1 @c Copyright 2000, 2003 Free Software Foundation, Inc.
2 @c This is part of the GAS manual.
3 @c For copying conditions, see the file as.texinfo.
7 @chapter Intel i860 Dependent Features
10 @node Machine Dependencies
11 @chapter Intel i860 Dependent Features
15 @c FIXME: This is basically a stub for i860. There is tons more information
16 that I will add later (jle@cygnus.com).
21 * Notes-i860:: i860 Notes
22 * Options-i860:: i860 Command-line Options
23 * Directives-i860:: i860 Machine Directives
24 * Opcodes for i860:: i860 Opcodes
29 This is a fairly complete i860 assembler which is compatible with the
30 UNIX System V/860 Release 4 assembler. However, it does not currently
31 support SVR4 PIC (i.e., @code{@@GOT, @@GOTOFF, @@PLT}).
33 Like the SVR4/860 assembler, the output object format is ELF32. Currently,
34 this is the only supported object format. If there is sufficient interest,
35 other formats such as COFF may be implemented.
37 Both the Intel and AT&T/SVR4 syntaxes are supported, with the latter
38 being the default. One difference is that AT&T syntax requires the '%'
39 prefix on register names while Intel syntax does not. Another difference
40 is in the specification of relocatable expressions. The Intel syntax
41 is @code{ha%expression} whereas the SVR4 syntax is @code{[expression]@@ha}
42 (and similarly for the "l" and "h" selectors).
44 @section i860 Command-line Options
45 @subsection SVR4 compatibility options
48 Print assembler version.
54 @subsection Other options
57 Select little endian output (this is the default).
59 Select big endian output. Note that the i860 always reads instructions
60 as little endian data, so this option only effects data and not
63 Emit a warning message if any pseudo-instruction expansions occurred.
64 For example, a @code{or} instruction with an immediate larger than 16-bits
65 will be expanded into two instructions. This is a very undesirable feature to
66 rely on, so this flag can help detect any code where it happens. One
67 use of it, for instance, has been to find and eliminate any place
68 where @code{gcc} may emit these pseudo-instructions.
70 Enable support for the i860XP instructions and control registers. By default,
71 this option is disabled so that only the base instruction set (i.e., i860XR)
74 The i860 assembler defaults to AT&T/SVR4 syntax. This option enables the
79 @section i860 Machine Directives
81 @cindex machine directives, i860
82 @cindex i860 machine directives
85 @cindex @code{dual} directive, i860
87 Enter dual instruction mode. While this directive is supported, the
88 preferred way to use dual instruction mode is to explicitly code
89 the dual bit with the @code{d.} prefix.
93 @cindex @code{enddual} directive, i860
95 Exit dual instruction mode. While this directive is supported, the
96 preferred way to use dual instruction mode is to explicitly code
97 the dual bit with the @code{d.} prefix.
101 @cindex @code{atmp} directive, i860
103 Change the temporary register used when expanding pseudo operations. The
104 default register is @code{r31}.
107 The @code{.dual}, @code{.enddual}, and @code{.atmp} directives are available only in the Intel syntax mode.
109 Both syntaxes allow for the standard @code{.align} directive. However,
110 the Intel syntax additionally allows keywords for the alignment
111 parameter: "@code{.align type}", where `type' is one of @code{.short}, @code{.long},
112 @code{.quad}, @code{.single}, @code{.double} representing alignments of 2, 4,
113 16, 4, and 8, respectively.
115 @node Opcodes for i860
116 @section i860 Opcodes
118 @cindex opcodes, i860
120 All of the Intel i860XR and i860XP machine instructions are supported. Please see
121 either @emph{i860 Microprocessor Programmer's Reference Manual} or @emph{i860 Microprocessor Architecture} for more information.
122 @subsection Other instruction support (pseudo-instructions)
123 For compatibility with some other i860 assemblers, a number of
124 pseudo-instructions are supported. While these are supported, they are
125 a very undesirable feature that should be avoided -- in particular, when
126 they result in an expansion to multiple actual i860 instructions. Below
127 are the pseudo-instructions that result in expansions.
129 @item Load large immediate into general register:
131 The pseudo-instruction @code{mov imm,%rn} (where the immediate does
132 not fit within a signed 16-bit field) will be expanded into:
134 orh large_imm@@h,%r0,%rn
135 or large_imm@@l,%rn,%rn
137 @item Load/store with relocatable address expression:
139 For example, the pseudo-instruction @code{ld.b addr_exp(%rx),%rn}
140 will be expanded into:
142 orh addr_exp@@ha,%rx,%r31
143 ld.l addr_exp@@l(%r31),%rn
146 The analogous expansions apply to @code{ld.x, st.x, fld.x, pfld.x, fst.x}, and @code{pst.x} as well.
147 @item Signed large immediate with add/subtract:
149 If any of the arithmetic operations @code{adds, addu, subs, subu} are used
150 with an immediate larger than 16-bits (signed), then they will be expanded.
151 For instance, the pseudo-instruction @code{adds large_imm,%rx,%rn} expands to:
153 orh large_imm@@h,%r0,%r31
154 or large_imm@@l,%r31,%r31
157 @item Unsigned large immediate with logical operations:
159 Logical operations (@code{or, andnot, or, xor}) also result in expansions.
160 The pseudo-instruction @code{or large_imm,%rx,%rn} results in:
162 orh large_imm@@h,%rx,%r31
163 or large_imm@@l,%r31,%rn
166 Similarly for the others, except for @code{and} which expands to:
168 andnot (-1 - large_imm)@@h,%rx,%r31
169 andnot (-1 - large_imm)@@l,%r31,%rn