1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
5 <title>TableGen Fundamentals</title>
6 <link rel="stylesheet" href="llvm.css" type="text/css">
10 <div class="doc_title">TableGen Fundamentals</div>
12 <div class="doc_text">
14 <li><a href="#introduction">Introduction</a>
16 <li><a href="#concepts">Basic concepts</a></li>
17 <li><a href="#example">An example record</a></li>
18 <li><a href="#running">Running TableGen</a></li>
20 <li><a href="#syntax">TableGen syntax</a>
22 <li><a href="#primitives">TableGen primitives</a>
24 <li><a href="#comments">TableGen comments</a></li>
25 <li><a href="#types">The TableGen type system</a></li>
26 <li><a href="#values">TableGen values and expressions</a></li>
28 <li><a href="#classesdefs">Classes and definitions</a>
30 <li><a href="#valuedef">Value definitions</a></li>
31 <li><a href="#recordlet">'let' expressions</a></li>
32 <li><a href="#templateargs">Class template arguments</a></li>
33 <li><a href="#multiclass">Multiclass definitions and instances</a></li>
35 <li><a href="#filescope">File scope entities</a>
37 <li><a href="#include">File inclusion</a></li>
38 <li><a href="#globallet">'let' expressions</a></li>
41 <li><a href="#backends">TableGen backends</a>
43 <li><a href="#">todo</a></li>
48 <div class="doc_author">
49 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a></p>
52 <!-- *********************************************************************** -->
53 <div class="doc_section"><a name="introduction">Introduction</a></div>
54 <!-- *********************************************************************** -->
56 <div class="doc_text">
58 <p>TableGen's purpose is to help a human develop and maintain records of
59 domain-specific information. Because there may be a large number of these
60 records, it is specifically designed to allow writing flexible descriptions and
61 for common features of these records to be factored out. This reduces the
62 amount of duplication in the description, reduces the chance of error, and
63 makes it easier to structure domain specific information.</p>
65 <p>The core part of TableGen <a href="#syntax">parses a file</a>, instantiates
66 the declarations, and hands the result off to a domain-specific "<a
67 href="#backends">TableGen backend</a>" for processing. The current major user
68 of TableGen is the <a href="CodeGenerator.html">LLVM code generator</a>.</p>
70 <p>Note that if you work on TableGen much, and use emacs or vim, that you can
71 find an emacs "TableGen mode" and a vim language file in the
72 <tt>llvm/utils/emacs</tt> and <tt>llvm/utils/vim</tt> directories of your LLVM
73 distribution, respectively.</p>
77 <!-- ======================================================================= -->
78 <div class="doc_subsection"><a name="concepts">Basic concepts</a></div>
80 <div class="doc_text">
82 <p>TableGen files consist of two key parts: 'classes' and 'definitions', both
83 of which are considered 'records'.</p>
85 <p><b>TableGen records</b> have a unique name, a list of values, and a list of
86 superclasses. The list of values is the main data that TableGen builds for each
87 record; it is this that holds the domain specific information for the
88 application. The interpretation of this data is left to a specific <a
89 href="#backends">TableGen backend</a>, but the structure and format rules are
90 taken care of and are fixed by TableGen.</p>
92 <p><b>TableGen definitions</b> are the concrete form of 'records'. These
93 generally do not have any undefined values, and are marked with the
94 '<tt>def</tt>' keyword.</p>
96 <p><b>TableGen classes</b> are abstract records that are used to build and
97 describe other records. These 'classes' allow the end-user to build
98 abstractions for either the domain they are targeting (such as "Register",
99 "RegisterClass", and "Instruction" in the LLVM code generator) or for the
100 implementor to help factor out common properties of records (such as "FPInst",
101 which is used to represent floating point instructions in the X86 backend).
