[DO NOT MERGE] LMC: Mavlink ULog rate control support #612
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This would prevent client to stop working if agent restarts in middle of connection initiation. If entity creation fails the client stop connecting and waiting for next ping response from agent before next connection attempt.
Mavlink start supports '-i <x>' flag to read IP address from MAV_<x>_REMOTE_IPn params.
-T is the command line argument for LD, it should not be in the script name
The type of main (entry point) is now relevant, as the entry point is no longer explicitly casted to main_t
For some reason, the type of the linker defined symbols was changed. The reasoning was "they typically represent pointers" which is absolutely not true at all. They typically define memory locations, but they can be whatever. It is just a way to provide symbols from the linker to the executable, which is done via relocation symbols.
- Allocate a separate wait semaphore for each process (one in PROTECTED_BUILD). Allocate the semaphore in "REGISTER" ioctl, free it in "UNREGISTER" if process exits - Duplicate the user side call events to kernel memory to make them accessible by kernel at all times. Keep track of the allocated items in "freelist" queue - Replace fixed size callout queue with an "inflight" queue Signed-off-by: Jukka Laitinen <[email protected]>
…versions These give errors on some compilers Signed-off-by: Jukka Laitinen <[email protected]>
microdds_client dds topic generator to support CamelCase topic names
Mavlink msg generator (pymavlink) does not handle mavlink2 message extensions properly. Receiver does not zero-fill the receiver buffer, so due to the extension zero suppression featyure the last "suppressed" fields may contain garbage data. HITL_SENSOR message last field is sensor_id, so it is explicitely set zero to avoid issues.
dds_topics.yaml has support for adding optional rate parameter for
each publication topics
e.g. Limit SensorCombined topic pub rate to 10 Hz:
- topic: /fmu/out/sensor_combined
type: px4_msgs::msg::SensorCombined
rate: 10
Not usable from user space
The ioctl parameter type was wrong, need to give the handle as well.
The shutdown hook is run when the system is shut down, what we want is to call hrt_stop() when the process exits -> use atexit().
- Pop the entry from any queue when fiddling around with it, this prevents
two threads accessing the same data via the queue
- When looking for any item in any queue, the search should be performed
with BOTH user entry, and the handle provided. This is because user
entries are ambiquous (user virtual addresses) and thus the kernel
handle (which is unique) is needed
- Fix HRT_CANCEL on the kernel side
- Fix HRT_UNREGISTER on the kernel side
- Remove potential double booking in callout_queue and callout_inflight.
- Do not free items from the freelist, as they do not belong to any
thread specifically (as they have been freed)
- Remove kernel_entry from the hrt_call structure, the same information
can be done by comparing user entry + handle
The only thing needed from there was the definition of struct builtin_s. The builtin functions are not meant to be used this way anyhow, it is supposed to be a way to inject user-space code into the kernel (the builtin list is _copied_ into kernel memory). The board_ioctl module is is kernel-space only. Also, remove the usage of the "FAR" keyword, we use processors from this millennium that do not use memory segmentation into 64K chunks... The real reason for this change is that CONFIG_BUILD_KERNEL does not allow setting CONFIG_BUILTIN (makes sense, injecting user code into the kernel is not feasible _at all_ with address environments).
Allow starting a process with its full path i.e. strip the path from argv[0] and use the base name to find the entry point.
Do not link the soft float math library, it is not needed for anything and it confuses the linker which does relocations to the FP instructions
Signed-off-by: Jukka Laitinen <[email protected]>
Signed-off-by: Jukka Laitinen <[email protected]>
…gnal Sigaction lets us specify that only the wanted signal is handled with the callback Signed-off-by: Jukka Laitinen <[email protected]>
CA_ROTOR and PWM_MAIN_FUNC params added to init.d-posix/4400 airframe to make actuator outputs working for sitl
Bumps [docker/login-action](https://github.com/docker/login-action) from 2 to 3. - [Release notes](https://github.com/docker/login-action/releases) - [Commits](docker/login-action@v2...v3) --- updated-dependencies: - dependency-name: docker/login-action dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <[email protected]>
Pull latest changes + get rid of merge commits
Signed-off-by: Jukka Laitinen <[email protected]>
…ates not set If update not called, it may lead to infinite loop in case check_for_updates not set and there are new messages in ulog_stream_acked uorb.
Signed-off-by: Jukka Laitinen <[email protected]>
An EEPROM behind and i2c media is not 100% failure proof. But any errors in e.g. mtd_params reading can cause catastrophic failures in flight. Mitigate any possible errors in px4 mtd_params reading by adding crc code to every page on EEPROM In addition: - Read every block into a temporary buffer before copying to clients buffer; this fixes some issues with memory protected builds when writing directly to user buffers from i2c driver interrupt handler - If a read request fails, don't return any partially filled buffer to user Signed-off-by: Jukka Laitinen <[email protected]>
Like __dso_handle, this symbol must NOT be pulled from the user libc. The __cxa_atexitatexit() functions in kernel are no-ops as we don't support shared library destructors.
…_for_updates not set" This reverts commit f3a8744. Updating and sometimes ignoring ulog_stream_acked reduces the load, but may cause log to be corrupted due to missing definition data. Need to revert the patch to avoid broken log files.
jnippula
force-pushed
the
DP-8103-ulog-rate-control
branch
from
a72b210
to
e17c4cb
Compare
jlaitine
approved these changes
Jan 30, 2024
maseabunikie
approved these changes
Jan 30, 2024
jnippula
changed the title
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Introduce parameter MAV_ULOG_MAX_RT to tune mavlink logging upload message rate (default value: 0.2f).
This makes possible to reduce mavlink data rate in LMC device where CPU load is already very high.