* Re-implement PKI from #1509
co-authored-by: edinnen <ethanjdinnen@protonmail.com>
* Set the key lengnth to actually make PKI work.
* Remove unused variable and initialize keys to null
* move printBytes() to meshUtils
* Don't reset PKI key son reboot unless needed.
* Remove double encryption for PKI messages
* Cleanup encrypt logic
* Add the MESHTASTIC_EXCLUDE_PKI option, and set it for minimal builds. Required for STM32 targets for now.
* Use SHA-256 for PKI key hashing, and add MESHTASTIC_EXCLUDE_PKI_KEYGEN for STM32
* Fix a crash when node is null
* Don't send PKI encrypted packets while licensed
* use chIndex 8 for PKI
* Don't be so clever, that you corrupt incoming packets
* Pass on channel 8 for now
* Typo
* Lock keys once non-zero
* We in fact need 2 scratch buffers, to store the encrypted bytes, unencrypted bytes, and decoded protobuf.
* Lighter approach to retaining known key
* Attach the public key to PKI decrypted packets in device memory
* Turn PKI back off for STM32 :(
* Don't just memcp over a protobuf
* Don't PKI encrypt nodeinfo packets
* Add a bit more memory logging around nodeDB
* Use the proper macro to refer to NODENUM_BROADCAST
* Typo fix
* Don't PKI encrypt ROUTING (naks and acks)
* Adds SecurityConfig protobuf
* Add admin messages over PKI
* Disable PKI for the WIO-e5
* Add MINIMUM_SAFE_FREE_HEAP macro and set to safe 1.5k
* Add missed "has_security"
* Add the admin_channel_enabled option
* STM32 again
* add missed configuration.h at the top of files
* Add EXCLUDE_TZ and RTC
* Enable PKI build on STM32 once again
* Attempt 1 at moving PKI to aes-ccm
* Fix buffers for encrypt/decrypt
* Eliminate unused aes variable
* Add debugging lines
* Set hash to 0 for PKI
* Fix debug lines so they don't print pointers.
* logic fix and more debug
* Rather important typo
* Check for short packets before attempting decrypt
* Don't forget to give cryptoEngine the keys!
* Use the right scratch buffer
* Cleanup
* moar cleanups
* Minor hardening
* Remove some in-progress stuff
* Turn PKI back off on STM32
* Return false
* 2.5 protos
* Sync up protos
* Add initial cryptography test vector tests
* re-add MINIMUM_SAFE_FREE_HEAP
* Housekeeping and comment fixes
* Add explanatory comment about weak dh25519 keys
---------
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
There are no calls to the functions defined in Nimble from this
class. See also older comment on line 8 about the dream to seperate
mesh and bluetooth :)
* Add GPS detection code for Airoha AG3335
Airoha AG3335 is used in Seeed T-1000E Tracker
* Add support for Airoha AG3335
Airoha AG3335 is used in Seeed T-1000E Tracker. This adds
detection code, and code to configure its use.
---------
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
* Add "Scan and Select" input method for canned messages
* Adapt canned message drawing if USE_EINK
* Indicate current selection with indent rather than inverse text
* Avoid large text on "sending" and delivery report pop-ups
* Fit SNR and RSSI details on screen
* Change hash function which detects changes in E-Ink images
The old function struggled to distingush between images on the canned-message frame, failing to update when scrolling between messages. No real justification for the new algorithm, other than "it works" and doesn't seem "too expensive". For context, this function runs once a second.
* Use canned messages (scan and select) by default with HT-VME213 and HT-VME290
* Guard for HAS_SCREEN
* Use SafeFile for atomic file writing (with xor checksum readback)
* Write db.proto last because it could be the largest file on the FS (and less critical)
* Don't keep a tmp file around while writing db.proto (because too big to fit two files in the filesystem)
* generate a new critial fault if we encounter errors writing to flash
either CriticalErrorCode_FLASH_CORRUPTION_RECOVERABLE or CriticalErrorCode_FLASH_CORRUPTION_UNRECOVERABLE
(depending on if the second write attempt worked)
* reformat the filesystem if we detect it is corrupted (then rewrite our config files) (only on nrf52 - not sure
yet if we should bother on ESP32)
* If we have to format the FS, make sure to preserve the oem.proto if it exists
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
* Turn off vscode cmake prompt - we don't use cmake on meshtastic
* Add rak4631_dap variant for debugging with NanoDAP debug probe device.
* The rak device can also run freertos (which is underneath nrf52 arduino)
* Add semihosting support for nrf52840 devices
Initial platformio.ini file only supports rak4630
Default to non TCP for the semihosting log output for now...