102 TableGen keeps track of all of the classes that are used to build up a
103 definition, so the backend can find all definitions of a particular class, such
104 as "Instruction".</p>
106 <p><b>TableGen multiclasses</b> are groups of abstract records that are
107 instantiated all at once. Each instantiation can result in multiple
108 TableGen definitions. If a multiclass inherits from another multiclass,
109 the definitions in the sub-multiclass become part of the current
110 multiclass, as if they were declared in the current multiclass.</p>
114 <!-- ======================================================================= -->
115 <div class="doc_subsection"><a name="example">An example record</a></div>
117 <div class="doc_text">
119 <p>With no other arguments, TableGen parses the specified file and prints out
120 all of the classes, then all of the definitions. This is a good way to see what
121 the various definitions expand to fully. Running this on the <tt>X86.td</tt>
122 file prints this (at the time of this writing):</p>
124 <div class="doc_code">
127 <b>def</b> ADD32rr { <i>// Instruction X86Inst I</i>
128 <b>string</b> Namespace = "X86";
129 <b>dag</b> OutOperandList = (outs GR32:$dst);
130 <b>dag</b> InOperandList = (ins GR32:$src1, GR32:$src2);
131 <b>string</b> AsmString = "add{l}\t{$src2, $dst|$dst, $src2}";
132 <b>list</b><dag> Pattern = [(set GR32:$dst, (add GR32:$src1, GR32:$src2))];
133 <b>list</b><Register> Uses = [];
134 <b>list</b><Register> Defs = [EFLAGS];
135 <b>list</b><Predicate> Predicates = [];
136 <b>int</b> CodeSize = 3;
137 <b>int</b> AddedComplexity = 0;
138 <b>bit</b> isReturn = 0;
139 <b>bit</b> isBranch = 0;
140 <b>bit</b> isIndirectBranch = 0;
141 <b>bit</b> isBarrier = 0;
142 <b>bit</b> isCall = 0;
143 <b>bit</b> canFoldAsLoad = 0;
144 <b>bit</b> mayLoad = 0;
145 <b>bit</b> mayStore = 0;
146 <b>bit</b> isImplicitDef = 0;
147 <b>bit</b> isConvertibleToThreeAddress = 1;
148 <b>bit</b> isCommutable = 1;
149 <b>bit</b> isTerminator = 0;
150 <b>bit</b> isReMaterializable = 0;
151 <b>bit</b> isPredicable = 0;
152 <b>bit</b> hasDelaySlot = 0;
153 <b>bit</b> usesCustomInserter = 0;
154 <b>bit</b> hasCtrlDep = 0;
155 <b>bit</b> isNotDuplicable = 0;
156 <b>bit</b> hasSideEffects = 0;
157 <b>bit</b> neverHasSideEffects = 0;
158 InstrItinClass Itinerary = NoItinerary;
159 <b>string</b> Constraints = "";
160 <b>string</b> DisableEncoding = "";
161 <b>bits</b><8> Opcode = { 0, 0, 0, 0, 0, 0, 0, 1 };
162 Format Form = MRMDestReg;
163 <b>bits</b><6> FormBits = { 0, 0, 0, 0, 1, 1 };
164 ImmType ImmT = NoImm;
165 <b>bits</b><3> ImmTypeBits = { 0, 0, 0 };
166 <b>bit</b> hasOpSizePrefix = 0;
167 <b>bit</b> hasAdSizePrefix = 0;
168 <b>bits</b><4> Prefix = { 0, 0, 0, 0 };
169 <b>bit</b> hasREX_WPrefix = 0;
171 <b>bits</b><3> FPFormBits = { 0, 0, 0 };
177 <p>This definition corresponds to a 32-bit register-register add instruction in
178 the X86. The string after the '<tt>def</tt>' string indicates the name of the
179 record—"<tt>ADD32rr</tt>" in this case—and the comment at the end of
180 the line indicates the superclasses of the definition. The body of the record
181 contains all of the data that TableGen assembled for the record, indicating that
182 the instruction is part of the "X86" namespace, the pattern indicating how the
183 the instruction should be emitted into the assembly file, that it is a
184 two-address instruction, has a particular encoding, etc. The contents and
185 semantics of the information in the record is specific to the needs of the X86
186 backend, and is only shown as an example.</p>
188 <p>As you can see, a lot of information is needed for every instruction
189 supported by the code generator, and specifying it all manually would be
190 unmaintainable, prone to bugs, and tiring to do in the first place. Because we
191 are using TableGen, all of the information was derived from the following
194 <div class="doc_code">
197 isCommutable = 1, <i>// X = ADD Y,Z --> X = ADD Z,Y</i>
198 isConvertibleToThreeAddress = 1 <b>in</b> <i>// Can transform into LEA.</i>
199 def ADD32rr : I<0x01, MRMDestReg, (outs GR32:$dst),
200 (ins GR32:$src1, GR32:$src2),
201 "add{l}\t{$src2, $dst|$dst, $src2}",
202 [(set GR32:$dst, (add GR32:$src1, GR32:$src2))]>;
206 <p>This definition makes use of the custom class <tt>I</tt> (extended from the
