Move option bits from SERIALn_OPTIONS to MAVn_OPTIONS #30094
Conversation
peterbarker
force-pushed
the
pr/move-uartdriver-options-to-MAVx
branch
from
b41badd to
3c4ec3b
Compare
libraries/GCS_MAVLink/GCS_Common.cpp
Outdated
| // PARAMETER_CONVERSION - Added: May-2025 for ArduPilot-4.7 | ||
| // convert parameters; we used to use bits in the UARTDriver to | ||
| // remember whether the mavlink connection on that interface was | ||
| // "private" or not, and whether to ignore streamrate sets GCSs. |
There was a problem hiding this comment.
Again - not sure if this is just stream rate sets from a GCS only.
There was a problem hiding this comment.
Reworded this too - thanks!
peterbarker
force-pushed
the
pr/move-uartdriver-options-to-MAVx
branch
from
3c4ec3b to
0a41da0
Compare
| @@ -202,6 +202,24 @@ const AP_Param::GroupInfo GCS_MAVLINK::var_info[] = { | |||
| // @RebootRequired: True | |||
| // @User: Advanced | |||
| AP_GROUPINFO("_ADSB", 10, GCS_MAVLINK, streamRates[GCS_MAVLINK::STREAM_ADSB], DRATE(GCS_MAVLINK::STREAM_ADSB)), | |||
|
|
|||
| // ------------ | |||
| // IMPORTANT: Add new stream rates *before* the _OPTIONS parameter. | |||
There was a problem hiding this comment.
Just to group them together. I know it's not required, but it's nicer when you're looking for things.
| // PARAMETER_CONVERSION - Added: May-2025 for ArduPilot-4.7 | ||
| // Hidden param used as a flag for param conversion | ||
| // This allows one time conversion while allowing user to flash between versions with and without converted params | ||
| AP_GROUPINFO_FLAGS("_OPTIONSCNV", 20, GCS_MAVLINK, options_were_converted, 0, AP_PARAM_FLAG_HIDDEN), |
There was a problem hiding this comment.
duplicate 20? how is this not crashing on startup?
There was a problem hiding this comment.
We do not call AP_Param::check_group_info on objects loaded via load_object_from_eeprom - so only validation on the base structures.
It looks like actually calling check_group_info would require us passing more information into load_object_from_eeprom - the current parameter-name-prefix length and the currently shift value - so adding this sanity check is a non-trivial task!
Further complicating things is that these problems may arise because a script is trying to define parameters. Our current FATAL("...") approach for invalid structures isn't suitable in this case!
As to why Everything Worked: they both get shoved into eeprom, differentiated by parameter type:
00d2: type 1 (INT8) key 258 group_element 1292 size 1 value 1
01
00d7: type 2 (INT16) key 258 group_element 1292 size 2 value 0
00 00
Probably not something to rely on... I've moved _OPTIONSCNV to parameter 21 and re-tested parameter upgrade
peterbarker
force-pushed
the
pr/move-uartdriver-options-to-MAVx
branch
from
0a41da0 to
494d43a
Compare
we recently got options on a per-mavlink-connection basis. Use them
we recently got options on a per-mavlink-connection basis. Use them
we recently got options on a per-mavlink-connection basis. Use them
peterbarker
force-pushed
the
pr/move-uartdriver-options-to-MAVx
branch
from
494d43a to
a05d0f9
Compare
This is built on top of #29546