Fixes https://github.com/meshtastic/firmware/issues/4135
* powermon WIP (for https://github.com/meshtastic/firmware/issues/4136 )
* oops - mean't to mark the _dbg variant as an 'extra' board.
* powermon wip
* Make serial port on wio-sdk-wm1110 board work
By disabling the (inaccessible) adafruit USB
* Instrument (radiolib only for now) lora for powermon
per https://github.com/meshtastic/firmware/issues/4136
* powermon gps support
https://github.com/meshtastic/firmware/issues/4136
* Add CPU deep and light sleep powermon states
https://github.com/meshtastic/firmware/issues/4136
* Change the board/swversion bootstring so it is a new "structured" log msg.
* powermon wip
* add example script for getting esp S3 debugging working
Not yet used but I didn't want these nasty tricks to get lost yet.
* Add PowerMon reporting for screen and bluetooth pwr.
* make power.powermon_enables config setting work.
* update to latest protobufs
* fix bogus shellcheck warning
* make powermon optional (but default enabled because tiny and no runtime impact)
* tell vscode, if formatting, use whatever our trunk formatter wants
without this flag if the user has set some other formatter (clang)
in their user level settings, it will be looking in the wrong directory
for the clang options (we want the options in .trunk/clang)
Note: formatOnSave is true in master, which means a bunch of our older
files are non compliant and if you edit them it will generate lots of
formatting related diffs. I guess I'll start letting that happen with
my future commits ;-).
* add PowerStress module
* nrf52 arduino is built upon freertos, so let platformio debug it
* don't accidentally try to Segger ICE if we are using another ICE
* clean up RedirectablePrint::log so it doesn't have three very different implementations inline.
* remove NoopPrint - it is no longer needed
* when talking to API clients via serial, don't turn off log msgs instead encapsuate them
* fix the build - would loop forever if there were no files to send
* don't use Segger code if not talking to a Segger debugger
* when encapsulating logs, make sure the strings always has nul terminators
* nrf52 soft device will watchdog if you use ICE while BT on...
so have debugger disable bluetooth.
* Important to not print debug messages while writing to the toPhone scratch buffer
* don't include newlines if encapsulating log records as protobufs
* update to latest protobufs (needed for powermon goo)
* PowerStress WIP
* for #4154 and #4136 add concept of dependent gpios...
Which is currently only tested with the LED but eventually
will be used for shared GPIO/screen power rail enable
and LED forcing (which is a sanity check in the power stress
testing)
* fix linter warning
* Transformer is a better name for the LED input > operation > output classes
* PMW led changes to work on esp32-s3
* power stress improvements
* allow ble logrecords to be fetched either by NOTIFY or INDICATE ble types
This allows 'lossless' log reading. If client has requested INDICATE
(rather than NOTIFY) each log record emitted via log() will have to fetched
by the client device before the meshtastic node can continue.
* Fix serious problem with nrf52 BLE logging.
When doing notifies of LogRecords it is important to use the
binary write routines - writing using the 'string' write won't work.
Because protobufs can contain \0 nuls inside of them which if being
parsed as a string will cause only a portion of the protobuf to be sent.
I noticed this because some log messages were not getting through.
* fix gpio transformer stuff to work correctly with LED_INVERTED
Thanks @todd-herbert for noticing this and the great stack trace.
The root cause was that I had accidentially shadowed outPin in a subclass
with an unneeded override. It would break on any board that had inverted
LED power.
fixes
https://github.com/meshtastic/firmware/pull/4230#pullrequestreview-2217389099
* Support driving multiple output gpios from one input.
While investigating https://github.com/meshtastic/firmware/pull/4230#pullrequestreview-2217389099
I noticed in variant.h that there are now apparently newer TBEAMs than mine
that have _both_ a GPIO based power LED and the PMU based LED. Add a splitter
so that we can drive two output GPIOs from one logical signal.
---------
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
As reported by @mrekin, the previous changes to the platformio
custom build script may not work on windows. Change to use
python3 instead of a call to /usr/bin/env python3.
The call to generate UF2 files in the platformio custom script was
a bare call to python. In some environments, this command won't
exist in this way. Instead, use the standard env approach to find
the right python.
Additionally, add the shebang line on line 1 so this script can be
executed standalone if needed.
* Fix wio-tracker-1110 lora radio reset GPIO assignment
This fixes flaky lora radio init on this board.