207 custom class <tt>X86Inst</tt>), which is defined in the X86-specific TableGen
208 file, to factor out the common features that instructions of its class share. A
209 key feature of TableGen is that it allows the end-user to define the
210 abstractions they prefer to use when describing their information.</p>
214 <!-- ======================================================================= -->
215 <div class="doc_subsection"><a name="running">Running TableGen</a></div>
217 <div class="doc_text">
219 <p>TableGen runs just like any other LLVM tool. The first (optional) argument
220 specifies the file to read. If a filename is not specified, <tt>tblgen</tt>
221 reads from standard input.</p>
223 <p>To be useful, one of the <a href="#backends">TableGen backends</a> must be
224 used. These backends are selectable on the command line (type '<tt>tblgen
225 -help</tt>' for a list). For example, to get a list of all of the definitions
226 that subclass a particular type (which can be useful for building up an enum
227 list of these records), use the <tt>-print-enums</tt> option:</p>
229 <div class="doc_code">
231 $ tblgen X86.td -print-enums -class=Register
232 AH, AL, AX, BH, BL, BP, BPL, BX, CH, CL, CX, DH, DI, DIL, DL, DX, EAX, EBP, EBX,
233 ECX, EDI, EDX, EFLAGS, EIP, ESI, ESP, FP0, FP1, FP2, FP3, FP4, FP5, FP6, IP,
234 MM0, MM1, MM2, MM3, MM4, MM5, MM6, MM7, R10, R10B, R10D, R10W, R11, R11B, R11D,
235 R11W, R12, R12B, R12D, R12W, R13, R13B, R13D, R13W, R14, R14B, R14D, R14W, R15,
236 R15B, R15D, R15W, R8, R8B, R8D, R8W, R9, R9B, R9D, R9W, RAX, RBP, RBX, RCX, RDI,
237 RDX, RIP, RSI, RSP, SI, SIL, SP, SPL, ST0, ST1, ST2, ST3, ST4, ST5, ST6, ST7,
238 XMM0, XMM1, XMM10, XMM11, XMM12, XMM13, XMM14, XMM15, XMM2, XMM3, XMM4, XMM5,
239 XMM6, XMM7, XMM8, XMM9,
241 $ tblgen X86.td -print-enums -class=Instruction
242 ABS_F, ABS_Fp32, ABS_Fp64, ABS_Fp80, ADC32mi, ADC32mi8, ADC32mr, ADC32ri,
243 ADC32ri8, ADC32rm, ADC32rr, ADC64mi32, ADC64mi8, ADC64mr, ADC64ri32, ADC64ri8,
244 ADC64rm, ADC64rr, ADD16mi, ADD16mi8, ADD16mr, ADD16ri, ADD16ri8, ADD16rm,
245 ADD16rr, ADD32mi, ADD32mi8, ADD32mr, ADD32ri, ADD32ri8, ADD32rm, ADD32rr,
246 ADD64mi32, ADD64mi8, ADD64mr, ADD64ri32, ...
250 <p>The default backend prints out all of the records, as described <a
251 href="#example">above</a>.</p>
253 <p>If you plan to use TableGen, you will most likely have to <a
254 href="#backends">write a backend</a> that extracts the information specific to
255 what you need and formats it in the appropriate way.</p>
260 <!-- *********************************************************************** -->
261 <div class="doc_section"><a name="syntax">TableGen syntax</a></div>
262 <!-- *********************************************************************** -->
264 <div class="doc_text">
266 <p>TableGen doesn't care about the meaning of data (that is up to the backend to
267 define), but it does care about syntax, and it enforces a simple type system.
268 This section describes the syntax and the constructs allowed in a TableGen file.
273 <!-- ======================================================================= -->
274 <div class="doc_subsection"><a name="primitives">TableGen primitives</a></div>
276 <!-- -------------------------------------------------------------------------->
277 <div class="doc_subsubsection"><a name="comments">TableGen comments</a></div>
279 <div class="doc_text">
281 <p>TableGen supports BCPL style "<tt>//</tt>" comments, which run to the end of
282 the line, and it also supports <b>nestable</b> "<tt>/* */</tt>" comments.</p>
286 <!-- -------------------------------------------------------------------------->
287 <div class="doc_subsubsection">
288 <a name="types">The TableGen type system</a>
291 <div class="doc_text">
293 <p>TableGen files are strongly typed, in a simple (but complete) type-system.
294 These types are used to perform automatic conversions, check for errors, and to
295 help interface designers constrain the input that they allow. Every <a
296 href="#valuedef">value definition</a> is required to have an associated type.
299 <p>TableGen supports a mixture of very low-level types (such as <tt>bit</tt>)
300 and very high-level types (such as <tt>dag</tt>). This flexibility is what
301 allows it to describe a wide range of information conveniently and compactly.
