.\" $Header$
.TH I386_AS 6ACK
.ad
.SH NAME
i386_as \- assembler for Intel 80386
.SH SYNOPSIS
~em/lib/i386/as [options] argument ...
.SH DESCRIPTION
This assembler is made with the general framework
described in \fIuni_ass\fP(6). It is an assembler generating relocatable
object code in \fIack.out\fP(5) format.
.SH SYNTAX
.IP segments
An address on the Intel 80386 consists of two pieces:
a segment number and an offset.
Usually, the segment number resides in a segment register, and
assembly language addresses only give the offset, with the exception of
the address of an inter-segment jump or call (see \fIaddressing modes\fP
below).
.IP registers
The Intel 80386 has the following 32-bit registers:
.br
Four general registers: eax (accumulator), ebx (base), ecx (count), and edx (data).
The low- and high order bytes of the low order words of these registers
are separately addressable as ah, bh, ch, dh, and al, bl, cl, dl respectively.
.br
Two pointer registers: esp (stack pointer) and ebp (base pointer).
.br
Two index registers: esi (source index) and edi (destination index).
.br
Six segment registers: cs (code), ds (data), ss (stack), es (extra),
fs (extra), and gs (extra).
.IP "addressing modes"
.nf
.ta 8n 16n 24n 32n 40n 48n
syntax		meaning

expr		the value of \fIexpr\fP is immediate data or
		an address offset. There is no special
		notation for immediate data.

register	one of the aforementioned general registers
		or their upper or lower halves, or one of the
		four segment registers.

(expr)		the value of expr is the address of the operand.

(reg)
expr (reg)	the value of \fIexpr\fP (if present) + the contents of
		\fIreg\fP (which must be a pointer or an index register)
		is the address of the operand.

(reg1) (reg2)
expr (reg1) (reg2)
		the value of \fIexpr\fP (if present) + the contents of
		\fIreg1\fP + the
		contents of \fIreg2\fP is the address of the operand.

(reg1) (reg2 * scale)
expr (reg1) (reg2 * scale)
		the value of \fIexpr\fP (if present) + the contents of
		\fIreg1\fP + the
		contents of \fIreg2\fP multiplied by \fIscale\fP,
		is the address of the operand.
		\fIscale\fP can be either 1, 2, 4, or 8.
		This mode is only allowed for 32-bit addressing.

The next addressing mode is only allowed with the instructions
"callf" or "jmpf".

expr : expr	the value of the first \fIexpr\fP is a segment number,
		the value of the second \fIexpr\fP is an address offset.

The following two addressing modes are only allowed with Intel 80[23]87 floating
point processor instructions:

st
st(num)		addresses the floating point processor stack. \fInum\fP
		must be between 0 and 7. st is the same as st(0).

.fi

.IP prefixes
Each time an address is computed the processor decides which segment register
to use. You can override the processor's choice by prefixing the instruction
with one of eseg, cseg, sseg, dseg, fseg, or gseg; these prefixes indicate that the
processor should choose es, cs, ss, ds, fs, or gs instead.
.br
Example: 
.ti +8
dseg movs
.IP ""
There is also an address size toggle, which switches between 32-bit and
16-bit address generation: a16 or a32. Normally, the assembler generates
32-bit addresses; both of these toggles make it generate 16-bit addresses
for the next instruction, and also generate code to set the processor 
temporarily in 16-bit address mode.
.IP ""
There is also an operand size toggle, which switches between 32-bit and
16-bit operands: o16 or o32. Normally, the assembler generates
32-bit operands; both of these toggles make it generate 16-bit operands
for the next instruction, and also generate code to set the processor 
temporarily in 16-bit operand mode.
.IP ""
Prefixes only affect the next instruction.
.IP ""
There are also the .use32 and .use16 assembler directives, which do not
generate code, but change the default for operand and address sizes.
Obviously, .use16 gives 16-bit modes, .use32 gives 32-bit modes.
This is useful for assembling real mode 80386 code, or pure 16-bit
modules (that do not have the D-bit set in the segment descriptor).
.SH "SEE ALSO"
uni_ass(6),
ack(1),
ack.out(5),
.br
80386 Programmer's Reference Manual, 1986, Intel Corporation