| 1 | @c Copyright 1991, 1992, 1993, 1994, 1995, 1996, 2002 |
| 2 | @c Free Software Foundation, Inc. |
| 3 | @c This is part of the GAS manual. |
| 4 | @c For copying conditions, see the file as.texinfo. |
| 5 | @ifset GENERIC |
| 6 | @page |
| 7 | @node i960-Dependent |
| 8 | @chapter Intel 80960 Dependent Features |
| 9 | @end ifset |
| 10 | @ifclear GENERIC |
| 11 | @node Machine Dependencies |
| 12 | @chapter Intel 80960 Dependent Features |
| 13 | @end ifclear |
| 14 | |
| 15 | @cindex i960 support |
| 16 | @menu |
| 17 | * Options-i960:: i960 Command-line Options |
| 18 | * Floating Point-i960:: Floating Point |
| 19 | * Directives-i960:: i960 Machine Directives |
| 20 | * Opcodes for i960:: i960 Opcodes |
| 21 | @end menu |
| 22 | |
| 23 | @c FIXME! Add Syntax sec with discussion of bitfields here, at least so |
| 24 | @c long as they're not turned on for other machines than 960. |
| 25 | |
| 26 | @node Options-i960 |
| 27 | |
| 28 | @section i960 Command-line Options |
| 29 | |
| 30 | @cindex i960 options |
| 31 | @cindex options, i960 |
| 32 | @table @code |
| 33 | |
| 34 | @cindex i960 architecture options |
| 35 | @cindex architecture options, i960 |
| 36 | @cindex @code{-A} options, i960 |
| 37 | @item -ACA | -ACA_A | -ACB | -ACC | -AKA | -AKB | -AKC | -AMC |
| 38 | Select the 80960 architecture. Instructions or features not supported |
| 39 | by the selected architecture cause fatal errors. |
| 40 | |
| 41 | @samp{-ACA} is equivalent to @samp{-ACA_A}; @samp{-AKC} is equivalent to |
| 42 | @samp{-AMC}. Synonyms are provided for compatibility with other tools. |
| 43 | |
| 44 | If you do not specify any of these options, @code{@value{AS}} generates code |
| 45 | for any instruction or feature that is supported by @emph{some} version of the |
| 46 | 960 (even if this means mixing architectures!). In principle, |
| 47 | @code{@value{AS}} attempts to deduce the minimal sufficient processor type if |
| 48 | none is specified; depending on the object code format, the processor type may |
| 49 | be recorded in the object file. If it is critical that the @code{@value{AS}} |
| 50 | output match a specific architecture, specify that architecture explicitly. |
| 51 | |
| 52 | @cindex @code{-b} option, i960 |
| 53 | @cindex branch recording, i960 |
| 54 | @cindex i960 branch recording |
| 55 | @item -b |
| 56 | Add code to collect information about conditional branches taken, for |
| 57 | later optimization using branch prediction bits. (The conditional branch |
| 58 | instructions have branch prediction bits in the CA, CB, and CC |
| 59 | architectures.) If @var{BR} represents a conditional branch instruction, |
| 60 | the following represents the code generated by the assembler when |
| 61 | @samp{-b} is specified: |
| 62 | |
| 63 | @smallexample |
| 64 | call @var{increment routine} |
| 65 | .word 0 # pre-counter |
| 66 | Label: @var{BR} |
| 67 | call @var{increment routine} |
| 68 | .word 0 # post-counter |
| 69 | @end smallexample |
| 70 | |
| 71 | The counter following a branch records the number of times that branch |
| 72 | was @emph{not} taken; the difference between the two counters is the |
| 73 | number of times the branch @emph{was} taken. |
| 74 | |
| 75 | @cindex @code{gbr960}, i960 postprocessor |
| 76 | @cindex branch statistics table, i960 |
| 77 | A table of every such @code{Label} is also generated, so that the |
| 78 | external postprocessor @code{gbr960} (supplied by Intel) can locate all |