302 The TableGen types are:</p>
305 <dt><tt><b>bit</b></tt></dt>
306 <dd>A 'bit' is a boolean value that can hold either 0 or 1.</dd>
308 <dt><tt><b>int</b></tt></dt>
309 <dd>The 'int' type represents a simple 32-bit integer value, such as 5.</dd>
311 <dt><tt><b>string</b></tt></dt>
312 <dd>The 'string' type represents an ordered sequence of characters of
313 arbitrary length.</dd>
315 <dt><tt><b>bits</b><n></tt></dt>
316 <dd>A 'bits' type is an arbitrary, but fixed, size integer that is broken up
317 into individual bits. This type is useful because it can handle some bits
318 being defined while others are undefined.</dd>
320 <dt><tt><b>list</b><ty></tt></dt>
321 <dd>This type represents a list whose elements are some other type. The
322 contained type is arbitrary: it can even be another list type.</dd>
325 <dd>Specifying a class name in a type context means that the defined value
326 must be a subclass of the specified class. This is useful in conjunction with
327 the <b><tt>list</tt></b> type, for example, to constrain the elements of the
328 list to a common base class (e.g., a <tt><b>list</b><Register></tt> can
329 only contain definitions derived from the "<tt>Register</tt>" class).</dd>
331 <dt><tt><b>dag</b></tt></dt>
332 <dd>This type represents a nestable directed graph of elements.</dd>
334 <dt><tt><b>code</b></tt></dt>
335 <dd>This represents a big hunk of text. This is lexically distinct from
336 string values because it doesn't require escapeing double quotes and other
337 common characters that occur in code.</dd>
340 <p>To date, these types have been sufficient for describing things that
341 TableGen has been used for, but it is straight-forward to extend this list if
346 <!-- -------------------------------------------------------------------------->
347 <div class="doc_subsubsection">
348 <a name="values">TableGen values and expressions</a>
351 <div class="doc_text">
353 <p>TableGen allows for a pretty reasonable number of different expression forms
354 when building up values. These forms allow the TableGen file to be written in a
355 natural syntax and flavor for the application. The current expression forms
356 supported include:</p>
360 <dd>uninitialized field</dd>
361 <dt><tt>0b1001011</tt></dt>
362 <dd>binary integer value</dd>
363 <dt><tt>07654321</tt></dt>
364 <dd>octal integer value (indicated by a leading 0)</dd>
366 <dd>decimal integer value</dd>
367 <dt><tt>0x7F</tt></dt>
368 <dd>hexadecimal integer value</dd>
369 <dt><tt>"foo"</tt></dt>
370 <dd>string value</dd>
371 <dt><tt>[{ ... }]</tt></dt>
372 <dd>code fragment</dd>
373 <dt><tt>[ X, Y, Z ]<type></tt></dt>
374 <dd>list value. <type> is the type of the list
375 element and is usually optional. In rare cases,
376 TableGen is unable to deduce the element type in
377 which case the user must specify it explicitly.</dd>
378 <dt><tt>{ a, b, c }</tt></dt>
379 <dd>initializer for a "bits<3>" value</dd>
380 <dt><tt>value</tt></dt>
381 <dd>value reference</dd>
382 <dt><tt>value{17}</tt></dt>
383 <dd>access to one bit of a value</dd>
384 <dt><tt>value{15-17}</tt></dt>
385 <dd>access to multiple bits of a value</dd>
386 <dt><tt>DEF</tt></dt>
387 <dd>reference to a record definition</dd>
388 <dt><tt>CLASS<val list></tt></dt>
389 <dd>reference to a new anonymous definition of CLASS with the specified
390 template arguments.</dd>
391 <dt><tt>X.Y</tt></dt>
392 <dd>reference to the subfield of a value</dd>
393 <dt><tt>list[4-7,17,2-3]</tt></dt>
394 <dd>A slice of the 'list' list, including elements 4,5,6,7,17,2, and 3 from
395 it. Elements may be included multiple times.</dd>
396 <dt><tt>(DEF a, b)</tt></dt>
397 <dd>a dag value. The first element is required to be a record definition, the
398 remaining elements in the list may be arbitrary other values, including nested
399 `<tt>dag</tt>' values.</dd>
400 <dt><tt>!strconcat(a, b)</tt></dt>
