26 80x86 cpu module

This chapter documents the Backend for the 80x86 microprocessor family.

26.1 Legal

This module is written in 2005-2006,2011,2015-2016 by Frank Wille and is covered by the vasm copyright without modifications.

26.2 Additional options for this module

This module provides the following additional options:

‘-cpudebug=<n>’: Enables debugging output.
‘-m8086’: Generate code for the 8086 CPU.
‘-mi186’: Generate code for the 80186 CPU.
‘-mi286’: Generate code for the 80286 CPU.
‘-mi386’: Generate code for the 80386 CPU.
‘-mi486’: Generate code for the 80486 CPU.
‘-mi586’: Generate code for the Pentium.
‘-mi686’: Generate code for the PentiumPro.
‘-mpentium’: Generate code for the Pentium.
‘-mpentiumpro’: Generate code for the PentiumPro.
‘-mk6’: Generate code for the AMD K6.
‘-mathlon’: Generate code for the AMD Athlon.
‘-msledgehammer’: Generate code for the Sledgehammer CPU.
‘-m64’: Generate code for 64-bit architectures (x86_64).

26.3 General

This backend accepts 80x86 instructions as described in the Intel Architecture Software Developer’s Manual.

The target address type is 32 bits. It is 64 bits when the x86_64 architecture was selected (‘-m64’). Floating point constants in instructions and data are supported and encoded in IEEE format.

Instructions do not need any alignment. Data is aligned to its natural alignment by default.

The backend uses AT&T-syntax! This means the left operands are always the source and the right operand is the destination. Register names have to be prefixed by a ’%’.

The operation size is indicated by a ’b’, ’w’, ’l’, etc. suffix directly appended to the mnemonic. The assembler can also determine the operation size from the size of the registers being used.

26.4 Extensions

Predefined register symbols in this backend:

- 8-bit registers: al cl dl bl ah ch dh bh axl cxl dxl spl bpl sil dil r8b r9b r10b r11b r12b r13b r14b r15b
- 16-bit registers: ax cx dx bx sp bp si di r8w r9w r10w r11w r12w r13w r14w r15w
- 32-bit registers: eax ecx edx ebx esp ebp esi edi r8d r9d r10d r11d r12d r13d r14d r15d
- 64-bit registers: rax rcx rdx rbx rsp ebp rsi rdi r8 r9 r10 r11 r12 r13 r14 r15
- segment registers: es cs ss ds fs gs
- control registers: cr0 cr1 cr2 cr3 cr4 cr5 cr6 cr7 cr8 cr9 cr10 cr11 cr12 cr13 cr14 cr15
- debug registers: dr0 dr1 dr2 dr3 dr4 dr5 dr6 dr7 dr8 dr9 dr10 dr11 dr12 dr13 dr14 dr15
- test registers: tr0 tr1 tr2 tr3 tr4 tr5 tr6 tr7
- MMX and SIMD registers: mm0 mm1 mm2 mm3 mm4 mm5 mm6 mm7 xmm0 xmm1 xmm2 xmm3 xmm4 xmm5 xmm6 xmm7 xmm8 xmm9 xmm10 xmm11 xmm12 xmm13 xmm14 xmm15
- FPU registers: st st(0) st(1) st(2) st(3) st(4) st(5) st(6) st(7)

This backend extends the selected syntax module by the following directives:

.code16: Sets the assembler to 16-bit addressing mode.
.code32: Sets the assembler to 32-bit addressing mode, which is the default.
.code64: Sets the assembler to 64-bit addressing mode.

26.5 Optimizations

This backend performs the following optimizations:

- Immediate operands are optimized to the smallest size which can still represent the absolute value.
- Displacement operands are optimized to the smallest size which can still represent the absolute value.
- Jump instructions are optimized to 8-bit displacements, when possible.

26.6 Known Problems

Some known problems of this module at the moment:

- 64-bit operations are incomplete and experimental.

26.7 Error Messages

This module has the following error messages:

- 2001: instruction not supported on selected architecture
- 2002: trailing garbage in operand
- 2003: same type of prefix used twice
- 2004: immediate operand illegal with absolute jump
- 2005: base register expected
- 2006: scale factor without index register
- 2007: missing ’)’ in baseindex addressing mode
- 2008: redundant %s prefix ignored
- 2009: unknown register specified
- 2010: using register %%%s instead of %%%s due to ’%c’ suffix
- 2011: %%%s not allowed with ’%c’ suffix
- 2012: illegal suffix ’%c’
- 2013: instruction has no suffix and no register operands - size is unknown
- 2015: memory operand expected
- 2016: you cannot pop %%%s
- 2017: translating to %s %%%s,%%%s
- 2018: translating to %s %%%s
- 2019: absolute scale factor required
- 2020: illegal scale factor (valid: 1,2,4,8)
- 2021: data objects with %d bits size are not supported
- 2022: need at least %d bits for a relocatable symbol
- 2023: pc-relative jump destination out of range (%lld)
- 2024: instruction doesn’t support these operand sizes
- 2025: cannot determine immediate operand size without a suffix
- 2026: displacement doesn’t fit into %d bits