(const char*)’:
../../libraries/AP_HAL_Linux/SPIDevice.cpp:337:27: warning: comparison is always false due to limited range of data type [-Wtype-limits]
for (uint8_t i = 0; i < _n_device_desc; i++) {
~~^~~~~~~~~~~~~~~~
Define a dummy device to remove warning.
This centralized private header encourages centralizing things on
umbrella headers that are a pain to maintain. Force each part of
AP_HAL_Linux to include what is used.
Make some member variables protected to follow what we do in other
places (and there's no reason to be private).
Move defines to .cpp to reduce their visibility.
This centralized namespace header encourages centralizing things on
umbrella headers that are a pain to maintain. Force each part of
AP_HAL_Linux to include what is used.
While at it, do some whitespace cleanups and minor changes to adhere to
coding style.
This is selectable by a define and is never changed. Just remove
everything referencing it: we can come up with a better solution if it
is actually used later.
There's little point in having the Linux::AnalogIn just to implement and
empty interface. All implementations inside AP_HAL_Linux are already
inheriting directly from AP_HAL, so just remove it.
In a case ArduPilot is launched as a background process without
detaching with *nohup* like this ./arduplane -C /dev/ttyAMA0 ConsoleDevice
is created and an attempt to read from it is made. This yields in a stopped
process. This is an endeavour to overcome this problem.
Using factory method maked it easier to grasp the lifetime of all object
that get created and destroyed. Instead of spanning this thoughout whole
source file we have it nicely encapsulated in this a little horrendeous
_parseDevicePath that is of course to improve more.
Otherwise we're going to leak memory without any need.
Before this fix we've created ConsoleDevice 4 times in case -A switch hadn't been supplied,
but we hadn't ever deleted those. Now there's no memory leak here.
Minor changes to follow coding style and improve readability:
- sort headers
- move struct definition to compilation unit rather than header
- Add braces to if, for, etc
../../libraries/AP_HAL_Linux/Perf.cpp: In member function ‘void Linux::Perf::_debug_counters()’:
../../libraries/AP_HAL_Linux/Perf.cpp:85:36: warning: format ‘%llu’ expects argument of type ‘long long unsigned int’, but argument 4 has type ‘uint64_t {aka long unsigned int}’ [-Wformat=]
c.name, c.count);
^
Test code for integration with another thread to pull data from internal
perf counters. Since we are using the timer thread here, there's no
retry mechanism and we only print that data can be corrupted.
Instead of creating a new object Perf_Lttng copying the necessaries
fields, just make a tighter integration with the internal perf counters
and re-use the same fields.
The idea is to leave the internal perf enabled all the time, like it is
in PX4, and then allow the integration with lttng on top. Next step
would be to runtime enable/disable only the perf counters we are
interested in.
This also changes the structure so it's easy to allow another thread to
pull data from the Perf object. A rw lock protects from addition of new
counters and an atomic unsigned int allows other threads to do a
lockless copy of the data.
In order for this to work the allocation was changed to use a single
memory pool instead of returning a calloc'ed data for each perf counter.
Since most of our counters are of ' elapsed' type, don't bother using a
smaller struct for the 'count' type
If the RPi version detection fails, the standard version is now RPi2/3 instead of RPi1.
I think this is useful, because the RPi1 is not really supported (performance reasons).
Like others, use HAVE_ prefix and name it HAVE_LTTNG_UST to be the same
name as exported by pkg-config While at it remove wrong comment with
_HELLO_TP_H.
Implementation of alloca() is very much architecture and compiler
dependent. Avoid the case in which it could return a non-aligned
pointer, which would mean Thread::_poison_stack() would do the wrong
thing.
../../libraries/AP_HAL_Linux/Thread.cpp: In member function ‘void Linux::Thread::_poison_stack()’:
../../libraries/AP_HAL_Linux/Thread.cpp:63:20: warning: declaration of ‘start’ shadows a member of 'this' [-Wshadow]
uint8_t *end, *start, *curr;
^
Sort include alphabetically and make them in order:
Main header
system headers
library headers
local headers
While reordering, change a include of endian.h to our sparse-endian.h
which is more reliant to toolchain changes.