401 <dd>A string value that is the result of concatenating the 'a' and 'b'
403 <dt><tt>!cast<type>(a)</tt></dt>
404 <dd>A symbol of type <em>type</em> obtained by looking up the string 'a' in
405 the symbol table. If the type of 'a' does not match <em>type</em>, TableGen
406 aborts with an error. !cast<string> is a special case in that the argument must
407 be an object defined by a 'def' construct.</dd>
408 <dt><tt>!nameconcat<type>(a, b)</tt></dt>
409 <dd>Shorthand for !cast<type>(!strconcat(a, b))</dd>
410 <dt><tt>!subst(a, b, c)</tt></dt>
411 <dd>If 'a' and 'b' are of string type or are symbol references, substitute
412 'b' for 'a' in 'c.' This operation is analogous to $(subst) in GNU make.</dd>
413 <dt><tt>!foreach(a, b, c)</tt></dt>
414 <dd>For each member 'b' of dag or list 'a' apply operator 'c.' 'b' is a
415 dummy variable that should be declared as a member variable of an instantiated
416 class. This operation is analogous to $(foreach) in GNU make.</dd>
417 <dt><tt>!car(a)</tt></dt>
418 <dd>The first element of list 'a.'</dd>
419 <dt><tt>!cdr(a)</tt></dt>
420 <dd>The 2nd-N elements of list 'a.'</dd>
421 <dt><tt>!null(a)</tt></dt>
422 <dd>An integer {0,1} indicating whether list 'a' is empty.</dd>
423 <dt><tt>!if(a,b,c)</tt></dt>
424 <dd>'b' if the result of 'int' or 'bit' operator 'a' is nonzero,
426 <dt><tt>!eq(a,b)</tt></dt>
427 <dd>Integer one if string a is equal to string b, zero otherwise. This
428 only operates on string, int and bit objects. Use !cast<string> to
429 compare other types of objects.</dd>
432 <p>Note that all of the values have rules specifying how they convert to values
433 for different types. These rules allow you to assign a value like "<tt>7</tt>"
434 to a "<tt>bits<4></tt>" value, for example.</p>
438 <!-- ======================================================================= -->
439 <div class="doc_subsection">
440 <a name="classesdefs">Classes and definitions</a>
443 <div class="doc_text">
445 <p>As mentioned in the <a href="#concepts">intro</a>, classes and definitions
446 (collectively known as 'records') in TableGen are the main high-level unit of
447 information that TableGen collects. Records are defined with a <tt>def</tt> or
448 <tt>class</tt> keyword, the record name, and an optional list of "<a
449 href="#templateargs">template arguments</a>". If the record has superclasses,
450 they are specified as a comma separated list that starts with a colon character
451 ("<tt>:</tt>"). If <a href="#valuedef">value definitions</a> or <a
452 href="#recordlet">let expressions</a> are needed for the class, they are
453 enclosed in curly braces ("<tt>{}</tt>"); otherwise, the record ends with a
456 <p>Here is a simple TableGen file:</p>
458 <div class="doc_code">
460 <b>class</b> C { <b>bit</b> V = 1; }
463 <b>string</b> Greeting = "hello";
468 <p>This example defines two definitions, <tt>X</tt> and <tt>Y</tt>, both of
469 which derive from the <tt>C</tt> class. Because of this, they both get the
470 <tt>V</tt> bit value. The <tt>Y</tt> definition also gets the Greeting member
473 <p>In general, classes are useful for collecting together the commonality
474 between a group of records and isolating it in a single place. Also, classes
475 permit the specification of default values for their subclasses, allowing the
476 subclasses to override them as they wish.</p>
480 <!---------------------------------------------------------------------------->
481 <div class="doc_subsubsection">
482 <a name="valuedef">Value definitions</a>
485 <div class="doc_text">
487 <p>Value definitions define named entries in records. A value must be defined
488 before it can be referred to as the operand for another value definition or
489 before the value is reset with a <a href="#recordlet">let expression</a>. A
490 value is defined by specifying a <a href="#types">TableGen type</a> and a name.
