Hello everyone,
I'm facing a HardFault issue when jumping from a bootloader to an application on the STM32L476RG.
The fault occurs right at the beginning of the application, specifically at the instruction bl __libc_init_array inside the Reset_Handler (from startup_stm32l476xx.s).
Memory layout:
Bootloader: starts at 0x08000000, size: 12 kB
Application: starts at 0x08004000, max size: 150 kB (currently using ~95%)
RAM usage: around 30%
Bootloader jump sequence:
typedef void (*p_function)(void);
static void app_start(boot_t *self) {
HAL_DeInit();
self->boot.address = *(volatile uint32_t *)(self->address.app + 4);
self->boot.jump_to_app = (p_function) self->boot.address;
__set_MSP(*(volatile uint32_t *) self->address.app);
SCB->VTOR = self->address.app;
self->boot.jump_to_app();
}
Here, self->address.app = 0x08004000.
The bootloader reads the application’s initial stack pointer and reset vector from 0x08004000 and 0x08004004, respectively, and then jumps to the reset handler of the application.
Observed behavior:
A HardFault occurs as soon as the application starts, and the CFSR register indicates IACCVIOL (Instruction Access Violation).
The faulting instruction is bl __libc_init_array inside Reset_Handler.
The application works perfectly if I start it directly via debugger.
If I comment out random application functions — even ones that are never executed — the HardFault disappears.
It feels like some flash layout, alignment, or size constraint is being silently violated.
What I've tried:
Verified that the application’s vector table and Reset_Handler are correctly located at the beginning of the application flash (0x08004000).
Increased application FLASH size in linker script from 150 kB to 200 kB → no effect.
Increased _Min_Heap_Size and _Min_Stack_Size up to 3× → no effect.
Verified that SCB->VTOR is set to the correct base address before jumping.
Any insight or help would be greatly appreciated — this issue has been very tricky to track down.
Thank you!