| 79 | the counters. This table is always labeled @samp{__BRANCH_TABLE__}; |
| 80 | this is a local symbol to permit collecting statistics for many separate |
| 81 | object files. The table is word aligned, and begins with a two-word |
| 82 | header. The first word, initialized to 0, is used in maintaining linked |
| 83 | lists of branch tables. The second word is a count of the number of |
| 84 | entries in the table, which follow immediately: each is a word, pointing |
| 85 | to one of the labels illustrated above. |
| 86 | |
| 87 | @c TEXI2ROFF-KILL |
| 88 | @ifinfo |
| 89 | @c END TEXI2ROFF-KILL |
| 90 | @example |
| 91 | +------------+------------+------------+ ... +------------+ |
| 92 | | | | | | | |
| 93 | | *NEXT | COUNT: N | *BRLAB 1 | | *BRLAB N | |
| 94 | | | | | | | |
| 95 | +------------+------------+------------+ ... +------------+ |
| 96 | |
| 97 | __BRANCH_TABLE__ layout |
| 98 | @end example |
| 99 | @c TEXI2ROFF-KILL |
| 100 | @end ifinfo |
| 101 | @need 2000 |
| 102 | @tex |
| 103 | \vskip 1pc |
| 104 | \line{\leftskip=0pt\hskip\tableindent |
| 105 | \boxit{2cm}{\tt *NEXT}\boxit{2cm}{\tt COUNT: \it N}\boxit{2cm}{\tt |
| 106 | *BRLAB 1}\ibox{1cm}{\quad\dots}\boxit{2cm}{\tt *BRLAB \it N}\hfil} |
| 107 | \centerline{\it {\tt \_\_BRANCH\_TABLE\_\_} layout} |
| 108 | @end tex |
| 109 | @c END TEXI2ROFF-KILL |
| 110 | |
| 111 | The first word of the header is used to locate multiple branch tables, |
| 112 | since each object file may contain one. Normally the links are |
| 113 | maintained with a call to an initialization routine, placed at the |
| 114 | beginning of each function in the file. The @sc{gnu} C compiler |
| 115 | generates these calls automatically when you give it a @samp{-b} option. |
| 116 | For further details, see the documentation of @samp{gbr960}. |
| 117 | |
| 118 | @cindex @code{-no-relax} option, i960 |
| 119 | @item -no-relax |
| 120 | Normally, Compare-and-Branch instructions with targets that require |
| 121 | displacements greater than 13 bits (or that have external targets) are |
| 122 | replaced with the corresponding compare (or @samp{chkbit}) and branch |
| 123 | instructions. You can use the @samp{-no-relax} option to specify that |
| 124 | @code{@value{AS}} should generate errors instead, if the target displacement |
| 125 | is larger than 13 bits. |
| 126 | |
| 127 | This option does not affect the Compare-and-Jump instructions; the code |
| 128 | emitted for them is @emph{always} adjusted when necessary (depending on |
| 129 | displacement size), regardless of whether you use @samp{-no-relax}. |
| 130 | @end table |
| 131 | |
| 132 | @node Floating Point-i960 |
| 133 | @section Floating Point |
| 134 | |
| 135 | @cindex floating point, i960 (@sc{ieee}) |
| 136 | @cindex i960 floating point (@sc{ieee}) |
| 137 | @code{@value{AS}} generates @sc{ieee} floating-point numbers for the directives |
| 138 | @samp{.float}, @samp{.double}, @samp{.extended}, and @samp{.single}. |
| 139 | |
| 140 | @node Directives-i960 |
| 141 | @section i960 Machine Directives |
| 142 | |
| 143 | @cindex machine directives, i960 |
| 144 | @cindex i960 machine directives |
| 145 | |
| 146 | @table @code |
| 147 | @cindex @code{bss} directive, i960 |
| 148 | @item .bss @var{symbol}, @var{length}, @var{align} |
| 149 | Reserve @var{length} bytes in the bss section for a local @var{symbol}, |
| 150 | aligned to the power of two specified by @var{align}. @var{length} and |