* No need to keep lr11x0 radio config during sleep anymore, also stop TCXO
I think the problem (at least on the board I'm using for power testing
a wio tracker 1110) was that actually the RESET GPIO was not correct for
the radio. This led to the radio not being properly reinited after exiting
sleep mode.
Now that the GPIO is fixed I can enter deep sleep (fully shutting down
radio) and then later when the CPU resets, it can successfully init the
radio and send packets.
After this seeming success, I also turned off the TCXO during sleep and that
worked as well.
While working on #4378 I noticed a funny problem: the blinking system
LED was on during deep-sleep. Initially I thought it was some weird
sleep hw config thing but it turns out it was easier but more pervasive.
We had two different preprocessor symbols which both meant approximately the same
thing LED_INVERTED and LED_STATE_ON (though their polarity was opposite).
Some variant files were setting one, others were setting the other, and others were
setting both. heh.
In the case of the board I was testing (seeed tracker wio 1100) it was only setting one
and the default behavior for the other (for all boards) was incorrect. So I did a grep
and it seems like LED_STATE_ON was used more often, so I kept that one and removed
LED_INVERTED everywhere.
* add bootloader install script for wio tracker 1110 board
Mostly for documentation purposes for future devs.
* Generalize nrf52 hw debugging support so it works on all nrf52 targets
for wio tracker 1110 and 1000-E and possibly other nrf52 boards.
The problem was that nrf52 power stuff wasn't generating regular
powerstatus notifications (because that code was guarded by a batteryLevel
check which was null for those boards). So I've cleaned up the battery status stuff
a bit and we now have fewer special cases.
Tested on a 1000-E, tracker 1110 and a rak4631 board.
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
* Automatically generate .uf2 files (which are often used by nrf52 bootloaders for installing
app loads) anytime we generate a new hex file. This tool takes very little time to run and it is handy
for development
* Remove an old custom target I had tried to add to autogen uf2 files (that never worked)
Build output now looks like:
$ pio run --environment tracker-t1000-e
Processing tracker-t1000-e (board: tracker-t1000-e; platform: platformio/nordicnrf52@^10.5.0; framework: arduino)
...
Generating UF2 file
Converting to uf2, output size: 1395200, start address: 0x27000
Wrote 1395200 bytes to /home/kevinh/development/meshtastic/firmware/.pio/build/tracker-t1000-e/firmware.uf2
Building .pio/build/tracker-t1000-e/firmware.zip
Zip created at .pio/build/tracker-t1000-e/firmware.zip
=================================================================================== [SUCCESS] Took 9.33 seconds ===================================================================================
Environment Status Duration
--------------- -------- ------------
tracker-t1000-e SUCCESS 00:00:09.327
=================================================================================== 1 succeeded in 00:00:09.327 ===================================================================================
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
* Fix build to not use incorrect OneButton version
OneButton pushed out a new update today that has a different API
rather than just use whichever new version they push, stay on 2.5.x until
someone sees a need to update. Fixes build for wm1100 tracker.
* Update stm32.ini
* 2.6.1
* Try github tag instead?
* Update stm32.ini
---------
Co-authored-by: Ben Meadors <benmmeadors@gmail.com>
* Fix I2C pindefs
* Initial driver testing for HT-VME290
* E-Ink full refresh after canned message pop up
* Tidy variant folders
* Clean ESP32 cpuDeepSleep method
Merge sections, and remove the random assortment of gpio holds left behind.
* Enable 32kHz in variant.h
* Orient E290 with LoRa antenna facing up
* Revert "Clean ESP32 cpuDeepSleep method"
This reverts commit cb8ee508ec2d6bc27a8e228021fd1efbd034c4a0.
* Reduce deep-sleep current for VME213 (non-intrusive)
Originally I made an attempt at tidying up the cpuDeepSleep method, but have reverted that. New commit makes only the changes needed to support VME213. Don't really want the headache of breaking sleep for other variants, especially when this PR is just about implementing new boards.
* Update lib_deps; remove board_level extra
* Make SPI frequency configurable on Native
* Make the tophone queue size configurable for Portduino
* The modified SPISettings must be configured in setup(), after config.yaml is processed
* make MeshService a pointer, so we can configure MAX_RX_TOPHONE at run time
* Got a little over excited with refactoring
* Silence a warning
* More compatible
Changed a RADIOMASTER_900_BANDIT_NANO to DISPLAY_FLIP_SCREEN that is responsible for flipping the OLED screen for better compatible with other devices.
* Update variant.h Radiomaster
Remove a un-used SCREEN_ROTATE and added DISPLAY_FLIP_SCREEN
