EmbDev.net

Forum: ARM programming with GCC/GNU tools Is ENTRY(_Reset) redundant here?


von Splee S. (Company: ss) (splee)


Rate this post
useful
not useful
I read a code somewhere:

startup.s:
=========================================
.section INTERRUPT_VECTOR, "x"
.global _Reset
_Reset:
    B Reset_Handler /* Reset */
    B.
    ...

Reset_Handler:
    LDR sp, =stack_top
    ...

test.ld:
=========================================
ENTRY(_Reset)
SECTIONS
{
    . = 0x0;
    .text: {
    startup.o (INTERRUPT_VECTOR)
    *(.text)
    }
    ...
}

I understand that ENTRY(_Reset) set the program entry point to _Reset.

Q1. What if .text does not start from 0x0? Will the first instruction to 
be executed still be "B Reset_Handler"?

Q2. In the original test.ld, is ENTRY(_Reset) still necessary, or is it 
redundant? Since we already set startup.o (INTERRUPT_VECTOR) to start 
from 0x0 already.

von Markus F. (mfro)


Rate this post
useful
not useful
Splee S. wrote:
> I understand that ENTRY(_Reset) set the program entry point to _Reset.

Basically the only thing ENTRY(...) does is to place the information 
about this entry point into the header of your executable, provided such 
thing (header) is available and can hold such information (which is the 
case for ELF, for example).

This will only have any effect if you have a loader (i.e., an ELF 
loader, in our case) that will use this information at runtime. In 
embedded coding, more often than not you won't have such luxury, i.e. 
there isn't any header since you basically convert your ELF file into a 
flat binary during the flash process.

Bottom line: in most use cases, the ENTRY() statement is basically 
useless for embedded development.

Please log in before posting. Registration is free and takes only a minute.
Existing account
Do you have a Google/GoogleMail account? No registration required!
Log in with Google account
No account? Register here.