This reverts commit 17b4aa2466.
Removing the -nostartfiles caused the Taranis bootloader to include
a number of strange functions. Instead define _fini to avoid the linker
issues.
This reverts commit 17b4aa2466.
Removing the -nostartfiles caused the Taranis bootloader to include
a number of strange functions. Instead define _fini to avoid the linker
issues.
We already do averaging in the normal anaIn/getADC().
Also according to Mike's and my own research we don't need the averaging.
Tested on two X-Lites with PMW sticks
QT 5.11 finally removes the qt5_use_modules macro.
Instead target_link_libraries with the Qt5::Lib added should be used.
Since we already use that in one place, it should work everywhere. Since
the common companion lib already depends on Core, widges and multimedia
components that link against it, also implicitly link against those
libraries.
* Save 128 byte by moving variable to stack
* Make usb mass storage buffer smaller on F2 processors and put into reuseablebuffer
- Lower USB mass storage buffer to 512 bytes per default on F2
targets, makes usb speed slightly slower ~600 kB/s instead ~700 KB/s
in my tests, saving 3584 bytes
- Save another ~300 bytes by putting that buffer into reusableBuffer
* Don't put id into reusuableBuffer
All uses of id can be on the stack as well. Makes no difference for ram usage but makes it safer to reuse reusableBuffer
* Only link cdc serial when needed
We always linked usb cdc even in non debug builds. CDC uses about ~600 byte of RAM for its buffers and structures.
* Add firmware lun target for non EEPROM platforms
* Fix Fat fat tables and make FAT 1024 byte big
* Niceify some constants
* Cleanup (saves 1024 byte of flash) and fix reporting wrong Size of drive
* Silence compiler warning, set right size for flash on X12/X10
* Add firmware.txt/bootload.txt to virtual drive that displays version information
* Report also version of other component (bootloader/firmware)
* Show version also in bootloader and fix F4 platforms
* Save space on X7
* Fix X7 logic
* Compile fix for AVR
* avr-gcc does not like no newline after ifdef
* Really fix avr
* Switch backlight on when displaying a warning.
This is especially visible on the power off RSSI warning. This will also show the other warnings at start with backlight, even when blocklight is set to always off. But my X9D+ will switch its backlight on after boot once even with backlight on, so backlight=off is not so off as it seems.
* Rework power rssi code a bit
- Cleanup shutdown_confirm code
- allow x9d+ to be compiled as power button too to be able to test these things on the x9d+
- Fix shutoff after 10s in warning dialog on x9d
- Make warning dialog always turn on backlight
* Revert CMakeLists.txt
* Update multi protocols
* SFHSS also has failsafe implemented
* Show failsafe setting in Companion
* Fix missing protocol names
* Update failsafe format and only show bind when module reports to support it
* implementing syncing with multi
* Send config command
* Implement setting failsafe for multimodule
* Fix
* Fix last bugs. Multi sync now working as expected.
* Add eww multi protocol
* Show telemetry errors in Debug statistics
This introduces a second debug page on X7 and X9 platform since the original page is almost full.
Also fix the X7 menu navigation to be able to go back and forth
* Fix compile error on platforms without key_page
* avr-gcc bails on missing eolf
* Fix navigation
* Just keep pushing ....
* compile fixes of 9x and use chainMenu
* Fixes
* Align x9 behavior to x7 (for once !)
* Allow total timer reset on both long menu and long enter
* Improve 9x handling
* Remove most of the special logic associated with RF_PROTO_OFF
This simplifies code in many areas since the internal module does not require special handling and also fixes a few bugs with INTERNAL_PPM
* Disable R9M sport option if internal module has telemetry
* Fix X12s build
* Use macros instead of direct comparisions
* Fix sport bit sent the wrong way. Rename S.port out to Telemetry
* Compile fix
* Implement R9M protocol according to PXX spec 1.7
Some of the things here are guesswork as I don’t have a R9M at hand. Also Companion support for the extra R9M options is not implemented yet.
* Fix build errors
* Fix issue #4961
Fix yet another reference to HORUS_OR_TARANIS that should be IS_ARM
Also rename C9X function to export/import to be inline with the rest
* Another two IS_ARM/HORUS_TARANIS mixups
* Fix condition
* Simplify/fix two if statements
* Really fix if statement
* Fix clang warnings
- signed char on AVR vs unsigned char on ARM
- use references instead of points in calculation (same code size)
- Ignore packed address warning for mavlink includes
- Turn error on packed addresses to warning for Horus (Cannot fix that without major refactoring, might even need eeprom change)
* Revert fixes for clang on AVR as the old AVR gcc does not like them
* Allow to disable telemetry alarams
This mainly useful for cheap transceivers that have telemetry but no power amplifier and very limited range of often 10-20 meters. This includes fishpepper's tinyFISH FC (and its legal/illegal clones) and also Spektrum BNF models.
* Always disable a warning when RSSI Warnings are disabled
* Compile fix
* Fix typos, also mute sensor lost
Refactor the FrskyRSSIAlarm[2] into a EEPROM compatible RssiAlarm struct
* Compile fixes
* Also refactor Companion
* Cleanup unused datastruct on ARM
* Compile fix for AVR
* Change int to uin8_t for avr
* Fix strings
* Implement setCurve for lua
* Bug fixes after actual testing of API
* Fix documentation
* Even more error checking
* Add example code how to use this function
* Silence gcc warnings
* Drop points and start array index at 1 and note about wrong index in getCurve
* Add standard example
* Fixes
* Don’t use an invalid number in example
* fix missing 0
Thanks Mike for noticing: In pxx_arm.cpp, the pulse duration is set to either 48 or 32. This value gets put into the ARR register of the timer (via DMA). Counters count from 0 to the ARR value, then go back to 0. This will result in either 49 or 33 counts for the pulse width, thus making the pulses 0.5uS too long.
Tested only with external PXX
* Add RAS to X7 analogs screen
* Also fix SWR->RAS in simulator UI
* Only display RAS for external XJT module
The „this is almost useless but I have done the work, now it stays commit“
Clang complains about startP pointer pointing to a member of a packed structure, resulting on in a unaligned pointered address. The compiler will optimise the pointer out anyway