491 If an initial value is available, it may be specified after the type with an
492 equal sign. Value definitions require terminating semicolons.</p>
496 <!-- -------------------------------------------------------------------------->
497 <div class="doc_subsubsection">
498 <a name="recordlet">'let' expressions</a>
501 <div class="doc_text">
503 <p>A record-level let expression is used to change the value of a value
504 definition in a record. This is primarily useful when a superclass defines a
505 value that a derived class or definition wants to override. Let expressions
506 consist of the '<tt>let</tt>' keyword followed by a value name, an equal sign
507 ("<tt>=</tt>"), and a new value. For example, a new class could be added to the
508 example above, redefining the <tt>V</tt> field for all of its subclasses:</p>
510 <div class="doc_code">
512 <b>class</b> D : C { let V = 0; }
517 <p>In this case, the <tt>Z</tt> definition will have a zero value for its "V"
518 value, despite the fact that it derives (indirectly) from the <tt>C</tt> class,
519 because the <tt>D</tt> class overrode its value.</p>
523 <!-- -------------------------------------------------------------------------->
524 <div class="doc_subsubsection">
525 <a name="templateargs">Class template arguments</a>
528 <div class="doc_text">
530 <p>TableGen permits the definition of parameterized classes as well as normal
531 concrete classes. Parameterized TableGen classes specify a list of variable
532 bindings (which may optionally have defaults) that are bound when used. Here is
533 a simple example:</p>
535 <div class="doc_code">
537 <b>class</b> FPFormat<<b>bits</b><3> val> {
538 <b>bits</b><3> Value = val;
540 <b>def</b> NotFP : FPFormat<0>;
541 <b>def</b> ZeroArgFP : FPFormat<1>;
542 <b>def</b> OneArgFP : FPFormat<2>;
543 <b>def</b> OneArgFPRW : FPFormat<3>;
544 <b>def</b> TwoArgFP : FPFormat<4>;
545 <b>def</b> CompareFP : FPFormat<5>;
546 <b>def</b> CondMovFP : FPFormat<6>;
547 <b>def</b> SpecialFP : FPFormat<7>;
551 <p>In this case, template arguments are used as a space efficient way to specify
552 a list of "enumeration values", each with a "<tt>Value</tt>" field set to the
553 specified integer.</p>
555 <p>The more esoteric forms of <a href="#values">TableGen expressions</a> are
556 useful in conjunction with template arguments. As an example:</p>
558 <div class="doc_code">
560 <b>class</b> ModRefVal<<b>bits</b><2> val> {
561 <b>bits</b><2> Value = val;
564 <b>def</b> None : ModRefVal<0>;
565 <b>def</b> Mod : ModRefVal<1>;
566 <b>def</b> Ref : ModRefVal<2>;
567 <b>def</b> ModRef : ModRefVal<3>;
569 <b>class</b> Value<ModRefVal MR> {
570 <i>// Decode some information into a more convenient format, while providing
571 // a nice interface to the user of the "Value" class.</i>
572 <b>bit</b> isMod = MR.Value{0};
573 <b>bit</b> isRef = MR.Value{1};
575 <i>// other stuff...</i>
578 <i>// Example uses</i>
579 <b>def</b> bork : Value<Mod>;
580 <b>def</b> zork : Value<Ref>;
581 <b>def</b> hork : Value<ModRef>;
585 <p>This is obviously a contrived example, but it shows how template arguments
586 can be used to decouple the interface provided to the user of the class from the
587 actual internal data representation expected by the class. In this case,
588 running <tt>tblgen</tt> on the example prints the following definitions:</p>
590 <div class="doc_code">
592 <b>def</b> bork { <i>// Value</i>
593 <b>bit</b> isMod = 1;
594 <b>bit</b> isRef = 0;
596 <b>def</b> hork { <i>// Value</i>
597 <b>bit</b> isMod = 1;
598 <b>bit</b> isRef = 1;
600 <b>def</b> zork { <i>// Value</i>
601 <b>bit</b> isMod = 0;
602 <b>bit</b> isRef = 1;
607 <p> This shows that TableGen was able to dig into the argument and extract a
608 piece of information that was requested by the designer of the "Value" class.
609 For more realistic examples, please see existing users of TableGen, such as the
614 <!-- -------------------------------------------------------------------------->
615 <div class="doc_subsubsection">
616 <a name="multiclass">Multiclass definitions and instances</a>
619 <div class="doc_text">
622 While classes with template arguments are a good way to factor commonality
623 between two instances of a definition, multiclasses allow a convenient notation
624 for defining multiple definitions at once (instances of implicitly constructed
625 classes). For example, consider an 3-address instruction set whose instructions
626 come in two forms: "<tt>reg = reg op reg</tt>" and "<tt>reg = reg op imm</tt>"
627 (e.g. SPARC). In this case, you'd like to specify in one place that this
628 commonality exists, then in a separate place indicate what all the ops are.
