The BOOTRST fuse was disabled, but bootloader code is still running after chip reset

Question

I am experimenting with an ATmega328P bootloader, and I created an Intel HEX (.hex) file with a program with an infinite loop in the beginning and with Optiboot code at the end of the hex file. (Code loaded into the chip is here: http://relliks.php5.cz/stackexchange/DUMP.bin.)

Then I disabled the BOOTRST fuse and pressed the restart button on Arduino Uno R3, but the bootloader still blinks with the LED on digital port 13.

How is it possible, that the bootloader is executed after chip restart and BOOTRST is disabled? My fuse settings are: Only SPIEN and BODLEVEL0 are set.

I found that when I disable the BODLEVEL0 fuse, the bootloader turns off. BODLEVEL0 is used to set the brown-out voltage and shouldn't affect bootloader start. Where is the issue?

Answer 1

You don't seem to have filled in all the interrupt vectors. I get this from your .bin file:

00000000 <.sec1>:
       0:       00 c0           rjmp    .+0             ;  0x2
       2:       00 00           nop
       4:       00 00           nop
       6:       00 00           nop
       8:       fc cf           rjmp    .-8             ;  0x2
       a:       fb cf           rjmp    .-10            ;  0x2
       c:       fa cf           rjmp    .-12            ;  0x2
       e:       f9 cf           rjmp    .-14            ;  0x2
      10:       f8 cf           rjmp    .-16            ;  0x2
      12:       f7 cf           rjmp    .-18            ;  0x2
      14:       f6 cf           rjmp    .-20            ;  0x2
      16:       f5 cf           rjmp    .-22            ;  0x2
      18:       ff ff           .word   0xffff  ; ????
      1a:       ff ff           .word   0xffff  ; ????
      1c:       ff ff           .word   0xffff  ; ????
      1e:       ff ff           .word   0xffff  ; ????
      20:       ff ff           .word   0xffff  ; ????
      22:       ff ff           .word   0xffff  ; ????
      24:       ff ff           .word   0xffff  ; ????
...
    7dfc:       ff ff           .word   0xffff  ; ????
    7dfe:       ff ff           .word   0xffff  ; ????
    7e00:       11 24           eor     r1, r1           <--- bootloader
    7e02:       94 b7           in      r25, 0x34       ; 52
    7e04:       14 be           out     0x34, r1        ; 52
    7e06:       89 2f           mov     r24, r25
    7e08:       8d 70           andi    r24, 0x0D       ; 13
    7e0a:       11 f0           breq    .+4             ;  0x7e10
    7e0c:       89 2f           mov     r24, r25
    7e0e:       de d0           rcall   .+444           ;  0x7fcc
    7e10:       85 e0           ldi     r24, 0x05       ; 5

Now I can't quite see the relationship between the brown-out and those higher interrupts, but if one of them happened to fire I think it would waddle its way past all those 0xFF instructions and hit the bootloader.

The Atmega328P has 26 interrupt vectors (including Reset).

Answer 2

Thank you very much!

I found that when I set fuses using program PONYPROG2000 and native RS232 programmer (http://3.bp.blogspot.com/-2D_0B7...) then I read fuses again using another program (eXtreme Burner) and another programmer(USBASP 2.0), there was difference. The PONYPROG2000 probably sets the fuses incorrectly. Now the Atmega328P works correctly yet.