| 151 | @var{align} must be positive absolute expressions. This directive |
| 152 | differs from @samp{.lcomm} only in that it permits you to specify |
| 153 | an alignment. @xref{Lcomm,,@code{.lcomm}}. |
| 154 | @end table |
| 155 | |
| 156 | @table @code |
| 157 | @cindex @code{extended} directive, i960 |
| 158 | @item .extended @var{flonums} |
| 159 | @code{.extended} expects zero or more flonums, separated by commas; for |
| 160 | each flonum, @samp{.extended} emits an @sc{ieee} extended-format (80-bit) |
| 161 | floating-point number. |
| 162 | |
| 163 | @cindex @code{leafproc} directive, i960 |
| 164 | @item .leafproc @var{call-lab}, @var{bal-lab} |
| 165 | You can use the @samp{.leafproc} directive in conjunction with the |
| 166 | optimized @code{callj} instruction to enable faster calls of leaf |
| 167 | procedures. If a procedure is known to call no other procedures, you |
| 168 | may define an entry point that skips procedure prolog code (and that does |
| 169 | not depend on system-supplied saved context), and declare it as the |
| 170 | @var{bal-lab} using @samp{.leafproc}. If the procedure also has an |
| 171 | entry point that goes through the normal prolog, you can specify that |
| 172 | entry point as @var{call-lab}. |
| 173 | |
| 174 | A @samp{.leafproc} declaration is meant for use in conjunction with the |
| 175 | optimized call instruction @samp{callj}; the directive records the data |
| 176 | needed later to choose between converting the @samp{callj} into a |
| 177 | @code{bal} or a @code{call}. |
| 178 | |
| 179 | @var{call-lab} is optional; if only one argument is present, or if the |
| 180 | two arguments are identical, the single argument is assumed to be the |
| 181 | @code{bal} entry point. |
| 182 | |
| 183 | @cindex @code{sysproc} directive, i960 |
| 184 | @item .sysproc @var{name}, @var{index} |
| 185 | The @samp{.sysproc} directive defines a name for a system procedure. |
| 186 | After you define it using @samp{.sysproc}, you can use @var{name} to |
| 187 | refer to the system procedure identified by @var{index} when calling |
| 188 | procedures with the optimized call instruction @samp{callj}. |
| 189 | |
| 190 | Both arguments are required; @var{index} must be between 0 and 31 |
| 191 | (inclusive). |
| 192 | @end table |
| 193 | |
| 194 | @node Opcodes for i960 |
| 195 | @section i960 Opcodes |
| 196 | |
| 197 | @cindex opcodes, i960 |
| 198 | @cindex i960 opcodes |
| 199 | All Intel 960 machine instructions are supported; |
| 200 | @pxref{Options-i960,,i960 Command-line Options} for a discussion of |
| 201 | selecting the instruction subset for a particular 960 |
| 202 | architecture.@refill |
| 203 | |
| 204 | Some opcodes are processed beyond simply emitting a single corresponding |
| 205 | instruction: @samp{callj}, and Compare-and-Branch or Compare-and-Jump |
| 206 | instructions with target displacements larger than 13 bits. |
| 207 | |
| 208 | @menu |
| 209 | * callj-i960:: @code{callj} |
| 210 | * Compare-and-branch-i960:: Compare-and-Branch |
| 211 | @end menu |
| 212 | |
| 213 | @node callj-i960 |
| 214 | @subsection @code{callj} |
| 215 | |
| 216 | @cindex @code{callj}, i960 pseudo-opcode |
| 217 | @cindex i960 @code{callj} pseudo-opcode |
| 218 | You can write @code{callj} to have the assembler or the linker determine |
| 219 | the most appropriate form of subroutine call: @samp{call}, |
| 220 | @samp{bal}, or @samp{calls}. If the assembly source contains |