632 Here is an example TableGen fragment that shows this idea:
635 <div class="doc_code">
640 <b>class</b> inst<<b>int</b> opc, <b>string</b> asmstr, <b>dag</b> operandlist>;
642 <b>multiclass</b> ri_inst<<b>int</b> opc, <b>string</b> asmstr> {
643 def _rr : inst<opc, !strconcat(asmstr, " $dst, $src1, $src2"),
644 (ops GPR:$dst, GPR:$src1, GPR:$src2)>;
645 def _ri : inst<opc, !strconcat(asmstr, " $dst, $src1, $src2"),
646 (ops GPR:$dst, GPR:$src1, Imm:$src2)>;
649 <i>// Instantiations of the ri_inst multiclass.</i>
650 <b>defm</b> ADD : ri_inst<0b111, "add">;
651 <b>defm</b> SUB : ri_inst<0b101, "sub">;
652 <b>defm</b> MUL : ri_inst<0b100, "mul">;
657 <p>The name of the resultant definitions has the multidef fragment names
658 appended to them, so this defines <tt>ADD_rr</tt>, <tt>ADD_ri</tt>,
659 <tt>SUB_rr</tt>, etc. A defm may inherit from multiple multiclasses,
660 instantiating definitions from each multiclass. Using a multiclass
661 this way is exactly equivalent to instantiating the classes multiple
662 times yourself, e.g. by writing:</p>
664 <div class="doc_code">
669 <b>class</b> inst<<b>int</b> opc, <b>string</b> asmstr, <b>dag</b> operandlist>;
671 <b>class</b> rrinst<<b>int</b> opc, <b>string</b> asmstr>
672 : inst<opc, !strconcat(asmstr, " $dst, $src1, $src2"),
673 (ops GPR:$dst, GPR:$src1, GPR:$src2)>;
675 <b>class</b> riinst<<b>int</b> opc, <b>string</b> asmstr>
676 : inst<opc, !strconcat(asmstr, " $dst, $src1, $src2"),
677 (ops GPR:$dst, GPR:$src1, Imm:$src2)>;
679 <i>// Instantiations of the ri_inst multiclass.</i>
680 <b>def</b> ADD_rr : rrinst<0b111, "add">;
681 <b>def</b> ADD_ri : riinst<0b111, "add">;
682 <b>def</b> SUB_rr : rrinst<0b101, "sub">;
683 <b>def</b> SUB_ri : riinst<0b101, "sub">;
684 <b>def</b> MUL_rr : rrinst<0b100, "mul">;
685 <b>def</b> MUL_ri : riinst<0b100, "mul">;
691 A defm can also be used inside a multiclass providing several levels of
692 multiclass instanciations.
695 <div class="doc_code">
697 <b>class</b> Instruction<bits<4> opc, string Name> {
698 bits<4> opcode = opc;
702 <b>multiclass</b> basic_r<bits<4> opc> {
703 <b>def</b> rr : Instruction<opc, "rr">;
704 <b>def</b> rm : Instruction<opc, "rm">;
707 <b>multiclass</b> basic_s<bits<4> opc> {
708 <b>defm</b> SS : basic_r<opc>;
709 <b>defm</b> SD : basic_r<opc>;
710 <b>def</b> X : Instruction<opc, "x">;
713 <b>multiclass</b> basic_p<bits<4> opc> {
714 <b>defm</b> PS : basic_r<opc>;
715 <b>defm</b> PD : basic_r<opc>;
716 <b>def</b> Y : Instruction<opc, "y">;
719 <b>defm</b> ADD : basic_s<0xf>, basic_p<0xf>;
723 <b>def</b> ADDPDrm { ...
724 <b>def</b> ADDPDrr { ...
725 <b>def</b> ADDPSrm { ...
726 <b>def</b> ADDPSrr { ...
727 <b>def</b> ADDSDrm { ...
728 <b>def</b> ADDSDrr { ...
729 <b>def</b> ADDY { ...
730 <b>def</b> ADDX { ...
735 defm declarations can inherit from classes too, the
736 rule to follow is that the class list must start after the
737 last multiclass, and there must be at least one multiclass
741 <div class="doc_code">
743 <b>class</b> XD { bits<4> Prefix = 11; }
744 <b>class</b> XS { bits<4> Prefix = 12; }
746 <b>class</b> I<bits<4> op> {
747 bits<4> opcode = op;
750 <b>multiclass</b> R {
751 <b>def</b> rr : I<4>;
752 <b>def</b> rm : I<2>;
755 <b>multiclass</b> Y {
756 <b>defm</b> SS : R, XD;
757 <b>defm</b> SD : R, XS;
760 <b>defm</b> Instr : Y;
763 <b>def</b> InstrSDrm {
764 bits<4> opcode = { 0, 0, 1, 0 };
765 bits<4> Prefix = { 1, 1, 0, 0 };
768 <b>def</b> InstrSSrr {
769 bits<4> opcode = { 0, 1, 0, 0 };
770 bits<4> Prefix = { 1, 0, 1, 1 };
777 <!-- ======================================================================= -->
778 <div class="doc_subsection">
779 <a name="filescope">File scope entities</a>
782 <!-- -------------------------------------------------------------------------->
783 <div class="doc_subsubsection">
784 <a name="include">File inclusion</a>
787 <div class="doc_text">
788 <p>TableGen supports the '<tt>include</tt>' token, which textually substitutes
789 the specified file in place of the include directive. The filename should be
790 specified as a double quoted string immediately after the '<tt>include</tt>'
791 keyword. Example:</p>
793 <div class="doc_code">
795 <b>include</b> "foo.td"
801 <!-- -------------------------------------------------------------------------->
802 <div class="doc_subsubsection">
803 <a name="globallet">'let' expressions</a>
806 <div class="doc_text">
808 <p>"Let" expressions at file scope are similar to <a href="#recordlet">"let"
809 expressions within a record</a>, except they can specify a value binding for
810 multiple records at a time, and may be useful in certain other cases.
