Re: grub thing

[Aleph One <amerei@xxxxxxxxx>]

On Wed, 1 Mar 2006 15:42:50 +0100
Marcin Kościelnicki <markosc@xxxxxxxxxx> wrote:

~> i'm looking at a disassembly of grub's stage 1. and i wanna know why it
~> need's to CLI before setting up the environment??
~
~It's because it sets the stack address -- otherwise, interrupt could occur 
~between mov to ss and mov to sp, which would use invalid stack address and be 
~lethal.

ahh! i see i see. it's for protection.

~
~Also, it has NOTHING to do with lack of interrupt handlers -- BIOS has 
~handlers for all interrupts needed for bootstrap, and the others are 
~disabled. No bad thing can happen. In fact, it re-enables interrupts right 
~after setting the stack (sti)
~
~> i tried to interpret it 
~> line by line and this is what i have so far
~>
~>  00007C4A  FA			cli			; disable interrupt
~>  00007C4B  80CA80		or dl,0x80          ; correct boot drive byte
~>  00007C4E  EA537C0000	jmp 0x0:0x7c53	; jmp to next inst
~>  00007C53  31C0		xor ax,ax           ; zero ax
~>  00007C55  8ED8		mov ds,ax		; zero data segment
~>  00007C57  8ED0		mov ss,ax		; zero stack segment
~>  00007C59  BC0020		mov sp,0x2000	; 8 kilobytes
~>  00007C5C  FB			sti			; enable interrupt
~>
~> also, since the origin is at 0000:7c00h i figured that must imply that "mov
~> sp, 0x2000" means that the stack starts at 0x7c00+0x2000 = 9c000 right?
~
~No.
~
~>  00007C57  8ED0		mov ss,ax		; zero stack segment
~>  00007C59  BC0020		mov sp,0x2000	; 8 kilobytes
~
~Here, ss==0 and sp==0x2000. Stack starts at ss*0x10+sp == 0x0*0x10+0x2000 == 
~0x2000. Also, remember it grows down.

one note. why must (ss*0x10)?

~
~----------------------------------------------------------------------
~Kliknij po wiecej! >>> http://link.interia.pl/f18ed
~
~-
~: send the line "unsubscribe linux-assembly" in
~the body of a message to majordomo@xxxxxxxxxxxxxxx
~More majordomo info at  http://vger.kernel.org/majordomo-info.html
-
: send the line "unsubscribe linux-assembly" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html