If the package is used for a controller that supports the HCI UART H4 transport layer protocol, the functions implemented in `nimble/transport/common/hci_h4` are very useful to deal with H4 formatted packages. If required, they can be enabled by module `nimble_transport_hci_h4`.
Using a `BLE_ADDR_RANDOM` as the peer address type allows connecting only to peers that have a static random address. Deriving the peer address type from the peer address is simple and allows connections to be established with peers that have either a public or
static random address.
If the Bluetooth 5 Advertising Extension is enabled by the `nimble_adv_ext` module, up to 3 events come in from the controller during connection establishment before they are processed by the host. The default size of the event queue `MYNEWT_VAL_BLE_TRANSPORT_EVT_COUNT` with only 2 entries is therefore too small and the connection establishment fails.
To control the compilation of NimBLE modules `nimble_*_ext` that require the Bluetooth 5 Advertising Extension, the `ble_adv_ext` is introduced to indicate that a platform supports this feature.
The patches include the following changes:
- define ARRAY_SIZE in `component/spi_flash/spi_flash_timing_tuning.c` only if it is not yet defined by RIOT macros
- add alternative implementations for`spi_flash_disable_interrupts_caches_and_other_cpu` and `spi_flash_enable_interrupts_caches_and_other_cpu` if compiled for RIOT
- fix the undefined reference to `rtc_gpio_force_hold_en_all` in `components/driver/gpio.c`
- rename the bootloader patch to fix the serial number
This mostly converts switch statements from double indent style to
Linux kernel style, as required per the coding convention.
From the C compiler perspective, this is a whitespace only change.
It turns out that #18287 needs to be somewhat reverted.
The make behaviour requires use to add psuedomodules for packages
that are not compiled (ie header only).
The only difference from the revert is that the module is no longer
selected in Kconfig so it will match the modules used list.
There was a mismatch between Kconfig and make, after some digging it
appears that the make never used the emlearn module, only the package.
This removes the emlearn pseudomodule from make since nothing selects it
and removes the MODULE_EMLEARN from Kconfig to match the make dependency
resolution.
