Re: [avr-gcc-list] Interrupt vector defenition with avr-libc

[Theodore A. Roth]

On Tue, 10 Dec 2002, Jeff Rose wrote:

:) Hi guys,
:)      Does anyone know how the SIGNAL macro works in avr-libc?  It
:) looks like the #define just outputs __vector_# where number is the number
:) for the vector.  When I disassemble my final elf file the vector table is
:) still empty, (All entries point to __bad_interrupt) and the label before
:) my signal handler is __vector_11.  (Its the timer0 overflow interrupt)
:)      At what stage is the vector table supposed to be filled out?  Has
:) anyone used the gcc and avr-libc with the atmega323?

It works fine for me. Here's some of the disassm from a uart test program
I wrote (compiled for m323 with today's binutils, gcc and avr-libc from
cvs):

Disassembly of section .text:

00000000 <__vectors>:
   0:   0c 94 28 00     jmp     0x50
   4:   0c 94 43 00     jmp     0x86
   8:   0c 94 43 00     jmp     0x86
   c:   0c 94 43 00     jmp     0x86
  10:   0c 94 43 00     jmp     0x86
  14:   0c 94 43 00     jmp     0x86
  18:   0c 94 43 00     jmp     0x86
  1c:   0c 94 43 00     jmp     0x86
  20:   0c 94 43 00     jmp     0x86
  24:   0c 94 43 00     jmp     0x86
  28:   0c 94 43 00     jmp     0x86
  2c:   0c 94 60 01     jmp     0x2c0
  30:   0c 94 43 00     jmp     0x86
  34:   0c 94 aa 00     jmp     0x154
  38:   0c 94 f4 00     jmp     0x1e8
  3c:   0c 94 43 00     jmp     0x86
  40:   0c 94 43 00     jmp     0x86
  44:   0c 94 43 00     jmp     0x86
  48:   0c 94 43 00     jmp     0x86
  4c:   0c 94 43 00     jmp     0x86

00000050 <__ctors_end>:

<snip>

000002c0 <__vector_11>:
    }

    return(0);
}

SIGNAL (SIG_OVERFLOW0)
{
 2c0:   1f 92           push    r1
 2c2:   0f 92           push    r0
 2c4:   0f b6           in      r0, 0x3f        ; 63
 2c6:   0f 92           push    r0
 2c8:   11 24           eor     r1, r1
}
 2ca:   0f 90           pop     r0
 2cc:   0f be           out     0x3f, r0        ; 63
 2ce:   0f 90           pop     r0
 2d0:   1f 90           pop     r1
 2d2:   18 95           reti


My guess is that you mis-speeled SIG_OVERFLOW0. If I do that, here's what
the disassm looks like:

Disassembly of section .text:

00000000 <__vectors>:
   0:   0c 94 28 00     jmp     0x50
   4:   0c 94 43 00     jmp     0x86
   8:   0c 94 43 00     jmp     0x86
   c:   0c 94 43 00     jmp     0x86
  10:   0c 94 43 00     jmp     0x86
  14:   0c 94 43 00     jmp     0x86
  18:   0c 94 43 00     jmp     0x86
  1c:   0c 94 43 00     jmp     0x86
  20:   0c 94 43 00     jmp     0x86
  24:   0c 94 43 00     jmp     0x86
  28:   0c 94 43 00     jmp     0x86
  2c:   0c 94 43 00     jmp     0x86
  30:   0c 94 43 00     jmp     0x86
  34:   0c 94 aa 00     jmp     0x154
  38:   0c 94 f4 00     jmp     0x1e8
  3c:   0c 94 43 00     jmp     0x86
  40:   0c 94 43 00     jmp     0x86
  44:   0c 94 43 00     jmp     0x86
  48:   0c 94 43 00     jmp     0x86
  4c:   0c 94 43 00     jmp     0x86

<snip>

000002c0 <SIG_a_speeling_err_OVERFLOW0>:
    }

    return(0);
}

SIGNAL (SIG_a_speeling_err_OVERFLOW0)
{
 2c0:   1f 92           push    r1
 2c2:   0f 92           push    r0
 2c4:   0f b6           in      r0, 0x3f        ; 63
 2c6:   0f 92           push    r0
 2c8:   11 24           eor     r1, r1
}
 2ca:   0f 90           pop     r0
 2cc:   0f be           out     0x3f, r0        ; 63
 2ce:   0f 90           pop     r0
 2d0:   1f 90           pop     r1
 2d2:   18 95           reti


Ted Roth

avr-gcc-list at http://avr1.org