| 221 | enough information---a @samp{.leafproc} or @samp{.sysproc} directive |
| 222 | defining the operand---then @code{@value{AS}} translates the |
| 223 | @code{callj}; if not, it simply emits the @code{callj}, leaving it |
| 224 | for the linker to resolve. |
| 225 | |
| 226 | @node Compare-and-branch-i960 |
| 227 | @subsection Compare-and-Branch |
| 228 | |
| 229 | @cindex i960 compare/branch instructions |
| 230 | @cindex compare/branch instructions, i960 |
| 231 | The 960 architectures provide combined Compare-and-Branch instructions |
| 232 | that permit you to store the branch target in the lower 13 bits of the |
| 233 | instruction word itself. However, if you specify a branch target far |
| 234 | enough away that its address won't fit in 13 bits, the assembler can |
| 235 | either issue an error, or convert your Compare-and-Branch instruction |
| 236 | into separate instructions to do the compare and the branch. |
| 237 | |
| 238 | @cindex compare and jump expansions, i960 |
| 239 | @cindex i960 compare and jump expansions |
| 240 | Whether @code{@value{AS}} gives an error or expands the instruction depends |
| 241 | on two choices you can make: whether you use the @samp{-no-relax} option, |
| 242 | and whether you use a ``Compare and Branch'' instruction or a ``Compare |
| 243 | and Jump'' instruction. The ``Jump'' instructions are @emph{always} |
| 244 | expanded if necessary; the ``Branch'' instructions are expanded when |
| 245 | necessary @emph{unless} you specify @code{-no-relax}---in which case |
| 246 | @code{@value{AS}} gives an error instead. |
| 247 | |
| 248 | These are the Compare-and-Branch instructions, their ``Jump'' variants, |
| 249 | and the instruction pairs they may expand into: |
| 250 | |
| 251 | @c TEXI2ROFF-KILL |
| 252 | @ifinfo |
| 253 | @c END TEXI2ROFF-KILL |
| 254 | @example |
| 255 | Compare and |
| 256 | Branch Jump Expanded to |
| 257 | ------ ------ ------------ |
| 258 | bbc chkbit; bno |
| 259 | bbs chkbit; bo |
| 260 | cmpibe cmpije cmpi; be |
| 261 | cmpibg cmpijg cmpi; bg |
| 262 | cmpibge cmpijge cmpi; bge |
| 263 | cmpibl cmpijl cmpi; bl |
| 264 | cmpible cmpijle cmpi; ble |
| 265 | cmpibno cmpijno cmpi; bno |
| 266 | cmpibne cmpijne cmpi; bne |
| 267 | cmpibo cmpijo cmpi; bo |
| 268 | cmpobe cmpoje cmpo; be |
| 269 | cmpobg cmpojg cmpo; bg |
| 270 | cmpobge cmpojge cmpo; bge |
| 271 | cmpobl cmpojl cmpo; bl |
| 272 | cmpoble cmpojle cmpo; ble |
| 273 | cmpobne cmpojne cmpo; bne |
| 274 | @end example |
| 275 | @c TEXI2ROFF-KILL |
| 276 | @end ifinfo |
| 277 | @tex |
| 278 | \hskip\tableindent |
| 279 | \halign{\hfil {\tt #}\quad&\hfil {\tt #}\qquad&{\tt #}\hfil\cr |
| 280 | \omit{\hfil\it Compare and\hfil}\span\omit&\cr |
| 281 | {\it Branch}&{\it Jump}&{\it Expanded to}\cr |
| 282 | bbc& & chkbit; bno\cr |
| 283 | bbs& & chkbit; bo\cr |
| 284 | cmpibe& cmpije& cmpi; be\cr |
| 285 | cmpibg& cmpijg& cmpi; bg\cr |
| 286 | cmpibge& cmpijge& cmpi; bge\cr |
| 287 | cmpibl& cmpijl& cmpi; bl\cr |
| 288 | cmpible& cmpijle& cmpi; ble\cr |
| 289 | cmpibno& cmpijno& cmpi; bno\cr |
| 290 | cmpibne& cmpijne& cmpi; bne\cr |
| 291 | cmpibo& cmpijo& cmpi; bo\cr |
| 292 | cmpobe& cmpoje& cmpo; be\cr |
| 293 | cmpobg& cmpojg& cmpo; bg\cr |
| 294 | cmpobge& cmpojge& cmpo; bge\cr |
| 295 | cmpobl& cmpojl& cmpo; bl\cr |
| 296 | cmpoble& cmpojle& cmpo; ble\cr |
| 297 | cmpobne& cmpojne& cmpo; bne\cr} |
| 298 | @end tex |
| 299 | @c END TEXI2ROFF-KILL |