debug (trace) Espruino core run under gdb-multiarch (arm-none-eabi-gdb)

[espruino-discuss]

Posted at 2024-12-01 by dponyatov

Good day

I built espruino_2v24_stm32f4discovery.elf which this command line:

~/Espruino > BOARD=STM32F4DISCOVERY DEBUG=1 make

my .gdbinit:

## on load
b Reset_Handler
b SystemInit
b main
monitor reset halt

When I try to load .elf file into the gdb I got some strange address binding for source code lines:
0x3930 (without startup_stm32f4xx.s link)
0x58324
0x35076

It looks like address bindings to Flash ROM are totally incorrect.

At the same time, when I build my own empty startup code (generated by latest CubeMX), there are no any problems with addresses or source code.

Next, another problem is I always got system stopped at 0x800076c address that looks like start of EXTI0_IRQHandler

---

Attachments:

• 241201_123519.png

[espruino-discuss]

Posted at 2024-12-01 by dponyatov

targets/stm32/jshardware.c

  // Remap system memory to address 0x0000 0000 in address space
#ifdef STM32F4
    SYSCFG->MEMRMP = 0x01;
#endif

---

Posted at 2024-12-02 by @fanoush

Yes, mapping to address 0 is not "totally incorrect", it is normal, so is there something else that is wrong? Maybe it just stops at breakpoints that you have set?

---

Posted at 2024-12-02 by @gfwilliams

You can change the linker file to get all the elf files to point to 0x00.. like this https://github.com/espruino/Espruino/blob/d6e543beec98b55cb4d00bed39cb9dab7fd53790/boards/NUCLEOF401RE.py#L51

But IIRC it's better having them 0-based because then the addresses only take up ~18 bits or so and are easier to store