bootloader stuck in wait_for_events if started programatically

[Tamas Varga]

Hi,

I have modified slightly the OTA bootloader sample in SDK6.1 and SoftDevice 7.1.0. After flashing the bootloader code to the nrf51822 it works like charm, I can update the firmware on the module OTA with MCP. If I upload OTA a modified version of hrs example code in SDK with OTA DFU support, the bootloader gets started by writing 0x01 to the DFU control characteristic, but afterwards it is stuck in wait_for_events function in bootloader.c. I use gcc 4.9.3 as compiler on Linux. Do you have any ideas what I should check?

I have checked with the unmodified codes in SDK v6.1. DFU does not work with MCP with those codes either. Is gcc problematic with bootloader?

Thanks,

Tamas

[Hung Bui]

Hi Tamas,

MCP v3.7 should work with DFU from SDK v6.1. Could you try and test with stock example (ble_app_hrs_dfu) in SDK v6.1 and let me know the result ?

Note that the ble_app_hrs_dfu won't do a chip reset, but jump directly from the application code to the bootloader when writing 1 to DFU control characteristic. So in the bootloader you should not reinitialize the softdevice when triggered from application.

Attached file: ble_app_hrs - TestDFU.zip ble_app_hrs_dfu.hex

[Tamas Varga]

Hi,

I have checked the unmodified sources as well, but the problem persisted. Speaking of unmodified sources I should note that I have moved the bootloader to address 0x35000 instead of 0x3C000 by altering linker script and bootloader address in header files. May it be relevant to the issue? Are you aware of any gcc issues that may cause this behaviour?

As a temporary workaround I have modified the bootloader and the application code to do a system reset and reinitialize the SD upon a DFU start command and then I could do OTA updates, but as I have read (and as you have written in your answer above), the SDK6.1 way of doing DFU update is that does not need a system reset nor reinitializing SD, so I would like to revert back to ble_app_hrs_dfu example's original intended behaviour.

Thanks for your help,

Tamas