Pinetime board gets stuck in cpu_init()

[arteeh]

Description

When creating an application for the Pinetime, cpu_init() and consequently board_init() will get stuck somewhere, meaning any code after that does not seem to run.

Steps to reproduce the issue

Create a simple application, with the RIOT folder checked out to 2021.01-branch. Make with the following definitions:

BOARD ?= pinetime
USEMODULE += ili9341

Create a main function where the following code is run:

ili9341_t display;
board_init(); // or cpu_init() and setting the display registers manually
BACKLIGHT_ON;
ili9341_init(&display, &ili9341_params[0]);
ili9341_fill(&display, 0, 319, 0, 239, 0x0000);
ili9341_fill(&display, 10, 59, 10, 109, 0x001F);

I then run PROGRAMMER=openocd DEBUG_ADAPTER=stlink make flash with an ST-Link V2 wired up to my Pinetime to flash the program.

Expected results

The Pinetime display shows a blue rectangle on a black background, at low brightness.

Actual results

The Pinetime doesn't show any activity, and nothing happens on the screen.

Versions

Operating System: Fedora Silverblue 34 GCC: arm-none-eabi-gcc (Fedora 10.2.0-4.fc34) 10.2.0 CLIB: arm-none-eabi-newlib: "4.1.0" Make: GNU Make 4.3 Debugger: Open On-Chip Debugger 0.11.0

[kaspar030]

does tests/driver_ili9341 display anything?

[arteeh]

does tests/driver_ili9341 display anything?

The test works well on my device, but it doesn't use board_init() or cpu_init() which is what this issue is about. I just tried to add cpu_init() to the beginning of the test, and nothing shows up on the display.

[kaspar030]

board_init() and cpu_init() are called automatically by RIOT's startup procedure, on every board, before main(). Are you calling it again in your application?

[arteeh]

I didn't know board_init() gets run before main(). This issue is a misunderstanding of mine then, closing.