Running the vehicles we check the stack size doesn't grow too much by
enabling the DEBUG_STACK in the scheduler. Even on 64bit boards the
stack is consistent around 4k. Just to be a little conservative, let it
be a little bit more that that: 256kB.
Since we have RT prio and we call mlock(), the memory for the stack of
each thread is locked in memory. This means we are effectively taking
that much memory. The default stack size varies per distro, but it's
common to have 8MB for 64 bit boards and 4MB for 32 bit boards. Here is
the output of ps -L -o 'comm,rtprio,rss $(pidof arducopter-quad)', showing the
RSS of arducopter-quad before and after this change:
Before:
COMMAND RTPRIO RSS
arducopter-quad 12 46960
sched-timer 15 46960
sched-uart 14 46960
sched-rcin 13 46960
sched-tonealarm 11 46960
sched-io 10 46960
After:
COMMAND RTPRIO RSS
arducopter-quad 12 7320
sched-timer 15 7320
sched-uart 14 7320
sched-rcin 13 7320
sched-tonealarm 11 7320
sched-io 10 7320
We don't need all the comments in the array declaration and we can
inline its declaration in the function call. This makes it easier to
copy it to other places.
This bug led to issues for us so it may help others to resolve it.
Currently, the AP_HAL_Linux RCInput::read(uint16_t*,uint8_t) function
only returns the first x nonzero channels. Once it hits a channel that
is set to zero, it stops and all remaining channels are returned as
zero, even if they are set. This causes discrepancies between the raw RC
input sent to the GCS and the RC input that is actually used on the
vehicle.
The fixes this issue and makes it behave exactly as it does on the
PX4_HAL code. We ran into this issue when sending rc_override messages
in which there were some channels set to zero.
0-length arrays are supported in C but forbidden in C++. GCC allows it
but clang is more strict:
../../libraries/AP_HAL_Linux/SPIDriver.cpp:75:35: fatal error: no matching constructor for initialization of 'Linux::SPIDeviceDriver [0]'
SPIDeviceDriver SPIDeviceManager::_device[0];
^
../../libraries/AP_HAL_Linux/SPIDriver.h:20:7: note: candidate constructor (the implicit move constructor) not viable: requires 1 argument, but 0 were provided
class SPIDeviceDriver : public AP_HAL::SPIDeviceDriver {
^
../../libraries/AP_HAL_Linux/SPIDriver.h:20:7: note: candidate constructor (the implicit copy constructor) not viable: requires 1 argument, but 0 were provided
../../libraries/AP_HAL_Linux/SPIDriver.h:25:5: note: candidate constructor not viable: requires 9 arguments, but 0 were provided
SPIDeviceDriver(const char *name, uint16_t bus, uint16_t subdev, enum AP_HAL::SPIDeviceType type, uint8_t mode, uint8_t bitsPerWord, int16_t cs_pin, uint32_t lowspeed, uint32_t highspeed);
^
1 error generated.
This allows us to re-use SPIDevice from SPIDeviceDriver (the
to-become-SPIDeviceProperties) while the drivers are
converted. We create a fake device by calling the temporary
SPIDeviceManager::get_device() method passing the descriptor. The
transfer and assert logic is still using the old code.
Now we can interoperate SPIDeviceDriver with the ones based in
SPIDevice since they are going to use the same semaphore and bus.
The way this code is structured is a little bit different from the
SPIDriver implementation:
- We only open the bus once, no matter how many devices we have in it
- There's a single transfer() method which uses half-duplex mode
instead of full duplex. The reason is that for all cases in the
codebase we are using half-duplex transfers using the full-duplex
API, i.e. a single SPI msg with both tx and rx buffers. This is
cumbersome because the buffers need to be of the same size and the
receive buffer using an offset of the same length as the actux data
being written. This means the high level APIs need to copy buffers
around.
