GNU ASM .section directive not working/linker issue

Question

I'm trying to move my _start function to 0x0, as it is the bootloader.

Flash ROM exists from 0x0 to the first 128MB (=1Gb), other memory is DDR3 RAM but we will map RAM to 0x80000000 to 0xFFFFFFFF.

The issue is with the directive .section ".vect", the _start address is not going into the .vect section, it is going into the .text section.

_start:
    .section ".vect"  /* I've tried .section .vect and I've tried moving above _start */
    b ResetHandler
    b UndefHandler
    b SVC_Handler
    b PrefetchAbortHandler
    b DataAbortHandler
    b NotUsedHandler
    b IRQ_Handler
    b FIQ_Handler
1:
    b 1b  /* Hang and don't return */

In my linker script, the MEMORY command and then the start of SECTIONS is:

MEMORY
{
    vect (rx) : o = 0x0, l = 1M
    rom (rx) : o = 1M, l = 127M
    ram (wx) : o = 0x80000000, l = 0x80000000
}


SECTIONS 
{
    .vect : ALIGN(64) 
    {
        . = 0x0;
        *(.vect*)
        VectTableEnd = .;
        VectTableSize = SIZEOF(.vect);
    } > vect 
    .... (.text, .bss, .data, .stack, etc are other SECTIONS entries)
}

But no matter what, the _start assembly function code gets shoved into the standard .text section, not at address 0x0. Does anyone know what I'm going wrong?

The code is targetted at bare-metal ARMv7A machines, compiled/linked with arm-none-eabi-as/ld

Cheers.

Answer 1

No surprises here if your _start label ends up in .text:

/* implicitly default section .text */
_start:
/* still the same section .text */
        .section .vect
/* explicitly the section .vect */
        b       ResetHandler

You probably want to switch the section before defining the _start label to make that label end up in the intended section.

/* implicitly default section .text */
        .section .vect
/* explicitly the section .vect */
_start:
        b       ResetHandler

Using the name _start as the symbol for a vector table is a bit weird (I would have expected a symbol like __vectors or vector_table), but that is beyond the scope of this question.