On May 1, 8:43 am, Frank Kotler <spamt...@[EMAIL PROTECTED]
> wrote:
> leilei wrote:
> > I am writting a program for target board which have a 486 cpu, 512K
> > ram(0x0 to 0x7ffff), 512k flash (0x80000 to 0xFFFFF).My program will
> > be burned into flash.
> > My program is to initialize the GDT, IDT, TSS, move them to memory.
> > Now I can enter protected model and mov GDT, IDT correctly.But when I
> > am about to mov TSS, some exception came out, and the CPU reset
> > automaticly. The code casue the problem is like this:
> > mov cx, gdt_idx
> > mov gs, cx
> > when cpu run to the instuction 'mov gs, cx', CPU will reset.
> > i can assure the value in cx is correctly.
>
> Your system appears to have a different opinion. Better show us how you
> calculate "gdt_idx".
>
> I wouldn't "expect" the system to reboot until you try to *use* gs. Does
> it still reboot if you put "here: jmp here" right after "mov gs, cx"? I
> suppose you've tried that... Only thing I can think of...
>
> Best,
> Frank
Hi, I calculate the gdt_idx like this:
gdt_idx equ 1 * SIZEOF(DESC) ;DESC is a struct of Segment
Descriptor.
....
mov cx, gdt_idx
mov gs, cx
;CPU can not even run to here after 'mov gs, cx', because it already
restarted.
I tried to change gs to es, but it still not work.
I even use a valide selector which executed successfully before to
replace gdt_idx, the CPU still reboot at this place.
|
