CODEWITHSUNDEEP
assemblyx86-16bootloaderas86
systolicDrake
systolicDrake

Reputation: 101

x86 assembly mov instruction LILO

I was reading the source code of LILO for a project and I stumbled across this line

mov dh, [d_dev](bp)

I wished to know what the mov instruction is doing here, I know that if it is 

mov dh, [d_dev]

then the value pointed by d_dev is placed in dh but what happens with the (bp).

Any help would be appreciated.

Source Link: https://github.com/a2o/lilo/blob/master/src/first.S line 205

Upvotes: 0

Views: 125

Answers (1)

Peter Cordes
Peter Cordes

Reputation: 363980

LILO still uses AS86 (note the get common.s /* as86 "include" will bypass the CPP */) line at the top.

AS86 apparently has op dst, src operand order, but memory-operand syntax looks like a cross between AT&T and Intel. [d_dev](bp) is AT&T d_dev(%bp) or NASM [d_dev + bp], i.e. base register = BP, with the address of d_dev as a disp8 or disp16.

An earlier line in the same file zeros BP:

xor     bp,bp       ! shorted addressing

Presumably d_dev is an offset that fits in a signed 8-bit displacement. Yes, the label appears pretty soon after a .org 6, so its address is a small displacement, and mov dh, [bp + disp8] is only a 3 byte instruction, vs. mov dh, [disp16] being a 4 byte instruction (opcode + modrm + disp16).

So mov dh, [d_dev](bp) does the same thing as mov dh, [d_dev], but in one less byte of machine code, because BP=0.

Upvotes: 1

Related Questions