Insufficient RAM allocated for the SoftDevice in nrf_sdh_ble_enable

[Daniel Reisfeld]

Hi,

I've defined:

#define NRF_SDH_BLE_GATT_MAX_MTU_SIZE 120 (in sdk_config.h)

I get "Insufficient RAM allocated for the SoftDevice" in nrf_sdh_ble_enable

nrf_sdh_ble_enable parameter *p_app_ram_start == 0

I'm using the last SDK version (15.0) on nRF52840

Please advise.

Thanks!,

Daniel

[butch]

For users of other IDE's such as Eclipse, this is my summary of the process, but I could be wrong....  the call to nrf_sdh_ble_enable has one argument that is an OUT variable (*p_app_ram_start, a pointer to an integer) telling you the address where application RAM SHOULD start (from calculations that nrf_sdh_ble_enable() does based on how you have configured the SD in sdk_config.h)  The SDK code also knows where you have already declared application RAM to start, and gives this error if they differ.  If you are logging, nrf_sdh_ble_enable() logs the OUT value (should be.)  Alternatively, you can use a debugger and step through the call and then read the value of the OUT variable.  Then you take the value, enter it in your linker script (Linux terminology, e.g. gcc_nrf52.ld; SES calls it the flash_placement.xml file) and rebuild.   This process is applicable to recent versions of the SDK (since v14.0?).   Whenever you change the configuration of the SD, you might need to repeat this process.  You will know whether your configuration change has affected SD RAM usage because you will get this error again.

[Daniel Reisfeld]

Thanks for this detailed and clear explanation!

Is there a similar procedure to determine the flash start and size? 

[butch]

I don't think so.  The Softdevice text is in low flash ( starting at address zero in ROM) and its length is fixed once a version and model (i.e. s112 or s132) of the Softdevice is published?  Application ROM can start just after the SD.  So you can get the values from any example linker script published for that version and model of the SDK.   But best to read the documentation for the Softdevice.

[Daniel Reisfeld]

Thanks!

Daniel