If later we have uses for a real full duplex case it's just a matter
of adding another transfer_fullduplex() method or something like
this.
- The methods are implemented in the SPIDevice class instead of having
proxy methods to SPIDeviceManager as is the case of SPIDriver
Also from now on we refer to the SPIDriver objects as "descriptors"
because they have the parameters of each device in the
SPIDeviceManager::devices[] table. When SPIDeviceDriver is completely
replaced we can rename them to SPIDeviceProperties.
Save in the manager the number of devices so it can be used in other
places like the SPIDevice implementation. This is a temporary storage
while we migrate to SPIDevice.
While at it use protected rather than private.
This allows us to re-use I2CDevice from I2CDriver while the drivers are
converted. We create a fake device with addr = 0 for each I2CDriver but
we only use the register/unregister logic. The transfer logic still uses
the methods from I2CDriver in order to use the right address.
Now we can interoperate I2CDevice drivers with the ones base in
I2CDriver since they are going to use the same semaphore and bus.
The I2CDriver constructors were changed to re-use the logic in I2CDevice
(it uses a number rather than an string) and the semaphore doesn't live
outside anymore, its embedded in the fake I2CDevice, as well as the
bus's file descritor.
This is a similar function to what we have in I2CDriver, but it can
receive a nullptr to recv or send. It will create 2 i2c_msg structs to
send and receive data to/from the I2C slave.
These are very similar to their counterparts in I2CDriver. The changes
were:
- Don't use fixed buffer with PATH_MAX length: allocate the string
- Change the interface to use std::vector so we can simplify the
implementation
Use pthread's barrier so we don't keep waking up threads with possibly
higher priority during initialization phase.
This also synchronizes all of them to a single point. With the previous
approach it was possible (but unlikely) that a thread hadn't reach the
synchronization point when main thread signalize "system initialized".
A string name allows to more easily expand the table: the board is
responsible for the name and doesn't have to declare it as a enum or
define. It's also easier to add 2 sensors of the same type.
This reverts support for RCInput via PWM. This is causing trouble in
some RPI-based boards, receiving a SIGSEGV. Let's revert it for now and
retry this later.
This reverts commit 5629f38b2c.
This reverts commit 51fd0b3d55.
This reverts commit 79d56073f7.
../../libraries/AP_HAL_Linux/RCInput_RPI.cpp: In member function ‘virtual void Linux::RCInput_RPI::_timer_tick()’:
../../libraries/AP_HAL_Linux/RCInput_RPI.cpp:489:127: warning: ‘x’ may be used uninitialized in this function [-Wmaybe-uninitialized]
counter = circle_buffer->bytes_available(curr_pointer, circle_buffer->get_offset(circle_buffer->_virt_pages, (uintptr_t)x));
^
This shrink must be used when the output camera sizes doesn't fit
the expected output.
We don't need to crop it even when the camera sizes aren't squared since
the shrink_8bpp() function shrinks a selected area.
This function shrinks a selected area on a 8bpp image.
The focus in this function was the performance, so this may not be the
clearer or the most understandable way to write it. The performance
was measured using GoogleBenchmark[1].
[1] - https://github.com/google/benchmark.git
Support for perf api using lttng.
Some additional build tricks needed for bebop because lttng uses dl_open
which is not compatible with a static link on a different libc as used
on the bebop
Georgii Staroselskii
Lucas De Marchi
Peter Barker
Siddharth Bharat Purohit
Andrew Tridgell
hiro2233
Pierre Kancir
davidaroyer
Eugene Shamaev
mirkix
murata
Mathieu OTHACEHE
Julien Beraud
Ivan Smirnov
Mathieu OTHACEHE
Ralf Ramsauer
Murilo Belluzzo
Martin Evans
William Liu
Gustavo Jose de Sousa
Patrick J.P
Leandro Pereira
Luiz Ywata
dgrat
Aaron Wang Shi
Ricardo de Almeida Gonzaga
Rustom Jehangir
Randy Mackay
raspilot
Pritam Ghanghas
mirkix