811 File-scope let expressions are really just another way that TableGen allows the
812 end-user to factor out commonality from the records.</p>
814 <p>File-scope "let" expressions take a comma-separated list of bindings to
815 apply, and one or more records to bind the values in. Here are some
818 <div class="doc_code">
820 <b>let</b> isTerminator = 1, isReturn = 1, isBarrier = 1, hasCtrlDep = 1 <b>in</b>
821 <b>def</b> RET : I<0xC3, RawFrm, (outs), (ins), "ret", [(X86retflag 0)]>;
823 <b>let</b> isCall = 1 <b>in</b>
824 <i>// All calls clobber the non-callee saved registers...</i>
825 <b>let</b> Defs = [EAX, ECX, EDX, FP0, FP1, FP2, FP3, FP4, FP5, FP6, ST0,
826 MM0, MM1, MM2, MM3, MM4, MM5, MM6, MM7,
827 XMM0, XMM1, XMM2, XMM3, XMM4, XMM5, XMM6, XMM7, EFLAGS] <b>in</b> {
828 <b>def</b> CALLpcrel32 : Ii32<0xE8, RawFrm, (outs), (ins i32imm:$dst,variable_ops),
829 "call\t${dst:call}", []>;
830 <b>def</b> CALL32r : I<0xFF, MRM2r, (outs), (ins GR32:$dst, variable_ops),
831 "call\t{*}$dst", [(X86call GR32:$dst)]>;
832 <b>def</b> CALL32m : I<0xFF, MRM2m, (outs), (ins i32mem:$dst, variable_ops),
833 "call\t{*}$dst", []>;
838 <p>File-scope "let" expressions are often useful when a couple of definitions
839 need to be added to several records, and the records do not otherwise need to be
840 opened, as in the case with the <tt>CALL*</tt> instructions above.</p>
842 <p>It's also possible to use "let" expressions inside multiclasses, providing
843 more ways to factor out commonality from the records, specially if using
844 several levels of multiclass instanciations. This also avoids the need of using
845 "let" expressions within subsequent records inside a multiclass.</p>
847 <div class="doc_code">
849 <b>multiclass </b>basic_r<bits<4> opc> {
850 <b>let </b>Predicates = [HasSSE2] in {
851 <b>def </b>rr : Instruction<opc, "rr">;
852 <b>def </b>rm : Instruction<opc, "rm">;
854 <b>let </b>Predicates = [HasSSE3] in
855 <b>def </b>rx : Instruction<opc, "rx">;
858 <b>multiclass </b>basic_ss<bits<4> opc> {
859 <b>let </b>IsDouble = 0 in
860 <b>defm </b>SS : basic_r<opc>;
862 <b>let </b>IsDouble = 1 in
863 <b>defm </b>SD : basic_r<opc>;
866 <b>defm </b>ADD : basic_ss<0xf>;
870 <!-- *********************************************************************** -->
871 <div class="doc_section"><a name="codegen">Code Generator backend info</a></div>
872 <!-- *********************************************************************** -->
874 <p>Expressions used by code generator to describe instructions and isel
877 <div class="doc_text">
879 <dt><tt>(implicit a)</tt></dt>
880 <dd>an implicitly defined physical register. This tells the dag instruction
881 selection emitter the input pattern's extra definitions matches implicit
882 physical register definitions.</dd>
886 <!-- *********************************************************************** -->
887 <div class="doc_section"><a name="backends">TableGen backends</a></div>
888 <!-- *********************************************************************** -->
890 <div class="doc_text">
892 <p>TODO: How they work, how to write one. This section should not contain
893 details about any particular backend, except maybe -print-enums as an example.
894 This should highlight the APIs in <tt>TableGen/Record.h</tt>.</p>
898 <!-- *********************************************************************** -->
902 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
903 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
904 <a href="http://validator.w3.org/check/referer"><img
905 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
907 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
908 <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br>
909 Last modified: $Date$