Re: Interrupt Handling

Li-Ta Lo  writes:
> Eric,
>
> Why there are two ways for interrupt handling in c_start.S

There are two ways for hardware exception handling.  Interrupts
are not handled.  And that is because on x86 exception handlers
don't always give you an error code.
> vec9:
>         pushl   $0 /* error code */
>         pushl   $9 /* vector */
>         jmp int_hand
>
> vec10:
>         /* error code */
>         pushl   $10 /* vector */
>         jmp     int_hand
>         .word   0x9090
>
> Vector 9 push the error code and vector but vector 10 only push the 
> vector number.
>
> BTW, are those handler actually used ? The idt is uninited.

This hunk below initializes it.  It seems to work out better to initialize the idt
with code instead of data.

/* Initialize the Interrupt Descriptor table */
        leal    _idt, %edi
        leal    vec0, %ebx
        movl    $(0x10 << 16), %eax     /* cs selector */

1:      movw    %bx, %ax
        movl    %ebx, %edx
        movw    $0x8E00, %dx            /* Interrupt gate - dpl=0, present */
        movl    %eax, 0(%edi)
        movl    %edx, 4(%edi)
        addl    $6, %ebx
        addl    $8, %edi
        cmpl    $_idt_end, %edi
        jne     1b

/* Load the Interrupt descriptor table */
        lidt    idtarg

Eric

Interrupt Handling

[Stefan Reinauer]

* Eric W. Biederman  [041103 00:26]:
> This hunk below initializes it.  It seems to work out better to initialize the idt
> with code instead of data.
>    /* Initialize the Interrupt Descriptor table */
>     leal    _idt, %edi
>     leal    vec0, %ebx
>     movl    $(0x10 << 16), %eax     /* cs selector */
>
> 1:      movw    %bx, %ax
>         movl    %ebx, %edx
>         movw    $0x8E00, %dx            /* Interrupt gate - dpl=0, present */
>         movl    %eax, 0(%edi)
>         movl    %edx, 4(%edi)
>         addl    $6, %ebx
>         addl    $8, %edi
>         cmpl    $_idt_end, %edi
>         jne     1b
>     /* Load the Interrupt descriptor table */
>     lidt    idtarg

What would btw be the correct way of resuming operation after, say, a
division by zero exception? I tried to set a new info->eip in
x86_exception() to a helper function, but I get the honour of a GPF
right after returning from that function... finding it jumps into the
stack at some point.

Stefan