We will reserve BOARD_FLASH_SIZE for the internal flash on stm32 flash processors, use HAL_PROGRAM_SIZE_LIMIT_KB in the general code base.
Notable change here is that boards with external flash will start to get features only available with more than 2MB of program storage
57a3bc1397 changed the code from "internal" to "in-flight
It seems the old value of "1" was no longer valid
It also changed things to that the learning system saved the offsets.
* remove unnecessary nullptr check, these are always called from an
initialized AP_DroneCAN so if it's nullptr something has gone
horrifically wrong
* pass in driver index instead of repeatedly calling function to get it
* simplify error handling; knowing exactly which allocation failed is not
super helpful and one failing likely means subsequent ones will too,
as it can only fail due to being out of memory
the WAI (whoami) register is writeable. Not only is it writeable, but
the written value is persistent across a power cycle. You have to
remove power for about 30s for it to finally go back to the right
default value of 0x10
this makes using WAI as a test for finding a IST8310 problematic. The
best we can do is send a soft reset which will reset it to default for
us to then check
Two structs with the same name must have exactly the same definition, no
matter where they occur in the program, otherwise the program is undefined.
Move each sample register struct definition into the associated class
definition so they are in a different namespace and no longer
identically named, thus fixing this issue.
this fixes an issue with the following sequence:
- new board (or board with FORMAT_VERSION reset) starts up with only internal compasses
- internal compasses are detected and devids saved
- an external compass is added and the board is rebooted
- the external compass will not be the first compass
- user then calibrates and flies, but has internal as primary
this can lead to a very bad experience for new users. At least one
vehicle has crashed due to this sequence.
The fix is to not save device IDs during the Compass::init() if we
have never been calibrated. This means that when an external compass
is added it will come up as the first compass.
This also removes the saving of the extra device ID. It was never
intended that these be saved (there is a comment to that effect in the
code), but actually they were saved.
